644    

Background:  Both self-reported adherence and sub-optimal plasma drug concentration have been independently associated with virologic failure, however a threshold has not yet been defined in these 2 determinants for an increasing risk of rebound in patients with virologic success.

Methods:  Longitudinal cohort analysis. Self-reported non-adherence rate was calculated as the number of pills taken divided by the number prescribed over the last 3 days. Plasma levels of nucleoside reverse transcriptase inhibitors (NRTI), protease inhibitors (PI), and non-NRTI (NNRTI) were concurrently assessed. A pharmacokinetic score from 0 to 3 was derived for NNRTI/PI, identifying undetectability (0) and sub-optimal (1), optimal/expected (2), and higher than optimal/expected (3) ranges of concentration. NRTI backbone was scored 0 or 1 according to detectability of drugs adjusted for half-life. An overall marker of plasma drug concentration was calculated as the number of classes with pharmacokinetic score >1 in current regimen.

Results:  Included in this analysis are 142 patients with a viral load £ 80 copies/mL at baseline (time of plasma concentration test). Crude rates (per 100 person-years) of virologic rebound were 11.0 (95% CI 6.3 to 17.9) in patients with 100% self-reported non-adherence, 24.1 (95% CI 9.7 to 49.7) in those with 90 to 99% adherence, and 8.3 (95% CI 0.2 to 4.6) in those with < 90% adherence. According to plasma drug concentration, patients without any class with pharmacokinetic score ≥1 had the greatest incidence of VR (75.0; 95%CI 34.3 to 142.4) compared with those with 1 (11.5; 95% CI 4.2 to 25.1) or 2 (8.0; 95% CI 3.7 to 15.3) classes with pharmacokinetic score ≥ 1. In a multivariable Poisson model, the only factors associated with virologic response were reporting 90 to 99% adherence (8.14 compared with 100% adherence; 95%CI 2.18 to 30.30, p = 0.002) and classes of antiretrovirals with detectable concentrations (0.27 per each class with pharmacokinetic score ≥ 1; 95%CI 0.11 to 0.63, p = 0.003). None of the scores for individual drug classes were significantly associated with the outcome if considered alone. There was a trend for the association between non-adherence and virologic response to be stronger in patients receiving single-PI (RR = 8.14, 95%CI: 2.18 to 30.30) than in those receiving NNRTI (RR = 2.98, 95% CI 0.36 to 24.8, p value for interaction = 0.17).

Conclusions:  Patients reporting 90% to 99% adherence, even if with undetectable viremia at questionnaire, were at remarkably increased risk of subsequent failure compared with patients declaring perfect adherence. A global score of plasma drug concentration that incorporates all antiretroviral classes, including a half-life–adjusted score for NRTI, seems to be more predictive than concentration from any singular drug class.

Keywords: drug plasma level; adherence; haart