Appraisal of a model for prediction of prognosis in advanced colorectal cancer.

Previously identified prognostic factors in advanced colorectal cancer were tested in an independent population for their relationship to survival by univariate and multivariate analyses. The new population comprised 198 patients included in a randomised chemotherapy trial. The earlier identified prognostic variables were: (1) haemoglobin level (B-Hb), (2) disease-free interval, (3) Karnofsky performance status (KPS), (4) number of symptoms, and (5) whether the primary tumour was removed or not. In the new population, variables (1-3) had significant relationships to survival in both univariate and multivariate analyses, whereas variable (4) was significant only in the univariate analysis. Variable (5) was not significantly related to survival in any analysis. When a group of additional variables (white blood cell count, B-thrombocytes, S-creatinine and liver function tests) was included, S-aspartate aminotransferase (S-ASAT) was a strong predictor of survival. The independent predictive value of B-Hb, disease-free interval and KPS was confirmed in this study. S-ASAT might be an additional important prognostic factor in advanced colorectal cancer. However, the results of this study indicate that any new prognostic factor should be viewed as preliminary until verified in an independent population.