Background:  The administration of antiretroviral drugs prior to HIV exposure (pre-exposure prophylaxis or PrEP) has gained considerable attention as a potential biomedical intervention to protect high-risk HIV^– people against infection. Several efficacy trials of daily PrEP with tenofovir (TDF) alone or with emtricitabine (FTC) are now ongoing among high-risk populations. A major concern is that drug resistance may develop in persons who fail PrEP or if PrEP is used by HIV-infected individuals.

Methods:  We constructed a complex mathematical model to predict the effects of PrEP on infections prevented and levels of transmitted drug resistance (TDR) in men who have sex with men (MSM) in San Francisco, California. Levels of TDR in this community are currently ~10% due to widespread usage of treatment over the past 2 decades. Our model was parameterized using clinical, behavioral, and epidemiological data from San Francisco. We also estimated parameters from virologic data from acute and chronic HIV infections in humans. We assumed that PrEP would differentially protect susceptible individuals against infection with wild-type or drug-resistant strains. Based on data from monkey models, we assumed that if individuals become infected and remain on PrEP they would have reduced acute viremia (hence be less infectious), a decreased disease progression rate but potentially develop drug resistance. To make predictions we employed uncertainty analyses to investigate a wide range of parameter values. We then used response hyper-surface modeling to conduct multivariate sensitivity analyses. To identify the impact of PrEP regimens, without confounding effects, we assumed that risk behavior remained constant.

Results:  We predict that PrEP, if used by 70% (median) of MSM and has a median efficacy of 60%, could increase TDR from 10 to a median of 31% while preventing ~58% (median) of infections over the next 10 years in San Francisco. Paradoxically, we find that increasing PrEP efficacy increases both the number of infections prevented and the proportion of cases of TDR. This occurs because increasing efficacy of PrEP does not reduce the number of infections with drug-resistant strains, but preferentially reduces the number of infections with wild-type strains.

Conclusions:  We show that, under certain conditions, PrEP could substantially reduce HIV incidence in San Francisco, but could also increase levels of TDR. Our results imply that it will be essential to develop new PrEP regimens with increased efficacy against drug-resistant strains.