Preoperative clinical and pathological variables in prognostic evaluation of patients with rectal cancer. A prospective study of 327 consecutive patients.

The prognostic information provided by a number of easily identified and preoperatively available characteristics was recorded prospectively and evaluated in 327 consecutive patients with rectal carcinoma. With use of the Cox regression model, the two variables indicating surgical non-curability, namely immobility of the tumour to the adjacent tissues and preoperatively diagnosed metastatic spread, showed the strongest relation to prognosis. Other variables predictive of a poorer outcome in all patients were abnormal liver function tests, large tumour size (number of degrees of the bowel wall circumference affected by the tumour), non-polypoid tumour growth, tumour ulceration, tumour stricture, tumour growth anteriorly and low histological differentiation in the preoperative biopsy. These variables, together with age, also gave information concerning the group of patients of interest for pre- or peroperatively initiated adjuvant therapy, i.e. patients potentially curable by surgery (locally resectable tumour and no known metastases). In this group, the two preoperatively available variables with the best prognostic value, polypoid tumour growth and age, appeared inferior to the postoperatively determined tumour stage (Dukes' staging). However, some clinical variables gave information additional to that provided by the tumour stage.