Abstract
BACKGROUND: The preoperative assessment of lymph node metastasis (LNM) in papillary thyroid carcinoma (PTC) determines the surgical approach adopted for patients. Central lymph nodes are the most common site of metastasis and pose significant evaluation challenges. This study aimed to identify the preoperative multiparametric ultrasound (MULTI-US) risk factors for predicting central lymph node metastasis (CLNM) in PTC. METHODS: This retrospective study included 764 PTC patients with CLNM from our institution, who were randomly divided into a training set (n=534) and test set (n=230) at a ratio of 7:3. Univariable and multivariable analyses were conducted to identify significant predictors from the MULTI-US features, including B-mode, color Doppler imaging, contrast-enhanced ultrasound, and shear wave elastography. A MULTI-US model was constructed as a nomogram to predict CLNM risk. The diagnostic performance and clinical utility of the model were evaluated by receiver operating characteristic curve analysis and decision curve analysis (DCA). RESULTS: Our study identified extrathyroidal extension (ETE) [2.175, 95% confidence interval (CI): 1.317-3.583; P=0.002], multifocality (2.040, 95% CI: 1.356-3.068; P<0.001), macrocalcifications (5.139, 95% CI: 2.118-12.471; P<0.001), clustered microcalcifications (6.926, 95% CI: 2.646-18.133, P<0.001), hypo-enhancement (3.405, 95% CI: 1.202-8.898, P=0.012), and the elasticity maximum value (1.097, 95% CI: 1.042-1.153, P=0.006) as significant independent predictors of CLNM. The MULTI-US model demonstrated superior predictive performance, with area under the curve (AUC) values of 0.780 (95% CI: 0.741-0.819) in the training set and 0.737 (95% CI: 0.692-0.807) in the test set. The DCA showed the high clinical applicability of the MULTI-US model. A comparison of the AUC values of the MULTI-US model for different tumor sizes revealed no significant differences between the tumors <10 and ≥10 mm in diameter (P=0.410). CONCLUSIONS: The nomogram based on the MULTI-US model showed potential in the preoperative risk stratification of CLNM. This model may serve as a useful clinical method for improving PTC management.