1. Exploring the potential impact of ART in reducing HIV transmission. Geoff Garnett, Jeff Eaton, Tim Hallett & Ide Cremin Imperial College London

2. Contents Potential impact of increased treatment at CD4 < 200 and < 350 on spread of infection. Potential impact of pre-exposure prophylaxis (PrEP). When treatment and when PrEP?

3. Factors decreasing the role of later stages of HIV infection and the potential of treatment to reduce transmission Rapid spread and saturation of HIV in the at risk population (i.e. little ongoing spread of infection). Decreasing number of contacts as a function of time since infection - 1) concurrency leads to more potential contacts early infection; 2) people reducing numbers of partners over time; 3) Saturation in age cohorts Poor adherence; poor suppression of viral load; treatment failure and resistance. Slower progression to low CD4 counts. Increased risk behaviour of those on treatment. Increased risk behaviour amongst those not on treatment – including susceptibles.

4. Model - Eaton et al AIDS & Behaviour (In Press) : Transmission model (Stochastic individual based) representing generalised heterosexual epidemic – including: concurrency in sexual partnerships; Heterogeneity in propensity to acquire new partnerships; Transmission risk within partnerships as a function of time since infection. Movement from high activity to moderate activity and moderate activity to low activity over time. Population size 50,000; seed 1% prevalence; results average of 100 runs.

5. Transmission risk by stage of HIV infection

6. Proportion of infections generated as a function of time since infection.

7. Proportion of transmission by stage of infection as epidemic progresses.

8. Population size 50,000; seed 1% prevalence; results average of 100 runs.

9. Generalised epidemic – concurrency driving epidemic CD4< 350 after mean 4.5 years

10. Slower progression to CD4 <350 More infections in earlier stages. Mean duration to <350 7 years.

11. Epidemic drive by small (2%) high risk group (prevalence 1.5%) More sensitive to movement from high to low risk.

12. PrEP model developed by Tim Hallett and Ide Cremin The first model of PrEP for West Africa Detailed Representation of PrEP Detailed patterns of adherence Targeting Duration on PrEP PrEP in Combination Prevention »Treatment for clinical need »Increases in condom use & reductions in numbers of partners »‘Early’ treatment initiation

13. The model captures many important features of HIV transmission in Cotonou: The Mathematical Model PrEP for prevention – preliminary results Sex workers Regular clients Women Men

14. Coverage, Adherence & Duration PrEP for prevention – preliminary results “Optimistic”“Realistic” Coverage Uniform with respect to risk group and gender % of PrEP users with good adherence 80%50% Mean duration on PrEP 10 years5 years Years to reach coverage 2 years5 years

15. “Optimistic”

16. “Realistic”

17. Effective Targeting For the same number of people staring PrEP, effective targeting to those at most risk can substantially amplify impact. 10% of population start PrEP Good Targeting No Targeting Some Targeting

18. PrEP in Combination Prevention Status quo Intervention to scale (incr. condom use and prompt treatment initiation) + Targeted effective PreP + The missing piece? Numbers based on extrapolation to Urban Benin; *PreP intervention is to 60% of sex workers & clients; 70% efficacy and 80% adherence, for 10 years. ** The missing piece required to reduce incidence by 90% in 2031 and eventually stop the epidemic is a 60% efficacy vaccine delivered to half the population.

19. 52%* 13%** 91%*

20. ART initiation at CD4<350 PrEP and ART initiation at CD4<200 Domain where PrEP averts more infections that treatment in couples. Need PreP effectiveness>60%

21. PrEP and ART initiation at CD4<350 ART initiation immediately Domain where PrEP averts more infections that treatment in couples. Need PreP effectiveness>85%

22. Conclusions Good coverage of those with CD4 < 200 could avert around 25% of new infections and with CD4 < 350 a further 15% could be averted. Reductions in risk behaviour associated with treatment could improve this; increases in risk behaviour could undermine it. PrEP can reduce incidence but needs high efficacy, coverage and adherence - and needs appropriate targeting to be efficient. Earlier treatment reduces role of PrEP; its effectiveness per partnership relative to treatment of the infected partner determines how useful it would be in discordant couples.