Establishment Of Predictive Models For Overall Survival And Cancer-specific Survival Of Patients Withadvanced Gastric Cancer

[Objective]Gastric cancer is a malignant tumorworldwide.Many patients suffer from distant metastasis at the time of diagnosis,which seriously affectsthe quality of life and shortenssurvival time.So far,individualized treatment of patients with advanced gastric cancer still plagues clinicians.Doctors usually develop a treatment plan based on the patient’s expected survival time.Therefore,accurate and scientific prediction of the survival time of patients with advanced gastric cancer will benefit both doctors and patients.This study aimsto construct and validate a prognostic model-nomogram for estimating overall survival(OS)and cancer-specific survival(CSS)of patients with advanced gastric cancer.[Methods]Statistical and clinical information onpatients diagnosed with advanced gastric cancer between 2010 and 2014 was extracted from the Surveillance,Epidemiology,and End Results(SEER)database.Seventy percent of patients were randomly assigned to the training set and thirty percentof patients were randomly assigned to the validation set.Firstly,the statistically significant variables(p<0.05)were included in the multivariate Cox proportional hazards analysis after Kaplan-Meierunivariate survivalanalysis,and then independent variables associated with overall survival or cancer-specific survival were obtained.Secondly,using these variables to construct a 1-year and 3-year overall survival or cancer-specific survival nomogram by R language.Thirdly,the predictiveability of models was verified internally(training set)and externally(validation set)through the concordance index(c-index)and the calibration plots.Finally,the receiver operating characteristic(ROC)curves and area under the curves(AUC)were created to compare the predictive performance of the nomograms and the routinestaging systems.[Results]Information on 2687 patients with advanced gastric cancer who met the screening criteria was collected from the SEER database.Age,household income,T stage,histological Grade,bone metastasis,liver metastasis,lung metastasis and surgical treatment were considered to be independently associated with overall survival or cancer-specific survival,and nomograms were successfully constructed.In the internal validation,the c-indexes of the overall survival nomogram and the cancer-specific survival nomogram were 0.630 and 0.632,respectively.In the external validation,the c-indexes of the overall nomogram and the cancer-specific nomogram were 0.628 and 0.632,respectively.Both the internal and external calibration plots tended to 45 degrees.The AUCs of nomograms were higher than those of routinestaging systems.[Conclusions]1)Older age,lower household income,higher T stage,higher histological Grade,bone metastasis,liver metastasis,lung metastasis and no surgical treatment were negatively correlated with the prognosis of patients with advanced gastric cancer;2)The prognostic nomograms for OS and CSS of patients with advanced gastric cancer could predict the outcomesaccurately and intuitively;3)The nomograms could predict outcomes of patients with advanced gastric cancer better than routinestaging systems.