Arterial stiffness and renal function decline: are all stiffness measurements equally predictive?

One important issue in cardiovascular prevention is to better detect individuals at high risk of cardiovascular events (CVE) to earlier apply preventive therapeutic strategies. As arterial stiffness is an important pathophysiological factor linking cardiovascular disease and kidney disease [1], several studies have investigated the predictive value of arterial stiffness for the decline in renal function. In most cross-sectional and longitudinal studies, arterial stiffening occurs in parallel with renal function decline. Findings depend on the stage of chronic kidney disease (CKD) at entry, criteria for the assessment of renal function (estimated or measured GFR, incident proteinuria, or progression to dialysis), and population. In primary care population with normal renal function and/or mild renal dysfunction at the start of the study (eGFR ≥60 ml/min per 1.73m2 and no proteinuria), arterial stiffness measured by aortic pulse wave velocity (PWV) has an independent predictive value for incident CKD [2–6]. These longitudinal studies used either carotid–femoral PWV [3–5], brachial–ankle PWV [2] or cardioankle vascular index (CAVI) [6]. Consequently, another important issue is to determine, which arterial stiffness measurement has the best predictive value. However, no study had yet compared them. All the above measurements determined arterial stiffness from PWV, according to the foot-to-foot velocity method, that is, pressure or flow waveforms are obtained transcutaneously on two arterial sites, the time delay (transit time) is calculated between the proximal and distal ‘foot waveforms’, and then PWV is calculated as the ratio of the travelled distance to the transit time [7]. Carotid–femoral PWV (cfPWV) is considered as the gold standard for determining aortic stiffness [8]. With cfPWV, the pressure wave travels mostly along an aortic segment, including the thoracic descending aorta and the abdominal aorta, and ultimately travels along the iliac and common femoral arteries. Several studies and a consensus statement have determined the correction factor, which should be applied to the carotid–femoral distance, to take into account the fact that when the pressure wave is recorded at the carotid level, it has already reached the descending thoracic aorta. In order to increase easiness and acceptability, automatic cuff-based methods have been developed. With brachial–ankle PWV, the travelled distance is automatically calculated based on patient's height. Brachial and post-tibial arterial pressure waveforms are simultaneously detected by cuffs connected to a plethysmographic sensor and an oscillometric pressure sensor wrapped on both arms and ankles [9]. The measurement of brachial–ankle PWV includes a much longer trajectory of the pressure wave along the muscular arteries of the upper and lower limbs than along the aortic pathway, and thus may not reflect the true ageing of the aorta. However, the main assumption of the developers of this method was that the transit times of the pressure waves in the upper and lower limbs were comparable. Thus, the net transit time that is measured reflects mainly the aortic pulse transit time. Using a similar cuff-based methodology for detecting the pressure waveforms and an ECG recording, a cardio-ankle (or heart–ankle) PWV can be calculated. A feature of the cardio-ankle PWV is that it shortcuts the subclavian and brachial artery pathways, compared with carotid–femoral PWV and brachial–ankle PWV. Cardio-ankle PWV reflects the stiffness of the aorta, iliac artery, femoral artery and tibial artery [10]. Finally, a last issue is to take into account the BP dependency of arterial stiffness measurement. The arterial wall of large arteries like the aorta is a complex entanglement of distensible elastic fibers, stiff collagen fibers, and vascular smooth muscle cells connected to the extracellular matrix, which includes a number of other components [11]. All of them are influenced by BP – the higher the BP, the higher the arterial stiffness. Thus, arterial stiffness measurements should be compared at a given BP. Shirai et al. [10] derived from the Bramwell and Hill equation and heart–ankle PWV a cardio-ankle vascular index (CAVI) as a BP-independent stiffness parameter. However, CAVI calculation still includes BP, even if this is as the log value of SBP/DBP, and the true BP-independency of CAVI has been debated. Spronck et al. [12] proposed a novel index, that is, CAVI0 as a variant form of CAVI that theoretically excludes dependence on BP. Thus, one may expect that CAVI0 has a better predictive value for outcome than CAVI or haPWV when no adjustment on BP is performed. The article by Nagayama et al. [13] in Journal of Hypertension, provides an important contribution with regard to these issues. Nagayama et al. [13] present data from a cohort of 27 864 Japanese urban residents without renal impairment at baseline who participated in two to eight consecutive annual health examinations, had a mean follow-up of 2.5 years, and had measurements of arterial stiffness at baseline. Arterial stiffness was measured according to three methods: CAVI, haPWV, and CAVI0. During the study period, 6.6% of participants developed renal function decline (estimated glomerular filtration rate <60 ml/min per1.73 m2). The predictive values of arterial stiffness for the decline in renal function were compared by using receiver-operating characteristic curve analyses. Findings are novel, timely, and important. First, the C-statistic of CAVI was higher than those of haPWV and CAVI0 (C-statistic: 0.740 vs. 0.734 vs. 0.726), suggesting a better predictive value of CAVI than haPWV, and a better predictive value of haPWV than CAVI0 for the decline in renal function. Second, when the discriminatory powers of the three parameters were compared using C-statistics, the net reclassification index (NRI) of CAVI was higher than those of haPWV and CAVI0, which means that the percentage of individuals with renal function decline correctly reclassified by CAVI was 15.9% (11.2–20.6%) compared with haPWV, and 47.7% (43.0–52.3%) compared with CAVI0. This finding was unexpected as CAVI0 theoretically excludes dependence on BP, and thus, should be more predictive for organ damage, including renal function decline, beyond BP. The study by Nagayama et al.[13] has several strengths. First, it has been done in a very large cohort of individuals. Second, several annual examinations were performed and the mean follow-up was 2.5 years. Third, arterial stiffness measurements were performed automatically, thus avoiding any observer bias, and the three arterial stiffness parameters were calculated from cardio-ankle PWV. Finally, the study was performed by experts in the field. As with all stimulating research, many questions arise from the work by Nagayama et al.[13]. We will raise two issues. The first one relates to the BP dependency of arterial stiffness. How to take into account the confounding effect of BP when measuring PWV? It may be sufficient to adjust PWV on BP using a multivariate analysis when studying a large number of individuals. But it cannot be applied to a measurement in one individual, for instance, during clinical practice. This is why researchers have suggested to incorporate the SBP/DBP values, recorded during the PWV measurement, into the calculation of arterial stiffness. As discussed by the authors, the value of haPWV fluctuates with BP throughout the cardiac cycle, from DBP to SBP. CAVI0 is based on PWV at DBP as the foot-to-foot method calculates the time delay between the feet of the pressure waveform, that is, at DBP, whereas CAVI corresponds to PWV measured at MBP [14], and not at DBP. But CAVI0 did not prove to have a better predictive value than CAVI in the study by Nagayama et al.[13]. In addition, the simplest measurement of arterial stiffness, haPWV, proved to be as predictive as CAVI in a multivariate analysis where BP was not included (Table 3, model 2). Thus, why not make it simple and calculate only haPWV? However, it is important to compare the predictive value of these three indices of arterial stiffness for other events than the decline in renal function, particularly for cardiovascular events. A second issue is the comparison of the predictive values of other methods of arterial stiffness measurement for cardiorenal events. Indeed, although meta-analyses are available for the predictive values of carotid–femoral PWV [15], brachial–ankle PWV [16], and cardio-ankle PWV [17] for cardiovascular events, to our knowledge no comparison has yet been performed between these measurements of arterial stiffness. At the end of the day, what matters is that there should be enough scientific evidence to choose the right arterial stiffness parameter, that is, the one, which has the best predictive value for cardiorenal events beyond BP and other cardiovascular risk factors. In conclusion, the study by Nagayama et al. [13] provides a timely and valuable contribution to the study of arterial stiffness’ predictive value for cardiorenal events, and the comparison of various methods based on cardio-ankle PWV, among which CAVI proved to be the best predictor for renal function decline in individuals without renal impairment at baseline. ACKNOWLEDGEMENTS Conflicts of interest There are no conflicts of interest.