Imaging of Myocardial Fibrosis and Its
Imaging of Myocardial Fibrosis and Its
Imaging of Myocardial Fibrosis and Its
a CardiovascularCenter Aalst, OLV Clinic, Aalst, Belgium; b Department of Advanced Biomedical Sciences,
Abstract
Patients with severe aortic stenosis (AS) show progressive Introduction
fibrotic changes in the myocardium, which may impair car-
diac function and patient outcomes even after successful The appropriate timing of aortic valve replacement
aortic valve replacement. Detection of patients who need an (AVR) in asymptomatic patients with severe aortic steno-
early operation remains a diagnostic challenge as myocar- sis (AS) remains challenging [1, 2]. Several of these pa-
dial functional changes may be subtle. In recent years, speck- tients show progressive fibrosis of the left ventricular (LV)
le tracking echocardiography (STE) and cardiac magnetic myocardium, which may impair cardiac function and
resonance mapping have been shown to provide comple- clinical outcomes even after successful AVR [3–5]. These
mentary information for the assessment of left ventricular individuals may benefit from early AVR before the devel-
mechanics and identification of subtle damage by focal or opment of irreversible myocardial fibrosis. The identifica-
diffuse myocardial fibrosis, respectively. Little is known, tion of myocardial damage at an early stage remains chal-
however, about how focal and diffuse myocardial fibrosis oc- lenging. Indices provided by standard echocardiography
curring in severe AS are related to measurable functional show a low sensitivity as myocardial structural and func-
changes by echocardiography and to which extent both pa- tional changes may be subtle. Cardiac magnetic resonance
rameters have prognostic and diagnostic value. The aims of (CMR) and speckle tracking echocardiography (STE)
this review are to discuss the occurrence of focal and diffuse have been recently shown to provide complementary in-
Advantages Low cost, more availability, rapid measurement offline after Ability to image on any plane, full visualization of the
adequate image acquisition myocardium, valve inflow/outflow tracts
Non-Doppler, angle-independent, myocardial deformation Direct measurement of the valve area and
evaluated in 2-D and 3-D, good reproducibility characterization of the associated great vessel anatomy
Objective quantification of myocardial systolic dynamics Gold standard to quantify valve flow, cardiac volumes,
and mass
Recent data support GLS derived by STE as a sensitive CMR techniques such as LGE and T1 mapping are
marker to detect subclinical myocardial dysfunction in AS promising markers to detect focal and diffuse
patients myocardial fibrosis, respectively
Limitations Lower temporal resolution, need for good image quality High cost, limited availability
Tracking affected by out-of-plane cardiac motion Adverse reaction to gadolinium
Intervendor variability Relative complexity of acquisitions, time-consuming
image analysis
AS, aortic stenosis; CMR, cardiac magnetic resonance; DMF, diffuse myocardial fibrosis; ECV, extracellular volume; GLS, global
longitudinal strain; LGE, late gadolinium enhancement; STE, speckle tracking echocardiography.
increase in comparison with normal myocardium. Both [6, 15, 19]. Accordingly, the LV ejection fraction, i.e., the
native T1 relaxation time and ECV have been significant- class I guideline recommendation for AVR, cannot be
ly associated with DMF at myocardial histology [25–27]. used for early risk stratification in asymptomatic AS pa-
We recently reported the high accuracy of both native T1 tients. In contrast, STE-derived 2-D global longitudinal
relaxation time with a cut-off value ≥1,010 ms (Ss = 90%, strain (GLS) is a validated and sensitive parameter to
Sp = 73%, AUC = 0.82) and ECV with a cut-off value quantify LV longitudinal systolic function [8, 9]. Several
≥0.315 (Ss = 80%, Sp = 90%, AUC = 0.85) to identify ex- studies have demonstrated a reduced magnitude of GLS
tensive (>30%) DMF at histology [24]. Moreover, corre- in AS patients compared to controls despite a preserved
lations between both native T1 and ECV with prognostic LV ejection fraction [16–18, 31–33]. In asymptomatic AS,
markers such as NT-pro-BNP or troponin have been re- GLS at rest has been shown to be independently associ-
ported [28, 29]. CMR-T1 has therefore been proposed as ated with development of symptoms, an abnormal exer-
a promising technique to identify early structural chang- cise tolerance, a need for AVR, and mortality [34–37].
es in patients with AS. The advantages and limitations of Furthermore, a magnitude of the longitudinal strain of
CMR in AS assessment are shown in Table 1. LV basal segments below −13% has been found to be as-
sociated with a higher rate of cardiac events at follow-up
[32]. It has also been shown that a GLS below −18% pre-
Imaging of Early LV Dysfunction in AS dicts an abnormal exercise response with a sensitivity of
68% and a specificity of 77% [38]. In another study, the
LV ejection fraction by echocardiography is routinely assessment of GLS during exercise had a higher accuracy
used to assess LV systolic chamber function in patients than the LV ejection fraction to detect latent LV systolic
with AS. However, increasing evidence demonstrates that dysfunction [39]. Finally, even the decrease in circumfer-
irreversible myocardial damage might occur before ential strain may be a marker of advanced disease with
changes in the ejection fraction become apparent [8]. It is unfavorable course, particularly when it is associated with
noteworthy that AS-induced DMF starts at the subendo- a low-flow state in AS patients [40]. These findings sug-
cardial level, affecting mainly longitudinal LV function. gest that both regional and GLS have a greater and earlier
Since it is predominantly determined by radial function, diagnostic power than the LV ejection fraction in this
the LV ejection fraction can be normal for a long time clinical setting [41]. The advantages and limitations of the
even in the presence of extensive subendocardial fibrosis STE-derived GLS assessment are summarized in Table 1.
Weidemann et al. [14] 85 FMF LGE CMR, histology, The extent of histologically determined that cardiac fibrosis
GLS at baseline correlated closely with markers of LS function
(all p < 0.001) but not global LVEF
Milano et al. [3] 99 FMF Histology, MF was inversely related to LV fractional shortening
LVEF (r = −0.64, p < 0.001), LVEF (r = −0.53, p < 0.001), and
LV relative wall thickness (r = −0.70, p < 0.001)
Treibel et al. [15] 133 FMF, LGE CMR, ECV, High ECV was associated with worse LV remodeling, LVEF,
DMF histology and functional capacity
Dweck et al. [18] 143 MF LGE CMR, Midwall fibrosis has an incremental prognostic value to
LVEF LVEF and may provide a useful method of risk stratification
Chin et al. [19] 166 FMF, LGE CMR, ECV, Index ECV demonstrated a good correlation with DMF on
DMF histology myocardial biopsies; there was evidence of increasing
hypertrophy, myocardial injury, diastolic dysfunction, and
LS dysfunction consistent with progressive LV
decompensation (all p < 0.05)
Kockova et al. [24] 40 DMF CMR T1, ECV, Both native T1 relaxation time with a cutoff value ≥1,010 ms
histology, and ECV with a cutoff value ≥0.32 showed a high accuracy in
identifying severe (>30%) DMF
Native T1 relaxation time showed a significant correlation
with LV mass (p < 0.01)
Fabiani et al. [29] 36 MF Histology, MF is associated with alterations of regional and GLS
GLS Plasmatic miRNA-21 is directly related to MF and associated
with LV structural and functional impairment
Hoffmann et al. [44] 30 FMF LGE CMR, GLS There was a negative correlation between the amount of
MF determined by LGE CMR and peak systolic longitudinal
strain for the total LV (r = –0.538, p = 0.007)
Lee et al. [45] 80 DMF CMR T1, GLS Native T1 correlated significantly with GLS measured with
2-D STE (r = 0.598, p < 0.001)
Bull et al. [46] 109 DMF CMR T1, T1 values increased with greater LV mass indices and
histology correlated with the degree of biopsy-quantified fibrosis
(r = 0.36, p = 0.008)
LVEF, left ventricle ejection fraction; STE, speckle tracking echocardiography; GLS, global longitudinal strain; LS, longitudinal sys-
tolic; CMR, cardiac magnetic resonance; MF, nonspecific myocardial fibrosis; DMF, diffuse myocardial fibrosis; FMF, focal myocardial
fibrosis; LGE, late gadolinium enhancement; ECV, extracellular volume.
Relationship between Myocardial Fibrosis and LV myocardial histology [3, 14, 15, 18, 19, 24, 29, 44–46] (Ta-
Systolic Function ble 2). Former studies have investigated the relationship
between FMF and LV contractile function [14, 44, 45]. It
Different kinds of observations have shown that GLS is has been shown that both the presence and the extent of
a functional marker of myocardial fibrosis. First of all, GLS FMF are inversely related to echocardiographic parame-
was found to be related to biomarkers of myocardial fibro- ters such as relative wall thickness, LV fractional shorten-
sis such as those expressing calcification, collagen forma- ing, and ejection fraction and to STE-derived indices of LV
tion, or breakdown and inflammation [42, 43]. Several myocardial function [18, 28, 44]. A GLS ≤−11.6% showed
studies have also reported significant associations between a sensitivity of 65% and a specificity of 75% to predict sig-
LV systolic function and both FMF and DMF at CMR or nificant FMF (LGE >10%) [43]. The majority of studies
Fig. 1. Examples of resting 2-D GLS compared with the extent of DMF on myocardial histology. a Patient with a
preserved magnitude of 2-D GLS (–21.1%) and a negligible extent of DMF (7.4%). b Patient with a reduced mag-
nitude of 2-D GLS (–14.9%) and extensive DMF (31.2%). DMF, diffuse myocardial fibrosis; GLS, global longitu-
dinal strain. The images are shown with permission from the research work group of the Cardiovascular Center
Aalst (Belgium) [23, 24].
dealing with this issue have focused on DMF [15, 19, 24, exercise and native T1 relaxation time (Fig. 3) [23, 24]. Fi-
46, 47]. Of the conventional echocardiography-derived nally, the native T1 relaxation time showed a high accu-
parameters, DMF seems to show a significant, though racy in predicting the limited LV contractile reserve [23,
weak, correlation only with LV mass and the LV mass in- 24]. All together these results strongly support the concept
dex [23, 24]. In contrast to FMF, none of the other conven- that GLS could be considered as an accurate functional
tional parameters including LV ejection fraction or aortic marker of DMF in AS.
valve area had a significant association with the degree of
DMF [39]. This emphasizes the need to use a highly sensi-
tive technique to assess DMF. Recent investigations have Limitations
reported a significant relationship among DMF at histol-
ogy, the CMR-T1-derived native T1 relaxation time or Although both CMR-T1 and STE seem to have great
ECV, and STE-derived deformation indices [15, 18, 45]. In clinical potential in various cardiovascular diseases, these
our study, a GLS <–15% showed excellent accuracy to pre- techniques also have several limitations (Table 1). One of
dict extensive (>30%) DMF (Fig. 1, 2) [23, 24]. Moreover, the major shortcomings of both methods is the great inter-
we observed a significant correlation between GLS during scanner or intervendor variability of normal values. This
–5 80
Sensitivity, %
2-D GLS, %
–10 + + 60
+ +
+ ++
–15 + + 40
+ +++ Sensitivity 100%
+ +
+ + ++ Specificity 88%
–20 r = 0.7072 20
+ ++ AUC 0.96
p = 0.0002 Cutoff <–15%*
+
–25 0
0 20 40 60 80 0 20 40 60 80 100
a Myocardial collagen, % b 100% – specificity, %
Fig. 2. a Correlation between 2-D GLS and the percentage of myocardial collagen on myocardial histology. b Ac-
curacy of resting 2-D GLS to identify extensive (>30%) DMF on myocardial histology. DMF, diffuse myocardial
fibrosis; GLS, global longitudinal strain. The images are shown w permission from the research work group of
the Cardiovascular Center Aalst [23, 24].
100
50
80
Δ 2-D GLS, %
Sensitivity, %
60
0
40
Sensitivity 100%
Specificity 87%
20 AUC 0.93
–50 T1 cutoff >1,259 ms
0
1,100 1,200 1,259 1,300 1,400 0 20 40 60 80 100
a Native T1 relaxation time, ms b 100% – specificity, %
Fig. 3. a Correlation between exercise-induced Δ 2-D GLS and native T1 relaxation time on a 3-T scan. b Accu-
racy of native T1 relaxation time on a 3-T scan to predict a reduced LV contractile reserve. DMF, diffuse myo-
cardial fibrosis; GLS, global longitudinal strain. The images are shown with permission from research work group
of the Cardiovascular Center Aalst [23, 24].
disadvantage requires definition of normal values for each and expertise, the associated high costs, and the need to
individual scanner or echo device when assessing healthy administer a contrast agent. In contrast, echocardiography
subjects. This procedure should be repeated after each ma- is more widely available, faster, and cheaper than CMR.
jor update of equipment or hardware. Other limitations GLS, a relatively operator-independent parameter, has a
need also mentioned. First of all, CMR-derived assessment higher reproducibility compared to LV ejection fraction
of FMF using LGE has a wide interobserver variability, de- and other echocardiographic parameters of LV systolic
pends on the technical setting of the scanner, and does not function [6]. However, due to the difference among differ-
allow detection of DMF [47]. The CMR-T1-derived T1 re- ent vendors, the same software should be used in individ-
laxation time and ECV are dependent on a specific CMR- ual patients over time [48–50]. The load dependency of the
T1 sequence, magnetic field strength, and homogeneity. In STE-derived indices may represent another challenge for
addition, there is a significant overlap between T1 map- routine clinical use in AS, as they are largely influenced by
ping values in healthy and diseased myocardia, making the both preload and afterload changes [27, 38, 39, 51]. Ac-
interpretation challenging [15, 30, 39, 40]. Other limita- cording to recent published studies in animal models, STE-
tions of CMR include the limited availability of equipment derived indices correlate strongly with pressure-volume
References
1 Baumgartner H, Falk V, Bax JJ, De Bonis M, 4 Treibel TA, Kozor R, Schofield R, Benedetti 7 Podlesnikar T, Delgado V, Bax JJ. Cardiovas-
Hamm C, Holm PJ, et al.; ESC Scientific Doc- G, Fontana M, Bhuva AN, et al. Reverse Myo- cular magnetic resonance imaging to assess
ument Group. 2017 ESC/EACTS Guidelines cardial Remodeling Following Valve Replace- myocardial fibrosis in valvular heart disease.
for the management of valvular heart disease. ment in Patients With Aortic Stenosis. J Am Int J Cardiovasc Imaging. 2018 Jan;34(1):97–
Eur Heart J. 2017 Sep;38(36):2739–91. Coll Cardiol. 2018 Feb;71(8):860–71. 112.
2 Nishimura RA, Otto CM, Bonow RO, Cara- 5 Kwiecinski J, Chin CW, Everett RJ, White AC, 8 Kempny A, Diller GP, Kaleschke G, Orwat S,
bello BA, Erwin JP 3rd, Fleisher LA, et al. 2017 Semple S, Yeung E, et al. Adverse prognosis Funke A, Radke R, et al. Longitudinal left ven-
AHA/ACC Focused Update of the 2014 associated with asymmetric myocardial thick- tricular 2D strain is superior to ejection frac-
AHA/ACC Guideline for the Management of ening in aortic stenosis. Eur Heart J Cardio- tion in predicting myocardial recovery and
Patients With Valvular Heart Disease: A Re- vasc Imaging. 2018 Mar;19(3):347–56. symptomatic improvement after aortic valve
port of the American College of Cardiology/ 6 Kvernby S, Rönnerfalk M, Warntjes M, Carl- implantation. Int J Cardiol. 2013 Sep; 167(5):
American Heart Association Task Force on häll CJ, Nylander E, Engvall J, et al. Longitu- 2239–43.
Clinical Practice Guidelines. J Am Coll Car- dinal changes in myocardial T1 and T2 relax- 9 Cengiz B, Şahin ŞT, Yurdakul S, Kahraman S,
diol. 2017 Jul;70(2):252–89. ation times related to diffuse myocardial fi- Bozkurt A, Aytekin S. Subclinical left ventric-
3 Milano AD, Faggian G, Dodonov M, Golia G, brosis in aortic stenosis; before and after ular systolic dysfunction in patients with se-
Tomezzoli A, Bortolotti U, et al. Prognostic aortic valve replacement. J Magn Reson Imag- vere aortic stenosis: A speckle-tracking echo-
value of myocardial fibrosis in patients with ing. 2018 Feb;48(3):799–807. cardiography study. Turk Kardiyol Dern Ars.
severe aortic valve stenosis. J Thorac Cardio- 2018 Jan;46(1):18–24.
vasc Surg. 2012 Oct;144(4):830–7.