Maternal Body Mass Index, Diabetes, and Gestational Weight Gain and Risk for Pediatric Cancer in Offspring: A Systematic Review and Meta-Analysis

Abstract

Background: Pediatric cancer incidence has steadily increased concurrent with rising adult obesity, but associations between maternal obesity and associated comorbidities and pediatric cancer risk remain understudied. We aimed to quantitatively characterize associations of pediatric cancer risk with maternal prepregnancy body mass index (BMI), gestational weight gain, and maternal diabetes.

Methods: We performed a comprehensive and systematic literature search in Ovid and EMBASE from their inception to March 15, 2021. Eligible studies reported risk estimates and sample sizes and provided sufficient description of outcome and exposure ascertainment. Random effects models were used to estimate pooled effects.

Results: Thirty-four studies were included in the analysis. Prepregnancy BMI was positively associated with leukemia risk in offspring (odds ratio [OR] per 5-unit BMI increase =1.07, 95% confidence intervals [CI] = 1.04 to 1.11; I2 = 0.0%). Any maternal diabetes was positively associated with acute lymphoblastic leukemia risk (OR = 1.46, 95% CI = 1.28 to 1.67; I2 = 0.0%), even after restricting to birthweight-adjusted analyses (OR = 1.74, 95% CI = 1.29 to 2.34; I2 = 0.0%), and inversely associated with risk of central nervous system tumors (OR = 0.73, 95% CI = 0.55 to 0.97; I2 = 0.0%). Pregestational diabetes (OR = 1.57, 95% CI = 1.11 to 2.24; I2 = 26.8%) and gestational diabetes (OR = 1.40, 95% CI = 1.12 to 1.75; I2 = 0.0%) were also positively associated with acute lymphoblastic leukemia risk. No statistically significant associations were observed for gestational weight gain.

Conclusions: Maternal obesity and diabetes may be etiologically linked to pediatric cancer, particularly leukemia and central nervous system tumors. Our findings support weight management and glycemic control as important components of maternal and offspring health. Further validation is warranted.

Figure 1.

Preferred Reporting Items for Systematic reviews and Meta-analyses flow diagram of study selection and study identification.

Figure 2.

Heat map of the associations between prepregnancy BMI, gestational weight gain, maternal diabetes, and risk of pediatric cancers. Cells depicting statistically significant or suggestive associations display meta-analyzed odds ratios (95% confidence intervals). † Considered suggestive if upper- and lower-bound confidence intervals were 1.00 to 1.05 and 0.95 to 1.00, respectively. *For inadequate gestational weight gain (Institute of Medicine defined guidelines) only. ALL = acute lymphoblastic leukemia; AML = acute myeloid leukemia; BMI = body mass index; CNS = central nervous system.

Figure 3.

Forest plots: meta-analysis of the association between a 5-unit increase in prepregnancy BMI and risk of (A) any leukemia, (B) central nervous system tumors, and (C) embryonal central nervous system tumors. The error bars represent the 95% confidence intervals (CIs). Fixed effects models (inverse variance method) were used for panels A and C, and random effects models (DerSimonian and Laird method) were used for panel B. Tests were 2-sided. DL = DerSimonian-Laird; IV = inverse variance.

Figure 4.

Forest plots: meta-analysis of the association between any diabetes and risk of (A) any leukemia, (B) acute lymphoblastic leukemia, (C) central nervous system tumors, (D) lymphoma, and (E) Wilms tumor. The error bars represent the 95% confidence intervals (CIs). Fixed effects models (inverse variance method) were used for statistical analyses. Tests were 2-sided. ALL = acute lymphoblastic leukemia; AML = acute myeloid leukemia; IV = inverse variance.

Figure 4.

(Continued)

Figure 5.

Forest plots: meta-analysis of the association between pregestational diabetes and (A) any leukemia and (B) acute lymphoblastic leukemia. The error bars represent the 95% confidence intervals (CIs). Random effects models (DerSimonian and Laird method) were used for statistical analyses. DL = DerSimonian-Laird.

Figure 6.

Forest plots: meta-analysis of the association between gestational diabetes and risk of (A) any leukemia and (B) acute lymphoblastic leukemia. The error bars represent the 95% confidence intervals (CIs). Fixed effects models (inverse variance method) were used for statistical analyses. Tests were 2-sided. ALL = acute lymphoblastic leukemia; AML = acute myeloid leukemia; IV = inverse variance.

Figure 7.

Forest plot: meta-analysis of the association between inadequate gestational weight gain and risk of central nervous system tumors. The error bars represent the 95% confidence intervals (CIs). Random effects models (DerSimonian and Laird method) were used for statistical analyses. Tests were 2-sided. DL = DerSimonian-Laird.

Figure 8.

Funnel plots for meta-analysis results of the associations between prepregnancy BMI, maternal diabetes, and gestational weight gain and risk of pediatric cancers. A) Funnel plot for the association between prepregnancy BMI and risk of any leukemia; (B) funnel plot for the association between prepregnancy BMI and risk of central nervous system tumors; (C) funnel plot for the association between prepregnancy BMI and risk of embryonal central nervous system tumors; (D) funnel plot for the association between any diabetes and risk of any leukemia; (E) funnel plot for the association between any diabetes and risk of acute lymphoblastic leukemia; (F) funnel plot for the association between any diabetes and risk of central nervous system tumors; (G) funnel plot for the association between any diabetes and risk of lymphoma; (H) funnel plot for the association between any diabetes and Wilms tumor risk; (I) funnel plot for the association between pregestational diabetes and risk of any leukemia; (J) funnel plot for the association between pregestational diabetes and risk of acute lymphoblastic leukemia; (K) funnel plot for the association between gestational diabetes and risk of any leukemia; (L) funnel plot for the association between gestational diabetes and risk of acute lymphoblastic leukemia; (M) funnel plot for the association between inadequate gestational weight gain and risk of central nervous system tumors. BMI = body mass index; logor = natural log of the odds ratio; s.e. = standard error.

Figure 8.

(Continued)