Aspirin ‘resistance’: role of pre-existent platelet reactivity and correlation between tests

Authors


Alan D. Michelson, Center for Platelet Function Studies, University of Massachusetts Medical School, Room S5-846, 55 Lake Avenue North, Worcester, MA 01655, USA.
Tel.: +1 508 856 0056; fax: +1 508 856 4282.
E-mail: michelson@platelets.org

Abstract

Summary. Background: Aspirin ‘resistance’ is a widely used term for hyporesponsiveness to aspirin in a platelet function test. Serum thromboxane (TX) B2 is the most specific test of aspirin’s effect on platelets. Objectives: (i) To examine the role of pre-existent platelet hyperreactivity in aspirin ‘resistance’. (ii) To determine the correlation between aspirin resistance defined by serum TXB2 and other assays of platelet function. Methods: To enable pre-aspirin samples to be drawn, platelet function was measured in normal subjects (n = 165) before and after aspirin 81 mg daily for seven days. Results: The proportion of the post-aspirin platelet function predicted by the pre-aspirin platelet function was 28.3 ± 7.5% (mean ± asymptotic standard error) for serum TXB2, 39.3 ± 6.8% for urinary 11-dehydro TXB2, 4.4 ± 7.7% for arachidonic acid-induced platelet aggregation, 40.4 ± 7.1% for adenosine diphosphate-induced platelet aggregation, 26.3 ± 9.2% for the VerifyNow Aspirin Assay®, and 45.0 ± 10.9% for the TEG® PlateletMapping™ System with arachidonic acid. There was poor agreement between aspirin-resistant subjects identified by serum TXB2 vs. aspirin-resistant subjects identified by the other five assays, irrespective of whether the analysis was based on categorical or continuous variables. Platelet count correlated with pre-aspirin serum TXB2 and VerifyNow Aspirin Assay, but not with any post-aspirin platelet function test. Conclusions: (i) Aspirin ‘resistance’ (i.e. hyporesponsiveness to aspirin in a laboratory test) is in part unrelated to aspirin but is the result of underlying platelet hyperreactivity prior to the institution of aspirin therapy. (ii) Aspirin resistance defined by serum TXB2 shows a poor correlation with aspirin resistance defined by other commonly used assays.

Introduction

Platelet activation plays an important role in acute myocardial infarction, the most common cause of death in Western society [1]. Aspirin, an inhibitor of platelet activation, has therefore been demonstrated to have a beneficial antithrombotic effect [2]. Approximately 36% of the adult population of the United States (more than 50 million people) take aspirin regularly for the prevention of cardiovascular disease [3]. The possibility that some people are hyporesponsive or ‘resistant’ to aspirin is therefore potentially of enormous importance to public health, and ‘aspirin resistance’ has become a topic of intense interest [4–11]. Although the definition of this term unfortunately remains non-standardized [4–8], aspirin resistance refers to the less than expected inhibition of a platelet function test by aspirin. The mechanism of action of aspirin’s effect is irreversible acetylation of serine 529 of cyclooxygenase (COX)-1, resulting in inhibition of the activation-dependent generation of the platelet agonist thromboxane (TX) A2 [12]. As recently summarized by Cattaneo [8], the term ‘resistance’ to a drug should optimally be used when a drug is unable to hit its pharmacological target, as a result of inability to reach it (as a consequence of reduced bioavailability, in vivo inactivation, or negative interactions with other substances) or alterations of the target. Based on this definition, the term ‘aspirin resistance’ should be limited to situations in which aspirin is unable to inhibit COX-1-dependent TXA2 production [7,8]. Serum TXB2, the stable metabolite of TXA2, is therefore the most definitive test of ‘aspirin resistance’ [4–8]. Even when arachidonic acid, the precursor of TXA2, is used as a platelet agonist, the results may overestimate the prevalence of aspirin resistance [7,8]. The concentration of 11-dehydro TXB2, a urinary metabolite of TXA2, may be partly dependent on extraplatelet sources of TXA2 [13]. Nevertheless, ‘aspirin resistance’ has been measured by assays other than serum TXB2, including urinary 11-dehydro TXB2 [14], arachidonic acid induced platelet aggregation [15], adenosine diphosphate (ADP)-induced platelet aggregation [15], VerifyNow® Aspirin Assay [16], and the thromboelastogram (TEG) Platelet Mapping System® [17]. One goal of the present study was therefore to determine the correlation between aspirin resistance defined by serum TXB2 and these other assays.

Published evidence suggests that causes of aspirin ‘resistance’ may include: non-compliance; underdosing; poor absorption; interference by other non-steroidal anti-inflammatory drugs; accelerated platelet turnover with introduction into the blood stream of newly formed, drug-unaffected platelets; stress-induced generation of COX-2 in platelets; single nucleotide polymorphisms in COX-1 and/or other molecules; and bypass of platelet COX-1 by endothelial cell or monocyte COX-1 and/or inducible COX-2 [4,5]. However, we have hypothesized [18] that an additional potentially important but poorly studied possibility is that ‘aspirin resistance’ is, at least in part, the result of an underlying platelet hyperreactivity that predates the aspirin therapy [4,5]. Indeed, in the comparable setting of hyporesponsiveness or ‘resistance’ to clopidogrel, we have demonstrated that at least part of this phenomenon is the result of an underlying variability in platelet response that is not increased by clopidogrel administration [19].

However, a major limitation of virtually all studies reporting an association between aspirin ‘resistance’ (i.e. hyporesponsiveness to aspirin in a platelet function test) and poor clinical outcomes [10,11] is that they have not truly measured platelet responsiveness to aspirin, but rather the platelet function of patients already taking aspirin. Therefore, the second major goal of the present study was to examine the hypothesis that aspirin ‘resistance’ is, at least in part, the result of an underlying platelet hyperreactivity prior to the institution of aspirin therapy.

To address these two study goals, platelet function – as determined by serum TXB2, urinary 11-dehydro TXB2, arachidonic acid-induced platelet aggregation, ADP-induced platelet aggregation, VerifyNow Aspirin Assay, and the TEG PlateletMapping System – was evaluated pre- and post-aspirin treatment. To enable pre-aspirin samples to be drawn, this study was performed in 165 normal subjects rather than in patients, because virtually all clinically relevant patients are already taking aspirin. An additional advantage of the use of normal subjects rather than patients is that the platelet response to stimuli of normal subjects is not influenced (with resultant increased scatter of the data) by an underlying disease (e.g. coronary artery disease), which is well known to cause platelet hyperreactivity [1,20,21].

Methods

Study population

After evaluation and written approval of the study by the University of Massachusetts Medical School Investigational Review Board (IRB), written informed consent was obtained from each study subject. Normal subjects (n = 167) who had not taken aspirin or a thienopyridine during the prior 10 days, or a non-steroidal anti-inflammatory drug (NSAID) during the prior three days, and who had a normal complete blood count, were given a St Joseph’s aspirin (McNeil Consumer Healthcare) 81 mg orally daily for one week. Subjects with known allergy or hypersensitivity to aspirin were excluded. For each enrolled subject, a clinical history and a 10-day medication history were recorded on a case report form. Two subjects did not complete the seven-day course of aspirin and were not studied further, leaving 165 subjects in the study.

Assays

In the 165 normal subjects, urine and peripheral venous blood were obtained on day 0 (pre-aspirin) and day 7 (by which time steady-state inhibition by aspirin 81 mg daily is achieved [22]). The blood was drawn into Becton Dickinson (BD Biosciences, San Jose, CA, USA) evacuated tubes containing either 3.2% (0.105 m) sodium citrate or no anticoagulant. The non-anticoagulated tube was incubated at 37 °C for 1 h, and then the serum was separated by centrifugation at 1500 × g for 15 min and then stored at −80 °C until analysis for serum TXB2 by an ELISA kit (R&D Systems, Minneapolis, MN, USA) according to the manufacturer’s recommendations. Platelet aggregation with 0.5 mg mL−1 arachidonic acid and 10 μm ADP was performed on platelet-rich plasma obtained from the citrated blood (without adjustment of the platelet count), as previously described [15].

Urine samples were stabilized by addition of Chlorstat tablets (Bio-Medical Products Corp., Mendham, NJ, USA), then frozen at <−20 °C until analysis. Urinary 11-dehydro TXB2 was assayed by the AspirinWorks® Test Kit (Corgenix Inc., Westminster, CO, USA), as described in the product insert. Urinary 11-dehydro TXB2 results were normalized to urinary creatinine to generate the urinary 11-dehydro TXB2/creatinine ratio (expressed in ng mmol−1), as previously described [14].

One subject had a pre-aspirin (day 0) serum TXB2 of 10.5 ng mL−1 and a post-aspirin (day 7) serum TXB2 of 1.3 ng mL−1. This subject was assumed to be taking aspirin or another NSAID prior to enrollment in the study and was therefore excluded from analysis. Thus, 164 subjects were analyzed for serum TXB2, urinary 11-dehydro TXB2, arachidonic acid-induced platelet aggregation, and ADP-induced platelet aggregation. There were 73 males and 91 females, with an age of 32.8 ± 11.3 (mean ± SD) years.

In a subgroup of 99 consecutive normal subjects (47 males, 52 females, age 32.0 ± 9.3 years), the VerifyNow® Aspirin Assay (Accumetrics, San Diego, CA, USA) was used according to the manufacturer’s recommendations.

In a subgroup of 49 consecutive normal subjects (26 males, 23 females, age 32.3 ± 8.6 years), the TEG® PlateletMapping System (Haemoscope, Niles, IL, USA) was used with arachidonic acid as an agonist, as previously described [17]. Results are expressed as the percent of a control of kaolin-treated, recalcified, citrated whole blood (which generates maximal thrombin-induced clot formation).

Statistical analysis

Sample size justification was performed prior to study initiation. To achieve statistical significance for detecting a 10% difference in variability between groups above typical scatter in the data (coefficients of variation [CVs] for these assays are estimated to be ∼15 ± 4%) given the proportional distribution between groups, power analysis [with α = 0.05 and a statistical power of 0.90 (i.e. β = 0.10)] indicated that a total of 145 subjects would need to be entered into the study. To take into account subject drop-out, non-compliance, and the possibility that results would not be normally distributed (which would require a slightly larger sample for proper statistical analysis), an additional 15% were planned to be tested – for a total of 167 subjects.

Statistical analyses were performed with SAS® 9.1.3 (SAS Institute, Inc., Cary, NC, USA). Graphs were prepared with Prism® 4.00 (GraphPad Software, San Diego, CA, USA). The results of each assay, pre-and post-aspirin treatment were compared by Wilcoxon signed rank test and correlated by Spearman’s rank method. Spearman’s rank correlation was also used for inter-assay comparison of results post-aspirin treatment. A SAS macro was used to calculate sensitivity and specificity.

Results

Relationship between platelet response variability pre-aspirin and platelet response variability post-aspirin

Figure 1 shows the relationship between platelet response variability pre-aspirin and platelet response variability after seven days of aspirin 81 mg daily, as determined by six different tests of platelet function. The degree to which pre-aspirin platelet function predicts post-aspirin platelet function for each test is shown by the correlation coefficients displayed in Table 1. Thus, the proportion of the post-aspirin platelet function predicted by the pre-aspirin platelet function was 28.3 ± 7.5% (mean ± asymptotic standard error) for serum TXB2, 39.3 ± 6.8% for urinary 11-dehydro TXB2, 4.4 ± 7.7% for arachidonic acid-induced platelet aggregation, 40.4 ± 7.1% for ADP-induced platelet aggregation, 26.3 ± 9.2% for the VerifyNow Aspirin Assay, and 45.0 ± 10.9% for the TEG PlateletMapping System with arachidonic acid as the agonist (Table 1).

Figure 1.

 The relationship between platelet response variability pre-aspirin and platelet response variability post-aspirin. Each dot represents one subject. The lines are linear regression plots. n = 164 subjects for serum thromboxane (TX) B2, urinary 11-dehydro TXB2, arachidonic acid-induced platelet aggregation, and adenosine diphosphate-induced platelet aggregation. n = 99 subjects for VerifyNow Aspirin Assay. n = 49 for TEG PlateletMapping System.

Table 1.   Wilcoxon signed rank test and Spearman correlation analysis of pre-aspirin vs. post-aspirin results in six different aspirin-sensitive assays
TestN Pre-ASA (mean ± SD) Post-ASA (mean ± SD)P (Rank T)Correlation
coefficient (mean ± ASE)
P (correlation)
  1. AA, arachidonic acid; ADP, adenosine diphosphate; ARU, aspirin resistance units; ASA, aspirin; ASE, asymptotic standard error; LTA, light transmittance aggregometry; TEG, thromboelastogram; TXB2, thromboxane B2.

Serum TXB2 (ng mL−1)164376.9 ± 173.83.6 ± 16.0<0.0010.283 ± 0.075<0.001
Urinary 11-dehydro TXB2 (ng mmol−1 creatinine)164105.6 ± 49.829.2 ± 20.9<0.0010.393 ± 0.068<0.001
LTA (AA) (%)16476.7 ± 10.83.7 ± 6.4<0.0010.044 ± 0.0770.646
LTA (ADP) (%)16479.0 ± 11.666.0 ± 11.6<0.0010.404 ± 0.071<0.001
VerifyNow® Aspirin Assay (ARU)99656.7 ± 13.3425.9 ± 40.8<0.0010.263 ± 0.0920.008
TEG PlateletMapping System® AA – induced Clot Strength (% of control)4984.3 ± 13.913.7 ± 17.8<0.0010.450 ± 0.1090.002

Relationship between aspirin resistance defined by post-aspirin serum TXB2 and aspirin resistance defined by five other assays post-aspirin

Consistent with recommendations in the literature [4–8], we prospectively considered serum TXB2 to be the most definitive marker of aspirin ‘resistance’, because aspirin specifically inhibits COX-1 and therefore directly inhibits the generation of TXA2, and its stable metabolite TXB2. However, there is no standard definition in the literature for the specific cutoff of serum TXB2 concentration to define ‘aspirin resistance’. In this study, we therefore used two different methods to define the TXB2 cutoff.

First, we used a receiver operator characteristic (ROC) analysis, which has the advantage that it specifically applies to our study’s subject population and aspirin dose. Based on this ROC curve, a post-aspirin TXB2 cutoff of >12 ng mL−1 was selected: sensitivity 100% [95% confidence interval (CI) 97.8–100]; specificity 98% (95% CI 93.5–99.2). Using this cutoff, there were 159 aspirin sensitive subjects (Fig. 2A, red squares) and only five aspirin-resistant subjects (Fig. 2A, black squares) – an incidence of aspirin resistance of 3.0%. Based on this definition of aspirin resistance (serum TXB2 > 12 ng mL−1), the sensitivity and specificity of the other five assays of platelet function are shown in Table 2. Because of the low number of aspirin-resistant subjects as defined by serum TXB2 >12 ng mL−1, point estimates of sensitivities and specificities are not provided in Table 2.

Figure 2.

 The relationship between aspirin resistance defined by post-aspirin serum thromboxane (TX) B2 >12 ng mL−1 and aspirin resistance defined by five other assays post-aspirin. Subjects were tested after receiving aspirin 81 mg daily for seven days. Aspirin-sensitive (below blue line) and aspirin-resistant (above blue line) subjects were defined by the cutoffs listed in the text. In all panels, the red squares indicate subjects who are aspirin-sensitive as defined by a serum TXB2 <12 ng mL−1 and the black squares indicate subjects who are aspirin-resistant as defined by a serum TXB2 >12 ng mL−1. (A)–(D) n = 164; (E) n = 99; (F) n = 49.

Table 2.   Sensitivity and specificity of tests for the identification of inadequate inhibition of platelet cyclooxygenase (COX)-1, defined as serum thromboxane (TX) B2 >12 ng mL−1
TestNNSensitivity (95% CI)Specificity (95% CI)
  1. AA, arachidonic acid; ADP, adenosine diphosphate; CI, confidence interval; LTA, light transmission aggregometry; N, total number of subjects analyzed; n, number of aspirin-resistant subjects as defined by serum TXB2 >12 ng mL−1; TEG, thromboelastogram.

Urinary 11-dehydro TXB216450.01–0.720.95–1.00
LTA (AA)16450.01–0.720.97–1.00
LTA (ADP)16450.28–0.990.52–0.67
VerifyNow Aspirin9920.01–0.990.93–1.00
TEG Platelet Mapping (AA)4920.01–0.990.92–1.00

Secondly, we used a post-aspirin TXB2 cutoff of >2.2 ng mL−1, based on the published data of Maree et al. [23], resulting in a sensitivity of 100% (95% CI 97.8–100.0) and a specificity of 76.2 (95% CI 69.0–82.5) in our ROC curve. Based on this cutoff, there were 125 aspirin-sensitive subjects (Fig. 3A, red squares) and 39 aspirin-resistant subjects (Fig. 3A, black squares) – an incidence of aspirin resistance of 23.8%. Based on this definition of aspirin resistance (serum TXB2 > 2.2 ng mL−1), the sensitivity and specificity of the other five assays of platelet function are shown in Table 3.

Figure 3.

 The relationship between aspirin resistance defined by post-aspirin serum thromboxane (TX) B2 >2.2 ng mL−1 and aspirin resistance defined by five other assays post-aspirin. All data are identical to those in Fig. 2, except that in all panels of Fig. 3 the red squares indicate subjects who are aspirin-sensitive as defined by a serum TXB2 <2.2 ng mL−1 and the black squares indicate subjects who are aspirin-resistant as defined by a serum TXB2 >2.2 ng mL−1.

Table 3.   Sensitivity and specificity of tests for the identification of inadequate inhibition of platelet cyclooxygenase (COX)-1, defined as serum thromboxane (TX) B2 >2.2 ng mL−1
TestNnSensitivity (95% CI)Specificity (95% CI)
  1. AA, arachidonic acid; ADP, adenosine diphosphate; CI, confidence interval; LTA, light transmission aggregometry; N, total number of subjects analyzed; n, number of aspirin-resistant subjects as defined by serum TXB2 >2.2 ng mL−1; TEG, thromboelastogram.

Urinary 11-dehydro TXB2164380.001–0.140.93–1.00
LTA (AA)164380.001–0.130.96–1.00
LTA (ADP)164380.13–0.420.48–0.66
VerifyNow Aspirin99250.001–0.200.90–1.00
TEG Platelet Mapping (AA)49140.018–0.430.90–1.00

Figure 2 illustrates the relationship between aspirin resistance defined by post-aspirin serum TXB2 >12 ng mL−1 and aspirin resistance defined by five other assays post-aspirin. Aspirin-sensitive (below blue line) and aspirin-resistant (above blue line) subjects were defined by the following cutoffs: serum TXB2 12 ng mL−1 (Fig. 2A), based on our ROC analysis; urinary 11-dehydro TXB2 67 ng mmol−1 creatinine (Fig. 2B), based on Lordkipanidze et al. [24] [an alternative cutoff would be 170 ng mmol−1 (1500 pg mg−1), as recommended in the manufacturer’s product insert]; 20% arachidonic acid-induced platelet aggregation (Fig. 2C), based on Gum et al. [15]; 70% ADP-induced platelet aggregation (Fig. 2D), also based on Gum et al. [15]; 550 aspirin resistance units (ARUs) in the VerifyNow Aspirin Assay (Fig. 2E), based on the manufacturer’s recommendations; and 50% of control clot strength in the TEG PlateletMapping System with arachidonic acid, based on Tantry et al. [17] (Fig. 2F). In each panel of Fig. 2, the red squares indicate subjects who are aspirin-sensitive as defined by a serum TXB2 <12 ng mL−1 and the black squares indicate subjects who are aspirin-resistant as defined by a serum TXB2 >12 ng mL−1. Aspirin resistance defined by serum TXB2 >12 ng mL−1 shows a poor correlation with aspirin resistance defined by the other five platelet function assays, as illustrated by the high proportion of black squares below the blue lines in Figs 2B–F.

The data in Fig. 3 are identical to those in Fig. 2, except for the following: (i) the serum TXB2 cutoff in Fig. 3A is 2.2 ng mL−1 rather than 12 ng mL−1; (ii) in all panels of Fig. 3, the red squares indicate subjects who are aspirin-sensitive as defined by a serum TXB2 <2.2 ng mL−1 and the black squares indicate subjects who are aspirin-resistant as defined by a serum TXB2 >2.2 ng mL−1. Aspirin resistance defined by serum TXB2 >2.2 ng mL−1 shows a poor correlation with aspirin resistance defined by the other five platelet function assays, as illustrated by the high proportion of black squares below the blue lines in Figs 3B–F.

Correlations between aspirin resistance defined by six different platelet function tests

Table 1 shows that each of the six assays of platelet function is aspirin sensitive (compare pre-aspirin to post-aspirin). Figure 4 shows the relationship between the six different assays of platelet function for both pre-aspirin samples (black dots) and post-aspirin samples (red dots).

Figure 4.

 The relationship between pre-aspirin samples (black dots) and post-aspirin samples (red dots) as determined by six different platelet function assays. Dashed lines represent the receiver operator characteristic (ROC) curve-defined serum thromboxane B2 cutoff of 12 ng mL−1 and cutoffs for the other assays as defined in the text.

Inspection of Fig. 4 reveals that the relationship between some pairs of assays is non-linear. However, Spearman rank order correlations were highly significant for comparisons between assays when both pre-aspirin and post-aspirin results were included in the analysis (Table 4A). There was good separation between pre-aspirin and post-aspirin values (Fig. 4, panels in top row; black dots vs. red dots on horizontal axis) for serum TXB2, arachidonic acid-induced platelet aggregation, and the VerifyNow Aspirin Assay, but less so for urinary 11-dehydro-TXB2, TEG PlateletMapping System, and, especially, ADP-induced platelet aggregation.

Table 4.   Spearman correlation between platelet function tests
 Urinary 11-dehdro TXB2 [correlation coefficient (P)/N]LTA (AA) [correlation coefficient (P)/N]LTA (ADP) [correlation coefficient (P)/N]VerifyNow Aspirin [correlation coefficient (P)/N]TEG Platelet Mapping (AA) [correlation coefficient (P)/N]
  1. AA, arachidonic acid; LTA, light transmission aggregometry; N, total number of subjects analyzed; TXB2, thromboxane B2.

(A) Pre- and post-aspirin data
  Serum TXB20.733 (<0.001)/3280.769 (<0.001)/3280.523 (<0.001)/3280.820 (<0.001)/1990.752 (<0.001)/98
  Urinary 11-dehydro TXB2 0.748 (<0.001)/3280.515 (<0.001)/3280.772 (<0.001)/1990.727 (<0.001)/98
  LTA (AA)  0.646 (<0.001)/3280.820 (<0.001)/1990.760 (<0.001)/98
  LTA (ADP)   0.566 (<0.001)/1990.465 (<0.001)/98
  VerifyNow Aspirin    0.754 (<0.001)/96
(B) Post-aspirin data
  Serum TXB2−0.039 (0.618)/164−0.069 (0.377)/164−0.026 (0.741)/1640.170 (0.093)/990.045 (0.759)/49
  Urinary 11-dehydro TXB2 0.036 (0.644)/1640.090 (0.251)/1640.083 (0.417)/990.084 (0.564)/49
  LTA (AA)  0.477 (<0.001)/1640.378 (<0.001)/990.133 (0.361)/49
  LTA (ADP)   0.453 (<0.001)/990.087 (0.553)/49
  VerifyNow Aspirin    0.194 (0.186)/48

The poor agreement between aspirin-resistant subjects identified by serum TXB2 vs. aspirin-resistant subjects identified by the other five assays (Figs 2 and 3, Tables 2 and 3) may have been due, in part, to the relatively low frequency of aspirin ‘resistance’ as defined by the serum TXB2 cutoff (5/164 for 12 ng mL−1 and 38/164 for 2.2 ng mL−1) and the required categorical analysis. We therefore performed an additional analysis wherein, rather than categorizing subjects as aspirin-resistant or aspirin-sensitive, we used the continuous variable of residual platelet function in each post-aspirin assay to obtain the correlation matrix shown in Table 4B. Residual serum TXB2 levels post-aspirin treatment were not significantly correlated with any of the other assays (Table 4B, first row). Residual urinary 11-dehydro TXB2 levels post-aspirin treatment were also not significantly correlated with any of the other assays (Table 4B, second row). However, residual post-aspirin platelet functions measured by arachidonic acid-stimulated platelet aggregation, ADP-stimulated platelet aggregation, and VerifyNow Aspirin Assay all correlated with each other (Table 4B).

Effect of platelet count on platelet function assays

Platelet count has previously been reported to contribute to variation in serum TXB2 levels in non-aspirinated subjects [25]. The effect of platelet count on the pre-existent variation in the presently reported platelet function assays was therefore analyzed. Pre-aspirin platelet count showed a significant correlation with pre-treatment serum TXB2 (Spearman = 0.3575, CI 0.2117–0.4877, P < 0.0001) and pre-treatment VerifyNow Aspirin Assay (Spearman = 0.3379, CI 0.1457–0.5055, P = 0.0006), but did not correlate with pre-treatment urinary 11-dehydro TXB2, arachidonic acid-induced platelet aggregation, ADP-induced platelet aggregation, or TEG PlateletMapping assay. No correlation was found between the post-aspirin platelet count and any of the six platelet function tests. As expected, post-aspirin platelet count correlated with pre-aspirin platelet count (Spearman r = 0.7091, CI 0.6208–0.7797, P < 0.0001).

Non-compliance vs. aspirin absorption or metabolic problem

The subject with the outlying, highest post-aspirin arachidonic acid-induced platelet aggregation of 71% (Fig. 2C) was the same subject with the outlying, highest post-aspirin serum TXB2 of 200 ng mL−1 (Fig. 2A), and the highest post-aspirin VerifyNow ARU of 657 (Fig. 2E). This subject had a post-aspirin urinary 11-dehydro TXB2 of 98 ng mmol−1 creatinine and a post-aspirin ADP-induced platelet aggregation of 76.5% (but was not part of the subgroup of subjects analyzed by the TEG PlateletMapping System). Addition of in vitro aspirin 56 μm (a dose we have previously determined to be a tenfold excess of the usual dose required to inhibit arachidonic acid-induced platelet aggregation [26]) to the post-aspirin sample of this subject resulted in a reduction of arachidonic acid-induced platelet aggregation from 71% to 0% (average of two determinations). This subject was therefore either non-compliant with aspirin therapy or had an aspirin absorption or metabolic problem.

Addition of in vitro aspirin 56 μm to the post-aspirin sample of the subject with the second highest post-aspirin arachidonic acid-induced platelet aggregation of 20% (Fig. 2C) resulted in a reduction of arachidonic acid-induced platelet aggregation to 11% (average of two determinations). This subject was therefore either partially non-compliant with aspirin therapy or had an aspirin absorption or metabolic problem.

Discussion

The main findings in this study are as follows. (i) Aspirin ‘resistance’ (i.e. hyporesponsiveness to aspirin in a laboratory test) is in part unrelated to aspirin but is the result of underlying platelet hyperreactivity prior to the institution of aspirin therapy. (ii) The platelet count contributes to pre-aspirin platelet reactivity as measured by serum TXB2 and the VerifyNow Aspirin Assay, but does not contribute to post-aspirin platelet reactivity. (iii) Aspirin resistance defined by serum TXB2 shows a poor correlation with aspirin resistance defined by five other widely used assays: urinary 11-dehydro TXB2, arachidonic acid-induced platelet aggregation, ADP-induced platelet aggregation, VerifyNow Aspirin Assay, and the TEG PlateletMapping System.

Aspirin ‘resistance’: role of pre-existent platelet reactivity

We have previously demonstrated, in a study of 700 aspirin-treated patients with coronary artery disease, that residual platelet activation after the in vitro addition of arachidonic acid occurred in direct proportion to the degree of baseline platelet activation (i.e. no in vitro addition of arachidonic acid) [26]. This finding raised the question as to whether measurements of ‘aspirin resistance’ reflect, at least in part, baseline heterogeneity in platelet response [9]. Therefore, a major goal of the present study was to determine whether the variability in platelet response after aspirin is partly explained by a pre-existent variability in platelet response. To enable pre-aspirin samples to be drawn, this study was performed in normal subjects rather than in patients, because virtually all clinically relevant patients are already taking aspirin. An additional advantage of the use of normal subjects rather than patients is that the platelet response to stimuli of normal subjects is not influenced (with resultant increased scatter of the data) by an underlying disease (e.g. coronary artery disease), which is well known to cause platelet hyperreactivity [1,20,21]. Thus, the use of normal volunteers in this study rather than patients enabled us to distinguish true aspirin resistance from underlying platelet hyperreactivity, an important point of confusion in the literature [8]. For example, two recent meta-analyses of published studies provided evidence for an association of laboratory-defined aspirin resistance with a higher risk of recurrent cardiovascular events [10,11]. However, these meta-analyses and their 20 included studies did not distinguish between (i) underlying platelet hyperreactivity with resultant high residual platelet reactivity and (ii) true aspirin resistance. In the present study, by analyzing blood samples pre- and post-aspirin, we demonstrate that the phenomenon of ‘aspirin resistance’ is, in part, the result of underlying platelet hyperreactivity, and therefore unrelated to aspirin. The present study does not address the question as to whether or not a higher dose of aspirin, or other additional antiplatelet therapy, would be clinically beneficial in patients with high residual platelet reactivity after aspirin therapy. Definitive evidence in this regard will require a randomized trial that guides antiplatelet therapy based on the results of a platelet function test [4].

The proportion of post-aspirin platelet function predicted by pre-aspirin platelet function is 28.3 ± 7.5% (mean ± asymptotic standard error) for serum TXB2, 39.3 ± 6.8% for urinary 11-dehydro TXB2, 4.4 ± 7.7% for arachidonic acid-induced platelet aggregation, 40.4 ± 7.1% for ADP-induced platelet aggregation, 26.3 ± 9.2% for the VerifyNow Aspirin Assay, and 45.0 ± 10.9% for the TEG PlateletMapping System with arachidonic acid as the agonist (Table 1). (The explanation for the low correlation coefficient with arachidonic acid-induced platelet aggregation is that the post-aspirin samples were so inhibited that there is little variation in these values.) The factors that contribute to this pre-existent variability in platelet function are unclear. In this study, platelet count was found to contribute to pre-existent variability in two of the assays (serum TXB2 and the VerifyNow Aspirin Assay) but not to contribute to post-aspirin variability in any of the assays. A number of other factors have been shown to contribute to variations in platelet reactivity, including diet [27], platelet turnover [28], and single nucleotide polymorphisms (especially in platelet signaling molecules) [29]. However, further studies are needed to identify the specific factors that contribute to both pre-aspirin and post-aspirin platelet response variability.

Aspirin ‘resistance’: correlation between platelet function tests

As recently summarized by Cattaneo [8], the term ‘resistance’ to a drug should be used when a drug is unable to hit its pharmacological target, as a result of inability to reach it (as a consequence of reduced bioavailability, in vivo inactivation, or negative interactions with other substances) or alterations of the target. Based on this definition, the term ‘aspirin resistance’ should be limited to situations in which aspirin is unable to inhibit COX-1-dependent TXA2 production [7,8]. Even when arachidonic acid, the precursor of TXA2, is used as a platelet agonist, the results may overestimate the prevalence of aspirin resistance [7,8]. The concentration of 11-dehydro TXB2, a urinary metabolite of TXB2, may be partly dependent on extraplatelet sources of TXA2 [13]. Therefore, the most specific test for ‘aspirin resistance’ is serum TXB2, which directly reflects the capacity of platelets to synthesize TXA2, of which TXB2 is a stable metabolite [7,8]. For this reason, in the present study we prospectively considered serum TXB2 to be the most definitive marker of aspirin ‘resistance’, because aspirin specifically inhibits COX-1 and therefore directly inhibits the generation of TXA2, and its stable metabolite TXB2 [12]. However, there is no standard definition in the literature for the specific cutoff of serum TXB2 concentration to define ‘aspirin resistance’. In this study, we therefore used two different methods to define the TXB2 cutoff. First, we used an ROC curve-generated post-aspirin TXB2 cutoff of >12 ng mL−1, which has the advantage that it specifically applies to our study’s subject population and aspirin dose. Secondly, we used a post-aspirin TXB2 cutoff of >2.2 ng mL−1, based on the experimental data of Maree et al. [23]. In patients taking aspirin 75 mg daily, these investigators [23] demonstrated that when serum TXB2 levels exceed 2.2 ng mL−1, platelets continue to generate TX from exogenous arachidonic acid, indicating the presence of uninhibited COX-1 and therefore an incomplete response to aspirin.

Irrespective of the serum TXB2 cutoff (12 ng mL−1 or 2.2 ng mL−1), we found that those subjects defined as aspirin-resistant by serum TXB2 were not the same subpopulation of subjects who were identified as aspirin-resistant by other widely used assays: urinary 11-dehydro TXB2, arachidonic acid-induced platelet aggregation, ADP-induced platelet aggregation, VerifyNow Aspirin Assay, and the TEG PlateletMapping Assay [see Figs 2 and 3, in which the black squares in each panel indicate subjects who are aspirin-resistant defined by a serum TXB2 of either >12 ng mL−1 (Fig. 2) or >2.2 ng mL−1 (Fig. 3)]. Even when residual post-aspirin platelet function was assessed by correlation of continuous results rather than categorical analysis using a cutoff, none of the platelet function assays showed a significant correlation with the gold standard serum TXB2 (Table 4B). A number of previous studies [17,24,30–35] have shown a poor correlation between different assays for aspirin resistance but, unlike the present study, most of these studies did not use serum TXB2 as a point of comparison and, also unlike the present study, the dose of aspirin varied. Of these studies, only the small study of Gonzalez-Conejero et al. [32] and the large study of Becker et al. [34] examined whether the variability in platelet response after aspirin is partly explained by the pre-existent variability in platelet response; both found that pre-aspirin platelet function contributed to the variance in platelet reactivity after aspirin therapy. However, unlike the present study, neither the study by Gonzalez-Conejero et al. [32] nor the study by Becker et al. [34] measured serum TXB2.

Clinically relevant definitions of aspirin resistance can only be based on data linking laboratory tests to poor clinical outcomes in patients [4,10,11]. There is evidence from small clinical studies of an association between subsequent poor clinical outcomes in aspirin-treated patients and urinary 11-dehydro TXB2 [14,36], arachidonic acid-induced platelet aggregation [15,37], ADP-induced platelet aggregation [15,37], VerifyNow Aspirin Assay [16,37,38], and the TEG PlateletMapping System [36]. However, the present study demonstrates that in aspirin-treated subjects none of these five assays is significantly correlated with the most definitive marker of aspirin resistance: serum TXB2 (Table 4B, first row). However, three of these assays – arachidonic acid-induced platelet aggregation, ADP-induced platelet aggregation, and the VerifyNow Aspirin Assay – correlated with each other (Table 4B). These findings suggest that the residual post-aspirin activity measured by these three assays may be partly COX-1-independent, consistent with other recent data from our group [26,39] and others [35,40]. Furthermore, given that arachidonic acid-induced platelet aggregation, ADP-induced platelet aggregation, and the VerifyNow Aspirin Assay results have been shown to correlate with clinical outcomes [15,16,37,38], but serum TXB2 has not [39], our present data demonstrate that hyporesponsiveness to aspirin as defined by platelet aggregation and/or the VerifyNow Aspirin Assay cannot be assumed to reflect aspirin resistance as defined by aspirin’s inability to suppress TXB2.

Acknowledgements

The authors thank Thomas J. McLaughlin, Sc.D. for statistical advice. This study was funded in part by research grants to the University of Massachusetts Medical School by McNeil Consumer Healthcare, Accumetrics, and Haemoscope.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.

Ancillary