Study Population

We prospectively recruited 251 participants through advertisement among patients receiving hemodialysis for at least 3 months at two Japanese hospitals (Hidaka Hospital and Gunma University Hospital). Patients were hemodynamically stable, and hemodialysis was performed 3 times weekly (3–5 hours/day). Patients with > mild left heart valvular heart diseases ( n = 10) and limited image quality ( n = 7) were excluded, leaving 234 patients for the final analysis. Among the 234 participants, 150 subjects (64%) also participated in another prospective observational study focusing on novel blood biomarkers in dialysis patients. The unique insights from this present study compared with the previous one are summarized in the Discussion section of this article. The study protocol was approved by the institutional medical ethics committee of the two hospitals, and written informed consent was obtained from all patients.

Data Collection

We collected data about patients’ demographic characteristics and clinical variables related to the delivery of hemodialysis from the hospital charts. Bloods were sampled before starting dialysis sessions. A recently validated 2-year mortality risk score in the hemodialysis cohort (the second Analyzing Data, Recognizing Excellence and Optimizing Outcomes cohort score or AROii score) was used as a clinical risk parameter. This score was calculated from 13 clinical and laboratory parameters (age, smoking status, body mass index, cardiovascular disease history, cancer history, chronic kidney disease etiology, vascular assess, blood flow rates, and levels of hemoglobin, ferritin, C-reactive protein, serum albumin, and creatinine).

Cardiovascular Function Analysis

Echocardiographic examinations were performed using commercially available ultrasound systems. Because loading conditions can change during an interdialytic interval, all participants were studied just before a hemodialysis session after the 1-day interdialytic interval. The LV end-diastolic (EDV) and end-systolic (ESV) volumes and EF were determined from the biplane method of disks. Stroke volume (SV) was also determined from the biplane method of discs because 14% of subjects had prominent sigmoid septum, which would affect LV outflow Doppler profile. Left atrial (LA) volume was calculated using the biapical area-length method. LV volume and mass and LA volume were subsequently indexed by body surface area. The early filling (E-wave), the peak late diastolic (A-wave) velocities, and deceleration time were assessed from transmitral flow. The peak systolic (S′), early diastolic (E′), and late diastolic (A′) mitral annular velocities were measured at septal annulus using spectral Doppler. All measures represent the mean of measurements from three beats for subjects in atrial fibrillation.

The following hemodynamic parameters were estimated noninvasively. End-systolic blood pressure (BP; 0.9 × systolic BP) was calculated as described elsewhere. Total afterload was defined by the Ea (0.9 × systolic BP/SV). SVR index (SVRI), the nonpulsatile component of afterload, was determined by calculating mean BP/cardiac index × 79.9. Arterial compliance was estimated by the ratio of SV to pulse pressure. LV elastance was assessed using the modified single-beat method Ees (determined from BP, SV, and preejection and total systolic periods determined on LV outflow Doppler, EF, and an estimated normalized ventricular elastance at arterial end-diastole). Single-beat PRSW was determined from stroke work/[EDV − k × EDV + (1 − k ) × LV wall volume, where stroke work = mean BP × SV, LV wall volume = LV mass/1.05, and k was assumed to be 0.7. V ₀ , the intercept of the end-systolic pressure volume relationship at an end-systolic pressure of 0 mmHg, was also estimated from Ees, end-systolic BP, and ESV. Ventricular-arterial coupling was assessed using the coupling ratio (Ea/Ees).

In a subgroup of 207 patients (88% of the whole population) who had DICOM-format echo movies, myocardial deformation was also assessed as global longitudinal strain (GLS) using vendor-independent two-dimensional speckle-tracking software (TomTec Imaging Systems, Unterschleissheim, Germany) as described elsewhere. The LV endocardial borders were traced manually in the apical four- and two-chamber views. GLS was obtained by averaging peak longitudinal strains from the four- and two-chamber views. According to the new definition from the standardization task force, we reported the absolute values of GLS in this manuscript and tables in order to avoid unnecessary confusion. An experienced echocardiologist, who was unaware of patients’ data, analyzed all echocardiographic measurements.

The reproducibility of Ees, Ea, Ea/Ees, and GLS was assessed in 15 randomly selected patients. An intra- and interobserver agreement was evaluated after the same observer and another experienced reader repeated the analysis using intraclass correlation coefficients (ICCs) and coefficient of variation.

Endpoint and Follow-Up

Our primary endpoint was a composite of all-cause death, nonfatal myocardial infarction, and hospitalization due to worsening heart failure (HF). The secondary endpoint was a composite of cardiovascular death, nonfatal myocardial infarction, and hospitalization for HF. These endpoints were manually collected from the medical records and adjudicated by experienced cardiologists. Diagnosis of acute myocardial infarction was defined as a cardiac event accompanied by typical electrocardiographic changes and elevation of cardiac biomarkers of myocardial necrosis (cardiac troponin I). HF was defined as dyspnea and pulmonary edema on a chest radiograph requiring extrahemodialysis and hospitalization. Patient follow-up was initiated on the day of echocardiographic examination. When patients did not develop these endpoints, they were censored on October 31, 2015.

Statistical Analysis

All continuous variables are presented as mean ± SD or median with interquartile range based on their distributions. Event-free survival for each of the ventricular-arterial coupling parameters was assessed using Kaplan-Meier analysis and compared with a log-rank test, where each index was dichotomized according to the median of its distribution. The prognostic value was determined using the Cox proportional hazards model. Nonnormally distributed data were log-transformed when put into the models. Some echocardiographic parameters are derived from one another, physiologically interdependent, and correlated with each other, and we placed one echocardiographic parameter in a single multivariable model to avoid multicollinearity. In addition, due to the limited number of events ( n = 30), the independence and robustness of each echocardiographic parameter were assessed using three different multivariable models (model 1, adjusted for age and sex; model 2, adjusted for AROii score and atrial fibrillation; and model 3, adjusted for AROii score and prior HF admission), limiting to put up to three variables at once, to avoid over-fitting statistical models. The proportional hazards assumption, assessed by the goodness-of-fit approach for each independent variable, was not violated. The incremental prognostic value was assessed by comparing −2 log likelihood values of the models with and without the parameter to χ ² distribution at degree of freedom of 1. In addition, because all independent variables do not necessarily share the same level of prognostic information, the decision-tree model was created to see whether there are any hierarchical relationships among these variables. For this, conditional inference trees method of classification and regression tree (CART) analysis was used, whereby patients were split into binary groups with the highest contrast for the composite endpoints. Input variables were AROii score, EF, S′, E/E′, GLS, and Ea/Ees. In each level of the tree, the variable with the strongest relationship to the endpoint was selected and the optimal cutoff was used. Two-sided P < .05 was accepted as indicating statistical significance. All data were statistically analyzed using SPSS version 23.0 (SPSS Inc., Chicago, IL) and R version 3.2.3. (R Foundation for Statistical Computing, Vienna, Austria) with the “party” package.

Clinical Characteristics

During a median follow-up of 776 days (interquartile range, 701–964), the composite endpoint occurred in 30 patients including 21 all-cause deaths, one nonfatal myocardial infarction, and eight HF hospitalizations ( Supplemental Table 1 , available at www.onlinejase.com ). Comparisons of baseline clinical characteristics between the groups are shown in Table 1 . Patients with the composite outcomes had lower body mass index and tended to be older without statistical significance. There were no significant differences in dialysis vintage and the proportion of causes of end-stage renal disease between the groups. Patients with adverse events had a higher prevalence of prior HF hospitalization and greater AROii scores than those without. Usage of antiplatelet drugs was more frequent in patients with the outcomes. Levels of serum albumin were significantly lower and levels of C-reactive protein were higher in patients with the adverse events.

Cardiovascular Functions

Table 2 depicts comparisons of cardiovascular functions between patients with and without the outcomes. BPs including pulse pressure were similar between the groups, while heart rates were significantly higher in patients who had adverse events. Although there was no significant difference in LV mass index, EDV index (EDVI), and ESV index (ESVI) between the groups, patients with adverse outcomes had significantly lower SV index and EF than those without, despite EFs being preserved. LV contractility was reduced in patients with adverse events, as evidenced by lower PRSW, GLS, and S′ velocity. Ees was elevated in both groups, and it tended to be lower in patients with adverse events. Regarding other factors affecting EF, Ea was significantly higher in patients with the adverse events while EDVI, SVRI, and arterial compliance were similar between the groups. Patients with adverse events had significantly higher E-waves and E/e′ ratios than those without, whereas LA volume did not differ between the groups. For integrated parameters, patients with adverse events had higher Ea/Ees ratios than those without events, but it was still <1.0. Although most study patients preserved EF (≥50%), and the prevalence of patients with preserved EF did not differ between the groups, GLS was reduced from normal in both groups (14.4% ± 4.8% and 16.7% ± 4.9%, respectively).

The intraobserver ICCs were as follows: Ees, 0.99; Ea, 0.89; Ea/Ees, 0.91; and GLS, 0.94. The corresponding interobserver ICCs were 0.98, 0.94, 0.98, and 0.76, respectively. The coefficient of variation for intraobserver variabilities were as follows: Ees, 5.6%; Ea, 10.9%; Ea/Ees, 10.5%; and GLS, 5.7%. The corresponding interobserver variabilities were 4.3%, 7.2%, 4.3%, and 12.1%, respectively.

Survival Analysis

Table 3 summarizes univariable and multivariable Cox proportional hazard analyses of echocardiographic parameter. Worse EF was a significant predictor of the primary endpoint ( Figure 1 ) and remained an independent predictor in each of the uni- and multivariable Cox models ( Table 3 ). Regarding physiologic determinants of EF, contractility parameters (reduced PRSW, Ees, and S′) were significantly associated with adverse events throughout the models, while EDVI and the afterload parameters (Ea, SVRI, and arterial compliance) did not show significant associations with adverse outcomes. In diastolic functional indices, the E/E′ ratio was associated with adverse events in each of the multivariable models, however, the LA volume index was not in any of them. Among the other integrated parameters, SV and Ea/Ees ratio were significant predictors of the adverse events ( Figure 1 ), where patients with above the median levels of the Ea/Ees ratio had a threefold increased risk of adverse outcomes compared with those who had below-median levels (hazard ratio [HR], 3.83; 95% CI, [1.64–8.95]; P = .002). In addition, both the AROii score and prior HF hospitalization remained independent predictors of the adverse events in each model (models 2 and 3).

Table 3

Univariable and multivariable Cox proportional hazard models for prediction of adverse events

Variables	Unadjusted		Model 1		Model 2		Model 3
	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value
LV volume and mass
Ln EDV index	0.91 (0.25–3.27)	.88	1.08 (0.28–4.15)	.91	1.46 (0.38–5.55)	.58	0.85 (0.24–2.98)	.80
LV mass index, 1 g/m 2	1.00 (0.99–1.01)	.68	1.00 (0.99–1.02)	.55	1.01 (0.99–1.02)	.37	1.01 (0.99–1.02)	.25
Arterial afterload
Ln Ea	2.10 (0.69–6.40)	.19	2.52 (0.76–8.38)	.13	1.92 (0.55–6.76)	.31	2.16 (0.61–7.61)	.23
Ln SVRI	0.84 (0.22–3.26)	.80	0.89 (0.23–3.42)	.86	0.99 (0.25–3.86)	.99	1.30 (0.32–5.34)	.72
Ln arterial compliance	0.91 (0.35–2.37)	.85	0.92 (0.34–2.52)	.88	1.02 (0.41–2.56)	.97	0.99 (0.39–2.52)	.99
Contractility
Ln Ees	0.42 (0.18–0.97)	.041	0.36 (0.15–0.85)	.020	0.33 (0.14–0.79)	.012	0.37 (0.15–0.92)	.032
PRSW, 1 g/cm 2	0.96 (0.94–0.98)	.001	0.96 (0.94–0.98)	.001	0.97 (0.94–0.99)	.010	0.97 (0.94–0.99)	.007
S′, 1 cm/sec	0.61 (0.48–0.77)	<.001	0.56 (0.43–0.73)	<.001	0.67 (0.51–0.88)	.004	0.68 (0.52–0.90)	.006
GLS ∗ , 1%	0.92 (0.85–0.99)	.025	0.92 (0.85–0.99)	.022	0.91 (0.84–0.98)	.017	0.91 (0.84–0.98)	.016
Diastolic dysfunction
Ln E/E′ ratio	3.63 (1.81–7.31)	<.001	5.29 (2.21–12.7)	<.001	3.02 (1.29–7.05)	.011	2.98 (1.26–7.06)	.013
Ln LA volume index	1.93 (0.72–5.16)	.19	1.86 (0.68–5.06)	.23	1.19 (0.42–3.40)	.74	1.52 (0.58–4.00)	.39
Integrated parameters
Ln SV index	0.24 (0.07–0.77)	.016	0.20 (0.06–0.73)	.014	0.25 (0.06–1.07)	.06	0.21 (0.05–0.84)	.028
Ln EF	0.08 (0.02–0.38)	.002	0.04 (0.01–0.24)	<.001	0.09 (0.02–0.41)	.002	0.14 (0.02–0.75)	.022
Ln Ea/Ees ratio, 1 SD	1.89 (1.38–2.59)	<.001	1.90 (1.41–2.57)	<.001	1.94 (1.42–2.64)	<.001	1.86 (1.35–2.57)	<.001

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

1. Continuing Education and Meeting Calendar
2. New Parameter Derived from Three-Dimensional Speckle-Tracking Echocardiography for the Estimation of Left Ventricular Filling Pressure in Nondilated Hearts
3. Quantification of Multiple Mitral Regurgitant Jets: An In Vitro Validation Study Comparing Two- and Three-Dimensional Proximal Isovelocity Surface Area Methods
4. The 2017 ASE Vascular Tract Baltimore: A Voyage Beyond

Stay updated, free articles. Join our Telegram channel

Join