Arterial Calcifications, Arterial Stiffness, and Cardiovascular Risk in End-Stage Renal Disease
To test the predictive values of and independent contributions to cardiovascular and all-cause mortality of various arterial parameters exploring characteristics of the arterial wall at different sites, we studied prospectively 110 stable end-stage renal disease patients on hemodialysis. These parameters involved carotid diameter, carotid intima-media thickness, carotid compliance, carotid distensibility, carotid incremental elastic modulus, aortic diameter, aortic pulse wave velocity, and the presence of arterial calcifications measured at the sites of the carotid artery, abdominal aorta, iliofemoral axis, and legs. The presence of calcifications was analyzed semiquantitatively as a score (0 to 4) according to the number of arterial sites with calcifications. During a follow-up of 53±21 months (mean±SD), 25 cardiovascular and 14 noncardiovascular deaths occurred. In univariate analysis, the carotid incremental elastic modulus was the most closely related to prognosis. Risk of death increased with the number of vascular sites involved by calcifications. Moreover, information (in terms of prediction) given by carotid elastic incremental modulus was additive to the presence and extent of vascular calcification-related prediction value. Adjusted hazard ratios of all-cause and cardiovascular mortality for an increase of 1 unit in calcification score were 1.9 (95% confidence interval [CI], 1.4 to 2.6) and 2.6 (95% CI, 1.5 to 4.4), respectively (P<0.001 for both). Adjusted hazard ratios of all-cause and cardiovascular mortality for a 1-SD increase in carotid incremental elastic modulus were 1.6 (95% CI, 1.2 to 2.2) and 1.7 (95% CI, 1.2 to 2.4), respectively (P<0.01 for both). The results of this study showed that the presence and extent of vascular calcifications were strong predictors of cardiovascular and all-cause mortality. Carotid incremental elastic modulus gave additional predictive value.
Autopsy studies have shown that arterial plaques and calcified lesions were frequent after 30 years of age.1 In 1986, Witteman et al2 reported a relationship between aortic calcifications and cardiovascular (CV) mortality. Wilson et al3 recently reported, on Framingham populations, that aortic calcifications were independently predictive of subsequent vascular morbidity and mortality beyond established risk factors. Calcification of elastic lamellae and increased calcium content are frequently observed in the arteries of uremic patients, alterations quantitatively associated with age, fibrinogen levels, and duration of hemodialysis.4 It has recently been reported that the presence of vascular calcifications in end-stage renal disease (ESRD) patients was associated with increased stiffness of large, capacitive, elastic-type arteries like the aorta and common carotid artery (CCA).4 Because CCA incremental elastic modulus (Einc) and aortic pulse wave velocity (PWV) were both identified as strong predictors of prognosis in ESRD patients5,6 and because, to the best of our knowledge, no prospective study has reported associations between arterial calcifications and prognosis in ESRD populations, we designed this study to test the predictive value of various arterial parameters exploring characteristics of the arterial wall at different sites. These parameters involved CCA diameter, CCA intima-media thickness (IMT), CCA compliance, CCA distensibility, CCA Einc, aortic diameter, aortic PWV, and the presence of arterial calcifications measured at the sites of the CCA, abdominal aorta, iliofemoral axis, and legs. Subsequently, we examined whether these different parameters had independent contributions to CV and all-cause mortality.
Patients were eligible for inclusion if they (1) had been on hemodialysis for ≥3 months (81±79 months, mean±SD), (2) had no clinical CV disease during the 6 months preceding entry, and (3) agreed to participate in the follow-up study, which was approved by our institutional review board. The study began in September 1994; recruitment was closed in June 1998; and follow-up ended December 31, 2000. We enrolled 110 patients who fulfilled the entry criteria. Patients who underwent renal transplantation and patients who moved were censored on the day of transplantation or departure. Of the participants, 62% were male and 8% had type 1 diabetes mellitus. The follow-up was 53±21 months (mean±SD) (range, 6 to 75 months). Data on mortality were obtained for the entire cohort. The age of the cohort at inclusion was 54±16 years. During follow-up, all patients were dialyzed with the same techniques as previously detailed.6
The CCA systolic and diastolic diameters, IMT, and wall motion were measured by a high-resolution B-mode (7.5-MHz transducer) echo-tracking system (Wall Track System, Neurodata). A detailed description of this system has already been published.7 A localized echo structure encroaching into the vessel lumen was considered to be a plaque if the CCA IMT was >50% thicker than that of neighboring sites.7,8 CCA pulse pressure was recorded noninvasively by applanation tonometry with a pencil-type probe incorporating a high-fidelity Millar strain-gauge transducer (SPT-301, Millar Instruments). A detailed description of this system has been published previously.7 Calculation, repeatability, and reproducibility of CCA-lumen cross-sectional area, intima-media cross-sectional area, CCA compliance, CCA distensibility, and CCA Einc have been published in detail elsewhere.5,7
Diastolic internal aortic diameter was measured ultrasonographically (Sonel 300, Compagnie Générale de Radiologie) with 3.5-MHz transducers as previously described.4 Aortic PWV was determined with transcutaneous Doppler flow recordings and the foot-to-foot method.9 A detailed description of this measurement has been published previously.6,7 The intraobserver repeatability of aortic PWV measurements was 5.8±1% (mean±SD).10
The presence of arterial calcifications was evaluated ultrasonographically in the CCA, abdominal aorta, iliofemoral axis, and legs as previously described.4 Highly echogenic plaques producing bright white echoes with shadowing were considered to be calcifications.11 Assessment of the presence of calcifications was complemented with posteroanterior and lateral fine-detail radiographs of the abdomen and pelvis. Arterial calcifications of the femorotibial arterial axis were evaluated by soft-tissue native radiographs. Arterial calcification at each arterial site was quantified qualitatively as absent (0) or present (1). The final overall score was obtained by the addition of calcifications from all studied zones. The final score ranged from 0 (absence of calcium deposit) to 4 (calcifications present in all arterial segments examined). The calcification score was independently checked by 2 observers with good reproducibility as previously published.4
The outcome events studied were all-cause and CV mortality. The primary analysis concerned the survival curves and Cox proportional-hazards regression model. Survival was estimated with the Kaplan-Meier product-limit method and compared by use of the Mantel (log-rank) test. Significant predictors (P<0.05) identified by Cox proportional-hazards regression analysis were age at inclusion, smoking, previous CV events, peripheral systolic, diastolic and pulse pressures, carotid pulse pressure, albumin, fibrinogen, C-reactive protein, arterial calcifications, calcification score, CCA diameter, CCA IMT, CCA compliance, CCA distensibility, CCA Einc, aortic diameter, and aortic PWV. The multivariate Cox proportional-hazards regression model was then used to test for the independent relationship between these parameters and the outcomes. Variable selection was automatically performed by the statistical software (NCSS 6.0.21); no variable was forced in the model. Reproducibility of the methods was defined as recommended by the British Standards Institution.12
Table 1 shows the characteristics of patients at inclusion according to calcification score. There is a positive association between calcification score and the following variables: age, smoking, time on dialysis before inclusion, previous CV events, pulse pressure, fibrinogen, C-reactive protein, aortic diameter, aortic PWV, carotid PP, CCA diameter, CCA IMT, and CCA Einc. We noted a negative association between calcification score and the following variables: diastolic blood pressure, albumin, CCA compliance, and CCA distensibility. Risk of death increased with the number of vascular sites having calcifications (for 0 to 4 vascular sites with calcifications, risk of all-cause mortality was 3%, 17%, 31%, 50%, and 73%, respectively; P<0.001).
During a follow-up of 53±21 months, 25 CV and 14 non-CV deaths occurred. Univariate proportional-hazards regression analyses of all-cause and CV mortality are shown in Table 2. Only statistically significant univariate models are shown. We noted that the best univariate models in terms of variance of proportional-hazards regression explained were those containing CCA Einc for both all-cause and CV mortality.
Table 3 shows the multivariate proportional-hazards regression analysis of all-cause and CV mortality (best models in terms of variance of proportional-hazards regression explained). For both outcomes, only calcification score and CCA Einc entered the models. There was no significant interaction between those 2 parameters (data not shown). None of the other prognostic variables in univariate analysis—age at inclusion, smoking, previous CV events, peripheral systolic, diastolic and pulse pressures, carotid pulse pressure, albumin, fibrinogen, C-reactive protein, CCA diameter, CCA IMT, CCA compliance, CCA distensibility, aortic diameter, and aortic PWV—entered the final models for either CV or all-cause mortality. Adjusted hazard ratios of all-cause and CV mortality for an increase of 1 unit in calcification score were 1.9 (95% confidence interval [CI], 1.4 to 2.6) and 2.6 (95% CI, 1.5 to 4.4), respectively (P<0.001 for both). Adjusted hazard ratios of all-cause and CV mortality for a 1-SD increase in carotid incremental elastic modulus were 1.6 (95% CI, 1.2 to 2.2) and 1.7 (95% CI, 1.2 to 2.4), respectively (P<0.01 for both).
The Figure shows the probability of all-cause survival according to calcification score. Comparison between curves was highly significant (χ2=42.66, P<0.0001).
The principal finding of the present study was that both presence of arterial calcifications and increased CCA Einc were strongly and independently predictive of outcome in ESRD. Furthermore, calcification score and CCA Einc summarized the integrality of predictive information in the population studied.
In the literature, reported associations between arterial stiffness and atherosclerosis have led to conflicting results,13,14 partly because of the chosen surrogate markers (eg, CCA plaques, CCA IMT, femoral IMT, aortic PWV, CV risk factors, arterial calcifications, CV events) and measurement methods. Several possibilities for the association between arterial stiffness and atherosclerosis can be hypothesized. First, the presence of atherosclerosis could lead to stiffening of the arteries. Second, increased arterial stiffness could lead to vessel wall damage and atherosclerosis. Third, both mechanisms could apply, and atherosclerosis not only would be a consequence of arterial stiffness but may by itself in advanced stages also increase arterial stiffness. This would result in a self-perpetuating reinforcing process. A final possibility is that arterial stiffness and atherosclerosis are independent processes that frequently occur at similar sites in the arteries without a causal relationship.13 The present study is not designed to elucidate the mechanisms; future long-term longitudinal studies, preferably starting in young subjects, are needed to elucidate the temporal relationship between arterial stiffness and atherosclerosis. Whatever the mechanisms involved, in the present study, there were strong intercorrelations between parameters related to atherosclerosis burden, such as calcification score, and parameters related to arterial stiffness, such as CCA Einc (r=0.50, P<0.001), but although closely related, each parameter had incremental predictive value in terms of CV and all-cause mortality beyond that of the other. There was no significant interaction between those 2 parameters; ie, the effect of each parameter was constant whatever the level of the other. The absence of interaction is important to consider in the present study. Arterial stiffness had incremental predictive value in terms of CV and all-cause mortality even in patients presenting extensive calcifications. Nevertheless, the power of the interaction analysis is low in the present study; only a strong interaction could be assessed.
Whether enhanced arterial stiffness is a risk factor contributing to the development of CV disease or is a marker of established CV disease is a matter of debate.6 A study in Chinese and Australians15 has suggested that morphological and structural alterations of the aorta may be influenced by both environmental and mostly genetic factors, suggesting that changes in biomechanical properties of major arteries may precede the development of clinically overt disease. Guérin et al16 have recently shown that the loss of aortic PWV sensitivity to blood pressure was predictive of adverse outcome, indicating that arterial stiffness not only is a risk factor contributing to the development of CV disease but also is a marker of established, more advanced, less reversible arterial changes. Taken together, these results support the hypothesis that measurement of arterial parameters exploring both structural and functional properties such as CCA Einc, in addition to quantification of arterial calcium deposit, could then help not only in risk assessment strategies but also in risk reduction strategies by monitoring those arterial parameters under different drug regimens.
The principal limitation of the present study is the semiquantitative evaluation of vascular calcification. Using the score to evaluate arterial calcification is rather crude and is likely to underestimate the true calcium load. It also does not allow a quantitative assessment of the calcium concentration in the arteries. More sensitive methods (helical or electron-beam computed tomography imaging) should be used in the future to quantify vascular calcifications and to assess the time-related alterations of vascular calcium content. However, using a simple score is reproducible and specific, and the method is inexpensive and readily available. These results suggest that vascular imaging will improve our ability to predict and then prevent CV events, and further research using newer technologies should help us to define the utility of vascular calcium measures beyond that of established risk factors.
In conclusion, results of this study showed that the presence and extent of vascular calcifications were strong predictors of CV and all-cause mortality. Measurement of CCA Einc gave additional predictive value. Finally, the extrapolation of the conclusions reached in the present study may be limited because of the particular clinical characteristics of ESRD patients, who are at very high risk of CV complications; thus, further studies are needed to extend these findings to other populations.
This work was supported by the Groupe d’Etude de Pathophysiologie de l’Insuffisance Renale, the Société Française d’Hypertension Artérielle, Daniel Brun for Organica Association, the Fédération Française de Cardiologie, and the Groupe de Pharmacologie et d’Hémodynamique Cardio-vasculaire. We are deeply indebted to Professor Michel E. Safar, whose concepts, analysis, and reflections were the foundations for this work. We thank Wendy Kay Johnson for her helpful linguistic assistance.
- Received April 28, 2001.
- Revision received July 2, 2001.
- Accepted July 2, 2001.
Solberg LA, Strong JP. Risk factors and atherosclerotic lesions: a review of autopsy studies. Arteriosclerosis. 1983; 3: 187–198.
Wilson PWF, Kauppila LI, O’Donnell CJ, Kiel DP, Hannan M, Polak JM, Cupples LA. Abdominal aortic calcific deposits are an important predictor of vascular morbidity and mortality. Circulation. 2001; 103: 1529–1534.
Guérin AP, London GM, Marchais SJ, Metivier F. Arterial stiffening and vascular calcifications in end-stage renal disease. Nephrol Dial Transplant. 2000; 15: 1014–1021.
Blacher J, Pannier B, Guerin A, Marchais S, Safar M, London G. Carotid arterial stiffness as a predictor of cardiovascular and all-cause mortality in end-stage renal disease. Hypertension. 1998; 32: 570–574.
Blacher J, Guerin A, Pannier B, Marchais S, Safar M, London G. Impact of aortic stiffness on survival in end-stage renal disease. Circulation. 1999; 99: 2434–2439.
Roman MJ, Saba PS, Pini R, Spitzer M, Pickering TG, Rosen S, Alderman MH, Devereux RB. Parallel cardiac and vascular adaptation in hypertension. Circulation. 1992; 86: 1909–1918.
Avolio AP, Chen S, Wang R, Zhang C, Li M, O’Rourke MF. Effects of aging on changing arterial compliance and left ventricular load in a northern Chinese urban community. Circulation. 1983; 68: 50–58.
London GM, Marchais SJ, Guerin AP, Metivier F, Safar ME, Fabiani F, Froment L. Salt and water retention and calcium blockade in uremia. Circulation. 1990; 82: 105–113.
Savage T, Clarke AL, Giles M, Tomson CRV, Raine AEG. Calcified plaque is more common in the carotid and femoral arteries of dialysis patients without clinical vascular disease. Nephrol Dial Transplant. 1998; 13: 204–212.
Precision of Test Methods: Guide for the Determination of Reproducibility for a Standard Test Method. London, England: British Standards Institution; 1979:BS 5497,pt 1.
Van Popele NM, Grobbee DE, Bots ML, Asmar R, Topouchian J, Reneman RS, Hoeks APG, van der Kuip DAM, Hofman A, Witteman JCM. Association between arterial stiffness and atherosclerosis: the Rotterdam study. Stroke. 2001; 32: 454–460.
Avolio A. Genetic and environmental factors in the function and structure of the arterial wall. Hypertension. 1995; 26: 34–37.
Guérin A, Blacher J, Pannier B, Marchais S, Safar M, London G. Impact of aortic stiffness attenuation on survival of patients in end-stage renal disease. Circulation. 2001; 103: 987–992.