Nutritional biomarkers and three-month symptom changes in antidepressant-free adolescents with major depressive disorder: a retrospective cohort study.
TL;DR
Adolescents with MDD showed lower vitamin D and folate than controls, but nutritional biomarkers did not independently predict symptom severity, and the three-month symptom improvement-occurring without antidepressants-was not explained by biomarker changes and likely reflects multiple factors including regression to the mean.
Abstract
Objectives
Major depressive disorder (MDD) in adolescence is influenced by genetic, psychosocial, and biological factors. Nutritional deficits-particularly vitamin D, vitamin B12, folate, and iron-are linked to mood regulation, though inconsistently. This study investigated associations between nutritional biomarkers and depressive/anxiety symptoms in adolescents with MDD and tracked three-month changes.
Methods
In this retrospective cohort, 125 adolescents with MDD and 125 healthy controls were identified from records between January 2024 and September 2025. MDD was confirmed with K-SADS-PL, and only patients not receiving ongoing antidepressant treatment during the baseline-to-follow-up interval were included. Symptoms (RCADS) and baseline biochemical markers (vitamin D, B12, folate, ferritin, iron indices, thyroid hormones) were obtained for both groups. For the MDD group, anonymized routine-care follow-up records obtained 75-105 days after baseline were also reviewed, whereas controls had baseline records only.
Results
Compared with controls, adolescents with MDD had higher BMI percentile (p = .012) and lower vitamin D (median 15.0 vs. 18.0 ng/mL, p < .001) and folate (6.2 vs. 7.1 ng/mL, p < .001); both survived Bonferroni correction. Ferritin was lower at the unadjusted level (13.6 vs. 17.1 ng/mL, p = .019) but not after correction. Vitamin B12, iron, and hemoglobin did not differ. Over three months, vitamin D, B12, and folate showed modest increases that did not remain significant after correction, whereas RCADS depressive and anxiety scores declined markedly across all domains (all p < .001). In adjusted models, no biomarker independently predicted symptom severity, and biomarker changes were not consistently associated with symptom changes; only Δfolate showed a modest nominal association with improvement in the adjusted change model.
Conclusions
Adolescents with MDD showed lower vitamin D and folate than controls, but nutritional biomarkers did not independently predict symptom severity, and the three-month symptom improvement-occurring without antidepressants-was not explained by biomarker changes and likely reflects multiple factors including regression to the mean. Routine biochemical screening may be clinically useful, but causal claims are not warranted; larger controlled trials are needed.