Abstract Body

Mechanisms for long-term HIV persistence despite antiretroviral therapy (ART) continue to be debated. Many studies have offered evidence that ART halts self-sustaining viral replication in most patients and that long-lived resting memory CD4+ T-cells support a stable latent reservoir (LR) of integrated virus. However, Lorenzo-Redondo et al. [Nature, 2016] deep-sequenced lymph node and blood samples from three participants during the first six months of ART, finding viral genetic signals of persistent replication and evolution. We show that these signals are expected outcomes of the known multi-phasic decay of infected cells during ART and do not provide evidence of ongoing replication.

We designed a simulation of HIV infection before and after ART, including changes in viral population size, multiple infected cell types with different lifespans, and mutation/selection within and outside sequenced regions. To test the hypothesis that the observed genetic signals could arise without replication, ART was assumed to block all new infection. To estimate infected cell lifespans, we fit a multi-phase decay model to longitudinal LR measurements from early-treated individuals. Roughly 12,000 simulations were run with a range of plausible parameter values; those with realistic levels of genetic diversity and divergence were included in analyses. To reduce reliance on assumed parameter values, we repeated our analysis using actual sequence data from acute infection to seed a hypothetical LR.

At zero, three, and six months following ART initiation, short-lived viral populations comprise >99%, 96%, and 76% of infected resting cells, masking the persistent reservoir. Applying the same population genetic and phylogenetic methods previously used to support claims of ongoing replication in lymphoid tissue, we found that up to 57% of simulations generated a false signal of ongoing replication in all tests. Time-structured phylogenies can also yield a misleading appearance of evolution (Figure). Analysis of acute infection data suggests that individuals with major sequence changes before treatment (CTL escape) produce a false signal of ongoing replication 92% of the time.

Investigation of ongoing replication during ART must wait >1 year following the start of ART; earlier analysis is unreliable. Where possible, population genetic studies of viral evolution should be conducted with reference to specific models of viral dynamics and literature on growth/decay of subpopulations.