Abstract Body

The FDA Snapshot algorithm defines success of antiretroviral therapy (ART) by viral load < 50 cp/mL without ART change (mostly for adverse events [AEs]). Notably, a viral load <50 cp/mL but with a drug-related AE is regarded as treatment success. However, drug-related AEs might increase the risk of ART failure.

We hypothesised that an efficacy algorithm incorporating ART-related AEs would be clinically meaningful, as they would predict ART failure. We analysed individual-patient data from SINGLE, a placebo-controlled trial of abacavir-lamivudine-dolutegravir (ABC-3TC-DTG) vs. tenofovir-emtricitabine-efavirenz (TDF-FTC-EFV). Data were obtained via clinicalstudydatarequest.com/ after independent protocol review by the Wellcome Trust (but not the study sponsor). We investigated if drug-related AEs through Week 12 predicted subsequent ART failure, and failure rates using the Snapshot algorithm vs. a ‘Snapshot-Plus’ algorithm that additionally regards Grade-2+ drug-related AEs as ART failure. Groups were compared through Week 144, as were strata ≥ or <100,000 cp/mL at baseline. Data were analysed by calculating the proportion of responders (95%CI) with both algorithms. Chi-square and McNemar tests compared responders within and between algorithms, respectively. Logistic regression determined the odds of failure by the Snapshot algorithm for subjects experiencing drug-related AEs to week 12.

Grade 2-4, drug-related AEs through Week 12 significantly increased the risk of ART failure at Week 48 by Snapshot (OR 2.68 [95%CI 1.75, 4.11]), and at Weeks 96 and 144; the relationship with any-grade, drug-related AEs through Week 12 was not significant. At Week 48, Snapshot efficacy was 87.9% with ABC-3TC-DTG and 80.1% with TDF-FTC-EFV (Table). ‘Snapshot-Plus’ response rates were substantially lower (76.6% and 62.8%, respectively). DTG minus EFV group differences were significantly greater with ‘Snapshot-Plus’ at each time point (13.8% vs. 7.3% at Week 48; McNemar’s P<0.001), including in the stratum with baseline viral load <100,000 cp/mL (15.0% vs. 7.7% at Week 48; P<0.001).

In this trial, grade 2-4, AE-free ART efficacy was lower than estimated by the standard Snapshot algorithm. The ‘Snapshot-Plus’ algorithm may be a more clinically relevant measure of ART efficacy, and may better distinguish regimens when used in patients with viral load <100,000 cp/mL. The algorithms should be compared using data from other trials.