Mechanisms of Virologic Failure in Previously Untreated HIV-Infected Patients From a Trial of Induction-Maintenance Therapy

CONTEXT: In the Trilège trial, following induction with a zidovudine, lamivudine, and indinavir regimen, human immunodeficiency virus (HIV) replication was less suppressed by 2-drug maintenance therapy than by triple-drug therapy. OBJECTIVE: To identify mechanisms of virologic failure in the 3 arms of the Trilège trial. DESIGN: Case-control study conducted from February to October 1998. SETTING: Three urban hospitals in Paris, France. PATIENTS: Fifty-eight case patients with virologic failure (HIV RNA rebound to >500 copies/mL in 2 consecutive samples) randomized to 3 therapy groups: triple drug (zidovudine, lamivudine, and indinavir), 8; zidovudine-lamivudine, 29; and zidovudine-indinavir, 21; the case patients were randomly matched with 58 control patients with sustained viral suppression. MAIN OUTCOME MEASURES: At virologic failure (S1 sample) and 6 weeks later (S2 sample), assessment of protease and reverse transcriptase gene mutations, plasma indinavir level, and degree of viral load rebound; pill count during induction and maintenance periods. RESULTS: Only 1 primary resistance mutation, M184V, was detected in S1 plasma samples from 4 of 6 patients in the triple-drug and in all 22 in the zidovudine-lamivudine therapy groups and in S2 plasma samples from 3 of 6 in the triple-drug and 20 of 21 in the zidovudine-lamivudine groups. Of controls, M184V was detected in 11 of 13 S1 plasma samples and in 10 of 11 S2 plasma samples. Indinavir levels were undetectable in all S1 samples but 2 in 7 triple-drug cases tested and in the expected range in 11 of 18 S1 and 5 of 12 S2 zidovudine-indinavir case plasma samples tested. Maintenance adherence rates were lower for cases vs controls for zidovudine (P = .05) and indinavir (P = .05). Low indinavir levels, lower adherence rates for zidovudine (P = .04) and lamivudine (P = .03), and rebound to near-baseline values suggested adherence as cause of early failure for 4 of 8 triple-drug cases. In the zidovudine-lamivudine arm, for which case and control adherence rates did not differ significantly (P = .96), most failures occurred late with low rebound, suggesting suboptimal drug potency. In the zidovudine-indinavir arm, virologic failures may be related to both mechanisms. CONCLUSIONS: During the maintenance phase early and late virologic failures appeared to be related more to problems of adherence and antiretroviral treatment potency, respectively, than to selection of resistant mutant viruses.