Portal de Periódicos da CAPES

Does red‐cell T activation matter?

2001; Wiley; Volume: 114; Issue: 1 Linguagem: Inglês

10.1046/j.1365-2141.2001.02886.x

1365-2141

Anne F. Eder, Catherine S. Manno,

Erythrocyte Function and Pathophysiology

Debate regarding appropriate transfusion management of patients demonstrating red-cell T activation centres on whether the infusion of plasma-containing blood components causes haemolysis. Red-cell T activation results from enzymatic exposure of the T antigen or other carbohydrate determinants on the surface of the red cell. Ubiquitous IgM antibodies in donor plasma directed against the T antigen and related epitopes have been temporally associated with variable degrees of haemolysis in some patients with T-activated red cells. Current practice among transfusion medicine specialists ranges from no special precautions for patients at high risk of red-cell T activation to specific measures to identify patients with T activation and provide blood or plasma components with absent or low titres of anti-T antibodies (Williams et al, 1989; Rodwell & Tudehope, 1993; Engelfriet & Reesink, 1999). Some practitioners have advocated completely avoiding the administration of plasma-containing products to patients with T activation (Novak et al, 1993), while others caution that plasma should not be withheld from such patients if they require coagulation factors to restore haemostasis (Crookston et al, 2000). Based on the available published information, the question posed by the title, 'does T activation matter?', will be explored in this Annotation. Although arguments have been formulated on both sides, definitive clinical data on which to draw firm conclusions are not available. Rather than advocate specific transfusion practice based on their experience and interpretation of the available evidence, the authors instead address practice points for each side to consider. Named after the original investigators, T activation, or the Thomsen–Hübener–Friedenreich phenomenon, describes the enzymatic modification of the red-cell membrane to expose the T, or Thomsen, antigen (Judd, 1992). The T antigen is present on the surface of all red cells but remains masked until N-acetyl-neuraminic acid (sialic acid) residues are removed by neuraminidase, an enzyme produced by a wide variety of microorganisms, most notably Clostridia species, Streptococcus pneumonia and influenza viruses. Infectious agents similarly induce other acquired red-cell modifications by exposing T-related carbohydrate antigens, such as Th, Tk and Tx. Th may represent early and mild T activation, when only a small amount of sialic acid has been cleaved from the red-cell surface, although qualitative differences between T and Th antigens have been demonstrated (Herman et al, 1987). Th activation also has a non-infectious aetiology, occurring in children with hypoplastic anaemias and otherwise healthy pregnant women and newborn infants (Wahl et al, 1989). Tk activation results from the activity of endo- and exo-β-galactosidases and has been linked to infections with Bacteroides fragilis, Serratia marcescens and Candida albicans. Tx has been described in children with pneumococcal infection (Bird et al, 1982). The different types of T activation can be distinguished by the reactivity of the exposed antigens with various plant lectins. For example T, Th, Tk and Tx antigens bind the peanut lectin Arachis hypogaea, but only T antigen also binds the soybean lectin, Glycine soja (Issitt & Anstee, 1998). While an interesting academic pursuit and research endeavour, discriminating among these types of T activation may not be practical or useful in a clinical setting. The clinical course of necrotizing enterocolitis (NEC) in infants with T-activated red cells did not differ significantly from infants with Tk-activated red cells (Osborn et al, 1999). Most investigators use only peanut lectin to identify T activation, which reacts with both T and Tk antigens indiscriminately. T activation may cause polyagglutination in vitro, which is the unexpected agglutination of the red cells by sera from ABO blood group-compatible adults; however, not all examples of T activation defined by reactions with lectins demonstrate this property (Judd, 1992; Crookston et al, 2000). Polyagglutination occurs as a result of IgM antibodies in donor plasma directed against the exposed carbohydrate determinants of T-activated red cells. Anti-T appears in the plasma of almost all individuals by about 3 months of age and rises to adult levels by 2 years of age. Although often referred to as 'naturally occurring', these antibodies are probably generated following exposure to ubiquitous bacteria or other environmental stimuli (Springer & Tegtmeyer, 1981). Anti-T has been identified as a possible cause of intravascular haemolysis in adults and older children with T-activated cells who possess these antibodies, and in infants with T-activated red cells who lack these antibodies but receive blood components containing plasma from adult donors (Moores et al, 1975; Seges et al, 1981; Klein et al, 1986; Williams et al, 1989; Novak et al, 1993). Although case reports have suggested an association between passively acquired anti-T and haemolysis, the importance of anti-T has been questioned based on serological grounds, animal studies and other clinical observations, described in greater detail below (Eversole et al, 1986; Judd, 1992; Issitt & Anstee, 1998; Crookston et al, 2000). T activation, defined by red cell–lectin reactivity, was observed in 0·5% of an unselected hospitalized adult population (Rawlinson & Stratton, 1984), in 7·6% of a hospital population considered to be at higher risk for T activation than the general population because of malignancy, sepsis and anaemia (Buskila et al, 1987), and in 7% of acquired immune deficiency syndrome (AIDS) patients (Adams et al, 1989). Th activation has also been reported in healthy populations, occurring in 0·5–1·5% of adult blood donors, 6% of pregnant women and 11% of healthy neonates, with a high degree of concordance between mothers and infants (Herman et al, 1987). Polyagglutination, which depends on the extent of T activation and the titre of anti-T in donor plasma, occurs at a lower frequency than lectin-detectable T activation, estimated at 1:10 000 of hospitalized adults (Rawlinson & Stratton, 1984). In paediatric hospitals, T activation is most commonly detected in infants with NEC and children with invasive pneumococcal infections, reflecting the ability of most bacterial isolates of Clostridium perfringens and S. pneumoniae, respectively, to produce enzymes that expose red-cell T antigens in vitro. Among 1672 infants admitted to a neonatal intensive care unit over a period of 2 years, 10 (0·6%) had detectable T activation defined by peanut lectin reactivity. In the subset with NEC, 8 out of 72 (11%) infants demonstrated T activation, with an even higher proportion (27%) of T activation among those infants with NEC who required surgical intervention (Williams et al, 1989). The correlation between T activation and disease severity in NEC is a consistent observation in clinical studies and may reflect bacterial load. Klein et al (1986) reported T activation in 17 out of 62 (27%) infants with NEC, with 76% of these infants requiring surgery compared with only 18% of infants who did not demonstrate T activation. Clostridia species were isolated in culture from the vast majority of infants with T activation and advanced disease (88%) compared with only 24% of infants without T activation. Osborn et al (1999) detected T or Tk antigen in 17 out of 201 (19%) infants with confirmed NEC and reported significantly increased morbidity and mortality among infants with T activation. Finally, T activation was detected in three out of seven children with haemolytic–uraemic syndrome (HUS) associated with S. pneumoniae bacteraemia (Mizusawa et al, 1996). The temptation to invoke an immune-mediated mechanism for clinical haemolysis exists whenever an antibody against a specific red-cell epitope is identified in the laboratory. Anti-T antibodies can cause red-cell agglutination and are plausible suspects for intravascular haemolysis of T-activated red cells. Most arguments implicating anti-T, however, are based on the temporal association of intravascular haemolysis with the transfusion of plasma-containing blood products. Numerous case reports and series, dating back to 1955, describe an association between passively transferred anti-T and haemolysis in patients with T-activated red cells (Crookston et al, 2000). In a 4 year study, the outcomes of 62 infants with NEC managed with knowledge of their T-activation status were compared retrospectively to 66 historical controls whose T-activation status was not determined (Klein et al, 1986; Novak et al, 1993). Screening revealed 17 infants (27%) with T activation, and transfusion of plasma products was completely avoided in 13 of these patients. Four infants, however, were transfused with plasma-containing components prior to diagnosis of NEC and developed haemolysis, which contrasted with 16 infants in the comparison group who developed post-transfusion haemolysis. There were three deaths in the screened group and 12 deaths in the comparison group. The authors reported these differences as statistically significant and concluded that the administration of plasma products should be strictly avoided in infants with NEC and T activation. A major shortcoming of this study is the fact that T activation in the control group was not assessed; consequently, the role of red-cell modification in these cases of transfusion-associated haemolysis cannot be determined and should not be inferred. Haemolysis could have occurred in infants whose red cells were not T activated. Furthermore, the retrospective design of this study confounds interpretation because of the possibility of unbalanced cohorts with respect to severity of disease, co-morbid conditions or treatment differences over time. In a prospective series of 1672 infants admitted for intensive care over a 3 year period, Williams et al (1989) implemented routine screening and a transfusion protocol for patients with T activation, restricting cellular components to washed red cells or washed platelets, selecting donors with low titre anti-T for plasma-containing blood products for patients with mild T activation, and avoiding plasma transfusion to patients with strong T activation, if possible. T activation occurred in 10 infants (0·6%) overall and in 8 out of 72 infants with NEC (11%). All 10 infants with T-activated red cells were transfused during the course of their illness. Four of the 10 patients received unselected plasma-containing blood products before T activation was detected; one patient with strong (4+) T activation had massive haemolysis; the other three patients with weaker (1+ to 2+) T activation had mild haemolysis. The authors did not define their criteria for the degrees of haemolysis reported. Two patients developed no haemolysis following infusion of low titre anti-T reduced-volume platelets or low titre anti-T fresh-frozen plasma; four patients developed no haemolysis despite transfusion of washed red cells or washed platelets. T activation was transient in all patients, lasting from 1 to 12 d. The incidence of haemolysis in infants without T activation was not reported. There was no increase in the mortality rate (13%) for infants with NEC and red-cell T activation who were managed on the conservative transfusion protocol compared with the mortality rate (14%) for all patients with NEC. The authors concluded that haemolysis occurred only in those patients with T activation who received plasma-containing blood products, and advocated screening for T activation and use of a similar transfusion protocol. Although passively transferred anti-T was implicated in these studies, other possible explanations for haemolysis in the setting of T activation include direct haemolytic action of bacterial enzymes and toxins (Hübl et al, 1993), altered interaction of complement components with desialyated red cells (Marshall et al, 1996) and shortened survival of red cells with reduced membrane sialic acid content (Durocher et al, 1975; Perret et al, 1980; Bratosin et al, 1998). The mechanism of immune-mediated haemolysis as a result of anti-T has been questioned based on practical considerations, serological grounds, animal studies and clinical reports (Issitt & Anstee, 1998; Engelfriet & Reesink, 1999; Crookston et al, 2000). T activation is not uncommon, occurring in 10–30% of infants with NEC, yet clinically significant haemolysis is rarely attributed to plasma transfusion, despite the frequent need for plasma in these critically ill infants (Engelfriet & Reesink, 1999). Because many infants are not screened for T activation prior to transfusion, some patients with T-activated red cells probably receive plasma without clinically detectable problems. Several transfusion centres, even those that serve large obstetric and neonatal populations, have concluded that haemolysis after plasma transfusion is a rare event and one which may not be prevented even with pretransfusion screening for T activation (Engelfriet & Reesink, 1999). For example, haemolysis occurring in patients with T activation on screening protocols has been attributed to transfusion of plasma prior to detection of T antigens (Williams et al, 1989; Osborn et al, 1999). Moreover, haemolysis can occur in infants with NEC in the absence of T activation and anti-T antibodies (Osborn et al, 1999; Crookston et al, 2000), suggesting causes other than immune-mediated haemolysis. In addition to these practical considerations, the serological properties of anti-T make it an improbable culprit in intravascular haemolysis. Although thought to arise from environmental stimuli such as ABO isohaemagglutinins, anti-T antibodies have serological characteristics similar to incidental and clinically insignificant antibodies (Issitt & Anstee, 1998). Anti-T antibodies are invariably cold-reactive, IgM agglutinins, optimally reacting with red cells below room temperature. In tests performed at 37°C, anti-T causes neither red-cell agglutination nor complement activation (Issitt & Anstee, 1998; Crookston et al, 2000). The direct antiglobulin test (DAT) in cases of suspected anti-T-mediated haemolysis are often negative. Studies that report positive DATs are difficult to interpret without appropriate controls or additional information regarding the specific type of reagent used in the test, because polyclonal reagents may contain anti-T and yield false positive results (Judd, 1992; Issitt & Anstee, 1998). Furthermore, the physiological significance of the antibody is not supported by animal studies which show that treatment of red cells with neuraminidase sharply decreases red-cell survival independent of anti-T titre (Perret et al, 1980; Crookston et al, 2000). Finally, clinical observations do not consistently support a relationship between transfusion and haemolysis in patients with T-activated red cells. There are reports describing cases of T activation in which administration of blood components, including plasma, did not lead to haemolysis (Heddle et al, 1977; Davis et al, 1985; Eversole et al, 1986) or haemolysis began before blood components were given (Novak et al, 1983; Sayas et al, 1990; Squire et al, 1992). Data are also conflicting on the clinical benefit derived from screening for T activation. Osborn et al (1999) retrospectively compared the outcomes of 110 infants diagnosed with NEC before implementation of universal T activation screening with 91 infants diagnosed and managed subsequently. This historical comparison revealed an increase in the incidence of surgical intervention (32% versus 56%) and mortality (11% versus 24%) from NEC after introduction of routine testing that were not significant differences after adjustment for several perinatal risk factors. These results contradict the findings of Novak et al (1993), who attributed decreased mortality (18% versus 5%) from NEC to the introduction of screening for T activation and avoidance of plasma-containing blood products, without considering possible confounding variables. Osborn et al (1999) also conducted a case–control study among infants with NEC, matching 17 infants with T activation to 28 controls, with respect to birthweight, gestational age and birth date. Antenatal and post-natal characteristics revealed no significant differences between the two groups. All infants with known T activation were transfused with plasma-free or low titre T blood products. 'Low titre anti-T' was not further defined; haemolysis was defined as the presence of red-cell fragments or spherocytes on a peripheral blood smear. Despite the use of low titre anti-T plasma components, a significantly higher mortality rate was observed in T-activated infants (35%) than in control infants (7%; P = 0·04), as well as a higher incidence of haemolysis in T-activated infants (71%) than in matched control infants (21%; P = 0·002). Unlike previous investigators, these authors could not correlate haemolysis in the infants with T-activated red cells with the use of fresh-frozen plasma. Further comparison between studies is hampered by different definitions of haemolysis and study design, but these results are not consistent with the suggestion of Williams et al (1989) that the use of low titre anti-T blood components may prevent haemolysis and improve the outcome for T-activated infants with NEC. A noticeable difference among studies is the proportion of infants who received plasma-containing products prior to diagnosis of NEC and detection of T activation. Osborn et al (1999) reported a greater proportion (14/17 infants) than Novak et al (1993) (4/17 infants) or Williams et al (1989) (4/10 infants), raising the possibility that anti-T persisted in the circulation until T activation occurred and subsequently caused haemolysis. This observation could be interpreted as suggesting that avoidance of plasma-containing blood components or provision of low titre anti-T blood components may be important in infants with NEC even prior to T activation. Additional observations of Osborn et al (1999), however, are not consistent with this speculation, as infants who received only albumin still required surgical intervention and demonstrated moderate to severe haemolysis prior to death. Clinicians have either accepted or rejected the relevance of T activation in transfusion decisions, based on their experience, interpretation of the available evidence or both. Definitive data to support clinical decisions, however, is lacking. Consequently, the advantages and limitations of the available options, taking no preventive measures or implementing screening and transfusion protocols for T activation, are discussed. Although a temporal relationship exists between the presence of anti-T in donor plasma and haemolysis of T-activated red cells in many case reports, a causal role has not been established. Non-immune mechanisms may cause or contribute to the haemolysis observed in patients with T-activated cells. Those transfusion services that have concluded that anti-T is not a probable cause of haemolysis in patients whose red cells are T-activated are charged to remain vigilant for unexplained cases of haemolysis and investigate all diagnostic possibilities. Communication between the neonatologists and transfusion service is essential for proper investigation. High-risk infants, with suspected NEC or sepsis, should be promptly evaluated for clinically evident haemolysis or failure to achieve an increase in haemoglobin concentration following transfusion. Laboratory evaluation of haemolysis in critically ill infants, especially preterm infants, is often complicated by several factors, including fluctuating haemoglobin levels owing to phlebotomy losses or intravascular fluid shifts, age-specific variability in reticulocyte count and lower than expected serum haptoglobin concentrations (Pisciotto & Luban, 1996). Hyperbilirubinaemia is common in premature infants and may be as a result of causes other than haemolytic disease, such as immaturity of the fetal liver or hepatic injury owing to hypotension, endotoxins or other physiological insults. Definitive diagnosis of immune haemolysis in infants may require careful correlation of clinical and laboratory data. Possible causes of intravascular haemolysis following transfusion include ABO plasma incompatibility (Boothe et al, 1995; Larsson et al, 2000), mechanical destruction of red cells resulting from rapid transfusion through small lumen catheters or osmotic lysis owing to simultaneous infusion or inappropriate addition of hypotonic solutions to blood components (Vengelen-Tyler, 1999). Other causes unrelated to transfusion include glucose-6-phosphate deficiency, sepsis, disseminated intravascular coagulation or antibiotics (i.e. ceftriaxone) (Pisciotto & Luban, 1996). Despite the low probability of haemolysis owing to passively transferred anti-T, clinicians in paediatric facilities may be unwilling to accept even small risks of further worsening the clinical condition of critically ill infants if there are alternative treatment options. Prudence dictates caution when haemolysis is clearly related in time to the transfusion of plasma-containing blood products and all other possible causes of haemolysis are eliminated. Measures should be taken, if possible, to limit transfusion of plasma-containing components to these infants. Transfusion services concluding that passively transferred anti-T has not been eliminated as a potential cause of haemolysis in infants with T-activated red cells may consider implementation of a testing strategy and transfusion management protocol. Universal screening of all infants upon admission to the intensive care unit is not warranted, and would have an unacceptably low yield and predictive value. Although advocated by some investigators (Novak et al, 1993), there is not even justification for routine screening of all infants with NEC or sepsis or for routine pretransfusion testing of these infants, based on the high incidence of T activation but the less frequent observation of clinically significant haemolysis. A rational approach to a selective screening strategy is to perform tests for T activation for infants with NEC or sepsis with symptoms of haemolysis (Engelfriet & Reesink, 1999) or to further restrict investigation to transfusion-associated events in infants who demonstrate (i) signs of intravascular haemolysis (i.e. haemoglobinuria, haemoglobinaemia and reduced serum haptoglobin concentration) following transfusion, or (ii) failure to achieve the expected post-transfusion haemoglobin increment following red-cell transfusion (Pisciotto & Luban, 1996). T activation is assessed in the laboratory by incubating the patients' red cells with commercially available lectins such as Arachis hypogaea; polyagglutination is assessed by incubating the patient's red cells with plasma from ABO-compatible adults. For controls, both negative and positive (neuraminidase-treated) red cells should be included in these studies. If the results of these screening tests reveal marked T activation and/or polyagglutination, a minor cross-match may be performed to select 'least incompatible' or 'low titre anti-T' plasma-containing components for infants. A difficulty in proffering practice guidelines is the lack of a standard definition of 'low titre anti-T plasma' and the restricted availability of such plasma products from Blood Banks in the USA. In addition, the minor cross-match is affected not only by the titre of anti-T in the plasma but also by the degree of T activation of the patient's red cells, and is relatively insensitive, detecting only two out of nine infants (22%) with weak T activation and five out of eight infants (63%) with strong T activation in one series (Klein et al, 1986). However, screening donors against neuraminidase-treated red cells is too sensitive, as all donor sera react and would be considered incompatible with the patient, and high levels of T-antigen expression (peanut lectin titre > 256 or 3+ to 4+) are probably more clinically significant than lower levels of T-antigen expression (Seges et al, 1981; Klein et al, 1986; Williams et al, 1989; Engelfriet & Reesink, 1999). For cellular blood components, residual plasma can be efficiently removed by saline washing. Saline washing does not adversely affect the function or survival of the transfused red cells; however, it results in loss of approximately 20% of the unit, is time-consuming and may delay the availability of the units (Vengelen-Tyler, 1999). In addition, washing violates the integrity of the unit, shortens the shelf-life of a unit from 35 or 42 d to 24 h and may increase the risk of bacterial contamination. The necessity of washing red-cell units stored in extended storage media (e.g. Adsol) has been questioned because nearly all residual plasma is removed during preparation of the component and further diluted with the addition of the preservative solution. In contrast, a platelet unit, collected by apheresis from a single donor, contains between 100 and 500 ml of plasma. Platelets may also be saline-washed to remove plasma, however, both yield and viability of the platelets is compromised with this manipulation (Pineda et al, 1989). Volume reduction could be used to decrease the amount of plasma in a unit, but may also be associated with unfavourable changes in the recovery of the platelets. Cryoprecipitate contains approximately 20 ml of plasma, depending on the amount of plasma supernatant left in the final product (Vengelen-Tyler, 1999). Other plasma-derived components, such as albumin, coagulation factor concentrates and intravenous immunoglobulin preparations, contain negligible amounts of IgM; hence, these products are not sources of anti-T antibodies. Infants with NEC and brisk, life-threatening haemolysis may be treated with red cell-exchange transfusion. Some of these infants may also require plasma, cryoprecipitate and platelet concentrates. The precise risk of exchange transfusion and use of plasma components in infants with NEC and T activation is difficult to assess owing to the critical nature of the illness, the small number of published cases and the anecdotal nature of reporting. Exchange transfusion and cautious use of plasma-containing blood products has been used successfully to treat an infant with NEC, marked T activation and massive haemolysis (Crookston et al, 2000). Similarly, limited evidence is available to guide transfusion decisions in infants and children with HUS, and the use of washed red cells and low titre anti-T plasma has been advocated based on the theoretical risk of T activation and plasma transfusion (Mizusawa et al, 1996). T antigen is exposed on the surface of platelets and renal epithelial cells during infection, as well as red cells, and anti-T antibody-mediated interactions with these cells have been postulated as contributory factors in haemolysis, thrombocytopenia and renal failure in HUS (Novak & Martin, 1983; Crookston et al, 2000). Direct evidence to support this hypothesis is lacking. Although caution may be warranted, strict avoidance of plasma-containing blood products in infants or children with T-activated red cells is unfounded and potentially dangerous. Transfusion medicine specialists are justifiably concerned with the overutilization of blood components; however, underutilization may also be associated with adverse clinical outcomes. In a survey conducted by the College of American Pathology in 1985, more than 11% of responding institutions said they would have 'refused to provide any component' to a patient who had red-cell T activation and to whom a 'plasma-containing product must be given' (Mintz, 1999). The possibility that urgently required components would not be released in a timely manner is disquieting. Although some case reports suggest a temporal association between anti-T in plasma-containing blood components and haemolysis of T-activated red cells, a causal relationship has not been established. There is conflicting clinical evidence regarding the effect of routine testing for T activation on the mortality associated with NEC and the use of low titre anti-T blood components. To date, the only clinical study of infants with NEC and T activation that attempted to compare T-activated infants with control infants and analyse confounding variables did not demonstrate any benefit of these practices. The balance of the available information currently favours the interpretation that routine screening protocols and strict avoidance of plasma-containing blood products in individuals with T activation are not warranted. A carefully designed and well-executed randomized controlled trial of screening for T activation in high-risk infants and provision of low titre anti-T plasma components may provide definitive data on which to base practice recommendations. Any further studies should include an appropriate and explicit definition of haemolysis, the incidence of haemolysis in T-activated and control infants, and detailed descriptions of the temporal relationship to the specific blood components administered, and should adhere to sound methodological principles including a power calculation, explicit eligibility criteria and appropriate statistical analysis. Until such evidence is available, transfusion services must understand the limitations of the current information and use their best judgement in practice. Those transfusion services that have rejected the hypothesis that passively transferred anti-T antibodies mediate haemolysis in T-activated infants should ensure appropriate investigation of unexplained cases of haemolysis associated with plasma transfusion. Those transfusion services that still consider anti-T as a potential cause of haemolysis may implement a selective testing strategy and transfusion management protocol. However, strict avoidance of plasma-containing blood components in infants with T-activated red cells in an attempt to avoid a potential risk of haemolysis may introduce a real risk of neglecting the need for coagulation factors in critically ill infants.