Reference: N Engl J Med. 2024 Mar 14;390(11):1009-1021
Practice Point: In a failed attempt to compete with Pfizer’s FDA-approved Abrysvo, GSK’s maternal RSV vaccine proves effective but not safe in pregnancy, associated with higher rates of preterm birth.
EBM Focus: When trying to understand or explain unexpected trial results, careful attention to how well randomization was executed is worth a million commentators’ opinions.
A recent phase 3 trial of RSVPreF3-Mat, a newer maternal vaccine for RSV prevention in infants, was stopped early due to an observed increase in preterm births in infants of vaccine recipients. This outcome was obviously unexpected, not demonstrated in the Phase 3 trial of FDA-approved Abrysvo, a different bivalent RSVpreF vaccine that showed efficacy of 69% in infants within 180 days of birth. According to the [GSK-sponsored] investigators of the newer RSVPreF3-Mat vaccine, the reasons for the increase in preterm birth rates are as yet unexplained, but it really comes down to just two possibilities: premature births were either related to the vaccine, or to something else. Critical appraisal of study design, execution, and analysis can help us determine if there is a ‘something else’ at play.
Let’s talk study design first. Pregnant patients with a singleton gestation at low risk for preterm birth were randomly assigned in a 2:1 ratio (as is often done when investigators are looking to uncover any safety issues) to receive either the RSVPreF3-Mat vaccine or a placebo injection between 24-34 weeks of pregnancy. The primary outcome was RSV-associated lower respiratory tract disease in infants up to 6 months, and safety (adverse events) in infants from birth to 12 months. The study protocol allowed for administration of other vaccines during the pregnancy, just not within 2 weeks of the study injection.
Next, let’s talk study execution. In total, 5,328 pregnant patients from 24 countries across 6 continents were randomized. The planned enrollment of 10,000 was not met because the trial was halted due to the observed safety signal. From an execution standpoint, we can be pretty certain that randomization was carried out properly because 1) allocation was concealed, 2) there were no baseline between-group differences, and 3) some of the outcomes such as neonatal death occurred at rates reflecting the exact 2:1 randomization pattern. The importance of this prognostic equivalence cannot be overstated, as it increases the certainty that outcome differences can be attributed to the intervention rather than to confounders or chance. Notably, the study was carried out during COVID, which theoretically introduces the possibility of confounding due to factors related to the pandemic. However, because randomization was carried out appropriately, these effects were presumably proportionally dispersed between the groups, making it unlikely that COVID accounts for the differences in preterm birth rates between groups.
By 6 months, infants randomized to the RSVPreF3-Mat group were less likely to have RSV (vaccine efficacy 65.5% and 69% for any or severe disease respectively). However, at a planned interim assessment for safety signals, rates of preterm birth were higher in the vaccine group (237 of 3,494, 6.8%) compared to the placebo group (86 of 1,739, 4.9%), NNH=52. Even more concerning, preterm births at <32 weeks occurred in 5.5% of infants in the vaccine group compared to 2.3% in the placebo group. Overall rates of neonatal death at 12 months were 0.4% and 0.2%, respectively, however among preterm infants, death occurred in 7 in the vaccine group and 0 in the placebo group; among full-term infants, death occurred in 6 in the vaccine group and 3 in the placebo group, reflecting the 2:1 randomization. Outcomes were appropriately analyzed by intention-to-treat using a Bayesian model.
The authors and various commentators have questioned other possible explanations for the increased risk of premature birth including socioeconomic status and a striking temporal distribution of premature births. Preterm births were higher in the months between April and December 2021, and peaked in August-December 2021 in low-income countries. This suggests a time-limited cofactor, most likely a wave of the delta variant of COVID that was prevalent during that time frame. Again, there was no discrepancy between groups in rates of COVID or complications thereof (because randomization was well-executed), making it nearly impossible for COVID to be the cause of the higher rate of preterm birth in the vaccine group. In addition, 50% of those in each group were enrolled in low- or middle-income countries, and higher rates of RSV and preterm birth were observed in infants there compared to high-income countries. However, this occurred in line with the 2:1 randomization and does not appear to explain the discrepancy in preterm births between groups either. There was no temporal association between when the vaccine was given and premature birth, and while getting other vaccines was associated with a lower risk of preterm birth, (suggesting more vaccines=adherence bias), again these differences were proportional and do not explain the higher rates of premature birth in the vaccine group.
Based on careful review of the trial and the outcome data, the most likely explanation is that the RSVPreF3-Mat vaccine itself somehow caused the increased rates of preterm birth seen in this trial, even if the exact mechanism is unknown. GSK put the unfortunately-named RSVPreF3-Mat (premature is in the name!) to bed, as future studies of this vaccine in pregnant patients would be impossible, given that it’s hard enough to study any intervention in pregnant patients or infants, let alone one with demonstrated harms of the intervention.
On another note, the trial of e-cigarettes we wrote about in a February 2024 edition of the EBM Focus, “Electronic Cigarettes for Smoking Cessation: Trading Up on a Bad Habit” has been retracted. The authors report there were “significant coding errors that are difficult to rectify…and discrepancies in the calculation process that cast doubt on the accuracy and reliability of the reporting findings.” While the trial results should be completely disregarded, we stand by our advice as stated in the Focus, which was “to [still] recommend varenicline as the first line.”
For more information, see the topic Respiratory Syncytial Virus (RSV) Infection in Adults in DynaMed.
DynaMed EBM Focus Editorial Team
This EBM Focus was written by Katharine DeGeorge, MD, MS, Senior Deputy Editor at DynaMed and Associate Professor of Family Medicine at the University of Virginia. Edited by Alan Ehrlich, MD, FAAFP, Executive Editor at DynaMed and Associate Professor in Family Medicine at the University of Massachusetts Medical School; Dan Randall, MD, MPH, FACP, Deputy Editor at DynaMed; Nicole Jensen, MD, Family Physician at WholeHealth Medical; Vincent Lemaitre, PhD, Medical Editor at DynaMed; Hannah Ekeh, MA, Senior Associate Editor at DynaMed; and Jennifer Wallace, BA, Senior Associate Editor at DynaMed.