The reappraisal of prognostic classifictions for colorectal cancer

Publisher Website
Google Scholar
Abstract
The prognostic usefulness of four variables and four staging systems were evaluated by the Akaike Information Criterion (AIC) and conventional statistical tests for survival time analysis (log rank test, generalized Kruskal-Wallis test, and linear trend test), based on the data of 394 colorectal cancer patients who had been followed for more than 5 years after surgery. The four variables were the histopathologic findings of cancer penetration through the bowel wall (w), lymphnode metastasis (n), intraoperative observations of liver metastasis (H), and peritoneal dissemination (P). They were graded according to the general rules proposed by the Japanese Society for Cancer of the Colon and Rectum (JRSCCR). The stagin systems were Turnbull's modification of Duke's classification, UICC's TNM system, the Japanese classification proposed by JRSCCR, and the authors' modification of the latter. AIC was calculated by changing the cutoff points of survival time, as well as by reclassifying predictor variables into two categories by different levels. The results of the standard tests differed from those of AIC in the evaluation of the four variables. The former statistics showed that the intraoperative observations were of more prognostic importance than the histopathologic findings, whereas AIC revealed that the histopathologic variables have more prognostic information except for prediction of early outcome. Further analysis of the w-factor showed that survival is more drastically changed by serosal involvement or extensive extramuscular invasion than by penetration of the muscularis propria. Analysis of the n-factor indicated that a major change in survival time occurs twice, i.e., when lymphonode metastasis occurs and when it extends beyound the paracolic level. Compared with single variables, the staging system using multiple factors gives more information on both early and long-term outcomes. The authors conclude that AIC analysis provides prognostic information that cannot be obtained with conventional tests.