Risk factors for early mortality among patients with gastrointestinal malignancy in the C-CAT database.

Abstract

Background: Comprehensive genomic profiling (CGP) is essential for precision medicine, but early mortality remains a concern for patients undergoing CGP. This study aimed to identify risk factors for early mortality and develop a prediction model for gastrointestinal (GI) malignancies on the basis of data from the Japanese C-CAT database.

Methods: Data from 18,657 patients with pancreatic, biliary, colorectal, and upper GI cancers were collected from the C-CAT database and retrospectively analyzed. Early mortality was defined as mortality within 90 days after CGP submission. A prediction model was constructed via weighted scoring of clinical factors, and the model was subsequently validated. Survival analysis was conducted to assess the utility of this model for prognostic stratification.

Results: The early mortality rate was 14.2%. Independent predictors of early mortality included cancer type (pancreatic/biliary), Eastern Cooperative Oncology Group performance status (ECOG-PS) ≥2, metastases, disease progression, and male sex. The prediction model stratified patients into low- (6.1%), intermediate- (17.6%), high-risk (39.2%), and very high-risk (75.6%) groups with a moderate level of discrimination (C statistic: 0.70-0.73). Survival analysis revealed that the median survival times after CGP submission for each group were 384.0 days, 199.0 days, 114.0 days, and 48.0 days, respectively. We developed a web-based application for the prediction of early mortality via the link: https://mortality-within-90days-cgp.shinyapps.io/mortality_treatment_20250130/ .

Conclusions: The prediction model effectively stratified patients on the basis of the risk of early mortality, thus supporting better patient selection and CGP timing.