Abstract
Purpose: To evaluate the effect of perioperative blood transfusions on colorectal cancer recurrence and patient survival. Patients and Methods: A total of 1,051 patients treated with curative surgery for stage II or III colorectal adenocarcinoma were retrospectively studied for the effect of perioperative blood transfusions on disease recurrence and patient survival. Forty-two percent of patients received perioperative blood components. Results: Perioperative transfusions had no effect on disease progression in univariate or multivariate analysis. Tumor stage (P = .0001), locally advanced tumor characteristics (adherence, involvement of adjacent structure, or perforation; P = .0001), location (rectal v colon; P = .0002), grade (P <.001), and cell kinetic profile (nondiploid or high percent synthetic phase [%S] + percent gap 2 mitosis phase [%G2M]; P = .0003) were the most powerful independent predictors of tumor recurrence. Use of transfusions was associated with an adverse effect on overall survival (P <.004) using multivariate analysis, as well as tumor stage (P = .0001), location(P = .004), grade (P = .001), patient age (P = .0001), sex (P <.04), and cell kinetic profile (P = .0001). In further evaluation of the prognostic effects of transfusions, there was no increased risk of disease recurrence after whole-blood transfusions (P = .14) as compared with packed RBC or no transfusions, although the disease-specific survival for patients who received whole blood was lower than for nontransfused patients (P <.005) and patients who received other blood components (P <.03). Conclusion: With transfusion practices that use blood components, most commonly RBCs, medically indicated transfusions to patients with colorectal carcinoma seem to have no impact on disease recurrence. The adverse impact of transfusions on cancer patient survival is more likely due to ether unevaluated tumor variables or underlying illness rather than tumor recurrence enhancement by immunosuppression induced by transfusion of blood components.