1. Injectable contraception and HIV-1 risk in women in HIV-1 serodiscordant partnerships: persistence of effect in multiple sensitivity analyses Renee Heffron 1, Deborah Donnell 2, Helen Rees 3, Connie Celum 1, Nelly Mugo 4, Edwin Were 5, Guy de Bruyn 3, Edith Nakku-Joloba 6, Kenneth Ngure 5, James Kiarie 5, Robert Coombs 1 and Jared Baeten 1 for the Partners in Prevention HSV/HIV Transmission Team from 1 University of Washington, Seattle, WA, USA; 2 Fred Hutchinson Cancer Research Center, Seattle, WA, USA; 3 University of the Witswatersrand, Johannesburg, South Africa; 4 Kenyatta National Hospital, Nairobi, Kenya; 5 Moi University, Eldoret, Kenya; 6 Makere University College of Health Sciences, Kampala, Uganda

2. Background Hormonal contraceptives are widely used Some epidemiologic and laboratory studies have suggested that hormonal contraceptives may alter HIV-1 susceptibility in women Results across studies have been inconsistent

3. In October 2011, we published an analysis from a prospective cohort study of African HIV-1 serodiscordant couples where injectable contraceptives were associated with a doubling of the risk for HIV-1 acquisition (adjusted hazard ratio 2.05, p=0.04) WHO has recently recommended that women using injectable contraceptives be especially counseled to use condoms for dual protection Results from observational studies of hormonal contraception and HIV- 1 risk can be challenging to interpret; associations could be explained by behavioral and/or biological factors Background

4. Objective Perform sensitivity analyses to explore the effect of analytic assumptions on our primary results Analyses include: – Adjusting for sexual behavior using alternative and additional factors – Reduce possible exposure misclassification – Isolate effect of DMPA injectable contraception from other injectable contraceptives

5. Study population Prospective cohort study of 3790 HIV-1 serodiscordant couples from East and southern Africa Couples recruited as part of 2 studies conducted between 2004 and 2010 – Partners in Prevention HSV/HIV Transmission Study Randomized trial of acyclovir herpes suppression to reduce HIV-1 transmission (n=3321) 1 – Couples Observational Study Prospective cohort study of immune correlates of HIV-1 protection (n=469) 1 Celum et al. NEJM 2010 5

6. Participants ≥18 years old and sexually active HIV-1 infected partners not eligible, at enrollment, for ART, under national guidelines For HIV-1 negative partners, HIV-1 testing done quarterly; for HIV-1 positive partners, CD4 counts measured every 6 months and plasma viral load measured at enrollment and 6 months later Contraceptive use and sexual behavior measured quarterly with standardized questionnaires – We did not collect data on adherence or brand of contraception used Methods 6

7. Primary statistical model Multivariate Cox proportional hazards model Adjusted for Age Plasma viral load Sex without a condom (time dependent) Pregnancy (time dependent) 7

8. Multivariate** Cox Proportional hazards regression analysis Incidence rate* HR (95% CI)p-value No hormonal contraception3.781.00 Any hormonal contraception6.61 1.98 (1.06-3.68) 0.03 Injectable6.85 2.05 (1.04-4.04) 0.04 Oral5.94 1.80 (0.55-5.82) 0.33 *per 100 person years **Adjusted for age, enrollment plasma viral load level of the HIV-1 infected partner and time dependent any unprotected sex and pregnancy. 8 Primary analysis results Lancet Infectious Diseases, 2012

9. Sensitivity analyses Minimize confounding by sexual behavior 1.Include additional sexual behavior covariate – sexual frequency 2.Include alternate sexual behavior covariate – male report of unprotected sex 3.Restrict to periods with unprotected sex Reduce misclassification among women who switched hormonal contraception during follow up 4.Restrict to periods prior to a hormonal contraceptive switch Isolate the effect of DMPA from other injectable contraceptives 5.Restrict to consistent non-South African injectable users (consistent DMPA users)

10. Sensitivity analysis results Primary analysis results HR (95% CI)p-value Injectable vs. no hormonal contraception 2.05 (1.04-4.04)0.04 % of primary analysis follow up time included

11. Adjusting for additional sexual behavior factors – number of sex acts Primary analysis results HR (95% CI)p-value Injectable vs. no hormonal contraception 2.05 (1.04-4.04)0.04 Multivariate** Cox Proportional hazards regression analysis HR (95% CI)p-value No hormonal contraception1.00 Injectable 2.06 (1.04-4.07) 0.04 **Adjusted for age, enrollment plasma viral load level of the HIV-1 infected partner and time dependent report of number of sex acts, any unprotected sex and pregnancy. % of primary analysis follow up time included 100%

12. Adjusting for alternative sexual behavior factors – male report of unprotected sex Primary analysis results HR (95% CI)p-value Injectable vs. no hormonal contraception 2.05 (1.04-4.04)0.04 Multivariate** Cox Proportional hazards regression analysis HR (95% CI)p-value No hormonal contraception1.00 Injectable 2.03 (0.95-4.32) 0.07 **Adjusted for age, enrollment plasma viral load level of the HIV-1 infected partner and time dependent report of pregnancy and the male partner’s report of any unprotected sex. % of primary analysis follow up time included 98.8%

13. In the subgroup of women who reported unprotected sex Primary analysis results HR (95% CI)p-value Injectable vs. no hormonal contraception 2.05 (1.04-4.04)0.04 Multivariate** Cox Proportional hazards regression analysis HR (95% CI)p-value No hormonal contraception1.00 Injectable 2.29 (0.70-7.53) 0.17 **Adjusted for age, enrollment plasma viral load level of the HIV-1 infected partner and time dependent report of any unprotected sex and pregnancy. % of primary analysis follow up time included 9.4%

14. Restricting to periods prior to hormonal contraceptive method switch Primary analysis results HR (95% CI)p-value Injectable vs. no hormonal contraception 2.05 (1.04-4.04)0.04 Multivariate** Cox Proportional hazards regression analysis HR (95% CI)p-value No hormonal contraception1.00 Injectable 2.62 (0.93-7.33) 0.07 **Adjusted for age, enrollment plasma viral load level of the HIV-1 infected partner and time dependent report of any unprotected sex and pregnancy. % of primary analysis follow up time included 75.3%

15. Restrict to consistent non-South African injectable users (consistent DMPA users) Primary analysis results HR (95% CI)p-value Injectable vs. no hormonal contraception 2.05 (1.04-4.04)0.04 Multivariate** Cox Proportional hazards regression analysis HR (95% CI)p-value No hormonal contraception1.00 DMPA 3.39 (1.38-11.22) 0.01 **Adjusted for age, enrollment plasma viral load level of the HIV-1 infected partner and time dependent report of any unprotected sex and pregnancy. % of primary analysis follow up time included 68.1%

16. Summary An approximate 2-fold HIV-1 risk with injectable contraceptive use persists in multiple sensitivity analyses – Some analyses had diminished power for precision due to decreased sample size but the adjusted HR always indicated a trend towards increased risk

17. Conclusions The benefits of injectable contraceptives are unequivocal and must be balanced with the potential risk for HIV-1 infection More high quality studies of hormonal contraceptives and HIV-1 risk are needed Integration of reproductive health and HIV-1 prevention programs is extremely important

18. Acknowledgements Partners in Prevention HSV/HIV Transmission Study Team University of Washington Coordinating Center and Central Laboratories - Seattle, WA Connie Celum, Anna Wald, Jairam Lingappa, Jared Baeten, Mary Campbell, Lawrence Corey, Robert Coombs, James Hughes, Amalia Magaret, M.Juliana McElrath, Rhoda Morrow, James Mullins Site Principal Investigators Botswana : Max Essex, Joseph Makhema Kenya : Elizabeth Bukusi, Kenneth Fife, James Kiarie, Nelly Rwamba Mugo, Edwin Were, Craig Cohen, Carey Farquhar, Grace John-Stewart Rwanda : Etienne Karita, Kayitesi Kayitenkore, Susan Allen South Africa : David Coetzee, Guy de Bruyn, Sinead Delany- Moretlwe, Glenda Gray, James McIntyre, Helen Rees Tanzania : Rachel Manongi, Saidi Kapiga Uganda : Elly Katabira, Allan Ronald Zambia : Mubiana Inambao, William Kanweka, Bellington Vwalika, Susan Allen Study participants and staff IAS 2012 International Scholarship Programme NIH (R03 HD068143 and T32 AI007140) Bill & Melinda Gates Foundation