Large bowel cancer: the effect of perioperative blood transfusion on outcome.

Perioperative blood transfusion has been reported to adversely affect survival in cancer patients, but the evidence is inconclusive and may be an epiphenomenon. From the Large Bowel Cancer Project, 961 patients who underwent curative resection and left hospital alive have been reviewed to compare the effect of perioperative blood transfusion on outcome; 591 patients (61%) had been given a blood transfusion while 370 (39%) had not been transfused. Some clinical variables were equally distributed between the two groups; ie age, sex, obstruction, perforation, tumour differentiation. Three other variables known to influence patient prognosis were not equally distributed, ie tumour site, Dukes' stage and tumour mobility. Patients with tumours of the rectum and rectosigmoid, with Dukes' stage C lesions and with some degree of tumour fixation were more likely to have received blood transfusions. Using the logrank method of multivariate analysis to allow for differences in distribution of all those variables known to affect prognosis, there was no survival disadvantage for those patients who had received perioperative blood transfusion. Furthermore, there were no overall differences between the two groups of patients in their risk of developing local tumour recurrence or distant metastases. The distribution of metastases differed: in the 'transfused' group only 37% of distant metastases were found in the liver, while 71% were found in this site in the 'not transfused' group (chi 2 = 18.46, d.f. = 1, P less than 0.001). By contrast, there was a larger proportion of patients with lung metastases in the transfused group (27% vs 11%) (chi 2 = 5.59, d.f. = 1, P less than 0.05). Therefore, these data do not support the concept of an overall deleterious effect of blood transfusion on patient survival, but suggest that blood given in the perioperative period may change the biology of the metastatic process.