Is It Time to Reconsider the Concepts of “Universal Donor” and “ABO Compatible” Transfusions?

The concept of “universal donor” group O red blood cells (RBC) and “universal donor” group AB plasma antedate the modern era of evidence-based medicine. The absence of gross hemolysis led to the assumption that the small amounts of incompatible anti-A and anti-B (in O RBC concentrates) and the absence of ABO antibodies (in AB plasma) are almost always benign and of no consequence for the recipient. These assumptions, never tested in observational studies, much less randomized trials, have been called into question in recent decades. Evidence that ABO compatible (but nonidentical) transfusions of blood components may be less effective and less safe than ABO identical has accumulated steadily over the past 2 decades. Although transfusions of ABO “compatible” blood components have long been considered safe to use, ABO identical transfusions have been associated with better outcomes in terms of morbidity and mortality in multiple studies.1–4 Clinical studies, bedside observations, randomized clinical trials, and in vitro studies strongly suggest that ABO identical, and not ABO “compatible,” transfusions are likely the safest practice (Table).1,8,10,13–18Table.: A Summarized List of Some of the ABO Identical and ABO Nonidentical StudiesMost blood components, regardless of the methods of collection, contain significant amounts of donor plasma carrying soluble A and/or B antigens (except group O), in addition to anti-A and/or anti-B (except group AB). The amount of plasma in each blood component varies from 200 to 300 mL in platelet or plasma units to 30–70 mL in RBC concentrates, to 10–20 mL in a unit of cryoprecipitate. In addition, cellular components (ie, RBCs and platelets) express ABO antigens on their membranes. The recipient frequently possesses cell surface and soluble ABO antigens in considerable amounts (except group O) and anti-A/B (except group AB). The transfused anti-A and anti-B are rapidly bound to circulating soluble A and B antigens and to cell surface A and B antigens. Likewise, the transfused soluble antigen(s) also binds to circulating anti-A/B forming soluble high-molecular weight immune complexes.7 These complexes circulate for hours to days after nonidentical transfusions and can bind to RBCs, causing hemolysis, and to platelets, resulting in their activation and premature removal from the circulation by monocytes and macrophages via their FC receptors.7,19 The formation of anti-A and/or anti-B immune complexes may initiate inflammation, tissue injury, and/or modulate cellular immunity and potentially predispose to infection.8 Depending on their avidity, transfused anti-A and/or anti-B present in ABO nonidentical components can also bind to the tissue-bound A and B antigens, resulting in RBC hemolysis, platelet activation, and alteration of the normal hemostasis.20 In addition, similar to human leukocyte antigen antigens, A and B antigens are widely expressed on the endothelium, thus transfused anti-A/B are likely capable of causing direct injury to organs. The acute rejection of ABO-incompatible organs and bone marrows is well documented in literature.9,21,22 ABO NONIDENTICAL RBC TRANSFUSIONS The presence of anti-A and anti-B in ABO nonidentical RBC has been associated with higher morbidity and mortality in many subsets of patients.13,14,16 Most recently, in a large multihospital registry (Transfusion Registry for Utilization, Surveillance, and Tracking), the association between exposures to at least 1 unit of ABO nonidentical RBC increased in-hospital mortality.13 Pai et al13 reported that of all nongroup O patients (n = 18,843) admitted between 2002 and 2011, group A patients who received group O RBCs had significantly increased risk of in-hospital death (relative risk, 1.79; 95% confidence interval, 1.20–2.67; P = .005). This remarkable and important finding raises for the first time the issue of whether group O RBCs are truly “universal donor,” at least for group A recipients. Possible contributing factors are the higher titer, avidity, and biologic activity of anti-A, compared with anti-B, and/or the formation of the circulating immune complexes, or the induction of iatrogenic “autoimmune like process.”7 ABO NONIDENTICAL PLATELET TRANSFUSIONS The importance of ABO compatibility in platelet transfusion, in particular, has long been a subject of debate. Although it is not usual practice to transfuse ABO nonidentical plasma (usually “universal” donor group AB plasma), with the exception of trauma patients of unknown ABO blood group, platelets are often transfused across ABO barriers due to the difficulty in inventory management of a component with a short shelf-life. Approximately 10%–40% of all platelet transfusions in the United States are ABO nonidentical.14,15 Greater morbidity and mortality have been reported in several observational studies of ABO nonidentical platelet transfusion.1,8,10,16–18 Early randomized trials, although small, clearly demonstrated that use of ABO nonidentical platelets leads to dramatic increases in platelet transfusion refractoriness, a potentially lethal outcome.5,6 Intentional transfusion of ABO-mismatched platelets increased refractoriness by 5-fold and increased sensitization to human leukocyte antigen as well. The use of unselected ABO blood group platelets doubled the refractoriness rate compared with use of ABO identical platelets. This evidence was largely ignored because of the long standing conviction that ABO was not clinically important, and due to difficulties in managing blood bank inventories to provide only ABO identical platelets. Severe, even fatal hemolytic transfusion reactions have been reported after transfusion of ABO nonidentical (compatible and incompatible) platelets.16–18 Transfusion of ABO nonidentical platelets in transfusion-dependent patients with hematologic diseases revealed earlier platelet refractoriness than in patients receiving ABO identical platelets.5 Consistently higher platelet increments were also observed in a systematic review assessing outcomes of ABO identical versus ABO nonidentical platelet transfusion; however, no consistent benefit in clinical outcomes was considered proven.1 After transfusion of ABO nonidentical platelets in bone marrow transplant patients, Heal et al10 hypothesized an association with increased mortality due to bleeding, multiorgan failure, and sepsis, seen in immune complex diseases such as lupus and rheumatoid arthritis. Multivariate analysis of 1-hour platelet recovery in pediatric patients (N = 400) showed superiority of efficacy of ABO identical versus ABO major-mismatched platelet transfusions (odds ratio, 3.97; 95% confidence interval, 1.52–10.39; P = .005).11 Using flow cytometry and fluorescence microscopy, Julmy et al11 demonstrated a rapid clearance of group A platelets from the circulation of group O or B recipients. However, major-mismatched transfusions of group A platelets that expose A2 antigen were as efficient as ABO identical transfusions (P = .90), indicating a remarkable effect of the circulating anti-A on platelet recovery. In the multicenter Platelet Dose “PLADO” study, where 1102 pediatric and adult hematology-oncology patients with hypoproliferative thrombocytopenia were randomized to receive different doses of prophylactic platelet transfusions (for platelet count of <10 × 109/L), Kaufman et al23 found no correlation between platelet source and ABO matching with overall transfusion reaction risks. However, the absence of a substantial control group of patients receiving solely ABO identical platelets makes it impossible to ascertain the role of ABO matching of platelet transfusions in these patients. In a post hoc analysis of the same study (PLADO study), including 1272 hematology-oncology patients who received 6031 prophylactic platelet transfusions, Triulzi et al12 reported that randomized dose strategy, platelet source, ABO compatibility, and duration of storage have no impact on prevention of clinical bleeding. However, authors noted that platelet increments were generally higher after transfusions of apheresis platelets, ABO identical platelets, and platelets stored 3 vs 4–5 days.12 The major limitation of these data is that few patients received solely ABO identical platelets, thus there is no substantive control group for comparison. ABO NONIDENTICAL PLASMA TRANSFUSIONS Inaba et al2 compared the rate of mortality and complications (acute respiratory distress syndrome, sepsis, renal failure, and liver failure) in a retrospective study of trauma patients (N = 284) exposed to ABO nonidentical (but compatible) plasma who were matched 1:1 with patients receiving solely ABO identical plasma. Although mortality was not different between the groups (P = .66), the overall complication rate was significantly higher for ABO nonidentical plasma recipients (P = .002), in particular for acute respiratory distress syndrome and sepsis. A dose-dependent increase in complication rates was detected as the transfused volume of ABO nonidentical plasma increased.2 In a retrospective, nationwide cohort study, Shanwell et al3 evaluated the mortality rate at 14 days posttransfusion in Swedish recipients (N = 86,082) exposed to plasma between 1990 and 2002. Higher mortality rates were found among patients receiving ABO nonidentical (but compatible) plasma. This risk was found to be dose dependent. The excess risk was most strongly associated with transfusion of 5 or more AB plasma units to group O recipients. This suggests a likely mechanism for mortality caused by circulating ABO immune complexes made up of donor soluble A and B antigens and recipient anti-A and anti-B3. In summary, there is evidence that anti-A/B, along with soluble A/B antigens present in the donor and/or recipient plasmas, may cause substantial dose-dependent harm after ABO nonidentical transfusions (Table). Unless transfusion is urgent and lifesaving, we propose that ABO identical transfusions should be the standard of care for all relevant blood components when feasible. Although adherence to a strict ABO identical transfusion policy, for platelets in particular, may lead to difficulties in maintaining adequate inventory and increase wastage, patient safety should outweigh these logistic challenges. It may be necessary to overproduce platelets from whole blood (a relatively inexpensive blood component), so that adequate inventories of ABO identical platelets are available. Of note, in a survey of North American laboratories that was performed almost a decade ago, <22% of participating laboratories (N = 3156) reported the existence of a transfusion policy of only ABO identical plasma and platelet products.15 While we have introduced an ABO identical policy successfully with minimal effects on outdating,24 this is made possible by the use of washed group O RBCs and platelets, the proximity of our regional blood center, and our large inventory/transfusion volume. It is still difficult to provide ABO identical platelets to group B and AB patients in emergency situations, particularly patients of group AB, as washed platelets are not available at a moment’s notice. The availability of platelet additive solutions that remove 95% of plasma will likely help solve this remaining issue. The availability of closed system washed platelets and RBCs that can be stored for perhaps 2–5 days would make it possible to supply minimally toxic RBCs and platelets at distances from regional blood centers and to smaller hospitals. It has been suggested that components be titered to exclude higher titers of anti-A and anti-B. Unfortunately there is no evidence base for the efficacy and safety of this practice. In any case, unless we are prepared to titer antigen in donor plasma and antibody in recipients, this approach is not feasible to mitigate the formation of ABO immune complexes based on recipient antibody and donor soluble antigen. An alternate approach might be to remove the supernatant from donor RBCs and platelets before transfusion when isoagglutinins that react with the recipient’s A/B antigens are present. None of these approaches have been rigorously evaluated as yet. There is an urgent need for randomized trials and other definitive studies to confirm or refute the hypothesis that there really are no truly safe “universal donor” transfusions. These long established practices, based on “expert opinion,” reflect historical and traditional concepts which may no longer be tenable in the modern era. ABO immune complexes and direct binding of anti-A to group A RBCs may contribute to low level hemolysis, among other biologic phenomena. Low level hemolysis may be deleterious to patients in the same manner that it is in sickle cell disease and paroxysmal nocturnal hemoglobinuria, where it contributes to inflammation, vasculopathy, thrombosis, and infection. Further research into these possibilities will help transfusion services determine if and how we move forward from the likely outdated concept of “ABO compatible” transfusions and “universal donor” blood components. A number of approaches can be taken to assess the benefits and risks of an ABO identical transfusion policy. A randomized trial comparing current practice with ABO identical only (using washed or plasma-reduced platelets and RBCs when ABO identical are not available) is 1 design that is both feasible and unlikely to dramatically affect inventory practice. Another possibility would be to compare group O universal donor RBCs with washed or plasma reduced, or low-titer group O RBCs, in elective transfusions, because these are standard practice in many centers. Finally, an implementation of a clinical trial with before and after analysis of current practice versus ABO identical only (using washed, plasma reduced, or low-titer group O platelets/RBCs when necessary) would be feasible. The terminology “ABO compatible” is potentially confusing, only relevant for RBCs, likely incorrect, and should be discarded and replaced by the more accurate terms “ABO identical” and “ABO nonidentical.” Similarly, the concept of “universal donors,” although time honored, is no longer one which we can be confident is clinically sound. The terms provide the illusion of safety to health care providers, but not the reality of safety for the recipient. DISCLOSURES Name: Majed A. Refaai, MD. Contribution: This author helped collect and analyze data, and draft and revise the manuscript. Name: Christine Cahill, MS, RN. Contribution: This author helped collect the data, and review and edit the manuscript. Name: Debra Masel, MT, (ASCP) SBB. Contribution: This author helped revise the manuscript and approve the final version. Name: Amy E. Schmidt, MD, PhD. Contribution: This author helped revise the manuscript and approve the final version. Name: Joanna M. Heal, MD. Contribution: This author helped revise the manuscript and approve the final version. Name: Scott A. Kirkley, MD. Contribution: This author helped revise the manuscript and approve the final version. Name: Neil Blumberg, MD. Contribution: This author helped revise the manuscript and approve the final version. This manuscript was handled by: Marisa B. Marques, MD.