Background Virologic failure (VF) is still a major concern in the use of cabotegravir (CAB) and rilpivirine (RPV) long-acting (LA) for many healthcare professionals (HCP). While many results from clinical trials have been published, there is suspicion that they might underestimate the risk under less-controlled real-life conditions. This study aimed to estimate the probability of VF (primary objective) as well as discontinuation for any reason (secondary objective) among people with HIV (PWH) on CAB and RPV LA every two months (Q2M) in real life using Bayesian methodology. Methods Bayesian estimation of VF based on prior knowledge about VF and discontinuation on CAB and RPV Q2M from randomized controlled trials (ATLAS-2M, SOLAR) and real-world data from CAB and RPV LA from a large single-center cohort. Results Among 175 PWH, two (1.1%) met the criteria of VF through week 48 (W48), resulting in an estimated risk of VF at W48 of 1.2% [0.6%; 1.9%] using Bayesian estimation. In one of the PWH, two-class resistance was observed at the time of VF, most likely being therapy emerging. The probability of discontinuation for any reason by W48 was 21.4%, leading to a Bayesian risk estimate of 8.9% [7.3%; 10.5%]. The main reasons for discontinuation were injection-site reactions (n=10). Conclusions Risk of VF on CAB and RPV LA under real-life conditions seems to be comparable to results in clinical trials. This finding can be reassuring for both PWH and HCPs considering CAB and RPV LA as an alternative to oral antiretroviral treatment, particularly when considering the risk factors for VF that have been identified from the cases of VF in the clinical trials for patient selection. At the same time, rates of discontinuation may be considerably higher. However, this does not seem to be an indicator of a worse safety profile outside clinical trials, but probably could be the result of making CAB and RPV LA available to a wider population of PWH.
Keywords: dual antiretroviral therapy; hiv-1; injection-site reaction; long-acting injectable; real-world data; virologic failure.
Copyright © 2025, Noe et al.