Lack of effect of Imrecoxib, an innovative and moderate COX-2 inhibitor, on pharmacokinetics and pharmacodynamics of warfarin in healthy volunteers

Imrecoxib is a registered treatment for osteoarthritis pain symptoms in China. This study aims to assess the effect of imrecoxib on the pharmacodynamics and pharmacokinetics of warfarin. 12 healthy male volunteers with CYP2C9*3 AA and VKORC1 AA genotypes took a 5 mg dose of warfarin both alone and concomitantly with steady-state imrecoxib. Both warfarin alone and concomitantly with imrecoxib have safey and good tolerance across the trial. Following warfarin and imrecoxib co-administration, neither Cmax, AUC0-t and t1/2 of warfarin enantiomers nor AUC of international normalized ratio (INR) were markedly different from those of warfarin alone. The geometric mean ratios (GMRs) (warfarin + imrecoxib: warfarin alone) of INR(AUC) was 1 (0.99, 1.01). The GMRs of warfarin AUC0-∞ (90% confidence interval, CIs) for warfarin + imrecoxib: warfarin alone were 1.12 (1.08, 1.16) for R-warfarin and 1.13 (1.07, 1.18) for S- warfarin. The 90% CIs of the GMRs of AUC0-∞, Cmax and INR (AUC) were all within a 0.8–1.25 interval. The combination of warfarin and imrecoxib did not impact the pharmacodynamics and pharmacokinetics of single-dose warfarin; therefore, when treating a patient with imrecoxib and warfarin, it is not required to adjust the dosage of warfarin.

Warfarin is effective for preventing intravenous thromboembolism, cardiovascular and cerebrovascular infarction, and other thromboembolic disorders. It is a racemic mixture of two isomers, CYP2C9 enzyme metabolizes S-warfarin, and CYP1A2 and CYP3A4 are responsible for metabolism of R-warfarin 9 , which make it susceptible to interaction with numerous inhibitors and inducers of CYPs. This interaction might lead to either an inability to achieve the expected anticoagulant effects or an enhanced bleeding risk induced by excessive anticoagulation 10 . In a US retrospective prescription analysis, nonsteroidal anti-inflammatory drugs (NSAIDs) with warfarin was the most frequently occurring medication pair in drug-drug interactions (DDI) reports 8 , and 24% of warfarin recipients would be given NSAIDs treatment within two years 8,11 . NSAIDs impair the gastrointestinal mucosa and aggregation of platelets by inhibiting the COX-1 isozyme 12,13 , which significantly enhances the risks of hemorrhage in patients taking warfarin [14][15][16][17] . Specific inhibitors of COX-2 have been approved for OA therapy. COX-2 specific inhibitors do not cause severe bleeding and are thus considered potentially safe for warfarin-treated patients 18 . However, increasing evidences of myocardial infarction, as well as cardiovascular secondary action relate to COX-2 specific inhibitors, such as rofecoxib and valdecoxib, lead to their retreat from the market 19,20 . Therefore, a moderate COX-2 selective inhibitor with decreased bleeding risk than NSAIDs and reduced cardiovascular secondary action compared with COX-2 specific inhibitors, would be appropriate for management of OA.
Clinical trials have demonstrated that imrecoxib shows 50% inhibitory concentration (IC 50 ) of COX-1 and COX-2 isozymes by 115 ± 28 nmol/L and 18 ± 4 nmol/L, respectively 5 . The selective index (IC 50 , COX-1/COX2 ) was 6.39, which was between that of meloxicam and celecoxib 21 . From a clinical perspective, whilst the lack of pharmacokinetic and pharmacodynamics effects on warfarin are important in terms of dose adjustment etc, the risk of bleeding due to GI irritation is still significant with NSAIDs (including a drug of relative COX-2 specificity) plus warfarin, particularly in the elderly. In addition, both S-warfarin, the more potent enantiomer of warfarin, and imrecoxib are metabolised by the CYP2C9 enzyme 8,22 . However, whether co-administration of imrecoxib and warfarin would result in DDI was not investigated. In this study, we evaluated the potential DDI of imrecoxib and warfarin by comparing the pharmacodynamic and pharmacokinetic parameters of warfarin with and without co-administration of imreocxib in healthy male volunteers. We also tested the safety and tolerability of study drugs across the trial 23 .

Methods
Ethics. Current study was conducted in conformity to the Declaration of Helsinki (as revised in Brazil, 2013) 24 , Good Clinical Practice (GCP) guidelines of China Food and Drug Administration (CFDA) 25 and the technical guidelines for clinical pharmacokinetic study of chemical drugs 26 . CFDA (no. 2011S00434) and the independent ethics committee (Tongji Medical College, Huazhong University of Science and Technology, no. (2014)185] reviewed and approved this study protocol. Written informed consent was required for every volunteer before any study procedures 27 . Subjects. Twelve subjects were enrolled in this study. The inclusion requirements were (i) male; (ii) BMI ranged from 19 to 24 kg ⁄m 2,26 ; (iii) aged between 18 to 40; (iv) qualified for complete health examination, including vital signs, electrocardiograms, routine blood test, urinalysis, biochemistry laboratory parameters, chest X-ray, liver and renal function tests are normal or not clinical significantly abnormal. (v) a condition of normal coagulation function (prothrombin time -PT, INR and fibrinogen) and negative serological test (HBsAg, HCV and HIV antibodies); (vi) voluntary signing of informed consent forms 27 .
As we previous reported 27 , subject would be excluded if he met these criterions: (i) hypersensitivity or allergy to the study drugs; (ii) any diseases or unstable medical history that may disturb the safety or the in vivo process of the study drugs, including cardiovascular, hepatic, renal, gastrointestinal, endocrine or immune system. (iii) a history of any bleeding disorders. (iv) diseases of nervous system or muscle diseases, that might affect subjects Study design. Current study is phase I clinical trial, which was designed as open-labeled and fixed-sequence, and all the information/data were collected from a single center. This study contained two phases (Fig. 2). In phase one, the volunteers received a 5 mg dose of warfarin alone at 8:00 a.m. on day 1. In the other phase, they orally took imrecoxib to steady-state (200 mg imrecoxib at 8:00 a.m. on day 8, and a 100 mg dose q.12 hours from day 8 to 10, 6 times in total), followed by a 5 mg dosage of warfarin co-administered at 8:00 a.m. on day 10. The volunteers were hospitalized on day-1 (the day before the study), 10 hours of fasting was required before administration 27 . Subjects should avoid any activities involved in risks of haemorrhage 9 . Blood samples (4 mL each) for analysis of pharmacokinetic parameters were obtained 60 minutes before dose of warfarin and 0. 5

Analytic methods. A stable LC-MS/MS method was established for detecting S-and R-warfarin plasma
concentrations. The chromatographic separation was carried out on an LC system (Shimadzu LC-20AD, Tokyo, Japan) using water and acetonitrile, and AB QTRAP 4000 system (AB Sciex, Foster City, CA, USA) in positive electrospray ion mode was hired for quantification [29][30][31] . Warfarin-d5 was used as the internal standard. Liquid-liquid extraction with 3 mL dichloromethane: diethyl ether: (2:3, v/v) was employed for 200 μL human plasma. Good linearity was obtained between 5.00-1000 ng/ml for each enantiomer 32 . The inter-and intra-precision (CVs% for 10, 100 and 800 ng/ml) were ≤5.2% for R-warfarin and ≤5.0% for S-warfarin, respectively. Inaccuracy for R-warfarin was between −6.4% to +4.2%, and ranged from −5.9% to +5.1% for S-warfarin. The mean absolute recovery was ≥87.3% (CVs <6.0%) 27,33 . Pharmacokinetics and pharmacodynamics analysis. As our previous studies reported 27 , pharmacokinetic analysis was performed base on plasma concentrations of warfarin enantiomers at each time-point by hiring Drug and Statistics Software version 3.1.5. The measurement outcomes contained area under the profile (AUC 0-t ), the terminal half-life (t 1/2 ), maximum plasma concentration observed (C max ), time of maximum concentration (T max ). AUC from 0 to infinity (AUC 0-∞ ). Parameters of pharmacodynamic were estimated from the INR data on each period. PT (INR) was measured with the use of prothrombin complex assay (STA-R, SPA 50 Reagent, Diagnostica stago) 34  Safety evaluations. The safety assessments were conducted on account of clinical examinations, such as evaluation of general subject appearance, vital signs and routine hematology and biochemistry assays 35 , together with adverse events evaluation (AEs), conducted at screening, pretreatment, post-treatment (day 7) and end of trial (day 16). Signs and symptoms relate to study drugs, such as nausea, diarrhea, vomiting, headache and dizziness, were observed and documented by the study physicians 36 . AEs were defined as mild, moderate or severe 37 . Determination of causal relationship between AEs and study drugs followed the criterions announced by the World Health Organization 27 .
Statistical methods. EpiData 3.0 software was used for data entry and management, statistical analysis was conducted on SAS 9.3 software programming (SAS Institute Inc., Cary, NC). The statistical significance was accepted with two-sided p < 0.05 38 . Pharmacokinetic and pharmacodynamic analyses were based on the subjects who finished trial without great program violation which have a major impact on pharmacokinetic and pharmacodynamic parameters. Descriptive statistics such as mean, median, range, and standard deviation were calculated for observed variables.
Log-transformation of pharmacodynamic parameters INR max and INR (AUC) were applied. Comparing the difference between warfarin treatment and combination treatment for T max , logINR max and logINR (AUC) used the F test in ANOVA analysis. The GMR and 90%CIs were calculated by back-transforming for AUC 0-∞ , AUC 0-144h , C max , INR max , T max and INR (AUC) . The 90% CIs felled within the acceptance range of 0.80-1.25 suggest a lack of drugs interaction 28 .  Safety and tolerability. Both warfarin alone and concomitantly with imrecoxib have safey and good tolerance in healthy volunteers across the trial. Neither severe AEs nor accidental bleeding events occurred during the trial. All the data or information of physical examination, vital signs, laboratory test results or 12-lead ECG were not meaningful altered compare to before administration. In period 1, one subject had transient elevated direct bilirubin (9.6 μmol/L, upper limit of normal = 6.8 μmol/L) on day 7, which met the definition of grade 1 AEs (>ULN -1.5 × ULN in direct bilirubin). However, the subject did not have any associated signs or symptoms and the level of direct bilirubin stayed normal on day 16 (period 2). Therefore, the investigator considered it to be unrelated to study drugs. In period 2, one subject was observed to be experiencing mild abdominal/upper abdominal discomfort on Day 8 after the first dose of imrecoxib, which continued for about 1.5 hours and disappeared without medical treatment. This event was regarded as possibly related to the drugs. No volunteer dropped out from the trial due to adverse experiences.
Pharmacokinetics. The pharmacokinetic parameters and pharmacokinetic curves of warfarin enantiomer both warfarin alone and concomitantly with imrecoxib are listed below (Table 2, Fig. 3). Concomitant administration of imrecoxib and warfarin did not change the median T max value of R-and S-warfarin 0.8 (0.5~2.0) hours compared to 1.0 (0.5~3.0) hours with administration of warfarin alone (P > 0.300). The t 1⁄2 of R-warfarin for recipients of warfarin alone and recipients of co-administration of imrecoxib and warfarin were 64.08 ± 15.97 hours and 59.02 ± 9.39 hours, respectively (P > 0.1). In the absence and presence of imrecoxib, the t 1⁄2 for S-warfarin were 57.00 ± 15.27 hours and 51.63 ± 7.59 hours respectively (P > 0.1). Receiving imrecoxib did not change V z /F of R-warfarin, however, a decrease of 16% was observed for V z /F of S-warfarin, the mean V z /F slightly decreased from 12.65 ± 2.55 to 10.61 ± 1.89 (P = 0.01 for co-administration of imrecoxib versus warfarin alone treatment). As summarized in Table 3, compare imrecoxib and warfarin in combination to warfarin alone, the GMR of R-warfarin AUC 0-144h and C max were 1.14 and 1.06, respectively, and the 90% CIs ranged from 0.93-1.13 and 0.98-1.15, both of which were within 0.8-1.25. For the S-warfarin enantiomer, the GMR of C max and AUC 0-t were 1.03 and 1.14, and the corresponding 90% CI were 0.93-1.13 and 1.09-1.20. All 90% CIs were in the range of 0.80-1.25. These results suggest that the pharmacokinetic profiles of S-and R-warfarin were not significantly impacted by co-administration of imrecoxib.  www.nature.com/scientificreports www.nature.com/scientificreports/ no significant difference for log INR max , logINR(AUC) and T max during concurrent imrecoxib treatment compared with warfarin alone treatment (Table 5).

Discussion
This study revealed the pharmacodynamics and pharmacokinetics of warfarin would not be altered by concomitant administration of imrecoxib with the clinically recommended dosage. As an innovative and mild selective COX-2 inhibitor, imrecoxib can probably be prescribed to patients with cardiovascular disease and stable long-term warfarin therapy 4 . Several studies indicated an increasing INR value of healthy volunteers and accidental bleeding in patients stable on warfarin therapy after dosing celecoxib, which with similar therapeutic efficacy and side effects to imrecoxib [39][40][41][42][43] . Monitoring the INR of long-term warfarin recipients is required to optimize effective dosage because of a large inter-individual variation and narrow therapeutic window 44 . S-warfarin is     www.nature.com/scientificreports www.nature.com/scientificreports/ metabolized by CYP2C9 enzyme 8,22,45 , as well as imrecoxib. S-warfarin directly inhibits vitamin K-dependent coagulation factors 46 , and accounts for 85% anticoagulant activity of warfarin 22 . Competition of the CYP2C9 metabolic enzyme may occur when patients receive warfarin and imrecoxib, which prevents S-warfarin from being metabolized to S-7-hydroxywarfarin, resulting in an increase of plasma concentration and anticoagulant effects of S-warfarin. Both warfarin (99%) and imrecoxib (96%) are highly protein bound in plasma. Imrecoxib may competitively displace warfarin from the protein-binding sites, enhancing blood concentration of free warfarin and increasing bleeding risks. Thus, we speculated that imrecoxib and warfarin may interact.
Inconsistent with our speculation, the results indicated the pharmacokinetic profiles of warfarin enantiomers were not significantly changed by co-administration of imrecoxib. Comparing co-administered warfarin and imrecoxib with warfarin alone, for S-warfarin, the outer bound of 90% CIs of AUC 0-144h increased to 20% (Table 3), but AUC 0-144h and AUC 0-∞ were not significantly changed (Table 2), and the GMRs for and 90% CIs for AUC 0-∞ , AUC 0-144h and C max were all within 0.80-1.25 (Table 3). PT were expressed by an INR value in this study. Monitoring of PT is required for individualized dosage adjustments in clinical warfarin use. There were no meaningful disparities in T max , logINR (AUC) and logINR max observed between two treatments. Although the mean INR at 12 hour, near the T max , was significantly reduced when dosed with imrecoxib, the GMR and 90% CI of INR AUC 0-144 h for warfarin + imrecoxib: warfarin only were near identical, 1 (0.99, 1.01) ( Table 4), and no significant difference in logINR max was observed (Table 5). These results suggested imrecoxib would not alter the pharmacokinetics parameters and anticoagulation activity of warfarin, but greater caution should be taken in the wider applicability of the results.    www.nature.com/scientificreports www.nature.com/scientificreports/ It has been widely agreed that the anticoagulant efficiency of warfarin is highly related to genetic polymorphisms. Among these genes, CYP2C9 and VKORC1 are responsible for 30% to 40% of the warfarin efficiency differentiation [47][48][49][50][51] . People with these polymorphisms show a significant difference in warfarin pharmacodynamic and pharmacokinetic profiles compared to wildtype subjects. For a better evaluation, all volunteers enrolled in this study were CYP2C9*3 AA genotype and VKORC1 (G-1639A) with homozygous AA genotype. Several studies have reported the frequency of CYP2C9 *3 AA and AC genotypes were 95% and 5%, respectively, and mutation frequency of VKORC1-1639 AG and AA were 7.4% and 92.6% in the Han-Chinese population [52][53][54][55][56][57] . Warfarin-induced hemorrhage associated with age 58 . Significant reduction in clearance of warfarin with age was also reported 59 . Healthy volunteers aged from 18 to 45 years old are recommended by guideline 26 . However, in a large Japanese reports analysis, the reporting odds ratio of hemorrhagic events associated warfarin in patients age 40-49 significantly lower than those aged ≤40 or those aged ≥50 60 . In addition, many studies focusing on age and warfarin's efficiency divided the volunteers' age into young and elderly groups. The age range of these groups was 18-40 and 65-90 respectively. Therefore, we enrolled the volunteers between 18 and 40 years old to rule out the influence of age on warfarin.
A loading dose of 200 mg imrecoxib was chosen, then subsequently taking 5 continuous 100 mg doses of imrecoxib in order, to guarantee imrecoxib reaches its steady-state concentration prior to warfarin dose in this study. Clinically recommended dosage is 5 mg for warfarin and 100 mg for imrecoxib. The dosage of warfarin used in some studies was 25 mg 61-63 . We used 5 mg warfarin in both periods, to ensure adequate plasma drug levels close to common clinic levels while avoiding exposing participants to unnecessary bleeding risks caused by excessive use of warfarin. Consistent with our study, the existence of an interaction between warfarin (5 mg/d) and celecoxib was evaluated in a study with 24 healthy volunteers 64-66 , and 7.5 mg warfarin was used to examine potential drug interactions with celecoxib in healthy volunteer studies 28,67 .
In conclusion, this study revealed that co-administration of imrecoxib did not affect the pharmacokinetic parameters or anticoagulant properties of warfarin. Thus, we concluded that adjusting dosage is not necessary when administering imrecoxib concomitantly with warfarin. However, we only conducted a single dose study of warfarin, so the possibility that a higher dosage or multiple doses of warfarin would alter its pharmacokinetic or pharmacodynamic profiles during co-administration with imrecoxib could not be excluded, though the clinical relevance would be doubtful.