Perioperative blood transfusions for the recurrence of colorectal cancer

The improvement of renal allograft survival by pre-transplantation transfusions alerted the medical community to the potential detrimental effect of transfusions in patients being treated for cancer.

The present meta-analysis aims to evaluate the role of perioperative blood transfusions (PBT) on colorectal cancer recurrence. This is accomplished by validating the results of a previously published meta-analysis (Amato 1998); and by updating it to December 2004.

Published papers were retrieved using Medline, EMBASE, the Cochrane Library, controlled trials web-based registries, or the CCG Trial Database. The search strategy used was: {colon OR rectal OR colorectal} WITH {cancer OR tumor OR neoplasm} AND transfusion. The tendency not to publish negative trials was balanced by inspecting the proceedings of international congresses.

Patients undergoing curative resection of colorectal cancer (classified either as Dukes stages A-C, Astler-Coller stages A-C2, or TNM stages T1-3a/N0-1/M0) were included if they had received any amount of blood products within one month of surgery. Excluded were patients with distant metastases at surgery, and studies with short follow-up or with no data.

A specific form was developed for data collection. Data extraction was cross-checked, using the most recent publication in case of repetitive ones. Papers' quality was ranked using the method by Evans and Pollock. Odds ratios (OR, with 95% confidence intervals) were computed for each study, and pooled estimates were generated by RevMan (version 4.2). When available, data were stratified for risk factors of cancer recurrence.

The findings of the 1998 meta-analysis were confirmed, with small variations in some estimates. Updating it through December 2004 led to the identification of 237 references. Two-hundred and one of them were excluded because they analyzed survival (n=22), were repetitive (n=26), letters/reviews (n=66) or had no data (n=87). Thirty-six studies on 12,127 patients were included: 23 showed a detrimental effect of PBT; 22 used also multivariable analyses, and 14 found PBT to be an independent prognostic factor. Pooled estimates of PBT effect on colorectal cancer recurrence yielded overall OR of 1.42 (95% CI, 1.20 to 1.67) against transfused patients in randomized controlled studies. Stratified meta-analyses confirmed these findings, also when stratifying patients by site and stage of disease. The PBT effect was observed regardless of timing, type, and in a dose-related fashion, although heterogeneity was detected. Data on surgical techniques was not available for further analysis.

This updated meta-analysis confirms the previous findings. All analyses support the hypothesis that PBT have a detrimental effect on the recurrence of curable colorectal cancers. However, since heterogeneity was detected and conclusions on the effect of surgical technique could not be drawn, a causal relationship cannot still be claimed. Carefully restricted indications for PBT seems necessary.