Article Text

Download PDFPDF

Original article
Evolution of hypertrophic cardiomyopathy in sarcomere mutation carriers
  1. Carolyn Y Ho1,
  2. Allison L Cirino1,
  3. Neal K Lakdawala1,
  4. John Groarke1,
  5. Anne Marie Valente2,
  6. Christopher Semsarian3,
  7. Steven D Colan2,
  8. E John Orav4
  1. 1Cardiovascular Division, Brigham and Women's Hospital, Boston, Massachusetts, USA
  2. 2Department of Cardiology, Boston Children's Hospital, Boston, Massachusetts, USA
  3. 3Agnes Ginges Centre for Molecular Cardiology, Centenary Institute, University of Sydney, and Department of Cardiology, Royal Prince Alfred Hospital, Sydney, Australia
  4. 4Department of Medicine, Brigham and Women's Hospital, Boston, Massachusetts, USA
  1. Correspondence to Dr Carolyn Y Ho, Cardiovascular Division, Brigham and Women's Hospital, 75 Francis Street, Boston, MA 02115, USA; cho{at}partners.org

Abstract

Objective The early natural history of sarcomere mutations and the evolution to hypertrophic cardiomyopathy (HCM) are poorly characterised. To describe phenotypic progression, we compared mutation carriers who developed HCM to those who did not during prospective longitudinal investigation.

Methods Sarcomere mutation carriers without baseline left ventricular hypertrophy (LVH) were studied during participation in a pilot clinical trial testing diltiazem versus placebo. 38 participants (mean±SD age 15.8±8.6 years) were followed for a median of 2.9 years (range 1.0–5.1 years) with imaging and biomarker analysis. 4 participants (mean baseline age 13.8±3.9 years) developed HCM and were compared to those without phenotypic progression.

Results Participants who developed HCM were all children/adolescents and members of families with more highly penetrant mutations. At baseline, participants who developed HCM had a higher left ventricular (LV) ejection fraction (74±2% vs 69±1%, p=0.02), lower global E′ velocity (11.2±0.5 vs 14.8±0.4 cm/s, p<0.0001), higher N terminal pro peptide of B-type natriuretic peptide (NT-proBNP) values (208±72 vs 57±13 pg/mL, p=0.04), longer posterior mitral leaflets, and more prevalent ECG abnormalities. During follow-up, these parameters and cardiac troponin values continued to diverge in participants who developed HCM, although LV wall thickness stabilised.

Conclusions LV relaxation, ECG changes, mitral leaflet length, and serum NT-proBNP concentrations appeared more prominently abnormal at baseline in preclinical sarcomere mutation carriers who imminently progressed to HCM. LVH appears to stabilise within 2 years of onset. Further investigation is needed to improve our understanding of the evolution of this disease.

Trial registration number NCT00319982; Post-results.

View Full Text

Statistics from Altmetric.com

Introduction

Hypertrophic cardiomyopathy (HCM) was the first inherited cardiomyopathy to be described at the molecular level. In the ∼25 years since landmark studies established that sarcomere mutations cause the disease,1 ,2 genetic testing has become readily available and today reliably determines the genetic aetiology in affected families. Genetic testing also identifies individuals at a preclinical or subclinical stage—when a mutation is present but before the defining clinical phenotype, left ventricular hypertrophy (LVH), has developed.

Prior cross-sectional investigations of sarcomere mutation carriers (G+) with normal left ventricular (LV) wall thickness (denoted preclinical HCM or G+/LVH−) have helped identify early phenotypic manifestations. These studies have indicated that mutation carriers may have impaired LV relaxation,3 altered myocardial energetics,4 ECG abnormalities,5 mitral valve elongation,6 myocardial crypts,6–9 and evidence of a profibrotic state,10 ,11 without having diagnostic features of HCM. However, it is unclear how these early abnormalities relate to the future development of clinically overt disease.

Very few longitudinal studies have been performed on genotyped populations and there is particularly scant data on preclinical mutation carriers. Consequently, the phenotypic evolution and natural history of sarcomere mutations remain ill-defined. Some mutation carriers develop overt HCM during childhood, while others have normal LV wall thickness until middle age,12–16 or even indefinitely. The factors that predict or drive disease penetrance, expression, and progression have not yet been characterised. The goal of this study was to take advantage of longitudinal phenotypic data to gain insight into which ‘subclinical’ findings are indeed ‘preclinical’ by capturing the transition to disease.

We analysed prospective information on sarcomere mutation carriers who developed overt HCM during participation in a pilot randomised trial testing diltiazem as disease-modifying therapy in preclinical disease.17 Participants did not have diagnostic features for HCM at enrolment and were closely followed over a median of 2.9 years (range 1.0–5.1 years). Four of 38 participants (10%) developed LVH, leading to a new diagnosis of HCM during the trial. Here, we present a comprehensive description of baseline differences between mutation carriers who did and did not develop HCM, as well as how features changed over time. We also investigated potentially influential factors related to genetic background and family history. Improved understanding of the initial stages of phenotypic development will help enable early recognition of mutation carriers who are at highest risk for impending disease expression, and provide crucial guidance in deciding when novel disease-modifying interventions should be initiated and which individuals to target for therapy.

Methods

Study participants

Participants were enrolled in a pilot, placebo-controlled, randomised clinical trial testing diltiazem as disease-modifying therapy in preclinical HCM (ClinicalTrials.gov NCT00319982). The results of the trial have been previously reported.17 Briefly, participants were at least 5 years old, carried the pathogenic or likely pathogenic sarcomere mutation assumed to cause HCM in their family (a listing of all mutations is provided in the online supplementary table), and did not have pathologic LVH by echocardiography at enrolment (maximal LV wall thickness ≤12 mm in adults or z-score ≤3 in children <18 years old). Participants were randomised to receive sustained release diltiazem 360 mg daily (5 mg/kg in children) or placebo and followed for up to 5 years (maximum treatment duration 3 years). Study visits occurred at baseline, 6 months after enrolment, and then annually. Participants aged 8.5–16.5 years at enrolment were also assessed at 18 months. At each visit, physical examination, 12-lead ECG, echocardiography, and blood draw for biomarker analysis were performed. Cardiac magnetic resonance (CMR) imaging was performed at baseline and at the end of treatment when possible.

For this study, we focused on development of LVH, defined as the new appearance of maximal LV wall thickness ≥12 mm in adults or z-score ≥3 in children <18 years old, using echocardiography as the primary imaging modality.

The institutional review and ethics boards of three collaborating medical centres approved the study protocol. All participants provided written informed consent or assent if <18 years old at enrolment.

Echocardiographic analysis

Standard two-dimensional images, spectral and colour Doppler, and tissue Doppler interrogation were obtained serially. All echocardiographic studies were stored digitally for blinded offline analysis. Additional analyses performed for this study included measurement of mitral valve anterior and posterior leaflet lengths from the parasternal long axis view during diastole with the leaflet maximally elongated, as the distance from mitral annulus to leaflet tip. Mitral valve leaflet lengths were also indexed to long axis LV end-diastolic diameter.

CMR analysis

CMR was performed at enrolment, final visit on study medication, ∼1 year after cessation of study medication for the earliest enrolling participants, and occasionally during the treatment period to investigate when echocardiography suggested increased LV wall thickness. Additional analyses performed for this study included determining the number of myocardial crypts present in the LV. Crypts were defined as a focal myocardial defect showing at least partial systolic obliteration, and having a depth ≥50% the thickness of the adjacent compact myocardium in diastole. Three standard long-axis cines (with cross-validation on the short-axis cines, where appropriate) were used to locate crypts.6

Serum biomarkers analysis

Blood samples (serum and K3-EDTA plasma) were collected at each study visit. Personnel blinded to clinical and genetic status performed biomarker measurements using commercial assays. Markers analysed in this study included: carboxy-terminal propeptide of procollagen type I (PICP; Quidel Corporation, San Diego, California, USA), amino terminal propeptide of B-type natriuretic peptide (NT-proBNP; Roche, Indianapolis, Indiana, USA), and supersensitive cardiac troponin I (Singulex, Atlanta, Georgia, USA).

Statistical analysis

Age, gender and causal gene were summarised using means and standard deviations, or counts, and compared between participants who progressed to HCM (progressors) and those who did not (non-progressors) using a t-test or Fisher exact test, as appropriate. Continuous baseline clinical measures were compared using linear regression models accounting for within-family correlation as well as age, sex, causal gene, treatment arm, and the primary predictor, development of HCM. For binary ECG outcomes, an analogous clustered logistic regression model was used but, because of the limited sample size and risk of over-fitting, the model was only adjusted for age and within-family correlation. Both types of models were implemented using the generalised estimating equation (GEE) approach, through the Genmod procedure in the SAS statistical package (SAS V.9.2, Cary, North Carolina, USA). This approach does not require any assumptions about the normality of the continuous outcomes and uses robust standard errors based on the empirical distribution of the data. If an ECG abnormality appeared in 0% or 100% of either group so logistic regression models would not converge, a simple Fisher exact test was used.

We also explored differences between mutation carriers that did and did not develop HCM by analysing the longitudinal change between final and baseline imaging and biomarker parameters. Both z-scores and raw values of echocardiographic measures were analysed.

To interpret results as average change per year of follow-up, the longitudinal change in each continuous clinical measure was calculated for each participant and divided by the number of years of participation. Using GEE-based linear regression models as described above, we compared longitudinal changes between participants who did and did not develop HCM, and assessed for significant changes over time within each group. For binary ECG outcomes, we compared abnormality rates between the two groups as of last follow-up using the same clustered logistic regression models as baseline ECG data.

Family characteristics of participants who did and did not develop HCM were compared using Poisson regression with each family as the unit of analysis, the total number of characterised family members as the exposure offset, and the number of family members with the characteristic of interest (HCM; myectomy; transplant; and sudden death) as the outcome.

For all analyses, a two-sided p value <0.05 was considered significant. Because we considered it essential to investigate all possible relationships to phenotypic progression, no adjustment was made to the p values and false positive findings are possible.

Results

Study population

Thirty-eight participants were enrolled in the trial with a mean±SD age of 15.8±8.6 years at baseline (range, 5–39 years); 58% were female. All were healthy and had no cardiovascular symptoms. The median duration of total study participation and follow-up was 2.9 years (range 1.0–5.1 years; interquartile range 2.3–3.7 years).

During the study, four unrelated participants (10%) had substantial increases in LV wall thickness, leading to a clinical diagnosis of HCM. The baseline characteristics of participants who did and did not develop HCM are summarised in table 1. At enrolment, participants who developed HCM had 7% higher LV ejection fraction (LVEF) (mean±SE 74±2% versus 69±1%; p=0.02), and 24% lower global E′ velocity (11.2±0.5 versus 14.8±0.4 cm/s, p<0.0001). Baseline maximal echocardiographic LV wall thickness was also 13% higher in participants who developed HCM (p<0.001), although there was no significant difference in corresponding z-scores. Mitral valve posterior leaflet length was also 27% longer in participants who developed HCM. Two of four participants who developed HCM had ≥1 mm ST segment depression on baseline ECG tracings, compared to 0 of 34 participants who did not develop HCM (p=0.01).

Table 1

Baseline characteristics of mutation carriers who did and did not develop HCM

Participants who developed HCM had >3-fold higher baseline serum NT-proBNP values compared to others (208±72 vs 57±13 pg/mL, p=0.04). This difference was influenced by progressor 4, who had a baseline NT-proBNP value of 597 pg/mL. The presence of myocardial crypts did not predict phenotypic progression in this cohort, although six participants who did not develop HCM declined CMR imaging. Myocardial crypts were present on baseline CMR in two of four participants who developed HCM (with four and eight crypts each) and two of 28 participants who did not develop HCM (with three and four crypts each) (p=0.07).

Longitudinal evolution of phenotype and progression to HCM

Four unrelated participants developed HCM during the trial, including a 9-year-old female (randomised to diltiazem) and a 12-year-old male (randomised to placebo) who both carried the myosin heavy chain (MYH7) mutation p.Arg719Gln; and two 17-year-old males, one carrying the MYH7 mutation p.Arg663Cys (randomised to diltiazem) and the other with the troponin T (TNNT2) mutation p.Arg92Trp (randomised to placebo). Phenotypic evolution by serial echocardiographic and CMR images is shown in figures 14. Representative serial images from a non-progressor are shown in the online supplementary figure for comparison. In progressors, LVH initially developed over a 12–24 month timeframe, and then LV wall thickness appeared to stabilise with no substantial increase subsequently. Late gadolinium enhancement on CMR was detected only in progressor 4 (quantified as ∼2% of total LV mass) when LVH emerged.

Figure 1

Serial echocardiographic and cardiac magnetic resonance (CMR) images in participants who progressed to hypertrophic cardiomyopathy (HCM) are shown in figures 14. Arrows indicate the region of maximal left ventricular (LV) wall thickness. Progressor 1: Female, 9 years of age at enrolment (myosin heavy chain (MYH7) mutation p.Arg719Gln)—randomised to diltiazem. Baseline echocardiogram suggested borderline focal septal hypertrophy that appeared more prominent on CMR (CMR maximal LV wall thickness of 12 mm involving one segment). Echocardiograms at 6 and 12 months of follow-up showed no definitive change. Echocardiography at 18 months showed an increase in basal septal hypertrophy to 14 mm. CMR was performed at 18-month follow-up for careful reassessment of LV morphology and showed more prominent and extensive LV hypertrophy, with a maximal wall thickness of 18 mm now involving three segments. There was no late gadolinium enhancement. At 24 months, echocardiography showed localised hypertrophy, measuring up to 17 mm (z-score 10.9). CMR was unchanged from the 18-month study. At 36-month follow-up (1 year after stopping study medication), there was no significant change in LV wall thickness by echocardiography or CMR, and no visible late gadolinium enhancement.

Figure 2

Progressor 2: Male, 12 years of age at enrolment (myosin heavy chain (MYH7) mutation p.Arg719Gln)—randomised to placebo. Baseline images showed normal left ventricular (LV) wall thickness by echocardiography and cardiac magnetic resonance (CMR). The inferior septum appeared subtly thickened by echocardiography at 12-month follow-up, therefore CMR was performed at 18-month follow-up to more carefully assess for phenotypic progression. This CMR showed an increase in mid inferior septal wall thickness to 15 mm. Concomitant echocardiography identified a maximal LV wall thickness of 13 mm (z-score 3.9). At 24-month follow-up, both echocardiography and CMR showed progression of LV hypertrophy with mid-inferoseptal thickness increasing to 16 mm (z-score 8.0). No late gadolinium enhancement was seen by CMR. There was no significant change in echocardiographic LV wall thickness at 36 month follow-up (1 year after stopping study medication).

Figure 3

Progressor 3: Male, 17 years of age at enrolment (myosin heavy chain (MYH7) mutation p.Arg663Cys)—randomised to diltiazem. Baseline echocardiography and cardiac magnetic resonance (CMR) showed normal left ventricular (LV) wall thickness (10 mm). Echocardiography at 12-month follow-up suggested an area of focal hypertrophy at the basal to mid inferior septum, measuring ∼15 mm. This region had more prominent thickening at 24-month follow-up, but imaging was off-axis and difficult to compare. Echocardiography at 36-month follow-up clearly demonstrated basal to mid inferior septal hypertrophy, measuring 18 mm. CMR at 36 months similarly showed asymmetric septal hypertrophy with a maximal LV wall thickness of 22 mm at the mid inferoseptum. Echocardiography at 48 month follow-up (1 year after stopping study medication) showed no further progression of LV hypertrophy with a maximal LV wall thickness of 18 mm.

Figure 4

Progressor 4: Male, 17 years of age at enrolment (troponin T (TNNT2) mutation p.Arg92Trp)—randomised to placebo. Baseline echocardiography showed normal left ventricular (LV) wall thickness, although cardiac magnetic resonance (CMR) suggested a focal area of borderline LV hypertrophy at the basal to mid septum (11–12 mm). Focal septal hypertrophy by echocardiography was less prominent and remained unchanged at 6- and 12-month follow-up. Echocardiography and CMR at 24-month follow-up both showed equivocally increased LV wall thickness (11–12 mm) involving a very focal portion of the basal inferior septum. Echocardiography and CMR at 36 months were unchanged and there was no late gadolinium enhancement (LGE) on CMR. Echocardiography and CMR were then repeated at 47 months of follow-up (11 months after stopping study medication). Echocardiography showed normal parasternal long axis measures of LV wall thickness, but there was hypertrophy of the mid-septum appreciated on short axis views, measuring up to 17 mm. CMR showed a maximal wall thickness of 16 mm at the inferior septum. There were 2–3 small foci of LGE, quantified as ∼2% of total LV mass.

Longitudinal data were analysed to evaluate how phenotype changed during follow-up (table 2). Participants who developed HCM showed further decreases in E′ and S′ velocities (suggesting decreased longitudinal systolic function), while LVEF increased (suggesting increased circumferential systolic function). Figure 5 shows the trend lines for maximal LV wall thickness and global E′ velocity for the individual participants who developed HCM compared to mean values for the remaining cohort.

Table 2

Longitudinal changes in participants who did and did not develop HCM

Figure 5

Changes during follow-up in (A) maximal echocardiographic left ventricular (LV) wall thickness and (B) global E′ velocity in participants who did and did not progress to hypertrophic cardiomyopathy. Time is given in months following baseline studies at enrolment.

NT-proBNP and troponin values also increased significantly in participants who developed HCM but decreased slightly in the remainder of the cohort. All participants who progressed to HCM developed Q waves and/or ST segment depression, whereas ECG findings remained relatively stable in the remainder of participants.

Individuals who did not have crypts on baseline CMR did not develop them during follow-up. Notably, throughout the study, including after the development of pathologic LVH, all participants remained asymptomatic and did not endorse subjective change in exercise tolerance or alter their participation in recreational activities.

Family history, genetic background, and penetrance

We compared family history and genotype in mutation carriers who developed HCM with those who did not (table 3). Participants who developed HCM all carried well-characterised, definitively pathogenic mutations in MYH7 or TNNT2. In contrast, seven of 34 participants who did not develop HCM had variants that were downgraded from being classified as likely pathogenic to being of unclear significance during the trial period. Analyses were repeated, excluding these seven individuals, and showed no pronounced change, although the divergence in baseline LVEF and NT-proBNP values between the cohorts was no longer statistically significant.

Table 3

Genotype and family features of mutation carriers who did and did not develop HCM

The penetrance of sarcomere mutation was analysed using two methods: (1) based solely on clinical screening of evaluated family members; and (2) using combined results of genetic testing and clinical screening of genotyped relatives. Penetrance was higher in the families of the four participants who developed HCM than in the families of those who did not. Based on clinical screening alone, 66% of relatives in the families of progressors had a diagnosis of HCM, compared to 35% of relatives in the families of non-progressors (p=0.006). Looking only at genotyped relatives, penetrance was 100% in the families of participants who developed HCM versus 52% in the families of participants who did not develop HCM (p<0.001). Specifically, by the end of the study, progressor families no longer had any G+/LVH− relatives; all mutation carriers developed HCM. The clinical profiles of families were otherwise similar with no significant difference in the proportion of relatives with sudden death, myectomy or cardiac transplantation.

Discussion

In this study, we analysed comprehensive, prospective data on HCM sarcomere mutation carriers to evaluate phenotypic evolution and capture the transition to disease. Although none of the participants had pathologic LVH at enrolment, 10% developed HCM during the course of the trial, providing a valuable opportunity to investigate the phenotypic changes that may accompany or herald impending evolution of clinically overt disease.

Reduced E′ velocity,3 ,18 increased LVEF,3 altered myocardial energetics,4 ECG abnormalities,5 increased mitral leaflet length,6 myocardial crypts by CMR imaging,6–9 and a profibrotic state10 ,11 can be detected in sarcomere mutation carriers when LV wall thickness is normal, but their relationship to disease development has not been previously characterised. The results of this longitudinal study suggest that increased LVEF, reduced E′ velocities, elongated posterior mitral leaflets, and ECG abnormalities are early phenotypes more strongly associated with impending disease progression. Baseline abnormalities in these parameters were more prominent in individuals who developed HCM over the next 12–24 months than in those who did not. Indeed, all participants who progressed to HCM had reduced E′ velocity at baseline. Furthermore, all of these parameters, as well as S′ velocities and serum cardiac troponin T concentrations, continued to diverge over time in these two populations as clinical HCM became established. Other previously reported early phenotypes of HCM, including serum biomarkers of collagen metabolism and myocardial crypts on CMR, were not significantly different at baseline and did not diverge over time between participants who developed HCM and those who did not.

A small number of longitudinal studies have reported similar phenotypic progression to overt HCM in ∼6–11% of mutation carriers followed over medium-term clinical follow-up.12 ,13 ,15 ,16 In contrast to previous reports, this investigation included more frequent, systematic, and comprehensive evaluation, thus providing new insights into disease development. Early phenotypic abnormalities were uniformly present at baseline in mutation carriers with impending disease emergence. These findings are consistent with a recent study indicating that sarcomere mutations may even have an embryonic phenotype in animal models of HCM.19 We postulate that HCM develops along a continuum, with phenotypic abnormalities progressively accumulating as disease evolves. The rate of increase in LV wall thickness was similar in each of the participants who developed HCM. Once subtle hypertrophy was detected, maximal LV wall thickness was reached within approximately 12–24 months. After that time, LV wall thickness appeared to plateau. Similar to our previous experience,20 CMR imaging may be more sensitive and definitive than echocardiography in identifying the earliest emergence of LVH, particularly if non-septal segments are involved, although final measures of LV wall thickness did not differ substantially between these imaging modalities.

Several observations emerged from family and genotype analysis. Participants who developed HCM had MYH7 or TNNT2 mutations; no MYBPC3 mutation carriers (32% of the study cohort) developed HCM during follow-up. This observation is consistent with other studies suggesting more delayed and/or reduced penetrance in MYBPC3 mutation carriers.21 All participants who developed HCM did so during adolescence, the stereotypical time for HCM to develop. Adult mutation carriers did not demonstrate obvious phenotypic progression during the study period, suggesting that these individuals may possess unidentified ‘protective’ factors attenuating disease progression. Furthermore, mutations appeared to be more highly penetrant and manifest at a younger age in the families of participants who developed HCM. Clinical disease expression and outcomes, such as sudden cardiac death, need for myectomy or cardiac transplantation, otherwise did not appear more severe in the families of those who developed HCM with more penetrant disease.

The small cohort size and relatively short duration of follow-up are major limitations to this study, precluding the ability to draw definitive conclusions about disease pathogenesis. However, no previous reports have been able to carefully chronicle the evolution of clinically overt disease in HCM sarcomere mutation carriers. While much remains to be learned, this study is the first to identify early phenotypic features of sarcomere mutations that may be most predictive of impending development of disease.

Conclusions

Relatives who inherit the sarcomere mutation that causes HCM in their family are at high risk for developing disease. However, the penetrance of these mutations is age-dependent, disease expression is variable, and the features associated with phenotypic progression are largely unknown. This study suggests that several early phenotypic features of sarcomere mutations may be more strongly associated with impending disease progression in mutation carriers, including reduced E′ velocities, increased LVEF, elongated posterior mitral leaflets, ST depression on ECG, and higher NT-proBNP values. Participants from families with more highly penetrant mutations were also more likely to have phenotypic progression in the short term. Continued basic investigation and complementary longitudinal human studies are needed to improve understanding of the factors that drive both disease evolution and clinical outcomes. By being able to better predict which apparently healthy sarcomere mutation carriers are most likely to progress to clinical disease, we can refine both family management and emerging strategies for disease modifying therapies, including guiding whom to target and when to start treatment. Ultimately, such advances will substantively advance care of HCM patients and their families.

Key messages

What is already known on this subject?

  • Hypertrophic cardiomyopathy (HCM) was the first inherited cardiomyopathy to be characterised at the molecular level. With genetic testing, we can identify at-risk individuals who carry pathogenic sarcomere mutations before a clinical diagnosis of HCM can be made. However, little is known about how at-risk mutation carriers progress to developing clinically overt HCM.

What might this study add?

  • There have been no prior studies that have systematically described phenotypic evolution and the transition to disease. Therefore this prospective study provides new insights regarding the pathogenesis of HCM. More prominent abnormalities in left ventricular relaxation, serum natriuretic peptide values, electrocardiography, and mitral leaflet elongation appeared to predict impending development of clinically overt HCM, and diverged further as phenotypic evolution progressed. The full extent of left ventricular hypertrophy appears to be established within 18–24 months of initial emergence.

How might this impact on clinical practice?

  • Results from this study and further investigation into the evolution of HCM will provide important information to guide the management of families and at-risk relatives by identifying individuals who may have more penetrant disease and are more likely to progress imminently to HCM.

Acknowledgments

The authors wish to thank David Annese, RT, Emily Harris, Andrew J Powell, MD, Faranak Farrohi, and Jose Rivero for their contributions to this study.

References

View Abstract

Footnotes

  • Contributors CYH planned the study, drafted the manuscript, and is responsible for the overall content. ALC coordinated the study, and collected and analysed the data. EJO conducted the statistical analyses. NKL, SDC and CS assisted in study planning and provided critical comments. AMV performed CMR analyses. JG performed mitral valve analyses.

  • Funding This study was funded by the National Institutes of Health (K23 HL078901 to CYH) with additional support provided by the Arthur L. Lenahan Sr. Family Foundation (to CYH). Neither the NIH nor the Foundation played any role in the design or analysis of this study.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval Partners Healthcare Institutional Review Board.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles