Background: A cure for HIV requires the interruption of antiretroviral therapy (ART) to assess the efficacy of curative interventions. Study designs using time to virologic rebound (TTR) as the primary endpoint aim to improve patient safety compared to strategies that expose participants to prolonged periods of viremia, but the safety of treatment interruption (TI) in women and ethnically diverse populations remains poorly characterized. We took advantage of a completed randomized trial in order to describe adverse events (AEs) during the aviremic period preceding TTR in women stopping ART compared to women who continued ART postpartum and clinical predictors of time to rebound.
Methods: 1,691 asymptomatic women living with HIV with baseline CD4+ counts ≥ 350 cells/mm3 and undetectable viral loads were randomized to stop or continue triple ART postpartum in the PROMISE trial. Primary outcomes included comparison of time to first composite AE prior to viral rebound between the two study arms, as well as clinical predictors of TTR in women stopping therapy. Secondary outcomes included comparison of baseline characteristics between post-treatment controllers (PTCs) and non-controllers and rates of acute retroviral syndrome (ARS). Follow up time was 48 � 2 weeks.
Results: After adjusting for CD4+ nadir, duration of ART, and pre-treatment viral load, women in the stop arm had a lower risk of early events compared to women who continued ART (HR 3.56 (95% CI 1.81, 7.02), p<0.001). There were no serious adverse events observed. Being from African regions was also protective of early events compared to other regions. AEs after stopping therapy were characterized mostly by mild neutropenia, anemia, headache, and diarrhea, with a combined incidence of 8 per 1000 person-years in the first 2 weeks after stopping therapy. Events beyond 6 weeks were rare. Estimated TTR was 1.86 weeks, with the projected proportion remaining virally suppressed at 12, 24, and 48 weeks 5.5% (4.3, 5.9%), 2.0% (1.4, 2.7%), and 0.7% (0.4, 1.1%), respectively. Higher CD4+ nadir (HR 0.4 (0.28, 0.57), p<0.001) and older age (HR 0.97 (0.96, 0.99), p=0.001) were predictive of delayed TTR; however, in a subgroup analysis (N=412) that included pre-treatment viral load, low pre-treatment HIV-1 RNA was the only predictor of delayed TTR (HR 1.7 (1.48, 1.95), p<0.001). PTCs made up 16.7% of the study population, were slightly older with higher BMI at baseline, and had significantly lower pre-ART viral loads compared to non-controllers. Incidence of ARS was <5%.
Conclusion: In young, healthy, postpartum women living with HIV with high CD4+ counts, adverse events were rare in the period after TI preceding viral rebound. Pre-treatment viral load was the strongest predictor of TTR, and TTR was similar to historical cohorts despite brief ART duration and chronic infection. Future cure studies with brief TIs using TTR as the primary endpoint may be considered safe in low-risk populations, and inclusion of diverse participant populations is necessary to characterize the impact of race/ethnicity and sex hormones on the viral reservoir.