This study in Cameroon showed a higher hepatotoxicity in HBV or HCV-coinfected patients receiving a nevirapine-based antiretroviral therapy than in their HIV-monoinfected homologues. However, this adverse event did not impact negatively on the effectiveness of treatment.
Altogether, our analyses showed that the risk of hepatotoxicity was 2-fold higher in hepatitis B or C coinfected patients. This finding is consistent with previous reports [11, 17, 27, 28], especially with a study in patients receiving a nevirapine-based treatment in United States . Of note, HCV-coinfected patients accounted for 76% of overall coinfected patients with liver enzymes elevations >2.5 times the ULN (ALT and/or AST levels) during treatment and 100% of those with liver enzymes elevations >5 times the ULN (as compared with 60% of the study's coinfected group). It should however be noted that HCV infection was not treated while HBV infection was treated with lamivudine monotherapy. Although anti-HBV lamivudine monotherapy has been shown to lead to frequent emergence of drug resistance  and, consequently, to possible acute hepatitis, fulminant hepatic failure, and death , lamivudine was likely to be effective against HBV in the first months of treatment when most episodes of hepatotoxicity occurred. Importantly, up to half the coinfected patients experienced liver enzymes elevations >2.5 times the ULN.
Overall, the hepatic tolerability of the nevirapine-based antiretroviral therapy was as expected. Seven percents of patients experienced liver enzymes elevations >5 times the ULN. In a comprehensive analysis of 17 randomized clinical trials of nevirapine, the rate was 10% after one year of treatment . A higher rate of 17% has been found in South African patients who had a mean baseline CD4 count of 398 cells/mm3 . Our lower rate should be explained by the profound immunodeficiency of our patients at the time of treatment initiation (median CD4 count of 135 cells/mm3). Indeed, transaminases elevations >5 times the ULN in patients receiving nevirapine has been associated with sex-dependant CD4 cell count . Thus, nevirapine should be avoided in women with a CD4 count above 250 cells/mm3 and in men with a CD4 count above 400 cells/mm3. In our study, only 13% of women and no man had a CD4 cell count above these respective cuttoffs. Most transaminases elevations in our patients were asymptomatic. Only 2% of patients had a rash following treatment initiation. In their analysis, Dieterich et al found a rash in 2.2% of patients . On the other hand, the incidence rates of liver enzymes elevations >5 times the ULN in our monoinfected and coinfected patients were comparable to those reported in South African patients receiving an efavirenz-based antiretroviral therapy .
The immunological response to the nevirapine-based antiretroviral therapy was comparable between coinfected and monoinfected patients, in accordance with several studies irrespective of the treatment prescribed [7, 11, 13, 14, 18, 20]. In contrast, other studies found delayed CD4 cell count recovery in coinfected patients especially in those coinfected with HCV [8–10, 21]. However, this lower immunological response was not sustained beyond the first few months in the Thai and Swiss studies [10, 33].
The virological response also was comparable between patients coinfected with an hepatitis virus and patients infected with HIV only. Increased risk for virological failure has been reported in HBV-coinfected patients in Taiwan . However, most other studies like ours did not find any difference [7, 9, 10, 17, 21].
The clinical progression did not differ significantly between coinfected and monoinfected patients but treatment was initiated late (median CD4 count of 135 cells/mm3). Surprisingly, deaths or new AIDS-defining events were less common in coinfected patients. A possible explanation could be a close medical supervision of coinfected patients owing to frequent liver enzymes elevations. Higher rates of deaths (from any cause, liver-related causes and/or AIDS-related causes) and/or new AIDS-defining events in patients coinfected with HBV or HCV has been reported by some [8, 11–16, 18, 20] but not all studies [7, 9, 10, 17]. In our study, only one death was from a liver-related cause and this was recorded in a monoinfected patient. Intake of traditional medicines which have a potential for hepatic toxicity was however reported in two of the three coinfected patients who died.
The main limitation to our study was the relatively small sample size. However, a significant higher risk of hepatotoxicity in coinfected patients was demonstrated. Also, the absence of viral hepatitis impact on immunological and virological responses to antiretroviral therapy was strongly suggested by the regression coefficients and hazard ratio, respectively. Finally, the risk of clinical progression tended to decrease in coinfected patients which is likely to be related to an unmeasured confounding factor. The exclusion of patients presenting with serum liver enzyme levels higher than three or five times the ULN values (depending on the initial study) could have led to a selection bias (exclusion of patients with an advanced hepatitis clinical stage) but this is concordant with the recommendations for nevirapine use. Classification biases for hepatitis status could also not be ruled out because seronegative blood samples were not tested for HBV DNA or HCV RNA. Higher rates of negative HBsAg or anti-HCV results in viraemic samples have been observed in immunocompromised HIV-infected patients [3, 4]. In addition, unrepeated testing of HBV and HCV during follow-up did not allow to identify new or reactivated hepatitis infections. New infections, if any, were however likely to be rare owing to the study's duration and, for HBV, the usual childhood infections in Africa through close contacts within households and, to a lesser extent, vertical transmission (unclear modes of transmission for HCV) . On the other hand, no symptomatic reactivated hepatitis infections were recorded. In contrast, classification of patients on the basis of HBV DNA and HCV RNA is clearly a strength of our study while most other studies only used serological assays.