Brachial-Ankle Pulse Wave Velocity and the Cardio-Ankle Vascular Index as a Predictor of Cardiovascular Outcomes in Patients on Regular Hemodialysis

Authors


Dr Akihiko Kato, Blood Purification Unit, Hamamatsu University School of Medicine, 1-20-1 Handayama, Higashi-ku, Hamamatsu, Shizuoka 431-3192, Japan. Email: a.kato@hama-med.ac.jp

Abstract

Brachial-ankle pulse wave velocity (baPWV) and the cardio-ankle vascular index (CAVI) are both used to evaluate arterial stiffness. The aim of the present study is to determine whether baPWV or CAVI is superior as a marker of arterial stiffness in hemodialysis (HD) patients. Of 194 patients, 59 patients had been excluded from the study due to advanced age over 76 years old (n = 29), or abnormal ankle-brachial pressure index (ABI) (<0.90 or ≥1.30) (n = 30). We then followed the 135 patients (age: 60 ± 11 years, time on HD: 110 ± 93 months) for the 63 ± 4 (55–70) months. Thirty-two (23.7%) patients had expired, 22 of them of cardiovascular (CV) causes. There were 37 fatal and non-fatal CV events. Kaplan–Meier analysis revealed that the patients with the highest tertile of baPWV (≥16.6 m/s) had a significantly lower survival rate (P < 0.01) when compared with the second (13.4 ≤ baPWV < 16.6 m/s) and the lowest tertiles (<13.4 m/s). Cox hazards analysis after adjustment for comorbid risk factors revealed that the top tertile of baPWV was a determinant of CV death (hazards ratio [HR]: 16.9 [1.1–251.8], P < 0.05) In contrast, CAVI did not associate with CV mortality or events. These findings suggest that baPWV is superior to CAVI as a predictor of CV outcomes in patients on regular HD.

Cardiovascular disease (CVD) is the main cause of mortality and morbidity in HD patients. To screen the patients at high risk for CVD, several non-invasive methods have been developed, including pulse wave velocity (PWV), the cardio-ankle vascular index (CAVI) and ankle-brachial pressure index (ABI).

Brachial-ankle PWV (baPWV) is widely applied to assess arterial stiffness in the clinical setting, owing to its time efficacy and technical simplicity. It is demonstrated that baPWV is well correlated with carotid-femoral (aortic) PWV (1), and its power to predict the presence of CVD is comparable to aortic PWV in general population (1). Increased baPWV is an independent determinant of carotid atherosclerosis (2) and diastolic left ventricular dysfunction (3) in HD patients.

The CAVI has been developed as a novel indicator of arterial distensibility. CAVI is measured from an electrocardiogram (ECG), phonocardiogram, brachial artery waveform, and ankle artery waveform, and adjusted for blood pressure (BP) based on the stiffness parameter β(4). So, CAVI is theoretically less influenced by BP during measurement. In fact, CAVI is demonstrated to be superior to baPWV (5) and carotid artery intima-medial thickness (CA-IMT) as an index of coronary arteriosclerosis (6). CAVI has been demonstrated to be better correlated with the parameters of left ventricular diastolic indices than baPWV in patients with chest pain syndrome (7).

In HD patients, CAVI is increased, and CAVI over 7.6 is shown as an indicator of the presence of CVD with both sensitivity and specificity of 79% (8). CAVI is also correlated with the histological arterial fibrosis of vascular access better than baPWV (9). In contrast, Ueyama et al. (10) demonstrated that CAVI and baPWV had a similar power to predict carotid atherosclerosis in HD patients. However, whether baPWV or CAVI is a better indicator of clinical outcomes remains to be determined.

It has been demonstrated that both baPWV (11) and CAVI (12) cannot be estimated properly when ABI is less than 0.9 or greater than 1.3. In the affected legs with peripheral arterial disease (PAD), baPWV and CAVI are apparently decreased. So, the impact of baPWV and CAVI became less when analyzed together with ABI (13). A recent cohort study also demonstrated that baPWV is not an independent predictor for clinical outcomes without any exclusion criteria for abnormal ABI (14). Thus, when estimating the significance of arterial stiffness, it is necessary to select the patients with normal ABI (13,15).

The main aim of the present study was to compare the impact of baPWV and CAVI on cardiovascular (CV) outcomes in HD patients having normal ABI (0.9 to 1.3). We measured baPWV, CAVI and ABI using a VeraSera VS-1000 (Fukuda Denshi, Tokyo, Japan) at the same time, and examined which marker of arterial stiffness is superior as a predictor of CV outcomes.

PATIENTS AND METHODS

Study populations

The patient population was 194 patients (age: 64 ± 12 [25–96] years old, time on HD: 112 ± 100 [2–396] months, male/female : 127/67) who had been undergoing regular HD at two dialysis units (Maruyama Hospital and Maruyama Clinic, Hamamatsu, Japan). Since mean life expectancy is estimated within 5 years in dialysis patients aged over 75 years (14), we first excluded 29 elderly patients older than 76 years from the study. We also excluded another 30 patients whose ABI was lower than 0.9 or higher than 1.3.

After determining basal parameters, we followed them up for 63 ± 4 (55–70) months. The diagnosis of CVD (history of acute myocardial infarction (AMI), angina pectoris (AP), stroke and PAD) or diabetes mellitus (DM), the presence of traditional major CV risk factors and drug prescription were recorded from medical charts, and then analyzed. Information on mortality and CV events was identified during outpatient follow-up visits or hospitalization in the two units, by reviewing hospital records or by directly contacting the referring physicians.

During the follow-up, we had treated the patients to achieve the target levels of hemoglobin more than 10 g/dL, serum phosphorous lower than 6.0 mg/dL, intact parathyroid hormone (PTH) below 180 pg/mL, and systolic and diastolic blood pressure (BP) below 140/90 mm Hg according to the clinical practice guidelines from Japanese Society for Dialysis Therapy (16–18) and Kidney Disease: Improving Global Outcomes (KDIGO) (19).

During the follow-up, no patient had transferred to other institutes or had received kidney transplantation. So, we analyzed data using all of the patients.

HD procedures

All patients had been subjected to regular HD for 3.5–5.0 h three times per week at a blood flow rate of 180–250 mL/min. All patients used bicarbonate dialysate at a dialysate flow rate of 500 mL/min. There was no patient who had received hemodiafiltration. We used either polysulfone or cellulose triacetate membrane during the follow-up. Neither bacteria nor pyrogens were detected in the dialysate prepared from water obtained by reverse osmosis. Using an endotoxin removal filter, endotoxin concentration in dialysate was below 0.05 mEU/mL by routine analysis with Limulus amebocyte lysate assay (Wako Junyaku endotoxin measurement kit, Tokyo, Japan).

Blood samples

Blood samples were drawn from the arterial site of the arteriovenous fistula at the start of each HD session after the 2-day interval Serum electrolytes, urea nitrogen, creatinine, albumin, total cholesterol and triglycerides were measured by standard laboratory techniques using an autoanalyzer. Intact PTH was determined by immunoradiometric assay. Serum ferritin was measured by the latex agglutination method. Highly-sensitive C-reactive protein (hs-CRP) was measured by latex photometric immunoassay (Wako Junyaku, Tokyo, Japan). Efficacy of dialysis was assessed based on the delivered dose of dialysis (Kt/Vurea) using a single-pool urea kinetic model. Normalized protein catabolic rate (nPCR) was calculated from dialysis urea removal and serum urea levels.

Assessment of ABI, baPWV and CAVI

Measurements of ABI, baPWV and CAVI were conducted before the midweek HD cycle in supine position at rest for 10 min as described previously (20–22). We applied cuffs at both legs and a brachium, which was not used for blood access, and monitored ECG and heart sounds during the measurement. PWV from the heart to the ankle was obtained by calculating the superficial path lengths from the elbow to the suprasternal notch (Da) and from the suprasternal to the femur to the ankle (Db) based on anthropometric data for the Japanese population. To detect the brachial and ankle pulse waves with cuffs, the pressure of the cuffs was kept low at 30 to 50 mm Hg to ensure a minimal effect of cuff pressure on the hemodynamics. We measured the time interval between the initial increase in brachial and tibial waveforms (Ta), and obtained baPWV as follows: baPWV = (Db − Da)/Ta. CAVI is automatically calculated using the formula; a[ρP × [ln Ps/Pd] PWV2] + b (a,b, constant; ρ, blood density; ΔP, difference in systolic and diastolic pressure; Ps, systolic pressure; Pd, diastolic pressure; PWV, heart-ankle pulse wave velocity). After obtaining bilateral baPWV or CAVI, the higher one was used for analysis. The validation of this device and its reproducibility was previously published (20).

All measurements and calculations were made together and automatically in CAVI-VaSera VS-1000. We repeatedly measured these parameters at both legs in each patient, and expressed those as the means.

Statistical analysis

Values were expressed as the means ± SD. The χ2 test was used for categorical variables including gender, underlying kidney disease, prevalence of CVD and smoking habits. Univariate correlations between baPWV and CAVI with laboratory variables were tested using a non-parametric Spearman's rank analysis, then used anova to compare distributed variables. Populations of baPWV and CAVI were divided into their three tertiles and analyzed as a categorical variable (baPWV: <14.3, 14.3 ≤ baPWV < 16.6, baPWV ≥ 16.6 m/s; CAVI: <8.0, 8.0 ≤ CAVI < 9.9, CAVI ≥ 9.9).

The primary outcome studied was all-cause and CV mortalities. Fatal and non-fatal CV events were also used as a secondary outcome. In addition, CV events especially related to BP control such as CVD, sudden death, HF and others were also examined as another outcome. Survival rate was estimated using the Kaplan–Meier curves, and compared by using the log-rank test. Cox proportional hazards models were applied to calculate hazard ratio (HR) and adjusted survival curve for time to death; no adjusting (model 1), adjusting for age, gender and DM (model 2), and further adjusting for 12 comorbid risk factors (time on HD, current smoking habit, usage of angiotensin converting enzyme inhibitor (ACEI) /angiotensin II receptor blocker (ARB), mean BP, serum creatinine, calcium, phosphorous, albumin, total cholesterol, hemoglobin, intact PTH, log-transformed ferritin) (model 3). Results of the Cox proportional hazards analysis were presented as the HR and the 95% CI. A P-value less than 0.05 was considered statistically significant. All statistical calculations were performed with StatView 5J software (SAS Institute, Cary, NC, USA).

RESULTS

Baseline characteristics

The underlying kidney diseases were chronic glomerulonephritis (n = 81), diabetic nephropathy (n = 21), polycystic kidney disease (n = 8), nephrosclerosis (n = 4), others (n = 12), and unknown (n = 9). There was no patient who had complications from atrial fibrillation on ECG at entry.

Table 1 presents basal characteristics of the study population. There was a higher baPWV (17.7 ± 4.3 vs. 15.8 ± 3.5 m/s, P < 0.05) and CAVI (10.9 ± 4.1 vs. 9.5 ± 2.8, P = 0.06) in patients with previous history of CVD than in those without. A significantly higher baPWV was found in patients with statin treatment than those without (19.7 ± 5.4 vs. 15.6 ± 3.2, P < 0.01). Usage of anti-platelet agents was also associated with increased CAVI (10.3 ± 3.5 vs. 9.3 ± 2.3, P < 0.05). There was no difference in baPWV (15.9 ± 3.0 vs. 15.8 ± 3.0 m/s) and CAVI (9.8 ± 3.1 vs. 9.5 ± 2.6) between those with and without ARB/ACEI treatment. In this study, β-blockers had been prescribed only in three patients, so we did not evaluate its association.

Table 1. Baseline characteristics of the study population
ParametersValueMedian (range)
  1. Data are expressed as means ± SD. ACEI, angiotensin converting enzyme inhibitor; ARB, angiotensin II receptor blocker; baPWV, brachial-ankle pulse wave velocity; BP, blood pressure, ankle-brachial pressure index; CAVI, cardio-ankle vascular index; HD, hemodialysis; hs-CRP, highly sensitive C-reactive protein; nPCR, normalized protein catabolic rate; PTH, parathyroid hormone.

Characteristics  
 Age (years)60 ± 1162 (25–75)
 Time on HD (months)110 ± 9378 (2–363)
 Male/female91/44 
 Diabetes (%)36.4 
 Cardiovascular disease (%)12.5 
 Smoker: current/past (%)22.4/25.4 
Medications  
 ACEI/ARB (%)30.5 
 Anti-platelet therapy (%)31.3 
 Stain (%)3.8 
Laboratory parameters  
 Blood urea nitrogen (mg/dL)72.6 ± 14.071.3 (25.4–112.9)
 Creatinine (mg/dL)12.47 ± 2.6512.44 (5.54–20.44)
 Albumin (g/L)37 ± 337 (28–43)
 Calcium (mg/dL)9.0 ± 0.98.8 (7.0–11.3)
 Phosphorous (mg/dL)6.0 ± 1.35.9 (3.3–10.8)
 Total cholesterol (mg/dL)148 ± 34146 (82–271)
 Triglycerides (mg/dL)121 ± 82102 (25–514)
 Intact PTH (pg/mL)244 ± 173212 (6–978)
 Ferritin (ng/mL)150 ± 86138 (9–445)
 Hemoglobin (g/dL)10.6 ± 1.110.5 (7.5–14.3)
 Hs-CRP (mg/L)3.1 ± 6.50.4 (0.0–22.0)
 Body mass index (kg/m2)20.7 ± 2.820.5 (15.1–30.4)
 Kt/Vurea1.21 ± 0.221.15 (0.47–1.64)
 nPCR (g/kg per day)0.99 ± 0.180.97 (0.47–1.64)
Atherosclerotic parameters  
 Systolic BP (mm Hg)139 ± 23138 (81–207)
 Diastolic BP (mm Hg)83 ± 1384 (43–138)
 Mean BP (mm Hg)105 ± 11105 (54–165)
 ABI1.09 ± 0.081.09 (0.91–1.28)
 baPWV (m/s)16.0 ± 3.615.2 (9.8–29.1)
 CAVI9.7 ± 3.09.0 (6.0–24.9)

Table 2 shows univariate correlations between baPWV and CAVI with laboratory variables. baPWV was positively correlated with age, systolic and diastolic BPs (P < 0.01) and log-transformed hs-CRP (P < 0.05), while negatively with serum albumin (ρ < 0.05), phosphorous (P < 0.01) and Kt/Vurea (P < 0.01). anova analysis revealed that an association of baPWV with age, systolic and diastolic BP and Kt/Vurea. CAVI was correlated with age (ρ < 0.01), systolic BP (P < 0.01), time on HD (P < 0.05), serum albumin (P = 0.05) and triglyceride (P = 0.05). There was a significant correlation between CAVI and age, time on HD and systolic BP following anova.

Table 2. Univariate correlations of arteriosclerotic parameters with laboratory data
 baPWVCAVI
ρP-valueρP-value
  1. baPWV, brachial-ankle pulse wave velocity; BMI, body mass index; BP, blood pressure; CAVI, cardio-ankle vascular index; HD, hemodialysis; Log hs-CRP, log-transformed highly-sensitive C-reactive protein; MAP, mean arterial pressure; nPCR, normalized protein catabolic rate; PTH, parathyroid hormone.

Age0.53<0.010.48<0.01
Time on HD−0.140.11−0.20<0.05
Creatinine−0.130.12−0.110.20
Calcium−0.090.30−0.130.15
Phosphorous−0.26<0.01−0.140.12
Alkaline phosphatase0.110.200.060.65
Total cholesterol−0.010.880.070.41
Triglycerides0.19<0.050.170.05
Albumin−0.18<0.05−0.170.05
Log hs-CRP0.23<0.050.190.09
Hemoglobin−0.120.180.030.75
Intact PTH0.140.12−0.030.71
Ferritin0.080.920.080.33
Kt/Vurea−0.26<0.01−0.150.08
nPCR0.020.85−0.800.38
Systolic BP0.61<0.010.51<0.01
Diastolic BP0.34<0.010.150.09
Mean BP0.60<0.010.43<0.01
BMI0.040.610.030.70

Table 3 portrays the characteristics of study population divided into three tertiles of baPWV and CAVI. Increased baPWV and CAVI levels were associated with increased age, a higher prevalence of diabetes and male gender.

Table 3. Characteristics of study population divided into the three tertiles of brachial-ankle pulse wave velocity (baPWV) and cardio-ankle vascular index (CAVI)
NbaPWV (m/s)CAVI
<14.314.3 ≤ < 16.6≥16.6P-value<8.08.0 ≤ < 9.9≥9.9P-value
454446454446
  1. ACEI, angiotensin converting enzyme inhibitor; ARB, angiotensin II receptor blocker; CAD, coronary artery disease; CVD, cerebrovascular disease; HD, hemodialysis.

Age (years)53 ± 1159 ± 1067 ± 6<0.0154 ± 1261 ± 965 ± 7<0.01
Time on HD (months)133 ± 109105 ± 8992 ± 110.10138 ± 10399 ± 9195 ± 810.06
Male (%)53.376.273.9<0.0547.776.177.3<0.01
Diabetes (%)4.431.815.2<0.014.513.031.8<0.01
Current smoking (%)27.723.822.70.9412.220.036.80.03
Previous CAD (%)4.4013.00.032.38.76.80.37
Previous CVD (%)4.54.58.70.644.76.56.80.90
ACEI/ARB usage (%)22.245.526.2<0.0521.439.131.00.19

Mortality and CV events

During the follow-up, 32 patients (23.7%) had died, 22 of them due to CV causes. The causes of CV death were sudden death in six, cerebrovascular disease (CVD) in six, AMI in four, congestive heart failure (HF) in three, intestinal ischemia in two and PAD in one patient. Non-CVD death was due to infectious diseases in three, malignancies in three, gastrointestinal bleeding in two, uremia in one and unknown in one patient.

There were totally 37 fatal and non-fatal CV events during the follow-up. The causes were coronary artery disease in nine (AMI in six and AP in three), CVD in nine, sudden death in six, congestive HF in six, valvular heart disease in three, intestinal ischemia in two and PAD in two cases. One patient had received limb amputation. There was no patient who had received coronary artery bypass grafts during the observation.

Table 4 lists differences in basal characteristics between the expired and the surviving patients. In the expired group, basal baPWV was significantly higher (P < 0.01) while CAVI was not. Expired patients were significantly older, and had lower serum creatinine and albumin values (P < 0.01).

Table 4. Baseline characteristics in the expired and the surviving groups
ParametersAliveExpiredP-value
  1. Data are expressed as means ± SD. ABI, ankle-brachial pressure index; ACEI, angiotensin converting enzyme inhibitor; ARB, angiotensin II receptor blocker; baPWV, brachial-ankle pulse wave velocity; BMI, body mass index; BP, blood pressure; CAVI, cardio-ankle vascular index; HD, hemodialysis; hs-CRP, highly sensitive C-reactive protein; nPCR, normalized protein catabolic rate; PCR, protein catabolic rate; PTH, parathyroid hormone.

N10233 
Patient's characteristics   
 Age (years)58 ± 1165 ± 8<0.01
 Time on HD (months)109 ± 92113 ± 990.81
 Male (%)61700.45
 Diabetes (%)25300.75
 Current smoking habit (%)20300.45
 Cardiovascular disease (%)924<0.05
 ACEI/ARB (%)33240.47
 Anti-platelet agents (%)29380.52
 Statin (%)370.66
Laboratory data   
 Creatinine (mg/dL)12.9 ± 2.511.1 ± 2.7<0.01
 Calcium (mg/dL)9.0 ± 0.98.9 ± 1.00.44
 Phosphorous (mg/dL)6.1 ± 1.35.9 ± 1.50.45
 Alkaline phosphatase (IU/L)236 ± 102258 ± 1000.29
 Total cholesterol (mg/dL)150 ± 35142 ± 290.23
 Triglycerides (mg/dL)125 ± 91107 ± 440.26
 Albumin (g/L)38 ± 336 ± 3<0.01
 Hemoglobin (g/dL)10.7 ± 1.110.4 ± 1.10.15
 Intact PTH (pg/mL)245 ± 177277 ± 1780.38
 Ferritin (ng/mL)146 ± 88164 ± 820.31
 Hs-CRP (mg/L)3.8 ± 7.75.5 ± 6.40.21
 Kt/Vurea1.20 ± 0.221.22 ± 0.240.76
 nPCR1.00 ± 0.170.96 ± 0.210.18
 BMI (kg/m2)20.9 ± 2.920.2 ± 2.40.28
Atherosclerotic parameters   
 Systolic BP (mm Hg)137 ± 22145 ± 270.11
 Diastolic BP (mm Hg)84 ± 1381 ± 140.27
 Mean BP (mm Hg)103 ± 17111 ± 210.05
 ABI1.10 ± 0.081.08 ± 0.090.14
 baPWV (m/s)15.4 ± 3.118.0 ± 4.5<0.01
 CAVI9.4 ± 2.910.4 ± 3.30.10

Kaplan–Meier analysis revealed that total survival rate was significantly lower in patients with baPWV higher than 16.6 m/s (58.7%, n = 46) when compared with those with baPWV from 14.3 to 16.6 (84.1%, n = 44) and those with baPWV lower than 14.3 (86.7%, n = 45) (P < 0.01) (Fig. 1A). The highest baPWV tertile had a significantly lower free rate from CV death (65.2%) compared with the middle (90.9%) and the lowest tertile (95.6%) (P < 0.01) (Fig. 2A). In contrast, CAVI tertiles did not associate with survival rates (Figs 1B,2B).

Figure 1.

Kaplan–Meier analysis for overall survival according to the levels of brachial-ankle pulse wave velocity (baPWV) and cardio-ankle vascular index (CAVI).

Figure 2.

Survival rate free from cardiovascular mortality according to tertiles of brachial-ankle pulse wave velocity (baPWV) and cardio-ankle vascular index (CAVI).

Predictors of mortality and CVD events (Table 5)

Table 5. Cox hazard analysis for total mortality, cardiovascular (CV) mortality and CV events
Total mortality
 NModel 1Model 2Model 3
HR [95% CI]P-valueHR [95% CI]P-valueHR [95% CI]P-value
baPWV (m/s)       
 baPWV < 14.3451.0 1.0   
 14.3 ≤ baPWV < 16.6441.1 [0.4–3.0]0.940.8 [0.3–2.4]0.73
 baPWV ≥ 16.6463.2 [1.3–7.5]<0.011.7 [0.7–4.7]0.27
CAVI       
 CAVI < 8.0451.0     
 8.0 ≤ CAVI < 9.9461.2 [0.5–3.1]0.68
 CAVI ≥ 9.9442.0 [0.9–4.8]0.11
CV mortality
 NModel 1Model 2Model 3
HR [95% CI]P-valueHR [95% CI]P-valueHR [95% CI]P-value
baPWV (m/s)       
 baPWV < 14.3451.0 1.0 1.0 
 14.3 ≤ baPWV < 16.6442.1 [0.4–11.4]0.941.8 [0.3–10.1]0.517.5 [0.6–95.2]0.12
 baPWV ≥ 16.6469.0 [2.1–40.1]<0.015.6 [1.2–27.4]0.0316.9 [1.1–251.8]0.04
CAVI       
 CAVI < 8.0451.0 1.0   
 8.0 ≤ CAVI < 9.9460.98 [0.28–3.37]0.970.69 [0.28–1.70]0.63
 CAVI ≥ 9.9442.59 [0.91–7.34]0.071.51 [0.47–4.85]0.49
CV events
 Model 1Model 2Model 3
HR [95% CI]P-valueHR [95% CI]P-valueHR [95% CI]P-value
  1. Model 1 includes tertile of brachial-ankle pulse wave velocity (baPWV); and cardio-ankle vascular index (CAVI); model 2 adjusted for age, gender and diabetes mellitus; model 3 further adjusted for time on HD, current smoking habit, usage of angiotensin converting enzyme inhibitor (ACEI) /angiotensin II receptor blocker (ARB), mean arterial pressure, serum creatinine, calcium, phosphorous, albumin, total cholesterol, hemoglobin, intact PTH, log-transformed ferritin. HR, hazard ratio.

baPWV (m/s)      
 baPWV < 14.31.0 1.0 1.0 
 14.3 ≤ baPWV < 16.61.7 [0.6–4.7]0.331.6 [0.5–4.7]0.402.9 [0.7–11.4]0.14
 baPWV ≥ 16.64.0 [1.6–9.9]<0.013.5 [1.2–9.9]0.023.4 [0.7–16.2]0.13
CAVI      
 CAVI < 8.01.0 1.0   
 8.0 ≤ CAVI < 9.91.01 [0.40–2.54]0.990.91 [0.34–2.44]0.86
 CAVI ≥ 9.92.02 [0.90–4.54]0.091.50 [0.59–3.84]0.40

Cox regression analysis not adjusted for risk factors (model 1) revealed that the patients with the highest tertile of baPWV had a significantly greater risk for CV mortalities and CV events when compared with those with the lowest and middle tertiles of baPWV. In contrast, CAVI did not correlate with total and CV mortalities.

When adjusted for age, gender and DM (model 2), the patients with the top tertile of baPWV had a 5.6- and a 3.5-time higher risk for CV mortality and CV events when compared with those with the bottom and middle tertiles of baPWV. The highest tertile of baPWV also became a significant determinant of CV death (HR [95% CI]: 16.9 [1.1–251.8], P < 0.05) when further adjusted for the 12 comorbid risk factors (model 3). The top tertile of baPWV did not relate to CV events (Table 5). However, there was a significantly higher risk for CV events in relation to BP control (e.g. stroke, CVD, congestive HF and others) in patients with the top tertile of baPWV (HR [95% CI]: 34.5 [2.4–493.4], P < 0.01) and those with the middle tertile (HR [95% CI]: 15.0 [2.3–176.4], P < 0.05), respectively.

DISCUSSION

There are several studies providing evidence that baPWV is an independent predictor of morbidity and mortality (22–26). In general population, increased baPWV (≥17 m/s) is associated with a 6.8-fold higher risk of all-cause mortality as compared with those lower than 14 m/s during the 6.5-year follow-up (23). A baPWV higher than 17.3 m/s is also independently associated with poor prognosis in type 2 diabetic patients with coronary artery disease (24). In patients with acute coronary syndrome, baPWV of 18 m/s or higher is a predictor for major CV events (e.g. stroke, re-admission for HF or cardiac death) (25). A baPWV higher than 19.6 m/s is also potently related to progression to commencement of dialysis or death in patients with chronic kidney disease (CKD) stages 3 to 5 (26). In HD patients, baPWV is associated with the number of lacuna infarctions on brain magnetic resonance imaging (27).

Cardio-ankle vascular index is a new index of the overall stiffness of the artery from the origin of the aorta to the ankle (4). CAVI is demonstrated superior to baPWV to predict coronary atherosclerosis in patients who underwent coronary angiography (5–7). CAVI has been demonstrated to be superior to baPWV to assess CV risks in metabolic syndrome (28). In CKD patients, CAVI increases corresponding to decreased estimated glomerular filtration rate (eGFR) (29). An increase in CAVI over time is also observed in HD patients (30). However, there was no study to compare the impact of baPWV and CAVI as a predictor of survival prognosis in dialysis population.

In this study, we showed that a modest increase of baPWV (≥16.6 m/s) was a significant predictor of CV death in HD patients. A higher baPWV was associated with CV events including stroke, congestive HF, sudden death and others. In addition, in patients without any CVD history (n = 117), the top tertile of baPWV was associated with an increased risk of CV death (HR: 10.4, P = 0.08) (data not shown). In contrast, CAVI did not associate clinical outcomes. These findings collectively suggest that baPWV is useful as a surrogate marker of CV events in dialysis population.

It is demonstrated that baPWV and CAVI cannot be estimated properly when arterial occlusion is present at the lower extremities. baPWV is inaccurate when the ABI is reduced lower than 0.95 (31). CAVI becomes also not useful in HD patients with PAD (8). Tanaka et al. (14) demonstrated that baPWV is not a predictor for CV morbidity and mortality when analyzed in combination with ABI. So, in this study, we first excluded patients whose ABIs were lower than 0.9 or greater than 1.3 from analyses. Previous studies (13,15) have demonstrated that increased baPWV is a significant predictor of all-cause and CV mortalities in patients with ABI ranging from 0.9 to 1.3. A small reduction of ABI (<1.0) is also associated with increased risk for CV death (13,21). So, it follows from these findings that baPWV is useful in identifying a high-risk population of HD patients with normal ABI.

Theoretically, there are some possible explanations why CAVI did not relate to clinical outcomes. First, CAVI is calculated from PWV at the origin of the aorta to the ankle portion of the tibial artery, and adjusted by systolic and diastolic BP at the upper brachial artery. In this study, we measured CAVI just before a HD session, so fluid volume excess may preferentially increase systolic BP, and may provoke a greater difference in systolic and diastolic BP at measurement, thereby leading to an underestimation of CAVI value. Second, CAVI is adjusted by the equation of stiffness β, which is essentially applied to a portion of the aorta. We previously demonstrated that CAVI was inversely correlated with calcified area of the aortic wall in HD patients (20), suggesting that advanced calcification may change smooth muscle contractile in the aorta, and falsely decrease CAVI. Third, CAVI is more affected by smoking habit than baPWV (32). In this study, we found that the prevalence of current smoking habit was more prominent in the top tertile of CAVI. So, cigarette smoking habit might affect the CAVI value.

The time points of baPWV and CAVI measurements are somewhat different among HD patients (8–15). We measured immediately before a HD cycle as described previously (9,10), while others performed these at 30 to 60 min during HD session (12), after the completion of HD (8,13), or at non-dialysis day (14,15). A single HD decreases baPWV independently of changes of body fluid volume, while it rather increases CAVI especially in patients with a higher removal rate (>5% of dry weight) (10). The concentrated blood after HD may increase the CAVI, because the density of blood (ρ) is one of the determinants of CAVI. So, the difference in time point could influence the results in HD patients. Further studies will be needed to determine the appropriate time point that predicts clinical outcomes more reliably.

There are several limitations to this study. First, we measured baPWV and CAVI only at enrollment, and thus, we could not assess the relationship between longitudinal changes of both parameters and outcomes. Second, the small number of patients may not provide enough statistical power to avoid the existence of confounding variables. Third, we did not assess the effect of food intake on pulse wave indices. Postprandial state is shown to underestimate baPWV and CAVI in type 2 diabetic patients (33). Increased vascular tone by sympathetic nerve activation also reduces CAVI (34). In this study, about half of the patients had taken dietary food 1 to 2 h before measurement. Although no patient had suffered from excess fluid overload clinically, we cannot exclude a possibility that food intake before measurement may affect the values. Finally, our cut-off value is lower than those in the previous studies (≥18; or ≥20.1 m/s) (13,15). In this study, since no appropriate cut-off point was present in dialysis population, we divided patients into the tertiles of baPWV according to the previous studies (14,15). Our cut-off level (≥16.6 m/s) had a sensitivity of 72.7% and a specificity of 70.3% for predicting CV death. The reasons for this discrepancy may be partly related to the difference in measuring devices. We used CAVI-VaSera VS-1000, while others did form PWV/ABI (Colin Medical Technology, Komaki, Japan) (13–15). In addition, we measured pulse wave indices just before HD, while other studies did after HD session (13) or at non-dialysis day (15).

CONCLUSION

We found that a brachial-ankle pulse wave velocity higher than 16.6 m/s was an independent predictor for cardiovascular death during a mean follow-up of 63 months in hemodialysis patients. A higher baPWV was also associated with the CV events possibly related to blood pressure control. In contrast, CAVI did not associate with any adverse outcomes. These findings suggest that baPWV is superior to CAVI as a predictor of adverse outcomes in HD patients.

Acknowledgment

The results presented have not been published previously in whole or part. There was no conflict to declare in this study.

Ancillary