Comparative Impact on Prostanoid Biosynthesis of Celecoxib and the Novel Nonsteroidal Anti-Inflammatory Drug CG100649

Abstract

Nonsteroidal anti-inflammatory drugs (NSAIDs) elevate cardiovascular risk by disrupting cyclooxygenase-2 (COX-2)-dependent biosynthesis of prostacyclin (PGI₂). CG100649 is a novel NSAID proposed to inhibit both COX-2 and carbonic anhydrase (CA)-I/-II. We compared its impact on prostanoid biosynthesis with that of celecoxib, an NSAID purposefully designed to selectively inhibit COX-2. In a controlled, double-blind randomized trial, single oral doses of 2 or 8 mg CG100649, 200 mg celecoxib, or placebo were well tolerated by healthy volunteers (n = 23). Both CG100649 and celecoxib had the effect of depressing urinary excretion of 2,3-dinor-6-keto-PGF_1α (PGI-M); the effect of CG100649 was dose-dependent and more sustained (up to 240 h after the dose) than that of celecoxib. Neither CG100649 nor celecoxib significantly inhibited COX-1-dependent prostanoid formation. CA inhibition was not detected after administration of CG100649, despite its partitioning asymmetrically into erythrocytes. CG100649 and celecoxib are both relatively selective inhibitors of COX-2, but they differ in duration of action. Whether they have similar impact on cardiovascular events remains to be determined.

INTRODUCTION

Evidence consistent with a mechanism-based cardiovascular hazard from nonsteroidal anti-inflammatory drugs (NSAIDs) due to inhibition of COX-2-dependent formation of prostacyclin (PGI₂) has emerged from clinical pharmacology, proof-of-concept studies in rodents and other species, observational studies, and human genetics. PGI₂ acts as a restraint on endogenous stimuli that promote thrombosis, hypertension, atherogenesis, and cardiac damage in vivo.¹ Since the withdrawal of rofecoxib in 2004, seven placebo-controlled trials have proven that three structurally distinct NSAIDs purposefully developed to inhibit selectively COX-2 (pdNSAIDs)—rofecoxib, valdecoxib, and celecoxib—are associated with an increase in cardiovascular risk. Less conclusive evidence suggests that this may also be true of some traditional NSAIDs (tNSAIDs), such as diclofenac.¹ Although no placebo-controlled trials of tNSAIDs have measured cardiovascular risk in large populations, the results of comparative randomized trials among tNSAIDs are consistent with what would be expected based on this mechanism.²

Hypertension is a feature of the NSAID-induced hazard, and both experimental³ and clinical⁴ evidence is compatible with the notion that the rise in blood pressure (BP) in patients taking an NSAID is reflective of inhibition of COX-2 and the selectivity with which it is attained. Indeed, selective deletion of COX-2 in vascular cells augments the hypertensive response to a high-salt diet in rodents, and this response is inversely correlated with suppression of biosynthesis of PGI₂.⁵ However, although hypertension is a consequence of disruption of vascular COX-2-dependent PGI₂ formation, myocardial infarction is more clearly related to pdNSAID intake than stroke is,⁶ suggesting that enhanced thrombogenesis is the more clinically dominant consequence of PGI₂ suppression.² CG100649 is a novel pdNSAID that, with its postulated dual inhibition of COX-enzymes and carbonic anhydrase-I/-II (CA-I/-II), might limit the propensity for NSAID-induced hypertension and thereby attenuate the related cardiovascular risk. Its biodistribution profile in preclinical models (Investigational Brochure CG100649 v3.4) shows a marked partitioning into erythrocytes due to extensive binding to CA-I/-II. This is evident in pharmacokinetic and tolerability studies in humans as a disproportionate accumulation of CG100649 in erythrocytes, resulting in 85- to 100-fold higher concentrations of CG100649 in whole blood as compared with plasma. The current study was designed to address more comprehensively the impact of CG100649 on prostanoid biosynthesis and CA-I/-II function. The objective was to determine whether CG100649 possesses features suggestive of a cardiovascular risk profile different from that of the approved pdN-SAID, celecoxib.

RESULTS

Subjects

Twenty-seven healthy volunteers were enrolled in this randomized, double-blind, parallel-design study to compare celecoxib (200 mg) with placebo in period I followed by a parallel-group comparison of CG100649 (2 mg), CG100649 (8 mg), celecoxib (200 mg), and placebo in period II (Figure 1, Table 1). The planned crossover analysis between periods I and II could not be completed because of a randomization error in subject allocation. One subject was excluded from the analysis for period II data because of a celecoxib medication error (1,600 mg, n = 1, female, African-American, 32 years of age). In the period II treatment arms CG100649 (8 mg) and placebo, urine and blood specimens for one subject in each group at post-treatment time points of ≥746 h and ≥168 h, respectively, were not included in the data analysis because of protocol violations, namely, intake of nonapproved medications at those time points (Figure 1, Table 1).

Figure 1.

CONSORT flow diagram. *Randomization error, that is, period II treatment assignments, were unlinked to period I treatments, thus not achieving the planned crossover group sizes of n = 6.

Table 1.

Demographic profile of the study participants

	Treatment assignments	Period I	Period I	Period I	Period II	Period II	Period II	Period II	Period II
		Celecoxib (200 mg)	Placebo	Total	Celecoxib (200 mg)	Placebo	CG100648 (2 mg)	CG100648 (8 mg)	Total
Number of subjects		17	9	26	6	6	7	6	25
Sex	Male	9 (53%)	3 (33%)	12 (46%)	6 (100%)	3 (50%)	2 (29%)	1 (17%)	12 (48%)
Age	Mean ± SD	31.2 ± 7.4	30.6 ± 8.9	31.0 ± 7.7	31.5 ± 6.6	29.8 ± 9.9	34.4 ± 8.5	27.3 ± 6.1	30.9 ± 7.9
	Median (Q1, Q3)	31.0 (26.0,37.0)	30.0 (23.0,31.0)	30.5 (25.0,37.0)	32.5 (27.0,37.0)	28.0 (25.0,31.0)	31.0 (27.0,44.0)	26.5 (23.0,30.0)	30.0 (25.0,37.0)
	Range	19–44	21–48	19–48	21–39	19–48	23–44	21–37	19–48
Ethnicity	Non-Hispanic	16 (94%)	9 (100%)	25 (96%)	6 (100%)	6 (100%)	6 (86%)	6 (100%)	24 (96%)
Race	Asian/Asian American	2 (12%)	1 (11%)	3 (12%)	1 (17%)	1 (17%)	0 (0%)	1 (17%)	3 (12%)
	Black/African American	4 (24%)	0 (0%)	4 (15%)	1 (17%)	0 (0%)	2 (29%)	0 (0%)	3 (12%)
	White/Caucasian	11 (65%)	8 (89%)	19 (73%)	4 (67%)	5 (83%)	5 (71%)	5 (83%)	19 (76%)

Safety

The single oral doses of CG100649 (2 mg/8 mg) and celecoxib (200 mg) were well tolerated. The overall reported cumulative adverse events (AEs) in subjects experiencing at least one AE were n = 8 and n = 15 for periods I and II, respectively (Table 2). The number of subjects with AEs was similar between active and inactive treatment groups in the interindividual comparison of period II: n = 5 for placebo vs. n = 5 for CG100649 (8 mg), n = 4 for CG100649 (2 mg), and n = 2 for celecoxib (Supplementary Table S2 online). The between-subject comparison of celecoxib (200 mg) vs. placebo in period I resulted in n = 5 and n = 3, respectively, for the highest grade of AEs (Supplementary Table S1 online). Left ventricular hypertrophy was reported in two subjects receiving CG100649 (8 mg), in one as a pre-existing condition and in the other at study exit; both were interpreted as benign and reflective of electrocardiographic signs of regular physical activity.⁷ Serious AEs did not occur. The single trial medication error of 1,600 mg celecoxib was not associated with signs of acute toxicity, consistent with doses up to 2,400 mg being well tolerated.⁸ The subject’s BP on this treatment remained within the range (systolic 113–122 mm Hg and diastolic 68–77 mm Hg) observed after administration of placebo.

Table 2.

Adverse events per body system and toxicity grade

Body system	Period I toxicity grade (n = 26)					Period II toxicity grade (n = 25)
	None	Normal (Gr0)	Mild (Gr1)	Moderate (Gr2)	Total with at least one adverse eventa	None	Normal (Gr0)	Mild (Gr1)	Moderate (Gr2)	Total with at least one adverse eventa
Highest grade of adverse event	18	2	5	1	8 (30.8%)	10	1	12	2	15 (60.0%)
Cardiac disorders	24	0	2	0	2 (7.7%)	23	0	2	0	2 (8.0%)
Eye disorders	26	0	0	0	0	24	1	0	0	1 (4.0%)
Gastrointestinal disorders	26	0	0	0	0	21	0	4	0	4 (16.0%)
General disorders and administration-site conditions	24	2	0	0	2 (7.7%)	23	0	2	0	2 (8.0%)
Immune system disorders	26	0	0	0	0	22	0	3	0	3 (12.0%)
Infections and infestations	25	0	1	0	1 (3.8%)	21	0	3	1	4 (16.0%)
Injury, poisoning, and procedural complications	25	0	1	0	1 (3.8%)	25	0	0	0	0
Musculoskeletal and connective tissue disorders	26	0	0	0	0	23	0	1	1	2 (8.0%)
Nervous system disorders	23	1	2	0	3 (11.5%)	22	1	2	0	3 (12.0%)
Respiratory, thoracic, and mediastinal disorders	22	3	1	0	4 (15.4%)	18	4	3	0	7 (28.0%)
Skin and subcutaneous tissue disorders	25	0	0	1	1 (3.8%)	23	0	2	0	2 (8.0%)

^aThe total number of subjects who have at least one adverse event.

Drug concentrations

CG100649 (2 mg) resulted in an area under the time–concentration curve (AUC)_{(0–1,176 h post-dose)} of 632.9 ± 162.1 ng/ml × h, whereas CG100649 (8 mg) achieved 2,366.8 ± 761.9 ng/ml × h, corresponding to 3.7 times that with the low dose. This was consistent with dose linearity in drug exposure (Figure 2, left). For celecoxib (200 mg), the AUC was 5,010.5 ± 1468.6 ng/ml × h (Figure 2, right; period I AUC_{(0–72 h)} and period II AUC_{(0–1,176 h)} pooled as AUC_{celecoxib, plasma}, less than the lower limit of quantification (LLOQ) for >72 h), which is consistent with earlier reports.^9,10 C_max-plasma was 3.5 ± 0.9 and 14.1 ± 3.7 ng/ml for CG100649 (2 mg) and (8 mg), respectively, supporting the hypothesis of dose linearity for peak plasma drug concentrations. In comparison, C_max-plasma was 889.9 ±209.3 ng/ml for celecoxib (200 mg). The t_max values were 5.6 ± 1.0 and 5.0 ± 1.7 h for CG100649 (2 mg) and (8 mg), respectively, as compared with 2.9 ± 1.0 h for celecoxib (200 mg). CG100649 concentration showed a whole-blood/plasma ratio of 78.6 ± 16.6 (range 117.6–54.3) in the n = 13 subjects with exposure to CG100649. This ratio was 0.9 ± 0.23 (range 1.17–0.66) for celecoxib data from a subset of 5 subjects. The single dose of celecoxib (1,600 mg) in period II resulted in a C_max-plasma of 2,086 ng/ml, t_max at 2 h, and an AUC_{(0–1,176 h)} of 43,094.1 ng/ml × h, the latter being 7.3-fold higher than the exposure attained by the 200-mg dose in period I for the same subject.

Figure 2.

CG100659 drug concentrations in plasma and whole blood (left) after single oral administrations of CG100659 (2 mg) and (8 mg). Both dose levels demonstrate the marked partition of CG100649 into erythrocytes. Celecoxib plasma concentrations after single oral administration of 200 mg show comparable drug exposure profiles for periods I and II (right). Note that the ordinate for CG100649 drug concentrations as well as both abscissas are log-scaled to capture the wide range.

Urinary excretion of PGI₂ metabolite, 2,3-dinor-6-keto PGF_1α (PGI-M)

The percentage change from baseline for PGI-M AUCs tended to differ among the four treatment groups (P = 0.015) but did not reach statistical significance. This seems to be driven mostly by the difference in the median (range) change in AUCs between CG100649 (8 mg), −338.1 (−423.4 to −116.4)% × log(h), and placebo, −55.0 (−253.8 to 33.8)% × log(h). Celecoxib and CG100649 (2 mg) were associated with AUCs of −142.4 (−379.3 to 18.6) and −176.0 (−239.5 to −33.9)% × log(h), respectively. Group sample sizes were inadequate for further statistical analyses.

CG100649 (2 mg) depressed PGI-M excretion (Figure 3) by −53.3 (−69.6 to 12.5)% and −44.4 (−73.3 to −20.0)% at 4–6 and 12–24 h, respectively, after the dose; at the same postdose intervals, CG100649 (8 mg) depressed PGI-M by −66.7 (−78.9 to −46.2)% and −74.1 (−80.0 to −30.8)%. Notably, this magnitude of PGI-M inhibition was sustained up to 228–240 h (−57.9 (−75.0 to −39.1)%) after a single dose of CG100649 (8 mg). In comparison, celecoxib (200 mg) maximally suppressed PGI-M at 4–6 and 12–24 h after the dose by −61.7 (−80.0 to −37.5)% and −41.7 (−81.8 to 0.0)%, respectively. There was considerable variability in PGI-M levels in urine after placebo, amounting to −25.0 (−37.5 to −9.1)% and −15.9 (−45.8 to 36.4)% change from baseline for the collection intervals 4–6 and 12–24 h after the dose.

Figure 3.

Urinary PGI-M concentrations over time expressed as percentage change from baseline. The horizontal reference line indicates the percentage change from zero. Box plots indicate mean (dot), median (center line of the box), first and third interquartile ranges (lower and upper lines), and minimum and maximum values (lower and upper whiskers). PGI, metabolite of prostacyclin.

Celecoxib (200 mg) maximally inhibited PGI-M by −58.3 (−76.09 to 50)% at 4–6 h after administration, as compared with a −9.3 (−65.22 to 100)% change after placebo in the parallel-group comparison (period I; P = 0.085; Supplementary Figure S15 online). No significant difference was evident in a comparison of the AUCs_{(0–72 h)} (AUC_celecoxib of −80.8 (−221.0 to 260.3)% × log(h) vs. AUC_placebo of −26.6 (−215.1 to 158.9)% × log(h); P = 0.83).

TxB₂ generation

The percent change from baseline of serum thromboxane (TxB₂) AUCs did not differ significantly among the four treatment groups (P = 0.12). The AUC was 4.5 (−281.9 to 567.4)% × log(h) for the treatment condition CG100649 (2 mg), −194.8 (−299.2 to −19.4)% × log(h) for CG100649 (8 mg), 19.2 (−169.0 to 388.6)% × log(h) for celecoxib, and −44.6 (−158.0 to 94.6)% × log(h) for placebo.

The temporal profile showed a decrease in TxB₂ formation by −22.8 (−48.0 to 4.1)% and −23.4 (−29.6 to 50.9)% at 2 and 3 h, respectively, after celecoxib administration, as compared with changes of −9.8 (−30.2 to 8.8)% and 7.2 (−31.6 to 25.3)%, respectively, after placebo. CG100649 (8 mg) depressed TxB₂ formation by −23.9 (−49.1 to −5.0)% and −17.9 (−44.8 to 2.7)%, respectively, at the two time points, and this effect persisted at the same level throughout the entire period of observation. In the case of CG100649 (2 mg), the observed decrease in TxB₂ formation by −13.1 (−31.5 to 191.3)% and −14.1(−36.4 to 227.7)% at 2 and 3 h after the dose is suggestive of drug-induced inhibition of COX-1. However, the mean ± SD drug effect data were biased by outliers (Figure 4, Table 3). These changes in TxB₂ formation failed to attain statistical significance in the comparison of summary measures of AUC_{(0–1,176 h)} across treatment groups (P = 0.12).

Figure 4.

Serum TxB₂ concentrations over time expressed as percentage change from baseline. The reference line indicates the percentage change from zero. TxB₂, serum thromboxane.

Skarke et al. Page 13

Clin Pharmacol

Table 3.

Concentrations of prostaglandins, prostaglandin metabolites, and indexes of carbonic anhydrase activity at predose baseline, as well as cumulative aggregate (area under the curve) to quantify drug effects

Treatment groups	Primary end points		Secondary end points
	PGI-M baseline (ng/mg cre)	PGI-M AUC0–1,176 (% change from baseline); P = 0.015	PGE-M baseline (ng/mg cre)	PGE-M AUC0–1,176 (% change from baseline); P = 0.228	11-deh-TxB2 baseline (ng/mg cre)	11-deh-TxB2 AUC0–1,176 (% change from baseline); P = 0.032	2,3-dinor-TxB2 baseline (ng/mg cre)	2,3-dinor-TxB2 AUC0–1,176 (% change from baseline); P = 0.521	PGD-M baseline (ng/mg cre)	PGD-M AUC0–1,176 (% change from baseline); P = 0.409
	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD
Celecoxib (200 mg), n = 6	0.21 ± 0.19	−175.2 ± 142.8	20.8 ± 16.9	−30.5 ± 164.1	0.5 ± 0.2	56.2 ± 221.0	0.2 ± 0.1	102.1 ± 212.3	3.1 ± 2.1	−1.2 ± 204.9
Placebo, n = 6	0.14 ± 0.07	−71.7 ± 107.0	14.6 ± 11.0	−41.1 ± 101.2	0.7 ± 0.4	37.5 ± 56.7	0.3 ± 0.1	−4.1 ± 50.8	4.4 ± 1.7	−107.9 ± 163.1
CG100649 (2 mg), n = 7	0.16 ± 0.05	−152.1 ± 73.1	11.4 ± 8.2	150.6 ± 484.6	0.6 ± 0.3	−35.6 ± 138.5	0.2 ± 0.1	−1.5 ± 248.1	3.3 ± 1.8	−13.3 ± 187.9
CG100649 (8 mg), n = 6	0.18 ± 0.04	−315.3 ± 108.2	11.9 ± 6.9	−210.9 ± 166.1	0.7 ± 0.1	−155.6 ± 74.1	0.2 ± 0.1	−70.1 ± 129.2	3.5 ± 1.5	−155.9 ± 256.2
Treatment groups	TxB2 baseline (ng/ml)	TxB2 AUC0–1,176 (% change from baseline); P = 0.120	LPS-PGE2 Baseline (ng/ml)	LPS-PGE2 AUC0–1,176 (% change from baseline); P = 0.669	HCO3− baseline (mmol/l)	HCO3− AUC0–1,176 (% change from baseline); P = 0.194	pH baseline	pH AUC0–1,176 (% change from baseline); P = 0.648	Cl− baseline (mmol/l)	Cl− AUC0–1,176 (% change from baseline); P = 0.935
	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD
Celecoxib (200 mg), n = 6	541.5 ± 248.7	45.0 ± 202.4	27.9 ± 23.2	−2.7 ± 436.6	24.3 ± 1.9	0.0 ± 47.5	7.4 ± 0.0	−0.1 ± 6.1	101.8 ± 3.0	1.8 ± 13.0
Placebo, n = 6	401.3 ± 201.1	−41.0 ± 96.0	13.3 ± 6.3	−3.3 ± 229.9	24.9 ± 2.7	−80.5 ± 46.3	7.4 ± 0.0	−3.8 ± 4.9	103.2 ± 2.8	3.5 ± 25.3
CG100649 (2 mg), n = 7	489.8 ± 225.9	6.2 ± 280.0	23.4 ± 16.9	−173.7 ± 190.3	20.5 ± 5.5	82.4 ± 252.9	7.4 ± 0.1	0.2 ± 8.8	102.9 ± 2.7	8.1 ± 12.6
CG100649 (8 mg), n = 6	454.2 ± 201.9	−177.8 ± 104.7	14.9 ± 7.1	−68.6 ± 63.6	20.2 ± 6.8	175.9 ± 495.7	7.4 ± 0.1	1.2 ± 7.6	104.3 ± 4.5	4.6 ± 25.9

AUC, area under the time–concentration curve; deh, dehydro; LPS, lipopolysaccharide; PGE₂, prostaglandin E₂; PGI, prostacyclin; TxB₂, serum thromboxane.

The parallel-group comparison (period I) did not suggest a celecoxib-induced inhibition of TxB₂ (Supplementary Figure S15 online); at 2 and 3 h after the dose, decreases in TxB₂ formation by 9.1 ± 27.7% and 7.4 ± 29.6%, respectively, were observed for celecoxib. For placebo, these changes were −26.8 ± 28.6% and −26.2 ± 31.2%, respectively (AUC_{(0–72 h-celecoxib)} −15.6 (−196.8 to 101.9)% × log(h) vs. AUC_{(0–72 h-placebo)} −100.0 (−195.3 to 68.6)% × log(h); P = 0.21).

Secondary outcome measures

The percentage change of the AUCs from baseline values did not differ significantly among the four treatment groups for PGE-M (9,15-dioxo-11α-hydroxy-2,3,4,5-tetranor-prostan-1,20-dioic acid), urinary excretion of 2,3-dinor-TxB₂, PGD-M (9α-hydroxy-11,15-dioxo-2,3,4,5-tetranor-prostan-1,20-dioic acid), and lipopolysaccharide (LPS)-induced PGE₂ formation ex vivo. Only urinary excretion of 11-dehydro-TxB₂ tended to differ among the four treatment groups (P = 0.032); this seems to be driven mostly by the difference in the change in AUCs between the modest inhibitory effect of CG100649 (8 mg), −160.2 (−258.6 to −45.6)% × log(h), and the neutral effect of celecoxib (200 mg), 51.0 (−179.5 to 331.4)% × log(h).

CG100649 (8 mg) depressed PGE-M excretion at 4–6 h after the dose (−65.5 (−71.9 to −36.0)%) and at 12–24 h after the dose (−53.3 (−73.7 to −22.1)%) (Supplementary Figure S1 online). Furthermore, CG100649 (8 mg) depressed the urinary excretion of 11-dehydro-TxB₂ at 4–6 h (−37.6 (−46.4 to −8.2)%), at 12–24 h (−38.9 (−46.4 to −20.0)%), and at 60–72 h (31.3 (−56.5 to −2.0)%) after the dose (Supplementary Figure S3 online). A similar pattern of drug-induced suppression was observed for urinary excretion of 2,3-dinor-TxB₂ (Supplementary Figure S4 online) and PGD-M (Supplementary Figure S5 online). LPS-induced PGE₂ formation ex vivo—a capacity-related index of COX-2 function¹¹—was depressed by CG100649 (2 mg) at 3, 6, and 8 h after the dose by −49.3 (−57.9 to −6.9)%, −54.5 (−80.7 to −14.2)%, and −44.5 (−71.0 to 17.3)%, respectively, and by CG100649 (8 mg) at 6 h and 8 h after the dose by −30.2 (−67.4 to 32.3)% and −27.1 (−65.1 to 22.9)%, respectively (Supplementary Figure S2 online). Table 3 shows baseline concentrations and treatment effects on prostanoid formation.

Inhibition of carbonic anhydrase

There were no differences in arterialized bicarbonate, pH, and chloride among any of the four treatment groups (Table 3, Supplementary Figure S6–S8 online).

DISCUSSION

Placebo-controlled trials have established that the pdNSAIDs rofecoxib, celecoxib and valdecoxib are associated with cardiovascular hazard. Mechanistically, this has been attributed to suppression of COX-2-dependent production of cardioprotective prostanoids, especially PGI₂.1 Although this hazard may apply to only 1–2% of patients exposed, it has led to denial of approval or removal from the market of several pdNSAIDs. In our study, we sought to determine whether a novel compound, CG100649, might offer a more benign pharmacologic profile than that of other pdNSAIDs. Preclinical studies revealed that CG100649 partitioned asymmetrically to red blood cells in rats and bound to CA; it was hypothesized that functional inhibition of CA might attenuate the hypertensive propensity of COX-2 inhibition.

The study was designed to compare CG100649 with celecoxib in a placebo-controlled design between subjects, but it failed because of a randomization error—additional celecoxib or placebo administrations were not matched to allow for within-subject comparisons. This study was planned as the first step in an iterative drug development process to compare CG100649 against active comparators of differing COX-1/COX-2 selectivities as well as to a CA inhibitor.

CG100649 had a longer half-life than celecoxib in this acute-dosing study, and both drugs suppressed urinary excretion of PGI-M. In the case of celecoxib, this was a relatively transient effect, although it could be expected to be sustained when the typical clinical dose of 200 mg twice daily is administered under steady-state conditions. CG100649 suppressed urinary excretion of PGI-M in a dose-dependent manner and, as expected from its extended half-life, for a longer period as compared with celecoxib. Peak suppression attained after CG100649 (8 mg) corresponded to peak suppression after celecoxib (200 mg). Repeated single-dose exposures to celecoxib (200 mg) consistently caused a transient depression of urinary excretion of PGI-M, thereby supporting the robustness of the observed drug responses despite the high degree of variability in prostaglandin biosynthesis.¹²

The observed suppression of urinary PGI-M consequent to dosing in healthy volunteers was the initial basis for predicting that cardiovascular hazard is associated with celecoxib¹³ and rofecoxib.¹⁴ Deletion of COX-2 in endothelial or vascular smooth muscle cells, resulting in a propensity to thrombosis and hypertension, suppresses urinary PGI-M.⁵ Unlike that with tNSAIDs, the inhibition of platelet COX-1-derived TxA₂ with these drugs is minimal and irrelevant with respect to platelet function, because suppression of platelet Tx-dependent aggregation requires inhibition of >95% of the capacity of platelets to elaborate this prostanoid.¹⁵ In the case of CG100649, the measurement of a urinary Tx metabolite suggested a modest inhibition of TxA₂ biosynthesis. The relatively low magnitude of this effect and the lack of a significant impact on serum TxB₂ make it unlikely that this is relevant to platelet inhibition or, specifically, that it represents a restraint on thrombogenesis consequent to suppression of PGI₂. Consistent with this observation, urinary PGD-M, which also derives largely from COX-1,¹⁶ was also modestly suppressed by CG100649.

The eightfold overdose of celecoxib (1,600 mg) in one volunteer did not produce inhibition of prostaglandin or thromboxane to an extent distinguishable from that caused by the 200 mg dose, despite the dose-dependent exposure kinetics of the drug; however, our observation is in concordance with the known effects of celecoxib on these parameters.^13,17

Using a variety of indirect approaches we found no evidence consistent with functional inhibition of CA, but we found evidence consistent with a high affinity of CG100649 for erythrocyte CAs, as judged from its very high concentration in whole blood vs. plasma. A caveat is that we did not include a positive control, such as acetazolamide;¹⁸ it is also possible that a CA-inhibiting effect may be detectable at higher doses or under chronic dosing conditions. The study was not designed to evaluate drug impact on BP, but there was a tendency for BP to be lower after CG100649, whereas no such trend was apparent, per visual inspection, after celecoxib (Supplementary Figure S10–S13 online). If there is actually such a blood pressure–lowering effect, the finding of a more pronounced inhibition of COX-1 after CG100649 than after celecoxib might have greater relevance; inhibition of Tx synthesis may relate more linearly to an impact on BP than the inhibition of platelets. This possibility could be readily investigated under chronic dosing conditions in patients with mild or evoked hypertension.

In summary, CG100649 is a novel, long-acting NSAID. It is relatively selective for inhibition of COX2, exerting a more sustained inhibitory eftect on PGI₂ biosynthesis as compared with celecoxib, consistent with comparative drug exposure. Its relative impact on COX-1-derived prostanoids is unlikely to restrain platelet activation. Whether this property restrains the propensity of pdNSAIDs such as celecoxib to elevate BP in individuals who are predisposed to such a response⁴ remains to be determined.

METHODS

Study design

The effects of CG100649 (2 and 8 mg administered as single doses) were compared with those of celecoxib (200 mg) and placebo, by randomizing subjects to either celecoxib or placebo in period I, followed ≥7 days later by random allocation to CG100649 (2 mg) or CG100649 (8 mg) or celecoxib (200 mg) or placebo. There were no preliminary data on CG100649’s impact on prostanoid metabolite excretion in humans; hence, a formal power analysis was not executed, and group (period II) and sequence (period I > period II) sample sizes were set to n = 6 (Supplementary Figure S14 online). However, because of a departure from the randomization schedule, the sequence sample sizes were not met. This study was compliant with International Conference on Harmonisation and Good Clinical Practice (ICH-GCP) standards as well as with those of the Declaration of Helsinki (1996). The study protocol and subject information and consent were reviewed and approved by the institutional review board of the University of Pennsylvania. This trial was registered at ClinicalTrials.gov (NCT00780325) before subject enrollment.

Subjects

We enrolled nonsmoking healthy volunteers (18–60 years of age) who had abstained from prescribed and over-the-counter medications in the 2 weeks before screening (30 days for experimental drug/medical devices and CYP450-modulating drugs) and throughout the study. Pregnant or breastfeeding women were not enrolled. They provided written informed consent and were eligible for study recruitment based on medical history, physical examination, and routine hematology, biochemistry, and urinalysis results. They were enrolled and studied under controlled conditions in the Clinical Translational Research Center. The Investigational Drug Service generated the randomization scheme for periods I and II separately using Randomization.com (http://www.randomization.com) and implemented treatment allocation.

Investigational medicinal products

CG100649 (strength 1.0 mg) and placebo capsules identical in size and appearance were manufactured by Aptuit (Edinburgh, Scotland); matching celecoxib capsules were prepared by the Investigational Drug Service of the Institute for Translational Medicine and Therapeutics to assign treatments in a triple-blinded manner (for subjects, investigators and sponsor, and data managers and biostatisticians).

Drug concentrations: COX/CA inhibition

Blood samples were collected predose and at 2, 3, 4 (period I only), 6, 8 (period II only), 24, and 72 h. In period II, sampling was extended by the addition of 240-h, 576-h, and 1,176-h samples. Plasma and whole-blood samples were processed immediately after sample collection and stored at −80 °C. CG100649 concentrations were determined in plasma (LLOQ 0.5 ng/ml) and whole blood (LLOQ 2 ng/ml) using liquid chromatography–mass spectrometry (LC-MS/MS) (QPS, Newark, DE) with valdecoxib as the internal standard. Celecoxib concentrations in plasma (LLOQ 5 ng/ml) were determined using LC-MS/MS with rofecoxib as internal standard; the concentration in whole blood was assessed only in an exploratory fashion. The concentration levels of serum TxB₂ and whole-blood LPS-PGE₂ were determined using radioimmunoassays (Amersham, Piscataway, NJ). Gas analysis of arterialized earlobe capillary blood was carried out on site (Rapidlab 348; Siemens, Deerfield, IL) within 15–30 seconds after sampling; hyperemia (Supplementary Figure S9 online) was induced by applying a topical vasodilator gel for 15 minutes (Finalgon Strong Ointment; Boehringer Ingelheim, Ingelheim am Rhein, Germany).

Urinary prostaglandin metabolite concentrations

Urine was collected before (−12, 0 h) and at 4–6, 12–24, and 60–72 h after each drug administration in periods I and II, as well as at 228–240, 564–576, and 1,164–1,176 h in period II. These time intervals corresponded to wash-out phases of 2, 5, and 10 times the half-life of CG100649 as observed in earlier pharmacokinetic trials. Metabolites of prostacyclin (PGI-M), prostaglandin E₂ (PGE-M), prostaglandin D₂ (PGD-M), and thromboxane (measured both as 11-dehydro TxB₂ and as 2,3-dinor TxB₂) were analyzed as previously described.¹⁹

Outcomes

Investigational medicinal product–induced changes in PGI-M and serum TxB₂ were prespecified as primary end points; the remaining outcome measures were prespecified as secondary end points. Stationary-position BP measurements were recorded for the purpose of drug safety assessment at the time points when blood samples were drawn for pharmacokinetic analysis.

Data and statistics

Blinded experimental data were entered into electronic case report forms using Oracle Clinical Remote Data Capture (version 4.5.2), and queries arising from the review and from issues relating to data outliers were resolved before unblinding as to the treatment assignments. AEs were coded to MedDRA terminology using CTCAE3, version 3.0 (NCI). Summarized statistics categorized by treatment groups were generated for selected baseline and outcome measures (SAS version 9.1; SAS Institute, Cary, NC) The data are reported as median (ranges) where not otherwise stated; mean ± SD values are provided in Supplementary Table S3 online. Areas under the curve (calculated using the trapezoidal rule) were generated, reflecting the percentage change from baseline, to describe the cumulative drug effects over multiple time intervals (for urine samples) or multiple time points (for blood specimens). Zero percentage change served as the reference (line) to differentiate between “inhibition” (below) and “disinhibition” (above); the latter was subtracted from the former to quantify the drug’s net effect, that is, the actual degree of suppression in the indexes of COX-1 and COX-2 activity attained after active treatment.

Kruskal–Wallis tests were used for k-group comparisons for AUC measures. Post hoc tests comparing individual groups were not performed because of the small sample sizes. No formal adjustment to P-values was made to account for the inflation in type I error due to the large number of statistical tests; therefore, only P-values that would be deemed highly significant (e.g., P < 0.001) should be interpreted as offering evidence of a statistically significant effect.