The association of dietary total antioxidant capacity and gestational diabetes: a prospective cohort study from the Mothers and their children's health (MATCH)

Gestational diabetes mellitus (GDM) is glucose intolerance first recognized during pregnancy and affects approximately 4–9% of pregnancies, with short- and long-term adverse effects for mothers and offspring. Known GDM risk factors include advanced maternal age, overweight/obesity, family history of diabetes, dietary patterns, micronutrient deficiencies, and oxidative stress. Nutritional factors—particularly plant-based dietary patterns rich in fruits and vegetables—are linked to lower chronic disease risk and have anti-inflammatory and antioxidant properties. While fruits and vegetables are major antioxidant sources, other foods (whole grains, nuts, coffee, chocolate) also contribute substantially. Studying single antioxidants or food groups may not capture synergistic effects; therefore, dietary total antioxidant capacity (DTAC) was developed to reflect the cumulative free-radical–scavenging potential of the overall diet. High DTAC has been associated with health benefits, but its relationship with GDM risk has been understudied. This study aimed to examine whether higher DTAC in early pregnancy is associated with reduced GDM risk in a prospective cohort using rigorous confounding control.

Prior work links higher intake of antioxidant-rich foods and nutrients to reduced inflammation and oxidative stress and suggests lower GDM risk with higher dietary antioxidants or vitamin C intake. DTAC, as an aggregate measure of dietary antioxidant potential (often calculated by FRAP), has been associated with favorable metabolic profiles and reduced odds of prediabetes and insulin resistance in various populations. Evidence directly evaluating DTAC and GDM has been limited; a case-control study found higher FRAP-based DTAC associated with markedly reduced GDM odds, while other TAC assays (TRAP, TEAC) were not significant. Additional studies show inverse associations between DTAC and prediabetes and insulin resistance, and between dietary antioxidant quality and type 2 diabetes prevalence, supporting a potential protective role of antioxidant-rich dietary patterns in glucose metabolism.

Design and population: Prospective cohort study (MATCH) of pregnant women aged 18–45 years initiated in 2020. Eligibility included gestational age <12 weeks confirmed by β-hCG. Exclusions: metabolic/chronic diseases, multiple pregnancy, special diets, certain supplements (except routine prenatal iron/folate), and physical/mental/cognitive disabilities. Ethics approvals obtained; informed consent collected and renewed biannually. From 3285 screened, 2103 eligible; 1856 with complete data included. Data collection: Four time points—baseline (<12 weeks) captured maternal/paternal demographics, lifestyle, medical/reproductive history, nutrition, sleep; blood samples and anthropometry collected at baseline; follow-ups at ~24 and ~32 weeks (physical activity, sleep), and 12–72 h postpartum (birth outcomes). Dietary assessment: A validated 168-item semi-quantitative FFQ (Iranian population; TLGS validation) assessed usual intake in the year prior to pregnancy. Intakes converted to grams; energy and nutrient intakes calculated (Nutrition IV). DTAC estimation used the ferric reducing ability of plasma (FRAP) method with values from published databases (Halvorsen et al., Guo et al.). For each food, Antioxidant Intake = frequency × portion size × FRAP per unit; DTAC = sum across foods (expressed per 100 g units). Anthropometry: Height, waist, and hip at enrollment; weight at all four visits. Measurements taken twice (and third if discordant) following standardized protocol. Lifestyle and other variables: Standardized questionnaire on sociodemographics, reproductive history, physical activity, smoking and alcohol, medical and family history, sleep disorders. Biospecimens and outcome: Blood analyzed at a single lab within 4 h. GDM defined per ADA criteria: fasting glucose ≥92 mg/dL, 1-h ≥180 mg/dL, or 2-h ≥153 mg/dL during OGTT. Quality control: Comprehensive participant tracking; weekly checks for interviewer bias; monthly independent review of 10% of data; rigorous interviewer training and standardized protocols. Sample size: Planned n=1590 (318/group) allowing 10% dropout to detect 40% difference (RR=1.4) in GDM between DTAC quartiles with 80% power. Statistical analysis: Descriptive statistics with ANOVA/t-test/Chi-square; normality assessed via Kolmogorov–Smirnov and Q–Q plots. Missing data addressed via multiple imputation by chained equations. Because DTAC quartiles differed on baseline covariates, propensity scores were estimated using generalized boosted models (TWANG; 10,000 trees) to derive inverse probability of treatment weights (IPTW). Minimal sufficient adjustment sets identified via directed acyclic graphs (dagitty). Associations between DTAC (quartiles with Q1 reference and continuous per 1 SD) and incident GDM estimated using weighted modified Poisson regression to obtain risk ratios (RRs) with 95% CIs. Analyses conducted in Stata 17 and R 4.2.1.

- Cohort: 1856 pregnant women with complete data; mean age 32.9 ± 6.1 years; mean BMI 25.9 ± 8.3 kg/m². - GDM cases: 369/1856 (19.9%). - Mean DTAC: 2.82 ± 2.56 mmol/100 g (range 0.01–18.55). - GDM incidence by DTAC quartile: Q1 25.16% (117/465), Q2 21.01% (97/462), Q3 18.22% (84/461), Q4 15.17% (71/468). - Crude RRs vs Q1: Q2 0.83 (95% CI 0.65–1.05), Q3 0.72 (0.56–0.92), Q4 0.61 (0.46–0.78); p-trend < 0.001. - Adjusted RRs (IPTW) vs Q1: Q2 0.32 (95% CI 0.14–0.73, p=0.007), Q3 0.26 (0.11–0.60, p=0.002), Q4 0.29 (0.12–0.68, p=0.005); p-trend < 0.001. - Continuous DTAC (per 1 SD increase): adjusted RR 0.66 (95% CI 0.48–0.90, p=0.023), indicating a 34% lower GDM risk per SD increase in DTAC. - Women in the highest DTAC quartile had about 71% lower adjusted risk than those in the lowest quartile.

Higher DTAC in early pregnancy was associated with a clinically meaningful reduction in GDM risk, with consistent dose-response across quartiles and robustness to adjustment for key confounders identified via DAGs (BMI, age, education, occupation/shift work, pre-existing hypertension and diabetes). These findings align with prior evidence linking antioxidant-rich diets and specific antioxidants (e.g., vitamin C) to lower GDM risk, and with case-control evidence showing strong inverse associations between FRAP-based DTAC and GDM. Additional literature shows DTAC inversely associated with prediabetes and insulin resistance and improved glycemic profiles, supporting biological plausibility. Potential mechanisms include antioxidant-driven modulation of carbohydrate digestion (alpha-amylase/glucosidase inhibition), effects on glucose transport (e.g., GLUT2 interactions), activation of Nrf2-mediated antioxidant pathways, and protection of pancreatic beta cells from oxidative damage, which may improve insulin secretion and reduce insulin resistance. The prospective design, assessment of diet before GDM onset, comprehensive covariate collection, use of propensity score weighting with machine learning–based GBM, and explicit confounder selection using DAGs strengthen causal inference. Using RRs from a cohort improves interpretability compared with ORs from case-control designs. Overall, the results support a protective association between higher DTAC and GDM risk.

In a large prospective cohort of pregnant women, higher dietary total antioxidant capacity in early pregnancy was associated with substantially lower risk of gestational diabetes mellitus, with a clear dose-response. These findings underscore the potential importance of antioxidant-rich dietary patterns in GDM prevention. Future research should include multi-center cohorts, incorporate insulin resistance biomarkers to elucidate mechanisms, and consider interventional studies to confirm causality and inform dietary recommendations.

- Selection/generalizability: Participants were recruited from a single specialized gynecological hospital in eastern Tehran, which may limit external validity and introduce selection bias. - Lack of insulin resistance measures: Absence of indices such as HOMA-IR limits mechanistic insights into DTAC’s effects on insulin sensitivity and beta-cell function. - Potential performance/information bias: While mitigated through rigorous QA/QC, standardized protocols, and centralized laboratory analyses, observational design remains susceptible to residual bias and confounding. - Missing data handled via multiple imputation; unmeasured confounding cannot be fully excluded.