The association of maternal fat-soluble antioxidants in early pregnancy with gestational diabetes mellitus: a prospective cohort study

Gestational diabetes mellitus (GDM) is a common pregnancy complication with rising global incidence and substantial long-term health impacts on mothers and offspring. Diagnosis typically occurs at 24–28 weeks via a 75-g OGTT, underscoring the need for early-pregnancy predictors and modifiable risk factors. Oxidative stress contributes to GDM pathophysiology through damage to proteins, lipids, and DNA, potentially impairing vascular and pancreatic β-cell function. Given that fat-soluble antioxidant vitamins A and E can neutralize free radicals and are critical in maternal-fetal health, the study hypothesized that insufficient antioxidant defenses in early pregnancy might predispose to GDM. The primary research question was whether early-pregnancy serum vitamins A and E are associated with subsequent GDM risk and whether they can serve as early predictive biomarkers.

Prior research links oxidative stress to adverse pregnancy outcomes including GDM, with evidence of impaired radical scavenging and elevated oxidative markers in GDM. Studies assessing antioxidant vitamins in mid-to-late pregnancy have yielded inconsistent findings, and few evaluated early pregnancy levels. Some reports observed lower fat-soluble antioxidant levels in GDM, though causality is unclear due to timing of measurement and observational design. There is mixed evidence regarding vitamin A: some studies reported higher mid-pregnancy retinol associated with preeclampsia, while others suggested low first-trimester retinol predicted insulin-treated GDM. For vitamin E, a systematic review found lower levels in GDM versus controls, but causality remained uncertain. The vitamin E to cholesterol ratio has been proposed as a better indicator of vitamin E status and was associated with reduced risk of early-onset preeclampsia. These heterogeneous findings motivated assessing early-pregnancy antioxidant vitamin status for predicting GDM.

Design: Prospective cohort within the Maternal and Infant Health birth cohort in Beijing. Recruitment of singleton pregnancies <13 gestational weeks at Beijing Daxing Maternal and Child Care Hospital from Nov 2016 to Dec 2017. Eligibility: ≥18 years, planned delivery and infant care at the hospital. Of 982 enrolled, 805 delivered live singletons; after excluding 3 with preexisting diabetes and those without required measurements, 667 women remained for analysis. Ethics approval obtained (Capital Institute of Pediatrics, SHERLL-2016034); written informed consent obtained. Exposure assessment: Fasting serum collected at first prenatal visit (median 9 weeks; IQR 8–10) between 08:00–10:00. Vitamin A (retinol) and vitamin E (α-tocopherol) measured by HPLC (Agilent, USA) with Westgard multirule QC; standards from Sigma. Routine labs included fasting blood glucose (FBG) and total cholesterol (Olympus AU640). Given interpretive value, vitamin E/cholesterol ratio was also analyzed. Outcome: GDM diagnosis at 24–28 weeks using 75-g OGTT; venous glucose at 0, 1, 2 h. GDM defined if any threshold met or exceeded: fasting ≥5.10 mmol/L; 1 h ≥10.00 mmol/L; 2 h ≥8.50 mmol/L (IADPSG criteria). Covariates: Baseline characteristics via standardized questionnaire; pre-pregnancy BMI from self-reported weight and height; FBG and total cholesterol at enrollment, OGTT glucose values, and early gestational weight gain (GWG; before 18 weeks) from electronic records. Vitamin supplementation information collected at enrollment. Statistical analysis: Continuous variables summarized as mean±SD or median (IQR); categorical as percentages. Between-group comparisons used χ², t test, or Wilcoxon test. Vitamin A and E were skewed; transformed to z scores (based on non-GDM SD). Analyzed as continuous (z score) and quartiles. Missing covariate data were not imputed (family history of diabetes n=4; GWG n=22; FBG n=3); affected observations excluded listwise in relevant models. Associations with GDM assessed via logistic regression: Model 1 univariate; Model 2 adjusted for maternal age, family history of diabetes, pre-pregnancy BMI, GWG in early pregnancy, and FBG at enrollment. Vitamin E/cholesterol ratio analyzed similarly (z score and quartiles). Sensitivity analysis excluded participants reporting multivitamin use in early pregnancy. Correlations between vitamins and FBG/OGTT glucose were examined using Pearson correlation. Predictive performance evaluated using ROC analysis and AUC. Analyses performed in R 3.5.1; two-sided p<0.05 considered significant.

- Cohort: 667 pregnant women; 93 developed GDM (13.94%). Median early-pregnancy serum concentrations: vitamin A 0.47 mg/L (IQR 0.41–0.53); vitamin E 10.6 mg/L (IQR 9.3–12.1). - Baseline differences (GDM vs non-GDM): higher pre-pregnancy BMI (24.16±3.92 vs 22.34±3.51 kg/m²; p<0.0001), higher early GWG (0.227 vs 0.172 kg/wk; p=0.010), more family history of diabetes (20.4% vs 12.5%; p=0.037), higher FBG at enrollment (5.07±0.43 vs 4.82±0.34 mmol/L; p<0.0001). Vitamin A higher in GDM (0.51 vs 0.46 mg/L; p<0.0001); vitamin E similar (p=0.50). - Associations with GDM: • Vitamin A (retinol) per SD (z score): OR 1.75 (95% CI 1.42–2.17; p<0.0001) unadjusted; OR 1.46 (95% CI 1.14–1.88; p=0.0032) adjusted. • Vitamin A quartiles (adjusted): Q4 vs Q1 OR 2.25 (95% CI 1.02–4.98; p=0.046); significant trend across quartiles (Ptrend=0.016). • Vitamin E (α-tocopherol): no significant association as continuous (adjusted OR 1.21; 95% CI 0.97–1.51; p=0.094) or by quartiles (Ptrend=0.99). • Vitamin E/cholesterol ratio: continuous not significant (adjusted OR 1.04; 95% CI 0.82–1.31; p=0.74), but quartiles showed a protective trend (Ptrend=0.043); Q4 vs Q1 adjusted OR 0.52 (95% CI 0.26–1.05). - Correlations: Vitamin A positively correlated with FBG at enrollment (r=0.090; p<0.05) and with OGTT glucose (FBG r=0.22; 1-h r=0.18; 2-h r=0.24; all p<0.0001). Vitamin E negatively correlated with FBG at enrollment (r=−0.085; p<0.05); no significant correlation with OGTT FBG. - Predictive performance: Vitamin A alone AUC 0.649–0.653; comparable to FBG at enrollment (AUC 0.679). Best performance with multivariate model including vitamin A, maternal age, family history of diabetes, pre-pregnancy BMI, early GWG, and early FBG: AUC 0.760 (95% CI 0.705–0.815; p<0.001), significantly higher than vitamin A alone (AUC difference 0.111; p=0.0002). - Sensitivity analyses excluding multivitamin users showed similar directions and magnitudes of associations; vitamin A remained positively associated with GDM risk, with an attenuated Q4 vs Q1 effect likely due to reduced sample size.

The study demonstrates that higher early-pregnancy serum vitamin A is associated with increased odds of subsequent GDM, supporting a link between antioxidant status and glucose dysregulation early in gestation. The positive correlations between vitamin A and fasting and post-load glucose measures reinforce its relationship with glycemia. Vitamin E showed no overall association, although the vitamin E/cholesterol ratio exhibited a trend toward reduced GDM risk across quartiles, aligning with the concept that the ratio better reflects vitamin E status. These findings refine current understanding by highlighting that, contrary to an initial hypothesis of deficiency, elevated vitamin A may relate to GDM pathogenesis. Potential mechanisms include retinol transport and signaling pathways (e.g., RBP4/TTR axis) and hepatic metabolism alterations during pregnancy, though causality cannot be established here. The predictive modeling indicates that vitamin A adds discriminative value when combined with established clinical risk factors, yielding moderate accuracy (AUC 0.760), suggesting utility as part of a multi-marker early screening strategy. Results are consistent with some prior studies reporting higher vitamin A in adverse pregnancy outcomes, while differing from reports of lower antioxidants in GDM measured later in pregnancy, underscoring the importance of timing and biomarker selection (e.g., α-tocopherol/cholesterol).

Higher vitamin A concentrations in early pregnancy are significantly associated with increased GDM risk, and vitamin A may serve as a useful biomarker for early identification when combined with clinical risk factors. Vitamin E alone was not associated with GDM, but the vitamin E/cholesterol ratio showed a protective trend. Future research should investigate mechanistic pathways (including RBP4 and transthyretin), incorporate detailed dietary and lifestyle assessments, validate findings in diverse populations, and develop multi-biomarker models to enhance early prediction and guide interventions.

- Retinol transport proteins (RBP4) and transthyretin were not measured. - No detailed dietary intake or physical activity data, leaving potential residual confounding. - Single-center recruitment in a district-level women and children’s hospital in Beijing may limit generalizability. - The study was ancillary to a larger cohort without a prespecified power calculation; some analyses may be underpowered. - Self-reported pre-pregnancy weight for BMI may introduce measurement error; small amounts of missing covariate data were not imputed.