As reported here on aidsmap, the big news at the Microbicides 2004 Conference in London was the imminent start of five large efficacy trials of six different candidate microbicides. These could produce significant results by 2007, if the trials are successful in retaining the vast majority of volunteers (no mean feat given the degree of commitment these long studies imply for women who participate).
Investigators are keeping their fingers crossed that significant results will emerge, despite the fact that an unusually large number of unknowns could confound the results and produce noise rather than information.
Part of the problem is simply that, as with vaccines, trials to prevent conditions rather than treat them have to recruit many more participants.
However, vaccines at least have surrogate markers that can be measured in laboratory tests, such as the levels of cellular and humoral immune activation. While these may not translate into clinical benefit, candidate vaccines that do not produce such immune responses can be eliminated. With microbicides, there is no measure of success so far other than the incidence of HIV in the study population.
For this reason, studies have to be very large. Andrew Nunn of the Medical Research Council told the Conference that a 12 month randomised study recruiting 2,000 women in a country with 5% annual HIV incidence (possibly high, even for Africa) would have a one in three chance of failing to detect a halving of the incidence.
This is because the actual difference in numbers between 50 and 25 women infected hovers very near the boundary of statistical significance. If the microbicide was 40 per cent effective and the incidence 4%, no meaningful data could be generated, Nunn said.
There are numerous behavioural problems that could confound the data. One of the most significant is adherence. Norman Hearst of UCSF pointed out that a theoretical efficacy of 98-99% for condoms had never translated into a ‘real world’ actual effectiveness of more than 85-90%, due to the difficulty of maintaining 100% use. Given that an optimistic prediction for the efficacy of the first generation of microbicides is 60%, the actual proportion of infections prevented could be small enough to be undetectable except in the most high-powered studies.
Because adherence tends to weaken over time, opinion is turning away from huge trials lasting two to three years. If the aim of the first phase III trials is to demonstrate the efficacy of the compound under ideal conditions, more meaningful data would be generated by a group of very high-risk women receiving intensive monitoring for a shorter period. For this reason, a couple of the phase III trials changed their follow-up period from two years to one year or, in the case of the South African MDP trial of PRO-2000 and dextran sulphate, nine months (this is also the biggest trial, with a planned recruitment of 12,600 women in six countries).
Alan Stone of the MRC summarised other possible problems expected in a poster presentation.
Microbicide trials are a thorny ethical area. There is the problem of providing enough monitoring and support to participants to ensure adherence while not providing so much that condom use increases enough to make results meaningless.
Conversely, over-confidence by participants because they are using the microbicide could lead to ‘condom migration’; and this could result in the trial producing more, rather than fewer, cases of HIV.
There is also the problem of placebo control. We may only have one ‘go’ at placebo-controlled microbicide trials, because if a compound produces marginally significant results it would not then be ethical to re-test it in a second placebo trial. It would have to be tested against another active compound and this would require an even larger trial.
Another area of great controversy at the conference was whether to include a no-treatment or ‘condom-only’ arm in the trial. The US Food and Drug Agency is at present insisting on a three-arm design: microbicide plus condom/placebo plus condom/condom only.
A ‘condom-only’ arm is desirable, firstly because some placebo gels might have slight anti-HIV properties themselves and this can be measured against a condom-only arm; and secondly because it would then be possible to measure whether the microbicide/placebo becomes a disincentive to condom use.
But it is essentially adding an unblinded arm to a blinded trial. There is no guarantee that trials would not be biased by the fact that one group is receiving no compound changing their sexual behaviour. People in the microbicide and placebo arms could give trial compound to the no-treatment people (this has confounded results in trials of breast-milk substitutes before). And, of course, another arm either adds to the size of the trial by 50% or reduces its power by 33%.
It is thought that the HIV Prevention Treatment Network 035 trial of Buffergel and PRO-2000, which is the only one to be regulated by the FDA, was delayed and had to be scaled down into a less powerful phase II/III trial (where safety is pre-assessed in a small preliminary study) because the FDA insist on a condom-only arm as a condition of any product receiving a licence.
Microbicide trials involve screening people for HIV, and in high-prevalence countries this results in a lot of women finding out they have HIV. Some presenters commented that this was leading to counsellor burnout and stretched support resources, particularly in rural areas with poor infrastructure.
Anal sex (a subject just beginning to be researched in Africa) could confound results, as could IV drug use and other high-risk activities. However, the exclusion of people who report these activities could simply lead to people concealing them in the ‘coital diaries’ that are a feature of many of the trials.
Finally there is the question of ‘standard of care’, which was the subject of a whole plenary. What are researchers’ obligations to trial participants who become HIV-positive or suffer adverse events?
This has become a very live issue in the last few days, with potential recruits to the Bill Gates/UCSF tenofovir pre-exposure prophylaxis study of female sex workers in Cambodia refusing to join unless they receive insurance to cover tenofovir-related side-effects for up to 30 years.
There was a lively debate as to whether researchers were obliged to offer the best possible care, the standard of care in the host country, or something in between. A consensus seemed to be that developed-world care standards were not realistic and that the best that could be hoped for was that the process of doing a trial would ‘ratchet up’ local skills and infrastructure enough to benefit the whole population as well as trial failures.
Estimates for the cost of bringing a successful microbicide or choice of microbicides to market from the current pipeline range from $750m to $1bn; and that is without accounting for unforeseen events like trial failures or new classes of compound.
Researchers are therefore hoping that by the time the next Microbicides Conference takes place in South Africa in 2006, some of these questions will have been answered and the phase III trials will have started to produce meaningful results.