Microbicides 2008: Will adherence issues affect all the first-generation trials?

There are still two ongoing trials of the ‘first generation’ non-ARV-based microbicide candidates ongoing, and at the Microbicides 2008 Conference last week Sharon Hillier, Principal Investigator of the Microbicides Trial Network (MTN) gave delegates an update of data from these studies.

The data collected so far suggest quite high use of both condoms and the study gel. However at the same conference Elof Johannson, Vice President of Biomedical Research for the Population Council gave a post-mortem of the Carraguard trial which was recently closed – see this report – and confirmed that the trial largely lost the power to determine whether Carraguard had any efficacy because of huge differences between claimed levels of gel use among trial participants and actual use.

HPTN 035 is four-arm, multi-centre Phase II/IIb randomised controlled trial that aims to determine the safety and effectiveness of two different candidate microbicides, BufferGel and 0.5% PRO 2000 gel for the prevention of HIV in women. It is supported by the US National Institutes of Health and is part of the MTN’s series of studies.

PRO 2000 is a sulphonated polymer similar to cellulose sulphate, but, as studies from the Conference found, less toxicity. It acts as an entry inhibitor. BufferGel works by lowering the pH of the vagina – in other words keeping the vaginal environment more acid, which has been shown to inhibit HIV transmission.

Participants are randomly assigned to one of four study groups: BufferGel, PRO 2000 gel, placebo gel, and no gel. Women assigned to the three gel groups apply gel up to one hour before sexual intercourse using pre-filled applicators.

The study design was much criticised at its inception for the incorporation of a ‘no-gel’ arm, Hillier noted. Critics had said it would be difficult to retain participants for the study arm that gave its participants no gel and just told them to use condoms, but Hillier said that retention had been as good in that arm as any other and that it was the only study that enabled researchers to detect any possible toxicity or efficacy of the placebo gel, hydroxyethylcellulose (HEC).

However the decision did have the effect of reducing the study’s power. As a results although the study was powered to detect and effect of either intervention against placebo if their joint efficacy was more than 43.6%, it was not powered to compare the interventions against each other.

The study enrolled its first participant in February 2005 and recruitment was completed in July 2007. At this time a total of 3,100 sexually active HIV-negative women had been enrolled at seven sites in Malawi, South Africa, Zambia, Zimbabwe, and Philadelphia in the United States. Each study participant is followed for at least one year.

The study is endpoint-driven, meaning that it will finish when it has reached 192 HIV infections amongst participants. At the current rate it is expected to be completed in July 2008, and results of the study available in early 2009.

The average age of participants was 27. Fifty-eight per cent of them were married and 64% living with a partner. The mean number of vaginal sex acts in a week was 2.7, and 65% of participants claimed condom use last time they had sex.

Self-reported adherence was high – though not as high as in the Carraguard trial, where 94% of participants claimed to use the gel. Condom use has remained at the same level in the study, and sex involving both condoms and gel has also remained at the same levels, with 83% of participants saying they used the gel when they used a condom.

However the percentage of women who used the gel when they did not use a condom increased as the trial went on.

Early on in the trial the percentage of unprotected sex acts that featured gel use was only 47% but as it has gone along that has increased to a cumulative average of gel use in 71% of unprotected sex acts. This may improve the possibility of a meaningful result, as early on in the trial there were concerns that too few women were relying on the gel for any statistical difference to be detected – see this report.

This study, supported by the Microbicides Development Programme of the UK Medical Research Council was, when set up, intended to compare two doses of PRO 2000 – 2% and 0.5% - to placebo. A larger phase III study, with 9673 participants at six sites in South Africa, Zambia, Tanzania and Uganda, it will be fully recruited this month (March 2008) and is expected to report results in late 2009.

Last month it was announced that the 2% trial arm would be closed – see this report – because it had been decided that there was no possibility of the trial arm demonstrating an effect when compared to placebo. The reason the higher dose might demonstrate less effect is because, while PRO 2000 appears less toxic than some other candidates, a protective effect at this dose might be offset by a local irritant effect.

The 0.5% arm is continuing and Hillier commented: “We may still find that 0.5% PRO 2000 prevents HIV”.

Trial participants in MDP 301 were a little older than in the previous trial – mean age 30 – and reported condom use was lower, with 55% of sex acts protected. However the median number of vaginal sex acts a week was also lower, at 2.0. Retention was high, at 88% if withdrawal for reason of pregnancy was not counted at 84% if it was included. Women could rejoin the study after giving birth.

Claimed adherence to gel use was high, at 84%, though has declined during the study from 91%. The proportion of protected sex acts that also featured gel has increased, from 57% to 67%, but the proportion of unprotected sex acts that featured gel decreased, from 91% to 83%. However, unlike in the HPTN 035 study, gel use was more common in unprotected than in protected sex.

There was a huge range of condom usage between sites: a fact that may help generate meaningful data – ranging from 80% in South Africa to 17% in Zambia.

One interesting confounder may be anal sex. One per cent of women at the monthly visits said they’d had anal sex in the last month, and condom use was lower in anal sex, at 44%. Given that HIV is more efficiently transmitted anally and that participants were told that the gel was for vaginal use, anal sex could give rise to a number of infections.

Elof Johansson of the Population Council reported on the Carraguard trial. Much of the data on the trial closure can be found in a previous report and we will only report what Johansson added here.

Johansson started by commenting: “We old guys who came from the contraceptive world have been humbled by the problems in microbicide studies.”

“In trials of contraceptives we had endpoints in phase II studies and could measure drug levels and efficacy. In microbicides phase III is the trial; all others are just safety trials.”

As previously reported, while participants claimed 94% usage of Carraguard, a technique that used a dye that reacted to vaginal mucus showed that only 61% of the returned applicators had actually been used and that Carraguard was only used in 43% of sex acts.

Furthermore, said Johansson, this test was not included in the protocol and was only introduced some way into study. In addition, when it was introduced participants would drop off applicators at the research centre reception, giving no chance to interact with staff. Halfway through the study they started leaving them with counsellors.

There was a huge range of gel use amongst participants, ranging from women who never used it to 10% of women who managed 100% use, and subgroup analyses of women who never used it, those who used it less than 35% of the time, less than 80% of the time and over 80% of the time are ongoing.

Low adherence was not the only reason the trial ended up having little power to generate a statistically meaningful result, said Johansson. Although the drop-out rate was only 13%, nine per cent were in addition lost to pregnancy, and pregnant women did not return to the study after giving birth.

One factor that may yet generate meaningful data is that trial participants at the three South African sites – at Cape Town, Pretoria and Durban – were very different. Durban participants has less sex – 1.3 acts per week compared with 2.2-2.3 at the other two sites. However HIV prevalence was much higher, at 43% of the adult population, compared with 24% at Pretoria and 18% at Cape Town. Circumcision rates in men were also very different, with 97% of men in the Cape Town area circumcised, 54% in Pretoria but only 24% in Durban.

This generated interestingly different results for efficacy in the Durban site. In that site there were actually more seroconversions amongst women using Carraguard than in the placebo arm – 48 (3.3% of participants) against 42 (2.8%).

In contrast there were more seroconversions in placebo users in the other two sites: 86 (1.9%) of Carraguard users and 103 (2.3%) of placebo users. This was still not significant but may indicate the protective effect of circumcision for women in the study.

As in MDP 301, there was a small but still significant minority of 2% of women who practised anal sex.

Lastly, Carraguard was popular: women said it improved sex and they liked the feel. Other studies at the conference reported similar findings, to the extent that participants at one site in the prematurely terminated cellulose sulphate trial refused to hand their gel back until they were given supplies of placebo.

Johansson said that with future trials close monitoring of adherence was essential. With this, it would be possible to do a ‘per protocol’ analysis only looking at efficacy amongst actual users rather than the ‘intention to treat’ analysis used.

Johansson commented that “using an intention-to-treat analysis in people who do not feel sick is very difficult”, and noted that trials of blood pressure drugs had run into the same problems. He said that in future trials he would also suggest a sub-study of the intervention in HIV-positive women.