Development And Validation Of A Prediction Model For The Risk Of Malignancy In 5-15mm Solitary Pulmonary Nodules

Objective:This study aims to develop and validate a prediction model for the risk of malignancy in 5-15mm solitary pulmonary nodules(SPN).This prediction model might help identify patients the probability of malignancy,avoid the over-treatment,and increase the overall survival of 5-15mm SPN patientsMethods:958 patients with 5-15 mm SPN were enrolled retrospectively,who had undergone surgical treatment with definite pathological diagnosis from December 2012 to March 2018 in Yunan Cancer Hospital.The clinical characters and preoperative laboratory inspection data included age,gender,height,weight,smoking history,history of respiratory disease,history of malignancy,family tumor history,carcinoembryonic antigen,carbohydrate antigen 125,carbohydrate antigen 724,and neutrophil to lymphocyte ratio(NLR).Besides,the imaging characters based on the axial chest CT with thin layer images including the SPN position,maximum diameter,minimum diameter,shape,edge,ground-glass opacity(GGO)composition,cavity,vacuole,air bronchus sign,vascular converge sign,pleural retraction sign,calcification,emphysema background,Fiber strip were evaluated.All 958 patients were randomly into development set(n=670)and validation set(n=288)in a 7:3 ratio The chi-square test was used to compare the count data,while the independent sample T test or rank sum test was used to compare the measurement data between the two groups.In the development set,independent predictors for the probability of malignancy were selected with multivariate logistic regression.Simultaneously,logistic regression was used to develop the predictive model using the clinical characters,laboratory inspection data and imaging signs.The discrimination degree and calibration degree of the prediction model were assessed using receiver operating characteristic(ROC)curve and calibration curves respectively.In addition,the diagnostic performance of the predictive model was compared in the development set and the validation set,concurrently,while the diagnostic performance was compared between the above-mention model and the internationally recognized Mayo model in the development set.Finally,the decision curve analysis was used to evaluate the clinical value of the constructed predictive modelResults:Univariate logistic analysis revealed that there were statistical differences between benign and malignant SPN in age,history of malignancy,maximum diameter of lesion,upper lobe,edge burr,vacuole sign,vascular converge sign,air bronchus sign,pleural retraction sign,GGO composition,emphysema background and NLR(P<0.10).Multivariate logistic regression analysis showed that age,vacuole sign,vascular converge sign,air bronchus sign,pleural retraction sign,GGO composition and NLR were independent predictors of malignancy in patients with 5-15mm SPN(P<0.05).The mathematical prediction model to estimate the probability of malignancy of 5-5mm SPN was as follow:Logit(P)=ex/(1+ex),x=-3.14282+(0.03737xage)+(0.51374×vacuole sign)+(1.87897×air bronchus sign)+(1.15055xpleural retraction sign)+(1.57413×GGO component)-(0.84182×NLR).At the cut-off value P=0.234,the Jordan index is the largest at 0.592.At this time,the area under the ROC curve(AUC)of the prediction model in the development set was 0.860,(95%CI:0.832,0.888),the sensitivity was 0.908,the specificity was 0.684,the positive likelihood ratio was 2.869,the negative likelihood ratio was 0.135,the positive predictive value was 0.766,the negative predictive value was 0.867,and the accuracy was 0.803.In the validation set,the AUC value was 0.864(95%CI:0.820,0.907),the sensitivity was 0.772,specificity was 0.863,positive predictive value was 0.858,negative predictive value was 0.779,positive likelihood ratio was 5.646,negative likelihood ratio was 0.264,accuracy was 0.816.The diagnostic efficacy of the Mayo model in the validation set was as follows:the AUC value 0.621(95%CI 0.585,0.657),,sensitivity was 0.844,specificity 0.361,positive likelihood ratio 1.320,negative likelihood ratio 0.433,positive predictive value 0.597,negative predictive value 0.674,and accuracy 0.616.In the validation set,the AUC value(0.864)of the prediction model constructed in this study was greater than that of the Mayo model(0.621)(P<0.001).Decision curve analysis showed that the prediction model in this study has clinical application value in the range of threshold probability 0.16-0.94.Conclusions:1.Age,NLR,vacuole sign,vascular converge sign,air bronchus sign,pleural retraction sign and GGO composition are independent predictors for differentiating benign and malignant SPN with a diameter range of 5-15mm;2.The diameter range of 5-15mm SPN predictive model constructed in this study has high predictive efficacy;3.This model has good clinical practical value,and it may be used to assist clinicians in the clinical development of individualized treatment plan for patients with diameter range of 5-15mm SPN.