The association between perioperative blood transfusion and survival of patients with colorectal cancer

Background. The authors undertook this study to test the hypothesis that perioperative blood transfusion has an adverse effect on survival of patients with colorectal cancer. Methods. A retrospective analysis was performed on 725 patients who underwent curative resection for Dukes Stage B and C colorectal cancers in our institution between 1981–1985 and who were followed for 5–11 years. Results. Rectal cancers occurred more frequently in the transfused than in the nontransfused patients (64% versus 47%; P = 0.0001). In the patients with colon cancer, no significant effect of transfusion on recurrence‐free survival was seen (P = 0.8, log‐rank; P = 0.49, Cox regression analysis). The cumulative 5‐year survival rate was 77% (95% confidence interval, 69–85%) for the nontransfused and 78% (72–83%) for the transfused patients. In patients with rectal cancer, the 5‐year survival rate was 79% (67–87%) for the nontransfused and 67% (59–71%) for the transfused patients. A significant association was noted in patients with Stage B cancer (P = 0.002, log‐rank; P = 0.02, Cox regression analysis), but not in those with Stage C cancer (P = 0.05, log‐rank; P = 0.15, Cox regression analysis). In patients with Stage B rectal cancer, more frequent abdominoperineal resections (APR) were performed among the transfused patients (65% versus 32%; P = 0.0001). This subgroup was further stratified by operative procedure and reanalyzed using the same Cox regression model. Transfusion had no effect on survival of patients treated by APR (P = 0.31) or of those having a sphincter‐saving procedure (P = 0.53). Conclusion. The seemingly adverse effect of perioperative blood transfusion on the survival of patients with colorectal cancer may be explained by other covariates.