Receipt of hepatitis E vaccine and fetal loss in rural Bangladesh: further analysis of a double-blind, cluster-randomised, controlled trial

Abstract

Background: Vaccination constitutes an attractive control measure for hepatitis E virus (HEV), a major cause of maternal and perinatal mortality globally. Analysis of pregnant participants in an effectiveness trial of the HEV vaccine HEV239 showed possible HEV239-associated fetal losses. We aimed to conduct a detailed analysis of this safety signal.

Methods: In a double-blind, cluster-randomised trial, 67 villages in Matlab, Bangladesh, were randomly allocated (1:1) to two vaccine groups, in which non-pregnant women aged 16-39 years received either HEV239 (HEV239 group) or Hepa-B (a hepatitis B vaccine; control group). We implemented weekly surveillance for pregnancy detection, and follow-up of pregnancies once every 2 weeks, using physician-confirmed diagnoses to evaluate fetal loss outcomes (miscarriage [spontaneous abortion], stillbirth, and elective termination). Data from a parallel system of reproductive health surveillance in Matlab were used to clarify study diagnoses when necessary. Miscarriage was assessed only among participants whose first positive pregnancy test and vaccination date (for whichever dose was closest to the date of last menstrual period [LMP]) were before 20 weeks' gestation. We defined the following analysis periods of interest: from 90 days before the LMP until the pregnancy outcome (the proximal period); from the LMP date until the pregnancy outcome (the pregnancy period); from 90 days before the LMP until the LMP date (90 days pre-LMP period); and from enrolment until 90 days before the LMP (the distal period). Both Poisson and Cox regression models were used to assess the associations between receipt of HEV239 and fetal loss outcomes. The trial was registered with ClinicalTrials.gov (NCT02759991).

Findings: Among the 19 460 non-pregnant participants enrolled in the trial, 5011 were identified as having pregnancies within 2 years following vaccination and met the criteria for analysis (2407 in the HEV239 group and 2604 in the control group). Among participants vaccinated in the proximal period and evaluated for miscarriage, miscarriage occurred in 54 (8·9%) of 607 in the HEV239 group and 32 (4·5%) of 719 in the control group (adjusted relative risk [aRR] 2·0 [95% CI 1·3-3·1], p=0·0009). Similarly, the risk of miscarriages was increased in the HEV239 group versus the control group among participants inadvertently vaccinated during pregnancy (22 [10·5%] miscarriages among 209 participants in the HEV239 group vs 14 [5·3%] of 266 in the control group; aRR 2·1 [95% CI 1·1-4·1], p=0·036) and among those vaccinated within 90 days pre-LMP (32 [8·0%] of 398 vs 18 [4·0%] of 453; 1·9 [1·1-3·2], p=0·013). No increased risk of miscarriage was observed in those who received HEV239 in the distal period (93 [5·6%] of 1647 vs 80 [4·5%] of 1773; 1·3 [0·8-1·9], p=0·295). Stillbirth and elective termination showed no increased risk among women administered HEV239 versus those administered Hepa-B in any of the analysis periods.

Interpretation: HEV239 given shortly before or during pregnancy was associated with an elevated risk of miscarriage. This association poses a possible safety concern for programmatic use of HEV239 in women of childbearing age.

Funding: Research Council of Norway and Innovax.