|Home | About | Journals | Submit | Contact Us | Français|
Cardiovascular disease among hemodialysis (HD) patients is linked to poor outcomes. The Acute Dialysis Quality Initiative Workgroup proposed echocardiographic (ECHO) criteria for structural heart disease (SHD) in dialysis patients. The association of SHD with patient-important outcomes is not well defined.
We determined prevalence of ECHO determined SHD and its association with survival among incident HD patients.
We analyzed patients initiating chronic HD from 2001-2013 who underwent ECHO ≤ 1 month prior to or ≤ 3 months following HD initiation (n = 654).
Mean patient age was 66 ± 16 years, and 60% of patients were male. ECHO findings that met 1 or more and ≥ 3 of the new criteria were discovered in 87% and 54% of patients, respectively. Over a median of 2.4 years, 415 patients died: 108 (26%) died within 6 months. Five-year mortality was 62%. Age- and gender-adjusted structural heart disease variables associated with death were left ventricular ejection fraction (LEVF) ≤ 45% (HR 1.48, CI 1.20-1.83) and right ventricular (RV) systolic dysfunction (HR 1.68, CI 1.35-2.07). An additive of higher death risk included LVEF ≤ 45% and RV systolic dysfunction rather than neither (HR 2.04, CI 1.57-2.67; p = 0.53 for test for interaction). Following adjustment for age, gender, race, diabetic kidney disease, and dialysis access, RV dysfunction was independently associated with death (HR 1.66; CI 1.34-2.06; p < 0.001).
SHD was common in our hemodialysis study population, and RV systolic dysfunction independently predicted mortality.
Heart failure (HF) contributes significantly to morbidity in patients with end-stage renal disease (ESRD) requiring dialysis (1-5). However, etiology of HF and HF symptoms in dialysis patients are often poorly defined and/or misclassified. Dyspnea or volume overload can be multifactorial and may not be related to underlying structural heart disease (SHD) (6). The Acute Dialysis Quality Initiative (ADQI) XI Workgroup recently proposed a new classification of HF in patients with ESRD that specifically excludes patients with volume overload and a normal heart, and focuses on those with underlying SHD as defined by echocardiography (ECHO). The goal was to optimize diagnostic and therapeutic approaches to HF and address the unique complexities associated with non-physiologic periodic volume removal (6). The 3 elements of the proposed staging system include: 1) standardized ECHO evidence of structural and/or functional heart abnormalities, 2) dyspnea occurring in the absence of primary lung disease, including isolated pulmonary hypertension, and 3) response of congestive symptoms to dialysis/ultrafiltration. Standardized ECHO criteria, adapted from the American Society of Echocardiography (ASE) and European Association of Echocardiography (EAE) consensus guidelines (7-10), assess 8 ECHO abnormalities, of which at least 1 must be present to fulfill the diagnosis of SHD.
The prevalence of SHD based on the proposed criteria and its impact on overall survival in dialysis patients are unknown, and the ADQI XI Workgroup called for research focused on the epidemiology and prognosis of dialysis patients classified with this new scheme. Our study was undertaken to ascertain prevalence of SHD based on the proposed ADQI criteria and examine prognostic implications of SHD in ESRD patients initiating maintenance hemodialysis (HD) therapy.
Mayo Clinic Dialysis Services provides all HD in the Mayo Clinic Health System, a comprehensive integrated healthcare network for 395,000 residents in southeastern Minnesota, northern Iowa, and southwestern Wisconsin, through 8 community-based outpatient HD facilities as previously described (11,12). This study included patients ≥ 18 years of age who initiated chronic outpatient HD between January 1, 2001 and May 31, 2013 (n = 1357) and remained on dialysis ≥ 30 days (n = 1187), and who underwent ECHO examination at Mayo Clinic ≤ 1 month prior or ≤ 3 months following HD initiation (n = 654). Primary outcome was all-cause mortality by study end (December 31, 2013). Minnesota Research Authorization was provided for all participants. The Mayo Clinic Institutional Review Board approved this study.
Baseline characteristics, comorbidities, and laboratory tests were collected through review of the electronic medical record. Data included cause of ESRD, type of initial dialysis access and initial dialysis location. First dialysis access was categorized as arteriovenous fistula, arteriovenous graft, or central venous catheter. Catheters were further classified as temporary (non-tunneled, non-cuffed) or tunneled (cuffed). HF diagnosis included congestive HF, systolic HF, diastolic HF, or cardiomyopathy based on manual review of medical records. The Charlson Comorbidity Index score, consisting of 19 comorbid conditions, was obtained by a previously validated automatic note search strategy (automated digital algorithm) (13). HD initiation was catagorized as occurring before or after the release of the 2005 Kidney Disease Outcomes Quality Initiative clinical practice guidelines (14).
ECHOs performed ≤ 1 month prior to or ≤ 3 months following dialysis initiation were identified through the Mayo Clinic Echocardiographic Laboratory database. Indications for ECHO are included in Online Table 1. ECHO was performed according to ASE and EAE guidelines for assessment of valves and chamber size and function (7-10). SHD was defined according to the proposed criteria from the ADQI XI Workgroup (6), except left ventricular (LV) regional wall motion abnormalities (RWMA). The 16-segment model was utilized for RWMA assessment, and any RWMA was included (instead of > 10% of the myocardium), and right ventricular (RV) systolic dysfunction included semiquantitative assessment. Measurement definitions matched the proposed ADQI XI criteria. Left ventricular hypertrophy (LVH) was defined as LV mass index > 110 g/m2 for women and > 130 g/m2 for men or LV mass index > 47g/m2.7 for women and > 50 g/m2.7 for men. Increased LV volume index was defined as > 86 ml/m 2 end-diastolic volume or > 37 ml/m2 end-systolic volume. RV systolic dysfunction was defined as lateral tricuspid annulus velocity (S′) < 9.5 cm/sec or abnormal systolic function by semiquantitative assessment, and LV systolic dysfunction as LV ejection fraction (LVEF) ≤ 45%. Other ADQI XI criteria included left atrial (LA) enlargement (volume index ≥ 34 ml/m2), diastolic dysfunction grade ≥ 2, and mitral and/or aortic valvular disease with moderate to severe stenosis or regurgitation. Methods and references for quantitation of the above ECHO variables are provided in the Supplemental Materials.
Continuous variables were reported as mean ± standard deviation (SD) or median with inter-quartile ranges (IQR) for non-normally distributed variables. Categorical variables were expressed as count (percent). Comparison of proportions between groups was made using the Chi square test. Comparison of continuous variables was done using 2-sample t-test or Wilcoxon Rank Sum test. Primary outcome was all-cause mortality between dialysis start and study period end. Subjects were censored at time of kidney transplantation, transfer to a non-Mayo Clinic Dialysis Services dialysis facility, transition to home dialysis therapies such as peritoneal dialysis or home HD, and study period end. Kaplan-Meier methods were used to summarize event rates, and comparison between groups was done using log-rank test. Age- and gender-adjusted survival curves were created using a semi-parametric approach, assuming age and gender follow the proportional hazards assumption while not requiring proportional hazards for the grouping variable (15). The association of SHD with mortality was assessed by Cox proportional hazards regression models for long-term outcomes after adjustment for age and gender. Additional models further adjusted for other predictors of mortality (race, diabetes as etiology of ESRD, type of dialysis access). For SHD variables with missing values, a missing value indicator was included to estimate the effect of missing values. Multivariable models included only SHD variables that were statistically significant after age and gender adjustment. SHD variables significant in multivariable analysis were simplified into groups for ease of interpretation.
We conducted a series of additional analyses to further evaluate parameters beyond the ADQI proposed criteria. To determine whether pulmonary hypertension (pulmonary artery systolic pressures > 35 mmHg) was also a predictor of survival, Cox regression analyses were performed in those with available pulmonary artery systolic pressures. To minimize the confounding effect of non-physiologic volume overload either prior to dialysis start or between treatments, Cox regression analyses were repeated among those who had ECHO performed after dialysis initiation and further subgrouped to those undergoing dialysis within 24 hours before ECHO. Statistical significance was set at p < 0.05 (2-sided), and statistical analyses were performed with SAS 9.4 (SAS Institute Inc., Cary, North Carolina).
From January 2001 to May 2013, 1187 patients initiated and remained on HD ≥ 1 month. Among these patients, 654 (55%) underwent ECHO ≤ 1 month prior to or ≤ 3 months following HD initiation. Baseline demographic characteristics of patients without ECHOs meeting study entry criteria (n = 533 [45%]) are shown in Table 1 and compared to the study population. Patients without ECHOs had fewer comorbidities including coronary artery disease (48% vs. 56%, p = 0.003), HF (36% vs. 56%, p < 0.001), and Charlson score ≥ 8 (45% vs. 52%, p = 0.02), but were more likely to have an AV fistula or graft dialysis access present at first dialysis treatment (39% vs. 20%, p < 0.001) (Online Table 2). Overall survival also differed between groups (Figure 1A). Study patients who underwent ECHOs and did not have SHD had similar survival to patients without ECHOs who were not included in the study (p = 0.97) (Figure 1B). However, those with SHD had worse overall survival compared to those without ECHOs (p < 0.001).
For patients with ECHOs who met study inclusion criteria (n = 654 [55%]), baseline demographics are shown for the overall study group and also divided by the presence or absence of SHD (Table 1). Mean age of the study was 66 ± 16 years (median 68; IQR: 56-78), and 56% had HF. The mean Charlson comorbidity index score was 7.5 ± 3.4 (median 8, IQR 5-10). The most common causes of kidney disease were diabetes mellitus (29%), hypertension (15%), and glomerulonephritis (12%). One or more ECHO findings of SHD were present in 567 (87%) of patients. Compared to those without SHD, patients with ≥ 1 ECHO finding of SHD at baseline were older (66.7 ± 15.8 vs. 58.5 ± 16.4, p < 0.001), had more prevalent coronary artery disease (60% vs. 34%, p < 0.001), more prevalent HF (60% vs. 26%, p < 0.001), and a higher Charlson comorbidity score (7.7 ± 3.3 vs. 6.1 ± 3.4, p < 0.001).
Baseline ECHO findings among patients with SHD are shown in Table 2. Median time between ECHO and HD initiation was 0 days (IQR -4 to 8). Most patients (85%) had ≥ 5 baseline SHD variables documented, and 53% had 7 or 8. Two hundred sixty-six patients (44%) had other ECHOs performed within the year preceding the index ECHO meeting inclusion criteria, which indicates that repeat data acquisition for some of the SHD variables may not have been performed. Within available data the prevalence of SHD, defined by the presence of ≥ 1 SHD finding, was high (87%, n = 567). Among patients with SHD, the most common abnormalities were LA volume index ≥ 34 ml/m2 (81%), diastolic dysfunction ≥ grade 2 (78%), LVH by g/m2 (49%) and by g/m2.7(71%), and RWMA (50%). Of patients with valvular SHD (16%), mitral regurgitation was the most common (13%) followed by aortic stenosis (3%). RV systolic dysfunction was present in 34% by semiquantitative assessment and 28% by S′ criteria.
Over a median follow up of 2.4 years (IQR 0.9-4.9) 239 subjects were censored, of which 61 (26%) underwent kidney transplantation, 46 (19%) were transferred to an outside dialysis unit, 32 (13%) had renal recovery, 4 (2%) were transitioned to home HD or PD, and 96 (40%) were alive at the end of study. Overall, 415 patients died; 108 (26%) of these deaths occurred within 6 months. The cumulative death rates were 17%, 38% and 62% at 6 months, 2 years, 5 years, respectively. Causes of death were sudden cardiac arrest (29%), dialysis withdrawal (inpatient or outpatient) leading to death (25%), other (10%), sepsis (7%), trauma (0.2%), and unknown cause (28%). In unadjusted analysis, compared to patients with 0-1 SHD abnormalities, survival was reduced by the presence of > 1 abnormality at baseline (p = 0.002). However, this association was no longer significant following adjustment for age and gender (0-1 SHD vs. 2-3 SHD, p = 0.75, 0-1 SHD vs. > 3 SHD, p = 0.13, and 2-3 SHD vs. 2 SHD, p = 0.05, with p-trend = 0.12). The 2 age- and gender-adjusted SHD variables significantly associated with mortality were LVEF ≤ 45% (HR 1.48, CI 1.20-1.83; p < 0.001) and RV systolic dysfunction (HR 1.68, CI 1.35-2.07, p < 0.001). RV systolic dysfunction was an important predictor of mortality, with the absence of RV dysfunction among SHD patients conferring no evident difference in survival to patients with no SHD at baseline, Central Illustration. In age-and gender-adjusted analyses using patients with no SHD as the reference group, patients with SHD but without RV dysfunction experienced no difference in survival (HR 0.84, CI 0.61-1.14, p = 0.25) whereas SHD patients with RV dysfunction had reduced survival (HR 1.41, CI 1.01-1.96, p = 0.04). The combination of RV and LV systolic dysfunction was also a strong determinant of survival. The age- and gender-adjusted relationships between RV systolic dysfunction, LVEF, and survival were further explored (Figure 2, Table 4). Patients with impaired LV function (LVEF ≤ 45%) and RV systolic dysfunction had an increased risk of death (HR 2.0, CI 1.6-2.7, p < 0.001) compared to patients with LVEF ≥ 45% and normal RV systolic function. The combination of these 2 SHD factors was simply additive (test for interaction p = 0.53). We further assessed RV systolic dysfunction in a model with age, gender, race, diabetic kidney disease and AV fistula access. RV dysfunction remained associated with death by study end (HR 1.66, CI 1.34-2.06; p < 0.001).
Sensitivity analyses examined whether our findings were supported among patients who initiated dialysis prior to ECHO (n = 358). In this cohort, Cox regression analyses revealed 3 important predictors of death, including moderate diastolic dysfunction (HR1.49,CI 1.03-2.17, p = 0.04), RV dysfunction (HR 1.75, CI 1.30-2.36, p < 0.001), and LV dysfunction (HR1.74, CI 1.29-2.33, p < 0.001). Analyses were limited to patients undergoing dialysis treatments ≤ 24 hours prior to their ECHO (n = 190) and revealed similar findings to the original study cohort wherein both RV and LV systolic dysfunction remained the only predictors of death. Separate analyses examined the relationship between pulmonary hypertension and death. Among those with measured pulmonary artery systolic pressure (n = 530), the mean pressure was 46.8 ± 14.8mmHg and 75% of patients had evidence of pulmonary hypertension (pulmonary artery systolic pressure > 35 mmHg). Compared to those without any SHD, those with ≥ 1 SHD abnormality had higher pulmonary artery systolic pressure (47.9 ± 14.7 vs. 35.0 ± 10.3, p < 0.001) and higher prevalence of pulmonary hypertension (78% vs. 44%, p < 0.001). Pulmonary hypertension was associated with reduced survival after adjusting for age and gender (HR 1.41, CI 1.08-1.84, p = 0.01), but was not independently associated with overall survival when adjusted for RV systolic dysfunction and LVEF ≤ 45% (HR 1.26, CI 0.96-1.66, p = 0.09).
In our study cohort SHD was common, with a substantial majority of the patients having ≥ 1 baseline ECHO abnormality and more than half having ≥ 3. The presence of SHD was also associated with more comorbidities. Patients with impaired LVEF and RV dysfunction had a 2-fold increased risk of death compared to patients with LVEF > 45% and normal RV function. Overall, RV dysfunction appeared to have the strongest association with mortality in this cohort.
The prevalence of underlying SHD is influenced by patient demographics and associated comorbidities, such as the higher prevalence of LVH in patients with long-standing or poorly controlled hypertension, either contributing to, or a result of their kidney disease. Patients with ESRD have a demonstrated substrate for rapid progression of a myriad of SHD subtypes (e.g., calcific valvular lesions) and known risks for ischemic cardiac dysfunction (16).
Several studies have shown impaired LVEF, diastolic dysfunction, and LVH as predictors of mortality in dialysis patients (17-25). Yamada et al. (23) found that, in 1254 incident HD patients similar in age to those in the present study (62 ± 14 years) and followed for 7 years, reduced LVEF was present in 13% and was a strong predictor of death from cardiovascular events as well as all-cause mortality. Progressive decline of LVEF was associated with increasing risk of death. We also found LV dysfunction was prevalent and associated with higher risk of mortality.
Diastolic dysfunction of grade 2 or higher was not independently associated with excess mortality in the age- and gender-adjusted analyses of our entire study cohort. However, an association with mortality was seen in patients on maintenance dialysis therapy prior to ever undergoing ECHO. Barberato et al. (20) examined a somewhat younger patient group (52 ± 16 years) with no previous cardiovascular disease such as HF, myocardial infarction or valvular disease. They found that advanced diastolic dysfunction was independently and significantly predictive of cardiovascular events (including sudden death, acute myocardial infarction, and decompensated HF requiring hospitalization) as well as increased mortality compared to those with normal diastolic function or mild diastolic dysfunction. Han et al. (21) found a strong impact of diastolic dysfunction on cardiovascular events in dialysis patients, although their study included only patients with preserved LV systolic function and defined diastolic dysfunction based on Doppler criteria (mitral early inflow velocity to early mitral annular velocity ratio E/E′ >15). Dubin and colleagues (26) demonstrated in a study of 40 patients that E/E′, but not traditional measures of diastolic dysfunction (i.e., classification into impaired relaxation, pseudonormal and restrictive filling patterns), was associated with elevated NT-pro-brain natriuretic peptide and high-sensitivity troponin. This suggests that Doppler-based data may have greater utility in evaluating cardiac function in the ESRD population. The aforementioned discordant findings may be explained by different patient populations and suggest that further evaluation of how to define diastolic function in ESRD and its impact on patient outcomes is needed.
Silberberg et al. (25), found that LVH (LV mass index > 125 g/m2) upon HD initiation was an independent determinant of all-cause mortality in 91 patients (age 55 ± 15 years) followed for ≤ 5 years. However, this study excluded patients with pre-existing malignancy and those with valvular disease. Paoletti et al. (24) also showed LVH to be predictive of subsequent sudden cardiac death in 123 patients (age range 29-79 years) who were on HD for at least 6 months and followed over a 10-year period. Our study confirmed the link between impaired LV systolic function–but not LVH or diastolic dysfunction–and poor outcome.
In this study, RV systolic dysfunction was relatively common (27%), and a minority of patients had moderate-severe dysfunction (10%). RV systolic dysfunction was independently associated with poor survival even when modeled with age, gender, diabetes, ESRD cause, and AV fistula access in a multivariable analysis. RV dysfunction in HD patients is thought to result from chronic volume overload and is exacerbated by AV fistula access, especially in the brachial position (27,28). RV dysfunction in turn leads to an impaired LV. This was shown by Paneni et al. (27) in 120 chronic dialysis patients with preserved LVEF (> 50%) in which RV systolic dysfunction correlated with indices of LV systolic and diastolic function and was independently associated with reduced LVEF. The RV-to-LV interdependence and impact of RV dysfunction is known in several cardiac diseases (29-35); RV systolic dysfunction is associated with poor outcomes in severe native aortic valve stenosis (29), is an independent predictor of short- and long-term mortality in patients with HF (31), predicts transplant-free survival in patients with dilated cardiomyopathy (32), is predictive of outcomes in patients who have undergone primary percutaneous coronary intervention for acute myocardial infarction (34), and is associated with clinical and ECHO evidence of more advanced HF predictive of poorer outcomes (36). Our study supports the interdependent relationship between RV and LV, as biventricular failure was associated with a significantly increased risk of death.
This is a retrospective study of ECHO examinations performed over a several-year timeframe in which practice patterns and guidelines changed, including the recommendation for more quantitative RV functional assessment (37,38). However, while semiquantitative RV assessment has limitations (31,37,39), it does afford useful information about RV size and function, provided the ECHO is sufficiently detailed (40,41). Semiquantitative RV assessment correlated well with quantitative assessment in our study. Referral for ECHO < 1 month prior to or ≤ 3 months after HD initiation was not systematic, as shown by the different survivorship between those who had an ECHO and those that did not. Our study nonetheless provides insight into the need for a standardized approach to SHD assessment, emphasizing the position of the ADQI to adopt consistent methodology for collection and documentation of ECHO data in dialysis patients. The predominantly white population of our study may limit generalizability, and a lack of data regarding dialysis-specific factors (adequacy, bone and mineral metabolism, anemia, hypoalbuminemia, and AV fistula duration) may have affected the association of SHD with mortality.
SHD is common among incident HD patients. Both impaired LV and RV systolic function are associated with poor outcomes and death. RV systolic dysfunction appears to have important prognostic implications, however additional studies are needed to confirm these findings. Implementation of a standardized comprehensive screening ECHO examination to completely evaluate all SHD variables may help identify HD patients at highest risk of death, inform the need for early intervention to improve patient outcomes, and generate critical conversations with patients related to their prognosis.
Guidelines recommend screening ECHO to evaluate cardiac structure and function in patients with end-stage renal disease initiating dialysis.
Patients with end-stage renal disease and RV dysfunction as assessed by ECHO are at higher risk of death during the first 6 months after initiating dialysis than patients with normal RV function.
Additional research is needed to understand the prevalence of SHD and identify predictors of early mortality in patients initiating dialysis.
Supplemental Table 1. Indications for Echocardiograms among incident hemodialysis patients (n=654).
Supplemental Table 2. Baseline patient characteristics and echocardiography findings for patients with (Echo) and without echocardiograms (No Echo) among incident hemodialysis patients.
Figure: Doppler Criteria for Classification of Diastolic Function
Participants with atrial fibrillation with DT >140 ms, other arrhythmia, fusion of E and A, or in whom diastolic parameters were not obtained, who had only 1 criterion suggesting moderate or severe diastolic dysfunction, or in whom diastolic parameters were borderline and suggestive of but not diagnostic of abnormality were classified as having indeterminate diastolic dysfunction. E, peak early filling velocity; A, velocity at atrial contraction; DT, deceleration time; Adur, A duration; ARdur; AR duration; S, systolic forward flow; D, diastolic forward flow; AR, pulmonary venous atrial reversal flow; e′, velocity of mitral annulus early diastolic motion; a′, velocity of mitral annulus motion with atrial systole; DT, mitral E velocity deceleration time
This project was supported by the Mary Kathryn and Michael B. Panitch Career Development Award (L.J.H.), Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery (L.J.H.), Mayo Clinic Rochester-Mayo Clinic Health System Integration award (M.O. and L.J.H.), Division of Cardiovascular Diseases research grant program (V.T.N., L.J.H., and C.G.S.), and Grant Number UL1 TR000135 from the National Center for Advancing Translational Sciences (NCATS). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH. Study content was presented in abstract form at Kidney Week 2014 in Philadelphia, Pennsylvania
Disclosures: A.W.W. American Society of Nephrology Public Policy Board; P.A.P. is funded through a research grant from the Intersocietal Accreditation Commission. None of the other coauthors have any disclosures.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.