Lessons learned from the first generation of microbicide studies

People in microbicides trials are never encouraged to have unprotected sex. Biomedical ethics require that all trial participants be counselled to use condoms every time they have sex, as well as the test product they receive (either the candidate microbicides or the comparison product, if they are in the placebo arm of the trial). The trial cannot measure whether the product being tested is better than nothing. It can only measure the extent to which it is better than the best available prevention package in terms of reducing participants’ HIV risk.

Although we refer to the placebo arm of a microbicide trial, it isn’t really a ‘placebo’ arm at all. Instead of comparing participants’ responses to either an active medication or an inactive look-alike (as occurs in a treatment trial), microbicide trials compare one very effective intervention (condoms and condom promotion) to condoms plus an extra intervention (the candidate microbicide). Clearly, it is more difficult to prove microbicide effectiveness in this circumstance. 

The difficulty is further exacerbated by the fact that trial participants frequently receive free medical treatment to which they had no access previously, including treatment of STIs that further lowers their risk of HIV acquisition. We are not suggesting that trial participants should not be as fully protected from HIV as possible during the trial, but simply noting that the need to do so heightens the challenge of demonstrating the efficacy, if any, of a test product. As Professor Elof Johansson of the Population Council commented: “For ethical reasons we have to promote condom use within the trial. In my 35 years of experience working with clinical trials, I’ve never been in such a difficult situation where you have to promote another treatment that will work as well, and probably better, than the product you are testing.”¹

Some trial participants acquire HIV during the course of a trial because they continue to have unprotected sex. This may occur by choice, or it may be because they are unable, despite assistance and counselling, to insist on consistent condom use with their partners. The trial assumes that the percentage of sex acts in which people do not use the condom, but do use the test product, is likely to be the same in both arms of the trial. Thus, if the rate of HIV seroconversion is lower in the candidate microbicide arm than in the comparison arm, that difference can be regarded as the effectiveness of the product.

Interestingly, the reverse of that behaviour - better adherence to condom use than use of the test products - is also being reported in some studies. In the cellulose-sulfate trial in Nigeria, for example, participants reported using a condom for 90% of sexual acts in the past week, but using the microbicide for only 83% of the acts in that time. Reports from the Savvy Nigeria study are similar (88% using condoms, 78% using gel for sexual acts in the past week). If such reported condom-use rates are accurate and sustained, it can lead to situations in which seroconversion rates are so low that it is impossible for the trial to prove anything.

Because of the ethical need to reduce risk among trial participants, and the resulting difficulty of measuring real product effectiveness, we learned during the first generation of trials that techniques for promoting adherence and gathering accurate adherence data are essential. Without reliable information on what participants used (condoms, the test product, both together, or neither one) and how often, it is impossible to measure the impact that the test product may (or may not) have had on HIV-seroconversion rates and other outcomes.

When drugs are swallowed (as in treatment or PrEP trials), adherence can be monitored simply by measuring drug levels in the blood stream. But, as discussed in the pharmacokenetics section below, this may not be a reliable measure of adherence when drugs are applied topically. In response to the overall unreliability of self-reporting, a number of innovative approaches have been developed to promote and monitor adherence in microbicide trials.

Telephone and web-based technology:  In one trial, participants were given cell phones and asked to call a toll-free number to report (through an interactive voice-response system) their sexual behaviour and product use.²  Audio Computer-Assisted Self-Interviewing (ACASI) is another computer-based approach. In trials using­ ACASI, participants respond (verbally or by selecting an image on the screen) to behavioural questions asked by a computer during their clinic visits. While ACASI seems to facilitate more accurate reporting of highly stigmatised behaviour than standard face-to-face interviewing does, it still does not lead to full disclosure regarding adherence.³

Applicator collection and testing:  Most trials involving the administration of a vaginal gel by applicator require the participant to return their used and unused applicators, as an objective way of measuring product use. In a study associated with the Carraguard Phase 3 trial, a dye applied to each empty applicator after collection showed which ones had actually been inserted vaginally. The resulting data highlighted a decline in adherence to gel use over time, as had also been noted in the larger trial.⁴  

More elaborate versions of this approach have also emerged. The ‘Wisebag’ is a device blending adherence-promotion messaging with product-use documentation. Participants are given gel applicators in a Wisebag. Every time the bag is opened, the date and time are recorded and an electronic text message is sent to the participant reminding her to use the gel.⁵  A device called the ‘Smart Applicator’ omits the reminder but records even more data about applicator use. The Smart Applicator is designed to record the time and date of its use, how much gel was expelled from it and under what conditions (e.g., temperature at time of use). After preliminary testing, this model is now being refined to be smaller, reusable and better adapted for use in clinical-trial settings.⁶

In addition to creating tools and techniques to collect and validate adherence data, the body of social and behavioural knowledge about adherence is also growing as researchers strive to better understand the factors that influence adherence and how improved adherence can be encouraged.

Predictors of adherence: One sub-study of participants in the Carraguard trial, for example, found that: “the same factors associated with risk of HIV acquisition – being aged 22 - 29, having sex regularly, having an abusive partner, or having sex in exchange for money – may also be associated with difficulty using a coitally dependent microbicide.”⁴

Daily monitoring of adherence (DMA):  The International Partnership for Microbicides is revisiting the 'directly observed therapy' (DOT) approach used in tuberculosis treatment to see if it can be translated successfully to boost adherence in microbicide trials. Participants in IPM’s six-week DMA trial were contacted daily by home visit or by attendance at a drop-off centre (a separate, convenient meeting point). Used gel applicators were collected daily during these visits. More than 95% of all applicators were collected within the expected time range and attrition from the study was very low. IPM is currently considering scale up of DMA for a large Phase III trial.⁷

As sociologist Judith Auerbach noted in her plenary address at Microbicides 2010, people do not passively accept new products. Rather, they actively choose to take them up or not, and often appropriate and adapt them to better fit the context of their lives. Adherence to product use within a clinical trial (or in real life) cannot be understood without an ‘insider insight’ into how people see both the trial and the attendant product. Such understanding can only be cultivated through non-trial-focused research, which is an essential tool for developing effective responses to the challenges of adherence.⁸

The third major challenge observed during the first-generation trials has been the incidence of pregnancy among trial participants. Women participants who fall pregnant during a clinical trial are immediately required to discontinue their use of the investigational product, since little or no human data exist as to how continued use could affect the woman or the foetus. A high pregnancy rate in a trial, therefore, means a high drop-out rate and this has implications for interpretation of trial data. Trials can become underpowered (not have enough participants to generate a statistically significant result) if pregnancy rates have a large impact on participant retention.

The dominant response to this has been expanded efforts to make contraception widely available and consistently used by trial participants. This is a relatively low-cost way to both provide additional health benefits to participants and enhance the study’s chances of yielding meaningful data.

The acceptability and uptake of contraception varies, however, across countries and regions. Strategies for improving contraceptive access are also inconsistent across trial sites. Some strategies involve referring women to local clinics and healthcare providers, whilst others provide contraceptive care, counselling and supplies on site. In African trial-host communities, hormonal contraceptives (taken by injection or orally) tend to be the most popular methods locally and, therefore, the most commonly requested by trial participants.

Discussion of emergency contraception and abortion with trial participants has also raised controversy. Trial networks are not only constrained by local laws regarding abortion, but also by local sentiment and the risk that they could be perceived as encouraging abortion among trial participants. Simultaneously, however, they are ethically obliged to address participants’ needs and questions as they arise. After studying the standards of care provided to trial participants across several trial sites, the Global Campaign for Microbicides included the following among its recommendations: “Site staff should be trained to inform women of all pregnancy options, including termination of pregnancy where abortion is legal, and to counsel on the dangers of unsafe abortion and when and where to seek care in case of post abortion complications.”⁹

Another response to high pregnancy rates in trials has been the launch of research to determine what impact candidate microbicides may have on the health of pregnant women and their foetuses. When microbicides become commercially available, some women will (knowingly or unknowingly) use them during pregnancy. It is essential, therefore, that ethical methods be developed for testing the effects of such use. The Microbicides Trials Network (MTN) has taken an innovative step in this direction.

Sixteen healthy, HIV-negative, pregnant MTN volunteers agreed to accept a single dose of tenofovir gel vaginally approximately two hours before giving birth via scheduled Caesarean section.¹⁰ Blood samples were subsequently collected from each woman’s blood stream, amniotic fluid surrounding the foetus, and umbilical cord. In comparing the tenofovir levels in these samples to blood provided to non-pregnant women using oral tenofovir, researchers found 40 times less drug in the cord-blood samples and up to 100 times less in the maternal blood than in the blood of oral tenofovir users.

While this study was small, it breaks new ground in a much-needed research area. Larger, and equally carefully designed, studies of the effects of tenofovir gel and other candidate microbicides on pregnant and breast-feeding women are now needed.