Current methods of assessing platelet function: relevance to transfusion medicine

Abstract

There are three areas in transfusion medicine where the assessment of platelet function may be informative: To identify platelet dysfunction in donors of platelet components. To assess whether changing processing or storage methodologies has an impact on platelet function in platelet concentrates (PC). To determine which patients require platelet transfusions and whether these platelet transfusions are effective by assessing platelet function in the recipient. Traditionally, platelet function tests are time-consuming to perform, difficult to quality control, lack sensitivity and usually require specialist laboratory support. However, during the last 10 years, significant developments have occurred in this field. Tests that provide more information on global haemostasis, rather than a single facet of platelet function, are now available, and some tests can be applied in a near-patient manner, rather than by specialized laboratories, so that results can be used to tailor therapy. We do not intend this review to be a ‘shopping list’ of tests that blood services should apply to PC, but aim to review methods for assessing platelet function that we believe might be relevant to transfusion and whether they have been applied in this setting. Platelet enumeration is reviewed elsewhere [1]. Overviews of normal platelet function have been published previously [2, 3]. Platelets are characteristically small and discoid in shape with dimensions of ≈ 2·0–4·0 × 0·5 microns and a mean volume of 7–11 fl. They are the second most numerous corpuscle in the blood, normally circulating at levels of 150–400 × 109/l. Platelets are anucleated cells derived from bone marrow megakaryocytes that contain cytoplasmic organelles, cytoskeletal components and platelet-specific granules. They normally circulate in the bloodstream in a quiescent state for a maximum of 10 days, but are primed to undergo ‘explosive’ activation following damage to the vessel wall. This leads to the rapid formation of a vascular plug to occlude the site of damage. Platelets are therefore enriched in signalling proteins and surface receptors that enable them to efficiently respond to vessel wall injury. Their shape and small size allows them to be forced along the edge of the vessel wall, placing them in the optimum position of maximal shear rate to constantly monitor vascular integrity. The normal vascular endothelium produces potent platelet inhibitors, such as nitric oxide, prostacyclin and a natural ADPase (CD39). However, once vessel wall integrity is compromised and subendothelial components [e.g. collagen, von Willebrand factor (VWF), fibronectin and laminin] become exposed, platelets undergo a highly regulated set of functional responses, including adhesion, spreading, release reactions, aggregation, induction of procoagulant activity, microparticle formation and clot retraction (Fig. 1). Platelet-subendothelial and platelet–platelet interactions. Reproduced with permission from S. Watson and P. Harrison P: The vascular function of platelets; in Hoffbrand V, Tuddenham E, Catovsky D (eds): Postgraduate Haematology, edn 5. Oxford, Blackwell Publications, 2004:817 (in press). Under conditions of high shear, as found in the arterial circulation, the initial platelet–collagen interactions are exclusively mediated via VWF present in plasma bridging between collagen and the glycoprotein (Gp)Ib/IX/V complex on the platelet surface. This facilitates initial adhesion or tethering, enabling platelets to slow down and roll along the damaged area of the vessel wall. Subsequent activation and firm adhesion is then mediated by other receptor–ligand interactions, e.g. direct binding of collagen to its platelet receptors GpIa/IIa and GpVI. However, at lower rates of shear found in the venous circulation, initial adhesion can also occur directly to subendothelial matrix proteins such as collagen and fibrinogen. At all shear conditions, adhesion is strengthened considerably through activation of platelet surface integrins, e.g. GpIa/IIa and GpIIb/IIIa, leading to an increase in affinity for their ligands and static adhesion. Adhesion also results in platelet activation via various signal transduction pathways and the elevation of cytoplasmic calcium levels. This causes cytoskeletal changes that mediate shape change, pseudopod formation and platelet spreading. Simultaneous release of granule components (e.g. serotonin, ADP, VWF) and thromboxane results in further recruitment, activation and aggregation of other platelets near to the growing haemostatic plug. Aggregation is mediated either by fibrinogen and/or VWF (under high shear) bridging between activated GpIIb/IIIa receptors on adjacent platelets. Platelet activation also results in the translocation of internal negatively charged phospholipids onto the outer platelet membranes, which provides binding sites for several coagulation complexes. The resultant burst of thrombin generation activates and recruits more platelets into the vicinity, but also converts soluble fibrinogen into insoluble fibrin, which is cross-linked by factor XIII, to stabilize the otherwise fragile haemostatic plug. When there is a defect in any one or more of these functions and/or in platelet number, haemostasis often becomes impaired and there is an increased risk of bleeding. The relationship between platelet count and risk of bleeding is complex. The ‘threshold’ platelet count at which platelet transfusion is required depends on the clinical status of the patient [4]. PCs are transfused to treat or prevent haemorrhage in patients with thrombocytopenia or platelet defects. These two scenarios may place different demands upon the quality and functionality of PC. For prophylaxis, platelets must be able to circulate and function as normally as possible and for as long as possible. But in patients who are haemorrhagic, an immediate haemostatic effect is desired and the potential circulating half-life might be less important. In transfusion the main application of in vitro tests has been to determine platelet function in PC. In PC, platelets are removed from the physiological environment in which they normally circulate and function: the majority of red cells and white cells are removed and, more importantly, there is no endothelium to maintain platelet quiescence. The concentration of PC (1000–1500 × 109/l) also far exceeds that of normal whole blood (150–400 × 109/l), on which many of these tests have been optimized. It may also be important to dilute control and test PC in the same media, as this is known to influence results when PC are stored in platelet additive solutions [5]. Finally, during platelet storage the ageing of the PC supernatant may also have an impact on a variety of plasma proteins and factors that are important for normal platelet homeostasis. The major disadvantage of in vitro tests of platelet function is that we do not know how the results in PC translate to platelet function in vivo following transfusion. For some tests there may be a relationship with recovery [6-10], but this has been shown mainly with PC stored in plasma; the same may not be true for PC stored in additive solutions. For many of the tests listed below, studies correlating results in vitro with platelet recovery do not exist, and indeed some have not yet been applied to PC at all. Moreover, no test has been adequately shown to correlate with the haemostatic function of platelets following transfusion in humans, partly because gold standards for assessing the clinical function of PC do not exist. The clinical utility of many of these tests (summarized in Table 1) is therefore unproven for use with PC. Nonetheless, in vitro data does provide a level of reassurance prior to the clinical use of the component, and severe lack of functionality in vitro would certainly be a major cause for concern and merit further clinical studies prior to routine production. We also need to differentiate between tests that are useful in validation studies and those that are beneficial as an ongoing check of PC quality; the majority below belong to the former category. The extent of validation is generally based upon the magnitude of the processing change and its probable impact. Small changes might require in vitro data only, whereas major changes would usually require clinical studies. The hierarchy of studies performed is usually: (1) in vitro tests; (2) recovery and survival in normal volunteers; (3) count increment studies in thrombocytopenic patients; and (4) an assessment of haemostasis (e.g. quantifying bleeding) in thrombocytopenic patients. On activation, platelets undergo a shape change, become spherical and start to project pseudopodia. When exposed to a light source, discoid platelets reflect the light and have a ‘shimmering’ or ‘swirling’ effect that disappears when platelets lose their discoid shape. Swirling is more reproducible than morphology scores [11], but the technique lacks sensitivity because assessment of a grade other than ‘yes’ or ‘no’ is subjective. PC that fail to swirl have pH values below 6·4 or above 7·6 in most, if not all, cases [12]. Thus, measurement of pH may provide similar information, but swirling is non-invasive and can be performed on all PC as a preissue check, as currently carried out in some countries. Deviation from normal discoid platelet morphology can be assessed by oil-phase microscopy and a scoring system [13]; however, the grading may not be reproducible between observers and it is a very time-consuming procedure. The extent of shape change (ESC) test uses an aggregometer to measure the shape change induced by ADP in the presence of EDTA to prevent platelets aggregating. ESC correlates well with morphology scores [14]. Although transformation from a disc to a sphere may be associated with an increase in the mean platelet volume (MPV), this in itself is not a reliable indicator, as microparticles released from activated platelets may cause a decrease in MPV [15]. There is a strong relationship (r = 0·71) between ESC and recovery in vivo, with a recovery of < 50% corresponding to ESC levels of < 60% of day-1 values [6]. However, in an animal model, changes in platelet morphology showed no correlation with loss of haemostatic function in vivo[16]. Aggregation is usually assessed by stirring platelet-rich plasma at 37 °C in a cuvette between a light source and a detector. Aggregation is initiated by using agonists such as collagen, thrombin, ADP, ristocetin, arachidonic acid, or epinephrine (either alone or in combination). As platelets start to aggregate there is an increase in light transmission, and results are recorded as the rate of aggregation and maximal response. By comparing responses to different agonists, platelet defects can be identified and differentiated from each other, and aggregation studies therefore remain an important diagnostic tool. Platelet aggregation in whole blood can also be measured by using techniques based on electrical impedance and by measuring the decrease in free platelets that occurs in response to agonists. The latter can be measured using flow cytometry [17] or by a purpose-designed instrument (Ichor Full Blood Counter; Helena Biosciences, Beaumont, Texas, USA). A disadvantage of aggregation is that it tests platelets under low shear and in response to single agonists – conditions that are not reflective of primary haemostasis in vivo. Platelet aggregation in PC does not appear to predict platelet recovery: PC stored at 4 °C maintain better aggregation responses than those stored at 22 °C [18], despite exhibiting lower recovery and survival. Numerous studies have shown declining aggregatory responses to single agonists during PC storage. However, this effect is reduced when pairs of agonists are used, and the poor ability of stored platelets to respond to ADP or collagen may recover following transfusion [19-21]. Furthermore, the aggregatory response of platelets stored for 5 days is increased if diluted in fresh-frozen plasma rather than in plasma from the PC [22], which may, in part, be caused by the correction of low pH. The platelet function analyser (PFA) was originally developed by Kratzer & Born to measure platelet adhesion, activation and aggregation under conditions of high shear [23]. The prototype instrument (Thrombostat-4000) has been superseded by the commercially available PFA-100 (Dade-Behring, Marburg, Germany). A citrated whole-blood sample is applied to a disposable cartridge containing a membrane coated with either collagen/epinephrine or collagen/ADP and which has a microscopic (147 µm) aperture cut into it. Under high shear rates (5000–6000 s−1), contact of blood with the membrane causes platelets to aggregate and occlude the aperture. The time taken to occlude the aperture or closure time (CT) is the end-point that is recorded in an analogous manner to the bleeding time. The maximal CT that can be recorded is 300 s. The technique is simple and rapid, but sensitive to a number of variables including platelet number, haematocrit, ABO blood group and plasma VWF levels. As a normal result has a high negative predictive value, it is beginning to become more popular as a screening test for platelet dysfunction, although false positive and negative results may occur. The test is more sensitive than the bleeding time in von Willebrands disease and various other platelet defects and is also sensitive to various anti-platelet drugs [24]. The PFA-100 has been used to show that a significant percentage (16–36%) of apheresis donors or whole-blood donors have platelet dysfunction at the time of donation [25-27]. This appears to be partly because of an undeclared intake of aspirin or other cyclooxygenase inhibitors, as well as dietary factors, and is largely transient in the same donors [27]. Moreover, in apheresis donors, decreasing the donation interval appears to decrease platelet function, as assessed by the PFA-100 [25]. These studies suggest that current policies for deferring donors who have consumed aspirin do not prevent this, and that techniques such as the PFA might be a useful predonation check, particularly for single-donor products. However, it is not yet apparent whether PC produced from donors with abnormal PFA-100 results are clinically less efficacious. In a small number of patients, transfusion of PC from donors with prolonged closure times did not correlate with increased blood loss [25] and although platelets from aspirinated donors initially give prolonged bleeding times in vivo, the effect was only transient and bleeding times were eventually corrected [28]. Furthermore, providing that 10–20% of circulating platelets have not been aspirinated, PC from donors who have taken aspirin appear to be haemostatically active [29]. These studies are particularly interesting as relatively little attention has been paid to the contribution of donor-related variables to platelet function in PC. Because the PFA-100 is designed for use with whole blood and high shear conditions, samples from PC cannot be assessed directly. This can be overcome by reconstituting PC samples into whole blood by adding leucocyte-depleted (and hence platelet-depleted) whole blood or red cells and plasma. This has been used to show that closure times increase during platelet storage, suggesting a loss of platelet function [30-32]. However, closure times can become prolonged beyond 300 s towards the end of platelet storage, although these can be shortened by adding additional agonists to the membrane of the cartridge [33]. Furthermore, manipulation of blood samples and PC increases the probability of premature closure times when the instrument detects a rapid blockage caused by microaggregates occluding the aperture too rapidly compared with normal platelet adhesion/aggregation. The original annular perfusion chamber was described by Baumgartner [34], and uses a blood vessel, usually rabbit aorta, mechanically or chemically denuded of endothelial cells. Anticoagulated blood is recirculated through the chamber containing the vessel. This was developed into a system that does not require the use of blood vessels. The parallel plate chamber consists of two parallel plates, one of which accepts a glass coverslip coated with purified proteins (e.g. fibrinogen), extracellular matrix, or cultured endothelial cells. Results are determined after a set period of perfusion by grading single platelets, platelets forming mono or bilayers, or larger aggregates (thrombi). Such analysis normally requires experience; however, video microscopy, confocal microscopy and computer analysis are increasingly being used to speed up this process. Most systems allow rheological properties to be altered by varying the distance between the plates, the flow rate, or the viscosity of the perfusate. The main advantage of perfusion chambers is that platelet deposition and thrombus formation can be examined real-time under conditions of well-defined flow [35]. These techniques are not used in clinical diagnosis, but have found application in the detailed analysis of platelet function and the action of various anti-platelet drugs. Some studies show that adhesion to collagen, fibrinogen, extracellular matrix or denuded blood vessels is minimally affected by storage of PC, whereas others show a decrease in platelet deposition of up to 50% by day 8 [36-41]. Two tests that have been used to detect inherited or acquired platelet defects and anti-platelet therapy are the cone and plate analyser (CPA), and the Ultegra Rapid Platelet Function Analyser (now called the Verify-Now; Accumetrics, Sandiego, California, USA), but these have not been applied to PC. The Verify-Now instrument measures platelet aggregation based on optical light transmittance [42], whereas the CPA measures the adhesion of platelets to extracellular matrix or plastic using whole blood exposed to high shear (1800 s−1) [43], which is generated by a rotating cone in the sample cup. The CPA has now been developed as an automated commercial instrument (IMPACT; Diamed, Cressier, Sor Morat, Switzerland). Flow cytometry permits the rapid analysis of large numbers of platelets within relatively small quantities of sample. There are a wide number of commercial anti-platelet antibodies and dyes/reagents that can be used to study platelets by flow cytometry. Examples of their application are shown in Table 2. Flow cytometry can be used to assess changes in platelet properties that occur during storage of PC, including platelet glycoprotein expression (e.g. GpIb, GpIIb/IIIa), the generation of platelet activation markers (CD62P, CD63, CD40L, platelet–leucocyte complexes), the exposure of negatively charged phospholipids, and the formation of platelet-derived microparticles. Most studies examining platelet glycoproteins that are important in platelet adhesion and aggregation in PC during storage have focused on GpIb and GpIIb/IIIa. When platelets become activated, GpIIb/IIIa undergoes a conformational change allowing it to bind fibrinogen and VWF. Antibodies are now available that recognize epitopes on GpIIb/IIIa which become exposed during this process rather than in the resting state. Alternatively, fibrinogen bound to the platelet surface, or the ability of fibrinogen to bind to receptors in response to agonists, can be measured. The latter may be more informative: expression of GpIIb/IIIa does not tend to change during the storage of PC, whereas there is a reduction in fibrinogen binding to platelets in response to thrombin [44]. The N-terminal portion of GpIbα (glycocalicin) can be cleaved by various proteases, resulting in a reduced ability to bind VWF and thrombin. There is a strong negative correlation (r= −0·91) between the percentage of platelets negative for GpIb and in vivo viability [45]. Recently the importance of GpIb in maintaining platelet viability has clearly been demonstrated in animal models. When platelets are stored in the cold, there is clustering of GpIb on the platelet surface, which results in recognition by complement receptors on liver macrophages and rapid removal from the circulation [46]. The expression of CD62P (P-selectin) on the surface of platelets is frequently used as a marker of platelet ‘activation’, yet its clinical relevance is uncertain. CD62P is contained in platelet alpha granules, but is rapidly translocated to the surface when platelets become activated. It has been postulated that increased expression of CD62P may be responsible for an increased rate of clearance of platelets from the circulation [47-49]. However, in animal models, platelets treated with sufficient thrombin to cause secretion of granule contents have a survival time in the circulation that is not significantly different to that of unstimulated platelets [50, 51]. Furthermore, thrombin-activated platelets are not only able to circulate, but appear to be functional once transfused, with generation of soluble-P-selectin and platelet–leucocyte complexes [50]. Within a mouse model, CD62P expression has been shown to be unimportant in platelet clearance [52]. Recent data from a thrombocytopenic rabbit model suggests that changes in CD62P expression may be involved in the rapid (< 1 h) removal of platelets from the circulation, whereas changes in GpIb expression are involved in longer-term-mediated mechanisms of platelet clearance [53]. An advantage of flow cytometry is that it provides information on the percentage of cells expressing a particular protein and the extent to which it is expressed. Yet, the latter is rarely reported. A drawback to only measuring cell-surface expression is that proteins can be cleaved from the surface by the action of various proteases, and this will not be measured. For example, both CD40L and CD62P can be shed from the surface of the platelet and it has been suggested that the measurement of soluble CD62P in the supernatant may be a better marker of platelet activation [54]. Moreover, platelets are generally assessed in their resting state. The expression of CD62P and GpIIb/IIIa (as well as fibrinogen bound to it) increases in response to physiological agonists, but this is impaired during platelet storage [41, 55]. Interestingly, Rinder also showed that if this is induced by metabolic stress it can be stabilized, or even reversed, by the addition of plasma and appropriate storage conditions. The responsiveness of platelets to stimuli, rather than spontaneous glycoprotein expression, might be a better predictor of in vivo efficacy, but this remains to be established. The exposure of anionic phospholipids on the platelet surface, and the generation of microparticles following activation, results in a 300 000-fold increase in the rate at which factor Xa can convert prothrombin to thrombin [56] and provides a surface on which factor XI can be activated by trace amounts of thrombin [57]. Exposure of phosphatidylserine on the platelet surface can be measured by the binding of Annexin V using flow cytometry. In the past, platelet PCA has been assessed using modifications of the activated-partial thromboplastin time, where platelets, rather than exogenous phospholipid, provide cofactor activity. Thrombin generation and inhibition in plasma, platelet-rich plasma or whole blood can now be measured using chromogenic or fluorogenic substrates [58]. Alternatively, purified coagulation factors can be added to washed platelets and thrombin generation assessed by using a suitable chromogenic substrate. Data on platelet PCA during storage of PC is inconsistent; some studies show an increase and others a decrease, probably because of the differing methods used [41, 59-61]. Platelets contain adenine nucleotides that are essential energy sources for platelet function. Approximately 60% of ADP and ATP are stored within platelet-dense bodies (‘storage pool’). The remaining ‘metabolic pool’ is present in mitochondria, bound to actin, or free in the cytoplasm. There are two major metabolic pathways in platelets: glycolysis; and oxidative metabolism via the tricarboxylic acid pathway. Preferred substrates for oxidative metabolism are fatty acids and acetate, rather than glucose. The majority of ATP production in platelets is derived from oxidative metabolism, which produces CO2 as an end product that can diffuse out of the storage bag. However, the end product of glycolysis is lactate and hydrogen ions that cannot diffuse out. The latter is buffered in plasma by bicarbonate, which is exhausted when lactate levels reach ≈ 20 mm, after which point there is a rapid decrease in pH. When PC are stored in plasma, pH values below 6·0–6·2 are associated with poor in vivo recovery [29, 62, 63]. For this reason, pH is used as a routine quality-monitoring test by most blood services. The relationship between high pH and viability is less clear. Some studies have shown a loss of recovery in vivo at pH values of > 7·2–7·7 [64, 65], whereas others [48] showed that PC with a pH of up to 7·6 do not show a loss of in vitro function or poor in vivo recovery. Another marker of metabolic activity is the hypotonic shock response (HSR), which is based on the ability of platelets to extrude water and electrolytes after the rapid swelling that occurs when placed in a hypotonic environment, and can be measured by using an aggregometer. There is a reasonable correlation between HSR (r = 0·57) or ATP levels and in vivo viability [6]. An in vivo recovery of < 50% appears to correspond to an HSR or an ATP level of < 70–75% of fresh platelets. There is a good correlation (r = 0·43) between lactate production per platelet and survival [6]. Apoptosis, or programmed cell death, has been well described within nucleated cells and has been recently suggested to be an important mechanism during platelet formation from megakaryocytes [66]. Indeed, platelets may represent a specialized form of apoptotic bodies that are not immediately detected and removed by macrophages until further ageing in the circulation occurs or by in vitro cold-induced clustering of GpIb. The exact signal(s) for removal of senescent platelets from the circulation remain poorly defined, but recently apoptotic mechanisms have also been described within platelets. Functional platelets normally retain both inner mitochondrial membrane potential and a classical phospholipid asymmetry. However, upon in vitro storage of platelets, a number of apoptotic changes have been described, including platelet shrinkage, cytoplasmic condensation, plasma membrane blebbing or microparticle generation, extension of filipodia, depolarization of mitochondrial membrane potential and PS exposure [67]. Many of these important changes can be monitored by flow cytometry (Table 2). Apoptosis has been described in stored PCs, but there are conflicting data to the exact mechanism(s) involved and the time at which it is initiated; in most studies this occurs late in, or beyond, the current shelf-life of platelets [68-70]. Research in this area is very preliminary, but may help to define important aspects of the platelet storage lesion and platelet senescence. These tests do not measure platelet function in isolation, but rather in the context of clot formation and retraction. Consequently, they are also dependent on plasma coagulation and fibrinolytic systems. Either PRP or whole blood can be studied. The Haemostasis Analysis system (Hemodyne Inc., Richmond, Virginia) has been developed as a commercial instrument based on a method described originally by Carr [71] and measures three variables: platelet contractile force (PCF), clot elastic modulus (CEM) and thrombin generation time (TGT). A small (700 µl) sample of whole blood is trapped between two parallel surfaces and clotting is initiated by various agents. As the clot forms, platelets begin to pull within the network and a downward force is transmitted to the upper plate, movement of which is measured by a transducer. The distance moved can be converted into PCF by means of a calibration constant, the CEM. The PCF does not alter over 5 days of platelet storage, and is virtually absent in platelets stored at 4 °C [72, 73]. Treatment of PC with inhibitors to prevent rearrangement of the cytoskeleton increases the PCF of 4 °C PC to 30% of that of fresh PC, yet has little effect on blood loss, as demonstrated in a rabbit kidney injury model [16]. This suggests that this variable in vitro does not correspond to haemostasis following transfusion. The TEG was developed more than 50 years ago (for recent reviews see ref. 73). Whole blood is incubated in a heated sample cup in which a pin is suspended that is connected to a chart recorder or computer. The cup oscillates 5° in each direction. In flowing blood the pin is unaffected, but as the blood clots, the motion of the cup is transmitted to the pin. Native whole blood or recalcified plasma can be used, with or without activators of the tissue factor or contact factor pathways. The TEG is rapid to perform (< 30 min), can be conducted in a near-patient manner and it provides various data relating to clot formation and lysis: the lag time before the clot starts to form, the rate at which clotting occurs; the maximal amplitude of the trace; and the extent and rate of amplitude reduction. Rotational TEG (RoTEG) is an adaptation of the TEG where the cup is stationary and the pin oscillates. TEG and RoTEG show that clot strength is minimally effected after storage of PC for 5 days [72, 73]. Measurement of platelet recovery and survival following autologous transfusion to normal volunteers is considered by many in the field to be the best assessment of PC quality and is usually performed by radiolabelling an aliquot of PC and reinfusion to the donor. By taking samples 10 min after infusion, and daily thereafter, recovery and survival can be estimated. There are a number of drawbacks to performing such studies, as follows: first, the radiolabelling process is technically demanding, and care must be taken to avoid red cell contamination (which can also become labelled) of the platelet sample; second, it is difficult to perform on buffy coat-derived PC, as they are produced from more than one donor; and, third, some countries are not permitted to carry out either radiolabelling or biotinylation studies. Previously, no recommendation has been made as to what constitutes acceptable values for platelet recovery and survival following transfusion, but a new standard has been proposed which states that the mean recovery should be > 66% of fresh platelets and survival > 50% of fresh platelets [75]. However, the exact requirements of the Food and Drug Administration (FDA) and other regulatory authorities are yet to be published. Recovery and survival can also be assessed in thrombocytopenic patients but, more commonly, count increment studies are performed in this setting. With either approach, the data provides little information on the functionality of the cells, although it is clearly a prerequisite for platelets to be able to circulate in order to function. At platelet counts of < 100 × 109/l, there is an inverse relationship between platelet count and the bleeding time in some patient groups [28] and in animal models [76]. Until recently, bleeding time was considered to be the best screening test for platelet function, but is now considered redundant as it is insensitive, invasive, poorly reproducible between operators and not ideal for repetitive tests within the same individual. Quantifying bleeding as an outcome measure of platelet transfusion is attractive because it is clinically meaningful, but there are a number of difficulties in achieving this [77, 78]. Key variables are the site and severity of bleeding, and whether data are collected in a prospective or retrospective manner. Variability in these factors can contribute to significant differences between studies. Although most studies use the World Health Organization (WHO) grade 1–4 bleeding scale, or an adaptation of it, there is a lack of a standardized grading system and agreement over which grades are the most relevant end-points. Furthermore, large numbers of patients may need to be studied, particularly at higher bleeding grades where events occur rarely. An assessment of bleeding has been used in studies examining platelet dose, the threshold ‘trigger’ for platelet transfusion, and more recently for photochemically treated platelets [79]. Several animal models have been developed to assess platelet function in PC, the most commonly used being the rabbit [76]. As human platelets transfused into rabbits are rapidly cleared by the reticuloendothelial system, this must first be inhibited. Animals are rendered thrombocytopenic by the use of anti-platelet antibodies or other agents. Recovery and survival can be assessed, even in animals that are not rendered thrombocytopenic, by flow cytometry using appropriate anti-human platelet antibodies. The main end-point and advantage of this technique is an assessment of haemostasis by observing cessation of bleeding from an induced injury. This is usually a cut in the ear, but other techniques have been used, such as kidney injury. Animal models have been particularly useful in assessing platelet substitutes, which are notoriously difficult to assess in vitro[80]. They have also been applied to extending the storage period of platelets [80], photochemical treatment of platelets [81] and storage of platelets, either cold or frozen [16, 46, 83]. Two studies have also used the Thrombostat-4000 or PFA-100 to evaluate the effectiveness of PC transfusion [83, 84]. In both studies, many patients have closure times that are too prolonged to measure before transfusion, presumably as a result of either thrombocytopenia and/or platelet dysfunction. However, in some patients there is a shortening of the closure time following transfusion of PC. This is higher in fresh (rather than stored) platelets [84] and appears to show a relationship to improvement in the control of bleeding [85]. Interestingly, the latter study also showed that some patients had normal PFA results before they were transfused, indicating that they may actually not have required a platelet transfusion. Unsurprisingly, both studies showed that the closure times do not correlate very well with platelet count increments. Indeed, some patients failed to produce a good count increment, but did show a shortening in closure times, suggesting that platelet function was improved. A further study has shown that there is a correlation between results following transfusion and those obtained if a pretransfusion sample from the patient is supplemented with the same PC that was transfused [86]. Platelet adhesion to subendothelium under flow is significantly worse in platelets stored at 4 °C compared with 22 °C, despite a shortening of the bleeding time following transfusion of either [87] and higher fibrin deposition with 4 °C PC. A number of significant questions remain in terms of assessing platelet function in transfusion medicine. Do we really know why we transfuse platelets and consequently what function we need them to perform? We have started to question whether prophylactic platelet transfusions are beneficial, or whether aggressive therapy in patients who are actively bleeding might be more appropriate [87]. Clinical studies in this area may challenge our concepts of what the functional characteristics of an optimal PC should be. If PC were only transfused in a therapeutic setting, some characteristics that are traditionally thought of as ‘bad’, such as platelet activation and microvesicle formation, might actually be highly desirable, and products which show poor recovery or survival, but are haemostatically active, may increase in interest (such as frozen platelets and platelet substitutes). What do we want in vitro tests to predict: recovery, survival or haemostatic responsiveness? Studies have quite clearly shown that tests which predict one might not predict the other; it is unreasonable to assume that one test will provide all the answers. Which defects in platelet function in PC are reversible and which are irreversible? The milieu in which we test PC may be critical to the results obtained and we may need to focus more attention in this area. What is the frequency and clinical relevance of platelet dysfunction in blood donors? Can we better predict which patients require platelet transfusions and assess whether they are effective by testing before or after surgery? Some of the newer tests of platelet function may help us to answer these questions, but not many have been applied in a transfusion setting and some are difficult to adapt to study PC. They will require careful validation against relevant end-points in animal and clinical studies.