A12 Modeling residual HIV replication and the emergence of drug resistance on ART

Abstract

There are conflicting reports regarding the presence of low-level HIV replication during suppressive antiretroviral therapy (ART). We simulated varying levels of replication and estimated the number of generations needed to obtain linked, drug resistance mutations to explore the effects of replication during ART. HIV replication was simulated with varying population sizes (10 to 3,000,000). Each population size was modeled ten times. Each genome was given a Poisson-distributed number of mutations according to its length and the average replication error rate (3.4 × 10−5 sub/nt/cycle). Simulations were run a maximum of 20,000 generations with endpoints defined as detection of a variant with resistance mutations to at least two ARVs. In all simulations, variants that were resistant to all three ARVs emerged in less than 20,000 generations. The time to emergence ranged from 148–16,156 generations in the various simulations, depending on the replicating population size (4.8 months to 44.3 years if the generation time is 1 day). Clinically detectable virologic failure can result from linkage of two mutations conferring resistance to two ARVs in a regimen. In our simulations, two linked mutations emerged in from 9 to 6,429 generations (9 days to 17.6 years). Our simulations suggest that in patients continually suppressed on ART for at least 10 years, the replicating population size would have to be less than ten, or virologic failure would have occurred from emergence of two ARV-resistant variants. Because most patients on ART do not experience virologic failure, our simulations suggest that any residual replicating population on ART is very small and thus not likely to either sustain or significantly contribute to the HIV reservoir.