A retrospective analysis of incident pregnancy in phase 1 and 2a HIV-1 vaccine study participants does not support concern for adverse pregnancy or birth outcomes

Some significant progress has been made this decade to develop a safe and effective vaccine that protects against HIV infection. UNAIDS estimates there were 1,600,000 new adult HIV infections in 2018, the majority in those aged 15–24 years, with young women in Sub-Saharan Africa bearing the brunt of the epidemic [1]. Future deployment of an effective HIV vaccine will likely target adolescents and youth in an effort to provide vaccine-elicited protection prior to the onset of sexual activity. To ensure durable protection throughout high-risk periods, booster vaccination may also be necessary.

For females of reproductive potential, sex that could lead to HIV infection may also result in pregnancy. Conversely, the perinatal period is a time of heightened risk for new HIV infection [2‐5]. Additionally, acute HIV infection during pregnancy or while breastfeeding carries a 10-fold increased risk of transmission to the child [6, 7]. Aligning with the UNAIDS ethical guidance pertaining to HIV vaccine trials [8], women are purposely included to obtain safety, immunogenicity, and efficacy data for vaccine candidates in both sexes. However, pregnancy is actively discouraged in vaccine trial participants because of unknown safety implications for the fetus. Investigators therefore exclude pregnant and breastfeeding women from participating and require women of reproductive potential to use effective contraception throughout the vaccination phase of the trial. These restrictions may deter women from enrolling in HIV vaccine trials or lead to discontinuation of vaccination prior to receipt of the full regimen or early study attrition for those who do enroll [9, 10]. Pregnancy testing is conducted regularly for females of reproductive potential who enroll in a vaccine trial, including prior to every scheduled vaccination. If a participant becomes pregnant, study vaccine administration is discontinued and she is encouraged to remain in follow-up for ongoing safety monitoring and to periodically collect systematic pregnancy and birth outcome data in person or via e-mail or telephone. Some protocols may allow resumption of the vaccination schedule after a participant is no longer pregnant. Hence, intensive contraception counselling is performed with women of reproductive potential and on-site provision of contraception services is a preferred approach in many settings [11].

Since its formation in 2000, the HIV Vaccine Trials Network (HVTN) has opened more than 80 vaccine trials with over 20,000 volunteers enrolled [12]. Study participants are typically healthy individuals 18 through 50 years of age, with increasing numbers of females of reproductive potential being involved. On-study pregnancies do occur despite pregnancy being discouraged and contraception being required, albeit rarely in early-phase trials. The requirement to report pregnancy outcomes provides some limited data regarding the incidence of pregnancy under these conditions as well as pregnancy and birth outcomes amongst study participants. As a result, a large data set is available with which to conduct a cross-sectional analysis of pregnancy outcomes in recipients of a diverse group of study vaccines compared with placebo. In this paper, we present a retrospective analysis of incident pregnancies which occurred during early phase (Phase 1 or 2a) HIV vaccine studies conducted by the HVTN.

This retrospective, cross-protocol analysis included all pregnancies reported during Phase 1 and Phase 2a HVTN vaccine clinical trials, including during long-term follow-up (LTFU) periods. Studies that did not involve active vaccination with an HIV vaccine were excluded, and only those studies for which participants and study staff were unblinded to treatment assignment prior to December 31, 2018 were included. Efficacy (Phase 2b or Phase 3) studies, some of which only enrolled participants born male, were excluded from this analysis. Of note, though efficacy studies are not included in this analysis, the HVTN 503 efficacy study data have previously been published [11].

Study participants were healthy volunteers without HIV-1 infection ages 18 through 50 years who were at low risk for acquiring HIV as per standard criteria [13]. The individual study protocols were approved by the institutional review boards and, when applicable, biosafety committees at all sites and written informed consent was obtained from each participant.

Pregnancy report and outcome data gathered from study Case Report Forms (CRFs) were tabulated from 53 phase 1 and 2a HIV vaccine trials conducted by the HVTN between 2002 and 2018. These trials were conducted in multiple countries, though the majority of participants were from the United States. Pregnancies were reported from time of enrolment through final study visit. The vaccination phase varied by study from a single injection up to 36 months.

Additionally, 13 studies also contained a long-term follow up (LTFU) period after the vaccination phase was completed for continued safety monitoring during the next 28 to 42 months during which pregnancies were also reported. To normalize across studies and different study phases, pregnancy rates per 100 woman-years (w-yr) of study participation were calculated.

Female participants of reproductive potential were identified according to the individual study criteria. This included being assigned female sex at birth, and not having reached menopause (no menses for 1 year) or having undergone a hysterectomy, bilateral oophorectomy or tubal ligation. All reported pregnancies were verified by clinical research site staff using urine or serum testing.

The date of the participant’s last menstrual period (LMP) was obtained from CRF data. In the case where only the month and year were reported, a date of the 15th of the known month was used as an estimate. Additionally, if an LMP was not reported, it was estimated using the outcome date and reported gestational age at outcome. Similarly, if an outcome was reported without an outcome date, the outcome date was estimated from the LMP and the reported gestational age. All pregnancies had either an LMP or outcome date reported.

Sites were encouraged to attempt to obtain outcome information from pregnant participants as soon as possible, even after the participants’ study participation had concluded. As most studies were blinded until after the final scheduled visit by all participants had occurred, outcomes reported during the active phase of studies were generally collected in a blinded fashion. Pregnancy outcome data, where known, were categorized as follows: therapeutic/elective abortion (as reported on the CRF), spontaneous abortion (< 20 weeks gestation), spontaneous fetal demise and/or stillbirth (≥20 weeks), premature live birth (< 37 weeks), or full-term live birth (≥37 weeks). Outcomes of spontaneous abortion, spontaneous fetal demise and/or stillbirth, and premature live birth were considered to be adverse outcomes. Adverse events of moderate or greater severity, including congenital anomalies, occurring within a window of 7 days prior to the pregnancy outcome and 56 days following the outcome were also compiled and tabulated from CRF data.

Treatment assignment information (receipt of study product or placebo) was classified according to type of product administered (viral vector, protein/peptide, or DNA), viral vector type and adjuvant. Participants were considered in receipt of product if they received at least one study product administration. Placebo recipients were grouped for analysis. Participants were also separated according to whether vaccinations were discontinued due to the identification of the incident pregnancy or whether the full course of vaccinations was completed prior to the identification.

Participants born female who were not considered of reproductive potential according to the protocol in which they participated were excluded. Pregnancies reported via standardized CRFs were compiled and tabulated. The total number of participants reporting being assigned female sex at birth, of reproductive potential and having received at least one dose of an experimental HIV vaccine or placebo served as the denominator. Descriptive statistics were used to summarize the participant characteristics, pregnancy rates, and outcome data. The associations between pregnancy outcomes and study factors, including treatment assignment, product type, adjuvant, injection status, and participant geographic region, were evaluated using Fisher’s exact test [14].

Increasing attention is being paid to the development of vaccines for adolescents prior to initiation of sexual activity to prevent incident infection and to the vaccination of pregnant women to confer protection to the newborn. Vaccine safety, particularly during pregnancy, is of considerable interest to the field [15‐17], and we anticipate this analysis will provide valuable information for future HIV vaccine trials. Eliminating vertical transmission of HIV-1 is a high priority for prevention research and since young women and adolescent girls are likely to be a target for deployment of an effective HIV vaccine once one is developed, vaccine safety with respect to pregnancy will need to be established.

Our analysis revealed that 18% of conceptions among vaccine recipients occurred during the main study phase, though the overall pregnancy rate during study conduct was low (3.09 per 100 w-yr). A similar finding was described in the groundbreaking RV 144 trail in Thailand [18] where 1922 incident pregnancies were recorded in 6334 women (30%) in equal proportions among vaccine and placebo recipients, 362 of those (18%) occurring within the first 6 months of participation [19]. Not surprisingly, younger women aged 25 years and below contributed over 70% of the pregnancies. Conception was also more frequent after the vaccination period in the 352 South African women of reproductive potential enrolled in the HVTN 503 trial [20]; the overall incident pregnancy rate was 9.6/100 w-yr but 6.8/100 w-yr during the vaccination period vs 11.3/100 w-yr thereafter [11]. In HVTN 503, women agreed to avoid pregnancy for “one month after last vaccination”. After vaccinations were stopped for safety concerns, all pregnancies occurring within 6 months after enrollment were considered to be within the “vaccination period” even in those women who had received only a single vaccination, and had delayed conception for the subsequent 6 months [11]. The studies included in our analysis that took place during the same time period as the HVTN 503 trial recorded an average incident pregnancy rate of 4.6/100 w-yr, which was higher than the 2.3/100 w-yr recorded in subsequent trials and may in part be due to increasing availability of contraception options and, at least at some sites, a more structured approach to providing on-site effective contraception during this time. Indeed, HIV prevention researchers have actively sought correlates of incident pregnancy to identify potential participants with higher pregnancy risk prior to study entry and exclude them for reasons of trial efficiency and due to the lack of pre-clinical safety data in pregnancy [10, 20‐22]. Increasing attention is being given to addressing vaccine safety during pregnancy and the ethics of conducting vaccine research among pregnant women [8, 15, 16, 23], including the rights and needs of high-risk populations who generally form the study population for efficacy trials of novel HIV prevention products [24]. Our findings contribute to the evidence base for such deliberations.

The Global Alignment of Immunization safety Assessment in pregnancy (GAIA) consortium [25] is one of the initiatives working to standardize tools for monitoring the safety of immunization programs in pregnancy, focusing on low- and middle-income countries. Another is the Brighton Collaboration [26], whose efforts to systematically collect potentially informative data internationally facilitate detection of rare events to improve global vaccine safety. To allow a more thorough analysis of pregnancy outcomes in terms of infant health along these guidelines, we propose to augment the information currently being collected in HIV-1 vaccine trials by adding the following variables to routinely-collected pregnancy outcome datasets: precise gestational age estimates and infant birth weight [27]. For vaccines nearing licensure, a pregnancy exposure registry could be established to collect pregnancy outcome data and possibly additional data on infants born to women vaccinated during pregnancy [28]; several licensed vaccines have such registries that could be used as models. Amongst other variables that would be of potential interest, more longitudinal infant data, including infant feeding method, immunological response to routine childhood vaccinations, anthropometric growth parameters, neurodevelopment and adverse events (including death) through age 6 months could be collected, potentially in post-licensure studies.

Our analyses provide a descriptive overview of the safety of prior HIV vaccines in pregnancy and provide reassuring data for estimating pregnancy safety in the large HVTN efficacy trials which are currently ongoing. Taken together, our data do not support an increased frequency of adverse pregnancy outcomes or pregnancy-related AEs in participants who have previously received experimental HIV-1 vaccines.

However, our study has limitations. Due to the diversity of the study populations across all included protocols, it is difficult to provide a comparator population for relative frequencies of pregnancies or adverse events to what is seen outside of the clinical trial context. The vaccine study products used were quite diverse and the number of pregnancies in recipients of any specific product was quite low. All participants who were capable of becoming pregnant were counselled to avoid pregnancy, contraception methods were reviewed throughout the trial periods, and negative pregnancy tests were confirmed prior to study product administration on the day of all vaccinations. Most pregnancies occurred after product administration had been completed, with many occurring over a year following last vaccination. Therefore, these data do not definitively establish that any of the products studied in HVTN trials are safe to administer during pregnancy.

However, these data do provide an estimated safety event rate which can inform sample size calculations for any future HIV-1 vaccine studies which specifically recruit pregnant women or participants who are considering pregnancy in the near future. As has been argued in the context of COVID-19 [29‐31], when pregnant women and their infants are at specific risk of infectious diseases, earlier participation in clinical trials can be done ethically with an emphasis on the informed consent process. An important component of this will be increasing involvement of pregnant women and advocates on the community advisory boards to strengthen their involvement in designing and implementing HIV vaccine and prevention studies [32, 33]. Our study also identified additional variables for which data should be collected in future vaccine studies.

We did not identify any apparent indications of increased risk of adverse pregnancy or birth outcomes in the early phase HIV-1 vaccine studies we analyzed. As there was considerable heterogeneity amongst the different vaccine trials, we suggest that more complete data on pregnancy outcomes should be collected in early phase HIV-1 vaccine clinical trials to better inform eventual efficacy trials.