Nomogram of conditional survival probability of long-term Survival for Metastatic Colorectal Cancer: A Real-World Data Retrospective Cohort Study from SEER database

Abstract

BackgroundMany patients with metastatic colorectal cancer (mCRC) have better prognosis than the prediction at diagnosis. Compared with invariable traditional Kaplan Meier assessment, conditional survival (CS) assessment has become a more accurate and informative assessment method to predict survival time. Materials and methodsPatients with mCRC between 2010 and 2015 were extracted from Surveillance, Epidemiology and End Results linked database. CS analysis was applied to depict exact survival for patients who have survived for specific year and standardized difference (d) was used to evaluate the differences between subgroups in CS analysis. Based on variables selected by Lasso analysis, nomograms for each year after diagnosis were fitted to estimate 3-year survival of stage IV CRC, respectively. ResultsOf 9732 patients, overall actuarial survival (OS) decreased from 24% at 4-year to 16% at 6-year, while corresponding 3-year CS (CS3) increased from 33% at 1-year to 48% at 3-year. Overall, CS3 was higher than corresponding actuarial survival. All clinicopathological characteristics were associated with actuarial survival (p < 0.05). However, in CS3 analysis, survival difference caused by gender, race and tumor size gradually disappeared over time (|d|>0.1→ |d|<0.1). Furthermore, survival difference caused by histological type, brain metastasis and chemotherapy reversed over time (d > 0→d<0 or d<0→d>0). Based on lasso analysis, nomograms for 1st, 2nd and 3rd year after diagnosis were conducted respectively. The AUC of nomogram for 1st year was 0.705, for 2nd year was 0.675, and for 3rd year was 0.648. ConclusionPatients with mCRC demonstrated a substantial increase in CS over time. Risk factors collected at diagnosis may change gradually. Nomograms constructed by survival time can predict more accurate survival for patients with mCRC. Conditional survival assessments provide important quantitative information about the probability of survival and are therefore of great value to patients and health care professionals.