The treatment for papillary thyroid microcarcinoma (PTMC) is controversial. Central lymph node metastasis (CLNM) is one of the main predictors of recurrence and survival, accurate preoperative identification of CLNM is essential for surgical protocol establishment for PTMC. The objective of this study was to establish a nomogram to predict the possibility of CLNM in PTMC patients. A total of 3023 PTMC patients were randomly divided into two groups by a ratio of 7 to 3, the training group (n = 2116) and validation group (n = 907). The LASSO regression model and multivariate logistic regression analysis were performed to examine risk factors associated with CLNM. A nomogram for predicting CLNM was established and internally validated. Meanwhile, we follow-up the serum thyroid function FT3, FT4, TSH, Tg, TGAb and TPOAb in 789 PTMC patients for 4 years after surgery and compared the differences between the CLNM (+) and CLNM (-) groups, respectively. The LASSO regression model and multivariate logistic regression analysis showed that younger age, lower BMI, being male, location in the lower pole, calcification, 1 ≥ diameter ≥ 0.5 cm, multifocality lesions, extra thyroidal extension (ETE), enlargement of central lymph node (ECLN), lateral lymph node metastasis (LLNM) and higher carcinoembryonic antigen were the ultimate risk factors for determining CLNM. A nomogram for predicting CLNM was constructed based on the influencing factors and internally validated. By establishing the prediction model, the AUC of CLNM in the training and validation groups were 0.73 (95% CI, 0.70-0.76) and 0.75 (95% CI, 0.71-0.79) respectively. Results of the DCA showed that the model is clinically useful when deciding on intervention in the most range of the threshold probability. A 4-year follow-up of thyroid function showed that FT3 and FT4 remained at stable levels after 3 months postoperative and were higher in the CLNM (+) group than in the CLNM (-) group. Hypothyroidism appeared predominantly within 3 months after surgery. The overall incidence of the CLNM (+) group and CLNM (-) groups were 16.46% and 12.04%, respectively. The nomogram model constructed in this study has a good predictive effect on CLNM in PTMC patients and provides a reasonable reference for clinical treatment.
Read full abstract