Clopidogrel is an antiplatelet drug that has recently joined the list of drugs being investigated for the secondary prevention of atherosclerotic and thrombotic events related to ischemia. 1 Current studies have reported some differences between clopidogrel and its chemical counterpart ticlopidine. The former has fewer side effects, is more potent as an antiplatelet drug, and is more convenient to administer, as a single dose of 75 mg day−1 achieves the same effect as 2 daily doses of 250 mg ticlopidine. 2
Clopidogrel and ticlopidine are members of the thienopyridine group that act by blocking the platelet ADP receptor, an effect that inhibits platelet activation of the IIb/IIIa glycoprotein complex (fibrinogen receptor). 3 However, acetylsalicylic acid (aspirin), which continues to be the reference antiplatelet agent and drug of choice in preventing fatal and non-fatal thrombotic complications of atherosclerosis, 4 acts preferentially by blocking cyclooxygenase (COX) in platelets and the endothelium by acetylating its active center. 5 Aspirin also reportedly increases nitric oxide (NO) production in neutrophils 6,7 and in the arterial wall. 8
A number of studies have shown that in platelet-enriched plasma, clopidogrel does not inhibit platelet aggregation after in vitro incubation at therapeutic concentrations. 9 However, it has been shown that clopidogrel is able to specifically inhibit platelet aggregation induced in vitro by ADP in washed platelets. 10
Little attention has been given to the effect of thienopyridines on the arachidonic acid pathway or platelet activation by collagen, the main inducer of platelet activation in human blood vessels. However, one recent study showed that ticlopidine increased NO production in human neutrophils 8; these results indicated that another antiplatelet mechanism exists in addition to blockage of the platelet ADP receptor.
The aim of the present study was to determine the in vitro influence of clopidogrel both in whole blood and in platelet-rich plasma platelet aggregation induced with collagen. Both ticlopidine and clopidogrel are antagonists of the platelet ADP P2Y12 receptor; however, collagen is the main inducer of platelet activation in vivo. For that reason we explore in this study collagen-induced platelet activation in static and flowing conditions. We compare the influence of clopidogrel on thromboxane A2 (TXA2), prostacyclin (PGI2), and NO production with its chemical-related compound ticlopidine, and with aspirin as the control drug. We also studied the influence of the inhibition of NO synthesis on the platelet effects of this drug.
Whole blood for this in vitro study was obtained from healthy men (mean age 37.6 ± 1.5 years, range 19–47 years) who had not taken any medication for at least 15 days previously. One sample was obtained from each donor; all samples were collected with 3.8% sodium citrate at a proportion of 1:10 (v v−1) as an anticoagulant. All blood samples were obtained between 9 hours and 10 hours before the donor had anything to eat. Each subject gave his informed consent to participate in the study.
The experiments to investigate the platelet-subendothelium interaction were performed in a perfusion chamber; the subendothelium was obtained from human umbilical vein endothelial cells (HUVEC) in culture. The umbilical cords were obtained from eutocic labors after maternal informed consent was obtained.
Acetylsalicylic acid (Sigma Chemical Corp, St. Louis, IL), ticlopidine (Sigma Chemical Corp, St. Louis, IL), and clopidogrel (Sanofi-Synthelabo, Barcelona, Spain), were incubated at different concentrations. N-nitro-L-arginine methyl ester hydrochloride (L-NAME) was incubated in each type of experiment as an inhibitor of nitric oxide synthase. Eight to ten different samples were run in each of the experiments detailed below.
Platelet aggregation was measured in whole blood and in platelet-rich plasma (PRP). PRP was obtained from whole blood after centrifugation at 180 ×g during 10 minutes at 20°C. We measured platelet aggregation with the electronic impedance method described by Cardinal and Flower 11. We used a Chrono-Log 540 aggregometer (Chrono-Log Corp., Haverton, PA), with ADP (2.5 μM) and collagen (1 μg ml−1) (Menarini Diagnostica, Barcelona, Spain) to induce aggregation. Drugs were incubated at 37°C for 5 minutes or 20 minutes before the aggregation inducer was added, and aggregation was recorded for 10 minutes. Maximum intensity of aggregation was quantified as the maximum change in electronic impedance in samples without the drug or with a given concentration of each drug.
The concentrations of the aggregating agent were chosen according to previous experiments in which EC50 values were as follows: 2.10 ± 0.37 μM for ADP (N = 10) and 1.10 ± 0.14 μg ml−1 for collagen (N = 10).
Platelet Production of Thromboxane B2
The production of thromboxane A2 was measured as the synthesis of its metabolite TxB2. Samples of whole blood stimulated with collagen (1 μg ml−1) were centrifuged at 10,000 ×g for 3 minutes, and the amount of TxB2 in the supernatant was determined with an enzyme immunoassay (Biotrak® RPN 220, Amersham International plc, Little Chalfont, Buckinghamshire, UK). The sensitivity of this method was 3.6 pg ml−1 for TxB2; within-assay variability for duplicate determinations was 2.8% and between-assay variability was 9.7%.
Leukocyte and Endothelial Production of Nitric Oxide
Nitric oxide production was induced with the calcium A 23187 ionophore (1 μM) and quantified in concentrates of polymorphonuclear leukocytes obtained by centrifugation of whole blood on a Ficoll gradient (Hystopaque 1077 and Hystopaque 1119, Sigma Diagnostics, Inc., St. Louis), and in cultured HUVEC. The mean number of leukocytes in whole blood was 6.8 ± 0.3 × 109 cells l−1, and the mean number of polymorphonuclear leukocytes in samples used for NO detection was 3.5 ± 0.2 × 109 cells l−1. The amount of NO was quantified with an electrochemical method 12 that used a specific electrode coupled to an ISO-NO detector (World Precision Instruments, Aston, Stevenage, Hertsforshire, UK).
Nitric oxide production was initially measured under basal conditions with the drug under study or saline solution (control sample) for 5 minutes or 20 minutes at 37°C. After this step, calcium A 23187 ionophore was added, and the increase in NO production induced by activation of constitutive or calcium-dependent NO synthase was recorded.
Endothelial Production of 6-keto-PGF1α
The production of prostacyclin was measured as the synthesis of its stable metabolite (6-keto-PGF1α). Endothelial cells harvested from human umbilical vein were stimulated with 1 μM calcium ionophore A 23187 for 3 minutes at 37°C. The amount of 6-keto-PGF1α in the supernatant of cultured endothelial cells was determined with an enzyme immunoassay (Biotrak® RPN 220). The sensitivity of this method was 3.1 pg ml−1; within-assay variability for duplicate determinations was 2.8% and between-assay variability was 9.7%.
Platelet-Vascular Subendothelium Interaction
To investigate platelet-vascular subendothelium interactions we used blood perfusion experiments with human subendothelial matrix preparations in a flat perfusion chamber. 13 The subendothelial matrix preparations were obtained from cultures of HUVEC basically as described by Jaffé et al 14.
The umbilical cord was cannulated at both ends, and endothelial cells were obtained by flushing the cord with a collagenase solution (0.05 g in 10 mL Hanks solution, pH 7.2). The resulting liquid was centrifuged to obtain the cell pellet, which was resuspended in a culture dish with MEM 199 culture medium supplemented with 20% deproteinated human serum and 2% penicillin-streptomycin (Bio-Whittaker Europe, Verviers, Belgium). Cell cultures were maintained by washing with Hanks solution (pH 7.2) and adding supplemented MEM 199 medium. Once the desired concentration of cells was reached, the cells were detached with a trypsin solution (pH 7.4) and cultured on pretreated coverslips (30 minutes with 0.1% gelatin, 30 minutes with 0.5% glutaraldehyde) to obtain subendothelial matrix preparations. To harvest the subendothelial matrices the cell growths were detached with 3% EGTA (pH 7.4).
For perfusion studies the blood samples were incubated for 5 or 20 minutes at 37°C with different concentrations of the drugs under study. The perfusion chamber was coupled to a peristaltic pump such that the incubated blood with or without (control samples) the drug circulated for 5 minutes at a shear rate of 800 seconds−1 in contact with the subendothelial matrix. The matrices were perfused in the presence and in the absence of the endothelium to test the possible influence of this part of the vessel wall on the platelet-vascular subendothelium interaction.
After perfusion each coverslip with the subendothelial matrix was washed with phosphate-buffered saline solution (pH 7.4) and platelet structures were fixed with an 0.5% solution of glutaraldehyde for 24 hours, then stained with 0.25% toluidine blue for morphometric analysis. In each perfused matrix we calculated the total percentage of subendothelium covered by platelets and the percentages of different platelet structures. For each perfused matrix we examined 20 fields, and for each concentration of drug we examined matrices from at least 4 different perfusions. Version 5.0 of the Visilog program (Noesis, Orsay-Cedex, France) was used with an inverted microscope to which a Sony B/W CCD camera was coupled.
In these experiments drugs were incubated in concentrations near to the plasmatic levels described in humans: 10 to 15 μg/ml (60–85 μM) for aspirin, 15 3 to 6 μg/ml (10–17 μM) for ticlopidine, 16 and 3 to 6 μg/ml (11–19 μM) for clopidogrel;17 we have chosen the lower limit of these intervals (60 μM for aspirin and 10 μM for ticlopidine and clopidogrel).
All data in the text, tables, and figures are the mean ± standard error of the mean (SEM) of all values for each experiment. The results were tested with one-way analysis of variance followed by Least Significant Difference test. All analyses were done with version 10.0 of the SPSS program (SPSS Co., Chicago, IL). The minimum value used to establish statistical significance was P < 0.05.
All 3 drugs inhibited platelet aggregation induced by ADP and by collagen (Figs. 1 and 2) in different ranges of concentrations: (i) aspirin had the lowest IC50 (concentration that inhibited platelet aggregation by 50%) when platelets were stimulated with collagen (Table 1), whereas the 2 thienopyridines had the lowest IC50 when platelets were stimulated with ADP needed 20 minutes of incubation to reach their maximum effect (Figs. 1 and 2); (ii) after incubation for 20 minutes, both thienopyridines influenced collagen-induced aggregation, although their effects were weaker than aspirin's (Fig. 1); (iii) the effect of clopidogrel was greater than that of ticlopidine regardless of whether aggregation was induced with collagen or ADP (Table 1). Values of IC50 for clopidogrel in platelet-rich plasma were: 832 ± 38.2 μM in experiments induced with ADP and 840 ± 41.1 μM in experiments induced with collagen.
All 3 drugs inhibited TxA2 production in a concentration-dependent way (Fig. 3), although the effect of aspirin was greater than that of either of the thienopyridines. Clopidogrel inhibited thromboxane production more efficiently than did ticlopidine (Table 1).
When we extrapolated the results for the concentrations usually attained in human plasma during chronic treatment with aspirin (60 μM), ticlopidine (10 μM), or clopidogrel (10 μM), we found that all 3 drugs inhibited platelet aggregation induced by collagen, and also inhibited platelet TxA2 production (Table 1).
The endothelial production of 6-keto-PGF1α was inhibited in a concentration-dependent way by aspirin, whereas ticlopidine and clopidogrel did not affect production (Fig. 4).
All 3 drugs increased leukocyte and endothelial production of NO in a concentration-dependent way (Fig. 5). The increase in endothelial NO production was similar with all 3 drugs at concentrations similar to those found in human plasma; however, the effect of clopidogrel was greater than that of ticlopidine at high concentrations. The thienopyridines increased leukocyte NO production more efficiently than did aspirin.
In blood perfusion experiments with subendothelial matrix preparations, all 3 drugs significantly reduced the surface area occupied by platelets in comparison with control samples. Aspirin reduced the area by 45.24%, ticlopidine by 76.75%, and clopidogrel by 51.27% (Fig. 6A).
The presence of endothelium in the flow experiment preparations increased the reduction in subendothelial surface occupied by platelets. In control samples (incubated with no drugs) the area was 28.97 ± 0.3% without endothelium, and 17.32 ± 0.54% with endothelium (P = 0.0001). The effect of ticlopidine and clopidogrel was not modified by the endothelium; the area occupied by platelets was reduced by 79.61 ± 61% with the former and by 46.22% with the latter, in comparison with control perfusions. The effect of aspirin on the area occupied by platelets was 33.00% weaker in the presence of endothelium (30% inhibition in comparison with control perfusions) (Fig. 6B).
Under flow conditions all 3 drugs affected mainly the larger platelet structures (> 61 μm2) both in the absence and in the presence of endothelium (Table 2). The greatest effect of ticlopidine was to reduce the area of the largest structures, whereas the effect of clopidogrel increased the most in the presence of endothelium (Table 2).
Figure 7 shows representative images of the structures formed in different platelet-vascular subendothelium interactions after incubation with the different drugs.
The incubation of l-NAME increased maximum intensity of platelet aggregation and subendothelial surface occupied with platelets (Table 3). The in vitro effect of acetylsalicylic acid and ticlopidine was reduced in the presence of l-NAME, while clopidogrel did not modify its antiplatelet effects (Table 4).
This study shows that clopidogrel has antiplatelet action in vitro when whole blood is induced with ADP or collagen, although the effect is greater when ADP is used. An earlier study found that in platelet-enriched plasma, clopidogrel did not inhibit platelet aggregation after incubation in vitro with concentrations lower than 100 μM. 9 The authors proposed that biotransformation in the liver was needed for the in vitro effects to appear; in other words, the antiplatelet effect might be due to a metabolite of the drug rather than to clopidogrel itself. 9 This finding is in accord with our results for clopidogrel in platelet-rich plasma, showing IC50 values 4-fold higher for collagen-induced aggregation and 138-fold higher for ADP-induced aggregation. However, these experiments were performed in animals, and the doses used were approximately 40-fold higher than the equivalent therapeutic doses in humans. Thus the hypothesis that a metabolite causes the anti-aggregant effect remains untested in humans. We found only 1 published study that showed that clopidogrel was able to specifically inhibit ADP-induced platelet aggregation in vitro, but the experiments were performed in washed platelets. 10 The lower effect of clopidogrel in platelet-rich plasma (presence of protein) may be explained by a stimulation of ADP secretion, eventually resulting in desensitization of the P2Y1 receptor. 18 In the present study the antiplatelet effect of clopidogrel has been found in whole blood (presence of protein); we can speculate that red blood cells, leukocytes, or even platelets could biotransform clopidogrel into its main active metabolite.
The present results also show that with in vitro tests, an incubation period of at least 20 minutes is needed for thienopyridines to fully exert their action, a finding consistent with the results of Weber et al 10. Thus in experiments designed to measure TxA2, prostacyclin, and NO production and the effect of blood perfusion we used a maximum incubation period of 20 minutes.
When collagen was used to induce platelet aggregation, the effect appeared mainly via the intraplatelet phosphoinositol-TxA2 pathway. Consequently any drug that inhibits collagen-induced platelet aggregation would be expected to have an effect on thromboxane production. This has been shown to be the case for aspirin, an irreversible inhibitor of COX. In an earlier study with diabetic rats, we showed that the chronic oral administration of clopidogrel at 10 mg kg−1 day−1 inhibited platelet thromboxane synthesis to a greater extent than did ticlopidine. 19 However, we found no earlier studies that documented the in vitro effect of clopidogrel on TxA2 production. The ability of thienopyridines to inhibit TxA2 production may be explained from the finding that ticlopidine stimulates the release of arachidonic acid from membrane phospholipids. 20 This gives rise to the formation of cyclic endoperoxides, which are used by ticlopidine to potentiate prostaglandin E2 formation, 21 a process that depletes the available substrate for TxA2 formation.
Clopidogrel may act through a similar mechanism, and in fact inhibits TxA2 production more efficiently than does ticlopidine. This may be because the former drug appears to act on COX, the target enzyme of aspirin. The chemical structure of both drugs contains a methylcarboxy group, the structure responsible for COX inhibition. However, any possible effect on COX is probably weak, as clopidogrel does not modify prostacyclin synthesis, possibly because any increase in synthesis would be offset by the increased release of arachidonic acid. 20
The effect of acetylsalicylic acid on NO production has been related to its greater effect in whole blood 6,22 and its neuroprotective action. 23 In principle, thienopyridine derivatives might be assigned the same applications as aspirin, in as much as the former also enhance an endogenous anti-aggregant, vasodilating, and cytoprotective substance; moreover, the very low effect of clopidogrel in platelet-rich plasma enhances the importance of its action on NO production in whole blood. We found no earlier publications that reported a direct relationship between clopidogrel and an increase NO production. One in vitro study from our laboratory showed that ticlopidine increased calcium-dependent NO production in neutrophils. To add to this line of evidence, the present study shows that both thienopyridine derivatives stimulate calcium-dependent NO production at concentrations that are clearly able to inhibit platelet aggregation in vitro, and that are equivalent to plasma concentrations in humans on chronic oral treatment with these drugs (179% for aspirin, 188% for ticlopidine and clopidogrel). Moreover, this effect may account for the beneficial effect seen after the administration of thienopyridines (in comparison with aspirin) in experimental diabetic retinopathy, 19 a situation in which vascular calcium-dependent NO production is deficient. 24 The absence of influence of the NOS inhibition on the clopidogrel effect increases the importance of this drug in these pathologic conditions.
The anti-aggregant effect of clopidogrel in vitro is greater than that of ticlopidine. However, in the flow system experiments, designed to investigate platelet-vascular subendothelium interactions, we found ticlopidine to be the more active of the two, leading to greater reductions in the height of large platelet structures. This discrepancy may reflect the fact that in the flow system, clopidogrel acts more slowly and requires a longer incubation period to fully exert its effects. Other studies that used the same experimental technique and drugs, but tested different thrombogenic surfaces at a different shear rate (we used 800 seconds−1), found different percentage rates of inhibition in subendothelial surface coverage by platelets. 25,26 (We note that 800 seconds−1 is equivalent to the rheological conditions in medium-caliber blood vessels such as the femoral artery.) However, our findings are similar to those of Roald et al 27, who found that at a shear rate of 2600 seconds−1, clopidogrel reduced platelet-vascular subendothelium interactions by 51%, a value close to the 56% reduction we found at 800 seconds−1.
In an attempt to reproduce conditions in the living artery, a separate set of experiments included the endothelium in the perfusion set-up to study the possible influence of these cells on platelet-vascular subendothelium interactions. The results lead us to conclude that the endothelium did not modify the antiplatelet action of clopidogrel or ticlopidine, but did reduce the effect of aspirin by 33%. This may reflect the inhibition by aspirin of prostacyclin synthesis in the endothelium, an effect that would remove one of the biochemical components that prevents platelet aggregation.
In conclusion, we found that clopidogrel inhibits platelet aggregation under in vitro conditions and in whole blood, and also inhibits TxA2 production. This latter effect, however, can be looked upon as a co-adjuvant of its main mechanism of action. Moreover, clopidogrel respects prostacyclin production. Clopidogrel (10 μM) increases endothelial NO production to a degree similar to aspirin (60 μM), and increases leukocyte NO production more efficiently than aspirin. Although the presence of the endothelium can partly condition the effects of aspirin, it appears to have no such effect on the action of clopidogrel.
The authors thank Antonio Pino Blanes and Manuela Vega for their valuable technical assistance. Our appreciation is expressed to the Centro Regional de Donantes de Sangre of Málaga for their help with collecting blood samples, the auxiliary staff of the Hospital Galvez in Málaga for obtaining umbilical cords, and K. Shashok for translating parts of the manuscript into English.
1. Caprie Steering Committee. A randomised, blinded trial of clopidogrel
versus aspirin in patients at risk of ischaemic events (CAPRIE). Lancet. 1996; 348:1329–1339.
2. Hirsh J, Weitz JI. New antithrombotic agents. Lancet. 1999; 353:1431–1436.
3. Foster CJ, Prosser DM, Agans JM, et al. Molecular identification and characterization of the platelet ADP receptor targeted by thienopyridine antithrombotic drugs. J Clin Invest. 2001; 107:1591–1598.
4. Antithrombotic Trialists' Collaboration. Collaborative meta-analysis of randomised trials of antiplatelet therapy for prevention of death, myocardial infarction, and stroke in high risk patients. BMJ. 2002; 324:71–86.
5. Vane JR. Inhibition of prostaglandin synthesis as a mechanism of action for aspirin-like drugs. Nat New Biol. 1975; 231:232–235.
6. Lopez-Farré A, Caramelo C, Esteban A, et al. Effects of aspirin on platelet-neutrophil interactions. Role of nitric oxide
and endothelin-1. Circulation. 1995; 91:2080–2088.
7. De La Cruz JP, Blanco E, Sanchez De La Cuesta F. Effect of dipyridamole and aspirin on the platelet-neutrophil interaction via the nitric oxide
pathway. Eur J Pharmacol. 2000; 397:35–41.
8. De La Cruz JP, Arrebola MM, Guerrero A, et al. Influence of nitric oxide
on the in vitro antiaggregant effect of ticlopidine. Vascul Pharmacol. 2002; 38:183–186.
9. Savi P, Herbert JM, Pflieger AM, et al. Importance of hepatic metabolism in the antiaggregating activity of the thienopyridine clopidogrel
. Biochem Pharmacol. 1992; 44:527–532.
10. Weber AA, Reimann S, Schrör K. Specific inhibition of ADP-induced platelet aggregation by clopidogrel
in vitro. Br J Pharmacol. 1999; 126:415–420.
11. Cardinal DC, Flower RJ. The electronic aggregometer: a novel device for assessing platelet behaviour in blood. J Pharmacol Methods. 1980; 1:135–158.
12. Shibuki K. Detection of nitric oxide
by an electrochemical micropobe. A comparison to methods in neuroscience. Neuroprotocols. 1992; 1:151–157.
13. Sakariassen KS, Aarts PA, De Groot PG, et al. A perfusion chamber developed to investigate platelet interaction in flowing blood with human vessel wall cells, their extracellular matrix, and purified components. J Lab Clin Med. 1983; 102:522–535.
14. Jaffe EA, Nachman RL, Becker CG, et al. Culture of human endothelial cells derived from umbilical veins. J Clin Invest. 1973; 147:2745–2756.
15. Hart FD, Huskisson EC. Nonsteroidal anti-inflammatory drugs. Current status and rational therapeutic use. Drugs. 1984; 27:232–255.
16. Knudsen JB, Bastain W, Sefton CM, et al. Pharmacokinetics of ticlopidine during chronic oral administration to healthy volunteers and its effects on antipyrine pharmacokinetics. Xenobiotica. 1992; 22:579–589.
17. McEwen J, Strauch G, Perles P, et al. Clopidogrel
bioavailability: absence of influence of food or antacids. Semin Thromb Hemost. 1999; 25:47–50.
18. Herbert JM, Savi P. Non-specific inhibition of ADP-induced platelet antiaggregation by clopidogrel
in vitro. Thromb Haemostas. 1999; 82:156–157.
19. De La Cruz JP, Moreno A, Martinez-Cerdan E, et al. Effect of clopidogrel
and ticlopidine on experimental diabetic ischemic retinopathy in rats. Naunyn Schmiedebergs Arch Pharmacol. 2003; 367:204–210.
20. Chap H, Simon MF, Douste-Blazy L. Effect of ticlopidine on arachidonic acid metabolism in platelet phospholipids in vitro. J Pharmacol. 1981; 30:1463–1468.
21. Lagarde M, Ghazi I, Dechayanne M. Effects of ticlopidine on platelet prostaglandin metabolism. Possible consequences for prostacyclin
production. Prostaglandins Med. 1979; 2:433–439.
22. De La Cruz JP, Camara S, Bellido I, et al. Platelet aggregation in human whole blood after chronic administration of aspirin. Thromb Res. 1987; 46:133–140.
23. Moro MA, De Alba J, Cardenas A, et al. Mechanisms of the neuroprotective effect of aspirin after oxygen and glucose deprivation in rat forebrain slices. Neuropharmacology. 2000; 39:1309–1318.
24. De La Cruz JP, Guerrero A, Paniego MJ, et al. Effect of aspirin on prostanoids and nitric oxide
production in streptozocin-diabetic rats with ischemic retinopathy. Naunyn Schmiedebergs Arch Pharmacol. 2002; 365:96–101.
25. Bossavy JP, Thalamas C, Sagnard L, et al. A double-blink randomized comparison of combined aspirin and ticlopidine therapy versus aspirin or ticlopidine alone on experimental arterial thrombogenesis in humans. Blood. 1998; 5:1518–1525.
26. Cadroy Y, Bossavy JP, Thalamas C, et al. Early potent antithrombotic effect with combined aspirin and a loading dose of clopidogrel
on experimental arterial thrombogenesis in humans. Circulation. 2000; 101:2823–2828.
27. Roald HE, Barstad RM, Kierulf P, et al. Clopidogrel
—a platelet inhibitor which inhibits thrombogenesis in non-anticoagulated human blood independently of the blood flow conditions. Thromb Haemost. 1994; 71:655–662.
Keywords:© 2004 Lippincott Williams & Wilkins, Inc.
clopidogrel; nitric oxide; platelet-subendothelium interaction; prostacyclin; thromboxane A2