Patients’ recruitment, anthropometric, biochemical, and clinical variables were registered continuously in the electronic health register of Metabolic Diseases of the Department of Interdisciplinary Medicine—Internal Medicine Division—”Aldo Moro” University of Bari (Bari, Italy) from January 2018 to September 2022. Participants underwent physical examination, biochemical assessment, and thyroid ultrasound. The study was conducted in a group of 142 consecutive outpatients (47 men and 95 women) who were diagnosed with thyroid nodules and underwent ultrasound-guided FNAB/C due to suspicion of malignancy according to ultrasound characteristics (1). Patients with acute diseases and neoplastic diseases with recent onset (less than 10 years) and/or under chemotherapy were excluded. The study was approved by the Ethics Committee (n.311, MSC/PBMC/2015) of the Azienda Ospedaliero-Universitaria Policlinico di Bari (Bari, Italy) in accordance with the requirements of the Declaration of Helsinki. Written informed consent for the use of clinical data was obtained from all participants in the study. In accordance with the approval of the Ethics Committee, only patients who were already 18 years or older were included.

All participants underwent a detailed anamnesis and physical examination. Anthropometric assessment was performed using standardized procedures. To sum up, waist circumference (WC) was measured at the midpoint between the inferior part of the 12th costa and the anterior–superior iliac crest. Body mass index (BMI) was computed as weight (kg) divided by the height squared (m²), and BMI values (kg/m²) between 25 and 29.9 and over 30.0 were considered as overweight and obesity conditions, respectively. Morning blood samples were obtained from the antecubital veins after 12 hours of fasting. After blood clotting and centrifugation, serum was processed for analysis of biochemical markers of glucose and lipid metabolism. Liver and thyroid markers were also studied following standardized biochemical procedures. All biochemical measurements were centralized and performed in the ISO 9001-certified laboratories of the University Hospital of Bari. Diabetes mellitus (DM) was diagnosed according to international criteria: HbA1c (percentage of glycosylated hemoglobin) ≥6.5%, fasting plasma glucose (FPG) ≥126 mg/dl, and/or active treatment for DM. Non-invasive liver fibrosis scores were determined according to published formulas. AAR was calculated as AST divided by ALT. The cutoff value adopted was AAR < 1 for the identification of a fibrosis-free liver and AAR ≥ 1 for probable fibrosis. APRI score was calculated as AST/platelet count (×10⁹/L) × 100. The cutoff values were as follows: APRI < 0.5 for no fibrosis, APRI ≥ 0.5 for liver fibrosis, and APRI ≥ 1.5 for probable cirrhosis. FIB-4 index was calculated as age × AST/platelet count (×10⁹/L) × √ALT. The cutoff values adopted were as follows: FIB-4 < 1.45 for no or moderate fibrosis, 1.45 < FIB-4 < 3.25 for moderate fibrosis, and FIB-4 > 3.25 for extensive fibrosis or cirrhosis (15).

Thyroid ultrasound was performed with an Esaote MyLab 70 Gold ultrasound system with 10–15-MHz linear transducers. Thyroid nodules were defined according to international criteria as discrete lesions within the thyroid gland radiologically distinct from the surrounding thyroid parenchyma (1). The maximal diameter of nodules was measured in three planes: longitudinal, transverse, and anteroposterior. Fine-needle aspiration (FNA) was used after ultrasound for cytological diagnosis of nodules. The cytologic assessment was performed according to the SIAPEC-SIE 2014 guidelines. Malignancy was assumed in those diagnostic categories for which surgery was the suggested action (TIR3B, TIR4, and TIR5) (16).

Descriptive statistical analyses of the study sample were performed, and results were expressed as mean ± standard deviation (SD) for numerical data and as counts and percentages for categorical data. Comparisons of continuous clinical variables between two groups were performed by Student’s t-test.

The receiver operating characteristic (ROC) curves were used to determine the optimum cutoff level of AAR, APRI, BMI, and FPG in discriminating malignancy of thyroid nodules. Empirical ROC curves were plotted for these variables along with a calculation of the area under the curve (AUC) with 95% confidence intervals and one-sided upper p-values for the null hypothesis AUC = 0.5. Youden’s index, or equivalently, the highest sensitivity + specificity, was used to determine the optimal cutoff of each variable for the prediction of malignancy.

Contingency tables, chi-squared test, and Fisher’s exact test, if required, were used to study the association between categorical variables. The results were expressed as odds ratios with their relative 95% confidence interval (95% CI) and were graphically plotted in a forest plot.

Correlations among continuous variables were also analyzed and estimated using Spearman’s correlation coefficient (r). p-Values (p) lower than 0.05 were considered statistically significant. All analyses were performed using the NCSS 12 Statistical Software, version 12.0.2018 (NCSS, LLC Company, Kaysville, UT, USA) and GraphPad Prism, version 9.1.0 (GraphPad Software, San Diego, CA, USA).

A total of 142 subjects (47 men and 95 women) with putative risk of metabolic diseases underwent thyroid ultrasound exam and were diagnosed with suspected malignant thyroid nodules, thus FNAB/C was performed. After FNAB/C, 121 patients (85%; 39 men and 82 women) were diagnosed with benign thyroid nodules, while 21 individuals (15%) were diagnosed with thyroid cancer (8 men and 13 women). As shown in Table 1 , 106 (75%) patients had pathological WC according to the International Diabetes Federation (IDF) criterion for MetS diagnosis (17). A total of 71 patients (50%) were overweight (BMI ≥ 25 kg/m²), and 23 (16%) were obese (BMI ≥ 30 kg/m²). Moreover, 28 patients (20%) were affected by diabetes mellitus.

With the aim of identifying metabolic biomarkers associated with thyroid nodule malignancy, we compared groups of patients with benign and malignant nodules. First of all, we investigated whether sex could influence thyroid cancer susceptibility. We performed a chi-squared test finding that there was no significant association (p = 0.598, Supplementary Table 1 ). Age as well as WC did not significantly differ between the two groups, while patients with thyroid cancer exhibited increased BMI (p = 0.048) and FPG levels (p = 0.046). Lipid profile, 25-OH vitamin D, and thyroid function markers were not found to be statistically different between the two groups. Regarding transaminase levels, only AST was found to be significantly higher in the malignant group (p < 0.001) compared to subjects with benign nodules, albeit in the normal range of values. Intriguingly, considering non-invasive scores for liver fibrosis, subjects with thyroid cancer presented increased AAR (p < 0.001) and APRI scores (p = 0.007), while no difference was found for FIB-4 ( Table 2 ).

We performed ROC curve analyses of AAR, APRI score, BMI, and FPG to define cutoff values for discrimination of thyroid nodule malignancy.

The empirical ROC curve of AAR ( Figure 1A ) was characterized by the largest AUC as well as the highest Youden’s index (sensitivity = 52%, specificity = 87%), with a cutoff value of 1.12 (p < 0.01). APRI score ( Figure 1B ) presented intermediate sensitivity and high specificity (48% and 85%, respectively), with a significant AUC (p < 0.01) and a cutoff value of 0.45. The empirical ROC curve of BMI ( Figure 1C ) showed high specificity (83%) but lower sensitivity than AAR and APRI scores in discriminating thyroid cancer, with a cutoff value of 29.01 (p = 0.058). Despite very high specificity (90%), FPG levels showed the lowest sensitivity (33%) and the smallest AUC (p = 0.131) in discriminating between benign and malignant nodules, with a cutoff value of 103 mg/dl ( Figure 1D ). Overall, these data highlight the strong ability of AAR and APRI to accurately discriminate thyroid nodule malignancy.

To understand whether and to what degree dysmetabolic conditions increased the risk of thyroid nodule malignancy, we calculated the odds ratio (OR) of main biomarkers, stratifying our population according to both internationally validated cutoff values and cutoff values that have been calculated through the analysis of our ROC curves even if not significant ( Figure 2 ) (18).

Values of AAR above the international cutoff for the identification of liver fibrosis (AAR ≥ 1) indicate increased risk of malignancy with an OR of 2.13 (95% CI: 1.2–3.8), although the association in the chi-squared test was not significant (p = 0.077), while AAR values above the ROC cutoff of 1.12 showed the highest OR (7.1, 95% CI: 2.4–18.5), with a significant association (p < 0.001). Considering APRI, values above the international cutoff of 0.5 and the ROC cutoff of 0.45 were both associated (p < 0.05) with malignancy, with ORs of 6.6 (95% CI: 2.0–20.0) and 5 (95% CI: 1.9–12.6), respectively.

Regarding BMI, overweight condition (BMI ≥ 25 kg/m²) was not associated with thyroid cancer (p = 0.345). Conversely, systemic obesity (BMI ≥ 30 kg/m²) was associated (p = 0.047) with a risk of nodule malignancy (OR = 3.3, 95% CI: 1.2–8.9). Interestingly, an even higher risk (OR = 3.8, 95% CI: 1.5–10.5) was associated (p = 0.015) with BMI values above the ROC cutoff of 29.01 kg/m². Surprisingly, visceral obesity assessed by WC was not associated (p = 0.785) with the risk of thyroid cancer in our study population.

Finally, FPG levels above the ROC cutoff value of 103 mg/dl were also found to significantly (p = 0.01) increase the risk of malignant nodules (OR = 4.5, 95% CI: 1.6–12.8).

To further disclose how dysmetabolic conditions could affect the nature of thyroid nodules, we performed correlations among AAR, APRI, and biochemical metabolic biomarkers in patients with benign nodules and with cancer, separately ( Figure 3 ).

Regarding AAR and APRI, in the benign group, they both significantly (p < 0.05) correlated with FPG, although inversely (r = −0.22) and directly (r = 0.23), respectively. These results are in line with previous studies in subjects with a risk of diabetes (19, 20). Intriguingly, these correlations were both lost when considering cancer patients, suggesting that a metabolic derangement could be present in patients with cancer. To further investigate the other pathways that may be involved in cancer pathogenesis, we performed correlations among the other metabolic biomarkers. We found significant correlations of BMI with FPG ( Figure 4A ), WC ( Figure 4B ), and high-density lipoprotein cholesterol (HDL-c) ( Figure 4C ) in subjects with benign nodules, as well as in those with malignant ones. Furthermore, HDL-c showed statistically significant correlations with FPG ( Figure 4D ) and WC ( Figure 4E ) only in subjects with benign nodules, while HDL-c levels were observed to significantly correlate with BMI ( Figure 4F ) in both groups. Conversely, 25-OH vitamin D significantly correlated with FPG ( Figure 4G ), WC ( Figure 4H ), and HDL-c ( Figure 4I ) only in patients with malignant nodules, thus suggesting that its deficiency could play a crucial role in the development of thyroid malignancy.