Red blood cell non–ABO‐identical transfusions are harmful: really ?

P revention of ABO-incompatible red blood cell (RBC) transfusion to avoid life-threatening transfusion reactions is of utmost importance in the practice of transfusion medicine. While every effort is made to provide ABO-identical blood, donor RBCs that are ABO compatible but not ABO identical (e.g., group O donor to a group A, group B, or group AB recipient) are transfused when ABO identical donor units are not available, in emergencies, or to avoid wastage of outdated blood. Based on immunologic considerations, provision of ABO-compatible blood should not put the patient at risk of transfusion reactions. However, the possibility remains that in ABO-nonidentical transfusions, the plasma in the RBC units may contain ABOincompatible antibodies (e.g., anti-A or anti-B) that can target recipient ABO antigens on RBCs, causing hemolysis. Since platelets (PLTs) also express ABO antigens, the anti-A/B in non–ABO-identical RBC transfusions may target the corresponding ABO antigens on PLTs, possibly leading to their activation and clearance. These anti-A/B can also potentially interact with the free soluble circulating ABO antigens present in donor or recipient plasma, forming immune complexes that have been implicated as the drivers of adverse events in non–ABO-identical plasma transfusions. Since RBCs contain minimal plasma (<30 mL), the risk of adverse events due to passive transfer of incompatible ABO antibodies or soluble ABO antigens is deemed very low. Nevertheless, in an observational study of 4241 trauma patients previously published, those receiving ABO-matched blood had a lower mortality rate than patients who received uncrossmatched (universal donor group O) transfusions. The authors concluded that the increased mortality was not due to uncrossmatched RBCs itself but rather “a surrogate marker for acute, active bleeding” in patients for whom there is no time to perform a crossmatch. In a separate multihospital Canadian registry study of 18,843 non-group O patients published in this issue of TRANSFUSION, a significant increase in in-hospital mortality was associated with ABO-nonidentical RBCs compared to ABO-identical RBC transfusions only among blood group A recipients. A strength of the study is that the analyses controlled for a number of confounders including age, creatinine, hemoglobin, in-hospital intervention, the number of units transfused and their storage age, and year of admissions. The finding that transfusion of group O units increases the risk of in-hospital deaths in group A recipients, but not group B or AB, argues against donor group O RBC itself being the culprit since you would expect to see this adverse effect equally in all non-group O recipients including group B and AB. When trauma patients were excluded from the analyses, transfusion of ABO-nonidentical blood increased the risk of in-hospital deaths in group A, but reduced it in group B patients. The detrimental effect of receiving universal group O RBCs only in group A recipients is likely due to patient-specific characteristics and the authors speculate that factors present in group A recipient plasma (identity unknown) rather than donor plasma (e.g., high-titer antiA) may be the instigator of the adverse events associated with receipt of group O RBCs. Possible biologic reason(s) for why transfusion of group O RBCs to group B patients may be protective against in-hospital death in trauma patients are not discussed. This article presents a considerable amount of research analyses and results. The breadth and effort in examining this research data notwithstanding, we would like to address some of the limitations and how they may impact the findings. One major limitation is that group A has a substantially greater number of patients with a broader distribution of medical diagnoses, thereby, allowing for a more comprehensive examination in A than in groups B and AB. Although there were significant interaction terms for blood group and exposure to non-identical blood, the stratified analyses by groups B and AB did not have sufficient sample size for the majority of disease categories to calculate hazard ratios. This shortcoming makes it difficult to fully compare the effects in the three groups. The receipt of PLTs, plasma, and cryoprecipitate, which could potentially indicate disease severity, was also not factored in the analyses and as such the question remains as to whether the higher risk of in-hospital death in group A recipients was due to severity of disease in this patient group. Furthermore, it is unclear whether these products were ABO identical or not. Patients receiving ABO-compatible, but -nonidentical, plasma have been reported to have higher mortality and increased complications, including respiratory distress syndrome and sepsis, compared to patients who received ABO-identical plasma. Thus, controlling for ABO type of plasma products that patients may have doi:10.1111/trf.13505