Association of Coagulation Factors and Inhibitors with Carotid Artery Atherosclerosis Early Results of the Atherosclerosis (ARIC) Study

Risk in Communities

Kenneth K. Wu, MD, Aaron R. Folsom, MD, Gerard0 Heiss, MD, C.E. Davis, PhD, Maureen G. Conlan, MD, and Ralph Barnes, PhD, for the ARIC Study Investigators Several population studies have shown that plasma levels offbrinogen and factor VII are significantly associated with ischemic cardiovascular events. However, there is little information regarding the association of hemostatic factors with early atherosclerosis. To evaluate this, we compared the plasma concentrations of several coagulation proteins (fibrinogen, factor VII, factor VIII, won Wilkbrand factor, protein C, and antithrombin III) between 385 case patients, defined by highresolution B-mode ultrasonography as having carotid arterial wall thickening, and 385 age-, race-, and sex-matched control subjects, These case patients and control subjects were selected from participants in a prospective population investigation, the Atherosclerosis Risk in Communities (ARK) Study, who were examined between May 1987 and May 1989. Plasma fibrinogen, factor VII, protein C, and antirhrombin JJJ levels were significantly higher in case patients than in control subjects (P < 0.05). Factor VJIJ and uon Wilebrand factor were not different. These jindings were supported by quartile distribution and univariate analysis. Howewer, only jibrinogen remained significantly associated with carotid atherosclerosis on multiuariate analysis taking other atherosclerosisrisk factors into consideration. A one standard deviation increase in fibrinogen (67 mg/dL) was associated with a I. 6-fold increase in the odds of carotid atherosclerosis univariately (P < 0.001) and with a I. 3,fold zncrease in the odds multivariately (P = 0.010). Further analysis revealed that the association offbrinogen with carotid atherosclerosis was somewhat stronger in cigarette smokers than in nonsmokers. This early case-control analysis of the ARK Study demonstrates a significant association between plasma fbrinogen concentration and early atherosclerosis in the carotid arteries. In the context of published findings from popdation studies, our results indicate that plasma fbrinogen concentrations may be a useful marker for identifying individuals at high risk of developing arterial thrombotic disorders. Ann Epidemiol J 992;2:47J -480. KEY WORDS:

C, antithrombin

Fibrinogen, III, fator

coagulation factors, hemostasis, ARJC Study, factor VII, protein VIII, von Wilebrand factor.

INTRODUCTION Hemostasis

is a complex

(procoagulants)

physiologic

counterbalanced

process involving a myriad of promoting

by naturally occurring

inhibitors.

factors

Derangement

of

this balance is considered to play an important role in the pathogenesis of thrombosis. Consequently, there has been intense interest in determining whether plasma levels of hemostatic

factors serve as markers for arterial prothrombotic

states. Recent observa-

From the University of Texas Medical School at Houston, Houston, TX (K.K.W., M.G.C.); University of Minnesota School of Public Health, Minneapolis, MN (A.R.F.); U mversity of North Carolina, Chapel Hill, NC (G.H., C.E.D.); and Bo wman Gray-School of Medicine, Winstonkalem, NC (R.B.). _ Address reprint requests to: Kenneth

K. Wu. MD, Universitv

of Texas Medical School at Houston,

643 1

Fannin, Ho&ton, TX 77030. Accepted September 4, 1991. 0 1992 Elsevier Science Publishing Co.. Inc.

1047.2797/92/$05.00

472

wu et al. FIBRINOGEN

AND CAROTID

AEP Vol. 2. No. 4 Jul> 1992: 471-480

ATHEROSCLEROSIS

tions that both plasma fibrinogen level and factor VII coagulant activity are significantly associated

with ischemic

cardiovascular

However, there is little information with early atherosclerosis. This

report presents

events lend support to this notion

regarding the association

associations

between

of hemostatic

several key coagulation

inhibitors with carotid artery atherosclerosis in 385 case-control the cohort in the Atherosclerosis Risk in Communities (ARIC) indicate that the plasma concentration

(l-5). factors

factors and

pairs selected from Study. The results

of fibrinogen is independently

associated with

early carotid atherosclerosis.

METHODS

AND

MATERIALS

The design and methods of the ARIC study were detailed elsewhere (6) and described in this journal by Sharrett

Ultrasound

(7) and Folsom and colleagues (8).

Examination

Carotid artery atherosclerosis

of the Carotid Arteries was determined

by a high-resolution

B-mode ultrasound

imaging method, described in detail in the ARIC Study Procedural Manual 6 (9). Sonographers from each of the four field centers received training at the ARIC Ultrasound Reading Center (Winston-Salem, tion according to a standardized protocol.

NC) to perform the ultrasound examinaThe extracranial carotid arteries were

scanned bilaterally, and one popliteal artery, selected at random, was also scanned. The carotid arteries were divided into three segments: the distal, l.O-cm, straight portion of the common cm of the internal

carotid artery; the carotid bifurcation;

carotid artery. The

and the proximal

1.0~cm length of the distal common

1.0

carotid

artery was defined in relation to the beginning of the dilatation of the carotid artery bifurcation, as the reference point. The carotid bifurcation was defined as the 1.0~cm segment proximal to the tip of the flow divider separating the internal and external carotid arteries, that is, the lowest point of the V-shaped structure. The same reference point was used to define the l.O-cm segment of internal carotid artery. To identify a 1 .O-cm portion of the popliteal artery, the popliteal skin crease was used as a reference point. The B-mode ultrasound data were recorded on a high-resolution and sent to the Ultrasound Reading Center for central measurement

Y+in cassette, of arterial wall

thickness. Readers, who were also trained centrally and certified, measured the wall thicknesses according to the standardized protocol described in the ARIC Study Procedural Manual 6 (9).

Selection

of Cases and Controls

Case patients were selected based on an arterial far-wall intima-media thickness indicative of atherosclerosis; controls subjects, by contrast, constituted a sample of study participants whose arterial far- and near-wall measurements at all examined carotid and popliteal artery sites indicated the absence of intima-media thickening. Case patients had at least two measurements of carotid artery far-wall thickness of more than 2.5 mm, or bilateral thickening corresponding to a maximum intima-media thickness of at least 1.7 mm at the internal carotid, and/or at least 1.8 mm of thickness in the carotid bifurcation, and/or at least 1.6 mm of thickness in the common carotid arteries. These cutpoints exceeded a value approximating the 90th percentile for the respective carotid artery segments in the cohort. Control subjects had no evidence of

Wu et al. FIBRINOGEN AND CAROTID ATHEROSCLEROSIS

AEP Vol. 2, No. 4

July 1992: 471-480

473

atherosclerotic thickening, that is, their maximum far- and near-wall thickness was below a value approximating the 75th percentile of intima-media thickness on all carotid artery segments visualized and the popliteal artery. To reduce the possibility of extraneous differences between case patients and their paired control subjects, both groups met similar criteria of minimum visualization of arterial wall thickness boundaries, defined as the number of measurements

of intima-media

thickness placed by the

ultrasound readers on each arterial site. After selection of candidate cases and controls, those with a history of symptomatic heart disease or stroke were excluded from analysis. Each patient was then matched to a control subject within strata defined by field center, race, sex, and IO-year age groups. For each case, the control subject with the closest date of examination was chosen, first searching the group of controls with the highest degree of arterial wall visualization, that is, the largest number of arterial wall thickness measurements. If no control was available in this group, the matching algorithm proceeded in sequence through up to five groups of decreasing completeness of arterial wall visualization. Seventy-five percent of the case-control pairs were matched in the group with the highest degree of arterial wall visualization. If more than one control was eligible within any stratum, one was chosen at random.

Assay of Hemostatic

Factors

Plasma fibrinogen,

factor VII, factor VIII,

and antithrombin

III (AT-III)

von Willebrand

factor (vWF),

were measured in each participant.

protein C,

Subjects

were

asked to fast for 12 hours prior to examination. The blood collection and processing procedures were described previously (10). For hemostatic determinations, blood was drawn from an antecubital vein and collected into tubes containing r/lo vol. of 3.8% sodium citrate. All samples were centrifuged at 3000 g for 10 minutes at 4°C. The titrated plasma samples were divided into aliquots at the field center laboratory, temporarily stored at - 70°C and then shipped to the Central Hemostasis Laboratory in Houston within 1 week after blood collection. On arrival at the hemostasis laboratory, samples were inspected and stored at - 70°C until assay, within 2 weeks of blood collection. Plasma fibrinogen level was measured by a thrombin time titration method described by Clauss (11). Reagents for the fibrinogen assay were obtained from General Diagnostic (Organon-Teknika, Morris Plains, NJ). Factor VII and VIII coagulant activities were measured by determining the ability of the testing sample to correct the clotting time of human factor VII- or VIII-deficient plasma obtained from George King Biomedical (Overland Parks, KS). The plasma factor VII and VIII levels were determined by relating the clotting time to a calibration curve, constructed for each batch of samples using a lyophilized reference plasma obtained from Pacific Hemostasis (Curtin Matheson, Houston, TX). vWF was measured by an enzyme-linked immunosorbent assay (ELISA) from American Bioproducts (Parsippany, NJ). Protein C antigen level was also measured by ELISA using a kit from American Diagnostica (Greenwich, CT). AT-III activity was measured by a chromogenic substrate method. Strict internal and external quality controls were instituted to ensure the accuracy and precision of these assays. The intra-assay and interassay coefficients of variation (CV) of all the assays were below 5%, except for the AT-III assay which had a CV of approximately 10%.

Other Measurements Plasma cholesterol level was measured by an enzymatic assay at the Central Lipid Laboratory at Baylor College of Medicine (12). Body mass index (kg/m’) was computed

474

wu et al. FIBRINOGEN

AND CAROTID

AEP Vol. 2, No. 4 .luly 1992: 471-480

ATHEROSCLEROSIS

TABLE 1 Differences in hemostatic factors between cases and controls, aged 45 to 64 years

Variables

No. of pairs

Fibrinogen (mg/dL) Factor VII (%) Factor VIII (%) vWF (%) Protein C (yg/mL) AT-III (%)

385 385 385 385 385 385

Cases mean (SD) 316 118 127 120 3.20 112

(73) (28) (34) (52) (0.62) (21)

Matched controls mean (SD) 287 115 124 115 3.10 108

(56) (25) (33) (43) (0.59) (20)

Mean difference

95% Confidence intervals of difference

28.8 3.6 3.2 4.5 0.09 4.0

(19, 38) (0.2, 6.9) (-1.2, 7.7) (- 1.7, 10.7) (0.012, 0.17) (1.1, 6.9)

from height and weight taken while the patient was without shoes and in light clothing. Cigarette

smoking status was assessed by a questionnaire.

Seated blood pressure was

measured after a 5-minute rest, by trained technicians using random-zero sphygmomanometers. Three successive measurements of systolic and phase 5 diastolic pressures were taken,

and the average of the last two was used for analysis.

Statistical

Methods

Three hundred eighty-five cases and 385 matched controls were selected for this study from the ARIC participants examined between May 15, 1987, and May 14, 1989. Differences in mean values of hemostatic factors were compared between case patients and control subjects, and 95% confidence intervals for the differences were calculated. The data were then cross-tabulated plus controls,

and marginal

by quartiles of the hemostatic

homogeneity

variable in the cases

was tested by Cochran’s

test (13).

The

conditional logistic regression model (14) was used for univariate and multivariate analyses. Odds ratios for a one standard deviation interval were computed from the regression coefficients.

Statistical

Analysis Systems (SAS)

(15) was used for computa-

tions. To test the assumption that the log odds ratios were linearly related to a hemostatic variable, a conditional logistic regression with a restricted cubic spline term (16) was fit and compared with the corresponding conditional logistic regression containing only a linear term by a likelihood ratio test. None of the associations showed significant

deviation

from linearity.

RESULTS Mean ages were 55.5 years for controls and 56.8 years for cases. There were 231 male and 154 female pairs. Table 1 shows the differences between the cases and controls in the mean level of each of the six coagulation proteins. The mean fibrinogen level was 28.8 mg/dL higher in the cases than in the controls. The mean factor VII, protein C, and AT-III levels were also higher in cases than in controls. The 95% confidence intervals for these differences excluded zero. On the other hand, mean levels of factor VIII and vWF did not exhibit a significant difference between cases and controls. Table 2 shows the distribution of case-control pairs according to quartiles of fibrinogen. Case patients tended to cluster in the upper quartiles and control subjects in the lower quartiles (P < 0.001). A similar analysis for factor VII, protein C, and AT-III levels (not shown) also revealed an upward shift of distributions in cases compared to controls, P values being 0.01,

0.02, and 0.04 for factor VII, protein C, and AT-III,

respectively.

AEP Vol. 2, No. 4 July 1992: 471-480

wu et al. FIBRINOGEN AND CAROTID ATHEROSCLEROSIS

TABLE 2 Cross-tabulation concentration

of matched

cases and controls

Quartile

by quartiles”

of fibrinogen

of cases

Quartile of control

I

II

III

IV

I 11 III IV Total (%)

20 20 19 11 70 (18)

27 18 19 22 86 (22)

29 42 20 14 105 (27)

42 31 30 21 124 (32)

a Quartdes ranges

were:

were

based

I, 142-256

b X’ (Cochran’s

test)

on the

distribution

mg/dL; =

33.0

II. (3 df).

257-292 P
0.09). We further evaluated the influence of cigarette smoking and other risk factors on the association of fibrinogen with carotid atherosclerosis, using logistic regression models. Adjusting for cigarette smoking alone reduced the univariate regression coefficient for fibrinogen from 0.0067 ( see Table 3) to 0.0050 (Table 5). As shown in Table 4, adjustment also for cholesterol, blood pressure, and body mass index dropped it to 0.0036. These data suggest that the fibrinogen association with carotid atherosclerosis

Univariate conditional logistic regression TABLE 3 hemostatic factors with carotid atherosclerosis Standard deviation

Variable

66.85 26.53 33.79 47.83 0.61 20.86

Fibrinogen (mg/dL) Factor VII (%) Factor VIII (%) vWF (%) Protein C (pg/mL) AT-III (%) a The deviation

odds

ratio

increase

represents in the

Regression coefficient

the

hemostatic

predicted factor.

analysis of the association

Standard error of coefficient

P value

Odds ratio’

0.0013 0.0031 0.0023 0.0017 0.1275 0.0036

< 0.001 0.038 0.158 0.158 0.031 0.008

1.6 1.2 1.1 1.1 1.2 1.2

0.0067 0.0064 0.0033 0.0024 0.2757 0.0095 mcrease

III the

odds

of

of developing

carotid

atherosclerosis

for a one

standard

476

AEP Vol. 2, No. 4 July 1992: 471-480

wu et al. FIBRINOGEN

AND CAROTID

ATHEROSCLEROSIS

TABLE 4 Multivariate conditional logistic regression analysis of the association fibrinogen and other risk factors with carotid atherosclerosis Standard deviation

Variable Fibrmogen (mg/dL) Plasma cholesterol(mgidL) Blood pressure (mm Hg) Body mass m&x (kg/m’) Current smoker (yes/no) Exsmoker (yes/no)

Regression coefficient

66.85 41.77 18.58 4.38 0.46 0.48

of

Standard error of coefficient

P value

Odds ratio’

0.0014 0.0023 0.0060 0.0216 0.2417 0.2196

0.010 < 0.001 < 0.001 0.043 < 0.001 0.009

1.3 1.5 1.9 1.2 2.0 1.3

0.0036 0.0100 0.0356 0.0437 1.5190 0.5703

u The odds raw represents the predlcted mcrease ,n the odds of developing carotld atherosclerosis for a one standard dewanon

mcrease m the hemostatic

was partly confounding

factor.

independent, partly the result of cigarette smoking, and partly due by other risk factors. We sought to further clarify the independence

the fibrinogen

effect by examining

nonsmokers, 0.0019 with

the nonsmoking

the univariate coefficient adjustment for cholesterol,

case-control

pairs.

to of

In 184 pairs of

for fibrinogen was 0.0046, and was reduced to blood pressure, and body mass index (Table

6). Although total sample

an interaction appeared likely, because the regression coefficient for the in Table 4 (0.0036) was almost twice that for nonsmokers (0.0019), a formal test of interaction by the log likelihood ratio test showed the finding to not be statistically significant. The relation ated further shows

that

smokers,

between

by looking

in pairs where the

average

both

fibrinogen

When the case patient difference was amplified a smoker

cigarette

smoking,

at pairs concordant

fibrinogen,

case patients level

and the case patient

was about

a nonsmoker,

on smoking

and control

subjects

18 mg/dL

was a smoker and the control to about 53 mg/dL. In contrast,

mg/dL higher in the controls. subjects were current smokers,

and atherosclerosis

or discordant

the average

were

higher

Table

7

not current

in case patients.

was a nonsmoker, when the control fibrinogen

was evalu-

status.

the mean subject was

10

level was about

In the 27 pairs where both case patients and control the difference in mean fibrinogen levels was quite high

(5 1 mg/dL); this appears to be partly explained by a higher daily cigarette consumption (7 cigarettes) in case patients (see footnote to Table 7). Similar results were obtained when

we analyzed

shown). Elevation of the effect

smoking

of leukocyte of cigarette

status

as never-smokers

counts has been hypothesized smoking on atherosclerosis.

influence of leukocyte count on the fibrinogen was significantly associated with atherosclerosis

TABLE 5 Conditional association of fibrinogen Variable Fibrinogen (mg/dL) Current smoker (yes/no) Exsmoker (yes/no)

versus

ever-smokers

0.0050 1.2043 0.5195

not

to be an important mediator We, therefore, analyzed the

association. Although leukocyte count in univariate analysis, its association

logistic regression analysis of the effect of cigarette with carotid atherosclerosis Regression coefficient

(data

smoking on the

Standard error of coefficient

P value

0.0013 0.2073 0.1916

< 0.001 < 0.001 0.006

477

wu et al. FIBRINOGEN AND CAROTID ATHEROSCLEROSIS

AEP Vol. 2, No. 4 July 1992: 471-480

TABLE 6 Conditional logistic regression analysis of the association carotid atherosclerosis in 184 nonsmoking pairs of cases and controls Regression coefficient

Model Univariate Fibrinogen (mg/dL) Multivariate Fibrinogen (mg/dL) Cholesterol (mg/dL) Blood pressure (mm Hg) Body mass index (kg/m’)

was not

significant

between

fibrinogen

with

Standard error of coefficient

0.0046

0.0018

0.0019 0.0106 0.0392 0.0496

0.0020 0.0031 0.0083 0.0302

in multivariate and carotid

of fibrinogen

analysis

and

atherosclerosis

did not

(data

P value 0.009 0.33 < 0.001 < 0.001 0.100

influence

the

association

not shown).

DISCUSSION In this study, we assessed the association of several key coagulation proteins thickening of the extracranial carotid arteries as measured by a high-resolution ultrasound

technique.

Previous

pathologic

correlational

studies

validated

with wall B-mode the use of

B-mode ultrasound as a sensitive method for detecting early atherosclerotic lesions (17). The ultrasound technique for measuring extracranial carotid arterial wall thickening was shown to be reproducible, as evaluated by several stringent, quality-control programs implemented by the ARIC Study (results to be reported elsewhere). Highresolution B-mode sonography is, hence, a reliable and sensitive technique for quantifying early atherosclerosis. The demonstration of a strong association between several well-known cardiovascular risk factors and carotid wall thickening further supports the validity

of this procedure.

and colleagues Hemostatic

In this respect,

( 18). mechanisms

play a critical

our findings

are similar

role in thrombus

to those

formation

of Salonen

on atheroscle-

rotic arterial surfaces, and may cause acute arterial occlusion, leading to myocardial infarction and thrombotic stroke. Prospective population studies from the United Kingdom,

United

States,

and Sweden

provided

solid evidence

indicating

a significant

association between plasma fibrinogen levels and acute vascular events such as myocardial infarction and stroke (l-5). As fibrinogen is the substrate for fibrin formation and

TABLE 7 Analysis of the relationship and controls”

between

cigarette smoking and fibrinogen

Case smoking status

Control smoking status

No. of pairs

N S N S

N N S S

184 131 43 27

’ The mean - 19; s-s ’ Mean N

=

=

differences

in cigarette

consumption

7.

difference noncurrent

in fibrinogen smoker;

(mg/dL).

S = current

smoker.

(no.

of agarettesid)

Mean casecontrol differenceb

level in cases 95% Confidence intervals for difference

17.5 52.7 -9.9 51.0 between

cases and controls

(4.8, 30.2) (37.1, 68.3) (-38.5, 18.7) (2.2, 99.8) in each

category

were:

S-N

=

25; N-S

=

478

AEP Vol. 2, No. 4 July 1992: 471-480

wu et al. FIBRINOGEN AND CAROTID ATHEROSCLEROSIS

is a mediator of platelet aggregate formation (19), this association has been interpreted to indicate the contribution of high fibrinogen levels to prothrombotic states. The finding

in the ARIC

fibrinogen

Study

and the study of Salonen

and colleagues

(18) that

a high

level also is associated

with early atherosclerosis is, hence, intriguing. The and 1.3 (multivariately) for a one relative odds of atherosclerosis of 1.6 ( univariately) increase in fibrinogen are comparable to relative risks standard deviation (67-mg/dL) reported by others for ischemic cardiovascular events (I-5) and for atherosclerosis

measured by ultrasound (18). It remains to be determined whether this association is directly related to the biologic activity of fibrinogen on vessel wall and platelets (19). Recent

studies

suggested

that

cigarette

smoking

and plasma

fibrinogen

are inti-

mately related. Meade and coworkers noted that plasma fibrinogen levels are elevated by cigarette smoking (20). Kannel and associates further suggested that the effect of cigarette through factors,

smoking fibrinogen

on acute atherosclerotic cardiovascular events may be mediated (21). We also found a close relationship between these two risk

but this relation

appears

complex.

Both carried

some independent

risk for early

atherosclerosis and yet part of the effects were interrelated. In fact, the association of fibrinogen with carotid atherosclerosis was enhanced by concomitant cigarette smoking, although a formal statistical test for interaction proved this to be nonsignificant. This nevertheless raises an interesting possibility that cigarette smoking and fibrinogen may have

synergistic

effects

on atherosclerosis

and thrombosis.

The association of factor VII, protein C, and AT-III with carotid atherosclerosis was weaker than that of fibrinogen. In fact, after adjustment for major risk factors, these associations in the Northwick independently

were no longer statistically Park Heart Study (NPHS) associated

with

the

significant. Meade and colleagues showed that factor VII activity was positively and

incidence

of ischemic

heart

disease

(1, 2). The

reason for the discrepancy between this study and the NPHS is unclear but may relate to several differences between the two studies. The study designs and disease end points were quite different. Furthermore, VII activity were also different detecting

the plasma

the technique and, therefore,

level of factor

resolve this important issue. Both AT-III and protein

VII activity

C are important

and reagents used for measuring factor may have different sensitivities for (22).

Further

anticoagulant

studies proteins.

are needed

to

A deficiency

state of either protein is a strong risk factor for deep vein thrombosis and pulmonary embolism (23, 24). Their association with arterial thrombosis is less clear. Our study revealed a positive association of AT-III univariately but not multivariately. The too quickly,

because

of potential

and protein C with carotid atherosclerosis univariate findings should not be dismissed

weaknesses

in the case-control

design,

as discussed

below. Furthermore, the ARIC Study has not yet examined clinical atherosclerotic disease, which may be more related to these anticoagulant proteins than is early carotid atherosclerosis. Several aspects of this case-control analysis warrant consideration. First, as these data are cross-sectional, a cause-effect relation between fibrinogen and atherosclerosis is not established. It is possible that a high fibrinogen level is consequential to atherosclerosis rather than vice versa. However, a strength of the ARIC Study is the investigation of an early manifestation of atherosclerosis. Moreover, exclusion of participants with symptomatic cardiovascular disease makes this study less susceptible than most to such ambiguous time relationships. Second, some atherosclerotic lesions are likely to be present in most middle-aged individuals in the population sampled by the ARIC Study, yet, for practical reasons, we evaluated only the carotid arteries. It is possible that the control subjects had atherosclerotic lesions elsewhere, despite

AEP Vol. 2, No. 4 July 1992: 471-480

wu et al. FIBRINOGEN AND CAROTID ATHEROSCLEROSIS

normal

carotid

arteries.

This

misclassification

error

would

serve

479

to attenuate

the

fibrinogen-atherosclerosis association rather than exaggerate it. Third, the analysis was based on a single measurement of the hemostatic factors. Although studies by Thompson and associates (25) and us (unpublished data, 1992) indicated relative low intraindividual factors, carried

variability of repeated measurements of fibrinogen and some other hemostatic the findings will be strengthened by follow-up measurements, which will be out in the next

phase

of the ARIC

Study.

SUMMARY In summary,

this early case-control

cant association

between

plasma

analysis

of an ARIC

fibrinogen

cohort

concentration

demonstrates

a signifi-

and early atherosclerosis

in

the carotid arteries. This association is influenced greatly by cigarette smoking. In the context of published findings from population studies, our results lead us to conclude

that plasma fibrinogen concentrations play an important role in human atherosclerosis and its clinical disorders and may be a useful marker for identifying high-risk individuals.

The work was supported by National Heart, Lung, and Blood Institute contracts NOl-HC5501522. The authors wish to thank the staff at the ARIC field centers, lipid and hemostasis laboratories, and Ultrasound Reading Center for their excellent contributions. We wish to thank particularly Sonia Davis, Yuan Li Shen, and Ding Yi Zhao at the ARIC Coordinating Center (University of North Carolina, Chapel Hill, NC) for programming support; Andrea Finch, Valarie Stinson, and Charlene Tanner at the ARIC Central Hemostasis Laboratory (University of Texas at Houston, TX) for performing laboratory assays; and the following field center technicians for sample preparation: Elsie Bacon, Karen Barr, Carol Christman, Lisa Field, Amy Haire, Sharada Iyer, Bryna Lester, Stella Loehr, Barbara Mariotti, Catherine McCormick, Gail Murton, Joan Nelling, Virginia Overman, Delilah Posey, Cathy Rachui, Sue Ware, Shirley Willis, Virginia Wyum. The critical reviews by the Publication and Steering Committees of ARIC are highly appreciated.

REFERENCES 1. Meade TW, North WRS, Chakrabarti RR, et al. Haemostatic function and cardiovascular death: Early results of a prospective study, Lancet. 1980;1:1050-4. 2. Meade TW, Brozovic M, Chakrabarti RR, et al. Haemostatic function and ischemic heart disease: 3.

Principal

Wilhelmsen

stroke and myocardial 4.

Stone

Gen Pratt. 5.

results of the Northwick

MC,

infarction, Thorp

JM.

1986;2:533-537. as a risk factor for

1984;311:501-6.

Plasma fibrinogen-A

Kannel WB, Wolf PA, Castelli WP, D’Agostino ARIC

Investigators.

Study, JAMA.

major coronary

risk factor,

J R Co11

RB. Fibrinogen

and risk of cardiovas-

Am J Epidemiol.

7.

AR, ARIC

Sharrett Introduction

and

1987;258:1183-6.

The Atherosclerosis

and objectives, Study.

N Engl J Med.

Lancet.

K, et al. Fibrinogen

1985;35:565-9.

cular disease: The Framingham 6.

Park Heart Study,

L, Svardsudd K, Korsan-Bengtsen

Risk in Communities

(ARIC)

Study: Design

1989;129:687-702.

Investigators. objectives

of

The Atherosclerosis the

hemostatic

Risk in Communities component,

Ann

(ARIC)

Epidemiol.

1992;2:00-00. 8. variables 9.

Folsom

AR,

Conlan

and cardiovascular National

Heart,

MC,

Davis CE,

and Wu KK. Relations

risk factors in middle-aged Lung,

and Blood

Institute.

between

adults, Ann Epidemiol. Atherosclerosis

hemostasis

1992;2:00-00.

Risk in Communities

480

wu et al.

AEP Vol. 2, No. 4 July 1992: 471-480

FIBRINOGEN AND CAROTID ATHEROSCLEROSIS

(ARIC) Study

Procedural

Coordinating 10. of blood

Manual

Hill,

NC: ARIC

University

of North

Carolina;

1987.

H, Bracey A, Wu KK. ARIC

Hemostasis

Study-I.

School

Papp AC,

6: Ultrasound

of Public

Hatzakis

collection

Haemost.

and processing

Examinations.

Health,

systems

suitable

Chapel

for multicenter

Study

Development

hemostatic

studies,

Thromb

1989;61:15-9.

11. gens,

Center,

Acta 12.

Clauss

A. Gennungsphysiologische

Haematol. Siedel

determination

Schnell

Methode

Zur Bestimmung

des Fibrino-

1957;17:237-46.

J, Hagele of serum

EO, total

Ziegenhoun

J, Wahlefeld

cholesterol

with

AW.

improved

Reagent

lipolytic

for the

efficiency,

enzymatic

Clin

Chem.

1983;29:1075-80. MA:

13.

Bishop

YMM,

MIT

Press;

1975.

14.

Breslow

of Case-Control SAS

16.

Harrell

17.

User’s

Salonen

19.

Doolittle Meade

21. 22. risk,

Cary,

SAS

vol.

for Research

Institute; models

1. The Analysis 1980.

1985.

Inst.

PJ. Noninvasive

Cambridge,

of Cancer;

in clinical

J Nat1 Cancer

JA, Curry

Analysis.

Research.

Agency

BG. Regression

K, Rauramaa

levels Imeson

studies:

Determining

1988;80:1198-202.

assessment

KG.

D’Agostino Insights

Annu

of atherosclerosis,

Factor

heart

RB, Belanger

VII assays,

1989;9:783-4.

23.

RD,

Rosenberg

plasma

JS. Natural

1984;53:195-229. and other

Lancet.

1987;2:986-8.

AJ. Fibrinogen,

cigarette

Am Heart

triglyceride

levels

anticoagulant

atherosclero-

in smoking

disease, Study,

of carotid

1988;8:788-92.

Rev Biochem.

Y. Effects of changes

from the Framingham

Atherosclerosis. Rosenberg

and fibrin,

J, Stirling

JT. Prevalence

Atherosclerosis.

and the risk of ischemic

WB,

disease:

R, Salonen

in east Finland,

RF. Fibrinogen

factors

Mann

Rooney

in Cancer

NC:

and responses,

R, Seppanen

TW,

Kannel

Methods

Multivariate

1984;26:295-307.

cholesterol

20.

cardiovascular

DH,

Dis.

PW. Discrete

International

Statistics.

predictors

Blankenhom

tics on clotting

France:

FE Jr, Lee KL, Pollock

Prog Cardiovasc 18.

Lyon,

Guide:

between

sis and serum

SE, Holland

NE, Day NE. Statistical Studies.

15. relationships

Feinberg

characteris-

smoking

and risk of

J. 1987;113:1006-10.

and cardiovascular mechanisms,

J Clin

disease Invest.

1984;74:1-6. 24.

Esmon

CT.

Protein

C: The

regulation

of natural

anticoagulant

Source

of variability

pathways,

Science.

1987;235:1348-52. 25. assays,

Thompson Thromb

SO,

Haemost.

Martin 1987;58:

JC,

Meade

1073-7.

TW.

in coagulation

factor