Can HIV-1 incidence be estimated from plasma viral load and sexual behaviour?

To estimate the impact of sexual practices and HIV-1 viral load status on HIV-1 incidence, we surveyed 38 men who have sex with men (MSM) on their sexual practices in the past three months. HIV-1 RNA load was measured in semen and blood contemporaneously. A Bernoulli model was developed incorporating seminal plasma viral load (SPVL), number and serostatus of partners, and number of protected and unprotected episodes of anal intercourse. Probability of transmission according to SPVL was determined by sensitivity analysis based on the correlation between blood plasma viral load (BPVL) and SPVL. There was a BPVL threshold below which SPVL was low or undetectable and above which SPVL increased geometrically. Seven subjects infected 0.93 HIV-1-negative partners and 13 subjects infected 4.28 unknown serostatus partners. Probability estimates were heavily skewed by a small number of subjects with high rates of unprotected sex and multiple sexual partners. We conclude that more HIV-1 infections may occur from increased episodes of unprotected sex with multiple partners of unknown HIV-1 serostatus. The model can be used to counsel individuals or predict epidemics, and to assess behaviour change or the impact of public health interventions.