BACKGROUND: Much of the prior research on viral failure (VF) with integrase inhibitor (INSTI) therapy is based on results from trials rather than clinical care settings and little is known about recently approved medications such as dolutegravir (DTG). We compared VF in persons living with HIV (PLWH) who initiated DTG-based vs. other guideline recommended regimens in clinical care across the U.S.
METHODS: PLWH from 8 CFAR Network of Integrated Clinical Systems (CNICS) sites who started a recommended regimen between 8/2013-8/2016 were included. We compared DTG vs. other INSTI, and vs. darunavir-based (DRV) regimens included in current guidelines for initiating antiretroviral therapy (ART). VF was defined as a viral load of >400 copies/mL >6 months after initiation. We used Cox models adjusting for age, sex, race/ethnicity, hepatitis B, hepatitis C, tuberculosis, HIV risk factor, CD4 count, days since last HIV viral load, and site. PLWH were censored at death, regimen change or loss to follow-up (LTFU) with sensitivity analyses varying LTFU definitions from 0-12 months after last activity and including/excluding inverse probability censoring weights based on variables in the main models.
RESULTS: Among 6636 PLWH who initiated a recommended regimen, a lower proportion on DTG-based regimens experienced VF during follow-up (Figure). The adjusted hazard ratio (HR) for VF for DTG vs. DRV-based regimens was 0.56 (95% confidence interval 0.37-0.86). In sensitivity models, the HR for VF for DTG vs. other INSTI regimens ranged from 0.73-1.07 depending on LTFU definitions. The HR for DTG vs. DRV-based regimens ranged from 0.38-0.63 depending on LTFU definitions. In sensitivity analyses among the 1229 PLWH known to be ART-naïve at initiation, a similar pattern was found with a lower HR of VF among those who initiated DTG vs. DRV-based regimens (HR 0.25, 95% CI 0.11-0.56).
CONCLUSION: The observed rate of VF during follow-up was lower among PLWH initiating DTG-based vs. DRV-based regimens in routine clinical care at sites across the US. Results also demonstrated that different definitions of LTFU can have a large impact on the results and highlight the importance of sensitivity analyses in informing study definitions to minimize bias.
