Author + information
- Received October 28, 2015
- Revision received December 1, 2015
- Accepted December 1, 2015
- Published online January 1, 2016.
- Valentina O. Puntmann, MD, PhD∗,†,‡,§∗ (, )
- Gerry Carr-White, MBBS, PhD∗,†,
- Andrew Jabbour, MBBS, PhD‖,
- Chung-Yao Yu, MBBS‖,
- Rolf Gebker, MD, PhD¶,
- Sebastian Kelle, MD, PhD¶,
- Rocio Hinojar, MD, Mres∗,#,
- Adelina Doltra, MD, PhD¶,
- Niharika Varma, MD∗,§,
- Nicholas Child, MBBS, PhD∗,§,
- Toby Rogers, MD†,§,
- Gonca Suna, MD†∗∗,
- Eduardo Arroyo Ucar, MD∗,
- Ben Goodman, MSc∗,
- Sitara Khan, MD, PhD†∗∗,
- Darius Dabir, MD∗,††,
- Eva Herrmann, PhD‡‡,
- Andreas M. Zeiher, MD, PhD‡,
- Eike Nagel, MD, PhD∗,†,‡,§,§§,
- International T1 Multicentre CMR Outcome Study
- ∗Guys and St Thomas’ NHS Trust, London, England
- †King’s College Hospital NHS Trust, London, England
- ‡Department of Cardiology, University Hospital Frankfurt, Frankfurt-am Main, Germany
- §Department of Cardiac Imaging, King’s College London, London, England
- ‖St Vincent’s University, Sydney, Australia
- ¶German Heart Institute Berlin, Berlin, Germany
- #Department of Cardiology, University Hospital Ramón y Cajal, Madrid, Spain
- ∗∗Cardiovascular Division, King’s College London, London, England
- ††Department of Radiology, University of Bonn, Bonn, Germany
- ‡‡Institute of Biostatistics and Mathematical Modelling at Goethe University Frankfurt; Frankfurt am Main, Germany
- §§Institute of Experimental and Translational Cardiac Imaging, DZHK Centre for Cardiovascular Imaging, Goethe University Frankfurt, Frankfurt am Main, Germany
- ↵∗Reprint requests and correspondence:
Dr. Valentina O. Puntmann, Department of Cardiology, Guys and St Thomas’ Hospital, Westminster Bridge Road, London, United Kingdom.
Objectives The study sought to examine prognostic relevance of T1 mapping parameters (based on a T1 mapping method) in nonischemic dilated cardiomyopathy (NIDCM) and compare them with conventional markers of adverse outcome.
Background NIDCM is a recognized cause of poor clinical outcome. NIDCM is characterized by intrinsic myocardial remodeling due to complex pathophysiological processes affecting myocardium diffusely. Lack of accurate and noninvasive characterization of diffuse myocardial disease limits recognition of early cardiomyopathy and effective clinical management in NIDCM. Cardiac magnetic resonance (CMR) supports detection of diffuse myocardial disease by T1 mapping.
Methods This is a prospective observational multicenter longitudinal study in 637 consecutive patients with dilated NIDCM (mean age 50 years [interquartile range: 37 to 76 years]; 395 males [62%]) undergoing CMR with T1 mapping and late gadolinium enhancement (LGE) at 1.5-T and 3.0-T. The primary endpoint was all-cause mortality. A composite of heart failure (HF) mortality and hospitalization was a secondary endpoint.
Results During a median follow-up period of 22 months (interquartile range: 19 to 25 months), we observed a total of 28 deaths (22 cardiac) and 68 composite HF events. T1 mapping indices (native T1 and extracellular volume fraction), as well as the presence and extent of LGE, were predictive of all-cause mortality and HF endpoint (p < 0.001 for all). In multivariable analyses, native T1 was the sole independent predictor of all-cause and HF composite endpoints (hazard ratio: 1.1; 95% confidence interval: 1.06 to 1.15; hazard ratio: 1.1; 95% confidence interval: 1.05 to 1.1; p < 0.001 for both), followed by the models including the extent of LGE and right ventricular ejection fraction, respectively.
Conclusions Noninvasive measures of diffuse myocardial disease by T1 mapping are significantly predictive of all-cause mortality and HF events in NIDCM. We provide a basis for a novel algorithm of risk stratification in NIDCM using a complementary assessment of diffuse and regional disease by T1 mapping and LGE, respectively.
Nonischemic dilated cardiomyopathy (NIDCM) is an increasingly recognized cause of cardiovascular morbidity and mortality (1–3). In NIDCM, a number of diverse influences promote intrinsic myocardial impairment and remodeling via complex pathophysiological processes including extracellular matrix remodeling, myofibroblast transformation, and cardiomyocyte cell loss, affecting the myocardium diffusely (4,5). The lack of accurate and noninvasive characterization of diffuse myocardial disease limits its early recognition and effective clinical management. Endomyocardial biopsy is the suggested gold standard for detection and classification of myocardial tissue abnormalities, yet its invasiveness, low diagnostic yield, and paucity of proven consequential management pathways limit its widespread use in guiding clinical management (6). Cardiac magnetic resonance (CMR) is able to visualize regional myocardial disease by late gadolinium enhancement (LGE) and has gained relevance in clinical management of cardiomyopathies by informing on the underlying etiology and supporting risk stratification in NIDCM (7,8). Because LGE relies on regional differences in tissue composition, it is an imperfect measure of diffuse interstitial disease underlying myocardial impairment in NIDCMs (9,10). Myocardial T1 mapping is emerging as the noninvasive method of choice in assessment of diffuse myocardial disease allowing quantification of altered magnetization properties, which relate to the pathophysiological changes in the myocardium. Studies have shown that T1 mapping measurements correlate with extracellular collagen volume fraction (9,11–14), are raised in a number of NIDCMs and relate to the severity of left ventricular (LV) remodeling in NIDCM (9–11,15). The relationship with outcome of these novel parameters in NIDCM and their comparative value against conventional markers of adverse outcome remain unknown.
This is a prospective longitudinal observational multicenter investigator-led study of the prognostic value of noninvasive T1 mapping measures in a cohort of adult patients with NIDCM. The multicenter consortium has been described previously (16). Standardization of T1 mapping acquisition was performed at all participating centers prior to the onset of patient recruitment. Participating centers support large CMR clinical service (>1,000 patients a year) and provide clinical care compliant with international guidelines and recommendations on patient management. The study protocol was reviewed and approved by the respective institutional ethics committees and written informed consent was obtained from all participants. All procedures were carried out in accordance with the Declaration of Helsinki (2000).
Consecutive subjects (n = 713) fulfilling the accepted diagnostic criteria for NIDCM (1–3) were enrolled between January 2011 and July 2014. Prior to enrolment, the diagnosis was confirmed by CMR on the basis of increased LV end-diastolic volume indexed to body surface area and reduced LV ejection fraction (EF) compared with published reference ranges normalized for age and sex (7). Patients were excluded (based on previous medical history, other investigations or CMR findings) if they had evidence of: 1) ischemic heart disease, defined as significant documented coronary artery disease, previous coronary revascularization, previous history of myocardial infarction, or evidence of ischemic type LGE, or inducible ischemia on stress testing (17); 2) myocardial infiltration due to amyloidosis, iron accumulation, lipid-storage disease, hypertrophic or arrhythmogenic right ventricular (RV) cardiomyopathy (1–3), or myocardial inflammation (18); or 3) significant primary valvular heart disease (1–3).
Additional exclusion criteria were the generally accepted contraindications to CMR (implantable devices, cerebral aneurysm clips, cochlear implants, severe claustrophobia), history of renal disease with a current estimated glomerular filtration rate <30 ml/min/1.73 m2, unable to receive gadolinium contrast agent, and inability to provide informed consent. Clinical metadata were collected for all subjects, as summarized in Tables 1 and 2. MAGGIC integer score was used to approximate the pretest likelihood of mortality due to heart failure (HF) in the present cohort (19).
All subjects underwent a standardized CMR protocol for routine assessment of cardiac volumes, mass, and LGE, at 1.5-T or 3.0-T Philips scanners (details of acquisition and post-processing available in the Online Appendix) (20). Modified Look-Locker imaging 3(2)3(2)5 was employed for T1 mapping and performed in a single midventricular short-axis (SAX) slice at mid-diastole, prior to and ∼15 min after administration of gadobutrol (Gadovist, Bayer, Leverkusen, Germany) (Figure 1) (10,13,15,16). The details of sequence parameters are provided in the Online Appendix. LV and RV volumes and function and the presence of LGE were interpreted locally, following standardized recommendations, to guide subsequent management decisions (17). Quantitative analysis of LGE extent and T1 mapping was performed centrally and not used in clinical management. LGE was quantified by a semiautomatic detection method using a previously validated method of full-width at half maximum and reported as a percentage of total LV mass (7,8,20). The distribution of LGE was characterized as midwall, epicardial, focal/involving the RV insertion points, or diffuse, based on the predominant pattern (7,8,17). Recovery rate of T1 relaxation was measured in a midventricular SAX slice conservatively within the septal myocardium (septal) as well as in the whole SAX myocardium, as previously described and validated (10,13,15,16,21). Extracellular volume fraction (ECV), a marker of interstitial contrast agent accumulation was calculated using T1 measurements of septal myocardium and blood pool pre- and post-contrast, and hematocrit value (22). Hematocrit was derived from routine bloods nearest to the CMR examination.
Follow-up was performed by review of electronic databases and telephone interviews after a minimum of 6 months. Medical records were examined for details on clinical presentation, entries of outpatient visits, hospitalizations, and medical procedures. A total of 53 patients (8%), lost to follow-up due to relocation (n = 37) or loss of contact (n = 16), and a further 23 with nondiagnostic images (significant breathing motion, mistriggering due to arrhythmia) were not included in the final analysis.
The predefined primary endpoint was death from any cause (all-cause mortality). Secondary endpoint was a HF composite endpoint (HF death or unplanned HF hospitalization) whereby the first single event per patient was included in the analysis (23). Primary endpoint events were adjudicated by a committee of independent physicians, blinded to the imaging results. Cause of death was established from a combination of death certification, available postmortem data, patients’ physicians, and review of medical records for patients who died while hospitalized. Mode of death was classified according to a modified Hinkle-Thaler system (7). Sudden cardiac death was defined as unexpected death either within 1 h of cardiac symptoms in the absence of progressive cardiac deterioration, during sleep, or within 24 h of last being seen alive. HF death was defined as death associated with unstable, progressive deterioration of pump function despite active therapy. Aborted sudden cardiac death was diagnosed in patients who received an appropriate implantable cardioverter-defibrillator shock for ventricular arrhythmia, or had a documented nonfatal episode of ventricular fibrillation or spontaneous sustained ventricular tachycardia (>30 s in duration) causing hemodynamic compromise and requiring cardioversion. HF hospitalization was categorized in patients admitted to the hospital with signs and symptoms of decompensated HF requiring treatment with an intravenous HF medication (diuretics, vasodilators, or inotropic agents).
Statistical analysis was performed using SPSS version 22 (IBM, Chicago, Illinois; details in the Online Appendix). All tests were 2-tailed and a p value of <0.05 was statistically significant. Baseline subject characteristics (Tables 1 and 2), grouped by the dichotomous primary endpoint, are presented as frequency (percentage) for categorical data and median (interquartile range [IQR]) for continuous data. Time to event was measured from the date of CMR study. Missing data for hematocrit were solved using recommended approaches (24). Univariate Cox proportional hazards models were used to test the association between the endpoints and baseline covariates (unadjusted hazard ratio and 95% confidence interval). Multivariable analysis was performed with a forward selection (likelihood ratio) modeling to determine independent associations with outcome (adjusted hazard ratio and 95% confidence interval), accounting for the rule of thumb for logistic and Cox models with a minimum of 10 outcome events per predictor variable (3 for all-cause mortality, 6 for HF endpoint), as well as interrelatedness of variables, using the best-of-the-group approach (further details in the Online Appendix).
We transformed the T1 mapping variables into categorical variables using: 1) the cutoff values as 2 standard deviations (SD) above the mean of the reference range (identification of abnormal myocardium), as well as ranking by 2n-times SD of the normal range (see the Online Appendix for details) (16); and 2) the classification into lower-to-middle versus upper tertile (patients with high risk of events).
Comparative performance of clinical decision-making (classification of subjects and events) against clinical standards based on LVEF <35% and LGE (25,26) were assessed using multivariate Cox regression.
Patient characteristics are summarized in Tables 1 and 2. A total of 637 subjects (mean age 50 years [IQR: 37 to 76 years]; 395 [62%] males) were included in the final analysis. The most common lead symptom included dyspnea (45%) and atypical chest discomfort (30%). Sixty-two patients (7%) presented with malignant ventricular arrhythmias. Seventy-four patients received an implantable cardioverter-defibrillator during the time of the follow-up. At the time of the CMR study, 71% of patients were New York Heart Association (NYHA) functional class II or less, and a total of 408 (64%) of subjects were taking regular cardiac medications. Hematocrit was available in 84.1% of subject and obtained on the same day in 49% (n = 312; overall mean time interval 8 ± 21 days, maximum 41 days). Subjects previously diagnosed with chronic kidney disease (n = 102) received a reduced dose of gadobutrol (0.01 mmol).
During a median follow-up of 22 months (IQR: 7 months) we observed a total of 28 deaths (cumulative event rate 4.4%). Cardiac mortality (n = 22, 3.5%) was the principal cause of the overall mortality. Patients who died were more likely to have evidence of adverse LV remodeling (Table 2), as well as raised T1 mapping indices, as well as presence and extent of LGE. HF endpoint consisted of 68 events (10.7%) due to HF death (n = 8), and unanticipated HF hospitalizations (n = 60). Whereas no patient underwent cardiac transplantation, 2 subjects were referred for LV assist device, but not implanted within the follow-up period. The median MAGGIC integer score was similar between the patients who survived and died, however, the score was higher in subjects who sustained a composite HF endpoint (median no event vs. event: 13 [IQR: 10 to 18] vs. 15 [IQR: 11 to 22]; p = 0.003).
In univariate Cox regression analyses, native T1 (septal and SAX), ECV, the presence and extent of LGE, and RVEF showed significant predictive associations with all-cause mortality and HF endpoint (p < 0.001) (Tables 3 and 4, Figures 2 and 3). LV end-diastolic volume, LVEF, and LV mass were less strongly associated with survival (p < 0.05). NYHA functional class >II and MAGGIC score were not associated with all-cause mortality; however, there was a relationship with the HF endpoint (p < 0.01). Other patient characteristics had no significant relationship with outcome. In multivariate stepwise analyses, native T1(septal) was the sole independent predictor of outcome, followed by the models with native T1(septal) and the extent of LGE for all-cause mortality, and native T1(septal) and RVEF for HF endpoint, respectively (Tables 3 and 4). Dichotomized variables, native T12SD (normal/abnormal) (16), and native T14SD and native T1tertiles (high risk) compared favorably to conventional markers of clinical decision making, using the presence of LGE (LGEpresence) and LVEF <35% for classifying subjects as high risk of poor outcome and HF event (Tables 3 and 4, Figures 2 and 3). Combination of LVEF <35% or LGE with native T1 did not improve predictive value, indicating the independent pathophysiological role of diffuse myocardial disease (4).
We demonstrate that noninvasive measures of diffuse myocardial disease by T1 mapping measurements based on this specific T1 mapping sequence, are significantly predictive of all-cause mortality and a composite HF endpoint of HF death and HF hospitalization in NIDCM. The predictive associations are independent of conventional markers of function, structure, and regional myocardial disease by LGE, supporting the prominent and independent pathophysiological role of diffuse myocardial disease in NIDCM. In multivariate analyses, native T1 measurement was the sole independent predictor of all-cause mortality and the composite HF endpoint. Our findings using this specific T1 mapping sequence provide a basis for a novel algorithm of clinical assessment and risk stratification of patients with NIDCM with a central role for native T1.
To the best of our knowledge this is the first report on the outcome associations with T1 mapping parameters in NIDCM in a large and multicenter cohort study, providing insight into the predictive relationships of markers of diffuse myocardial disease by T1 mapping and comparisons with standards of clinical decision making (25,26). Our results corroborate the predictive association of LGE with survival and HF, reported previously in NIDCM (7,8). The presence of any LGE was associated with all-cause mortality and HF endpoints; notably, midwall striae were the prevalent LGE pattern observed in our cohort. The stronger associations for the LGE extent compared to the presence alone reiterate the utility of this marker in improving risk stratification within the LGE group, as also shown previously (7,8).
Native T1 and ECV among T1 mapping indices were both strongly associated with the endpoints, substantiating their direct relationship to diffuse myocardial disease as the driver of poor outcome, independent of regional myocardial disease visualized by LGE. The differences in performance of T1 indices largely relate to the choice of sequence and the imaging parameters; the type of a modified Look-Locker imaging sequence used in this study is not optimized to accurately determine the true T1 signal. Because it is influenced by T2 decay, it is more sensitive to abnormalities in the water-rich myocardial milieu, explaining the greater effect detected by native T1 compared to post-contrast T1 (27,28). The difficulty in obtaining contemporaneous blood measurement for hematocrit, as well as different gadolinium doses to accommodate for patients with reduced estimated glomerular filtration rate, may be compromising the accuracy of our ECV results, however, the predictive associations for ECV are similar to those previously reported (on a basis of % of change) (29), supporting the viability of this complex marker in a real-life clinical scenario (further discussion in the Online Appendix). Results of ECV (based on this T1 mapping sequence) mainly rely on native T1 as the driver and to a lesser extent on the post-contrast T1 measurements resulting in a close interrelatedness of native T1 and ECV. Whereas the inclusion of larger volumes of myocardium into the T1 measurements seems more intuitive, septal measurements compared to full SAX sampling afford a higher precision as well as reduce the inclusion of degraded signal due to noise in the lateral segments diluting the relevant information (21,30).
Our findings may help to overcome an important gap in clinical management and discovery of therapies in NIDCM, and provide a basis for prospective studies of improved clinical pathways guided by T1 mapping. Native T1—based on this specific T1 sequence—provides a simple measurement in a single short and highly reproducible acquisition. Combined with rapid septal sampling this method offers a robust and simple standard for clinical routine, capturing the effects of conditions affecting myocardium diffusely, as providing a quantifiable marker whose magnitude directly relates to prognosis (Figures 2 and 3B). In contrast, LGE reflects (extracellular) myocardial disease, which is visualized once it is sufficiently regionalized, to afford a contrast relative to the nonenhanced reference (5). By way of an immediate clinical utility, native T1 is able to inform on the presence of abnormal myocardium (native T1 >2 SD above the mean of the normal range), and to detect those subjects where events are expected with a greater likelihood (native T1 in the upper tertile or >4 SD above the mean of the normal range), prior to significant functional impairment (EF <35%) and irrespective of the presence of LGE. Similarly, the low likelihood of events in those with normal test provides reassurance by exclusion of relevant myocardial disease. Native T1 (based on this T1 mapping method) may assume a central role in clinical management pathway of patients with suspected or known NIDCM.
The performance of T1 mapping indices relates to their ability to closely approximate the complex underlying pathophysiology in NIDCM. Due to the nonspecific drivers of a change in the T1 imaging signal, and the plethora of the pathophysiological mechanisms underlying the myocardial impairment in NIDCM, eventually culminating into diffuse interstitial fibrosis (1–5), the direct inference to the underlying histological substrate is not absolute. We and others have previously shown the relevance of T1 mapping in detection of subclinical disease (31–33) and sensitivity to myocardial inflammation (34). Thus, native T1 and LGE act as independent yet complementary imaging measures, informing on important distinction between diffuse and regional myocardial disease.
As for any diagnostic test, standardization of data acquisition and post-processing, as well as predefined reference ranges, are prerequisite for application of quantifiable imaging biomarkers in clinical routine (35). We achieved this by using a single-vendor platform, unifying the imaging parameters, using identical contrast type at all sites, performing quality control of the acquired data, and employing centralized post-processing. We benefited from the previously defined reference ranges (16), utilized the concept of SD for (field-strength independent) classification into normal/abnormal and ranking of disease expression, and defined a group at higher risk using the upper tertile. This allowed us to identify, firstly the presence of (prognostically) relevant disease, secondly, subjects at high risk of adverse outcome, and thirdly, to draw comparisons against standard means of disease detection and risk stratification (25,26). Baseline characteristics, including age, sex, NYHA functional class and EF, and consequently MAGGIC score, were less powerful in predicting survival; however, their relationship with the HF endpoint persisted in this study, providing an independent validation of the score in a NIDCM population. Yet, predictive associations of T1 mapping indices were notably stronger compared to LGE for the HF endpoint, lending support to the premise that unlike fixed, irreversible injury (seen by LGE), the activity of diffuse disease (detected by T1 mapping) portraits the compensatory capacity within the remaining viable myocardium.
A few limitations apply (details in the Online Appendix). Loss to follow-up is a major limitation of prognostic studies, including the present study: upon review the subjects lost were similar to the overall cohort in terms of heart risk score (MAGGIC) and T1 mapping indices and, therefore, it is unlikely we missed a significant number of events. Imaging was conducted on clinical CMR scanners with limited research funding, thus contemporaneous blood measurement for hematocrit could not be performed. We took care to record the value of hematocrit as near to the CMR scan as possible, as well as to avoid any major change of overall health status or treatment in the interval between blood sampling to CMR. We believe that our results are a very close reflection of a clinical reality and suggest a pathway for clinical viability of ECV (further discussion in the Online Appendix). The low overall rate of events may be explained by the guideline-based therapy, the lesser representation of advanced disease due to implantable devices, and the relatively short follow-up time. Post hoc power analyses of our results suggest that it was adequately powered (Online Appendix). For wider translation of our findings, an external cohort validation based using the same imaging method, as well as a multivendor cross-reference is required.
In summary, noninvasive measures of diffuse myocardial disease by T1 mapping are significantly predictive of all-cause mortality and HF events in NIDCM. Our findings provide a basis for a novel algorithm of clinical assessment and risk stratification of patients with NIDCM—based on this sequence, with a central role for native T1.
COMPETENCY IN MEDICAL KNOWLEDGE: In this observational longitudinal multicenter study in patients with NIDCM, noninvasive measures of diffuse myocardial disease by T1 mapping are able to identify patients at risk of all-cause mortality as well as heart failure. Using this particular T1 mapping methodology, native T1 is the strongest as well as independent predictor. Our findings provide a basis for a novel algorithm of clinical assessment and risk stratification of patients with NIDCM with a central role for native T1.
TRANSLATIONAL OUTLOOK: Successful clinical translation of T1 mapping by CMR may represent one of the most important advances in clinical management of NIDCM, allowing noninvasive detection of myocardial impairment and treatment discovery. For wider translation of our findings, an external cohort validation using the same imaging method, as well as a multivendor cross-reference is required.
The authors acknowledge the support of local Cardiology departments at Guy’s and St Thomas’ NHS Trust, King’s College Hospital NHST Trust, the German Heart Institute Berlin, and St Vincent’s University Hospital. Furthermore, cardiac radiographers at the respective sites in obtaining the high-quality imaging studies; Philips Clinical Scientists for support with setting up the sites: David M. Higgins, PhD; Bernhard Schnackenburg, PhD; Christian Stehning, PhD; Eltjo Haselhoff, PhD; Ian Ball, PhD; Contribution to clinical data collection: Mr. Banher Sandhu; Miss Tootsie Buser; Mr. Julian Bostock.
For supplemental materials, please see the online version of this article.
Funding was received from the Department of Health via the National Institute for Health Research (NIHR) comprehensive Biomedical Research Centre (BRC) award to Guy's & St Thomas' NHS Foundation Trust in partnership with King's College London and King's College Hospital NHS Foundation Trust. Drs. Puntmann, Hermann, Zeiher, and Nagel have received funding from the German Ministry of Education and Research via the German Centre for Cardiovascular Research (DZHK). Drs. Puntmann and Nagel hold a patent of invention for a method for differentiation of normal myocardium from diffuse disease using T1 mapping in nonischemic cardiomyopathies and others (based on PR-MS 33.297, PR-MS 33.837, PR-MS 33.654; with no financial interest). Drs. Jabbour and Yu were funded by the Victor Chang Cardiac Research Institute. Dr. Hinojar was funded by a Spanish Cardiology Society fellowship. Dr. Child has received funding from St. Jude Medical. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Javier Sanz, MD, served as Guest Editor for this paper.
- Abbreviations and Acronyms
- cardiac magnetic resonance
- extracellular volume fraction
- ejection fraction
- heart failure
- interquartile range
- late gadolinium enhancement
- left ventricular
- nonischemic dilated cardiomyopathy
- New York Heart Association
- right ventricular
- short axis
- standard deviation
- Received October 28, 2015.
- Revision received December 1, 2015.
- Accepted December 1, 2015.
- American College of Cardiology Foundation
- ↵World Health Organization. International Classification of Diseases (ICD). Available at: http://www.who.int/classifications/icd/en/. Accessed August 3, 2015.
- Elliott P.,
- Andersson B.,
- Arbustini E.,
- et al.
- Arbustini E.,
- Narula N.,
- Tavazzi L.,
- et al.
- Mewton N.,
- Liu C.Y.,
- Croisille P.,
- Bluemke D.,
- Lima J.A.
- Cooper L.T.,
- Baughman K.L.,
- Feldman A.M.,
- et al.
- Neilan T.G.,
- Coelho-Filho O.R.,
- Danik S.B.,
- et al.
- Iles L.,
- Pfluger H.,
- Phrommintikul A.,
- et al.
- Puntmann V.O.,
- Voigt T.,
- Chen Z.,
- et al.
- Bull S.,
- White S.K.,
- Piechnik S.K.,
- et al.
- Schulz-Menger J.,
- Bluemke D.A.,
- Bremerich J.,
- et al.
- Caforio A.L.,
- Pankuweit S.,
- Arbustini E.,
- et al.
- Pocock S.J.,
- Ariti C.A.,
- McMurray J.J.,
- et al.,
- Meta-Analysis Global Group in Chronic Heart Failure
- Jerosch-Herold M.,
- Sheridan D.C.,
- Kushner J.D.,
- et al.
- Hicks K.A.,
- Tcheng J.E.,
- Bozkurt B.,
- et al.
- Steyerberg E.W.
- ↵National Institute for Health and Care Excellence. Implantable Cardioverter Defibrillators and Cardiac Resynchronization Therapy for Arrhythmias and Heart Failure. NICE Technology Appraisal Guidelines. June 2014. Available at: https://www.nice.org.uk/guidance/ta314. Accessed August 3, 2015.
- McMurray J.J.,
- Adamopoulos S.,
- Anker S.D.,
- et al.,
- ESC Committee for Practice Guidelines
- Wong T.C.,
- Piehler K.,
- Meier C.G.,
- et al.
- Puntmann V.O.,
- D'Cruz D.,
- Smith Z.,
- et al.
- Hinojar R.,
- Varma N.,
- Arroyo Ucar E.,
- et al.
- Ntusi N.A.,
- Piechnik S.K.,
- Francis J.M.,
- et al.
- Hinojar R.,
- Foote L.,
- Arroyo Ucar E.,
- et al.
- ↵U.S. Food and Drug Administration. Biomarker Qualification Program. 2015. Available at: http://www.fda.gov/drugs/developmentapprovalprocess/drugdevelopmenttoolsqualificationprogram/ucm284076.htm. Accessed August 3, 2015.