Gestational vitamin D deficiency and autism-related traits: the Generation R Study

Population-based prospective cohort study (Generation R) in Rotterdam, The Netherlands. Included 4229 children with maternal mid-gestation and/or cord blood 25OHD measurements and Social Responsiveness Scale (SRS) scores at ~6 years; 3169 children had offspring genotype data; 2489 had 25OHD data at both time points. Vitamin D status was defined as deficient (<25.0 nmol/L) or insufficient/sufficient per thresholds. Analyses used linear mixed models adjusting for covariates including ethnicity, sex, age at SRS, birth weight, gestational age at birth, maternal age, paternal age, smoking during pregnancy, educational level, and maternal BMI; genome-wide genotype data were used to account for sample structure and relatedness; 25OHD measured by LC-MS/MS; timepoints: mid-gestation and birth (cord blood).

Gestational 25OHD deficiency (<25 nmol/L) at mid-gestation or at birth was associated with higher SRS scores at ~6 years, indicating more autism-related traits. Compared with 25OHD sufficient, mid-gestation deficiency showed β = 0.06 (SE 0.01, P < 0.001); cord blood deficiency β = 0.03 (SE 0.01, P = 0.01). Insufficiency at mid-gestation also associated (β = 0.02, SE 0.01, P = 0.007). Deficiency at both time points predicted higher SRS (β = 0.07, SE = 0.02, P < 0.001). When SRS was treated as a dichotomous outcome, mid-gestation deficiency increased odds of high-SRS by 3.8-fold. In continuous analyses, lower 25OHD concentrations were associated with higher SRS at both time points (mid-gestation β = -0.07, P < 0.001; cord β = -0.05, P = 0.002). Patterns persisted after adjusting for ethnicity using genetic data and after adjusting for season of sampling. Interpretation: Prenatal vitamin D deficiency may represent a graded risk factor for autism-related traits in the general population, suggesting that safe, inexpensive prenatal vitamin D supplementation could have public health relevance; randomized controlled trials are needed to establish causality.

Observational design cannot establish causality. Vitamin D was measured at only two gestational time points; no data on other stages of gestation or early life to identify a precise critical window. Differential attrition and unmeasured confounding may bias results; however, analyses adjusting for ancestry and season remained significant. SRS is a parent-reported screening tool, not a clinical autism diagnosis, and findings require replication in independent cohorts.