Oral PrEP can work just as well for women as men, modelling studies conclude

Oral PrEP for people whose exposure to HIV is through vaginal sex is just as effective as it is for exposure through anal sex, two re-analyses of the data from PrEP efficacy trials conclude. Lower efficacy among cisgender women, and zero efficacy in a couple of large trials, is almost entirely due to differences in adherence – not to any biological differences in the way PrEP is taken up by women’s bodies, or by vaginal rather than rectal tissues. In particular, the studies suggest that four or more doses of PrEP per week confer 95-100% efficacy, as they do in gay men.

One of the analyses comes from Dr Mia Moore and colleagues linked with the HIV Prevention Trials Network (HPTN), while the other is from Lanxin Zhang of the Robert Koch Institute in Berlin and colleagues. Both papers were published in the November issue of Nature Medicine.

Background

Ten years’ worth of data from the pivotal efficacy studies of PrEP demonstrated convincingly that oral PrEP using tenofovir disoproxil (TDF) plus emtricitabine (FTC) prevented most HIV infections in gay and bisexual men exposed to the virus. Initially, PrEP’s effectiveness seemed rather modest, with only 44% fewer HIV infections in men taking TDF/FTC versus men taking a placebo in 2010’s pioneering iPrEx study. But even then, efficacy was 73% in men who reported taking all their pills, and the researchers estimated it was as much as 92% in men who, as measured by drug levels in their blood, actually did take all their pills.

Two early studies in heterosexual men and women also seemed promising. Effectiveness in the Partners PrEP study in heterosexual African couples was 63% in women and 83% in men, and in the TDF2 study among mainly younger heterosexuals in Botswana was 75.5% in women (among those who returned for prescriptions) and 83% in men. (These two randomised studies from 2011 also remain our sole evidence for PrEP’s efficacy in heterosexual men.)

But two large studies conducted with women only delivered a shock. The FEM-PrEP study found only 6% fewer HIV infections among women given PrEP versus placebo, and 2013’s VOICE study actually found 4.4% more infections in women given PrEP. Both these results were statistically equivalent to zero effectiveness.

Aidsmap commented at the time that VOICE’s result “posed big questions for PrEP”. These were answered for gay men in 2015, when the PROUD and IPERGAY studies both reported that PrEP was 86% effective – even when, as in IPERGAY, it was taken with event-based dosing rather than daily.

It seemed clear at the time that the negative results of FEM-PrEP and VOICE were largely due to poor adherence, driven mainly by distrust of PrEP and of the research process among participants.

But there were also questions about whether PrEP was intrinsically less effective for exposure through vaginal rather than anal sex.

Some results seemed to show that while even moderate adherence of 3-4 pills a week was enough to stop HIV infection via anal sex, it needed 6-7 pills a week to work in vaginal sex. It was found that tenofovir (but not emtricitabine) takes considerably longer to reach preventive levels in vaginal, rather than rectal tissue, and never reaches the same levels. Could this explain poorer effectiveness in women?

Animal studies suggested this might be related to the fact that vaginal cells contain higher levels of naturally-occurring nucleosides that compete with the nucleoside-based NRTI drugs used in PrEP and get preferentially incorporated into viral components. In addition, some studies suggested that bacterial vaginosis could impact PrEP efficacy.

All this would not only rule out event-driven PrEP in women but might also mean that PrEP for them is generally less forgiving of missed doses.

The extremely positive results of the HPTN 084 study of injectable PrEP put paid to the idea that PrEP in general was less effective in women, but because HPTN 084 compared injectable cabotegravir with TDF/FTC and not with a placebo, it left open the question of whether the pills were intrinsically less efficacious in women.

Calculating biological efficacy for women

We know the level of PrEP drugs in the body in gay and bisexual men that is sufficient to prevent HIV infection. This is because a decade ago, researchers were able to calibrate the results of the STRAND study (which related the frequency of doses to long-term drug levels measured in the hair) against the results of the PrEP efficacy studies (which related those levels to the number of HIV infections seen.)

Levels in hair or in cellular specimens such as dried blood spots (DBS) are more reliable because they provide evidence of long-term, consistent usage over weeks or even months, whereas blood plasma levels only show adherence for the last few days, and are susceptible to phenomena such as ‘white coat dosing’ (where people resume a treatment just before their adherence is due to be checked).

In women there was nothing equivalent to STRAND, which especially matters if – because of lower drug levels in vaginal tissue – blood plasma or intracellular DBS levels do not reliably predict efficacy.

But the authors of one paper (Moore et al) realised that there was a near-equivalent in the shape of HPTN 082. This was primarily a study to see if youth-friendly print media, videos, community consultations and other engagement activities encouraged PrEP uptake and adherence in adolescent girls and young women. In this it was successful, achieving 95% adherence in participants. But the important factor for the Moore et al paper was that it measured both long-term drug levels in DBS samples and shorter term levels in blood plasma. This enabled the researchers to make a correlation between the intracellular levels measured in HPTN 082 and the blood plasma levels measured in the earlier efficacy studies. Then, crucially, they could relate that to the HIV incidence seen in women with low and high drug levels and determine the degree of protection different levels offered.

Because full adherence was so rare in FEM-PrEP and VOICE, they could not use data from those studies, but they could use drug levels measured in Partners PrEP and to a lesser extent in TDF2. Importantly they could also now establish the biological efficacy or TDF/FTC in the women allocated to it in HPTN 084, even though there was no placebo to compare it to.

In the case of HPTN 084, HIV incidence in women taking TDF/FTC was only 15% lower than incidence in the placebo arm of the VOICE study (after adjustment to reflect the different makeup of the two study groups). This looks as if was driven by poor adherence – 44% of all samples taken in HPTN 084 found no detectable drug in women allocated to TDF/FTC. But until this careful comparison was made, no one could say to what degree biological processes contributed.

Moore et al’s bottom-line finding was that adherence equivalent to two pills a week should provide an average efficacy against HIV of 58%; four pills a week should have 84% efficacy; and seven pills a week should provide 96% efficacy.

This is still lower than the efficacy calculated for gay and bisexual men, and roughly the same as that calculated by another analysis of HPTN 084 published last January.

However the second study (Zhang) suggests these may still be underestimates of the biological efficacy of different doses of oral PrEP in women.

This study also used what it called a ‘top down’ approach to calculate the efficacy of oral PrEP, basing it on the correlation between drug levels and efficacy detected in HPTN 084, Partners PrEP and to a lesser extent TDF2. However they did not try relate different doses per week to different efficacies, but just divided study participants into ‘no drug detected’, which implied zero efficacy, and ‘some drug detected’, which implied efficacy in the region of 90 to 100%. These weren’t guesses, but based on an analysis of the adherence-to-efficacy curve seen in the three studies. They saw efficacy dropping off sharply at around two pills a week: very few infections were seen in women with drug levels suggesting two or more pills, but levels any lower than that were associated with efficacy that was statistically zero.

Moore et al. commented in their paper that this approach neglects the partial efficacy offered by partial adherence, but this was only part of the model used by Zhang et al. They then tested the reliability of this 90-100% efficacy figure, based on ‘top down’ data by calibrating it against the predictions of PrEP efficacy offered by a ‘bottom-up’ approach.

This approach uses what are called counterfactual scenarios. The researchers tested the different explanations for the lower effectiveness of oral in PrEP in women by asking: “If the lower drug levels seen in vaginal tissues were the most important determinant of efficacy, what PrEP efficacy would we expect? What if the levels of competing host nucleosides in cells were significant? Is PrEP more effective in women who have some anal sex versus women who have none? Or is adherence the sole and sufficient explanation for lower oral PrEP effectiveness in women?”

By testing different combinations of these factors, they found that:

• Anal sex by itself had little influence, though incorporating a small percentage of anal sex slightly increased efficacy.
• If competing host nucleosides had an influence, it would actually be a slightly positive one, raising the efficacy of drug levels equivalent to two doses a week from about 90% to 95%.
• But if the lower drug levels in vaginal and cervical tissues were the prime determinant of PrEP efficacy, the effectiveness we actually see would be very much lower, to the extent of only 50% efficacy with seven doses a week and 20% efficacy with two.
• If the effects of host nucleosides and a proportion of anal sex were added together, efficacy would somewhat increase to over 75% with seven doses a week, and 55% with two, but it would still be much still lower than the data suggested by the trials.
• And the bottom-line figure calculated by Zhang et al. is that four doses or more a week are close to 100% efficacious – exactly as they are in gay and bisexual men.

To illustrate this further: their model computes that for women with any detectable level of drug in their blood, one would have expected to see three to four HIV infections in HPTN 084 – if drug levels in lymphocytes was the most important determinant of efficacy. Whereas if drug levels in vaginal tissue were crucial, then 15 to 27 HIV infections would be expected.

The former was the truth of the matter: only four of the 36 HIV infections seen in women allocated to TDF/FTC in HPTN 084 were in women with any detectable drug in their blood plasma at all.

What the studies found: a summary

The two modelling studies between them suggest that oral PrEP is just as biologically efficacious as injections in women. By calibrating the intracellular tenofovir and emtricitabine levels seen among women against their efficacy in recent trials, and then transferring the adherence/efficacy relationship found back to the earlier placebo-controlled studies (which did not measure drug levels as comprehensively), they found that PrEP’s effectiveness could be solely explained by adherence. Its intrinsic efficacy in women with good adherence was pretty much the same as it was in gay men.

The second study added in ‘what-if’ scenarios to predict what level of efficacy would be expected if lower drug levels in vaginal membrane cells were significant drivers of low efficacy and if host-cell nucleosides did compete with them. This study found PrEP efficacy would be much lower in women with good adherence than their model predicted if absorption by cells had that much effect.

In doing so, they have answered one of the key questions about PrEP that has remained unsolved: do drug levels in the epithelial cells of the vaginal and rectal membranes, where the virus first enters, matter at all? Or are levels in the body’s lymphocytes (white blood cells), which are the preferred host of HIV and where it replicates, more important? The models finds that levels in lymphocytes are what predicts efficacy. This suggests that topical PrEP, which only provides high level of drug at the sites of entry, such as the vaginal ring or microbicides, may be intrinsically less efficacious than systemic PrEP given orally or by injection.

Although the studies conclude that there is no reason oral PrEP would be less effective in cisgender women, they are not able to say that event-driven, ‘on demand’ PrEP would be just as effective as it is in gay and bisexual men, due to the complete lack of data in women.

The studies also do not address why, up till now, adherence to oral PrEP has generally been lower in cisgender women than in gay and bisexual men and transgender women. But studies like HPTN 082 suggest that this is not an inevitable state of affairs. Interest, uptake and adherence to PrEP could all be raised to high levels if it is portrayed in a way that women and especially young women can relate to.

A commentary published in Nature Medicine by experts in South Africa, Zimbabwe and the UK notes that a question often asked by people considering starting or, more often, stopping oral PrEP is “Why take a pill a day to stop having to take a pill a day?”. In other words, the burden of having to maintain 100% adherence to PrEP was a disincentive, especially for women, to protect themselves against what they saw as a smaller chance of catching HIV. The commentators suggest that if women could be assured that four or more doses a week would be protective, oral PrEP might seem a better option.

They also, however, note that the models in these studies assume that women face a constant risk of HIV. The models cannot account for event-driven dosing, or for longer on/off episodic dosing in women who, for example, may have male partners who work away and only come home from time to time. However these modelling studies do suggest that event-driven PrEP may not be inherently more risky in cisgender women than it is in gay and bisexual men and trans women.