Analysis Of Prognostic Factors And Establishment Of Survival Prediction Model Of Pulmonary Sarcomatoid Carcinoma Based On SEER Database

Background and objective:Pulmonary sarcomatoid carcinoma(PSC)is a rare and aggressive tumor,accounting for only 0.1-0.4%of non-small cell lung cancer(NSCLC).Due to the aggressive nature of PSC,even with early surgery,the patient’s prognosis is still poor,and local infiltration and distant metastasis are extremely prone to occur.Due to its rarity and heterogeneity,clinical studies on PSC are mostly case reports or small series of reports,and the prognosis assessment of PSC patients is usually only based on the personal experience of clinicians.We use data from the Surveillance,Epidemiology,and End Results(SEER)database to explore the factors affecting the prognosis of PSC patients,and construct a nomogram to predict the 3-year and 5-year survival rates of PSC patients,to help clinicians evaluate the prognosis based on the patient’s personal situation.Methods:The data of patients diagnosed with PSC from 1988 to 2015 were extracted from the SEER database,and their clinicopathological characteristics were summarized.The PSC patients were divided into a modeling group and a test group in a ratio of 7:3 using the"Caret package"of RStudio software.Univariate and multivariate Cox regression analysis was performed on patients in the modeling group to determine the independent factors affecting the prognosis of PSC patients.The results of the multivariate analysis of the modeling group were used to construct a nomogram to predict 3-years and 5-year overall survival(OS)rate in PSC patients.The prediction ability and consistency of the model were evaluated by concordance index,receiver operating characteristic curve(ROC curve)and calibration curve in the modeling group and the test group,respectively.The Kaplan-Meier method was used for survival analysis,the Graph Pad software was used to draw the survival curve,and the Log-rank method was used to test the survival difference between the groups.The difference of P value<0.05 was statistically significant.Results:PSC patients accounted for 0.43%of NSCLC.The median survival time of PSC patients and traditional NSCLC patients were 6 months and 11 months,respectively,and the prognosis of PSC patients was worse than that of traditional NSCLC.The median age of PSC patients was 67 years,58.7%were males,80.3%were whites,54.5%were married,and 51.9%were the primary lesions located in the upper lobe of the lung.Most patients are poorly differentiated or undifferentiated.A total of75.3%of patients were in stage III/IV.The prognosis of PSC patients of different subtypes were also different.The median survival time of 132 months patients with pulmonary blastoma,and 10 months for pleomorphic carcinoma,8 months for carcinosarcoma,6 months for giant cell carcinoma and 5 months for spindle cell carcinoma,respectively.The median survival time of PSC patients in the stage I-IV modeling group was 71 months,19 months,9 months,and 2 months.Multivariate regression analysis based on the modeling group found that age,gender,histological type,tumor diameter over 50 millimeter,TNM stage,surgery,radiotherapy,and chemotherapy were independent factors that affect the prognosis of patients.Based on the results of the multivariate analysis of the modeling group,a nomogram prediction model was constructed and verified.In the modeling group,the C index was calculated to be 0.790(95%CI:0.776-0.804),and the ROC curve of the prediction model of the modeling group showed that the area under the curve(AUC)for 3 years and 5 years was0.872 and 0.865,respectively.In the test group,the C index calculated was0.781(95%CI:0.759-0.803),and the ROC curve predicted by the nomogram showed that the AUC of 3-year and 5-year overall survival is 0.860 and 0.888.The 3-year and5-year calibration curves predicted by the nomogram of the modeling group and the test module are close to 45~°indicating that the predicted survival rate is basically consistent with the actual survival rate,which ensured the reliability of the nomogram prediction model.Conclusion:(1)Compared with traditional NSLCL,PSC has a low incidence,poor differentiation,and poor prognosis.PSC is more common in elderly male patients,and most patients are already at an advanced stage when diagnosed.Multivariate analysis showed that age,gender,histological type,tumor diameter over 50 millimeter,TNM stage,surgery,radiotherapy,and chemotherapy were independent factors that affect the prognosis of patients.(2)The nomogram established based on the results of multivariate analysis could predict the 3-year and 5-year survival rates of PSC patients.The C index,ROC curve and calibration curve verified the high accuracy and consistency of the prediction model,which could provide a reference for clinicians to judge the prognosis of patients with PSC.