1. 1 Case Studies in Prevention Trial Design Deborah Donnell, PhD SCHARP Vaccine and Infectious Disease Institute Fred Hutchinson Cancer Research Center

2. 2 Case Studies HPTN035 – Microbicide 4 arm HPTN039 – HSV-2 Acyclovir HPTN052 – Discordant couples HAART HPTN058 – IDU treatment Suboxone Tom Fleming – University of Washington Jim Hughes – University of Washington Barbra Richardson – University of Washington Ben Mâsse – FHCRC Ying Chen - FHCRC

3. 3 “How can we be so different and feel so much alike?” mused Flitter “And how can we feel so different and be so much alike?” wondered Pip.

4. 4 Selection criteria Randomized Clinical Trials HIV seroconversion endpoint Approved and in the field Biomedical interventions Long-term adherence for on-going risk

5. 5 HPTN035: Microbicide Trial Population: Women at sexual risk for HIV 4 arms: gel arms blinded ActiveInactive Arm 1: Pro 2000Arm 3: Placebo Gel vs. vs. Arm 2: BufferGel Arm 4: Condom Only Intended Mechanism: Microbicide = “Kill the microbe” Protect/enhance natural defenses Hypothesis: Active vs. Placebo Gel reduces HIV infection by 33% Active Gel vs. condom only reduces HIV infection by 33%

6. 6 HPTN039 : Acyclovir to prevent HIV acquisition Population: HSV-2+, HIV uninfected Women (in Southern Africa) MSM (in Americas) Two arms, double blinded Arm1: Acyclovir twice daily Arm 2: Placebo twice daily Intended mechanism: HSV-2 a suspected cofactor in HIV acquisition HSV-2 a suspected cofactor in HIV acquisition Daily Acyclovir reduces herpes activation reduced lesions and/or lymphocyte activation reduced lesions and/or lymphocyte activation reducing risk of HIV acquisition reducing risk of HIV acquisitionHypothesis: Daily Acyclovir will reduce HIV acquisition by 50%

7. 7 HPTN052: Delayed vs. Immediate HAART in Discordant Couples Population: Discordant couples (index HIV infected; partner HIV uninfected) with CD4 350-550 in index Two arm; unblinded Arm 1: Immediate HAART for index Arm 2: HAART initiated when clinically indicated for index Mechanism: Immediate therapy Potential benefits Partner: Lower risk of early transmission Index : More sustained virologic response Lower risk of irreversible CD4 drop Lower risk of irreversible CD4 dropHypothesis: Index on immediate vs. delayed strategy reduces HIV transmission to partner by 37%. Potential risks Partner : Transmission of resistant virus Partner : Transmission of resistant virus Index : Longer course of ART may Index : Longer course of ART may reduce treatment options reduce treatment options Poorer adherence/high resistance Poorer adherence/high resistance

8. 8 HPTN058: Suboxone treatment in opiate injectors Targeted population: Active opiate injectors (China and Thailand) Two arm, unblinded Arm 1: One year of Suboxone substitution therapy + one year counseling support + one year counseling support Arm 2: Suboxone detoxification + one year counseling support + one year counseling support Intended Mechanism: Biomedical + Behavioral Biomedical support for cessation of injection drug use Counseling support for long term behavior change Hypothesis: One year of Suboxone treatment and counseling will reduce death and HIV infection rate at 2 years by 42%

9. 9 ComparisonsComparisons 1.Proof of Concept vs. Efficacy trial 2.Effect size 3.Adherence requirements 4.Durability of effect

10. 10 Proof of Concept vs Effectiveness (aka Phase IIb vs Phase III) HPTN035 – Microbicide Phase II/IIb? ? HPTN039 – Acyclovir Phase III90% 2.5% HPTN052 – HAART strategy Phase III90 % 2.5% HPTN058 – Suboxone for IDU Phase III90% 2.5% PowerSize

11. 11 What is a Phase IIb design, anyway, and when would you use it?

12. 12 Goal of a clinical trial Obtain a reliable answer to a clinically relevant question Clinically relevant improvement in clinically relevant endpoint Definitive (Phase III) vs. Proof of concept (Phase IIb)

13. 13 Phase III 0%33%

14. 14 Phase III 0%33% 2.5%

15. 15 Phase III 0%33%22% 10% 256 endpoints 2.5%

16. 16 Why consider a Phase IIb ? No reliable information about efficacy risk of taking too long, using too many resources on something that may not work, or something in early development: …take a two step approach to establish proof of concept Phase IIb questions: 1.Does this look like it might work? 2.Does this not work at all ? 3.Does this work better than we hoped?

17. 17 Phase III Phase IIb 0% 15% 33%22% 33% Fleming TR, Richardson BA, JID 2004. 256 endpoints 96 endpoints

18. 18 Phase III Phase IIb 0% 15% 33%22% 33% 52% 12% 20% 256 endpoints 96 endpoints 10% 2.5%

19. 19 Phase III Phase IIb 0% 15% 33%22% 33% 52% 256 endpoints 96 endpoints 10% 2.5% 10% 2.5%

20. 20 Phase III Phase IIb 0% 15% 33%22% 33% 52% 256 endpoints 96 endpoints

21. 21 Factors in play for Phase IIb vs. Phase III: Risk and Resources Maturity/Definitiveness of the intervention and its intended mechanism Microbicide vs. Acyclovir; Suboxone; HAART State of results in the field Success in other settings: perinatal vs. microbicides Effectiveness on intended mediators; surrogates Concurrent trials Resources Participants, staff and money to conduct a high quality trial

22. 22 Characteristics of Phase IIb designs High probability of deciding an intervention not promising is discarded early Have 80% probability, if it doesn’t work at all that we would reject it High probability of not rejecting a promising intervention Have a 88% probability, if it works at a clinically relevant level, that we would not reject it Alternative and null from Phase III trial “Promising” and “Not-promising”

23. 23 ComparisonsComparisons 1.Proof of Concept vs. Efficacy trial 2.Effect size 3.Adherence requirements 4.Durability of effect

24. 24 Targeted Effectiveness in HIV Prevention HPTN035 – Microbicide33% vs. 0% Phase II/IIb HPTN039 – Acyclovir49% vs. 0% Phase III HPTN052 – HAART strategy37% vs. 0% Phase III HPTN058 – Suboxone for IDU42% vs. 0% Phase III 3100 96 [256] 3300 93 1750 cpls ~188 1500 ~132 Number of people Number of Events Targeted Effectiveness

25. 25 Targeted Effectiveness in HIV Prevention What is plausible ? HPTN035 – Microbicide 33% vs. 0% Phase II/IIb HPTN039 – Acyclovir 49% vs. 0% Phase III HPTN052 – HAART strategy 37% vs. 0% Phase III HPTN058 – Suboxone for IDU 42% vs. 0% Phase III No evidence of efficacy Time off drug in pregnancy Condom use and adherence Highly effective in treating HSV2 Time off drug in pregnancy Proven to reduce viral load Concept proven in perinatal trials Crossover to HAART Effective in treating opiate dependence HIV: 50% reduction Death: 25% effective reduction Targeted Effectiveness

26. 26 Targeted Effectiveness in HIV Prevention Is it relevant for implementation? HPTN035 – Microbicide33% Phase II/IIb HPTN039 – Acyclovir49% Phase III HPTN052 – HAART strategy37% Phase III HPTN058 – Suboxone for IDU42% Phase III Low cost, low tech, low risk Woman controlled Low cost Lifetime prophylaxis High cost/infrastructure needs Significant side effects High cost/infrastructure needs Targeted Effectiveness

27. 27 ComparisonsComparisons 1.Proof of Concept vs. Efficacy trial 2.Effect size 3.Adherence requirements 4.Durability of effect

28. 28 Burden of Adherence HPTN035 – Microbicide Phase II/IIb HPTN039 – Acyclovir Phase III HPTN052 – HAART strategy Phase III HPTN058 – Suboxone for IDU Phase III With each sex act 12 – 30 mo Twice daily pill12 – 18 mo Daily HAART regimen5 – 7 yrs Three times/week (DOT)1 yr Duration Adherence

29. 29 Best Achievable Adherence for Effectiveness HPTN035 – Microbicide Phase II/IIb HPTN039 – Acyclovir Phase III HPTN052 – HAART strategy Phase III HPTN058 – Suboxone for IDU Phase III 83-85% adherence With each sex act 83-85% adherence 87-92% retention 87-92% retention 90-95% adherence Twice daily pill 90-95% adherence 91-96% retention 91-96% retention Daily HAART regimenNA Three times/week (DOT)NA Self reported Adherence Targeted Adherence

30. 30 Effectiveness vs. Efficacy (a.k.a. “Intent to Treat”(ITT) vs. “Per Protocol” (PP)) Biological efficacy (Theoretical: Does the intended mechanism, when delivered, work?) vs. Effectiveness (Does the intervention, with best achievable adherence, as delivered in clinical trial, work?) vs. Implemented (Real world: Does the intervention, when made available to people, work)

31. 31 Illustration: Microbicides Biological Efficacy: Unprotected vaginal sex with HIV infected man Biological Efficacy (“Per protocol”) Woman supplied with gel, with monthly counseling support for use of condoms and gel, who uses gel consistently. Effectiveness (ITT) Woman offered an ongoing supply of gel, with monthly counseling support for use of condoms and gel.

32. 32 Randomization and Effectiveness Randomized trials are the gold standard for objective evidence Effectiveness is the most relevant prevention measure Treatment Placebo Balanced in Risk factors: measured and unmeasured R

33. 33 Post-Randomization Selection Bias Treatment Placebo Adheren t Risk factors: measured and unmeasured Treatment Placebo Treatment Placebo Non- Adheren t R

34. 34 The Experiment Conducted Determine people who are eligible Randomize to each arm: experimental and control Follow all participants High retention Highest achievable adherence Drug withheld for Pregnancy, AEsDrug withheld for Pregnancy, AEs Objective assessment of the results of the experiment includes all the follow-up in both arms References

35. 35 Examples of ITT vs. Per Protocol analysis MIRA (July 2007) Intent to treat: RR = 1.05 (0.84, 1.32) Intent to treat: RR = 1.05 (CI: 0.84, 1.32) Per-protocol (last sex act): RR = 0.90 (0.68, 1.17) Per-protocol (last sex act): RR = 0.90 (CI: 0.68, 1.17) Per-protocol (consistent use): RR = 0.83 (0.61,1.14) Per-protocol (consistent use): RR = 0.83 (CI: 0.61,1.14) Acyclovir trial in Mwanza (Aug 2007) Intent to treatRR = 1.02 P-P (Pregnancy censored)RR = 1.08 (CI: 0.64-1.83) P-P (90% adherent) RR = 0.58 (CI: 0.25-1.38) STEP trial (Nov 2007) STEP trial (Nov 2007) ITT: Vaccine vs. Placebo24 vs. 21 Per-protocol: 19 vs. 11

36. 36 ComparisonsComparisons 1.Proof of Concept vs. Efficacy trial 2.Effect size 3.Adherence requirements 4.Durability of effect

37. 37 Targeted Duration for Effectiveness HPTN035 – Microbicide Phase II/IIb HPTN039 – Acyclovir Phase III HPTN052 – HAART strategy Phase III HPTN058 – Suboxone for IDU Phase III With each sex act 12-30 mo Twice daily pill12-18 mo Daily HAART regimen5-7yrs Three times/week (DOT)1yr Duration Targeted Adherence

38. 38 HPTN035 Microbicide HPTN039 Acyclovir HPTN052 HAART strategy HPTN058 Suboxone IDU Years

39. 39 Duration of effect Survival design maxim of “fixed number of events” 88 events: mean 3.3 years with 300 people 88 events: mean 1 year with 1000 people Equivalent assuming proportional hazards Inherent in design Longevity of effect Estimate effect is “average” reduction in infection (weighted by time on study) Real world vs. Clinical trial setting Intended use Associated safety data Adherence over time Retention

40. 40 Non-proportional Hazard Designs HPTN052 – HAART strategy Immediate strategy likely to have early advantage; Delayed strategy may defray this with better viral control from conserved treatment options; Delayed strategy may be “almost as good” as immediate for preventing transmission over 5-7 years Trial designed with long follow-up, i.e. estimate decreased HR over 5-7 years, to deliberately average out the early effect HPTN058 – IDU Suboxone Likely early advantage while on suboxone (up to 1 year); potentially defrayed when suboxone removed Trial assesses difference in cumulative survival at 2 years Interim monitoring requires special attention

41. 41 Interim monitoring in HPTN052: HAART strategy Assess joint prevention benefit to partner and treatment benefit to index “The benefit of early ART will have to be considerable to offset the considerable logistical and financial costs” Long duration to ensure ability to assess both early and later effects of ART – HIV infected persons live with HIV for many years, on HAART for many years; Partner remains at risk Joint endpoint for monitoring: HIV infection in the partner or Life threatening event or Death in Index Hypothesis: Reduce number of significant clinical events and/or infections by at least 44%, ruling out benefits less than 20%

42. 42 ConclusionConclusion Trial design requires good collaboration Determining design parameters: effectiveness, duration, endpoint analysis Resources are finite but we need definitive answers Givens: Clinically relevant questions Definitive answers Role of adherence in achieving success requires ITT Variables Design depend on the prevention modality Strategies for advancing the science differ because of the science Each prevention modality has its own idiom

43. 43 “How can we be so different and feel so much alike?” mused Flitter “And how can we feel so different and be so much alike?” wondered Pip. “ I wish you could see in the dark, too.” “We wish you could land on your feet,” Flitter replied.