Comparison of classic and quantitative prognostic factors in colorectal cancer.

Despite the many pathologic and clinical variables shown to influence survival rates in patients with colorectal cancer, the prediction of outcome after curative resection is still not completely reliable. Dukes-classification has been historically the strongest prognostic indicator and the most frequently employed method. In this study our aim was to search for significant independent histological and morphometric factors that could possibly be used in predicting the outcome of different stages of colorectal cancer. We analysed the clinical follow-up data of 308 patients with colorectal adenocarcinoma (followed-up for a mean of 14.4 years). The clinical findings were correlated to histological and morphometric factors to establish their value as predictors of colorectal adenocarcinoma Clinical, histological and morphometric factors were significantly interrelated. The large invasive tumours had larger nuclei, a larger variation in nuclear size, and they were also rapidly proliferating. In univariate survival analysis Dukes, the mean nuclear area (NA), standard deviation of nuclear area (SDNA), nuclear perimetry (PE) and the mean of the longest nuclear axis (Dmax) were the most important predictors of recurrence-free survival (RFS). TNM-categories, Dukes, histological grade and all the quantitative variables were significant predictors of cancer related survival. In a multivariate analysis of Tl-4N0-3M0 tumours (n = 164) N-category, nuclear area and the vear of operation and in local tumours (Tl-2N0M0) (n = 70) Dmax and Dukes predicted RFS. Important determinants of survival in all 269 cases were M-category. Dukes and Dmax and in local tumours (Tl-2N0M0) (n = 67) Dmax. These results indicate that although an accurate prognostic evaluation of colorectal carcinomas can be based on TNM- and Dukes-classification, nuclear morphometry can give additional prognostic information.