Key words
cardiac - heart - tissue characterization - MR imaging
Introduction
Late gadolinium enhancement (LGE), also known as delayed enhancement (DE) or late
enhancement (LE), is a widely used cardiac magnetic resonance imaging (MRI) technique
and the current noninvasive reference standard for the assessment of myocardial viability
[1]. Since its initial validation against histology approximately two decades ago [2]
[3], its ability to distinguish areas of myocardial infarction from normal myocardium
has gained wide acceptance. Over the years, the clinical utility of LGE has broadened
considerably, including the detection of non-ischemic enhancement patterns [4]
[5], pre-procedural assessment of arrhythmogenic substrate [6]
[7], and visualization of radiofrequency-induced ablation lesions [8]
[9]. More recently, LGE emerged as a promising tool for guiding subsequent electrophysiology
procedures to treat cardiac arrhythmias [10]
[11]. Nowadays, LGE is used to diagnose a broad range of ischemic and non-ischemic cardiomyopathies
in everyday clinical practice.
In this review, we not only discuss the basic concepts of LGE imaging and its diagnostic
and prognostic value, but also elaborate on emerging methods in the field of LGE,
including dark-blood techniques, high-resolution (isotropic) 3 D imaging, water/fat-separated
methods, motion correction techniques, and semi-automated scar quantification.
Basic LGE concepts
LGE imaging requires the use of a gadolinium-based contrast agent (GBCA) which is
intravenously administered to the patient. GBCAs are routinely used in MRI examinations
for their strong T1-shortening effect and tissue dependent distribution within the body. Particularly
for LGE MRI, the time required for the GBCA to reach the cardiac tissue compartments
after injection plays a major role in the acquired image contrast. As various compartments
within the myocardium have different wash-in and wash-out kinetics, the GBCA concentration
in these individual compartments will constantly vary after contrast injection [12]. While the GBCA reaches the normal myocardium already soon after injection, areas
of fibrosis are reached at a much slower rate ([Fig. 1]). A similar effect is observed for the clearance of the GBCA. While the GBCA is
cleared quickly after reaching its peak concentration in the normal myocardium, it
resides in areas of fibrosis for a much longer time.
Fig. 1 Schematic illustration of the gadolinium-based contrast agent (GBCA) concentration
in both healthy myocardium (green) and scar tissue (cyan) after a bolus injection.
While the GBCA quickly washes in and out of the normal myocardium, delayed wash-in
and wash-out is observed for scar tissue. At approximately ten minutes post-injection,
a phase known as “late (gadolinium) enhancement” commences (gray area), where scar
tissue contains a significantly larger GBCA concentration compared to normal myocardium.
Abb. 1 Schematische Darstellung der Gadolinium-Kontrastmittel (GBCA)-Konzentration im gesunden
Myokard (grün) als auch im Narbengewebe (cyan) nach einer Bolus-Injektion. Während
das GBCA schnell in das normale Myokard ein- und ausgewaschen wird, wird im Narbengewebe
ein verzögertes Ein- und Auswaschen beobachtet. Ungefähr zehn Minuten nach der Injektion
beginnt eine als „Late (Gadolinium)-Enhancement“ bekannte Phase (grauer Bereich),
in der Narbengewebe im Vergleich zu normalem Myokard eine signifikant höhere GBCA-Konzentration
enthält.
Aside from the different wash-in and wash-out kinetics, the local distribution is
of equal importance for LGE. The used GBCAs are extracellular and cannot cross the
intact cell membranes of the normal myocardium, thereby limiting their distribution
volume to the interstitial space. However, in patients with acute myocardial injury,
the cell membranes of the affected myocytes have ruptured and the GBCA can now also
access the ‘intracellular’ space, leading to an increased distribution volume [12]. Over time, scar tissue forms and the affected myocytes are replaced by a collagen
matrix. As a result, the interstitial space is increased and thus also the distribution
volume for GBCAs. Ultimately, both in patients with acute and chronic myocardial injury,
the distribution volume for GBCAs is increased compared to that in healthy subjects.
The combined effect of reduced wash-in/wash-out kinetics and the increased distribution
volume leads to a delayed accumulation of the GBCA in areas of myocardial injury approximately
ten minutes post-injection [13]. Because of the strong T1-shortening effect of GBCAs, these areas of accumulation lead to a delayed hyperenhancement
of the signal, hence the name “late enhancement” or “delayed enhancement”.
Standard LGE sequence
While the areas of myocardial injury experience a strong decrease in T1 relaxation time due to prolonged GBCA accumulation, after approximately ten minutes
most of the GBCA has already been cleared from the normal myocardium, resulting in
a much longer T1 relaxation time there [13]. In order to maximize contrast between the normal myocardium and areas of myocardial
injury, a heavily T1-weighted segmented inversion-recovery (IR) gradient-echo pulse sequence is used ([Fig. 2]) [3]. IR pulse sequences start with a non-selective 180-degree inversion radiofrequency
(RF) pulse that inverts the longitudinal magnetization of all tissues. The IR pulse
is followed by a time delay, called the “inversion time” or simply TI. During this
delay, the longitudinal magnetization of all tissues will recover to its equilibrium
with a rate determined by the tissue-specific T1 relaxation time. Finally, after the TI, a spoiled gradient-echo readout of approximately
100–150 ms is performed to acquire the IR-prepared MR signal.
Fig. 2 Conventional inversion-recovery sequence used for late gadolinium enhancement (LGE).
A non-selective 180 degrees inversion radiofrequency pulse inverts all magnetization
levels. A time delay follows, called the inversion time (TI), in which all longitudinal
magnetization (Mz) levels recover with a rate determined by the tissue-dependent T1 relaxation time. Following the TI, acquisition of the signal takes place. Note that
for conventional LGE the TI is set to null the magnetization level of normal myocardium
(green curve at 0), leading to a black appearance of the normal myocardium in the
resulting (magnitude) image.
Abb. 2 Konventionelle Inversion-Recovery-Sequenz für „Late Gadolinium-Enhancement“ (LGE).
Ein nicht-selektiver 180-Grad Inversions-Radiofrequenzimpuls invertiert alle Magnetisierung.
Es folgt eine Zeitverzögerung, die als Inversionszeit (TI) bezeichnet wird, in der
sich alle Längsmagnetisierungen (Mz) abhängig von der gewebeabhängigen T1-Relaxationszeit erholen. Nach der TI erfolgt die Akquisition des Signals. Für das
konventionelle LGE wird die TI so eingestellt, dass die Magnetisierung des normalen
Myokards (grüne Kurve bei 0) genullt ist, was zu einem schwarzen Erscheinungsbild
des normalen Myokards im resultierenden (Magnituden-) Bild führt.
A “Look-Locker” scan, also known as a TI scout scan, provides a series of low-resolution
images with an increasing TI and is usually performed prior to LGE imaging to determine
the desired TI ([Fig. 3]) [14]. Traditionally, the TI is set so that the longitudinal magnetization of the normal
myocardium is zero when the signal is acquired [15]. Nulling the normal myocardium will then lead to a black appearance on magnitude
images, while the blood pool and areas of myocardial injury appear enhanced due to
their shortened T1 relaxation time.
Fig. 3 Example of a “Look-Locker” or scout scan that provides a series of low-resolution
images with increasing inversion time (TI). After injection of the contrast agent,
the scar tissue and blood pool have a shorter T1 relaxation time compared to the normal myocardium, leading to a faster recovery of
the longitudinal magnetization (Mz). Therefore, the scar tissue (cyan curve) reaches
the zero-magnetization level first, shown as a black appearance in the magnitude image
(cyan arrowhead in image 1). Soon after, the blood pool follows (orange curve + dot
in image 2), and, finally, the normal myocardium (green curve + dot in image 4). In
this example, the TI that corresponds to image 4 should be set for conventional myocardium-nulled
(bright-blood) LGE.
Abb. 3 Beispiel eines „Look-Locker“-Scans, der eine Reihe von Bildern mit niedriger Auflösung
mit zunehmender Inversionszeit (TI) liefert. Nach Injektion des Kontrastmittels haben
das Narbengewebe und der Blutpool im Vergleich zum gesunden Myokard eine kürzere T1-Relaxationszeit, was zu einer schnelleren Erholung der Längsmagnetisierung (Mz) führt.
Daher erreicht das Narbengewebe (cyanfarbene Kurve) zuerst das Nullmagnetisierungsniveau,
was als schwarze Erscheinung auf dem Magnitudenbild angezeigt wird (cyanfarbene Pfeilspitze
in Bild 1). Bald darauf folgt der Blutpool (orange Kurve und Punkt in Bild 2) und
schließlich das normale Myokard (grüne Kurve und Punkt in Bild 4). In diesem Beispiel
sollte die TI, die Bild 4 entspricht, für das konventionelle („bright-blood“) LGE
eingestellt werden.
As an alternative to a conventional IR sequence, it is possible to use a phase-sensitive
inversion-recovery (PSIR) sequence that applies a 180 degrees inversion RF pulse only
once every two heartbeats ([Fig. 4]). During every second heartbeat, a reference image is acquired using a low flip
angle excitation (usually only 5 degrees). This reference image is used to determine
the phase, and thus the sign, of the MR signals acquired every first heartbeat. PSIR
is therefore able to distinguish positive from negative magnetization levels. Negative
magnetization levels will be represented differently in the corrected real (or PSIR)
image that is reconstructed: negative magnetization levels appear darkest, nulled
magnetization levels appear mid-gray, and positive magnetization levels appear bright
[16]. This is in contrast to conventional magnitude image reconstruction, where nulled
magnetization levels appear darkest, and both negative and positive magnetization
levels appear bright. When PSIR is used in combination with conventional myocardium-nulled
LGE, the clinical observer may adjust window levels to further darken normal myocardial
tissue, mimicking a magnitude image representation. In clinical practice, PSIR is
recommended to make LGE image quality less sensitive to the chosen TI, leading to
a reduction in image artifacts and potential misinterpretations [1]. Additionally, in contrast to standard IR, PSIR is a so-called “two-beat” sequence
with only a single inversion RF pulse every two heartbeats. Therefore, PSIR is more
robust to heart rate variations and cardiac arrhythmias since it relies to a lesser
extent on a constant time delay between successive inversion RF pulses, averaging
irregular heartbeats over two heart beats. Inherently, however, scan duration doubles
when using PSIR instead of standard IR.
Fig. 4 Phase-sensitive inversion-recovery (PSIR) versus standard inversion-recovery (IR):
whereas a 180 degrees inversion pulse is applied in IR every heartbeat, this same
pulse is applied only once every two heartbeats in PSIR. In the second heartbeat,
a low flip angle (5 vs. 25 degrees) reference acquisition is performed at exactly
the same trigger delay time as the actual image acquisition (in the first heartbeat).
This reference acquisition helps to distinguish between positive and negative longitudinal
magnetization (Mz) levels and enables the construction of the additional corrected
real (PSIR) image. This resulting PSIR image uses a different grayscale compared to
the conventional magnitude grayscale used for standard IR LGE. Negative magnetization
levels appear darkest, nulled magnetization levels appear mid-gray, and positive magnetization
levels appear bright.
Abb. 4 Phasen-sensitive Inversion-Recovery (PSIR) versus Standard Inversion-Recovery (IR):
Während in der IR bei jedem Herzschlag ein 180-Grad-Inversionspuls appliziert wird,
wird dieser Impuls in der PSIR nur einmal alle zwei Herzschläge appliziert. Beim zweiten
Herzschlag wird eine Referenzaufnahme mit niedrigem Flip-Winkel (5 vs. 25 Grad) mit
genau derselben Trigger-Verzögerung wie die eigentliche Bildaufnahme (des ersten Herzschlags)
aufgenommen. Diese Referenzaufnahme hilft zwischen positiver und negativer Längsmagnetisierung
(Mz) zu unterscheiden und ermöglicht die Konstruktion eines zusätzlichen korrigierten
realen (PSIR-) Bildes. Dieses resultierende PSIR-Bild verwendet eine andere Grauskala
als die herkömmliche Grauskala, die für den Standard verwendet wird. Negative Längsmagnetisierung
wird am dunkelsten, genullte Längsmagnetisierung mittelgrau und positive Längsmagnetisierung
hell dargestellt.
Free-breathing 3 D methods
Free-breathing 3 D methods
Currently used clinical LGE techniques employ 2 D breath-hold sequences, which are
performed in 12–15 subsequent breath holds to cover the entire left ventricle (LV)
in the short-axis view. Additional 2 D breath-hold LGE scans can be acquired in the
long-axis views for a complete LV evaluation. Such 2 D LGE sequences generally provide
very good image quality but may miss small infarcts due to the limited spatial resolution
and associated partial volume effects. In addition, the repetitive breath holding
can be challenging for elderly patients or patients with heart failure. Alternatively,
free-breathing 3 D LGE imaging that is able to provide higher spatial resolution required
for the visualization of small scar regions, and can be performed in patients who
are unable to hold their breath, has been proposed. In addition, acquiring a whole-heart
3 D volume with (near) isotropic resolution enables the use of multiplanar reformatting
(MPR). Instead of being limited to prescribed short- and long-axis views, one can
freely reconstruct any desired image orientation with sufficiently high spatial resolution
([Fig. 5]).
Fig. 5 3 D late gadolinium enhancement imaging with high isotropic resolution (1.6 × 1.6 × 1.6 mm
acquired) in a patient with myocardial infarction. The high isotropic resolution enables
the use of multiplanar reformatting (MPR), allowing for the selection of desired imaging
planes after acquisition. Note that the three cardiac imaging planes (right) are all
reconstructed from the same 3 D volume acquired as transverse slices (left).
Abb. 5 3D-Bildgebung mit Late Gadolinium Enhancement mit hoher isotroper Auflösung (1.6 × 1.6 × 1.6 mm
erworben) bei einem Patienten mit Myokardinfarkt. Die hohe isotrope Auflösung ermöglicht
multiplanare Reformationen (MPR), was die Auswahl der gewünschten Bildgebungsebenen
nach der Aufnahme ermöglicht. Die drei kardialen Bildgebungsebenen (rechts) wurden
alle aus demselben 3D-Volumen rekonstruiert, das als transversales Schichtpaket (links)
erfasst wurde.
Initial 3 D implementations employed diaphragmatic navigators for respiratory motion
correction and shorter acquisition windows to minimize cardiac motion but used the
same magnetization preparation scheme and imaging sequence as the clinically used
conventional 2 D LGE sequences [17]
[18]. As the IR preparation pulse is set to null normal myocardium, it also leads to
a very low liver signal which can compromise the performance of the respiratory navigator.
To overcome this challenge, the use of a navigator restore pulse that re-inverted
the longitudinal signal in the liver immediately after the IR preparation pulse was
proposed [18]. While this approach was successful in improving the performance of the navigator,
the restore pulse also labelled blood in the vena cava that could lead to a high blood
signal in the right heart chambers, which was especially challenging for LGE imaging
of the right atrium. To address the blood labelling effect, a time delay between the
navigator and the imaging sequence has been proposed. However, this solution prohibits
motion correction due to the prolonged delay between navigator and imaging sequence
[19]. Another approach employed an adaptive flip angle for the navigator restore pulse
and effectively reduced the blood labelling artifacts [20]. Another challenge of 3 D LGE are frequent arrythmias which can disturb the steady
state signal and consequently lead to severe ghosting artifacts. One solution is the
use of an adaptive flip angle of the imaging sequence which has been shown to improve
myocardial nulling and effectively reduced ghosting artifacts in the 3 D LGE images
[20]. Finally, the prolonged acquisition times required for high-resolution 3 D LGE imaging
are accompanied by a progressive decline in contrast agent concentration in the blood
pool. As a result, the ideal TI to null the normal myocardium gradually increases.
This limitation can be addressed by using a steadily increasing inversion time that
optimizes tissue nulling for improved contrast [21].
While 3 D LGE imaging has become the reference for atrial scar imaging, it has not
been widely adapted for LV scar imaging due to the unpredictable scan time related
to respiratory motion gating. To address the relatively long scan time and associated
washout of GBCA, single breath-hold 3 D LGE sequences have been proposed with and
without [22]
[23] undersampled reconstruction. While these approaches have provided promising results,
they have not replaced 2 D LGE MRI due to their lower spatial resolution and lower
scar to myocardium contrast. The main limitations of all 3 D implementations have
been the prolonged and unpredictable scan time and ghosting artifacts from unsuppressed
subcutaneous fat signal where shimming is less effective.
Fat/water separation methods
Fat/water separation methods
Although fat suppression in 2 D LGE MRI is not very common, it is crucial in free-breathing
3 D LGE to minimize ghosting artifacts from subcutaneous fat which can be pronounced
in patients with irregular and/or deep breathing. Initial free-breathing navigator-gated
3 D LGE implementations employed a frequency selective fat saturation pulse to suppress
epicardial, pericardial, and subcutaneous fat. Those sequences have been very effective
on 1.5 T MRI scanners. However, at a higher field strength, fat suppression can sometimes
be suboptimal due to larger B0 inhomogeneity. This can cause unwanted ghosting artifacts from subcutaneous fat,
which are more pronounced in LGE scans due to the relatively low overall signal to
noise ratio (SNR). An alternative method for dealing with fat-related artifacts is
the use of water-fat separation using the 2- or 3-point Dixon method, which allows
the effective separation of the water from the fat signal, and which is relatively
robust with respect to B0 inhomogeneity at a higher field strength ([Fig. 6]). The Dixon water-fat separation methods acquire both an in-phase and out-of-phase
image, where the water and fat signals are either in- or out-of-phase. Several reconstruction
methods have been proposed (e. g., IDEAL [24], graph cut [25]), which allow for separation of the water and fat signals and consequently generation
of a water-only and a fat-only image or volume. An added benefit of Dixon-based water-fat
separation is the increased SNR due to the acquisition of two or three echoes. Moreover,
Dixon LGE allows the detection of intramyocardial fatty infiltration, which is considered
a risk factor for future cardiac events. The fat image also has been used to estimate
respiratory motion, which can be beneficial, as the water signal is typically suppressed
by the initial inversion pulse of the 3 D LGE sequence.
Fig. 6 Water/fat-separated 3 D LGE images acquired used the 2-point Dixon method in a patient
with myocardial infarction. After acquiring the in- and out-of-phase images, water/fat
separation can be performed to obtain a water- and fat-only volume (coronal view shown).
The high-resolution 3 D acquisition enables reconstruction of every imaging plane
after acquisition (see also [Fig. 5]), with the short-axis, 2-chamber and 4-chamber views being shown here. Image navigation
was used to correct for respiratory motion. The blood-nulled PSIR LGE approach was
used to obtain dark-blood contrast.
Abb. 6 Wasser/Fett-getrennte 3D LGE-Bilder, die mit der 2-Punkt-Dixon-Methode bei einem
Patienten mit Myokardinfarkt aufgenommen wurden. Nach dem Erfassen der „In-“ und „Out-of-Phase“-Bilder
kann eine Wasser-Fett-Trennung durchgeführt werden, um ein reines Wasser- und Fettbild
zu erhalten (koronare Ansicht). Die hochauflösende 3D-Akquisition ermöglicht die Rekonstruktion
jeder Bildebene nach der Aufnahme (siehe auch Abb. 5), wobei hier die Kurzachsen-,
2-Kammer- und 4-Kammer-Ansicht gezeigt wird. Die Atembewegung wurde durch die Navigator-Technik
korrigiert. Die blut-genullte PSIR LGE-Methode wurde verwendet um einen „dark-blood“
Kontrast zu erhalten.
Motion correction methods
Motion correction methods
As mentioned previously, free-breathing 3 D LGE MRI requires respiratory motion correction.
Initial approaches employing respiratory navigators suffer from long and unpredictable
scan times. For these reasons, self-navigation [26] and image-navigation [27]
[28] approaches have been proposed to enable 100 % scan efficiency without the need for
data rejection. The self-navigation approach proposed by Rutz et al. demonstrated
improved myocardial border sharpness in comparison to the standard clinical 2 D LGE
sequence [26]. Image navigators in concert with non-rigid motion correction, a 2-point Dixon readout,
and undersampled low-rank patch-based multi-contrast reconstruction (HD-PROST) enabled
scar imaging with high spatial resolution (1.3 mm isotropic) in a scan time of approximately
8 minutes with good agreement with breath-hold 2 D LGE MRI [28]. Motion tracking was facilitated by using the out-of-phase image navigator, which
provided the best motion tracking fidelity in comparison to the in-phase, water-only,
and fat-only navigator images. The 3 D LGE imaging sequence was also evaluated in
patients with an atrial scar and showed similar image quality compared to he diaphragmatic
navigator (dNAV) technique, however, with a lower scan time compared to the dNAV (4.5 ± 1.2 min
vs. 10.9 ± 3.9 min, p < 0.0001) [29].
Dark-blood methods
Although LGE can adequately distinguish hyperenhanced scar regions from normal myocardium,
conventional (myocardium-nulled) LGE has its limitations. Due to the almost equally
bright signal of the adjacent blood pool, detection of these small areas of ischemic
scar is challenging. Even when successfully detecting such areas, identifying the
exact scar-blood border remains difficult. Already more than 15 years ago, advances
were made to increase scar-to-blood contrast using various additional magnetization
preparation mechanisms. Over the years, numerous new LGE approaches have emerged to
improve subendocardial scar conspicuity ([Fig. 7]) [108]. Although these methods achieve superior scar-to-blood contrast compared to conventional
LGE, the majority require additional magnetization preparation schemes, including
T2 preparation [30]
[31]
[32]
[33], magnetization transfer [34]
[35], spin-locking [35], and the utilization of multiple inversion pulses [36]
[37]. These additional schemes, however, require adjustments to the MRI system software
and/or configuration, optimizations for new sequence and patient-specific parameters,
and additional training for radiographers, radiologists, and cardiologists, thus hampering
their introduction into routine clinical practices. Alternatively, a dark-blood LGE
method that achieves increased scar-to-blood contrast without using additional magnetization
preparation (blood-nulled PSIR LGE) was proposed, making it readily and widely available
in every clinical routine setting [38]
[39]. While most novel dark-blood methods have been compared with conventional LGE, only
flow-independent dark-blood delayed enhancement (FIDDLE) and blood-nulled PSIR LGE
have been validated against histology [34]
[57]. Both methods demonstrated superior visualization and quantification of ischemic
scar patterns compared to the current in-vivo reference standard; conventional myocardium-nulled
LGE.
Fig. 7 Short-axis LGE images of three patients and one swine with (suspicion of) myocardial
infarction acquired using conventional bright-blood LGE (top row) and dark-blood LGE
(bottom row). Due to the almost equally high signal of the blood pool, conventional
LGE suffers from poor scar-to-blood contrast which hampers the detection and delineation
of enhanced regions (cyan arrowheads). Note that these indicated regions can be accurately
visualised using dark-blood LGE. The shown dark-blood LGE images were acquired using
the blood-nulled PSIR LGE approach [38].
Abb. 7 Kurzachsen-LGE-Bilder von drei Patienten und einem Schwein mit (Verdacht auf) Myokardinfarkt,
aufgenommen mit konventionellen „bright-blood“ LGE (obere Reihe) und „dark-blood“
LGE (untere Reihe). Aufgrund des fast gleich hohen Signals des Blutpools leidet das
konventionelle LGE unter einem schlechten Narben-zu-Blut-Kontrast, der die Erkennung
und Abgrenzung von kontrastmittelaufnehmenden Abschnitten (cyanfarbene Pfeilspitzen)
limitiert. Diese Abschnitte können mit dem „dark-blood“ LGE akkurat abgegrenzt werden.
Die gezeigten „dark-blood“ LGE-Bilder wurden mit dem blut-genullten PSIR-LGE-Ansatz
aufgenommen [38].
Scar quantification methods
Scar quantification methods
For most clinical indications, visual assessment of LGE images is sufficient and quantitative
analysis is primarily performed to measure LGE extent for research purposes. By today,
no method for LGE quantification is universally accepted. Several techniques have
been proposed including manual contouring, semi-automated approaches using signal
intensity thresholds between two to six times the standard deviation from the normal
myocardium, non-binary techniques, and the full width at half maximum (FWHM) approach
[40]
[41]. While manual contouring methods are time-consuming, semi-automated approaches require
the definition of regions of interest in the remote and/or enhanced myocardium. Notably,
different methods lead to somewhat varying results in regard to reproducibility and
histology as the reference standard [42]. Therefore, accurate and objective LGE quantification remains a research tool without
current clinical implementation.
Diagnostic value
The LGE technique has dramatically changed the role of CMR in the evaluation of cardiomyopathies
as it is able to differentiate between ischemic and non-ischemic cardiomyopathies
by identifying specific patterns of LGE [43]. Ischemic cardiomyopathies are characterized by an LGE pattern that involves the
subendocardium and can extend up to the epicardium to form a transmural scar (so-called
“wave front phenomenon” of ischemic cell death) in a territory of a coronary artery
([Fig. 8]). LGE imaging also provides insight into irreversible damage of the microvascular
circulation by visualizing a microvascular obstruction (MVO), also known as the no-reflow
phenomenon ([Fig. 9]). MVO can be identified as low or absent signal areas in LGE images and is typically
located within the central portions of the infarcted tissue.
Fig. 8 Short-axis LGE images of three patients with myocardial infarction showing various
degrees of scar transmurality. The scar regions are indicated using cyan arrowheads.
Abb. 8 Kurzachsen-LGE-Bilder von drei Patienten mit Myokardinfarkt, die verschiedene Grade
der Narben-Transmuralität zeigen. Die Narbenregionen sind mit cyanfarbenen Pfeilspitzen
gekennzeichnet.
Fig. 9 LGE images in a case of acute myocardial infarction (MI). The left two LGE images
show a short-axis and four-chamber view 3 days post-MI (symptom to recanalization
time > 24 hours). Extensive regions of microvascular obstruction (MVO) can be observed
as dark areas within the enhanced areas. The right LGE image shows a short-axis view
4 weeks post-MI. Left ventricular remodeling can be observed with shrinkage of the
infarct area, replacement fibrosis, and an increase of the end-diastolic left-ventricular
volume (cine imaging, not shown). The cyan arrowheads indicate the affected area.
Abb. 9 LGE-Bilder bei einem Patienten mit akutem Myokardinfarkt (MI). Die beiden linken
LGE-Bilder zeigen eine Kurzachsen- und Vierkammeransicht 3 Tage nach MI (Zeit von
Symptom bis Rekanalisation > 24 Stunden). Ausgedehnte Bereiche mikrovaskulärer Obstruktion
(MVO) können als dunkle Bereiche innerhalb der kontrastmittelaufnehmenden Abschnitte
abgegrenzt werden. Das rechte LGE-Bild zeigt eine Kurzachsenansicht 4 Wochen nach
MI. Das linksventrikuläres Remodeling führt zu einer Schrumpfung des Infarktbereichs,
Gewebe-Ersatz-Fibrosierung und einer Zunahme des enddiastolischen linksventrikulären
Volumens (Cine-Bildgebung, nicht gezeigt). Die cyanfarbenen Pfeilspitzen zeigen den
betroffenen Bereich an.
In contrast, non-ischemic cardiomyopathies are characterized by mid-wall, epicardial,
or global endocardial LGE. Around 30 % of patients with heart failure caused by dilated
cardiomyopathy (DCM) present with mid-wall enhancement predominantly in the interventricular
septum [44]
[45]. Beside patients with heart failure, LGE – in conjunction with other anatomic and
functional CMR parameters – is an incremental diagnostic tool for the assessment of
various acute and chronic cardiomyopathies. LGE is a cornerstone technique to identify
focal areas of myocardial injury – beside evidence for myocardial edema, the major
criteria for the detection of acute myocarditis in the recently updated Lake Louise
criteria [46]. Myocarditis typically affects the epicardium with variable extension to the midmyocardial
wall and sparing of the subendocardial portions of the myocardium. A dominance in
the lateral and inferolateral wall is frequently described [47]. In more chronic cardiomyopathies such as hypertrophic cardiomyopathy (HCM) and
amyloidosis, LGE can be used to identify foci of diffuse scarring. The pattern of
LGE in HCM is typically patchy and predominantly located in hypertrophic parts of
the myocardium and sometimes within a mid-wall location that involves the junction
zones between the left and right ventricle as a sign of myocardial disarray. Amyloidosis,
a systemic disease with infiltration of various organs including the heart, can be
diagnosed by CMR using specific patterns of functional and structural changes including
myocardial hypertrophy, thickened interatrial septum and bi-atria dilation [48]. In addition to those anatomical changes, LGE imaging often demonstrates a unique
appearance in amyloidosis, as wash-in and wash-out kinetics of gadolinium are dramatically
changed by accumulation of amyloid fibrils in the myocardial interstitium. This results
in the inability to properly null the myocardium using TI scouts, leading to a zebra-striped
appearance with global subendocardial enhancement of LV and RV, and atrial walls,
and a relative dark appearing blood pool, indicating high myocardial contrast uptake
and fast blood pool washout [49]. Around 50 % of patients with valvular heart disease, such as severe aortic stenosis,
present with evidence of LGE [50]. In aortic stenosis, LGE is a result of chronic pressure overload of the left ventricle
[51]. Both infarct and non-infarct-like patterns can be present in aortic stenosis. However,
due to the absence of randomized clinical trials using LGE in valvular heart disease,
the assessment of myocardial fibrosis by LGE is not included in the routine evaluation
of patients with severe aortic stenosis [50]. A pictorial overview of LGE patterns of non-ischemic cardiomyopathies is provided
in [Fig. 10].
Fig. 10 Case examples of various non-ischemic cardiomyopathies. The scar regions are indicated
by cyan arrowheads. ARVC: arrhythmogenic right-ventricular cardiomyopathy; DCM: dilated
cardiomyopathy; HCM: hypertrophic cardiomyopathy.
Abb. 10 Fallbeispiele verschiedener nicht-ischämischer Kardiomyopathien. Die Narbenregionen
sind durch cyanfarbene Pfeilspitzen gekennzeichnet. ARVC: arrhythmogene rechtsventrikuläre
Kardiomyopathie; DCM: dilatative Kardiomyopathie; HCM: hypertrophe Kardiomyopathie.
Irrespectively of the etiology of myocardial disease, several studies demonstrated
that distinct patterns of LGE correspond to focal fibrosis at histopathology/autopsy
in human [52]
[53]
[54]
[55]
[56] and animal studies [2]
[57] in a wide range of clinical scenarios.
Prognostic value
LGE is the current CMR standard to assess myocardial viability in ischemic cardiomyopathy.
LGE allows the detection of even small subendocardial infarcts due to its superior
spatial resolution, which may be missed by other imaging techniques with lower spatial
resolution such as single-photon emission computed tomography [58]
[59]
[60]. LGE is also a reliable tool for the assessment of myocardial viability and the
likelihood of recovery after revascularization [61]. Kim et al. could initially demonstrate an inverse relation between the transmurality
of LGE and the recovery of segmental contractile function after revascularization
[62]. A transmurality of < 50 % has been shown to predict a very high likelihood for
functional recovery, while functional improvement in segments with scar transmurality
of > 50 % was only 8 %. Various studies have confirmed these results in both acute
and chronic infarct settings [63]
[64].
The assessment of the extent of LGE in acute myocardial infarction has significant
prognostic value. A meta-analysis incorporating 10 randomized trials with 2632 patients
with acute myocardial infarction demonstrated that a 5 % increase in infarct size
correlates to a 20 % increase in rates for all-cause mortality and heart failure-related
hospitalization [65]. In addition, LGE extent was a strong predictor of major adverse cardiovascular
events (MACE), independent of LV function [61]. However, MVO was a stronger independent predictor of LV dysfunction and post-infarct
complications compared to LGE extent [66]
[67]. Infarct size and MVO outperformed clinical risk scores and LV ejection fraction
for the prediction of adverse events [67]
[68]. MVO most likely represents irreversible tissue destruction and clinically relevant
reperfusion injury, which are both parameters that have been linked to negative LV
remodeling [69]. Hence, besides quantifying the size of the infarct, the evaluation of vascular
integrity may have a relevant role for prognosis estimation in acute myocardial infarction.
In chronic ischemic heart disease, LGE is a stronger predictor of clinical outcome
than contractile reserve [70]
[71].
The presence of LGE in patients with non-ischemic myocardial injury and cardiomyopathies
is strongly associated with poor prognosis, including increased risk of all-cause
mortality, heart failure hospitalization, and sudden cardiac death [72].
In DCM, LGE is associated with abnormal function and serves as a potential substrate
for re-entrant ventricular arrhythmias [73]. Several studies have evaluated the prognostic implications of LGE in DCM patients
[44]
[74]
[75]
[76]. Recently, a large, multi-institutional study has followed 1672 patients with DCM
over a median of 2.3 years and found a 1.5-fold increased risk of LGE for all-cause
mortality, heart transplantation, or left ventricular assist device implantation [77]. In addition, LGE identifies patients who may not respond optimally to medical therapies,
and therefore might play a role as a gatekeeper for intensified medical treatments
or procedures such as ICD implantation [78]. However, there are currently no clinical guidelines supporting LGE as a relevant
criterion for clinical decision-making and therapy indication.
In myocarditis, LGE has been identified as the best indicator of all-cause mortality
in a series of 202 consecutive biopsy-proven viral myocarditis patients. Over a time
period of 4.7 years, the presence of LGE increased all-cause and cardiac mortality
by 8.4- and 12.8-fold, independent of clinical symptoms [79]. Furthermore, other larger prospective studies demonstrated that the location and
extent of LGE were independent predictors of adverse cardiac events [80]
[81]
[82].
Several studies have examined the relationship between the presence of LGE and adverse
outcomes in HCM. For example, the presence of LGE increased the risk for all-cause
and cardiac mortality 5.5- and 8-fold over 3 years in 243 HCM patients [83]. A recent meta-analysis including 2993 patients with a follow-up over three years
identified LGE as a powerful predictor of sudden cardiac death, all-cause mortality,
and cardiovascular mortality [84]. In addition, this and another meta-analysis could also link the extent of LGE to
a continuously increased risk of adverse events with higher amounts of scarring [84]
[85].
The prognosis of amyloidosis is clearly influenced by the presence and severity of
cardiac involvement [45]. The diagnostic performance of CMR and LGE has been well validated in cardiac amyloidosis.
However, its prognostic value is controversial. Maceira et al. as well as Ruberg et
al. could not demonstrate a relevant influence of the presence and extent of LGE on
the short-term prognosis of cardiac amyloid patients in smaller study cohorts [48]
[86]. Another study by Mekinian et al. could demonstrate a relationship between LGE and
markers of ventricular dysfunction but failed to demonstrate the incremental prognostic
value on mortality after the adjustment for confounding parameters [87]. Conversely, other studies with larger study populations identified LGE to be the
strongest predictor of all-cause mortality [88]
[89]
[90]. These conflicting findings may be explained by the diffuse nature of amyloid as
a disease, as LGE may miss diffuse changes of the myocardium.
Valvular heart disease such as aortic stenosis is an important and increasing public
health problem, in particular with increasing prevalence in aging western population
[91]. LGE imaging has been used for risk prediction in patients with aortic stenosis
[92]
[93]
[94]. In aortic stenosis, fibrosis on LGE imaging is a useful biomarker of LV remodeling,
and its presence is associated with a worse long-term outcome after aortic valve intervention
[94]. A systematic review and meta-analysis indicate that LGE is a powerful prognostic
marker, conveying > 2-fold higher risk of all-cause mortality in patients with aortic
stenosis, even after adjusting for baseline characteristics [50].
Recently, an “umbrella” review of meta-analyses in various heart diseases using 33
meta-analyses demonstrated the highly suggestive evidence for a prognostic role of
LGE in heart disease, which may be used to impact guidelines and therapeutic strategies
in the future [95].
Discussion
As early as 1984, GBCAs have been used to study differences in normal and infarcted
myocardium using MRI [96]. Ever since, the role of LGE imaging for the evaluation of myocardial viability
has been extensively investigated in both animal models [2]
[97]
[98] and patient studies [99]
[100]
[101]. As of today, LGE imaging is the main component of cardiac MRI protocols on both
1.5 T and 3 T systems [102] and is performed in clinical routine settings all over the world on a daily basis.
As the use of LGE imaging has matured and the range of applications has broadened,
the demand for increased resolution has grown and the need for improved contrast increased.
In fact, for the assessment of thin-walled structures with a thickness of just a few
millimeters, such as the atria and right ventricles, smaller voxels are required to
reduce partial volume effects, while improved scar-to-blood contrast is crucial to
detect small subendocardial lesions. Even though these demands have been partly addressed
by recent developments such as the dark-blood and high-resolution 3 D techniques described
earlier, these have not yet been widely implemented in clinical protocols outside
of academic centers.
Despite the promising diagnostic and prognostic implications, LGE imaging has a limited
capability to detect diffuse myocardial involvement, as myocardial enhancement must
be compared to normal reference tissue. Mapping techniques, such as T1 and T2 mapping, however, measure intrinsic tissue properties, and therefore do not rely
on reference tissue. Thus, mapping techniques are able to identify diffuse processes,
such as inflammation in myocarditis or fibrosis in dilated cardiomyopathy, and have
emerged as clinical tools for myocardial characterization [103]
[104]. In the specific context of LGE imaging, T1 mapping has become a promising alternative technique to provide myocardial characterization.
On the other hand, LGE will play an increasingly important future role in the detection
and assessment of atrial fibrosis, a common pathophysiological factor to the onset
and maintenance of atrial fibrillation. The combination of high isotropic resolution
and optimized dark-blood contrast is a promising approach to tackle the current problems
with the thin atrial wall and poor scar-to-blood contrast. The use of high-resolution
3 D LGE may be evaluated for various indications, ranging from predicting the risk
of cardiovascular events in patients with atrial fibrillation and predicting post-ablation
arrhythmia recurrence, to guiding atrial fibrillation ablations and mapping post-ablation
scar patterns to guide potential repeat procedures. High-resolution 3 D LGE, however,
may also prove beneficial for assessing the larger ventricles. Potential clinical
benefits that should be evaluated include the improved ability to detect microstructural
fibrosis which is important for patients with presumed idiopathic ventricular fibrillation
and the ability to perform accurate multiplanar reconstructions for the assessment
of papillary muscle scar. This may pave the way for improved diagnostic accuracy and
prognosis but also plays an important role in the work-up towards personalized treatment
of patients with (supra)ventricular tachyarrhythmias.
As the number of patients with cardiac devices is increasing, future research may
focus on LGE techniques that are robust for local disturbances in the main magnetic
field caused by such devices. The use of wideband LGE sequences may help minimize
image artifacts caused by these disturbances at both 1.5 T and 3 T [105]
[106]. However, since clinical evidence is limited, additional research is required to
enable widespread clinical adoption.
In addition, the recent introduction of compressed sensing by the major vendors enabled
widespread use of sparse imaging techniques, achieving acceleration factors that have
not previously been possible to attain with parallel imaging alone. Artificial intelligence-based
reconstruction techniques may be used to further push the use of compressed sensing
methods [107], achieving high-resolution 3 D LGE acquisitions of the entire heart in a matter
of minutes.
In conclusion, LGE imaging remains the current noninvasive reference standard for
the assessment of myocardial viability. Improvements in spatial resolution, scar-to-blood
contrast, and water/fat-separated imaging further strengthened this position. LGE
will play an increasingly important role in the diagnosis, prognosis, and treatment
planning of patients with (supra)ventricular cardiac arrhythmias.
