Development and validation of a nomogram based on clinicopathological characteristics and multimodal ultrasound parameters for predicting lateral lymph node metastasis in papillary thyroid carcinoma.

Background: Tumor neovascularization and increased extracellular matrix stiffness have been confirmed to be crucial for oncology research, however, they are rarely integrated into diagnostic prediction models for predicting lateral cervical lymph node metastasis (LLNM). This study aimed to explore the correlation between these ultrasound parameters, clinicopathological characteristics and LLNM in papillary thyroid carcinoma (PTC), and construct a nomogram prediction model, as well as estimate its preoperative diagnosis values for LLNM. Methods: The clinical and ultrasound imaging data of 703 patients with postoperative histopathologically confirmed PTC were retrospectively analyzed. Conventional ultrasound, superb micro-vascular imaging (SMI) and strain ultrasound elastography (SUE) were performed for all patients, and they were stratified into training and validation cohorts based on the chronological sequence of surgery with a ratio of 7:3. Comprehensive evaluations of clinicopathological and ultrasonic features were conducted using least absolute shrinkage and selection operator (LASSO) regression and multivariate logistic regression, with the aim of identifying independent predictors of LLNM, and a nomogram prediction model was constructed. All LLNM patients were confirmed by postoperative pathology. Receiver operating characteristic curves (ROC) and calibration curves were drawn. Decision curve analysis (DCA) was performed to calculate the predictive efficiency, consistency, and clinical practicality of the model. Results: Among the 703 patients, 98 patients were diagnosed with LLNM (13.9%). According to the results of LASSO regression and multivariable logistic regression model, eight independent risk variables were screened to construct a prediction model, including sex, tumor size, multifocality, capsular invasion, microcalcification, perforator vessel, strain rate ratio (SRR) and ratio of metastatic central lymph nodes (LNR). The consistency index of the prediction model in the training cohort was 0.895, and it was 0.866 in the validation cohort. The optimal cutoff value (0.149) showed the balance between the sensitivity (80.6%) and specificity (84.5%). In both the training cohort and the validation cohort, the calibration curves were close to the standard curve, and the DCA curves showed that more than 90% of PTC patients could benefit from the prediction model. Conclusions: Compared with conventional imaging modalities used alone, the integrated application of novel ultrasonographic technologies, including SMI and SUE, demonstrates superior diagnostic performance in predicting LLNM in PTC patients. This nomogram incorporating the aforementioned ultrasound parameters might be helpful for accurate preoperative risk stratification of LLNM, thereby assisting surgeons in formulating individualized surgical strategies prior to intervention.