Twenty-four-week safety and tolerability of nevirapine vs. abacavir in combination with zidovudine/lamivudine as first-line antiretroviral therapy: a randomized double-blind trial (NORA)

OBJECTIVE: To compare the safety/tolerability of abacavir and nevirapine in HIV-infected adults starting antiretroviral (ARV) therapy in Uganda. METHODS: Twenty-four-week randomized double-blind trial conducted with 600 symptomatic ARV-naive adults with CD4 <200 cells/mm(3) allocated to zidovudine/lamivudine plus 300 mg abacavir (A) and nevirapine placebo (n = 300) or 200 mg nevirapine (N) and abacavir placebo (n = 300) twice daily. The primary endpoint was any serious adverse event (SAE) definitely/probably or uncertain whether related to blinded nevirapine/abacavir. Secondary endpoints were adverse events leading to permanent discontinuation of blinded nevirapine/abacavir, and grade 4 events. RESULTS: Seventy-two per cent participants were women; 19% had WHO stage 4 disease; the median age was 37 years (range 18–66); the median baseline CD4 count was 99 cells/mm(3) (1–199). Ninety-five per cent completed 24 weeks: 4% died and 1% were lost to follow-up. Thirty-seven SAEs occurred on blinded drug in 36 participants. Twenty events [6 (2.0%) abacavir, 14 (4.7%) nevirapine participants] were considered serious adverse reactions definitely/probably/uncertain whether related to blinded abacavir/nevirapine [HR = 0.42 (95% CI 0.16–1.09) P = 0.06]. Only 2.0% of abacavir participants [six patients (0.7–4.3%)] experienced a suspected hypersensitivity reaction (HSR). In total 14 (4.7%) abacavir and 30 (10.0%) nevirapine participants discontinued blinded abacavir/nevirapine (P = 0.02): because of toxicity (6A, 15N; P = 0.07, all rash/possible HSR and/or hepatotoxicity), anti-tuberculosis therapy (6A, 13N), or for other reasons (2A, 2N). CONCLUSIONS: There was a trend towards a lower rate of serious adverse reactions in Ugandan adults with low CD4 starting ARV regimens with abacavir than with nevirapine. This suggests that abacavir could be used more widely in resource-limited settings without major safety concerns.