Article Text
Abstract
Objective The effects of cardiovascular risk factors on the vascular anatomy at differing sites of the arterial vasculature have not been well described. The aim of this study was to compare the effect of cardiovascular risk factors on the intima media thickness (IMT) of the wall of the right and left common carotid artery (CCA) at their bifurcation and proximal from their bifurcation, and the effects on the presence of plaque at carotid and femoral arteries.
Design Cross-sectional population-based study.
Setting/participants Random samples of men (n=425) and women (n=367) aged 56–77 years were recruited from two general practices participating in the British Regional Heart Study.
Main outcome measures Ultrasound examination ascertained IMT and the presence of atheromatous plaque. A model for correlated outcomes was used to simultaneously model all risk factor on all measured vascular sites.
Results All cardiovascular risk factors (HbA1c, waist-to-hip ratio, hypertension, LDL and smoking) showed a larger association with IMT thickening at the wall of the CCA at its bifurcation than proximal to its bifurcation. The IMT was greater on the left wall of the CCA than on the right. The association between hypertension with the wall of the CCA depended on age. Smoking was the only risk factor that demonstrated an increased odds (45%, 95% CI 14% to 65%) of the presence of plaque at the femoral arteries when compared with the CCA.
Conclusions The associations of cardiovascular risk factors with the vascular anatomy are not uniform for IMT thickening or the presence of plaque. These differences in local arterial anatomy may result in differences between trial outcomes that investigate surrogate endpoints such as IMT.
- Atherosclerosis
- IMT
- multivariate
- risk
- cardiovascular
- risk stratification
Statistics from Altmetric.com
Introduction
A small number of common, modifiable cardiovascular risk factors account for the burden of coronary artery disease in most regions of the world.1 These include raised blood lipids, smoking, hypertension, diabetes, abdominal obesity, psychosocial factors, consumption of fruits, vegetables, alcohol and regular physical activity. Risk factors for other vascular disease are broadly similar to those of coronary artery disease.2
Epidemiological studies have reported associations between a range of cardiovascular risk factors and intima medial thickness (IMT).3–7 IMT has been proposed as a quantitative index of atherosclerosis of value in monitoring disease progression, and as a surrogate outcome marker in clinical trials.8–11 B-mode ultrasound offers a non-invasive method of examining the peripheral arteries for IMT at other sites than the heart.12 13 The measured IMT comprises endothelial cells, connective tissue and smooth muscle and is the site of lipid deposition in plaque formation.14 In healthy adults, IMT ranges from 0.25 mm to 1.5 mm,13 and values of >1.0 mm are often regarded as abnormal.14 Lipid deposition leads to lipid core plaques with the lipid contained in foam cells. These can subsequently become fibrous and/or calcified to form fibrocalicific plaques.15 However atherosclerosis is a multifactorial disease resulting in the formation of heterogeneous plaques. Not all plaques will progress in the same linear manner as different plaque developmental pathways probably exist. The dominating pathway is probably determined by the local artery factors (geometry/anatomy) and individual risk factors. Hence plaques at the common carotid artery (CCA) are predominantly foam cell lesions and those at the femoral arteries (FA) more often fibrous.16 17
It is conceivable that certain risk factors might have a different association with IMT and the presence of plaque measured at different regions of the arterial tree. Indeed, our previous study suggested that most risk factors were associated more strongly with the IMT at the carotid bifurcation than at the common carotid, possibly due to greater turbulence of blood flow at artery bifurcations.7 18–21 However, these original analyses did not allow for the correlation within individuals of IMT measured at different vascular sites. It was also not possible to compare the effects of risk factors on IMT at each vascular site because data were averaged across right and left carotid arteries.
The purpose of this study was to evaluate coronary disease risk factors for their associations with the presence of plaque in the carotid and femoral arteries (FA), and to examine the association of the risk factors with the IMT in right and left carotid bifurcations and common carotid arteries. We hypothesised that associations between risk factors and plaque/IMT at different arterial sites would be similar. This hypothesis can only be tested in vivo by examining all arterial sites within a person and testing whether within a person the effect of risk factors differ by site or not.
Method
Study population
Data were used from the Dewsbury and Maidstone sub-study of the British Regional Heart Study (BRHS). This is a secondary analysis of these data from which previous work has been published.7 18–21 The BRHS is a prospective study of cardiovascular disease involving 7735 men aged 40–59 years who were randomly selected from the age-sex registers of one general practice in each of 24 British towns and screened between 1978 and 1980. In 1996, surviving men from two of the BRHS towns, Dewsbury in northern England and Maidstone in southern England, were invited for detailed follow-up including ultrasonography of carotid and FA. A random sample of women of similar age to the men (56–75 years) was also selected from each general practice in both towns and they were also invited to attend an interview and examination. In the UK the vast majority of the population (98%) are on the general practice register.22
Risk factor measurements
Waist measurements were taken using the midpoint between the lowest rib and iliac crest and hip measurements using the largest circumference below the waist. A flexible metal tape was used and two measurements taken to the nearest millimetre. A Dinamap 1846SX vital signs monitor was used to measure blood pressue (BP) using standard procedures. The Dinamax 1846SX monitor systematically overestimates systolic BP by 8 mm Hg and all measurements were corrected for this error. A sample of fasting venous blood was taken for analysis of lipids and glycated haemoglobin. A self-administered questionnaire was used to obtain information on smoking behaviour. Subjects were classified into never, former and current smokers.
Ultrasound
Measurements were made with an Advanced Technology Laboratories HDI (high definition imaging) 3000 duplex scanner with a high resolution broadband width linear array transducer L 7–4 MHz. All ultrasound measurements were made by two experienced vascular ultrasonographers, and images were recorded on magneto-optical disks and VHS videotapes for further analysis. The carotid bifurcation was examined over a length of 3 cm (1.5 cm proximal and distal to the flow divider) for plaques. The common carotid IMT was measured at its thickest point on the distal (far) wall of the CCA, approximately 1.5 cm proximal to the flow divider. An artery was classified as being affected by plaque if there was a localised thickening >1.2 mm that did not uniformly involve the whole vessel circumference with or without flow disturbance.12 23 IMT was also measured at the origin of the bulb which was defined as the point at which the arterial wall diverges to form the bulb. In the presence of a plaque, the maximum thickness of the plaque was measured and this was taken as the bifurcation IMT. In the absence of a plaque, the IMT measured at the bulb origin was the thickest part of the intima-media complex and was defined as IMT. The within- technologist and between-technologist coefficients of variation for IMT measurements were assessed in pilot studies and were between 3% and 5%, which was consistent with previous reproducibility estimates from this laboratory. The mean difference between technicians showed no evidence of bias and was −0.005 mm with 95% limits of agreement of −0.12 mm to+0.11 mm. Full details of the methods have been published previously.7
Statistical analysis
Two multivariate outcome models were fitted in MlwiN.24–26 (For relevant models see appendix.) The first model included all four IMT measurements (ie, far walls of the right and left CCA proximal to their bifurcation and also at their bifurcation). The second model used the presence or absence of plaque at the right and left CCA and the right and left FA as the outcome variable.
The risk factors considered were HbA1c, waist-to-hip ratio, smoking status, LDL levels, hypertension, sex and age. As both systolic and diastolic pressures have showed association with IMT we used the mean (diastolic+1/3(systolic-diastolic)) BP as a continuous variable in our modelling. To compare the relative size of the association between the risk factors and IMT/plaque we also fitted effects of standardised risk factors. Standardisation was undertaken by centring the continuous risk factors on their mean value and dividing them by their SD.
The simultaneous modelling allowed the effects of risk factors on each of the outcome measurements to be estimated while taking the correlation between risk factors and measurements within a given individual into account. The advantage of this approach over an analysis examining each outcome separately is that it is possible to formally test for differences in the effect of a given vascular risk factor at each vascular site. For example, it is possible to test in the first model whether LDL has a stronger effect at the bifurcation of the right than the left carotid artery. The association of the risk factors with the outcomes was tested with a joint significance test. The inclusion of an interaction would allow for the formal comparison of any difference in the interplay of the risk factors between different vascular sites. An a priori decision was therefore made to undertake a single comparison of the difference in the effect of BP, with age, between the right and left CCA by testing the significance of an interaction term between age and hypertension (appendix 2). The right and left carotids have different vascular geometry and the test of this interaction would demonstrate if there is any difference in the effect of BP with increasing age. As women have more cardiovascular disease (CVD) than men the possibility existed that the effect of BP over time may differ between the sexes. To investigate this, interaction terms between sex with BP and sex with age were fitted one at a time to the model.
Result
Patient characteristics
A total of 425 men and 367 women were studied from both towns, representing response rates of 79% in Dewsbury and 72% in Maidstone. The prevalence of plaques was similar in men and women (see table 1). Forty-two per cent of men and women had no CCA plaque but 32% of men and 30% of women had bilateral CCA plaques. In men 11.5% had plaques only in the left CCA and 12% only in the right. In women 12.5% had plaque in the left CCA and 13.9% in the right CCA. The mean IMT in men was 0.9 mm in the left and 0.86 mm in the right CCA. In women IMTs in the CCA were 0.79 mm and 0.78 mm, respectively. At the bifurcation of the CCA the IMT was 1.99 mm in the left and 1.93 mm in the right in men and in women it was 1.79 mm and 1.72 mm, respectively.
The age distribution was similar between males and females (65.8 years vs 65.2 years). The waist-to-hip ratio was higher in males than females (0.93 vs 0.80). The HbA1c, LDL and average mean BP in males were 4.9%, 3.6 mmol/l and 108.6 mm Hg and 4.7%, 4.1 mmol/l and 102.0 mm Hg in females, respectively. Approximately half of the females had not smoked previously and 29% were ex-smokers compared to males where 28% had never smoked previously and approximately half were ex-smokers. A similar percentage of males (20.3%) and females (19.7%) were still current smokers. Men had higher IMT readings at all sites when compared to women of similar age (p=0.03). Male sex was associated with larger IMT at the bifurcation than the IMT at the CCA (p=0.011) when compared to the same effect in women. However, there was no evidence that a male sex predisposed to plaque formation (p=0.132 for all plaque sites) when compared to women.
Association of modifiable cardiovascular risk factors with IMT at all CCA sites
The association of the cardiovascular risk factors with the IMT at each CCA site are displayed in detail in table 2. The upper part of the table shows the effect at each CCA site with joint p values, which are testing for an overall association of a given risk factor with all sites simultaneously. Table 3 shows the difference in the associations of a risk factor with the carotid bifurcation when compared to its association proximal to the bifurcation. The association estimates for the risk factors expressed as a 1 SD change on outcome are also presented.
Joint p values showed strong evidence of associations with outcomes for all risk factors except for waist-to-hip ratio, which only had a borderline effect. When examining the individual regression coefficients in the upper portion of the table, there appeared to be a pattern of risk factors having much larger associations with the bifurcation than proximal to it. For example, on average a 1% increase in HbA1c was positively associated with IMT at all carotid sites (joint p value 0.004). A 1% increase in HBA1c was associated with a 0.067 mm greater association (p=0.004) with the bifurcation IMT than common carotid IMT. A 10% increase in waist-to-hip ratio was associated with a 0.091 mm larger association (p=0.014) with bifurcation IMT than common carotid IMT. On average current smokers had increased association with IMT thickening at all vascular sites but with bifurcation IMT it was substantially larger by 0.246 mm (p<0.001) compared with common carotid IMT. Similarly, an increase of 1 mmol/l LDL cholesterol level was associated with a significantly larger association in the bifurcation IMT of 0.077 mm (p=0.001) compared with common carotid IMT. A 5 mm Hg increase in the mean BP was on average associated with a larger association with the IMT by 0.02 mm (p=0.017).
The standardised measures of association highlight that one SD change in waist-to-hip ratio and HbA1c have, on average, similar association with the bifurcation IMT, with effects of 1 SD change in LDL and mean BP having slightly less pronounced associations (upper part of table 2). In terms of SD changes of risk factors, the association with CCA IMT was largest for waist-to-hip ratio, followed by HbA1c, LDL cholesterol and mean BP. These associations resulted in a relatively larger effect on the IMT of the CCA at its bifurcation than of the CCA proximal to its bifurcation.
Association of BP with the IMT in the right and left carotid arteries
Increases in both age and BP were positively associated with greater IMT. However, there was a need to fit an interaction between age and BP to achieve adequate model fit (log likelihood test p value = 0.002). This meant that the higher the BP and the higher the age, there was a multiplicative effect of both on IMT at the bifurcations (appendix 2). The dependence of the age-effect on underlying BP was weaker on the left side at the proximal wall (with a negative interaction term) compared to the right proximal wall (a positive interaction term). The effect of this age-hypertension interaction on right and left CCA is illustrated in figure 1. In those individuals with mean BP below 90 mm Hg (chosen for illustration), the estimated association of increasing age with the IMT of the right and left CCA proximal to their bifurcation is identical—the curves are parallel; however, for each age group examined, the right CCA appears to be thinner than the left CCA to start with. On the other hand, in patients with mean BP more than 90 mm Hg the right and left CCA seem to have equivalent IMT thickness and the curves appear to overlap. The increase of IMT with age is steeper in those with lower mean BP and less steep in those who have a higher mean BP. These comparisons were done within a person of a given age and sex, and adjusted for sex. This means that in those with a low mean BP, there was evidence for a difference between left and right side, while there was no such difference observed for people with high mean BP. Figure 1 also shows that there was a lot of individual variability and that the small and significant differences between left and right side and age are in the range of tenth of millimetres.
The inclusion of neither interaction terms, sex with BP or sex with age, significantly improved the model fit.
Presence of plaque in the CCA and the FA
The lower part of table 2 summarises the association between the risk factors and the presence of plaque at the right and left CCA and the right and left FA. The estimated difference in the associations of the risk factors with the presence of plaque in the carotid and FA are shown in table 4. In contrast to analyses on IMT, HbA1c, waist-to-hip ratio and LDL cholesterol had similar association across the sites examined. As displayed in table 4, there was no evidence of a difference in the association between all risk factors, except smoking, and the presence of plaque at the CCA and FA.
Smokers, when compared to non-smokers, were much more likely to be associated with the presence of plaque at any site examined (upper part of table 3, joint p value <0.001), but within individuals there was a 45% decrease in the odds of association with the presence of plaque in the CCA from that in the FA (p=0.009).
Of note was the difference in the effect of BP on the presence of plaque between the right and left CCA, with BP increases associated with increased presence of plaque in the right CCA (9%) when compared to the left CCA (2%) (p=0.016, not shown on table). Otherwise, BP had comparable effects on the presence of plaque in both FA.
Discussion
We investigated the association between cardiovascular risk factors and IMT measurement at different vascular sites in a joint analysis. As expected, there was strong evidence that all well-established risk factors were associated with IMT, independent of age and sex. Associations between the risk factors and IMT at the bifurcation were stronger than with the CCA proximal from the bifurcation, confirming our earlier simpler analyses.7 All risk factors were significantly associated with the presence of plaque at the CCA and FA. All risk factors examined had similar effects on the presence of plaque in the CCA and FA with the exception of smoking. Smokers showed a predilection for plaques predominantly occurring in the FA. Figure 1 shows graphically that age and BP interact and have differential effects according to the side of the body—for example, a person with mean arterial pressure of 90 mm Hg or below has good chances of a bilateral CCA IMT increase but the left IMT will always be thicker than the right IMT, while a person with high BP (mean arterial pressure of >90 mm Hg) will have weaker increases with age, without obvious difference in right and left CCA intima. Therefore, our hypothesis that risk factors would be associated with IMT and plaque to a similar extent at all sites of the arterial tree was not supported. The association of cardiovascular risk factors assessed with IMT and the presence of plaque is not homogeneous across the vasculature.
Differential associations between risk factors and IMT and the presence of common carotid plaques have been reported.27–29 Rubba et al,30 Polak et al31 and Tell et al28 have all reported different risk factors with differing strengths of association with different segments of the carotid artery. Espeland et al27 suggested that blood glucose and smoking had an asymmetrical effect on IMT along segments of the carotid artery, whereas the effect of LDL and hypertension was more homogeneous. Similar to the findings in other studies the IMT in our study was more in the left CCA than the right and in both it increased in a linear fashion with age.32–34 Also smoking showed a relative stronger association with the presence of peripheral vascular plaques as also seen by Leng et al35 when compared to ischaemic heart disease.
The relatively larger differential effect of traditional risk factors on the IMT at the bifurcation compared with the CCA raises further questions about their pathophysiological role in the development of atherosclerosis. Both the sites differ in their geometry, shear stress, extracellular matrix and cell composition.36 Dalager et al16 looked at 4756 microscopic sections of the vasculature from 100 individuals from the ages of 20 to 82 years in an autopsy study. They found the initiation, development and phenotypic expression of atherosclerotic plaques was artery-related, such that the femoral artery were least affected by atherosclerosis and the carotid bifurcation the most likely to develop plaques. Possible explanations can be hypothesised as to why the bifurcation demonstrated a significantly larger association with risk factors than the CCA. First, the difference could be the result of biological differences between the wall and the bifurcation of the carotids. The bifurcation has more macrophages and hence may be more susceptible to lipid accumulation36 and, therefore, the relatively larger effect of LDL cholesterol at the bifurcation than the proximal wall of the CCA.
Matrix metalloproteinases (MMPs) and their inhibitors (TIMPs) play a central part in arterial wall remodelling,37 affecting the stability of the fibrous caps covering atherosclerotic plaques. Choudary et al38 investigated the MMP and TIMP distribution and activity in carotid endarterectomy tissue. Segments at or near the bifurcation of the CCA contained higher MMP levels and activity than segments distant from the bifurcation. This would suggest that signals associated with the remodelling of the arterial wall are dependent on the vascular anatomy. Blood flow exerts shear stress on the vessel wall, which changes as the blood flow divides at the bifurcation. Haemodynamic studies have shown that differences in branch angle and asymmetry in three-dimensional curvature have a key role in determining the resulting wall shear stress patterns.39 Berceli et al40 41 showed by changing the arterial flow, by the unilateral ligation of the internal carotid artery and three of four branches of the external carotid artery, that the expression of MMPs and its regulation in arterial wall is influenced by haemodynamic changes. This may explain the above mentioned differences in abundance of MMPs according to vascular anatomy.
Apart from MMP and TIMP, interleukin-1β (IL-1β) appears to be a factor that modulates intimal hyperplasia.42–44 Using IL-1 receptor I null mice Jiang et al45 demonstrated that perturbation in the arterial blood flow induced the local expression of IL-1β in a time-dependant fashion. Thus difference in the forces experienced at the different vascular sites could interplay with the risk factors to show varying effects. In this study the interplay between hypertension and age was investigated and the increase in IMT with age depended on underlying BP, and in those with lower mean BP, the right CCA had less IMT than the left, with no such difference in those with higher mean BP. The difference in the geometry between the right and left CCA may offer a possible explanation for this difference. A higher BP, generating greater forces in the arterial vasculature, may over time show more IMT on the right side than the left owing to their differing local geometry. This might explain why in our cross-sectional data we did not find a larger IMT in the left compared to the right side in those with higher BP. The findings are supported from an earlier correlation study, which found only a 25% correlation between the right and left CCA in IMT by ultrasound. They concluded that the low correlation may be due to different focal component factors in the development of atherosclerosis.46 Mean arterial pressure was chosen for multivariate analysis because it is constant along the arterial tree and hence reflects the impact of pressure on the different carotid segments more reliably. Interestingly in this study the odds of the presence of plaque with increasing BP was also significantly higher for the right CCA compared to the left.
Certain limitations of the study need to be mentioned. First, the data precede the more common use of statins and ACE inhibitors seen today. Second, since this is a cross-sectional study the associations observed cannot be inferred to explain disease progression or event outcomes. As this is a secondary analysis we are limited by the numbers in the dataset. To detect differences of effects between risk factors, apart from smoking on plaque formation, we would have needed a larger dataset. The strength of our study is that it simultaneously evaluated all exposures and outcomes at once with all effects adjusting for each other. All correlations between the outcomes were taken into account when associations were investigated with the cardiovascular risk factors and any differences in such associations were inferred from the single model fitted.
Multivariate outcomes are very common in modern clinical studies. Many studies collect a wide range of variables on individuals, and some of these variables may measure correlated entities. There are several advantages to fitting a multivariate outcome model as the ones used here. First, the effect of the risk factors on the array of outcomes under investigation is adjusted for the correlation that exists between the outcomes within a given individual. The total effect of exposures on a range of outcomes of a similar meaning can be formally tested without the need to decide on how to aggregate data into outcome scores or combining information from the left and right side, which in turn are hard to interpret.
Our paper highlights the fact that the established cardiovascular risk factors do not have a uniform risk profile across the arterial vasculature within the same person and also with age, as seen previously.34 This has implications for trials, such as the lipid lowering trials, which have used carotid IMT measures as a surrogate outcome for cardiovascular disease and the added practical concern of measuring both carotid sides, particularly in older patients.34 47 Beneficial results of disease regression or the lack of it at this site from such trials will be non-generalisable to other vascular sites.
Acknowledgments
We thank Andrew Nicolaides, who organised all the ultrasound.
References
Supplementary materials
Web Only Data hrt.2009.188219
Files in this Data Supplement:
Footnotes
Funding Stroke Association.
Competing interests None.
Ethics approval This study was conducted with the approval of the university.
Provenance and peer review Not commissioned; externally peer reviewed.