Endothelial Reactivity and Cardiac RiskFactors in Older Patients With Peripheral

Arterial DiseaseAlberto Rafael Yataco, MD, Mary Concepta Corretti, MD,

Andrew William Gardner, PhD, Christopher Joseph Womack, PhD, andLeslie Ira Katzel, MD, PhD

Peripheral arterial disease (PAD) is a major cause ofmorbidity and mortality. Endothelium-dependent vaso-reactivity, which is advocated as a measure of vascularhealth, is impaired in persons with cardiac risk factorsand coronary artery disease. Few studies have exam-ined the degree of endothelial dysfunction in patientswith PAD. Using high-resolution external vascular ultra-sound, we measured brachial artery diameter and flowat rest, and in response to reactive hyperemia (flow-mediated dilation) in 50 older patients (age 69 6 1 year)with PAD (ankle-to-brachial artery index of 0.67 60.03), and 50 age-matched non-PAD patients. Coronaryartery disease was more prevalent in PAD than in non-PAD patients (40% vs 4%, p <0.001). Systolic bloodpressure (153 6 4 vs 141 1 3 mm Hg, p <0.01), fastingglucose (129 6 6 vs 109 6 5 mg/dl, p <0.001), andpack-years smoked (54 6 7 vs 25 6 3, p <0.01) werehigher in the PAD than in non-PAD patients. There wereno differences in baseline brachial artery diameter,blood velocity, or flow between the 2 groups. However,

the 1-minute postocclusion percent change in diameter(6.5 6 0.7% vs 9.8 6 0.7%, p <0.001) and the changein diameter (0.22 6 0.02 vs 0.33 6 0.02 mm, p<0.001) were lower in PAD than in non-PAD patients,suggesting impaired endothelium-dependent dilation.The postocclusion hyperemic velocity and blood flowwere also lower in PAD than in non-PAD patients. Inmultiple regression analyses the low-density lipopro-tein–to–high-density lipoprotein cholesterol ratio, ele-vated fasting glucose, and high systolic blood pressurewere independent predictors of percent change in bra-chial artery diameter (r2 5 0.37, p <0.001). Thus, olderpatients with PAD had impaired endothelial dependentvasodilation compared with controls that was associ-ated with the presence of cardiac risk factors. The effectof cardiac risk factor intervention on endothelial functionin patients with PAD remains to be determined.Q1999 by Excerpta Medica, Inc.

(Am J Cardiol 1999;83:754–758)

Peripheral arterial disease (PAD), a manifestationof atherosclerosis, increases exponentially as a

function of age.1,2 Patients with PAD often have sig-nificant comorbid medical conditions, such as coro-nary artery disease, systemic hypertension, and diabe-tes mellitus.2 Chronic cigarette smoking, which isassociated with progression of atherosclerosis,3 is usu-ally prevalent in this population. Endothelium-depen-dent vasoreactivity, which is advocated as a measureof vascular health, is impaired in persons with cardiacrisk factors and coronary artery disease. Thus, poorflow-mediated vasoreactivity is expected in PAD pa-tients. Few studies have examined the degree of en-dothelial dysfunction in patients with PAD.4,5 Thepurpose of this study was to determine the degree andthe major determinants of endothelial dysfunction in

PAD patients compared with age-matched non-PADpatients.

METHODSSubjects: This study was approved by the Institu-

tional Review Board of the University of Maryland,and all patients provided informed consent beforetheir participation in the study. Outpatient volunteerswith a history of PAD and intermittent claudication(Fontaine stage II) were recruited from vascular clin-ics at the Baltimore Veterans Affairs Medical Center,University of Maryland Hospital, health-care provid-ers in the Baltimore metropolitan area, and via mediaadvertisements for participation in exercise interven-tion trials. These subjects were interviewed by tele-phone using the London School of Hygiene Cardio-vascular Rose Questionnaires6 to ascertain the pres-ence of intermittent claudication. Fifty patients (44men and 6 women) aged.55 years (mean 696 1,mean6 SEM) with history of PAD and intermittentclaudication were recruited. All had resting ankle-to-brachial artery index,0.97 (range 0.28 to 0.96).Twenty of the patients (40%) with PAD had a historyof coronary artery disease, 34 (68%) were on medi-cations for hypertension, 23 (46%) were on lipid-lowering medications, 11 (22%) were on oral hypo-glycemic agents, and 4 (8%) were on insulin fordiabetes mellitus. Ninety percent of the PAD patients

From the Divisions of Gerontology and Cardiology, Department ofMedicine, University of Maryland School of Medicine and BaltimoreVeteran Affairs Medical Center, Baltimore, Maryland. This work wassupported by the Department of Veteran Affairs Baltimore GeriatricResearch, Education and Clinical Center, Baltimore, Maryland; aVeteran Affairs Merit grant, Baltimore, Maryland; the University ofMaryland Claude D. Pepper Older Americans Independence Center(NIH/NIA P60-AG-12583), Baltimore, Maryland; and NIH/NIA-5-KO1-AG00657, Baltimore, Maryland. Manuscript received July 1,1998; revised manuscript received and accepted October 5, 1998.

Address for reprints: Leslie Ira Katzel, MD, PhD, Baltimore VeteransAffairs Medical Center, Geriatric Research, Education and ClinicalCenter (18), 10 North Greene Street, Baltimore, Maryland 21201.

754 ©1999 by Excerpta Medica, Inc. 0002-9149/99/$–see front matterAll rights reserved. PII S0002-9149(98)00984-9

were either former (n5 28) or current smokers (n517).

Fifty patients (42 men and 8 women) aged.55years without a history of claudication (non-PAD)were recruited from the community via media adver-tisement for participation in weight loss and exerciseinterventions. Two of the non-PAD patients (4%) hada history of coronary artery disease, 16 (32%) were onmedications for hypertension, 11 (22%) were on lipid-lowering medications, and 2 (4%) were on oral hypo-glycemic agents for diabetes mellitus. Fifty-six per-cent of the non-PAD patients were either former (n526) or current smokers (n5 2). All participants hadcomprehensive initial medical evaluations, includinghistory and physical examinations performed by aphysician or nurse practitioner, resting 12-lead elec-trocardiogram, and fasting blood for serum electro-lytes, glucose, liver function tests, and complete bloodcount.

Disease criteria: The presence of coronary arterydisease was defined as a history of previous myocar-dial infarction, history of angioplasty, coronary arterybypass graft surgery, or use of antianginal medica-tions. Diabetes was defined as a fasting blood glucose.7.00 mmol/L (126 mg/dl), or use of hypoglycemicagents or insulin. Hypertension was defined as systolicblood pressure.140 mm Hg, or diastolic.90 mmHg on$2 occasions, or use of antihypertensive med-ications. Dyslipidemia was defined as a low-densitylipoprotein cholesterol concentration.4.14 mmol/L(160 mg/dl), triglyceride concentration.4.52 mmol/L(400 mg/dl), high-density lipoprotein cholesterol con-centration,0.91 mmol/L (35 mg/dl), or use of lipid-lowering medications.7

Endothelial reactivity: Endothelial function in theform of flow-mediated brachial artery vasodilationwas measured using high-frequency ultrasound, aspreviously described.8 Subjects were imaged by asingle dedicated sonographer after an overnight fast ina temperature-controlled room (22°C) in the recum-bent, resting state. Vasoactive medications, such ascalcium antagonists, angiotensin-converting enzymeinhibitors, andb blockers were held for 24 to 48 hoursbefore the study. Caffeinated beverages and smokingwere not allowed the day of the study. Blood pressureand heart rate were recorded automatically (DatascopeAccutor 3 SAT, Paramus, New Jersey) from the sub-ject’s right arm every 5 minutes, along with 1-leadcontinuous electrocardiographic monitoring. Flow-mediated brachial artery vasoactivity was assessedfrom the subject’s left arm once it was comfortablyimmobilized in the extended position. Brachial arterydiameter and Doppler blood flow velocity were ob-tained from a longitudinal orientation approximately 5cm above the antecubital fossa using a linear arraybroad-band frequency (7.5 to 11 MHz) transducerattached to an ATL Apogee 800 ultrasound system(Seattle, Washington). Baseline brachial artery diam-eter and blood flow velocity were recorded twice and1 minute after release of 5 minutes upper arm bloodpressure cuff (12.5 cm) occlusion (180 to 200 mmHg). Blood flow velocity was immediately (within 30

to 45 seconds) recorded by Doppler after cuff defla-tion (reactive hyperemia). Brachial artery diameterand flow velocity were recorded at 1 minute after cuffrelease. Ultrasound images of end-diastolic frameswere obtained for off-line analysis on the same ultra-sound system, and arterial diameter and blood flowwere subsequently measured. Percent change in bra-chial artery diameter was determined as the percentdiameter change of the postocclusion arterial diametermeasurement relative to the mean of the correspond-ing baseline measurements. Arterial blood flow wasdetermined as arterial cross-sectional area (pDiame-ter2/4) times mean Doppler flow velocity integratedover 5 consecutive cardiac cycles.

Ankle/brachial systolic pressure index: Ankle sys-tolic blood pressure was measured in the posteriortibialis and dorsalis pedis arteries of both legs using anondirectional Doppler flow detector (Parks MedicalElectronics, Inc., model 810-A, Aloha, Oregon) and apencil probe (9.3 MHz).9 The artery yielding thehigher pressures in the more diseased leg was re-corded as the resting ankle systolic pressure. Thebrachial systolic and diastolic blood pressures wereobtained by Dinamap Vital Signs Monitor (model1846-SX; Critikon, Inc., Tampa, Florida) in botharms. The arm yielding the higher systolic pressurewas used to calculate the ankle/brachial systolic bloodpressure index. These procedures yield high test-retestreliability for ankle and brachial systolic pressures andankle/brachial index.9

Metabolic testing: Blood samples were drawn intochilled ethylenediaminetetraacetic acid (1-mg/ml)tubes after a 12- to 14-hour overnight fast. Plasmatriglyceride and cholesterol levels were measured en-zymatically. High-density lipoprotein cholesterol con-centrations were measured in the supernatant afterprecipitation of apolipoprotein B–containing lipopro-teins with dextran sulfate. The low-density lipoproteincholesterol concentration was calculated using theFriedewald equation. Fasting plasma glucose levelswere measured enzymatically using the glucose oxi-dase method. Two-hour oral glucose tolerance testswere performed in 23 PAD and 36 non-PAD subjectswho had fasting glucose concentrations,140 mg/dl.All reported blood pressure, lipoprotein, and glucosevalues are for patients on their regular medications.

Statistics: Data from this study were entered intoStatview for analysis (Abacus Concepts, Berkeley,California). The distribution of all variables was testedfor normality. The smoking pack-years was not nor-mally distributed. Unpaired 2-tailedt tests were usedto compare measures between PAD and non-PAD forvariables that were normally distributed. The Wilcoxrank test was used to compare the number of pack-years smoked between the 2 groups. In a subgroupanalysis, we used analysis of variance with Fisher’sleast significant difference post-hoc testing to compareparameters among PAD subjects classified by smok-ing status into never smokers, former smokers, andcurrent smokers. Pearson product moment correlationand Spearman rank correlation coefficients were cal-culated between cardiac risk factors and percent

MISCELLANEOUS/ENDOTHELIAL REACTIVITY IN PERIPHERAL ARTERIAL DISEASE 755

change in diameter at 1 minute. Stepwise multipleregression analyses were performed to determine theindependent contribution of cardiac risk factors to thepercent change in diameter at 1 minute. Statisticalsignificance was set at p,0.05. Data are presented asmean6 SEM.

RESULTSPhysical characteristics and the measured blood

pressures, glucose tolerance, and lipoprotein lipids forthe subjects with PAD and non-PAD are summarizedin Table I. The 2 groups were of comparable age,weight, and body mass index. Systolic blood pressurewas significantly higher in subjects with PAD thannon-PAD (p,0.01). However, diastolic blood pres-sures were the same in the 2 groups. Fasting glucoseconcentrations were significantly higher in PAD thanin non-PAD patients (p,0.001), as were the 2-hourpostprandial concentrations obtained in the subset ofsubjects without evidence of fasting diabetes. Theprevalence of current and former smokers, and meanpack-years smoked was also higher in PAD than non-PAD subjects (p,0.01). Lipoprotein lipid profileswere comparable in the 2 groups.

At baseline, there were no significant differences inbrachial artery diameter, blood velocity, or flow be-tween the 2 groups (Table II). However, the flow-mediated 1-minute postocclusion percent change inbrachial artery diameter, and the absolute change indiameter at 1 minute, were lower in PAD than non-PAD subjects, suggesting an impaired endothelium-dependent dilation. Based on percent change in diam-eter value of 6%, equivalent to 1 SD below the normfor young healthy controls in our laboratory, 60% ofsubjects with PAD had impaired endothelium-depen-dent dilation compared with 32% of older non-PADsubjects (p,0.01). The postocclusion hyperemic ve-locity and blood flow were also lower in PAD than

non-PAD subjects. A subgroup anal-ysis demonstrated that within sub-jects with PAD, the posthyperemicpercent change in diameter was com-parable in current and former smok-ers (6.86 1.3% vs 5.86 0.8%, p5NS), but lower than that observed inthe PAD never smokers (10.761.8%, p,0.05). There was no sig-nificant difference in the posthypere-mic percent change in diameter inthe 20 PAD subjects with coronaryartery disease compared with the 30PAD subjects without a history ofcoronary artery disease (7.1%61.1% vs 6.0%6 0.8%, p5 NS).

Regression analyses were per-formed to determine the relation be-tween coronary artery disease riskfactors and percent change in bra-chial artery diameter in data pooledfrom the 2 groups of subjects. In biva-riate analysis, elevated low-density li-poprotein cholesterol (r5 20.23, p

,0.05), increased ratio of low-density–to–high-densitylipoprotein cholesterol (r5 20.37, p ,0.001), highbody mass index (r5 20.37, p,0.001), elevated fast-ing glucose (r 5 20.36, p ,0.001), pack-yearssmoked (r5 20.33, p,0.001), and elevated systolic(r 5 20.47, p,0.001) and diastolic blood pressures(r 5 20.33, p,0.001) were associated with reducedpercent change in brachial artery diameter. By con-trast, increased high-density lipoprotein cholesterolconcentrations were associated with an increased per-cent change in diameter (r5 0.26, p,0.01). Withinthe subjects with PAD, there was no significant rela-tion between ankle/brachial index and percent changediameter (r5 20.01, p 5 NS). In the combinedgroups, in stepwise multiple regression analyses, sys-tolic blood pressure, fasting glucose, and the low-density–to–high-density lipoprotein cholesterol ratiowere independent predictors of percent change in di-ameter (Table III). The presence of PAD and smokingpack-years were not independent determinants of thepercent change in diameter.

DISCUSSIONThis study demonstrates that older persons with

symptomatic PAD have impaired flow-mediated bra-chial artery endothelial vasodilatory responses com-pared with age-matched persons without PAD. Theendothelium-dependent vasodilation of the brachialartery was abnormal in 60% of older PAD subjects,suggestive of widespread endothelial dysfunction.Similar to the present study, Harris et al4 reporteddecreased vascular reactivity in 16 patients with PAD.Bode-Boger et al5 examined the effects of intravenousinfusion of L-arginine on nitric oxide–dependent va-sodilation in 10 patients with PAD with critical limbischemia (2 with Fontaine stage III and 8 with Fon-taine stage IV).5 In their study intravenous L-arginineinduced peripheral dilation in the diseased limb, per-

TABLE I Physical Characteristics and Cardiac Risk Factors of PAD and Non-PADPatients

PAD(n 5 50)

Non-PAD(n 5 50)

Age (yr) 69.9 6 1.1 68.1 6 0.8Weight (kg) 84.3 6 2.3 83.7 6 2.3Body mass index (kg/m2) 29.9 6 0.7 28.1 6 0.6Systolic blood pressure (mm Hg) 153 6 4* 141 6 3Diastolic blood pressure (mm Hg) 77 6 1 80 6 2Ankle-to-brachial index 0.67 6 0.03 —Fasting glucose 129 6 6† 109 6 5Glucose 2 hours (mg/dl) 156 6 11‡ 132 6 7Triglycerides 142 6 10 140 6 9Total cholesterol 191 6 5 191 6 4Low-density lipoprotein cholesterol (mg/dl) 122 6 5 120 6 4High-density lipoprotein cholesterol (mg/dl) 41 6 2 44 6 2Low-density to high-density lipoprotein cholesterol

ratio (mg/dl)3.2 6 0.2 3.0 6 0.1

Smoking pack-years 54 6 7* 25 6 3

*p ,0.01; †p ,0.001; ‡p ,0.05.Values expressed as mean 6 SEM.Two-hour glucose value during oral glucose tolerance test measured in 23 PAD and 36 non-PAD

patients with fasting glucose ,140 mg/dl.

756 THE AMERICAN JOURNAL OF CARDIOLOGYT VOL. 83 MARCH 1, 1999

haps due to increased nitric oxide synthesis. Collec-tively, results from the present study and the work ofHarris and Bode-Boger suggest that patients withPAD have defects in nitric oxide–dependent periph-eral vasodilation.

In bivariate analyses performed on the total popu-lation of subjects, the presence of hypertension, dys-lipidemia, diabetes mellitus, and cigarette smokingwere associated with impaired flow-mediated endo-thelial reactivity. In multiple regression analyses, sys-tolic blood pressure accounted for the largest percent-age of the variance in the percent change in diameterat 1 minute, with glucose intolerance and dyslipidemiaalso independently contributing to the variance. Ourfindings are consistent with those reported by otherinvestigators that demonstrate impaired endothelialfunction in patients with essential hypertension,10–12

the insulin resistance syndrome,13 smokers,14,15 andhyperlipidemia.16–18 It has been proposed that thesedisorders result in decreased endothelium-derived ni-tric oxide production and/or release.19,20 Given themultiple metabolic abnormalities present in this pop-ulation, and the propensity for a clustering of theserisk factors in a given subject, it is difficult to ascertainthe independent contributions of these disorders toendothelial dysfunction.14,15 For the clinician, it isimportant to note that, because risk factors for athero-sclerosis are modifiable, it is plausible that risk factorintervention may improve endothelial function in pa-tients with PAD. The effect of aggressive, multiplerisk factor intervention on endothelial function in thispopulation of older claudicants remains to be deter-mined.

Although the major focus of this study is on im-

paired flow-mediated endothelial vasodilation in pa-tients with PAD, one must not lose sight of the pres-ence of the markedly impaired endothelial functionpresent in many of the non-PAD patients. The mean6SD for the posthyperemic percent change in diameterat 1 minute for young healthy controls studied in ourlaboratory is 12.66 6%.21 Based on the value of 6%equivalent to 1 SD below the norm, 32% of oldernon-PAD patients had impaired endothelium-depen-dent vasodilation of their brachial arteries. This im-paired endothelial-dependent vasodilatory response isattributable in part to the presence of multiple riskfactors for atherosclerosis in the non-PAD population.There may also be a primary age-associated decline inendothelium-dependent vasodilation that contributesto their impaired functioning.22–24Nevertheless, thesemeasures of vascular health are abnormal in a signif-icant percentage of this nonclaudicant older popula-tion.

The limitations of this study warrant comment.One limitation is the potential confounding effects ofmedications on endothelial reactivity. We were unableto withdraw these chronically ill patients from theirmedications for extended periods of time. Some va-soactive medications (including calcium antagonists,angiotensin-converting enzyme inhibitors, nitrates)may directly impact on endothelial function, whereasother medications (including lipid-lowering medica-tions, oral hypoglycemic agents) indirectly modulateendothelial function through their effects on cardiacrisk factors. Because of the broad spectrum of drugsadministered and the marked variation in the numberand doses of the medications, we were unable toexamine the interaction between medication status,cardiac risk factors, and endothelial function. Second,there are a number of other metabolic factors, such ashomocysteine, that may impact on the endothelialfunction of these patients.21 Homocysteine levels werenot measured in this study, and its contribution toendothelial dysfunction in this population remains tobe determined. Finally, our population was predomi-nantly men.

Acknowledgment: We want to thank Charles Man-gano for his technical assistance in collecting theendothelial reactivity measurements.

1. Kannel WB, McGee. Update on some epidemiologic features of intermittentclaudication: the Framingham Study.J Am Geriatr Soc1985;33:13–18.2. Murabito JM, D’Aostino RB, Silbershatz H, Wilson PWF. Intermittent clau-dication. A risk factor profile from the Framingham Heart Study.Circulation1997;96:44–49.3. Howard G, Wagenknecht LE, Burke GL, Diez-Roux A, Evans GW, McGovernP, Nieto J, Tell GS. Cigarette smoking and progression of atherosclerosis. TheAtherosclerosis Risk in Communities (ARIC) study.JAMA 1998;279:119–124.4. Harris LM, Faggioli GL, Shah R, Koerner N, Lillis L, Dandona R, Izzo JL,Snyder B, Ricotta JL. Vascular reactivity in patients with peripheral vasculardisease.Am J Cardiol1995;76:207–212.5. Bode-Boger SM, Boger RH, Alfke H, Heinzel D, Tsikas D, Creung A,Alexander K, Frolisch JC. L-arginine induces nitric-oxide dependent vasodilationin patients with critical limb ischemia. A randomized, controlled study.Circu-lation 1996;93:85–90.6. Rose GA, Blackburn H. Cardiovascular survey methods.WHO Monograph1968;56:162–168.7. National Cholesterol Education Program. Summary of the second report of the

TABLE II Endothelial Function Measurements of PAD andNon-PAD Patients

PAD(n 5 50)

Non-PAD(n 5 50)

Baseline diameter (mm) 3.70 6 0.09 3.59 6 0.08Baseline velocity (m/min) 9.3 6 0.8 11.1 6 1.3Baseline flow (ml/min) 62.4 6 6.6 70.1 6 8.9Hyperemia velocity (m/min) 84.0 6 4.5* 98.8 6 3.8Hyperemia flow (ml/min) 598 6 36* 722 6 381-Minute diameter (mm) 3.92 6 0.09 3.92 6 0.08Change in diameter (mm) (1 min) 0.22 6 0.02† 0.33 6 0.02% Change in diameter (1 min) 6.45 6 0.68† 9.79 6 0.73

*p ,0.05; †p ,0.001.Values expressed as mean 6 SEM.

TABLE III Stepwise Multiple Regression of Predictors of Flow-Mediated Percent Change in Brachial Artery Diameter in100 Patients*

Parameter Value SEM Cumulative r2

Intercept 27.5* 2.7*Systolic blood pressure 20.081* 0.02* 0.23Fasting glucose 20.31* 0.01* 0.32Low-density to high-density

lipoprotein cholesterol ratio21.16† 0.42† 0.37

*p 5 0.001; p 5 0.01.

MISCELLANEOUS/ENDOTHELIAL REACTIVITY IN PERIPHERAL ARTERIAL DISEASE 757

expert panel on detection, evaluation, and treatment of high blood cholesterol inadults (adult treatment panel II).JAMA 1993;269:3015–3023.8. Corretti MC, Plotnick GD, Vogel RA. Technical aspects of evaluating brachialartery vasodilatation using high-frequency ultrasound.Am J Physiol 1995;268(Heart Circ Physiol 37):H1397–H1404.9. Gardner AW, Skinner JS, Cantwell BW, Smith LK. Progressive vs. single-stage treadmill tests for evaluation of claudication.Med Sci Sports Exerc1991;23:402–408.10. Laurent S, Lacolley P, Brunel P, Laloux B, Pannier B, Safar M. Flow-dependent vasodilatation of brachial artery in essential hypertension.Am JPhysiol1990;258(Heart Circ Physiol 27):H1004–H1011.11. Safar ME, Peronneau PA, Levenson JA, Toto-Moukouo JA, Simon AC. PulseDoppler: diameter, blood flow velocity and volumic flow of the brachial artery insustained essential hypertension.Circulation 1981;63:393–400.12. Panza JA, Quyyumy A, Brush JE Jr, Epstein SE. Abnormal endothelium-dependent vascular relaxation in patients with essential hypertension.N EnglJ Med1990;323:22–27.13. McVeigh GE, Brennan GM, Johnston GD, McDermott BJ, McGrath LT,Andrews JW, Hayes JR. Impaired endothelium-dependent and independent va-sodilatation in patients with type 2 (non-insulin-dependent) diabetes mellitus.Diabetologia1992;35:771–776.14. Celermajer DS, Sorensen KE, Georgakopoulos D, Bull C, Thomas O,Robinson J, Deanfield JE. Cigarette smoking is associated with dose-related andpotentially reversible impairment of endothelium-dependent dilation in healthyyoung adults.Circulation 1993;88:2149–2155.15. Celermajer DS, Sorensen KE, Bull D, Robinson J, Deanfield JE. Endotheli-um-dependent dilatation in the systemic arteries of asymptomatic subjects relatesto coronary risk factors and their interaction.J Am Coll Cardiol.1994;24:1468–1474.

16. Chowienczyk PJ, Watts JR, Cockcroft JR, Ritter JM. Impaired endothelium-dependent vasodilation of forearm resistance vessels in hypercholesterolemia.Lancet1992;340:1430–1432.17. Creager MA, Cooke JP, Mendelsohn ME, Gallagher SJ, Coleman SM,Loscalzo J, Dzau VJ. Impaired vasodilatation of forearm resistance vessels inhypercholesterolemic humans.J Clin Invest1990;86:228–234.18. Celermajer DS, Sorensen KE, Gooch VM, Spiegelhalter DJ, Miller OI,Sullivan ID, Lloyd JK, Deanfield JE. Non-invasive detection of endothelialdysfunction in children and adults at risk of atherosclerosis.Lancet1992;340:1111–1115.19. Oemar BS, Tschueli MR, Godoy N, Brokovich V, Malinski T, Lu¨sher TF.Reduced endothelial nitric oxide synthase expression and production in humanatherosclerosis.Circulation 1998;97:2494–2498.20. Luscher TF, Barton M. Biology of the endothelium.Clin Cardiol 1997;20(11suppl 2):II-3–II-10.21. Woo KS, Chook P, Lolin YI, Cheung ASP, Chan LT, Sun YY, Sanderson JE,Metreweli C, Celermajer DS. Hyperhomsyst(e)inemia is a risk factor for arterialendothelial dysfunction in humans.Circulation 1997;96:2542–2544.22. Celermajer DS, Sorensen KE, Spiegelhalter J, Georgakopoulos D, RobinsonJ, Deanfield JE. Aging is associated with endothelial dysfunction in healthy menyears before the age-related decline in women.J Am Coll Cardiol1994;24:471–476.23. Chauhan A, More RS, Mullins PA, Taylor G, Petch MC, Schofield PM.Aging-associated endothelial dysfunction in humans is reversed by L-arginine.J Am Coll Cardiol1996;28:1796–1804.24. Gerhard M, Roddy M-A, Creager SJ, Creager MA. Aging progressivelyimpairs endothelium-dependent vasodilation in forearm resistance vessels ofhumans.Hypertension1996;27:849–853.

758 THE AMERICAN JOURNAL OF CARDIOLOGYT VOL. 83 MARCH 1, 1999