Differential impairment of aspirin-dependent platelet cyclooxygenase acetylation by nonsteroidal antiinflammatory drugs

Abstract

The cardiovascular safety of nonsteroidal antiinflammatory drugs (NSAIDs) may be influenced by interactions with antiplatelet doses of aspirin. We sought to quantitate precisely the propensity of commonly consumed NSAIDs—ibuprofen, naproxen, and celecoxib—to cause a drug-drug interaction with aspirin in vivo by measuring the target engagement of aspirin directly by MS. We developed a novel assay of cyclooxygenase-1 (COX-1) acetylation in platelets isolated from volunteers who were administered aspirin and used conventional and microfluidic assays to evaluate platelet function. Although ibuprofen, naproxen, and celecoxib all had the potential to compete with the access of aspirin to the substrate binding channel of COX-1 in vitro, exposure of volunteers to a single therapeutic dose of each NSAID followed by 325 mg aspirin revealed a potent drug-drug interaction between ibuprofen and aspirin and between naproxen and aspirin but not between celecoxib and aspirin. The imprecision of estimates of aspirin consumption and the differential impact on the ability of aspirin to inactivate platelet COX-1 will confound head-to-head comparisons of distinct NSAIDs in ongoing clinical studies designed to measure their cardiovascular risk.

Keywords: MS; acetylation; aspirin; cyclooxygenase; nonsteroidal antiinflammatory drugs.

Fig. 1.

Study designs. (A) Qualification of COX-1 acetylation as a pharmacodynamic marker of the antiplatelet action of aspirin. Subjects received a single dose of 325 mg uncoated aspirin; blood was drawn and urine was collected at the indicated time points. Established markers of the antiplatelet effect of aspirin [serum TxB₂, urinary 11-dehydro-TxB₂ (a stable Tx metabolite), and arachidonic acid-induced platelet aggregation] were compared with COX-1 acetylation. (B) The NSAID–aspirin interaction was assessed in an open-label two-period trial. Aspirin responsiveness was tested in the absence (period 1) and presence of ibuprofen, naproxen, or celecoxib (period 2). COX-1 acetylation, serum TxB₂, urinary 11-dehydro-TxB₂ (a stable Tx metabolite), arachidonic acid-induced platelet aggregation, and microfluidic platelet deposition on collagen in whole blood were measured at each time point.

Fig. 2.

The human pharmacology of platelet COX-1 acetylation. Healthy volunteers (n = 8) received a single dose of 325 mg aspirin under supervision and were followed up for 10 d (0–240 h plotted on a log scale). Platelet COX-1 acetylation and established markers of COX-1 pathway inhibition were measured. (A) Platelet COX-1 acetylation in vivo was quantified before and after aspirin administration (orange). The maximally achievable COX-1 acetylation was determined by exposing washed platelets to a supra therapeutic concentration of aspirin (500 μmol/L) ex vivo (gray). (B) The enzymatic activity of platelet COX-1 was assessed by measurement of serum TxB₂ ex vivo (green). (C) Urinary excretion of the major Tx metabolite, 11-dehydro TxB₂, was measured as an index of COX function in vivo (blue). (D) COX-1–dependent platelet function was determined by arachidonic acid-induced platelet aggregometry (red). Box and whiskers plots show medians (horizontal lines), interquartile ranges (boxes), and ranges (whiskers) of data at each time point.

Fig. 3.

Ibuprofen, naproxen, and celecoxib blunt aspirin acetylation of platelet COX-1 dose-dependently in vitro. Washed human platelets were preincubated with (A) ibuprofen, (B) naproxen, or (C) celecoxib at concentrations that inhibited enzymatic function in the higher concentration ranges (gray). Platelet COX-1 acetylation was assessed after the addition of 50 (n = 7 donors; dashed orange lines) or 250 μmol/L (n = 5 donors; solid orange lines) aspirin. All three NSAIDs abated COX-1 acetylation dose-dependently. The higher concentration of aspirin was less susceptible to the drug–drug interaction indicated by a right shift of the concentration–response curve. The reduction in COX-1 acetylation resulted in increased platelet Tx formation compared with aspirin alone (the green curve shows the Tx response with 250 μmol/L aspirin). Data are medians and interquartile ranges (quartiles 1 and 3).

Fig. 4.

Coadministration of NSAIDs and aspirin in volunteers. Healthy subjects were administered aspirin alone (period 1) and aspirin 2 h after ibuprofen (n = 7), naproxen (n = 7), or celecoxib (n = 7; period 2) in a cross-over design. (A) COX-1 acetylation by aspirin alone (period 1) was significantly reduced when aspirin was given after ibuprofen or naproxen but not celecoxib. (B) Inhibition of serum Tx 24 h after the aspirin dose displayed as the post- to predose ratio of serum Tx levels (a ratio = 1 is no inhibition). Ibuprofen blunted inhibition of Tx formation by aspirin markedly. Naproxen reduced inhibition of serum Tx less potently, presumably because of the long-lasting inhibition of Tx formation of naproxen. The presence of celecoxib did not affect the activity of aspirin. (C) Inhibition of arachidonic acid-induced platelet aggregation 24 h after the aspirin dose displayed as the post- to predose ratio of arachidonic acid-induced maximal aggregation (a ratio = 1 is no inhibition). Ibuprofen prevented the inhibition by aspirin of platelet aggregation completely. Platelets were fully inhibited in the presence of naproxen or celecoxib. (D) Assessment of platelet function using a microfluidic platelet deposition assay that detects secondary aggregation under venous flow conditions (34). R values below one represent inhibited COX-dependent platelet function. Exposure to ibuprofen before aspirin prevented inhibition of secondary aggregation completely, whereas platelet function remained inhibited in the presence of naproxen or celecoxib. Data are medians, interquartile ranges (boxes), and ranges (whiskers). Aspirin alone (period 1) vs. NSAID followed by aspirin (period 2) was compared within each treatment group using the Wilcoxon matched pairs signed rank test (P values above the plots). The three NSAID treatment groups were compared using a Kruskal–Wallis analysis on the ratios of period 2 vs. period 1 data followed by the Nemenyi–Damico–Wolfe–Dunn posthoc test (P values below the plots).