Healthy volunteers including both men and women 20 to 30 years of age were recruited for this study. Subjects were proved to have no abnormalities on physical examination, 12-lead electrocardiogram, and routine laboratory tests. Subjects with a history of cardiovascular disease, or hemostatic disorder, and hypersensitivity to NSAIDs and clopidogrel, were excluded. Women with childbearing potential were tested for pregnancy, and women with a positive test result were excluded. Others were excluded if they were smokers, chronic drinkers or overweight (exceeding 20% of standard body weight). Subjects had to abstain from alcohol, beverages containing caffeine, and other drugs beginning one week before the study.

The study was approved by the review board of Seoul National University Hospital. Written and informed consent was obtained from all volunteers before enrollment in the study.

We designed a single-center, open label, parallel-group and randomized study. Young healthy volunteers (n=40) were randomized into 5 groups: celecoxib only (CCX), aspirin only (ASA), celecoxib+aspirin (CCX+ASA), aspirin+clopidogrel (CCX+CPD), and celecoxib+aspirin+clopidogrel group (CCX+ASA+CPD). Screening tests were performed 3 days before the initiation of the study. A celecoxib (Celebrex®, Pfizer Korea, Seoul, Korea) dose of 200 mg twice a day was chosen since this is the usual dose in clinical practice for treating arthritis and pain. Aspirin (Rhonal®, Kun-Wha Pharmaceuticals, Seoul, Korea) 100 mg daily, and/or clopidogrel (Plavix®, Sanofi-Aventis, Seoul, Korea) 75 mg daily were used because they are the standard doses for treating patients with ischemic heart disease. Each subject received their assigned medications for six consecutive days. Blood samples were collected at day 0 (pre) and day 7 (post) for assessment of platelet aggregation. Urine samples were collected at the same time to evaluate prostacyclin and thromboxane levels.

Whole blood samples were drawn into four standard sodium citrate containing tubes. Samples were centrifuged at 1,200 rpm for 2 minutes and 30 seconds. Platelet aggregation was induced by adenosine 5'-diphosphate (adenosine diphosphase, ADP) or collagen. It was measured using a Chronolog Lumi-Aggregometer (model 560-Ca, Chrono-log Corp., Havertown, PA, USA). The extent of platelet aggregation in samples was expressed by the maximum increase in light transmittance compared to baseline values.

Urine 6-keto PGF_1α, a stable metabolite of prostaglandin I₂ (PGI₂), was measured as described in previous studies.12)13) Urine samples were centrifuged to remove any precipitates. Correlate-EIA™ Immunoassay kits (Assay Designs, Inc., Ann Arbor, MI, USA) were used to measure urine levels of 6-keto PGF_1α. Values obtained using the kit were adjusted by VAGUE. the urine creatinine level of each subject. Urine samples were also used to measure urinary 11-dehydro-TXB₂, which is a major metabolite of TXB₂. Correlate-EIA™ Immunoassay kits (Assay Designs, Inc., Ann Arbor, MI, USA) designed for urinary 11-dehydro-TXB₂ were used.

Platelet aggregation results are presented as means and standard deviations. Changes from baseline values were calculated to detect the average differences in the platelet aggregation for each treatment group. The Wilcoxon rank sum test was used to analyze for between group differences. Results for urinary 6-keto PGF_1α were calculated to show changes from baseline values. The Mann-Whitney test was performed to show whether any differences in prostacyclin between groups was statistically significant. Statistical Package for the Social Sciences (SPSS) software (Version 12.0 for Windows, SPSS Inc, Chicago, IL, USA) was used for statistical analysis. A p<0.05 was considered to indicate statistical significance.

Fig. 1A shows ADP-induced platelet aggregation measured before and after each treatment. The CCX group showed no significant change in platelet aggregation after treatment (pretreatment 86.9±11.0% vs. post-treatment 74.6±18.5%, p=0.29). ASA and ASA+CPD groups showed significantly reduced ADP-induced platelet aggregation compared to baseline (p<0.05), and this reduction was not affected by addition of celecoxib.

Changes in collagen-induced platelet aggregation are shown in Fig. 1B. Celecoxib treatment alone did not affect collagen-induced platelet aggregation (pretreatment 85.9±15.0% vs. post-treatment 76.5±14.4%, p=0.12). ASA or ASA+CPD group showed reduced collagen-induced platelet aggregation compared to baseline (p<0.05), which was not affected by celecoxib.

The reduction in ADP-induced platelet aggregation by ASA (Δ aggregation, Pretreatment inhibition %-Post-treatment inhibition %) was not retarded by addition of CCX; rather, it was potentiated (9.8±9.2% vs. 30.9±16.8%, p=0.021) (Fig. 2A). The reduction in collagen-induced platelet aggregation by aspirin was not significantly retarded by addition of celecoxib (ASA group 38.6±12.1% vs. ASA+CCX group 29.1±20.9%, p=0.29) (Fig. 2A).

We also compared the 'Δ aggregation' between ASA+CPD and ASA+CPD+CCX groups. The reduction in ADP-induced platelet aggregation by ASA+CPD was not significantly affected by addition of celecoxib (ASA+CPD group 31.3±6.9% vs. ASA+CPD+CCX group 32.4±12.2%, p=0.83) (Fig. 2B). Inhibition of collagen-induced platelet aggregation by aspirin and clopidogrel was also not affected by addition of celecoxib (47.6±13.4% vs. 51.6±23.7%, p=0.69) (Fig. 2B).

Pretreatment mean urine 6-keto PGF_1α values adjusted by urine creatinine were not different among groups (Fig. 3). All groups tended to have decreased adjusted urine 6-keto PGF_1α levels after administration of study medications compared to pretreatment values. These trends were not statistically significant except for the ASA+CPD+CCX group (Fig. 3E) (Pretreatment 296.8±231.4 pg/mg creatinine vs. post-treatment 102.0±43.1 pg/mg creatinine, p=0.025). When we compared the changes in urine 6-keto PGF_1α levels (pretreatment value-post-treatment value) (Fig. 4), there were no significant differences between ASA and ASA+CCX groups (75.8±199.7 pg/mg creatinine vs. 108.8±295.5 pg/mg creatinine, p=1.0) or between ASA+CPD and ASA+CPD+CCX groups (49.5±406.6 pg/mg creatinine vs. 194.8±219.9 pg/mg creatinine, p=0.20) (Fig. 4A).

In all five groups, treatment with anti-platelet agents with or without celecoxib showed a tendency towards decreased levels of urinary 11-dehydro-TXB₂ (Fig. 3). In the ASA+CCX group there was a statistically significant decrease in the level of urine 11-dehydro-TXB₂ after treatment (Fig. 3C) (Pretreatment 409.5±181.6 pg/mg creatinine vs. post-treatment 152.6±82.1 pg/mg creatinine, p=0.012).

The reduction in urine 11-dehydro-TXB₂ was more profound after ASA+CCX than after ASA treatment alone (Fig. 4B) (256.96±213.1 pg/mg creatinine vs. 17.2±114.1 pg/mg creatinine, p=0.005). Reductions in urinary 11-dehydro-TXB₂ were not significantly different between ASA+CPD and ASA+CPD+CCX groups.

There has been a concern that COX-2 inhibitors might increase thrombogenicity and are associated with adverse cardiovascular events. The rationale was that COX-2 inhibitors suppress the synthesis of prostacyclin in endothelial cells while they cannot inhibit thromboxane A₂ formation due to unopposed expression of COX-1 in platelets.14) It was demonstrated that urinary excretion of 2, 3 dinor-6 keto PGF_1α and 6-keto PGF_1α, which represents prostacyclin biosynthesis, was reduced by celecoxib at 4-6 hours after dosing, and partially recovered after 12-24 hours.13) The effect of aspirin and clopidogrel on prostacyclin production was assessed in vitro in a previous study,15) where aspirin reduced endothelial production of 6-keto PGF_1α significantly in a concentration-dependent way, whereas clopidogrel did not. However, celecoxib did not interfere with normal mechanisms of platelet aggregation and hemostasis and did not alter serum thromboxane B₂ level in healthy adults.16)

In our study, prostacyclin production tended to decrease in all five groups after treatment. Although it was not statistically significant, prostacyclin production seemed to decrease more in the ASA+CCX group than in the ASA group, and more in the ASA+CPD+CCX than in the ASA+CPD group. But these changes in prostacyclin production were paralleled by changes in thromboxane. Our results shows that urinary 11-dehydro-TXB₂ levels were also decreased significantly more in the ASA+CCX group than in the ASA only group, and tended to decrease more in the ASA+CPD+CCX group compared to the ASA+CPD group. Together, these results suggest that celecoxib as a COX-2 inhibitor has a neutral effect on platelet aggregation since it has mild inhibitory effects not only on prostacyclin production but also on thromboxane production, thus not changing the balance of prostacyclin and thromboxane.

In this study, aspirin inhibited ADP-induced platelet aggregation and collagen-induced platelet aggregation by 13% and 55% respectively. This result is consistent with a previous study that reported values of 14% and 50%.17) We confirmed that celecoxib did not affect the platelet aggregation response to collagen in healthy volunteers receiving aspirin as Wilner et al.18) previously reported. Also in patients who were taking aspirin and had both osteoarthritis and ischemic heart disease, celecoxib, unlike ibuprofen, did not influence aspirin's action to irreversibly inactivate platelet COX-1.19)

In addition, celecoxib is known to have COX-independent biological effects. We previously reported that celecoxib in animal studies inhibits neointimal hyperplasia after balloon injury by blocking Akt signaling.20) Celecoxib also decreases inflammatory responses and neointimal hyperplasia, possibly through inhibition of MCP-1 expression.21) It also suppressed aortic endothelial cell tissue factor expression, leading to prevention of thrombus formation in the vasculature, which was observed for celecoxib but not for other COX-2 inhibitors.22)

In the current study, we evaluated the effect of dual anti-platelet therapy (aspirin and clopidogrel) on platelet aggregation and whether celecoxib affects the antiplatelet aggregation activity of aspirin and clopidogrel. Clopidogrel has a potent anti-thrombotic action by blocking P2Y12, a platelet ADP receptor, and also inhibits platelet aggregation induced by collagen.15)23)

In our study, we demonstrated a significant additional inhibitory effect of clopidogrel on both ADP- and collagen-induced platelet aggregation when added to aspirin. More importantly, we observed that the enhanced antiplatelet effect of dual antiplatelet therapy compared to aspirin alone is not altered by concomitant celecoxib use. This result is consistent with our previous prospective randomized clinical trial,3) where additional of celecoxib treatment for 6 months after coronary stenting significantly reduced late loss of the stented segment and the target lesion revascularization rate without any increase in adverse thrombotic events. It is noteworthy that all patients in this trial were taking both aspirin and clopidogrel for more than six months after stenting.

Despite these results indicating that celecoxib does not attenuate the antiplatelet action of aspirin and clopidogrel, and that celecoxib might even have a favorable effect on the cardiovascular system, several clinical studies have questioned the safety of celecoxib treatment for the cardiovascular system. The Cross Trial Safety Analysis which assessed cardiovascular risk based on >16,000 patient-years of follow-up from six randomized placebocontrolled trials including APC,24) Pre-SAP, and ADAPT concluded that celecoxib increases the number of adverse cardiovascular events. But such risk may not be celecoxib-specific. Rather, it may that the increased risk is associated with the entire class of NSAIDs. The mechanism of class risk of NSAIDs may be multi-factorial, i.e., induction of blood pressure elevation and platelet aggregability, etc. Actually, celecoxib's cardiovascular risk has been reported to be relatively low compared with other NSAIDs.25-27) Addition of celecoxib to patients who take dual anti-platelet agents due to coronary stenting, do not appear to be associated with thrombotic risk.

The current study recruited young healthy volunteers instead of old patients at risk for developing cardiovascular disease. In the setting of atherosclerosis, suppression of PGI₂ production in endothelial cells by celecoxib is pronounced, since COX-2 plays a greater role as a source of PGI₂.28)29) As a result, celecoxib might have a more profound effect on prostanoid balance under atherosclerotic conditions, promoting platelet-dependent thrombosis.30)

The number of volunteers allocated to each treatment group was rather small. However, previous reports evaluating effects of celecoxib on antiplatelet activity and prostanoids had similar numbers of subjects. And results of platelet function tests in this study were consistent with those from previous studies.

Celecoxib does not interfere with the antiplatelet action of dual antiplatelet therapy, and it does not affect the balance of prostacyclin and thromboxane production. Recent evidence suggests that celecoxib may increase cardiovascular events in high risk patients and that aspirin may not be protective. Since patients with coronary artery stenting take aspirin and clopidogrel together, and some of them take it indefinitely, our study results are worth consideration. Celecoxib can be administered safely in these patients during coverage of dual antiplatelet therapy. Current warnings for celecoxib now mandate larger clinical trials that evaluate the safety of celecoxib during dual antiplatelet therapy following stenting.