Severe Aortic Stenosis in Dialysis Patients.

BACKGROUND: Characteristics and prognosis of hemodialysis patients with severe aortic stenosis have not yet been well defined. METHODS AND
RESULTS: The CURRENT AS (contemporary outcomes after surgery and medical treatment in patients with severe aortic stenosis) registry, a Japanese multicenter registry, enrolled 3815 consecutive patients with severe aortic stenosis. There were 405 hemodialysis patients (initial aortic valve replacement [AVR] group: N=135 [33.3%], and conservative group: N=270) and 3410 nonhemodialysis patients (initial AVR group: N=1062 [31.1%], and conservative group: N=2348). The median follow-up duration after the index echocardiography was 1361 days, with 90% follow-up rate at 2 years. The cumulative 5-year incidence of all-cause death was significantly higher in hemodialysis patients than in nonhemodialysis patients in both the entire cohort (71% versus 40%, P<0.001) and in the initial AVR group (63.2% versus 17.9%, P<0.001). Among hemodialysis patients, the initial AVR group as compared with the conservative group was associated with significantly lower cumulative 5-year incidences of all-cause death (60.6% versus 75.5%, P<0.001) and sudden death (10.2% versus 31.7%, P<0.001). Nevertheless, the rate of aortic valve procedure-related death, which predominantly occurred within 6 months of the AVR procedure, was markedly higher in the hemodialysis patients than in the nonhemodialysis patients (21.2% and 2.3%, P<0.001).
CONCLUSIONS: Among hemodialysis patients with severe aortic stenosis, the initial AVR strategy as compared with the conservative strategy was associated with significantly lower long-term mortality risk, particularly the risk for sudden death, although the effect size for the survival benefit of the initial AVR strategy was smaller than that in the nonhemodialysis patients.

Clinical Perspective

What Is New?

Although the initial aortic valve replacement strategy improves prognosis of hemodialysis patients with severe aortic stenosis, the magnitude of its benefit is smaller in hemodialysis patients than in nonhemodialysis patients, which could at least in part be explained by the extremely high rate of aortic valve procedure–related death in hemodialysis patients.

What Are the Clinical Implications?

The initial aortic valve replacement is an effective procedure to treat hemodialysis patients with severe aortic stenosis, but there is an urgent unmet need to improve the procedural outcomes of aortic valve replacement in hemodialysis patients.

Introduction

The number of patients with end‐stage renal disease requiring hemodialysis has been gradually increasing. In 2012, more than 400 000 American patients with end‐stage renal disease were on hemodialysis, including over 98 000 new patients.1 In Japan, more than 300 000 patients with end‐stage renal disease are on hemodialysis, and the candidate pool increases by over 30 000 patients per year.2 Dialysis patients are associated with a higher risk for the development of cardiovascular diseases with calcified degeneration of cardiac valves and/or heavy calcification of coronary arteries.3, 4, 5, 6 Calcified aortic stenosis (AS) is particularly common in hemodialysis patients, and has been found to be an independent risk factor for death in hemodialysis patients.6, 7 Regarding the clinical outcomes of hemodialysis patients with severe AS, there are only a few small studies suggesting the higher mortality and morbidity rates following surgical aortic valve replacement (AVR).8, 9, 10 However, the characteristics and prognosis of hemodialysis patients with severe AS have not yet been well defined in comparison with those in nonhemodialysis patients with severe AS. Also, it is not clear whether the effect size of the initial AVR strategy relative to the conservative strategy differs between hemodialysis patients and nonhemodialysis patients, although we previously reported that the initial AVR strategy was associated with markedly lower mortality risk than the conservative strategy in the asymptomatic patients with severe AS.11 Therefore, we sought to investigate the long‐term outcomes of hemodialysis patients with severe AS and to evaluate the effect of the initial AVR strategy relative to the conservative strategy in clinical outcomes compared between hemodialysis and nonhemodialysis patients in a large Japanese observational database of consecutive patients with severe AS.

Methods

Study Design and Patient Population

The study design of the CURRENT AS (contemporary outcomes after surgery and medical treatment in patients with severe aortic stenosis) registry was previously described in detail.11 In brief, the CURRENT AS registry is a retrospective, multicenter registry that enrolled 3815 consecutive patients with severe AS from 27 centers (Appendix S1, on‐site surgical facility in 20 centers) in Japan between January 2003 and December 2011. We examined the hospital database of transthoracic echocardiography and enrolled consecutive patients meeting the definition of severe AS (peak aortic jet velocity >4.0 m/s, mean aortic pressure gradient >40 mm Hg, or aortic valve area <1.0 cm2) for the first time during the study period. We excluded patients with a history of aortic valve repair/replacement/plasty or percutaneous aortic balloon valvuloplasty. The institutional review boards in all 27 participating centers approved the protocol. Written informed consent from each patient was waived because clinical information was obtained from the routine practice, and no patient refused to participate in the study when contacted for follow‐up.

Among the 3815 study patients, there were 405 hemodialysis patients (initial AVR group: N=135 [33.3%], and conservative group: N=270) and 3410 nonhemodialysis patients (initial AVR group: N=1062 [31.1%], and conservative group: N=2348) at the time of the index echocardiography (Figure 1). The conservative group consisted of all the patients other than those in whom AVR was planned based on the index echocardiographic findings. The conservative management included watchful waiting for symptoms in asymptomatic patients, and medical management of angina and/or heart failure in symptomatic patients. There were 3 main reasons why AVR was not performed in the conservative group. First, patients were considered to have no formal AVR indications by their attending physician (1388 patients, 53.0%) either because of absence of symptoms (1067 patients) or improvement of symptoms by medication (237 patients). Second, patients had unacceptably high surgical risk (798 patients, 30.5%). Third, patients refused to undergo AVR (403 patients, 15.4%).

Figure 1

Study flow chart. AS indicates aortic stenosis; AVR, aortic valve replacement.

The effect of the initial AVR strategy relative to the conservative strategy was assessed based on the initial strategies regardless of the actual performance of AVR. The follow‐up was commenced on the day of the index echocardiography except for the analysis comparing the clinical outcomes after AVR between hemodialysis and nonhemodialysis patients, in which the follow‐up was commenced on the day of AVR.

Echocardiography

All patients underwent a comprehensive 2‐dimensional and Doppler echocardiographic evaluation in each participating center. Peak aortic jet velocity and mean pressure gradient were obtained with the use of the simplified Bernoulli equation. Aortic valve area was calculated with the use of the standard continuity equation and was indexed by body surface area.12

Data Collections and Definitions

Baseline clinical information was collected by reviewing the hospital charts or database. Follow‐up information was collected mainly by reviewing the hospital charts, and additional information was collected through contact with patients, their families, and/or referring physicians by sending mails regarding survival status, symptoms, and subsequent hospitalizations.

The primary outcome measure in the present analysis was all‐cause death. The secondary outcome measures included cardiovascular death, aortic valve–related death, aortic valve procedure–related death, sudden death, noncardiovascular death, hospitalization for heart failure (HF), and a composite of aortic valve–related death or hospitalization for HF. The causes of death were classified according to the Valve Academic Research Consortium definitions and were adjudicated by a clinical event committee (Data S1).13, 14 Aortic valve–related death included aortic valve procedure–related death, sudden death, and death attributable to HF possibly related to the aortic valve. Hospitalization for HF was defined as hospitalization because of worsening HF requiring intravenous drug therapy.

Statistical Analysis

Categorical variables are presented as numbers and percentages and were compared with the χ2 test or Fisher exact test. Continuous variables are presented as the mean and SD or median and interquartile range. Continuous variables were compared using Student t test or Wilcoxon rank sum test based on their distributions. The Kaplan–Meier method was used to calculate the cumulative incidence of events, and the differences were assessed with the log‐rank test.

The Cox proportional hazard models were used to estimate the adjusted risk of hemodialysis patients relative to nonhemodialysis patients or the adjusted risk of initial AVR strategy relative to conservative strategy for all‐cause death, cardiovascular death, aortic valve–related death, aortic valve procedure–related death, sudden death, noncardiovascular death, hospitalization for HF, and a composite of aortic valve–related death and hospitalization for HF. We selected 20 clinically relevant factors listed in Table 1 and Tables S1 and S2 as the risk‐adjusting variables consistent with the previous report. We constructed parsimonious models with a limited number of variables for aortic valve procedure–related death, because the number of patients with this event was too small for the nonparsimonious model. We selected 6 clinically relevant variables as the risk‐adjusted variables as listed in Table 1 and Tables S1 and S2 in the parsimonious models. The centers were incorporated into all the models as the stratification variable. With the exception of age, the continuous variables were dichotomized by the median values or clinically meaningful reference values. Because the differences in the age distributions between the hemodialysis and nonhemodialysis groups and between the initial AVR and conservative strategies were too large to allow the dichotomous approach, we treated age as a continuous variable in the Cox proportional hazard models. The risks for the clinical end points were expressed as hazard ratios and their 95% CI. We also evaluated the interaction between hemodialysis/nonhemodialysis subgroups and the effect of the initial AVR strategy relative to the conservative strategy for the primary and secondary outcome measures.

Table 1

Baseline Characteristics: Dialysis VS Nondialysis Patients

Variables	Dialysis Patients (N=405)	Nondialysis Patients (N=3410)	P Value
Age, y	73.2±8.6	78.3±9.7	<0.001
Age ≥80 ya, b	93 (23)	1636 (48)	<0.001
Mena	242 (60)	1201 (35)	<0.001
Body mass indexc	20.5±3.6	21.9±3.8	<0.001
Body mass index <22a	298 (74)	2028 (59)	<0.001
Body surface area, m2	1.46±0.18	1.46±0.19	0.47
Initial treatment strategies			0.37
Initial AVR strategya	135 (33)	1062 (31)
Conservative strategya	270 (67)	2348 (69)
Any symptoms possibly related to aortic stenosis	198 (49)	1807 (53)	0.12
Angina	60 (15)	438 (13)	0.27
Syncope	25 (6)	173 (5)	0.34
Heart failure	143 (35)	1460 (43)	0.004
Admission for heart failure at index echocardiographya, b	60 (15)	730 (21)	0.002
Hypertensiona	287 (71)	2380 (70)	0.66
Current smokinga	22 (5)	174 (5)	0.78
History of smoking	120 (30)	710 (21)	<0.001
Dyslipidemia	78 (19)	1249 (37)	<0.001
On statin therapy	45 (11)	925 (27)	<0.001
Diabetes mellitus	127 (31)	770 (23)	<0.001
On insulin therapya	36 (9)	152 (4)	<0.001
Prior myocardial infarctiona	48 (12)	275 (8)	0.01
Prior percutaneous coronary intervention	97 (24)	405 (12)	<0.001
Prior coronary artery bypass grafting	45 (11)	154 (5)	<0.001
Prior open heart surgery	51 (13)	265 (8)	<0.001
Prior symptomatic strokea	70 (17)	433 (13)	0.01
Atrial fibrillation or fluttera, b	100 (25)	728 (21)	0.12
Aortic/peripheral vascular diseasea	82 (20)	200 (6)	<0.001
Serum creatinine, mg/dL		0.83 (0.68–1.1)
Years from dialysis introduction
<1 y	33 (8)
≥1 y, <5 y	116 (29)
≥5 y, <10 y	87 (21)
≥10 y	169 (42)
Anemiaa, d	335 (83)	1782 (52)	<0.001
Liver cirrhosis (Child‐Pugh B or C)a	6 (1)	32 (1)	0.29
Malignancy	59 (15)	458 (13)	0.53
Malignancy currently under treatmenta, b	13 (3)	136 (4)	0.44
Chest wall irradiation	2 (0.5)	23 (1)	1.0
Immunosuppressive therapy	17 (4)	114 (3)	0.37
Chronic lung disease	40 (10)	360 (11)	0.67
Chronic lung disease (moderate or severe)a	11 (3)	101 (3)	0.78
Coronary artery diseasea	195 (48)	949 (28)	<0.001
Logistic EuroSCORE, %	13.4 (7.9–23.0)	9.4 (5.5–16.1)	<0.001
EuroSCORE II, %	2.7 (1.9–4.6)	2.9 (1.6–4.9)	0.10
STS‐PROM score, %	8.5 (5.6–12.3)	3.5 (2.1–5.7)	<0.001
Etiology of aortic stenosis			<0.001
Degenerative	394 (97)	2985 (88)
Congenital (unicuspid, bicuspid, quadricuspid)	4 (1)	254 (7)
Rheumatic	5 (1)	145 (4)
Infective endocarditis	1 (0.3)	6 (0.2)
Others	1 (0.3)	20 (0.6)
Echocardiographic variables
Left ventricular end‐diastolic diameter, mm	48.9±7.2	45.6±6.9	<0.001
Left ventricular end‐systolic diameter, mm	34.2±8.8	29.8±7.7	<0.001
Left ventricular ejection fraction, %	56.9±14.0	63.5±13.3	<0.001
Left ventricular ejection fraction <40%	57 (14)	236 (7)	<0.001
Left ventricular ejection fraction <50%	105 (26)	488 (14)	<0.001
Interventricular septum thickness in diastole, mm	11.7±2.3	11.3±2.3	<0.001
Posterior wall thickness in diastole, mm	11.4±2.0	10.9±2.0	<0.001
Peak aortic jet velocity, m/s	3.96±0.87	4.16±0.92	<0.001
Peak aortic jet velocity ≥5 m/s	52 (13)	646 (19)	0.003
Peak aortic jet velocity ≥4 m/sa, b	211 (52)	1974 (58)	0.03
Peak aortic pressure gradient, mm Hg	66±28	72±32	<0.001
Mean aortic pressure gradient, mm Hg	37±17	41±20	<0.001
Aortic valve area (equation of continuity), cm2	0.74±0.17	0.72±0.19	0.03
Aortic valve area index, cm2/m2	0.51±0.12	0.50±0.13	0.18
Any combined valvular disease (moderate or severe)a, b	166 (41)	1392 (41)	0.95
Moderate or severe aortic regurgitation	77 (19)	714 (21)	0.37
Moderate or severe mitral stenosis	14 (3)	119 (3)	0.97
Moderate or severe mitral regurgitation	102 (25)	661 (19)	0.006
Moderate or severe tricuspid regurgitation	64 (16)	564 (17)	0.71
Tricuspid regurgitation pressure gradient ≥40 mm Hga	80 (20)	526 (15)	0.02

Values are number (%), mean±SD, or median (interquartile range) unless otherwise stated. AVR indicates aortic valve replacement; PROM, predicted risk of mortality; STS, Society of Thoracic Surgeons.

Potential independent variables selected in the Cox proportional hazard models for all‐cause death, cardiovascular death, aortic valve–related death, sudden death, noncardiovascular death, hospitalization for heart failure, and a composite of aortic valve–related death or hospitalization for heart failure.

Potential independent variables selected in the Cox proportional hazard models for aortic valve procedure–related death.

Body mass index was calculated as weight in kilograms divided by height in meters squared.

Anemia was defined by the World Health Organization criteria (hemoglobin <12.0 g/dL in women and <13.0 g/dL in men).

All statistical analyses were conducted by 2 physicians (Y.K. and T.T.) and a statistician (T.M.) with the use of JMP 10.0.2 or SAS 9.4. All reported P values were 2‐tailed, and P<0.05 was considered statistically significant.

Results

Baseline Characteristics and Clinical Outcomes: Hemodialysis Versus Nonhemodialysis Patients

Baseline characteristics were significantly different between hemodialysis patients and nonhemodialysis patients (Table 1).

The median follow‐up period after the index echocardiography was 1361 (interquartile range: 1055–1697) days, with a 90% follow‐up rate at 2 years. Cumulative 5‐year incidence of the first surgical AVR or transcatheter aortic valve implantation (TAVI) was significantly higher in hemodialysis patients than in nonhemodialysis patients in the entire study population as well as in the conservatively managed population (Figure 2 and Figure S1). TAVI was performed in 40 patients including 1 hemodialysis patient during the follow‐up period.

Figure 2

Cumulative incidence of surgical AVR or TAVI: dialysis vs nondialysis patients. Follow‐up was commenced on the day of the index echocardiography. AVR indicates aortic valve replacement; TAVI, transcatheter aortic valve implantation.

Cumulative 5‐year incidence of the primary outcome measure (all‐cause death) was significantly higher in hemodialysis patients than in nonhemodialysis patients (71% versus 40%, P<0.001) with a notably higher rate in the first year (35.9% versus 13.4%) after the index echocardiography (Table 2 and Figure 3). The risk for sudden death was markedly higher in hemodialysis patients than in nonhemodialysis patients (Table 2 and Figure 3). Cumulative 5‐year incidence of hospitalization for HF was not significantly different between hemodialysis and nonhemodialysis patients (Table 3). The results from the adjusted analysis were fully consistent with those from the unadjusted analysis (Table 3).

Table 2

Clinical Outcomes: Dialysis VS Nondialysis Patients

	Dialysis Patients	Nondialysis Patients	Unadjusted HR (95% CI)	P Value	Adjusted HR (95% CI)	P Value
	Number of Patients With at Least 1 Event (Cumulative 5‐Y Incidence [%])	Number of Patients With at Least 1 Event (Cumulative 5‐Y Incidence [%])
	N=405	N=3410
All‐cause death	248 (71.0)	1201 (40.0)	2.69 (2.34–3.08)	<0.001	2.76 (2.35–3.25)	<0.001
Cardiovascular death	183 (59.5)	769 (27.8)	3.07 (2.60–3.60)	<0.001	3.18 (2.62–3.86)	<0.001
Aortic valve–related death	107 (36.1)	502 (19.2)	2.69 (2.17–3.31)	<0.001	3.10 (2.41–3.98)	<0.001
Aortic valve procedure–related death	32 (9.9)	49 (1.9)	7.02 (4.44–10.93)	<0.001	6.74 (4.17–10.91)	<0.001
Sudden death	52 (24.7)	144 (5.7)	4.70 (3.38–6.43)	<0.001	4.80 (3.23–7.14)	<0.001
Noncardiovascular death	65 (28.5)	432 (16.9)	2.00 (1.52–2.57)	<0.001	2.05 (1.52–2.77)	<0.001
Hospitalization for heart failure	57 (31.9)	755 (29.2)	0.96 (0.72–1.24)	0.76	0.84 (0.63–1.13)	0.26
A composite of aortic valve–related death or hospitalization for heart failure	136 (51.6)	936 (34.1)	1.78 (1.48–2.12)	<0.001	1.63 (1.33–2.00)	<0.001

Number of patients with at least 1 event was counted through the entire follow‐up period, while the cumulative incidence was truncated at 5 years. Follow‐up was commenced on the day of the index echocardiography. HR indicates hazard ratio.

Figure 3

Cumulative incidence of all‐cause death and sudden death: dialysis vs nondialysis patients. Follow‐up was commenced on the day of the index echocardiography.

Table 3

Baseline Characteristics in Dialysis and Nondialysis Patients: Initial AVR VS Conservative Groups

Variables	Dialysis Patients			Nondialysis Patients
	Initial AVR Group (N=135)	Conservative Group (N=270)	P Value	Initial AVR Group (N=1062)	Conservative Group (N=2348)	P Value
Age, y	70.3±8.2	74.7±8.4	<0.001	73.7±8.9	80.3±9.4	<0.001
Age ≥80 ya, b	17 (13)	76 (28)	<0.001	282 (27)	1354 (58)	<0.001
Mena	87 (64)	155 (57)	0.17	420 (40)	781 (33)	<0.001
Body mass indexc	20.8±3.1	20.3±3.9	0.24	22.5±3.6	21.7±3.9	<0.001
Body mass index <22a	95 (70)	203 (75)	0.30	527 (50)	1501 (64)	<0.001
Any symptoms possibly related to aortic stenosis	103 (76)	95 (35)	<0.001	802 (76)	1005 (43)	<0.001
Angina	33 (24)	27 (10)	<0.001	258 (24)	180 (8)	<0.001
Syncope	15 (11)	10 (4)	0.007	95 (9)	78 (3)	<0.001
Heart failure	72 (53)	71 (26)	<0.001	587 (55)	873 (37)	<0.001
Admission for heart failure at index echocardiographya, b	28 (21)	32 (12)	0.02	242 (23)	488 (21)	0.19
Hypertensiona	97 (72)	190 (70)	0.76	710 (67)	1670 (71)	0.01
Current smokinga	12 (9)	10 (4)	0.03	71 (7)	103 (4)	0.005
Diabetes mellitus on insulin therapya	11 (8)	25 (9)	0.71	47 (4)	105 (4)	0.95
Prior myocardial infarctiona	11 (8)	37 (14)	0.10	40 (4)	235 (10)	<0.001
Prior open heart surgery	12 (9)	39 (14)	0.11	36 (3)	229 (10)	<0.001
Prior symptomatic strokea	19 (14)	51 (19)	0.23	88 (8)	345 (15)	<0.001
Atrial fibrillation or fluttera, b	29 (21)	71 (26)	0.29	178 (17)	550 (23)	<0.001
Aortic/peripheral vascular diseasea	12 (9)	70 (26)	<0.001	58 (5)	142 (6)	0.50
Years from dialysis introduction			0.21
<1 y	8 (6)	25 (9)
≥1 y, <5 y	33 (24)	83 (31)
≥5 y, <10 y	29 (21)	58 (21)
≥10 y	65 (48)	104 (39)
Anemiaa, d	121 (90)	214 (79)	0.009	508 (48)	1274 (54)	<0.001
Liver cirrhosis (Child‐Pugh B or C)a	2 (1)	4 (1)	1.0	4 (0.4)	28 (1)	0.02
Malignancy currently under treatmenta, b	2 (1)	11 (4)	0.23	22 (2)	114 (5)	<0.001
Chronic lung disease (moderate or severe)a	5 (4)	6 (2)	0.52	14 (1)	87 (4)	<0.001
Coronary artery diseasea	76 (56)	119 (44)	0.02	322 (30)	627 (27)	0.03
STS‐PROM score, %	7.3 (4.8–11.4)	8.9 (5.9–13.0)	0.007	2.5 (1.6–4.1)	3.9 (2.4–6.7)	<0.001
Echocardiographic variables
Left ventricular ejection fraction, %a	58.1±13.2	56.4±14.4	0.25	63.4±14.0	63.5±12.9	0.72
Left ventricular ejection fraction <50%	32 (24)	73 (27)	0.47	173 (16)	315 (13)	0.03
Peak aortic jet velocity, m/s	4.41±0.82	3.74±0.80	<0.001	4.74±0.82	3.89±0.84	<0.001
Peak aortic jet velocity ≥4 m/sa, b	100 (74)	111 (41)	<0.001	894 (84)	1080 (46)	<0.001
Mean aortic pressure gradient, mm Hg	49±18	32±15	<0.001	54±20	36±17	<0.001
Aortic valve area (equation of continuity), cm2	0.69±0.16	0.76±0.16	<0.001	0.64±0.17	0.75±0.18	<0.001
Any combined valvular disease (moderate or severe)a, b	65 (48)	101 (37)	0.04	414 (39)	978 (42)	0.14
Tricuspid regurgitation pressure gradient ≥40 mm Hga	32 (24)	48 (18)	0.16	148 (14)	378 (16)	0.11

Values are number (%), mean±SD, or median (interquartile range) unless otherwise stated. AVR indicates aortic valve replacement; PROM, predicted risk of mortality; STS, Society of Thoracic Surgeons.

Potential independent variables selected in the Cox proportional hazard models for all‐cause death, cardiovascular death, and a composite of aortic valve–related death or hospitalization for heart failure.

Potential independent variables selected in the Cox proportional hazard models for aortic valve–related death, aortic valve–procedure death, sudden death, noncardiovascular death, and hospitalization for heart failure.

Body mass index was calculated as weight in kilograms divided by height in meters squared.

Anemia was defined by the World Health Organization criteria (hemoglobin <12.0 g/dL in women and <13.0 g/dL in men).

Baseline Characteristics and Clinical Outcomes in Hemodialysis and Nonhemodialysis Patients: Initial AVR Versus Conservative Strategies

The differences in baseline characteristics between the initial AVR and conservative groups were basically consistent in both hemodialysis and nonhemodialysis patients. Surgical risk scores were significantly lower in the initial AVR group than in the conservative group. Patients in the initial AVR group had greater echocardiographic AS severity and left ventricular wall thickness than those in the conservative group (Table 3, Table S1).

In both hemodialysis and nonhemodialysis patients, cumulative 5‐year incidence of all‐cause death was significantly lower in the initial AVR group than in the conservative group (Table 4 and Figure 4). However, the effect size of the initial AVR strategy relative to the conservative strategy on mortality was smaller in hemodialysis patients than in nonhemodialysis patients (adjusted hazard ratio 0.62, 95% CI 0.43–0.90, P=0.01, and adjusted hazard ratio 0.40, 95% CI 0.33–0.48, P<0.001, P interaction=0.001) (Table 4). In both hemodialysis and nonhemodialysis patients, the lower risks of the initial AVR group relative to the conservative group were highly significant for sudden death and HF hospitalization (Table 4 and Figure 4). In addition, we have conducted a propensity‐score matched analysis as a sensitivity analysis and analyzed the initial AVR versus conservative groups in hemodialysis and nonhemodialysis patients, respectively. The results of sensitivity analysis were similar to those of the primary analysis (Tables S3 through S6 and Figures S2 and S3).

Table 4

Clinical Outcomes in Dialysis and Nondialysis Patients: Initial AVR VS Conservative Groups

	Initial AVR Group	Conservative Group	Unadjusted HR (95% CI)	P Value	Adjusted HR (95% CI)	P Value	P Interaction
	Number of Patients With at Least 1 Event (Cumulative 5‐Y Incidence [%])	Number of Patients With at Least 1 Event (Cumulative 5‐Y Incidence [%])
	Dialysis: N=135 Nondialysis: N=1062	Dialysis: N=270 Nondialysis: N=2348
All‐cause death							0.001
Dialysis patients	66 (60.6)	182 (75.5)	0.61 (0.46–0.81)	<0.001	0.62 (0.43–0.90)	0.01
Nondialysis patients	170 (18.7)	1031 (49.3)	0.30 (0.25–0.35)	<0.001	0.40 (0.33–0.48)	<0.001
Cardiovascular death							0.004
Dialysis patients	48 (48.6)	135 (64.5)	0.61 (0.43–0.84)	0.002	0.64 (0.41–0.98)	0.04
Nondialysis patients	103 (11.3)	666 (35.6)	0.28 (0.23–0.35)	<0.001	0.36 (0.28–0.45)	<0.001
Aortic valve–related death							<0.001
Dialysis patients	27 (21.8)	80 (44.6)	0.60 (0.38–0.91)	0.02	0.57 (0.34–0.94)	0.03
Nondialysis patients	41 (4.2)	461 (26.7)	0.16 (0.12–0.23)	<0.001	0.18 (0.13–0.26)	<0.001
Aortic valve procedure–related death							0.005
Dialysis patients	26 (21.2)	6 (3.4)	8.19 (3.60–22.00)	<0.001	7.99 (2.91–21.93)	<0.001
Nondialysis patients	24 (2.3)	25 (1.8)	1.87 (1.07–3.28)	0.03	1.43 (0.75–2.73)	0.28
Sudden death							0.06
Dialysis patients	4 (10.2)	48 (31.7)	0.14 (0.04–0.35)	<0.001	0.15 (0.05–0.44)	<0.001
Nondialysis patients	23 (3.0)	121 (7.2)	0.35 (0.22–0.54)	<0.001	0.38 (0.23–0.62)	<0.001
Noncardiovascular death							0.11
Dialysis patients	18 (23.4)	47 (30.9)	0.64 (0.36–1.09)	0.10	0.73 (0.37–1.45)	0.37
Nondialysis patients	67 (8.3)	365 (21.3)	0.33 (0.26–0.43)	<0.001	0.50 (0.37–0.67)	<0.001
Hospitalization for heart failure							0.66
Dialysis patients	10 (26.9)	47 (31.7)	0.32 (0.15–0.61)	<0.001	0.16 (0.06–0.42)	<0.001
Nondialysis patients	92 (10.5)	663 (38.7)	0.23 (0.19–0.29)	<0.001	0.24 (0.19–0.31)	<0.001
A composite of aortic valve–related death or hospitalization because of heart failure							0.01
Dialysis patients	36 (42.2)	100 (56.1)	0.59 (0.40–0.85)	0.005	0.62 (0.38–1.02)	0.06
Nondialysis patients	129 (13.9)	807 (44.0)	0.27 (0.23–0.33)	<0.001	0.27 (0.22–0.33)	<0.001

Number of patients with at least 1 event was counted through the entire follow‐up period, while the cumulative incidence was truncated at 5 years. Follow‐up was commenced on the day of the index echocardiography. AVR indicates aortic valve replacement; HR, hazard ratio.

Figure 4

Cumulative incidence of all‐cause death (A) and sudden death (B) in dialysis patients and nondialysis patients: initial AVR vs conservative strategies. Follow‐up was commenced on the day of the index echocardiography. AVR indicates aortic valve replacement.

Baseline Characteristics and Clinical Outcomes After AVR: Hemodialysis Versus Nonhemodialysis Patients

Among the 135 hemodialysis and 1062 nonhemodialysis patients in the initial AVR group, surgical AVR or TAVI was actually performed in 131 patients (97%) and 1043 patients (98%), respectively. TAVI was performed only in 11 nonhemodialysis patients. The differences in baseline clinical and echocardiographic characteristics between hemodialysis and nonhemodialysis patients in the initial AVR group were consistent with those in the entire study population (Table S2). The procedural characteristics in 1174 patients who actually underwent AVR based on the initial treatment strategy are presented in Table S3. Hemodialysis patients more often underwent AVR combined with coronary artery bypass grafting and less often AVR combined with replacement of ascending aorta than nonhemodialysis patients. A bioprosthetic valve was less often used in hemodialysis patients than in nonhemodialysis patients (Table S7). Hemodialysis patients were associated with markedly higher risk for all‐cause death and aortic valve procedure–related death after AVR than nonhemodialysis patients (Table 5, Figure 5). The 30‐day mortality rate after AVR was 7.6% in hemodialysis patients, which was markedly higher than 1.3% in nonhemodialysis patients (Figure 5). The rate of aortic valve procedure–related death, which predominantly occurred within 6 months of the AVR procedure, was markedly higher in the hemodialysis patients than in the nonhemodialysis patients (21.2% and 2.3%, P<0.001) (Table 5 and Figure 5). In hemodialysis patients, the cumulative 5‐year incidences of all‐cause death and hospitalization for HF were not significantly different between the bioprosthetic and mechanical valves (Figure S4). Only 2 patients underwent redo AVR in the 80 hemodialysis patients who received the bioprosthetic valve.

Table 5

Clinical Outcomes After AVR: Dialysis VS Nondialysis Patients

	Dialysis Patients	Nondialysis Patients	Unadjusted HR (95% CI)	P Value	Adjusted HR (95% CI)	P Value
	Number of Patients With at Least 1 Event (Cumulative 5‐Y Incidence [%])	Number of Patients With at Least 1 Event (Cumulative 5‐Y Incidence [%])
	N=131	N=1043
All‐cause death	64 (63.2)	156 (17.9)	5.19 (3.84–6.93)	<0.001	4.00 (2.75–5.81)	<0.001
Cardiovascular death	47 (52.6)	92 (10.7)	6.24 (4.34–8.86)	<0.001	5.30 (3.32–8.48)	<0.001
Aortic valve–related death	26 (21.2)	30 (3.0)	8.22 (4.82–13.94)	<0.001	9.31 (5.20–16.68)	<0.001
Aortic valve procedure–related death	26 (21.2)	24 (2.3)	10.12 (5.78–17.78)	<0.001	10.86 (5.87–20.07)	<0.001
Noncardiovascular death	17 (22.4)	64 (8.0)	3.57 (2.02–5.99)	<0.001	4.87 (2.69–8.82)	<0.001
Hospitalization for heart failure	9 (25.6)	84 (9.8)	1.43 (0.67–2.70)	0.33	1.65 (0.80–3.40)	0.18
A composite of aortic valve–related death or hospitalization for heart failure	34 (40.8)	112 (12.3)	3.63 (2.43–5.28)	<0.001	2.90 (1.80–4.66)	<0.001

Number of patients with at least 1 event was counted through the entire follow‐up period, while the cumulative incidence was truncated at 5 years. Follow‐up was commenced on the day of surgical AVR or TAVI. AVR indicates aortic valve replacement; HR, hazard ratio; TAVI, transcatheter aortic valve implantation.

Figure 5

Cumulative incidence of all‐cause death and aortic valve procedure–related death after surgical AVR/TAVI in the initial AVR group: dialysis vs nondialysis patients. In the initial AVR group, 1174 of 1197 patients actually underwent surgical AVR or TAVI. Follow‐up was commenced on the day of surgical AVR or TAVI. AVR indicates aortic valve replacement; TAVI, transcatheter aortic valve implantation.

Discussion

The main findings of the present study are as follows: (1) Hemodialysis patients with severe AS have a significantly poorer prognosis than nonhemodialysis patients with severe AS; (2) The initial AVR strategy as compared with the conservative strategy in hemodialysis patients was associated with significantly lower long‐term mortality risk, particularly the risk for sudden death; (3) However, the AVR procedure‐related mortality was much higher in hemodialysis patients than that in nonhemodialysis patients.

Previous small studies have suggested the poor prognosis after AVR in hemodialysis patients with severe AS.6, 7, 8, 9, 10 The present study including a large number of hemodialysis patients with severe AS clearly demonstrated that hemodialysis patients had a significantly higher 5‐year cumulative mortality rate than nonhemodialysis patients (71% versus 40%) with a notably higher rate in the first year (35.9% versus 13.4%), although AVR was more often performed in hemodialysis patients than in nonhemodialysis patients. Compared with the survival rates of 89.7% at 1 year and 60.5% at 5 years after hemodialysis introduction reported from the Society for Dialysis Therapy, the mortality rate of our study is considerably higher, suggesting the grimmer prognosis of hemodialysis patients with severe AS, although the durations after the hemodialysis introduction were variable in the present study.2 Furthermore, the rate of sudden death in hemodialysis patients was remarkably high (9.3% at 1 year and 24.7% at 5 years) in the present study. The 5‐year rate of sudden death was much higher than 10.5% at 4 years after percutaneous coronary intervention reported in Japanese hemodialysis patients.15

AVR is the only definitive treatment in patients with severe AS. In the present study, the initial AVR strategy as compared with the conservative strategy was associated with a significantly lower long‐term mortality rate in hemodialysis patients as well as in nonhemodialysis patients. A dramatically lower risk for sudden death in the initial AVR group would have been one of the major contributors for the lower long‐term mortality in hemodialysis patients with severe AS. Furthermore, markedly lower risk for HF hospitalization in the initial AVR strategy might also have contributed to the lower long‐term mortality in hemodialysis patients with severe AS. On the other hand, it should be noted that the long‐term mortality rate of hemodialysis patients with severe AS is very high even after AVR as compared with that in nonhemodialysis patients. Thourani et al reported the 5‐year mortality rate of 71.5% after AVR in 114 hemodialysis patients, which was comparable to 63.2% in 131 hemodialysis patients in the present study.8 The effect size of the initial AVR strategy relative to the conservative strategy for all‐cause death was smaller in hemodialysis patients than in nonhemodialysis patients, which could be explained by the extremely high rate of aortic valve procedure–related death in hemodialysis patients, predominantly occurring within 6 months of the AVR procedure. The major causes of aortic valve procedure–related death were infection and hemorrhage. When undergoing surgical AVR, hemodialysis patients are more likely to have an increased risk of perioperative infection and hemorrhage because of having higher preoperative morbidity, more concomitant diseases, and more severe atherosclerosis than nonhemodialysis patients, which may have increased postoperative mortality rates in hemodialysis patients. Therefore, reduction in the procedural mortality of AVR is essential to improve the overall mortality outcome of hemodialysis patients with severe AS. In this context, less invasive TAVI might be an attractive alternative to surgical AVR in hemodialysis patients with severe AS. Currently, there are only a few small studies reporting outcomes of TAVI in hemodialysis patients; reported 30‐day mortality rates included 14.0% (6/43) (transfemoral: 6.5%, and nontransfemoral: 33.3%) by Szerlip et al, 15.2% (5/33) by Dumonteil et al, and 0% (0/17) by Maeda et al.16, 17, 18 Further investigation on the role of TAVI in hemodialysis patients with severe AS is urgently needed.

Study Limitations

The present study has several limitations. First, its observational study design was prone to inherent bias. Particularly, the comparison between the initial AVR and conservative strategies would have been influenced by selection bias toward choosing less morbid patients in the initial AVR strategy. Also, the conservative group is a heterogeneous population and includes those with indications for surgery who were thought to be too risky and those not meeting indications for AVR because they were without symptoms. This could be noted as the Society of Thoracic Surgeons predicted risk of mortality is higher in the conservative group, yet the mean aortic pressure gradient is lower in this group. Despite an extensive multivariable adjustment, we could not deny the possibility of unmeasured confounders and selection bias. Also, we did not take the competing risk of death into account for the nonfatal outcomes and also the competing risks of various causes of death. Secondly, we did not evaluate the influence of the duration after introduction of hemodialysis on long‐term outcomes. Thirdly, the threshold to choose the AVR strategy and procedural outcomes of AVR might be different across the participating centers in this multicenter registry. Finally, the number of patients undergoing TAVI in our study was too small to clarify prognosis of hemodialysis patients undergoing TAVI.

Conclusions

Among hemodialysis patients with severe AS, the initial AVR strategy as compared with the conservative strategy was associated with significantly lower long‐term mortality risk, particularly the risk for sudden death, although the effect size for the survival benefit of the initial AVR strategy was smaller than that in the nonhemodialysis patients.

Disclosures

None.

Appendix S1. CURRENT AS Registry Investigators.

Data S1.

Table S1. Baseline Characteristics of Dialysis and Nondialysis Patients: Initial AVR Versus Conservative Groups

Table S2. Baseline Characteristics of Patients in the Initial AVR Group: Dialysis Versus Nondialysis Patients

Table S3. Baseline Characteristics in the Entire Cohort and in the Propensity‐Score Matched Cohort in Dialysis Patients: Initial AVR Versus Conservative Groups

Table S4. Baseline Characteristics in the Entire Cohort and in the Propensity‐Score Matched Cohort in Nondialysis Patients: Initial AVR Versus Conservative Groups

Table S5. Clinical Outcomes in the Propensity‐Score Matched Cohort in Dialysis Patients: Initial AVR Versus Conservative Groups

Table S6. Clinical Outcomes in the Propensity‐Score Matched Cohort in Nondialysis Patients: Initial AVR Versus Conservative Groups

Table S7. Procedural Characteristics of AVR in the Initial AVR Group: Dialysis Versus Nondialysis Patients

Figure S1. Cumulative incidence of surgical AVR or TAVI in the conservative group: dialysis vs nondialysis patients. AVR indicates aortic valve replacement; TAVI, transcatheter aortic valve replacement.

Figure S2. Propensity‐score matched cohort in dialysis and nondialysis patients for the sensitivity analysis. We used logistic regression model to develop propensity score for the choice of initial AVR with 15 independent variables relevant to the choice of initial AVR listed in Table S3. Patients in the conservative group were matched to those in the initial AVR group using a 1:1 greedy matching technique. AS indicates aortic stenosis; AVR, aortic valve replacement.

Figure S3. Cumulative incidence of all‐cause death and sudden death in the propensity‐score matched cohort in (A) dialysis patients and (B) nondialysis patients: initial AVR vs conservative strategies. AVR indicates aortic valve replacement.

Figure S4. Cumulative incidences of (A) all‐cause death and (B) hospitalization for heart failure after surgical AVR in the initial AVR group in dialysis patients: mechanical valve vs bioprosthetic valve. Follow‐up was commenced on the day of surgical AVR. AVR indicates aortic valve replacement.

Click here for additional data file.