INTRODUCTION

With interest in the development of maternal immunization strategies to address vaccine-preventable infections in pregnant women and their fetuses and newborns has come the need to rigorously evaluate their impact. While the approach can vary from pathogen to pathogen, evaluating vaccine efficacy in the controlled setting of a trial and the effectiveness of these vaccine candidates in a real-world setting requires expert discussion on the range of intended outcomes targeted by the vaccine candidates. Herein, we will review vaccine efficacy and vaccine effectiveness as well as review case studies, or examples, of vaccines that are currently under development or already in use. We will highlight how different study designs and assessments can be used to assess the expected benefits of these vaccines.

EFFICACY VERSUS EFFECTIVENESS

The efficacy of an intervention is defined as the performance of the intervention under controlled conditions. Efficacy studies of vaccines utilize randomized comparison groups and are referred to as randomized clinical trials (RCTs). Vaccine development goes through multiple “phases” with phase I RCTs assessing the safety of the vaccine in a relatively small group of people; the desired dose may be ascertained as a result. Phase II trials continue to assess safety as well as immunogenicity (i.e., the level of immune activation in response to the vaccine) in a relatively small and targeted population. The immune response may be one that is well-known to be protective against a particular infection or disease or may be a “correlate of protection,” an accepted proxy of a protective immune response. Phase III RCTs assess vaccine efficacy in a sufficiently large population to measure protection from disease or infection; these studies may require a large number of participants based on the frequency of the disease and the desired level of protection that the study is designed to demonstrate. If the disease is rare and it would be impractical or infeasible to monitor for the disease of interest, then a “correlate of protection” may be measured and compared among groups. If a vaccine is demonstrated to be safe, immunogenic, and efficacious from these three RCT phases, it may be licensed for use. If the vaccine is licensed for use and implemented at a population level, phase IV (post-licensure) studies may be performed to study the effectiveness of the vaccine under real-world circumstances as compared to the controlled setting under which RCTs are conducted.

Phase III efficacy studies may be restricted to healthy, non-pregnant individuals. For example, since influenza vaccines were not initially licensed for pregnant women, later RCTs measuring efficacy in this population would be considered phase IV, or post-licensure trials. In this case, efficacy studies may elect to enroll women with uncomplicated pregnancies at a pre-specified gestational age. These design choices may be based on the objectives of the study.

In phase IV effectiveness studies, individuals are less likely to be individually randomized, though cluster randomization of populations may be used. For effectiveness studies, often vaccines are made available to a community – through public health measures (i.e., immunization programs or campaigns). However, effectiveness will depend on how many individuals actually get vaccinated, the timing of vaccination, and the number of doses received. The effectiveness of a vaccine will depend partially on the efficacy of the vaccine and on the success of implementation, which will depend on whether the community is well-informed of the program, whether or not individuals choose to partake, and other factors. Effectiveness of a vaccine will additionally vary from its efficacy due to the change in population – by removing the restrictions on enrollment into efficacy trials, the vaccine may have more or less of an impact. Efficacy studies may overestimate the value of an intervention by imposing perfect conditions. In the context of vaccine evaluations, many vaccines require multiple doses or specific timing of doses to ensure high levels of protection. If a substantial proportion of the population fails to meet these requirements in implementation, effectiveness may be far lower than efficacy. Alternatively, efficacy studies may underestimate the impact of a vaccine as vaccines can provide protection for unvaccinated individuals. A vaccine can have greater effect when a greater proportion of the population is vaccinated or immune – this effect can be seen at the individual and population level and is a result of herd immunity, whereby unvaccinated individuals are protected because those that are vaccinated do not become infectious.

DEFINING OUTCOMES FOR EVALUATING EFFICACY AND EFFECTIVENESS

In the case of common infections, direct measures of disease should be used to assess the impact of a vaccine. With maternal vaccination, there may be additional metrics to assess. Maternal vaccines would prevent disease in pregnant women and, potentially, their newborn infants. In this case, maternal and infant cases of disease would be the primary outcomes of interest. Some infections during pregnancy can have negative consequences in infants regardless of clinical disease in pregnant women in which case the primary outcome of interest may be focused on the infant’s outcome. Such is the case in congenital cytomegalovirus.

Rare outcomes

As mentioned previously, effectiveness and efficacy studies evaluate the incidence of disease among study groups, however, case ascertainment may be difficult when the primary outcome of interest is rare. For example, few randomized studies of maternal tetanus-diphtheria-acellular pertussis (Tdap) vaccination have been conducted,¹ and those that have, did not have infant pertussis cases identified as part of the study design.^2,3 The evidence for recommending maternal Tdap vaccination in the United States thus relied on evaluating the immunogenicity of maternal Tdap vaccination in infants as described below.⁴ Effectiveness of Tdap vaccination for preventing infant pertussis can now be measured, as the vaccine is used widely by pregnant women, using observational studies including retrospective cohort studies and case-control studies.^5,6

Immunogenicity (correlates of protection)

Identifying correlates of protection is frequently one of the first steps in vaccine design but is additionally useful for assessing vaccine efficacy when the pathogen of interest is relatively rare. In these cases, licensure based on a proxy metric such as immune correlates of protection is desirable as it can accelerate vaccine development and lower the cost of trials.^7,8 Understanding immune correlates of infection requires research to identify specific antibody markers and levels of antibody which are protective against infection. To determine if these markers are predictive of highly effective vaccine protection, additional studies are needed to confirm both the correlation between antibody markers and the functionality of those markers as well as to compare the kinetics between vaccine-induced and naturally-induced antibodies to ensure that the longevity and behavior of vaccine-induced antibodies are comparable to naturally-induced protection.⁹ To design maternal vaccines, data are ideally required on both antibody levels in pregnant women as well as cord-blood and longitudinal early-life infant samples. With these data, researchers can estimate the rate of antibody transfer from mother to child for pathogen-specific antibodies as well as estimate the rate of maternally-acquired antibody decay.^10,11

Other proxy outcomes

Maternal infections may lead to numerous adverse outcomes, which can serve as additional metrics for assessing vaccine efficacy. Some common outcomes include rate of fetal death, stillbirths (babies born with no signs of life after 28 weeks' gestation), preterm births (babies born before 37 weeks' gestation), low birthweight births (less than 2500 g), and small for gestational age births.

Strain-specific efficacy

Evaluating vaccine efficacy is often complicated by the fact that many targets, including group B streptococcus and influenza, include groups of pathogens with large numbers of heterogenous strains. While there may be some cross-reactivity between strains of a given pathogen (i.e., a flu vaccine designed against one strain may provide some protection against another), vaccines will not work as well to protect against strains that do not “match” the vaccine strain. Efficacy can thus be calculated as either the effect of reducing total infections or the vaccine’s effect on strain-specific infections. The strain-specific efficacy may be significantly higher than total efficacy, potentially highlighting a need for more broadly immunogenic vaccine designs.

Temporal trends

It is important to account for time when considering the impact of a vaccine. Any estimate must incorporate time, either by determining a pre-set period over which the vaccine will be evaluated (i.e., efficacy over 5 years), or by using a time-sensitive analysis method such as survival analysis, which allows researchers to determine how efficacy varies over time. Immunity due to vaccination is anticipated to wane over time in most cases, but the rapidity of decline must be evaluated either from clinical trials or post-licensure studies to determine the duration of efficacy levels. The timing is particularly important in maternal vaccination when months may pass between the vaccination of the pregnant women and the birth of the infant the vaccination is intended to protect. Often, as seen when assessing the vaccine against SARS-CoV-2, vaccination later in pregnancy is associated with higher levels of maternal antibody detected in infants,^12,13 however vaccination must be done with sufficient time to prompt a strong antibody response in the pregnant woman before delivery.¹⁴

Lastly, analyses must account for timing of vaccination relative to when the protective effect is desired as well as to when antibodies can be expected to transfer to the fetus. Influenza disease and the availability of hemisphere-specific vaccine is seasonal, thus the efficacy and effectiveness of influenza vaccine can be affected by the time it is administered to the pregnant woman, relative to influenza season.¹⁵ Ideally to maximize impact, women would be vaccinated immediately before the influenza season begins, particularly among women who may give birth while the virus is circulating.

CASE STUDIES

Group B streptococcus

Invasive group B streptococcus (GBS) infections were estimated to cause over 90,000 deaths among young infants and 46,000 stillbirths globally in 2020.¹⁶ Invasive GBS occurs most often within the first few days of life, arising from ascending GBS infection from the maternal genito-urinary tract.¹⁷ If infection occurs in utero, it has been known to lead to stillbirth. While an estimated 10–40% of pregnant women are colonized with GBS,¹⁸ the proportion of GBS-infected women, which give birth to infants with invasive GBS is relatively low, with estimates of invasive GBS incidence occurring in fewer than 1 in 10,000 births in Europe and North America.¹⁹ GBS infection can additionally lead to severe disease in pregnant women and is a frequent contributor to maternal sepsis along with being associated with preterm birth.^20,21

Due to the burden of GBS in infants, the World Health Organization (WHO) declared development of maternal GBS vaccines a priority in 2015.⁷ The WHO declared goal is to develop and license a safe, effective GBS vaccine for immunization of pregnant women during the second or third trimester to prevent GBS-related stillbirth and invasive GBS disease in neonates with a target of 80% protection against combined outcomes.²² Currently, there is no vaccine available to pregnant women, and intrapartum antibiotic prophylaxis (IAP) is used in some high-income settings. While IAP is effective at preventing early onset GBS disease, it is not effective at preventing late onset disease (occurring after 7 days of age) and is not widely available, particularly in low-resource settings.

There are multiple vaccines currently in development or being tested for maternal GBS vaccination and many have completed phase I/II trials. Due to the substantial safety concerns associated with testing novel vaccines in pregnant women, some of these trials are first conducted in non-pregnant adults, such as a recent phase I/II trial of a hexavalent GBS vaccine tested in the USA.²³ Looking at safety and immunogenicity, this vaccine was well tolerated and immunogenic for up to 6 months. Trials among pregnant women have shown similar promising results. A 2016 trial among 320 healthy pregnant women in South Africa tested a trivalent GBS vaccine for GBS serotype-specific antibody responses to vaccination in pregnant women and their infants, observed maternal GBS vaginal colonization at delivery and followed infants for up to 12 months of age for safety outcomes..²⁴ While vaccinated pregnant women and their infants had significantly higher long-lasting GBS-specific antibody concentrations against all serotypes than controls, maternal GBS colonization levels increased at delivery compared to screening.²⁴ More recent phase II trials have been conducted on pregnant women in Belgium, Canada, South Africa, Malawi, and the US – vaccinating 635 women in total and showing tolerability and transplacental antibody transfer lasting up to 3 months in infants.²⁵

Despite the number of promising phase II trials, designing phase III trials to evaluate the effectiveness of a maternal GBS vaccine is complicated, as the most compelling primary endpoint – culture confirmed invasive GBS in young infants – has relatively low incidence in the population,¹⁹ thus requiring extremely large numbers of participants to detect the effect of a vaccine in a research study. Due to the low incidence of invasive GBS in young infants, some phase II studies have examined maternal GBS colonization, which, while not predictive of infant outcomes, is a prerequisite for invasive GBS in young infants. A composite endpoint – comprised of all serious and fatal events including cases of invasive GBS diseases in neonates, infants, and stillbirths – could be another suitable outcome for clinical trials.⁸

Like the process used for maternal Tdap vaccination approval, there is potential for a maternal GBS vaccine to be licensed based on immune correlates of protection alone, without pre-licensure efficacy trials.⁹ In the case of maternal GBS vaccination, if immunological correlates of protection alone were to be used for vaccine licensure, multiple additional conditions would have to be met. Limited data currently exist on specific antibody markers and levels of antibody, in either pregnant women or infants, which are predictive of protection against GBS.^26,27,28 Additional studies are currently needed to confirm both the correlation between these identified antibody markers and the functionality of those markers as well as to compare the kinetics between vaccine-induced and naturally-induced antibodies to ensure that the longevity and behavior of vaccine-induced antibodies are comparable to naturally-induced protection.⁹ Additionally, if licensure is based on immune correlates of protection alone, post approval effectiveness studies would be critical. These studies would need to demonstrate maternal GBS vaccine impact on invasive GBS disease burden in young infants. Based on the study design, these evaluations could require baseline data on the burden of GBS disease in the target populations, including data from low-and-middle income settings.

Strain-specific efficacy

An additional challenge in designing a GBS vaccine is identifying sufficiently broadly protective neutralizing antibodies to protect against the majority of GBS strains causing invasive GBS disease. As mentioned previously, current vaccines in development are designed to protect against anywhere from three to five strains of GBS.^23,25 Whether or not this number is sufficient is currently unknown. Studies examining immune correlates of protection show a broad range in protective efficacy of serotype-specific antibodies against invasive GBS disease.²⁶

Influenza

Influenza has a disproportionate impact on pregnant women and neonates. Influenza infection during pregnancy is associated with increased risk of fetal loss and maternal hospitalization,²⁹ and influenza in the first 6 months of life is associated with a high risk of hospitalization and death.³⁰ As a result of this increased risk, the WHO recommended that pregnant women be prioritized to receive influenza vaccine in 2012.³¹

Efficacy of maternal influenza vaccination is frequently assessed for multiple outcomes in both the pregnant women and their infants, including influenza-like respiratory infections and laboratory confirmed influenza infections. Lab-confirmed influenza infections including both a clinical presentation of flu-like symptoms and molecular detection of the influenza virus is the gold standard for assessing efficacy, however some studies rely on detection of acute respiratory infections alone (without molecular confirmation) due to resource limitations. Due to the range of complications associated with influenza infection during pregnancy, there is additionally substantial interest in the efficacy of maternal influenza vaccination to decrease the risk of fetal death, stillbirths, and adverse birth outcomes, such as preterm birth, low birthweight, and small for gestational age births.

Relatively few clinical trials exist on the efficacy of maternal influenza vaccination, and these have shown varied results across this range of outcomes, as expected given the level of variation in populations, circulating virus strains, and specific vaccines tested. For the gold standard outcome of laboratory confirmed influenza infections in either mothers or infants, some studies have found high vaccine efficacy,³² such as a 2005 study in Bangladesh, which identified a vaccine efficacy of 63% for preventing lab-confirmed influenza infections in infants.³³ Alternatively, other studies have found relatively low, or non-significant vaccine efficacy such as a 2013 study from Nepal, which found minimal, non-significant efficacy against both maternal and infant lab-confirmed influenza.^34,35

Unsurprisingly, the same level of variation exists when examining the efficacy of maternal influenza vaccination for preventing adverse infant birth outcomes, including the frequency of infants born small for gestational age, infants born with low birthweight, preterm birth, and fetal death. Few clinical trials have found high vaccine efficacy³⁶ and many have found either low, or null efficacy at preventing infant birth outcomes.^34,35,37 The reasons for this variation may be related to variations in population demographics, circulating virus types, or season-specific vaccines. However, it is also possible that many clinical trials are underpowered to evaluate differences in these types of outcomes, particularly under randomized clinical trial conditions. As mentioned at the beginning of the chapter, clinical trials are often restricted to relatively healthy populations, potentially limiting the possible number of adverse outcomes in study participants. Additionally, clinical trial participants are closely monitored and treated for any health-related issues, potentially receiving higher than average medical care and again limiting the frequency of adverse outcomes occurring during the study. Observational studies of population-level vaccine effectiveness can be helpful by both increasing the numbers of included participants and expanding the participant pool to include a broad range of individuals with varying levels of risk for complications.

Studies of population-level effectiveness of maternal influenza vaccination often rely on large cohort studies or nations with population-wide medical registries, such as Norway or Denmark.^38,39 These types of studies can be valuable as they are substantially cheaper than randomized clinical trials, and include large populations (i.e., millions of paired women and children). At this scale, researchers are better able to detect the varying levels of effect of vaccination on relatively rare outcomes, such as stillbirth; for example, a population-based registry study found maternal influenza vaccination halved the risk of stillbirth.³⁸ Similarly, an observational study using a database including ~12,000 pregnant women in France was able to identify an astonishing reduction in SGA among infants born to vaccinated women (0R = 0.36, 95% CI = 0.17,0.78).⁴⁰ Observational studies have additionally been able to detect large effects of maternal vaccination for preventing preterm births⁴¹ and lab-confirmed influenza infections in infants⁴² and mothers.^39,41,43 Despite the promising findings from population-level studies, it is important to acknowledge that effectiveness estimates from observational research such as the cohort and registry-based studies described above may suffer from confounding, or bias, due to unmeasurable variables: because participants are not randomized to receiving maternal influenza vaccine, the effects found by researchers may be a product of generally healthier or more health-conscious mothers choosing to get vaccinated, which can be difficult to account for in analysis.

The window of observation included in analyses may be another major factor confounding estimates of the efficacy of maternal influenza vaccination. Ideally, a maternal influenza vaccine would protect infants out to 6 months of life, the period where infants are most at risk of influenza complications. However, at least two studies have shown that maternal influenza vaccination has high efficacy in the first 2 months of life, estimates of which are attenuated significantly by including longer follow-up time.^35,44 Stratifying by time when examining the efficacy of maternal influenza vaccination not only allows researchers to identify periods with high vaccine efficacy, but additionally indicates that duration of efficacy is low and wanes quickly.

Strain-specific efficacy

Influenza viruses are a diverse class of viruses with infinite potential for variation, but each year, influenza vaccine is designed to protect against only three to four viral strains. These strains are updated annually by compiling antigenic characterization and genetic data to predict the best options in the months leading up to the influenza season.⁴⁵ Influenza vaccines have the highest vaccine efficacy preventing infections caused by strains which have antigenic “matches” to one of the strains included in the vaccine. However, the prediction process is not perfect, and circulating viruses may not match the vaccine. Influenza vaccination still provides some, although lesser, protection against mismatched strains,⁴⁶ however estimates of vaccine efficacy that include all influenza infections as outcomes will inherently be lower than strain-specific analyses restricting to strains which match the vaccine.

Hospitalization as an outcome

As mentioned previously, infants in the first 6 months of life have increased risk of influenza-related hospitalization. This risk is higher than at any other time of life, thus decreasing the rate of hospitalizations is a primary goal of maternal influenza vaccination. Due to the relatively low rate at which influenza-related hospitalizations occur, estimating the efficacy of maternal vaccination at reducing influenza-associated hospitalizations is challenging. One clinical trial examining all-cause infant hospitalizations for respiratory infections as an outcome found relatively low maternal influenza vaccination efficacy of 43% (or 58% when restricting to the first 90 days of life).⁴⁷ This low value is unsurprising as all cause respiratory hospitalizations is a non-specific outcome, and many of the respiratory infections included as outcomes may have been caused by other pathogens, unaffected by influenza vaccination. However multiple observational studies have attempted to directly estimate the effectiveness of maternal vaccination for decreasing influenza hospitalizations. A meta-analysis including four observational studies found maternal influenza vaccination was associated with large decreases in influenza-associated hospitalizations, VE = 72% (95% CI: 39%, 87%).⁴⁸ Studies included in the meta-analysis were all conducted in the US and England, and vaccine efficacy in these studies ranged from 45% to 92%^49,50,51 with included studies subject to varying levels of bias. More recently, a multi-country study estimated 40% maternal vaccine efficacy at preventing infant influenza hospitalizations in the influenza season after a pregnancy.⁵² Again, the wide span of estimates is not surprising, as influenza vaccine efficacy is expected to vary by season and population, particularly depending on the proportion of circulating viruses that match the vaccine strains for that season.⁵³

SUMMARY/CONCLUSIONS

Maternal vaccination is a potent tool in preserving the health of pregnant women. The use of maternal vaccination in the prevention of infections and related complications in the fetus and/or newborn can be lifesaving. Assessing vaccine efficacy and vaccine effectiveness includes observing the full breadth of the disease burden. This evaluation can include the spectrum of outcomes like the incidence of disease, antibodies measured in the woman and infant, and neonatal outcomes such as prematurity. When designed appropriately, either in a well-defined study population or in large meta-analyses or registry population, the results can lead to the licensure of these products for specific use in pregnant women.

PRACTICE RECOMMENDATIONS