Transfusion-related acute lung injury: a literature review
Professor P. C. A. Kam
Transfusion-related acute lung injury (TRALI) is a serious and potentially fatal complication of transfusion of blood and blood components. TRALI is under-diagnosed and under-reported because of a lack of awareness. A number of models have been proposed to explain the pathogenesis of TRALI: an antibody mediated model; a two-event biologically active mediator model; and a combined model. TRALI can occur with any type of blood product and can occur with as little as one unit. Its presentation is similar to other forms of acute lung injury and management is predominantly supportive. The main strategy in combating TRALI is prevention both through manipulation of the donor pool and through clinical strategies directed at reducing transfusion of blood products including, but not limited to, evidence-based lower transfusion thresholds. This article presents a review of TRALI and addresses the definition, pathology, pathogenesis, clinical manifestations, treatment and prevention of the syndrome.
Blood supplies have become safer with regard to the transmission of infective agents and the attention of transfusion medicine is now focused on understanding and eliminating other serious complications associated with blood component transfusion. Transfusion-related acute lung injury (TRALI) is a serious and potentially fatal complication of transfusion of blood and blood components. TRALI was not recognised as a clinical entity until the 1980s. The United States Food and Drug Administration (FDA) currently estimates it to be the leading cause of transfusion-related mortality . However, TRALI is under-diagnosed and under-reported because of a lack of awareness [2, 3].
The aim of this article is to review the definition, pathology, pathogenesis, clinical manifestations and the treatment and prevention of TRALI. The available articles retrieved from Pub-Med, Medline and EMBASE consisted of case reports, consensus statements from conferences, animal research, retrospective case series, observational studies from haemovigilance networks and one randomised controlled trial of multiparous donors. Despite the fact that it is estimated to be the leading cause of transfusion-associated death [1, 4], little research towards optimising management strategies has been undertaken because it is a relatively rare condition.
Respiratory complications following blood transfusion that resemble the syndrome of TRALI have been reported since the 1950s . The syndrome has previously been called pulmonary hypersensitivity reaction, allergic pulmonary oedema, non-cardiogenic pulmonary oedema, and pulmonary leucoagglutinin reaction. The term ‘transfusion-related acute lung injury’ was proposed by Popovsky to refer to pulmonary oedema complicating blood transfusion. The lack of a consensus definition of TRALI contributed to its under-recognition as a cause of acute lung injury and has inhibited research into TRALI [2, 3]. A definition emerged from the TRALI consensus conference in 2004 and from the US National Heart, Lung and Blood Institute [6,7]. The definition was based on the widely used definition of acute lung injury proposed by the American-European Consensus Committee (AECC) . Acute lung injury (ALI) is defined by the AECC as a syndrome based on clinical and radiological findings rather than a pathological or a pathophysiological diagnosis. The AECC definition of ALI is characterised by: an acute onset; hypoxaemia (Pao2/Fio2 = 300 mmHg (40 kPa) at sea level (adjusted downward depending on altitude) regardless of level of positive end-expiratory pressure or Spo2 < 90% on room air); bilateral pulmonary infiltrates on the frontal chest radiograph; and no clinical evidence of left atrial hypertension or circulatory overload, or the pulmonary artery occlusion pressure ≤ 18 mmHg . ALI is a clinical syndrome which may be caused by several direct or indirect injuries to the lung [9–12]. The potential causes of ALI include aspiration, pneumonia, toxic inhalation, lung contusion, near drowning, severe sepsis, shock states, multiple trauma, burn injury, massive transfusion, acute pancreatitis, cardiopulmonary bypass and drug overdose [9–13]. Risk factors associated with the development of ALI include increased age, ethanol or tobacco abuse, severe illness and the presence of the variant surfactant B gene [11–14].
The TRALI consensus conference and the US National Heart, Lung and Blood Institute defined TRALI as a form of acute lung injury meeting the criteria proposed by the AECC that is temporally and mechanistically related to transfusion of blood or blood components [6, 7, 15]. For a diagnosis of TRALI to be made there must be no pre-existing ALI before transfusion, the onset of signs and symptoms must occur during or within 6 h of transfusion and there must be no temporal relationship to an alternative risk factor for ALI. If both transfusion and another cause for ALI are temporally related, the consensus conference recommended that the term ‘possible TRALI’ be used in these cases [6, 7, 15]. ‘Possible TRALI’ is defined when the following criteria exist: evidence of ALI; absence of a pre-existing ALI before transfusion; onset of symptoms or signs of ALI within 6 h after the transfusion; and the presence of an alternative risk factor for ALI.
Multiple blood transfusion is a risk factor for the development of ALI [11–13,16]. However, massive transfusion (defined as the replacement of one or more blood volumes within a 24-h period with either packed red cells or whole blood) is not a requirement of the committee's definition of TRALI [6, 7, 15]. An acute lung injury should be ascribed to TRALI or possible TRALI if other factors for ALI are present, provided that the signs and symptoms commence within 6 h of the last blood component transfused.
TRALI has been reported following transfusion of plasma-containing blood components [13, 17–25]. Estimates of the incidence of TRALI include: 1 in 5000 blood and blood components transfused ; 1 in 2000 plasma-containing components ; 1 in 7900 units of fresh frozen plasma ; 1 in 432 units of whole blood-derived platelet concentrates ; and 1 in 1323 blood components transfused . The fact that packed red cells and cryoprecipitate can cause TRALI suggests that small quantities of plasma (as little as 10% of a blood component unit) are sufficient to cause TRALI .
However, TRALI is believed to be under-diagnosed because of difficulties in diagnosis and lack of awareness [2, 3, 28]. Various haemovigilance networks have reported a greater incidence of TRALI than was previously thought. The FDA estimated that TRALI is the leading cause of transfusion-related death in the US [1, 29, 30]. Similarly the United Kingdom Serious Hazards of Transfusion Committee (SHOT) reported that TRALI was a leading cause of transfusion-related morbidity and mortality in the UK . The French haemovigilance network reported that 15% of transfusion related fatalities were caused by TRALI . In Germany, 101 out of 765 cases associated with complications of blood transfusion that were not related to infections were caused by TRALI . In 23 of these cases, antileucocyte antibodies directed against antigens on host leucocytes were demonstrated in donor plasma .
Two theories have been proposed to explain the pathogenesis of TRALI. The first is an antibody-mediated reaction following transfusion of antigranulocyte antibodies into patients who have leucocytes that express the cognate antigens. The second postulates that it is mediated by an interaction between biologically active mediators and the lung.
The antibody-mediated model
The antibody-mediated model postulates that the reaction is secondary to antibodies in donor plasma against antigens present on the recipient's leucocytes. These may be antibodies to the human leucocyte antigen (HLA) or to other leucocyte antigens. HLA antibodies may be directed against either HLA class I antigens that are present in all leucocytes or HLA class II antigens found on B lymphocytes and monocytes [7, 19, 33–38] or infusion of donor leucocytes into a recipient whose antibodies are directed against these donor leucocytes [17, 39, 40]. Most reactions are caused by antibodies in donor plasma and less than 10% of reactions are caused by plasma antibodies in the recipient that agglutinate the donor leucocytes [17, 39, 40]. The antibody–antigen interaction causes complement activation, resulting in the pulmonary sequestration and activation of neutrophils, endothelial cell damage and a capillary leak syndrome in the lungs leading to TRALI.
Rabbit and rat models for antibody-mediated TRALI have been developed [41–43]. The rabbit model is an ex vivo model utilising isolated rabbit lungs perfused with a mixture containing human granulocytes (5b-positive), antibodies to granulocyte 5b, and rabbit plasma as a source of complement . In this model, pulmonary oedema occurred 3–6 h after infusion of the mixture. There was no pulmonary oedema if any one of the three components was absent . The rat model is an ex vivo model of TRALI in which rat lungs are perfused with neutrophils that bear HNA-2a antigen and HNA-2a antibodies. This caused increased pulmonary capillary leakage, increased total lung water and increased lung neutrophil accumulation [42, 43]. The activation of neutrophils with subsequent release of reactive oxygen species led to TRALI . In this model TRALI developed without the addition of complement, thus bringing into question its role in the reaction [42, 43]. The pulmonary changes observed in the rat and rabbit models are similar to those observed in a study of patients with TRALI at autopsy . This study also demonstrated granulocyte sequestration within the pulmonary capillaries, extravasation of granulocytes into the alveoli and pulmonary oedema with proteinaceous material in the alveoli .
Clinical studies of patients who experienced TRALI have demonstrated the presence of either anti-HLA or antigranulocyte antibodies in donor plasma [7, 17, 25, 33–39, 45]. In two large series of patients with suspected TRALI, specific anti-HLA and antigranulocyte antibodies have been found [36, 46]. In a case series of 46 patients Popovsky and Haley  identified antibodies to human leucocyte antigens in 28% and granulocyte antibodies in 41% of the patients. An earlier study reported leucocyte antibodies in 89% and lymphocyte antibodies in 72% of 36 patients with a clinical syndrome resembling TRALI . Other studies have demonstrated antigranulocyte antibodies specific for HNA-3a (5b) [40, 47], anti-NB2 , HNA-2a (NB1)  and anti-A2  in the implicated donor units.
Several ‘look-back’ studies followed up recipients of blood components from a donor implicated in a TRALI reaction to determine the risk associated with subsequent transfusions from such donors [28, 50–53]. Kopko et al. examined the records of previous recipients of a multiparous female frequent blood donor who had a granulocyte 5b antibody . Seven mild to moderate reactions were identified in six recipients and eight severe reactions were identified in eight recipients. Cooling examined the case records of recipients of blood products from a multiparous donor who had anti-HLA class I antibodies and that found 15% of recipients experienced TRALI . Win et al.  examined the clinical records of previous recipients of blood products from six donors who had antileucocyte antibodies. They found that each donor was associated with one documented case of TRALI but there was no evidence of TRALI reactions in other recipients. A recent study by Toy et al.  examined the records of 103 patients who had received blood from a donor associated with a known TRALI reaction. Only one patient developed TRALI even though 54 patients had known HLA antigens to the donor's antibodies. Nicolle et al. found that only one out of the 18 patients who received blood components from two donors who were implicated in TRALI reactions experienced TRALI . In summary these ‘look-back’ studies demonstrated that single donors could cause TRALI reactions in more than one patient, lending support for the antibody-mediated hypothesis. However, blood containing antileucocyte antibodies may be transfused without causing TRALI, indicating that the antibody alone is insufficient to cause the syndrome.
There is also evidence that TRALI is commoner in recipients of blood products from multiparous donors who are more likely to possess anti-HLA and anti-HNA antibodies [54–58]. A randomised controlled crossover trial of plasma obtained from multiparous donors compared with plasma obtained from women with no history of transfusion or pregnancy reported that one patient developed TRALI after transfusion of plasma obtained from a multiparous donor . IgG antineutrophil antibodies were found in the multiparous donor's plasma . A significant reduction in oxygenation followed transfusion of plasma obtained from the multiparous donors but not the control donors. Insunza et al.  examined the effect of excluding plasma derived from multiparous donors which tested positive for anti-HLA antibodies. There was a reduction in reported episodes of TRALI, from five cases in the 25 months before their intervention to no cases in 15 months after the intervention. However, a major criticism of this study was the potential for under-reporting the incidence of TRALI because it relied on voluntary reporting by clinicians to the regional blood bank.
The two-event (biologically active mediator) model
The biologically active mediator model postulates that TRALI is the result of two events. The first event is the clinical condition of the patient, resulting in pulmonary endothelial activation and neutrophil sequestration, and the second event is the transfusion of biologically active mediators (lipids, antigranulocyte antibodies) that activate adherent neutrophils leading to endothelial damage, capillary leak and TRALI [18, 60–62].
The first event can be caused by a variety of insults to the pulmonary vascular endothelium such as sepsis, cardiopulmonary bypass, haematological malignancy, thermal injury and trauma [17, 18, 41, 63]. In an isolated rat lung model non-cardiogenic pulmonary oedema developed after the pulmonary vasculature was primed by lipopolysaccharide with subsequent infusion of supernatant from 42-day-old red cell concentrates and from 5-day-old platelet concentrates [60, 64]. ALI did not develop when fresh (day 0) red cell or platelet supernatants were used or when the tissue was not initially primed with lipopolysaccharide [60, 64].
Pre-existing clinical conditions contribute to the first event of the two-event model. In a retrospective case-control study by Silliman et al. , 10 patients who experienced TRALI were compared with 10 patients with febrile or allergic transfusion reactions. All the patients with TRALI had one or more predisposing clinical conditions: active infection in five patients; recent surgery in five patients; cytokine administration in two patients and massive blood transfusion in one patient. Only two patients out of the 10 control patients had predisposing factors.
In a prospective case-control study of TRALI , the roles of cytotoxic HLA class I, class II and antigranulocyte antibodies were examined. The investigators examined 81 patients who experienced 90 TRALI reactions. The first 46 reactions were analysed in a nested case control study. Patient and blood transfusion data in these patients were compared with a control group of 225 recipients of platelet concentrates. TRALI was associated with cardiac disease or a haematological malignancy in the patients, and an increasing age of the platelet concentrate transfused. TRALI was not correlated with the patients' age or gender, the number of previous blood transfusions, or the type or number of previous transfusion reactions. There was increased neutrophil priming activity (neutral lipids and lysophosphatidylcholines) in all TRALI patients. This study showed that TRALI was caused by two events: the first was the clinical condition of the patient, and the second was the infusion of active lipids in the stored blood component . Inflammatory cytokines and lipids accumulate during storage of platelets and red blood cells [18, 26, 60, 65–69]. Platelet-derived CD40 ligand may be a cofactor in the pathogenesis of TRALI. CD40 ligand accumulates during routine storage of packed red blood cells, whole blood and platelet concentrates (especially apheresis platelets). CD40 ligand primes neutrophils and this leads to pulmonary endothelial damage . The inflammatory cytokines increase with the duration of storage and are decreased by pre-storage leucodepletion . However, leucodepletion has not been shown to reduce the incidence of TRALI . Studies examining the relationship between patient morbidity and duration of storage of blood products have, however, reported conflicting results such as no impact on morbidity [71–74], an increased risk of nosocomial pneumonia , an increased risk of multiple organ failure  and an increased overall mortality .
Kopko et al. demonstrated that the antineutrophil [HNA-3 (5b)] antibody can prime neutrophils and cause a respiratory burst effect, similar to the effects of biologically active lipids (that accumulate during routine blood storage) on neutrophils . They suggested that TRALI occurred in patients when the pulmonary vascular bed was primed by sepsis, trauma or transfusion followed by the activation of primed neutrophils by either biologically active mediators or by antigranulocyte antibodies leading to pulmonary vascular endothelial damage .
Autopsy of patients who have died from TRALI demonstrates dilation of pulmonary capillaries associated with sequestration of granulocytes within the capillaries, extravasation of granulocytes into the alveoli, interstitial and intra-alveolar oedema and the presence of proteinaceous material in the alveoli . The oedema fluid has a high protein concentration (oedema fluid/plasma protein ratio = 0.75) [78–80].
Clinical manifestations and diagnosis
The diagnosis of TRALI is based on the criteria recommended by the TRALI consensus conference: there must be no pre-existing ALI before transfusion, the onset of signs and symptoms must occur during or within 6 h of transfusion and there must be no temporal relationship with an alternative risk factor for ALI [6, 7]. It is important to remember that TRALI may be caused by single unit transfusions of any blood product [6, 15]. Common symptoms and signs of TRALI include progressive dyspnoea, tachypnoea, frothy sputum, hypoxaemia and hypotension or (rarely) hypertension [6, 7, 17, 18, 28, 36, 46, 81–83]. The chest radiograph commonly shows bilateral pulmonary infiltrates consistent with non-cardiogenic pulmonary oedema. It is difficult to differentiate TRALI from other causes of ALI other than on the basis of history because the clinical features of TRALI are non-specific . It is important to exclude elevated left atrial pressures and signs of left ventricular failure. Surveillance by SHOT and the FDA suggest that TRALI is more common than expected and should be considered when these symptoms and signs occur within 6 h of the completion of a transfusion.
There are no specific laboratory markers to confirm TRALI. However, neutropenia  and the finding of antileucocyte antibody–antigen pairs support the diagnosis [18, 33]. A high protein content in pulmonary oedema fluid can help to differentiate TRALI from fluid overload and cardiogenic pulmonary oedema [78–80, 85].
TRALI reaction should be reported to the transfusion service within the hospital so that antileucocyte antibody screening tests can be performed on the donor plasma and donors who have anti-leucocyte antibodies excluded from the donor pool.
The management of TRALI is supportive. Transfusion of the suspected blood product should cease immediately when TRALI is suspected [7, 82, 83, 86]. Respiratory support is dictated by the clinical picture (varying from oxygen supplementation in mild TRALI to non-invasive ventilation or tracheal intubation and mechanical ventilation in severe TRALI) . The optimal method of ventilation is unknown [7, 82, 83, 86, 87]. In ALI and the acute respiratory distress syndrome (ARDS) randomised controlled trials support the use of small volume, high frequency ventilation with limits set on inspired pressures, optimising positive end-expiratory pressure (PEEP) and fraction of inspired oxygen, and allowing permissive hypercapnia . In a series of 37 patients with TRALI, all patients required oxygen supplements and 72% required mechanical ventilation . Haemodynamic support may be required. The use of diuretics is controversial [7, 18, 28, 82, 83, 86, 88, 89] and hasty and uncontrolled use of diuretics may be harmful [22, 90, 91]. The efficacy of corticosteroids in TRALI is unproven [7, 18, 28, 82, 83, 86, 88, 89, 92, 93], although there are anecdotal case studies that have reported the use of high dose steroids as part of a successful treatment regimen [4, 94, 95].
Most patients with TRALI recover within 48–96 h [17, 18, 27, 96] but hypoxaemia and radiological evidence of pulmonary infiltration can persist for 7 days in 20% of patients . Approximately 70% of patients require mechanical ventilation. The in-hospital mortality is estimated at 5–10%[17, 46, 63].
The most important step in reducing episodes of TRALI is to limit the transfusion of blood products using [16, 86, 97–100]. In a multicentre randomised controlled clinical trial of transfusion requirements in critical care, critically ill patients were randomised to maintain a haemoglobin of 7–9 g.dl−1 or 10–12 g.dl−1. The patients who were randomly assigned to the lower haemoglobin group received fewer units of red cells. Although the two groups had a similar 30-day mortality, there was a trend towards a lower in-hospital mortality in the restrictive transfusion group (22.2%vs 28.1%, p = 0.05). Younger (age < 55 years) and less critically ill patients (Acute Physiology and Chronic Health Evaluation II score ≤ 20) had a significantly lower mortality (8.7% in the restrictive-strategy group and 16.1 in the liberal-strategy group; p = 0.03). The number of patients with multi-organ dysfunction was not statistically different between the two groups nor was the incidence of ARDS (7.7% in the restrictive group vs 11.4% in the liberal strategy group) . The risk of TRALI provides further stimulus to utilise transfusion guidelines at the lower threshold. The British Committee for Standards in Haematology Guidelines supports the use of strict transfusion thresholds in chronic and acute anaemia . Peri-operative strategies that reduce transfusion include pre-operative optimisation using dietary supplements or erythropoietin, prevention of hypothermia, the use of antifibrinolytic drugs and cell salvage [102–104]. Coagulation factor and platelet replacement should be guided by formal laboratory tests or thromboelastography wherever possible. The British Committee for Standards in Haematology have issued guidelines outlining the appropriate, evidenced-based use of platelets and fresh frozen plasma in a variety of clinical settings to limit blood product use [105, 106].
The management of donors to prevent episodes of TRALI is a difficult and contentious issue [7, 15, 86]. The preservation of blood supplies whilst maintaining appropriate patient safety remains a problem. At present there is no universally agreed approach to donor management. Given that there is evidence of TRALI occurring in multiple recipients linked to single donors [28, 50], it is suggested that donors implicated in TRALI and who have demonstrable antibodies should be permanently disqualified from the donor pool [33, 81, 86, 88, 107]. This prudent approach is supported by the 2004 consensus conference on TRALI [7, 15].
Several studies have demonstrated that multiparous donors are more likely to possess antileucocyte antibodies [54–58]. Studies also suggest that there is a higher incidence of TRALI associated with transfusions of plasma derived from multiparous women [46, 58, 96], although a causal link between plasma from multiparous donors and TRALI is not always clearly established [55, 108]. Insunza et al. reported a reduction in the incidence of TRALI by excluding multiparous donors from contributing to the plasma pool . The simple exclusion of multiparous donors would substantially reduce the donor pool. In 2002, SHOT recommended that consideration be given to excluding female donors . It is estimated that at present 90% of plasma products are derived from male donors in the UK as a result of this policy . It has not yet been demonstrated that this policy has reduced the incidence of TRALI in the UK.
There have been arguments for shorter periods of blood product storage [18, 58, 81, 110] because inflammatory cytokines and lipids accumulate during storage of platelets and red blood cells [18, 26, 60, 65–69]. However, current evidence does not demonstrate a clinically significant increase in TRALI associated with the increased duration of storage of blood products [71–74]. Although pre-storage leucodepletion may decrease biologically active mediators associated with prolonged storage [60, 69, 86], the incidence of TRALI is not significantly reduced with the use of leucodepleted blood products [68, 111]. The use of pooled solvent/detergent-treated plasma or only male donor plasma minimises the risk of TRALI from fresh frozen plasma but does not obviate the risk of TRALI from cellular blood components.
TRALI presents as a spectrum of transfusion reactions that range from mild respiratory impairment to severe fulminant and fatal pulmonary injury. Cases of TRALI reported to blood banks may represent just the tip of an iceberg, and transfusion may play an important role in more cases of ALI than currently realised. TRALI has been largely unrecognised because of a lack of appreciation of the clinical picture by clinicians and transfusion medicine specialists alike. The diagnosis and prevention of TRALI requires a close working relationship between the clinical medical staff and the blood bank.