Development And Validation Of A Novel Risk Stratification Model Of Multiple Primary Lung Cancer Based On SEER Database

Objective To further study the survival prognosis of Multiple Primary Lung Cancer(MPLC)and investigate the factors influencing the prognosis of patients with MPLC.Based on the above factors,establishing and verifying the nomogram in order to realize the quantitative assessment of risk of death in patients with MPLC,which provide a new direction for the clinical work individually.Methods Through the surveillance,epidemiology and end results(SEER),we collected4177 patients diagnosed with MPLC from 2007 to 2015 which randomly divided into training set and verification set in a ratio of 7:3.The training set was used as the modeling group.The risk factors of patients were screened by univariate Cox regression analysis.And then the independent risk factors screened by multivariate Cox regression analysis were included in the nomogram.The validation set was used as the validation group to verify the accuracy of the nomogram.Kaplan-Meier survival curves were used to compare the survival differences among the groups.Index of concordance and the Area Under Curve were used to evaluate the differentiation of nomogram and calibration curves were used to evaluate the fit of the prediction model.Then Net reclassification improvement and Integrated discrimination improvement compared the accuracy of nomogram.Lastly,decision curve analysis assesses the actual clinical benefits.Results1.Univariate Cox analysis showed that age at diagnosis,sex,AJCC stage,histology of first primary lung cancer(FPLC)and second primary lung cancer(SPLC),tumor size of MPLC,surgery of FPLC and SPLC,average family income,marital status,race(all P <0.05)were significantly correlated with prognosis of MPLC patients.2.According to the results of the Cox multifactor analysis,there were 11 independent risk factors included in the nomogram eventually,including diagnosis at age,sex,AJCC stage,histology of FPLC and SPLC,tumor size of SPLC,surgery of FPLC and SPLC,average family income,marital status,race(P < 0.05).3.The C-index of nomogram in the training set was 0.702,which was 0.699 in the verification set.The AUC used to predict 3-year,5-year and 8-year overall survival in the training set were 0.743,0.751 and 0.759,respectively.In the validation set,they were 0.737,0.734 and 0.695,respectively.4.The calibration curves of the nomogram was close to the reference line and falls along the 45-degree diagonal,which indicates that the nomogram has a good fitting degree.5.In the training set,the 3-year,5-year and 8-year NRI values were 0.590(95%CI=0.505-0.664),0.576(95%CI=0.522-0.688)and 0.577(95%CI=0.467-0.762),respectively.In the training set,the 3-year,5-year and 8-year IDI values were 0.097,0.091 and 0.070,respectively.In the validation set,the 3-year,5-year and 8-year IDI values were 0.102,0.090 and 0.066,and all the IDI values were > 0,showing the excellent prediction performance of the nomogram.Clinical decision curves had a benefits in both the training set and the validation set.6.The prediction accuracy of the nomogram and the traditional prediction model were compared by the AUC value.The AUC value of the training set and the verification set were higher than that of the traditional prediction model regardless of the 3-year,5-year or 8-year nomogram prediction model.The AUC values of the nomogram and the traditional prediction model were 0.743vs0.622,0.751vs0.612 and 0.759vs0.583 in the training set,respectively.And the verification sets were: 0.737vs0.60,0.734vs0.572 and0.695vs0.569,respectively.Conclusions The nomogram of the MPLC patients through SEER database has good differentiation and fitting degree,which can predict the individual survival prognosis of MPLC patients more efficiently,accurately and easily to help clinicians to better evaluate the prognosis of patients than before and provide treatment individually.