Residual phenotypic susceptibility to doravirine in multidrug-resistant HIV-1 from subjects enrolled in the PRESTIGIO Registry

Objectives: Doravirine shows a rather distinct resistance profile within the nonnucleoside reverse tran- scriptase inhibitor (NNRTI) class. This study aimed to evaluate the phenotypic susceptibility to doravirine, rilpivirine and etravirine in a panel of multidrug-resistant (MDR) HIV-1 isolates collected from people living with HIV (PLWH) enrolled in the PRESTIGIO Registry. Methods: Recombinant viruses expressing PLWH-derived protease, reverse transcriptase coding regions were generated from plasma samples at virological failure with documented resistance to protease in- hibitors, nucleoside reverse transcriptase inhibitors, NNRTIs and integrase strand transfer inhibitors. In vitro susceptibility was assessed through a phenotypic assay measuring fold-change values with respect to the reference NL4-3 virus. Genotypic susceptibility was computed by the Stanford HIVdb algorithm 8.9-1. Results: Plasma samples were collected from 22 PLWH: 20 (91%) were male, median age 55 years (IQR 50–58), time since HIV-1 diagnosis 27 years (23–31) and time on antiretroviral treatment 23 years (22– 26). Median doravirine, etravirine and rilpivirine fold-change values were 9.8 (2.9–40.4), 42.9 (3.1–100.0) and 100.0 (17.9–100.0), respectively. According to the fold-change cut-offs, full susceptibility was observed in five (23%), four (18%) and one (5%) cases with doravirine, etravirine and rilpivirine, respectively. Ir- respective of the presence of specific doravirine mutations, higher numbers of NNRTI mutations corre- lated with higher fold-change values for doravirine. By comparing the distribution of fold-change values with the Stanford HIVdb predicted susceptibility, a significant correlation was detected for doravirine and rilpivirine but not etravirine. Conclusion: Despite extensive cross-resistance among NNRTIs, doravirine can be a valid option in a pro- portion of PLWH with MDR HIV-1. Doravirine activity appeared to be inferred with fair accuracy by the HIVdb algorithm.