This is the first study to examine the relationship between AD assessed with CMR and outcome in ESRD. Patients with ESRD have previously been demonstrated to have increased aortic stiffness compared to controls, represented by reduced AD and VAS with associated vessel dilatation[6]. In keeping with the relationship between arteriosclerosis and advancing age, there was a significant relationship between reduced AD and VAS and age. AD and VAS were both lower in patients at highest CV risk, i.e. those with diabetes, ischaemic heart disease and/or peripheral vascular disease. No other factors such as drug therapy, adequacy of dialysis, duration of renal replacement therapy, dialysis modality or any laboratory parameter appeared to be significantly associated with AD or VAS.
One major potential limitation of our study is that calculation of AD is based on cross-sectional volume of the vessel wall and pulse pressure. For practical reasons, direct aortic blood pressure measurement has been substituted by non-invasive indirect brachial blood pressure. Brachial blood pressure is not entirely representative of central haemodynamics and our methods do not permit assessment of the phenomenon of central to peripheral systolic and pulse pressure amplification, as can be measured with tonometry. It is difficult to dissociate any clinical effect associated with changes in AD from that due directly to pulse pressure, which is a well described predictor of outcome in ESRD. Nonetheless, as it is otherwise impossible to assess aortic stiffness without documentation of pressure within the vessel lumen, the approach used in this study has been used widely[7, 9–11]. For this reason, aortic VAS, the fractional increase in aortic volume during the cardiac cycle was also used. This does not depend on blood pressure variables and in this study displayed similar relationships with clinical variables and outcome. However, there is limited literature to support its use. Other studies have demonstrated that PWV can be measured with CMR. There are advantages to using measuring PWV compared to AD as it is load-independent, negating the necessity for blood pressure measurements. However, at the time of initiation of this study, there were few studies in any patient group other than healthy volunteers[10, 12], and unfortunately CMR derived PWV was not measured in this study cohort.
The relationship between these markers of aortic function and LV dimensions show a weak but significant relationship between increased aortic stiffness and increasing LVMI and end systolic volume, suggesting that reduced AD may increase LV wall tension and hence cardiac hypertrophy. Aortic VAS, which is independent of blood pressure, correlated with markers of systolic function, namely ejection function and stroke volume, suggesting that in the failing heart where systolic blood pressure is lower, increased arterial stiffness remains deleterious to ventricular performance.
A variety of studies have addressed the role of AD as a marker of either CV risk or relating AD to cardiac performance in other groups at high risk of CV disease. Data relating CMR measures of arterial function to long term outcome are scarce. In otherwise healthy individuals, obese subjects have been shown by CMR to have increased aortic cross sectional area and decreased aortic elasticity[13, 14]. AD is reduced in patients with heart failure and correlates with exercise capacity[9]. One study in hypertensive patients has demonstrated that AD increases following treatment with nicardipine or alacepril but not trichlormethiazide, independent of changes in pulse pressure[15]. More recently, in patients with non diabetic CKD, AD has been shown to correlate with GFR demonstrating that CMR can quantify the impact of reduced kidney function on vascular function in early CKD[7]. Additionally, endothelial function can be studied using CMR to by assessing cross sectional flow mediated dilation of the brachial artery[16].
Independent predictors of mortality during the follow up period were diabetes, AD and systolic blood pressure. Due to their close interdependence, systolic blood pressure and aortic distensibilty could not be independently assessed. When a combined end point of death, non-fatal myocardial infarction, cardiac revascularisation, amputation for peripheral vascular disease and cerebrovascular event was used, only diabetes and AD or VAS were independent predictors of events. Surprisingly, age was not a predictor of outcome, suggesting that since age and AD were closely correlated, vascular, rather than temporal, aging is a more important determinant of survival. In keeping with other studies, time on renal replacement therapy was significantly longer in those patients who died[4], but this was not independently associated with outcome. Therefore in this patient group, AD was a predictor of mortality and/or vascular events independent of age and dialysis vintage.
Haemodynamic factors which promote arterial remodelling present in ESRD, although not specific to kidney disease, include age and blood pressure. Factors specific to ESRD include arteriovenous shunts and chronic volume overload[17]. Increased blood flow through a shunt promotes arterial remodelling although may not lead to detectible changes in measures of vascular function[18]. We could find no difference in AD or VAS in patients with an arteriovenous shunt and those without. As novel marker of increased CV risk, no data exist on the natural history of progression of AD, or whether it can be improved by therapeutic intervention. Our results suggest that as there was no significant difference between those CKD 5 patients on dialysis compared to pre-dialysis patients, the major derangement in vascular function occurs earlier during the progression of CKD. One assumes that reduced AD is a consequence of arterial calcification[19], although we cannot confirm this as CMR is unable to display calcific lesions due to their absence of water content.
A large number of studies have been performed to assess the determinants of arterial stiffness in ESRD, and its relation to long term survival, using either incremental elastic modulus, PWV or augmentation index as markers of arterial stiffness. In keeping with our results, factors associated with increased arterial stiffness include age, diabetes, and systolic blood pressure (or pulse pressure) [20–22]. Other factors associated with arterial stiffness include serum calcium or the presence of inflammation[22]. Surprisingly, these factors were not predictors of vascular function in our study. Arterial stiffness has been repeatedly demonstrated to be an independent predictor of all cause and cardiovascular mortality in haemodialysis patients[3, 4, 23].
Aside from the issue of use of brachial blood pressure to calculate central haemodynamics, our study has some other limitations. Although direct comparison between CMR and calcification is not possible, CMR has emerged as a technique for imaging atheroma. A direct comparison between vascular function and dark blood imaging looking at aortic wall thickness and atheroma would be of interest but unfortunately was not performed. Measurement of AD is still a relatively novel technique and its widespread reproducibility is yet to be established. In house blinded analysis has demonstrated an inter-observer variability of 4.6%, although this needs to be reproduced in other centres, particularly as there are potential sources of error in both data acquisition and analysis as well as in blood pressure measurements. As we have previously used CMR with late gadolinium enhancement (LGE) to analyse myocardial tissue composition in ESRD[5], it would have been of interest to compare myocardial tissue abnormalities with vascular function. However, as a link between gadolinium and nephrogenic systemic fibrosis has emerged, this comparison is impossible[24]. It is unfortunate that it was not possible to acquire brachial blood pressure on every patient due to some patients having multiple vascular procedures on both upper limbs. Finally, CMR technique is relatively expensive and time consuming to analyse. With future development of automated analysis software, faster analysis of AD will be possible.