Prognostic Analysis And Validation Of Multiple Primary Lung Cancers Based On SEER Database

Objective:Because there is a huge difference in prognosis and survival between multiple primary lung cancers and intrapulmonary metastatic cancers.The original 8th edition of the AJCC TNM staging for lung cancer designated multifocal lung cancer as IM,homolobular lung cancer as T3,ipsilateral different lobe cancer as T4,and bilateral lung cancer as M1.The 8th edition of the AJCC staging guidelines has been further corrected to include separate T,N,and M staging for each tumor for second primary lung malignancies.For intrapulmonary metastatic lesions,the same staging criteria as in the 7th edition were used.Studies on the prognosis of multiple primary lung cancers do not yet fully account for the risk factors for multiple primary lung cancers.The aim of this study was to:1.To study the baseline information of patients with multiple primary lung cancer through SEER database.We also summarized the risk factors affecting the OS of patients with multiple primary lung cancers by multi-factor COX regression analysis.2.The prognostic analysis of multiple primary lung cancers was plotted by R-language algorithm to predict the survival rate of patients at3 and 5 years.Internal validation was also performed by C-index,ROC curve and K-M curve.3.To collect data from postoperative thoracic surgery cases in the Sino-Japanese Consortium and compare them with data from the SEER database,and to externally validate the accuracy of the prognostic analysis column line graphs for multiple primary lung cancers obtained from the SEER database.Methods:（1）Data sourcesSEER database patients were screened from the Surveillance,Epidemiology,and End Results（SEER）public use database.Thoracic surgery patient data were obtained from a random selection of postoperative thoracic surgery patient information from the Sino-Japanese Friendship Hospital of Jilin University.（2）Study parametersInitial screening of the SEER database included all patients aged 20 years and older who were diagnosed with first primary lung cancer between 2009 and 2017.Follow-up for SPLC diagnoses was through2017.To further differentiate SPLCs from IPLCs with advanced metastatic disease,patients with multiple primary tumors that were complicated by the presence of malignancy at other sites were excluded.Information on age,sex,race,TNM stage,and histological type at diagnosis for IPLCs and SPLCs was obtained from the SEER database.For the thoracic surgery data,patients’ gender,age,number of lung cancers,pathological type,and TNM stage（version 7）were counted.Telephone follow-up was also performed to obtain survival data for patients with multiple primary lung cancers.（3）Data analysisStatistical analysis to determine risk factors was performed with SPSS 26 for Windows.multivariate analysis was performed using multifactorial Cox regression analysis.Categorical variables were grouped according to the results of clinical studies,and the groups were analyzed and modeled before modeling.Results were compared using theχ2 test.Continuous variables were compared using the Student t test.Survival curves were described using the Kaplan-Meier method and compared using the log-rank test（Logrank test）.The results of the multivariate analysis were developed by using the rms package（http://www.r-project.org/）in R language（version 4.0.3）,and a column line graph was used for prognostic prediction.The consistency index（C-index）was used to measure the performance of the prognostic plot.The survival ROC package in R language was used to evaluate the model using ROC.And the performance of each risk factor was evaluated by Kaplan-Meier curves of each risk factor.The data were divided into high risk group and low risk group by prognostic analysis using this model,and the Kaplan-Meier curves of this model were evaluated.The accuracy of the model was externally validated using the Calibrate curve using the Survival package in R.P < 0.05 was considered to be statistically significant.Results:1.SEER database data analysis of multiple primary lung cancers（1）Patient characteristicsIn the SEER database,2365 patients who were surgically first diagnosed with IPLC and 1376 patients with SPLC were screened from2009 to 2017.the gender composition was similar in both groups,with a female over-representation of 55.14% and 52.47%,respectively.There were no statistically significant differences between races.The predominant histological type in IPLC and SPLC was adenocarcinoma,accounting for 66.0% and 67.9%,respectively,and squamous carcinoma was also highly represented,accounting for 28.6% and 24.7%,respectively.The clinical characteristics were compared as follows（Table4-1）.It can be seen that age,TNM stage,and histological type were independent of each other in IPLC and SPLC,and the differences were all statistically significant.（P < 0.05）（2）Multi-factor COX analysis of multiple primary lung cancersThe C-index was 0.559（95% CI,0.530 to 0.588）.The HR ratio showed that gender,race,histological type and TNM stage were the risk factors that had a greater effect on multiple primary lung cancers.And the effects were statistically significant in Asians and whites and in adenocarcinoma and small cell carcinoma.（P < 0.05）.（3）The prognostic prediction of multiple primary lung cancer column line graphThe prognostic graph also showed that the most influential prognostic factors were the pathological type and N stage of the tumor.Surprisingly,an age of <49 years predicted a worse prognosis.Moreover,the gender of the patient and the M-stage of the main lesion had little impact on the prognosis.It is also clear from the figure that early-stage multiple primary lung cancers basically have good 3-year survival rates（>90%）.（4）Comparison of prediction accuracy: AUC curveThe AUCs of the model for predicting 3-year survival and 5-year survival were 0.615 and 0.531,respectively,which indicated that the model could predict 3-year survival and 5-year survival from the major lesions of multiple primary tumors.（5）Evaluation of the prediction model: K-M survival curvesThe survival curves of the two groups were significantly separated and the difference was statistically significant（P<0.01）.It proved that the model could predict the survival of patients well.（6）Comparison of K-M survival curves for each risk factorAge,gender,and NM stage had little difference on patient survival.While race,histological type and T-stage started to separate the curves more significantly between different groups as the survival time of patients increased.The results indicated that race,histological type and T stage had a greater impact on patient survival.（P < 0.05）.2.Retrospective data analysis of patients undergoing thoracic surgery at China-Japan Friendship Hospital（1）Patient characteristicsAmong patients with multiple primary lung cancers,38.0%（16patients）were male and 62.0%（26 patients）were female.The histological type in the primary lesion was mainly adenocarcinoma in 39cases（90.4%）,with squamous and small cell carcinoma in only 2 and 1cases.In three cases,the primary and secondary lesions had different histologic types,（two cases of adenocarcinoma and squamous carcinoma and one case of adenocarcinoma and small cell carcinoma,respectively）.Isolated lung cancer patients accounted for 39.6%（21 cases）of male patients and 60.3%（32 cases）of female patients.The histological type of lesion was mainly adenocarcinoma in 51 cases（96.2%）and squamous carcinoma in only 2 cases.Among them,age,T and M stages were statistically independent of each other in multiple primary lung cancer and isolated lung cancer,and the differences were all statistically significant.（P < 0.05）（2）Multi-factor COX analysis of multiple primary lung cancersC-index was 0.767（95% CI,0.663 to 0.871）.The HR ratio showed that histological type and M stage were the risk factors that had a greater effect on multiple primary lung cancer.And the effect of gender difference and T stage was statistically significant.（P < 0.05）.（3）External validation of prognostic prediction model: Calibrate curveThe externally validated 3-year survival rate and 5-year survival rate had good prediction results.Conclusion:This study initially summarized the characteristics of multiple primary lung cancer through the SEER database,and summarized gender,race,histological type and TNM stage as risk factors that have a strong impact on multiple primary lung cancer.In this study,a prognostic line chart was constructed for patients with multi-cell lung cancer to predict survival at the next 3 and 5 years.The line graph performed well in predicting survival,and its prediction was supported by internal validation of the C-index,ROC curve,K-M curve,and external validation of the Calibrate curve.This study also collected data from postoperative patients with multiple primary lung cancer in the Department of Thoracic Surgery at China-Japan Union Hospital,and compared them with the SEER database data to investigate the remaining problems in the SEER database data and some hot issues of multiple primary lung cancer that remain to be solved.