The cardiovascular safety of cyclooxygenase-2–selective (coxibs) and nonselective nonsteroidal anti-inflammatory drugs (NSAIDs) is of concern, although most users remain free of adverse outcomes. A gene-drug interaction could modulate this cardiovascular risk through prostaglandin synthesis or inflammatory pathways. From an existing acute coronary syndrome cohort (Recurrence and Inflammation in the Acute Coronary Syndromes Study) (n = 1,210), a case-only study was performed by identifying 115 patients exposed to NSAIDs (rofecoxib [n = 43], celecoxib [n = 49], or nonselective NSAIDs [n = 23]) and 345 unexposed patients matched for age, gender, and hospital center. These patients were genotyped for 115 candidate single-nucleotide polymorphisms (SNPs). Statistically significant associations between NSAID exposure and 9 SNPs in 6 genes were observed. Analyzing patients exposed only to coxibs and their matched unexposed cases, significant associations remained for 5 SNPs at 4 loci (prostaglandin-endoperoxide synthase–1 [PTGS1], chromosome 9p21.3, C-reactive protein [CRP], and klotho [KL]). Two independent SNPs from the PTGS1 gene gave similar results under a recessive model, with odds ratios for the association with NSAID exposure of 6.94 (95% confidence interval 1.35 to 35.65, p = 0.016) and 7.11 (95% confidence interval 1.38 to 36.74, p = 0.033). A significant association was also observed for a SNP in the CRP gene (rs1205) (additive odds ratio 1.64, 95% confidence interval 1.18 to 2.27, p = 0.003). In conclusion, these findings suggest that genetic variability may contribute to the susceptibility for acute coronary syndromes observed in some NSAID users. In particular, genetic polymorphisms in the PTGS1 and CRP genes appear to be candidates for a possible gene-drug interaction influencing the acute coronary risk associated with NSAID use, but these findings will require confirmation in larger cohorts.
Selective cyclooxygenase (COX)–2 inhibitors are effective for the treatment of pain but are associated with an increase in adverse cardiovascular outcomes. However, the absolute rate of patients experiencing thrombotic events in the Adenomatous Polyp Prevention on Vioxx (APPROVe) trial was low (1.5/100 patient-years), suggesting that most patients were treated without incident. The biologic mechanism that triggers these cardiovascular events is still unclear. The favored hypothesis centers on a potential disequilibrium in prostaglandin synthesis between prothrombotic thromboxane A 2 and antithrombotic prostacyclin, which are regulated by the COX-1 and COX-2 enzymes, respectively. However, the use of low-dose aspirin does not appear to attenuate the risk, suggesting that it is not due solely to the induction of a prothrombotic state. It is possible that only some nonsteroidal anti-inflammatory drug (NSAID) users may be genetically susceptible to increased cardiovascular risk, but this hypothesis has yet to be tested.
Supporting our hypothesis that cardiovascular safety may be influenced by genetic variants, recent studies have shown that polymorphisms in key prostaglandin metabolism genes may lead to different levels of therapeutic response to NSAIDs as well as variation in prostaglandin synthesis and the development of hypertension. For example, the level of prostaglandin-endoperoxide synthase–2 (PTGS2) expression and the degree of pain relief seen with rofecoxib and ibuprofen varied by allele. Furthermore, there is evidence that genetic variants in the prostaglandin-endoperoxide synthase–1 (PTGS1) gene alter the metabolism of arachidonic acid to prostaglandin H 2 and prostaglandin F 2α , the precursors of prostacyclin, thromboxane, and several other prostaglandin metabolites. Finally, variation in aspirin response is associated with 2 single-nucleotide polymorphisms (SNPs) in PTGS1. Therefore, we sought to investigate potential gene-drug interactions between SNPs in select candidate genes and NSAID exposure that may contribute to the cardiovascular risk associated with this drug class. Our primary hypothesis focused on the 2 prostaglandin synthesis enzyme targets of NSAIDs, COX-1 and COX-2, but we also examined several other candidate genes and SNPs previously associated with drug metabolism, inflammation, thrombosis, or other cardiovascular disease processes.
Methods
We studied 460 (38%) patients from the previously established Recurrence and Inflammation in the Acute Coronary Syndromes Study cohort who were admitted in 2001–03 for an acute coronary syndrome (ACS) including either myocardial infarction or unstable angina pectoris. This cohort was from 4 tertiary and 4 community hospitals, 7 in Quebec and 1 in New Brunswick, Canada. Patients were recruited within 24 hours of symptom onset. Myocardial infarction was defined as characteristic chest discomfort or pain with an elevation of creatine kinase-MB to ≥1.5 times the upper normal limit or positive cardiac troponin and included incident and recurrent cases. Unstable angina pectoris was defined as either 1 episode lasting ≥10 minutes or ≥2 episodes lasting ≥5 minutes within 24 hours of characteristic chest discomfort or pain at rest or with minimal exertion and ≥1 of these features: ischemic electrocardiographic changes, cardiac troponin I or T considered positive for myocardial necrosis with creatine kinase but under the threshold definition of myocardial infarction, history of cardiovascular disease (myocardial infarction, coronary revascularization, coronary angiography with ≥1 artery with ≥50% stenosis, positive noninvasive test results, peripheral vascular disease, or cerebrovascular disease), or diabetes.
On study admission, demographic data, risk factors, and clinical information including recurrent ischemia, cardiac function, and the use of invasive cardiac procedures were collected. Baseline medications, including specific NSAIDs, regularly prescribed were recorded by research study nurses. Because even short-term continuous NSAID exposure is associated with increased risk, we considered patients treated with NSAIDs within the 10 days preceding admission to be exposed. Low-dose aspirin treatment (≤325 mg) was not considered as NSAID use. Local principal investigators confirmed all data entry, and on study completion, data were systematically verified by on-site visits and assessment by a central adjudication committee. The study protocol was approved by the institutional review committee of each of the participating hospitals, and all subjects gave informed consent for blood sample collection and genetic testing.
Our primary hypothesis involved candidate genes controlling the COX-1 and COX-2 enzymes. COX-1 is encoded by the constitutively expressed PTGS1 gene and produces prostaglandins involved in the regulation of stomach mucosa, platelet aggregation, and kidney function. COX-2 is encoded by the inducible PTGS2 gene, is induced by inflammatory cytokines and mitogens, and is believed to produce most prostaglandins involved in inflammation and cancer. These enzymes are mediators of the key prostaglandins involved in vascular homeostasis, including prostacyclin and thromboxane A 2 . Our primary hypothesis was thus to investigate associations between genetic variants of PTGS1 (11 tagging SNPs) and PTGS2 (14 tagging SNPs) and exposure to NSAIDs in patients with documented ACS.
In addition to PTGS1 and PTGS2, we selected candidate genes from a review of published evidence on the basis of the plausibility of their interaction with NSAIDs to increase the risk for cardiovascular events. These included the prostaglandin I 2 synthase gene (PTGIS), which facilitates the isomerization of prostaglandin H 2 to prostacyclin, the most potent inhibitor of platelet aggregation and a strong vasodilator. PTGIS genetic variations have been associated with hypertension and may therefore modulate cardiovascular risk due to NSAID exposure. We also selected genes regulating the matrix metalloproteinase (MMP) pathway (MMP1, MMP3, and MMP9) as well as a gene involved in the metabolism of NSAIDs (CYP2C9). These selections are supported by evidence of rofecoxib and ibuprofen altering the gene expression of MMP1 and MMP3, a putative role of MMP9 in coronary artery disease, and the role of CYP2C9 in the metabolism of celecoxib. We identified 69 tagging SNPs for these 5 prospective candidate genes: 12 for PTGIS, 20 for MMP1, 8 for MMP3, 17 for MMP9, and 12 for CYP2C9.
Finally, we selected 21 additional SNPs in 12 genes on the basis of their previous associations with conditions or processes related to cardiovascular disease. These included C-reactive protein (CRP; rs1205), angiotensinogen (AGT; rs943580, rs699), interleukin-18 (IL18; rs543810, rs360722), klotho (KL; rs211247), estrogen receptor–1 (ESR1; rs3853248), estrogen receptor–2 (ESR2; rs71454455, rs3020450), endothelin-1 (EDN1; rs9369217, rs93808973, rs5370), apolipoprotein E (APOE; rs429359, rs7412), paraoxonase-1 (PON; rs854542), resistin (RETN; rs3219177), phospholipase A2 group VII (PLA2G7; rs1805018), thrombomodulin (THBD; rs13306848), and the recently identified chromosome 9 locus (9p21.3; rs10757274). One SNP in the chromosome 9p21.3 region and 1 in ESR1 were found to be in high linkage disequilibrium with another selected SNP and were excluded from our analysis.
SNP genotype data were extracted from the HapMap Project ( http://www.hapmap.org ) for PTGS1, PTGS2, PTGIS, MMP1, MMP3, MMP9, and CYP2C9. We included 5,000 base pairs upstream and downstream of each gene to provide more complete coverage and capture variation in potential regulatory regions. Tagger software ( http://www.broad.mit.edu/mpg/haploview ) was used to select tagging SNPs that would best capture the variation within each gene. We performed tagging with a minimum correlation coefficient (r 2 ) of 0.8. All SNPs included in the tagging process had a minimum minor allele frequency of 0.05. Genotyping was performed at the McGill University and Genome Quebec Innovation Centre using the Sequenom iPLEX Gold Assay (Sequenom, Cambridge, Massachusetts).
We used the case-only study design to investigate the association between admission for ACS and the interaction of genes and NSAID use. In this design, all NSAID-exposed patients with ACS were compared with non-NSAID-exposed patients with ACS. Odds ratios derived from a case-only study indicate the multiplicative interaction between genotype and NSAID exposure. The design relies on the assumption of independence between genotype and NSAID prescription and provides greater statistical efficiency by eliminating the variance and potential biases associated with controls.
We used conditional logistic regression and matched on age, gender, and hospital center to compare genotype frequencies among patients with ACS exposed to NSAIDs versus an unexposed ACS group. SNPs were tested under additive, dominant, and recessive models. In a planned subgroup analysis, we limited our case-only study group to patients exposed to COX-2 inhibitors and the matched unexposed group.
Results
We identified 115 patients with ACS who were treated with rofecoxib (n = 43), celecoxib (n = 49), or nonselective NSAIDs (n = 23) <10 days before hospital admission and compared them to 345 patients with ACS unexposed to NSAIDs and matched for age (±10 years), gender, and hospital center. The nonselective NSAIDs included ibuprofen (n = 7), naproxen (n = 7), diclofenac (n = 4), mesalamine (n = 2), floctafenine (n = 1), meloxicam (n = 1), and high-dose aspirin (n = 1). Baseline patient characteristics are listed in Table 1 . Our cohort was typical of most ACS studies, with a mean age of 65 years, a predominance of men, and a high prevalence of conventional risk factors. Baseline characteristics were comparable between exposed and unexposed subjects, but exposed patients were more likely to have hypertension and less likely to have congestive heart failure. Treatment with regular or low-dose aspirin was similar between exposed (52.2%) and unexposed (48.7%) patients.
| Variable | NSAID Exposed | NSAID Unexposed | Absolute Differences (95% CI) |
|---|---|---|---|
| (n = 115) | (n = 345) | ||
| Mean age (years) | 64.5 | 64.7 | — |
| Men | 79 (68.7%) | 237 (68.7%) | — |
| Women | 36 (31.3%) | 108 (31.3%) | — |
| Myocardial infarction | 70 (60.9%) | 214 (62.0%) | 1.1% (−9.2% to 11.4%) |
| Unstable angina pectoris | 45 (39.1%) | 131 (38.0%) | 1.1% (−9.2% to 11.4%) |
| Mean body mass index (kg/m 2 ) | 27.7 | 26.9 | 0.8 (−0.2 to 1.8) |
| Current smokers | 25 (21.7%) | 101 (29.3%) | 7.6% (−1.3% to 16.5%) |
| Past smokers | 67 (58.3%) | 162 (47.0%) | 11.3% (0.8% to 21.8%) |
| Never smokers | 23 (20.0%) | 82 (23.8%) | 3.8% (−4.8% to 12.4%) |
| Diabetes mellitus | 27 (23.5%) | 66 (19.1%) | 4.4% (−4.1% to 12.9%) |
| Hypercholesterolemia | 71 (61.7%) | 220 (63.8%) | 2.1% (−8.1% to 12.3%) |
| Hypertension | 76 (66.1%) | 179 (52.2%) | 13.9% (3.5% to 24.3%) |
| Previous myocardial infarction | 34 (29.6%) | 102 (29.6%) | 0% (−9.6% to 9.6%) |
| Previous heart failure | 2 (1.7%) | 25 (7.3%) | 5.6% (1.9% to 9.2%) |
Of the 115 candidate SNPs, 105 were successfully genotyped in >90.0% of patients. Of these, 99 SNPs were in Hardy-Weinberg equilibrium in unexposed cases. The exceptions were in the genes AGT (rs699, p = 0.0345; rs943580, p = 0.0191), CYP2C9 (rs9332197, p = 0.0320), endothelin-1 (rs9369217, p = 0.0141), MMP1 (rs2408489, p = 0.0001), and paraoxonase-1 (rs854543, p = 0.0357).
In the primary analysis of patients exposed to any NSAID, 9 SNPs significantly interacted with NSAID exposure ( Table 2 ). The largest odds ratios (ORs) were observed for 2 PTGS1 SNPs tested under the recessive model (rs10306135: OR 7.33, 95% confidence interval [CI] 1.46 to 36.88, p = 0.016; rs12353214: OR 4.77, 95% CI 1.14 to 19.99, p= 0.033). These SNPs were not in linkage disequilibrium with each other (r 2 = 0). Significant associations were also observed with 2 SNPs in the MMP1 gene, 2 SNPs in the AGT gene, 1 SNP in the chromosome 9p21.3 region, and 1 SNP in the klotho gene. Furthermore, significant associations were observed for all genetic models for 1 SNP in the CRP gene (rs1205).
| Gene and SNP ID | Genotypes | Exposed | Unexposed | MAF | Additive | Dominant | Recessive | |||
|---|---|---|---|---|---|---|---|---|---|---|
| OR (95% CI) | p Value | OR (95% CI) | p Value | OR (95% CI) | p Value | |||||
| PTGS1 | ||||||||||
| rs10306135 | AA | 79 | 234 | |||||||
| AT | 27 | 95 | ||||||||
| TT | 6 | 3 | 0.158 | 1.20 (0.77–1.87) | 0.422 | 1.00 (0.61–1.63) | 0.983 | 7.33 (1.46–36.88) | 0.016 | |
| rs12353214 | CC | 97 | 284 | |||||||
| CT | 10 | 45 | ||||||||
| TT | 5 | 3 | 0.080 | 1.12 (0.69–1.82) | 0.655 | 0.90 (0.49–1.64) | 0.726 | 4.77 (1.14–19.99) | 0.033 | |
| MMP1 | ||||||||||
| rs7945189 | CC | 94 | 249 | |||||||
| CT | 13 | 76 | ||||||||
| TT | 3 | 3 | 0.115 | 0.67 (0.40–1.13) | 0.134 | 0.55 (0.30–0.992) | 0.047 | 2.52 (0.49–12.89) | 0.266 | |
| rs2071230 | AA | 89 | 287 | |||||||
| AG | 21 | 35 | ||||||||
| GG | 0 | 1 | 0.067 | 1.80 (1.01–3.20) | 0.046 | 1.87 (1.04–3.36) | 0.036 | — | — | |
| AGT | ||||||||||
| rs943580 | AA | 43 | 115 | |||||||
| AG | 53 | 142 | ||||||||
| GG | 15 | 75 | 0.423 | 0.75 (0.55–1.02) | 0.071 | 0.82 (0.53–1.27) | 0.368 | 0.49 (0.26–0.93) | 0.028 | |
| rs699 | TT | 40 | 110 | |||||||
| TC | 57 | 144 | ||||||||
| CC | 15 | 76 | 0.433 | 0.81 (0.59–1.10) | 0.168 | 0.92 (0.59–1.43) | 0.703 | 0.52 (0.28–0.98) | 0.041 | |
| Chr9p21.3 | ||||||||||
| rs10757274 | GG | 44 | 87 | |||||||
| GA | 49 | 171 | ||||||||
| AA | 17 | 74 | 0.455 | 0.64 (0.46–0.89) | 0.007 | 0.51 (0.31–0.82) | 0.006 | 0.61 (0.33–1.10) | 0.101 | |
| CRP | ||||||||||
| rs1205 | CC | 40 | 162 | |||||||
| CT | 54 | 147 | ||||||||
| TT | 18 | 23 | 0.319 | 1.64 (1.18–2.27) | 0.003 | 1.62 (1.05–2.49) | 0.030 | 2.54 (1.33–4.87) | 0.005 | |
| KL | ||||||||||
| rs211247 | CC | 65 | 232 | |||||||
| CG | 43 | 92 | ||||||||
| GG | 5 | 8 | 0.181 | 1.69 (1.12–2.55) | 0.013 | 1.76 (1.11–2.80) | 0.017 | 1.90 (0.57–6.39) | 0.299 | |
To test for SNPs that interacted with selective COX-2 inhibitors, we selected subjects exposed to rofecoxib (n = 43) or celecoxib (n = 49) and their matched unexposed subjects (n = 276). In this subgroup analysis, an interaction between COX-2 exposure and genotype was observed for 5 SNPs ( Table 3 ). The recessive model yielded statistically significant ORs for the same SNPs in PTGS1 (rs10306135: OR 6.94, 95% CI 1.35 to 35.65, p = 0.016; rs12353214: OR 7.11, 95% CI 1.38 to 36.74, p = 0.019), as did all models for the same SNP in CRP (rs1205: OR 2.94, 95% CI 1.41 to 6.11, p = 0.004). Furthermore, in this subgroup of COX-2 users, 2 SNPs in PTGS2 showed a significant interaction under the dominant model (rs4648276: OR 1.80, 95% CI 1.05 to 3.10, p = 0.034; rs20417: OR 1.69, 95% CI 1.02 to 2.81, p = 0.044) that was not present in the primary analysis of all NSAID users.
| Gene and SNP ID | Genotypes | Exposed | Unexposed | Additive | Dominant | Recessive | |||
|---|---|---|---|---|---|---|---|---|---|
| OR (95% CI) | p Value | OR (95% CI) | p Value | OR (95% CI) | p Value | ||||
| PTGS1 | |||||||||
| rs10306135 | AA | 61 | 183 | ||||||
| AT | 22 | 78 | |||||||
| TT | 6 | 3 | 1.28 (0.79–2.09) | 0.315 | 1.03 (0.60–1.80) | 0.906 | 7.33 (1.46–36.88) | 0.016 | |
| rs12353214 | CC | 75 | 226 | ||||||
| CT | 9 | 35 | |||||||
| TT | 5 | 2 | 1.34 (0.80–2.25) | 0.264 | 1.10 (0.58–2.08) | 0.781 | 7.11 (1.38–36.74) | 0.019 | |
| MMP1 | |||||||||
| rs7945189 | CC | 73 | 196 | ||||||
| CT | 11 | 65 | |||||||
| TT | 3 | 2 | 0.72 (0.42–1.26) | 0.249 | 0.58 (0.31–1.08) | 0.087 | 3.71 (0.60–22.98) | 0.158 | |
| rs2071230 | AA | 72 | 226 | ||||||
| AG | 16 | 29 | |||||||
| GG | 0 | 1 | 1.65 (0.87–3.13) | 0.129 | 1.72 (0.90–3.31) | 0.102 | — | — | |
| AGT | |||||||||
| rs943580 | AA | 27 | 91 | ||||||
| AG | 48 | 115 | |||||||
| GG | 14 | 58 | 0.95 (0.67–1.35) | 0.767 | 1.20 (0.72–1.99) | 0.495 | 0.60 (0.30–1.22) | 0.158 | |
| rs699 | TT | 24 | 87 | ||||||
| TC | 50 | 119 | |||||||
| CC | 15 | 58 | 1.05 (0.74–1.50) | 0.787 | 1.34 (0.79–2.28) | 0.277 | 0.75 (0.39–1.45) | 0.389 | |
| Chr9p21.3 | |||||||||
| rs10757274 | GG | 34 | 73 | ||||||
| GA | 40 | 133 | |||||||
| AA | 14 | 58 | 0.68 (0.48–0.98) | 0.040 | 0.577 (0.34–0.99) | 0.045 | 0.62 (0.32–1.22) | 0.166 | |
| CRP | |||||||||
| rs1205 | CC | 30 | 128 | ||||||
| CT | 45 | 120 | |||||||
| TT | 15 | 16 | 1.76 (1.23–2.54) | 0.002 | 1.77 (1.08–2.89) | 0.023 | 2.94 (1.41–6.11) | 0.004 | |
| KL | |||||||||
| rs211247 | CC | 47 | 188 | ||||||
| CG | 39 | 69 | |||||||
| GG | 4 | 6 | 2.11 (1.34–3.34) | 0.001 | 2.35 (1.40–3.93) | 0.001 | 1.94 (0.51–7.37) | 0.33 | |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
