Implications of myocardial strain in primary mitral regurgitation—a cardiovascular magnetic resonance study

Abstract

Aims

Chronic primary mitral regurgitation (MR) results in progressive left ventricular (LV) remodelling. Abnormal myocardial deformation (strain) can be present despite preserved ejection fraction (EF). Cardiovascular magnetic resonance (CMR) feature-tracking techniques allow assessment of global longitudinal strain (GLS) from routine cine images. The aim of this study is to evaluate the prognostic value of CMR feature tracking–derived GLS in patients with primary MR.

Methods and results

Consecutive patients undergoing CMR for chronic MR from January 2012 to June 2018 were enrolled. Patients with LVEF <50% were excluded. The composite primary outcome aiming to detect decompensation related to MR comprised (i) referral for mitral surgery owing to symptoms or LV systolic dysfunction or (ii) cardiovascular death. The secondary outcome was all-cause death. A total of 422 patients were followed for a median of 2.7 years, and the primary endpoint was met in 93 patients (34 patients reported symptoms at baseline). At multivariable analysis, GLS≥ −16.6% was associated with primary outcome [hazard ratio (HR) 1.90, P = 0.01]. In moderate MR cohort, patients with GLS≥ −16.6% had worse event-free survival, whereas there was no significant difference in mild or severe MR groups. GLS≥ −16.0% remained associated with all-cause death after adjusting for other covariates including the MR severity (HR 2.24, P = 0.02).

Conclusion

In patients with primary MR with preserved systolic function, GLS was associated with our composite outcomes and all-cause death. GLS may serve as a marker of cardiac dysfunction in the patients with primary MR with preserved systolic function allowing identification of patients likely to decompensate during observation.

Graphical Abstract

Open in new tabDownload slide

Introduction

Primary mitral regurgitation (MR) is caused by a structural abnormality of the mitral valve.¹ It is the most common valvular disease in the USA, with a prevalence increasing with advanced age and the second most common valvular disease requiring intervention, only superseded by aortic stenosis.^2,3 Chronic MR results in progressive remodelling of the left ventricle, manifesting initially as volume overload with preserved contractile function. Favourable loading conditions in MR increase left ventricular (LV) ejection fraction (EF) but do not affect the extent of myocardial shortening.⁴ Thus, LVEF should be considered a late marker of LV dysfunction and a ‘normal’ LVEF may mask subclinical LV dysfunction. The significance of this conundrum is acknowledged in the recently published ACC/AHA Guidelines on valvular heart disease which highlight the need for studies examining the role of earlier markers of myocardial dysfunction to define the optimal timing of intervention.⁴

Direct assessment of myocardial fibre deformation with global longitudinal strain (GLS) imaging has shown promise in providing diagnostic and prognostic information that is incremental to LVEF.⁵ GLS has been recognized as a sensitive marker of LV dysfunction, regardless of LV function.⁶ While GLS is also load-dependent, it evaluates the myocardial longitudinal deformation and is more sensitive in identifying myocardial dysfunction. The recent development of cardiovascular magnetic resonance (CMR) feature-tracking technology shows promises in allowing measurement of longitudinal strain using routine cine images in the clinical setting without specialized pulse sequences.⁷ The underlying principle is based on the recognition of ‘patterns of features’ or ‘irregularities’ in the image that are tracked and followed in successive frames.⁷ Importantly, this approach can be applied to routine cine CMR acquisitions, thus avoiding the need for dedicated pulse sequences, which are required for other specialized CMR strain techniques. GLS has been shown to provide prognostic value in several clinical conditions.^8–12 However, the data on prognostic value of GLS in primary MR are scarce and limited to GLS derived from echocardiographic technique.^13–15 Therefore, the aim of the present study is to evaluate the prognostic role of CMR feature tracking–derived GLS in primary MR population with preserved systolic function.

Methods

Study population

Consecutive patients undergoing contrast-enhanced CMR for assessment of chronic MR at the Houston Methodist Hospital (Houston, TX, USA) from January 2012 to June 2018 were enrolled into the DEBAKEY CMR Registry (NCT04281823), a prospective longitudinal registry designed to assess the utility of CMR in various disease states. We excluded patients with the following confounding causes of myocardial structural abnormality by clinical history as reported by the patients or available medical records: (i) coronary artery disease, (ii) cardiomyopathy, (iii) infiltrative/inflammatory heart disease (e.g. amyloidosis, sarcoidosis or myocarditis), (iv) prior cardiac surgery, or (v) congenital heart disease. Additionally, patients with the following findings on CMR were excluded: (i) LVEF < 50%; (ii) coexisting valvular disease that was greater than mild in severity (except for tricuspid regurgitation that is secondary to MR; all primary tricuspid regurgitations related to structural leaflet abnormalities such as endocarditis or carcinoid heart disease were excluded); or (iii) greater than mild concentric LV hypertrophy (LV wall thickness >1.3 cm and relative wall thickness >0.42). By their inherent nature, the aforementioned criteria also excluded patients with secondary MR, thus leaving only patients with primary MR in the study cohort. The patient enrolment process is summarized in Figure 1. The likely aetiology of primary MR was deduced using all available medical history and diagnostic testing, including CMR findings. Mitral valve prolapse was defined as >2-mm displacement of any mitral valve scallops into the left atrium, as indicated on a three-chamber view.^16,17 The study was approved by the institutional review board at the Houston Methodist Research Institute, and the patients gave written informed consent (IRB number: Pro00001568).

CMR imaging protocol

CMR images were acquired using either 1.5-T or 3.0-T clinical scanners (Siemens Avanto, Aera, or Verio; Siemens, Erlangen, Germany) with phased-array coil systems. A standard examination consisted of a cine CMR for anatomic and functional assessment in a short-axis stack and two-, three-, and four-chamber views using a steady-state free-precession sequence Additionally, to evaluate mechanism of MR, a high-resolution stack of small field-of-view cine CMR of the mitral valve was performed ‘en-face’ along with sequential three-chamber views to cover all mitral scallops. Flow across the aortic valve was ascertained using phase-contrast imaging. Standard late gadolinium enhancement (LGE) CMR identified replacement fibrosis using a magnitude and phase-sensitive segmented inversion recovery sequence ∼10 min after intravenous gadolinium contrast administration (gadopentetate dimeglumine or gadoterate meglumine, at a constant dose of 0.15 mmol/kg). Short- and long-axis LGE images (two-, three-, and four-chamber) were acquired in identical orientations to the cine images. An electrocardiographically gated modified Look-Locker inversion recovery sequence (MOLLI) with motion correction was performed at a mid-LV short-axis level in a matching position for pre- and post-contrast administration (∼15 min). The pre-contrast MOLLI acquisition was performed using a 5(3)3 sampling scheme, and the post-contrast acquisition used a 4(1)3(1)2 sampling scheme.

CMR image analysis

The LV and right ventricular (RV) volumes were measured by planimetry of the endocardial borders, on a stack of short-axis cine images covering both ventricles from base to apex. Papillary muscles and trabeculae were excluded from the LV cavity (i.e. included in LV mass). LV end-diastolic volume, LV end-systolic volume, RV end-diastolic volume, and RV end-systolic volume were calculated by summation of these images. Mitral regurgitant volume was calculated as the difference between LV stroke volume and aortic-forward stroke volume. Mitral regurgitant fraction was calculated as follows: mitral regurgitant volume/(LV stroke volume—aortic regurgitant volume).¹⁸ The presence and extent of replacement fibrosis, as demonstrated by LGE, was assessed in all LV segments according to the ACC/AHA 17-myocardial-segment model by a consensus of two readers who were blinded to clinical history and other imaging information. LGE was only considered present if it was identified on two contiguous or orthogonal slices and seen on both magnitude and phase-sensitive image reconstruction. A semi-quantitative method was used to calculate the burden of replacement fibrosis as a percentage of the left ventricle. The extent of LGE involvement in each 17-LV segment was scored as 0 for no LGE detected, 1 for 1–25%, 2 for 26–50%, 3 for 51–75%, and 4 for 76–100%. The total per cent LV scar volume was calculated from the summed score divided by 68 (maximum score) assuming that each sub-segment equally represented 1/68th of the total LV myocardial volume.¹⁹ For extracellular volume (ECV) analysis, quantitative parametric images of myocardial T1 were generated with manual contouring to define a region of interest in the mid-LV septum at the same location for pre- and post-contrast. We excluded areas where LGE or artefacts were present in the septum; for this situation, the region of interest was placed in either the anteroseptum or inferoseptum where no LGE or artefact was evident. ECV was calculated using the following validated formula: ECV = [(ΔR1 myocardium)/(ΔR1 blood pool) × (1 − Hct)], where Hct refers to the haematocrit recorded on a venous blood sample obtained at the time of CMR and ΔR1 = 1/T1 post-contrast − 1/T1 pre-contrast.

CMR strain analysis

Feature-tracking analysis for assessment of GLS was performed in a core lab after the database of clinical, imaging, and follow-up events was locked. The analysis was performed by a single physician (SR) who was blinded to all other study data. GLS was calculated by averaging the values obtained in the apical two- to three- and four-chamber views (18 segments) using the Qstrain package (Medis Medical Imaging Systems, Leiden, The Netherlands). LV endocardial borders were automatically traced at the end of the diastole; subsequently, the software tracked the endocardial layer throughout the cardiac cycle. When necessary, the operator manually adjusted the endocardial border (Figure 2). Measurement of GLS required ∼15 min of analysis time per case. GLS measurements showed good reproducibility. The Bland–Altman analysis of inter-observer repeatability showed a bias of 0.16% (95% limits of agreement: −2.08 to 2.40%), whereas the Bland–Altman analysis of intra-observer repeatability showed a bias of −0.07% (95% limits of agreement: −1.59 to 1.45%).⁶

Symptoms related to MR and study outcomes

Self-reported symptoms of dyspnoea on exertion or declining exercise tolerance were obtained from patient interviews at the time of CMR examination. To confirm the clinical significance, as these symptoms can be non-specific, we only considered that they were related to MR if the patients were subsequently referred for mitral surgery owing to their symptoms, or the patient experienced cardiovascular death within 90 days after CMR.

All patients regardless of symptoms related to MR at the time of CMR were followed longitudinally for clinical events. Our composite primary outcome aiming to detect decompensation related to MR included (i) referral for mitral surgery owing to symptomatic MR or progressive LV systolic dysfunction or (ii) cardiovascular death that occurred prior to mitral surgery. We excluded patients who underwent mitral valve surgery within 30 days after CMR to avoid potential bias of a predesignated CMR scan ‘en-route’ to surgery. The decisions for mitral valve surgical referral were made by the treating physicians who were blinded to GLS data. The secondary outcome was all-cause death. Follow-up data were collected from review of electronic medical record, structured telephone interviews with the patients, relatives, and/or their healthcare providers, and the National Death Index. For ascertainment of cardiovascular-related death, in patients whose cause of death was equivocal, the patient data were presented to an adjudication committee consisting of two board-certified cardiologists blinded to imaging data for further determination. The last round of follow-up data collection was performed in January 2020.

Statistical analysis

The patient characteristics were reported as frequencies and proportions for categorical variables and as median and inter-quartile range for continuous variables. The optimal cut-points of GLS in discriminating primary and secondary outcome were determined by the receiver-operating characteristic curve analysis with a Youden index.²⁰ Cox regression modelling was conducted to determine the characteristics associated with a higher risk of composite events and all-cause death. Variable selection for the models was conducted on the basis of established clinically important variables that known from prior studies and clinical judgement (a priori selection) and the Stata’s Lasso method with the cross-validation selection option.²¹ To avoid over fitting, some variables which were significant in the univariate analysis, but insignificant in multivariable modelling, were not selected in the final model if their exclusion did not affect the diagnostic performance of the final model. The discrimination power of predictive models was assessed using the C-statistic. The model’s good calibration was determined by a non-significant Hosmer–Lemeshow’s goodness-of-fit test. Event-free survival for the composite primary outcome was depicted by the Kaplan–Meier curves. Differences between subgroups of mitral regurgitant fraction and GLS were compared by the log-rank test. All analyses were performed on Stata version 17 (StataCorp LLC, College Station, TX, USA). A P-value of <0.05 was considered statistically significant.

Results

Baseline characteristics

Following the clinical history screening process, a total of 827 patients with chronic MR were enrolled. After excluding patients with imaging findings that could represent secondary MR or confounding aetiologies for myocardial structural abnormalities, 422 patients remained and comprised the primary MR study cohort (Figure 1). Data on the prevalence of replacement fibrosis and ECV of the patient in the current cohort were reported in our prior studies.^22,23 Imaging was performed with 1.5- and 3-T scanners in 147 patients (35%) and 275 patients (65%), respectively. Baseline characteristics and CMR data are summarized in Table 1 and Supplementary data online, Table S1. Of the total cohort, 47.9% were men with a median age of 61.8 years (inter-quartile 51.7, 70.7 years). Mitral valve prolapse was the most common aetiology of primary MR (228 patients, 54.0%; posterior mitral leaflet prolapse in 133 patients, 31.5%) followed by mitral calcification and/or thickening resulting in restriction or malcoaptation of leaflets (91 patients, 21.6%), rheumatic MR (15 patients, 3.6%), prior infective endocarditis (7 patients, 1.7%), suspected radiation related or connective tissue disease (3 patients, 0.7%), and indeterminate in 78 patients (18.5%). There was a spectrum of MR severity with a median regurgitant volume of 29.0 mL (inter-quartile 13.0 mL, 52.0 mL) and regurgitant fraction of 30% (inter-quartile 16.0%, 45%). LGE was found in 90 patients (21.3%) with LGE burden of 1.9% ± 0.9% of the left ventricle. The median ECV was 25.8% (inter-quartile 24.2%, 28.1%). The optimal threshold of GLS for identifying the primary composite outcome (referral for mitral surgery or cardiovascular death) and secondary outcome (all-cause death) was −16.6% (area under the curve 0.54, sensitivity 0.46, specificity 0.61) and −16.0% (area under the curve 0.60, sensitivity 0.50, specificity 0.70), respectively (see Supplementary data online, Figure  S1).

Symptoms and clinical events related to MR

Our criteria for symptomatic MR at the time of CMR procedure were met in 89 patients. Of them, 88 were referred for mitral surgery and 1 experienced cardiovascular death within 90 days of CMR. The average mitral regurgitant volume and fraction in the symptomatic MR group were 68.8 ± 27.0 mL and 51.1 ± 8.8%, respectively. The prevalence of symptomatic MR patients was highest in the worst GLS tertile (Figure 3).

During the follow-up period [a median follow-up time 2.7 years (inter-quartile 0.5, 4.6 years)], the primary outcome was met in 93 patients (22.0%). Of note, 34 out of 93 patients met our definition of symptomatic MR at the time of CMR. Among 93 patients with primary outcome, 89 patients were referred for mitral valve surgery and 4 suffered cardiovascular death. In the group of 89 patients referred for mitral surgery, 72 patients (80.9%) developed symptoms that led to referral, whereas 17 patients (19.1%) underwent MV surgery owing to progressive LV systolic dysfunction. Of all 89 patients referred for mitral surgery, 81 patients underwent mitral repairs, 5 underwent mitral replacement, and 3 were referred for procedure but declined. The median time to composite clinical events was 5.6 months (inter-quartile 2.0, 18.0 months). Male gender, the severity of MR, presence of mitral valve prolapse, ECV, and GLS were found to be associated with a composite of referral for mitral surgery and cardiovascular death on multivariable analysis (Table 2 and Supplementary data online, Table S2). Furthermore, adding ECV and GLS to traditional parameters (MR severity and LV volume) provided incremental value as demonstrated in Figure 4. Of note, the left atrial volume index and RVEF were not included in the final model to avoid overfitting and collinearity, and their exclusion did not impact diagnostic performance. In our cohort of patients (all of whom had preserved systolic function), the presence of LGE and LVEF was not significantly associated with the primary composite outcome. Kaplan–Meier curves for event-free survival from mitral surgery referral and cardiovascular death, stratified by severity of MR and GLS of −16.6% as a threshold, are demonstrated in Figure 5. Patients with mild MR demonstrated favourable event-free survival irrespective of GLS. In patients with moderate MR, those with GLS ≥ −16.6% demonstrated worse event-free survival, whereas in patients with severe MR, using GLS for stratification did not demonstrate statistically significant difference.

All-cause death was found in 32 patients during the follow-up period. Multivariable Cox regression models analysing factors associated with all-cause death in primary MR patients are given in Table 3 (univariate analysis in Supplementary data online, Table S3). Worsening GLS (using GLS ≥ −16.0% as a threshold) remained independently associated with all-cause death after adjusting for other covariates including the severity of MR.

Discussion

Our study is the first investigation to demonstrate the prognostic relevance of CMR-assessed longitudinal function in patients with chronic primary MR with preserved EF.

Our data show that CMR feature-tracking GLS is independently associated with composite adverse outcomes and all-cause mortality in patients with chronic primary MR with preserved systolic function. In addition, GLS provides incremental predictive value over traditional parameters.

Knowledge gaps in patients with valvular heart disease

Several knowledge gaps in patients with valvular heart disease exist. The optimal timing of a valvular intervention remains unknown in patients with asymptomatic severe valvular disease or symptomatic moderate valvular disease. In patients with MR, a meta-analysis suggests that a strategy of early surgery may improve survival and increase the likelihood of mitral valve repair compared with watchful waiting strategy.²⁴ However, it is unlikely that all patients will be candidates for earlier intervention; thus, identification of the subgroup most likely to benefit would be essential for optimal patient management. Newer approaches such as early markers of cardiac dysfunction using advanced imaging or circulating blood markers to identify high-risk patients are needed. Despite LVEF being the conventional echocardiographic parameter to estimate LV function, its impairment may be evident at a later stage in the disease process. Moreover, it does not take into consideration the direction of the blood flow and the intrinsic contractile properties of the myocardium. For these reasons, whereas LVEF provides a global measure of LV function, it fails to give an accurate estimate of the real LV inotropic status and ability to distinguish the longitudinal and circumferential functions.

GLS in MR

Likely because of their subendocardial location, the myocardial fibres are highly sensitive to disturbance by various noxious stimuli.²⁵ In an experimental model, a rapid reduction in mitral annular motion has been demonstrated after ischaemia induction.²⁶ Manaka et al.²⁷ demonstrated that myocardial systolic impairment may occur on the endocardial side and extends to the epicardium. Longitudinal function plays a fundamental role in cardiac systole by reducing the longitudinal LV size as the mitral annulus is pulled towards the apex; in fact, it has been demonstrated that longitudinal motion can mobilize as much as 60% of stroke volume.²⁸

The geometrical change that occurs owing to MR can directly impair the longitudinal fibres and may have an impact on LV longitudinal function regardless of myocardial contractility. The myocardial fibre geometry of the left ventricle changes gradually from a right-handed helix in the subendocardium to a left-handed helix in the subepicardium.²⁹ Additionally, during the remodelling phase of MR, the LV geometry becomes more globular following the LV dilatation. Therefore, the alignment of the muscle fibres becomes more transverse, preserving the myocardial contractility but impairing the longitudinal contraction. Kim et al.³⁰ demonstrated that speckle-tracking longitudinal strain becomes depressed earlier than the LVEF in MR cohort. In patients with MR, a proportion of the regurgitant volume is emptied into the left atrium before the aortic valve opens.³¹ In addition, the lower wall stress during early and late systole and the reduced afterload result in an increased total LV stroke volume.³² These two mechanisms can further mask an existing LV dysfunction that LVEF cannot detect but may potentially be uncovered by assessment of longitudinal function.

The results of our study compliment previous echocardiography-derived GLS investigations in primary MR. Mentias et al.³³ showed that resting and exercise LV-GLS provided incremental prognostic utility in asymptomatic patients with primary MR and preserved systolic function. They also demonstrated that abnormal resting LV-GLS was associated with higher mortality in this patient population.³³ In addition, Kim et al.³⁴ have shown that in patients with severe primary MR, GLS is an independent predictor of all-cause mortality after surgery in comparison with conventional parameters. A recent systematic review evaluating the role of echocardiography-derived GLS in severe primary MR after mitral repair or replacement concluded that impaired baseline GLS was associated with higher mortality rates.³⁵ In patients with secondary MR, Namazi et al.³⁶ showed that impaired GLS was independently associated with an increased risk of all-cause mortality, whereas LVEF was not. In addition, Kamperidis et al.³⁷ concluded that in patients with severe secondary MR, speckle-tracking GLS identified more deterioration in LV systolic function than LVEF.

Limitations

Valvular patients are frequently referred to CMR as a second-line test, and as such, a referral bias might have affected the prevalence of disease, comorbidities, and rate of adverse clinical events in the patients in this study. Symptoms related to MR reporting by the patients were subjective and objective measurement of symptoms such as exercise stress testing was not systematically performed in our study. However, the addition of other criteria, including surgical referral, although it has its own limitation, should minimize and allow us to include patients with clinically relevant MR-related symptoms for further analysis. Additional information from echocardiogram, despite beyond the scope of current study, would be of value; however, as a referral centre, the echocardiograms for many of our patients were performed in the referring cardiologist’s office, not available to us, and the reports frequently did not include a complete quantification of MR. For these reasons, our study did not compare GLS by echocardiogram with GLS by CMR, but the future studies that assess this will be important to perform. Despite stringent study design, we acknowledged that it may not be possible to entirely exclude certain secondary MR such as atrial function MR or functional MR in dilated cardiomyopathy with mild systolic dysfunction given that there were some overlap features with our primary MR cohorts and transoesophageal echocardiogram was not performed in all patients to determine aetiology of MR. However, these specific populations should comprise a small fraction of our study given that among 78 patients with indeterminate MR aetiology, there are four patients with LVEF between 50 and 60% and LV end-diastolic volume index >100 mL/m² and 7 patients with severely enlarged left atrium (LA volume index >72 mL/m²). MR severity is known and found to be a strong predictor of adverse outcome in our patients with primary MR. When the patients were stratified by a severity of MR (Figure 5), a statistical significance for association of GLS and our composite outcome only achieved in moderate but not severe MR group despite similar magnitude of the difference between those with a GLS above or below −16.6%. The lack of a statistically significance for the severe MR group is likely related to a smaller number of subjects in severe MR group, and the use of GLS in this population requires further investigation and should not be disregarded. The prognostic and clinical implication of GLS in patients with primary MR needs to be confirmed in future studies that should emphasize on answering the questions whether (i) earlier intervention in patients with MR with impaired GLS will successfully improve long-term outcomes, (ii) there is inter-vendor variability in GLS measurements, and (iii) the numerical thresholds of GLS may trigger earlier mitral intervention.

Conclusion

In patients with primary MR with preserved systolic function, GLS was associated with our composite outcomes of mitral surgery referral owing to symptoms or LV systolic dysfunction and cardiovascular death. Additionally, GLS is independently associated with all-cause mortality in this population. Our findings suggest that GLS may represent a marker of cardiac dysfunction in patients with primary MR with preserved systolic function allowing for the identification of patients likely to decompensate during observation.

Supplementary data

Supplementary data are available at European Heart Journal - Cardiovascular Imaging online.

Acknowledgements

The authors thank the technologists and nurses of the DeBakey CMR laboratory for their dedication and the patients of Houston Methodist Hospital for their participation in this study.

Funding

D.J.S. received support from the National Science Foundation (CNS-1931884), the Beverly B. and Daniel C. Arnold Distinguished Centennial Chair Endowment at the Houston Methodist Hospital Foundation, and the National Institutes of Health (R01HL171150).

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

References

Author notes