Pfizer's RSV Vaccine in Pregnant Women
Respiratory Syncytial Virus (RSV) is a ubiquitous virus that infects everyone repeatedly throughout their lifetime. The burden of this virus falls mostly on infants and elderly adults. RSV is the most common cause of bronchiolitis, a lower respiratory tract infection that itself is the most common cause of hospitalization in infants and young children.
RSV was discovered in 1956, in a time when Jonas Salk’s polio vaccine was in the process of reducing the cases of paralytic polio in the United States from tens of thousands to less than 100 each year. A whole-inactivated RSV vaccine was created and studied for use in children in 1965 and 1966. The youngest cohort (<6mo) had no history of infection with RSV. 31 infants were vaccinated, 20 were infected, 16 were hospitalized, and 2 infants died. This tragedy explains why over half a century later we still have no RSV vaccine product for infants.
Sensible Medicine is a reader-supported publication. To receive new posts and support our work, consider becoming a free or paid subscriber.
The current approach being tested is vaccination of pregnant women. Maternal vaccination increases antibodies against RSV, which can then be transferred via the placenta (before birth) and breast milk (after birth). Pfizer’s maternal RSV vaccine trial was published on April 5th in the NEJM. We will look at what they found.
Before reading a study, it is helpful to ask yourself what questions you want answered. I like to break them down with respect to efficacy, effectiveness, and safety.
Efficacy refers to how well the intervention works in ideal situations (clinical trial). This we will be able to assess directly based on the data.
Effectiveness refers to how well the intervention works in real-world conditions. A study like this won’t directly tell us, but we can make inferences based on things like inclusion/exclusion criteria, protocol, etc.
I have a few questions in mind going into this:
Did the vaccine reduce RSV hospitalizations?
Did the vaccine reduce all-cause hospitalizations?
How selective was the study?
Did they require that mothers commit to breastfeeding, how did breastfeeding impact the efficacy within the trial?
What were the rates of pre-term birth, stillbirth, neonatal death, etc.?
Note: This is by no means comprehensive, but I’d like to explain why these are some of the first questions that come to mind.
Efficacy: Infant RSV hospitalizations and death are the reason this study matters. I don’t expect it to be powered to measure a difference in infant deaths, but hospitalizations should have good data.
Effectiveness: If the inclusion/exclusion criteria are excessively strict, this makes the results more difficult to generalize. Breast feeding represents a major question mark in my mind. Rates of breast feeding vary by country but are quite low in the US, so if the efficacy of this approach is contingent on breast feeding, this would once again limit the expected real-world value.
Safety: Given the safety signal GSK* found with increased preterm births, this is of heightened interest in this data set.
*GSK has seemingly abandoned their maternal RSV vaccine efforts to date, though they had better data than Pfizer with their RSV vaccine for older adults.
Study: “Bivalent Prefusion F Vaccine in Pregnancy to Prevent RSV Illness in Infants”
Design: Double-blind, randomized (1:1), placebo-controlled trial.
Population: 7,392 healthy women, age <49, at 24-36 weeks gestation on day of planned injection. Participants had an uncomplicated, singleton pregnancy with no known increased risk of complications.
Intervention: Single intramuscular injection of 120mcg of bivalent RSVpreF vaccine or placebo.
Primary: Both measured in infants within 90, 120, 150, and 180 days after birth.
Medically attended severe RSV-associated lower respiratory tract illness
Medically attended RSV-associated lower respiratory tract illness
Secondary: All measured within 360 days after birth.
Medically attended RSV-associated lower respiratory tract illness
RSV associated hospitalization
Medically attended lower respiratory tract illness of any cause
Note: I will refer to “severe RSV” vs. “all RSV” infections for simplicity, but I am referring to these end points.
What we see above are the two primary end points. They met the endpoint for severe RSV infection at 90 days (graph A) but did not meet criteria for all RSV infections (graph B). The pre-specified lower limit of the confidence interval was >20%, which was not met at 90 days for all RSV infections. These graphs look pretty good. I will reserve my critique of this for now, let’s see the safety data.
Here we see what looks like a reassuring safety figure. This is obviously a poor way to present bar graphs, and you can often tell the difference between efficacy and safety figures by simply looking at scale. In this case they display a full 100 percent scale, which compresses any differences between the vaccine and placebo. Imagine how unimpressive the first figure (efficacy) would have looked if they took the same approach with cumulative incidence. More on the safety details after we wrap up the conclusions from the authors.
Authors’ Conclusion: “RSVpreF vaccine had a favorable safety profile and efficacy against medically attended severe RSV-associated lower respiratory tract illness and RSV-associated hospitalization in infants through 6 months of age.”
My Critique: I have several concerns and lingering questions after reading this study. We will go through them in the same framework presented at the outset.
Efficacy: How well did this vaccine work?
RSV-associated hospitalizations: RSV-associated hospitalization was a secondary endpoint. The author’s note that RSV-associated hospitalizations occurred less frequently in the vaccinated group. In their paper they only report the 90- and 180-day results for hospitalization, even though all secondary end points were measured out to 360 days. This is convenient, as the supplemental table below demonstrates:
All cause hospitalizations: There is no data provided regarding all-cause hospitalizations, which is concerning given the results of all cause lower respiratory tract infection (LRTI).
RSV-associated LRTI: This comprised both primary endpoints. At 90 days, severe infections were reduced significantly, all infections were not. I have one main complaint regarding the severe infection data at 90 days.
In the paper the author’s note: “In each group, data were missing for 3 additional infants who would have qualified as having cases of medically attended severe RSV-associated lower respiratory tract illness within 90 days after birth. Imputation showed that missing results had no effect on the conclusions.”
The omitted data was symmetric across the groups, so it doesn’t matter, right? No, and leaving the data out makes the vaccine look better than it was. To appreciate this, we must see the numbers. At the 90-day cutoff, there were 6 cases of severe infection in the vaccine group and 33 in placebo. Including these 3 additional cases for each group would increase the severe infection rate by 50% in the vaccine group, but only 10% in the placebo group.
I’m not saying this cancels out any of the benefit, but clearly this makes the vaccine look better than it was. The lower limit of the confidence interval they calculated was 40.6%, how close to 20% would it have been with those data points?
All cause LRTI: There was unambiguously zero effect on all-cause LRTI:
Effectiveness: How well might this vaccine work in the general population?
Here the glaring omission in my mind is breastfeeding. The protocol explicitly states on page 122 that breastfeeding information will be collected at each post birth visit, and at each MA-RTI visit. Breastfeeding goes completely unmentioned in both the paper and the supplementary appendix.
It is very important to know whether the efficacy of this vaccine is impacted by breastfeeding (any/none, sometime/always, duration, etc). I don’t really understand how they don’t mention it. This surprised and disappointed me.
The authors note the limitations inherent in excluding high risk pregnancies, but the inclusion/exclusion criteria seemed reasonable overall.
Safety: What safety concerns remain?
I will focus on the issue of preterm births. Here again the authors’ writing is misleading. They say in the paper: “The incidences of premature delivery were similar in the two groups (28 cases [0.8%] in the vaccine group and 23 cases [0.6%] in the placebo group).”
This is highly misleading, given they are using the data from severe maternal adverse events (S13). They don’t mention the increased risk of premature births (<37wks) which we will look at next.
There were 202 (5.7%) premature babies born to vaccinated mothers. There were 169 (4.7%) in the placebo group. This was difficult to appreciate in the misleading figure on safety, and someone casually skimming this paper could easily have missed this. The data is clear in Table 1, which I hope most people will have noticed.
Given that GSK found a safety signal of increased preterm births in their maternal RSV vaccine study, I think it is crucial to determine whether the 1% increase in absolute risk of preterm birth is real, which this study has not done.
Efficacy: There was modest vaccine efficacy against RSV-associated LRTI. That efficacy was decreased against RSV-associated hospitalization, which showed no difference at one year of life. Protection from RSV-associated hospitalization in the first 6 months of life might be sufficient cause to look favorably on these data, but the lack of all-cause LRTI hospitalization data makes the value impossible to assess. There was no effect on all-cause LRTI at any time point measured. An FDA advisory committee would presumably demand this data during consideration; its absence here is noteworthy. Overall, I’m unable to assess whether these infants conferred benefit from the maternal vaccination.
Effectiveness: I have no idea. Breastfeeding data was completely absent, which is potentially critical to vaccine efficacy. Maybe I’m wrong and it’s not, but they just omitted the data entirely.
Safety: I am not reassured that this vaccine is safe. If the 1% absolute increased risk of premature births is real, that is a major blow to the cost/benefit of this intervention. Need more randomized data. The only other company that got to this stage halted their trial for that exact safety signal. The increased risk of premature births in the GSK trial led to increased neonatal death. In my opinion, the FDA should require much more data from Pfizer before considering approval.