Prediction model for lymph node metastasis in the right lateral region of papillary thyroid carcinoma

To compare the differences among patients with papillary thyroid carcinoma (PTC) whose right lateral lymph node biopsy results indicating positive (pN1b) or negative lymph node metastasis (pN0 or pN1a) so as to explore whether there is a high risk of long-term metastasis in right lateral lymph node among patients who are classified as pN0 or pN1a at present. Only those whose preoperative ultrasound result indicating lymph node metastasis in the right lateral region (cN1b) are included. Meanwhile, to establish predictive models to help clinicians distinguish high-risk and low-risk populations and adopt different follow-up strategies.

A retrospective analysis was conducted on clinical data of patients who underwent right neck side area biopsy or dissection (cN1b) at the Affiliated Drum Tower Hospital, Medical School of Nanjing University from January 2017 to August 2022 to establish a predictive model. The model will then be validated on patients who undergo right neck lateral area biopsy or clearance from September 2022 to August 2023. Data including demographic characteristics, pathological results, and laboratory examinations were collected. The predictive model was established and evaluated by SPSS 26.0 and R 4.3.1.

This study includes a total of 316 patients diagnosed with papillary thyroid carcinoma and with preoperative ultrasound indicating lymph node metastasis in the neck lateral compartment (cN1b), who were treated at the Affiliated Drum Tower Hospital, Medical School of Nanjing University from January 2017 to August 2023. The training set included 264 patients, and the validation set included 52 patients.The area under the ROC curve for the prediction model is 0.83, with a diagnostic optimal threshold of 0.595. At this threshold, the model achieves a sensitivity of 81.6% and a specificity of 75.2%. Similar results were obtained when utilizing the model on the validation set.

There are distinct disparities in demographics, pathological result, and laboratory examinations that can differentiate between high-risk and low-risk populations for lymph node metastasis in the right lateral region. These differences can be effectively identified by developing predictive models.