An antiretroviral-containing microbicide could have some very paradoxical effects, according to epidemiologist Sally Blower, speaking at the Microbicides 2008 conference last week in Delhi.
In particular, a moderately-effective ARV microbicide which had a high level of systemic absorption and produced sustained levels of ARVs in the blood could end up reducing HIV infections in men more than in women.
The potential for ARV-based microbicides to cause resistance is one of the biggest unanswered questions about this class. The conference featured several studies of pharmacokinetics and found that two of the candidate microbicide drugs, achieved measurable levels in the blood – see this report.
Although these are far below the levels achieved by oral dosing, it is as yet an unanswered question as to whether someone using a microbicide containing one of these drugs who was HIV-positive and didn’t know it, or who caught HIV despite using the microbicide, might develop drug resistance.
Blower gave preliminary figures from a mathematical model which calculated the risk of resistance and associated HIV incidence and mortality for a ‘high risk’ and ‘low risk’ microbicide, under various conditions of efficacy and use. (These figures are from a soon to-be-published paper and could change after peer review.)
The model showed that, because of drug resistance, a microbicide of 50% efficacy, which was used 50% of the time without condoms, and featured a high risk of producing HIV drug resistance, would end up reducing AIDS deaths more in men than in women.
If women used this microbicide without condoms less than 30% of the time, it would have to have 90% efficacy if women were not to benefit less than men; and, as Blower commented, if a microbicide proved that efficacious in trials, there might well be pressure to license it even if there was significant systemic absorption.
Blower acknowledged that cases of HIV drug resistance in seroconverters in microbicide trials have so far proved to be extremely rare (about 0.3%); however this is because trial volunteers are screened for HIV and seroconverters are resistance-tested and get appropriate therapy if they do have primary resistance. Such conditions would not apply after licensing in many developing countries.
These paradoxical effects arise because men would only be vulnerable to resistant virus that they caught. Women, on the other hand, would develop it directly. In the UK at present, the risk of developing resistance on treatment is five times higher than the risk of being infected with resistant HIV.
Blower’s model, assuming 50% use, predicted something similar; 22% of women would have resistant HIV whereas only 5% of men would develop resistance. Some of the resistance the men would develop would be due to their bodies absorbing the microbicide through the penis and urethra (another set of assumptions built into the model), but most drug-resistant HIV (3%) would be transmitted.
Because resistant virus is transmitted more rarely than wild-type, only 0.2% of women who did not use the microbicide would acquire primary HIV infections with resistance directly due to the microbicide. HIV incidence would decrease more in men than in women (by 14% compared with 11%), and whereas there would be more than six HIV infections prevented per case of resistant HIV infection in men, there would be two cases of resistant HIV infection per case prevented in women.
This is a model with a number of pessimistic assumptions built into it.
Firstly, a strongly systemically-absorbed ‘high risk’ microbicide is unlikely to be licensed. However such risk might well apply should a single ARV be licensed for PrEP, especially in the absence of widespread testing.
The other assumption is that the model assumed ARVs were only available as microbicides and did not factor in the effect of increasing treatment access for people with HIV.
The session also included another couple of models that produced predictions somewhat contrary to expectation. In one, microbicides with low and high degrees of protection against HIV and STIs were mathematically ‘tested’ against the real-life HIV and STI prevalences in Johannesburg, south Africa and Cotonou, Benin – which has less than 10% of South Africa’s HIV prevalence.
The results showed that microbicides are likely to have a much larger impact in low-prevalence, concentrated epidemics than they are in high-prevalence, generalised ones. A microbicide with 40% efficacy against HIV and 40% efficacy against STIs would reduce HIV infections by 48% in Cotonou but only 8% in Johannesburg – because the steepest part of the epidemic curve in South Africa has already happened.
Finally, a model that plotted the risks of HIV infection in serodiscordant couples who used a 50% effective microbicide and/or condoms found that switching from using condoms half the time to using the microbicide three-quarters of the time would result in an increased risk for the HIV-negative partner (using current estimates for HIV transmission rates, condom efficacy and typical sex frequency).
The full table of predicted effects looks like this:
| Condom use | 0 | 30% | 0 | 50% | 30% | 50% |
| Microbicide use | 0 | 0 | 75% | 0 | 75% | 75% |
| Annual HIV incidence, negative partner | 23% | 16% | 13% | 12.5% | 12% | 7% |
| Relative reduction in HIV risk | 0 | 18% | 40% | 42% | 48% | 63% |
Mathematical models are not predictions of how microbicides will work in the real world. But they can serve to forewarn public health workers of effects that might seem counterintuitive. And they do suggest that for any microbicide to have a good chance of reducing HIV infections, especially in the women using it, it must have had a high degree of efficacy under trial conditions, be used consistently, and not be likely to cause much, if any, drug resistance.
Blower S. Modelling the impact of microbicide introduction. Key note address, track D, Microbicides 2008, Delhi. 2008.