Objective:To investigate the value of preoperative CT in diagnosis of non-small cell lung cancer(NSCLC)with visceral pleural invasion(VPI).Methods:A total of 489 NSCLC patients were retrospectively included and divided into the training set(362 cases)and the validation set(127 cases).Each set was divided into VPI group and non-VPI group according to the results of elastic fiber staining.In the training set,there were 57 cases in VPI group and 305 cases in non-VPI group.And in the validation set,there were 15 cases in VPI group and 112 cases in non-VPI group.The pathological features,clinical data and CT characteristics of VPI group and non-VPI group were summarized.Through univariate analysis and binary logistic regression,the independent risk factors were screened and combined models were established.The ROC curve was used to compare the diagnostic efficiency of each model.Results:In the training set,there were statistical differences between VPI group and non-VPI group in nodular type,air bronchial sign,type of pleural contact,length of solid contact,long diameter,short diameter,CTmin value and CTmean value(all P<0.05).After multivariate logistic regression was used to eliminate the confounding factors,the length of solid contact and type of pleural contact were independent risk factors for predicting VPI(both P<0.01).The best cut-off value of solid contact length was 8.5 mm,with a sensitivity of 80.7%and a specificity of 55.4%.In the training set,the AUCs of models among the combined model 1 based on the statistically significant factors in the univariate analysis,the combined model 2 established by two independent risk factors and the model established by solid contact length were 0.775,0.755 and 0.734,respectively,and in the validation set,the AUCs of the three models were 0.789,0.740 and 0.682,respectively.There were no significant differences among the three models either in the training set or in the validation set(all P>0.05).Conclusion:The length of solid contact and the type of pleural contact can be used as independent risk factors for distinguishing VPI,which can provide non-invasive guidance for clinical preoperative decision-making.
国家科技期刊平台
NSTL
维普
万方数据
