Author + information
- Received March 21, 2016
- Revision received May 20, 2016
- Accepted May 25, 2016
- Published online April 3, 2017.
- G. Cameron Coleman, MDa,
- Peter W. Shaw, MDa,
- Pelbreton C. Balfour Jr., MD, ScMa,
- Jorge A. Gonzalez, MDa,
- Christopher M. Kramer, MDa,b,
- Amit R. Patel, MDc,d and
- Michael Salerno, MD, PhD, MSa,e,∗ ()
- aDepartment of Medicine, University of Virginia, Charlottesville, Virginia
- bDepartment of Radiology and Medical Imaging, University of Virginia, Charlottesville, Virginia
- cDepartment of Medicine, University of Chicago Medicine, Chicago, Illinois
- dDepartment of Radiology, University of Chicago Medicine, Chicago, Illinois
- eDepartment of Biomedical Engineering, University of Virginia, Charlottesville, Virginia
- ↵∗Address for correspondence:
Dr. Michael Salerno, Department of Medicine, Cardiovascular Division, University of Virginia Health System, 1215 Lee Street, Box 800158, Charlottesville, Virginia 22908.
Objectives This study sought to perform a systematic review and meta-analysis to understand the prognostic value of myocardial scarring as evidenced by late gadolinium enhancement (LGE) on cardiac magnetic resonance (CMR) imaging in patients with known or suspected cardiac sarcoidosis.
Background Although CMR is increasingly used for the diagnosis of cardiac sarcoidosis, the prognostic value of CMR has been less well described in this population.
Methods PubMed, Cochrane CENTRAL, and metaRegister of Controlled Trials were searched for CMR studies with ≥1 year of prognostic data. Primary endpoints were all-cause mortality and a composite outcome of arrhythmogenic events (ventricular arrhythmia, implantable cardioverter-defibrillator shock, sudden cardiac death) plus all-cause mortality during follow-up. Summary effect estimates were generated with random-effects modeling.
Results Ten studies were included, involving a total of 760 patients with a mean follow-up of 3.0 ± 1.1 years. Patients had a mean age of 53 years, 41% were male, 95.3% had known extracardiac sarcoidosis, and 21.6% had known cardiac sarcoidosis. The average ejection fraction was 57.8 ± 9.1%. Patients with LGE had higher odds for all-cause mortality (odds ratio [OR]: 3.06; p < 0.03) and higher odds of the composite outcome (OR: 10.74; p < 0.00001) than those without LGE. Patients with LGE had an increased annualized event rate of the composite outcome (11.9% vs. 1.1%; p < 0.0001).
Conclusions In patients with known or suspected cardiac sarcoidosis, the presence of LGE on CMR imaging is associated with increased odds of both all-cause mortality and arrhythmogenic events.
Sarcoidosis is an inflammatory granulomatous disease of unknown origin, characterized histologically by noncaseating granulomas in multiple organs, including the lungs, skin, lymphatics, and central nervous system (1). Cardiac involvement is associated with ventricular arrhythmias, sudden cardiac death (SCD), and congestive heart failure. It is thought that two-thirds of sarcoid-related deaths are attributable to involvement of the myocardium (2–4). Thus, clinical diagnosis of cardiac sarcoidosis (CS) is crucial for timely therapeutic management and consideration of immunosuppressive therapies. Furthermore, a better understanding of cardiovascular risk factors in this population could have implications for device therapy for the prevention of SCD.
Cardiac magnetic resonance (CMR) has been shown to have excellent diagnostic accuracy for detection of CS and is becoming the gold standard for its diagnosis (5,6). CMR may detect myocardial edema and inflammation using T2-weighted imaging as well as detect myocardial scarring and fibrosis using late gadolinium enhancement (LGE) (7).
Multiple recent studies have been published regarding CMR assessment of prognosis in CS, in particular examining the presence of LGE and its association with adverse outcomes (8–10). However, the broad applicability of many of these studies is limited because they are small and single-centered. Prognostic validation of CMR is crucial, as the presence of LGE is thought to confer a higher risk of major adverse cardiac events such as new or worsening heart failure, life-threatening arrhythmias, and SCD resulting from myocardial scarring and fibrosis as has been demonstrated in other cardiac pathologies (11–13).
In the current environment of escalating medical costs, the prognostic utility of CMR may help justify its use and guide therapies in patients with sarcoidosis. Prognostic CMR data might provide valuable information for risk stratification and resource allocation, such as clarifying which patients benefit from implantable cardioverter-defibrillator (ICD) placement or when immunosuppressive medications, which have significant patient side effects, are indicated.
Given the multiple small and single-centered studies, we performed a systematic review and meta-analysis of studies reporting prognostic data from patients undergoing CMR for evaluation of known or suspected CS.
To identify eligible studies for inclusion in the current systematic review and meta-analysis, 3 independent reviewers (G.C.C., P.S., and P.B.) systematically searched (July 2015) PubMed, Cochrane CENTRAL, and metaRegister of Controlled Trials for studies assessing prognosis in patients undergoing CMR with known or suspected CS. Keywords used were “sarcoid late gadolinium enhancement,” “sarcoid delayed enhancement,” and “cardiac MRI and sarcoid.” Since the initial search, no further articles have been identified as of December 2015.
Studies were considered eligible for inclusion if CMR was used (alone or in addition to other imaging modalities) to assess for myocardial scarring from biopsy-proven or clinically suspected sarcoidosis; in cohorts of ≥5 patients; with ≥1 year of prognostic follow-up data, including event data for ventricular arrhythmia, SCD, aborted cardiac death and/or appropriate ICD discharge, hospital admission for congestive heart failure, cardiac mortality, and all-cause mortality. Studies with populations known to have coronary artery disease or cardiomyopathies of nonsarcoid etiology were excluded.
In addition, we consulted experts, reviewed citations from eligible studies, and contacted some investigators for additional unpublished data. The search was limited to studies published in peer-reviewed journals and therefore excluded trials presented in abstract form only. We restricted the review to studies that enrolled adults only with no language restriction. The current systematic review and meta-analysis was performed in accordance with guidelines of the MOOSE (Meta-Analysis of Observational Studies in Epidemiology) and PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) groups (14,15).
Three authors (G.C.C., P.S., and P.B.) independently and in duplicate scanned all abstracts and obtained full-text reports of articles that indicated or suggested eligibility. After obtaining full reports, the same reviewers independently assessed eligibility from the full-text articles, with divergences resolved after consensus.
The quality of included studies was assessed by 2 investigators (J.A.G., G.C.C.) using the Newcastle-Ottawa Quality Assessment Scale for Cohort Studies (16), in which the quality of the selected trials was determined on the basis of selection of the study groups (0 to 4 points), comparability of the study groups (0 to 2 points), and ascertainment of the outcome of interest (0 to 3 points).
Data abstraction and study appraisal were performed by the same aforementioned authors. Clinical outcomes of interest were cardiovascular death, all-cause mortality, and a composite of arrhythmogenic events defined as ventricular arrhythmia (ventricular tachycardia or ventricular fibrillation), SCD, or aborted SCD (appropriate ICD discharge) during follow-up. Clinical outcomes data was directly abstracted. Annualized event rates were calculated for studies by dividing the number of events by the follow-up duration.
Dichotomous variables are reported as proportions (percentages); continuous variables are reported as mean ± SD or median (range). Binary outcomes from individual studies were combined with a random-effects model, leading to computations of odds ratios (ORs) and 95% confidence intervals (CIs). I2 was calculated as a measure of statistical heterogeneity, with values of 25%, 50%, and 75% representing mild, moderate, and severe inconsistency, respectively. Small study or publication bias was explored with funnel plots and Egger test. Finally, meta-regression and sensitivity analyses (including exclusion of 1 study at a time) were conducted to explore heterogeneity.
Statistical analysis was performed using Review Manager (RevMan) version 5.3.5 (The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark) freeware package and R version 3.2.2 (R Foundation for Statistical Computing, Vienna, Austria), with statistical significance for hypothesis testing set at the α <0.05, 2-tailed level. Meta-regression analysis was performed using the package “metafor” in R. For studies with zero events in a group, the convention of adding 0.5 events to all cells was adopted (17).
Results of the literature search
The literature search identified 519 relevant abstracts of full-text articles: 58 unique articles were abstracted for review; 27 of these warranted full-text review; 17 articles were excluded for various reasons including cohort overlap with other articles, lack of specified outcomes, or incomplete CMR data (18–33). Ten articles remained for detailed study (6,8–10,34–39). Details of the search strategy are outlined in the QUORUM (Quality of Reporting of Meta-Analyses Standards) diagram in Figure 1.
Study characteristics are presented in Table 1. The 10 studies included a total of 760 patients with known or suspected CS undergoing CMR. Four studies were prospective and 3 studies were multicenter. The follow-up duration ranged from 1.5 years to 4.9 years with a weighted mean follow-up duration of 3.0 ± 1.1 years. Baseline patient characteristics are shown in Table 2. Patients had a weighted mean age of 53.0 ± 10.0 years and 41% were male. The weighted average ejection fraction was 57.8 ± 9.1%. Of the total, 95.3% of patients had known extracardiac sarcoidosis and 21.6% had known CS. The prevalence of LGE ranged from 13% to 89% with a weighted mean prevalence of LGE of 33%. The prevalence of LGE in each study had a strong negative correlation with the mean left ventricular ejection fraction (LVEF) (R = 0.95; p < 0.001) (Figure 2). Eight of the studies included patients undergoing CMR at 1.5-T; the remaining 2 studies do not report CMR imaging field strength. Data for immunosuppressive therapy including corticosteroid use was inconsistently reported.
Overall, the included studies were of high quality, with all 10 studies receiving maximal scores on the Newcastle-Ottawa Quality Assessment Scale in the areas of study group selection and ascertainment of the desired outcome (Table 1). Seven of 10 studies also received maximal scores in the third domain of comparability of study groups. Thus, the pooled data from these high-quality studies are collectively robust.
Late gadolinium enhancement and cardiovascular outcomes
Of the 10 studies reporting outcome data for ventricular arrhythmias, SCD, appropriate ICD discharge/aborted SCD, and all-cause mortality, patients with LGE had greater odds of having the combined outcome of arrhythmogenic events plus all-cause mortality compared with those without LGE (overall OR: 10.74; 95% CI: 4.12 to 27.90; p < 0.00001, I2 = 45%) (Figure 3). When comparing annualized event rates for the composite endpoint, patients with LGE had significantly higher rates of events than did patients without LGE (11.9% vs. 1.1%; p < 0.001) (Figure 4).
Moderate heterogeneity (I2 = 45%) was noted in the meta-analysis. To investigate this heterogeneity, we performed meta-regression to determine whether any clinical variables were associated with the composite cardiovascular outcome. There was adequate data to explore the effects of sex, age, LVEF, percentage of patients with known extracardiac sarcoidosis, and duration of follow-up using a mixed-model approach. LVEF was the only significant covariate, and inclusion of LVEF in the meta-regression model accounted for all remaining heterogeneity (I2 = 0%).
The OR for the association of LGE with adverse events was higher in studies with greater mean LVEF. However, the total prevalence of events was higher in studies with a mean LVEF <50% (24%) than those with mean LVEF ≥50% (11%). Two of the larger studies (10,36) had a pre-specified LVEF cutoff of ≥50%, and to explore this association further, we performed a stratified analysis based on this LVEF cutoff (Figure 3). Among studies with a mean LVEF ≥50%, the presence of LGE was associated with greater odds of the combined endpoint (OR: 19.43; 95% CI: 7.62 to 49.56; p < 0.00001), with only mild-to-moderate residual heterogeneity (I2 = 28%). In this population, the annualized event rate for the composite outcome was significantly greater for those with LGE than for those without LGE (11.59% vs. 0.69%; p = 0.0011). In contrast, among studies with mean LVEF <50%, where there was a very high prevalence of LGE positivity, patients with LGE were not at increased odds of having the composite endpoint.
From 7 studies reporting all-cause mortality, patients with LGE had significantly greater odds of death from any cause than did patients without LGE (OR: 3.06; 95% CI: 1.14 to 8.20; p = 0.03; I2 = 37%) (Figure 5). Of these studies, only 1 (Watanabe et al. ) had a mean LVEF <50%. A trend toward a higher annualized event rate for all-cause mortality was also noted in patients with LGE versus those without LGE (4.0% vs. 1.2%; p = 0.07) (Figure 4).
Only 3 studies provided specific data for cardiovascular mortality including 151 patients. No significant association between the presence of LGE and increased odds of cardiovascular death (OR: 3.24; 95% CI: 0.43 to 24.63; p = 0.26; I2 = 31%) was seen.
Two studies included in the analysis (10,39) individually showed near-neutral OR for the composite outcome. This discordance may be explained on the basis of individual study characteristics. Watanabe et al. (39) retrospectively studied 19 subjects with cardiac sarcoid; 17 of 19 patients (89%) demonstrated LGE and the mean EF was 37.5%. Only 2 events among all 19 subjects were noted, both in LGE+ patients. The neutral OR from this study is likely due to the small sample size and biased distribution. Nagai et al. (10) prospectively studied 61 patients with known extracardiac sarcoid, no evidence of cardiac involvement, and LVEF >50%. In this cohort, only 13% of patients had LGE and the overall event rate was low in both groups (there were 3 total events, all noncardiac in nature among patients without LGE), which likely led to the neutral OR for the composite endpoint. These 2 studies only contributed 16% weight to the overall meta-analysis.
Assessment of bias
Visual inspection of funnel plots and Egger test for funnel plot asymmetry did not demonstrate significant asymmetry. Sensitivity analysis, which was performed by excluding 1 study at a time from the outcomes analysis, demonstrated that the measured effect for the composite cardiovascular outcome was not sensitive to any individual studies. However, sensitivity analysis in the model for all-cause mortality did demonstrate sensitivity to 3 included studies (6,8,36), as exclusion of 1 study at a time no longer rendered the model significant. Notably, these are 3 of the larger studies and therefore had the largest effect sizes.
The findings of this systematic review and meta-analysis show that the presence of myocardial scarring as evidenced by the presence of LGE in CMR provides meaningful prognostic information in patients with known or suspected CS. The data demonstrate that patients with LGE have increased likelihood of death from any cause as well as increased odds of future arrhythmogenic events. The correlation of LGE and adverse outcomes seen in this meta-analysis supports the role of CMR imaging for detection of cardiac involvement in patients with sarcoidosis when cardiac involvement may not be evident clinically. Our findings also support previous work advocating CMR imaging in patients with suspected CS and normal LVEF (18).
Multiple previous studies have shown equivocal or insignificant associations between LGE and future risk of death or ventricular arrhythmias (25,40,41), which may be due to population differences or differences in CMR techniques as pointed out in the 2014 Heart Rhythm Society (HRS) Expert Consensus Statement (42). The studies that do show an association between myocardial scarring and worse prognosis are small (8,9,38). Despite this limited data, the HRS reached a consensus that CMR imaging for the purpose of sudden death risk stratification was reasonable in patients with CS, even in those with LVEF >35%. This meta-analysis helps to validate the HRS position statement by bolstering the growing body of evidence showing an association between LGE and adverse outcomes and justifying the role for CMR in patients with known or suspected CS, including those with near-normal LVEF. The current analysis shows that the presence of LGE in sarcoid patients with normal or near-normal LVEF is prognostically significant and greatly increases the likelihood of adverse events.
Implications for ICD and future directions
The 2014 HRS guidelines indicate that sarcoid patients with LGE on CMR and normal LVEF should have an electrophysiological (EP) study; if the EP study is positive, then an ICD may be indicated (Class IIa). The results of this meta-analysis may justify consideration of device therapy without further EP testing. Further prospective studies are needed to clarify the role of both CMR and EP testing with regard to ICD implantation in patients with CS. Although key concerns regarding inappropriate shocks and adverse events related to device therapy remain (43), this new data should be considered when deciding on ICD implantation given the adverse prognosis associated with myocardial scarring in patients with cardiac sarcoid.
As the optimal management of CS patients continues to evolve, there is a need for prospective studies enrolling patients with normal EF and reduced EF to further evaluate the interaction between LVEF and myocardial scarring on cardiovascular outcomes. Outcomes analysis adjusted for LVEF was only available for 2 of the included studies and was inadequate for pooled analysis. In the study by Nadel et al. (9) (N = 106), adjusted Cox analysis including LVEF and the presence of LGE demonstrated that the presence of LGE was the only independent variable that was predictive of the composite cardiovascular outcome (hazard ratio [HR]: 12.52; 95% CI: 1.35 to 116.18; p < 0.03). Multivariate Cox regression analysis by Greulich et al. (8) (N = 155) including the presence of LGE and the initial LVEF demonstrated that LGE presence was the best independent predictor of the composite endpoint (HR: 31.6; p = 0.0014). Patient-level data was available for the cohort in Murtagh et al. (36) (N = 205), and we performed a multivariate analysis including LVEF and LGE and found that LGE was an independent predictor of adverse outcomes (HR 29.79; 95% CI: 6.05 to 146.76; p < 0.0001). In the current analysis, the majority of adverse cardiovascular events (73%) were in LGE+ patients with a mean LVEF ≥50%, suggesting that LGE provides risk stratification for adverse events in patients with CS beyond LVEF assessment alone.
Future prospective studies using quantitative assessment of LGE may provide a more nuanced risk stratification model. Furthermore, as the inflammation and fibrosis may be more diffuse in sarcoid, there may be a role for parametric mapping techniques such as T1 or T2 mapping (44,45).
Certain limitations inherent to systematic reviews are pertinent to the current analysis, including nonuniform reporting of data from included studies and variable duration of follow-up. Additional limitations include heterogeneity of methods for quantifying EF, lack of LGE quantification or pattern data, and variable inclusion criteria, such as a pre-specified LVEF cutoff ≥50% in some studies. As only study-level covariates were available for analysis, the relationship between LVEF and LGE could not be assessed on a per patient basis; future prospective studies may help mitigate selection bias and provide patient-level insights. Despite these differences among studies, we demonstrate that meta-regression analysis showed no residual heterogeneity when LVEF was accounted for (I2 = 0%).
Finally, we were only able to include a composite of all-cause mortality and arrhythmogenic events due to insufficient breakdown of events in some of the studies. However, the direction of effect was similar to that of all-cause mortality. In the studies separately reporting arrhythmogenic events, the effect size for the OR was similar to the composite endpoint.
CMR imaging with LGE provides important prognostic risk stratification for patients with known or suspected CS. Patients with the presence of LGE are at increased risk of death from any cause and arrhythmogenic events, even if their cardiac function is normal or near normal. This study illustrates how the presence or absence of LGE likely has important implications for optimizing therapy in patients with known or suspected CS.
COMPETENCY IN MEDICAL KNOWLEDGE 1: CMR imaging is excellent in the diagnosis of CS and the presence of LGE provides prognostic risk stratification.
COMPETENCY IN MEDICAL KNOWLEDGE 2: The presence of LGE confers an increased risk of death by any cause and arrhythmogenic events in patients with known or suspected CS.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: With the ability to provide both diagnostic and prognostic information, CMR imaging should be strongly considered in the management of patients with suspected CS.
COMPETENCY IN INTERPERSONAL AND COMMUNICATION SKILLS: CMR imaging provides important prognostic information to providers that can help direct patient management.
TRANSLATIONAL OUTLOOK 1: A prospective registry with strict entry criteria for patients with CS would be helpful to better define the association between LGE and other adverse prognostic factors.
TRANSLATIONAL OUTLOOK 2: Additional research in the quantification of LGE in patients with CS may provide more nuanced risk stratification.
Drs. Shaw, Balfour, and Gonzalez has received grant support from the National Institutes of Health (T32 5T32EB003841). Dr. Patel has received research support from Philips Healthcare, Astellas Pharma, and the American Society of Echocardiography. Dr. Salerno has received grant support from the National Institutes of Health (K23 HL112910); and research support from Siemens Healthcare. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Coleman and Shaw contributed equally to this work. David Bluemke, MD, served as Guest Editor for this paper.
- Abbreviations and Acronyms
- confidence interval
- cardiac magnetic resonance
- cardiac sarcoidosis
- hazard ratio
- implantable cardioverter-defibrillator
- late gadolinium enhancement
- left ventricular ejection fraction
- odds ratio
- sudden cardiac death
- Received March 21, 2016.
- Revision received May 20, 2016.
- Accepted May 25, 2016.
- 2017 American College of Cardiology Foundation
- ↵(1999) Statement on sarcoidosis: joint statement of the American Thoracic Society (ATS), the European Respiratory Society (ERS) and the World Association of Sarcoidosis and Other Granulomatous Disorders (WASOG) adopted by the ATS board of directors and by the ERS executive committee, February 1999. Am J Respir Crit Care Med 160:736–755.
- Smedema J.P.,
- Snoep G.,
- van Kroonenburgh M.P.,
- et al.
- Patel M.R.,
- Cawley P.J.,
- Heitner J.F.,
- et al.
- Greulich S.,
- Deluigi C.C.,
- Gloekler S.,
- et al.
- Nadel J.,
- Lancefield T.,
- Voskoboinik A.,
- Taylor A.J.
- Kuruvilla S.,
- Adenaw N.,
- Katwal A.B.,
- Lipinski M.J.,
- Kramer C.M.,
- Salerno M.
- Hamirani Y.S.,
- Wong A.,
- Kramer C.M.,
- Salerno M.
- ↵Wells GA, Shea B, O'Connell D, et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. Available at: http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp. Accessed December 9, 2015.
- ↵Studies with zero-cell counts (section 16.9.2). In: Higgins J, Green S, editors. Cochrane Handbook for Systematic Reviews of Interventions. 5.1.0 [updated March 2011]. The Cochrane Collaboration, 2011. Available at: http://handbook.cochrane.org/. Accessed May 17, 2016.
- McArdle B.A.,
- Birnie D.H.,
- Klein R.,
- et al.
- Panda S.,
- Kaur D.,
- Lalukota K.,
- Sundar G.,
- Pavri B.B.,
- Narasimhan C.
- Kron J.,
- Sauer W.,
- Mueller G.,
- et al.
- Mehta D.,
- Mori N.,
- Goldbarg S.H.,
- Lubitz S.,
- Wisnivesky J.P.,
- Teirstein A.
- Orii M.,
- Hirata K.,
- Tanimoto T.,
- et al.
- Paz Y.E.,
- Bokhari S.
- Tezuka D.,
- Terashima M.,
- Kato Y.,
- et al.
- Yokoyama R.,
- Miyagawa M.,
- Okayama H.,
- et al.
- Ise T.,
- Hasegawa T.,
- Morita Y.,
- et al.
- Blankstein R.,
- Osborne M.,
- Naya M.,
- et al.
- Crawford T.,
- Mueller G.,
- Sarsam S.,
- et al.
- Murtagh G.,
- Laffin L.J.,
- Beshai J.F.,
- et al.
- Poyhonen P.,
- Holmstrom M.,
- Kivisto S.,
- Hanninen H.
- Kron J.,
- Sauer W.,
- Schuller J.,
- et al.
- Salerno M.,
- Kramer C.M.