Construction And Validation Of Preoperatively Predictive Model For Low-risk Papillary Thyroid Carcinoma

Object:We aim to explore the potential preoperative predictors of low-risk PTC patients and to construct a nomogram model which can separate low-risk PTC patients who have excellent prognosis from medium and high-risk PTC patients preoperatively and can be employed to make individualized treatment regime in clinical practice.Methods:We respectively reviewed and analyzed the preoperative clinical and pathological characters of patients who underwent open surgery and confirmed as papillary thyroid carcinoma by pathology at Department of thyroid Surgery,China-Japan Union Hospital,Jilin University during 2020.A total 1875 patients were randomly divided into training set and validation set as 7:3.We decided the prognostic stratification standard with full reference to ATA guidelines,AMES system and MACIS system.Logistic regression and Lasso regression were used to analyze the potential indicators for low-risk PTC.The preoperative prediction model was constructed based on the results of multivariable analyses.A nomogram was drawn for clinical utility.We used ROC curve to determine cutoff value of risk score for prognostic stratification.The predictive value of model was verified in the validation set.The discrimination of model was verified using the receiver operating characteristic(ROC)curve,area under the curve and calibration plot.To evaluate the clinical utility of the model,decision curve analysis and clinical impact curve was performed to calculate the net benefits.Results:Our research obtained 1,875 PTC patients,there were 1,313 patients in training set,562 patients in validation set.8 potential preoperative indicators were selected by univariable analysis and lasso regression,including sex,age,number of foci,maximum diameter on US,calcification,capsule,lymph node status on US,TPO antibody.The model showed good discrimination,AUC were 0.777 [95% CI(0.752,0.803)] and 0.769 [95% CI(0.729,0.809)] in training set and validation set,respectively.The calibration curve exhibited to be rather good consistency with the perfect prediction in both training set and validation set.Results of decision curve analysis and clinical impact curve showed the model had good efficacy in predicting the prognostic risk of PTC.Conclusion:1.Male gender,an increase in tumor number and size,tumor close to or invading the capsule,and the presence of calcification and suspicious lymph nodes were identified as independent risk factors of middle–high-risk PTC;2.Older age and a higher level of TPO antibodies(≥26.45 pmol/L)were protective factors of low risk PTC;3.The constructed nomogram aids in predicting the prognostic risk preoperatively,which helps clinical practitioners select individualized treatment plans for each patient.