Introduction

Asthma is one of the common chronic respiratory diseases, with an estimated prevalence of 334 million worldwide1. Recently, the prevalence of asthma increased markedly among countries with Western lifestyles2. Asthma is also one of the top-10 chronic diseases for disability-adjusted life for 5- to 14-year-old children. Globally, asthma causes approximately 250,000 deaths a year, imposing a heavy burden on health system3. Therefore, the identification of risk factors for infant asthma is of major significance for early intervention and treatment of asthma4,5.

Asthma is considered to be the result of a combination of genetic and environmental risk factors. Changes in nutrition are linked to the development of asthma6. Folate, an essential B vitamin of nutrition, participates in the carrying and chemical activation (as tetrahydrofolates) of one-carbon units for further biosynthesis7. Such folate-mediated one-carbon metabolism plays a crucial role in purine and thymidylate synthesis, amino acid metabolism and S-adenosyl methionine formation8. Through this pathway, folic acid is involved in the synthesis of nucleic acids, methylation of DNA and regulation of cell growth9. Consequently, it plays an irreplaceable role in all of our activities in daily life, especially in the early stages of uterine growth and development10. Due to the growth of the foetus in utero, folate levels in pregnant women may be insufficient, which could lead to a number of birth defects, such as neural tube defect, growth retardation, cardiac defects and oral clefts11,12,13,14. Thus, the supplementation and fortification of food with folic acid are recommended among pregnant women15. In recent years, some countries have implemented mandatory folic acid fortifications16. However, high folate intake during pregnancy is considered to be the cause of some adverse effects in newborn and child health, such as large-for-gestational-age birth and respiratory illness17,18.

Given that pregnant women may consume a relatively higher dose of folic acid, the potential adverse effects on foetal development warrant evaluation. Emerging studies explored the relationship between maternal folic acid intake and the risk of childhood asthma. However, their results are not consistent. Therefore, we reviewed the available studies and performed a meta-analysis to better estimate the association between folate intake and childhood asthma risk.

Methods

We followed the Meta-analysis of the Observational Studies in Epidemiology (MOOSE) guidelines when conducting and reporting this meta-analysis19.

Search strategy

We searched PubMed and SCOPUS databases for relevant studies published through 1 August 2018. The SCOPUS database is an abstract and citation database containing all of the EMBASE databases20. We used the following keywords: “folate” OR “folic acid” in combination with “maternal” OR “pregnancy” in combination with “asthma”. No language or time restrictions were applied. A manual was used for all references of qualified research to identify other potentially relevant studies.

Selection criteria

We included the eligible studies according to the following criteria: (1) the study was peer-reviewed original research; (2) the study was a cohort study; (3) the study provided the risk estimates of asthma associated with maternal folate intake or maternal folate concentration during pregnancy. Studies that provided only a crude estimate were excluded. For articles including the same study, the latest one was selected.

Data extraction and quality assessment

For the included studies, study information, participants, exposure and outcome measurements, effect sizes and related statistics were extracted by two investigators. Disagreements were resolved through discussions. The quality of the eligible studies was assessed by the Newcastle-Ottawa Scale (NOS)21. For each cohort study, the highest score was 9 stars, and studies with 6 or more stars were considered to be of high quality.

Statistical analysis

Multivariable-adjusted odds ratios (ORs), prevalence rates (PRs) or relative risks (RRs) with 95% confidence intervals (95% CIs) were included in the meta-analysis. For studies that provided multiple exposure periods of maternal folate intake, we chose the first trimester because the first trimester is the most critical period of DNA methylation during pregnancy22. For studies that provided RRs with 95% CIs from different lengths of follow-up for asthma, we chose the RRs from the longest length of follow-up for the outcome. Statistical heterogeneity across studies was estimated via the χ2-based Q-statistic, and we considered P < 0.05 to indicate significant heterogeneity. We conducted stratified analyses to search for potential differences in RRs in subgroups by exposure assessment, folate source, exposure period, geographic area, sample size, quality score, publication year, and adjustment for potential confounders (yes or no). In addition, we performed a sensitivity analysis by removing omitting one study at a time and calculating the overall RR for the remaining studies. A dose-response meta-analysis was conducted to explore the trend between folate intake and asthma risk. Furthermore, we used generalized least-squares trend (GLST) estimation to calculate the trend from the relevant log-RR estimates across folate intake category23,24. The publication bias was assessed by Egger’s test and Begg’s visual inspection of funnel plots25,26. All statistical tests were two-sided and performed by STATA software (version 11.2, StataCorp, College Station, TX, USA). P values < 0.05 were considered significant.

Results

Literature search

The flow chart with the literature selection details is presented in Fig. 1. In total, 26 articles were considered for further estimation: 5 articles did not report on infant asthma; 7 articles did not report ORs, RRs, or PRs; and 2 articles included a repeated study, and we chose the latest one and excluded a case control study27. Finally, 12 articles28,29,30,31,32,33,34,35,36,37,38 with 10 studies28,29,30,31,32,33,34,35,36,39 on folate intake and 5 studies29,30,35,37,38 on blood folate concentration were included in our meta-analysis. In addition, the data of the Avon longitudinal birth cohort39 were derived from a review40.

Figure 1
figure 1

Flowchart of the search strategy and study selection process.

Study characteristics

The information of the 10 studies assessing folate intake is shown in Table 1. These studies were published from 2008 to 2018 and included 201,248 participants. Seven studies were conducted in Europe29,30,32,33,35,39,41, two in North America31,34, and one in Australia36. Among the 10 studies that reported folate intake, 5 studies were related to supplemental folate intake29,31,32,33,35, and 5 studies were related to total folic acid intake from diet and supplements28,30,34,36,39. The studies included in the analysis were adjusted for a number of potential confounders, such as maternal age, maternal smoking, maternal asthma history, infant gender. The characteristics of 5 studies focusing on blood folate concentrations were summarized in Table 2. The details of the NOS evaluation method are shown in Supplementary Table 1. Consequently, four studies received 6 stars, five studies received 7 stars, and one study received 8 stars.

Table 1 Baseline characteristics of individual studies on maternal folate intake and infant asthma.
Table 2 Baseline characteristics of individual studies on blood folate concentration and infant asthma.

Maternal folate intake, blood folate concentration, and infant asthma risk

The adjusted RRs of maternal folic acid intake and the risk of infant asthma for each study are shown in Fig. 2. The relationship between maternal folate intake during pregnancy and childhood asthma risk is inconsistent. In short, a summary RR of maternal folate intake was 1.11 in the fixed effects model (95% CI = 1.06–1.07; P = 4.664 × 10−5), revealing that maternal folate intake during pregnancy was significantly associated with the risk of infant asthma. We did not conduct meta-regression analyses to identify the sources of heterogeneity due to the low heterogeneity (P = 0.087).

Figure 2
figure 2

Forest plot showing pooled relative risks and corresponding 95% CIs of infant asthma according to maternal folate intake. The grey squares indicate study-specific relative risks, the horizontal lines represent the 95% CI, and the size of each square is proportional to its weight in the analysis. The diamond represents the summary relative risk estimate with its 95% CI.

In addition, 5 studies29,30,35,37,38 reported the blood folate concentration. Figure 3 shows the RRs for the association of maternal blood folate concentration with infant asthma risk. The heterogeneity of the results was high (P = 0.018), and the pooled RR was 1.04 (95% CI = 0.81–1.35; P = 0.737) in the random models compared with the reference category.

Figure 3
figure 3

Forest plot showing pooled relative risks and corresponding 95% CIs of infant asthma according to blood folate concentration.

In the subgroup analyses of geographic region, we found a significantly increased risk of folic acid intake for Europe (RR = 1.08; 95% CI = 1.01–1.16) and North America (RR = 1.20; 95% CI = 1.11–1.30) (Table 3). The analysis by publication year yielded pooled RRs of 1.03 (95% CI = 0.94–1.12) in 5 studies published before 2013 and 1.15 (95% CI = 1.08–1.22) in 5 studies published after 2013. In addition, the stratified analysis by sample size revealed pooled RRs of 1.00 (95% CI = 0.91–1.10) in 5 studies with under 5000 participants and 1.16 (95% CI = 1.09–1.23) in 5 studies with more than 5000 participants. Furthermore, when we stratified the analysis by quality score, the RR was 1.02 (95% CI = 0.94–1.12) in 4 studies with scores <7, while the RR was 1.15 (95% CI = 1.08–1.22) in 6 studies with scores ≥7. In the stratification studies of folate source intake and asthma risk, we found that folate intake from supplements increased the infant asthma risk (RR 1.12; 95% CI = 1.05–1.18), while the effect was not significant for folate intake from diet and supplements (RR = 1.09; 95% CI = 0.99–1.21).

Table 3 Results of the subgroup analysis for the association between maternal folate intake and infant asthma risk.

Sensitivity analyses

Sensitivity analyses indicated that the overall RR was not markedly influenced by the removal of any single study except for Veeranki S. P. et al.31. (Supplementary Fig. 1).

Dose-response relationship between folate intake and asthma

Studies with available related data were selected to conduct a dose-response analysis30,34. The dose-response relationship between maternal folate intake and infant asthma is shown in Fig. 4. Because no evidence of departure from linearity was found (P = 0.824), we finally assumed a linear relationship in a fixed-effect model (Pheterogeneity = 0.055) with a linear dose-response relationship (Plinearity = 0.042). The result of our dose-response analysis suggests that each 100-mg/day increment in maternal folate intake was associated with a 0.02% higher risk of infant asthma.

Figure 4
figure 4

Dose-response relationship between maternal folate intake and infant asthma. The solid line and the dash line represent the estimated relative risks and corresponding 95% confidence intervals. Folic acid intake was modeled with a linear trend (P-value for non-linearity = 0.82) in a fixed-effects model.

Publication bias

No publication bias was noted with Egger’s test (P = 0.788) or Begg’s funnel plot (Supplementary Fig. 2).

Discussion

The results of this meta-analysis suggested that maternal folate intake is associated with an increased risk of infant asthma. In stratified analyses, we found that such an effect was significant in Europe and North America. The influence remains statistically significant in studies with later years of publication or larger sample sizes or in studies of higher quality.

The study of the relationship between folate intake during pregnancy and asthma during pregnancy was first performed by Granell R. et al. in39. After that, several studies were conducted to further explore the relationship, but their results are not consistent. Thus, Krista S. Crider et al.22 conducted a meta-analysis including 5 studies and ultimately provided no evidence of a significant relationship between maternal folic acid supplement use and infant asthma in offspring (RR = 1.01; 95% CI = 0.78–1.30). Liu Yang et al.27 also performed a meta-analysis including 5 studies in 2015 and found that the connection between maternal folate intake during pregnancy and infant asthma risk was not significant (RR = 1.06, 95% CI = 0.99–1.14). Afterwards, Parr C. L. et al.30 and Veeranki S. P. et al.31 conducted corresponding cohort studies, and found a significant relationship between folic acid intake during pregnancy and infant asthma. Given the increasing number of studies on the relationship between maternal folate intake and infant asthma, we included more literature for further meta-analysis to more thoroughly explore this relationship and to find a significant association between maternal folate intake and infant asthma. In addition, several differences between our meta-analysis and the previous analyses were markedly observed. First, the previous analyses included no more than 5 studies, and the publication year was limited to 2012. Our analysis included 10 studies related to maternal folate intake and 5 studies related to blood folate concentration. Our meta-analysis is more statistically convincing than previous studies due to a relatively larger number of studies and sample sizes. Second, a previous meta-analysis on prenatal folate intake and infant asthma only summed the available evidence qualitatively. In our analysis, we made full use of the available dose data and performed a dose-response analysis, which quantitatively reveals the relationship between maternal folic acid intake during pregnancy and the risk of infant asthma. Last, we further conducted subgroup analyses based on the characteristics of included studies, such as analyses by geographic region, publication year, folate source and other significant factors.

The mechanism by which folic acid affects the development of asthma in children may be achieved via variable DNA methylation in the mother’s uterus. In DNA methylation, a methyl group is transferred from s-adenosylmethionine to cytosine by the action of a transmethylase, which plays a vital role in regulating cell growth42. Hollingsworth et al. found that Runt-related transcription factor 3 (Runx3), a gene known for the prevention of allergic airway disease, was excessively methylated in progeny exposed to a high-methylation diet, resulting in the suppression of Runx3 mRNA and protein levels. Methyl sources are supplemented in pregnant mice to alter DNA methylation and to ultimately predisposed the mice to allergic airway disease by inducing the differentiation of T-lymphocytes to a TH2 phenotype43. Subsequently, a serious number of epidemiologic studies were performed to assess the potential link between maternal folate intake and infant asthma. However, we recognized that the underlying mechanism of folate-induced infant asthma is still limited.

Several strengths were observed in our study. First, our analysis was based on a comprehensive bibliographic search including 201,248 participants, which provide sufficient statistical power for our research. Second, subgroup analyses of included studies were conducted by geographic region, publication year, folate source, and other factors, which indicated the influence of different covariates on maternal folate intake and infant asthma risk. Finally, there was no publication bias in our analysis. However, the limitations of this analysis should be considered in the interpretation of our findings. First, all the studies included in our analysis were adjusted for known infant asthma risk factors, but these factors were not consistent. Second, it is difficult to accurately determine how much folate in natural food and in its synthetic form was consumed (used in multivitamins, prenatal fortified supplement) during pregnancy. Because of misclassifications of folate sources or inaccurate measurements of blood folate concentration, the included studies may have potential bias. Third, the dose-response analysis demonstrated a linear association between maternal folate intake and asthma risk; however, due to the lack of included studies, more dose-response studies are needed to further confirm this linear relationship.

Our meta-analysis showed that maternal folate intake during pregnancy increases the risk of infant asthma. Meanwhile, the dose-response analysis confirmed a linear association between maternal folic acid intake and the risk of infant asthma. Therefore, the adverse effect of folic acid on infant asthma should not be ignored when it was supplemented during pregnancy to prevent birth defects. Further studies are warranted to determine a critical intake dose of folate in pregnancy that can effectively prevent the adverse effect of infant asthma while also preventing birth defects.