Antiretroviral Hair Levels, Self-Reported Adherence and Virologic Failure in Second-Line Regimen Patients in Resource-Limited Settings.

OBJECTIVE To evaluate associations between hair antiretroviral (ARV) hair concentrations as an objective, cumulative adherence metric, with self-reported adherence and virologic outcomes. DESIGN Analysis of cohort A of the ACTG-A5288 study. These patients in resource-limited settings were failing second-line PI-based ART, but were susceptible to ≥1 NRTI and their PI, and continued taking their PI-based regimen. METHODS ARV hair concentrations participants taking 2 NRTIs with boosted atazanavir (n = 69) or lopinavir (n = 112) were analyzed at weeks 12, 24, 36 and 48 using liquid-chromatography-tandem-mass-spectrometry assays. Participants self-reported percentage of doses taken in the previous month; virologic failure (VF) was confirmed HIV-1 RNA≥1000 copies/mL at week 24 or 48. RESULTS From 181 participants with hair samples, (61% female median age: 39y; CD4 count: 167 cells/uL; HIV-1 RNA: 18,648 copies/mL), 91 (50%) experienced VF at either visit. At 24 weeks, median hair concentrations were 2.95 (IQR 0.49-4.60) ng/mg for atazanavir, 2.64 (IQR 0.73-7.16) for lopinavir, and 0.44 (IQR 0.11-0.76) for ritonavir. Plasma HIV-1 RNA demonstrated inverse correlations with hair levels (rs -0.46 to -0.74) at weeks 24 and 48. Weaker associations were seen with self-reported adherence (rs -0.03 to -0.24). Decreasing hair concentrations were significantly associated with VF, the hazard ratio (HR) (95%CI) for ATV, LPV and RTV were 0.69 (0.56-0.86), 0.77 (0.68-0.87), and 0.12 (0.06-0.27) respectively. CONCLUSIONS PI hair concentrations showed stronger associations with subsequent virologic outcomes than self-reported adherence in this cohort. Hair adherence measures could identify individuals at risk of 2nd-line treatment failure in need of interventions.