Prediction of cervical lymph node metastasis in papillary thyroid carcinoma using model based on thyroglobulin and clinical features / 中华内分泌外科杂志

ABSTRACT

Objective:To investigate the relationship between preoperative serum thyroglobulin (Tg) and clinical data with the risk of cervical lymph node metastasis in patients with papillary thyroid carcinoma (PTC) .Methods:Data of 395 PTC patients who underwent surgery from Feb. 2016 to Jun. 2019 at the Second Affiliated Hospital of Soochow University were retrospectively analyzed. Based on whether cervical lymph nodes had metastasis, patients were classified into central lymph node metastasis positive group ( n=195 cases) , central lymph node metastasis negative group ( n=200 cases) , lateral lymph node metastasis positive group ( n=72 cases) , and lateral lymph node metastasis negative group ( n=323 cases) .Then the relationship between age, sex, multifocality, tumor diameter, capsular invasion, preoperative TSH and preoperative Tg with lymph node metastasis were analyzed by SPSS. Comparisons between groups were performed by χ2 test and rank sum test. Prediction efficiency of the preoperative Tg and Logistic regression model was estimated by receiver operating characteristic (ROC) curve. A total of 100 PTC patients confirmed by pathological results in the Second Affiliated Hospital of Soochow University from Jul. 2019 to Apr. 2020 were selected as the validation data. Results:Multi-factor Logistic regression showed that age, tumor diameter, capsular invasion and preoperative Tg were independent risk factors of central cervical lymphatic metastasis ( P<0.05) ; Tumor diameter, capsular invasion, central cervical lymphatic metastasis and preoperative Tg were independent risk factors of lateral cervical lymphatic metastasis ( P<0.05) . The area under the ROC curve (AUC) for diagnosing central lymph node metastasis by preoperative Tg was 0.710, with a sensitivity of 49.2%, and specificity of 88.5%. The AUC for diagnosing lateral lymph node metastasis by preoperative Tg was 0.728, with a sensitivity of 59.7%, and specificity of 89.5%. The AUC for diagnosing central lymph node metastasis by the prediction model was 0.773, with a sensitivity of 78.5%, and specificity of 64.5%.The AUC for diagnosing lateral lymph node metastasis by the prediction model was 0.869, with a sensitivity of 84.7%, and specificity of 70.3%. Conclusions:The preoperative serum Tg level is correlated with cervical lymph node metastasis in PTC patients. But the Logistic regression model based on preoperative Tg and other independent risk factors shows a better predictive value.