Preoperative prediction of late cancer-specific deaths in patients with rectal and rectosigmoid carcinoma.

The possibility of predicting late cancer-specific deaths from (a) the preoperative serum levels of three tumour markers, carcinoembryonic antigen (CEA), tissue polypeptide antigen (TPA) and an antigen defined by the C-50 antibody (CA-50), from (b) one clinical factor of independent prognostic relevance, polypoid tumour growth, and from (c) Dukes' stage was evaluated in 276 patients with rectal carcinoma operated upon with curative intent ("potentially curable"), and in the 251 of those patients who were considered to be "potentially cured" after surgery. Using the Cox regression model, the preoperative serum levels of the tumour markers strongly predicted the cancer-specific mortality within the first year after surgery. This ability of S-CEA and S-CA-50 diminished for the mortality during the second year after surgery, and virtually disappeared thereafter. The ability of S-TPA to predict cancer-specific deaths did not change as dramatically with time as that of the other two markers, particularly in the group of "potentially cured" patients. Patients with polypoid tumour growth had a good prognosis which did not appear to change with time. Similarly, the prognostic information provided by Dukes' staging system was valid at all studied time intervals after surgery, although it declined after the second year. The importance of these results in relation to the selection of patients for adjuvant treatment is discussed.