Emergence of severe liver toxicity in this cohort of women was infrequent. Women with either coinfection had a higher risk of mild hepatoxicity, mainly occurring in the first 6 months of treatment, and among the HBV-positive women baseline levels of HBV-DNA above 10,000 IU/ml were significantly associated with the emergence of toxicity. However, it has to be underlined that a significant association was found only for grade ≥ 1 ALT increases, while no effect was found for the occurrence of moderate to severe liver toxicity, suggesting an overall impact of modest magnitude.
More frequent ALT elevations were observed in coinfected patients in other reports [2–4, 6, 7] and HBV-DNA was found to be associated with liver toxicity in some studies  but not in others . Our finding of a correlation between detectable levels of HBV-DNA during follow-up and values of ALT supports the role of actively replicating HBV infection in the occurrence of liver toxicity. In our study both incomplete response to therapy and interruption of treatment 6 months postpartum for women not meeting the criteria for treatment, could be responsible for high HBV-DNA levels during follow-up. At this regard, recent World Health Organization guidelines for the prevention of vertical transmission of HIV , recommending life-long antiretroviral therapy containing two anti-HBV active agents for all pregnant women, represent an important step to minimize the risk of hepatic flares in coinfected patients.
High levels of CD4+ are known to be associated with nevirapine-induced liver toxicity , and in our cohort of nevirapine-treated women baseline levels of CD4+ > 250/mm3 were indeed correlated to the emergence of grade ≥ 2 ALT elevation. Liver toxicity associated with nevirapine is supposed to occur more frequently in the first 18 weeks of treatment, and in our study grade ≥ 2 early toxicity occurred almost exclusively among women with baseline CD4+ count > 250/mm3. However, the overall incidence of moderate or severe liver toxicity (9%) was relatively low, considering the long-term follow-up (and less than 1% of the women discontinued nevirapine because of liver toxicity), thus confirming our previous observation of a relative safety of using nevirapine in this context . Also, only 2.5% of the women in our study had grade ≥ 2 events during pregnancy in line with the findings of a recent meta-analysis , indicating that drug adverse events were not greater in pregnant women receiving nevirapine than in the general population. The findings of a study performed in Cameroon , in patients with CD4+ < 350/mm3, have also suggested that a nevirapine-based antiretroviral regimen could be used safely as first-line therapy despite frequent coinfections with the hepatitis viruses.
Previous studies have evaluated the impact of HBV or HCV on the response to ART in resource-limited settings. Although a limited impact of coinfections has been shown on the antiretroviral response, some studies [4, 16] have shown a negative effect on the CD4+ response in HBV-coinfected patients, especially in those with HBe or high HBV-DNA levels, and some a slower virological response . The low numbers in our study limited the power to detect significant differences in the viro-immunological parameters however, we did not observe any negative trend in terms of long-term virological suppression and immunological response.
Our study also confirmed that in Malawi HBV infection is highly prevalent in the HIV-positive population  with more than 8% of the HIV-infected women positive for HBsAg in our cohort, and that HCV is quite rare, in line with other findings . Among the HBV-infected women, HBe antigen was present in about one third of patients, in agreement with previous reports , while the proportion of those with > 10,000 IU/ml of HBV-DNA was low compared to other studies [3, 16]. A role of different genotypes can be hypothesized since genotypes C, D and F are associated to higher DNA levels compared to other genotypes  and in Malawi genotype A is highly dominant .
Limitations of this study include: the small number of coinfected women in the cohort that often did not allow to draw statistically significant conclusions; the lack of treatment adherence measures to determine the possible impact of unscheduled suspension of drugs; and incomplete data on the HBV-DNA levels during follow-up in the HBV-infected women, although the available data support the role of actively replicating infection in determining liver enzymes elevations.