- Email: [email protected]
Endothelial function and cardiovascular events in chronic kidney disease
International Journal of Cardiology 173 (2014) 481–486
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Endothelial function and cardiovascular events in chronic kidney disease Yoshihiro Hirata a,1, Seigo Sugiyama a,b,⁎,1, Eiichiro Yamamoto a, Yasushi Matsuzawa c, Eiichi Akiyama a,c, Hiroaki Kusaka a, Koichiro Fujisue a, Hirofumi Kurokawa a, Junichi Matsubara a, Koichi Sugamura a, Hirofumi Maeda a, Satomi Iwashita a, Hideaki Jinnouchi d,e, Kunihiko Matsui f, Hisao Ogawa a a
Faculty of Life Sciences, Department of Cardiovascular Medicine, Graduate School of Medical Science, Kumamoto University Kumamoto, Japan Cardiovascular Division Diabetes Care Center, Jinnouchi Hospital Kumamoto, Japan Division of Cardiology, Yokohama City University Medical Center Yokohama, Japan d Diabetes Care Center, Jinnouchi Hospital Kumamoto, Japan e Division of Preventive Cardiology, Department of Cardiovascular Medicine, Kumamoto University Hospital Kumamoto, Japan f Department of General Internal Medicine, Yamaguchi University Hospital Ube, Japan b c
a r t i c l e
i n f o
Article history: Received 11 February 2014 Accepted 12 March 2014 Available online 20 March 2014 Keywords: Endothelial dysfunction Chronic kidney disease Coronary artery disease Cardiovascular events Cohort study
a b s t r a c t Background: As patients with chronic kidney disease (CKD) are at high risk of developing coronary artery disease (CAD), it is important to stratify their cardiovascular risk. We investigated whether peripheral endothelial dysfunction is associated with the presence of CAD in patients with CKD and is a predictor of cardiovascular events. Methods: We enrolled 383 CKD patients with at least one coronary risk factor. Peripheral endothelial function was assessed by reactive hyperemia peripheral arterial tonometry index (RHI). The presence of CAD was determined by coronary angiography. Cardiovascular events were assessed during follow-up. Results: Ln-RHI was significantly lower in risk factor-matched CKD patients (n = 323) than risk factormatched non-CKD patients (n = 323) (0.527 ± 0.192 vs. 0.580 ± 0.218, p = 0.001). In CKD patients (n = 383), Ln-RHI was significantly lower in CAD (0.499 ± 0.183, n = 262) than non-CAD (0.582 ± 0.206, n = 121) (p b0.001) patients. Multivariate logistic regression analysis identified Ln-RHI as an independent factor associated with the presence of CAD (p = 0.001). During a mean follow-up period of 30 months, 90 cardiovascular events were recorded in CKD patients. Multivariate Cox hazard analysis identified low-Ln-RHI as an independent predictor of cardiovascular events (hazard ratio = 2.70, 95% confidence interval = 1.62–4.51, p b 0.001). The predictive value of combined Ln-RHI and Framingham risk score (FRS) was evaluated by net reclassification index (NRI) and C-statistics, which showed significant improvement (NRI = 22%, p b 0.001) (C-statistics: FRS = 0.49, FRS + Ln-RHI = 0.62, p = 0.005). Conclusions: Endothelial function was significantly impaired in CKD patients and correlated with the presence of CAD. Severe endothelial dysfunction was an independent and incremental predictor of cardiovascular events in CKD. © 2014 Elsevier Ireland Ltd. All rights reserved.
Trial Registration: Unique identifier: UMIN000010432. URL: http://www.umin.ac.jp/ctr/. 1. Introduction Patients with chronic kidney disease (CKD) are at increased risk of coronary artery disease (CAD), and CAD is a major cause of death for these patients [1]. The increasing number of CKD patients places an importance on stratifying cardiovascular risks in patients with CKD. While
⁎ Corresponding author at: Faculty of Life Sciences, Department of Cardiovascular Medicine, Graduate School of Medical Science, Kumamoto University, 1-1-1 Honjo, Chuo-ku, Kumamoto 860-8556, Japan. Tel.: +81 96 373 5175; fax: +81 96 362 3256. E-mail address: [email protected] (S. Sugiyama). 1 Yoshihiro Hirata and Seigo Sugiyama contributed equally to the study.
http://dx.doi.org/10.1016/j.ijcard.2014.03.085 0167-5273/© 2014 Elsevier Ireland Ltd. All rights reserved.
patients with CKD are at high risk of cardiovascular events, not all CKD patients are equally predisposed to CAD and cardiovascular complications. Physicians sometimes hesitate to perform coronary angiography to assess CAD in CKD patients because of potential contrast mediainduced kidney dysfunction. Various methods for cardiovascular risk stratification have been proposed in patients with CKD, although the Framingham risk model lack predictive power for CKD patients and there are currently no useful predictive parameters for the future occurrence of cardiovascular events [2,3]. Endothelial dysfunction correlates with cardiovascular diseases and plays an important role in all stages of atherosclerosis, leading to obstructive CAD [4] and cardiovascular events. Furthermore, endothelial dysfunction is implicated in increased frequency of cardiac events, even in the absence of CAD, suggesting that endothelial dysfunction could be involved in any cardiovascular manifestations [5,6]. The role of endothelial dysfunction in CKD has been described [7], but to our
482
Y. Hirata et al. / International Journal of Cardiology 173 (2014) 481–486
knowledge, there is no information on correlations between endothelial dysfunction and CAD/future cardiovascular events in patients with CKD. In this study, we hypothesized that peripheral endothelial dysfunction is associated with the presence of CAD in patients with CKD and investigated the prognostic significance of peripheral endothelial function using reactive hyperemia peripheral arterial tonometry (RH-PAT) examination for risk stratification in stable patients with CKD. 2. Methods 2.1. Study subjects and protocol We performed a prospective cohort study to investigate the clinical significance of peripheral endothelial function in high-risk CKD patients. We recruited 920 consecutive stable patients with at least one coronary risk factor (hypertension, diabetes mellitus [DM], dyslipidemia, current smoking, and family history of CAD) who were referred to Kumamoto University Hospital for coronary angiography (CAG) because of suspected CAD or abnormal electrocardiography and/or echocardiography between August 2006 and January 2013. Patients with at least one coronary risk factor included those with and without CKD for cross-sectional investigation of the effects of peripheral endothelial function on the presence of CKD. The RH-PAT index (RHI) was used to assess peripheral endothelial function and was measured in all study participants before CAG using fingertip RH-PAT by Endo-PAT2000. RHI was further compared between CKD patients and nonCKD patients after matching risk factors: number of patients, age, sex, equal incidence of hypertension, DM, dyslipidemia, and CAD (risk factor-matched CKD patients and risk factor-matched non-CKD patients). Exclusion criteria included patients on hemodialysis, as RH-PAT examination cannot be performed correctly in these patients because of the A-V shunt in the arm. The presence of CAD was determined by CAG. CKD patients were prospectively followed until October 2013 or until the occurrence of cardiovascular events (Fig. 1 and Supplemental Fig. 1). The study protocol conformed to the principles of the Declaration of Helsinki. Written informed consent was obtained from all patients. This study is registered at the University Hospital Medical Information Network (UMIN) Clinical Trials Registry (UMIN000010432).
2.3. RH-PAT examination RH-PAT has been described previously [10]. RH-PAT was conducted in the morning after subjects had fasted, before taking medications and before CAG. Non-invasive RH-PAT was measured with a blood pressure cuff placed on an upper arm (study arm), while the contralateral arm served as a control (control arm). The PAT probe was placed on a finger of each hand. After a 5-min equilibration period, the blood pressure cuff was inflated on the study arm to 60 mmHg above the systolic pressure or 200 mmHg for 5 min, and then the cuff was deflated to induce reactive hyperemia. RH-PAT data were digitally analyzed online (Endo-PAT2000 software, version 3.0.4 and 3.4.4, Itamar Medical Ltd, Caesarea, Israel). RHI was calculated as the ratio of the average amplitude of the PAT signal over 1 min starting 1.5 min after cuff deflation (control arm, A; occluded arm, C) divided by the average amplitude of the PAT signal of a 2.5-min period before cuff inflation (baseline) (control arm, B; occluded arm, D). RHI values were automatically calculated by the online computer based on the ratio of (C/D)/(A/B). Because RHI values are not normally distributed, we calculated the natural logarithmic transformed RHI values as the Ln-RHI for use in regression analyses, as reported previously [11,12]. The reproducibility of RH-PAT technology has been confirmed in previous studies [11–14]. 2.4. Coronary angiography and CAD classification CAD was defined as coronary artery stenosis (N75% narrowing of the arterial diameter) in at least one coronary artery using quantitative coronary angiography. Based on the CAG results, CKD + CAD patients were divided into two groups according to the Gensini score and synergy between percutaneous coronary intervention with TAXUS and cardiac surgery (SYNTAX) score. We divided CKD + CAD patients using a cutoff value for the Gensini score of ≥40, as reported previously [15], and used a cutoff value for the SYNTAX score of ≥12, representing the median value of CKD + CAD patients. 2.5. Follow-up and cardiovascular events After RH-PAT and CAG, CKD patients were prospectively followed at outpatient clinics until October 2013 or until the occurrence of the end point (composite cardiovascular events), defined as cardiovascular death, non-fatal myocardial infarction, unstable angina pectoris, non-fatal ischemic stroke, hospitalization for heart failure decompensation, or coronary revascularization. Based on the RH-PAT results, the mean value of Ln-RHI (Ln-RHI: 0.525) was used to divide CKD patients into the low-Ln-RHI and high-Ln-RHI groups.
2.2. eGFR/urinary protein assessment and definition of CKD
2.6. Statistical analysis
The estimated glomerular filtration ratio (eGFR) was calculated using the Japanese Society of Nephrology formula [8]. Urinary protein was measured semi-quantitatively by a urine dipstick test (Uro-Labstix; Siemens Japan, Tokyo, Japan) (negative and ±: urinary protein b30 mg/dl; 1+: urinary protein 30–99 mg/dl; 2+: urinary protein 100–299 mg/dl; 3+: urinary protein 300–999 mg/dl; and 4+: urinary protein N1000 mg/dl). Proteinuria was defined as urinary protein excretion of N30 mg/dl. CKD was defined as a low eGFR of b60 ml/min/1.73 m2 or a high eGFR of N60 ml/min/1.73 m2 with proteinuria [9]. Non-CKD was defined as a high eGFR of ≥60 ml/min/1.73 m2 without proteinuria. CKD patients were divided into two groups depending on eGFR (mild CKD: eGFR 30 ≤eGFR b 60 ml/min/1.73 m2 or higher eGFR of N60 ml/min/1.73 m2 with proteinuria; moderate CKD: eGFR b30 ml/min/1.73 m2).
Continuous variables with normal distribution are expressed as mean ± standard deviation. Categorical data are presented as frequencies and percentages. The Shapiro–Wilk test was used to evaluate the distribution of continuous data. Skewed variables are expressed as median values (25–75%). Significant clinical parameters associated with CAD in simple logistic regression analysis, and several factors reported previously to correlate significantly with CAD, were entered into multivariate logistic regression analysis. The Hosmer–Lemeshow test for goodness of fit was used for model calibration. We also calculated the cumulative incidence of cardiovascular events using the Kaplan–Meier method and compared such incidence to the log-rank test. To account for confounding variables, a propensity score was calculated for each patient using a logistic regression model in which the dependent variable was low-Ln-RHI (≤0.525). Independent variables
Fig. 1. Flow chart showing the protocol used in the study.
Y. Hirata et al. / International Journal of Cardiology 173 (2014) 481–486 included in the propensity score model were age, sex, body mass index, hypertension, systolic blood pressure, diastolic blood pressure, Ln-B-type natriuretic peptide (BNP), Ln-high-sensitivity C-reactive protein (hs-CRP), eGFR, total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, Ln-triglycerides, dyslipidemia, DM, left ventricular ejection fraction, current smoking, family history of CAD, FRS, Gensini score, SYNTAX score, treatment with aspirin, calcium channel blockers, angiotensinconverting enzyme inhibitors (ACE-I) or angiotensin II receptor blockers (ARB), β-blockers, and hydroxymethylglutaryl coenzyme A reductase inhibitors. The Cox proportional hazard model was used to estimate the cardiovascular event hazard ratio (HR) and its 95% confidence interval (CI) in CKD patients by simple and multivariate analyses with forced inclusion models. The negative predictive value of RH-PAT examination at a cutoff value of Ln-RHI 0.525 was also assessed. Proportional hazards assumption was confirmed using the Schoenfeld test. We also estimated C-statistics for the Cox proportional hazards regression models [16]. C-statistics were compared after the addition of Ln-RHI to FRS. The incremental effects of adding Ln-RHI to FRS to predict cardiovascular events were also evaluated using the net reclassification index (NRI) statistic [17]. A p value of b0.05 was considered statistically significant. An expanded method is available in the online Appendix.
3. Results 3.1. Baseline characteristics of 857 patients with coronary risk factors Sixty-three patients were excluded from the initial 920 patients because of the presence of severe valvular disease (n = 19), severe collagen disease (n = 8), history of malignancy (n = 23), and active infection (n = 13). The final study subjects were 857 stable patients with at least one coronary risk factor (Supplemental Fig. 1). Coronary risk prevalence for hypertension, DM, dyslipidemia, current smoking, and family history of CAD were 88.1%, 39.9%, 80.8%, 12.4%, and 19.1%, respectively. Among the study patients, 474 were classified as non-CKD and 383 were diagnosed as CKD (Fig. 1 and Supplemental Fig. 1). CKD patients were older than non-CKD patients. The Ln-RHI was significantly lower, and hs-CRP, BNP, and FRS were significantly higher in CKD patients compared with non-CKD patients. The proportion of patients with hypertension, DM, or treated with ACE-I or ARB was significantly higher in CKD patients than in non-CKD patients (Supplemental Table 1). 3.2. Baseline characteristics of 383 patients with CKD The baseline characteristics of 383 CKD patients are shown in Table 1. We divided CKD patients into the low-Ln-RHI and high-Ln-RHI groups using the mean value of Ln-RHI (Ln-RHI: 0.525). BNP was significantly higher in low-Ln-RHI patients than high-Ln-RHI patients. In patients with CKD (n = 383), 262 patients had CAD. The CAD + CKD patients comprised mainly men and were older and had DM,
483
dyslipidemia, and high FRS compared with the non-CAD + CKD group. The proportion of patients treated with aspirin, β-blockers, ACE-I/ARB, and HMG-CoA reductase inhibitors was significantly higher in CAD than non-CAD patients (Supplemental Table 1). In CAD + CKD patients, the Gensini score was 41.3 ± 35.1 and the SYNTAX score was 13.5 ± 9.1. 3.3. Endothelial function and CKD Ln-RHI was significantly lower in CKD patients (0.525 ± 0.194, n = 383) compared with non-CKD patients (0.588 ± 0.216, n = 474, p b 0.001) (Supplemental Fig. 2A), and Ln-RHI was significantly lower in risk factor-matched CKD patients (0.527 ± 0.192, n = 323) than risk factor-matched non-CKD patients (0.580 ± 0.218, n = 323, p = 0.001) (Supplemental Table 2, Supplemental Fig. 2B). Ln-RHI significantly decreased with increased CKD severity (non-CKD: 0.588 ± 0.216, mild CKD: 0.528 ± 0.192, moderate CKD: 0.491 ± 0.222, ANOVA, p b 0.001) (Supplemental Fig. 2C). 3.4. Endothelial function and CAD in patients with/without CKD Ln-RHI was significantly lower in CAD + non-CKD patients (0.574 ± 0.212, n = 321) compared with non-CAD + non-CKD patients (0.617 ± 0.222, n = 153, p = 0.04) (Supplemental Fig. 3A). Ln-RHI was significantly lower in CAD + CKD patients (0.499 ± 0.183, n = 262) compared with non-CAD + CKD patients (0.582 ± 0.206, n = 121, p b 0.001) (Supplemental Fig. 3B). Simple logistic regression analysis demonstrated that age, sex, DM, dyslipidemia, and Ln-RHI correlated significantly with the presence of CAD in patients with CKD. Multivariate logistic regression analysis, including significant factors in simple regression, identified LnRHI as an independent and significant factor associated with the presence of CAD in patients with CKD (per 0.1; odds ratio [OR] = 0.81, 95% CI = 0.72–0.91, p = 0.001) (Supplemental Table 3). FRS correlated significantly with the presence of CAD in CKD patients (per 1.0; OR = 1.04, 95% CI = 1.01–1.08, p = 0.007). Multivariate logistic regression analysis with FRS and Ln-RHI identified Ln-RHI as an independent factor associated with the presence of CAD (per 0.1; OR = 0.81, 95% CI = 0.72–0.90, p b 0.001). Receiver-operating characteristics analysis showed that Ln-RHI correlated significantly with the presence of CAD (area under the curve [AUC] = 0.647, 95% CI = 0.583–0.710, p b 0.001). FRS non-significantly correlated with the presence of CAD in CKD patients (AUC = 0.520, 95% CI = 0.451–0.589, p =
Table 1 Baseline characteristics of 383 CKD patients.
Age (years) Sex (male, %) BMI (kg/m2) Hypertension (yes, %) DM (yes, %) Dyslipidemia (yes, %) Current smoking (yes, %) Family history of IHD (yes, %) Ln-RHI LVEF (%) Hs-CRP (mg/dl) BNP (pg/ml) eGFR (ml/min/1.73 m2) Aspirin (%) β-blockers (%) ACE-I or ARB (%) CCB (%) HMG-CoA RIs (%)
All CKD patients n = 383
High-Ln-RHI patients n = 167
Low-Ln-RHI patients n = 216
p value*
72.0 (9.5) 64.2 24.3 (3.6) 90.6 44.9 83.0 7.8 18.0 0.525 (0.194) 62.1 (7.4) 0.08 (0.04–0.17) 49.4 (20.6–99.6) 49.4 (12.9) 80.6 62.0 72.0 67.8 81.9
72.2 (9.0) 62.3 24.3 (3.7) 93.4 44.3 82.6 5.4 19.2 0.701 (0.140) 62.2 (7.1) 0.07 (0.03–0.14) 42.9 (18.7–85.0) 50.1 (12.1) 77.8 61.1 72.5 71.3 80.8
71.8 (10.0) 65.7 24.2 (3.6) 88.4 45.4 83.3 9.7 17.1 0.389 (0.097) 61.9 (7.6) 0.08 (0.04–0.21) 56.0 (23.4–107.5) 48.9 (13.5) 82.8 62.8 71.6 65.1 82.8
0.66 0.52 0.83 0.11 0.92 0.89 0.13 0.69 b0.001 0.69 0.24 0.005 0.33 0.24 0.75 0.91 0.23 0.69
Data are mean (standard deviation), median (25th to 75th percentile range), or numbers (percentages). *Between high-Ln-RHI patients and low-Ln-RHI patients. CKD: chronic kidney disease; CAD: coronary artery disease; BMI: body mass index; DM: diabetes mellitus; IHD: ischemic heart diseases; RHI: reactive hyperemia index; LVEF: left ventricular ejection fraction; Hs-CRP: high-sensitivity C-reactive protein; BNP: b-type natriuretic peptide; eGFR: estimated glomerular filtration rate; ACE-I: angiotensin-converting enzyme inhibitor; ARB: angiotensin II receptor blocker; CCB: calcium channel blocker; HMG-CoA RIs: hydroxy methylglutaryl coenzyme A reductase inhibitors.
484
Y. Hirata et al. / International Journal of Cardiology 173 (2014) 481–486
0.51). The correlation between Ln-RHI + FRS and the presence of CAD was more significant (AUC = 0.653, 95% CI = 0.590–0.717, p =0.003) than FRS alone (Supplemental Fig. 4). 3.5. Endothelial function and coronary atherosclerosis in CAD + CKD patients Ln-RHI was significantly lower in the high-Gensini score group (Gensini score ≥ 40: 0.447 ± 0.195) compared with the low-Gensini score group (0.532 ± 0.168, p b 0.001) (Supplemental Table 4, Supplemental Fig. 3C). Analysis of CAD + CKD patient data by multivariate logistic regression analysis, including significant factors in simple logistic regression, identified Ln-RHI as an independent and significant factor associated with the severity of coronary atherosclerotic burden assessed by a high Gensini score (per 0.1; OR = 0.76, 95% CI = 0.65–0.89, p = 0.001) (Supplemental Table 5). Ln-RHI was significantly lower in the high-SYNTAX score group (≥ 12) (0.433 ± 0.169) compared with the low-SYNTAX score group (0.564 ± 0.174, p b 0.001) (Supplemental Table 6, Supplemental Fig. 3D). Multivariate logistic regression analysis of CAD + CKD patient data, including significant factors in simple logistic regression, identified Ln-RHI as an independent and significant factor associated with complex coronary atherosclerosis lesions assessed by a high SYNTAX score (per 0.1; OR = 0.62, 95% CI = 0.51–0.74, p b 0.001) (Supplemental Table 7). 3.6. Follow-up cardiovascular events Data from 383 patients with CKD were available for analysis of cardiovascular events. There was no loss to follow-up. The mean follow-up period was 30 ± 19 months. Ninety cardiovascular events were recorded in patients with CKD during the follow-up period. Cardiovascular event details are summarized in Supplemental Table 8. Kaplan–Meier analysis demonstrated a significantly higher probability of cardiovascular events in the low-Ln-RHI group compared with the high-Ln-RHI group (cutoff value of Ln-RHI = 0.525, log-rank test p b 0.001) (Fig. 2A). Patients with cardiovascular events during the follow-up period had significantly lower Ln-RHI than patients without cardiovascular events (0.451 ± 0.172 vs. 0.548 ± 0.195, p b 0.001) (Fig. 2B). 3.7. Cox proportional hazard analysis for cardiovascular events The results of simple and multivariate Cox proportional hazard analyses for cardiovascular events are summarized in Table 2. Multivariate Cox hazard analysis, including significant predictors in simple Cox hazard analysis and various established coronary risk factors, identified low-Ln-RHI as a significant and independent predictor of cardiovascular events (HR: 2.70, 95% CI = 1.62–4.51, p b 0.001) (Table 2). Low-Ln-RHI was also significantly associated with cardiovascular events in the model with propensity score adjustment (HR: 2.09, 95% CI = 1.21–3.61, p = 0.008). Using the model with FRS and Ln-RHI, low-LnRHI was identified as an independent predictor of cardiovascular events (HR: 2.93, 95% CI = 1.77–4.87, p b 0.001). The negative predictive value of RH-PAT examination for the occurrence of future cardiovascular events was 88.6% at Ln-RHI 0.525. 3.8. C-statistics and NRI for Cox proportional hazard model to predict cardiovascular events We estimated C-statistics of the FRS alone and FRS + Ln-RHI. A significant incremental change was observed in C-statistics when Ln-RHI was added to FRS (C-statistics, 95% CI = FRS 0.49 [0.41–0.57], FRS + Ln-RHI 0.62 [0.55–0.69], p = 0.005) (Table 3). The Schoefield test indicated that the proportional hazards assumptions were appropriate (FRS: p = 0.12, FRS + Ln-RHI: p = 0.15). We also
Fig. 2. Follow-up analysis. (A) Kaplan–Meier analysis demonstrated a significantly higher probability of cardiovascular events in the low-Ln-RHI group compared with the high-LnRHI group. (B) Ln-RHI was significantly lower in patients with cardiovascular events than without cardiovascular events.
confirmed good calibration for the analysis using Grønnesby and Borgan statistics (FRS: p = 0.21, FRS + Ln-RHI: p = 0.71). We also evaluated the use of Ln-RHI for CKD patient classification for the prediction of cardiovascular events using the NRI statistic. The predictive value of Ln-RHI + FRS was evaluated by NRI. The NRI was significant with the inclusion of Ln-RHI (NRI = 22%, p b 0.001) (Table 3). 4. Discussion The study findings were as follows: (1) Ln-RHI was significantly lower in patients with CKD compared with non-CKD; (2) in patients with CKD, Ln-RHI was significantly lower in CAD patients compared with non-CAD patients; (3) multivariate logistic regression analysis identified that Ln-RHI independently and significantly correlated with CAD; and (4) peripheral endothelial dysfunction was a significant predictor of cardiovascular events, and the combined use of Ln-RHI and FRS improved risk classification of CKD patients, as indicated by the NRI and a significant increase in C-statistics. These results indicate that the non-invasive assessment of endothelial dysfunction by RH-PAT examination may provide information on incremental risk stratification to identify vulnerable CKD patients. Cardiovascular disease is a major cause of death in CKD patients [18]; thus, the risk stratification of these high-risk CKD patients is clinically important. Classic cardiovascular risk factors such as FRS cannot adequately assess risk stratification in CKD patients [2,3]. In CKD patients, there was currently no useful predictive parameter for the future occurrence of cardiovascular events, and no previous report on
Y. Hirata et al. / International Journal of Cardiology 173 (2014) 481–486
485
Table 2 Cox proportional hazards analysis for cardiovascular events in CKD patients. Variable
Age (per years) Sex (male, yes) BMI (per kg/m2) Hypertension (yes) DM (yes) Dyslipidemia (yes) Current smoking (yes) Family history of IHD (yes) Low-Ln-RHI (≤0.525) (yes) LVEF (per %) Ln-Hs-CRP (per 1) Ln-BNP (per 1) eGFR (per 1)
Simple regression
Multiple regression with significant factors in simple analysis and previously reported risk factors
HR
95% CI
p value
HR
95% CI
p value
1.03 0.75 0.99 0.97 1.03 0.90 1.32 1.29 2.94 0.98 1.14 1.61 0.98
1.01–1.06 0.49–1.14 0.93–1.04 0.49–1.93 0.78–1.56 0.53–1.52 0.64–2.73 0.77–2.17 1.77–4.88 0.96–1.01 0.99–1.31 1.32–1.96 0.97–0.99
0.02 0.18 0.61 0.93 0.89 0.69 0.46 0.33 b0.001 0.13 0.07 b0.001 0.03
1.02 – – 1.25 0.92 0.95 – – 2.70 – 1.11 1.48 0.99
0.99–1.04 – – 0.61–2.56 0.60–1.40 0.55–1.63 – – 1.62–4.51 – 0.96–1.27 1.20–1.81 0.98–1.01
0.18 – – 0.54 0.69 0.84 – – b0.001 – 0.16 b0.001 0.49
HR: hazard ratio; CI: confidence interval. Other abbreviations as in Table 1.
correlation between peripheral endothelial dysfunction and future cardiovascular events. In this study, we firstly reported that peripheral endothelial function assessed by RH-PAT was a significant and independent predictor of future cardiovascular events in CKD patients. Our results demonstrated that low-Ln-RHI can predict cardiovascular events, and the combination of Ln-RHI and FRS was more powerful at predicting cardiovascular events. The negative predictive value of RH-PAT examination with a cutoff value of Ln-RHI 0.525 was 88.6%, which suggests that CKD patients who have Ln-RHI more than 0.525 could avoid future cardiovascular events with a higher probability. The non-invasive information of endothelial function assessed by RH-PAT would provide clinically useful information for risk stratification with improved risk discrimination in CKD patients. The results also showed that endothelial dysfunction, as assessed by Ln-RHI, was independently associated with the severity and complexity of coronary atherosclerosis in CAD + CKD patients using Gensini and SYNTAX scores. It is possible that severe and complex coronary atherosclerosis assessed by CAG contributes to the poor cardiovascular prognosis in patients with low-Ln-RHI and CKD. The assessment of endothelial function by RH-PAT could potentially suggest non-invasive information for the assessment of the severity and complexity of coronary atherosclerosis. Further studies are needed to evaluate the benefits and drawbacks of the non-invasive cardiovascular risk assessment of CKD patients using RH-PAT test compared with invasive examination using contrast media.
The use of contrast medium is necessary to evaluate the presence of CAD and coronary atherosclerotic stenosis, although it may be detrimental to renal function in patients with CKD. Non-invasive RH-PAT could provide valuable risk stratification to identify vulnerable patients with severe and complex CAD and subsequent cardiovascular events in CKD patients. It is currently difficult to improve renal function, to induce regression of coronary stenosis plaques, and to change plaque morphology. Since endothelial dysfunction can be reversed or improved by lifestyle modifications [19] and medical therapies [20], intensive and multidisciplinary therapies may improve endothelial function [21,22], which might be beneficial for patients with CKD. Decreased nitric oxide (NO) levels are closely related to endothelial dysfunction. NO is synthesized from L-arginine by NO synthesis (NOS). Asymmetric dimethylarginine (ADMA) is an inhibitor of NOS, and recently, Sesti G et al. [23] reported that excess ADMA causes endothelial dysfunction by reducing NO availability and increases microvascular damage in CKD patients. Lack of tetrahydrobiopterin, which is a cofactor of NOS, leads to endothelial NOS uncoupling and produces superoxide. These processes may also be related to endothelial dysfunction in CKD patients. ADMA-lowering and tetrahydrobiopterin supplementation may be a therapeutic target in vascular diseases [24]; however, we did not investigate these molecules in the present study. Further experimental and clinical studies would be necessary to determine the molecular mechanisms of endothelial dysfunction in CKD patients. Our study has several limitations. First, it was a one-center design with a small patient population. Second, all patients were referred for
Table 3 Models to predict cardiovascular events. C-statistics for Cox proportional hazard model to predict cardiovascular events NRI models to predict cardiovascular events Risk category by FRS
Classification by FRS alone
New risk category using FRS + Ln-RHI
CKD patients (n = 383) FRS FRS + Ln-RHI
C-statistics (95% CI) 0.49 (0.41–0.57) 0.62 (0.55–0.69)
Increment in C-statistics
p value
0.13 (0.04–0.22)
0.005
0 0 293
0 0 24
0 0 96
0 0 173
0 0 90
0 0 3
0 0 14
0 0 73
Low risk
Intermediate risk
High risk
Patients without cardiovascular events Low risk Intermediate risk High risk Patients with cardiovascular risk Low risk Intermediate risk High risk
FRS: Framingham risk score; NRI: net reclassification index. Other abbreviations as in Table 1.
486
Y. Hirata et al. / International Journal of Cardiology 173 (2014) 481–486
a coronary angiography, which may present a selection bias. Third, the number of moderate CKD patients (eGFR b 30 ml/min/1.73 m2) was small (n = 30) to avoid complications of renal dysfunction by coronary angiography in the study protocol. Fourth, endothelial dysfunction in CKD patients may be influenced by the severity and duration of hypertension and DM, and CKD patients are considered to have more severe and longer duration of hypertension and DM than non-CKD patients. We had insufficient objective information on the severity and duration of hypertension and DM. Sixth, cardiovascular events included heart failure. CKD patients tend to show a fluid overload because of CKD itself, which can lead to heart failure. It is possible that CKD itself influenced the occurrence of heart failure, although we think heart failure hospitalization is an important clinical cardiovascular event in CKD patients and peripheral endothelial dysfunction could be associated with heart failure occurrence [12]. Further multicenter studies involving larger numbers of patients are necessary. In conclusion, this study demonstrated that peripheral endothelial dysfunction as assessed by RH-PAT correlated significantly with the presence of CAD and advanced coronary atherosclerosis in stable CKD patients. Endothelial dysfunction was identified as an independent predictor of cardiovascular events in CKD patients. Identifying high-risk CKD patients with endothelial dysfunction by RH-PAT could provide benefits for incremental risk stratification. Acknowledgments This work was supported by Grants-in Aid for Scientific Research (grant number C25461086 to S. Sugiyama, grant number B24790770 to E. Yamamoto and grant number B25860610 to J. Matsubara) from the Japanese Ministry of Education, Science, and Culture; and a grant from the West Japan Vascular Function Society. Conflict of interestDr. Ogawa received lecture fees and research grant from Astellas, AstraZeneca, Bayer, Boehringer Ingelheim, Chugai, Daiichi Sankyo, Dainippon Sumitomo Pharma, Eisai, Kowa, Kyowa Hakko Kirin, Mitsubishi Tanabe, MSD, Novartis, Otsuka, Pfizer, Sanofi, Sionogi, Takeda and Mochida. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ijcard.2014.03.085. References [1] Tonelli M, Muntner P, Lloyd A, et al. Risk of coronary events in people with chronic kidney disease compared with those with diabetes: a population-level cohort study. Lancet 2012;380(9844):807–14.
[2] Weiner DE, Tighiouart H, Elsayed EF, et al. The Framingham predictive instrument in chronic kidney disease. J Am Coll Cardiol 2007;50(3):217–24. [3] Chen SC, Su HM, Tsai YC, et al. Framingham risk score with cardiovascular events in chronic kidney disease. PLoS One 2013;8(3):e60008. [4] Ross R. Atherosclerosis—an inflammatory disease. N Engl J Med 1999;340(2):115–26. [5] Suwaidi JA, Hamasaki S, Higano ST, et al. Long-term follow-up of patients with mild coronary artery disease and endothelial dysfunction. Circulation 2000;101(9):948–54. [6] Rubinshtein R, Kuvin JT, Soffler M, et al. Assessment of endothelial function by noninvasive peripheral arterial tonometry predicts late cardiovascular adverse events. Eur Heart J 2010;31(9):1142–8. [7] Moody WE, Edwards NC, Madhani M, et al. Endothelial dysfunction and cardiovascular disease in early-stage chronic kidney disease: cause or association? Atherosclerosis 2012;223(1):86–94. [8] Matsuo S, Imai E, Horio M, et al. Revised equations for estimated GFR from serum creatinine in Japan. Am J Kidney Dis 2009;53(6):982–92. [9] K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39(2 Suppl. 1):S1–S266. [10] Bonetti PO, Pumper GM, Higano ST, et al. Noninvasive identification of patients with early coronary atherosclerosis by assessment of digital reactive hyperemia. J Am Coll Cardiol 2004;44(11):2137–41. [11] Matsuzawa Y, Sugiyama S, Sugamura K, et al. Digital assessment of endothelial function and ischemic heart disease in women. J Am Coll Cardiol 2010;55(16):1688–96. [12] Akiyama E, Sugiyama S, Matsuzawa Y, et al. Incremental prognostic significance of peripheral endothelial dysfunction in patients with heart failure with normal left ventricular ejection fraction. J Am Coll Cardiol 2012;60(18):1778–86. [13] Salamet Tierney ES, Newburger JW, Gauvreau K, et al. Endothelial pulse amplitude testing: feasibility and reproductibility in adolescents. J Pediatr 2009;154(6):901–5. [14] Bonetti PO, Barsness GW, keelan PC, et al. Enhanced external counterpulsation improves endothelial function in patients with symptomatic coronary artery disease. J Am Coll Cardiol 2003;41(10):1761–8. [15] Murata M, Matsubara Y, Kawano K, et al. Coronary artery disease and polymorphisms in a receptor mediating shear stress-dependent platelet activation. Circulation 1997;96(10):3281–6. [16] Pencina MJ, D'Agostino RB. Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Stat Med 2004;23(13):2109–23. [17] Pencina MJ, D'Agostino Sr RB, D'Agostino Jr RB, et al. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27(2):157–72. [18] Sarnak MJ, Levey AS, Schoolwerth AC, et al. Kidney disease as a risk factor for development of cardiovascular disease: a statement from the American Heart Association Councils on Kidney in Cardiovascular Disease, High Blood Pressure Research, Clinical Cardiology, and Epidemiology and Prevention. Circulation 2003;108(17):2154–69. [19] Matsuzawa Y, Sugiyama S, Sugamura K, et al. Successful diet and exercise therapy as evaluated on self-assessment score significantly improves endothelial function in metabolic syndrome patients. Circ J 2013;77(11):2807–15. [20] Tonetti MS, D'Aiuto F, Nibali L, et al. Treatment of periodontitis and endothelial function. N Engl J Med 2007;356(9):911–20. [21] Tian J, Wang J, Li Y, et al. Endothelial function in patients with newly diagnosed type 2 diabetes receiving early intensive insulin therapy. Am J Hypertens 2012;25(12):1242–8. [22] Matsubara J, Sugiyama S, Akiyama E, et al. Dipeptidyl peptidase-4 inhibitor, sitagliptin, improves endothelial dysfunction in association with its antiinflammatory effects in patients with coronary artery disease and uncontrolled diabetes. Circ J 2013;77(5):1337–44. [23] Sesti G, Mannino GC, De Lorenzo C, et al. A functional variant of the dimethylarginine dimethylaminohydrolase-2 gene is associated with chronic kidney disease. Atherosclerosis 2013;231(1):141–4. [24] Rochette L, Lorin J, Zeller M, et al. Nitric oxide synthase inhibition and oxidative stress in cardiovascular diseases: possible therapeutic targets? Pharmacol Ther 2013;140(3):239–57.
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Endothelial function and cardiovascular events in chronic kidney disease Yoshihiro Hirata a,1, Seigo Sugiyama a,b,⁎,1, Eiichiro Yamamoto a, Yasushi Matsuzawa c, Eiichi Akiyama a,c, Hiroaki Kusaka a, Koichiro Fujisue a, Hirofumi Kurokawa a, Junichi Matsubara a, Koichi Sugamura a, Hirofumi Maeda a, Satomi Iwashita a, Hideaki Jinnouchi d,e, Kunihiko Matsui f, Hisao Ogawa a a
Faculty of Life Sciences, Department of Cardiovascular Medicine, Graduate School of Medical Science, Kumamoto University Kumamoto, Japan Cardiovascular Division Diabetes Care Center, Jinnouchi Hospital Kumamoto, Japan Division of Cardiology, Yokohama City University Medical Center Yokohama, Japan d Diabetes Care Center, Jinnouchi Hospital Kumamoto, Japan e Division of Preventive Cardiology, Department of Cardiovascular Medicine, Kumamoto University Hospital Kumamoto, Japan f Department of General Internal Medicine, Yamaguchi University Hospital Ube, Japan b c
a r t i c l e
i n f o
Article history: Received 11 February 2014 Accepted 12 March 2014 Available online 20 March 2014 Keywords: Endothelial dysfunction Chronic kidney disease Coronary artery disease Cardiovascular events Cohort study
a b s t r a c t Background: As patients with chronic kidney disease (CKD) are at high risk of developing coronary artery disease (CAD), it is important to stratify their cardiovascular risk. We investigated whether peripheral endothelial dysfunction is associated with the presence of CAD in patients with CKD and is a predictor of cardiovascular events. Methods: We enrolled 383 CKD patients with at least one coronary risk factor. Peripheral endothelial function was assessed by reactive hyperemia peripheral arterial tonometry index (RHI). The presence of CAD was determined by coronary angiography. Cardiovascular events were assessed during follow-up. Results: Ln-RHI was significantly lower in risk factor-matched CKD patients (n = 323) than risk factormatched non-CKD patients (n = 323) (0.527 ± 0.192 vs. 0.580 ± 0.218, p = 0.001). In CKD patients (n = 383), Ln-RHI was significantly lower in CAD (0.499 ± 0.183, n = 262) than non-CAD (0.582 ± 0.206, n = 121) (p b0.001) patients. Multivariate logistic regression analysis identified Ln-RHI as an independent factor associated with the presence of CAD (p = 0.001). During a mean follow-up period of 30 months, 90 cardiovascular events were recorded in CKD patients. Multivariate Cox hazard analysis identified low-Ln-RHI as an independent predictor of cardiovascular events (hazard ratio = 2.70, 95% confidence interval = 1.62–4.51, p b 0.001). The predictive value of combined Ln-RHI and Framingham risk score (FRS) was evaluated by net reclassification index (NRI) and C-statistics, which showed significant improvement (NRI = 22%, p b 0.001) (C-statistics: FRS = 0.49, FRS + Ln-RHI = 0.62, p = 0.005). Conclusions: Endothelial function was significantly impaired in CKD patients and correlated with the presence of CAD. Severe endothelial dysfunction was an independent and incremental predictor of cardiovascular events in CKD. © 2014 Elsevier Ireland Ltd. All rights reserved.
Trial Registration: Unique identifier: UMIN000010432. URL: http://www.umin.ac.jp/ctr/. 1. Introduction Patients with chronic kidney disease (CKD) are at increased risk of coronary artery disease (CAD), and CAD is a major cause of death for these patients [1]. The increasing number of CKD patients places an importance on stratifying cardiovascular risks in patients with CKD. While
⁎ Corresponding author at: Faculty of Life Sciences, Department of Cardiovascular Medicine, Graduate School of Medical Science, Kumamoto University, 1-1-1 Honjo, Chuo-ku, Kumamoto 860-8556, Japan. Tel.: +81 96 373 5175; fax: +81 96 362 3256. E-mail address: [email protected] (S. Sugiyama). 1 Yoshihiro Hirata and Seigo Sugiyama contributed equally to the study.
http://dx.doi.org/10.1016/j.ijcard.2014.03.085 0167-5273/© 2014 Elsevier Ireland Ltd. All rights reserved.
patients with CKD are at high risk of cardiovascular events, not all CKD patients are equally predisposed to CAD and cardiovascular complications. Physicians sometimes hesitate to perform coronary angiography to assess CAD in CKD patients because of potential contrast mediainduced kidney dysfunction. Various methods for cardiovascular risk stratification have been proposed in patients with CKD, although the Framingham risk model lack predictive power for CKD patients and there are currently no useful predictive parameters for the future occurrence of cardiovascular events [2,3]. Endothelial dysfunction correlates with cardiovascular diseases and plays an important role in all stages of atherosclerosis, leading to obstructive CAD [4] and cardiovascular events. Furthermore, endothelial dysfunction is implicated in increased frequency of cardiac events, even in the absence of CAD, suggesting that endothelial dysfunction could be involved in any cardiovascular manifestations [5,6]. The role of endothelial dysfunction in CKD has been described [7], but to our
482
Y. Hirata et al. / International Journal of Cardiology 173 (2014) 481–486
knowledge, there is no information on correlations between endothelial dysfunction and CAD/future cardiovascular events in patients with CKD. In this study, we hypothesized that peripheral endothelial dysfunction is associated with the presence of CAD in patients with CKD and investigated the prognostic significance of peripheral endothelial function using reactive hyperemia peripheral arterial tonometry (RH-PAT) examination for risk stratification in stable patients with CKD. 2. Methods 2.1. Study subjects and protocol We performed a prospective cohort study to investigate the clinical significance of peripheral endothelial function in high-risk CKD patients. We recruited 920 consecutive stable patients with at least one coronary risk factor (hypertension, diabetes mellitus [DM], dyslipidemia, current smoking, and family history of CAD) who were referred to Kumamoto University Hospital for coronary angiography (CAG) because of suspected CAD or abnormal electrocardiography and/or echocardiography between August 2006 and January 2013. Patients with at least one coronary risk factor included those with and without CKD for cross-sectional investigation of the effects of peripheral endothelial function on the presence of CKD. The RH-PAT index (RHI) was used to assess peripheral endothelial function and was measured in all study participants before CAG using fingertip RH-PAT by Endo-PAT2000. RHI was further compared between CKD patients and nonCKD patients after matching risk factors: number of patients, age, sex, equal incidence of hypertension, DM, dyslipidemia, and CAD (risk factor-matched CKD patients and risk factor-matched non-CKD patients). Exclusion criteria included patients on hemodialysis, as RH-PAT examination cannot be performed correctly in these patients because of the A-V shunt in the arm. The presence of CAD was determined by CAG. CKD patients were prospectively followed until October 2013 or until the occurrence of cardiovascular events (Fig. 1 and Supplemental Fig. 1). The study protocol conformed to the principles of the Declaration of Helsinki. Written informed consent was obtained from all patients. This study is registered at the University Hospital Medical Information Network (UMIN) Clinical Trials Registry (UMIN000010432).
2.3. RH-PAT examination RH-PAT has been described previously [10]. RH-PAT was conducted in the morning after subjects had fasted, before taking medications and before CAG. Non-invasive RH-PAT was measured with a blood pressure cuff placed on an upper arm (study arm), while the contralateral arm served as a control (control arm). The PAT probe was placed on a finger of each hand. After a 5-min equilibration period, the blood pressure cuff was inflated on the study arm to 60 mmHg above the systolic pressure or 200 mmHg for 5 min, and then the cuff was deflated to induce reactive hyperemia. RH-PAT data were digitally analyzed online (Endo-PAT2000 software, version 3.0.4 and 3.4.4, Itamar Medical Ltd, Caesarea, Israel). RHI was calculated as the ratio of the average amplitude of the PAT signal over 1 min starting 1.5 min after cuff deflation (control arm, A; occluded arm, C) divided by the average amplitude of the PAT signal of a 2.5-min period before cuff inflation (baseline) (control arm, B; occluded arm, D). RHI values were automatically calculated by the online computer based on the ratio of (C/D)/(A/B). Because RHI values are not normally distributed, we calculated the natural logarithmic transformed RHI values as the Ln-RHI for use in regression analyses, as reported previously [11,12]. The reproducibility of RH-PAT technology has been confirmed in previous studies [11–14]. 2.4. Coronary angiography and CAD classification CAD was defined as coronary artery stenosis (N75% narrowing of the arterial diameter) in at least one coronary artery using quantitative coronary angiography. Based on the CAG results, CKD + CAD patients were divided into two groups according to the Gensini score and synergy between percutaneous coronary intervention with TAXUS and cardiac surgery (SYNTAX) score. We divided CKD + CAD patients using a cutoff value for the Gensini score of ≥40, as reported previously [15], and used a cutoff value for the SYNTAX score of ≥12, representing the median value of CKD + CAD patients. 2.5. Follow-up and cardiovascular events After RH-PAT and CAG, CKD patients were prospectively followed at outpatient clinics until October 2013 or until the occurrence of the end point (composite cardiovascular events), defined as cardiovascular death, non-fatal myocardial infarction, unstable angina pectoris, non-fatal ischemic stroke, hospitalization for heart failure decompensation, or coronary revascularization. Based on the RH-PAT results, the mean value of Ln-RHI (Ln-RHI: 0.525) was used to divide CKD patients into the low-Ln-RHI and high-Ln-RHI groups.
2.2. eGFR/urinary protein assessment and definition of CKD
2.6. Statistical analysis
The estimated glomerular filtration ratio (eGFR) was calculated using the Japanese Society of Nephrology formula [8]. Urinary protein was measured semi-quantitatively by a urine dipstick test (Uro-Labstix; Siemens Japan, Tokyo, Japan) (negative and ±: urinary protein b30 mg/dl; 1+: urinary protein 30–99 mg/dl; 2+: urinary protein 100–299 mg/dl; 3+: urinary protein 300–999 mg/dl; and 4+: urinary protein N1000 mg/dl). Proteinuria was defined as urinary protein excretion of N30 mg/dl. CKD was defined as a low eGFR of b60 ml/min/1.73 m2 or a high eGFR of N60 ml/min/1.73 m2 with proteinuria [9]. Non-CKD was defined as a high eGFR of ≥60 ml/min/1.73 m2 without proteinuria. CKD patients were divided into two groups depending on eGFR (mild CKD: eGFR 30 ≤eGFR b 60 ml/min/1.73 m2 or higher eGFR of N60 ml/min/1.73 m2 with proteinuria; moderate CKD: eGFR b30 ml/min/1.73 m2).
Continuous variables with normal distribution are expressed as mean ± standard deviation. Categorical data are presented as frequencies and percentages. The Shapiro–Wilk test was used to evaluate the distribution of continuous data. Skewed variables are expressed as median values (25–75%). Significant clinical parameters associated with CAD in simple logistic regression analysis, and several factors reported previously to correlate significantly with CAD, were entered into multivariate logistic regression analysis. The Hosmer–Lemeshow test for goodness of fit was used for model calibration. We also calculated the cumulative incidence of cardiovascular events using the Kaplan–Meier method and compared such incidence to the log-rank test. To account for confounding variables, a propensity score was calculated for each patient using a logistic regression model in which the dependent variable was low-Ln-RHI (≤0.525). Independent variables
Fig. 1. Flow chart showing the protocol used in the study.
Y. Hirata et al. / International Journal of Cardiology 173 (2014) 481–486 included in the propensity score model were age, sex, body mass index, hypertension, systolic blood pressure, diastolic blood pressure, Ln-B-type natriuretic peptide (BNP), Ln-high-sensitivity C-reactive protein (hs-CRP), eGFR, total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, Ln-triglycerides, dyslipidemia, DM, left ventricular ejection fraction, current smoking, family history of CAD, FRS, Gensini score, SYNTAX score, treatment with aspirin, calcium channel blockers, angiotensinconverting enzyme inhibitors (ACE-I) or angiotensin II receptor blockers (ARB), β-blockers, and hydroxymethylglutaryl coenzyme A reductase inhibitors. The Cox proportional hazard model was used to estimate the cardiovascular event hazard ratio (HR) and its 95% confidence interval (CI) in CKD patients by simple and multivariate analyses with forced inclusion models. The negative predictive value of RH-PAT examination at a cutoff value of Ln-RHI 0.525 was also assessed. Proportional hazards assumption was confirmed using the Schoenfeld test. We also estimated C-statistics for the Cox proportional hazards regression models [16]. C-statistics were compared after the addition of Ln-RHI to FRS. The incremental effects of adding Ln-RHI to FRS to predict cardiovascular events were also evaluated using the net reclassification index (NRI) statistic [17]. A p value of b0.05 was considered statistically significant. An expanded method is available in the online Appendix.
3. Results 3.1. Baseline characteristics of 857 patients with coronary risk factors Sixty-three patients were excluded from the initial 920 patients because of the presence of severe valvular disease (n = 19), severe collagen disease (n = 8), history of malignancy (n = 23), and active infection (n = 13). The final study subjects were 857 stable patients with at least one coronary risk factor (Supplemental Fig. 1). Coronary risk prevalence for hypertension, DM, dyslipidemia, current smoking, and family history of CAD were 88.1%, 39.9%, 80.8%, 12.4%, and 19.1%, respectively. Among the study patients, 474 were classified as non-CKD and 383 were diagnosed as CKD (Fig. 1 and Supplemental Fig. 1). CKD patients were older than non-CKD patients. The Ln-RHI was significantly lower, and hs-CRP, BNP, and FRS were significantly higher in CKD patients compared with non-CKD patients. The proportion of patients with hypertension, DM, or treated with ACE-I or ARB was significantly higher in CKD patients than in non-CKD patients (Supplemental Table 1). 3.2. Baseline characteristics of 383 patients with CKD The baseline characteristics of 383 CKD patients are shown in Table 1. We divided CKD patients into the low-Ln-RHI and high-Ln-RHI groups using the mean value of Ln-RHI (Ln-RHI: 0.525). BNP was significantly higher in low-Ln-RHI patients than high-Ln-RHI patients. In patients with CKD (n = 383), 262 patients had CAD. The CAD + CKD patients comprised mainly men and were older and had DM,
483
dyslipidemia, and high FRS compared with the non-CAD + CKD group. The proportion of patients treated with aspirin, β-blockers, ACE-I/ARB, and HMG-CoA reductase inhibitors was significantly higher in CAD than non-CAD patients (Supplemental Table 1). In CAD + CKD patients, the Gensini score was 41.3 ± 35.1 and the SYNTAX score was 13.5 ± 9.1. 3.3. Endothelial function and CKD Ln-RHI was significantly lower in CKD patients (0.525 ± 0.194, n = 383) compared with non-CKD patients (0.588 ± 0.216, n = 474, p b 0.001) (Supplemental Fig. 2A), and Ln-RHI was significantly lower in risk factor-matched CKD patients (0.527 ± 0.192, n = 323) than risk factor-matched non-CKD patients (0.580 ± 0.218, n = 323, p = 0.001) (Supplemental Table 2, Supplemental Fig. 2B). Ln-RHI significantly decreased with increased CKD severity (non-CKD: 0.588 ± 0.216, mild CKD: 0.528 ± 0.192, moderate CKD: 0.491 ± 0.222, ANOVA, p b 0.001) (Supplemental Fig. 2C). 3.4. Endothelial function and CAD in patients with/without CKD Ln-RHI was significantly lower in CAD + non-CKD patients (0.574 ± 0.212, n = 321) compared with non-CAD + non-CKD patients (0.617 ± 0.222, n = 153, p = 0.04) (Supplemental Fig. 3A). Ln-RHI was significantly lower in CAD + CKD patients (0.499 ± 0.183, n = 262) compared with non-CAD + CKD patients (0.582 ± 0.206, n = 121, p b 0.001) (Supplemental Fig. 3B). Simple logistic regression analysis demonstrated that age, sex, DM, dyslipidemia, and Ln-RHI correlated significantly with the presence of CAD in patients with CKD. Multivariate logistic regression analysis, including significant factors in simple regression, identified LnRHI as an independent and significant factor associated with the presence of CAD in patients with CKD (per 0.1; odds ratio [OR] = 0.81, 95% CI = 0.72–0.91, p = 0.001) (Supplemental Table 3). FRS correlated significantly with the presence of CAD in CKD patients (per 1.0; OR = 1.04, 95% CI = 1.01–1.08, p = 0.007). Multivariate logistic regression analysis with FRS and Ln-RHI identified Ln-RHI as an independent factor associated with the presence of CAD (per 0.1; OR = 0.81, 95% CI = 0.72–0.90, p b 0.001). Receiver-operating characteristics analysis showed that Ln-RHI correlated significantly with the presence of CAD (area under the curve [AUC] = 0.647, 95% CI = 0.583–0.710, p b 0.001). FRS non-significantly correlated with the presence of CAD in CKD patients (AUC = 0.520, 95% CI = 0.451–0.589, p =
Table 1 Baseline characteristics of 383 CKD patients.
Age (years) Sex (male, %) BMI (kg/m2) Hypertension (yes, %) DM (yes, %) Dyslipidemia (yes, %) Current smoking (yes, %) Family history of IHD (yes, %) Ln-RHI LVEF (%) Hs-CRP (mg/dl) BNP (pg/ml) eGFR (ml/min/1.73 m2) Aspirin (%) β-blockers (%) ACE-I or ARB (%) CCB (%) HMG-CoA RIs (%)
All CKD patients n = 383
High-Ln-RHI patients n = 167
Low-Ln-RHI patients n = 216
p value*
72.0 (9.5) 64.2 24.3 (3.6) 90.6 44.9 83.0 7.8 18.0 0.525 (0.194) 62.1 (7.4) 0.08 (0.04–0.17) 49.4 (20.6–99.6) 49.4 (12.9) 80.6 62.0 72.0 67.8 81.9
72.2 (9.0) 62.3 24.3 (3.7) 93.4 44.3 82.6 5.4 19.2 0.701 (0.140) 62.2 (7.1) 0.07 (0.03–0.14) 42.9 (18.7–85.0) 50.1 (12.1) 77.8 61.1 72.5 71.3 80.8
71.8 (10.0) 65.7 24.2 (3.6) 88.4 45.4 83.3 9.7 17.1 0.389 (0.097) 61.9 (7.6) 0.08 (0.04–0.21) 56.0 (23.4–107.5) 48.9 (13.5) 82.8 62.8 71.6 65.1 82.8
0.66 0.52 0.83 0.11 0.92 0.89 0.13 0.69 b0.001 0.69 0.24 0.005 0.33 0.24 0.75 0.91 0.23 0.69
Data are mean (standard deviation), median (25th to 75th percentile range), or numbers (percentages). *Between high-Ln-RHI patients and low-Ln-RHI patients. CKD: chronic kidney disease; CAD: coronary artery disease; BMI: body mass index; DM: diabetes mellitus; IHD: ischemic heart diseases; RHI: reactive hyperemia index; LVEF: left ventricular ejection fraction; Hs-CRP: high-sensitivity C-reactive protein; BNP: b-type natriuretic peptide; eGFR: estimated glomerular filtration rate; ACE-I: angiotensin-converting enzyme inhibitor; ARB: angiotensin II receptor blocker; CCB: calcium channel blocker; HMG-CoA RIs: hydroxy methylglutaryl coenzyme A reductase inhibitors.
484
Y. Hirata et al. / International Journal of Cardiology 173 (2014) 481–486
0.51). The correlation between Ln-RHI + FRS and the presence of CAD was more significant (AUC = 0.653, 95% CI = 0.590–0.717, p =0.003) than FRS alone (Supplemental Fig. 4). 3.5. Endothelial function and coronary atherosclerosis in CAD + CKD patients Ln-RHI was significantly lower in the high-Gensini score group (Gensini score ≥ 40: 0.447 ± 0.195) compared with the low-Gensini score group (0.532 ± 0.168, p b 0.001) (Supplemental Table 4, Supplemental Fig. 3C). Analysis of CAD + CKD patient data by multivariate logistic regression analysis, including significant factors in simple logistic regression, identified Ln-RHI as an independent and significant factor associated with the severity of coronary atherosclerotic burden assessed by a high Gensini score (per 0.1; OR = 0.76, 95% CI = 0.65–0.89, p = 0.001) (Supplemental Table 5). Ln-RHI was significantly lower in the high-SYNTAX score group (≥ 12) (0.433 ± 0.169) compared with the low-SYNTAX score group (0.564 ± 0.174, p b 0.001) (Supplemental Table 6, Supplemental Fig. 3D). Multivariate logistic regression analysis of CAD + CKD patient data, including significant factors in simple logistic regression, identified Ln-RHI as an independent and significant factor associated with complex coronary atherosclerosis lesions assessed by a high SYNTAX score (per 0.1; OR = 0.62, 95% CI = 0.51–0.74, p b 0.001) (Supplemental Table 7). 3.6. Follow-up cardiovascular events Data from 383 patients with CKD were available for analysis of cardiovascular events. There was no loss to follow-up. The mean follow-up period was 30 ± 19 months. Ninety cardiovascular events were recorded in patients with CKD during the follow-up period. Cardiovascular event details are summarized in Supplemental Table 8. Kaplan–Meier analysis demonstrated a significantly higher probability of cardiovascular events in the low-Ln-RHI group compared with the high-Ln-RHI group (cutoff value of Ln-RHI = 0.525, log-rank test p b 0.001) (Fig. 2A). Patients with cardiovascular events during the follow-up period had significantly lower Ln-RHI than patients without cardiovascular events (0.451 ± 0.172 vs. 0.548 ± 0.195, p b 0.001) (Fig. 2B). 3.7. Cox proportional hazard analysis for cardiovascular events The results of simple and multivariate Cox proportional hazard analyses for cardiovascular events are summarized in Table 2. Multivariate Cox hazard analysis, including significant predictors in simple Cox hazard analysis and various established coronary risk factors, identified low-Ln-RHI as a significant and independent predictor of cardiovascular events (HR: 2.70, 95% CI = 1.62–4.51, p b 0.001) (Table 2). Low-Ln-RHI was also significantly associated with cardiovascular events in the model with propensity score adjustment (HR: 2.09, 95% CI = 1.21–3.61, p = 0.008). Using the model with FRS and Ln-RHI, low-LnRHI was identified as an independent predictor of cardiovascular events (HR: 2.93, 95% CI = 1.77–4.87, p b 0.001). The negative predictive value of RH-PAT examination for the occurrence of future cardiovascular events was 88.6% at Ln-RHI 0.525. 3.8. C-statistics and NRI for Cox proportional hazard model to predict cardiovascular events We estimated C-statistics of the FRS alone and FRS + Ln-RHI. A significant incremental change was observed in C-statistics when Ln-RHI was added to FRS (C-statistics, 95% CI = FRS 0.49 [0.41–0.57], FRS + Ln-RHI 0.62 [0.55–0.69], p = 0.005) (Table 3). The Schoefield test indicated that the proportional hazards assumptions were appropriate (FRS: p = 0.12, FRS + Ln-RHI: p = 0.15). We also
Fig. 2. Follow-up analysis. (A) Kaplan–Meier analysis demonstrated a significantly higher probability of cardiovascular events in the low-Ln-RHI group compared with the high-LnRHI group. (B) Ln-RHI was significantly lower in patients with cardiovascular events than without cardiovascular events.
confirmed good calibration for the analysis using Grønnesby and Borgan statistics (FRS: p = 0.21, FRS + Ln-RHI: p = 0.71). We also evaluated the use of Ln-RHI for CKD patient classification for the prediction of cardiovascular events using the NRI statistic. The predictive value of Ln-RHI + FRS was evaluated by NRI. The NRI was significant with the inclusion of Ln-RHI (NRI = 22%, p b 0.001) (Table 3). 4. Discussion The study findings were as follows: (1) Ln-RHI was significantly lower in patients with CKD compared with non-CKD; (2) in patients with CKD, Ln-RHI was significantly lower in CAD patients compared with non-CAD patients; (3) multivariate logistic regression analysis identified that Ln-RHI independently and significantly correlated with CAD; and (4) peripheral endothelial dysfunction was a significant predictor of cardiovascular events, and the combined use of Ln-RHI and FRS improved risk classification of CKD patients, as indicated by the NRI and a significant increase in C-statistics. These results indicate that the non-invasive assessment of endothelial dysfunction by RH-PAT examination may provide information on incremental risk stratification to identify vulnerable CKD patients. Cardiovascular disease is a major cause of death in CKD patients [18]; thus, the risk stratification of these high-risk CKD patients is clinically important. Classic cardiovascular risk factors such as FRS cannot adequately assess risk stratification in CKD patients [2,3]. In CKD patients, there was currently no useful predictive parameter for the future occurrence of cardiovascular events, and no previous report on
Y. Hirata et al. / International Journal of Cardiology 173 (2014) 481–486
485
Table 2 Cox proportional hazards analysis for cardiovascular events in CKD patients. Variable
Age (per years) Sex (male, yes) BMI (per kg/m2) Hypertension (yes) DM (yes) Dyslipidemia (yes) Current smoking (yes) Family history of IHD (yes) Low-Ln-RHI (≤0.525) (yes) LVEF (per %) Ln-Hs-CRP (per 1) Ln-BNP (per 1) eGFR (per 1)
Simple regression
Multiple regression with significant factors in simple analysis and previously reported risk factors
HR
95% CI
p value
HR
95% CI
p value
1.03 0.75 0.99 0.97 1.03 0.90 1.32 1.29 2.94 0.98 1.14 1.61 0.98
1.01–1.06 0.49–1.14 0.93–1.04 0.49–1.93 0.78–1.56 0.53–1.52 0.64–2.73 0.77–2.17 1.77–4.88 0.96–1.01 0.99–1.31 1.32–1.96 0.97–0.99
0.02 0.18 0.61 0.93 0.89 0.69 0.46 0.33 b0.001 0.13 0.07 b0.001 0.03
1.02 – – 1.25 0.92 0.95 – – 2.70 – 1.11 1.48 0.99
0.99–1.04 – – 0.61–2.56 0.60–1.40 0.55–1.63 – – 1.62–4.51 – 0.96–1.27 1.20–1.81 0.98–1.01
0.18 – – 0.54 0.69 0.84 – – b0.001 – 0.16 b0.001 0.49
HR: hazard ratio; CI: confidence interval. Other abbreviations as in Table 1.
correlation between peripheral endothelial dysfunction and future cardiovascular events. In this study, we firstly reported that peripheral endothelial function assessed by RH-PAT was a significant and independent predictor of future cardiovascular events in CKD patients. Our results demonstrated that low-Ln-RHI can predict cardiovascular events, and the combination of Ln-RHI and FRS was more powerful at predicting cardiovascular events. The negative predictive value of RH-PAT examination with a cutoff value of Ln-RHI 0.525 was 88.6%, which suggests that CKD patients who have Ln-RHI more than 0.525 could avoid future cardiovascular events with a higher probability. The non-invasive information of endothelial function assessed by RH-PAT would provide clinically useful information for risk stratification with improved risk discrimination in CKD patients. The results also showed that endothelial dysfunction, as assessed by Ln-RHI, was independently associated with the severity and complexity of coronary atherosclerosis in CAD + CKD patients using Gensini and SYNTAX scores. It is possible that severe and complex coronary atherosclerosis assessed by CAG contributes to the poor cardiovascular prognosis in patients with low-Ln-RHI and CKD. The assessment of endothelial function by RH-PAT could potentially suggest non-invasive information for the assessment of the severity and complexity of coronary atherosclerosis. Further studies are needed to evaluate the benefits and drawbacks of the non-invasive cardiovascular risk assessment of CKD patients using RH-PAT test compared with invasive examination using contrast media.
The use of contrast medium is necessary to evaluate the presence of CAD and coronary atherosclerotic stenosis, although it may be detrimental to renal function in patients with CKD. Non-invasive RH-PAT could provide valuable risk stratification to identify vulnerable patients with severe and complex CAD and subsequent cardiovascular events in CKD patients. It is currently difficult to improve renal function, to induce regression of coronary stenosis plaques, and to change plaque morphology. Since endothelial dysfunction can be reversed or improved by lifestyle modifications [19] and medical therapies [20], intensive and multidisciplinary therapies may improve endothelial function [21,22], which might be beneficial for patients with CKD. Decreased nitric oxide (NO) levels are closely related to endothelial dysfunction. NO is synthesized from L-arginine by NO synthesis (NOS). Asymmetric dimethylarginine (ADMA) is an inhibitor of NOS, and recently, Sesti G et al. [23] reported that excess ADMA causes endothelial dysfunction by reducing NO availability and increases microvascular damage in CKD patients. Lack of tetrahydrobiopterin, which is a cofactor of NOS, leads to endothelial NOS uncoupling and produces superoxide. These processes may also be related to endothelial dysfunction in CKD patients. ADMA-lowering and tetrahydrobiopterin supplementation may be a therapeutic target in vascular diseases [24]; however, we did not investigate these molecules in the present study. Further experimental and clinical studies would be necessary to determine the molecular mechanisms of endothelial dysfunction in CKD patients. Our study has several limitations. First, it was a one-center design with a small patient population. Second, all patients were referred for
Table 3 Models to predict cardiovascular events. C-statistics for Cox proportional hazard model to predict cardiovascular events NRI models to predict cardiovascular events Risk category by FRS
Classification by FRS alone
New risk category using FRS + Ln-RHI
CKD patients (n = 383) FRS FRS + Ln-RHI
C-statistics (95% CI) 0.49 (0.41–0.57) 0.62 (0.55–0.69)
Increment in C-statistics
p value
0.13 (0.04–0.22)
0.005
0 0 293
0 0 24
0 0 96
0 0 173
0 0 90
0 0 3
0 0 14
0 0 73
Low risk
Intermediate risk
High risk
Patients without cardiovascular events Low risk Intermediate risk High risk Patients with cardiovascular risk Low risk Intermediate risk High risk
FRS: Framingham risk score; NRI: net reclassification index. Other abbreviations as in Table 1.
486
Y. Hirata et al. / International Journal of Cardiology 173 (2014) 481–486
a coronary angiography, which may present a selection bias. Third, the number of moderate CKD patients (eGFR b 30 ml/min/1.73 m2) was small (n = 30) to avoid complications of renal dysfunction by coronary angiography in the study protocol. Fourth, endothelial dysfunction in CKD patients may be influenced by the severity and duration of hypertension and DM, and CKD patients are considered to have more severe and longer duration of hypertension and DM than non-CKD patients. We had insufficient objective information on the severity and duration of hypertension and DM. Sixth, cardiovascular events included heart failure. CKD patients tend to show a fluid overload because of CKD itself, which can lead to heart failure. It is possible that CKD itself influenced the occurrence of heart failure, although we think heart failure hospitalization is an important clinical cardiovascular event in CKD patients and peripheral endothelial dysfunction could be associated with heart failure occurrence [12]. Further multicenter studies involving larger numbers of patients are necessary. In conclusion, this study demonstrated that peripheral endothelial dysfunction as assessed by RH-PAT correlated significantly with the presence of CAD and advanced coronary atherosclerosis in stable CKD patients. Endothelial dysfunction was identified as an independent predictor of cardiovascular events in CKD patients. Identifying high-risk CKD patients with endothelial dysfunction by RH-PAT could provide benefits for incremental risk stratification. Acknowledgments This work was supported by Grants-in Aid for Scientific Research (grant number C25461086 to S. Sugiyama, grant number B24790770 to E. Yamamoto and grant number B25860610 to J. Matsubara) from the Japanese Ministry of Education, Science, and Culture; and a grant from the West Japan Vascular Function Society. Conflict of interestDr. Ogawa received lecture fees and research grant from Astellas, AstraZeneca, Bayer, Boehringer Ingelheim, Chugai, Daiichi Sankyo, Dainippon Sumitomo Pharma, Eisai, Kowa, Kyowa Hakko Kirin, Mitsubishi Tanabe, MSD, Novartis, Otsuka, Pfizer, Sanofi, Sionogi, Takeda and Mochida. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ijcard.2014.03.085. References [1] Tonelli M, Muntner P, Lloyd A, et al. Risk of coronary events in people with chronic kidney disease compared with those with diabetes: a population-level cohort study. Lancet 2012;380(9844):807–14.
[2] Weiner DE, Tighiouart H, Elsayed EF, et al. The Framingham predictive instrument in chronic kidney disease. J Am Coll Cardiol 2007;50(3):217–24. [3] Chen SC, Su HM, Tsai YC, et al. Framingham risk score with cardiovascular events in chronic kidney disease. PLoS One 2013;8(3):e60008. [4] Ross R. Atherosclerosis—an inflammatory disease. N Engl J Med 1999;340(2):115–26. [5] Suwaidi JA, Hamasaki S, Higano ST, et al. Long-term follow-up of patients with mild coronary artery disease and endothelial dysfunction. Circulation 2000;101(9):948–54. [6] Rubinshtein R, Kuvin JT, Soffler M, et al. Assessment of endothelial function by noninvasive peripheral arterial tonometry predicts late cardiovascular adverse events. Eur Heart J 2010;31(9):1142–8. [7] Moody WE, Edwards NC, Madhani M, et al. Endothelial dysfunction and cardiovascular disease in early-stage chronic kidney disease: cause or association? Atherosclerosis 2012;223(1):86–94. [8] Matsuo S, Imai E, Horio M, et al. Revised equations for estimated GFR from serum creatinine in Japan. Am J Kidney Dis 2009;53(6):982–92. [9] K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39(2 Suppl. 1):S1–S266. [10] Bonetti PO, Pumper GM, Higano ST, et al. Noninvasive identification of patients with early coronary atherosclerosis by assessment of digital reactive hyperemia. J Am Coll Cardiol 2004;44(11):2137–41. [11] Matsuzawa Y, Sugiyama S, Sugamura K, et al. Digital assessment of endothelial function and ischemic heart disease in women. J Am Coll Cardiol 2010;55(16):1688–96. [12] Akiyama E, Sugiyama S, Matsuzawa Y, et al. Incremental prognostic significance of peripheral endothelial dysfunction in patients with heart failure with normal left ventricular ejection fraction. J Am Coll Cardiol 2012;60(18):1778–86. [13] Salamet Tierney ES, Newburger JW, Gauvreau K, et al. Endothelial pulse amplitude testing: feasibility and reproductibility in adolescents. J Pediatr 2009;154(6):901–5. [14] Bonetti PO, Barsness GW, keelan PC, et al. Enhanced external counterpulsation improves endothelial function in patients with symptomatic coronary artery disease. J Am Coll Cardiol 2003;41(10):1761–8. [15] Murata M, Matsubara Y, Kawano K, et al. Coronary artery disease and polymorphisms in a receptor mediating shear stress-dependent platelet activation. Circulation 1997;96(10):3281–6. [16] Pencina MJ, D'Agostino RB. Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Stat Med 2004;23(13):2109–23. [17] Pencina MJ, D'Agostino Sr RB, D'Agostino Jr RB, et al. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27(2):157–72. [18] Sarnak MJ, Levey AS, Schoolwerth AC, et al. Kidney disease as a risk factor for development of cardiovascular disease: a statement from the American Heart Association Councils on Kidney in Cardiovascular Disease, High Blood Pressure Research, Clinical Cardiology, and Epidemiology and Prevention. Circulation 2003;108(17):2154–69. [19] Matsuzawa Y, Sugiyama S, Sugamura K, et al. Successful diet and exercise therapy as evaluated on self-assessment score significantly improves endothelial function in metabolic syndrome patients. Circ J 2013;77(11):2807–15. [20] Tonetti MS, D'Aiuto F, Nibali L, et al. Treatment of periodontitis and endothelial function. N Engl J Med 2007;356(9):911–20. [21] Tian J, Wang J, Li Y, et al. Endothelial function in patients with newly diagnosed type 2 diabetes receiving early intensive insulin therapy. Am J Hypertens 2012;25(12):1242–8. [22] Matsubara J, Sugiyama S, Akiyama E, et al. Dipeptidyl peptidase-4 inhibitor, sitagliptin, improves endothelial dysfunction in association with its antiinflammatory effects in patients with coronary artery disease and uncontrolled diabetes. Circ J 2013;77(5):1337–44. [23] Sesti G, Mannino GC, De Lorenzo C, et al. A functional variant of the dimethylarginine dimethylaminohydrolase-2 gene is associated with chronic kidney disease. Atherosclerosis 2013;231(1):141–4. [24] Rochette L, Lorin J, Zeller M, et al. Nitric oxide synthase inhibition and oxidative stress in cardiovascular diseases: possible therapeutic targets? Pharmacol Ther 2013;140(3):239–57.