Background: Understanding the spread of HIV during HIV prevention trials has high potential to inform future intervention programming. Here, we provide a first characterisation of the residual age- and gender-specific transmission dynamics during the HPTN 071 (PopART) trial using orthogonal methods, and investigate the potential impact of suppressing transmissions from inferred source populations.
Methods: First, epidemic predictions were made using an individual-based model (IBM) calibrated on trial data. Model parameters of the sexual network were derived from surveys on sexual behaviour. Second, model predictions were tested against phylogenetic estimates obtained with phyloscanner from viral short-read next-generation sequencing (NGS) data from three trial communities in Zambia.
Results: Phylogenetic analysis identified 180 probable transmission pairs and the direction of transmission between them. 62% of these transmissions were from men to women, the same as was predicted by the IBM. The phylogenetic analysis predicted men to be 5.4 years than women in male-to-female transmission (IBM predicted 5.5 years), and 3.9 years older in female-to-male transmissions (the IBM predicted 2.9 years). The age distribution of transmitters agreed with modelling predictions, more closely after adjusting for sampling bias (figure 1). According to both the model and phylogenetics analysis, onwards transmissions peaked in 25-29 year old (y.o.) men and 20-24 y.o. women. Modelling the prevention of all transmissions from 25-29 y.o. men and 20-24 y.o. women reduced cumulative incidence over the trial period (mid-2014 to 2018) by 20% and 19% respectively.
Conclusions: Our results validate predictions of a mathematical model using phylogenetic data. These results support observations that there is a significant contribution of young people to HIV transmission in sub-Saharan Africa, especially 25 to 29 y.o. men. These results highlight that if universal testing-and-treatment (UTT) does not reach young people, and 25 to 29 y.o. men in particular, then the effect of UTT on reducing incidence may be limited.

Figure 1
[Figure 1]

Download the e-Poster