Degree of Cyclooxygenase-2 Inhibition Modulates Blood Pressure Response

Authors: Katherine N. Theken, Soumita Ghosh, Carsten Skarke, Susanne Fries, Nicholas F. Lahens, Dimitra Sarantopoulou, Gregory R. Grant, Garret A. FitzGerald, Tilo Grosser

Abstract

BACKGROUND:

Large clinical trials compared distinct nonsteroidal anti-inflammatory drugs in terms of their risk of adverse cardiovascular events. However, whether pharmacologically equipotent doses were used, that is, whether a similar degree of COX (cyclooxygenase)-2 inhibition was achieved, was not considered. We compared drug target inhibition and blood pressure (BP) response to celecoxib and naproxen.

METHODS:

Sixteen healthy participants were treated with celecoxib (200 mg/d), naproxen (500 mg/d), or placebo for 7 days in a double-blind, crossover design. The degree of COX inhibition was assessed ex vivo using established whole blood assays and in vivo by quantifying urinary metabolites of thromboxane A2 (COX-1) and prostacyclin (COX-2). Ambulatory BP was measured throughout the final dosing interval.

RESULTS:

Both nonsteroidal anti-inflammatory drugs inhibited COX-2 activity relative to placebo, but naproxen inhibited COX-2 activity to a greater degree (62.9±21.7%) than celecoxib (35.7±25.2%; P<0.05). Similarly, naproxen treatment inhibited prostacyclin formation in vivo (48.0±24.9%) to a greater degree than celecoxib (26.7±24.6%; P<0.05). Naproxen significantly increased BP compared with celecoxib (mean arterial pressure, +2.5 [95% CI, 1.5–3.5] mm Hg; systolic BP, +4.0 [95% CI, 2.9–5.1] mm Hg; and diastolic BP, +1.8 [95% CI, 0.8–2.8] mm Hg; P<0.05 for all). The difference in systolic BP relative to placebo was associated with the degree of COX-2 inhibition (P<0.05).

CONCLUSIONS:

Future studies should consider pharmacokinetic and pharmacodynamic properties, as well as patient-specific factors that may modulate the cardiovascular risk of nonsteroidal anti-inflammatory drug use.

REGISTRATION:

URL: https://www.clinicaltrials.gov; Unique NCT02502006.

Nonsteroidal anti-inflammatory drugs (NSAIDs) are among the most commonly used medications worldwide, and the opioid crisis has placed a new emphasis on their use.^1,2^ Approximately 20% of adults in the United States receive at least 1 NSAID prescription per year,^3^ and 12% of adults in the United States reported using NSAIDs chronically, that is, >3 times weekly for >3 months.^4^ Consumption of NSAIDs in individuals at risk for musculoskeletal injuries is even more common.^5–7^ Given the high prevalence of chronic pain in the United States, >100 million Americans experience chronic pain^8^; NSAIDs are an important nonaddictive option for pain relief. Optimizing NSAID therapy is one strategy to address the current opioid crisis,^2^ and rates of NSAID use have increased in recent years as opioid prescriptions have declined.^9^ Although they lack the addictive potential of opioids, NSAIDs have the potential to cause serious and, in some cases, life-threatening adverse events, including gastrointestinal bleeding, renal dysfunction, hypertension, and thrombotic cardiovascular events.^10^ Clinically meaningful blood pressure (BP) increases on NSAIDs are a common complication.^11^

NSAIDs exert their analgesic and anti-inflammatory effects via inhibition of COX (cyclooxygenase)-1 and COX-2, enzymes that catalyze the first committed step in prostaglandin synthesis. Prostaglandins produce a diverse array of biologic effects via activation of prostanoid receptors and play important roles in a variety of pathological and homeostatic processes.^12^ The risk of thrombotic events associated with the use of NSAIDs, particularly those selective for COX-2, is mediated via suppression of COX-2-derived prostacyclin formation in endothelial and vascular smooth muscle cells.^13,14^ Prostacyclin possesses potent antithrombotic and vasodilatory effects and, thus, acts as a general inhibitor of platelet activation in vivo.^12^ Nonselective NSAIDs also inhibit COX-2 in the vasculature, but the associated risk of thrombosis is mitigated to some extent by inhibition of the formation of thromboxane A2, a COX-1-derived prostaglandin released by activated platelets that promote platelet activation and aggregation.^10,15^ In addition to their effects on vascular prostaglandin production, NSAIDs inhibit renal prostaglandin formation, resulting in sodium retention and BP increases, which may further augment cardiovascular risk.^10,15,16^

One of the largest (N=24 081) outcome studies to date, the PRECISION (Prospective Randomized Evaluation of Celecoxib Integrated Safety Versus Ibuprofen or Naproxen), compared the safety of the COX-2 selective NSAID, celecoxib (100–200 mg twice a day), and 2 traditional NSAIDs, naproxen (375–500 mg twice a day) and ibuprofen (600–800 mg 3 times a day), in patients with osteoarthritis (90%) and rheumatoid arthritis (10%). PRECISION concluded that the cardiovascular safety of moderate doses of celecoxib (average daily dose, 209±37 mg) was noninferior to naproxen (average daily dose, 852±103 mg) or ibuprofen (average daily dose, 2045±246 mg).^17^ However, a secondary on-treatment analysis of PRECISION showed a lower risk of cardiorenal events in the celecoxib group than the ibuprofen and naproxen groups.^18^ Similarly, a prespecified substudy (PRECISION-ABPM [Prospective Randomized Evaluation of Celecoxib Integrated Safety Versus Ibuprofen or Naproxen Ambulatory Blood Pressure Measurement] trial) reported that the percentage of patients with normal baseline BP who developed hypertension was significantly greater in patients treated with naproxen (19%) or ibuprofen (23%) than with celecoxib (10%).^19^ However, the degree of COX-2 inhibition attained has never been assessed in clinical outcome trials, and thus, it is unknown whether the doses used in PRECISION were equipotent.

Here, we compared the pharmacological potency of celecoxib (100 mg twice a day) with naproxen (250 mg twice a day) and how this relates to BP response to NSAIDs in a highly controlled study in apparently healthy volunteers. Both are the lowest recommended daily doses of celecoxib and naproxen for osteoarthritis. We hypothesized that a greater degree of COX-2 inhibition would be associated with a greater increase in BP in response to NSAID treatment.

Methods

Data Availability

Aggregate data that support the findings of this study have been made publicly available in ClinicalTrials.gov (https://www.clinicaltrials.gov; Unique NCT02502006). Subject-level data cannot be publicly shared due to privacy restrictions. Requests to access the data sets should be directed to the corresponding authors.

Participants

Healthy volunteers were recruited by advertisement and word of mouth. Men and women (≥18 years of age) who were in good health based on medical history, physical examination, vital signs, and laboratory tests were eligible for inclusion. Participants were excluded if they were pregnant or nursing a child, smoked or used nicotine-containing products. They were also excluded, if they were obese (body mass index >30 kg/m^2^) had a history of significant cardiovascular, gastrointestinal, renal, hepatic, respiratory, immune, endocrine, hematologic, or neurological disease, or a history of cancer within the last 5 years, a coagulation or bleeding disorder. Volunteers were also excluded, if they were sensitive or allergic to celecoxib, naproxen, aspirin, or other NSAIDs, or had used NSAIDs (including aspirin and acetaminophen), dietary or herbal supplements containing salicylates, vitamin E, fish oil, or any other herbal supplements, within 14 days of study drug administration.

Study Procedures

The study protocol was approved by the University of Pennsylvania institutional review board (820715; https://www.clinicaltrials.gov; Unique NCT02502006), and all participants provided informed consent. The study was a randomized, double-blind, 3-way crossover study comparing the degree of COX inhibition and the BP response at steady state following treatment with celecoxib (100 mg twice daily), naproxen (250 mg twice daily), or placebo (twice daily) for 7 days. Before beginning treatment, all participants attended a screening visit to obtain a complete medical history and confirm eligibility. They were asked to abstain from analgesics, including products containing NSAIDs (including aspirin and acetaminophen), high-dose vitamins, and nutritional supplements until study completion.

Randomization was performed using a computer-generated sequence to assign participants to 1 of 6 treatment orders to ensure that each treatment occurred with equal frequency and any carryover effects were balanced by sequence. Allocation was provided to the University of Pennsylvania Investigational Drug Service by an investigator not involved in participant enrollment.

On the first day of each treatment phase, baseline blood and urine samples were collected, and participants were given a blister pack with blinded study medication to be taken by mouth twice daily on an outpatient basis. Study medication was blinded by overencapsulation by the University of Pennsylvania Investigational Drug Service. On the morning of day 7, participants returned to the clinical research unit for a 12-hour visit for pharmacokinetic-pharmacodynamic sampling. Ambulatory BP (ABP) was monitored every 15 minutes, while the participants were seated in the clinical research unit, using an automated ABP monitor (Spacelabs 90207) placed on the nondominant upper arm. Blood and urine samples were collected (T=0), and the final dose of study medication was administered. Additional samples were collected 0.5 (blood only), 1, 2, 4, 8, and 12 hours after study medication administration. Participants were discharged after the 12-hour sample collection. These study visits were repeated for the next 2 treatment phases, with a washout period of at least 2 weeks between each treatment phase.

Study data were collected and managed using Research Electronic Data Capture hosted at the University of Pennsylvania Perelman School of Medicine.^20,21^

Quantification of COX Activity and Plasma Drug Concentrations

Whole blood assays were used to assess the degree of COX inhibition ex vivo under conditions of maximal stimulation. COX-1 activity ex vivo was evaluated by quantifying serum thromboxane B2 levels, as previously described.^22^ Briefly, whole blood was collected into vacuum tubes containing clot activator and incubated in a water bath at 37 °C for 1 hour. Serum was separated by centrifugation and stored at −80 °C until analysis by liquid chromatography-tandem mass spectrometry (LC-MS/MS).

COX-2 activity ex vivo was evaluated by quantifying plasma prostaglandin E2 levels following lipopolysaccharide stimulation in whole blood, as previously described.^23^ Briefly, heparinized whole blood was treated with aspirin (1 mmol/L) and incubated at room temperature for 15 minutes. Lipopolysaccharide (Escherichia coli, serotype O111:B4, 10-µg/mL whole blood) was added, and the sample was incubated in a water bath at 37 °C for 24 hours. Plasma was separated by centrifugation and stored at −80 °C until analysis by LC-MS/MS.

COX activity in vivo was determined by quantification of urinary prostanoid metabolites by LC-MS/MS as previously described.^14^ This allows for assessment of the degree of COX inhibition under physiological conditions. Systemic production of prostacyclin, prostaglandin E2, prostaglandin D2, and thromboxane A2 was determined by quantifying their major urinary 2,3-dinor 6-keto-PGF1α, 7-hydroxy-5,11-diketotetranorprostane-1,16-dioic acid, 11,15-dioxo-9α-hydroxy-2,3,4,5-tetranorprostan-1,20-dioic acid, and 2,3-dinor thromboxane B2, respectively. Results were normalized to urinary creatinine measured by LC-MS/MS. Urinary 2,3-dinor thromboxane B2 was used as an index of COX-1 activity in vivo, and urinary 2,3-dinor 6-keto-PGF1α was used as an index of COX-2 activity in vivo, as previously described.^16^

Plasma concentrations of celecoxib and naproxen were quantified by LC-MS/MS as previously described.^24^

Statistical Analysis

Measurements of COX activity ex vivo and urinary prostaglandin metabolite levels were normalized to the mean value during the placebo phase for each subject to calculate the percent of COX inhibition relative to placebo. The area under the curve from T=0 to T=12 hours was calculated as a measure of the degree of COX inhibition throughout the dosing interval. The estimated glomerular filtration rate was calculated using the Chronic Kidney Disease Epidemiology Collaboration creatinine equation as described by Inker et al.^25^ The degree of COX inhibition was compared by paired t test. The effect of treatment on COX activity and ABP over time was analyzed by linear mixed effect modeling using the lme4 R package,^26^ including time, treatment, and time×treatment as main effects and participant as a random effect. Time was treated as a continuous variable, and all BP measurements, without averaging, were included in the analysis. The effect of treatment was evaluated by comparing the differences in least-squares (LS) means for each treatment averaged over time. The relationship between change in systolic BP (SBP) and COX inhibition at each time point relative to placebo was evaluated by linear mixed effects modeling including COX inhibition as a main effect and participant as a random effect. P<0.05 was considered statistically significant. At an α=0.05 level, a sample size of 16 participants provided >90% power to detect a difference of 25% in the degree of COX inhibition and 3 mm Hg in SBP, assuming a coefficient of variation of 100%. Statistical analyses were performed in R (version 4.3.1). Figures were created in BioRender and GraphPad Prism.

Results

The flow of participants is shown in Figure 1. The analysis cohort included 16 healthy adults (9 men and 7 women) with a mean age of 34.7±13.4 years. Baseline demographic and clinical characteristics are shown in the Table. Biochemical measures were evaluated in all participants. One participant was excluded from ABP analysis due to an ABP monitor malfunction and incomplete data.

Naproxen 250 mg twice daily inhibited COX-1 activity and COX-2 activity ex vivo by 95.3±4.4% and 62.9±21.7%, respectively, while celecoxib 100 mg twice daily had minimal effects on COX-1 activity ex vivo and inhibited COX-2 activity ex vivo by 35.7±25.2% over the 12-hour dosing interval (Figure 2). Similar results were observed for COX-1 and COX-2 activity in vivo, assessed by urinary 2,3-dinor thromboxane B2 and 2,3-dinor 6-keto-PGF1α concentrations, respectively. COX-1 activity in vivo was inhibited by 68.2±18.7% with naproxen treatment and 8.9±35.7% with celecoxib treatment. COX-2 activity in vivo was inhibited by 48.0±24.9% with naproxen treatment and 26.7±24.6% with celecoxib treatment. With all functional parameters, the degree of COX inhibition was significantly greater with naproxen treatment than with celecoxib treatment (P<0.05). The maximum plasma concentration of naproxen was 228.0±45.1 µmol/L, and the time to maximum plasma concentration was 1.8±1.2 hours after administration. For celecoxib, the maximum plasma concentration was 1.28±0.55 µmol/L, and the time to maximum plasma concentration was 2.3±1.2 hours.

During the placebo phase, the average mean arterial pressure (MAP), SBP, and diastolic BP (DBP) were 89.7±8.7, 124.1±11.2, and 73.3±8.7 mm Hg, respectively. NSAID treatment affected MAP, SBP, and DBP over the 12-hour dosing interval (Figure 3). Naproxen treatment significantly increased MAP (difference in LS means, 3.1 [95% CI, 2.1–4.1] mm Hg; P<0.05), SBP (difference in LS means, 2.9 [95% CI, 1.8–4.0] mm Hg; P<0.05), and DBP (difference in LS means, 3.2 [95% CI, 2.2–4.2] mm Hg; P<0.05) relative to placebo. In contrast, celecoxib treatment did not affect MAP (difference in LS means, 0.6 [95% CI, −0.3 to 1.6] mm Hg; P=0.28) but significantly decreased SBP (difference in LS means, −1.1 [95% CI, −2.2 to 0.04] mm Hg; P<0.05) and increased DBP (difference in LS means, 1.4 [95% CI, 0.4–2.4] mm Hg; P<0.05) relative to placebo. Compared with celecoxib, naproxen significantly increased MAP (difference in LS means, 2.5 [95% CI, 1.5–3.5] mm Hg; P<0.05), SBP (difference in LS means, 4.0 [95% CI, 2.9–5.1] mm Hg; P<0.05), and DBP (difference in LS means, 1.8 [95% CI, 0.8–2.8] mm Hg; P<0.05). For SBP, a significant time×treatment interaction was observed for naproxen (β = 0.40 [95% CI, 0.16–0.65] mm Hg/h; P<0.05) but not for celecoxib treatment (β = 0.04 [95% CI, −0.21 to 0.28] mm Hg/h; P=0.78). No significant time×treatment effects were observed for MAP or DBP.

Mixed effects modeling was performed to determine whether the degree of COX-1 or COX-2 inhibition ex vivo predicted the difference in SBP with NSAID treatment relative to placebo (Figure 4). COX-2 inhibition ex vivo was a predictor of difference in SBP (β = −5.64 [95% CI, −9.36 to −1.92]; P=0.003), but COX-1 inhibition ex vivo was not significantly associated with difference in SBP (β = −0.850 [95% CI, −2.99 to 1.30]; P=0.43).

Discussion

Given the high prevalence of chronic pain in the United States,^8^ NSAIDs are an important nonaddictive option for pain relief.^2^ Currently, it is recommended that NSAIDs be avoided or used only for a limited duration at the lowest possible dose in patients considered at high cardiovascular risk.^27^ As COX-2 inhibition mechanistically underlies the cardiovascular risk, a key question remains whether differences in the cardiovascular safety profile exist between traditional NSAIDs (ie, naproxen), which inhibit both COX-1 and COX-2, and selective inhibitors of COX-2 (ie, celecoxib).^28^ The largest comparator trial to address this question, PRECISION, indicated that celecoxib is noninferior to naproxen and ibuprofen with regard to cardiovascular risk.^17^ Importantly, the dose of celecoxib was limited to 200 mg/d (100 mg twice per day) in osteoarthritis patients enrolled in PRECISION, which made up the vast majority of the study population, while a dose up to 400 mg/d (200 mg twice per day) was allowed in patients with rheumatoid arthritis. Limiting the dose in patients with osteoarthritis had been a regulatory response to the cardiovascular hazard detected in previous randomized controlled trials. Overall, this dosing regimen resulted in an average daily dose of 209±37 mg across all patients in the celecoxib arm.^17^ In patients randomized to naproxen treatment in PRECISION, the starting dose was 750 mg/d (375 mg twice per day) with the option to increase to 1000 mg/d (500 mg twice per day) in patients with rheumatoid arthritis, resulting in an average dose of 852±103 mg in the naproxen arm.^17^ Although the pharmacological potency of celecoxib measured with isolated COX-2 enzyme or cellular preparations in vitro is higher than that of naproxen,^29^ here, we demonstrate that 200 mg/d of celecoxib (approximately the average daily dose used in the PRECISION trial) inhibited COX-2 activity in vivo to a lesser degree than 500 mg/d of naproxen (a dose that was 33% lower than the average daily dose used in PRECISION), and this impacted the BP response to NSAID treatment. Our results underscore the importance of considering dose and pharmacoequivalence in vivo in comparisons of safety among drugs of the same class.

Only a small number of clinical trials have prospectively assessed the effects of COX inhibition on BP control. Based on these studies, acute increases of 3 to 5 mm Hg in SBP can be expected within 1 to 2 weeks of treatment.^16,30,31^ In our cohort, naproxen treatment for 1 week increased SBP relative to placebo (2.9 [95% CI, 1.8–4.0] mm Hg) to a greater extent than celecoxib treatment (−1.1 [95% CI, −2.2 to −0.04] mm Hg) over the final 12-hour dosing interval. This difference is similar to what was observed in PRECISION-ABPM, where the change in SBP from baseline after 4 months of treatment was 1.91±9.796 mm Hg among naproxen-treated patients and −0.18±9.400 mm Hg among celecoxib-treated patients.^19^ We also observed a significant time×treatment effect on SBP for naproxen, but not celecoxib, treatment, which may reflect the longer half-life and duration of COX inhibition with naproxen in addition to differences in potency in vivo. Notably, we observed that the difference in SBP was associated with the degree of COX-2 inhibition on NSAID treatment, consistent with the inhibition of COX-2-mediated prostacyclin formation as the primary mechanism underlying the increased cardiovascular risk associated with NSAID use.

Although naproxen increased SBP to a greater extent than celecoxib in our study cohort, the BP response to NSAID treatment was heterogeneous among individual patients. The effects of COX inhibition on BP control are complex, which may contribute to the variable occurrence of hypertension on NSAIDs.^10,15,16,32^ In the renal cortex, the production of vasodilatory prostaglandin E2 and prostacyclin maintains the patency of adjacent afferent arterioles,^16,33^ and COX-2 expression in renal medullary interstitial cells plays an important role in the adaptive regulation of BP in response to high-salt diet and dehydration.^34–36^ COX inhibition in these regions of the kidney contributes to the decline in glomerular filtration rate and elevations in BP observed in patients who take NSAIDs. However, dynamic expression of COX-2 in the macula densa system is a component of the tubuloglomerular feedback mechanism, which promotes renin release.^37–39^ Inhibition of COX-2 in these cells would counteract the hypertensive effects of renin-angiotensin system activation. Thus, the effect of NSAID treatment on BP reflects the complex interplay among these regulatory systems. Interestingly, we observed that an individual participant’s response was similar to both drugs, suggesting that patient-specific factors may contribute to interindividual heterogeneity in the response to NSAIDs. Future studies are necessary to elucidate the factors that contribute to an individual patient’s risk of hypertension and other cardiovascular adverse effects with NSAID treatment.

There are limitations to our study. The small sample size limits our ability to comprehensively investigate the factors that contribute to the BP response to NSAIDs. The study cohort included only healthy adults, most of whom were relatively young. Although this limits potential confounding due to effects of age and comorbidities, it precludes interrogation of the influence of these factors on the BP response to NSAID treatment. Prior studies have demonstrated that older individuals (age>65 years) and patients with hypertension have greater elevations in BP with NSAID treatment.^40^ NSAIDs also can decrease the efficacy of antihypertensive drugs and given the role of renal COX-2 in the activation of the renin-angiotensin-aldosterone system,^37–39,41,42^ perhaps particularly for renin-angiotensin-aldosterone system inhibitors. Thus, we would hypothesize that the degree of BP elevation would be greater in patients with these cardiovascular risk factors. Although quantification of urinary prostanoid metabolites provides a measurement of the degree of COX inhibition systemically in vivo, we are unable to determine the tissue source of these metabolites. This study observed acute changes in BP following 7 days of treatment and linked them to COX-2 inhibition. Different mechanisms may contribute to more steady BP increases over time, observed in long-duration clinical trials.^43^ Finally, we compared only 1 dose level of naproxen and celecoxib, which limits our ability to extrapolate our results to higher doses or other NSAIDs. Despite these limitations, our results provide mechanistic insight into the outcome of PRECISION.

In conclusion, our results demonstrate that naproxen 500 mg/d inhibits COX-2 activity to a greater degree than celecoxib 200 mg/d, and the degree of COX-2 inhibition is associated with the BP response to NSAID treatment. While PRECISION concluded noninferiority of celecoxib compared with naproxen with regard to cardiovascular risk, this is based on a comparison of doses that are not equipotent. Future studies should consider the pharmacokinetic and pharmacodynamic properties of the drugs, as well as patient-specific risk factors when seeking to compare the cardiovascular risk associated with the use of specific NSAIDs.

Perspectives

• Naproxen 250 mg twice a day inhibited COX-2 activity to a greater degree than celecoxib 100 mg twice a day.
• The degree of COX-2 inhibition was associated with the increase in systolic BP with NSAID treatment relative to placebo.
• Dose and its pharmacological potency achieved in vivo should be considered when evaluating the relative cardiovascular safety of COX-2-selective versus nonselective NSAIDs.

ARTICLE INFORMATION

Acknowledgments

G.A. FitzGerald is the McNeil Professor of Translational Medicine and Therapeutics and holds a Merit Award from the American Heart Association. C. Skarke is the Robert L. McNeil Jr. Fellow in Translational Medicine and Therapeutics.

Sources of Funding

Research reported in this publication was supported by a Translational Medicine and Therapeutics Postdoctoral Fellowship from the PhRMA Foundation and funding from the National Heart, Lung, and Blood Institute (grant HL117798) and the National Center for Advancing Translational Sciences of the National Institutes of Health (grant UL1TR001878). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Disclosures

None.