Predicting postoperative metastasis of non-small cell lung cancer based on logistic regression and currently constructed nomogram models
Objective:To investigate the risk factors for postoperative metastasis of non-small cell lung cancer(NSCLC),and construct a nomogram model for predicting postoperative metastasis of this tumor.Methods:265 NSCLC patients surgically treated in our hospital between May 2018 and February 2021 were included in the model group,and another 96 NSCLC patients undergone surgical treatment in our hospital between March 2021 and June 2022 were recruited in validation group.LASSO analysis was performed,and logistic regression model was used to respectively screen the predictive factors and risk factors for postoperative metastasis of NSCLC.Nomograph model for postoperative metastasis of NSCLC was constructed using software R(version:4.2.0),and validated as well.Results:Logistic regression analysis showed that the risk factors for postoperative metastasis of NSCLC were associated with maximum tumor diameter>3cm,differentiation degree,clinical stage,level of carcinoembryonic antigen(CEA),D-dimer and MGMT gene promoter methylation(all P<0.05).The calibration curve demonstrated better fit between the predicted and actual values in both groups.The AUC was 0.732 and 0.843,respectively for the model group and the validation group.The decision curve of the model group showed that currently constructed nomogram model had better clinical efficacy for estimation of postoperative metastasis of NSCLC.Conclusion:Tumor maximum diameter,differentiation degree,clinical stage,CEA level,D-dimer and MGMT gene promoter methylation are influencing factors for postoperative metastasis of NSCLC.The nomogram model constructed in current study can be clinically conductive to early prevention and treatment of postoperative metastasis of this neoplasm.