1. HPV genotype attribution of anal neoplasia in HIV-positive MSM: estimating the preventable fraction and disease misclassification Vikrant V. Sahasrabuddhe, MBBS, MPH, DrPH Hormonal and Reproductive Epidemiology Branch Division of Cancer Epidemiology and Genetics National Cancer Institute National Institutes of Health Rockville, MD, USA XIX International AIDS Conference, Washington, D.C. July 25, 2012

2. I have no conflicts of interest to declare

3. Co-Authors Vikrant V. Sahasrabuddhe 1, Philip E. Castle 2, Stephen Follansbee 3, Sylvia Borgonovo 3, Brandon J. LaMere 3, Diane Tokugawa 3, Teresa Darragh 4, Sean Boyle 5, Mark Sadorra 5, Scott Tang 5, Nicolas Wentzensen 1 1 National Cancer Institute, Rockville, MD, USA, 2 American Society for Clinical Pathology, Washington, DC, USA, 3 Kaiser Permanente Northern California, San Francisco, CA, USA, 4 University of California at San Francisco, San Francisco, CA, USA, 5 Roche Molecular Systems, Pleasanton, CA, USA

4. Background Reducing the burden of HPV-related cancers in HIV+ populations is an important global public health agenda HIV+ MSM are at exceedingly high-risk of HPV-induced anal cancer and its precursor- high-grade anal intraepithelial neoplasia (HGAIN)

5. Background Potential approaches to prevention of anal cancer in HIV+ MSM Primary prevention: HPV prophylactic vaccination Licensed vaccines: Bivalent HPV vaccine: HPV16/18 (Cervarix®, GSK) Quadrivalent HPV vaccine: HPV16/18/6/11 (Gardasil®, Merck) Investigational vaccine: Nonavalent HPV vaccine: HPV16/18/6/11/31/33/45/52/58 (V503, Merck) No efficacy trials to-date in HIV+ MSM Safety & immunogenicity data are reassuring Quadrivalent vaccine licensed for HIV+/MSM/immunocompromised males: ages 9 to 26 years Ref: CDC MMWR February 3, 2012 / 61(04);1-7

6. Secondary prevention: screening and treatment of HGAIN Screening by anal cytology (Pap) Diagnosis by high-resolution anoscopy + biopsy Treatment of HGAIN [e.g., by ablation/infrared coagulation] No consensus guidelines for anal cancer screening Clinicians in specialized centers offer anal cancer screening – strong similarities between anal and cervical cancer and their precursors & demonstrated success of cervical cancer prevention program – high proportion of HGAIN lesions can be treated using office-based procedures – anal screening and treatment in targeted populations shown to be cost- effective Ref: Palefsky 2012, Goldstone et al 2007, Stier et al 2008, Singh et al 2009, Fox et al 2010, Goldie et al 2000 Background Potential approaches to prevention of anal cancer in HIV+ MSM

7. Background Challenges in HPV clinical epidemiology studies in HIV+ MSM with a bearing on clinical practice – Anal cytology and high-resolution anoscopy are sub- optimally sensitive  how to minimize disease misclassification? – Multiple HPV infections and multiple carcinogenic HPV infections are very common  how to attribute etiologic fraction to individual HPV genotypes?

8. Study Objectives – Characterization of HPV genotype distribution across AIN categories in HIV+ MSM – Unsupervised hierarchical clustering to explore histology-cytology disease groups based on distribution of carcinogenic HPV genotypes – Attribution modeling to estimate fractions of HGAIN preventable by HPV vaccination

9. Methods Study design – Cross-sectional study at Kaiser Permanente Northern California (KPNC)-Anal Cancer screening clinic in San Francisco, California Study population – n=363 HIV+ MSM participating in KPNC HMO plan Study investigations – Anal sample by wetted, flocked nylon swab HPV genotyping: Roche Linear Array HPV (LA-HPV) PCR assay Anal cytology – Digital anorectal examination – High resolution anoscopy (HRA) with anal biopsy (histology)

10. Results Participant characteristics HPV results available342/363 Demographic & behavioral Age (Mean; +SD)53 (+9.2) Age at first anal sex (Mean; +SD)21 (+6.4) Proportion with 40+ lifetime anal sexual partners39.1% Proportion with 3+ anal partners in last 6 months15.1% Condom use during anal sex80.6% Ever smoker55.2% Clinical Current CD4+ <350/µl19.1% Current CD4+ <200/µl3% HIV-1 Viral load <75 copies/ml90% Taking cART93.4%

11. Results HPV genotype prevalence and distribution (n=342) High prevalence of HPV genotypes & carcinogenic HPV genotypes – Any HPV: 94.4% – Any carcinogenic HPV: 75.4% – 36 of 37 detectable genotypes present High proportion with multiple genotypes & multiple carcinogenic genotypes – Multiple HPV genotypes: 85.4% – Multiple carcinogenic HPV genotypes: 47.4% ‘Carcinogenic’ HPV: HPV16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 68; Ref: Bouvard et al 2009 Lancet Oncol

12. Results Cytology & HRA-Histology Cytology NILM = 154 ASC-US = 33 LSIL = 80 ASC-H = 13 HSIL-AIN2 = 19 HSIL-AIN3 = 43 HRA-Histology Normal HRA = 92 Normal Histology =81 AIN1 =106 AIN2 = 41 AIN3 = 22 AIN: Anal intraepithelial neoplasia; NILM: Negative for intraepithelial lesions or malignancy ASC-US: Atypical squamous cells of undetermined significance ASC-H: Atypical-squamous cells cannot exclude high-grade squamous intraepithelial lesion LSIL: low-grade squamous intraepithelial lesions ; HSIL: high-grade squamous intraepithelial lesions ; HSIL-AIN2: HSIL-favor AIN2 ; HSIL-AIN3: HSIL-favor AIN3 Normal HRA: No Biopsy taken; NILM includes cases with no cytology results were available but for which HRA/histology results were available

13. Methods Unsupervised hierarchical clustering Categorization: Separated participants into 11 anal disease categories by crossing cytology/HRA/histology results Clustering: Simultaneous clustering of both disease combinations and HPV genotypes – Complete linkage and Euclidean distance matrix – Software: Cluster 3.0 for clustering of HPV frequency arrays Java TreeView for heatmap and dendrogram visualization

14. Results Heatmap & Dendrograms: Unsupervised Hierarchical Clustering

15. Results Clusters of Anal Disease Categories by Unsupervised Hierarchical Clustering Cluster 1 Cluster 3 Cluster 2

16. Cytology HRA/Histology NILMASC-USLSILASC-HHSIL-AIN2HSIL-AIN3 Normal HRA 61196 56 Normal Histology 50205 AIN1 Histology 334912 AIN2 Histology 31 AIN3 Histology Results Disease ascertainment for genotype attribution: Creation of composite cytology-histologic endpoints (n=342)

17. Cytology HRA/Histology NILMASC-USLSILASC-HHSIL-AIN2HSIL-AIN3 Normal HRA ‘No AIN’ AIN1 121 (35.4%) AIN2 54 (15.8%) AIN3 56 (16.4%) Normal Histology 111 (32.4%) AIN1 Histology AIN2 Histology AIN3 Histology HGAIN (AIN2 + AIN3) prevalence = 110 out of 342 = 32.2% (95%CI: 27.4-37.3)

18. Results Composite disease categories and HPV prevalence Total‘No AIN’AIN1AIN2AIN3HGAIN 3421111215456110 Any HPV323 (94.4%) 99 (89.2%) 115 (95%) 53 (98.1%) 56 (100%) 109 (99%) Carcinogenic HPV 258 (75.4%) 57 (51.4%) 95 (78.5%) 51 (94.4%) 55 (98.2%) 106 (96.4%) HPV1696 (28.1%) 10 (9.0%) 29 (24.0%) 22 (40.7%) 35 (62.5%) 57 (51.8%) Multiple Carcinogenic HPV 162 (47.4%) 29 (26.1%) 57 (47.1%) 36 (66.7%) 40 (71.4%) 76 (69.1%) ‘Carcinogenic’ HPV: HPV16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 68 All p for trend<0.01 (‘No AIN’ to AIN3)

19. Methods Attributing HGAIN fraction to individual HPV genotypes Low-end estimate – Prevalence in cases with single carcinogenic HPV genotype Irrespective of other non/possible carcinogenic genotypes High-end estimate – Prevalence in HGAIN cases with any HPV genotype Irrespective of multiple genotypes Hierarchical attribution model estimate – HPV genotypes are attributed to cases by hierarchical frequency of carcinogenic genotypes in HGAIN

20. Methods Attribution modeling: assumptions and caveats HPV vaccination would happen before exposure to HPV genotypes included in the vaccines Cross-genotype protection and synergy between genotypes not considered Full immune response to HPV VLP vaccines

21. Results Estimation models of HPV vaccine preventable fractions of HGAIN (n=110) Bivalent vaccine (16/18) 36.7% (95%CI: 21.9-54.5) 56.4% (95%CI: 47.0-65.3) 61.8% (95%CI: 52.5-70.3) Quadrivalent vaccine (16/18/6/11) 40.0% (95%CI: 24.6-57.7) 56.4% (95%CI: 47.0-65.3) 70.0% (95%CI: 60.9-77.8) Nonavalent vaccine (16/18/6/11/31/33/45/52/58) 86.7% (95%CI: 70.3-94.7) 89.1% (95%CI: 81.9-93.7) 92.7% (95%CI: 86.3-96.3)

22. Conclusions Genotype attribution models suggest that HPV prophylactic vaccines may potentially prevent a large fraction of HGAIN in HIV-positive MSM HGAIN is clinically & etiologically heterogeneous in HIV-positive MSM HPV16 is the most common and etiologically dominant genotype in HIV- positive MSM Combining cytology and histology endpoints may improve disease ascertainment in cross sectional studies in HIV-positive MSM Molecular characterization by HPV genotyping or biomarkers may reduce misclassification of HGAIN in HIV-positive MSM

23. Future directions Methodology: Analytical framework needs to be replicated and for use in evaluation and comparison of attribution fractions and disease ascertainment across populations and across disease sites. Informing prevention clinical trials: Model results can be used to inform the design of future efficacy trials of HPV vaccines in HIV+ individuals and trials of best approaches to anal cancer screening. Prospective cohort studies: Risk of HGAIN associated with clusters of anal disease categories and HPV genotypes in HIV- positive MSM needs to be evaluated prospectively.

24. Acknowledgments NCI/NIH Nicolas Wentzensen Louise Brinton Katherine McGlynn Lauren Schwartz Lauren Wilson Mark Sherman Saloni Nayar Jennifer Loukissas ASCP Philip Castle UCSF Teresa Darragh Joel Palefsky KPNC Steven Follansbee Roche Sean Boyle Vanderbilt Sten Vermund *Study participants Funding: Intramural Research Program of the National Cancer Institute (NCI), National Institutes of Health (NIH), Department of Health and Human Services (DHHS) Oak Ridge Institute of Science and Education (ORISE) Research Participation Program

25. Thank you for your attention

26. Extra slides

27. Results Distribution of cytology-HRA-histology (n=342) Cytology HRA/Histology NILM 1 ASC-USLSILASC-HHSIL-AIN2HSIL-AIN3 Normal HRA 2 61811426 Normal Histology 50515326 AIN1 Histology 3313364812 AIN2 Histology 76132310 AIN3 Histology 315049 1 Category includes cases with no cytology results available, for which HRA/histology results were available 2 Normal HRA results indicate that no Biopsy was taken Normal HRA: No Biopsy taken AIN: Anal intraepithelial neoplasia NILM: Negative for intraepithelial lesions or malignancy ASC-US: Atypical squamous cells of undetermined significance ASC-H: Atypical-squamous cells cannot exclude high-grade squamous intraepithelial lesion LSIL: low-grade squamous intraepithelial lesions HSIL: high-grade squamous intraepithelial lesions HSIL-AIN2: HSIL-favor AIN2 HSIL-AIN3: HSIL-favor AIN3

28. Results High prevalence of multiple HPV genotypes across AIN disease categories All HPV genotypes Median: 4 Range: 0-14 IQR: 2-6 Carcinogenic HPV genotypes Median: 1 Range: 0-7 IQR: 1-3 62.2 % 86 % 94.4 % 92.9 % 26.1 % 47.1 % 62.2 % 71.4% Multiple types %

29. Results Clusters of Anal Disease by Unsupervised Hierarchical Clustering Cytology HRA/Histology NILMASC-USLSILASC-HHSIL-AIN2HSIL-AIN3 Normal HRA Normal Histology AIN1 Histology AIN2 Histology AIN3 Histology Cluster 3 Cluster 1 Cluster 2 Cluster 1{ASC-H/HSIL-AIN2 cytology and abnormal HRA} + {AIN2+ histology} Cluster 2{<=LSIL cytology and <=AIN1 histology} + {NILM cytology/normal HRA} Cluster 3{Normal HRA + <=ASC-H/HSIL-AIN2cytology}

30. Results Participant characteristics by dendrogram clusters of anal disease Dendrogram clusterCluster 1Cluster 2Cluster 3 Mean Age (years)535553 HPV +99.0%93.5%92.5% Carcinogenic HPV+96.9%80.0%64.8% Multiple (2+) Carcinogenic HPV+71.4%58.1%34.7% HPV1655.1%12.9%17.8%

31. Results Differences in HPV genotypes prevalence in HGAIN by CD4 counts CD4<350 (n=63) CD4>=350 (n=266) p-value Any HPV100%98.7%0.99 Any Carcinogenic HPV100%94.9%0.57 HPV1670.4%43.6%0.03 Bivalent HPV vaccine genotypes 77.8%55.1%0.04 Quadrivalent HPV vaccine genotypes 81.5%61.1%0.15 Nonavalent HPV vaccine genotypes 92.6%92.3%0.99 Bivalent HPV vaccine genotypes: HPV16, 18 ; Quadrivalent HPV vaccine genotypes: HPV 16, 18, 6, 11; Nonavalent HPV vaccine genotypes: HPV16, 18, 6, 11, 31, 33, 45, 52, 58

32. Cluster 1 Cluster 2 Results Clusters of Carcinogenic HPV genotypes by Unsupervised Hierarchical Clustering HPV16 Non-16 carcinogenic HPV

33. Results Comparison of main findings with other published research Evidence from this study in HIV+ MSM Pooled evidence in HIV+ MSM (Machelek et al 2012 Lancet Oncol) Carcinogenic HPV prevalence 75.4% (95%CI: 70.6-79.7) 73·5 % (95%CI: 63·9–83·0) HGAIN prevalence32.2% (95%CI: 27.4-37.3) (Composite diagnosis) 29·1% (95%CI: 22·8–35·4) (Variable methods of diagnosis) Range of preventable fraction of HGAIN from this study in HIV+ MSM Vaccine trial efficacy against HGAIN in HIV- MSM (Palefsky et al 2011 NEJM) Prevention by quadrivalent HPV vaccine (16/18/6/11) 70.0% (95%CI: 60.9-77.8) (Hierarchical attribution model) 74.9% (95%CI: 8.8 to 95.4) (Per-protocol efficacy population)