• Surgery
• Ovarian Cancer
• Postoperative complications
• Surgical Wound Infection
• Quality of Life (PRO)/Palliative Care

“It is no easy task to understand unfamiliar blood” Friedrich Nietzsche

Perioperative transfusion of packed red blood cells is frequently used in patients undergoing surgery for advanced ovarian cancer to improve tissue oxygen delivery and mitigate symptoms of anemia. In the early 1980s the observation that packed red cell transfusion in renal transplant recipients improved cadaveric graft survival, and the implication that transfusion was immunosuppressive, led investigators to question whether transfusing patients with cancer might adversely affect prognosis.1 2 A plethora of studies supported this hypothesis and led to the concept of ‘transfusion-related immune modulation’ (TRIM).2 TRIM is postulated to involve adaptive immunological mechanisms affecting T and B cells, and the innate immune system including alteration of natural killer cells, and macrophage and monocyte function. Meta-analyses of patients undergoing surgery for colorectal cancer have found that the administration of blood products is associated with shorter recurrence-free survival and overall survival.3 Studies investigating the impact of blood transfusion on survival in patients undergoing surgery for ovarian cancer have had inconsistent findings and are limited by their single center, retrospective design.

Prescott et al present the results of an ancillary analysis of the European Organization for Research and Treatment (EORTC) 55 971 phase III trial.4 EORTC 55971 was a randomized non-inferiority trial that compared primary debulking surgery followed by platinum-based chemotherapy to platinum-based neoadjuvant chemotherapy followed by interval debulking surgery in patients with stage IIIC and IV ovarian carcinoma.5 Prescott et al categorized participants in the per-protocol analysis by perioperative transfusion defined as receipt of transfusion intraoperatively to 28 days post-operatively. Over half of 612 participants with available data received a transfusion. No difference in median overall survival was observed between the no transfusion and transfusion groups (34.0 vs 35.2 months respectively (p=0.97); adjusted hazard ratio (HR) for death 1.18 (95% confidence interval (CI) 0.94, 1.48) and no difference in median progression free survival (13.6 vs 12.6 months respectively (p=0.99); adjusted HR for progression 1.14 (95% CI 0.91, 1.43)). No differences were observed in global quality of life. Notably, grade 3 and 4 surgical site infections were more frequent in participants who were transfused.

The study has important limitations that are appropriately acknowledged by the authors and of which the reader should be cognizant. Findings of post hoc analyses need to be interpreted with caution because they are prone to type I errors (rejecting the null hypothesis when it is true) and to the observed consequences of interventions being incorrectly construed as having causal rather than associative effects. Although the authors included residual disease and operative time as surrogates for surgical complexity in a multivariate model, surgical complexity was not an independent variable in their analysis. Transfusion should not be viewed as the cause of increased surgical site infections because more patients undergoing primary cytoreduction were transfused compared with patients undergoing interval debulking surgery and the former were likely to have been more complex procedures. Variables such as estimated blood loss, number of transfusions and packed cell units transfused, and participant hemoglobin levels were not included in the analysis. Further, differences in quality of life data compliance among recruiting sites in EORTC 55971 limits interpretation of the quality of life findings.

Despite their limitations, post hoc analyses that are well conducted may yield valuable insights and Prescott et al are to be congratulated on making an important contribution to the literature. Their study is the first to have used data from a large, multicenter randomized trial to assess the impact of transfusion on survival, quality of life and perioperative outcomes. An important finding was the high variation in transfusion practice between countries that recruited to EORTC 55971, ranging from 17–91% for countries that entered five or more participants. As noted by the authors, interventions with high degrees of variation often indicate either a lack of implementation of evidence-based guidelines, or a lack of evidence, or both, and present opportunities to reduce costs and improve health care delivery through generation and implementation of evidence.

Transfusion is, of course, critically important in the context of severe intraoperative hemorrhage, but excessive use of transfusion is potentially problematic and should be limited. Preoperative patient optimization including correcting anemia and nutritional screening are valuable strategies to minimize transfusion and should be used according to established guidelines.6 Prescott et al provide some reassurance that judicious use of transfusion is not associated with worse oncologic outcomes in patients undergoing surgery for advanced ovarian cancer, but their findings should be viewed as hypothesis generating. Future surgical trial protocols that include prespecified statistical analyses to assess the impact of transfusion on survival and quality of life outcomes, and meta-analysis of individual participant data from such trials, will contribute novel evidence to inform decisions about perioperative transfusion in patients with advanced ovarian cancer.