| Objective:The purpose of our research was to analyze relevant clinicopathological characteristics that may affect the prognosis of patients with resected T1-3N0-2M0[according to the eighth edition of American Joint Committee on Cancer(AJCC)/Union for International Cancer Control(UICC)lung cancer tumor-node-metastasis(TNM)staging]non-small cell lung cancer(NSCLC),and develop software to predict the prognostic risk and survival of patients,so as to provide reference for clinicians to evaluate the prognosis of patients with resected T1-3N0-2M0 NSCLC to develop comprehensive treatment plans.Methods:(1)NSCLC patients who had received operation only with T1-3N0-2M0 stage were screened in the U.S.surveillance,epidemiology,and end results(SEER)database.(2)The clinicopathological features that may affect the prognosis of patients and the survival outcome and survival time were collected to establish a survival data set.(3)The clinicopathological characteristics(variables)were coded and assigned.(4)Based on survival data set,Kaplan-Meier analysis and Log-rank test were used to estimate the survival rate and compare the survival curves of different subgroups of clinicopathological features,so as to screen the clinicopathological features affecting the prognosis of patients.(5)In the survival data set,70%of patients were randomly selected to be the training set and the rest to be the test set.(6)Based on the training set,Cox proportional risk regression model was used to identify the independent prognostic factors of patients.(7)Based on the training set,the importance of prognostic prediction of each prognostic factor was ranked by tree model analysis to further verify the statistical results of Cox proportional risk regression model.(8)Based on the training set,Cox proportional risk regression model was used to construct the prognosis index(PI)equation,and then the PI value of each patient was obtained.According to the quantile of PI value,Kaplan-Meier analysis and Log-rank test were used to divide patients into three risk groups with significantly different survival rates:low-,intermediate-and high-risk group,and the estimated mean,median survival time and 1-5 year survival rates of each risk group were obtained,and a prognostic risk and survival prediction model for patients was constructed.(9)We validated the model based on the test set.(10)RNSCLC-PRSP software was developed to realize model prediction with patients prognostic risk and survival.(11)After logging in the specific website,clinicians register and log in the software for free,and select the corresponding clinicopathological characteristics of the patient,the prognostic risk and survival prediction results are obtained.Result:(1)After survival rate estimation and survival curves comparison among different subgroups of clinicopathologic characteristics,the survival rate of female is better than that of male(P<0.05).The survival rate of patients ?65 years old is better than that of>65 years old(P<0.05).The survival rate of other races are better than White or Black(P<0.017).The survival rate are stepwise better from N2 to N1 to No stage(P<0.017).Based on the number of positive regional lymph nodes(NPLNs),the survival rate are stepwise better from N?4 tol?N?3 to N=0(P<0.017).The survival rate of those patients who had underwent the lobectomy(LET)are better than patients who had underwent sublobectomy(SLET)or pneumonectomy(PET)(P<0.017).The survival rate of primary tumors located in the middle lobe(ML)is higher than the lower lobe(LL)(P<0.008).For histologic grade,the survival rate are stepwise worse from well differentiated(?)to moderately differentiated(?)to poorly differentiated(?)(P<0.008),the survival rate of well differentiated(?)is better than undifferentiated(?)(P<0.008).For histology,the survival rate are stepwise better from squamous cell carcinoma(S)to adenocarcinomaC AC)to bronchoalveolar adenocarcinoma(BAA)(P<0.005),the survival rate was stepwise better from S to AC to others(P<0.005),while the survival rate of BAA or others are all better than adenosquamous cell carcinoma(ASC)(P<0.005).The survival rate of single or married or divorced are better than widowed(P<0.005).For tumor extension(T stage),from T3 Inv to T2 Centr to T2 Visc PI to Tla ss,the survival rate are stepwise better(P<0.005),from T3 Satell to T2 Visc PI to T1a ss,the survival rate were stepwise better(P<0.005).For tumor size(T stage),from T2a>3-4 or T2b>4-5 or T3>5-7 to T1c>2-3 to T1b>1-2 or T1a?1,the survival rate are stepwise better(P<0.003).(2)Based on the training set and using Cox proportional risk regression model,we found that gender,age,N1 stage,number of examined regional lymph nodes(NELNs)(6<N?12),NELNs(N>12),NPLNs(1?N?3),NPLNs(N?4),LET,primary tumors located in upper lobe(UL),primary tumors located in ML,?,?,?,AC,S,ASC,married,divorced,T3 Inv,T3 Satell,T2a>3-4,T2b>4-5,T3>5-7 are independent prognostic factors for patients.(3)Based on the training set,the importance of prognostic prediction of each prognostic factor was ranked by tree model analysis,and the result is consistent with the conclusion of Cox proportional risk regression model.(4)Based on the training set,Cox proportional risk regression model was used to construct PI eq.PI=??ixi=0.379X1-0.403X2-0.267X51-0.167X61-0.298X62+0.460X71+0.617X72-0.344 X81-0.105X91-0.243X92+0.305X101+0.508X102+0.754X103+0.143X111+0.170X112+0.434X113-0.327X122-0.247X123+0.517X133+0.340X134+0.457X143+0.419X144+0.407X145.(5)Based on the training set,the PI value of each patient in the training set is obtained.According to the value of PI quantile,patients were divided into low-,intermediate-,and high-risk groups.PI<0.11 for low risk group,0.11?PI?0.79 for intermediate risk group,PI?0.79 for high-risk group.Using Kaplan-Meier analysis and Log-rank test,we discovered from low-to intermediate-to high-risk group,the survival rate of patients are stepwise worse(p<0.017).At same time,the estimated mean,median survival time and 1-5 year survival rate of each risk group were obtained.The estimated mean and median survival time are 90.16 months and 115.00 months in the low risk group,63.86 months and 47.00 months in the intermediate risk group,and 42.93 months and 24.00 months in the high risk group.The estimated 1-,2-,3-,4-and 5-year survival rates are 94.1%,87.0%,79.0%,73.5%,68.2%respectively in the low risk group,83.9%,69.3%,58.9%,49.1%,43.8%respectively in the intermediate risk group,and 68.6%,49.7%,41.6%,32.6%,26.8%respectively in the high risk group.Furthermore,prognostic risk and survival prediction model was constructed.(6)Based on the test set model validation,we discovered the model is effective.(7)We developed RNSCLC-PRSP software to implement the model for prognosis risk and survival prediction of resected T1-3N0-2M0 NSCLC patients.Clinicians can login web site http://www.rnsclcpps.com,register and login the software for free,select clinicopathological characteristics of patient,prognostic risk and survival prediction results of patient are obtained.Conlusions:By analyzing the prognostic factors of resected T1-3N0-2M0 NSCLC patients,grouping the prognostic risks and corresponding survival prediction,we developed a new prognostic prediction software.This software enables clinicians to conveniently,quickly,comprehensively and accurately predict the prognostic risk and corresponding survival of resected T1-3N0-2M0 NSCLC patients,providing a reference for clinicians to make comprehensive treatment plans for patients. |
PDF Full Text Request