More evidence that PrEP works as well for women

In a presentation at last month’s Conference on Retroviruses and Opportunistic Infections (CROI 2024), the case was made that event-driven pre-exposure prophylaxis (PrEP) might work just as well for women as men, and should be offered to them as an option.

One of the pieces of evidence Dr Jenell Stewart drew on to make the case was published on 1 March in the Journal of the American Medical Association (JAMA), just before CROI 2024 opened. In it, Dr Jeanne Marrazzo of the University of Alabama showed that in 11 studies of PrEP that included women, there was a strong association between the women’s adherence to PrEP and their likelihood of acquiring HIV.

She found that non-daily but high adherence, taking four to six pills a week, was almost as effective at preventing HIV infection taking them every day.

This study, which pooled adherence and efficacy evidence from 11 PrEP trials, strengthens the evidence from modelling studies recently published that perfect adherence to oral PrEP is not required for effectiveness in women.

The studies analysed

The 11 studies included 6296 cisgender female participants, with the first study starting in November 2012 and the last one ending in December 2020. Nearly half (46%) of the participants were from Kenya, owing to the inclusion of two large demonstration projects. Twenty-eight per cent were from South Africa and 21% from India, with smaller numbers from Uganda, Botswana and the US. The average length of study follow-up was 31.5 months, ranging from 12 to 56 months.

Among the participants 2954 (47%) had adherence measures. Most were what Marrazzo and colleagues call ‘subjective’: self-reports, estimates made by doctors, or techniques such as pill counts and electronic pill bottle caps. Experience has shown that none of these adherence measures are reliable as participants tend to overstate their adherence.

The alternative is to measure adherence objectively using drug levels. This was done less often and less consistently because it is costly and requires more time at clinics. Just 237 of participants, in six of the 11 studies, had their adherence measured by drug levels – 8% of all those with adherence measures.

What adherence measures can tell you – and not tell you

Drug level measurements tell you different things according to whether the samples taken are of blood plasma or of cells. Measuring drug concentrations in dried blood spots (DBS), i.e. red blood cells, can tell you what people’s average adherence levels have been over the last eight to 12 weeks. They’re not good at picking out the kind of short-term variations you’d find in someone who was, for instance, using event-driven PrEP.

Conversely, blood plasma levels can tell you someone’s adherence level over the last two to seven days but can’t give you a longer-term average and are susceptible to ‘white-coat dosing’, i.e. taking PrEP just before a clinic appointment. DBS measurements were taken in 209 women, plasma in 46, and both in 18.

While objective drug level measurements cannot tell you everything about adherence, the results from subjective adherence may appear to tell us little. For instance, the proportion of women who said they took every dose was in the range of 40- 60%, whereas drug levels revealed that in fact 60-70% of participants actually took less than two doses a week, and most of them none.

But subjective measurements may give away more information than that. If the women in these studies were overstating their adherence consistently – both within a study and between studies – then objective tests could be used to calibrate the meaning of subjective declarations of adherence.

In addition, while subjective adherence measures greatly exaggerate the proportion who took PrEP daily and under-report the proportion who took hardly any, the proportion of  women who said they took middling amounts of PrEP are more in line with drug level data. For instance, about 20% of women said they took 4-6 pills a week – and drug level measurements show that the proportion of women in that adherence stratum was, indeed, about 20% (which does not mean they were the same women). The proportion who took 2-3 pills a week was 5-10% (though it tended to fall to zero after two years) and the proportion who said they took 2-3 was, throughout, about 5-8%.

Marrazzo and colleagues didn’t directly derive assumed adherence levels from declared ones. Instead, they took every subjective and objective adherence level measured or declared throughout each study and assigned an ordinal number to each measurement. In other words, if each women’s adherence was displayed in a list from highest to lowest, the ordinal number would say where each figure lay on that list.

They then asked, if a trial participant’s adherence is at the xth position at time zero, what is the probability it will still be at or near that point the next time it is measured?

This gave them a measure of for how consistent the women’s ascertained or declared adherence was, and they were able to sort the trial participants’ consistency of adherence into four groups.

Three groups had consistent daily or near-daily adherence (close to 7 doses a week), consistently high adherence (meaning 4-6 doses a week), and consistently low adherence (zero to two doses a week). The fourth group was also consistent, but in a different way; their adherence started off relatively high but then declined, usually by about 2-3 pills a week; in other words, if they started by taking 4-6 pills a week, they ended by taking 2-3. But they might go from three a week to zero, or seven a week to four.

There were 498 participants in the daily-adherence group, 658 in the 4-6 pills a week group, 1166 in the high-but-declining incidence group, and 632 in the 0-2 pills a day group.

Relating adherence to infection

Sorting adherence out into levels of consistency to some extent freed the researchers from having to state a correspondence between the likelihood of infection and an exact adherence figure.

Among the 6296 participants in the 11 studies, 32 acquired HIV. There were no infections in six studies; these tended to be the smaller ones and/or in lower-prevalence areas. This yielded an average annual incidence figure of 0.72%.

The only demographic factor that influenced incidence was age. Incidence in women under 25 was 1.33% and in women 25 or over it was 0.24% - a large and highly statistically significant difference.

The only other influence on the likelihood of becoming HIV positive was what adherence stratum someone belonged to.

There were no infections in the 498 women who took PrEP daily or near-daily. There was only one in 658 women in the 4 -6 pills a week incidence group. This equates to an incidence of 0.13%. Incidence in the high-but-declining group was 0.49%, and in the women who took little or no PrEP, it was 1.27%.

If we take this last figure as the ‘background’ (i.e. zero-PrEP) incidence, then this implies that PrEP stopped all the infections that would otherwise have happened in the daily PrEP group, 90% in the 4-6 pills a week group and 61% in the high-but-declining group.

Marrazzo and colleagues observe that the 90% effectiveness of 4-6 pills a week was about the same as the 88% effectiveness seen in participants who took 4-6 pills a week in the oral arm of the HPTN 084 study.

Comments and reservations

In a commentary of this study, Professor Anandi Sheth from Emory University in Atlanta and colleagues say that, with the majority of HIV infections globally happening in cisgender women, any evidence that ‘shifts the narrative’ told to women – that perfect adherence is needed to achieve efficacy – is welcome.

However they add that the evidence is incomplete. Studies that incorporate into their design detailed pharmacological measures of both long- and short-term adherence in all or most participants, and correlate them against behavioural data, are still urgently needed.

Furthermore, they add, although they find the evidence that four or more doses of PrEP a week offer high protection “reassuring”, only 40% of women in the 11 studies achieved this adherence level (and 36% in HPTN 084). If high-enough levels of adherence to oral PrEP, despite this reassurance, can’t be achieved by many women, then this adds urgency to the widespread implementation of injectable PrEP which, as another presentation at CROI 2024 emphasised, faces delays due barriers of cost and regulation.

So stronger evidence as to the true and achievable effectiveness of oral PrEP in women is still needed, so that, in the words of Jeanne Marrazzo and colleagues, women’s “broader…prevention needs [can] be integrated into decisions about PrEP use, and…cisgender women need not be restricted to a rigid daily regimen.”