Author + information
- Received August 17, 2009
- Revision received January 19, 2010
- Accepted January 20, 2010
- Published online May 18, 2010.
- Arnold F. Jacobson, MD, PhD⁎,
- Roxy Senior, MD†,
- Manuel D. Cerqueira, MD‡,
- Nathan D. Wong, PhD§,
- Gregory S. Thomas, MD, MPH§,
- Victor A. Lopez, BS§,
- Denis Agostini, MD, PhD∥,
- Fred Weiland, MD¶,
- Harish Chandna, MD#,
- Jagat Narula, MD, PhD§,⁎ (, )
- ADMIRE-HF Investigators
- ↵⁎Reprint requests and correspondence:
Dr. Jagat Narula, University of California at Irvine, 333 City Boulevard West, Suite 400, Orange, California 92868-4080
Objectives The ADMIRE-HF (AdreView Myocardial Imaging for Risk Evaluation in Heart Failure) study prospectively evaluated iodine-123 meta-iodobenzylguanidine (123I-mIBG) imaging for identifying symptomatic heart failure (HF) patients most likely to experience cardiac events.
Background Single-center studies have demonstrated the poorer prognosis of HF patients with reduced 123I-mIBG myocardial uptake, but these observations have not been validated in large multicenter trials.
Methods A total of 961 subjects with New York Heart Association (NYHA) functional class II/III HF and left ventricular ejection fraction (LVEF) ≤35% were studied. Subjects underwent 123I-mIBG myocardial imaging (sympathetic neuronal integrity quantified as the heart/mediastinum uptake ratio [H/M] on 4-h delayed planar images) and myocardial perfusion imaging and were then followed up for up to 2 years. Time to first occurrence of NYHA functional class progression, potentially life-threatening arrhythmic event, or cardiac death was compared with H/M (either in relation to estimated lower limit of normal [1.60] or as a continuous variable) using Cox proportional hazards regression. Multivariable analyses using clinical, laboratory, and imaging data were also performed.
Results A total of 237 subjects (25%) experienced events (median follow-up 17 months). The hazard ratio for H/M ≥1.60 was 0.40 (p < 0.001); the hazard ratio for continuous H/M was 0.22 (p < 0.001). Two-year event rate was 15% for H/M ≥1.60 and 37% for H/M <1.60; hazard ratios for individual event categories were as follows: HF progression, 0.49 (p = 0.002); arrhythmic events, 0.37 (p = 0.02); and cardiac death, 0.14 (p = 0.006). Significant contributors to the multivariable model were H/M, LVEF, B-type natriuretic peptide, and NYHA functional class. 123I-mIBG imaging also provided additional discrimination in analyses of interactions between B-type natriuretic peptide, LVEF, and H/M.
Conclusions ADMIRE-HF provides prospective validation of the independent prognostic value of 123I-mIBG scintigraphy in assessment of patients with HF. (Meta-Iodobenzylguanidine Scintigraphy Imaging in Patients With Heart Failure and Control Subjects Without Cardiovascular Disease, NCT00126425; Meta-Iodobenzylguanidine [123I-mIBG] Scintigraphy Imaging in Patients With Heart Failure and Control Subjects Without Cardiovascular Disease, NCT00126438)
Increased myocardial sympathetic activity is a prominent feature of heart failure (HF) and is associated with progressive myocardial remodeling, inexorable decline in left ventricular function, and worsening symptoms (1–3). Alterations in myocardial sympathetic nerve activity also play an important role in the generation of ventricular arrhythmias and sudden cardiac death (SCD) (4,5). Increased neuronal release of norepinephrine (NE) is usually accompanied by decreased neuronal NE reuptake due to post-transcriptional downregulation of the cardiac NE transporter (6–8). The resultant increase in NE concentration in the sympathetic synaptic cleft induces desensitization of myocardial beta-adrenoceptors (9,10). Adrenoreceptor inhibitors counter such alterations and improve survival by retarding HF progression and preventing tachyarrhythmias (11,12). Accordingly, interrogation of myocardial sympathetic nervous system activity has been suggested as an aid to assessment of prognosis and clinical management of HF patients (13,14).
The decrease in the NE reuptake mechanism has been successfully assessed by radionuclide imaging with the iodine-123–labeled NE analog meta-iodobenzylguanidine (123I-mIBG), which has demonstrated an excellent safety profile during more than 20 years of clinical use (5,13–15). Uptake of 123I-mIBG into myocardial sympathetic nerve endings is mediated by the NE transporter, and because the compound is not metabolized, the amount of 123I-mIBG retention over several hours after administration is a reflection of neuronal integrity (15). Reduced myocardial 123I-mIBG uptake has been demonstrated to be an independent predictor of adverse long-term outcome, and improvement in 123I-mIBG uptake is observed in response to effective HF therapy (13,16–21). Although there is extensive literature on 123I-mIBG imaging in both ischemic and nonischemic cardiomyopathy, most studies have been conducted at single centers involving relatively small numbers of patients and have not been performed under rigorous clinical trial conditions. As such, the potential usefulness of 123I-mIBG imaging in the clinical management of HF patients has remained uncertain.
The ADMIRE-HF (AdreView Myocardial Imaging for Risk Evaluation in Heart Failure) study consisted of 2 identical open-label phase III studies to provide prospective validation of the prognostic role of quantitation of sympathetic innervation of the myocardium using 123I-mIBG. This paper presents the combined primary efficacy results from the 2 ADMIRE-HF studies.
Ninety-six sites in North America (U.S. and Canada) and Europe participated in ADMIRE-HF. The study was approved by the institutional review boards and ethics committees at each center, and all subjects signed informed consent before performance of any procedures. The first subject was imaged on July 27, 2005; the last subject was imaged on February 20, 2008.
The methods used in the trial have been described in detail previously (22). The primary inclusion criteria were HF (New York Heart Association [NYHA] functional class II or III) due to ischemic or nonischemic cardiomyopathy; left ventricular ejection fraction (LVEF) ≤35%; and guidelines-based optimum pharmacotherapy including beta-blocker, angiotensin-converting enzyme inhibitor, and/or angiotensin receptor blocker (ARB). Major exclusion criteria were functioning ventricular pacemaker; history of defibrillation to treat a previous ventricular arrhythmic event; cardiac revascularization, implantable cardioverter-defibrillator (ICD) implantation or acute myocardial infarction within previous 30 days; and serum creatinine >3.0 mg/dl (265 μmol/l) (22). Within 30 days before imaging, all subjects underwent a complete clinical evaluation, NYHA functional class assessment, echocardiography, and blood draw for plasma B-type natriuretic peptide (BNP) and NE levels. The echocardiographic data and blood work were submitted to separate core laboratories for analysis.
Radionuclide imaging and analysis procedures
Imaging procedures are described in detail in the Online Appendix. Briefly, all subjects received 10 mCi (370 MBq; ±10%) of 123I-mIBG (AdreView, GE Healthcare) and underwent anterior planar and single-photon emission computed tomography (SPECT) imaging of the thorax beginning at 15 min (“early”) and 3 h 50 min (“late”) post-injection (Fig. 1).On a separate day, myocardial perfusion imaging (MPI) with technetium-99m (99mTc)-tetrofosmin (MyoView, GE Healthcare) was performed as previously described (22). All images were processed by certified nuclear medicine technologists and interpreted at an independent core laboratory (Icon Medical Imaging, Warrington, Pennsylvania). All planar and SPECT images were reviewed by 3 expert independent nuclear cardiologists who were blinded to clinical data.
The heart/mediastinum ratio (H/M) was determined from the counts/pixel in a visually drawn heart region of interest divided by the counts/pixel in a 7×7 pixel mediastinum region of interest in the mid-line upper chest positioned to reflect the location with lowest activity (i.e., nonspecific background). Each reader validated the H/M determined for each planar image and scored all SPECT image sets employing a 17-segment model and a scale of 0 to 4 (23).
Clinical follow-up and event adjudication
All subjects received standard clinical care and were followed up until: 1) the subject completed a 2-year period after the 123I-mIBG study; 2) subject death was confirmed; 3) the subject withdrew from the study or was lost to follow-up; or 4) the trial was terminated because the protocol-specified number of cardiac events (CEs) had been confirmed. Complete details are provided in the Online Appendix.
The Clinical Adjudication Committee reviewed data from case report forms and source documents to confirm occurrence of CEs, specifically: 1) HF progression: increase in symptomatic status from NYHA functional class II to III or IV, or increase from NYHA class III to class IV; 2) potentially life-threatening arrhythmic event, including documented episode of spontaneous sustained (>30 s) ventricular tachyarrhythmia, resuscitated cardiac arrest, or appropriate ICD discharge (antitachycardia pacing or defibrillation); or 3) cardiac death (further classified as due to HF progression, sudden cardiac death [SCD], or other causes).
Statistical analyses were performed at the University of California, Irvine (by V.A.L. and N.D.W.). The univariate primary and secondary analyses were performed on the pooled ADMIRE-HF dataset using consensus H/M results for each subject. The pre-specified primary end point was the late H/M in relation to time-to-occurrence of the first CE in trial subjects. This single 123I-mIBG value was selected for the primary analysis because of the extensive literature supporting its prognostic significance (24,25). Complete 123I-mIBG and MPI SPECT data were collected for exploratory evaluations, and these data were included in preplanned secondary analyses for each of the 2 trials. Complete details of statistical analyses are provided in the Online Appendix.
The primary analysis was based on subjects classified into 2 pre-specified groups, late H/M <1.6 and ≥1.6 (22). The analysis employed a univariate Cox proportional hazards model fitted to the time to first occurrence of a CE (composite of 3 CE categories) (26). Secondary Cox proportional hazards analyses were performed using the continuous numerical H/M rather than the binary division. Additional Cox proportional hazards analyses were performed for each of the 3 event categories, providing estimates of the hazard ratios for occurrence of each event category independent of the others (e.g., comparing subjects who had an HF progression CE vs. those who did not). Only the time to occurrence of the first event in a category for a given subject was used. Kaplan-Meier survival analyses were also performed on the primary and secondary end point data.
All multivariable analyses employed Cox proportional hazards methods with backwards elimination. For each phase III study, separate multivariable analyses were performed using the SPECT interpretations for each blinded reader. Additional multivariable analyses were performed on the pooled ADMIRE-HF data using only the planar (early and late) 123I-mIBG results; clinical variables such as age, sex, serum creatinine, systolic blood pressure; and cardiac risk factors such as hypertension, dyslipidemia, and diabetes. In the protocol-specified multivariable analyses, myocardial washout was not included because it is derived from the same data used in calculating early and late H/M. However, as the prognostic significance of washout is widely cited in the literature (24,25), additional analyses examining the impact of this variable were performed.
Statistical significance for the univariate Cox proportional hazards analyses of the composite primary end point was set at p < 0.025 (alpha = 0.05 split between the analyses using H/M divided at 1.60 and as a continuous variable). All other analyses used a significance level of p < 0.05.
123I-mIBG studies were performed in 985 subjects (453 and 532 subjects in the 2 phase III trials). Of these 985 subjects, 21 subjects were withdrawn because of protocol violations (only discovered post-dosing; n = 17), adverse events (n = 3; nausea; hypoglycemia secondary to insulin administration; acute myocardial infarction in subject with ischemic electrocardiography changes before 123I-mIBG dosing), or subject request on the day of dosing (n = 1). In addition, images from 2 subjects were not submitted for central reading, and planar images from 1 subject were considered nondiagnostic by the readers. Therefore, the evaluable efficacy population consisted of 961 subjects. The demographic information for those subjects is summarized in Table 1.
During the follow-up period of 2 days to 30.4 months (median 17 months), 237 first CEs were observed: HF progression in 163 subjects, arrhythmic events in 50 subjects, and cardiac death in 24 subjects. Fifty-two subjects had CEs in more than 1 category during follow-up, resulting in the following subject totals in each CE category: HF progression, n = 176; arrhythmic events, n = 64; cardiac death, n = 53. The total number of deaths during the trial was 81 (8%; 28 noncardiac). Characteristics of subjects based on occurrence of CEs are compared in Table 2.There was no difference in the age, sex, and medication usage of subjects who did and did not experience CEs. Subjects who experienced events were more likely to have NYHA functional class III symptoms, HF of nonischemic etiology, lower LVEF, and higher levels of BNP and NE.
The mean late H/M was 1.44 (standard deviation 0.20), and the distribution of this parameter was symmetric (first, second, and third quartiles of 1.30, 1.42, and 1.57, respectively). A total of 201 subjects (21%) had H/M ≥1.60 (protocol-defined binary division) (22).
The CE risk (primary end point) was significantly lower for subjects with H/M ≥1.60, with hazard ratio of 0.40 (97.5% CI: 0.25 to 0.64; p < 0.001). Survival analysis showed 2-year event rates of 38% versus 15% for the 2 H/M groups (Fig. 2).The Cox proportional hazards analysis based on the continuous numerical H/M demonstrated an even lower hazard ratio (0.22; 97.5% CI: 0.10 to 0.47; p < 0.001). The analyses based on the individual event categories showed significant differences for all 3, with the lowest hazard ratio for subjects with high H/M being for cardiac death (0.14; 95% CI: 0.03 to 0.58; p = 0.006) (Fig. 2, Table 3).
Additional analyses considering cardiac death and all-cause mortality are presented in Figures 2 and 3⇓and Table 4.For H/M <1.60, 2-year probabilities of cardiac death and all-cause mortality were 11.2% and 16.1% versus 1.8% and 3.0% for the group with H/M ≥1.60. Treated as a continuous variable, there was a progressive decline in both cardiac and all-cause mortality from >20% for H/M <1.10 to 0% for H/M ≥1.80. The 2-year cardiac death rate in the total population was 9.1%, compared with the rate in the first decile (H/M <1.20; n = 92) of 19.1%.
The multivariable analyses from the individual phase III studies demonstrated a consistently significant contribution for the late H/M across blinded readers. These same analyses showed no consistent contribution to event prediction for the 123I-mIBG and 99mTc-tetrofosmin SPECT results. The multivariable analysis of the pooled ADMIRE-HF data using only the planar 123I-mIBG imaging results produced a model with 4 independent variables contributing to the prediction of the primary outcome events: late H/M, LVEF, NYHA functional class, and plasma BNP (Table 5).In univariate Cox proportional hazards analyses, the early and late H/M and washout all were significantly associated with risk of CEs. In multivariable analyses with these 3 parameters, the late H/M was consistently the only parameter that retained statistical significance.
Comparison with other prognostic markers
The utility of addition of the H/M to 2 frequently used markers of prognosis in HF patients, BNP and LVEF, was examined in subanalyses. The median BNP level in the 926 subjects with adequate blood samples was 140 ng/l, and there was a small but statistically significant negative correlation between plasma BNP and H/M (Online Fig. 1). There were highly significant differences in the total CE and cardiac death rates between subjects with BNP levels below and above the population median (Figs. 4A and 4B). Addition of the binary H/M result provided further stratification for total CEs and cardiac death in the high-risk BNP population and a difference of borderline significance for total CEs in subjects with BNP <140 ng/l (Figs. 4C and 4D). The 2-year CE rate for subjects with BNP >140 ng/l was 42%, but among the subset with H/M ≥1.60, the rate was 20.5%. There were no cardiac deaths among the 57 subjects with BNP >140 ng/l and 123I-mIBG H/M ≥1.60, compared with 42 cardiac deaths among the 406 subjects (10.3%) with above-median BNP and H/M <1.60.
Median LVEF in the 961 subjects was 29%, and there was a modest but statistically significant positive correlation between LVEF and H/M (Online Fig. 2). There were highly significant differences in total event and cardiac death rates between subjects with LVEF levels below and above the population median (Figs. 4E and 4F). Addition of the binary H/M result provided further stratification for HF subjects in both LVEF groups (Figs. 4G and 4H). The 2-year CE rate for subjects with LVEF <30% and H/M ≥1.60 was less than half that of all subjects with LVEF <30% (17.6% vs. 40.3%). There were 2 cardiac deaths among the 81 subjects (2.5%) with LVEF <30% and H/M ≥1.60 as compared with 39 cardiac deaths among the 409 (9.5%) with LVEF <30% and H/M <1.60. There were no cardiac deaths among the 120 subjects with LVEF ≥30% and H/M ≥1.60.
Further analyses examined the interaction of H/M, LVEF, and BNP in the 926 subjects with BNP results. Among the 273 subjects with above-median values for BNP and below-median LVEF values, there were significantly lower 2-year primary event and cardiac death rates for subjects with H/M ≥1.60 compared with H/M <1.60 (Fig. 5A).Considering values of all 3 parameters in the most abnormal quartile (H/M ≤1.30, LVEF ≤23%, BNP >311.4 ng/l), the highest and lowest rates of primary events and cardiac deaths were among the 38 and 456 subjects with all 3 parameters or none in the most abnormal quartile, respectively (Figs. 5B and 5C).
Among the 10% of subjects with the lowest LVEF (<20%; n = 98), H/M values still discriminated between higher- and lower-risk subpopulations. Thirty-five of 61 subjects (57%) with H/M <1.40 had outcome events, compared with 10 of 37 (27%) with H/M ≥1.40 (p = 0.004). Considering this low LVEF decile in conjunction with subjects in the highest-risk deciles for H/M (<1.20) and BNP (>595.9 ng/l), 31 cardiac deaths (58% of the study total) occurred among the 233 subjects (24%) with at least 1 such high-risk value (p < 0.001 compared with 22 cardiac deaths among the remainder of the population [n = 727]). Each of the 3 parameters alone identified a subset of the subjects who experienced cardiac death (BNP, n =18; H/M, n = 13; LVEF, n = 13).
A total of 86 subjects (9%) experienced either nonfatal arrhythmic events (self-limited ventricular tachycardia [VT], n = 12; resuscitated cardiac arrest, n = 6; appropriate ICD discharges, n = 45) or SCD (n = 23). These combined “arrhythmic” events were significantly more common in subjects with H/M <1.60 (79 of 760 [10.4%] vs. 7 of 201 [3.5%]; p < 0.01). The highest prevalence of arrhythmic events was in the H/M range 1.30 to 1.39 (27 of 206; 13.1%). The highest H/M in a subject who experienced a fatal arrhythmic event was 1.60. Only 5 arrhythmic events occurred in the 191 subjects with H/M >1.60 (2.6%); 2 of 137 subjects with no ICD experienced self-limited episodes of VT, whereas 3 of 54 subjects with ICDs had device activations (2 antitachycardia pacing, 1 direct current shock).
123I-mIBG has been in clinical use in Japan and Europe for 2 decades, and a large number of published reports have documented abnormalities of myocardial sympathetic innervation in various cardiovascular diseases (13,14,16–21,24,25). A recent meta-analysis of 18 studies and 1,755 patients confirmed that HF patients with reduced late H/M or increased 123I-mIBG washout rate have a worse prognosis as compared with patients with normal 123I-mIBG uptake and washout parameters (24). In addition to prediction of HF progression and death, abnormal 123I-mIBG uptake and washout have also been shown to be associated with increased incidence of SCD and appropriate ICD discharges (27,28). These reports have supported the usefulness of 123I-mIBG imaging for predicting potentially fatal outcomes in HF patients over follow-up periods of months to years. ADMIRE-HF provides the first large, prospective confirmation of the strong prognostic value of quantitation of cardiac adrenergic neuronal activity in HF patients. The current results also suggest a potential beneficial application for the 123I-mIBG imaging procedure for identifying HF patients at very low and very high risks for near-term morbidity and mortality.
Role of neurohormonal regulation in HF
Observations that patients with HF have increased circulating catecholamines and downregulation of beta-adrenergic receptors (6–10) have focused particular attention on the potential for molecular imaging of in vivo myocardial NE kinetics to provide improved risk stratification for the HF patient population (24,25). Direct examination of cardiac neuronal status may be more informative than assessment of systemic adrenergic activity in that the 123I-mIBG H/M uptake ratio was a significant contributor to the ADMIRE-HF multivariable risk model, but plasma NE was not.
Preserved neuronal uptake of 123I-mIBG identified a very low-risk HF population, with late H/M ≥1.60 (21% of trial subjects) associated with <1%/year incidence of cardiac death. In contrast, among the 10% of subjects with H/M <1.20, annual rate of cardiac mortality (9.6%) was 10-fold greater. The results of the 123I-mIBG imaging procedure could therefore potentially identify approximately one-third of the HF population studied as either at substantial risk for near-term mortality or to be at low risk on the current level of therapy.
Clinical implications of ADMIRE-HF
ADMIRE-HF did not directly evaluate the potential benefit of 123I-mIBG imaging as an aid to clinical management of HF patients, but the study suggests that in appropriately selected patients, this imaging procedure could alert clinicians to the potential need for considering additional treatments. For example, there were 74 ADMIRE-HF subjects who were <50 years old and had BNP <100 ng/l. Forty-seven of these subjects had H/M <1.60, of whom 13 developed CE, including 3 ICD activations, 1 episode of sustained VT, and 1 SCD. Nine of these subjects had HF progression (including 1 who had previously experienced an ICD shock). Twenty-seven of these 47 subjects did not have ICDs, including the subject who died suddenly. Knowledge of the higher risk associated with abnormal cardiac innervation might have facilitated consideration of more aggressive treatment (such as earlier use of resynchronization therapy) in these subjects.
As a marker of adverse prognosis and cardiac death, BNP had the strongest predictive power (highest chi-square value) in the multivariable analyses. However, as demonstrated in those analyses and in the analyses restricted to H/M, BNP, and LVEF, the H/M provided significant information to complement BNP for identifying subjects at the highest risk for CEs and cardiac death. H/M also stratified risk among subjects with the lowest LVEF values (<20%), including identifying a small low-risk cohort with H/M ≥1.60 (12 of 98; 12%). The likelihood of cardiac death was more than 4 times greater in the subpopulation of subjects with extreme deciles for at least 1 of the 3 parameters of BNP, H/M, and LVEF compared with the remainder of the trial population (13.4% vs. 3.0%).
Failure of the 123I-mIBG and 99mTc-tetrofosmin SPECT results to contribute to the multivariable risk models may have been a direct consequence of the rigorous design of the phase III clinical trials. The readers scored all 123I-mIBG SPECT exams without reference to either the planar images (and the calculated H/Ms) or the 99mTc-tetrofosmin SPECT studies. The blinded mixture of ischemic and nonischemic cardiomyopathy subjects also made it more challenging for the readers to assign defect severity to the myocardial segments based on global versus regional reduction in 123I-mIBG uptake. Determination of the potential value of MPI SPECT, included in this study to assess whether quantitation of innervation/perfusion mismatch would augment prediction of arrhythmic CEs (10,19), was also affected by the inconsistency in the visual 123I-mIBG SPECT assessments. The planar H/M succeeded because it was a reproducible quantitative result; it is likely that similar success with SPECT will require use of automated quantitative analysis procedures.
As the present study involved only a single 123I-mIBG imaging procedure, short-term reproducibility of the H/M determinations was not assessed. Most published studies that included replicate 123I-mIBG exams were examining the effect of medications or devices and involved inter-scan intervals of 6 to 12 months (20,21). However, even modest intra-subject variability in the H/M measurements would be unlikely to change the highly significant results obtained in this study, particularly for individuals with either severely abnormal or normal cardiac uptake.
The ADMIRE-HF demonstrated the capacity of quantitation of sympathetic innervation of the myocardium, measured by the H/M on 123I-mIBG scintigraphy, for predicting prognosis for significant cardiac events in subjects with HF and significant left ventricular dysfunction. This is the population to which guidelines for use of implanted devices for management of HF and arrhythmic event risk usually apply (29). ADMIRE-HF showed a highly significant relationship between time to HF-related events and the H/M, which was independent of other commonly measured parameters such as LVEF and BNP, as well as demographic parameters such as age and renal function, in an HF population on guidelines-based contemporary therapy. The study also showed a clear association between severity of myocardial sympathetic neuronal dysfunction and risk for subsequent cardiac death.
For an expanded Methods section, supplementary figures, and Acknowledgments, please see the online version of this article.
The study was supported entirely by GE Healthcare. Dr. Jacobson is an employee of GE Healthcare and owns shares in the General Electric Company. Dr. Cerqueira is a consultant for GE Healthcare, Astellas Pharma USA, Siemens Molecular Imaging, and MDS Norton; is on the Speakers' Bureau of Astellas Pharma USA; and has received research grants from Perceptive InformaticsUSA. Dr. Thomas is a consultant for GE Healthcare and receives research grants from GE Healthcare and Siemens Medical Systems. Dr. Agostini receives research support and honoraria from GE Healthcare. Dr. Weiland receives research support and is a consultant for GE Healthcare; and is a principal investigator for GE Healthcare, Astellas Pharma, CardioKline, CardioDx, and Schering Plough. Stephen Nissen, MD, served as Guest Editor for this article.
- Abbreviations and Acronyms
- B-type natriuretic peptide
- cardiac event
- heart failure
- heart/mediastinum ratio
- implantable cardioverter-defibrillator
- iodine-123 meta-iodobenzylguanidine
- left ventricular ejection fraction
- myocardial perfusion imaging
- New York Heart Association
- sudden cardiac death
- single-photon emission computed tomography
- ventricular tachycardia
- Received August 17, 2009.
- Revision received January 19, 2010.
- Accepted January 20, 2010.
- American College of Cardiology Foundation
- Brunner-La Rocca H.P.,
- Esler M.D.,
- Jennings G.L.,
- Kay D.M.
- Leimbach W.N. Jr..,
- Wallin B.G.,
- Victor R.G.,
- Aylward P.E.,
- Sundlof G.,
- Mark A.L.
- Bohm M.,
- La Rosee K.,
- Schwinger R.H.,
- Erdmann E.
- Ungerer M.,
- Hartmann F.,
- Karoglan M.,
- et al.
- Mardon K.,
- Montagne O.,
- Elbaz N.,
- et al.
- Merlet P.,
- Benvenuti C.,
- Moyse D.,
- et al.
- Simoes M.V.,
- Barthel P.,
- Matsunari I.,
- et al.
- Wakabayashi T.,
- Nakata T.,
- Hashimoto A.,
- et al.
- Merlet P.,
- Valette H.,
- Dubois-Rande J.L.,
- et al.
- Cohen-Solal A.,
- Esanu Y.,
- Logeart D.,
- et al.
- Kasama S.,
- Toyama T.,
- Kumakura H.,
- et al.
- Agostini D.,
- Belin A.,
- Amar M.H.,
- et al.
- Cerqueira M.D.,
- Weissman N.J.,
- Dilsizian V.,
- et al.
- Verberne H.J.,
- Brewster L.M.,
- Somsen G.A.,
- van Eck-Smit B.L.
- Hosmer D.W.,
- Lemeshow S.
- Nagarahara D.,
- Nakata T.,
- Hashimoto A.,
- et al.
- Tamaki S.,
- Yamada T.,
- Okuyama Y.,
- et al.
- Hunt S.A.,
- Abraham W.T.,
- Chin M.H.,
- et al.