The Ethical Odyssey in Testing HIV Treatment as Prevention

Study Design

Among others, necessary conditions for research to be ethically acceptable are that a study appropriately and rigorously designed to address relevant research questions.⁴ In randomized clinical trials, such research may rightly involve examining testing interventions for which there is a lack of expert consensus on whether one study arm is superior to another. Some scholars believe that such a state of affairs can be described as “clinical equipoise”,⁵ yet others find this notion to be problematic⁶. While it is beyond the scope of this paper to defend a particular conceptual approach, it seems fair to say that those on both sides of the conceptual debate would endorse the ethical requirements delineated above^6,7.

As the HPTN 052 protocol was being developed, it was clear that there was a need to develop means for preventing the sexual transmission of HIV infection as well as to offer the best treatment for those infected with HIV. While limited data were available concerning the research questions the study addressed, some experts at the time believed that the primary research question regarding transmission to a sexual partner had already been answered based upon a single retrospective observational study that associated blood HIV viral load and transmission probability in Uganda.⁸ In other words, the strong biological plausibility of the idea combined with the observational data suggested to some that a randomized trial was not required. This belief had to be explicitly considered to ensure that conducting the proposed research would still be ethically appropriate This in part arguable involves whether there was clinical equipoise regarding each arm of the study in terms of the main outcome measures. Note that clinical equipoise permits particular individuals to disagree that uncertainty remains since under this approach the crucial assessment is whether there is uncertainty among experts all things considered. Accordingly, it was critical to recognize the limitations of the findings from the observational Uganda study. First, the Uganda study did not involve the use of ART, nor did it determine whether apparent HIV transmission events were virally linked. Since HIV can be acquired outside individual partnerships¹, demonstrating viral linkage is critical. Most importantly, the observational data from Uganda could not provide insight about the magnitude or durability of benefit afforded by intervening with ART; however, such data are critical for developing sound prevention strategies. Ultimately, this information from HPTN 052 regarding the extent to which ART prevented HIV transmission became a catalytic event for establishing that ART can help prevent infection with HIV⁹; at its outset HPTN 052 was poised to help answer research questions about which there were substantial uncertainty.

Site Selection

In order to answer the primary research question, it was essential to conduct HPTN 052 at sites that were experiencing high incidence rates of HIV infection. Across much of the world at the time the study was being designed, high HIV incidence rates were commonplace in resource constrained settings. As a consequence, as the trial was being designed, ART was not generally available at most of the research sites, generating claims that the trial might be “coercive”, since participants would have access to life-saving drugs that otherwise would not be available to them. While the unavailability of ART at the research sites reflected longstanding issues of global justice, labeling the trial as coercive was misplaced since coercion implies a forced bad choice.¹⁰ Given what was known about the use of ART at the time, deciding whether to participate in HPTN 052 did not necessarily require making a forced bad choice. Similarly, offering enrollment in the trial would not be properly considered an undue inducement since the offer of participation including access to ART would not reasonably seem to adversely affect decision-making of potential subjects.¹¹ However, if there was unavailability of ART at the research sites, enrollment would understandably be considered a strong incentive. Fortunately, the World Health Organization’s Three by Five Campaign¹², the Global Fund to Fight AIDS, Tuberculosis, and Malaria¹³ and the President’s Emergency Plan for AIDS Relief (PEPFAR)¹⁴ made ART far more available than it had been before HPTN 052 fully started, so patients with low CD4 counts who unequivocally needed ART generally had access to treatment according to country guidelines without having to enroll in HPTN 052. It is important to note, however, that this typically meant that ART was available for those with very low CD4 counts who would de facto be ineligible for enrollment in the HPTN 052 study. In addition, in coordination with NIAID, a separate study (AIDS Clinical Trials Group 5175) was launched to offer the option of participating in a treatment clinical trial for potential participants who screened out of HPTN 052 due to CD4 count falling below the threshold required for study entry.¹⁵

The prevention package

Until quite recently, HIV prevention in adults depended entirely on reducing higher risk behavior through counseling and the provision of condoms. To provide ART without offering safer sex counseling and condoms in HPTN 052 would suggest that the investigators and sponsor were not concerned with the well-being of the participants and were willing to tolerate HIV transmission to address the research questions. Conversely, if HIV transmission was decreased greatly as a result of counseling and behavior change, the trial might well prove futile. This tension is commonplace in HIV-prevention trials, yet including a “prevention package” is ethically obligatory. As discussed in more detail elsewhere, developing a particular prevention package for a trial should be established with broad input and include interventions that are known to be effective, reasonably accessible, and practically achievable.¹⁶ In order to ensure the integrity of a trial, it should be powered appropriately to account for the anticipated prevention benefit from the package. In HPTN 052, couples and adherence counseling was optimized so that the intervention could reasonably be expected to be effective, accessible and achievable. In addition, the study was powered to anticipate the very low level of HIV transmission that was actually observed.¹

ART and HIV transmission

Prior to the results from HPTN 052, two lines of evidence were used to examine the effects of ART on sexual transmission of HIV: analyses of HIV transmission among serodiscordant couples in observational studies,^{20, 21} and population-level ecological studies.²² While most observational studies support the idea that ART can decrease HIV transmission, not all population study results have reported a benefit from ART.^{23, 24} Further, there are clear limitations of observational and ecologic studies.²⁵ Accordingly, although observational studies can be useful, the potential for confounding and systematic error should cause great pause when and if such studies are used to support changes in medical or public health policy. The randomized controlled trial method is used specifically to reduce many of these concerns, leading to a higher level of evidence¹⁸. Thus, despite biological plausibility, evolving data from population studies, and growing interest in using ART for prevention during the course of the HPTN 052, it remained unclear if this approach was truly safe and effective, warranting continuation of the trial. While it may seem intuitive that early initiation of ART might be desirable, untoward side effects in new populations always require consideration. In addition, the effects of ART on sexual behavior have long been a concern, with the fear that increased risk behavior might offset the prevention benefit provided by ART.²⁶ Ultimately, he desire for rigorous data needed to be balanced against the interests of the subjects; accordingly HPTN 052 included not only plausible approaches to the prevention of HIV transmission, but also regular systematic monitoring by the DSMB.

When to start ART

At least five observational retrospective studies addressing the optimal time to initiate ART have been published.^{24, 27–30} These studies all suggest that ART started at a CD4 count higher than 200 cells/mm3 reduces morbidity and mortality from HIV infection, but to differing degrees depending on the methods and the cohorts. The greatest disagreements in these reports surface when ART is initiated at CD4 counts greater than 500 cells/mm³ where differences observed are smaller, and sometimes negligible. As the first of these reports surfaced as meeting abstracts or publications, many organizational (normative) treatment guidelines underwent review and were modified in response to them.^3,31–33 The WHO guidelines were of greatest importance to HPTN 052 because of the multinational nature of the study.^3,31 However, it should be emphasized that the countries that participated in HPTN 052 did not uniformly follow the evolving WHO Guidelines (see Table). The country decisions were predicated on their own expert response to the strength of evidence, and the general lack of availability of enough ART to follow WHO guidelines if they were put into place. In 2009, investigators working in Haiti reported that subjects initiating ART at a median CD4 count of 287 cells/mm3 did considerably better than those delaying ART until a median CD4 count of 160 cells/mm3.¹⁷ While the data from this Comprehensive International Program of Research on AIDS (CIPRA) Haiti study were compelling, understanding the potential implications of these data for HPTN 052 required considering a variety of questions, beginning with the external validity of the findings. Were the populations in the CIPRA Haiti and HPTN 052 studies so similar that the results applied to the HPTN 052 study? Were the study designs so similar that the CIPRA Haiti results addressed the “when to start” question in HPTN 052?

On careful examination it became clear that there were important differences in the two studies. In short, ART was initiated in subjects in HPTN 052 at higher CD4 counts than in the CIPRA Haiti study in both the immediate and delayed arm. Ultimately, in HPTN 052, the median CD4 count for the infected partner was 442 cells/mm3 in the immediate arm and 428 cells/mm3 in the delayed arm; ART was initiated in the delayed arm at a CD4 count median of 221 cells/mm3 (Range: 199–249 cells/mm3)¹. Nevertheless, the CIPRA Haiti results and the anticipated changes in WHO guidelines expected to specify earlier initiation of ART, led NIAID to request an urgent DSMB meeting in July 2009 to consider the aggregate data. The primary purpose of this deliberation was to protect the well-being of the subjects enrolled in the trial; here was a secondary but critical obligation to protect the integrity of the trial, if appropriate. These deliberations emphasize the constant tension between the need for modification of clinical and public health guidelines in response the best available evidence, and the generation of better evidence: if randomized controlled trials were routinely discontinued because of observational data or changing expert belief, many critical trials would not come to completion. In other words, the opportunity to obtain a higher level of evidence to address critical questions would be foreclosed.

Upon review of all the internal and external data, the DSMB saw no evidence of harm to the HPTN 052 subjects and the study was permitted to continue without modification. However, in November of 2009, when new “Rapid Advice” WHO guidelines were released, which recommended that ART be initiated at or before a CD4 count falls below 350 cells/mm3,³¹ all couples enrolled in the study were provided information about the new guidelines.

Despite uniformity in this approach across the study sites, Kenya, one of the participating countries in HPTN 052 with a site in partnership with the Centers for Disease Control-sponsored Kenya Medical Research Institute, adopted the new WHO guidelines in late 2010. The CDC felt that a change in country guidelines to comport with the international guidelines should be endorsed, and that the study participants should be provided with ART to be consistent with these guidelines. Given the desire to maintain uniformity across all sites, consideration was given to discontinuation of the study site, but the CDC declined this option. As an alternative, participants at the site were given information about the WHO and country guidelines and index participants in the delayed arm were told how to obtain ART outside the study at the country-recommended CD4 count level should they choose to seek it. Eight (8) participants elected to do so; for purposes of analysis of data, those changes in treatment were considered protocol violations.

Linked and Unlinked Transmission

The primary endpoint of HPTN 052 requires the establishment of virally linked and unlinked transmission events.¹ While transmission linkage information is important in understanding the study results, whether to communicate such results to participants is another matter. After considering the merits of disclosing linkage information to participants, a decision was made to keep it confidential. The justification for doing so is based on the potential harms related to disclosure or non-disclosure. Any potential benefits for the disclosure to the index subject would derive from having information about sexual activity outside the partnership that would then translate into engaging in safer sexual practices to minimize the chance of sexually transmitted infections. Conversely, revealing an unlinked transmission could lead to harm and even physical violence to the newly infected subject. In addition, each member of the couples participating in the study gave individual consent, which stipulated that any information about sexual activity would be kept confidential from the other member of the couple. Parenthetically, couples with unlinked transmission were more likely to acknowledge multiple sexual partners.^{37, 38}