Prediction of distant metastases after curative surgery for rectal cancer

Background. This study was performed to define selection criteria for adjuvant therapy in rectal cancer. Materials and methods. An immunohistochemical analysis using nine monoclonal antibodies against CEA, CD15s, CD44v6, DCC, E-cadherin, EGF-R, NM23, PAI-1, and P53 was performed on paraffin sections of two matched (age, gender, UICC stage [I-III], year of operation [1982-1991]) groups of patients (n = 2 x 64) with rectal carcinoma curatively treated by surgery alone. The two groups differed only with regard to metachronous distant metastatic spread. In order to exclude the influence of surgery, all patients had to meet the selection criterion "free of locoregional disease." Follow-up was prospective (median 80 months). Conventional staining procedures and immunohistochemical evaluation were used. Tumor grading and lymphatic and extramural venous invasion were also investigated. Analysis was performed with Fisher's exact test and Kaplan-Meier estimates of disease-free survival (log rank). The Cox model was used for multivariate analysis. Results. In univariate analysis only grading (P < 0.001) and extramural venous invasion (P < 0.001) correlated significantly with metachronous metastases. In multivariate analysis, beside grading (P = 0.010) and extramural venous invasion (P = 0.011), CD15s (P = 0.042) was also of significance. All other immunohistochemical markers failed. Conclusions. The histopathological parameters grading and extramural venous invasion appear to be acceptable predictors of metachronous distant spread in curatively resected rectal cancer. In contrast to the immunohistochemical markers, grading seems to better reflect the individual tumor phenotype and its behavior. (C) 2002 Elsevier Science (USA).