Nomogram for predicting the risk of cervical lymph node metastases and recurrence in papillary thyroid carcinoma based on the thyroid differentiation score system and clinical characteristics.

Abstract

Background: This study aimed to identify independent predictors of cervical central lymph node metastasis (CLNM), cervical lateral lymph node metastasis (LLNM), and recurrence in patients with PTC, which could help guide the surgical management of these patients.

Methods: This retrospective study analyzed data from 542 patients with PTC, who underwent thyroid surgery and were enrolled in The Cancer Genome Atlas and Gene Expression Omnibus databases. Patients were categorized into two groups based on the presence or absence of cervical LNM, classified as CLNM or LLNM. Data were randomly partitioned into training and validation sets in a ratio of 7:3. Age, sex, thyroid differentiation score (TDS), and other relevant attributes, were compared between the two groups using univariate and multivariate analyses and reported as odds ratios (OR) or hazard ratios (HR) with 95% confidence intervals (CI). Independent predictors were identified and used to develop nomograms. To assess the accuracy, discrimination, and clinical utility of the prediction model, calibration, receiver operating characteristic (ROC), and decision curve analysis (DCA) were performed for both the training and validation sets.

Results: Of the 542 patients, 261 (48.15%) and 130 (23.99%) presented with CLNM and LLNM, respectively. The analyses identified several independent predictors for CLNM, including the presence of extrathyroidal invasion (OR 2.53, 95% CI 1.60-4.00), larger tumor dimension (OR 1.17, 95% CI 1.02-1.34), age over 55 years (OR 0.52, 95% CI 0.33-0.82), non-classic papillary subtype (OR 0.38, 95% CI 0.23-0.61), and lower TDS (OR 0.50, 95% CI 0.33-0.76). A greater number of excised cervical LNs (OR 12.30, 95% CI 4.35-34.77), the presence of CLNM (OR 1.07, 95% CI 1.04-1.10), and lower TDS (OR 0.09, 95% CI 0.04-0.21) were independent predictors for LLNM. Additionally, the independent predictors for relapse included age greater than 55 years (HR 1.87, 95% CI 1.00-3.49) and lower TDS (HR 0.35, 95% CI 0.20-0.62). These predictors were used to develop nomograms for CLNM, LLNM, and recurrence. ROC and DCA confirmed the discrimination and clinical utility of the models.

Conclusions: This study identified independent predictors of cervical CLNM, LLNM, and recurrence. Clinically relevant nomograms were developed that can assist in guiding cervical lymph node dissection and prediction of recurrence in patients with PTC.