Preoperative Malignant Risk Assessment Of Thyroid Nodules:Development And Validation Of Nomogram Prediction Model

Objective :This topic aims to establish a practical and simplified method to analyze clinical thyroid nodules,construct and verify a simple and reliable prediction model,and use the ultrasonic characteristics of thyroid nodules to stratify the risk of malignant tumors.Methods:1.3095 cases of thyroid nodules who met the inclusion criteria and were treated in Yunnan Cancer Hospital’s head and neck surgery from January 2020 to December2020 were selected as the study subjects,and were randomly divided into a training set(2163)and a validation set(932)based on a 7:3 ratio.Collect the ultrasound features of thyroid nodules(shape,margin,echo,composition,calcification,blood flow signals,capsule,whether the glands are diffuse lesions,whether there is a comet tail sign,etc.).The patients were divided into malignant tumor group and benign tumor group according to the postoperative pathological examination results.2.All data were statistically described and analyzed for differences.The categorical variables were described using n(%),and the differences were tested using Chi Square test.P<0.05 indicates that the differences were statistically significant.A multivariate binary logistic regression was conducted to screen out independent predictors by including all variables with differences.3.The least absolute contraction and selection operator method(Lasso regression)was used to screen the predictive factors,and a nomograph prediction model was constructed on this basis.The prediction model was used to predict the validation set and the training set,respectively.The area under the subject work characteristic(ROC)curve was used to verify the predictive differentiation of the prediction model,and the consistency of the model was verified by drawing a correction curve.The decision curve analysis(DCA)was used to verify the clinical effectiveness of the prediction model.P>0.05 indicates a statistically significant difference.All data were statistically analyzed using SPSS27.0 and R software.4.By comparing the predictive model of the nomogram in this study with the American College of Radiology TIRADS guidelines and the China TIRADS(C-TIRADS)guidelines on the ROC for the diagnosis of thyroid malignant nodules,we can more objectively understand the accuracy differences in the diagnostic results of malignant nodules among the three,thereby achieving a more accurate assessment of the malignant risk of thyroid nodules.Results:1.This study shows that there is a significant difference in the morphological basis for predicting benign and malignant tumors by two-dimensional ultrasound examination(The difference was statistically significant(P<0.05)).There was no statistically significant difference between the other factors(P>0.05).In the case of thyroid nodules with diffuse diseases of the gland(Hashimoto’s disease or subacute thyroiditis,etc.),there was no significant difference in the above comparison(P>0.05).Further binary logistic regression analysis revealed significant differences in the shape,margin,calcification,composition,and echo of the nodules between the two groups,with a statistically significant difference(P<0.05).There was no statistically significant difference between the other factors(P>0.05).In the case of thyroid nodules with diffuse diseases of the gland(Hashimoto’s disease or subacute thyroiditis,etc.),there was no significant difference in the above comparison(P>0.05).2.Perform Lasso regression analysis using whether malignant tumors occur in the training set as the dependent variable,and select the quintuple cross validation error as the minimum λ+1 As the optimal value of the model,the selected variables include shape,edge,calcification,envelope,echo,composition,and comet tail sign.3.Through screening,predictive variables are included in the constructed nomograph model.Using a prediction model to predict the training set,the prediction accuracy rate(i.e.,predicted outcome=actual outcome)is 0.814,and the confidence interval(CI)is 95% CI 0.791-0.837(Figure 3).The accuracy rate for predicting the validation set was 0.818,and the confidence interval(CI)was 95% CI 0.784-0.853.As can be seen from the calibration curve,the trend of the simulated curve and the actual curve are basically consistent.Decision curve analysis shows that predictive models have high clinical benefits in predicting the risk of thyroid nodule malignancy.4.Compared with the ACR TIRADS and C-TIRADS guidelines,the nomogram prediction model had the highest AUC,with an AUC of 0.818(95% CI 0.770-0.857),significantly higher than the diagnostic efficacy of the ACR TIRADS(AUC: 0.718)and C-TIRADS(AUC: 0.774)guidelines(p<0.05).Conclusions:1.The ultrasound signs of thyroid malignant nodules are more prone to irregular shapes and edges,solid hypoechogenicity,microcalcification,and capsule invasion,consistent with previous studies.This study also found that compared to benign nodules,malignant nodules are more likely to have ultrasound signs adjacent to the capsule,which can be used as a predictor of preoperative thyroid malignancy risk assessment,but it is not recommended to use them alone as a predictor of thyroid malignancy risk.2.Thyroid nodules with irregular shape,irregular edge,aspect ratio>1,microcalcification,solid components,and low echo are more likely to be malignant.The shape,boundary,calcification,component,and echo of thyroid nodules can be used as independent predictors of preoperative malignant risk assessment.3.As a convenient,practical,intuitive,and visual tool,nomogram can accurately predict the risk probability of malignant thyroid nodules and provide reference for clinical decision-making.4.Compared to the ACR TIRADS and C-TIRADS guidelines,the nomogram prediction model has higher diagnostic accuracy and is currently a recommended method for predicting thyroid risk.