`Efavirenz plasma levels can predict treatment failure and central nervous system side effects in HIV-1-infected patients'

The results from our study on efavirenz, showing an association between plasma levels and treatment outcomes, were derived from a prospective observational study. So they are not sufficient to impose a change in therapeutic practice, but they call for validation through an intervention trial. It had been determined that the analysis would only include patients under efavirenz for at least 3 months (mean duration of efavirenz treatment was 8 months). So they all had the chance to reach steady-state for efavirenz pharmacokinetics, liver enzyme induction and viral kinetics. Treatment failure was defined as viraemia rebound to over 400 copies/ml, and therefore represented actual instances of virological failure. The choice of 400 copies/ml as the cut-off to define viral suppression was imposed by the detection limit of the Amplicor test (Roche Diagnostics, Basel, Switzerland) used in the majority of patients. In a log–log regression, the association between efavirenz concentrations and viraemia values persisted (r = 0.34, P = 0.0001). Although we used body mass index in the model, the use of body weight did not demonstrate any influence on efavirenz pharmacokinetics (P = 0.97). López-Cortés et al.[1] indicated a half-life of 14 h, whereas our regression shows 13 h. This means that the concentrations are expected to decrease by a factor of approximately 2 during the sampling interval of 8–20 h post-dosing. This is less than the relative width defining the ‘optimal range’ (1000–4000 μg/l), making this influence of limited importance in concentration–outcome relationships. The mean sampling times across the efavirenz concentration ranges defined post-hoc (< 1000; 1000–4000; > 4000 μg/l) were of the same order (15.9 ± 3.1; 13.8 ± 2.5; 14.4 ± 3.1 h, respectively). Including the sampling time in the analysis did not change the observed relationships between drug levels and outcomes. However, interpreting a plasma level in a therapeutic drug monitoring programme based on population pharmacokinetics will surely require that the sampling time is taken into account. Poor adherence is certainly a relevant factor explaining both low efavirenz concentrations and viral failures. Therapeutic drug monitoring could therefore represent a valuable tool to ameliorate the patient's adherence, response to treatment and general condition. We believe that future studies will confirm the usefulness of efavirenz dose individualization based on plasma determinations. Catia Marzolini Amalio Telenti Laurent Decosterd Jérôme Biollaz Thierry Buclin