relation of carotid artery wall thickness to diabetes mellitus

66

Relation of Carotid Artery Wall Thickness toDiabetes Mellitus, Fasting Glucose and Insulin,

Body Size, and Physical Activity

Aaron R. Folsom, MD; John H. Eckfeldt, MD, PhD; Shimon Weitzman, MD; Jing Ma, MD;Lloyd E. Chambless, PhD; Ralph W. Barnes, PhD; Kenneth B. Cram, MD;

Richard G. Hutchinson, MD, for the Atherosclerosis Risk in Communities(ARIC) Study Investigators

Background and Purpose We tested the hypothesis thatbody mass, waist-to-hip circumference ratio, physical inactiv-ity, diabetes, hyperglycemia, and fasting insulin are eachpositively associated with asymptomatic carotid artery wallthickness.

Methods Average intimal-medial carotid wall thickness (anindicator of atherosclerosis) was measured noninvasively byB-mode ultrasonography in cross-sectional samples of 45- to64 -year-old adults, both blacks and whites, free of symptom-atic cardiovascular disease, in four US communities.

Results Sample mean carotid wall thickness was approxi-mately 0.7 mm in women (n=7956) and 0.8 mm in men(n=6474). Body mass, waist-to-hip ratio, work physical activ-ity, diabetes, and fasting insulin were associated (P<.05) withcarotid wall thickness in the hypothesized direction. Adjustedfor age, race, smoking, body mass index, artery depth, and

Atherosclerosis Risk in Communities field center, mean wallthickness was greater by 0.02 mm in women and 0.03 mm inmen for a 0.07-unit (one SD) larger waist-to-hip ratio.Adjusted mean wall thickness was about 0.07 mm thicker inparticipants with diabetes mellitus and 0.02 mm thicker inparticipants with hyperglycemia (fasting glucose 6.4 to 7.7mmol/L) than in subjects with fasting glucose <6.4 mmol/L.Adjusted mean wall thickness increased by about 0.02 mm withan increase of 100 mmol/L in fasting serum insulin.

Conclusions Abdominal adiposity, physical inactivity, andabnormal glucose metabolism are associated positively withcarotid intimal-medial wall thickness, suggesting these factorscontribute to atherogenesis. (Stroke. 1994;25:66-73.)

Key Words • atherosclerosis • insulin • carotid arteries• diabetes mellitus • obesity

Epidemiologic studies have consistently demon-strated a positive association between diabetesmellitus and atherosclerotic cardiovascular dis-

ease (CVD). The association seems to be stronger inwomen than men.1 Impaired glucose tolerance short ofdiabetes, on the other hand, is not a firmly establishedrisk factor; there may be a nonlinear or thresholdrelation between fasting or postload hyperglycemia andCVD incidence.12 Insulin resistance is another possibleCVD risk factor.3 Prospective studies have reportedfasting and/or postglucose insulin concentrations posi-tively associated with incidence of CVD.1-4

Physical inactivity and increased body size are impor-tant contributors to insulin resistance and impairedglucose tolerance. Physical inactivity confers almost a

Received July 19, 1993; final revision received September 16,1993; accepted September 16, 1993.

From the Division of Epidemiology, School of Public Health(A.R.F., J.M.), and the Department of Laboratory Medicine andPathology (J.H.E.), University of Minnesota, Minneapolis, andNorth Memorial Medical Center (K.B.C.), Robbinsdale, Minn;Epidemiology Unit, Faculty of Health Sciences (S.W.), Ben-Gurion University, Beer-Sheva, Israel; Collaborative Studies Co-ordinating Center (L.E.C.), University of North Carolina, ChapelHill, and Ultrasound Reading Center (R.W.B.), Winston-Salem,NC; and the Department of Medicine (R.G.H.), University ofMississippi Medical Center, Jackson, Miss.

Correspondence and reprint requests to Aaron R. Folsom, MD,Division of Epidemiology, School of Public Health, University ofMinnesota, Suite 300, 1300 S Second St, Minneapolis, MN 55454.

two-fold increase in CVD incidence.5 Body mass hasbeen associated positively with CVD in many studies,although other studies reported a U-shaped or noassociation.6 More recently, studies of fat distributionhave shown that abdominal adiposity is an importantdeterminant of CVD risk.7

A shortcoming of epidemiologic studies employingclinical CVD end points is an inability to disentanglethe two processes related to most atherosclerotic events,namely atherosclerosis and thrombosis. Noninvasivetechniques such as B-mode ultrasound can more di-rectly assess atherosclerosis. We report here the rela-tion of carotid artery wall thickness to diabetes, fastinghyperglycemia, fasting insulin, body shape and size, andphysical activity in a large sample of middle-aged adults.

Subjects and Methods

Study PopulationThe Atherosclerosis Risk in Communities (ARIC) Study is

a prospective investigation of atherosclerosis and clinical ath-erosclerotic diseases in four US communities: Forsyth County,NC; Jackson, Miss; the northwest suburbs of Minneapolis,Minn; and Washington County, Md. A population sampletotaling 15 800 persons aged 45 to 64 years was selected; onlyblacks were sampled in Jackson. The complete study design,sampling strategy, and examination techniques have beenpublished.8 The response rate was 46% of all eligible subjectsin Jackson and 65% to 67% in the other three communities.

by guest on February 10, 2018http://stroke.ahajournals.org/

Dow

nloaded from

http://stroke.ahajournals.org/

Folsom et al Correlates of Carotid Wall Thickness 67

MeasurementsAfter informed consent, a home interview and clinic exam-

ination were conducted. The home interview included assess-ment of socioeconomic factors, education level, smoking sta-tus, medication use, and history of cardiovascular disease anddiabetes mellitus. The clinic examination included measure-ments of cardiovascular risk factors and a B-mode ultrasoundexamination of selected arterial sites.

Subjects were asked to fast for 12 hours before the clinicexamination. Blood specimens were collected into vacuumtubes containing serum separator gel (insulin and glucose) andEDTA (lipids). The tubes were centrifuged at 3000g for 10minutes at 4°C. After separation, aliquots were quickly frozenat -70°C until analysis within a few weeks. Serum glucose wasassessed by the hexokinase method. Serum insulin was as-sessed using a radioimmunoassay (125Insulin Kit; CambridgeMedical Diagnostics, Inc, Billerica, Mass). Total cholesterol9

and triglycerides10 were measured by enzymatic methods, andlow-density lipoprotein (LDL) cholesterol was calculated.11

High-density lipoprotein (HDL) cholesterol was measuredafter dextran-magnesium precipitation.12 The lipid laboratoryparticipated in the Centers for Disease Control Standardiza-tion Program throughout the study.

Body mass index (kg/m2) was computed from height andweight. Circumferences of the waist (umbilical level) and hip(maximum buttocks) were measured to the centimeter. Theratio of waist-to-hip circumferences was calculated as a mea-sure of fat distribution.

Physical activity was assessed by interview using a question-naire developed by Baecke.13 The questionnaire included 16items about usual exertion, and three indexes ranging from 1(low) to 5 (high) were derived for physical activity at work,during leisure time, and in sports. Leisure time physicalactivity showed no relation with carotid atherosclerosis, soassociations are shown for the work and sport indexes only.Systolic and diastolic fifth phase blood pressures were mea-sured three times in the right arm of seated participants. Themean of the last two measurements was used in analysis.

Prevalent hypertension was defined as systolic pressurea 140 mm Hg or diastolic pressure >90 mm Hg or use ofantihypertensives. Because an oral glucose tolerance test wasnot performed, glycemia status was defined by fasting valuesand self-reported medical history. Prevalent diabetes mellituswas defined as a fasting glucose >7.8 mmol/L (>140 mg/dL),a nonfasting glucose >11.1 mmol/L (>200 mg/dL), and/or ahistory of or treatment for diabetes. Hyperglycemia was de-fined as a fasting glucose of 6.4 to 7.7 mmol/L (115 to 139mg/dL) and no history of or treatment for diabetes. PrevalentCVD was defined as a positive history of angina or intermit-tent claudication by the Rose questionnaire14; a self-reportedphysician-diagnosed history of a heart attack or stroke; evi-dence of old myocardial infarction by electrocardiogram; or aself-reported history of cardiovascular surgery or angioplasty.

Carotid artery atherosclerosis was determined by high res-olution B-mode ultrasound.15 Trained technicians in each fieldcenter scanned the extracranial carotid arteries bilaterally(and one popliteal artery, not reported here). The carotidarteries were divided into three segments: the distal 1.0-cmstraight portion of the common carotid artery, the carotidbifurcation, and the proximal 1.0 cm of the internal carotidartery. Thus, including both sides, six artery segments werescanned. B-mode ultrasound data were recorded onto high-resolution 3/4-in. cassettes and interpreted by trained readersaccording to a standardized protocol.16 Using magnified im-ages, readers recorded for each segment up to 11 coordinatepoints from each echogenic interface that identified an arterialboundary. Between-reader and within-reader coefficients ofreliability for average artery wall thickness in the threesegments were all r>.90.16

Our primary interest was the average intimal-medial thick-ness of the far wall of the six carotid artery segments. The

rationale for this measure as an indicator of early atheroscle-rosis has been described,17 as has its strong association withmajor cardiovascular risk factors.18 Poor visualization of ca-rotid boundaries led to missing information in part of the sixsegments on some scans, with a greater degree of missingnessfor deeper arteries or participants with greater body massindex. However, controlling for race, sex, body mass index, andartery depth, degree of visualization was not strongly associ-ated with standard coronary heart disease risk factors. Wetherefore assumed that, conditional on race, sex, body massindex, and artery depth, missingness was random.

Statistical MethodsThis analysis was restricted to participants with usable

ultrasound data who reported their race as black or white. Weexcluded glucose and insulin values from analysis for subjectswho had not fasted at least 8 hours. Insulin values were alsoexcluded if the subject reported current pharmacologic treat-ment for diabetes. The final sample size was 14 430 of the15 800 original participants.

The characteristics of the entire ARIC cohort were de-scribed as sex- and race-specific means and standard devia-tions or prevalence estimates. Differences in characteristics ofthose with prevalent CVD versus no CVD at baseline werecompared using x1 or ' tests. Analysis of the B-mode ultra-sound data were restricted to those without prevalent CVD atbaseline. To include in analysis all participants with wallthickness data in at least one of the six sites, multivariatenormal maximal likelihood techniques were applied to fitmodels of site-specific wall thickness as a function of potentialrisk factors and adjusting variables, using the EM algorithm19

in BMDP5V.20 Models were performed sex-specific and alwayscontrolled for body mass index, artery depth, and race to morenearly assure that missingness in the B-mode ultrasound datawas random. Risk factor stratum-specific age-adjusted (imput-ed) mean wall thickness was computed by ANCOVA usingSAS PROC GLM.21

ResultsA description of ARIC participants is given in Table 1.

Among women, blacks had greater adiposity than whites,lower sport physical activity, and higher levels of workactivity, fasting glucose, insulin, and diabetes prevalence.Among men, there were virtually no racial differences inadiposity and insulin. However, similar to women, blackmen had lower sport physical activity than white men andhigher glucose levels and diabetes prevalence.

Compared with those who were free of CVD atbaseline (Table 2), those who reported CVD had agreater mean body mass index and waist-to-hip ratio,lower sport and work activity indexes, higher fastingglucose and insulin concentrations, and a 1.4- to 2.6-fold greater prevalence of diabetes. These findings werestatistically significant (P< .05) for most race-sex groups,given the large ARIC sample size.

The remaining analyses were performed after exclud-ing those with prevalent CVD to focus on early carotidatherosclerosis in asymptomatic subjects. The distribu-tion of carotid artery intimal-medial wall thickness wasslightly right-skewed. The sample mean (SD) wall thick-ness in those without CVD was 0.69 (0.16) mm in whitewomen (n=4340), 0.72 (0.15) mm in black women(n=1682), 0.79 (0.20) mm in white men (n=3645), and0.77 (0.17) mm in black men (n=1047). The respectivemedians were 0.66, 0.69, 0.75, and 0.75 mm.

Figs 1 and 2 illustrate for the four race-sex groups theassociation of age-adjusted average carotid wall thick-ness with body mass, waist-to-hip ratio, the sport index,


Dow

nloaded from


68 Stroke Vol 25, No 1 January 1994

TABLE 1 . Selected Baseline Characteristics of Participants by Race and Sex in the ARIC Study,1987 to 1989

Characteristic

Age, y

Body mass index, kg/m2

Waist-to-hip ratio

Sport index*

Work index*

Fasting serum glucose, mmol/L

Fasting serum insulin, mmol/Lt

Diabetes mellitus, °/<4

Plasma LDL cholesterol, mmol/L

Plasma HDL cholesterol, mmol/L

Cigarette smoking, pack-years

Hypertension, %

White(n=5520)

54.0±5.7

26.6±5.5

0.89±0.08

2.4±0.8

2.0±0.9

5.7±1.6

68 ±54

6.3

3.5±1.0

1.5+0.5

12±18

26.1

Women

Black(n=2436)

53.4+5.7

30.8+6.5

0.90 ±0.08

2.1 ±0.7

2.3±1.0

6.3±2.7

105±75

14.7

3.6±1.1

1.5+0.5

8+15

55.7

White(n=4964)

54.8+5.7

27.4+4.0

0.97+0.05

2.7±0.8

2.3+0.9

6.0+1.5

80+57

7.4

3.6+0.9

1.1+0.3

24+25

28.5

Men

Black(n=1510)

53.9±6.0

27.6±4.9

0.94±0.06

2.3±0.8

2.3±1.0

6.2±2.2

82±61

12.6

3.6±1.1

1.3±0.4

20±25

54.8

ARIC indicates Atherosclerosis Risk in Communities; LDL, low-density lipoprotein; and HDL, high-densitylipoprotein. Values are mean±SD or percent.

*Range 1 to 5 (low to high).tExcludes pharmacologically treated diabetics.•Reported and/or treated diabetes, fasting glucose >7.8 mmol/L, or nonfasting glucose >11.1 mmol/L.

fasting insulin, fasting glucose, and diabetes. Tables 3and 4 show the associations by sex from regressionmodeling. Twelve separate models were run, two (mod-els 1 and 2) for each characteristic considered sepa-rately. Simultaneous adjustment was viewed as inappro-priate because body size, physical activity, and glucoseand insulin metabolism are causally intertwined; yetadjustment for body mass index was required to accountfor possible nonrandom missing wall thickness data (see

"Methods"). Therefore, the most appropriate assess-ment of the overall associations is model 1, whichadjusted for age, race, pack-years of smoking, arterydepth, ARIC field center, and body mass index. Model2 assessed whether the associations in model 1 might beexplained by several mediating factors, ie, HDL choles-terol, LDL cholesterol, and hypertension. The regres-sion coefficients, for the most part, were larger for menthan for women. Most of the associations were statisti-

TABLE 2. Characteristics of Participants With or Without Cardiovascular Disease* at Baseline, by Race and Sex inthe ARIC Study, 1987 to 1989

Characteristic

Age, y

Body mass index, kg/m2

Waist-to-hip ratio

Sport index

Work index

Fasting serum glucose,mmol/L

Fasting serum insulin,mmol/L+

Diabetes mellitus, %§

Women

White

CVD(n=728)

55.0

27.9

0.916

2.3

1.9

6.1

81

11.3

t

t

t

t

t

t

t

t

No CVD(n=4677)

53.8

26.4

0.887

2.4

2.0

5.7

66

5.4

CVD(n=318)

54.6

31.6

0.927

2.1

2.0

6.4

103

20.0

Black

t

t

t

t

t

No CVD(n=2079)

53.2

30.6

0.901

2.1

2.3

6.3

106

14.2

CVD(n=857)

56.3

28.0

0.980

2.6

2.2

6.4

88

15.0

White

t

t

t

t

t

t

t

t

No CVD(n=4014)

54.5

27.3

0.967

2.7

2.4

5.9

79

5.7

Men

CVD(n=226)

55.7

27.8

0.955

2.2

1.8

6.5

79

25.7

Black

t

t

t

t

t

No CVD(n=1256)

53.6

27.6

0.939

2.3

2.4

6.1

82

10.5

ARIC indicates Atherosclerosis Risk in Communities; CVD, cardiovascular disease. Values are mean or percent.*Self-reported heart attack, stroke, angioplasty, cardiovascular surgery; positive Rose questionnaire for angina or claudication; or

electrocardiograph^ evidence of myocardial infarction.tP<.05 for sex- and race-specific comparison between CVD and no CVD.$ Excludes pharmacologically treated diabetic subjects.§Reported and/or treated diabetes, fasting glucose >7.8 mmol/L, or nonfasting glucose >11.1 mmol/L.


Dow

nloaded from



« 2 2 2 2 - 2 4 - 2 6 - 2 1 - J O - 3 2 - 1 4 - 3 6 .

BODY MASS INDEX (KG/M2)

WHITE WOMEN

BUCK WOMEN

WHITE MEN

BLACK MEN

s

inin

zuXt—

-t

3

ME

AN

0.8-

0.7-

0.6-

serfs'~>^^^^ . **^

* ~~~&**^^~—*—~~—<'^"^

/ ^

< 0 . B 0 0 . S 0 - 0 . S 5 - O . g o - 0 . » 5 - 1 . 0 0 - 1 . 0 5 .

WAIST/HIP RATIO

WHITE WOMEN

BUCK WOMEN

WHITE MEN

BLACK MEN

en

1WHITE WOMENBLACK WOMENWHTEMENBLACK MEN

en<nuizo

z4

4 1 . 5 1 . 5 - 2 . 0 - 2 . 5 - 3 . 0 - 3 . 5 -

SPORT INDEX

WHITE WOMENBLACK WOMENWHITE MENBLACK MEN

70 - 1 0 5 - 1 4 0 -

FASTING INSULIN (MMOL/L)

FIG 1. Sex- and race-specific associations of average carotid artery intimal-medial wall thickness with body mass index, waist-to-hipratio, sport index, and fasting insulin (excluding pharmacologically treated diabetics), in participants free of cardiovascular disease in theAtherosclerosis Risk in Communities (ARIC) Study, 1987 to 1989.

cally significant in model 1, given the large sample size.Adjustment for hypertension, LDL cholesterol, andHDL cholesterol reduced the associations.

Age-adjusted average carotid intimal-medial wallthickness increased steadily with increasing body massand waist-to-hip ratio in each sex-race group. Frommodel 1, the predicted increase in wall thickness for a5-kg/m2 (1 SD) increase in body mass was about 0.017mm for women and about 0.026 mm for men. Thepredicted increase in wall thickness for an increase of

0.07 unit (1 SD) in waist-to-hip ratio was 0.020 mm inwomen and 0.029 mm in men. (To place these inperspective, a 0.02-mm increase would represent abouta 3% increase in average carotid wall thickness.) Boththe body mass and waist-to-hip ratio associations withwall thickness were reduced but remained statisticallysignificant (P<.01) after adjustment for other major riskfactors.

Wall thickness decreased with increasing physicalexertion at work for both men and women; by model 1

mui

yz

ill2

FASTING GLUCOSE

• < 6.4 MMOLL

E3 6.4-7.7 MMOLL

0 DIABETES MELUTUS

FIG 2. Bar graph of sex- and race-specificassociations of average carotid artery intimal-medial wall thickness with fasting glucose anddiabetes mellitus, in participants free of cardio-vascular disease in the Atherosclerosis Risk inCommunities (ARIC) Study, 1987 to 1989.

WHITE WOMEN BLACK WOMEN WHITE MEN HACK M£N


Dow

nloaded from



TABLE 3. Adjusted Regression Coefficients (p) for Multiple Linear Regressions of Average Carotid ArteryIntimal-Medial Wall Thickness (mm) on Selected Characteristics in Women Free of CardiovascularDisease in the ARIC Study, 1987 to 1989

Model 1 Model 2

Characteristic* (separately considered) p0.0033

0.2885

-0.0063

-0.0009

0.0716

0.0203

0.00017

P<.01

<.01

.03

.78

<.01

.08

< 01

P0.0021

0.2135

-0.0060

0.0011

0.0605

0.0113

0.00009

P<.01

<.01

.03

.75

<.01

.32

.06

(1) Body mass index, kg/m2

(2) Waist-to-hip ratio, 1 unit

(3) Work index, 1 unit

(4) Sport index, 1 unit

(5) Diabetes (yes vs no)

Hyperglycemia (yes vs no)t

(6) Fasting insulin, mmol/14

* Average wall thickness was the dependent variable, and 12 separate regression models were run with the six listedcharacteristics as separately considered independent variables adjusted successively for two sets of covariates.

Model 1 coefficients were adjusted for age, race, pack-years of smoking, artery depth, Atherosclerosis Risk inCommunities (ARIC) field center, and body mass index [except (1)]; Model 2 coefficients were adjusted also forhypertension (yes, no), plasma low-density lipoprotein cholesterol, and plasma high-density lipoprotein cholesterol.

tHyperglycemia=fasting glucose 6.4 to 7.7 mmol/L.$ Excludes pharmacologically treated diabetes.

estimation, the decrease was 0.006 to 0.010 mm per unitincrease in the work index. Inclusion of the other majorrisk factors (model 2) did not appreciably alter thisestimate. The association for sports participation, how-ever, was not statistically significant in either sex.

By model 1 estimation, in both sexes, diabetic subjectshad carotid artery walls about 0.07 mm thicker andsubjects with fasting glucose equal to 6.4 to 7.7 mmol/Lhad artery walls about 0.02 mm thicker than subjectswith fasting glucose <6.4 mmol/L (Fig 2 and Tables 3and 4). Adjustment for hypertension and plasma lipidsreduced these estimates somewhat. Among those notpharmacologically treated for diabetes, a 100 -mmol/Lgreater fasting insulin was associated in model 1 withthicker carotid artery walls by about 0.017 mm inwomen and 0.016 mm in men.

It has been suggested that the sex difference incardiovascular disease may be due to a lower waist-to-

hip ratio in women19 and is greatly diminished indiabetic subjects.20 However, the waist-to-hip ratio anddiabetes graphs in Figs 1 and 2 suggest that carotidatherosclerosis is higher in men than women both foreach level of waist-to-hip ratio and for participants withdiabetes. In a linear regression model that included age,race, pack-years of smoking, body mass index, arterydepth, and ARIC field center, women were predicted tohave a 0.08-mm lower average wall thickness than men.When a term for waist-to-hip ratio was added to themodel, this male/female difference was mildly reducedto 0.06 mm. Alternatively, when terms for hyperglyce-mia and diabetes were added, the male/female differ-ence of 0.08 mm was unchanged.

We also ran model 1 regressions separately for theaverage wall thickness of the common carotid, thebifurcation, and the internal carotid artery segments.Associations tended to be in the same direction for all

TABLE 4. Adjusted Regression Coefficients (P) for Multiple Linear Regressions of Average Carotid ArteryIntimal-Medial Wall Thickness (mm) on Selected Characteristics in Men Free of Cardiovascular Diseasein the ARIC Study, 1987 to 1989

Model 1 Model 2

Characteristic* (separately considered) P0.0052

0.4155

-0.0102

-0.0043

0.0731

0.0184

0.00016

P<.01

<.01

<.01

.31

<.01

.12

.01

P0.0037

0.3632

-0.0093

-0.0034

0.0698

0.0147

0.00013

P<.01

<.01

.01

.42

<.01

.21

.04

(1) Body mass index, kg/m2

(2) Waist-to-hip ratio, 1 unit

(3) Work index, 1 unit

(4) Sport index, 1 unit

(5) Diabetes (yes vs no)

Hyperglycemia (yes vs no)t

(6) Fasting insulin, mmol/L*

* Average wall thickness was the dependent variable and 12 separate regression models were run with the six listedcharacteristics as separately considered independent variables adjusted successively for two sets of covariates.

Model 1 coefficients were adjusted for age, race, pack-years of smoking, artery depth, Atherosclerosis Risk inCommunities (ARIC) field center, and body mass index [except (1)]; Model 2 coefficients were adjusted also forhypertension (yes, no), plasma low-density lipoprotein cholesterol, and plasma high-density lipoprotein cholesterol.

tHyperglycemia indicates fasting glucose of 6.4 to 7.7 mmol/L.^Excludes pharmacologically treated diabetes.


Dow

nloaded from



three arterial segments, but there was some variation inthe strength. For example, associations of wall thicknesswith glycemia status, the work index, and waist-to-hipratio were generally stronger for the bifurcation andinternal carotid segments than for the common carotid.

DiscussionEpidemiologic studies of atherosclerotic diseases typ-

ically study clinical end points, such as myocardialinfarction or coronary heart disease death. However,among those with underlying atherosclerosis, clinicalatherosclerotic events occur only infrequently, usuallybecause of rupture of a plaque and subsequent throm-bosis.22 A theoretical advantage of B-mode ultrasoundis that risk factors for atherosclerosis might be directlyquantifiable. Indeed, the clinical use of B-mode inmeasuring symptomatic carotid artery stenoses is wellestablished. However, because most ARIC participantsdo not have stenotic carotid lesions, we measuredaverage carotid intimal-medial wall thickness. It is pos-sible that greater than average wall thickness mayrepresent not only early atherosclerosis, but also medialhypertrophy or hyperplasia from other causes such ashypertension. The validity of intimal-medial wall thick-ness as a measure of atherosclerosis and its reliability inthe ARIC Study have been described.1617 Wall thick-ness, in ARIC, is strongly associated with atherogeniclipids and smoking, as well as hypertension,18 suggestingthat the atherosclerotic process is reflected in intimal-medial wall thickness measures. Preliminary ARIC dataalso suggest that carotid intimal-medial thickness pre-dicts coronary disease prospectively (unpublished data).

There are, nevertheless, drawbacks of B-mode ca-rotid wall thickness as an end point. Although athero-sclerosis in the carotid arteries is correlated with ath-erosclerosis in the coronaries and elsewhere,2324 thecarotid represents only a small segment of the arterialsystem. Frequently, the boundaries of the arterial tis-sues were not fully identifiable in the B-mode image(see "Methods"), so the ARIC investigators had toimpute missing boundaries from the existing bound-aries. Finally, this particular analysis may be susceptibleto the usual biases of cross-sectional studies; however,we believe their likelihood was reduced by the exclusionof participants with symptomatic CVD.

The strengths of ARIC are its size and its biracialpopulation-based samples. We found that greater bodymass and waist-to-hip ratio, less physical activity,greater fasting insulin and glucose, and diabetes were allrelated to prevalent CVD, as is generally accepted.1-46-7

More importantly, our main findings were that thesefactors were similarly associated with greater carotidartery wall thickness in participants with no previousclinical manifestation of CVD. The associations withwall thickness were similar for the most part in blacksand whites, were statistically significant, and demon-strated a dose-response. The associations appeared tobe modest in magnitude, causing alterations on theorder of tenths of millimeters. On the other hand, itshould be remembered that the average intimal-medialthickness was only about 0.7 mm.

The combination of age, race, and any single riskfactor explained only 11% to 15% of the variance in wallthickness. There are several possible explanations, notmutually exclusive, for the small amount of variance

explained: (1) the associations of these factors withcarotid atherosclerosis may in truth be weaker thantheir associations with coronary atherosclerosis, clinicalCVD, or thrombosis; (2) the associations may in fact bestrong but we measured them imprecisely; or (3) un-known biases were operating. The first explanation hassome appeal. Increased body mass,25 physical inactivi-ty,26 and fasting insulin4 are less consistently implicatedas risk factors for stroke than for coronary disease. Also,body mass index is not consistently related to carotidatherosclerosis2732 or coronary atherosclerotic dis-ease.33 Waist-to-hip ratio and diabetes are generallyassociated with carotid atherosclerosis and stroke25-28-34

but not without exceptions.35-36 Furthermore, all ofthese characteristics have been associated with hemo-static variables,3740 suggesting that they may act byenhancing thrombosis as well as atherosclerosis. Thesecond possibility for the little variance in wall thicknessbeing explained may also be true. Error in measure-ments of risk factors may have contributed to underes-timation of associations.

The study of Salonen and Salonen32 has the mostcomparably reported measures of carotid intimal-me-dial thickening from population-based samples. Theseinvestigators found in 1224 Eastern Finnish men thatambulatory pulse pressure, cigarette smoking, serumLDL cholesterol, history of coronary heart disease,systolic blood pressure, and diabetes were the strongestrisk factors for maximal intimal-medial thickening. Bodymass index was not an independent risk factor. Wereported average intimal-medial thickness here, and con-firmed that along with atherogenic lipids, smoking, andhypertension,22 diabetes was a risk factor. To a lesserdegree so were body mass index, waist-to-hip ratio, fastinginsulin, and physical exertion at work, although lack ofsports participation was not. Salonen and Salonen32 didnot report whether waist-to-hip ratio, exercise, or fastinginsulin concentrations were related to carotid wall thick-ness. However, these investigators did demonstrate anassociation between carotid atherosclerosis and reducedwhole-body glucose uptake (a measure of insulin resis-tance) in a recent substudy of 30 nonobese subjects withatherosclerosis and 13 control subjects.41

Atherosclerosis has a greater predilection for thecarotid bifurcation than the internal or common carotidsegments. Tell et al35 have recently reported that riskfactors for carotid plaques vary somewhat by segment.Giral et al42 have shown variations among risk factorsfor plaques in the carotid, aorta, and the femoralarteries among hypercholesterolemic men. We also ob-served some differences in the association of examinedrisk factors with the wall thickness of three carotidsegments.

Comparisons of carotid wall thickness between menand women should be made cautiously because normalartery wall thickness may be greater in men thanwomen. We nevertheless tested in regression modelswhether sex differences in waist-to-hip ratio, or hyper-glycemia and diabetes, may explain the sex difference inwall thickness. Accounting for hyperglycemia and dia-betes had no effect on the sex difference in wallthickness. Accounting for waist-to-hip ratio reduced thesex difference in carotid thickness somewhat but did noteliminate it. Thus, the data are consistent with this


Dow

nloaded from



factor being partly responsible for sex differences inCVD.

There are several mechanisms by which the examinedrisk factors may be associated with atherosclerosis. Theo-ries of atherosclerosis suggest that two crucial steps areendothelial injury and LDL uptake by subendothelialmacrophages to form foam cells and eventually atheroscle-rotic plaques. Greater body mass, waist-to-hip ratio, phys-ical inactivity, fasting insulin, and diabetes all tend to beassociated with higher blood pressure, which may injureendothelium. They are also associated with higher plasmatriglycerides and LDL, and with lower HDL.43-44 Observedassociations with wall thickness were weakened whenhypertension and plasma lipids were included in theregressions (model 2), suggesting that hypertension andatherogenic lipids may indeed partly mediate the associa-tions. It is also possible that the examined risk factorsdirectly promote endothelial injury, LDL uptake, orsmooth muscle proliferation. For example, insulin hasbeen shown to enhance cholesterol transport into arteri-olar smooth muscle cells, and to increase the proliferationand cholesterol synthesis of these cells.45 Regardless of thecellular mechanisms, this study provides additional sup-port for the assertion that abdominal obesity, physicalinactivity, and abnormal glucose metabolism are athero-genic and that efforts for reducing them must beconsidered.

AcknowledgmentsThis study was supported by grant contracts N01-HC-55015,

N01-HC-55016, N01-HC-55018, N01-HC-55019, N01-HC-55020, N01-HC-55021, and N01-HC-55022 from the NationalHeart, Lung, and Blood Institute. The authors thank theARIC coinvestigators and staff for their important contribu-tions and Laura Kemmis, Larry Crum, and Gretchen Marcuccifor their help in manuscript preparation.

References1. Pyorala K, Laakso M. Macrovascular disease in diabetes mellitus.

In: Mann JI, Pyorala K, Teuscher A, eds. Diabetes in EpidemiologicPerspective. Edinburgh, Scotland: Churchill Livingstone; 1983:183-247.

2. Stamler R, Stamler J. Asymptomatic hyperglycemia and coronaryheart disease: a series of papers by the International CollaborativeGroup, based on studies on fifteen populations. J Chron Dis. 1979;32:683-691.

3. Reaven GM. Role of insulin resistance in human disease. Diabetes.1988;37:1595-1607.

4. Stout RW. Overview of the association between insulin and ath-erosclerosis. Metabolism. 1985;23(suppl 1):7-12.

5. Powell KE, Thompson PD, Caspersen CJ, Kendrick JS. Physicalactivity and the incidence of coronary heart disease. Anna RevPublic Health. 1987;8:253-287.

6. Barrett-Connor EL. Obesity, atherosclerosis, and coronary arterydisease. Ann Intern Med. 1985;103:1010-1019.

7. Lapidus L, Bengtsson C, Larsson B, Pennert K, Rybo E, SjostromL. Distribution of adipose tissue and risk of cardiovascular diseaseand death: a 12-year follow-up of participants in the populationstudy of women in Gothenburg, Sweden. Br Med J. 1984;289:1257-1261.

8. The ARIC Investigators. The Atherosclerosis Risk in Communities(ARIC) Study: design and objectives. Am J Epidemiol. 1989; 129:687-702.

9. Siedel J, Hagele EO, Ziegenhorn J, Wahlefeld AW. Reagent forthe enzymatic determination of serum total cholesterol withimproved lipolytic efficiency. Clin Chem. 1983;29:1075-1080.

10. Nagele U, Hagele EO, Sauer G, Wiedemann E, Lehmann P,Wahlefeld AW, Grube W. Reagent for the enzymatic determi-nation of serum total triglycerides with improved lipolytic effi-ciency. J Clin Chem Clin Biochem. 1984;22:165-174.

11. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the con-centration of low-density lipoprotein cholesterol in plasma,

without use of the preparative ultracentrifuge. Clin Chem. 1972;18:499-502.

12. Warnick GR, Benderson JM, Albers JJ. Ouantitation of high-density-lipoprotein subclasses after separation by dextran sulfateand Mg2+ precipitation. Clin Chem. 1982;28:1574. Abstract.

13. Baecke JAH, Burema J, Frijters JER. A short questionnaire forthe measurement of habitual physical activity in epidemiologicstudies. Am J Clin Nutr. 1982;36:936-942.

14. Rose GA, Blackburn H, Gillum RF, Prineas RJ. CardiovascularSurvey Methods, 2nd ed. World Health Organization MonographSeries; 1982;56.

15. Bond MG, Barnes R, Riley WA, Wilmoth SK, Chambless LE,Howard G, Owens B. High resolution B-mode ultrasound scanningmethods in the Atherosclerosis Risk in Communities Study(ARIC). J Neuroimag. 1991;l:68-73.

16. Riley WA, Barnes R, Bond MG, Evans GW, Chambless L, HeissG. High resolution B-mode ultrasound reading methods in theAtherosclerosis Risk in Communities Study (ARIC). / Neuroimag.1991;1:168-172.

17. Pignoli P, Tremoli E, Poli A, Oreste P, Paoletti R. Intimal plusmedial thickness of the arterial wall: a direct measurement withultrasound imaging. Circulation. 1986;74:1399-1406.

18. Heiss G, Sharrett AR, Barnes R, Chambless LE, Szklo M, AlzolaC, and the ARIC Investigators. Case-control analysis of athero-sclerosis and established risk factors. Am J Epidemiol. 1991;134:250-256.

19. Dempster AP, Laird NM, Rubin DB. Maximum likelihood fromincomplete data via the EM Algorithm (with discussion). J R StatSoc. 1977;B39:l-38.

20. Dixon WJ, ed. BMDP Statistical Software Manual, Vol 2. Berkeley,Calif: University of California Press; 1990:1385.

21. SAS Institute Inc. SAS/STA T User's Guide, Vols 1 and 2. Cary, NC:SAS Institute; 1985.

22. DeWood MA, Spores J, Notske R, Mouser LT, Burroughs R,Golden MS, Lang HT. Prevalence of total coronary occlusionduring the early hours of transmural myocardial infarction. N EnglJ Med. 1980;303:897-902.

23. Strong JP, Solberg LA, Restrepo C. Atherosclerosis in personswith coronary heart disease. Lab Invest. 1968;18:527-537.

24. Young W, Gofman JW, Tandy R, Malamud N, Waters ESG. Thequantitation of atherosclerosis. III. The extent of correlation ofdegrees of atherosclerosis within and between the coronary andcerebral vascular beds. Am J Cardiol. 1960;6:300-308.

25. Kannel WB, Wolf PA. Epidemiology of cerebrovascular disease.In: Russell RWR, ed. Vascular Disease of the Central NervousSystem. New York, NY: Churchill Livingstone; 1983:1-24.

26. Folsom AR, Ensrud KE. Cardiovascular benefits of enduranceexercise. In: Astrand PO, Shephard RJ, eds. Sports and HumanEndurance. The Encyclopaedia of Sports Medicine. Oxford,England: Blackwell Scientific Publications; 1992:420-437.

27. Crouse JR, Toole JF, McKinney WM, Dignan MB, Howard G,Kahl FR, McMahan MR, Harpold GH. Risk factors for extra-cranial carotid artery atherosclerosis. Stroke. 1987;18:990-996.

28. Thompson CJ, Ryu JE, Craven TE, Kahl FR, Crouse JR III.Central adipose distribution is related to coronary atherosclerosis.Arterioscler Thromb. 1991;ll:327-333.

29. Okumiya N, Tanaka K, Ueda K, Omae T. Coronary athero-sclerosis and antecedent risk factors: pathologic and epidemiologicstudy in Hisayama, Japan. Am J Cardiol. 1985;56:62-66.

30. Bonithon-Kopp C, Scarabin P-Y, Taquet A, Touboul P-J,Malmejac A, Guize L. Risk factors for early carotid atherosclerosisin middle-aged French women. Arterioscler Thromb. 1991;11:966-972.

31. Giral P, Filitti V, Levenson J, Pithois-Merli I, Plainfosse M-C,Mainardi C, Gold A, Simon A, and PCV Metra Group. Relation ofrisk factors for cardiovascular disease to early atherosclerosisdetected by ultrasonography in middle-aged normotensive hyper-cholesterolemic men. Atherosclerosis. 1990;85:151-159.

32. Salonen R, Salonen JT. Determinants of carotid intima-mediathickness: a population-based ultrasonography study in EasternFinnish men. / Intern Med. 1991;229:225-231.

33. Barrett-Connor EL. Obesity, atherosclerosis, and coronary arterydisease. Ann Intern Med. 1985;103:10I0-1019.

34. Welin L, Svardsudd K, Wilhelmssen L, Larsson B, Tibblin G.Analysis of risk factors for stroke in a cohort of men born in 1913.N Engl] Med. 1987;317:521-526.

35. Tell GS, Howard G, McKinney WM. Risk factors for site specificextracranial carotid artery plaque distribution as measured byB-mode ultrasound. / Clin Epidemiol. 1989;42:551-559.


Dow

nloaded from



36. Salonen JT, Seppanen K, Rauramaa R, Salonen R. Risk factors forcarotid atherosclerosis: the Kuopio Ischaemic Heart Disease RiskFactor Study. Ann Med. 1989;21:227-229.

37. Eichner ER. Antithrombotic effects of exercise. Am Fam Physician.1987;36:207-211.

38. Sundell IB, Nilsson TK, Hallmans G, Nygren C. The effect ofbody build, diet and endocrine factors on the extrinsic fibrinolyticsystem in healthy young women. Scand J Gin Lab Invest. 1988;48:557-564.

39. Fuller JH, Keen H, Jarrett RJ, Omer T, Meade TW, ChakrabartiR, North WRS, Stirling Y. Haemostatic variables associated withdiabetes and its complications. Br Med J. 1979;2:964-966.

40. Meade TW. Clotting factors and ischaemic heart disease: the epi-demiological evidence. In: Meade TW, ed. Anticoagulants andMyocardial Infarction: A Reappraisal. New York, NY: John Wiley &Sons, Ltd; 1984:91-111.

41. Laakso M, Sarlund H, Salonen R, Suhonen M, Pyorala K, SalonenJT, Karhapaa P. Asymptomatic atherosclerosis and insulinresistance. Arterioscler Thromb. 1991;11:1068-1076.

42. Giral P, Pithois-Merli I, Filitti V, Levenson J, Plainfosse MC,Mainardi C, Simon AC. Risk factors and early extracoronary ath-erosclerotic plaques detected by three-site ultrasound imaging inhypercholesterolemic men. Arch Intern Med. 1991;151:950-956.

43. Folsom AR, Burke GL, Byers CL, Hutchinson RG, Heiss G,Jacobs DR, Flack JM, Caan B. Implications of obesity for cardio-vascular disease in blacks: the CARDIA and ARIC studies. Am JClin Nutr. 1991;53:1604S-1611S.

44. Howard BV. Lipoprotein metabolism in diabetes mellitus. J LipidRes. 1987;28:613-628.

45. DeFronzo RA, Ferrannini E. Insulin resistance: a multifacetedsyndrome responsible for NIDDM, obesity, hypertension, dyslip-idemia, and atherosclerotic cardiovascular disease. Diabetes Care.1991;14:173-194.


Dow

nloaded from


HutchinsonA R Folsom, J H Eckfeldt, S Weitzman, J Ma, L E Chambless, R W Barnes, K B Cram and R G

Investigators.body size, and physical activity. Atherosclerosis Risk in Communities (ARIC) Study

Relation of carotid artery wall thickness to diabetes mellitus, fasting glucose and insulin,

Print ISSN: 0039-2499. Online ISSN: 1524-4628 Copyright © 1994 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Stroke doi: 10.1161/01.STR.25.1.66

1994;25:66-73Stroke.

http://stroke.ahajournals.org/content/25/1/66World Wide Web at:

The online version of this article, along with updated information and services, is located on the

http://stroke.ahajournals.org//subscriptions/

is online at: Stroke Information about subscribing to Subscriptions:

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

document. Permissions and Rights Question and Answer available in the

Permissions in the middle column of the Web page under Services. Further information about this process isOnce the online version of the published article for which permission is being requested is located, click Request

can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.Stroke Requests for permissions to reproduce figures, tables, or portions of articles originally published inPermissions:


Dow

nloaded from

http://stroke.ahajournals.org/content/25/1/66

http://www.ahajournals.org/site/rights/

http://www.lww.com/reprints

http://stroke.ahajournals.org//subscriptions/


relation of carotid artery wall thickness to diabetes mellitus

Documents