A perspective on transfusion‐related acute lung injury two years after the Canadian Consensus Conference

Research on various aspects of transfusion-related acute lung injury (TRALI) including its incidence in various settings, epidemiologic associations, pathophysiology, and potential risk reduction strategies has been hampered by lack of agreement on a TRALI case definition. One of the goals of the April 2004 Canadian Consensus Conference on TRALI was to agree upon a TRALI case definition so that standardized research approaches could be used and results between studies could be compared.1,2 Because of the recognition that the pathophysiology of TRALI was not completely understood and was highly likely to be due to several different mechanisms, a case definition based solely on clinical and radiological parameters was proposed by the Consensus Panel. Although this definition appears straightforward, one recognized problem with the definition was the difficult clinical task of ruling out left atrial hypertension (e.g., circulatory overload, congestive heart failure) in patients with pulmonary edema. A second complicating factor in establishing a workable definition was that the characteristics of the acute lung injury (ALI) of TRALI are indistinguishable from ALI due to multiple other etiologies. Thus, if ALI developed in a patient with an underlying ALI risk factor who also received a transfusion, it would not be possible to conclude whether the ALI arose as a direct result of the transfusion or for reasons related to the patient’s underlying clinical condition. To capture such problematic cases for further scientific study and to distinguish such cases from ALI without other associated risk factors, the Consensus Conference Panel developed the term possible TRALI for these cases. Unfortunately, the Consensus Panel could not delineate definitive guidelines as to what case characteristics would favor transfusion as the causative factor versus a competing etiology. An NIH expert working group that also dealt with this issue felt that the distinction as to probable causation should be made by physicians caring for the patient but did not specify how this would be done; hence, according to their classification scheme, cases of possible TRALI would already be classified into TRALI or not TRALI (and in some cases indeterminate TRALI), and only the TRALI and indeterminate cases would be reported to the transfusion service.3 The article by Rana and coworkers4 in this issue of TRANSFUSION is the first study to apply the definitions proposed by the Consensus Conference Panel. As per the Consensus Panel recommendations, the authors distinguish TRALI cases without other ALI risk factors (they call these suspected TRALI) from cases in which transfusion was temporally related to the development of ALI in a patient with other ALI risk factors (they call these possible TRALI cases as recommended by the Consensus Panel). Although in retrospect the Consensus Panel terminology is a little confusing (i.e., all adverse transfusion reactions reported to the transfusion service are possible until they undergo further review, thus making the terminology of possible TRALI less than ideal), the concept of distinguishing possible TRALI from TRALI in a case reporting system remains a fundamentally sound idea. The authors of the current study have developed an excellent surveillance tool for TRALI case ascertainment in the intensive care unit (ICU) by making use of their ability to link several real-time databases that contain the actual timing of events such as transfusion and alteration in respiratory function. Furthermore, many of the authors of this article were ICU and pulmonary critical care specialists reflecting the absolute essential collaboration of clinicians and transfusion medicine specialists in this area of study. The study was conducted in an institution with a long-standing interest in monitoring TRALI and other adverse transfusion events and was able to build upon expertise acquired from previous studies in the same institution that evaluated ALI and transfusion in ICU patients.5 Several interesting and important findings clearly emerge from the study. The incidence of TRALI in the ICU setting in this study (1 per 1271 transfused units or 1 per 193 recipients) is higher than incidence rates reported in the literature in other settings.1 The incidence of possible TRALI was even higher (1 per 534 transfused units or 1 per 79 recipients), which is not surprising given the frequency of other ALI risk factors in this severely ill population. Likewise, circulatory overload attributable to transfusion (TACO) also occurred with very high frequency (1 per 356 transfused units or 1 per 54 recipients), although it is not clear whether all of these cases were caused by excessive transfusion versus worsening of cardiac function or coincident disruption of underlying fluid balance for other reasons. Of great significance to transfusion medicine specialists is that not a single case of TRALI or possible TRALI detected during this retrospective review had been reported from any of the four participating ICUs to the transfusion service—and this occurred in an institution that has performed some of the most intensive TRALI research studies over the past two decades. This lack of reporting supports previous findings in the literature as well as a number of statements made by participants at the Consensus Conference.6 It is not known, however, whether this degree of nonreporting also occurs in clinical services where the patients do not have as many ALI risk factors or whether case reporting has increased in the past few years since the publication of two review articles and an AABB association bulletin designed to increase awareness of TRALI.1,3,7 Despite the careful case ascertainment strategy, there are many limitations of this study, as acknowledged by the authors. The study was retrospective and therefore subject to confounding. Importantly, TRALI incidence data in the ICU setting should not be extrapolated to other clinical settings, because it is quite plausible that recipient factors influence the development of TRALI, as discussed in detail in the two-hit model of TRALI.8 The algorithm for distinguishing pulmonary edema due to ALI from that due to other causes is not given in detail. It appears that expert judgment was used to evaluate the many pieces of clinical data that contributed to this assessment. We need to assume that this judgment was accurate given the participation of multiple skilled clinicians; however, it is interesting to note that Rana and coworkers’ Table 14 shows few differences between the characteristics of patients with TRALI or possible TRALI and those with TACO. Furthermore, for the cases of possible TRALI, we do not get an inkling of how often the clinical judgment of the panel favors transfusion versus other ALI risk factors as the most likely etiology; it would have been helpful if illustrative details had been given for some of the cases so that the reader could determine whether such inferences are even possible. Due to small numbers of cases, the authors analyze TRALI cases together with possible TRALI cases in their logistic regression models; however, because these cases have different degrees of certainty with regard to diagnosis, this may not be methodologically appropriate. An additional conclusion of the study is that when TRALI and possible TRALI cases are grouped together, these patients had received a greater total volume of plasma (expressed in milliliters rather than number of components transfused) as well as a higher volume of plasma from both female and male donors. Furthermore, based on logistical regression modeling, the authors report that the transfusion of higher volumes of total plasma or plasma from women was a risk factor for the development of TRALI or possible TRALI. This association, however, is not readily apparent when examining the raw data (Rana and coworkers’ Tables 1 and 24) on either the total number of fresh-frozen plasma (FFP) and single-donor plateletpheresis units or the number of such units from female donors transfused to TRALI and/or possible TRALI cases versus matched controls. Even if the association is accepted as valid, it is clear that causality has not been established as this study did not evaluate donors or recipients for white blood cell (WBC) antibodies and cognate antigens. Based on these limitations, the authors are cautious in projecting how their data should be used to influence policies that are designed to reduce the risk of TRALI. In 2004, the Canadian Consensus Conference panel indicated that blood collection agencies should evaluate whether interventions to reduce TRALI risk would have a projected benefit that would be in excess of any potential adverse impact, such as a decrease in the availability of needed blood components.1 This recommendation was made because TRALI was one of the leading causes of transfusion mortality; subsequently, it has become the leading cause as reflected by transfusion-related fatality data reported to the FDA.2,9,10 A mean of 16 such cases were reported annually from 2001 to 2003; assuming that there are 3.2 million transfusion recipients annually and that there is not underreporting of TRALI fatalities to FDA, this leads to an estimate that TRALI contributes to mortality in 1 in 200,000 transfusion recipients.10 The Consensus Panel did not recommend a particular risk reduction and/or prevention strategy but enumerated several options including the use of male plasma rather than female plasma for transfusion, excluding plasma from multiparous females from transfusion, or excluding plasma from donors with HLA antibodies. At the time of the Consensus Conference, the UK had performed a detailed programmatic analysis that resulted in their adoption of a policy to minimize the transfusion of FFP from female donors; no data on the efficacy of that policy were yet available.2 Recently, data on the UK experience was published as part of the 2004 UK Serious Hazards of Transfusion (SHOT) report. Comparison of 2004 TRALI incidence data (excluding cases from January 2004 when FFP from female donors was still in hospital inventory) with historical data before the implementation of the restricted female plasma strategy showed that the number of TRALI cases markedly decreased.11 Despite what appears to be a clear change in TRALI incidence, the authors of the SHOT report urged caution in drawing any definitive conclusions from their first-year experience with female-restricted plasma. Are there additional data that could help inform our decision making regarding TRALI risk reduction policies? We should get some additional data in the very near term. The 2005 SHOT data should be available by year’s end. At the time of this writing, major US blood suppliers are reviewing their TRALI case data to ascertain how frequently implicated donors can be identified (based on a donor antibody matching a cognate antigen in the recipient) and analyzing whether this concurrence occurs more frequently after transfusion of FFP from women rather men. Based on the passive reporting system of TRALI in the United States and the incomplete clinical information that is often available to blood centers when cases are reported, it is almost certain that these US data will be subject to criticism and will not provide definitive evidence for the projected efficacy of any particular risk reduction strategy. Even without seeing these data, however, the well-documented higher frequency of HLA antibodies in women unequivocally predicts that, for the antibody-mediated mechanism of TRALI, FFP units from female donors will be associated with TRALI more often than will similar units from male donors.12,13 Although it is not known what percentage of TRALI cases are due to the WBC antibody mechanism, several recent publications have emphasized that antibody-mediated TRALI is the predominant mechanism in patients with severe cases of TRALI requiring ventilator support or resulting in death.4,14,15 Although it may not be possible to quantitate the effect of a risk reduction strategy on overall TRALI incidence, it seems reasonable to conclude that the number of severe cases will decrease. Therefore, in my mind, the question is not whether some risk reduction will be achieved by modifying our source of FFP units (it will), but how much will be achieved and at what consequence (if any) to the ready availability of transfusable blood components. Component availability will be influenced by how a risk reduction strategy is implemented. An important policy question is whether a risk reduction policy can be incremental (e.g., supplying a greater percentage of FFP units from male donors if that is the strategy adopted) and partial (e.g., applied only to FFP [and then maybe not to group AB FFP] and not to plateletpheresis due to a greater impact on product availability) without either the regulatory or the medicolegal climate requiring a 100 percent solution. The precedent for implementing incremental TRALI risk reduction strategies exists in the UK and had previously been adopted by a regional blood center in Spain.2,16 These issues are complex, and many nuances need to be considered before effective or partially effective TRALI risk reduction programs can be implemented in the United States. To this end, AABB has convened an expert working group to explore the effects on blood safety and availability of potential intervention strategies to prevent or reduce the incidence of TRALI. In addition, NHLBI has funded two major studies addressing several aspects of TRALI that may be pertinent to risk reduction strategies. The largest study is a 5-year Specialized Centers of Clinically Oriented Research (SCCOR) award to the University of California at San Francisco (P.I. Pearl Toy) with major collaborations with the Mayo Clinic. The SCCOR includes a prospective protocol using a computerized surveillance system to study the incidence of TRALI and to determine risk factors either in the recipient or in the donor unit.17 It is likely that definitive data from this study are still several years away. In addition, the new Retrovirus Epidemiology Donor Study-II (REDS-II) is implementing a major protocol entitled the Leukocyte Antibody Prevalence Study (LAPS), which will over the course of the next year determine the prevalence of anti-HLA Class I and II antibodies in contemporary female donors stratified by pregnancy history and in transfused and untransfused male donors. The LAPS data should be helpful if strategies to screen donors for HLA antibodies are being considered as a risk reduction method. So, have we yet reached the juncture where the transfusion medicine community should take action? As the concept of TRALI risk reduction was introduced more than a decade ago, why is it now being seriously reconsidered?18 There appear to be at least two reasons. First, TRALI now clearly stands as the statistically highest severe transfusion risk. Second, over the past 7 years, the transfusion medicine community has shown the ability and willingness to implement logistically complicated and very expensive interventions (minipool NAT for HIV and HCV, automated culturing of plateletpheresis) to further reduce risks that were as low or lower than the current TRALI risk. Although implementing a TRALI risk reduction strategy involving altering the release of blood components into inventory based on either donor sex, donor demographics (i.e., past parity history), or laboratory findings (presence of HLA antibodies) will pose operational challenges, such a program does not seem to be inherently more complicated or expensive than these other successfully implemented interventions. In my opinion, we currently have or shortly will have enough understanding of the issues to decide whether the benefits of implementing a TRALI risk reduction strategy outweigh the downsides of taking such action. Active debate over the next months will likely determine whether such new risk reduction strategies will be implemented.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.