Microparticles in Stored RBC as Potential Mediators of Transfusion Complications

Abstract

This article reviews evidence for the involvement of cell-derived microparticles (MP) in transfusion-related adverse events. The controversy concerning possible added risk of older vs. fresher stored blood is also reviewed, and is consistent with the hypothesis that MP are involved with adverse events. Although all types of circulating MP are discussed, the emphasis is on red cell-derived MP (RMP). The evidence is particularly strong for involvement of RMP in transfusion-related acute lung injury (TRALI), but also for post-operative thrombosis. However, this evidence is largely circumstantial. Work in progress to directly test the hypothesis is also briefly reviewed.

Introduction

Red blood cells (RBC) stored in the blood bank undergo a series of physical and chemical changes and release many potentially hazardous products, increasing with time, resulting in the so-called “storage lesion”¹. In view of some but not all recent studies, it is widely believed that transfusion with younger blood carries less risk of adverse reactions than older blood^2–6. However, there is little agreement on the “safe” age of blood, nor is it clearly understood why older blood may carry increased risks.

It has been known since the 1970’s that stored whole blood, platelet concentrates, and RBC release submicron-sized fragments of the cells’ plasma membranes to the supernatant, and that their numbers increase with time of storage ^{7, 8}. These fragments, or vesicles, are commonly termed “microparticles” (MP) and constitute one aspect of the storage lesion. A series of studies has shown that MP released from blood cells exhibited strong procoagulant and proinflammatory activities^9–12. In addition, red cell microparticles (RMP) also contain hemoglobin, a potent scavenger of nitric oxide (NO), which has been shown to modulate vascular contractility via NO pathway¹³. Several relevant reviews have appeared in recent years^9–12. This review will focus on the generation of MP during blood storage, and on evidence supporting the hypothesis that MP act as a mediator of transfusion-related inflammatory and thrombotic complications.

Risk Associated with Age of RBC

It is our working hypothesis that RMP or other MP released from stored RBC may contribute to at least some of the adverse effects of transfusion that occasionally occur. Since RMP are released steadily with time and accumulate, it is expected that associated adverse effects would also increase with time. In this section, we first briefly review reports of increased risk of older blood, then more recent reports doubting those findings, and finally draw some tentative conclusions.

In 2006, the first of a series of reports appeared which challenged the safety of blood aged up to the allowed 42 days. These reports, chiefly by Koch and colleagues, presented evidence of significantly increased mortality and adverse events associated with blood aged >14 days, and suggested that such risks increase continuously with time of storage^2–4.

More recently, major questions about the validity of the findings of Koch et al have appeared. It was pointed out that patients in the study of Koch et al who received older blood (>14 days) also most frequently received 6 or more units, which is known to be independently associated with increased mortality¹⁴. After adjustment for this and other confounders, the significance of the findings of Koch et al almost disappeared¹⁴.

Edgren et al⁵, in the largest and most recent study yet published, analyzed data on 404,959 transfusions between 1995–2002. Contrary to other reports, they found no significant relation at all between age of blood and 7-day mortality. A “tendency” (5% increase in mortality rate) was noted for blood aged 30–42 days. They conclude that any excess mortality associated with blood age is <5%, which they point out is much less than found in previous smaller studies, suggesting that confounding factors have distorted results in previous studies.

Also in 2010, Van Stratten et al¹⁵ analyzed 9 years of transfusion data for coronary bypass at a large hospital, classified in 3 groups, (A) n=1422 who received blood age 0–14 days, (B) n=1719 received blood stored >14 days, (C) n=2175 received mixed but at least 1U >14 days. All were followed for average 170 days. They found no significant association of age of RBC transfused and mortality, either early or later. Similarly, Robinson et al¹⁶ analyzed data on 32,580 patients who underwent percutaneous coronary intervention (PCI), of whom 909 received blood, of mean age 25 +/− 10 days. They then divided the 909 into those who received blood <25 days old (n = 352) vs. >25 days (n = 360). They detected no significant difference in 30-day mortality between these groups. In other findings, they confirmed that mortality was associated with volume of blood transfused; and that any transfusion is associated with higher mortality compared to no transfusion. Elkelboom et al⁶ analyzed data on about 7,000 patients receiving a total of 21,400 units of blood of median age 17 days. They divided recipients into quartiles by blood age (0–13, 14–17, 18–22, 23–42 days) and detected “a modest independent association” between duration of storage of RBC and in-hospital mortality.

In view of these large retrospective studies, it now appears that the reports of Koch et al and related studies greatly exaggerated the risk associated with older blood, at least with respect to mortality statistics. Adverse events (other than mortality) are more difficult quantitatively evaluate by retrospective analysis. This issue needs to be addressed by prospective controlled studies with well-defined patient groups. For example, Gauvin et al¹⁷ showed that stable critically ill children who receive RBC units with storage times longer than 2 to 3 weeks may be at greater risk of developing new or progressive multiple organ dysfunction syndrome (MODS) in a prospective randomized controlled trial.

Risk of older blood limited to specific circumstances?

Although the recent reports listed above found little added risk of older blood, reports of increased risk of older blood continue to appear. If the more recent reports of adverse effects of older blood are free of confounding factors, then it is possible that specific patient groups, such as those with traumatic bleeding requiring massive transfusions¹⁸, or critically ill children¹⁹, are more sensitive to adverse reactions to factors in aged blood. The US Army has reported significant problems with blood of age >30 days²⁰. Tsai et al found that adverse effects of time of blood storage was limited to the microvasculature, and would not necessarily be evident at the systemic level²¹.

It is clear that the hazards of aged blood were overstated by Koch et al⁴ but some recent prospective controlled studies suggest that at least some degree of risk increases with time. The extent of this risk may depend on particular patient populations or the type of situation requiring transfusion, e.g. trauma.

MP generation during blood storage

Studies of RMP in blood bank products

In the late 1970’s and early 1980’s, a number of studies of MP released from blood stored for transfusion appeared, particularly on MP from RBC, here termed RMP. It was reported that the release of RMP from stored blood over time was related to the shape transition from discoid to echinocytic to spherocytic, supported by scanning EM micrographs; and both paralleled sensitivity of the RBC to phospholipase C-induced shape change^{7, 8}. The rate of these changes was almost zero for the first 10 days, followed by a steep rise from days 10 to 15, then little change until a further steady rise at 5–8 weeks. These effects were largely blocked or reversed up to 30 days by a “rejuvenating” additive consisting of glucose, pyruvate, inosine, adenine, and phosphate. In addition, RMP are known to exist in two size classes, which differ in physical and chemical properties^{22, 23}. The “large” ones (MP, 0.1–1.1 um) are those normally measured in clinical studies while the “small” ones (nanoparticles, 60–80 nm) are more difficult to detect in flow cytometry and to isolate by centrifuge. Both size classes are produced by erythrocytes. The compositional difference of these two types of particles will be discussed in a later section.

Multiple species of MP in stored PC

The presence of other MP species such as platelet MP (PMP) or leukocyte MP (LMP) in stored RBC has been mostly ignored. A recent study showed that multiple species of MP were generated in stored non-leukoreduced PC²⁴. Although RMP are the predominant species, significant amount of PMP and LMP are also generated during blood storage. The time-course of generation of the MP subtypes varied considerably. For RMP, there was little increase up to day 10, but thereafter rose steadily with time even after day 42. For PMP, counts rose steadily from day 0 and peaked at day 20. For LMP, there was no significant change in the first 20 days but continuous increases after day 30. The degree of neutrophil activation correlated well with PMP levels and the thrombin generation correlated with PMP and RMP. These data suggest increasing proinflammatory and procoagulant potential with time of storage.

However, Sugawara et al²⁵ showed that pre-storage leukofiltration of whole blood significantly decreased post-storage PMP count. We have confirmed that finding, and in addition, found that leukoreduction also substantially reduced RMP generation (unpublished). Thus, leukoreduction not only removes immunogenic leukocytes but also reduces MP generation during storage. This effect may contribute additional benefits of leukoreduction for reducing risks of transfusion.

Composition of RMP

Cole et al demonstrated that for most of blood groups (A, B, H, P₁), the antigens are steadily released for up to 6–8 weeks in association with membrane vesicles (MP), since blood group reactivity could be sedimented at 100,000 ×g²⁶. Kriebardis et al performed a detailed analysis of the components of RMP over time in stored PC and compared them with ghosts of whole RBC²⁷. That study included electron microscopy (EM) of colloidal gold stained RMP, which were 0.1–0.2 um. RMP constituents testing positive by blotting were stomatin, synexin, flotillins 1 and 2, sorcin, band 3, aquaporin, CD47, caspase and procaspase 3 and 8, Fas, FADD and abundant IgG. Nearly all increased steadily with time of storage, while declining in the whole RBC ghosts. Several were lipid raft proteins, and the proteins involved with apoptosis tend to confirm that RBC undergo a kind of apoptosis. There was notable evidence of oxidized products increasing to 5-fold in proportion to that of the RBC ghosts by day 15. They concluded that RMP release functions mainly to dispose of harmful agents from senescent cells. Earlier study by Muller and Lutz demonstrated preferential binding of autologous IgG to RMP²⁸. The view that RMP function mainly to dispose of damaged or harmful agents has been suggested by others, recently by Willekens et al²⁹, but it is unlikely that is the only or main function of MP shedding.

Salzer et al³⁰ studied RMP from stored blood at intervals to 50 days and compared their properties to the corresponding RBC, and to RMP produced in vitro by calcium ionophore. RMP from stored blood were similar to those made in vitro, with notable exception of some proteins. Their work is novel in employing atomic force microscopy (AFM) to obtain size histograms of RMP, which ranged from 50–200 nm; but conditions of these measurements may alter the particles compared to flow cytometry. They also measured thrombin generation and phosphotidylserine (PS) exposure. They performed density fractionations in which they assessed by blotting acetylcholinesterase (AChE), band 3, stomatin, CD55 (a.k.a. DAF), flotillin-2 and Duffy antigens. They emphasized new insights on the process of storage-induced RBC vesiculation, i.e.. that it is “raft-based” and “stomatin-specific”.

Bosman et al³¹ performed a semi-quantitative proteomic analysis of RMP and the parent RBC membranes. Although a large number of identified peptide fragments could not be assigned to individual proteins (data online lists about 20,000 fragments), the authors identified a total of 257 different proteins. For example, semaphorin 7A and peroxiredoxins decreased in the parent RBC with age of storage but simultaneously increased in the microparticles (RMP) and nanoparticles (RNP). Many complement components were identified on the RMP, including fragments of C1q, C1r, C1s, C3, C4 and C9. Of interest, a comparison of RMP and RNP showed, unexpectedly, that RNP were 100-fold enriched in complement proteins compared to the RBC membrane, and 10-fold enriched compared to RMP. Immunoglobulins were also greatly enriched in the RNP. When taken together with the proteomic study of whole RBC (membrane and cytosolic components) by Pasini et al³², this study is highly suggestive that RMP have proinflammatory potential owing to the presence of complement and immunoglobulins. Several proteins involved in coagulation such as phospholipid scramblase 1, plasminogen precursor, fibrinogen beta chain precursor, and beta-2-glycoprotein 1 were also detected on RMP.

MP-mediated proinflammatory activities

RMP as mediators of the complement (C) system

The complement (C) system is a major mediator of inflammation via many pathways. Erythrocytes have complement receptors and play vital roles in innate immunity and inflammation via the C system. For example, dysregulation of C in relation to RBC was recently linked to venous insufficiency of the lower extremeties, involving CD35 (CR1)³³. A broad consensus indicates that C is the underlying cause of many adverse effects of blood transfusion including hemolytic reaction^34–36, anaphylactic reaction^{37, 38}, and TRALI³⁹.

It has been shown that C receptor 1 (CR1, a.k.a. CD35) is enriched on RMP from ATP-depleted RBC, which was proposed to explain the progressive loss of CR1 on RBC with storage⁴⁰. CR1 is a central player in RBC immune function, chiefly by elimination of immune complex^{41, 42}. The C5b-9 membrane attack complex (MAC) is selectively shed on MP from platelets⁴³, endothelial cells⁴⁴ and other cells as a defensive mechanism, and is probably true also for RBC⁴⁵. Several authors have proposed that a major function of MP shedding is to dispose of harmful agents from cells, such as are likely to be opsonized by C^{29, 46}. The previously mentioned proteomic studies of RMP showed great enrichment in complement and IgG³¹. These observations suggest that RMP in stored blood may be involved in some types of C-mediated adverse effects of transfusion.

Interactions of MP with leukocytes

MP from platelets (PMP) and from endothelial cells (EMP) were shown to adhere to and activate leukocytes ^{47, 48}, and many authors have since confirmed and extended those findings. Comparatively little research has been done on possibly similar interactions of RMP, probably because RMP do not possess selectins (C62P, CD62E, CD62L) and therefore cannot interact via those adhesins. However, other mechanisms for interaction can occur. Gasser and Schifferli⁴⁹ demonstrated that leukocyte MP (LMP) bind to intact RBC in a C-dependent fashion; i.e., binding of C3 followed by activation of the classical pathway on LMP yielded C3 fragments, resulting in capture by RBC via CR1 (similar to capture of immune complex). This binding may well act in reverse, i.e., C fixation occurring on RMP followed by binding to leukocytes. In addition, expression of PS on any MP can lead to binding to leukocytes since they have a specific PS receptor⁵⁰.

MP as a potential mediator for TRALI

Transfusion-related acute lung injury (TRALI) is among the most serious of transfusion-related adverse events with high morbidity and mortality. The underlying mechanism has been unclear, but recent work persuasively demonstrates a 2-hit scenario. Using a sheep model, Tung et al have shown that if the animal is first sensitized by inducing inflammation via lipopolysaccharide (LPS), which primes neutrophils, then administration of either the supernatant of 5-day aged platelets or the supernatant of 42-day aged RBC results in neutrophil sequestration in the lungs and full-blown TRALI in 80–90% of animals^{51, 52}. These findings are consistent with current consensus that excessive neutrophil activation plays a pivotal role in the pathogenesis of TRALI⁵³. Although the identity of the substance in the supernatants responsible for this outcome has not been established, we postulate that PMP and/or RMP are likely to be the main mediators for TRALI, by the following evidence. First, it has been shown that PMP can bind and activate neutrophils via P-selectin – PSGL-1 interaction ^{47, 54}. Second, complement and IgG are enriched in RMP from storage lesion³¹. The RMP-bound IgG and complement can activate neutrophils via Fc receptors of neutrophils. Third, it has been reported that CD40L released from stored platelet concentrates is also a potential mediator for TRALI⁵⁵. Studies have demonstrated that the majority of CD40L in blood is actually MP-bound^56–58. Our group showed that much of CD40L could be removed by 0.1 um filtration⁵⁷.

MP-mediated thrombotic complications

Because microparticles have been implicated in thrombosis by work from our lab^59–61 and others^62–65, and exhibit potent procoagulant activity^{59, 62}, they may contribute significantly to post-transfusion thrombosis. The most important procoagulant property of MP is their expression of the anionic phospholipid, phosphatidyl serine (PS), which serves for assembly of the coagulation factors into active complexes for thrombin generation. We developed an assay for MP-mediated thrombin generation⁶⁶, and compared in vitro the relative profiles of PMP, EMP and RMP⁶⁷. Marked differences were observed. EMP exhibited short lag time (2–3 min) but low thrombin peak amplitude. In contrast, RMP show a large thrombin peak but very long lag time (>15 min). PMP exhibited intermediate lag time (8–10 min) and highest thrombin peak. The lag time is inversely correlated with tissue factor (TF) expression, and thrombin peak seems to be proportional to PS expression.

The thrombogenic potential of PMP is well-recognized but some evidence suggests also involvement of RMP. Binder’s group measured formation of thrombin-antithrombin (TAT) to measure prothrombotic activity in mice, and observed that RMP injected in normal mice had no effect on TAT, but if injected into mice fed a high-fat diet, TAT increased markedly⁶⁸. This observation is consistent with findings concerning TRALI, in the sense that MP-mediated adverse effects of transfusion may occur only in specific patient groups, i.e., pre-existing inflammation for TRALI, high cholesterol for thrombosis due to RMP. Therefore, the greater risk of aged blood may apply only to sensitive or sicker patient groups.

Since we have observed significant amounts of PMP in packed cells (PC), but not in leuko-reduced PC (unpublished), we postulate that PMP may work synergistically with RMP in mediating some transfusion-related thrombotic complications. In a recent study, Spinella et al reported a higher incidence (34%) of DVT in trauma patients receiving older blood (28 days or more) compared to blood <28 days old (16%); p<0.02¹⁸. This observation is consistent with the hypothesis that cell-derived MP may contribute to the DVT since it is known that MP increase with time of storage. They also observed increased in hospital mortality with the older blood (16% vs. 7% for younger blood).

Post-operative cognitive impairment (POCI) is a well recognized complication^69–71. To our knowledge, it is not established if this complication is exacerbated by transfusion, but there is evidence of MP involvement in ischemic brain disease⁷². We demonstrated correlation between PMP levels and ischemic brain disease ^73–75. The frequency of POCI was reduced in off-pump vs. on-pump CABG procedures⁷⁶, which is also consistent with possible involvement of MP, as it is known that exposure of platelets to artificial surfaces induces their activation and PMP release⁷⁷. However, the hypothesis that blood transfusions contribute to post-surgical thrombosis requires direct testing.

Testing the Hypothesis of MP Involvement in Transfusion Complications

The work reviewed suggests but does not prove that cell-derived MP may contribute to transfusion-related complications. In view of concerns about older blood, our laboratory will more directly test this hypothesis. The main strategy for the in vivo component of the study is to compare clinical outcomes and proinflammatory/procoagulant biomarkers in CABG patients who received either normal PC (n = 100–150 target) or washed PC (n = 100–150 target), using the rationale that washing will remove the accumulated MP from the storage lesion. A second component of the study is to investigate in vitro the rate of MP production and the MP properties in stored blood over time, and their composition in proteomic analysis, by methods designed to improve on previous studies.

A major challenge, which was anticipated, is the fact that significant transfusion complications are very rare. Our patient population is unlikely to be large enough to detect significant differences in adverse events between these two groups. To address this short coming, we are looking at laboratory biomarkers that could suggest more serious adverse effects in some patients. These are markers of inflammation, oxidation, procoagulant activity, and others. Even if it is not possible to draw clear conclusions about the role of MP in adverse effects of transfusion, a wealth of additional data is being compiled that is expected to elucidate important aspects of transfusion medicine.