Development and performance evaluation of a risk assessment model for cervical lymph node metastasis in patients with Hashimoto’s thyroiditis coexisting with papillary thyroid carcinoma

Introduction

In recent years, the incidence of concurrent Hashimoto’s thyroiditis (HT) and papillary thyroid carcinoma (PTC) has been steadily increasing, with a coexistence rate of approximately 10–58% (1). However, the risk of cervical lymph node metastasis (CLNM) in patients with HT and PTC remains controversial. Some studies suggest that HT exerts a protective effect on PTC prognosis, with fewer cases of extrathyroidal extension (ETE) and distant metastasis (1-3); conversely, other studies indicate that the persistent accumulation of cytokines under chronic inflammatory conditions in HT may increase the risk of CLNM in PTC (4-7). HT is characterized by extensive lymphocytic infiltration and reactive lymphadenopathy, often leading to enlargement of multiple cervical lymph nodes, with ultrasound features overlapping with those of metastatic lymph nodes (e.g., loss of hilum, cortical thickening), thereby posing challenges for accurate preoperative diagnosis of CLNM in the presence of both diseases. Currently, ultrasound remains the first-line modality for CLNM evaluation, but conventional ultrasound has limited sensitivity and diagnostic performance (8), and a single parameter is insufficient to meet clinical demands.

To improve the accuracy of preoperative CLNM assessment, several research teams have developed risk assessment models for patients with HT and PTC. Wang et al. (9) constructed a clinical-ultrasound nomogram incorporating age, thyroglobulin antibody (TGAb), tumor size, punctate hyperechogenicity, and vascularity grade [area under the curve (AUC) =0.76], but its specificity was only 51%, and lymph node ultrasound features (LN-US) were not included. Chen et al. (10) developed a comprehensive prediction model based on sex, maximum tumor diameter, multifocality, margin, and thyrotropin receptor antibody (TRAb) (AUC =0.82), which likewise lacked LN-US. In summary, existing models have the following limitations: insufficient integration of LN-US; inadequate combination of clinical, serological, and ultrasound parameters; and a lack of systematic comparisons of diagnostic performance among different feature combinations.

Therefore, this study aims to integrate multidimensional data including clinical characteristics, serum biomarkers, and ultrasound imaging features, construct risk assessment models with different feature combinations, and compare their diagnostic performance to identify the optimal model for evaluating CLNM risk in patients with PTC coexisting with HT, thereby providing a more reliable basis for preoperative risk assessment and surgical decision-making. We present this article in accordance with the TRIPOD reporting checklist (available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0205/rc).

Methods

Patients and methods

This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Medical Science Research Ethics Committee of the Affiliated Hospital of Guizhou Medical University (approval No. 2022, ethics review No. 346). Due to the retrospective design, written informed consent was waived. A total of 261 patients who underwent thyroid surgery at the Affiliated Hospital of Guizhou Medical University from January 2018 to December 2023 and were postoperatively pathologically diagnosed with HT coexisting with PTC were retrospectively enrolled. All patients completed preoperative thyroid and cervical lymph node ultrasonography. Based on postoperative pathological results, patients were divided into a lymph node metastasis group and a non-metastasis group (Figure 1).

Figure 1 Flowchart of patient selection for differentiating CLNM of PTC patients with HT. CLNM, cervical lymph node metastasis; HT, Hashimoto’s thyroiditis; PTC, papillary thyroid carcinoma.

Inclusion criteria: (I) age 18–80 years; (II) complete preoperative thyroid and cervical lymph node ultrasound and serological examinations [including alkaline phosphatase (ALP), hemoglobin (Hb), thyroid function, and related antibodies]; (III) thyroid nodules classified as ≥ Chinese Thyroid Imaging Reporting and Data System (C-TIRADS) 4a according to the 2020 Chinese Guidelines for Ultrasound Malignancy Risk Stratification of Thyroid Nodules (11); (IV) lymph node metastasis status confirmed by postoperative pathology as the gold standard, with negative results determined by combining postoperative pathology and preoperative fine-needle aspiration cytology; in case of discrepancy, postoperative pathology (regional lymph node dissection/biopsy) was considered definitive.

Exclusion criteria: (I) incomplete clinical or ultrasound data; (II) history of neck surgery, radiotherapy, or chemotherapy; (III) postoperative pathology confirming a type of thyroid cancer other than PTC; (IV) concurrent Graves’ disease.

Instruments and methods

Ultrasound examination

Ultrasound examinations were performed using Philips iU22, EPIQ5 (L12-5 probe, 5–12 MHz), and Mindray Resona R9 (L14-3 probe, 3–14 MHz) color Doppler ultrasound systems. Patients were placed in the supine position with the neck adequately exposed. Routine multi-plane scanning of the thyroid and cervical lymph nodes was conducted, and ultrasound features of target nodules and lymph nodes were recorded.

Data collection

Three categories of data were collected:

• Clinical information: sex, age, body mass index (BMI).
• Laboratory indicators: preoperative thyroid function and related antibodies [thyroid peroxidase antibody (TPOAb), TGAb], ALP (reference range, 35–100 U/L), Hb (115–150 g/L), among others.

Ultrasound features: location, margin, aspect ratio, microcalcifications, capsular invasion, vascularity grade [Adler semi-quantitative grading (12): grade 0–1 = not abundant, grade 2–3 = abundant], and ETE (defined as tumor destruction of the thyroid capsule or invasion of adjacent tissues). In cases of multiple nodules, imaging features of the largest tumor were analyzed. Cervical LN-US included location, long‑toshort axis ratio (≤2), hilum, cortical echogenicity, microcalcifications, cystic change, and vascularity (Figure S1).

Quality control

To reduce selection bias, all consecutive patients meeting the eligibility criteria during the study period were enrolled. To minimize information bias, data extraction was performed using a standardized form, and ultrasound features were independently evaluated by two physicians (with >5 years of experience); disagreements were resolved by consensus. No formal inter-rater reliability test (e.g., Kappa) was performed, which is acknowledged as a limitation.

Sample size

The final model in this study included seven predictors, and the number of CLNM-positive events was 141. Thus, events per variable (EPV) =141/7≈20.1.

Statistical analysis

Statistical analyses were performed using SPSS 25.0 and Stata 12 software. Continuous variables were expressed as mean ± standard deviation, and intergroup comparisons were conducted using the Mann-Whitney U test. Categorical variables were presented as number (%), and group comparisons were performed using the Chi-squared test or Fisher’s exact test (mean imputation for continuous variables, mode imputation for binary variables). Univariate and multivariate logistic regression analyses were used to identify independent predictors of CLNM. Features with P<0.05 in the multivariate analysis were incorporated into model development. Four models were constructed: based on tumor nodule ultrasound features (TN-US), LN-US, their combination (TN-LN-US), and a full-feature model integrating clinical, laboratory, and ultrasound parameters (Clin-US). Receiver operating characteristic (ROC) curves were plotted, and the AUC was compared using the DeLong test. The model with the best diagnostic performance was selected based on AUC, specificity, and sensitivity. Internal validation was performed using the bootstrap method. Model performance was further evaluated using calibration curves, decision curve analysis (DCA), and clinical impact curve (CIC). P value <0.05 was considered statistically significant.

Results

Baseline characteristics

This study included 261 patients with PTC coexisting with HT, including 141 patients (54.0%) in the lymph node metastasis group and 120 patients (46.0%) in the non-metastasis group. There were 223 females (85.4%) and 38 males (14.6%). The mean age was 37.1±9.81 years in the metastasis group and 42.5±11.5 years in the non-metastasis group. Univariate analysis showed that several factors were significantly associated with CLNM in patients with PTC and HT (P<0.05) (Table 1).

Logistic regression analysis

Based on the univariate analysis, variables with statistical significance were included in multivariate logistic regression analysis. The results showed that age [odds ratio (OR) =0.953, P<0.05], ALP (OR =0.983, P<0.05), tumor calcification (OR =3.697, P<0.05), tumor vascularity (OR =3.005, P<0.05), lymph node morphology (OR =3.908, P<0.05), lymph node cortical echogenicity (OR =0.105, P<0.05), and lymph node calcification (OR =0.240, P<0.05) were independent predictors of CLNM. Features with P<0.05 in the multivariate analysis were included in subsequent model development (Table 2).

Model development and comparison of diagnostic performance

To evaluate the predictive value of different feature combinations for CLNM, four models were constructed in this study: TN-US, LN-US, TN-LN-US, and Clin-US. ROC curves were used to assess the AUC, sensitivity, specificity, and 95% confidence interval of each model (Table 3, Figure 2). The DeLong test results showed that the Clin-US model had a significantly higher AUC than the other combined models and individual parameters (P<0.001), with an AUC of 0.842, sensitivity of 70.92%, and specificity of 84.17%, indicating optimal diagnostic performance. Internal validation using the bootstrap method (1,000 resampling iterations) yielded a 95% confidence interval of 0.792–0.885 for this model (Figure 3).

Figure 2 Comparison of diagnostic efficacy for different grouping models. ALP, alkaline phosphatase; Clin-US, a full-feature model integrating clinical, laboratory, and ultrasound parameters; LN-US, lymph node ultrasound features; TN-LN-US, the combination of TN-US and LN-US; TN-US, tumor nodule ultrasound features.

Figure 3 Internal validation using the bootstrap method (1,000 resampling iterations). AUC, area under the curve; Clin-US, a full-feature model integrating clinical, laboratory, and ultrasound parameters; ROC, receiver operating characteristic

Model performance evaluation

The calibration curve showed good agreement between the predicted probability of CLNM and the observed incidence (Figure 4A). The Hosmer‑Lemeshow test yielded a P value of 0.883, further confirming satisfactory model calibration. DCA demonstrated that the model provided a positive net benefit across a risk threshold range of 0.20–0.90, indicating its clinical utility over a wide range of thresholds (Figure 4B). The CIC revealed good consistency between the high‑risk population identified by the model and the actual number of CLNM cases (Figure 4C). Overall, the Clin‑US model exhibited favorable risk stratification capability and clinical applicability across different thresholds.

Figure 4 Diagnostic performance of Clin-US models in HT-PTC patients. (A) Calibration curve; (B) decision curve analysis; (C) clinical impact curve. Clin-US, a full-feature model integrating clinical, laboratory, and ultrasound parameters; HT, Hashimoto’s thyroiditis; PTC, papillary thyroid carcinoma.

Discussion

Based on clinical information, laboratory indicators, and ultrasound features, this study constructed four predictive models to evaluate the risk of CLNM in patients with HT coexisting with PTC. The results showed that the Clin-US model demonstrated the best diagnostic performance (AUC =0.842), with a sensitivity of 70.92% and specificity of 84.17%. Previous models often lacked adequate integration of LN-US, insufficient combination of clinical, serological, and ultrasound parameters, and systematic comparisons of diagnostic performance among different feature combinations. Wang et al. (9) developed a clinical‑ultrasound nomogram (AUC =0.76) with a specificity of only 51%, and did not include LN-US. The comprehensive prediction model by Chen et al. (10) (AUC =0.82) similarly lacked sufficient integration of LN-US. The model by Lin et al. (13) was applicable only to cN0 stage papillary thyroid microcarcinoma patients. On this basis, the Clin-US model in our study fully incorporated clinical data and LN-US, and its superiority was validated through multi‑model comparisons.

Our results show that age is an important influencing factor for CLNM, with younger patients being more susceptible to lymph node metastasis, which is consistent with previous reports (14,15). This may be related to the more aggressive biological behavior of tumors in younger patients. Therefore, more active treatment strategies and enhanced follow-up surveillance may be necessary for younger patients.

Contrary to previous studies, the present study found that decreased serum ALP level was significantly associated with an increased risk of CLNM in patients with HT coexisting with PTC. Previous studies have reported that ALP is highly expressed in thyroid carcinoma tissues, particularly in older patients, tall cell variant, and those with poor prognosis (16,17). We speculate that the reasons for these discrepancies may include: first, differences in study populations—this study focused on patients with PTC coexisting with HT, whereas previous studies predominantly included patients with PTC alone; second, differences in the types of specimens tested. Additionally, the source heterogeneity and functional diversity of ALP isoenzymes may also play an important role. We believe that these differences are not contradictory but rather complement the understanding of the complex biological role of ALP in thyroid cancer from different perspectives. Further studies with larger sample sizes are needed to validate the underlying mechanisms.

Regarding primary tumor-related features, some studies have reported a positive correlation between CLNM and nodule size (18-20), which is consistent with our findings. This may be attributed to the secretion of vascular endothelial growth factor by malignant nodules, which stimulates angiogenesis, thereby accelerating nodule growth and promoting further invasion. Regarding other primary tumor features, microcalcifications and abundant vascularity were identified as independent risk factors for CLNM. Hypoxia and necrosis within the tumor microenvironment contribute to the formation of calcifications, and these environmental changes accelerate tumor cell invasiveness and metastatic potential (21). Abundant vascularity increases intratumoral neovascularization, which on one hand enhances contact between actively proliferating cells and lymphatic vessels, and on the other hand increases the number of lymphaticvenous connections, thereby promoting lymphatic metastasis (22). Notably, ETE did not differ significantly between the two groups in this study, which differs from the findings of Mao (23). Possible explanations include the limited sample size, the study population restricted to PTC patients with coexisting HT (whose biological behavior may differ from that of PTC alone), and the fact that chronic inflammationinduced interstitial fibrosis in the context of HT may penetrate the true thyroid capsule, resulting in false ETE that can be easily confused with true invasion (24).

Regarding LN-US, abnormal lymph node morphology, hyperechoic cortex, and microcalcifications were identified as independent risk factors for CLNM, consistent with previous studies (25). The researchers suggest that hyperechoic cortex may be attributed to thyroglobulin deposition (26-28); microcalcifications mainly reflect psammoma bodies and serve as strong evidence of lymph node metastasis; metastatic lymph nodes may exhibit capsular disruption and irregular margins due to tumor cell infiltration and destruction of reticular fibers (29). In this study, the specificity of cortical hyperechogenicity and microcalcifications reached 95% and 84.17%, respectively, and the specificity of abnormal lymph node morphology was 88.33%, indicating that preoperative ultrasound should emphasize comprehensive analysis of multiple signs to avoid misjudgment based on a single parameter.

Currently, most models rely on a single examination modality. In this study, we integrated multidimensional routine preoperative parameters, including clinical, laboratory, and ultrasound features, to construct the Clin‑US model. The calibration curve showed excellent agreement between the predicted probabilities and the actual incidence. DCA and CIC indicated that the net benefit of this model was superior to either the strategy of total lymph node dissection or complete observation. In summary, the Clin‑US model demonstrates favorable diagnostic performance for assessing CLNM risk in patients with PTC coexisting with HT, facilitating preoperative individualized risk assessment.

This study has several limitations. First, the single‑center retrospective design and the relatively small sample size may introduce selection bias and information bias. Second, the assessment of ultrasound features is somewhat subjective; although quality control measures were implemented, inter‑observer variability may still have an impact. Third, the use of fine‑needle aspiration cytology as one of the reference standards is susceptible to variations in puncture technique and the pathologist’s judgment, potentially introducing bias. Fourth, the model underwent only internal validation without multi‑center external validation, so caution is needed when extrapolating the results, and there is a risk of overfitting. Nevertheless, the ultrasound features and clinical variables incorporated in this study are all routinely available in clinical practice, which endows the model with certain potential for generalization. Future prospective, multi‑center external validation studies are needed to further evaluate the robustness of this model.

Conclusions

Age, ALP, primary tumor microcalcification, primary tumor vascularity, lymph node morphology, lymph node cortical echogenicity, and lymph node microcalcification were independent influencing factors for CLNM in patients with PTC coexisting with HT. Compared with the other models, the Clin-US model demonstrated promising diagnostic performance in predicting lymph node metastasis in PTC patients with coexisting HT, which may assist clinicians in preoperative risk assessment for this patient population.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0205/rc

Data Sharing Statement: Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0205/dss

Peer Review File: Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0205/prf

Funding: The study was supported by the National Natural Science Foundation of China (No. 82260160), Science and Technology Fund Project of the Health Commission of Guizhou Province (No. gzwkj2021-376), and Science and Technology Planning Project of Guizhou Province (No. ZK [2022] General Program 440).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0205/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Medical Science Research Ethics Committee of Affiliated Hospital of Guizhou Medical University (approval No. 2022 Ethics Review No. 346). Due to the retrospective design, written informed consent was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Xu S, Huang H, Qian J, et al. Prevalence of Hashimoto Thyroiditis in Adults With Papillary Thyroid Cancer and Its Association With Cancer Recurrence and Outcomes. JAMA Netw Open 2021;4:e2118526. [Crossref] [PubMed]
2. Antonaci A, Consorti F, Mardente S, et al. Clinical and biological relationship between chronic lymphocytic thyroiditis and papillary thyroid carcinoma. Oncol Res 2009;17:495-503. [Crossref] [PubMed]
3. Marotta V, Sciammarella C, Chiofalo MG, et al. Hashimoto's thyroiditis predicts outcome in intrathyroidal papillary thyroid cancer. Endocr Relat Cancer 2017;24:485-93. [Crossref] [PubMed]
4. Hu X, Wang X, Liang Y, et al. Cancer Risk in Hashimoto's Thyroiditis: a Systematic Review and Meta-Analysis. Front Endocrinol (Lausanne) 2022;13:937871. [Crossref] [PubMed]
5. Xu SJ, Jin B, Zhao WJ, et al. The Specifically Androgen-Regulated Gene (SARG) Promotes Papillary Thyroid Carcinoma (PTC) Lymphatic Metastasis Through Vascular Endothelial Growth Factor C (VEGF-C) and VEGF Receptor 3 (VEGFR-3) Axis. Front Oncol 2022;12:817660. [Crossref] [PubMed]
6. Kobawala TP, Trivedi TI, Gajjar KK, et al. Significance of TNF-α and the Adhesion Molecules: L-Selectin and VCAM-1 in Papillary Thyroid Carcinoma. J Thyroid Res 2016;2016:8143695. [Crossref] [PubMed]
7. Wang Y, Zong H, Zhou H. Circular RNA circ_0062389 modulates papillary thyroid carcinoma progression via the miR-1179/high mobility group box 1 axis. Bioengineered 2021;12:1484-94. [Crossref] [PubMed]
8. Yu XQ, Fu XH, Shen Y, et al. Analysis of related factors of ultrasonic features in predicting cervical lymph node metastasis of papillary thyroid microcarcinoma. Chinese Journal of Ultrasound in Medicine 2023;39:1212-5.
9. Wang L, Zhang L, Wang D, et al. Predicting central cervical lymph node metastasis in papillary thyroid carcinoma with Hashimoto's thyroiditis: a practical nomogram based on retrospective study. PeerJ 2024;12:e17108. [Crossref] [PubMed]
10. Chen Y, Zhao S, Zhang Z, et al. A comprehensive prediction model for central lymph node metastasis in papillary thyroid carcinoma with Hashimoto's thyroiditis: BRAF may not be a valuable predictor. Front Endocrinol (Lausanne) 2024;15:1429382. [Crossref] [PubMed]
11. Zhou J, Yin L, Wei X, et al. 2020 Chinese guidelines for ultrasound malignancy risk stratification of thyroid nodules: the C-TIRADS. Endocrine 2020;70:256-79. [Crossref] [PubMed]
12. Huang J, Zhao J. Quantitative Diagnosis Progress of Ultrasound Imaging Technology in Thyroid Diffuse Diseases. Diagnostics (Basel) 2023;13:700. [Crossref] [PubMed]
13. Lin Y, Cui N, Li F, et al. The model for predicting the central lymph node metastasis in cN0 papillary thyroid microcarcinoma with Hashimoto's thyroiditis. Front Endocrinol (Lausanne) 2024;15:1330896. [Crossref] [PubMed]
14. Yang Z, Heng Y, Lin J, et al. Nomogram for Predicting Central Lymph Node Metastasis in Papillary Thyroid Cancer: A Retrospective Cohort Study of Two Clinical Centers. Cancer Res Treat 2020;52:1010-8. [Crossref] [PubMed]
15. Zhao L, Sun X, Luo Y, et al. Clinical and pathologic predictors of lymph node metastasis in papillary thyroid microcarcinomas. Ann Diagn Pathol 2020;49:151647. [Crossref] [PubMed]
16. Zheng K, Mao J. Comparison and Analysis of Clinical Features of Papillary Thyroid Cancer Complicated With Hashimoto's Thyroiditis. Clin Med Insights Oncol 2024;18:11795549241287085. [Crossref] [PubMed]
17. Liu WC, Li MP, Hong WY, et al. A practical dynamic nomogram model for predicting bone metastasis in patients with thyroid cancer. Front Endocrinol (Lausanne) 2023;14:1142796. [Crossref] [PubMed]
18. Zhao W, He L, Zhu J, et al. A nomogram model based on the preoperative clinical characteristics of papillary thyroid carcinoma with Hashimoto's thyroiditis to predict central lymph node metastasis. Clin Endocrinol (Oxf) 2021;94:310-21. [Crossref] [PubMed]
19. Zhong M, Zhang Z, Xiao Y, et al. The Predictive Value of ACR TI-RADS Classification for Central Lymph Node Metastasis of Papillary Thyroid Carcinoma: A Retrospective Study. Int J Endocrinol 2022;2022:4412725. [Crossref] [PubMed]
20. Liu C, Xiao C, Chen J, et al. Risk factor analysis for predicting cervical lymph node metastasis in papillary thyroid carcinoma: a study of 966 patients. BMC Cancer 2019;19:622. [Crossref] [PubMed]
21. Wang JZ, Zhu W, Han J, et al. The role of the HIF-1α/ALYREF/PKM2 axis in glycolysis and tumorigenesis of bladder cancer. Cancer Commun (Lond) 2021;41:560-75. [Crossref] [PubMed]
22. Rosario PW, Côrtes MCS, Franco Mourão G. Follow-up of patients with thyroid cancer and antithyroglobulin antibodies: a review for clinicians. Endocr Relat Cancer 2021;28:R111-9. [Crossref] [PubMed]
23. Mao J, Zhang Q, Zhang H, et al. Risk Factors for Lymph Node Metastasis in Papillary Thyroid Carcinoma: A Systematic Review and Meta-Analysis. Front Endocrinol (Lausanne) 2020;11:265. [Crossref] [PubMed]
24. Dioufa N, Baloch ZW. Encapsulated neoplasms of the thyroid gland. Virchows Arch 2026;488:95-111. [Crossref] [PubMed]
25. Ke X, Shen H, Lv G, et al. Construction and Value of Ultrasonic Risk Prediction Model for Benign and Malignant Cervical Lymph Nodes. Chinese Journal of Ultrasound in Medicine 2020;36:314-7.
26. Ahuja A, Ying M. Sonography of neck lymph nodes. Part II: abnormal lymph nodes. Clin Radiol 2003;58:359-66.
27. Jia X, Wang Y, Liu Y, et al. Thyroglobulin Measurement Through Fine-Needle Aspiration for Optimizing Neck Node Dissection in Papillary Thyroid Cancer. Ann Surg Oncol 2022;29:88-96. [Crossref] [PubMed]
28. Chung SR, Baek JH, Rho YH, et al. Sonographic Diagnosis of Cervical Lymph Node Metastasis in Patients with Thyroid Cancer and Comparison of European and Korean Guidelines for Stratifying the Risk of Malignant Lymph Node. Korean J Radiol 2022;23:1102-11. [Crossref] [PubMed]
29. Randolph GW, Duh QY, Heller KS, et al. The prognostic significance of nodal metastases from papillary thyroid carcinoma can be stratified based on the size and number of metastatic lymph nodes, as well as the presence of extranodal extension. Thyroid 2012;22:1144-52. [Crossref] [PubMed]