Abstract Body

The effectiveness of male circumcision (MC) as an HIV prevention measure among heterosexual men has been demonstrated in clinical trials. However, the efficacy and population-level effectiveness of MC among men who have sex with men (MSM) remains uncertain, and is likely to depend on the behavioral and demographic characteristics of specific MSM populations. We assessed the potential impact of MC among MSM in different settings worldwide to help determine the conditions in which it could be an effective HIV prevention measure for MSM.

We developed a deterministic compartmental model of HIV transmission among MSM and simulated the HIV epidemic in nine selected countries. The model incorporates infectivity by type of sex act, sexual mixing by role preference, condom use, three stages of HIV with varying infectivity, and assumes a 40%-67% MC efficacy during insertive anal sex. The model was calibrated to country-specific HIV prevalence and current coverage of MC (Figure), and accounted for the uncertainty in other parameter values based on literature review. We compared intervention strategies of MC scale-up where 25%, 50% and 100% of uncircumcised, uninfected MSM engaging in insertive anal intercourse more than 50% of the time were circumcised over the course of 5 years. Impact was measured as cumulative fraction of new HIV infections averted over 10 years.

The predicted impact of MC varied substantially across settings (0%-17%). Countries with high existing levels of MC (e.g. USA or Ghana) would see a minimal impact (<3%/6% when circumcising 25%/50% of insertive MSM). The maximum impact of 8%/16% was observed in countries with low existing levels of MC (e.g. Peru and India). An upper bound of 38% impact is predicted using the 100% coverage intervention. The impact was most pronounced among MSM receiving MC, although herd effects were also observed among their partners. In uncertainty analysis, the intervention impact was positively correlated with role segregation in each setting (greater impact was seen when MSM had a strong preference for either insertive or receptive intercourse).

MC among MSM is likely to have the greatest impact in highly role-segregated settings with low MC coverage, such as Peru or India. However, our results suggest that the public health benefits among MSM would likely be modest, with the intervention unlikely to avert more than 17% of infections even in the most favorable of settings under realistically achievable coverage.