Ex vivo evaluation of erythrocytosis-enhanced platelet thrombus formation using the cone and plate(let) analyzer: effect of platelet antagonists


Dr Ellinor I. B. Peerschke, New York Presbyterian Hospital, 525 East 68th Street, Room F 715, New York, NY 10021, USA.
E-mail: epeersch@med.cornell.edu


Red blood cells (RBC) contribute significantly to haemostasis and thrombosis under oscillatory flow conditions, and erythrocytosis has been associated with increased thrombotic risk. To measure the dynamic influences of RBC on platelets, we used a recently described cone and plate(let) analyzer (CPA), evaluating the effect of haematocrit (Hct) on platelet function in whole blood under arterial flow conditions (1800/s, 2 min, 25°C). Anticoagulated blood, reconstituted to varying haematocrits with autologous RBC, demonstrated a significant increase in adherent platelet aggregate formation at Hct levels >45%. This increase was not affected by pretreatment of blood with 0·05 mmol/l aspirin, but was prevented by antagonists of P2Y1, P2Y12, or P2X1, ADP and ATP receptors, and by converting exogenous ADP to ATP with creatine phosphate/creatine phosphokinase. As negligible platelet granule secretion was measured during CPA analysis, but metabolic inhibition of RBC with sodium azide or glutaraldehyde fixation inhibited erythrocytosis-enhanced increases in platelet aggregate size, adenine nucleotides contributing to shear-induced platelet aggregate formation appear to be derived from erythrocytes. These findings support the use of CPA for ex vivo evaluation of the contribution of RBC to platelet function and its inhibition under physiological shear conditions.

Erythrocytes contribute significantly to haemostasis and thrombosis by enhancing platelet function under dynamic flow conditions. Red blood cells (RBC) physically enhance the interaction of platelets with the vascular surface, and increase platelet eicosanoid formation to promote the recruitment of additional platelets from the microenvironment into the forming thrombus (Turitto & Weiss, 1980, 1998; Santos et al, 1991; Valles et al, 1991). RBC also potentiate shear induced platelet aggregation, particularly at high shear rates and in the presence of increasing haematocrit (Hct), through the release of adenine nucleotides (ADP, ATP) (Stormorken, 1971; Born et al, 1976; Born & Wehmeier, 1979; Schmid-Schonbein et al, 1981; Reimers et al, 1984).

At high shear rates of 2600/s, both platelet adhesion and thrombus formation have been observed to increase with increases in Hct (Turitto & Weiss, 1980). Moreover, erythrocytosis is clinically associated with increased thrombosis risk (Kessler, 2004), and aspirin was recently shown to enhance cardiovascular event-free survival in patients with polycythaemia vera (Landolfi et al, 2004). The increased thrombotic risk observed in patients with myeloproliferative disease is thought to reflect altered interactions between platelets, other blood cells and vascular endothelial cells under dynamic blood flow conditions (Kessler, 2004).

Most conventional laboratory tests of platelet function fail to capture the effect of red cells on platelet function, and do not evaluate platelet function under oscillatory flow conditions. Thus, it is difficult to predict arterial thrombotic risk, which is estimated to represent 50–80% of thrombotic complications associated with essential thrombocythaemia or polycythaemia vera (Amitrano et al, 2003). To overcome limitations of in vitro platelet function analysis, and particularly to examine the effect of RBC on platelet thrombus formation in whole blood under dynamic flow conditions and to evaluate the potential efficacy of antiplatelet agents, we used a recently developed cone and plate(let) analyzer (CPA).

The CPA method for studying platelet function has been well described (Varon et al, 1997; Shenkman et al, 2000). Briefly, an aliquot of anticoagulated whole blood is placed into a polystyrene well, and a defined shear rate is applied using the cone and plate device. Platelets adherent to the surface are visualized after removal of blood from the wells, washing, and staining with May–Grünwald solution. The percentage of the well surface covered (SC) by the stained platelets and the average size (AS) of adherent platelet aggregates is quantified using an image analyzer. Under the conditions used, only platelets adhere to the well surface. Platelet deposition is both shear and time dependent, reaching maximal levels within 2 min at high shear rates (1800/s), typical of arterial blood flow.

On polystyrene surfaces, platelet adhesion requires plasma von Willebrand factor (VWF) and fibrinogen immobilization on the plastic surface, the expression of quantitatively and qualitatively normal levels of platelet membrane glycoprotein (GP)Ib-IX and GPIIb-IIIa, and platelet activation (Shenkman et al, 2000). As in the bleeding time and ex vivo tests of platelet function using whole blood under high shear stress conditions, platelet deposition during CPA is affected by thrombocytopenia and anaemia (Hellem et al, 1961; Harrison et al, 1999). The effect of elevated haematocrit on platelet function in this system has not previously been assessed.


Blood collection

Blood samples were obtained from healthy volunteers according to local Institutional Review Board approved protocols. Blood was collected routinely into 3·2% sodium citrate (9:1, blood:anticoagulant ratio). To rule out artefacts associated with calcium chelation, additional studies were performed using blood anticoagulated with the direct thrombin inhibitor, D-phe-pro-arg-chloromethylketone (PPACK; 40 μmol/l). In some studies, anticoagulated blood was incubated at 37°C for 15 min with 50 μmol/l acetyl salicylic acid (aspirin) before commencement of studies to inhibit cyclooxygenase-dependent platelet granule secretion (Peerschke, 1982).

Processing and reconstitution of blood samples

To obtain blood samples with elevated RBC or platelet counts, washed platelets, packed RBC and platelet poor plasma (PPP) were prepared from the same individual's blood sample. Whole blood was first centrifuged (160 g, 8–10 min) and the platelet rich plasma (PRP) was removed. The remaining portion of the blood sample was submitted to further centrifugation (1000 g, 20 min) to obtain PPP and packed RBC. The packed RBC concentration was measured using a microhaematocrit determination (Bull et al, 2003). Washed platelet concentrates were prepared from PRP by centrifugation (20 min, 1000 g) in the presence of ethylenediaminetetraacetic acid (EDTA) (13 mmol/l final concentration). The resulting platelet pellet was resuspended in a small volume of 0·01 mol/l HEPES-buffered modified Tyrode's solution to achieve a 5–10-fold increase in platelet count relative to PRP. Finally, the desired volumes of packed RBC, PPP and washed platelets were combined to reconstitute samples for CPA. Automated complete blood cell counts (CBC) (Sysmex R3000; Sysmex America, Inc., Mundelein, IL, USA) were obtained on the reconstituted whole blood specimens for confirmation of cell counts.

To distinguish between physical and chemical contributions of RBC to platelet adhesion and thrombus formation during CPA, RBC were depleted of ATP either by fixation in 0·25% glutaraldehyde or 1% paraformaldehyde or by treatment with 1% sodium azide in 0·15 mol/l NaCl, for 2–3 h at room temperature. The treated RBC were washed extensively with 0·15 mol/l NaCl by centrifugation (1000 g, 5 min) and resuspension, and were ultimately suspended in an appropriate volume of autologous PPP to achieve the desired Hct. Washed platelet concentrates were added to reconstitute samples for CPA.

Cone and plate analysis

Cone and plate analysis enables the measurement of platelet adhesion and aggregation under conditions of physiological shear wherein a cone and plate viscometer induces laminar flow with a uniform shear stress over a plastic plate SC by the rotating cone. A small volume (200 μl) of anticoagulated native or reconstituted blood was applied to a polystyrene plate (NuncTMΔ Surface; Nalge Nunc International, distributed by VWR Scientific, West Chester, PA, USA) and subjected to a defined shear rate of 1800/s for 2 min. After removal of the blood, the wells were washed with tap water, and their surface exposed to May–Grünwald stain, as previously described (Varon et al, 1997; Shenkman et al, 2000). Adherent platelets were visualized microscopically (100×) and quantified using an image analysis program to calculate the percentage of well SC and the AS of the stained platelet aggregates. Each blood sample was sheared in duplicate wells. Platelet deposition in each well was analysed in four separate quadrants. The data reflect the average of these measurements. The intra-assay variability of CPA analysis for SC and AS ranged from 21 to 36% and 6–15% respectively. The inter-assay variability was <10% for both parameters.

Platelet secretion

Cell-free supernatants were prepared by centrifugation (12 000 g, 5 min) of the blood samples pre- and post-CPA, and examined for secretion of dense and alpha granule contents. Dense granule secretion was quantified using platelets that had been prelabelled with 14C-serotonin (5-hydroxytryptamine creatine sulphate; Amersham Corp., Piscataway, NJ, USA) (185 kBq) in whole blood for 30 min at 37°C, as described (Peerschke, 1982). Alpha granule secretion was evaluated by measuring platelet factor 4 (PF4) release using a commercial enzyme-linked immunosorbent assay (ELISA) kit (Diagnostica Stago, Inc., Parsippany, NJ, USA) according to the manufacturer's instructions. Further studies examined CPA-induced soluble CD40 ligand antigen expression (Lindmark et al, 2000) using a commercial ELISA (Kamiya Biomedical Co., Seattle, WA, USA).

Platelet activating potential of post-CPA supernatants

Cell free supernatants were examined for their ability to support the aggregation of fresh platelets in PRP using laser light scatter aggregometry (KOWA PA-200; Kowa Co. Ltd, Tokyo, Japan) to quantify small, medium and large-sized aggregates in a defined observational volume (48 × 140 × 20 μm). For these studies, 30 μl of pre- or postshear, cell-free supernatant was added to 270 μl of PRP in an aggregometer cuvette. Samples were analysed for 5 min with constant stirring.

The principles of the laser light scatter aggregation method have been described (Ozaki et al, 1994). Briefly, a diode laser-light beam is passed through PRP stirred in a cylindrical glass cuvette with a 5-mm internal diameter. The light scattered from the observation volume is detected by a photocell array. Light intensity corresponds to particle size. Signal frequency for every 10 s represents the number of aggregates (counts per 10 s). Particles with an intensity of 25–400 mV represent small aggregates (9–25 μm). Those with an intensity of 400–1000 mV represent medium aggregates (25–50 μm), and those with an intensity of 1000–2048 mV represent large aggregates (50–70 μm). Small aggregates contain approximately 70–1400 platelets, medium aggregates contain approximately 1000–11 000 platelets, and large aggregates contain approximately 11 000–31 000 platelets.

Platelet antagonism

A variety of platelet antagonists were selected to gain insight into the molecular mechanisms governing platelet function during CPA in the presence of a high Hct. Platelets in PRP were pretreated (20 min, 37°C) with inhibitors of outside-in or inside-out signalling events (Shattil et al, 1994; Kovacsovics et al, 1995; van Willigen et al, 1996), such as inhibitors of phosphatidylinositol 3-kinase (PI-3 kinase) (Ly294002, 25 μmol/l), tyrosine kinase (Herbimycin A, 10 μmol/l), protein kinase C (PKC; Bis-indolylmaleimide, 10 μmol/l), and myosin light chain kinase (ML-7, 6 μmol/l). All inhibitors were purchased from Biomol (Plymouth Meeting, PA, USA). Dimethyl sulphoxide (DMSO) served as a vehicle control.

In addition, studies were performed to assess the role of adenine nucleotides in platelet function during CPA. For these studies, platelets were sheared in the presence of the combination of 2 mmol/l creatine phosphate (CP) and 20 U/ml creatine phosphokinase (CPK) to convert exogenous ADP to ATP, or the following ADP and ATP receptor antagonists (Gachet, 2001; Hollopeter et al, 2001): MRS 2179 (P2Y1, P2X1) (200 μmol/l), MRS 2159 (P2X1) (200 μmol/l), or 2 methylthioadenosine 5′-monophosphate (2MeSAMP) (P2Y12) (10–100 μmol/l), respectively, from Sigma Chemical Corp. (St Louis, MO, USA). The inhibitory activity of CP/CPK, and ADP receptor antagonists was confirmed by platelet aggregation studies using PRP stimulated with 20 μmol/l ADP in a Kowa PA 200 aggregometer. MRS 2159 (200 μmol/l) failed to inhibit ADP-induced platelet aggregation.


Effect of Hct on platelet function under high shear stress using CPA

CPA measurements of platelet deposition from whole blood under high shear stress conditions at a normal Hct (30–45%) was dependent on platelet count, such that platelet deposition/surface coverage (SC) on polystyrene surfaces increased linearly with platelet counts from 100 × 109/l to 1200 × 109/l. No significant change in the AS of platelet deposits was noted over this range of platelet counts (Fig 1A and B). These results are consistent with reports of CPA on extracellular matrix-coated surfaces (Varon et al, 1997).

Figure 1.

Effect of platelet count on platelet deposition during high shear flow conditions using CPA. Reconstituted whole blood (Hct 33–46%), anticoagulated with 3·2% sodium citrate, was circulated on polystyrene surfaces at a shear rate of 1800/s for 2 min. Surfaces were washed, stained and analysed as described in Methods. Changes in (A) platelet adhesion (surface coverage) (n = 20) and (B) aggregate size (average size) (n = 93) as a function of platelet count are shown. Each data point represents results from duplicate specimens.

The effect of high Hct on platelet adhesion and aggregation during CPA has not been studied previously. Using CPA on polystyrene surfaces, a distinct change in the platelet deposition pattern was seen with increasing Hct (>45%) (Fig 2). This was most dramatic in test samples with a Hct ≥50%, which consistently demonstrated a marked increase in platelet aggregate formation with decreased single platelet deposition. These morphological observations correlated with quantifiable increases of 30–50% in platelet aggregate size (Fig 3). Slight increases (10–20%) in SC were also noted, but these did not reach statistical significance using paired Student's t-test analysis.

Figure 2.

Typical effects of erythrocytosis and thrombocytosis on platelet deposition patterns seen on polystyrene surfaces following CPA. Reconstituted whole blood samples were sheared over polystyrene for 2 min at 1800/s. Surfaces were washed, stained and analysed as described in Methods. Images were obtained by light microscopy (100×). (A) Control (Hct 38%, platelet count 252 × 109/l). (B) Thrombocytosis (Hct 38%, platelet count 705 × 109/l). (C) Erythrocytosis (Hct 62%, platelet count 252 × 109/l).

Figure 3.

Effect of haematocrit on platelet aggregate size (average size) measured following CPA (n = 80). Reconstituted blood samples (Hct 30–65%, platelet count 150–280 × 109/l), anticoagulated with 0·32% sodium citrate, were circulated on polystyrene for 2 min at 1800/s. Surfaces were washed, stained and analysed as described in Methods. Each data point represents results from discrete reconstituted blood samples analysed in duplicate.

As platelets are concentrated in a decreasing plasma volume as the RBC count/Hct increases, studies were performed also with samples exhibiting thrombocytosis but not erythrocytosis. A comparison of platelet deposition patterns obtained using reconstituted blood samples with erythrocytosis (Fig 2B) or thrombocytosis (Fig 2C) suggests that increases in the AS of platelet deposits (thrombus size) appears to be a direct effect of increasing Hct, but not of increasing platelet count.

The lack of effect of thrombocytosis on platelet aggregate size may reflect both the physical and biochemical contribution of RBC to platelet function at high shear conditions. RBC appear to increase platelet aggregate size by physically concentrating platelets to adhesive surfaces (Turitto & Weiss, 1998), and by chemically activating platelets (Reimers et al, 1984; Valles et al, 1991), producing increased platelet–platelet interactions, i.e. aggregation.

To rule out artefacts induced by calcium chelation in whole blood when sodium citrate is used for anticoagulation, additional studies were performed using the direct thrombin inhibitor PPACK. Similar increases in the AS of platelet deposits were noted in samples with high Hct compared with normal Hct. These observations are consistent with the observation that, except for EDTA, which prevents platelet adhesion and aggregate formation, most anticoagulants have little effect on shear stress-induced responses in a cone and plate viscometer (Alveriadou et al, 1993).

Increases in average/aggregate size of platelet deposits following CPA under high shear stress conditions in the presence of an elevated Hct were not accompanied by significant dense granule secretion (Table I). Minimal soluble alpha granule PF4 secretion and soluble CD-40 ligand expression was detected in the plasma of blood samples sheared at a high Hct. In contrast, a significant increase in the proaggregatory activity of the post-CPA cell-free supernatant from samples sheared at a high compared with a normal Hct was noted (Table I). This proaggregatory activity was characterized by the ability of cell-free supernatants to aggregate naïve platelets in PRP using a laser light scatter aggregometer. The number of small aggregates nearly doubled following addition of supernatants from sheared samples with an elevated Hct, when compared with a normal Hct.

Table I.  Effect of haematocrit on shear-induced elaboration of proaggregatory activity.
TestNormal HctHigh Hct
  1. Reconstituted whole blood samples were circulated on polystyrene surfaces at a shear rate of 1800/s for 2 min. Surfaces were washed, stained and analysed as described in Methods. Cell counts in reconstituted samples were as follows: normal Hct 33–48%, platelet counts 122–254 × 109/l, and high Hct 58–59%, platelet counts 148–244 × 109/l.

  2. *Statistical significance (P < 0·05), comparing samples with a normal and a high Hct (n = 5) using the paired Student's t-test analysis.

14C-serotonin release (%)6 ± 80 ± 0
PF4 (ng/ml)8 ± 518 ± 6*
Soluble-CD40 ligand (ng/ml)00·166 ± 0·112*
Small aggregate formation (no./observation volume 48 × 140 × 20 μm 10/s)5333 ± 11549333 ± 1527

Inhibition of erythrocytosis enhanced platelet aggregate formation during CPA

To better understand molecular mechanisms contributing to increased platelet aggregate/thrombus formation during CPA under high shear conditions, particularly at a high Hct, further experiments were conducted using well characterized inhibitors of platelet function. Data summarized in Fig 4 demonstrate that inhibition of intracellular signalling pathways involving PKC or myosin-light chain kinase resulted in statistically significant inhibition of erythrocytosis-enhanced platelet thrombus formation. The observed reduction in aggregate size of approximately 30%, returned aggregate size back to levels observed at a normal Hct. Inhibition of PKC by bisindolyl maleimide also decreased platelet adhesion (SC) by 61 ± 23%, n = 5, P < 0·05. In contrast, inhibition of PI-3 kinase or protein tyrosine kinase had no measurable effect on either SC or AS in samples sheared at a high Hct. Interestingly, pretreatment of whole blood with aspirin, an inhibitor of cyclooxygenase-mediated platelet granule secretion and thromboxane A2 formation, had no effect on shear-induced platelet deposition (SC) or AS using CPA, regardless of Hct.

Figure 4.

Effect of platelet antagonists on erythrocytosis-induced increases in platelet aggregate size (average size) during 2-min shear at 1800/s on polystyrene. Platelets were preincubated (20 min 37°C) with DMSO or 0·15 mol/l NaCl, 25 μmol/l PI 3-kinase inhibitor LY294002, 10 μmol/l PKC inhibitor bis-indolylmaleimide (BIM), 6 μmol/l myosin light chain kinase inhibitor ML-7, 10 μmol/l protein tyrosine kinase inhibitor Herbimycin A (HA), or 0·05 mmol/l aspirin, as described in Methods. Results are expressed as a percentage, relative to the average size of platelet thrombi formed by samples sheared at normal haematocrit after treatment with appropriate vehicle controls (DMSO or saline). *Statistical significance with P < 0·05 comparing high haematocrit vehicle control samples with high haematocrit samples incubated with various inhibitors, n = 5.

To investigate the role of adenine nucleotide ADP/ATP receptors in platelet aggregate formation during CPA at a high Hct, studies were performed in the presence of the following ADP receptor antagonists: MRS 2179, recognizing P2Y1 and P2X1, MRS 2159 recognizing P2X1, or 2MeSAMP antagonizing the P2Y12 receptor. All three adenine nucleotide receptor antagonists produced a statistically significant reduction in platelet thrombus formation following CPA at a high Hct (Table II). Under these conditions, aggregate size remained at levels observed in the presence of a normal Hct. Blockade of P2X1 or P2Y12 also had significant effects on thrombus formation at a normal Hct (Table II). Moreover, P2X1 and P2Y12 receptor antagonists markedly reduced platelet adhesion (SC) during CPA, whereas inhibition of P2Y1 failed to cause a statistically significant inhibition of platelet adhesion either at a normal or high Hct.

Table II.  Effect of adenine nucleotides/receptors on platelet deposition under high shear stress conditions using CPA.
 Surface coverage (% of control)Average size (% of control)
Normal HctHigh HctNormal HctHigh Hct
  1. Reconstituted whole blood samples were circulated on polystyrene surfaces at a shear rate of 1800/s for 2 min in the presence or absence (control) of MRS 2179 (200 μmol/l) recognizing the P2Y1 ADP receptor, MRS 2159 (200 μmol/l) recognizing the P2X1 ATP receptor, 2MeSAMP (20 μmol/l) antagonizing the P2Y12 ADP receptor, or the combination of 2 mmol/l CP/20 U/ml CPK converting exogenous ADP to ATP. Surfaces were washed, stained and analyzed as described in Methods. Normal Hct ranged from 36% to 39% with platelet counts of 120–180 × 109/l. High Hct ranged from 54% to 56%, with platelet counts of 145–210 × 109/l. Data for surface coverage (SC) and average aggregate size (AS) are expressed as a percentage relative to untreated, normal Hct controls.

  2. *Statistically significant difference using relevant Hct controls, determined using paired Student's t-test analysis (P < 0·05, n = 4).

Control100111 ± 14100151 ± 14
MRS 217977 ± 2060 ± 2191 ± 1092 ± 9*
MRS 215925 ± 10*33 ± 13*70 ± 5*66 ± 15*
2MeSAMP31 ± 5*44 ± 11*80 ± 5*90 ± 10*
CP/CPK23 ± 5*73 ± 17*70 ± 5*90 ± 4*

To further examine the role of exogenous ADP in platelet deposition during CPA, studies were performed with the combination of CP/CPK to convert exogenous adenine nucleotides to ATP. At doses causing complete inhibition of 20 μmol/l ADP-induced platelet aggregation in PRP, CP/CPK significantly inhibited platelet adhesion (SC) and thrombus formation (AS) during CPA, regardless of Hct.

Effect of RBC on platelet activation by CPA

As RBC contribute both physically and chemically to platelet deposition on surfaces in flowing blood (Stormorken, 1971; Born et al, 1976; Born & Wehmeier, 1979; Turitto & Weiss, 1980, 1998; Schmid-Schonbein et al, 1981; Reimers et al, 1984; Santos et al, 1991; Valles et al, 1991), and negligible platelet dense granule secretion was observed during CPA, further studies explored RBC as a source of exogenous adenine nucleotides supporting platelet thrombus formation. Results are summarized in Table III. Metabolically inactive, glutaraldehyde-fixed RBC, or RBC depleted of ATP following incubation with saline-azide, prevented the observed increases in platelet aggregate size at a high Hct. Azide treatment of RBC, however, had no statistically significant effect on platelet deposition when samples were sheared at a normal Hct. In contrast, fixed RBC reduced both platelet adhesion and aggregate formation during shear, regardless of Hct.

Table III.  Effect of RBC ATP depletion on platelet deposition.
RBC treatmentSurface coverage (% of control)Average size (% of control)
Normal HctHigh HctNormal HctHigh Hct
  1. Reconstituted blood samples were circulated on polystyrene surfaces at a shear rate of 1875/s for 2 min. Red blood cells were either pretreated (3 h at 37°C) with 1% sodium azide (Azide) or fixed with 0·25% glutaraldehyde (Fixed). Surfaces were washed, stained and analysed as described in Methods. Cell counts were as follows: normal Hct ranged from 32% to 45% with platelet counts of 140–180 × 109/l; high Hct ranged from 50% to 60% with platelet counts of 152–200 × 109/l.

  2. *Statistically significant differences using relevant Hct controls (P < 0·05 calculated using the paired Student's t-test, n = 4).

Control100121 ± 31100 140 ± 16
Azide87 ± 1071 ± 1699 ± 1697 ± 21*
Fixed3·5 ± 7*18 ± 16*22 ± 39*50 ± 43*


Considerable insight has been gained over the past 20 years into the physical and molecular mechanisms involved in platelet function in flowing blood. Overall, platelet interaction with the subendothelium or adhesive proteins on surfaces depends on two independent mechanisms: transport of platelets from the blood to the vessel wall or surface, and reaction of the platelets with vascular components. At low shear rates (50–200/s), platelet adhesion increases as the Hct increased from 10 to 40%, but is independent of a Hct above 40% (Turitto & Weiss, 1980, 1998). Under these low shear conditions, virtually no platelet thrombi form, regardless of the Hct. At high shear rates, 800–10 000/s, however, platelet adhesion is strongly influenced by Hct. Indeed, at blood shear rates of 2600/s both adhesion and thrombus formation increased continuously as the Hct increased from 10 to 70% (Turitto & Weiss, 1980, 1998). Moreover, correction of the platelet adhesion defect in delta storage pool deficiency has been described with elevation of the Hct (Weiss et al, 1996). This effect of red cells appears to be chemical rather than physical.

Using CPA as a convenient test of platelet function in whole blood under arterial flow conditions, adherent platelet aggregate size was found to increase as a function of Hct above approximately 45–50%. No change in aggregate size (AS) was evident as the Hct varied within the normal range, however. These observations extend previous findings (Varon et al, 1997) demonstrating a reduction in platelet adhesion (SC) during CPA below a Hct of 30%, with a nearly 60% decrease in platelet adhesion noted below a Hct of 20%. These data are consistent with reported increases in the bleeding time in patients with a low Hct (17–25%), and the restoration of normal bleeding time following correction of the Hct (Editorial, 1984).

In the present study, pathologically elevated Hct levels (>50%) increased platelet aggregate size by 30–50% following CPA. This was accompanied by the release of increased proaggregatory activity in the cell-free, postshear supernatant, suggesting the presence of increased platelet recruitment potential. Indeed, cell–cell interactions between platelets and erythrocytes can significantly alter platelet reactivity, through modulation of both platelet activation and recruitment (Stormorken, 1971; Born et al, 1976; Born & Wehmeier, 1979; Schmid-Schonbein et al, 1981; Reimers et al, 1984).

In the present study, metabolically intact erythrocytes enhanced platelet adhesion and aggregation on polystyrene during high shear stress conditions. Substitution of fixed RBC or azide-treated RBC decreased platelet adhesion and aggregate formation. Interestingly, RBC treated with sodium azide to deplete their ATP supply remained more active than fixed RBC in supporting platelet function during CPA. This observation is consistent with metabolic and physical contributions of RBC to platelet function under arterial flow conditions, with less deformable fixed RBC supporting platelet interactions with surfaces under oscillatory flow conditions less well.

As aggregate size may reflect thrombotic potential, further studies were performed to evaluate the mechanism of erythrocyte-induced increases in platelet aggregate formation in the CPA system. The results suggested an important role for adenine nucleotides, both ADP and ATP, in thrombus growth during high shear stress at an elevated Hct. As minimal platelet dense granule secretion was detected during CPA, the majority of responsible adenine nucleotides appear to be derived from RBC. This may account for the lack of effect of aspirin on platelet aggregate size in the present study, despite good clinical efficacy in patients with polycythaemia vera (Landolfi et al, 2004). Indeed, it has been reported that the enhancing effect of erythrocytes on platelets is preserved in the complete absence of thromboxane generation following aspirin ingestion (Valles et al, 1998).

These observations are consistent also with reports (Moake et al, 1988; Ikeda et al, 1991) indicating a lack of effect by aspirin on the initiation of platelet aggregation in response to shear stress. Interestingly, it has been suggested that therapeutic aspirin regimens directed towards inhibition of platelet function, as evaluated in vitro in PRP, might not reflect therapeutic efficacy in vivo because of the prothrombotic effects of erythrocytes (Turner et al, 2001).

A variety of intracellular signalling pathways are involved in platelet adhesion and thrombus formation (Shattil et al, 1994; Kovacsovics et al, 1995; van Willigen et al, 1996). The signalling events induced by GPIb-IX-V binding to VWF include elevation in cytosolic calcium and activation of PKC, and tyrosine kinases (Moake et al, 1988; Ikeda et al, 1991; Ruggeri, 1993; Berndt et al, 2001). The observed bisindolyl-induced inhibition of platelet adhesion (reduced surface coverage) is consistent with these findings. The lack of inhibition of platelet thrombus growth at a high Hct by Herbimycin A, a protein tyrosine kinase inhibitor, is consistent with its lack of effect on ADP-induced platelet aggregation. Inhibition of myosin light chain kinase not only inhibited thrombus growth at a high Hct, but was also associated with inhibition of ADP-induced aggregation. Interestingly, myosin light chain kinase activation is common to platelet activation via signalling through both P2X1 and P2Y1 (Gachet, 2001; Toth-Zsamboki et al, 2003). The lack of effect of PI 3-kinase in this system is puzzling, as it fully prevented ADP-induced platelet aggregation, but may suggest that this pathway is not involved under high shear conditions and a pathologically high Hct.

The rheological effects of erythrocytes are often disregarded in conventional laboratory tests of platelet function to assess haemorrhagic or thrombotic risk. Results obtained in the present study using CPA offer strong evidence that under arterial shear conditions, engagement of P2Y1 and P2Y12 ADP receptors and P2X1 ATP receptors play a role in stable platelet adhesion and thrombus formation, particularly at a high Hct. These findings are consistent with in vivo studies demonstrating that blockade of ADP receptors P2Y12 and P2Y1 is required for the inhibition of platelet aggregation in whole blood under flow (Turner et al, 2001), and potentiation of platelet activation, adhesion and thrombus formation through stimulation of P2X1 receptors (Erhardt et al, 2003; Hechler et al, 2003; Oury et al, 2003). Moreover, stimulation of P2Y12 has been shown to be involved in platelet activation initiated by the binding of VWF to GPIbα induced by high shear rates (Goto et al, 2002), and is consistent with the present observations using CPA, of markedly decreased platelet adhesion and thrombus formation at a normal and high Hct.

In conclusion, the present study demonstrated that CPA detected many of the effects of RBC on platelet function occurring in vivo under dynamic flow conditions. CPA may offer an important tool for evaluating platelet RBC interactions and understanding signalling pathways coupling shear-induced GPIb-VWF interaction with adenine nucleotide receptor activation, and the evaluation and identification of novel anti-thrombotic therapy.

Conflict of Interest

Professor Savion and Dr Varon are founders of Matis-Medical Inc., which has developed the CPA technology.


This work was supported in part by grant HL 067211 from the National Institutes of Health, National Heart Lung and Blood Institute, and a grant from the Doctor's Cancer Foundation, New Rochelle, New York.