Author + information
- Received November 7, 2014
- Revision received February 11, 2015
- Accepted February 12, 2015
- Published online May 1, 2015.
- ∗British Heart Foundation Glasgow Cardiovascular Research Centre, Institute of Cardiovascular and Medical Sciences, University of Glasgow, Glasgow, United Kingdom
- †Harrington Heart and Vascular Institute, University Hospitals Case Medical Center, Case Western Reserve University, Cleveland, Ohio
- ↵∗Reprint requests and correspondence:
Dr. Ify Mordi, Institute of Cardiovascular and Medical Sciences, British Heart Foundation Glasgow Cardiovascular Research Centre, University of Glasgow, Glasgow G12 8TA, United Kingdom.
Objectives This study aimed to assess the incremental prognostic value of global circumferential strain (GCS), as measured using cardiac magnetic resonance (CMR) tagging, in addition to baseline clinical characteristics, left ventricular ejection fraction (LVEF), and late gadolinium enhancement (LGE), in the prediction of major adverse cardiovascular events (MACE) in an unselected cohort of patients.
Background LVEF is a powerful predictor of mortality and is used for guiding treatment decisions. It is, however, subject to limitations. The value of GCS measured by CMR tagging in patients with suspected cardiac disease has not been fully explored despite its being considered as the gold standard noninvasive method of assessment of LV deformation.
Methods We prospectively evaluated data from 539 consecutive patients referred for CMR who underwent a CMR protocol that included cine imaging, tagging, and LGE. The primary endpoint was the prevalence of MACE, defined as a composite of all-cause mortality, heart failure–related hospitalization, and aborted sudden cardiac death.
Results MACE occurred in 62 of 539 patients (11.5%) over a mean follow-up period of 2.2 years. History of ischemic heart disease (IHD) and beta-blocker use were both significant clinical predictors of adverse outcomes. All 3 CMR parameters were significant multivariate predictors of the primary outcome when added to significant clinical predictors (LVEF, hazard ratio [HR]: 0.96 [95% confidence interval [CI]: 0.94 to 0.99; p = 0.005]; presence of LGE, HR: 2.07 [95% CI: 1.03 to 4.14; p = 0.04]; GCS, HR: 1.11 [95% CI: 1.02 to 1.21; p = 0.041]). Global chi-square increased significantly with the addition of both LGE and GCS. Both the presence of LGE and reduced GCS had independent prognostic value in the overall cohort. Patients with LVEF ≥35% but LGE present and reduced GCS had a poor outcome similar to that in those with LVEF <35%.
Conclusions We found, in a large-scale cohort of patients, that GCS, in addition to clinical variables, LVEF, and LGE, had incremental independent prognostic value. This measure could provide further risk stratification, especially in patients with mild LV impairment.
The assessment of myocardial function by left ventricular ejection fraction (LVEF) has an important role in the evaluation and management of patients in cardiology, providing significant prognostic information and guiding treatment decisions (1–3). Although myocardial function is commonly evaluated using echocardiography, cardiac magnetic resonance (CMR) is becoming increasingly utilized. CMR is currently recognized as the gold standard noninvasive strategy for the assessment of LVEF due to its advantages of providing unobstructed views of the heart in any plane and increased reproducibility (4).
A further advantage of CMR is tissue characterization by late gadolinium enhancement (LGE) imaging. The acquisition of images obtained 10 min after intravenous gadolinium contrast injection allows for the visualization of myocardial fibrosis. The presence of fibrosis is seen in numerous conditions and has also been shown to have significant prognostic value independent of LVEF (5–7).
Despite LVEF assessment by CMR being well validated, LVEF can be insensitive for the assessment of myocardial contractility. LVEF is dependent on other factors, including preload and afterload, and as such does not always identify subtle but important changes in LV systolic function (8). Early changes in myocardial contractility that are not identified by a decline in LVEF may still have an important clinical impact. The assessment of myocardial contractility, most commonly reported using strain as a measure of deformation, can be carried out using tagging sequences during CMR (9). This technique is regarded as the gold standard noninvasive strategy for deformation imaging. The assessment of global circumferential strain (GCS) using tagging has been shown to identify myocardial dysfunction in numerous conditions independent of LVEF. Recently, GCS has been shown to be an independent prognostic indicator in both asymptomatic patients and those with heart failure (10,11).
As yet, no study has assessed the prognostic value of GCS, in addition to LVEF, LGE, and in a combined CMR protocol. The aim of this study was to explore the prognostic value of these parameters in addition to baseline clinical risk factors in the prediction of major adverse cardiovascular events (MACE) in an unselected cohort of patients with suspected cardiac disease referred for clinical assessment.
Over the inclusion period, we prospectively evaluated data from 570 consecutive patients referred to the Golden Jubilee National Hospital, Clydebank, Glasgow, United Kingdom, for clinically indicated CMR and without contraindications to gadolinium contrast. Eligible patients were able to undergo the complete CMR protocol. Clinical management of the patients was left to the discretion of the referring physician. Baseline characteristics were obtained at the time of referral using each patient's electronic medical record. The study was approved by the local ethics committee.
All patients underwent scanning using a 1.5-T magnetic resonance imaging scanner (Avanto, Siemens, Erlangen, Germany) and underwent a standardized CMR protocol. Detailed imaging-acquisition methods have been previously described (7). Briefly, after initial localizers, cine imaging was performed using a steady-state free precession technique in 3 long-axis views (2-, 3-, and 4-chamber) and a series of short-axis slices (8 to 10, typically) covering the entire left ventricle, from base to apex (imaging parameters: repetition time/echo time/flip angle, 1.4 ms/3.5 ms/50°; spatial resolution, 1.7 × 2 mm; slice thickness, 8 mm).
Tagged CMR images were acquired using a spatial modulation of magnetization sequence in 3 short-axis slices selected to represent the basal, mid, and apical levels of the left ventricle (12). Grid tags were applied at the start of the electrocardiograph R-wave, and gradient echo cine images were acquired to follow myocardial motion using the tags. Care was taken to ensure adequate diastolic phase covering in patients with arrhythmia. Imaging parameters for tagging were as follows: repetition time, 4.1 ms; echo time, 3.9 ms; flip angle, 14°; slice thickness, 6 mm; field of view, 380 mm; grid distance, 5 mm. The number of cardiac phases was dependent on the patient’s heart rate. Mean temporal resolution was 32 ± 4 ms.
For LGE imaging, intravenous gadolinium contrast was injected (total dose of gadolinium–diethylenetriamine penta-acetic acid, 0.15 mmol/kg). Ten minutes later, images were acquired using a phase-sensitive inversion recovery technique (13).
LVEF was calculated from the short-axis cine images with post-processing using proprietary software (Argus; Siemens, Erlangen, Germany). Endocardial contours were drawn at both end-diastole (defined by the software) and end-systole (manually defined as the image with the smallest area of myocardium). This technique was repeated for each short-axis slice, and LVEF was calculated by the software using the summation-of-discs method.
Tagged CMR images were analyzed using the harmonic phase method (HARP version 5.03, Diagnosoft, Durham, North Carolina). Endocardial and epicardial borders were drawn for each slice, allowing for tracking of each tag throughout the cardiac cycle (14). The tags were then manually adjusted to ensure accurate tracking. Circumferential (Eulerian) strain was calculated for each slice. GCS was calculated as the peak circumferential strain, using the mean of the 3 slices. As GCS is a measure of circumferential shortening from baseline, it is typically a negative value. Greater amounts of circumferential shortening are indicated by a more negative value.
LGE images were also analyzed using Argus as previously described (7). LGE was considered to be present in any areas with a signal intensity 5 SD above that of normal myocardium. If LGE was present, the percentage of myocardial LGE was quantified by calculating the area of LGE each slice. The total percentage of myocardial LGE was calculated as (13): (total area of myocardial LGE/total area of myocardium) × 100.
The primary endpoint in this study was the prevalence of MACE, defined as a composite of all-cause mortality, heart failure–related hospitalization, and aborted sudden cardiac death (SCD). All patients were followed up using a computed record-linkage system, which allowed us to identify survival status and hospital admissions using access to patients' records (7). In cases in which electronic records were not up to date, the primary care practitioner was contacted to ensure adequate follow-up status. Events were adjudicated by an independent observer blinded to the results (N.T.).
All statistical analyses were carried out using SPSS version 22.0 (IBM, Armonk, New York). Continuous variables are reported as mean ± SD; categorical data, as n (%). Comparisons between continuous variables were carried out using a 2-tailed Student t test; categorical variables, the chi-square test. Correlations between CMR parameters were assessed using the Pearson correlation coefficient. Outcomes analysis was conducted using a Cox proportional hazards model, and time-to-event curves were drawn using the Kaplan-Meier method. All variables were evaluated using univariate Cox regression analysis to ascertain their prognostic power in predicting the primary outcome. Hazard ratios (HRs) (95% confidence intervals [CIs]) and chi-square values were obtained. All significant clinical univariate predictors (p < 0.10) were then entered into a multivariate Cox model to identify significant multivariate clinical predictors. To evaluate the incremental prognostic value of CMR, significant univariate CMR predictors (p < 0.05) were then added to the significant multivariate clinical predictors, with further multivariate models using each CMR parameter singly created. Optimal cutoff points for LVEF, GCS, and LGE were calculated using receiver-operating characteristic (ROC) curves, as well as division of the groups into tertiles as necessary. The chi-square value of each model was subsequently calculated. This process was then repeated in patients without severe LV systolic dysfunction (LVSD) (≥35%). Reproducibility was assessed using the Pearson correlation coefficient and the intraclass coefficient.
In total, 539 patients were included in the final analysis. Of the 31 patients excluded, 12 had arrhythmias that impaired scan quality (e.g., uncontrolled atrial fibrillation), 11 patients had inadequate image quality due to poor breath-holding ability, and 8 were unable to undergo LGE imaging due to renal impairment. The baseline clinical characteristics of the 539 patients are shown in Table 1. The mean age of the cohort was 48.1 ± 15.4 years; 63.6% were male. The majority of patients did not have any clinical risk factors. The reasons for referral for CMR are shown in Table 2. The majority of patients (56.8%) were referred for further clinical risk stratification of suspected heart failure (accurate LV function assessment and the presence of LGE).
Baseline CMR characteristics are shown in Table 3. Mean LVEF measured by CMR was 55.9 ± 14.1% and mean GCS was –13.4 ± 4.6%. A total of 164 patients had LGE present (30.4%). The mean volume of LGE was 4.1 ± 10.4%. There was a strong correlation between all 3 parameters (correlation between LVEF and LGE percentage, r = –0.43; between LVEF and GCS, r = –0.58; between GCS and LGE percentage, r = 0.33; all, p < 0.001). Mean acquisition time of tagged CMR was 104 ± 22 s (1.7 ± 0.37 min); mean time of post-processing analysis of tagged CMR was 3.6 ± 0.7 min.
The mean follow-up duration was 2.2 ± 1.2 years. MACE occurred in 62 of 539 patients (11.5%). The post-CMR diagnoses in patients with MACE are given in Table 4. There were 20 deaths (16 cardiac related, 4 noncardiac related), 30 admissions for heart failure, and 12 aborted SCDs. Age; history of IHD; the presence of diabetes; smoking habit; and treatment with angiotensin-converting enzyme inhibitors, beta-blockers, aspirin, and/or statins were all significant univariate clinical predictors of the primary outcome (Table 5). On multivariate analysis of significant clinical predictors, only a history of IHD (HR: 3.58; 95% CI: 1.77 to 7.22; p < 0.001) and beta-blocker use (HR: 2.29; 95% CI: 1.01 to 5.20; p = 0.048) remained significant predictors of the primary outcome (Table 5).
Incremental prognostic value of CMR parameters
All 3 CMR parameters were significant univariate predictors of the primary outcome (LVEF, HR: 0.92 [95% CI: 0.91 to 0.94; p < 0.001]; presence of LGE, HR: 5.47 [95% CI: 3.16 to 9.48; p < 0.001]; GCS, HR: 1.21 [95% CI: 1.16 to 1.27; p < 0.001]) (Figure 3). When individually added to significant clinical predictors on multivariate analysis, all 3 CMR parameters remained significant (Table 6). In the final multivariate model, all 3 CMR parameters remained significant predictors of the primary outcome (LVEF, HR: 0.96 [95% CI: 0.94 to 0.99; p = 0.005]; presence of LGE, HR: 2.07 [95% CI: 1.03 to 4.14; p = 0.040]; GCS, HR: 1.11 [95% CI: 1.02 to 1.21; p = 0.041]). The addition of both LGE and GCS had incremental prognostic value when added to clinical predictors and LVEF.
All 3 CMR parameters had reasonable accuracy in the prediction of AEs. The area under the ROC curve for LVEF was 0.834; for LGE, 0.699; and GCS, 0.820 (all, p < 0.001).
In patients with nonischemic cardiomyopathy, both the presence of LGE (HR: 2.65; 95% CI: 1.18 to 5.95; p = 0.018) and GCS (HR: 1.13; 95% CI: 1.01 to 1.27; p = 0.031) remained significant multivariate predictors of adverse outcome.
CMR predictors in patients with ischemic cardiomyopathy or prior myocardial infarction
There were 90 patients with a CMR-confirmed diagnosis of ischemic cardiomyopathy or evidence of a prior myocardial infarction, 26 of whom experienced MACE. Both LVEF (HR: 0.95; 95% CI: 0.92 to 0.98; p = 0.001) and GCS (HR: 1.18; 95% CI: 1.05 to 1.32; p = 0.007) were significant univariate CMR predictors of the primary outcome in this group.
Patients without severely impaired LV function
In total, there were 474 patients with LVEF ≥35%, of whom 35 experienced the primary outcome (7.4%) (11 deaths, 14 admissions for heart failure, and 10 aborted SCDs). Using ROC analysis, optimal cutoffs for LVEF and GCS were 50.2% and –12.1%, respectively. In this group, both the presence of LGE (HR: 3.88; 95% CI: 1.86 to 8.09; p < 0.001) and GCS (HR: 1.09; 95% CI: 1.00 to 1.19; p = 0.046) remained significant multivariate predictors of the primary outcome when added to LVEF (HR: 0.93; 95% CI: 0.90 to 0.97; p = 0.001) (Figure 4).
Patients with LVEF ≥35% but with LGE present and GCS ≥12.1% had a poor outcome similar to that in patients with LVEF <35%, whereas patients with LVEF ≥35%, no LGE, and GCS <–12.1% had a much better outcome (Figure 5).
Reproducibility of GCS measurements using tagging
The reproducibility of GCS measurements was assessed in 30 patients. There was excellent intraobserver variability (r = 0.96; p < 0.01) and interobserver agreement (r = 0.94; p < 0.01). The intraclass correlation coefficient for intraobserver agreement was 0.96 (95% CI: 0.93 to 0.98; p < 0.001) and interobserver agreement was 0.95 (95% CI: 0.88 to 0.98; p < 0.001).
For the first time in a large-scale, prospectively evaluated cohort of unselected patients, we have shown that the assessment of GCS using CMR tagging has independent and incremental value in the prediction of MACE when added to the clinical predictors LVEF and LGE as a part of a routine CMR protocol. Importantly, this incremental predictive value of GCS held true in the absence of significant LVSD, extending its applicability.
LVEF has been shown to be a strong predictor of adverse outcomes (1,15). The presence of severe LVSD, defined by an LVEF <35%, is a marker of poor prognosis and is commonly used for guiding treatment decisions regarding the use of therapies such as implantable cardioverter-defibrillators and cardiac resynchronization therapy (2,3). Patients with reduced LVEF (and/or previous myocardial infarction) are also commonly prescribed beta-blockers, which are known to improve LVSD, thus explaining the increased hazard ratio in this study in patients prescribed this class of drugs (16,17). Despite these factors, the use of LVEF alone is subject to some limitations. Firstly, it is known that a strict cutoff on the basis of LVEF may still miss patients at higher risk, as patients with LVEF >35% can still have admissions for heart failure and cardiac death (18–20). Secondly, LVEF is not a measure purely of contractility, as it is affected by volumes, loading conditions, heart rate, and valvular function, among other factors, meaning that patients with a preserved or mildly reduced LVEF may still have an adverse prognosis (21), whereas some healthy patients may actually have reduced LVEF (22). These limitations have led to the search for other parameters that may improve risk stratification.
LGE has emerged as a powerful predictor of adverse outcomes in patients with both reduced and preserved LVEF (23–26). Its absence may provide assurance to clinicians about patients who might already be at high risk for MACE (7). The presence of LGE signifies myocardial fibrosis, which, in addition to reducing overall myocardial contractility, might represent a substrate for ventricular arrhythmias that might lead to SCD (27–29). This study adds to the increasing evidence suggesting that the presence of LGE is an adverse prognostic indicator not only in conventionally high-risk patients (LVEF <35%) but also in those who might be thought to have a reduced risk (LVEF ≥35%).
The measurement of circumferential strain may aim to resolve some of the problems associated with LVEF. GCS is perhaps a more sensitive measure of myocardial contractile function (9). Strain measured using echocardiography has been shown to be an independent marker of adverse prognosis that adds incremental value to LVEF (11). However, although CMR tagging is accepted as the gold standard noninvasive strategy for the assessment of strain, only one study has evaluated the prognostic value of tagging. In the MESA (Multi-Ethnic Study of Atherosclerosis) cohort, Choi et al. (10) evaluated data from 1,768 asymptomatic patients who underwent CMR tagging. The authors found that GCS provided incremental prognostic value when added to baseline clinical variables and LVEF, including in patients with preserved LVEF.
Our study, on one hand, parallels the results of Choi et al. (10), and on the other extends its observations to patients with suspected cardiac disease, underscoring the additional clinical value of CMR tagging. Similar to Choi et al. (10), we also found that GCS was an independent predictor of MACE in a large-scale, unselected cohort of patients undergoing CMR examination. In addition, we found that an incremental benefit of GCS when the presence of LGE was included. This incremental value provides information to suggest that the addition of tagging to routine CMR protocols that already includes LVEF and LGE assessment may have some clinical benefit.
We hypothesize that the incremental benefit of GCS may be due to 2 reasons. First, strain is a more sensitive measure of contractile dysfunction than LVEF (8,30). Indeed, GCS is the major contributor to LV stroke volume (8). Second, LGE often does not identify diffuse myocardial fibrosis due to the need for identifying areas of “nulled” healthy myocardium (31). We postulate that the assessment of GCS may identify areas of reduced contractility caused by diffuse fibrosis not picked up by LGE (which is a marker of focal fibrosis), although further study is needed for confirming or refuting this hypothesis (32). Additionally, T1 mapping techniques are becoming increasingly utilized for the assessment of diffuse myocardial fibrosis not identified by LGE and have been shown to correlate well with GCS (33). These 2 reasons may provide some explanation for the incremental prognostic value of GCS in patients with moderate and preserved LV function. Interestingly, the presence of LGE was a stronger predictor of outcome in this group, which might have implications on the use of defibrillator therapies in this group, as it might suggest that the more common cause of adverse outcomes in this group is ventricular arrhythmia rather than pump failure. Our results suggest that perhaps this might have been the case.
Intriguingly, there appears to be a trend toward worse outcomes over time in the group of patients with LVEF >35% but either LGE present or reduced GCS. We speculate that the presence of LGE or reduced strain might be earlier indicators of adverse outcome than LVEF, and a larger-scale study with a longer follow-up period may identify the possible mechanism of this.
We found that the addition of CMR tagging did not add too much time to our standard CMR protocol. Ultimately, the assessment of GCS could be incorporated into routine CMR examinations and provides incremental prognostic value in addition to LVEF and LGE. The identification of patients with LVEF ≥35% but reduced GCS and LGE may identify a group that could benefit from more advanced therapies. Although current guidelines recommend the use of advanced heart failure therapies and devices in patients with LVEF <35%, it is increasingly accepted that most SCDs actually occur in patients with LVEF ≥35% (20). Nonetheless, it is important to recognize that in this study, LVEF was still by far the strongest predictor of outcome, and so both LGE and GCS should be used as additional, rather than replacement, diagnostic tools.
This study may have an element of referral bias. CMR is not yet a routine part of clinical care (for most conditions), and so we may have only been referred patients about whom the clinician wished to have had further information. Nonetheless, we feel that by having included consecutive patients with very few exclusions, we were able to obtain generalizable data.
The inclusion of all-comers in our study also added a potential limitation in terms of analysis of our results, as the effects of different treatments in different diseases and their effects on CMR parameters were not analyzed. A larger-scale study that allows for subgroup analyses, or several studies in cohorts of patients with one pathology (e.g., ischemic cardiomyopathy), would be required. We do believe, though, that the mixed cohort in this study does provide some advantages, however, and perhaps allows for a level of generalizability to the practicing clinician.
We assessed only GCS in this study; to keep scan times to a minimum, we did not obtain long-axis views for the assessment of global longitudinal strain in this cohort. However, global longitudinal strain has been shown to be a strong echocardiographic predictor of outcome (34). Additionally, we did not assess strain rate or diastolic parameters in this cohort, which might also have prognostic value.
The assessment of GCS by CMR tagging provides incremental prognostic value for the prediction of MACE when added to baseline clinical variables, LVEF, and LGE in a routine CMR protocol in patients with and without severe LV systolic dysfunction. The assessment of myocardial strain may be a useful additional parameter to include in CMR-scanning protocols and could be used for further risk stratification in patients with mildly impaired LV systolic function.
COMPETENCY IN MEDICAL KNOWLEDGE: The additional assessment of myocardial deformation by using tagged cardiac magnetic resonance can provide incremental prognostic value over and above left ventricular ejection fraction and the presence of late gadolinium enhancement in a variety of conditions.
TRANSLATIONAL OUTLOOK: Larger, multicenter trials are needed to identify the impact of tagging in specific conditions in order to investigate its clinical utility and potentially identify new therapeutic targets.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- cardiac magnetic resonance
- global circumferential strain
- ischemic heart disease
- late gadolinium enhancement
- left ventricular ejection fraction
- left ventricular systolic dysfunction
- major adverse cardiovascular events
- sudden cardiac death
- Received November 7, 2014.
- Revision received February 11, 2015.
- Accepted February 12, 2015.
- 2015 American College of Cardiology Foundation
- Solomon S.D.,
- Anavekar N.,
- Skali H.,
- et al.
- Yancy C.W.,
- Jessup M.,
- Bozkurt B.,
- et al.
- McMurray J.J.,
- Adamopoulos S.,
- Anker S.D.,
- et al.
- Kuruvilla S.,
- Adenaw N.,
- Katwal A.B.,
- Lipinski M.J.,
- Kramer C.M.,
- Salerno M.
- Mordi I.,
- Jhund P.S.,
- Gardner R.S.,
- et al.
- Buckberg G.,
- Hoffman J.I.,
- Mahajan A.,
- Saleh S.,
- Coghlan C.
- Cho G.Y.,
- Marwick T.H.,
- Kim H.S.,
- Kim M.K.,
- Hong K.S.,
- Oh D.J.
- Mordi I.,
- Stanton T.,
- Carrick D.,
- et al.
- Stecker E.C.,
- Vickers C.,
- Waltz J.,
- et al.
- Abergel E.,
- Chatellier G.,
- Hagege A.A.,
- et al.
- Kwong R.Y.,
- Chan A.K.,
- Brown K.A.,
- et al.
- Wu K.C.,
- Weiss R.G.,
- Thiemann D.R.,
- et al.
- Iles L.,
- Pfluger H.,
- Lefkovits L.,
- et al.
- Scott P.A.,
- Morgan J.M.,
- Carroll N.,
- et al.
- Dawson D.K.,
- Hawlisch K.,
- Prescott G.,
- et al.
- Rosen B.D.,
- Edvardsen T.,
- Lai S.,
- et al.
- Mewton N.,
- Liu C.Y.,
- Croisille P.,
- Bluemke D.,
- Lima J.A.
- Dass S.,
- Suttie J.J.,
- Piechnik S.K.,
- et al.
- Kuruvilla S.,
- Janardhanan R.,
- Antkowiak P.,
- et al.
- Stanton T.,
- Leano R.,
- Marwick T.H.