Analysis Of Risk Factors And Establishment Of Prediction Model For Unifocal Papillary Thyroid Carcinoma With Central Lymph Node Metastasis

Objective:The aim of this study was to explore the risk factors of central lymph node metastasis(CLNM),and construct a combined nomogram model based on clinical characteristics and ultrasound(US)features in predicting CLNM in unifocal papillary thyroid carcinoma(PTC)patients before surgery.Methods:The clinicopathological information of 450 patients with unifocal PTC from January2018 to December 2020 in the Second Hospital of Jilin University were retrospectively review.Retrospective review of clinicopathologic information and ultrasonography images.The patients were divided into training sets(n=315)and testing sets(n=135)with a ratio of 7:3,and divided into negative and positive groups according to postoperative pathologic lymph node diagnosis in training sets and testing sets,respectively.The univariate analysis and multivariate logistic regression analysis were performed to screen for significant predictors of positive lymph nodes,which were then included in the nomogram construction using training sets data.Then,the prediction model was validated in the testing sets.Finally,the discrimination,calibration ability,and clinical usefulness were used to evaluate the model’s performance according to the TRIPOD transparent reporting.Results:Among 450 patients with unifocal PTC,124 were positive for CLNM and 326 were negative for CLNM,the incidence of CLNM was 27.6%.Multivariable logistic regression analysis showed that unifocal PTC patients with younger age(OR=0.952,p=0.001),male gender(OR=2.123,p=0.041),microcalcification(OR=1.800,p=0.042),larger tumor size(OR= 3.571,p<0.001)and US reported suspicious lymph metastasis(OR=3.565,p=0.042)were independent predictors of CLNM(P<0.05).And these predictors were served as the basis of the nomogram.Our model showed good predictive capacity in the training sets and testing sets,with an AUC of 0.758(95%CI,0.696-0.820)and 0.744(95%CI,0.649-0.838),respectively.The calibration curve revealed that the predicted and observed probabilities of CLNM were in a good agreement,mean absolute error in training sets and testing sets were0.013 and 0.025,respectively.The decision curve analysis and clinical impact curve revealed that the nomogram model can benefit for the PTC patients with CLNM.Conclusion:Younger age,male gender,lager tumor size,microcalcification,and US reported lymph metastasis were considered as independent predictors for CLNM in unifocal PTC patients,which indicate the surgeons should make reasonable central lymph node dissection according to different factors.It is feasible to develop prediction model of CLNM in PTC patients by incorporating clinical characteristics and US features.The nomogram model can be used to predict the risk of CLNM in PTC patients rapidly,accurately,non-invasively and individually,which is helpful to guide clinical decision-making.