Children (< 18 years of age) with blood samples at the time of the T1DM diagnosis at Seoul National University Children’s Hospital were included in this study. T1DM was defined as insulin-dependence and having one or more autoimmune markers, including autoantibodies to glutamic acid decarboxylase, islet cells, insulin, or islet antigen-2 (20). Islet antigen-2 autoantibody positivity and titers have been recorded in a limited number of patients since it became measurable in 2018. Among the 298 blood samples obtained from T1DM children between April 2008 and April 2021, 87 were taken at the time of T1DM diagnosis. Finally, 84 T1DM cases were included in this study, after excluding 3 patients for a lack of biochemical information about ketoacidosis at diagnosis ( Supplementary Figure 1 ). We used previous clinical and biochemical data from healthy control children (13, 21), and age- and sex-matched children (n = 84) were included in this study. Case–control matches were performed using the matchControls function in the e1071 package in R software (The R Foundation for Statistical Computing, Vienna, Austria), which computes pairwise dissimilarities between observations using Gower’s coefficients (22).

Clinical information on height, weight, and pubertal status were collected. Body mass index (BMI) was calculated as the body weight divided by the height squared. The BMI and weight-for-height percentiles were assigned based on the 2017 Korean National Growth Chart (23). For children aged > 2 years, overweight and obesity were defined as ≥ 85th and< 95th percentiles of the BMI for age and sex and ≥ 95th percentile of the BMI for age and sex, respectively. For those aged< 2 years old, overweight was defined as ≥ 95th percentile of weight-for-height for age and sex. Seasonality at the time of T1DM diagnosis was classified as spring or summer (March – August) and fall or winter (September – February).

The blood samples were centrifuged at 1500 × g for 10 min within 6 h of collection and stored at −70°C for analysis. 25-Hydroxyvitamin D2 (25OHD₂), 25-hydroxyvitamin D3 (25OHD₃), 24,25-dihydroxyvitamin D3 (24,25OH₂D₃), and the VDBP isoforms were simultaneously quantified via multiplex LC-MS/MS. In brief, samples pretreated with hexane and trypsin were analyzed with an ACQUITY ultra-performance liquid chromatography (UPLC) system (Waters, Milford, MA, USA) using an HSS T3 column (2.1 × 50 mm, 1.8 μm) and Xevo TQ-S mass analyzer (Waters). The coefficient of variation for the three quality control materials was ± 10% and ± 20% at the lower limit of quantification. Additional details can be found elsewhere (19).

Vitamin D sufficiency was defined when total 25OHD (sum of 25OHD₂ and 25OHD₃) was ≥20 ng/mL, and low vitamin D (either vitamin D deficiency or insufficiency) was defined when total 25OHD was<20 ng/mL (24). Bioavailable vitamin D was defined as the sum of free vitamin D and albumin-bound vitamin D (25). Bikle’s formula (26) was used to calculate free and bioavailable vitamin D using the concentrations of serum albumin, VDBP, and total 25OHD, and the binding affinities for albumin and VDBP. Kalb and KVDBP represent the affinity constants between 25OHD and albumin (6 × 10⁵ M^-1), and between 25OHD and VDBP (7 × 10⁸ M^-1), respectively. The vitamin D metabolite ratio (VMR) was calculated as the ratio of 24,25OH₂D₃ to total 25OHD (27). According to the newly established reference of free 25OHD by Zeng et al. (28), free vitamin D sufficiency was defined when free 25OHD was ≥5.67 pg/mL.

Bioavailable 25 OHD

The relative concentrations of free 25OHD to total 25OHD (free to total 25OHD ratio) and the relative concentrations of bioavailable 25OHD to total 25OHD (bioavailable to total 25OHD ratio) were expressed as percentages after matching the units.

Free to total 25 OHD ratio 

Bioavailable to total 25 OHD ratio (%)

Serum concentrations of calcium, phosphorus, albumin, and creatinine were collected. The estimated glomerular filtration rate (eGFR) was calculated using the bedside Schwartz Estimate Equation [eGFR = 0.413 × height (cm)/serum creatinine (mg/dL)] (29). Laboratory results for glycated hemoglobin (HbA1c), pH, bicarbonate (HCO3-), serum C-peptide, and the positivity status and titers of diabetes-associated autoantibodies were collected for the T1DM cases. HbA1c was measured using the National Glycohemoglobin Standardization Program certified method. DKA was defined when the pH of the venous blood was< 7.3 and ketonuria or ketonemia was confirmed when the serum glucose level was > 200 mg/dL. DKA severity was classified as severe (pH< 7.1 or HCO3-< 5 mmol/L), moderate (7.1 ≤ pH< 7.2 or 5 mmol/L ≤ HCO3-< 10 mmol/L), or mild (7.2 ≤ pH< 7.3 or 10 mmol/L ≤ HCO3-< 15 mmol/L).

All statistical analyses were performed using R version 4.0.5 software. All continuous variables are expressed as median [25th–75th percentiles] or mean ± standard deviation. The Shapiro-Wilk test was used to assess normality. Variables with a skewed distribution were natural log-transformed. Student’s t-test or the Mann-Whitney U test was used to compare continuous variables, and the chi-square test or Fisher’s exact test was used to compare categorical variables between the two groups. Comparisons among three groups were conducted with the Kruskal-Wallis test for non-parametric variables, and one-way analysis of variance (ANOVA) or Welch’s ANOVA for parametric variables with homogeneous or heterogeneous variances, respectively. Bonferroni’s method was used for the posthoc analysis. Spearman’s correlation analysis was conducted to evaluate the relationships between vitamin D metabolites and clinical and biochemical parameters. P-values< 0.05 were considered significant.

The clinical and biochemical data of the T1DM cases and healthy controls are described in Table 1 . Age, sex, pubertal status, the proportion of overweight or obesity, and seasonality did not differ between the two groups. Total, free, and bioavailable 25OHD levels and VDBP concentrations did not differ according to normal weight and overweight or obesity status within the T1DM cases and healthy controls, respectively. T1DM patients had significantly lower levels of albumin (3.5 ± 0.5 vs. 4.5 ± 0.4 g/dL, P< 0.001) and eGFR (82.9 ± 20.9 vs. 99.5 ± 15.0 mL/min/1.73 m², P< 0.001) than controls. The total 25OHD levels were significantly lower (16.3 ± 5.1 vs. 19.9 ± 6.5 ng/mL, P< 0.001) with a lower proportion of vitamin D sufficiency (20.2 vs. 46.4%, P = 0.001) in children with T1DM than the controls. Children with T1DM had significantly lower concentrations of VDBP (146.0 ± 27.8 vs. 224.9 ± 36.1 µg/mL, P< 0.001) but higher levels of free 25OHD (8.0 ± 2.5 vs. 6.5 ± 2.3 pg/mL, P< 0.001) with a higher frequency of free vitamin D sufficiency (83.3 vs. 59.5%, P = 0.001) than the controls, although bioavailable 25OHD levels did not differ between the two groups. The free to total 25OHD ratio (0.050 ± 0.009 vs. 0.033 ± 0.005%) and bioavailable to total 25OHD ratio (15.5 ± 2.9 vs. 13.2 ± 1.8%) were significantly higher in the T1DM group than the controls, respectively (P< 0.001 for both). The concentration of 24,25OH₂D₃ was lower in T1DM patients (0.9 ± 0.5 vs. 1.1 ± 0.6 ng/mL, P = 0.006), but the VMR was similar between the two groups ( Table 1 ).

Next, vitamin D metabolites in blood samples taken at the time of the T1DM diagnosis were compared according to whether the T1DM patients presented with DKA [DKA (+) T1DM, n = 41] or not [DKA (–) T1DM, n = 43] at diagnosis. The DKA severity at diagnosis was classified as mild (n = 24, 58.5%), moderate (n = 7, 17.1%), or severe (n = 10, 24.4%) ( Table 1 ). Although the initial HbA1c levels did not differ between the two groups, the serum C-peptide levels were lower in the DKA (+) than in the DKA (–) group (P = 0.027). Although the number of positive autoantibodies did not differ between the two groups, the DKA (+) group had a higher proportion of insulin autoantibody positivity compared to the DKA (–) group (55.0 vs. 27.9%, P = 0.022).

Figure 1 shows a comparison of the vitamin D metabolites among the three groups [healthy controls, DKA (–) T1DM, and DKA (+) T1DM groups]. The DKA (+) group had the lowest levels of total 25OHD compared to the control (15.0 ± 4.6 vs. 19.9 ± 6.5 ng/mL, P< 0.001) and DKA (–) (15.0 ± 4.6 vs. 17.6 ± 5.2 ng/mL, P = 0.020) groups ( Figure 1A ). The serum concentrations of VDBP in the DKA (–) and DKA (+) groups were significantly lower than those in the control group (147.7 ± 26.7 in the DKA (–) and 144.2 ± 29.1 in the DKA (+) vs. 224.9 ± 36.1 µg/mL in the control group, P< 0.001 for both) without a difference between the DKA (+) and DKA (–) groups ( Figure 1B ). The free and bioavailable 25OHD levels were highest in the DKA (–) group among the three groups. The free 25OHD levels increased significantly in the DKA (–) and DKA (+) groups compared to the control group (8.4 ± 2.4 in DKA (–) and 7.5 ± 2.6 in DKA (+) vs. 6.5 ± 2.3 pg/mL in the control group, P< 0.001 and P = 0.047, respectively). Bioavailable 25OHD was lower in the DKA (+) than the DKA (–) group (2.3 ± 0.9 vs. 2.8 ± 0.8 ng/mL, P = 0.014), with no difference between the control and DKA (–) groups ( Figures 1C, D ). The free to total 25OHD ratio and the bioavailable to total 25OHD ratio were significantly higher in the DKA (–) and DKA (+) groups than in the controls (P< 0.001 for both), without differences between the DKA (–) and DKA (+) groups ( Figures 1E, F ). The concentrations of 24,25OH₂D₃ and VMR were significantly lower in the DKA (+) group than in the control and DKA (–) groups (24,25OH₂D₃: 0.7 ± 0.4 in the DKA (+) vs. 1.1 ± 0.6 in the control and 1.1 ± 0.6 ng/mL in the DKA (–) group, P< 0.001 and P = 0.001, respectively) (VMR: 4.6 ± 1.6 in the DKA (+) vs. 5.6 ± 1.5 in the control and 6.0 ± 2.1% in the DKA (–) group, P = 0.002 and P = 0.003, respectively) ( Figures 1G, H and Table 1 ).

Table 2 shows the association between vitamin D metabolites and biochemical parameters reflecting disease severity at the time of the T1DM diagnosis. Total, free, and bioavailable 25OHD levels were positively correlated with pH (P = 0.011 for total 25OHD, P = 0.027 for free 25OHD, and P = 0.006 for bioavailable 25OHD) and HCO3- (P = 0.033 for total 25OHD, P = 0.037 for free 25OHD, and P = 0.031 for bioavailable 25OHD). The serum VDBP concentration was inversely associated with the initial HbA1c value (P = 0.012).

When the VDBP genotypes were analyzed, Gc1f/2 and Gc1s/1f were equally detected in 21.8% (n = 19 each) of the T1DM patients, and Gc1f/1f was observed in 33.3% (n = 29) of the healthy children ( Supplementary Figure 2 ). No significant differences in the proportions of the VDBP genotypes were found between the T1DM and control groups. The presence of DKA and the biochemical parameters reflecting disease severity (initial HbA1c, pH, and HCO3-) at the time of T1DM diagnosis were not significantly different according to the VDBP genotype ( Supplementary Table 1 ).