VOICE trial's disappointing result poses big questions for PrEP

Gus Cairns reports on some recent disappointment in HIV prevention research.

The failure of the VOICE trial¹, one of the largest trials yet conducted of HIV-drug based prevention methods, poses questions for how to turn vaginal microbicides and oral pre-exposure prophylaxis (PrEP) into methods people can use in real life.

Final efficacy results from VOICE (Vaginal and Oral Interventions to Control the Epidemic) were presented at the 20th Conference on Retroviruses and Opportunistic Infections (CROI) in March.

The trial recruited 5029 women from three sites in South Africa, two in Zimbabwe and one in Uganda; Durban, in South Africa, provided more than half the participants. Women were randomised to use one of three prevention methods or two placebos (dummy methods):

• A daily Truvada (tenofovir plus emtricitabine) pill as PrEP;
• A daily tenofovir-only pill as PrEP;
• A daily placebo pill looking like Truvada;
• A tenofovir-containing gel, similar to that used in the CAPRISA 004 study, to be used as a vaginal microbicide;
• An inert gel as a placebo microbicide.

The tenofovir oral PrEP arm, and the tenofovir vaginal gel and placebo arms, of the trial were stopped due to futility in September and November 2011 respectively. ('Futility' means that the trial's data and safety monitoring board realised that there was no possibility that continuing these arms of the trial would produce a positive result.) The Truvada PrEP and placebo-pill arms, however, were continued.

But CROI heard that Truvada had also not proven effective in preventing HIV in the study and, therefore, that all three methods had proved no better than placebo.

No reduction seen in HIV infections

This was a group of relatively young (average age 25) and largely unmarried (79% single) women. Retention was good, with only 9% dropping out of the study.

Self-reported condom use at last vaginal sex was very high, at 85%, but needs to be regarded with a degree of scepticism given the extreme disconnect (see below) between self-reported and actual adherence to the methods being studied. Twenty-two per cent had had more than one male partner in the previous three months. Quite a high proportion of women (17%) reported anal sex in the last three months.

During the trial, 334 of the women tested HIV positive, but 22 of them turned out to have entered the trial while actually having been infected with HIV very shortly beforehand. This means 6.2% of participants became infected during the trial, an annual incidence rate (infection rate per year) of 5.7%, with strong geographical variance by site from 0.8 to 9.9%.

In the women using the tenofovir-gel microbicide there were 15% fewer infections versus placebo, but this was not statistically significant (which means the result could have been due to chance). In the oral PrEP arms there were actually more infections in women taking PrEP compared to placebo. Women taking Truvada were 4% more likely and women taking tenofovir alone 49% more likely to become HIV positive than women taking placebo; in the latter case, this was almost statistically significant (95% confidence interval 0.97-2.29, p = 0.07).

Adherence – much lower than reported

Adherence was assessed both by counting returned pill bottles and gel applicators and by a computer-based questionnaire. According to these two different methods, women took their PrEP pill or used the microbicide on average nine out of ten times.

However, drug levels in the blood, and in the case of the microbicide in vaginal fluids, were also analysed in a randomised subset of participants (about 15%), plus in all women who acquired HIV. This showed a very different story, and that is starting to be familiar in PrEP and microbicide studies: only 28 to 29% of women taking tenofovir or Truvada PrEP had measurable drug levels in their blood, and only 25% of women using tenofovir microbicide.

The assays could detect whether drug had been taken in the last two days on oral PrEP, and used in the last three days in the case of microbicide gel, so some may possibly have used PrEP or microbicide some time since the previous trial visit, but 50 to 58% of women, depending on which arm they were in, had no detectable drug in their blood at any trial visit.

Women who were married, were aged over 25, or who had a primary partner aged over 28, were more likely to have detectable drug levels. 

Married women were also very considerably less likely to acquire HIV than unmarried women: annual incidence in married women in South Africa was 0.9%, compared with 7.5% in unmarried women. Women over 25 were half as likely to acquire HIV as women under 25.

There was another disconnect between adherence and result, too: using contraception was a requirement for entering the trial. Seventy-one per cent of women were using an injectable contraceptive and 23% took an oral contraceptive pill (the remainder used other methods, including condoms). Or so they said – and yet the annual pregnancy rate was 7.8%. This suggest that a high proportion of women – especially those on non-injectable methods – were not using contraception.

It was also not the case that women started with high adherence but were unable to keep it up. Just 38% of women allocated to oral PrEP and 34% allocated to the microbicide gel had detectable drug the first time it was measured, and adherence only got worse after that.

Why did they join the trial?

These results seem to back up what was found in the FEM-PrEP trial and in a recent pilot trial of PrEP in young gay men: young people seem to find it particularly difficult to adhere to these biomedical prevention methods, suggesting to the investigators that "products that are long-lasting and require minimal daily adherence may be more suitable for this population".  Investigators also called for more social research to determine which populations might benefit.

But the VOICE study results raise even more difficult questions. Thousands of women went to the bother of signing up for a large clinical trial and often travelling to attend day-long clinic visits every three months, but the majority never took a single pill or used a dose of microbicide – and perhaps never intended to.

Yet there was excellent retention. As one audience member commented, the benefits of joining a trial like VOICE in a resource-poor setting may be so large as to make the disclosure of non-adherence feel very difficult for participants, who may fear being excluded from the trial.

These benefits include financial incentives: participants in clinical trials in South Africa are entitled to 150 rand per visit. Given the average black South African earns R5050 (£360) a month, this represents a fair sum.

The audience member was social scientist Dr Judy Auerbach, who comments in her blog: “A lot of folks in really resource-limited settings might quite sensibly say, ‘Well, I could get money for participating in this trial, and I’d get a lot of health benefits – monitoring of my health and wellness, good counseling, access to condoms – so it’s a good deal. I have no intention of taking the pill, but I’m not going to say that.’”

She adds: “In these clinical trials, researchers try to control out all the ‘noise’: You want similar women in similar circumstances so you can tease out the product effect. But social scientists say that the differences about women and their context – their community, their lives, their choices, their psychology, their culture, their age, all that stuff you’re trying to control for – are exactly what’s really important in these trials because they enter into how individuals think and (what they say) about taking a product.”