You are on page 1of 6

Atherosclerosis 203 (2009) 604–609

Contents lists available at ScienceDirect

journal homepage:

Metabolic syndrome, haemostatic and inflammatory markers, cerebrovascular

and peripheral arterial disease: The Edinburgh Artery Study
Sarah H. Wild a,∗ , Christopher D. Byrne b , Ioanna Tzoulaki c , Amanda J. Lee d ,
Ann Rumley e , Gordon D.O. Lowe e , F. Gerald R. Fowkes a
Public Health Sciences, University of Edinburgh, Teviot Place, Edinburgh EH8 9AG, United Kingdom
The Institute of Developmental Sciences, University of Southampton, Southampton General Hospital, Southampton SO16 6YD, United Kingdom
Division of Epidemiology, Public Health and Primary Care, Imperial College, St. Mary’s Campus, London W2 1PG, United Kingdom
Department of General Practice and Primary Care, Foresterhill Health Centre, Westburn Road, Aberdeen AB25 2AY, United Kingdom
Division of Cardiovascular and Medical Sciences, University of Glasgow, Royal Infirmary, Glasgow G31 2ER, United Kingdom

a r t i c l e i n f o a b s t r a c t

Article history: The role of metabolic syndrome and associated haemostatic and inflammatory markers in risk of
Received 14 December 2007 atherosclerosis in different vascular beds is controversial. We used modified National Cholesterol Educa-
Received in revised form 10 July 2008 tion Program criteria to define metabolic syndrome in a population-based cohort of men and women aged
Accepted 25 July 2008
55–74 years with up to 15 years of follow-up to investigate whether metabolic syndrome is associated
Available online 3 August 2008
with risk of cerebrovascular and peripheral arterial disease and the role of inflammatory and haemostatic
factors in these relationships. Data were available for 762 participants, of whom 267 (35%) had metabolic
syndrome at baseline and 69 (9.0%) and 108 (14%) had cerebrovascular and peripheral arterial disease
Metabolic syndrome
Cerebrovascular disease
events, respectively, during follow-up. We used Cox proportional hazards modelling to estimate hazard
Peripheral arterial disease ratios (HRs). Metabolic syndrome was associated with several haemostatic and inflammatory variables
Haemostatic factors and with cerebrovascular disease both after adjusting for age and sex (HR 2.12 (1.31–3.41) and after further
Inflammation adjustment for conventional cardiovascular risk factors and inflammatory and haemostatic markers (HR
1.77 (1.05–2.96). The association between metabolic syndrome and peripheral arterial disease was not
statistically significant either with adjustment for age and sex (HR 1.33 (0.90–1.96) or after full adjustment
(HR 0.89 (0.57–1.38).
We conclude that metabolic syndrome was more strongly related to risk of atherosclerosis in the cere-
brovascular than the peripheral circulation and the association was independent of conventional risk
factors, haemostatic and inflammatory markers in this population. Improving insulin sensitivity may
reduce cerebrovascular disease risk.
© 2008 Elsevier Ireland Ltd. All rights reserved.

1. Introduction have shown that increased risk of cardiovascular disease mortal-

ity associated with the metabolic syndrome is independent of age,
The role of the metabolic syndrome, a clustering of three or more sex, LDL-cholesterol levels and smoking. We have previously shown
of central obesity, dyslipidemia, hypertension and hyperglycemia, a statistically significant association between metabolic syndrome
as a risk factor for cardiovascular disease has been widely debated and cardiovascular disease in Edinburgh Artery Study participants
and several definitions exist [1,2]. The pragmatic definition pro- (hazard ratio 1.20, 95% CI 1.00–1.43 after adjusting for age, sex, base-
posed by the third National Cholesterol Education Program Adult line cardiovascular disease, diabetes, LDL-cholesterol, smoking and
Treatment Panel (NCEP-ATPIII) [3] has been widely used in epi- low ankle-brachial pressure index) [7].
demiological studies of cardiovascular disease and diabetes. The The relevance of the metabolic syndrome to coronary artery,
majority of published studies and meta-analyses suggest that pres- cerebrovascular and peripheral arterial disease (PAD) may vary.
ence of the metabolic syndrome increases the risk of cardiovascular Relatively few studies have investigated the relationship between
disease [4–6]. In some studies this relationship has been con- metabolic syndrome and cerebrovascular or peripheral arterial dis-
founded by conventional cardiovascular risk factors while others ease compared to the number of studies of the association between
metabolic syndrome and coronary heart disease or combined car-
diovascular disease outcomes. Most prospective studies suggest
∗ Corresponding author. Tel.: +44 131 651 1630; fax: +44 131 650 6909. that metabolic syndrome is associated with a 50–100% higher risk
E-mail address: (S.H. Wild). of stroke even after adjusting for conventional risk factors [8–12]. A

0021-9150/$ – see front matter © 2008 Elsevier Ireland Ltd. All rights reserved.

Downloaded for Mahasiswa KOAS UKDW ( at Universitas Kristen Duta Wacana from by Elsevier on January 05, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
S.H. Wild et al. / Atherosclerosis 203 (2009) 604–609 605

Dutch study of people with familial hypercholesterolemia reported of death for fatal events or for approximately 15 years until the end
that the risk of peripheral arterial disease among people with of April 2003 for people who did not have a cardiovascular event.
the metabolic syndrome was almost double that of people with- The study was approved by the Lothian Health Board Ethics Com-
out the metabolic syndrome after adjusting for sex, smoking, LDL mittee and informed consent was obtained from each participant.
cholesterol and statin therapy [13]. However, age and prevalence Participants completed a self-administered questionnaire at
of diabetes were also statistically significantly higher among the baseline which included the World Health Organization (WHO)
group with metabolic syndrome in this study and are major poten- angina and intermittent claudication questionnaires [18]. Height
tial confounding factors. and weight were measured using standard methods. Participants
All of the above studies did not investigate the role of haemo- were asked to indicate whether they had received a diagnosis of
static or inflammatory markers in the relationship between diabetes from a doctor. Systolic blood pressure (SBP) was measured
metabolic syndrome and stroke or peripheral vascular disease. once in the right arm in the supine position after 10 min rest using a
The link between metabolic syndrome and a pro-thrombotic and random zero sphygmomanometer. Ankle SBP at the posterior tibial
pro-inflammatory phenotype is well-established and this pheno- artery was measured once in each leg using a random zero sphyg-
type could potentially mediate the increased risk of cardiovascular momanometer and a Doppler probe. The ankle-brachial index (ABI)
disease associated with the metabolic syndrome [14]. A Swedish for each leg was calculated by dividing the ankle SBP by the brachial
cohort study of people without diabetes reported a hazard ratio SBP and the lower value obtained for the two legs was used as the
(95% CI) of 1.55 (1.20–2.00) for NCEP-ATPIII defined metabolic syn- measure of disease severity in all subsequent analysis. Low ABI was
drome for fatal and non-fatal cardiovascular events after adjusting defined as ABI ≤ 0.9. Participants with an ABI of >1.4 (n = 14) were
for conventional risk factors and attenuation of the hazard ratio excluded because of possible erroneously high levels due to arterial
to 1.45 (1.12–1.88) after further adjustment for C-reactive pro- stiffness.
tein (CRP) levels [15]. This finding provides some support that the Glucose, total cholesterol, high-density lipoprotein (HDL)
effect of metabolic syndrome may be at least partially mediated cholesterol and triglycerides were measured in a fasting blood sam-
by this single measure of inflammation. There has been consider- ple using a Cobas Bio analyser (Roche Products, Welwyn Garden
able debate about whether the presence of the metabolic syndrome City, UK) and standard kits. Participants who were not known to
confers an additional risk beyond the sum of its core components have diabetes received an oral glucose tolerance test (OGTT) in the
of central obesity, dyslipidemia, hyperglycemia and hypertension. form of 75 g of glucose in 335 ml of Solripe Gluctoza Health Drink
This relationship which is difficult to establish in small or mod- (Strathmore Mineral Water Co., Angus, Forfar, UK) and a further
erate sized datasets, may vary between populations and may be blood specimen was collected 2 h later for measurement of glu-
influenced by associations with other components of metabolic cose using the same method as for the fasting sample. CRP was
syndrome including haemostatic and inflammatory markers. measured immunologically using a high-sensitivity assay in a BN
It remains unclear whether metabolic syndrome has a different ProSpec nephelometer (Dade Behring, Milton Keynes, United King-
impact on risk of developing atherosclerotic disease in different dom). Fibrinogen was measured in citrated plasma by means of a
parts of the circulatory system. The aim of this study was to thrombin-clotting turbidimetric method in a centrifugal analyzer
examine the relationships between metabolic syndrome and both [19]. Plasma levels of von Willebrand factor (vWF), tissue plas-
cerebrovascular disease and PAD in men and women followed minogen activator (t-PA), interleukin (IL)-6, intracellular adhesion
up over 15 years. We have also investigated the role of mark- molecule (ICAM)-1, vascular adhesion molecule (VCAM)-1 and E-
ers of hemostasis and inflammation in the relationship between selectin, were measured using high sensitivity ELISA kits (Dako,
metabolic syndrome and cerebrovascular disease or peripheral Milton Keynes, UK (vWF); Biopool, Umea, Sweden (t-PA); R&D Sys-
arterial disease. tems, Oxford, UK (others)) as previously described [20,21].
Prevalent cardiovascular disease (CVD) at baseline was defined
2. Methods as one or more of: stroke (recall of a doctor’s diagnosis), PAD (WHO
intermittent claudication questionnaire evidence and/or ABI < 0.9),
The Edinburgh Artery Study recruited 1592 people (809 men and myocardial infarction (two out of three of recall of a doctor’s diag-
783 women) of 55–74 years of age in 1988 from a general north- nosis, evidence from responses to the WHO questionnaire, and
ern European white population sample. Full details of recruitment, electrocardiographic evidence). Diabetes at baseline was defined
data collection at baseline in 1988/1989, methods of follow-up and using recall of a doctor’s diagnosis or WHO 1999 criteria (fasting
definitions of outcomes have been described previously [16,17]. A glucose of >7 mmol/l or 2 h glucose >11.1 mmol/l) [22] applied to
population-based sample was generated by random identification data collected during the OGTT. Smoking status was defined as
of eligible individuals in 5 year age bands from 11 general practices self-report of current smoking habit at baseline.
across the city of Edinburgh that provide primary care for popula- The most recent NCEP-ATP III criteria for the metabolic syn-
tions of varying socio-economic status. The response rate was 65%. drome are the presence of three or more of the following:
Comparisons with a random sample of 20% of the non-responders waist circumference ≥102 cm (men), ≥88 cm (women); blood
did not detect substantial bias and suggested that the sample was pressure ≥130/85 mm Hg or treatment for hypertension; triglyc-
representative of the general population [16]. Baseline data were erides ≥1.7 mmol/l (≥150 mg/dl); HDL-cholesterol <1.03 mmol/l
collected during participants’ attendance at a research clinic using (<40 mg/dl) (men), <1.29 mmol/l (<50 mg/dl) (women); fasting
a combination of questionnaire and examination data. plasma glucose ≥5.6 mmol/l (≥100 mg/dl) [3]. Waist circumference
Information on non-fatal cardiovascular events (including data were not available in the Edinburgh Artery Study and therefore
stroke, transient ischemic attack, intermittent claudication and the waist circumference criteria used in the above definition were
revascularisation procedures) was obtained from primary care replaced with cut-points for body mass index (BMI) of 28.8 kg/m2
physicians, hospital data and from participants responses to an for men and 26.7 kg/m2 for women, as used in previous studies
annual questionnaire. Complete mortality follow-up was achieved [23,24]. The cut-point for men was identified as being equivalent
by linkage to death data from the National Health Service Central to a waist measurement of 102 cm in a regression analysis of a
Registry in addition to notification of deaths by general prac- cross-sectional study [23]. The criterion for women was devel-
titioners and family members. For this analysis, follow-up was oped from the Women’s Health Study using the value for BMI that
undertaken until the date of first event for non-fatal events or date corresponded to the same percentile cutpoint for a waist circum-

Downloaded for Mahasiswa KOAS UKDW ( at Universitas Kristen Duta Wacana from by Elsevier on January 05, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
606 S.H. Wild et al. / Atherosclerosis 203 (2009) 604–609

Table 1
Demographic characteristics of participants and distribution of conventional cardiovascular risk factors at baseline by absence or presence of metabolic syndrome (see text
for definition)

Variable No metabolic syndrome (N = 498) Metabolic syndrome (N = 264) P-value

Age (mean [S.D.], yr) 64.8 [5.6] 65.4 [5.7] 0.140

Men (%, [numerator]) 49 [246] 48 [126] 0.661
Diabetes (%, [numerator]) 5.4 [27] 18.2 [48] <0.001
History of cardiovascular disease (%, [numerator]) 6.6 [33] 12.1 [32] 0.002
Ankle-brachial index < 0.9 (%, [numerator]) 15 [73] 21 [54] 0.041
Body mass index (mean [S.D.], kg/m2 ) 24.4 [3.2] 28.4 [4.1] <0.001
Smoker at baseline (%, [numerator]) 27 [133] 23 [60] 0.229
Systolic blood pressure (mean [S.D.], mmHg) 140 [24] 155 [23] <0.001
Total cholesterol (mean [S.D.], mmol/l) 7.0 [1.3] 7.2 [1.4] 0.01
LDL cholesterol (mean [S.D.], mmol/l) 5.2 [1.2] 5.6 [1.3] <0.001
HDL cholesterol (mean [S.D.], mmol/l) 1.6 [0.4] 1.2 [0.3] <0.001
Triglyceride (mmol/l) Geometric mean [transformed 95% confidence intervals] 1.20 [1.16–1.24] 2.01 [1.91–2.11] <0.001

P-values refer to t-tests for continuous variables and for chi squared tests for categorical variables.

ference of 88 cm measured at the same time during follow-up [24]. drome variable. The time to event was taken as the time (in years)
No participants were taking lipid-lowering therapy when baseline from the date of entry to the study to the date of the first event of
measurements were performed. relevance for each analysis. Further models examining the associa-
Criteria used to define cardiovascular outcomes were adapted tion between individual components of the metabolic syndrome
from international diagnostic criteria developed by the Ameri- were run. Likelihood ratio tests were used to compare models
can Heart Association [25] and have been described in detail that included and excluded metabolic syndrome, haemostatic and
elsewhere [17]. All possible cardiovascular outcomes were inves- inflammatory factors to determine whether they contributed to the
tigated using medical records and only those that fulfilled the risk of cardiovascular disease outcomes.
protocol criteria for a definite event were included in the anal-
ysis. Outcomes were defined as cerebrovascular disease (stroke 3. Results
or transient ischemic attack, TIA) and peripheral arterial disease
(intermittent claudication or rest pain, ulcer or gangrene or vascu- Of the 1592 original participants in the Edinburgh Artery Study,
lar surgery or amputation). Questionnaire and clinic measurement 762 people with data on all relevant variables were included in
data were checked by clinic staff, coded, and entered onto a DBASE this analysis as a consequence of the large number of variables
IV database. Double data entry was used and discrepant entries used in the analysis and the random nature of missing data for
were checked against original records. different variables. The sub-group for whom all data were avail-
Analysis was performed using Stata software (Stata Corp., Col- able were similar in terms of age and sex distribution, prevalence
lege Station, Texas) for participants with full data on all variables. of the metabolic syndrome at baseline and the proportions devel-
Geometric means and transformed confidence intervals are pre- oping the outcomes of interest to participants excluded from this
sented for triglyceride, CRP, IL-6, ICAM-1, VCAM-1 and E-selectin analysis suggesting that data were missing at random.
levels which were positively skewed. Student’s t-tests and 2 tests Over one third of participants (35%) had metabolic syndrome
were used to test for differences between groups with and without as defined using the modified revised ATP-III criteria as described
the metabolic syndrome for continuous and categorical variables above. Characteristics of groups defined by presence or absence
respectively. Hazard ratios for the outcomes of interest were esti- of the metabolic syndrome by this definition are given for demo-
mated using Cox proportional hazards modelling with adjustment graphic information and conventional cardiovascular disease risk
for age and sex in a basic model and with adjustment for various factors in Table 1 and for haemostatic and inflammatory factors
other cardiovascular disease risk factors in subsequent models. The and outcome variables in Table 2. People with metabolic syndrome
final model of the first set was run again after excluding people with tended to have a more adverse cardiovascular risk factor profile
cardiovascular disease or diabetes at baseline. The first set of mod- than those without metabolic syndrome, other than for sex dis-
els included metabolic syndrome as a binary variable and further tribution, smoking status and von Willebrand factor levels and
models were run using individual components of the syndrome were statistically significantly more likely to have a history of
as continuous variables both with and without the metabolic syn- cardiovascular disease at baseline than those without metabolic

Table 2
Distribution of haemostatic and inflammatory factors at baseline and percentage [number of people] developing cerebrovascular disease or peripheral arterial disease during
follow-up by absence or presence of metabolic syndrome (see text for definition)

Variable No metabolic syndrome (N = 498) Metabolic syndrome (N = 264) P-value

Fibrinogen (mean [S.D.], g/l) 2.64 [0.63] 2.83 [0.71] <0.001

Tissue plasminogen activator (mean [S.D.], ng/ml) 6.93 [3.2] 8.92 [3.1] <0.001
von Willebrand factor (mean [S.D.], IU/dl) 112 [44] 117 [50] 0.162
C-reactive protein (mg/l) geometric mean [transformed 95% confidence intervals] 1.74 [1.58–1.93] 2.59 [2.28–2.94] <0.001
Interleukin-6 (pg/ml) geometric mean [transformed 95% confidence intervals] 2.23 [2.09–2.39] 2.87 [2.52–3.10] <0.001
ICAM-1 (ng/ml) geometric mean [transformed 95% confidence intervals] 215 [210–220] 232 [225–239] <0.001
VCAM-1 (ng/ml) geometric mean [transformed 95% confidence intervals] 376 [369–384] 396 [385–408] 0.002
e-Selectin (ng/ml) geometric mean [transformed 95% confidence intervals] 38 [37–39] 43 [41–45] <0.001
Cerebrovascular disease during follow-up (% [n]) 6.6 [33] 13.3 [35] 0.002
Stroke during follow-up (% [n]) 4.2 [21] 11.0 [29] <0.001
Peripheral arterial disease during follow-up (% [n]) 13.1 [65] 16.3 [43] 0.223

P-values refer to t-tests for continuous variables and for chi squared tests for categorical variables.

Downloaded for Mahasiswa KOAS UKDW ( at Universitas Kristen Duta Wacana from by Elsevier on January 05, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
S.H. Wild et al. / Atherosclerosis 203 (2009) 604–609 607

Table 3
Adjusted hazard ratios (95% confidence intervals) for metabolic syndrome for cerebrovascular disease and peripheral arterial disease

Factors included in models CeVDa (N = 97) PADb (N = 139)

Model 1: age and sex 2.12 (1.31–3.41) 1.33 (0.90–1.96)

Model 2: age, sex, baseline cardiovascular disease, known and newly diagnosed diabetes 2.12 (1.30–3.45) 1.15 (0.77–1.72)
Model 3: model 2 variables + low ABIc , LDLd cholesterol, smoking at baseline 2.10 (1.28–3.45) 1.07 (0.71–1.61)
Model 4: model 3 variables + haemostatic factors (fibrinogen + + t-PA + vWF) 1.90 (1.13–3.20) 0.96 (0.63–1.47)
Model 5: model 3 variables + inflammatory markers (log C-reactive protein + log IL-6 + ICAM1 + VCAM-1 + E-selectin) 1.82 (1.10–3.01) 1.06 (0.69–1.62)
Model 6: model 3 variables + haemostatic factors (as above) + inflammatory markers (as above) 1.77 (1.05–2.96) 0.97 (0.62–1.49)
CeVD, cerebrovascular disease.
PAD, peripheral arterial disease.
ABI, ankle-brachial index.
LDL, low-density lipoprotein.

syndrome. The proportions of people that had a stroke or had any between predictors and outcome may not persist in such a multi-
cerebrovascular event during follow-up were significantly higher variate model. There was no evidence of multicollinearity in either
among people with the metabolic syndrome than among people of the models as the standard errors of the regression coefficients
without the metabolic syndrome at baseline. The proportion of did not dramatically increase as each predictor was added into the
people developing PAD was not statistically significantly different model.
between the two groups.
Table 3 shows hazard ratios (HR) for each outcome derived
from Cox proportional hazards models of increasing complexity. 4. Discussion
There was no evidence of an interaction between sex and metabolic
syndrome (p = 0.5 and 0.3 for fully adjusted models for cerebrovas- We have shown that that metabolic syndrome was associated
cular disease and PAD respectively) and all models were adjusted with an approximate 75% increase in hazard for cerebrovascu-
for age and sex. The metabolic syndrome was a statistically sig- lar disease, even after adjusting for a range of haemostatic and
nificant risk factor for cerebrovascular disease independent of all inflammatory markers as well as conventional cardiovascular risk
cardiovascular risk factors included in the model. Hazard ratios factors. Our findings are consistent with, and extend, the results of
were greater for stroke than for cerebrovascular disease (stroke previous studies of the relationship between metabolic syndrome
and TIA combined)—for example the fully adjusted HR (95% con- and stroke [8–12]. We found no evidence of a significant associa-
fidence interval) for stroke was 2.15 (1.18–3.92) compared to 1.77 tion between metabolic syndrome and PAD. The discrepancy with
(1.05–2.96) for cerebrovascular disease. Metabolic syndrome was the findings of a previous study which reported an association
associated with a non-statistically significantly elevated hazard between metabolic syndrome and peripheral vascular disease may
ratio for PAD in the age and sex adjusted model. The HR for PAD be explained by confounding by age and prevalent diabetes in the
was attenuated after adjustment for conventional and haemostatic previous study [13]. Our findings suggest that the metabolic syn-
factors but the addition of inflammatory factors had little effect. drome phenotype may have a differential impact on atherosclerosis
Inclusion or exclusion of the low ABI variable as a measure of pre- in the cerebrovascular and the peripheral arterial circulations and
clinical PAD had little effect on the association between metabolic that metabolic syndrome is a risk factor for cerebrovascular disease
syndrome and PAD (fully adjusted HR after excluding low ABI from independently of conventional cardiovascular risk factors and the
model 6 0.95 95% CI 0.55–1.64) Likelihood ratio tests provided measured haemostatic and inflammatory factors.
evidence that metabolic syndrome is significantly associated with
cerebrovascular disease in models including both conventional risk
factors (age, sex, diabetes and cardiovascular disease at baseline, Table 4
smoking, LDL cholesterol and low ABI) alone and conventional Hazard ratios (95% confidence intervals) for cerebrovascular disease and peripheral
risk factors together with haemostatic and inflammatory markers arterial disease derived from a model containing age, sex and specified compo-
nents of the metabolic syndrome, conventional risk factors, haemostatic factors and
(p = 0.004 and 0.03 respectively). Limiting the analyses to the 634
markers of inflammation (see text for definitions)
people who were not known to have cardiovascular disease or dia-
betes at baseline gave similar results for the effect of metabolic Risk factor HR (95% CI) for HR (95% CI) for
cerebrovascular peripheral arterial
syndrome on cardiovascular disease in fully adjusted models (HR
disease disease
(1.68 (0.95–2.95)) to the analysis of data for the whole group (HR
1.77 (1.05–2.96)) although the reduced power of the sub-group Systolic blood pressure (mmHg) 1.02 (1.00–1.03) 1.01 (0.99–1.01)
HDL cholesterol (mmol/l) 0.48 (0.20–1.17) 0.77 (0.39–1.50)
analysis meant that the hazard ratio was no longer statistically
Log triglyceride (mmol/l) 2.24 (1.11–4.49) 0.72 (0.39–1.30)
significant at the 5% level. Glucose (mmol/l) 0.93 (0.71–1.22) 1.05 (0.91–1.21)
Hazard ratios for the associations between individual compo- Body mass index (kg/m2 ) 0.98 (0.91–1.04) 0.96 (0.90–1.02)
nents of the metabolic syndrome, conventional risk factors and Smoker 2.00 (1.10–3.65) 1.70 (1.04–2.76)
LDL (mmol/l) 1.00 (0.80–1.24) 1.30 (1.09–1.55)
the two major outcomes of cerebrovascular disease and periph-
ABI < 0.9 1.08 (0.59–1.92) 1.51 (0.94–2.42)
eral arterial disease derived from Cox multiple regression models Baseline CVD 1.22 (1.55–2.66) 1.64 (0.89–3.02)
are shown in Table 4. Of the individual metabolic syndrome com- Diabetes 0.68 (0.22–2.15) 1.71 (0.80–3.66)
ponents, the only statistically significant independent associations Fibrinogen 1.10 (0.70–1.74) 1.31 (0.93–1.84)
were between fasting plasma triglyceride concentration and sys- t-PA (ng/ml) 1.00 (0.92–1.09) 1.06 (0.99–2.13)
vWF (IU/dl) 1.00 (0.99–1.00) 1.00 (0.99–1.00)
tolic blood pressure and cerebrovascular disease. For haemostatic
Log CRP (mg/l) 0.97 (0.71–1.33) 1.15 (0.90-1.47)
and inflammatory variables, the only statistically significant inde- Log IL-6 (pg/ml) 1.43 (1.03–1.98) 0.85 (0.61–1.18)
pendent association was for IL-6 and cerebrovascular disease. As Log ICAM-1 (ng/ml) 0.65 (0.18-2.30) 1.84 (0.69–4.94)
each of the hazard ratios in this table is adjusted for all of the Log VCAM-1 (ng/ml) 1.52 (0.41–5.68) 0.99 (0.32–3.02)
Log E-selectin (ng/ml) 1.91 (0.89–4.06) 0.50 (0.28–0.90)
other variables in the model, some of the univariate relationships

Downloaded for Mahasiswa KOAS UKDW ( at Universitas Kristen Duta Wacana from by Elsevier on January 05, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
608 S.H. Wild et al. / Atherosclerosis 203 (2009) 604–609

Established risk factors for cerebrovascular disease include outcomes. We were not able to differentiate between ischemic and
hypertension, smoking and diabetes, but the role of the metabolic hemorrhagic strokes in this study but the pattern of type of strokes
syndrome as an independent risk factor for stroke is unclear [26]. in the UK [28] suggests that ischemic strokes are likely to predom-
Much of the controversy surrounding the value of the metabolic inate in this population. Other factors such as physical inactivity,
syndrome involves discussion as to whether identification of psychosocial stress and low fruit and vegetable intake may also be
the syndrome adds to risk prediction over and above the sum risk factors for cerebrovascular disease. These factors are also asso-
of the component parts [1]. We have explored the associations ciated with metabolic syndrome but it was not possible to explore
between the individual components of the syndrome and cere- these relationships further in this analysis as data on these factors
brovascular disease (Table 4). Our results show that only plasma were not available in the Edinburgh Artery Study.
triglyceride concentration and systolic blood pressure were inde- In summary, we have found that increased risk of cerebrovas-
pendently associated with cerebrovascular disease. The hazard cular disease is independently associated with the metabolic
ratio for metabolic syndrome in a model including all variables syndrome and the hazard was increased by approximately 75% even
listed in Table 4 in addition to age and sex was 1.11 (0.49–2.50) sug- after adjusting for multiple other factors. The excess risk of cere-
gesting a modest (but non-statistically significant) additional effect brovascular disease associated with the metabolic syndrome could
of metabolic syndrome beyond the components included in the only be partially explained by its association with pro-thrombotic
model. The higher hazard ratio for metabolic syndrome observed and pro-inflammatory factors. We did not find a statistically signif-
in a model containing the individual core components of the icant independent association between metabolic syndrome and
metabolic syndrome but excluding haemostatic and inflammatory PAD. We suggest that interventions that attenuate the metabolic
variables of 1.22 (0.55–2.71) suggests that some of the additional syndrome phenotype such as weight loss (in overweight and obese
effect of the metabolic syndrome is mediated by haemostatic and individuals) and increased physical activity, may have a more
inflammatory factors. Further analysis of larger datasets is required marked effect on reducing risk of cerebrovascular disease than of
to establish whether the presence of metabolic syndrome confers PAD.
additional risk beyond that associated with its individual compo-
nents. Data from the Framingham offspring study suggest that the Acknowledgment
population attributable risk for stroke of metabolic syndrome is
19% [9]. Using a similar approach, we have estimated that the pop- The British Heart Foundation provided financial support for the
ulation attributable risk for stroke with metabolic syndrome in the Edinburgh Artery Study.
Edinburgh Artery Study cohort is 26%.
Other studies have reported that abdominal body fat distribu-
tion and obesity are risk factors for stroke [26]. The explanation for a References
link between obesity and stroke is uncertain but the association of a
[1] Kahn R, Buse J, Ferrannini E, Stern M. The metabolic syndrome: time for
pro-inflammatory phenotype with obesity may be important. Adi- a critical appraisal: joint statement from the American Diabetes Associa-
pose tissue produces IL-6 and both increased body fat and increased tion and the European Association for the Study of Diabetes. Diabetes Care
IL-6 concentrations are well-recognised features of the metabolic 2005;28:2289–304.
[2] Grundy SM. Does the metabolic syndrome exist? Diabetes Care 2006;29:
syndrome [27]. Higher concentrations of all measured inflamma- 1689–92.
tory and haemostatic (except vWF) factors were associated with the [3] Grundy SM, Cleeman JI, Daniels SR, et al. Diagnosis and management of the
metabolic syndrome in univariate analyses of our data (Table 2) and metabolic syndrome. An American Heart Association/National Heart, Lung, and
Blood Institute Scientific Statement. Circulation 2005;112:2735–52.
IL-6 was independently associated with cerebrovascular disease [4] Ford ES. Risks for all-cause mortality, cardiovascular disease, and diabetes asso-
(Table 4). The attenuation of the hazard of the metabolic syn- ciated with the metabolic syndrome: a summary of the evidence. Diabetes Care
drome for cerebrovascular disease after inclusion of inflammatory 2005;28:1769–78.
[5] Galassi A, Reynolds K, He J. Metabolic syndrome and risk of cardiovascular
and haemostatic factors in the regression model suggests that the
disease: a meta-analysis. Am J Med 2006;119:812–9.
presence of a low-grade inflammatory phenotype associated with [6] Gami AS, Witt BJ, Howard DE, et al. Metabolic syndrome and risk of incident
metabolic syndrome contributes to the aetiology of cerebrovascular cardiovascular events and death: a systematic review and meta-analysis of
longitudinal studies. J Am Coll Cardiol 2007;49:403–14.
[7] Wild SH, Byrne CD, Smith FB, Lee AJ, Fowkes FG. Low ankle-brachial pres-
The strengths of our study include the inclusion of both men sure index predicts increased risk of cardiovascular disease independent of
and women, the long follow-up period, the large number of out- the metabolic syndrome and conventional cardiovascular risk factors in the
comes and the availability of data on conventional cardiovascular Edinburgh Artery Study. Diabetes Care 2006;29:637–42.
[8] Koren-Morag N, Goldbourt U, Tanne D. Relation between the metabolic
risk factors, haemostatic and inflammatory factors. Limitations of syndrome and ischemic stroke or transient ischemic attack: a prospective
our study include the use of baseline data which does not allow cohort study in patients with atherosclerotic cardiovascular disease. Stroke
adjustment for regression dilution bias or changes in prevalence 2005;36:1366–71.
[9] Najarian RM, Sullivan LM, Kannel WB, Wilson PW, D’Agostino RB, Wolf PA.
of risk factors during follow-up. As waist circumference measure- Metabolic syndrome compared with type 2 diabetes mellitus as a risk fac-
ments were not available in the Edinburgh Artery Study we used tor for stroke: the Framingham Offspring Study. Arch Int Med 2006;166:
a proxy measure of BMI that has been shown to reflect central 106–11.
[10] Kurl S, Laukkanen JA, Niskanen L, et al. Metabolic syndrome and the risk of
obesity in other populations [23,24], one of which was a cohort stroke in middle-aged men. Stroke 2006;37:806–11.
of similarly aged Scottish men [23]. Unfortunately when the Edin- [11] Chen HJ, Bai CH, Yeh WT, Chiu HC, Pan WH. Influence of metabolic syndrome
burgh Artery Study started the importance of central obesity had and general obesity on the risk of ischemic stroke. Stroke 2006;37:1060–4.
[12] Liu J, Grundy SM, Wang W, et al. Ten-year risk of cardiovascular incidence
not been recognised and so waist circumference was not recorded.
related to diabetes, prediabetes, and the metabolic syndrome. Am Heart J
It is difficult to find data from a similar time point and the dis- 2007;153:552–8.
proportionate increase in waist circumference over time compared [13] Rana JS, Jansen AC, Zwinderman AH, et al. Metabolic syndrome and risk of
coronary, cerebral, and peripheral vascular disease in a large Dutch population
to BMI trends, particularly among women, means that it would be
with familial hypercholesterolemia. Diabetes Care 2006;29:1125–7.
inappropriate to use more recent data for validation of the BMI cut- [14] Devaraj S, Rosenson RS, Jialal I. Metabolic syndrome: an appraisal of the
point as a measure of central obesity in women. Using BMI rather pro-inflammatory and procoagulant status. Endocrinol Metab Clin North Am
than waist as one of the criteria for the metabolic syndrome means 2004;33:431–53, table.
[15] Nilsson PM, Engstrom G, Hedblad B. The metabolic syndrome and incidence
that it is likely that these analyses may underestimate the associa- of cardiovascular disease in non-diabetic subjects—a population-based study
tion between the metabolic syndrome and cardiovascular disease comparing three different definitions. Diabet Med 2007;24:464–72.

Downloaded for Mahasiswa KOAS UKDW ( at Universitas Kristen Duta Wacana from by Elsevier on January 05, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.
S.H. Wild et al. / Atherosclerosis 203 (2009) 604–609 609

[16] Fowkes FG, Housley E, Cawood EH, Macintyre CC, Ruckley CV, Prescott RJ. Edin- [23] Sattar N, Gaw A, Scherbakova O, et al. Metabolic syndrome with and without
burgh Artery Study: prevalence of asymptomatic and symptomatic peripheral C-reactive protein as a predictor of coronary heart disease and diabetes in the
arterial disease in the general population. Int J Epidemiol 1991;20:384–92. West of Scotland Coronary Prevention Study. Circulation 2003;108:414–9.
[17] Leng GC, Lee AJ, Fowkes FG, et al. Incidence, natural history and cardiovascular [24] Ridker PM, Buring JE, Cook NR, Rifai N. C-reactive protein, the metabolic syn-
events in symptomatic and asymptomatic peripheral arterial disease in the drome, and risk of incident cardiovascular events: an 8-year follow-up of 14
general population. Int J Epidemiol 1996;25:1172–81. 719 initially healthy American women. Circulation 2003;107:391–7.
[18] Rose GA, Blackburn H. Cardiovascular survey methods. Monogr Ser World [25] Gillum RF, Fortmann SP, Prineas RJ, Kottke TE. International diagnostic crite-
Health Organ 1968;56:1–188. ria for acute myocardial infarction and acute stroke. Am Heart J 1984;108:
[19] Lowe GD, Wood DA, Douglas JT, et al. Relationships of plasma viscosity, coag- 150–8.
ulation and fibrinolysis to coronary risk factors and angina. Thromb Haemost [26] Goldstein LB, Adams R, Alberts MJ, et al. Primary prevention of ischemic stroke:
1991;65:339–43. a guideline from the American Heart Association/American Stroke Association
[20] Smith FB, Lee AJ, Fowkes FG, Price JF, Rumley A, Lowe GD. Haemostatic factors as Stroke Council: cosponsored by the Atherosclerotic Peripheral Vascular Dis-
predictors of ischemic heart disease and stroke in the Edinburgh Artery Study. ease Interdisciplinary Working Group; Cardiovascular Nursing Council; Clinical
Arterioscler Thromb Vasc Biol 1997;17:3321–5. Cardiology Council; Nutrition, Physical Activity, and Metabolism Council; and
[21] Tzoulaki I, Murray GD, Lee AJ, Rumley A, Lowe GD, Fowkes FG. C-reactive pro- the Quality of Care and Outcomes Research Interdisciplinary Working Group.
tein, interleukin-6, and soluble adhesion molecules as predictors of progressive Circulation 2006;113:e873–923.
peripheral atherosclerosis in the general population: Edinburgh Artery Study. [27] Rutter MK, Meigs JB, Sullivan LM, D’Agostino Sr RB, Wilson PW. C-reactive pro-
Circulation 2005;112:976–83. tein, the metabolic syndrome, and prediction of cardiovascular events in the
[22] World Health Organization Consultation. Definition, diagnosis and classi- Framingham Offspring Study. Circulation 2004;110:380–5.
fication of diabetes mellitus and its complications. part 1: diagnosis and [28] Wolfe CD, Rudd AG, Howard R, et al. Incidence and case fatality rates of stroke
classification of diabetes mellitus. Geneva: World Health Organization; subtypes in a multiethnic population: the South London Stroke Register. J Neu-
1999. rol Neurosurg Psychiatry 2002;72:211–6.

Downloaded for Mahasiswa KOAS UKDW ( at Universitas Kristen Duta Wacana from by Elsevier on January 05, 2018.
For personal use only. No other uses without permission. Copyright ©2018. Elsevier Inc. All rights reserved.