Targeting HIV Prevention Based on Molecular Epidemiology Among Deeply Sampled Subnetworks of Men Who Have Sex With Men

Abstract

Background: Molecular epidemiology can be useful in identifying clusters of human immunodeficiency virus (HIV) transmission that can be targeted for prevention.

Methods: Regular screening of 2000 men who have sex with men (MSM) in Beijing, China, for HIV infection every 2 months identified 179 primary infections (2007-2010). HIV-1 pol sequences were obtained and used to infer the transmission network and identify transmitted drug resistance (TDR) among these individuals. We evaluated the use of clinical and network information to target prevention efforts. Prevention efficiency was calculated as the number of infections saved per number of interventions.

Results: This cohort was infected with HIV-1 subtype B (28%), circulating recombinant form (CRF)_01 AE (53%), and CRF_07 BC (16%). The overall rate of TDR was low (5%), but the rate of clustering was high (64%), suggesting deep sampling of the subnetwork. Provision of a theoretically high-efficacy intervention like antiretroviral therapy to all participants had a prevention efficiency of 23%. The efficiency of targeting prevention based on lower CD4 counts (<200 cells/mL, <350 cells/mL, or <500 cells/mL) and higher viral loads (>100 000 copies/mL and >50 000 copies/mL) was between 10% and 18%. The efficiency of targeting prevention based on number of network connections was much higher (30%-42%). For example, treating the 33 participants with ≥5 connections in 2009 would have theoretically prevented 14 infections in 2010 (42% prevention efficiency).

Conclusions: Regular HIV testing of MSM in Beijing can deeply sample the local transmission subnetwork, and targeting prevention efforts based on network connectivity may be an efficient way to deliver prevention interventions.

Keywords: China; HIV; MSM; molecular epidemiology; targeted prevention.

Figure 1.

Maximum likelihood trees for human immunodeficiency virus type 1 pol sequences generated from our cohort. Sequences classified as (A) CRF01_AE (green) and subtype B (blue) and (B) CRF07_BC (red) are shown. Phylogenetic bootstrap values ≥95% are denoted by an asterisk (*).

Figure 2.

Antiretroviral therapy (ART) targeted based on network connectivity. Here, we show that when ART is targeted to individuals (left y-axis) with higher numbers of connections in 2009 (x-axis) the targeted prevention efficiency (right y-axis) of ART increases. For example, if all 134 individuals had been provided ART in 2009 (♦), then 31 new infections would have been predicted to be prevented in 2010 (□), with a targeted prevention efficiency (▴) of 23% (number of infections saved/number of people treated with ART). If ART had only been provided to those 33 individuals with ≥5 connections in 2009, then 14 infections would have been predicted to be prevented in 2010 with a targeted prevention efficiency of 42%.