Occupational heavy metal exposure and endocrine–metabolic biomarkers: a retrospective analysis of 34,595 screened individuals in Türkiye Authors: Arzu Or Koca, Mustafa Cesur, Erman Çakal, Tuğba Akkale, İlhan Yetkin Published: 2026 ABSTRACT BackgroundHeavy metals and trace elements can act as endocrine-disrupting chemicals and have been linked to thyroid, glucose, lipid, and hepatic abnormalities. We aimed to explore associations between routinely measured metal concentrations and biochemical/endocrinological parameters in a large occupational screening dataset from Türkiye.MethodsRecords from 37,763 individuals evaluated at a national occupational and environmental diseases hospital between November 2021 and October 2023 were retrospectively reviewed; after prespecified exclusions, 34,595 unique participants were included. Metal measurements were performed in whole blood, serum, and urine samples as requested in routine surveillance; for each analyte, results were categorized as within or above the interpretive limit by the laboratory information system. Primary comparisons focused on arsenic, cadmium, lead, manganese, and zinc because of occupational relevance and sample size. Routine biochemistry, lipid profile, and thyroid function tests were extracted when available, and estimated glomerular filtration rate (eGFR, CKD‑EPI) was calculated.ResultsAmong participants with available thyroid function tests (n = 9,956), thyroid dysfunction was observed in 14.3% (subclinical hypothyroidism 11.9%). Among participants with available fasting glucose and/or HbA1c data (n = 20,738), diabetes mellitus and prediabetes criteria were met in 12.8% and 10.0%, respectively. In univariable comparisons, several metals above the interpretive limit were associated with differences in metabolic biomarkers, including lower HDL with elevated cadmium, lower HDL/LDL with elevated manganese, and higher ALT/AST and triglycerides with elevated zinc. Elevated metal strata differed in age and sex distribution. After within-metal false discovery rate correction, only a limited subset of biomarker differences remained significant.ConclusionAmong tested participants, thyroid dysfunction and dysglycemia were observed, and selected metal measurements above interpretive limits were associated with lipid and liver enzyme patterns. Given the clinically driven testing and limited covariate data, these findings should be interpreted as exploratory surveillance signals rather than workforce-level prevalence estimates. Heavy metal exposure may exert endocrine-disrupting effects on workers in relevant industries, particularly on thyroid function and glycemic regulation, suggesting that periodic surveillance of these parameters is warranted in exposure-prone workforces.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12902-026-02388-7. Background There has been increased awareness and understanding of endocrine-disrupting chemicals (EDCs) in recent years, which comprise a heterogeneous group of compounds with complex metabolic effects [1]. Exposure to EDCs can induce various adverse effects, including immune and endocrine system disruption, respiratory problems, metabolic problems, type 2 diabetes mellitus (DM), obesity, cardiovascular disease, neurological disorders, and cancer [2]. EDCs are classified into many subgroups, including heavy metals, parabens, and phytoestrogens. Heavy metals constitute a significant portion of EDCs. It is well established that EDCs, including heavy metals, mainly affect steroid and thyroid hormones. The endocrine-disrupting effects of heavy metals may occur through direct action on target glands or indirectly along the endocrine axis, resulting in either hypo- or hyperfunction, and may also affect endocrine–nonendocrine tissue interactions and cognition-related pathways [3, 4]. Recently, a study suggested exposure to EDCs as an additional and independent risk factor for metabolic disorders. Several previous studies have investigated arsenic, cadmium, and other heavy metals, providing increasing evidence that these metals are associated with metabolic dysfunction. Arsenic is a heavy metal associated with type 2 DM and hypertension and may cause low high-density lipoprotein (HDL) and high low-density lipoprotein (LDL) levels [5–8]. Nevertheless, these effects have not yet been fully understood in epidemiological studies so far. Similarly, a previous study revealed that there may or may not exist an association between mercury and metabolic syndrome parameters [9]. According to an earlier study, individuals in Europe were exposed to 10–20 µg of cadmium daily, which was associated with various metabolic disorders [10]. Metal ions accumulate in thyroid tissue and may disrupt thyroid hormone homeostasis through multiple mechanisms [3, 11]. These include reductions in free thyroid hormone levels with cadmium exposure, impaired deiodination with lead exposure, altered pituitary TSH secretion, and triggering of thyroid autoimmunity, particularly with aluminum [3, 12, 13]. Exposure to endocrine-disrupting chemicals (EDCs) begins during pregnancy and persists throughout the human lifespan [14]. Groups at high risk of exposure to EDCs include workers involved in certain occupations and/or individuals with low socioeconomic status [2]. Certain occupational groups in Türkiye are at high risk of heavy metal exposure; therefore, blood and/or urine heavy metal levels of individuals in these occupations are periodically measured. Examples of such occupational groups in the Turkish metal sector include welding, mining, iron and steel production, the metal recycling industry, foundries, accumulator production, and dye and glass production. Similar to many other EDCs, the adverse endocrinological effects of heavy metals become visible upon prolonged exposure above a certain level. By the time the consequences of heavy metal exposure are observed, often due to accumulation in endocrine glands, it is usually too late to implement measures or manage affected individuals. Although individual associations between specific heavy metals and metabolic or thyroid outcomes have been reported, large-scale occupational surveillance datasets combining multiple routinely measured metals with concurrent endocrine and metabolic biomarkers remain limited, particularly from middle-income countries with active industrial workforces. Such surveillance-derived signals may help identify priorities for occupational health monitoring even when causal inference is not feasible. This study aimed to explore the associations between routinely measured heavy metal and trace element concentrations and basic biochemical and endocrinological parameters in individuals undergoing regulatory periodic screening for occupational exposure at Ankara Gazi Mustafa Kemal Occupational and Environmental Diseases Hospital, Türkiye’s national referral center for occupational diseases with an advanced toxicology laboratory. Materials and methods In this retrospective, cross-sectional study, the laboratory records of 37,763 individuals who underwent regulatory periodic examinations at Ankara Gazi Mustafa Kemal Occupational and Environmental Diseases Hospital (Ankara, Türkiye) between November 2021 and October 2023 were reviewed. Metal analyses were requested as part of occupational surveillance. After excluding records with missing metal results (n = 2,999), inconsistent hormonal analysis results (n = 157), and missing age and/or sex data (n = 12), the final analytical sample consisted of 34,595 participants. Only one eligible record per individual was included in the final analytical dataset; therefore, no participant contributed more than one observation to the analyses. Depending on the analyte and the clinical or occupational request, specimens consisted of whole blood, serum, or urine. The specimen matrix recorded in the electronic archive was preserved for each analyte, and comparisons were made within analyte-specific categories rather than by pooling different biological matrices. For each analyte, results were categorized by the laboratory information system as within or above the interpretive limit. The term “interpretive limit” refers to the clinical or occupational interpretation threshold used in routine laboratory reporting for a given analyte and specimen matrix. These thresholds were not selected post hoc by the authors and should not be interpreted as analytical limits of detection or quantification. Rather, they represent reference intervals, upper reference limits, or biological exposure guidance values used to classify results as within or above the expected interpretive range in occupational health surveillance. The interpretive limits, specimen matrices, units, and frequencies of results above the interpretive limits are summarized in Table 1. For essential trace elements, particularly zinc, values above the reference interval were not interpreted as toxic exposure. Such values were described as “above the reference interval” or “above the interpretive limit” and interpreted separately from toxic metals. Because the retrospective electronic archive did not consistently capture the timing of biological sampling in relation to work shift or work week, thresholds that may have timing-dependent occupational interpretations were used as laboratory reporting thresholds rather than as definitive biological exposure index compliance assessments. This limitation was considered when interpreting urinary metal results, particularly spot urine measurements. Urinary metal concentrations were analyzed using the concentration units reported by the laboratory information system. Creatinine-normalized urinary metal concentrations were not used in the primary analyses because paired urine creatinine values corresponding to the same sampling time as the metal measurements were not consistently available in the extracted archive. Therefore, urinary arsenic and cadmium results should be interpreted as spot concentration-based screening values rather than fully normalized exposure biomarkers. Selection of prespecified metals and trace elements Arsenic, cadmium, lead, and manganese were prespecified because they are occupationally relevant metals/metalloids, were sufficiently represented in the surveillance archive, and have established toxicological or endocrine–metabolic relevance in previous literature. Zinc was retained as a separate prespecified trace-element analysis because it was measured within the same occupational surveillance workflow and had a sufficient number of results above the reference value for comparison. However, zinc was not conceptualized as equivalent to toxic heavy metals. Zinc-related findings were interpreted in the context of its role as an essential trace element and the potential influence of nutritional status, supplementation, inflammation, and physiological regulation. Analytical methods and quality assurance Metal and trace-element determinations were performed by the hospital toxicology laboratory as part of routine occupational surveillance, in accordance with the laboratory’s standard operating procedures for specimen acceptance, handling, storage, analysis, calibration, and internal quality control. The retrospective electronic extract retained the analyte name, specimen matrix, unit, numerical result, and laboratory interpretive category. However, it did not consistently retain full analytical method metadata, including instrument platform/model, analytical method, calibration range, limits of detection, limits of quantification, lot-specific quality-control information, or external proficiency-testing identifiers. Therefore, the present analysis was based on laboratory-reported numerical values and interpretive categories rather than on independently reconstructed analytical performance characteristics. This limitation has been explicitly acknowledged, and the findings are interpreted as exploratory occupational surveillance findings rather than definitive exposure–response estimates. The laboratory parameters analyzed in the present study included fasting plasma glucose, fasting insulin, glycated hemoglobin (HbA1c), creatinine, urea, calcium (Ca), albumin (Alb), sodium (Na), potassium (K), alanine aminotransferase (ALT), aspartate aminotransferase (AST), TSH, fT3, fT4, anti-thyroid peroxidase antibody (anti-TPO), anti-thyroglobulin antibody (anti-TG), TSH receptor antibody (TRAB), LDL, HDL, triglyceride (TG), total cholesterol, alkaline phosphatase (ALP), gamma-glutamyl transferase (GGT), total bilirubin, direct bilirubin, C-reactive protein (CRP), sedimentation, 25-hydroxy cholecalciferol (25-OH vitamin D), spot urine protein, and spot urine creatinine levels as well as complete blood count and complete urinalysis. Furthermore, fibrosis-4 (FIB-4) scores, subgrouped by FIB-4 risk categories [15] and glomerular filtration rates (GFR) [using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formula] were calculated using age and sex data and/or laboratory parameters [16]. Age and sex data were available for all participants. However, information on occupation, duration of employment/exposure, smoking and alcohol use, comorbidities, medications, and anthropometric measurements (height, weight, body mass index) was not available in the electronic archive. Because laboratory tests were ordered according to routine screening workflows and clinical indications, not all biochemical or endocrinological parameters were available for all participants; therefore, denominators vary across analyses and are reported for each outcome. In addition, radiological data (e.g., thyroid or abdominal ultrasonography/CT findings) could not be retrieved due to limitations in electronic transfer. Of the 37,763 records screened, 3,168 were excluded due to missing metal measurements (n = 2,999), inconsistencies in hormonal analyses (n = 157), and missing sex and/or age data (n = 12), leaving a final analytic cohort of 34,595 participants. Statistical analysis The Statistical Package for the Social Sciences (SPSS), version 25.0 (IBM Corp., Armonk, NY, USA), was used for statistical analyses. Normality of continuous variables was assessed using the Shapiro–Wilk test. Normally distributed variables were presented as mean ± standard deviation, whereas non-normally distributed variables were presented as median (interquartile range [IQR]; 25th–75th percentiles). Categorical variables were presented as frequencies and percentages. Between-group comparisons were performed using the independent-samples. t-test or Mann–Whitney U test for continuous variables, and Pearson’s chi-squared test, continuity correction, or Fisher’s exact test for categorical variables, as appropriate. Given the large number of analytes and outcomes in routine surveillance data, we prespecified a focused set of primary comparisons for commonly measured occupationally relevant metals (arsenic, cadmium, lead, manganese, and zinc) and primary clinical outcomes (thyroid dysfunction and dysglycemia) among tested participants. Additional laboratory comparisons were considered secondary/exploratory. All tests were two-sided, and p < 0.05 was considered statistically significant, interpreted in the context of multiplicity and effect size. To address multiplicity, we applied the Benjamini–Hochberg false discovery rate (FDR) procedure. Adjustment was performed separately for each prespecified metal (arsenic, cadmium, lead, manganese, and zinc), treating all outcome comparisons within a given metal as a single family of hypotheses. This within-metal family approach was chosen because the prespecified analyses were structured by metal exposure rather than by outcome, and because the biological interpretation of the findings is metal-specific. For each comparison, both raw p-values and Benjamini–Hochberg-adjusted q-values were reported, and findings were presented separately for primary (thyroid dysfunction and dysglycemia) and secondary/exploratory analyses (additional laboratory parameters). Statistical significance after FDR adjustment was defined as a q-value < 0.05. The FDR procedure was applied as a correction for multiple comparisons and should not be interpreted as an adjustment for confounding. Because the study was designed around occupational interpretive thresholds used in routine surveillance practice, primary analyses were prespecified using categorical exposure definitions rather than continuous exposure modeling. Additionally, continuous measurements were not uniformly available across all analytes and biological matrices in the retrospective archive. Subsequent multivariable linear regression analyses were conducted for selected biomarkers that demonstrated clinically or statistically significant exploratory associations. Variables with skewed distributions were natural log-transformed before regression analyses. Age and sex were included as covariates in adjusted models, except for FIB-4 analyses, in which age was not additionally included because age is mathematically incorporated into the FIB-4 equation. Adjusted regression coefficients (B) with 95% confidence intervals (CI) were reported. Results A total of 34,595 participants were included. The median age was 42 years; 45.6% were female, and 54.4% were male. For each heavy metal or trace element, the number of requested tests and the proportion of results above the interpretive limit are shown in Table 1. Table 1Participant characteristics and distribution of heavy metal and trace element measurements and interpretive limits (n = 34,595)All patientsAge (median, IQR)42.0 (20.0)Sex (n/%) Female Male15,770 (45.6%)18,825 (54.4%)Heavy metalsAge (median, IQR)Sex (F/M; %)N/%Level (mean ± SD or median (IQR)Interpretive limit used for classificationFrequency above interpretive limit among requested tests* (n/%)Antimony (Serum: µg/L)42.0 (15.0)6.7/93.3661 (1.9)3.2 ± 1.468 (1.2)Arsenic (urinary: (µg/L)40.0 (16.0)6.3/93.75030 (14.5)17.9 (18.7)351353 (26.9)Copper (µg/dL)41.0 (21.0)4.5/95.51571 (4.5)93.0 (21.0)> 14511 (0.7)Boron oxide (µg/L)40.0 (14.0)6.6/93.42801 (8.1)40.0 (60.0)100474 (18.1)Mercury (random urine, µg/L)40.0 (15.0)46.2/53.81595 (4.6)0.6 (0.6)102 (0.1)Cadmium (urine, µg/L)42.0 (20.0)5.2/94.81965 (5.7)0.5 (0.9)0-2.694 (4.8)Cobalt (µg/L)41.0 (14.0)5.0/95.01176 (3.4)0.2 (0.1)0–14 (0.3)Chromium (serum, µg/L)41.0 (14.0)4.5/95.51121 (3.5)3.1 (3.4)0–1028 (2.3)Lead (serum, µg/dL)41.0 (18.0)4.4/95.63205 (9.3)1.1 (0.8)0–5171 (5.3)Lithium (serum, mmol/L)39.0 (14.0)5.9/94.1999 (2.9)0.07 (0.03)1.60 (0.0)Manganese (serum, µg/L)40.0 (18.0)4.4/95.63298 (9.5)0.97 (0.48)0–253 (1.6)Nickel (serum, µg/L)41.0 (15.0)4.7/95.31432 (4.1)2.5 (2.6)0–102 (0.1)Molybdenum (serum, µg/L)42.0 (15.0)6.9/93.1635 (1.8)0.9 (0.4)0.3-26 (0.9)Selenium (serum, µg/L)42.0 (15.0)7.4/92.6648 (1.9)84.3 (17.0)60–1501 (0.2)Thallium (urine, µg/L)42.0 (14.0)6.7/93.3615 (1.8)0.2 (0.2)0–13 (0.5)Zinc (serum, µg/dL)35.0 (21.0)73.9/26.1394 (1.1)98.0 (23.3)60–106136 (34.5)*Interpretive limits indicate the clinical or occupational interpretation thresholds used by the laboratory information system for routine reporting. They are not analytical limits of detection or quantification and were not selected post hoc by the authors. Depending on the analyte and specimen matrix, these thresholds may represent reference intervals, upper reference limits, or biological exposure guidance values. The timing of sample collection relative to the work shift or work week was not consistently available in the retrospective archive; therefore, timing-dependent thresholds were interpreted as laboratory reporting thresholds rather than definitive compliance assessments. Threshold interpretation was based on the Mayo Clinic Laboratories heavy metals test catalog: https://www.mayocliniclabs.com/test-catalog/overview/7735#Clinical-and-Interpretive Table 1Participant characteristics and distribution of heavy metal and trace element measurements and interpretive limits (n = 34,595)All patientsAge (median, IQR)42.0 (20.0)Sex (n/%) Female Male15,770 (45.6%)18,825 (54.4%)Heavy metalsAge (median, IQR)Sex (F/M; %)N/%Level (mean ± SD or median (IQR)Interpretive limit used for classificationFrequency above interpretive limit among requested tests* (n/%)Antimony (Serum: µg/L)42.0 (15.0)6.7/93.3661 (1.9)3.2 ± 1.468 (1.2)Arsenic (urinary: (µg/L)40.0 (16.0)6.3/93.75030 (14.5)17.9 (18.7)351353 (26.9)Copper (µg/dL)41.0 (21.0)4.5/95.51571 (4.5)93.0 (21.0)> 14511 (0.7)Boron oxide (µg/L)40.0 (14.0)6.6/93.42801 (8.1)40.0 (60.0)100474 (18.1)Mercury (random urine, µg/L)40.0 (15.0)46.2/53.81595 (4.6)0.6 (0.6)102 (0.1)Cadmium (urine, µg/L)42.0 (20.0)5.2/94.81965 (5.7)0.5 (0.9)0-2.694 (4.8)Cobalt (µg/L)41.0 (14.0)5.0/95.01176 (3.4)0.2 (0.1)0–14 (0.3)Chromium (serum, µg/L)41.0 (14.0)4.5/95.51121 (3.5)3.1 (3.4)0–1028 (2.3)Lead (serum, µg/dL)41.0 (18.0)4.4/95.63205 (9.3)1.1 (0.8)0–5171 (5.3)Lithium (serum, mmol/L)39.0 (14.0)5.9/94.1999 (2.9)0.07 (0.03)1.60 (0.0)Manganese (serum, µg/L)40.0 (18.0)4.4/95.63298 (9.5)0.97 (0.48)0–253 (1.6)Nickel (serum, µg/L)41.0 (15.0)4.7/95.31432 (4.1)2.5 (2.6)0–102 (0.1)Molybdenum (serum, µg/L)42.0 (15.0)6.9/93.1635 (1.8)0.9 (0.4)0.3-26 (0.9)Selenium (serum, µg/L)42.0 (15.0)7.4/92.6648 (1.9)84.3 (17.0)60–1501 (0.2)Thallium (urine, µg/L)42.0 (14.0)6.7/93.3615 (1.8)0.2 (0.2)0–13 (0.5)Zinc (serum, µg/dL)35.0 (21.0)73.9/26.1394 (1.1)98.0 (23.3)60–106136 (34.5)*Interpretive limits indicate the clinical or occupational interpretation thresholds used by the laboratory information system for routine reporting. They are not analytical limits of detection or quantification and were not selected post hoc by the authors. Depending on the analyte and specimen matrix, these thresholds may represent reference intervals, upper reference limits, or biological exposure guidance values. The timing of sample collection relative to the work shift or work week was not consistently available in the retrospective archive; therefore, timing-dependent thresholds were interpreted as laboratory reporting thresholds rather than definitive compliance assessments. Threshold interpretation was based on the Mayo Clinic Laboratories heavy metals test catalog: https://www.mayocliniclabs.com/test-catalog/overview/7735#Clinical-and-Interpretive Participant characteristics and distribution of heavy metal and trace element measurements and interpretive limits (n = 34,595) Female Male 15,770 (45.6%) 18,825 (54.4%) *Interpretive limits indicate the clinical or occupational interpretation thresholds used by the laboratory information system for routine reporting. They are not analytical limits of detection or quantification and were not selected post hoc by the authors. Depending on the analyte and specimen matrix, these thresholds may represent reference intervals, upper reference limits, or biological exposure guidance values. The timing of sample collection relative to the work shift or work week was not consistently available in the retrospective archive; therefore, timing-dependent thresholds were interpreted as laboratory reporting thresholds rather than definitive compliance assessments. Threshold interpretation was based on the Mayo Clinic Laboratories heavy metals test catalog: https://www.mayocliniclabs.com/test-catalog/overview/7735#Clinical-and-Interpretive The mean or median values of various laboratory parameters, FIB-4 scores, and GFR were calculated based on the number of workers who were tested (Table 2). Table 2Summary of routine laboratory parameters among tested participantsN/%Mean ± SD or Median (IQR)Hemoglobin (g/dL)7703 (22.3)13.7 ± 1.7Hematocrit (%)7703 (22.3)44.1 ± 7.4WBC (/mm³)7703 (22.3)6920.0 (2430.0)Platelet count (x103/mm³)7703 (22.3)243.0 (78.0)CRP (mg/dL)7703 (22.3)2.5 (5.6)Sedimentation (mm/hr)7703 (22.3)13.0 (15.0)FBG (mg/dL)17,752 (51.3)98.0 (18.0)HbA1c (%)8405 (24.3)5.5 (0.9)Insulin (µU/mL)2057 (5.9)10.1 (8.3)Creatinine (mg/dL)23,836 (68.8)0.8 (0.3)eGFR (CKD-EPI), mL/min/1.73 m²30,257 (87.5)114.6 (21.0)Urea (mg/dL)19,425 (56.1)26 (10.0)Albumin (g/dL)80 (0.2)4.2 (0.1)Sodium (mmol/L)10,435 (30.1)138 (3.0)Potassium (mmol/L)10,449 (30.2)4.3 (0.5)ALT (U/L)7704 (22.2)19 (14.0)AST (U/L)7704 (22.2)20.0 (8.0)FIB-4 score21,164 (61.2)0.78 (0.56)FIB-4 category (n/%) Normal range Intermediate-range High range19,371 (56.0)1799 (5.2)91 (0.3)ALP (U/L)7704 (22.2)72.0 (30.0)GGT (U/L)7704 (22.2)19.0 (17.0)Total bilirubin (mg/dL)1841 (5.3)0.6 (0.3)Direct bilirubin (mg/dL)1841 (5.3)0.1 (0.05)TSH (µIU/L)9956 (28.8)1.8 (1.6)fT3 (pg/mL)9956 (28.8)3.1 (1.6)fT4 (ng/dL)9956 (28.8)1.1 (0.3)25-OH vitamin D (ng/ml)4980 (14.4)14.2 (12.3)Triglyceride (mg/dL)9373 (27.1)122.0 (93.0)HDL (mg/dL)9373 (27.1)49.0 (16.0)LDL (mg/dL)9373 (27.1)118.0 (48.0)Urine analysis (n = 34,000)Pyuria3623 (10.7)Ketone (positive)136 (0.4)Protein (positive)13 (0.04)Nitrite (positive)204 (0.6)ALT: Alanine aminotransferase, AST: Aspartate aminotransferase, * CKD-EPI: Calculated pursuant to Chronic Kidney Disease Epidemiology Collaboration, CRP: C-reactive protein, FBG: Fasting blood glucose, FIB-4: Fibrosis-4 score, GGT: Gamma-glutamyl transferase, HDL: High-density lipoprotein, LDL: Low-density lipoprotein, FT3: free triiodothyronine, FT4: free thyroxine, TG: Triglyceride, TSH: thyroid-stimulating hormone, WBC: White blood cell Table 2Summary of routine laboratory parameters among tested participantsN/%Mean ± SD or Median (IQR)Hemoglobin (g/dL)7703 (22.3)13.7 ± 1.7Hematocrit (%)7703 (22.3)44.1 ± 7.4WBC (/mm³)7703 (22.3)6920.0 (2430.0)Platelet count (x103/mm³)7703 (22.3)243.0 (78.0)CRP (mg/dL)7703 (22.3)2.5 (5.6)Sedimentation (mm/hr)7703 (22.3)13.0 (15.0)FBG (mg/dL)17,752 (51.3)98.0 (18.0)HbA1c (%)8405 (24.3)5.5 (0.9)Insulin (µU/mL)2057 (5.9)10.1 (8.3)Creatinine (mg/dL)23,836 (68.8)0.8 (0.3)eGFR (CKD-EPI), mL/min/1.73 m²30,257 (87.5)114.6 (21.0)Urea (mg/dL)19,425 (56.1)26 (10.0)Albumin (g/dL)80 (0.2)4.2 (0.1)Sodium (mmol/L)10,435 (30.1)138 (3.0)Potassium (mmol/L)10,449 (30.2)4.3 (0.5)ALT (U/L)7704 (22.2)19 (14.0)AST (U/L)7704 (22.2)20.0 (8.0)FIB-4 score21,164 (61.2)0.78 (0.56)FIB-4 category (n/%) Normal range Intermediate-range High range19,371 (56.0)1799 (5.2)91 (0.3)ALP (U/L)7704 (22.2)72.0 (30.0)GGT (U/L)7704 (22.2)19.0 (17.0)Total bilirubin (mg/dL)1841 (5.3)0.6 (0.3)Direct bilirubin (mg/dL)1841 (5.3)0.1 (0.05)TSH (µIU/L)9956 (28.8)1.8 (1.6)fT3 (pg/mL)9956 (28.8)3.1 (1.6)fT4 (ng/dL)9956 (28.8)1.1 (0.3)25-OH vitamin D (ng/ml)4980 (14.4)14.2 (12.3)Triglyceride (mg/dL)9373 (27.1)122.0 (93.0)HDL (mg/dL)9373 (27.1)49.0 (16.0)LDL (mg/dL)9373 (27.1)118.0 (48.0)Urine analysis (n = 34,000)Pyuria3623 (10.7)Ketone (positive)136 (0.4)Protein (positive)13 (0.04)Nitrite (positive)204 (0.6)ALT: Alanine aminotransferase, AST: Aspartate aminotransferase, * CKD-EPI: Calculated pursuant to Chronic Kidney Disease Epidemiology Collaboration, CRP: C-reactive protein, FBG: Fasting blood glucose, FIB-4: Fibrosis-4 score, GGT: Gamma-glutamyl transferase, HDL: High-density lipoprotein, LDL: Low-density lipoprotein, FT3: free triiodothyronine, FT4: free thyroxine, TG: Triglyceride, TSH: thyroid-stimulating hormone, WBC: White blood cell Summary of routine laboratory parameters among tested participants Normal range Intermediate-range High range 19,371 (56.0) 1799 (5.2) 91 (0.3) ALT: Alanine aminotransferase, AST: Aspartate aminotransferase, * CKD-EPI: Calculated pursuant to Chronic Kidney Disease Epidemiology Collaboration, CRP: C-reactive protein, FBG: Fasting blood glucose, FIB-4: Fibrosis-4 score, GGT: Gamma-glutamyl transferase, HDL: High-density lipoprotein, LDL: Low-density lipoprotein, FT3: free triiodothyronine, FT4: free thyroxine, TG: Triglyceride, TSH: thyroid-stimulating hormone, WBC: White blood cell Among the 9,956 participants with available thyroid function tests (TSH, fT3, and fT4), 24 (0.24%) had overt hypothyroidism, 1,193 (11.9%) had subclinical hypothyroidism, 58 (0.6%) had thyrotoxicosis, and 164 (1.6%) had subclinical hyperthyroidism (Table 3; Fig. 1). Thyroid dysfunction categories were defined using biochemical criteria as follows: overt hypothyroidism, elevated TSH (> 4 mIU/L; >5 mIU/L for individuals aged > 65 years) with low fT4 and/or fT3; subclinical hypothyroidism, elevated TSH with normal fT3 and fT4; overt hyperthyroidism, elevated fT3 and/or fT4 with suppressed TSH (TSH < 0.35 mIU/L); and subclinical hyperthyroidism, normal fT3 and fT4 with suppressed TSH (TSH < 0.35 mIU/L) [17, 18]. Among the 20,738 participants (59.9% of the cohort) with available fasting plasma glucose and/or HbA1c results, 2,653 (12.8%) met criteria for diabetes mellitus and 2,064 (10.0%) met criteria for prediabetes (Table 3; Fig. 1). Prediabetes was defined as HbA1c 5.7–6.4% and/or fasting plasma glucose 100–125 mg/dL, and diabetes mellitus as HbA1c ≥ 6.5% and/or fasting plasma glucose ≥ 126 mg/dL [19]. Overall, 32 (1.2%) of individuals with diabetes mellitus had ketonuria. Table 3Frequencies of thyroid dysfunction, diabetes mellitus, and prediabetes among tested participantsThyroid diseasesTested for thyroid function (TSH, fT3, and fT4) (n = 9956)HypothyroidismSubclinical hypothyroidismThyrotoxicosisSubclinical hyperthyroidism24 (0.24%)1193 (11.9%)58 (0.6%)164 (1.6%) Tests for HbA1c or fasting glucose ( n = 20738, 59.9%) DMPrediabetes Low risk High risk2653 (12.8%)*2064 (10.0%)**1230 (59.6%)834 (40.4%)*Diabetes mellitus and prediabetes proportions were calculated among participants with fasting plasma glucose and/or HbA1c available (n = 20,738). **Low- and high-risk prediabetes subcategories were calculated among participants with HbA1c available (n = 9,667)DM: Diabetes Mellitus, HbA1c: Glycated hemoglobin; FT3: free triiodothyronine; FT4: free thyroxine; TSH: thyroid-stimulating hormone Table 3Frequencies of thyroid dysfunction, diabetes mellitus, and prediabetes among tested participantsThyroid diseasesTested for thyroid function (TSH, fT3, and fT4) (n = 9956)HypothyroidismSubclinical hypothyroidismThyrotoxicosisSubclinical hyperthyroidism24 (0.24%)1193 (11.9%)58 (0.6%)164 (1.6%) Tests for HbA1c or fasting glucose ( n = 20738, 59.9%) DMPrediabetes Low risk High risk2653 (12.8%)*2064 (10.0%)**1230 (59.6%)834 (40.4%)*Diabetes mellitus and prediabetes proportions were calculated among participants with fasting plasma glucose and/or HbA1c available (n = 20,738). **Low- and high-risk prediabetes subcategories were calculated among participants with HbA1c available (n = 9,667)DM: Diabetes Mellitus, HbA1c: Glycated hemoglobin; FT3: free triiodothyronine; FT4: free thyroxine; TSH: thyroid-stimulating hormone Frequencies of thyroid dysfunction, diabetes mellitus, and prediabetes among tested participants Hypothyroidism Subclinical hypothyroidism Thyrotoxicosis Subclinical hyperthyroidism 24 (0.24%) 1193 (11.9%) 58 (0.6%) 164 (1.6%) DM Prediabetes Low risk High risk 2653 (12.8%)* 2064 (10.0%)** 1230 (59.6%) 834 (40.4%) *Diabetes mellitus and prediabetes proportions were calculated among participants with fasting plasma glucose and/or HbA1c available (n = 20,738). **Low- and high-risk prediabetes subcategories were calculated among participants with HbA1c available (n = 9,667) DM: Diabetes Mellitus, HbA1c: Glycated hemoglobin; FT3: free triiodothyronine; FT4: free thyroxine; TSH: thyroid-stimulating hormone Fig. 1Thyroid function and glycemic status of the participants Thyroid function and glycemic status of the participants Participants with values above the interpretive limit for cadmium, lead, and manganese were younger than those with values within the interpretive limit, and sex distributions differed across several metals, including arsenic and zinc (Table 4). Table 4Comparison of age and sex between participants with values within versus above the interpretive limit for prespecified heavy metals and trace elements (arsenic, cadmium, lead, manganese, and zinc)Age (Median-IQR)Sex (n/%)Arsenic (urinary: (µg/L)Within the interpretive limit40.0 (16.0)FemaleMale266 (7.2)3411 (92.8)Above interpretive limit40.0 (14.0)FemaleMale52 (3.8)1301 (96.2)p0.835a< 0.001bCadmium (urine, µg/L)Within the interpretive limit42.0 (20.0)FemaleMale102 (5.5)1769 (94.5)Above interpretive limit36.0 (15.0)FemaleMale0 (0.0)94 (100.0)p< 0.001a0.014bLead (serum, µg/dL)Within the interpretive limit41.0 (17.0)FemaleMale137 (4.5)2897 (95.5)Above interpretive limit35.0 (14.0)FemaleMale4 (2.3)167 (97.7)p< 0.001a0.247bManganese (serum, µg/L)Within the interpretive limit40.0 (18.0)FemaleMale143 (4.4)3102 (95.6)Above interpretive limit30.0 (13.0)FemaleMale3 (5.7)50 (94.3)p< 0.001a0.508bZinc (serum, µg/dL)Within the interpretive limit35.0 (21.0)FemaleMale211 (81.8)47 (18.2)Above interpretive limit33.5 (21.0)FemaleMale80 (58.8)56 (41.2)p0.935a< 0.001ba Mann-Whitney U Test, b Fisher’s Exact Test Table 4Comparison of age and sex between participants with values within versus above the interpretive limit for prespecified heavy metals and trace elements (arsenic, cadmium, lead, manganese, and zinc)Age (Median-IQR)Sex (n/%)Arsenic (urinary: (µg/L)Within the interpretive limit40.0 (16.0)FemaleMale266 (7.2)3411 (92.8)Above interpretive limit40.0 (14.0)FemaleMale52 (3.8)1301 (96.2)p0.835a< 0.001bCadmium (urine, µg/L)Within the interpretive limit42.0 (20.0)FemaleMale102 (5.5)1769 (94.5)Above interpretive limit36.0 (15.0)FemaleMale0 (0.0)94 (100.0)p< 0.001a0.014bLead (serum, µg/dL)Within the interpretive limit41.0 (17.0)FemaleMale137 (4.5)2897 (95.5)Above interpretive limit35.0 (14.0)FemaleMale4 (2.3)167 (97.7)p< 0.001a0.247bManganese (serum, µg/L)Within the interpretive limit40.0 (18.0)FemaleMale143 (4.4)3102 (95.6)Above interpretive limit30.0 (13.0)FemaleMale3 (5.7)50 (94.3)p< 0.001a0.508bZinc (serum, µg/dL)Within the interpretive limit35.0 (21.0)FemaleMale211 (81.8)47 (18.2)Above interpretive limit33.5 (21.0)FemaleMale80 (58.8)56 (41.2)p0.935a< 0.001ba Mann-Whitney U Test, b Fisher’s Exact Test Comparison of age and sex between participants with values within versus above the interpretive limit for prespecified heavy metals and trace elements (arsenic, cadmium, lead, manganese, and zinc) Female Male 266 (7.2) 3411 (92.8) Female Male 52 (3.8) 1301 (96.2) Female Male 102 (5.5) 1769 (94.5) Female Male 0 (0.0) 94 (100.0) Female Male 137 (4.5) 2897 (95.5) Female Male 4 (2.3) 167 (97.7) Female Male 143 (4.4) 3102 (95.6) Female Male 3 (5.7) 50 (94.3) Female Male 211 (81.8) 47 (18.2) Female Male 80 (58.8) 56 (41.2) a Mann-Whitney U Test, b Fisher’s Exact Test Among participants with the relevant laboratory tests available, thyroid dysfunction, diabetes mellitus, and prediabetes were compared between those with values within versus above the interpretive limit for each prespecified metal (Table 5). In univariable comparisons, overt hypothyroidism was numerically more frequent in the elevated-lead stratum than in the within-limit stratum; however, this finding was based on only two cases in the elevated-lead stratum and should be interpreted with caution. The intermediate/high FIB-4 category, shown in Table 6, was less frequent in the elevated-lead stratum. Because age is included in the FIB-4 calculation and participants with elevated lead were substantially younger than those within the interpretive limit, this difference most likely reflects the age distribution of the strata rather than a true association between lead status and lower liver fibrosis risk. These findings should not be interpreted as population-level prevalence estimates or as evidence of causal associations. Table 5Comparison of thyroid dysfunction, diabetes mellitus, and prediabetes among participants with available relevant laboratory tests, stratified by within versus above interpretive limit status for prespecified heavy metals and trace elementsHypothyroidism (n = 24, 0.2%)Subclinical hypothyroidism (n = 1193, 12.0%)Thyrotoxicosis (n = 58, 0.6%)Subclinical hyperthyroidism (n = 164, 1.6%)DM (n = 2653, 12.8%)Prediabetes (n = 2064, 10.0%)Arsenic (urinary) (µg/L)Within the interpretive limit2 (0.3)14 (2.1)-11 (1.7)215 (7.0)746 (24.2)Above interpretive limit0 (0.0)3 (5.0)-1 (1.7)86 (6.5)319 (24.2)p0.952a0.389a*0.260b0.603c1.000cCadmium (urine, µg/L)Within the interpretive limit2 (0.3)16 (2.8)-11 (1.9)132 (8.5)436 (28.1)Above interpretive limit0 (0.0)1 (1.9)-0 (0.0)5 (6.1)24 (29.3)p0.751b0.831b-0.456b0.544c0.802cLead (serum, µg/dL)Within the interpretive limit0 (0.0)16 (2.8)-9 (1.6)162 (7.3)566 (25.4)Above interpretive limit2 (1.4)3 (2.1)-2 (1.4)5 (3.3)28 (18.4)p< 0.001b0.128a*0.220b0.070c0.053cManganese (serum, µg/L)Within the interpretive limit2 (0.3)16 (2.6)-10 (1.6)179 (6.9)690 (26.5)Above interpretive limit0 (0.0)1 (2.4)-1 (2.4)2 (4.1)14 (28.6)p0.857a0.160b*0.438a0.442b0.745cZinc (serum, µg/dL)Within the interpretive limit1 (0.2)15 (3.1)-1 (0.4)72 (5.8)283 (22.7)Above interpretive limit----9 (10.0)35 (27.6)p*0.154c**0.109c1.000cPercentages are calculated within participants with available relevant testsa Continuity correction, b Pearson chi-squared test, c Fisher’s exact test, * No test or not sufficient data for analysis; DM: Diabetes mellitus Table 5Comparison of thyroid dysfunction, diabetes mellitus, and prediabetes among participants with available relevant laboratory tests, stratified by within versus above interpretive limit status for prespecified heavy metals and trace elementsHypothyroidism (n = 24, 0.2%)Subclinical hypothyroidism (n = 1193, 12.0%)Thyrotoxicosis (n = 58, 0.6%)Subclinical hyperthyroidism (n = 164, 1.6%)DM (n = 2653, 12.8%)Prediabetes (n = 2064, 10.0%)Arsenic (urinary) (µg/L)Within the interpretive limit2 (0.3)14 (2.1)-11 (1.7)215 (7.0)746 (24.2)Above interpretive limit0 (0.0)3 (5.0)-1 (1.7)86 (6.5)319 (24.2)p0.952a0.389a*0.260b0.603c1.000cCadmium (urine, µg/L)Within the interpretive limit2 (0.3)16 (2.8)-11 (1.9)132 (8.5)436 (28.1)Above interpretive limit0 (0.0)1 (1.9)-0 (0.0)5 (6.1)24 (29.3)p0.751b0.831b-0.456b0.544c0.802cLead (serum, µg/dL)Within the interpretive limit0 (0.0)16 (2.8)-9 (1.6)162 (7.3)566 (25.4)Above interpretive limit2 (1.4)3 (2.1)-2 (1.4)5 (3.3)28 (18.4)p< 0.001b0.128a*0.220b0.070c0.053cManganese (serum, µg/L)Within the interpretive limit2 (0.3)16 (2.6)-10 (1.6)179 (6.9)690 (26.5)Above interpretive limit0 (0.0)1 (2.4)-1 (2.4)2 (4.1)14 (28.6)p0.857a0.160b*0.438a0.442b0.745cZinc (serum, µg/dL)Within the interpretive limit1 (0.2)15 (3.1)-1 (0.4)72 (5.8)283 (22.7)Above interpretive limit----9 (10.0)35 (27.6)p*0.154c**0.109c1.000cPercentages are calculated within participants with available relevant testsa Continuity correction, b Pearson chi-squared test, c Fisher’s exact test, * No test or not sufficient data for analysis; DM: Diabetes mellitus Comparison of thyroid dysfunction, diabetes mellitus, and prediabetes among participants with available relevant laboratory tests, stratified by within versus above interpretive limit status for prespecified heavy metals and trace elements Percentages are calculated within participants with available relevant tests a Continuity correction, b Pearson chi-squared test, c Fisher’s exact test, * No test or not sufficient data for analysis; DM: Diabetes mellitus Comparisons of selected laboratory parameters between participants with values within versus above the interpretive limits are presented in Table 6. For arsenic, no metabolic, hepatic, lipid, thyroid, hematological, or renal biomarker difference remained statistically significant after within-metal FDR correction (all q-values ≥ 0.197). Elevated cadmium levels were associated with higher WBC counts after FDR adjustment (adjusted p = 0.013). Differences in FIB-4 (adjusted p = 0.066) and HDL (adjusted p = 0.074) did not remain statistically significant after correction, and no significant associations were observed for glycemic, hepatic, renal, or thyroid parameters. Elevated lead levels were associated with lower hemoglobin (q = 0.012), lower FBG (q = 0.003), lower ALT (q = 0.003), lower FIB-4 values (q = 0.003), lower LDL levels (q = 0.003), and higher eGFR (q = 0.003) after FDR correction. No significant associations were observed for HbA1c (adjusted p = 0.086), AST (adjusted p = 0.086), HDL (adjusted p = 0.326), TG (adjusted p = 0.644), TSH (adjusted p = 0.288), WBC (adjusted p = 0.288), platelet count (adjusted p = 0.953), or fT4 (adjusted p = 0.288). Elevated manganese levels were associated with lower FIB-4 scores (adjusted p = 0.046) and lower LDL levels (adjusted p = 0.039) after FDR correction. No significant differences were observed for FBG (adjusted p = 0.058), HDL (adjusted p = 0.058), HbA1c (adjusted p = 0.224), ALT (adjusted p = 0.908), AST (adjusted p = 0.317), TG (adjusted p = 0.506), TSH (adjusted p = 0.953), hematological indices, or eGFR (adjusted p = 0.053). Participants above the zinc interpretive limit demonstrated higher hemoglobin (adjusted p = 0.003), higher fT4 (adjusted p = 0.042), higher ALT (adjusted p = 0.003), higher AST (adjusted p = 0.003), and higher TG levels (adjusted p = 0.003) after FDR adjustment. No significant associations were observed for FBG (adjusted p = 0.261), HbA1c (adjusted p = 0.297), FIB-4 (adjusted p = 0.084), LDL (adjusted p = 0.698), HDL (adjusted p = 0.698), WBC (adjusted p = 0.698), platelet count (adjusted p = 0.083), eGFR (adjusted p = 0.087), fT3 (adjusted p = 0.093), or TSH (adjusted p = 0.409). Hgb, WBC, Platelet count, eGFR (CKD-EPI), fT4, and fT3 levels are presented in Supplementary Table S1. For each metal, raw p-values and adjusted p-values are shown in Supplementary Table S2-6. Table 6Comparison of selected laboratory parameters between participants with values within versus above the interpretive limits for prespecified metals/trace elementsFBG (mg/dL)HbA1c (%)ALT (U/L)AST(U/L)FIB-4LDL (mg/dL)HDL (mg/dL)TG (mg/dL)TSH (µIU/L)Arsenic (urinary: µg/L)Within the interpretive limit95.0 (15.0)5.3 (1.0)22.0 (15.0)21.0 (7.0)0.79 (0.49)116.0 (44.0)45.0 (12.0)122.0 (94.0)1.4 (1.0)Above interpretive limit99.0 (17.0)5.3 (1.7)22.0 (11.0)21.0 (6.0)0.77 (0.44)117.0 (50.5)44.0 (12.8)117.0 (99.3)1.5 (0.8)q†0.1970.3410.8270.9840.3330.9840.9840.9840.984Cadmium (urine, µg/L)Within the interpretive limit96.0 (15.0)*22.0 (15.0)21.0 (7.0)0.80 (0.49)116.0 (44.0)45.0 (12.0)123.0 (100.0)1.4 (1.0)Above interpretive limit94.0 (16.3)*21.0 (13.0)20.0 (7.0)0.70 (0.41)109.0 (51.3)42.0 (10.3)133.0 (111.3)1.6 (1.1)q†0.246*0.2340.2460.0660.3530.0740.5230.758Lead (serum, µg/dL)Within the interpretive limit96.0 (15.0)5.2 (1.6)23.0 (14.0)21.0 (7.0)0.81 (0.48)116.0 (45.0)45.0 (12.0)121.0 (92.0)1.4 (0.9)Above interpretive limit90.0 (14.3)5.0 (1.8)19.0 (14.5)20.0 (7.0)0.64 (0.35)104.0 (77.0)45.0 (11.0)127.0 (105.5)1.4 (1.0)q†0.0030.0860.0030.0860.0030.0030.3260.6440.288Manganese (serum, µg/L)Within the interpretive limit96.0 (14.0)5.3 (1.7)23.0 (15.0)21.0 (7.0)0.79 (0.48)116.0 (45.0)45.0 (11.0)121.0 (92.4)1.6 (1.2)Above interpretive limit91.0 (18.5)5.3 (0.7)22.0 (26.0)23.0 (10.0)0.63 (0.40)103.3 (53.0)40.5 (12.0)148.0 (118.0)1.8 (1.6)q†0.0580.2240.9080.3170.0460.0390.0580.5060.953Zinc (serum, µg/dL)Within the interpretive limit93.0 (13.0)5.5 (0.6)15.0 (9.0)18.0 (7.0)0.79 (0.40)126.1 ± 38.155.2 ± 13.589.0 (58.5)2.0 (1.5)Above interpretive limit95.0 (14.0)5.5 (0.6)18.0 (13.0)20.0 (6.0)0.75 (0.50)124.8 ± 32.955.3 ± 13.2118 (102.3)1.8 (1.3)q†0.2610.2970.0030.0030.0840.6980.6980.0030.409Values are presented as mean ± SD or median (IQR), † Benjamini–Hochberg-False Discovery Rate-adjusted q-values derived from Mann–Whitney U test or independent-samples t-test according to distribution* No test or not sufficient for statistical analysisFBG: fasting blood glucose, HbA1c: Glycated Hemoglobin, ALT: Alanine aminotransferase, AST: Aspartate aminotransferase, FIB-4: Fibrosis-4 score, LDL: Low-density lipoprotein, HDL: High-density lipoprotein, TSH: Thyroid-stimulating hormone Table 6Comparison of selected laboratory parameters between participants with values within versus above the interpretive limits for prespecified metals/trace elementsFBG (mg/dL)HbA1c (%)ALT (U/L)AST(U/L)FIB-4LDL (mg/dL)HDL (mg/dL)TG (mg/dL)TSH (µIU/L)Arsenic (urinary: µg/L)Within the interpretive limit95.0 (15.0)5.3 (1.0)22.0 (15.0)21.0 (7.0)0.79 (0.49)116.0 (44.0)45.0 (12.0)122.0 (94.0)1.4 (1.0)Above interpretive limit99.0 (17.0)5.3 (1.7)22.0 (11.0)21.0 (6.0)0.77 (0.44)117.0 (50.5)44.0 (12.8)117.0 (99.3)1.5 (0.8)q†0.1970.3410.8270.9840.3330.9840.9840.9840.984Cadmium (urine, µg/L)Within the interpretive limit96.0 (15.0)*22.0 (15.0)21.0 (7.0)0.80 (0.49)116.0 (44.0)45.0 (12.0)123.0 (100.0)1.4 (1.0)Above interpretive limit94.0 (16.3)*21.0 (13.0)20.0 (7.0)0.70 (0.41)109.0 (51.3)42.0 (10.3)133.0 (111.3)1.6 (1.1)q†0.246*0.2340.2460.0660.3530.0740.5230.758Lead (serum, µg/dL)Within the interpretive limit96.0 (15.0)5.2 (1.6)23.0 (14.0)21.0 (7.0)0.81 (0.48)116.0 (45.0)45.0 (12.0)121.0 (92.0)1.4 (0.9)Above interpretive limit90.0 (14.3)5.0 (1.8)19.0 (14.5)20.0 (7.0)0.64 (0.35)104.0 (77.0)45.0 (11.0)127.0 (105.5)1.4 (1.0)q†0.0030.0860.0030.0860.0030.0030.3260.6440.288Manganese (serum, µg/L)Within the interpretive limit96.0 (14.0)5.3 (1.7)23.0 (15.0)21.0 (7.0)0.79 (0.48)116.0 (45.0)45.0 (11.0)121.0 (92.4)1.6 (1.2)Above interpretive limit91.0 (18.5)5.3 (0.7)22.0 (26.0)23.0 (10.0)0.63 (0.40)103.3 (53.0)40.5 (12.0)148.0 (118.0)1.8 (1.6)q†0.0580.2240.9080.3170.0460.0390.0580.5060.953Zinc (serum, µg/dL)Within the interpretive limit93.0 (13.0)5.5 (0.6)15.0 (9.0)18.0 (7.0)0.79 (0.40)126.1 ± 38.155.2 ± 13.589.0 (58.5)2.0 (1.5)Above interpretive limit95.0 (14.0)5.5 (0.6)18.0 (13.0)20.0 (6.0)0.75 (0.50)124.8 ± 32.955.3 ± 13.2118 (102.3)1.8 (1.3)q†0.2610.2970.0030.0030.0840.6980.6980.0030.409Values are presented as mean ± SD or median (IQR), † Benjamini–Hochberg-False Discovery Rate-adjusted q-values derived from Mann–Whitney U test or independent-samples t-test according to distribution* No test or not sufficient for statistical analysisFBG: fasting blood glucose, HbA1c: Glycated Hemoglobin, ALT: Alanine aminotransferase, AST: Aspartate aminotransferase, FIB-4: Fibrosis-4 score, LDL: Low-density lipoprotein, HDL: High-density lipoprotein, TSH: Thyroid-stimulating hormone Comparison of selected laboratory parameters between participants with values within versus above the interpretive limits for prespecified metals/trace elements Values are presented as mean ± SD or median (IQR), † Benjamini–Hochberg-False Discovery Rate-adjusted q-values derived from Mann–Whitney U test or independent-samples t-test according to distribution * No test or not sufficient for statistical analysis FBG: fasting blood glucose, HbA1c: Glycated Hemoglobin, ALT: Alanine aminotransferase, AST: Aspartate aminotransferase, FIB-4: Fibrosis-4 score, LDL: Low-density lipoprotein, HDL: High-density lipoprotein, TSH: Thyroid-stimulating hormone Variables demonstrating significant between-group differences and non-normal distribution were natural log-transformed before age- and/or sex-adjusted linear regression analyses. Elevated lead status remained independently associated with lower ln(FBG) levels (adjusted B − 0.062, 95% CI − 0.097 to − 0.027; p < 0.001), lower ln(ALT) levels (adjusted B − 0.192, 95% CI − 0.265 to − 0.119; p < 0.001), and lower ln(LDL) levels (adjusted B − 0.073, 95% CI − 0.124 to − 0.023; p = 0.005). Elevated lead status also remained associated with lower ln(FIB-4) values after sex adjustment (adjusted B − 0.205, 95% CI − 0.272 to − 0.138; p < 0.001); age was not additionally included because age is mathematically incorporated into the FIB-4 equation (Supplementary Table S7, Fig. 2). For zinc, associations with ln(ALT) and ln(AST) attenuated after adjustment and were no longer statistically significant, whereas elevated zinc status remained independently associated with higher ln(TG) levels (adjusted B 0.208, 95% CI 0.083 to 0.333; p = 0.001) (Supplementary Table S7, Fig. 2). Fig. 2Forest plot of adjusted regression coefficients for selected exploratory biomarker associations according to elevated lead and zinc status Forest plot of adjusted regression coefficients for selected exploratory biomarker associations according to elevated lead and zinc status Discussion In this large occupational screening cohort, thyroid dysfunction and dysglycemia were frequently observed among tested participants. In univariable comparisons, several metals above the interpretive limit were associated with differences in metabolic biomarkers (e.g., lipid parameters and liver enzymes). When multiplicity was addressed using within‑metal false discovery rate adjustment, the most consistent signals were observed for lead- and zinc-associated differences, whereas other nominal associations should be interpreted as exploratory. Participants with elevated levels of cadmium, lead, or manganese tended to be younger, and elevated arsenic, cadmium, and zinc levels were more prevalent among males, highlighting the importance of considering demographic distributions when interpreting surveillance data. Overall, these findings should be regarded as exploratory surveillance signals that may help inform priorities for further investigation and ongoing biomarker monitoring within occupational health programs, rather than as evidence of metal-specific causal effects. To our knowledge, this is one of the largest single-center archive-based analyses of occupational metal/trace element measurements linked to endocrine–metabolic laboratory parameters in Türkiye. As metals are not biodegradable, they can accumulate in the body with chronic exposure and have been associated with a wide range of health outcomes. The thyroid gland, pancreas, liver, heart, brain, gonads, and kidneys are considered primary target organs. The liver is a key organ for xenobiotic metabolism; therefore, endocrine disruptor–related injury is frequently discussed in this organ [20]. An animal study reported that the liver shows substantial accumulation of several metals, with an order of accumulation reported as lead > cadmium > nickel > chromium [21]. In the present study, we evaluated FIB-4 as a noninvasive indicator of liver fibrosis risk. FIB-4 values and the proportion of participants in the intermediate/high FIB-4 category were not higher — and for lead, were lower — in strata with metal results above the interpretive limit. Because age is a key component of the FIB-4 formula and participants with elevated cadmium, lead, and manganese tended to be younger than those within the interpretive limit, the lower FIB-4 values in these strata are most likely explained by age differences confounding rather than by a protective effect of metal exposure. Accordingly, the present FIB-4 comparisons across exposure strata should not be interpreted as evidence of differential liver fibrosis risk, and any interpretation requires age-adjusted modeling. In the multivariable analysis, age was deliberately not included as a covariate for FIB-4 because age is mathematically incorporated into the FIB-4 equation; therefore, the FIB-4 model was adjusted only for sex. This sex-adjusted analysis cannot disentangle the mathematical contribution of age from other factors, and the lower FIB-4 values in the elevated-lead stratum should therefore remain interpreted as exploratory rather than as evidence of lower fibrosis risk or a protective association. The present study indicated lower HDL cholesterol levels in participants with elevated cadmium and manganese levels and higher triglyceride levels in those with elevated zinc levels. For arsenic, fasting glucose values were numerically higher in the elevated group, but the difference was small in univariable analyses. As with many endocrine-disrupting exposures, it is difficult to directly attribute lipid or glycemic differences to a single metal because of concomitant occupational and lifestyle factors, which likely contribute to heterogeneity across epidemiological studies. In age- and sex-adjusted multivariable linear regression analyses (Supplementary Table S7), elevated lead status remained associated with lower fasting plasma glucose, ALT, and LDL values. The FIB-4 analysis was adjusted only for sex because age is incorporated into the FIB-4 formula and therefore should be interpreted separately. For zinc, the previously observed associations with ALT and AST attenuated and were no longer statistically significant after adjustment, whereas the association with higher triglycerides persisted. For arsenic, cadmium, and manganese, exploratory findings did not warrant additional multivariable modeling because univariable associations either did not survive within-metal FDR correction or were considered insufficient for stable modeling. Additional FDR-adjusted findings, including higher white blood cell counts in the elevated-cadmium stratum and higher fT4 levels in the elevated-zinc stratum, were considered exploratory and were not interpreted further because relevant covariate data were unavailable. These adjusted findings refine the univariable signals but remain exploratory because adjustment was limited to age and sex. Heavy metals and metalloids have been implicated in the pathogenesis of diabetes mellitus and cardiovascular disease through inflammatory, lipoprotein-metabolic, and oxidative stress pathways, and combined exposure to multiple metals may further increase cardiovascular risk [22–25]. Among the elements assessed in the present study, arsenic is notable because of its widespread environmental distribution. Arsenic is a metalloid endocrine disruptor that can contaminate drinking water and represents a global public health concern. Experimental studies suggest that arsenic exposure can impair pancreatic β‑cell function and contribute to dysglycemia [26–28]. In the present occupational screening dataset, fasting glucose values were slightly higher in the elevated arsenic group, whereas HbA1c differences were not apparent. Given the relatively young age of the cohort, the limited availability of detailed exposure duration data, and the cross-sectional design, any long-term cumulative associations between arsenic exposure and glucose metabolism cannot be evaluated in the present dataset and would require longitudinal studies with quantitative exposure assessment. To place the surveillance findings in a broader context, the overall prevalence of diabetes mellitus among participants with available fasting plasma glucose and/or HbA1c data (12.8% in 20,738 individuals; median age approximately 40 years) was higher than the diabetes prevalence reported for the corresponding age groups in the Turkish general population based on national electronic health records (approximately 2.4% at 30–34 years, 3.9% at 35–39 years, and 6.3% at 40–44 years) [29]. Although this difference cannot be attributed to a specific occupational exposure given the limited covariate information, retrospective design, and the exploratory nature of the present analyses, it is consistent with the hypothesis that workforces with potential exposure to heavy metals—including arsenic, cadmium, lead, and manganese—may carry an increased burden of dysglycemia and possibly thyroid dysfunction compared with the general age- and sex-matched population. This observation supports prior evidence suggesting that occupational exposure to arsenic and other heavy metals may contribute to an elevated risk of diabetes and warrants further investigation in prospective, dose–response studies. In contrast to the other prespecified metals, zinc is an essential trace element with tightly regulated homeostasis, and serum zinc concentrations are influenced by nutritional status, recent dietary intake, acute-phase responses, physiological variation, and exposure. Within the present occupational screening dataset, values above the upper reference interval were therefore interpreted as ‘above the reference interval’ rather than as evidence of toxic exposure, and findings involving zinc should be considered separately from those involving toxic metals such as arsenic, cadmium, lead, and manganese. Zinc was included in the prespecified set because it was one of the most frequently measured elements in this surveillance dataset and because both deficiency and elevations have been linked to lipid and glucose metabolism in the existing literature [23, 24]. The observed higher triglyceride and aminotransferase values in the above-reference zinc stratum are exploratory and may reflect nutritional, anthropometric, or comorbidity-related factors that could not be assessed in this retrospective archive. These signals require confirmation in studies with complete dietary, anthropometric, and clinical data. Thyroid function test data were available for 9,956 participants, and thyroid dysfunction was observed in 14.3% of this tested subgroup. This proportion should not be interpreted as the prevalence of thyroid dysfunction in the entire screened workforce. Because thyroid tests were ordered according to routine clinical or occupational workflows rather than a standardized research protocol, the tested subgroup may differ systematically from untested participants. Therefore, these findings support cautious consideration of thyroid monitoring within occupational surveillance, but do not establish population-level prevalence or a causal relationship with metal measurements. This study has several limitations inherent to retrospective archive-based analyses. First, information on occupation type, duration, and intensity of exposure, smoking, alcohol use, comorbidities, medications, and anthropometrics was not available, limiting confounder adjustment. Second, laboratory testing was not protocolized. Thyroid function tests, HbA1c, lipid parameters, liver enzymes, and other biochemical tests were available only in clinically or occupationally tested subsets. This clinically driven testing pattern may have introduced selection bias if participants who underwent additional testing differed systematically from those who did not; therefore, the observed proportions should not be interpreted as population-level prevalence estimates. Third, because only one eligible record per individual was included, non-independence due to repeated measurements was minimized; however, the retrospective cross-sectional design still precluded evaluation of within-person longitudinal changes. Finally, detailed preanalytical and analytical metadata, including urine collection timing, instrument platform, calibration range, limits of detection, limits of quantification, and external proficiency-testing identifiers, could not be uniformly retrieved from the retrospective electronic archive. Accordingly, the observed associations should be interpreted as exploratory and do not establish causality. Despite these limitations, the large sample size provides a useful overview of occupational surveillance data in Türkiye. From a practical occupational surveillance perspective, the present findings highlight several pragmatic priorities: (i) routine endocrine biomarker monitoring (thyroid function, fasting glucose, HbA1c, and lipid profile) appears feasible and informative within existing periodic examination workflows, particularly for workers with metal measurements above interpretive limits; (ii) demographic distributions of exposure (younger age in lead, cadmium, and manganese strata; male predominance in arsenic, cadmium, and zinc strata) should inform targeted monitoring; and (iii) integration of standardized covariate capture (smoking, body mass index, occupation type, exposure duration) into electronic surveillance archives would substantially increase the analytic value of such datasets. The relatively young age of the screened workforce also underscores the importance of continued monitoring and exposure reduction to mitigate longer-term endocrine and metabolic complications. Future studies with detailed exposure assessment, longitudinal follow-up, and complete covariate information are warranted to clarify causal pathways and clinical relevance. Conclusions In this large occupational surveillance cohort, thyroid dysfunction and dysglycemia were frequent among tested participants, and values above interpretive limits for selected metals showed univariable associations with lipid and liver enzyme parameters. Age- and sex-adjusted sensitivity analyses attenuated some findings (notably zinc-related ALT and AST differences), while selected lead-related biomarker patterns and the zinc–triglyceride association persisted. Given the retrospective design, clinically driven testing, and adjustment limited to age and sex, these results should be interpreted as exploratory, hypothesis-generating surveillance signals rather than as evidence of causal or fully confounder-adjusted associations, and warrant further investigation in prospective studies with quantitative exposure assessment and complete covariate information. The patterns observed, together with prior evidence, suggest that occupational exposure to heavy metals may have implications for endocrine and metabolic health in relevant industries. Accordingly, structured occupational surveillance programs incorporating periodic assessment of thyroid hormones, fasting glucose, and HbA1c appear warranted as an integral component of occupational health practice in exposure-prone workforces. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3 Supplementary Material 4 Supplementary Material 4 Supplementary Material 5 Supplementary Material 5 Supplementary Material 6 Supplementary Material 6 Supplementary Material 7 Supplementary Material 7