Myocardial fibrosis is one of the hallmarks of a failing heart (1). In non-ischemic etiologies, the accumulation of collagen in the interstitium is progressive: from reactive interstitial (microscopic) fibrosis in the early stages to replacement (macroscopic) fibrosis in more advanced stages of heart failure (2, 3). These changes in collagen composition result in increased myocardial stiffness, left ventricular (LV) diastolic and systolic dysfunction (4, 5). Therefore, there is increasing clinical interest in assessing the myocardial interstitium as a marker of decompensation and a potential therapeutic target in heart failure (6–8).

Regardless of etiology, impaired LV ejection fraction (LVEF) is commonly used as a measure of global LV systolic dysfunction to guide heart failure management. However, LVEF is a late marker of cardiac decompensation. Global longitudinal strain (GLS) using speckle tracking echocardiography is more sensitive in detecting early LV dysfunction and correlates with myocardial fibrosis on histology (9). Despite the potential utility and accuracy in detecting fibrosis, LV GLS thresholds associated with myocardial fibrosis as markers of cardiac decompensation and prognosis have not been well-investigated (9).

In Stage B heart failure (structural heart abnormalities without signs and symptoms of heart failure), adverse myocardial remodeling in left ventricular hypertrophy (LVH) conditions such as hypertensive heart disease and aortic stenosis (AS) is accompanied by myocardial fibrosis that is mediated in part by myocardial injury/ischemia, neurohormones and the renin-angiotensin-aldosterone system (10, 11). In these conditions, we have previously demonstrated that markers of myocardial fibrosis and advanced left ventricular hypertrophy (LVH) predicted worse outcomes (12–17).

Independent of underlying etiology, LV GLS reflects the functional consequences of myocardial fibrosis. In this study, we aim to examine the mechanistic association between LV GLS and the myocardium; and establish optimal LV GLS thresholds associated with myocardial fibrosis on cardiovascular magnetic resonance (CMR) in a cohort of hypertensive patients who have undergone both CMR and echocardiography. Subsequently, we will validate the prognostic value of these LV GLS thresholds in a separate cohort of asymptomatic/minimally symptomatic (NYHA functional class I–II) patients with moderate to severe AS and preserved LVEF.

The derivation cohort consisted of 151 patients with hypertension (57 ± 10 years old; 58% males). Patients with replacement myocardial fibrosis (n = 55) had a higher systolic blood pressure compared to those without (138 ± 15 vs. 130 ± 13 mmHg, respectively; P = 0.001). These patients had higher indexed LV mass (69 ± 22 vs. 52 ± 10 g/m², respectively; P < 0.001), and higher measures of reactive interstitial myocardial fibrosis (ECV and native T1; P < 0.001 for all; Table 1).

LV GLS demonstrated a moderately strong correlation with indexed LV mass (r = 0.62; P < 0.001) but a weaker correlation with native T1 (r = 0.34; P < 0.001) and ECV (r = 0.15; P = 0.071). Despite similar LVEF, patients with replacement myocardial fibrosis demonstrated impaired LV GLS on echocardiography (−15.4 ± 3.7 vs. −18.5 ± 2.6%, respectively; P < 0.001, Figure 1). Indexed LV mass (standardized coefficient = 0.48; P < 0.001) and replacement myocardial fibrosis (standardized coefficient = 0.18; P = 0.013) were independently associated with LV GLS, after adjusting for age, sex and systolic blood pressure.

The current study demonstrates that LV mass and replacement myocardial fibrosis on CMR were independently associated with echocardiographic LV GLS; while CMR measures of reactive interstitial fibrosis had weaker correlations with LV GLS. LV GLS thresholds of replacement myocardial fibrosis risk-stratified patients with AS. Despite absent/minimal symptoms and preserved LVEF, patients with moderate or severe AS deemed at high risk (LV GLS > −15.0%) experienced increased cardiovascular events compared to those at intermediate- (LV GLS −15.0 to −21.0%) and low-risk (LV GLS < −21.0%) over medium term follow-up. LV GLS was independently associated with adverse cardiovascular outcomes after adjusting for the effects of potential confounders (Graphical Abstract).

In the pathophysiology of LVH, increased wall thickness and decreased chamber size initially normalize elevated wall stress. Collagen accumulates in a diffuse pattern within the interstitium and progresses to replacement fibrosis that corresponds to regions of cellular necrosis/apoptosis (3, 24). Therefore, changes affecting the interstitium and intrinsic myocardial contractility can influence myocardial deformation (25). In our study, LV GLS is independently associated with LV mass and to a lesser extent, replacement fibrosis. This observation can partly be explained by the technique of LV GLS by speckle tracking. The echocardiographic technique of assessing GLS by speckle tracking relies on tracking the displacement of speckles generated by the interaction between ultrasound waves and myocardial fibers (26). A recent study in young pigs demonstrated that LVH remodeling (as opposed to myocardial fibrosis) contributed more to reduced LV GLS after 19 weeks of aortic banding (27). This, along with our current study, would suggest LV GLS is mediated to a greater extent by increased LV mass; and interstitial abnormalities (myocardial fibrosis) affect LV GLS as a function of reduced intrinsic contractility only when LVH is more advanced. Of relevance, the myocardial architecture is a complex array of fibers organized in layers: subendocardial layer oriented longitudinally (more susceptible to wall stress and reduced perfusion) and subepicardial layer oriented more obliquely (28). Recent evidence demonstrated subendocardial GLS as a marker of disease severity, symptoms and prognosis (29, 30). The association between specific layer GLS, myocardial fibrosis and prognosis is of topical interest and warrants further investigation.

There is increasing recognition that AS is a condition that not only affects the aortic valve but also the myocardium (31, 32). The magnitude of LVH in response to AS is heterogeneous and this may partially account for the weak association between AS severity and LV mass (33, 34). Similarly, the association between LV GLS and AS severity has not been consistently demonstrated (35–37). In our study, we did not observe a strong correlation between LV GLS and measures of AS severity. An integrated multimodality approach of assessing the valve and myocardium is therefore necessary in the optimal management of patients with AS, particularly before the onset of symptoms and development of heart failure (38).

Conventional and novel CMR techniques have increased our ability to characterize the myocardium (2). The limited availability and high costs of CMR may make routine surveillance impractical and not cost-effective. Consequently, markers of myocardial fibrosis, particularly those with prognostic value are potentially attractive. In our previous studies, we have developed a clinical risk score (consisting of age, sex, AS severity, high-sensitive cardiac troponin I concentration and electrocardiographic strain pattern) based on its mechanistic association with replacement myocardial fibrosis on CMR (12–14). In that study, asymptomatic patients with moderate to severe AS and a high-risk score for myocardial fibrosis have the worst outcomes compared to patients at intermediate- and low-risk (14). In the current study, we have validated the prognostic potential of echocardiographic LV GLS, another marker associated with replacement myocardial fibrosis. A highly specific LV GLS value for replacement fibrosis (GLS > −15.0%) predicts worse cardiovascular outcomes compared to a threshold that is highly sensitive for replacement fibrosis (GLS < −21.0%). These GLS values are also consistent with data-driven thresholds from studies in moderate disease and asymptomatic severe AS (39, 40). Our studies highlight the novel approach of risk-stratifying patients based on the association with a pathophysiologically relevant substrate (myocardial fibrosis) of cardiac failure.

In our study, 14 patients experienced myocardial infarction as the primary outcome. Unlike heart failure, the association between myocardial fibrosis and myocardial infarction may be less obvious. Coronary flow reserve (CFR) is impaired in patients with aortic stenosis, even in the absence of epicardial coronary artery disease (41). Although the exact mechanisms are unknown, increased myocardial wall stress and left ventricular hypertrophy are associated with impaired CFR (42). Impaired CFR results in myocardial ischemia that is manifested clinically as angina and myocardial infarction. In a recent study, we have demonstrated elevated high sensitivity cardiac troponin I concentrations in patients with aortic stenosis and replacement myocardial fibrosis on CMR (12). These findings support a mechanistic association between myocardial fibrosis and ischemia.

Echocardiographic LV GLS has an independent association with replacement myocardial fibrosis and LV mass on CMR. Asymptomatic patients with significant AS and LV GLS thresholds specific for myocardial fibrosis (GLS > −15.0%) are at high-risk of adverse cardiovascular outcomes. Conversely, those at low-risk (GLS < −21.0%) have very favorable prognosis. LV GLS thresholds of replacement fibrosis is a novel approach to risk-stratify patients with moderate to severe AS and preserved LVEF who may benefit from early AVR.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Singhealth CIRB and National Healthcare Group DSRB. The patients/participants provided their written informed consent to participate in this study.

T-TL, WH, L-HL, and CC conceived the study idea. T-TL and WH contributed in data analysis, manuscript drafting, and preparation. BA, JAB, and D-FT recruited the patients, collected study/imaging data, and curated the data in the Singapore derivation cohort. SE, HT, ET, JY, QY, EL, PY, SC, PG, LL, HO, AR, and L-HL recruited aortic stenosis patients in the multi-center study, collected and curated the study data in their respective sites in the Singapore validation cohort. WS, YO, LG, and JL assisted in the data/imaging analysis of patients in the Singapore validation cohort. GS, VD, and JJB collected, analyzed, and curated the clinical and imaging data in the Leiden validation cohort. L-HL and CC contributed in supervising the writing and interpretation of the data. All authors have made critical feedback and contributed to the final manuscript.

This study was funded by the National Medical Research Council, Singapore (NMRC/CG/NUHCS/2010, NMRC/CG12Aug14, NMRC/CGAug16/C006, MOH-CSAINV17nov-0002), and the Charles Toh Cardiovascular Fellowship Fund.

The department of Cardiology at Leiden University receives unrestricted research grants from Abbott Vascular, Bayer, Biotronik, BIoventrix, Boston Scientific, Edwards Lifesciences, GE Healthcare, Ionis and Medtronic. VD received speaker fees from Abbott Vascular, Edwards Lifesciences, GE Healthcare, MSD, Medtronic and Novartis. JJB received speaker fees from Abbott Vascular. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.