Development of a Nomogram for Predicting Survival in Microsatellite Stable Patients with Resected Colorectal Cancer

Y. Li,Jiayuan Peng,Dan Huang,Xu Ma,San-Jun Cai

DOI: https://doi.org/10.1093/annonc/mdy431.037

IF: 51.769

2018-01-01

Annals of Oncology

Abstract:Background: The benefits of adjuvant chemotherapy remain controversial in microsatellite stable (MSS) stage II resected colorectal cancer (CRC) patients. In this study, we constructed an overall survival (OS) prediction model for this subgroup, integrating a combination of molecular and clinical predictors. Methods: Variables with p-values < 0.05 were entered into multivariate analyses using Cox stepwise regression model to select independent predictors as input for OS prediction nomogram. A final model was selected using a backward step-down process, which used Akaike Information Criterion as a stopping rule. The probability of 3-year and 5-year survival can be obtained by summing up the total score and locating it on the total point scale. Results: We performed targeted sequencing on surgically resected tumor tissue of 122 stage II MSS patients, defined as having less than 15% length-instable loci, using a panel which allows for simultaneous detection of MSI status and mutation in 41 CRC-related genes. Among them, 23, 30 and 69 patients were diagnosed with proximal, distal and rectal cancer, respectively. To predict the probability of 3-year and 5-year survival, we constructed a nomogram incorporating the significant prognostic factors, including APC, ATM, BRAF, PTEN, TP53 (LOF), mutation count (high: >3, low: < =3), age, CEA, and the location of the tumor. The actual and predicted survivals were in an excellent agreement, reflected by a C-index of 0.887 (95% CI: 0.816-0.957). Furthermore, Kaplan-Meier curves for survival outcomes showed significant distinction (p < 0.001) after stratifying our cohort into low, median and high risk groups according to total score obtained from our nomogram. Our data also demonstrated high risk patients who received adjuvant chemotherapy are associated with a better OS (p = 0.004). Low and median risk patients did not benefit from adjuvant chemotherapy, reflected by comparable OS. Conclusions: We developed a nomogram model for predicting survival of MSS patients with stage II resected CRC. It can potentially serve as complementary method for clinicians to identify subgroups necessitating adjuvant therapy. Legal entity responsible for the study: Fudan University Shanghai Cancer Center. Funding: Has not received any funding. Disclosure: All authors have declared no conflicts of interest.