Introduction
Left ventricular non-compaction cardiomyopathy (LVNC) is a potentially life-threatening disease, characterized by a thin, compacted outer layer and a thick, non-compacted inner layer with deep recesses between prominent trabeculations [1, 2]. Symptomatic patients typically present with heart failure, ventricular arrhythmias or thromboembolic events [3–5]. However, the clinical course of patients is variable and the need for identifying factors which relate to adverse outcomes and subsequent mortality remains crucial. Previous studies showed, that left ventricular ejection fraction (LVEF), right ventricular size and systolic function, N-terminal prohormone of B-type natriuretic peptide (NT-proBNP), exercise capacity and heart failure symptoms can predict outcome [6–8].
In patients with chronic heart failure impaired renal function seems to be consistently associated with a worse prognosis independent of other risk factors [9–15]. In non-ischemic cardiomyopathy, moderate renal insufficiency was an independent risk factor for cardiac events, even in patients with only mild to moderate symptoms [16–18]. A similar relationship was observed in takotsubo cardiomyopathy, where lower estimated glomerular filtration rate (eGFR) values during hospitalization were associated with longer hospitalizations and higher rates of adverse events [19].
According to available research, no studies have evaluated renal function and its prognostic value among patients with LVNC. The aim of this study was to evaluate renal function and its prognostic role in patients with LVNC.
Methods
Patients and data collection
All patients diagnosed with LVNC between 1988 and 2016, fulfilling the echocardiographic criteria described by Jenni et al. [20] were identified from databases at the University Hospitals Zurich, Basel, Geneva, and St. Gallen. Patients with at least 1 measurement of creatinine were included in this retrospective analysis. The study was approved by the local ethical committees and informed consent was obtained from all participants.
Demographic and clinical data as well as echocardiography data were collected retrospectively and entered into a web-based database (SecuTrial, Berlin, Germany) hosted by the Clinical Trial Center at the University of Zurich. Entry into study was defined as the first visit in one of the study hospitals when at least one parameter was recorded. All values for creatinine and urea at baseline and follow-ups were collected and eGFR was assessed by the CKD-EPI formula [21]. Serum creatinine level was measured in each center based on certified protocols [22–24].
The endpoint was defined as the occurrence of death by any cause or need for heart transplantation as assessed in hospital records as well as by telephone survey.
Statistical analysis
Statistical analysis of the time-to-event data was performed using the Cox proportional hazard models with age as time scale (with or without adjustment for age and gender), so patients were treated as left-truncated at their age of entry. In order to check for non-informative late entry, the age at entry of a patient was included in the adjusted models, without showing a significant effect [25]. Time-dependent variation of the covariates creatinine, urea, and eGFR was taken into account by creating a data set listing the time-dependent covariates for each follow-up visit of a patient and the time span during which the values of the covariates did not change [26]. In case of a violation of the proportional hazards assumption, it was examined by plotting the scaled Schoenfeld residuals against time (age of patient) and by applying the test developed by Grambsch and Therneau [27], the covariate was modeled using a time-dependent coefficient (linear time scale), and hazard ratios were assessed for different ages separately. In the analysis of different cut-offs, a minimum of two events were considered per group, and was necessary to avoid complete separation. Statistical software for the R programming language were used.
Results
Patients and sample size
During 1025 person-years (longest follow-up 18.7 years) 23 (18%) patients died or underwent heart transplantation. An overview of the study population and its baseline characteristics is provided in Table 1. Table 2 provides data of patients reaching the endpoint. All 126 patients had in total 888 creatinine measurements resulting in 888 eGFR calculations. A subset of 94 patients had in total 667 urea measurements. All data points were included in the analysis.
All patients |
Patients not reaching |
Patients reaching |
|
Number of patients |
126 |
103 |
23 |
Age [years] |
45.7 ± 16.9 |
44.6 ± 17.1 |
50.8 ± 15.6 |
Female |
41 (32.5%) |
36 (34.9%) |
5 (21.7%) |
Systolic blood pressure [mmHg] |
122 ± 20.90 |
124 ± 19.05 |
111 ± 26.08 |
Diastolic blood pressure [mmHg] |
75 ± 12.01 |
75 ± 10.80 |
70 ± 17.74 |
Heart rate [bpm] |
75 ± 18.72 |
74 ± 17.60 |
79 ± 22.50 |
Left ventricular ejection fraction [%] |
41 ± 17.30 |
43 ± 16.67 |
28 ± 14.40 |
Medication: |
|||
Beta-blockers |
50 (39.7%) |
38 (36.9%) |
11 (47.8%) |
ACE-inhibitor |
47 (37.3%) |
29 (28.1%) |
17 (74.0%) |
AT-2 antagonist |
14 (11.1%) |
10 (9.7%) |
4 (17.4%) |
Aldosterone antagonist |
19 (15.1%) |
11(10.7%) |
8 (34.8%) |
Calcium antagonist |
5 (3.9%) |
4 (3.9%) |
0 (0.0%) |
Diuretics |
53 (42.1%) |
34 (33.0%) |
18 (78.2%) |
Digitalis |
10 (7.9%) |
5 (4.8%) |
5 (21.7%) |
ASA |
22 (17.4%) |
17 (16.5%) |
5 (21.7%) |
Statin |
11 (8.7%) |
6 (5.8%) |
4 (17.4%) |
Anticoagulant |
38 (30.2%) |
24 (23.3%) |
13 (56.5%) |
Baseline |
Last measurement before |
|
Systolic blood pressure [mmHg] |
111 ± 26.08 |
94 ± 24.35 |
Diastolic blood pressure [mmHg] |
70 ± 17.74 |
61 ± 16.83 |
Heart rate [bpm] |
79 ± 22.50 |
77.60 ± 18.57 |
Left ventricular ejection fraction [%] |
28 ± 14.40 |
28 ± 14.01 |
Medications: |
||
Beta-blockers |
11 (47.8%) |
13 (56.5%) |
ACE-inhibitor |
17 (74.0%) |
9 (39.1%) |
AT-2 antagonist |
4 (17.4%) |
6 (26.1%) |
Aldosterone antagonist |
8 (34.8%) |
7 (30.4%) |
Calcium antagonist |
0 (0.0%) |
1 (4.3%) |
Diuretics |
18 (78.2%) |
13 (56.5%) |
Digitalis |
5 (21.7%) |
1 (33.3%) |
ASA |
5 (21.7%) |
1 (4.3%) |
Statin |
4 (17.4%) |
5 (21.7%) |
Anticoagulant |
13 (56.5%) |
14 (60.8%) |
Baseline and event-preceding kidney function
The median creatinine, urea, and eGFR levels at baseline of all patients, as well as separately for patients not reaching the endpoint and patients reaching the endpoint are depicted in Table 3. Median last measurements of patients reaching the endpoint was for creatinine 112 (interquartile range [IQR] 82.5–146) µmol/L, for urea 6.8 (5.1–9.7) mmol/L, and for eGFR 68 (50–76) mL/min (Table 3, Figs. 1, 2).
Baseline measurement |
Last measurement |
|
All patients (n = 126) |
||
Creatinine [µmol/L] |
87 (74–106) |
85 (75.5–102) |
Urea [mmol/L] |
6.1 (4.4–8.1) |
5.9 (4.5–7.4) |
eGFR [mL/min] |
85.5 (68.2–95.8) |
82.5 (64.5–97.8) |
Patients not reaching the endpoint (n = 103) |
||
Creatinine [µmol/L] |
84 (73–102) |
83 (73–95.5) |
Urea [mmol/L] |
5.5 (4.2–7.0) |
5.8 (4.25–7) |
eGFR [mL/min] |
89 (70.5–96.5) |
89 (68–101) |
Patients reaching the endpoint (n = 23) |
||
Creatinine [µmol/L] |
94 (84.5–106) |
112 (82.5–146) |
Urea [mmol/L] |
7.9 (5.8–9.4) |
6.8 (5.1–9.7) |
eGFR [mL/min] |
73 (63.5–89.5) |
68 (50–76) |
Survival analysis
Cox regression analysis revealed a highly significant relationship between the creatinine level and the risk of death or heart transplantation. The risk of reaching the endpoint was substantially increased in both the unadjusted analysis (hazard ratio [HR] 1.9 for every increase of 30 µmol/L, 95% confidence interval [CI] 1.39–2.57, p < 0.001) and after adjustment for age and gender (adjusted HR 1.9, 95% CI 1.37–2.57, p < 0.001, Fig. 3A). Doubling of creatinine (log2 analysis) resulted in an almost 8 times higher risk of death or transplantation in both unadjusted analysis (HR 7.8, 95% CI 3.09–19.76, p < 0.001) and after adjustment for age and gender (adjusted HR 7.7, 95% CI 2.96–19.85, p < 0.001, Fig. 3B).
Similarly, for eGFR a highly significant relationship with the risk of death or heart transplantation was observed. The risk of reaching the endpoint was about twice as high with every 15 mL/min decrease in both the unadjusted analysis (HR 1.8, 95% CI 1.34–2.54, p = 0.0002) and after adjustment for age and gender (adjusted HR 1.9, 95% CI 1.36–3.62, p = 0.0001, Fig. 3A). Bisection of eGFR (log2 analysis) was associated with a 5 times higher risk of death or transplantation in both unadjusted analysis (HR 4.9, 95% CI 2.30–10.52, p = 0.0002) and after adjustment for age and gender (adjusted HR 5.3, 95% CI 2.40–11.60, p < 0.0001, Fig. 3B).
In addition, the prognostic relevance of clinically used cut-off values was assessed. An eGFR ≤ 60 mL/min was associated with a 4 times higher risk of death or transplantation in both the unadjusted analysis (HR 3.9, 95% CI 1.5–10, p = 0.005) and after adjustment for age and gender (adjusted HR 4.0, 95% CI 1.55–10.6, p = 0.004, Fig. 3C). An even stronger effect was observed for an eGFR ≤ 30 mL/min with an 8 to 10 times higher risk in the unadjusted analysis (HR 8.2, 95% CI 2.34–28.4, p = 0.001) and after adjustment for age and gender (adjusted HR 10.5, 95% CI 2.87–38.2, p = 0.0004, Fig. 3C).
Comparably, a highly significant relationship between urea levels and the risk of death or heart transplantation was seen. The risk of reaching the endpoint was increased with higher urea in both the unadjusted analysis (HR 1.6 for every increase of 5 mmol/L, 95% CI 1.15–2.16, p = 0.005) and after adjustment for age and gender (adjusted HR 1.6 for every increase of 5 mmol/L, 95% CI 1.15–2.19, p = 0.004, Fig. 3A). Doubling of urea (log2 analysis) resulted in an increased risk of death or transplantation by factor 2.5 in both the unadjusted analysis (HR 2.5, 95% CI 1.49–4.23, p = 0.0006) and after adjustment for age and gender (adjusted HR 2.5, 95% CI 1.50–4.32, p = 0.0006, Fig. 3B).
Additional analysis blanking older values
Owing to the retrospective study design the interval between the assessment of kidney function and the endpoint was variable (median distance [days] to event for creatinine/eGFR measurements 121 [IQR 6.0–370.5]; for urea 121 [IQR 6.0–370.5]). Therefore, an additional analysis was performed where all values were measured more than 1 year before an event were blanked. Also with this approach, results were consistent (Suppl. Table S1). For eGFR ≤ 30 mL/min it was no longer possible to calculate HRs due to complete separation (empty cells as a result of fewer data points).
Discussion
Predictors of mortality remain scarce in patients with LVNC. Renal dysfunction is common in patients with heart disease, occurring due to individual combinations of pre-existing renal damage, impaired perfusion, and venous congestion [28]. Impaired renal function has been observed to predict poor outcome in various cardiomyopathies [16–18]. This study determined the prognostic value of renal function in one of the largest LVNC cohorts published to date, with 126 patients and a median follow-up duration of more than 7.9 years. The overall mortality and heart transplantation rate was 18%, which is in the range of previous studies reporting rates were between 2% and 35% [7, 29–32].
Kidney function was a strong predictor of death or heart transplantation in LVNC patients. Elevated creatinine was associated with a substantially higher risk of death or heart transplantation in our cohort. Doubling of creatinine resulted in an almost 8 times higher risk of death or transplantation. Since creatinine is freely filtered by the glomerulus, it allows direct estimation of GFR in some cases. However, multiple sources of bias (such as age and gender) lead to an inaccurate estimation of GFR [33]. Therefore, in the present study the CKD-EPI formula was used, which includes age, race, and serum creatinine and estimates GFR more accurately in patients with chronic systolic heart failure compared with the MDRD or Cockcroft-Gault equation [34]. Every bisection of kidney function assessed by eGFR was associated with a 5-times higher risk. An even 8 to 10 times higher risk was observed for patients with an eGFR of less than 30 mL/min. Similar associations of impaired kidney function and clinical outcome have been documented for other cardiomyopathies. In non-ischemic dilated cardiomyopathy, an eGFR < 60 mL/min was an independent predictor of death or the need for heart transplantation [17]. A very similar relationship was described in children with dilated cardiomyopathy [16]. In patients with takotsubo cardiomyopathy, lower eGFR values during hospitalization were associated with longer hospitalizations and higher rates of adverse events [18]. These findings suggest that the cardio-renal association is similar in LVNC and other cardiomyopathies. Heart failure interacts with kidney function via numerous pathways in both an acute and chronic setting. This complex interplay involves hemodynamic, (neuro-)humoral, and direct cardiovascular disease-associated mechanisms [35]. A possible reason why eGFR is such a powerful and consistent predictor of outcome in different heart diseases is its dependency on renal perfusion and thus on cardiac output [36–38].
In line with the described findings for creatinine, elevated urea was associated with a higher risk of death and heart transplantation. Doubling of urea resulted in a 2.5-times increased risk. In acute heart failure patients, it was even observed to be the strongest predictor of mortality amongst all renal function parameters [39]. Possibly, this is due to the fact that urea is not only dependent on renal perfusion, but also on tubular function (up to 50% of urea is passively reabsorbed in the renal tubules) and is closely related to neurohumoral activity such as renin–angiotensin–aldosterone system activity [40]. Thus, compared with creatinine, urea may be more sensitive to changes in diuretic therapy, venous congestion and volume status.
Limitations of the study
Even though this was one of the largest LVNC cohorts studied, this cardiomyopathy is still rare and thus conclusions are limited by the relatively small number of events. Further on, referral bias, and — given the observational retrospective study design — possible confounding bias as well as missing measurements of patients not requiring medical care may have affected the results.
Conclusions
This study provides evidence that a decrease in kidney function as assessed by creatinine, eGFR, and urea is associated with an increased risk of death and heart transplantation in patients with LVNC. In light of this observation, it is suggested herein, that renal function should be included in follow-up and risk assessment of LVNC patients.