https://orcid.org/0000-0003-3791-2305
Figures & data
Figure 1. Schematic diagram for data collection and the association between maternal smoking during pregnancy and BMI z-scores of infants.
(a) A schematic diagram for data collection. Questionnaires were collected prenatally and throughout the first three years of life. BMI scores were calculated based on weight and length/height data collected at age 1 year and age 3 years. Stool samples were collected at 3 and 12 months, and analyzed by amplicon sequencing of the V4 region of 16S rRNA and magnetic resonance spectroscopy for quantification of metabolites. (b) Distributions of BMI z-scores at 1 and 3 years. Weight status was categorized into “Wasted” (BMI z-score < -2), “Normal” (-2 ≤ BMI z-score ≤1), “At-risk overweight” (1 < BMI z-score ≤2), “Overweight” (2 < BMI z-score ≤3) and “Obese” (BMI z-score >3). (c) Association between maternal smoking during pregnancy and BMI z-scores of infants. Maternal race/ethnicity, socio-economic status, and maternal pre-pregnancy BMI (except in the models assessing joint effects of maternal BMI and smoking status) were adjusted in Model 1; prenatal and postnatal pet exposure, breastfeeding status at three months (except in the models assessing joint effects of maternal smoking status and breastfeeding status at three months), solid food introduction at three months, and household disinfectant use were additionally adjusted in Model 2; and prenatal diet calories, infant sex, birth mode and antibiotic use in the first three months were further adjusted in Model 3. Maternal education level was used as a proxy to the socio-economic status (SES) of the family. Maternal smoking status during pregnancy was categorized into 1) never smoker, 2) quit smoking before conception, 3) quit smoking during pregnancy, 4) cut the number of cigarettes during pregnancy, and 5) had the same number of cigarettes during pregnancy. In modeling of the effect of detailed maternal smoking statuses during pregnancy, cutting the number of cigarettes during pregnancy and having the same number of cigarettes during pregnancy were combined into one category (“Yes”) due to small sample sizes. These detailed statuses were also dichotomized into 1) “Yes/Smoker”: those who quit smoking during pregnancy, reduced the number of cigarettes, or had the same number of cigarettes (as before conception) during pregnancy; and 2) “No/Non-smoker”: those who quit smoking before conception, or never smoked. NW, normal weight. OWOB, overweight or obese. AROW, at-risk of overweight.
Table 1. Characteristics of the study subjects by exposure to maternal smoking during pregnancy.
Figure 2. Associations between maternal smoking during pregnancy and profiles of gut microbiota in infancy.
(a) Principal coordinate analysis (PCoA) of beta diversity among the collected samples based on Bray-Curtis dissimilarity. Diagrams at the top side and on the right side represent the density of samples along the PCo1 and PCo2, respectively. P values were given by adonis tests. (b) Differentially abundant taxa identified by LEfSe. “Yes”: samples from infants whose mothers quit smoking during pregnancy, reduced the number of cigarettes, or had the same number of cigarettes (as before conception) during pregnancy. “No”: samples from infants whose mothers quit smoking before conception, or never smoked.
Figure 3. Early-infancy microbial mediators in the pathway from maternal smoking during pregnancy to weight outcomes at 1 and 3 years.
(a) Relative abundance or value of indices of the mediators by exposure of maternal smoking during pregnancy, year-1 weight status, and year-3 weight status. The category “wasted” was included here to show the trend, but not included in the downstream analysis because of its small sample size and being out of the scope of the study. P values: “***”, <0.001; “**”, <0.01; “*”, < 0.05; “.” < 0.1. P values in were FDR-corrected. (b) The proportions of the effect of maternal smoking during pregnancy on childhood weight outcome mediated through the microbial mediators. In the mediation analysis, maternal pre-pregnancy BMI, maternal race/ethnicity, prenatal and postnatal pet exposure, breastfeeding status at 3 months, solid food introduction at three months, and household disinfectant use, prenatal diet calories, infant sex, birth mode and antibiotic use in the first 3 months and age of sample collection (in months) were adjusted in the mediator (gut microbiota)-outcome associations, while maternal race/ethnicity and maternal education level was considered in the exposure (maternal smoking during pregnancy)-outcome association. The sizes of orange triangles reflected the mediation proportions. Significant mediation effects were marked in red.
Figure 4. Metabolite mediators.
(a) Fecal concentration of butyrate, pyruvate, and succinate in early infancy by maternal pre-pregnancy weight status and maternal smoking during pregnancy. (b) Correlations between butyrate, pyruvate, and succinate and related microbial pathways. c, Distributions of selected microbial pathways by maternal smoking during pregnancy. (d) Correlations between the selected microbial pathways and BMI z-scores at 1 and 3 years. P values: “***”, <0.001; “**”, <0.01; “*”, < 0.05; “.” < 0.1. P values in B, C, D were FDR-corrected.
Supplemental material
R2_Supplementary_materials_clean.docx
Download MS Word (2.8 MB)Data availability statement
De-identified data are available from the corresponding author upon reasonable request. Registration and/or proposal applications may be needed depending on the tier of access requested (https://childstudy.ca/for-researchers/data-access/).