The prognostic impact of pre-implantation hyponatremia on morbidity and mortality among patients with left ventricular dysfunction and implantable cardioverter-defibrillators

Abstract

Aims

Hyponatremia is commonly observed among patients with left ventricular (LV) dysfunction and is a marker for adverse outcomes. We aimed to determine the prognostic significance of pre-implant hyponatremia on the outcomes of death, acute decompensated heart failure (ADHF) and appropriate implantable cardioverter-defibrillator (ICD) therapy for ventricular arrhythmias among patients with ICDs.

Methods and results

The study population consisted of patients with an ejection fraction ≤40% undergoing ICD implantation (n = 911) for the primary or secondary prevention of sudden cardiac death from 1997 to 2007. The predictive value of the severity of pre-implantation hyponatremia stratified into mild hyponatremia (n = 268, sodium 134–136 mmol/L), moderate hyponatremia (n = 105, sodium 131–133 mmol/L), and severe hyponatremia (n = 31, sodium ≤130 mmol/L) on the risk of death, ADHF, and appropriate ICD therapy for ventricular arrhythmias as compared with patients a normal serum sodium (n = 507, sodium ≥ 137 mmol/L), was calculated using multivariable Cox proportional hazards analyses. During a mean follow-up of 775 ± 750 days as the severity of hyponatremia (from a normal sodium to severe hyponatremia) increased an incremental incidence of death (25% to 61%, P < 0.001) and ADHF (11% to 26%, P = 0.004) was observed with a reduced incidence of ICD therapy for ventricular tachycardia/ventricular fibrillation (37–29%, P = 0.037). Compared with the normal sodium cohort, patients with severe hyponatremia demonstrated an increased risk of death [adjusted hazard ratio (AHR) 2.69 (95% confidence interval, CI 1.57–4.59), P = 0.004] and ADHF [AHR 2.98 (95% CI 1.41–6.30), P = 0.004], with a lower probability of appropriate ICD therapy [AHR 0.68 (95% CI 0.27–0.88), P = 0.031].

Conclusion

Hyponatremia is commonly observed among ICD recipients with LV dysfunction. Patients with an increasing severity of hyponatremia are at increased risk of death and HF related morbidity with a reduced incidence of appropriate ICD therapy particularly among patients with severe hyponatremia.

Introduction

Hyponatremia is commonly observed among patients with heart failure (HF) and left ventricular (LV) dysfunction and is a strong predictor of adverse outcomes among a broad spectrum of HF patients.^1–4 The complex adaptive physiologic process resulting in hyponatremia is related to activation of the sympathetic nervous system, renin–angiotensin–aldosterone system and arginine vasopressin release^5,6 and parallels the deterioration in HF clinical status and progression of pump dysfunction as the severity of hyponatremia increases.⁷

The prognosis among patients with LV dysfunction is heterogeneous with some experiencing death due to ventricular arrhythmias and others with progressive pump failure.⁸ The mode of death in implantable cardioverter-defibrillator (ICD) clinical trials has demonstrated that the reduction in mortality with ICD-based therapy is exclusively due to a reduction in sudden cardiac death (SCD) and termination of ventricular arrhythmias without an impact in death resulting from progressive HF or non-cardiac-related mortality.^9–11

Current guidelines broadly recognize the limitations of non-invasive risk stratification testing to determine the prognosis among individuals with LV dysfunction and ICDs.^12,13 Clinicians are therefore left with a careful evaluation of the severity of LV dysfunction and HF clinical status to guide the risk assessment prior to ICD implantation.

In this context, we sought to determine the prognostic significance of the severity of pre-implantation hyponatremia on the risk of death, HF decompensation and appropriate ICD therapy for ventricular arrhythmias among a heterogeneous cohort of individuals with LV dysfunction undergoing ICD implantation for the prevention of SCD.

Methods

Study design and population

Preparation of this report was in accordance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement for reporting of observational studies.¹⁴ The study population consisted of all patients undergoing ICD implantation (n = 911) with a left ventricular ejection fraction (LVEF) ≤40% and an available serum sodium assessment at the time of ICD implant for the primary (n = 556) or secondary (n = 335) prevention of SCD between 1 December 1997 through 31 January 2007 at a tertiary care, community-based teaching hospital. The type of ICD device implanted (bi-ventricular, single chamber, or dual chamber), pacemaker functionality, back-up pacing rate, programmed detection zones for ventricular and atrial tachyarrhythmias and programming changes on follow-up were heterogeneous and occurred at the discretion of an electrophysiologist (R.K.Y, C.A.C, J.K.), and were implanted according to published guidelines. No patients with class IV HF status or active treatment with intravenous vasoactive or inotropic medications were implanted with an ICD. These data were extracted from a prospectively collected database and include information on pre-implantation co-morbidities, laboratory values, medications, procedural characteristics, tachyarrhythmia programming algorithms, and ICD therapy and outcomes that had been entered by device clinicians into an electronic medical record designed in Microsoft Access© (Redman WA). Patients were evaluated 1 week after ICD implantation and at approximately 3 months intervals until the end of follow-up or loss to follow-up. Each outpatient follow-up included a detailed clinical evaluation, an electrocardiogram and ICD interrogation. All authors verify data integrity and all analyses conducted. The study received institutional review board approval from our institution.

Serum sodium assessment and hyponatremia stratification groups

Data on baseline laboratory values and serum sodium concentrations were obtained 24–48 h prior to or at the time of ICD implantation. Hyponatremia was defined as a serum sodium ≤136 mmol/L and based on the definition and outcomes determined by Lee and Packer.¹⁵ The severity of hyponatremia (normal, mild, moderate, and severe) and corresponding serum sodium concentrations were stratified into the following groups:

• Overall population (n = 911)

  Mean serum sodium 137.0 ± 3.0. Serum sodium range 121–143 mmol/L

• Normal sodium ≥137 mmol/L (n = 507)

  Mean serum sodium 139.0 ± 2.0 mmol/L. Serum sodium range 137–143 mmol/L

• Mild hyponatremia 134–136 mmol/L (n = 268)

  Mean serum sodium 135.0 ± 0.8 mmol/L. Serum sodium range 134–136 mmol/L

• Moderate hyponatremia 131–133 mmol/L (n = 105)

  Mean serum sodium 132.3 ± 0.7 mmol/L. Serum sodium range 131–133 mmol/L

• Severe hyponatremia ≤130 mmol/L (n = 31)

  Mean serum sodium 128.0 ± 2.0. Serum sodium range 121–130.

Endpoints of the study

The individual endpoints of this study were all-cause mortality, hospitalization for acute decompensated heart failure (ADHF) and appropriate ICD therapy for ventricular arrhythmias during follow-up.

Mortality

All-cause mortality was assessed via a comprehensive search of the Social Security Death Registry last accessed September 2009 and adjudicated entries into the database.

Hospitalization for acute decompensated heart failure

Hospitalization for ADHF was identified at the time of ICD interrogation during incident hospitalization at our institution or at time of out-patient follow-up. All ADHF hospitalizations were then verified with a review of the inpatient medical record searched with ICD-9 codes 428.0 (congestive HF unspecified), 428.1 (left HF), and 428.20 (unspecified systolic HF).

Appropriate implantable cardioverter-defibrillator therapy

Time to the first appropriate ICD therapy was identified and defined as an episode of ventricular tachycardia/ventricular fibrillation (VT/VF) resulting in anti-tachycardia pacing or single/multiple shocks for arrhythmia termination. Uniform programming at the time of implantation was not required but the ICD was generally programmed with two to three detection zones with at least one VT zone that included ≥2 attempts at anti-tachycardia pacing followed by a shock. Ventricular fibrillation zones were set with detection of arrhythmia at least 200 b.p.m. and defibrillation therapy set at least 10 joules greater than the defibrillation threshold at implant or at the maximum energy output. Episodes of self terminating non-sustained VT were not included in this analysis. Ventricular tachycardia was defined by a uniform regular electrogram different from the baseline rhythm. Ventricular fibrillation was characterized by the presence of fibrillatory R waves and termination by defibrillation. Three electrophysiologists (R.K.Y, C.A.C, J.K.) analyzed stored electrograms at the time of ICD interrogation for all episodes of VT/VF resulting in therapy prior to inclusion into the database.

Statistical analysis

Continuous variables are presented as means with standard deviations and were compared using Student's t-test, analysis of variance (ANOVA) or the Mann–Whitney test where appropriate. Dichotomous variables are presented as percentages and standard deviations and compared using the χ² or Fisher's exact test where appropriate. Crude outcome rates (mortality, ADHF and appropriate ICD therapy) across hyponatremia stratification groups were compared by χ² tests or ANOVA where appropriate. The cumulative hazard of each outcome was examined with the use of multivariable Cox proportional hazard modelling and was conducted to control for potential confounders. All baseline variables demonstrating a significant association upon univariate analysis (P ≤ 0.10 for inclusion) between the occurrence of the endpoint (dependent variable = mortality or ADHF or appropriate ICD therapy) and treatment and co-morbidity characteristics including the severity of hyponatremia (independent variables) were entered into the multivariable model. The model was ultimately adjusted for age, pre-existing atrial tachyarrhythmias, LVEF, New York Heart Association (NYHA) class II-III HF, primary prevention indication for ICD implantation, implantation of a cardiac-resynchronization therapy defibrillator, prior coronary revascularization procedure medications including beta-blockers, angiotensin converting enzyme (ACE) inhibitors/angiotensin receptor blockers (ARBs), aldosterone receptor antagonist, furosemide, coumadin, statins, and antiarrhythmic therapy and laboratory values including serum sodium, blood–urea nitrogen (BUN), creatinine, white blood cell count, and red-cell distribution width.¹⁶ In the multivariable model, variables were selected by stepwise backward elimination and a P value of ≤0.05 was considered significant. Adjusted hazard ratios (AHR) and 95% confidence intervals (CIs) were calculated for all independent predictors. All analyses were performed with SPSS 15.0 (SPSS Inc.).

Results

Baseline characteristics

Baseline characteristics and laboratory values in the study population and hyponatremia stratification cohorts are listed in Table 1. Overall, the mean age of the study population was 67 ± 12 years, mean LVEF of 24 ± 8%, 78% were male and 61% implanted for primary prevention indications and 71% receiving non-cardiac resynchronization therapy defibrillators (CRT-D). A majority, 82% were receiving therapy with beta-blockers, 92% receiving ACE-inhibitors, ARBs or aldosterone-receptor antagonists and 60% receiving furosemide. The mean serum sodium was 137 ± 3 mmol/L, BUN 27 ± 16 mg/dL and creatinine 1.4 ± 1.1 mg/dL. Hyponatremia (serum sodium ≤136 mmol/L) was present among 44% (404 of 911) of patients and severe hyponatremia (serum sodium ≤130 mmol/L) was present among 3.4% (31 of 911).

A greater severity of hyponatremia was observed in patients with a lower LVEF, a higher incidence of NYHA class II-III HF symptoms, more CRT-D devices, a higher daily dose of furosemide, and higher concentrations of BUN and creatinine. A weak correlation was observed between LVEF and serum sodium concentration (r = 0.13, P = 0.01). Similarly a weak temporal effect over time and serum sodium was observed (r = −0.093, P = 0.005, Figure 1).

Prognostic association of hyponatremia and outcomes: overall and adjusted analyses

During a mean follow-up of 775 ± 750 days the overall incidence of death was 29% (n = 265), ADHF 12% (n = 108), and ICD therapy for VT/VF 36% (n = 332), respectively. As the severity of hyponatremia increased across the stratification cohorts (from a normal serum sodium to severe hyponatremia) an incremental incidence of death (25% to 61%, P < 0.001) and ADHF (11% to 26%, P = 0.004) was observed with a reduced incidence of ICD therapy for VT/VF (37–29%, P = 0.037, Figure 2).

Compared to individuals with a normal serum sodium, patients with either moderate or severe hyponatremia demonstrated a nearly two- and three-fold increased risk of death [AHR 1.73 (95% CI 1.18–2.52), P = 0.005] and [AHR 2.69 (95% CI 1.57–4.59), P = 0.004, Figure 3] and a similar risk increase in the probability of ADHF [AHR 1.85 (95% CI 1.10–3.13), P = 0.021] and [AHR 2.98 (95% CI 1.41–6.30), P = 0.004, Figure 4] upon multivariable analysis, respectively. Patients with severe hyponatremia demonstrated a lower probability of ICD therapy for VT/VF [AHR 0.68 (95% CI 0.27–0.88), P = 0.031, Figure 5] as compared to individuals with a normal serum sodium. Analyzed as a continuous variable, a 1 mmol/L reduction in serum sodium was associated with a 9% increase in the probability of death [AHR 1.09 (95% CI 1.03–1.14), P < 0.001] and an 8% increase in the probability of ADHF [AHR 1.08 (95% CI 1.01–1.16), P = 0.023]. In contrast, a 1 mmol/L reduction in serum sodium was associated with a 4% reduction in the probability of ICD therapy for VT/VF [AHR 0.96 (95% CI 0.92–0.99), P = 0.031].

The continuous relationship between serum sodium concentrations and outcomes of death, ADHF and ICD therapy for VT/VF censored at 3 years is illustrated in Figure 6. As the serum sodium decreased an increased cumulative incidence of death and ADHF with lower incidence of ICD therapy for VT/VF was demonstrated with the greatest effect among patients with severe hyponatremia. The associated independent predictors of death in addition to serum sodium are listed in Table 2.

Primary and secondary prevention populations

Analyses were conducted to refine these results and demonstrate the prognostic association of hyponatremia among patients with LV dysfunction and primary or secondary prevention implantation indications. Overall, 61% of patients were implanted with primary prevention (mean LVEF of 22 ± 8%, mean serum sodium 137 ± 3 mmol/L) and 39% with secondary prevention (mean LVEF of 25 ± 9%, mean serum sodium 136 ± 3 mmol/L) implantation indications. The cumulative probability of death during follow-up was 26% and 34% in the primary and secondary prevention subgroups, respectively. Patients with severe hyponatremia demonstrated a higher total mortality than those with normal serum sodium both among the cohort with primary prevention [67% vs. 20% AHR 2.87 (95% CI 1.46–5.66), P = 0.002] and secondary prevention [54% vs. 34%, AHR 3.16 (95% CI 1.41–7.10), P = 0.005] implantation indications.

Discussion

We describe an association between the severity of hyponatremia and the risk of death, HF hospitalization and appropriate ICD therapy for ventricular arrhythmias among a heterogeneous cohort of patients with LV dysfunction and ICDs. The presence of moderate and severe pre-implantation hyponatremia was strongly associated with an increased risk of death and ADHF. In contrast, the presence of severe hyponatremia was associated with a lower risk of appropriate ICD therapy for VT/VF. Hyponatremia is commonly observed among patients undergoing ICD implantation and the severity of hyponatremia is the strongest independent predictor of death.

Hyponatremia is commonly observed and remains an ominous marker for the progressions of myocardial remodeling among a broad spectrum of HF patients including those with decompensated HF¹, ambulatory and compensated HF², advanced end-stage HF treated with inotropic therapy³ and among a general populace with and without HF⁴. Hyponatremia is a maker for the progression of myocardial remodelling and is reflected by a multitude of physiologic mechanisms between the up-regulation of the sympathetic nervous system and an increase in plasma norepinephrine, activation of the renin–angiotensin–aldosterone system with increased circulating plasma renin and angiotensin-II⁵ and non-osmotic release of arginine vasopressin⁶ that lead to an inability of the body to excrete free water with subsequent fluid retention and hyponatremia. This complex interaction is in part the explanation of why the targeted treatment of total body water excess in HF with aquaretic therapy and vasopressin-receptor antagonists to improve long term outcomes has remained challenging.^17,18

A tenuous relationship exists between competing comorbidities, HF status and outcomes among ICD-eligible patients with LV dysfunction.^19–24 Important observations have demonstrated that there was no benefit of ICD therapy in reducing SCD among HF subjects with NYHA class III HF, whereas subjects with less severe, NYHA class II HF demonstrated a reduction in mortality with ICD therapy.²⁵ These findings may be explained by an increase in HF and non-arrhythmic related mortality observed among individuals with a more advanced HF status and emphasizes the continuous nature of HF progression among patients with LV dysfunction that extends beyond the risk of malignant ventricular arrhythmias.

In this context, Levy et al.²⁶ provide compelling evidence of risk heterogeneity among primary prevention ICD recipients by application of the Seattle Heart Failure Model (SHFM) score to the SCD-HeFT (Sudden Cardiac Death in Heart Failure Trial) cohort. The authors created a model to examine the relationship of baseline, predicted mortality, and survival benefit with ICD therapy and identified five categories of increasing mortality risk, ranging from 12% (quintile 1) to 50% (quintile 5) at 4 years by utilizing baseline characteristics, serum sodium, creatinine, and medical therapies. The fifth and highest risk group (mean serum sodium of 138 mEq/L and creatinine of 1.4 mg/dL) demonstrated an increased mortality with no benefit of ICD treatment despite the greatest incidence of appropriate ICD therapy for ventricular arrhythmias. Furthermore, an interaction of the SHFM score on total mortality suggested that the benefit of the ICD approached zero at approximately a 20–25% annual mortality rate. Extrapolated to our population, the severe hyponatremia cohort demonstrated an annual mortality rate of greater than 30%.

Important observations within the demographic profile among our cohort with severe hyponatremia revealed a lower EF, more symptomatic HF, CRT-D implantation, and a greater daily dosage of furosemide which suggest a more advanced HF status. In addition, less ACE-inhibitor usage and a higher BUN and creatinine may reflect intolerance to reverse-remodelling poly-pharmacy and may be markers of an increased severity of pump dysfunction. Among patients with decompensated NYHA class IV HF and availability of invasive hemodynamic measurements, Gheorghiade et al.⁷ reported that patients with persistent hyponatremia (mean serum sodium 130 mEq/L) demonstrated an elevated pulmonary capillary wedge pressure (mean 28 mmHg), cardiogenic shock (mean cardiac index of 1.8 L/min/m²) and hypotension (mean systolic blood pressure of 100 mmHg), high-risk findings for both 6 month⁷ and 1 year mortality.²⁷

There are two possible interpretations of our results. Firstly, compared to patients with LV dysfunction and normal serum sodium, those with severe hyponatremia are at increased risk of death and HF-related morbidity with a reduced incidence of appropriate ICD therapy and suggests that patients with severe hyponatremia may experience death not related to ventricular arrhythmias. Secondly, after a mean follow-up duration of 2 years, the cumulative incidence of appropriate ICD therapy was 29% among patients with severe hyponatremia. Such recipients of appropriate ICD shocks subsequently become a unique cohort at a higher mortality risk. We,²⁸ and others^29,30 have demonstrated an increase in mortality among appropriate shock therapy recipients, and that this risk is primarily mediated by an increase in HF decompensation.

Four distinct approaches have been suggested that may enhance the therapeutic efficiency of ICD implantation: improved risk stratification, improved therapy for HF, improved ICD programming and improved ICD technology.³¹ Comprehensive HF functional status assessments, particularly among patients with severe hyponatremia, are easily attainable metrics that may improve the risk–benefit ratio prior to ICD implantation and identify a vulnerable cohort of patients with LV dysfunction that may benefit with optimization of conventional medical therapy³² or advanced therapies such as ventricular assist devices or transplantation.

Limitations

While we employed methods to adjust for known confounders, we cannot control for unknown variables that may be different between the hyponatremia stratification groups. We are unable to examine continuous clinical data (e.g. 6 minute walk, cardiopulmonary stress testing), baseline biometrics (e.g. heart rate and blood pressure), or echocardiographic indices. A precise assessment of the NYHA HF functional status was not obtained. Changes in serum sodium concentrations or medical therapy were not determined and therefore the impact of a decrease or increase in serum sodium upon outcomes was not assessed. Different types of devices (single/dual chamber, CRT-D) were implanted without uniform back-up pacing settings. Programming strategies for ventricular arrhythmias was heterogeneous and hence we are unable to determine the impact of more or less aggressive anti-tachycardia pacing on the efficacy of VT therapy or the differential programming, e.g. number of intervals to detect, duration of the ventricular arrhythmia, rate cut-off that may impact the number of events detected. Although the occurrences of VT/VF were reduced among patients with severe hyponatremia, we cannot conclusively state that the risk of SCD is also reduced. Finally, the specific cause of death cannot be elucidated by the Social Security Death Registry.

Conclusions

Hyponatremia is commonly observed among ICD recipients with LV dysfunction. Patients with an increasing severity of hyponatremia become a unique ICD cohort at increased risk of death and HF-related morbidity with a reduced incidence of appropriate ICD therapy particularly among patients with severe hyponatremia.

Acknowledgements

We are indebted to Angel Rentas APRN, Thea Ling RN and David McComas RN, the ICD device clinicians that incorporated information into the ICD database.

Conflict of interest: Dr Yarlagadda has received honoraria and travel support from Medtronic; Dr Clyne receives research support from St Jude Medical; Dr Kluger receives research support from Boston Scientific and St Jude Medical and has received honoraria from Medtronic and Boston Scientific. Drs Bhavnani, Coleman, White and Mrs Guertin report no disclosures.

References