1. NATIONAL INSTITUTES OF HEALTH: National Institute of Allergy and Infectious Diseases National Institute of Mental Health National Institute on Drug Abuse Choosing the Primary Endpoint for HIV prevention Trials; The example of HPTN071/PopART Helen Ayles, Sian Floyd, Ab Schaap, Anne Cori, Mike Pickles, Christophe Fraser, Deborah Donnell, Nulda Beyers, Sarah Fidler, Richard Hayes and the HPTN 071 (PopART) Team.

2. Population effect of universal testing and immediate ART therapy to Reduce HIV Transmission PopART: 2

3. Universal voluntary HIV testing with appropriate combination prevention offered to all those testing HIV negative - in addition to immediate ART for all those testing HIV positive - will have a substantial impact on HIV incidence at population level Hypothesis 3

4. Lancet 2009 373: 48-57 4

5. Not known whether a UTT intervention can be delivered with high uptake and acceptability Many uncertainties in model parameters Population-level impact of intervention package is not known Potential adverse effects such as sexual risk disinhibition, HIV-related stigma, overload of health services, toxicity, and drug resistance A rigorously designed trial can measure the costs and benefits of this strategy and provide reliable evidence on cost-effectiveness for health policy makers Why is a Trial Needed? 5

6. This trial was designed to ask –“Does a strategy of combination HIV prevention including universal HIV testing and treatment reduce HIV transmission (incidence) at community level?” Primary outcome was clearly incidence but how to measure? Choosing the Primary Endpoint 6

7. HIV incidence will be estimated by assessing HIV seroconversion in a longitudinal cohort (the PC) Advantages Gold standard approach for HIV incidence estimation Uses routine HIV test methods Provides interim and cumulative incidence estimates Cohort allows for measurement of other indicators such as sexual behaviour, HSV2 Disadvantages Requires longitudinal cohort follow-up –Impacted by loss-to-follow up, including differential loss to follow-up (e.g., of those at higher risk of HIV acquisition) –Hawthorne effect –Complex sampling is needed to ensure that the cohort reflects the population as a whole HIV incidence prior to enrollment may also be estimated using a multi-assay algorithm (cross-sectional incidence estimate) This approach was recently used for primary endpoint determination in a large, community- randomized clinical trial (NIMH Project Accept [HPTN 043]) Coates et al., Lancet Global Health 2014; 2:e267-277) Provides an estimate of HIV incidence in the months prior to study enrollment Measuring HIV Incidence

8. Design issues What should the combination prevention package contain? –HCT- universal uptake –Linkage to care and provision of ART –Sexual risk reduction –VMMC –PMTCT –STI –TB What scenarios would be useful to policy makers? –Universal test Vs Universal test and treat Vs current –Costs of each –Delivery under routine programmatic conditions as far as possible What effectiveness is possible? 8

9. Health centre VMMC facility Universal testing: annual door-to-door HBT Follow-up on referral Support for: - Retention in care - Adherence to treatment CHiPs: Community HIV-care Providers PMTCT: Prevention of Mother to Child Transmission VMMC: Voluntary Medical Male Circumcision TB: Tuberculosis STI: Sexually Transmitted Infections Service promotion and referral for - HIV care for HIV +ve including PMTCT - VMMC - TB - STI Universal treatment for HIV +ve irrespective of CD4 count Facilitated by CHiPs PopART Intervention Package

10. Final Trial Design Communities matched into 7 triplets on geographical area and HIV prevalence Average of ~50,000 in each cluster (~ 50% adults) Incidence measured in Population Cohort: 2,500 adults in each cluster, followed up after 1, 2 and 3 years 9 communities in South Africa 12 communities in Zambia 21 communities in 3 arms 10

11. Calculating sample size: The role of mathematical modelling Deterministic compartmental model of individuals aged 15+ Heterosexual mixing Three risk groups Cori et al. PLoS One, 2014 11

12. Model calibration: - to national HIV prevalence estimates from UNAIDS - and ART coverage data from Zamstar Best fit, national guidelines CD4<350, central target Zambia South Africa 12

13. What is the influence of process parameters? treatment drop-out/failure Relative reduction in 3-year HIV incidence in arms A and B Linear model efficacy of ART in blocking transmission uptake of testing, ART and circumcision % sex acts with partners from other communities Effect of counselling on infectivity Delays in linkage to care 13

14. uptake of testing, ART and circumcision treatment drop-out/failure efficacy of ART in blocking transmission Effect of counselling on infectivity Delays in linkage to care Arm A Arm B Zambia % sex acts with partners from other communities What is the influence of process parameters? 14

15. uptake of testing, ART and circumcision treatment drop-out/failure efficacy of ART in blocking transmission Effect of counselling on infectivity Delays in linkage to care Arm A Arm B Zambia R 2 >0.98 Contributions to variability in outcome % sex acts with partners from other communities efficacy of ART in blocking transmission uptake of testing, ART and circumcision 1 1 2 2 1 2 1 2 What is the influence of process parameters? 15

16. Summary : projected reduction in 3-year cumulative HIV incidence ZambiaSouth Africa Relative reduction Arm A Relative reduction Arm B Mean annual incidence Arm C Relative reduction Arm A Relative reduction Arm B Mean annual incidence Arm C Best fit, central target 61%25%1.85%62%26%1.36% 95% variability due to uptake level (pessimistic-optimistic targets) 42-75%15-33%1.85%44-75%16-33%1.36% Power calculations, best fit, central target (incidence 1.5% k 0.2): –A versus C: 100% –B versus C: 48% –A versus B: 96% 16

17. Impact of the new WHO guidelines: scenarios 1- 4 ; projected reduction in 3-year cumulative HIV incidence Scenario 1Scenario 2Scenario 3Scenario 4 Late adoption Early adoption Small eligibility Large eligibility Small eligibility Large eligibility 2 hypotheses regarding starting date: –Early adoption: 1 st January 2014 –Late adoption: 1 st January 2015 2 hypotheses regarding population affected by new guidelines: –Large eligibility: 90% testing and linkage to care in ANC & 30% HIV+ individuals with CD4>500 in a serodiscordant couple or co-infected with TB or Hep B –Small eligibility: 40% testing and linkage to care in ANC & 5% HIV+ individuals with CD4>500 in a serodiscordant couple or co-infected with TB or Hep B 17

18. Impact of the new WHO guidelines: scenarios 1- 4 ; projected reduction in 3-year cumulative HIV incidence ZambiaSouth Africa Rel. red. Arm A Rel. red. Arm B Mean annual incid. Arm C Rel. red. Arm A Rel. red. Arm B Mean annual incid. Arm C Best fit, central target 61% 59% 58% 56%* 25% 31% 32% 35% 37% 1.85% 1.75% 1.68% 1.67% 1.53% 62% 61% 60% 58%* 26% 32% 33% 37% 39% 1.36% 1.32% 1.28% 1.20% 95% variability due to intervention uptake level (pessimistic-optimistic targets) 42-75%15-33%1.85%44-75%16-33%1.36% 18 * Also validated by independent calculation based on Eaton et al., Lancet Global Health, 2014

19. Using sample size of 2500 per cluster, incidence over 3 years, Study power under new scenarios 19 HIV incidence rate/ 100py (control arm) Between- cluster coefficient of variation (k) Reduction arm A Reduction arm B Arm A vs Arm C Arm B Vs Arm C Arm A Vs Arm B 1.00.255%30%99%60%71% 1.00.260%35%100%71%77% 1.50.255%30%100%65%78% 1.50.260%35%100%80%83%

20. Conclusions HPTN071 will use a cohort measure of HIV incidence to assess the effectiveness of a package of combination HIV prevention including a “universal test and treat” approach Adoption of new consolidated WHO guidelines in Zambia and South Africa should only moderately affect ability to detect differences between arms in the HPTN-071 (PopART) trial Main trial outcome mostly depends on –Community-level changes in behaviours –Efficacy of ART in blocking transmission (adherence) –Uptake of HIV testing, treatment and circumcision All of these process measures are being actively measured during the trial 20

21. ACKNOWLEDGEMENTS Sponsored by the National Institute of Allergy and Infectious Diseases (NIAID) under Cooperative Agreements # UM1 AI068619, UM1-AI068617, and UM1-AI068613 Funded by: –The U.S. President's Emergency Plan for AIDS Relief (PEPFAR) –The International Initiative for Impact Evaluation (3ie) with support from the Bill & Melinda Gates Foundation –NIAID, the National Institute of Mental Health (NIMH), and the National Institute on Drug Abuse (NIDA) all part of the U.S. National Institutes of Health (NIH)

22. The HPTN 071 Study Team, led by: Dr. Richard Hayes Dr. Sarah Fidler Dr. Helen Ayles Dr. Nulda Beyers Implementing Partners: Government Agencies:

23. The HPTN 071 Study Team, led by: Dr. Richard Hayes Dr. Sarah Fidler Dr. Helen Ayles Dr. Nulda Beyers Implementing Partners: Government Agencies:

24. All research participants and their families The 21 research communities and their religious, traditional, secular and civil leadership structures Volunteers in the community advisory board structures With thanks to: 24