Comparing CMR Mapping Methods and Myocardial Patterns Toward Heart Failure Outcomes in Nonischemic Dilated Cardiomyopathy
Author + information
- Received June 25, 2018
- Revision received August 13, 2018
- Accepted August 15, 2018
- Published online November 14, 2018.
Article Versions
- You are currently viewing a previous version of this article (November 14, 2018 - 11:01).
- View the most recent version of this article.
Author Information
- Tomas Vita, MD, MPHa,∗,
- Christoph Gräni, MD, PhDb,∗,
- Siddique A. Abbasi, MD, MSca,
- Tomas G. Neilan, MDc,
- Ethan Rowin, MDa,
- Kyoichi Kaneko, MD, PhDb,
- Otavio Coelho-Filho, MD, MPH, PhDb,
- Eri Watanabe, MD, PhDa,
- Francois-Pierre Mongeon, MD, SMd,
- Hoshang Farhad, MDb,
- Carlos Henrique Rassi, MDb,
- Yuna L. Choi, BAb,
- Kathleen Cheng, BAb,
- Michael M. Givertz, MDb,
- Ron Blankstein, MDa,b,
- Michael Steigner, MDa,
- Ayaz Aghayev, MDa,
- Michael Jerosch-Herold, PhDa and
- Raymond Y. Kwong, MD, MPHa,b,∗ (rykwong{at}bwh.harvard.edu)
- aNoninvasive Cardiovascular Imaging Program, Department of Radiology, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts
- bCardiovascular Division, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts
- cCardiac MR PET CT Program and Division of Cardiology, Department of Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts
- dDepartment of Medicine, Montreal Heart Institute, Université de Montréal, Montréal, Quebec, Canada
- ↵∗Address for correspondence:
Dr. Raymond Y. Kwong, Brigham and Women's Hospital, Cardiovascular Division, Department of Medicine, Harvard Medical School, 75 Francis Street, Boston, Massachusetts 02115.
Graphical abstract
Abstract
Objectives In patients with nonischemic dilated cardiomyopathy (NIDCM), native T1, partition coefficient (λGd), and extracellular volume fraction (ECV) mapping may offer prognostic values beyond late gadolinium enhancement (LGE), by scaling the range of myocardial changes.
Background In patients with NIDCM, LGE is seen in 30% of patients and it indicates adverse prognosis.
Methods The study mapped 6 anatomical locations using all 4 cardiac magnetic resonance (CMR) tissue-characterizing methods and associated with outcome. The authors performed T1 mapping of the myocardium and the blood pool, before and serially after contrast injection, using a Look-Locker cine gradient-echo technique to obtain T1 and the corresponding reciprocal R1 values. λGd values were derived from the slopes of the least-squares regression lines for myocardial versus blood R1, then adjusted to serum hematocrit to yield ECV.
Results Consecutive 240 NIDCM patients (49 ± 16 years of age; 38% women) underwent CMR for cardiac function, LGE, native T1, λGd, and ECV. After a median of 3.8 years, 36 (15%) experienced major adverse cardiac events (MACE), including 22 heart failure hospitalizations and 14 deaths. Nonischemic LGE was detected in 34%, whereas ECV was elevated (≥1 location) in 58%. Comparing the 4 methods, mean ECV and λGd both demonstrated strong association with MACE (both p < 0.001). In contrast to native T1 and LGE, ECV values from all 6 locations were associated with MACE and death, with the anteroseptum being the most significant (p < 0.0001). The number of abnormal ECV locations correlated linearly with annual MACE rates (p = 0.0003). Mean ECV was the only predictor to enter a prognostic model that contained age, sex, New York Heart Association functional class, and left ventricular ejection fraction. For every 10% increase, mean ECV portended to a 2.8-fold adjusted increase risk to MACE (p < 0.001).
Conclusions In this study of patients with NIDCM, mapping the myocardial extent of abnormality using ECV offers prognostication toward heart failure outcomes incremental to LGE or native T1 mapping.
Risk stratification of patients with nonischemic dilated cardiomyopathy (NIDCM) relies primarily on left ventricular ejection fraction (LVEF), and it inadequately reflects myocardial pathophysiology. Patients with NIDCM at risk of heart failure decompensation may benefit from novel noninvasive risk stratification (1). Diffuse fibrosis plays a key role altering mechanical, electrical, and vasomotor functions in this condition (2). Prior work had shown that midwall fibrosis detected on late gadolinium enhancement (LGE) by cardiac magnetic resonance (CMR) imaging is detected in approximately one-third of NIDCM patients and it had strong association with patient mortality (3). However, LGE imaging does not capture the full spectrum of diffuse fibrosis in NIDCM. Newer noninvasive imaging techniques, namely serial T1 mapping imaging characterizing extracellular volume fraction (ECV) has been validated against diffuse interstitial fibrosis by histology in NIDCM (4,5). Previous studies (6,7) have demonstrated a correlation between ECV and cardiac events in general patient cohort but the prognostic value of ECV and measurement of the partition coefficient (λGd) in a consecutive cohort of patients with NIDCM remains limited. In addition, LV myocardial anatomical location patterns of high risk of these respective methods had not been systemically compared. In the present study, we first tested the hypothesis that various CMR tissue characterization methods, namely LGE, native T1 mapping, λGd, and ECV, provide different strengths of prognostic association with major adverse cardiovascular events (MACE) in patients with NIDCM. In addition, we compared the strengths of prognostic association from 6 LV anatomical locations (anterior, anteroseptal, inferoseptal, inferior, inferolateral, and anterolateral) of abnormal tissue characteristics across these 4 methods, with MACE.
Methods
Study population
Subjects were enrolled at our center between January 2009 and December 2015. The study cohort included consecutive patients who fulfilled all of the following inclusion criteria: 1) CMR referral primarily for assessing for etiology of cardiomyopathy; 2) symptoms or signs consistent of heart failure as identified by the referring physician; 3) LVEF <60%; and 4) left ventricular end-diastolic volume index (LVEDVI) >80 ml/m2. Exclusion criteria included any of the following: 1) any medical documentation of coronary artery disease (by angiography or history) including imaging evidence of myocardial infarction or ischemia; 2) any history or imaging evidence of cardiac amyloidosis, cardiac sarcoidosis, hypertrophic cardiomyopathy, and Chagas disease; 3) moderate-to-severe valvulopathy; and 4) any absolute contraindications to performing CMR.
CMR imaging protocol and post-processing
A 3-T CMR system (Tim Trio, Siemens, Erlangen, Germany) was used for imaging of all subjects. The CMR protocol consisted of cine steady-state free precession imaging (repetition time [TR] 3.4 ms; echo time [TE] 1.2 ms; in-plane spatial resolution 1.6 mm × 2 mm) for LV function and mass, LGE, and serial T1 mapping. Cine imaging was obtained in 8 to 14 matching short-axis (8-mm thick with no gap) and 3 radial long-axis planes. LGE imaging (TR 4.8 ms; TE 1.3 ms; inversion time 200 to 300 ms) was performed to detect focal myocardial scars, using a segmented inversion recovery pulse sequence starting 10 to 15 min after a weight-based injection (cumulative dose 0.15 mmol/kg) of gadolinium diethylenetriaminepentaacetic acid (Magnevist, Bayer HealthCare Pharmaceuticals, Wayne, New Jersey). In patients with estimate glomerular filtration rate <60 ml/min/1.73 m2, contrast dose was restricted to 0.1 mmol/kg or 20 ml, whichever lower in volume based on our institutional policy (8). T1 measurements were acquired in a 3 LV short-axis cuts covering the 18-segment model using a validated cine Look-Locker sequence (5,9), with a non–slice-selective adiabatic inversion pulse, followed by segmented gradient-echo acquisition for 17 cardiac phases or times after inversion (TE 2.5 ms; TR 5.5 ms; flip angle 10°; 192 × 128 matrix; 6-mm slice), spread over 2 cardiac cycles (Inversion Time increments for T1 measurements of 100 ms pre-contrast, and 55 ms post-contrast, slice thickness 8 mm). T1 mapping imaging was acquired once before and 3 times after the injection of gadolinium spanning across a post-contrast period of approximately 30 min, to reduce the variance of the mean ECV values (9). A commercially available software (MASS v15, 2008, Medis, Leiden, the Netherlands) was used to post-process and quantify all CMR images. Epicardial and endocardial contours were placed manually on all LGE images, then LGE mass (in grams) was quantified by using signal intensity threshold cutoff of 2, 3, 4, and 6 SDs above the mean intensity of remote myocardium in the same slice.
Native T1, λGd, and ECV measurements
Figure 1 demonstrates the method of native T1, λGd, and ECV measurements. Endo- and epicardial borders of the LV of all T1 mapping images were traced with care to exclude any epicardial fat and endocardial blood. For each of LGE, native T1, λGd, and ECV, the 18 myocardial segments from 3 short-axis cuts were grouped and averaged into 6 anatomical locations (anterior, anteroseptal, inferoseptal, inferior, inferolateral, and anterolateral) for purpose of comparison across these methods. Using commercially available software (QMASS MR, Medis), signal intensity versus time curves were generated from regions of interest in the LV and blood pool. The signal intensity versus time curves for each myocardial segments and the blood pool were used for fitting to an analytical expression for the inversion recovery to obtain T1*, and correction for the effects of radiofrequency pulse during inversion recovery to calculate T1. The reproducibility of native T1 was tested and resulted in an intraclass coefficient for the intraobserver T1 measurement of 0.979 (n = 20; 95% confidence interval [CI]: 0.945 to 0.992). Septal versus lateral wall mean native T1 values of the whole cohort were significant different (1,102 ± 134 vs. 1,080 ± 136; p < 0.0001). The reciprocal of T1 (R1 = 1/T1) was used to plot the myocardial R1 against the R1 in the blood pool. Subsequently, the slope of least-squares regression line for R1 in tissue versus R1 in blood (limited data points with R1 in blood <3.5 s−1) was used to estimate λGd. This represents an extension of the formula: λGd = (1/T1Myo Post − 1/T1Myo Pre) / (1/T1Blood Post − 1/T1Blood Pre). λGd was then multiplied by blood plasma fraction (1 − the hematocrit expressed as value between 0 and 1), to obtain anatomical location myocardial ECV. The global mean ECV was calculated by averaging the ECV values across all segments. Hematocrit was derived from routine blood laboratory testing within 60 days to the CMR examination without any interval clinical event. The native T1 normal values were derived from a group of healthy volunteers who underwent research 3-T CMR studies at our center. The normal mean native T1 was 947.5 ± 62.1 ms. A cutoff of 1,072 ms (2 SD) was used for a normal T1 value. Further, the normal reference value of ECV was reported to be 28 ± 3% and we defined abnormal elevation of ECV by a 2-SD cutoff of >34% (9). The observers who performed the CMR analyses of LGE and T1 mapping were blinded to the clinical parameters, clinical outcome, and other CMR variables such as LV size and function.
Image Processing of Look-Locker T1 Imaging
Look-Locker T1 imaging using a low flip angle segmented gradient-echo pulse sequence (flip angle = 10°, repetition time/echo time = 5.5/2.8 ms). (A) Images for a total of 17 times after inversion (TI) were acquired, and every second one is displayed in this montage. Three short-axis cuts were imaged for 1 pre- and 3 post-contrast T1 measurements. (B) Bullseye view representing color-encoded T1 values for midring locations by Look-Locker imaging. The average signal intensity for each of the 18 segments was used, with inversion recovery phase restoration, for fitting to a parametrized representation of the inversion recovery, A + B⋅exp(-TI/T1*), where A, B, and T1* are adjustable parameters, and TI the time after application of the adiabatic inversion pulse. T1 was calculated from A, B, and T1* to correct for radiofrequency pulse effects during the Look-Locker readouts. (C) The data and best fit inversion recovery curve are shown in magnitude mode. (D) A total of 4 T1 measurements were used for estimation of extracellular volume fraction (ECV) by least-squares linear regression. ECV corresponds to the slope of the regression line, multiplied by (1 − hematocrit). (E) ECV values estimated are color coded and displayed in the 6 anatomical locations.
Follow-up of clinical endpoints
Clinical events of all subjects were assessed by review of the Social Security Death Index of the United States, then followed by a detailed review of all available electronic medical records. When electronic medical record were insufficient, subjects were evaluated by a standardized checklist-based patient questionnaire by mail or a scripted telephone interview based on the same standardized checklist. An informed consent was sent to subjects before initiation of any telephone contact. We defined MACE based on first heart failure related outcomes: 1) death from any cause; or 2) heart failure decompensation requiring hospitalization. Heart failure hospitalization was defined, in accord with prior trials (10,11), when all of the following criteria were met: 1) worsening of heart failure signs or symptoms with at least 1 of dyspnea on exertion, orthopnea, nocturnal dyspnea, pulmonary edema, increasing peripheral edema, increasing fatigue or decreasing exercise tolerance, raised jugular venous pressure, and radiological signs of heart failure; 2) inpatient hospitalization; and 3) any addition or increase of any heart failure intravenous treatments including diuretics, inotropes, or afterload-reducing vasodilator drugs (12). All study procedures were reviewed and approved by our Institutional Review Board in accordance with institutional guidelines.
Statistical analysis
Continuous variables were compared by t tests or Wilcoxcon rank sum tests as appropriate whereas categorical variables by the Fisher exact test. A 2-sided p value of <0.05 was used. Univariable and multivariable associations of risk covariates with clinical events were determined by Cox proportional hazards regression with time to event measured from the day of CMR. Survival curves were displayed using Kaplan-Meier estimates. Event rates were compared using chi-square or Cochran-Armitage trend tests as appropriate. We performed 3 multivariable models by inclusion of key risk variables including patient age, New York Heart Association functional class, and LVEF, and each of mean T1, LGE presence, and LGE mass. Mean ECV was then added to each model to determine the incremental prognostic value of mean ECV. Proportional hazards assumption of each covariate was assessed by visualization of a log-log plot then tested by including an interaction term of the covariate with time. SAS was used for all statistical analysis (version 9.4, SAS Institute, Cary, North Carolina).
Results
Patient demographics and CMR characteristics
The study included 289 consecutive patient subjects. Ten subjects were excluded due to imaging failure: breath-holding problem (n = 3), failure in cardiac gating (n = 4), and severe imaging artifacts (n = 3). Another 39 subjects were excluded due to CMR findings of a myocardial infarction (n = 25), cardiac amyloidosis (n = 8), or hypertrophic cardiomyopathy (n = 6). The remaining 240 subjects formed the study cohort. Baseline characteristics are summarized in Table 1. Mean age of the cohort was 49 ± 16 years and 62% of subjects were men. Ischemic causes were ruled out by: coronary angiography 154 (64%), 27 (11%) had a negative exercise/pharmacologic stress nuclear imaging, 18 (8%) by a negative stress CMR, and 41 (17%) by the clinical evaluation of the referring cardiologists before CMR study. No patient had thyroid or inflammatory causes of cardiomyopathy by available serological tests ordered by the caring physicians. Prevalence of coronary risk factors in the cohort was low, with hypertension, dyslipidemia, and diabetes occurring in 30%, 21%, and 13%, respectively. The median hematocrit was 40% (interquartile range: 4%) and the median time between hematocrit and CMR acquisition was 0 (interquartile range: 4) days. In 40 (17%) patients, hematocrit values were not available from clinical testing and they were imputed to 41%. The cohort resembles patients with moderate LV dysfunction and dilatation, evident by a mean LVEF of 43 ± 15% and LVEDVI of 114 ± 31 ml/m2, respectively.
Patient Demographics
Similar to published reports, nonischemic LGE was seen in 81 of 239 (34%) patients with septal LGE seen in 68 (28%) patients (3). The myocardial extent of LGE (4-SD method) was small, at median 2.9 (interquartile range: 5.8) in LGE positive patients. By comparison, ECV and native T1 were abnormal in at least 1 anatomical location in 58% and 77% of patients, respectively.
Clinical contact for assessing MACE was successful in 238 of the 240 subjects (0.8% lost to follow-up). Although, the remaining 2 lost to follow-up patients were reported alive based on interrogation of Social Security Death Index. During the mean follow-up of 3.8 ± 1.6 years, 36 (15%) subjects experienced MACE including first events of heart failure hospitalizations (n = 22, 9%) and deaths (n = 14, 6%). As shown in Table 1, a hypertensive history was noted in 42% of patients who experienced MACE and it demonstrated a trend association with MACE. Patients who experienced MACE had higher LGE extent and mean ECV than those free from MACE (4.9 g and 38% vs. 1.7 g and 31%; p = 0.03 and p < 0.001, respectively), but not mean native T1.
Kaplan-Meier curves for event-free survival, stratified by the ECV tertiles, are shown in Figure 2. ECV tertiles corresponded to a progressive reduction of cardiac event–free survival (p = 0.004). Table 2 demonstrate the associations of clinical and CMR variables with MACE. Patient age, New York Heart Association functional class, LV end-systolic volume, LVEF, and right ventricular ejection fraction demonstrated association with MACE. Both mean ECV and mean λGd were associated with MACE (all p < 0.001). For every 10% increase in mean ECV, hazard of MACE increased by 2.8-fold (p < 0.001). LGE presence was a significant covariate in the associations with MACE. Native T1 was not a significant covariate in the association with MACE. In the Online Table 1 the univariable association with MACE in patients with no LGE is represented. Figure 3 shows the annualized event rates by ECV tertiles stratified by a 30% LVEF cutoff. Annualized MACE rates escalated from 1.4% to 2.9% to 6.6%, across the first, second, and third ECV tertiles, respectively (p < 0.001). Annualized MACE rates among patients with LVEF ≥30% escalated modestly from 1.5% to 1.9% to 3.7%, respectively. On the other hand, in patients with LVEF <30%, annualized event rates escalated from 2.6% to 5.9% to 18.3% from the lowest to the highest ECV tertile, respectively (p < 0.001). Subjects with LVEF <30% suffered markedly high rates of MACE among those in the second and third ECV tertiles.
Kaplan-Meier Curve Stratified by ECV Tertile
Kaplan-Meier survival curves demonstrating progressive reduction of major adverse cardiovascular event (MACE)–free survival from the lowest to the highest extracellular volume fraction (ECV) tertile.
Annualized Event Rates of MACE Stratified by LVEF <30% and ECV Tertiles
There was progressive increase in the annual MACE rates across the escalating ECV tertiles, regardless of patients’ status of left ventricular ejection fraction (LVEF) <30%. Abbreviations as in Figure 2.
Univariable Association With MACE in the Whole Cohort
Multivariable models for MACE
Table 3 shows the multivariable models. When mean ECV was added to the models that contained each of native T1, LGE presence, and LGE mass (2 SD), mean ECV maintained strong association with MACE adjusted to each of the 3 variables. Patient age, New York Heart Association functional class, and LVEF were included in each of these 3 models so that the comparisons between mean ECV and each of 3 variables were properly adjusted. All 3 variables lost their significance in their respective models. Mean ECV did not maintain association with MACE when adjusted against λGd in the same model. In 156 patients with moderate LV dysfunction or dilatation defined by the recent published reference ranges (LVEF <45%, male with LVEDVI >112 ml/m2 or female with LVEDVI >99 ml/m2), mean ECV maintained strong association with MACE when added to a model containing patient age, New York Association functional class, and LVEF (adjusted hazard ratio [HR]: 1.08; p = 0.004). All covariates in these multivariable models fulfilled the proportional hazards assumption.
Multivariable Analysis of Association With MACE in the Whole Cohort
Comparing anatomical locations across the tissue characterizing methods
The association of anatomical locations with MACE by each of the 4 methods is shown in Figure 4. For LGE, both presence or absence and percentage extent by anatomical locations were analyzed. The anteroseptal segment represented the location of most closely associated with MACE across all 4 methods. Although ECV in the anteroseptal location had strongest association with MACE (chi-square 29.1; HR: 1.11; 95% CI: 1.07 to 1.16; p < 0.0001), ECV in all locations demonstrated strong association with MACE. Such prognostic association involving all anatomical locations was most robust by ECV when the 4 methods were compared.
Univariable Associations of the Anatomical Locations Across the Tissue Characterization Methods With MACE
Mean ECV values in all 6 anatomical locations had robust association with MACE, which is in contrast to native T1 mapping and late gadolinium enhancement (LGE) presence. Abnormalities in anteroseptal and inferoseptal locations were more closely linked with MACE. Abbreviations as in Figure 2.
Figure 5 illustrates the annual MACE event rates among native T1 mean, λGd, and ECV, stratified by the number of abnormal anatomical locations. Mapping the number of abnormal ECV locations corresponded to a linear increase in annual MACE rates (p = 0.0003 for trend), ranging from 2.4% in patients with only 1 location with abnormal ECV to 7.7% in patients with all 6 ECV locations abnormal. For each abnormal ECV location, the risk increased by 18% (HR: 1.18; 95% CI: 1.01 to 1.39; p = 0.035).
Annualized Event-Rates Based on the Extent of Abnormal Anatomical Locations in LGE, Native T1, and ECV
The total number of abnormal anatomical locations by (A) LGE and (B) partition coefficient (λGd) were not significantly associated with MACE. (C) The extent of abnormal ECV locations showed a linear trend toward the risk of MACE (p < 0.01 for trend). This suggests the diffuse myocardial pattern mapped by ECV characterized the risk to MACE more closely than LGE or native T1. For every additional abnormal ECV myocardial location, the risk increases by 18% (hazard ratio: 1.18; 95% confidence interval: 1.01 to 1.39; p = 0.0003 for trend). Abbreviations as in Figures 2 and 4.
Prognostic values of tissue charactization in patients with no LGE
The association of anatomical locations with MACE, by native T1 mean, λGd, and ECV, are shown in Figure 6. In the LGE negative patients, only ECV maintained its significant association, with MACE, in all anatomical locations. Such pattern of significant prognostication was not observed by native T1 mean and λGd.
Univariable Analysis Associations of the Anatomical Locations Across the Tissue Characterization Methods With MACE in Patients With No LGE
In LGE-negative patients, all locations by ECV maintained robust prognostic association with MACE. In comparison, locations by native T1, λGd, and LGE were not associated with MACE regardless of the anatomical location of abnormality. Abbreviations as in Figures 2, 4, and 5.
Discussion
In a consecutive cohort of patients with NIDCM, we observed that the extent of abnormality by ECV mapping portended to progressive increased risk to heart failure outcomes. ECV was significantly associated with MACE in any anatomical location (with highest association in the anteroseptal location). By increasing numbers of abnormal anatomical location, a corresponding increasing higher annualized event-rates for MACE was observed. In addition, ECV provides incremental prognostic association with MACE over native T1, LGE presence, and LGE mass, individually.
As heart failure hospitalization is costly, better risk stratification is imperatively needed and LGE presence might be a promising marker. Indeed, previous studies have reported that LGE was associated with adverse cardiac events and have incremental prognostic information beyond LVEF (3,13–15). However, as LGE, in line with our study, is only present in 25% to 40% of patients with NIDCM (3,16), potentially novel stratification tools to risk assess heart failure outcomes with the potential window of closer monitoring of high-risk patients and modify heart failure medication is needed. Mean ECV and λGd allows mapping of the entire heart and provided association with MACE, incremental to LVEF and LGE presence or size. LGE relies on regional contrast between normal and fibrotic myocardium, thus is an indirect measure of focal macroscopic myocardial scar. Our findings support the notion that LGE presence and size represent the “tip of the iceberg” of the underlying myocardial fibrosis and they are insensitive to characterizing the spectrum of diffuse myocardial fibrosis in NIDCM. Furthermore, we could show that anatomical location of involved abnormal anatomical locations helps in the risk stratification. Indeed, anteroseptal abnormal ECV, in line with anteroseptal LGE presence from previous reports, seems to pose the patient at high risk (17). Oloriz et al. (18) showed that the anteroseptal location of scarring, based on endocardial unipolar voltage, is most associated with ventricular tachycardial recurrence in patients with NIDCM. This is in line with our study, showing that fibrosis, based on ECV finding is mostly associated with MACE when located anteroseptal. It has been postulated that the combination of excess basal mechanical stress and underlying cardiomyopathic processes lead to increased arrhythmogenic substrate in this region (15). Prior evidence supports our finding that ECV provides incremental prognostic association than native T1 mapping, in part due to ECV’s better estimation of myocardial collagen volume fraction. In a recent small study of patients with symptomatic severe aortic stenosis scheduled for surgical valve replacement, native T1 showed only a moderate correlation with histologically defined fibrosis, whereas the λGd and ECV both showed strong correlations with fibrosis in histology (19). It is important to note that ECV and all measures derived from pre- and post-contrast T1 mapping are sensitive to fibrosis and edema (20). Compared with native T1 mapping, λGd and ECV is less sensitive to the effects of field strength and the interindividual variability because it is normalized by the change of the T1 relaxation rate in blood between pre- and post-contrast imaging (21).
Prior studies had reported a strong association of ECV with mortality and heart failure outcomes (22). However, this was assessed in a large cohort of patients consecutively referred to undergo a clinical CMR study with diverse study indications and different underlying cardiac conditions. Puntmann et al. (14) studied 637 patients with NIDCM and found that native T1 and LGE, but not ECV, provided strong association with outcomes in their multivariable models for clinical outcomes, and they commented on the logistic problems of obtaining hematocrit measurements. In addition to Puntmann et al. (14), we observed significant prognostic association of ECV (and λGd) with MACE in patients with NIDCM. Although there are discrepancies between the study, which are likely represented in differences in the CMR techniques, study cohort, and designs at this stage of clinical adaptation of ECV imaging. As normal values differ between scanner types and are therefore not interchangeable (23), the study by Puntmann et al. using 2 different field strengths (namely 1.5-T and 3.0-T scanners), make generalizability of the results difficult, whereas in our center, we used our own derived normal values of a single state of the art 3.0-T scanner for all patients. However, based on the results of these 2 studies, they are in some way complementary on account of the differences in patient characteristics, and specifically differences in the extent of diffuse interstitial fibrosis (native T1 and ECV) versus focal fibrosis (LGE), it seems that both native T1, ECV, and LGE are likely to play a significant role in NICM patient for the prediction of outcomes.
Study limitations
Our study has several limitations. First, we were not able to adjudicate the cause of death in some patients followed in an outpatient setting. In addition, the number of mortality events was small, thus our study cannot inform regarding any association of ECV and sudden cardiac death or decision toward implantable cardioverter-defibrillator therapy. Furthermore, due to the small numbers of events, our results should be seen as preliminary results. Second, CMR has been a part of the clinical workup of cardiomyopathy at our center. Consequently, a subset of our current cohort were considered to have a nonischemic cause of cardiomyopathy based on combined clinical evaluation and CMR evidence alone without the confirmation from an invasive coronary angiography. Prior investigations had indicated that approach could reliably ascertain nonischemic etiology of heart failure (24,25). Third, different T1 mapping methods extract different tissue information. Thus, generalization or comparison of results obtained by our cine Look-Locker gradient echo T1-mapping method, to other more commonly used T1 mapping methods, may not apply. Abnormal cutoff values of T1 and ECV remain investigational and cannot be universally applied across different pulse sequences. In addition, given that at the time of the study, our institution did not have the capability of performing hematocrit using a point-of-care bedside device and as a result 17% of subjects had missing hematocrit. Fourth, patients in our cohort represent a mild/early phenotype of heart failure. Thus, our results may not be applicable to those with severe phenotypes of heart failure. CMR referral itself introduced selection bias and our NIDCM patients were likely more hemodynamically stable and have less severe renal dysfunction or LVEF impairment given the exclusion of those with an implanted cardiac defibrillator. Our study thus cannot currently assess the impact from these referral biases onto the current results. Fifth, our cohort may have included patients with chronic myocarditis with fibrosis or scars which account for a higher T1 and ECV and may have influenced the outcomes. Finally, adjudication of heart failure hospitalization was subject to the clinical discretions of the admitting cardiologists. We minimized this bias by using standardized criteria of heart failure decompensation (10,11).
Conclusions
In patients with NIDCM, mapping the myocardial extent of abnormality using ECV characterizes the risk of heart failure outcomes, incremental to LGE or native T1 mapping in this study.
COMPETENCY IN MEDICAL KNOWLEDGE: Comparing 4 methods of tissue characterization, mapping the extent of ECV abnormality across all anatomical locations provides the high prognostic value to heart failure events in patients with NIDCM.
TRANSLATIONAL OUTLOOK: Extent of ECV abnormality as an estimate of diffuse myocardial fibrosis in patients with NIDCM provides significant incremental prognostic value over the current routine using LGE imaging.
Appendix
Footnotes
↵∗ Drs. Vita and Gräni contributed equally to this paper and are joint first authors. Dudley Pennell, MD, served as Guest Editor for this paper.
Dr. Vita was supported by National Institutes of Health training grant T32-HL094301-05. Dr. Kwong was supported by National Institutes of Health award 1UH2 TR000901; and has received research grant support from GlaxoSmithKline, Myokardia, and the Society for Cardiovascular Magnetic Resonance. Dr. Coelho-Filho is supported by a National Council for Scientific and Technological Development Productivity in Research award grant (303366/2015-0); and is supported by a Young Investigators Grant from the São Paulo Research Foundation (2015/15402-2). Dr. Mongeon has received research funding from Bracco and the Montreal Heart Institute has a research agreement with Siemens. Dr. Jerosch-Herold was supported by National Institutes of Health award R01 HL090634. Dr. Abbasi is a currently an employee of Amgen. Dr. Gräni has received funding support from the Novartis Foundation for Medical-Biological Research, Bangerter-Rhyner Foundation, Swiss Sports Medicine Society, and Kreislauf Kardiologie Foundation.
- Abbreviations and Acronyms
- CI
- confidence interval
- CMR
- cardiac magnetic resonance
- ECV
- extracellular volume fraction
- HR
- hazard ratio
- LGE
- late gadolinium enhancement
- LV
- left ventricular
- LVEDVI
- left ventricular end-diastolic volume index
- LVEF
- left ventricular ejection fraction
- MACE
- major adverse cardiovascular event
- NIDCM
- nonischemic dilated cardiomyopathy
- TE
- echo time
- TR
- repetition time
- λGd
- partition coefficient
- Received June 25, 2018.
- Revision received August 13, 2018.
- Accepted August 15, 2018.
- 2018 American College of Cardiology Foundation
References
- ↵
- Heidenreich P.A.,
- Albert N.M.,
- Allen L.A.,
- et al.
- ↵
- Wu T.J.,
- Ong J.J.,
- Hwang C.,
- et al.
- ↵
- ↵
- ↵
- Coelho-Filho O.R.,
- Mongeon F.P.,
- Mitchell R.,
- et al.
- ↵
- Wong T.C.,
- Schelbert E.B.
- ↵
- Wong T.C.,
- Piehler K.,
- Meier C.G.,
- et al.
- ↵
- ↵
- Neilan T.G.,
- Coelho-Filho O.R.,
- Shah R.V.,
- et al.
- ↵
- ↵
- ↵
- Ponikowski P.,
- Voors A.A.,
- Anker S.D.,
- et al.
- ↵
- Neilan T.G.,
- Coelho-Filho O.R.,
- Danik S.B.,
- et al.
- ↵
- Puntmann V.O.,
- Carr-White G.,
- Jabbour A.,
- et al.
- ↵
- Halliday B.P.,
- Gulati A.,
- Ali A.,
- et al.
- ↵
- Iles L.,
- Pfluger H.,
- Phrommintikul A.,
- et al.
- ↵
- Castelvecchio S.,
- Careri G.,
- Ambrogi F.,
- et al.
- ↵
- Oloriz T.,
- Silberbauer J.,
- Maccabelli G.,
- et al.
- ↵
- Vassiliou V.S.,
- Wassilew K.,
- Cameron D.,
- et al.
- ↵
- Moon J.C.,
- Messroghli D.R.,
- Kellman P.,
- et al.
- ↵
- ↵
- Schelbert E.B.,
- Piehler K.M.,
- Zareba K.M.,
- et al.
- ↵
- Messroghli D.R.,
- Moon J.C.,
- Ferreira V.M.,
- et al.
- ↵
- Assomull R.G.,
- Shakespeare C.,
- Kalra P.R.,
- et al.
- ↵
- McCrohon J.A.,
- Moon J.C.,
- Prasad S.K.,
- et al.
Toolbox
Citation Manager Formats
Article Outline
Figures in Article
Supplementary Materials
- Online Table 1[S1936878X18307496_mmc1.docx]