Zhu CY, Fang ZR, Hua LY, Jin H, Xu XY, Liu XL, Zhang YM, Rao ZC. Serum 25-hydroxyvitamin D and peripheral thyroid hormone sensitivity in euthyroid type 2 diabetes: Linear estimates and low-range threshold. World J Diabetes 2026; 17(4): 117489 [DOI: 10.4239/wjd.v17.i4.117489]
Corresponding Author of This Article
Zi-Chen Rao, MD, Department of Endocrinology, The Quzhou Affiliated Hospital of Wenzhou Medical University, Quzhou People’s Hospital, No. 100 Minjiang Dadao, West District, Quzhou 324000, Zhejiang Province, China. rzc1522@wmu.edu.cn
Research Domain of This Article
Endocrinology & Metabolism
Article-Type of This Article
Observational Study
Open-Access Policy of This Article
This article is an open-access article which was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Apr 15, 2026 (publication date) through Apr 14, 2026
Times Cited of This Article
Times Cited (0)
Journal Information of This Article
Publication Name
World Journal of Diabetes
ISSN
1948-9358
Publisher of This Article
Baishideng Publishing Group Inc, 7041 Koll Center Parkway, Suite 160, Pleasanton, CA 94566, USA
Share the Article
Zhu CY, Fang ZR, Hua LY, Jin H, Xu XY, Liu XL, Zhang YM, Rao ZC. Serum 25-hydroxyvitamin D and peripheral thyroid hormone sensitivity in euthyroid type 2 diabetes: Linear estimates and low-range threshold. World J Diabetes 2026; 17(4): 117489 [DOI: 10.4239/wjd.v17.i4.117489]
Chun-Yan Zhu, Zi-Ru Fang, Liang-Yan Hua, Xin-Yi Xu, Xiang-Lan Liu, Yi-Ming Zhang, Zi-Chen Rao, Department of Endocrinology, The Quzhou Affiliated Hospital of Wenzhou Medical University, Quzhou People’s Hospital, Quzhou 324000, Zhejiang Province, China
Hui Jin, Department of Clinical Nutrition, The Quzhou Affiliated Hospital of Wenzhou Medical University, Quzhou People’s Hospital, Quzhou 324000, Zhejiang Province, China
Co-corresponding authors: Yi-Ming Zhang and Zi-Chen Rao.
Author contributions: Zhu CY, Fang ZR, Zhang YM and Rao ZC conceptualized the study; Zhang YM, Rao ZC and Zhu CY developed the methodology; Fang ZR performed the software analysis (EmpowerStats); Zhang YM and Rao ZC conducted the validation; Zhu CY and Fang ZR performed the formal analysis; Xu XY, Hua LY, Jin H, Zhu CY and Liu XL carried out the investigation; Hua LY, Zhu CY and Liu XL provided resources; Xu XY, Hua LY, Jin H, Zhu CY and Liu XL curated the data; Zhu CY prepared the visualization; Zhu CY and Fang ZR drafted the manuscript; Zhang YM, Rao ZC, Zhu CY and Fang ZR reviewed and edited the manuscript; Zhang YM and Rao ZC supervised the study; Rao ZC and Hua LY managed project administration; Zhu CY and Fang ZR contributed equally and share first authorship; Rao ZC and Zhang YM contributed equally and share corresponding authorship.
Institutional review board statement: This study was approved by the Ethics Committee of Quzhou People’s Hospital, Wenzhou Medical University (Approval No. 2022-110).
Informed consent statement: Written informed consent was waived by the Ethics Committee because this retrospective study used de-identified data and involved minimal risk.
Conflict-of-interest statement: The authors declare that they have no conflict of interest.
STROBE statement: The authors have read the STROBE Statement—checklist of items, and the manuscript was prepared and revised according to the STROBE Statement—checklist of items.
Data sharing statement: De-identified data are available from the corresponding author upon reasonable request and subject to institutional and privacy restrictions.
Corresponding author: Zi-Chen Rao, MD, Department of Endocrinology, The Quzhou Affiliated Hospital of Wenzhou Medical University, Quzhou People’s Hospital, No. 100 Minjiang Dadao, West District, Quzhou 324000, Zhejiang Province, China. rzc1522@wmu.edu.cn
Received: December 9, 2025 Revised: January 10, 2026 Accepted: February 11, 2026 Published online: April 15, 2026 Processing time: 127 Days and 2.8 Hours
Abstract
BACKGROUND
Vitamin D insufficiency is common in type 2 diabetes (T2D) and may relate to peripheral thyroid hormone sensitivity. We hypothesized that higher serum 25-hydroxyvitamin D [25(OH)D] is associated with a higher free triiodothyronine-to-free thyroxine ratio in euthyroid adults, with a nonlinear pattern.
AIM
To evaluate the association between serum 25(OH)D and the free triiodothyronine-to-free thyroxine ratio in euthyroid adults with T2D.
METHODS
We conducted a cross-sectional study in a real-world cohort from a metabolic management center in China. Euthyroid adults with T2D were enrolled (n = 1408). Free triiodothyronine, free thyroxine, thyroid-stimulating hormone, and 25(OH)D were measured at the same visit. Analyses of 25(OH)D were restricted to participants with available measurements (n = 1271). Multivariable linear regression modeled the ratio scaled by 100. Restricted cubic splines and two-piecewise models assessed nonlinearity and estimated a threshold. Prespecified subgroups included age, sex, body mass index, current smoking, and current drinking.
RESULTS
Higher 25(OH)D was associated with a higher scaled ratio in the fully adjusted model (β = 0.088 per 1 ng/mL; 95%CI: 0.037-0.139; P < 0.001), equivalent to +0.88 units per 10 ng/mL. Spline and segmented analyses indicated a low-range threshold at 14.7 ng/mL (95%CI: 12.85-16.11). The slope was steeper below the threshold and not significant above it. Estimates were numerically larger in participants aged ≥ 60 years, but age interaction was not significant (P for interaction = 0.113). A drinking interaction was observed: The association was positive in non-drinkers but null in current drinkers (P for interaction = 0.0115). Interaction testing was exploratory and not adjusted for multiple comparisons.
CONCLUSION
In euthyroid adults with T2D, higher 25(OH)D is associated with a higher free triiodothyronine-to-free thyroxine ratio, mainly at low 25(OH)D concentrations.
Core Tip: In a real-world cohort of euthyroid adults with type 2 diabetes, we examined the association between serum 25-hydroxyvitamin D and the free triiodothyronine-to-free thyroxine (FT3/FT4) ratio, a pragmatic circulating proxy of peripheral thyroid hormone sensitivity. Using multivariable models with restricted cubic splines and segmented regression, we identified a low-range threshold around 14.7 ng/mL: The positive gradient was steeper below this level and was attenuated above it. The overall pattern was broadly consistent across prespecified subgroups, with attenuation among current drinkers. These findings are associative and require longitudinal or interventional confirmation.
Citation: Zhu CY, Fang ZR, Hua LY, Jin H, Xu XY, Liu XL, Zhang YM, Rao ZC. Serum 25-hydroxyvitamin D and peripheral thyroid hormone sensitivity in euthyroid type 2 diabetes: Linear estimates and low-range threshold. World J Diabetes 2026; 17(4): 117489
Thyroid hormone action in peripheral tissues reflects not only circulating levels but also local activation and cellular responsiveness[1]. The FT3/FT4 ratio has been widely used as a pragmatic circulating proxy of peripheral T4-to-T3 conversion within the euthyroid range, capturing variation in deiodination that isolated hormone values may miss[2]. However, it is an indirect surrogate and can be influenced by non-thyroidal conditions and measurement factors; therefore, it should not be interpreted as direct evidence of tissue-level thyroid hormone action. Across euthyroid populations, variation in this ratio has been associated with adiposity and insulin resistance, hepatic steatosis and dyslipidemia, low-grade inflammation and blood pressure, and broader cardiorenal phenotypes—positioning FT3/FT4 as a clinically relevant lens on metabolic heterogeneity, particularly in type 2 diabetes (T2D)[3,4].
Vitamin D, also known as 25-hydroxyvitamin D [25(OH)D], insufficiency is frequent in T2D and aligns with pathways central to metabolic health, including immune-inflammatory signaling, adipocyte biology, hepatic lipid handling, and insulin sensitivity[5,6]. These processes intersect with determinants of peripheral thyroid hormone sensitivity, such as deiodinase (DIO1/DIO2)-mediated T4-to-T3 conversion, transport, and tissue responsiveness, providing a biologically plausible bridge between 25(OH)D status and the FT3/FT4 balance[7,8]. Consistent with this biology, observational studies have reported inverse associations between vitamin D status and thyroid-stimulating hormone (TSH), and positive associations with FT3 or the FT3/FT4 ratio, although null or conflicting findings are also described across specific settings (e.g., obesity, pregnancy, autoimmune thyroiditis, pediatric cohorts)[9,10]. Interventional data are mixed: Some small trials suggest modest shifts in thyroid indices with vitamin D supplementation, while others show no material change[11,12]. Overall, clinical evidence directly relating 25(OH)D to FT3/FT4 in euthyroid adults with T2D remains limited and heterogeneous, underscoring the need for focused analyses in this group[13,14].
Against this background, we designed a study to clarify how 25(OH)D relates to the FT3/FT4 ratio in euthyroid adults with T2D. Our primary aim was to provide a clinically interpretable description of this association in a real-world cohort under rigorous adjustment, and to examine potential heterogeneity across patient characteristics. Given the cross-sectional design, all inferences were framed as associations rather than causation.
MATERIALS AND METHODS
Study design and participants
This cross-sectional study was conducted at the metabolic management center, Department of Endocrinology, Quzhou People’s Hospital, Wenzhou Medical University, Zhejiang Province, China. Consecutive adult patients with T2D mellitus (T2DM) who attended the metabolic management center between December 2022 and June 2025 were assessed for eligibility. The diagnosis of T2DM followed the 2023 American Diabetes Association criteria.
Inclusion criteria: (1) Age ≥ 18 years; (2) Euthyroid status (TSH, FT4, and FT3 within laboratory reference ranges); and (3) Available thyroid function tests.
Exclusion criteria: Recent vitamin D supplementation (within 3 months) and conditions known to markedly alter vitamin D metabolism. Additional exclusions were thyroid disease or thyroid-related therapy, prior thyroid surgery or radioiodine, acute diabetic complications at sampling, severe hepatic or renal dysfunction, pregnancy or lactation, age < 18 years, and missing key data.
Final sample: After applying these criteria, 1408 euthyroid adults with T2DM were included. Analyses involving 25(OH)D were restricted to participants with available 25(OH)D measurements. Enrollment and data capture followed standardized metabolic management center procedures. Ethics approval and informed consent are detailed in the Ethics section.
Measurements
Baseline data: Demographic characteristics (age, sex), medical history, smoking (current vs no) and drinking (current vs no) were obtained at baseline using a standardized case report form.
Anthropometry: Weight and height were measured without shoes and heavy clothing. Body mass index (BMI) was calculated as weight/height2 (kg/m2). Waist circumference was measured at the midpoint between the lowest rib and the iliac crest after normal expiration. Visceral fat area (VFA) and subcutaneous fat area (SFA) were assessed with a calibrated body-composition analyzer in the metabolic management center according to the manufacturer’s protocol.
Blood pressure: Seated blood pressure was recorded after ≥ 5 minutes of rest using an automated sphygmomanometer; the mean of two readings was used for analysis.
Laboratory assays: Fasting venous blood was drawn in the morning according to standardized metabolic management center procedures. The analytic dataset did not capture the exact date (month/season) of blood sampling; therefore, the distribution of sampling month/season could not be reported, and seasonality could not be adjusted for in the models. Participants without an available 25(OH)D measurement were excluded from vitamin D-related analyses (analytic sample n = 1271). Glycated hemoglobin (HbA1c) was measured by high-performance liquid chromatography. Lipid profile [total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C)], liver enzymes [aspartate aminotransferase (AST), alanine aminotransferase (ALT)], serum creatinine, uric acid (UA), and high-sensitivity C-reactive protein (CRP) were determined on automated analyzers under internal and external quality control. Estimated glomerular filtration rate (eGFR) was calculated from serum creatinine using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation with age and sex.
Thyroid function: FT3, FT4, and TSH were measured on the same blood draw by chemiluminescence immunoassay in the hospital laboratory (routine quality control in place). Reference ranges were TSH: 0.35-4.94 μIU/mL, FT4: 9.01-19.05 pmol/L, and FT3: 2.43-6.01 pmol/L. The FT3/FT4 ratio was then calculated; details on outcome scaling are provided in the Statistical Analysis section.
Serum 25(OH)D was quantified in ng/mL using a validated immunoassay (laboratory routine QC; 1 ng/mL = approximately 2.5 nmol/L for unit conversion).
Definitions
Exposure: Serum 25(OH)D, analyzed as a continuous variable; effects additionally expressed per 10 ng/mL higher 25(OH)D for clinical readability.
Outcome: The FT3/FT4 ratio was scaled by a factor of 100 for modeling (FT3/FT4 × 100) because raw effect sizes were numerically small; all primary models used the scaled outcome.
Descriptive quartiles: For baseline characterization (Table 1), participants were grouped by FT3/FT4 quartiles (Q1-Q4, lowest to highest).
Table 1 Baseline characteristics across quartiles of the FT3/FT4 ratio.
Euthyroid status: TSH, FT4, and FT3 within laboratory reference ranges (TSH: 0.35-4.94 μIU/mL, FT4: 9.01-19.05 pmol/L, and FT3: 2.43-6.01 pmol/L).
Subgroups: Age (< 60 years vs ≥ 60 years), sex, BMI (< 28 kg/m2 vs ≥ 28 kg/m2), current smoking (yes/no), and current drinking (yes/no).
Covariates: Prespecified set included age, sex, BMI, systolic blood pressure (SBP), diastolic blood pressure (DBP), HbA1c, TC, HDL-C, LDL-C, TG, eGFR, UA, AST, ALT, CRP, white blood cell count (WBC), VFA, SFA, smoking, and drinking. Variables with skewed distributions were Box-Cox transformed prior to modeling.
eGFR: The eGFR was calculated using the CKD-EPI equation with age and sex.
Statistical analysis
Continuous variables were summarized as mean ± SD if approximately normal or median (interquartile range) if skewed; categorical variables as n (%). Missingness of analysis variables is summarized in Supplementary Table 1. Table 1 describes baseline characteristics across FT3/FT4 quartiles (Q1-Q4). Between-group comparisons used one-way ANOVA or Kruskal-Wallis tests for continuous variables and χ² tests for categorical variables, as appropriate; these comparisons are descriptive.
The primary outcome was FT3/FT4 × 100. The FT3/FT4 ratio was scaled by 100 because raw effect sizes were numerically small; all regression models used the scaled outcome. The exposure was serum 25(OH)D, analyzed continuously, with effects additionally expressed per 10 ng/mL for interpretability. In secondary analyses, 25(OH)D was modeled by quartiles (Q1-Q4; Q1 as reference), and P for trend was tested by entering quartile medians as a continuous term. We additionally analyzed 25(OH)D using clinical categories (< 20 ng/mL, 20-30 ng/mL, and ≥ 30 ng/mL; < 20 ng/mL as the reference) to improve clinical interpretability (Supplementary Table 2).
We fitted linear regression models with progressive adjustment: Model 1, unadjusted; model 2, adjusted for age and sex; model 3 (fully adjusted), additionally including BMI, SBP, DBP, HbA1c (Box-Cox), TC (Box-Cox), HDL-C (Box-Cox), LDL-C, TG (Box-Cox), eGFR (Box-Cox), UA (Box-Cox), AST (Box-Cox), ALT (Box-Cox), CRP (Box-Cox), WBC (Box-Cox), VFA (Box-Cox), SFA (Box-Cox), current smoking, and current drinking. Information on the month/season of blood sampling was unavailable; thus, seasonality of serum 25(OH)D could not be additionally adjusted for in the regression models. As a robustness check, we performed supplementary models using alternative covariate sets and summarised the assumed confounding structure in a DAG (Supplementary Table 3 and Supplementary Figure 1).
Prespecified subgroup analyses were conducted for age (< 60 years vs ≥ 60 years), sex, BMI (< 28 kg/m2 vs ≥ 28 kg/m2), current smoking, and current drinking. P-interaction values were obtained by adding a cross-product term [25(OH)D × subgroup] to the fully adjusted model; all interaction tests were considered exploratory and were not adjusted for multiple comparisons.
Non-linearity was assessed using restricted cubic splines in fully adjusted models. When non-linearity was suggested, a two-piecewise linear (segmented) model was fitted to estimate the inflection point (knot) and the slopes below and above the threshold; likelihood-ratio tests compared segmented vs single-line fit. The breakpoint was identified using an iterative grid-search procedure that evaluated candidate cut-points across the observed 25(OH)D distribution (restricted to central quantiles) and selected the value that maximized the model log-likelihood (profile likelihood). A 95%CI for K was obtained by bootstrap resampling, as implemented in EmpowerStats. To assess robustness, we compared the segmented model with the restricted cubic spline curve in the fully adjusted model; the estimated K aligned with the point where the spline began to plateau.
Variables with skewed distributions were Box-Cox transformed prior to modeling. Characteristics of complete-case vs non-complete-case participants were compared as a missing-data diagnostic (Supplementary Table 4). Model assumptions (linearity, homoscedasticity, and normality of residuals) were assessed using standard regression diagnostic plots (residuals vs fitted, normal Q-Q, scale-location, and residuals vs leverage with Cook’s distance contours; Supplementary Figure 2), and multicollinearity was evaluated using variance inflation factors (Supplementary Table 5). Analyses used a complete-case approach; therefore, analytic sample sizes varied across models and subgroups due to missing covariates. As a sensitivity analysis, missing covariates were imputed using multiple imputation (m = 30), and estimates were pooled using Rubin’s rules (Supplementary Table 6). Box-Cox transformations were applied in the imputed analyses using parameters derived from the original dataset. The imputation model included the exposure, outcome, and all covariates in the fully adjusted model. Quartiles of 25(OH)D were defined using cut-points from the original dataset and applied consistently across imputations. All tests were two-sided with P < 0.05 considered statistically significant. Analyses were performed using EmpowerStats (www.empowerstats.com) and R software (version 4.2.2), with multiple imputation implemented using the mice package.
RESULTS
Baseline characteristics
We included 1408 euthyroid adults with T2D; 25(OH)D was available for 1271 participants for vitamin D-related analyses. Across FT3/FT4 quartiles, participants were younger and more often male at higher quartiles (Q1 vs Q4: 55.43 ± 13.13 years vs 51.23 ± 12.39 years; P < 0.001; males 56.8%-69.2%, P < 0.001). Adiposity indices increased from Q1 to Q4 (BMI, VFA, SFA; all P ≤ 0.038). CRP decreased (median 2.04-1.44 mg/L, P < 0.001), while AST and ALT were higher (both P < 0.001). The 25(OH)D did not differ across quartiles in crude comparisons among participants with available 25(OH)D data (P = 0.105). All comparisons are descriptive and reflect distributional differences across quartiles.
Primary association of 25(OH)D with FT3/FT4 × 100
Analyses of 25(OH)D were restricted to participants with available 25(OH)D (n = 1271), and complete-case sample sizes varied across models due to missing covariates. Table 2 presents the associations of 25(OH)D with FT3/FT4 × 100. Higher 25(OH)D was positively associated with FT3/FT4 × 100 in the unadjusted model (β = 0.045, 95%CI: 0.011-0.079; P = 0.010) and after adjustment for age and sex (β = 0.058, 95%CI: 0.022-0.093; P = 0.0016). The association remained in the fully adjusted model—accounting for adiposity, lipids, glycemia, renal and hepatic function, inflammation, VFA, and SFA (β = 0.088, 95%CI: 0.037-0.139; P = 0.00081). Model diagnostics did not indicate major violations of linear model assumptions, and multicollinearity was acceptable (Supplementary Figure 2 and Supplementary Table 5). On a clinical scale, this corresponds to approximately +0.88 in FT3/FT4 × 100 per 10 ng/mL higher 25(OH)D. When 25(OH)D was modeled by quartiles, Q3 and Q4 showed higher FT3/FT4 × 100 vs Q1 (β = 1.517, 95%CI: 0.362-2.671, P = 0.010; β = 1.246, 95%CI: 0.074-2.419, P = 0.038), with a linear trend (P = 0.024). When 25(OH)D was categorized using clinical cutoffs (< 20 ng/mL, 20-30 ng/mL, and ≥ 30 ng/mL; < 20 ng/mL as reference), the direction was consistent with the primary analyses (Supplementary Table 3). In multiple-imputation sensitivity analyses (Supplementary Table 6), the association remained positive (β = 0.062, 95%CI: 0.024-0.101; P = 0.002), and the quartile trend was consistent (P = 0.034). Findings were consistent in supplementary models using alternative covariate sets (Supplementary Table 3).
Table 2 Association between 25-hydroxyvitamin D and the FT3/FT4 ratio in multivariable linear models.
Table 3 summarizes fully adjusted subgroup associations between 25(OH)D and FT3/FT4 × 100 in complete-case analyses, with sample sizes varying by subgroup due to missing covariates. The pattern was broadly consistent: Higher 25(OH)D aligned with higher FT3/FT4 × 100 across most strata. The point estimate was larger in older adults (≥ 60 years: β = 0.13, 95%CI: 0.05-0.21; P = 0.0015) than in those < 60 years (β = 0.04, not significant); however, there was no statistical evidence of effect modification by age (P-interaction = 0.113). Associations were similar in women (β = 0.08; P = 0.022) and men (β = 0.12; P = 0.0087), with no sex interaction (P-interaction = 0.534). Findings were also comparable by smoking status (non-smokers β = 0.10; P = 0.0010; current smokers β = 0.05; P = 0.367; P-interaction = 0.389) and by BMI category (< 28 kg/m2 vs ≥ 28 kg/m2; P-interaction = 0.673). A single interaction pattern was observed for drinking status: A positive association in non-drinkers (β = 0.13, P < 0.0001) contrasted with a null association in current drinkers (β = -0.03; P = 0.633), with P-interaction = 0.0115. This interaction analysis was exploratory and was not adjusted for multiple comparisons. Overall, the subgroup results were broadly consistent with a stable positive association, with attenuation among current drinkers.
Table 3 Subgroup analyses of the association between 25-hydroxyvitamin D and the FT3/FT4 ratio, with tests for interaction.
Table 4 and Figure 1 jointly depict a non-linear relationship between 25(OH)D and FT3/FT4 × 100 based on the fully adjusted complete-case model. A threshold at 14.7 ng/mL was identified in two-piecewise linear models, with a bootstrap-derived 95%CI of 12.85-16.11 ng/mL. Below 14.7, 25(OH)D showed a positive slope (β = 0.61, 95%CI: 0.31-0.92; P < 0.0001). At 25(OH)D ≥ 14.7, the slope was not significant (β = 0.04, 95%CI: -0.02 to 0.10; P = 0.205). Model fit improved vs a single-line specification (likelihood-ratio P < 0.001), and the difference between segment slopes was significant (Δβ = -0.57, 95%CI: -0.91 to -0.24; P = 0.0007). The smoothed curve (Figure 1) illustrates a steeper increase at lower 25(OH)D with attenuation toward a plateau at higher levels, aligning with the piecewise results.
Figure 1 Smoothed curve from the fully adjusted linear model with restricted cubic splines.
The shaded band indicates 95%CIs. The vertical line marks the estimated inflection (approximately 14.7 ng/mL). The outcome was modeled as FT3/FT4 × 100 for numerical scaling; inference is unaffected. 25(OH)D: 25-hydroxyvitamin D.
Table 4 Threshold effect of 25-hydroxyvitamin D on the FT3/FT4 ratio from two-piecewise linear regression.
In this study, among euthyroid adults with T2D, higher 25(OH)D was positively associated with a higher FT3/FT4 ratio after multivariable adjustment. Notably, the association was concentrated at low 25(OH)D, with a low-range inflection around 14.7 ng/mL (95%CI: 12.85-16.11) and an apparent plateau thereafter. Moreover, the direction was consistent across complete-case models and prespecified subgroups. This pattern was also supported by multiple-imputation sensitivity analyses. In subgroup analyses, the association appeared weaker among current drinkers. However, interaction tests were exploratory and should be interpreted cautiously given multiple comparisons. Although exploratory, this heterogeneity is biologically plausible and is discussed in the mechanistic section. This may also explain the attenuation in the fully adjusted clinical-category analysis (Supplementary Table 2). The spline and piecewise models suggest that the association is concentrated below the low-range threshold and plateaus thereafter (Table 4 and Figure 1). Because the FT3/FT4 ratio is an indirect and non-specific proxy, our findings should not be interpreted as evidence of tissue-level thyroid hormone action.
Echoing our observations, multiple studies in euthyroid populations have likewise reported a positive association between 25(OH)D and the FT3/FT4 ratio. Community-based analyses, after accounting for age, overall and regional adiposity, and glycemic status, generally found higher FT3/FT4 among participants with higher 25(OH)D. Several reports using spline or segmented models further suggested that the signal concentrates at lower 25(OH)D levels and then tapers toward a plateau[15]. In hospital cohorts enriched for metabolic risk (e.g., T2D or insulin resistance), multivariable models that included adiposity measures, lipids, hepatic and renal indices, inflammatory markers, and lifestyle factors still showed a positive association between 25(OH)D and FT3/FT4[16]. Analyses in euthyroid overweight/obese adults similarly observed a stepwise increase in FT3/FT4 across higher 25(OH)D strata, consistent with the view that vitamin D status may track peripheral thyroid hormone sensitivity[17].
By contrast, evidence in certain settings has been null or mixed for the 25(OH)D-FT3/FT4 link. In obesity-focused cohorts, associations often became non-significant after stringent control for adiposity and inflammation[18,19]. Pregnancy studies, influenced by trimester-specific physiology and reference intervals, reported inconsistent trends when strata were pooled or differently defined[20,21]. Samples with autoimmune thyroiditis and pediatric cohorts also showed unstable directions or non-significant results, likely reflecting greater biological heterogeneity[22,23]. Overall, differences across studies may reflect variation in population structure, exposure distribution, and assay or adjustment strategies. In addition, failing to model nonlinearity may obscure associations at lower 25(OH)D levels[24,25].
Vitamin D may be related to peripheral thyroid hormone sensitivity through converging anti-inflammatory, hepatic, and adipose-muscle pathways[26]. At the molecular level, vitamin D receptor (VDR)-retinoid X receptor signaling inhibits nuclear factor kappa-β activity and modulates cytokine balance[27,28]. This environment may help maintain DIO1/DIO2 activity while suppressing DIO3. Such changes could align with a higher FT3/FT4 ratio within the euthyroid range[29,30]. In the liver, a major site of DIO1 expression, vitamin D may alleviate steatosis and dampen inflammatory signaling[31,32]. These effects may be linked to changes in deiodinase expression and thyroid hormone clearance[33]. Once sufficient vitamin D levels are achieved, VDR-dependent transcription likely reaches saturation, resulting in a plateau effect at higher 25(OH)D levels[34,35]. In adipose tissue and skeletal muscle, vitamin D may influence adipocyte differentiation and adipokine secretion (leptin and adiponectin). It may also affect mitochondrial function and local cytokine signaling[36,37]. These changes may influence tissue deiodinases and thyroid hormone receptors (TRα and TRβ). This could modify local T3 action without altering TSH, consistent with FT3/FT4 as a marker of peripheral thyroid sensitivity[38,39]. Context can blunt this alignment. One potential explanation is reduced hepatic DIO1 activity in drinkers, which may dampen peripheral T4-to-T3 conversion efficiency. Alcohol use may also alter vitamin D metabolism through hepatic enzyme induction (e.g., CYP24A1 or CYP3A4) and is often accompanied by suboptimal nutritional intake, which could weaken the observable 25(OH)D-FT3/FT4 association. Alcohol exposure may induce hepatic inflammation and steatosis, increase oxidative stress, and disrupt selenium availability for deiodinase selenoproteins[40,41]. It also affects hormone transport and binding, as well as drug-nutrient interactions. Together, these factors may attenuate the observed 25(OH)D-FT3/FT4 association, as seen among current drinkers[42]. These mechanisms provide a coherent biological context for our findings: A stronger association at low 25(OH)D levels, a plateau beyond sufficiency, and attenuation in adverse metabolic contexts[43].
From a nutrition standpoint, these findings refine how to read low 25(OH)D in euthyroid adults with T2D when routine thyroid tests are normal[44,45]. The concentration of the association at lower 25(OH)D suggests that the insufficiency range is where variation in peripheral thyroid hormone sensitivity is most evident. Beyond the low-range threshold, additional increases may add little within the euthyroid window[46]. In practice, this suggests attention to deficiency screening and to modifiable determinants of vitamin D status—dietary intake and safe sunlight exposure in particular[47,48]. Moreover, the attenuation in current drinkers indicates that lifestyle context can blunt the observable gradient and should be considered during counseling. These observations may inform nutrition-focused interpretation and patient education; they do not imply changes to thyroid therapy[49].
Our dose-response results may also have clinical relevance. The association was concentrated at low 25(OH)D levels and then appeared to plateau, suggesting that patients with T2D and marked vitamin D deficiency may be the subgroup in whom vitamin D status is most informative with respect to variation in peripheral thyroid hormone indices. In this context, vitamin D repletion could plausibly be linked to improvements in thyroid hormone metabolism and energy homeostasis, but our cross-sectional design cannot establish causality or treatment effects. Prospective studies and randomized trials focusing on severely deficient patients are needed to test this hypothesis.
Several features support the robustness of our findings. These include restriction to euthyroid adults, consecutive enrollment in a metabolic management center setting, a large sample, and exclusion of recent vitamin D supplementation. We also summarized missingness (Supplementary Table 1), compared complete-case vs non-complete-case participants (Supplementary Table 4), and conducted multiple-imputation sensitivity analyses (Supplementary Table 6). As a robustness check, we evaluated alternative covariate sets guided by a prespecified DAG, including adiposity-extended and inflammation-extended models. The association remained directionally consistent (Supplementary Table 3 and Supplementary Figure 1). Model diagnostics suggested no major departures from linear-model assumptions, and multicollinearity was acceptable (Supplementary Figure 2 and Supplementary Table 5). Because exposure, covariates, and outcome were measured at the same visit, we did not perform mediation analyses. Temporal ordering cannot be established, and causal interpretation would be inappropriate. However, the cross-sectional design precludes temporality, and residual confounding remains possible. Fasting blood samples were collected in the morning under standardized metabolic management center procedures. However, the analytic dataset lacked the exact sampling date (month/season), so we could not report month/season distributions or adjust for 25(OH)D seasonality. Because model 3 relied on complete-case data, selection bias is possible; baseline characteristics differed between complete-case and non-complete-case participants (Supplementary Table 4). Multiple imputation-pooled estimates were directionally consistent with complete-case results (Supplementary Table 6). However, multiple imputation assumes missing at random and may not fully remove bias if this assumption is violated. In addition, detailed dietary intake and sunlight exposure were not directly measured. Information on concurrent medication use (e.g., lipid-lowering and antidiabetic agents) was not available. Therefore, we could not adjust for medication use or conduct medication-stratified analyses. We also lacked data on alcohol dose/intensity and biomarkers or imaging related to hepatic function or enzyme activity; therefore, the proposed mechanisms underlying the weaker association in current drinkers cannot be directly tested in our dataset. This limitation may contribute to residual confounding, alongside the lack of blood sampling date (month/season) information. Looking ahead, longitudinal cohorts and targeted trials in low-25(OH)D T2D populations should test whether changes in 25(OH)D track FT3/FT4. Dose-response shape and contextual modifiers such as alcohol should be prespecified. Mechanistic work integrating inflammation, hepatic measures or imaging, deiodinase activity, and selenium status may help identify responsive subgroups.
CONCLUSION
In euthyroid adults with T2D, higher 25(OH)D was positively associated with a higher FT3/FT4 ratio after multivariable adjustment. The association was concentrated at low 25(OH)D, with an inflection at 14.7 ng/mL and a plateau thereafter, and appeared attenuated in current drinkers in exploratory subgroup analyses. These results are associative; confirmation in longitudinal cohorts and targeted trials—especially at low 25(OH)D—is warranted.
ACKNOWLEDGEMENTS
We thank Zheng-Ping Zhu for his valuable support and constructive input during the preparation of this manuscript.
Wang XC, Huang YY, Qu S. 625-P: Association between Vitamin D Status and Thyroid Function in Individuals with Type 2 Diabetes Mellitus.Diabetes. 2024;73:625-P.
[PubMed] [DOI] [Full Text]
Waterhouse M, Pham H, Rahman ST, Baxter C, Duarte Romero B, Armstrong BK, Ebeling PR, English DR, Hartel G, van der Pols JC, Venn AJ, Webb PM, Whiteman DC, McLeod DSA, Neale RE. The Effect of Vitamin D Supplementation on Hypothyroidism in the Randomized Controlled D-Health Trial.Thyroid. 2023;33:1302-1310.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 2][Cited by in RCA: 8][Article Influence: 2.7][Reference Citation Analysis (0)]
Islam N, Majumder J, Yesmin F, Habib A, Sayeeda S, Parvin T, Akhter N. Assessment of trimester-specific reference ranges for thyroid hormones in normal pregnant women.Int J Clin Obstet Gynaecol. 2025;9:145-150.
[PubMed] [DOI] [Full Text]
Schleicher RL, Sternberg MR, Looker AC, Yetley EA, Lacher DA, Sempos CT, Taylor CL, Durazo-Arvizu RA, Maw KL, Chaudhary-Webb M, Johnson CL, Pfeiffer CM. National Estimates of Serum Total 25-Hydroxyvitamin D and Metabolite Concentrations Measured by Liquid Chromatography-Tandem Mass Spectrometry in the US Population during 2007-2010.J Nutr. 2016;146:1051-1061.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 153][Cited by in RCA: 170][Article Influence: 17.0][Reference Citation Analysis (0)]
Ma J, Cao H, Rodrigues RM, Xu M, Ren T, He Y, Hwang S, Feng D, Ren R, Yang P, Liangpunsakul S, Sun J, Gao B. Chronic-plus-binge alcohol intake induces production of proinflammatory mtDNA-enriched extracellular vesicles and steatohepatitis via ASK1/p38MAPKα-dependent mechanisms.JCI Insight. 2020;5:e136496.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 20][Cited by in RCA: 55][Article Influence: 9.2][Reference Citation Analysis (0)]
Jayedi A, Daneshvar M, Jibril AT, Sluyter JD, Waterhouse M, Romero BD, Neale RE, Manson JE, Shab-Bidar S. Serum 25(OH)D Concentration, Vitamin D Supplementation, and Risk of Cardiovascular Disease and Mortality in Patients with Type 2 Diabetes or Prediabetes: a Systematic Review and Dose-Response Meta-Analysis.Am J Clin Nutr. 2023;118:697-707.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 15][Cited by in RCA: 30][Article Influence: 10.0][Reference Citation Analysis (0)]
Scientific quality: Grade A, Grade B, Grade B, Grade C
Novelty: Grade B, Grade C
Creativity or innovation: Grade B, Grade C
Scientific significance: Grade C, Grade C
P-Reviewer: Dabla PK, MD, Professor, India; Papazafiropoulou A, MD, PhD, Greece; Song YF, MD, PhD, China; You W, PhD, Chief Nurse, China S-Editor: Lin C L-Editor: A P-Editor: Xu ZH
