Nevirapine‐associated toxicity in Niger

Abstract

The results reported by Phanuphak et al. [1] showed that the incidence of hepatitis and rash attributable to nevirapine (NVP) in Thai adults is similar to that reported in other populations. One of the main findings of that study was that patients' sex and baseline CD4 cell counts were not associated with the risk of hepatotoxicity or rash [1]. We conducted a similarly designed prospective cohort study in Niger where the first antiretroviral regimen is the generic fixed-dose combination (FDC) stavudine/lamivudine/nevirapine (d4T/3TC/NVP). All HIV-1-infected adult patients who were seen consecutively between November 2004 and January 2006 and for whom FDC was indicated were included in our study, with clinical and biological evaluation at baseline and during follow-up, including alanine aminotransferase (ALT) tests. Pregnant women were excluded. Hepatitis B and C tests were not available. Cutaneous and hepatic reactions were identified and potential risk factors for cutaneous reactions, hepatotoxicity overall and severe hepatotoxicity (ALT level ≥5-fold the upper limit of normal) were investigated using Cox proportional hazards modelling. The median age of the patients was 37 years [interquartile range (IQR) 31–43 years] and 46% were women. The median nadir CD4 count was 126 cells/μL (IQR 59–202 cells/μL). The median duration of follow-up was 13 months (IQR 7–18.4 months). Of the 540 patients included in the study, seven had cutaneous adverse reactions (1.3%), including five cases of rash and two cases of pruritus. The outcome was always favourable. The time to onset was 18 days (IQR 14–29 days). The incidence of rash was not significantly associated with sex, age or CD4 cell count. Of the 497 patients with at least two ALT tests, 78 episodes of hepatotoxicity (15.7%), including six cases of grade III hepatotoxicity (1.2%), were identified. There was no grade IV hepatotoxicity or symptomatic hepatotoxicity. The time to onset was 29 days (IQR 14–181 days). The risk of hepatotoxicity was lower in men with CD4 counts >50 cells/μL (P=0.05). Abnormal baseline ALT level increased the risk of severe hepatotoxicity (P=0.003). The incidence of NVP-associated rash in our study is thus surprisingly low compared with other series (4.3–36%) [2]. The difference among the incidences of cutaneous adverse events in Niger (1.3%), Uganda (6%) and Thailand (16.1%) has no clear explanation [1, 3]. Loss to follow-up bias does not seem significant (9% of our patients). There may be some pharmacogenetic differences among Asian, Europeans and Africans. However, our study was not sufficiently powered to detect differences in risk factors. The incidence of NVP-associated severe hepatotoxicity also seems to be surprisingly low in our study, as compared with values reported by others (6–8%) [4]. Particularly, the absence of symptomatic hepatotoxicity, the incidence of which was reported as approximately 5% in the review of Dietrich [4] and 4.6% in Thailand [1], is striking. Nonetheless, this value is quite similar to those reported in studies undertaken in other resource-poor settings such as in the multicentric observational cohort of Medecins Sans Frontieres (MSF) (2.2% severe hepatotoxicity) [5], in Uganda (≤0.5%) [3] and in Thailand (1.7%) [1]. This indicates that NVP is safe if monitoring includes ALT tests and if NVP is discontinued in the case of toxicity. In addition, the tolerance of NVP could be improved if hepatitis B and C status were known, given that the incidence of these infections is high in resource-poor settings and that they have been associated with hepatic adverse events in patients on antiretroviral therapy [4]. Patients should also be tested for abnormal baseline transaminase levels, another well-known risk factor for hepatotoxicity [6]. In conclusion, this study shows that NVP is better tolerated in Niger than in Western countries and other developing countries. It is therefore suggested that NVP should still be used in sub-Saharan countries as a first-line treatment. The authors of the original paper [1] were invited to comment, but they declined to do so. The Editors