Patient-Reported Symptoms Over 48 Weeks Among Participants in Randomized, Double-Blind, Phase III Non-inferiority Trials of Adults with HIV on Co-formulated Bictegravir, Emtricitabine, and Tenofovir Alafenamide versus Co-formulated Abacavir, Dolutegravir, and Lamivudine

Background Integrase strand transfer inhibitors (INSTIs) are recommended for first-line antiretroviral therapy in combination with two nucleos(t)ide reverse transcriptase inhibitors. Co-formulated bictegravir, emtricitabine, and tenofovir alafenamide (B/F/TAF), a novel, INSTI-based regimen, is currently approved in the US and EU for the treatment of HIV-1 infection and recommended as first-line treatment in current guidelines. In our current analysis, we aimed to determine changes in patient-reported symptoms over time among HIV-1-infected adults who initiated or switched to B/F/TAF versus another INSTI-based regimen, co-formulated abacavir, dolutegravir, and lamivudine (ABC/DTG/3TC). Methods A planned secondary analysis of patient-reported outcomes was conducted for two double-blind, randomized, phase III studies in HIV-1-infected adults comparing B/F/TAF with ABC/DTG/3TC: one in treatment-naïve individuals (GS-US-380-1489, ClinicalTrials.gov NCT02607930) and the other in virologically suppressed participants (GS-US-380-1844, ClinicalTrials.gov NCT02603120). In both studies, the HIV symptoms distress module (HIV-SI) was administered at baseline (BL) and weeks 4, 12, and 48. Responses to each of the 20 items were dichotomized as bothersome or not bothersome. Treatment differences were assessed using unadjusted and adjusted logistic regression models (adjusted for BL HIV-SI count, age, sex, BL Veterans Aging Cohort Study [VACS] Index, medical history of serious mental illness, BL Short Form [SF]-36 Physical Component Summary [PCS], BL SF-36 Mental Component Summary [MCS], and, for virologically suppressed participants only, years since HIV diagnosis). We conducted longitudinal modeling of bothersome symptoms using a generalized mixed model including treatment, time, time-by-treatment, and additional covariates from the adjusted logistic regression model as described above. The Pittsburgh Sleep Quality Index (PSQI) was administered at the same frequency as the HIV-SI, and the total score was dichotomized as good or poor sleep quality. Similar models to those used for HIV-SI were applied, using BL sleep quality and BL SF-36 MCS as covariates. Statistical significance was assessed using p