Author + information
- Received November 11, 2013
- Revision received January 26, 2014
- Accepted February 4, 2014
- Published online June 3, 2014.
- Ge Gao, MD, PhD∗,
- Vikram Brahmanandam, MD∗,
- Mihai Raicu, MS∗,
- Lianzhi Gu, MD, PhD∗,
- Li Zhou, MD, PhD∗,
- Srinivasan Kasturirangan, MD∗,
- Anish Shah, BS†,
- Smita I. Negi, MD‡,
- Melissa R. Wood, MD∗,
- Ankit A. Desai, MD∗,§,
- Antone Tatooles, MD‖,
- Alan Schwartz, PhD¶ and
- Samuel C. Dudley Jr., MD, PhD∗∗ ()
- ∗Section of Cardiology and the Jesse Brown VAMC, University of Illinois at Chicago, Chicago, Illinois
- †University of Illinois at Chicago College of Medicine, Chicago, Illinois
- ‡The University of Texas Health Science Center, Houston, Texas
- §Institute for Personalized Respiratory Medicine and Center for Cardiovascular Research, University of Illinois at Chicago, Chicago, Illinois
- ‖Transplant/Mechanical Assist, Advocate Christ Medical Center, Oak Lawn, Illinois
- ¶Departments of Medical Education and Pediatrics, University of Illinois at Chicago, Chicago, Illinois
- ↵∗Reprint requests and correspondence:
Dr. Samuel C. Dudley, Lifespan Cardiovascular Institute, The Warren Alpert Medical School of Brown University, 593 Eddy Street, APC 730, Providence, Rhode Island 02903.
Objectives The aim of this study was to determine the association of SCN5A cardiac sodium (Na+) channel mRNA splice variants in white blood cells (WBCs) with risk of arrhythmias in heart failure (HF).
Background HF is associated with upregulation of two cardiac SCN5A mRNA splice variants that encode prematurely truncated, nonfunctional Na+ channels. Because circulating WBCs demonstrate similar SCN5A splicing patterns, we hypothesized that these WBC-derived splice variants might further stratify patients with HF who are at risk for arrhythmias.
Methods Simultaneously obtained myocardial core samples and WBCs were compared for SCN5A variants C (VC) and D (VD). Circulating variant levels were compared among patients with HF, divided into three groups: HF without an implantable cardioverter-defibrillator (ICD), HF with an ICD without appropriate intervention, and HF with an ICD with appropriate intervention.
Results Myocardial tissue–derived SCN5A variant expression levels strongly correlated with circulating WBC samples for both VC and VD variants (r = 0.78 and 0.75, respectively). After controlling for covariates, patients with HF who had received an appropriate ICD intervention had higher expression levels of both WBC-derived SCN5A variants compared with patients with HF with ICDs who had not received appropriate ICD intervention (odds ratio, 3.25; 95% CI, 1.64–6.45; p = 0.001). Receiver operating characteristic analysis revealed that circulating SCN5A variant levels were highly associated with the risk for appropriate ICD intervention (area under the curve ≥0.97).
Conclusions Circulating expression levels of SCN5A variants were strongly associated with myocardial tissue levels. Furthermore, circulating variant levels were correlative with arrhythmic risk as measured by ICD events in an HF population within 1 year. (Sodium Channel Splicing in Heart Failure Trial [SOCS-HEFT]; NCT01185587)
Heart failure (HF) represents a growing global healthcare concern. HF is increasing in prevalence, and up to half of all patients with HF have arrhythmic sudden death (1,2). Currently, placement of an implanted cardioverter-defibrillator (ICD) is an established interventional therapy to decrease the risk of arrhythmia-related sudden death in patients with HF. Both the American College of Cardiology and the American Heart Association endorse the placement of ICDs to reduce total mortality as part of their national guidelines for primary prevention of sudden cardiac death in high-risk patients with HF (3). On the basis of the criteria for determining “high risk” by these guidelines, however, up to 70% of patients who receive an ICD never have a malignant arrhythmia (4,5), and somewhere between 15% and 40% of patients who are eligible for an ICD never receive one (6). Moreover, the majority of sudden deaths occur in patients with HF who do not meet the current criteria for ICD implantation (7–9). These data suggest that current risk stratification with the use of markers such as left ventricular ejection fraction alone is suboptimal (10). Other methods used for risk stratification include signal-averaged electrocardiography (sensitivity of 62.4% and specificity of 77.4% at 2 years) (11), T-wave alternans (sensitivity of 74% and specificity of 44% at 1 year) (12), and invasive electrophysiological testing (sensitivity of 62% and specificity of 62% at 1 year) (11), techniques that are not widely used because of poor accuracy as well as equipment and personnel costs required for implementation. In addition, although risk may change over time, these more demanding techniques are often limited to a single assessment per patient. Therefore, there is an unmet need for a convenient, inexpensive, and non-invasive test to stratify risk for sudden cardiac death and arrhythmias in the HF population.
Alternative mRNA splicing is a post-transcriptional mechanism that can change substantially the pattern of gene expression by creating a variety of gene products from a single DNA message. Up to 95% of multi-exon human genes have alternative spliced forms, which suggests that alternative splicing is one of the most significant components of the functional complexity of the human genome (13,14). We previously reported that both angiotensin II and hypoxia, signals common to HF, increase two myocardial splicing factors, RBM25 and associated factor LUC7L3 (15,16). The activated RMB25/LUC7L3 splicing complex increases SCN5A C (VC) and D (VD) variants, decreases the full-length SCN5A transcript and protein, and decreases Na+ current (17). Interestingly, HF results in Na+ current reductions in the range of those seen in Brugada syndrome, an inherited arrhythmogenic condition at high risk for sudden cardiac death (16,18). SCN5A variants result from splicing at cryptic splice sequences in the terminal exon of SCN5A (exon 28) (16,19). SCN5A variants are shorter and encode prematurely truncated, nonfunctional Na+ channel proteins missing part of the C terminus and can represent >50% of the SCN5A transcripts during HF (16,19). A mouse model of this degree of variant expression showed an 80% reduction in cardiac Na+ current, a significant reduction in myocardial conduction velocity, and an increase in arrhythmic risk (19).
SCN5A transcripts and variants have been noted in circulating white blood cells (WBCs) (16,19). Circulating molecular biomarkers are attractive because of the convenience and comfort of access compared with more traditional methods. Therefore, we sought to determine the association of expression levels of SCN5A cardiac sodium (Na+) channel mRNA splice variants in WBCs with risk for arrhythmias in HF.
Correlation of cardiac tissue and circulating levels of VC and VD variants
Simultaneous human blood and heart tissue were obtained with an institutional review board–approved protocol (2009–0881) at Christ Advocate Hospital from patients undergoing left ventricular assist device implantation. These patients were not included in the clinical trial. The characteristics of these patients are presented in Table 1.
Clinical characteristics of the study population and recruitment criteria
This was a cross-sectional, cohort, comparison trial entitled “Sodium Channel Splicing in Heart Failure Trial” (SOCS-HEFT, NCT01185587) and conducted at the University of Illinois at Chicago and the Jesse Brown Veterans Administration Medical Center (JBVAMC) in Chicago, Illinois. The study was approved by the Collaborative University of Illinois at Chicago/Northwestern/JBVAMC Institutional Review Board. All study subjects signed a written, informed consent before enrollment. Subjects were adult patients (age ≥18 years) with systolic HF (defined by echocardiography-derived left ventricular ejection fraction or LVEF ≤50%). The subjects were assigned to four groups: those who did not have HF (i.e., control); those with HF without an ICD (HF); those with HF, an ICD, and no evidence of appropriate event-driven therapy [ICD(–)Event]; and finally, those with HF, an ICD, and evidence of appropriate event-driven therapy [ICD(+)Event]. Control patients were defined by normal left ventricular systolic and diastolic function by echocardiographic assessment. The pre-enrollment evaluation for all groups included reviewing the electronic medical records and subject interviews for history, a physical examination, and current medication at the time of enrollment. Demographic data obtained included age, race, body mass index, and New York Heart Association (NYHA) functional class.
ICD implantation was performed at least 1 year before enrollment. An appropriate ICD event was adjudicated by an independent, blinded, clinical cardiac electrophysiologist. An “event” was defined as any device therapy delivered to interrupt ventricular fibrillation or ventricular tachycardia excluding antitachycardia pacing. ICD programming was at the discretion of the attending physician. The ICD implant indication was predominantly primary prevention (77%). All LVEF determinations were made by means of echocardiography or cardiac magnetic resonance imaging. LVEF was determined in a 2-year window before enrollment.
Any patient with a history of congenital heart disease or use of illicit drugs was excluded. Patients receiving immunosuppressive medications or who had evidence of a chronic infection, acute or chronic inflammatory illness, or any illness expected to result in death within 18 months of enrollment were excluded. Control patients had to be free of HF symptoms, diastolic dysfunction, and left ventricular systolic dysfunction documented by any cardiac imaging or diagnosis in the electronic medical record within 1 year of study enrollment. Other exclusion criteria for the control group included long-QT syndrome, Brugada syndrome, or a history of significant illness (i.e., myocardial infarction, cardiac hospitalization, cardiac arrhythmia, infection, or cancer) within 12 months of study enrollment.
Blood samples were collected in PAXgene Blood RNA tubes (Becton Dickinson, Franklin Lakes, New Jersey), following the manufacturer’s procedure. Samples were stored for up to 3 days at room temperature or 5 days at 2°C to 8°C. Total RNA was isolated with the use of the PAXgene Blood RNA isolation kit and was then converted to cDNA with the use of the High Capacity cDNA Reverse Transcription Kit (Qiagen, Valencia, California).
Total RNA was isolated from WBCs and human heart tissue with the use of the RNeasy Mini and RNeasy Lipid Tissue Mini Kits, respectively (Qiagen), and then were converted to cDNA with the use of the High Capacity cDNA Reverse Transcription Kit (Qiagen). Only samples with an optical density of 260/280 >1.8, an optical density of 260/230 >1.5, and total RNA >6 μg were used. Both tissue and blood samples were stored in liquid nitrogen. Under these conditions, repeated measures of the same sample varied by <2%. Quantitative real-time reverse-transcriptase polymerase chain reaction (qRT-PCR) was performed to detect the abundance of SCN5A variants by use of iQ SYBR Green Supermix (Bio-Rad) and 7500 Fast Real-Time PCR System (Life Technologies). The primer sequences used were HE27F (5′-CTGCGCCACTACTACTTCACCAACA-3′); HSCN5AE28A/R (5′-GGAAGAGCGTCGGGGAGAAGAAGTA-3′); HSCN5AE28C/R (5′-TCTCTTCTCCCCTCCTGCTGGTCA-3′); and HSCN5AE28D/R (5′-GGAAGAGCGTCGGGGAGAAGAAGTA-3′). qRT-PCR thermal cycling conditions were an initial uracil-N-glycosylase incubation at 50°C for 2 minutes. iTaq DNA polymerase was activated with an initial denaturation step at 95°C for 5 minutes, followed by cycles of denaturation at 95°C for 15 seconds and annealing and extension at 60°C for 1 minute. Each sample was measured for the target gene SCN5A, VC, VD, and β-actin. Samples were run in triplicate and averaged. Representative qRT-PCR amplification plots and the sample data are shown in Online Figure 1 and Online Table 1. To correct for WBC SCN5A expression between subjects, variants levels were expressed as a percentage of the variant with respect to the total Na+ channel mRNA including variants and normalized to the level of β-actin.
Age, sex, race, ischemia (defined as a chart review revealing a diagnosis of coronary artery disease, ischemic cardiomyopathy, previous coronary bypass surgery, previous percutaneous coronary intervention, or test results indicating obstructive coronary artery disease), LVEF, medications, NYHA class, and QRS duration measurements were recorded. Clinical characteristics were reported as mean ± SD for continuous variables and frequencies for categorical variables. Differences between the groups were examined by means of t tests and chi-square tests for continuous and categorical variables, respectively. Results with values of p < 0.05 were considered statistically significant in all analyses.
Linear regression, on the basis of ordinary least squares, was used to determine the degree of correlation between normalized variant levels in the ventricle and blood. A probability value <0.05 was taken to indicate statistical correlation. The diagnostic odds ratio (DOR) is an overall measure of diagnostic accuracy that combines both sensitivity and specificity: [sensitivity/(1–sensitivity)]/[(1–specificity)/specificity]. We compared the summary DORs and their corresponding 95% CIs across different diagnostic predictors: normalized variants VC and VD in the blood, NYHA class III/IV, angiotensin-converting enzyme (ACE) inhibitors, antiarrhythmic drugs, LVEF ≤20%, and QRS duration ≥120 ms. Univariate analysis was performed to calculate DORs and their corresponding 95% CIs.
Receiver operating characteristic (ROC) curves were generated for both splicing variants and LVEF ≤20%. Sensitivity (the proportion of true-positive ICD patients with an event) and the specificity (the proportion of ICD patients without an event) were evaluated. A commonly used measure of overall diagnostic effectiveness is the Youden index, which is defined as (sensitivity+specificity)–1. We determined the optimal cutoff value by maximizing the Youden index. The sensitivities and specificities were calculated from the data across all possible cutoff values within the range of the test results, and we selected the cutoff value leading to the highest Youden index.
Correlation of cardiac tissue and circulating levels of VC and VD variants
Paired analysis of SCN5A variants from circulating WBC and ventricular tissue demonstrated strong correlation (Fig. 1) (r = 0.78 and 0.75, respectively for variants VC and VD), which demonstrates that WBC-derived expression of SCN5A variants are correlative of levels in myocardial tissue. Online Figure 2 shows that the variant levels were independent of the WBC count.
SCN5A variant expression in HF
Table 2 shows the clinical characteristics of patients enrolled in the study. By omnibus F test in the one-way analysis of variance, both VC and VD varied with NYHA class, β-blocker use, and antiarrhythmic drug use (amiodarone in all cases except for three patients in the ICD(+)Event group whose records indicated use of digoxin, sotalol, and an unspecified drug, respectively) (Online Table 2). VC but not VD was influenced by QRS duration. Figure 2 displays median and interquartile ranges (IQR) for VC and VD across the three groups. Patients in the HF and ICD(–)Event group exhibited significantly lower median and IQR for both WBC-derived VC and VD variants compared with patients in the ICD(+)Event group [VC: HF, 2.1 (IQR: 1.7 to 2.5); ICD(–)Event, 2.5 (IQR: 2.1 to 3.1); ICD(+)Event, 7.3 (IQR: 6.8 to 8.4); VD: HF, 2.8 (IQR: 2.5 to 3.1); ICD(–)Event, 1.7 (IQR: 1.1 to 2.0); ICD(+)Event, 6.6 (IQR: 6.3 to 7.8). The expression of VC and VD variants were also significantly increased in all three HF groups (HF, ICD(–)Event, ICD(+)Event) compared with the control group (p < 0.05).
Effect of population characteristics on SCN5A variant expression
There was no difference in the expression of VC and VD across races (p = NS) (Figs. 3A and 3B), between sexes (p = NS) (Figs. 3C and 3D), between subjects with an ischemia history and those without an ischemia history (p = NS) (Figs. 3E and 3F), or between subjects with an LVEF ≤20% and those with LVEF >20% (p = NS) (Figs. 3G and 3H). Worsening NYHA class, however, was associated with an induction of both WBC-derived SCN5A variants (NYHA class I to II versus NYHA class III to IV: 2.8 ± 1.7 vs. 4.1 ± 2.7 and 2.5 ± 1.6 vs. 3.8 ± 2.4 for VC and VD, respectively, p < 0.05 for each) (Figs. 3I and 3J). VC expression also demonstrated significant changes between QRS duration ≤120 ms vs. >120 ms (3.0 ± 2.1 vs. 4.2 ± 2.6, respectively, p < 0.05) (Fig. 3K). A similar trend toward significance was evident in VD expression (2.9 ± 1.9 vs. 3.6 ± 2.4, p > 0.05) (Fig. 3L).
Predictors of ICD events
By univariate analysis (Fig. 4), NYHA class III/IV (DOR, 7.65; 95% CI: 2.17 to 27.00), antiarrhythmic drug use (DOR, 5.85; 95% CI: 1.51 to 22.70), WBC-derived VC expression (DOR, 3.85; 95% CI: 2.00 to 7.43), WBC-derived VD expression (DOR, 3.04; 95% CI: 1.83 to 5.05), and QRS duration ≥120 ms (DOR, 3.19; 95% CI: 1.03 to 9.89) demonstrated association with increased risk of an ICD event. In contrast, an LVEF ≤20% (DOR, 2.11; 95% CI: 0.22 to 20.12) and ACE inhibitors (DOR, 1.12; 95% CI: 0.36 to 3.54) were not associated with ICD events.
Sensitivity and specificity of SCN5A variants for determination of ICD events
ROC curves were generated to evaluate the performance of the expression of the WBC-derived variants in distinguishing between the ICD patients with and without the events and then were compared with those generated for an LVEF ≤20%. The area under the ROC curve was 0.98 (95% CI: 0.95 to 1.00), 0.97 (95% CI: 0.93 to 1.00), and 0.56 (95% CI: 0.41 to 0.71) for VC, VD, and LVEF ≤20%, respectively (Fig. 5). The values for the optimal Youden index and cutoff as well as the corresponding maximum sensitivities and specificities are shown in Table 3. To address any model overfitting, we performed a 7-fold cross-validation of each logistic regression associated with ICD intervention (from VC, VD, and LVEF <20%) and plotted the distribution of sensitivity and specificity values for cutoffs on the basis of the regressions on the ROC curves. The plots (Online Fig. 3) illustrate that both VC and VD variants provided performance superior to LVEF in our data set.
Alterations in sodium current, the main current for cardiac conduction, are associated with arrhythmogenesis (20). Since the cloning of SCN5A, encoding the α-subunit of the Na+ channel (21), hundreds of mutations have been reported to cause inherited sudden death syndromes such as Brugada syndrome, the third variant of long-QT syndrome, and sudden infant death (20,22). Moreover, we have shown previously that abnormal mRNA splicing results in SCN5A variants that can contribute to arrhythmic risk and that these variants are increased in HF (16,17,19).
With the use of left ventricular assist device core samples and blood samples from the same patient, we now show a significant correlation between normalized variant expression levels in the heart and blood. The results also indicate a graded association between levels of circulating SCN5A variants and an increasing risk of ICD events in patients from control to HF to HF with ICD events. Patients with HF who had received appropriate ICD intervention had significantly higher levels of SCN5A splice variants compared with control patients and patients who had not received an intervention. As expected, patients with HF with an ICD and those without an ICD but with no intervention had similar variant levels. Moreover, the separation between groups allowed for further discriminatory power to risk-stratify patients with HF.
The discriminatory capacity of the variant expression to identify groups with and groups without ICD events was independent of sex, race, etiology of myopathy, or severity of LVEF. SCN5A variant expression was increased with higher NYHA class, consistent with the notion that these patients have worsening HF symptoms and are at higher risk from HF complications such as arrhythmias. Because longer QRS duration is a manifestation of cardiac conduction disease and is associated with reduced functional sodium channels (23–26), observations of higher variant levels in patients with longer QRS duration were also consistent. Interestingly, variant levels were not associated with severity of LVEF. Because LVEF severity beyond the initial threshold of 30% or 35% has not been a reliable indicator for further risk stratification for ICD implantation (27–29), the current data indicate that measures of circulating variant expression levels may give added information to risk reflected by left ventricular function. Because of the cardiac-specific nature of the pathophysiological role of SCN5A variants and the high degree of correlation between WBC-derived SCN5A variant levels and ICD interventions, we speculate that circulating SCN5A variants may supplement current methods to improve discrimination of patients who will most probably benefit from ICD implantation.
There are a number of limitations to this study. The small sample size may limit the applicability of the findings to larger population-level cohorts. For example, sample size may have masked weak associations of variants with other potential covariates or identification of other predictors of arrhythmic risk. Because of the retrospective imaging and ICD data, unforeseen biases may have been introduced that might reduce the power of WBC-derived variants. Analysis of patients with ICD device interventions for nonlethal arrhythmias, however, revealed that device intervention alone did not alter SCN5A variants levels (data not shown). Additionally, ICD programming was not controlled in this study design. Therefore, potential nonsustained events may have been counted, altering the calculated power of the variant levels. Exclusion of anti-tachycardia pacing as an event mitigates some of this concern. Although this study suggested that the association of SCN5A mRNA splice variants was independent of race, the total number of patients and limited numbers of Caucasian, Hispanic, or Asian patients make it difficult to be certain that the findings apply similarly to all racial groups. We did not evaluate the correlation of variants with multiple ICD interventions in the ICD(+)Event group or over longer than the defined 1-year period. Despite the correlation between elevated levels of VC and VD with ICD events, it is possible that not all arrhythmic conditions will be similarly associated. Finally, the cost/benefit ratio of any combination of predictive parameters remains to be determined.
We have shown that levels of circulating WBC-derived SCN5A mRNA variants are representative of levels in the myocardium. Moreover, the SCN5A variant levels increased with risk for sudden cardiac death, and variant levels were significantly elevated in patients who received an ICD intervention. The degree of separation of variant levels between patients with HF with and those without an ICD intervention suggests that variant levels had a strong power to discriminate between these two groups. If true in prospective validation trials, WBC SCN5A variant level determinations may help identify which patients with HF might benefit most from device implantation.
This research was funded by the National Institutes of Health grants P01 HL058000 (SCD), R01 HL1024025 (SCD), R01 HL106592 (SCD), Veterans Administration Merit Award (SCD), R41 HL112355 (3PrimeDx), and National Center for Research Resources/National Center for Advancing Translational Sciences UL1RR029879 (AAD). Dr. Dudley is the inventor on patent applications: 1) SCN5A Splice Variants for Use in Methods Relating to Sudden Cardiac Death and Need for Implanted Cardiac Defibrillators, PCT/US2012/20564; and 2) SCN5A Splicing Factors and Splice Variants for Use in Diagnostic and Prognostic Methods, 13/291,826.
All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- angiotensin-converting enzyme
- diagnostic odds ratio
- heart failure
- implantable cardioverter-defibrillator
- interquartile range
- left ventricular ejection fraction
- New York Heart Association
- receiver operating characteristic
- SCN5A variant C
- SCN5A variant D
- quantitative real-time reverse-transcriptase polymerase chain reaction
- white blood cell
- Received November 11, 2013.
- Revision received January 26, 2014.
- Accepted February 4, 2014.
- American College of Cardiology Foundation
- Roger V.L.,
- Go A.S.,
- Lloyd-Jones D.M.,
- et al.
- Hunt S.A.,
- Abraham W.T.,
- Chin M.H.,
- et al.
- LaPointe N.M.,
- Al-Khatib S.M.,
- Piccini J.P.,
- et al.
- Goldenberg I.,
- Vyas A.K.,
- Hall W.J.,
- et al.
- Levy W.C.,
- Lee K.L.,
- Hellkamp A.S.,
- et al.
- Bilchick K.C.,
- Stukenborg G.J.,
- Kamath S.,
- Cheng A.
- Huikuri H.V.,
- Makikallio T.H.,
- Raatikainen M.J.,
- et al.
- Bailey J.J.,
- Berson A.S.,
- Handelsman H.,
- Hodges M.
- Costantini O.,
- Hohnloser S.H.,
- Kirk M.M.,
- et al.
- Zhou A.,
- Ou A.C.,
- Cho A.,
- et al.
- Gao G.,
- Xie A.,
- Huang S.C.,
- et al.
- Gao G.,
- Xie A.,
- Zhang J.,
- et al.
- Shang L.L.,
- Pfahnl A.E.,
- Sanyal S.,
- et al.
- Gellens M.E.,
- George A.L. Jr..,
- Chen L.Q.,
- et al.
- Santos L.F.,
- Rodrigues B.,
- Moreira D.,
- et al.
- Kanter R.J.,
- Pfeiffer R.,
- Hu D.,
- et al.
- Royer A.,
- van Veen T.A.,
- Le Bouter S.,
- et al.
- Buxton A.E.,
- Lee K.L.,
- Hafley G.E.,
- et al.