Author + information
- Received October 28, 2015
- Revision received January 5, 2016
- Accepted January 14, 2016
- Published online November 1, 2016.
- Nicola Gaibazzi, MD, PhDa,∗ (, )
- Michele Bianconcini, MDa,
- Nicola Marziliano, PhDb,
- Iris Parrini, MDc,
- Maria Rosa Conte, MDc,
- Carmine Siniscalchi, MDa,
- Giacomo Faden, MDd,
- Pompilio Faggiano, MDe,
- Filippo Pigazzani, MDa,
- Francesca Grassi, MDa and
- Lisa Albertini, MDa
- aParma University Hospital, Parma, Italy
- bUOC Cardiologia-ASL3 Ospedale San Francesco, Nuoro, Italy
- cMauriziano Hospital, Torino, Italy
- dPoliambulanza Hospital, Brescia, Italy
- eBrescia Civili University Hospital, Brescia, Italy
- ↵∗Reprint requests and correspondence:
Dr. Nicola Gaibazzi, Parma University Hospital, Via Gramsci 14, Parma 43123, Italy.
Objectives This study sought to assess an echocardiographic approach (scar imaging echocardiography with ultrasound multipulse scheme [eSCAR]), based on existing multipulse ultrasound scheme, as a marker of myocardial scar in humans, compared with cardiac magnetic resonance assessing late gadolinium enhancement (CMR-LGE).
Background The detection of myocardial scar impacts patient prognosis and management in coronary artery disease and other types of cardiac disease. The clinical experience with echocardiography suggests that the reflected ultrasound signal is often significantly enhanced in infarcted myocardial segments.
Methods Twenty patients with a recent ST-segment elevation myocardial infarction (STEMI) (cases) and 15 patients with absent CMR-LGE (negative controls) were imaged with both the eSCAR pulse-cancellation echocardiography and CMR-LGE to assess their potential association.
Results Scar was detectable at CMR-LGE in 19 of 20 STEMI patients (91%), whereas all (100%) demonstrated eSCAR at echocardiography. In the 19 STEMI patients in whom CMR-LGE was detected, regional matching between eSCAR and CMR-LGE was total, although the segmental extent of detected scar was not always superimposable, particularly in the most apical segments, a region in which eSCAR demonstrated undersensitivity for the true extent of scar.
Conclusions A 2-dimensional multipulse echocardiography allows detection of myocardial scar, reliably matching the presence and site of CMR-LGE at 30 days after STEMI, or its absence in negative controls.
The detection of myocardial scar or severely fibrotic areas profoundly impacts patient management and prognostic stratification, both in coronary artery disease and other cardiac conditions (1–5). Myocardial scar detection currently requires nuclear medicine techniques or cardiac magnetic resonance assessing late gadolinium enhancement (CMR-LGE). These methods are costly, time consuming, and unfit for wide population studies; in addition, CMR-LGE is generally contraindicated in patients with cardiac devices. All of these factors limit the widespread clinical use of scar imaging. Clinical experience with echocardiography suggests that patients with a prior myocardial infarction not only demonstrate wall motion abnormalities, but the reflected ultrasound signal is often enhanced significantly in the infarcted myocardial segments (6–8).
We tested a pulse cancellation ultrasound technique (scar imaging echocardiography with ultrasound multipulse scheme [eSCAR]) to differentiate normal from scarred myocardium; the technique is in principle not different from CMR-LGE pulse-inversion technique: 2 ultrasound signals—phase or amplitude shifted—are transmitted and cancel each other when reflected by normal myocardium, but not in case of abnormally fibrotic or disarrayed myocardium, which returns a measurable, nonzero signal, due to nonlinear ultrasound response. In time, eSCAR would bring scar detection to the bedside, potentially expanding scar assessment to any type of patient.
We aimed to establish whether eSCAR is an accurate surrogate for the presence of CMR-LGE in humans, testing patients with a recent ST-segment elevation myocardial infarction (STEMI), versus patients with nonischemic suspect cardiomyopathies and absence of CMR-LGE, who were chosen as negative controls.
Study design and population
This prospective pilot study enrolled 21 consecutive patients with recent STEMI treated with coronary angiography and primary percutaneous angioplasty (if required), who consented to participate (group A, cases), and 15 patients who underwent CMR-LGE during the same study period for a clinical indication of suspect cardiomyopathy with known normal coronary arteries (or <30% coronary artery stenosis) at recent coronary angiogram performed <6 months before (group B, negative controls). Exclusion criteria were common to both groups and were the presence of prior history of acute coronary syndrome, prior coronary revascularization, severe chronic kidney disease, intracardiac pacing leads or other devices precluding CMR-LGE, hemodynamic instability, or known claustrophobia. The study was performed at the Parma University Tertiary Medical Center, Parma, Italy. Demographic data, clinical characteristics, and concomitant drug therapy were recorded. In the STEMI group, echocardiography was performed between 28 and 32 days after STEMI and CMR-LGE was performed within 72 h after echocardiogram.
In group B (negative controls) the absence of scar at CMR-LGE was part of inclusion criteria, so that echocardiography was performed after their CMR study, within 72 h. Echocardiography was performed using both standard 2-dimensional (2D) and the newly devised eSCAR ultrasound setting, which is detailed below. The same echocardiography imaging protocol was used for both STEMI cases (group A) and control patients (group B). The ultrasound eSCAR data were compared with CMR-LGE results in both groups. The study complies with the Declaration of Helsinki and the locally appointed ethics committee has approved the research protocol and informed consent has been obtained from the subjects.
Scar imaging echocardiography with ultrasound multipulse scheme
A standard echocardiography machine (Philips ie33, Philips Medical Systems, Best, the Netherlands), equipped with standard S5 phased array 2D transthoracic probe was used for ultrasound examinations. Although the built-in setting for left ventricle opacification (LVO) is provided by vendors to be used in conjunction with ultrasound contrast, thanks to cancellation of “linear” signals back from normal myocardium, it is incidentally very efficient to enhance signals from abnormal myocardial tissue, such as fibrotic (or calcified) tissues, which on the contrary show “nonlinear” response (similarly to the nonlinear acoustic behavior of microbubbles) (Figure 1).
Starting from the 2D standard setting, the “iscan button” on the machine panel was pushed once (set at 0 dB) to automatically normalize gains, and then the built-in 2D LVO setting, originally devised for left ventricle contrast opacification, was activated, exploiting power-modulation/pulse inversion harmonic imaging (transmit 1.6 MHz/receive 3.2 MHz). The LVO setting was tuned to an intermediate mechanical index, between 0.40 and 0.47, and general gain between 70% and 77%, depending on individual patient echogenicity.
Each echocardiographic examination comprised at least 2 clips of apical 4-, 2-, and 3-chamber views, which were saved on a hard disk as DICOM and AVI files for offline reading.
Patients were examined supine in a 1.5-T imaging unit (Achieva, Philips Medical Systems) equipped with master gradients (30 mT/m peak gradients; 150 mT/m/ms slew rate) and a 5-element cardiac phased-array receiver coil. Images were acquired with the use of electrocardiographic gating and expiratory breath holds. All data were acquired in the true LV short-axis, with 10 to 12 contiguous sections as required to cover the entire LV. A dose of 0.2 mmol/kg of body weight of gadopentetate dimeglumine (Magnevist; Bayer Schering Health Care, Cambridge, United Kingdom) was administered intravenously at a rate of 5 ml/s with a power injector. Ten minutes after contrast agent injection, a Look-Locker sequence was performed to obtain the most appropriate inversion time to null the signal intensity of normal myocardium. This was immediately followed by acquisition of LGE images, with an inversion-recovery prepared T1-weighted gradient-echo sequence (4.9/1.9; flip angle, 15°; turbo field-echo factor, 30; spatial resolution, 1.35 × 1.35 × 10 mm).
Visual analysis of eSCAR images was used for the assessment of the presence and segmental distribution of scar by consensus of 2 expert echocardiographers, unaware of patient clinical history, based on the 17-segment model (9). We defined eSCAR as any myocardial area demonstrating a continuous series of nonblack pixels, taking care not to include the pericardium, valve leaflets, or other nonmyocardial fibrous/calcific structures, which are also enhanced when using the eSCAR setting (Figure 2), using standard 2D images as a guidance if needed. Apical 4-, 2-, and 3-chamber view clips were assessed in diastole, in the clearest still image just before, during, or right after the P-wave of the electrocardiogram. Isolated mid-myocardial interventricular septal signals in the 4-chamber view (also known as septal stripes) were not considered, if present, because they do not to represent true scar according to CMR-LGE (5).
Visual analysis was used similarly for the assessment, distribution, and scoring of CMR-LGE, based on the 17-segment model; CMR-LGE was interpreted as present or absent by the consensus of 2 CMR-trained physicians. CMR-LGE was considered present only if confirmed on both short-axis and matching long-axis myocardial locations. Long-axis images (4-, 2-, and 3-chamber equivalent) were used for comparison with eSCAR data. The distribution of eSCAR or CMR-LGE was defined as either subendocardial (<50% myocardial thickness) or transmural (≥50%). If both patterns were present, the distribution was characterized by the predominant pattern.
Scar area quantitation
Both echocardiographic eSCAR and CMR-LGE images were additionally assessed quantitatively in group A (STEMI patients) (Figure 3), using only 4- and 2-chamber views. ImageJ software version 1.49 (National Institutes of Health, Bethesda, Maryland) was used for image processing and measurement of videointensity in manually traced regions of interest. First, the image contrast, already visually apparent with the eSCAR setting, between enhanced whiter areas and normal myocardium (fully black) was enhanced further by the use of the software binary filter (“default” thresholding method), producing only black or white pixels. Regions of interest were then positioned onto each myocardial area demonstrating white pixels, starting from the endocardium to the epicardium, taking care not to include the pericardium, valve, or other nonmyocardial structures, which are characterized by strong signals when using eSCAR setting. The eSCAR area was then quantified by the area measurement function of the software, after proper scale was set. Isolated mid-myocardial interventricular septum signals in the 4-chamber view were avoided, if present. The 4- and 2-chamber view clips were always measured, and measurements performed in diastole, by selecting the clearest image just before, during or right after the P wave of the electrocardiogram as a reference. A frame in which the pericardium was differentiated clearly from the myocardium was preferred. The same method was applied for CMR-LGE images.
Basic descriptive statistics was used, and association between CMR-LGE and eSCAR data was simply described by means of tables. Interobserver and test–retest agreement were assessed by means of kappa (weighted by 1 − [(i − j)/(1 − k)]2) between the number of eSCAR segments adjudicated by 2 readers, or by the same reader who repeated the eSCAR assessment after 3 days.
Twenty-one patients who presented with STEMI between June and October 2014 and between August and October 2015, who underwent coronary angiography and primary angioplasty (if indicated) within 6 h from symptom onset were enrolled; however, 1 patient did not complete the diagnostic protocol due to CMR-LGE intolerance. Twenty patients with both ultrasound eSCAR and CMR-LGE data formed the final STEMI study group (Table 1). Table 2 shows the clinical, CMR-LGE, and echocardiographic eSCAR data for the STEMI patients, indicating the presence or absence (dominant subendocardial or transmural myocardial involvement) of CMR-LGE, eSCAR, and respective segmental involvement of the left ventricle.
Scar was detectable at CMR-LGE in 19 of 20 STEMI patients (95%), and all 20 patients (100%) demonstrated eSCAR at echocardiography. In fact, there was 1 case (Patient #9) with no demonstrable CMR-LGE, but showing single-segment subendocardial eSCAR; this patient had an anterior STEMI with transient ST-segment elevation and underwent coronary angiography and angioplasty after chest pain had already subsided, with spontaneous partial reperfusion and ST-segment tract recovery. Cardiac enzymes were only mildly increased in this patient (Figure 4).
In all STEMI patients, the scarred territory, according to either scar detection method (CMR-LGE or eSCAR), was linked to the STEMI culprit artery treated with primary angioplasty, although in 1 case (Patient #11) a culprit plaque amenable to angioplasty was not identified, and hence angioplasty not performed, due to possible microembolic etiology (Figure 5). This young female patient demonstrated interatrial defect in the context of an interatrial septal aneurysm, predisposing to microembolization, which was evident angiographically in some distal marginal branches of the circumflex artery; the electrocardiographic ST-segment elevation was mainly inferior and both eSCAR and CMR-LGE were detected in the anterolateral and inferior myocardium.
In the 19 STEMI patients in whom CMR-LGE was detectable, regional (i.e., coronary territory) matching between eSCAR and CMR-LGE was total, although the segmental extent and myocardial thickness percentage extension of the scar was not always superimposable, as expected using different methods (Table 2). In most apical segments, eSCAR demonstrated low sensitivity for scar; as an example, only 2 of 9 patients with CMR-LGE detected in the true apex (segment 17) were also noted to have eSCAR in that segment. Quantitative data (Figure 6) confirmed a significant underestimation of scar area by eSCAR method in the left anterior descending coronary artery territory, whereas in the posterior circulation (either left circumflex or right coronary artery territories) quantitative matching between eSCAR and CMR-LGE was much closer. Although the relationship (Pearson linear correlation) between quantitatively determined eSCAR and CMR-LGE was only moderate (r = 0.47) in the left anterior descending coronary artery territory, it was almost perfect (r = 0.98 and r = 0.97, respectively) in the circumflex and right coronary artery territories. See the Online Appendix for detailed rank correlation data and intraclass coefficients (with Bland and Altman plots) split based on individual coronary territory involved.
Figure 7 shows patient-by-patient visual analysis of scar tissue in the STEMI group, with bull’s eye segmental comparison of CMR-LGE and eSCAR findings.
The eSCAR visual analysis agreement between 2 observers for the 20 STEMI cases (number of segments with eSCAR detected) was substantial, with weighted κ = 0.62, and an optimal agreement for test-retest (weighted κ = 0.95). All controls were deemed fully negative for eSCAR by both readers.
Negative control patients
Fifteen patients who underwent CMR for suspected or known nonischemic cardiomyopathy with no detectable CMR-LGE were enrolled and underwent additional echocardiography with eSCAR assessment within 72 h after their CMR-LGE (Table 1). All of these patients did not demonstrate the presence of eSCAR, as defined in the current study, although 10 of 15 (67%) showed some degree of mid-myocardial septal stripe.
Online Table 1 shows the clinical, CMR-LGE and echocardiographic eSCAR data in negative controls, in all of whom no eSCAR was apparent. Figure 8 demonstrates the parallel behavior of eSCAR and CMR-LGE in one of the cardiomyopathy patients, per-protocol all without demonstrable CMR-LGE.
In this proof-of-concept study, we demonstrate that the presence of positive eSCAR findings, is able to accurately identify true myocardial scar, using CMR-LGE as a gold standard. This method is an application of long-standing existing technology, widely available in most echolabs, slightly modified in its presets and applied in a different field with respect to contrast echocardiography, which was the original aim of this technology.
The eSCAR method was tested in patients 1 month after STEMI and in controls who underwent CMR for suspect or known nonischemic cardiomyopathy, but without CMR-LGE findings. The presence, coronary territory, and segmental extension of eSCAR and CMR-LGE matched almost completely, apart from the apical segments, in which eSCAR was less sensitive than CMR-LGE to pick up scar, although this quantitative underestimation did not preclude the detection of the presence of scar in such left anterior descending coronary artery territory. We speculate that the undersensitivity in the most apical segments might be caused by the partial dependence of eSCAR on returning harmonics, because harmonics are generated by ultrasound waves travelling through tissue and harmonic signal is hence significantly reduced from more superficial myocardial areas (short distance between probe and myocardium), compared with deeper regions as the ones pertinent to the posterior coronary circulation. Scar was detectable in the territory of the culprit coronary artery in 19 of 20 STEMI patients when using CMR-LGE and in all 20 when using eSCAR, raising the hypothesis that, under ideal image quality, the high temporal and spatial resolution of echocardiography makes it even more sensitive than CMR-LGE for very small, subendocardial thin areas of scar, as in the case of the only scar apparently missed by CMR-LGE. Importantly, there were no false-positive results with eSCAR, either in nonculprit territories of STEMI patients or in group B negative controls with suspect cardiomyopathies and no CMR-LGE.
The main finding of the current study is that scar can be detected by echocardiography, easily, reliably, at the bedside, and without the need for radiation or contrast media. This makes the eSCAR method extremely attractive for widespread clinical use, particularly for large-scale screenings, and in each type of patient in whom the presence of significant scar would possibly reclassify their prognosis, or guide management and medical or device therapy, with no constraint regarding the exclusion of subjects who carry intracardiac devices, who are often precluded from CMR-LGE.
The idea of using echocardiography for tissue characterization dates back to 1975, and it was later embraced by researchers in the 1980s, mainly by the use of integrated radiofrequency backscatter analysis or videointensity in standard 2D images. Histopathologic and histochemical studies verified that the increase in echocardiography amplitude correlated with the evolution of infarction healing and myocardial scar formation. The integrated backscatter approach failed, however, to demonstrate feasibility in clinical practice.
More recently, Montant et al. (8) revived the concept of hyperechoic scar tissue by the use of 3-dimensional (3D) echocardiography with contrast media used for endocardial border delineation. They tested 3D echocardiography versus CMR-LGE and concluded that standard harmonic imaging with transmission/receive 1.6/3.2 MHz frequency at an intermediate 0.5 mechanical index (and no contrast-specific multipulse scheme) was the best combination to differentiate normal myocardium from scar. Unfortunately, the same 0.5-mechanical index, 1.6/3.2 MHz setting was not tested with pulse modulation/inversion scheme, probably because such frequencies were not supported by the 3D probe they used or possibly because contrast-specific pulse schemes in conjunction with 3D imaging increased artifacts. The addition of pulse modulation/inversion in conjunction with 2D imaging, as we used in our study, greatly benefits the differentiation of scar from normal myocardium compared with standard harmonic imaging. The basic LVO setting is a factory setting in the echocardiography machine we used, which simply needs trimming to a slightly higher mechanical index. We expect modified LVO settings from other vendors to perform similarly to the one used in our study, but this could not be tested in the current study using equipment from a single vendor.
Scar quantitation and 2D versus 3D echocardiography
We believe full and accurate quantitation of scar is inherently out of reach of ultrasound eSCAR due to higher grade of subjectivity in its application and interpretation when compared with CMR-LGE. Our study relied mainly on the visual semiquantitative assessment of eSCAR, simply (but robustly) based on the number of scarred segments, although we also present quantitative data.
Three-dimensional imaging would give theoretical advantages compared with 2D imaging, that is, the potential for volumetric scar measurement, resulting in better comparability with CMR-LGE data. Nonetheless, the advantages of full-volume acquisition would be counterbalanced by a worse signal-to-noise ratio and lower spatial and temporal definition, which would complicate, together with “stitch” artifacts, such a demanding technique as contrast-specific multipulse cancellation schemes.
Clearly, CMR-LGE remains the gold standard for scar detection, specifically due to its precise quantitation, but eSCAR by echocardiography could be a promising alternative when semiquantitation (number of scarred segments) or binary “scar: yes/no” assessment is deemed sufficient or CMR-LGE is not technically feasible (intracardiac leads, hemodynamic instability, claustrophobia), as frequently happens in clinical practice.
Septal stripe is a line or islets of hyperechoic pixels longitudinally oriented from base to apex in the middle of the interventricular septum, frequently detected in 4-chamber view, but also in the apical 3-chamber view. This structure (excluded from the eSCAR definition) was common in STEMI patients and controls, but should not be confounded with true scar, similar to nonmyocardial structures such as the pericardium, valve chordae, leaflets, and annulus. The interpretation of eSCAR images has a basic learning curve to recognize true scar from these cardiac structures. In the current study, eSCAR was compared with CMR-LGE only in STEMI patients, and suspect or known cardiomyopathy patients used only when showing no CMR-LGE. eSCAR parallel behavior with CMR-LGE should not be extrapolated to cardiomyopathy or myocarditis patients showing CMR-LGE.
A validation study involving pathologic confirmation, similar to the initial validation of CMR-LGE in animal studies, would be welcome for eSCAR, but is out of the scope of the current study. Such a study would be conceptually difficult, because the adaptation of ultrasound power and other settings to the animal model would make it difficult to translate back to humans. In the end, validation of eSCAR against CMR-LGE may be sufficient, because CMR-LGE is considered the current gold standard for in vivo evaluation of cardiac scars.
Gain dependency and image quality remain as issues in echocardiography, because excessive increase of gain may enhance also normal or marginally fibrotic myocardium, whereas low echogenicity may lead to false-negative results.
During eSCAR imaging, endocardial and epicardial borders can be identified by simply temporarily and briefly increasing gain or the mechanical index, so that the use of contrast was not deemed necessary. Moreover, strictly subendocardial scars could theoretically be missed using contrast, possibly due to shading or contrast complicating the visual differentiation between subendocardial scar and contrasted cavity (8).
A 2D multipulse echocardiography allows detection of myocardial scar, matching the presence and specific territory of CMR-LGE at 30 days after STEMI, or its absence in negative controls, although eSCAR suffered from undersensitivity in the most apical myocardial segments. If the current findings are confirmed by other independent studies, this could offer an easy and cost-effective strategy to detect myocardial scar that is available in most echocardiography laboratories and also suitable for patients with implanted cardiac devices.
COMPETENCY IN MEDICAL KNOWLEDGE: In a small prospective cohort of patients with recent STEMI and in negative controls with suspect cardiomyopathy without scar at late CMR-LGE, a newly devised echocardiography method demonstrated the ability to detect scarred myocardium, similar to CMR-LGE, although with less sensitivity in the apical segments.
TRANSLATIONAL OUTLOOK: Further studies are needed to better appreciate the yield of this new and simple method and determine whether this method can extend scar detection out of the current boundaries imposed by the complexity of CMR.
For supplemental tables, figures, and videos, please see the online version of this article.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- cardiac magnetic resonance assessing late gadolinium enhancement
- scar imaging echocardiography with ultrasound multipulse scheme
- left ventricle opacification
- ST-segment elevation myocardial infarction
- Received October 28, 2015.
- Revision received January 5, 2016.
- Accepted January 14, 2016.
- American College of Cardiology Foundation
- Kwong R.Y.,
- Chan A.K.,
- Brown K.A.,
- et al.
- Kuruvilla S.,
- Adenaw N.,
- Katwal A.B.,
- Lipinski M.J.,
- Kramer C.M.,
- Salerno M.
- Chan R.H.,
- Maron B.J.,
- Olivotto I.,
- et al.
- Neilan T.G.,
- Coelho-Filho O.R.,
- Danik S.B.,
- et al.
- Hoyt R.H.,
- Collins S.M.,
- Skorton D.J.,
- Ericksen E.E.,
- Conyers D.
- Montant P.,
- Chenot F.,
- Goffinet C.,
- et al.
- Ortiz-Pérez J.T.,
- Rodríguez J.,
- Meyers S.N.,
- Lee D.C.,
- Davidson C.,
- Wu E.