Mortality using raltegravir versus other integrase strand-transfer inhibitors in people with HIV in Europe and Australia: a prospective multicentre study.

Background: Integrase strand-transfer inhibitors (INSTI) are a key part of contemporary antiretroviral therapy (ART). Raltegravir (RAL) was the first INSTI and remains recommended for some people with HIV. We investigated all-cause mortality between RAL-based ART and other INSTIs in the RESPOND cohort consortium among both ART-naïve and treatment experienced individuals. Methods: RESPOND, a multicenter prospective cohort study, includes approximately 40,000 adults (≥18 years) with HIV from 17 cohorts across Europe and Australia. Individuals eligible for inclusion into RESPOND had ≥1 clinical visit at a site participating in RESPOND after January 01, 2012, and a CD4 count and HIV viral load measurement available at inclusion. Participants in RESPOND who started their first INSTI between JAN 01, 2012 and DEC 31, 2021 were included. All-cause mortality among those starting RAL was compared to those starting any other INSTI using Cox proportional hazards regressions: one model adjusting for age and another weighted by inverse probability of treatment weights (IPTW). Predictors of starting RAL were estimated by logistic regression. Findings: Among 20,349 participants starting an INSTI (15,429 (75.8%) male, 4879 (24.0%) female, and 41 (0.2%) transgender), 938 (4.6%) died during 94,677 person-years of follow-up (PYFU). Crude mortality rates (MR) were higher for participants starting RAL (MR 12.9 per 1000 PYFU; 95% CI 11.5-14.5) than other INSTIs (MR 9.1 per 1000 PYFU; 95% CI 8.4, 9.8). Starting RAL was significantly associated with higher mortality when controlling for age (adjusted hazard ratio (aHR) 1.43; 95% CI 1.25, 1.65). However, after applying IPTW, there was insufficient evidence for a difference in mortality in the full cohort (hazard ratio (HR) 1.13; 95% CI 0.93, 1.34) or among ART-naïve participants (HR 1.23; 95% CI 0.71, 2.12). Starting RAL was associated with higher HIV viral load, hepatitis C positive status (aOR 2.07; 95% CI 1.82, 2.37), prevalent end-stage renal disease (aOR 2.58; 95% CI 1.58, 4.19), chemotherapy near baseline (aOR 1.58; 1.01, 2.48), and cardiovascular disease (aOR 1.58; 95% CI 1.30, 1.91). Interpretation: In this large and well-characterised cohort we found no evidence of an association between all-cause mortality and use of RAL compared to other INSTIs after accounting for confounding at the time of starting the INSTI. Our findings suggest that prior reports of such an association could have been confounded by indication and channelling bias. While a large number of potential confounders were accounted for, the results presented are an estimation of average treatment effect using IPTW which is still vulnerable to uncontrolled confounding. Funding: CHU St Pierre Brussels HIV Cohort, Austrian HIV Cohort Study, Australian HIV Observational Database, AIDS Therapy Evaluation in the Netherlands National Observational HIV cohort, EuroSIDA cohort, Frankfurt HIV Cohort Study, Georgian National AIDS Health Information System, Nice HIV Cohort, ICONA Foundation, Modena HIV Cohort, PISCIS Cohort Study, Swiss HIV Cohort Study, Swedish InfCare HIV Cohort, Royal Free HIV Cohort Study, San Raffaele Scientific Institute, University Hospital Bonn HIV Cohort, University of Cologne HIV Cohort, Brighton HIV Cohort, and the National Croatian HIV cohort, ViiV Healthcare, Merck Life Sciences, Gilead Sciences, and the Centre of Excellence for Health, Immunity Infections (CHIP).