UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

HIV genomics : trends in antiretroviral resistance and future directions for pharmacogenetic testing Rocheleau, Genevieve 2017

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

Item Metadata

Download

Media
24-ubc_2018_february_rocheleau_genevieve.pdf [ 4.61MB ]
Metadata
JSON: 24-1.0357604.json
JSON-LD: 24-1.0357604-ld.json
RDF/XML (Pretty): 24-1.0357604-rdf.xml
RDF/JSON: 24-1.0357604-rdf.json
Turtle: 24-1.0357604-turtle.txt
N-Triples: 24-1.0357604-rdf-ntriples.txt
Original Record: 24-1.0357604-source.json
Full Text
24-1.0357604-fulltext.txt
Citation
24-1.0357604.ris

Full Text

iHIV GENOMICS: TRENDS IN ANTIRETROVIRAL RESISTANCE AND FUTURE DIRECTIONS FOR PHARMACOGENETIC TESTING by Genevieve RocheleauB.Sc., University of British Columbia, 2014A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCEinTHE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES(Experimental Medicine)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)November 2017© Genevieve Rocheleau, 2017iiAbstract Antiretroviral drugs are fundamental to the treatment of Human Immunodeficiency Virus (HIV), effectively inhibiting viral replication, the emergence of Acquired Immune Deficiency Syndrome (AIDS), and subsequent mortality. Combined antiretroviral therapy (cART) is the cornerstone of HIV treatment, and has been adopted by agencies around the world. An estimated 18.2 million individuals accessed cART worldwide in 2016. However, HIV drug resistance to cART leads to ineffective HIV treatment, is associated with AIDS-related morbidity and mortality, and the potential onward transmission of drug resistant HIV strains. Since cART became available in BC in the mid-1990s, antiretroviral drugs and clinical guidelines for HIV management have evolved to reflect best practices. Over the past two decades, genetic testing for HIV drug resistance has become an important tool for HIV care. Next generation massively parallel sequencing has proven to be a powerful sequencing tool rivaling the gold standard Sanger sequencing method, however it is not yet widely adopted for HIV-related genetic testing in BC. There are three primary objectives discussed in this thesis: 1) identification of long-term trends in transmitted and acquired HIV drug resistance in BC, Canada; 2) determination of sociodemographic covariates of drug resistance development and testing uptake; and 3) validation and application of an HIV-related next generation sequencing (NGS) assay for abacavir hypersensitivity screening.Prevalence data of acquired and transmitted drug resistance over the past two decades are presented. Acquired resistance was examined in further detail in order to assess the effect of iiitherapy duration on drug resistance, as well as temporal effects and other factors. Covariates of acquired drug resistance were also examined over calendar time, with a particular focus on adherence to treatment regimens, including sociodemographic predictive factors, as well as sociodemographic covariates of resistance testing uptake. After characterizing historical trends of drug resistance, a glimpse at the future of HIV-related genetic testing is presented: an NGS assay for abacavir hypersensitivity screening was validated and applied as a proof of principle on the Illumina MiSeq platform. This assay was shown to be highly accurate and reliable, providing higher resolution sequencing compared to currently used methods, and expediting testing. ivLay SummaryCombinations of antiretroviral drugs (cART) are an effective treatment for Human immunodeficiency virus (HIV) infection: preventing viral replication, acquired immunodeficiency syndrome (AIDS), and HIV transmission. However, HIV mutates rapidly and is able to escape the effects of antiretroviral drugs. Current practice is to sequence the viral genome, and prescribe a personalized cART regimen to which the virus remains susceptible. New technologies, such as next generation sequencing (NGS), are more powerful and have the ability to generate more precise genetic sequences. HIV care could therefore be improved by NGS.Three primary objectives are discussed in this thesis; 1) to identify long-term trends of HIV drug resistance in BC, Canada; 2) to determine sociodemographic factors that may be associated with drug resistance and access to drug resistance testing; and 3) to validate and apply a HIV-related next generation sequencing (NGS) assay. vPrefaceThe main body of this thesis features two peer-reviewed articles published in the scientific literature, and a clinical validation of a next generation sequencing pharmacogenetic screening assay. The first published work was reprinted here with permission from the copyright holders at Clinical Microbiology and Infection. The candidate was the primary author, responsible for the writing, and collaborated with the co-authors to determine the study design. The second published work has been published by PLOS ONE, and was reprinted here under the terms of the Creative Commons Attribution License. The candidate was the primary author, responsible for the writing, and collaborated with co-authors to determine the study design. In addition, the candidate was principally responsible for writing, method modifications, laboratory work leading to data generation, and data analysis for the clinical validation of the screening assay.Co-authors within the candidate’s research laboratory include research scientist Dr. Chanson Brumme, and the candidate’s graduate supervisor Dr. Richard Harrigan. Senior laboratory assistant Winnie Dong was the principal designer of the laboratory method used in Chapter Four. A version of Chapter Two has been previously published as:Rocheleau G, Brumme CJ, Shoveller J, Lima VD, Harrigan PR. Longitudinal trends of HIV drug resistance in a large Canadian cohort; 1996-2016. Clinical Microbiology and Infection, (2017): in press, corrected proof available online (doi: 10.1016/j.cmi.2017.06.014).viThe right to reproduce this article was granted by Elsevier, license number 4190330274154. GR, CJB, VDL and PRH conceived and designed the study, JS managed data collection, CJB and VDL completed data analysis, GR, CJB, VDL, PRH edited, GR was the primary author and wrote the manuscript. A version of Chapter Three has been previously published by PLOS ONE:Rocheleau G, Franco-Villalobos C, Oliveira N, Brumme ZL, Rusch M, Shoveller J, Brumme CJ, Harrigan PR. Sociodemographic correlates of HIV drug resistance and access to drug resistance testing in British Columbia, Canada. PLOS ONE 12(9): e0184848 (2017). The right to reproduce this article was granted by the Creative Commons Attribution License. MR, CJB, ZLB and PRH conceived and designed the study, JS and NO managed data collection, CFV, NO, GR, and CJB completed data analysis, GR, ZLB, MR, CJB, and PRH edited, GR was the primary author and wrote the manuscript.Ethical approval for use of HOMER cohort research was granted by the University of British Columbia Research Ethics Board at St. Paul’s, Providence Health Care site (H05-50123). Ethical approval for use of residual clinical and research specimens for assay development, as in Chapter Four, was also granted by the University of British Columbia Research Ethics Board at St. Paul’s, Providence Health Care site (H04-50276).viiTable of ContentsAbstract ..........................................................................................................................................iiLay Summary................................................................................................................................ivPreface.............................................................................................................................................vTable of Contents.........................................................................................................................viiList of Tables...............................................................................................................................xiiiList of Figures .............................................................................................................................xivList of Abbreviations ...................................................................................................................xvAcknowledgements ....................................................................................................................xviiDedication..................................................................................................................................xviiiChapter One: Introduction ...........................................................................................................11.1 History of HIV...........................................................................................................................11.1.1 Discovery....................................................................................................................11.1.2 The BC Epidemic .......................................................................................................21.1.3 Virology......................................................................................................................41.2 Emergence of HIV Drug Resistance........................................................................................61.3 Antiretroviral Drugs; An Overview........................................................................................81.3.1 Nucleoside and Nucleotide Reverse Transcriptase Inhibitors ....................................81.3.2 Non-Nucleoside Reverse Transcriptase Inhibitors ...................................................101.3.3 Protease Inhibitors ....................................................................................................101.3.4 Entry, Fusion and Integrase Inhibitors......................................................................111.3.5 Acquired vs Transmitted Drug Resistance ...............................................................12viii1.4 HIV Genotypic Testing...........................................................................................................131.5 Relevant Sequencing Technologies........................................................................................141.5.1 Sanger Sequencing....................................................................................................141.5.2 Illumina MiSeq Sequencing .....................................................................................151.6 Research Objectives and Thesis Organization.....................................................................171.6.1 Chapter Descriptions ................................................................................................18Chapter Two: Longitudinal Trends of HIV Drug Resistance in a Large British Columbia Cohort, 1996-2016........................................................................................................................212.1 Background and Introduction ...............................................................................................212.2 Methods....................................................................................................................................222.2.1 Data Collection .........................................................................................................222.2.2 Plasma Viral Load Testing .......................................................................................232.2.3 Drug Resistance Genotyping ....................................................................................232.2.4 Estimation of Adherence ..........................................................................................242.2.5 Statistical Analysis: Percentage of Participants with Transmitted or Acquired Drug Resistance ..........................................................................................................................242.2.6 Statistical Analysis: Kaplan-Meier Analysis ............................................................252.2.7 Statistical analysis: Multivariable Logistic Regression Model for Acquired Drug Resistance ..........................................................................................................................262.3 Results ......................................................................................................................................272.3.1 Cohort Profile ...........................................................................................................272.3.2 Percentage of Participants with Transmitted or Acquired Drug Resistance.............292.3.3 Prevalence of Acquired Drug Resistance by Time Interval of cART Initiation.......302.3.4 Prevalence of Acquired Drug Resistance After Two Decades on Therapy..............32ix2.3.5 Multivariable Model: Variables Associated with Acquired Drug Resistance Mutations ...........................................................................................................................322.4 Discussion ................................................................................................................................34Chapter Three: Sociodemographic Correlates of HIV Drug Resistance and Access to Drug Resistance Testing in British Columbia, Canada .....................................................................383.1 Background and Introduction ...............................................................................................383.2 Methods: Accessing Drug Resistance Testing Cohort.........................................................393.2.1 Data Collection .........................................................................................................393.2.2 Drug Resistance Testing ...........................................................................................393.2.3 Linking of Census Data and Clinical Data ...............................................................403.2.4 Plasma Viral Load Testing .......................................................................................403.2.5 Quantification of Adherence to Prescribed Drug Regimen......................................403.2.6 Statistical Methods....................................................................................................413.3 Methods: Development of Drug Resistance Cohort ............................................................423.3.1 Data Collection .........................................................................................................423.3.2 Clinical Data and Census Linkages ..........................................................................433.3.3 Drug Resistance Testing ...........................................................................................433.3.4 Statistical Methods....................................................................................................443.4 Results: Access Drug Resistance Testing Cohort.................................................................443.4.1 Cohort Description....................................................................................................443.4.2 Odds Ratios of Correlates for Accessing Drug Resistance Testing .........................473.5 Results: Development of Drug Resistance Cohort...............................................................493.5.1 Cohort Description....................................................................................................493.5.2 Hazard Ratios of Correlates for Development of Drug Resistance..........................503.6 Discussion ................................................................................................................................52xChapter Four: The Future of HIV-Related Genetic Testing; a Next Generation Sequencing Assay Validation and Proof of Principle....................................................................................574.1 Background and Introduction ...............................................................................................574.2 Methods....................................................................................................................................584.2.1 Data Collection .........................................................................................................584.2.2 Sequencing................................................................................................................594.2.3 Data Analysis............................................................................................................624.2.4 Accuracy ...................................................................................................................634.2.5 Repeatability .............................................................................................................644.2.6 Reproducibility .........................................................................................................644.2.7 Cross-Contamination ................................................................................................654.2.8 Application of HLA Assay: A Proof of Principle Study ..........................................654.3 Results ......................................................................................................................................684.3.1 Accuracy ...................................................................................................................684.3.2 Repeatability .............................................................................................................704.3.3 Reproducibility .........................................................................................................714.3.4 Cross-Contamination ................................................................................................734.3.5 Proof of Principle......................................................................................................744.4 Discussion ................................................................................................................................76Chapter Five: General Discussion and Conclusion ..................................................................805.1 Thesis Summary......................................................................................................................805.1.1 Longitudinal Trends of HIV Drug Resistance..........................................................805.1.2 Sociodemographic Correlates of Access to Resistance Testing and Development of HIV Drug Resistance.........................................................................................................815.1.3 HIV-Related NGS Assay; Abacavir Hypersensitivity Prescreening ........................82xi5.2 Overall Impact, Applications, and Future Directions .........................................................825.3 Limitations...............................................................................................................................85Bibliography .................................................................................................................................86Appendix.....................................................................................................................................100Appendix I. Adjusted Odds Ratios, Stratified by Time Interval............................................100Appendix II. Adjusted Odds Ratios, Stratified by Level of Adherence to Drug Regimen ..101Appendix III. Top Ten Baseline Regimen, by Interval of cART Initiation...........................102Appendix IV. Comparison of Baseline Characteristics Among Those Included, and Excluded Based on Exclusion Criteria......................................................................................104Appendix V. Sensitivity Analysis Comparing Inclusion and Exclusion of those with Unknown Transmitted Drug Resistance...................................................................................105Appendix VI. Percentage of Participants with Transmitted Drug Resistance Mutations...108Appendix VII. Percentage of Transmitted HIV Drug Resistance Among Participants Initiating cART in a Given Year ...............................................................................................109Appendix VIII. Proportions in Each Adherence Strata, by Time Interval of cART Initiation......................................................................................................................................................110Appendix IX. Detection of Drug Resistance After cART Initiation.......................................111Appendix X. The Percentage of Individuals with Resistance to 0, 1, 2, and 3 or More Drug Categories, by Time Interval of cART Initiation.....................................................................112Appendix XI. Proportion of Individuals on First Line Therapy with Specified Third Drug......................................................................................................................................................113Appendix XII. Percentage of Participants Without Baseline Drug Resistance Testing by Year of cART Initiation .............................................................................................................114Appendix XIII. Percentage of Patients Accessing Drug Resistance Testing Per Year in BC......................................................................................................................................................115xiiAppendix XIV. Hazard Ratios of Developing 3TC/FTC Resistance......................................116Appendix XV. Hazard Ratios of Developing NNRTI Resistance...........................................117Appendix XVI. Hazard Ratios of Developing “Other” nRTI Resistance (Excluding 3TC/FTC) ....................................................................................................................................118Appendix XVII. Hazard Ratios of Developing PI Resistance.................................................119Appendix XVIII. Adjusted Odds Ratios of Access to Drug Resistance Testing, Stratified by Adherence ....................................................................................................................................120Appendix XIX. Adjusted Hazard Ratios of Development of Drug Resistance, Stratified by Adherence ....................................................................................................................................121Appendix XX. Access to Drug Resistance Resting Cohort: Comparison Between Excluded and Included Individuals. ..........................................................................................................122Appendix XXI. Adjusted Odds Ratios of Access to Drug Resistance Testing, Stratified by People Who Inject Drugs (PWID).............................................................................................123Appendix XXII. Adjusted Hazard Ratios of Development of Drug Resistance, Stratified by People Who Inject Drugs (PWID).............................................................................................124Appendix XXIII. Development of Drug Resistance Cohort: Comparison Between Excluded and Included Individuals. ..........................................................................................................125Appendix XXIV. Access to Drug Resistance Testing Cohort: Comprehensive Univariable and Multivariable Models ..........................................................................................................126Appendix XXV. Development of Drug Resistance Cohort: Comprehensive Univariable and Multivariable Models .................................................................................................................128xiiiList of TablesTable 2.1       Ranked Censoring Criteria ......................................................................................25Table 2.2       Description of Cohort Baseline Characteristics Grouped by Baseline Resistance Status, and Associated Bivariable Comparisons............................................................................28Table 3.1       Clinical Baseline Characteristics .............................................................................46Table 3.2       Census Tract-Level Sociodemographic Characteristics ..........................................47Table 4.1       Accuracy Analysis: MiSeq-Sanger Comparison Summary.....................................68Table 4.2       Repeatability Sequence Analysis.............................................................................71Table 4.3       Reproducibility Analysis: Concordance Results .....................................................72Table 4.4       Cross-Contamination Analysis: MiSeq-Sanger Comparison Summary..................74xivList of FiguresFigure 1.1       Estimated Prevalence of HIV in Vancouver, BC, by Risk Group ...........................3Figure 1.2       Number of HIV Diagnoses and Number Active on HAART in BC, 1996-2009.....8Figure 1.3       HIV Life Cycle and Genome Map: Resistance Genes of Interest............................9Figure 1.4       British Columbia Distribution of Third Drug in HAART, 2011-2016 ..................12Figure 1.5       General Protocol for HIV-Related Genetic Testing in BC.....................................13Figure 2.1       Percentage of Transmitted HIV Drug Resistance, and Acquired HIV Drug Resistance (Detected as of April 2016) Among Treatment Naïve Participants Initiating cART in a Given Year (N=6543) .................................................................................................................30Figure 2.2       Percentage of Participants with Acquired Drug Resistant Mutations Detected in a Given Drug Category, by Years After cART Initiation.................................................................31Figure 2.3       Multivariable Odds Ratios for the Development of Acquired Drug Resistance Associated with Regimen Adherence, per Time Interval of cART Initiation ...............................33Figure 3.1       Odds Ratio and 95% Confidence Intervals Assessing the Odds of Accessing Drug Resistance Testing, by Model ........................................................................................................48Figure 3.2       Hazard Ratio and 95% Confidence Intervals Assessing the Likelihood of Developing Drug Resistance in a Treatment Naïve Cohort Subset, by Model .............................51Figure 4.1       First and Second Round PCR Maps .......................................................................60Figure 4.2       Second PCR Primer Design ...................................................................................60Figure 4.3       Observed and Expected Allele Group Frequency ..................................................67Figure 4.4       Accuracy Analysis: Detailed Concordance Comparison .......................................70Figure 4.5       Observed and Expected Allele Group Frequency: Confirmation of Results .........75xvList of Abbreviations3TC – LamivudineACD – Acid Citrate DextroseADR – Acquired Drug Resistance AIC – Akaike Information Criterion: for use in statistical model selection, determines the quality of each model relative to the other models. AIDS – Acquired Immune Deficiency SyndromeaHR – adjusted Hazard RatioaOR – adjusted Odds RatioAZT – ZidovudineBC – British ColumbiacART – Combined Antiretroviral TherapyCCR5 – C-C Motif Receptor 5CD4 – Cluster of Differentiation 4CDC – Centers for Disease ControlCXCR4 – C-X-C-Motif Receptor 4ddATP – Dideoxyadenine TriphosphateddCTP – Dideoxycytosine TriphosphateddGTP – Dideoxyguanine TriphosphateddTTP – Dideoxythymidine TriphosphateDEPC – Diethyl pyrocarbonateDNA – Deoxyribonucleic AcidEDTA – Ethylenediaminetetraacetic acidFDA – United States of America Food and Drug AdministrationFSW – Female Sex WorkersFTC – EmtricitabineGEE – Generalized Estimating Equation: estimates parameters of a generalized linear model with unknown correlation between outcomes. Gp120 – Envelope Glycoprotein 120Gp41 – Envelope Glycoprotein 41HAART – Highly Active Antiretroviral TherapyHCV – Hepatitis C VirusHIV – Human Immunodeficiency VirusHLA – Human Leukocyte AntigenHOMER – HAART Observational Medical Evaluation and ResearchHR – unadjusted Hazard RatioHTLV – Human T-Lymphotrophic VirusIAS-USA – International AIDS Society-United States of America: non-profit organization promoting HIV/AIDS research and education since 1992. Accredited by the Accreditation Council for Continuing Medical Education. IDU – Injection Drug UserxviINI – Integrase InhibitorIPD-IMGT – Immuno Polymorphism Database-International ImMunoGeneTics projectLAV – Lymphadenopathy-Associated VirusMDRT – Multiple Drug Rescue TherapyMHC – Major Histocompatibility ComplexMSM – Men Who Have Sex With MenNIH – National Institutes of HealthNNRTI – Non-Nucleoside Reverse Transcriptase InhibitornRTI – Nucleoside/Nucleotide Reverse Transcriptase InhibitorNS5a – Nonstructural protein 5ANS5b – Nonstructural protein 5BOR – unadjusted Odds RatioPCR – Polymerase Chain ReactionPEPFAR – Presidents Emergency Plan for AIDS Relief PI – Protease InhibitorPR-RT – Protease-Reverse TranscriptasepVL – Plasma Viral LoadPWID – People Who Inject DrugsQC – Quality ControlRNA – Ribonucleic AcidSAS – Statistical Analysis SoftwareTDR – Transmitted Drug ResistanceV3-Loop – Third Hypervariable LoopWHO – World Health OrganizationxviiAcknowledgementsTo my supervisor, Richard Harrigan, you have my deepest gratitude for propping me up on your shoulders, being incredibly supportive, all the student beers, and providing some seriously valuable life lessons in your unique and hilarious manner. I feel very fortunate to have had you as a teacher and mentor. I would like to thank the other members of my supervisory committee: Hélène Côté and Mel Krajden, for sharing their valuable time and lending me a guiding hand. Thank you to all the study participants, without whom this research would not be possible. Additionally, I would also like to thank study collaborators and co-authors for the opportunity to work and learn alongside you. I am deeply grateful to the BC Centre for Excellence in HIV/AIDS laboratory team for their guidance and support over the years. I’d like to thank Liliana Barrios and Celia Chui for their patience, instruction and for being incredibly good at what they do. Particularly, I’d like to thank Chanson Brumme for tolerating my plethora of questions, and being an inspiring mentor and friend. Finally, to my Mom, for being a strong woman who is unapologetically herself; Dad, for constantly pushing me out of my comfort zone and teaching me my own courage; and Thom, for patience and partnership – wherever you are is home. xviiiDedicationI dedicate this work to my family.1Chapter One: Introduction 1.1 History of HIV1.1.1 Discovery In 1981, the U.S. Centers for Disease Control (CDC) documented a sharp increase in the number of deaths from Kaposi’s Sarcoma and opportunistic infections. However, it was not until the following year when the term autoimmune deficiency syndrome (AIDS) was first used.1,2 By 1983 a human retrovirus was isolated from the blood of patients diagnosed with AIDS, similar to the recently discovered Human T-cell Leukemia Viruses (HTLV). This new virus was discovered by Dr. Luc Montagnier’s laboratory at the Pasteur Institute, and was named lymphadenopathy-associated virus (LAV).3,4 In 1986, LAV was renamed Human Immunodeficiency Virus (HIV) by the International Committee on Taxonomy of Viruses.5 The U.S. CDC documented 270 cases of severe immunodeficiency in men who have sex with men (MSM), leading to 121 deaths in 1981; however, the CDC estimated that tens of thousands of people were already infected with HIV in 1982.5 In 2015, the World Health Organization (WHO) estimated 36.7 million people were living with HIV worldwide, leading to 1.2 million AIDS-related deaths annually.6 Though the number of individuals living with HIV/AIDS continues to rise globally, the annual number of new infections has been stable at approximately 1.9 million between 2010 and 2015.7The discovery of HIV/AIDS was characterized by a surge in public fear, further stigmatization of MSM, and a significant loss of life, which continues to the present day. A 2group of highly organized activists, largely lead by members of the MSM community, arose from this crisis; calling for action from select U.S. government officials, hosting protests, and focusing public pressure by outlining specific treatment demands. These activists ultimately played a fundamental role in motivating government funding for treatment research, leading to the discovery of the first antiretroviral drug (zidovudine) in 1987.8 When activists pushed for faster access to better drugs, the FDA created a program expanding access and expediting approval of therapies related to HIV/AIDS.8 The proceeding decade saw a flurry of antiretroviral drug development. To date, there are dozens of approved antiretroviral drugs, and an estimated 18.2 million individuals received antiretroviral therapy globally in 2016.7,9 1.1.2 The BC EpidemicDuring the late 1980s, the BC HIV epidemic followed a similar trajectory to the epidemic in the United States. There was a documented surge in the number of new HIV infections among MSM, with an estimated incidence of 8.41 per 100 person-years among MSM in Vancouver between 1982-1985.10–12 The late 1990s saw a second wave in BC, with rapid increase in the number of new infections among people who inject drugs (PWID).10 One study estimated HIV incidence to be 18.6 per 100 person-years in Vancouver between 1996-1997 for PWID, and even higher among female sex workers (FSW).133Figure 1.1 Estimated Prevalence of HIV in Vancouver, BC, by Risk Group The estimated prevalence of HIV in each risk group is illustrated, as shown by percentage of HIV-positive individuals within a risk group, by year of infection. FSW; female sex worker, PWID; people who injects drugs, MSM; men who have sex with men. This figure is from McInnes CW, et al., Harm Reduct J. (2009). This figure is used under the terms of the Creative Commons Attribution License. © BioMed Central Ltd. 2009; McInnes et al.10  In response to this health crisis, the BC Centre for Excellence in HIV/AIDS was formed in 1992, and became the custodian of the Drug Treatment Program. This is a monitoring, research and treatment program designed to ensure individuals diagnosed with HIV in BC have access to antiretroviral drugs free of charge.14 The BC Centre for Excellence in HIV/AIDS is involved with additional ongoing research activities through several longitudinal observational cohorts, as well as the development and dissemination of evidence-based clinical treatment guidelines for HIV/AIDS care.As of June 2016, approximately 13,500 individuals were reported to have received treatment for HIV/AIDS in BC since the Drug Treatment Program was established in 1992. The majority 4of these were men (84%), 37% were MSM, and 28% were PWID. In recent years, 12,000 individuals were estimated to be infected with HIV within BC (in 2011). A subset of 7146 were diagnosed and receiving antiretroviral medications, and therefore were being monitored by the Drug Treatment Program (in 2016).15,16 1.1.3 VirologyHIV is transmitted through blood (e.g. needle sharing among PWID) or contact between mucosal membranes (e.g. vaginal or anal sex) and certain other bodily fluids (e.g. semen, breast milk).17 HIV targets CD4-positive cells, including T helper cells, monocytes, macrophages, and dendritic cells.18 These cells are essential to both adaptive and innate immunity, as they send signals to other immune cells, such as CD8 T cells, which target and kill invading pathogens.18 HIV enters these cells by associating with the CD4 glycoprotein on the cell surface, as well as one of two co-receptors; CXCR4 or CCR5.19 Once inside the host cell, HIV avoids detection by down-regulating innate immune inflammatory responses.20 After initial infection, HIV can become latent, further evading detection and destruction by CD8 killer cells. The primary HIV target cells, CD4 T cells, can reside in a quiescent state for long periods of time, shielding HIV from detection and contributing to the viral reservoir.21 The high rate of HIV mutation enables evasion of antibody recognition of the virus particle outside the cell by impeding the presentation of consistent HIV peptides on Major Histocompatibility Complex (MHC). Due to the time delay of the adaptive immune response, this perpetual change prevents infected cells from being flagged for destruction by the immunesystem.18 The rate of mutation is largely attributed to the high error rate of the viral 5reverse transcriptase enzyme, which generates single mutations for every 10,000 bases (the equivalent length of the HIV genome, per single round of replication), combined with the rapid replication rate of HIV; in the order of 1010 to 1011 virions per day during uncontrolled infection.18Individuals experience high levels of viral replication in the first one to twelve weeks of infection, resulting in non-specific symptoms that resemble mononucleosis or flu (e.g. fatigue, fever, sore throat, head and body aches, swollen lymph nodes, rash).22 Seroconversion occurs several weeks after and the immune system is able to initiate a strong response, reducing viraemia. As a result of the immune response viral load plummets, typically reaching a set point between 103 and 105 RNA copies per mL of plasma, and the patient becomes asymptomatic.18 However while the patient may not experience symptoms during this prolonged period (years to potentially more than a decade in duration), in most cases there is continued depletion of CD4 T cells as a result of viral replication. As the CD4 T cells become depleted, and the rate of CD4 T cell production eventually drops below the rate of CD4 T cell loss. In this way, HIV slowly cripples the body’s ability to mount an effective immune response. CD4 cell count eventually drops below 200 cells per mL of blood, resulting in the emergence of opportunistic infections and the development of AIDS.18 As a result of the development of antiretroviral drugs and combination antiretroviral therapy (cART), such as highly active antiretroviral therapy (HAART) (see section 1.2), HIV replication can be suppressed sufficiently to prevent the decay of the CD4 T cell population and allow CD4 T cell recovery. The life expectancy of individuals who initiate therapy at an early 6stage of infection (CD4 cell counts >350 cells/μL) is nearing that of the general population.23–25 Antiretroviral drugs have been the most effective treatment for HIV/AIDS, and it is estimated that their use prevented 4.2 million deaths world-wide as of 2012.26 1.2 Emergence of HIV Drug ResistanceShortly after the first antiretroviral (zidovudine) mono-therapy was approved by the FDA in 1987, research showed the rapid emergence of resistance due to genetic mutations in the viral reverse transcriptase gene, nullifying the drug’s suppression of viral replication shortly after initiation.27–29 When selection pressure was applied, a more replication-fit mutant variant filled the ecological niche left by the susceptible wild-type variant.18 In this manner viral genomic mutations rendered a drug ineffective at viral suppression. Once drug resistance emerged, genotypically resistant provirus was incorporated into the viral reservoir.30 Fortunately, when enough selection pressure was applied, the low rate of viral replication decreased the likelihood of drug resistance mutations. However, if the selection pressure were interrupted or antiretroviral medication discontinued, viral replication would resume, and the likelihood of resistant viral variant selection would increase.31 With the development of a diverse array of drugs, it was theorized that a combination of drugs could be more efficacious and prevent the emergence of drug resistance. Thus dual-therapy and triple-therapy were attempted and shown to be more effective alternatives.32–35 HAART was developed from this research and became the gold standard for HIV treatment.36,37 HAART was a structured form of combination antiretroviral therapy where two nucleoside reverse transcriptase inhibitors (nRTI) were used in conjunction with a third antiretroviral from another 7drug class. These other drug classes were non-nucleoside reverse transcriptase inhibitors (NNRTI), protease inhibitors (PI), entry inhibitors, fusion inhibitors, or integrase inhibitors (INI). Compared to previous treatments, HAART was an effective tool in suppressing HIV replication to undetectable levels and preventing AIDS-related mortality and the emergence of resistance.38,39 However HAART was not a cure; HIV was able to lay dormant due to latency, or host cell quiescence.18 HAART targeted actively replicating virus, therefore this replication competent viral reservoir remained unaffected such that when HAART was discontinued, viraemia returned.40,41 HAART has been the cornerstone of the “Treatment as Prevention” health policy, which has demonstrated that successful treatment with HAART reduces HIV viral loads to undetectable levels, in turn reducing the risk of HIV transmission to near zero.32,42 HAART is also a fundamental pillar of the WHO “90-90-90” treatment target: by 2020, 90% of HIV-infected individuals will know their status, 90% of these individuals will be receiving HAART, and 90% of treated individuals will achieve viral suppression. These health policies have been adopted around the world.43–45 Between 1996 and 2009, HAART expanded more than 5-fold in BC (Figure 1.2).46 However, as with the overuse of antibiotics and development of antibiotic resistance, the long term effect of increasing HAART uptake makes the emergence of drug resistance an ongoing concern.47,48 8Figure 1.2 Number of HIV Diagnoses and Number Active on HAART in BC, 1996-2009  This figure shows the estimated number of new HIV diagnosis in BC, by year of detection. Additionally, the number of BC residents actively taking HAART is shown in each year between 1996-2009. This figure was adapted from Montaner, et al., Lancet (2010), with modifications by Genevieve Rocheleau. The right to produce this figure was granted by Elsevier Inc. via RightsLink (license number 4171630998215) © 2010 Elsevier Inc., Montaner et al.461.3 Antiretroviral Drugs; An Overview1.3.1 Nucleoside and Nucleotide Reverse Transcriptase InhibitorsZidovudine (AZT) was the first antiretroviral drug approved by the FDA in 1987, and is a member of a class of HIV drugs called nucleoside/nucleotide reverse transcriptase inhibitors (nRTIs). Reverse transcription is an essential step in the HIV life cycle, as it converts the HIV single-stranded RNA genome into DNA within the cytosol (Figure 1.3). This step occurs after the viral envelope has merged with the cellular membrane, and it produces DNA to be transported into the nucleus and integrated into the host cell genome. nRTIs are analogues of natural nucleosides or nucleotides, where the –OH group at the 3` position has been removed or 9replaced with a non-functional alternative. nRTIs are incorporated into the newly synthesized DNA by the reverse transcriptase enzyme, which catalyzes the covalent bonding to the previous nucleotide in the growing chain. In nRTIs, the missing 3` –OH group inhibits reverse transcriptase enzyme function and no downstream nucleosides can be covalently bound. This causes the chain to be truncated and non-functional. Without a functional reverse transcription process, HIV genetic material cannot be integrated in to the host cell DNA and no viable viral progeny can be produced. Figure 1.3 HIV Life Cycle and Genome Map: Resistance Genes of InterestA) Map of HIV genome highlighting the resistance genes of interest, their relative location on the genome, and precursor polyprotein. B) The life cycle of HIV and targets for antiretroviral suppression of viral replication. Figure A was created by Genevieve Rocheleau, using data from Richman, et al. (2003) 18. Figure B was adapted from Laskey and Siliciano, Nature Reviews Microbiology (2014), with modifications by Genevieve Rocheleau. The right to produce this figure was granted by Nature Publishing Group via RightsLink (license number 4171640171600) © 2014 Nature Publishing Group, Laskey and Siliciano 49101.3.2 Non-Nucleoside Reverse Transcriptase InhibitorsNon-nucleoside reverse transcriptase inhibitors (NNRTIs) also target the reverse transcriptase enzyme, but use allosteric inhibition instead of substrate analogues. Generally, NNRTIs target the binding pockets adjacent to the active sites, and prevent transcription by causing disassociation of the reverse transcriptase-template primer complex.18 The first NNRTI, nevirapine, was approved for use in 1996.91.3.3 Protease InhibitorsThe approval of the first protease inhibitor (PI) saquinavir in 1995 ushered in the HAART era.50 The HIV genome is transcribed and translated as several large precursor polyproteins (Figure 1.3). These precursor polyproteins are cleaved by a viral protease enzyme before becoming functional.18 These protease-facilitated steps in the viral life cycle are necessary for the synthesis of infectious virus particles. PIs prevented this essential function through competitive inhibition of the active site.30 PIs were typically co-prescribed with pharmacokinetic boosting agents, such as cobicistat or ritonavir.51,52 These boosters enhance the concentration of PIs by reducing the rate of PI metabolism.53 However the toxicity of these drugs commonly produced serious side effects, such as lipodystrophy, dyslipidemia, insulin resistance and elevated cardiovascular disease risk, as well as drug resistance, which quickly became a concern.18,52 111.3.4 Entry, Fusion and Integrase InhibitorsAfter reverse transcription of HIV RNA in the cytosol, the HIV DNA product is transported into the nucleus and inserted into the host cell genome using the HIV integrase enzyme (Figure 1.3).18 The HIV genome is then transcribed and translated using host cell machinery. Integrase inhibitors (INI) interfere with HIV integrase enzyme functioning. The integrase inhibitors currently in use prevent integration by binding at the integrase active site, and block strand transfer activity.54,55 The first approved integrase inhibitor was raltegravir (2007).56 Entry inhibitors generally interfere with the cellular process of HIV entry into the cell. In 2007, maraviroc was the first approved entry inhibitor.57 Maraviroc binds to the CCR5 co-receptor for HIV entry located on the surface of CD4 T cells, thus blocking cell entry and preventing viral replication.20 Since maraviroc selectively binds CCR5, the tropism of a patient’s viral population must be determined prior to use. The viral tropism is determined by sequencing the V3-Loop region on the viral genome, a small section of the gp120 viral surface protein that is approximately 35 amino acids in length (Figure 1.3).58 Alternatively, fusion inhibitors prevent the virus from entering host cells by binding to proteins on the viral envelope surface necessary for fusion with the cell membrane. One of the primary examples of this is gp41 - a critical protein that penetrates the cell membrane immediately before HIV envelope fusion (Figure 1.3).18 Fusion inhibitors targeting gp41 include enfuvirtide, which was approved for use in 2003.57 Due to their cost and inconvenient dosing schedule, fusion and entry inhibitors are generally used as last resort ‘salvage therapies’ for patients with high levels of accumulated resistance mutations.59 12Use of these antiretroviral drug classes, especially integrase inhibitors, have become more prevalent within BC in the past few years (Figure 1.4). However since the recruitment of participants for this research ceased in 2014, an insufficient number of participants were prescribed these drugs to assess the risk of developing drug resistance. Therefore, entry, fusion and integrase inhibitors will not be extensively examined in the primary research herein. Figure 1.4 British Columbia Distribution of Third Drug in HAART, 2011-2016The image shows the proportion of individuals receiving antiretroviral drugs in BC by the third (non-NRTI) drug class in the HAART regimen. MDRT; multiple drug rescue therapy, Other; antiretroviral regimens other than HAART or MDRT (e.g. triple nucleoside regimens). Figure 1.4 includes information from the BC Centre for Excellence in HIV/AIDS Drug Treatment Program Monthly Report dated October 2016, reproduced with permission. © 2016 BC Centre for Excellence in HIV/AIDS 1.3.5 Acquired vs Transmitted Drug ResistanceHIV drug resistance can be grouped in to two categories: Acquired drug resistance (ADR) and transmitted drug resistance (TDR). ADR occurs when an individual is infected with a wild-type virus, but the virus is allowed to replicate in the presence of antiretroviral drugs. This could 13be due to imperfect adherence to the prescribed regimen, or in the early days, when monotherapy was prescribed but it was not effective at sufficiently limiting viral replication. The virus becomes drug resistant through escape mutations as a result of the drug’s selection pressure, and this is called acquired drug resistance. Alternatively, individuals can also be infected with a strain of virus that already harbours resistant mutations. This is called transmitted drug resistance.1.4 HIV Genotypic TestingWith the discovery of HIV drug resistance came the urgent need to develop rapid personalized testing for individuals undergoing virologic failure. In BC, these tests primarily involved genetic testing of the unique viral population within each individual patient. This was done by extracting viral RNA from human plasma, reverse transcribing the RNA to DNA, running nested polymerase chain reaction (PCR) to amplify the genes of interest that are targeted by antiretroviral drugs, such as reverse transcriptase (nRTIs/NNRTIs), protease (PIs), integrase (INI), gp41 (fusion inhibitors) and V3-Loop located in gp120 (entry inhibitors) (see Figure 1.4). These genes of interest were then sequenced (see section 1.5). See Figure 1.5 for a visualization of the general protocol for HIV-related genotypic testing in BC. Results for these tests were typically available within 10 days of the laboratory receiving the specimen, and all tests in BC were completed at the BC Centre for Excellence in HIV/AIDS Laboratory, located at St. Paul’s Hospital Vancouver, Canada. 14Figure 1.5 General Protocol for HIV-Related Genetic Testing in BC This figure describes the general procedure for HIV-related genetic testing in BC, from when the sample reaches the laboratory to when the physician receives the results. This image was created by Genevieve Rocheleau. 1.5 Relevant Sequencing Technologies1.5.1 Sanger Sequencing Sanger sequencing was first developed by Frederick Sanger in 1975.60 Since then it has become the gold standard for both clinical and research-based sequencing applications around the world. Modern Sanger sequencing uses florescent-dyed terminating nucleotides. During thermocycling, these nucleotides are incorporated by several rounds of DNA polymerase elongation using the target template strand and primers. Terminating nucleotides, such as ddATP, ddTTP, ddGTP, and ddCTP, each have a unique fluorescent label. The addition of the fluorophore-bound nucleotide prevents polymerase-catalyzed –OH cleavage necessary to add a subsequent nucleotide, truncating the DNA and leaving the terminal fluorophore. This is done randomly, creating DNA sequences of many sizes, each with a terminal fluorophore. By the colour of the fluorophore, it is possible to determine the terminal nucleotide. In conjunction with 15capillary gel electrophoresis, the size of these DNA fragments and subsequently the location of the fluorophore in the DNA strand, as well as the colour are determined simultaneously. Raw chromatograms are generated using Phred software, subsequently sequences are aligned to a standard using RECall freely accessible software, and trained lab technicians quality check the sequence before being run through an interpretation algorithm.61,62 Sanger sequencing is a “population-based” sequencing approach. The limit of detection for Sanger sequencing is approximately 15-20% prevalence for unique viral genetic variants.58 Therefore minority variants with less than 15% prevalence generally remain undetected. This limitation is problematic when the undetected minority variant might harbour drug resistance.63 The ABI 3730xl DNA Analyzer is used for Sanger sequencing throughout this thesis.1.5.2 Illumina MiSeq Sequencing Illumina is a biotechnology company with a focus on development of “next generation sequencing” platforms. The MiSeq is a bench top device released in 2011 that focuses on small whole genome or target “amplicon” sequencing.64 In this thesis, amplicon sequencing will be the primary focus.60 In preparation for sequencing, PCR amplifies specific target DNA. Primers are synthesized with Illumina-generated sequence called ‘adaptor regions’ on one end, and the target sequence on the other. Through PCR, these adaptors are added to the terminus of each target DNA strand. The adaptors contain specific sequence that adheres to the flow cell, sequencing primers, as well as indices that are specific to the sample and the direction (5` to 3`, or 3` to 5`) of the read. In this way, many samples can be run on the same flow cell and are de-multiplexed 16during data analysis. An ‘amplicon’ is a target DNA sequence, flanked by two adaptor regions on either end. In short, the terminal ends of the single stranded amplicon adhere to the flow cell, and then are selectively cleaved off, resulting in only the forward read DNA template remaining.60 These replicate while still attached to the flow cell, creating ‘clusters’ of the same DNA sequence. Similar to Sanger sequencing, fluorescently dyed terminal nucleotides are then used. However, on the MiSeq, the fluorophore cleaved after optical imaging, allowing for polymerization of the DNA strand to continue. All four nucleotides are washed over the flow cell simultaneously, and bind to their complimentary nucleotide using polymerase. Clusters emitting the same fluorescent colour can then be detected using the highly sensitive camera. The fluorophore is then cleaved and washed away, leaving behind the nucleotide. The next consecutive nucleotide is then added and sequenced in the same manner. This process is called sequencing by synthesis.60 This thesis was completed using “paired-end sequencing”, where this process is repeated for the reverse read. Both read directions are compared for concordance and added accuracy.  The sequence data is then analyzed through an in-house, test-specific algorithm. Generally, the algorithm cleans up the sequence data through various quality control checks, aligns the remaining sequences to a standard reference sequence, and based on the number of reads, produces either the most common unique sequence, or the consensus sequence, depending on the downstream application.17Due to the large amount of data generated (approximately 8 Gb per run, or 24-30 million reads), the MiSeq has a much greater sequencing depth compared to Sanger sequencing.65 Where Sanger sequencing has a limit of detection at about 15-20% prevalence for unique viral variants, the MiSeq comfortably detects variants at 2-5% prevalence, making it a more sensitive instrument, better able to detect minority variants that may contain drug resistance. 58,66  1.6 Research Objectives and Thesis OrganizationThe objectives of this thesis are to determine longitudinal trends of HIV drug resistance over the past two decades, to describe sociodemographic factors associated with access to resistance testing and development of acquired drug resistance, and to demonstrate the effectiveness of a newly developed Illumina MiSeq assay to determine the host HLA-B*57:01 genotype for clinical abacavir hypersensitivity prescreening. Through longitudinal observation of cohorts that are treatment naïve upon initiation of HAART, this thesis aims to describe the trends in HIV drug resistance in BC. This includes prevalence of acquired and transmitted drug resistance between 1996 and 2014, how prevalence of drug resistance has changed with duration of treatment, and how associations of clinical and treatment variables to the development of acquired drug resistance have changed over calendar time. Similarly, drug resistance data and census tract-level social and demographic data are used to assess sociodemographic correlates of drug resistance testing access, as well as development of acquired drug resistance. This exploratory study aims to identify sociodemographic factors that lead to experience gaps in the HIV cascade of care, which might affect the risk of developing drug resistance. Additionally, this thesis presents the validation of an HIV-related next generation sequencing (NGS) assay of host HLA-B*57:01 correlates of abacavir 18hypersensitivity screening. Using the MiSeq, this clinical validation is a proof of concept for future HIV-related tests, demonstrating they can be readily performed on NGS platforms. 1.6.1 Chapter DescriptionsThis thesis is divided into five chapters. Chapter One provides a general overview of background information on HIV, including a brief history of its discovery, BC-specific epidemiology, molecular virology, drug resistance, an overview of genetic testing for HIV drug resistance within BC, and finally, a description of the research objectives.Chapters Two to Four are the primary research chapters. Chapter Two takes a retrospective view of HIV in BC by describing trends in HIV drug resistance observed based on follow up of a large cohort for nearly two decades. Samples were collected as part of routine patient care and archived in a biorepository. All genotypic drug resistance test results were anonymized and maintained in a database, along with other HIV-relevant administrative data such as prescription details, adherence measures, risk group, CD4 cell count, and viral load. The proportion of individuals with acquired drug resistance was compared to those with transmitted drug resistance in each calendar year. Also, the probability of acquiring drug resistance de novo was estimated based on the number of years on therapy. Finally, a multivariable model stratified by 5-year time intervals highlights the changes in how clinical variables are associated with acquired drug resistance over time. As millions rely on daily intake of cART for the suppression of HIV and the prevention of AIDS, ongoing monitoring of drug resistance remains important. Furthermore, there is conflicting evidence on the prevalence of transmitted resistance in the literature, and in 2012, the WHO called for further research on the declining rate of acquired 19drug resistance. Retrospective observation of drug resistance trends addresses these ongoing concerns in a large cohort with access to ARVs and resistance testing free of charge, followed for nearly 20 years. Chapter Three describes an exploratory study where sociodemographic correlates of acquired drug resistance and access to resistance testing were assessed to determine if unique sociodemographic strata were disproportionately affected. Access to physician prescribed drug resistance testing was associated with sociodemographic correlates using a large cohort of HIV infected BC residents active in the Drug Treatment Program. Individual-level clinical variables and drug resistance testing data were combined with census tract-level sociodemographic data based on postal code of residence from the Canadian census closest to the year of cART initiation in BC. A logistic regression model accounting for clinical and sociodemographic variables determined sociodemographic correlates of access to drug resistance testing. Similarly, a subset of this cohort (HOMER) consisting of initially treatment naïve individuals living with HIV was followed between 1996-2013, and sociodemographic correlates of development of HIV drug resistance were determined. Identification of significant disparities in access to HIV drug resistance testing and development of drug resistance between sociodemographic groups, could highlight groups that require targeted health programming, such as education on available resources or hiring peer navigators to help guide members of specific societal groups through the health care system. Access to drug resistance testing could help ensure prompt diagnosis of HIV drug resistance and appropriate treatment, by adjusting the prescribed antiretroviral regimen to include virus susceptible drugs. In this way, the risk of morbidity and mortality associated with AIDS and transmission of resistant strains could be reduced. 20While Chapter Two and Three described HIV drug resistance retrospectively, Chapter Four aimed to describe the future of HIV-related genetic testing by outlining the clinical validation of a next generation sequencing test. As per standard clinical validation practice, MiSeq sequencing assay for host HLA-B*57:01 was compared to the Sanger sequencing gold standard currently in clinical use. Accuracy, specificity, sensitivity, and precision were determined and reported. Additionally, a large sample set of HIV exposed uninfected children were tested for host HLA-B*57:01 as a proof of principle. As technologies with greater throughput and genetic depth become available in the clinical setting, they offer an opportunity to provide more information and a better quality of care for individuals with HIV. Therefore, it is important to describe how existing tests can be transitioned to new sequencing platforms in this emerging period of genetics-based personalized medicine. Chapter Four was the final chapter, and presents the conclusions, limitations, future applications, and implications of the findings in the context of the current body of research in the field. 21Chapter Two: Longitudinal Trends of HIV Drug Resistance in a Large British Columbia Cohort, 1996-2016 2.1 Background and IntroductionIn 2015 an estimated 36.7 million people were infected with human immunodeficiency virus (HIV) worldwide, of whom 17 million were undergoing antiretroviral drug therapy.67 Combined antiretroviral therapy (cART) suppresses HIV virus replication and disease progression.68,69 However, drug resistance has hindered the efficacy of antiretroviral drugs since the first nucleoside reverse transcriptase inhibitor (nRTI) monotherapies and dual nucleoside therapies became available.27,70 Once drug resistance mutations are detected, cART regimens are typically changed; drugs that are no longer effective are switched for drugs that will likely suppress the virus. This approach has significantly reduced pVL, and significantly increased CD4 cell count in randomized control trials.71,72 The trial results led to sequenced-based drug resistance testing in BC (1998), and has become the standard of care for both treated and untreated individuals, free of charge. The European multi-centre cohort (2013) and the Swiss HIV cohort (2016) studies both reported declines in HIV resistance incidence based on 10 and 15 years of retrospective observation, respectively.73,74 Citing this evidence, the World Health Organization HIV Drug Resistance Report (2012) declared a need for more research on this topic.75 Our study aims to expand upon these findings using nearly 20 years of data on long-term trends of drug resistance – an unprecedented duration in this field. Additionally, there exists contradictory reports on rates of transmitted drug resistance (TDR) in high-income countries, and seldom are TDR and acquired drug resistance (ADR) compared.76–78 Using our longitudinal retrospective analysis on a 22large British Columbia (BC) cohort (N=6543) observed between 1996-2016, we aimed to identify long-term trends in drug resistance. The BC cohort is ideal to study long term trends of resistance due to the duration of follow-up, free access to care, and the high amount of testing that is recommended in BC treatment guidelines. 2.2 Methods2.2.1 Data CollectionBetween 1992-2015, 13,281 HIV-positive BC residents enrolled in the Drug Treatment Program at St. Paul’s Hospital, Vancouver. The HAART (Highly Active Antiretroviral Therapy) Observational Medical Evaluation and Research (HOMER) cohort is a subset of this population, however it is restricted to initially treatment naïve adults (≥19 years of age) receiving triple drug therapy, who initiated cART in BC between August 1996 and December 2014, and who’s drug resistance information was collected until April 2016.79 Initial cART must have consisted of two nRTIs, and either a NNRTI, a PI (boosted or unboosted), or an alternative according to treatment guidelines.80 To be included in HOMER, pre-therapy CD4 count and pVL must have been measured within six months prior to therapy initiation. Appendix IV compares HOMER participants to those who were treatment naïve at baseline but excluded from HOMER due to missing data, using Cochran-Mantel-Haenszel Statistics. The primary outcome of this study was detection of drug resistance mutations in HIV pVL samples. Patient follow-up was conducted through administrative data collection of lab test results and prescriptions. No formal consent is required for using anonymized administrative data; an information sheet was provided at enrolment. The University of British Columbia Research Ethics Board at St. Paul’s, Providence Health Care site granted ethics approval for the HOMER cohort (H05-50123).232.2.2 Plasma Viral Load TestingpVL testing is free of charge to patients in BC when ordered by a physician, and ordered according to therapeutic guidelines.80 pVL testing was completed at the St. Paul’s Hospital Virology Laboratory, and both clinical and research samples were included.2.2.3 Drug Resistance GenotypingGenotyping was completed at the BC Centre for Excellence in HIV/AIDS (BC CfE) using Sanger sequencing as described previously.81 A total of 23,271 HIV PR-RT sequences from pVL samples were compared to major mutations on the IAS-USA (2015) drug resistance mutations list.82 Sequences were then grouped into four drug resistance categories: lamivudine/emtricitabine (3TC/FTC), other nRTI, NNRTI, and PI. Since the 3TC/FTC resistance mutation M184V is common, but does not confer resistance to other nRTIs, it is categorized separately to avoid over-representing other nRTI resistance. The clinical guidelines describe a temporal framework for pVL and drug resistance tests, however these were not strictly followed, and have changed several times over this longitudinal study. Individual physicians largely determined the frequency of tests. Physicians also were able to request tests on stored samples retrospectively. If a sample was tested more than once, results from all tests were considered. If a participant tested positive for resistance in one drug category, all subsequent tests were still considered for the other categories.242.2.4 Estimation of AdherenceAs previously validated through blood drug level testing, the number of days in first year of follow-up covered by an antiretroviral prescription estimated individual adherence (expressed as a percentage).31 Participants were then stratified into six categories of adherence: 0-<40%, 40-<60%, 60-<80%, 80-<90%, 90-<95%, ≥95%.2.2.5 Statistical Analysis: Percentage of Participants with Transmitted or Acquired Drug ResistanceA two-sample test for equality of proportions with continuity correction was used to compare the change in percentage of participants with ADR or TDR in 1996, to those in 2014 using R 3.2.3 software.83 Individuals were counted in year of cART initiation, and resistance was determined using test results up to April 2016. TDR was broken down in an inset plot by resistance category. Individuals who had never received a drug resistance test were included in this analysis; those without a test before therapy initiation were assumed not to have TDR (see sensitivity analysis in section 2.3.1), and those with a negative TDR test but without a test after therapy initiation were assumed not to have ADR. TDR was defined using World Health Organization transmitted mutation list (2009) shown in Appendix VI for reference.84 To assess the pattern of TDR as of 2015, Appendix VII used a subset of more recently tracked Drug Treatment Program participants (N=6818) with the same inclusion, exclusion, and censoring criteria as the HOMER cohort.252.2.6 Statistical Analysis: Kaplan-Meier AnalysisKaplan-Meier analysis estimated prevalence of ADR within the cohort. The primary outcome was detection of a drug resistance mutation at the time of resistance testing, using cART initiation date as reference (N=1092). Participants who did not have the primary outcome were censored at the most recent test meeting either the censoring criteria in Table 2.1 (N=4769), or prior to moving away from BC, starting a placebo-blinded trial, death, or the last available contact with participant. Results were compared using the log-rank test.Table 2.1 Ranked Censoring CriteriaAmong participants who were censored (did not have primary outcome), the criteria are applicable to data collected prior to meeting exclusion criteria (moving away from BC, starting a placebo blind trial, death, or the last available contact with participant). Criteria are ranked in order of importance; should an individual not meet criterion 1, they would be assessed for criterion 2, and so on. The individual would be censored at the earliest time point when a criterion is met. The number censored for each criteria are given in the right-most column. Characteristic N (%)Without baseline resistance 5861(100)1. Any detectable resistance (primary outcome) 1092 (19)2. Resistance tests performed, but no resistance detected 1903 (32)3. No resistance tests performed; censored at the first viral load test with ≥1000 copies/ml collected more than 180 days after the start of therapy 258 (4)4. Consistently virally suppressed (<1000 copies/ml); censored at the date of last viral load 2521 (43)5. No follow-up viral load; censored at the date of first cART prescription 87 (2)Total censored without any drug resistance detected 4769 (81)262.2.7 Statistical analysis: Multivariable Logistic Regression Model for Acquired Drug ResistanceThe multivariable logistic regression model was generated using a modified backward stepwise technique, based on Akaike Information Criterion (AIC) and Type III p-values, to select explanatory variables. Using the area under the receiver operating characteristic curve, goodness-of-fit was assessed to determine the model’s ability to determine a positive or negative outcome.85 Either Fisher’s exact test (2x2 tables) or Cochran-Mantel-Haenszel test (other table sizes) was used to compare categorical variables, while the Wilcoxon rank-sum test was used to compare continuous variables.86 All analyses were performed using SAS version 9.4 (SAS, Cary, NC). The final model controlled for sex at birth (male, female), CD4 count (<200 cells/μL, 200-349 cells/μL, 350-499 cells/μL, >500 cells/μL), third drug in baseline regimen (NNRTI, boosted PI, or other drugs; such as integrase inhibitors, unboosted PIs, or fusion inhibitors), adherence (see above in methods), AIDS-related illness ever reported, ever been an IDU (yes, no, unknown), time interval of cART initiation (1996-2000, 2001-2005, 2006-2010, 2011-2014), age at first cART, and baseline pVL. The primary outcome was measured as the detection of any mutation conferring resistance to any drug resistance category. The model was stratified by intervals of cART initiation to observe changes in drug resistance over time. Participants not achieving the detection of drug resistance endpoint were censored as described in Kaplan-Meier Analysis.27 2.3 Results2.3.1 Cohort ProfileThis study included 6543 initially treatment naïve adults, where 80% of the cohort were male at birth, median age was 41 years, and 36% had a history of injection drug use. A subset had TDR (N=682), and was excluded from Kaplan-Meier and multivariable regression analyses. Pre-therapy cohort characteristics and bivariable comparisons of included versus excluded are shown on Table 2.2. Participants that were never tested for resistance prior to treatment were assumed not to have TDR mutations (N=1260) and were included in ADR analyses. To assess confounding, a sensitivity analysis was conducted where participants with unknown TDR were censored. While no major changes in effect magnitude or direction were observed, selection for inclusion of explanatory variables changed over time; AIDS in 1996-2000, sex and CD4 cell count in 2000-2005, AIDS in 2006-2010, and the third drug in baseline regimen in 2011-2014 (Appendix V). However, censoring removed 19% (1260/6543) of our study population, or 42% (570/1251) from 2000-2005 (Appendix XI). Other time intervals were less affected: 25% (363/1441) in 1996-2000, 16% (301/1843) in 2006-2010, and 2% (26/1326) in 2011-2014.   As longitudinal trends were the primary objective of this study, we decided not to censor unknown TDR and increase the potential for temporal bias, but report this possible limitation. The primary outcome (ADR) occurred in 19% (1092/5861) of patients during follow-up, and 51% (2995/5861) had a resistance test after therapy initiation. The remaining 4769 participants were censored without resistance detected (Table 2.1). Median follow-up time was 6.5 years (interquartile range; 3.2-10.9) with a maximum of 19.4 years, and the median time between cART initiation and first resistance test was 20 months (interquartile range: 3.9-63 months). There was a median of five samples genotyped for drug resistance per patient.  28Table 2.2 Description of Cohort Baseline Characteristics Grouped by Baseline Resistance Status, and Associated Bivariable Comparisons. Q1-Q3 refers to the 25th-75th percentiles. * denotes reference group for bivariable associations. Not everyone in the HOMER cohort received a genotypic test, therefore 8% of participants with no baseline resistance and 1% with baseline resistance are missing subtype data. Bivariable AssociationsBaseline Characteristics No Baseline Resistance*Baseline Resistance P-ValueN (%) 5861(100) 682(100)Sex at birth     Male - n (%)* 4765(81) 548(80)     Female - n (%) 1096(19) 134(20)0.53Subtype     Not B - n (%)* 378(7) 37(5)     B - n (%) 4950(85) 649(94)0.11Median age at cART1 initiation in years (Q1-Q3) 41(33-48) 40(34-47) <0.0001Median CD4 cell count at baseline per μL (Q1-Q3) 230(120-370) 270(150-410) <0.0001Median pVL2 log value at baseline (Q1-Q3) 4.94(4.43-5.00) 4.82(4.28-5.00) <0.0001Ever had AIDS3-related illness      No - n (%)* 4536(77) 548(80)     Yes - n (%) 1325(23) 134(20)0.08History of IDU4 - n (%)      No - n (%)* 2532(43) 304(45)      Yes - n (%) 2098(36) 235(34)      Unknown - n (%) 1231(21) 143(21)0.75Interval of cART initiation    1996-2000 - n (%) 1441(25) 163(24)    2001-2005 - n (%) 1251(21) 93(14)    2006-2010 - n (%) 1843(31) 214(31)    2011-2014 - n (%) 1326(23) 212(31)<0.0001Baseline regimen includes;    Boosted PI5 - n (%)* 2234(38) 330(48)    NNRTI6 - n (%) 2449(42) 201(29)    Other - n (%) 1178(20) 151(22)<0.0001Median number of pre-therapy pVL tests (Q1-Q3) 3(1-7) 3(1-10) <0.0001Median number of pre-therapy resistance tests (Q1-Q3)1(1-2) 1(1-2) <0.00011. Combination antiretroviral therapy (cART)2. Plasma viral load (pVL)3. Autoimmune Deficiency Syndrome (AIDS)4. Injection Drug Use (IDU)5. Protease Inhibitor (PI)6. Non-nucleoside Reverse Transcriptase Inhibitor (NNRTI)292.3.2 Percentage of Participants with Transmitted or Acquired Drug Resistance Of note, the proportion of individuals with ADR declined from 39% in 1996, to 3% in 2014 (p<0.0001) (Figure 2.1). In contrast, a slight increase in TDR was observed over the same period: from 12% in 1996, to 18% in 2014 (p=0.14). Transmitted NNRTI resistance especially increased in 2005-2014, from 7% to 14% (p=0.005). Using a separate BC cohort with more recent data, NNRTI TDR and TDR overall dropped slightly from 14% and 18% in 2014, to 11% (p=0.45) and 17% (p=0.78) in 2015, respectively (Appendix VII). Less than 12% of participants had acquired resistance to three or four drug categories, even up to 17 years of being on cART (Appendix IX). Nearly all multi-class resistance was observed in individuals who started cART in 2010 or earlier; between 2011-2014, only one person acquired resistance to more than two categories of drugs out of the 1326 participants who started therapy in that time (Appendix X).30Figure 2.1 Percentage of Transmitted HIV Drug Resistance, and Acquired HIV Drug Resistance (Detected as of April 2016) Among Treatment Naïve Participants Initiating cART in a Given Year (N=6543). In the primary plot, drug resistance mutations were determined using major mutations in the IAS-USA (2015) mutation list. The inset plot shows the percentage of participants with drug resistance detected pre-therapy initiation in a given year, broken down by category of drug resistance. Drug categories include lamivudine/emtricitabine (3TC/FTC), other nucleoside reverse transcriptase inhibitors (Other nRTI), non-nucleoside reverse transcriptase inhibitors (NNRTI), and protease inhibitors (PI). Individuals who have never received a resistance test (N=1260) were assumed to have no drug resistance.2.3.3 Prevalence of Acquired Drug Resistance by Time Interval of cART InitiationPrevalence of ADR in all drug categories has declined markedly in each successive time interval (Figure 2.2, A-D). However, within the first two years on therapy prevalence of drug resistance increases in all time intervals of this study, though magnitudes are smaller over time. For example, NNRTI resistance increased the most; 7.0% of individuals initiating cART in 1996-2000 at 12 months on therapy, compared to 12% at 24 months and 16% at 36 months. In 31contrast, among those who initiated cART in 2011-2014, these values were 2.0%, 2.7% and 3.2%, respectively. Note that in Figure 2.2, small segments at the end of each curve were excluded to remove censoring artifacts. For each drug category, log-rank tests compared the survival curves between different time intervals of cART initiation; all gave p<0.0001 showing that the drug categories were statistically different between eras. Figure 2.2 Percentage of Participants with Acquired Drug Resistance Mutations Detected in a Given Drug Category, by Years After cART Initiation. Drug resistance mutations are as defined in Figure 1. Kaplan-Meier plots are shown for individuals that initiated cART between A) 1996-2000, B) 2001-2005, C) 2006-2 010, and D) 2011-2014. For each drug category, log-rank tests compared the survival curves between different eras; all gave p<0.0001. E) Shows acquired drug resistance for entire duration of study. Log-rank test compared the survival curves between drug categories, giving a p-value of 0.055.322.3.4 Prevalence of Acquired Drug Resistance After Two Decades on TherapyTo estimate the prevalence of ADR in BC between 1996-2016, Kaplan-Meier estimates of survival were used (Figure 2.2, E). While the prevalence of detected resistance increases rapidly within the first two years of therapy, it then slowed but continued to gradually increase over time.  Notably, the prevalence of resistance to NNRTIs and 3TC/FTC was similar throughout almost two decades on therapy, but the prevalence of developing resistance to PIs or other nRTI was lower, but not statistically lower: Log-rank test compared the survival curves between drug categories, giving a p-value of 0.055 and showing that the curves were not statistically different.2.3.5 Multivariable Model: Variables Associated with Acquired Drug Resistance MutationsMultivariable logistic regression showed that the odds of developing ADR was associated with adherence in each time interval of therapy initiation, as well as other treatment variables (Appendix I). The highest odds of developing drug resistance within a time interval (model stratified by time interval of therapy initiation) shifted from those with 60-<80% adherence in 1996-2010, to those with <40% adherence in 2011-2014 (Figure 2.3 B, Appendix I). Under the same model, use of NNRTI as third drug in baseline regimen, <200 CD4 count at baseline, and IDU were each associated with significantly higher odds of developing drug resistance in every time interval (Appendix I). The model was then stratified by adherence level to compare the changes within a level of adherence over time (Figure 2.3 B, Appendix II). Within each level of adherence, the odds ratio of developing drug resistance dropped consistently in 1996-2014 by 5-3316 fold, with the exception of <40% adherence, where the odds ratio doubled from 0.24 (0.12-0.48) in 2006-2010 to 0.48 (0.20-1.12) in 2011-2014.Figure 2.3 Multivariable Odds Ratios for the Development of Acquired Drug Resistance Associated with Regimen Adherence, per Time Interval of cART Initiation. A) A cross-interval comparison of odds ratios for developing drug resistance within a single adherence level. The multivariable logistic regression model for A was stratified by level of adherence. B) A cross-adherence level comparison of odds ratios for developing drug resistance within each era. The multivariable logistic regression model for B was stratified by time interval of therapy initiation. Variables that were included in the models used in A and B were the same as listed in Appendix I.34 2.4 Discussion This longitudinal retrospective study followed a large BC cohort for nearly two decades. A small increase in TDR occurred concomitant with a marked decline in ADR, a decrease in the prevalence of ADR in all drug categories, and a shift in the impact of adherence intensity on ADR. Any TDR increased from 12% in 1996, to 18% in 2014, with NNRTI TDR being the largest contributor. During the same time period, ADR dropped from 39% to 3%. However, prevalence of ADR gradually rises with time on therapy; even nearly 20 years after cART initiation. After adjusting for years of follow-up, those who started therapy more recently continue to have a lower prevalence of ADR compared to individuals who started during the initial time intervals discussed in this study. The role of adherence in the development of ADR has also changed over time, with the highest risk burden shifting from intermediate adherence to low adherence. Among other treatment variables, use of NNRTI as third drug in baseline regimen, <200 baseline CD4 count, and IDU were consistently associated with significant likelihood of ADR between 1996-2016. This is the longest retrospective observational cohort study on HIV drug resistance to date. We found that the prevalence of ADR by year of therapy initiation decreased with more recent cART initiation by 13-fold over the course of the study. Similar longitudinal findings were observed in Switzerland (1999-2013), and a multi-centre European cohort (1997-2008).73,74 Increasing efficacy of antiretroviral regimens over the past two decades, as well as more readily accessible combination regimens and improved patient management, could explain increasing levels of adherence (Appendix VIII) and the subsequent decline in ADR.31,87,88 Conversely, 35conflicting reports exist on whether TDR is increasing or stabilizing in high-income countries.76–78 We found a small increase in TDR between 1996-2014, in alignment with Rhee et al.’s findings.76 In BC, this seems to be largely attributable to an increasing amount of transmitted NNRTI resistance. Currently more people are starting cART with preexisting drug resistance than eventually developing resistance after starting therapy, however that number remains relatively small.Prior to 2011 participants with 60-<80% adherence had the highest odds ratio of developing drug resistance, as reflected in the literature.89,90 The highest odds of drug resistance shifted after 2011 to <40% adherence. Further research is needed to understand the mechanism behind this shift in adherence, but biological reasoning and clinical evidence points to increasing antiretroviral efficacy; as drugs became more effective, lower levels of adherence are sufficient for viral suppression.88 Though the relationship between adherence and drug resistance is unique for each drug category, this study is limited due to an insufficient N for regimen-specific analysis.91,92 Drug categories in first line cART have been relatively consistent throughout most of this study (Appendix XI). A list of the top ten combination regimens used in each time interval is shown in Appendix III for reference. While the association of IDUs and the risk of transmitting resistance has been well characterized, ADR in this group is studied less frequently.93,94 In a meta-analysis (N=9055), Werb et al. found the risk of developing drug resistance not significantly different between IDU and non-IDU.95 In contrast, we observed that IDUs have maintained significantly higher odds of developing drug resistance between 1996-2016 compared to non-IDU. Adherence could be a 36strong factor in this difference, as it was reported to be lower in individuals who inject heroin and cocaine.96In this study, some inherent bias may affect the results; the proportion of drug resistance detected prior to therapy initiation between 2000-2007 may be biased by a decline in TDR testing, creating the potential for selection bias. Concurrently, the number of participants became more adherent, most likely resulting from changes to patient management in clinical practice, decreasing the likelihood of ADR (Appendix VIII). Additionally, changes to clinical practice during the course of the study may increase the likelihood of ADR, such as removal from therapy or ‘treatment holidays’. Moderate bias may be introduced through longer viral exposure to cART, as this increases the likelihood of drug resistance variants, skewing detection of drug resistance towards the initial therapeutic time intervals of the study. To address this, follow-up time was controlled in the multivariable model. Also, Kaplan-Meier analysis was used to confirm the decline ADR, normalizing the results by duration on therapy while stratifiying by date of therapy initiation.  In the past two decades, HIV drug resistance has shifted from being selected in patients after therapy initiation, to primarily being driven by resistance transmitted upon infection. Resistance to antiretroviral drugs has become increasingly rare, and is now observed mostly in those with the lowest levels of cART regimen adherence. These findings show a substantial decline in ADR, and support feasibility of treatment as prevention health policies to ultimately prevent the spread of HIV/AIDS without excessive selection of resistance from this treated population. However we have also found increasing transmitted drug resistance, especially for 37the NNRTI drug class. As transmitted resistance becomes increasingly common worldwide, in settings where no resistance testing is available the final goal of 90-90-90 (where 90% of people on treatment are virologically supressed) is at risk. Changing health policies to improve access to drug resistance testing at the clinical level – especially access to baseline resistance testing, access to a variety of alternative regimens (e.g. integrase-based regimens), and ongoing funding of cART programs such as Presidents Emergency Plan for AIDS Relief (PEPFAR), could help the 90-90-90 targets to be realized. 38Chapter Three: Sociodemographic Correlates of HIV Drug Resistance and Access to Drug Resistance Testing in British Columbia, Canada3.1 Background and IntroductionIn resource rich settings with uninterrupted access to combination antiretroviral therapy (cART) and ongoing HIV-related care, human immunodeficiency virus (HIV) infection has become a manageable chronic illness with life expectancy nearing that of the general population.23,25,24 However, the negative impact of drug resistance on treatment response is well-established.97–102 The use of drug resistance testing to guide clinical decision-making has yielded improved treatment outcomes in randomized clinical trials.71,72,80,103 As a result, drug resistance testing is the current standard of care in BC and elsewhere.37,104Despite these advances, research indicates low socioeconomic status not only increases vulnerability to HIV infection, but also impedes engagement and retention of HIV-infected persons in clinical care.105–113 It is conceivable therefore that social and demographic factors associated with reduced access to HIV clinical services such as drug resistance testing could lead to elevated risks of drug resistance, and thus adverse health outcomes, in certain demographic groups. However, studies explicitly linking HIV drug resistance, and access to HIV drug resistance testing to sociodemographic factors are lacking. This research could therefore inform a more nuanced understanding of the changing HIV epidemic. The objective of this study is to examine the sociodemographic correlates of the development of drug resistance and access to drug resistance testing in a province-wide sample of HIV-positive patients receiving cART.393.2 Methods: Accessing Drug Resistance Testing Cohort3.2.1 Data CollectionIn the province of British Columbia (BC), Canada, antiretroviral (ARV) therapy is distributed through the provincial Drug Treatment Program, operated through the BC Centre for Excellence in HIV/AIDS.80 This study followed 11,801 Drug Treatment Program patients between 1996-2014. Antiretroviral medication was prescribed according to BC guidelines and was provided free of charge to the patient. Administrative data such as prescriptions and lab test results were collected on an ongoing basis until the patient was lost to follow-up by moving out of BC, passing away, or entered into a clinical trial. In these cases, patients were censored at the most recent data collected. Due to the administrative nature of the data, the University of British Columbia Research Ethics Board at Providence Health Care Research Institute waived the requirement for consent under protocol H05-50123. This study was reported in accordance with the STROBE statement.1143.2.2 Drug Resistance TestingThe BC treatment guidelines recommend testing for ARV resistance in all individuals prior to therapy as well as at virologic failure (pVL>250 copies/mL).80 In each calendar year patients were considered eligible for testing when ≥1 sample was above the lower limit of detection of the DRT assay in use at that time (usually pVL >250 copies/mL).  Patients with a physician-ordered DRT result available were considered to have accessed testing during that calendar year. Among the 11,801 patients initially considered, 9456 had eligible pVL in any year of the study. Among those eligible, a mean number of 2.1 (Q1-Q3; 0-3.0) resistance tests were ordered per patient over the course of follow-up.403.2.3 Linking of Census Data and Clinical DataPatient postal code or city of residence at time of Drug Treatment Program enrollment determined their census tract. Clinical data was linked with census tract-level data from the Census Canada Survey that was conducted closest to Drug Treatment Program enrollment date (census 1996, 2001, 2006, or 2011). If insufficient information was available to determine census tract, census metropolitan area/census agglomeration (or local health area) was used. Among patients with eligible pVL, 8398 had clinical data linked to census tract data. See Appendix XX for a comparison of characteristics of included and excluded individuals.3.2.4 Plasma Viral Load TestingTesting is free of charge in BC when ordered by a physician, and occurs regularly as part of routine patient care.80 Testing was completed at the St. Paul’s Hospital Virology Laboratory using Roche Molecular Diagnostic kits.3.2.5 Quantification of Adherence to Prescribed Drug RegimenPrescription refill percentage, obtained from Drug Treatment Program ARV prescription records and calculated as the number of days with antiretroviral drugs dispensed divided by the number of days of follow up in the first year on therapy, was used as a crude estimate of adherence and has been demonstrated as a good predictor of future adherence.115413.2.6 Statistical MethodsUnadjusted odds ratios (uOR), adjusted odds ratios (aOR), and 95% confidence intervals (CIs) were determined by Generalized Estimating Equations (GEE) logistic regression. The optimal multivariable explanatory model was selected using an Akaike Information Criterion (AIC)-based backward elimination procedure. Covariate selection was completed through backward elimination to minimize the Quasi-likelihood Information Criterion (QICu). The model was adjusted for individual-level clinical covariates and census tract-level sociodemographic covariates. All clinical covariates were treated as categorical, and included sex at birth (male, female), transmission risk factor (men who have sex with men (MSM), people who inject drugs (PWID), heterosexual), ever diagnosed with hepatitis C virus (HCV) infection, drug in first recorded regimen (only nRTI -including mono-, dual-, or triple-therapy; NNRTI as third drug in regimen; or PI as third drug in regimen), suboptimal adherence (<95% for first year on therapy), age at enrollment (years), baseline CD4 count (cells/μL), baseline pVL (copies/mL) – including baseline pVL unknown (first ARV before 1997; prior to baseline pVL testing implementation in BC), whether a patient was eligible for drug resistance test (per year), and physician experience (number of HIV patients treated in the previous two years). Models were adjusted for unknown baseline CD4, unknown baseline pVL and unknown adherence, however these are not shown in the final figure, for readability and due to low N in these categories. See Appendix XXIV for the comprehensive model.42Sociodemographic covariates were calculated at the census tract-level, including percentage of one-family households among total number of private households in the area (per 10% increment), percentage of single people (per 10%), population density (per 10k inhabitants), percentage of immigrants (per 10%), median income (per $10k), percentage with post-secondary certificate (per 10%), unemployment rate (per 10%), and percentage of census tract residents that have aboriginal ancestry (<5%, 5%-<10%, and ≥10%). Pearson correlation coefficients (PCC) were calculated for all possible pairs of census-level covariates. In cases where correlation was observed between two covariates, median income, percent of single family households, percent of immigrants, and the percent of census tract residents with aboriginal ancestry were given priority; thus avoiding multicollinearity. Data manipulation was done in SAS 9.4 (SAS Institute, Cary NC) and statistical analyses in R 3.3.1.116,1173.3 Methods: Development of Drug Resistance Cohort3.3.1 Data CollectionThe HAART (Highly Active Antiretroviral Therapy) Observational Medical Evaluation and Research (HOMER) cohort is a subset of the Drug Treatment Program (N= 11,801). HOMER participants are ≥19 year old, initially treatment naïve individuals who did not harbor baseline resistance, started HAART (defined as triple-drug combination therapy with regimens consisting of two nucleos(t)ide reverse transcriptase inhibitors (nRTIs) plus a protease inhibitor (PI), non-nucleoside reverse transcriptase inhibitor (NNRTI), or integrase inhibitor (INI)), 43between 1996-2013, had baseline pVL and baseline CD4 cell count measured within six months prior to therapy initiation, and had a minimum of one year of follow-up (N=5703).3.3.2 Clinical Data and Census LinkagesLinkage was conducted in the same manner as the access to DRT cohort described above. Within HOMER, 5175 patients had clinical data linked to census tract data. See Appendix XXIII for a comparison of included and excluded individuals. Plasma viral load testing and adherence estimation were conducted in the same manner as the access to DRT cohort described above.3.3.3 Drug Resistance TestingResistance testing in BC is provided at no cost to the patient. Drug resistance genotyping was attempted on all available plasma samples; these included physician-ordered tests and genotyping performed for research purposes. HIV RNA was extracted and genotyping of the protease and reverse transcriptase genes was performed as previously described (mean 1.5; Q1-Q3: 0-2.0 resistance tests/patient).31,61,118 Resistance was defined as the presence of ‘key’ resistance mutation from the IAS-USA (2013) mutation list.119 Detection of any resistance was considered the endpoint and patients were censored after the date of the positive resistance test. Patients without baseline resistance tests were assumed not to harbor resistance. Integrase resistance was not included since no individuals in this study initiated therapy with integrase inhibitors.443.3.4 Statistical MethodsUnadjusted hazard ratios (uHR), adjusted hazard ratios (aHR), and 95% confidence intervals (CI) of time to any drug resistance among those with eligible pVLs were determined by Cox proportional hazards (PH) regression. Covariate selection was completed through backward elimination to minimize the Quasi-likelihood Information Criterion (QICu). Violation of the proportional hazard assumption was tested using the global test for proportionality.Covariates that were accounted for in the development of drug resistance model were identical to the access to DRT model, with the exception of the following individual-level clinical covariates: eligibility for drug resistance test (per year) was replaced with year of therapy initiation, drug in first recorded regimen (Only nRTI including mono-, dual-, or triple-therapy; NNRTI as third drug in regimen; or PI as third drug in regimen) was replaced with third drug in baseline regimen (NNRTI, or PI), and within the baseline pVL category; baseline pVL unknown (first ARV before 1997) was removed as baseline viral load was a HOMER inclusion criterion. The comprehensive model can be found in Appendix XXV. The procedure described in the access to DRT cohort statistical methods section was followed for covariate multicollinearity.3.4 Results: Access Drug Resistance Testing Cohort3.4.1 Cohort DescriptionBetween 1996-2014, 11,801 HIV-positive patients were observed longitudinally; 9,456 had ≥1 pVL test above the limit of detection for DRT (usually pVL >250 copies/mL) after therapy initiation in that time period, and were therefore eligible for a physician-ordered DRT. 45Due to missing data, census data could not be linked to clinical data for 1058 participants, resulting in a final tally of 8398 individuals included in this analysis. Baseline characteristics of those included were found to be statistically different from those excluded, with the exception of sex at birth (p=0.078); see Appendix XX. Most patients were male at birth (82%), and over one third were MSM (34%), PWID (35%), or co-infected with HCV (37%). The majority was ≥95% adherent to the prescribed drug regimen during the first year of therapy (54%), and half had ever received a physician-ordered DRT (49%). The median age at study entry was 40 years (1st quartile-3rd quartile; 33-47). See Table 3.1 for complete description of clinical characteristics. See Table 3.2 for census-tract level sociodemographic characteristics. Among participants eligible for a drug resistance test, the proportion of those who accessed testing increased gradually over the course of the study (Appendix XIII), from 29% in 1996 to 54% in 2013.46Table 3.1 Clinical Baseline Characteristics.Baseline Clinical Characteristics Accessing DRT Any Drug ResistanceN (%) 8398 (100) 5175 (100)Sex at birth     Male - n (%) 6916 (82) 4232 (82)     Female - n (%) 1482 (18) 943 (18)MSM risk     No - n (%) 3501 (42) 2303 (45)     Yes - n (%) 2842 (34) 1491 (29)     Unknown - n (%) 2055 (25) 1381 (27)Heterosexual risk     No - n (%) 4108 (49) 2380 (46)     Yes - n (%) 1773 (21) 1137 (22)     Unknown - n (%) 2517 (30) 1658 (32)PWID risk     No - n (%) 3852 (46) 2325 (45)     Yes - n (%) 2919 (35) 1888 (36)     Unknown - n (%) 1627 (19) 962 (19)Hepatitis C positive     No - n (%) 4266 (51) 2734 (53)     Yes - n (%) 3132 (37) 2109 (41)     Unknown - n (%) 1000 (12) 332 (6)Baseline CD4     <200 cells/μL - n (%) 3273 (39) 2310 (45)     200-<350 cells/μL - n (%) 2516 (30) 1558 (30)     ≥350 cells/μL - n (%) 2505 (30) 1307 (25)Baseline Viral Load    <10,000 copies/mL - n (%) 958 (11) 553 (11)    10,000-<100,000 copies/mL - n (%) 2865 (34) 2118 (41)    ≥100,000 copies/mL - n (%) 3043 (36) 2504 (48)Baseline regimen third drug class    PI - n (%) 3679 (44) 3064 (59)    NNRTI - n (%) 2472 (29) 2111 (41)    nRTI Only – n (%) 2077 (25) N/A    Other – n (%) 170 (2) N/AAdherence in first 12 months of therapy <95%     No - n (%) 4543 (54) 3215 (62)     Yes - n (%) 3550 (42) 1960 (38)Patients ever having a drug resistance test     No - n (%) 4271 (51) 2904 (56)     Yes - n (%) 4127 (49) 2271 (44)Median year of ARV initiation (Q1-Q3) 2002 (1997-2008) 2005 (2000-2009)Median age at ARV initiation in years (Q1-Q3) 40 (33-47) 34 (31-48)47Table 3.2 Census Tract-Level Sociodemographic Characteristics Sociodemographic Census Tract-Level CharacteristicsAccessing DRTMedian (Q1-Q3)Any Drug Resistance Median (Q1-Q3)Percentage of single-family households 51 (30-66) 53 (30-66)Population density (per 10K) 5150 (2560-10,800) 5150 (2500-10,500)Percentage of immigrants 31 (21-39) 31 (21-40)Median income ($) 23 200 (19 200-27,300) 24 300 (20,000-28,000)Percentage of single people 38 (30-54) 37 (30-54)Percentage with post-secondary certification 53 (44-64) 54 (44-65)Percentage unemployed 63 (56-67) 63 (56-68)Percentage aboriginal ancestry 3.0 (1.0-5.0) 3.0 (1.0-6.0)3.4.2 Odds Ratios of Correlates for Accessing Drug Resistance TestingThe univariable and multivariable ORs for the covariates (see methods) included in GEE logistic regression are presented in Figure 3.1. Pearson correlation coefficients (PCC) were calculated for all possible pairs of census-level covariates. Strong correlation was observed between percent of single family households and percent of single individuals (PCC=-0.85), median income and post-secondary education (PCC=0.74), median income and unemployment rate (PCC=0.72), post-secondary education and unemployment rate (PCC=0.67). Moderate correlation was observed between population density and percent of single individuals (PCC=0.62). As determined prior to running the model, median income, percent of single family households, percent of immigrants, and the percent of census tract residents with aboriginal ancestry were given priority; thus post-secondary certificate, unemployment rate, and single people were excluded from the final model to avoid multicollinearity. Among the sociodemographic covariates as determined from census tract-level data, every $10k increase in median income was associated with 17% lower odds ratio for accessing drug resistance testing (aOR: 0.83, 0.77-0.89), while census tracts with 5%-<10% of people reporting aboriginal 48ancestry were associated with 15% lower odds ratio for accessing drug resistance testing when compared regions with <5% of people self-reporting aboriginal ancestry (aOR: 0.85, 0.76-0.95).Figure 3.1 Odds Ratio and 95% Confidence Intervals Assessing the Odds of Accessing Drug Resistance Testing, by Model. Odds ratios do not exist for covariates not selected under the multivariable model (see methods).49Among the individual-level clinical covariates, odds ratios associated with access to drug resistance testing were 20% higher for women compared to men (aOR: 1.2, 1.1-1.3), 53% lower for those with unknown PWID status (aOR: 0.47, 0.42-0.53), and 50% higher when baseline CD4 was <200 cells/μL (aOR: 1.5, 1.3-1.6). Similarly they were 20% higher for those with baseline CD4 of 200-<350 cells/μL (aOR: 1.2, 1.1-1.3), 80% higher when baseline pVL was unknown due to ARV initiation before pVL testing becoming available in 1997 (aOR: 1.8, 1.5-2.1), 40% higher among those with first recorded regimen as mono-, dual-, or triple-nRTIs (aOR: 1.4, 1.3-1.6), and finally, 30% higher for those with adherence <95% (aOR: 1.3, 1.2-1.4). Baseline pVL >100,000 copies/mL (aOR: 1.2, 1.1-1.4) and physicians with unknown amount of experience treating HIV (aOR: 0.68, 0.58-0.80) were significant only after adjusting for other covariates.Low adherence has been associated with low socio-economic status as well as PWID and the development of drug resistance, thus accounting for adherence could theoretically mask effects of these covariates.120,121 However, we found that accounting for <95% adherence did not substantially change the results of the model, therefore adherence was included (see Appendix XVIII).3.5 Results: Development of Drug Resistance Cohort3.5.1 Cohort DescriptionData on 5175 eligible individuals (see methods) between 1996-2014 was analyzed retrospectively. Baseline characteristics of those excluded were found to be statistically different from those included (see Appendix XXIII). Similar to the access to DRT cohort described above, 5082% of patients were male at birth, and 36% were PWID. Only 29% were MSM, while 41% were co-infected with HCV. Overall 62% were ≥95% adherent, but only 44% had ever been ordered a drug resistance test by a physician. The median age was 41 years (1st quartile-3rd quartile; 34-48). See Table 3.1 for complete description of clinical characteristics. See Table 3.2 for complete description of census-tract level sociodemographic characteristics.3.5.2 Hazard Ratios of Correlates for Development of Drug ResistanceThe univariable and multivariable HRs for the covariates in the Cox PH regression are presented in Figure 3.2. The global test for proportionality gave a non-significant p-value (p=0.11), also the interaction between time and each individual variable gave non-significant p-values ranging from 0.072 to 0.84, showing no evidence the proportional hazard assumption was violated. Pearson correlation coefficients (PCC) results were the same as the access to resistance testing model, and the same variables were excluded to avoid multicolinearity. A higher proportion of residents reporting aboriginal ancestry was the only census tract-level sociodemographic covariate significantly associated with the development of drug resistance. Individuals in a census tract with ≥10% people self-reporting aboriginal ancestry were 20% more likely to develop drug resistance compared to regions with <5% self-reported aboriginal ancestry (aHR: 1.2, 1.1-1.5).51Figure 3.2 Hazard Ratio and 95% Confidence Intervals Assessing the Likelihood of Developing Drug Resistance in a Treatment Naïve Cohort Subset, by Model. Hazard ratios do not exist for covariates not selected under the multivariable model (see methods).Among individual-level clinical covariates, the likelihood of developing drug resistance was 90% higher in individuals with baseline CD4 <200 cells/μL (aHR: 1.9, 1.6-2.3) and 30% higher for those were 200-<350 cells/μL (aHR: 1.3, 1.1-1.6) compared to those with ≥350 cells/μL, 100% higher among those with baseline pVL >100,000 copies/mL compared to those 52with <10,000 copies/mL (aHR: 2.0, 1.6-2.6), 30% higher in PWIDs (aHR: 1.3, 1.1-1.5), and 120% higher among those with <95% adherence (aHR: 2.2, 1.9-2.5). More recent ARV initiation was associated with decreased likelihood of developing drug resistance. Individuals who initiated therapy between 2000-2003 were 18% less likely to develop drug resistance compared to those who started between 1996-1999 (aHR: 0.82, 0.71-0.95); whereas, participants who initiated therapy between 2004-2007 (aHR: 0.52, 0.44-0.62), and 2008-2013 (aHR: 0.44, 0.36-0.53) were 48% and 56% less likely to develop drug resistance, respectively. Baseline CD4 200-<350 cells/μL, and those who first started ARVs between 2000-2003 became significant only after adjusting for other covariates. These results were broadly consistent when broken down by categories of drug resistance; emtricitabine/lamivudine (3TC/FTC) (Appendix XIV), non-nucleoside reverse transcriptase inhibitors (NNRTI) (Appendix XV), other nucleoside reverse transcriptase inhibitors (nRTI) (Appendix XVI), and protease inhibitors (PI) (Appendix XVII). Similar to the access to DRT model, adherence was tested as a confounder. Accounting for those who had <95% adherence did not substantially change the results (see Appendix XIX), therefore adherence was included in the final model.3.6 DiscussionThis longitudinal observational study describes an exploratory analysis of sociodemographic correlates of drug resistance and access to drug resistance testing among patients receiving ARVs in BC between 1996-2014. The sociodemographic variables were determined using census tract-level data, while clinical variables used individual-level data. The results indicate that living in census tracts with high median income and high rates of aboriginal 53ancestry remain weakly correlated with decreased access to drug resistance testing after adjusting for clinical factors. However, only census-tracts reporting higher proportion of aboriginal ancestry had an elevated likelihood of developing HIV drug resistance. The combined census tract-level sociodemographic results indicate census-tracts with a high proportion of aboriginal ancestry may benefit from targeted public health interventions, such as more DRT-specific physician training or public awareness campaigns regarding HIV drug resistance.The difference in effect size between sociodemographic and clinical covariates associated with access to drug resistance testing was notable; odds ratios of clinical covariates had up to 4 times stronger effect size than sociodemographic covariates. A similar disparity was observed in development of drug resistance; clinical covariates had between 4.5-6 times larger effect size than the strongest sociodemographic covariate. The results suggest that at the individual-level, clinical correlates are better determinants for the development of drug resistance or access to drug resistance testing in BC compared to census tract-level sociodemographic covariates. More recent ARV initiation was associated with decreased likelihood of developing drug resistance. This is likely to do multiple factors, potentially including patient management, an improvement in cohort-wide adherence, and improvements in regimens resulting in greater genetic barriers to resistance. Research expanding upon this finding has recently been published, but more work is required to determine causation.122 PWIDs did not access resistance testing significantly more than non-PWIDs in this study. This result is counterintuitive and could result from effect modification, as imperfect adherence, 54development of drug resistance, and low median income have been previously associated with PWID.122–124 Imperfect adherence has been strongly associated with the development of acquired drug resistance, leading to elevated viral load; the BC primary guidelines for the treatment of HIV recommends a drug resistance test be ordered under these circumstances.31,37,122 Therefore adherence and drug resistance are on the causal pathway for drug resistance testing. To test for effect modification, the models were stratified by PWID status: No strong differences were found between PWID and non-PWID (see Appendix XXI and Appendix XXII).Aboriginal ancestry was associated with moderately lower access to testing as well as higher likelihood of developing drug resistance.125 These results highlight the complex ways indigenous peoples interact with HIV-related health care, and more work is needed to address barriers to HIV care and related co-morbidities indigenous peoples face in Canada. Living in a census-tract with high median income was also marginally associated with reduced access to DRT. This finding does not support our hypothesis that low socioeconomic standing is associated with barriers to care. This is counter to results that have been reported in other areas of HIV research.110–113,126Nosyk et al (2013) observed access to HIV related care in BC has improved between 1996-2011; infected, undiagnosed HIV cases dropped 36-58%, while viral suppression increased from 1% to 35%, and individuals who are fully adherent but not virally suppressed dropped from 95% to 22%.127 Our research implies these improvements in care access, retention, and improvements in cART treatments occurred relatively equally over different sociodemographic strata. Identifying potential causes of these results require further, more detailed study. While no 55studies to our knowledge have directly investigated sociodemographic correlates of HIV drug resistance, previous studies have reported no strong association between MDR-TB and sociodemographic covariates in both high income and low income settings.128–132 A limitation of this study was missing census data –as many as 373 patients had census data linked, but had incomplete or absent data for a given census question resulting from the change to a voluntary National Household Survey in 2011. Due to the low quality of data collected from a voluntary survey some data was deemed not fit for release, or flagged to be used with caution.133 Additionally, those excluded were statistically different from those included in both cohorts of this study, which may bias the results. Drug resistance monitoring was not included in the clinical guidelines until the early 2000s, potentially skewing the results towards low access to resistance testing.134,135 The extrapolation of results are limited to settings where testing is free of charge. Since sociodemographic data was gathered at the census tract level, there exists potential for ecological fallacy; this should be born in mind when interpreting the results. Especially in areas of vast socioeconomic disparities, such as the downtown eastside of Vancouver, variables have a decreased likelihood of being accurately descriptive of the population in the region, thus confounding the results. Ecological fallacy does not apply for clinical variables, as this data was gathered for each individual patient. Our results suggest regions with high proportion of people with aboriginal ancestry may experience increased barriers to accessing testing and be at higher risk of developing resistance. Working with indigenous communities to promote culturally sensitive and appropriate programs could help to reduce some of these barriers. However, vast disparities in drug resistance were not 56observed between sociodemographic strata. The complex relationships between sociodemographic covariates make statistical modeling a challenge, and therefore readers should be cautious when interpreting and extrapolating these results to real world contexts. While evidence exists linking infectious diseases with poor socio-economic standing, there are few examples for drug resistant variants of HIV. Clinical covariates, particularly low CD4 cell count, high pVL, and imperfect adherence continue to be effective individual-level predictors of developing drug resistance and access to drug resistance testing across all sociodemographic strata.57Chapter Four: The Future of HIV-Related Genetic Testing; a Next Generation Sequencing Assay Validation and Proof of Principle4.1 Background and IntroductionAbacavir is an nRTI, first approved for use within Canada in 1999, and is on the 2015 World Health Organization (WHO)’s List of Essential Medicines.136,137 Abacavir is currently available in several combination tablets, such as Triumeq, Kivexa, and Trizivir, or available on its own for customized combination therapy.138 Between 2000 and 2010, approximately 18% of treatment naïve individuals across Canada initiated therapy on abacavir-containing regimens.139 Abacavir hypersensitivity is an immune response to abacavir antiretroviral medication, characterized by fever, rash, fatigue, malaise, muscle and joint pain, swollen lymph nodes, inflammation and ulceration of the mucosal layers lining the gut, inflammation of the lungs, the heart, and the liver, interstitial nephritis, atypical lymphocytosis, and eosinophilia.140 Continuation or re-challenge with abacavir after symptoms of hypersensitivity elevates the risk of a serious systemic and potentially fatal immune reaction.140 Previous studies demonstrated that HLA-B*57:01, an allele of the Major Histocompatibility Complex (MHC) on human chromosome 6, has a high positive predictive value for abacavir hypersensitivity.141 The prevalence of HLA-B*57:01 in the North American population is estimated to be between 2-8%.142 Genetic testing for this allele is cost effective and can be used to guide treatment choices, thereby reducing the incidence of abacavir hypersensitivity reactions by avoiding use in those with HLA-B*57:01.58The preferred method for HLA-B*57:01 typing in recent years has been Sanger sequencing.143 However due to inherent limitations of population-level sequencing, Sanger sequencing is often unable to distinguish between HLA-B*57 subtypes without additional testing. The MiSeq method is able to generate full exon sequence as it exists on independent strands of chromosome 6. The HLA interpretation algorithm is then able to analyze each exon as a whole, as opposed to each unique base. This allows the MiSeq to better distinguish between HLA-B*57 subtypes and gives results with up to six digits of resolution, where digits of resolution refers to the number of digits as per the IPD-IMGT/HLA Database nomenclature.144 Through the comparison of Sanger sequencing and MiSeq sequencing results, our intent was to validate a more precise and rapid HLA-B typing method for use in clinical abacavir screening. 4.2 Methods4.2.1 Data CollectionThe BC Centre for Excellence in HIV/AIDS Laboratory completes HLA-B*57:01 typing for HIV infected individuals for all Provinces within Canada, with the exception of Quebec, and the Yukon. After testing using Sanger sequencing, the remaining samples were stored indefinitely at -20°C. These archived samples were reused for this validation, and include previously tested clinical samples of human whole blood combined with anticoagulent; EDTA, ACD, or heparin; and proficiency testing panel samples in the form of nucleic acid extracts (College of American Pathologists External Quality Assurance/Proficiency Testing, University of California at Los Angeles International HLA DNA Exchange, Royal College of Pathologists of Australia Quality Assurance Program). Note that some samples were included more than once as they were tested individually, but at multiple time points for different aspects of the validation 59(e.g. inter- and intra- variability testing). Extracts from previous tests were used when possible, and were also stored at -20°C. In the absence of available DNA extract, whole blood was re-extracted. Ethical approval for use of residual clinical and research specimens for assay development was granted by the University of British Columbia Research Ethics Board at St. Paul’s, Providence Health Care site (H04-50276).4.2.2 SequencingNucleic acid was extracted from previously frozen whole blood using standard procedures on the NucliSENS easyMag (bioMérieux, St Lauren, Canada). Water was included in each run as a negative control. Sanger sequencing was completed in accordance with standard protocols, as described previously (section 1.5.1). In MiSeq sequencing, the DNA extract was amplified by the first round PCR reaction using HLA-BF_mod forward primer (GGGAGGAGMRAGGGGACCSCAG) and HLA-BR reverse primer (GGAGGCCATCCCCGGCGACCTAT), producing an 1869 base pair (bp) amplicon. The amplicon included exon 2, exon 3, and the intervening intron (Figure 4.1A). This 1869bp amplicon was then used as a template in a second round of PCR, where exon 2 and exon 3 were amplified separately (Figure 4.1B). 60Figure 4.1 First and Second Round PCR MapsA) First round of PCR reaction – schematic of primer alignment. B) Second round of PCR reaction – schematic of primer alignment. The second round PCR primers contained Illumina-generated adaptor regions, which include sample-specific and direction-specific indexes (Figure 4.2). One of the adaptor regions contained the compliment of the Illumina sequencing primer, however this information was proprietary. During second round PCR amplification, these indexes were added to both ends of the exon 2 and exon 3 HLA-B amplicons. Indexing was crucial to tracking each sample during data analysis.Figure 4.2 Second PCR Primer DesignIllustrates the standard primer design for Illumina MiSeq amplicon sequencing. 61The second round PCR products (also known as tagged amplicons) were purified and normalized using magnetic beads (Agencourt® AMPure XP) to remove PCR reagents, and to ensure that an equal amount of DNA for each sample was loaded on the sequencing machine. This was done by adding PCR product to a specific amount of beads, saturating them with tagged DNA amplicons. Each bead bound to the same amount of DNA.  The beads were then pooled and washed to remove PCR reagents and excess DNA. The tagged DNA amplicons were eluted from the beads with water. As the same amount of beads was used for each sample, equal amounts of DNA from individual samples contribute this pooled DNA library. The purified exon 2 and exon 3 amplicon libraries were both quantified using the Invitrogen Quant-iT PicoGreen® dsDNA Assay and diluted to the appropriate concentration for loading on to the MiSeq (1.0 ng/uL). The 1.0 ng/uL exon 2 and exon 3 libraries are combined in a 4:1 ratio (exon 2: exon 3). For cost saving purposes, every MiSeq run described here included several unrelated genetic tests, including some or all of the following; hepatitis C (HCV) drug resistance testing in NS5a, NS5b, or whole genome; HIV drug resistance or tropism testing in protease, reverse transcriptase, V3-Loop, or integrase genes; or other tests that have been prepared in a similar manner for MiSeq sequencing. The libraries were stored at -20°C until sequenced on the MiSeq. Just before sequencing, the libraries were denatured with sodium hydroxide, diluted into hybridization buffer and loaded onto a MiSeq cartridge. MiSeq sequencing has a run time of approximately two days, after which the data was analyzed using an in-house pipeline.624.2.3 Data AnalysisThe HLA MiSeq interpretation algorithm used here was an in-house software program designed to process the raw short read data generated by the MiSeq. The HLA MiSeq interpretation algorithm analyzes short-read sequences by first trimming low quality bases from the end of the reads, aligning reads together to create one merged sequence, and applying a strict set of criteria on these merged sequences. These criteria include: perfect agreement between reads in the overlapping regions, a minimum of three other identical merged sequences must exist in the run, and the merged sequence must align to either exon 2 or exon 3 of the HLA standard sequence (HLA-B*07:02:01). When merged sequences do not match any of the above criteria, they are excluded from downstream analysis. All remaining merged sequences are then compared to the merged sequence with the highest read count, called the top merged sequence. Only the merged sequences that have >30% the number of reads as the top merged sequence are kept for further analysis.  These remaining sequences are compared to World Health Organization recognized HLA allele sequences on the IMGT/HLA database to generate the final HLA allele interpretations using standard nomenclature.Consensus sequence results obtained using the new MiSeq method and the previously validated Sanger method were compared. Complete concordance was considered to be present when the base call for each position in the sequence was the same; for instance, the base call was A by both methods (A  A), or the potential amino acid calls for a position was the same for both methods ([A, D, K]  [A, D, K]). Partial concordance was considered when one base call was a mixture and the corresponding base call was one of the components of the mixture, for example A  R (A or G), or [A, D, K]  [A, K]. This was considered a compatible difference. 63Complete discordance, or an incompatible difference, was considered present when each base call was a different, non-ambiguous nucleotide at the same position, for example A  G, or [A, D, K]  [A, M, K].Four comparison metrics were calculated for each component of the validation, each of which compared concordance of the same sample between methods for 1) all nucleotides and 2) all amino acids, 3) HLA allele interpretations to the fourth digit of resolution where Sanger four-digit resolution was available (using the standard nomenclature as determined by the World Health Organization Nomenclature Committee for Factors of the HLA System), as well as 4) the primary outcome; HLA-B*57:01 positive or HLA-B*57:01 negative. The 95% CIs were determined in R using the exact binomial test under the alternative hypothesis that the true probability of success is not equal to 1.0. The acceptable level of concordance was arbitrarily predefined as one nucleic acid difference in each HLA allele sequence (1bp difference out of the 546bp allele amplicon generates a concordance of 99.8%), or ≥99.8%.1454.2.4 Accuracy Accuracy is a broad term that is used to describe the amount of agreement between a new test method (MiSeq) and a reference method (Sanger).146 There were 396 samples (as described in section 4.2.1) included in this analysis. This is well over the recommended minimum of 50 samples.147,148 Samples were compared between methods using the methods described in the data analysis section (4.2.3). All the samples used in the repeatability and reproducibility assessments were included in this accuracy analysis. 644.2.5 RepeatabilityTo test intra-assay variability, 12 replicates of seven samples (96 well plate with a negative control) heterozygous for HLA-B*57:01 were prepared at the same time and run on the MiSeq all at once. In order to have enough extract volume, samples were extracted twice. For each of the seven samples, the nucleic acid and amino acid concordance of the 12 sample replicates was compared in R using Muscle alignment method in the msa package. Similarly, the HLA-B*57:01 concordance, and the four digits of resolution concordance, were compared using excel.4.2.6 Reproducibility Inter-assay variability was determined using 24 samples plus a negative control. The same sample set was run in five separate batches by five separate lab staff. Lab staff comprised of a range of experience levels, both in laboratory science as a whole, as well as in regards to the HLA-B MiSeq method. These runs shared the same nucleic acid extract, but were run separately during PCR, tagging, and sequencing. Of the 24 samples, 12 were HLA-B*57:01 heterozygous, and 12 were HLA-B*57:01 negative. Among the 12 samples that were HLA-B*57:01 negative, two were homozygous, and 10 were heterozygous. An individual alignment for each sample across the five inter-variability runs was completed. The nucleic acid and amino acid multiple sequence alignments were completed in R using Muscle alignment method in the msa package. Similarly an individual comparison of the HLA-B*57:01 outcome, and the four digits of resolution outcome, were completed for each sample across the five inter-variability runs using Microsoft Excel 2011.654.2.7 Cross-Contamination To determine how cross-contamination between samples would affect the results, extracts were chosen and deliberately combined. Combined extracts were heterozygous, and neither extract shared a common allele. The extracts were quantified using the Invitrogen Quant-iT PicoGreen® dsDNA Assay and diluted to 1500ng/mL. Unmixed samples (N=14) were included as positive controls, and from those 14 samples seven sample “pairs” were combined in the following ratios; 1:1, 1:3, 3:1, 1:6, 6:1, and run in duplicate. The sample with the larger ratio was called the ‘majority sample’. Water was used as the negative control. Samples that passed quality control checks within the HLA Interpretation Algorithm were compared. The majority sample in each mixture as well as the positive controls were compared to their Sanger sequences as described in the Data Analysis section (4.2.4), then the majority sample in each mixture was compared to its unmixed positive control for nucleic acid concordance.After these quality checks were complete, a different set of criteria was applied. A mixture was said to have given a ‘compatible result’ when the majority sample in the mixture was called, or when a sample with equal proportions of two extracts was failed by the algorithm (i.e. ‘too many alleles’). A sample was not considered to have a ‘compatible result’ if it failed to report a result due to other quality control errors, including ‘not enough alleles’ or ‘incomplete heterozygous’, or reported a result from the 1:1 mixture or the minority sample of any mixture. 4.2.8 Application of HLA Assay: A Proof of Principle StudyAmong HIV positive mothers in Canada, 95.5% were receiving cART treatment as of 2013.149 HIV positive women on cART have 1.4-fold increased odds for preterm delivery and 66two-fold increased odds of severe preterm delivery when compared to women on monotherapy.150 Given these findings, a deeper look in to the association of cART and pregnancy was warranted. It has been shown that abacavir can cross the placenta.151 Mothers taking abacavir will have been screened as HLA-B*57:01 negative, however the allele can also be passed to the infant from the father. Since the vast majority of HIV positive mothers in Canada are on cART, and abacavir is a very common first line antiretroviral, it is possible that HLA-B*57:01 positive foetuses have been prenatally exposed to abacavir, and foetal abacavir hypersensitivity may influence preterm births. MiSeq HLA typing was previously performed on whole blood samples from 365 HEU children in the Canadian cohorts (BC Pregnancy, Centre Maternel et Infantile sur le SIDA mother–infant cohort at Centre Hospitalier Universitaire Sainte-Justine, Children and Women: Antiretroviral therapy and Markers of Aging), however only 0.82% (3/365) were found to be HLA-B*57:01 positive and thus the study was underpowered to detect any association between HLA type, abacavir exposure and pre-term birth. Typically within the North American population, this allele occurs in 2-8% of the population, therefore this finding was notable.142 This study also collected the self reported ethnicities of the parents for each child. In an attempt to explain the results, the self reported ethnicities were used to estimate the expected allele frequency of HLA-B*57:01 in the children. Data available on the NCBI dbMHC database (maintained in cooperation with the Medical University Graz, Austria) associated allele frequencies with ethnic populations, and was used for this estimation.152 However, due to the lack of data for certain alleles, the comparison was only for the first two digits of resolution, specifying allele group. Figure 4.3 shows the observed allele group frequency of this study, compared against the expected allele group frequency estimated 67from the self-reported ethnicity of the parents. The observed and expected allele frequencies were compared for each allele using Fisher’s exact test. Figure 4.3 Observed and Expected Allele Group FrequencyThe observed allele group frequency was indicated in grey, while the expected allele group frequency was blue. Fisher’s exact test was used to determine statistical difference between the observed alleles (N=730) and expected results (N=646). The difference between Ns is due to rounding error when calculating expected allele frequency of the children based on the parents’ ethnicity. * Denotes allele groups with a p-value<0.05. Similar to the majority of alleles found in this study (90%) the difference between observed and expected allele group frequency for HLA-B*57 was not statistically significant (p=0.72), however B*50, B*49, and B*44 were found to be overrepresented. Since this 68comparison looked only at allele group and not four digits of resolution, the low HLA-B*57:01 frequency was not directly addressed. Therefore a new hypothesis was proposed: foetuses with HLA-B*57:01 from HIV infected mothers were being selected against in utero, resulting in extremely early pregnancy termination and a much lower frequency of HLA-B*57:01 than expected in this population. To test this theory and confirm the previous finding was not due to random chance, another cohort of children (N=241) potentially exposed to HIV in utero, during birth, or through breast milk from HIV infected mothers but remained uninfected were HLA typed, and the allele frequency was determined. The observed and expected alleles were also compared, as described previously, with the exception that the expected allele frequency was calculated based on the mother’s self-reported ethnicity alone due to missing paternal data. 4.3 Results4.3.1 Accuracy Table 4.1 Accuracy Analysis: MiSeq-Sanger Comparison Summary Comparison between gold standard Sanger sequencing and newly developed MiSeq sequencing for HLA-B exons 2 and 3. Concordance of nucleotides, amino acids, four digits of resolution and HLA-B*57:01.Number of samples passing QC (N=451) 396 HLA-B*57:01 (positive/negative) concordant (95% CI) (N=396) 100% (99.1-100%)4-digit resolution concordance (N=287) 100% (98.7-100%)Total nucleic acid number 216,216Nucleotide differences 142Incompatible nucleotide differences 28Nucleic acid concordance (N=216,216) 99.9% (99.9-99.9%)Total amino acid number 72,072Amino acid differences 97Incompatible amino acid differences 26Amino acid concordance (N=72,072) 99.9% (99.8-99.9%)69The MiSeq method had 100% (95% CI: 98.6-100%) sensitivity and 100% (95% CI: 97.2-100%) specificity in determining the primary outcome: HLA-B*57:01 positive or HLA-B*57:01 negative when using Sanger sequencing results as the gold standard. Similarly, the MiSeq method was 100% (95% CI: 99.1-100%) concordant when comparing HLA-B*57:01 results, 100% (95% CI: 98.7-100%) concordant when comparing four-digit resolution, 99.9% (95% CI: 99.9-99.9%) concordant when comparing nucleic acids, and 99.9% (95% CI: 99.8-99.9%) concordant when comparing amino acids. The nucleic acid comparison is described in detail in Figure 4.4, highlighting compatible and non-compatible differences as described in section 3.2.3. Compatible differences made up 80% (114/142) of all nucleic acid differences between methods, the most common included K (Sanger) being called as a G or a T (MiSeq), a T (Sanger) being called as a W (MiSeq), and a C (Sanger) being called as an M (MiSeq).Note that only samples that passed the HLA MiSeq interpretation algorithm quality checks were included in this analysis (N=396). Excluded from the analysis were 55 (12%) samples that failed quality control. The reason for failure in 38 (69% of failures) samples was 'not enough reads'. Among the samples that failed (N=55), 24 (44%) contained an HLA-B*57:01 allele, whereas among those that passed (N=396) 266 (68%) contained a HLA-B*57:01 allele. One sample was sequenced independently eight times, and every time the quality warning 'possibly incomplete' was given alongside an allele interpretation. The allele interpretation given for this sample was the same to six digits of resolution in each of the eight samples. It is unclear why this particular sample generated a quality warning each time, and the allele combination has been identified as unique (HLA-B*35:01:01G-HLA-B*58:01:01G) among this sample set; occurring at an estimated rate of 0.05% in the North American population among those of 70European decent.153 More data is needed and this allele combination has been flagged for further analysis should another be tested in the future. Due to the lack of false negatives and false positives in these results, sensitivity and specificity could not be calculated. Figure 4.4 Accuracy Analysis: Detailed Concordance ComparisonThe differences between nucleic acids determined by Sanger sequencing, and newly developed MiSeq sequencing for HLA-B exons 2 and 3 (N=396) are described in detail.4.3.2 RepeatabilityFor each set of unique sample repeats, the nucleotide sequences that passed the quality control checks were 100% (95% CI: 99.8-100%) concordant (See Table 4.2). Similarly, for each set of repeats the amino acid sequences were 100% concordant (95% CI: 99.3-100%), four-digit resolution was 100% (95% CI: 29.2-100%) concordant, and HLA-B*57:01 outcome was 100% (95% CI: 29.2-100%) concordant. 71Table 4.2 Repeatability Sequence AnalysisSample IDReplicates passing QCHLA-B*57:01 concordance (95% CI)4 digit resolution concordance (95% CI)Number of nucleic acids Nucleic acid concordance (95% CI)Number of amino acids Amino acid concordance (95% CI)1 3 100% (29.2-100%)100% (29.2-100%) 1,638100% (99.8-100%) 546100% (99.3-100%)2 12 100%(73.5-100%)100%(73.5-100%) 6,552100% (99.9-100%) 2,184100% (99.8-100%)3 12 100%(73.5-100%)100%(73.5-100%) 6,552100% (99.9-100%) 2,184100% (99.8-100%)4 12 100% (73.5-100%100% (73.5-100% 6,552100% (99.9-100%) 2,184100% (99.8-100%)5 11 100%(71.5-100%)100%(71.5-100%) 6,006100% (99.9-100%) 2,002100% (99.8-100%)6 12 100%(73.5-100%)100%(73.5-100%) 6,552100% (99.9-100%) 2,184100% (99.8-100%)7 11 100%(71.5-100%100%(71.5-100% 6,006100% (99.9-100%) 2,002100% (99.8-100%)Only samples that passed the HLA MiSeq interpretation algorithm quality control checks were included in this analysis. Excluded from the analysis were 11 (13%) samples that failed quality control. Of these failures, nine were from the same sample and had the same reason for failing: ‘Not enough reads’. This result suggests that this particular sample might have been compromised or the nucleic acids were slightly degraded, resulting in a low amount of usable DNA. All seven samples used were heterozygous for HLA-B*57:01. All 12 replicates from the negative control failed due to no reads detected, as expected. 4.3.3 ReproducibilityConcordance of each of the 24 samples was compared across all five runs (see Table 4.3) for each outcome described in the Data Analysis section: HLA-B*57:01 concordance, four digit resolution concordance, nucleic acid concordance and amino acid concordance. All were 100% concordant (see Table 4.3 for 95% CIs). 72Table 4.3 Reproducibility Analysis: Concordance ResultsSample IDReplicates Passing QCHLA-B*57:01 concordance (95% CI)4 digit concordance (95% CI)Number of nucleic acidsNucleic acid concordance (95% CI)Number of amino acidsAmino acid concordance (95% CI)1 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)2 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)3 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)4 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)5 4 100%(39.8-100%)100%(39.8-100%) 2184100%(99.8-100%) 728100%(99.5-100%)6 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)7 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)8 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)9 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)10 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)11 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)12 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)13 2 100%(15.8-100%)100%(15.8-100%) 1092100%(99.7-100%) 364100%(99.0-100%)14 3 100%(29.2-100%)100%(29.2-100%) 1638100%(99.8-100%) 546100%(99.3-100%)15 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)16 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)17 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)18 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)19 4 100%(39.8-100%)100%(39.8-100%) 2184100%(99.8-100%) 728100%(99.5-100%)20 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)21 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)22 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)23 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)24 5 100%(47.8-100%)100%(47.8-100%) 2730100%(99.9-100%) 910100%(99.6-100%)734.3.4 Cross-ContaminationFor all samples that passed HLA MiSeq interpretation algorithm QC (N=65), the product of the interpretation algorithm was the majority sample in the mixture. The majority sample sequences were compared to their Sanger sequence as described in the Data Analysis section (4.2.3). Overall, identification of HLA-B*57:01 alleles was found to be 100% (95% CI: 94.5-100%) concordant between sequencing methods. Similarly, 100% concordance (95% CI: 83.2-100%) was observed for all HLA alleles at 4-digit resolution, 99.96% overall concordance (95% CI: 99.93-100%) for nucleotide sequence, and 99.9% (95% CI: 99.8-100%) concordance in amino acid sequence (Table 4.4).  The detailed concordance analysis showed that all 15 nucleotide differences were compatible with the Sanger reference sequence as described in section 4.2.3. Interestingly, all 15 differences were called as a C by Sanger sequencing, but called as an M by the MiSeq. Each occurred at the same position (base 532, amino acid 178) of the combined exon 2 and exon 3 sequence, however this had no effect on the interpretation of the allele: Some alleles produced by Sanger sequencing only had two digits of resolution, and therefore this amino acid difference was not applicable, and some had six digit resolution, in which case the Sanger sequences had a single base pair difference (base 532) to the closest known allele from the IMGT/HLA database, and were assigned this known allele interpretation per standard operating procedure. 74Table 4.4 Cross-Contamination Analysis: MiSeq-Sanger Comparison SummaryNumber of samples passing QC (N=82) 65HLA-B*57:01 (positive/negative) concordance (N=65) 100% (94.5-100%)Number of samples with 4-digit resolution by both sequencing methods 204-digit resolution concordance (N=20) 100% (83.2-100%)Total nucleic acid number 35,490Nucleotide differences 15Incompatible nucleotide differences 0Nucleic acid concordance (N=35,490) 99.96% (99.93-100%)Total amino acid number 11,830Amino acid differences 15Incompatible amino acid differences 0Amino acid concordance (N=11,830) 99.9% (99.8-100%)For each mixture, the majority sample was 100% (95% CI: 99.3-100%) concordant with the nucleotide sequence of its unmixed positive control (N=546bp). All 1:1 sample mixtures failed sequencing as expected (N=12), and no samples were given an incorrect allele call. The HLA MiSeq interpretation algorithm failed five samples mixed at a 3:1 ratio due to QC issues. In no instance was an allele assigned from a minority sample.4.3.5 Proof of PrincipleSamples were tested using the MiSeq method from HIV exposed uninfected children (N=241) to confirm a previous observation of a lower than expected HLA-B*57:01 allele frequency in children born to HIV infected mothers. Six samples (2.5%) failed sequencing once, and 15 samples (6.2%) failed sequencing twice. After a second failed attempt, no further attempts were made.  A total of N=226 samples (93.8%) were successfully HLA typed by the MiSeq method. The observed allele frequency was compared to the expected allele frequency based on their mothers’ self-reported ethnicity. The frequency of the HLA-B*57:01 allele was found to be 4% in this cohort, which is comparable to the 2-8% reported in the North American 75population.142 Furthermore, when comparing the expected allele group frequency of HLA-B*57 to that observed, there was a marginal difference in the opposite direction (p=0.037) with the frequency observed exceeding the expected frequency (Figure 4.5). The frequency of B*42 and B*08 was significantly higher than expected; however these were not consistent between cohorts and are believed to be outliers. Figure 4.5 Observed and Expected Allele Group Frequency: Confirmation of ResultsThe observed allele group frequency is indicated in grey, while the expected allele group frequency is blue. Fisher’s exact test was used to determine statistical difference between the observed alleles (N=441) and expected alleles (N=438). Failures and ambiguous alleles are not included in this figure. The difference between Ns is due to rounding error when calculating expected allele frequency of the children based on the parents’ ethnicity. * Denotes allele groups with a p-value<0.05. 764.4 DiscussionIn this chapter, a pharmacogenetic screening assay was validated and developed for the MiSeq NGS platform. This assay determined HLA-B genotype as a screening tool for abacavir hypersensitivity, and was found to be 100% (95% CI: 98.6-100%) sensitive and 100% (95% CI: 97.2-100%) specific in determining whether a patient was positive or negative for HLA-B*57:01 versus the existing Sanger-sequencing based methodology. Furthermore, concordance within an assay and between assays, including results across batches, lab staff, and lab staff experience level, was found to be 100% (see results section for 95% CIs). When Sanger results were available, the assay was 100% (95% CI: 98.7-100%) accurate to four-digits of resolution, removing the need for time-consuming resolution protocols. Importantly, the MiSeq methods produces 4-digit HLA allele calls for all samples that pass QC standards. The BC Centre for Excellence in HIV/AIDS has released nearly 10,300 clinical HLA results since its foundation. Prior to this method’s development, 4-digit HLA allele resolution was only able to be determined in 33% (3409) of samples tested using Sanger sequencing. With antiretroviral drugs recommended for an estimated 36.7 million (2015) individuals infected with HIV, genomic testing for HIV drug hypersensitivity reactions remains a highly important component of clinical care in regions where carriage of HLA-B*57:01 is common in the population.67 For example, in 2017, approximately 40% of individuals on therapy in BC were prescribed abacavir-containing regimens. In a double blind, prospective, randomized study (N=1956) across 19 countries, hypersensitivity was observed in 2.7% of the control group, and in 770% of the group screened for HLA-B*57:01 (p<0.001).154 Similar results have been observed in French and North American trials.155,156HLA-B is the most widely variable MHC class I gene known, with the IMGT/HLA database reporting 4828 different alleles and 3501 different proteins from the HLA-B gene alone.157 Additionally, HLA-B*57:01 is one of a select few HLA genes strongly associated with long-term non-progressors and elite controllers.158 This assay could therefore be useful for future research in understanding the virus-host relationship, particularly in the pursuit of developing new clinical treatments for HIV and emulating the AIDS-refractory phenotype.  Next generation sequencing has rapidly expanded and advanced clinical genomics in the past decade, the utilization of massively parallel sequencing for HIV-related genetic testing was a logical next step.159 This has been previously demonstrated in several settings within the context of HIV, however the assay presented herein was developed with the growing prevalence of next generation sequencing assays in mind: The HLA MiSeq assay can be easily incorporated alongside numerous other HIV and HCV-related tests on the sequencing platform, in order to maximize use of MiSeq space and therefore minimize per-sample costs.160The assay was also implemented in a pharmacogenetic study assessing the potential effects of foetal abacavir hypersensitivity on preterm birth, by determining the HLA-B*57:01 allele among a large cohort of uninfected children born to HIV infected mothers and linking genetic results to obstetric health data. Initial results gave a very low frequency for HLA-B*57:01 (0.82%; N=365) compared to the reported North American frequency (2-8%), resulting 78in insufficient power to reject the null hypothesis.142 These results can be understood on the basis that the ethnic makeup of the study group (HIV infected mothers) differs markedly from that of the typical Canadian population, where about 7% of individuals would be expected to harbour HLA B*57:01.  Thus, in retrospect, the results obtained were not unusual, and we do not have sufficient statistical power to draw conclusions from this study, even with several hundred participants.  Regardless, the large amount of data generated from this study provided a proof of principle that the assay can be used successfully in large numbers of patients.Though this assay was found to be highly accurate and has substantial practical applicability in research and clinical care, it is limited by the relatively high failure rate of 6-12%. This limitation should be addressed, either through optimization of the assay, or in conjunction with other existing sequencing methods (e.g. send failures for Sanger sequencing). It should be noted that the primary reason for failure is the inability to obtain sufficient sequence reads to perform a reliable interpretation - which may be caused by poor PCR amplification or problems in NGS library preparation.  A second limitation is an inherent deficiency in the HLA interpretation algorithm: in cases where two alleles share identical sequences in one of the two sequenced exons, it is impossible to unambiguously determine whether the inability to observe two distinct sequences per allele is due to sequence identity, or allele dropout. These issues would likely be addressed by increasing the number of usable reads. Potential solutions include allotting more space on the MiSeq to HLA testing when running with other tests, or creating more nuance in the HLA MiSeq interpretation algorithm. For example, in the proof of principle, 6-25% of reads on average were used in determining the final result. The other 75-94% of reads were removed from analysis due to the strict criteria of the HLA interpretation algorithm. 79Reworking the interpretation algorithm and salvaging some of the removed sequences could increase the read count. Finally, while the frequency of HLA-B*57:01 is between 2-8% in the North American population, our analysis used a cohort that was skewed towards a positive result, with 67% previously testing HLA-B*57:01 positive using Sanger sequencing. This was done to adequately test the assay, while remaining economical. The negative and positive predictive values of the assay were therefore not able to be determined. Enriching the sample set for HLA-B*57:01 positive samples also creates the possibility of validation bias, since the test results of the samples were known, prior to applying the HLA MiSeq assay. In summary, as a result of the frequent use of abacavir within the BC population, screening individuals for an increased likelihood of developing a potentially life threatening hypersensitivity reaction continues to be critical. Next generation sequencing is more efficient than the previously used Sanger method at determining HLA-B*57:01 since it does not require further testing for allelic resolutions, as shown in this thorough validation and proof of principle using two large cohorts.80Chapter Five: General Discussion and Conclusion5.1 Thesis Summary 5.1.1 Longitudinal Trends of HIV Drug ResistanceThere has been remarkable progress in clinical outcomes for individuals diagnosed with HIV in the past two decades, largely attributable to advances in antiretroviral therapy. In the early days of cART the development of drug resistance that compromised treatment options was frequently observed; however, evidence presented in Chapter Two suggests that the incidence of acquired drug resistance in resource-rich settings has declined drastically. Poor adherence had been previously reported as the strongest correlate of acquiring drug resistance. In the past, intermediate levels of adherence put patients at the greatest risk for acquired resistance, but recently this risk has shifted to those with very poor adherence. Furthermore, adherence has increased during the course of this study. These findings on adherence are potentially attributable to many factors, such as improved patient management by frontline care workers and increased efficacy of antiretroviral drugs. Further research is required to determine the cause of these shifts. While acquired drug resistance has become much less prevalent during the course of this study, transmitted drug resistance has gradually risen over the past decade and since 2009, rates have overcome that of acquired drug resistance in BC. In recent years elevated levels of transmitted drug resistance have been observed in several high-resource settings, and therefore it remains a growing concern. 815.1.2 Sociodemographic Correlates of Access to Resistance Testing and Development of HIV Drug ResistanceChapter Three describes an study that attempted to identify social and demographic factors associated with decreased access to drug resistance testing in BC; a public health setting where individuals can access antiretrovirals and drug resistance testing free of charge. No major disparities were found between sociodemographic strata; however, this analysis was conducted using census tract-level data and not individual patient data. Therefore, it was possible that marginal groups were not adequately considered as described by ecological fallacy. Preliminary results of the STOP HIV/AIDS Program Evaluation (SHAPE) study assessing engagement in the HIV cascade of care found 21% (94/454) of participants had issues accessing HIV care.161 Many participants were from outside of Vancouver, suggesting geography contributes to care access. Evidence for limited access to HIV care in rural areas can be found in the literature.162,163 Therefore a potential limitation of this study was the lack of adjustment for rurality, which should be addressed in future research. Limited access to testing could lead to under reporting of drug resistance and poor health outcomes related to development of drug resistance. Similar to the access to testing results, no major disparities were found between groups in regards to the development of acquired drug resistance. Regions with a higher concentration of indigenous peoples were associated with lower odds of access to testing and higher likelihoods of developing drug resistance in this study. Discrimination towards indigenous peoples in the Canadian health care system has been documented as a barrier to care (30% of survey respondents, Jackson and Reimer 2008), as had geographic isolation from prescribing physicians. These barriers to care could contribute our 82findings regarding access to HIV drug resistance testing among indigenous peoples.164,165 Though in association with indigenous leaders programs were designed that targeted indigenous HIV/AIDS care in BC, such as The Red Road; Pathways to Wholeness: An Aboriginal Strategy for HIV/AIDS in BC and An Aboriginal Strategy on HIV/AIDS for Northern British Columbia, further work is required to address the low access to drug resistance testing and elevated drug resistance observed among Canadian indigenous peoples.166,167 5.1.3 HIV-Related NGS Assay; Abacavir Hypersensitivity Prescreening In Chapter Four a highly accurate, reproducible and repeatable method using the Illumina MiSeq to determine HLA-B*57:01 was validated. The HLA-B*57:01 allele has been associated with hypersensitivity to abacavir: a common antiretroviral drug. Additionally, HLA-B*57:01 has been associated with long-term non-progressors, and is also the focus of “HIV cure” research. The HLA-B alleles of a large cohort were determined as a proof of principle for the assay. This validation serves as an example for the potential future incorporation of clinical genomics, and specifically HIV-related genetic tests, with next generation sequencing platforms. 5.2 Overall Impact, Applications, and Future DirectionsAs cART uptake increased in BC and worldwide, transmitted resistance has emerged as the next hurdle for HIV treatment. Once transmitted resistance has reached more than 15% of the infected population the WHO recommends a systematic reassessment of the regions cART regimens.168 In Chapter Two, up to 18% transmitted drug resistance was reported in a large sample of individuals infected with HIV in BC. While systematic testing for transmitted drug resistance was recommended in the BC clinical care guidelines, monthly BC surveillance reports 83currently discuss detection of any drug resistance and do not distinguish transmitted resistance as its own outcome. Therefore, the research in Chapter Two supports the BC HIV clinical care guideline recommendation for drug resistance screening prior to initiating cART, but efforts should be made to improve the ability to monitor transmitted drug resistance in BC, such as including transmitted drug resistance in routine monthly surveillance reports. Additionally, reassessing the available cART regimens could be beneficial in the context of this new data on transmitted resistance based on the WHO recommendation. Finally, increasing transmitted drug resistance threatens the final target of the 90-90-90 campaign, where 90% of individuals on treatment globally are virologically suppressed, by decreasing antiretroviral efficacy. BC has routine drug resistance testing and monitoring, as well as a wide number of antiretroviral drugs available free of charge. Switzerland is a similar setting where transmitted drug resistance was 16% in 2012, and in parts of the US 16% of new diagnoses had transmitted drug resistance between 2007 and 2010.169,170 These results mirror our findings in BC, and identify transmitted drug resistance as a rising global threat within regions that practice regular clinical monitoring of viral load and drug resistance. Recent evidence from low resource settings such as Southern Africa, Central Africa, and Southeast Asia, where regular clinical monitoring is not conducted, show transmitted drug resistance to be increasing with cART uptake.75 This threat should be addressed when implementing the Treatment as Prevention strategy through systematic or periodic pre-cART resistance testing in settings where it is not yet standard practice. The findings largely show substantial improvements in the status of acquired HIV drug resistance and treatment adherence in BC, however this should not be interpreted as an excuse for complacency. HIV remains a chronic lifelong illness, whose only treatment continues to be 84effective antiretroviral drugs. Ongoing monitoring of drug resistance and its covariates remains an important step to ensure successful treatment of HIV and AIDS, and future efforts should maintain and improve this important work, and expand it in to resource limited settings. While Chapter Two highlights the importance of ongoing HIV genetic resistance testing, Chapter Three subsequently discussed the future of HIV-related NGS testing through the example of abacavir hypersensitivity prescreening validation and proof of principle. Abacavir continues to be a commonly used antiretroviral drug and is recommended as a backbone nRTI in first line regimens within BC. Prescreening for a genetic predisposition to abacavir hypersensitivity can decrease morbidity and mortality associated with this condition, and has been implemented in BC since 2006.171 While Sanger sequencing was the gold standard at that time, next generation sequencing is quickly expanding the field of clinical genomics. Clonal sequence amplification paired with massively parallel sequencing on the Illumina MiSeq has been particularly beneficial for HLA-B*57:01 detection, allowing for each allele to be determined separately. As other HIV-related NGS assays are transferred from Sanger sequencing to the more powerful MiSeq platform, HLA-B*57:01 prescreening for abacavir will be able to be assayed on the same MiSeq run, improving efficiency and decreasing costs. This high throughput ability and economical design could particularly benefit HIV resistance testing facilities in resource-limited settings where the prevalence of HIV, and therefore the demand for testing, is high. 855.3 LimitationsThough the findings of this thesis have broad implications for HIV drug resistance and HIV-related NGS testing, it is important to note that extrapolating these results outside of a resource-rich setting could be inappropriate. This research was conducted within the context of a public health care system where patients diagnosed with HIV can access cART and any prescribed testing free of charge; this should be taken into account when interpreting the results. While the cost of NGS has declined in recent years it remains restrictive; this limitation should be kept in mind when assessing the applicability and impact of the validation. Furthermore, evidence in the literature suggests each class of antiretrovirals has a different relationship to adherence and the associated likelihood of developing drug resistance. Further research should focus on these relationships, to inform a more thorough understanding of how adherence and drug class affects drug resistance. 86Bibliography1. Haverkos, H. W. & Curran, J. W. The Current Outbreak of Kaposi’s Sarcoma and Opportunistic Infections. CA. Cancer J. Clin. 32, 330–339 (1982).2. United States Centre for Disease Control. Current Trends Update on Acquired Immune Deficiency Syndrome (AIDS): United States. Morbidity and Mortality Weekly Report (1982). Available at: https://www.cdc.gov/mmwr/preview/mmwrhtml/00001163.htm. (Accessed: 22nd June 2017)3. Popovic, M. et al. Isolation and Transmission of Human Retrovirus (Human T-Cell Leukemia Virus). Source Sci. New Ser. 219, 856–859 (1983).4. Barré-Sinoussi, F. et al. Isolation of a T-lymphotropic retrovirus from a patient at risk for acquired immune deficiency syndrome (AIDS). Science 220, 868–71 (1983).5. Federal HIV/AIDS Web Council. A Timeline of HIV/AIDS. (2016).6. UNAIDS. Fact sheet. (2015).7. UNAIDS. Fact sheet. (2016).8. U.S. Food and Drug Administration. Expanded Access and Expedited Approval of New Therapies Related to HIV/AIDS. (2014). Available at: https://www.fda.gov/ForPatients/Illness/HIVAIDS/Treatment/. (Accessed: 28th June 2017)9. Bowersox, J. Nevirapine approved by FDA. Food and Drug Administration. NIAID AIDS agenda 10 (1996).10. McInnes, C. W. et al. HIV/AIDS in Vancouver, British Columbia: a growing epidemic. Harm Reduct. J. 6, 5 (2009).11. Schechter, M. T. et al. The Vancouver Lymphadenopathy-AIDS Study: 6. HIV seroconversion in a cohort of homosexual men. CMAJ 135, 1355–60 (1986).12. Macdougall, R. G., Rekart, M. L., Knowles, L., Spencer, D. & Elliott, L. HIV / AIDS UPDATE Year End 2000. (2000).13. Strathdee, S. et al. Needle exchange is not enough: lessons from the Vancouver injecting drug use study. AIDS 11, 59–65 (1997).14. BC Centre for Excellence in HIV/AIDS. Drug Treatment Program. (2017). Available at: http://www.cfenet.ubc.ca/drug-treatment-program. (Accessed: 29th June 2017)8715. BC Centre for Disease Control. HIV in British Columbia: Annual Surveillance Report 2014. (2015).16. BC Centre for Excellence in HIV/AIDS. BC Centre for Excellence in HIV/AIDS Drug Treatment Program Monthly Report: October 2016. (2016).17. Division of HIV/AIDS Prevention, National Center for HIV/AIDS, Viral Hepatitis, S. T. D. and T. P. HIV Transmission. Centers for Disease Control and Prevention (2017). Available at: https://www.cdc.gov/hiv/basics/transmission.html. (Accessed: 10th July 2017)18. Richman, D. D. et al. Human Immunodeficiency Virus. (International Medical Press, 2003).19. Wilen, C. B., Tilton, J. C. & Doms, R. W. HIV: cell binding and entry. Cold Spring Harb. Perspect. Med. 2, (2012).20. Dorr, P. et al. Maraviroc (UK-427,857), a Potent, Orally Bioavailable, and Selective Small-Molecule Inhibitor of Chemokine Receptor CCR5 with Broad-Spectrum Anti-Human Immunodeficiency Virus Type 1 Activity. Antimicrob. Agents Chemother. 49, 4721–4732 (2005).21. Tran, T. A. et al. Resting Regulatory CD4 T Cells: A Site of HIV Persistence in Patients on Long-Term Effective Antiretroviral Therapy. PLoS One 3, e3305 (2008).22. US Centers for Disease Control and Prevention. Mononucleosis. (2016). Available at: https://www.cdc.gov/epstein-barr/about-mono.html. (Accessed: 21st July 2017)23. Trickey, A. et al. Survival of HIV-positive patients starting antiretroviral therapy between 1996 and 2013: a collaborative analysis of cohort studies. Lancet HIV (2017). doi:10.1016/S2352-3018(17)30066-824. The Antiretroviral Therapy Cohort Collaboration. Life expectancy of individuals on combination antiretroviral therapy in high-income countries: a collaborative analysis of 14 cohort studies. Lancet 372, 293–299 (2008).25. Wandeler, G., Johnson, L. F. & Egger, M. Trends in life expectancy of HIV-positive adults on antiretroviral therapy across the globe. Curr. Opin. HIV AIDS 11, 492–500 (2016).26. World Health Organization, United Nations AIDS & United Nations Children’s Fund. Presentation on Global Update on HIV Treatment 2013: Results, Impact and Opportunities. (2013).27. Rooke, R. et al. Isolation of drug-resistant variants of HIV-1 from patients on long-term zidovudine therapy. AIDS 3, 411–415 (1989).8828. Larder, B. A., Darby, G. & Richman, D. D. HIV with reduced sensitivity to Zidovudine (AZT) isolated during prolonged therapy. Science 234, 1731–1734 (1989).29. Larder, B. a & Kemp, S. D. Multiple mutations in HIV-1 reverse transcriptase confer high-level resistance to zidovudine (AZT). Science 246, 1155–1158 (1989).30. Lv, Z., Chu, Y. & Wang, Y. HIV protease inhibitors: a review of molecular selectivity and toxicity. HIV. AIDS. (Auckl). 7, 95–104 (2015).31. Harrigan, P. R. et al. Predictors of HIV drug-resistance mutations in a large antiretroviral-naive cohort initiating triple antiretroviral therapy. J. Infect. Dis. 191, 339–47 (2005).32. Montaner, J. S. G. et al. A Randomized, Double-blind Trial Comparing Combinations of Nevirapine, Didanosine, and Zidovudine for HIV-Infected Patients. JAMA 279, 930 (1998).33. Cott, S. et al. A controlled trial of two nucleoside analogues plus indinavir in persons with human immunodeficiency virus infection and CD4 cell counts of 200 pre cubic millimeter or less. N. Engl. J. Med. 337, 725–733 (1997).34. Hammer, S. M. et al. A Trial Comparing Nucleoside Monotherapy with Combination Therapy in HIV-Infected Adults with CD4 Cell Counts from 200 to 500 per Cubic Millimeter. N. Engl. J. Med. 335, 1081–1090 (1996).35. CAESAR Coordinating Committee. Randomised trial of addition of lamivudine or lamivudine plus loviride to zidovudine-containing regimens for patients with HIV-1 infection: the CAESAR trial. Lancet 349, 1413–1421 (1997).36. World Health Organization. in (WHO Press, World Health Organization, 2016).37. British Columbia Centre for Excellence in HIV/AIDS. Primary Care Guidelines for the Management of HIV/AIDS in British Columbia On behalf of the Primary Care Guidelines Panel BC Centre for Excellence in HIV/AIDS. (2015).38. Reijers, M. H. et al. Maintenance therapy after quadruple induction therapy in HIV-1 infected individuals: Amsterdam Duration of Antiretroviral Medication (ADAM) study. Lancet 352, 185–190 (1998).39. Crum, N. F. et al. Comparisons of Causes of Death and Mortality Rates Among HIV-Infected Persons. JAIDS J. Acquir. Immune Defic. Syndr. 41, 194–200 (2006).40. Finzi, D. et al. Identification of a Reservoir for HIV-1 in Patients on Highly Active Antiretroviral Therapy. Science 278, 1295–1300 (1997).8941. Wong, J. K. et al. Recovery of Replication-Competent HIV Despite Prolonged Suppression of Plasma Viremia. Science 278, 1291–1295 (1997).42. Montaner, J. S. G. et al. Expansion of HAART Coverage Is Associated with Sustained Decreases in HIV/AIDS Morbidity, Mortality and HIV Transmission: The ‘HIV Treatment as Prevention’ Experience in a Canadian Setting. PLoS One 9, e87872 (2014).43. UNAIDS. 90-90-90: An ambitious treatment target to help end the AIDS epidemic. (2014).44. Sidibé, M., Loures, L. & Samb, B. The UNAIDS 90-90-90 target: a clear choice for ending AIDS and for sustainable health and development. J. Int. AIDS Soc. 19, 21133 (2016).45. Hull, M. W. & Montaner, J. S. G. HIV treatment as prevention: the key to an AIDS-free generation. J. food drug Anal. 21, 95–101 (2013).46. Montaner, S. G. et al. Association of highly active antiretroviral therapy coverage, population viral load, and yearly new HIV diagnoses in British Columbia, Canada: a population-based study. Lancet 6736, (2010).47. Laxminarayan, R. et al. Antibiotic resistance—the need for global solutions. Lancet Infect. Dis. 13, 1057–1098 (2013).48. Bennett, D. E., Myatt, M., Bertagnolio, S., Sutherland, D. & Gilks, C. F. Recommendations for surveillance of transmitted HIV drug resistance in countries scaling up antiretroviral treatment. Antivir. Ther. 13, 25–36 (2008).49. Laskey, S. B. & Siliciano, R. F. A mechanistic theory to explain the efficacy of antiretroviral therapy. Nat. Rev. Microbiol. 12, 772–780 (2014).50. U.S. Department of Health and Human Services. FDA Approves First Protease Inhibitor Drug for Treatment of HIV. (1995). Available at: https://archive.hhs.gov/news/press/1995pres/951207.html. (Accessed: 28th June 2017)51. FDA. FDA Product Label: Tybost (cobicistat). (2012).52. Lu, Z. Second generation HIV protease inhibitors against resistant virus. Expert Opin. Drug Discov. 3, 775–786 (2008).53. Zeldin, R. K. & Petruschke, R. A. Pharmacological and therapeutic properties of ritonavir-boosted protease inhibitor therapy in HIV-infected patients. J. Antimicrob. Chemother. 53, 4–9 (2003).9054. Summa, V. et al. Discovery of Raltegravir, a Potent, Selective Orally Bioavailable HIV-Integrase Inhibitor for the Treatment of HIV-AIDS Infection. J. Med. Chem. 51, 5843–5855 (2008).55. Pendri, A., Meanwell, N. A., Peese, K. M. & Walker, M. A. New first and second generation inhibitors of human immunodeficiency virus-1 integrase. Expert Opin. Ther. Pat. 21, 1173–1189 (2011).56. Center for Drug Evaluation and Research & Food and Drug Administration. Isentress (raltegravir) 400 mg Approval Letter. (2007).57. Centre for Drug Evaluation and Research. Fuzeon (enfuvirtide) 90 mg Approval Letter. (2003).58. Mcgovern, R. A. The Use of Genetic Sequencing Technologies to Determine HIV-1 Viral Tropism and to Evaluate the Effects of Maraviroc on Patient Viral Populations. (University of British Columbia, 2015).59. Imaz, A., Falcó, V. & Ribera, E. Antiretroviral salvage therapy for multiclass drug-resistant HIV-1-infected patients: from clinical trials to daily clinical practice. AIDS Rev. 13, 180–93 (2011).60. Sanger, F. & Coulson, A. R. A rapid method for determining sequences in DNA by primed synthesis with DNA polymerase. J. Mol. Biol. 94, 441–8 (1975).61. Woods, C. K. et al. Automating HIV drug resistance genotyping with RECall, a freely accessible sequence analysis tool. J. Clin. Microbiol. 50, 1936–42 (2012).62. Ewing, B., Hillier, L., Wendl, M. C. & Green, P. Base-calling of automated sequencer traces using phred. I. Accuracy assessment. Genome Res. 8, 175–85 (1998).63. Swenson, L. C. et al. HIV drug resistance detected during low-level viraemia is associated with subsequent virologic failure. AIDS 28, 1125–34 (2014).64. Liu, L. et al. Comparison of Next-Generation Sequencing Systems. J. Biomed. Biotechnol. 2012, 1–11 (2012).65. Illumina. MiSeq Specifications: Key performance parameters. (2017). Available at: https://www.illumina.com/systems/sequencing-platforms/miseq/specifications.html. (Accessed: 10th August 2017)66. Brumme, C. J. Tools and Applications for DNA Sequencing in the Clinical Management of Viral Infectious Diseases: Examples From HIV-1 and HCV. (University of British Columbia, 2015).9167. Joint United Nations Programme on HIV/AIDS. Global AIDS Update 2016. (2016).68. Montaner, J. S. et al. 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 279, 930–937 (1998).69. Hammer, S. M. et al. A controlled trial of two nucleoside analogues plus indinavir in persons with human immunodeficiency virus infection and CD4 cell counts of 200 per cubic millimeter or less. N. Engl. J. Med. 337, 725–733 (1997).70. Clavel, F. & Hance, A. J. HIV drug resistance. N. Engl. J. Med. 350, 1023–1035 (2004).71. Durant, J. et al. Drug-resistance genotyping in HIV-1 therapy: The VIRADAPT randomised controlled trial. Lancet 353, 2195–2199 (1999).72. Baxter, J. D. et al. A randomized study of antiretroviral management based on plasma genotypic antiretroviral resistance testing in patients failing therapy. AIDS 14, F83-93 (2000).73. De Luca, A. et al. Declining prevalence of HIV-1 drug resistance in antiretroviral treatment-exposed individuals in Western Europe. J. Infect. Dis. 207, 1216–20 (2013).74. Scherrer, A. U. et al. Emergence of Acquired HIV-1 Drug Resistance Almost Stopped in Switzerland: A 15-Year Prospective Cohort Analysis. Clin. Infect. Dis. 62, 1310–7 (2016).75. World Health Organization; HIV/AIDS Programme. HIV Drug Resistance Report 2012.76. Rhee, S. Y. et al. Geographic and Temporal Trends in the Molecular Epidemiology and Genetic Mechanisms of Transmitted HIV-1 Drug Resistance: An Individual-Patient- and Sequence-Level Meta-Analysis. PLoS Med. 12, (2015).77. Yang, W. L. et al. Assessing the Paradox Between Transmitted and Acquired HIV Type 1 Drug Resistance Mutations in the Swiss HIV Cohort Study From 1998 to 2012. J Infect Dis 212, 28–38 (2015).78. Frentz, D. et al. Increase in transmitted resistance to non-nucleoside reverse transcriptase inhibitors among newly diagnosed HIV-1 infections in Europe. BMC Infect. Dis. 14, 407 (2014).79. Patterson, S. et al. Cohort Profile: HAART Observational Medical Evaluation and Research (HOMER) cohort. Int. J. Epidemiol. 44, 58–67 (2015).80. British Columbia Centre for Excellence in HIV/AIDS. Primary Care Guidelines for the Management of HIV/AIDS in British Columbia on behalf of the Primary Care Guidelines Panel BC Centre for Excellence in HIV/AIDS. (2015).9281. Swenson, L. C. et al. HIV drug resistance detected during low-level viraemia is associated with subsequent virologic failure. AIDS 28, 1125–34 (2014).82. Wensing, A. M. et al. 2015 Update of the Drug Resistance Mutations in HIV-1. Top. Antivir. Med. 23, 132–141 (2015).83. Newcombe, R. G. Two-sided confidence intervals for the single proportion: comparison of seven methods. Stat. Med. 17, 857–72 (1998).84. Bennett, D. E. et al. Drug Resistance Mutations for Surveillance of Transmitted HIV-1 Drug-Resistance: 2009 Update. Updat. PLoS ONE 4, (2009).85. Hosmer, D. W., Lemeshow, S. & Sturdivant, R. X. Applied logistic regression. (John Wiley & Sons, 2013).86. Mcdonald, J. H. Handbook of Biological Statistics. (Sparky House Publishing, 2009).87. Lima, V. D. et al. Development and Validation of a Composite Programmatic Assessment Tool for HIV Therapy. PLoS One 7, e47859 (2012).88. Lee, F. J., Amin, J. & Carr, A. Efficacy of Initial Antiretroviral Therapy for HIV-1 Infection in Adults: A Systematic Review and Meta- Analysis of 114 Studies with up to 144 Weeks’ Follow-Up. PLoS One 9, e97482 (2014).89. Kobin, A. B. & Sheth, N. U. Levels of adherence required for virologic suppression among newer antiretroviral medications. Annals of Pharmacotherapy 45, 372–379 (2011).90. Sethi, A. K., Celentano, D. D., Gange, S. J., Moore, R. D. & Gallant, J. E. Association between adherence to antiretroviral therapy and human immunodeficiency virus drug resistance. Clin. Infect. Dis. 37, 1112–8 (2003).91. Tam, L. W. Y. et al. The relationship between resistance and adherence in drug-naive individuals initiating HAART is specific to individual drug classes. J. Acquir. Immune Defic. Syndr. 49, 266–71 (2008).92. Bangsberg, D. R., Moss, A. R. & Deeks, S. G. Paradoxes of adherence and drug resistance to HIV antiretroviral therapy. J. Antimicrob. Chemother. 5353, 696–699 (2004).93. Mathers, B. M. et al. Global epidemiology of injecting drug use and HIV among people who inject drugs: a systematic review. Lancet 372, 1733–1745 (2008).94. Cohn, J. A. HIV-1 infection in injection drug users. Infectious Disease Clinics of North America 16, 745–770 (2002).9395. Werb, D., Mills, E. J., Montaner, J. S. G. & Wood, E. Risk of resistance to highly active antiretroviral therapy among HIV-positive injecting drug users: a meta-analysis. Lancet. Infect. Dis. 10, 464–9 (2010).96. Azar, P. et al. Drug use patterns associated with risk of non-adherence to antiretroviral therapy among HIV-positive illicit drug users in a Canadian setting: a longitudinal analysis. BMC Infect. Dis. 15, 193 (2015).97. Walmsley, S. L., Becker, M. I., Zhang, M., Humar, A. & Harrigan, P. R. Predictors of virological response in HIV-infected patients to salvage antiretroviral therapy that includes nelfinavir. Antivir. Ther. 6, 47–54 (2001).98. Montaner, J. S. et al. Multiple drug rescue therapy for HIV-infected individuals with prior virologic failure to multiple regimens. AIDS 15, 61–69 (2001).99. Harrigan, P. R. et al. Baseline HIV drug resistance profile predicts response to ritonavir-saquinavir protease inhibitor therapy in a community setting. AIDS 13, 1863–71 (1999).100. Montaner, J. S. et al. Human immunodeficiency virus-infected persons with mutations conferring resistance to zidovudine show reduced virologic responses to hydroxyurea and stavudine-lamivudine. J Infect Dis 181, 729–732 (2000).101. Miller, V. et al. HIV-1 reverse transcriptase (RT) genotype and susceptibility to RT inhibitors during abacavir monotherapy and combination therapy. AIDS 14, 163–71 (2000).102. Alexander, C. S. et al. HIV protease and reverse transcriptase variation and therapy outcome in antiretroviral-naive individuals from a large North American cohort. AIDS 15, 601–607 (2001).103. National Institutes of Health (NIH). Guidelines for the Use of Antiretroviral Agents in HIV-1-Infected Adults and Adolescents Drug-Resistance Testing Genotypic and Phenotypic Resistance Assays. (2014).104. DHHS Panel on Antiretroviral Guidelines for Adults and Adolescents & Office of AIDS Research Advisory Council (OARAC). Guidelines for the Use of Antiretroviral Agents in HIV-1-Infected Adults and Adolescents. (2016).105. Kinsler, J. J., Wong, M. D., Sayles, J. N., Davis, C. & Cunningham, W. E. The Effect of Perceived Stigma from a Health Care Provider on Access to Care Among a Low-Income HIV-Positive Population. AIDS Patient Care STDS 21, 584–592 (2007).106. Cunningham, C. O. et al. Utilization of Health Care Services in Hard-to-Reach Marginalized HIV-Infected Individuals. AIDS Patient Care STDS 21, 177–186 (2007).94107. Leaver, C. A., Bargh, G., Dunn, J. R. & Hwang, S. W. The Effects of Housing Status on Health-Related Outcomes in People living with HIV: A Systematic Review of the Literature. AIDS Behav. 11, 85–100 (2007).108. Chambers, L. A. et al. Stigma, HIV and health: a qualitative synthesis. BMC Public Health 15, 848 (2015).109. Gwadz, M. et al. Exploring How Substance Use Impedes Engagement along the HIV Care Continuum: A Qualitative Study. Front. Public Heal. 4, (2016).110. Bolsewicz, K., Vallely, A., Debattista, J., Whittaker, A. & Fitzgerald, L. Factors impacting HIV testing: a review – perspectives from Australia, Canada, and the UK. AIDS Care 27, 570–580 (2015).111. Giordano, T. P. Retention in HIV care: what the clinician needs to know. Top. Antivir. Med. 19, 12–6 (2010).112. Denning, P. & Dinenno, E. Communities in crisis: Is there a generalized HIV epidemic in impoverished urban areas of the United States? Center for Disease Control and Prevention (2015).113. Joy, R. et al. Impact of Neighborhood-Level Socioeconomic Status on HIV Disease Progression in a Universal Health Care Setting. JAIDS J. Acquir. Immune Defic. Syndr. 47, 500–505 (2008).114. Vandenbroucke, J. P. et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): Explanation and Elaboration for the STROBE Initiative. (2007). doi:10.1371/journal.pmed115. Low-Beer, S., Yip, B., O’Shaughnessy, M. V, Hogg, R. S. & Montaner, J. S. Adherence to triple therapy and viral load response. J. Acquir. Immune Defic. Syndr. 23, 360–1 (2000).116. R Development Core Team. R: A language and environment for statistical computing. (2008).117. SAS Institute Inc. SAS 9.4 Software. (2013).118. Gene Codes Corperation. Sequencher® version 5.4.6 DNA sequence analysis software. Ann Arbor, MI USA Available at: http://www.genecodes.com. (Accessed: 25th April 2017)119. Johnson, V. A. et al. Update of the Drug Resistance Mutations in HIV-1: March 2013. Top. Antiretrovir. Med. 21, 6–14 (2013).120. Godin, G. et al. Prediction of adherence to antiretroviral therapy: A one-year longitudinal study. AIDS Care 17, 493–504 (2005).95121. Wood, E. et al. Adherence and plasma HIV RNA responses to highly active antiretroviral therapy among HIV-1 infected injection drug users. C.  Can. Med. Assoc. J. 169, 656–661 (2003).122. Rocheleau, G., Brumme, C. J., Shoveller, J., Lima, V. & Harrigan, R. Longitudinal Trends of HIV Drug Resistance in a Large Canadian Cohort; 1996-2016. Clin. Microbiol. Infect. [In Press, Corrected Proof], (2017).123. Wood, E. et al. Adherence and plasma HIV RNA responses to highly active antiretroviral therapy among HIV-1 infected injection drug users. CMAJ 169, 656–661 (2003).124. Long, C. et al. Income level and drug related harm among people who use injection drugs in a Canadian setting. Int. J. Drug Policy 25, 458–464 (2014).125. Statistics Canada. 2011 Census; Aboriginal Peoples in Canada: First Nations People, Métis and Inuit. (2011).126. Gordillo, V., del Amo, J., Soriano, V. & González-Lahoz, J. Sociodemographic and psychological variables influencing adherence to antiretroviral therapy. AIDS 13, 1763–1769 (1999).127. Nosyk, B. et al. The cascade of HIV care in British Columbia, Canada, 1996–2011: a population-based retrospective cohort study. Lancet Infect. Dis. 14, 40–49 (2014).128. Pande, J. N., Singh, U. B., Sinha, S., Agarwal, R. C. & Singh, S. P. N. Evaluation of risk factors and prevalence of drug resistant tuberculosis in North India. Chest. 128, 404 (2005).129. Ruddy, M. et al. Rates of drug resistance and risk factor analysis in civilian and prison patients with tuberculosis in Samara Region, Russia. Thorax 60, 130–135 (2005).130. Suárez-García, I. et al. Risk factors for multidrug-resistant tuberculosis in a tuberculosis unit in Madrid, Spain. Eur. J. Clin. Microbiol. Infect. Dis. 28, 325–330 (2009).131. Barroso, E. C. et al. Risk factors for acquired multidrug-resistant tuberculosis. J Pneumol J Pneumol 29, 89–97 (2003).132. Faustini, A., Hall, A. J. & Perucci, C. A. Risk factors for multidrug resistant tuberculosis in Europe: a systematic review. Thorax 61, 158–63 (2006).133. Smith, W. R. The 2011 National Household Survey: the complete statistical story. StatCan Blog (2015). Available at: http://www.statcan.gc.ca/eng/blog-blogue/cs-sc/2011NHSstory. (Accessed: 31st October 2016)96134. British Columbia Centre for Excellence in HIV/AIDS. Therapeutic Guidelines for the Treatment of HIV/AIDS and Related Conditions. (1999).135. British Columbia Centre for Excellence in HIV/AIDS. Therapeutic Guidelines Antiretroviral Therapy for HIV-1 Infected Adults. (2006).136. 19th WHO Model List of Essential Medicines. (2015).137. Health Canada. Drug Product Database Online Query: Ziagen. Government of Canada (2017). Available at: https://health-products.canada.ca/dpd-bdpp/. (Accessed: 3rd March 2017)138. Appendix B: Common HIV drugs available in Canada for adults. Canada’s source for HIV and hepatitis C information (CATIE). Supported by the Public Health Agency of Canada Available at: http://www.catie.ca/en/practical-guides/hiv-drug-treatment/appendices/b. (Accessed: 3rd March 2017)139. Antoniou, T. et al. Impact of the Data Collection on Adverse Events of Anti-HIV Drugs Cohort Study on Abacavir Prescription among Treatment-Naive, HIV-Infected Patients in Canada. J. Int. Assoc. Provid. AIDS Care 13, 153–159 (2014).140. Hewitt, R. G. Abacavir Hypersensitivity Reaction. Clin. Infect. Dis. 34, 1137–1142 (2002).141. Mallal, S. et al. Association between presence of HLA-B*5701, HLA-DR7, and HLA-DQ3 and hypersensitivity to HIV-1 reverse-transcriptase inhibitor abacavir. Lancet (London, England) 359, 727–32 (2002).142. Small, C. B., Margolis, D. A., Shaefer, M. S. & Ross, L. L. HLA-B*57:01 allele prevalence in HIV-infected North American subjects and the impact of allele testing on the incidence of abacavir-associated hypersensitivity reaction in HLA-B*57:01-negative subjects. BMC Infect. Dis. 17, 256 (2017).143. Sanger, F., Nicklen, S. & Coulson, A. R. DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74, 5463–5467 (1977).144. Immuno Polymorphism Database, HLA Informatics Group & Anthony Nolan Research Institute. HLA Nomenclature. IPD-IMGT/HLA Database (2017). Available at: http://hla.alleles.org/nomenclature/naming.html. (Accessed: 11th September 2017)145. Shafer, R. W. et al. High degree of interlaboratory reproducibility of human immunodeficiency virus type 1 protease and reverse transcriptase sequencing of plasma samples from heavily treated patients. J. Clin. Microbiol. 39, 1522–9 (2001).97146. International Organization for Standardization. Statistics Vocabulary and symbols Part 1: General statistical terms and terms used in probability. (2006). Available at: http://www.iso.org/iso/catalogue_detail.htm%3Fcsnumber=40145. (Accessed: 15th February 2017)147. Clinical and Laboratory Standards Institute: National Committee for Clinical Laboratory Standards. EP09-A2 Measurement Procedure Comparison and Bias Estimation Using Patient Samples: Approved Guideline. Second Edition. (2002).148. Koch, D. D. & Peters, T. J. in Tietz Textbook of Clincial Chemistry (eds. Birtis, C. A. & Ashwood, E. R.) 320–335 (W.B. Saunders Co., 1999).149. Public Health Agency of Canada. HIV and AIDS in Canada: Surveillance Report to December 31, 2013. (2014).150. Townsend, C. L., Tookey, P. A., Newell, M.-L. & Cortina-Borja, M. Antiretroviral therapy in pregnancy: balancing the risk of preterm delivery with prevention of mother-to-child HIV transmission. Antivir. Ther. 15, 775–783 (2010).151. Food and Drug Administration. Abacavir (Ziagen) package insert. Available at: https://www.accessdata.fda.gov/drugsatfda_docs/label/2008/020977s019,020978s022lbl.pdf. (Accessed: 2nd October 2017)152. MHC database. Information National Center for Biotechnology, Medical University Graz, Available at: https://www.ncbi.nlm.nih.gov/projects/gv/mhc/ihwg.cgi?cmd=PRJOV&ID=9. (Accessed: 15th August 2017)153. EUROSTAM - FP7-HEALTH collaborative project. The Allele Frequency Net Database - Allele, haplotype and genotype frequencies in Worldwide Populations. (2016). Available at: http://www.allelefrequencies.net/. (Accessed: 4th May 2017)154. Mallal, S. et al. HLA-B*5701 Screening for Hypersensitivity to Abacavir. N. Engl. J. Med. 358, 568–579 (2008).155. Zucman, D., Truchis, P. de, Majerholc, C., Stegman, S. & Caillat-Zucman, S. Prospective Screening for Human Leukocyte Antigen-B*5701 Avoids Abacavir Hypersensitivity Reaction in the Ethnically Mixed French HIV Population. JAIDS J. Acquir. Immune Defic. Syndr. 45, 1–3 (2007).156. Young, B. et al. First large, multicenter, open-label study utilizing HLA-B*5701 screening for abacavir hypersensitivity in North America. AIDS 22, 1673–1675 (2008).157. European Molecular Biology Laboratory. Statistics: IMGT/HLA Database. (2017). Available at: http://www.ebi.ac.uk/ipd/imgt/hla/stats.html. (Accessed: 1st August 2017)98158. Emu, B. et al. HLA class I-restricted T-cell responses may contribute to the control of human immunodeficiency virus infection, but such responses are not always necessary for long-term virus control. J. Virol. 82, 5398–407 (2008).159. Koboldt, D. C., Steinberg, K. M., Larson, D. E., Wilson, R. K. & Mardis, E. R. The next-generation sequencing revolution and its impact on genomics. Cell 155, 27–38 (2013).160. Casadellà, M. & Paredes, R. Deep sequencing for HIV-1 clinical management. Virus Res. (2016). doi:10.1016/j.virusres.2016.10.019161. Coulter, E. HIV testing increases in BC, but access to medical care emerging issue. News 1130 (2017).162. MacKenzie, L. J. et al. Is there a rural/urban gap in the quality of HIV care for treatment-naïve HIV-positive individuals initiating antiretroviral therapy in British Columbia? AIDS Care 29, 1218–1226 (2017).163. Albritton, T., Martinez, I., Gibson, C., Angley, M. & Grandelski, V. R. What about Us? Economic and Policy Changes Affecting Rural HIV/AIDS Services and Care. Soc. Work Public Health 32, 273–289 (2017).164. Jackson, R. & Reimer, G. Canadian Aboriginal people living with HIV/AIDS : care, treatment, and support issues. (2008).165. Public Health Agency of Canada. Population-Specific HIV/AIDS Status Report: Aboriginal Peoples. (2010).166. Northern Aboriginal HIV/AIDS Task Force. An Aboriginal Strategy on HIV/AIDS for northern British Columbia. (2008).167. BC Aboriginal HIV/AIDS Task Force. The Red Road; Pathways to Wholeness: An Aboriginal Strategy for HIV/AIDS in BC. (1999).168. Bennett, D. E., Myatt, M., Bertagnolio, S., Sutherland, D. & Gilks, C. F. Recommendations for surveillance of transmitted HIV drug resistance in countries scaling up antiretroviral treatment. Antivir. Ther. 13, 25–36 (2008).169. Yang, W.-L. et al. Assessing the Paradox Between Transmitted and Acquired HIV Type 1 Drug Resistance Mutations in the Swiss HIV Cohort Study From 1998 to 2012. J. Infect. Dis. 212, 28–38 (2015).170. Kim, D. Trend in transmitted HIV-1 antiretroviral drug resistance-associated mutations, 10 HIV Surveillance areas, United States, 2007-2010. Conference on Retroviruses and Opportunistic Infections (CROI) 2013 Oral Abstract, Session 43, #149 (2013). Available at: https://www.cdc.gov/nchhstp/newsroom/2013/croi-studies-of-interest.html. (Accessed: 11th September 2017)99171. Ma, J. D., Lee, K. C. & Kuo, G. M. HLA-B*5701 testing to predict abacavir hypersensitivity. PLoS Curr. 2, RRN1203 (2010).100Appendix Appendix I. Adjusted Odds Ratios, Stratified by Time Interval Within the ‘Third drug in initial cART’ variable, ‘Other’ includes one of fusion inhibitors, integrase inhibitors, or unboosted protease inhibitors.  Model is stratified by time interval of cART initiation.Variable Associated with Drug Resistance Multivariable Model by Time Interval of cART InitiationInterval of cART Initiation 1996-2000 2001-2005 2006-2010 2011-2014AIDS, ever diagnosedNo 1.0 1.0 1.0 1.0Yes 1.3 (0.98-1.6) 1.89 (1.3-2.6) 1.34 (0.95-2.0) 3.0 (1.4-6.3)IDUNo 1.0 1.0 1.0 1.0Yes 2.1 (1.6-2.8) 1.6 (1.2-2.3) 1.9 (1.3-2.8) 2.9 (1.4-5.8)Unknown 0.67 (0.47-0.94) 0.60 (0.36-0.99) 0.79 (0.40-1.5) 0.92 (0.41-2.0)CD4 cell count>500 cells/μL Not Selected 1.0 1.0 1.0350-499 cells/μL Not Selected 2.4 (0.51-17.6) 1.6 (0.52-6.1) 1.2 (0.45-3.1)200-349 cells/μL Not Selected 4.6 (1.3-29.9) 2.8 (1.1-9.8) 0.72 (0.26-2.0)<200 cells/μL Not Selected 9.1 (2.5-58.5) 4.0 (1.5-13.9) 2.7 (1.2-6.6)Sex at birthFemale 1.0 1.0 1.0 1.0Male Not Selected Not Selected 1.5 (0.99-2.1) 2.1 (1.1-4.0)pVL at baseline (per log) 1.7 (1.3-2.2) Not Selected 1.4 (1.0-2.1) 1.8 (1.0-3.5)Age at baseline (years) 0.97 (0.96-0.98) 0.96 (0.94-0.98) 0.98 (0.96-0.99) 0.98 (0.95-1.0)Third drug in initial cARTBoosted PI 1.0 1.0 1.0 1.0NNRTI Not Selected 2.7 (1.9-3.8) 1.7 (1.2-2.4) 3.7 (2.0-7.1)Other Not Selected 2.5 (1.4-4.5) 0.75 (0.11-2.8) 1.4 (0.42-3.9)Follow-up time (years) 0.87 (0.85-0.90) 0.85 (0.82-0.88) 0.81 (0.76-0.86) 0.55 (0.43-0.69)Adherence≥95% 1.0 1.0 1.0 1.090-<95% 1.7 (0.94-2.9) 1.4 (0.67-2.8) 2.2 (1.1-4.0) 3.8 (1.3-10.0)80-<90% 1.8 (1.1-2.7) 2.2 (1.3-3.7) 2.8 (1.7-4.6) 5.3 (2.3-12.0)60-<80% 1.8 (1.2-2.6) 2.5 (1.5-3.9) 3.1 (1.8-5.1) 4.5 (1.7-11.0)40-<60% 1.4 (0.93-2.2) 1.6 (0.93-2.9) 2.5 (1.4-4.6) 4.2 (1.2-12.2)0-<40% 0.90 (0.63-1.3) 0.88 (0.53-1.5) 2.4 (1.3-4.2) 8.4 (3.1-22.3)101Appendix II. Adjusted Odds Ratios, Stratified by Level of Adherence to Drug Regimen For simplicity, only data for each time interval of cART initiation is shown, stratified by adherence level.Variable Associated with Drug Resistance Multivariable Model by Adherence LevelAdherence ≥95% 90-<95% 80-<95% 60-<80% 40-<60% <40%Interval of cART initiation1996-2000 1.0 1.0 1.0 1.0 1.0 1.02001-2005 0.53 (0.37-0.75)0.38 (0.15-0.87)0.83 (0.38-1.8)0.64 (0.37-1.1)0.51 (0.22-1.1)0.33 (0.18-0.60)2006-2010 0.16 (0.11-0.23)0.16 (0.07-0.37)0.29 (0.13-0.62)0.25 (0.14-0.43)0.25 (0.10-0.59)0.24 (0.12-0.48)2011-2014 0.06 (0.03-0.09)0.10 (0.03-0.25)0.17 (0.07-0.40)0.13 (0.06-0.28)0.18 (0.05-0.53)0.48 (0.20-1.12)102Appendix III. Top Ten Baseline Regimen, by Interval of cART InitiationTop 10 Baseline Regimens in Each EraInterval of cART initiationRank Baseline regimens N (%)nRTI1 Backbone NNRTI2 PI3 INT41 lamivudine + stavudine - indinavir - 364 (23)2 zidovudine + lamivudine - indinavir - 313 (20)3 lamivudine + stavudine nevirapine - 241 (15)4 lamivudine + stavudine - nelfinavir - 116 (7)5 zidovudine + lamivudine nevirapine - - 78 (5)6 didanosine + stavudine nevirapine - - 69 (4)7 didanosine + lamivudine nevirapine - - 63 (4)8 zidovudine + lamivudine - saquinavir - 52 (3)9 zidovudine + lamivudine - nelfinavir - 49 (3)1996-2000(N=1604)10 lamivudine + stavudine -indinavir + ritonavir - 47 (3)1 lamivudine + tenofovir -atazanavir + ritonavir - 174 (13)2 zidovudine + lamivudine nevirapine - - 170 (13)3 lamivudine + stavudine -lopinavir + ritonavir - 135 (10)4 zidovudine + lamivudine -lopinavir + ritonavir - 101 (8)5 didanosine + lamivudine nevirapine - - 96 (7)6 lamivudine + stavudine nevirapine - - 94 (7)7 zidovudine + lamivudine efavirenz - - 52 (4)8 didanosine + lamivudine efavirenz - - 51 (4)9 zidovudine + lamivudine - nelfinavir - 46 (3)2001-2005(N=1344)10 lamivudine + tenofovir -lopinavir + ritonavir - 41 (3)1. Nucleoside/nucleotide reverse transcriptase inhibitor (nRTI)2. Non-nucleoside reverse transcriptase inhibitor (NNRTI)3. Protease inhibitor (PI)4. Integrase inhibitor (INT)103Appendix III continued.Interval of cART initiationRank Baseline regimens N (%)nRTI1 Backbone NNRTI2 PI3 INT41 tenofovir + emtricitabine efavirenz - - 751 (37)2 tenofovir + emtricitabine -atazanavir + ritonavir - 572 (28)3 tenofovir + emtricitabine -lopinavir + ritonavir - 191 (9)4 lamivudine + abacavir -atazanavir + ritonavir - 112 (5)5 lamivudine + tenofovir -atazanavir + ritonavir - 92 (4)6 lamivudine + abacavirlopinavir + ritonavir - 57 (3)7 lamivudine + abacavir efavirenz - - 55 (3)8 lamivudine + tenofovir efavirenz - - 40 (2)9 zidovudine + lamivudine -lopinavir + ritonavir - 39 (2)2006-2010(N=2057)10 lamivudine + tenofovir -lopinavir + ritonavir - 18 (1)1 tenofovir + emtricitabine efavirenz - - 505 (33)2 tenofovir + emtricitabine -atazanavir + ritonavir - 436 (28)3 tenofovir + emtricitabine -darunavir + ritonavir - 117 (8)4 lamivudine + abacavir -atazanavir + ritonavir - 98 (6)5 tenofovir + emtricitabine - -elvitegravir + cobicistat 76 (5)6 tenofovir + emtricitabine rilpivirine - - 68 (4)7 tenofovir + emtricitabine - - raltegravir 48 (3)8 lamivudine + abacavir - - dolutegravir 46 (3)9 tenofovir + emtricitabine -lopinavir + ritonavir - 21 (1)2011-2014(N=1538)10 lamivudine + abacavir - - raltegravir 19 (1)104Appendix IV. Comparison of Baseline Characteristics Among Those Included, and Excluded Based on Exclusion Criteria See methods section for exclusion criteria. Comparison was done using Cochran-Mantel-Haensznel Statistics. HAART Observational Medical Evaluation and Research (HOMER). Bivariable AssociationsBaseline Characteristics Included in HOMER*Treatment Naïve Excluded from HOMERP-ValueN (%) 6543(100) 1530(100)Sex at birth     Male - n (%)* 5313(81) 1222(80)     Female - n (%) 1230(19) 308(20)0.23Median age at cART1 initiation in years (Q1-Q3) 41(34-48) 37(31-45)<0.0001Median CD4 cell count at baseline per μL (Q1-Q3) 230(120-380) 330(180-500)<0.0001Median pVL2 log value at baseline (Q1-Q3) 4.9(4.4-5.0) 4.7(4.1-5.0)<0.0001Ever had AIDS3-related illness      No - n (%)* 5084(78) 1206(79)     Yes - n (%) 1459(22) 324(21)0.34History of IDU4 - n (%)      No - n (%)* 2836(43) 661(43)      Yes - n (%) 2333(36) 535(35)      Unknown - n (%) 1374(21) 334(22)0.75Interval of cART initiation    1996-2000 - n (%) 1604(25) 799(52)    2001-2005 - n (%) 1344(21) 178(12)    2006-2010 - n (%) 2057(31) 147(10)    2011-2014 - n (%) 1538(24) 406(27)<0.0001Baseline regimen includes;    Boosted PI5 - n (%)* 2564(39) 189(12)    NNRTI6 - n (%) 2650(41) 160(10)    Other - n (%) 1329(20) 1181(77)<0.00011. Combination antiretroviral therapy (cART)2. Plasma viral load (pVL)3. Autoimmune Deficiency Syndrome (AIDS)4. Injection Drug Use (IDU)5. Protease Inhibitor (PI)6. Non-nucleoside Reverse Transcriptase Inhibitor (NNRTI)105Appendix V. Sensitivity Analysis Comparing Inclusion and Exclusion of those with Unknown Transmitted Drug Resistance Variable Multivariable OR with unknown TDR censoredMultivariable OR including unknown TDR1996-2000Adherence ≥95% 1.0 (1.0-1.0) 1.0 (1.0-1.0)90-<95% 1.8 (0.89-3.5) 1.7 (0.94-2.9)80-<90% 1.7 (1.0-2.8) 1.8 (1.1-2.7)60-<80% 1.5 (0.96-2.4) 1.8 (1.2-2.6)40-<60% 1.5 (0.95-2.5) 1.4 (0.93-2.2)0-<40% 0.83 (0.55-1.2) 0.90 (0.63-1.3)AIDS, ever diagnosedNo Not Selected 1.0 (1.0-1.0)Yes Not Selected 1.3 (0.98-1.6)IDU  No 1.0 (1.0-1.0) 1.0 (1.0-1.0)Yes 2.2 (1.6-3.0) 2.1 (1.6-2.8)Unknown 0.70 (0.44-0.99) 0.67 (0.47-0.94)Age at baseline (years) 0.97 (0.96-0.99) 0.97 (0.96-0.98)pVL at baseline (per log) 2.3 (1.5-3.3) 1.7 (1.3-2.2)Follow-up time (years) 0.87 (0.84-0.90) 0.87 (0.85-0.90)2001-2005Sex at birthFemale 1.0 (1.0-1.0) 1.0 (1.0-1.0)Male 0.63 (0.36-1.1) Not SelectedCD4 cell count>500 cells/μL Not Selected 1.0 (1.0-1.0)350-499 cells/μL Not Selected 2.4 (0.51-17.6)200-349 cells/μL Not Selected 4.6 (1.3-29.9)<200 cells/μL Not Selected 9.1 (2.5-58.5)Third drug in initial cARTBoosted PI 1.0 (1.0-1.0) 1.0 (1.0-1.0)NNRTI 2.6 (1.6-4.2) 2.7 (1.9-3.8)Other 4.9 (1.8-13) 2.5 (1.4-4.5)Adherence≥95% 1.0 (1.0-1.0) 1.0 (1.0-1.0)90-<95% 1.7 (0.61-5.1) 1.4 (0.67-2.8)80-<90% 3.7 (1.7-8.3) 2.2 (1.3-3.7)60-<80% 5.0 (2.5-9.9) 2.5 (1.5-3.9)40-<60% 2.8 (1.3-6.2) 1.6 (0.93-2.9)0-<40% 0.55 (0.25-1.2) 0.88 (0.53-1.5)AIDS, ever diagnosedNo 1.0 (1.0-1.0) 1.0 (1.0-1.0)Yes 2.8 (1.8-4.5) 1.89 (1.3-2.6)IDUNo 1.0 (1.0-1.0) 1.0 (1.0-1.0)Yes 2.1 (1.3-3.5) 1.6 (1.2-2.3)Unknown 0.48 (0.21-1.1) 0.60 (0.36-0.99)Age at baseline (years) 0.98 (0.95-1.0) 0.96 (0.94-0.98)pVL at baseline (per log) Not Selected Not SelectedFollow-up time (years) 0.84 (0.80-0.89) 0.85 (0.82-0.88)106Appendix V continued.Variable Multivariable OR unknown TDR censoredMultivariable OR unknown TDR included2006-2010Sex at birthFemale 1.0 (1.0-1.0) 1.0 (1.0-1.0)Male 1.5 (0.95-2.4) 1.5 (0.99-2.1)CD4 cell count>500 cells/μL 1.0 (1.0-1.0) 1.0 (1.0-1.0)350-499 cells/μL 1.2 (0.32-4.2) 1.6 (0.52-6.1)200-349 cells/μL 2.2 (0.73-6.8) 2.8 (1.1-9.8)<200 cells/μL 3.2 (1.1-9.6) 4.0 (1.5-13.9)Third drug in initial cARTBoosted PI 1.0 (1.0-1.0) 1.0 (1.0-1.0)NNRTI 1.9 (1.3-2.9) 1.7 (1.2-2.4)Other 1.1 (0.13-9.2) 0.75 (0.11-2.8)Adherence≥95% 1.0 (1.0-1.0) 1.0 (1.0-1.0)90-<95% 3.3 (1.7-6.4) 2.2 (1.1-4.0)80-<90% 3.3 (1.8-5.9) 2.8 (1.7-4.6)60-<80% 2.8 (1.5-5.3) 3.1 (1.8-5.1)40-<60% 3.0 (1.5-6.1) 2.5 (1.4-4.6)0-<40% 2.4 (1.2-5.0) 2.4 (1.3-4.2)AIDS, ever diagnosedNo 1.0 (1.0-1.0) 1.0 (1.0-1.0)Yes Not Selected 1.34 (0.95-2.0)IDUNo 1.0 (1.0-1.0) 1.0 (1.0-1.0)Yes 2.3 (1.5-3.6) 1.9 (1.3-2.8)Unknown 0.89 (0.42-1.9) 0.79 (0.40-1.5)Age at baseline (years) 0.98 (0.96-1.0) 0.98 (0.96-0.99)pVL at baseline (per log) 2.1 (1.3-3.4) 1.4 (1.0-2.1)Follow-up time (years) 0.79 (0.74-0.85) 0.81 (0.76-0.86)2011-2014Sex at birthFemale 1.0 (1.0-1.0) 1.0 (1.0-1.0)Male 1.9 (1.0-3.8) 2.1 (1.1-4.0)CD4 cell count>500 cells/μL 1.0 (1.0-1.0) 1.0 (1.0-1.0)350-499 cells/μL 1.1 (0.42-3.0) 1.2 (0.45-3.1)200-349 cells/μL 0.73 (0.27-2.0) 0.72 (0.26-2.0)<200 cells/μL 2.7 (1.1-6.3) 2.7 (1.2-6.6)Third drug in initial cARTBoosted PI Not Selected 1.0 (1.0-1.0)NNRTI Not Selected 3.7 (2.0-7.1)Other Not Selected 1.4 (0.42-3.9)Adherence≥95% 1.0 (1.0-1.0) 1.0 (1.0-1.0)90-<95% 3.6 (1.3-9.8) 3.8 (1.3-10.0)80-<90% 5.3 (2.3-12) 5.3 (2.3-12.0)60-<80% 4.3 (1.7-11) 4.5 (1.7-11.0)40-<60% 4.0 (1.3-12) 4.2 (1.2-12.2)0-<40% 6.9 (2.5-19) 8.4 (3.1-22.3)107Appendix V continued.Variable Multivariable OR unknown TDR censoredMultivariable OR unknown TDR included2011-2014AIDS, ever diagnosedNo 1.0 (1.0-1.0) 1.0 (1.0-1.0)Yes 3.0 (1.4-6.4) 3.0 (1.4-6.3)IDUNo 1.0 (1.0-1.0) 1.0 (1.0-1.0)Yes 3.2 (1.6-6.6) 2.9 (1.4-5.8)Unknown 1.0 (0.45-2.3) 0.92 (0.41-2.0)Age at baseline (years) 0.98 (0.95-1.0) 0.98 (0.95-1.0)pVL at baseline (per log) 1.9 (1.0-3.7) 1.8 (1.0-3.5)Follow-up time (years) 0.54 (0.43-0.69) 0.55 (0.43-0.69)108Appendix VI. Percentage of Participants with Transmitted Drug Resistance MutationsFigure was produced using the World Health Organization Transmitted Mutation List (2009) as Reference.109Appendix VII. Percentage of Transmitted HIV Drug Resistance Among Participants Initiating cART in a Given Year In the primary plot, drug resistance mutations were determined using major mutations in the IAS-USA (2015) mutation list among a subset of adult Drug Treatment Program participants (≥19 years old) who were initially treatment naïve, then started on cART according to therapeutic guidelines between 1996-2015, and had a pVL or CD4 test within six months of starting therapy. Transmitted Drug Resisatnce (TDR) mutations were reported if detected prior to April 2016. Participants were censored at the last available test before moving out of BC, dying, or entering a placebo controlled blind trial. The inset plot shows the percentage of participants with drug resistance detected pre-therapy initiation in a given year, broken down by category of drug resistance. Drug categories include lamivudine/emtricitabine (3TC/FTC), other nucleoside reverse transcriptase inhibitors (Other nRTI), non-nucleoside reverse transcriptase inhibitors (NNRTI), and protease inhibitors (PI). 110Appendix VIII. Proportions in Each Adherence Strata, by Time Interval of cART Initiation111Appendix IX. Detection of Drug Resistance After cART InitiationIndividuals were grouped by the number of drug categories to which they harbored resistance. Analyses were completed independently for each group.  Log-rank p-value for each group is <0.0001, indicating each line is statistically different from the others.112Appendix X. The Percentage of Individuals with Resistance to 0, 1, 2, and 3 or More Drug Categories, by Time Interval of cART InitiationNote that there is only one individual resistant to three drug categories or more (0.08%) in 2011-2014.113Appendix XI. Proportion of Individuals on First Line Therapy with Specified Third Drug The therapy included two NRTIs and one of either NNRTI, PI (boosted, or unboosted), or Other (integrase inhibitors, entry inhibitors, or fusion inhibitors), by time interval of therapy initiation. 114Appendix XII. Percentage of Participants Without Baseline Drug Resistance Testing by Year of cART InitiationThose With Transmitted Drug Resistance were excluded. Total N=5643.115Appendix XIII. Percentage of Patients Accessing Drug Resistance Testing Per Year in BCPatients were considered eligible to access testing when plasma viral load (pVL) was above the lower limit of detection of the drug resistance test in a calendar year. This changed year to year, but was generally higher than pVL of 250 copies/mL. Patients were considered to have accessed testing when a physician ordered a drug resistance test.116Appendix XIV. Hazard Ratios of Developing 3TC/FTC Resistance117Appendix XV. Hazard Ratios of Developing NNRTI Resistance118Appendix XVI. Hazard Ratios of Developing “Other” nRTI Resistance (Excluding 3TC/FTC) 119Appendix XVII. Hazard Ratios of Developing PI Resistance 120Appendix XVIII. Adjusted Odds Ratios of Access to Drug Resistance Testing, Stratified by AdherenceModel was generated using General Estimating Equation (GEE) Logistic Regression.Multivariable Covariates of Accessing Drug Resistance TestingWithout Adherence aOR (95% CI) N=8398 With Adherence aOR (95% CI) N=8398Sex     Female (vs Male) 1.2 (1.1-1.3) 1.2 (1.1-1.3)Hepatitis C     Positive (vs Negative) 1.2 (1.1-1.4) 0.92 (0.80-1.1)     Unknown (vs Negative) 0.38 (0.30-0.47) 0.35 (0.25-0.49)Baseline regimen third drug class    PI (vs NNRTI) 1.1 (0.96-1.1) 1.1 (0.97-1.1)    nRTI Only (vs NNRTI) 1.5 (1.3-1.6) 1.4 (1.3-1.6)    Other (vs NNRTI) 1.2 (0.90-1.6) 1.2 (0.86-1.5)Adherence in first 12 months of therapy <95% (vs ≥95%) Not Selected 1.3 (1.2-1.4)Baseline CD4     <200 cells/μL 1.4 (1.3-1.6) 1.5 (1.3-1.6)     200-<350 cells/μL 1.1 (1.1-1.2) 1.2 (1.1-1.3)     ≥350 cells/μL Reference ReferenceBaseline pVL      ≥100,000 copies/mL 1.1 (1.0-1.3) 1.2 (1.1-1.4)     10,000-<100,000 copies/mL 0.96 (0.85-1.1) 1.0 (0.90-1.2)      <10,000 copies/mL Reference ReferenceEligible for drug resistance test (per year) 1.1 (1.1-1.1) 1.1 (1.1-1.1)Physician experience (last 2 years)      ≥100 patients 0.90 (0.82-0.99) 0.92 (0.84-1.0)      20-100 patients 1.1 (0.98-1.2) 1.1 (0.98-1.2)      Unknown 0.99 (0.87-1.1) 0.68 (0.58-0.80)      <20 patients Reference ReferenceImmigrants (per 10%) 1.0 (0.99-1.0) 1.0 (0.99-1.1)Median Income (per $10k) 0.82 (0.77-0.88) 0.83 (0.77-0.89)Percentage aboriginal ancestry      ≥10% 0.87 (0.77-0.99) 0.88 (0.78-1.0)      5%-<10% 0.85 (0.76-0.95) 0.85 (0.76-0.95)      <5% Reference Reference121Appendix XIX. Adjusted Hazard Ratios of Development of Drug Resistance, Stratified by Adherence Model was generated using Cox Proportional Hazards logistic regression.Multivariable Covariates of Developing Drug ResistanceWithout Adherence  aHR (95% CI) N=5175With Adherence aHR (95% CI) N=5175Age     ≥50 Years 0.72 (0.57-0.91) 0.82 (0.65-1.0)     40-<50 Years 0.96 (0.80-1.2) 1.0 (0.87-1.3)     30-<40 Years 1.0 (0.88-1.2) 1.1 (0.95-1.3)     <30 Years Reference ReferenceSex     Female (vs Male) 1.2 (1.1-1.4) 1.1 (0.95-1.3)PWID     Yes (vs No) 1.6 (1.4-1.8) 1.3 (1.1-1.5)     Unknown (vs No) 1.3 (1.1-1.6) 1.2 (0.98-1.4)Adherence in first 12 months of therapy <95% (vs ≥95%) N/A 2.2 (1.9-2.5)Baseline CD4     <200 cells/μL 1.8 (1.5-2.1) 1.9 (1.6-2.3)     200-<350 cells/μL 1.3 (1.1-1.5) 1.3 (1.1-1.6)     ≥350 cells/μL Reference ReferenceBaseline pVL      ≥100,000 copies/mL 2.0 (1.5-2.5) 2.0 (1.6-2.6)     10,000-<100,000 copies/mL 1.3 (1.0-1.7) 1.3 (1.0-1.7)      <10,000 copies/mL Reference ReferenceFirst Year of ARV      2008-2013 0.38 (0.32-0.46) 0.44 (0.36-0.53)      2004-2007 0.48 (0.40-0.56) 0.52 (0.44-0.62)      2000-2003 0.78 (0.67-0.90) 0.82 (0.71-0.95)      1996-1999 Reference ReferenceImmigrants (per 10%) Not Selected 1.1 (1.0-1.1)Median Income (per $10k) 0.84 (0.77-0.92) 0.82 (0.76-0.89)Percentage aboriginal ancestry      ≥10% 1.3 (1.1-1.6) 1.2 (1.1-1.5)      5%-<10% 0.89 (0.74-1.1) 0.87 (0.72-1.1)      <5% Reference Reference122Appendix XX. Access to Drug Resistance Resting Cohort: Comparison Between Excluded and Included Individuals. Individuals were excluded due to missing data. Cohorts were compared using Chi-squared test. Multivariable Covariates of Accessing Drug Resistance TestingIncluded N(%) N=8398Excluded N(%) N=1058 P-valueSex     Female 1482 (17.7) 210 (19.9)     Male 6916 (82.4) 848 (80.2)0.078Age at First ARV (years)     <30 1322 (15.74) 231 (21.83)     30-≤39 3110 (37.03) 421 (39.79)     40-≤50 2619 (31.19) 279 (26.37)     >50 1347 (16.04) 127 (12)<0.0001Hepatitis C     Positive 3132 (37.29) 441 (41.68)     Negative 4266 (50.8) 455 (43.01)     Unknown 1000 (11.91) 162 (15.31)<0.0001Baseline regimen third drug class    NNRTI 2472 (29.44) 301 (28.45)    PI 3679 (43.81) 348 (32.89)    nRTI Only 1.5 (1.3-1.6) 1.4 (1.3-1.6)    Other 1.2 (0.90-1.6) 1.2 (0.86-1.5)<0.0001Adherence in first 12 months of therapy     <95% 3550 (42.27) 336 (31.76)    ≥95% 4543 (54.1) 353 (33.36)    Unknown 305 (3.63) 369 (34.88)0.013Baseline CD4     <200 cells/μL 3273 (38.97) 288 (27.22)     200-349 cells/μL 2516 (29.96) 283 (26.75)     ≥350 cells/μL 2505 (29.83) 445 (42.06)     Unknown 104 (1.24) 42 (3.97)<0.0001Baseline pVL       <9,999 copies/mL 3273 (38.97) 288 (27.22)     10,000-99,999 copies/mL 2516 (29.96) 283 (26.75)      ≥100,000 copies/mL 2505 (29.83) 445 (42.06)      Unknown 1532 (18.24) 367 (34.69)<0.0001Ever having a test (with eligible pVL)      No 4271 (50.86) 463 (43.76)      Yes 4127 (49.14) 595 (56.24)<0.0001Physician experience (last 2 years)      <20 patients 2402 (28.6) 129 (12.19)      20-100 patients 2591 (30.85) 235 (22.21)      ≥100 patients 2685 (31.97) 285 (26.94)      Unknown 720 (8.57) 409 (38.66)<0.0001123Appendix XXI. Adjusted Odds Ratios of Access to Drug Resistance Testing, Stratified by People Who Inject Drugs (PWID). Individuals with unknown PWID status were excluded from this analysis (N=1627). Models were generated using General Estimating Equation logistic regression.Multivariable Covariates of Accessing Drug Resistance TestingNon-PWIDaOR (95% CI)N= 3852PWID aOR aOR (95% CI)N= 2919Sex     Female (vs Male) 1.3 (1.1-1.5) 1.1 (1.0-1.3)Hepatitis C     Positive (vs Negative) 1.2 (1.1-1.4) 0.92 (0.80-1.1)     Unknown (vs Negative) 0.38 (0.30-0.47) 0.35 (0.25-0.49)Baseline regimen third drug class    PI (vs NNRTI) 1.2 (1.1-1.4) 0.95 (0.84-1.1)    nRTI Only (vs NNRTI) 2.1 (1.7-2.6) 1.2 (0.97-1.4)    Other (vs NNRTI) 1.6 (1.1-2.3) 1.2 (0.65-2.3)Adherence in first 12 months of therapy <95% (vs ≥95%) 1.3 (1.2-1.5) 1.2 (1.1-1.3)Baseline CD4     <200 cells/μL 1.6 (1.4-1.8) 1.3 (1.2-1.5)     200-<350 cells/μL 1.2 (1.0-1.3) 1.1 (0.99-1.3)     ≥350 cells/μL Reference ReferenceBaseline pVL      ≥100,000 copies/mL 1.3 (1.1-1.6) 1.2 (1.0-1.5)     10,000-<100,000 copies/mL 0.97 (0.80-1.2) 1.1 (0.90-1.3)      <10,000 copies/mL Reference ReferenceEligible for drug resistance test (per year) 1.1 (1.1-1.1) 1.1 (1.1-1.1)Physician experience (last 2 years)      ≥100 patients 0.92 (0.80-1.1) Not Selected      20-100 patients 1.1 (0.93-1.2) Not Selected      Unknown 0.71 (0.53-0.95) Not Selected      <20 patients Reference Not SelectedImmigrants (per 10%) Not Selected 1.1 (1.0-1.1)Median Income (per $10k) 0.84 (0.77-0.92) 0.82 (0.76-0.89)Percentage aboriginal ancestry      ≥10% 0.73 (0.58-0.91) Not Selected       5%-<10% 0.88 (0.75-1.0) Not Selected      <5% Reference Not Selected124Appendix XXII. Adjusted Hazard Ratios of Development of Drug Resistance, Stratified by People Who Inject Drugs (PWID). Individuals with unknown PWID status were excluded from this analysis (N=962). Model generated using Cox Proportional Hazards logistic regression.Multivariable Covariates of Developing Drug ResistanceNon-PWID aHR (95% CI) N=2325PWID aHR (95% CI) N=1888Age     >50 Years 0.87 (0.59-1.3) 0.64 (0.45-0.91)     40-<50 Years 1.2 (0.82-1.6) 0.91 (0.71-1.2)     30-<40 Years 1.4 (1.0-2.0) 0.96 (0.77-1.2)     <30 Years Reference ReferenceSex     Female (vs Male) 1.0 (0.76-1.4) Not SelectedHepatitis C     Positive (vs Negative) Not Selected 1.1 (0.87-1.5)     Unknown (vs Negative) Not Selected 2.0 (1.1-3.7)Baseline regimen third drug class    PI (vs NNRTI) 1.2 (0.98-1.5) 0.76 (0.64-0.91)Adherence in first 12 months of therapy <95% (vs ≥95%) 2.6 (2.1-3.1) 1.9 (1.6-2.3)Baseline CD4     <200 cells/μL 1.7 (1.3-2.2) 2.2 (1.7-2.7)     200-<350 cells/μL 1.3 (0.98-1.8) 1.3 (1.0-1.7)     ≥350 cells/μL Reference ReferenceBaseline pVL      ≥100,000 copies/mL 1.6 (1.1-2.5) 2.5 (1.8-3.5)     10,000-<100,000 copies/mL 1.0 (0.67-1.6) 1.7 (1.2-2.5)      <10,000 copies/mL Reference ReferenceFirst Year on cART      2008-2013 0.41 (0.30-0.56) 0.64 (0.49-0.83)      2004-2007 0.63 (0.47-0.84) 0.49 (0.39-0.63)      2000-2003 0.88 (0.68-1.1) 0.74 (0.59-0.92)      1996-1999 Reference ReferenceImmigrants (per 10%) 1.1 (0.99-1.1) Not SelectedPercentage aboriginal ancestry      ≥10% Not Selected 1.3 (1.1-1.5)      5%-<10% Not Selected 0.90 (0.70-1.2)      <5% Not Selected Reference125Appendix XXIII. Development of Drug Resistance Cohort: Comparison Between Excluded and Included Individuals. Individuals were excluded due to missing data resulting in the inability to link census-level sociodemographic data to individual clinical data. Included individuals were compared to excluded individuals using Chi-squared test.Multivariable Covariates of Developing Drug Resistance Included  N (%) N=5175Excluded N (%) N=528 P-valueSex     Female 943 (18.22) 135 (25.57)     Male 4232 (81.78) 393 (74.43)<0.0001Age at First ARV (years)     <30 669 (12.93) 100 (18.94)     30-<40 1799 (34.76) 183 (34.66)     40-<50 1743 (33.68) 167 (31.63)     >50 964 (18.63) 78 (14.77)0.0006Hepatitis C     Positive 2109 (40.75) 285 (53.98)     Negative 2734 (52.83) 203 (38.45)     Unknown 332 (6.42) 40 (7.58)<0.0001Baseline regimen third drug class    NNRTI 2111 (40.79) 255 (48.3)    PI 3064 (59.21) 273 (51.7)0.0009Adherence in first 12 months of therapy     <95% 1960 (37.87) 245 (46.4)    ≥95% 3215 (62.13) 283 (53.6)0.0001Baseline CD4     <200 cells/μL 2310 (44.64) 212 (40.15)     200-<350 cells/μL 1558 (30.11) 139 (26.33)     ≥350 cells/μL 1307 (25.26) 177 (33.52)0.0002Baseline pVL       <10,000 copies/mL 553 (10.69) 78 (14.77)     10,000-<100,000 copies/mL 2118 (40.93) 229 (43.37)      ≥100,000 copies/mL 2504 (48.39) 221 (41.86)0.0022Ever DRT (with eligible pVL)      No 2904 (56.12) 268 (50.76)      Yes 2271 (43.88) 260 (49.24)0.018Physician experience (last 2 years)      <20 patients 1402 (27.09) 96 (18.18)      20-100 patients 1637 (31.63) 177 (33.52)      ≥100 patients 1944 (37.57) 239 (45.27)      Unknown 192 (3.71) 16 (3.03)<0.0001126Appendix XXIV. Access to Drug Resistance Testing Cohort: Comprehensive Univariable and Multivariable Models Models generated using General Estimating Equation logistic regression. Multivariable Covariates of Accessing Drug Resistance TestingUnivariable OR (95% CI) N=8398 Multivariable aOR (95% CI) N=8398Age     ≥50 Years of age 1.9 (1.7-2.2) Not Selected     40-<50 Years of age 1.5 (1.3-1.7) Not Selected     30-<40 Years of age 1.2 (1.1-1.4) Not Selected     <30 Years of age Reference Not SelectedSex      Female (vs Male) 1.3 (1.2-1.4) 1.2 (1.1-1.3)MSM Risk      MSM (vs non-MSM) 1.0 (0.94-1.1) Not Selected      MSM risk unknown (vs non-MSM) 0.43 (0.39-0.48) Not SelectedHeterosexual Risk      Heterosexual (vs non-heterosexual) 1.4 (1.3-1.5) Not Selected      Heterosexual risk unknown 0.54 (0.49-0.59) Not SelectedPWID Risk      PWID (vs non-PWID) 1.2 (1.1-1.3) 1.1 (1.0-1.2)      PWID risk unknown (vs non-PWID) 0.43 (0.38-0.48) 0.47 (0.42-0.53)Hepatitis C     Positive (vs Negative) 1.2 (1.1-1.3) Not Selected     Unknown (vs Negative) 0.44 (0.38-0.51) Not SelectedBaseline regimen third drug class    PI (vs NNRTI) 1.0 (0.94-1.1) 1.1 (0.97-1.1)    nRTI Only (vs NNRTI) 1.3 (1.2-1.4) 1.4 (1.3-1.6)    Other (vs NNRTI) 1.3 (0.97-1.7) 1.2 (0.86-1.5)Adherence     First 12 months of therapy <95% 1.3 (1.2-1.4) 1.3 (1.2-1.4)     First 12 months of therapy Unknown 1.7 (1.5-2.1) 2.6 (2.1-3.3)     First 12 months of therapy >95% Reference ReferenceBaseline CD4     <200 cells/μL 1.4 (1.3-1.5) 1.5 (1.3-1.6)     200-<350 cells/μL 1.2 (1.1-1.3) 1.2 (1.1-1.3)     Unknown 1.3 (1.0-1.6) 0.94 (0.72-1.2)     ≥350 cells/μL Reference ReferenceBaseline pVL      ≥100,000 copies/mL 1.1 (0.98-1.2) 1.2 (1.1-1.4)     10,000-<100,000 copies/mL 0.94 (0.84-1.1) 1.0 (0.90-1.2)      Unknown (First ARV before 1997) 1.5 (1.3-1.7) 1.8 (1.5-2.1)      Unknown (Other) 2.0 (1.5-2.6) 2.2 (1.6-2.9)      <10,000 copies/mL Reference Reference127Appendix XXIV continued.Multivariable Covariates of Accessing Drug Resistance TestingUnivariableOR (95% CI) N=8398MultivariableaOR (95% CI) N=8398Year eligible for drug resistance test (per year) 1.1 (1.1-1.1) 1.1 (1.1-1.1)Physician experience (last 2 years)      ≥100 patients 0.75 (0.69-0.82) 0.92 (0.84-1.0)      20-100 patients 1.1 (0.97-1.1) 1.1 (0.98-1.2)      Unknown 0.88 (0.78-1.0) 0.68 (0.58-0.80)      <20 patients Reference ReferenceOne-family households (per 10%) 1.0 (0.99-1.0) Not SelectedPopulation density (per 10k) 1.0 (0.95-1.1) Not SelectedImmigrants (per 10%) 1.0 (1.0-1.0) 1.0 (0.99-1.1)Median Income (per $10k) 0.87 (0.83-0.91) 0.83 (0.77-0.89)Single people (per 10%) 1.0 (1.0-1.0) Not SelectedPost-secondary certificate (per 10%) 0.94 (0.92-0.97) Not SelectedUnemployment rate (per 10%) 0.97 (0.94-1.0) Not SelectedPercentage aboriginal ancestry      ≥10% 1.1 (1.0-1.2) 0.88 (0.78-1.0)      5%-<10% 1.0 (0.92-1.1) 0.85 (0.76-0.95)      <5% Reference Reference128Appendix XXV. Development of Drug Resistance Cohort: Comprehensive Univariable and Multivariable Models Model generated using Cox Proportional Hazards logistic regression. Multivariable Covariates of Developing Drug ResistanceUnivariable HR (95% CI) N=5175 Multivariable aHR (95% CI) N=5175Age     ≥50 Years of age 0.60 (0.48-0.76) 0.82 (0.65-1.0)     40-<50 Years of age 0.90 (0.75-1.1) 1.0 (0.87-1.3)     30-<40 Years of age 1.1 (0.90-1.3) 1.1 (0.95-1.3)     <30 Years of age Reference ReferenceSex      Female (vs Male) 1.3 (1.2-1.5) 1.1 (0.95-1.3)MSM Risk      MSM (vs non-MSM) 0.76 (0.67-0.87) Not Selected      MSM risk unknown (vs non-MSM) 0.53 (0.45-0.62) Not SelectedHeterosexual Risk      Heterosexual (vs non-heterosexual) 1.1 (0.95-1.2) Not Selected      Heterosexual risk unknown 0.66 (0.57-0.77) Not SelectedPWID Risk      PWID (vs non-PWID) 1.8 (1.6-2.0) 1.3 (1.1-1.5)      PWID risk unknown (vs non-PWID) 1.2 (1.0-1.5) 1.2 (0.98-1.4)Hepatitis C     Positive (vs Negative) 1.8 (1.6-2.0) Not Selected     Unknown (vs Negative) 1.2 (0.89-1.6) Not SelectedBaseline regimen third drug class    PI (vs NNRTI) 1.2 (1.0-1.3) Not SelectedAdherence     First 12 months of therapy <95% (vs >95%) 2.6 (2.3-2.9) 2.2 (1.9-2.5)Baseline CD4     <200 cells/μL 1.9 (1.6-2.2) 1.9 (1.6-2.3)     200-<350 cells/μL 1.1 (0.93-1.3) 1.3 (1.1-1.6)     ≥350 cells/μL Reference ReferenceBaseline pVL      ≥100,000 copies/mL 2.6 (2.1-3.3) 2.0 (1.6-2.6)     10,000-<100,000 copies/mL 1.4 (1.1-1.8) 1.3 (1.0-1.7)      <10,000 copies/mL Reference ReferenceFirst year of ARV      2008-2013 0.31 (0.25-0.37) 0.44 (0.36-0.53)      2004-2007 0.53 (0.45-0.63) 0.52 (0.44-0.62)      2000-2003 0.89 (0.77-1.0) 0.82 (0.71-0.95)      1996-1999 Reference Reference129Appendix XXV continued. Multivariable Covariates of Developing Drug ResistanceUnivariable HR (95% CI) N=5175 Multivariable aHR (95% CI) N=5175Physician experience (last 2 years)      ≥100 patients 0.74 (0.64-0.86) Not Selected      20-100 patients 0.98 (0.85-1.1) Not Selected      Unknown 0.98 (0.70-1.4) Not Selected      <20 patients Reference Not SelectedOne-family households (per 10%) 0.97 (0.95-1.0) Not SelectedPopulation density (per 10k) 0.89 (0.82-0.97) Not SelectedImmigrants (per 10%) 0.97 (0.94-1.0) Not SelectedMedian Income (per $10k) 0.65 (0.60-0.71) Not SelectedSingle people (per 10%) 1.0 (1.0-1.1) Not SelectedPost-secondary certificate (per 10%) 0.81 (0.77-0.85) Not SelectedUnemployment rate (per 10%) 0.82 (0.78-0.86) Not SelectedPercentage aboriginal ancestry      ≥10% 1.5 (1.3-1.8) 1.2 (1.1-1.5)      5%-<10% 0.85 (0.71-1.0) 0.87 (0.72-1.1)      <5% Reference Reference

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.24.1-0357604/manifest

Comment

Related Items