[The value of blood transfusion in tumor surgery. Experiences from randomized trials]

Since the first publication by Tartter in 1982 [3] multiple studies were performed concerning the association between blood transfusion and outcome in surgical oncology. The hypothesis was that blood transfusion-associated immunosuppression influences postoperative infection rates and long-term prognosis. Due to the fact that most of these studies were uncontrolled observational the decisive question about the causal relationship could not be answered so long. In randomized studies different methodological approaches were performed. Jeekel from Rotterdam and our group tried to control the allogeneic transfusion-associated immunosuppression by use of autologous transfusions. We found in our monocentric study with a very homogenous study population a significantly reduced infection rate in the autologous blood group which was not observed in the multicentric Dutch trial. The second approach was chosen by Jensen from Denmark and van de Velde from the Netherlands. They used leucocyte-depletion to create a less immunosuppressive blood product, which was then randomized and compared with conventional allogeneic transfusions. Whereas the monocentric Dutch study showed a significant effect in the pilot study and also in the later larger confirmational study, again this was not found in the multicentric Dutch trial. Concerning the association between prognosis and transfusion the results of two randomized studies are available. The multicentric study from Jeekel from Rotterdam and our monocentric study found also in this endpoint different results. Whereas in our study allogeneic transfusions were an independent risk factor demonstrated in the intention-to-treat comparison by a clear trend (p = 0.067), this could not be detected in the multicentric Dutch study. In conclusion, the results from actually available randomized studies are not compulsive. Whereas the homogeneous monocentric studies showed a significant transfusion effect for the postoperative infection rate and also for long-term prognosis, this could not be verified in two multicentric trials. Also the cumulative meta-analyses are not able to answer the decisive question concerning the causal relationship between transfusion and long-term prognosis. Therefore new and innovative study designs are mandatory.