744    

Background:  Existing methods for therapeutic drug monitoring (TDM) provide information on treatment exposure over short intervals. Determination of drug levels in hair has been used to assess longer-term exposure for a broad array of drugs. Antiretroviral (ART) drug levels in hair could potentially measure cumulative drug exposure over weeks to months. Indinavir (IDV) levels in hair have previously been shown to correlate with virologic response and genotype. We developed methods to measure protease inhibitor (PI) levels in small samples of hair and assessed the association of those levels with treatment outcomes.

Methods:  Analytical methods for quantitating PI were developed using a liquid chromatography-tandem mass spectrometry system. Hair specimens were pulverized by cryo-milling at liquid nitrogen temperatures, and drug extraction was performed via 40% aqueous acetonitrile prior to analysis. Methods were validated using standard QA/QC protocols and applied to hair specimens collected from HIV-infected women on PI-based HAART. For 46 patients on lopinavir (LPV)/ritonavir (RTV)-based regimens, we examined the correlation of LPV and RTV levels in hair with virologic response. Responders achieved undetectable viral loads 6 months after initiating therapy and nonresponders had persistently detectable viral loads at 6 months.

Results:  For optimal analysis, 10 mg (approximately 10 hairs) was required; the limit of quantitation for all PI in hair was 0.5 ng/mg. Inter- and intra-day coefficient of variation for all analytes averaged < 15%. In terms of correlation with outcomes, LPV and RTV concentrations in hair were significantly higher in virologic responders versus nonresponders (Mann-Whitney exact test, p < 0.0001 for each drug). The odds of viral non-response was 5.3 (95% CI 1.5 to 18.8) times higher in patients with hair LPV levels below the median than in those at or above the median for all 46 patients. The odds of viral non-response was 21.5 (95% CI 4.5 to 103.9) times higher in patients with hair RTV levels below the median than in those above the median for all patients.

Conclusions:  Methods for determining PI levels in hair have been developed; these levels may correlate with long-term treatment outcomes on HAART. Hair can be stored at room temperature and can be analyzed months after collection. Measurement of hair ARV levels may provide an inexpensive, noninvasive, field-friendly method for TDM in the domestic and international HIV treatment and research settings.

Keywords: therapeutic drug monitoring; hair levels; lopinavir/ritonavir