Cardiovascular risk factors and endothelial ... - Clinical Science

Clinical Science (2004) 107, 609–615 (Printed in Great Britain)

Cardiovascular risk factors and endothelial dysfunction Anthony S. WIERZBICKI∗ , Philip J. CHOWIENCZYK†, John R. COCKCROFT†1 , Sally E. BRETT†, Gerald F. WATTS∗2 , B. Stephen JENKINS‡ and James M. RITTER† ∗

Department of Chemical Pathology, King’s College London (King’s, Guy’s & St. Thomas’ Medical School), St. Thomas’ Hospital Campus, Lambeth Palace Road, London SE1 7EH, U.K., †Department of Clinical Pharmacology, King’s College London (King’s, Guy’s & St. Thomas’ Medical School), St. Thomas’ Hospital Campus, Lambeth Palace Road, London SE1 7EH, U.K., and ‡Department of Cardiology, King’s College London (King’s, Guy’s & St. Thomas’ Medical School), St. Thomas’ Hospital Campus, Lambeth Palace Road, London SE1 7EH, U.K.

A

B

S

T

R

A

C

T

Endothelial dysfunction is a feature of atherosclerosis and is associated with CHD (coronary heart disease) risk factors. This study aimed to determine the relationship between the degree of endothelial dysfunction and calculated cardiovascular risk. Endothelial function, as determined by the ACh/NP (acetycholine/sodium nitroprusside response) ratio on brachial plethysmography, was compared with cardiovascular risk as calculated from the Framingham, PROCAM (Prospective Cardiovascular Munster) and MRFIT (Multiple Risk Factor Intervention Trial) algorithms in 246 (187 male) patients, including 44 (22 %) with established CHD. Endothelial dysfunction correlated with the total number of risk factors (r2 = 0.22; P = 0.002) and was related to LDL (low-density lipoprotein)-cholesterol in men and triacylglycerols (triglycerides) in women. The ACh/NP ratio correlated with the occurrence of diabetes, CHD and the LDL-cholesterol concentration (r2 = 0.58; P < 0.001). Endothelial dysfunction was associated with presence of CHD on receiveroperating characteristic plot analysis (area = 0.706 + − 0.04; P = 0.001). There was no correlation between ACh/NP ratio and CHD risk calculated with the Framingham algorithm in men, although both ACh and NP response correlated separately with risk in women. The endothelial ACh/NP ratio correlated with absolute risk in the PROCAM algorithm (r2 = 0.41; P < 0.005). Intermediate results were obtained with MRFIT. Individual risk factors make different contributions to endothelial dysfunction compared with their role in risk calculators. The stronger relationship of endothelial dysfunction with PROCAM risk reflects the contribution of male sex, LDL-cholesterol and triacylglycerols to risk calculated by this algorithm.

INTRODUCTION Endothelial dysfunction is a feature of atherosclerosis [1] and is associated with CHD (coronary heart disease) risk factors, including smoking, hypercholesterolaemia [2], family history of CHD [3], diabetes mellitus [4],

smoking [5] and in some [6,7], but not all, reports with hypertension [8]. Large-scale long-term epidemiological studies of cardiovascular disease have led to the introduction of risk calculation algorithms for the prediction of cardiovascular disease in individuals [9]. These form the basis of many policy statements for the therapeutic

Key words: brachial plethysmography, cardiovascular disease, endothelial dysfunction, epidemiological algorithm, risk factor. Abbreviations: ACh, acetycholine; CHD, coronary heart disease; HDL, high-density lipoprotein; LDL, low-density lipoprotein; MRFIT, Multiple Risk Factor Intervention Trial; NP, sodium nitroprusside; PROCAM, Prospective Cardiovascular Munster; SBP, systolic blood pressure. 1 Present address: Department of Cardiology, University Hospital of Wales, Cardiff CF4 4FW, U.K. 2 Present address: Department of Medicine, University of Western Australia, Perth WA 6001, Australia. Correspondence: Dr Anthony S. Wierzbicki (email [email protected]).

C

2004 The Biochemical Society

609

610

A. S. Wierzbicki and others

management of cardiovascular disease [10]. However, the relationship between the degree of endothelial dysfunction, the magnitude of cardiovascular risk factors and epidemiologically validated algorithms for prediction of CHD risk in a population is unclear [11]. The present study investigated the relationship between endothelial dysfunction and cardiovascular risk factors and also its correlation with calculated cardiovascular risk derived from epidemiological algorithms.

METHODS Data was assembled from a cohort of new patients who had been investigated between 1993–1999. These studies included subgroups from populations with CHD [12], hypercholesterolaemia [2], hypertension [8] and hypertriglyceridaemia [13] and a cohort of normal controls [14]. Factors surveyed included anthropometry, smoking and diabetes status, blood pressure and biochemical risk factor profiles, including lipid subfractions. All investigations had been performed to a consistent protocol with informed consent after ethical approval had been granted. Subjects were studied after an overnight 10-h fast in a warm room (24–26 ◦ C). Aspirin and antihypertensive therapy (n = 48) was discontinued 24 h prior to investigation. Smoking that morning was discouraged. Medications were continued unchanged if used. This applied only to patients with hypertriglyceridaemia who had been treated with fibrate therapy (n = 2) or those with coronary artery disease who received β-blockers and patients with diabetes receiving oral hypoglycaemic therapy. Patients on oral nitrates and statins were excluded from the study. Endothelial function was measured by a standard brachial plethysmography technique in which forearm blood flow responses to brachial artery infusions of ACh (acetylcholine; 7.5–15.0 µg/min) and NP (sodium nitroprusside; 3–10 µg/min) were obtained and could be expressed in absolute terms or as the ACh/NP ratio [15]. Results were referenced by comparison with flow in the other arm. Cardiovascular risk factors were determined by NCEP (National Cholesterol Education Program) criteria [10], and relative (to age- and sex-matched controls) and absolute risks were calculated using the Framingham [9], PROCAM (Prospective Cardiovascular Munster) [16] and MRFIT (Multiple Risk Factor Intervention Trial) ([17], see http://www.norvasc.com/professional/grimm/ grimm.svf) algorithms. Presence of coronary artery disease was defined as 50 % stenosis in one or more coronary arteries on angiography, previous confirmed myocardial infarction or interventional procedure. Blood pressure was measured on three occasions by mercury sphygmomanometry and was averaged. Hypertension was included as categorical variable in the analysis in all patients C


Table 1 Demographic characteristics of patients whose endothelial function was studied and their prospective risks of cardiovascular disease as determined by the Framingham, PROCAM and MRFIT algorithms

DM, diabetes; F, female; LVH, left ventricular hypertrophy; M, male. Risk factor

Framingham

PROCAM

MRFIT

Age (years) Sex CHD (family) DM Smoking Total cholesterol Triacylglycerols HDL LDL SBP LVH

20–75 M and F No Yes Yes Yes No Yes No Yes Yes

35–65 M only Yes Yes Yes No Yes Yes Yes Yes No

30–70 M and F No Yes Number of cigarettes No No Yes Yes Yes No

on prior antihypertensive therapy. Biochemical analyses of lipids were conducted by standard automated methods on a Cobas Fara 2 analyser (Roche, Welwyn Garden City, Herts, U.K.). LDL (low-density lipoprotein)-cholesterol was calculated using the Friedwald equation except when triacylglycerols (triglycerides) exceeded 4.5 mmol/l, where the LDL-cholesterol was measured directly by sequential density ultracentrifugation [13]. Cardiovascular risk was calculated from epidemiological-derived calculators used within their defined limits. Risk factors and accepted inclusion criteria used for risk calculation for each algorithm are shown in Table 1. Statistical analysis was performed using GB Stat 10.0 (Dynamic Microsystems, Silver Spring, MA, U.S.A.). Data are presented as means and S.D. Non-normally distributed data are presented as medians and range and were analysed after log transformation. A P value < 0.05 was considered significant. Forward multiple regression analysis was conducted using all variables and continued until addition of variables resulted in any parameter attaining a P value of 0.10.

RESULTS The clinical characteristics of the cohort of 246 patients are shown in Table 2. The study group included 20 % of patients with CHD, 28 % with hypertension, 19 % smokers and 10 % with Type II diabetes. Endothelial dysfunction correlated with the total number of risk factors (P = 0.002; Figure 1), and individually with presence of CHD, diabetes, smoking, mean log triacylglycerols, but not mean cholesterol. Endothelial dysfunction was associated with the presence of CHD in the present

Cardiovascular risk factors and endothelial dysfunction

Table 2 Comparison of cardiovascular risk factors in the study population compared with the populations recruited for the Framingham and PROCAM studies

Values are means + − S.D. or medians (range). P < 0.05 compared with the other studies. Parameter

Present study

Framingham

PROCAM

Male (%) Age (years) Body mass index (kg/m2 ) Smokers (%) SBP (mmHg) Total cholesterol (mmol/l) Triacylglycerols (mmol/l) HDL (mmol/l) LDL (mmol/l) Framingham risk (% 10 years) Framingham relative risk PROCAM risk (% 8 years) PROCAM relative risk

76 41.8 + − 13.5 25.9 + − 4.04

∗ 49.2 + − 11.8 26.3

40 + − 11 26.4

19∗ 136 + − 23/77 + − 15 6.06 + 1.64 −

32 144/91 5.52

31 134/87 5.90

1.75 (0.41–41.7)

−

1.50

1.21 + − 0.50 3.60 (0.40–8.80) 5.4 (0.01–60)

1.16 3.70

1.24 3.93

MRFIT risk (% 5 years)

2.11 (0.06–32.0)

1.74 (0.05–90) 7.4 (0.3–55)

whole population, ACh response correlated with CHD, gender, diabetes and SBP (systolic blood pressure). NP response correlated with CHD, diabetes and SBP. The ACh/NP response ratio correlated with sex, CHD, SBP and diabetes. A very similar relationship was found in the male subgroup. However, in women, the strongest associations occurred with diabetes, smoking and SBP. No women with CHD were recruited in the present study. The effects of defined changes in significant risk factors on endothelial function are shown in Table 4 to indicate the magnitude of any effects. Endothelial response, defined as the response to ACh, NP or the ratio of ACh/NP responses, showed variable correlations with cardiovascular risk calculated by different algorithms (Table 5) and were notable for their wide variation. The risk algorithm derived from MRFIT correlated with endothelial function in all populations and gender subgroups. However, the Framingham algorithm showed a weak correlation in the whole population and correlated with both ACh and NP response, but not the ACh/NP ratio, in women. The Framingham algorithm risk did not correlate with the ACh, NP or the ACh/NP response ratio in men. In contrast, the PROCAM algorithm, which cannot be used in women, correlated strongly with endothelial function in men.

4.0 (1.25–7.83)

DISCUSSION Endothelial dysfunction is associated with the presence of individual cardiovascular risk factors and correlates, as confirmed in the present study, with the total number of risk factors present in a population [1,18]. Yet, despite this, few studies have examined the correlation between endothelial dysfunction and epidemiologically validated measures of cardiovascular risk.

Endothelial function and cardiovascular risk factors

Figure 1 Relationship of ACh/NP response to the number of cardiovascular risk factors in any patient

study on a receiver-operating characteristic plot (area = 0.706 + − 0.04; P = 0.001). Multiple regression analysis of ACh, NP or ACh/NP response ratio with cardiovascular risk factors, including age, lipids, blood pressure, smoking, diabetes, CHD, forearm flow and length, was performed (Table 3). In the

The correlation of the degree of endothelial dysfunction with individual risk factors and principally to the level of plasma LDL and presence of CHD is consistent with other large studies that have examined the individual determinants of endothelial response [1,19]. Endothelial dysfunction is highly correlated with established cardiovascular disease as it shows a 76 % predictive capacity for established CHD on receiver-operator characteristic curve analysis. Similar results were obtained in a 7.7 year prospective trial of 147 individuals, where previous CHD was associated with the greatest endothelial dysfunction and highest prospective risk [20], and in a 10 year follow-up study of 37 women with normal coronary angiograms, where the presence of endothelial dysfunction predicted a poor prognosis [21], and in other studies of patients with established or possible CHD [22]. C


611

612


Table 3 Correlation of individual cardiovascular risk factors (β) with ACh, NP or ACh/NP ratio in multiple regression analysis

Response Risk factors All (n = 246) Correlation Sex CHD SBP Diabetes Men (n = 183) Correlation CHD SBP Diabetes LDL-cholesterol Women (n = 63) Correlation SBP Triacylglycerol Smoking

ACh

P

P

ACh/NP ratio

P

0.55 − 3.14 − 3.81 0.05 − 3.59

< 0.001 < 0.001 < 0.001 0.003 0.004

2.83 0.027 − 2.17

< 0.001 – < 0.001 0.05 0.04

0.55 − 0.366 0.225 0.005 0.199

< 0.001 < 0.001 < 0.001 < 0.001 0.02

0.57 − 4.16 0.04 − 4.93 − 0.01

< 0.001 < 0.001 0.02 < 0.001 0.01

0.48 − 3.87 – − 2.68 –

< 0.001 < 0.001 – 0.002 –

0.45 − 3.87 0.002 − 0.26 − 0.0006

< 0.001 < 0.001 0.03 < 0.001 0.02

0.54 0.011 − 0.63 − 0.30

< 0.001 0.005 0.02 0.03

0.46 – − 4.75 –

< 0.001 – 0.03 –

0.53 – − 8.73 –

< 0.001 – 0.003 –

NP

0.46 –

Table 4 Effect of changes in cardiovascular risk factor status on the calculated ratio of response to ACh compared with NP

DBP, diastolic blood pressure. Risk factor

Change

Baseline

Outcome

Change

% change

Diabetes CHD Smoking Total cholesterol Triacylglycerol HDL-cholesterol SBP DBP

Absent/present Absent/present Absent/present + 1 mmol/l + 1 mmol/l − 0.25 mmol/l + 10 mmHg + 5 mmHg

0.73 0.82 0.71 0.71 0.71 0.71 0.71 0.71

0.49 0.61 0.64 0.65 0.71 0.71 0.74 0.68

− 0.24 − 0.21 − 0.07 − 0.06 0.00 0.00 + 0.03 − 0.03

− 33.0 − 26.0 − 10.0 − 8.5 0.00 0.00 + 4.2 − 4.2

(0.20–1.45) (0.19–1.44) (0.20–1.44) (0.19–1.22) (0.19–1.22) (0.19–1.21) (0.19–1.21) (0.19–1.21)

In primary prevention studies, impaired flow-mediated dilation was associated with a 1.3-fold excess risk for CHD [19], and in a prospective study of 400 postmenopausal women over a mean of 5 years using the flowmediated dilation method of determining endothelial function, impaired endothelial function correlated with poorer prognosis [23]. Analysis of endothelial function with respect to cardiovascular risk factors revealed a clear association of ACh and NP response with CHD and diabetes, although no relationship was seen with these risk factors in women due to the low numbers of women investigated with these risk factors. A study of similar design to the present one using brachial plethysmography in 69 patients in which response to ACh, bradykinin, glyceryltrinitrate C


(0.03–1.00) (0.00–1.23) (0.13–1.16) (0.13–1.16) (0.19–1.28) (0.20–1.23) (0.20–1.25) (0.16–1.20)

and l-NMMA (NG -monomethyl-l-arginine) was assessed also showed weak correlation of endothelial function with risk factors after exclusion of age, gender and basal flow parameters with a tendency towards a stronger correlation in women [24]. In that study, ACh response correlated with smoking, HDL (high-density lipoprotein)-cholesterol and body mass index, while nitrate response correlated with the total cholesterol/HDLcholesterol ratio and body mass index. Difference have been seen in the determinants of endothelial response between men [25] and women with a stronger role for LDL in men and triacylglycerols in women [26,27]. Although the point in the menstrual cycle has been suggested as a factor influencing endothelial function [28], not all studies show this effect [14], and measurements


Table 5 Correlations between endothelial response to ACh and NP and risk predictions from the Framingham, PROCAM and MRFIT studies ∗

P = 0.01–0.05,

∗∗

P = 0.001–0.01 and

∗∗∗

P = 0.001.

Risk algorithm Endothelial function measure Men ACh response NP response ACh/NP ratio Women ACh response NP response ACh/NP ratio All ACh response NP response ACh/NP ratio

Framingham

PROCAM

MRFIT

− 0.078 − 0.070 − 0.079

− 0.52∗∗∗ − 0.33∗ − 0.414∗∗∗

− 0.338∗∗∗ − 0.195∗∗ − 0.263∗∗∗

− 0.192∗∗∗ − 0.33∗∗∗ − 0.08

– – –

− 0.304∗∗∗ − 0.049 − 0.56∗∗∗

− 0.175∗∗ − 0.104 − 0.263∗∗∗

– – –

− 0.352∗∗∗ − 0.230∗∗ − 0.148∗

made in the present study were not correlated with the point in the cycle. Chan et al. [24] showed similar results with LDL being the principal factor in men and a stronger correlation of the total cholesterol/HDL-cholesterol ratio (a strong co-correlate of triacylglycerols) in women. Another study in the Framingham cohort of 2883 individuals (53 % female) using flow-mediated dilation as a measure of endothelial response has shown a relationship with age, sex, blood pressure, lipids and body mass index [6].

Endothelial function and cardiovascular risk estimates However, although endothelial dysfunction is strongly associated with the presence of CHD, only a moderate association was found between the degree of endothelial dysfunction and absolute risk as predicted from epidemiologically validated cardiovascular risk prediction algorithms in patients without established CHD. The Framingham algorithm has a predictive capacity of 70–80 % when applied prospectively to the NHANES 1 and NHANES 3 cohorts [29,30]. The Framingham and PROCAM calculators measure broadly similar degrees of risk in epidemiological studies with a good, although not perfect, cross-correlation [31–33]. However, although they may identify similar proportions of high-risk patients, it is uncertain as to whether they are completely concordant in individuals [34]. The MRFIT risk calculator has not been evaluated in other cohorts and relies on data from initially screened patients to compensate for the exclusion of diabetes in the final study, but correlated reasonably well with risk calculated by other techniques

in the present study. The discrepancy in the results obtained between the algorithms in higher risk patients in the present study is likely to be due to the PROCAM calculator being useful in patients with a family history of CHD where the degree of endothelial dysfunction is more profound, unlike the Framingham calculator, which only applies to index individuals, and similarly for MRFIT. The presence of a correlation with CHD risk in patients with established disease would agree with the results of one prospective 28 month study of 157 patients with mild established CHD [35], which has shown an association of endothelial dysfunction with new CHD events, and with multiple studies that have documented endothelial dysfunction in patients with established cardiovascular disease [35]. The MRFIT calculator, in contrast with Framingham, relies on quantified cigarette consumption in risk calculation rather than smoking as a dichotomous variable and this may account for its stronger relationship with endothelial function given the persistent association between cigarette smoke exposure and endothelial dysfunction [5].

The role of biological variation This association between endothelial function and cardiovascular risk, although moderate, was highly statistically significant given the numbers studied in the present investigation. The relatively poor correlation could arise for a number of reasons. Both the Framingham algorithm [32] and endothelial function are subject to significant effects from biological variation. Endothelial function responds acutely to cigarette smoke, mental stress, postprandial lipaemia or infusion of fat emulsions and thus even the use of averaged measurements may not be enough to reduce variation to acceptable limits [36]. Also, other work has shown [24] that 30–45 % of the differences in change in flow are dependent on basal flow and changes in flow in the control arm. Risk algorithms are subject to variation induced by differing definitions [33] and biological variation in their underlying risk variables, especially lipids and blood pressure, and multiple determinations are required to achieve accurate risk prediction [32]; even minor modifications can profoundly affect the properties of the high-risk individuals identified by these techniques. Cardiovascular risk factors were only measured once in the present study, but, given the size of the study, this is unlikely to have had a significant effect, but a difference in concordance for identification of high-risk individuals could explain the better correlation seen with the PROCAM compared with other algorithms.

Conclusions The results of the present study suggest that the individual risk factors make different relative contributions to endothelial dysfunction compared with their role as C


613

614


determined in long-term epidemiological studies of cardiovascular risk. Thus either endothelial dysfunction is related to components of risk not captured in the algorithms and, hence, may be a useful additional predictive factor [11,35], or any correlation between labile endothelial function and long-term determinants of risk is too weak to be of practical use. In summary, endothelial dysfunction is strongly correlated with some cardiovascular risk factors, in particular LDL-cholesterol and established CHD. However, its correlation with cardiovascular risk, as calculated from epidemiologically based calculators, was weak, except in patients with higher CHD risk and, therefore, especially in those patients with established CHD. Large-scale prospective studies [37] are required to identify whether the assessment of endothelial function is likely to be a useful physiological marker of the prospective risk of atherosclerotic disease.

REFERENCES 1 Anderson, T. J. (1999) Assessment and treatment of endothelial dysfunction in humans. J. Am. Coll. Cardiol. 34, 631–638 2 Chowienczyk, P. J., Watts, G. F., Cockcroft, J. R. and Ritter, J. M. (1992) Impaired endothelium-dependent vasodilation of forearm resistance vessels in hypercholesterolaemia. Lancet 340, 1430–1432 3 Clarkson, P., Celermajer, D. S., Powe, A. J., Donald, A. E., Henry, R. M. and Deanfield, J. E. (1997) Endotheliumdependent dilatation is impaired in young healthy subjects with a family history of premature coronary disease. Circulation 96, 3378–3383 4 Guerci, B., Bohme, P., Kearney-Schwartz, A., Zannad, F. and Drouin, P. (2001) Endothelial dysfunction and type 2 diabetes. Part 2: altered endothelial function and the effects of treatments in type 2 diabetes mellitus. Diabetes Metab. 27, 436–447 5 Celermajer, D. S., Sorensen, K. E., Georgakopoulos, D. et al. (1993) Cigarette smoking is associated with doserelated and potentially reversible impairment of endothelium-dependent dilation in healthy young adults. Circulation 88, 2149–2155 6 Benjamin, E. J., Larson, M. G., Keyes, M. J. et al. (2004) Clinical correlates and heritability of flow-mediated dilation in the community: the Framingham Heart Study. Circulation 109, 613–619 7 Felmeden, D. C., Spencer, C. G., Blann, A. D., Beevers, D. G. and Lip, G. Y. (2003) Low-density lipoprotein subfractions and cardiovascular risk in hypertension: relationship to endothelial dysfunction and effects of treatment. Hypertension 41, 528–533 8 Cockcroft, J. R., Chowienczyk, P. J., Benjamin, N. and Ritter, J. M. (1994) Preserved endothelium-dependent vasodilatation in patients with essential hypertension. N. Engl. J Med. 330, 1036–1040 9 Anderson, K. M., Odell, P. M., Wilson, P. W. and Kannel, W. B. (1991) Cardiovascular disease risk profiles. Am. Heart J. 121, 293–298 10 Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (2001) Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA, J. Am. Med. Assoc. 285, 2486–2497 11 Elliott, H. L. (1998) Endothelial dysfunction in cardiovascular disease: risk factor, risk marker, or surrogate end point? J. Cardiovasc. Pharmacol. 32 (Suppl. 3), S74–S77

C


12 Walker, H. A., McGing, E., Fisher, I. et al. (2001) Endothelium-dependent vasodilation is independent of the plasma l-arginine/ADMA ratio in men with stable angina: lack of effect of oral l-arginine on endothelial function, oxidative stress and exercise performance. J. Am. Coll. Cardiol. 38, 499–505 13 Chowienczyk, P. J., Watts, G. F., Wierzbicki, A. S., Cockcroft, J. R., Brett, S. E. and Ritter, J. M. (1997) Preserved endothelial function in patients with severe hypertriglyceridemia and low functional lipoprotein lipase activity. J. Am. Coll. Cardiol. 29, 964–968 14 Chowienczyk, P. J., Watts, G. F., Cockcroft, J. R., Brett, S. E. and Ritter, J. M. (1994) Sex differences in endothelial function in normal and hypercholesterolaemic subjects. Lancet 344, 305–306 15 Chowienczyk, P. J., Cockcroft, J. R. and Ritter, J. M. (1994) Blood flow responses to intra-arterial acetylcholine in man: effects of basal flow and conduit vessel length. Clin. Sci. 87, 45–51 16 Assmann, G., Cullen, P. and Schulte, H. (1998) The Munster Heart Study (PROCAM). Results of follow-up at 8 years. Eur. Heart J. 19 (Suppl. A), A2–A11 17 Grimm, R. H. and Svendsen, K. (2000) The ‘Grimm Meter’: a device for rapid estimation of absolute CHD mortality for use in clinical settings: implications for treatment based on absolute risk, 18th Scientific Meeting of the International Society of Hypertension, Chicago 18 Davignon, J. and Ganz, P. (2004) Role of endothelial function in atherosclerosis. Circulation 109 (Suppl. III), III-27–III-32 19 Celermajer, D. S., Sorensen, K. E., Bull, C., Robinson, J. and Deanfield, J. E. (1994) Endothelium-dependent dilation in the systemic arteries of asymptomatic subjects relates to coronary risk factors and their interaction. J. Am. Coll. Cardiol. 24, 1468–1474 20 Schachinger, V., Britten, M. B. and Zeiher, A. M. (2000) Prognostic impact of coronary vasodilator dysfunction on adverse long-term outcome of coronary heart disease. Circulation 101, 1899–1906 21 Bugiardini, R., Manfrini, O., Pizzi, C., Fontana, F. and Morgagni, G. (2004) Endothelial function predicts future development of coronary artery disease. A study of women with chest pain and normal coronary angiograms. Circulation 109, 2518–2523 22 Landmesser, U., Hornig, B. and Drexler, H. (2004) Endothelial function. A critical determinant in atherosclerosis? Circulation 109 (Suppl. II), II-27–II-33 23 Modena, M. G., Bonetti, L., Coppi, F., Bursi, F. and Rossi, R. (2002) Prognostic role of reversible endothelial dysfunction in hypertensive postmenopausal women. J. Am. Coll. Cardiol. 40, 505–510 24 Chan, N. N., Colhoun, H. M. and Vallance, P. (2001) Cardiovascular risk factors as determinants of endothelium-dependent and endothelium-independent vascular reactivity in the general population. J. Am. Coll. Cardiol. 38, 1814–1820 25 Sorensen, K. E., Celermajer, D. S., Georgakopoulos, D., Hatcher, G., Betteridge, D. J. and Deanfield, J. E. (1994) Impairment of endothelium-dependent dilation is an early event in children with familial hypercholesterolemia and is related to the lipoprotein(a) level. J. Clin. Invest. 93, 50–55 26 Forte, P., Kneale, B. J., Milne, E. et al. (1998) Evidence for a difference in nitric oxide biosynthesis between healthy women and men. Hypertension 32, 730–734 27 Kneale, B. J., Chowienczyk, P. J., Cockcroft, J. R., Coltart, D. J. and Ritter, J. M. (1997) Vasoconstrictor sensitivity to noradrenaline and NG -monomethyl-l-arginine in men and women. Clin. Sci. 93, 513–518 28 Chan, N. N., MacAllister, R. J., Colhoun, H. M., Vallance, P. and Hingorani, A. D. (2001) Changes in endothelium-dependent vasodilatation and α-adrenergic responses in resistance vessels during the menstrual cycle in healthy women. J. Clin. Endocrinol. Metab. 86, 2499–2504 29 Haq, I. U., Ramsay, L. E., Jackson, P. R. and Wallis, E. J. (1999) Prediction of coronary risk for primary prevention of coronary heart disease: a comparison of methods. Q. J. Med. 92, 379–385


30 Liao, Y., McGee, D. L., Cooper, R. S. and Sutkowski, M. B. (1999) How generalizable are coronary risk prediction models? Comparison of Framingham and two national cohorts. Am. Heart J. 137, 837–845 31 Empana, J. P., Ducimetiere, P., Arveiler, D. et al. (2003) Are the Framingham and PROCAM coronary heart disease risk functions applicable to different European populations? The PRIME study. Eur. Heart J. 24, 1903–1911 32 Reynolds, T. M., Twomey, P. and Wierzbicki, A. S. (2002) Accuracy of cardiovascular risk estimation for primary prevention in patients without diabetes. J. Cardiovasc. Risk 9, 183–190 33 Wierzbicki, A. S., Reynolds, T. M., Gill, K., Alg, S. and Crook, M. A. (2000) A comparison of algorithms for initiation of lipid lowering therapy in primary prevention of coronary heart disease. J. Cardiovasc. Risk 7, 63–71

34 Reynolds, T. M., Twomey, P. and Wierzbicki, A. S. (2004) Concordance evaluation of coronary risk scores: implications for CHD risk screening. Curr. Med. Res. Opin. 20, 211–219 35 Suwaidi, J. A., Hamasaki, S., Higano, S. T., Nishimura, R. A., Holmes, Jr, D. R. and Lerman, A. (2000) Longterm follow-up of patients with mild coronary artery disease and endothelial dysfunction. Circulation 101, 948–954 36 Sorensen, K. E., Celermajer, D. S., Spiegelhalter, D. J. et al. (1995) Non-invasive measurement of human endothelium dependent arterial responses: accuracy and reproducibility. Br. Heart J. 74, 247–253 37 Anderson, T. J., Roberts, A. C., Hildebrand, K. et al. (2003) The fate of endothelial function testing: rationale and design of the Firefighters And Their Endothelium (FATE) study. Can. J. Cardiol. 19, 61–66

Received 12 March 2004/19 July 2004; accepted 28 September 2004 Published as Immediate Publication 28 September 2004, DOI 10.1042/CS20040078

C


615