Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Improving risk prediction for the malignant transformation of low-grade oral dysplasia - clinicopathological… Rock, Leigha Duree 2018

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_november_rock_leigha.pdf [ 5.02MB ]
Metadata
JSON: 24-1.0372321.json
JSON-LD: 24-1.0372321-ld.json
RDF/XML (Pretty): 24-1.0372321-rdf.xml
RDF/JSON: 24-1.0372321-rdf.json
Turtle: 24-1.0372321-turtle.txt
N-Triples: 24-1.0372321-rdf-ntriples.txt
Original Record: 24-1.0372321-source.json
Full Text
24-1.0372321-fulltext.txt
Citation
24-1.0372321.ris

Full Text

IMPROVING RISK PREDICTION FOR THE MALIGNANT TRANSFORMATION OF LOW-GRADE ORAL DYSPLASIA – CLINICOPATHOLOGICAL FEATURES AND LOSS OF HETEROZYGOSITY  by  Leigha Duree Rock  B.D.Sc., The University of British Columbia, 2014  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Craniofacial Science)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)   September 2018 © Leigha Duree Rock, 2018 ii The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled: Improving risk prediction for the malignant transformation of low-grade oral dysplasia – Clinicopathological features and loss of heterozygosity   submitted by Leigha D. Rock in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Craniofacial Science  Examining Committee: Dr. Denise M. Laronde Supervisor  Dr. Miriam P. Rosin  Supervisory Committee Member  Dr. Lewei Zhang Supervisory Committee Member Dr. Ian Matthew  University Examiner Dr. Kristin Campbell University Examiner  Additional Supervisory Committee Members: Dr. Batoul Shariati Supervisory Committee Member iii Abstract  A major barrier to oral cancer prevention is the lack of risk predictors for the malignant progression of oral potentially malignant lesions (OPML). OPML with evidence of dysplasia are at risk of progressing to oral cancer. However, not all will progress and predicting which low-grade dysplasia (LGD; mild/moderate dysplasia) are at risk of progression is challenging. The overall goal of this thesis was to advance risk stratification and to improve the prediction of malignant progression in LGD. Three research projects were developed to accomplish this goal. Each identified important insights into the phenotypic changes associated with malignant transformation and advanced risk prediction by exploring the association between histological, clinical and molecular biomarkers and malignant progression. The first project revealed that dysplasia with or without lichenoid mucositis (LM) had similar cancer risk and that pathologists and clinicians should not discount dysplasia in the presence of LM. The second project compared the clinical and molecular features of LGD in smokers in contrast to those of non-smokers (NS) and confirmed that NS possess an increased risk of progression, and progressed more quickly, than smokers. These findings emphasize the need for clinicians to consider smoking history (or the lack thereof) and molecular profiles in the triage and management of LGD. The final project aimed to advance a risk prediction model using microsatellite analysis for loss of heterozygosity (LOH) and repeated measures of clinicopathological features. Multivariable analysis showed that after LOH risk category, temporal repeated measures of toluidine blue status was the most significant predictor of progression. Two risk prediction models are presented and provide a systematic decision-making process for these very heterogeneous group of lesions. Patients at higher risk could be offered intensified surveillance or targeted interventions based on their iv predicted risk of disease, while patients at low risk would be spared from excessive screening and treatment. This body of work has advanced the risk stratification of LGD and presents an important framework to give scientists and clinicians a better view into the natural history of the disease and a novel approach to integrate repeated measurements of change over time into risk models.   v Lay Summary A major barrier to oral cancer prevention is the lack of ability to predict the risk of cancer developing from precancerous lesions. The overall goal of this thesis was to improve risk assessment in oral precancerous lesions so that appropriate treatment and management can be customized to each patient based on their individualized risk. Three research projects are presented; each advances risk prediction by studying different connections between microscopic diagnosis, molecular features, risk habits, clinical lesion characteristics and progression to cancer. The risk models presented provide a systematic decision-making process for the management of this very diverse group of lesions and has the potential to improve outcome while maximizing health system resources and cost-effectiveness. It also provides a framework to give scientists and clinicians a better view into the natural history of the disease and a new way to integrate repeated measurements of change over time into risk models.  vi Preface  This thesis is an original intellectual product of the author L. Rock. All works presented in this thesis involved a longitudinal cohort enrolled in the Oral Cancer Prevention Longitudinal (OCPL) study, developed by the BC Oral Cancer Prediction Program. The University of British Columbia (UBC) and British Columbia Cancer Agency (BCCA) joint Research Ethics Board approved the work contained within this thesis, certificate number H98-61224, entitled “Clonal Changes in Oral Lesions of High-Risk Patients.” In addition to all institutional safety training and certification, I completed certification in the Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans Course on Research Ethics (TCPS 2: CORE), from the Panel on Research Ethics.   This thesis consists of six chapters. The first chapter provides the introduction and background. The second chapter provides details on the methodology and the techniques used in the projects presented in this thesis.  The next three chapters present three papers that have ensued from the work completed for this thesis. The final chapter is a general discussion.   A version of Chapter 3 has been published. Rock, L. D., Laronde, D. M., Lin, I., Rosin, M. P., Chan, B., Shariati, B. and Zhang, L. Dysplasia should not be ignored in lichenoid mucositis. J Dent Res. 2018 Jul;97(7):767-772. I was responsible for the study inception, study design and for obtaining the necessary ethical and institutional certificates. The University of British Columbia (UBC) and British Columbia Cancer Agency (BCCA) joint Research Ethics Board approved the work contained within this chapter - certificate number H17-01452, entitled vii “Malignant Progression of Oral Lichenoid Dysplasia.” I participated in data acquisition and supervised Lin, I. during the data analysis, statistical analysis and interpretation phases; I contributed significantly to the drafting of the manuscript. Zhang, L. contributed substantially to the Introduction and Discussion sections of the manuscript.   Chapter 4 presents an article that has been published: Rock, L., Rosin, M. P., Zhang, L., Chan, B., Shariati, B., and Laronde, D. M. Characterization of epithelial oral dysplasia in non-smokers: First steps towards precision medicine. Oral Oncol. 2018; (78): 119- 125. I was responsible for the data mining, data analysis, statistical analysis and interpretation and as lead author, I contributed extensively to the manuscript preparation and editing.     Chapter 5 is based on work conducted in the British Columbia Oral Cancer Prevention Program, in conjunction with the University of British Columbia and Simon Fraser University by Dr. D.M. Laronde, Dr. M. P. Rosin, Dr. L. Zhang. Dr. B. Shariati and Leigha D. Rock. In completing this work, I spent several years in the clinic and in the lab. In addition to learning and assuming many administrative and managerial roles in the program, I was responsible for performing clinical assessments and contributing to clinical data gathering in the last five years of follow-up of the longitudinal cohort. I also contributed to three years of the molecular analysis by performing microdissection, deoxyribonucleic acid (DNA) extraction, running the microsatellite assay, and scoring the results. I contributed substantively to the study design, data management and statistical analysis of the research data of this project. Dr. F. Pourmelek provided consultation and assistance with the advanced statistical modelling required.   viii Table of Contents  Abstract ......................................................................................................................................... iii Lay Summary .................................................................................................................................v Preface ........................................................................................................................................... vi Table of Contents ....................................................................................................................... viii List of Tables .............................................................................................................................. xvi List of Figures ........................................................................................................................... xviii List of Symbols ........................................................................................................................... xix List of Abbreviations ................................................................................................................. xxi Acknowledgements ....................................................................................................................xxv Dedication ................................................................................................................................. xxvi Chapter 1: Introduction and Literature Review.........................................................................1 1.1 Overview ......................................................................................................................... 1 1.2 Epidemiology .................................................................................................................. 2 1.2.1 Epidemiology of Oral Cancer ................................................................................. 3 1.2.1.1 Global .................................................................................................................. 3 1.2.1.2 North America .................................................................................................... 4 1.2.1.3 British Columbia ................................................................................................. 5 1.2.2 Epidemiology of Oral Potentially Malignant Lesions ............................................ 5 1.3 Etiology ........................................................................................................................... 7 1.3.1 Tobacco and Alcohol .............................................................................................. 8 1.3.2 Betel Nut ............................................................................................................... 10 ix 1.3.3 Infectious Agents .................................................................................................. 10 1.3.4 Immunosuppression .............................................................................................. 11 1.3.5 Age and Gender .................................................................................................... 12 1.3.6 Other Risk Factors ................................................................................................ 12 1.4 Histopathology and Histological Progression Model of Malignant Transformation.... 13 1.5 Oral Potentially Malignant Lesions .............................................................................. 17 1.5.1 Leukoplakia........................................................................................................... 18 1.5.2 Erythroplakia and Erythroleukoplakia .................................................................. 19 1.5.3 Oral Lichen Planus and Lichenoid Mucositis ....................................................... 19 1.5.4 Other Oral Potentially Malignant Lesions ............................................................ 20 1.6 Malignant Transformation of Oral Potentially Malignant Lesions ............................... 21 1.6.1 Transformation Rates ............................................................................................ 21 1.6.1.1 Histological Grade and Risk of Malignant Transformation ............................. 22 1.6.1.2 Clinicopathological Features and Association with Malignant Progression .... 23 1.6.1.2.1 Anatomical site ........................................................................................... 23 1.6.1.2.2 Appearance ................................................................................................. 24 1.6.1.2.3 Lesion Size and Number of Lesions ........................................................... 25 1.6.1.2.4 Duration ...................................................................................................... 26 1.6.2 Field cancerization ................................................................................................ 26 1.6.3 Invasion and Metastasis ........................................................................................ 27 1.6.4 Tumour Staging .................................................................................................... 28 1.7 Management of Oral Epithelial Dysplasia .................................................................... 30 1.7.1 Risk Factor Modification ...................................................................................... 31 x 1.7.2 Surgical Treatment ................................................................................................ 31 1.7.3 Medical Treatment ................................................................................................ 33 1.7.3.1 Chemotherapeutics ............................................................................................ 33 1.7.3.2 Chemoprevention .............................................................................................. 33 1.7.3.3 Immunoprevention ............................................................................................ 34 1.7.4 Surveillance........................................................................................................... 34 1.8 Adjunctive Clinical Aids............................................................................................... 35 1.8.1.1 Toluidine Blue .................................................................................................. 35 1.8.1.2 Fluorescence Visualization ............................................................................... 36 1.9 Biomarkers of Risk Prediction ...................................................................................... 38 1.9.1.1 Overview of Molecular Biomarkers ................................................................. 38 1.9.1.2 Mutation as a Driving Force for Carcinogenesis and Loss of Heterozygosity . 40 Chapter 2: Methodology..............................................................................................................43 2.1 OCPL Study and Patient Population ............................................................................. 43 2.1.1 Participant Selection ............................................................................................. 44 2.2 Histological Evaluation ................................................................................................. 44 2.3 Molecular Evaluation .................................................................................................... 45 2.3.1 Sample Cutting and Preparation ........................................................................... 45 2.3.1.1 Hematoxylin and Eosin (H&E) Slide Preparation ............................................ 45 2.3.1.2 Methyl Green Slide Preparation........................................................................ 46 2.3.2 Microdissection ..................................................................................................... 46 2.3.3 DNA Extraction .................................................................................................... 47 2.3.4 Loss of Heterozygosity Microsatellite Assay ....................................................... 48 xi 2.3.4.1 Microsatellite Markers ...................................................................................... 48 2.3.4.2 End-Labelling and Dinucleotide PCR Reaction ............................................... 50 2.3.4.3 Casting the Polyacrylamide Gel........................................................................ 50 2.3.4.4 Running LOH.................................................................................................... 51 2.3.4.5 Scoring LOH ..................................................................................................... 52 2.4 Demographic Data ........................................................................................................ 53 2.5 Risk Habit Assessment ................................................................................................. 53 2.6 Clinical Evaluation........................................................................................................ 54 2.6.1 Initial Visit ............................................................................................................ 54 2.6.1.1 White Light Examination .................................................................................. 54 2.6.1.2 Fluorescence Visualization Examination .......................................................... 55 2.6.1.3 Toluidine Blue Examination ............................................................................. 55 2.6.1.4 Digital Photographs .......................................................................................... 55 2.6.1.5 Exfoliative Cytology ......................................................................................... 56 2.6.2 Follow Up Visits ................................................................................................... 56 2.7 Outcome ........................................................................................................................ 57 2.7.1 Biopsy ................................................................................................................... 57 2.7.2 Histological Evaluation ......................................................................................... 57 Chapter 3: (Paper 1). “Dysplasia should not be ignored in lichenoid mucositis.” Journal of Dental Research. 2018 Jul;97(7)767-772. .....................................................................................59 3.1 Synopsis ........................................................................................................................ 59 3.2 Objective ....................................................................................................................... 60 3.3 Hypotheses .................................................................................................................... 60 xii 3.4 Published Paper ............................................................................................................. 61 3.4.1 Introduction ........................................................................................................... 61 3.4.2 Materials and Methods .......................................................................................... 63 3.4.2.1 Patient Population ............................................................................................. 63 3.4.2.2 Clinical pathological data, treatment and follow-up ......................................... 64 3.4.2.3 Statistical analysis and Reporting ..................................................................... 64 3.4.3 Results ................................................................................................................... 65 3.4.4 Discussion ............................................................................................................. 72 3.4.5 Conclusion ............................................................................................................ 75 3.4.6 Acknowledgments................................................................................................. 75 Chapter 4: (Paper 2). “Characterization of epithelial oral dysplasia in non-smokers: First steps towards precision medicine”. Oral Oncology. 2018; 78:119-125. ....................................76 4.1 Synopsis ........................................................................................................................ 76 4.2 Objectives ..................................................................................................................... 77 4.3 Hypotheses .................................................................................................................... 77 4.4 Published Paper ............................................................................................................. 77 4.4.1 Introduction ........................................................................................................... 77 4.4.2 Materials and Methods .......................................................................................... 80 4.4.3 Results ................................................................................................................... 83 4.4.3.1 Sociodemographic and Lifestyle Characteristics .............................................. 83 4.4.3.2 Clinicopathological Features ............................................................................ 84 4.4.3.3 Outcome ............................................................................................................ 88 4.4.4 Discussion ............................................................................................................. 91 xiii 4.4.5 Conclusion ............................................................................................................ 95 4.4.6 Acknowledgement ................................................................................................ 96 Chapter 5: (Paper 3). “Molecular analysis and repeated measures of clinical risk indicators – building a framework to predict the malignant transformation of low-grade oral dysplasia”. .....................................................................................................................................97 5.1 Synopsis ........................................................................................................................ 97 5.2 Introduction ................................................................................................................... 98 5.2.1 Background and Rationale .................................................................................... 98 5.2.2 Objectives ........................................................................................................... 101 5.2.3 Hypotheses .......................................................................................................... 101 5.3 Methods....................................................................................................................... 102 5.3.1 Study Design ....................................................................................................... 102 5.3.2 Setting and Participants....................................................................................... 102 5.3.3 Data Collection ................................................................................................... 103 5.3.3.1 Histological Diagnosis .................................................................................... 103 5.3.3.2 Molecular Assessment .................................................................................... 103 5.3.3.3 Clinicopathological Variables ......................................................................... 104 5.3.3.4 Outcome .......................................................................................................... 105 5.3.4 Statistical Methods .............................................................................................. 105 5.3.4.1 Sample Size ..................................................................................................... 105 5.3.4.2 Univariate Analysis ......................................................................................... 106 5.3.4.2.1 Pure Longitudinal Analysis....................................................................... 107 5.3.4.2.2 Cumulative Counts Longitudinal Analysis ............................................... 107 xiv 5.3.4.3 Multivariable Analysis .................................................................................... 108 5.3.4.4 Risk Classification Models ............................................................................. 108 5.4 Results ......................................................................................................................... 109 5.4.1 Participant Selection ........................................................................................... 109 5.4.2 Descriptive Data.................................................................................................. 110 5.4.3 Outcome Data ..................................................................................................... 114 5.4.3.1 Time-invariant Clinicopathological Characteristics ....................................... 114 5.4.3.2 Time-variant Clinicopathological Characteristics .......................................... 117 5.4.3.2.1 Pure – Longitudinal Analysis.................................................................... 117 5.4.3.2.2 Cumulative Counts Analysis..................................................................... 122 5.4.3.3 Comparison of Baseline and Repeated Clinicopathological Measurements in the Prediction of Outcome .............................................................................................. 128 5.4.4 Correlation of Time-Variant Clinicopathological Characteristics and LOH ...... 129 5.4.5 Risk Classification Modeling .............................................................................. 131 5.4.5.1 Pure Longitudinal Analysis............................................................................. 132 5.4.5.2 Cumulative Counts Analysis........................................................................... 135 5.5 Discussion ................................................................................................................... 141 5.6 Conclusions ................................................................................................................. 149 Chapter 6: General Discussion .................................................................................................151 6.1 Summary of goals and main findings ......................................................................... 151 6.2 Integration and Significance ....................................................................................... 154 6.3 Comments on strengths and limitations of the thesis research ................................... 158 6.4 Future Directions ........................................................................................................ 161 xv References ...................................................................................................................................164 Appendix A Clinicopathological Data Collection Tools ........................................................ 187 A.1 Sample Oral Biopsy Service Pathology Report ...................................................... 187 A.2 Initial Questionnaire................................................................................................ 188 A.3 Standardized Oral Map ........................................................................................... 192 A.4 Lesion Tracking Sheet ............................................................................................ 193 A.5 Oral Biopsy Service Requisition Form ................................................................... 195 Appendix B Additional Publications ...................................................................................... 196  xvi List of Tables  Table 1.1 Architectural and cytological changes associated with dysplasia ................................ 14 Table 1.2 Histopathological stages in oral epithelial lesions ........................................................ 15 Table 1.3 Classification systems for oral epithelial lesions .......................................................... 17 Table 1.4 TNM staging of oral cancer .......................................................................................... 29 Table 1.5 Stage grouping .............................................................................................................. 30 Table 2.1 Primer details ................................................................................................................ 49 Table 3.1 Comparison of oral epithelial dysplasia with and without lichenoid mucositis according to demographic, risk habit information and clinical features ....................................... 65 Table 3.2 Distribution of cases according to outcome .................................................................. 68 Table 3.3 Probability of progression in low-grade dysplasia with and without lichenoid mucositis....................................................................................................................................................... 72 Table 4.1 Distribution of cases according to sociodemographic and lifestyle variables .............. 84 Table 4.2 Clinicopathological and histopathological features according to smoking status and according to outcome .................................................................................................................... 86 Table 4.3 Distribution of risk factor variables according to outcome .......................................... 88 Table 4.4 Probability of progression in smokers versus non-smokers ......................................... 90 Table 5.1 Sociodemographic characteristics of patients by progression .................................... 112 Table 5.2 Time-invariant clinicopathological characteristics by progression ............................ 115 Table 5.3 Time-variant clinicopathological characteristics by progression – Pure longitudinal analysis£ ...................................................................................................................................... 119 xvii Table 5.4 Determinants of progression in Cox proportional hazards regression – Pure longitudinal analysis ................................................................................................................... 121 Table 5.5 Time-variant clinicopathological characteristics by progression – Cumulative counts analysis£ ...................................................................................................................................... 124 Table 5.6 Determinants of progression in Cox proportional hazards regression – Cumulative counts analysis ............................................................................................................................ 127 Table 5.7 Risk prediction of baseline measurements compared to repeated measurements in clinicopathological features ........................................................................................................ 129 Table 5.8 Time-variant clinicopathological characteristics by molecular risk category ............ 131 Table 5.9 Performance of the pure longitudinal analysis model ................................................ 134 Table 5.10 Probability of progression - pure longitudinal analysis risk categories .................... 135 Table 5.11 Performance of the cumulative counts analysis model ............................................. 138 Table 5.12 Probability of progression – cumulative counts analysis.......................................... 140 Table 5.13 Cumulative counts analysis detailed report of sensitivity and specificity ................ 148 xviii List of Figures  Figure 3.1 The proportion of malignant progression was similar between low-grade dysplasia with lichenoid mucositis (LM) and those without LM. ................................................................ 70 Figure 3.2 Low-grade dysplasia with or without lichenoid mucositis possess similar cancer risk....................................................................................................................................................... 71 Figure 4.1 Kaplan-Meier plot of time to progression in smokers vs. non-smokers...................... 90 Figure 4.2 Cox proportional hazard regression model analysis for LOH risk pattern in non-smokers compared to smokers. ..................................................................................................... 91 Figure 5.1 Participant Selection Process ..................................................................................... 110 Figure 5.2 Pure longitudinal analysis risk stratification model .................................................. 134 Figure 5.3 Pure longitudinal analysis Kaplan-Meier survival estimates .................................... 135 Figure 5.4 Cumulative counts analysis risk stratification model ................................................ 138 Figure 5.5 Cumulative counts analysis Kaplan-Meier survival estimates .................................. 139 Figure 5.6 Comparison of previously validated 2012 LOH risk model and cumulative counts temporal longitudinal model ....................................................................................................... 146 Figure 6.1 LOH and smoking history risk stratification model .................................................. 155 Figure 6.2 LOH and smoking history risk stratification Kaplan-Meier survival estimates ........ 156 Figure 6.3 Comparison of each of the project cohorts with each other ...................................... 157 Figure 6.4 Comparison on each of the project cohorts with that of the cohort from Zhang et al., 2012............................................................................................................................................. 157 Figure 6.5 2015 Global smoking prevalence .............................................................................. 160  xix List of Symbols °C   degrees Celsius ddH20  double distilled water  E6   E6 protein encoded by human papillomavirus E7   E7 protein encoded by human papillomavirus INFα   interferon alpha Ki-67  protein Ki-67 L  litre L0/2  loss on top allele, loss of heterozygosity L1/0  loss on bottom allele, loss of heterozygosity mA   milliamps mL  millilitre mm   millimeters mmol   millimoles NH4  ammonium ng  nanograms pH   power of hydrogen  p16  protein 16 p53  protein 53 pRB  protein retinoblastoma R  retention, no loss of heterozygosity  µl  microlitres µm  micrometre xx V  volts W   watts  [ γ-32P] gamma phosphate group with radioactive phosphorus - isotope 32 xxi List of Abbreviations  95% CI   95% Confidence Interval Ag-NOR  argyrophilic nucleolar organizer region APS   ammonium persulfate ASIR   age-standardized incidence rate ANOVA  analysis of variance ATP, [ γ-32P]  adenosine triphosphate, labelled on the gamma phosphate group with 32P BCCA   British Columbia Cancer Agency BCCRC  British Columbia Cancer Research Centre  BC OCPP  British Columbia Oral Cancer Prevention Program BSA    bovine serum albumin CDK2NA  Cyclin-dependent kinase inhibitor 2A CIS   carcinoma in situ COX-2  cyclooxygenase-2 CS   continuing smoker DNA    deoxyribonucleic acid dNTP   deoxyribose nucleoside triphosphate dATP   deoxyribose adenosine triphosphate  dGTP   deoxyribose guanosine triphosphate dCTP    deoxyribose cytidine triphosphate dTTP   deoxyribose thymidine triphosphate EGFR   epidermal growth factor receptor xxii EDTA   ethylenediaminetetraacetic acid FAD   flavin adenine dinucleotide FFPE   formalin-fixed paraffin-embedded  FOD   Faculty of Dentistry FS   former smoker FV   florescence visualization  gMART  genomic marker-based test H&E   haematoxylin and eosin HIV   human immunodeficiency virus HNSCC   head and neck squamous cell carcinoma HPV   human papillomavirus HR    hazard ratio IARC   International Agency for Research on Cancer LD   lichenoid dysplasia LGD   low-grade dysplasia; mild or moderate dysplasia LM   lichenoid mucositis LOH   loss of heterozygosity LSA   lesion site A LSB    lesion site B  MMP   matrix metallopeptidase NADH   nicotinamide adenine dinucleotide NI    non-informative  NS   non-smoker xxiii OBS    Oral Biopsy Service OCC   oral cavity cancer OCPL    Oral Cancer Prediction Longitudinal (Study) OED    oral epithelial dysplasia OLP    oral lichen planus OPC   oropharyngeal cancer OPML   oral potentially malignant lesion OSCC    oral squamous cell carcinoma OSF    oral submucous fibrosis OR    odds ratio PCNA   proliferating cell nuclear antigen  PCR   polymerase chain reaction  PK   proteinase K  Rb   Retinoblastoma  SCC   squamous cell carcinoma SEER    Surveillance Epidemiology and End Results SES    socioeconomic status TB   toluidine blue TB+   toluidine blue positive  TB-    toluidine blue negative  TEMED  tetramethylethylenediamine TNM   Tumour Node Metastases staging system TP53   Tumour protein p53 gene xxiv Tris    tris-hydroxymethyl aminomethane TSG   tumour suppressor gene UBC   University of British Columbia VCC   Vancouver Cancer Centre VGH   Vancouver General Hospital  WHO    World Health Organization WLE    white light examination   xxv Acknowledgements  I would like to express my sincerest gratitude to my supervisor, Dr. Denise Laronde, for her guidance, continuous support, encouragement, and the long, late nights she puts in. I owe special thanks to Dr. Miriam Rosin, whose committed mentorship allowed me to achieve more than I could have ever imagined. I would like to thank committee members, Dr. Lewei Zhang and Dr. Batoul Shariati for their assistance and encouragement. You are all brilliant, dedicated researchers and teachers. This has been much more than just an academic journey. “Thank you" is not good enough to show how grateful I am.   The work presented in this thesis was supported by grants from the British Columbia Cancer Foundation, the National Institutes of Health (R01DE13124), the National Institute of Dental and Craniofacial Research (R01DE17013), the Canadian Institutes of Health Research Doctoral Award (379723), the Canadian Foundation for Dental Hygiene Research and Education, the Alpha Omega Foundation, and the University of British Columbia Faculty of Dentistry Summer Research Studentship program.   I offer my gratitude to the faculty, staff and my fellow students at UBC and the British Columbia Oral Cancer Prevention Program. I extend special thanks to Dr. Bertrand Chan for his clinical mentorship, and to Dr. Farshad Pourmelek for his assistance in the statistical analysis.   Last, but not least, I would like to especially thank my family. I could not have done it without you.  xxvi Dedication  To Keith. Thank you for letting me have the plums…they were so sweet, and so, so cold.  1 Chapter 1: Introduction and Literature Review 1.1 Overview The global burden of oral cancer is high, with an estimated 300,000 new cases and 145,000 deaths in 2012,(1) and a poor survival rate mainly due to late stage diagnosis.(2, 3) Early detection is vital to the improvement of this prognosis.(4) A clinically visible oral potentially malignant lesion (OPML) often precedes malignant disease.(5, 6) Even when identified histologically, knowing which OPML to treat can be difficult. OPMLs with evidence of oral epithelial dysplasia (OED) are at greatest risk of progressing to oral cancer, with the risk rising with increasing degree of dysplasia. However, not all OED will progress to cancer.(7-9) While severe dysplasia has been shown to be at an elevated risk of malignant transformation only a small proportion of low-grade dysplasia (LGD, mild or moderate dysplasia) will progress.(10-12)  This creates a challenge for the clinical management of these lesions, which represent the majority of dysplasia.(13) Treatment can bear significant morbidity, and given that most low-grade lesions will not progress, it is correspondingly important to avoid overtreatment.   Differentiating between those LGD that are at high risk of progression from those at low risk of progressing to squamous cell carcinoma (SCC) is difficult,(14-16) and is a major barrier to improving outcome in this disease.(13, 17) Equally, deciding when to do a comparative biopsy of an OPML under surveillance is challenging. Finally, the prediction of time to progression is uncertain with some lesions progressing quickly, as others progress slowly. Most cancer has a window of opportunity during which it can be detected by screening. However, screening tests are more effective at detecting slowly growing neoplasms than those that progress rapidly.(18) Over diagnosis occurs when neoplasms that would never progress, or progress so slowly they 2 would not be a cause of death or cause symptoms in one’s lifetime, are identified and treated.(19) Currently, limited prognosticators exist which can identify those lesions that are likely to progress and require intervention from those that will naturally regress or remain stable. A gap continues to exist in the current knowledge of malignant risk prediction and management of LGD.  The overall goal of this thesis is to summarize what is currently known about the risk and management of LGD and to advance the current risk prediction model for the malignant progression of oral LGD. Three studies are detailed herein. The objectives and specific aims of each study, and how these objectives work towards this overall goal, are presented within each of the specific chapters.   By identifying and understanding the early indicators of progression, regression and invasion, we will advance the establishment of early detection biomarkers of high-risk disease that can be used to more effectively intervene in the disease course. The search for additional markers and the refining of risk models is critical to establishing such intervention strategies. It will also allow better treatment and a more successful targeting of interventions to high-risk lesions in order to intercept disease.   1.2 Epidemiology  Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer in the world.(1, 20) Cancers of the oral cavity include malignant neoplasms of the lip, tongue, gingiva, palate, floor of mouth, cheek, vestibule, and other non-specified parts in the mouth (coded ICD-3 10: C00-06), and oropharynx (coded ICD-10: C09-10 by the International Statistical Classification of Diseases and Related Health Problems).(21)  More than 90% of all oral cavity cancers (OCC) and oropharyngeal cancers (OPC) arise from the stratified squamous epithelium of the lining mucosa and are SCC.(22, 23) Although the oral cavity is easily accessible and can be directly examined, almost two thirds of patients are diagnosed at late stage, with metastasis to lymph nodes or other parts of the body, leading to high morbidity and mortality.(23) Overall, five-year survival rates for cancers of the oral cavity and oropharynx are approximately 50% to 60%,(20, 22, 24-26) and survival rates have not improved over the past three decades.(2, 22) Stage at presentation significantly affects five-year survival, and the poor prognosis for oral cancer is mainly due to late stage diagnosis.(20) In general, prognosis decreases with advanced disease, advanced age, low socioeconomic status (SES) and the continuing usage of known risk factors such as tobacco, excessive alcohol, and betel quid.(20)   1.2.1 Epidemiology of Oral Cancer 1.2.1.1 Global Worldwide, the annual estimated incidence is approximately 300,000 for oral cavity and 142,000 for oropharyngeal cancers, which represents about 3.1% of the world’s cancer burden.(1) There is a wide geographical variation in the incidence of these cancers. Two-thirds of OCC cases occur in developing countries.(20) The highest incidence rates are found in South and Southeast Asia (Sri Lanka, India, Pakistan, and Taiwan), where oral SCC (OSCC) can account for up to one quarter of all malignancies.(1, 27) Parts of Eastern Europe (Hungary, Slovakia and Slovenia), Latin America and the Caribbean (Brazil, Uruguay and Puerto Rico), and France also possess high rates.(20) In contrast, the highest incidence rates for OPC comes from economically 4 developed countries, and the incidence of OPC is increasing significantly in these countries.(28) Globally, oral cavity and oropharyngeal cancers are more common in men than in women; two-thirds of OCC and roughly 80% of OPC occurred in men.(1) Worldwide, 125,000 OCC deaths and 97,000 OPC deaths occurred (3.0% of the world estimated cancer deaths) in 2012.(1) Mortality is disproportionately higher in less developed regions as compared to more developed regions of the world.(1)   1.2.1.2 North America In the United States, OCC and OPC are the eighth most common cancer among men, and the fourteenth most common among women.(29) It has been estimated that 51,540 new cases of OCC and OPC will be diagnosed in the United States in 2018, and approximately 10,030 people will die from this disease.(4, 30) These figures represent approximately 3.0% of all malignancies, and 1.6% of all cancer deaths in the United States. The age-standardized incidence rate (ASIR) for these cancers in the United States is 11.2 new cases per 100,000 people. Based on 2013 – 2015 data, roughly 1.2% of Americans will be diagnosed with OCC and OPC during their lifetime.(30) Overall, the 5-year survival rate is 64.8%. For localized disease the rate is 83.7%, and for distant (metastatic) disease the rate declines steeply to 38.5%.(30) The median age of diagnosis is 63 years, and the number of new cases in males is about 2.5 times greater than in females.(30) With respect to prevalence, in 2015, there was an estimated 359,718 people in the United States living with OCC and OPC.(30)   Canadian statistics are comparable to those of the United States; in Canada, OCC is the ninth most common cancer among males and the fourteenth most common cancer among females.(31) 5 The estimated incidence for OCC in 2017 is 4700 (ASIR of 11.9 cases per 100,000 people), and the estimated number of deaths is 1200. Twice as many males are diagnosed with OCC as compared to females. In 2009, the estimated prevalence of oral cancer in Canada was 19,510.(31)   1.2.1.3 British Columbia In British Columbia, the incidence and mortality rates are consistent with the rest of the country.(32) In 2015, the estimated incidence for OCC and OPC in British Columbia was 675. The ratio of males to females was 2.4:1. The ASIR was 20.4 per 100,000 people for males and 8.4 per 100,000 people for females. The number of OCC and OPC deaths in 2015 was estimated to be 185, which computed to a mortality rate of 3.9 per 100,000 people.(32)   1.2.2 Epidemiology of Oral Potentially Malignant Lesions  A clinically visible premalignant, or potentially malignant, lesion often precedes OSCC.(5, 7, 14, 22, 33, 34) An OPML is a “morphologically altered tissue in which cancer is more likely to occur than in its apparently normal counterpart”(35); such lesions are characterized by histological and /or genetically altered tissue changes and have increased risk to develop cancer than a normal tissue.(6, 36, 37) According to Warnakulasuriya et al.(6), in the past, “the terms ‘pre-cancer’, ‘precursor lesions’, ‘premalignant’, ‘intra epithelial neoplasia’ and ‘potentially malignant’ have been used in the literature to describe various clinical lesions that may have the potential to become cancer.” At a symposium organized by the World Health Organization (WHO) Collaborating Centre for Oral Cancer and Precancer, an expert group examined the terminology, definitions and classification of oral precancers. The recommendation of the working group was 6 to adopt the term ‘potentially malignant’ to refer to precancer, “as it conveys that not all lesions and conditions described under this term may transform to cancer.”(6) It is established that cancers may arise from longstanding OPML, but it is not known whether this is true for all oral cancers.(5, 6, 15)   OPML include leukoplakia, erythroplakia, erythroleukoplakia, oral submucous fibrosis (OSF), actinic keratosis, and oral lichen planus (OLP).(6) The most prevalent OMPL is leukoplakia, which at present, is defined as “a white plaque of questionable risk having excluded (other) known diseases or disorders that carry no increased risk for cancer”.(14) Erythroplakia can be defined as a “fiery red patch that cannot be characterized clinically or pathologically as any other definable disease.”(36) Erythroleukoplakia are a combination of leukoplakia and erythroplakia, possessing both red and white components.(6) Leukoplakia, erythroplakia, erythroleukoplakia, and their clinical sub-types are discussed in further detail in section 1.5. OSF is a chronic disorder characterized by fibrosis of the submucosal connective tissues of the oral cavity and oropharynx. The fibrosis causes the oral tissues to become rigid and results in limited opening.(38) It is associated with betel quid chewing, a practice that is more common in South Asian communities worldwide.(39) OLP is a chronic inflammatory disorder of the skin and mucous membranes. It is considered to be an autoimmune condition in which T-lymphocytes accumulate beneath the epithelium of the oral mucosa resulting in hyperkeratosis, erythema, and ulceration. Various subtypes exist, some of which are associated with increased risk of malignant transformation.(5-7, 40)   7 It is difficult to estimate the incidence rate, or number of new OPML developing each year, as there are few studies in the literature. Prevalence, on the other hand, which measures the proportion of a population that have an OPML at a particular time, has been documented to a somewhat greater degree. A systematic review carried out by Petti (41) in 2003, aimed at reporting the global estimated pooled prevalence of leukoplakia based on 26 primary studies published between 1986 and 2002. Prevalence values ranged between 0.5% and 26.92%. The pooled point prevalence and 95% confidence interval was estimated to be 1.49% (95% CI, 1.42 -1.56) using a weighted average, inverse variance method, and 2.60% (95% CI, 1.72 - 2.74) using a random effect method. The first method does not account for the potential between-studies heterogeneity, and therefore the author suggests that the second estimate is more likely to possess higher reliability.(41) A relatively recent, large study conducted by Villa et al.,(42) aimed to estimate the prevalence of OPML in a large population of dental patients. Subjects were accrued from oral medicine clinics at Boston University School of Dental Medicine (n=3142). Using the WHO diagnostic criteria, (6, 43) 0.9% of these subjects were diagnosed with a biopsy confirmed OPML (one OSF, three dysplasia, 14 hyperplasia and nine OLP).(42)   1.3 Etiology The development of oral cancer and its purported precursor lesion, OPML, is multifaceted and is interrelated to both external factors such as nutrition and exposure to potential carcinogens such as tobacco and alcohol, as well as internal factors such as systemic health, nutrition, age and genetics. In western countries, the most significant risk factors identified so far are tobacco use and alcohol consumption, which appear to act synergistically.(7, 27, 44) The use of betel quid is also associated significantly with both OSCC and OPML.(27, 45, 46) More recently, oral human 8 papillomavirus (HPV) infection has been recognized as a cause of a distinct subset of OPC that is rising drastically in incidence, and presents with non-traditional risk factors.(47-50) A number of OPML have no obvious cause and have no association with these etiological factors and can be classified as idiopathic.(5, 51, 52)   1.3.1 Tobacco and Alcohol Tobacco is considered one of the most significant risk factors for OSCC.(20, 53-55) The International Agency for Research on Cancer (IARC) has evaluated the evidence on tobacco and has stated that there is ‘sufficient’ evidence to show a causal relationship and that tobacco smoking and smokeless tobacco increase the risk of oral cancer.(55) Chemical analysis has shown that tobacco and tobacco smoke contain thousands of chemicals that come into contact with the oral mucosa, including those that are, genotoxic, mutagenic, carcinogenic, or have immunomodulatory effects.(56, 57) Some of the carcinogens found in tobacco and tobacco smoke include nitrosamines, polycyclic aromatic hydrocarbons and aromatic amines.(58) Risk is dose-dependent and increases with both quantity (i.e. number of cigarettes per day) and duration (i.e. number of years) of tobacco use.(57, 59) Whole-exome sequencing has also shown that smoking is associated with increased tumour suppressor gene TP53 and other mutations found in HNSCC.(60, 61) OPML are also associated with the use of tobacco.(5, 16, 62, 63) The amount and frequency of tobacco use is associated with OMPL; heavy smokers are seven times more likely to have OPML than non-smokers.(5, 63) There is a considerable reduction in the risk of OSCC with smoking cessation. After ten years of abstinence, former smokers possess a similar risk of OSCC as non-smokers.(64) The disappearance or regression of many OPML following 9 tobacco cessation further demonstrates the significance of tobacco in the etiology of OPML and OSCC.(5)   Alcohol consumption is recognized as an independent risk factor for OSCC(53, 54, 65-67) and has shown a dose-response relationship with respect to the risk for OSCC.(53, 54, 66-68) Although alcohol itself is not a direct carcinogen, one of its metabolites, acetaldehyde, may act as an indirect carcinogen by forming DNA adducts, which interfere with DNA replication and repair.(68) Heavy alcohol consumption has been associated with somatic copy-number alterations of oncogenes and tumour suppressor genes that have been reported to occur frequently in HNSCC.(61) The evidence about the role of alcohol in the etiology of OPML is mixed. A prospective study, which followed 41,458 male US health professionals, examined alcohol consumption and risk of OPML, and found that alcohol is an independent risk factor for the development of an OPML.(69) In contrast, the 1988-1994 US, National Health and Nutrition Examination Survey (n=15,811), found no independent role in the risk of OPML development.(70)   Some studies have found that tobacco smoking and alcohol consumption act synergistically to contribute to OSCC risk.(27, 44, 54, 71) These studies found that the risk of cancer development among heavy smokers and drinkers was considerably higher than the additive effect of the individual risks, suggesting that the joint effect is multiplicative. This interaction has also been shown to be associated with the development of OPML.(69) However, a recent large North American, prospective cohort study (n=101,182), reported that their findings did not suggest an interaction between cigarette smoking and alcohol drinking on a multiplicative scale.(53)  10  Tobacco cessation efforts have resulted in a drop in oral cancer rates associated with this habit,(72) leading to a growing interest in the increased proportion of cases occurring among non-smokers (NS).(73) A better understanding into the differences in the natural history of the disease in NS as compared to that of smokers is required.   1.3.2 Betel Nut The use of betel-quid, a mixture of areca nut, betel leaf, and slaked lime, with or without smokeless tobacco, is a significant risk factor in South and Southeast Asian populations.(45, 46) Like tobacco, a significant dose-response exists.(45) IARC has stated that there is carcinogenic risk to humans.(74) The components contain carcinogens and genotoxic agents which have a role in carcinogenesis. Betel nut contains alkaloids, including arecoline, and nitrosamines; lime provides reactive oxygen radicals; smokeless tobacco contains tobacco specific nitrosamines and nicotine-derived nitrosamine ketones.(75)   1.3.3 Infectious Agents Recently, HPV has been recognized as an etiological factor for the development of OPC, particularly those arising in the tonsils and base of the tongue.(48, 49, 76) IARC has stated that there is evidence to conclude that HPV is ‘sufficient, but not necessary’ as a cause of OPC. (50) HPV strains are categorized as being either high-risk (oncogenic) and associated with malignancy, or low-risk (non-oncogenic) and associated with benign diseases. HPV-16 is the most prevalent high-risk genotype and accounts for 71% to 95% of all HPV-positive OPC.(77-82) HPV is a DNA virus that infects the basal layer through microwounds or abrasions.(50) 11 High-risk HPV types cause cancer by expressing two viral oncoproteins, E6 and E7.  These oncoproteins degrade and destabilize the major tumour suppressor proteins, p53 and pRb. The net result is loss of cell control and unrestricted cell proliferation.(83)   There is no consensus in the literature whether Candida infection is a contributory factor in the development of oral cancer. Although a higher malignant transformation rate has been reported in Candida-infected OPML, a carcinogenic mechanism is not clear.(15) Candida can produce carcinogenic compounds such as nitrosamines and acetaldehyde. These compounds then can bind with DNA, forming adducts, and causing irregularities with DNA replication.(15, 84) However, there is no causal evidence for this association, and there continues to be considerable debate whether Candida infection is a cause of OPML, or if it is a coincidental infection within a preexisting lesion.(14, 15)   1.3.4 Immunosuppression A few studies have examined the incidence of cancer in organ transplant recipients. Patients on drugs for immune suppression to prevent rejection of transplanted organs are at a higher risk for virus-related cancers, including HPV-positive OPC.(85) The extent of immunosuppression varies according to factors such as type of organ transplant, time since transplant, and drug regimen.   Individuals with human immunodeficiency virus (HIV) have a two- to six-fold increase in risk for OCC and OPC relative to the general population.(86-88) HIV-positive individuals have an increased risk of HPV-positive cancers of the oral cavity and oropharynx,(48, 89-91) as well as other cancers, including Kaposi’s sarcoma and non-Hodgkin’s lymphoma.(92)  12  1.3.5 Age and Gender As with many cancers, the likelihood of developing OCC and OPC increases with age. The average age at diagnosis is 62. Half of all cases are in persons older than 65, and 90% are older than age 45.(30) The association of age with cancer risk is complex. Increasing age allows for more time for individuals to have been exposed to potential carcinogens(93) The underlying biological mechanism that drives the malignant transformation of OPML and carcinogenesis is thought to be the accumulation of genetic damage in key regulatory genes over time.(44, 94) The subsequent decline in immune and DNA repair systems with age may also contribute to overall risk.(94)   Oral and oropharyngeal cancer are more common among men. Men are twice as likely to develop oral cancer as women and four times more likely to develop cancer of the oropharynx.(1, 30) Males also have a higher prevalence ratio of OPML(15, 21), as compared to females.(41) This difference may be related to the increased use of alcohol and tobacco.  However, the gender difference is decreasing as trends in tobacco and alcohol consumption equalize between the sexes.(95)   1.3.6 Other Risk Factors Chronic exposure to sunlight, or actinic radiation, is considered a significant risk factor in the development of cancer of the lip. The lower lip, which receives more direct exposure, is more frequently affected than the upper lip. Lip cancer is more common in fair-skinned individuals 13 than in individuals with darker pigmentation, and it appears that melanin is protective against its development.(96)   Oral cancer risk is associated with low SES. Even when adjusted for potential confounders, this association remains globally.(97, 98) A diet low in fruits and vegetables is also associated with an increased risk of developing oral cancer.(99, 100) A recent pooled analysis of ten case-control studies (n=18,207 subjects) performed by the International Head and Neck Cancer Epidemiology Consortium, concluded that a low carotenoid intake, raises the risk of head and neck cancer substantially. This risk is extenuated with either tobacco or alcohol consumption. (101)  A family history of oral or oropharyngeal cancer is associated with an increased risk of OCC and OPC.(27, 102)   A certain number of cases have no obvious etiologies and are classified as idiopathic.(5, 51, 103)   1.4 Histopathology and Histological Progression Model of Malignant Transformation The presence of epithelial dysplasia is generally recognized as a potential predictor of malignant development in OPML. This is based on findings from longitudinal studies that oral lesions with dysplasia more often develop into SCC than those without dysplasia.(9, 12, 15, 16, 36, 37, 104-111) In addition, dysplasia is frequently seen in epithelium adjacent to oral SCC.(15)   The term dysplasia is used to describe histopathological changes associated with an increased risk of malignant transformation.(15) Under microscopic examination, dysplasia appears as both 14 cellular and architectural changes within the epithelial strata.(7, 10, 15, 93, 112, 113) The criteria used for diagnosing and grading dysplasia are outlined in Table 1.1.  Table 1.1 Architectural and cytological changes associated with dysplasia Architectural (Tissue) Changes Cytological (Cellular) Changes • Irregular epithelial stratification • Loss of polarity  • Presence of more than one layer of cells having a basaloid appearance • Drop-shaped rete ridges • Premature keratinization in single cells (dyskeratosis) • Keratin pearls within rete pegs  • Abnormal variation in nuclear size and shape (anisonucleosis and pleomorphism) • Abnormal variation in cell size and shape (anisocytosis and pleomorphism) • Increased nuclear/cytoplasmic ratio • Enlarged nuclei and cells • Nuclear hyperchromatism  • Increased number of mitotic figures • Abnormal mitotic figures (abnormal in shape or location)  Adapted from Pindborg et al., 1997 and Warnakulasuriya et al., 2008.(93, 112)   The WHO 2005 histopathological classification system uses a five-tiered system of grading of oral epithelial lesions (Table 1.2).(113) The grade of dysplasia is based on the extent of architectural changes and the degree of cellular changes across the epithelium thickness. Mild dysplasia is categorized as architectural changes limited to the lower third of the epithelium (basal and parabasal layers) with minimal cytological atypia. Moderate dysplasia is defined as architectural changes restricted to the lower two-thirds of the epithelium with moderate cytological atypia; if the cytological atypia is minimal, the lesion is downgraded to mild dysplasia. Abnormal rete pegs, cellular and nuclear pleomorphism, hyperchromatism, and 15 increased and abnormal mitoses may be seen. Severe dysplasia is characterized by architectural changes to more than two thirds of the epithelium with associated cytological atypia or by the architectural changes to the middle third of the epithelium with marked cytological atypia. The alterations seen in mild and moderate dysplasia may be present, but with more severity. Apoptotic bodies may also be noticeable. There is often a complete loss of stratification. Bulbous rete pegs, deep abnormal keratinization, and keratin pearls may be seen.(7, 10, 14, 112, 113)   Carcinoma in situ (CIS) is defined as "a lesion in which the full thickness, or almost the full thickness, of squamous epithelium shows the cellular features of carcinoma without stromal invasion".(112) SCC is diagnosed when the changes expand beyond the basement membrane and extend into the lamina propria.(7, 10, 14, 112, 113)   Table 1.2 Histopathological stages in oral epithelial lesions  Extent of Architectural and Cytological Changes Architectural Cytological Hyperplasia Thickened epithelium (hyperkeratosis) None Mild Dysplasia  Limited to lower third of epithelium  Mild atypia Moderate Dysplasia  Limited to lower two-thirds of epithelium  Mild atypia Severe Dysplasia More than two-thirds of epithelium  Moderate atypia Carcinoma in situ Full thickness, basement membrane intact, no stromal invasion Pronounced atypia  It should be noted that the histopathological stages in oral epithelial lesions are a continuum and 16 cannot be precisely divided into mild, moderate and severe categories. Diagnosis and grading is based on a combination of architectural and cytological changes. There is pronounced variability in the interpretation of the presence, degree, and significance of the specific individual criteria.  It has been shown that diagnosis is subjective and lacks intra- and inter-observer reproducibility.(93, 114-119)   It has been proposed that better agreement may be reached by modification of the WHO five-tier classification system into a binary system.(93, 120-122) This system classifies lesions into either low-risk or high-risk categories, based on a composite score from the individual criteria used for diagnosing and grading dysplasia. Lesions considered as having no dysplasia or mild dysplastic changes are categorized as low-grade. Moderate or severe changes are classified as high-grade (Table 1.3).(93) Studies have shown improved kappa values when grading with this system.(120, 121) It has also been suggested that a binary system more clearly demarcates those lesions that require treatment from those that fit into a “wait and watch” category.(122) An expert working group has stated that reducing the number of categories to two may increase the likelihood of agreement and that the utility of this classification system should be tested in future studies and would require validation before being adopted.(93, 122)    17 Table 1.3 Classification systems for oral epithelial lesions Adapted from Barnes et al., 2005.(113)  It can be difficult to differentiate between the earliest SCC that has invaded the underlying connective tissue and a severe dysplasia or CIS. Interpretation can be difficult; extensive inflammatory infiltrates can make the identification of breaks in the basement membrane challenging.(10) Another problem is that the ability to provide an accurate diagnosis may be compromised if the biopsy tissue is of poor quality, or is not representative of the lesion.(7) Biopsies should always be taken from the most representative area of a lesion, and if a lesion is extensive or non-homogeneous, more than one sample should be obtained.(123)   1.5 Oral Potentially Malignant Lesions   A clinically visible premalignant, or potentially malignant, lesion often precedes OSCC.(5-7, 14, 22, 34) As mentioned in section 1.2.2, OPML are ‘‘morphologically altered tissue in which cancer is more likely to occur than in its apparently normal counterpart”(36); they are characterized by histological and /or genetically altered tissue changes and have increased risk to develop cancer than a normal tissue.(6, 36, 37) Not all OPML will progress into cancer, and a Oral epithelial dysplasia Squamous intraepithelial neoplasia (SIN) Binary grading system Hyperplasia  Low Grade Mild dysplasia  SIN1 Moderate dysplasia SIN2 High Grade Severe dysplasia  SIN3 Carcinoma-in-situ SIN3 18 certain number of lesions will regress.(15, 34) Furthermore, it should not be presumed that an OPML always precedes oral cancer.  Oral carcinoma has been shown to develop shortly after a clinical examination with no lesion present or after a lesion has been histologically proven to be a benign non-dysplastic hyperkeratosis.(15)   The clinically apparent OPML that can precede the development of OSCC include leukoplakia, erythroplakia, OLP, tobacco pouch keratosis, and oral submucous fibrosis. The most common are leukoplakia, erythroplakia, or a mixture of both, known as erythroleukoplakia.(5-7, 14)   1.5.1 Leukoplakia Originally defined by the World Health Organization (WHO) in 1978 as “a white patch or plaque that cannot be characterized clinically or pathologically as any other disease”,(36) leukoplakia is presently defined as ‘‘a white plaque of questionable risk having excluded (other) known diseases or disorders that carry no increased risk for cancer”.(6, 14) If a lesion can be diagnosed as some other condition, such as OLP, candidiasis, et cetera, then the lesion should not be categorized as a leukoplakia.  The term leukoplakia is not diagnostic; it is a clinical, descriptive term. It is not a specific disease entity, and as such, clinical and histological appearances vary.(6, 14)   The incidence of dysplasia or neoplasia in oral leukoplakia ranges from 15.6% to 39.2%.(11, 16, 22, 51, 124-126) The clinicopathological features can vary and may be associated with the likelihood of dysplasia or malignancy.  19 1.5.2 Erythroplakia and Erythroleukoplakia An erythroplakia is a “fiery red patch or plaque that cannot be characterized clinically or pathologically as any other disease”.(36) Erythroplakia often present with a flat, smooth, velvety or granular texture.(6) Like leukoplakia, the term erythroplakia should only be used as a clinical term.  It is not diagnostic and has no histopathological meaning. Erythroplakia are less common than leukoplakia.(6) However, they are more often found to be dysplastic or neoplastic and are considered to have the highest risk of malignant progression amongst OPML.(6, 62)   Lesions that contain both red and white areas are termed speckled leukoplakia or erythroleukoplakia and tend to have a more irregular surface texture as compared to more homogeneous lesions.  1.5.3 Oral Lichen Planus and Lichenoid Mucositis  Lichenoid mucositis (LM) refers to a group of mucosal lesions (e.g. OLP, lupus erythematous and LM from contact with dental materials or intake of drugs) that are characterized by a band-like lympho-histiocytic inflammation in the immediate subepithelial region.(127) It is hypothesized that such inflammation results from both antigen-specific cell-mediated immunity in response to antigenicity changes in the oral epithelial lining cells as well as non-specific mechanisms such as mast cell degranulation and matrix metalloproteinase (MMP) activation in oral lichen planus lesions (128, 129). If an allergen can be identified then a diagnosis of LM can be made. A diagnosis of OLP can only be made after ruling out LM, and after fulfilling the criteria of presence of bilateral symmetrical lesions. Consequently, LM from allergic contact or drugs could be cured by withdrawal of the allergen, but not OPL. The diagnosis of OPL therefore 20 requires both histological assessment and clinical information (bilateral symmetrical lesions).   Clinical presentations can range from white lacy, reticular striae to plaque-like lesions, or erosive erythema and ulcerations.(128) OLP is categorized by the WHO as a potentially malignant condition (113); however, there is controversy in the literature over this statement.(130) Some argue that only OLP or LM with dysplasia – referred to as lichenoid dysplasia (LD) – have malignant potential.(127, 131, 132) Others maintain that there remains some evidence that patients with OLP may be at a greater risk of malignant progression.(133) It has been noted that erosive OLP lesions may have a greater potential for malignant transformation than reticular OLP.(134, 135) Although there is research that compares the rate of progression between OLP and LD,(40, 127, 133, 136) there has been no research that compares malignant transformation of LD compared to OED.   1.5.4 Other Oral Potentially Malignant Lesions Other OPML include nicotinic stomatitis and tobacco pouch keratosis and OSF. OSF is a chronic disorder of the oral mucosa characterized by fibrosis of the submucosa and whitening of the oral epithelium. The fibrotic bands results in stiffness and limited opening. The condition is thought to have a complex multifactorial etiology and is directly associated with areca nut or betel quid chewing, a habit akin to smokeless tobacco.(137, 138) It is encountered less frequently in North America and is more common in certain parts of the world where betel quid chewing is practiced among the populations of South and Southeast Asia.(22, 38, 45, 46)   21 1.6 Malignant Transformation of Oral Potentially Malignant Lesions Not all OPML will progress into cancer. Some lesions will remain static and a certain number of lesions may even regress.(15, 34, 51, 104, 105, 112)  In spite of progress in the field of oral cancer biology there is no single marker that can reliably predict the risk of malignant transformation in an OPML. The clinicopathological features of a lesion are insufficient to predict the likelihood of malignant transformation. A biopsy and histopathological examination are necessary to establish a definitive diagnosis and may be valuable in predicting malignant development.   1.6.1 Transformation Rates The evidence that OPML undergo malignant transformation is largely derived from hospital-based, follow-up studies.(15) Studies have shown that between 0.13% and 36.5% of OPML will develop into oral cancer.(7, 9, 15, 104, 107, 109, 139). The rate of transformation varies widely with case selection, geographical location and habits, indicating that study heterogeneity plays a significant role in the wide range reported.(15, 104, 105) The wide variation in the malignant transformation rate is likely due to the definition of OPML. As mentioned before, the definition of oral leukoplakia should include histological assessment but in many studies, the diagnosis of OPML was based on a clinical diagnosis of leukoplakia or white keratotic lesion only. It is likely that many of these white lesions are not premalignant but rather reactive hyperkeratotic or acanthotic lesions. Lower rates may contain a considerable number of OPML without dysplastic features. After considering heterogeneity amongst studies, a systematic review and meta-analysis carried out by Mehanna et al.(12) estimated the mean malignant transformation rate to be 12.1% (95% CI, 8.1 - 17.9).  22  One of the inherent difficulties in measuring the rate of malignant transformation of OPML is that the outcome can be influenced by treatment interventions. Even the biopsy, required to make a diagnosis can be curative or promotive to alter the natural history of the lesion. Although biopsy and treatment influence end-point, withholding treatment is not an option for ethical reasons, which limits prospective follow up studies.(15, 52)   1.6.1.1 Histological Grade and Risk of Malignant Transformation OPML with evidence of dysplasia are at the greatest risk of progressing to oral cancer.(139) The severity of dysplasia is considered the “gold standard” predictor of progression; lesions with the highest grade of dysplasia are thought to have the greatest probability of malignant transformation.(12, 93, 108, 111) The presence of high-grade oral dysplasia (severe dysplasia/CIS) is a significant predictor for progression to malignancy. The malignant transformation rate for severe dysplasia ranges between 7% and 50%, with an estimated overall malignant transformation rate of about 16%.(10-12)   However, the ability to predict outcome in low-grade (mild/moderate) dysplasia is challenging as the majority will not progress and a certain number will even regress.(14, 15, 52, 104, 109, 140, 141); it is very difficult to predict which LGD will progress to SCC from histology alone.(142, 143) The malignant transformation rate for moderate dysplasia ranges between 3 and 15%, while the rate for mild dysplasia is estimated to be at approximately 5%.(10-12)  23 1.6.1.2 Clinicopathological Features and Association with Malignant Progression  The presentation or clinical characteristics of OPML can be somewhat predictive of a higher risk of malignant transformation. It has been shown that certain clinical sub-types of OPML possess a greater risk for malignant transformation than others.(15, 140) Of particular importance are lesion site, appearance, size and number of lesions.  1.6.1.2.1 Anatomical site The anatomic site of an OPML is associated with the risk of malignant transformation.  OPML can be located on any mucosal surface in the oral cavity. In western countries, smoking is the main etiological factor for oral cancer development and most oral cancers are located at the ventrolateral tongue and floor of mouth. It is therefore not surprising that lesions located on the ventrolateral tongue and the floor of mouth are at a higher risk of cancer progression; consequently these sites are called high-risk sites.(22, 33, 40, 63, 71, 125, 139, 144, 145) On the other hand, lesions located on the buccal mucosa, gingiva and hard palate are at lower risk of cancer progression; hence these sites are called low-risk sites.(37, 139, 146) Lesions on the soft palate are also regarded by some to have an increased cancer risk. However, our previous studies have shown similar cancer risk of lesions on the soft palate to those located on the buccal mucosa, gingiva and hard palate;(13, 147) consequently in this thesis ventrolateral tongue and floor of mouth were designated as high-risk sites; whereas the rest of the oral cavity as low-risk sites.  The reason for the increased cancer risk of lesions from the floor of mouth and ventrolateral tongue remain unclear. Proposed theories include: (1) thinner epithelium and lack of 24 keratinization allow easier penetration for carcinogens to reach the basal epithelial cells; (2) higher proliferation risk, which allows for a higher chance of mutation during DNA replication; (3) the location of these lesions at the lower part of the oral cavity allows for longer exposure to carcinogens dissolved in the saliva (reservoir theory).(148)   Of course, any sites with prolonged exposure to carcinogens will have an increased cancer risk, such as sites where chewing tobacco or betel quid are placed (such as the labial and buccal vestibule or gingiva).  In countries where these habits are prevalent, such as Southeast Asia, the high-risk sites are the gingiva, and the labial and buccal vestibules.(22, 149, 150) In areas where reverse smoking is practiced, most OPML are found on the palate.(22)   1.6.1.2.2 Appearance  As discussed previously, lesions can be classified by colour as either white (leukoplakia) or red (erythroplakia).(36, 151) Erythroplakia generally have a higher risk of malignant transformation.(15, 52, 56, 110) Leukoplakia can be further classified as homogeneous or nonhomogeneous.(151) The terms homogeneous and nonhomogeneous refer to the colour and the texture of the lesion.  Homogeneous lesions are uniform in both colour and texture. They are primarily white and have a smooth, thin texture. Nonhomogeneous lesions have variation in colour or texture, or both. Texture can be rough, leathery, granular, exophytic, papillary, nodular, indurated, verrucous, ulcerated, or present as a mixture of these.(5-7, 22, 33, 145, 151, 152) In general, homogeneous lesions are considered to possess a lower risk of malignant transformation than nonhomogeneous lesions.(5, 15, 22, 33) Speckled, or mixed leukoplakias (erythroleukoplakia) possess a significantly higher risk of dysplasia or malignant transformation 25 as compared to homogeneous lesions.(10, 15, 33, 51, 104, 105, 109, 124, 125, 140, 145, 153, 154) Similarly, the thicker the lesion is, the higher the risk of finding dysplastic changes or of malignant transformation. A smooth, thin lesion is less likely to progress than a thick, rough, verrucous, or nodular lesion.(8, 151) The clinical sub-type of verrucous hyperplasia, or proliferative verrucous leukoplakia, is a high-risk lesion with a high rate of transformation and should be treated aggressively.(37, 52, 155) A 10-year retrospective study by Casparis et al.,(40) found that an ulcerative appearance is also significantly associated with higher degrees of dysplasia or malignancy (P = 0.06).   1.6.1.2.3 Lesion Size and Number of Lesions Size of an OPML is also associated with risk of malignancy. Size of the lesion is measured from the greatest length by the greatest width.  In the case of multiple lesions, size is calculated by a field measurement, totaling the greatest length of all lesions by the greatest width of all lesions in a single site.(33) Lesions that are greater than 2 cm have shown to have a higher likelihood of malignant transformation than those that are less than 2 cm.(14, 125, 145)   The risk of malignant progression is greater with the presence of multiple lesions.(5, 145, 156) When large areas of tissue surface are exposed to carcinogenic exposures over a lengthy period, resultant genetic defects can give rise to multifocal OPML at different stages of carcinogenesis. This field effect is termed field cancerization.(156-160) Accordingly, the consequential multiple lesions that arise are significantly associated with malignant progression.(5, 145, 156, 158, 159)    26 1.6.1.2.4 Duration  Finally, duration of a lesion is associated with malignant transformation.(5, 16, 33, 109, 125) The longer an OPML is present, the more likely it has had time to accumulate genetic defects that can drive malignant transformation.  Persistent lesions should be considered suspicious.(161, 162)  The clinicopathological features of a lesion can increase the suspicion that a lesion may be a potentially malignant or a malignant lesion.  However, the clinical characteristics alone are insufficient. A biopsy and histopathological examination are essential to establish a definitive diagnosis.    1.6.2 Field cancerization The concept of  “field cancerization” was first described by Slaughter et al.,(158) to explain how synchronous, multiple OPML or malignant lesions develop. The authors proposed that when large areas of tissue surface are exposed to carcinogens over a lengthy period, it results in an increased susceptibility of an entire area.  Genetic mutations occur over a widespread multifocal field and can give rise to multifocal OPML at different stages of carcinogenesis. This field effect is termed “field cancerization”.(156-158)   Later, the concept was used to develop a multistep genetic progression model for the development of OSCC. Braakhuis et al.(159) suggested that a single stem cell acquires a critical genetic mutation. The alteration provides a growth advantage over adjacent cells and it proliferates.  This patch, or “clonal unit”, consists of the stem cell and its daughter cells, which all possess the genetic mutation. This patch then expands into a field, which replaces the normal 27 epithelium. The process repeats, with further genetic mutations and expansion within an already altered field.  Eventually, clonal selection causes the development of a carcinoma within an altered field.(157, 159, 163, 164)   An important clinical implication of field cancerization is that altered fields, which appear clinically normal, can remain after surgical excision of the tumour and could lead to a new cancer or local recurrence.(157, 163, 165)   1.6.3 Invasion and Metastasis Spreading of cancer cells through invasion and metastasis is a complex, multistep process. Only a small number of the cells in a malignant tumour will be able to carry out all the steps necessary for metastasis.(166) The sequential steps begin with proliferation and primary tumour formation.  Intercellular adhesion is reduced in the tumour cells because of loss of E-cadherin, which causes them to express proteins that promote cell elongation and interfere with cell polarity.(167) Changes in cell adhesion, detachment, motility, and local proteolysis, result in cell migration and local invasion.(23, 166, 167) At the same time, the cancer cells produce angiogenesis-stimulating molecules, leading to the formation of new blood vessels. The neovasculature is loose and permeable, providing the migrating tumour cells with easy access. Intravasation, or penetration of a cell into a blood or lymphatic vessel, allows the cancer cell to be transported to distant sites. Interaction between the malignant cell, platelets, lymphocytes and other blood components results in the arrest of the malignant cancer cell embolus in microvessels of various organs.(166) Extravasation occurs as the malignant cell attaches onto and retracts the endothelial cells and penetrates the basement membrane.  Local proteolysis and migration into the tissues produce a 28 micrometastasis that can then colonize and continue to proliferate.(23, 166, 167)   1.6.4 Tumour Staging Oral cancers are staged using the TNM classification system, which indicates the extent of the tumour spreading. This system uses the size of primary tumour (T), extent of lymph node involvement (N), and presence of distant metastases (M). TNM staging is used to determine treatment and predict prognosis (Table 1.4).(22, 168) Generally, the larger the size of the primary tumour (T), the higher the risk of regional neck disease.(23) Oral cancer frequently metastasizes to the cervical lymph nodes.  Malignant lymph nodes (N) involvement is determined by size and whether the nodes are hard and fixed.(23) Nodes become fixed and immoveable when the tumour penetrates the capsule and spreads into the surrounding connective tissue (extracapsular spread).(22, 169) Up to 30% of all oral cancers have cervical node metastases at the time of diagnosis; this statistic increases to 66% for cancer of the tongue.(22, 169) Distant metastases (M) can occur anywhere in the body but are most common in the lungs.(22) Staging of oral cancer is important for establishing proper treatment and determining prognosis.  29 Table 1.4 TNM staging of oral cancer Primary Tumour (T)  TX Primary tumour cannot be assessed T0 No evidence of primary tumour Tis Carcinoma in situ T1 Tumour 2 cm or less in greatest dimension T2 Tumour more than 2 cm but not more than 4 cm in greatest dimension  T3 Tumour more than 4 cm in greatest dimension T4 Tumour invades adjacent structures   T4a Tumour invades adjacent structures such as through cortical bone into deep extrinsic muscle of the tongue, maxillary sinus, or skin of face  T4b Tumour invades masticator space, pterygoid plates, or skull base and/or encases internal carotid artery Nodal Involvement (N) NX Regional lymph nodes cannot be assessed N0 No regional lymph node metastasis N1 Metastasis in a single, ipsilateral lymph node, 3 cm or less in greatest dimension N2 Metastasis in a single ipsilateral lymph node, more than 3 cm but not more than 6 cm in greatest dimension; or in a bilateral or contralateral lymph nodes, none more than 6 cm in greatest dimension  N2a Metastasis in a single ipsilateral lymph node, more than 3 cm but not more than 6 cm in greatest dimension  N2b Metastasis in multiple ipsilateral lymph nodes, none more than 6 cm in greatest dimension  N2c Metastasis in bilateral or contralateral lymph nodes, none more than 6 cm in greatest dimension N3 Metastasis in a lymph node more than 6 cm in greatest dimension Distant Metastasis (M) MX Distant metastasis cannot be assessed  M0 No distant metastasis M1 Distant metastasis  Modified from Neville BW, Day TA. Oral Cancer and Precancerous Lesions. CA: A Cancer Journal for Clinicians 2002;52(4):195. 30 Tumour staging is based on the TNM system (Table 1.5). Staging of oral cancer is important for establishing proper treatment and determining prognosis. Survival rate is related to the stage of disease. According to 2005 – 2011 SEER data from the National Cancer Institute, the five-year relative survival rate for patients with localized disease is 83.0%. This rate drops to 61.5% when regional spread to regional lymph nodes is present and to 37.7% when distant metastasis has occurred.(30)   Table 1.5 Stage grouping Stage TNM Grouping Stage 0 Tis N0 M0 Stage I T1 N0 M0 Stage II T2 N0 M0  Stage III T3 N0 M0 or any N1 Stage IV   IVa Any T4 or any N2  IVb Any T4b or any N3  IVc Any M1 lesion Modified from Neville BW, Day TA. Oral Cancer and Precancerous Lesions. CA: A Cancer Journal for Clinicians 2002; 52(4):195.  1.7 Management of Oral Epithelial Dysplasia The fundamental dilemma in the management of OED is whether to observe patients or to intervene and offer treatment. It is generally accepted that surgical management should be offered for “high-risk” lesions - currently defined by histological diagnosis of severe dysplasia or higher.(170-172) However, there is no agreement whether LGD should be surgically managed, 31 or how it should be surgically managed. On one hand, it is important to treat OMPL to prevent malignant transformation. On the other hand, most LGD will not progress. Treatment can bring significant patient morbidity and risks the over-treatment of lesions that are unlikely to progress.  1.7.1 Risk Factor Modification  Management protocol should begin with education of patients about the importance of eliminating risk factors, including tobacco, both smoked and smokeless.(148) This not only reduces the risk of malignant transformation but also decreases the risk of future precancerous lesion development.(173) Patients should also be counselled about the risk associated with excessive alcohol consumption and the synergistic effects between tobacco and alcohol on the risk of oral cancer.(148) Any dietary deficiencies should be addressed and patients should be counselled on the potential benefits of an adequate dietary intake of fresh fruit and vegetables.(148)   1.7.2 Surgical Treatment  There are no universally agreed upon guidelines for the treatment of OPML. This is mainly due to the fact that no high-level randomized control trials of surgical intervention are available and evidence is lacking that intervention halts carcinogenesis. (174) The majority of evidence is based on observational and retrospective data.(175) For high-grade lesions (severe dysplasia or CIS), surgery is the first choice for management by most specialists.(170, 176) A study from our lab has shown that surgical removal of severe dysplasia significantly lowers the risk of cancer progression.(177) When surgical treatment is employed, the intention is to remove the entire lesion(s). This can be accomplished with traditional scalpel surgery, with laser, or 32 cryosurgery.(148)   For low-grade lesions, the management tends to be more conservative. It might be assumed that removing premalignant lesions would eliminate the risk of cancer in the affected area. However, this assumption cannot be made with OPML. Retrospective studies have shown that surgical excision of low-grade OPML does not show benefit with respect to progression to cancer.(12, 109, 140, 178-184) Poor surgical outcomes may be due to field cancerization which considers that there are patches of genetically altered multiclonal cells beyond the removed lesion.(185) Another reason why low-grade lesions are managed conservatively is to prevent the harm of overtreatment.  Surgical treatment can bear significant patient morbidity. Complications following surgery may include pain, swelling, bleeding, paresthesia, speech difficulties, swallowing difficulties, and recurrent disease. Overtreatment can also be a strain on limited health-care resources.(186)   Another treatment approach is laser ablation. With this technique, a defocused laser beam is used to destroy the surface mucosa.(148) Laser ablation is not routinely recommended because the tissue is not preserved and cannot be subjected to histopathological examination. Also, abnormal basal epithelium may be left in situ. Some theorize that provoking proliferative activity and epithelial regeneration in dysplastic tissue, with interventions such as ablation, may even worsen the disease progress and encourage malignant transformation.(148) In spite of this, ablation may have a role in treating extensive, multifocal lesions or lesions on alveolar or gingival tissue where surgical excision could lead to areas of denuded non-vital alveolar bone.(148)   33 1.7.3 Medical Treatment  1.7.3.1 Chemotherapeutics A variety of interventions have been explored over the years as treatments for OPML. These have included the topical or systemic use of beta carotene, vitamins C and E, retinoic acid, bleomycin, and other anti-oxidants.(184) However, there has been no evidence of long-term effectiveness with these therapeutics.(184) Furthermore, recurrence was common following treatment cessation and side effects were frequent.(184)   1.7.3.2 Chemoprevention The rationale for chemoprevention is based on the concept that the early stages of carcinogenesis may be reversible and that the prevention or delay of progression to invasive cancer is beneficial. OSCC is a fitting model for chemoprevention studies because of its field cancerization effect. Attention has been focused on the development of chemopreventive agents targeted to specific molecular pathways in the progression from oral premalignancy to OSCC. Examples of molecularly targeted agents include cyclooxygenase-2 (COX-2) inhibitors and epidermal growth factor receptor (EGFR) inhibitors.(187-190) However, results from these studies have not shown a significantly beneficial response. 
  Recently, the diabetic medication metformin has been investigated for a repurposed use in cancer chemoprevention.(191-193) Preliminary results from a phase IIa chemoprevention trial examining the use of metformin in oral cancer prevention have shown a toxicity profile consistent with the known side effect profile of metformin, and of interest, a histologic response rate of 59%, including 13% complete responses.(194) Circulating and tissue biomarker analyses 34 are being analyzed.(194) Although chemoprevention appears to be a promising approach, larger, multi-centered, prospective clinical trials of a longer duration are needed to evaluate clinical, histological, and molecular efficacy.  1.7.3.3 Immunoprevention  Recent findings have drawn attention to the importance of the tumour microenvironment and its associated immune cells in cancer development and regression.(195-197) This raises the possibility that like chemoprevention, immune therapies could be used for the prevention of OSCC. Experimental pre-clinical research is currently being conducted which may possibly support the future development of approaches for preventive immune oncology.(198)   1.7.4 Surveillance  One point that is agreed upon by most is that regular patient follow up and repeated clinical examination is required for all OPML, irrespective of the mode of proposed treatment.(199) It is increasingly accepted that surveillance is probably the most pragmatic and efficacious approach to management of OPML.(176, 186) Careful and thorough examination of all oral mucosal surfaces at each follow up visit is important. A systematic approach to the assessment of OPML and detailed documentation is required.(145) This can be supplemented with the use of adjunctive visual tools, which are discussed in detail in the next section.  There are no clearly established guidelines for the follow up of OPML. It is best that follow up be tailored to individual patient risk. Ideally, those deemed to be at high-risk of malignant transformation should be followed very closely. However, such risk stratification is more 35 difficult than it might appear. There is a lack of biomarkers for the predicting risk of progression. This makes it a challenge to differentiate between those OPML at high-risk from those at low-risk of progression. A major barrier to oral cancer prevention has been the lack of validated risk predictors for OPML. There is a pressing need for the development of visual aids and biomarkers that will facilitate the detection of OPML with a high-risk of progression.  1.8 Adjunctive Clinical Aids   1.8.1.1 Toluidine Blue  Toluidine blue (TB) is an acidophilic stain that has been used primarily to identify SCC and dysplasia since the 1960s.(200, 201) It is believed to work in 2 ways: 1) by selectively binding to nucleic acids (found in high concentrations in rapidly growing cells); and 2) by penetrating between the cells as a result of defective cell barriers found in tissue undergoing carcinogenesis.(202, 203) It is a simple chairside test that involves the application of the stain using a cotton tip applicator with a subsequent application of acetic acid to remove any excess dye.  A positive test results in an intense royal blue colour. TB helps highlight the area of the lesion believed to have the highest degree of pathology; hence it aids in determining where within a lesion to biopsy.(204-206) TB has also been found to aid in the delineation of faint lesions, hence surgical margins and in finding satellite and occult lesions, and aid in the delineation of surgical margins.(204, 207-211)   TB, in experienced hands, has been reported to be sensitive in the detection of oral SCC.(212) Specificity varies across studies but improves when the test is repeated 10-14 days after the initial assessment. This allows sufficient time for reactions associated with ulceration to 36 resolve.(209) The use of TB in the detection of dysplasia has led to varied results.(213-215)  In the past TB has not been found to be helpful in detecting LGD (216, 217) and often complicated by a high-rate of false positives.(218) The question then asked was why some LGD stain positive while others do not. It has been found that TB positive LGD are more likely to have high-risk molecular patterns, (147, 219, 220) and in 2005, our research group found that TB positive LGD were four times more likely to progress to SCC than TB negative LGD.(147) Despite the significance of this study, the study results are based on one time point during the natural history of a lesion. From our experience within a prospective longitudinal study, we know that TB staining status can change over time (e.g. from positive to negative or vice versa). Although TB status at one point in time and its association with future malignant transformation has been examined, no research has been carried out to determine whether its status over time is associated with malignant transformation of LGD. A temporal analysis may well improve the prediction model and provide critical insight into the sequence of events that drive progression to invasive cancer.  Further research is needed to evaluate how a change in TB status during follow-up is associated with risk of progression.  1.8.1.2 Fluorescence Visualization  Tissue autofluorescence can be examined through direct fluorescence visualization (FV). With FV the oral mucosa is exposed to high-energy visible (blue) light, which excites fluorophores, such as nicotinamide adenine dinucleotide (NADH), flavin adenine dinucleotide (FAD), collagen and elastin, in normal tissue, causing them to fluoresce.(221) The fluorescent properties come from the ability of these fluorophores to absorb light energy of a particular wavelength and then to re-emit the light at a longer wavelength.(221) Tissue change that increases either the scattering 37 of the incoming light or decrease its reemission by the tissue can result in the loss of fluorescence. Such change includes underlying stromal disruption, breakdown of collagen matrix, or increased vascularity, or changes in the epithelium, such as an increase in tissue thickness, increased cellular hyperchromatism, and increased nuclear and cellular pleomorphism. All result in the appearance of loss of fluorescence in tissue under FV in comparison to surrounding normal tissue. These cellular and architectural changes are associated with premalignant and malignant changes in the epithelium and submucosa.(221) Visualization of tissue autofluorescence has been shown to be sensitive in detecting OPML and malignant lesions.(222-224) Unfortunately, some of these changes can also occur in reactive lesions in response to inflammation, infection, and trauma resulting in loss of FV as a result of increased vascularity or a thicker epithelium.(225). To reduce these confounding conditions, it is recommended that a positive FV assessment be confirmed two to three weeks after the initial assessment, allowing time for the tissue to heal.(162) Finally, research has also shown that when excision of an OPML is necessary, FV determined margins are superior to white light and TB margins.(226) This suggests that the approach may have additional value in defining clinically non-apparent regions undergoing cancerization that could put an area at increased risk for developing cancer.   Lesions that are being followed longitudinally can benefit from a temporal assessment of their positivity. A LGD that is always positive to adjunctive assessment, or is on and off positive, may have higher cancer risk than those LGD that are always negative; and a LGD that becomes positive after a period of negativity may be revealing a change in risk. Previous research has examined the association between FV and its association with histology.(200, 222, 227, 228) 38 However, no research has been carried out to determine whether its status over time is associated with malignant transformation of LGD. Further research is needed to evaluate how change in FV status during follow-up is associated with risk of progression.  1.9 Biomarkers of Risk Prediction   1.9.1.1 Overview of Molecular Biomarkers  Neither clinical appearance nor histopathology has been able to reliably predict the risk of malignant transformation in an OPML. Histological examination, considered to be the current gold standard, is reasonably effective in judging the malignant risk of high-grade lesions. However, it is a poor predictor for low-grade dysplasia, which represent the majority of oral dysplasia. This has led to a search for additional cellular or molecular markers for malignant progression in OPML.    The evaluation of DNA content by cytometry, ploidy status, and loss of heterozygosity (LOH) have shown capacity as potential biomarkers for malignant progression.(229-236) In 2009, a systematic review evaluating predictive biomarkers was conducted by Smith et al.(237) They scrutinized 288 publications analyzing 113 different predictive biomarkers in dysplasia. The most common biomarkers investigated were p53, proliferation markers Ki67 and proliferating cell nuclear antigen (PCNA), cell cycle proteins cyclin and argyrophilic nucleolar organizer region (Ag-NOR), and loss of heterozygosity (LOH). Only 13 studies met the criteria of longitudinal design with adequate follow-up and well-defined diagnostic criteria. Within those that met the criteria, only LOH, survivin, matrix metallopeptidase 9 (MMP-9), and DNA content were significantly associated with malignant progression. The authors conclude that research into 39 this field should concentrate on longitudinal design, with pooling of data from multiple centers to achieve larger cohorts.(237) In 2011, Nankivell and Mehanna (238) updated the review that was completed in 2009. They noted several markers which had statistically significant ability to predict malignant transformation in OED. However, they noted that the studies were small, retrospective cohort or case-controlled studies from single centers. Furthermore, many of the studies had not incorporated an analysis accounting for risk factors or confounders such as smoking or alcohol consumption. They concluded that despite some biomarkers having shown a potential for predictive ability in small cross-sectional studies, none have as yet been validated in independent prospective cohorts; hence there was little evidence to support their routine clinical use.(238)  Since these systematic reviews were published, the only risk predictors of malignant progression in OED that have been validated in independent prospective cohorts are LOH and p16 methylation.(13, 239) Another study has stated that there is work in progress to validate chromosomal instability and aneuploidy in a prospective study with clinical endpoints.(235).   In summary, the research to date that has examined biomarkers for their ability to predict malignant progression in OED has mainly used retrospective cohorts – there are very few prospective cohort studies. Few potential markers have been validated in independent studies. None have used a temporal repeated analysis. This is a major roadblock in our capacity to develop risk models of clinical utility. Finally, it is important to note that, although a few markers have been validated within longitudinal cohorts, such studies have only examined risk based on predictors measured at a single point in time, often baseline. It is well known that 40 lesions can change significantly over time, apparent clinically, histologically and, potentially genetically. There is a need for studies to “drill” deeper into such analyses, to examine predictors at repeated intervals to provide critical insight into the sequence of events that drive progression to invasive cancer and to determine whether such an analysis framework would further strengthen a risk model. The fifth chapter of this thesis explores the potential value of taking such a repeated approach to temporal analysis.  1.9.1.2 Mutation as a Driving Force for Carcinogenesis and Loss of Heterozygosity Cancer is a genetic disease that develops when a stem cell in a tissue sustains a mutation in a critical control gene, either through exposure to endogenous or exogenous factors. This mutation gives the stem cell a selective growth advantage, such that it can grow out to populate a greater portion of the stem cell niche. Repeated mutation and clonal expansion within this initiated cell’s progeny leads to the development of a multi-clonal cancerized field within a tissue, and eventually, to cancer development.(158)   The conversion of proto-oncogenes to oncogenes is one group of genetic alterations that occur during carcinogenesis.  Proto-oncogenes act as control genes that turn on key signalling pathways in normal cells, to trigger events such as cell proliferation, angiogenesis, DNA repair and others in response to a need. This process usually involves the activation of signal transduction pathways that result in gene transcription, new proteins and eventually alterations to normal cell structure and function. Mutation of proto-oncogenes results in a gain of function in the mutated cell with an uncontrolled activation of such key regulatory events. In contrast, tumour suppressor genes (TSG) are genes that act to negatively regulate key events by turning 41 them off when not needed or by identifying inappropriate behaviour and blocking such processes when they are dysregulated.(240-242) Mutations to TSG result in a loss of function of a gene, and consequently a dysregulation of a key process, for example, uncontrollable, inappropriate cell proliferation.(243) Oncogenes require the mutation of only one of the two chromosomal copies of a gene to be mutated for a gain of function to occur. In contrast, both copies of a TSG must be mutated for a function to be lost. The process of cancer development involves both the activation of oncogenes and the loss of TSGs, with this combination underlying the altered behaviors that characterize what has become known as the “hallmarks” of cancer.(244, 245)  This thesis focuses on a well-defined, robust assay, a PCR-based loss of heterozygosity test that has been developed and validated for measuring alteration to copy number in key regions of DNA from archival formalin-fixed tissue of premalignant lesions and cancers. The DNA from such sources is often of poor quality and in low quantity, and requires protocols tailored to assessing such samples. The assay identifies copy number change to key regions of DNA. Such change is commonly associated with the loss of function of TSGs, in the two-step process described above. The loss of genetic material from one chromosomal locus, in a chromosomal pair, is termed loss of heterozygosity (LOH).(17, 234, 246) Hence, LOH in TSGs is key to the loss of cell cycle regulatory function and the alteration to key biological processes that are associated with evolution of normal cells to cancer.(241, 242)  Many TSGs have been investigated for their role in OPMLs and oral cancer including those found at 9p21, an early marker of carcinogenesis, 3p21, 17p13 and 4q.(247-251)   42 The use of LOH to examine alterations in DNA copy-number at select chromosome sites in OPML was introduced in a hallmark publication in 1996 from Califano et al.(248) This paper was the first to present a set of molecular changes that associated with early premalignant stages, establishing the foundation for using early LOH events, such as those occurring at 9p21, 3p or 17p, as markers of progression.(248) Several studies have since confirmed that LOH occurring on 3p and 9p is associated with malignant transformation and progression.(163, 246, 251, 252) With mounting interest in validation of these progression markers, a suitable patient population for retrospective and prospective analyses was required. In 2000, our lab confirmed previously identified promising candidate LOH markers in a retrospective cohort.(17) Best predictors of outcome in this analysis was LOH at 3p14 and/or 9p21 – confirming the work of others. A high-risk profile (3p and/or 9p LOH) had a 22-fold increase in risk compared to a low-risk profile (3p and 9p retention).(17) Our lab then went on to validate these LOH markers in an independent prospective cohort. A refinement of this model was developed within this cohort with the addition of another two markers (loci on 4q/17p) that further improved the risk prediction. LOH at 9p, 4q and 17p had a 52-fold increased risk of oral cancer when compared to lesions which retained 9p.(13) The predictive value of this refined model was then reverse validated by using data from the previous retrospective cohort.(13) It should be noted that this assay is one of the very few tests that have been used in multiple laboratories for association with risk of premalignant disease and remains the only test that has prospective longitudinal analysis supporting its role in risk prediction. Although this model is highly predictive of oral cancer development in the high-risk molecular group, further refinement of the model is required, particularly for those in the molecular intermediate-risk category.  43 Chapter 2: Methodology   This chapter provides an overview of the materials and methods used in this thesis.   2.1 OCPL Study and Patient Population  The subjects for the analyses presented in this thesis were acquired through the Oral Cancer Prediction Longitudinal (OCPL) Study being conducted at the BC Cancer Agency in Vancouver (British Columbia, Canada). The OCPL study is an ongoing cohort study, funded by the National Institute of Dental and Craniofacial Research and the BC Cancer Foundation, and has prospectively enrolled and followed patients since 1997, with a goal of identifying biomarkers for the prediction and management of oral cancer. The OCPL study has two study arms: the first follows patients with a histological diagnosis of CIS or SCC, who have been treated with curative intent, and aims to predict reoccurrence and/or second primary tumours; the second follows patients with histologically confirmed OED (mild, moderate or severe) to predict malignant progression. Participants in the OCPL study were identified through a centralized population-based biopsy service, the BC Oral Biopsy Service (OBS), where community dentists and surgeons across British Columbia (estimated population 4.8 million, in 2017 (253)) send biopsies for histopathological diagnosis. Patients were referred for clinical follow up to Oral Dysplasia Clinics (Vancouver Cancer Centre, Fraser Valley Cancer Centre, Vancouver General Hospital, and UBC Specialty Clinics), where they were evaluated and, if eligible, invited to participate in the study. Patients are eligible for accrual to the OCPL study if they are 18 years or older, have a histologically confirmed diagnosis of OED, CIS or SCC, have no history of oral cancer and are able to attend regular follow-up appointments. The OCPL study is one of the 44 largest longitudinal cohorts ever conducted for the above purpose and is unique in that it draws from a community-based population as opposed to a high-risk hospital-based population.    Ethical approval was obtained from UBC and the BC Cancer Agency Research Ethics Board. Each patient was consented via a written informed consent form prior to enrolment in the study. Each participant was assigned a unique study identification number which was used for data collection, database storage, labelling of patient samples and for laboratory analysis.   2.1.1 Participant Selection  The analyses presented in this thesis used subgroups from the second arm of the OCPL study population that focused on prediction of risk for patients with OED. Participants used in these analyses were enrolled between January 1, 1997 and February 13, 2014 and were diagnosed with a histologically confirmed primary diagnosis of mild or moderate epithelial dysplasia. Detailed inclusion criteria and description of the sample population for each of the analyses are provided in each of the following chapters specific to each analysis.     2.2 Histological Evaluation  Biopsy specimens were formalin-fixed and paraffin-embedded by staff at the OBS. Sections were cut and stained with haematoxylin and eosin (H&E) and were submitted for histological examination and diagnosis to pathologists at the OBS. As part of the study protocol, a copy of the pathology report (Appendix A.1) and archival formalin-fixed paraffin-embedded (FFPE) tissue specimen blocks from index biopsies were requested and obtained for each participant by the OCPL study staff. Histologic diagnoses were reviewed and confirmed by the study 45 pathologist (L.Z.), using diagnostic criteria established by the WHO. (113) In the case of disagreement in diagnosis, the following protocol was followed: for discrepancies of one grade the study pathologist’s diagnosis was considered final; for discrepancies of two grades or more, diagnosis was achieved by consensus evaluation among three pathologists.   2.3 Molecular Evaluation    LOH analysis was performed on index biopsies collected at baseline.   2.3.1 Sample Cutting and Preparation  Each block was oriented on a microtome to align all axes in order to cut even sections. The tissues were cut on a microtome, using a new blade for each block. Tissue sections of 5 µm were cut for H&E staining, which serves to confirm histological diagnosis and to serve as a reference slide for microdissection. Ten to 15 sections of 10 µm thickness were cut for staining with methyl green (Sigma-Aldrich, St. Louis, MO, USA) for microdissection.  H&E staining provides a higher quality staining for reference for histology, and methyl green allows for a higher quality of DNA for downstream molecular analysis.   2.3.1.1 Hematoxylin and Eosin (H&E) Slide Preparation  Working in a fume hood, slides are immersed in xylene for 10 minutes, twice. They were then placed into 100% alcohol and submerged for two minutes, twice, followed by 95% alcohol for one minute, then 85% alcohol for one minute. The slides were then washed in running tap water, and placed in hematoxylin for five minutes.  Slides were washed in running tap water until the water was clear, and then placed in 1.5% sodium bicarbonate for 30 seconds. The slides were 46 washed in running tap water again and then placed into eosin for eight seconds. The slides were washed again and then placed into 75% alcohol, then 95% alcohol, then 100% alcohol for 30 seconds each. The slides were placed into xylene for five minutes, twice, and a coverslip was placed over the specimen using Permount (Sigma-Aldrich, St. Louis, MO, USA) mounting medium.   2.3.1.2 Methyl Green Slide Preparation  Working in a fume hood, slides were immersed in xylene for 10 minutes, twice. They were then placed into 100% alcohol and submerged for two minutes, twice, followed by 95% alcohol for one minute, then 85% alcohol for one minute. The slides were then washed in running tap water and placed in methyl green for five minutes.  Slides were washed in running tap water four times and then air dried briefly. The slides were then placed into 0.2% methyl green. Because methyl green is light sensitive, the jar was wrapped with tin foil and the light of the fume hood was turned off. The slides were washed again with running tap water four more times. Stained slides were then air dried.   2.3.2 Microdissection   The H&E slides were reviewed by the study pathologist (L.Z.) and were annotated with areas and grade of dysplasia to serve as the reference for microdissection. On samples stained with methyl green, areas of dysplasia were manually microdissected under an inverted microscope using a 23G needle, separating the epithelium from the underlying stroma. The epithelium with the pathology served as the experimental tissue, while the connective tissue served as a source of matched control DNA. Microdissected tissue was placed into labelled Eppendorf tubes and dried 47 completely. If there was minimal connective tissue, DNA from exfoliative cytology samples of the same patient were used (see section 2.6.5.1 for description of sample collection). All samples were assigned a unique code to blind research staff to sample identity.   2.3.3 DNA Extraction   Microdissected tissue was digested in a mixture of 270 µl TE-9 buffer (1 M Tris-HCl (pH 8.)/ 0.2 M EDTA (pH 7.5)/ 4 M NaCl/ddH20), 1% sodium dodecyl sulfate (SDS)and 20 µl of proteinase K (0.5 mg/mL) (PK) and incubated in a water bath at 48 °C, spiked twice daily with fresh PK, for 72 hours. DNA was extracted using a standard phenol–chloroform method.  Cold phenol-chloroform and TE-9 (PC-9) (-20°C) was added to each of the samples. After vortexing and centrifugation, the aqueous layer was removed and transferred to new tubes. The samples were vortexed and centrifuged once more and the aqueous layer was removed and transferred into 100% ethanol; 120 ul of 10M NH4 and 2 ul glycogen were added. Samples were placed in a -20 C freezer for 45 minutes to precipitate and protect the DNA. They were then centrifuged, the supernatants were removed, and the pellets were permitted to air dry. LoTE buffer (1 M Tris, pH 7.5/ 0.2 M EDTA, pH 7.5/ ddH20/Adjusted to pH 7.5) was used to re-suspend the sample prior to analysis.
  Quantification of the extracted nucleic acids was performed using a spectrophotometer (NanoDrop ND-100; PEQLAB Biotechnologie, Erlangen, Germany).  DNA quality was assessed by evaluating 260/280 and 260/230 ratios. Master stock was created by adding LoTE buffer to create a concentration of 50ng/ µl. A final working solution of 4ng/µl was used to perform the experiments. 48  2.3.4 Loss of Heterozygosity Microsatellite Assay   LOH analysis was performed on coded samples. Genomic DNA extracted from paired dysplastic epithelial tissue and normal control tissue were subjected to polymerase chain reaction (PCR) amplification of dinucleotide repeats containing sequence microsatellite markers labelled with [γ-32P] ATP. The PCR products were then separated using denaturing formamide-urea polyacrylamide gel electrophoresis and visualized by autoradiography.  2.3.4.1 Microsatellite Markers  The microsatellite markers used for DNA amplification were purchased from Research Genetics (Huntsville, AL, USA) and mapped to chromosome regions 9p21, 17p11.2, 17p13.1, 4q26, and 4q31.1 (see Table 2.1). These regions were chosen because they contain known or putative previously reported to be associated with malignant progression.(13, 17, 252, 254, 255)   Markers at the 9p21.3 region, surround the cyclin-dependent kinase inhibitor 2A (CDK2NA) gene (D9SIFNα, D9S1751, D9S1748, D9S171). CDK2NA is a TSG and codes for both p16INK4a and p14ARF proteins. Markers at 17p13.1 surround TP53 (D17SCHRNB1, D17STP53, D17S786). TP53 is also a TSG and codes for p53. Microsatellite marker D4SFABP2 was used to evaluate 4q26, while D4S243 was used to assess 4q33.  49 Table 2.1 Primer details  Chromosome Arm Primer Name Sequence  (5' -- 3') Annealing Temperature  (°C) Run  Distance  (cm) 9p 9p21.3 D9INFA F: TGCGCGTTAAGTTAATTGGTT  R: GTAAGGTGGAAACCCCCACT  55 32 D9S171 F: AGCTAAGTGAACCTCATCTCTGTCT  R: ACCCTAGCACTGATGGTATAGTCT  56 32 D9S1748 F: CACCTCAGAAGTCAGTGAGT  R: GTGCTTGAAATACACCTTTCC  58 22 D9S1751 F: TTGTTGATTCTGCCTTCAAAGTCTTTTAAC  R: CGTTAAGTCCTCTATTACACAGAG  56 32 17p 17p13.1 D17STP53 F: TGGATCCTCTTGCAGCAGCC  R: AACCCTTGTCCTTACCAGAA  60 20 D17SCHRNB1 F: CTCGAGCCCCCGCATTCAAGAA  R: AACTTTACTACAGGAGTTACACCC  55 30 D17S786 F: TACAGGGATAGGTAGCCGAG  R: GGATTTGGGCTCTTTTGTAA  59 32 4q 4q26 D4SFABP2 F: AACTCAGAACAGTGCCTGAC  R: ATTTCCCTCAAGGCTCCAGGT  55 36 4q33 D4S234 F: TCAGTCTCTCTTTCTCCTTGCA R: TAGGAGCCTGTGGTCCTGTT 57 25   50 2.3.4.2 End-Labelling and Dinucleotide PCR Reaction  One primer from each pair was end-labeled with [ γ-32P] ATP (20 mCi) (PerkinElmer®, Waltham, MA, USA) using the following protocol. For 1 – 20 samples, 19 µl of PCR-distilled water, 2.5 µl of 10x Polynucleotide Kinase buffer, 0.6 µl of 100x bovine serum albumin (BSA), and 1.5 µl of T4 polynucleotide kinase (New England BioLabs, Beverly, MA, USA) were added to one of each of the primer pairs.  Lastly, one microliter of [ γ-32P] ATP was added to the mixture before incubating at 37°C for one hour in a PCR thermal cycler.   Each PCR amplification used 10.1 µl  reaction volumes contained the following: 1.1 µl (4 ng/µl) genomic DNA, 5.4 µl PCR-distilled water, 1.125 µl PCR buffer (16.6 mmol/L ammonium sulfate, 67 mmol/L tris-hydroxymethy aminomethane (Tris) (pH 8.8), 6.7 mmol/L magnesium chloride, 10 mmol/L ß-mercaptoethanol, 6.7 mmol/L EDTA and 0.9% dimethylsulfoxide), 0.675 µl deoxynucleotide (dNTP) solution mix (100mM each dATP, dGTP, dCTP and dTTP (Thermo Fisher Scientific, Waltham, MA, USA)), 0.225 µl (100ng/µl) each of forward and reverse primer, 0.225 µl of TAQ DNA polymerase (GIBCO BRL, Gaithersburg, MD, USA), and 1.125 µl (100ng/ µl) of the labelled primer. PCR amplification was performed for 40 cycles consisting of denaturation at 95 °C for 30 seconds, annealing for 60 seconds (see Table 2.1 for individual primer annealing temperature) and extension at 70 °C for 60 seconds, with a final extension at 70 °C for 5 minutes.  2.3.4.3 Casting the Polyacrylamide Gel  After being cleaned thoroughly with anhydrous ethanol and KimwipesTM, two panes of glass were assembled to pour the gel. Spacers are placed along the long side of the larger glass, and the 51 smaller pane of glass, which had previously been treated with acrylease, was placed on top of the larger glass and spacers, treated side down. The lower edge and sides of the panes of glass were carefully aligned and gel clamps were placed along the sides of the glass assembly, while the lower edge was sealed with one continuous strip of gel sealing tape.   The assembled glass sandwich was placed upright to cast the gel. Sealing gel was created by combining 3.5 mL DINOC (175 mL 19:1 Acrylamide-Bis, 210 mL formamide, 226 g urea (Gibco BRL), 200 mL TBE (54 g Trizma® base,
27.5 g boric acid
2.93 g anhydrous EDTA made up to 1 L with ddH20 and filtered through coarse filter paper), 50 µl ammonium persulfate (APS) and 10 µl tetramethylethylenediamine (TEMED) in a 15 mL tube, and gently mixing with a disposable pipette. The sealing gel was gently squeezed between the glass, in a steady stream, and allowed to set. The loading gel was prepared by combining 75 mL DINOC, 1200 µl APS and 50 µl TEMED in an Erlenmeyer flask. The solution was gently swirled to mix, and transferred to a squeeze bottle. The solution was added between the panes of glass, taking care to avoid the formation of bubbles. When the gel reached the top, the comb was placed horizontally, teeth side upwards. The gel was then allowed to polymerize for at least four hours.  2.3.4.4 Running LOH    The comb was removed, and the resulting top of the gel was flushed with TPE buffer using a syringe. The gel was then placed into the gel box apparatus. The gel was run with five times diluted TBE buffer, set to a temperature of 47°C and power to 1800V, 85W and 60mA.  When the gel reached 30°C, the gel was flushed with TBE buffer again. The comb was then placed, teeth towards the gel, to a depth of 1-2mm, taking care not to puncture the gel surface.  When the 52 gel reached 38°C the wells were again rinsed with TBE buffer, using a syringe.   Next, 8.0 µl of gel loading dye (0.05 g 0.05% xylene cyanol, 0.05 g 0.05% bromophenol blue, 5mL of 0.2M EDTA, 95 mL 95% formamide) was added to each PCR product sample and mixed well.  Then, 2.9 µl of the PCR product was loaded into individual wells, and the gel was run to a specified distance (see Table 2.1 for distance run for each primer).   Once the gel has reached the required distance, the small glass was removed, and the gel was covered with chromatography paper (Anachemia, #3030-392). The gel and paper were flipped over and the gel surface was covered with Saran™ plastic wrap. Using a Geiger counter, the highest amount of radioactivity was recorded. This number was used to determine the duration of film exposure needed. The gel was placed into a film cassette and placed into an -80ºC freezer. Film (Fuji Medical X-ray Film, Super RX-N, 100 NIF, 35 x 43 cm; Ref. 47470 19339) was then placed inside the cassette, which was placed back into the -80ºC freezer and retrieved when the exposure time had elapsed. After the determined exposure time, films were removed and developed in dark room conditions, and were evaluated to determine whether additional exposures or sample reloads were needed. 
  2.3.4.5 Scoring LOH LOH scoring was done as a blind analysis on the coded samples, without knowledge of the sample diagnosis.   53 For informative cases, maternal and paternal alleles vary in the number of tandem repeats for the region spanned by a specific primer resulting in different fragment sizes that run to different lengths on the gel. The two alleles show as two separate bands on the x-ray film. Samples in which the copy number of either allele have been altered were identified by comparing the relative intensity of the two alleles of DNA from the experimental sample to that observed in the normal control tissue of that patient. The sample was scored as retained (R) if the intensity of the signals for the alleles was the same as those of the normal control. Allelic loss was recorded if either of allele band showed a reduction in band intensity as compared to the pattern seen for the normal control tissue of the patient. (256) A loss of intensity on the bottom allele was recorded as L1/0, while loss of the top allele was recorded as L0/2.  Homozygous cases, in which the tandem repeats on maternal and paternal alleles do not differ, appear as single bands on the film and were deemed as non-informative (NI).   2.4 Demographic Data   Demographic data, including date of birth, sex, and ethnicity were collected at study entry via a standardized study questionnaire (Appendix A.2).  2.5 Risk Habit Assessment  Self-reported information on risk habits, including the use of tobacco and alcohol, were collected via the same questionnaire (Appendix A.2). Both past and present tobacco use and alcohol consumption were recorded, with quantity, frequency and duration of usage recorded throughout the patient’s lifetime. The presence of a family history of head and neck cancer having been diagnosed in any biologically-related family members was also recorded.  54  2.6 Clinical Evaluation   2.6.1 Initial Visit  At the initial visit, the patient’s histological diagnosis and medical history were reviewed, and an extra-oral examination was performed to assess for any asymmetry or palpable lymph nodes. Next, a standardized intra-oral examination consisting of conventional white light examination FV examination and TB examination was conducted.   2.6.1.1 White Light Examination  Each of the histologically confirmed low-grade oral epithelial dysplasia were assigned a lesion code (for example, lesion site A (LSA), lesion site B (LSB), etcetera), and were mapped onto a standardized oral map (Appendix A.3). A lesion tracking sheet (Appendix A.4) was used to record clinicopathological data for each lesion. Lesion presence was recorded as present or absent (scar). Lesion site was recorded based on defined anatomical categories. Lesion size (longest length, by largest width, by greatest thickness) was measured using a calibrated probe, and recorded in millimeters. Lesion texture was noted as ulcerative, smooth, velvety, nodular, verrucous, nodular, fissured, and/or other. Colour was recorded as white (leukoplakia), red (erythroplakia), or mixed white and red (erythroleukoplakia). Lesion appearance was documented as either homogeneous (same colour and texture throughout) or as non-homogeneous (colour and texture not uniform). Lesion margins were either diffuse (ill-defined margins) or discrete (well-defined margins).   55 2.6.1.2 Fluorescence Visualization Examination With the room lighting turned off, tissue autofluorescence was examined with FV (VELscope™, LED Dental, Inc.). Lesions were recorded as FV loss (appearing dark, indicating loss of tissue autofluorescence), FV retained (retention of normal, green tissue autofluorescence), or equivocal (slightly darker or uncertain). The area of loss of fluorescence was also recorded using bidirectional measurement in millimeters.  2.6.1.3 Toluidine Blue Examination TB examination was conducted following the FV exam. Lesions were dried with gauze, followed by an application of 1% acetic acid via a cotton tipped applicator. The lesion area was then painted with a cotton tip applicator soaked in a 1% TB solution (1g of TB, 10 mL of acetic acid, 4.19 mL absolute alcohol), 86 mL distilled water and 125 drops 2 M NaOH; pH adjusted to 4.5). After 45 seconds, the area was swabbed with an additional cotton tip applicator soaked in 1% acetic acid to remove the dye. The lesion site was then rinsed with water. If the tissue was dark/royal blue, it was considered positive (TB +). If no stain was taken up, it was considered equivocal or negative (TB-), respectively.   2.6.1.4 Digital Photographs Digital images were taken after the white light examination (WLE), FV exam and TB staining (Nikon D7100 camera body, Nikon AF-S VR Micro-Nikkor 105mm f/2.8G lens, Metz Macablitz 15 MS-1macro ring flash). Photographs were taken using cheek retractors, and if possible, were taken perpendicular to the lesion site. Lesion surfaces were lightly dried using gauze, and rhodium plated intraoral photography mirrors were used for lesions that could not be imaged 56 directly.  FV photographs were taken using a Hoya K2 52mm yellow filter. These clinical images were used in the clinical follow-up of patients and were used to verify clinical data where required.   2.6.1.5 Exfoliative Cytology  Although not part of the analysis presented in this thesis, exfoliative cytology samples (brushings) are collected as part of the OCPL study protocols. Brushing are collected at the initial visit using a disposable cytology brush (Innotech, Vancouver). Exfoliated cells are collected from the lesion site, a high-risk normal control site (contralateral to lesion site), and a low-risk control site (non-lesion buccal mucosa site). The brush is broken off and placed into a one mL vial containing ThinPrep® PreservCyt® (Hologic, Marlborough, MA). These brushings of normal tissue were used to produce the LOH patterns that characterize normal tissue in which the quantity of DNA from the underlying stroma of a lesion biopsy is not sufficient to support its use as a control.    2.6.2 Follow Up Visits  Clinical follow-up visits occurred approximately every 6 months. The patient’s medical history and risk habit information were updated at each visit. Clinical examinations, including extra-oral examination, white light, FV and TB examination were also performed and clinicopathological data were collected as per the initial visit. WLE, FV, and TB digital photographs were taken and exfoliative cytology brushing of the lesion site was performed.   57 2.7 Outcome    Comparative biopsies of the index site for each lesion were collected approximately every 24 months, or upon significant clinical change, as judged by the expert opinion of the attending certified oral medicine specialist. Outcome was histologically proven progression to severe dysplasia, CIS, or SCC. Severe dysplasia was included as an endpoint, based on our previous findings that without treatment, progression occurred in 32% of patients with this diagnosis within 3 years; and 60% within 5 years.(177)   2.7.1 Biopsy Standard incisional biopsy procedures were followed. A topical application of 20% benzocaine was applied, and then the site was injected with 2% lidocaine with 1:50,000 epinephrine. Using a 5 mm biopsy punch (Integra® Miltex®), the clinically most suspicious area of the lesion was biopsied. The circular punched out tissue was grasped gently with fine tooth forceps and excised with a #15 scalpel. Care was taken to avoid crushing the specimen. The biopsy sample was placed mucosa side up onto 15 mm Whatman™ filter paper (Fischer Scientific). The specimen was placed into 10% buffered formalin and sent to OBS at Vancouver General Hospital for routine histopathological diagnosis (Appendix A.5). Hemostasis was achieved with direct pressure and the use of a silver nitrate stick (Henry Schein).  Participants were provided with oral wound care instructions.   2.7.2 Histological Evaluation  Histological evaluation was performed in the same manner as the initial evaluation at time of study entry. The biopsy specimens were formalin-fixed and paraffin-embedded by staff at the 58 OBS. Sections were cut and stained with H&E and histological diagnosis was provided by pathologists at the OBS (Appendix A.1). Histologic diagnoses were reviewed and confirmed by the study pathologist (L.Z.), using diagnostic criteria established by the WHO.(113)  59 Chapter 3: (Paper 1). “Dysplasia should not be ignored in lichenoid mucositis.” Journal of Dental Research. 2018 Jul;97(7)767-772.  3.1 Synopsis   Objectives: Oral lichen planus is categorized as a potentially malignant condition by the World Health Organization; however, some argue that only lichen planus with dysplasia have malignant potential. Many pathologists call lichen planus with dysplasia ‘dysplasia with lichenoid mucositis (LM)’, or LM with dysplasia’. Previous research has shown that certain high-risk patterns of loss of heterozygosity (LOH) in dysplastic lesions are associated with significantly increased cancer risk. However, LM without dysplasia lack such molecular patterns, supporting the hypothesis that LM, by itself, is not potentially malignant, and that only those with dysplasia have malignant potential. To further investigate the premalignant nature of LM with dysplasia, this study compared the rate of malignant progression of dysplasia with LM with that of dysplasia without LM.  Materials and Methods: Patients from a population-based prospective cohort study with more than 10 years of follow up were analyzed. Study eligibility included a histological diagnosis of a primary low-grade dysplasia with or without LM. A total of 446 lesions in 446 patients met the selection criteria; 373 (84%) were classified as dysplasia without LM, while 73 (16%) as dysplasia with LM. Demographic and habit information, clinical information and outcome (progression) were compared between the two groups.   Results: Forty-nine out of 373 dysplasia (13%) progressed compared to 8% (6/73) of dysplasia 60 with LM. However, the difference was not statistically different (P = 0.24). The 3- and 5-year rate of progression did not differ between the groups (6.7% and 12.5% for dysplasia without LM; and 2.9% and 6.6% for those with LM (P = 0.36)). Progression was associated with non-smoking, location at a high-risk site and diagnosis of moderate dysplasia regardless of whether LM was present or not.   Conclusion: Dysplasia with or without LM had similar cancer risk and dysplasia should not be discounted in the presence of LM.  3.2 Objective  1. To investigate the premalignant nature of LM with dysplasia by determining the risk of progression of LM with dysplasia, in order to better define this unique subset of patients and ultimately aid in the diagnosis and management of this disease.  3.3 Hypotheses  1. The proportion of malignant progression of LGD with LM will be the same as that of LGD without LM.  2. The time to malignant progression will be the same in LGD with LM with that of LGD without LM.  61 3.4 Published Paper  3.4.1 Introduction  Lichenoid mucositis (LM) refers to a group of mucosal lesions (e.g. lichen planus and LM from contact with dental materials or intake of drugs) that are characterized by a band-like lympho-histiocytic inflammation in the immediate subepithelial region.(127) It is hypothesized that such inflammation results from both antigen-specific cell-mediated immunity in response to antigenicity changes in the oral epithelial lining cells as well as non-specific mechanisms such as mast cell degranulation and matrix metalloproteinase (MMP) activation in oral lichen planus lesions (128, 129). If an allergen can be identified then a diagnosis of LM can be made. A diagnosis of lichen planus can only be made after ruling out LM, in which case one is able to identify the allergen. Consequently, LM from allergic contact or drugs could be cured by withdrawal of the allergen, but not lichen planus. The diagnosis of lichen planus therefore requires both histological assessment and clinical information. Unfortunately, pathologists often do not have the clinical information and therefore will tend to diagnose an LM as an oral lichen planus, and hence artificially inflate the incidence of lichen planus.   Oral lichen planus is categorized by the World Health Organization (WHO) as a potentially malignant condition (113); however, there are heated debates as to whether lichen planus should be considered premalignant. In 1985, Krutchkoff and Eisenberg reviewed literature on cancer progression of oral lichen planus and found that all reported oral lichen planus with cancer progression had shown low-grade dysplastic changes. Consequently, they concluded that lichen planus without dysplasia was not premalignant and only those with dysplasia were premalignant. They coined the term ‘lichenoid dysplasia’ for these dysplastic lesions with LM; however, the 62 term was not widely used and many called such lesions dysplasia with LM, LM with dysplasia or lichen planus with dysplasia. They hypothesized that the reason for lichen planus being classified as premalignant was due the failure to recognize low-grade dysplasia in the presence of lichenoid inflammation.   Previously, we developed and validated a risk prediction model that showed that certain patterns of loss of heterozygosity (LOH) are significantly associated with the risk of malignant progression in oral premalignant lesions. In this model, low-grade dysplasia (mild/moderate dysplasia) with intermediate-risk and high-risk LOH patterns demonstrated a 3.8-fold and 33-fold increased cancer risk respectively, as compared to those with a low-risk LOH pattern.(13) Previously, to investigate the possible premalignant nature of LM with or without dysplasia, we compared the LOH patterns of these lesions with reactive hyperplasia and low-grade dysplasia without LM. Our results showed that LM without dysplasia and reactive hyperplasia lacked the genetic characteristics of oral premalignant lesions (256); whereas LM with dysplastic changes showed similar genetic alterations to those of low-grade dysplasia without lichenoid changes (257), supporting the hypothesis that LM per se is not premalignant and that only LM with dysplasia is premalignant.   The objective of this study was to further investigate the premalignant nature of LM with dysplasia. This study compared the proportion and the rate of malignant progression of low-grade dysplasia with LM with that of low-grade dysplasia without LM. By determining the risk of progression of dysplasia with LM, we seek to better define this unique subset of patients and ultimately aid in the diagnosis, and management of this disease.  63  3.4.2 Materials and Methods 3.4.2.1 Patient Population  This cohort study involved patients who were enrolled in an ongoing Oral Cancer Prediction Longitudinal (OCPL) study between January 1, 1997 and October 4, 2013. Participants in the study were identified through a centralized population-based biopsy service, the BC Oral Biopsy Service, where community dentists and specialists across British Columbia (population 4.8 million, in 2016) send biopsies for histological diagnosis. Patients with a diagnosis of low-grade dysplasia (regardless of whether there are accompanying LM) were referred by these community clinicians, upon recommendation from the Oral Biopsy Service, for follow up in Oral Dysplasia Clinics, where they were invited to participate in the OCPL study. Study protocol and ethical approval were obtained from the University of British Columbia and BC Cancer Agency Research Ethics Board, and participants were recruited to the study using written informed consent. Study eligibility included a histological confirmed diagnosis of low-grade (mild or moderate) oral epithelial dysplasia with or without a LM and the patients had no prior history of oral cancer. A total of 446 lesions in 446 patients met the selection criteria, and were included in the present analysis, with a median follow-up time of 63.8 months (7.5 – 258.4 months). To address and reduce potential bias from inter-observer variability, the slides of the index biopsies as well as those of subsequent biopsies from the index biopsy sites were reviewed by an oral pathologist (LZ), and the diagnostic criteria were those of the WHO and papers from authorities in the field.(113, 127, 131, 258) Of the 446 lesions, 373 lesions were low-grade dysplasia without LM and 73 lesions were low-grade dysplasia with LM. LM was noted in 68 lesions in the subsequent biopsies from the 373 lesions that initially showed no lichenoid changes, resulting 64 in a total of 141 lesions with LM ever. Of the 446 lesions, 225 cases have been reported in a previous study, however that study was not aimed at comparing dysplasia with and without LM but rather at relationship between LOH pattern in dysplasia and outcome.(13)   3.4.2.2 Clinical pathological data, treatment and follow-up  The OCPL study collects demographic data, clinical information, as well as tobacco and alcohol habits at study entry. The primary endpoint of this study was time from index biopsy to histologically confirmed progression to severe dysplasia or higher, occurring at the same anatomical site as the index biopsy. Inclusion of severe dysplasia as the progression endpoint was based on our findings that without treatment, progression occurred in 32% of patients in 3 years; 59% in 5 years. (177)   3.4.2.3 Statistical analysis and Reporting  Data analyses were carried out using SPSS® Version 25.0 software (Armonk, NY: IBM Corp). The threshold for significance was set at P < 0.05, and all tests were 2-tailed. The inferential analysis included separate bivariate analyses between each independent and dependent variable. Categorical variables were tested using the Chi-square Test or Fisher’s Exact Test when more than 20% of cells contained expected frequencies of < 5. Quantitative variables were tested using an independent samples T-test; those that were not normally distributed were tested with the Mann-Whitney U test. To control for potential confounding, demographic, risk habit and clinical variables were assessed for significant differences between groups. Missing data was deemed to be minimal, likely unrelated to observed responses, and was handled by available case analysis.  Time-to-endpoint was calculated from date of the index biopsy to endpoint date or to last follow-65 up date (as of December 22, 2016), if no progression occurred. Since every patient had a different length of follow-up, time-to-progression curves and 3-year and 5-year progression rates were estimated using Kaplan–Meier analysis and the Log Rank test. Hazard ratios and the corresponding 95% confidence intervals (95% CI) were determined using the Cox proportional hazards regression model. This observational study conformed with Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for cohort studies.  3.4.3 Results  Of the 466 lesions that were included in the analysis, 373 (84%) were classified as low-grade dysplasia without LM, while 73 (16%) as low-grade dysplasia with LM. Table 3.1 compares patient characteristics between the two groups. There were no significant differences in age, sex, ethnicity and smoking habit between the groups, nor were there any significant differences in site, number of lesions or length of follow-up time.   Table 3.1 Comparison of oral epithelial dysplasia with and without lichenoid mucositis according to demographic, risk habit information and clinical features  All Low-Grade Dysplasia without LM£ (%)* Low-Grade Dysplasia  with LM£  (%)* P value Total 446 (100) 373  73   Age at diagnosis (n = 445) +  Mean (years ± SD) 57.3 ± 11.4 57.5 ± 11.6 56.0 ± 10.7 .289 Age Category (n = 445) +  <40 years 25(6) 22 (6) 3 (4) .719  40-60 years 246 (55) 203 (55) 43 (59)  >60 years 174 (39) 147 (39) 27 (37) 66  All Low-Grade Dysplasia without LM£ (%)* Low-Grade Dysplasia  with LM£  (%)* P value Sex  Male 229 (51) 199 (53) 30 (41) .055  Female 217 (49) 174 (47) 43 (59) Ethnicity (n = 445) +  Caucasian 375 (84) 318 (85/85) 57 (79) .194  Non-Caucasian 70 (16) 55 (79/15) 15 (21) Smoking Historya (n=440) +  Never 139 (32) 112 (81/30) 24 (38) .238  Ever 301 (68) 256 (85/70) 45 (62) Risk of Lesion Siteb  Low Risk 178 (40) 148 (40) 30 (41) .821  High Risk 268 (60) 225 (60) 43 (59) Multiple Lesion Sites  No 426 (96) 355 (83/95) 71 (97) .756  Yes 20 (4) 18 (90/5) 2 (3) Length of Follow-up§   Median months of follow-up (range) 63.8 (7.5 – 258.4) 64.1  (7.5 – 258.4) 63.0  (13.2 – 168.3) .798 £ Low-grade dysplasia = mild or moderate dysplasia; LM = lichenoid mucositis. * Column percentage reported. + One participant declined to provide date of birth; one participant declined to provide ethnicity; 5 participants declined to provide smoking history.  a Never smoker = less than 100 cigarettes in lifetime; Ever smoker = consumption of more than 100 cigarettes in lifetime.(259) b High Risk = floor of mouth and tongue; Low Risk = all other sites. § Months to last follow-up or progression, whichever occurred first.  During the study period, of the 446 lesions, 55 (12%) progressed; 26 to severe dysplasia, 4 to carcinoma in situ and 25 to squamous cell carcinoma. Age at diagnosis, sex, and ethnicity were not associated with progression (Table 3.2). A significantly higher proportion of progression occurred in never smokers. Never smokers were almost twice as likely to progress compared to 67 those who smoked (95% CI, 1.06 – 3.37; P = .03). Lesion site was significantly associated with progression. A lesion in a high-risk site (the floor of mouth or the tongue) possessed a greater than 3-fold increased risk of progression as compared to lower risk sites (such as the gingiva, palate, buccal or labial mucosa) (OR = 3.89; 95% CI, 1.85 – 8.17; P < .001).  A diagnosis of moderate dysplasia, regardless of whether LM was present, was associated with progression. Lesions with a diagnosis of moderate dysplasia were 2.3 times more likely to progress compared to those with a diagnosis of mild dysplasia (95% CI, 1.31 4.18; P = 0.003)68  Table 3.2 Distribution of cases according to outcome  All No Progression† (%)* Progression† (%)* P value Odds Ratio (95% CI) Total 446  391 (88) 55 (12)   Age at Diagnosis (n=445) +  Mean (years ± SD)  57.3 ± 11.4 57.5 ± 11.4 55.7 ± 11.4 0.27  Age Category (n=445) +  <40 years 25  21 (84) 4 (16)  0.69  1  40-60 years 246  214 (87) 32 (13) 0.79 (0.25 – 2.44)  >60 years 174  155 (89) 19 (11) 0.64 (0.20 – 2.08) Sex  Male 229  201 (88) 28 (12) 0.95 1  Female 217  190 (88) 27 (12) 1.02 (0.58 – 1.79) Ethnicity (n=445) +  Caucasian 375  330 (88) 45 (12) 0.59 1  Non-Caucasian 70  60 (86) 10 (14) 1.12 (0.58 – 2.56) Smoking History a (n=440) +  Never 139  115 (83) 24 (17) 0.03 1.89 (1.06 – 3.37)  Ever 301  271 (90) 30 (10) 1 Lesion Site b  Low Risk  178  169 (95) 9 (5) < 0.001 1   High Risk   268 222 (83) 46 (17) 3.89 (1.85 – 8.17)   69  All No Progression† (%)* Progression† (%)* P value Odds Ratio (95% CI) Diagnosis £  LGD without LM 373  324 (87) 49 (13) 0.24 1  LGD with LM 73  67 (92) 6 (8) 0.59 (0.24 – 1.44) Diagnosis £  D1 with or without LM 252  231 (92) 21 (8) 0.003 1  D2 with or without LM 194  160 (83) 34 (17) 2.34 (1.31 – 4.18) History of Lichenoid Diagnosis   Never 305  267 (86) 38 (13) 0.90 1  Ever 141  124 (88) 17 (12) 2.34 (1.31 – 4.18) Length of Follow-up§  Median months of follow-up (range) 63.8 (7.5 – 258.4) 68.9  (13.2 – 258.4) 38.0  (7.5 – 173.2) < 0.001  † Progression to severe dysplasia, carcinoma in-situ, or squamous cell carcinoma. * Row percentage reported. + One participant declined to provide date of birth; one participant declined to provide ethnicity; 5 participants declined to provide smoking history.  a Never smoker = less than 100 cigarettes in lifetime; Ever smoker = consumption of more than 100 cigarettes in lifetime.(259) b High Risk = floor of mouth and tongue; Low Risk = all other sites. £ LGD = low-grade dysplasia – mild or moderate dysplasia; LM – lichenoid mucositis; D1 = mild dysplasia; D2 = moderate dysplasia. § Months to last follow-up or progression, whichever occurred first.  70 The main objective of the study was to explore whether there were differences in the progression of low-grade dysplasia with LM compared to those with a straightforward diagnosis of low-grade dysplasia without LM.  Forty-nine out of 373 dysplasia (13%) progressed compared to 8% (6/73) of dysplasia with LM. However, the difference was not statistically different (P = 0.24) (Figure 3.1). Additionally, there was no significant difference in the risk of progression between low-grade dysplasia that had ever had a diagnosis of LM, and those that never possessed lichenoid features (P = 0.90).   Figure 3.1 The proportion of malignant progression was similar between low-grade dysplasia with lichenoid mucositis (LM) and those without LM.    Time to progression did not differ between the groups. Kaplan-Meier plots of progression by histological diagnosis showed very similar plots for the two groups of lesions, whether comparing index biopsies (P = 0.36), or whether there was ever a lichenoid diagnosis (P = 0.88) (Figure 3.2). Although the 3-year and 5-year probability of progression was higher for low-grade dysplasia without LM (6.7% and 12.5%. respectively) than it was for low-grade dysplasia with 71 LM (2.9% and 6.6%), it was not significantly significant (P = 0.36) (Table 3.3). The difference was even less for low-grade dysplasia with an ever-lichenoid diagnosis as compared to those with a never-lichenoid diagnosis (P = 0.88).   Figure 3.2 Low-grade dysplasia with or without lichenoid mucositis possess similar cancer risk     72 Table 3.3 Probability of progression in low-grade dysplasia with and without lichenoid mucositis £ Low-grade dysplasia = mild or moderate dysplasia; LM = lichenoid mucositis.  † Progression to severe dysplasia, carcinoma in situ, or squamous cell carcinoma. * Row percentage reported. § Months to last follow-up or progression, whichever occurred first.  3.4.4 Discussion  The results of this study indicate that low-grade oral epithelial dysplasia with LM possessed a similar risk of malignant transformation as those without LM. The limitations to the study are, like any prospective cohort study, it requires a large sample size and long follow up. This increases the study time, complexity and cost, and increases the potential of loss to follow-up. Although this study comprised the largest sample size to date to assess the malignant potential of low-grade dysplasia with LM compared to those without LM, there is a possibility of type II  All (%) Low-Grade Dysplasia  without LM£ (%)* Low-Grade Dysplasia   with LM£ (%)* P value Total 446 (100) 373 (84%) 73 (16%)  Months to progression† (n = 55)  Median (range)§ 38.0 (7.5 – 173.2) 36.3 (7.5 – 173.2) 38.3 (16.9 – 69.3) 0.63 Probability of Progression†    3-year (95% CI)  6.7 (5.4 – 8.0) 2.9 (0.9 – 4.9) 0.36   5-year (95% CI)  12.5 (10.6 – 14.4) 6.6 (3.4 – 9.8)  All (%) Low-Grade Dysplasia  without LM£ Ever (%)* Low-Grade Dysplasia  with LM£ Ever (%)* P value Total 446 (100) 305 (68%) 141 (32%)  Months to progression† (n = 55)  Median (range)§ 38.0 (7.5 – 173.2) 31.8 (7.5 – 173.2) 42.1 (7.7 – 139.2) 0.27 Probability of Progression†  3-year (95% CI)  6.9 (5.4 – 8.4) 4.4 (2.7 – 6.1) 0.88  5-year (95% CI)  12.7 (10.6 – 14.8) 9.4 (6.7 – 12.1) 73 error due to lack of sufficient power. However, it is difficult to obtain and follow larger numbers. Another potential limitation is the subjective nature of diagnosis. The critical element that allows separation of low-grade dysplasia with LM from LM or lichen planus is the additional presence of dysplastic features within the overlying epithelium. However, such features are often subtle and subjective, and it is difficult to sub-classify various inflammatory disorders accurately – histologic features often overlap one another. (117, 118)   It is well known that heavy inflammation could cause atypical epithelial changes resembling epithelial dysplasia. For this reason, many pathologists tend to discount low-grade epithelial dysplasia when there is heavy inflammation nearby. It is not clear however whether these reactive changes are limited to nonspecific inflammation, including the acute inflammation from epithelial ulceration and mixed chronic inflammation with plasma cells, lymphocytes and macrophages that are frequently seen in the oral cavity, such as gingivitis and periodontitis; or include both nonspecific and specific inflammation such as those caused by cell-mediated immunity as seen in LM. It is possible that in the case of LM, one should not discount any dysplasia despite the heavy specific inflammation, which in theory should only attack epithelial cells with antigenicity changes as opposed to nonspecific inflammation which attack all nearby cells without discrimination.   One of the dysplastic features as discussed by Krutchkoff and Eisenberg (1985) is prominent basal cells in many of the dysplasia with LM. We have also noted this frequently. It should be emphasized that one of the diagnostic features for LM or lichen planus is destruction or degeneration of basal cell layers since this is the first layer of cells to be attacked by the T 74 lymphocytes in the lamina propria. (127) The presence of prominent basal cells in areas of the lesion despite the heavy inflammation could suggest increased growth ability, a feature for carcinogenesis.  Recognition of dysplasia in the presence of LM is critical for appropriate management of such lesions. As shown in our data, there are no significant differences in cancer progression rate or speed between low-grade dysplasia with or without LM. In our experience in the Oral Biopsy Service, a number of oral cancer patients had dysplastic lesions prior to the oral cancer but these lesions were misdiagnosed as lichen planus by the pathologists and the dysplastic changes were discounted because of the inflammation, resulting in mismanagement of the lesions. In one case, a large non-homogeneous leukoplakia in the floor of mouth was diagnosed as lichen planus, even though there was moderate epithelial dysplasia when the slide was reviewed (unpublished data). This lesion later progressed into cancer. As discussed, our previous study had shown that dysplasia with LM had a similar high-risk LOH pattern, supporting the thesis that dysplasia is dysplasia regardless if there is accompanying LM.   In this study, we have shown that low-grade dysplasia with or without LM demonstrated similar cancer risk as judged by cancer progression rate and speed. This is a much higher level of supporting evidence than our previous study, particularly because this study was conducted within the framework of a longitudinal prospective cohort, the OCPL Study, the largest longitudinal study attempted to date. It is also unique in that it draws from a community-based rather than a high-risk population, thus the study results can be considered relevant to the population at large.  While this study strongly supports the premise that LM with dysplasia 75 should be regarded as premalignant, it does not rule out the possibility that LM or lichen planus could have a higher chance of becoming dysplasia as compared with normal mucosa. The presence of inflammation increases cell proliferation, which in turn may increase the chance of random mutations or replicative errors (260). It is therefore important to follow up these lesions as well.   3.4.5 Conclusion Low-grade dysplasia with or without LM had similar cancer risk and pathologists should not discount the dysplastic changes in the presence of LM. Clinicians should not disregard dysplasia in a pathology diagnosis, even if they believe the patient has lichen planus clinically. Lesions that demonstrate any dysplasia upon biopsy warrant careful follow-up.  3.4.6 Acknowledgments This work was supported by the BC Cancer Foundation and grants R01 DE13124 and R01 DE17013 from the National Institute of Dental and Craniofacial Research. L.D.R. is a recipient of a CIHR Doctoral Research Award (grant 379723) I.L. is a recipient of a University of British Columbia Faculty of Dentistry Summer Research Studentship. The authors declare no potential conflicts of interest with respect to the authorship and/or publication of this article.  76 Chapter 4:  (Paper 2). “Characterization of epithelial oral dysplasia in non-smokers: First steps towards precision medicine”. Oral Oncology. 2018; 78:119-125.  4.1 Synopsis   Objectives: Tobacco usage is the strongest risk factor in the development of oral squamous cell carcinoma (OSCC), which mandates careful screening for oral cancers in smokers. However, there are indications that oral potentially malignant lesions, such as oral epithelial dysplasia (OED), in non-smokers (NS) have a higher cancer risk than those in smokers. Without tobacco as an etiology, the development of these lesions in NS may suggest genetic susceptibility. The increasing incidence of OSCC in NS calls for a better understanding of the natural history of OED in NS as compared to that of smokers.
   Materials and Methods: Patients from a population-based longitudinal study with more than 10 years of follow up were analyzed. Of the 455 patients with primary OED (233 mild and 212 moderate dysplasia), 139 were NS and 306 were smokers. Demographic and habit information, clinical information (lesion site, size and appearance; toluidine blue and fluorescent visualization), microsatellite analysis for loss of heterozygosity (LOH) and outcome (progression) were compared between the two groups.   Results and Conclusions: The majority of patients with OED were smokers. Of these, more were males, Caucasians and heavy drinkers. A significantly higher number of OED in NS were in the tongue, whereas a significantly higher number of OED in smokers were in the floor of 77 mouth (FOM). OED in NS showed a greater than 2-fold increase in cancer progression. Strikingly, OED located in the FOM in NS showed a 38-fold increase in cancer progression as compared to those in smokers.   4.2 Objectives  1. To characterize the clinicopathological features and the genetic profile of LGD in NS.  2. To compare progression rates and time to progression between NS and smokers with OED.   4.3 Hypotheses  1. The clinicopathological features of LGD, including lesion size, site, appearance, margin characteristics, FV presentation and TB status, will appear different in smokers as compared to NS.  2. The proportion of malignant progression and the time to progression will differ between NS and smokers with LGD.   4.4 Published Paper 4.4.1 Introduction  Tobacco usage is the strongest risk factor for the development of oral squamous cell carcinoma (OSCC),(20, 53-55) which mandates careful screening for oral cancers in smokers. However, OSCC does develop in non-smokers (NS), and there are indications that oral potentially malignant lesions (OPML) in NS possess a higher cancer risk than 78 those in smokers.(13, 109, 140, 178) Without tobacco as an etiology, the development of these lesions in NS may suggest genetic susceptibility. Tobacco cessation efforts have resulted in a drop in oral cancer rates associated with this habit (72), leading to a growing interest in the increased proportion of cases occurring among NS. (73) The increasing incidence of oral cancer in NS petition a better understanding of the natural history of OPML in NS as compared to that of smokers.
  OPML with a histological diagnosis of oral epithelial dysplasia (OED) are at an increased risk of progressing to oral cancer than those without dysplasia.(9, 12, 111) Although the presence of dysplasia provides an indication of risk for higher grades of dysplasia (52, 177), it is a relatively poor predictor for OED with low-grade (mild/moderate) dysplasia, which represent the majority.(152)  A more precise risk stratification is required for low-grade lesions.   The study of OPMLs has been the focus of our research team for more than two decades, mainly with respect to the development of markers that would help in differentiating progressing from non-progressing mild/moderate dysplasia. The markers included clinical visual aids, such as toluidine blue (TB) staining (147), fluorescent visualization (FV) (221, 261) and microsatellite analysis of loss of heterozygosity (LOH).(13)   Microsatellite analysis for loss of heterozygosity (LOH) analysis is used to assess the loss of chromosomal regions that contain known or putative tumour suppressor genes. The Oral Cancer Prediction Longitudinal (OCPL) study being conducted at the BC Cancer Agency in Vancouver 79 (British Columbia, Canada) has reported a risk prediction model which uses LOH at key chromosomal loci to stratify lesions to risk of malignant progression.(13) To date, this PCR-based assay is the only marker that has been shown to predict malignant progression of low-grade OED and has been prospectively validated in an independent cohort of patients from community settings.(13, 262) Furthermore, it has been optimized for work with archival tissue and small DNA quantities.(248, 263-265)   Several studies have examined clinical characteristics and the prognosis of OSCC in NS. However, this question has not been explored thoroughly with respect to OED.(266-271) Not only is the natural history of OED in NS poorly understood, but the path to prevention and intervention of disease is not well defined in this group. There is a gap in the knowledge surrounding the clinicopathological and genetic characterization and the risk of progression in this growing category. This information is critical to the evolution of precision medicine in this subgroup by allowing for medical decisions, practices, and interventions to be tailored to the individual patient based on their predicted risk of disease.  This study reports on findings within the ongoing OCPL study, of which the overall goal is to establish a risk model for the malignant progression of low-grade OED. The purpose of the present study was to characterize the clinicopathological features and the genetic profile of low-grade OED in NS, as well as to compare progression rates and time to progression between NS and smokers with OED. By describing the clinical characteristics of OED in NS, we seek to better define this unique subset of patients and ultimately aid in the prevention, diagnosis, and management of this disease.  80  4.4.2 Materials and Methods    Since January 1, 1997, the OCPL study has prospectively enrolled and followed patients with low-grade OED to a primary endpoint of malignant progression to severe dysplasia, carcinoma in situ (CIS), or SCC. Participants in the study were identified through a centralized population-based biopsy service, the BC Oral Biopsy Service, where community dentists and specialists across British Columbia (population 4.6 million, in 2014) send biopsies for histological diagnosis. Patients with a diagnosis of low-grade OED were referred by these community clinicians, upon recommendation from the OBS, for follow up to Oral Dysplasia Clinics, where they were invited to participate in the OCPL study. Study protocol and ethical approval was obtained from the University of British Columbia/ BC Cancer Agency Research Ethics Board, and participants were accrued to the study using written informed consent.   The current study is a focused analysis which used a subgroup of the OCPL study population. Eligibility criteria for this analysis required a histologically confirmed primary mild or moderate OED with lesion clinicopathogical and tobacco history available and no prior history of oral cancer. Participants were followed a minimum of 12 months, or to progression, whichever occurred first. No participants were excluded, unless they did not meet the criteria. A total of 445 subjects met the selection criteria and were included in the present analysis, with a median follow-up time of 55.4 months (3.3 – 241.4 months). Of the 455 cases reported, 275 were reported in a previously published study involving patients with primary OPML.(13)  81 Detailed past and present tobacco and alcohol habits were collected by a standardized questionnaire at study entry. Past and current smoking status, as well as amount and form of tobacco (cigarette, pipe, cigar or smokeless tobacco), were documented. Pipe, cigar and smokeless tobacco were recorded if the subject indicated that they had used this form of tobacco more than once per week for one year or longer.(272) Cigarette equivalents were calculated as one pipe equaled three cigarettes, and one cigar equaled two cigarettes. Smoker was defined as having consumed more than 100 cigarettes (or the equivalent) in one’s life time.(259) Periods of time where a subject had temporarily or permanently quit smoking were recorded. Lifetime smoking history over the subject’s entire life, including amount smoked per day during specific age categories, was collated as a pack-year calculation. A pack-year was defined as the equivalent of smoking 20 cigarettes (1 pack) per day for 1 year.  Average weekly alcohol consumption was recorded. One alcoholic drink was defined as 8 ounces of beer, 4 ounces of wine or 1 ounce of spirits. Heavy drinker was defined as consumption of more than 14 drinks per week for women and 21 drinks per week for men.(273, 274)  Clinicopathological data, including lesion site, size, appearance, lesion margin characteristics, as well as information on FV retention and TB positivity were included in the analysis. Lesion size was measured using a calibrated probe and recorded with a bidirectional measurement in millimeters. Lesion appearance was documented as either homogeneous (same colour and texture throughout) or as non-homogeneous (colour and texture not uniform). Lesion margins were either ill-defined or well-defined. Index lesions were assessed for FV and TB status as previously described.(147, 221) LOH analysis was performed on index biopsies collected at 82 baseline, and lesions were classified as low, intermediate or high risk of progression, using previously published methods.(13, 256)   Clinical follow-up visits occurred every 6 months. Comparative biopsies of the index site were performed upon significant clinical change or approximately every 24 months if no significant change. Outcome was histologically proven progression to severe dysplasia, CIS, or SCC. Inclusion of severe dysplasia as the progression endpoint was based on our findings that without treatment, progression occurred in 32% of patients in three years; 60% in five years.(177)  Data analyses were carried out using SPSS® Version 24.0 software (Armonk, NY: IBM Corp). The threshold for significance was set at P < 0.05, and all tests were 2-tailed. The inferential analysis included separate bivariate analyses between each independent and dependent variable. Categorical variables were tested using the Chi-square Test or Fisher’s Exact Test when more than 20% of cells contained expected frequencies of greater than five. Quantitative variables were tested using an independent samples T-test; those that were not normally distributed were tested with the Mann-Whitney U test. Interaction effects between tobacco and gender, site and alcohol were evaluated with respect to progression, using a binomial logistic regression model. The main analyses were based on the time-to-event outcome. Time to endpoint was calculated from date of the index biopsy to endpoint date or to last follow-up date (as of Nov 15, 2016), if no progression occurred. Time-to-progression curves and 3-year and 5-year progression rates were estimated using Kaplan–Meier analysis and the Log Rank test. Hazard ratios and the corresponding 95% confidence intervals (95% CI) were determined using the Cox proportional hazards regression model. 83  4.4.3 Results  4.4.3.1 Sociodemographic and Lifestyle Characteristics  A total of 445 subjects were included in the analysis. Approximately one third (31%) of the subjects were NS. Sixty-nine percent of subjects were smokers; 3.4% had reported having used chewing tobacco, 6.5% reported using cigars and 4.9% reported smoking a pipe. Table 4.1 shows the distribution of cases of OED according to sociodemographic and lifestyle variables in NS as compared to smokers. The majority were Caucasian and over the age of 40, and males were more likely to be smokers than females were. Age at diagnosis was not significantly associated with smoking status. Gender and ethnicity were significant for smoking status (P = .01 and P < .001, respectively). Alcohol consumption was also associated with smoking status. Heavy consumers of alcohol were 6.6 times more likely to have smoked than those who were light drinkers or who abstained (95% CI, 2.58 – 16.76; P < 0.001). Gender, ethnicity and alcohol category were each tested in multivariable analysis to see if interaction with smoking status was predictive of malignant progression.  When combined with smoking status, neither gender (P = 0.36), ethnicity (P = 0.86), or alcohol consumption (P = 0.85), was significantly associated with progression.    84 Table 4.1 Distribution of cases according to sociodemographic and lifestyle variables  ALL  Non-Smokera (%)* Smokerb  (%)* P value Total 445(100) 139  306   Age at diagnosis (n= 444)   Mean (years ± SD) 58.8 ± 11.86 60.1 ± 12.43 58.2 ± 11.55  0.10 Age Category (n=444)   <40 years 18 (4) 6 (4) 12 (4) 0.35  40 – 60 years 227 (51) 64 (46) 163 (53)  ≥60 years 199 (45) 69 (50) 130 (43) Gender (n=445)  Female 220 (49) 81 (58) 139 (45) 0.01  Male 225 (51) 58 (42) 167 (55) Ethnicity (n=445)  Caucasian 368 (83) 99 (71) 269 (88) < 0.001  Asian 37 (8) 21 (15) 16 (5)  South Asian 29 (7) 14 (10) 15 (5)  Otherc 11 (2) 5 (4) 6 (2) Alcohol Categoryd (n=441)  None/Light 376 (85) 133 (96) 243 (80) < 0.001  Heavy 65 (15) 5 (4) 60 (20) *Column percentage reported.  a Non-smoker was defined as less than 100 cigarettes in life time.(259) b Smoker was defined as consumption of more than 100 cigarettes in life time.(259)
 c 3 Hispanic, 2 African American, 2 North American Aboriginal/First Nations, 1 Mixed, 1 Unknown.  d Heavy drinker is defined as consumption of more than 14 drinks per week for women and 21 drinks per week for men. 1 drink = 8oz beer or 4oz wine or 1oz spirits.(274, 275)  4.4.3.2 Clinicopathological Features The first aim of the study was to characterize the clinicopathological features of OED in NS. Clinical features, including lesion size, texture, colour, appearance, margin characteristics, FV status and TB status, did not differ significantly between smokers and 85 NS (Table 4.2). Smokers were more likely to have OED at the palate, retromolar trigone or floor of the mouth (FOM) (P < 0.001). Dysplastic lesions on the tongue were 7.3 times more likely to progress than OED elsewhere in the oral cavity (95% CI, 1.71 – 31.11; P < 0.001). Lesion size (P < 0.001), non-homogeneous appearance (P = 0.01), loss of FV (P = 0.01), TB positivity (P = 0.001), and grade of dysplasia (P = 0.002) were also significantly associated with progression. Strikingly, when lesion site was analyzed together with smoking status, interaction analysis revealed that NS with a lesion on the FOM possessed a 38-fold increased risk of progression as compared to smokers (95% CI, 3.35 – 440.26; P < 0.003).86 Table 4.2 Clinicopathological and histopathological features according to smoking status and according to outcome  ALL Non-Smokera  (%)*  Smokerb (%)* P value Odds Ratio (95%CI) No Progression (%) Progression (%) P value Odds Ratio (95%CI) Total 445 139 306   385 60   Size at diagnosis (n=402)   Median (mm2)  (IQRc) 160             (50 – 378) 160             (65 – 414) 155         (48 – 360)  0.54  135           (48– 324) 297         (108 – 600) (%)*  Site (n=445)  gingiva 54 (12) 18 (13) 36 (12) < 0.001 1 52 (14) 2 (1) < 0.001 1  buccal / vestibule mucosa  63 (14) 19 (14) 44 (14) 1.16 (0.53 – 2.53) 58 (15) 5 (8) 2.24 (0.42 – 12.1)   palate/ retromolar / trigone 54 (12) 9 (6) 45 (15) 2.50 (1.01 – 6.23) 49 (13) 5 (8) 2.65 (0.49 – 14.32)   tongue 201 (45) 85 (61) 116 (38) 0.68 (0.36 – 1.28) 157 (41) 44 (73) 7.29 (1.71 – 31.11)  FOMd 73 (16) 8 (6) 65 (21) 4.06 (1.61 – 10.27) 69 (18) 4 (1) 1.51 (0.27 – 8.55) Appearance (n=385)  Homogeneous 221 (57) 67 (54) 154 (59) 0.36 1 199 (60) 22 (40) 0.01 1  Non-homogeneous 164 (43) 57 (46) 107 (41) 0.82 (0.53 – 1.26) 131 (40) 33 (60) 2.28 (1.27 – 4.08) Marginse (n=343)  Well-defined 114 (33) 34 (30) 80 (35) 0.31 1 101 (34) 13 (30) 0.66 1  Ill-defined 229 (67) 81 (70) 148 (65) 0.77 (0.48 – 12.6) 199 (66) 30 (70) 1.17 (0.59 – 2.34) FVf Results (n=272)  FV retention  98 (36) 30 (32) 68 (38) 0.26 1 92 (39) 6 (39) 0.01 1  FV loss  174 (64) 65 (68) 109 (62) 0.74 (0.44 – 1.26) 145 (61) 29 (83) 3.07 (1.23 – 7.67) TBg Results (n=387)  TB negative 300 (78) 90 (73) 210 (80) 0.16 1 266 (80) 34 (61) 0.001 1  TB positive  87 (22) 33 (27) 54 (20) 0.70 (0.43 -1.16) 65 (20) 22 (39) 2.65 (1.45 – 4.83) Diagnosis (n=445)  Mild Dysplasia 233 (52) 71 (51) 162 (53) 0.72 1 213  20  0.002 1  Moderate Dysplasia 212 (48) 68 (49) 144 (47) 0.93 (0.62 – 1.39) 172  40 2.48 (1.40 – 4.39) 87  ALL Non-Smokera  (%)*  Smokerb (%)* P value Odds Ratio (95%CI) No Progression (%) Progression (%) P value Odds Ratio (95%CI) Length of Follow-up§  Median months of follow-up (range) 55.4           (3.3 – 241.4) 59.9          (3.3 – 222.7) 55.1        (3.6– 241.4) 0.51  59.9         (12.0 – 241.4) 32.0         (3.3 – 222.7) < 0.001  *Column percentage reported.  a Non-smoker was defined as less than 100 cigarettes in life time.(259) b Smoker was defined as consumption of more than 100 cigarettes in life time.(259) c IQR = interquartile range.
 d FOM = floor of mouth. e discrete = well-defined; diffuse = ill-defined. f FV = fluorescence visualization. gTB = toluidine blue. §Months to last follow-up or progression, whichever occurred first. 88 4.4.3.3 Outcome  The second aim of the study was to explore whether there were differences in progression between smokers and NS with OED. Out of 445 subjects, 60 (13%) cases progressed (Table 4.3); 33 to severe dysplasia (7%), 5 to CIS (1%), and 22 to SCC (5%). A significantly higher proportion of progression occurred in NS: NS were more than twice as likely to progress than those who smoked (95% CI, 1.24 – 3.76; P = 0.006). When smokers were further categorized into former smoker (FS) and continuing smoker (CS), NS possessed a 4-fold increased risk of progression as compared to that of CS (P = .004). Amount of smoking was also negatively associated with progression: NS possessed more than twice the risk of heavy smokers (HR = 2.31; 95% CI, 1.16 - 4.60; P = 0.02).   Table 4.3 Distribution of risk factor variables according to outcome  ALL No Progression  (%)* Progressed (%)* P value Odds Ratio  (95%CI) Total 445 (100) 385  60    Tobacco History (n=445)   NSa 139 (31) 111 (29) 28 (47) 0.006 2.16 (1.24 – 3.76)   Smokerb 306 (69) 274 (71) 32 (53) 1 Tobacco History (n=445)   NSa 139 (31) 111 (29) 28 (47) 0.004 3.93 (1.65 – 9.37)  FSc  190 (43) 165 (43) 25 (42) 2.34 (0.99 – 5.65)  CSd 116 (26) 109 (28) 7 (12) 1  Total Pack-yeare (n=445)  Median pack-year (IQRf) 10.5              (0.0 – 30.0) 12.8              (0.0 – 30.9) 0.0                (0.0 – 17.1) 0.05§  89  ALL No Progression  (%)* Progressed (%)* P value Odds Ratio  (95%CI) Tobacco Amount Category  (n=445)  NSa 139 (31) 111 (29) 28 (47) 0.02 2.31 (1.16 - 4.60)  Lightg 159 (36) 141 (37) 18 (30) 1.17 (0.56 – 2.44)  Heavyg 142 (32) 128 (34) 14 (23) 1 Alcohol Categoryh (n=441)  None/Light 376 (85) 324 (85) 52 (87) 0.74 1  Heavy 65 (15) 57 (15) 8 (13) 0.87 (0.40 – 1.94) *Column percentage reported.  § exponential distribution of data, logarithmic transformation applied. a NS = non-smoker; defined as less than 100 cigarettes in life time.(259) b Smoker was defined as consumption of more than 100 cigarettes in lifetime.(259)
 c FS = former smoker; defined as smoker who quit smoking at or before diagnosis. d CS = current smoker; defined as smoker who continued to smoke after diagnosis. e A pack-year is defined as the equivalent of smoking 20 cigarettes per day for 1 year. f IQR = interquartile range. g Light smoker was defined as smoker and pack-year total less than median (23.8); Heavy smoker was defined as smoker and pack-year total greater than median (23.8).  h Heavy drinker was defined as more than 14 units in females and 20 in males.(274) One unit was defined as 8 oz. beer, 4 oz. wine, or 1 oz. spirits.(275)  Time to progression occurred faster in NS as well (Figure 4.1). Table 4.4 compares the probability of progression in NS and in smokers, showing 3- and 5-year rates. Both 3-year and 5-year progression rates were higher in NS than those in smokers (3-year: 12.7% vs. 5.5%; 5-year: 16.6% vs. 10.1%, respectively) (P = 0.002). Length of follow up did not differ significantly between the groups (median time of 66.2 months for NS, 60.4 months for smokers; P = 0.07).     90 Figure 4.1 Kaplan-Meier plot of time to progression in smokers vs. non-smokers.    Table 4.4 Probability of progression in smokers versus non-smokers  ALL Non-Smokera  Smokerb P value Total 445  139 306  Probability of Progression†     3-year (95% CI)   12.7 (9.8 – 15.6) 5.5 (4.1 – 6.9) 0.02   5-year (95% CI)   16.6 (13.2 – 20) 10.1 (8.1 – 12.8) a Non-smoker was defined as less than 100 cigarettes in life time.(259) b Smoker was defined as consumption of more than 100 cigarettes in life time.(259) † Progression defined as progression to severe dysplasia, carcinoma in-situ, squamous cell carcinoma.  When the LOH risk model was used to examine outcome in NS compared to that in smokers, Cox regression analysis showed that LOH risk patterns were strongly associated 91 with progression and was sensitive in both groups. Overall, lesions in the high-risk category had a 25-fold increased risk of progression (95% CI 8.50 – 76.69; P < 0.001) as compared to those in a low-risk category. However, NS in a high-risk category possessed much higher risk (HR = 60.74; 95% CI, 7.17 – 514.51; P < 0.001) than smokers (HR=15.09; 95% CI, 3.98 – 57.25; P < 0.001) (Figure 4.2).   Figure 4.2 Cox proportional hazard regression model analysis for LOH risk pattern in non-smokers compared to smokers.    4.4.4 Discussion    This study characterizes both the clinicopathological features and the genetic profile of OED in 92 NS and associates these findings with outcome in a large number of patients in longitudinal follow up. Although several studies have explored the association between clinical or genomic characteristics and outcome of OSCC in NS, (266-269, 271, 275, 276) few studies have explored these considerations with respect to OED. Although previous studies have reported a higher transformation rate in NS,(13, 109, 125, 140, 178, 276) this study is more comprehensive in that the primary focus is to compare multiple parameters (histological, clinicopathological and genetic) between smokers and NS, as well as to evaluate the interaction of smoking status with these parameters in association with progression. In 2012, Ho et al.(125) found that non-smoking status and tongue subsite had the highest risk of transformation. Our study has supported the findings of previous studies in OPML, by confirming that NS with OED possess a significantly elevated risk for progression, and has presented new findings in interaction analysis with clinical features, the genetic risk models, as well as the proportion and time to progression among smokers and NS. This study was conducted within the framework of a prospective clinical trial, the OCPL Study, the largest longitudinal study attempted to date, and is unique in that it draws from a community-based population. The study design demonstrates clear temporal sequence between exposure and outcome. The limitations are the same inherent limitations as those of any prospective cohort study: it requires a large sample size and long follow up.  Long latency periods increase the study time, complexity, and cost, as well as increase the potential of loss to follow-up. Another potential limitation comes from the self-reported smoking data which requires the participant to recall and report this information accurately.    Tobacco use is considered one of the most significant risk factors for OSCC.(20, 53-55). However, this environmental exposure is not the only pathway to oral cancer. Alcohol 93 consumption is also recognized as an independent risk factor for OSCC.(53, 54, 65-67, 69) There is also evidence that suggests tobacco and alcohol act synergistically to contribute to OSCC risk.(27, 44, 53, 54, 71) Although alcohol was strongly associated with smoking status in this data set, alcohol alone had no association with progression (P = 0.65). Like other studies that have examined alcohol and tobacco interaction in the etiology of OSCC, results of this study are hampered by the low numbers of heavy drinkers who do not use tobacco (n = 5). The interaction between alcohol category (none/light vs. heavy) and smoking (NS vs. smoker) was not predictive of progression (P = 0.85).  Similarly, interaction between number of tobacco pack-years and weekly alcohol consumption was also not predictive of progression (P = 0.19).   With increasing evidence of the etiological role of human papilloma virus (HPV) in the development of cancers of many human organs and tissues, one could hypothesize that HPV may play an important etiological role for oral SCC in NS. However unpublished data from our lab has shown a higher percentage of oral SCC with HPV DNA in smokers (9%, 13/135) than that of NS (3%, 2/76) although the difference was not significant (P = 0.09).     The data presented in this analysis confirm that although smokers are more likely to develop OED, when OED does occur in NS, they are at higher risk for cancer progression. Our findings not only clearly demonstrate a significantly elevated risk for malignant progression in NS with OED, but also reveal that OED in NS progress more quickly than smoking associated OED. LOH markers can delineate high risk lesions, regardless of risk habits, and should be an important consideration in the management of OED.   94 Cancer development is believed to be underlined by accumulation of mutations of driver genes through exposure to environmental carcinogens or hereditary predisposition. Recently Tomasetti and Vogelstein (260) have proposed a third theory for the mutation - mutations resulting from the random mistakes made during normal DNA replication, or replicative errors. It has been proposed that up to two-thirds of human cancers are a result of such errors.(277) It is possible that OED in NS is driven either by inherited predisposition and/or by replicative errors, versus smokers whose OED is more likely attributable to mutations that are environmental in etiology. To test the hypothesis that the progression risk model would differ in the OED of NS and those of smokers, we examined the chromosomal changes in regions of hypothesized tumour suppressor genes at 3p, 4q, 8p, 9p, 11q, 13q and 17p. The previously published LOH risk model, which uses LOH at 9p, 17p and 4q to predict the cancer risk of OED, was still the best risk model and equally predictive of progression in both smokers and NS.(13) The similarities in the prediction models could be interpreted as showing that the genetic alterations are similar between smokers and NS, regardless of how these changes are acquired, i.e., through environmental carcinogens, genetic predisposition or replicative errors. On the other hand, OED in NS may involve unique genetic mutations, which are driving progression, which have not yet been identified. Further genomic characterization, using methods such as next genome sequencing (NGS), would be needed to provide valuable insight into the differences in the molecular pathogenesis of OSCC associated with cigarette smoking and that of NS.   It is generally accepted that OED is at risk of progression to SCC, (9, 12, 111) although no universally accepted guidelines for the management of low-grade OED exist. Therefore, it is suggested that the secondary prevention of SCC, from OED, should utilize, not only the 95 histological diagnosis of dysplasia, but also more objective biomarkers of the risk of transformation. The need to find molecular markers for the risk of OSCC, and the importance of the implications for the prevention and early detection, has been highlighted by others.(264, 278-280) The term precision medicine, or personalized medicine, refers to the ability to make medical decisions and offer treatment or interventions tailored to the individual patient based on their predicted risk of disease.(281) The ability to identify low-grade lesions that are at risk for progression paves the way for interception, or the idea that premalignant lesions (OPML) can actively be treated to reduce the risk of the lesion becoming a full blown cancer.(282). These high-risk individuals could be offered more aggressive treatment options and more intensive follow-up; they are also prime candidates to target for chemoprevention trials.  4.4.5 Conclusion   Clinicians should be diligent in screening for cancer in both smokers and NS. Tobacco remains one of the strongest risk factors for the development of OSCC, yet for patients with a histologically confirmed OED, NS have increased cancer risk. With smoking eliminated as an etiology, their development in these patients suggest either genetic susceptibility or replicative errors. These findings substantiate the risk of progression in NS and emphasize the need for clinicians to consider smoking history and the molecular profiles in the triage and management of OED. The consideration of smoking history and LOH risk category marks the evolution of a systematic decision-making process for this very heterogeneous group of lesions and an important move towards clinical application of these markers in a way that minimizes patient morbidity while maximizing health system and cost efficiency. This information is critical to the 96 evolution of precision medicine in this subgroup by allowing for medical decisions, practices, and interventions to be tailored to the individual patient based on their predicted risk of disease.   4.4.6 Acknowledgement   This work was supported by grants from the British Columbia Cancer Foundation, the National Institutes of Health (R01DE13124), and the National Institute of Dental and Craniofacial Research (R01DE17013).  97 Chapter 5: (Paper 3). “Molecular analysis and repeated measures of clinical risk indicators – building a framework to predict the malignant transformation of low-grade oral dysplasia”. 5.1 Synopsis Objectives: A major barrier to oral cancer prevention has been the lack of validated risk predictors to differentiate between those LGD at high-risk from those at low-risk of progressing to SCC. LOH at key regions has been established as a validated molecular marker for stratifying risk. However, improvement is needed in the model, particularly for the intermediate-risk category, which represents a substantial proportion of cases. Previous research has examined baseline clinicopathological features and their association with malignant progression; yet, the association with temporal repeated measurements over time has not been assessed. Phenotypic lesion changes over time captured as repeated assessments of clinical features may improve risk prediction. The aim of this research was to advance a risk prediction model for the malignant progression of oral LGD by fusing validated molecular risk predictors to repeated measurements of clinical change.   Materials and Methods: Analysis involved a prospective cohort of 306 patients with primary mild/ moderate oral dysplasia enrolled in the OCPL study that were followed to an outcome of malignant progression. Demographic, habit, and clinical information were collected at baseline. Index biopsies were examined by microsatellite analysis for LOH (9p, 17p 4q). Lesion presence, size, appearance; colour, texture, TB positivity and FV status were collected longitudinally, every six months. Multilevel analyses were developed to utilize repeated measurements of clinical features and to compare their association with stratified molecular risk category and 98 progression. Risk models were constructed and refined using clustered multivariable Cox regression and recursive partitioning analyses.  Results and Conclusions: Multilevel regression analysis demonstrated that repeated measurements of clinical features are stronger predictors of progression than a single baseline measurement. Multivariable analysis showed that after LOH risk category, temporal TB status was the most significant predictor of progression. Two models are presented; one which describes and predicts risk over the natural history of the disease, and the other with the potential utility to guide clinical decision-making during the course of the disease. This research explores clinical changes as they occur over time, and offers insight into underlying biological changes that may be happening in the microenvironment as lesions progress towards malignancy. This analysis into temporal patterns marks an evolution of a new concept and provides an important framework for how to integrate repeated measurements of change over time into risk models and the study of the natural history of malignant progression.   5.2 Introduction  5.2.1 Background and Rationale  With an estimated world-wide incidence of 300,000 new cases, and 145,000 deaths, OSCC is a major cause of cancer-associated morbidity and mortality.(1) OSCC may develop from clinically visible OPML. OPML with a histopathological diagnosis of OED are at risk of progressing to OSCC. However, not all OED will progress to cancer,(7-9) and predicting which LGD (the majority of dysplasia) is at risk of progression is difficult.(14-16) A major barrier to oral cancer prevention has been the lack of validated risk predictors to differentiate between those LGD at 99 high-risk from those at low-risk of progressing to SCC.(13, 17) This creates a challenge for the long-term clinical management of these lesions. Furthermore, deciding when to do a comparative biopsy can be difficult. There is a pressing need for the development of visual aids and biomarkers that will facilitate the detection of OPML with a high-risk of progression. The ability to detect lesions at high-risk of progression could allow for informed management approaches, and ultimately better patient outcomes.   Our lab has established a unique longitudinal cohort in British Columbia, which has led to the development of a LOH risk model for progression. In 2012, we reported a validated model based on a genome marker-based assay that categorizes the LOH patterns of LGD into groups at low, intermediate, and high-risk of progression.(13) The model has a strong predictive value: 1.5% of low-risk LGD progressed within five years as compared to 53.6% high-risk LGD (P < 0.001).The problem is that a considerable proportion of lesions (40.5%) fall into the intermediate-risk category. Further refinement of the model is required.   Clinicopathological features, such as lesion site, size, colour and texture, have shown some prognostic capability, with the strength of association ranging from weak to strong.(132) Research to date has been confined to measuring the correlation between these clinical features and the risk of progression at one point in time.  Our lab has reported a significant association in some clinical features and malignant progression, including lesion presence in a high-risk site (ventrolateral tongue or floor of mouth) (13) or with TB positivity (147). However, the significant results were limited to measuring these features at baseline only, and did not add value to risk prediction when added to the LOH model. Progressive lesions change over time. 100 These changes are driven by genetic mutations and alterations to other systemic components such as the micro-environment, and lead to a change in lesion appearance. Phenotypic changes captured as repeated assessments of clinical features over time may improve risk prediction and are valuable for examining the natural history of the disease and may give insight into what is important in the development of the disease.   A gap exists in the current knowledge of the natural history and malignant risk prediction of LGD. In order to accurately identify the population that would benefit the most from screening and then correctly distinguish the lesions that would benefit the most from treatment, there has been a call for studies that integrate clinical and genomic and molecular variables into our understanding of the underlying biology of neoplastic development.(18) Furthermore, longitudinal studies are required to examine malignant development over time, not at a single time point. The question is whether the accumulation of genetic changes and the presence of high-risk molecular clones translate to phenotypic changes resulting in changes in clinicopathological features over time, and can predict malignant progression. Research in other types of cancer have shown that repeated assessments can improve risk predictions.(283-286) Equally important is to begin the process of developing frameworks that can be used for future temporal analyses, to generate a better understanding of clinical and biological changes and facilitate research into the natural history of the disease, and can be used for patient follow up.   The goal of this research was to advance a risk prediction model for the malignant progression of oral LGD by fusing previously validated molecular risk predictors to temporal clinical markers, 101 and to develop a framework with utility for the future analyses of temporal clinical or molecular biomarkers.   5.2.2 Objectives 1.  To determine whether the repeated measurements of clinical features of LGD, including lesion presence, size, appearance, colour, texture, FV presentation and TB status, or a subset thereof, can predict lesion behaviour and eventual malignant progression.   2. To determine whether the above temporal clinicopathological patterns in LGD are associated with different molecular risk patterns.   3. To develop a clinically useful model to predict malignant progression of LGD that further stratifies and improves the risk prediction of the intermediate-risk group of a previously validated model, based on LOH and temporal clinicopathological features.   5.2.3 Hypotheses Hypothesis 1: Repeated measurements of lesion presence, lesion size, lesion appearance, lesion margin characteristics, TB status and FV presentation (or a subset of these features) will be predictive of malignant progression, and will be stronger predictors than measurements taken at one point in time.   Hypothesis 2: High-risk temporal clinical features are more likely to be associated with 102 intermediate or high-risk molecular risk patterns as compared to low-risk molecular risk patterns.   Hypothesis 3:  LGD with known LOH molecular risk patterns and a high-risk temporal clinical profile will be more likely to progress as compared to those with a low-risk temporal clinical profile.    5.3 Methods 5.3.1 Study Design  This study used a historical prospective cohort design involving participants of the OCPL study, described in section 2.1. Participants with a histologically confirmed hyperplasia, mild dysplasia, or moderate dysplasia were assessed via microsatellite analysis for LOH. The study database was used to collect information on demographics, risk habits and detailed measurements of clinical features (lesion presence, size, appearance, colour, margin characteristics, TB positivity and FV status), acquired through regular, standardized, repeated clinical assessment of lesions in each patient over time. Outcome, also in the database, was  progression to severe dysplasia, CIS or SCC. Study protocol and ethical approval were obtained from the University of British Columbia and BC Cancer Agency Research Ethics Board.   5.3.2 Setting and Participants Participants included in this analysis were enrolled in the OCPL study between January 1, 1997, and August 16, 2012. Eligibility criteria for this project required a histologically confirmed primary mild or moderate OED with no prior history of head and neck cancer. The availability of 103 a minimal amount of temporal clinicopathological data was required. Subjects with less than 2 follow-up visits were excluded. Participants were followed a minimum of 12 months, or to progression, whichever occurred first. Subjects with no progression and less than 12 months of longitudinal follow-up were excluded. Subjects who progressed within less than three months of index biopsy were also excluded. Inclusion also required that FFPE tissue blocks from index biopsies be available, and that viable DNA could be extracted for microsatellite analysis. A total of 306 subjects met the selection criteria and were included in the present analysis. Median follow-up time was 71.8 months (11.7 – 238.8 months). Of the 306 cases, 244 were reported in a previously published study that validated the LOH risk model. That study was not aimed at analyzing longitudinal repeated measures of clinicopathological patterns; it only examined clinical features measured at one point in time, in association with LOH and outcome.   5.3.3 Data Collection  5.3.3.1 Histological Diagnosis  After obtaining written informed consent, archival FFPE tissue specimen blocks from index biopsies were requested and obtained for each participant. Histologic diagnoses were reviewed and confirmed by the study pathologist (L.Z.), using diagnostic criteria established by the WHO. (113)   5.3.3.2 Molecular Assessment  LOH analysis was performed on index biopsies collected at baseline at 9p, 17p, and 4q using methods described in detail in section 2.3. Lesions were classified as low, intermediate or high risk of progression, using previously validated methods as described in section 2.3.4.(13) 104  5.3.3.3 Clinicopathological Variables A standardized initial questionnaire (Appendix A.2) was used to collect information from the participant on demographics, including age, gender, and ethnicity, and on variables associated with oral cancer risk, including tobacco use, alcohol consumption, and family history of oral cancer.  The participant’s medical history was reviewed and extra-oral and intra-oral examinations, including white light, FV, and TB examination were performed as detailed in section 2.6.1. White light, FV and TB digital photographs were taken of the lesions.  Clinical follow-up visits, following the same protocols as the initial visit, occurred every 6 months.   Clinicopathological data, including lesion site, size, texture, colour, appearance, lesion margin characteristics, as well as information on fluorescence visualization (FV) retention and toluidine blue (TB) positivity were collected as detailed in Section 2.6.1. Lesion presence was recorded as present or absent. Lesion site was marked onto an illustrated mouth map (Appendix A.3) and coded into defined anatomical categories (gingiva, buccal or vestibular mucosa, retromolar pad, hard palate, soft palate, tongue, floor or mouth). Lesion size was measured using a calibrated probe. Lesion length, width and thickness were recorded, in millimeters. Lesion texture was noted as ulcerative, smooth, velvety, nodular, verrucous, nodular, fissured, and/or other. Colour was recorded as white (leukoplakia), red (erythroplakia), or mixed white and red (erythroleukoplakia). Lesion appearance was documented as either homogeneous (same colour and texture throughout) or as non-homogeneous (colour and texture not uniform). Lesion margins were either diffuse (ill-defined margins) or discrete (well-defined margins). Lesions 105 were recorded as FV retained (retention of normal, green tissue autofluorescence), FV loss (appearing dark, with loss of tissue autofluorescence), or equivocal (slightly darker or uncertain). TB was recorded as positive (dark royal blue), negative (no stain taken up), or equivocal (weak or light uptake) (Appendix A.4).  5.3.3.4 Outcome The primary endpoint of this study was histologically confirmed progression to severe dysplasia CIS, or SCC (as of November 3, 2017), occurring at the same anatomical site as the index biopsy. The inclusion of severe dysplasia as the progression endpoint was based on our findings that without treatment, progression from severe dysplasia to CIS or SCC occurred in 32% of patients in 3 years; 59% in 5 years.(177)  5.3.4 Statistical Methods  P < .05 was chosen as statistical significance level. Statistical analyses were performed using Stata 14.2 (College Station, TX: StataCorp LLC). Recursive partitioning was performed using ‘rpart’ package in R 3.4.3 (Vienna, Austria: The R Foundation).   5.3.4.1 Sample Size  Sample size calculation was based on our lab’s previously published study data in which LGD in high-risk  (LOH on 9p, 4q and 17p) and intermediate-risk (LOH on 9p and 4q or 17p) categories were shown to have a 52.1-fold and 11.2-fold increase in risk of progression as compared to those in a low-risk category (retention on 9p)(13), as well as unpublished data which showed that LGD that were ever TB positive showed a 5.9-fold risk of progression as compared to those that 106 were never positive.(287) A sample size of 191 was required to detect a relatively small effect size of 0.2, with a significance level of 5% and 80% power on 2-tailed tests (G*Power® Version 3.1 software; Düsseldorf, Germany: University of Düsseldorf). Sample size calculation for multilevel logistic regression is a complex problem. Based on the method published by Peduzzi et al.(288), and assuming a proportion of 0.13 of positive cases in the population, a sample size of 231 was required for 3 covariates, and a sample size of 308 was required for 4 covariates.   5.3.4.2 Univariate Analysis  Univariate comparison of characteristics of patients with and without progression was done using the appropriate statistical test. Categorical variables were tested using the Chi-square Test or Fisher’s Exact Test when more than 20% of cells contained expected frequencies of <5. Quantitative variables were evaluated for normalcy. Normally distributed quantitative variables were tested using an independent samples T-test. Those that were not normally distributed were tested using the Mann-Whitney U Test. The threshold for significance was set at 0.05. All tests were 2-tailed. Dichotomization of categorical variables for comparisons was done based on clinical usefulness and significance of comparison results. Determinants of progression were identified using Cox proportional hazards regression. The proportional hazards assumption was assessed using the Schoenfeld residuals test. Missing values were imputed using chained equations with 100 iterations for the dichotomized categorical variables of family history of head and neck cancer, appearance, margin characteristics, TB, and FV, as well as for molecular risk group.   Comparison of time-variant characteristics of patients by progression was performed using a 107 pure longitudinal analysis, as well as an analysis based on cumulative counts. The rationale for examining both methods was that each informs different aspects of the analysis. Pure longitudinal analysis uses all measurements over a long follow-up period, and is valuable for examining the natural history of the disease. It describes what happened across all years and gives insight into what is important in the development of the disease.  However, we do not want patients to experience the natural history of the disease. The goal is to be able to offer early intervention. Information about what is happening in the first few visits may provide insight into which patients who might benefit from additional tests and/or early intervention.  The cumulative counts analysis provides valuable insight into early visits and is valuable for providing prognostication and informing early management.   5.3.4.2.1 Pure Longitudinal Analysis   Comparison of univariate time-variant characteristics of patients by progression group was performed using multilevel Cox regression on longitudinal data. To account for within and between patient changes, patient study ID was used to adjust for the cluster effect in estimating the standard errors and parameters. In this way, each person acts as their own control and adjusts for dependency of variables.(289)   5.3.4.2.2 Cumulative Counts Longitudinal Analysis   Time-variant characteristics of lesions were also counted in 5 visits at baseline, six months, one year, 2 years and 5 years from initial biopsy. These time intervals were chosen for use on later downstream prediction modeling based on potential clinical relevance for patient management and decision-making.  Patients who progressed and who did not progress were compared in 108 terms of zero, one, two, three, four or five times of presence of each time-variant characteristic. Each time-variant variable was compared using separate chi-squared or Fisher’s exact tests. The P-value for the overall trend was calculated using ordered logistic regression.  To eliminate the bi-directional risk which resulted from the counted time intervals analysis, the time-variant characteristics were then counted cumulatively in the 5 visit time points, as 0 times (never in 5 times), 1 time or more (ever in 5 times), 2 times or more, 3 times or more, 4 times or more, and all 5 time points. Comparisons between the two groups were made using chi-squared analysis. These associations were further assessed using univariate Cox regression.  5.3.4.3 Multivariable Analysis  Determinants of progression were identified using univariate Cox proportional hazards regression.  Variables with a P-value greater than 0.1 in univariate Cox regression were included in the multivariable Cox regression.   5.3.4.4 Risk Classification Models  Classification of risk of progression to cancer was assessed using recursive partitioning. Selection of variables for inclusion in the risk models was based on the significance of association with progression in multivariable Cox regression. The complexity parameter was set as 0.15 to reach optimal branching of the classification trees.  Hazard ratio of progression for risk categories of risk models were estimated using flexible parametric survival modelling to accommodate for departure from the proportionality of the hazards assumption among the risk categories. Wilcoxon trend test of equality of survivor 109 functions was used to assess non-overlapping risk categories from the risk model terminal nodes. If equality of survivor functions was observed, the terminal nodes of the risk model were combined. Survival analysis and 95% CI of the risk categories was assessed using Kaplan-Meier survival estimates. Accuracy of the models was assessed using Receiver Operating Characteristic (ROC) curves. Performance was measured using the area under the curve (AUC) summary statistic (C statistic) and its 95% CI.   Survival analysis was used because it can handle people entering and leaving at different times, and followed for varying durations. It can account for loss of participants and is used to summarize the results. Loss to follow-up was handled through censoring, which was deemed to be independent to outcome. Assumptions for survival analysis, including that censoring was deemed to be independent and no secular trends were detected.  5.4 Results  5.4.1 Participant Selection The participant selection process for the analysis is shown in Figure 5.1.  Of the 1106 participants enrolled in the OCPL study at the time of participant selection 522 participants were diagnosed with a histologically confirmed primary high-risk hyperplasia or mild or moderate OED with no prior history of head and neck cancer. The final eligibility for analysis was determined on December 22, 2016. A hyperplasia was considered high-risk if it had no known cause (e.g. reactive hyperplasia), was located at a high-risk site (tongue or floor of mouth), or had later progressed to a mild or moderate dysplasia.  A minimal amount two follow-up visits with temporal clinicopathological data available was required. The criteria of a minimum of two 110 follow-up visits was met by 494 participants. Of those, 491 participants had been followed for a minimum of 12 months of follow-up if there was no progression, or 3 months of follow-up if progression had occurred. Finally, extracted DNA for microsatellite analysis was available for 306 participants. A total of 306 subjects met the selection criteria and were included in the present analysis, with a median follow-up time of 71.8 months (11.7 – 238.8 months).  Figure 5.1 Participant Selection Process   5.4.2 Descriptive Data  Entry diagnoses from the 306 participants included 12 hyperplasia, 145 mild dysplasia, and 149 111 moderate dysplasia. Fifty-three (17.3%) progressed; 30 to severe dysplasia, 4 to CIS, and 19 to SCC. Median time to progression was 36.4 months (3.0 – 126.2 months). The first biopsy dates ranged from 1997 to 2012, and the last follow up dates ranged from 2000 to 2017.  The median length of follow-up was 71.8 months (11.7 – 238.8 months).   Table 5.1 shows the sociodemographic and risk habit characteristics of the 306 study participants. The median age at diagnosis was 59.4 years. In general, 150 (49%) of the participants were male, and 156 (51%) were female. Age at diagnosis and gender was not associated with progression. Eighty-four percent of the subjects were Caucasian. Being of Asian ethnicity was associated with progression (OR: 2.71; 95% CI, 1.03 – 7.14; P = 0.05).  Approximately one-third of the participants were never-smokers (34%). Although more ever-smokers had an OPML, a significant higher proportion of never-smokers underwent malignant progression (OR: 3.41; 95% CI, 1.85 – 6.26; P < 0.001). When considering all participants, smoking amount (pack-year calculation), was inversely associated with progression (P = 0.002). However, when taking only smokers into consideration, pack-year amount was not associated (P = 0.37). The amount of alcohol consumed per week and family history of head and neck cancer was not associated with progression (P = 0.56 and P = 0.39, respectively).   112  Table 5.1 Sociodemographic characteristics of patients by progression  ALL No Progression (%)* Progression  (%)* P value Odds Ratio (95% CI) Total 306 (100/100) 253 (83/100) 53 (17/100)   Age at diagnosis (n=306)   Median (IQR) Mean 59 (52-58) 60 59 (52-58) 60 59 (50-68) 59  0.62  Age Category (n=306)    <40 years 6 (100/2) 5 (83/2) 1 (17/2) 0.71 1  40 – 60 years 157 (100/52) 127 (81/50) 30 (19/57) 1.18 (0.13 – 10.49)  ≥60 years 142 (100/47) 120 (85/48) 22 (16/42) 0.92 (0.10 – 8.23)  Gender (n=306)  Female 156 (100/51) 128 (82/51) 28 (18/53) 0.77 1  Male 150 (100/49) 125 (83/49) 25 (17/47) 0.91 (0.51 – 1.65) Ethnicity (n=306)  Caucasian 257 (100/84) 217 (85/86) 40 (16/76) 0.05 1  Asian 21 (100/7) 14 (67/6) 7 (33/13) 2.71 (1.03 – 7.14)  South Asian 21 (100/7) 18 (86/7) 3 (14/6) 0.90 (0.25 – 3.21)  Other a 7 (100/2) 4 (57/2) 3 (43/6) 4.01 (0.88 – 18.87) Smoking Category b (n=306)  Never 105 (100/34) 74 (70/29) 31 (30/58) < 0.001 3.41 (1.85 – 6.27)  Ever  201 (100/66) 179 (89/71) 22 (11/42) 1 Smoking Amount (pack-year) (n=304) c 113  ALL No Progression (%)* Progression  (%)* P value Odds Ratio (95% CI)  Median (IQR) mean 9 (0-29) 17 11 (0-30) 19 0 (0-12) 10 0.002  Alcohol Consumed per week d (n=300)  Median (IQR) mean 4 (0-12) 8  4 (0-12) 8 3 (0-4) 8 0.56   Alcohol Category e (n=300)  None/Light 259 (100/86) 220 (85/85) 39 (15/92) 0.18 1  Heavy 41 (100/14) 37 (90/15) 4 (10/8) 0.45 (0.18 – 1.59) Family history of head and neck cancer f (n=290)   Negative  248 (100/86) 208 (84/86) 40 (16/82) 0.39 1  Positive  42 (100/14) 33 (79/14) 9 (21/18) 1.42 (0.63 – 3.91) Months follow up g  Median (IQR) mean  72 (46-107) 83 70 (46-104) 79 83 (51-117) 89 0.17  * (row% / column%)  a 3 Hispanic, 2 Black, 1 Mixed, 2 Other. b Never: smoked 0-100 cigarettes in life time; Ever: smoked >100 cigarettes in lifetime.(259) c Pack-year data missing for 2 participants (with no progression). d Average equivalent units per week (8 oz. beer, 4 oz. wine, or 1 oz. spirits).(275) e More than 14 units in females and 20 in males.(274)  f Diagnosis of head and neck cancer, excluding skin cancer, of a first or second degree birth-related relative.  g Last follow-up or progression, whichever occurred first.   114 5.4.3 Outcome Data    5.4.3.1 Time-invariant Clinicopathological Characteristics  Time-invariant clinicopathological characteristics of the OPML were compared based on progression (Table 5.2). A diagnosis of moderate dysplasia was more than twice as likely to undergo malignant progression as compared to an OMPL with a lesser diagnosis (hyperplasia or mild dysplasia) (OR: 2.37; 95% CI, 1.28 – 4.41; P = 0.005). Similarly, lesion presence in high-risk site (floor of mouth or tongue) possessed a greater than two-fold risk of progression (OR: 2.76; 95% CI, 1.38 – 5.50; P = 0.003). LOH in 9p, 17p, or 4q was significantly related with progression in univariate analysis (P < 0.001, P = 0.002, P < 0.001, respectively). The LOH risk progression model that was validated by Zhang et al. (13) in 2012, was also significantly related with progression in this study. Compared with the molecular low-risk group, the intermediate- and high-risk groups showed a 7.4-fold and 46.8-fold increase in risk (P < 0.001).   115 Table 5.2 Time-invariant clinicopathological characteristics by progression   ALL No Progression (%)* Progression  (%)* P value Odds Ratio (95% CI) Total 306 (100/100) 253 (83/100) 53 (17/100)   Biopsy Diagnosis (n=306)  Hyperplasia 12 (100/4) 10 (83/4) 2 (17/4) 0.02 1  Mild Dysplasia 145 (100/47) 129 (89/51) 16 (11/30) 0.62 (0.13 – 3.09)  Moderate Dysplasia 149 (100/48) 114 (77/45) 35 (24/66) 1.54 (0.32 – 7.34) Biopsy Diagnosis (n=306)  Hyperplasia + Mild Dysplasia 157 (100/51) 139 (88/55) 18 (12/34) 0.005 1  Moderate Dysplasia 149 (100/48) 114 (77/45) 35 (24/66) 2.37 (1.28 – 4.41) Site a (n=306)  Low-risk site 125 (100/41)  113 (90/45) 12 (10/23) 0.003 1  High-risk site 181 (100/59) 140 (77/55) 41 (23/77) 2.76 (1.38 – 5.50) 9p LOH b (n= 281)  9p Retained 121 (100/43) 117 (97/50) 4 (3/8) < 0.001 1  9p LOH 160 (100/57) 117 (73/50) 43 (27/92) 10.75 (3.74 – 30.91) 17p LOH b (n=287)  17 Retained 185 (100/65) 163 (88/68) 22 (12/45) 0.002 1  17p LOH 102(100/36) 75 (73/32) 27 (27/55) 2.67 (1.43 – 4.99) 116  ALL No Progression (%)* Progression  (%)* P value Odds Ratio (95% CI) 4q LOH b (n=248)  4q Retained 174 (100/70) 154 (89/75) 20 (11/46) < 0.001 1  4q LOH 74 (100/30) 51 (69/25) 23(31/54) 3.47 (1.76 – 6.84) Molecular Risk Category based on a previously validated model c (n=266)  Low-risk 121 (100/45) 117 (97/53) 4 (3/9) < 0.001 1  Intermediate-risk 119 (100/45) 95 (80/43) 24 (20/55) 7.39 (2.48 – 22.04)  High-risk 26 (100/10) 10 (39/5) 16 (62/36) 46.8 (13.12 – 166.95)  * (row% / column%) a High-risk site: tongue or floor of mouth; Low-risk site:  all other sites; b LOH: loss of heterozygosity versus retention (non-informative not included); c Risk category based on risk model by Zhang et al. (2012). Low risk: 9pRet, Intermediate risk 9pLOH or 9pL0H+4qLOH or 9pLOH+17pLOH, High risk: 9pLOH+4qLOH+17pLOH (non-informative cases not included); 117 5.4.3.2 Time-variant Clinicopathological Characteristics Time-variant clinicopathological features were compared by both a pure longitudinal analysis method and by a cumulative counts analysis.   5.4.3.2.1 Pure – Longitudinal Analysis  5.4.3 2.1.1 Univariate Analysis Lesion presence (HR: 19.17; 95% CI, 2.64 – 138.96; P < 0.001), area (P < 0.001), non-homogeneous appearance (HR: 3.64; 95% CI, 1.79 – 7.40; P < 0.001), mixed or red colour (HR: 3.71; 95% CI, 1.80 – 7.65; P < .001), texture other than smooth (HR: 2.90; 95% CI, 2.90; P < 0.001), TB+ (HR: 11.37; 95% CI, 6.25 – 20.70; P < 0.001) and FV+ (HR: 7.78; 95% CI, 3.02 – 20.03; P < 0.001) were the time varying variables with a significant association with progression in univariate analysis (Table 5.3). Longitudinal lesion margin characteristics were not significantly associated with progression. HR is the ratio of the hazard rates between the group that experience malignant progression and the group that did not. It measures the effect of a variable on malignant progression over time. For example, at any particular time, 3.6 times as many patients with a lesion with a non-homogeneous appearance will have experienced malignant progression as compared to those with homogeneous lesions at that same period of time.   5.4.3.2.1.2 Multivariable Analysis Time-invariant and time-variant variables with a P-value of ≤ 0.1were moved forward into multivariable analysis. After controlling for all independent variables, multivariable Cox regression analysis showed that the previously validated molecular risk categories(13) (LOH 118 Intermediate Risk adjusted HR: 4.55, 95% CI: 0.97 – 22.47, P = 0.05; LOH High Risk adjusted HR: 19.15, 95% CI: 9.94 – 93.12, P < 0.001) were significantly associated with progression (Table 5.4). With respect to time-variant variables, only temporal TB status was significantly associated with progression (adjusted HR: 3.25, 95% CI: 1.24 – 8.51, P = 0.02). Lesion area (mm2) had an association of very small size in univariate regression (HR: 1.0008; 95%CI, 1.0003-1.001; P = 0.002). In multivariable regression, this small association was statistically insignificant (P = 0.66). Other configurations of lesion size (area in 200 mm2 and area > 400 mm2 (binary) were tested in multivariable models and failed to demonstrate significant relation with outcome. Appearance is considered to be a combined depiction of colour and texture (homogeneous appearance being uniform in colour and texture; non-homogeneous having variation in colour or texture, or both). Hence, the model was tested with the variables appearance, colour and texture, and without the variables of colour or texture. Excluding colour and texture from the model did not affect the other variables. Therefore, further pure longitudinal analyses included only the variable appearance. 119 Table 5.3 Time-variant clinicopathological characteristics by progression – Pure longitudinal analysis£  ALL† No Progression† (%)* Progression†  (%)* P value Hazard Ratio (95% CI) Total 306 (100/100) 253 (83/100) 53 (17/100)   Lesion Area (mm2) (n=2266)   Median (IQR)  189 (72-432) 180 (60-420)  273 (108-540) < 0.001  Lesion present (n=3339)    Not present 759 (100/22.7) 747 (98.4/25.6) 12 (1.6/2.8) < 0.001  1  Lesion present 2580 (100/77.3) 2169 (84.1/74.4) 411 (15.1/94.2) 19.17 (2.64 – 138.96) Lesion appearance (n=2242)  Homogeneous  1326 (100/59.1) 1172 (88.4/61.5) 154 (11.6/45.7) < 0.001 1  Non-homogeneous 916 (100/40.9) 733 (79.9/38.5) 183 (20.3/54.3) 3.64 (1.79 – 7.40) Lesion margins (n=2097)  Discrete (well-defined)  723 (100/34.6) 626 (86.6/34.6) 97 (13.4/34.6) 0.93 1  Diffuse (Ill-defined) 1374 (100/65.) 1188 (86.5/65.5)  186 (13.5/65.7) 1.15 (0.54 – 2.44) Lesion Colour (n=2214)  All white 1382 (100/62.4) 1193 (86.3/63.3)  189 (13.7/57.4) 0.04 1  Mixed or Red 832(100/37.6) 692 (83.2/36.7) 140 (16.8/42.6) 3.71 (1.80 – 7.65) Lesion Texture (n=2198)  Smooth 1423 (100/64.7) 1268 (89.1/67.9) 155 (10.9/46.6) < 0.001 1  Other than smooth  775 (100/35.3) 599 (77.3/32.1) 176 (22.7/53.2) 2.90 (1.45 – 5.81) 120  ALL† No Progression† (%)* Progression†  (%)* P value Hazard Ratio (95% CI) Toluidine Blue (n=3018)  Negative 2793 (100/92.5) 2508 (89.8/94.7) 285 (10.2/77.0) < 0.001 1  Positive 225 (100/7.5) 140 (62.2/5.3) 85 (37.8/23.0)  11.37 (6.25 – 20.70) Florescence Visualization (n=2553)  Retention 1317 (100/51.6) 1241 (94.2/54.3) 76 (5.8/28.6) < 0.001 1  Loss of fluorescence 1236 (100/48.4)  1046 (84.6/45.7) 190 (15.4/71.4) 7.78 (3.02 – 20.03) £ Pure longitudinal analysis uses all measurements over a long follow-up period, and is valuable for examining the natural history of the disease. It describes what happened across all years and gives insight into what is important in the development of the disease. The cumulative counts analysis provides valuable insight into the first few visits and how to differentiate between patients who might benefit from additional tests and/or early intervention.  † Numbers reported are raw numbers, however P values and hazard ratios are adjusted for clustering and imputation. Multilevel Cox regression was used on longitudinal data. Patient study ID was used to adjust for the cluster effect and adjust for dependency of the variables.   * (row% / column%)    121 Table 5.4 Determinants of progression in Cox proportional hazards regression – Pure longitudinal analysis  Determinants Univariate Cox Multivariable Cox Hazard Ratio (95%CI) P value Hazard Ratio (95%CI) P value Non-Caucasian ethnicity 1.84 (0.98 – 3.43) 0.06 1.05 (0.38 – 2.90)  0.93 Never smoker 2.74 (1.59 – 4.74) < 0.001 2.18 (0.80 – 5.93) 0.13 Heavy Alcohol Consumption  0.48 (0.12 – 1.45) 0.18 1.09 (0.28 – 4.23)  0.90 Site FOMa or tongue 2.73 (1.43 – 5.21) 0.002 2.07 (0.80 – 5.33)  0.13 LOH Intermediate Risk † 2.29 (1.31 – 3.98) 0.004   4.66 (0.97 – 22.47) 0.055 LOH High Risk †  8.34 (3.06 – 22.76) < 0.001 19.15 (3.94 – 93.12)  < 0.001  Lesion present  19.17 (2.65 – 138.96) 0.003 * * Lesion area (mm2) b  1.0008 (1.0003 – 1.001) 0.002 1.000 (0.998 – 1.0001)  0.60 Non-homogeneous Appearance 3.64 (1.79 – 7.40) < 0.001 1.38 (0.48 – 3.99)  0.55 Toluidine Blue positive 11.38 (6.25 – 20.70) < 0.001 3.25 (1.24 – 8.51)  0.02 Loss of Fluorescence Visualization  7.78 (3.02 – 20.03) < 0.001 3.23 (0.86 – 12.21)  0.08 Diagnosis of Moderate Dysplasia or VH 2.37 (1.22 – 4.68) .001 0.79 (0.30 – 2.08) 0.63 a Floor of mouth.  b Note:  other configurations of the variable “lesion area” were tested in multivariable models to assess if they demonstrate significant relation with the outcome.  † LOH Risk category based on risk model by Zhang et al. (2012 (13)). Intermediate risk = 9pLOH or 9pL0H+4qLOH or 9pLOH+17pLOH; High risk = 9pLOH+4qLOH+17pLOH (non-informative cases not included); *Omitted by model because of inter-dependency of independent variables.   122 5.4.3.2.2 Cumulative Counts Analysis  5.4.3.2.2.1   Univariate Analysis For the univariate analysis, time-variant characteristics of lesions were counted for each patient at five time points: baseline, six months, one year, two years and five years from initial biopsy. These time intervals were chosen based on potential clinical relevance for patient management and decision-making. Patients who progressed and who did not progress were compared in terms each time-variant characteristic being present zero times (never in five time points), one time or more (ever in 5 time points), two time points or more, three time points or more, four time points or more, and at all five time points. Details on the statistical tests applied are reported in section 5.3.4.2.2.  In the cumulative counts analysis, lesion presence in 3 or more of the time points (HR: 3.80; 95% CI, 1.13 – 12.71; P = 0.02), a non-homogeneous appearance in one time point or more (HR: 4.07; 95% CI, 1.90 – 8.69; P < 0.001), a mixed or red colour in two time points or more (HR: 1.82; 95% CI, 1.01 – 3.33; P = 0.04), lesion texture other than smooth in two time points or more (HR: 2.15; 95% CI, 1.18 – 3.94; P = 0.01), or TB positivity in two time points or more (HR: 4.47; 95% CI, 1.90 – 10.51; P < 0.001) were significantly associated with progression (Table 5.5). Margin characteristics and FV positive cumulative times were not associated with progression.   The cumulative counts analysis was performed to provide insight into early visits and is valuable for providing prognostication and informing early management. However, a limitation to this analysis is that not all participants had data for all time periods. Although the median length of 123 follow-up was 71.8 months (11.7 – 238.8 months), not all participants had data for the time periods of 24 or 60 months. Having a lesion characteristic in four or five of the time periods or more may not have achieved significance in some of the characteristics due to low numbers and low power.   5.4.3.2.2.2   Multivariable Analysis After controlling for all other variables, the variables with adjusted significant relation with progression were never smoking status (HR: 2.77; 95% CI, 1.43 – 5.35; P = 0.002), high-risk site (tongue or FOM) (HR: 2.12; 95% CI, 1.02 – 4.40; P = 0.04), LOH intermediate-risk group (HR: 6.44; 95% CI, 2.21 – 18.80; P = 0.001), and LOH high-risk group (HR: 22.32, 95% CI, 7.02 – 70.96; P < 0.001) (Table 5.6). Additionally, TB positivity at two or more time points (out of time points at baseline, six months, 12 months, 24 months, and 60 months from initial biopsy) (HR: 3.02; 95% CI, 1.26 – 7.21; P = 0.01) or a non-homogeneous appearance at one or more time points (HR: 2.39, 95% CI 0.99 – 5.81; P = 0.05) were associated with progression. Other time-variant clinicopathological features did not demonstrate significant association with progression when adjusted for the presence of all other variables  124 Table 5.5 Time-variant clinicopathological characteristics by progression – Cumulative counts analysis£   ALL No Progression (%)* Progression  (%)* P value Hazard Ratio (95% CI) Total 306 (100/100) 253 (83/100) 53 (17/100)   Lesion Presence (number of time periods§ or more (cumulative))    0 times    0 (100 / 0) 0 (- / 0) 0 (- / 0) - -  1 time or more 306 (100/100) 253 (82.7 / 100) 53 (17.3 / 100) † -  2 times or more 283 (100 / 92.5) 230 (81.3 / 90.9) 53 (18.7 / 100) † -  3 times or more 256 (100 / 83.7) 206 (80.5 / 81.4) 50 (19.5 / 94.3) 0.02 3.80 (1.13 – 12.71)  4 times or more 214 (100 / 69.9) 175 (81.8 / 69.2) 39 (18.2 / 73.6) 0.52 1.24 (0.63 – 2.41)  5 times or more  102 (100 / 33.3) 86 (84.3 / 34.0) 16 (15.7 / 30.2) 0.59 0.83 (0.44 – 1.59) Non-homogeneous Appearance (number of time periods§ or more (cumulative))  0 times    124 (100 / 40.5) 115 (92.7 / 45.5) 9 (7.3 / 17.0) < 0.001 -  1 time or more 182 (100 / 59.5) 138 (75.8 / 54.5) 44 (24.2 / 83.0) < 0.001 4.07 (1.90 – 8.69)  2 times or more 110 (100 / 35.9) 86 (78.2 / 34.0) 24 (21.8 / 45.3) 0.11 1.60 (0.88 – 2.92)  3 times or more 71 (100 / 23.2) 59 (83.1 / 23.3) 12 (16.9 / 22.6) 0.91 0.96 (0.47 – 1.95)  4 times or more 20 (100 / 6.5) 14 (70.0 / 5.5) 6 (30.0 / 11.3) 0.12 2.17 (0.79 – 5.96)  5 times or more 2 (100 / 0.7) 2 (100 / 0.8) 0 (0 / 0) 0.52 - Ill-defined Margins (number of time periods§ or more (cumulative))  0 times    101 (100 / 33.0) 85 (84.2 / 33.6) 16 (15.8 / 30.2) 0.63 -  1 time or more 205 (100 / 66.9) 168 (81.9 / 66.4) 37 (18.1 / 69.8) 0.63 1.17 (0.61 – 2.22)  2 times or more 145 (100 / 47.4) 121 (83.5 / 47.8) 24 (16.5 / 45.3) 0.73 0.90 (0.49 – 1.63) 125  ALL No Progression (%)* Progression  (%)* P value Hazard Ratio (95% CI)  3 times or more 101 (100 / 33.1) 86 (85.2 /34.0)  15 (14.8 / 28.3) 0.42 0.76 (0.39 – 1.47)  4 times or more 46 (100 / 15.0) 42 (91.3 / 16.6) 4 (8.7 / 7.5) 0.19 0.41 (0.14 – 1.19)  5 times or more 6 (100 / 2.0) 6 (100 / 2.4) 0 (0 / 0) 0.26 - Colour Mixed or Red (number of time periods§ or more (cumulative))  0 times  142 (100 / 46.4) 123 (86.6 / 48.6) 19 (13.4 / 35.8) 0.09 -  1 time or more 164 (100 / 53.6) 130 (79.3 / 51.4) .04 0.09 1.69 (0.91 – 3.12)  2 times or more 103 (100 / 33.7) 79 (76.7 / 31.2) 24 (23.3 / 45.3) 0.04 1.82 (1.00 – 3.33)  3 times or more 66 (100 / 21.6) 53 (80.3 / 20.9) 13 (19.7 / 24.5) 0.56 1.22 (0.61 – 2.45)  4 times or more 22 (100 / 7.2) 17 (77.3 / 6.7) 5 (22.7 / 9.4) .048 1.44 (0.50 – 4.10)  5 times or more 2 (100 / 0.7) 2 (100 / 0.8) 0 (0 / 0) 0.52 - Texture not smooth (number of time periods§ or more (cumulative))  0 times 155 (100 / 50.7) 142 (91.6 / 56.1) 13 (8.4 / 24.5) < 0.001 -  1 time or more 151 (100 / 49.3) 111 (73.5 / 43.9) 40 (26.5 / 75.5) <. 0001 3.93 (2.00 – 7.71)  2 times or more 99 (100 / 32.3) 74 (74.8 / 29.2) 25 (25.3 / 47.2) 0.01 2.15 (1.18 – 3.94)  3 times or more 59 (100 / 19.3) 48 (81.4 / 19.0) 11 (18.6 / 20.8) 0.76 1.11 (0.53 – 2.33)  4 times or more 18 (100 / 5.9) 13 (72.2 / 5.1) 5 (27.8 / 9.4) 0.22 1.92 (0.65 – 5.64)  5 times or more 2 (100 / 0.7) 2 (100 / 0.8) 0 (0 / 0) 0.52 - Toluidine Blue Positive (number of time periods§ or more (cumulative))  0 times  228 (100 / 74.5) 208 (91.2 / 82.2) 20 (8.8 / 37.7) < 0.001 -  1 time or more 78 (100 / 25.5) 45 (57.7 / 17.8) 33 (42.3 / 62.3) < 0.001 7.62 (4.02 – 14.49) 126  ALL No Progression (%)* Progression  (%)* P value Hazard Ratio (95% CI)  2 times or more 25 (100 / 8.2) 14 (56.0 / 5.5) 11 (44.0 / 20.8) < 0.001 4.47 (1.90 – 10.51)  3 times or more 4 (100 / 1.3) 3 (75.0 / 1.2) 1 (25.0 / 1.9) 0.68 1.60 (0.16 – 15.71)  4 times or more 1 (100 / 0.3) 1 (100 / 0.4) 0 (0 / 0) 0.65 -  5 times or more 1 (100 / 0.3) 1 (100 / 0.4) 0 (0 / 0) 0.65 - Loss of Fluorescence Visualization (number of time periods§ or more (cumulative))  0 times 100 (100 / 32.7) 85 (85.0 / 33.6) 15 (15.0 / 28.3) 0.46 -  1 time or more 206 (100 / 67.3) 168 (81.5 / 66.4) 38 (18.5 / 71.7) 0.46 1.28 (0.66 – 2.46)  2 times or more 158 (100 / 51.6) 129 (81.7 / 51.0) 29 (18.3 / 54.7) 0.62 1.16 (0.64 – 2.10)  3 times or more 92 (100 / 30.1) 77 (83.7 / 30.4) 15 (16.3 / 28.3) 0.49 0.90 (0.46 – 1.73)  4 times or more 33 (100 / 10.8) 30 (90.9 / 12.6) 3 (7.1 / 5.7) 0.18. 0.44 (0.13 – 1.51)  5 times or more 5 (100 / 1.6) 4 (80.0 / 1.6) 1 (20.0 / 1.9) 0.87 1.19 (0.13 – 10.92) * (row% / column%). £ Pure longitudinal analysis uses all measurements over a long follow-up period, and is valuable for examining the natural history of the disease. It describes what happened across all years and gives insight into what is important in the development of the disease. The cumulative counts analysis provides valuable insight into the first few visits and how to differentiate between patients who might benefit from additional tests and/or early intervention.   § time points = baseline, 6 months, 1 year, 2 years and 5 years from initial biopsy; chosen based on clinical relevance. † Perfect predictor (variables perfectly predict outcome).   127 Table 5.6 Determinants of progression in Cox proportional hazards regression – Cumulative counts analysis  Determinants Univariate Cox Multivariable Cox Hazard Ratio (95%CI) P value Hazard Ratio (95%CI) P value Non-Caucasian ethnicity 1.84 (0.98 – 3.43) 0.06 1.67 (0.81 – 3.45) 0.16 Never smoker 2.74 (1.59 – 4.74) < 0.001 2.77 (1.43 – 5.35) 0.002 Site FOMa or tongue 2.73 (1.43 – 5.21) 0.002 2.12 (1.02 – 4.40) 0.04 LOH Intermediate Risk † 7.30 (2.53 – 21.08) < 0.001  6.44 (2.21 – 18.80) 0.001 LOH High Risk †  33.98 (11.22 – 102.90) < 0.001  22.32 (7.02 – 70.96) < 0.001 Lesion present 3 times or more  2.97 (0.92 – 9.53) 0.06 0.75 (0.21 – 2.66) 0.65 Non-homogeneous Appearance 1 time or more 3.58 (1.74 – 7.33) < 0.001 2.39 (0.98 – 5.81) 0.054 Mixed Red and White or Red Colour 2 times or more  1.54 (0.89 – 2.65) 0.11 0.76 (0.37 – 1.53) 0.43 Texture not smooth 2 times or more 1.84 (1.07 – 3.15) 0.02 1.30 (0.66 – 2.57) 0.45 Toluidine Blue positive 2 times or more  3.61 (1.85 – 7.02) < 0.001 3.01 (1.26 – 7.21) 0.01 a Floor of mouth. † LOH Risk category based on risk model by Zhang et al. (13). Intermediate risk = 9pLOH or 9pL0H+4qLOH or 9pLOH+17pLOH; High risk = 9pLOH+4qLOH+17pLOH (non-informative cases not included).  128 5.4.3.3 Comparison of Baseline and Repeated Clinicopathological Measurements in the Prediction of Outcome Univariate Cox regression analysis showed that repeated temporal measurements of lesion presence, lesion size, lesion appearance, TB status and FV presentation will be predictive of malignant progression, and were stronger predictors than measurements taken at one point in time only (Table 5.7). Lesion margin characteristics did not demonstrate significant association with progression with baseline or with repeated temporal measurements. When assessing the significant variables with multivariable regression, and controlling for all other variables, temporal TB remained a significant predictor (HR: 3.25; 95% CI, 1.24 – 8.51; P = 0.02).     129 Table 5.7 Risk prediction of baseline measurements compared to repeated measurements in clinicopathological features Determinant Univariate Analysis Baseline Measurement Univariate Analysis Temporal Measurements Hazard Ratio  (95% CI) P value Hazard Ratio  (95% CI) P value Lesion size (mm2) 1.001 (1.000-1.002) 0.002 1.0008 (1.0003 – 1.001) 0.002 Lesion present * * 19.17 (2.64 – 138.96) < 0.001 Non-homogeneous appearance  1.78 (1.02 – 3.08) 0.04 3.64 (1.79 – 7.40) < 0.001 Ill-defined margins 1.26 (0.62 – 2.59) 0.53 1.15 (0.54 – 2.44) 0.93 Toluidine blue positive 2.22 (1.26 – 3.90) 0.006 11.37 (6.25 – 20.70) < 0.001 FVa loss 2.73 (1.17 – 6.37) 0.02 7.78 (3.02 – 20.03) < 0.001   Multivariable Analysis  Baseline Measurement Multivariable Analysis Temporal Measurements Hazard Ratio  (95% CI) P value Hazard Ratio  (95% CI) P value Lesion size (mm2) 1.002 (1.000-1.003) 0.008 1.000 (0.998 – 1.001) 0.60 Lesion present § § § § Non-homogeneous appearance  1.14 (0.43 – 3.00) 0.80 1.38 (0.48 – 3.99 0.55 Toluidine blue positive 1.40 (0.62 -3.18) 0.42 3.25 (1.24 – 8.51) 0.02 FVa loss 2.38 (0.97 – 5.79) 0.06 3.23 (0.86 – 12.21) 0.08 * Perfect predictor. a FV = fluorescence visualization. §Omitted by model because of inter-dependency of independent variables.  5.4.4 Correlation of Time-Variant Clinicopathological Characteristics and LOH Temporal patterns in high-risk clinicopathological features, including lesion presence, lesion area ≥ 200 mm2, non-homogeneous appearance, ill-defined margins, mixed or red colour, texture other than smooth, TB positivity and loss of FV were significantly associated with intermediate 130 and high-risk molecular risk patterns as compared to low-risk molecular risk patterns (Table 5.8). Strength of the association of the temporal variables was also assessed in multivariable analysis. After controlling for all variables, lesion size of greater than 2 cm2 (HR: 1.81; 95% CI, 1.60 – 2.05; P < 0.001), ill-defined margins (HR: 27.17; 95% CI, 19.31- 38.23; P < 0.001), mixed or red lesion colour (HR: 4.68, 95% CI, 4.12 – 5.31; P < 0.001), texture other than smooth (HR: 6.23; 95% CI, 5.55 – 7.02; P <.001) and TB positivity (HR: 2.70, 95% CI, 2.45 – 2.99; P < 0.001), were significantly associated with a high-risk LOH pattern.     131 Table 5.8 Time-variant clinicopathological characteristics by molecular risk category Determinant Univariate Analysis LOH† Intermediate Risk Compared to Low Risk LOH† High Risk Compared to Low Risk OR (95% CI) P value OR (95% CI) P value Lesion presence 1.43 (1.41-146) < 0.001 3.67 (3.49-3.86) < 0.001 Area ≥ 200 mm2  1.24 (1.22-1.25) < 0.001 3.15 (3.09-3.19) < 0.001 Appearance non-homogeneous 1.21 (1.18-1.25) < 0.001 3.07 (2.91-3.24) < 0.001 Ill-defined margins  1.98 (1.92-2.04) < 0.001 3.27 (3.04-3.51) < 0.001 Colour mixed or red 1.38 (1.34-1.42) < 0.001 3.33 (3.16-3.52) < 0.001 Texture other than smooth 0.69 (0.68-0.72) < 0.001 2.11 (2.00-2.23) < 0.001 Toluidine blue positive 1.54 (1.48-1.61) < 0.001 7.01 (6.62-7.42) < 0.001 Loss of fluorescence visualization 1.55 (1.52-1.59) < 0.001 5.93 (5.55-6.33) < .001  Determinant Multivariable Analysis LOH† Intermediate Risk Compared to Low Risk LOH† High Risk Compared to Low Risk OR (95% CI) P value OR (95% CI) P value Lesion presence 1 (omitted) - 1 (omitted)  - Area ≥ 200 mm2  0.52 (0.50 -0.55) < 0.001 1.81 (1.60 – 2.05) < 0.001 Appearance non-homogeneous 1.39 (1.32 – 1.47) < 0.001 0.90 (0.77 – 1.06) 0.20 Ill-defined margins  1.13 (1.09 – 1.19) < 0.001 27.17 (19.31 -38.23) < 0.001 Colour mixed or red 0.86 (0.82 – 0.91) < 0.001 4.68 (4.12 – 5.31) < 0.001 Texture other than smooth 0.51 (0.49 – 0.54) < 0.001 6.23 (5.55 – 7.02) < 0.001 TB positive 1.66 (1.56 – 1.78)  < 0.001 2.70 (2.45 – 2.99) < 0.001 Loss of fluorescence visualization 1.00 (0.96 – 1.05) 0.84 0.62 (0.55 – 0.71) < 0.001 † LOH = loss of heterozygosity; LOH Risk category based on risk model by Zhang et al. (13). Intermediate risk = 9pLOH or 9pL0H+4qLOH or 9pLOH+17pLOH; High risk = 9pLOH+4qLOH+17pLOH (non-informative cases not included).  5.4.5 Risk Classification Modeling  To develop a model to predict malignant progression of LGD that further stratifies and improves 132 the risk prediction of the intermediate-risk group of the previously validated LOH model developed by Zhang et al.(13) in 2012, recursive partitioning analysis was used to construct a new classification model that used LOH risk category and the significant time-invariant and time-variant variables from the temporal multivariable analyses. Two models were developed. The first incorporated significant variables as determined by the multivariable pure longitudinal analysis; the second used those variables established as significant by the multivariable cumulative counts analysis method.  The rationale for examining both methods was to inform different aspects of the analysis.   5.4.5.1 Pure Longitudinal Analysis Pure longitudinal analysis uses all measurements over the follow-up period. It is valuable for examining the natural history of the disease. It gives valuable insight into what is important in the natural history of the disease.   The input variables into the model were validated LOH risk category, TB, FV, and lesion site category. With the complexity parameter set at 0.10, the model retained validated LOH risk category, TB and high-risk site as covariates. The model produced five terminal nodes. The validated LOH risk category was the first most significant split. For cases showing intermediate-risk LOH, a second split involved TB status, whereas among TB negative cases, a third split involved lesion site (Figure 5.2). Based on the output of this analysis, study patients were placed into three categories with respect to the risk of progression. Terminal nodes one and two, and terminal nodes four and five were combined to form three terminal nodes.  The new risk level one included cases that were LOH low-risk only, or LOH intermediate-risk and TB negative and 133 low-risk site (59.4% of informative cases).  The new risk level two included those cases that were LOH intermediate-risk and TB positive and high-risk site (25.6% of informative cases). Risk level three included those cases that were LOH intermediate-risk and TB positive, or LOH high-risk only (13.5% of informative cases). The proportion of progression for new risk categories one, two and three was 5.1%, 20.1% and 58.3%, respectively.   Performance of the model was assessed in Table 5.9. Area Under the Curve (AUC) of Receiver Operating Characteristic (ROC) curve and its 95% CI were used to assess the prediction accuracy of the risk model (AUC = 0.88; 95% CI, 0.81-0.94). Time to progression curves were examined based on the new risk classification categories. Compared with low-risk lesions, the HR for intermediate-risk lesions was 3.29 (95% CI, 1.10 – 9.82); P = 0.03) and for high-risk 36.30 (95% CI, 14.45 – 91.2; P < 0.001) (Figure 5.3). The 3-year and 5-year probability of progression were 1.4% and 3.1% for the new low-risk category, 6.1% and 13.9% for the new intermediate-risk category, and 40.1% and 53.8% for the new high-risk category. (P < 0.001) (Table 5.10).   134 Figure 5.2 Pure longitudinal analysis risk stratification model    Table 5.9 Performance of the pure longitudinal analysis model Input Variables New Risk Category Hazard ratio (95% CI) P value AUC (95%CI) Validated LOH Risk Category, Temporal TB, Site. 1 Low 1  0.88 (0.81 – 0.94) 2 Intermediate 3.29 (1.10 – 9.82) 0.03 3 High 36.30 (14.45 – 91.20) < 0.001 † LOH = loss of heterozygosity.  § TB = toluidine blue.     135 Figure 5.3 Pure longitudinal analysis Kaplan-Meier survival estimates   Table 5.10 Probability of progression - pure longitudinal analysis risk categories  † Progression to severe dysplasia, carcinoma in-situ, or squamous cell carcinoma. § Months to last follow up or progression, whichever occurred first.  5.4.5.2 Cumulative Counts Analysis Although pure longitudinal analysis offers us some insight in the natural history of the disease as  Risk Category 1 Risk Category 2 Risk Category 3 P value Total (n=262) 158 (60.3%) 68 (26.0%)  36 (13.7%)  Months to progression†   Median (range)§ 61.0 (19.3 – 115.8) 50.2 (3.3 – 126.2) 27.4 (4.3 – 78.0) 0.02 Probability of Progression†    3-year (95% CI) 1.4 (0.4 – 2.4)  6.1 (2.7 – 9.5) 40.1 (31.8 – 48.4) < 0.001  5-year (95% CI) 3.1 (1.6 – 4.6) 13.9 (9.3 – 18.5) 53.8 (44.9 – 62.7) 136 it goes untreated, we do not want patients to go on to experience the natural history of the disease. The goal is to be able to offer early intervention. Information about what is happening in the first few visits may provide valuable insight into how to differentiate between patients who might benefit from additional tests and/or early intervention.  The cumulative counts analysis provides valuable insight into early visits and is valuable for providing prognostication and informing early management.   In an effort to establish a clinically useful model to predict malignant progression of LGD and further stratify the intermediate-risk group of the previously validated 2012 LOH model,(13), the significant variables of validated LOH risk category, TB positivity within two or more time periods, smoking status, lesion site category, non-homogeneous appearance more than once were entered into recursive partitioning. With the complexity parameter set at 0.0005, the model retained validated LOH risk category, TB positivity within two or more time periods, smoking status, and appearance as measured over one or more time periods, as covariates, and produced five terminal nodes. Validated LOH risk category was the first most significant split. For cases showing intermediate-risk LOH, a second split involved smoking status, which moved to a third split involving appearance over one or more time periods. Ever-smokers with LGD with a non-homogeneous appearance in one or more time periods moved to a fourth split based on TB positivity within two or more time periods (Figure 5.4). Based on this output of this analysis, terminal nodes were combined to form four categories with respect to the risk of progression. This new risk model keeps the previously validated LOH low-risk and LOH high-risk as the lowest and highest risk categories (45.5% and 9.8% of informative cases, respectively), and separates the LOH intermediate-risk category into new low-intermediate (33.5% of informative 137 cases) and high-intermediate (11.3% of informative cases) risk categories. The proportion of progression for the risk in the cumulative counts model was 3.3%, 14.6%, 36.7%, and 61.5%, respectively.   Performance of the model was assessed in Table 5.11. Area Under the Curve (AUC) of Receiver Operating Characteristic (ROC) curve and its 95% CI were used to evaluate the quality of the performance of the model (AUC = 0.81; 95% CI, 0.80-0.83). Time to progression curves were examined based on these new risk classification categories. Compared with low-risk lesions, the HR was 5.15 (95% CI, 1.67 – 15.84; P = 0.004), 14.28 (95% CI, 4.54 – 44.89; P < 0.001), and 33.88 (95% CI, 11.19 – 102.58; P < 0.001) for new risk categories two, three, and four. (Figure 5.5). The 3-year and 5-year probability of progression were 0.0% and 1.2% for the new Risk Category 1, 7.2% and 10.0% for new Risk Category 2, 17.1% and 34.4% for new Risk Category 3, and 36.5% and 55.4% for new Risk Category 4. (P < 0.001) (Table 5.12).     138 Figure 5.4 Cumulative counts analysis risk stratification model   Table 5.11 Performance of the cumulative counts analysis model  Input Variables Risk Category Hazard ratio (95% CI) P value AUC (95%CI) Validated LOH† Risk Category, Smoking status, Non-homogeneous appearance count ≥ 1, TB+ count ≥ 2. 1 1  0.81 (0.80-0.83) 2 5.15 (1.67-15.84) 0.004 3 14.28 (4.54-44.89) < 0.001 4 33.88 (11.19-102.58) < 0.001 † LOH = loss of heterozygosity.  § TB+ = toluidine blue positive.   139 Figure 5.5 Cumulative counts analysis Kaplan-Meier survival estimates  140 Table 5.12 Probability of progression – cumulative counts analysis            † Progression to severe dysplasia, carcinoma in-situ, or squamous cell carcinoma. § Months to last follow up or progression, whichever occurred first.   Risk Category 1 Risk Category 2 Risk Category 3 Risk Category 4 P value Total (n=266) 121 (45.5%) 89 (33.4%)  30 (11.3%) 26 (9.8%)   Months to progression†   Median (range)§ 107.9 (51.5 – 115.8) 44.2 (15.9 – 100.8) 36.4 (3.3 – 126.2) 32.1 (4.3 – 78.0) 0.04 Probability of Progression†    3-year (95% CI) 0 (0 – 0) 7.2 (4.4 – 10.0) 17.1 (10.1 – 24.1) 36.5 (26.8 – 46.2) < 0.001  5-year (95% CI) 1.2 (0 – 2.4) 10.0 (6.6 – 13.4) 34.4 (24.8 – 44.0) 55.4 (44.8 – 66.0) 141 5.5 Discussion A major barrier to oral cancer prevention continues to be the lack of validated markers that can predict for the behaviour of individual OPML and stratify them to low- and high-risk of progression. The goal of this study was to advance an established molecular risk prediction model for the malignant progression of oral LGD, by integrating clinical patterns of change over time and to determine whether it better predicted outcome. The data presented in this study is, to the best of my knowledge, the first analysis to evaluate repeated measurements of clinicopathological features as predictive markers of progression for LGD.   The first aim of the study was to determine whether the repeated measurements of specific clinical features of LGD (lesion presence, size, appearance, colour, texture, FV presentation and TB status), or a sub-set of such features, would predict malignant progression. Although many studies have reported a significant association between clinical characteristics and malignant progression of oral LGD, sometimes in combination with other biomarkers, these analyses have all been based on data obtained at a single point in time. (13, 109, 125, 140, 147, 178, 275, 290, 291) The effect of repeated measures and a temporal analysis on the ability to predict outcome in LGD has not yet been examined. This study clearly showed that repeated measurements of clinical features were significant predictors of malignant progression, as demonstrated separately in both pure longitudinal and cumulative counts analyses.  The importance and independency of these temporal clinical markers, in conjunction with time-invariant clinical and molecular features, was confirmed, using both univariate and multivariable analysis. After controlling for all variables, LOH status, smoking status, lesion site, appearance and TB status were significantly associated with progression in a temporal analysis. Comparison of the strength of 142 risk prediction between a single baseline measurement and repeated measurements revealed that clinicopathological features, including lesion size, lesion presence, appearance, TB status and FV status, become much stronger risk predictors when their features were observed over longer periods of time. These are novel findings that have not been reported in the literature previously.   In both the pure longitudinal analysis and the cumulative counts analysis, a single baseline measurement of LOH status was still the strongest predictor of risk to progression of all the features included in the multivariable regression models. In pure longitudinal analysis, lesions with an intermediate-risk LOH status possessed more than four times the risk, and those with a high-risk LOH status were almost 20 times more likely to progress to cancer. In the cumulative counts analysis, LOH risk was similar at 6.5 times and 22.3 times more likely to progress for the intermediate- and high-risk categories, over the low-risk lesions, respectively. Temporal TB status was the next strongest predictor. After controlling for all other variables, a baseline measurement of TB status was not significant; however, when repeated measurements were considered, temporal TB status was a significant predictor in both the pure longitudinal and the cumulative counts models (HR: 3.25 and 3.01, respectively). This finding suggests that TB status may not only offer prediction in early disease, but may also offer important phenotypic clues to genetic changes and expansion of molecular clones over the natural history of the disease. Further research into this area is warranted through integration of additional molecular biomarkers into this temporal framework.   The second study objective was to determine whether temporal clinicopathological patterns in LGD are associated with different LOH risk patterns. In other words, do lesions that are categorized molecularly to different risk categories display different patterns of clinical change 143 over time? The analysis has revealed that temporal clinical features are significantly correlated to LOH. Multivariable analysis showed that repeated measurements of lesion size, colour, texture, definition of lesion borders, and TB status are each associated with a high-risk molecular profile. This suggests that there is value in continuing to use these features to follow and manage LGD, even in the absence of knowledge of LOH status. It also suggests that there is value in further exploring other potential phenotypic markers for assessment of altered behavior of LGD over time.   The final objective was to develop a clinically useful model to predict malignant progression of LGD that further stratifies and improves the risk prediction of the intermediate-risk group of a previously validated model, based on LOH and temporal clinicopathological features. The ideal model would utilize the time-variant longitudinal data, provide added value over the 2012 validated model, provide two or three additional risk stratification models to the IR group, provide significant P-values and non-overlapping KM survival curves with a AUC ROC equal or greater than that reported in the 2012 paper. Although many models were explored to find a model that could provide further stratification of risk in the IR risk group, two risk classification models satisfied these criteria and were presented in this study, each with a potential use; one to examine the natural history of the disease, the other to guide clinical intervention.   When evaluating accuracy of a model or test with multiple risk categories, it is more appropriate to use the area under the curve (AUC) of a time-dependent ROC analysis rather than sensitivity or specificity analyses, which are more suited to binary decision making. The AUC summarizes all the estimates for all risk levels and is inherent to that model. The model prediction accuracy 144 for the pure longitudinal analysis was assessed at an AUC of 0.88 (95% CI, 0.81 – 0.94), which was substantially higher than the AUC of 0.79 and 0.81 that was reported in the 2012 model.(13) This model (Figure 5.2) provides separation of the LOH intermediate-risk group, with temporal TB status shifting the intermediate-risk group into low- and high-risk categories, with lesion site category providing further separation. When applying the 2012 LOH model to this cohort, 119 subjects (44.7%) fell into the intermediate-risk category. With the new longitudinal model, only 68 subjects (25.6%) fell into the intermediate-risk category. These data suggest the possibility that temporal TB positivity could be tied into the biological changes that are happening as the disease progresses over time. The biology underlying such an association requires further study.  The cumulative counts risk classification model was developed to offer insight into what happens in the first few visits and how to differentiate between patients who might benefit from additional tests and/or early intervention and for informing early management in the first five years of the disease. This model also provides further stratification of the LOH intermediate-risk group (Figure 5.6). This model suggests that when a clinician sees a non-homogeneous appearance, or a shift to a non-homogeneous appearance, one or more times during follow-up visits, a repeat biopsy is warranted, as the risk of progression may have shifted to a higher risk category. The AUC for this model was 0.81 (95% CI, 0.80 – 0.83), which was equal to or greater than that reported in the 2012 model.(13)  A review by conducted by Prince et al. (292) in 2016 reported AUC of 0.65 to 0.71 on models for outcomes on oral cancer patients. However, these prognostic models were for patients with an untreated oral cancer diagnosis, not for prognosticating patients with OED. Zarate et al. (293) 145 presented 3 prediction models with an AUC ranging from 0.81 to 0.96. However, their sample size was small (n = 10 OPML, n=10 oral cancer, and n= 8 control) and was cross-sectional in design. Their prediction models correlated to histological diagnosis and not to longitudinally followed progression of OED to cancer. The present study represents a large number (n = 306) of biopsy confirmed OED that has been followed longitudinally for a median of 72 months (range of 12 months to 239 months) to progression or last follow-up, whichever occurred first, and represents first of its kind in the field of OPML.  The results of this study may be generalizable to other North American populations with similar ethnicity and risk habits. This study is unique in that the study sample draws on patients in a community setting. Most previous studies are based on samples drawn from high-risk hospital settings, thus limiting their generalizability to lesions under surveillance in community settings. It can be very difficult to study low-grade OPML, as patients are typically seen in community dental offices instead of research hospitals. Potential study participants can be difficult to identify and recruit. Additionally, the retention of patients for longitudinal study can be quite challenging. The OCPL study, from which this analysis draws upon, is a valuable and unique cohort in that it has managed to overcome these barriers and follow over 450 such lesions since 1999.   146 Figure 5.6 Comparison of previously validated 2012 LOH risk model and cumulative counts temporal longitudinal model   LOH = loss of heterozygosity; LOH Risk category based on risk model by Zhang et al. (13)(2012). Intermediate risk = 9pLOH or 9pL0H+4qLOH or 9pLOH+17pLOH; High risk = 9pLOH+4qLOH+17pLOH (non-informative cases not included)   147 This study did show a higher risk of progression in subjects of Asian ethnicity. However, this could be due to lower numbers of non-Caucasians compared to Caucasians. To explore potential reasons why Asians were more likely to progress compared to other ethnicities, participant age was examined. Median age at did not differ between different ethnicities (P = 0.982). However, based on self-reported smoking history, Asians in this cohort were significantly less likely to be NS (P = 0.001). Previous research has shown a higher risk of progression in NS (13, 294), and this may be one reason why we are seeing increased progression in this group.  Whenever a longitudinal study spans a long period of time, there is a possibility that there may systemic differences between participants who entered the study early on, versus those who entered the study later. The proportion of malignant progression and smoking habits were assessed to see if there were any systemic differences between early study participants who enrolled in the late 1990s and those who enrolled later in the study. There were no significant differences between the earlier participants and later participants in either smoking habit (P = 0.90) or progression (P = 0.17). (Table 5.13)   148 Table 5.13 Cumulative counts analysis detailed report of sensitivity and specificity   ALL Early Cohort (Participants 1 – 153) (%)* Later Cohort (Participants 134 - 306) (%)* P value Odds Ratio (95% CI) Smoking Category b (n=306)  Never 105  52 (49.5) 53 (50.5) 0.90 1  Ever 201 101 (50.2) 100 (49.8) 0.97 (0.61 – 1.56) Smoking Amount (pack-year) (n=304) c  Median (mean)  21.8 (26.3)  25.3 (28.1)  20.0 (24.6) 0.23  Progression (n=304)  No progression 253  122 (48.2) 131 (51.8) 0.17 1  Progression  53  31 (58.5) 22 (41.5) 0.66 (0.36 – 1.20) * Row% reported. b Never: smoked 0-100 cigarettes in life time; Ever: smoked >100 cigarettes in lifetime.(259) c Pack-year data missing for 2 participants (with no progression).  One limitation to the analysis is the fact that FV data was only collected from 2004 onwards.  Earlier visits (1997 – 2004) did not have FV data available and this meant that a smaller proportion of FV data points were available for analysis with status at each of the time periods as compared to other variables. This may have limited the power of the FV analysis. FV status was significant in univariate pure longitudinal analysis (HR: 7.78; 95% CI, 3.02 – 20.03; P < 0.001), but failed to achieve significance in cumulative counts and multivariable analyses. The decision to include this relatively newer technology in the analysis was meant to be exploratory. Further future analysis is required to fully understand the potential of temporal FV status and different FV patterns as a prediction tool.   149 The search for additional markers for malignant progression is necessary and important. The ability to detect lesions at high-risk of malignant transformation lesions holds significant potential for secondary prevention of OSCC by allowing for informed management on an individual basis, including increased surveillance or therapeutic interventions. Individuals with LGD at high-risk of progression are ideal candidates for interception through chemoprevention trials. This study had reassessed clinicopathological features and advanced a risk prediction model for progression of oral LGD by fusing molecular risk predictors (LOH), to temporal patterns of clinical change observed during longitudinal follow-up. This is the first time clinicopathological changes over time in LGD has been analyzed to further facilitate the prediction of outcome and guide management. This research has the potential to build a translational bridge to the community to improve oral cancer control while maximizing health care resources and cost efficiency. Future directions should include validation of these new models in an independent cohort. Larger scale, prospective studies are also necessary. This will require multi-institutional collaboration to increase numbers of patients and geographic regions.   5.6 Conclusions  In summary, this study provides the first models utilizing temporal clinicopathological data for differentiating LGD at low-risk for progression from those with greater risk, via the largest longitudinal study of low-grade OPML from a population-based patient group. These models represent a significant first step in the evolution of precision medicine by means of a systematic decision-making process for this very heterogeneous group of lesions and an important move towards a new framework for patient follow-up where a clinical application of these markers may improve patient outcome, minimize patient morbidity and allocate health system resources 150 more effectively. In addition, the data supports the need for comparative biopsies of high-risk LGD to monitor lesion histology and to inform management. 151 Chapter 6: General Discussion With an estimated 300,000 cases reported annually, oral cancer is a significant disease. Early diagnosis is associated with a substantially better prognosis. Proficient screening can lead to the identification of OPML and early diagnosis of dysplasia, which is at risk of progressing to cancer. However, a significant obstacle is that even when identified histologically, knowing how to manage LGD, which represent the majority of dysplasia, is challenging. Differentiating between those low-grade lesions that are at high-risk of progressing to cancer from those at low-risk of progressing is difficult and a major barrier to improving outcome in this disease. Equally, deciding when to do a comparative biopsy on LGD under surveillance is challenging. Despite remarkable advancement in the field of molecular biology there is currently no single marker that can reliably predict malignant transformation in an individual patient. Limited prognosticators exist which can identify those lesions that are likely to progress and require intervention from those that will naturally regress or remain stable. The overall goal of this thesis was to advance a risk prediction model for the malignant progression of oral LGD. This was addressed by the development of novel analyses and models through 3 research projects, each of which identified important risk factors and which provide further insight into the risk stratification and management of LGD. This chapter summarizes the thesis goals and main findings and provides an overall analysis and integration of these findings considering current research in the field. Finally, overall strengths and limitations of the work and future directions are discussed.   6.1 Summary of goals and main findings The first project aimed to further investigate the premalignant nature of LM with dysplasia, and to compare that risk with oral epithelial dysplasia without LM. The results showed that dysplasia 152 with or without LM have a similar cancer risk These findings support the overall thesis goal to advance the risk prediction model for LGD by informing pathologists and clinicans that the architectural and cellular changes seen in LM with dysplasia are indicative of true dysplastic change, and should not be discounted as part of the reactive and inflammatory process occurring in OLP.   The second project sought to further investigate potential differences between LGD that occurs in smokers and LGD that occurs in NS. The question was whether the disease is the same, with respect to the clinicopathological and genetic characteristics, in smokers and NS, and whether the risk of progression is the same between these subsets. The results showed that although two-thirds of the patients were ever smokers, NS were more than twice as likely to undergo malignant transformation. Not only did a higher proportion of NS progress, but time to progression was significantly faster. Ever smokers with LGD were more likely to be male, Caucasian, and heavy drinkers. Ever smokers were more likely to have LGD at the floor of mouth; whereas a significantly higher number of LGD in NS were on the tongue. Remarkably, LGD located in the FOM in NS showed a 38-fold increase in cancer progression as compared to those in smokers. These findings support the overall thesis goal to advance the risk prediction model for LGD by confirming the risk of progression in NS and emphasizing the need for clinicians to consider smoking history, or lack thereof, and the molecular profiles in the triage and management of LGD.  The final project aimed to determine whether repeated measurements of clinicopathological features of LGD could improve the prediction of malignant progression in LGD as compared to a 153 single baseline measurement. Clustered univariate regression analysis established that repeated measurements of lesion presence, lesion size, appearance, and TB status are significantly stronger predictors of progression as compared to a single baseline measurement. Additional clustered multivariable regression analysis confirmed that measurements of TB status over time is a strong predictor of malignant transformation and should be considered in triage and risk stratification. The second aim of the study was to determine whether temporal clinicopathological patterns are associated with different molecular risk patterns. Clustered univariate and multivariable analyses confirmed that repeated high-risk clinical features were strongly associated with high-risk LOH patterns. This finding suggests that genetic alterations lead to the accumulation and progression of molecular mutations that translate to phenotypic change and sets the stage for temporal analyses of LOH. The final aim was to develop a clinically useful model to predict malignant progression of LGD that further stratifies and improves the risk prediction of the previously validated intermediate-risk group. Two models were presented. The first considered all measurements taken over all visits and considerably improved risk stratification. The Zhang et al.,(13) model placed 119 patients into intermediate risk, while this new model placed only 68 into intermediate risk, placing the remaining patients into either low- (n=158) or high-risk (n = 36) categories. The model accuracy was significantly improved over the original model that used LOH only (AUC = 0.88 and 0.81, respectively). The second model used data at clinically relevant time points in the first five years of follow-up only. This model aims to provide clinicians with information on the triage and management of disease in the first 5 years of follow-up. This model also provides further stratification of the LOH intermediate-risk group by assigning a low-intermediate risk and a high-intermediate risk to this group. These models provide patient-specific risk information that may be helpful in assessing 154 risk and benefits of repeated biopsies, follow-up interval, and other therapeutics.   6.2 Integration and Significance  All three projects have substantially built on what was previously known about the risk of progression in LGD.   One remarkable point that has come from these analyses is that despite the availability of large amount of longitudinal data and multiple repeated clinical measurements in many features, a single assessment of LOH is still the strongest single predictor of progression. Improving the previously validated model was difficult. One would think that having access to such large amount of temporally collected longitudinal data would make the development of a refined model relatively straight-forward. Yet despite the strong analysis of longitudinal clinical data on detailed lesion behaviour, the single molecular risk-predictor taken at baseline, remains the strongest single predictor of future progression.  This analysis substantiates the use of LOH as a strong tool for assessing risk in LGD, and emphasizes the need for clinicians to consider molecular genomic profiles in the triage and management of LGD.   The second project, “Characterization of epithelial oral dysplasia in non-smokers: First steps towards precision medicine”, was completed prior to the sophisticated analysis done in the third, temporal project. To further the knowledge acquired from that study, recursive partitioning utilizing only the time-invariant variables of LOH and smoking status was applied to the temporal cohort (Figure 6.1) This model shows risk of NS in a more visual format, and stresses the importance of considering smoking history in the triage and management of LGD. This 155 model improves stratification of the LOH intermediate-risk group and may be helpful for informing risk management early on in the disease. Survival estimates and corresponding hazard ratios for each risk category are displayed in Figure 6.2. The model accuracy was assessed at an AUC under ROC curve of 0.81 (95% CI, 0.80 – 82).  Figure 6.1 LOH and smoking history risk stratification model     156 Figure 6.2 LOH and smoking history risk stratification Kaplan-Meier survival estimates   The OCPL study contains more than 500 participants with a primary diagnosis of hyperplasia, mild dysplasia or moderate dysplasia diagnosed before December 22, 2016. Each of the analyses described in this thesis drew upon this prospective cohort. There was some overlap in each of the individual study cohorts. Overlap depended on the research question and the respective inclusion and exclusion criteria, which was detailed in each of the project’s respective chapters. Figure 6.3 shows the comparison of overlap between the thesis cohorts with each other. Total overlap for all three analyses was 260 participants. Figure 6.4 demonstrates the overlap between each of the thesis cohorts and the cohort reported in the LOH model reported by Zhang et al. in 2012.(13) Out of the 446, 455, and 306 study participants reported on in this thesis, 225, 275, and 244 cases, respectively, were reported in the 2012 LOH model. However, the aims of each of the thesis studies were all different.  157 Figure 6.3 Comparison of each of the project cohorts with each other    Figure 6.4 Comparison on each of the project cohorts with that of the cohort from Zhang et al., 2012  158 Clinicians need strategies to guide them when a comparative biopsy is necessary. Research has shown a lack of concordance between clinical impression and definitive biopsy diagnosis.(295) Clinical impression is not an acceptable alternative to definitive biopsy findings.(296) Management of LGD requires long-term follow-up. Knowing when to do a comparative biopsy is challenging. Early diagnosis is critical, yet biopsy is also an invasive and costly procedure. The findings presented in this thesis provide clinicians with a strategy to identify lesions at high risk of progression. Individuals in these high-risk categories should receive timely routine comparative biopsy examination according to their systematically determined risk regardless of clinical presentation. The overall significance of this work is that it has reassessed smoking and clinicopathological features as risk predictors and has advanced risk prediction for the progression of oral LGD. This research triages and improves ‘over diagnosis’ and ‘over treatment’ in low-risk categories by allowing for better target interventions to intercept disease in high-risk categories. This will reduce the number of individuals with aggressive advanced stage disease. This research has the potential to build a translational bridge to the community and is a significant step towards a new framework for patient follow-up to build upon.  6.3 Comments on strengths and limitations of the thesis research  The temporal analysis in chapter 5 was the first analysis of its kind in the field of OPML. It used very sophisticated statistical analyses that not only included multiple parameters (histological, clinical and molecular biomarkers) to predict risk, but it utilized repeated measurements of variables. The multilevel analysis accounts for both within and between patient differences. Not only were univariate analyses performed, but the effect of the independent variables on 159 malignant progression was assessed using multivariable regression analyses to control for dependency in one model.  Prospective cohort studies require large sample sizes and long follow up, which increases the study time, cost, and potential loss to follow-up. Multivariable analyses require very large sample sizes and there is a possibility of a type II error due to lack of power. Many of the univariate analyses indicated significant results. With higher numbers, it is possible that more variables within the multivariable analysis would have achieved significance. However, it is very difficult to obtain and to retain larger numbers of participants. The OCPL study is the largest and longest cohort study to date to follow LGD to progression from a community-based population. The findings from these analyses are valuable, but should be interpreted within this limitation.   A potential limitation to the generalizability of the findings is that the smoking prevalence in Vancouver, Canada, is lower than the rest of the country, and Canada has a lower prevelance compared to many other countries in the world. (Figure 6.5) In 2015, the WHO reported the global average prevalence of smoking any tobacco product among persons aged 15 years or older to be 36.1%.(297) In 2013, 20.3% of Canadians age 12 and older report daily or occasional smoking.(298) British Columbians reports a lower prevalence rate for daily or occasional smoking than the rest of the country (16.6%), and Greater Vancouver has the lowest prevalence in the country (14.5%).(298) In 2015, the majority of Canadian smokers report smoking daily, while 3.7% reported non-daily prevalence. Prevalence was higher among males (15.6%) than females (10.4%). Prevalence was highest amongst those aged 20-24 (18.5%), and generally declined with age. Prevalence was lowest among youth aged 15-19 (9.7%) and adults aged 55 160 and older (10.6%). (299) If we believe that there are different root causes or sub-types of oral cancer, depending on the etiology, then the proportion of a NS subtype of disease may differ considerably incountries with significantly different environmental (smoking) risk habits.   Figure 6.5 2015 Global smoking prevalence    The prevalence of smoking in Canada has been declining considerably since the mid-1960s. In 1965, about half of all Canadians smoked daily or occasionally, compared with 17% in 2011.(299) In 1999, 51% of Canadians self-reported ever smoking, and 25% reported current smoking. While in 2012, 44% reported being an ever smoker and 16% reported being current 161 smokers. With respect to smoking intensity, in a 2007 study published by Pierce et al. it was noted that In the United States, in the 1960s, 56% of smokers had more than 20 cigarettes a day. In 1964, the Surgeon General issued the first report linking smoking to cancer. This, and numerous public tobacco control programs and legislation led to declining smoking rates over the next decades. In 2007, the national percentage of smokers in the United States who smoked 20 cigarettes a day or more, was down to 40%, and to just 23% in California.(300)   6.4 Future Directions The findings presented in this study also have the potential to open new doors for research in finding further ways to stratify risk of cancer development. One such possibility is to explore temporal LOH patterns with respect to risk prediction. Repeated measurements of this biomarker may provide valuable insight into disease progression and may improve risk prediction and stratification.   Further research should also be aimed at further understanding the relationship between the pattern of intermittent TB positivity. For example, after controlling for all other variables, temporal TB status was a significant predictor in both the pure longitudinal and the cumulative counts models (HR: 3.25 and 3.01, respectively). This finding suggests that TB status may not only offer prediction in early disease, but may also offer important phenotypic clues to the biology of the underlying genetic changes and expansion of molecular clones over the natural history of the disease. One possible approach for future research into this area would be to explore the value of quantitative tissue and cytology assessment to determine whether high-risk temporal TB patterns are associated with DNA ploidy and nuclei associated alterations to DNA 162 content and tissue architecture.(301, 302) Such approaches could help fill in our understanding of the natural history of the disease.   Future research should be directed at finding and validating immune biomarkers to predict malignant progression of LGD. It is understood that the dysregulation and evasion of the immune system is key in the development and progression of oral cancer.(303, 304) The search for emerging biomarkers in this area has the potential to better elucidate both the order of somatic alterations as well as the corresponding changes specific to the premalignant microenvironment that enable transformation and invasion. The fact that LM sometimes present with dysplasia (or dysplasia with LM) has raised the question on how the premalignant microenvironment affects transformation and invasion, and whether immune factors in the tumour microenvironment may have prognostic value to inform clinical management and improve patient prognosis.  The development of high-throughput sequencing technologies has allowed for the ability to sequence large numbers of genes very quickly, and the costs associated with this technology has come down substantially in recent years. Next-generation sequencing is a powerful tool and has the potential to elucidate the somatic alterations that drive initiation, progression, regression and invasion in OSCC. This information is also critical in the understanding of disease stratification and subtypes. Data presented in this thesis showed that although smokers developed an OPML with OED, when OED did occur in non-smokers (NS), they were at higher risk of malignant transformation. This finding suggests that although tobacco use is considered one of the most significant risk factors for OSCC, it does not necessarily follow that this environmental exposure 163 is the only pathway to oral cancer. The development of OSCC in NS may differ from the carcinogenesis mechanisms in smoking-related malignancies, and may involve unique genetic cancer susceptibility mutations, which are driving progression and have not yet been identified. This hypothesis could be tested by the comparison of mutations in NS compared to smokers. This information will provide new insights into the pathogenesis of OSCC and further the knowledge of the genetic oral cancer progression model.   Another potential area of exploration resulting from this analysis is data visualization. Data visualization is a form of visual communication and can elucidate patterns in very large datasets, which may not be easily found in traditional statistical methods. Data visualization not only helps end users understand the results, but it can identify new patterns and generate hypotheses for further analysis.(305) These techniques are currently being explored within our lab.   Lastly, like all cancers, oral cancer is a heterogeneous disease. Improving the ability to predict malignant progression in OPML cannot be achieved solely by using histopathology, clinicopathological measurements, or a single biomarker. Instead, the development of an easy to follow triage framework which will facilitate decision making, will undoubtedly require a combined approach incorporating all of these parameters. Investigation into longitudinally sampled premalignant lesions as they progress toward cancer can provide critical insight into the sequence of events that drive progression to invasive cancer.(306) By identifying and understanding the initial events that drive initiation, progression, regression and invasion, we will advance early detection biomarkers and identify effective intervention strategies to reduce the number of individuals with aggressive advanced stage disease. 164 References 1. Ferlay J, Soerjomataram I, Dikshit R, Eser S, Mathers C, Rebelo M, et al. Cancer incidence and mortality worldwide: Sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer. 2015;136(5):E359-86. 2. Warnakulasuriya S. Global epidemiology of oral and oropharyngeal cancer. Oral oncol. 2009;45(4-5):309-16. 3. Brocklehurst P, Kujan O, O'Malley LA, Ogden G, Shepherd S, Glenny A-M. Screening programmes for the early detection and prevention of oral cancer. Cochrane Database Syst Rev. 2013 Nov 19;(11):CD004150. 4. Siegel R, Ma J, Zou Z, Jemal A. Cancer statistics, 2014. CA: A Cancer Journal for Clinicians. 2014;64(1):9-29. 5. Napier SS, Speight PM. Natural history of potentially malignant oral lesions and conditions: an overview of the literature. J Oral Pathol Med. 2008;37(1):1-10. 6. Warnakulasuriya S, Johnson NW, van der Waal I. Nomenclature and classification of potentially malignant disorders of the oral mucosa. J Oal Pathol Med. 2007;36(10):575-80. 7. Dionne KR, Warnakulasuriya S, Binti Zain R, Cheong SC. Potentially malignant disorders of the oral cavity: Current practice and future directions in the clinic and laboratory. Int J Cancer. 2015;136(3):503-15. 8. Amagasa T, Yamashiro M, Ishikawa H. Oral Leukoplakia Related to Malignant Transformation. Int J Oral Sci. 2006;3(2):45-55. 9. Lumerman H, Freedman P, Kerpel S. Oral epithelial dysplasia and the development of invasive squamous cell carcinoma. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 1995;79(3):321-9. 10. Speight PM. Update on Oral Epithelial Dysplasia and Progression to Cancer. Head Neck Pathol. 2007;1(1):61-6. 11. Bouquot JE, Gorlin RJ. Leukoplakia, lichen planus, and other oral keratoses in 23,616 white Americans over the age of 35 years. Oral Surg Oral Med Oral Pathol. 1986;61(4):373-81. 12. Mehanna HM, Rattay T, Smith J, McConkey CC. Treatment and follow-up of oral dysplasia - a systematic review and meta-analysis. Head Neck. 2009;31(12):1600-9. 13. Zhang L, Poh CF, Williams M, Laronde DM, Berean K, Gardner PJ, et al. Loss of heterozygosity (LOH) profiles--validated risk predictors for progression to oral cancer. Cancer Prev Res (Phila). 2012;5(9):1081-9. 165 14. van der Waal I. Potentially malignant disorders of the oral and oropharyngeal mucosa; terminology, classification and present concepts of management. Oral Oncol. 2009;45(4):317-23. 15. Reibel J. Prognosis of oral pre-malignant lesions: significance of clinical, histopathological, and molecular biological characteristics.Crit Rev Oral Biol Med. 2003;14(1):47-62. 16. Silverman JS, Gorsky M, Lozada F. Oral leukoplakia and malignant transformation. A follow-up study of 257 patients. Cancer. 1984;53(3):563-8. 17. Rosin MP, Cheng X, Poh C, Lam WL, Huang Y, Lovas J, et al. Use of Allelic Loss to Predict Malignant Risk for Low-grade Oral Epithelial Dysplasia. Clin Cancer Res. 2000;6(2):357-62. 18. Srivastava S, Reid BJ, Ghosh S, Kramer BS. Research Needs for Understanding the Biology of Overdiagnosis in Cancer Screening. J Cell Physiol. 2016;231(9):1870-5. 19. Esserman LJ, Thompson IM, Reid B, Nelson P, Ransohoff DF, Welch HG, et al. Addressing overdiagnosis and overtreatment in cancer: a prescription for change. Lancet Oncol. 2014;15(6):e234-42. 20. Warnakulasuriya S. Living with oral cancer: Epidemiology with particular reference to prevalence and life-style changes that influence survival. Oral Oncol. 2010;46(6):407-10. 21. World Health Organization. International Classification of Diseases and Related Health Problems 10th Revision (ICD-10). World Health Organization: Geneva, Switerzland. Version: 2015. 2015. 22. Neville BW, Day TA. Oral Cancer and Precancerous Lesions. CA Cancer J Clin. 2002;52(4):195. 23. Noguti J, De Moura CFG, De Jesus GPP, Da Silva VHP, Hossaka TA, Oshima CTF, et al. Metastasis from oral cancer: an overview. Cancer Genomics Proteomics. 2012;9(5):329-35. 24. Listl S, Jansen L, Stenzinger A, Freier K, Emrich K, Holleczek B, et al. Survival of Patients with Oral Cavity Cancer in Germany. PLoS One. 2013;8(1):e53415. 25. Jemal A, Clegg LX, Ward E, Ries LAG, Wu X, Jamison PM, et al. Annual report to the nation on the status of cancer, 1975-2001, with a special feature regarding survival. Cancer. 2004;101(1):3-27. 26. Lozano R, Naghavi M, Foreman K, Lim S, Shibuya K, Aboyans V, et al. Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet. 2012;380(9859):2095-128. 166 27.    Tanaka T, Ishigamori R. Understanding carcinogenesis for fighting oral cancer. J Oncol. 2011;2011:603740. 28. Chaturvedi AK, Anderson WF, Lortet-Tieulent J, Curado MP, Ferlay J, Franceschi S, et al. Worldwide trends in incidence rates for oral cavity and oropharyngeal cancers. J Clin Oncol. 2013;31(36):4550-9. 29. Jemal A, Simard EP, Dorell C, Noone AM, Markowitz LE, Kohler B, et al. Annual Report to the Nation on the Status of Cancer, 1975-2009, featuring the burden and trends in human papillomavirus(HPV)-associated cancers and HPV vaccination coverage levels. J Natl Cancer Inst. 2013;105(3):175-201. 30. National Cancer I. Surveillance, Epidemiology, and End Results Program. SEER Stat Fact Sheets: Oral Cavity and Pharynx Cancer. [Online].Available from: https://seer.cancer.gov/statfacts/html/oralcav.html. [cited 2018 May 27]. 31. Canadian Cancer Society's Advisory Committee on Cancer Statistics. Canadian Cancer Statistics 2017. Toronto, ON: Canadian Cancer Society; 2017 32. BC Cancer. BC Cancer Statistics. Statistics by Cancer Type - Oral. 2015. [Online]. Available from: http://www.bccancer.bc.ca/statistics-and-reports-site/Documents/Cancer_Type_Oral_2015_20180427.pdf. [cited 2018 May 27] 33. van der Waal I, Johnson NW, Warnakulasuriya S. Nomenclature and classification of potentially malignant disorders of the oral mucosa. J Oral Pathol Med. 2007;36(10):575-80. 34. Mincer HH, Coleman SA, Hopkins KP. Observations on the clinical characteristics of oral lesions showing histologic epithelial dysplasia. Oral Surg Oral Med Oral Pathol. 1972;33(3):389-99. 35. Pindborg JJ. Oral Cancer and Precancer. Bristol: John Wright and Sons Ltd.; 1980. 36. Kramer IR, Lucas RB, Pindborg JJ, Sobin LH. Definition of leukoplakia and related lesions: An aid to studies on oral precancer. Oral Surg Oral Med Oral Pathol. 1978;46(4):518,39-37,39. 37. Ho P-S, Chen P-L, Warnakulasuriya S, Shieh T-Y, Chen Y-K, Huang IY. Malignant transformation of oral potentially malignant disorders in males: a retrospective cohort study. BMC Cancer. 2009;9(1):260. 38.    Oral Submucous Fibrosis Experts Symposium held at the 5th International Congress on Oral Cancer. London, 26th September 1997. Abstracts. Oral diseases. 1997;3(4):276. 39.    Zain RB, Ikeda N, Razak IA, Axéll T, Majid ZA, Gupta PC, et al. A national epidemiological survey of oral mucosal lesions in Malaysia. Community Dent Oral Epidemiol. 1997;25(5):377-83. 167 40.    Casparis S, Borm JM, Tektas S, Kamarachev J, Locher MC, Damerau G, et al. Oral lichen planus (OLP), oral lichenoid lesions (OLL), oral dysplasia, and oral cancer: retrospective analysis of clinicopathological data from 2002-2011. Oral Maxillofac Surg. 2015;19(2):149-56. 41.    Petti S. Pooled estimate of world leukoplakia prevalence: a systematic review. Oral Oncol. 2003;39(8):770-80. 42.    Villa A, Gohel A. Oral potentially malignant disorders in a large dental population. J Appl Oral Sci. 2014;22(6):473-6. 43.    Kramer IR, Pindborg JJ, Bezroukov V, Infirri JS. Guide to epidemiology and diagnosis of oral mucosal diseases and conditions. World Health Organization. Community Dent Oral Epidemiol. 1980;8(1):1-24. 44.    Leemans CR, Braakhuis BJM, Brakenhoff RH. The molecular biology of head and neck cancer. Nat Rev Cancer. 2011;11(1):9-22. 45.    Amarasinghe HK, Usgodaarachchi US, Johnson NW, Lalloo R, Warnakulasuriya S. Betel-quid chewing with or without tobacco is a major risk factor for oral potentially malignant disorders in Sri Lanka: A case-control study. Oral Oncol. 2010;46(4):297-301. 46. Lee CH, Ko AMS, Warnakulasuriya S, Yin BL, Sunarjo, Zain RB, et al. Intercountry prevalences and practices of betel‐quid use in south, southeast and eastern asia regions and associated oral preneoplastic disorders: An international collaborative study by asian betel‐quid consortium of south and east Asia. Int J Cancer. 2011;129(7):1741-51. 47. Herrero R, Castellsagué X, Pawlita M, Lissowska J, Kee F, Balaram P, et al. Human papillomavirus and oral cancer: the International Agency for Research on Cancer multicenter study. J Natl Cancer Inst. 2003;95(23):1772-83. 48. D'Souza G, Kreimer AR, Viscidi R, Pawlita M, Fakhry C, Koch WM, et al. Case-control study of human papillomavirus and oropharyngeal cancer. N Engl J Med. 2007;356(19):1944-56. 49. Gillison ML. Human papillomavirus-associated head and neck cancer is a distinct epidemiologic, clinical, and molecular entity. Mein Oncol. 2004;31(6):744-54. 50. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Human Papillomaviruses. IARC Monogr Aval Carcinog Risks Hum. 2007;90:1-689.   51. Banoczy J, Csiba A. Occurrence of epithelial dysplasia in oral leukoplakia. Analysis and follow-up study of 12 cases. Oral Surg Oral Med Oral Pathol. 1976;42(6):766-74. 52. Silverman S, Gorsky M, Kaugars GE. Leukoplakia, dysplasia, and malignant transformation. Oral Surg Oral Med Oral Pathol Oral Radiol and Endod. 1996;82(2):117. 168 53. Hashibe M, Hunt J, Wei M, Buys S, Gren L, Lee YCA. Tobacco, alcohol, body mass index, physical activity, and the risk of head and neck cancer in the prostate, lung, colorectal, and ovarian (PLCO) cohort. Head Neck. 2013;35(7):914-22. 54. Hashibe M, Brennan P, Benhamou S, Castellsague X, Chu C, Curado MP, et al. Alcohol drinking in never users of tobacco, cigarette smoking in never drinkers, and the risk of head and neck cancer: Pooled analysis in the international head and neck cancer epidemiology consortium. J Natl Cancer Inst. 2007;99(10):777-89. 55. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Tobacco smoke and involuntary smoking. IARC Monogr Aval Carcinog Risks Hum. 2004;83:1-1438. 56. Pai SI, Westra WH. Molecular pathology of head and neck cancer: implications for diagnosis, prognosis, and treatment. Annu Rev Pathol. 2009;4(1):49-70. 57. Lee J, Taneja V, Vassallo R. Cigarette smoking and inflammation: cellular and molecular mechanisms. J Dent Res. 2012;91(2):142-9. 58. Hecht SS. Tobacco carcinogens, their biomarkers and tobacco-induced cancer. Nat Rev Cancer. 2003;3(10):733-44. 59. Petti S. Lifestyle risk factors for oral cancer. Oral Oncol. 2009;45(4–5):340-50. 60. Stransky N, Egloff AM, Tward AD, Kostic AD, Cibulskis K, Sivachenko A, et al. The mutational landscape of head and neck squamous cell carcinoma. Science. 2011;333(6046):1157-60. 61. Urashima M, Hama T, Suda T, Suzuki Y, Ikegami M, Sakanashi C, et al. Distinct effects of alcohol consumption and smoking on genetic alterations in head and neck carcinoma. PLoS One. 2013;8(11):e80828. 62. Villa A, Villa C, Abati S. Oral cancer and oral erythroplakia: an update and implication for clinicians. Aust Dent J. 2011;56(3):253-6. 63. Waldron CA, Shafer WG. Leukoplakia revisited. A clinicopathologic study 3256 oral leukoplakias. Cancer. 1975;36(4):1386. 64. Lee Y-CA, Marron M, Benhamou S, Bouchardy C, Ahrens W, Pohlabeln H, et al. Active and Involuntary Tobacco Smoking and Upper Aerodigestive Tract Cancer Risks in a Multicenter Case-Control Study. Cancer Epidemiol Biomarkers Prev. 2009;18(12):3353. 65. Lubin JH, Purdue M, Kelsey K, Zhang Z-F, Winn D, Wei Q, et al. Total Exposure and Exposure Rate Effects for Alcohol and Smoking and Risk of Head and Neck Cancer: A Pooled Analysis of Case-Control Studies. Am J Epidemiol. 2009;170(8):937-47. 169 66. Turati F, Garavello W, Tramacere I, Pelucchi C, Galeone C, Bagnardi V, et al. A meta-analysis of alcohol drinking and oral and pharyngeal cancers: results from subgroup analyses. Alcohol Alcohol. 2013;48(1):107-18. 67. Bagnardi V, Rota M, Botteri E, Tramacere I, Islami F, Fedirko V, et al. Alcohol consumption and site-specific cancer risk: a comprehensive dose-response meta-analysis. Br J Cancer. 2015;112(3):580. 68. Seitz HK, Stickel F. Molecular mechanisms of alcohol-mediated carcinogenesis. Nat Rev Cancer. 2007;7(8):599-612. 69. Maserejian NN, Joshipura KJ, Rosner BA, Giovannucci E, Zavras AI. Prospective study of alcohol consumption and risk of oral premalignant lesions in men. Cancer Epidemiol Biomarkers Prev. 2006;15(4):774-81. 70. Dietrich T, Reichart PA, Scheifele C. Clinical risk factors of oral leukoplakia in a representative sample of the US population. Oral Oncol. 2004;40(2):158-63. 71. Mashberg A, Boffetta P, Winkelman R, Garfinkel L. Tobacco smoking, alcohol drinking, and cancer of the oral cavity and oropharynx among U.S. veterans. Cancer. 1993;72(4):1369-75. 72. Centers for Disease Control and Prevention. Smoking and tobacco use: data and statistics. 2017. [Online]. Available from: https://www.cdc.gov/tobacco/data_statistics/index.htm. [cited 2018 May 27]. 73. Llewellyn CD, Johnson NW, Warnakulasuriya K. Risk factors for squamous cell carcinoma of the oral cavity in young people - a comprehensive literature review. Oral Oncol. 2001;37(5):401-18. 74.    IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Smokeless tobacco and some tobacco-specific N-nitrosamines. IARC Monogra Eval Carcinog Risks Hum. 2007;89:1-592. 75. IARC Working Group on the Evaluations of Carcinogenic Risks to Humans. Personal habits and indoor combustions. Volume 100 E. A review of human carcinogens. IARC Monogr Aval Carcinog Risks Hum. 2012;100(PtE):1-538. 76. Haddad RI, Shin DM. Recent Advances in Head and Neck Cancer. N Engl J Med. 2008;359(11):1143-54. 77. Kreimer AR, Clifford GM, Boyle P, Franceschi S. Human papillomavirus types in head and neck squamous cell carcinomas worldwide: A systematic review. Cancer Epidemiol Biomarkers Prev. 2005;14(2):467-75. 78. Tural D, Elicin O, Batur S, Arslan D, Oz B, Serdengecti S, et al. Increase in the Rate of HPV Positive Oropharyngeal Cancers During 1996-2011 in a Case Study in Turkey. Asian Pac J Cancer Prev. 2013;14(10):6065-8. 170 79. Nordfors C, Vlastos A, Du J, Ahrlund-Richter A, Tertipis N, Grun N, et al. Human papillomavirus prevalence is high in oral samples of patients with tonsillar and base of tongue cancer. Oral Oncol. 2014;50(5):491-7. 80.    Steinau M, Hariri S, Gillison ML, Broutian TR, Dunne EF, Tong ZY, et al. Prevalence of Cervical and Oral Human Papillomavirus Infections Among US Women. J Infect Dis. 2014;209(11):1739-43. 81. Saulle R, Semyonov L, Mannocci A, Careri A, Saburri F, Ottolenghi L, et al. Human papillomavirus and cancerous diseases of the head and neck: a systematic review and meta-analysis. Oral Dis. 2015;21(4):417-31. 82. Kuhs KAL, Gonzalez P, Struijk L, Castro F, Hildesheim A, van Doorn LJ, et al. Prevalence of and Risk Factors for Oral Human Papillomavirus Among Young Women in Costa Rica. J of Infect Dis. 2013;208(10):1643-52. 83. Thomas M, Pim D, Banks L. The role of the E6-p53 interaction in the molecular pathogenesis of HPV. Oncogene. 1999;18(53):7690-700. 84. Sankari SL, Gayathri K, Balachander N, Malathi L. Candida in potentially malignant oral disorders. J Pharm Bioallied Sci. 2015;7(Suppl 1):S162-4. 85. Madeleine MM, Finch JL, Lynch CF, Goodman MT, Engels EA. HPV-related cancers after solid organ transplantation in the United States. Am J Transplant. 2013;13(12):3202-9. 86. Grulich AE, van Leeuwen MT, Falster MO, Vajdic CM, Engels EA, Biggar RJ, et al. Incidence of cancers in people with HIV/AIDS compared with immunosuppressed transplant recipients: a meta-analysis. Lancet. 2007;370(9581):59-67. 87. Engels EA, Biggar RJ, Hall HI, Cross H, Crutchfield A, Finch JL, et al. Cancer risk in people infected with human immunodeficiency virus in the United States. Int J Cancer. 2008;123(1):187-94. 88. Shiels MS, Cole SR, Kirk GD, Poole C. A meta-analysis of the incidence of non-AIDS cancers in HIV-infected individuals. J Acquir Immune Defic Syndr. 2009;52(5):611-22. 89. Gillison ML, D'Souza G, Westra W, Sugar E, Xiao W, Begum S, et al. Distinct risk factor profiles for human papillomavirus type 16-positive and human papillomavirus type 16-negative head and neck cancers. J Natl Cancer Inst. 2008;100(6):407-20. 90. Darwich L, Canadas MP, Videla S, Coll J, Molina-Lopez RA, Cobarsi P, et al. Oral human papillomavirus type-specific infection in HIV-infected men: a prospective cohort study among men who have sex with men and heterosexual men. Clin Microbiol Infect. 2014;20(9):O585-9. 171 91. Mooij SH, Boot HJ, Speksnijder AG, Stolte IG, Meijer CJ, Snijders PJ, et al. Oral human papillomavirus infection in HIV-negative and HIV-infected MSM. Aids. 2013;27(13):2117-28. 92. Chen C-H, Chung C-Y, Wang L-H, Lin C, Lin H-L, Lin H-C. Risk of cancer among HIV-infected patients from a population-based nested case-control study: implications for cancer prevention. BMC Cancer. 2015;15(1):1099-133. 93. Warnakulasuriya S, Reibel J, Bouquot J, Dabelsteen E. Oral epithelial dysplasia classification systems: predictive value, utility, weaknesses and scope for improvement. J Oral Pathol Med. 2008;37(3):127-33. 94. Serrano M, Finkel T, Blasco MA. The common biology of cancer and ageing. Nature. 2007;448(7155):767-74. 95. American Cancer Society. Oral cavity and oropharyngeal cancer. 2018. [Online] Available from: https://www.cancer.org/cancer/oral-cavity-and-oropharyngeal-cancer.html. [cited 2018 May 27] 96. Moore SR, Johnson NW, Pierce AM, Wilson DF. The epidemiology of lip cancer: a review of global incidence and aetiology. Oral Dis. 1999;5(3):185-95. 97. Warnakulasuriya S. Significant oral cancer risk associated with low socioeconomic status. Evid Based Dent. 2009;10(1):4-5. 98. Conway DI, Petticrew M, Marlborough H, Bertbiller J, Hashibe M, Macpherson LMD. Socioeconomic inequalities and oral cancer risk: A systematic review and meta-analysis of case-control studies. Int J Cancer. 2008;122(12):2811-9. 99.. McLaughlin JK, Gridley G, Block G, Winn DM, Preston-Martin S, Schoenberg JB, et al. Dietary factors in oral and pharyngeal cancer. J Nat Cancer Inst. 1988;80(15):1237-43. 100. Garewal H. Chemoprevention of oral cancer: beta-carotene and vitamin E in leukoplakia. Dur J Cancer Prev. 1994;3(2):101-8. 101. Leoncini E, Edefonti V, Hashibe M, Parpinel M, Cadoni G, Ferraroni M, et al. Carotenoid intake and head and neck cancer: a pooled analysis in the International Head and Neck Cancer Epidemiology Consortium. Eur J Epidemiol. 2015. 102. Garavello W, Foschi R, Talamini R, La Vecchia C, Rossi M, Dal Maso L, et al. Family history and the risk of oral and pharyngeal cancer. Int J Cancer. 2008;122(8):1827-31. 103. Epstein JB, Silverman S, Jr., Epstein JD, Lonky SA, Bride MA. Analysis of oral lesion biopsies identified and evaluated by visual examination, chemiluminescence and toluidine blue. Oral Oncol. 2008;44(6):538-44. 172 104. Silverman S, Bhargava K, Smith LW, Malaowalla AM. Malignant transformation and natural history of oral leukoplakia in 57,518 industrial workers of Gujarat, India. Cancer. 1976;38(4):1790-5. 105. Gupta PC, Mehta FS, Daftary DK, Pindborg JJ, Bhonsle RB, Jalnawalla PN, et al. Incidence rates of oral cancer and natural history of oral precancerous lesions in a 10-year follow-up study of Indian villagers. Community Dent Oral Epidemiol. 1980;8(6):283. 106. Lee JJ, Hong WK, Hittelman WN, Mao L, Lotan R, Shin DM, et al. Predicting Cancer Development in Oral Leukoplakia: Ten Years of Translational Research. Clin Cancer Res. 2000;6(5):1702. 107. Hsue SS, Wang WC, Chen CH, Lin CC, Chen YK, Lin LM. Malignant transformation in 1458 patients with potentially malignant oral mucosal disorders: a follow‐up study based in a Taiwanese hospital. J Oral Pathol Med. 2007;36(1):25-9. 108. Liu W, Wang Y-F, Zhou H-W, Shi P, Zhou Z-T, Tang G-Y. Malignant transformation of oral leukoplakia: a retrospective cohort study of 218 Chinese patients. BMC Cancer. 2010;10(1):685. 109. Schepman KP, van der Meij EH, Smeele LE, van der Waal I. Malignant transformation of oral leukoplakia: a follow-up study of a hospital-based population of 166 patients with oral leukoplakia from The Netherlands. Oral Oncol. 1998;34(4):270-5. 110. Shafer WG, Waldron CA. Erythroplakia of the oral cavity. Cancer. 1975;36(3):1021-8. 111. Warnakulasuriya S, Kovacevic T, Madden P, Coupland VH, Sperandio M, Odell E, et al. Factors predicting malignant transformation in oral potentially malignant disorders among patients accrued over a 10‐year period in South East England. J Oral Pathol Med. 2011;40(9):677-83. 112. Pindborg JJ, World Health Organization. Histological typing of cancer and precancer of the oral mucosa. 2nd ed. Berlin, Germany: Springer; 1997. 113. Barnes L, Eveson HW, Reichart P, Sidransky D. World Health Organization Classification of Tumours. Pathology and genetics of head and neck tumours 2005; Lyon: IARC Press.  114. Abbey LM, Kaugars GE, Gunsolley JC, Burns JC, Page DG, Svirsky JA, et al. Intraexaminer and interexaminer reliability in the diagnosis of oral epithelial dysplasia. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 1995;80(2):188-91. 115. Bosman FT. Dysplasia classification: pathology in disgrace? J Pathol. 2001;194(2):143-4. 116. Pindborg JJ, Reibel J, Holmstrup P. Subjectivity in evaluating oral epithelial dysplasia, carcinoma in situ and initial carcinoma. J Oral Pathol Med. 1985;14(9):698-708. 173 117. Karabulut A, Reibel J, Therkildsen MH, Praetorius F, Nielsen HW, Dabelsteen E. Observer variability in the histologic assessment of oral premalignant lesions. J Oral Pathol Med. 1995;24(5):198-200. 118. Fleskens S, Slootweg P. Grading systems in head and neck dysplasia: their prognostic value, weaknesses and utility. Head Neck Oncol. 2009;1:11-3284-1-11. 119. Shubhasini AR, Praveen BN, Hegde U, Uma K, Subha G. Keerthi G, et al. Inter- and Intra-Observer Variability in Diagnosis of Oral Dysplasia. Asian Pac J Cancer Prev. 2017;18(12):3251-4. 120. Kujan O, Oliver RJ, Khattab A, Roberts SA, Thakker N, Sloan P. Evaluation of a new binary system of grading oral epithelial dysplasia for prediction of malignant transformation. Oral Oncol. 2006;42(10):987-93. 121. Kujan O, Khattab A, Oliver RJ, Roberts SA, Thakker N, Sloan P. Why oral histopathology suffers inter-observer variability on grading oral epithelial dysplasia: An attempt to understand the sources of variation. Oral Oncol. 2007;43(3):224-31. 122. Warnakulasuriya S, Tilakaratne WM, Ranganathan K, Kuriakose MA. Report of a consensus meeting of a group of oral and general pathologists in India on grading of oral epithelial dysplasia. Oral Oncol. 2018. 123. Avon S-L, Klieb HBE. Oral soft-tissue biopsy: an overview. J Can Dent Assoc. 2012;78:c75. 124. Shafer WG, Waldron CA. A clinical and histopathologic study of oral leukoplakia. Surge Gynecol Obstet. 1961;112:411-20. 125. Ho MW, Risk JM, Woolgar JA, Field EA, Field JK, Steele JC, et al. The clinical determinants of malignant transformation in oral epithelial dysplasia. Oral Oncol. 2012;48(10):969-76. 126. Silverman JS. Observations on the clinical characteristics and natural history of oral leukoplakia. J Am Dent Assoc. 1968;76(4):772-7. 127. van der Meij EH, Schepman KP, van der Waal I. The possible premalignant character of oral lichen planus and oral lichenoid lesions: a prospective study. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2003;96(2):164-71. 128. Sugerman PB, Savage NW, Walsh LJ, Zhao ZZ, Zhou XJ, Khan A, et al. The pathogenesis of oral lichen planus. Crit Rev Oral Biol Med. 2002;13(4):350-65. 129. Epstein JB, Wan LS, Gorsky M, Zhang LW. Oral lichen planus: Progress in understanding its malignant potential and the implications for clinical management. Oral Surgery Oral Medicine Oral Pathology Oral Radiology and Endodontics. 2003;96(1):32-7. 174 130. Lodi G, Scully C, Carrozzo M, Griffiths M, Sugerman PB, Thongprasom K. Current controversies in oral lichen planus: report of an international consensus meeting. Part 2. Clinical management and malignant transformation. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2005;100(2):164-78. 131. Krutchkoff DJ, Eisenberg E. Lichenoid dysplasia: a distinct histopathologic entity. Oral Surg Oral Med Oral Pathol. 1985;60(3):308-15. 132. Speight PM, Khurram SA, Kujan O. Oral potentially malignant disorders: risk of progression to malignancy. Oral Surg Oral Med Oral Pathol Oral Radiol. 2017.[Epub ahead of print] 133. Aghbari SMH, Abushouk AI, Attia A, Elmaraezy A, Menshawy A, Ahmed MS, et al. Malignant transformation of oral lichen planus and oral lichenoid lesions: A meta-analysis of 20095 patient data. Oral Oncol. 2017;68:92-102. 134. Gonzalez-Moles MA, Scully C, Gil-Montoya JA. Oral lichen planus: controversies surrounding malignant transformation. Oral Dis. 2008;14(3):229-43. 135. Eisen D. The clinical features, malignant potential, and systemic associations of oral lichen planus: a study of 723 patients. J Am Acad Dermatol. 2002;46(2):207-14. 136. Fitzpatrick SG, Hirsch SA, Gordon SC. The malignant transformation of oral lichen planus and oral lichenoid lesions: a systematic review. J Am Dent Assoc. 2014;145(1):45-56. 137. Tilakaratne WM, Ekanayaka RP, Warnakulasuriya S. Oral submucous fibrosis: a historical perspective and a review on etiology and pathogenesis. Oral Surg Oral Med Oral Pathol Oral Radiol. 2016;122(2):178-91. 138. Ekanayaka RP, Tilakaratne WM. Oral submucous fibrosis: review on mechanisms of malignant transformation. Oral Surg Oral Med Oral Pathol Oral Radiol. 2016;122(2):192-9. 139. Wang Y-Y, Tail Y-H, Wang W-C, Chen C-Y, Kao Y-H, Chen Y-K, et al. Malignant transformation in 5071 southern Taiwanese patients with potentially malignant oral mucosal disorders. BMC Oral Health. 2014;14:99. 140. Holmstrup P, Vedtofte P, Reibel J, Stoltze K. Long-term treatment outcome of oral premalignant lesions. Oral Oncol. 2006;42(5):461-74. 141. Holmstrup P, Vedtofte P, Reibel J, Stoltze K. Oral premalignant lesions: is a biopsy reliable? J Oral Pathol Med. 2007;36(5):262-6. 142. Gale N PB, Sidransky D. Epithelial precursor lesions. In: Press I, editor.: World Health Organization classification of tumours: pathology and genetics of tumours of the head and neck. Lyon: IARC Press. 2005. p. 143.   175 143. Lind PO. Malignant transformation in oral leukoplakia. Scand J Dent Res. 1987;95(6):449-55. 144. Mashberg A, Merletti F, Boffetta P, Gandolfo S, Ozzello F, Fracchia F, et al. Appearance, site of occurrence, and physical and clinical characteristics of oral carcinoma in Torino, Italy. Cancer. 1989;63(12):2522-7. 145. Williams PM, Poh CF, Hovan AJ, Ng S, Rosin MP. Evaluation of a suspicious oral mucosal lesion. J Can Dent Assoc. 2008;74(3):275-80. 146. Bagan J, Sarrion G, Jimenez Y. Oral cancer: Clinical features. Oral Oncol. 2010;46(6):414-7. 147. Zhang L, Williams M, Poh CF, Laronde D, Epstein JB, Durham S, et al. Toluidine blue staining identifies high-risk primary oral premalignant lesions with poor outcome. Cancer research. 2005;65(17):8017-21 148. Thomson PJ. Oral precancer: diagnosis and management of potentially malignant disorders. Hoboken, NJ: John Wiley & Sons, Incorporated; 2012. 149. Simi S, Nandakumar G, Anish T. White lesions in the oral cavity: a clinicopathological study from a tertiary care dermatology centre in kerala, India. Indian J Dermatol. 2013;58(4):269-74. 150. Mehta FS, Shroff BC, Gupta PC, Daftary DK. Oral leukoplakia in relation to tobacco habits. A ten-year follow-up study of Bombay policemen. Oral Surg Oral Med Oral Pathol. 1972;34(3):426-33. 151.  Axell T, Pindborg JJ, Smith CJ, Waal I, an International Collaborative Group on Oral White L. Oral white lesions with special reference to precancerous and tobacco-related lesions: conclusions of an international symposium held in Uppsala, Sweden, May 18-21 1994. J Oral Pathol Med. 1996;25(2):49-54. 152. Amagasa T, Yamashiro M, Uzawa N. Oral premalignant lesions: from a clinical perspective. Int J of Clin Oncol. 2011;16(1):5-14. 153. Pindborg JJ, Renstrup G, Poulsen HE, Silverman S. Studies in Oral Leukoplakias. V. Clinical and Histologic Signs of Malignancy. Acta Odontologica. 1963;21(5):407-14. 154. Mehta FS, Pindborg JJ, Gupta PC, Daftary DK. Epidemiologic and histologic study of oral cancer and leukoplakia among 50,915 villagers in India. Cancer. 1969;24(4):832-49. 155. Silverman JS, Gorsky M. Proliferative verrucous leukoplakia: a follow-up study of 54 cases. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 1997;84(2):154-7. 156. Thomas G, Hashibe M, Jacob BJ, Ramadas K, Mathew B, Sankaranarayanan R, et al. Risk factors for multiple oral premalignant lesions. Int J Cancer. 2003;107(2):285-91. 176 157. Angadi PV, Savitha JK, Rao SS, Sivaranjini Y. Oral field cancerization: current evidence and future perspectives. Oral Maxillofac Surg. 2012;16(2):171-80. 158. Slaughter DP, Southwick HW, Smejkal W. Field cancerization in oral stratified squamous epithelium; clinical implications of multicentric origin. Cancer. 1953;6(5):963-8. 159. Braakhuis BJM, Tabor MP, Kummer JA, Leemans CR, Brakenhoff RH. A genetic explanation of Slaughter's concept of field cancerization: evidence and clinical implications. Cancer Res. 2003;63(8):1727-30. 160. Simple M, Suresh A, Das D, Kuriakose MA. Cancer stem cells and field cancerization of Oral squamous cell carcinoma. Oral Oncol. 2015;51(7):643-51. 161. British Columbia Oral Cancer Prevention Program, BC Cancer Agency; College of Dental Surgeons of British Columbia. Guideline for the early detection of oral cancer in British Columbia 2008. J Can  Dent Assoc. 2008;74(3):245. 162. Laronde DM, Williams PM, Hislop TG, Poh C, Ng S, Zhang L, et al. Decision making on detection and triage of oral mucosa lesions in community dental practices: screening decisions and referral. Community Dent Oral Epidemiol. 2014;42(4):375-84. 163. Partridge M, Pateromichelakis S, Phillips E, Emilion GG, A'Hern RP, Langdon JD. A case-control study confirms that microsatellite assay can identify patients at risk of developing oral squamous cell carcinoma within a field of cancerization. Cancer Res. 2000;60(14):3893-8. 164. Braakhuis BJM, Leemans CR, Brakenhoff RH. Expanding fields of genetically altered cells in head and neck squamous carcinogenesis. Semin Cancer Biol. 2005;15(2):113-20. 165. Braakhuis BJ, Tabor MP, Leemans CR, van der Waal I, Snow GB, Brakenhoff RH. Second primary tumors and field cancerization in oral and oropharyngeal cancer: molecular techniques provide new insights and definitions. Head Neck. 2002;24(2):198-206. 166. Kleinsmith LJ. Principles of cancer biology. San Francisco: Pearson/Benjamin Cummings; 2006. 167. Hirshberg A, Shnaiderman-Shapiro A, Kaplan I, Berger R. Metastatic tumours to the oral cavity – Pathogenesis and analysis of 673 cases. Oral Oncol. 2008;44(8):743-52. 168. Patel SG, Shah JP. TNM Staging of Cancers of the Head and Neck: Striving for Uniformity Among Diversity. CA Cancer J Clin. 2005;55(4):242-58. 169. Shah JP, Candela FC, Poddar AK. The patterns of cervical lymph node metastases from squamous carcinoma of the oral cavity. Cancer. 1990;66(1):109-13. 170. Marley JJ, Linden GJ, Cowan CG, Lamey PJ, Johnson NW, Warnakulasuriya K, et al. A comparison of the management of potentially malignant oral mucosal lesions by oral 177 medicine practitioners and oral & maxillofacial surgeons in the UK. J Oral Pathol Med. 1998;27(10):489-95. 171. Thomson PJ, Goodson ML, Hamadah O. Cohort studies in oral precancer management: Intervention vs. observation. Oral Oncol. 2009 3(Suppl 1):116. 172. Scully C. Oral Precancer: preventive and medical appraoches to management. Eur J Cancer B Oral Oncol. 1995;31B(1):16-26. 173. Hamadah O, Hepburn S, Thomson PJ. Effects of active non-smoking programmes on smoking behaviour in oral precancer patients. Int J Oral Maxillofac Surg. 2007;36(8):706-11. 174. Lodi G, Porter S. Management of potentially malignant disorders: evidence and critique. J Oral Pathol Med. 2008;37(2):63-9. 175. Lodi G, Sardella A, Bez C, Demarosi F, Carrassi A. Interventions for treating oral leukoplakia. Cochrane Database Syst Rev. 2006;(4):CD001829. 176. Marley JJ, Cowan CG, Lamey PJ, Linden GJ, Johnson NW, Warnakulasuriya K. Management of potentially malignant oral mucosal lesions by consultant UK oral and maxillofacial surgeons. Br J Oral Maxillofacl Surg. 1996;34(1):28-36. 177. Zhang L, Lubpairee T, Laronde DM, Rosin MP. Should severe epithelial dysplasia be treated? Oral Oncology. 2016;60:125-9. 178. Einhorn J, Wersall J. Incidence of oral carcinoma in patients with leukoplakia of the oral mucosa. Cancer. 1967;20(12):2189-93. 179. Vedtofte P, Holmstrup P, Hjortinghhansen E, Pindborg JJ. Surigcal treatment of premalignant lesions of the oral mucosa. Int J Oral Maxillofac Surg.1987;16(6):656-64. 180. Schoelch ML, Sekandari N, Regezi JA, Silverman S. Laser management of oral leukoplakias: A follow-up study of 70 patients. Laryngoscope. 1999;109(6):949-53. 181. Holmstrup P. Can we prevent malignancy by treating premalignant lesions? Oral Oncol. 2009;45(7):549-50. 182. van der Waal I, Schepman KP, vanderMeij EH, Smeele LE. Oral leukoplakia: a clinicopathological review. Oral Oncol. 1997;33(5):291-301. 183. Arduino PG, Surace A, Carbone M, Elia A, Massolini G, Gandolfo S, et al. Outcome of oral dysplasia: a retrospective hospital-based study of 207 patients with a long follow-up. J Oral Pathol Med 2009;38(6):540-4. 184. Lodi G, Franchini R, Warnakulasuriya S, Varoni EM, Sardella A, Kerr AR, et al. Interventions for treating oral leukoplakia to prevent oral cancer. Cochrane Database Syst Rev. 2016;(7):CD001829. 178 185. Thomson PJ. Field change and oral cancer: new evidence for widespread carcinogenesis? Int J Oral Maxillofac Surg. 2002;31(3):262-6. 186. Guneri P, Epstein JB. Why are we still unable to accurately determine the malignant potential or the behavior of oral mucosal lesions? Oral Oncol. 2017;71:177-9. 187. Mulshine JL, Atkinson JC, Greer RO, Papadimitrakopoulou VA, Van Waes C, Rudy S, et al. Randomized, double-blind, placebo-controlled phase IIB trial of the cyclooxygenase inhibitor ketorolac as an oral rinse in oropharyngeal leukoplakia. Clin Cancer Res. 2004;10(5):1565-73. 188. Lippman SM, Gibson N, Subbaramaiah K, Dannenberg AJ. Combined targeting of the epidermal growth factor receptor and cyclooxygenase-2 pathways. Clin Cancer Res. 2005;11(17):6097-9. 189. Papaclimitrakopoulou VA, William WN, Dannenberg A, Lippman SM, Lee JJ, Ondrey FG, et al. Pilot randomized phase II study of celecoxib in oral premalignant lesions. Clini Cancer Res. 2008;14(7):2095-101. 190. William WN, Jr., Papadimitrakopoulou V, Lee JJ, Mao L, Cohen EEW, Lin HY, et al. Erlotinib and the Risk of Oral Cancer The Erlotinib Prevention of Oral Cancer (EPOC) Randomized Clinical Trial. JAMA Oncol. 2016;2(2):209-16. 191. Pollak M. Metformin and Other Biguanides in Oncology: Advancing the Research Agenda. Cancer Prev Res. 2010;3(9):1060-5. 192. Quinn BJ, Kitagawa H, Memmott RM, Gills JJ, Dennis PA. Repositioning metformin for cancer prevention and treatment. Trends Endocrinol Metabol. 2013;24(9):469-80. 193. Pollak MN. Investigating Metformin for Cancer Prevention and Treatment: The End of the Beginning. Cancer Discov. 2012;2(9):778-90. 194. Gutkind JS, Ondrey F, Laronde DM, Rosin MP, Molinolo A, Coffey C, et al. M4OC-prevent: clinical evaluation of metformin for oral cancer precision. Cancer Res. 2018; In: Proceedings of the AACR Annual Meeting; 2018; Chicago, IL: Abstract 4985. 195. Whiteside TL. The tumor microenvironment and its role in promoting tumor growth. Oncogene. 2008;27(45):5904-12. 196. de Visser KE, Eichten A, Coussens LM. Paradoxical roles of the immune system during cancer development. Nature Rev Cancer. 2006;6(1):24-37. 197. Grivennikov SI, Greten FR, Karin M. Immunity, Inflammation, and Cancer. Cell. 2010;140(6):883-99. 198. Gutkind JS, Bui JD. The Next Frontier: Head and Neck Cancer Immunoprevention. Cancer Prev Res. 2017;10(12):681-3. 179 199. Ribeiro AS, Salles PR, da Silva TA, Mesquita RA. A review of the nonsurgical treatment of oral leukoplakia. Int J Dent. 2010;2010:186018. 200. Patton LL. The effectiveness of community-based visual screening and utility of adjunctive diagnostic aids in the early detection of oral cancer. Oral Oncol. 2003;39(7):708-23. 201. Richart RM. A clinical staining test for the in vivo delineation of dysplasia and carcinoma in situ. Am J Obstet Gynecol. 1963;86:703-12. 202. Dunipace AJ, Beaven R, Noblitt T, Li Y, Zunt S, Stookey G. Mutagenic potential of toluidine blue evaluated in the Ames test. Mutat Res. 1992;279(4):255-9. 203. Herlin P, Marnay J, Jacob JH, Ollivier JM, Mandard AM. A study of the mechanism of the toluidine blue dye test. Endoscopy. 1983;15(1):4-7. 204. Epstein JB, Scully C, Spinelli J. Toluidine blue and Lugol's iodine application in the assessment of oral malignant disease and lesions at risk of malignancy. J Oral Pathol Med. 1992;21(4):160-3. 205. Sridharan G, Shankar AA. Toluidine blue: A review of its chemistry and clinical utility. J Oral Maxillofac Pathol. 2012;16(2):251-5. 206. Pallagatti S, Sheikh S, Aggarwal A, Gupta D, Singh R, Handa R, et al. Toluidine blue staining as an adjunctive tool for early diagnosis of dysplastic changes in the oral mucosa. J Clin Exp Dent. 2013;5(4):e187-91. 207. Pizer M, Dubois D. An assessment of toluidine blue for the diagnosis of lip lesions. Va Med. 1979;106:860-2. 208. Myers ND, Brincks AM, Ames AJ, Prado GJ, Penedo FJ, Benedict C. Multilevel modeling in psychosomatic medicine research. Psychosom Med. 2012;74(9):925-36. 209. Rosenberg D, Cretin S. Use of meta-analysis to evaluate tolonium chloride in oral cancer screening. Oral Surg Oral Med Oral Pathol. 1989;67(5):621-7. 210. Shedd DP, Gaeta JF. In vivo staining of pharyngeal and laryngeal cancer. Arch Surg. 1971;102(5):442-6. 211. Portugal LG, Wilson KM, Biddinger PW, Gluckman JL. The role of toluidine blue in assessing margin status after resection of squamous cell carcinomas of the upper aerodigestive tract. Arch Otolaryngol Head Neck Surg. 1996;122(5):517-9. 212. Warnakulasuriya KA, Johnson NW. Sensitivity and specificity of OraScan (R) toluidine blue mouthrinse in the detection of oral cancer and precancer. J Oral Pathol Med. 1996;25(3):97-103. 213. Shedd DP, Hukill PB, Bahn S. In vivo staining properties of oral cancer. Am J Surg. 1965;110(4):631-4. 180 214. Shedd DP, Hukill PB, Bahn S, Farraro RH. Further appraisal of in vivo staining properties of oral cancer. Arch Surg. 1967;95(1):16-22. 215. Parakh MK, Jagat Reddy RC, Subramani P. Toluidine Blue Staining in Identification of a Biopsy Site in Potentially Malignant Lesions: A Case-control Study. Asia Pac J Oncol Nurs. 2017;4(4):356-60. 216. Epstein JB, Oakley C, Millner A, Emerton S, van der Meij E, Le N. The utility of toluidine blue application as a diagnostic aid in patients previously treated for upper oropharyngeal carcinoma. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 1997;83(5):537-47. 217. Mashberg A. Tolonium (toluidine blue) rinse--a screening method for recognition of squamous carcinoma. Continuing study of oral cancer IV. JAMA. 1981;245(23):2408-10. 218. Onofre MA, Sposto MR, Navarro CM. Reliability of toluidine blue application in the detection of oral epithelial dysplasia and in situ and invasive squamous cell carcinomas. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2001;91(5):535-40. 219. Epstein JB, Zhang L, Poh C, Nakamura H, Berean K, Rosin M. Increased allelic loss in toluidine blue-positive oral premalignant lesions. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2003;95(1):45-50. 220. Guo Z, Yamaguchi K, Sanchez-Cespedes M, Westra WH, Koch WM, Sidransky D. Allelic losses in OraTest-directed biopsies of patients with prior upper aerodigestive tract malignancy. Clin Cancer Res. 2001;7(7):1963-8. 221. Lane PM, Gilhuly T, Whitehead P, Zeng H, Poh CF, Ng S, et al. Simple device for the direct visualization of oral-cavity tissue fluorescence. J Biomed Opt. 2006;11(2):024006. 222. Awan KH, Patil S. Efficacy of Autofluorescence Imaging as an Adjunctive Technique for Examination and Detection of Oral Potentially Malignant Disorders: A Systematic Review. J Contemp Dent Pract. 2015;16(9):744-9. 223. Lalla Y, Matias MA, Farah CS. Assessment of oral mucosal lesions with autofluorescence imaging and reflectance spectroscopy. J Am Dent Assoc. 2016;147(8):650-60. 224. Yamamoto N, Kawaguchi K, Fujihara H, Hasebe M, Kishi Y, Yasukawa M, et al. Detection accuracy for epithelial dysplasia using an objective autofluorescence visualization method based on the luminance ratio. Int J Oral Sci. 2017;9(11):e2. 225. Awan KH, Morgan PR, Warnakulasuriya S. Evaluation of an autofluorescence based imaging system (VELscope™) in the detection of oral potentially malignant disorders and benign keratoses. Oral Oncol. 2011;47(4):274-7. 226. Farah CS, Kordbacheh F, John K, Bennett N, Fox SA. Molecular classification of autofluorescence excision margins in oral potentially malignant disorders. Oral Dis. 2017.[Epub ahead of print] 181 227. Petruzzi M, Lucchese A, Nardi GM, Lauritano D, Favia G, Serpico R, et al. Evaluation of autofluorescence and toluidine blue in the differentiation of oral dysplastic and neoplastic lesions from non dysplastic and neoplastic lesions: a cross-sectional study. J Biomed Opt. 2014;19(7):76003. 228. Macey R, Walsh T, Brocklehurst P, Kerr AR, Liu JL, Lingen MW, et al. Diagnostic tests for oral cancer and potentially malignant disorders in patients presenting with clinically evident lesions. Cochrane Database Syst Rev. 2015;(5):CD010276.  229. Bradley G, Odell EW, Raphael S, Ho J, Le LW, Benchimol S, et al. Abnormal DNA content in oral epithelial dysplasia is associated with increased risk of progression to carcinoma. Br J Cancer. 2010;103(9):1432-42. 230. Bremmer JF, Brakenhoff RH, Broeckaert MAM, Beliën JAM, Leemans CR, Bloemena E, et al. Prognostic value of DNA ploidy status in patients with oral leukoplakia. Oral Oncol. 2011;47(10):956-60. 231. Fabarius A, Hehlmann R, Duesberg PH. Instability of chromosome structure in cancer cells increases exponentially with degrees of aneuploidy.Cancer Genet Cytogenet. 2003;143(1):59-72. 232. Sperandio M, Brown AL, Lock C, Morgan PR, Coupland VH, Madden PB, et al. Predictive value of dysplasia grading and DNA ploidy in malignant transformation of oral potentially malignant disorders. Cancer Prev Res (Phila). 2013;6(8):822-31. 233. Torres-Rendon A, Stewart R, Craig GT, Wells M, Speight PM. DNA ploidy analysis by image cytometry helps to identify oral epithelial dysplasias with a high risk of malignant progression. Oral Oncol. 2009;45(6):468-73. 234. Zhang L, Rosin MP. Loss of heterozygosity: a potential tool in management of oral premalignant lesions? J Oral Pathol Med. 2001;30(9):513-20. 235. Giaretti W, Monteghirfo S, Pentenero M, Gandolfo S, Malacarne D, Castagnola P. Chromosomal Instability, DNA Index, Dysplasia, and Subsite in Oral Premalignancy as Intermediate Endpoints of Risk of Cancer. Cancer Epidemiol Biomarkers Prev. 2013;22(6):1133-41. 236. Nikitakis NG, Pentenero M, Georgaki M, Poh CF, Peterson DE, Edwards P, et al. Molecular markers associated with development and progression of potentially premalignant oral epithelial lesions: Current knowledge and future implications. Oral Surg Oral Med Oral Pathol Oral Radiol. 2018. 237. Smith J, Rattay T, McConkey C, Helliwell T, Mehanna H. Biomarkers in dysplasia of the oral cavity: A systematic review. Oral Oncol. 2009;45(8):647-53. 238. Nankivell P, Mehanna H. Oral Dysplasia: Biomarkers, Treatment, and Follow-up. Curr Oncol Rep. 2011;13(2):145-52. 182 239. Liu HW, Liu XW, Dong GY, Zhou J, Liu Y, Gao Y, et al. P16 Methylation as an Early Predictor for Cancer Development From Oral Epithelial Dysplasia: A Double-blind Multicentre Prospective Study. EBioMedicine. 2015;2(5):432-7. 240. Jorde LB. Medical genetics. 2nd rev. ed. St. Louis;Toronto: Mosby; 2000. 241. Scully C, Field JK, Tanzawa H. Genetic aberrations in oral or head and neck squamous cell carcinoma 2: chromosomal aberrations. Oral Oncol. 2000;36(4):311-27. 242. Kasamatsu A, Uzawa K, Usukura K, Koike K, Nakashima D, Ishigami T, et al. Loss of heterozygosity in oral cancer. Oral Science International. 2011;8(2):37-43. 243. Bettendorf O, Piffko J, Bankfalvi A. Prognostic and predictive factors in oral squamous cell cancer: important tools for planning individual therapy? Oral Oncol. 2004;40(2):110-9. 244.   Hanahan D, Weinberg RA. The hallmarks of cancer. Cell. 2000;100(1):57-70. 245.   Hanahan D, Weinberg RA. Hallmarks of Cancer: The Next Generation. Cell. 2011;144(5):646-74. 246. Mao L, Lee JS, Fan YH, Ro JY, Batsakis JG, Lippman S, et al. Frequent microsatellite alterations at chromosomes 9p21 and 3p14 in oral premalignant lesions and their value in cancer risk assessment. Nat Med. 1996;2(6):682-5. 247. van der Riet P, Nawroz H, Hruban RH, Corio R, Tokino K, Koch W, et al. Frequent loss of chromosome 9p21-22 early in head and neck cancer progression. Cancer Res. 1994;54(5):1156-8. 248. Califano J, van der Riet P, Westra W, Nawroz H, Clayman G, Piantadosi S, et al. Genetic progression model for head and neck cancer: implications for field cancerization. Cancer Res. 1996;56(11):2488. 249. el-Naggar AK, Hurr K, Batsakis JG, Luna MA, Goepfert H, Huff V. Sequential loss of heterozygosity at microsatellite motifs in preinvasive and invasive head and neck squamous carcinoma. Cancer Res. 1995;55(12):2656-9. 250. Emilion G, Langdon JD, Speight P, Partridge M. Frequent gene deletions in potentially malignant oral lesions. Br J Cancer. 1996;73(6):809-13. 251. Roz L, Wu CL, Porter S, Scully C, Speight P, Read A, et al. Allelic imbalance on chromosome 3p in oral dysplastic lesions: an early event in oral carcinogenesis. Cancer Res. 1996;56(6):1228. 252. Papadimitrakopoulou VA, Izzo J, Mao L, Keck J, Hamilton D, Shin DM, et al. Cyclin D1 and p16 alterations in advanced premalignant lesions of the upper aerodigestive tract: role in response to chemoprevention and cancer development. Clin Cancer Res. 2001;7(10):3127. 183 253. Government of British Columbia. Population Estimates. 2017. [Online]. Available from: https://www2.gov.bc.ca/gov/content/data/statistics/people-population-community/population/population-estimates. [cited 2018 May 27] 254. Miracca EC, Kowalski LP, Nagai MA. High prevalence of p16 genetic alterations in head and neck tumours. Br J Cancer. 1999;81(4):677-83. 255. Pershouse MA, El-Naggar AK, Hurr K, Lin H, Yung WK, Steck PA. Deletion mapping of chromosome 4 in head and neck squamous cell carcinoma. Oncogene. 1997;14(3):369-73. 256. Zhang L, Michelsen C, Cheng X, Zeng T, Priddy R, Rosin MP. Molecular analysis of oral lichen planus. A premalignant lesion? Am J Pathol. 1997;151(2):323. 257. Zhang LW, Cheng X, Li YH, Poh C, Zeng T, Priddy R, et al. High frequency of allelic loss in dysplastic lichenoid lesions. Lab Invest. 2000;80(2):233-7. 258. Patil S, Rao RS, Sanketh DS, Warnakulasuriya S. Lichenoid dysplasia revisited - evidence from a review of Indian archives. J Oral Pathol Med. 2015;44(7):507-14. 259.  Bondy SJ, Victor JC, Diemert LM. Origin and use of the 100 cigarette criterion in tobacco surveys. Tobacco Control. 2009;18(4):317-23. 260. Tomasetti C, Vogelstein B. Variation in cancer risk among tissues can be explained by the number of stem cell divisions. Science. 2015;347(6217):78-81. 261. Poh CF, Ng SP, Williams PM, Zhang L, Laronde DM, Lane P, et al. Direct fluorescence visualization of clinically occult high-risk oral premalignant disease using a simple hand-held device. Head Neck. 2007;29(1):71-6. 262. Lingen MW, Szabo E. Validation of LOH Profiles for Assessing Oral Cancer Risk. Cancer Prev Res. 2012;5(9):1075-7. 263. Lippman S, Ro JY, Batsakis JG, Lee JS, Fan YH, Hong WK, et al. Frequent microsatellite alterations at chromosomes 9p21 and 3p14 in oral premalignant lesions and their value in cancer risk assessment. Nat Med. 1996;2(6):682-5. 264. Lippman SM, Hong WK. Molecular markers of the risk of oral cancer. N Engl J Med. 2001;344(17):1323-6. 265. Sidransky D. Molecular genetics of head and neck cancer. Curr Opin Oncol. 1995;7(3):229-33. 266. Barrowman R, Koo K, Wiesenfeld D, Nastri A, McCullough M. Oral cancer in non-smokers, non-drinkers: a systematic review. Intl Jf Oral Maxillofac Surg. 2015;44:e34-e5. 267. Durr ML, Li D, Wang SJ. Oral cavity squamous cell carcinoma in never smokers: analysis of clinicopathologic characteristics and survival. Am J Otolaryngol. 2013;34(5):388. 184 268. Koch WM, Lango M, Sewell D, Zahurak M, Sidransky D. Head and neck cancer in nonsmokers: a distinct clinical and molecular entity. The Laryngoscope. 1999;109(10):1544-51. 269. Li R, Faden DL, Fakhry C, Langelier C, Jiao Y, Wang Y, et al. Clinical, genomic, and metagenomic characterization of oral tongue squamous cell carcinoma in patients who do not smoke: Clinical and genomic study of nonsmokers with oral tongue cancer. Head Neck. 2015;37(11):1642-9. 270. Morse DE, Psoter WJ, Baek LS, Eisenberg E, Cohen D, Cleveland D, et al. Smoking and drinking in relation to depressive symptoms among persons with oral cancer or oral epithelial dysplasia. Head Neck. 2010;32(5):578-87. 271. Pickering CR, Zhang J, Neskey DM, Zhao M, Jasser SA, Wang J, et al. Squamous cell carcinoma of the oral tongue in young non-smokers is genomically similar to tumors in older smokers. Clin Cancer Res. 2014;20(14):3842-8. 272. National Cancer Institute. Cigars: Health effects and trends. Monograph No. 9. Bethesda, Maryland: National Cancer Institute. 1998.. 273. Lim K, Moles DR, Downer MC, Speight PM. Opportunistic screening for oral cancer and precancer in general dental practice: results of a demonstration study. Br Dent J. 2003;194(9):497-502. 274. Allen NE, Beral V, Casabonne D, Kan SW, Reeves GK, Brown A, et al. Moderate Alcohol Intake and Cancer Incidence in Women. J Nat Cancer Inst. 2009;101(5):296-305. 275. Jaber MA. Oral epithelial dysplasia in non-users of tobacco and alcohol: an analysis of clinicopathologic characteristics and treatment outcome. J of Oral Sci. 2010;52(1):13-21. 276. Perry BJ, Zammit AP, Lewandowski AW, Bashford JJ, Dragovic AS, Perry EJ, et al. Sites of origin of oral cavity cancer in nonsmokers vs smokers: possible evidence of dental trauma carcinogenesis and its importance compared with human papillomavirus. JAMA Otolaryngol Head Neck Surg. 2015;141(1):5-11. 277. Tomasetti C, Li L, Vogelstein B. Stem cell divisions, somatic mutations, cancer etiology, and cancer prevention. Science. 2017;355(6331):1330-4. 278. Lippman SM, Sudbo J, Hong WK. Oral cancer prevention and the evolution of molecular-targeted drug development. J Clin Oncol. 2005;23(2):346-56. 279. Pitiyage G, Tilakaratne WM, Tavassoli M, Warnakulasuriya S. Molecular markers in oral epithelial dysplasia: review. J Oral Pathol Med. 2009;38(10):737-52. 280. Chimenos-Kustner E, Font-Costa I, Lopez-Lopez J. Oral cancer risk and molecular markers. Med Oral Patol Oral Cir Bucal. 2004;9(5):381-4; 77-80. 185 281. Burke W, Psaty BM. Personalized medicine in the era of genomics. JAMA.2007;298(14):1682-4. 282. Blackburn EH. Cancer Interception. Cancer Prev Res. 2011;4(6):787-92. 283. Su J, Barbera L, Sutradhar R. Do repeated assessments of performance status improve predictions for risk of death among patients with cancer? A population-based cohort study. Palliat Med. 2015;29(6):547-53. 284. Sutradhar R, Atzema C, Seow H, Earle C, Porter J, Barbera L. Repeated assessments of symptom severity improve predictions for risk of death among patients with cancer. J Pain Symptom Manage. 2014;48(6):1041-9. 285. Dranitsaris G, Molassiotis A, Clemons M, Roeland E, Schwartzberg L, Dielenseger P, et al. The development of a prediction tool to identify cancer patients at high risk for chemotherapy-induced nausea and vomiting. Ann Oncol. 2017;28(6):1260-7. 286. Futoma J, Sendak M, Cameron CB, Heller K. Predicting disease progression with a model for multivariate longitudinal clinical data.  Proceedings of the 1st Machine Learning for Healthcare Conference. 2016;56:42-54. . 287. Rock L, Rosin M, Zhang L, Shariati B, Laronde D. Molecular analysis and changes in toluidine blue over time: predicting malignant progression. J Dent Res. 2016;95(A):1729. 288. Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol. 1996;49(12):1373-9. 289. Jayatillake RV, Sooriyarachchi MR, Senarathna DLP. Adjusting for a cluster effect in the logistic regression model: an illustration of theory and its application. Journal of the National Science Foundation of Sri Lanka. 2011;39(3):211-8. 290. Maia HC, Pinto NA, Pereira Jdos S, de Medeiros AM, da Silveira EJ, Miguel MC. Potentially malignant oral lesions: clinicopathological correlations. Einstein (Sao Paulo). 2016;14(1):35-40. 291. Zhang X, Kim KY, Zheng Z, Bazarsad S, Kim J. Nomogram for risk prediction of malignant transformation in oral leukoplakia patients using combined biomarkers. Oral Oncol. 2017;72:132-9. 292. Prince V, Bellile EL, Sun Y, Wolf GT, Hoban CW, Shuman AG, et al. Individualized risk prediction of outcomes for oral cavity cancer patients. Oral Oncol. 2016;63:66-73. 293. Zarate AM, Brezzo MM, Secchi DG, Barra JL, Brunotto M. Malignancy risk models for oral lesions. Medicina Oral Patologia Oral Y Cirugia Bucal. 2013;18(5):E759-E65 186 294. Rock LD, Rosin MP, Zhang L, Chan B, Shariati B, Laronde DM. Characterization of epithelial oral dysplasia in non-smokers: First steps towards precision medicine. Oral Oncol. 2018;78:119-25. 295. Forman MS, Chuang SK, August M. The Accuracy of Clinical Diagnosis of Oral Lesions and Patient-Specific Risk Factors that Affect Diagnosis. J Oral Maxillofac Surg. 2015;73(10):1932-7. 296. Lingen MW, Kalmar JR, Karrison T, Speight PM. Critical evaluation of diagnostic aids for the detection of oral cancer. Oral oncology. 2008;44(1):10-22. 297. World Health Organiztion. Prevalence of tobacco smoking. 2015. [Online]. Available from: http://gamapserver.who.int/gho/interactive_charts/tobacco/use/atlas.html. [cited 2018 May 27]. 298. Canadian Partnership Against Cancer. Population health in Canada's largest cities: A cancer system performance spotlight report. Toronto, ON: Canadian Partnership Against Cancer; 2013. 299. Reid J, Hammond D, Rynard V, Madill C, Burkhalter R. Tobacco use in Canada: patterns and trends. 2017 Edition ed. Waterloo, ON: Propel Centre for Population Health Impact, University Waterloo; 2017. 300. Pierce JP, Messer K, White MM, Cowling DW, Thomas DP. Prevalence of Heavy Smoking in California and the United States, 1965-2007. JAMA. 2011;305(11):1106-12. 301. Guillaud M, Zhang L, Poh C, Rosin MP, MacAulay C. Potential use of quantitative tissue phenotype to predict malignant risk for oral premalignant lesions. Cancer Res. 2008;68(9):3099-107. 302. MacAulay C, Poh CF, Guillaud M, Williams PM, Laronde DM, Zhang L, et al. High throughput image cytometry for detection of suspicious lesions in the oral cavity. J Biomedical Opt. 2012;17(8). 303. Ferris RL. Immunology and Immunotherapy of Head and Neck Cancer. J Clin Oncol. 2015;33(29):3293-304. 304. Levingston CA, Young MRI. Local Immune Responsiveness of Mice Bearing Premalignant Oral Lesions to PD-1 Antibody Treatment. Cancers (Basel). 2017;9(6):62. 305. Ola O, Sedig K. The challenge of big data in public health: an opportunity for visual analytics. Online J Public Health Inform. 2014;5(3):223 306. Campbell JD, Mazzilli SA, Reid ME, Dhillon SS, Platero S, Beane J, et al. The Case for a Pre-Cancer Genome Atlas (PCGA). Cancer Prev Res (Phila). 2016;9(2):119-24.   187 Appendix A  Clinicopathological Data Collection Tools  A.1 Sample Oral Biopsy Service Pathology Report    188 A.2 Initial Questionnaire  189 190 191  192 A.3 Standardized Oral Map  193 A.4 Lesion Tracking Sheet   194    195 A.5 Oral Biopsy Service Requisition Form    196 Appendix B  Additional Publications The molecular laboratory techniques used in these projects are sensitive, and the recovery of nucleic acids from formalin-fixed paraffin-embedded (FFPE) is challenging and required much optimizing and troubleshooting. Although not presented formally as part of this thesis, a publication by Maraschin BJ, Silva VP, Rock L, Sun H, Visioli F, Rados PV, Rosin MP.  Optimizing fixation protocols to improve molecular analysis from FFPE tissues. Braz Dent J. 2017 Jan-Feb;28(1):82-84, resulted from this work. My role in this publication, in addition to optimizing and refining the molecular techniques, was to perform manuscript editing and review.   The publication of Chapter 3, “Dysplasia should not be ignored in lichenoid mucositis.” elicited a letter to the editor, and a response was invited. The invited publication, Zhang L, Rock LD, Rosin MP, Laronde DM. Lichenoid mucositis: The chicken or the egg? J Dent Res. 2018 Sept;97(10):1179, ensued from this work. My role in this publication was to perform manuscript editing and review.   

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.24.1-0372321/manifest

Comment

Related Items