The Effects of Cyclooxygenase-2 Inhibitors and Nonsteroidal Anti-inflammatory Therapy on 24-Hour Blood Pressure in Patients With Hypertension, Osteoarthritis, and Type 2 Diabetes Mellitus

Background  Nonsteroidal anti-inflammatory drugs (NSAIDs) and cyclooxygenase-2 (COX-2) inhibitors may attenuate the efficacy of antihypertensive agents in high-risk patients. Therefore, we conducted a double-blind, randomized trial to evaluate the effects of celecoxib, rofecoxib, and naproxen on 24-hour blood pressure (BP) in patients with type 2 diabetes, hypertension, and osteoarthritis.

Methods  Patients were randomly assigned to treatment with 200 mg of celecoxib once daily (n = 136), 25 mg of rofecoxib once daily (n = 138), or 500 mg of naproxen twice daily (n = 130) for 12 weeks. Twenty-four–hour ambulatory BP monitoring and validated arthritis efficacy assessments were conducted at randomization and at weeks 6 and 12 of treatment. The primary end point was the mean change from baseline in average 24-hour systolic BP at week 6.

Results:  Reductions in osteoarthritis symptoms, including pain, mobility, and stiffness, were similar in all treatment groups. The mean ± SE 24-hour systolic BP following 6 weeks of therapy was increased significantly by rofecoxib (from 130.3 ± 1.2 to 134.5 ± 1.4 mm Hg; P<.001) but not by celecoxib (132.0 ± 1.3 to 131.9 ± 1.3 mm Hg; P = .54) or naproxen (133.7 ± 1.5 to 133.0 ± 1.4 mm Hg; P = .74). The BP difference between rofecoxib and celecoxib was 3.78 mm Hg (95% confidence interval, 1.18-6.38; P = .005); between rofecoxib and naproxen, 3.85 mm Hg (95% confidence interval, 1.15-6.55; P = .005). The proportion of patients with controlled hypertension at baseline who developed ambulatory hypertension by week 6 (24-hour systolic BP>135 mm Hg) was significantly greater with rofecoxib (30%) than with celecoxib (16%) (P = .05) but not significantly greater than with naproxen (19%).

Conclusions  At equally effective doses for osteoarthritis management, treatment with rofecoxib but not celecoxib or naproxen induced a significant increase in 24-hour systolic BP. However, destabilization of hypertension control occurred to some extent in all 3 treatment groups; this phenomenon was seen more often in patients treated with rofecoxib than with the other therapies.

Elevated systolic blood pressure (SBP) places patients with type 2 diabetes at risk for premature renal and cardiovascular disease.^1-3 With aging, patients with diabetes and hypertension often have coexisting osteoarthritis, which is most frequently managed with nonsteroidal anti-inflammatory drugs (NSAIDs) and cyclooxygenase-2 (COX-2) inhibitors.⁴

Nonspecific NSAIDs have been reported to increase blood pressure (BP) in treated populations with hypertension, especially among users of angiotensin-converting enzyme (ACE) inhibitors and β-blockers.^5-8 Furthermore, rofecoxib has been reported to induce clinically significant increases in ambulatory SBP in ACE inhibitor–treated patients with hypertension,^9,10 whereas celecoxib has shown effects on ambulatory BP similar to placebo.^8,11 Angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers (ARBs) are integral components of antihypertensive therapy in the diabetic population.¹² The Celecoxib Rofecoxib Efficacy and Safety in Comorbidities Evaluation Trial (CRESCENT) was designed to investigate the effects of these agents compared with naproxen on 24-hour BP in patients with type 2 diabetes who were receiving stable regimens of antihypertensive therapy that included at least an ACE inhibitor or an ARB. To avoid concerns with dose disparities among these osteoarthritis treatments, we performed standardized and validated measurements of efficacy for osteoarthritis.

Patients were randomized if they had osteoarthritis of the knee¹³ or hip¹⁴ (meeting American College of Rheumatology criteria) requiring daily NSAID therapy; type 2 diabetes being treated with oral hypoglycemic agents and/or insulin; glycosylated hemoglobin values lower than 9%; hypertension being treated with a stable antihypertensive regimen (at least 6 weeks) of at least an ACE inhibitor (or an ARB if ACE inhibitor was not tolerated); and a seated SBP not greater than 150 mm Hg. Patients were required to have completed at least 7 days without their current NSAID therapy prior to baseline osteoarthritis assessments. Ongoing treatment with estrogen, progesterone, and/or testosterone (if used for at least 2 months before randomization) and 81 to 325 mg of aspirin daily (if used for at least 30 days before randomization) was permitted if the treatment could be maintained at a stable dose throughout the study.

Important criteria for exclusion were type 1 diabetes, rheumatoid arthritis, night or rotating shift work that included night-shift schedules, serum creatinine concentration greater than 1.5 mg/dL (132.6 μmol/L), and serum potassium concentration greater than 5.4 mmol/L. Treatment with drugs that could affect BP, weight, or edema (excluding each patient’s stable regimen of antihypertensive agents) was not permitted (eg, orlistat, sibutramine, thiazolidinediones, and oral corticosteroids). The institutional review board or ethics committee at each center approved the protocol, and all patients provided written informed consent before enrollment.

This was a multicenter, international, randomized, double-blind trial. Patients were randomly assigned to receive 200 mg of celecoxib once daily (Celebrex; Pfizer Inc, New York, NY), 25 mg of rofecoxib once daily (Vioxx; Merck & Co, West Point, Pa), or 500 mg of naproxen twice daily (Naprosyn; Roche Pharmaceuticals, Basel, Switzerland) for 12 weeks. There was a single, computer-generated, randomization code, and each site was given block sizes of 3. Patients were randomized in a 1:1:1 fashion, and there was no stratification. Prior to the start of double-blind therapy, each patient received treatment for 7 to 10 days with 1000 mg of acetaminophen (Tylenol; McNeil Consumer & Specialty Pharmaceuticals, Fort Washington, Pa) 3 times daily for control of osteoarthritis symptoms. Patient evaluations were conducted at baseline and 1, 2, 6, and 12 weeks after randomization. At each visit, seated BP measurements (in duplicate) and determination of adverse events were performed. Ambulatory BP measurements and osteoarthritis efficacy assessments were conducted at baseline and at weeks 6 and 12 after randomization.

Seated BP measurements were taken in the nondominant arm with a mercury sphygmomanometer. Patients were to be discontinued from the study in cases of investigator-determined uncontrolled hypertension that led to any change in antihypertensive therapy.

Ambulatory BP measurements were obtained with the SpaceLabs 90207 monitor (Redmond, Wash). Quality criteria used for an acceptable ambulatory BP recording included a minimum of 80% valid readings obtained within 24 hours after monitor hookup and a minimum of 2 valid readings per hour. If these criteria were not met, the patient was asked to repeat the study within 3 days. If the repeat study failed to meet the quality control criteria, the data were considered nonevaluable. Blood pressure was measured every 20 minutes during the entire 24-hour ambulatory monitoring period. Monitoring hookup was initiated between 7 and 11 AM, and patients were given the study medication within 5 minutes of monitor hookup. Study coordinators recorded times of sleep, awakening, medication administration, and monitor hookup.

Assessments of osteoarthritis treatment efficacy were performed using standard validated measurements: (1) Patient’s Assessment of Arthritis Pain–Visual Analog Scale, graded on a scale of 0 mm (no pain) to 100 mm (very severe pain)¹⁵; (2) Patient’s Global Assessment of Arthritis, graded on a scale of 1 (very good) to 5 (very poor)¹⁵; and (3) The Western Ontario and McMaster Universities (WOMAC) Osteoarthritis Index, a 24-component patient assessment of osteoarthritis pain, joint stiffness, and physical function.^15,16 The outcome of any patient who withdrew from the study because of symptoms from osteoarthritis (lack of efficacy) was reported as a treatment failure, not an adverse event.

The primary end point was the mean change from baseline in average 24-hour SBP at week 6. Six weeks was chosen to assess the primary end point to avoid the impact of later selective patient dropout from 1 of the study groups, whether owing to adverse effects, loss of BP control, or lack of efficacy. Secondary end points for BP included mean change from baseline at week 6 for the average 24-hour diastolic BP, mean change from baseline at week 12 for average 24-hour BP and pulse pressure, changes from baseline in daytime (6 AM to 10 PM) and nighttime (10 PM to 6 AM) BPs, and number of patients discontinued from the study owing to a change in medication for loss of BP control. Additionally, the 3 osteoarthritis efficacy assessments were evaluated.

The comparability of treatment groups with respect to baseline characteristics and the primary and secondary end points were assessed using the Fisher exact test for categorical variables. For continuous variables, a 2-way analysis of covariance was used with investigational site and baseline score of the variable being analyzed (for the primary and secondary end points) as possible covariates and treatment as a factor. Pairwise comparisons of the least square means were used to evaluate treatment differences. Baseline BP values were not imputed to either week 6 or week 12, and there was no imputation at week 6. Any missing postbaseline BP value was imputed, and a last-observation-carried-forward analysis was performed. There were multiple secondary end points, and no adjustments for multiplicity were preplanned for these end points. No interim analyses were performed. The type I (α) error for all hypothesis testing was set at .05, and all tests were 2-tailed tests. All confidence intervals (CIs) were 95%.

The sample size was calculated assuming a standard deviation of approximately 7.5 mm Hg in 24-hour SBP differences for the treatment groups and a dropout rate or lack of evaluability of 20%, using a 2-tailed test (.05 significance level). Thus, a sample size of 375 patients (125 per treatment group) was required to provide power of 80% to detect a 3–mm Hg or greater difference in 24-hour SBP among the treatment groups.

A total of 65 centers from 7 countries participated in this trial. Between May 2001 and April 2002, 404 patients were randomized: 136 to celecoxib, 138 to rofecoxib, and 130 to naproxen. Eight randomized patients (6 to rofecoxib and 2 to naproxen) never took any study medication and were not included in either the BP or efficacy analyses.

The 3 treatment groups had similar baseline characteristics, including BP, serum creatinine, plasma glucose, and glycosylated hemoglobin levels (Table 1). More than 98% of the patients were treated with ACE inhibitors or, in cases of ACE-inhibitor intolerance, ARBs, and more than 60% were receiving multiple antihypertensive therapies. Six (1.5%) of 396 randomized patients were not taking an ACE inhibitor or ARB. The study results were similar when these 6 patients were either included or excluded from the analyses.

A total of 81.6% (323) of 396 patients completed 6 weeks of the trial, which included the primary outcome ambulatory BP assessment. The remaining 73 patients (22 celecoxib, 24 rofecoxib, and 27 naproxen) did not have a valid week 6 ambulatory BP assessment because it was either not performed or not evaluable. A total of 29 of these 73 patients were withdrawn from the study owing to an adverse event and did not complete a week 6 ambulatory BP assessment: 7 (5%) in the celecoxib group, 11 (8%) in the rofecoxib group, and 11 (9%) in the naproxen group (P = .48 for comparison of the proportions).

All 3 treatments produced significant improvements from baseline in the signs and symptoms of osteoarthritis at both 6 and 12 weeks, and pairwise comparisons showed no significant differences between the treatments except for a significant difference in favor of celecoxib vs naproxen in the stiffness subdomain of the WOMAC (P = .02) at week 12 (Table 2). The number of patients discontinued for lack of arthritis efficacy was low for all treatments (3 for celecoxib, 1 for rofecoxib, and 3 for naproxen).

As detailed in Table 3, rofecoxib increased the 24-hour systolic BP at week 6, but celecoxib and naproxen did not. Additionally, the difference in the mean change from baseline to week 6 in 24-hour ambulatory SBP between rofecoxib and celecoxib was 3.78 mm Hg (95% CI, 1.18-6.38; P = .005); between rofecoxib and naproxen, 3.85 mm Hg (95% CI, 1.15-6.55; P = .005) (Table 3). There was no significant difference between celecoxib and naproxen. The distribution of ambulatory SBP changes from baseline at week 6 is shown in Figure 1. The rofecoxib treatment group had larger proportions of patients whose 24-hour SBP increased compared with the other 2 treatment groups. In addition, the percentage of patients with baseline ambulatory SBPs lower than 135 mm Hg (normotension) who developed hypertension was significantly greater for rofecoxib than for celecoxib (Figure 2). The hourly ambulatory SBP curves over 24 hours at baseline and at week 6 for the 3 treatment groups are shown in Figure 3. A consistent increase from baseline in SBP was observed only in the rofecoxib treatment group.

Three patients discontinued the study owing to a change in medication for loss of BP control before week 6 (1 celecoxib, 1 rofecoxib, and 1 naproxen), and 5 discontinued for this reason after week 6 (4 rofecoxib and 1 naproxen). The ambulatory and clinic BP results are summarized in Table 3. The clinic SBP measurements were similar to the ambulatory BP results (Table 3). However, at week 12, naproxen significantly increased the mean ± SE clinic SBP (+3.6 ± 1.8 mm Hg; P = .02). The difference in the mean change from baseline to week 6 in clinic SBP between rofecoxib and celecoxib was 4.20 mm Hg (95% confidence interval, 0.48-7.93; P = .03); between rofecoxib and naproxen, 3.70 mm Hg (95% confidence interval, −0.30 to 7.71; P = .07). There was no significant difference between celecoxib and naproxen.

This is the first clinical trial comparing the COX-2–specific inhibitors and a nonspecific NSAID on BP control in which well-defined measures of osteoarthritis efficacy were also assessed. Our study used 24-hour BP recordings, which improve the reproducibility of BP values in clinical trials,¹⁷ avoid observer bias, and allow for the BP assessments over the entire 24-hour period.

Treatment with 25 mg of rofecoxib once daily significantly increased average 24-hour SBP and pulse pressure in patients with type 2 diabetes who were receiving stable regimens of antihypertensive therapy consisting of ACE inhibitors or ARBs. In contrast, treatment with 200mg of celecoxib once daily or 500 mg of naproxen twice daily had no significant effect on the mean 24-hour BP following both 6 and 12 weeks of double-blind therapy. These BP findings occurred with doses typically prescribed for osteoarthritis¹⁸ and where the assessments of osteoarthritis efficacy showed equivalence for all 3 treatments.

Two previous trials evaluated the effects of COX-2–specific inhibitors on ambulatory BP in ACE inhibitor–treated patients with hypertension.^9,11 In 1 study, 200 mg of celecoxib twice daily vs placebo was given to 178 patients stabilized with lisinopril therapy; there was no significant difference in 24-hour SBP between celecoxib and placebo (+2.6 mm Hg vs +1.0 mm Hg; P = .34).¹¹ In the other study, 25 mg of rofecoxib daily was compared with placebo and sustained-release indomethacin in a crossover design trial involving 36 patients in whom ambulatory BP was measured before and after 4 weeks of benazepril therapy.⁹ Rofecoxib induced a significant 4.5– mm Hg increase in 24-hour SBP over placebo and a 2.5– mm Hg increase over indomethacin.⁹ These findings are consistent with those in our trial.

Clinic SBP results in our study are also in agreement with results from 2 previous comparative trials of rofecoxib and celecoxib in older patients treated for hypertension and osteoarthritis.^10,19 In the present study, there was a significant treatment difference in clinic SBP for rofecoxib compared with celecoxib (+4.2 mm Hg) at week 6, while in the previous studies, 25 mg of rofecoxib daily was associated with an approximate +3.4–mm Hg difference in clinic SBP compared with 200 mg of celecoxib over a 6-week period.^10,19

Our results suggest that there may be a hemodynamic difference not only between celecoxib and rofecoxib but also between individual COX-2–specific inhibitors and nonspecific NSAIDs. While the mechanism(s) for these differences is unknown at this time, differences in disposition, metabolites, effects on intrarenal prostaglandin production, and possible molecular distinctions may be contributing factors. Beyond these possible mechanisms, celecoxib may reduce endothelial dysfunction, which could reduce vasoconstriction in susceptible patients.^20,21

The comparative efficacy of the 2 COX-2–specific inhibitors in the treatment of osteoarthritis has been controversial. In 2 studies,^22,23 25 mg of rofecoxib and 200 mg of celecoxib demonstrated similar efficacy, while in 1 study,²⁴ 25 mg of rofecoxib was determined to be statistically superior in efficacy to 200 mg of celecoxib. Our trial represents the third such study showing similar osteoarthritis efficacy between these 2 doses of COX-2–specific inhibitors. We compared the efficacy of rofecoxib, celecoxib, and naproxen using well-defined instruments of osteoarthritis efficacy at 6 and 12 weeks of treatment. All 3 drugs, under conditions of a randomized, blinded protocol, produced comparable improvements in these standard measures at both time points, which enhances the validity of the study with regard to dose parity for assessing the impact of these agents on ambulatory BP in patients with osteoarthritis, type 2 diabetes, and hypertension. Owing to drug cost and the lack of a hypertensive effect, naproxen could be considered first-line treatment. However, as recommended by the American College of Rheumatology²⁵ and others,^26,27 clinicians should consider other risks, including the risk factors for upper gastrointestinal events, in determining appropriate therapies.

Small increases in both clinic and ambulatory SBP in patients with hypertension and type 2 diabetes are associated with substantial increases in the risk of cardiovascular morbidity.^1,2,28-30 Additionally, ambulatory BP appears to be a better predictor of cardiovascular and cerebrovascular morbidity and mortality than clinic BP, including in the patient treated for hypertension.^30-33 Specifically, it was recently demonstrated that ambulatory SBPs of 135 mm Hg or more (hypertension) predict a higher incidence of cardiovascular events than SBPs lower than 135 mm Hg (normotension). While all 3 treatment groups had a minority of patients whose 24-hour SBPs increased to hypertensive values, the findings from the present study show that significantly more rofecoxib patients than celecoxib patients attained SBP levels in the hypertensive range (Figure 2). This finding along with an increase in 24-hour SBP of more than 4 mm Hg for rofecoxib but not for the other 2 osteoarthritis therapies (Table 3) must be considered clinically relevant. While further study is warranted,^34,35 these results suggest the need for careful monitoring and control of BP when NSAIDs or COX-2 inhibitors are chosen for osteoarthritis management in patients with hypertension and type 2 diabetes and further suggest need for careful evaluation of currently available as well as future COX-2–specific inhibitors and nonspecific NSAIDs in this population.

Box Section Ref ID

Correspondence: William B. White, MD, Division of Hypertension and Clinical Pharmacology, Pat and Jim Calhoun Cardiology Center, University of Connecticut School of Medicine, Farmington, CT 06030-3940 (wwhite@nso1.uchc.edu).

Accepted for Publication: September 16, 2004.

Financial Disclosure: Drs Sowers, White, Simon, Pitt, and Whelton have been consultants to Pharmacia Corporation and Pfizer. Dr Pitt has been a consultant to Merck. Drs Sowers, White, Simon, Pitt, and Whelton have received honoraria from Pfizer for occasional lectures during the past 3 years. Drs van Ingen and Fort are former employees of Pharmacia and Pfizer.

Funding/Support: This study was supported by a grant from Pharmacia Corporation/Pfizer Inc.

Author Contributions: Drs White and Sowers contributed equally in the senior authorship.

Acknowledgment: We are indebted to Reinhard Schuller, MS, for his expert help in the statistical analyses and to Lorraine R. Baer, PharmD, for her invaluable editorial contributions and assistance in the preparation of the manuscript.

Box Section Ref ID