Establishing A Prediction Model For Pulmonary Nodule Malignancy Risk In Yunnan Province,China

Objective:This research retrospectively analyzed clinical characteristics,imaging data,tumor markers,and other factors of patients with pulmonary nodules in Yunnan Province,China,to establish a prediction model by studying independent risk factors for predicting pulmonary nodule malignancy,using uni-and multi-variate analysis of each factor.Methods:This study selected 1139 patients in Yunnan Province,China,who underwent surgical treatments and pathological examinations for pulmonary nodules at the First Affiliated Hospital of Kunming Medical University from May 2017 to December 2020.Clinical features,imaging characteristics,and tumor markers of the patients were collected.The patients were divided into 2 groups with a ratio of 7:3.Patients admitted between May 2017 and June 2020 served as the model group(778 cases),and data collected from July 2020 to December 2020 was categorized as the verification group(361 cases).IBM SPSS19.0 was used in statistical analysis.First,all patients’ data were analyzed to determine factors that have a significant influence on the judgment of benign or malignant nodules.Univariate and multivariate logistic regressions were applied to evaluate the risk factors from the model group and to figure out factors closely related to the prediction of pulmonary nodules malignancy.The prediction model of pulmonary nodule malignancy risk was built based on the results obtained.Finally,the data from the verification group was introduced into the model to verify the accuracy of the prediction model.Results:Univariate and multivariate regression analysis showed age,Xuanwei population,maximum nodule diameter,spicule sign,vascular convergence sign,vacuole sign,mGGO,pGGO,CEA,and NSE are independent risk factors for pulmonary nodule malignancy.Gender is an independent protective factor.Smoking history,family history of malignant tumor,nodule location,lobulation sign,pleural stretch sign,and CYFRA21-1 are not independent risk factors.The regression equation of the model group is P=e/(1+ex);e is the natural logarithm.x=-3.289+(age*0.031)+(maximum nodule diameter*0.458)+(spicule sign*0.820)+(vascular convergence sign*1.109)+(vacuole sign*0.590)+(pGGO*2.429)+(mGGO*3.314)+(CEA*1.645)+(NSE*0.881)-(gender*0.560).The AUC value of the ROC curve calculated,using data from the model group,was 0.844(95%CI:0.816-0.868).The maximum value of the Youden index from the ROC curve was 0.56,and the diagnostic threshold has a p-value of 0.132.The sensitivity,specificity,positive predictive value,and negative predictive value of the model were 83.86%,72.10%,84.53%,and 71.01%,respectively.The results calculated from the verification group were obtained by introducing the data from the model group into the equation.The AUC of the ROC curve was 0.842(95%CI:0.800-0.878),the sensitivity was 75.84%,the specificity was 93.22%,the positive predictive value was 95.23%,and the negative predictive value was 57.00%.Conclusion:1.This study discovered statistical significance in age,Xuanwei population,maximum nodule diameter,spicule sign,vascular convergence sign,vacuole sign,pGGO mGGO,CEA and NSE among all the risk factor.Multivariate logistic analysis showed that age,maximum nodule diameter,spicule sign,vascular convergence sign,vacuole sign,pGGO,mGGO,CEA,and NSE are independent risk factors of pulmonary nodule malignancy,while gender is a protective factor of malignant pulmonary nodules.Smoking history,family history of malignant cancer,and lobulation sign are not independent risk factors of pulmonary nodule malignancy.2.The data for benign and malignant nodules of the validation group were calculated and analyzed in this prediction model separately.The diagnostic threshold p-value was 0.613.Using the prediction model,when the diagnostic p-value is greater than 0.613,the probability of malignancy is high.When the diagnostic p-value is less than 0.613,a benign pulmonary nodule is more likely.3.Different prediction models of pulmonary nodule malignancy are used at each treatment center,which may be a result of regional differences.It is necessary to integrate imaging examination,laboratory results,clinical data,and relevant regional factors into the establishment of a prediction model of pulmonary nodule malignancy risk,as such a prediction model might help healthcare professionals in making accurate predictions of the nodules in the locality.