A comprehensive prediction model for central lymph node metastasis in papillary thyroid carcinoma with Hashimoto's thyroiditis: BRAF may not be a valuable predictor
Yanwei Chen,Shuangshuang Zhao,Zheng Zhang,Zheming Chen,Bingxin Jiang,Maohui An,Mengyuan Shang,Xincai Wu,Xin Zhang,Baoding Chen
DOI: https://doi.org/10.3389/fendo.2024.1429382
2024-09-19
Abstract:Purpose: Papillary thyroid carcinoma (PTC) frequently coexists with Hashimoto's thyroiditis (HT), which poses challenges in detecting central lymph node metastasis (CLNM) and determining optimal surgical management. Our study aimed to identify the independent predictors for CLNM in PTC patients with HT and develop a comprehensive prediction model for individualized clinical decision-making. Patients and methods: In this retrospective study, a total of 242 consecutive PTC patients who underwent thyroid surgery and central lymph node dissection between February 2019 and December 2021 were included. 129 patients with HT were enrolled as the case group and 113 patients without HT as control. The results of patients' general information, laboratory examination, ultrasound features, pathological evaluation, and BRAF mutation were collected. Multivariate logistic regression analysis was used to identify independent predictors, and the prediction model and nomogram were developed for PTC patients with HT. The performance of the model was assessed using the receiver operating characteristic curve, calibration curve, decision curve analysis, and clinical impact curve. In addition, the impact of the factor BRAF mutation was further evaluated. Results: Multivariate analysis revealed that gender (OR = 8.341, P = 0.013, 95% CI: 1.572, 44.266), maximum diameter (OR = 0.316, P = 0.029, 95% CI: 0.113, 0.888), multifocality (OR = 3.238, P = 0.010, 95% CI: 1.319, 7.948), margin (OR = 2.750, P = 0.046, 95% CI: 1.020, 7.416), and thyrotropin receptor antibody (TR-Ab) (OR = 0.054, P = 0.003, 95% CI: 0.008, 0.374) were identified as independent predictors for CLNM in PTC patients with HT. The area under the curve of the model was 0.82, with accuracy, sensitivity, and specificity of 77.5%, 80.3% and 75.0%, respectively. Meanwhile, the model showed satisfactory performance in the internal validation. Moreover, the results revealed that BRAF mutation cannot further improve the efficacy of the prediction model. Conclusion: Male, maximum diameter > 10mm, multifocal tumors, irregular margin, and lower TR-Ab level have significant predictive value for CLNM in PTC patients with HT. Meanwhile, BRAF mutation may not have a valuable predictive role for CLNM in these cases. The nomogram constructed offers a convenient and valuable tool for clinicians to determine surgical decision and prognostication for patients.