Transfusion‐related acute lung injury: three decades of progress but miles to go before we sleep

In the first decade, approximately 1985 to 1995, we learned that the classic presentation of TRALI is life-threatening and indistinguishable from adult respiratory distress syndrome (ARDS) and that because of severe, diffuse permeability pulmonary edema, hypoxemia, and hypotension, these patients require immediate mechanical ventilation as well as blood pressure support. Unlike those with ARDS, fully 80% recover within 96 hours. Any type of plasma-containing product, irrespective of volume, can cause TRALI. While there is consensus that the vast majority of classic TRALI occurs within 1 to 2 hours and the remainder within 6 hours, some investigators believe that there are cases that present up to 24 hours from initiation of transfusion.3 Although it is the severe cases which require resources and clinical skill to manage, the most frequent presentation may be relatively benign, limited to mild hypoxemia in the presence of white blood cell antibodies.4 After the Popovsky and Moore publication in 1985, a plethora of articles from around the globe, representing hundreds of cases, appeared over the next 10 years. Investigators confirmed the Mayo findings of a prominent role for passively acquired antibody-laden plasma. In the majority of cases, antibody corresponded to recipient HLA or HNA epitopes.5 Donors implicated in the HLA-positive cases were typically multiparous females. The reason that this complication of hemotherapy moved to center stage in the transfusion medicine world was its significant morbidity and mortality. The fatality rate of 5% to 20% exceeds that of any transfusion reaction.6 In the setting of the intensive care unit, the death rate may be as high as 47%.7 Once the Food and Drug Administration (FDA) initiated the collection of data for TRALI, it was clear that this syndrome is the leading cause of transfusion-associated death. In 2012 TRALI accounted for 43% of transfusion-related deaths in the United States, and in the preceding decade, it had accounted for up to 65%.8 The European experience is similar, where TRALI is the first or second most frequently reported transfusion-associated fatality. The significant resources expended to deal with its morbidity increase its importance, particularly in an era of health care cost constraint. Because of the prevention strategies now in place, the answer is much different today than it was in the 1980s. At the Mayo Clinic, where bedside transfusions are under the control of highly experienced transfusion medicine nurses, an incidence of 1 in 5000 plasma-containing transfusions was established as the early benchmark.1 Subsequent hospital-based, albeit retrospective, studies reported similar frequencies although the published rates of TRALI from European hemovigilance systems were lower by orders of magnitude. There were many reasons for the disparity, including the lack of common diagnostic criteria, differences in type of blood product involved reported, and passive versus active collection of data. The publication of case definitions from the 2004 Canadian consensus conference and National Heart, Lung, and Blood Institute Working Group were milestone events,9, 10 as they provided a common framework to diagnose the syndrome. The case definitions were based on clinical and radiologic parameters. Yet, some countries use an alternative approach in which imputability is scored.11 Complicating matters further is how some transfusion services and donor centers distinguish and report TRALI, possible TRALI, and delayed TRALI (presentation beyond 24 hr).12 These differences result in different thresholds for which a definitive diagnosis is made. In summary, we have a Tower of Babel in establishing a common assessment of the frequency. Before the implementation of risk reduction measures, evidence existed that TRALI is more frequent than the 1 in 5000 Mayo benchmark. In a prospective study in patients undergoing cardiopulmonary bypass, TRALI developed in 2.4% within 30 hours postsurgery.13 In the medical intensive care unit, 8% of transfused patients followed prospectively developed acute lung injury.14 In an active surveillance study of two tertiary care centers, the TRALI incidence in 2006, just before implementation of plasma risk reduction measures, was 2.57 per 10,000 units transfused (1:4000 units), dramatically higher than what was reported from passive hemovigilance systems,15 but comparable to the earliest observations. Because the diagnosis of TRALI involves consideration of other confounders, most frequently hydrostatic pulmonary edema (transfusion-associated circulatory overload [TACO]), many cases of TRALI are most certainly overlooked or misdiagnosed. Supporting this view is the widely known case study of Kopko and coworkers6 in which of 36 evaluable patients transfused from an index donor, only seven of 15 TRALI reactions were reported to the transfusion service, and of these, only two were reported to the regional blood center that collected the implicated plasma products. As TRALI and TACO can occur concurrently and because these pulmonary complications are frequently confused, the belief that many cases are undetected now receives additional credence Both retrospective and prospective studies have identified common patient risk factors for TRALI. Most striking is mechanical ventilation, in which up to 33% of transfused patients develop acute lung injury.7 A working assumption is that positive pressure ventilation causes neutrophil priming, which leads to pulmonary microvascular endothelial injury. This theory may explain why so many critical care patients develop TRALI.16 Prospectively collected data indicate that massive transfusion, shock, positive fluid balance, chronic alcohol abuse, severe liver disease, and elevated pretransfusion levels of interleukin-8 are risk factors for the syndrome. Additionally, sepsis, hematologic malignancy, and emergency cardiac surgery have been implicated in retrospective studies.13, 16 In cardiac surgery extracorporeal circuits may serve to activate neutrophils. On the other hand, the cases at high risk for TRALI are also among the sickest patients who are commonly transfused, making proof of causality difficult. Transfusion risk factors have been found to play a role as well. These include plasma, whole blood, volume of HLA Class II antibody, and volume of anti-human neutrophil antigen.15 The antibody predictors apply to all blood components, but it appears that larger plasma volume confers higher risk. From the study of Toy and colleagues,16 no risk or little risk was associated with older red blood cell (RBC) units, noncognate or weak cognate Class II antibody, or Class I antibody. These data must be interpreted with caution as the number of studies that have investigated these issues are few and the level of evidence mixed. Caution should be applied to the relationship of RBC storage age, platelet (PLT) age, and the presence of bioactive lipids to TRALI, inasmuch as these factors have either been implicated or dismissed. Recent studies shed light on two demographic groups—the elderly and children—neither previously thought to have a prominent place in TRALI. A multivariate analysis of inpatient Medicare beneficiaries during 2007 through 2011 found that there was increased TRALI risk with plasma or PLT transfusions, in females and the young elderly (65-79 years). The inpatient mortality of 21% is at the upper end of what is reported in the literature.17 The authors hypothesize that the decreased TRALI rate among the older elderly (>79 years) may reflect the age-related decline in innate immunity and, with it, impairment of neutrophil function. TRALI cases in children were thought to be rare. However, Canadian investigators reviewing cases over an 11-year period found that children were at neither greater nor lesser risk for TRALI.18 Clinical presentation and morbidity were the same as for adults. Cases clustered around teenagers or those less than 1 year of age. Since 2004 a new TRALI era—the use of low-risk plasma—was entered. For most blood collectors, this approach has dictated the use of male plasma for transfusion but some have utilized HLA antibody screening of female donors. The results of these efforts are impressive. In the United Kingdom, the risk of death from TRALI nearly disappeared while the incidence decreased by almost two-thirds.19, 20 Data from the American Red Cross hemovigilance program showed that the risk of TRALI from the general inventory of distributed plasma decreased significantly from 18.6 cases per million units in 2006 to 4.2 cases per million units from 2008 to 2011 (using predominantly male plasma).21 In a multihospital investigation, once male-predominant plasma was introduced, the incidence of TRALI decreased from 0.0084% to zero over a 16-month period.22 But the most notable of these studies was a prospective, active surveillance model at two US medical centers. These investigations showed a decrease in TRALI incidence from 2.57 per 10,000 (1:4000) in 2006, before implementation of TRALI risk reduction measures versus 0.81 per 10,000 (1:12,000) in 2009 after implementation.15 Most importantly, TRALI deaths attributable to plasma decreased substantially over the past decade. In the United Kingdom there were seven known deaths in 2003 recorded by the Serious Hazards of Transfusion Program and in 2012 there were none.23 FDA data are equally gratifying. There was a reduction by 63% of reported deaths after the 2007 introduction of low-risk plasma. While these data are encouraging, it is noteworthy that since 2008, the number of deaths (16-18 per year) and proportion of transfusion-related fatalities due to TRALI has leveled off at 37%, illustrating that there remains considerable room for improvement.8 What do these data tell us? The removal of HLA, HNA, and other antibody-laden plasma from the transfusion supply chain has had a hugely beneficial effect on transfusion safety, but it has not eliminated the risk. The current status is best illustrated by American Red Cross hemovigilance data, which indicate that despite a risk reduction plasma strategy there is a 14-fold higher risk for TRALI in recipients of AB plasma than for recipients of A, B, and O plasma.21 Additionally, RBCs now account for the largest number of TRALI cases in the United States, despite the fact that RBCs are considered a low-risk product. Since the 1990s the understanding of the pathogenesis of TRALI has been built on two pillars: antibody-mediated and two-hit models. In both these models there is a final common pathway that entails pulmonary endothelial cell damage and capillary leakage.24 In antibody-mediated TRALI, antibodies to HLA Class I or Class II antigens and HNA bind either neutrophils or monocytes, which leads to neutrophil activation and subsequent pulmonary edema. In the two-hit model, a patient is predisposed to TRALI by underlying factors (e.g., cardiac surgery, hematologic malignancy), which causes polymorphonuclear cell attraction, rolling, and adherence to pulmonary endothelium.25 Transfusion of a blood component containing antibodies, cytokines, or bioactive lipids causes the release of cytokines and neutrophil activation. In the past decade elegant models have been developed, which support both models. Looney and coworkers26 showed that TRALI is highly dependent on neutrophils. In an in vivo mouse model, depletion of neutrophils protected the animals from lung injury after MHC I monoclonal antibody challenge. In a recently developed murine antibody–dependent TRALI model, lung damage was shown to require the binding of anti-MHC antibodies to their cognate antigen, stimulation of chemokines from monocytes, and a subsequent monocyte Fc-dependent injury.27 Bux and Sachs28 took the two-step approach further by developing a “threshold model.” These investigators proposed that a certain threshold must be overcome to induce a TRALI reaction. Whether TRALI occurs and whether or not it is severe depends on patient predisposition and the strength of the transfusion-related mediators. In this issue of TRANSFUSION two provocative articles challenge our current thinking on TRALI. Middelburg and van der Bom29 argue for a new type of threshold model and a sufficient cause model. They critique the current threshold model as “lumping” patient and transfusion risks together and, in so doing, limit the ability to tease apart the influences of individual risk factors. They believe that TRALI is a result of multiple causal factors that together reach a pathogenic threshold for TRALI. To counteract the perceived weaknesses of the threshold model of Bux and Sachs, they propose a sufficient cause model. It is an abstract construct but it provides for representation of the interaction and effect modification of individual risk factors. It is not clear how investigators will utilize this model, but time will tell. The article by Warkentin and colleagues30 is a bold reappraisal of the current clinically based TRALI definitions. Using heparin-induced thrombocytopenia (HIT) as the model, they propose a new classification scheme that hinges on the presence or absence of leukoreactive alloantibodies, in other words, immune versus nonimmune. The authors would designate “TRALI-I” in patients in which alloantibodies are identified; “TRALI-II” would be for antibody-negative cases. In this schema, TRALI-1a might be for HNA alloantibody positivity and TRALI-1b and TRALI-1c could implicate HLA Class I and II antibodies, respectively. They posit that this approach would reframe the thinking of what is today a clinical diagnosis and redirect the focus on developing better tests and perhaps new streams of research. This objective is worthy, given the fact there is no way to distinguish which antibodies are pathologic and there is no pathognomonic way today to make the diagnosis. One can envision how such a schema would potentially resolve those cases currently identified as possible TRALI. There is considerable merit to this proposal but one can foresee some obstacles. Will clinicians wait for test results before initiating treatment? Fulminant TRALI requires rapid intervention. What about antibody-negative cases? Will these be considered something other than TRALI? Would this approach supplant the current clinical definitions? The importance of both articles in this issue is that despite the effectiveness of risk reduction measures that have already been implemented, TRALI remains an exciting and important area of research. It is conceivable that these new approaches will lead to new ways of looking at what remains as a significant clinical problem, which could lead to creative and effective diagnostic and therapeutic solutions. Assuming that the new baseline of 1 in 12,000 transfusions approximates the postplasma mitigation incidence rate and that a considerable number of cases of TRALI are not identified for reasons previously discussed, TRALI is far from eliminated. When coupled with the high mortality rate reported in elderly patients, TRALI remains an important clinical problem. The most cost-effective strategy to reduce its incidence builds on the patient blood management momentum created worldwide over the past several years. The four “rights” of right product, right dose, right time, and right patient certainly apply in the TRALI world. Better decision making for the transfusion of all types of blood components will reduce the incidence of TRALI. The other possibilities for further risk reduction must each be weighed for their cost-benefit. These include broad screening for HNA antibodies (in addition to HLA), PLT additive solutions, and plasma-reduced PLT transfusions, for whole blood and apheresis. The use of whole blood–derived PLT pools is viewed as low risk for TRALI.20 Filters have been developed that absorb antibodies and lipids, reducing TRALI-associated antibodies and neutrophil-priming activity, mitigating TRALI in an animal model. The experimental filter removes 96% of IgG as well as HLA and HNA antibodies.31 Whether the costs of such an advance would permit widespread implementation remains to be seen. Finally, the recent approval of pathogen reduction technology of PLTs and plasma in the United States offers the potential of a new and exciting approach in the prevention of TRALI. Clearly more data and experience will be required to draw firm conclusions about its effectiveness. The work over the past 30 years has resulted in dramatic improvements in transfusion safety overall, both for infectious and for immunologic causes of transfusion sequelae. There is little doubt that much progress has been made, but there is both room for improvement as well as optimism about the tools that the transfusionist can apply to the patient who will benefit from them. In summary, we still have miles to go before we sleep. Mark A. Popovsky, MD e-mail: [email protected] Vice President and Chief Medical Officer Haemonetics Corporation Braintree, MA The author has disclosed no conflicts of interest.