Contrast-enhanced cardiac Magnetic Resonance: distinction between cardiac sarcoidosis and infarction scar

Abstract

Objectives: To review the value of delayed contrast-enhanced cardiac magnetic resonance (CECMR) in differentiating patients with cardiac sarcoidosis (CS) from those with coronary artery disease and recent myocardial infarctions. Background: Late gadolinium enhancement (LGE) accurately delineates myocardial necrosis or fibrosis. The pattern of LGE in ischemic and non-ischemic myocardial disease is different, and might be helpful in distinguishing CS from ischemic disease. Methods: The CECMR studies of 30 patients with CS were compared to those performed in 30 consecutive infarct patients, who had been managed with primary coronary interventions, and 10 healthy controls. Two experienced blinded observers classified patients by assessing the distribution of LGE. Results: LV LGE was present in 29/30 CS (mean 3.8 segments, range 0-12), all infarct (mean 4.3 segments, range 0-9), and none of the patients in the control group. The amount of LV LGE did not differ significantly between CS and infarct patients (19 ± 11% and 19 ± 12%, P= 0.8). The CS group exhibited a predominantly patchy, 3 layer LGE (P = 0.01), whereas confluent transmural LGE (P = 0.04) with a vascular distribution (P < 0.001) was prevalent in the infarct group. Significantly more RV LGE (P = 0.01) and dilation (P = 0.02) were found in the CS group. The two observers classified patients correctly as CS in 72% and 83% of cases, as ischemic in nature in 77% and 80% of cases, and as normal in 90% and 100% respectively. Conclusions: Gadolinium CMR was helpful in differentiating patients with CS from patients with ischemic heart disease and previous myocardial infarctions. In a subgroup of ischemic patients the pattern of LGE was atypical, and suggestive of non-ischemic etiology. (Sarcoidosis Vasc Diffuse Lung Dis 2017; 34: 307-314)

Introduction

Sarcoidosis is a multi-system granulomatous disorder of unknown etiology with cardiac involvement in approximately twenty to thirty percent of patients (1). The clinical features of sarcoid heart disease include congestive heart failure, cor pulmonale, supraventricular and ventricular arrhythmias, atrioventricular and intraventricular conduction disease, ventricular aneurysms, pericardial effusion and sudden death (1). The diagnosis of cardiac sarcoidosis (CS) is made in the co-existence of non-caseating granulomas on myocardial biopsy or biopsies of any extra-cardiac tissue (with the exclusion of other causes for granulomatous inflammation such as mycobacterial or fungal infection) and cardiovascular abnormalities for which other possible causes have been excluded (2). The value of gadolinium-enhanced cardiac magnetic resonance (CMR) in the diagnosis and management of this condition has been demonstrated (2-4). Late gadolinium enhancement (LGE) by CMR is the most accurate non-invasive method to evaluate myocardial necrosis or fibrosis caused by acute myocardial infarction, chronic myocardial infarction or non-ischemic myocardial disease (5). The distribution of LGE was valuable in differentiating between ischemic and non-ischemic myocardial scarring (6-9). Since a significant number of patients with CS present with symptoms of heart failure or chest pain, similar in nature to those in patients with coronary artery disease, we aimed to determine whether CMR, and specifically the pattern of MDE, would allow us to distinguish the CS patients from patients with coronary artery disease and recent myocardial infarcts.

Patients and methods

Patient population

Between July 1998 and November 2004 thirty patients were diagnosed with CS according to modified guidelines, based on the Study Report on Diffuse Pulmonary Diseases of the Japanese Ministry of Health and Welfare (1993) (2). We compared the CMR findings in the CS patients with those of 30 consecutive patients who had presented with myocardial infarcts, and who had CECMR studies during the study period. We included CECMR studies of 10 healthy control subjects. All patients in the infarct group had recently been diagnosed with myocardial infarction as defined by the European Society of Cardiology and American College of Cardiology, and underwent coronary angiography with (primary) percutaneous coronary interventions. (10) Twenty patients had myocardial infarction in the distribution of a single coronary artery (right coronary artery: 6 patients, left anterior descending artery: 7 patients, left circumflex artery: 7 patients), while ten patients had infarcts in the distribution of two coronary arteries (right coronary artery and left anterior descending artery: 4 patients, left anterior descending artery and left circumflex artery: 4 patients, right coronary artery and left circumflex artery: 2 patients). Since the study concerned retrospective analysis of clinical data in the CS and infarct groups, the Institutional Review Board waived the need for consent according to Dutch legislation. The Board approved the CMR studies in the control subjects, and all controls provided written informed consent.

CMR protocol

Studies were performed using a 1.5 Tesla MRI scanner [(Philips, Best, The Netherlands (53 patients), Siemens, Erlangen, Germany (3 patients, 10 controls) and General Electric, Milwaukee, Wisconsin, USA (4 patients)] with a cardiac-dedicated phased-array coil. The CMR studies were ECG triggered by standard software. All patients underwent steady-state-free precession studies of short axis, vertical long axis and four chamber views, to assess regional wall motion abnormalities. Before and ten minutes after the administration of 0.1 mmol/kg gadolinium-DTPA (Schering, Berlin), short axis, vertical long axis and 4 chamber images were obtained with Spin Echo in 5 patients (slice thickness 8 mm, gap .8 mm, matrix 512 × 512, FOV 380 mm, voxel size 0.7 mm × 0.7 mm × 8 mm ), and 3-D breath hold inversion recovery-gradient echo (IR-GRE) sequences, obtained in diastole to minimize artifact due to cardiac motion, in the remaining 55 patients and 10 controls (slice thickness 10 mm, no gap, matrix 256 × 256, FOV 400 mm, voxel size 1.6 mm × 1.6 mm × 10 mm) to assess for the presence of LGE. The inversion time (250-400 msec) was determined on an individual basis to obtain optimal nulling of the unenhanced myocardial signal. The total time required for the investigation was 30-45 minutes.

CMR Analysis

CMR studies were analyzed off-line using commercially available software (CAAS MRV 3, Pie Medical Imaging, Maastricht, The Netherlands). Ventricular parameters were assessed in a standard way (11). Regional wall motion abnormalities, loss of wall thickness and left ventricular (LV) LGE were localized according to the standard 17 segment model (12). After delineating the endocardial and epicardial LV contours manually, the hyper-enhanced myocardium was depicted by changing the threshold setting for signal intensity. Signal intensities of LV and right ventricular (RV) LGE, and remote unenhanced LV and RV myocardium were measured in the short axis slice with the highest level of enhancement. The cut off signal intensity value for LGE was two times the standard deviation of remote unenhanced myocardium. The absolute and relative amounts of LV LGE were computed by the post-processing software. RV LGE was considered to be present when seen in both the short axis and four chamber views. The distribution of LV LGE was classified as sub-endocardial, mid-wall, sub-epicardial, patchy three-layer involvement, or confluent transmural involvement. Two blinded, independent observers (RT, BMM) were asked to differentiate between the three groups based on the presence and distribution of LGE. To test for intraobserver variability, one observer (RT) repeated the assessment after a month, while blinded to previous results.

Statistical analysis

All statistical analyses were performed with a commercially available statistical software program (SSPS for Windows, version 21; SPSS, Chicago, Ill). Group data were expressed as mean ± SD. Continuous variables were assessed using the parametric t test for independent samples or Mann Whitney test where appropriate, and all categorical variables were assessed using the chi-square test. Statistical significance was defined as a p value less than 0.05. Bland-Altman analyses and intra-class correlation coefficients were used to determine intra-observer variability in the assessment of ventricular masses, volumes, ejection fractions, and the amount of LGE. We used kappa values to assess interobserver variability in determining the presence and localization of LGE, the diagnostic accuracy of the observers and the intra-observer variability in diagnosing CS and infarcts by LGE CMR.

Results

Patient characteristics

The demographic and background medical data are summarized in Table 1. The patients suffering from coronary artery disease were significantly older, had more cardiovascular risk factors, and pathological Q-waves on their 12-lead ECG’s.

Table 1.

Summary of demographic and background medical data

Cardiac Magnetic Resonance

The CMR findings are summarized in Table 2. The LV ejection fraction and LV end-diastolic volume index differed significantly between the CS and infarct groups, with more loss of regional LV wall thickness (P = 0.009) and wall motion abnormalities (P = 0.008) resulting in poorer systolic function (P = 0.047) and larger ventricles (P = 0.01) in the latter group. LGE of the LV was present in 29/30 CS patients (mean 3.8 segments/patient, range 0-12), 30/30 infarct patients (mean 4.3 segments/patient, range 0-9), and none of the control group. The amount of LGE did not significantly differ between the CS and infarct groups (P = 0.8). Most enhancing lesions in the CS group were located in the basal and anterolateral LV (segments 1-6, 12 and 16; 74 out of 117 enhancing segments (64%), compared to 58/127 (46%) for the MI group). In the majority of CS patients (19/30, 63%) patchy LGE involved all three myocardial layers. In 2/30 (7%) confluent transmural LGE was present, suggesting co-existing coronary artery disease (Figure 1A). But coronary angiography showed unobstructed epicardial coronary arteries. In 25/30 (83%) patients of the infarct group the distribution of LGE was suggestive of underlying coronary artery disease, since hyper-enhancement started at the subendocardium and involved segments restricted to the vascular territory of specific coronary arteries. Sole mid-layer involvement was only seen in patients with CS, while LGE involving both the subendocardial and mid myocardial layers was predominantly seen in infarct patients. When considering sole mid layer, patchy subendocardial and three layer LGE to be diagnostic of CS, 26/30 (87%) would have been correctly classified (Figure 2A, C, E). Significantly more RV LGE (8 (27%) versus 3 (7%) patients, P = 0.01) and RV dilation (10 (33%) versus 2 (7%) patients, P = 0.02) was found in the CS group (Figure 3). The observers correctly diagnosed CS in 21/29 (72%), 24/29 (83%), 21/29 (72%), ischemic heart disease in 23/30 (77%), 23/30 (77%), 24/30 (80%) and normal controls in 9/10 (90%), 10/10 (100%), 10/10 (100%). The kappa values for interobserver agreement in differentiating between CS, infarcts and normal controls by assessing the LGE CMR studies were 0.91 (0.86, 0.95), and 0.86 (0.81, 0.92). The kappa value for intra-observer agreement in differentiating between CS, infarcts and normal controls by assessing the LGE CMR studies was 0.86 (0.81, 0.92). There was excellent intra- and interobserver correlation for ventricular volumes, masses, and ejection fractions. The intra-class correlation coefficient for LV LGE was 0.989 [0.981-0.993] (p=0.001).

Table 2.

Summary of the findings with gadolinium-enhanced cardiac magnetic resonance

Fig. 1.

Delayed-enhanced three-dimensional breath-hold inversion-recovery gradient echo studies (short axis views) demonstrating multi-focal transmural hyper-enhancement involving the inferior and lateral LV segments in a patient with sarcoidosis (A), and patchy (B) respectively confluent (C) transmural inferior LV (arrows) and RV (arrowheads) wall late gadolinium enhancement secondary to infarctions in patients with coronary artery disease of the RCA

Fig. 2.

Delayed-enhanced three-dimensional breath-hold inversion-recovery gradient echo studies (short axis views) of three patients with cardiac sarcoidosis, demonstrating: focal (A) and patchy, multi-segmental three layer hyper-enhancement (C and E), and three patients with coronary artery disease demonstrating: focal sub-epicardial infarctions in the distribution of respectively the LCX (B) and RCA (D), and two-vessel disease (F) involving the LAD (subendocardial infarction) and RCA (mid-layer infarction)

Fig. 3.

Delayed-enhanced three-dimensional breath-hold inversion-recovery gradient echo studies (A: 4 chamber view; B, D: short axis views) demonstrating multi-segmental, patchy three layer hyper-enhancement in a patients with sarcoidosis (A,B) and confluent transmural hyper-enhancement in the vascular distribution of the RCA (D). The delayed-enhanced spin echo study (C: short axis view) shows patchy, three layer involvement of the lateral LV segments. Transmural right ventricular enhancement secondary to sarcoidosis (C) and coronary artery disease (D) is demonstrated by the arrowheads

Discussion

Our study is the first to systematically compare the distribution of LGE in CS with the findings in patients with coronary artery disease. Our findings suggest that CECMR is helpful in the non-invasive differentiation between patients with CS and patients with coronary artery disease and previous myocardial infarcts. However, in approximately a third of patients with coronary artery disease, subepicardial or patchy, three layer LGE was found, a pattern that was suggestive of a non-ischemic etiology such as CS (6-8). Additional angiographic information was needed to correctly classify 11 patients with coronary artery disease and 3 with CS.

LGE results from the sustained presence of gadolinium in the myocardium, secondary to expansion of the interstitial space or impaired microvascular wash out (“no-reflow phenomena”), and has been reported in a variety of conditions, such as coronary artery disease, cardiomyopathies, myocarditis, and myocardial infiltration, that are characterized by necrosis, fibrosis or inflammation (2-9). In CS the presence of active, granulomatous inflammation and the resulting myocardial fibrosis are considered to be the underlying histopathological substrate resulting in the accumulation of gadolinium (2-4). Recent studies have determined subendocardial and transmural LGE to be diagnostic of underlying coronary artery disease, while sole mid-layer and epicardial LGE were predominantly seen in patients with non-ischemic cardiomyopathies (2-9). Although sole subepicardial myocardial fibrosis is considered rare in coronary artery disease, both our patients with epicardial LGE happened to be infarct patients (13) (Figure 3). Early presentation and primary percutaneous interventions may have resulted in smaller, myocardial scars with an atypical mural distribution (14) (Figure 2B, D, F, 4B, D).

The pattern of LGE in CS has been reported to be patchy, multi-segmental, not related to coronary artery territories, while predominantly involving the mid myocardial and subepicardial layers (2-4). When considering sole mid-layer, and patchy subendocardial or patchy transmural LGE to be diagnostic of CS, and confluent subendocardial or confluent transmural LGE diagnostic of coronary artery disease, the sensitivity, specificity and overall accuracy of LGE CMR for these conditions respectively would be 87%, 70%, 78%, and 63%, 90% and 77%. The significantly higher number of patients with loss of regional wall thickness, wall motion abnormalities, and generally poorer systolic LV function in the infarct group is explained by more extensive, confluent, transmural LGE in this group (15).

RV LGE and dilation was present in a substantial number of CS patients. These findings may be explained by primary RV myocardial involvement, but alternatively the presence of pulmonary arterial hypertension, resulting from extensive pulmonary sarcoidosis, has been considered a possible cause of RV fibrosis (16). The junction of right and left ventricle and right-sided interventricular septum seemed to be predominantly affected (Figure 4). However, further studies are needed to elucidate the underlying mechanisms and significance of RV LGE in patients with sarcoidosis. The absense of LGE in one patients with CS, with a dilated, globally hypokinetic LV, may be explained by the presence of diffuse myocardial fibrosis, which is not detected with the IR-GRE technique.

Fig. 4.

Delayed-enhanced three-dimensional breath-hold inversion-recovery gradient echo studies demonstrating confluent apical hyper-enhancement secondary to sarcoidosis (A: vertical long axis view, C: 4 chamber view) and apical patchy three-layer hyper-enhancement in a patient with coronary artery disease and myocardial infarction secondary to obstruction of the LAD (B: vertical long axis view, D: 4 chamber view)

IR-GRE is the current gold standard technique in the assessment of myocardial fibrosis, and we may have underestimated the amount of myocardial scar tissue by relying on Spin Echo sequences in 5 patients. The distribution of LGE in these particular patients was however strongly suggestive of CS, in that it concerned patchy subendocardial or three layer hyperenhancement. Since the majority of patients were not assessed with first-pass myocardial perfusion studies, we did not include these data. The presence of flow limiting coronary artery disease in the infarct group might have improved the diagnostic accuracy of CMR in this group.

In conclusion, the presence of sole mid-layer, and patchy, subendocardial or three layer LGE with a non-vascular distribution, in patients with extra-cardiac sarcoidosis suggests cardiac involvement, and differentiates these patients from patients with coronary artery disease.

Abbreviations

CECMR

Contrast-Enhanced Cardiovascular Magnetic Resonance

CS

cardiac sarcoidosis

IR-GRE

inversion-recovery gradient echo

LV

left ventricle

LGE

late gadolinium enhancement

RV

right ventricle