Data from the MMC cohort at Taizhou Central Hospital (Taizhou University Hospital) and Yuhuan Second People's Hospital from September 2017 to February 2025 were used in this cross-sectional analysis. Initially, 14,345 people with T2DM were screened. Lacking thyroid autoantibody or BUN values or a history of end-stage renal disease were the two exclusion criteria. Participants were excluded if they met any of the following criteria: (1) missing data for BUN or thyroid autoantibodies (TG-Ab and/or TPO-Ab), which were essential for exposure assessment and covariate adjustment; (2) a documented history of end-stage renal disease (ESRD), given that severe renal impairment independently alters BUN metabolism and could confound the studied association. In the end, 2,054 individuals were included in the study (the flow chart is shown in Figure 1). The Clinical Research Ethics Committees of both participating hospitals accepted the study procedure. Prior to recruitment, all participants provided written informed permission, and the study was carried out in compliance with the Declaration of Helsinki's guiding principles.

Each person underwent a thorough clinical assessment that included behavioral aspects, biological indicators, and sociodemographic traits. Age, sex, education level, duration of diabetes, history of hypertension and hyperlipidemia, smoking and drinking status, height, weight, systolic blood pressure (SBP), diastolic blood pressure (DBP), fasting blood glucose (FBG), fasting C-peptide (FCp), glycated hemoglobin (HbA1c), total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), uric acid (UA), serum creatinine (Scr), BUN, free triiodothyronine (FT3), free thyroxine (FT4), thyroid-stimulating hormone (TSH), anti-thyroglobulin antibody (TG-Ab), and anti-thyroid peroxidase antibody (TPO-Ab). Blood samples were collected after an overnight fast of at least 8 h and processed using standardized laboratory protocols. FBG was measured using the hexokinase method; FCp was quantified by chemiluminescent microparticle immunoassay (CMIA); and HbA1c was analyzed by high-performance liquid chromatography (HPLC) calibrated to National Glycohemoglobin Standardization Program (NGSP) standards. For renal function assessment, BUN was determined using the enzymatic continuous monitoring method (urease-glutamate dehydrogenase), Scr was measured by the sarcosine oxidase method, and UA was assayed using the uricase-peroxidase (Uricase-PAP) method. Lipid profiles were assessed as follows: TG by the glycerol kinase-glycerol phosphate oxidase-peroxidase method (GK-GPO-PAP), TC by the cholesterol oxidase-peroxidase method (CHOD-PAP), and both HDL-C and LDL-C by direct homogeneous endpoint methods. Thyroid function parameters were measured using distinct immunoassay methodologies: FT3 and FT4 by competitive methods, TSH by a two-site sandwich method, and both TG-Ab and TPO-Ab by competitive chemiluminescent microparticle immunoassays. The body mass index (BMI) was computed by weight (kg)/height (m²). Based on Scr levels, the Chronic Kidney Disease Epidemiology Collaboration equation was used to determine the estimated glomerular filtration rate (eGFR) (12).

Thyroid autoantibodies in serum were used to define HT (13). Antibody positivity was determined as TPOAb > 34 IU/ml and/or TgAb > 115 IU/ml based on the manufacturer's recommended limit. The HT-positive group consisted of individuals who satisfied the positivity criteria for either antibody.

While continuous variables were shown as mean (standard deviation, SD) or median (interquartile range, IQR), depending on their distribution, categorical variables were reported as proportions (%). One-way ANOVA for normally distributed data, Kruskal-Wallis tests for skewed data, or chi-square testing for categorical data were used to evaluate group differences. Multiple imputation was used to manage missing data; Supplementary Table 1 contains information on covariate missingness. The relationship between BUN and HT was assessed using logistic regression models, which produced odds ratios (OR) with 95% confidence intervals (95% CI). BUN was classified into tertiles and analyzed as a continuous variable. A crude model with no adjustments was created; Model 1 was adjusted for age and sex; Model 2 was further adjusted for education level, diabetes duration, body mass index, and HbA1c; and Model 3 was fully adjusted for age, sex, education level, diabetes duration, BMI, HbA1c, smoking, drinking, hypertension, and hyperlipidemia. Based on Model 3, restricted cubic spline (RCS) regression was used to evaluate the linearity of the BUN–HT association. Likelihood-ratio tests and bootstrap resampling were performed to discover possible inflection points in a two-piecewise logistic regression model with smoothing if non-linearity was apparent. Using multivariate logistic regression, potential impact modification was investigated among subgroups defined by sex, age (< 60 vs. ≥60 years), BMI (< 24 vs. ≥24 kg/m²), and study center. Likelihood-ratio tests were used to examine interactions between subgroups and BUN. Only full instances were used in sensitivity studies to assess any potential bias caused by imputation. R 4.2.2 R Statistical Software (Version 4.2.2, http://www.R-project.org, The R Foundation) and Free Statistics software (Version 2.3.1, Beijing, China, http://www.clinicalscientists.cn/freestatistics) were used for all analyses, and the threshold for statistical significance was set at a two-sided P < 0.05.

The final analysis of this cross-sectional study includes 2,054 individuals. Table 1 displays the study population's baseline characteristics. 414 people, or 20.2% of the study population, were diagnosed with HT out of the entire cohort. BUN tertiles were used to stratify the participants into three groups: T1 (BUN ≤ 4.7 mmol/L), T2 (BUN > 4.7 to ≤ 6.0 mmol/L), and T3 (BUN > 6.0 mmol/L). Several baseline factors showed statistically significant differences (P < 0.05) in inter-group comparisons. Participants showed a higher percentage of males, older age, longer duration of diabetes, and a higher prevalence of hypertension (P < 0.001) as BUN tertiles increased. Renal function indicators also changed at the same time: eGFR dropped and Scr levels rose dramatically (P < 0.001). Additionally, there were differences in HT prevalence between the tertiles (T1: 15.2%; T2: 23.3%; T3: 21.9%, P < 0.001). There were no statistically significant differences between the BUN groups in other metabolic and thyroid function indicators, including FBG, lipid profiles, FT3, and TSH.

Supplementary Table 2 displays the findings of the univariate logistic regression study. Elevated BUN was substantially linked to a higher incidence of HT among all the factors examined (OR = 1.06, 95% CI: 1.01–1.12, P = 0.020). BUN's relationship with HT became clearer when it was examined as a categorical variable (stratified by tertiles). The middle and highest BUN tertiles were linked to a considerably higher prevalence of HT (P < 0.05) when the lowest BUN tertile was used as the reference. A higher incidence of HT was linked to lower educational level, DBP, and eGFR (P < 0.05). In this univariate study, additional factors such as age, sex, blood pressure, diabetes duration, lipid profiles, and other thyroid function indicators did not reach statistical significance.

Multivariate logistic regression analysis assessed the independent relationship between BUN and HT following stepwise correction for possible confounders, as indicated in Table 2. Both when BUN was examined as a continuous variable and as a categorical variable (tertiles), the positive correlation between BUN and HT persisted across all correction models. When analyzed as a continuous variable, each 1 mmol/L increase in BUN was consistently associated with approximately a 7% increase in prevalence of HT after adjusting for age and sex (Model 1), with further adjustments for education level, diabetes duration, BMI, and HbA1c (Model 2, OR = 1.06, 95% CI: 1.01–1.13, P = 0.030), and with full adjustment for smoking, drinking, hypertension, and hyperlipidemia (Model 3, OR = 1.07, 95% CI: 1.01–1.13, P = 0.031). In Model 3, both the middle (T2) and highest (T3) BUN tertiles were significantly associated with higher odds of HT, corresponding to ORs of 1.67 (95% CI: 1.26–2.22, P < 0.001) and 1.56 (95% CI: 1.16–2.11, P = 0.003), respectively, when compared with the lowest tertile (T1).

The prevalence of HT and BUN levels had a strong nonlinear association, according to RCS analysis (Figure 2, P for nonlinearity = 0.022). The nonlinear association between BUN and HT was validated by threshold effect analysis (Table 3). 5.299 mmol/L was found to be a major tipping point. A dose-response association was evident on the left side of the inflection point (BUN ≤ 5.299 mmol/L), where a 1 mmol/L rise in BUN was linked to a 52.1% higher prevalence of HT (OR = 1.521, 95% CI: 1.196–1.934, P < 0.001). However, there was no statistically significant correlation between BUN and HT prevalence on the right side of the inflection point (BUN > 5.299 mmol/L; OR = 0.977, 95% CI: 0.868–1.099, P = 0.700).

Subgroup studies were carried out to evaluate the generalizability of the relationship between BUN and HT. The results (Figure 3) consistently demonstrated a positive correlation between elevated BUN and increased HT prevalence across all categories after stratification by sex, age, BMI, and research center. In particular, higher BUN levels were linked to an increased prevalence of HT among each subgroup, with effects trending in the same direction. The strength of the link between BUN and HT was not substantially altered by sex, age, BMI, or research center, according to interaction analyses (all interaction P-values > 0.05).

To confirm that our primary findings were robust, sensitivity analyses were carried out. Both analyses limited to full cases and studies utilizing datasets with multiple imputation for missing data showed the same results (Supplementary Table 3).