Background:  Access to ART is increasing in South Africa with tenofovir (TDF)+lamivudine (3TC)/emtricitabine (FTC)+efavirenz (EFV)/nevirapine (NVP) as the first-line regimen. TDF and TDF+FTC are promising pre-exposure prophylaxis (PrEP) agents, but they overlap with first-line ART raising concerns about drug resistance. Here we predict the combined impact of ART and PrEP rollout on the spread of HIV and drug resistance using mathematical modeling.

Methods:  We developed and analyzed deterministic and stochastic models that incorporate heterogeneity in sexual behavior, HIV transmission, HIV disease progression, and emergence of drug resistance to represent the heterosexual HIV epidemic in South Africa with ART and PrEP rollout. Model output includes:  cumulative new infections prevented, and prevalence of transmitted and acquired resistance from ART and PrEP. Uncertainty and sensitivity analyses were performed to determine prediction uncertainty and sensitivity to parameter variability. Scenario analyses were used to examine the effect on outcomes of specific inputs as well as inadvertent PrEP use.

Results:  ART alone initiated at CD4 counts ≤200 cell/µL with 80% coverage decreases cumulative new infections over 10 years by 21% but increases prevalence of drug resistance to 8.5% (3% transmitted and 5.5% acquired resistance). In an optimistic scenario of ART combined with PrEP that assumes 70% PrEP efficacy, 80% adherence and 50% coverage, new infections decrease by 38% and resistance prevalence increases to 11.4% (4.1% transmitted and 6.3% acquired resistance from ART; 0.4% transmitted and 0.7% acquired resistance from PrEP). Including inadvertent PrEP use among acutely infected persons in the antibody-negative phase has minimal effect on outcomes. By contrast, inadvertent PrEP use among 5% of infected persons increases resistance prevalence to 14.6%. A realistic scenario of ART plus PrEP assuming 50% PrEP efficacy, adherence and coverage with PrEP dropout rate of 20%/year, decreases infections by 25% and increases resistance prevalence to 9.5%. Uncertainty analysis of 50,000 simulations predicts a median 28% (IQR 25 to 31%) decline in infections and 9% (IQR 8 to 11%) prevalence of drug resistance after 10 years of ART and PrEP rollout.

Conclusions:  ART combined with PrEP is likely to have a bigger HIV prevention impact than either strategy alone, but overlapping drugs will increase drug resistance prevalence. ART is predicted to contribute more to resistance than is PrEP. However, inadvertent PrEP use by persons with established HIV could increase resistance from PrEP.