Protected to death: systematic exclusion of pregnant women from Ebola virus disease trials

Table 1 summarizes published data from EBOV outbreaks on maternal and pregnancy outcomes, Section A for the previous epidemics, Section B for the 2013-6 epidemic. In the 1976 outbreak, the case fatality rate (CFR) in EBOV-infected pregnant women was 89% (73/82) [17]. Almost half of all EBOV-infected women were pregnant (46%: 82/177). The high risk in pregnancy was later attributed to the repeated use of needles for vitamin injections in routine antenatal care without sterilization between patients [17, 18]. In the 1995 outbreak, 15/105 (14%) EBOV-infected women were pregnant [17]. The CFR for EBOV-infected pregnant women was 93% (14/15) compared with 70% (28/40) for EBOV-infected non-pregnant women and an overall 77.5% CFR (245/316) [17]. The differences in CFR are not statistically significant.

In aggregate, any EBOV-infected pregnant woman had survived only after spontaneous miscarriage, elective abortion, stillbirth, or with a neonatal death (Table 1 SectionA). All EBOV-infected pregnant women developed vaginal and uterine bleeding and were at high risk for spontaneous abortion and pregnancy-related hemorrhage [19]. In the 1976 outbreak, the rate of spontaneous abortions was 23% (19/82). The remaining pregnancy outcomes were stillbirths or neonatal deaths—no neonate survived longer than 19 days [18]. In 1995, the spontaneous abortion rate was 67% (10/15), with three elective abortions, one premature stillborn, and one live-born, full-term neonate who died at three days [17]; one of the three elective abortions followed an incomplete spontaneous abortion and the woman survived [17]. Four EBOV-infected mother-baby pairs were traced after the 2000–2001 outbreak in Uganda: all mothers and babies had died [20, 21].

At the time of this epidemic, there were no approved specific treatments or vaccines for Ebola virus disease (EVD). Clinical management consisted of supportive care, particularly fluid and electrolyte management, correction of coagulopathy, treatment of secondary infections, and management of other complications [19]. Treatments proposed had not undergone clinical trials in EBOV populations or at all [22‐24]. Vaccines were in very early development with few having entered Phase I safety and immunogenicity trials [24‐29].

Table 2 lists the trials conducted during the 2013–2016 epidemic in Liberia, Guinea, and Sierra Leone and their pregnancy-related eligibility criteria. All drug and vaccine trials excluded pregnant women. Two of three convalescent plasma studies, funded by the European Union, included pregnant women [30]. Pregnant women were granted access to new treatments only within ‘Monitored Emergency Use of Unregistered Interventions’ (MEURI) [31] protocols implemented by Médecins Sans Frontières (MSF) for MIL77 (three chimeric monoclonal antibodies targeting different epitopes on the surface of EBOV glycoprotein) and favipiravir [16, 31].

The WHO-ERC reviewed all protocols for studies supported or sponsored by WHO, four protocols at the request of the Médecins Sans Frontières Ethics Review Board and one for the Rapid Assessment of Potential Interventions & Drugs for Ebola (RAPIDE) Consortium [16, 31, 32]. The WHO-ERC applied the Council for International Organizations of Medical Sciences guidelines [33] and followed the recommendations of a WHO panel of external experts convened to provide ethical guidance on use of unregistered interventions for treatment or prevention of EBOV in a context in which patients were managed with no, or limited, clinical trial data [34]. The WHO panel counselled use of unregistered interventions in the epidemic, conditional upon evidence from laboratory and animal studies. The panel also emphasized that in prioritizing and allocating interventions “children and pregnant women should be considered particularly vulnerable [because of their higher mortality rates]... and given special protection when receiving such interventions” [34]. In the face of the long history of exclusion of pregnant women from clinical trials [9, 15, 35‐37], this recommendation was remarkable and important; the WHO-ERC understood that these groups were to be provided preferential access to interventions. For the WHO-ERC, the virtual certainty of fetal/neonatal loss invalidated exclusion of pregnant women because of risk to the fetus; the high maternal mortality in past Ebola outbreaks favored their inclusion for clinical and ethical reasons. Other ethical considerations dictated that pregnant women should be accorded the same autonomy as non-pregnant adults: pregnant women had a greater interest in and right to decide about their own and their fetus’ health than sponsors, researchers, regulators or ethics committees. The WHO-ERC considered these points equally applicable to vaccine trials in EVD-affected countries that would enroll uninfected participants based on data from Phase I safety and immunogenicity trials and noted that other ethics committees took the same view [31].

By the end of the EBOV epidemic, the WHO-ERC had reviewed 14 protocols for interventional trials as well as two MEURI protocols [16]. These included studies of brincidofovir [32] and favipiravir [38], a study with convalescent plasma [39] and several phases of the rVSVΔG/ZEBOV-GP vaccine [40‐43] and the ChAd3-EBO-Z vaccine [44, 45]. All vaccine protocols, including those in affected countries, excluded pregnant women. The brincidofovir trial excluded pregnant women on the basis of embryotoxicity in animal studies without comment on the relevance of these data for a disease resulting in 100% human fetal loss; the favipiravir trial could not include pregnant women because the sponsors were unable to get insurance coverage despite strong recommendations for inclusion from the WHO-ERC, MSF Ethics Review Board, and Inserm Institutional Review Board [31]. The WHO-ERC requested investigators of all treatment and vaccine trials in the affected countries to reconsider exclusion of pregnant women based on benefit-risk assessment, but the requested amendments were not submitted. Therefore, the WHO-ERC faced the difficult dilemma of granting approval for immediate trial start (with potential benefit for the many participants the protocols included) or withholding approval until pregnant women were either included or their exclusion justified. The latter choice would delay trial start in the context of an epidemic for which mortality was high and speed of intervention was essential. Since agreement to include pregnant women would require consensus between numerous parties (including sponsors), which would take time and delay trial start substantially (or possibly indefinitely), WHO-ERC did not make inclusion of pregnant women a prerequisite for WHO-ERC clearance [16]. When interim analysis of the efficacy and safety data in non-pregnant adults in the rVSVΔG/ ZEBOV-GP vaccine trial showed benefit, the WHO-ERC (and the Data Safety Monitoring Board) pointed out the high incidence and mortality rates in children and pregnancy and unsuccessfully sought inclusion of the latter, or justification of exclusion. Forty-two pregnant women were denied participation [43]. However, since the trial excluded pregnant women on the basis of self-reported pregnancy status (pregnancy tests were offered, but not obligatory), more than 20 other pregnant women received the vaccine [Henao-Restrepo and MSF, personal communication] [43].

The 2013–2016 EVD epidemic permitted better estimates of CFRs and factors impacting survival rates and determination of persistence of EBOV in different body fluids.

The CFR across both sexes was 62.9% (95% CI: 61.9–64.0%) declining from 69.8% (95% CI: 58.6–79.2) to around 39% (95% CI: 25.7–54.3%) from July 2015 to September 2015. Survival was highest in those under 5 years (75.6%) and above 75 years (83.8%), a pattern similar in all three countries [2]. Both sexes were equally susceptible to infection [46]. In all countries, time from initial symptoms to hospitalization was approximately 0.5 days shorter for women [46]. CFRs were significantly lower for women: 63% (95% CI: 61.6–64.4, n = 4756) versus 67.1% for men (95% CI, 65.8–68.5, n = 4637), p < 0.001; the survival difference was significant after adjustment for age, clinical symptoms, and intervals between onset and hospitalization [46].

Despite the size of the epidemic, and the opportunity, information on pregnancy and pregnancy outcomes was not systematically obtained. Available data are shown in Table 1B; some analyses are ongoing. The maternal CFR estimated from these data is 55% (44/80) [46‐58] excluding approximately 20 pregnant vaccinated women [43]; maternal CFR is not statistically significantly different from the CFR of women overall. All surviving mothers experienced miscarriages or stillbirths [49‐51, 57, 58] and two women died with the fetus in utero [47, 48]. The only surviving baby was born to a woman who had received favipiravir under a MEURI protocol and died. Authorization was given to MSF to treat the newborn, but not the mother, with ZMapp [56] [59]. The reasons for high fetal mortality may be related to EBOV placental preference and consequent high viral load in utero, as samples from amniotic fluid, placenta, and fetuses tested positive for EBOV [52]. Live-born babies appear to have been preterm births and preterm babies normally have a high mortality risk. In EBOV-affected countries where babies are often exclusively breastfed immediately after birth (and there may not be a safe alternative to breastfeeding available), the absence of a surviving mother or the inability of an EBOV-infected survivor to breastfeed places a surviving baby at risk of death.

As of 2 February 2016, between 10,000 and 17,000 EBOV survivors were reported compared with 1000 survivors from all previous epidemics combined [2, 60, 61]. Compared to blood used for determining cure, clearance of EBOV is delayed (sometimes for months) in immunologically protected fluids/body compartments including semen [62, 63], ocular tissues [64], breastmilk [65], vaginal secretions [66] and the central nervous system [60, 67]. Mother-to-child transmission of EBOV can occur through body fluids in utero, during delivery, contact after birth, and breastmilk, even when the woman is asymptomatic [48]. Among 70 EBOV survivors who conceived post-recovery, 15/68 miscarried and two survivors elected to terminate their pregnancies; four neonates were stillborn (3/4 conceived within two months of discharge of the Ebola Treatment Unit). While still sparse, the data suggest that pregnancies shortly after recovery also increase the risk for poor outcomes [68].