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

Objective:Patient clinical information,serological examination,ultrasound examination,and quantitative analysis of contrast-enhanced ultrasound were combined to screened out the risk factors associated with lymph node metastasis(LNM)in papillary thyroid carcinoma(PTC).And a clinical nomogram prediction model was established to predict the risk probability of LNM,which is helpful for non-invasive,convenient and effective preoperative LNM risk assessment in PTC patients.Methods:A total of 252 patients with PTC,with or without LNM,were retrospectively analyzed in Xiangya Hospital of Central South University who underwent preoperative thyroid contrast-enhanced ultrasonography,and then underwent thyroidectomy and lymph node dissection.Patient information were collected and recorded.Two-category univariate logistic regression analysis was used to preliminarily screen out the statistically significant indicators.After multivariate logistic regression analysis,the risk factors that were highly correlated with the occurrence of LNM in PTC were screened out.A clinical prediction model was established,and the visualization of the predictive model is a nomogram.The prediction model was evaluated and validated by ROC curve,calibration curve,cross-validation,clinical decision curve and clinical impact curve.Results:The risk factors associated with LNM in PTC with statistical significance(P<0.05)were screened out by Univariate analysis: gender,age,longest diameter of nodule,halo around nodule,thyroid capsule invasion,lymph node microcalcification,lymph node hyperechoic area,peak intensity of contrast(Peak Intensity,PI),area under the curve of the time intensity curve of contrast(Area Under Curve,AUC).After multivariate analysis,four risk factors that were highly correlated with the occurrence of LNM in PTC were screened out: Gender,age,thyroid capsule invasion,lymph node microcalcification.The prediction model formula was: logit(P)=2.380-1.037*sex-0.044*age+1.627*thyroid capsule invasion + 2.009*lymph node microcalcification.The AUC of the ROC curve was 0.800(95%CI: 0.747-0.853),P<0.001.The calibration curve shows that the probability of the prediction model predicting LNM was consistent with the probability of the actual LNM.The result of cross-validation was: accuracy = 0.757.This shows that the predicted results was good.Kappa=0.508,the consistency was acceptable.In the DCA curve and the clinical impact curve,the model performed well in clinical efficacy.Conclusion:Male gender,young patients,thyroid capsule invasion,and cervical lymph node microcalcification are highly correlated risk factors for LNM in PTC.The nomogram prediction model was established based on clinical,serological and ultrasound information,and it had a good ability to predict lymph node metastasis.It is helpful for effective preoperative LNM risk assessment in PTC patients.