The research team, led by investigators at the Affiliated Women and Children’s Hospital of Ningbo University and collaborating institutions across Ningbo and Hangzhou, set out to resolve a question that earlier work had left open. Vitamin D exists in two supplemental isoforms: vitamin D3, or cholecalciferol, which is the form used exclusively in China’s national pediatric supplementation program, and vitamin D2, or ergocalciferol, which is negligible in East Asian diets. Prior studies typically measured total 25-hydroxyvitamin D, the combined circulating marker of both isoforms, which effectively averages a potent, abundant molecule with a nearly absent one. Because China delivers D3 monotherapy, the researchers argued that isoform-specific thresholds may offer superior clinical resolution for guiding supplementation and predicting micronutrient status.

To isolate the D3 signal, the team employed liquid chromatography–tandem mass spectrometry, a gold-standard analytical platform that chromatographically separates 25-hydroxyvitamin D3 from 25-hydroxyvitamin D2 before quantification. The technique resolves the two isoforms with a retention-time difference exceeding 0.4 minutes, ensuring that the D3 measurement reflects the metabolically dominant circulating form rather than a diluted combined signal. Serum retinol, the principal circulating form of vitamin A, was measured concurrently on the same instrument. Quality control procedures using certified reference materials yielded intra-assay coefficients of variation of 4.2 percent for D3 and 3.8 percent for retinol, with inter-assay variation of 6.1 and 5.5 percent respectively across 15 analytical runs.

The cohort comprised children who attended routine preventive health surveillance at 22 community health-service centers across Ningbo’s Yinzhou and Zhenhai districts between January 2019 and December 2023. The age range was deliberately restricted to 0 to 10 years to capture pre-pubertal children during peak windows of skeletal growth and dietary transition, while excluding older adolescents in whom hormonal changes and divergent lifestyle patterns might confound the association. Children using pharmacological vitamin D2 or mixed D2/D3 supplements within the preceding four weeks were excluded, as were those with acute infection or inflammation, because retinol falls as an acute-phase reactant during illness. Chronic kidney or liver disease, malabsorption syndromes, hyperparathyroidism, and use of anticonvulsants, glucocorticoids, or rifampin were also grounds for exclusion.

Baseline findings revealed a cohort in generally good nutritional standing, with notable secular improvement compared with earlier cycles. Mean serum 25-hydroxyvitamin D3 was 30.9 ng/mL, and mean retinol was 272.5 ng/mL. Severe D3 deficiency was rare at 0.2 percent of participants, roughly one-seventh the prevalence observed in the team’s previous 2018–2021 analysis of 4,752 Ningbo children, which the authors attribute to heightened post-pandemic parental awareness of micronutrient supplementation alongside sustained national D3 fortification. Nevertheless, the spring nadir persisted: 42.6 percent of children in the combined severe deficiency and deficiency group were sampled in spring, reflecting the seasonal metabolic memory driven by winter ultraviolet B deprivation and D3’s approximately 15-day circulating half-life. An age gradient also endured, with deficient children older, at a median of 1.2 years, than sufficient children, at 0.7 years, underscoring waning supplementation adherence as toddlers transition to family diets.

The central analytical result concerns the shape of the D3–retinol dose–response curve. Restricted cubic splines, a flexible non-linear modeling technique, revealed a significant non-linear, saturable relationship with p-values below 0.001 for both overall association and departure from linearity. Two-piecewise linear regression identified a transition region centered at 28.9 ng/mL. Below this threshold, each 1 ng/mL increment in D3 corresponded to a 3.16 ng/mL rise in retinol, with a 95 percent confidence interval of 2.66 to 3.65. Above the transition, the slope attenuated to 1.80, with a confidence interval of 1.46 to 2.15, but remained statistically positive, indicating that the association persists even at higher D3 concentrations, albeit with diminished marginal returns. Bootstrap resampling with 1,000 iterations yielded a median transition point of 28.7 ng/mL with a confidence interval of 26.4 to 31.2, confirming the robustness of the saturation profile.

Subgroup analyses added seasonal texture that the authors regard as among the most striking findings. After adjustment for age, sex, calendar year, and season, the overall association was 2.11 ng/mL of retinol per 1 ng/mL of D3. The association was strongest in winter, at 2.70, and weakest in summer, at 1.79, with a statistically significant seasonal interaction at p equal to 0.002. The pattern is consistent with ultraviolet B-driven cutaneous D3 synthesis during summer months raising baseline D3 levels independently of dietary intake, thereby compressing the observable range of the D3–retinol association. Children examined during winter, when cutaneous production is scarce, appear to operate in the steep portion of the dose–response curve, where the retinol response to D3 variation is most pronounced.

Tertile-stratified modeling offered additional granularity beyond a simple dichotomy. In the lowest D3 tertile, the adjusted slope was 3.11, indicating a robust positive association in the lower repletion range. In the middle tertile, the estimate attenuated markedly to 1.02, with a confidence interval spanning the null. In the highest tertile, the slope was 1.17, remaining statistically positive but substantially weaker than in the lowest tertile. The interaction between D3 tertile and continuous D3 was highly significant in both crude and adjusted models, formally confirming that the D3–retinol relationship varies non-uniformly across the nutritional spectrum. Rather than a single breakpoint, the data support a continuous gradient of association strength that weakens as D3 repletion improves, a pattern consistent with a saturable nutritional interaction in which the marginal contribution of additional D3 to retinol status diminishes once adequate repletion is approached.

The biological plausibility of this saturable pattern rests on well-characterized molecular machinery. Vitamin D-dependent transcription operates through the vitamin D receptor, which signals exclusively as a heterodimer with the retinoid X receptor, meaning that vitamin D signaling is constrained by retinoid availability and the two pathways may cross-regulate most strongly at low repletion. Recent genomic mapping has identified regulatory regions occupied by both retinoic acid receptor alpha and the vitamin D receptor that participate in the convergence and cooperation of retinoid and vitamin D signaling. In addition, D3 and retinol share intestinal uptake and transport via chylomicrons and lipid transporters including CD36, SR-BI, and NPC1L1, along with hepatic secretion mediated by retinol-binding protein 4, which would be expected to co-vary when intake is low and to saturate as repletion improves.

From a public health perspective, the non-linear D3–retinol association identifies an observational transition point near 29 ng/mL that is relevant to China’s D3-only national supplementation program. Current guidelines often cite total 25-hydroxyvitamin D sufficiency above 20 ng/mL, yet the present data suggest that the association with retinol status is steepest in the upper insufficient to lower sufficient range. The authors emphasize that this isoform-specific estimate should be regarded as hypothesis-generating rather than a validated clinical target, and that whether maintaining D3 above this observational transition point improves retinol status requires confirmation in prospective intervention trials. Several limitations warrant caution: the cross-sectional design precludes causal inference, the cohort was strongly skewed toward infancy with a median age of 0.7 years, the small samples in the youngest and oldest age subgroups rendered their estimates imprecise, and the single coastal region at 29 degrees north latitude with high seafood consumption limits generalizability to higher-latitude or inland populations. Longitudinal trials are now needed to test whether maintaining D3 within the optimal repletion range, particularly during winter, prevents retinol insufficiency and its downstream infectious and allergic morbidities in children.

Subject of Research: Isoform-specific cross-sectional analysis of the non-linear association between serum 25-hydroxyvitamin D3 and retinol in Chinese children aged 0–10 years.

Article Title: Association between serum 25-hydroxyvitamin D₃ and retinol in Chinese children (2019–2023): an isoform-specific cross-sectional analysis

Article References: Bi, Y., Yi, J., Wang, Q., Chen, H., Zheng, J., Wang, S., & Liu, W. (2026). Association between serum 25-hydroxyvitamin D₃ and retinol in Chinese children (2019–2023): an isoform-specific cross-sectional analysis. Frontiers in Nutrition, 13, Article 1985336. https://doi.org/10.3389/fnut.2026.1985336

Image Credits: AI Generated

DOI: 10.3389/fnut.2026.1985336

Keywords: vitamin D3, retinol, children, China, non-linear dose–response, 25-hydroxyvitamin D3, LC–MS/MS, seasonal variation, supplementation, pediatric nutrition, Frontiers in Nutrition, Ningbo