Development and validation of a preoperative nomogram for predicting contralateral occult carcinoma and central lymph...

Thyroid cancer is one of the fastest-growing malignancies worldwide, with papillary thyroid carcinoma (PTC) accounting for approximately 85-90% of all cases. For patients with T1b-T2 PTC, the choice between thyroid lobectomy and total thyroidectomy remains controversial. Postoperative pathological examination reveals contralateral occult carcinoma in 20-40% of patients, while contralateral central lymph node metastasis (CLNM) represents a key prognostic factor. However, reliable and practical preoperative tools to support individualized surgical decision-making are currently lacking. This study aimed to develop and validate a preoperative prediction model for assessing the risk of contralateral occult carcinoma (OTC) and contralateral CLNM in patients with T1b-T2 PTC, thereby facilitating individualized surgical management.

This retrospective study enrolled 932 patients with T1b-T2 PTC who presented with unilateral nodules on preoperative ultrasound and underwent total thyroidectomy with bilateral central lymph node dissection. All 932 patients were included in the contralateral OTC prediction model, and 730 patients were included in the contralateral CLNM prediction model. Nine machine learning algorithms and logistic regression models were trained and compared using five-fold cross-validation. Key predictive features were selected through multivariate logistic regression and random forest. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration curves, the Hosmer Lemeshow test, and decision curve analysis.

Feature selection identified four independent predictors of contralateral OTC: malignant isthmic nodule, sum of the longest diameter (SLD), number of malignant nodules, and capsular disruption. Five predictors were identified for contralateral CLNM: isthmus adjacent nodule, sex, SLD, age, and chronic lymphocytic thyroiditis (CLT), with CLT showing a negative association with CLNM.Logistic regression model demonstrated the best performance for OTC prediction (AUC = 82.00%). For contralateral CLNM, logistic regression achieved an AUC of 79.00%, comparable to that of the best-performing machine learning model (glmBoost, AUC = 79.10%), with no statistically significant difference observed between the two models. Model calibration was satisfactory, as demonstrated by calibration curves and the Hosmer-Lemeshow test (P = 0.5097). Decision curve analysis further indicated favorable clinical net benefit.

This study developed a logistic regression-based preoperative predictive tool to assess the individualized risk of contralateral occult carcinoma and contralateral CLNM in patients with T1b-T2 PTC. To our knowledge, this is the first dual-outcome prediction model specifically designed for this patient population. The prediction results were visualized using two nomograms and two web-based applications, which may assist clinicians in making individualized surgical decisions before surgery. Specifically, low-risk patients may be considered for thyroid lobectomy to avoid overtreatment, whereas high risk-patients may undergo total thyroidectomy to reduce the risk of recurrence.