Protease inhibitors (PI) in combination therapy reduce HIV-related morbidity and mortality, although viral or cellular drug resistance can cause inadequate drug levels and therefore treatment failures. PI in particular have individually variable metabolization rates and therapeutic plasma drug level monitoring is currently under evaluation. As indinavir accumulates gradually over time in the hair, retrospective assessments are feasible .
The correlation between indinavir hair levels and treatment outcome with respect to drug resistance was longitudinally assessed. In February 1999, three indinavir-treated patients (stavudine, lamivudine, n = 2; zidovudine, zalcitabine, n = 1) and two patients taken off indinavir 3 months earlier (stavudine, lamivudine, saquinavir hard gel, n = 1; stavudine, didanosine, hydroxyurea, n = 1) had hair samples taken (mean viral load of 3.29 ± 1.02 log10 copies/ml). The hair was cut close to the scalp (1–2 cm area, more than 5 cm long) and crushed to powder. Indinavir concentrations were measured chromatographically . PI resistance was studied using direct sequencing (TruGene HIV-1 Assay, Visible Genetics, Toronto, Canada; GenBank/EMBL AJ401: 856, 857, 879, 892, 925, 937, 952, 956, 974, 966, 975), a line probe assay (INNO-LiPA HIV-1 PI, research version, Innogenetics, Ghent, Belgium) and recombinant virus drug susceptibility assay (indinavir from Merck Research Laboratories, West Point, USA). Viral load was measured by Quantiplex 2.0. (Chiron, Cergy-Pontoise, France) with a detection limit of 500 copies/ml. Specimens below this limit were assigned a viraemia level of 250 copies/ml. The slope of a weighted linear mixed-effects regression (SPSS, Chicago, USA; 1999) measured the association between indinavir hair concentration and the viral load baseline changes or the number of PI resistance mutations detected by both assays (46I, 48V, 54V, 82A/F/T, 84V, 90M; no 30N or 50V encountered), both assessed 3 monthly during follow-up, i.e. the years 1998–1999 (Fig. 1). Baseline data were obtained at the first PI administration (in 1996).
Two patients had concentrations of less than 5 μg/g. One of them had discontinued indinavir 3 months earlier. Patient 1 (□, 27.3 μg/g, Centers for Disease Control and Prevention stage C3) harboured a phenotypically susceptible virus to indinavir and did not develop any new clinical events during follow-up. In contrast, patient 4 (▿, 0.83 μg/g) had resistant virus (10-fold) but did not progress clinically from stage A3 during follow-up. From the three other patients (0.8–8.9 μg/g) with stages C2 and B3 at the beginning of follow-up, two experimented a new clinical event. The hair concentration was inversely associated with the virological response (P < 0.001, Fig. 1a), even after the exclusion of patient 1 or adjustment for baseline confounders (CD4 cell count, the number of previous nucleoside analogues used or Centers for Disease Control and Prevention clinical stage). The indinavir concentration was also inversely related to the number of mutations detected either by sequencing or line probe assay (P = 0.001, Fig. 1b). Both variables were dependent predictors of virological outcome (not shown).
Indinavir hair monitoring may be a useful tool to monitor indinavir impregnation on a wide window . This might allow the avoidance of potential dose-dependent side-effects and the adaptation of treatment strategies to evade the progressive emergence of resistance and failure. Larger studies are needed to evaluate whether the monitoring of indinavir in hair increases the predictive value for treatment outcome in addition to resistance testing.
The authors would like to thank J.M. Zimmer for editorial help.
1. Graham K, Koren G, Klein J. Determination of gestational cocaine exposure by hair analysis. JAMA 1989, 262: 3328 –3330.
2. Bernard L, Peytavin G, Vuagnat A, Truchis PD, Perrone C. Indinavir concentrations in hair from patients receiving highly active antiretroviral therapy. Lancet 1998, 352: 1757 –1758.