Construction of a survival prediction columnar plot for postoperative rectal cancer patients based on SEER database and its validation
Objective The influencing factors for the overall survival of postoperative rectal cancer patients were analyzed,and the nomogram model of risk prediction was constructed.Method A total of 14,201 patients diagnosed with rectal cancer and undergoing surgical treatment,from 2010 to 2015,was screened from the SEER(surveillance,epidemiology,and end results)database.According to the ratio of 7∶3,9,941 cases were randomly divided into a training set and 4,260 cases in a validation set;In addition,376 patients with postoperative rectal cancer consistent with the SEER database extraction criteria from January 1,2017 to December 31,2021 in the Second Hospital of Lanzhou University were collected as an external validation set.Univariate and multivariate COX regression analysis was used to determine the independent risk factors affecting the overall survival of rectal cancer patients after surgery,and a nomogram prediction model was constructed.The model's discriminatory ability was evaluated through the C-index and area under curve(AUC),while the calibration curve evaluated the degree of closeness between the predicted value and the actu-al value(ideal curve)of the predictive model.Results Multivariate COX regression analysis showed that there were 13 independent risk factors for the prognosis of postoperative rectal cancer patients,including general demographics(gender,age,marital status),tumor characteristics(primary tumor site,tumor stage,T stage,N stage,nerve invasion,carcinoembryonic antigen),distant metastasis(liver metastasis,lung metastasis,bone metastasis)and chemotherapy(P<0.05).A column chart model for predicting the survival of postoperative rec-tal cancer patients was constructed based on these factors.The training and validation sets of C-index,AUC and calibration curves showed that the predicted value of the model was in good agreement with the actual value.Conclusion In this study,we constructed a good column-line graph for survival prediction of postopera-tive rectal cancer patients based on the SEER database,and for the first time,we analyzed the further division of rectal cancer tumors into their primary sites to conclude that tumors occurring in the rectum have a worse prognosis than tumors at the rectosigmoid junction.
rectal cancersurveillance,epidemiology and end results databasenomogramprognosis
国家科技期刊平台
NSTL
万方数据
