A new British mathematical modelling study published in The Lancet HIV has found that adding pre-exposure prophylaxis (PrEP) for gay men at high risk of HIV to relatively modest increases in HIV testing, and immediate treatment for those diagnosed, could substantially cut the number of gay men infected by 2020. The researchers conclude that without these interventions, the number of gay men acquiring HIV is unlikely to decrease by 2020, even if the UK achieves the ‘90-90-90’ target of 73% of all people with HIV virally suppressed by this time.
The study, which was funded by the UK Health Protection Agency (now Public Health England) also finds that behavioural changes such as reducing the number of regular sexual partners could also make substantial inroads into HIV incidence but are less powerful as individual interventions.
It finds that increases in sexual partner numbers or condomless sex would substantially reduce the effectiveness of PrEP, testing and treatment, but even a complete cessation of condom use would not totally negate the effect of other interventions: only this, combined with a decrease in HIV testing, would do so.
Ideal impact of individual and combination interventions
The study uses a large set of inputs in which gay men in the UK were subdivided into less and more sexually active categories, older and younger. Sexual encounters were divided into one-off casual encounters and repeated sex with the same partner. The researchers assumed – in line with current data – that gay men picked partners of approximately their age and also serosorted to some degree, preferring partners of their own HIV status. They drew their estimates of risk behaviour from three different surveys, the NATSAL nationwide survey, the Gay Men’s Sex Surveys and the series of London gym surveys, which studied progressively smaller and more focused groups of gay men.
The model looked at seven different individual interventions which were first studied in isolation: testing for HIV once a year; testing for HIV twice a year; treatment of all diagnosed with HIV immediately after diagnosis (in this model, for those with CD4 counts under 350 cells/mm3, in line with the BHIVA guidelines at the time the model was devised); pre-exposure prophylaxis; reduction of the number of one-off partners by 50%; reduction of repeat partners by 50%; and a reduction in condomless sex with repeat sexual partners by 50% without a change in number of partners.
In order to gauge the power of individual interventions, the researchers first modelled unrealistic scenarios in which 100% of all gay men eligible for that intervention adopted it. They also modelled 100% take-up only in gay men under 45, and 100% take-up only in ‘high-risk’ gay men (which actually means more than one new sex partner a year).
They found that PrEP had the most powerful individual effect. If all HIV-negative gay men took PrEP, the proportion of men infected with HIV between 2014 and 2020 would be reduced by 59%, and if only high-risk gay men took it – an estimated 294,100 men – it would be reduced by 51%. The only other intervention in which HIV infections would be cut by more than one-third was if all gay men tested for HIV twice a year (42% reduction) or all high-risk gay men (39%) did so. Reducing sexual partners by 50% in high-risk gay men would reduce it by 32.5%; testing once a year would reduce infections in all high-risk gay men by 30%; and decreasing condomless sex by 50% in all high-risk gay men would reduce infections by nearly 25%. Treating all diagnosed high-risk gay men (17% reduction) and decreasing casual sex by 50% (11%) would have smaller effects.
Combined scenarios were more powerful, however. If all gay men reduced their number of repeat partners by 50%, and all HIV-negative gay men took PrEP, HIV infections would fall by 75% up to 2020; combining PrEP with treating all HIV-positive men would reduce infections by 70%.
No intervention is going to be adopted by 100% of the men eligible for it, of course. The researchers then modelled a range of scenarios in which the coverage of each intervention – the proportion of eligible men adopting it – was progressively reduced to 25%. This had a large effect on most interventions, but not as large for some as one might expect. If only a quarter of gay men halved their number of repeated sexual partners, it would cut the number of infections prevented by 75%; but infections were only reduced by 55% if only a quarter of gay men tested once a year and by 37% if only 25% of gay men tested twice a year.
Although immediate treatment had a relatively small effect by itself, it had a larger effect when combined with other interventions, and a synergistic effect if combined with regular testing. It was also the least affected by decreases in coverage or increases in risk behaviour; this is due to the fact that, as it reduces infectiousness anyway, such increases had less effect on HIV transmissions. At 25% coverage of both, ‘test-and-treat’ had the same effect as an individual intervention as PrEP would do.
In a probably achievable scenario a combination of PrEP in 25% of 'high-risk' men, 25% uptake of annual HIV testing in the 75% who did not use PrEP, and 25% more men diagnosed with HIV put on immediate treatment reduced the number of infections acquired up to 2020 by 44%.
If risk behaviours increase, then the reductions in HIV infections will be much smaller, the model finds, but will not disappear entirely. For instance if condom use with repeat sexual partners fell to zero, then combining PrEP (in 25% of HIV-negative men) and annual testing (in the other 75%) would only result in a 12% decrease in infections to 2020. On the other hand, if, in the same scenario, men doubled the number of repeat sexual partners they had instead of dropping condoms, then there would still be a 28% drop in HIV infections; this is because, paradoxically, more repeat sexual partners would result in increased serosorting behaviour in the 75% of men who were not on PrEP. However it would take at least a 75% increase in most ‘risk compensation’ behaviours – including a halving of the annual HIV testing rate – to produce a situation in which HIV infections did not decrease. In other words the adoption of nearly any combination of the interventions or behaviour changes proposed would be better than what we have at the moment.
The provision of PrEP within a combination strategy as above – especially if combined with increased HIV testing in those not on PrEP – would, in a likely scenario, prevent more than 7000 of a projected 17,000 infections by the end of the decade.
All models stand or fall by their inputs. For instance, if PrEP adherence was very low, it would decrease the effect of this intervention. However the effect of PrEP is actually likely to be greater than that predicted in this model: this is because the researchers took 44% as their base-case effectiveness rate of PrEP, from the iPrEx study. But, the 86% effectiveness found in the PROUD study would considerably increase the contribution of PrEP to reducing infection. However the researchers also found that increasing the number of gay men on PrEP had a greater impact on new infections than an increase in its effectiveness.
Conclusions and comments: rolling out PrEP
The researchers point out that with 84% of gay men knowing their HIV status, 90% of those diagnosed with HIV on treatment, and over 90% virally suppressed, the UK is within reach of achieving the UNAIDS/WHO ‘90-90-90’ target for HIV suppression. However, they predict that because of high transmission rates by the minority of undiagnosed men with HIV, achieving this would not be sufficient to reduce infections in gay men. However if it was combined with PrEP, this could increase the proportion of gay men aware of their infection to 95%.
In an accompanying commentary, the researchers Emily Arnold and Wayne Steward, take note of the finding that increasing the number of gay men on PrEP would make more difference than increasing its effectiveness, and ponder the challenges involved in scaling up PrEP, which requires “a unique combination of expertise and engagement”.
They point out that delivery of PrEP “does not fit neatly into the existing health delivery system: it is an intervention for HIV-negative individuals that requires medical expertise and ongoing interaction with the healthcare system.”
They suggest that the public health system needs to strengthen ties between community-based organisations and sexual health clinics and providers, with a view to adopting ideas such as medical workers delivering PrEP in community settings, as has already been done to some extent with testing. Community-based organisations can assist with publicity and outreach, but will need the “enhanced in-house expertise about biomedical forms of intervention” that PrEP programmes may provide. Medical providers also need guidance about how to manage people receiving PrEP.
Finally, although this model predicts that PrEP could make a substantial difference to HIV infections, it is not a cost-effectiveness model: the new predictions it provides of the effect of PrEP, especially when combined with other interventions, will need to be combined with cost-effectiveness models in order to establish the affordability of widespread PrEP provision.
Punyacharoensin N et al. Effect of pre-exposure prophylaxis and combination HIV prevention for men who have sex wth men in the UK: a mathematical modelling study. The Lancet HIV, early online publication (free, but requires registration). 2016.
Arnold EA and Steward WT. HIV transmission in MSM: considerations for PrEP scale-up. The Lancet HIV, early online publication (free, but requires registration). 2016.