Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Deep sequencing of HIV-1 envelope determines coreceptor usage and predicts virologic responses to antiretroviral… Swenson, Luke Christopher 2014

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2014_september_swenson_luke.pdf [ 9.23MB ]
Metadata
JSON: 24-1.0167457.json
JSON-LD: 24-1.0167457-ld.json
RDF/XML (Pretty): 24-1.0167457-rdf.xml
RDF/JSON: 24-1.0167457-rdf.json
Turtle: 24-1.0167457-turtle.txt
N-Triples: 24-1.0167457-rdf-ntriples.txt
Original Record: 24-1.0167457-source.json
Full Text
24-1.0167457-fulltext.txt
Citation
24-1.0167457.ris

Full Text

DEEP SEQUENCING OF HIV-1 ENVELOPE DETERMINES CORECEPTOR USAGE AND PREDICTS VIROLOGIC RESPONSES TO ANTIRETROVIRAL TREATMENT WITH CCR5 ANTAGONIST MEDICATIONbyLuke Christopher SwensonB.Sc., The University of British Columbia, 2008A THESIS SUBMITTED IN PARTIAL FULFILMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES(Experimental Medicine)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)May 2014© Luke Christopher Swenson, 2014iAbstractNext-generation  sequencing  can  be  used  to  genotype  an  array  of  HIV  variants  within  clinicalspecimens, in a process referred to as deep sequencing. When directed at the gene for HIV envelope,this approach can be used to generate a high-sensitivity overview of the viral tropism of the HIVquasispecies in an infected individual. Since HIV variants with tropism for the CXCR4 coreceptor arenot susceptible to the CCR5 antagonist agent maraviroc, their detection is crucial in order to screenout patients unlikely to achieve viral load declines on maraviroc. There are several assays that testHIV tropism, but it has been unclear as to which are most useful in the clinical setting. The aim ofthis thesis is to compare next-generation sequencing using massively parallel pyrosequencing againstseveral alternative tropism assays in a total of four large randomized clinical trials of maraviroc. Themethodologies are refned and optimized in the early chapters and applied to clinical  specimensfrom a total of 2864 HIV-infected individuals.  The  concordance  of  next-generation  sequencing  with  phenotypic  tropism  assays  reached  87%.Relative to both the original Trofle phenotypic assay and standard population-based sequencing,deep sequencing had higher sensitivity to detect minority non-R5 HIV. Where assays gave discordantresults,  deep  sequencing  tended  to  outperform  the  comparator  assay  and  was  able  to  betterdiscriminate maraviroc responders from non-responders. Next-generation sequencing had excellentperformance in populations of both treatment-experienced and treatment-naïve individuals. It wasconsistently able to determine coreceptor usage, and to predict which patients would respond tomaraviroc.  It  could be performed using either  HIV RNA from blood plasma or  HIV DNA fromperipheral blood mononuclear cells. Additionally, longitudinal deep sequencing was performed onsamples  taken  prior  to  maraviroc  administration  and  again  at  treatment  failure.  Phylogeneticanalyses  confrmed that  the non-R5 variants present  at  time of maraviroc treatment failure  werederived from variants detected by deep sequencing before treatment was initiated. In conclusion,next-generation sequencing was applied to thousands of samples from phase III clinical trials, andwas  a  superior  screening  tool  to  those  originally  used  during  trial  enrollment.  This  thesisdemonstrates the clinical utility of next-generation sequencing.iiPrefaceVersions or components of Chapters 1 through 6 have been published in the scientifc literature. Thecandidate is the primary author on all seven of the published manuscripts, and their contents havebeen reprinted with permission from their respective journals. This statement is to certify that the candidate was the major contributor to study design, collectionand/or supervision of collection of laboratory data, and performed the majority of the data analysis.The candidate also researched and wrote all seven of the manuscripts. Primary co-authors of these manuscripts include the thesis supervisor (Richard Harrigan), statisticaland data analysts  (Chanson Brumme, Dennison Chan,  Chris Glascock,  Art  Poon,  Conan Woods,Brian  Wynhoven,  and Xiaoyin  Zhong),  laboratory  technicians  (Celia  Chui,  Winnie  Dong,  DavidKnapp, Theresa Mo, and Andrew Moores), and external collaborators.The details of these publications are as follows:Swenson, L.C., Boehme, R., Thielen, A., McGovern, R.A., Harrigan, P.R. Genotypic determination ofHIV-1 tropism in the clinical setting. HIV Therapy 2010; 4(3): 293-303. © 2010 Future Medicine, Inc.Swenson, L.C., Däumer, M., Paredes, R. Next-Generation Sequencing to Assess HIV Tropism. CurrentOpinion in HIV and AIDS 2012; 7(5): 478-485. © 2012 Wolters Kluwer Health.Swenson, L.C., Moores, A., Low, A.J., Thielen, A., Dong, W., Woods, C., Jensen, M.A., Wynhoven, B.,Chan, D., Glascock, C., Harrigan, P.R. Improved Detection of CXCR4-Using HIV by V3 Genotyping:Application of Population-based and Deep Sequencing to Plasma RNA and Proviral DNA. Journal ofAcquired Immune Deficiency Syndromes 2010; 54(5): 506-510. © 2010 Wolters Kluwer Health.iiiSwenson, L.C., Mo, T., Dong, W.W.Y., Zhong, X., Woods, C.K., Jensen, M.A., Thielen, A., Chapman,D., Lewis, M., James, I.,  Heera, J., Valdez, H., Harrigan, P.R. Deep Sequencing to Infer HIV-1 Co-Receptor Usage: Application to Three Clinical Trials of Maraviroc in Treatment-Experienced Patients.Journal of Infectious Diseases 2011; 203(2): 237-245. © 2011 Oxford University Press.Swenson, L.C., Dong, W.W.Y., Mo, T., Demarest,  J.,  Chapman, D., Ellery, S., Heera, J.,  Valdez, H.,Poon, A.F.Y., Harrigan, P.R. Use of Cellular HIV DNA to Predict Virologic Response to Maraviroc:Performance of Population-based and Deep Sequencing. Clinical Infectious Diseases 2013; 56(11): 1659-1666. © 2013 Oxford University Press.Swenson L.C., Chui C.K.S., Brumme C.J., Chan D., Woods C.K., Mo T., Dong W., Chapman D., LewisM., Demarest J.F., James I., Portsmouth S., Goodrich J., Heera J., Valdez H., Harrigan P.R.. GenotypicAnalysis  of  the  HIV V3 Region in  Virologic  Non-Responders  to  Maraviroc-containing RegimensReveals Distinct Patterns of Failure. Antimicrobial Agents and Chemotherapy 2013; 57(12): 6122-6130. ©2013 American Society for Microbiology.Swenson, L.C., Mo, T.,  Dong, W.W.Y.,  Zhong,  X.,  Woods,  C.K.,  Thielen,  A.,  Jensen, M.A.,  Knapp,D.J.H.F., Chapman, D., Portsmouth, S., Lewis, M., James, I., Heera, J., Valdez, H., Harrigan, P.R.. DeepV3 Sequencing for HIV Type 1 Tropism in Treatment-Naïve Patients: A Reanalysis of the MERIT Trialof Maraviroc. Clinical Infectious Diseases 2011; 53(7): 732-742. © 2011 Oxford University Press.These  studies  received  ethical  approval  from the  Providence  Health  Care–University  of  BritishColumbia Research Ethics Board: H07-00987 & H07-01901.ivTable of ContentsAbstract............................................................................................................. iiPreface.............................................................................................................. iiiTable of Contents.............................................................................................vList of Tables.................................................................................................. xiiList of Figures............................................................................................... xiiiList of Abbreviations...................................................................................xviAcknowledgments........................................................................................xixForeword........................................................................................................ xxiChapter 1:     General Introduction & Thesis Objectives.........................11.1     Background.............................................................................................................. 11.1.1     The Origins & Current State of the HIV Epidemic.................................................................. 11.1.2     The Structure & Replication Cycle of HIV-1............................................................................. 21.1.3     HIV Target Cells............................................................................................................................ 71.1.4     Natural History & Pathogenesis of HIV-1 Infection in Humans...........................................91.2     HIV-1 Coreceptor Usage...................................................................................... 101.2.1     Chemokine Receptors.................................................................................................................101.2.2     Interactions of the Viral Surface Glycoproteins with Cellular Surface Receptors & Coreceptors............................................................................................................................................... 121.2.3     History of Cellular HIV-1 Tropism........................................................................................... 16v1.2.4     Clinical Relevance of HIV-1 Tropism....................................................................................... 171.2.5     Phenotypic Assays to Detect HIV-1 Coreceptor Usage......................................................... 191.2.6     Sequencing-Based Genotypic Assays to Determine HIV-1 Tropism Using Bioinformatic Interpretation............................................................................................................................................201.2.7     The Development & Performance of the Bioinformatic Algorithms PSSMX4/R5 & Geno2pheno..............................................................................................................................................211.2.8     Alternative Genotypic Assays & Comparisons between Genotypic & Phenotypic Assays.................................................................................................................................................................... 241.2.9     Next-Generation Sequencing to Detect Minority HIV-1 Variants........................................251.3     Treatment of HIV.................................................................................................. 281.3.1     Antiretroviral Treatment of HIV Infection & Development of Resistance.........................281.3.2     Coreceptor Usage & Antiretroviral Therapy...........................................................................301.3.3     Relevance of Minority Variants for Antiretroviral Resistance & Viral Coreceptor Usage311.4     Thesis Overview................................................................................................... 331.4.1     Thesis Organization & Objectives............................................................................................ 331.4.2     Overview of Data Sources......................................................................................................... 34Chapter 2:     Improved Detection of CXCR4-Using HIV by V3 Genotyping: Application of Population-Based & Deep Sequencing to Plasma HIV RNA & Proviral HIV DNA...................................................352.1     Background & Introduction................................................................................352.2     Materials & Methods............................................................................................372.2.1     Cohort Description & Patients.................................................................................................. 372.2.2     Extraction & Population-Based Sequencing........................................................................... 372.2.3     Deep Sequencing & Emulsion PCR Methods......................................................................... 382.2.4     Sequence Analysis & Coreceptor Usage Determination by Bioinformatic Algorithms. . .402.2.5     Independent Validation..............................................................................................................412.3     Results..................................................................................................................... 41vi2.3.1     Standard Sequencing of V3 to Infer Tropism.......................................................................... 412.3.2     Proviral DNA to Infer Tropism................................................................................................. 432.3.3     Deep Sequencing of HIV RNA & DNA................................................................................... 452.3.4     Results of Independent Validation........................................................................................... 462.4     Discussion & Conclusions.................................................................................. 47Chapter 3:     Deep Sequencing to Infer HIV-1 Coreceptor Usage: Application to Three Clinical Trials of Maraviroc in Treatment-Experienced Patients..................................................................................... 513.1     Background & Introduction................................................................................513.2     Materials & Methods............................................................................................523.2.1     Trial Patients, Samples & Amplifcation Methods................................................................. 523.2.2     Emulsion Polymerase Chain Reaction & Pyrosequencing...................................................543.2.3     Optimizing Bioinformatic Cutoffs for Deep Sequencing...................................................... 543.2.4     Bioinformatic Algorithms for Inferring Tropism from Genotypic Data.............................553.2.5     Population-Based Sequencing as a Comparator.................................................................... 593.2.6     Data Analysis...............................................................................................................................593.3     Results..................................................................................................................... 603.3.1     Tropism Screening by Deep Sequencing Relative to Trofle & Population-Based Sequencing................................................................................................................................................ 603.3.2     Early Virologic Response to Maraviroc................................................................................... 623.3.3     Longer-Term Virologic Effcacy................................................................................................ 633.3.4     Changes in Viral Tropism...........................................................................................................663.3.5     Response Stratifed by Background Drug Activity................................................................ 683.3.6     Discordance Amongst Bioinformatic Algorithms.................................................................. 683.3.7     Assay Discordance...................................................................................................................... 713.3.8     Comparison with Independent Replication by an External Laboratory............................743.3.9     Comparison to the Enhanced Sensitivity Trofle Assay........................................................ 76vii3.4     Discussion & Conclusions.................................................................................. 77Chapter 4:     Use of Cellular HIV DNA to Predict Virologic Responses to Maraviroc: Performance of Population-Based & Deep Sequencing........................................................................................................................... 824.1     Background & Introduction................................................................................824.2     Materials & Methods............................................................................................834.2.1     Samples & Patient Composition............................................................................................... 834.2.2     V3 Amplifcation & Sequencing................................................................................................844.2.3     Tropism Prediction......................................................................................................................844.2.4     Ethics Statement.......................................................................................................................... 854.2.5     Data Analysis...............................................................................................................................854.3     Results..................................................................................................................... 864.3.1     Prediction of Virologic Effcacy on Maraviroc........................................................................ 864.3.2     Similar Performance Regardless of Background Regimen Activity.................................... 894.3.3     Prediction of Future Tropism Changes on Maraviroc........................................................... 894.3.4     Diagnostic Performance of Tropism Assays............................................................................894.3.5     Compartmental Differences...................................................................................................... 924.3.6     Virologic Responses with Screening by DNA-Based versus RNA-Based Approaches....934.4     Discussion & Conclusions.................................................................................. 96Chapter 5:     Genotypic Analysis of the HIV-1 V3 Region in Virologic Non-Responders to Maraviroc-Containing Regimens Reveals Distinct Patterns of Failure........................................................................................1005.1     Background & Introduction..............................................................................1005.2     Materials & Methods..........................................................................................1025.2.1     Patient & Sample Selection...................................................................................................... 102viii5.2.2     Genotypic Tropism Testing......................................................................................................1035.2.3     Statistical Analyses....................................................................................................................1045.3     Results................................................................................................................... 1045.3.1     Patient & Sample Composition...............................................................................................1045.3.2     Performance of Population-Based Genotyping for Determining HIV Tropism..............1065.3.3     Change in V3 Sequence & Geno2pheno Value after Maraviroc Treatment......................1075.3.4     Classical Substitutions in Patients with Non-R5 HIV at Failure but Limited Evidence of Maraviroc Resistance in Those with R5 HIV......................................................................................1105.3.5     Change in Non-R5 Viral Population as Determined by Deep Sequencing.......................1115.3.6     Phylogenetic Relationship between Screening & Failure Sequences................................1135.3.7     Comparison of Tropism Methods........................................................................................... 1145.3.8     Virologic Responses to Maraviroc.......................................................................................... 1195.3.9     Comparison to the Enhanced Sensitivity Trofle Assay...................................................... 1215.4     Discussion & Conclusions................................................................................ 121Chapter 6:     Deep V3 Sequencing for HIV-1 Tropism in Treatment-Naïve Patients: A Reanalysis of the MERIT Trial of Maraviroc.........1286.1     Background & Introduction..............................................................................1286.2     Materials & Methods..........................................................................................1296.2.1     Samples & MERIT Trial Design.............................................................................................. 1296.2.2     V3 Amplifcation Method........................................................................................................ 1306.2.3     Bioinformatic Analyses............................................................................................................ 1306.2.4     Ethics Statement........................................................................................................................ 1316.2.5     Data Analysis.............................................................................................................................1316.3     Results................................................................................................................... 1326.3.1     Patient Characteristics.............................................................................................................. 1326.3.2     Identifcation of Non-R5 Screening Samples Using Deep Sequencing.............................1326.3.3     Viral Load Decline from Baseline........................................................................................... 134ix6.3.4     Virologic Suppression...............................................................................................................1346.3.5     Non-Inferiority Analysis.......................................................................................................... 1386.3.6     Changes in HIV Tropism..........................................................................................................1396.3.7     Effects of HIV Subtype............................................................................................................. 1396.3.8     Comparison of Deep Sequencing to the Enhanced Sensitivity Trofle Assay & Population-Based Sequencing..............................................................................................................1426.3.9     Maraviroc Once-Daily Arm..................................................................................................... 1486.4     Discussion & Conclusions................................................................................ 148Chapter 7:     General Discussion & Conclusions..................................1527.1     Thesis Summary & Overall Conclusions....................................................... 1527.2     Specific Conclusions of the Thesis.................................................................. 1537.2.1     Conclusions for Chapter Two..................................................................................................1537.2.2     Conclusions for Chapter Three............................................................................................... 1537.2.3     Conclusions for Chapter Four.................................................................................................1547.2.4     Conclusions for Chapter Five..................................................................................................1557.2.5     Conclusions for Chapter Six.................................................................................................... 1567.2.6     Summary of Specifc Conclusions.......................................................................................... 1577.3     Future Directions & Applications....................................................................157Bibliography................................................................................................. 161Appendices....................................................................................................213Appendix I:     Primers for Chapter 2.................................................................................................. 213Appendix II:     Thermal Cycler Protocols for Chapter 2..................................................................215Appendix III:     Primers for Chapter 3............................................................................................... 216Appendix IV:     Description of Deep Sequencing Data Processing Pipeline................................ 218Appendix V:     Phylogenetic Tree from a Maraviroc Recipient with a Small Pre-Treatment X4 Population Related to the Failure Sequence...................................................................................... 219xAppendix VI:     Phylogenetic Tree from a Maraviroc Recipient with a Small Pre-Treatment X4 Population Related to the Failure Sequence...................................................................................... 220Appendix VII:     Phylogenetic Tree from a Maraviroc Recipient with a Small Pre-Treatment X4 Population Related to the Failure Sequence...................................................................................... 221Appendix VIII:     Phylogenetic Tree from a Maraviroc Recipient with a Large Pre-Treatment X4 Population Related to the Failure Sequence...................................................................................... 222Appendix IX:     Phylogenetic Tree from a Maraviroc Recipient with a Large Pre-Treatment X4 Population Related to the Failure Sequence...................................................................................... 223Appendix X:     Phylogenetic Tree from a Maraviroc Recipient with a Large Pre-Treatment X4 Population Related to the Failure Sequence...................................................................................... 224Appendix XI:     Phylogenetic Tree from a Maraviroc Recipient with a Large Pre-Treatment X4 Population Related to the Failure Sequence...................................................................................... 225Appendix XII:     Phylogenetic Tree from a Maraviroc Recipient for Whom Deep Sequencing Failed to Detect a Pre-Treatment X4 Population Despite Failure with X4 HIV............................226Appendix XIII:     Phylogenetic Tree from a Maraviroc Recipient Who Failed with R5 HIV.....227Appendix XIV:     Phylogenetic Tree from a Maraviroc Recipient Who Failed with R5 HIV.....228xiList of TablesTable 2.1: Deep Sequencing of HIV RNA & DNA Compared with Standard Population-Based Sequencing & the Trofle Assay.............................................................................................................. 45Table 3.1: Baseline Characteristics of Treated Population, Stratifed by Tropism Status by Genotype and Phenotype.......................................................................................................................................... 62Table 3.2: Discordance amongst Bioinformatic Algorithms in the Maraviroc Arms — PSSMX4/R5 versus Geno2pheno.............................................................................................................................................. 70Table 3.3: Discordance amongst Bioinformatic Algorithms in the Maraviroc Arms — PSSMX4/R5 versus PSSMSI/NSI................................................................................................................................................... 70Table 3.4: Discordance amongst Bioinformatic Algorithms in the Maraviroc Arms — PSSMSI/NSI versus Geno2pheno.............................................................................................................................................. 71Table 4.1: Short- & Long-Term Virologic Responses to Maraviroc as Predicted by All Tropism Assays in the Full Dataset.....................................................................................................................................88Table 4.2: Short- & Long-Term Virologic Responses to Maraviroc in Patients with Compromised Background Regimens.............................................................................................................................90Table 4.3: Performance of Sequencing-Based Tropism Assays from Peripheral Blood Mononuclear Cells & Plasma Relative to the Original Trofle Assay in Plasma......................................................92Table 5.1: Patient Characteristics in the Current Study Compared to the MOTIVATE Studies Overall...................................................................................................................................................................105Table 6.1: Baseline Patient Characteristics in the MERIT trial................................................................... 133Table 6.2: Non-Inferiority Analysis between the Maraviroc & Efavirenz Arms..................................... 138Table 6.3: Virologic Outcomes in Maraviroc Recipients Infected with Subtype B HIV-1, Stratifed by Tropism Assessment by Deep Sequencing & the Enhanced Sensitivity Trofle Assay.................141Table 6.4: Virologic Outcomes in Maraviroc Recipients Infected with Non-Subtype B HIV-1, Stratifed by Tropism Assessment with Deep Sequencing & the Enhanced Sensitivity Trofle Assay........141Table 6.5: Overall Virologic Responses of Maraviroc Recipients Grouped by Discordance between Deep Sequencing & the Enhanced Sensitivity Trofle Assay or Population-Based Sequencing.145   xiiList of FiguresFigure 1.1: HIV-1 Replication Cycle & Drug Targets...................................................................................... 3Figure 1.2: Schematic of Native & Bound Structures of HIV-1 gp120........................................................ 14Figure 1.3: Density Maps of Native & Bound Structures of HIV-1 gp120................................................. 15Figure 2.1: PSSM Scores from Standard Population-based Sequencing of Independent Triplicate PCRs of V3 Amplifed from Plasma HIV RNA.............................................................................................. 42Figure 2.2: PSSM Scores from HIV Proviral DNA by Standard Population-Based Sequencing............44Figure 2.3: Distribution of PSSM Scores for Variants Detected by Deep Sequencing of Plasma Samples.....................................................................................................................................................................48Figure 3.1: Sample & Patient Distribution......................................................................................................53Figure 3.2: Areas under Receiver Operating Characteristic Curves for Various Bioinformatic Cutoffs.....................................................................................................................................................................56Figure 3.3: Receiver Operating Characteristic Curves for Optimal Geno2pheno & PSSM Cutoffs — Prediction of Original Trofle Assay Results.........................................................................................57Figure 3.4: Receiver Operating Characteristic Curves for Optimal Geno2pheno & PSSM Cutoffs — Prediction of Week Eight Virologic Success..........................................................................................58Figure 3.5: Sensitivity of Population-Based V3 Sequencing & Original Trofle Assay with Deep Sequencing as the Reference................................................................................................................... 61Figure 3.6: Median Change in Plasma Viral Load from Baseline in the Maraviroc & Placebo Arms....64Figure 3.7: Percentage of Patients with Viral Loads Less than 50 HIV RNA Copies/mL in the Maraviroc & Placebo Arms..................................................................................................................... 65Figure 3.8: Time to Change in Tropism from R5 to Dual/Mixed or X4 in the Maraviroc & Placebo Arms........................................................................................................................................................... 67Figure 3.9: Median Change in Plasma Viral Load from Baseline in Patients with R5 Virus Stratifed by Their Weighted Optimized Background Therapy Susceptibility Score (wOBTss).........................69Figure 3.10: Where Assay Results Were Discordant, the Virologic Responses Tended to Favour Deep Sequencing Results...................................................................................................................................72xiiiFigure 3.11: Virologic Response of Maraviroc Recipients as a Function of the Proportion of Non-R5 HIV at Screening.......................................................................................................................................73Figure 3.12: Correlation of Deep Sequencing Between Two Independent Laboratories.........................75Figure 3.13: Bland-Altman Plot Comparing the Assay Results from Two Independent Laboratories. 76Figure 3.14: Similar Predictions of Plasma Viral Load Changes on Maraviroc by Deep Sequencing & the Enhanced Sensitivity Trofle Assay................................................................................................. 78Figure 3.15: Intermediate Plasma Viral Load Changes Where Deep Sequencing & the Enhanced Sensitivity Trofle Assay Gave Discordant Results..............................................................................79Figure 4.1: Percentage of Patients with Plasma Viral Loads below 50 Copies/mL Was Similar by All Cellular or Plasma-Based Methods........................................................................................................87Figure 4.2: All Genotypic Tropism Testing Methods Predicted Future Phenotypic Tropism Changes While Receiving Maraviroc.....................................................................................................................91Figure 4.3: Effect of Geno2pheno Cutoffs on Prediction of Response to Maraviroc in Patients with Compromised Background Regimens.................................................................................................. 94Figure 4.4: Patients with Discordant Tropism Results Between Compartments Had Virologic Responses Which Favoured the Plasma Prediction............................................................................ 95Figure 5.1: Overall Change in Geno2pheno False-Positive Rate between Screening & Failure...........108Figure 5.2: Individual Geno2pheno False-Positive Rate Values at Screening & Failure........................109Figure 5.3: Individual Changes in Geno2pheno False-Positive Rate Values between Screening & Failure.......................................................................................................................................................110Figure 5.4: The Percentage of Non-R5 Variants in Deep Sequencing Results at Screening & Failure. 113Figure 5.5: Phylogenetic Tree from a Patient Who Had a Small Pre-Treatment X4 Population...........115Figure 5.6: Phylogenetic Tree from a Patient Who Had a Large Pre-Treatment X4 Population...........116Figure 5.7: Phylogenetic Tree from a Patient for Whom Deep Sequencing Failed to Detect X4 HIV. .117Figure 5.8: Phylogenetic Tree from a Patient Who Failed with R5 HIV................................................... 118Figure 5.9: Virologic Responses Were Reduced among Patients with Non-R5 Genotype Results at Failure...................................................................................................................................................... 120Figure 5.10: Overall Change in Geno2pheno False-Positive Rate between Screening & Failure in a Subset of Patients with R5 Results at Screening by the Enhanced Sensitivity Trofle Assay......122xivFigure 5.11: Individual Geno2pheno False-Positive Rate Values at Screening & Failure in a Subset of Patients with R5 Results at Screening by the Enhanced Sensitivity Trofle Assay.......................123Figure 5.12: Individual Changes in Geno2pheno False-Positive Rate Values between Screening & Failure in a Subset of Patients with R5 Results at Screening by the Enhanced Sensitivity Trofle Assay........................................................................................................................................................ 124Figure 6.1: Median Decline in Plasma Viral Load from Baseline in Patients Screened with R5 HIV by Deep Sequencing Who Received Either Maraviroc Twice-Daily or Efavirenz..............................135Figure 6.2: Median Decline in Plasma Viral Load from Baseline in Patients Screened with Non-R5 HIV by Deep Sequencing Who Received Either Maraviroc Twice-Daily or Efavirenz................136Figure 6.3: Percentage of Maraviroc Twice-Daily & Efavirenz Recipients with Plasma Viral Loads Less than 50 Copies/mL with R5 HIV at Screening by Deep Sequencing.....................................137Figure 6.4: Time to Change in Tropism for Maraviroc Twice-Daily Recipients...................................... 140Figure 6.5: Median Decline in Plasma Viral Load from Baseline in Maraviroc Twice-Daily Recipients with Screening by Deep Sequencing & the Enhanced Sensitivity Trofle Assay...........................143Figure 6.6: Percentage of Maraviroc Twice-Daily Recipients with Plasma Viral Loads below 50 Copies/mL with Screening by Deep Sequencing & the Enhanced Sensitivity Trofle Assay.....144Figure 6.7: Declines in Plasma Viral Load from Baseline in Patients with Concordant & Discordant Results between Tropism Assays......................................................................................................... 146Figure 6.8: Proportion of Patients with Plasma Viral Loads below 50 Copies/mL in Groups with Concordant & Discordant Results between Tropism Assays...........................................................147Figure 6.9: Median Decline in Plasma Viral Load from Baseline in Maraviroc Once-Daily Recipients...................................................................................................................................................................149xvList of Abbreviations3TC — LamivudineA — AdenosineAIDS — Acquired Immune Defciency SyndromeAPOBEC3G — apolipoprotein B mRNA-editing enzyme-catalytic polypeptide-like-3GARV — AntiretroviralATP — Adenosine TriphosphateAZT — ZidovudineBID — Twice dailyC — CytidineCA — Capsid proteinCCR5 — C-C motif Chemokine Receptor 5CD4 — Cluster of Differentiation 4CTL — Cytotoxic T LymphocyteCRF — Circulating Recombinant FormCXCR4 — C-X-C motif Chemokine Receptor 4D/M — Dual/Mixed tropicDNA — Deoxyribonucleic AciddNTP — Deoxyribonucleoside triphosphatesEFV — EfavirenzemPCR — Emulsion Polymerase Chain ReactionEnv — EnvelopeESCRT — Endosomal Sorting Complex Required for TransportESTA — Enhanced Sensitivity Trofle AssayFDA — U.S. Food and Drug AdministrationFPR — False-Positive RateG — GuanidineGag — Group-specifc antigenxviGALT — Gut-Associated Lymphoid Tissuegp41 — Glycoprotein 41 (kilodaltons)gp120 — Glycoprotein 120 (kilodaltons)gp160 — Glycoprotein 160 (kilodaltons)GS-FLX — Genome Sequencer FLXHAART — Highly Active Antiretroviral TherapyHIV — Human Immunodefciency VirusHOMER — HAART Observational Medical Evaluation and Research cohortIN — Integrase proteinIQR — Interquartile RangeLCB — Lower Confdence BoundLTR — Long Terminal RepeatMA — Matrix proteinMERIT — Maraviroc versus Efavirenz Regimens as Initial TherapyMIP-1α/MIP-1β — Macrophage Inflammatory Protein 1 α or βMOTIVATE — Maraviroc versus Optimized Therapy in Viremic Antiretroviral Treatment-Experienced PatientsmRNA — messenger Ribonucleic AcidMSM — Men who have Sex with MenMVC — MaravirocNC — Nucleocapsid proteinNef — Negative regulatory factorNRTI — Nucleoside Reverse Transcriptase InhibitorNNRTI — Non-Nucleoside Reverse Transcriptase Inhibitorn.s. — Not SignifcantOTA — Original Trofle AssayPBMC — Peripheral Blood Mononuclear CellPCR — Polymerase Chain ReactionxviiPDVF — Protocol-Defned Virologic FailurePEP — Post-Exposure ProphylaxisPI — Protease InhibitorPIC — Pre-Integration ComplexPol — PolymerasePR — Protease proteinPrEP — Pre-Exposure ProphylaxisPSSM — Position Specifc Scoring MatrixpVL — Plasma Viral LoadQD — Once dailyR5 — CCR5-usingRANTES — Regulated-upon-Activation, Normal T Expressed and SecretedRev — Regulator of expression of virion proteinsRLU — Relative Light UnitsRNA — Ribonucleic AcidRT — Reverse Transcriptase proteinRT–PCR — Reverse Transcriptase Polymerase Chain ReactionSDF-1 — Stroma-Derived Factor 1SU — Surface proteinT — ThymidineTat — Trans-activator of transcriptionTM — Transmembrane proteinV3 — Third Variable loop Vif — Virion infectivity factorVpr — Viral protein RVpu — Viral protein uniquewOBTss — Weighted Optimized Background Therapy Susceptibility ScoreX4 — CXCR4-usingxviiiAcknowledgmentsThere are many people to whom I am indebted for their help, input, conversations, collaborations,support, friendship, and generosity. First, I thank the patients who participated in all the studies I had the privilege of exploring. I amhumbled that thousands of people whom I have never met have made my research possible. I am also indebted to the many co-authors who contributed to the studies contained in this thesis andwho provided outstanding data, analyses, and insight. I  am  extremely  thankful  for  grants  and  fnancial  support  during  my  studies  from  the  BritishColumbia  Centre  for  Excellence  in  HIV/AIDS,  the  Canadian  Institutes  of  Health  Research,  theCanadian Observational Cohort, the National Sciences and Engineering Research Council of Canada,and the  University  of  British Columbia.  I  would also like  to  thank the industry colleagues  andpartners who provided access to samples and data, and who supported much of this research bothfnancially and intellectually.I am grateful to all the outstanding members, both past and present, of the British Columbia Centrefor Excellence in HIV/AIDS Laboratory, Liliana Barrios, Carolyn Beatty, Zabrina Brumme, DennisonChan,  Peter  Cheung,  Celia  Chui,  Paul  Crosson,  Weiyan Dong,  Winnie  Dong,  Erin  Gillan,  ChrisGlascock, Alejandro Gonzalez-Serna, Rob Hollebakken, Susie Johnstone, Jeffrey Joy, Aram Karakas,David Knapp, Chris  Lachowski,  Richard Liang,  Andrew Low, Eric  Martin,  Chelsea  McCullough,Theresa  Mo,  Andrew Moores,  Laurie  Nichol,  Carmond Ng,  Emma Preston,  Leslie  Rae,  MarjorieRobbins, Alissa Sadler, Jenna Spring, Emilie Stevens, Sharon Szeto, Lily Tam, Iris Tao, Jeremy Taylor,Art Poon, Birgit Watson, Conan Woods, Brian Wynhoven, and Xiaoyin Zhong. Special thanks to my fellow graduate students Chanson Brumme, Vikram Gill, Guinevere Lee, andRachel McGovern, who are brilliant, kind, and talented. I’ve truly enjoyed our time learning togetheras students. xixJoel Kaake for being a constant and loyal friend. Amadou Issacs for his many years of support andpartnership. Michael Moorhead, who helped me to get the summer job that would turn into a thesis. My committee  — Janet  Chantler,  Hélène Côté,  and Julio  Montaner  — who gave  me invaluableguidance and encouragement throughout my studies. I  feel  extremely  lucky  to  have  been  mentored  by  my  supervisor,  Richard  Harrigan,  who  bothchallenged me and rooted for my success. Thank you, Richard.Finally, I want to thank my beautiful family. Mom, Dad, Emily, Daniel, and James: thank you for yourincredible support, steadfast belief, and enduring love.xxForewordThere was a point in time, not so very long ago by some estimates, when a miniscule virus brushedagainst a fnger-like protein; twisting ever so slightly, it tethered a second protein, and slipped itsway into the cell of a new species.Since that frst point, this viral entry has been endlessly and immeasurably repeated, in millions ofpeople throughout the world.  The way in which HIV enters cells  has been a focus of basic andclinical research for decades. The study of that second protein it tethers — the HIV coreceptor — hasfar-reaching implications, from immune decline, to drug therapy and even hope for a cure.By better understanding this early step in the life cycle of HIV, science and medicine have developedways  of  slowing  it  down  and  even  preventing  the  entry  from  occurring  at  all.  Cutting-edgeapproaches have been applied to this line of research in the hopes that with each advance, and otherslike it, we come closer to the end of HIV.xxiChapter 1:     General Introduction & Thesis Objectives1.1     Background1.1.1     The Origins & Current State of the HIV EpidemicThere were an estimated 35.3 million people living with Human Immunodefciency Virus (HIV) in2013 1. Each year, 2.3 million people are estimated to become newly infected with HIV, and each year1.6 million people die from Acquired Immune Defciency Syndrome (AIDS) 1. The frst report of AIDSwas published in 1981 2, and its causative agent, HIV, was subsequently characterized in 1983 3. Sincethat time, there have been a myriad of studies characterizing almost every aspect of the virus, from itsstructure and biology to its epidemiology, treatment, and prevention. There are two major types of HIV. HIV type 1 (HIV-1) comprises the bulk of the epidemic. HIV type 2(HIV-2) is mostly isolated to Western Africa, with an estimated prevalence of 1 to 2 million people 4,5.Both HIV types belong to the Lentivirus subfamily of the family Retroviridae. These are viruses withRNA genomes  that  replicate  through  a  DNA intermediate  6.  HIV-1  can  be  further  classifed  byphylogeny into three groups: M (main), O (outlier), or N (non-M/non-O) 7. Group M can be furthersubdivided into subtypes or clades. These are lettered A through K (with some exceptions), or arelabelled with more than one letter to indicate circulating recombinant forms 7,8. Globally, subtype C isthe most prevalent HIV-1 subtype, at close to half of all infections 9. Subtypes A and B are the nextmost prevalent infections, at approximately 12% each. Subtype A is most common in Eastern Europe,Central Asia, and East Africa, while subtype B is most prevalent in Western Europe, the Americas,and Australia. Two circulating recombinant forms (CRF) exist at global prevalence between 5% and8%. These are CRF01_AE, which is most common in South-East Asia, and CRF02_AG, which is mostcommon in West Africa  9. Central Africa — where the HIV epidemic is likely to have frst crossedover into humans 10–12 — has the greatest diversity of subtypes 9. 1The zoonotic origins of HIV-1 and HIV-2 have been mostly elucidated 13. Both epidemics arose fromfrom transmission of Simian Immunodefciency Virus (SIV) infections of non-human primates 11,12,14.HIV-1 is likely to have crossed over into humans in at least two cross-species transmission events ofSIVcpz, the virus that infects some chimpanzees (Pan troglodytes troglodytes), and likely gave rise toHIV-1 groups M, and N 12. HIV-1 group O is related to SIVgor and may have arisen from cross-speciesinfection from gorillas (Gorilla gorilla gorilla) or chimpanzees  15,16, and a recent potential new HIV-1group that is distinct from the other three also appears to have crossed over from gorillas 17. The HIV-2 epidemic  appears  to  have been derived from SIVsm,  which infects sooty mangabeys (Cerocebustorquatus atys) 14. As is customary in the HIV feld, hereafter, references to HIV can be assumed to bespecifcally to HIV-1 unless otherwise indicated. 1.1.2     The Structure & Replication Cycle of HIV-1Virions of HIV-1 have a diameter of approximately 110 nm, with an internal protein capsid envelopedby a lipid bilayer. Inside the capsid are most of the proteins required for the early stages of infection,as  well  as  two copies  of  an  approximately  9  kilobase-long single-stranded  RNA genome.  Thesegenome encodes a total of 15 viral proteins 18. There are nine open reading frames: gag, pol, vif, vpr,vpu, tat, rev, env, and nef. The Gag, Pol, and Env polyproteins are each cleaved into their smaller activeconstituents. The gag gene encodes for the main viral structural components: the MA (matrix) protein,CA (capsid),  NC  (nucleocapsid),  and  p6.  The  env  gene  encodes  the  viral  proteins  of  the  outermembrane envelope: SU (surface), and TM (transmembrane). The primary viral enzymatic functionsare performed by PR (protease), RT (reverse transcriptase), and IN (integrase). These are encoded bythe pol gene. Additionally, there are six accessory proteins: Vif, Vpr, Nef, Tat, Rev, and Vpu, whichhave  various specialized roles  in viral  replication  19.  The functions of each these aforementionedproteins  will  be  addressed  in  turn  as  the  replication  cycle  of  HIV  is  described  henceforth.  Anillustration of the replication cycle can be found in Figure 1.1. Steps targeted by antiretroviral agentsare also denoted in the illustration. Treatment for HIV will be discussed in more detail in Section 1.3.2Figure 1.1: HIV-1 Replication Cycle & Drug TargetsPLoS Pathogens 9(3): e1003241. Figure 1.1: HIV-1 Replication Cycle & Drug Targets Depicted is the replication cycle of HIV-1 from cellularentry to progeny release and maturation. Specifc antiretroviral classes are indicated adjacent to the step in the life cycle atwhich they target. This fgure is adapted from Wong et al, PLoS Pathogens, with modifcation courtesy of Hélène Côté. Thisfgure is used under the terms of the Creative Commons Attribution License. © 2013 Wong et al 20. Wong RW, Balachandran A,Ostrowski MA, Cochrane A (2013) Digoxin Suppresses HIV-1 Replication by Altering Viral RNA Processing. PLoS Pathogens9(3): e1003241. 3Briefly,  viral entry into a target cell  begins when the viral  SU glycoprotein (gp120),  binds to thecellular surface receptor CD4, initiating a series of conformational changes in gp120 21,22. This exposescertain sites in the glycoprotein — a primary site being the variable V3 loop. Exposure of V3 leads tobinding of a chemokine receptor, generally either CCR5 or CXCR4 22–26. Host restriction factors suchas beta-defensins or RANTES may provide some protection against binding  27,28, but if coreceptor-binding occurs, the entry step can proceed. HIV entry will be described in further detail later in thischapter since this is the key step related to HIV tropism. After conformational changes in gp120 associated with coreceptor binding occur, the hydrophobicfusion peptide of gp41 is exposed and inserts into the cellular membrane 29. In addition to the fusionpeptide of gp41, there is also the N-terminus-linked heptad repeat region (repeating patterns of sevenamino acids), the C-terminus-linked heptad repeat and the transmembrane peptide. In trimeric gp41,the N heptads form a coiled-coil structure onto which the C heptads then bind, thus collapsing theoverall structure of gp41 into a six-helix hairpin structure  29.  This conformational  change in gp41decreases the effective distance between its fusion and transmembrane peptides.  In this way, thecellular membrane (within which the fusion peptide has been inserted),  and the viral membrane(within with the transmembrane peptide exists) are brought into close proximity. This allows the viralmembrane to fuse with the target cell membrane 22,30,31. Fusion of the two membranes releases the viral contents into the cytoplasm. The most importantcontent is the conical viral capsid which comprises multimers of the CA protein 32. The capsid encasesthe viral genome and its proteins. A process called uncoating then occurs whereby the capsid shellbegins to disassociate, releasing its contents. At this stage, the genome is released in the form of a pre-integration complex (PIC). The PIC comprises the RNA genome, closely associated with several viralproteins 33. Its release triggers the next step in the life cycle: reverse transcription 22,34. In order establish a productive infection of a host cell, HIV must convert its RNA-based genome intoDNA. This is achieved by the enzyme reverse transcriptase (RT). RT contains two active sites: an N-4terminal RNA- and DNA-dependent DNA polymerase to convert the RNA genome into DNA, and aC-terminal  RNase H to digest  the RNA strand during complementary DNA strand synthesis  22,35.First,  a short stretch of single-stranded DNA is produced at the 5’ end of the viral RNA, and thecomplementary RNA is degraded 36. Next, there is a “template switch” whereby this short stretch ofDNA binds at the 3’ end of the viral RNA. Single-stranded DNA is produced complementary to theRNA, and the RNA is degraded by the RNase H activity of reverse transcriptase. DNA-dependentDNA polymerase activity of reverse transcriptase is then used to produce double stranded DNA.During this process, long terminal repeats (LTR) of repeated sequences of DNA are added flankingthe viral genome, resulting in a DNA genome of approximately 10 kilobases in length 18. The activityof reverse transcriptase is highly error-prone. This is a signifcant contributor to the high degree ofvariation seen for HIV  37.  The reverse transcription step is  targeted by the host  restriction factorAPOBEC3G, which affects elongation of the reverse transcripts and introduces hypermutation in thegrowing viral DNA genome  38,39.  These activities are, in turn, antagonized by the viral protein Vifthrough induction of APOBEC3G degradation 40,41. Upon successful completion of reverse transcription, the pre-integration complex with its now DNAgenome is then imported into the cell nucleus, probably through binding of cellular nuclear importfactors or nuclear import signals on the Vpr, IN, CA and/or MA components of the pre-integrationcomplex 33,42,43. The viral Integrase protein exposes short stretches of DNA at the 3’ ends of the viralgenome LTRs. The exposed hydroxyl groups are then used to cut the host cell’s chromosomal DNAand join the viral genome to the cellular DNA, with host cell enzymes completing the process  22,44.Interestingly, the host cell factor, nuclear pore protein (Nup) 358 is recruited by the capsid protein inthe pre-integration complex. Nup358 then actually targets the proviral DNA to regions of highertranscriptional  activity, thus ensuring that  the virus will  be actively replicated  33.  The integrationprocess thus results in a stable provirus within the cell’s DNA. In some cases the cell may go into aresting state, resulting in a latently infected cell which can carry this provirus over a long period oftime 45. Alternatively, the replication process can continue with the manufacture of progeny virionsfrom the integrated provirus.5Progeny virions are produced by transcription of the DNA provirus. A promoter within the LTRinitiates the cell’s own transcription process to create plus-strand RNAs from the integrated HIVgenome.  These  plus-strand  RNAs  can  act  either  as  mRNAs  for  viral  peptide  synthesis  or  beencapsidated into progeny virions. The viral Tat protein is required for effcient elongation of theviral plus-strand RNA through recruitment of the cell’s positive transcription elongation factor (P-TEF) b. Smaller transcripts can directly enter the cytoplasm from the nucleus. Longer transcripts arebound by Rev, which recruits the host nuclear export factor CRM1 to facilitate the entry of thesetranscripts into the cytoplasm 22. The plus-strand RNAs produced are used as templates for makingviral proteins, and full-length transcripts can be assembled into viral progeny. In the early phase ofviral RNA production, the HIV RNA transcripts are spliced at multiple sites by host cell spliceosomes19. Translation of the viral mRNA is performed by the cell’s ribosomes 46. As the concentration of theviral Rev protein increases, it inhibits host cell splicing of viral RNA transcripts, resulting in singly-spliced and unspliced RNA transcripts which are used for translation of larger polyproteins, as wellas for the genomic RNA 19. Translated viral accessory proteins Nef, Vif, Vpu, and Vpr all contribute tomodifying the cellular environment to ensure effcient viral replication, persistence, and release. Thisis achieved through such varied functions as downregulation of CD4 receptors, modulation of thecell cycle, antagonism of host restriction factors, and enabling of immune evasion 47. The HIV envelope polypeptide, called gp160, is cleaved into gp120 and gp41 by the host protease,furin prior to being transported to the cell surface  48,49. Translated Gag proteins accumulate on theinner surface of the cellular membrane 46. A subset of Gag units exists as Gag-Pol polyproteins due toan infrequently occurring ribosomal  frame shift  50.  The other  viral  proteins,  as  well  as  the RNAgenomes, all co-associate with different portions of the Gag polyprotein to form the components of acomplete assembled virion  51,52. These Gag units also bind a class of cellular proteins called ESCRTcomplexes. These proteins are usually involved in the formation of vesicles, but are co-opted by thevirus to induce budding of the viral progeny from the cell 22,53. Budding is antagonized, however, bythe  host  restriction  factor  tetherin,  which  harnesses  the  budding  virions  to  the  cell  membrane,preventing their release 22,41. This, in turn, is inhibited by HIV’s Vpu protein through sequestration oftetherin to an intracellular site distal from the cell surface 41,54.  6A lipid membrane derived from the former host cell now surrounds virions that have successfullybudded, and these virions then begin a process of maturation. The viral Protease cleaves the Gag andGag-Pol precursors into their active structural and enzymatic components  22,55,56.  The MA proteinsassemble as a matrix on the inner surface of the viral lipid bilayer, while the CA proteins join to formthe cone-shaped capsid surrounding the viral  proteins  and RNA genomes  57.  Once maturation iscomplete, the virion is infectious and able to initiate a new round of replication in another target cell. 1.1.3     HIV Target CellsHIV acquisition generally occurs via sexual, perinatal, oral, or intravenous transmission of the virusfrom an infected individual to an uninfected individual 58,59. Viral entry is dependent on the presenceof CD4 as well as a coreceptor protein — typically CCR5 or CXCR4. Thus, generally only cells withthese proteins on their surfaces can be infected by HIV. The cell type frst encountered by the virusdepends primarily on the route of transmission. In general, there are three main cell types which HIVinfects: CD4+ T-cells, macrophages, and dendritic cells 60. Cells capable of being infected by HIV existin  the  genital  epithelium,  gastrointestinal  tract,  placenta,  central  nervous  system,  and blood andlymphoid circulation. All of these are sites of exposure and transmission of the virus, and representcompartments of  viral reservoirs  58,61.  Dendritic  cells  are not productively infected by HIV in theclassical sense, but rather engulf virions and can then transmit them to uninfected cells which comein contact with them. Accordingly, HIV can exploit the normal behaviour of dendritic cells, wherebytheir function as antigen-presenting cells is hijacked and used to expose HIV target cells to the virus58.CD4+ T-cells are the primary target of HIV. These are lymphocytes which coordinate the immuneresponses to infecting microbes and antigens through signalling to both the cellular and humoralimmune systems  62.  CD4+ T-cell  counts decline over the course of HIV infection, and decreases inCD4+ T  cell  counts  are  associated  with  increases  in  risk  for  opportunistic  infections,  AIDS,  andmortality 63–67. A CD4+ cell count below 200 cells/mm3 is associated with AIDS 68. The mechanism by7which HIV causes depletion of CD4+ T cells is not fully understood. However there are a number ofproposed mechanisms involving both direct and indirect pathways. These include cytopathic deathof infected cells primarily via pyroptosis,  immune-mediated killing of infected cells,  apoptosis ofuninfected cells  from contact  with viral  proteins,  formation of syncytia  between infected  and/oruninfected  cells,  impaired  production  of  new T  cells  or  their  progenitors,  and  chronic  immuneactivation  60,69–72.  Apoptosis  occurs  in  productively  infected  cells,  whereas  pyroptosis  occurs  inabortively  infected  non-permissive  cells.  The  former  is  mediated  by  caspase-3  and  is  minimallyimmune activating, whereas the latter is mediated by caspase-1 and is extremely inflammatory. Thus,pyroptosis itself may be a driver of further immune activation, creating a cycle of further cell deathand inflammation 69,70.HIV also has a tendency to affect the very cells which respond to it. Memory CD4+ T cells that arespecifc for HIV and whose function may better coordinate the immune response to the virus areactually preferentially infected by HIV compared to memory T cells without HIV specifcity 73. HIV-specifc CD4+ T cells  become activated upon encountering viral antigens, remaining in prolongedclose contact with HIV containing cells and free virus in the lymph nodes. Furthermore, detection ofHIV antigen results in release of inflammatory chemokines which recruit HIV-specifc cells to the siteof infection. Thus, it has been hypothesized that this recruitment and extended proximity may allowthe observed preferential  infection  of  HIV-specifc T  cells  73.  In  addition to  CD4+ T  cells,  HIV  iscapable of  infecting two types of  antigen-presenting cells:  macrophages and dendritic  cells  58,60,74.Paradoxically, their immune function of antigen presentation may increase viral spread through theirinteractions with and signalling to T cells 60,75. A major challenge of HIV infection is that the virus persists long-term in infected individuals. Cellswhich have  been infected can  enter  a  resting state after  viral  integration,  such that  they remainlatently infected for years and can be subsequently reactivated to initiate new rounds of infection.Long-lived infected cells such as macrophages or resting CD4+ T cells thus act as viral reservoirs anda barriers to curing HIV infection 60,76–79. There are also several anatomical reservoirs in which HIV canbe found  61. The lymphoid organs, namely the spleen, lymph nodes, and gut-associated lymphoid8tissue  (GALT),  and  their  associated  lymphocytes  have  been  the  focus  of  most  research  efforts.Lymphocytes present in systemic circulation are also studied since they are more straightforward toobtain  from donors.  However  HIV replication  also  occurs  in  the  central  nervous  system withinmacrophages and microglia, as well as in T cells and macrophages in the genital and gastrointestinaltracts 61. Several efforts have been made both to study the latent reservoir as well as to activate it ordecrease its size in attempts to purge latently infected cells and establish a functional cure 45,61,78,80,81.1.1.4     Natural History & Pathogenesis of HIV-1 Infection in HumansHIV exposure has a fairly low probability of resulting in productive transmission,  and is  largelydependent on the viral load of the donor 82–85. HIV becomes established in the host during a phaseknown as acute infection. Following the transmission of HIV, there is an approximately 2 to 3 weekincubation period during which the virus establishes itself in the lymphatic tissue 58,82. Viral RNA canbe detected in the blood plasma approximately 10 days after initial infection, and as the number ofinfected cells increase, so too does the viral load, reaching a peak viraemia after approximately 3 to 4weeks 86. Concomitant with this increase in viral load is a steep depletion in CD4+ T cells due to HIVand CTL-mediated killing. Often the appearance of non-specifc constitutional  symptoms such asfever  or rash which are  similar to  symptoms in more common viral  illnesses  87,88.  The virus alsobecomes disseminated to various body tissues during this acute infection period. The immune response during the acute phase is mostly dominated by the innate immune system,especially cytokine, dendritic cell and natural killer cell responses 86,89,90. At the time of peak viraemia,infected individuals have viral loads often on the order of millions or hundreds of millions of HIVcopies  per  millilitre  36.  Thereafter,  the  individuals  pass  through several  stages  (known as  Fiebigstages) as progressively more viral infection markers become detectable by diagnostic assays 91. Thelater Fiebig stages are marked by the presence of HIV-specifc antibodies produced by B cells  86,92.However,  these  antibodies  tend  to  be  non-neutralizing,  and  along  with  the  extensive  B  celldysregulation  associated  with  HIV  infection,  the  humoral  immune  system  tends  to  be  poor  atcontaining infection 92. 9Following peak viraemia, there is a gradual decrease in viral load until it reaches a relatively stableviral  set  point  averaging  approximately  105 copies  per  millilitre.  This  decrease  is  temporallyassociated  with  elevated  CD8+ cytotoxic  T  lymphocyte  (CTL)  activity  86,93.  CTLs  are  capable  ofdetecting HIV-infected CD4+ cells through interactions with Human Leukocyte Antigen (HLA) class Imolecules displaying viral peptides. The most common viral peptides targeted by HLA are derivedfrom the HIV Nef, Gag, and Pol proteins  94.  Once detected, CTLs can lyse the infected cells  93,95,96.Certain HLA alleles are associated with lower viral set points and slower HIV disease progression,while other alleles are associated with higher set points and more rapid progression 97. This seems tobe driven largely by differential responses to viral peptide epitopes. Viral epitopes are themselvesbound and displayed to CTLs by different HLA molecules. There seems to be substantive immunecontrol of HIV by this component of the cellular immune system. This is evidenced by the decline inviral load and influence on disease progression that are associated with CTL responses. However,HIV is able to counteract and evade these responses through mutations in the HLA-targeted epitopes(termed escape mutations), and through downregulation of HLA molecules on the surface of infectedcells  97,98.  The  general  inadequacy  of  the  immune  system  in  controlling  HIV  infection  allowsexpansion  of  the  viral  population  and  depletion  of  CD4+ cells.  Eventually,  this  leads  to  themanifestation of HIV-related disease and AIDS.  1.2     HIV-1 Coreceptor Usage1.2.1     Chemokine ReceptorsThe human biological functions of the chemokine receptors CCR5 and CXCR4 are to serve as asnormal cell signalling proteins activated by the presence of their respective chemokines. Chemokinesare small proteins of 70 to 90 amino acids in length, and mediate leukocyte migration and activationat inflammation sites  99.  The natural  ligands for  CCR5  are macrophage inflammatory protein 1  α(MIP-1α), MIP-1β, and RANTES (regulated-upon-activation, normal T expressed and secreted). Theseare released by CD8+ T lymphocytes. The natural ligand for CXCR4 is stroma-derived factor 1 (SDF-1)1021,100,101.  Also  known  as  CXCL12  102,  it  is  released  by  fbroblasts  in  several  tissues  and  organsthroughout  the  body  103.  Both  chemokine  receptors  are  seven transmembrane  G-protein  coupledreceptors. This class of proteins acts via intracellular G proteins, which activate several pathways toaffect leukocyte chemotaxis and activation at sites of inflammation. For instance, binding of SDF-1 toCXCR4  activates  a  diverse  set  of  signalling  pathways  involved  in  chemotaxis,  cell  adhesion,transcriptional activation, and cell survival  104. Interestingly, it has been demonstrated  in vitro thatHIV gp120 can bind to these chemokine receptors and cause similar signal transduction cascades.Furthermore, co-administration of the chemokines with HIV has been shown to inhibit HIV infectionin cells 21,104.There is  differential  expression of  these chemokine receptors  depending on cell  type.  In general,CCR5 is primarily expressed on macrophages, dendritic cells, and activated or memory CD4+ T-cells,while CXCR4 is primarily expressed on naïve CD4+ T cells, B lymphocytes and several types of stemcells 101,103,105. Interestingly, the ligands for CCR5 can also act as ligands to other chemokine receptors(e.g., CCR1, CCR4). Therefore release of the chemokines can activate several receptors on a variety ofdifferent cell types, resulting in an array of activities. In contrast, SDF-1 binds only to CXCR4, andthus has more specifc effects solely on CXCR4-expressing cells 21,106. There is  a  genetic  polymorphism for CCR5 which results  in a non-functional  and non-expressedversion of the protein (this  will  be discussed in more detail  elsewhere in the thesis).  Individualshomozygous for this allele appear to have a relatively normal phenotype overall  100. However, theyare over-represented in cohorts of patients with symptomatic West Nile virus infection, and are morelikely to die from that infection than patients without the allele, so the defect is not completely benign107. In contrast, the CXCR4 protein appears to be essential for ontogenetic development, and it is lethalto ‘knockout’ the  CXCR4 gene in mice  101. Small-molecule antagonists of both chemokine receptorshave been developed, primarily for their anti-HIV activity 108–111, but also for anti-cancer proceduresinvolving CXCR4 and mobilization of hematopoietic stem cells 102,109,112,113. 111.2.2     Interactions of the Viral Surface Glycoproteins with Cellular Surface Receptors & CoreceptorsThe protein HIV uses to attach to human cells is called gp120. Also referred to as SU or surfaceprotein, gp120 is approximately 510 amino acids in length. It is frst synthesized at the endoplasmicreticulum as  part  of  the  polyprotein  gp160,  which  includes  both  gp120  and the  transmembraneprotein, gp41. Within the Golgi, the gp160 precursor polypeptide oligomerizes and is glycosylatedwith  N-linked  oligosaccharides  114.  Glycosylation  is  essential  for  the  correct  conformation  andstability of the fnal envelope trimer 114. The glycoprotein is transported through the Golgi, where itsoligosaccharides are trimmed and modifed. Gp160 is then cleaved by the cellular endoprotease furininto  its  constituents:  gp120 and gp41  49,114.  Despite  cleavage,  these  two glycoproteins  remain  co-associated through weak non-covalent  interactions.  Together,  they assemble in trimers  at  the cellmembrane, with gp41 components embedded within the phospholipid bilayer, and trimeric gp120protruding away from the cell. During budding, they become incorporated into the viral membrane57,114.  The  gp120  glycoprotein  itself  can  be  divided  into  fve  conserved  regions  (C1-C5),  which  haverelatively consistent sequences across many HIV isolates. Interspersed between the constant regionsare fve variable regions (V1-V5), which can vary widely in both their amino acid sequences andlengths. Gp120 can be roughly divided into three areas: the inner domain, the outer domain, and thebridging sheet linking the two domains. The inner domain is formed by the conserved regions C1and C5, which are the main contact areas for gp41, while C2, C3 and C4 form a hydrophobic corewithin gp120  114. The outer domain contains higher glycan concentrations, which makes the outerdomain  less  immunogenic,  thus  protecting  the  virus  from neutralizing  antibodies.  The  variableregions, especially V1/V2 and V3 are exposed on the trimer and provide partial shielding of the CD4binding cavity within gp120 114,115. A very important region of gp120 is the CD4-binding site, which isa cavity located at the interface between the inner and outer domains and the bridging sheet. Withinthe CD4-binding site there is a crucial hydrophobic cavity which interacts with phenylalanine-43 ofCD4 during cellular attachment 24,114–116. Disruption of this interaction limits HIV entry 117,118. 12In its native state, the envelope trimer exists with three of its variable regions V1/V2, and V3 at theapex of the structure 119,120. These regions are present as large loops due to intramolecular disulphidebonds 115. For example, the V3 loop is formed by disulphide bonding between its two cysteine residesat codon 1 and codon 35 of V3  121. Binding of the CD4 receptor to the trimer results in signifcantconformational  changes wherein each  gp120 monomer  rotates  outward  119.  The V1/V2 loops  arerotated away from the apex of the spike and the V3 loop extends directly towards the target cell119,120,122. Thus, rearrangement of the trimer upon CD4 binding exposes additional sites important forviral entry (Figures 1.2 & 1.3). The fact that these epitopes are only exposed after CD4 binding meansthey are partially shielded from possible neutralizing antibodies 119. The V3 loop is approximately 35 amino acids in length, and is the primary determinant of HIV-1coreceptor tropism  123,124.  There are two functional domains of V3: the crown (an antiparallel betasheet formed by codons 11 through 25), and the stem (formed by the amino acids surrounding thecrown).  It  has  been  demonstrated  that  V3  interacts  directly  with  CCR5  and  CXCR4  125–128,  andvariation in HIV env and V3 is a signifcant contributor to viral ftness 129–131. The V3 stem likely bindsto the N-terminus of the coreceptor, while the crown (the most important contributor to coreceptortropism) interacts with the extracellular loops of the coreceptor 126. Sequence analysis of HIV-1 V3 loops reveals that those derived from X4 viruses tend to have anoverall more positive charge than those derived from R5 viruses, and such charges are especiallycommon at codons 11, 24, and 25 132,133. This observation fts well with the putative V3 binding pocketson the HIV-1 coreceptors, since CXCR4 has a high density of negative charge at its V3 binding pocket,while CCR5 does not 134,135. Thus, the direct evidence of the interaction between coreceptors and V3,as  well  as  the association between coreceptor  usage and V3 sequence variation,  have led to thedevelopment of a number of tropism assays.  Whether phenotype-based or genotype-based, theseassays almost always include the V3 region.13Figure 1.2: Schematic of Native & Bound Structures of HIV-1 gp120Figure 1.2: Schematic of Native & Bound Structures of HIV-1 gp120. (A) Depicts the native structure of trimeric HIV-1 gp120 prior to binding CD4 or the viral coreceptor.(B) Depicts the conformational changes in gp120 which occur following CD4 binding. The V3 regions are shown in red, and become exposed during the conformational change. Thebinding of V3 to the CCR5 coreceptor is also depicted. The HIV-1 gp120 trimer is shown in yellow, and the gp41 protein is shown in blue. The CD4 binding site of gp120 (CD4bs) isshown in pink. This fgure is used under the terms of the Creative Commons Attribution License. © 2006 Zanetti et al 136. Zanetti G, Briggs JAG, Grünewald K, Sattentau QJ, Fuller SD(2006) Cryo-Electron Tomographic Structure of an Immunodefciency Virus Envelope Complex In Situ. PLoS Pathog 2(8): e83.14Figure 1.3: Density Maps of Native & Bound Structures of HIV-1 gp120     ABFigure 1.3: Density Maps of Native & Bound Structures of HIV-1 gp120 (A) Depicts  a density  map of  the native structure  of gp120 prior  to  CD4 or  coreceptorbinding. (B) Depicts a density map of HIV-1 gp120 when experimentally bound to constituents mimicking the receptor and coreceptor. These fgures are used under the terms of theCreative Commons Attribution License. © 2010 White et al  120. Adapted from White TA, Bartesaghi A, Borgnia MJ, Meyerson JR, de la Cruz MJV, Bess JW, Nandwani R, Hoxie JA,Lifson JD, Milne JLS, Subramaniam S. (2010) Molecular Architectures of Trimeric SIV and HIV-1 Envelope Glycoproteins on Intact Viruses: Strain-Dependent Variation in QuaternaryStructure. PLoS Pathog 6(12): e1001249.151.2.3     History of Cellular HIV-1 TropismEarly in the study of the epidemic, the actual mechanism for cellular entry by HIV-1 was unknown 100.In the same year as AIDS was frst identifed, it was discovered that CD4+ T-cells were depleted inpatients with AIDS-associated opportunistic infections 137. Soon, HIV was found to have tropism forCD4+ T-cells  138, though the presence of CD4 was found to be necessary but not suffcient for mostcases  of  HIV  entry  100.  Furthermore,  it  was  discovered  that  certain  strains  of  HIV  were  fasterreplicating and able to cause syncytia  induction (SI  isolates)  of  cells  in  vitro,  while  other  slowerreplicating isolates  were  not  able  to  induce  syncytia  (NSI  isolates)  139.  These  empirically  derivedphenotypes were  correlated  with more  effcient  replication in  either  T-cell  lines  or  macrophages,respectively, such that they were also referred to as T-tropic or M-tropic isolates. Furthermore, it was discovered that in vivo, progression to AIDS was faster for patients with SI versusNSI isolates  139. It was also noted that M-tropic infections tended to be more common in early HIVinfection, with the prevalence of T-tropic isolates increasing with the duration of infection  139. Theobservation of differential cellular tropism suggested unidentifed factors present in these cell typeswhich  rendered  them capable  of  productive  infection  by  one  type  of  strain  but  not  the  other.Eventually, it was determined that these factors were the human proteins, CXCR4 or CCR5 26,140–142.The natural functions of these viral  coreceptors are as receptors for certain chemokines,  and HIVentry has been demonstrated to be inhibited in the presence of these chemokines 100.Nomenclature for HIV tropism has a number of terms which overlap and have similar meanings butare used in different contexts. Broadly, T-tropic strains are also SI and use CXCR4, while M-tropicstrains are NSI and use CCR5 106. Dual-tropic strains of HIV are capable of using either coreceptor. Forconvenience, viruses that use CXCR4 for entry are called X4, and those that use CCR5 are called R5100. Occasionally, dual-tropic HIV is signifed by R5X4. However, often R5X4 is grouped in with X4viruses, or both are collectively called non-R5. Furthermore,  an eclectic  gallimaufry of alternative16proteins has been demonstrated to function as HIV coreceptors 143. Many of these are also chemokinereceptors though their relevance in vivo appears to be small with respect to HIV infection 144. 1.2.4     Clinical Relevance of HIV-1 TropismThere are a number of effects that HIV-1 coreceptor usage has on viral pathogenesis  and clinicaloutcomes during infection  145.  Studies  of  primary infection with subtype B HIV-1 have generallyfound that CCR5-using viruses tend to be more prevalent during this early phase of disease  146,147.Furthermore, R5 HIV tends to be transmitted to the recipient even if non-R5 variants are present inthe donor 148. The reasons for this seemingly preferential transmission are unclear. Some studies haveproposed a biological bottleneck such as the presence of CCR5-expressing cells such as Langerhanscells at the sites of sexual transmission. This biological bottleneck may select for CCR5-using virusesduring  transmission  149,150.  Moreover,  there  may  be  additional  biological  factors  which  allow forselective  expansion  of  R5 HIV  variants  following transmission  145.  However,  presence  of  non-R5variants in recent infections has been documented in a minority of patients 151,152. Alternatively, someauthors have proposed that transmission is a stochastic event and therefore that transmission of R5HIV is only more frequent because more donors tend to have R5 HIV in the frst place 153. As previously mentioned, human variation in the gene for CCR5 can also influence HIV infection106,154,155. A 32 base pair deletion within CCR5, denoted CCR5 Δ32, is relatively common (approximately10% prevalence) in individuals of European descent  154,156,157.  Homozygosity for  CCR5 Δ32  is rarer(approximately 1% prevalence) but is associated with almost complete resistance to infection by HIV156,157.  However, this mutation is  not protective against  HIV capable of  using CXCR4. Infection ofhomozygotes with X4 HIV has been documented but only rarely 158,159. Individuals heterozygous forCCR5 Δ32 do not seem to be signifcantly protected against infection, but may have a slower diseasecourse,  though  a  higher  risk  for  carrying  X4  variants  154,155,160.  Finally,  the  observation  thathomozygotes are mostly protected against HIV infection lends further evidence that HIV exposurestend to be with CCR5-using variants.17Regardless of the reason for the predominance of CCR5-using HIV in early infection, CXCR4-usingvariants can evolve over the course of disease,  in a process known as  coreceptor switching,  andquasispecies using both or either coreceptor may coexist within an individual. Non-R5 variants likelyevolve from R5 variants over the course of infection  145,161. This process may be gradual or sudden,and the evolution from R5 to X4 may traverse a ftness valley which may be an obstacle to coreceptorswitching  162–164.  Thus, there are constraints on the evolution of HIV from CCR5 to CXCR4 usage,which one group has commented requires the virus “to make the right amino acid substitution in theright place at the right time” 131 – not a simple matter. HIV virions with different coreceptor usage can infect different target cells more effciently, such thatthere may be compartmentalization of HIV coreceptor usage depending on the cell type  105,145,165,166.The evolution of CXCR4-using HIV is associated with an accelerated decline in CD4+ T cells, andfaster disease progression 145,167,168. This is probably related to an ability to infect a larger proportion ofCD4+ T-cells,  and also  due  to  the  higher  cytotoxicity  of  CXCR4-using viruses  compared to  theirCCR5-using counterparts 105,169. The ability of HIV to use CXCR4 is also an independent factor whichincreases the risk of progression to AIDS by approximately seven-fold 139,167,170. Non-R5 HIV infectionhas also been linked to higher mortality in some studies, though not in others 139,155,171–173. The discovery and characterization of the HIV coreceptors, as well as their association with clinicaloutcomes, led to the development of a number of assays designed to assess the coreceptor usage ofHIV obtained from clinical isolates. Phenotypic assays were the frst types developed to determinetropism. Later, by linking the results of these phenotypes with the envelope genotypes of multiple HIVvariants, there was an emergence of genotypic tropism assays 174,175. This pattern echoes that seen inthe earlier stages of HIV drug resistance research. Assays were originally developed to determinewhether  phenotypic  resistance  was  present,  and  these  assay  results  were  later  correlated  withmutation profles which could be used to infer the resistance phenotype 176–184. 181.2.5     Phenotypic Assays to Detect HIV-1 Coreceptor UsageThe earliest assay for HIV tropism was based on detection of the syncytium-inducing (SI) phenotypethrough co-culture of patient derived peripheral blood mononuclear cells (PBMC) with MT-2 cells invitro 185. This was an indirect phenotypic assay in that it could not determine which coreceptor wasbeing  used.  Indeed,  the  HIV-1  coreceptors  had  not  even  been  discovered  at  the  time  of  thedevelopment of the MT-2 assay. However, it was later concluded that viruses with an SI phenotypetend to use CXCR4, while those with an NSI phenotype tend to use CCR5  186. A number of otherphenotypic assays which are more direct measures of coreceptor usage have since been developedand described  175,187–193.  Generally,  these  phenotypic  assays  tend to  use  patient-derived viruses  orrecombinant viruses which are then used to infect reporter cell lines that constitutively express CD4and either CCR5 or CXCR4. The most widely used phenotypic tropism assay has been the Trofle assay  194. This assay has beenmodifed to improve the detection of low-prevalence X4 variants in clinical  samples,  and is nowknown as the Enhanced Sensitivity Trofle Assay (ESTA) 195. In the assay, gp160 envelope sequencesare amplifed from patient plasma samples by using reverse transcriptase polymerase chain reaction(RT-PCR).  These  gp160  sequences  are  then  inserted  into  pseudotyped viral  vectors  containing  aluciferase reporter gene. These are then used to infect cell lines that express either  CCR5 or CXCR4.Luciferase  activity  is  measured  in  relative  light  units  (RLUs),  and  is  elevated  upon  successfulinfection of a cell line by the patient-derived pseudoviruses. A coreceptor antagonist drug is thenadded and infection is confrmed by a subsequent decrease in RLUs. High luciferase activity in theCCR5 cell line or CXCR4 cell line is indicative of R5 or X4 virus, respectively. Activity in both celllines  indicates  a  dual-tropic  or  mixed-tropic  viral  population,  which  is  designated  D/M  194.Disadvantages of phenotypic assays include their high labour intensity and long turn-around time,which ranges from approximately two weeks for Trofle to up to fve weeks for the MT-2 assay  191.Accordingly,  there  have  been  attempts  to  reproduce  phenotypic  assay  results  using  genotypicmethods.191.2.6     Sequencing-Based Genotypic Assays to Determine HIV-1 Tropism Using Bioinformatic InterpretationAt  their  core,  genotypic  tropism  assays  typically  involve  amplifcation  of  a  portion  of  the  HIVenvelope gene, followed by sequencing, and interpretation of whether that sequence is likely to haveX4  or  R5  behaviour.  The  sequence  input  has  traditionally  been  obtained  by  population-based(Sanger) sequencing. The sequence must then be interpreted manually by a technician. Interpretationinvolves identifcation of secondary peaks in the sequence chromatogram, which are indicative of aquasispecies mixture in the sample. Alternatively, there is automated software called RECall whichcan perform this sequence interpretation step with high accuracy and without manual intervention.Use  of  RECall  for  sequence  interpretation  increases  both  productivity  and  reliability  196.  Aninterpretation  system  or  bioinformatic  algorithm  can  then  be  used  to  infer  the  likely  sequencetropism. Early,  more  rudimentary  genotypic  assay  interpretation  systems  tended to  be  “rulesbased”,  andfocused on identifying basic, positively charged amino acids at certain codons in V3 and/or the netcharge of the amino acid sequence 133. In addition to codons 11, 24 and 25, many other codons withinthe V3 loop also have different amino acid compositions. Variation in V3 amino acid substitutionstends to cluster with the use of one coreceptor or another. As a result, a number of bioinformaticalgorithms have been developed which use this genotypic information to infer a phenotype  197–202.Arguably the two most commonly used algorithms have been the Position Specifc Scoring Matrix(PSSM) 203 and geno2pheno 197,204. These algorithms are trained on a set of samples with both knownphenotype results and envelope sequences. These phenotypes may be from a Trofle-like assay suchas for geno2pheno and PSSMX4/R5 or may be from the MT-2 assay as in PSSMSI/NSI or PSSMC 197,203–205. Asequence submitted for interpretation by the algorithms is then assessed for its similarity to typicallyX4 or typically R5 sequences. Bioinformatic algorithms can be assembled and trained in a number of ways, each with its  ownstrengths  and  weaknesses  197.  Support  vector  machines  (SVM),  artifcial  neural  networks,  linearregression models, position-specifc scoring matrices (PSSM), and rules-based algorithms have been20most commonly used for bioinformatic prediction of HIV tropism  197,204.   Together, these representstatistical learning methods which ft a model based on a set of training data. This trained model isthen applied to “unseen” datasets to predict the outcome of interest (in this case, HIV coreceptorusage). In essence, the output of the algorithms is a rating of how X4 or how R5 a sequence is likely tobe. While this value is typically a continuous variable, a binary category can be applied using a pre-specifed cutoff. For instance, geno2pheno gives an output known as a false-positive rate (FPR). Thelower this value, the higher the likelihood is that the sequence is X4. Establishing a false-positive ratecutoff  of 10, for example, would categorize any sequence with an FPR ≤10 as being X4, and anysequence >10 as being R5. Various cutoffs for these algorithms have been proposed, and no cutoff hasbeen frmly  established  or  widely  adopted  for  any algorithm.  Often  a  cutoff  is  established as  abalance between sensitivity for correctly identifying an X4 sequence,  and specifcity  for correctlyidentifying an R5 sequence. The two most common algorithms for bioinformatic prediction of HIV tropism are PSSMX4/R5 andgeno2pheno. These algorithms were developed using position-specifc scoring matrices, or supportvector  machines,  respectively.  Given that  these  are  also the  main  algorithms used  in the  studiescontained in this thesis, their design and performance will be reviewed here in additional detail withinformation obtained from Jensen et al (2003) and Sing et al (2007) 203,204. 1.2.7     The Development & Performance of the Bioinformatic Algorithms PSSMX4/R5 & Geno2phenoPSSM-based models are used to distinguish signifcant differences in the distribution of amino acidsbetween groups of sequences  which have been sorted by an empirically-obtained character  – forexample,  phenotypic  CCR5-capable  or  CXCR4-capable  groups.  The  training  set  for  PSSMX4/R5comprised  a  total  of  213  V3  sequences  obtained  from 177  HIV-infected  individuals.  Phenotypicresults for these sequences revealed 168 R5 sequences (79%), 17 X4 sequences (8%), and 28 dual-tropicsequences (13%) capable of using either coreceptor. The latter two sequence sets were combined intoa single non-R5 category. This training set was used to generate a 35×20 matrix: 35 being the length ofa typical  V3 loop,  and 20 being the number of amino acids.  Each of  the resulting 700 sites  was21assigned a likelihood ratio score. This likelihood ratio is given by the natural log of the ratio betweentwo frequencies: the frequency that a particular amino acid is seen at that position in the training setof non-R5 sequences divided by the frequency that the same amino acid is seen at that position in theR5 sequences. Thus, if a particular amino acid (for example an arginine at codon 11) is seen morefrequently  in  non-R5  sequences  than  R5  sequences,  then  the  likelihood  ratio  assigned  to  thatparticular site would be positive. Conversely, sites which are more common in R5 sequences receive anegative likelihood scores. Then, when a given V3 sequence is submitted to the matrix, each observedsite receives a likelihood ratio score, and these are summed up to give a total PSSM score, with higherscores representing sequences more likely to be X4 and lower scores representing those more likely tobe R5. In order to classify sequences into the binary tropism categories, a PSSM cutoff value must be chosen,above which  a sequence is  classifed as X4 and below which a  sequence is  classifed as  R5.  ForPSSMX4/R5, a bootstrapping procedure was used whereby multiple sub-samplings of the dataset wereused to generate new matrices (essentially “mini” PSSMs) that were then applied to the dataset as awhole. For each of these new, smaller matrices, an optimized cutoff was calculated which maximizedthe number of true positives (X4 variants higher than the cutoff), maximized the number of truenegatives (R5 variants lower than the cutoff), and minimized the number of false positives and falsenegatives (R5 or X4 variants incorrectly falling above or below the cutoff, respectively). Each sub-sampling matrix gave an optimized cutoff value, and the distribution of these values was plotted.The  5th and  95th percentiles  for  the  distribution  of  the  optimized  cutoffs  were  -6.96  and  -2.88,respectively. Thus, for two sequences, one scoring below -6.96 or one scoring above -2.88, there is avery high likelihood that each sequence would be R5 or non-R5, respectively. In other words, thesevalues represent cutoffs whereby a given score can be confdently used to predict tropism status of asequence, while scores falling in between these cutoffs have less certain tropism associations andwere classifed according to the 11/25 rule. PSSMX4/R5 was subsequently applied to two independentdatasets for the purposes of validation. The algorithm was determined to have 84-89% sensitivity tocorrectly identify non-R5 sequences, and 96-100% specifcity to correctly identify R5 sequences. Acaveat of this performance is that it was on clonal sequences. Actual clinical samples are invariably22more  heterogeneous,  containing  multiple  quasispecies  with  different  sequences  and  possiblydifferent  coreceptor  usages.  Therefore,  applications  to  clinical  samples  will  tend  to  have  lowerperformance.In addition to PSSMX4/R5, the geno2pheno algorithm is also used in this thesis to infer the coreceptorusage of V3 sequences.  Geno2pheno was developed using support vector machine methodology,which is a machine learning method that categorizes input examples (like sequences) based on alarge number of features. Features that are found to be signifcant classifers (e.g.,  an arginine atcodon 11 of V3) are assigned weights based on their ability to classify inputs. The training set for theSVM which was used to generate geno2pheno comprised 1110 samples from 332 patients. A total of769  samples  had R5  phenotypes  (69%),  210  had X4  phenotypes  (19%),  and 131  had dual-tropicphenotypes (12%). Similar to PSSMX4/R5, there are approximately 700 features assessed in the SVM,each with a weight based on its ability to predict tropism status. To illustrate, signifcant X4 featuresin order of decreasing weight were the presence of the residues 13Y, 11R and 20V. Signifcant R5features were the presence of residues 24G, 18S, and 13P. Overall, the algorithm had 76.4% sensitivity and 92.5% specifcity for inferring the coreceptor usageof  clonal  isolates.  However,  when  assessed  on 952 clinical  isolates  (containing HIV quasispeciesmixtures),  the  SVM-based  predictions  performed  worse  compared  to  the  clonal  data:  39.8%sensitivity  and  93.5% specifcity.  Interestingly,  however,  this  performance  could  be  improved  byincorporating  other  features  into  the  SVM model  in  addition  to  the  amino  acid  features.  Fouradditional “clinical” features which improved performance were: the log10 of the CD4 percentage,host  heterozygosity  for  the  CCR5  Δ32 allele,  number  of  ambiguous  amino acids  in  V3,  and thepresence  of insertions or  deletions in V3 (“indels”).  Lower CD4%, presence  of  CCR5  Δ32,  highernumbers of ambiguous amino acids, and higher numbers of indels were all associated with a higherlikelihood of having CXCR4-capable HIV. Including these parameters increased sensitivity on clinicalspecimens to greater than 60%. These additional features are biologically plausible markers of CXCR4tropism.  For  instance,  sequence  diversity  (i.e.,  increased  numbers  of  amino  acid  ambiguities  orindels) has been linked to X4 tropism and disease progression 206,207. Additionally, CXCR4 propensity23is  also  associated  with  lower  CD4+ T  cell  counts,  and  therefore  lower  CD4  percentages  145,169.Furthermore, CCR5 Δ32 heterozygosity decreases the availability of CCR5 coreceptors for HIV entry,and it is plausible that this may manifest as a viral environment with increased selective pressure tochange coreceptor usage. Indeed, such an association has been demonstrated in a separate study 155. 1.2.8     Alternative Genotypic Assays & Comparisons between Genotypic & Phenotypic AssaysAside from population-based sequencing of HIV-1 envelope followed by sequence interpretation,there are a handful of other genotypic assays used for determining coreceptor tropism. For example,heteroduplex tracking assays use labelled probes to  form duplexes with PCR-amplifed envelopesequences, and these are then analyzed by electrophoresis to determine if X4 probes were bound208,209. Allele-specifc PCR has also been used to selectively detect X4 or R5 sequences from patientsamples using real-time PCR 210. Additionally, there have been efforts to determine HIV-1 coreceptorusage from integrated or cell-associated HIV DNA in order to test patients with low or undetectableplasma viraemia 211–215. Although most genotypic tropism assays tend to use V3 sequences as their input, other enveloperegions outside of V3 are also important for HIV-1 tropism and may give additional insight. Aminoacid substitutions in the other variable and constant regions of HIV-1 gp120 have been correlatedwith coreceptor tropism, as have substitutions in the gp41 transmembrane protein 216–220. Furthermore,variation in peptide length and the number of N-linked glycosylation sites in the variable regions ofgp120 tend to increase over time during HIV-1 infection. These factors have also been found to beassociated with tropism 207. Additionally, different HIV-1 subtypes can have different prevalences ofX4  or  R5  isolates,  and  bioinformatic  algorithms  may  need  to  interpret  the  tropism  of  differentsubtypes  in  different  manners  8,205,221,222.  There  is  even  a  tropism assay  that  is  neither  classicallygenotypic nor phenotypic. Called the maraviroc clinical test, patient tropism is assessed by responseto short-term monotherapy with the CCR5 antagonist maraviroc 192,223–226. In this scheme, an R5 resultis given if by day eight of maraviroc monotherapy, the patient achieves a viral load <40 copies/mL or24at least a 1 log10 decrease in viral load from day one. The concordance of this approach with theTrofle assay has been measured as being 94% 224.Since  genotypic  assays  are  designed  to  mimic  phenotypic  assay  results,  they are  often  assessedagainst phenotypic assay results in order to determine performance. Genotypic assays often performvery well on clonal sequences (i.e., not a mixture of quasispecies), and have given sensitivities as highas 99% 227. However, clinical isolates derived from patients are composed of heterogeneous mixturesof diverse sequences, where X4 sequences may exist at low prevalence relative to R5 sequences  145.Population-based sequencing has a reported detection threshold of approximately 20% of the viralpopulation, meaning that quasispecies present at lower prevalence are not reliably detected 228. Thus,while  clonal  sequences  are  often  predicted  easily  by  bioinformatic  algorithms,  clinical  isolatescontaining minority X4 variants can often be misclassifed as R5 by genotyping but still have an X4biological phenotype 229. This can result in genotypic approaches giving wildly varying sensitivitiesrelative to phenotypic assays, even while using the same algorithm 187,190,229,230. Furthermore, phenotypic assays themselves can give discordant results 190. However, the traditional“gold standard” for determining HIV tropism has been the phenotypic assay. Minority X4 variantsare often better  detected by phenotypic  assays than standard,  population-based sequencing,  andthese variants can lead to treatment failure on CCR5 antagonists 175,195,229,231,232. Therefore, a genotypicassay with higher sensitivity is likely to be needed in many clinical contexts and in order to bettercorrelate  with  phenotypic  assay  results.  A  major  advance  in  HIV  genotyping  has  been  thedevelopment of next-generation sequencing and its use for detecting low-level HIV quasispecies. 1.2.9     Next-Generation Sequencing to Detect Minority HIV-1 VariantsAs  alluded  to  in  the  previous  section,  genotypic  tropism  testing  occasionally  lacks  suffcientsensitivity  to  detect  minority  non-R5  variants  in  clinical  isolates.  However,  next-generationsequencing  technology  is  an  important  alternative  to  traditional  population-based  sequencingmethods.  Next-generation  sequencing  allows  for  high-resolution  detection  of  HIV-1  minority25variants,  such that it  can estimate the actual proportion of the viral population which may be ofparticular  interest.  For  example,  this  methodology  can  be  used  to  quantify  subpopulationsharbouring  resistance  mutations  or  CXCR4  tropism.  This  approach  is  also  referred  to  as  deepsequencing, since it can probe deep into the viral “swarm”. Minority drug resistant or CXCR4-usingvariants have been shown to impact clinical outcomes on antiretroviral treatment, so their detection isan important tool in optimizing and personalizing therapy 233,234.Currently, there are a number of next-generation sequencing platforms in development or operation.Many foundational studies in HIV-1 deep sequencing have been performed on platforms developedby 454 Life Sciences  232,235–241.  The next-generation instruments used to generate the results for thisthesis are all pyrosequencing-based. Therefore, a more detailed description of this process will begiven. In order to be sequenced on a Roche/454 Life Sciences Genome Sequencer FLX (GS-FLX) or GSJunior, a PCR-amplifed DNA library must be further amplifed in a step known as emulsion PCR.Emulsion PCR is performed by combining the DNA library with microbeads at a concentration suchthat  approximately  one  DNA molecule  is  associated  with  one  microbead  242.  This  allows  clonalamplifcation of DNA on the surface of microscopic beads, which increases the strength and purity ofthe available signal. The PCR primers used to create the library are designed with a short stretch ofnucleotides complementary to oligonucleotides on the surface of the beads, which allows the beadsto bind the DNA. The beads, DNA, and PCR reagents are mixed with oil and shaken to create anemulsion.  This  process  generates  small,  independent  micelle  “microreactions”  which  are  thensubjected to thermal cycling,  After emulsion PCR, these beads are then placed onto a plate withhundreds of thousands of small wells to hold their position. The sequence of the DNA on the beads isthen determined by pyrosequencing 242. The  name  pyrosequencing  is  derived  from  the  pyrophosphate  moiety  that  is  released  duringsequencing. DNA complementary to the target DNA library on the beads is produced with successivewashes of each of the four nucleotides (Thymidine (T), Adenosine (A), Cytidine (C), Guanidine (G))232. This process is a type of “sequencing-by-synthesis”, where the sequence of the DNA library isdetermined as it is being synthesized. The incorporation of a nucleotide (or nucleotides) by a DNA26polymerase results in the release of pyrophosphate. This is then converted to adenosine triphosphate(ATP) by ATP sulfurylase. The energy from the ATP is used by the luciferase enzyme to oxidizeluciferin, causing the release of photons which are detected by a camera 242,243. Each type of nucleotideis washed over the sequencing plate independently, and a wash containing apyrase degrades anyremaining unincorporated nucleotides before the next nucleotide wash. Thus, light generated fromany given nucleotide wash can be attributed to incorporation of that base only, thereby allowing thesequence to be determined. Since these nucleotides do not contain termination moieties,  multiplenucleotides may incorporate into the nascent DNA strand if there is a stretch of the same type, knownas a homopolymer. The number of bases is proportional to the light intensity; however, the camera ispoor  at  distinguishing  the  number  of  nucleotides  incorporated  when  the  homopolymer  exceedsapproximately six bases in length 242. Thus, this method has higher error rates within homopolymerregions 211,240,242,244,245.In addition  to  these pyrosequencing platforms however,  a  number  of  alternative  next-generationsequencing instruments are in use or development 246–249. These employ different principles in signalamplifcation and sequencing chemistry but have in common the capacity for simultaneous DNAsequencing  from multiple  samples.  For  example,  there  are  several  sequencers  manufactured  byIllumina which are being increasingly used for HIV applications. For these platforms, bridge PCRamplifes  the template,  creating a clonal  cluster tethered to a solid chip  246.  Next,  sequencing-by-synthesis occurs by washing four reversibly-terminating fluorescent-labelled nucleotides. This resultsin a single nucleotide binding to a growing DNA strand that is complementary to the input strand.Imaging of the fluorescent signal of the bound nucleotide takes place, followed by cleavage of thefluorescent tag and 3’ termination moiety to allow further extension and sequencing 247,250.The SOLiD platform uses emulsion PCR, where amplifcation occurs on beads within microreactor oilmicelles 247,251, followed by sequencing-by-synthesis using a DNA ligase. A total of 16 probes is used,with each carrying a two base-pair stretch and labelled with one of four fluorescent tags. Duringsequencing, one probe at a time is ligated to the complementary DNA strand. Since there are fourdifferent probes possible for each tag, ligation is repeated a number of times, each time beginning one27base pair downstream of the previous starting site. Thus, each base in the sequence is interrogated atleast twice, allowing for deconvolution of the actual sequence and resulting in potentially lower errorrates 247,250,252. Real-time sequencing approaches are able to observe DNA synthesis directly during sequencing. ThePacifc Biosciences platform uses immobilized DNA polymerase molecules which are recorded inreal-time while they synthesize DNA with fluorescently labeled deoxyribonucleoside triphosphates(dNTPs)  253.  Another real-time sequencing approach from Oxford Nanopore  also looks extremelypromising 244,246,254. Finally, sequencing approaches that do not use light or fluorescence have also beendeveloped. For instance, the Life Technologies Ion Torrent platform detects changes in pH caused byhydrogen ions that are released when nucleotides bind to a growing DNA strand during sequencing247,255. The read lengths for this approach are continuing to increase 256. When compared, next-generation sequencing platforms all have roughly comparable performance 248.Any  or  all  of  these  platforms  have  the  potential  to  revolutionize  sequencing  in  HIV and  otherdiseases and applications. Next-generation sequencing may have special signifcance in terms of HIVcoreceptor usage since minority non-R5 variants have been found to commonly exist within majorityR5  populations  232.  Minority  variants  have  also  been  shown  to  be  relevant  to  development  ofresistance to a number of antiretroviral medications 234,257,258. While treatment for HIV has made greatadvances, resistance to antiretrovirals remains a key barrier to successful therapy. 1.3     Treatment of HIV1.3.1     Antiretroviral Treatment of HIV Infection & Development of ResistanceIn the approximately 15 years after it was frst identifed, AIDS became the leading cause of deathamongst adults aged 25 to 44 in the United States 259,260. By 1985, an antibody test was developed toscreen for HIV infection, but vaccine candidates and treatment for the virus were slower to come tofruition 261,262. The frst nucleoside reverse transcriptase inhibitor (NRTI), zidovudine, was approved28in  1987 by the U.S.  Food and Drug Administration  (FDA).  NRTIs  lack  a  3’-hydroxyl  group andtherefore act as chain terminators to the nascent DNA strand during reverse transcription 263. Prior to1996, a handful of additional antiretroviral agents were approved for use by the FDA. These includedfour NRTIs, as well as the frst protease inhibitor (PI), a class of antiretrovirals which bind to thecatalytic domain of the viral protease, preventing its function  262,263. However, monotherapy with asingle antiretroviral  agent  proved to have disappointingly  limited effcacy  264.  There is  extremelyrapid production of virus during infection untreated, or suboptimally treated infection. It is estimatedthat approximately 1 billion virions are produced each day in untreated individuals 265. This fact, incombination with the error-prone nature of HIV reverse transcriptase led to the development of drugresistance in many patients 182,183. During  this  decade,  additional  antiretrovirals  were  developed,  including  a  new  class:  the  non-nucleoside reverse transcriptase inhibitors. NNRTIs inhibit the viral reverse transcriptase protein bybinding near its catalytic domain and allosterically inhibiting it. As new agents within existing classescontinued to be developed 262,263,266, they were ultimately tested in combinations. Finally, around themid-1990s, results of  trials  of  triple therapy were released.  These trials  showed that combinationregimens could both lower plasma viraemia and slow disease progression 18,262,267–269, with the formerbeing a surrogate predictor of the latter 66,270. However, as with monotherapy, a barrier to successfultreatment with highly active antiretroviral therapy (HAART) was development of drug resistance.Testing for HIV drug resistance can be used to identify alternative therapeutic options for patientsfailing  their  antiretroviral  therapy  regimen.  Both  cell-based  phenotypic  and  sequencing-basedgenotypic assays can be used to test for HIV resistance 177,271,272, and the utility of genotypic resistancetesting has been demonstrated in randomized clinical trials 176,181,273. There are currently six classes ofantiretroviral medications in clinical usage: the NRTIs, NNRTIs, PIs, fusion inhibitors,  co-receptorantagonists,  and integrase  inhibitors  (Figure  1.1)  67.  For  each  drug  class  there  are  specifc  drug-associated resistance mutations which can be selected for by suboptimal antiretroviral treatment 274.Naturally  occurring  polymorphisms  and  envelope  variation  are  also  associated  with  decreasedsusceptibility  to  certain  antiretrovirals  275,276.  Screening  for  antiretroviral  drug  resistance  is29recommended by most treatment guidelines prior to beginning therapy and again if therapy failureoccurs  67,277–280.  Screening  for  HIV  tropism/coreceptor  usage  is  also  recommended  prior  toadministration of the CCR5 antagonist maraviroc 67.1.3.2     Coreceptor Usage & Antiretroviral Therapy Patients receiving antiretroviral therapy are more likely to harbour X4 variants than those who aretreatment-naïve  281.  This  observation  is  likely  driven  by  CD4+ T-cell  count,  since  treatment-experienced patients often have lower current or nadir CD4+ cell counts, both of which are associatedwith viral CXCR4 usage 155,281. Patients with X4 HIV may be less likely or may take longer to achievevirologic suppression on antiretroviral therapy than those with R5 HIV  173,282.  Once suppressed bytherapy,  the  HIV  population  tends  to  be  slower  to  evolve  or  change  coreceptor  usage  282,283.Additionally, certain antiretrovirals have increased or decreased activity against HIV depending onthe viral tropism 145. The effect of HIV tropism on antiretroviral susceptibility is most signifcant for antiretrovirals thatdirectly  target  the  coreceptors  108,110,111,116,284,285.  There  are  several  small-molecule  antagonists  of  thecoreceptors  that  have  been  tested  in  HIV  applications  109,110,285,286.  Currently,  the  only  coreceptorantagonist approved is the CCR5 antagonist, maraviroc. This is a small-molecule agent which bindsto the CCR5 protein and changes its conformation, thereby reducing the ability of HIV to infect cells.Maraviroc has been shown to have good tolerability and high effcacy against HIV 108,287–290. However,the drug has suboptimal  activity against  strains  which are capable of  using CXCR4. Therefore atropism test  must  be performed prior  to  treatment  67,291.  Undetected  X4 variants  present  prior  totreatment  with  maraviroc  can  compromise  its  antiretroviral  activity.  This  is  because  maraviroctherapy selects for those variants,  potentially leading to treatment failure  232,238,290,292.  These non-R5viruses often exist  as a minority of the total viral population.  Therefore sensitive detection of X4variants is likely to be a key factor in therapy success with maraviroc and other CCR5 antagonists.This principle can also be applied to early detection of antiretroviral resistance.301.3.3     Relevance of Minority Variants for Antiretroviral Resistance & Viral Coreceptor UsageAntiretroviral  effcacy  is  compromised  by  the  presence  of  HIV  with  resistance  or  reducedsusceptibility  to  components of  the  treatment regimen  234,258.  Due to the error-prone nature of  itsreverse transcriptase enzyme, HIV exists as a swarm of quasispecies which differ in their geneticmakeup  293.This fact, combined with the high rate of HIV replication, can lead to a diverse set ofvariants, any of which may carry a resistance associated mutation simply by unfortunate coincidence.Resistance  to  antiretroviral  agents  is  generally associated with genetic  changes within  the actualtarget of the drug. Since these targets are usually all essential for effcient HIV replication, mutationswithin the  genes for  these  targets can be associated with reduced viral  ftness  in  the absence oftreatment  294.  The  reduced  ftness  of  resistant  variants  limits  their  population  size  within  anindividual,  which further underscores  the  potential  importance  for minority species  detection.  Inaddition to natural accumulation of potential resistance mutations, minority drug resistant variantsmay be transmitted from one individual to another (though this has been disputed 295) or they may betransmitted as the dominant variants but decline to minority levels in their new untreated host 296,297. Several studies have demonstrated that minority drug resistant variants exist in antiretroviral-naïvepatients  296–300,  though other  investigators  have postulated that  such variants  are only spuriouslydetected 295. Some minority mutations such as the NRTI resistance mutation K65R may be naturallymore common in the context of certain genetic backgrounds (e.g., subtype C) due to there being onlyone nucleotide change between the wildtype and the mutant codons  301. Indeed, sensitive minorityspecies detection has confrmed that such variants are more common in subtype C viruses than othersubtypes  302, which raises the possibility of a higher propensity to develop full resistance to manyNRTIs. In addition to treatment-naïve patients, those with treatment experience are also at risk forminority drug resistance both on and off therapy 303–308. Upon  exposure  to  antiretroviral  medications,  resistant  variants  gain  a  selective  advantage  oversusceptible quasispecies and expand in population size in a classical darwinian selection process 183.The  selection  of  drug  resistant  variants  appears  to  be  very  rapid.  Even  after  a  single  dose  of31antiretroviral medication (such as during administration of single-dose nevirapine to prevent mother-to-child-transmission) there is suffcient selection pressure to increase the population size of resistantvariants 309 and even lead to eventual therapy failure if later treated with nevirapine or other NNRTIs310.  Importantly,  the presence of minority variants before drug exposure has been associated withemergence of resistance and treatment failure in several independent studies 233,296,298,303,304,307. Thoughsome have not found such an association 297,300.CXCR4-usage  by  HIV  is  similar  to  resistance  in  that  it  compromises  treatment  with  CCR5antagonists. However, HIV tropism also contrasts with resistance because there is a constant selectivepressure that is present within the host immune and lymphatic systems. The ability of an individualvirion to use both CCR5 and CXCR4 (i.e., dual tropism) could be seen to have a theoretical selectiveadvantage due to an ability to infect a wider range of cells. However, there are conflicting argumentsas to whether this may be the case  145,149,222,311,312, especially since CXCR4-usage is associated with adecline in target CD4+ T cells 169,172,173. Nevertheless, CXCR4-usage does evolve in a proportion of HIV-infected patients over time. However,  perhaps as a result  of this balance between advantage anddisadvantage,  this happens in a minority of patients,  and X4 variants tend to exist  as minoritieswithin a majority CCR5-using viral population 153,281,313–315. Again, as with minority resistant variants,minority X4 subpopulations can lead to antiretroviral treatment failure 232,316,317. Consequently, there is a strong rationale to apply deep sequencing in order to detect minority X4variants. Additionally, the methodologies which are applied to sequencing the HIV envelope gene canbe readily extended to  other  parts  of  the  virus,  potentially  enabling detection  of  other  minorityvariants with reduced susceptibility to antiretrovirals. This thesis can therefore be considered as aspecifc application of a very flexible, multi-application tool which has implications that extend farbeyond viral coreceptor usage.  321.4     Thesis Overview1.4.1     Thesis Organization & ObjectivesThis thesis is organized into seven chapters. Chapter 1 introduces the HIV epidemic as a whole. Itcontains general information about the viral composition, life cycle, target cells, pathogenesis, andtreatment.  There  is  also  an  overview  of  HIV  coreceptor  usage,  tropism  assays,  next-generationsequencing,  and the clinical relevance of these topics.  Chapters  2 through 6 address the  primaryobjectives of the thesis, as detailed below. Chapter 7 summarizes the results of the research, discussesthese  results,  and comments  on  their  implications.  Figures  and tables  are  found after  their  frstmention in the text.  References for all  chapters are presented in a Bibliography at the end of thethesis. The general hypothesis of these studies is that deep sequencing is equivalent or superior toalternative tropism assays and that  it  can be used to better  assess HIV coreceptor  usage and itsimplication in treatment with maraviroc.The general aim of this thesis is to establish the clinical relevance of next-generation sequencingin HIV applications, with an emphasis on those relating to HIV tropism and treatment with CCR5antagonists. Chapter 2 introduces the methodology and performance of next-generation sequencing, while alsocomparing various other tropism assays. Chapters 3 through 5 all focus on three trials of treatment-experienced patients. Each chapter focuses on a different aspect of these trials. Chapter 3 applies deepsequencing  to  HIV  RNA  from  plasma.  Chapter  4  investigates  the  utility  of  next-generationsequencing in assessing the cellular compartment in HIV-infected individuals. Chapter 5 comprises adetailed longitudinal analysis of HIV envelope sequences prior to and after drug selection pressureby  maraviroc.  Finally,  Chapter  6  applies  deep  sequencing  to  an  entirely  different  treatmentpopulation of antiretroviral-naïve patients.  The thesis thus addresses a number of aspects of HIVinfection and treatment which can be assessed and predicted with next-generation sequencing. 331.4.2     Overview of Data SourcesThe primary data sources for this thesis have been four large clinical trials of maraviroc conductedinternationally.  Two  trials,  Maraviroc  versus  Optimized  Therapy  in  Viraemia  AntiretroviralTreatment-Experienced patients (MOTIVATE) 1 and MOTIVATE 2, investigated maraviroc or placebo,both with optimized background regimens in therapy-experienced patients with CCR5-using HIV-1as determined by the phenotypic Trofle assay. A separate but related trial, A4001029 had an identicalstudy design to the other  two,  but  enrolled patients  with phenotypically assessed non-R5 HIV-1infections.  The fourth  trial  was  conducted  in  treatment-naïve  patients  who  were  randomized toreceive the NRTIs zidovudine and lamivudine, plus either maraviroc or the NNRTI, efavirenz. The samples examined in this thesis were those drawn from patients when they were screened foreligibility in the trials, or were drawn on the frst day of treatment (baseline). Additional sampleswere  also  obtained  from later  time-points  from three  of  these  trials  for  patients  who  were  notresponding optimally to therapy. The sample types included were either blood plasma or peripheralblood mononuclear cells  from patients participating in these trials.  Additional samples  were alsoobtained from a subset of the HAART Observational Medical Evaluation and Research (HOMER)cohort, which is based in British Columbia. Ethical approval for all of the studies presented in this thesis was granted by the Providence HealthCare/University of British Columbia Research Ethics Board. Versions of Chapters 2 through 6 haveall been published in several international, peer-reviewed journals:  the Journal of Acquired ImmuneDeficiency  Syndromes,  the  Journal  of  Infectious  Diseases,  Clinical  Infectious Diseases,  and  AntimicrobialAgents and Chemotherapy. The candidate is the lead author on all of these manuscripts, and is leadauthor  on  two  additional  review  articles,  which  are  included  in  part  in  Chapter  1,  and  werepublished in HIV Therapy and Current Opinion in HIV and AIDS.34Chapter 2:     Improved Detection of CXCR4-Using HIV by V3 Genotyping: Application of Population-Based & Deep Sequencing to Plasma HIV RNA & Proviral HIV DNA 2.1     Background & IntroductionHIV gains entry into a cell through the use of its envelope protein, gp120. During entry, it binds to thehuman  CD4  receptor  and  a  coreceptor  —  either  CXCR4  (X4  HIV)  or  CCR5  (R5  HIV)  100.  Thecoreceptor used by a virus to gain cellular entry is referred to as its tropism or coreceptor phenotype(e.g.,  X4, R5,  or dual-tropic).  More advanced disease progression is often associated with CXCR4tropism and detectable X4 viral load  100. Furthermore, with the emergence of the CCR5 antagonistantiretroviral drug class (e.g., maraviroc  108), coreceptor usage has become more clinically relevant,since  the effcacy of  these  CCR5 antagonists  is  dependent  on the CCR5 coreceptor  being almostexclusively used by the patient’s virus 291,316. Many tests are available to screen for tropism, each withits own advantages and disadvantages  175,191.  One of the most commonly used tropism tests is theTrofle coreceptor assay (Monogram Biosciences) 194 and the Enhanced Sensitivity Trofle assay (ESTA)195,318.Genotypic screening methods for determining coreceptor usage have the potential to be faster, lessexpensive  and  more  easily  standardized  than  current  phenotypic  methods  197.  This  approach  ispossible because viral tropism is reflected in the genetic sequence of the gp160 protein, with its thirdvariable  domain,  or  V3  loop,  being particularly  predictive  of  coreceptor  usage  319.  Bioinformaticalgorithms use V3 sequence data to predict coreceptor phenotype 197,203,204. By analogy, drug resistancetesting is routinely performed by inferring a phenotype (in this case, degree of virologic response to adrug) from genotypic sequence data. However, standard, population-based V3 sequencing may lacksensitivity for minority X4 HIV 229. 35In a clinical setting, it may be useful to detect the presence of minority species at low concentrations,such as low levels of X4 virus that have the potential to emerge following treatment with a CCR5antagonist. Although previous efforts to determine viral tropism through more traditional genotypicmethods have appeared inadequate  229,  generating sequences through independent triplicate PCRamplifcation  and/or  by  deep  sequencing  may  improve  upon  the  sensitivity  of  these  results.Triplicate PCR amplifcation increases the probability that X4 minority species will be amplifed froma given sample extract. Deep sequencing can be performed on a Roche/454 Life Sciences GenomeSequencer  FLX System and generates  data  for  many individual  variants  within a  given sample,including  X4  variants.  Deep  sequencing  allows  amplifcation  and  increased  detection  of  raresubpopulations, thereby increasing the threshold for detection of low-level viral populations.An additional drawback of current tropism assays is that it is not currently possible to screen patientswho wish to switch to a CCR5 antagonist for reasons of tolerability or otherwise, but who currentlyhave  viral  loads  below  1000  copies/ml  or,  indeed,  undetectable  viral  loads.  This  stems  from alimitation of assays based on plasma HIV, such as the Trofle assay, which requires plasma viral loadsgreater  than 1000 HIV RNA copies/mL. As an alternative,  proviral  DNA can be amplifed fromperipheral  blood  mononuclear  cells  (PBMCs)  and  the  coreceptor  usage  of  these  species  can  beinferred using methods similar to those performed on plasma HIV RNA. Interestingly, such methodsmay be performed even when pVL is undetectable,  so it  may be advantageous to use genotypicprediction  methods  to  determine  viral  tropism  from  proviral  DNA  in  patients  with  low  orundetectable plasma viral loads. The aims of the current study were to improve detection of X4 HIV using a number of genotypictropism methods, and to compare them to phenotypic Trofle assay results. Standard, population-based sequencing and deep sequencing were performed on triplicate amplifcations of the HIV V3region. Amplifcations were made from both viral RNA in plasma and from integrated proviral DNAin peripheral blood mononuclear cells (PBMC), where plasma viral load was undetectable. Tropismwas inferred using bioinformatic algorithms, and the results from these various genotypic methods36were compared to those of the Trofle assay. This approach was then validated in an independentdataset of screening samples from the MOTIVATE 1 and MOTIVATE 2 trials of maraviroc 288,289,291. 2.2     Materials & Methods2.2.1     Cohort Description & PatientsV3 loop sequence variation was assessed in samples from a cohort of antiretroviral-naïve, chronicallyinfected individuals initiating antiretroviral  therapy. The primary study group represents a subset(N=63 patients) of the well-characterized HAART Observational Medical Evaluation and Research(“HOMER”) cohort 320. Individuals were included in the present study by convenience, based on theavailability of a peripheral blood sample for PCR amplifcation and a documented Trofle assay result155.  Ethical  approval  was  granted by the Providence Health Care/University  of  British  ColumbiaResearch Ethics Board.2.2.2     Extraction & Population-Based SequencingHIV RNA was extracted from previously frozen plasma samples, and HIV DNA was extracted frombuffy coat samples, both using a NucliSENS easyMAG (bioMerieux). Both RNA sequencing methods(i.e., population-based, and deep sequencing) followed the same procedures up to and including frstround PCR, but differed in later steps, such as using different second round PCR primers. The regionencoding the  HIV V3  loop was  amplifed independently  in  triplicate  by  nested  RT-PCRs set  upsimultaneously  from  extracts  using  a  multichannel  pipette  —  the  additional  effort  is  minimalcompared to a single PCR. Triplicates were chosen mostly arbitrarily as a compromise between potential increased probabilityfor amplifcation of X4 HIV and a procedure that is clinically feasible for a technician to perform on asingle PCR plate. Sequencing was performed in the 5' and 3' directions on an ABI 3730 automated37sequencer as previously described  173. All primers and thermal cycler protocols for all methods areavailable in Appendices I and II.2.2.3     Deep Sequencing & Emulsion PCR MethodsDeep sequencing on the Roche/454 Life Sciences Genome Sequencer FLX (GS-FLX) is a sensitivesequencing technique able to detect low-level subpopulations of virus and generate thousands ofsequences  from a  given  sample  232,248,321.  Second  round  PCR primers  were  designed  with  fusionprimers to fuse to the emulsion PCR beads required by the pyrosequencing technique. Also includedwere 12 unique multiplex “barcode” sequence tags to enable the identifcation of samples after thesequencing was complete. After PCR amplifcation, the concentrations of the PCR products were quantifed using a Quant-iTPicogreen dsDNA Assay Kit (Invitrogen) and a DTX 880 Multimode Detector (Beckman Coulter).After quantitation,  they were combined in equal proportions (2 x 1012 DNA molecules  from eachtriplicate sample), purifed with Agencourt Ampure PCR Purifcation beads (Beckman Coulter), andsubsequently re-quantifed. Note that only samples from which all three triplicates were successfullyamplifed were combined. Otherwise, RT-PCR amplifcation was re-performed until three triplicateswere available. This increased the probability that CXCR4-using minority species would be amplifedand sequenced from a given sample's RNA extract.Following  its  quantifcation,  the  combined,  purifed  amplicon  “library”  was  then  diluted  to  aconcentration of 2 x 105 molecules per millilitre, and combined at a ratio of 0.6 molecules to 1 DNAcapture  microbead  used  for  emulsion  PCR  (emPCR).  This  ratio  of  less  than  1:1  increased  thelikelihood  that  a  single  microbead  would  bind  a  single  DNA amplicon,  such  that  subsequentamplifcation by emulsion PCR would generate homogeneous clones on each bead.Two separate emPCR amplifcations were performed: one for the forward sequencing direction, andone for the reverse. Along with the DNA amplicons, the microbeads are mixed with amplifcation38buffer, primer and enzyme, as well as with oil.  These components are shaken with a TissueLyser(Qiagen/Retsch) to allow formation of oil microreactor micelles around the beads. This process givesmostly independent reaction sites to allow clonal amplifcation of the DNA amplicon, thus increasingthe signal for subsequent pyrosequencing.The emulsions are then broken and the beads are washed with isopropanol,  followed by two 454washing reagents.  The microbeads  are  then enriched with magnetic  beads such that  only  beadscoated in DNA are carried forward into subsequent steps.  After enrichment,  an annealing step isperformed to anneal the pyrosequencing primers onto the bead-bound DNA amplicons. The DNAbeads are then added onto a picotitre plate (divided into 4 regions) at a density of 2.5 × 105 beads perregion,  as  quantifed  with  a  Z1  Coulter  Particle  Counter  (Beckman  Coulter).  Control  beads  (togenerate quality scores for the sequencing run), packing beads (to hold the DNA beads in place in thepicotitre plate) and enzyme beads are all added over the plate as well. After this point, the plate isprepared for deep sequencing on the GS-FLX. The  sequence  amplifed  on  each  bead  was  determined  by pyrosequencing  on  the  GS-FLX.  Thisprocess generated ~200 base pairs of data in each direction per amplicon, with a typical V3 loopconsisting of 105 base pairs (35 amino acids). Truncated reads (defned as sequences missing ≥4 basesat the 5’ or 3’ end) were not included in the analysis. In total, 12 HOMER plasma samples underwentdeep sequencing with the GS-FLX.Proviral HIV DNA V3 sequences were assessed in a similar manner in 26 HOMER subjects with non-R5  HIV and 14  with  R5  HIV.  These  subjects  were  receiving  highly  active  antiretroviral  therapy(HAART) 155 and had undetectable plasma viral loads. Sample material consisted of PBMCs from thebuffy  coat  fraction of  centrifuged whole  blood.  Patient  phenotypic  tropism had previously beendetermined using Trofle prior to initiating HAART. Nested PCR was performed for bulk sequencingon an ABI 3730 sequencer. Deep sequencing used the same frst round primers as bulk sequencing,with different second round primers. Proviral DNA from a total of 12 buffy coat samples from theHOMER cohort underwent deep sequencing on the GS-FLX.392.2.4     Sequence Analysis & Coreceptor Usage Determination by Bioinformatic AlgorithmsAfter sequencing on the ABI 3730, data were analyzed using the custom software, RECall 196 with nomanual  intervention.  RECall  has  been  shown  to  have  ~99% concordance  with  human  calls  196.Nucleotide mixtures were automatically called if the  secondary peak height exceeded 12.5% of thedominant peak height.  Sequences were aligned to HIV-1 subtype B reference strain HXB2 (GenbankAcc. No. K03455) using a modifed NAP algorithm 322. HIV tropism was predicted from V3 genotype using position specifc scoring matrices (PSSMX4/R5) 203and/or  geno2pheno[coreceptor] 197 scoring.  Non-genotypic  factors  such  as  CD4+ cell  counts  were  notincluded in the bioinformatic analysis.  Results  were compared to  the Trofle data as  a reference.Standard sequencing replicates  with PSSM values below the  predetermined cut-off  of  -6.96 werecalled R5, while those with scores greater than or equal to -6.96 were called X4 203.  The geno2phenomethod 197,204 used a 5% false-positive rate, with samples also categorized as R5 or X4. Where sequence ambiguity occurred due to the presence of nucleotide mixtures, the permutationwith the highest PSSM score was used to assign the score for a given replicate. In other words, the“most”  X4  residues  were  retained  from  a  sequence  where  multiple  variants  contributed  to  theconsensus  population  sequence.  This  method  was  used  in  order  to  increase  the  sensitivity  fordetection of X4 variants 229. A similar system was used for geno2pheno. Where triplicate data differed,the most highly X4 (e.g., maximum PSSM score) replicate was assigned to a sample. Thus, R5 sampleshad all 3 replicates inferred as R5, and X4 samples had one or more replicates inferred as X4. For deepsequencing, each V3 variant detected received a tropism classifcation using the same two algorithms.This allowed the proportion of X4 virus within the sample to be determined, and samples  wereclassifed according to this  parameter.  Sensitivity was  defned as  the  prediction of  CXCR4-usagewhich correlated with the original Trofle assay.402.2.5     Independent ValidationThe performance of the triplicate method was also assessed in a blinded independent dataset (N=278)of  screening  samples  from  the  Maraviroc  versus  Optimized  Therapy  in  Viremic  AntiretroviralTreatment-Experienced Patients (MOTIVATE) studies  288,289. All patients in MOTIVATE consented toother tropism testing being performed on their samples. A sister trial,  the A4001029 study had apatient  criterion of  non-R5 (X4, Dual/Mixed,  non-reportable)  virus.  Samples  from this  trial  wereexcluded since these samples could not be blinded.The MOTIVATE screening samples were amplifed in triplicate and bulk-sequenced on an ABI3730.As  with  the  HOMER  subset,  MOTIVATE  samples  showing  evidence  of  CXCR4-usage  in  theirstandard sequencing results were classifed as X4. Subjects with R5 (by Trofle) at both screening andbaseline (just prior to treatment with study medication) were classifed as “confrmed R5s”. A subsetof 11 MOTIVATE samples were then amplifed independently in triplicate as above and underwentdeep sequencing on a GS-FLX.2.3     Results2.3.1     Standard Sequencing of V3 to Infer TropismStandard, population-based sequencing of triplicate amplifcations of the V3 loop, in combinationwith PSSM tropism inference, gave approximately 81% concordance with Trofle. Of samples calledR5 by Trofle, 31 of 34 (91%) were also identifed as R5 by standard sequencing. Of those called X4 byTrofle, 20 of 29 (69%) were inferred as X4 (Figure 2.1). Often, there was notable variation among the three independent amplifcations performed for eachsample. Indeed, 12/63 samples (19%) had at least one replicate indicate a different tropism than theothers. 41Figure 2.1: PSSM Scores from Standard Population-based Sequencing of Independent Triplicate PCRs of V3 Amplified from Plasma HIV RNA Figure 2.1: PSSM Scores from Standard Population-based Sequencing of Independent Triplicate PCRs of V3 Amplified from Plasma HIV RNA.  The  horizontal  axisrepresents the possible PSSM scores, with scores to the left of the vertical line (-6.96) indicating R5 virus, and scores to the right of -6.96 indicating X4 virus. Samples are arranged bytheir Trofle screening result, with the Trofle X4/DM samples in the upper region of the fgure, and the Trofle R5 samples in the lower region. Closed circles indicate the PSSM score ofeach of three replicate amplifcations of the V3 loop for different samples. Horizontal lines span the range of the three scores in order to give an indication of the diversity within arespective sample. Where three circles are not visible, this is either because of a failed amplifcation or because the points overlap because of similar or identical PSSM scores.42Almost half of Trofle X4 or Dual/Mixed samples that were called X4 by standard sequencing had atleast one amplifcation that would have been classifed as R5 if the triplicate approach had not beenused (9/20 samples,  45%).  It  is  currently  unknown as  to  what  effect  replicate  testing of  clinicalsamples would have had on Trofle assay results.Receiver  operator  characteristic  (ROC)  curves  were  plotted  using  either  the  maximum of  threetriplicates, or a “singleton” approach using only the frst of the triplicates. The area under the curve(AUC) of the ROC curve for the triplicate approach was 0.874 versus 0.828 for the singleton approach,indicating  improved  performance  of  the  triplicate  method  over  a  single  amplifcation.  Overall,comparing the result (R5 or X4) by standard sequencing with PSSM, and using the Trofle result as areference, the sensitivity for the HOMER samples was 69% and specifcity was 91%. In comparison,keeping specifcity constant, singleton approach would have given 48% sensitivity, and a duplicateapproach, 59% sensitivity, relative to Trofle. 2.3.2     Proviral DNA to Infer TropismPBMC  samples  were  retrieved  from  patients  who  were  currently  on  HAART  (without  CCR5antagonist medication), had undetectable plasma viral loads at the time of sampling, and for whom apre-therapy plasma sample and Trofle assay result were available. Of 46 samples initially attempted,40  samples  yielded  successful  amplifcations,  giving  an  87%  amplifcation  rate.  Proviral  DNAsamples  were  amplifed  in  triplicate,  bulk  sequenced  on  an  ABI  3730,  and  the  sequences  wereinferred as R5 or X4 by PSSM (Figure 2.2).   For samples called non-R5 by Trofle from plasma RNA, 20/26 (77%) had evidence of X4 HIV DNA intheir peripheral blood mononuclear cells. A total of 10/14 samples (71%) called R5 by Trofle had R5sequences in their corresponding proviral DNA, giving sensitivities and specifcities of 77% and 71%,respectively, for PSSM; or 77% and 93%, respectively, for geno2pheno (data not shown). The meanproviral DNA PSSM score of each sample was also correlated to the mean pre-treatment RNA PSSMscores (r2 = 0.35).43Figure 2.2: PSSM Scores from HIV Proviral DNA by Standard Population-Based Sequencing Figure 2.2: PSSM Scores from HIV Proviral DNA by Standard Population-Based Sequencing.  Possible PSSM scores on the X-axis. Scores left of the dashed vertical line (-6.96)indicate R5 virus; scores to the right indicate X4 virus. Samples are arranged by their Trofle status. Note that the Trofle result is based on plasma RNA and not proviral DNA. Closedcircles indicate the PSSM score of each V3 amplifcation for different samples. Horizontal lines span the range of the three scores for each sample. Where three circles are not visible,this is either because of a failed amplifcation or because the circles overlap due to similar or identical PSSM scores.442.3.3     Deep Sequencing of HIV RNA & DNAA subset of patients with matching plasma RNA and proviral DNA (N=12) samples were sequencedusing  the  GS-FLX.  The  RNA was  extracted  from  plasma  samples  drawn  prior  to  initiation  ofantiretroviral therapy, while the DNA was extracted from buffy coat samples drawn after patientsachieved  undetectable  viraemia,  after  a  median  of  36.5  months  (IQR:  30.5–39)  on  antiretroviraltherapy. These samples were assessed according to the percentage of X4 virus comprising their deepsequencing results. A total of 4 of the 12 patients (33%) had very similar proportions of X4 virus intheir plasma RNA and proviral DNA, (within ~1% of each other). For the remaining 8 samples, thepercent X4 in RNA and proviral DNA differed by a range of 6-72%, with proviral DNA tending toharbour a higher percentage of X4 variants (median 46% X4 in DNA versus 8% in RNA). Table 2.1shows the comparison of these 12 patients across the various tropism methods.  Table 2.1: Deep Sequencing of HIV RNA & DNA Compared with Standard Population-Based Sequencing & the Trofile AssaySample TrofileResultStandard Sequencing Deep SequencingPlasmaRNAProviralDNAPlasmaRNA (%X4)ProviralDNA (%X4)1 R5 R5 R5 0.05 0.62 R5 R5 R5 0.4 0.43 R5 X4 R5 0.4 0.14 R5 R5 X4 1.9 74.35 D/M R5 X4 6.8 53.26 D/M X4 R5 7.6 1.27 D/M X4 X4 8.7 48.48 D/M R5 X4 15.9 49.49 D/M X4 X4 19.5 39.810 D/M R5 X4 71.0 99.411 D/M X4 X4 84.5 43.112 D/M X4 X4 99.8 99.6Table 2.1: Deep Sequencing of HIV RNA & DNA Compared with Standard Population-Based Sequencing & the TrofileAssay. Deep sequencing was performed on 12 plasma (RNA) samples and 12 PBMC (DNA) samples, and thepercent  X4  in  each  sample  was  compared  to  the  corresponding  Trofle  result  and  maximum  PSSM score  by  standardsequencing. Standard sequencing classifed samples as X4 if the maximum PSSM score was ≥-6.96. All non-R5 results arebolded. For deep sequencing, samples are bolded if >2% X4 virus was detected by the GS-FLX. D/M — Dual/Mixed Tropic45Overall,  the deep sequencing percent  X4 from pre-treatment  plasma RNA, and post-suppressionproviral DNA were well correlated (r2=0.44), and also corresponded very well to the pre-treatmentTrofle results. Using RNA and DNA, respectively, 4/4 and 3/4 samples called R5 by Trofle had lessthan 2% X4 virus within their deep sequencing results, while 8/8 and 7/8 samples called non-R5 byTrofle had greater than 2% X4 virus comprising their deep sequencing results. Standard sequencingof RNA and DNA gave X4 calls in 11/12 samples (92%) that had 20% or more X4 by deep sequencing,and  gave  R5  calls  in  9  of  12  (75%)  samples  with  less  than 20% X4,  consistent  with  the  typicalsensitivity of standard sequencing in reliably detecting minority species. Indeed, the presence of low-level (below 20%) X4 variants could explain 3/4 (75%) Trofle-non-R5 samples which were apparentlymisclassifed as R5 by standard sequencing.2.3.4     Results of Independent ValidationThe sensitivity and specifcity of the current approach were ascertained on a blinded, independentsample  set  (N=278)  from the MOTIVATE trials,  which tested maraviroc in treatment-experiencedindividuals. A previous attempt to determine tropism by standard sequencing (not in triplicate) withPSSM methods  had only  24% sensitivity with 97% specifcity  when compared to Trofle  229.  Theindependent  validation  of  the  current  method  with  bioinformatic  analysis  using  PSSM yieldedsubstantially increased sensitivity (75%) with only a modest decline in specifcity (83%). Compared toPSSM, geno2pheno methods yielded a slightly worse sensitivity (61%) but had higher specifcity(93%), though there was limited power to distinguish either algorithm as superior. For “confrmed” R5 samples (i.e., those called R5 by Trofle at time of screening and again at baselinejust  before  starting  maraviroc),  genotyping  had  sensitivities  and  specifcities  of  71%  and  95%,respectively, for geno2pheno and 75% and 82%, respectively, for PSSM. A subset of 11 MOTIVATEsamples also underwent deep sequencing. Many of these samples had at least a minority of non-R5sequences present in their deep sequencing results. There was a wide distribution of PSSM scorespresent within these samples, representing a viral population that often comprised both CXCR4- andCCR5-using  variants  .  For  10  of  11  samples  (91%),  the  PSSM score  of  the  most  common deep46sequencing  variant  resulted  in  the  same  tropism  classifcation  as  the  Trofle  assay  classifcation(Figure 2.3). 2.4     Discussion & ConclusionsClinical samples were assessed using standard, population-based sequencing and deep sequencing ofthe  HIV  V3  region,  and  the  results  show higher  sensitivity  for  detecting  CXCR4-using  virus  insamples than previously achieved 229. Of additional signifcance was the use of proviral DNA to inferviral tropism in treated patients with undetectable plasma viral loads. Deep sequencing also seemedto be a good predictor of Trofle results, with a cut-off of 2% X4 giving good concordance with theoriginal Trofle assay. Relatively few studies performed prior to this study used proviral DNA 323 or deep sequencing 232,317to infer tropism. Genotypic tropism testing from proviral DNA suggests the possibility of screeningfor those with suppressed viraemia who may wish to switch to CCR5 antagonists for reasons such astolerability, whereas the Trofle assay requires a viral load of at least 1000 copies/mL 195. With theabove outlined approach, most patients harbouring X4 virus can be quickly screened out as beingineligible to receive maraviroc.  The differences in the results of the earlier study versus the current one may be attributable to anumber of factors, especially: triplicate amplifcation, better sequencing technology, and automaticbase-calling. The use of independent triplicate amplifcations here may be able to amplify a greaterproportion of minority species due to the inherently stochastic nature of PCR. This was evidenced inthe variability amongst the replicates, with approximately 20% of samples yielding replicates withdifferent inferred coreceptor usage. Also lending evidence to the utility of the triplicate approach wasthe larger area under the curve for the ROC curve plotting the triplicate approach versus that of asingleton  approach  (0.874  versus  0.828,  respectively).  Triplicate  PCRs  also  demonstrated  greateroverall sensitivity for detecting CXCR4-using variants (69% for the triplicate approach versus 48% fora singleton approach).  47Figure 2.3: Distribution of PSSM Scores for Variants Detected by Deep Sequencing of Plasma Samples Figure 2.3: Distribution of PSSM Scores for Variants Detected by Deep Sequencing of Plasma Samples There was a large degree of variation in PSSM scores withinsamples, according to deep sequencing. The large circle indicates the PSSM score of the most common variant detected by deep sequencing for each sample. Diamonds indicate thePSSM scores of the rest of the ten most common variants detected. The horizontal axis represents the possible PSSM scores, with scores to the left of the dashed vertical line (-6.96)indicating R5 variants, and scores to the right of -6.96 indicating X4 variants. Samples are arranged by their Trofle classifcations, with X4/DM calls by Trofle on the top and colouredin red, and R5 calls below and coloured in green.48The sequencing technology has improved, with the ABI 3730 used for the current study, and eitherthe ABI 3100 or 3700 used for the earlier study. The Roche/454 Life Sciences Genome Sequencer-FLXalso represents a further advance in sequencing technology. An additional advantage in the currentmethod is the automated nature of the sequence analysis. Due to analysis by RECall, sequence dataunderwent no manual intervention such as base-calling. Use of RECall made this method quick andeffcient  while  bypassing  the  inherently  inconsistent  and  labour-intensive  process  of  manualsequence analysis by a technician.Some  limitations  of  the  sample  population  should  be  noted.  The  frst  63  available  samples  asorganized by sample identifer were arbitrarily chosen, resulting in a potentially unknown selectionbias. Furthermore, the clinical test set from British Columbia is composed of 97.5% clade-B virus, thusskewing the results in favour of methods trained primarily on clade-B  203,204.  The PSSM algorithmused here may not be readily extendable to non-clade B sequences. It should also be noted that theoriginal Trofle assay was used for these analyses but not the enhanced sensitivity Trofle assay, ESTA.The  current  results  may have  differed  if  these  genotypic  methods  were compared to  ESTA.  Forinstance, some of the Trofle R5 samples may have yielded X4 or Dual/Mixed results if tested byESTA, which may have decreased specifcity. Interestingly, the deep sequencing genotypic methods were able to detect low levels of inferred-X4virus in almost all samples that underwent deep sequencing, regardless of their Trofle assay results.Indeed,  some  samples  which  had  R5  results  by  Trofle  gave  entirely  different  results  by  deepsequencing, which revealed a majority of X4 variants. Subsequent chapters will explore how suchdiscordant patients (e.g., Trofle R5/deep sequencing X4) perform virologically on CCR5 antagonisttherapy. These results do suggest, however, that tropism may not always fall into discrete categoriesof R5 or X4, and that the proportion of a patient’s virus that uses either coreceptor may be variable.The “true” sensitivity of genotypic tropism testing is confounded by the “gold standard” againstwhich  these  tests  are  compared.  Numerous  studies  have  compared  a  variety  of  genotypic  and49phenotypic  tests,  each  yielding  varying  sensitivities  and  specifcities  187,190,230.  Concordance  evenbetween phenotypic tropism assays is not necessarily 100% (e.g., 190). Furthermore, depending on thetests used, genotypic sensitivity for X4 variants has ranged from as low as 10%  187 for genotypicpredictors, such as the 11/25 charge rule 324, to ~70% for support vector machines 190 and geno2pheno187.  Even the  same algorithm used on different  datasets  can yield vastly different  sensitivities  230.Because of the wide range of sensitivities reported from genotypic testing, the results reported hereshould be taken in this context. This study yielded improved detection of X4 HIV in clinical samples and was a better predictor ofviral tropism than many previous attempts at genotypic approaches to determining coreceptor usage.These data also suggest that deep sequencing technology and genotypic analysis of proviral DNAmay prove useful in the determination of coreceptor usage. With the above outlined approach, mostpatients harbouring X4 virus can be quickly screened out.  Using the Trofle call as the reference may, however, be problematic. Ultimately, the best indicationagainst which results should be compared is the virologic outcome of patients who receive CCR5antagonist medication. Clinical  outcome, and not other assays,  may be the best candidate for the“gold standard” of comparison 325. Therefore, the next chapter of this thesis tests the ability to predictoutcomes on maraviroc in treatment-experienced patients entering three large clinical trials of theCCR5 antagonist. Methods developed above are optimized and refned in the following chapter, withthe primary focus being the performance of deep sequencing situated in a clinical context.50Chapter 3:     Deep Sequencing to Infer HIV-1 Coreceptor Usage: Application to Three Clinical Trials of Maraviroc in Treatment-Experienced Patients 3.1     Background & IntroductionHuman Immunodefciency Virus Type 1 (HIV-1) enters and infects a target cell by an interaction of itsenvelope glycoprotein,  gp120, with the cellular CD4 receptor and a co-receptor:  CCR5 or CXCR4100,116,326,327. CCR5 antagonists such as maraviroc inhibit HIV entry via CCR5. These agents work byallosterically altering the conformation of CCR5 at the cell surface, thereby disrupting its interactionwith  HIV  gp120  100,108,286.  However,  CCR5  antagonists  have  suboptimal  activity  against  viralpopulations capable of using CXCR4  291,316. Accordingly, before clinical use of CCR5 antagonists, atropism test is performed to rule out the presence of detectable non-CCR5-tropic (non-R5) virus.Some  of  the  most  widely-used  coreceptor  tropism tests  have  been  the  recombinant,  phenotypicTrofle assay (Monogram Biosciences) 194, or its newer iteration, the Enhanced Sensitivity Trofle assay(“ESTA”) 195. Despite their wide use, there are some practical limitations to these assays, including along turnaround time, restricted geographic access, and large sample volume required 328. Genotypictropism testing is an alternative method 329 that is possible because the sequence of the third variable(V3)  loop  of  HIV gp120  is  the  principal  determinant  of  tropism  23,123,220,324,330,331,  allowing  tropisminference using bioinformatic algorithms such as PSSMX4/R5  203 and geno2pheno[coreceptor] 197,204.However,  genotypic  assays  based on  standard,  population-based  V3  sequencing  have  often  hadapparently  poor  sensitivity  for  detection  of  non-R5  HIV (e.g.,  229),  especially  when  such  speciescomprise  minorities  in  the  viral  population  below  ~20%,  the  reliable  sensitivity  of  standardsequencing  180,228.  In  comparison,  next-generation  deep sequencing  approaches have  much highersensitivity and can detect minority HIV variants at much lower levels 240,332, including minority non-51R5  subpopulations  232.  Consequently,  this  method  can  capture  a  detailed  “cross-section”  of  co-receptor-usage across a patient’s viral population, and quantify the prevalence of non-R5 HIV withinthe patient.Presented  here  is  an  extensive  study  of  deep  V3  sequencing  as  a  tool  for  predicting  virologicoutcomes on maraviroc-based therapy in  treatment-experienced patients  in  the Maraviroc  versusOptimized Therapy in Viremic Antiretroviral Treatment-Experienced Patients (MOTIVATE) 1 and 2studies.  These  were  randomized,  phase  3,  placebo-controlled  studies  of  maraviroc  in  treatment-experienced patients with R5 HIV  288,289. Patients were originally screened using the original Trofleassay. Of those screened out due to non-R5 HIV, approximately 20% (186/955) entered the A4001029trial  291. This trial also assessed maraviroc versus placebo but in patients with non-R5 HIV results.Deep sequencing was retrospectively tested on a total of 1827 blinded screening samples from thesethree clinical trials, and assessed for its ability to predict virologic responses in maraviroc recipients.3.2     Materials & Methods3.2.1     Trial Patients, Samples & Amplification Methods Briefly,  the  V3  loop  of  HIV  gp120  was  amplifed  independently  in  triplicate  by  nested  RT-PCRmethods from a total of 1827 screening samples from the three trials. These were then sequenced byeither (A) standard, population-based sequencing 325, or (B) deep sequencing 333. Refer to Chapter 2for detailed methodologies. The current study focuses on the deep sequencing data, hereafter referredto as genotyping.In total, 1093 of 1827 patients examined in the current study were randomized into the three arms ofthe MOTIVATE (R5) and A4001029 (non-R5) trials (Figure 3.1). Informed consent was obtained fromall  individuals.  Treatment arms were  maraviroc once-daily (QD),  maraviroc  twice-daily (BID),  orplacebo,  plus an optimized background therapy of 3 to 6 agents, based on treatment history andresistance testing 288,289.52Figure 3.1: Sample & Patient DistributionFigure 3.1: Sample & Patient Distribution Of the current study population, patients were screened for entry into either MOTIVATE-1 (76%) or MOTIVATE-2 (24%).In both studies, a majority of patients had R5 HIV at screening by the original Trofle assay, while a minority had non-R5 results. A subset of these patients were enrolled into the trialsand received treatment in one of three arms: placebo (PBO), maraviroc once-daily (MVC QD), or maraviroc twice daily (MVC BID). *Patients with non-R5 HIV at screening weretreated in the A4001029 study, with 8 patients being screened and enrolled into A4001029 directly. Note that all phenotypic screening results were performed using the original Trofle assay (approximate 10% non-R5 cutoff 334) and not ESTA (0.3%cutoff  195). Figure 3.1 shows the distribution of patient samples tested in the current study. The primary analysis was based on all patients whoentered any study (MOTIVATE-1, MOTIVATE-2 or A4001029). Critically, this included all treated patients for whom Trofle gave a non-R5 result.For additional analyses with respect to tropism assessments by both assays, the patients screened for MOTIVATE-1 (including Trofle-non-R5patients),  but  who did not  enter  a  study were also  included.  However,  only patients  entering the studies  could be  examined for  virologicresponses.53HIV  RNA was  extracted  from  500  µL  of  plasma  per  sample  using  a  NucliSENS  easyMAG(bioMérieux). Three independent one-step RT-PCR amplifcations were performed with 4µL of extractper amplifcation, followed by a second-round amplifcation using customised primers that includeda V3-specifc PCR primer and a multiplex barcode (for distinguishing between samples). All primersand are listed in Appendix III.3.2.2     Emulsion Polymerase Chain Reaction & PyrosequencingAfter PCR amplifcation, PCR amplicon concentrations were quantifed using a Quant-iT PicogreendsDNA Assay Kit (Invitrogen) and a DTX-880 Multimode Detector (Beckman Coulter). These werecombined  in  equal  proportions  (2x1012 DNA molecules/amplifcation),  purifed  with  AgencourtAmpure PCR Purifcation beads (Beckman Coulter), and re-quantifed. The purifed products werethen diluted to 2x105 molecules/mL, and combined at a ratio of 0.6 molecules to one emulsion PCR(emPCR)  microbead.  Oil  and  emPCR  buffer  components  were  shaken  with  a  TissueLyser(Qiagen/Retsch) to allow formation of microreactor micelles around the beads. After emPCR, thebeads were  washed and enriched for  DNA-coated  beads  as  per  the  manufacturer’s  instructions.These were added onto a picotitre plate at 2.5  × 105 beads in each of four regions, and underwentsequencing with a Genome Sequencer-FLX (Roche/454 Life Sciences). 3.2.3     Optimizing Bioinformatic Cutoffs for Deep SequencingBecause of the limited prior experience in the literature with deep sequencing, an attempt was madeto frst identify an optimal cutoff for both the geno2pheno and PSSM bioinformatic algorithms 335–337.Using a large dataset of 1875 samples (1827 for the PSSM analyses), Receiver Operating Characteristic(ROC) curves  338 were  generated to compare the  performance of deep sequencing relative  to theoriginal Trofle assay. A variety of FPR cutoffs were tested, ranging from 1.0 to 6.5 for geno2phenoand a variety of scores ranging from -1.0 to -6.5 for PSSM. 54The area under the curve (AUC) of the ROC curves was also calculated for each cutoff and plotted asshown in Figure 3.2. Cutoffs which optimized performance relative to the original Trofle assay wereselected.  For geno2pheno, this was a cutoff of 3.5, whereby sequences with FPRs above 3.5 wereclassifed as R5 (Figure 3.3: Panel 1). PSSM had an optimized cutoff of -4.75, whereby sequences withscores  less  than  -4.75  were  classifed  as  X4  (Figure  3.3:  Panel  2).  These  particular  cutoffs  forgenp2pheno and PSSM had maximal AUCs of 0.8911 (95% confdence interval: 0.8755 – 0.9067) and0.8915 (95% confdence interval: 0.8765 – 0.9065), respectively.Additionally,  the cutoff  for the percentage of non-R5 variants was explored in order to optimallypredict virologic outcomes. A random 75% of the dataset was used for training and exploration ofcutoffs,  and once these were established,  they were tested in a validation dataset comprising theremaining 25%. For this approach, virologic success was defned as a ≥2 log10 decline in viral loadfrom baseline and/or a viral load <50 copies/mL at week 8.  Figure 3.4 shows the results of theseanalyses.  A cutoff  of  2%  non-R5  variants  was  able  to  correctly  predict  virologic  success  withsensitivities of 84% for geno2pheno and 81% for PSSM.  This cutoff was also found to give the bestperformance in Chapter 2. 3.2.4     Bioinformatic Algorithms for Inferring Tropism from Genotypic DataDeep sequencing generated read-lengths of approximately 250 base pairs of data in each direction. Atypical V3 loop was 105 base pairs long (35 amino acids). Truncated reads (missing 4 or more bases ateither end of V3) were excluded from the analysis, as were samples producing fewer than 750 usablereads. Genotyping generated a mean of over 3000 V3 sequences per sample.  The tropism of eachsequence was interpreted by the PSSMX4/R5 or geno2pheno bioinformatic algorithms (optimized cut-offs for  non-R5:  PSSMX4/R5,  ≥-4.75;  geno2pheno,  ≤3.5).  The selection of these cutoffs is  detailed inFigures 3.2 – 3.4. The overall sample tropism was expressed as the proportion of non-R5 sequenceswithin the sample’s viral population. Patients with samples harbouring ≥2% non-R5 variants wereclassifed as having non-R5 HIV, while those with <2% were classifed as having R5 HIV.55Figure 3.2: Areas under Receiver Operating Characteristic Curves for Various Bioinformatic CutoffsFigure 3.2: Areas under Receiver Operating Characteristic Curves for Various Bioinformatic Cutoffs. The areas under the receiver operating characteristic curves(AUC) were calculated for various bioinformatic cutoffs, and their values are plotted here. The cutoffs which maximized the AUC were a geno2pheno false-positive rate cutoff of 3.5,and a PSSM score cutoff of -4.75. For display purposes in the fgure, the PSSM scores have been changed to their absolute values rather than negative values. Maximal AUCs indicatedthat the highest numbers of samples were classifed “correctly” using the original Trofle assay as a comparator. 56Figure 3.3: Receiver Operating Characteristic Curves for Optimal Geno2pheno & PSSM Cutoffs — Prediction of Original Trofile Assay ResultsFigure 3.3: Receiver Operating Characteristic Curves for Optimal Geno2pheno & PSSM Cutoffs — Prediction of Original Trofile Assay Results. The  optimalgeno2pheno cutoff was a false-positive rate of 3.5%. The optimal PSSM cutoff was a score of -4.75. The areas under the curve (AUC) and their 95% confdence intervals (95% CI) aredisplayed under their respective curves. The specifc point of 2% non-R5 variants is emphasized on each graph.57Figure 3.4: Receiver Operating Characteristic Curves for Optimal Geno2pheno & PSSM Cutoffs — Prediction of Week Eight Virologic SuccessFigure 3.4: Receiver Operating Characteristic Curves for Optimal Geno2pheno & PSSM Cutoffs — Prediction of WeekEight Virologic Success. Receiver  Operating  Characteristic  (ROC)  curves  for  predicting  week  8  virologicsuccess on maraviroc in the training dataset (upper two panels) and validation dataset (lower two panels). The areas under thecurve (AUC) and their 95% confdence intervals (95% CI) are displayed under their respective curves. The specifc point of 2%non-R5 variants  is emphasized on each graph. Performance for predicting virologic  success was  lower than performancepredicting original Trofle assay results, likely due to the fact that clinical outcomes are inherently more diffcult to predict thanassay results.  The number of samples included in the validation dataset was 684 for geno2pheno and 648 for PSSM. Thenumber of samples in the validation dataset was 206 for geno2pheno and 200 for PSSM. Clinical sensitivity to predict virologicsuccess was 84% for geno2pheno in the training set and 82% in the validation set. For PSSM these were 81% and 78%. Thespecifcities  (corresponding  roughly  to  correct  prediction  of  virologic  non-response)  were  lower  at  46%/30%  for  thegeno2pheno training/validation datasets, and 45%/35% for the PSSM training/validation datasets.58These cutoffs were established by optimizing to week 8 virologic response in a random 75% of thedataset, and testing on the remaining 25%, as shown in section 3.2.3 335–337. Results for all patients arepresented in the text. This 2% cutoff also approaches the likely level of reproducibility for PCR-basedmethods.3.2.5     Population-Based Sequencing as a ComparatorAdditional  second-round  PCR amplifcations  were  also  performed prior  to  standard  population-based sequencing on an ABI 3730XL DNA analyzer, according to previously-described methods  325.The cut-offs used were -4.25 for PSSMX4/R5 and 5.75 for geno2pheno 335,336. 3.2.6     Data AnalysisWhere clinical data were missing for patients enrolled in the clinical trials, the last observation wascarried forward, except for the analysis of the proportion of patients with pVL <50 copies/mL, wherea missing result  was considered >50 copies/mL. Data from MOTIVATE-1,  -2 and A4001029 werepooled, and both maraviroc arms were combined into a single group. Analyses were restricted topatients with tropism results from both assays. Clinical parameters examined included: the median change in log10-transformed HIV plasma viralload (pVL) from baseline; the proportion of patients with a pVL <50 HIV RNA copies/mL; and timeto a tropism switch. The performance of genotyping on these parameters was compared against thatof the original Trofle assay. The above results could not be extensively compared to ESTA responserates  in this  population,  as ESTA results  were not  available  from most  patients  in these  studies.However  a  subset  of  patients  with  ESTA results  is  examined in  the  results,  as  is  a  subset  withreplicate  deep  sequencing  results  performed  by an  independent  laboratory.  Comparisons  of  theperformance between laboratories were assessed through X-Y correlation and Bland-Altman plots.593.3     Results3.3.1     Tropism Screening by Deep Sequencing Relative to Trofile & Population-Based SequencingOverall, genotyping identifed 1037 samples (57%) as R5 and 790 (43%) as non-R5 using the PSSMX4/R5algorithm. PSSM and geno2pheno had approximately 90% concordance with one another. For ease ofpresentation,  results will  be shown for the PSSMX4/R5  algorithm. When screened with Trofle, 1141samples (62%) were called R5, and 686 (38%) were called non-R5 (Dual-/Mixed-tropic or X4). Globalconcordance of genotyping with Trofle-defned tropism was 82%. Of the 686 samples identifed as non-R5 by Trofle, genotyping agreed in 575 (84%) of cases. Anadditional 215 samples were further identifed as non-R5 by genotyping. Using Trofle as a reference,sensitivity of genotyping was 84%, and specifcity was 81%. Using genotyping as a reference, thesensitivity of original Trofle assay was 73%, and specifcity was 89%. Comparing population-basedagainst deep sequencing, overall concordance was 80%, with 64% sensitivity and 93% specifcity. Thesensitivity of both Trofle and population-based sequencing was lower when the proportion of non-R5 variants in the viral population was lower according to deep sequencing (Figure 3.5). Of interest,  deep sequencing detected at least some non-R5 HIV in >90% of patients (1700/1827),regardless of tropism classifcation. Samples with R5 HIV by Trofle had a median of 0.1% variantsidentifed as  non-R5 at  screening,  according to deep sequencing results.  However,  non-R5 levelsbelow 1-2% likely have low reproducibility and should be considered with caution. The Trofle non-R5 group excluding Dual/Mixed samples (i.e., only “pure” X4 by Trofle, N=39) had a median of 93%non-R5-virus. Patients screened R5 by Trofle but non-R5 by genotyping had 12% non-R5 HIV present at screening(N=167), much higher than where both assays indicated R5: 0.1% (N=926). This suggests that theoriginal  Trofle  assay  did  not  reliably  detect  patients  with  low level  non-R5  variants  present  atscreening (Figure 3.5). 60Figure 3.5: Sensitivity of Population-Based V3 Sequencing & Original Trofile Assay with Deep Sequencing as the ReferenceFigure 3.5: Sensitivity of Population-Based V3 Sequencing & Original Trofile Assay with Deep Sequencing as the Reference.  The ability of the population-based V3 sequencing (black bars) and original Trofle (grey bars) to identify screening samples as non-R5 that deep sequencing had identifed as having ≥ 2% non-R5 virus, stratifed by different proportions of non-R5 virus identifed in the deep sequencing result. Both alternative assays seemed to have decreased sensitivity for non-R5 HIV when such variants were present at lower proportions of the viral population. When non-R5 was present at 2-10% according to deep sequencing, 31% (69/224) & 58% (130/224) of samples were also called non-R5 by population-based sequencing & Trofle, respectively. These were 49% (57/116) & 57% (66/116) in the 10-20% group; 69% (81/117) & 77% (90/117) in the 20-40% group; 81% (118/145) & 81% (118/145) in the 40-80% group; and 96% (180/188) & 91% (171/188) in the >80% group.This was  consistent with the fnding that 8% of patients had non-R5 results at baseline despite R5Trofle results at screening 288, and consistent with ESTA results for the MERIT trial of maraviroc 290.Of the 1827 patients screened, 1093 actually entered the maraviroc (N=851) or placebo (N=242) armsof the trials.  Baseline characteristics of patient groups screened by both methods are presented inTable 3.1. The R5 groups by either method were similar in terms of baseline viral load and CD4+ cell61count, as were the non-R5 groups. Amongst maraviroc recipients, genotyping identifed over twice asmany patients than Trofle as being unlikely to respond to maraviroc (N=240 versus 111).Table 3.1: Baseline Characteristics of Treated Population, Stratified by Tropism Status by Genotype and PhenotypeGeno R5(N=775)Trofile R5(N=925)Geno non-R5 (N=318)Trofile non-R5(N=168)Baseline pVL, median log10 HIV-RNA copies/mL 4.85 4.88 5.04 5.07Median CD4 cell count,cells/mm3 177 168 72 54Median percent non-R5 variantsin deep sequencing screeningresult, % (IQR)0.1%(0 – 0.2%)0.1%(0 – 0.7%)19%(7-54%)28%(7-63%)Table 3.1: Baseline Characteristics of Treated Population, Stratified by Tropism Status by Genotype & Phenotype. Geno  non-R5,  identifed  as  having  non-R5  virus  by  genotyping;  Geno  R5,  identifed  as  having  R5  virus  bygenotyping; IQR, interquartile range; pVL, plasma viral load 3.3.2     Early Virologic Response to MaravirocScreening  genotype  was  a  predictor  of  response  to  maraviroc-based  antiretroviral  therapy  intreatment-experienced  patients.  Maraviroc  recipients  screened  with  R5  HIV  by  genotyping  hadconsistently better virologic outcomes than those screened non-R5. Using a number of parameters,genotyping  performed  similarly  to,  or  marginally  out-performed  the  original  Trofle  assay  inpredicting virologic response. Virologic performance was slightly better in the maraviroc BID armcompared  to  the  QD  arm  (data  not  shown),  but  these  arms  have  been  pooled  to  simplify  thepresentation of the results. The median pVL change from baseline to week 8 of treatment was examined in order to minimize thenumber of patients who had discontinued the study due to reasons such as treatment failure, or loss-to-follow-up, but with suffcient time to measure the effcacy of maraviroc in patients. Maravirocrecipients screened R5 by genotyping had a combined median week 8 decrease in pVL from baseline62of 2.4 log10 (Inter-quartile Range [IQR]: 1.7 – 2.9; N=611).  This was twice as large as the 1.4 log 10decline (IQR: 0.2 – 2.7; N=240) for patients classifed non-R5 by genotyping. Results where missing patients were censored, or where data were restricted to only those screenedfor MOTIVATE-1, were largely similar (data not shown). As mentioned in Section 3.2.3 on optimizingbioinformatic cutoffs, clinical sensitivity to predict virologic success on maraviroc was approximately80% when patients were screened with a 2% non-R5 variant cutoff (Figure 3.4).Using Trofle, the corresponding week 8 viral load declines were similar: 2.4 log 10 (IQR: 1.3 – 2.8;N=740)  for  R5 patients  versus  1.3  log10 (IQR:  0.3  –  2.7;  N=111)  for  non-R5 patients.  For  placeborecipients, the week 8 pVL declines were modest (0.5 – 0.8 log10) and similar regardless of genotypictropism. Median pVL responses on maraviroc and placebo over the course of the studies are shownin Figures  3.6A and 3.6B,  where  prediction of  virologic  outcomes  can be compared between thegenotypic and phenotypic assays.3.3.3     Longer-Term Virologic EfficacyThe effcacy of maraviroc was sustained to week 48 in patients identifed by genotyping as having R5HIV. The primary endpoint for the MOTIVATE trials was the percentage of patients with viral loads<50  copies/mL  at  week  48.  The  proportion  of  patients  achieving  virologic  suppression  <50copies/mL was assessed throughout the study for both the maraviroc and placebo arms. Maravirocrecipients with R5 HIV at screening were more likely to achieve a pVL <50 copies/mL at week 48compared to the non-R5 group. In total, 49% (301/611) of the R5 group, and 26% (62/240) of the non-R5 group had virologic suppression at week 48 when screened by genotyping. By Trofle, these were46% (337/740) and 23% (26/111), respectively (Figure 3.7).63Figure 3.6: Median Change in Plasma Viral Load from Baseline in the Maraviroc & Placebo ArmsFigure 3.6: Median Change in Plasma Viral Load from Baseline in the Maraviroc & Placebo Arms. Left:  Panel  A shows  the  responses  in  the  maraviroc  arms.  Patientsscreened as R5 by either genotyping or Trofle had much larger median pVL declines from baseline relative to patients screened as non-R5. Green and red lines correspond to deepsequencing R5 (N=611) and non-R5 (N=240) groups respectively, while solid black and dotted black lines correspond to Trofle R5 (N=740) and non-R5 (N=111) groups. Right: Panel Bshows the pVL declines from baseline for patients receiving placebo were similar to MVC-receiving patients identifed as non-R5 or non-R5, and were small regardless of screeningtropism or assay used. Green and red lines correspond to deep sequencing R5 (N=164) and non-R5 (N=78) groups respectively, while solid black and dotted black lines correspond toTrofle R5 (N=185) and non-R5 (N=57) groups.  64Figure 3.7: Percentage of Patients with Viral Loads Less than 50 HIV RNA Copies/mL in the Maraviroc & Placebo ArmsFigure 3.7: Percentage of Patients with Viral Loads Less than 50 HIV RNA Copies/mL in the Maraviroc & Placebo Arms. Left: Panel A shows the proportion with virologicsuppression in the maraviroc arms. A higher proportion of maraviroc recipients screened by either method as R5 had a pVL <50 HIV RNA copies/mL compared to the non-R5patients. Green and red lines correspond to deep sequencing R5 (N=611) and non-R5 (N=240) groups respectively, while solid black and dotted black lines correspond to Trofle R5(N=740) and non-R5 (N=111) groups. Right: Panel B shows the proportion with virologic suppression in the placebo arms. Green and red lines correspond to deep sequencing R5(N=164) and non-R5 (N=78) groups respectively, while solid black and dotted black lines correspond to Trofle R5 (N=185) and non-R5 (N=57) groups.65The  genotypic  non-R5  group  could  be  divided  roughly  in  half,  with  127  patients  having  low-prevalence (2-20%) non-R5, and 113 having >20% non-R5 virus. The group of patients with 2-20%non-R5 according to deep sequencing had minority non-R5 variants that were not reliably detectedby standard population-based sequencing methods (Figure 3.3). Importantly, this group of patientshad a poor response to maraviroc, with 27% (34/127) of patients achieving virologic suppression atweek 48, similar  to the non-R5 group as a whole (26%) and to patients with >20% non-R5 (25%;28/113).  The virologic responses of placebo recipients  were similarly low to maraviroc recipientsidentifed as having non-R5 HIV, ranging from 17-23% depending on tropism or assay. Interestingly, the virologic outcomes of maraviroc recipients showed a general inverse relationshipwith the percentage of non-R5-virus present at screening according to genotyping. Patients with 0%non-R5 had the best success, showing a week 8 log10 pVL decline of 2.6, with 65% (58/89) of patientshaving week 48 virologic suppression. Patients with between 0 and 1% non-R5 had slightly pooreroutcomes: week 8 pVL log10  decline of 2.4, 48% (234/491) with virologic suppression. This declinedagain in patients with between 1-2% non-R5: 2.1 log10, 29% (9/31). Patients with >2% non-R5 (i.e., thegenotyping non-R5 group) all showed similar low virologic responses, as detailed above.3.3.4     Changes in Viral TropismAs a separate endpoint, patients were analysed according to whether they experienced a change intheir Trofle result from R5 to non-R5 (a tropism “switch”) over the course of the studies, as measuredby Kaplan-Meier analysis. This parameter is both clinically relevant for maraviroc-based therapy, andfunctioned as a measure separate from changes in viral load measurements. Amongst those patientsoriginally screened as R5 by Trofle, those identifed to be non-R5 by genotyping were almost twice aslikely to have non-R5 HIV emerge by week 24 compared to patients screened as R5 by both methods(Figure 3.8). A total of 40% (72/180) of maraviroc recipients who switched tropism were identifed bygenotyping as having ≥2% non-R5 virus. Tropism switches occurred in 18% (111/612) of the groupcalled R5 by genotyping, lower than in those called R5 by Trofle alone: 25% (180/724).  66Figure 3.8: Time to Change in Tropism from R5 to Dual/Mixed or X4 in the Maraviroc & Placebo ArmsFigure 3.8: Time to Change in Tropism from R5 to Dual/Mixed or X4 in the Maraviroc & Placebo Arms. Left: The change in tropism in the MVC arms where all patients were R5 at screening by Trofle and switched tropism to DM or X4 over the course of the studies according to the Trofle assay. Green and red lines correspond to genotyping R5 (N=605) and non-R5 (N=135) groups respectively, while the solid black lines correspond to the Trofle R5 (N=740) group.  Right: The change in tropism in the placebo arms for patients with Trofle R5 results at screening. Green and red lines correspond to genotyping R5 (N=153) and non-R5 (N=32) groups respectively, while the solid black lines correspond to the Trofle R5 (N=185) group.67Amongst patients who switched tropism, maraviroc recipients classifed as R5 by Trofle but non-R5by genotyping had been on treatment for a mean of 4.6 weeks before Trofle gave a non-R5 result,over twice as quickly (9.7 weeks) as where both tests indicated R5 (Figure 3.8).3.3.5     Response Stratified by Background Drug ActivityPatients were also classifed according to a weighted optimized background therapy susceptibilityscore (wOBTss). In general, wOBTss was defned as the number of active drugs in the backgroundregimen at baseline, with nucleoside reverse transcriptase inhibitors scoring 0.5 339. Genotyping waspredictive  of virologic  success on maraviroc-based therapy regardless of  wOBTss.  Maraviroc wassuccessful in either of the R5 groups where the wOBTss was between 1 and 2.  The proportions ofthese patients with a week 48 pVL <50 copies/mL were 58% (179/311) and 53% (205/389) whenscreened by genotyping and Trofle,  respectively.  The  predictive  ability  of  genotyping  was morepronounced at more compromised background regimens. The proportions with undetectable viralloads  were  33%  (81/232)  versus  29%  (78/271)  of  R5-classifed  patients  with  wOBTss  <1  bygenotyping or Trofle, respectively (Figure 3.9).3.3.6     Discordance Amongst Bioinformatic AlgorithmsThere was a high degree of concordance using alternative bioinformatic algorithms. Geno2pheno andPSSMSINSI were compared to PSSMX4R5 in terms of their ability to predict various virologic outcomes.Detailed  analyses  of  these  comparisons  are  presented  in  Tables  3.2  –  3.4.  In  general,  where  thealgorithms disagreed on tropism classifcations, the virologic responses were intermediate betweenthe  concordant  R5  and  concordant  non-R5  groups.  This  indicated  that  no  algorithm  clearlyoutperformed the others.  68Figure 3.9: Median Change in Plasma Viral Load from Baseline in Patients with R5 Virus Stratified by Their Weighted Optimized Background Therapy Susceptibility Score (wOBTss)Figure 3.9: Median Change in Plasma Viral Load from Baseline in Patients with R5 Virus Stratified by Their Weighted Optimized Background Therapy Susceptibility Score(wOBTss). Maraviroc-treated patients screened as R5 by genotyping (green lines) or Trofle (black lines). Patients screened as R5 by either method who also hadwOBTss > 2 (N=68 or 80, respectively) showed the largest pVL declines from baseline. Patients with wOBTss 1-2 (N=151 or 197) showed intermediate pVL decline and patients withwOBTss≤1 (N=392 or 463) showed poorer changes in pVL. The wOBTss > 2, 1-2, and ≤ 1 groups are indicated by thin, intermediate, and thick lines, respectively.69Table 3.2: Discordance amongst Bioinformatic Algorithms in the Maraviroc Arms — PSSMX4/R5 versus Geno2phenoTable 3.2: Discordance amongst Bioinformatic Algorithms in the Maraviroc Arms — PSSMX4/R5 versus Geno2pheno. Virologic outcomes for patients with discordant results between algorithms. PSSMX4R5 (non-R5 if ≥2% scored ≥-4.75)versus geno2pheno (non-R5 if ≥2% scored ≤3.5)Table 3.3: Discordance amongst Bioinformatic Algorithms in the Maraviroc Arms — PSSMX4/R5 versus PSSMSI/NSIConcordance orDiscordancePSSMX4R5/PSSMSINSI NWeek 8 log10pVL decline(IQR)%<50c/mL @week 48(N)% Changingtropism (N)R5/R5 557 2.43(1.78 – 2.86)49%(275/557)17%(92/553)R5/Non-R5 52 2.34(0.87 – 2.78)46%(24/52)38%(19/50)Non-R5/R5 17 2.44(0.47 – 3.05)24%(4/17)27%(4/15)Non-R5/Non-R5 217 1.27(0.21 – 2.63)26%(56/217)58%(70/120)Table 3.3: Discordance amongst Bioinformatic Algorithms in the Maraviroc Arms — PSSMX4/R5 versus PSSMSI/NSI. Virologic outcomes for patients with discordant results between algorithms. PSSMX4R5 (non-R5 if ≥2% scored ≥-4.75)versus PSSMSINSI (non-R5 if ≥2% scored ≥-3.5)Concordance or DiscordancePSSMX4R5/geno2pheno NWeek 8 log10pVL decline(IQR)%<50c/mL @week 48(N)% Changingtropism (N)R5/R5 583 2.43(1.79 – 2.86)50%(289/583)17%(97/577)R5/Non-R5 26 2.06(0.33 – 2.59)38%(10/26)54%(14/26)Non-R5/R5 55 2.19(0.68 – 2.79)36%(20/55)26%(12/46)Non-R5/Non-R5 179 0.98(0.12 – 2.61)22%(40/179)70%(62/89)70Table 3.4: Discordance amongst Bioinformatic Algorithms in the Maraviroc Arms — PSSMSI/NSI versus Geno2phenoConcordance or DiscordancePSSMSINSI/geno2pheno NWeek 8 log10pVL decline(IQR)%<50c/mL@ week 48(N)% Changingtropism(N)R5/R5 559 2.43(1.76 – 2.86)49%(273/559)16%(91/554)R5/Non-R5 15 2.39(1.48 – 2.70)40%(6/15)36%(5/14)Non-R5/R5 79 2.35(1.56 – 2.79)46%(36/79)26%(18/69)Non-R5/Non-R5 190 0.98(0.11 – 2.60)23%(44/190)70%(71/101)Table 3.4: Discordance amongst Bioinformatic Algorithms in the Maraviroc Arms — PSSMSI/NSI versus Geno2pheno. Virologic outcomes for patients with discordant results between algorithms. PSSMSINSI (non-R5 if ≥2% scored ≥-3.5)versus geno2pheno (non-R5 if ≥2% scored ≤3.5)3.3.7     Assay DiscordanceWhere screening assays differed, virologic outcomes on maraviroc slightly favoured the genotypingresults. Amongst the discordant patients, where Trofle indicated R5 but genotyping identifed >2%non-R5  virus  (N=135),  median  log10 pVL declines  were  lower,  at  1.8  log10.  For  comparison,  theconcordant non-R5 group (N=105) had 1.2 log10 pVL decline. Where genotyping screened patients ashaving R5 virus, but Trofle screened as non-R5, the median week 8 pVL decline was 2.6 log10 (N=6),similar to the concordant R5 group: 2.4 log10 (N=605). When deep sequencing was compared to population-based sequencing,  virologic  outcomes againfavoured the deep sequencing results. The results over 48 weeks of treatment, with patients groupedby assay concordance or  discordance are  shown in Figure 3.10.  Furthermore,  there  were a  largenumber of patients with 50% or lower non-R5 variants, which could potentially be diffcult to detectby standard population-based sequencing. Importantly, these patients had poor virologic outcomeson maraviroc compared to those with low non-R5 prevalence of <2% (Figure 3.11).71Figure 3.10: Where Assay Results Were Discordant, the Virologic Responses Tended to Favour Deep Sequencing ResultsFigure 3.10: Where Assay Results Were Discordant, the Virologic Responses Tended to Favour Deep Sequencing Results. Panel  A  shows  virologic  responses  tomaraviroc  for patients  stratifed by whether deep sequencing and the original  Trofle assay gave concordant or  discordant results.  The turquoise  line  represents  patients  withconcordant R5 results (N=605), the orange line represents concordant non-R5 results (N=105). The dashed black line represents patients where deep sequencing indicated R5 but Trofleindicated non-R5 (N=6), and the dashed grey line represents patients with non-R5 by deep sequencing but R5 by Trofle (N=135). Panel B shows virologic responses to maraviroc forpatients stratifed by whether deep sequencing and population-based sequencing gave concordant or discordant results. The turquoise line represents patients with concordant R5results (N=573); the orange line represents concordant non-R5 results (N=127). The dashed black line represents patients where deep sequencing indicated R5 but population-basedsequencing indicated non-R5 (N=38), and the dashed grey line represents patients with non-R5 by deep sequencing but R5 by population-based sequencing (N=113). Overall, thediscordant lines tended to favour the deep sequencing classifcations, especially when compared to population-based sequencing.72Figure 3.11: Virologic Response of Maraviroc Recipients as a Function of the Proportion of Non-R5 HIV at Screening. Figure 3.11: Virologic Response of Maraviroc Recipients as a Function of the Proportion of Non-R5 HIV at Screening. The  median  plasma  viral  load  change  frombaseline is shown for groups of treatment-experienced maraviroc recipients. Patients are grouped according to the percentage of non-R5 HIV detected at screening by deep sequencingusing a geno2pheno false-positive rate of 3.5. Patients with less than 2% non-R5 HIV had a median pVL decline from baseline of approximately 2.5 log10 copies/mL at week 48. Patientswith greater than 2% non-R5 all had similarly poor responses to maraviroc, with a median pVL decline of approximately 1 log 10. Patients with 2-20% non-R5 are less likely to have non-R5 detected using population-based sequencing, but still had poor virologic responses. This fgure is adapted with permission from Swenson et al, Current Opinion in HIV and AIDS.2012; 7(5): 478-485. © 2012 Wolters Kluwer Health.73Having  either  or  both  assays  indicate  non-R5  was  a  poor  prognostic  indicator  of  longer-termmaraviroc response. At week 48, the proportions of patients with suppressed viraemia were: 27%(36/135) for the Trofle R5_geno non-R5 group and 0% (0/6) for the Trofle non-R5_geno R5 group.Week 48 suppression was twice as high in the concordant-R5 group versus the concordant-non-R5group: 50% (301/605) versus 25% (26/105), respectively. In the Trofle-R5_geno-non-R5 group, 55%(74/135) of maraviroc recipients changed tropism, much higher than the concordant-R5 group: 18%(111/605).  Most patients screened by Trofle as non-R5 remained so over the course of the studyperiod, regardless of concordance with genotyping.3.3.8     Comparison with Independent Replication by an External Laboratory In order to assess the reproducibility of this method, a subset of 310 samples from this dataset werealso processed by an independent laboratory (Quest Diagnostics Nichols Institute, California).  Forthis comparison, the geno2pheno algorithm was used to infer tropism rather than PSSMX4/R5. This dataset was selected to represent an unbiased subpopulation of enrollees into the MOTIVATEand A4001029 trials  340,341. This was achieved by restricting to those patients entering the maravirocarms of MOTIVATE and A4001029 until the latter study had completely enrolled. Since at that point,patients with non-R5 original Trofle assay results were no longer able to enroll into a trial, then byrestricting to this subset of patients, the population is not biased by initial tropism screening. Inter-laboratory concordance was excellent between the two implementations of deep sequencing,with 92% of samples receiving the same tropism classifcation (Kappa=0.84). The percentage of non-R5 variants detected by the laboratories was also highly correlated in the samples, with a Pearson’scorrelation coeffcient of 0.92, and a median difference of only 0.3% (interquartile range: 0-3%) (Figure3.12). Bland-Altman analysis demonstrated that there was limited bias in the percentage of non-R5variants detected by either laboratory (Figure 3.13). 74Figure 3.12: Correlation of Deep Sequencing Between Two Independent LaboratoriesFigure 3.12: Correlation of Deep Sequencing Between Two Independent Laboratories. The  correlation  between  bothlaboratories in terms of the percentage of non-R5 variants was high (R2=0.85). Points are marked by whether they representconcordant or discordant results between laboratories (see key). The line of best-ft for the linear regression is shown as a blackline, and bordered by the 95% confdence intervals, shown with dashed lines. The slope of the line of best-ft was signifcantlynon-zero (p<0.0001). 75Figure 3.13: Bland-Altman Plot Comparing the Assay Results from Two Independent Laboratories Figure 3.13: Bland-Altman Plot Comparing the Assay Results from Two Independent Laboratories. There  was  minimal  bias  in  the  results  between  the  laboratories  according  to  the  Bland-Altman  analysis(bias=0.048%). A total of 95% of the data points were within ±22% of one another (dashed lines represent the 95% limits ofagreement). 3.3.9     Comparison to the Enhanced Sensitivity Trofile AssayOf the 310 samples processed by the two independent laboratories, 294 also had results from theEnhanced  Sensitivity  Trofle  Assay  (ESTA)  at  screening.  Because  ESTA has  replaced  the  originalTrofle assay, the performance of deep sequencing was compared to ESTA. Relative to ESTA, deep76sequencing had 76% sensitivity (87 of 114 samples), 89% specifcity (160 of 180 samples) and 82%concordance (247 of 294 samples). Furthermore,  both  assays  predicted  subsequent  virologic  outcomes  on  maraviroc  equally  well.Patients classifed with R5 HIV had plasma viral load declines from baseline to week 8 of 2.4 log 10copies/mL and 2.4 log10 copies/mL by ESTA and deep sequencing, respectively. Those with non-R5HIV at screening had week 8 declines of 0.9 log10 copies/mL and 0.9 log10 copies/mL, respectively(Figure 3.14). Where assays gave discordant results, virologic outcomes were intermediate betweenthe concordant R5 and non-R5 groups, with neither assay seeming to give superior predictions to theother (Figure 3.15).3.4     Discussion & ConclusionsThis  study  represents  one  of  the  largest  clinical  applications  of  next-generation  sequencingtechnology  to  date.  Deep  V3  sequencing  was  able  to  detect  and  quantify  low  prevalence  sub-populations of CXCR4-using HIV within a large set of clinical isolates. This method was predictive ofvirologic  response  to  a  maraviroc-containing  regimen  and  matched  or  surpassed  the  predictiveability of the original Trofle assay on a number of parameters, including the proportion of patientsachieving  a  plasma  viral  load  <50  copies/mL,  and  the  likelihood  of  switching  tropism  whilereceiving maraviroc.Retrospectively screening with genotyping led to over twice as many maraviroc recipients  beingidentifed with non-R5 HIV. Trofle-R5 patients screened as non-R5 by genotyping were more likely tochange their Trofle result to Dual/Mixed or X4 during the trials, suggesting earlier non-R5 detectionby  deep sequencing versus  the  original  Trofle  assay.   Of  maraviroc  recipients  who  experiencedtropism switches, genotyping would have identifed 40% as non-R5. Thus, the high sensitivity ofdeep sequencing was able to account for a substantial portion of tropism switches as being due to thepresence of low-level non-R5 variants that were not detected by Trofle.77Figure 3.14: Similar Predictions of Plasma Viral Load Changes on Maraviroc by Deep Sequencing & the Enhanced Sensitivity Trofile AssayFigure  3.14:  Similar  Predictions  of  Plasma Viral  Load  Changes  on  Maraviroc  by  Deep  Sequencing  &  the  EnhancedSensitivity Trofile Assay. The median viral  load changes from baseline are  shown for  maraviroc  recipients,stratifed by their deep sequencing genotype results or Enhanced Sensitivity Trofle Assay phenotype results. The median viralload change for patients classifed as having R5 HIV by deep sequencing is shown as a green line, and the patients with R5 byESTA are shown with a solid black line.  Patients with non-R5 results by genotype or phenotype are shown with red lines ordashed black lines, respectively. 78Figure 3.15: Intermediate Plasma Viral Load Changes Where Deep Sequencing & the Enhanced Sensitivity Trofile Assay Gave Discordant ResultsFigure 3.15: Intermediate Plasma Viral Load Changes Where Deep Sequencing & the Enhanced Sensitivity Trofile AssayGave Discordant Results. Virologic responses to maraviroc for patients stratifed by whether deep sequencingand the Enhanced Sensitivity Trofle assay (ESTA) gave concordant or discordant results. The turquoise line represents patientswith concordant  R5 results  (N=160);  the  orange line  represents  concordant non-R5 results  (N=97).  The dashed black linerepresents  patients  where  ESTA indicated  R5  but  deep  sequencing  indicated  non-R5  (N=20),  and  the  dashed  grey  linerepresents patients with non-R5 by ESTA but R5 by deep sequencing (N=17). Where the assays gave discordant results, neitherassay seemed to have consistently superior performance to the other. 79The emergence of minority non-R5 variants detected by deep sequencing but not Trofle has beenshown  previously  following  treatment  with  maraviroc  232.  In  the  MERIT  trial  of  maraviroc  intreatment-naïve  patients,  low X4  sensitivity of  the  original  Trofle assay was determined to be aprimary  reason  that  maraviroc  demonstrated  inferiority  to  efavirenz  290.  When  this  trial  wasretrospectively re-analysed with ESTA, more patients were identifed as harbouring non-R5 virus 290.A potential beneft of deep sequencing over standard population-based sequencing is that the lattercannot reliably detect variants present below approximately 20% of the viral population 180,228 , whiledeep sequencing can reliably detect quasi-species present at much lower levels, as shown here. Only37% of maraviroc recipients with 2-20% non-R5 variants were identifed by population-based V3sequencing (and only 58% were identifed by Trofle),  yet  these patients still  showed suboptimalvirologic responses. Thus, like ESTA, deep sequencing may represent an enhanced sensitivity tropismtest, able to detect minority non-R5 variants. However, the added clinical beneft of capturing low-prevalence non-R5 variants should be weighed against the accessibility and relative affordability ofstandard sequencing. Importantly, population-based sequencing had >80% concordance with deepsequencing in this same dataset.Most patients called R5 by genotyping still had low levels (<2%) of detectable non-R5 virus. Despitethis,  good virologic  responses  to  maraviroc  were  seen  in  this  population.  It  is  possible  that  thebackground antiretrovirals were able to suppress the non-R5 variants in these patients. Alternatively,it may be the case that a minimum threshold of non-R5 HIV must be surpassed before treatment withCCR5  antagonists  is  compromised.  The  data  reported  here  may  indicate  that  this  threshold  isapproximately 2% of the viral population. The activity of the background regimen is also likely amajor factor.Some limitations of this study and the deep sequencing method in general should be acknowledged.A major  limitation  is  the  relatively  high cost  of  deep  sequencing.  Also,  the  labour  in  preparingsamples  for  pyrosequencing is  intensive  and complex.  Third,  there  may be  correlates  of  tropism80outside V3 218,342,343, which this method is unable to capture. This study was retrospective in nature,and a  randomized  clinical  trial  with  exclusion  of  patients  screened non-R5  by  deep sequencinginstead of Trofle may have yielded different results. However, the results in an unbiased subset ofpatients  presented  in  Sections  3.3.8  and  3.3.9  suggest  that  the  performance  of  deep  sequencingremained excellent even without pre-screening.Surprisingly,  there  are  limited  ESTA data  available  for  these  studies.  This  is  a  concern,  sincemaraviroc is primarily prescribed for treatment-experienced patients yet ESTA, the most commonlyused tropism assay,  has  not  been  formally  validated  in  MOTIVATE.  This  is  a  limitation  for  theinterpretation of the results of the current study, however this and other studies have found goodconcordance between these methods 340,341. The pre-screening of patients with Trofle also limited thenumber of treated non-R5 patients examined in this study. These limitations coupled with the goodperformance of the above detailed method support a prospective trial evaluating the relative meritsof genotypic and phenotypic approaches. Such a trial would also establish the true sensitivities andspecifcities of both approaches 344.Overall, despite the study’s limitations, deep sequencing showed good performance in predicting avariety  of  clinical  parameters  including  viral  load  declines,  likelihood  of  achieving  virologicsuppression, and time to a tropism switch. This large study establishes deep sequence analysis of theHIV  envelope  V3  loop  as  an  extremely  promising  tool  for  identifying  treatment-experiencedindividuals who could receive clinical beneft from CCR5 antagonist-containing therapy regimens. The  above  chapter  exclusively  examined next-generation  sequencing  using  RNA amplifed fromblood plasma of HIV-infected patients. In a number of settings, however, current HIV RNA is notavailable due to the success of antiretroviral therapy at suppressing plasma viraemia 345. HIV exists inthe cells of its host as both integrated and cell-associated (e.g., episomal) DNA. Therefore, there areefforts aimed at evaluating the utility of cellular HIV DNA for determining HIV coreceptor usage214,344,346,347. While this approach was introduced briefly in Chapter 2, it is expanded on and examinedmore in-depth in the following chapter. 81Chapter 4:     Use of Cellular HIV DNA to Predict Virologic Responses to Maraviroc: Performance of Population-Based & Deep Sequencing 4.1     Background & IntroductionRecent advances in HIV treatment and curative strategies have led to the need for sensitive andaccurate HIV tropism assays. The CCR5 antagonist antiretroviral drug class, including maraviroc 288and others  111,318,  are most  successful  when used in patients with solely CCR5-using (R5) HIV, asdetermined  with  phenotypic  290,291 or  genotypic  tropism  assays  325,348,349.  Additionally,  fledglingattempts at establishing long-term remission from HIV disease have been developed  350,351, such asusing zinc-fnger nucleases for disruption of the CCR5 gene. These are related to the successful cureby stem cell transplantation from a  CCR5 Δ32-homozygous  156 donor to an HIV-infected patient  352.These  curative  approaches  will  probably  require  pre-screening  with  tropism  assays  to  identifycandidate patients with exclusively R5 HIV. This is because a viral population which uses CXCR4would likely be unaffected by reducing CCR5 protein levels.Sustained suppression of plasma viraemia with advances in antiretroviral therapy improves patientoutcomes  345 but  precludes resistance  and/or tropism testing from plasma due to low HIV copynumbers. Given the impracticalities and clinical consequences of treatment interruptions 353, tropismtesting from HIV RNA in successfully treated patients is not possible. Still, some patients may wish toincorporate  maraviroc into an already successful  antiretroviral regimen to manage side effects orsimplify the regimen. For such patients, a more feasible approach may be tropism testing from HIVDNA. HIV DNA is the product of successful infection of cells by HIV 214,333. Tropism testing from HIV DNAinvolves PCR amplifcation of a portion of the envelope gene followed by phenotypic testing in celllines or genotypic testing by sequencing. Generally, the V3 loop of HIV gp120 is the main target of82such approaches. Tropism testing from HIV DNA can allow patients to switch a component of theirantiretroviral regimen to a CCR5 antagonist without an interruption to their existing treatment. The  aim  of  this  study  was  to  assess  the  performance  of  cell-based  genotypic  tropism  testingapproaches  in  a  large  group  of  patients  entering  three  clinical  trials  of  the  CCR5  antagonist,maraviroc.  All  patients  were  viraemic  at  the  baseline  testing  visit,  allowing  parallel  testing  andcomparison of  plasma and cell-based  tropism assays,  as  well  as  examination  of  actual  virologicoutcomes to the medication. Both population-based and deep sequencing approaches were applied inboth compartments, giving a total of four different genotypic tropism tests. The abilities of these fourmethods to predict subsequent virologic response to maraviroc were compared with each other, andalso with the phenotypic, plasma-based original Trofle assay at the same time point.4.2     Materials & Methods4.2.1     Samples & Patient CompositionPeripheral blood mononuclear cell (PBMC) samples were obtained at baseline from 181 maravirocrecipients in the MOTIVATE-1 (N=48), MOTIVATE-2 (N=48) and A4001029 (N=85) studies 288,291. Notethat these samples were deliberately selected to include a large proportion of non-R5 Trofle results(N=89,  49%).  An approximate 1:1  ratio  of  R5 to non-R5 Trofle  results  mitigated any populationskewing through over-enrichment for CCR5-tropic samples — a criticism of past studies 354. A total of 156 (86%) had matching tropism results from plasma available. The baseline time point wasday 0  of  treatment  with  maraviroc.  All  participants  were  antiretroviral  therapy-experienced  andreceived at least one dose of maraviroc (once or twice daily) plus an optimized background therapyduring the trials. Patients were screened and their plasma samples were periodically tested while on-treatment using the original Trofle assay. Results using the enhanced sensitivity Trofle assay (ESTA)or DNA-based Trofle assay were not available. 834.2.2     V3 Amplification & SequencingV3 amplifcation and sequencing were performed by similar methods to those previously publishedand reported in the above chapters 348,349. Briefly, triplicate nested RT-PCR was used to amplify the V3region from HIV RNA in plasma. For HIV DNA, 500 µL of PBMC samples  were extracted withautomated  methods,  followed  by  triplicate  nested  PCR  targeting  V3.  Deep  sequencing  with  aRoche/454  Life  Sciences  Genome  Sequencer  FLX  was  performed  using  the  second-round  PCRproducts, which had multiplex tags, allowing 48 samples to be sequenced in each direction per run. The median read depths obtained were 2799 reads per DNA sample (interquartile range [IQR]: 2057 –3623)  and 2088 (IQR:  1783 –  2579)  reads per  RNA sample.  A description  of  the  data  processingpipeline  for  deep  sequencing  is  included  in  Appendix  IV.  In  addition,  a  second  round  PCRamplifcation was also performed using the same triplicate amplifed template. These PCR productsunderwent standard, population-based sequencing on an ABI 3730 XL DNA analyzer according topreviously described methods 325.4.2.3     Tropism PredictionThe geno2pheno algorithm generates a false positive rate (FPR) for each input sequence  204. Thosescoring above a certain pre-selected cutoff are classifed as R5. The false positive rate (FPR) cutoff forgeno2pheno  tropism  assignments  was  set  previously  325,336,348.  Optimization  of  these  cutoffs  wasperformed  using  a  random  75%  of  plasma  screening  samples  from  the  maraviroc  treatment-experienced  trials,  and  was  validated  on  the  remaining  25%.  Cutoffs  were  chosen  in  order  todistinguish maximal differences between early response and non-response to maraviroc at week 8 oftreatment. The maximum percentage of non-R5 variants allowed for a sample to be classifed as R5was also optimized and validated in a similar manner 335. A sample was considered R5 if the lowest of three population-based V3 sequences had a geno2phenoFPR cutoff greater than 5.75 325. For deep sequencing, a sample was considered R5 if fewer than 2% ofthe variants detected fell below an FPR of 3.5 348. Population-based sequencing required a higher FPR84cutoff than deep sequencing likely due to the reduced sensitivity of population-based sequencing todetect minority variants. There was an additional exploration of alternative geno2pheno cutoffs in thecurrent study.4.2.4     Ethics StatementWritten,  informed  consent  was  obtained  from all  individuals,  including  consent  to  allow  othertropism testing to be performed on their samples. The University of British Columbia-ProvidenceHealth Care Research Ethics Board reviewed the research project and granted ethical approval. Alldata were analyzed anonymously.4.2.5     Data AnalysisPatients were grouped according to the R5 or non-R5 result by each tropism assay. Concordance wascalculated from the number of samples with identical tropism calls by any two assays. The periodfrom baseline to week 24 was examined for all patients.  Each assay was assessed in its ability topredict responses to maraviroc plus optimized background therapy. Patients classifed as having R5HIV  would  be  expected  to  have  larger  virologic  responses  to  maraviroc  compared  to  patientsclassifed as having non-R5 HIV. Within  compartments  (plasma  or  PBMCs),  data  were  restricted  to  samples  with  results  by  allavailable assays; this was 181 PBMC samples and 156 plasma samples. When the two compartmentswere compared, analyses were restricted to the 156 samples with results by all fve assays. Tests forstatistical signifcance included the Mann-Whitney U test for comparisons of median pVL declines onmaraviroc, the Fisher’s exact test for comparisons of the proportion of patients achieving virologicresponses  at  weeks 8  and 24,  and the  log-rank test  for  differences  in  median time  to change  inphenotypic tropism. 854.3     Results4.3.1     Prediction of Virologic Efficacy on MaravirocMatched plasma and PBMC baseline  samples  were  assessed by two genotypic  methods and theoriginal Trofle assay in plasma, giving a total of fve tropism assays to compare. Both the short-term(to 8 weeks) and long-term (to 48 weeks) virologic effcacy of maraviroc in patients were assessed asprimary analyses. Patients were deemed to be correctly classifed as having R5 HIV if they werevirologic responders to maraviroc-based therapy. Patients were stratifed by whether they had R5 ornon-R5 results by each of the fve assays at baseline. Generally, these fve tropism methods were allsimilarly  predictive  of  virologic  responses  to  the  study  medication,  regardless  of  the  specifcapproach or compartment (Figure 4.1).Short-term virologic responses by week 8 were examined, as in a previous study of deep sequencing348.  A response  to  maraviroc-based  therapy  was  defned  as  a  plasma  viral  load  decline  ≥2  log 10copies/mL from baseline to week 8, or having an undetectable viral load at week 8. Odds ratios ofsuccess for groups identifed as R5 versus non-R5 ranged from 3.2 for deep sequencing in PBMCs to9.4 for deep sequencing in plasma.At week 24, the percentage of patients with undetectable viraemia ranged from 42-47% amongst theR5 groups, which was 16-22% higher than the non-R5 groups. By all assays, patients with R5 HIV hadapproximately 2 times the odds of virologic suppression at week 24 compared to those with non-R5.Virologic success of R5 groups was statistically signifcantly higher than non-R5 groups by all assaysat  week  8.  These  groups  also  had  signifcantly  different  virologic  suppression  by  week  24,  asclassifed by all assays (p<0.05), except the Trofle assay (which had a trend towards signifcance)(Table 4.1).  86Figure 4.1: Percentage of Patients with Plasma Viral Loads below 50 Copies/mL Was Similar by All Cellular or Plasma-Based MethodsFigure 4.1: Percentage of Patients with Plasma Viral Loads below 50 Copies/mL Was Similar by All Cellular or Plasma-Based Methods. Panel A shows the percentageof patients classifed with R5 HIV at baseline who had undetectable plasma viral loads over 24 weeks of treatment with maraviroc. Panel B shows the same but for patients classifedwith non-R5 HIV. Results are restricted for all groups to the 156 patients with baseline tropism results by all fve methods. Solid lines represent the DNA-based assays, dashed linesrepresent the RNA-based assays, and the Trofle assay results are represented with dotted lines. DeepSeq — deep sequencing; PopSeq — population-based sequencing; pVL — plasmaviral load87Table 4.1: Short- & Long-Term Virologic Responses to Maraviroc as Predicted by All Tropism Assays in the Full DatasetTable 4.1: Short- & Long-Term Virologic Responses to Maraviroc as Predicted by All Tropism Assays in the Full Dataset. The plasma viral load (pVL) responsesto maraviroc plus OBT are shown for patients classifed as having R5 and non-R5 HIV by all 5 tropism methods. The median pVL decline from baseline to week 8 is shown as well asthe percentage of patients with a week 8 virologic response — defned as a pVL decline ≥2 log 10 or a pVL of <50 copies/mL at week 8. The percentage of patients with undetectableviraemia is also shown. The odds ratios of week 8 and week 48 virologic responses are also displayed. This table is restricted to the 156 samples which had results for all assays.884.3.2     Similar Performance Regardless of Background Regimen ActivityThese  analyses  were  re-examined  in  a  subset  of  81  patients  who  had  compromised  treatmentbackground activity, and for whom maraviroc would be expected to have the largest impact. Thisapproach can best distinguish differences in assay performance by minimizing the contribution ofbackground antiretroviral agents to the activity of maraviroc. These patients, who had a weighted optimized background therapy susceptibility score (wOBTss) ≤1,had similar responses to the overall study group. In fact, the difference between the R5 and non-R5groups was exaggerated in this subset. The patients classifed as having R5 HIV by any of the tropismassays had much higher responses to maraviroc than did those with non-R5 HIV classifcations (Table4.2).4.3.3     Prediction of Future Tropism Changes on MaravirocVirologic failure on maraviroc is often accompanied by a “switch” in tropism from R5 to non-R5 289.Changes in phenotypic tropism by Trofle were also examined over the course of the trial. Analyseswere restricted to those patients who had R5 Trofle results at both screening and baseline, leaving atotal of 84 patients with DNA results and 71 with RNA results. Patients with non-R5 results by thegenotypic assays were signifcantly more likely to have subsequent Dual/Mixed or X4 results byTrofle in plasma over the course of the study (Figure 4.2).4.3.4     Diagnostic Performance of Tropism AssaysThe  performance  characteristics  of  all  four  sequencing  approaches  were  compared  against  thephenotypic original Trofle assay. Concordance of DNA-based tropism testing in cells was assessedrelative to tropism classifcations by the Trofle assay in the matching plasma sample.89Table 4.2: Short- & Long-Term Virologic Responses to Maraviroc in Patients with Compromised Background RegimensTable 4.2: Short- & Long-Term Virologic Responses to Maraviroc in Patients with Compromised Background Regimens. This  table  shows  a  subset  of  patients(N=81) with a weighted optimized background sensitivity score (wOBTss) ≤1. The plasma viral load (pVL) responses to maraviroc plus OBT are shown for patients classifed as havingR5 and non-R5 HIV by all 5 tropism methods. The median pVL decline from baseline to week 8 is shown as well as the percentage of patients with a week 8 virologic response —defned as a pVL decline ≥2 log10 or a pVL of <50 copies/mL at week 8. The percentage of patients with undetectable viraemia is also shown. The odds ratios of week 8 and week 48virologic responses are also displayed.90Figure 4.2: All Genotypic Tropism Testing Methods Predicted Future Phenotypic Tropism Changes While Receiving MaravirocFigure  4.2:  All  Genotypic  Tropism  Testing  Methods  Predicted  Future  Phenotypic  Tropism  Changes  While  ReceivingMaraviroc. Kaplan Meier plots show the percentage of patients who change phenotypic tropism status byTrofle over the course of the study. Analysis is restricted to patients with R5 Trofle results at both screening and baseline.Patients are stratifed by their baseline tropism results by each of the four genotypic assays, with the R5 groups in green andthe non-R5 groups in red. Panels A and B depict deep sequencing results in the PBMC and plasma compartments, respectively.Panels C and D depict the population-based sequencing results. The p-values by the log-rank test for the differences in mediantime to tropism change between groups were all <0.05. Ticks represent censored observations. pVL — plasma viral load91Genotypic DNA tropism testing by population-based sequencing had 80% concordance with Trofle;DNA-based deep sequencing had 77% concordance. The corresponding RNA-based approaches hadconcordance with Trofle of 80% and 86%, respectively. The sensitivity, specifcity, and concordance ofall approaches are shown in Table 4.3.Table 4.3: Performance of Sequencing-Based Tropism Assays from Peripheral Blood Mononuclear Cells & Plasma Relative to the Original Trofile Assay in PlasmaTable 4.3: Performance of Sequencing-Based Tropism Assays from Peripheral Blood Mononuclear Cells & Plasma Relativeto the Original Trofile Assay in Plasma. The sensitivity (percentage of correct non-R5 results), specifcity(percentage  of  correct  R5  results),  and  concordance  (overall  percentage  of  correct  results)  are  shown for  all  sequencingapproaches  relative  to  the  original  Trofle  assay  (not  ESTA).  PopSeq — Population-based Sequencing;  DeepSeq  — DeepSequencing.4.3.5     Compartmental DifferencesSince plasma approaches have been more thoroughly examined in the literature, the ability of theDNA-based  approaches  were  compared  against  their  corresponding  RNA-based  approaches.Relative to RNA, population-based and deep sequencing from DNA had sensitivities of 88% and86%, and specifcities of 72% and 80%. Overall rates of concordance between plasma and PBMCswere 78% using population-based sequencing and 83% using deep sequencing.Method Sensitivity, % (n/N) Specificity, % (n/N) Concordance, % (n/N)PopSeq DNA 78%(69/89)83%(76/92)80%(145/181)DeepSeq DNA 81%(72/89)74%(68/92)77%(140/181)PopSeq RNA 68%(53/78)92%(72/78)80%(125/156)DeepSeq RNA 91%(71/78)81%(63/78)86%(134/156)92To address the possibility that the bioinformatic algorithm cutoffs previously optimized for plasmawere not  optimized in  the  cellular  compartment,  exploratory  analyses  of  additional  geno2phenocutoffs were undertaken. The virologic responses of patients were evaluated using a range of FPRcutoffs: 2, 3.5, 5.75, 10, 20, and 50, in a group of 93 patients with a weighted optimized backgroundtherapy susceptibility score <1 (i.e., fewer than 1 drug in addition to maraviroc in their backgroundregimens). This analysis again confrmed that a cutoff in the range of approximately 5.75 to 10 wasable to distinguish the largest difference in week 8 pVL declines between tropism groups (Figure 4.3).Thus, the poorer performance of tropism testing in the cellular compartment was likely not an issuewith bioinformatic cutoffs having been optimized in a plasma-based context (Figure 4.3).4.3.6     Virologic Responses with Screening by DNA-Based versus RNA-Based ApproachesFor each sequencing approach,  the relative performance of the plasma or PBMC predictions wasassessed.  The  plasma  compartment  tended  to  outperform  the  PBMC  compartment  in  itspredictability  (Table  4.1).  For  both  deep and population-based sequencing,  patients  identifed  ashaving R5 HIV in both compartments had virologic declines on maraviroc of approximately 2.5 log10copies/mL by week 24. In contrast, where both compartments indicated non-R5 HIV, the medianviral load declines were approximately 1 log10. At week 24, the median pVL decline was ~1.5 log10 where plasma indicated R5 but PBMCs indicatednon-R5,  while  the  median  decline  was  ~0.5  log10 where  plasma  indicated  non-R5  but  PBMCsindicated R5 HIV. This suggests  that testing from the plasma compartment was able to  correctlyidentify  more patients  as  maraviroc  responders or  non-responders  than testing from the  cellularcompartment (Figure 4.4).93Figure 4.3: Effect of Geno2pheno Cutoffs on Prediction of Response to Maraviroc in Patients with Compromised Background RegimensFigure 4.3: Effect of Geno2pheno Cutoffs on Prediction of Response to Maraviroc in Patients with Compromised Background Regimens. Median  pVL  decline  onmaraviroc in patients with compromised background regimens, defned as a weighted Optimized Background Susceptibility Score (wOBTss) <1. Population-based sequencing resultswere interpreted using FPR cutoffs ranging from 2 to 50. Patients with FPRs less than or equal to each respective cutoff are classifed as having non-R5 HIV and are indicated with thedashed red lines. Patients with FPRs above the cutoff are classifed as having R5 HIV and are indicated with the solid green lines. A geno2pheno FPR cutoff of 5.75 – 10 seems to bestdiscriminate between responders and non-responders to maraviroc. Note that a cutoff of 3.5 is optimized for deep sequencing and not population-based sequencing.94Figure 4.4: Patients with Discordant Tropism Results Between Compartments Had Virologic Responses Which Favoured the Plasma PredictionFigure 4.4: Patients with Discordant Tropism Results Between Compartments Had Virologic Responses which Favoured the Plasma Prediction. Panel  A shows  thedeep sequencing results and Panel B shows the population-based sequencing results. Solid blue lines indicate the groups where RNA and DNA methods both indicated R5. Solidorange lines indicate the groups where RNA and DNA both indicated non-R5. Discordant groups are indicated by black dashed (RNA R5 and DNA non-R5), and grey dotted (RNAnon-R5 and DNA R5) lines. pVL — plasma viral load954.4     Discussion & ConclusionsThis study examined the performance of tropism classifcations from the cellular compartment in alarge  number  of  patients  initiating  maraviroc-based  therapy.  Two  independent  sequencingapproaches from both the plasma and peripheral blood mononuclear cell compartments were testedagainst  each  other  and  against  the  plasma-based  phenotypic  original  Trofle  assay.  Whileperformance was fairly good in PBMCs, tropism predictions from the plasma compartment tended tooutperform  the  DNA-based  methods.  Where  results  were  discordant,  longer-term  virologicsuppression was not predicted as well by DNA-based methods, suggesting some misclassifcation byDNA versus RNA approaches.Nevertheless,  this  approach  may  be  the  only  option  for  some  patients,  barring  deliberate  butinadvisable treatment interruptions  355 to raise  viraemia to  levels  needed for  RNA-based tropismtesting.  Additionally,  the  reasonably  high  concordance  (~80%)  between  the  plasma  and  cellularcompartments  should  give  some  confdence  that  DNA-based  approaches  give  useful  clinicalinformation. The use of DNA-based tropism testing is suggested in European guidelines 276, but betterguidance will likely result from an ongoing clinical trial of DNA-based tropism testing in patientswith suppressed viraemia 344. This study confrms and expands on the recent results from Vitiello and colleagues, who examined agroup of  20  patients  switching a  component  of  their  antiretroviral  regimens  to  maraviroc  whilevirologically suppressed  356. Both studies found that DNA tropism testing could be used to predictsuccessful treatment with maraviroc, arguably the most clinically relevant outcome of a tropism test.Compared to past reports, this study found roughly similar, if slightly worse, diagnostic accuracy ofcell-based genotypic tropism assays relative to phenotypic and plasma-based approaches 231,347,357,358. Aside from the ability to assess diagnostic performance, the primary advantage of the current studyis that the tropism classifcations could be additionally evaluated for how well they predict virologicresponse to maraviroc in a real  clinical setting.  The ability to predict these responses acted as an96independent,  objective confrmation of the diagnostic performance of these assays. An additionaladvantage of this study is the fact that paired plasma and PBMC fractions from the same blood drawscould be compared for their ability to predict virologic response, whereas other studies have tendedto compare later PBMC results with earlier pre-suppression plasma results  283,347,359. This study alsohas the advantage of the design of the clinical  trials from which these patients were drawn. Theinclusion  of  patients  enrolled  in  A4001029  who  had  non-R5  results  at  baseline  but  were  stillprescribed maraviroc gives additional confrmation on the utility of these methods. There are some diffculties inherent to tropism testing from HIV DNA. The cellular “buffy coat”fraction of whole blood is not routinely collected or stored, nor are peripheral blood mononuclearcells routinely separated for analysis. The cellular compartment has also been found to have highersequence variation  211,  and higher prevalence of CXCR4-tropic HIV  165 compared to blood serum.Importantly,  overestimation of CXCR4-usage may actually increase the likelihood of success withCCR5 antagonists, since more patients may be screened out by DNA-based approaches. However, thecurrent results are not defnitive in their support for this hypothesis, and a prospective clinical trialusing DNA tropism testing has yet to be completed. Low input copy number may also be an issue for testing from the cellular compartment compared tothe plasma  360.  Quantitation of HIV DNA and cell  number were not performed in this study,  sopotentially  low  copy  numbers  may  have  contributed  to  performance  issues.  However,  routinequantitation of cells  or  DNA copies  represents  a  fairly  signifcant  barrier  to  the implementation,availability,  and  turn-around  time  of  DNA-based  tropism  testing.  Input  copy  number  may  beaccounted for by the use of PCR “tags” accompanying each DNA strand amplifed, as recent workhas shown 361. However, this technique was not available at the time of testing. A major strength of this study was having access to paired plasma and PBMC samples from the sametime point. This enabled direct comparisons between compartments. Although these patients wereviraemic, this sample set is ideal for comparing DNA-based approaches to RNA-based ones, sincethere is much more clinical experience with RNA-based approaches. However, the fact that patients97did not have suppressed viraemia at baseline requires extrapolation of these results to patients withundetectable viral loads who may switch to maraviroc. This should be noted, since such patientsultimately comprise the target group for DNA-based tropism approaches. Cellular HIV DNA copynumber may decay with effective antiretroviral therapy 362,363, since lower replication may reduce thepool of HIV DNA. This could lead to diffculties in DNA-based tropism testing in aviraemic patients.However this issue could not be addressed with this sample set because all patients had detectableviral loads. The relatively small number of patients who had tropism classifcations that were discordant betweencompartments (e.g., R5 in plasma but non-R5 in PBMCs) meant it was not possible to defnitivelystate that plasma predicts maraviroc response better than the cellular compartment. Conversely, thesmall number of patients with discordant results also reflects the reassuring fact that a large majorityof patients in fact had concordant results between the compartments. Another potential limitation ofthis study is the fact that background therapy also affects response to maraviroc in addition to HIVtropism  364.  However,  this  would  presumably  affect  all  assays  equally  in  their  ability  to  predictvirologic response to maraviroc-based therapy, so it should not have greatly skewed the results. Despite the above-mentioned caveats, this study demonstrates promising potential for DNA-basedtropism methods. That the cell-based classifcations were not as clinically predictive as plasma-basedones should add a measure of caution to the routine use of this approach. However, the DNA-basedtesting  was  still  able  to  discriminate  between  responders  and  non-responders  to  maraviroc-containing  regimens,  and despite  some shortcomings,  may  be  the  best  course  of  action  prior  toprescribing maraviroc in patients with suppressed viraemia. This, and the previous chapter examined the utility of next-generation sequencing in evaluating HIVtropism and using the results to predict subsequent virologic failure over long-term follow-up. Thesechapters examined the performance of next-generation sequencing and V3 genotyping prior to CCR5antagonist treatment and assessed this performance during follow-up by employing results  fromother assays and virologic tests such as phenotypic assays and plasma viral  load tests.  The next98chapter  applies  genotypic  tropism  and  next-generation  sequencing  methods  in  a  more  detailedcontext with long-term, longitudinal follow-up. Changes in the  env sequence which are associatedwith treatment failure are examined and correlated with pre-treatment deep sequencing results. Thefollowing chapter thus explores the specifc genotypic factors associated with the virologic failurewhich the previous chapters have demonstrated can be predicted by next-generation sequencing. 99Chapter 5:     Genotypic Analysis of the HIV-1 V3 Region in Virologic Non-Responders to Maraviroc-Containing Regimens Reveals Distinct Patterns of Failure5.1     Background & IntroductionSuccessful antiretroviral treatment with the CCR5 antagonist, maraviroc, requires a tropism test toconfrm that  the patient’s  HIV uses the CCR5 coreceptor for cellular  entry (R5 HIV) rather  thanCXCR4 (non-R5 HIV) 288,289,291. In the Phase III clinical trials of maraviroc, patients were screened fortropism status using the original Trofle phenotypic coreceptor assay (OTA), which has subsequentlybeen replaced by the enhanced sensitivity Trofle assay (ESTA)  290,365. Recent re-screening of clinicaltrials of maraviroc has confrmed the utility of genotypic approaches for the determination of HIVtropism 325,340,348,349,366. Such approaches typically involve sequencing of the third variable (V3) region ofthe HIV envelope gene 200. Bioinformatic algorithms such as geno2pheno 204 are then used to infer the phenotypic tropism that islikely associated with a V3 genotype. Geno2pheno converts an input V3 sequence into an outputvalue in the form of a false-positive rate (FPR), ranging from 0 to 100. An FPR indicates how likely asequence is to be incorrectly identifed as non-R5. Therefore, sequences yielding low false-positiverates have a high likelihood of being non-R5.Historically, population-based sequencing has been the most commonly used genotypic approach forpredicting coreceptor usage  200. However, more sensitive tropism determination methods can moreaccurately predict response to maraviroc 290; thus, newer deep sequencing methods targeting the V3loop  are  becoming  increasingly  common  232,248,321,348,349,367.  These  next-generation  sequencingapproaches can identify low-level non-R5 subpopulations in clinical samples. Following treatment100with maraviroc, these minority non-R5 quasispecies may emerge to much higher prevalence, therebycompromising treatment effcacy 232,316.There are several  pathways by which patients may fail  a  maraviroc-containing therapy regimen.Most commonly, a minority non-R5 population in a patient’s HIV population may expand underdrug  pressure,  causing  an  overall  change  in  observed  tropism  289.  Less  commonly,  the  viralpopulation may retain its CCR5 tropism while evolving the ability to use maraviroc-bound CCR5protein for  cellular entry — a form of  maraviroc resistance  368.  Thirdly, the viral  population maydevelop resistance  to the other agents  in the  background regimen in the absence  of a change insusceptibility to maraviroc 364; this may be associated with either R5 or non-R5 tropism. Furthermore,as  with  other  agents,  adherence,  absorption,  and  other  patient-associated  and  pharmacokineticfactors can also lead to therapy failure.Early detection of tropism shifts  or  maraviroc resistance can accelerate the decision to substitutemaraviroc  with  another  antiretroviral  agent  and  potentially  prevent  further  accumulation  ofantiretroviral  drug  resistance  to  other  agents  in  the  regimen.  Thus,  patients  in  this  study  weresampled  relatively soon after  beginning maraviroc treatment to  determine the utility  of  an earlymonitoring approach. This study uses both population-based and deep sequencing approaches to assess changes in tropismand  V3  sequence  among  treatment-experienced,  R5-infected  patients  who  experienced  virologicfailure  while  receiving  maraviroc  in  the  MOTIVATE-1  and  -2  studies  288,289.   Patients  from  theA4001029 study which enrolled patients with non-R5 HIV 291 were not included in the current study.Thus, all patients studied were determined to have exclusively R5 HIV by the original Trofle assay(OTA). They are therefore a representative population of patients most likely to receive maraviroc.Phylogenetic methods were also used to assess whether sequences present at failure were derivedfrom pre-existing  minority  subpopulations,  and  next-generation  sequencing  was  used  to  assesschanges in non-R5 prevalence after treatment with maraviroc. 101Previous studies 232,316 have noted emergence of CXCR4-using virus from pre-existing subpopulations,and CCR5 antagonists have been known to inhibit R5-only, while selecting non-R5 subpopulations241,289, with such shifts appearing to occur very quickly 317. Furthermore, resistance to maraviroc hasbeen associated with genotypic changes in the HIV envelope gene 368,369. Thus, it was hypothesizedthat there are distinct mechanisms of failure that can be identifed by population-based and/or deepsequencing of the HIV V3 region. 5.2     Materials & Methods5.2.1     Patient & Sample SelectionA subset of patients was selected who had suboptimal responses to maraviroc in the MOTIVATEtrials (N=181). Patients were selected such that approximately the same proportion had non-R5 OTAresults at failure as was reported for the MOTIVATE trials overall (57% in MOTIVATE, 58% in thecurrent study) 289. All patients were treatment-experienced, 100% of patients had R5 results by OTA atscreening, and 69% had R5 results by ESTA (124/181). All received maraviroc (once or twice daily)plus an optimized background regimen of three to six other antiretroviral  agents.  All individualsgave written informed consent, including consent to allow other tropism testing to be performed ontheir samples. The University of British Columbia—Providence Health Care Research Ethics Boardreviewed the research project and granted ethical approval.Sequencing was performed on samples from two time points: one prior to receiving maraviroc (thescreening sample) and one while receiving treatment (the failure sample). This on-treatment failuresample was defned as the earliest available sample with a plasma viral load (pVL) greater than 500HIV RNA copies/mL, and an OTA result. The screening sample was drawn approximately 6 to 8weeks  prior  to  beginning  maraviroc;  the  failure  sample  was  drawn a  median  of  4  weeks  afterbeginning maraviroc (interquartile range [IQR]: 4-16 weeks),  and a median of 2 weeks (IQR: 2-10weeks) after the frst viral load ≥500 copies/mL. While phenotypic tropism results were available forall samples, phenotypic maraviroc resistance assay results were not available for these samples. ESTAresults were available at screening, but only OTA results were available at failure. 1025.2.2     Genotypic Tropism TestingThe third variable  (V3)  loop of  the  HIV envelope gene  was amplifed with nested RT-PCR. Thescreening  samples  were  amplifed  and  sequenced in  triplicate;  the  failure  samples  had  a  singlesequence  generated  per  sample.  Standard,  population-based  sequencing  was  performed  on  allscreening and failure samples, as previously described 325. Deep V3 sequencing was also performedon  all  screening  samples,  plus  a  subset  (N=73)  of  failure  samples,  with  methods  as  previouslydescribed  348,349. The 73 samples comprised the last batch of samples processed through populationsequencing, with no targeted selection.  The tropism associated with the V3 loop sequences was inferred using the geno2pheno algorithm 204with  false-positive  rate  (FPR)  cutoffs  of  5.75  for  population-based  sequencing  and  3.5  for  deepsequencing 336,337, below which sequences were categorized as non-R5. These cutoffs had previouslybeen  optimized  for  predicting  virologic  response  to  maraviroc  336,337.  The  percentage  of  non-R5variants in the viral population was defned as the proportion of sequences scoring below or equal toan FPR of 3.5 as observed by deep sequencing, and previous studies have defned an R5 sample ashaving <2% non-R5 variants 348,349. The screening and failure sequences were assessed for amino acid changes that may have appearedfollowing maraviroc-based therapy, as well as for a change in the geno2pheno FPR value. Neighbour-joining phylogenetic trees were constructed with ClustalX using deep sequencing data at screeningand failure  population-based or  deep sequencing  data.  Thus,  it  could be  determined whether  asequence present at failure may have already been present prior to treatment with maraviroc. Thechange in the percentage of non-R5 variants between screening and failure was also examined usingthe deep sequencing results. Sample phenotypes were obtained using OTA at all time points, andESTA at screening. 1035.2.3     Statistical AnalysesStatistical analyses performed included the Mann-Whitney U test and Kruskal-Wallis test for testingthe statistical  signifcance of  differences  between medians (e.g.,  median  plasma viral  loads).  TheFisher exact test was used for testing the statistical signifcance of differences in proportions (e.g., theproportion of patients who were R5 by ESTA at screening).5.3     Results5.3.1     Patient & Sample CompositionPatients in the current study were all treatment-experienced and received maraviroc once daily (89patients, 49%) or twice daily (92 patients, 51%), as per randomization at study entry. Most patients(91%) were  enrolled in the North American MOTIVATE-1 trial  370,  with the remaining 9% in theMOTIVATE-2 trial,  which  had an identical  study design.  Of the 181 patients  selected,  100 (55%)experienced virologic failure; 44 (24%) never achieved virologic suppression but completed 48 weeksof treatment; and 15 (8%) had a virologic rebound. Of the remaining 22 patients, 18 were lost tofollow up, two died, one experienced an adverse event, and one was withdrawn due to pregnancy. The mean age of subjects was 45 (range: 19-70), and the proportion of males in the study was 91%(165/181). These were similar to the maraviroc arms of MOTIVATE overall  288.  The proportion ofpatients reporting Black race or ethnicity was 19% (34/181), which was slightly elevated relative tothe larger trial overall (14%). This was likely due to a higher number of maraviroc non-responderswho reported Black race/ethnicity in MOTIVATE 289 (Table 5.1). As expected for a study on patientswho experience failure of therapy, the patients in the current study had higher plasma viral loads,lower  CD4 cell  counts,  and fewer  active  drugs  in  their  background  regimens  than those  in  theMOTIVATE studies 288,348 overall.104Table 5.1: Patient Characteristics in the Current Study Compared to the MOTIVATE Studies Overall105Table 5.1: Patient Characteristics in the Current Study Compared to the MOTIVATE Studies Overall. Thebaseline patient characteristics in the current study, as well as those from the maraviroc (MVC) arms of the MOTIVATE-1 and-2 studies. Most values shown are median values with the interquartile range (IQR) in parentheses, unless otherwise indicated.The MOTIVATE column was derived from a previously published dataset (Swenson et al, JID 2011 [ 348]) comprising a majorityof  maraviroc  recipients  in  the  MOTIVATE  studies  (94%,  788/840  patients).  Due to  low numbers  of  patients,  those  withrace/ethnicities other than White  or Black are not included in the table. pVL — plasma viral load; wOBTss — weightedoptimized background therapy sensitivity score; FPR — false-positive rate; ARVs — antiretrovirals. *Some values were derivedfrom Gulick et al, NEJM 2005 [288], and from Fätkenheuer et al, NEJM 2005 [289] and each is marked with an asterix.The failure sample was taken as the earliest available on-treatment sample with both a viral load>500 copies/mL and an OTA result from the same time-point. Samples with viral loads <500 copieswere not tested by OTA, and were therefore excluded from the study. The median viral load at failurewas 4.1  log10 copies/mL (IQR:  3.5  –  5.0  log10),  ranging from a minimum of  670 copies/mL to  amaximum of 10 million copies/mL. Most patients (55%) in the study experienced protocol-defned virologic failure (PDVF) over the 48weeks of the MOTIVATE trials. However, the samples tested were generally from earlier time pointsthan the week where PDVF was met. The median time to PDVF was approximately 17 weeks and 25weeks for the groups failing with non-R5 and R5 OTA phenotypes, respectively 289. In comparison, thesamples in the current study were from a median of 4 and 4 weeks, for those with non-R5 and R5phenotypes, respectively, since the earliest available failure samples were intentionally selected. 5.3.2     Performance of Population-Based Genotyping for Determining HIV TropismAt screening, all patients had R5 HIV by OTA, but 12% and 31% had non-R5 results by population-based sequencing and ESTA, respectively. At the failure time point, 91 patients had genotypic non-R5results  (50%)  and  90  patients  had  genotypic  R5  results  (50%)  by  population  sequencing.  Incomparison, the proportions reported by the phenotypic  OTA were 105 non-R5 (58%) and 76 R5(42%).  Approximately  half  of  the  patients  had  non-R5  HIV  by  both  genotypic  and  phenotypicmethods at failure (89 patients, 49%). Of the remaining patients, 41% had R5 HIV by both methods106(N=74), and 10% (N=18) had discordant results (with 16/18 having R5 as determined by genotypebut non-R5 by OTA). 5.3.3     Change in V3 Sequence & Geno2pheno Value after Maraviroc TreatmentFor a number of patients, there were large changes in the geno2pheno false-positive rate followingmaraviroc treatment. The overall change in geno2pheno for the population as a whole is shown inFigure 5.1. The median FPR for all patients regardless of tropism status fell from 31.0 at screening to5.3 at failure (Figure 5.1), owing to the large number of patients failing with non-R5 HIV. Importantly,these  patients  fell  into  two  distinct  categories:  those  who  maintained  essentially  the  samegeno2pheno FPR, and those for whom a large decrease in the FPR value between screening andfailure was observed (Figure 5.2). These categories of failure generally corresponded with failing withan R5 or non-R5 phenotype, respectively. The overall drop in geno2pheno FPR was driven by an increase in the number of patients with non-R5 genotypes, with this number increasing over 4-fold from 21 patients at screening to 91 patients atfailure (12% to 50%). Between screening and failure, the geno2pheno FPR fell by a median of 18.2units (IQR: -38.0 – -5.7) for those patients with concordant non-R5 results (Figure 5.3). These patientshad extremely low geno2pheno false-positive rates at failure, with a median FPR of 1.1 (IQR: 0.4 –1.7). In comparison, the median FPR of these same patients at screening was 20 (IQR: 6.9 – 38). Incontrast,  there was negligible  change in  geno2pheno FPR in patients  failing with concordant R5results. For these patients, the median FPR change was 2.2 (IQR: -0.5 – 16) (Figure 5.3, Table 5.1).107Figure 5.1: Overall Change in Geno2pheno False-Positive Rate between Screening & Failure Figure 5.1: Overall Change in Geno2pheno False-Positive Rate between Screening & Failure. The  overalldecrease in the geno2pheno false-positive rate value between screening and failure.  The distribution of geno2pheno false-positive rate (FPR) values is shown for the screening (left) and failure sequences (right). Boxes indicate the interquartile rangeof the values, with the median value indicated by a solid horizontal line. Whiskers correspond to 1.5 times the interquartilerange.108Figure 5.2: Individual Geno2pheno False-Positive Rate Values at Screening & Failure Figure 5.2: Individual Geno2pheno False-Positive Rate Values & Screening and Failure. A  scatterplot  of  thegeno2pheno false-positive rate (FPR) of all patients with coordinates at two time-points: screening value on the horizontal axisand failure  value on the vertical  axis.  Points  are  marked by whether tropism results  at  failure  were concordant betweenphenotype and genotype  (see  legend).  The  geno2pheno decreased by  a  large amount  between screening  and failure  forpatients in the non-R5 group, but changed very little for those in the R5 group.109Figure 5.3: Individual Changes in Geno2pheno False-Positive Rate Values between Screening & Failure Figure 5.3: Individual Changes in Geno2pheno False-Positive Rate Values between Screening & Failure. Thechange in geno2pheno false-positive rate between screening and failure for each patient. Horizontal lines denote the medianvalues, with error bars indicating the interquartile ranges. Patients with concordant R5 tropism at failure had a median FPRchange of 2 (IQR: -1 – 16), versus a median decline of 18 FPR units (IQR: -38 – -6) in the concordant non-R5 group (p<0.001).Patients with discordant results at failure had an overall intermediate FPR decline (median FPR change = 10; IQR: -39 – 7).Points are marked by whether tropism results at failure were concordant between phenotype and genotype (see legend).5.3.4     Classical Substitutions in Patients with Non-R5 HIV at Failure but Limited Evidence of Maraviroc Resistance in Those with R5 HIVWhen the sequences from patients with non-R5 genotypes at failure were examined, not surprisingly,the most common emergent amino acid substitutions among patients with non-R5 genotypes weresubstitutions  to  basic  amino  acids.  There  were  3  primary  codons  where  non-R5  substitutionsoccurred.  These occurred as follows:  11R (36  patients,  40%),  13R (23 patients,  25%),  and 25K (20patients, 22%). Consequently, the 11/25 rule 133 could identify a substantial proportion of sequences asnon-R5 (57/91, 63%) at the time of failure. 110In contrast to patients with non-R5 failure genotypes, patients who maintained genotypic R5 HIVthrough the study period (N=90) exhibited no clear accumulation of mutations at the failure timepoint. Among these patients, the median geno2pheno FPR at screening and failure showed very littlechange,  at  41.9  versus  48.9,  respectively.  In  21  patients  (23%)  no  V3  amino  acid  changes  wereobserved following maraviroc treatment, while substitutions in the remaining 70 patients (77%) wererestricted to partial amino acid changes (mixtures). Among these patients without a tropism change, the sites with the highest rates of substitutions werecodons 10, 13, 14, 18, and 25. The most common substitutions at these positions were 10R, 13H/P,14I/M, 18R, and 25D. Importantly, however, the prevalence of these substitutions was very low inthis  population,  ranging  between  8  and  14  samples  (9  –  16%)  depending  on  the  substitution.Prevalence was low even for substitutions previously documented to be associated with maravirocresistanceThese may indeed be simply natural polymorphisms unrelated to maraviroc resistance. Furthermore,since most of these samples did not have phenotypic maraviroc resistance assay results, the ability tointerpret the implications of these substitutions was limited, and many patients with R5 viruses maysimply have been non-adherent,  or  had viruses  that  were resistant  to  other components of  theirregimens.  5.3.5     Change in Non-R5 Viral Population as Determined by Deep Sequencing The viral population present prior to treatment with maraviroc was assessed by deep sequencing ofthe screening plasma samples. The deep sequencing data was then investigated for both the changein the percentage of non-R5 variants, as well as the phylogenetic relationship between the screeningand failure V3 sequences. 111All screening samples underwent deep V3 sequencing,  as  did a subset  of  73 failure samples.  Atscreening, the median percentage of non-R5 variants per patient was 0.1% (IQR: 0–3.1%) — reflectiveof the R5 phenotypes of all patients. However, a majority of patients had at least some level of non-R5sequences present at screening (96/181, 53%), and including one patient in whom 99.9% of recoveredsequences were interpreted as non-R5 at screening, who had R5 by OTA and ESTA at screening, butexperienced virologic failure with Dual-Mixed tropism by OTA at week 4. A total of 50 patients (28%)had ≥2% non-R5 HIV according to their deep sequencing screening result. This was over twice asmany as were detected at screening by population-based sequencing, despite all patients having R5OTA phenotypes at screening.Of the 50 patients with non-R5 variants present at ≥2% prevalence by deep sequencing, 42 (84%) wereconfrmed to have non-R5 at failure by population-based genotype. Where deep sequencing resultswere available at both time points (N=73),  the overall median percentage of non-R5 variants roseslightly from 0% (IQR: 0 – 1.2%) at screening to 0.8% (0 – 98.8%) at failure. When these patients wererestricted to those with non-R5 at failure by population-based sequencing, the median percentage ofnon-R5 variants rose to 99.4% (IQR: 95.4 – 99.9%) at failure.Strikingly, distribution of non-R5 variants in patients treated with maraviroc was nearly completelydichotomous. According to deep sequencing results, the vast majority of patients (65/73, 89%) hadtreatment failure with either less than 5% non-R5 variants or greater than 95% non-R5 variants, withvery few patients  falling in between.  As previously  mentioned,  the population-based sequencingresults  were  also  quite  unambiguous  in  their  interpretation.  Of  those  with  non-R5  populationsequencing results at failure, over three-quarters had extremely low geno2pheno FPRs of 2 or lower(70/91, 77%), indicative of “highly” non-R5 virus 336 (Figure 5.4). 112Figure 5.4: The Percentage of Non-R5 Variants in Deep Sequencing Results at Screening & FailureFigure 5.4: The Percentage of Non-R5 Variants in Deep Sequencing Results at Screening & Failure. A scatterplot of the percent non-R5 variants for all patients with deep sequencing results at screening and failure.Points are marked by whether tropism results in the same time-point were concordant between phenotype and population-based genotype (see legend).  The failure column illustrates how the majority of failure samples had very high or very lownon-R5 prevalence. Phenotypes were performed by ESTA at screening and original Trofle assay (OTA) at failure. A dashed lineat 2% non-R5 prevalence represents a cutoff for deep sequencing, above which a sample was classifed as having non-R5tropism5.3.6     Phylogenetic Relationship between Screening & Failure SequencesPhylogenetic  trees  were  generated  using  the  screening  deep  sequencing  data  and  the  failurepopulation-based sequence. For many patients, a distinct minority subpopulation of non-R5 variants113was detected by deep sequencing at screening. This minority subpopulation often emerged followingtreatment, and was detected with standard population-based sequencing methods. The trees wereinspected  manually to  assess  the  degree of  the phylogenetic  relationship between the  failure  V3sequence and sequences detected by deep sequencing prior to maraviroc treatment. Overall,  70%  of  patients  (64/91)  with  non-R5  HIV  at  failure  had  a  closely  related  non-R5subpopulation  present  prior  to  treatment  with  maraviroc,  confrming  previous  reports  of  theselection of pre-treatment non-R5 reservoirs by maraviroc 232,316. These CXCR4-using subpopulationswere present despite patients being pre-screened as having R5 HIV with OTA. A number of thesepatients were missed by population-based sequencing as well. A set of representative example trees isgiven in Figure 5.5-5.8 and Appendices V–XIV.5.3.7     Comparison of Tropism MethodsThe performance of population-based sequencing could be assessed at both screening and failure bycomparing the results using deep sequencing as the “gold standard”. ESTA results were available forcomparison  at  screening,  and  OTA at  failure.  When  the  two  genotypic  tropism  methods  werecompared at screening, population-based sequencing had 30% sensitivity (15/50 non-R5 samples)and 95% specifcity (125/131 R5 samples)  relative  to deep sequencing.  However,  performance ofpopulation-based  sequencing  was  dramatically  better  at  failure.  This  method  achieved  88%sensitivity (29/33 called non-R5) and 95% specifcity (38/40 called R5) relative to deep sequencing,likely due to the higher proportions of non-R5 variants after maraviroc treatment.114Figure 5.5: Phylogenetic Tree from a Patient Who Had a Small Pre-Treatment X4 PopulationFigure 5.5: Phylogenetic Tree from a Patient with a Small Pre-Treatment X4 Population. Phylogenetic tree generated from the deep sequencing data at screening and thepopulation-based genotype at failure. Screening R5 sequences are shown in green, X4 sequences are shown in red, and the failure sequence is shown in blue. Failure was due to a smallpre-treatment X4 population detected by deep sequencing (1% X4) but not by phenotyping (ESTA R5). The failure sample was X4 by population-based genotype and X4 by the originalTrofle assay.115Figure 5.6: Phylogenetic Tree from a Patient Who Had a Large Pre-Treatment X4 PopulationFigure 5.6: Phylogenetic Tree from a Patient with a Large Pre-Treatment X4 Population. Phylogenetic tree generated from the deep sequencing data at screening and thepopulation-based genotype at failure. Screening R5 sequences are shown in green, X4 sequences are shown in red, and the failure sequence is shown in blue. Failure was due to a largepre-treatment X4 population detected by deep sequencing (53% X4) but not by phenotyping (R5 by ESTA). The failure sample was X4 by population-based genotype and Dual-Mixedby OTA.116Figure 5.7: Phylogenetic Tree from a Patient for Whom Deep Sequencing Failed to Detect X4 HIV Figure 5.7: Phylogenetic Tree from a Patient for Whom Deep Sequencing Failed to Detect X4 HIV. Phylogenetic  tree  generated  from  the  deep  sequencing  data  atscreening and the population-based genotype at failure. Screening R5 sequences are shown in green, X4 sequences are shown in red, and the failure sequence is shown in blue. Pre-treatment X4 sequences were not detected by deep sequencing (0% X4) and screening phenotype was R5 by ESTA. However, failure occurred with a non-R5 genotype and phenotype.117Figure 5.8: Phylogenetic Tree from a Patient Who Failed with R5 HIV Figure 5.8: Phylogenetic Tree from a Patient Who Failed with R5 HIV. Phylogenetic tree generated from the deep sequencing data at screening and the population-basedgenotype at failure. Screening R5 sequences are shown in green, X4 sequences are shown in red, and the failure sequence is shown in blue. A patient who experienced failure with anR5 genotype and phenotype. One outlier branch has been truncated for display purposes. Both screening and failure time points were R5 by all genotypic or phenotypic testsperformed 118The genotypes were also compared to the phenotypes. At screening, population-based sequencinghad 19% sensitivity (11/57) and 92% specifcity (114/124) relative to ESTA. Deep sequencing had 53%sensitivity (30/57),  and 84% specifcity  (104/124)  relative  to  ESTA at  screening.  At failure,  whenpopulation-based  genotypes  were  compared to  the  OTA phenotypes  in  the  same time-point  theassays  were  90% concordant  (163/181  samples).  Overall  sensitivity  of  genotyping  compared  tophenotyping was 85% (89/105 non-R5), with 97% specifcity (74/76 R5) in these failure samples. Thisperformance was comparable to the performance of deep sequencing relative to OTA at failure: 83%sensitivity (30/36), 92% specifcity (34/37). 5.3.8     Virologic Responses to MaravirocWhile all patients were R5 by OTA at screening, they could be stratifed by their genotypic tropismresults  in  their  failure visit.  Patients with a non-R5 genotype by population-based sequencing atfailure had overall poorer virologic responses to maraviroc. At week 8, the median decline in plasmaviral load (pVL) from baseline was 2.0 log10 in those with R5 HIV but 0.4 log10 in those with non-R5HIV at failure (p<0.001). In contrast, the median change in pVL at week 8 for the maraviroc arms inthe MOTIVATE trials overall was approximately 2.4 log10 copies/mL. This was larger than the viralload decreases for either group in the current study (Figure 5.9, p<0.01).A total of 88% of patients with non-R5 population genotypes at failure (80/91) failed to achieve anundetectable viral load during the study, versus 77% (69/90) of those with R5 genotypes at failure.Protocol-defned virologic failure was documented for  69% of  patients  with non-R5 genotypes atfailure (63/91), compared to 41% of patients with R5 at failure (37/90). Patients with R5 at failure hadhigher rates of virologic rebound compared to those with non-R5: 14% (13/90) versus 2% (2/91).They were also twice as likely to have never suppressed throughout the study but remain enrolled:32% (29/90) versus 16% (15/91). 119Figure 5.9: Virologic Responses Were Reduced among Patients with Non-R5 Genotype Results at FailureFigure 5.9: Virologic Responses Were Reduced among Patients with Non-R5 Genotype Results at Failure. The  median  change  in  plasma  viral  load  from  baselineamong maraviroc recipients is shown. Patients are stratifed according to whether their frst available failure sample had an R5 (turquoise line) or non-R5 (red line) populationgenotype. For comparison, the median viral load change of the maraviroc arms in the MOTIVATE trials overall is also shown (black line).1205.3.9     Comparison to the Enhanced Sensitivity Trofile AssayAs stated above, sensitivities of population-based sequencing and deep sequencing were 19% and53% relative to ESTA at screening, respectively, with concordance of 69% and 74%. Of those where re-screening  by  ESTA indicated  pre-treatment  non-R5  phenotypes,  50  of  57  (88%)  patients  wereconfrmed to have OTA non-R5 phenotypes at failure, similar to the results by genotyping (47 of 57patients, 82%). Patients with pre-treatment R5 phenotypes by both OTA and ESTA were more likelyto fail therapy with R5 phenotypes or genotypes (56% or 65%) than non-R5 (44% or 35%). All follow-up results were tested with OTA, but it is important to note that this study indicates thatvery low minority non-R5 variants present only at the time of failure are not commonly associatedwith suboptimal maraviroc responses. Deep sequence analysis demonstrated that when phenotypictropism changes occurred, they were generally accompanied by very high non-R5 prevalence (Figure5.4).  Accordingly, the current results are likely to be unaffected by the fact that the failure phenotypeswere performed using OTA rather than ESTA. Furthermore,  largely similar  results were obtainedwhen analyses were restricted only to patients with R5 by ESTA at screening (Figures 5.10-5.12). 5.4     Discussion & ConclusionsIn this study, genotypic analysis indicated that failure on maraviroc followed two distinct pathways.Those patients who experienced an HIV tropism shift to non-R5 had a large decline in geno2phenofalse-positive rate and accumulated V3 substitutions at multiple codons, and had a large increase inthe prevalence of non-R5 variants to a median of 99% according to deep sequencing. Patients with R5results  at  failure  tended  to  have  very  similar  geno2pheno  values  to  their  screening  values,  andaccumulated few amino acid substitutions in V3 compared to the pre-treatment sequences. 121Figure 5.10: Overall Change in Geno2pheno False-Positive Rate between Screening & Failure in a Subset of Patients with R5 Results at Screening by the Enhanced Sensitivity Trofile AssayFigure 5.10: Overall Change in Geno2pheno False-Positive Rate between Screening & Failure in a Subset of Patients withR5 Results at Screening by the Enhanced Sensitivity Trofile Assay. The distribution of geno2pheno false-positiverate (FPR) values is shown for the screening (left) and failure sequences (right). Boxes indicate the interquartile range of thevalues, with the median value indicated by a solid horizontal line. Whiskers correspond to 1.5 times the interquartile range.122Figure 5.11: Individual Geno2pheno False-Positive Rate Values at Screening & Failure in a Subset of Patients with R5 Results at Screening by the Enhanced Sensitivity Trofile AssayFigure 5.11: Individual Geno2pheno False Positive Rate Values between Screening & Failure in a Subset of Patients withR5 Results at Screening by the Enhanced Sensitivity Trofile Assay. A  scatterplot  of  the  geno2pheno  false-positive rate (FPR) of all patients with coordinates at two time-points: screening value on the horizontal axis and failure valueon the vertical axis. Points are marked by whether tropism results at failure were concordant between phenotype and genotype(see legend). The geno2pheno decreased by a large amount between screening and failure for patients in the non-R5 group, butchanged very little for those in the R5 group.123Figure 5.12: Individual Changes in Geno2pheno False-Positive Rate Values between Screening & Failure in a Subset of Patients with R5 Results at Screening by the Enhanced Sensitivity Trofile AssayFigure 5.12: Individual Changes in Geno2pheno False-Positive Rate Values between Screening & Failure in a Subset ofPatients with R5 Results at Screening by the Enhanced Sensitivity Trofile Assay. A scatterplot of the false-positive rate change between screening and failure. Horizontal lines denote the median values, with error bars indicating theinterquartile ranges.In 70% of patients, deep sequencing was able to detect a pre-treatment non-R5 subpopulation whichemerged  at  failure.  This  study  also  demonstrated  that  standard  population-based  sequencing  iscapable of identifying on-treatment tropism changes accompanying maraviroc failure, and does sowith high sensitivity (85%) relative to phenotyping. The sensitivity reported here for  population-based sequencing is much higher than previously reported results 325. The sensitivity of population-based sequencing more than tripled after patients began treatment with maraviroc (88% sensitivityon-treatment  versus  24% at  screening  when  compared  to  deep sequencing).  Other  studies  havetypically reported much lower sensitivities for population-based sequencing 187,190,229,325,371. 124The high sensitivity reported for the failure samples in the current study is likely attributable to theselective effect of maraviroc treatment on patient HIV. For those who failed maraviroc-based therapywith  non-R5  HIV,  the  average  percentage  of  non-R5  variants  rose  to  99%  according  to  deepsequencing. This likely increased the ability of population-based sequencing to give a non-R5 result.Under these circumstances,  population-based methods performed better than usual since non-R5prevalence is usually masked by a predominantly R5 viral population, thus limiting sensitivity. Thisrapid emergence of high prevalence non-R5 variants has also been reported for cases of treatmentfailure with other CCR5 antagonists 317,372. Maraviroc  recipients  with  non-R5 HIV infection at  failure  had poorer  virologic  responses  to  themedication than those whose virus did not change tropism — even within this population of patientswho failed maraviroc-based therapy. This is likely due to the additional loss of maraviroc activity inpatients who fail therapy with non-R5 HIV, whereas failure with R5 HIV may have been due to anumber of reasons, including maraviroc resistance, resistance to the other background antiretroviralagents, and/or poor adherence. However, response to antiretroviral therapy in general may also beimpacted by the presence of non-R5 HIV infection 155,173.Some limitations of this study should be acknowledged. It may be diffcult to extend these fndingsgenerally to all maraviroc-treated populations as this study was conducted in a selected populationpre-screened for R5 HIV by OTA. However, the proportion of non-R5 OTA results at failure in thecurrent study (57%) was quite reflective of the proportion seen in the MOTIVATE trials overall (58%)289, so these results are likely generalizable to the larger MOTIVATE study population. While ESTAresults were available at screening, the follow-up phenotypes were performed using OTA. However,additional sensitivity of ESTA over OTA (reported detection limit of 0.3% versus 10% non-R5 HIV 365)likely had very little affect on the above results, since genotypic analyses indicated that phenotypictropism changes on maraviroc were associated with extremely high non-R5 prevalence well abovethe 10% detection limit of the original Trofle assay. 125Despite its high sensitivity, deep sequencing could not identify pre-existing non-R5 populations in30% of patients with non-R5 HIV at failure. Thus, this approach may still lack suffcient sensitivity, orthe sampling volume may have been insuffcient for detection of minority variants.  Alternatively,non-R5 variants may evolve from R5 populations more rapidly during maraviroc treatment, or mayhave  emerged  between  screening  and  enrollment,  as  previously  reported  in  8%  of  MOTIVATEparticipants  289.  Indeed,  among  the  studied  population,  25  patients  had  switched  to  non-R5phenotypes at maraviroc initiation (14%). It is also possible that non-R5 variants may also emergefrom compartments other than blood plasma. This study was limited in its ability to better characterize maraviroc resistance. To date, a reducedmaximal percentage inhibition (MPI) in  a phenotypic assay assessing susceptibility of the patientvirus  to  maraviroc  is  the  only  consistent  characteristic  of  maraviroc  resistance  373;  no  signaturemutations have been observed for maraviroc  374,375 or other CCR5 antagonists  376.   Similar to theseprior fndings, no consistent patterns of mutations were noted that were associated with virologicfailure  while  maintaining  an  R5  population.  This  may  be  due  to  a  number  of  factors,  such  asinsuffcient  time on the medication to induce resistance-associated mutations,  the possibility thatmutations  may  emerge  outside  the  V3  loop,  and/or  the  possibility  that  maraviroc  resistancemutations  are  patient-specifc  and  diffcult  to  generalize.  Furthermore,  only  a  small  number  ofpatients who experienced failure on maraviroc with R5 viruses have actually been shown to havemaraviroc-resistant  HIV in phenotypic  assays  374.  In the remaining patients,  other factors such asadherence or resistance to the other agents in their regimens may be involved. The  aforementioned analyses  indicate  that  maraviroc  treatment  dichotomized  V3  sequences  andinferred coreceptor usage. Genotypic tropism analyses demonstrated large decreases in geno2phenovalues, and large increases in the percentage of non-R5 variants. Patients with non-R5 HIV at failureexperienced suboptimal virologic responses to maraviroc in this study, likely driven by their non-R5status. However, genotypic analysis in patients with failure R5 results was not informative, with littlechange  in  geno2pheno  values,  and  few  amino  acid  substitutions  that  might  be  attributed  tomaraviroc  resistance.  In  contrast,  the  results  in  the  non-R5  population  were  unambiguous  and126striking,  suggesting  that  both  deep  and  population-based  sequencing  approaches  are  usefulmonitoring tools for patients receiving maraviroc. While  this  and all  previous chapters  have  assessed deep sequencing as  applied to patients  withprevious  treatment  experience,  Chapter  6  will  extend  this  application  to  patients  beginningantiretroviral therapy for the frst time. This treatment-naïve population may be especially relevantsince CCR5-using HIV-1 tends to be most prevalent early in infection  145,146.  It may therefore be apreferred population in which to prescribe maraviroc. Additionally, the next chapter aims to morethoroughly assess deep sequencing in comparison to a phenotypic assay with higher sensitivity thanthe original Trofle assay The following chapter also represents a completely independent validationof  next-generation  sequencing  methods  since  they  were  originally  optimized  in  the  treatment-experienced population described previously. 127Chapter 6:     Deep V3 Sequencing for HIV-1 Tropism in Treatment-Naïve Patients: A Reanalysis of the MERIT Trial of Maraviroc 6.1     Background & IntroductionHuman Immunodefciency Virus type 1 (HIV-1) infects cells using the CD4 receptor and a coreceptor.The  chemokine  receptor  CCR5 is  a  necessary  coreceptor  for  strains  of  HIV called  R5  140,  whichpredominate in antiretroviral-naïve individuals 139,155,173. The CCR5 coreceptor is also the target of theHIV entry inhibitor, maraviroc. Maraviroc inhibits the ability of HIV to interact with and infect cellsvia CCR5 108. The use of an alternative coreceptor emerges in approximately half of clade-B-infectedindividuals  100. Therefore, a tropism test is performed prior to maraviroc administration to excludepatients whose viral population (or some subpopulation of it) is non-R5 and unlikely to respond tomaraviroc 67. A number of genotypic HIV tropism approaches have been developed to provide alternatives  tophenotypic  tropism  assays  such  as  the  Monogram  Biosciences  Trofle  assay  377 and  EnhancedSensitivity Trofle Assay (ESTA) 195. Commonly, genotypic approaches use the sequence of the thirdvariable (V3) region of the HIV gp120 gene, since the V3 loop itself interacts with the HIV coreceptor122 and  mutations  encoded  by  V3  are  associated  with  measurable  changes  in  HIV-tropism  123,324.Tropism is then inferred using a bioinformatic algorithm such as geno2pheno 197,204.While population-based genotypic tropism assays can infer the coreceptor usage of a patient’s mostcommon HIV quasispecies,  these tests may miss non-R5 variants  comprising low-level minoritieswithin a predominantly R5 population 229. The ability to detect minority non-R5 variants is importantbecause these subpopulations may undergo selection by maraviroc treatment and lead to virologicfailure 232,291,316.128There have been four large clinical trials of maraviroc to date 288–291. The Maraviroc versus EfavirenzRegimens  as  Initial  Therapy  (MERIT)  trial  assessed  two  doses  of  maraviroc  (pluslamivudine/zidovudine) in antiretroviral-naïve patients, with a comparator arm of efavirenz (pluslamivudine/zidovudine)  290.  The trial consisted only of patients with R5 HIV at screening by theoriginal Trofle assay. The maraviroc once-daily (QD) arm was discontinued early after failing to meetpre-specifed effcacy criteria.Although superior to placebo in the MOTIVATE trials,  maraviroc was inferior to efavirenz in theprimary analysis of the MERIT trial using the original screening population. However, when patientsin MERIT were retrospectively re-screened using the higher-sensitivity ESTA, with exclusion of thosenow identifed with non-R5 HIV, maraviroc twice daily (BID) was non-inferior to efavirenz for theprimary study endpoint 290,292.Deep sequencing  refers  to  the  application  of  next-generation sequencing  technology  such  as  theGenome  Sequencer  FLX  (GS-FLX)  240.  The  GS-FLX  can  simultaneously  sequence  and  quantifythousands of individual variants within a viral population, allowing an in-depth quantifcation of theproportion  of  non-R5  variants  in  a  given  sample  232,352,  and therefore  the  proportion  unlikely  torespond to maraviroc  348. The aim of this study was to assess whether the high-sensitivity of deepsequencing could also have been a successful screening tool for the treatment-naïve patients in theMERIT trial.6.2     Materials & Methods6.2.1     Samples & MERIT Trial DesignA total of 859 plasma screening samples from the MERIT trial were examined. All samples were R5by the original  Trofle assay.  Most patients entered either  the maraviroc BID arm (N=347) or theefavirenz arm (N=346). The trial’s primary endpoint was the proportion of patients with a viral load<50 HIV RNA copies/mL at week 48. A third arm consisting of maraviroc QD was also partially129enrolled. Screening samples from those initially assigned to the maraviroc QD arm (N=166) were alsotested.6.2.2     V3 Amplification MethodHIV RNA was extracted from 500 µL of each  of the  859 stored screening plasma samples  usingautomated  extraction methods with a  NucliSENS easyMAG (bioMérieux).  One-step  RT-PCR wasperformed  in  triplicate  using  4  µL  of  sample  extract  per  amplifcation.  A  second-round  PCRamplifcation was then performed using customized primers to allow multiplexing (48 samples persequencing run). PCR amplifcations were then quantifed. Each amplifcation was combined in equalproportions with the others to a concentration of 2 × 1012 DNA molecules per sample. This combinedset of PCR products then underwent emulsion PCR and deep sequencing with a GS-FLX. A detailedmethodology has been published 333,348 and is detailed in previous chapters.In addition, a second round PCR amplifcation was also performed using the same triplicate RT-PCRtemplate. These PCR products underwent individual standard, population-based sequencing on anABI 3730 XL DNA analyzer according to previously described methods 325,333,348,366.6.2.3     Bioinformatic AnalysesThe  false  positive  rate  (FPR)  cutoff  for  geno2pheno  tropism  assignments  had  previously  beenoptimized and validated  in  the  maraviroc treatment-experienced trials,  as  had the  cutoff  for  thepercentage of non-R5 variants needed for a sample to be classifed as non-R5 335–337,348. A sample wasconsidered  R5  if  fewer  than  2%  of  the  variants  detected  using  deep  sequencing  fell  below  ageno2pheno FPR of 3.5 335. Population-based V3 sequencing used a geno2pheno FPR cutoff of 5.75 336.1306.2.4     Ethics StatementWritten,  informed  consent  was  obtained  from all  individuals,  including  consent  to  allow  othertropism testing to be performed on their samples. The University of British Columbia-ProvidenceHealth Care Research Ethics Board reviewed the research project and granted ethical approval. Alldata were analyzed anonymously.6.2.5     Data AnalysisThe maraviroc BID arm was the primary dataset for assessing the utility of deep sequencing. Theefavirenz  arm  served  as  a  comparator.  The  maraviroc  QD  arm  was  also  examined  as  acomplementary analysis. Unless otherwise stated, any reference to maraviroc should be taken as areference to maraviroc BID. Virologic outcomes examined included the viral load change from baseline, the percentage of patientswith virologic suppression, and a time to a change in a patient’s Trofle result from R5 to non-R5 (i.e.,a tropism “switch”). Where data were missing, the last observation was carried forward, except in thecase of the percentage of patients with a pVL <50 copies, where missing values were imputed to be>50 (“failures”). Deep sequencing was also compared to the performance of both ESTA and standardpopulation-based sequencing in the same dataset. Differences between tropism groups (R5 versus non-R5) were tested for statistical signifcance usingthree  tests.  The Mann-Whitney test  tested for  statistically  signifcant  differences  between medianmeasurements,  such as  median pVL declines.  The Fisher’s  exact test  examined differences in theproportions  of  patients,  such  as  differences  in  virologic  suppression  or  clade.  The  log-rank  testexamined differences in the Kaplan-Meier curves for tropism changes.  No statistical  comparisonsbetween the  populations  defned by Trofle  and deep sequencing  were  performed because  thesepopulations were not independent.1316.3     Results6.3.1     Patient CharacteristicsBaseline  characteristics  of  patients  stratifed  by  deep  sequencing  tropism  result  at  screening  arelargely  similar  to  the  original  MERIT population  290.  Those  found to  have  non-R5  HIV by deepsequencing were more likely to be white,  MSM, infected with clade-B HIV, and have lower CD4counts than those found to have R5 HIV by deep sequencing, though these differences were relativelyminor (Table 6.1).6.3.2     Identification of Non-R5 Screening Samples Using Deep SequencingDeep sequencing generated a mean of 5002 sequences per sample (median: 4529; Inter-quartile Range[IQR]:  3715  –  6024).  Sequence  depth  did  not  have  a  discernable  impact  on  deep  sequencing’ssensitivity or ability to predict virologic outcomes (data not shown). Overall, re-screening MERITpatients using deep sequencing classifed an additional 10% of maraviroc BID recipients (35/347) asbeing  unlikely  to  respond  to  their  regimens  due  to  the  presence  of  ≥2% non-R5  virus  prior  totreatment. Similarly, 13% (22/166) in the maraviroc QD arm, and 9% (30/346) in the efavirenz armwere classifed as having non-R5 HIV by deep sequencing.Samples screened non-R5 by deep sequencing had non-R5 variants at a median proportion of 20.9%(IQR: 5.4 – 44.1). Samples screened R5 had a median of 0% non-R5 HIV (IQR: 0 – 0). Seventy-fourpercent  of  patients  (511/693)  treated with maraviroc  BID or  efavirenz had no detectable non-R5variants at screening by deep sequencing. Additionally, 60% of all non-R5 samples had >10% non-R5variants by this method, despite having been already pre-screened with the Trofle assay, which has areported cut-off of 10% non-R5 virus 334. There were a total of 94 maraviroc recipients with detectablenon-R5 virus by deep sequencing. When the non-R5 prevalence was extended to the absolute amountof non-R5 at screening, these patients had a median non-R5 viral load of 2.9 log10 copies/mL (IQR: 2.2– 3.5).132Table 6.1: Baseline Patient Characteristics in the MERIT trialTable 6.1: Baseline Patient Characteristics in the MERIT Trial. Baseline patient characteristics for the study population. MERIT participants are shown for the groupas a whole (N=693), as well as for the groups who were classifed as R5 or non-R5 by deep sequencing. The difference in baseline characteristics between the R5 and non-R5 groupswere also tested for their statistical signifcance. Het — heterosexual; MSM — Men who have sex with men; IDU — Injection Drug Use; pVL — plasma viral load; IQR — interquartilerange; n.s. — not signifcant.Combined MVC BID and EFV arms (N=693)Deep Sequencing Non-R5 (N=65)Deep Sequencing R5 (N=628)Statisticalsignificancep valueAge — median(range) 36 (18 – 77) 39 (21 – 68) 36 (18 – 77) n.s. 0.09Male sex - no. (%) 503 (73) 53 (82) 450 (72) n.s. 0.11Race or ethnicity —no. (%)White - 394 (57); Black - 238(34), Asian, other – 61 (9)White - 45 (69); Black - 13(20); Asian, other – 7 (11)White - 349 (56); Black - 225(36); Asian, other – 54 (9)0.04(white versus non-white)Clade — no. (%) B 414 (60); C – 205 (30); Other - 74 (11)B – 48 (74); C – 10 (15); Other - 7 (11)B - 366 (58); C - 195 (31); Other - 67 (11)0.02 (B versus non-B)Mode ofTransmission —no. (%)Het - 328 (47); MSM - 292(42); IDU - 48 (7); Other - 25(4)Het - 23 (36); MSM - 37(58); IDU - 0 (0); Other - 4(6)Het- 305 (49); MSM - 254(40); IDU - 48 (8); Other - 21(3)0.01 (MSM versus non-MSM)Median baselinepVL —  log10copies/mL (IQR)5.0 (4.5 – 5.3) 5.0 (4.5 – 5.3) 4.9 (4.4 – 5.2) n.s. 0.54Median baselineCD4 count —cells/mm3 (IQR)251 (183 – 323) 236 (135 – 300) 252 (185 – 327) 0.031336.3.3     Viral Load Decline from BaselineOverall viral load declines from baseline through 96 weeks are shown for both arms in Figures 6.1and 6.2, with patients grouped according to their deep sequencing result. Patients with R5 results hadsimilar responses in both the maraviroc and efavirenz arms (Figure 6.1), but patients with non-R5results had expectedly poorer responses in the maraviroc arm compared to the efavirenz arm (Figure6.2). Maraviroc recipients with R5 HIV by deep sequencing showed a median 2.7 log 10 decline in pVL frombaseline to week 8 (IQR: 2.3 – 3.1), while the non-R5 group had a smaller decline: 2.3 log 10 (IQR: 1.9 –2.6),  p<0.0001.  The  efavirenz  arm  had  similar  virologic  responses  as  the  R5-infected  maravirocrecipients, regardless of tropism assessment by deep sequencing:  2.8 log10 (IQR: 2.4 – 3.2) for R5, and2.9 log10 (IQR: 2.5 – 3.2) for non-R5, p=0.56. 6.3.4     Virologic SuppressionThe larger pVL changes observed when patients were classifed using the deep sequencing methodwas also reflected in the percentage of patients who achieved an undetectable viral load at 48 weeks.Where deep sequencing had indicated R5 HIV at screening, a total of 67% of maraviroc recipients(208/312) had a pVL <50 HIV RNA copies/mL at week 48 (i.e., virologic suppression). In contrast,only 46% of non-R5-infected maraviroc recipients (16/35) achieved week 48 virologic suppression,p=0.02 (Figure 6.3). In terms of non-R5 viral load, the percentage of maraviroc recipients with week48 virologic suppression was: 68% (173/255) of those with <1 log10 non-R5 copies/mL; 77% (10/13)with 1-2 log10; 56% (22/39) with 2-3 log10; 52% (14/27) with 3-4 log10; and 38% (5/13) of those with >4log10 non-R5 copies/mL. In the efavirenz arm, suppression was 69% (219/316) in those with R5 HIV— similar to the maraviroc arm. This was 70% (21/30) in efavirenz recipients with non-R5 HIV bydeep sequencing, p=n.s.. The percentages of patients achieving virologic suppression for both armsare shown in Figure 6.3, with data to week 96.134Figure 6.1: Median Decline in Plasma Viral Load from Baseline in Patients Screened with R5 HIV by Deep Sequencing Who Received Either Maraviroc Twice-Daily or EfavirenzFigure 6.1: Median Decline in Plasma Viral Load from Baseline in Patients Screened with R5 HIV by Deep Sequencing Who Received Either Maraviroc Twice-Daily or Efavirenz. The log10-transformed decline in plasma viral load for R5 HIV-infected patients. The blue line indicates patients receiving maraviroc BID (N=312), and the yellowline indicates those receiving efavirenz (N=316). With screening by deep sequencing, both groups had a median pVL decline from baseline of approximately 3 log 10 HIV RNAcopies/mL, which was sustained to week 96. 135Figure 6.2: Median Decline in Plasma Viral Load from Baseline in Patients Screened with Non-R5 HIV by Deep Sequencing Who Received Either Maraviroc Twice-Daily or EfavirenzFigure 6.2: Median Decline in Plasma Viral Load from Baseline in Patients Screened with Non-R5 HIV by Deep Sequencing Who Received Either Maraviroc Twice-Daily orEfavirenz. The log10-transformed decline in plasma viral load for non-R5 HIV-infected patients. The blue line indicates patients receiving maraviroc BID (N=35),and the yellow line indicates those receiving efavirenz (N=30). With screening by deep sequencing, those found to have non-R5 HIV had lower pVL declines from baseline whentreated with maraviroc BID versus efavirenz.  136Figure 6.3: Percentage of Maraviroc Twice-Daily & Efavirenz Recipients with Plasma Viral Loads Less than 50 Copies/mL with R5 HIV at Screening by Deep SequencingFigure 6.3: Percentage of Maraviroc Twice-Daily & Efavirenz Recipients with Plasma Viral Loads Less than 50 Copies/mL with R5 HIV at Screening by Deep Sequencing.  The black line indicates R5 HIV-infected patients receiving maraviroc BID (N=312), and the dashed-dotted line indicates those receiving efavirenz (N=316). Similarpercentages of patients  had virologic suppression at  week 48 in the two treatment arms when patients with non-R5 HIV at  screening by deep sequencing were excluded.  Thepercentage of patients with virologic suppression at week 48 is indicated for both the maraviroc (MVC) and efavirenz (EFV) arms.1376.3.5     Non-Inferiority AnalysisIn the original MERIT study, maraviroc was found to be inferior to efavirenz. This analysis was basedon  comparing  the  differences  in  the  percentage  of  patients  achieving  virologic  suppression  <50copies/mL in the maraviroc and efavirenz arms, when these patients were screened by the originalTrofle  assay.  The  criterion  was  that  the  lower  bound  of  the  97.5%  confdence  interval  for  thedifference between arms must not fall below -10 for the maraviroc arm, but this criterion was not met.In contrast, when patients in the current study were tested using deep sequencing, the lower boundof the 97.5% confdence interval for the difference between arms was -8.67; less than the pre-specifedminimum value of -10 for determining non-inferiority of maraviroc at week 48 (Table 6.2). Table 6.2: Non-Inferiority Analysis between the Maraviroc & Efavirenz ArmsTable 6.2: Non-Inferiority Analysis between the Maraviroc & Efavirenz Arms. Non-inferiority analysis comparingthe maraviroc BID arm to the efavirenz arm. The lower confdence bound of the difference between drug arms for R5 screenedparticipants was smaller than -10%, indicating non-inferiority between arms. MVC — maraviroc; EFV — efavirenz; Adj Diff —adjusted difference; LCB — lower confdence bound; ESTA — enhanced sensitivity Trofle assay; DeepSeq — Deep SequencingNumber and proportion of virologic successes at 48 weeksMaraviroc BID arm Efavirenz arm Stratifiedn N % n N %Rawdiff(MVC -EFV) Adj Diff97.5%LCBAssayresultDeepSeq R5 210 312 67.31 217 316 68.67 -1.36 -1.48 -8.67DeepSeqnon-R5 17 35 48.57 21 30 70.00 -21.43 -42.19 -60.71ESTA R5 205 300 68.33 196 290 67.59 0.75 0.17 -7.21ESTA non-R5 22 47 46.81 42 56 75.00 -28.19 -31.15 -48.87Trofile R5 227 347 65.42 238 346 68.79 -3.37 -3.73 -10.61138This analysis also confrms the poor virologic response among maraviroc recipients screened withnon-R5 HIV by deep sequencing, compared to those who received efavirenz. Together, these analysessuggest that had patients been screened with deep sequencing rather than the original Trofle assay,the maraviroc BID arm would have likely been found to be non-inferior to the efavirenz arm.6.3.6     Changes in HIV TropismMaraviroc administration unmasks and can select non-R5 virus that was present prior to maravirocadministration  316.  Maraviroc-recipients with non-R5 HIV by deep sequencing were more likely tochange phenotypic tropism over the course of the study compared to those with R5 HIV by deepsequencing (p<0.0001). Of those with non-R5 HIV, 43% (15/35) changed their Trofle result from R5 tonon-R5 between screening and 96 weeks, versus only 7% (23/312) of the deep sequencing R5 group(Figure 6.4).In the non-R5 group, patients switched tropism a mean of 5 weeks after beginning treatment. Thiswas earlier than the 17 weeks seen in the R5 group. Maraviroc recipients who changed tropism alsohad a higher proportion of non-R5 variants present pre-treatment according to deep sequencing, witha median of 0.8% non-R5 variants (IQR: 0.0 – 7.4%) versus 0% (IQR: 0 – 0%) for those who did notchange tropism. Deep sequencing was able to detect at least low levels (>0%) of non-R5 HIV in amajority, 61%, of maraviroc recipients who switched tropism, versus 23% of those who did not switchtropism.6.3.7     Effects of HIV SubtypeFor all patients analysed in the current study, 60% had HIV-1 clade-B, 29% had clade-C, and 11% hadother  clades  of  HIV.  Non-R5  tropism  seemed  to  be  overrepresented  amongst  clade-B-infectedindividuals, with 74% of the deep sequencing non-R5 group consisting of clade-B-infected patients,higher  than  the  overall  clade-B  composition  of  60%,  p=0.02.  Conversely,  clade-C  wasunderrepresented amongst non-R5-infected patients, at 15%, p=0.001. 139Figure 6.4: Time to Change in Tropism for Maraviroc Twice-Daily RecipientsFigure 6.4: Time to Change in Tropism for Maraviroc Twice-Daily Recipients. This analysis examined the likelihood of a change from an original Trofle assay result of R5 to non-R5 over thecourse of the study. The upper green line indicates patients screened with R5 HIV by deep sequencing (N=312). The lower redline indicates those screened with non-R5 HIV by deep sequencing (N=35). The upper solid black and lower dotted black linesindicate the ESTA-R5 (N=300) and ESTA-non-R5 (N=47) groups. Patients screened with non-R5 HIV by either assay were morelikely to change Trofle results to non-R5 during the study. The numbers of patients remaining at risk for a change in theirTrofle result are shown below the week numbers. The scale of the Y axis has been modifed for display purposes.140Global concordance in the entire study population between deep sequencing and ESTA was 79% inthe clade B-infected population and 87% in the non-clade B-infected population. Importantly, bothdeep sequencing and ESTA had similar performance in predicting virologic outcome in both clade-Bnon-clade-B-infected patients (Tables 6.3 & 6.4).Table 6.3: Virologic Outcomes in Maraviroc Recipients Infected with Subtype B HIV-1, Stratified by Tropism Assessment by Deep Sequencing & the Enhanced Sensitivity Trofile AssayAssay Tropism CallWeek 8 pVL decline,log10 scaleMedian (IQR)Week 48 virologicsuppression%, n/NDeep sequencingR5 2.7 (2.4 – 3.1)72% (132/183)Non-R5 2.3 (2.0 – 2.6)46% (12/26)ESTAR5 2.7 (2.3 – 3.0)73% (128/176)Non-R5 2.6 (2.1 – 2.9)48% (16/33)Table 6.4: Virologic Outcomes in Maraviroc Recipients Infected with Non-Subtype B HIV-1, Stratified by Tropism Assessment with Deep Sequencing & the Enhanced Sensitivity Trofile AssayAssay Tropism CallWeek 8 pVL decline,log10 scaleMedian (IQR)Week 48 virologicsuppression%, n/NDeep sequencingR5 2.8 (2.2 – 3.1)59% (76/129)Non-R5 1.9 (1.9 – 2.4)44% (4/9)ESTAR5 2.8 (2.2 – 3.1)60% (75/124)Non-R5 2.2 (1.7 – 2.9)36% (5/14)1416.3.8     Comparison of Deep Sequencing to the Enhanced Sensitivity Trofile Assay & Population-Based Sequencing Maraviroc recipients rescreened with R5- and non-R5 HIV using ESTA had week 8 pVL declines of2.7  (IQR:  2.3 –  3.1)  and 2.4 (IQR:  1.9 –  3.0),  respectively,  similar  to  the  deep sequencing results.Likewise, the percentage of patients who achieved virologic suppression on maraviroc was similarregardless of the assay used to determine tropism (Figures 6.5 & 6.6). When compared with each other,  deep sequencing and ESTA had a  global  concordance of  82%.Perhaps surprisingly, only 22 samples of the 693 total (3%) were called non-R5 by both methods, oronly 15% of the 146 samples called non-R5 by either method. Consequently, both assays had lowsensitivity relative to the other. Deep sequencing had 21% sensitivity and 93% specifcity using ESTAas a reference; ESTA had 34% sensitivity and 87% specifcity using deep sequencing as a reference.Despite this,  the groups called R5 and non-R5 by either  method had similar  virologic  outcomesregardless  of  the  assay.  Retrospective  screening  by  ESTA identifed 14% of  maraviroc  recipients(47/347) as having non-R5 HIV. This was 16% (56/346) in the efavirenz arm. Overall, the ESTA-non-R5  group  had  a  median  of  0% non-R5  HIV (IQR:  0  –  0.8%;  mean  =  7.4%),  according  to  deepsequencing using the geno2pheno algorithm; the ESTA-R5 group also had a median of 0% (IQR: 0 –0%; mean = 2.2%). Deep sequencing was also compared to population-based sequencing where available. Population-based sequencing was concordant with deep sequencing in 93% of cases (638/688), and gave 54%sensitivity relative to deep sequencing. Samples identifed by population-based sequencing as non-R5had a median of 9.1% non-R5 variants in their deep sequencing result (IQR: 0.7 – 41.0%; mean =26.3%). Virologic responses of patients grouped by discordance of deep sequencing with ESTA orpopulation-based sequencing are shown in Table 6.5.142Figure 6.5: Median Decline in Plasma Viral Load from Baseline in Maraviroc Twice-Daily Recipients with Screening by Deep Sequencing & the Enhanced Sensitivity Trofile AssayFigure 6.5: Median Decline in Plasma Viral Load from Baseline in Maraviroc Twice-Daily Recipients with Screening by Deep Sequencing & the Enhanced Sensitivity TrofileAssay. The green line indicates maraviroc BID recipients screened with R5 HIV by deep sequencing (N=312). The red line indicates those screened with non-R5 HIV bydeep sequencing (N=35). The solid black and dashed black lines indicate the ESTA-R5 (N=300) and ESTA-non-R5 (N=47) groups, respectively. Deep sequencing and ESTA performedsimilarly in terms of distinguishing between virologic responders and non-responders on maraviroc.143Figure 6.6: Percentage of Maraviroc Twice-Daily Recipients with Plasma Viral Loads below 50 Copies/mL with Screening by Deep Sequencing & the Enhanced Sensitivity Trofile AssayFigure 6.6: Percentage of Maraviroc Twice-Daily Recipients with Plasma Viral Loads below 50 Copies/mL with Screening by Deep Sequencing & the Enhanced Sensitivity TrofileAssay. The green line indicates maraviroc BID recipients screened with R5 HIV by deep sequencing (N=312). The red line indicates those screened with non-R5 HIV bydeep sequencing (N=35). The solid black and dashed black lines indicate the ESTA-R5 (N=300) and ESTA-non-R5 (N=47) groups144Table 6.5: Overall Virologic Responses of Maraviroc Recipients Grouped by Discordance between Deep Sequencing & the Enhanced Sensitivity Trofile Assay or Population-Based SequencingConcordance orDiscordanceDeepSeq Resultversus Other assayresultMedian week 8 log10 pVLchange from baseline, (IQR)Deep Sequencing versus:Percentage of patients with week 48virologic suppression, (n)Deep Sequencing versus:ESTAPopulation-basedsequencingESTA Population-basedsequencingR5/R52.7 (2.3 – 3.1)2.7 (2.3 – 3.1)68% (188/276)67% (202/301)R5/Non-R52.6 (2.2 – 3.1)2.8 (2.1 – 3.1)56% (20/36)50% (4/8)Non-R5/R52.4 (2.2 – 2.7)2.4 (1.9 – 2.7)63% (15/24)47% (8/17)Non-R5/Non-R51.9 (1.3 – 2.1)2.1 (1.9 – 2.3)9% (1/11)44% (8/18)Table 6.5: Overall Virologic Responses of Maraviroc Recipients Grouped by Discordance between Deep Sequencing & the Enhanced Sensitivity Trofile Assay or Population-Based Sequencing. pVL — plasma viral load; IQR — interquartile range; ESTA — enhanced sensitivity Trofle assay; DeepSeq — Deep SequencingIn addition, virologic responses were compared with classifcation by deep sequencing versus ESTAor  population-based  sequencing.  Overall,  where  screening  assays  differed,  there  was  no  clearindication  as  to  which  assay  was  the  “gold  standard”.  Indeed,  deep  sequencing,  ESTA,  andpopulation-based sequencing all performed quite similarly in terms of predicting virologic responseto maraviroc in this population. There was a possible trend towards slightly superior predictions bydeep sequencing (Figures 6.7 & 6.8). 145Figure 6.7: Declines in Plasma Viral Load from Baseline in Patients with Concordant & Discordant Results between Tropism AssaysFigure 6.7: Declines in Plasma Viral Load from Baseline in Patients with Concordant & Discordant Results between Tropism Assays. Left:  Panel  A  shows  theplasma viral load decline on maraviroc for patient groups where deep sequencing and ESTA gave the same (concordant) and different (discordant) tropism calls. The solid blue andsolid orange lines indicate the concordant R5 and non-R5 groups, respectively. The black dashed line indicates the group called R5 by deep sequencing but non-R5 by ESTA. The greydashed line indicates the group called non-R5 by deep sequencing but R5 by ESTA. Right: Panel B shows the median plasma viral load change from baseline in maraviroc recipients.Patients are grouped where deep sequencing and population-based sequencing gave the same (concordant) or different (discordant) tropism calls. The solid blue and solid orange linesindicate the concordant R5 and non-R5 groups, respectively. The black dashed line indicates the group called R5 by deep sequencing but non-R5 by population-based sequencing. Thegrey dashed line indicates the group called non-R5 by deep sequencing but R5 by population-based sequencing.146Figure 6.8: Proportion of Patients with Plasma Viral Loads below 50 Copies/mL in Groups with Concordant & Discordant Results between Tropism AssaysFigure 6.8: Proportion of Patients with Plasma Viral Loads below 50 Copies/mL in Groups with Concordant & Discordant Results between Tropism Assays. Left:Panel A shows the percentage of patients on maraviroc who had viral loads below 50 copies/mL. Patients are grouped where deep sequencing and ESTA gave the same (concordant)or different (discordant) tropism calls. The solid blue and solid orange lines indicate the concordant R5 and non-R5 groups, respectively. The black dashed line indicates the groupcalled R5 by deep sequencing but non-R5 by ESTA. The dashed line indicates the group called non-R5 by deep sequencing but R5 by ESTA. Right: Panel B shows the percentage ofpatients on maraviroc who had viral  loads below 50 copies/mL. Patients are grouped where deep sequencing and population-based sequencing gave the same (concordant) ordifferent (discordant) tropism calls. The solid blue and solid orange lines indicate the concordant R5 and non-R5 groups, respectively. The dashed black line indicates the group calledR5 by deep sequencing but non-R5 by population-based sequencing.  The dashed grey line indicates the group called non-R5 by deep sequencing  but R5 by population-basedsequencing.1476.3.9     Maraviroc Once-Daily ArmThe  patients  who  were  randomized into  the  maraviroc  QD arm were  also  examined with deepsequencing (N=166). This dataset served as an independent validation of the deep V3 sequencingmethod. The maraviroc QD arm was originally discontinued partway through the MERIT study dueto a protocol-defned lack of demonstrated non-inferiority to efavirenz. Maraviroc QD recipients werethen allowed to switch to maraviroc BID for the remainder of the study. The performance of deepsequencing as a screening tool for tropism was assessed in this population (Figure 6.9). Analyses wereperformed where responses were censored or uncensored after patients switched to maraviroc BID.The week 8 log10 pVL declines from baseline were similar between the maraviroc QD and BID arms inthe uncensored analysis. The median decline of those screened as having R5 HIV (N=144) was 2.8log10 (IQR: 2.4 – 3.1) versus 2.6 log10 (IQR: 1.3 – 3.0) for those with non-R5 HIV (N=22). Note that 26patients  in  the  R5  group  (18%)  and 6  in  the  non-R5  group  (27%)  had  discontinued  therapy  orswitched to maraviroc BID by week 8. Viral load declines from baseline for the uncensored groupsare shown in Figure 6.9. The R5-group censored for those remaining on maraviroc QD is also shownand gave similar results (Figure 6.9).6.4     Discussion & ConclusionsThis study represents the frst large clinical comparison of two highly sensitive HIV tropism assays:deep sequencing and ESTA, and the largest application of deep sequencing in antiretroviral-naïvepatients to date. Retrospective screening by deep sequencing, with removal of patients classifed withnon-R5 HIV, led to similar rates of week 48 virologic suppression between the maraviroc BID andefavirenz arms. Maraviroc recipients screened with R5 HIV by this approach had larger on-treatmentpVL declines,  were  more  likely  to  achieve  virologic  suppression,  and were  less  likely  to  changetropism than those screened with non-R5 virus. 148Figure 6.9: Median Decline in Plasma Viral Load from Baseline in Maraviroc Once-Daily RecipientsFigure 6.9: Median Decline in Plasma Viral Load from Baseline in Maraviroc Once-Daily Recipients. The  viral  load  changes  for  patients  in  the  discontinuedmaraviroc once-daily (QD) arm. The red line indicates those for whom deep sequencing indicated non-R5 HIV at screening. The solid green line indicates those patients classifed ashaving R5 HIV at screening, with patients censored when they were switched to maraviroc twice-daily dosing. The dashed green line indicates the same patients as the R5 group, butwith continued follow-up after switching to twice-daily from once-daily dosing.149Deep sequencing also had similar performance to ESTA, which is widely used in the clinic. Virologicresponses were similar between groups that had discordant results by either assay, suggesting thatneither assay is signifcantly more “correct” than the other. Interestingly, the decline in viral loadfrom baseline was greater than 2 log10 copies/mL even in the maraviroc-treated non-R5 group. This islikely  due to  the  activity  of  the  background zidovudine-lamivudine,  and perhaps  some residualactivity of maraviroc. The additional clinical utility of deep sequencing over standard population-based sequencing wasnot clearly demonstrated in this study, despite a possible trend in a previous study in treatment-experienced patients 348. In fact, concordance was over 90% between the methods in the current study,though the sensitivity of population-based sequencing remained low relative to deep sequencing. Acommon critique of bioinformatic algorithms for HIV tropism is that most are trained primarily onclade  B  sequences.  However,  the  deep  sequencing  genotypic  assay  presented  here  performedsimilarly to the phenotypic ESTA assay in MERIT, including in non-clade-B-infected patients, lendingconfdence to the utility of this approach in such populations (see also 378).Some limitations of this study and the use of deep sequencing should be acknowledged. The MERITtrial itself only included patients pre-screened as having R5 HIV by the original Trofle assay, so ananalysis of maraviroc treatment in an antiretroviral-naïve population infected with non-R5 virus bythe Trofle assay was not possible,  though an analysis of deep sequencing in a non-R5 treatment-experienced trial has been published  348 and is presented in Chapter 3. The pre-screening of thesepatients may also have diminished the ability to demonstrate improved tropism prediction of anyassay over any other given the small number of patients  re-screened as having non-R5 HIV. Theanalysis of the maraviroc QD arm should also be examined with caution given the small number ofpatients continuing once-daily maraviroc treatment.  Finally,  the deep sequencing method itself  iscostly in both time and capital, which currently limits its utility in clinical settings.Overall, deep sequencing is a useful tool for distinguishing between probable responders and non-responders  to  maraviroc.  This  high  sensitivity  method  performed  similarly  to  ESTA,  which  is150currently the most commonly used clinical phenotypic tropism assay. Had deep sequencing beenused to screen patients, maraviroc would have likely been found to be non-inferior to efavirenz in theMERIT trial.The next and fnal chapter of this thesis will summarize the fndings of the previous chapters, andremark on conclusions and future implications that these results may have within the HIV treatmentfeld. 151Chapter 7:     General Discussion & Conclusions7.1     Thesis Summary & Overall ConclusionsThis thesis described a number of aspects of HIV tropism. Next-generation sequencing was used tobetter evaluate coreceptor usage and its clinical relevance in patients treated with medication thatspecifcally antagonizes the HIV-1 coreceptor, CCR5. This approach was able to accurately determineHIV tropism in a number of patient populations, including an observational cohort, and a total offour large-scale, randomized clinical trials of the CCR5 antagonist maraviroc. Deep sequencing eithermet or surpassed the performance of a number of alternative tropism assays, including population-based sequencing, the original Trofle assay, and the Enhanced Sensitivity Trofle assay. This approach could both accurately quantify the coreceptor usage of patients’ viral populations aswell as predict the virologic responses that those patients would experience while receiving treatmentwith  maraviroc.  Next-generation  sequencing  could  be  performed  using  both  the  plasma  andperipheral  blood mononuclear  cell  compartments to test  either  HIV RNA or DNA. Patients  withCXCR4-using HIV as  assessed by next-generation sequencing had smaller  viral  load declines  onmaraviroc,  were  less  likely  to  achieve  a  suppressed  viral  load,  and were  more  likely  to  switchphenotypic  tropism results  to non-R5 while on maraviroc-containing regimens.  Furthermore, in amajority of cases, the pre-treatment assessments were phylogenetically related to the viral outgrowthpopulations following maraviroc administration, indicating that the CXCR4-using variants detectedprior to treatment with maraviroc did in fact give rise to those which would later emerge duringtreatment failure. A major advantage of the use of next-generation sequencing in the treatment of HIV is the flexibilityof the technique. Not only can it be used to assess viral coreceptor usage, it can also simultaneouslyquantify drug resistance to all known classes of antiretrovirals. Increasingly, there are opportunitiesto sequence the entire genome of HIV on a single next-generation sequencing platform. Moreover, thegenome or exome of the host may also be sequenced in parallel with the virus, enabling clinicians to152gain insight into potential genetic factors that may influence response to therapy or risk for adverseevents.  This  flexible  and  high-output  approach  is  now  standard-of-care  for  HIV  tropismdetermination in many settings including Canada and much of Europe. It will likely also becomeincreasingly used to quantify other types of drug resistance in order to further personalize and tailorHIV treatment to the individual patient. 7.2     Specific Conclusions of the Thesis7.2.1     Conclusions for Chapter TwoChapter  2  provided the  foundation  for  all  of  the  studies  detailed  in  the  subsequent  chapters  Itestablished that genotypic tropism testing was viable for predicting phenotypic assay results. Thisstudy also established that higher sensitivities for detecting CXCR4-using HIV could be achievedthrough  the  use  of  triplicate  amplifcation  methods  and  deep  sequencing.  The  bioinformaticalgorithms used to infer coreceptor usage were observed to vary in their outputs with phenotypictropism results, such that it was feasible to assign them classifcation cutoffs. Furthermore, both theplasma RNA-based and peripheral blood mononuclear cell DNA-based amplifcations gave usefulresults. The methodologies were performed in patients from the observational cohort HOMER, andwere validated in a small subset of patients from the MOTIVATE clinical trials of maraviroc. 7.2.2     Conclusions for Chapter ThreeChapter  3  examined  an  extensive  dataset  of  patients  enrolling  in  the  phase  III  clinical  trials  ofmaraviroc  in  treatment-experienced  participants.  Using  75%  of  the  dataset,  cutoffs  for  thebioinformatic algorithms and the allowable percentage of CXCR4-using variants were established byoptimizing against phenotypic tropism assay results and virologic responses to maraviroc betweenbaseline and week eight. For deep sequencing, the primary cutoffs established for defning X4 HIVwere a PSSM cutoff of ≥-4.75 and a geno2pheno cutoff of ≤3.5. A sample was defned as non-R5 if 2%or more of the viral sequences were classifed as CXCR4-using by either of the optimized algorithms. 153These cutoffs were confrmed in the remaining quarter of patients in the dataset. Compared to thesensitivity of X4 detection with population-based sequencing in Chapter 2 (69%-75%), the sensitivityof next-generation sequencing relative to the phenotypic  Trofle assay was 84%, with comparablespecifcity. More importantly, next-generation sequencing was able to predict virologic responses tomaraviroc in all three of the clinical trials examined. Deep sequencing predicted viral load declines,rates of virologic suppression, and rates of tropism changes while on maraviroc. It was superior atpredicting these  responses  than either  the original  Trofle assay or  population-based sequencing.Results were reproducible when the samples were processed at an independent laboratory, and werecomparable to the Enhanced Sensitivity Trofle assay, where these results were available.There  has  been  past  criticism that  the  true  performance  of  next-generation  sequencing  for  HIVtropism cannot be determined from the maraviroc trials described in Chapter 3 234,354. This has beenlevelled at the fact that patients were pre-screened using the original Trofle assay, and thus are notrepresentative of a general treatment population. Importantly, however, this thesis demonstrates thatvery similar results were obtained when analyses were restricted to the unbiased sample set which isdescribed in Section 3.3.8 and Figure 3.14. The patients selected for this unbiased data set were alltreated regardless of their tropism status. Thus, the composition of this dataset was not impacted byscreening. Even in this unbiased dataset,  deep sequencing was an excellent predictor of virologicoutcomes on maraviroc and was comparable to the Enhanced Sensitivity Trofle Assay. Furthermore,similar results were obtained by an independent laboratory which replicated these methods on thesame dataset. 7.2.3     Conclusions for Chapter FourThe majority of the work presented in Chapters 2 and 3 focussed on using plasma HIV RNA fromviraemic individuals. Chapter 4 differs from but expands on the previous chapters because it presentsresults obtained from HIV DNA derived from peripheral blood mononuclear cells rather than freevirions. Approximately 2% of the overall lymphocyte population is present in the systemic circulation154at any one time 61, and these can be obtained during a blood draw. HIV DNA may be integrated in thecells’  chromosomes,  or  may otherwise  be  associated  with  them (e.g.,  episomal  DNA or  recentlyreverse-transcribed viral genomes). Since the HIV DNA is used as template for new viral progeny, itrepresents  useful  alternative  material  for  assessing  HIV  tropism.  A  tropism  test  is  stronglyrecommended for patients about to start maraviroc-containing regimens  67.   However, in order toperform  an  RNA-based  tropism  test,  plasma  viraemia  must  be  high  enough  to  have  suffcientmaterial. For patients with lower or suppressed viral loads, only the cellular compartment may haveenough material available for testing. Chapter  4  evaluates  this  DNA-based  tropism method and offers  a  comparison  with RNA-basedmethods.  Broadly,  these  methods performed very similarly,  and DNA-based  tropism testing wasshown to be a good indicator of successful treatment with maraviroc. In a subset of patients however,it appeared that where DNA- and RNA-based methods gave discordant results, that the RNA-basedresults tended to be slightly more indicative of therapy success. A major success of this study wasthat it directly helped to inform the design of an international clinical trial. The MARCH study is arandomized  clinical  trial  designed  to  evaluate  the  utility  of  DNA-based  tropism  testing  invirologically-suppressed individuals on antiretroviral therapy 344,346. 7.2.4     Conclusions for Chapter FiveChapter  5  examines the  longitudinal  changes  in  HIV coreceptor  usage as  assessed  by genotypictropism assays  and next-generation sequencing.  The study fnds that  a  majority  of patients whoexperience a change in tropism on maraviroc harboured CXCR4-using HIV prior to treatment andthat this can be detected with next-generation sequencing. Furthermore, the study confrms that theviral population present at treatment failure is phylogenetically related to and is likely derived fromthe pre-treatment CXCR4-using population. This is important because it provides evidence that thepre-treatment viral species are not merely artifacts of the next-generation sequencing methodology,but represent the actual viral populations which undergo selection by maraviroc. 155The experiments performed in Chapter  5 also revealed an intriguing aspect of  maraviroc failure.There were two distinct  patterns of failure.  Tropism changes were accompanied by a number ofmutations in the V3 loop and a massive increase in the proportion of non-R5 variants. In contrast,where virologic failure was not associated with a tropism change, there were few, if any, changes inV3 and the proportion of non-R5 variants remained low at both screening and failure. These changeswere so distinct and drastic that while X4 variants could only be detected at low levels by deepsequencing  at  screening,  they  could  be  easily  detected  at  failure  by  standard  population-basedsequencing due to their massive increase in prevalence. 7.2.5     Conclusions for Chapter SixAll  previous  chapters  described  the  application  of  deep  sequencing  to  treatment-experiencedpopulations.  The  fnal  chapter  applies  this  technology  in  a  completely  different  population  oftreatment-naïve patients entering the MERIT trial of maraviroc. Additionally, all previous analyseswere performed on patients with primarily subtype B HIV-1. Chapter 6 includes patients with othersubtypes of HIV-1. Interestingly, although deep sequencing was optimized on subtype B-infected,treatment-experienced patients, it was still an excellent discriminator of responses to maraviroc intreatment-naïve  patients,  many  of  whom  were  infected  with  non-subtype  B  HIV.  Thus,  themethodologies appear robust to very different bioinformatic approaches, treatment populations, andHIV subtypes. Next-generation  sequencing  had  similar  performance  to  ESTA  in  this  population.  Also,  aretrospective non-inferiority analysis of the trial indicated that had deep sequencing been used toscreen patients entering MERIT, the maraviroc arm would have likely been found to be non-inferiorto the efavirenz arm of the trial. Where results were discordant between assays, again next-generationsequencing  tended  to  give  better  results  relative  to  ESTA  or  population-based  sequencing.Furthermore, when the once-daily maraviroc arm was retrospectively re-tested using next-generationsequencing, the virologic responses in this arm were similar to those of the twice-daily arm. Hence,156better tropism screening of the maraviroc once-daily arm may have preventing its early terminationby the study’s data, safety and monitoring board. 7.2.6     Summary of Specific ConclusionsOverall,  this  thesis  establishes  that  next-generation  sequencing  is  a  useful  tool  in  the  clinicaltreatment of HIV infection. It can be used to accurately assess HIV tropism from both the plasmaRNA and  cellular  DNA compartments.  The  coreceptor  usage  assessments  by  next-generationsequencing  are  similar,  if  not  superior  to  population-based  sequencing  and  two  versions  of  aphenotypic tropism assay. The data obtained with this high-sensitivity genotyping approach can beused  to  predict  several  independent  outcomes  on  therapy,  including  viral  load  declines  andsuppression, as well as phenotypic changes in HIV tropism over time. In most cases, the variantsdetected by next-generation  sequencing are  those  which  experience  drug selection  pressure,  andcontribute to treatment failure. Deep sequencing is therefore a highly detailed, clinically relevant, andpredictive  application  of  next-generation  sequencing,  and  provides  valuable  insight  into  thetreatment of HIV infection. 7.3     Future Directions & ApplicationsThere are a number of applications and research avenues which can build upon the results describedin this thesis. Importantly, the studies described in this thesis have been key in implementing next-generation sequencing into HIV treatment. In fact, since beginning these studies, the use of next-generation sequencing to determine HIV-1 tropism has now become the standard-of-care for HIVtherapy, and is recommended by expert guidelines panels including the United States Department ofHealth and Human Services 276,277,379. The bulk of the above detailed studies were performed using an earlier example of a next-generationsequencing platform, the Roche/454 Life Sciences Genome Sequencer FLX. As cited in Chapter 1,there are a number of other platforms which can generate similar results, but differ in their chemistry157244,246–248,321. As the feld progresses, these platforms tend towards having increasingly long sequenceread-lengths,  higher read depths,  lower  error rates,  lower costs per  base, and faster turn-aroundtimes. Developments in single molecule and real-time sequencing herald exciting and unforeseeableadvancements in next-generation sequencing 253,380. Thus, there will surely be an increasing number ofapplications  of  next-generation  sequencing  in  a  vast  array  of  felds,  including  HIV.  Futureimplementation of next-generation sequencing must include transition from older platforms such asthe GS-FLX to newer ones. This must also include validation that the new methods are comparable tothe previously adopted ones. There are already a number of interesting applications of next-generation sequencing in HIV infectionother than for determining coreceptor usage. For example, next-generation sequencing has been usedfor detecting drug resistant HIV or hepatitis C virus; for investigating viral coinfections; for studyingthe development of HIV neutralizing antibodies; and for more accurately determining linked HIVtransmissions 234,381–384. Human genetic testing to personalize medical treatment is also possible withnext-generation sequencing 254, and the above applications combined with others have high potentialto lead to better treatment for people living with HIV. Although  this  thesis  specifcally  focuses  on  the  application  of  next-generation  sequencing  todetermining HIV tropism, the approach itself is readily transferrable to other types of antiretroviraldrug resistance and viral evolution in general. The thesis candidate has also used deep sequencing insuch a context.  For example,  deep sequencing can be used to detect drug resistance mutations inprotease, reverse transcriptase, and integrase 237,385,386. It can be used to monitor longitudinal evolutionof HIV after transmission of drug resistance, and this may have implications on the ftness costs ofvarious mutations 385. Deep sequencing can also be used to monitor viral evolution upon resumptionof antiretroviral therapy, though does not appear to have much utility to detect possibly archivedhistoric  mutations  386.  Finally,  deep sequencing can  be used to probe evolution within  genes  notcurrently associated with drug resistance, and the candidate has performed experiments to evaluateco-evolving sites within HIV-1 nef 387. 158In addition to providing insight into treatment with maraviroc, the candidate has also used deep V3sequencing to assess viral tropism in untreated individuals 161,164. This can provide insight into viralevolution and coreceptor switching. Deep V3 sequencing can also potentially be used to investigatethe hypothesis that CXCR4-using HIV may be protective against the development of breast cancer viainteraction with CXCR4-expressing tumour tissue 388,389. The approach is also useful in treatment withother CCR5 antagonists, including vicriviroc and cenicriviroc 110,111,317,390. Pre-exposure prophylaxis (PrEP) or post-exposure prophylaxis (PEP) regimens containing maravirochave been investigated for the prevention of HIV transmission  391,392. Microbicide formulations andvaginal rings containing maraviroc have also been considered as biomedical prevention technologies393–395. Success with these prevention strategies will likely be contingent upon the exposing virus beingCCR5-using, which appears to be common for the majority of HIV transmissions 149. Cases in whichtransmission still occurs could be examined using deep V3 sequencing, and may provide additionalinsight. For instance, it could be used to determine whether a donor harboured CXCR4-using viruswhich was transmitted to the recipient. Furthermore, next-generation sequencing could be used toretrospectively examine a recipient’s viral population to determine whether they were infected with aCXCR4-using virus, and also to examine how this virus evolves under maraviroc pressure duringearly infection prior to diagnosis. Finally, the absence of detection of any CXCR4-using virus by deepsequencing  could  indicate  that  non-adherence  was  the  primary  cause  of  transmission  undermaraviroc PrEP or PEP, rather than exposure to non-R5 virus. A number of novel approaches to actually cure HIV infection have involved the CCR5 gene. The frstwell-documented cure of HIV was published in 2009, with a follow-up confrmation study two yearslater 352,396. These studies described an HIV-infected patient who underwent stem-cell transplantationfor  treatment  of  acute  myeloid  leukemia.  The  donor  material  was  remarkable  in  that  it  wasspecifcally selected from an individual who was homozygous for the CCR5 Δ32 allele, which confersnear  total  resistance  to  HIV  infection.  Following  successful  transplantation  and  cessation  ofantiretroviral therapy, HIV RNA and DNA remained persistently undetectable in the patient severalyears  after  the  initial  procedure,  and  there  was  signifcant  immunological  recovery  396.  Deep159sequencing was performed on a sample from the patient prior to achieving an undetectable viralload, and the proportion of non-R5 variants was 2.9% 352, a value which is provocatively close to thecutoff established in this thesis. It is possible that if such a procedure were attempted in the future ona patient with a higher proportion of non-R5 variants,  that a cure might not be achieved in thiscontext. Hence it may be prudent to perform deep sequencing in these cases in order to confrm thepatient’s  R5  HIV  status  prior  to  transplantation.  Recent  attempts  at  replicating  this  cure  haveindicated  that  the  defective  CCR5 allele  seems  to  be  a  crucial  part  of  achieving  long-termremission/cure, rather than the stem cell transplant itself 397.Partially based on the fndings of the stem cell transplant cure, there have been efforts at mimickingthe stem-cell cure using gene therapy which targets CCR5 351,398–400. Generally, these approaches haveinvolved direct modifcation of the genomes of stem cells drawn from HIV-infected patients. Usingenzymatic proteins such as zinc-fnger nucleases, the CCR5 gene can be disrupted, and the modifedcells can be infused back into the patient. Cells where CCR5 has been disrupted will express lowerconcentrations  of  CCR5  on  their  surfaces,  and  will  be  less  susceptible  to  HIV  infection.  Again,however, there may be a requirement for a very low proportion of non-R5 HIV variants in thesepatients in order for the procedures to work. Therefore, deep V3 sequencing may also be used inthese situations to identify the best candidates.  Next-generation  sequencing  has  revolutionized  the  ability  to  describe  and  evaluate  the  viralpopulations  within  HIV-infected  individuals.  Pre-existing  minority  variants  can  be  selected  bysuboptimal  antiretroviral  therapy,  and  lead  to  treatment  failure.  However,  deep  sequencing  canaccurately detect such variants, and this information can be used to predict whether a patient willrespond to their regimen. Not only can next-generation sequencing lead to personalized treatment forHIV, it may also act as a component in strategies aimed at curing HIV infection entirely. 160Bibliography1. UNAIDS. Global report: UNAIDS report on the global AIDS epidemic 2013.; 2013.2. Gottlieb M, Schanker H, Fan P, Saxon A, Weisman J, Pozalski I. Pneumocystis Pneumonia -- Los Angeles. MMWR Morbidity and mortality weekly report. 1981;30:250–2.3. Barré-Sinoussi F, Chermann JC, Rey F, Nugeyre MT, Chamaret S, Gruest J, Dauguet C, Axler-Blin C, Vézinet-Brun F, Rouzioux C, Rozenbaum W, Montagnier L. Isolation of a T-lymphotropic retrovirus from a patient at risk for acquired immune defciency syndrome (AIDS). Science. 1983;220(4599):868–71.4. Hu DJ, Dondero TJ, Rayfeld MA, George JR, Schochetman G, Jaffe HW, Luo CC, Kalish ML, Weniger BG, Pau CP, Schable CA, Curran JW. The emerging genetic diversity of HIV. The importance of global surveillance for diagnostics, research, and prevention. JAMA: the journal of the American Medical Association. 1996;275(3):210–6.5. Campbell-Yesufu OT, Gandhi RT. Update on human immunodefciency virus (HIV)-2 infection. Clinical infectious diseases. 2011;52(6):780–7.6. Varmus H. Retroviruses. Science. 1988;240(4858):1427–1435.7. Robertson DL, Anderson JP, Bradac JA, Carr JK, Foley B, Funkhouser RK, Gao F, Hahn BH, Kalish ML, Kuiken C, Learn GH, Leitner T, McCutchan F, Osmanov S, Peeters M, Pieniazek D, Salminen M, Sharp PM, Wolinsky S, et al. HIV-1 nomenclature proposal. Science. 2000;288(5463):55–6.8. Taylor BS, Sobieszczyk ME, McCutchan FE, Hammer SM. The challenge of HIV-1 subtype diversity. The New England journal of medicine. 2008;358(15):1590–1602.1619. Hemelaar J, Gouws E, Ghys PD, Osmanov S, others. Global trends in molecular epidemiology of HIV-1 during 2000–2007. AIDS. 2011;25(5):679.10. Worobey M, Gemmel M, Teuwen DE, Haselkorn T, Kunstman K, Bunce M, Kalengayi M, Marck E Van, Muyembe J, Kabongo JM, Gilbert MTP, Wolinsky SM. Direct evidence of extensive diversity of HIV-1 in Kinshasa by 1960. Nature. 2008;455:661–665.11. Heeney JL, Dalgleish AG, Weiss R a. Origins of HIV and the evolution of resistance to AIDS. Science. 2006;313(5786):462–6.12. Keele BF, Van Heuverswyn F, Li Y, Bailes E, Takehisa J, Santiago ML, Bibollet-Ruche F, Chen Y, Wain L V, Liegeois F, Loul S, Ngole EM, Bienvenue Y, Delaporte E, Brookfeld JFY, Sharp PM, Shaw GM, Peeters M, Hahn BH. Chimpanzee reservoirs of pandemic and nonpandemic HIV-1. Science. 2006;313(5786):523–6.13. Sharp PM, Hahn BH. Origins of HIV and the AIDS pandemic. Cold Spring Harbor perspectives in medicine. 2011;1(1):a006841.14. Lemey P, Pybus OG, Wang B, Saksena NK, Salemi M, Vandamme A-M. Tracing the origin and history of the HIV-2 epidemic. Proceedings of the National Academy of Sciences of the United States of America. 2003;100(11):6588–92.15. Van Heuverswyn F, Li Y, Neel C, Bailes E, Keele BF, Liu W, Loul S, Butel C, Liegeois F, Bienvenue Y, Ngolle EM, Sharp PM, Shaw GM, Delaporte E, Hahn BH, Peeters M. Human immunodefciency viruses: SIV infection in wild gorillas. Nature. 2006;444(7116):164.16. Thomson MM, Pérez-Alvarez L, Nájera R. Molecular epidemiology of HIV-1 genetic forms and its signifcance for vaccine development and therapy. The Lancet infectious diseases. 2002;2(8):461–71.16217. Plantier J-C, Leoz M, Dickerson JE, De Oliveira F, Cordonnier F, Lemée V, Damond F, Robertson DL, Simon F. A new human immunodefciency virus derived from gorillas. Nature medicine. 2009;15(8):871–2.18. Richman DD, Kinchington D, Fitch N, Griffn S, Persaud D, Siliciano R, Stein SEG, Gazzard BG, Powderly WG, Tebas P, Race E, Rizzardi GP, Pantaleo G, Razavi B, Friedland G, Pawinski R, Govender M, Govender T. Human Immunodeficiency Virus. 1st ed. (Richman DD, ed.). London, , UK: International Medical Press; 2003.19. Frankel A, Young J. HIV-1: Fifteen Proteins and an RNA. Annual review of biochemistry. 1998;67:1–25.20. Wong RW, Balachandran A, Ostrowski MA, Cochrane A. Digoxin suppresses HIV-1 replication by altering viral RNA processing. PLoS pathogens. 2013;9(3):e1003241.21. Clapham PR. HIV and chemokines: ligands sharing cell-surface receptors. Trends in cell biology. 1997;7(7):264–8.22. Engelman A, Cherepanov P. The structural biology of HIV-1: mechanistic and therapeutic insights. Nature reviews Microbiology. 2012;10(4):279–90.23. Kwong PD, Wyatt R, Robinson J, Sweet RW, Sodroski J, Hendrickson WA. Structure of an HIV gp120 envelope glycoprotein in complex with the CD4 receptor and a neutralizing human antibody. Nature. 1998;393(6686):648–59.24. Rizzuto CD, Wyatt R, Hernández-Ramos N, Sun Y, Kwong PD, Hendrickson WA, Sodroski J. A conserved HIV gp120 glycoprotein structure involved in chemokine receptor binding. Science. 1998;280(5371):1949–53.25. Xiang S-H, Finzi A, Pacheco B, Alexander K, Yuan W, Rizzuto C, Huang C-C, Kwong PD, Sodroski J. A V3 loop-dependent gp120 element disrupted by CD4 binding stabilizes the 163human immunodefciency virus envelope glycoprotein trimer. Journal of virology. 2010;84(7):3147–61.26. Lapham CK, Ouyang J, Chandrasekhar B, Nguyen NY, Dimitrov DS, Golding H. Evidence for cell-surface association between fusin and the CD4-gp120 complex in human cell lines. Science. 1996;274(5287):602–5.27. Quiñones-Mateu ME, Lederman MM, Feng Z, Chakraborty B, Weber J, Rangel HR, Marotta ML, Mirza M, Jiang B, Kiser P, Medvik K, Sieg SF, Weinberg A. Human epithelial beta-defensins 2 and 3 inhibit HIV-1 replication. AIDS. 2003;17(16):F39–48.28. Coffey MJ, Woffendin C, Phare SM, Strieter RM, Markovitz DM. RANTES inhibits HIV-1 replication in human peripheral blood monocytes and alveolar macrophages. The American journal of physiology. 1997;272(5 Pt 1):L1025–9.29. Weiss CD. HIV-1 gp41: mediator of fusion and target for inhibition. AIDS reviews. 2003;5(4):214–21.30. Chan DC, Fass D, Berger JM, Kim PS. Core structure of gp41 from the HIV envelope glycoprotein. Cell. 1997;89(2):263–73.31. Weissenhorn W, Dessen A, Harrison SC, Skehel JJ, Wiley DC. Atomic structure of the ectodomain from HIV-1 gp41. Nature. 1997;387(6631):426–30.32. Pornillos O, Ganser-Pornillos BK, Yeager M. Atomic-level modelling of the HIV capsid. Nature. 2011;469(7330):424–7.33. Schaller T, Ocwieja KE, Rasaiyaah J, Price AJ, Brady TL, Roth SL, Hué S, Fletcher AJ, Lee K, KewalRamani VN, Noursadeghi M, Jenner RG, James LC, Bushman FD, Towers GJ. HIV-1 capsid-cyclophilin interactions determine nuclear import pathway, integration targeting and replication effciency. PLoS pathogens. 2011;7(12):e1002439.16434. Forshey BM, von Schwedler U, Sundquist WI, Aiken C. Formation of a human immunodefciency virus type 1 core of optimal stability is crucial for viral replication. Journal of virology. 2002;76(11):5667–77.35. Huang H, Chopra R, Verdine GL, Harrison SC. Structure of a covalently trapped catalytic complex of HIV-1 reverse transcriptase: implications for drug resistance. Science. 1998;282(5394):1669–75.36. Albrecht D, Altfeld M, Behrens G, Bredeek UF, Buhk T, Dickinson L, Lyall H, Meemken L. HIV Medicine. 15th ed. (Hoffmann C, Rockstroh JK, Kamps BS, eds.). Flying Publisher; 2007.37. Preston BD, Poiesz BJ, Loeb LA. Fidelity of HIV-1 reverse transcriptase. Science. 1988;242(4882):1168–71.38. Guo F, Cen S, Niu M, Saadatmand J, Kleiman L. Inhibition of tRNA(Lys)-primed reverse transcription by human APOBEC3G during human immunodefciency virus type 1 replication. Journal of virology. 2006;80(23):11710–22.39. Mangeat B, Turelli P, Caron G, Friedli M, Perrin L, Trono D. Broad antiretroviral defence by human APOBEC3G through lethal editing of nascent reverse transcripts. Nature. 2003;424(6944):99–103.40. Sheehy AM, Gaddis NC, Malim MH. The antiretroviral enzyme APOBEC3G is degraded by the proteasome in response to HIV-1 Vif. Nature medicine. 2003;9(11):1404–7.41. Strebel K, Luban J, Jeang K-T. Human cellular restriction factors that target HIV-1 replication. BMC medicine. 2009;7:48.42. Fassati A, Görlich D, Harrison I, Zaytseva L, Mingot J-M. Nuclear import of HIV-1 intracellular reverse transcription complexes is mediated by importin 7. The EMBO journal. 2003;22(14):3675–85.16543. Bukrinsky M, Adzhubei A. Viral protein R of HIV-1. Reviews in medical virology. 1999;9(1):39–49.44. Maertens GN, Hare S, Cherepanov P. The mechanism of retroviral integration from X-ray structures of its key intermediates. Nature. 2010;468(7321):326–9.45. Chun T-W, Fauci AS. HIV reservoirs: pathogenesis and obstacles to viral eradication and cure. AIDS. 2012;26(10):1261–8.46. Saad JS, Miller J, Tai J, Kim A, Ghanam RH, Summers MF. Structural basis for targeting HIV-1 Gag proteins to the plasma membrane for virus assembly. Proceedings of the National Academy of Sciences of the United States of America. 2006;103(30):11364–9.47. Malim MH, Emerman M. HIV-1 accessory proteins--ensuring viral survival in a hostile environment. Cell host & microbe. 2008;3(6):388–98.48. Hallenberger S, Bosch V, Angliker H, Shaw E, Klenk HD, Garten W. Inhibition of furin-mediated cleavage activation of HIV-1 glycoprotein gp160. Nature. 1992;360(6402):358–61.49. Moulard M, Hallenberger S, Garten W, Klenk HD. Processing and routage of HIV glycoproteins by furin to the cell surface. Virus research. 1999;60(1):55–65.50. Jacks T, Power MD, Masiarz FR, Luciw PA, Barr PJ, Varmus HE. Characterization of ribosomal frameshifting in HIV-1 gag-pol expression. Nature. 1988;331(6153):280–3.51. Scarlata S, Carter C. Role of HIV-1 Gag domains in viral assembly. Biochimica et biophysica acta. 2003;1614(1):62–72.52. Jouvenet N, Simon SM, Bieniasz PD. Imaging the interaction of HIV-1 genomes and Gag during assembly of individual viral particles. Proceedings of the National Academy of Sciences of the United States of America. 2009;106(45):19114–9.16653. Hurley JH, Hanson PI. Membrane budding and scission by the ESCRT machinery: it’s all in the neck. Nature reviews Molecular cell biology. 2010;11(8):556–66.54. Dubé M, Roy BB, Guiot-Guillain P, Binette J, Mercier J, Chiasson A, Cohen E a. Antagonism of tetherin restriction of HIV-1 release by Vpu involves binding and sequestration of the restriction factor in a perinuclear compartment. PLoS pathogens. 2010;6(4):e1000856.55. Fitzgerald PMD, Springer JP. Structure and function of retroviral proteases. Annual review of biophysics and biophysical chemistry. 1991;20(1):299–320.56. Huff JR. HIV protease: a novel chemotherapeutic target for AIDS. Journal of medicinal chemistry.1991;34(8):2305–14.57. Sundquist WI, Kräusslich H-G. HIV-1 assembly, budding, and maturation. Cold Spring Harbor perspectives in medicine. 2012;2(7):a006924.58. Hladik F, McElrath MJ. Setting the stage: host invasion by HIV. Nature reviews Immunology. 2008;8(6):447–57.59. Tebit DM, Ndembi N, Weinberg A, Quiñones-Mateu ME. Mucosal transmission of human immunodefciency virus. Current HIV research. 2012;10(1):3–8.60. Stevenson M. HIV-1 pathogenesis. Nature medicine. 2003;9(7):853–60.61. Blankson JN, Persaud D, Siliciano RF. The challenge of viral reservoirs in HIV-1 infection. Annual review of medicine. 2002;53(6):557–93.62. Romagnani S. The Th1/Th2 paradigm. Immunology today. 1997;18(6):263–6.63. Fahey JL, Taylor JM, Detels R, Hofmann B, Melmed R, Nishanian P, Giorgi J V. The prognostic value of cellular and serologic markers in infection with human immunodefciency virus type 1. The New England journal of medicine. 1990;322(3):166–72.16764. Puhan M a, Van Natta ML, Palella FJ, Addessi A, Meinert C. Excess mortality in patients with AIDS in the era of highly active antiretroviral therapy: temporal changes and risk factors. Clinical infectious diseases. 2010;51(8):947–56.65. The Opportunistic Infections Project Team of the Collaboration of Observational HIV Epidemiological Research in Europe (COHERE) in EuroCoord. CD4 Cell Count and the Risk of AIDS or Death in HIV-Infected Adults on Combination Antiretroviral Therapy with a Suppressed Viral Load: A Longitudinal Cohort Study from COHERE. PLoS medicine. 2012;9(3):e1001194.66. Mellors JW, Rinaldo CR, Gupta P, White RM, Todd JA, Kingsley LA. Prognosis in HIV-1 infection predicted by the quantity of virus in plasma. Science. 1996;272(5265):1167–70.67. Thompson MA, Aberg JA, Hoy JF, Telenti A, Benson C, Cahn P, Eron JJ, Günthard HF, Hammer SM, Reiss P, Richman DD, Rizzardini G, Thomas DL, Jacobsen DM, Volberding PA. Antiretroviral treatment of adult HIV infection: 2012 recommendations of the International Antiviral Society-USA panel. JAMA: the journal of the American Medical Association. 2012;308(4):387–402.68. Castro KG, Ward JW, Slutsker L, Buehler JW, Jaffe HW, Berkelman RL, Curran JW. 1993 revised classifcation system for HIV infection and expanded surveillance case defnition for AIDS among adolescents and adults. MMWR Morbidity and mortality weekly report. 1992;41(RR-17):1–19.69. Doitsh G, Galloway NLK, Geng X, Yang Z, Monroe KM, Zepeda O, Hunt PW, Hatano H, Sowinski S, Muñoz-Arias I, Greene WC. Cell death by pyroptosis drives CD4 T-cell depletion in HIV-1 infection. Nature. 2014;505(7484):509–14.70. Monroe KM, Yang Z, Johnson JR, Geng X, Doitsh G, Krogan NJ, Greene WC. IFI16 DNA Sensor Is Required for Death of Lymphoid CD4 T Cells Abortively Infected with HIV. Science. 2013;11(11):997–1004.16871. McCune JM. The dynamics of CD4+ T-cell depletion in HIV disease. Nature. 2001;410(6831):974–9.72. Grossman Z, Meier-Schellersheim M, Sousa AE, Victorino RMM, Paul WE. CD4+ T-cell depletion in HIV infection: are we closer to understanding the cause? Nature medicine. 2002;8(4):319–23.73. Douek DC, Brenchley JM, Betts MR, Ambrozak DR, Hill BJ, Okamoto Y, Casazza JP, Kuruppu J, Kunstman K, Wolinsky S, Grossman Z, Dybul M, Oxenius A, Price DA, Connors M, Koup RA. HIV preferentially infects HIV-specifc CD4+ T cells. Nature. 2002;417(6884):95–8.74. Embretson J, Zupancic M, Ribas JL, Burke A, Racz P, Tenner-Racz K, Haase AT. Massive covert infection of helper T lymphocytes and macrophages by HIV during the incubation period of AIDS. Nature. 1993;362(6418):359–62.75. Cameron PU, Freudenthal PS, Barker JM, Gezelter S, Inaba K, Steinman RM. Dendritic cells exposed to human immunodefciency virus type-1 transmit a vigorous cytopathic infection to CD4+ T cells. Science. 1992;257(5068):383–7.76. Trono D, Van Lint C, Rouzioux C, Verdin E, Barre-Sinoussi F, Chun T-W, Chomont N. HIV Persistence and the Prospect of Long-Term Drug-Free Remissions for HIV-Infected Individuals. Science. 2010;329(5988):174–180.77. Richman DD, Margolis DM, Delaney M, Greene WC, Hazuda D, Pomerantz RJ. The Challenge of Finding a Cure for HIV Infection: Can We Do Better Than HAART ? Science. 2009;(March):1304–1307.78. Han Y, Wind-Rotolo M, Yang H-C, Siliciano JD, Siliciano RF. Experimental approaches to the study of HIV-1 latency. Nature reviews Microbiology. 2007;5(2):95–106.79. Simon V, Ho DD. HIV-1 dynamics in vivo: implications for therapy. Nature reviews Microbiology. 2003;1(3):181–90.16980. Chomont N, El-Far M, Ancuta P, Trautmann L, Procopio FA, Yassine-Diab B, Boucher G, Boulassel M-R, Ghattas G, Brenchley JM, Schacker TW, Hill BJ, Douek DC, Routy J-P, Haddad EK, Sékaly R-P. HIV reservoir size and persistence are driven by T cell survival and homeostatic proliferation. Nature medicine. 2009;15(8):893–900.81. Hatano H, Hayes TL, Dahl V, Sinclair E, Lee T-H, Hoh R, Lampiris H, Hunt PW, Palmer S, McCune JM, Martin JN, Busch MP, Shacklett BL, Deeks SG. A Randomized, Controlled Trial of Raltegravir Intensifcation in Antiretroviral-treated, HIV-infected Patients with a Suboptimal CD4+ T Cell Response. The Journal of infectious diseases. 2011;203(7):960–8.82. Pope M, Haase AT. Transmission, acute HIV-1 infection and the quest for strategies to prevent infection. Nature medicine. 2003;9(7):847–52.83. Hughes JP, Baeten JM, Lingappa JR, Magaret AS, Wald A, de Bruyn G, Kiarie J, Inambao M, Kilembe W, Farquhar C, Celum C. Determinants of Per-Coital-Act HIV-1 Infectivity Among African HIV-1-Serodiscordant Couples. The Journal of infectious diseases. 2012;205(3):358–365.84. Quinn TCC, Wawer MJJ, Sewankambo N, Serwadda D, Li C, Wabwire-Mangen F, Meehan MOO, Lutalo T, Gray RHH, Rakai Project Study Group. Viral Load and Heterosexual Transmission of Human Immunodefciency Virus Type 1. New England Journal of Medicine. 2000;342(13):921–929.85. Gray RH, Wawer MJ, Brookmeyer R, Sewankambo NK, Serwadda D, Wabwire-Mangen F, Lutalo T, Li X, vanCott T, Quinn TC. Probability of HIV-1 transmission per coital act in monogamous, heterosexual, HIV-1-discordant couples in Rakai, Uganda. Lancet. 2001;357(9263):1149–53.86. McMichael AJ, Borrow P, Tomaras GD, Goonetilleke N, Haynes BF. The immune response during acute HIV-1 infection: clues for vaccine development. Nature reviews Immunology. 2010;10(1):11–23.17087. Pedersen C, Lindhardt BO, Jensen BL, Lauritzen E, Gerstoft J, Dickmeiss E, Gaub J, Scheibel E, Karlsmark T. Clinical course of primary HIV infection: consequences for subsequent course of infection. BMJ. 1989;299(6692):154–7.88. Mindel A, Tenant-Flowers M. ABC of AIDS: Natural history and management of early HIV infection. BMJ. 2001;322(7297):1290–3.89. Altfeld M, Fadda L, Frleta D, Bhardwaj N. DCs and NK cells: critical effectors in the immune response to HIV-1. Nature Reviews Immunology. 2011;11(3):176–186.90. Kawai T, Akira S. Innate immune recognition of viral infection. Nature immunology. 2006;7(2):131–7.91. Fiebig EW, Wright DJ, Rawal BD, Garrett PE, Schumacher RT, Peddada L, Heldebrant C, Smith R, Conrad A, Kleinman SH, Busch MP. Dynamics of HIV viremia and antibody seroconversion in plasma donors: implications for diagnosis and staging of primary HIV infection. AIDS. 2003;17(13):1871–9.92. Moir S, Fauci AS. B cells in HIV infection and disease. Nature reviews Immunology. 2009;9(4):235–45.93. McMichael A, Rowland-Jones SL. Cellular immune responses to HIV. Nature. 2001;410(6831):980–7.94. Brumme ZL, John M, Carlson JM, Brumme CJ, Chan D, Brockman MA, Swenson LC, Tao I, Szeto S, Rosato P, Sela J, Kadie CM, Frahm N, Brander C, Haas DW, Riddler SA, Haubrich R, Walker BD, Harrigan PR, et al. HLA-associated immune escape pathways in HIV-1 subtype B Gag, Pol and Nef proteins. PloS one. 2009;4(8):e6687.95. Yang OO, Kalams SA, Trocha A, Cao H, Luster A, Johnson RP, Walker BD. Suppression of human immunodefciency virus type 1 replication by CD8+ cells: evidence for HLA class I-171restricted triggering of cytolytic and noncytolytic mechanisms. Journal of virology. 1997;71(4):3120–8.96. Shankar P, Xu Z, Lieberman J. Viral-specifc cytotoxic T lymphocytes lyse human immunodefciency virus-infected primary T lymphocytes by the granule exocytosis pathway. Blood. 1999;94(9):3084–93.97. Goulder PJR, Watkins DI. Impact of MHC class I diversity on immune control of immunodefciency virus replication. Nature reviews Immunology. 2008;8(8):619–30.98. Collins KL, Chen BK, Kalams SA, Walker BD, Baltimore D. HIV-1 Nef protein protects infected primary cells against killing by cytotoxic T lymphocytes. Nature. 1998;391(6665):397–401.99. Raport CJ, Gosling J, Schweickart VL, Gray PW, Charo IF. Molecular cloning and functional characterization of a novel human CC chemokine receptor (CCR5) for RANTES, MIP-1beta, and MIP-1alpha. The Journal of biological chemistry. 1996;271(29):17161–6.100. Berger EA, Murphy PM, Farber JM. Chemokine receptors as HIV-1 coreceptors: roles in viral entry, tropism, and disease. Annual review of immunology. 1999;17:657–700.101. Clapham PR, McKnight A. HIV-1 receptors and cell tropism. British medical bulletin. 2001;58:43–59.102. De Clercq E. Recent advances on the use of the CXCR4 antagonist plerixafor (AMD3100, MozobilTM) and potential of other CXCR4 antagonists as stem cell mobilizers. Pharmacology & therapeutics. 2010;128(3):509–18.103. Kucia M, Jankowski K, Reca R, Wysoczynski M, Bandura L, Allendorf DJ, Zhang J, Ratajczak J, Ratajczak MZ. CXCR4-SDF-1 signalling, locomotion, chemotaxis and adhesion. Journal of molecular histology. 2004;35(3):233–45.104. Wu Y, Yoder A. Chemokine coreceptor signaling in HIV-1 infection and pathogenesis. PLoS pathogens. 2009;5(12):e1000520.172105. Bleul CC, Wu L, Hoxie JA, Springer TA, Mackay CR. The HIV coreceptors CXCR4 and CCR5 are differentially expressed and regulated on human T lymphocytes. Proceedings of the National Academy of Sciences of the United States of America. 1997;94(5):1925–30.106. D’Souza MP, Harden VA. Chemokines and HIV-1 second receptors. Confluence of two felds generates optimism in AIDS research. Nature medicine. 1996;2(12):1293–300.107. Glass WG, McDermott DH, Lim JK, Lekhong S, Yu SF, Frank W a, Pape J, Cheshier RC, Murphy PM. CCR5 defciency increases risk of symptomatic West Nile virus infection. The Journal of experimental medicine. 2006;203(1):35–40.108. Dorr P, Westby M, Dobbs S, Griffn P, Irvine B, Macartney M, Mori J, Rickett G, Smith-Burchnell C, Napier C, Webster R, Armour D, Price D, Stammen B, Wood A, Perros M. Maraviroc (UK-427,857), a potent, orally bioavailable, and selective small-molecule inhibitor of chemokine receptor CCR5 with broad-spectrum anti-human immunodefciency virus type 1 activity. Antimicrobial agents and chemotherapy. 2005;49(11):4721–32.109. De Clercq E. The bicyclam AMD3100 story. Nature reviews Drug discovery. 2003;2(7):581–7.110. Gulick RM, Su Z, Flexner C, Hughes MD, Skolnik PR, Wilkin TJ, Gross R, Krambrink A, Coakley E, Greaves WL, Zolopa A, Reichman R, Godfrey C, Hirsch M, Kuritzkes DR. Phase 2 study of the safety and effcacy of vicriviroc, a CCR5 inhibitor, in HIV-1-Infected, treatment-experienced patients: AIDS clinical trials group 5211. The Journal of infectious diseases. 2007;196(2):304–12.111. Lalezari J, Gathe J, Brinson C, Thompson M, Cohen C, Dejesus E, Galindez J, Ernst JA, Martin DE, Palleja SM. Safety, Effcacy, and Pharmacokinetics of TBR-652, a CCR5/CCR2 Antagonist, in HIV-1 – Infected, CCR5 Antagonist–Naive Subjects. Journal of acquired immune deficiency syndromes. 2011;57(2):118–125.112. DiPersio JF, Stadtmauer EA, Nademanee A, Micallef INM, Stiff PJ, Kaufman JL, Maziarz RT, Hosing C, Früehauf S, Horwitz M, Cooper D, Bridger G, Calandra G. Plerixafor and G-CSF 173versus placebo and G-CSF to mobilize hematopoietic stem cells for autologous stem cell transplantation in patients with multiple myeloma. Blood. 2009;113(23):5720–6.113. DiPersio JF, Uy GL, Yasothan U, Kirkpatrick P. Plerixafor. Nature reviews Drug discovery. 2009;8(2):105–6.114. Pantophlet R, Burton DR. GP120: target for neutralizing HIV-1 antibodies. Annual review of immunology. 2006;24:739–69.115. Wyatt R. The HIV-1 Envelope Glycoproteins: Fusogens, Antigens, and Immunogens. Science. 1998;280(5371):1884–1888.116. Briz V, Poveda E, Soriano V. HIV entry inhibitors: mechanisms of action and resistance pathways. The Journal of antimicrobial chemotherapy. 2006;57(4):619–27.117. Madani N, Perdigoto AL, Srinivasan K, Cox JM, Chruma JJ, LaLonde J, Head M, Smith AB, Sodroski JG. Localized changes in the gp120 envelope glycoprotein confer resistance to human immunodefciency virus entry inhibitors BMS-806 and #155. Journal of virology. 2004;78(7):3742–52.118. Xiang S, Kwong PD, Gupta R, Rizzuto CD, Casper DJ, Wyatt R, Wang L, Hendrickson WA, Doyle ML, Sodroski J. Mutagenic stabilization and/or disruption of a CD4-bound state reveals distinct conformations of the human immunodefciency virus type 1 gp120 envelope glycoprotein. Journal of virology. 2002;76(19):9888–99.119. Liu J, Bartesaghi A, Borgnia MJ, Sapiro G, Subramaniam S. Molecular architecture of native HIV-1 gp120 trimers. Nature. 2008;455(7209):109–13.120. White TA, Bartesaghi A, Borgnia MJ, Meyerson JR, de la Cruz MJ V, Bess JW, Nandwani R, Hoxie J a, Lifson JD, Milne JLS, Subramaniam S. Molecular Architectures of Trimeric SIV and HIV-1 Envelope Glycoproteins on Intact Viruses: Strain-Dependent Variation in Quaternary Structure. PLoS pathogens. 2010;6(12):e1001249.174121. Azimi I, Matthias LJ, Center RJ, Wong JWH, Hogg PJ. Disulfde bond that constrains the HIV-1 gp120 V3 domain is cleaved by thioredoxin. The Journal of biological chemistry. 2010;285(51):40072–80.122. Huang C, Tang M, Zhang M-Y, Majeed S, Montabana E, Stanfeld RL, Dimitrov DS, Korber B, Sodroski J, Wilson IA, Wyatt R, Kwong PD. Structure of a V3-containing HIV-1 gp120 core. Science. 2005;310(5750):1025–8.123. Shioda T, Levy JA, Cheng-Mayer C. Macrophage and T cell-line tropisms of HIV-1 are determined by specifc regions of the envelope gp120 gene. Nature. 1991;349(6305):167–9.124. Hwang SS, Boyle TJ, Lyerly HK, Cullen BR. Identifcation of the envelope V3 loop as the primary determinant of cell tropism in HIV-1. Science. 1991;253(5015):71–4.125. Yuan W, Craig S, Si Z, Farzan M, Sodroski J. CD4-induced T-20 binding to human immunodefciency virus type 1 gp120 blocks interaction with the CXCR4 coreceptor. Journal of virology. 2004;78(10):5448–57.126. Cormier E, Dragic T. The crown and stem of the V3 loop play distinct roles in human immunodefciency virus type 1 envelope glycoprotein interactions with the CCR5 coreceptor. Journal of virology. 2002;76(17):8953–8957.127. Huang C-C, Lam SN, Acharya P, Tang M, Xiang S-H, Hussan SS-U, Stanfeld RL, Robinson J, Sodroski J, Wilson IA, Wyatt R, Bewley CA, Kwong PD. Structures of the CCR5 N terminus and of a tyrosine-sulfated antibody with HIV-1 gp120 and CD4. Science. 2007;317(5846):1930–4.128. Basmaciogullari S, Babcock GJ, Van Ryk D, Wojtowicz W, Sodroski J. Identifcation of conserved and variable structures in the human immunodefciency virus gp120 glycoprotein of importance for CXCR4 binding. Journal of virology. 2002;76(21):10791.175129. Rangel HR, Weber J, Chakraborty B, Gutierrez A, Marotta ML, Mirza M, Kiser P, Martinez MA, Este JA, Quiñones-Mateu ME. Role of the human immunodefciency virus type 1 envelope gene in viral ftness. Journal of virology. 2003;77(16):9069–73.130. Lobritz MA, Marozsan AJ, Troyer RM, Arts EJ. Natural variation in the V3 crown of human immunodefciency virus type 1 affects replicative ftness and entry inhibitor sensitivity. Journal of virology. 2007;81(15):8258–69.131. Pastore C, Nedellec R, Ramos A, Pontow S, Ratner L, Mosier DE. Human immunodefciency virus type 1 coreceptor switching: V1/V2 gain-of-ftness mutations compensate for V3 loss-of-ftness mutations. Journal of virology. 2006;80(2):750–8.132. Cardozo T, Kimura T, Philpott S, Weiser B, Burger H, Zolla-Pazner S. Structural basis for coreceptor selectivity by the HIV type 1 V3 loop. AIDS research and human retroviruses. 2007;23(3):415–26.133. Resch W, Hoffman N, Swanstrom R. Improved success of phenotype prediction of the human immunodefciency virus type 1 from envelope variable loop 3 sequence using neural networks. Virology. 2001;288(1):51–62.134. Tan Q, Zhu Y, Li J, Chen Z, Han GW, Kufareva I, Li T, Ma L, Fenalti G, Li J, Zhang W, Xie X, Yang H, Jiang H, Cherezov V, Liu H, Stevens RC, Zhao Q, Wu B. Structure of the CCR5 chemokine receptor-HIV entry inhibitor maraviroc complex. Science. 2013;341(6152):1387–90.135. Kajumo F, Thompson DA, Guo Y, Dragic T. Entry of R5X4 and X4 human immunodefciency virus type 1 strains is mediated by negatively charged and tyrosine residues in the amino-terminal domain and the second extracellular loop of CXCR4. Virology. 2000;271(2):240–7.136. Zanetti G, Briggs J a G, Grünewald K, Sattentau QJ, Fuller SD. Cryo-electron tomographic structure of an immunodefciency virus envelope complex in situ. PLoS pathogens. 2006;2(8):e83.176137. Gottlieb MS, Schroff R, Schanker HM, Weisman JD, Fan PT, Wolf RA, Saxon A. Pneumocystis carinii Pneumonia and Mucosal Candidiasis in Previously Healthy Homosexual Men - Evidence of a New Acquired Cellular Immunodefciency. New England Journal of Medicine. 1981;305(24):1425–1431.138. Klatzmann D, Barré-Sinoussi F, Nugeyre MT, Danquet C, Vilmer E, Griscelli C, Brun-Veziret F, Rouzioux C, Gluckman JC, Chermann JC. Selective tropism of lymphadenopathy associated virus (LAV) for helper-inducer T lymphocytes. Science. 1984;225(4657):59–63.139. Miedema F, Meyaard L, Koot M, Klein MR, Roos MT, Groenink M, Fouchier RA, Van’t Wout A, Tersmette M, Schellekens PT, H S. Changing virus-host interactions in the course of HIV-1 infection. Immunological reviews. 1994;140(140):35–72.140. Dragic T, Litwin V, Allaway GP, Martin SR, Huang Y, Nagashima KA, Cayanan C, Maddon PJ, Koup RA, Moore JP, Paxton WA. HIV-1 entry into CD4+ cells is mediated by the chemokine receptor CC-CKR-5. Nature. 1996;381(6584):667–73.141. Deng H, Liu R, Ellmeier W, Choe S, Unutmaz D, Burkhart M, Di Marzio P, Marmon S, Sutton RE, Hill CM, Davis CB, Peiper SC, Schall TJ, Littman DR, Landau NR. Identifcation of a major co-receptor for primary isolates of HIV-1. Nature. 1996;381(6584):661–6.142. Feng Y, Broder CC, Kennedy PE, Berger E. HIV-1 entry cofactor: functional cDNA cloning of a seven-transmembrane, G protein-coupled receptor. Science. 1996;272(5263):872–7.143. Zhang YJ, Moore JP. Will multiple coreceptors need to be targeted by inhibitors of human immunodefciency virus type 1 entry? Journal of virology. 1999;73(4):3443–8.144. Pollakis G, Paxton W a. Use of (alternative) coreceptors for HIV entry. Current opinion in HIV and AIDS. 2012;7(5):440–9.145. Schuitemaker H, van ’t Wout AB, Lusso P. Clinical signifcance of HIV-1 coreceptor usage. Journal of translational medicine. 2011;9(Suppl 1):S5.177146. Keele BF, Giorgi EE, Salazar-Gonzalez JF, Decker JM, Pham KT, Salazar MG, Sun C, Grayson T, Wang S, Li H, Wei X, Jiang C, Kirchherr JL, Gao F, Anderson JA, Ping L-H, Swanstrom R, Tomaras GD, Blattner WA, et al. Identifcation and characterization of transmitted and early founder virus envelopes in primary HIV-1 infection. Proceedings of the National Academy of Sciences of the United States of America. 2008;105(21):7552–7.147. Salazar-Gonzalez JF, Bailes E, Pham KT, Salazar MG, Guffey MB, Keele BF, Derdeyn CA, Farmer P, Hunter E, Allen S, Manigart O, Mulenga J, Anderson JA, Swanstrom R, Haynes BF, Athreya GS, Korber BTM, Sharp PM, Shaw GM, et al. Deciphering human immunodefciency virus type 1 transmission and early envelope diversifcation by single-genome amplifcation and sequencing. Journal of virology. 2008;82(8):3952–70.148. van’t Wout AB, Kootstra NA, Mulder-Kampinga GA, Albrecht-van Lent N, Scherpbier HJ, Veenstra J, Boer K, Coutinho RA, Miedema F, Schuitemaker H. Macrophage-tropic variants initiate human immunodefciency virus type 1 infection after sexual, parenteral, and vertical transmission. The Journal of clinical investigation. 1994;94(5):2060–7.149. Margolis L, Shattock R. Selective transmission of CCR5-utilizing HIV-1: the “gatekeeper” problem resolved? Nature reviews Microbiology. 2006;4(4):312–7.150. Soto-Ramirez LE, Renjifo B, McLane MF, Marlink R, O’Hara C, Sutthent R, Wasi C, Vithayasai P, Vithayasai V, Apichartpiyakul C, Auewarakul P, Peña Cruz V, Chui DS, Osathanondh R, Mayer K, Lee TH, Essex M. HIV-1 Langerhans’ cell tropism associated with heterosexual transmission of HIV. Science. 1996;271(5253):1291–3.151. Chalmet K, Dauwe K, Foquet L, Baatz F, Seguin-Devaux C, Van Der Gucht B, Vogelaers D, Vandekerckhove L, Plum J, Verhofstede C. Presence of CXCR4-using HIV-1 in patients with recently diagnosed infection: correlates and evidence for transmission. The Journal of infectious diseases. 2012;205(2):174–84.152. De Mendoza C, Rodriguez C, García F, Eiros JM, Ruíz L, Caballero E, Aguilera A, Leiva P, Colomina J, Gutierrez F, del Romero J, Aguero J, Soriano V. Prevalence of X4 tropic viruses in 178patients recently infected with HIV-1 and lack of association with transmission of drug resistance. The Journal of antimicrobial chemotherapy. 2007;59(4):698–704.153. Hedskog C, Mild M, Albert J. Transmission of the X4 phenotype of HIV-1: is there evidence against the “random transmission” hypothesis? The Journal of infectious diseases. 2012;205(2):163–5.154. Martinson JJ, Chapman NH, Rees DC, Liu YT, Clegg JB. Global distribution of the CCR5 gene 32-basepair deletion. Nature genetics. 1997;16(1):100–103.155. Brumme ZL, Goodrich J, Mayer HB, Brumme CJ, Henrick BM, Wynhoven B, Asselin JJ, Cheung PK, Hogg RS, Montaner JSG, Harrigan PR. Molecular and clinical epidemiology of CXCR4-using HIV-1 in a large population of antiretroviral-naive individuals. The Journal of infectious diseases. 2005;192(3):466–74.156. Samson M, Libert F, Doranz BJ, Rucker J, Liesnard C, Farber CM, Saragosti S, Lapoumeroulie C, Cognaux J, Forceille C, Muyldermans G, Verhofstede C, Burtonboy G, Georges M, Imai T, Rana S, Yi Y, Smyth RJ, Collman RG, et al. Resistance to HIV-1 infection in caucasian individuals bearing mutant alleles of the CCR-5 chemokine receptor gene. Nature. 1996;382(6593):722–5.157. Liu R, Paxton W a, Choe S, Ceradini D, Martin SR, Horuk R, MacDonald ME, Stuhlmann H, Koup R a, Landau NR. Homozygous defect in HIV-1 coreceptor accounts for resistance of some multiply-exposed individuals to HIV-1 infection. Cell. 1996;86(3):367–77.158. O’Brien TR, Winkler C, Dean M, Nelson JA, Carrington M, Michael NL, White GC. HIV-1 infection in a man homozygous for CCR5 delta 32. Lancet. 1997;349(9060):1219.159. Michael NL, Nelson JA, KewalRamani VN, Chang G, O’Brien SJ, Mascola JR, Volsky B, Louder M, White GC, Littman DR, Swanstrom R, O’Brien TR. Exclusive and persistent use of the entry coreceptor CXCR4 by human immunodefciency virus type 1 from a subject homozygous for CCR5 delta32. Journal of virology. 1998;72(7):6040–7.179160. Michael NL, Chang G, Louie LG, Mascola JR, Dondero D, Birx DL, Sheppard HW. The role of viral phenotype and CCR-5 gene defects in HIV-1 transmission and disease progression. Nature medicine. 1997;3(3):338–40.161. Bunnik EM, Swenson LC, Edo-Matas D, Huang W, Dong W, Frantzell A, Petropoulos CJ, Coakley E, Schuitemaker H, Harrigan PR, van ’t Wout AB. Detection of Inferred CCR5- and CXCR4-Using HIV-1 Variants and Evolutionary Intermediates Using Ultra-Deep Pyrosequencing. Trkola A, ed. PLoS Pathogens. 2011;7(6):e1002106.162. Pastore C, Ramos A, Mosier DE. Intrinsic obstacles to human immunodefciency virus type 1 coreceptor switching. Journal of virology. 2004;78(14):7565–74.163. Alizon S, Boldin B. Within-host viral evolution in a heterogeneous environment: insights into the HIV co-receptor switch. Journal of evolutionary biology. 2010;23(2005):2625–2635.164. Poon AFY, Swenson LC, Bunnik EM, Edo-Matas D, Schuitemaker H, van ’t Wout AB, Harrigan PR. Reconstructing the Dynamics of HIV Evolution within Hosts from Serial Deep Sequence Data. Fraser C, ed. PLoS Computational Biology. 2012;8(11):e1002753.165. Verhofstede C, Vandekerckhove L, Eygen V Van, Demecheleer E, Vandenbroucke I, Winters B, Plum J, Vogelaers D, Stuyver L. CXCR4-using HIV type 1 variants are more commonly found in peripheral blood mononuclear cell DNA than in plasma RNA. Journal of acquired immune deficiency syndromes. 2009;50(2):126–36.166. Bachis A, Cruz MI, Mocchetti I. M-tropic HIV envelope protein gp120 exhibits a different neuropathological profle than T-tropic gp120 in rat striatum. The European journal of neuroscience. 2010;32(4):570–8.167. Koot M, Keet IPM, Vos AHV, de Goede REY, Roos MTL, Coutinho RA, Miedema F, Schellekens PTA, Tersmette M. Prognostic value of HIV-1 syncytium-inducing phenotype for rate of CD4+ cell depletion and progression to AIDS. Annals of internal medicine. 1993;118(9):681.180168. Poveda E, Briz V, Quiñones-Mateu M, Soriano V. HIV tropism: diagnostic tools and implications for disease progression and treatment with entry inhibitors. AIDS. 2006;20(10):1359–67.169. Penn ML, Grivel JC, Schramm B, Goldsmith MA, Margolis L. CXCR4 utilization is suffcient to trigger CD4+ T cell depletion in HIV-1-infected human lymphoid tissue. Proceedings of the National Academy of Sciences of the United States of America. 1999;96(2):663–8.170. Shepherd JC, Jacobson LP, Qiao W, Jamieson BD, Phair JP, Piazza P, Quinn TC, Margolick JB. Emergence and persistence of CXCR4-tropic HIV-1 in a population of men from the multicenter AIDS cohort study. The Journal of infectious diseases. 2008;198(8):1104–12.171. Tersmette M, Lange JM, de Goede RE, de Wolf F, Eeftink-Schattenkerk JK, Schellekens PT, Coutinho RA, Huisman JG, Goudsmit J, Miedema F. Association between biological properties of human immunodefciency virus variants and risk for AIDS and AIDS mortality. Lancet. 1989;1(8645):983–5.172. Richman DD, Bozzette SA. The impact of the syncytium-inducing phenotype of human immunodefciency virus on disease progression. The Journal of infectious diseases. 1994;169(5):968–74.173. Brumme ZL, Dong WWY, Yip B, Wynhoven B, Hoffman NG, Swanstrom R, Jensen MA, Mullins JI, Hogg RS, Montaner JSG, Harrigan PR. Clinical and immunological impact of HIV envelope V3 sequence variation after starting initial triple antiretroviral therapy. AIDS. 2004;18(4):F1–9.174. Lin NH, Kuritzkes DR. Tropism testing in the clinical management of HIV-1 infection. Current opinion in HIV and AIDS. 2009;4(6):481–7.175. Low AJ, Swenson LC, Harrigan PR. HIV coreceptor phenotyping in the clinical setting. AIDS reviews. 2008;10(3):143–51.181176. Durant J, Clevenbergh P, Halfon P, Delgiudice P, Porsin S, Simonet P, Montagne N, Boucher C, Schapiro JM, Dellamonica P. Drug-resistance genotyping in HIV-1 therapy: the VIRADAPT randomised controlled trial. Lancet. 1999;353(9171):2195–9.177. Winters B, Montaner J, Harrigan PR, Gazzard B, Pozniak A, Miller MD, Emery S, van Leth F, Robinson P, Baxter JD, Perez-Elias M, Castor D, Hammer S, Rinehart A, Vermeiren H, Van Craenenbroeck E, Bacheler L. Determination of clinically relevant cutoffs for HIV-1 phenotypic resistance estimates through a combined analysis of clinical trial and cohort data. Journal of acquired immune deficiency syndromes. 2008;48(1):26–34.178. Shafer RW, Dupnik K, Winters MA, Eshleman SH. A Guide to HIV-1 Reverse Transcriptase and Protease Sequencing for Drug Resistance Studies. HIV sequence compendium. 2001;2001:1–51.179. Winters B, Van Craenenbroeck E, Van Der Borght K, Lecocq P, Villacian J, Bacheler L. Clinical cut-offs for HIV-1 phenotypic resistance estimates: update based on recent pivotal clinical trial data and a revised approach to viral mixtures. Journal of Virological Methods. 2009;162:101–108.180. Schuurman R, Demeter L, Reichelderfer P, Tijnagel J, de Groot T, Boucher C. Worldwide Evaluation of DNA Sequencing Approaches for Identifcation of Drug Resistance Mutations in the Human Immunodefciency Virus Type 1 Reverse Transcriptase. Journal of clinical microbiology. 1999;37(7):2291–2296.181. Harrigan PR, Côté HC. Clinical utility of testing human immunodefciency virus for drug resistance. Clinical infectious diseases. 2000;30 Suppl 2:S117–22.182. Hirsch MS, Conway B, D’Aquila RT, Johnson VA, Brun-Vézinet F, Clotet B, Demeter LM, Hammer SM, Jacobsen DM, Kuritzkes DR, Loveday C, Mellors JW, Vella S, Richman DD. Antiretroviral drug resistance testing in adults with HIV infection: implications for clinical management. International AIDS Society--USA Panel. JAMA: the journal of the American Medical Association. 1998;279(24):1984–91.182183. Clavel F, Hance AJ. HIV drug resistance. The New England journal of medicine. 2004;350(10):1023–35.184. DeGruttola V, Dix L, D’Aquila R, Holder D, Phillips A, Ait-Khaled M, Baxter J, Clevenbergh P, Hammer S, Harrigan R, Katzenstein D, Lanier R, Miller M, Para M, Yerly S, Zolopa A, Murray J, Patick A, Miller V, et al. The relation between baseline HIV drug resistance and response to antiretroviral therapy: re-analysis of retrospective and prospective studies using a standardized data analysis plan. Antiviral therapy. 2000;5(1):41–8.185. Koot M, Vos AH, Keet RP, de Goede RE, Dercksen MW, Terpstra FG, Coutinho RA, Miedema F, Tersmette M. HIV-1 biological phenotype in long-term infected individuals evaluated with an MT-2 cocultivation assay. AIDS. 1992;6(1):49–54.186. Björndal A, Deng H, Jansson M, Fiore JR, Colognesi C, Karlsson A, Albert J, Scarlatti G, Littman DR, Fenyö EM. Coreceptor usage of primary human immunodefciency virus type 1 isolates varies according to biological phenotype. Journal of virology. 1997;71(10):7478–87.187. De Mendoza C, Van Baelen K, Poveda E, Rondelez E, Zahonero N, Stuyver L, Garrido C, Villacian J, Soriano V. Performance of a population-based HIV-1 tropism phenotypic assay and correlation with V3 genotypic prediction tools in recent HIV-1 seroconverters. Journal of acquired immune deficiency syndromes. 2008;48(3):241–4.188. Raymond S, Delobel P, Mavigner M, Cazabat M, Souyris C, Encinas S, Bruel P, Sandres-Sauné K, Marchou B, Massip P, Izopet J. Development and performance of a new recombinant virus phenotypic entry assay to determine HIV-1 coreceptor usage. Journal of clinical virology. 2010;47(2):126–30.189. Saliou A, Delobel P, Dubois M, Nicot F, Raymond S, Calvez V, Masquelier B, Izopet J. Concordance between two phenotypic assays and ultradeep pyrosequencing for determining HIV-1 tropism. Antimicrobial agents and chemotherapy. 2011;55(6):2831–6.183190. Skrabal K, Low AJ, Dong W, Sing T, Cheung PK, Mammano F, Harrigan PR. Determining human immunodefciency virus coreceptor use in a clinical setting: degree of correlation between two phenotypic assays and a bioinformatic model. Journal of clinical microbiology. 2007;45(2):279–84.191. Raymond S, Delobel P, Izopet J. Phenotyping methods for determining HIV tropism and applications in clinical settings. Current opinion in HIV and AIDS. 2012;7(5):463–9.192. Gonzalez-Serna A, Leal M, Genebat M, Abad MA, Garcia-Perganeda A, Ferrando-Martinez S, Ruiz-Mateos E. TROCAI (tropism coreceptor assay information): a new phenotypic tropism test and its correlation with Trofle enhanced sensitivity and genotypic approaches. Journal of clinical microbiology. 2010;48(12):4453–8.193. Weber J, Vazquez AC, Winner D, Gibson RM, Rhea AM, Rose JD, Wylie D, Henry K, Wright A, King K, Archer J, Poveda E, Soriano V, Robertson DL, Olivo PD, Arts EJ, Quiñones-Mateu ME. Sensitive cell-based assay for determination of human immunodefciency virus type 1 coreceptor tropism. Journal of clinical microbiology. 2013;51(5):1517–27.194. Whitcomb JM, Huang W, Fransen S, Limoli K, Toma J, Wrin T, Chappey C, Kiss LDB, Paxinos EE, Petropoulos CJ. Development and characterization of a novel single-cycle recombinant-virus assay to determine human immunodefciency virus type 1 coreceptor tropism. Antimicrobial agents and chemotherapy. 2007;51(2):566–75.195. Reeves JD, Coakley E, Petropoulos CJ, Whitcomb JM. An enhanced sensitivity Trofle HIV coreceptor tropism assay for selecting patients for therapy with entry inhibitors targeting CCR5: a review of analytical and clinical studies. Journal of Viral Entry. 2009;3:94–102.196. Woods CK, Brumme CJ, Liu TF, Chui CKS, Chu AL, Wynhoven B, Hall T a, Trevino C, Shafer RW, Harrigan PR. Automating HIV drug resistance genotyping with RECall, a freely accessible sequence analysis tool. Journal of clinical microbiology. 2012;50(6):1936–42.184197. Lengauer T, Sander O, Sierra S, Thielen A, Kaiser R. Bioinformatics prediction of HIV coreceptor usage. Nature biotechnology. 2007;25(12):1407–10.198. Garrido C, Roulet V, Chueca N, Poveda E, Aguilera A, Skrabal K, Zahonero N, Carlos S, García F, Faudon JL, Soriano V, de Mendoza C. Evaluation of eight different bioinformatics tools to predict viral tropism in different human immunodefciency virus type 1 subtypes. Journal of clinical microbiology. 2008;46(3):887–91.199. Pillai S, Good B, Richman D, Corbeil J. A new perspective on V3 phenotype prediction. AIDS research and human retroviruses. 2003;19(2):145–9.200. Swenson LC, Boehme R, Thielen A, McGovern RA, Harrigan PR. Genotypic determination of HIV-1 tropism in the clinical setting. HIV Therapy. 2010;4(3):293–303.201. Lee GQ, Cheung PK, Swenson LC, Harrigan PR. HIV tropism and CCR5 antagonists. Hot Topics in HIV and Other Retroviruses. 2012;3:1–21.202. Evans MC, Paquet AC, Huang W, Napolitano L, Frantzell A, Toma J, Stawiski EW, Goetz MB, Petropoulos CJ, Whitcomb J, Coakley E, Haddad M. A case-based reasoning system for genotypic prediction of HIV-1 co-receptor tropism. Journal of bioinformatics and computational biology. 2013;11(4):1350006.203. Jensen M, Li F, van’t Wout A, Nickle D. Improved coreceptor usage prediction and genotypic monitoring of R5-to-X4 transition by motif analysis of human immunodefciency virus type 1 env V3 loop. Journal of Virology. 2003;77(24):13376–13388.204. Sing T, Low AJ, Beerenwinkel N, Sander O, Cheung PK, Domingues FS, Büch J, Däumer M, Kaiser R, Lengauer T, Harrigan PR. Predicting HIV coreceptor usage on the basis of genetic and clinical covariates. Antiviral therapy. 2007;12(7):1097–106.185205. Jensen MA, Coetzer M, van’t Wout AB, Morris L, Mullins JI. A reliable phenotype predictor for human immunodefciency virus type 1 subtype C based on envelope V3 sequences. Journal of virology. 2006;80(10):4698.206. Shankarappa R, Margolick JB, Gange SJ, Rodrigo A, Upchurch D, Farzadegan H, Gupta P, Rinaldo CR, Learn GH, He X, Huang XL, Mullins JI. Consistent viral evolutionary changes associated with the progression of human immunodefciency virus type 1 infection. Journal of virology. 1999;73(12):10489–502.207. Curlin ME, Zioni R, Hawes SE, Liu Y, Deng W, Gottlieb GS, Zhu T, Mullins JI. HIV-1 Envelope Subregion Length Variation during Disease Progression. Trkola A, ed. PLoS Pathogens. 2010;6(12):e1001228.208. Weiser B, Philpott S, Klimkait T, Burger H, Kitchen C, Bürgisser P, Gorgievski M, Perrin L, Piffaretti J-C, Ledergerber B. HIV-1 coreceptor usage and CXCR4-specifc viral load predict clinical disease progression during combination antiretroviral therapy. AIDS. 2008;22(4):469–79.209. Nelson JAE, Fiscus SA, Swanstrom R. Evolutionary variants of the human immunodefciency virus type 1 V3 region characterized by using a heteroduplex tracking assay. Journal of virology. 1997;71(11):8750–8758.210. McLaughlin S, Swenson LC, Hu S, Hughes P, Harrigan PR, Coombs RW, Frenkel LM. Development of a novel codon-specifc polymerase chain reaction for the detection of CXCR4-utilizing HIV type 1 subtype B. AIDS research and human retroviruses. 2013;29(5):814–25.211. Abbate I, Rozera G, Tommasi C, Bruselles A, Bartolini B, Chillemi G, Nicastri E, Narciso P, Ippolito G, Capobianchi MR. Analysis of co-receptor usage of circulating viral and proviral HIV genome quasispecies by ultra-deep pyrosequencing in patients who are candidates for CCR5 antagonist treatment. Clinical microbiology and infection. 2011;17(5):725–31.186212. Uzgiris AJ, Pandit S, Shields A, Zhang N, Chapman D, Valdez H, Heera J, Kaiser R, Thielen A, Turczyn J. Performance Characteristics of a DNA Sequencing-Based HIV-1 Coreceptor Tropism Assay for Use in a Prospective Double-Blind Randomized Phase 3 Study (A4001095, MODERN). In: American Association for Clinical Chemistry Annual Meeting. Los Angeles, CA; 2012:Abstract E–97.213. Svicher V, Alteri C, Montano M, D’Arrigo R, Andreoni M, Angarano G, Antinori A, Antonelli G, Allice T, Bagnarelli P, Baldanti F, Bertoli A, Borderi M, Boeri E, Bon I, Bruzzone B, Callegaro AP, Capobianchi MR, Carosi G, et al. Performance of genotypic tropism testing on proviral DNA in clinical practice: results from the DIVA study group. New Microbiologica. 2012;35(1):17–25.214. Soulié C, Fourati S, Lambert-Niclot S, Malet I, Wirden M, Tubiana R, Valantin M-A, Katlama C, Calvez V, Marcelin A-G. Factors associated with proviral DNA HIV-1 tropism in antiretroviral therapy-treated patients with fully suppressed plasma HIV viral load: implications for the clinical use of CCR5 antagonists. Journal of antimicrobial chemotherapy. 2010;65(4):749–51.215. Soulie C, Calvez V, Marcelin A-G. Coreceptor usage in different reservoirs. Current opinion in HIV and AIDS. 2012;7(5):450–455.216. Monno L, Saracino A, Scudeller L, Punzi G, Brindicci G, Altamura M, Lagioia A, Ladisa N, Angarano G. Impact of mutations outside the V3 region on coreceptor tropism phenotypically assessed in patients infected with HIV-1 subtype B. Antimicrobial agents and chemotherapy. 2011;55(11):5078–84.217. Thielen A, Sichtig N, Kaiser R, Lam J, Harrigan PR, Lengauer T. Improved Prediction of HIV-1 Coreceptor Usage with Sequence Information from the Second Hypervariable Loop of gp120. The Journal of infectious diseases. 2010;202(9):1435–1443.187218. Thielen A, Lengauer T, Swenson LC, Dong WWY, McGovern RA, Lewis M, James I, Heera J, Valdez H, Harrigan PR. Mutations in gp41 are correlated with coreceptor tropism but do not improve prediction methods substantially. Antiviral Therapy. 2011;16:319–328.219. Huang W, Toma J, Fransen S, Stawiski E, Reeves JD, Whitcomb JM, Parkin N, Petropoulos CJ. Coreceptor tropism can be influenced by amino acid substitutions in the gp41 transmembrane subunit of human immunodefciency virus type 1 envelope protein. Journal of virology. 2008;82(11):5584–93.220. Hoffman NG, Seillier-Moiseiwitsch F, Ahn J, Walker JM, Swanstrom R. Variability in the human immunodefciency virus type 1 gp120 Env protein linked to phenotype-associated changes in the V3 loop. Journal of virology. 2002;76(8):3852–64.221. Esbjörnsson J, Månsson F, Martínez-Arias W, Vincic E, Biague AJ, da Silva ZJ, Fenyö EM, Norrgren H, Medstrand P. Frequent CXCR4 tropism of HIV-1 subtype A and CRF02_AG during late-stage disease - indication of an evolving epidemic in West Africa. Retrovirology. 2010;7:23.222. Huang W, Eshleman SH, Toma J, Fransen S, Stawiski E, Paxinos EE, Whitcomb JM, Young AM, Donnell D, Mmiro F, Musoke P, Guay LA, Jackson JB, Parkin NT, Petropoulos CJ. Coreceptor tropism in human immunodefciency virus type 1 subtype D: high prevalence of CXCR4 tropism and heterogeneous composition of viral populations. Journal of virology. 2007;81(15):7885–93.223. Gonzalez-Serna A, McGovern RA, Harrigan PR, Vidal F, Poon AFY, Ferrando-Martinez S, Abad MA, Genebat M, Leal M, Ruiz-Mateos E. Correlation of the virological response to short-term maraviroc monotherapy with standard and deep-sequencing-based genotypic tropism prediction methods. Antimicrobial agents and chemotherapy. 2012;56(3):1202–7.224. Genebat M, Ruiz-Mateos E, León JA, González-Serna A, Pulido I, Rivas I, Ferrando-Martínez S, Sánchez B, Muñoz-Fernández MA, Leal M. Correlation between the Trofle test and 188virological response to a short-term maraviroc exposure in HIV-infected patients. The Journal of antimicrobial chemotherapy. 2009;64(4):845–9.225. Gonzalez-Serna A, Romero-Sánchez MC, Ferrando-Martinez S, Genebat M, Vidal F, Muñoz-Fernández MÁ, Abad MA, Leal M, Ruiz-Mateos E. HIV-1 tropism evolution after short-term maraviroc monotherapy in HIV-1-infected patients. Antimicrobial agents and chemotherapy. 2012;56(7):3981–3.226. Genebat M, Pulido I, Romero-Sánchez MC, González-Serna A, Ferrando-Martínez S, Machmach K, Pacheco YM, Muñoz-Fernández MÁ, Ruiz-Mateos E, Leal M. Patients on a combined antiretroviral therapy after maraviroc clinical test show no immunovirological impairment. Antiviral research. 2012;95(3):207–11.227. Chueca N, Garrido C, Álvarez M. Improvement in the determination of HIV‐1 tropism using the V3 gene sequence and a combination of bioinformatic tools. Journal of medical virology. 2009;81:763–767.228. Palmer S, Kearney M, Maldarelli F, Halvas EK, Bixby CJ, Bazmi H, Rock D, Falloon J, Davey RT, Dewar RL, Metcalf JA, Hammer S, Mellors JW, Coffn JM. Multiple, linked human immunodefciency virus type 1 drug resistance mutations in treatment-experienced patients are missed by standard genotype analysis. Journal of clinical microbiology. 2005;43(1):406–13.229. Low AJ, Dong W, Chan D, Sing T, Swanstrom R, Jensen M, Pillai S, Good B, Harrigan PR. Current V3 genotyping algorithms are inadequate for predicting X4 co-receptor usage in clinical isolates. AIDS. 2007;21(14):F17–24.230. Raymond S, Delobel P, Mavigner M, Cazabat M, Souyris C, Sandres-Sauné K, Cuzin L, Marchou B, Massip P, Izopet J. Correlation between genotypic predictions based on V3 sequences and phenotypic determination of HIV-1 tropism. AIDS. 2008;22(14):F11–6.231. Prosperi MCF, Bracciale L, Fabbiani M, Di Giambenedetto S, Razzolini F, Meini G, Colafgli M, Marzocchetti A, Cauda R, Zazzi M, De Luca A. Comparative determination of HIV-1 co-189receptor tropism by Enhanced Sensitivity Trofle, gp120 V3-loop RNA and DNA genotyping. Retrovirology. 2010;7:56.232. Archer J, Braverman MS, Taillon BE, Desany B, James I, Harrigan PR, Lewis M, Robertson DL. Detection of low-frequency pretherapy chemokine (CXC motif) receptor 4 (CXCR4)-using HIV-1 with ultra-deep pyrosequencing. AIDS. 2009;23(10):1209–18.233. Li JZ, Paredes R, Ribaudo HJ, Svarovskaia ES, Metzner KJ, Kozal MJ, Hullsiek KH, Balduin M, Jakobsen MR, Geretti AM, Thiebaut R, Ostergaard L, Masquelier B, Johnson JA, Miller MD, Kuritzkes DR. Low-frequency HIV-1 drug resistance mutations and risk of NNRTI-based antiretroviral treatment failure: a systematic review and pooled analysis. JAMA: the journal of the American Medical Association. 2011;305(13):1327–35.234. Li JZ, Kuritzkes DR. Clinical Implications of HIV-1 Minority Variants. Clinical infectious diseases. 2013;56(11):1667–74.235. Vandenbroucke I, Van Marck H, Verhasselt P, Thys K, Mostmans W, Dumont S, Van Eygen V, Coen K, Tuefferd M, Aerssens J. Minor variant detection in amplicons using 454 massive parallel pyrosequencing: experiences and considerations for successful applications. BioTechniques. 2011;51(3):167–77.236. Lee GQ, Harrigan PR, Dong W, Poon AFY, Heera J, Demarest J, Rinehart A, Chapman D, Valdez H, Portsmouth S. Comparison of population and 454 “deep” sequence analysis for HIV type 1 tropism versus the original trofle assay in non-B subtypes. AIDS research and human retroviruses. 2013;29(6):979–84.237. Lee GQ, Swenson LC, Poon AFY, Martin JN, Hatano H, Deeks SG, Harrigan PR. Prolonged and Substantial Discordance in Prevalence of Raltegravir-Resistant HIV-1 in Plasma versus PBMC Samples Revealed by 454 “Deep” Sequencing. Wainberg M, ed. PLoS ONE. 2012;7(9):e46181.190238. Archer J, Rambaut A, Taillon BE, Harrigan PR, Lewis M, Robertson DL. The Evolutionary Analysis of Emerging Low Frequency HIV-1 CXCR4 Using Variants through Time—An Ultra-Deep Approach. Kosakovsky Pond SL, ed. PLoS Computational Biology. 2010;6(12):e1001022.239. Fischer W, Ganusov V V, Giorgi EE, Hraber PT, Keele BF, Leitner T, Han CS, Gleasner CD, Green L, Lo C-C, Nag A, Wallstrom TC, Wang S, McMichael AJ, Haynes BF, Hahn BH, Perelson AS, Borrow P, Shaw GM, et al. Transmission of Single HIV-1 Genomes and Dynamics of Early Immune Escape Revealed by Ultra-Deep Sequencing. PloS one. 2010;5(8).240. Droege M, Hill B. The Genome Sequencer FLX System--longer reads, more applications, straight forward bioinformatics and more complete data sets. Journal of biotechnology. 2008;136(1-2):3–10.241. McGovern RA, Symons J, Poon AFY, Harrigan PR, van Lelyveld SFL, Hoepelman AIM, van Ham PM, Dong W, Wensing AMJ, Nijhuis M. Maraviroc treatment in non-R5-HIV-1-infected patients results in the selection of extreme CXCR4-using variants with limited effect on the total viral setpoint. The Journal of antimicrobial chemotherapy. 2013;68(9):2007–14.242. Mardis ER. Next-generation DNA sequencing methods. Annual review of genomics and human genetics. 2008;9:387–402.243. Fakhrai-Rad H, Pourmand N, Ronaghi M. Pyrosequencing: an accurate detection platform for single nucleotide polymorphisms. Human mutation. 2002;19(5):479–85.244. Metzker ML. Sequencing technologies - the next generation. Nature reviews Genetics. 2010;11(1):31–46.245. Wang C, Mitsuya Y, Gharizadeh B, Ronaghi M, Shafer RW. Characterization of mutation spectra with ultra-deep pyrosequencing: application to HIV-1 drug resistance. Genome research. 2007;17(8):1195–201.191246. Glenn TC. Field guide to next-generation DNA sequencers. Molecular Ecology Resources. 2011;11(5):759–769.247. Shendure J, Ji H. Next-generation DNA sequencing. Nature biotechnology. 2008;26(10):1135–45.248. Archer J, Weber J, Henry K, Winner D, Gibson R, Lee L, Paxinos E, Arts EJ, Robertson DL, Mimms L, Quiñones-Mateu ME. Use of Four Next-Generation Sequencing Platforms to Determine HIV-1 Coreceptor Tropism. PloS one. 2012;7(11):e49602.249. Margulies M, Egholm M, Altman WE, Attiya S, Bader JS, Bemben L a, Berka J, Braverman MS, Chen Y-J, Chen Z, Dewell SB, Du L, Fierro JM, Gomes X V, Godwin BC, He W, Helgesen S, Ho CH, Ho CH, et al. Genome sequencing in microfabricated high-density picolitre reactors. Nature. 2005;437(7057):376–80.250. Zhou X, Ren L, Li Y, Zhang M, Yu Y, Yu J. The next-generation sequencing technology: a technology review and future perspective. Science China Life sciences. 2010;53(1):44–57.251. Fedurco M, Romieu A, Williams S, Lawrence I, Turcatti G. BTA, a novel reagent for DNA attachment on glass and effcient generation of solid-phase amplifed DNA colonies. Nucleic acids research. 2006;34(3):e22.252. Dressman D, Yan H, Traverso G, Kinzler KW, Vogelstein B. Transforming single DNA molecules into fluorescent magnetic particles for detection and enumeration of genetic variations. Proceedings of the National Academy of Sciences of the United States of America. 2003;100(15):8817–22.253. Eid J, Fehr A, Gray J, Luong K, Lyle J, Otto G, Peluso P, Rank D, Baybayan P, Bettman B, Bibillo A, Bjornson K, Chaudhuri B, Christians F, Cicero R, Clark S, Dalal R, Dewinter A, Dixon J, et al. Real-time DNA sequencing from single polymerase molecules. Science. 2009;323(5910):133–8.192254. Frese KS, Katus HA, Meder B. Next-Generation Sequencing: From Understanding Biology to Personalized Medicine. Biology. 2013;2:378–398.255. Peckham HE, Mclaughlin SF, Ni JN, Rhodes MD, Malek JA, Mckernan KJ, Blanchard AP. SOLiDTM Sequencing and 2-Base Encoding. SOLiDTM Sequencing Chemistry, Applied Biosystems. (2624). Available at: http://www3.appliedbiosystems.com/cms/groups/mcb_marketing/documents/generaldocuments/cms_057810.pdf. Accessed December 12, 2013.256. Applied Biosystems. Specification Sheet - 5500 Series Genetic Analysis Systems.; 2011.257. Shafer RW. Low-abundance drug-resistant HIV-1 variants: fnding signifcance in an era of abundant diagnostic and therapeutic options. The Journal of infectious diseases. 2009;199(5):610–2.258. Gianella S, Richman DD. Minority variants of drug-resistant HIV. The Journal of infectious diseases. 2010;202(5):657–66.259. Weiss RA. HIV and AIDS in relation to other pandemics. EMBO reports. 2003;4:S10–S14.260. Hogg RS, Schechter MT, Montaner JS, Goldstone I, Craib K, O’Shaughnessy M V. Impact of HIV infection and AIDS on death rates in British Columbia and Canada. CMAJ: Canadian Medical Association journal. 1994;150(5):711–7.261. McMichael AJ, Hanke T. HIV vaccines 1983-2003. Nature medicine. 2003;9(7):874–80.262. Pomerantz RJ, Horn DL. Twenty years of therapy for HIV-1 infection. Nature medicine. 2003;9(7):867–73.263. Richman DD. HIV chemotherapy. Nature. 2001;410(6831):995–1001.193264. Concorde Coordinating Committee. Concorde: MRC/ANRS randomised double-blind controlled trial of immediate and deferred zidovudine in symptom-free HIV infection. Concorde Coordinating Committee. Lancet. 1994;343(8902):871–81.265. Ho DD, Neumann AU, Perelson AS, Chen W, Leonard JM, Markowitz M. Rapid turnover of plasma virions and CD4 lymphocytes in HIV-1 infection. Nature. 1995;373(6510):123–6.266. Fauci AS. HIV and AIDS: 20 years of science. Nature medicine. 2003;9(7):839–43.267. Sepkowitz KA. AIDS--the frst 20 years. The New England journal of medicine. 2001;344(23):1764–72.268. Hammer SM, Squires KE, Hughes MD, Grimes JM, Demeter LM, Currier JS, Eron JJ, Feinberg JE, Balfour HH, Deyton LR, Chodakewitz JA, Fischl MA. A controlled trial of two nucleoside analogues plus indinavir in persons with human immunodefciency virus infection and CD4 cell counts of 200 per cubic millimeter or less. AIDS Clinical Trials Group 320 Study Team. The New England journal of medicine. 1997;337(11):725–33.269. Montaner JS, Reiss P, Cooper D, Vella S, Harris M, Conway B, Wainberg MA, Smith D, Robinson P, Hall D, Myers M, Lange JM. A randomized, double-blind trial comparing combinations of nevirapine, didanosine, and zidovudine for HIV-infected patients: the INCAS Trial. Italy, The Netherlands, Canada and Australia Study. JAMA: the journal of the American Medical Association. 1998;279(12):930–7.270. Mellors JW, Kingsley LA, Rinaldo CR, Todd JA, Hoo BS, Kokka RP, Gupta P. Quantitation of HIV-1 RNA in plasma predicts outcome after seroconversion. Annals of internal medicine. 1995;122(8):573–9.271. Weber J, Vazquez AC, Winner D, Rose JD, Wylie D, Rhea AM, Henry K, Pappas J, Wright A, Mohamed N, Gibson R, Rodriguez B, Soriano V, King K, Arts EJ, Olivo PD, Quiñones-Mateu ME. Novel method for simultaneous quantifcation of phenotypic resistance to maturation, protease, reverse transcriptase, and integrase HIV inhibitors based on 1943’Gag(p2/p7/p1/p6)/PR/RT/INT-recombinant viruses: a useful tool in the multitarget era of antiretro. Antimicrobial agents and chemotherapy. 2011;55(8):3729–42.272. Liu TF, Shafer RW. Web resources for HIV type 1 genotypic-resistance test interpretation. Clinical infectious diseases. 2006;42(11):1608–18.273. Baxter JD, Mayers DL, Wentworth DN, Neaton JD, Hoover ML, Winters MA, Mannheimer SB, Thompson MA, Abrams DI, Brizz BJ, Ioannidis JP, Merigan TC. A randomized study of antiretroviral management based on plasma genotypic antiretroviral resistance testing in patients failing therapy. CPCRA 046 Study Team for the Terry Beirn Community Programs for Clinical Research on AIDS. AIDS. 2000;14(9):F83–93.274. Hirsch MS, Günthard HF, Schapiro JM, Brun-Vézinet F, Clotet B, Hammer SM, Johnson V a, Kuritzkes DR, Mellors JW, Pillay D, Yeni PG, Jacobsen DM, Richman DD. Antiretroviral drug resistance testing in adult HIV-1 infection: 2008 recommendations of an International AIDS Society-USA panel. Clinical infectious diseases. 2008;47(2):266–85.275. Van Baelen K, Salzwedel K, Rondelez E, Van Eygen V, De Vos S, Verheyen A, Steegen K, Verlinden Y, Allaway GP, Stuyver LJ. Susceptibility of human immunodefciency virus type 1 to the maturation inhibitor bevirimat is modulated by baseline polymorphisms in Gag spacer peptide 1. Antimicrobial agents and chemotherapy. 2009;53(5):2185–8.276. Vandekerckhove L, Wensing A, Kaiser R, Brun-Vézinet F, Clotet B, De Luca A, Dressler S, Garcia F, Geretti A, Klimkait T, Korn K, Masquelier B, Perno C, Schapiro J, Soriano V, Sönnerborg A, Vandamme A-M, Verhofstede C, Walter H, et al. European guidelines on the clinical management of HIV-1 tropism testing. The Lancet infectious diseases. 2011;11(5):394–407.277. The Therapeutic Guidelines Committee of the BC Centre for Excellence in HIV/AIDS. Therapeutic Guidelines: Antiretroviral Treatment (ARV) of Adult HIV Infection. 2013;(February):1–48.195278. Williams I, Churchill D, Anderson J, Boffto M, Bower M, Cairns G, Cwynarski K, Edwards S, Fidler S, Fisher M, Freedman A, Geretti AM, Gilleece Y, Horne R, Johnson M, Khoo S, Leen C, Marshall N, Nelson M, et al. British HIV Association guidelines for the treatment of HIV-1-positive adults with antiretroviral therapy 2012 British HIV Association guidelines for the treatment of HIV-1-positive adults with antiretroviral therapy 2012. HIV Medicine. 2012;13(July):1–85.279. European AIDS Clinical Society, EACS. European AIDS Clinical Society Guidelines: Clinical Management and Treatment of HIV-infected Adults in Europe Version 7.0. 2013;(October):1–82.280. Dybul M, Fauci AS, Bartlett JG, Kaplan JE, Pau AK. Guidelines for using antiretroviral agents among HIV-infected adults and adolescents. Annals of internal medicine. 2002;137(5 Pt 2):381–433.281. Hunt PW, Harrigan PR, Huang W, Bates M, Williamson DW, McCune JM, Price RW, Spudich SS, Lampiris H, Hoh R, Leigler T, Martin JN, Deeks SG. Prevalence of CXCR4 tropism among antiretroviral-treated HIV-1-infected patients with detectable viremia. The Journal of infectious diseases. 2006;194(7):926–30.282. Seclén E, Soriano V, González MM, Martín-Carbonero L, Gellermann H, Distel M, Kadus W, Poveda E. Impact of Baseline HIV-1 Tropism on Viral Response and CD4 Cell Count Gains in HIV-Infected Patients Receiving First-line Antiretroviral Therapy. The Journal of infectious diseases. 2011;204(1):139–44.283. Waters LJ, Scourfeld AT, Marcano M, Gazzard BG, Bower M, Nelson M, Stebbing J. The Evolution of Coreceptor Tropism in HIV-infected Patients Interrupting Suppressive Antiretroviral Therapy. Clinical infectious diseases. 2011;52(5):671–3.284. Kuritzkes D, Kar S, Kirkpatrick P. Maraviroc. Nature Reviews Drug Discovery. 2008;7(1):15–16.196285. Marier J-F, Trinh M, Pheng LH, Palleja SM, Martin DE. Pharmacokinetics and pharmacodynamics of TBR-652, a novel CCR5 antagonist, in HIV-1-infected, antiretroviral treatment-experienced, CCR5 antagonist-naïve patients. Antimicrobial agents and chemotherapy. 2011;55(6):2768–74.286. Carter NJ, Keating GM. Maraviroc. Drugs. 2007;67(15):2277–88.287. Fätkenheuer G, Pozniak AL, Johnson M a, Plettenberg A, Staszewski S, Hoepelman AIM, Saag MS, Goebel FD, Rockstroh JK, Dezube BJ, Jenkins TM, Medhurst C, Sullivan JF, Ridgway C, Abel S, James IT, Youle M, van der Ryst E. Effcacy of short-term monotherapy with maraviroc, a new CCR5 antagonist, in patients infected with HIV-1. Nature medicine. 2005;11(11):1170–2.288. Gulick RM, Lalezari J, Goodrich J, Clumeck N, DeJesus E, Horban A, Nadler J, Clotet B, Karlsson A, Wohlfeiler M, Montana JB, McHale M, Sullivan J, Ridgway C, Felstead S, Dunne MW, van der Ryst E, Mayer H. Maraviroc for Previously Treated Patients with R5 HIV-1 Infection. New England Journal of Medicine. 2008;359(14):1429–1441.289. Fätkenheuer G, Nelson M, Lazzarin A, Konourina I, Hoepelman AIM, Lampiris H, Hirschel B, Tebas P, Raff F, Trottier B, Bellos N, Saag M, Cooper DA, Westby M, Tawadrous M, Sullivan JF, Ridgway C, Dunne MW, Felstead S, et al. Subgroup analyses of maraviroc in previously treated R5 HIV-1 infection. The New England journal of Medicine. 2008;359(14):1442–55.290. Cooper D, Heera J, Goodrich J, Tawadrous M, Saag M, Dejesus E, Clumeck N, Walmsley S, Ting N, Coakley E, Reeves JD, Reyes-Teran G, Westby M, Van Der Ryst E, Ive P, Mohapi L, Mingrone H, Horban A, Hackman F, et al. Maraviroc versus efavirenz, both in combination with zidovudine-lamivudine, for the treatment of antiretroviral-naive subjects with CCR5-tropic HIV-1 infection. The Journal of infectious diseases. 2010;201(6):803–13.291. Saag M, Goodrich J, Fätkenheuer G, Clotet B, Clumeck N, Sullivan J, Westby M, van der Ryst E, Mayer H. A double-blind, placebo-controlled trial of maraviroc in treatment-experienced patients infected with non-R5 HIV-1. The Journal of infectious diseases. 2009;199(11):1638–47.197292. Saag M, Heera J, Goodrich J, Dejesus E, Clumeck N, Cooper D, Walmsley S, Ting N, Coakley E, Reeves J, Westby M, van der Ryst E, Mayer H. Reanalysis of the MERIT Study with the Enhanced Trofle Assay (MERIT-ES). In: 48th Annual ICAAC/IDSA 46th Annual Meeting.Vol 88. Washington DC; 2008:Poster H–1232.293. Eigen M. Viral quasispecies. Scientific American. 1993;269(1):42–9.294. Nijhuis M, Deeks S, Boucher C. Implications of antiretroviral resistance on viral ftness. Current opinion in infectious diseases. 2001;14(1):23–8.295. Gianella S, Delport W, Pacold ME, Young J a, Choi JY, Little SJ, Richman DD, Kosakovsky Pond SL, Smith DM. Detection of minority resistance during early HIV-1 infection: natural variation and spurious detection rather than transmission and evolution of multiple viral variants. Journal of virology. 2011;85(16):8359–67.296. Johnson JA, Li J-F, Wei X, Lipscomb J, Irlbeck D, Craig C, Smith A, Bennett DE, Monsour M, Sandstrom P, Lanier ER, Heneine W. Minority HIV-1 drug resistance mutations are present in antiretroviral treatment-naïve populations and associate with reduced treatment effcacy. PLoS medicine. 2008;5(7):e158.297. Peuchant O, Thiébaut R, Capdepont S, Lavignolle-Aurillac V, Neau D, Morlat P, Dabis F, Fleury H, Masquelier B. Transmission of HIV-1 minority-resistant variants and response to frst-line antiretroviral therapy. AIDS. 2008;22(12):1417–23.298. Metzner KJ, Giulieri SG, Knoepfel S a, Rauch P, Burgisser P, Yerly S, Günthard HF, Cavassini M. Minority quasispecies of drug-resistant HIV-1 that lead to early therapy failure in treatment-naive and -adherent patients. Clinical infectious diseases. 2009;48(2):239–47.299. Metzner KJ, Rauch P, Walter H, Boesecke C, Zöllner B, Jessen H, Schewe K, Fenske S, Gellermann H, Stellbrink H-J. Detection of minor populations of drug-resistant HIV-1 in acute seroconverters. AIDS. 2005;19(16):1819–25.198300. Charpentier C, Laureillard D, Piketty C, Tisserand P, Batisse D, Karmochkine M, Si-Mohamed A, Weiss L. High frequency of integrase Q148R minority variants in HIV-infected patients naive of integrase inhibitors. AIDS. 2010;24(6):867–73.301. Doualla-Bell F, Avalos A, Brenner B, Gaolathe T, Mine M, Gaseitsiwe S, Oliveira M, Moisi D, Ndwapi N, Moffat H, Essex M, Wainberg M a. High prevalence of the K65R mutation in human immunodefciency virus type 1 subtype C isolates from infected patients in Botswana treated with didanosine-based regimens. Antimicrobial agents and chemotherapy. 2006;50(12):4182–5.302. Kozal MJ, Chiarella J, St John EP, Moreno EA, Simen BB, Arnold TE, Lataillade M. Prevalence of low-level HIV-1 variants with reverse transcriptase mutation K65R and the effect of antiretroviral drug exposure on variant levels. Antiviral therapy. 2011;16(6):925–9.303. Varghese V, Shahriar R, Rhee S-Y, Liu T, Simen BB, Egholm M, Hanczaruk B, Blake LA, Gharizadeh B, Babrzadeh F, Bachmann MH, Fessel WJ, Shafer RW. Minority variants associated with transmitted and acquired HIV-1 nonnucleoside reverse transcriptase inhibitor resistance: implications for the use of second-generation nonnucleoside reverse transcriptase inhibitors. Journal of acquired immune deficiency syndromes. 2009;52(3):309–15.304. Paredes R, Lalama CM, Ribaudo HJ, Schackman BR, Shikuma C, Giguel F, Meyer W a, Johnson V a, Fiscus S a, D’Aquila RT, Gulick RM, Kuritzkes DR. Pre-existing minority drug-resistant HIV-1 variants, adherence, and risk of antiretroviral treatment failure. The Journal of infectious diseases. 2010;201(5):662–71.305. Dykes C, Najjar J, Bosch RJ, Wantman M, Furtado M, Hart S, Hammer SM, Demeter LM. Detection of drug-resistant minority variants of HIV-1 during virologic failure of indinavir, lamivudine, and zidovudine. The Journal of infectious diseases. 2004;189(6):1091–6.306. Roquebert B, Malet I, Wirden M, Tubiana R, Valantin M-A, Simon A, Katlama C, Peytavin G, Calvez V, Marcelin A-G. Role of HIV-1 minority populations on resistance mutational pattern evolution and susceptibility to protease inhibitors. AIDS. 2006;20(2):287–9.199307. Le T, Chiarella J, Simen BB, Hanczaruk B, Egholm M, Landry ML, Dieckhaus K, Rosen MI, Kozal MJ. Low-abundance HIV drug-resistant viral variants in treatment-experienced persons correlate with historical antiretroviral use. PloS one. 2009;4(6):e6079.308. Lecossier D, Shulman NS, Morand-Joubert L, Shafer RW, Joly V, Zolopa AR, Clavel F, Hance AJ. Detection of minority populations of HIV-1 expressing the K103N resistance mutation in patients failing nevirapine. Journal of acquired immune deficiency syndromes. 2005;38(1):37–42.309. Flys TS, Chen S, Jones DC, Hoover DR, Church JD, Fiscus SA, Mwatha A, Guay LA, Mmiro F, Musoke P, Kumwenda N, Taha TE, Jackson JB, Eshleman SH. Quantitative analysis of HIV-1 variants with the K103N resistance mutation after single-dose nevirapine in women with HIV-1 subtypes A, C, and D. Journal of acquired immune deficiency syndromes. 2006;42(5):610–3.310. Coovadia A, Hunt G, Abrams EJ, Sherman G, Meyers T, Barry G, Malan E, Marais B, Stehlau R, Ledwaba J, Hammer SM, Morris L, Kuhn L. Persistent minority K103N mutations among women exposed to single-dose nevirapine and virologic response to nonnucleoside reverse-transcriptase inhibitor-based therapy. Clinical infectious diseases. 2009;48(4):462–72.311. Gorry PR, Zhang C, Wu S, Kunstman K, Trachtenberg E, Phair J, Wolinsky S, Gabuzda D. Persistence of dual-tropic HIV-1 in an individual homozygous for the CCR5 Delta 32 allele. Lancet. 2002;359(9320):1832–4.312. Grivel J-C, Shattock RJ, Margolis LB. Selective transmission of R5 HIV-1 variants: where is the gatekeeper? Journal of translational medicine. 2011;9 Suppl 1(Suppl 1):S6.313. Hoffmann C. The epidemiology of HIV coreceptor tropism. European journal of medical research. 2007;12(9):385–90.314. Low AJ, Marchant D, Brumme CJ, Brumme ZL, Dong W, Sing T, Hogg RS, Montaner JSG, Gill V, Cheung PK, Harrigan PR. CD4-dependent characteristics of coreceptor use and HIV type 1 V3 sequence in a large population of therapy-naive individuals. AIDS research and human retroviruses. 2008;24(2):219–28.200315. Irlbeck DM, Amrine-Madsen H, Kitrinos KM, Labranche CC, Demarest JF. Chemokine (C-C motif) receptor 5-using envelopes predominate in dual/mixed-tropic HIV from the plasma of drug-naive individuals. AIDS. 2008;22(12):1425–31.316. Westby M, Lewis M, Whitcomb J, Youle M, Pozniak AL, James IT, Jenkins TM, Perros M, Van Der Ryst E. Emergence of CXCR4-using human immunodefciency virus type 1 (HIV-1) variants in a minority of HIV-1-infected patients following treatment with the CCR5 antagonist maraviroc is from a pretreatment CXCR4-using virus reservoir. Journal of virology. 2006;80(10):4909–4920.317. Tsibris AMN, Korber B, Arnaout R, Russ C, Lo C-C, Leitner T, Gaschen B, Theiler J, Paredes R, Su Z, Hughes MD, Gulick RM, Greaves W, Coakley E, Flexner C, Nusbaum C, Kuritzkes DR. Quantitative deep sequencing reveals dynamic HIV-1 escape and large population shifts during CCR5 antagonist therapy in vivo. PloS one. 2009;4(5):e5683.318. Su Z, Gulick RM, Krambrink A, Coakley E, Hughes MD, Han D, Flexner C, Wilkin TJ, Skolnik PR, Greaves WL, Kuritzkes DR, Reeves JD. Response to vicriviroc in treatment-experienced subjects, as determined by an enhanced-sensitivity coreceptor tropism assay: reanalysis of AIDS clinical trials group A5211. The Journal of infectious diseases. 2009;200(11):1724–8.319. Briggs DR, Tuttle DL, Sleasman JW, Goodenow MM. Envelope V3 amino acid sequence predicts HIV-1 phenotype (co-receptor usage and tropism for macrophages). AIDS. 2000;14(18):2937–9.320. Hogg RS, Yip B, Chan KJ, Wood E, Craib KJ, O’Shaughnessy M V, Montaner JS. Rates of disease progression by baseline CD4 cell count and viral load after initiating triple-drug therapy. JAMA: the journal of the American Medical Association. 2001;286(20):2568–77.321. Swenson LC, Däumer M, Paredes R. Next-generation sequencing to assess HIV tropism. Current opinion in HIV and AIDS. 2012;7(5):478–485.201322. Huang X, Zhang J. Methods for comparing a DNA sequence with a protein sequence. Computer applications in the biosciences : CABIOS. 1996;12(6):497–506.323. Simmonds P, Zhang LQ, McOmish F, Balfe P, Ludlam CA, Brown AJ. Discontinuous sequence change of human immunodefciency virus (HIV) type 1 env sequences in plasma viral and lymphocyte-associated proviral populations in vivo: implications for models of HIV pathogenesis. Journal of virology. 1991;65(11):6266–76.324. De Jong JJ, De Ronde A, Keulen W, Tersmette M, Goudsmit J. Minimal requirements for the human immunodefciency virus type 1 V3 domain to support the syncytium-inducing phenotype: analysis by single amino acid substitution. Journal of virology. 1992;66(11):6777–80.325. McGovern RA, Thielen A, Mo T, Dong W, Woods CK, Chapman D, Lewis M, James I, Heera J, Valdez H, Harrigan PR. Population-based V3 genotypic tropism assay: a retrospective analysis using screening samples from the A4001029 and MOTIVATE studies. AIDS. 2010;24(16):2517–25.326. Soriano V, Geretti AM, Perno C-F, Fätkenheuer G, Pillay D, Reynes J, Tambussi G, Calvez V, Alcamí J, Rockstroh J. Optimal use of maraviroc in clinical practice. AIDS. 2008;22(17):2231–40.327. Esté JA, Telenti A. HIV entry inhibitors. Lancet. 2007;370(9581):81–8.328. Rose JD, Rhea AM, Weber J, Quiñones-Mateu ME. Current tests to evaluate HIV-1 coreceptor tropism. Current opinion in HIV and AIDS. 2009;4(2):136–42.329. Sierra S, Kaiser R, Thielen A. Genotypic coreceptor analysis. European journal of medical research. 2007;12:453–462.330. Fouchier RA, Brouwer M, Broersen SM, Schuitemaker H. Simple determination of human immunodefciency virus type 1 syncytium-inducing V3 genotype by PCR. Journal of clinical microbiology. 1995;33(4):906–11.202331. Chan SY, Speck RF, Power C, Gaffen SL, Chesebro B, Goldsmith MA. V3 recombinants indicate a central role for CCR5 as a coreceptor in tissue infection by human immunodefciency virus type 1. Journal of virology. 1999;73(3):2350–8.332. Bushman FD, Hoffmann C, Ronen K, Malani N, Minkah N, Rose HM, Tebas P, Wang GP. Massively parallel pyrosequencing in HIV research. AIDS. 2008;22(12):1411–5.333. Swenson LC, Moores A, Low AJ, Thielen A, Dong W, Woods C, Jensen MA, Wynhoven B, Chan D, Glascock C, Harrigan PR. Improved detection of CXCR4-using HIV by V3 genotyping: application of population-based and “deep” sequencing to plasma RNA and proviral DNA. Journal of acquired immune deficiency syndromes. 2010;54(5):506–10.334. Trinh L, Han D, Huang W, Wrin T, Larson J, Kiss L, Coakley E, Petropoulos C, Parkin N, Whitcomb J, Reeves J. Technical Validation of an Enhanced Sensitivity Trofle HIV Co-receptor Tropism Assay for Selecting Patients for Therapy with Entry Inhibitors Targeting CCR5. In: XVII HIV Drug Resistance Workshop. Sitges, Spain; 2008.335. Swenson LC, McGovern RA, Dong W, Mo T, Lee GQ, Thielen A, Jensen M, Heera J, Lewis M, James I, Chapman D, Valdez H, Harrigan PR. Optimization of Clinically Relevant Cutoffs for Determining HIV Co-Receptor Use by Population and Deep Sequencing Methods. In: Infectious Diseases Society of America 47th Annual Meeting. Philadelphia, Pennsylvania: IDSA.336. McGovern RA, Dong W, Mo T, Woods C, Zhong X, Thielen A, Jensen M, Heera J, Ellery S, Lewis M, James I, Biswas P, Chapman D, Valdez H, Harrigan R. Optimization of Clinically Relevant Cutpoints for the Determination of HIV Co-Receptor Usage to Predict Maraviroc Responses in Treatment Experienced (TE) Patients Using Population V3 Genotyping. In: European AIDS Clinical Society. Cologne, Germany; 2009:Poster PE3.4/8.337. Swenson LC, Dong W, Mo T, Woods C, Zhong X, Thielen A, Jensen M, Biswas P, Ellery S, Lewis M, James I, Chapman D, Valdez H, Harrigan R. “Deep” Sequencing to Identify Treatment-Experienced Patients Who Respond to Maraviroc (MVC). In: European AIDS Clinical Society. Cologne, Germany; 2009:Poster PE3.3/2.203338. Zweig MH, Campbell G. Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine. Clinical chemistry. 1993;39(4):561–77.339. Valdez H, Lewis M, Delogne C, Simpson P, Chapman D, Mcfadyen L, Coakley E, Ryst E Van Der, Westby M. Weighted OBT susceptibility score (wOBTSS) is a stronger predictor of virologic response at 48 weeks than baseline tropism result in MOTIVATE 1 and 2. In: 48th Annual ICAAC/IDSA 46th Meeting. Washington DC; 2008:Poster H–1221.340. Kagan RM, Johnson EP, Siaw M, Biswas P, Chapman DS, Su Z, Platt JL, Pesano RL. A Genotypic Test for HIV-1 Tropism Combining Sanger Sequencing with Ultradeep Sequencing Predicts Virologic Response in Treatment-Experienced Patients. Polis MA, ed. PLoS ONE. 2012;7(9):e46334.341. Brumme C, Wilkin T, Su Z, Schapiro J, Kagan R, Chapman D, Heera J, Valdez H, Harrigan PR. Relative performance of ESTA, Trofle, 454 deep sequencing and reflex testing for HIV tropism in the MOTIVATE screening population of therapy experienced patients. In: 18th Conference on Retroviruses and Opportunistic Infections. Boston, MA, USA: CROI; 2011:Abstract V–165.342. Nabatov AA, Pollakis G, Linnemann T, Kliphius A, Chalaby MIM, Paxton WA. Intrapatient alterations in the human immunodefciency virus type 1 gp120 V1V2 and V3 regions differentially modulate coreceptor usage, virus inhibition by CC/CXC chemokines, soluble CD4, and the b12 and 2G12 monoclonal antibodies. Journal of virology. 2004;78(1):524–30.343. Cho MW, Lee MK, Carney MC, Berson JF, Doms RW, Martin MA. Identifcation of determinants on a dualtropic human immunodefciency virus type 1 envelope glycoprotein that confer usage of CXCR4. Journal of virology. 1998;72(3):2509–15.344. Kirby Institute, ViiV Healthcare, Pfzer. Maraviroc Switch Collaborative Study (MARCH). ClinicalTrials.gov. 2012. Available at: http://clinicaltrials.gov/ct2/show/NCT01384682. Accessed December 12, 2013.204345. Gill VS, Lima VD, Zhang W, Wynhoven B, Yip B, Hogg RS, Montaner JSG, Harrigan PR. Improved virological outcomes in British Columbia concomitant with decreasing incidence of HIV type 1 drug resistance detection. Clinical infectious diseases. 2010;50(1):98–105.346. Tu E, Swenson LC, Land S, Pett S, Emery S, Marks K, Kelleher AD, Kaye S, Kaiser R, Schuelter E, Harrigan R. Results of external quality assessment for proviral DNA testing of HIV tropism in the Maraviroc Switch collaborative study. Journal of clinical microbiology. 2013;51(7):2063–71.347. Verhofstede C, Brudney D, Reynaerts J, Vaira D, Fransen K, De Bel A, Seguin-Devaux C, De Wit S, Vandekerckhove L, Geretti A-M. Concordance between HIV-1 genotypic coreceptor tropism predictions based on plasma RNA and proviral DNA. HIV medicine. 2011;12(9):544–552.348. Swenson LC, Mo T, Dong WWY, Zhong X, Woods CK, Jensen MA, Thielen A, Chapman D, Lewis M, James I, Heera J, Valdez H, Harrigan PR. Deep sequencing to infer HIV-1 co-receptor usage: application to three clinical trials of maraviroc in treatment-experienced patients. The Journal of infectious diseases. 2011;203(2):237–45.349. Swenson LC, Mo T, Dong WWY, Zhong X, Woods CK, Thielen A, Jensen M, Knapp DJHF, Chapman D, Portsmouth S, Lewis M, James I, Heera J, Valdez H, Harrigan PR. Deep V3 Sequencing for HIV Type 1 Tropism in Treatment-Naive Patients: A Reanalysis of the MERIT Trial of Maraviroc. Clinical infectious diseases. 2011;53(7):732–42.350. Lalezari J, Mitsuyasu R, Deeks S, Wang S, Lee G, Holmes M, Gregory P, Giedlin M, Tang W, Ando D. Successful and Persistent Engraftment of ZFN-M-R5-D Autologous CD4 T Cells (SB-728-T) in Aviremic HIV-infected Subjects on HAART. In: 18th Conference on Retroviruses and Opportunistic Infections. Boston, MA, USA; :Abstract 46.351. Ando D, WW T, D S. HAART Treatment Interruption following Adoptive Transfer of Zinc Finger Nuclease (ZFN) Modifed Autologous CD4 T-cells (SB-728-T) to HIV-infected Subjects Demonstrates Durable Engraftment and Suppression of Viral Load. In: 51st Interscience Conference on Antimicrobial Agents and Chemotherapy.Vol 3. Chicago, IL: ICAAC; 2011:H2–794a.205352. Hütter G, Nowak D, Mossner M, Ganepola S, Müssig A, Allers K, Schneider T, Hofmann J, Kücherer C, Blau O, Blau IW, Hofmann WK, Thiel E. Long-term control of HIV by CCR5 Delta32/Delta32 stem-cell transplantation. The New England journal of medicine. 2009;360(7):692–8.353. El-Sadr W, Lundgren JD, Neaton J, Gordin F, Abrams D, Arduino R, Babiker A, Burman W, Clumeck N, Cohen C, others. CD4+ Count-Guided Interruption of Antiretroviral Treatment. New England Journal of Medicine. 2006;355(22):2283–2296.354. Kuritzkes DR. Genotypic tests for determining coreceptor usage of HIV-1. The Journal of infectious diseases. 2011;203(2):146–8.355. Thompson MA, Aberg JA, Cahn P, Montaner JSG, Rizzardini G, Telenti A, Gatell JM, Günthard HF, Hammer SM, Hirsch MS, Jacobsen DM, Reiss P, Richman DD, Volberding PA, Yeni P, Schooley RT. Antiretroviral treatment of adult HIV infection: 2010 recommendations of the International AIDS Society-USA panel. JAMA: the journal of the American Medical Association. 2010;304(3):321–33.356. Vitiello P, Brudney D, MacCartney M, Garcia A, Smith C, Marshall N, Johnson M, Geretti AM. Responses to switching to maraviroc-based antiretroviral therapy in treated patients with suppressed plasma HIV-1-RNA load. Intervirology. 2012;55(2):172–8.357. Paar C, Geit M, Stekel H, Berg J. Genotypic prediction of human immunodefciency virus type 1 tropism by use of plasma and peripheral blood mononuclear cells in the routine clinical laboratory. Journal of clinical microbiology. 2011;49(7):2697–9.358. Parisi SG, Andreoni C, Sarmati L, Boldrin C, Buonomini AR, Andreis S, Scaggiante R, Cruciani M, Bosco O, Manfrin V, D’Ettorre G, Mengoli C, Vullo V, Palù G, Andreoni M. HIV coreceptor tropism in paired plasma, peripheral blood mononuclear cell, and cerebrospinal fluid isolates from antiretroviral-naïve subjects. Journal of clinical microbiology. 2011;49(4):1441–5.206359. Lee GQ, Dong W, Mo T, Woods C, Kanters S, Shen A. Limited Evolution of Inferred HIV-1 Tropism While Viremia Is Undetectable During Standard HAART Therapy. In: Infectious Diseases Society of America. Vancouver, BC, Canada: IDSA; 2011:Abstract 3646.360. Baldanti F, Paolucci S, Gulminetti R, Maserati R, Migliorino G, Pan A, Maggiolo F, Comolli G, Chiesa A, Gerna G. Higher levels of HIV DNA in memory and naive CD4(+) T cell subsets of viremic compared to non-viremic patients after 18 and 24 months of HAART. Antiviral research. 2001;50(3):197–206.361. Jabara CB, Jones CD, Roach J, Anderson J a, Swanstrom R. Accurate sampling and deep sequencing of the HIV-1 protease gene using a Primer ID. Proceedings of the National Academy of Sciences of the United States of America. 2011;108(50):20166–71.362. Andreoni M, Parisi SG, Sarmati L, Nicastri E, Ercoli L, Mancino G, Sotgiu G, Mannazzu M, Trevenzoli M, Tridente G, Concia E, Aceti A. Cellular proviral HIV-DNA decline and viral isolation in naïve subjects with <5000 copies/ml of HIV-RNA and >500 x 10(6)/l CD4 cells treated with highly active antiretroviral therapy. AIDS. 2000;14(1):23–9.363. Sarmati L, Parisi SG, Nicastri E, D’Ettorre G, Palmisano L, Andreotti M, Andreoni C, Giuliano M, Gatti F, Boldrin C, Palù G, Vullo V, Vella S, Andreoni M. Association between cellular human immunodefciency virus DNA level and immunological parameters in patients with undetectable plasma viremia level during highly active antiretroviral therapy. Journal of clinical microbiology. 2005;43(12):6183–5.364. Schapiro JM, Boucher CA, Kuritzkes DR, van de Vijver DA, Llibre JM, Lewis M, Simpson P, Delogne C, McFadyen L, Chapman D, Perros M, Valdez H, Ryst E van der, Westby M. Baseline CD4+ T-cell counts and weighted background susceptibility scores strongly predict response to maraviroc regimens in treatment-experienced patients. Antiviral therapy. 2011;16:395–404.365. Wilkin TJ, Goetz MB, Leduc R, Skowron G, Su Z, Chan ES, Heera J, Chapman D, Spritzler J, Reeves JD, Gulick RM, Coakley E. Reanalysis of coreceptor tropism in HIV-1-infected adults using a phenotypic assay with enhanced sensitivity. Clinical infectious diseases. 2011;52(7):925–8.207366. McGovern RA, Thielen A, Portsmouth S, Mo T, Dong W, Woods CK, Zhong X, Brumme CJ, Chapman D, Lewis M, James I, Heera J, Valdez H, Harrigan PR. Population-based sequencing of the V3-loop can predict the virological response to maraviroc in treatment-naive patients of the MERIT trial. Journal of acquired immune deficiency syndromes. 2012;61(3):279–86.367. Raymond S, Saliou A, Nicot F, Delobel P, Dubois M, Cazabat M, Sandres-Sauné K, Marchou B, Massip P, Izopet J. Frequency of CXCR4-using viruses in primary HIV-1 infections using ultra-deep pyrosequencing. AIDS. 2011;25(13):1668–1673.368. Tilton JC, Wilen CB, Didigu CA, Sinha R, Harrison JE, Agrawal-Gamse C, Henning EA, Bushman FD, Martin JN, Deeks SG, Doms RW. A maraviroc-resistant HIV-1 with narrow cross-resistance to other CCR5 antagonists depends on both N-terminal and extracellular loop domains of drug-bound CCR5. Journal of virology. 2010;84(20):10863–76.369. Roche M, Salimi H, Duncan R, Wilkinson BL, Chikere K, Moore MS, Webb NE, Zappi H, Sterjovski J, Flynn JK, Ellett A, Gray LR, Lee B, Jubb B, Westby M, Ramsland PA, Lewin SR, Payne RJ, Churchill MJ, et al. A common mechanism of clinical HIV-1 resistance to the CCR5 antagonist maraviroc despite divergent resistance levels and lack of common gp120 resistance mutations. Retrovirology. 2013;10:43.370. Lalezari J, Goodrich J, DeJesus E, Lampiris H, Gulick R, Saag M, Ridgway C, McHale M, Ryst E van der, Mayer H. Effcacy and Safety of Maraviroc plus Optimized Background Therapy In Viremic, ART- Experienced Patients Infected With CCR5-Tropic HIV-1: 24-Week Results of Phase 2b/3 Studies. In: Program and abstracts of the 14th Conference on Retroviruses and Opportunistic Infections, Los Angeles.; :Abstract 104bLB.371. Obermeier M, Symons J, Wensing AMJ. HIV population genotypic tropism testing and its clinical signifcance. Current opinion in HIV and AIDS. 2012;7(5):470–477.372. Demarest JF, Amrine-Madsen H, Irlbeck DM, Kitrinos KM. Virologic failure in frst-line human immunodefciency virus therapy with a CCR5 entry inhibitor, aplaviroc, plus a fxed-dose combination of lamivudine-zidovudine: nucleoside reverse transcriptase inhibitor 208resistance regardless of envelope tropism. Antimicrobial agents and chemotherapy. 2009;53(3):1116–23.373. Westby M, Smith-Burchnell C, Mori J, Lewis M, Mosley M, Stockdale M, Dorr P, Ciaramella G, Perros M. Reduced maximal inhibition in phenotypic susceptibility assays indicates that viral strains resistant to the CCR5 antagonist maraviroc utilize inhibitor-bound receptor for entry. Journal of virology. 2007;81(5):2359–71.374. Moore JP, Kuritzkes DR. A pièce de resistance: how HIV-1 escapes small molecule CCR5 inhibitors. Current Opinion in HIV and AIDS. 2009;4(2):118–124.375. Delobel P, Cazabat M, Saliou A, Loiseau C, Coassin L, Raymond S, Requena M, Marchou B, Massip P, Izopet J. Primary resistance of CCR5-tropic HIV-1 to maraviroc cannot be predicted by the V3 sequence. The Journal of antimicrobial chemotherapy. 2013;68(11):2506–14.376. McNicholas P, Vilchez RA, Greaves W, Kumar S, Onyebuchi C, Black T, Strizki JM. Detection of HIV-1 CXCR4 tropism and resistance in treatment experienced subjects receiving CCR5 antagonist-Vicriviroc. Journal of clinical virology. 2012;55(2):134–9.377. Coakley E, Reeves JD, Huang W, Mangas-Ruiz M, Maurer I, Harskamp AM, Gupta S, Lie Y, Petropoulos CJ, Schuitemaker H, van ’t Wout AB. Comparison of human immunodefciency virus type 1 tropism profles in clinical samples by the Trofle and MT-2 assays. Antimicrobial agents and chemotherapy. 2009;53(11):4686–93.378. Walter H, Wolf E, Braun P, Daumer M, Thielen A, Noah C, Sichtig N, Muller H, Stumer M, Lengauer T, Kaiser R, Obermeier M. Performance of genotypic coreceptor measurement using geno2pheno[coreceptor] in B- and non-B HIV subtypes in a large cohort of therapy-experienced patients in Germany. In: 7th European HIV Drug Resistance Workshop. Stockholm, Sweden; 2009.209379. Panel on Antiretroviral Guidelines for Adults and Adolescents. Guidelines for the use of antiretroviral agents in HIV-infected adults and adolescents. Department of Health and Human Services. 2013;5(2):1–267.380. Olasagasti F, Lieberman KR, Benner S, Cherf GM, Dahl JM, Deamer DW, Akeson M. Replication of individual DNA molecules under electronic control using a protein nanopore. Nature Nanotechnology. 2010;5(11):798–806.381. Bartolini B, Giombini E, Zaccaro P, Selleri M, Rozera G, Abbate I, Comandini UV, Ippolito G, Solmone M, Capobianchi MR. Extent of HCV NS3 protease variability and resistance-associated mutations assessed by next generation sequencing in HCV monoinfected and HIV/HCV coinfected patients. Virus Research. 2013;177(2):205–208.382. Cohen MS, Chen YQ, McCauley M, Gamble T, Hosseinipour MC, Kumarasamy N, Hakim JG, Kumwenda J, Grinsztejn B, Pilotto JHS, Godbole S V, Mehendale S, Chariyalertsak S, Santos BR, Mayer KH, Hoffman IF, Eshleman SH, Piwowar-Manning E, Wang L, et al. Prevention of HIV-1 infection with early antiretroviral therapy. The New England journal of medicine. 2011;365(6):493–505.383. Zhu J, Ofek G, Yang Y, Zhang B, Louder MK, Lu G, Mckee K, Pancera M, Skinner J, Zhang Z, Parks R, Eudailey J, Lloyd KE, Blinn J, Alam SM, Haynes BF, Simek M, Burton DR, Koff WC, et al. Mining the antibodyome for HIV-1-neutralizing antibodies with next-generation sequencing and phylogenetic pairing of heavy/light chains. Proceedings of the National Academy of Sciences of the United States of America. 2013;110(16):6470–6475.384. Svarovskaia ES, Martin R, McHutchison JG, Miller MD, Mo H. Abundant drug-resistant NS3 mutants detected by deep sequencing in hepatitis C virus-infected patients undergoing NS3 protease inhibitor monotherapy. Journal of clinical microbiology. 2012;50(10):3267–74.385. Swenson LC, Lee GQ, Mo T, Woods C, Harrigan PR. “Deep” Sequencing to Quantify in vivo Viral Fitness after Transmission of Resistant HIV. In: 49th Interscience Conference on Antimicrobial Agents and Chemotherapy. San Francisco, CA; 2009.210386. Swenson LC, Hudson E, Mo T, Woods C, Archer J, Robertson D, Montaner J, Harrigan PR. Measurement of Low-level Drug-resistance Mutations Using Deep Sequencing Technology upon Restarting HAART. In: Conference on Retroviruses and Opportunistic Infections. Montreal, QC; 2009.387. Poon AF, Swenson LC, Dong WW, Deng W, Kosakovsky Pond SL, Brumme ZL, Mullins JI, Richman DD, Harrigan PR, Frost SDW. Phylogenetic analysis of population-based and deep sequencing data to identify coevolving sites in the nef gene of HIV-1. Molecular biology and evolution. 2010;27(4):819–832.388. Hessol N, Napolitano L, Smith D, Lie Y, Levine A, Young M, Cohen M, Minkoff H, Anastos K, D’Souza G, Greenblatt RM, Goedert JJ. HIV tropism and decreased risk of breast cancer. PloS one. 2010;5(12):e14349.389. Velasco-Velazquez M, Jiao X, De La Fuente M, Pestell TG, Ertel A, Lisanti MP, Pestell RG. CCR5 antagonist blocks metastasis of basal breast cancer cells. Cancer research. 2012.390. Kagan RM, Johnson EP, Siaw MF, Van Baelen B, Ogden R, Platt JL, Pesano RL, Lefebvre E. Comparison of genotypic and phenotypic HIV type 1 tropism assay: results from the screening samples of Cenicriviroc Study 202, a randomized phase II trial in treatment-naive subjects. AIDS research and human retroviruses. 2014;30(2):151–9.391. Cohen MS, Muessig KE, Smith MK, Powers KA, Kashuba ADM. Antiviral agents and HIV prevention: controversies, conflicts, and consensus. AIDS. 2012;26(13):1585–98.392. Méchai F, Quertainmont Y, Sahali S, Delfraissy J-F, Ghosn J. Post-exposure prophylaxis with a maraviroc-containing regimen after occupational exposure to a multi-resistant HIV-infected source person. Journal of medical virology. 2008;80(1):9–10.393. Neff CP, Kurisu T, Ndolo T, Fox K, Akkina R. A topical microbicide gel formulation of CCR5 antagonist maraviroc prevents HIV-1 vaginal transmission in humanized RAG-hu mice. PloS one. 2011;6(6):e20209.211394. Malcolm RK, Veazey RS, Geer L, Lowry D, Fetherston SM, Murphy DJ, Boyd P, Major I, Shattock RJ, Klasse PJ, Doyle LA, Rasmussen KK, Goldman L, Ketas TJ, Moore JP. Sustained release of the CCR5 inhibitors CMPD167 and maraviroc from vaginal rings in rhesus macaques. Antimicrobial agents and chemotherapy. 2012;56(5):2251–8.395. Fetherston SM, Boyd P, McCoy CF, McBride MC, Edwards K-L, Ampofo S, Malcolm RK. A silicone elastomer vaginal ring for HIV prevention containing two microbicides with different mechanisms of action. European journal of pharmaceutical sciences. 2012;48(3):406–415.396. Allers K, Hütter G, Hofmann J, Loddenkemper C, Rieger K, Thiel E, Schneider T. Evidence for the cure of HIV infection by CCR5Δ32/Δ32 stem cell transplantation. Blood. 2011;117(10):2791–9.397. Check Hayden E. Hopes of HIV cure in “Boston patients” dashed. Nature. 2013;785:6–8.398. Wilen CB, Wang J, Tilton JC, Miller JC, Kim KA, Rebar EJ, Sherrill-Mix SA, Patro SC, Secreto AJ, Jordan APO, Lee G, Kahn J, Aye PP, Bunnell BA, Lackner AA, Hoxie JA, Danet-Desnoyers GA, Bushman FD, Riley JL, et al. Engineering HIV-resistant human CD4+ T cells with CXCR4-specifc zinc-fnger nucleases. PLoS pathogens. 2011;7(4):e1002020.399. Voit RA, McMahon MA, Sawyer SL, Porteus MH. Generation of an HIV resistant T-cell line by targeted “stacking” of restriction factors. Molecular therapy. 2013;21(4):786–95.400. Nazari R, Joshi S. CCR5 as target for HIV-1 gene therapy. Current gene therapy. 2008;8(4):264–72. 212AppendicesAppendix I:     Primers for Chapter 2Population-based sequencing primers:Standard, population-based sequencing from HIV RNAFirst-round PCR primer, forward: 5' GAGCCAATTCCCATACATTATTGT 3’First-round PCR primer, reverse: 5' TAAGTCTCTCAAGCGGTGGTAGCTGAA 3'Second-round PCR primer, forward: 5' TGTGCCCCAGCTGGTTTTGCGAT 3'Second-round PCR primer, reverse: 5' GGATCTGTCTCTGTCTCTCTCTCCA 3'Standard, population-based sequencing from HIV DNAFirst-round PCR primer, forward: 5' GAGCCAATTCCCATACATTATTGT 3’First-round PCR primer, reverse: 5' TGTGCCCCAGCTGGTTTTGCGAT 3'Second-round PCR primer, forward: 5' AGCACAGTACAATGTACACATGG 3’Second-round PCR primer, reverse: 5' GAAAAATTCCCTTCCACAATTAAA 3’Population-based sequencing primersSequencing primer, forward: 5' AATGTCAGYACAGTACAATGTACAC 3’Sequencing primer, reverse: 5' GAAAAATTCCCTTCCACAATTAAA 3’Deep sequencing primers:Deep sequencing from HIV RNASecond-round PCR primer, forward: 5’AATGCCAAAACCATAATAGTACA 3’Second-round PCR primer, reverse: 5'GAAAAATTCCCTTCCACAATTAAA 3’213Deep sequencing from HIV DNASecond-round PCR primer, forward: 5’ AATGCCAAAACCATAATAGTACA 3’Second-round PCR primer, reverse: 5' GAAAAATTCCCTTCCACAATTAAA 3’Deep sequencing fusion primer5' GCCTCCCTCGCGCCATCAG 3’Deep sequencing barcode tags:(A) ACGAGTGCGT; (B) ACGCTCGACA; (C) AGACGCACTC; (D) AGCACTGTAG; (E) ATCAGACACG; (F) CGTGTCTCTA; (G) CTCGCGTGTC; (H) TAGTATCAGC; (I) TCTCTATGCG; (J) TGATACGTCT; (K) TACTGAGCTA; (L) ATATCGCGAG.Example deep sequencing complete forward primer, tag A:(fusion, then barcode A, then PCR primer) 5'GCCTCCCTCGCGCCATCAGACGAGTGCGTAATGCCAAAACCATAATAGTACA3’214Appendix II:     Thermal Cycler Protocols for Chapter 2RT-PCR:30’@52°C; 2’@94°C; 40 cycles of (15”@94°C, 30”@55°C, 1’30”@68°C); 5’@68°C2nd round PCR for standard sequencing:2’@94°C; 35 cycles of (15”@94°C, 30”@55°C, 1’@72°C); 7’@72°C2nd round PCR for deep sequencing:2’@94°C; 35 cycles of (15”@94°C, 30”@55°C, 50”@72°C); 5’@72°C215Appendix III:     Primers for Chapter 3Primers for deep sequencingFirst-round PCR primer, forward: 5' GAGCCAATTCCCATACATTATTGT 3’First-round PCR primer, reverse: 5’ GCCCATAGTGCTTCCTGCTGCTCCCAAG AACC 3’Second-round PCR primer, forward: 5’ AATGCCAAAACCATAATAGTACA 3’Second-round PCR primer, reverse: 5' GAAAAATTCCCTTCCACAATTAAA 3’Fusion primer5' GCCTCCCTCGCGCCATCAG 3’Deep sequencing barcode tags:(A) ACGAGTGCGT; (B) ACGCTCGACA; (C) AGACGCACTC; (D) AGCACTGTAG; (E) ATCAGACACG; (F) CGTGTCTCTA; (G) CTCGCGTGTC; (H) TAGTATCAGC; (I) TCTCTATGCG; (J) TGATACGTCT; (K) TACTGAGCTA; (L) ATATCGCGAG.216Example of complete deep sequencing forward primer, tag A:(fusion primer, then barcode A, then PCR primer) 5'GCCTCCCTCGCGCCATCAGACGAGTGCGTAATGCCAAAACCATAATAGTACA3’Primers for population-based genotypingFirst-round PCR primer, forward: 5'  GAGCCAATTCCCATACATTATTGT 3’First-round PCR primer, reverse: 5’ GCCCATAGTGCTTCCTGCTGCTCCCAAG AACC 3’Second-round PCR primer, forward: 5' TGTGCCCCAGCTGGTTTTGCGAT 3’Second-round PCR primer, reverse: 5' TATAATTCACTTCTCCAATTGTCC 3’Sequencing primer, forward: 5’ AATGTCAGYACAGTACAATGTACAC 3’  Sequencing primer, reverse: 5' GAAAAATTCCCTTCCACAATTAAA 3’217Appendix IV:     Description of Deep Sequencing Data Processing PipelineThe output from the Genome Sequencer FLX platform was processed using a custom pipeline ofRuby and Python scripts. Identical nucleotide reads were merged into "variants" and the number ofreads per variant was recorded.   Variants were sorted by multiplexing tag and primer, tolerating amaximum of three nucleotide mismatches from the known primer sequence.   No differences weretolerated in the tag sequences.  A sample-specifc consensus sequence was generated from the three most abundant variants with agiven tag and primer combination.  Subsequently, all variants in the tag and primer-defned set werealigned pairwise against this consensus sequence, trimmed to the env V3 region, and screened forinsertions and deletions that induced shifts in the reading frame.  Geno2pheno scores were calculated from the V3 sequences by aligning the amino acid translationagainst  the  geno2pheno  reference  sequence  and  scoring  the  residues  at  each  reference  positionaccording to the geno2pheno support vector machine classifer.   Any V3 protein sequences that (1)contained a stop codon;  (2)  contained residues other  than cysteine on the 5'  or 3'  termini;  or  (3)comprised fewer than 33 or more than 40 residues did not receive a score. Samples with fewer than750 reads were discarded and resequenced.218Appendix V:     Phylogenetic Tree from a Maraviroc Recipient with a Small Pre-Treatment X4 Population Related to the Failure Sequence.  Appendix V: Phylogenetic Tree from a Maraviroc Recipient with a Small Pre-Treatment X4 Population Related to the Failure Sequence Screening sample: 0.1% X4 by deep sequencing, and R5 by ESTAFailure sample: X4 by population-based genotype and Dual/Mixed by the original Trofle assay.X4 sequences are shown in red, R5 sequences are shown in green, and the failure sequence is shown in blue.219Appendix VI:     Phylogenetic Tree from a Maraviroc Recipient with a Small Pre-Treatment X4 Population Related to the Failure SequenceAppendix VI: Phylogenetic Tree from a Maraviroc Recipient with a Small Pre-Treatment X4 Population Related to the Failure Sequence  Screening sample: 2% X4 by deep sequencing, and R5 by ESTAFailure sample: X4 by population-based genotype, and X4 by original Trofle assay. R5 sequences shown in green, X4 sequences shown in red, failure sequence shown in blue.220Appendix VII:     Phylogenetic Tree from a Maraviroc Recipient with a Small Pre-Treatment X4 Population Related to the Failure SequenceAppendix VII: Phylogenetic Tree from a Maraviroc Recipient with a Small Pre-Treatment X4 Population Related to the Failure Sequence Screening sample: 5% X4 by deep sequencing, and Dual/Mixed by ESTAFailure sample: X4 by population-based genotype, and X4 by original Trofle assayR5 sequences shown in green, X4 sequences shown in red, failure sequence shown in blue.221Appendix VIII:     Phylogenetic Tree from a Maraviroc Recipient with a Large Pre-Treatment X4 Population Related to the Failure SequenceAppendix VIII: Phylogenetic Tree from a Maraviroc Recipient with a Large Pre-Treatment X4 Population Related to the Failure Sequence Screening sample: 56% X4 by deep sequencing, and R5 by ESTAFailure sample: X4 by population-based genotype, and Dual/Mixed by original Trofle assay.X4 sequences are shown in red, R5 sequences are shown in green, and the failure sequence is shown in blue.222Appendix IX:     Phylogenetic Tree from a Maraviroc Recipient with a Large Pre-Treatment X4 Population Related to the Failure SequenceAppendix IX: Phylogenetic Tree from a Maraviroc Recipient with a Large Pre-Treatment X4 Population Related to the Failure Sequence Screening sample: 92% X4 by deep sequencing, and R5 by ESTAFailure sample: X4 by population-based genotype, and Dual/Mixed by original Trofle assay.R5 sequences shown in green, X4 sequences shown in red, failure sequence shown in blue. 223Appendix X:     Phylogenetic Tree from a Maraviroc Recipient with a Large Pre-Treatment X4 Population Related to the Failure SequenceAppendix X: Phylogenetic Tree from a Maraviroc Recipient with a Large Pre-Treatment X4 Population Related to the Failure Sequence Screening sample: 97% X4 by deep sequencing and R5 by ESTAFailure sample: X4 by population-based genotype, and Dual/Mixed by original Trofle assay.X4 sequences are shown in red, R5 sequences are shown in green, and the failure sequence is shown in blue.224Appendix XI:     Phylogenetic Tree from a Maraviroc Recipient with a Large Pre-Treatment X4 Population Related to the Failure SequenceAppendix XI: Phylogenetic Tree from a Maraviroc Recipient with a Large Pre-Treatment X4 Population Related to the Failure Sequence Screening sample: 99.9% X4 by deep sequencing, and Dual/Mixed by ESTAFailure sample: X4 by population-based genotype, and Dual/Mixed by original Trofle assayR5 sequences shown in green, X4 sequences shown in red, and failure sequence shown in blue.225Appendix XII:     Phylogenetic Tree from a Maraviroc Recipient for Whom Deep Sequencing Failed to Detect a Pre-Treatment X4 Population Despite Failure with X4 HIVAppendix XII: Phylogenetic Tree from a Maraviroc Recipient for Whom Deep Sequencing Failed to Detect a Pre-Treatment X4 Population Despite Failure with X4 HIV Screening sample: 0% X4 by deep sequencing, and R5 by ESTAFailure sample: X4 by population-based genotype, and Dual/Mixed by original Trofle assayR5 sequences shown in green, failure sequence shown in blue.226Appendix XIII:     Phylogenetic Tree from a Maraviroc Recipient Who Failed with R5 HIVAppendix XIII:     Phylogenetic Tree from a Maraviroc Recipient Who Failed with R5 HIV Screening sample: 0% X4 by deep sequencing, and Dual/Mixed by ESTAFailure sample: R5 by population-based genotype, and R5 by original Trofle assayR5 sequences shown in green, failure sequence shown in blue. 227Appendix XIV:     Phylogenetic Tree from a Maraviroc Recipient Who Failed with R5 HIVAppendix XIV: Phylogenetic Tree from a Maraviroc Recipient Who Failed with R5 HIVScreening sample: 0% X4 by deep sequencing, and R5 by ESTAFailure sample: R5 by population-based genotype, and R5 by original Trofle assay.R5 sequences shown in green, failure sequence shown in blue. 228

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0167457/manifest

Comment

Related Items