These authors contributed equally to this study.
Cyclooxygenase-1 haplotype modulates platelet response to aspirin
Article first published online: 9 SEP 2005
Journal of Thrombosis and Haemostasis
Volume 3, Issue 10, pages 2340–2345, October 2005
How to Cite
MAREE, A. O., CURTIN, R. J., CHUBB, A., DOLAN, C., COX, D., O'BRIEN, J., CREAN, P., SHIELDS, D. C. and FITZGERALD, D. J. (2005), Cyclooxygenase-1 haplotype modulates platelet response to aspirin. Journal of Thrombosis and Haemostasis, 3: 2340–2345. doi: 10.1111/j.1538-7836.2005.01555.x
- Issue published online: 9 SEP 2005
- Article first published online: 9 SEP 2005
- Received 24 February 2005, accepted 9 June 2005
- cardiovascular disease;
Summary. Background: Aspirin (acetylsalicylic acid) irreversibly inhibits platelet cyclooxygenase (COX)-1, the enzyme that converts arachidonic acid (AA) to the potent platelet agonist thromboxane (TX) A2. Despite clear benefit from aspirin in patients with cardiovascular disease (CAD), evidence of heterogeneity in the way individuals respond has given rise to the concept of ‘aspirin resistance.’Aims: To evaluate the hypothesis that incomplete suppression of platelet COX as a consequence of variation in the COX-1 gene may affect aspirin response and thus contribute to aspirin resistance. Patients and methods: Aspirin response, determined by serum TXB2 levels and AA-induced platelet aggregation, was prospectively studied in patients (n = 144) with stable CAD taking aspirin (75–300 mg). Patients were genotyped for five single nucleotide polymorphisms in COX-1 [A-842G, C22T (R8W), G128A (Q41Q), C644A (G213G) and C714A (L237M)]. Haplotype frequencies and effect of haplotype on two platelet phenotypes were estimated by maximum likelihood. The four most common haplotypes were considered separately and less common haplotypes pooled. Results: COX-1 haplotype was significantly associated with aspirin response determined by AA-induced platelet aggregation (P = 0.004; 4 d.f.). Serum TXB2 generation was also related to genotype (P = 0.02; 4 d.f.). Conclusion: Genetic variability in COX-1 appears to modulate both AA-induced platelet aggregation and thromboxane generation. Heterogeneity in the way patients respond to aspirin may in part reflect variation in COX-1 genotype.
Platelet aggregation at sites of atherosclerotic plaque rupture triggers arterial thrombosis, which underlies myocardial infarction and stroke. Aspirin (acetylsalicylic acid) reduces risk of cardiovascular morbidity and mortality in patients with pre-existing vascular disease and hence is widely prescribed for secondary prevention . Aspirin irreversibly inhibits platelet cyclooxygenase (COX)-1, the enzyme that converts arachidonic acid (AA) to the potent platelet agonist thromboxane TXA2. By acetylating serine residue 530 in COX-1's active site aspirin sterically inhibits AA metabolism and TX formation . The anucleate platelet cannot regenerate COX thus recovery of TX synthesis depends on new platelet formation, which occurs at a rate of 10% daily. Irreversible enzyme deactivation and slow rates of platelet turnover permit complete COX-1 inhibition with a single daily dose of aspirin .
Despite clear benefit from aspirin in cardiovascular disease (CAD) patients, evidence of heterogeneity in the way individuals respond has given rise to the concept of ‘aspirin resistance’ . Two recent studies suggest that aspirin resistance may be clinically significant. A (HOPE) sub-study determined that high levels of a TX metabolite, 11-dehydrothromboxane B2, in urine predicted myocardial infarction and cardiovascular death . Subsequently, Gum et al showed that persistent platelet aggregation to AA and adenosine di-phosphate in patients with stable CAD on aspirin was associated with a greater than threefold increased risk of major adverse events . However, to date no controlled prospective study has been performed comparing aspirin responsive and aspirin resistant subjects.
Heritable factors are greater determinants of platelet reactivity than environmental factors in patients with CAD . Thus, in the present study we investigated whether genetic factors influence platelet response to aspirin. We hypothesized that incomplete suppression of platelet COX as a consequence of variation in the COX-1 gene may affect aspirin response and thus contribute to aspirin resistance. Drug response is measured in terms of two platelet phenotypes; generation of TXB2 in serum and AA-induced platelet aggregation. Both assays depend on COX-1 function and thus reflect aspirin's pharmacological effect on the platelet .
The protocol was approved by the institutional Ethics Committees of Beaumont and St James's hospitals, Dublin, Ireland and all patients gave written informed consent.
Men and women aged 21 years and older were recruited consecutively while attending cardiology out-patient clinics between March and September 2002. Patients with CAD documented by cardiac catheterization or determined by medical history, electrocardiography and exercise testing and taking aspirin (75–300 mg) once daily for at least 2 weeks were eligible. Those with a recent history (within 6 weeks) of myocardial infarction, unstable angina, coronary angioplasty, coronary artery bypass grafting or major surgical procedure were ineligible. Patients with a personal or family history of bleeding, hematological disorder, a platelet count <150 000/μL or >450 000/μL or hemoglobin <8 g dL−1 were excluded. Recent ingestion (within 2 weeks) of non-steroidal anti-inflammatory drugs, other cyclooxygenase inhibitors, alternative anti-platelet therapy or anticoagulants was also criteria for exclusion. Patients were re-contacted in advance of blood sampling and drug compliance was emphasized and also recorded on the day.
Blood samples were drawn into vacuum tubes between 10am and 1pm on each occasion and always within 6 h of the last dose of aspirin. 4 × 9 mL blood samples were collected in 3.2% sodium citrate, 2 for leukocyte DNA extraction and storage and 2 for platelet aggregation studies. 4.5 mL of blood was collected into 3.8% sodium citrate for platelet count and hemoglobin determination. A further 5 mL sample was collected in a non-siliconized glass tube for serum thromboxane TXB2 assay.
Thromboxane B2 assay
Within 10 min of sampling, serum TXB2 samples in non-siliconized glass tubes were incubated at 37 °C on a heating block for 1 h to achieve full clot formation. Serum was separated by centrifugation at 1000 g relative centrifugal force (RCF) for 10 min and the supernatant removed and stored in a cryotube at −80 °C. Quantitative serum TXB2 assay was carried out subsequently by enzyme immunoassay (Assay Design Inc). The serum samples were brought to room temperature and diluted using the assay buffer provided to fall within the range of the standard curve (10 000–13.7 pg mL−1). A quality control program was carried out on each occasion. Samples and standards were run in triplicate in 96 well microplates coated with a goat anti- rabbit polyclonal antibody. Samples consisted of 100 μL of diluted sample, 50 μL of TXB2 conjugate and 50 μL of TXB2 antibody solution. The microplate was incubated at room temperature for 2 h on a horizontal orbital microplate reader set at 500 ± 50 rpm. Each well was decanted and washed three times with wash buffer. The plate was then incubated at room temperature for 1 h. Fifty microliter of stop solution was added to all wells and optical density determined using a microplate reader set to 405 nm with wavelength correction set between 600 and 690 nm. The average of each sample was calculated. A standard curve was created by plotting mean absorbance of each standard on a linear y-axis against the log concentration on the x-axis. Concentration of TXB2 was calculated using the standard curve and multiplied by the dilution factor. A population of 20 controls was used to determine the normal range (100–500 ng mL−1).
Platelet aggregation studies
Ex-vivo platelet aggregation studies were performed on whole blood collected in 3.2% sodium citrate at a final dilution of 1:10 within 20 min of sampling. Samples were then divided into 5 mL volumes and centrifuged at 160 g (RCF) for 10 min to procure platelet rich plasma (PRP). PRP was removed with a Pasteur pipette and the remaining plasma centrifuged at 2500 g for 5 mins to obtain platelet poor plasma (PPP). Five hundred microliter of PPP in a glass cuvette was used to set baseline PPP levels for the aggregometer. Four hundered and fifty microliter of PRP was then added to another glass cuvette containing a magnetic stir bar. Baseline percentage aggregation was set and recorded for 1 min before 50 μL of agonist was added. Aggregation response was recorded for 3 min and all agonists run in duplicate. Platelet aggregation was studied in response to AA (1.6 mm) and thrombin receptor activating peptide (TRAP) (5 μm) at 37 °C by light transmission (Biodata PAP-4, Biodata Corporation, Horsham, PA, USA). TRAP aggregation was used as a positive control for the assay.
DNA preparation and genotyping
Genomic DNA was isolated from whole blood by red cell lysis and subsequent digestion of pelleted leucocytes with Proteinase K/sodium dodecyl sulfate. Contaminating protein was eliminated using a ‘salting out’ technique .
Patients were genotyped for five single nucleotide polymorphisms in COX-1 [A-842G, C22T (R8W), G128A (Q41Q), C644A (G213G) and C714A (L237M)] (Table 1). The A842G variant is known to be in complete linkage disequilibrium with C50T in the signal peptide (Fig. 1). Genotyping was performed by KBioscience using Amplifluor technology (http://www.kbioscience.co.uk). Twenty microliter of total DNA was supplied at 2 ng/ μL concentration (3 ng is consumed per assay) in ‘v-bottomed’ 96-well microtitre plates. Repeat and blank samples were included as controls. Fifty to 400 base pair of sequence data on either side of the polymorphism was also provided.
|Gene region||Nucleotide change*||Protein change||Relationship to COX molecule||Genotype frequency (p/q)||Heterozygosity (2 pq)|
|Exon 2||C22T||R8W||Signal peptide||0.94/0.06||0.111|
|Exon 6||C644A||G213G||Synonymous molecule surface||0.88/0.12||0.214|
|Exon 7||C714A||L237M||Homodimer interface new hydrogen bond||0.98/0.02||0.034|
The primary endpoint was to determine if AA-induced platelet aggregation and TXB2 generation in serum were influenced by COX-1 genotype. A cut-off of persistent AA-induced platelet aggregation ≥20% despite aspirin therapy has previously been associated with adverse cardiovascular events and was thus deemed significant . For comparison, the relationship between COX-1 haplotype on TRAP-induced platelet aggregation was also evaluated. Serum TXB2 > 2.2 ng mL−1 was taken as evidence of suboptimal COX-1 inhibition as this predicts incomplete inhibition of platelet cyclooxygenase (unpublished data, Maree and Fitzgerald).
Haplotype frequencies and effect of haplotype on platelet phenotype were estimated by maximum likelihood. The four most common haplotypes were considered individually and less common haplotypes pooled. The HAPIPF program of the STATA software package was used to compare the haplotype frequencies between high aggregation and low aggregation patients, and between high TXB2 and low TXB2 patients . Significance was inferred when the overall haplotypic distribution differed from expectation with P ≤ 0.05. The method of Schaid et al.  was used to infer effects of individual haplotypes.
COX-1 haplotype and AA-induced platelet aggregation
Arachidonic acid-induced platelet aggregation (%) mean and standard error values were 13.59 ± 1.62 and 95% CI were 10.41 and 16.78. COX-1 allele frequencies were similar to those seen in previous studies of Caucasian populations (Table 1) [11,12]. COX-1 haplotypes were significantly associated with AA-induced platelet aggregation in patients on aspirin (P = 0.004; 4 d.f.; Table 2; Fig. 2). This primarily reflected an association between the haplotype GCGCC with increased aggregation (Table 3) . This haplotype carried by 12% of the population contains the minor allele of the promoter variant −A842G variant and is known to be in complete linkage disequilibrium with another variant, C50T in the signal peptide.
|COX-1 Haplotype||Platelet aggregation <20% (%)*||Platelet aggregation ≥20% (%)*||Serum TXB2≤2.2 ng mL−1 (%)*||Serum TXB2 >2.2 ng mL−1 (%)*|
|ACGCC||182.4 (75)||13.0 (65)||139.4 (76)||60.6 (71)|
|ACGAC||25.5 (10.5)||1.0 (5)||19.3 (10)||5.3 (6)|
|GCGCC||11.7 (5)||4.0 (20)||9.3 (5)||7.3 (9)|
|ATGCC||12.5 (5)||1.0 (5)||8.0 (4)||5.8 (7)|
|Rare haplotypes||11.9 (5)||1.0 (5)||7.87 (4)||7.0 (8)|
|Total||244.0 (100)||20.0 (100)||183.9 (100)||85.7 (100)|
|COX-1 Haplotype||Haplotype score for AA aggregation*||P-value||Haplotype score for serum TXB2*||P-value|
The TRAP-induced platelet aggregation was not significantly associated with COX-1 haplotype, as would be expected. Globally, the four common haplotypes did not have a significant impact on TRAP activation (P = 0.12; 4 d.f.). The only haplotype which showed any marked departure from expectation was the protective effect of the ATGCC haplotype (haplotype score of −1.235), but this haplotypic effect was not significant (P = 0.18).
COX-1 haplotype and serum thromboxane B2 formation
Serum TXB2 value (ng mL−1) mean and standard error was 6.99 ± 1.2 and 95% CI were 4.4 and 9.5. Serum TXB2 generation was also significantly modified by genotype (P = 0.02; 4 d.f.; Table 2; Fig. 3). This effect however appeared to reflect a broad influence of genetic variation on phenotype rather than a significant contribution of an individual common haplotype (Table 3).
Many patients who take aspirin for secondary prevention will have further vascular events . In some cases an incomplete pharmacological effect may be detected and thus resistance to therapy concluded. This is not surprising because drug doses are selected based on population dose-response analysis rather than individual responses. As with most drugs, ‘aspirin resistance’ may reflect either pharmacokinetic or pharmacodynamic mechanisms. Here, we describe a potential pharmacodynamic explanation as to why some individuals demonstrate an incomplete response to aspirin.
Genetic variation in COX-1 may affect enzyme expression, biochemical function or interaction with pharmacological agents. To the best of our knowledge, the present study is the first to associate COX-1 haplotype and platelet response to aspirin in a CAD population. The human COX-1 gene  encodes more than 20 variants however most are rare [11,12]. We investigated five common polymorphisms and found a significant haplotypic association with platelet aggregation and TXB2 generation. The promoter −842G allele appears to contribute significantly to an aspirin-resistant haplotype, although this may be due to known or unknown genetic variants in linkage disequilibrium. One such variant, C50T (P17L) in the signal peptide is in complete linkage disequilibrium with A-842G in Caucasians . A recent study by Ulrich et al.  investigated the association between COX-1 polymorphisms and colorectal cancer and also evaluated the protective effect of aspirin in this population. The authors determined that carriers of the common P17P (C50C) genotype taking aspirin were less likely to develop colorectal adenomas than those not taking aspirin [OR = 0.6 (0.5–0.8); P = 0.03]. They concluded that this may reflect aspirin pharmacogenetics . The apparent enhanced aspirin effect among P17P carriers is consistent with our findings. Because of the complete linkage disequilibrium between A-842G and C50T it can be inferred that patients in our population with the GCGCC haplotype (−842G) also carry the less common 50T allele (P17L) in the signal peptide. Interestingly these patients were significantly less sensitivity to aspirin as determined by AA-induced platelet aggregation (P = 0.009; Fig. 1; Table 3). Another study of 38 healthy volunteers associated A-842G/C50T heterozygosity with greater inhibition of prostaglandin F2α formation by aspirin, which appears to be at odds with our findings and those of Ulrich et al. .
In patients taking aspirin for secondary prevention of CAD, we have shown that two established platelet phenotypes, aggregation to AA and TXB2 formation in serum, are influenced by COX-1 genotype. Whether this is because of altered COX-1 inactivation by aspirin or reflects a relationship between the haplotype and an unidentified additional factor awaits further study. As much of the effect on AA-induced platelet aggregation could be accounted for by one common haplotype carried by 12% of the population, genetic variation does not appear to provide an explanation for all studies detecting aspirin resistance. Its incidence has been reported to be as high as 40% depending on definition applied and phenotype determined. It may however be one of several mechanisms contributing to a composite phenomenon .
The number of individuals in our population with elevated aggregation was relatively low, thus replication of these findings in a larger sample of patients with high aggregation despite aspirin therapy is required. An alternative approach might be to look at patients with greater platelet activation such as those with an acute coronary syndrome or postcoronary artery bypass grafting.
Genetic variability in COX-1 appears to modulate both AA-induced platelet aggregation and TX generation. Heterogeneity in the way patients respond to aspirin may in part reflect variation in COX-1 genotype. Functional characterization of COX-1 haplotype may facilitate risk prediction and individualized antiplatelet therapy in the future.
Andrew Maree was responsible for protocol development, grant funding application, patient recruitment, data collection and extraction, statistic and genetic analysis. Ronan Curtin was responsible for protocol development, grant funding application, patient recruitment, data collection and extraction. Anthony Chubb contributed to assay development and genetic analysis. Ciara Dolan contributed to the statistic and bioinformatics analysis. Dermot Cox was responsible for assay development and supervision and data extraction and analysis. Peter Crean contributed to protocol development, ethics approval and patient recruitment. John O'Brien was responsible for the genotyping assay and data analysis. Denis Shields was responsible for DNA collection, genetic data management, and statistic and bioinformatics analysis. Desmond Fitzgerald devised and supervised the study, was responsible for protocol development, assay design, data interpretation and analysis of results. Andrew Maree, Denis Shields and Desmond Fitzgerald wrote the first draft of the paper. All authors contributed to the final draft.
The study was sponsored by the Irish Heart Foundation and Higher Education Authority. Access to facilities at the Clinical Research Centre, Beaumont Hospital, Dublin is gratefully acknowledged. Technical assistance from Ms Michelle Dooley's with laboratory assay optimization and data collection is much appreciated. Supported by grants from the Irish Heart Foundation and Higher Education Authority.
- 9Haplotype analysis in population-based association studies. Stata J 2001; 1: 58–75.