Celebrex And Blood Pressure Medication

Hypertension and arthritis represent common comorbid conditions leading to the frequent coadministration of nonsteroidal anti-inflammatory agents (NSAIDs) and antihypertensive medications. Conventional NSAIDs used commonly in the management of the signs and symptoms of arthritis nonspecifically inhibit both isoforms of the cyclooxgenase (COX) system. As a result, nonselective NSAID therapy may reduce the efficacy of antihypertensive drug therapy, especially drugs that block the renin-angiotensin system.^1–4

Nonselective COX inhibition results in inhibition of prostaglandin synthesis and is associated with antinatriuretic and vasoconstrictor effects.^4,5 These effects have consequences on blood pressure (BP) control and are of particular relevance in patients with preexisting hypertension.^1,2 Meta-analyses of the effects of many of these agents (including naproxen, indomethacin, and ibuprofen) have reported average increases in mean arterial pressure of 5.5 mm Hg^1,2 However, the relative contribution of COX-1 or COX-2 to this effect is unknown.

Angiotensin-converting enzyme (ACE) inhibitors are widely prescribed antihypertensive medicines. Like other antihypertensive therapies (with the exception of the calcium channel blockers and possibly the alpha-blockers), an important component of the mechanism of action of ACE inhibitors is the increased synthesis of vasodilatory prostaglandins.⁴

Celecoxib is a COX-2-specific inhibitor indicated for the management of osteoarthritis or rheumatoid arthritis.^5,6 It specifically inhibits the COX-2 isoenzyme while sparing COX-1,^5,6 resulting in an improved gastrointestinal safety and tolerability profile compared with nonselective NSAIDs.^6,7 The COX-2 inhibitor rofecoxib has also been shown to reduce gastrointestinal toxicity compared with conventional NSAIDs⁸ but has recently been shown to significantly increase 24-hour systolic BP by 4.5 mm Hg over placebo in hypertensive patients on the ACE inhibitor benazepril.^9,10 Because of these reported interactions of the NSAIDs and rofecoxib with BP control in ACE inhibitor-treated patients, we studied the effects of celecoxib to determine whether a similar interaction occurred.

Methods

Patients

Men and women between the ages of 18 and 75 years were included if they had a history of uncomplicated hypertension and were treated with no more than 2 antihypertensive agents. Because aspirin has been shown to attenuate the hemodynamic effects of the ACE inhibitors,^11,12 aspirin was not allowed during the study. Acetaminophen (up to 4 grams/daily) was permitted.

Study Design

The study was a double-blind, placebo-controlled, parallel-group multicenter trial (Figure 1). The primary objective was to evaluate the effects of high doses (200 mg dosed twice daily) of celecoxib on 24-hour BP in hypertensive patients whose BP was controlled with the ACE inhibitor lisinopril. Current antihypertensive medication(s) were discontinued or tapered to discontinuation. When seated diastolic BP in the doctor's office or clinic was stable (defined as 2 consecutive measurements, taken 2 weeks apart and varying by no more than 8 mm Hg) at 95 to 109 mm Hg, patients were eligible for the open label lisinopril titration period.

**Figure 1.** Design of the clinical trial. ABPM indicates ambulatory blood pressure monitoring.

Lisinopril was titrated weekly (10, 20, to 40 mg once daily in the morning) to stabilize the hypertension (defined as a seated diastolic BP [DBP] <90 mm Hg if the untreated DBP was 95 to 104 mm Hg or <95 mm Hg if the untreated DBP was 105 to 109 mm Hg). During this ACE-inhibitor treatment period, all patients received single-blind placebo to match celecoxib, 200 mg twice daily. When patients met the criteria for diastolic BP stabilization at 2 separate weekly visits on a stable dose of lisinopril in the clinic or office, 24-hour ambulatory BP monitoring was performed, providing baseline measures. If BP could not be stabilized adequately with lisinopril 40 mg QD, or if ambulatory BP monitoring was not adequate after 2 attempts during the single-blind period, then patients were withdrawn from the study.

Patients with stable BP on lisinopril and complete 24-hour ambulatory BP recordings at baseline were randomized to treatment with either celecoxib 200 mg twice daily (twice the recommended dose for osteoarthritis) or placebo twice daily. During the double-blind period, the dose of lisinopril that had controlled the BP during the baseline period was maintained. Following 4 weeks of double-blind therapy, 24-hour ambulatory monitoring was repeated. During the double-blind period, the lisinopril dose could not be altered, and no other antihypertensive therapy was permitted.

Statistical Analyses

The comparability of patients in the 2 treatment groups was determined from the demographic data and baseline hemodynamic values. Continuous variables were analyzed using the nonparametric Kruskal-Wallis method. Discrete variables were compared using the χ² test method for homogeneity of proportions for categorical data. The primary variable for assessing the attenuation of ACE inhibitor control of hypertension was the change in 24-hour diastolic BP from baseline after 4 weeks of celecoxib or placebo. Secondary analyses included the changes from baseline in the 24-hour systolic BP and several other ambulatory monitoring parameters, including daytime mean, nighttime mean, and the changes in ambulatory heart rate. Furthermore, the proportion of patients whose 24-hour systolic and diastolic BP increased by <5, 5 to 9, 10 to 14, 15 to 19, and >20 mm Hg on celecoxib and placebo compared with the baseline values were calculated.

Treatment groups were compared with respect to change from baseline to the week 4 endpoint using an analysis of covariance (ANCOVA) that included treatment, center, and baseline measures as the covariates in the model. Treatment comparisons were based on the least square means obtained via a SAS Type III analysis (SAS 6.09 VMS operating system, SAS Institute). A power analysis was based on a single effect variable, ie, the change from baseline in the 24-hour mean diastolic BP as determined at the final visit of the double-blind period. It was determined that a sample size of 80 patients per treatment group would give 80% power to detect a 3 mm Hg difference between celecoxib and placebo in the 24-hour diastolic BP assuming a standard deviation of 6.5 mm Hg and a 2-tailed α of 0.05.

Results

Baseline Parameters

As shown in Figure 2, there were 242 patients who entered the lisinopril titration period; 64 patients (26%) were not randomized, the majority of the discontinuations before randomization occurred because of an inability to meet the BP stability criteria. Of the 178 patients who were randomized in the double-blind treatment period, there were no significant baseline differences in demographic characteristics, or in baseline ambulatory BPs or heart rates between the celecoxib (n=91) or placebo (n=87) groups (Table 1). The majority of the randomized patients completed the entire 4-week double-blind treatment period (97.7% in the placebo group and 94.5% in the celecoxib group). The average daily doses of lisinopril were similar in the celecoxib treatment group (23 mg) and the placebo group (21 mg).

**Figure 2.** Disposition of the study patients.

**Table 1.** Baseline Demographic Characteristics of All Randomized Patients
Characteristic	Celecoxib (n=91)	Placebo (n=87)
Values shown are mean±standard error of the mean; P>0.05 for all comparison between treatment groups. BP indicates blood pressure.
Gender (M/F)	51/40	49/38
Age, mean yr (range)	54.5 (30–70)	52.9 (29–69)
Weight, mean kg (range)	87.6 (55–153)	87.5 (54–140)
Ethnicity, n (%)
White	51 (56%)	55 (63%)
Black	17 (19%)	10 (12%)
Hispanic/Latin American	19 (21%)	19 (22%)
Asian	4 (4%)	2 (2%)
Other	0 (0%)	1 (1.1%)
Serum creatinine, μmol/L	70.1	73.7
Ambulatory BP measurements (mm Hg)
24-hour systolic BP	134.7 ±1.5	130.8 ±1.2
24-hour diastolic BP	83.8 ±0.9	82.3 ±0.8
Daytime systolic BP	137.4 ±1.4	133.7 ±1.2
Daytime diastolic BP	86.7 ±0.9	85.0 ±0.8
Nighttime systolic BP	125.8 ±1.8	121.4 ±1.4
Nighttime diastolic BP	74.6 ±1.0	73.5 ±1.1

Effects of Treatment on Ambulatory BP and Heart Rate

The effects of celecoxib and placebo on ambulatory BP and heart rate parameters are shown in Table 2. There were small increases in 24-hour ambulatory systolic and diastolic BP (2.6/1.5 mm Hg) in the celecoxib group compared with the placebo group (1.0/0.3 mm Hg). However, these were not statistically significant. These numerical differences between celecoxib and placebo were more notable during the daytime, awake hours than during the nighttime sleep period (Table 2). There were no changes in the 24-hour mean heart rates on placebo or celecoxib.

**Table 2.** Changes From Baseline in Ambulatory BP and Heart Rate (HR) Measures Following Celecoxib Versus Placebo in Lisinopril-Treated Patients
Ambulatory BP or HR Parameter	Treatment Differences
Ambulatory BP or HR Parameter	Celecoxib	Placebo	Mean	95% CI	P value
Data are expressed as the mean±standard error of the mean.
	(n=91)	(n=87)
24-hour systolic BP (mm Hg)	2.62 ±0.86	1.04 ±0.98	1.58	−0.88, 4.30	0.195
24-hour diastolic BP (mm Hg)	1.48 ±0.62	0.26 ±0.64	1.22	−0.38, 3.08	0.126
24-hour heart rate (bpm)	0.16 ±0.66	−0.19 ±0.63	0.35	−1.27, 2.36	0.555
Daytime systolic BP (mm Hg)	2.97 ±0.92	1.01 ±1.03	1.96	−0.68, 4.83	0.139
Daytime diastolic BP (mm Hg)	1.67 ±0.68	0.41 ±0.68	1.26	−0.44, 3.30	0.133
Daytime heart rate (bpm)	−0.05 ±0.64	−0.17 ±0.73	−0.12	−1.83, 2.02	0.923
Nighttime systolic BP (mm Hg)	1.37 ±1.03	1.10 ±1.20	0.27	−2.72, 3.52	0.801
Nighttime diastolic BP (mm Hg)	0.62 ±0.78	−0.28 ±0.90	0.90	−1.41, 3.28	0.433
Nighttime heart rate (bpm)	−0.72 ±0.77	1.17 ±0.71	1.89	−0.08, 4.08	0.059

Visual evaluation of the hourly ambulatory BP profiles showed a 2-to-4-mm Hg, transient increase in ambulatory systolic BP for approximately 4 hours post-dosing of celecoxib, both in the morning (9 am to 1 pm) and in the evening (7 pm to 10 pm) (Figure 3). These hourly changes from baseline coincide with the approximate dosing times of study drug in the morning and 12 hours later in the evening. The changes from baseline in diastolic BP were smaller on celecoxib.

**Figure 3.** Mean changes from baseline in 24-hour systolic (top) and diastolic (bottom) blood pressure following 4 weeks of treatment with placebo versus celecoxib in patients whose BP was controlled on lisinopril monotherapy. Celecoxib was dosed at 200 mg twice daily at 6–8 am and 6–8 pm There were no significant differences between celecoxib and placebo on ambulatory blood pressure. MN indicates midnight.

An analysis of the proportion of individuals whose 24-hour BP increased by clinically relevant threshold values is shown in Figure 4. Most of the patients whose BP rose in either the celecoxib or placebo group were in the range of 1 to 10 mm Hg. There were no significant differences in the proportion of patients whose BP was elevated at any particular threshold value on celecoxib versus placebo.

**Figure 4.** The proportion of patients whose 24-hour systolic (top) and diastolic (bottom) blood pressure increased by <5, 5–9, 10–14, 15–19, and >20 mm Hg above baseline following celecoxib 200 mg twice daily or placebo. There were no significant differences in the proportions at any of the given thresholds.

Safety Analyses

There were no deaths or serious adverse events in the celecoxib treatment group. There were 2 serious adverse events in the placebo group (suicide attempt and acute myocardial infarction that occurred post-study). There were no differences in nonserious adverse event rates between the celecoxib and placebo treatment groups. The mean change from baseline in body weight was −0.06±0.47 kg in the placebo group and 0.16±0.37 kg in the celecoxib group (P=NS). There were also no differences in the proportion of patients who developed a prespecified 5% increase or decrease in body weight. No patient in the placebo group developed significant peripheral edema, whereas one subject (1%) developed edema in the celecoxib group. There were no differences in any of the safety laboratory parameters with the exception of the blood urea nitrogen which increased 0.67±0.13 mmol/L on celecoxib versus decreasing by 0.24±0.13 mmol/L on placebo (P<0.01). However, no patient in either group developed a blood urea nitrogen above the upper limits of normal.

Discussion

Principal Findings of the Study

Our study demonstrates that celecoxib, a novel COX-2-specific inhibitor, had no clinically significant interaction with the 24-hour antihypertensive effect of lisinopril. In this study, change in the 24-hour ambulatory BP in an ACE inhibitor-treated population was chosen as the primary endpoint, because ambulatory BP has recently been shown to have an important and independent association with target organ disease and prognosis in patients with hypertension.^13–16 In addition, the 24-hour BP curves are superior to the clinic measurements as an effective method for assessing any potential pharmacodynamic interactions occurring during the entire dosing interval of the drugs under evaluation, ie, the ACE inhibitor and the COX-2 inhibitor (Figure 3). The other clinically relevant finding from our study was that celecoxib at maximal doses of 400 mg per day was not associated with increases in body weight, edema rates, or clinically important changes in renal function or potassium homeostasis in hypertensive patients treated with an ACE inhibitor.

Comparative Findings With Conventional NSAID and ACE Inhibitors

Celecoxib appears to have effects that are different from those reported with conventional NSAIDs on BP control in ACE inhibitor-treated patients. The changes observed in 24-hour BP (1.58/1.22 mm Hg, Table 2) are substantially less than those that have been reported for nonselective NSAIDs in ACE inhibitor-treated patients.^1–4 For example, in a study by Morgan et al,⁴ indomethacin was given at a dose of 50 mg twice daily to 60 patients whose BP had been controlled by the ACE inhibitor enalapril and the dihydropyridine calcium antagonist amlodipine. The difference caused by indomethacin between the 2 groups was a 10/5 mm Hg increase in 24-hour ambulatory BP in patients taking enalapril. Based on biochemical assessments in that study, it was suggested that indomethacin induced sodium retention, and in the patients taking enalapril, plasma renin activity fell but was not translated into a physiological effect because of the blockade of the angiotensin-converting enzyme. In contrast, in patients treated with amlodipine, the fall in plasma renin activity ameliorated the effect of sodium retention on the BP. Similarly, Klassen et al¹⁷ also showed that, whereas the NSAID naproxen induced a small weight gain in hypertensive patients treated with the calcium antagonist nicardipine, no changes in BP in the clinic were induced. In a large multicenter trial involving 216 patients, Conlin and coworkers³ reported that indomethacin (75 mg daily) attenuated the 24-hour ambulatory diastolic BP for both losartan and captopril by approximately 40%. Thus, the effects of an NSAID on the angiotensin II receptor blocker was similar to the effects observed with an ACE inhibitor.

One study by Thakur and colleagues¹⁸ suggested that combining a thiazide diuretic with an ACE inhibitor prevented the attenuation of BP reduction induced by an NSAID. They performed a study in 17 African-American women with arthritis and hypertension, who were receiving the ACE inhibitor fosinopril and hydrochlorothiazide, and co-administered 3 different NSAIDs (ibuprofen, sulindac, and nabumetone) in a crossover design. No changes in the clinic BP were observed on any of the 3 NSAIDs, despite 5 patients (1 on sulindac and 4 on ibuprofen) going into acute renal failure on the combination of drugs. The investigators suggested that the interference of the NSAID with the ACE inhibitor effects was lessened because the thiazide diuretic induced salt depletion and stimulated renin, thereby enhancing the ACE inhibitor.

Comparisons Between COX-2 Inhibitors Added to Antihypertensive Therapy

Our study represents the only trial to date that has evaluated the ambulatory BP effects of celecoxib in patients treated exclusively with ACE inhibitors. Whelton et al¹⁹ performed a randomized study in which 812 much older patients with osteoarthritis and controlled hypertension received either celecoxib (200 mg daily) or rofecoxib (25 mg daily). Approximately 40% of the patients randomized to the celecoxib group had been treated chronically with ACE inhibitors, whereas 30% of the rofecoxib-treated patients were on ACE inhibitors. The proportion of patients whose systolic BP increased by a clinically significant amount (>20 mm Hg) was significantly greater on rofecoxib (17%) compared with patients on celecoxib (11%). These findings are supported in part by a 36-patient crossover design study that evaluated the COX-2 inhibitor rofecoxib, 25 mg daily, indomethacin sustained-release, 75 mg daily, and placebo in hypertensive patients controlled with the ACE inhibitor benazepril.^9,10 In that study, rofecoxib was associated with a significant 4.5 mm Hg sustained increase in 24-hour systolic BP compared with placebo and an approximately 2.5 mm Hg increase in 24-hour systolic pressure compared with indomethacin.

Conclusions

Our study demonstrates that celecoxib, even at maximal doses, is not associated with clinically or statistically significant attenuation of the 24-hour BP control in patients treated exclusively with an ACE inhibitor. Based on several studies in the literature, celecoxib appears to be associated with a lower incidence of hypertension than nonselective NSAIDs using both clinic and ambulatory BP measurements.^1–4,8 Recent data directly comparing celecoxib to rofecoxib in older hypertensive patients also confirm a significant difference between these agents in terms of their effects on systolic BP in the clinic and edema.¹⁹ Differences in the level of BP control or stability in patients with hypertension clearly have the potential to drive overall cardiovascular morbidity, especially in terms of congestive heart failure and stroke.²⁰ The mechanisms associated with the apparent difference in the COX-2-specific inhibitors, as well as the differences between celecoxib and the conventional NSAIDs, remains to be elucidated. Nevertheless, variations in the cardiovascular effects of celecoxib versus conventional NSAIDs, which may be molecule-specific rather than mechanism-specific, are demonstrable.^8,21 Further, direct comparative trials with the COX-2 inhibitors and conventional NSAIDs in patients with hypertension are needed.

This work was supported by the Patrick and Catherine Weldon Donaghue Medical Research Foundation (Hartford, Conn), Pharmacia Clinical Research and Development (Skokie, Ill), and the University of Connecticut Clinical Trials Unit (Farmington).

Footnotes

Correspondence to William B. White, MD, Professor and Chief, Section of Hypertension and Clinical Pharmacology, University of Connecticut School of Medicine, 263 Farmington Avenue, Farmington, CT 06030-3940. E-mail [email protected]

References

1 Pope JE, Anderson JJ, Felson DT. A meta-analysis of the effects of nonsteroidal anti-inflammatory drugs on blood pressure. Arch Intern Med . 1993; 153: 477–484.CrossrefMedlineGoogle Scholar
2 Johnson AG, Nguyen TV, Day RO. Do nonsteroidal anti-inflammatory drugs affect blood pressure? A meta-analysis. Ann Intern Med . 1994; 121: 289–300.CrossrefMedlineGoogle Scholar
3 Conlin PR, Moore TJ, Swartz SL, Barr E, Gazdick L, Fletcher C, DeLucca P, Demopoulos L. Effect of indomethacin on blood pressure lowering by captopril and losartan in hypertensive patients. Hypertension . 2000; 36: 461–465.CrossrefMedlineGoogle Scholar
4 Morgan TO, Anderson A, Bertram D. Effect of indomethacin on blood pressure in elderly people with essential hypertension well controlled on amlodipine and enalapril. Am J Hypertens . 2000; 13: 1161–1167.CrossrefMedlineGoogle Scholar
5 Penning TD, Talley JJ, Bertenshaw SR, Carter JS, Collins PW, Docter S, Graneto MJ, Lee LF, Malecha JW, Miyashiro JM, Rogers RS, Rogier DJ, Yu SS, Anderson GD, Burton EG, Cogburn JN, Gregory SA, Koboldt CM, Perkins WE, Siebert K, Veenhuizen AW, Zhang YY, Isakson PC. Synthesis and biological evaluation of the 1,5-diarylpyrazole class of cyclooxygenase-2 inhibitors: identification of SC-58635 (celecoxib). J Med Chem . 1997; 40: 1347–1365.CrossrefMedlineGoogle Scholar
6 Simons LS, Lanza FL, Lipsky PE, Hubbard RC, Talwalker S, Schwartz BP, Isakson PC, Geis GS. Preliminary study of the safety and efficacy of SC-58635, a novel cyclooxygenase 2 inhibitor. Arthritis Rheum . 1998; 41: 1591–1602.CrossrefMedlineGoogle Scholar
7 Silverstein FE, Faich G, Goldstein JL, Simon LS, Pincus T, Whelton A, Makuch R, Eisen G, Agrawal NM, Stenson WF, Burr AM, Zhao WW, Kent JD, Lefkowith JB, Verburg KM, Geis GS. Gastrointestinal toxicity with celecoxib vs nonsteroidal anti-inflammatory drugs for osteoarthritis and rheumatoid arthritis: the CLASS study. A randomized controlled trial. Celecoxib Long-Term Arthritis Safety Study. JAMA . 2000; 284: 1247–1255.CrossrefMedlineGoogle Scholar
8 Bombardier C, Laine L, Reicin A, Shapiro D, Burgos-Vargas R, Davis B, Day R, Ferraz MB, Hawkey CJ, Hochberg MC, Kvien TK, Schnitzer TJ. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. VIGOR Study Group N Engl J Med . 2000; 343: 1520–1528.CrossrefMedlineGoogle Scholar
9 FDA. Cardiovascular-renal safety review: Rofecoxib NDA 21–042. Rockville, Md: Food and Drug Administration, 1999. Google Scholar
10 Merck & Co, Inc. VIOXX (rofecoxib) package insert. West Point, Pa: Merck & Co, Inc, 2001.Pamphlet.Google Scholar
11 Guazzi MD, Campodonico J, Celeste F, Guazzi M, Santambrogio G, Rossi M, Trabattoni D, Alimento M. Antihypertensive efficacy of angiotensin converting enzyme inhibition and aspirin counteraction. Clin Pharmacol Ther . 1998; 63: 79–86.CrossrefMedlineGoogle Scholar
12 Nawarskas JJ, Spinler SA. Does aspirin interfere with the therapeutic efficacy of angiotensin converting enzyme inhibitors in hypertension or congestive heart failure? Pharmacotherapy . 1998; 18: 1041–1052.MedlineGoogle Scholar
13 Verdecchia P, Porcellati C, Schillaci G, Borgioni C, Ciucci A, Battistelli M, Guerrieri M, Gatteschi C, Zampi I, Santucci A, Santucci C, Reboldi G. Ambulatory blood pressure: an independent predictor of prognosis in essential hypertension. Hypertension . 1994; 24: 793–801.LinkGoogle Scholar
14 Ohkubo T, Imai Y, Tsuji I, Nagai K, Watanabe N, Minami N, Itoh O, Bando T, Sakuma M, Fukao A, Satoh H, Hisamichi S, Abe K. Prediction of mortality by ambulatory blood pressure monitoring versus screening blood pressure measurements: a pilot study in Ohasama. J Hypertens . 1997; 15: 357–364.CrossrefMedlineGoogle Scholar
15 Redon J, Campos C, Narciso ML, Rodicio JL, Pascual JM, Ruilope LM. Prognostic value of ambulatory blood pressure monitoring in refractory hypertension: a prospective study. Hypertension . 1998; 31: 712–718.CrossrefMedlineGoogle Scholar
16 White WB, Schulman P, McCabe EJ, Dey HM. Average daily blood pressure, not office pressure, determines cardiac function in patients with hypertension. JAMA . 1989; 261: 873–877.CrossrefMedlineGoogle Scholar
17 Klassen DK, Jane LH, Young DY, Peterson CA. Assessment of blood pressure during naproxen therapy in hypertensive patients treated with nicardipine. Am J Hypertens . 1995; 8: 146–153.CrossrefMedlineGoogle Scholar
18 Thakur V, Cook E, Wallin JD. Antihypertensive effect of the combination of fosinopril and hydrochlorothiazide is resistant to interference by nonsteroidal antiinflammatory drugs. Am J Hypertens . 1999; 12: 925–928.CrossrefMedlineGoogle Scholar
19 Whelton A, Fort JG, Puma JA, Normandin D, Bello AE, Verburg KM. Cyclooxygenase-2-specific inhibitors and cardiorenal function: a randomized, controlled trial of celecoxib and rofecoxib in older hypertensive osteoarthritis patients. Am J Ther . 2001; 8: 85–96.CrossrefMedlineGoogle Scholar
20 Major cardiovascular events in hypertensive patients randomized to doxazosin vs chlorthalidone: the antihypertensive and lipid-lowering treatment to prevent heart attack trial (ALLHAT). ALLHAT Collaborative Research Group. JAMA . 2000; 283: 1967–1975.CrossrefMedlineGoogle Scholar
21 Fries JF, Williams CA, Bloch DA. The relative toxicity of nonsteroidal antiinflammatory drugs. Arthritis Rheum . 1991; 34: 1353–1360.MedlineGoogle Scholar