Late Gadolinium Enhancement Predicts Improvement in Global Longitudinal Strain after Aortic Valve Replacement in Aortic Stenosis

Myocardial fibrosis, as detected by late gadolinium enhancement (LGE) magnetic resonance imaging (MRI), is related to mortality after aortic valve replacement (AVR). This study aimed to determine whether LGEMRI predicts improvement in global longitudinal strain (GLS) after AVR in patients with severe aortic stenosis (AS). Twenty-nine patients with severe AS who were scheduled to undergo AVR were enrolled. Two-dimensional echocardiography and contrast-enhanced MRI were performed before AVR. GLS and LGEcore (g: > 5 SD of normal area), LGEgray (g: 2–5 SD), and LGEcore+gray (g) were measured. One year after AVR, GLS were examined by echocardiography to assess improvement in LV function. Preoperatively, GLS correlated with LGEcore (g) (r2 = 0.14, p < 0.05), LGEgray (g) (r2 = 0.32, p < 0.01) and LGEcore+gray (g) (r2 = 0.36, p < 0.01). LGEcore was significantly lower in patients with improved GLS after AVR (GLS1year ≥ −19.9%) compared to those with no improvement (1.34 g vs. 4.70 g, p < 0.01). LGE predicts improvement in LV systolic function after AVR.


Results
Baseline characteristics. Table 1 summarized preoperative baseline characteristics of the 29 patients (age, 73 years; 52% male) included in this study. Nine patients presented with symptoms of heart failure, and 16 patients presented with symptoms of AS. In this cohort, patients had several atherosclerotic risk factors (hypertension, diabetes mellitus, hyperlipidemia, and/or current smoking). Table 2 shows echocardiographic parameters at baseline. All patients had high-gradient severe AS. LVEF was well-preserved at 65.7%, while GLS was reduced at −16.5%. LV hypertrophy was observed, but no severe diastolic dysfunction with increased LA pressure was noted.

Follow-up echocardiography after AVR.
There was no all-cause death or hospitalization due to heart failure at one year after AVR. We examined echocardiography in 26 patients; reasons for not performing follow-up echocardiography included patient refusal, cost of echocardiography, and other socioeconomical reasons. The results of comparisons of echocardiographic parameters before and after AVR are summarized in Table 3. After AVR, aortic valve function was significantly improved in terms of peak velocity (4.73 to 2.55 m/s), mean pressure gradient (50.5 to 14.1 mmHg), aortic valve area (0.65 to 1.47 cm 2 ), and valvulo-arterial impedance (Zva) ( www.nature.com/scientificreports www.nature.com/scientificreports/ We divided the 26 patients who underwent follow-up echocardiography according to median postoperative GLS: the improvement group (≥−19.9%; n = 14) and the non-improvement group (<−19.9%; n = 12). The comparisons of patient characteristics, echocardiographic parameters, and MRI parameters between the two groups are shown in Table 4.
No significant differences were observed in age, implanted valve size, and blood pressure between the improvement and non-improvement groups. Preoperative echocardiographic parameters did not differ between the two groups. Postoperatively, however, significant improvements were observed in LV hypertrophy (interventricular septal thickness [IVS] and posterior wall thickness [PW]) and LV diastolic function (left ventricular mass index [LVMI] and e') in the improvement group compared to the non-improvement group.
LGEcore and LGEcore+gray were lower in the improvement group compared to the non-improvement group.
LGEgray did not differ between the two groups.
In the ROC analysis, the area under the curve was 0.81 for predicting postoperative GLS improvement (≥−19.9%) by LGEcore (g), with a cut-off value of 2.86 g (sensitivity, 78.6%; specificity, 83.3%) (Fig. 3A). Figure 3B shows changes in GLS before and after AVR for each patient. Patients with low LGEcore (<2.86 g) showed improved GLS after AVR compared to those with high LGEcore (≥2.86 g).

Discussion
In this study, we investigated whether preoperative LGEs could predict improvement in GLS after AVR in patients with preserved LVEF and reduced GLS. The major findings are as follows: 1) Preoperative examinations revealed significant correlations among GLS, LGEs, and FI; 2) One year after AVR, GLS was improved in a manner dependent on preoperative LGEcore; and 3) LGEcore can predict postoperative improvement in GLS with a

Relationships among GLS, LGEs, and FI. In the present study, GLS correlated with
LGEcore+gray. On the other hand, the correlation with LGEcore was weak.
LGEgray didn't correlate with FI. Because interstitial diffuse fibrosis, interstitial edema and myocardial hypertrophy weren't detected as FI, LGEgray containing them didn't correlate with FI. GLS correlated well with LGEcore+gray including focal and diffuse fibrosis because GLS reflected the state of the whole myocardium.
Microscopic changes in LV are characterized by cardiomyocyte hypertrophy and extracellular matrix expansion in patients with AS. These conditions are caused by either focal replacement fibrosis (scar) or reactive and interstitial diffuse fibrosis 2,3,7-11 . According to previous studies, LGEcore and LGEgray reflect focal fibrosis and diffuse mild interstitial fibrosis, respectively 12 .
In a previous study, histological findings suggested improved GLS in patients with mild fibrosis 9 . Lee et al. also reported that native T1 values by cardiac MRI as an index of diffuse interstitial fibrosis correlated with GLS 13 . Reduction of GLS correlates with several factors such as myocardial fibrosis 14 , pressure overload, and obesity 15 , improved after AVR if fibrosis was mild in previous report 9 . Although the main cause of GLS impairment is still unknown, as well as the prospect for GLS improvement after AVR, it suggests that myocardium of patients with severe AS contains reversible and irreversible fibrosis.
GLS is reduced in symptomatic patients with severe AS, and a decrease in GLS is a predictor of all-cause mortality 6 . GLS is also a predictor of future major adverse cardiac events in asymptomatic patients with severe AS and preserved LVEF 16 . Thus, assessing GLS is clinically important in patients with potential systolic dysfunction and preserved LVEF. In the present cohort study, patients had preserved LVEF with a slight decrease in GLS, and myocardial specimens showed mild fibrosis compared to severity of fibrosis in the previous reports 13 . However, given that not all patients showed improved GLS after AVR, predictors of GLS improvement after AVR need to be investigated further.

Prediction of improvement in GLS after AVR.
In the present study, LGEcore, but not LGEgray or LGEcore+gray, was found to be a predictor of GLS improvement after AVR. While LGEgray (i.e., mild interstitial fibrosis) can be reversible, LGEcore (i.e., focal fibrosis) is unlikely to improve after AVR. Thus, our findings suggest that the degree of focal fibrosis before AVR is a determining factor for GLS improvement after AVR in patients with severe AS.
A previous cohort study reported that LGE did not improve significantly 9 months after AVR 9 . Since focal fibrosis doesn't improve, the decrease in diffuse fibrosis contributes to the improvement of postoperative contractility. A recent prospective observational cohort study reported that focal fibrosis (scars) as detected by LGE does not resolve, while diffuse fibrosis and myocardial hypertrophy as assessed by extracellular volume (ECV) show significant regression after AVR in patients with symptomatic severe AS 7 . However, it remains unclear as to which type of LV myocardial fibrosis (i.e., focal or diffuse) plays an important role in persistent systolic dysfunction after AVR.
To the best of our knowledge, this is the first study to evaluate whether LGE as an index of focal fibrosis and/or diffuse fibrosis could predict improvement in GLS one year after AVR. Preoperative GLS strongly correlated with LGEgray, but weakly correlated with LGEcore. On the other hand, LGEcore was found to be a predictor of GLS improvement after AVR. The use of different thresholds, i.e., > 5 standard deviation (SD) for LGEcore and 2-5 SD www.nature.com/scientificreports www.nature.com/scientificreports/ for LGEgray, allowed us to detect potential systolic dysfunction with preserved LVEF (LGEgray), and to predict improvement in GLS (LGEcore) after AVR.
Clinical implication. Recent therapeutic strategies for asymptomatic severe AS include AVR, which is recommended only when LVEF is less than 50% 17 . However, severe AS patients with preserved LVEF already has LV myocardial fibrosis 18 . In patients with extensive focal fibrosis, myocardial damage persists even if LV afterload is decreased by AVR. Thus, myocardial fibrosis needs to be evaluated noninvasively in order to predict prognosis after AVR in a clinical setting. Since focal fibrosis as detected by LGEcore (<2.86 g) is an independent predictor of GLS improvement after AVR, surgical therapy should be considered before patients develop irreversible LV dysfunction.

Study limitations.
This study has several limitations. First, this study was conducted at a single center. Second, we excluded patients with chronic kidney disease (CKD) because of a contraindication to contrast-enhanced MRI. Thus, the results of the present study may not apply to patients with CKD, which is a common disorder in elderly patients. Other methods to assess LV myocardial fibrosis, e.g., ECV by MRI 19,20 , should be considered. Third, we assessed GLS by 2D echocardiography, not 3D echocardiography. A significant correlation has been reported between 2D GLS and 3D GLS in patients with AS, and 3D GLS as well as 2D GLS are reportedly predictors of major adverse cardiac events 13 . Forth, the myocardial specimens were endomyocardial and not transmural which raised issue of representability.
In conclusion, this prospective observational study demonstrated that LGEcore predicts improvement in GLS after AVR in patients with severe AS and preserved LVEF.

Study design and patient recruitment. This prospective observational study was conducted in 29 patients
with severe AS who underwent AVR from January 2014 to July 2017. Severe AS was defined as an aortic valve area < 1.0 cm 2 , peak aortic valve velocity > 4.0 m/s, and mean pressure gradient > 40 mmHg 21 . Exclusion criteria were patients with concomitant severe aortic regurgitation, moderate to severe mitral regurgitation, and a previous history of ischemic heart disease, atrial fibrillation, left bundle branch block, or CKD (eGFR < 30 ml/min/1.73 m 2 is a contraindication to gadolinium enhanced-MRI) 22 .
All patients underwent echocardiography and MRI prior to AVR. Myocardial biopsy specimens were collected intraoperatively from 13 patients who obtained informed consent. We evaluated relationships among the parameters of echocardiography and MRI and myocardial specimens. Echocardiography was also performed one year after AVR to assess the correlation between preoperative LGE and postoperative GLS improvement.
This study was approved by the institutional review board of Fukushima Medical University and was conducted in compliance with the principles of the Declaration of Helsinki. All patients provided written informed consent.
Echocardiography. We performed transthoracic echocardiography to assess aortic valve function and LV systolic and diastolic function using the Acuson SC2000 TM system (SIEMENS: Mountain View, CA, USA) with a 4-MHz transducer 23,24 . Echocardiographic parameters included LV wall thickness and dimension, LV volume  www.nature.com/scientificreports www.nature.com/scientificreports/ and LVEF, left atrium volume index (LAVI), peak early filling (E velocity) and late filling (A velocity) using pulsed-wave Doppler images, peak early diastolic velocity (mean E´of lateral and septal wall) using tissue Doppler images, E/e' and AS indices (aortic valve area, peak velocity, mean pressure gradient [MPG], and Zva). Zva was defined as the ratio of estimated LV systolic pressure (the sum of systolic arterial pressure [SAP] and MPG) to stroke volume index (SVi): Zva = (SAP + MPG)/SVi 25 . LV mass index was calculated by the cube formula in the parasternal long-axis view 23 .
2D-GLS was examined by 2D speckle tracking echocardiography using the SC2000 workplace system VVI TM (SIEMENS: Mountain View, CA, USA). We assessed endocardial GLS as the average of GLSs in apical 2-, 3-, and 4-chamber views 26 . Cardiac MRI. Cardiac MRI was performed on a 1.5-T scanner (Vantage Titan TM : Canon Medical Systems, Otawara, Japan) according to the standard LGE protocol 27 . Ten minutes before image acquisition, 1.0 M gadobutrol (Gadovist TM : Bayer, Berlin, Germany), a gadolinium-based contrast agent, was administered systemically to patients with eGFR ≥ 30 ml/min/1.73 m 2 .
Cardiac MRI was analyzed using a post-processing workstation (Ziostation2 TM : Ziosoft, Tokyo, Japan). The contours of the LV endocardium and epicardium were traced semi-automatically in short-axis slices. The region of interest (ROI) was selected within the remote reference myocardium to set the SD 28 . All LGE measurements were performed by the author (T.F), and 95% intraclass correlation coefficient was 0.858. We evaluated LGEs as parameters of fibrosis, calculated on the workstation as areas with the above-threshold signal intensity compared to the remote reference myocardium in the ROI (LGEcore: > 5 SD; LGEgray: 2-5 SD; LGEcore+gray: LGEcore plus LGEgray) (Fig. 1) 12,29 . These indices were expressed as absolute amounts (g), percentage of myocardial mass (% of LV), and amounts corrected by body surface area (g/BSA).
Intraoperative biopsy. Intraoperative myocardial biopsy specimens were taken from 13 of the 29 patients.
An endomyocardial specimen roughly 8 mm 3 in volume was obtained from the basal muscular septum 2 cm below the outflow tract following aortic valve resection. All specimens were preserved in 20% formalin, embedded in paraffin, cut into 5-µm-thick sections, and stained with Elastica-Masson stain. The myocardial muscle and fibrous tissue was observed at a magnification of 100x 30 . The FI was defined as the ratio (in percentage) of fibrosis tissue to the total myocardial field using Image J 31 . For each patient, FI was quantified in five different fields representative of all myocardial samples.