ABO and platelet transfusion revisited.

Historically, ABO compatibility between donor and recipient has been considered of minor importance for platelet transfusion. However, in a recent randomized trial we showed that provision of only ABO-identical platelets was associated with a significantly higher corrected count increment (CCI) in the early transfusions and with a twofold reduction in refractoriness, as compared with platelets given unmatched for ABO. In older studies this phenomenon was not observed. We wondered whether in these previous studies the use of the conventional term "ABO compatible", whereby ABO-identical platelet transfusions are grouped with those that are merely ABO compatible, might have obscured the advantages of the ABO-identical platelets. In this paper the CCI from our original study are compared with what would have been found had the conventional terminology of "ABO compatible" and "ABO incompatible" been used. In our original study the mean CCI in patients receiving only ABO-identical platelets was 64% higher than that achieved with ABO-unmatched products. However, when the transfusions were reanalyzed according to conventional terminology of ABO "compatible" and "incompatible," the actual benefits of ABO-identical platelets were no longer detected. Thus, had we used the conventional terminology in the original study we would have come to completely different conclusions about the importance of ABO in platelet transfusion. We hypothesize that transfusion of ABO-incompatible plasma leads to the formation of immune complexes that can destroy compatible platelets by indirect mechanisms. This model predicts that the success of a transfusion could be affected by previous incompatible transfusions. When transfusions were classified as to whether they were first, second, or third unmatched transfusion, increasing numbers of transfusions of ABO-incompatible plasma were associated with progressively poorer mean increments. The same was true for platelet ABO-incompatible transfusions. In contrast, increasing numbers of ABO-identical transfusions were not associated with poorer increments. The failure to appreciate that (a) transfusions containing ABO-incompatible plasma yield poor increments and (b) the effects of ABO-nonidentical transfusions are cumulative, and therefore affect the success of subsequent identical transfusions, may have led in the past to the incorrect conclusion that ABO matching was of minor importance in repetitively transfused patients.