Abstract Body

Antiretroviral therapy (ART) is recommended for all HIV-infected pregnant women and is highly effective in preventing vertical transmission of HIV. High rates of liver enzyme elevation (LEE) during pregnancy and some cases of severe hepatotoxicity have been reported, but the causes remain unclear. Surveillance including sequential liver function tests is recommended for all HIV+ pregnant women in France. Our objective was to estimate the prevalence, causes and risk factors of LEE in the national prospective French Perinatal Cohort.

We studied the 6264 HIV+ pregnant women treated with ART, enrolled in the French Perinatal Cohort between 2005 and 2014, who had >1 normal measure of liver enzymes during pregnancy. Adjusted hazard ratio (aHR) for the association of LEE with ART were estimated using multivariate Cox models with ART as time-dependent variable, separately for women on ART at conception and for those starting ART during pregnancy.

Liver enzyme elevation (grade >1) was observed in 17.4% (n=1087), grade 3-4 in 2%. Two third occurred in the third trimester. Among women with LEE, 13% had active hepatitis B or C, 7% preeclampsia, 11% intrahepatic cholestasis of pregnancy, and less than 1% other identified causes (sickle cell disease, malaria or bile duct obstruction). More than 2/3 of LEE were unexplained: some of them were the reason for hospitalizations (n=53), cesarean sections (14), inductions of labor (3), and changes in ART regimens (36). Unexplained LEE was significantly associated with higher risk of preterm births; p<0.001.

In the women already on ART at conception, the risk of unexplained LEE was lower with NNRTI-based regimens (n=633) than with ritonavir-boosted protease inhibitors (PI/r n=1557): aHR=0.48 [0.29-0.79]. The risk was similar with boosted and non-boosted PIs (n=127): aHRs = 0.80 [95%CI 0.37-1.72] with no difference among the various PI drugs.

Among the women initiating ART during pregnancy, most regimens included PIs (89%). Compared with lopinavir/r (the most prescribed PI), LEE were less frequent for nelfinavir, tended to be higher for darunavir/r, and similar for atazanavir/r (Table1). The few women who initiated NNRTI, NRTI monotherapy or dual therapy as first treatment were excluded from this analysis.

The rate of LEE among HIV-infected women is high and impacts obstetrical care management. In most case, the cause remained undetermined. Our results suggest a possible role of PIs which needs further investigation.