Persistent production of platelet thromboxane A2 in patients chronically treated with aspirin


Fabio M. Pulcinelli, Dipartimento di Medicina Sperimentale e Patologia, Università degli Studi ‘La Sapienza’, Viale Regina Elena 324, 00161 Roma, Italy.
Tel.: +39 6 49973002; fax: +39 6 4454820; e-mail:


Summary. Background: Patients treated with aspirin may have a reduced sensitivity to its antiplatelet effect. The mechanism accounting for such a reduced sensitivity might involve an impaired interaction of aspirin with cyclooxygenase-1 (COX)-1. Objective: We sought to investigate whether platelets from patients under chronic treatment with aspirin still produce TxA2 and whether there is any relationship between the eventual persistent TxA2 formation and platelet aggregation. Finally, whether platelet-derived TxA2 can be inhibited by in vitro addition of aspirin. Methods: Collagen-induced platelet aggregation and thromboxane-A2 (TxA2) were measured in 196 patients treated with aspirin (100–330 mg day−1) because of previous vascular events or presence of risk factors of atherosclerosis. Results: Collagen-induced TxA2 production of the entire cohort was 128.7 ± 21.6 pg 10−8 cells, and was significantly correlated with platelet aggregation (Spearman's correlation coefficient = 0.44; P < 0.0001). Patients in the highest quartile of TxA2 showed higher platelet response to collagen (P < 0.0001) when compared with those in the lowest quartile. In a subgroup of 96 patients, platelets were treated in vitro with a TxA2 receptor antagonist (13-azaprostanoic acid) or aspirin before stimulation with collagen. 13-APA acid significantly inhibited platelet aggregation. Aspirin reduced (−72.9%) TxA2 production in patients with TxA2 values above the median but it was ineffective in those with TxA2 values below the median. Conclusion: In some patients chronically treated with aspirin platelet production of TxA2 may persist and account for enhanced platelet aggregation. Incomplete inhibition of COX-1 seems to be implicated in persistent TxA2 production.


Acetylsalicylic acid (aspirin) is an antiplatelet agent that inhibits platelet cyclooxygenase-1 (COX-1), thus preventing the formation of thromboxane A2 (TxA2), potent aggregating and vasoconstrictor molecule [1].

Although the antithrombotic property of aspirin has been well documented in patients with acute coronary syndrome [2], recent clinical and laboratory studies evidenced that aspirin does not exert an equal antiplatelet action on all subjects (see Ref. [3], for review). In a recent in vitro study of our group, we observed, by monitoring collagen-induced platelet aggregation during a 24 months follow-up, that platelet sensitivity to the inhibitory function of aspirin progressively decreased with time [4].

The phenomenon of the insufficient inhibition of platelet function by aspirin has been termed ‘aspirin resistance’; however, it has been suggested that, clinically, ‘aspirin resistance’ should be better defined as aspirin failure to prevent an arterial thrombotic event [5]. An interesting approach based on biochemical methods and functional in vitro studies on healthy subjects and patients with established cardiovascular and cerebrovascular diseases (CVDs) has been proposed to better understand this phenomenon [6]. The authors found that in some patients the inhibition of platelet TxA2 formation, that has not been achieved by aspirin intake (in vivo), requires further in vitro addition of aspirin (pharmacokinetic resistance); conversely, in other patients aspirin fails to inhibit platelet TxA2 both in vivo and in vitro (pharmacodynamic resistance). Finally, a third group of patients displays platelet hyperaggregability despite low levels of TxA2 (pseudo-resistance).

We suggested that ‘aspirin resistance’ should identify those patients whose platelets, despite aspirin treatment, still produce TxA2 and could, thus, be more responsive to agonist stimulation [7]. The clinical implications of a persistent thromboxane production has been suggested by Eikelboom et al. [8] by demonstrating that patients with elevated urinary excretion of 11-dehydro-TxB2 were at higher risk of cardiovascular events. However, the authors have not established whether the observed TxA2 persistent production was of platelet or extraplatelet origin.

So far, the relationship between platelet-derived TxA2 and platelet function has never been investigated in a large population of long-term aspirin-treated patients.

Therefore, the aim of this study was to determine whether: (i) platelets from patients chronically treated with aspirin still produce TxA2; (ii) there is any relationship between the eventual persistent platelet TxA2 formation and platelet aggregation; (iii) such persistent TxA2 production can be inhibitable by in vitro addition of aspirin.

Materials and methods


Between September 2002 and June 2003, a total of 196 consecutive patients (60 males, 136 females, mean age 66.5 ± 12.6 years), with a history of coronary heart disease (8%), CVD (28%), or with one or more risk factors for atherothrombosis (hypertension 44%; dyslipidemia 36%; diabetes 5%; smoke 15%), who attended our laboratory to study platelet aggregation in order to monitor aspirin antiplatelet treatment (100–330 mg day−1) were enrolled in this study (Table 1). Treatment lasted 2–48 months (median 15 months). Patients were considered: hypertensive if blood pressure was >140/90 mmHg in three different measurements when patients were in a supine position for at least 10 min, or if they were treated with antihypertensive drugs; hypercholesterolemic if serum cholesterol was >240 mg dL−1 or if they were treated with lipid lowering agents; diabetics if serum glucose was >126 mg dL−1 or if on treatment with oral antidiabetics or insulin; smokers if they habitually smoked at least five cigarettes per day.

Table 1.  Baseline characteristic of patients in accordance to TxA2 quartiles
 I quartileII quartileIII quartileIV quartileTotal (n = 196)
  1. CVD, cerebrovascular disease; CHD, coronary heart disease.

Males (n)1319141460
Females (n)36303535136
Age, mean ± SD (years)68.2 ± 9.568.6 ± 13.764.8 ± 13,866.0 ± 13.166.5 ± 12.6
Clinical conditions
Hypertension (n)2123212186
Dyslipidemia (n)2316161671
Diabetes (n)222410
Smoke (n)489829
CVD (n)1316161156
CHD (n)444416
Antihypertensive (n)2123202084
Lipid lowering (n)1714141459
Antidiabetics (n)222410
Diuretics (n)652619
Aspirin dosage
100 mg day−1 (n)35323435136
160 mg day−1 (n)1010111041
330 mg day−1 (n)474419

In accordance to our previous paper [4], patients were excluded following these criteria: (i) concurrent therapy with any other drug known to interfere with platelet function (NSAID, ticlopidine, clopidogrel); (ii) presence of acute coronary syndromes or coronary intervention in the previous 3 months; (iii) patients with known platelet dysfunction; (iv) in vitro platelet aggregation in response to arachidonic acid (AA) higher than 10%.

During the time of the enrolment, 18 patients were excluded because they presented an elevated AA-induced platelet aggregation (33.8% ± 4.2%) and, as not being hospitalized, we could not exclude patient non-compliance.

Platelet aggregation and TxA2 formation

Ex vivo studies

Blood samples anticoagulated with Na-citrate (ratio 9:1) were taken from each patient (within 12–18 h from the last aspirin intake) after 12 h of fasting to study collagen- and AA-induced platelet aggregation and TxA2 production. Platelet-rich plasma (PRP) was prepared by centrifugation at 180  g for 15 min at room temperature. In order to minimize the presence of white blood cells (WBC), PRP was further centrifuged at 180  g for 5 min. Platelet concentration was determined in a hemocytometer (Celltac Auto, Sebia, Italy). A WBC count above 0.5 × 103 cells μL−1 was considered as an exclusion criterion. The mean number of WBC observed in our samples was 0.185 ± 0.015 × 103 cells μL−1 that was equally distributed among the quartiles of platelet TxA2 production. All samples were processed within 1 h after collection.

Platelet aggregation (Born's method) was evaluated on PRP as previously described [4], considering the maximal percentage in response to 2 μg mL−1 collagen and 1 mm AA, calculated as light transmission difference between PRP and platelet poor plasma (PPP) after 5 min of stimulation with the agonists. In response to collagen, the lag-phase, defined as the delay time occurring between the addition of collagen and the beginning of the aggregation curve, was also considered.

Platelet agonists as well as the aggregometer (APACT 4) were from Helena BioSciences (Sunderland, UK); concurrent controls were performed to ensure that all agonists retained the same level of activity during the whole study.

After 5 min of collagen-induced platelet activation, indomethacin (10 μm) was added to samples that were then centrifuged (8000 g for 1 min). TxB2, a stable metabolite of TxA2, was measured in the supernatant by an EIA commercial kit (Amersham Pharmacia, Biotech, Little Chalfont, UK), according to manufacturer instructions. All samples were assayed in duplicate and those showing values above the standard curve were re-probed with appropriate dilutions.

In order to assess the possible interference of AA on the immunoassay measurement, we added AA to PPP samples. All the values obtained were below the detectable range, confirming a negligible interference of AA immunoreactivity on the immunoassay.

In vitro studies

In 96 (31 males, 65 females, mean age 68.5 ± 10.4 years) out of 196 patients who were representative of the entire population in terms of risk factors and antiatherothrombotic therapy, collagen-induced platelet aggregation and TxA2 formation were also measured after incubation with the TxA2 receptor antagonist, 13-azaprostanoic acid (APA) (30 μm, 3 min at 37 °C), or with the COX-1 inhibitor aspirin (50 μm, 10 min at 37 °C). Samples pretreated only with the inhibitors’ solvent were used as controls.

In order to investigate the interplay between platelet TxA2 and AA-induced aggregation, TxA2 production and platelet aggregation were measured, as above reported, in platelets taken from seven healthy volunteers (three males, four females, age 61.3 ± 7.5 years) and incubated with scalar concentrations of aspirin (3–50 μm 10 min at 37 °C).

To verify whether the small amount of TxA2 produced in aspirin-treated patients is sufficient to enhance collagen induced-platelet aggregation, scalar doses of U46619, a synthetic agonist of the thromboxane receptor, were added to aspirin-treated (100 μm) platelets, from five healthy volunteers, activated with subthreshold collagen concentration. For this purpose, the concentrations of U46619 chosen were comprised between the median and the highest value (0.5–10 nm) of TxA2 obtained in the entire cohort.

Statistical analysis

All data are reported as mean ± SEM and, where appropriate, as median and ranges. Non-parametric Spearman's correlation coefficients were used to assess the relationship of thromboxane formation with collagen lag-phase and collagen maximal percentage of the entire cohort.

In in vitro studies, the differences between with and without the inhibitors were analyzed using non-parametric tests such as Kruskal–Wallis and Wilcoxon signed-rank. Significance was accepted at P < 0.05 level. All tests performed were two-sided. Data were analyzed using Stata version 6.0 (College Station, TX, USA; Stata Corporation, 1999).


Ex vivo studies

In the entire cohort (196) of patients, collagen-induced platelet TxA2 production was 128.7 ± 21.6 pg 10−8 cells, with values ranging from 0.5 to 616.8 pg 10−8 cells and a median value of 41.1 pg 10−8 cells. A positive, statistically significant, correlation was found between collagen-induced TxA2 formation and maximal percentage of platelet aggregation (ρ = 0.44; P < 0.0001).

Therefore, patients were divided into quartiles defined by the distribution of the TxA2 values. The results (Fig. 1) confirmed an association between either the lag-phase (panel A) or the maximal percentage (panel B) of platelet aggregation and TxA2 production. These findings were not influenced by aspirin dosage or by concomitant antiatherothrombotic treatment. In fact, the different aspirin dosages (100, 160 or 330 mg per die) as well as the risk factors for atherosclerosis were equally distributed among the quartiles (Table 1). Table 2 reports collagen-induced platelet aggregation and TxB2 values divided according to the different aspirin dosages.

Figure 1.

Association between TxA2 production and platelet aggregation. Box plot showing the association between quartiles of TxA2 production and either the lag-phase (panel A) or the maximal percentage (Mx%) (panel B) of platelet aggregation after collagen (2 μg mL−1) stimulation in patients (n = 196) treated with aspirin. *P < 0.01; **P < 0.002; ***P < 0.005; ****P < 0.0002; NS = not significant.

Table 2.  Collagen-induced platelet aggregation and TxA2 production in patients treated with different aspirin doses
Aspirin dosagePlatelet aggregation (Mx%)TxB2 production pg 10−8 cells
100 mg day−153.2 ± 2.4120.19 ± 14.4
160 mg day−149.0 ± 4.1116.42 ± 27.0
330 mg day−154.8 ± 5.5122.60 ± 40.9

In vitro studies

In the subgroup of 96 patients, collagen-induced TxA2 production (mean value 141.2 ± 17.7 pg 10−8 cells, median value 40.3 pg 10−8 cells) was similar to that found in the entire cohort (mean value 128.7 ± 21.6 pg 10−8 cells, median value 41.1 pg 10−8 cells).

Compared with untreated platelets, 13-APA acid induced a statistically significant inhibition of platelet aggregation (lag phase: 86.9 ± 3.5 s vs. 49.7 ± 2.2 s; Mx% 27.0 ± 1.6 vs. 49.9 ± 2.5) (Fig. 2) but did not affect TxA2 production (data not shown).

Figure 2.

Effect of TxA2 receptor antagonist on platelet aggregation. Platelet aggregation (mean ± SEM) analyzed as lag-phase and maximal aggregation in response to collagen (2 μg mL−1) in untreated platelets or in platelets (n = 96) incubated with the TxA2 receptor antagonist 13-APA. Significance between two groups was assessed using Wilcoxon's signed-rank test. *P < 0.0001.

Further evidence of a role of TxA2 in synergizing with collagen in causing platelet aggregation derives from experiments in which we added scalar concentrations (0.5–10 nm) of U46619 to aspirin-treated platelets from healthy volunteers, stimulated with subthreshold concentration of collagen. Figure 3 shows an evident, dose-dependent, platelet response when U46619 was used in combination with collagen. U46619, at the concentrations employed, was unable per se to elicit platelet aggregation (data not shown).

Figure 3.

Effect of U46619 on collagen-induced platelet aggregation. Platelet aggregation induced by scalar doses (0.5–10 nm) of U46619 added to subthreshold concentration of collagen in aspirin-treated platelets from healthy subjects. The results are expressed as the mean ± SEM (n = 5) of the maximal aggregation obtained.

To better clarify whether the persistent platelet TxA2 production depends on a residual COX-1 activity, platelets have been further treated in vitro with aspirin (50 μm).

Incubation of PRP with aspirin significantly reduced collagen-induced TxA2 production when compared with control (77.1 ± 14.8 pg 10−8 cells vs. 141.2 ± 19.1 pg 10−8 cells, P < 0.001). However, aspirin was ineffective in patients with TxA2 values below median (17.7 ± 2.1 pg 10−8 cells vs. 14.8 ± 1.7 pg 10−8 cells), while it significantly inhibited TxA2 (−72.9%) in patients with TxA2 values above median (66.9 ± 16.7 pg 10−8 cells vs. 246.7 ± 26.5 pg 10−8 cells; P < 0.0001) (Fig. 4).

Figure 4.

Effect of in vitro addition of aspirin on TxA2 production. Collagen-induced (2 μg mL−1) TxA2 production in control platelets and in platelets treated with aspirin in the population below and above the median value (40.3 pg 10−8 cells) of the TxA2 (see Materials and methods). *P < 0.0001; NS = not significant by Kruskall–Wallis and Wilcoxon's signed-rank test.

Arachidonic acid-induced platelet aggregation and TxA2 formation were measured in order to verify whether these tests could be considered useful tools to assess aspirin compliance. Platelets from healthy volunteers were thus incubated with scalar concentrations of aspirin (3–50 μm) prior to AA stimulation. A TxA2 production inhibited by >90% caused the suppression of platelet aggregation; conversely, an inhibition of TxA2 < 90% was associated with only a partial reduction of AA-induced platelet aggregation (Fig. 5).

Figure 5.

Aspirin abolishes AA-induced platelet aggregation when TxA2 production is inhibited by more than 90%. Effect of scalar concentrations of aspirin (3–50 μm) in reducing AA-induced (1 mm) platelet aggregation. The results are reported as the percentage of inhibition of the maximal aggregation (84.7% ± 4.2%) obtained in seven healthy subjects. For each concentration of aspirin, AA-induced TxA2 production (expressed as mean ± SEM of pg 10−8 cells) is reported.


This study provides evidence that patients under chronic treatment with aspirin present a wide inter-individual variability in terms of platelet sensitivity to inhibitory action of the drug and that this phenomenon depends on an insufficient inhibition of COX-1. In fact, in a certain number of patients, platelets still retain the capability to produce small amounts of TxA2 that is functionally relevant for platelet aggregation. The following observations are in favor of this hypothesis. Firstly, platelets in the third and fourth quartiles of TxA2 were significantly more responsive to collagen stimulation in that the lag phase was reduced and the maximal percentage was enhanced, when compared with platelets in the first and second quartiles. Secondly, in vitro pretreatment with a TxA2 receptor antagonist significantly inhibited collagen-induced platelet aggregation. Thirdly, small amounts of a TxA2 analog, U46619, ranging between the median and the highest values observed in our patient population, were able to induce platelet aggregation by synergizing with subthreshold collagen concentration.

An important implication of these results is that in case of persistent production of platelet TxA2, even low levels of this molecule may potentiate platelet activation. Indeed, in our cohort, values of TxA2 > 40 pg 10−8 cells identified patients whose platelets better responded to collagen, when compared with those with values <40 pg 10−8 cells. This interpretation is in accordance with a previous study that outlined the functional relevance of even a minimal production of TxA2 in the process of platelet activation [9]. Moreover, Di Minno et al. [10] demonstrated that early recovery of small amount of TxA2 after aspirin intake was associated with an enhanced platelet response to collagen.

Thus, the observed persistent platelet production of TxA2 in aspirin-treated patients prompted us to investigate whether this was dependent on incomplete or abnormal aspirin interaction with COX-1.

Our study suggests that, in a certain number of patients, aspirin intake is not sufficient to completely inhibit COX-1. In fact, in vitro addition of aspirin significantly reduced TxA2 formation by 72.9% in samples expressing TxA2 values above median, while, as expected, no additional inhibitory effect was obtained in samples with TxA2 values below median. The reason why aspirin does not completely inhibit COX-1 is a difficult issue and is beyond the aim of this study. However, as our data seem to exclude an abnormal interaction between aspirin and COX-1, a reduced bioavailability of aspirin in chronically treated patients should be considered.

Another possibility is that, in some patients, platelets over-express COX-1 as suggested by a recent, preliminary in vitro study [11].

A role for COX-2, that is insensitive to aspirin, in platelet TxA2 production has been suggested [12]; in this view, the incomplete inhibition of TxA2 formation by in vitro treatment with aspirin in our patients might depend on platelet COX-2 activation. However, expression of COX-2 by platelets as well as its relevance in terms of platelet TxA2 production is still a matter of debate [12,13] that deserves further investigation.

A potential limitation of this study was the lack of a direct proof of patient compliance to aspirin. However, we consider the use of AA-induced platelet aggregation useful, albeit indirect, evidence concerning aspirin intake. In fact, the in vitro demonstration that AA-induced platelet aggregation was abolished only when TxA2 production was inhibited by more than 90% could imply that our selected population was actually taking aspirin. The exclusion of the 18 patients whose platelets responded to AA stimulation at the time of the enrolment was, indeed, because of the fact that we could not exclude patient non-compliance. It should be also underscored that in our patients, despite AA-induced platelet aggregation was abolished, the residual TxA2 was still relevant for collagen-induced platelet aggregation. This could be dependent on the fact that while this residual TxA2 is probably not sufficient to induce platelet aggregation following AA stimulation, behaves as a synergistic agonist when platelets are activated by collagen.

On the basis of these considerations, we can conclude that some patients under chronic treatment with aspirin, display a persistent production of TxA2 that is mainly dependent upon impaired COX-1 inactivation and is responsible for enhanced platelet aggregation. The inter-individual variability of platelet response to aspirin may be a point of interest in the development of both future pharmacological studies and in new clinical trials concerning the efficacy of antiplatelet treatment in reducing cardiovascular events.


The authors thank Prof. Giovanni De Gaetano and Dr Chiara Cerletti for their critical review of the manuscript. This paper was partially supported by a Grant from the Italian Ministry of Education, University and Scientific Research (MIUR) ‘ex 60% Ateneo 2002, to F.M.P’ and by a grant from Italian Ministry of Health to F.V.