Two recent commentaries,1 1 in the current issue of this journal,2 have addressed pressing issues concerning a commonly used therapy, platelet transfusion. Both are timely and succinct, pointing out that current practices are making life difficult for transfusion services, clinicians, and almost certainly harming patients. One commentary documents that supply of platelets increasingly has become a problem, and the other that the supply problem is exacerbated by liberal and nonevidence-based platelet transfusions. This editorial will focus on what is going well (the good), what is not going well due to misunderstandings about the efficacy and safety of platelet transfusions (the bad), and current practices that are harmful to patients (the ugly). We will focus on issues in the United States, but much of what we discuss will be relevant in other nations.
THE GOOD
The very best about platelet transfusion is altruistic platelet donors. In exchange for movie tickets or a sandwich, they give up several hours of their day (if apheresis donors) and subject themselves to 2 large bore venipunctures. This is distinctly more challenging and uncomfortable than the altruistic donors who donate single units of whole blood, which can be fractionated into red cells, plasma, cryoprecipitate, and platelets. The donor population is a tiny fraction of those eligible and probably cannot be dramatically increased. Thus, methods to improve transfusion utilization, where excessive/harmful, and improve clinical efficacy will be the focus of this editorial.
The only convincing evidence for platelet transfusion’s benefits is that routine prophylactic transfusion when a patient’s platelet count is <10,000/µL reduces both minor and moderate bleeding during the period of thrombocytopenia.3 Severe bleeding, which is fortunately not common, is also likely mitigated. Medium and long-term effects on survival have not been studied thoroughly. There are no studies comparing clinical judgment assisted by platelet count, to counts alone, in assessing whether transfusions will improve outcomes. Since the best predictor of bleeding is the history and physical examination, not the platelet count, there is certainly room for improvement in our paradigms. The current approach to prophylactic transfusion has been primarily studied in patients with hematologic malignancies during initial induction therapy or stem cell transplant. Whether minimizing transfusions or incorporating the presence of personal bleeding history and physical examination findings to modify transfuse decisions could reduce transfusion needs and thus increase supplies has not been studied. There is some evidence that increasing numbers of platelet transfusions is associated with decreased long-term survival.4
There is a movement toward reducing the dose of platelets per transfusion which might provide equivalent clinical efficacy and reduced toxicity in many situations. Indeed, the standard platelet dose is substantially lower in some countries than in the United States. Reducing the routine dose would likely also increase the supply of platelets by increasing the number of available doses.
THE BAD
In the United States, for reasons that are not clear, apheresis platelets are the product of choice. This has led to the almost complete abandonment of whole blood-derived platelets, which are largely indistinguishable in their clinical efficacy and safety from the single-donor apheresis platelet. We agree with Riley et al1 that a return to use of whole blood platelets would substantially improve the supply of platelets, and be a more appropriate use of donor resources than a total reliance on apheresis. Apheresis is more expensive, time-consuming, and entails increased risks for the donor, with little or no advantage for the transfusion recipient.
Burns et al2 also make the case that our clinical practice patterns for platelet transfusions are overly aggressive, potentially harmful, and not evidence based. They note that there are now 3 randomized trials demonstrating that aggressive or liberal platelet transfusion is associated with increases in bleeding and mortality, much to the surprise of many in the field.5–7 More about the possible mechanisms underlying these findings have been discussed later in this editorial.
We should not have been surprised, since all allogeneic transfusions, including red cells and platelets, have been associated with increases in inflammation, thrombosis, multiorgan failure, and mortality in both observational studies and randomized trials.8 For plasma, there are only observational studies. An example of platelets is a study by Cashen et al,9 demonstrating that platelet transfusions are associated with increases in bleeding, thrombosis, and mortality in pediatric patients receiving extracorporeal support. As hemostatic and antithrombotic therapies are usually associated, respectively, with increased thrombosis or bleeding, but not both, this seems mystifying at first glance. However, platelet transfusion as currently practiced interferes with hemostasis and causes inflammation, which may explain these seemingly incompatible associations.10 This is partially explained by the fact that stored platelets are highly activated and do not resemble normal healthy platelets under healthy physiologic conditions. Dysregulated hemostasis and endothelial function predispose to bleeding, while inflammation predisposes to both bleeding and thrombosis.
For many years, platelets were thought of as purely hemostatic in function. However, approximately 20 to 25 years ago it became clear platelets were involved in host defense and contain large amounts of immunomodulatory mediators such as sCD40L, VEGF, IL-6, etc. Platelet transfusions have been shown to be proinflammatory and prothrombotic, both in vitro and in vivo in patients.10 Recent randomized trials, including the PlaNet2 trial in premature neonates, have demonstrated that aggressive platelet transfusions are associated with increased bleeding and mortality.6,7 One consideration, in light of mounting evidence of increased morbidity and mortality, would be to reduce prophylactic platelet transfusions for patients with no physical evidence of impaired hemostasis, and platelet counts >10,000/µL, contrary to long-standing practice. There is little or no evidence of efficacy and accumulating evidence of harm.
A recent study provides evidence that prophylactic transfusion of platelets to patients undergoing central venous line placement might reduce bleeding, but only in patients with counts <30,000/µL. Unusually, the extent of bleeding was greater in transfused patients, and the benefit was restricted to patients undergoing subclavian line placements.11 This study is limited by not accounting for recent previous platelet transfusions in the nontransfused cohort (which can impair hemostasis, if not ABO identical) nor the ABO matching of the index transfusion (see the next section).
THE UGLY
Platelet refractoriness is when it is not possible to raise the platelet count with transfusion, usually due to alloimmunization to HLA class I antigens from prior transfusions and pregnancy. It is a common problem in repetitively transfused patients, typically those with hematologic malignancies. There is an abiding dogma, unsupported by credible data, that the ABO blood group does not matter for platelet transfusions. This dogma ignores substantial scientific and clinical evidence, as 2 randomized trials and 1 large observational study demonstrate increased platelet transfusion refractoriness after ABO nonidentical platelet transfusions.12–14 The randomized trials are small, heterogeneous, and antedate universal leukoreduction. One study included diverse patients with hematologic disorders rather than, for example, a single type of leukemia. However, small studies have the power to demonstrate striking differences between recipients of ABO identical and ABO nonidentical transfusions because the effect is very large. Studies that include typical patients are more generalizable than studies focused on 1 particular disease.
When ABO identical and leukoreduction are combined, the refractoriness rate is close to 0,12 compared with 10% to 20% in the literature, and perhaps 50% before implementation of leukoreduction and use of ABO identical blood transfusions. The mechanism of refractoriness is the infusion of both ABO nonidentical antigen and antibody through repetitive transfusions that do not respect the patient’s ABO blood group. This leads to increases in human leukocyte antigen antibody alloimmunization, very large immune complexes of ABO antigen and antibody, and inability to raise the platelet count >5000/µL, much less 10,000/µL.15 Laboratory modeling demonstrates that ABO antibody and immune complexes impair platelet16 and endothelial function,17 increase inflammation and causes dysregulation of thrombin generation.16 The effects are cumulative and last days to weeks in our experience.13,15 Refractoriness is associated with early mortality.18
Additional observational and epidemiologic studies provide evidence that ABO nonidentical transfusions are also associated with increased bleeding, sepsis, lung injury, and mortality in a variety of clinical settings.19–21 Most concerning perhaps is the evidence that the favored type of ABO major incompatible platelet transfusions is associated with increased bleeding and mortality in patients on antiplatelet drugs who have intracranial hemorrhage.22,23 The number needed to harm is approximately 5.22
In randomized trials of platelet transfusion prophylaxis, the rate of bleeding was 70% in a study where no consideration was given to ABO identical transfusions,24 versus 2% to 3% when only ABO identical transfusions were transfused,25 a 20- to 35-fold difference in bleeding. This difference is likely attributable, at least in part, to transfusing without regard to ABO identicality versus making ABO identicality a priority. There are significant logistic challenges to providing platelets free of ABO-incompatible antigen and antibody, most notably the short 5 to 7 storage period. However, these challenges can be overcome by altering priorities and adding practices such as plasma reduction/washing, which have been successful in randomized and cohort studies.25 At least 3 large medical centers have been using washed platelets for decades to almost completely mitigate febrile and allergic reactions to platelet transfusions (personal communication). We have treated hundreds of patients diagnosed with hematologic malignancies with prophylactic-washed platelet transfusions and with a significant bleeding rate <5%.
Two observational studies demonstrating that ABO does not matter for platelet transfusion are methodologically flawed.26,27 The ABO identical group in each study had an unknown number of patients who received ABO nonidentical platelets. The number of ABO nonidentical platelets and recipients in the ABO identical group was not reported. Patients were classified solely by their first platelet transfusion, a scientifically invalid method of classification when subsequent transfusions were not identical.
Finally, avoiding infusion of ABO-incompatible antigen and antibody reduces the utilization of platelets in a randomized trial in hematology patients by approximately 50%.13 Thus, transfusing only ABO identical platelets will not only avert serious harm to patients but also reduce the need for platelet transfusions in the setting of hematologic disease, effectively increasing the supply of platelets for transfusion to all patients.
THE WAY FORWARD
Blood centers that are not making at least some whole blood-derived platelets out of their collections should reconsider this logistically and clinically disadvantageous practice, which wastes the platelets that could be manufactured from millions of collected units of whole blood. Pathogen-reduced platelets that are treated to reduce the risk of bacterial contamination and posttransfusion sepsis have become the product of choice in many hospitals. In the United States, in contrast to Europe, only apheresis platelets, not whole blood platelets that are pathogen reduced, are licensed for transfusion. The FDA should allow licensure of pathogen-reduced platelets derived from whole blood in the United States, based on the European data.
There is a need for studies evaluating whether restrictive platelet transfusions are as effective and safe as the nonevidence based liberal guidelines now in use. In particular, assessment of bleeding risk using history and physical examination, rather than strictly platelet counts, is an approach that has promise and a strong clinical rationale that can be tested.
Anesthesiologists and other practitioners need to insist their transfusion services make greater efforts to avoid infusing ABO nonidentical antibody and cellular/soluble antigen. This can be achieved by using some combination of blood component stewardship through patient blood management, ABO identical blood components, and removing ABO nonidentical supernatant through washing or other means. All physicians should consider adopting more conservative platelet transfusion practices (prophylaxis for counts of <10 to 30,000/µL, depending on the clinical findings), rather than higher platelet counts as at present. There is very little evidence that higher thresholds provide clinical benefit and there is accumulating evidence of harm, up to and including mortality.
Another potential approach to reducing platelet transfusion toxicity is to reduce the volume transfused to children from the typical 15 to 20 mL/kg to 3 to 5 mL/kg. The latter is the dose that has successfully reduced bleeding in adult randomized trials of prophylactic platelet transfusions. Extraordinarily high infused volumes very likely increase toxicity with no clinical benefit, and are not evidence based.
Transfusion services should strive to avoid providing ABO nonidentical platelets unless ABO nonidentical cellular/soluble antigen and antibody have been substantially reduced. Clinicians at the bedside should consider newer principles of patient blood management that allow even autologous stem cell transplants to be performed with no platelet transfusions.28 E
REFERENCES
Leave a Reply
You must be logged in to post a comment.