The effectiveness of vaccination against measles, the leading cause of vaccine-preventable deaths in infants globally, is greatly impacted by the level of maternal antibody to measles virus (or “measles maternal antibody”; MMA) during infancy. Variation in the prevalence of maternal antibody to measles virus between infant populations across countries and socio-demographic strata is poorly understood. We reviewed the literature on the prevalence of MMA, focusing on 3 principal determinants: starting level of maternal antibody, placental transfer of maternal antibody, and rate of decay of maternal antibody after birth. Our review identified placental transfer as an important determinant, with greater efficiency found in studies performed in developed countries. Placental transfer was influenced by gestational age, human immunodeficiency virus infection, and malaria. Antibody levels in mothers varied widely between countries, although predictably according to vaccination status within populations. Rates of antibody decay across studies were similar. Future studies should evaluate the utility of the cord blood level of MMA as a predictor of vaccine efficacy in infancy; inclusion of World Health Organization international reference sera will facilitate comparisons. Greater understanding of the determinants of the prevalence of MMA will help national policy makers determine the appropriate age for measles vaccination.
Measles is the principle cause of vaccine-preventable deaths in infants in the world. An estimated 1 million children die annually of measles, and millions more have its sequelae, which include diarrhea, pneumonia, encephalitis, and malnutrition . Before the measles vaccine era, there were ∼500,000 cases of measles and 500 deaths due to measles reported annually in the United States . Since the introduction of measles virus vaccines in 1963, the endemicity of measles has been dramatically reduced in many countries and eliminated in some [3, 4].
The optimal age for vaccination against measles continues to be debated. Except for a brief period between 1990 and 1993 (during which the recommendation to use high-titer Edmonston-Zagreb vaccine at age 6 months was introduced and withdrawn), the World Health Organization (WHO) has recommended a single dose of measles vaccine at 9 months [5, 6]. Several developed and developing countries follow a strategy that differs in timing and in the number of doses delivered either through routine immunization or supplemental mass immunization campaigns . The current recommendation in the United States is a 2-dose measles vaccination schedule: dose 1 given at age 12–15 months and dose 2 given at age 4–6 years .
In determining the age for vaccination, countries must balance the consequences of an older age (lack of protection in the early months of life) and a younger age (reduced effectiveness). Many countries, where morbidity and mortality due to measles are uncommon in infants, choose an older age for vaccination (e.g., age 12 or 15 months). In other countries, where a high number of deaths due to measles occur in children aged <9 months, a younger age for vaccination has been advocated [8, 9].
The ability of maternal antibody to measles virus (or “measles maternal antibody”; MMA) to interfere with seroconversion after vaccination is well known [10–13]. However, reasons for variation in the prevalence of MMA throughout infancy in different country populations are not well understood. Previous reports have noted a trend toward higher seroconversion rates following measles vaccination during infancy in developing countries, as compared with developing countries [12, 14]. Many researchers interpret this trend as circumstantial evidence that there is earlier loss of MMA in developing countries, but hard evidence is not available. A recent literature review of studies on the efficacy of measles vaccine compiled prevaccination point prevalences of MMA from various country studies, noting the difficulty in drawing comparisons and conclusions from them .
Learning about the factors that influence the loss of MMA early in life is important because of the significant impact that it has on the effectiveness of measles vaccination. In many countries, there are no data on the seroprevalence of MMA in infancy—seroprevalence studies are difficult and expensive. In the absence of these data, deeper understanding of the determinants of the seroprevalence of MMA might better inform policy makers facing the question of when to vaccinate. In this review, the importance of MMA and the tests used to measure it are described, and the literature on the 3 principal determinants of the seroprevalence of MMA (levels of antibody to measles virus in mothers, the degree of placental transfer of MMA, and the rate of decay of MMA after birth) are reviewed. A detailed discussion of other factors that relate to the effect of measles vaccination on seroconversion is beyond the scope of this review.
A MEDLINE search of the literature from 1967 to 1999 was conducted by using the National Library of Medicine's PubMed on-line search utility. Key words included measles, maternal, antibody, and decay. A “related articles” hyperlink for each of the articles that were initially retrieved significantly expanded the search. The reference lists from the retrieved articles were also used to identify other relevant literature. Except in a few instances, the search was limited to the scientific English-language literature.
Significance and measurement of MMA. MMA is type IgG, primarily IgG1, and is actively transported through the placenta from mother to fetus . MMA protects infants from measles during the early critical months of infancy. As with maternal antibodies to other antigens, the level of MMA declines, and it is typically absent by the end of the first year of life .
Immunity to measles after natural infection involves both humoral and cellular responses . The B cells produce IgG, IgA, and IgM antibodies. Typically, IgM wanes to undetectable levels within several weeks, IgA becomes undetectable within a few years, and IgG persists indefinitely. Although cellular immunity is not as well understood and more difficult to measure, its importance is highlighted by the fact that children with agammaglobulinemia can have uncomplicated measles virus infection and develop immunity to measles [18, 19].
Several studies have underscored the significance of MMA in decreasing seroconversion and antibody titer after measles vaccination . In a study that separated 34 infants into 4 groups on the basis of the value of their MMA titers, Albrecht et al.  showed an inverse relationship between prevaccination and postvaccination titers of antibody to measles virus as measured by the neutralization test. None of the infants in the highest prevaccination titer group seroconverted. Stewien et al.  used the hemagglutination inhibition (HI) test for 43 infants and had similar findings. In a large vaccine trial of 1061 infants aged 6 months and 299 infants aged 9 months that was conducted by Markowitz et al. , infants with preimmunization antibody titers <1:40 consistently seroconverted with higher titers than did those with preimmunization titers >1:40. This finding held true regardless of age, vaccine type, or dose.
Measurement of MMA. The tests that have been developed to detect antibodies to measles virus vary in their sensitivity, cost, and ease of administration. Traditionally, the presence of any detectable antibody has been thought to correlate with immunity; however, studies have shown that low levels of antibody detected by highly sensitive tests may not be protective . In addition, variability in testing (in assays and cutoff values) and lack of calibration of these tests against WHO international reference sera have made comparison of profiles of the seroprevalence of MMA across studies difficult . Although the lowest level of MMA that interferes with seroconversion after vaccination is not well established, this interference, as mentioned previously, occurs along a gradient.
The HI test measures antibody to the H protein of measles antigen and has been widely used until recently. The sensitivity of this test varies according to the method of treatment of the measles antigen and the type of RBCs used. The neutralization test is more sensitive than the HI test and measures the extent to which antibodies neutralize the cytopathic effect of measles virus in cell culture. The plaque reduction neutralization (PRN) test has maximum sensitivity and specificity and is considered the gold standard. The sensitivity of the PRN test greatly exceeds that of the standard HI test and varies by the type of measles antigen used in the assay. Chen et al.  demonstrated that antibody titers <120 milli-international reference units (mIRU), as measured by PRN testing, are detectable but might not be protective. The PRN test is seldom used in large surveys because it is expensive and time-consuming to perform.
The measles virus EIA is now the most widely used and ideal test for large field surveys assessing seroprevalence . The reagents are easily stored, and results are rapid, sensitive, specific, and reproducible.
Other variables to be considered in testing are the method for collecting serum (i.e., venipuncture vs. finger or heel stick) and the method used to calculate geometric mean titers (GMTs). With the HI test, greater sensitivity is achieved by testing venous blood. Novello et al.  showed a high correlation between EIA results attained by both methods of serum collection, although there were discrepancies at lower antibody levels.
Levels of antibody to measles virus in pregnant women. We found only 1 study in the literature that looked at maternal blood samples from several countries and tested them by using methods to maximize comparability . Serum samples were obtained from pregnant women in 15 different populations in 9 countries in North America, South America, Africa, and Asia (table 1). These specimens were obtained at various stages of pregnancy and at full term. A correction factor was incorporated into the results for serum specimens obtained before full term to adjust for dilutional effects of expanding blood volume during pregnancy. Mean log GMTs varied from 4.4 to 7.6 (although only differences of at least ≥0.75 were statistically significant). The United States was the only developed country represented in this study. Of note is the finding that mean log titers of MMA in women born after 1958 in Connecticut were in the middle of the range of titers observed in developing countries. In another study in the United Kingdom, researchers found that the log GMT was 5.4 among unvaccinated mothers and 3.6 among those who were vaccinated . A study by Gendrel et al.  demonstrated a log GMT of 5.9 among mothers in France, compared with 6.5 among mothers living in Gabon. Because of the lack of detailed information about the populations studied (e.g., vaccination and/or health status of mothers in the study by Black et al. ), few inferences could be drawn regarding the factors influencing maternal antibody levels across these populations.
Several factors have been studied to determine reasons for variation within populations. Well-documented determinants of the level of antibody to measles virus in pregnant women are prior measles exposure and vaccination history. It has been observed in several developed countries that women with natural exposure to measles have significantly higher levels of antibody to measles virus than do those who are vaccinated. Bromberg et al.  demonstrated that non-US-born mothers had significantly higher levels of antibody to measles virus than did US-born mothers likely to have had received measles vaccine. A study done in the United Kingdom by Brugha et al.  similarly showed a higher GMT of antibody to measles virus in mothers with a history of natural exposure. This pattern presumably occurs in developing countries and may become easier to document as increasing numbers of vaccinated females reach childbearing age without natural exposure to measles.
Another variable of interest is the boosting effect of repeated exposures to measles, as this effect could imply that there are sustained high levels of antibody to measles virus in pregnant women living in areas of high endemicity. The degree and long-term duration of the boosting effect from repeated exposures were first studied by Krugman et al. . These researchers compared 2 vaccinated cohorts of children with different levels of exposure to measles and found that 14 years after vaccination, the more highly exposed group had a higher GMT and seropositivity rate.
In an outbreak setting with pre-exposure and postexposure serum samples available, Chen et al.  found that 7 of 11 persons with preexposure antibody titers of 1:216 to 1:874 determined by PRN testing had a ≥4-fold increase in their antibody titer, whereas this finding was absent for 7 persons with preexposure antibody titers ≥1:1052. This result suggests a boosting effect that is limited by the magnitude of preexisting antibody titer.
Lennon and Black  showed that GMTs in women of childbearing age who were born from 1955 through 1967 decreased by advancing year of birth. This period coincided with a precipitous decline of measles circulation and, therefore, the boosting effect in the United States. These findings were confirmed in 2 recent studies by Maldonado et al.  and Kacica et al. .
Other variables that have been studied include socioeconomic status, education, race, parity, age, and nutrition. In a large study of pregnant women in Portugal, Guilherme de Almeida Goncalves  found that mothers with a higher socioeconomic status had a significantly higher GMT of antibodies to measles virus. In an analysis by Markowitz et al.  of US-born women, black mothers were found to have significantly higher GMTs than Hispanic or white mothers. Two studies, one in Nigeria  and the other in Bangladesh , found a relationship between increasing age and parity of mothers and lower GMTs, although the effect of confounding between the two variables could not be assessed. A study by Sinha  linked increasing age and/or parity of the mother to a heightened risk of measles before age 6 months. Kaur et al.  in India and Halsey et al.  in Haiti found no difference in GMTs between well-nourished and malnourished mothers.
Placental transfer of MMA. Antibody to measles virus is transferred from mother to fetus by an active transport mechanism during the third trimester . Comparisons of concentration ratios found in studies performed in different countries are less subject to variations in test sensitivity since both titers (that in the mother and that in the cord blood) that comprise the ratio would presumably be equally affected. As before, the study by Black et al.  provides the most comprehensive and comparable data; these data are listed below along with those from other studies in table 2.
Almost all studies done in various countries have demonstrated that GMTs of cord blood antibody were higher than GMTs of maternal antibody. Exceptions were found in Black's data from Taiwan  and the study by Gendrel et al.  in Gabon, where the concentration ratios were 0.48 and 0.92, respectively. We found a general trend toward more efficient placental transfer of MMA in developed countries, compared with developing countries, which held true across the different tests used to measure antibody to measles virus.
Many variables were analyzed in studies done within countries, including vaccination status, socioeconomic status, maternal age, gestational age, delivery by cesarean section, and baby's weight. Gestational age was found to significantly affect ratio of GMT of cord blood antibody to GMT of maternal antibody (i.e., longer gestation resulted in a higher ratio) [30, 34, 37, 43, 44]. Studies that looked at measles vaccination status found that the rate of placental transfer of antibody did not differ between vaccinated and unvaccinated mothers [30, 37, 45].
Because of the high prevalence of HIV infection and malaria in the developing world, literature on the effects of these diseases on placental function was of particular interest. One study by De Moraes-Pinto and Almeida  in Brazil demonstrated that the rate of placental transfer of IgG antibody to measles virus from HIV type 1—infected mothers to their newborns was significantly lower than among control subjects. In a study by De Moraes-Pinto et al.  in Malawi, reduced placental transfer was associated with placental malaria and maternal hypergammaglobulinemia.
Decay of MMA during infancy. The rate of decline of MMA during infancy is important to ascertain since steeper declines (other factors held equal) will result in lower levels of MMA at each month-old interval and, presumably, greater responsiveness to measles vaccination. We found no prospective studies that followed up infants in multiple countries throughout their first year of life with serial blood specimens to identify determinants of rates of decay of MMA across populations. We did identify 7 prospective studies and several cross-sectional seroprevalence studies done in single countries.
The prospective studies and the methods used to determine antibody half-life are summarized in table 3; these studies were performed in both developed and developing countries [19, 29, 42, 48]. The half-life of MMA ranged from ∼40 to 64 days (all studies) and between 46.1 and 60.8 days (limiting analysis to those studies using the HI test). Two of these studies reported a linkage between higher newborn antibody titer levels and steeper decay curves (i.e., shorter antibody half-life).
Studies of the seroprevalence of MMA that we identified were typically not undertaken for the purpose of determining rates of antibody decay. Rather, their goal was to determine the target age for vaccination and the efficacy of various types of measles vaccine. Results from vaccine trials generally did not provide enough datum points from which to construct a seroprevalence profile. With these caveats and the understanding that seroprevalence profiles are only an indirect measure of decay rates, a summary of seroprevalence studies that we identified is provided in table 4 and figure 1. These studies are grouped by the continents in which they were performed. To maximize comparability, we limited the studies to those that reported data as “percent seropositivity” and provided at least 2 (preferably 3) datum points when the infants were aged 0–12 months. Studies that used only GMTs or age intervals, as opposed to discrete ages, were not included [13, 19, 33, 56, 70–72]. To facilitate comparisons, information regarding the study population, design, and tests used is shown in table 4.
Visual inspection of the graphs of seroprevalence in the various countries does not reveal a distinct continental variation. An interesting pattern that appears is a slow decrease in the prevalence of MMA in the early months of life followed by a more abrupt decline. In essence, most seroprevalence curves reach a minimum level of seropositivity at age 7–9 months, and many increase at greater ages, probably due to increasing exposure to measles virus.
Many studies evaluated the impact of various factors on seroprevalence. The results of these studies are summarized in table 5. It should be noted that these studies varied greatly in sample size and power to detect differences. Apart from the most obvious factor, infant age, very few factors were identified that impacted significantly on seroprevalence. Although vaccination status of the mother was found to be significant in one study  and implied in another , it is likely that this factor impacts antibody titers at birth rather than the rate of antibody decay among the infant. The influences of maternal antibody titer, mother's age (often a correlate of vaccination status), and cord blood antibody titer are intuitive but provided little information regarding determinants of rates of antibody decay. There was no evidence that the nutritional status of infants was significantly related to seroprevalence.
Researchers have speculated on other various factors to explain differences in rates of decay of MMA between countries (e.g., incidence of diarrhea and respiratory infections and catabolic rate of IgG per capita income) [12, 17]. However, we found no data in the previously reported literature to explain the relationship between these factors and decay of MMA.
The literature review provided us with data that help to explain variations in the seroprevalence of MMA among infants around the world. With careful consideration of the limitations of cross-study comparisons, we were able to identify important factors to help guide future research and policy decisions.
The available data did not allow a determination as to whether mothers born in developing countries, where measles is typically endemic, have higher titers of antibody to measles virus than do those in developed countries. First, there were few comparative studies, and second, the extent to which natural disease contributed to immunity, as opposed to vaccination, was not well documented and difficult to quantify [26, 27, 32].
A factor within populations (in single-country studies) that was repeatedly found to affect antibody levels in pregnant women was history of natural or vaccine exposure to measles virus. This finding was further supported by several studies that documented the boosting effect of measles exposure. The broader geographic significance of other factors identified in single-country studies (e.g., age, race, parity, and socioeconomic status) is unknown.
Our review indicates a trend toward more efficient placental transfer of MMA to the newborn in developed countries, compared with developing countries. The fact that comparisons of concentration ratios (for antibody across the placenta) are less affected by variations in sensitivity and type of test adds more credibility to our observation.
We reviewed studies that evaluated the influence of placental disease, prematurity, maternal nutrition, and parity on placental efficiency. Gestational age was found to vary directly with the degree of placental transfer of antibody, with prematurity having an adverse effect. Could it be that a higher frequency of lower gestational age births in developing countries results in newborns starting out with relatively smaller amounts of antibody? Parity is certainly greater in developing countries, and it can be argued that there is a higher prevalence of certain diseases that might have an adverse effect on placental transfer. Although cross-country studies by Eghafona et al.  and Sinha  suggest that multiparity is inversely related to the seroprevalence of MMA, it remains an open question whether multiparity affects the starting maternal level or the efficiency of placental transfer. The adverse effect of maternal HIV infection and malaria on placental transfer of MMA could have significant implications on vaccine policy in countries where these diseases are common. Further study is especially needed in Africa, where premature delivery is common, HIV infection is widespread, and an estimated 24 million women become pregnant each year in areas where malaria is endemic .
In regard to rates of decay of MMA, of the 7 prospective studies identified, 3 were performed in developed countries. Although comparisons were somewhat limited because of the different methods used to determine antibody half-life, we observed that determinations of the half-life of MMA are remarkably similar and that there appears to be no observed relationship between decay rate and country development. From a mathematical analysis of age-stratified seroprevalence curves from studies in the United States and 7 developing countries, McLean and Anderson  also concluded that “there is no significant difference in the rate of loss of protection by maternal antibodies between children in developed and developing countries.” Antibody titer at birth has been identified in 2 of these studies to have an inverse relationship with the rate of antibody decay; this finding should be investigated in future studies. Review of many cross-sectional studies revealed a paucity of infant-related factors that could be determined to influence seroprevalence (the surrogate for decay rate) among the infant populations studied and lack of a large number of recent studies of diverse populations that would better describe contemporary trends in the prevalence of MMA. Most seroprevalence profiles revealed a nadir in the prevalence of antibody to measles virus at age ≤9 months. All other things being equal, this finding tends to support the current WHO strategy of measles vaccination at age 9 months.
Even though our review identified vaccination status and/or measles exposure as the major determinant of antibody titers in mothers within countries, data were not available to demonstrate whether vaccination status was the primary determinant of the variation in levels of antibody to measles virus in mothers from different countries. Although it would be generally expected that there will be a decreasing seroprevalence of MMA among infants as increasing numbers of vaccinated females in the developing world reach childbearing age, variability in several factors could influence this expectation in either direction (e.g., exposure to natural disease, prematurity, and parity). Changes in the seroprevalence of MMA among infants that are due to these and other unknown factors will influence susceptibility to measles, immunogenicity of measles vaccine, and transmission patterns in infancy. Therefore, increasing our understanding of how the prevalence of MMA is determined is crucial to the formulation of vaccination policy, especially for those populations where little other information is available.
In the absence of studies on the seroprevalence of MMA that are performed later in infancy, newborn cord blood analyses may provide the best and most accessible data by which to predict patterns of seroprevalence of MMA and vaccine efficacy later in infancy. Absolute levels of MMA in newborns that are expressed as GMTs would be more useful than rates of seropositivity of MMA in newborns. Cord blood is fairly easy to obtain, and its analysis would “factor out” the variations seen in maternal antibody levels and the efficiency of placental transfer and their impact on the seroprevalence of MMA. If newborn cord blood levels of antibody are validated as a predictor of interpopulation MMA seroprevalence profiles, vital information could be provided to national level policy makers who struggle with the decision of when to vaccinate against measles in infancy.
Future investigations should further evaluate the impact of common diseases, especially those prevalent in developing countries, on placental transfer of antibody to measles virus. Inclusion of WHO international reference sera in assays will facilitate comparisons across studies. Research should continue in areas that could mitigate the ability of MMA to interfere with seroconversion after measles vaccination (e.g., use of second-dose strategies, recombinant DNA technology, and vitamin A supplementation) [4, 75]. As population differences in MMA are further understood, policy on measles vaccination will become less empirical and better able to fulfill the objectives of measles control and elimination.
We thank Drs. Lauri Markowitz and Rita Helfand for their assistance in reviewing the manuscript.