JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY VOL. 78, NO.

11, 2021

ª 2021 THE AUTHORS. PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN

COLLEGE OF CARDIOLOGY FOUNDATION. THIS IS AN OPEN ACCESS ARTICLE UNDER

THE CC BY LICENSE (http://creativecommons.org/licenses/by/4.0/).

Phenotypic Expression and Outcomes in

Individuals With Rare Genetic Variants of
Hypertrophic Cardiomyopathy
Antonio de Marvao, MBCHB, PHD,a Kathryn A. McGurk, PHD,b Sean L. Zheng, BMBCH,b,c Marjola Thanaj, PHD,a
Wenjia Bai, PHD,d,e Jinming Duan, PHD,a,d,f Carlo Biffi, PHD,a,d Francesco Mazzarotto, PHD,b,c,g,h Ben Statton, MSC,a
Timothy J.W. Dawes, MB BCHIR, PHD,a,b Nicolò Savioli, PHD,a Brian P. Halliday, MBCHB, PHD,b,c Xiao Xu, PHD,b,c
Rachel J. Buchan, MSC,b,c A. John Baksi, MBBS, PHD,b,c Marina Quinlan, MSC,a Paweł Tokarczuk, PHD,a
Upasana Tayal, BMBCH, PHD,b,c Catherine Francis, BMBCH, PHD,b,c Nicola Whiffin, PHD,b,c,i
Pantazis I. Theotokis, MSC,a Xiaolei Zhang, PHD,b Mikyung Jang, PHD,b,c Alaine Berry, MSC,a Antonis Pantazis, MD,b,c
Paul J.R. Barton, PHD,a,b,c Daniel Rueckert, PHD,d,j Sanjay K. Prasad, MD,b,c Roddy Walsh, PHD,k Carolyn Y. Ho, MD,l
Stuart A. Cook, MBBS, PHD,a,b,c,m,n James S. Ware, MB BCHIR, PHD,a,b,c,* Declan P. O’Regan, MBBS, PHDa,*

ABSTRACT

BACKGROUND Hypertrophic cardiomyopathy (HCM) is caused by rare variants in sarcomere-encoding genes, but little
is known about the clinical significance of these variants in the general population.

OBJECTIVES The goal of this study was to compare lifetime outcomes and cardiovascular phenotypes according to the
presence of rare variants in sarcomere-encoding genes among middle-aged adults.

METHODS This study analyzed whole exome sequencing and cardiac magnetic resonance imaging in UK Biobank
participants stratified according to sarcomere-encoding variant status.

RESULTS The prevalence of rare variants (allele frequency <0.00004) in HCM-associated sarcomere-encoding genes in
200,584 participants was 2.9% (n ¼ 5,712; 1 in 35), and the prevalence of variants pathogenic or likely pathogenic for HCM
(SARC-HCM-P/LP) was 0.25% (n ¼ 493; 1 in 407). SARC-HCM-P/LP variants were associated with an increased risk of death
or major adverse cardiac events compared with controls (hazard ratio: 1.69; 95% confidence interval [CI]: 1.38-2.07;
P < 0.001), mainly due to heart failure endpoints (hazard ratio: 4.23; 95% CI: 3.07-5.83; P < 0.001). In 21,322 participants
with both cardiac magnetic resonance imaging and whole exome sequencing, SARC-HCM-P/LP variants were associated
with an asymmetric increase in left ventricular maximum wall thickness (10.9  2.7 mm vs 9.4  1.6 mm; P < 0.001), but
hypertrophy ($13 mm) was only present in 18.4% (n ¼ 9 of 49; 95% CI: 9%-32%). SARC-HCM-P/LP variants were still
associated with heart failure after adjustment for wall thickness (hazard ratio: 6.74; 95% CI: 2.43-18.7; P < 0.001).

CONCLUSIONS In this population of middle-aged adults, SARC-HCM-P/LP variants have low aggregate penetrance
for overt HCM but are associated with an increased risk of adverse cardiovascular outcomes and an attenuated
cardiomyopathic phenotype. Although absolute event rates are low, identification of these variants may enhance
risk stratification beyond familial disease. (J Am Coll Cardiol 2021;78:1097–1110) © 2021 The Authors. Published by
Elsevier on behalf of the American College of Cardiology Foundation. This is an open access article under the CC BY
license (http://creativecommons.org/licenses/by/4.0/).

From the aMRC London Institute of Medical Sciences, Imperial College London, Hammersmith Hospital Campus, London, United
Listen to this manuscript’s
Kingdom; bNational Heart and Lung Institute, Imperial College London, London, United Kingdom; cCardiovascular Research
audio summary by
Centre at Royal Brompton and Harefield Hospitals, London, United Kingdom; dBiomedical Image Analysis Group, Department of
Editor-in-Chief
Computing, Imperial College London, London, United Kingdom; eDepartment of Brain Sciences, Imperial College London, Lon-
Dr. Valentin Fuster on
don, United Kingdom; fSchool of Computer Science, University of Birmingham, Birmingham, United Kingdom; gDepartment of
JACC.org.
Experimental and Clinical Medicine, University of Florence, Florence, Italy; hCardiomyopathy Unit, Careggi University Hospital,
Florence, Italy; iWellcome Centre for Human Genetics, University of Oxford, Oxford, United Kingdom; jFaculty of Informatics and
Medicine, Klinikum Rechts der Isar, TU Munich, Munich, Germany; kDepartment of Experimental Cardiology, Amsterdam UMC,
AMC Heart Centre, Amsterdam, the Netherlands; lCardiovascular Division, Brigham and Women’s Hospital, Boston, Massachu-
setts, USA; mNational Heart Centre Singapore, Singapore; and the nDuke-NUS Graduate Medical School, Singapore. *Drs Ware and
O’Regan are joint senior authors.

ISSN 0735-1097 https://doi.org/10.1016/j.jacc.2021.07.017

1098 de Marvao et al. JACC VOL. 78, NO. 11, 2021

Impact of HCM-Associated Variants on the Adult Population SEPTEMBER 14, 2021:1097–1110

H
ABBREVIATIONS ypertrophic cardiomyopathy 2 groups of variants found in 8 genes with definitive
AND ACRONYMS (HCM) is characterized by clinical evidence for HCM (3): sarcomeric variants P/LP spe-
and genetic heterogeneity, incom- cifically for HCM (SARC-HCM-P/LP) and rare sarco-
ACMG = American College of
Medical Genetics and Genomics
plete and age-dependent penetrance, and meric variants of indeterminate significance (SARC-
variable expressivity (1). Most individuals IND) with the potential to cause HCM, dilated
CMR = cardiac magnetic
resonance imaging with HCM have a normal life expectancy cardiomyopathy (DCM), or have little impact on car-
DCM = dilated cardiomyopathy but are at increased risk of adverse outcomes diomyopathy risk. Using high-precision, deep
HCM = hypertrophic
such as heart failure, atrial fibrillation, learning phenotyping of cardiac magnetic resonance
cardiomyopathy stroke, or sudden cardiac death (2). imaging (CMR), we also characterized phenotypic
LVH = left ventricular SEE PAGE 1111
manifestations and estimated the prevalence of
hypertrophy penetrant disease. Lastly, we determined the preva-
MACE = major adverse
A recent expert-led assessment of the lence and genetic yield of sequencing in unexplained
cardiovascular events validity of reported gene associations with left ventricular hypertrophy (LVH) among this adult
P/LP = pathogenic or likely HCM identified 8 sarcomeric genes with population.
pathogenic definitive evidence for disease causation (3),
SARC-HCM-P/LP = including MYH7 and MYBPC3, that account METHODS
pathogenic or likely pathogenic
for the majority of genetically explained
variants for hypertrophic
disease (4). The American College of Medical STUDY COHORTS. The UKBB recruited 500,000 par-
cardiomyopathy in sarcomere-
encoding genes Genetics and Genomics (ACMG) includes this ticipants aged 40 to 69 years across the United
SARC-IND = indeterminate set of genes among those for which specific Kingdom between 2006 and 2010 (National Research
variants in hypertrophic
variants are known to be causative of disease Ethics Service, 11/NW/0382) (9). This study was
cardiomyopathy–associated
phenotypes and are clinically actionable (5). conducted under terms of access approval number
sarcomere-encoding genes
(rare variants that do not meet The ACMG recommends that these genes 40616. In each case, written informed consent was
criteria for pathogenic/likely
be analyzed whenever clinical exome provided. The results are reported in accordance with
pathogenic annotation)
sequencing is undertaken and that patho- the Strengthening the Reporting of Observational
SARC-NEG = genotype Studies in Epidemiology guidelines (10) and the
negative/no rare variants in
genic or likely pathogenic (P/LP) variants
sarcomere-encoding genes should be proactively reported to patients checklist provided in Supplemental Table 1.

UKBB = UK Biobank as secondary findings. With increasing CARDIAC PHENOTYPING USING MACHINE LEARNING.

WES = whole exome

availability of whole exome sequencing Participants in the imaging substudy were randomly
sequencing (WES), both in wider clinical settings and as invited, and the response rates with exclusion criteria
WT = wall thickness direct-to-consumer asymptomatic testing, have been previously reported (11). Each underwent
this raises questions regarding potential CMR at 1.5-T (12). Segmentation of the cine images
benefit as well as downstream risks (6). Evidence is was performed by using a deep learning neural
not currently available that would allow a critical network algorithm developed in-house and opti-
evaluation of genomic screening at the population mized on the UKBB cohort. The performance of image
level (7). Specifically, it is unclear what risk annotation using this algorithm is equivalent to a
cardiomyopathy-associated variants confer in the consensus of expert human readers and achieves
general adult population and their phenotypic subpixel accuracy for cardiac segmentation (13).
expression outside families with penetrant disease. Myocardial wall thickness (WT) was measured along
Current evidence is based on aggregating data from radial line segments connecting the endocardial and
small and often underpowered studies (8), using epicardial surfaces perpendicular to the myocardial
different variant classifications and relying on center-line and excluding trabeculae (Figure 1), an
inconsistent phenotyping. approach that also exceeds the reliability of human
Here, we sought to determine the population experts (14). Chamber volumes and mass were
prevalence of rare sarcomeric variants in a prospec- calculated from the segmentations according to
tively recruited cohort of >200,000 middle-age standard post-processing guidelines (15). Myocardial
participants drawn from the UK Biobank (UKBB) and strain analysis was performed by using nonrigid
to assess lifetime risk of adverse events. We analyzed free-form deformation image registration (16,17).

The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’
institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information,
visit the Author Center.

Manuscript received May 7, 2021; revised manuscript received July 1, 2021, accepted July 6, 2021.
JACC VOL. 78, NO. 11, 2021 de Marvao et al. 1099
SEPTEMBER 14, 2021:1097–1110 Impact of HCM-Associated Variants on the Adult Population

F I G U R E 1 Cardiac Image Analysis in the UK Biobank

A Cardiac Image Analysis

3-Dimensional cardiac segmentation

Wall Thickness/mm
100 10

Radial Strain %
80 8
60
40 6
20 4
0 2
−20 3-Dimensional
−40 wall thickness model 0

Cardiac strain analysis

B Distribution of Maximum Left Ventricular Wall Thickness (n = 21,272)

1,500

1,000
Count

500

5 10 15 20
Maximum LV Wall Thickness (mm)
Male Female

(A) Machine learning segmentation of the heart from cardiac magnetic resonance imaging (right atrium: light blue; right ventricle: dark blue; left atrium: yellow; left
ventricle: red; left ventricular myocardium: green). Motion analysis was used to derive strain and strain rates (radial strain in diastole and systole shown). Regional
analysis of left ventricular (LV) wall thickness was performed by using 3-dimensional modeling. Mean wall thickness for 21,322 UK Biobank participants is mapped onto
the LV surface; the right ventricle is shown as a mesh. (B) Histogram of maximum LV wall thickness according to sex.
1100 de Marvao et al. JACC VOL. 78, NO. 11, 2021

Impact of HCM-Associated Variants on the Adult Population SEPTEMBER 14, 2021:1097–1110

Trabecular traits were quantified by using fractal Appendix). Individuals harboring variants that did
dimension analysis in which a higher value indicates not fit into these 3 categories were removed from
more complex trabeculation (18). Parametric analyses, including those with rare variants in genes
3-dimensional analysis of the geometry of the left associated with HCM genocopies, those with
ventricle was performed to map regional patterns intermediate-frequency variants (0.00004 < allele
of remodeling and quantify the association with frequency <0.001), those with variant classes not
genetic and environmental predictors (16,19-21). robustly established as disease-causing (eg, trun-
Further details on phenotyping are given in the cating variants in MYH7), and those identified as P/LP
Supplemental Appendix. for DCM (n ¼ 7) (3). Details on the variant curation
pipelines are presented in the Supplemental
SEQUENCING AND VARIANT CATEGORIZATION
Appendix.
PIPELINE. UKBB participants underwent WES as
previously described (22). They were divided into 3 OUTCOME MEASURES. The effect of genotype strata

genetic strata. Individuals were classified as genotype on clinical outcomes was assessed by using lifetime
negative (SARC-NEG) if they had no rare protein- risk. The UKBB reports the date of first occurrence of
altering genetic variation (minor allele frequency a diagnosis, identified from self-reporting, primary
<0.001 in the UKBB and the Genome Aggregation care, hospital in-patient, and death register records.
Database) (23) in any of 25 genes that may cause or This permitted the identification of events preceding
mimic HCM. These 25 genes represent an inclusive recruitment to the UKBB. The primary clinical
list of 8 sarcomere-encoding genes with definitive outcome was a composite of all-cause mortality or
evidence of an association with HCM, 3 moderate- major adverse cardiovascular events (MACE) defined
evidence sarcomere-encoding genes, and 14 genes as a diagnostic code for heart failure (including car-
associated with syndromic phenotypes that can diomyopathy), arrhythmia, stroke, or cardiac arrest
include LVH (3). This SARC-NEG group was compared events. Secondary clinical outcomes were the indi-
with individuals with rare variants in 8 sarcomere- vidual components of the primary clinical outcome. A
encoding genes definitively associated with HCM full list of endpoint definitions and data fields used
(MYBPC3, MYH7, MYL2, MYL3, TNNI3, TNNT2, TPM1, from the UKBB database is presented in the
and ACTC1). Analysis was restricted to robustly Supplemental Appendix and Supplemental Table 2.
disease-associated variant classes for each gene and STATISTICAL ANALYSIS. Statistical analysis was
to variants sufficiently rare to cause penetrant performed with R version 3.6.0 (R Foundation for
disease (filtering allele frequency <0.00004) (24). Statistical Computing) and RStudio Server version
Variants were classified as pathogenic/likely path- 1.043, unless otherwise stated. Variables are
ogenic (SARC-HCM-P/LP) if reported as P/LP for HCM expressed as percentages if categorical, mean  SD if
in ClinVar and confirmed by manual review (n ¼ 81), continuous and normal, and median (interquartile
or if annotated as P/LP according to ACMG criteria, range) if continuous and non-normal. Baseline
using the semi-automated CardioClassifier decision anthropometric data were compared by using
support tool (24) (n ¼ 19); the curation is depicted in Kruskal-Wallis tests and, if differences were identi-
Supplemental Figure 1. Other variants that were fied, a Wilcoxon test was used for pairwise compari-
consistent with known disease mechanisms and suf- sons with Benjamini-Hochberg adjustment for
ficiently rare, but not recorded in ClinVar and without multiple testing. Imaging parameters in 2 or more
other computationally available data to robustly groups were compared by using analysis of covari-
classify as P/LP, were defined as indeterminate ance, adjusted for relevant clinical covariates. When
sarcomeric variants (SARC-IND) (25,26). Despite being differences were significant, a Tukey post hoc test
curated for HCM, given that 5 of the 8 genes (MYH7, was applied for pairwise multiple comparisons.
ACTC1, TNNT2, TNNI3, and TPM1) that have defini- Three-dimensional phenotypic regression modeling
tive evidence for HCM have also been implicated in applied threshold-free cluster enhancement and
DCM (27), it is possible that some SARC-IND variants permutation testing to derive the P values associated
have the potential to cause DCM. The SARC-IND with each regression coefficient following adjustment
strata differs from those classically termed variants to control the false discovery rate, as previously
of unknown significance as it likely contains addi- described (16,20).
tional variants that would be reported as P/LP if Clinical outcomes were analyzed in participants
subject to full manual curation, which is not feasible with WES stratified according to genotype categories.
in >5,000 individuals (further subgroup in- Cox proportional hazards were calculated for lifetime
vestigations are presented in the Supplemental risk of clinical events. For the primary outcome,
JACC VOL. 78, NO. 11, 2021 de Marvao et al. 1101
SEPTEMBER 14, 2021:1097–1110 Impact of HCM-Associated Variants on the Adult Population

F I G U R E 2 Flowchart of UKBB Participants

200,584 participants with WES data 39,551 participants with CMR data

157,922 SARC-NEG 5,219 SARC-IND 493 SARC-HCM-P/LP 38,348 WT <13 mm 763 WT = 13-15 mm 163 WT ≥15 mm

36,950 excludeda 277 excludedb

21,322 UKBB participants with CMR and WES data

16,844 SARC-NEG 513 SARC-IND 49 SARC-HCM-P/LP

3,916 excludeda

We included 200,548 participants with whole exome sequencing (WES) in the UK Biobank (UKBB) and stratified them according to variant pathogenicity for outcome
analysis. Machine learning was also used to characterize left ventricular traits in 39,551 participants, of whom 21,322 also had sequencing. aIndividuals excluded if
carriers of: 1) rare variants in genes associated with HCM genocopies or LV phenotype; 2) intermediate frequency variants (0.00004 < AF < 0.001); 3) variant classes
not robustly established as disease causing. bCMRs excluded from WT measure due to nondiagnostic imaging, incomplete sequences, and other technical reasons.
AF ¼ allele frequency; CMR ¼ cardiac magnetic resonance imaging; HCM ¼ hypertrophic cardiomyopathy; SARC-HCM-P/LP ¼ pathogenic or likely pathogenic variants
for hypertrophic cardiomyopathy in sarcomere-encoding genes; SARC-IND ¼ indeterminate variants in hypertrophic cardiomyopathy–associated sarcomere-encoding
genes (rare variants that do not meet criteria for pathogenic/likely pathogenic annotation); SARC-NEG ¼ genotype negative; WT ¼ wall thickness.

competing risk analysis was performed by using the (0.25%) heterozygotes for 100 SARC-HCM-P/LP vari-
cause-specific survival method (28). Secondary anal- ants, including 16 compound heterozygotes of a
ysis of incident clinical events from recruitment SARC-HCM-P/LP variant and a SARC-IND variant, and
excluded individuals with events preceding enroll- 5,219 (2.6%) heterozygotes with SARC-IND variants,
ment and was performed by using Cox proportional including 112 compound heterozygotes where neither
hazards adjusted for age at recruitment. Time-to- variant met criteria to be classified as SARC-HCM-
event was censored at first event for each outcome, P/LP (Supplemental Table 3).
death, or last recorded follow-up. The relationship LEFT VENTRICULAR WALL THICKNESS IN THE
between genotype or CMR phenotypes and outcomes GENERAL POPULATION. Of the 39,274 subjects who
was assessed with multivariable Cox proportional underwent CMR analysis, 763 (1.9% [1 in 51]) had mild
hazards models and robust SE estimates. Sex was hypertrophy (WT 13-15 mm) and 163 (0.4% [1 in 241])
included as a covariate in all models. Hazard pro- had at least moderate hypertrophy (WT $15 mm).
portionality assumption was tested by using Participants with WT $13 mm were older (66.3  7.4
Schoenfeld residuals. Sex was found to be in violation years vs 63.6  7.6 years; P < 0.001), more often male
of these assumptions, and therefore a sex-stratified (88.4% vs 46.9%; P < 0.001), and with higher systolic
analysis was conducted with interaction co- blood pressure (149.5.2  18.8 mm Hg vs 138 
efficients. Outcomes are reported as hazard ratios 18.2 mm Hg; P < 0.001) and body surface area (2.07 
(HRs) with 95% confidence intervals (CIs) and pre- 0.21 m 2 vs 1.86  0.21 m 2; P < 0.001). There were no
sented graphically as cumulative hazards and Cox differences in terms of race. The prevalence of
proportional hazards curves. phenotypic HCM, defined as WT $15 mm in the
RESULTS absence of hypertension and valve disease, was
0.19% (n ¼ 76 [1 in 517]).
PARTICIPANTS. We analyzed WES data from 200,584 PENETRANCE AND EXPRESSIVITY OF RARE
participants and CMR from 39,551 subjects. A total of SARCOMERIC VARIANTS. Analyses were restricted
21,322 had both CMR and WES data available to UKBB participants with both CMR and WES
(Figure 2, Table 1). (n ¼ 21,322). In this subset, 16,844 were denoted as
PREVALENCE OF RARE SARCOMERIC VARIANTS. SARC-NEG, 513 as SARC-IND, and 49 as SARC-HCM-P/
There were 157,922 SARC-NEG subjects, 493 LP. CMR-derived cardiac measurements by genotype
1102 de Marvao et al. JACC VOL. 78, NO. 11, 2021

Impact of HCM-Associated Variants on the Adult Population SEPTEMBER 14, 2021:1097–1110

T A B L E 1 Subject Characteristics and CMR-Derived Cardiac Measurements According to Genotype

SARC-NEG SARC-IND SARC-HCM-P/LP SARC-NEG vs SARC-NEG vs SARC-IND vs

(n ¼ 157,922) (n ¼ 5,219) (n ¼ 493) SARC-HCM-P/LP SARC-IND SARC-HCM-P/LP

Female 86,710 (54.9) 2,866 (54.9) 276 (56.0)

Age, y 56.5  8.1 56.3  8.2 56.2  8.1
White 150,403 (95.2) 4,756 (91.1) 461 (93.5) 0.11 <0.001 0.11
SBP, mm Hg 139.7  19.6 139.5  19.8 139.2  20.2
DBP, mm Hg 82.2  10.7 82.2  10.5 81.4  10.7
BSA, m2 1.9  0.22 1.9  0.22 1.9  0.22
eGFR, mL/min/1.73 m2 90.7  13.4 91.1  13.3 90.1  13.3
Hypercholesterolemia 29,137 (18.5) 950 (18.2) 94 (19.1)
Hypertension 52,356 (33.2) 1,723 (33.0) 168 (34.1)
Diabetes 8,429 (5.3) 309 (5.9) 25 (5.1)
Aortic valve disease 1,567 (1.0) 37 (0.7) 8 (1.6)
Known HCM 109 (0.07) 9 (0.17) 18 (3.65) <0.001 0.006 <0.001
Known DCM 265 (0.17) 9 (0.17) 1 (0.2)
On medication for cholesterol, 18,537 (11.7) 599 (11.5) 64 (13.0)
blood pressure, diabetes

SARC-NEG SARC-IND SARC-HCM-P/LP

(n ¼ 16,844) (n ¼ 513) (n ¼ 49)

LVEDV, mL 147.9  33.5 150.1  34.6 139.4  30.3 0.18 0.005 0.02
LVESV, mL 60.4  19 61.4  19.7 54.7  15.9 0.07 0.04 0.01
LVEF, % 59.5  6.1 59.5  6.1 61.1  5.9
LVM, g 86  22.1 87.1  23.6 86.2  26.6 0.43 0.001 0.99
LV maximum WT, mm 9.4  1.6 9.5  1.7 10.9  2.7 <0.001 0.008 <0.001
LV concentricity, g/mL 0.58  0.09 0.58  0.09 0.62  0.12 0.001 0.95 0.002
LV global longitudinal strain, % –18.5  2.8 –18.4  2.8 –19  2.5
LV global radial strain, % 45  8.3 44.8  8.1 46.7  9.3
LV global circumferential strain, % –22.3  3.4 –22.2  3.3 –22.8  3.4
LV longitudinal PDSR 1.7  0.6 1.6  0.6 1.7  0.6
LV radial PDSR –5.7  2.0 –5.7  1.9 –5.9  2.0
LAEDV, mL 72.7  23.1 73.6  27.8 81.8  26.8 0.004 0.29 0.02
LAESV, mL 29.2  15.3 30.2  20.8 34.4  15.4 0.03 0.18 0.12
LAEF, % 61.3  9.2 60.9  9.8 59.2  8.2
RVEDV, mL 156.6  37.4 157.7  37.6 138.6  32.2 <0.001 0.08 <0.001
RVESV, mL 67.6  21.3 68.1  21 55  15.3 <0.001 0.26 <0.001
RVEF, % 57.3  6.2 57.2  5.9 60.6  5.4 <0.001 0.99 <0.001
RAEDV, mL 86  27.7 89.1  32.9 80.6  27.9 0.41 0.007 0.08
RAESV, mL 46.1  19.1 47.9  24.4 42.2  19.4 0.33 0.03 0.08
RAEF, % 47.1  9.4 47.5  9.8 48.9  9.9
Mean apical FD 1.14  0.04 1.14  0.04 1.16  0.05 0.006 0.88 0.02
Mean basal FD 1.19  0.03 1.19  0.03 1.2  0.04
Mean global FD 1.17  0.03 1.17  0.03 1.18  0.03 0.002 0.96 0.003

Values are n (%) or mean  SD. Table includes data for genotype-negative individuals (SARC-NEG), individuals with pathogenic or likely pathogenic sarcomeric variants (SARC-
HCM-P/LP), or those with other rare indeterminate sarcomeric variants (SARC-IND). Bold P values are statistically significant. Kruskal-Wallis tests were conducted for each
variable to determine whether differences in participants’ characteristics between genotype group were significant; if so, a Wilcoxon test was used to perform pairwise
comparisons between groups, with Benjamini-Hochberg adjustment for multiple testing, and those adjusted P values are shown. For the cardiac magnetic resonance imaging
(CMR)-derived cardiac parameters, analysis was adjusted for age, sex, race, systolic blood pressure (SBP), and body surface area (BSA) using an analysis of covariance; when
differences between genotype groups were significant, a Tukey post hoc test was applied for pairwise multiple comparisons, and those P values are shown.
concentricity ¼ (left ventricular mass/left ventricular end-diastolic volume); DBP ¼ diastolic blood pressure; DCM ¼ dilated cardiomyopathy; eGFR ¼ estimated glomerular
filtration rate; FD ¼ fractal dimension; HCM ¼ hypertrophic cardiomyopathy; LAEDV ¼ left atrial end-diastolic volume; LAEF ¼ left atrial ejection fraction; LAESV ¼ left atrial
end-systolic volume; LV ¼ left ventricular; LVEDV ¼ left ventricular end-diastolic volume; LVEF ¼ left ventricular ejection fraction; LVESV ¼ left ventricular end-systolic volume;
LVM ¼ left ventricular mass; PDSR ¼ peak diastolic strain rate; RAEF ¼ right atrial ejection fraction; RVEDV ¼ right ventricular end-diastolic volume; RVESV ¼ right ventricular
end-systolic volume; WT ¼ wall thickness.

are summarized in Table 1. Wall thickness was greater showed evidence of concentric remodeling, smaller
in SARC-HCM-P/LP versus SARC-NEG (Figure 3) right ventricular volumes, higher left atrial
and marginally greater in SARC-IND after adjustment volume, and increased trabeculation. Independent,
for relevant clinical variables. Compared with 3-dimensional analysis of cardiac geometry showed
SARC-NEG, those harboring SARC-HCM-P/LP variants that SARC-IND variants were positively associated
JACC VOL. 78, NO. 11, 2021 de Marvao et al. 1103
SEPTEMBER 14, 2021:1097–1110 Impact of HCM-Associated Variants on the Adult Population

F I G U R E 3 Relationship Between Rare Variants in HCM-Associated Genes and WT

A ***
****
ns

20
Maximum LV Wall Thickness (mm)

15 mm
15

13 mm

10

SARC-NEG (n = 16,844) SARC-IND (n = 513) SARC-HCM-P/LP (n = 49)

Genotype

0.2
Beta Coefficients

0.1
0
SARC-IND
−.01
−.02
C

SARC-HCM-P/LP

(A) Dot and boxplots of maximum wall thickness according to genotype. (B and C) 3-dimensional modeling of LV geometry with standardized
beta-coefficients showing the strength of association between genotype and regional WT. Contour lines indicate significant regions
(P < 0.05) after correction for multiple testing. LV projections are septal (left) and anterior (right). ***P # 0.001; ****P # 0.0001. ns ¼ not
significant; other abbreviations as in Figures 1 and 2.
1104 de Marvao et al. JACC VOL. 78, NO. 11, 2021

Impact of HCM-Associated Variants on the Adult Population SEPTEMBER 14, 2021:1097–1110

with an asymmetric increase in WT (n ¼ 508; mean 95 (19.3%) primary clinical outcome events (all-cause
b ¼ 0.08; significant area ¼ 70.1%), predominantly mortality or MACE), and by age 70 years there were
across the mid-basal anterior septum in continuity 14,168 (cumulative incidence: 12%; 95% CI: 12%-12%),
with the anterior wall. There was a much stronger 461 (cumulative incidence: 12%; 95% CI: 11%-13%),
positive association between SARC-HCM-P/LP vari- and 80 events (cumulative incidence: 21%; 95%
ants and increased WT (n ¼ 48; mean b ¼ 0.31; signifi- CI: 16%-25%), respectively.
cant area ¼ 47%) in an asymmetric pattern involving Examining lifetime risks, SARC-HCM-P/LP variants
most of the septum, anterior wall, and apex. were associated with an increased risk of death or
In the 49 subjects harboring SARC-HCM-P/LP vari- MACE (HR: 1.69; 95% CI: 1.38-2.07; P < 0.001), heart
ants, LVH $13 mm was present in 9, corresponding to a failure (HR: 4.23; 95% CI: 3.07-5.83; P < 0.001), and
penetrance of 18.4% (95% CI: 9%-32%) for this arrhythmia (HR: 1.59; 95% CI: 1.22-2.09; P < 0.001)
phenotype. In 3 of these individuals, LVH was $15 mm (Figure 4, Supplemental Table 5). There was no dif-
(6.1%; 95% CI: 1.3%-16.8%). Of the 513 with SARC-IND ference in risk in any of the primary or secondary
variants, 15 individuals (2.9%; 95% CI: 1.6%-4.8%) clinical outcomes when comparing SARC-IND and
had WT $13 mm. In 390 of these 562 individuals with a SARC-NEG. Sex was independently associated with
SARC-IND or SARC-HCM-P/LP variant, there was no all clinical outcomes, with men having an increased
concomitant hypertension or valve disease (69.4%). risk. Although male subjects have a higher overall
Among these 390 individuals, the penetrance of risk of adverse outcomes (HR: 1.76; 95% CI: 1.71-1.81;
LVH $13 mm was 10.8% for SARC-HCM-P/LP (4 of 37) P < 0.001), the incremental genetic risk from SARC-
and 0.57% for SARC-IND (2 of 353). Details of variants HCM-P/LP mutations compared with SARC-NEG is
found in these 6 individuals with otherwise greater in female subjects (HR for female subjects:
unexplained LVH are shown in Supplemental Table 4. 2.24 [95% CI: 1.71-2.94; P < 0.001]; HR for male sub-
Among the 173 individuals with a rare sarcomere- jects: 1.31 [95% CI: 0.97-1.76; P ¼ 0.08]; HR for sex*-
encoding variant and hypertension or valve disease, SARC-HCM-P/LP interaction: 1.72 [95% CI: 1.15-2.58;
the penetrance of LVH was 41.7% for SARC-HCM-P/LP P ¼ 0.009]) (Supplemental Table 6, Supplemental
(5 of 12) and 8.1% for SARC-IND (13 of 161). Only 1 in- Figure 2). Sensitivity analyses of lifetime risk of
dividual of the 39 with a wall thickness $15 mm (2.6%), death or MACE, excluding participants with any car-
not accounted for by hypertension or valve disease, diomyopathy and excluding cardiomyopathy events
harbored a SARC-HCM-P/LP variant. There was no as an outcome, yielded concordant results to the
difference in the prevalence of left ventricular ejection primary analysis comparing SARC-HCM-P/LP with
fraction <50% (P ¼ 0.12) between subjects harboring SARC-NEG (excluding participants with cardiomyop-
SARC-HCM-P/LP variants (1 of 49 [2%]; 95% CI: 0.0%- athy HR: 1.29 [95% CI: 1.02-1.63; P ¼ 0.03]; excluding
10.9%), subjects with a SARC-IND variant (29 of 513 cardiomyopathy diagnosis from the HF composite
[5.7%]; 95% CI: 3.9%-8.0%), and those identified as outcome HR: 1.60 [95% CI: 1.25-2.04; P < 0.001])
SARC-NEG (888 of 16,844 [5.3%]; 95% CI: 4.9%-5.6%). (Supplemental Table 7), confirming that previous
The prevalence of left ventricular dilatation (defined findings were not solely driven by individuals with
by using the UKBB-derived normal ranges of left ven- known disease, nor only by a diagnostic label of
tricular end-diastolic volume [abnormal high >232 mL cardiomyopathy as an endpoint. Examining incident
in male subjects and >175 mL in female subjects]) (29) risks, primary clinical events (HR: 1.57; 95% CI: 1.2-
was also not different in SARC-NEG individuals (530 of 2.04; P < 0.001), including heart failure (HR: 3.15;
16,844 [3.15%]; 95% CI: 2.9%-3.4%) and those with 95% CI: 2.03-4.89; P < 0.001), were increased in
SARC-IND (16 of 513 [3.12%]; 95% CI: 1.8%-5.0%) or SARC-HCM-P/LP but not in SARC-IND (Supplemental
SARC-HCM-P/LP (0 of 49 [0%]; 95% CI: 0.0%-7.3%) Table 8, Supplemental Figure 3).
variants. In the 39,551 individuals who underwent CMR, the
CLINICAL OUTCOMES ASSOCIATED WITH RARE number of primary clinical events was 2,568 (6.7%),
SARCOMERIC VARIANTS. Clinical outcomes for 112 (14.7%), and 34 (20.9%) in the WT <13 mm,
163,634 participants were analyzed. Median age at 13-15 mm, and $15 mm subgroups, respectively.
recruitment was 58 years (interquartile range: 50-63 Increased WT was associated with an increased risk of
years), and participants were followed up for a me- death or MACE (per millimeter increase HR: 1.15;
dian of 10.8 years (interquartile range: 9.9-11.6 years) 95% CI: 1.11-1.20; P < 0.001) and each individual
with a total of 19,504 primary clinical events endpoint (Supplemental Table 9, Supplemental
reported. Among SARC-NEG (n ¼ 157,922), SARC-IND Figure 4), with similar findings in the imaging and
(n ¼ 5,219), and SARC-HCM-P/LP (n ¼ 493) in- WES subgroup (n ¼ 17,447) (Supplemental Table 10,
dividuals, there were 18,793 (11.9%), 616 (11.8%), and Supplemental Figure 5). Increased trabecular
JACC VOL. 78, NO. 11, 2021 de Marvao et al. 1105
SEPTEMBER 14, 2021:1097–1110 Impact of HCM-Associated Variants on the Adult Population

F I G U R E 4 Outcomes Stratified According to Variant Pathogenicity

A Death and MACE

0.40
1.0
0.35
0.9 0.30
0.25
Cumulative Hazards of Event

0.8 0.20
0.15
0.7 0.10
0.05
0.6
0.00
0.5 0 10 20 30 40 50 60 70 80
0.4
0.3
0.2
0.1
0.0

0 10 20 30 40 50 60 70 80
Years
Number at risk
Genotype

SARC-NEG 157,922 157,888 157,845 157,702 157,260 155,565 123,305 70,044 7,248
SARC-IND 5,219 5,219 5,216 5,213 5,199 5,140 4,005 2,264 208
SARC-HCM-P/LP 493 493 493 492 491 480 364 200 17

0 10 20 30 40 50 60 70 80
Genotype SARC-NEG SARC-IND SARC-HCM-P/LP

Cumulative hazard curves with zoomed plots for lifetime risk of: (A) death and major adverse cardiovascular events (MACE), consisting of heart failure, arrhythmia,
stroke, and cardiac arrest events, or (B) heart failure, stratified according to genotype, consisting of SARC-NEG, SARC-IND, or SARC-HCM-P/LP. (C) Forest plot of
comparative lifetime risk of clinical endpoints (Cox proportional hazards models adjusted for sex) according to genotype. Sex refers to male subjects compared with
female subjects. Abbreviations as in Figure 2.

Continued on the next page

complexity, as measured by fractal dimension anal- end-diastolic volume (HR: 8.05; 95% CI: 2.46-26.3;
ysis, was associated with adverse clinical outcomes P < 0.001) (Supplemental Tables 13 and 14).
(HR per 0.1 unit increase in mean global fractal
dimension: 1.04; 95% CI: 1.01-1.08; P ¼ 0.01) after DISCUSSION
adjustment for systolic blood pressure and WT
(Supplemental Table 11, Supplemental Figure 6). In this study of >200,000 adults, we found the preva-
Among 17,406 individuals in the 3 genotype strata lence of rare variants in HCM-associated sarcomere-
with imaging and WES data, there were 1,177 (7.0%), 44 encoding genes to be 2.9%, with the prevalence of SARC-
(8.3%), and 87 (16.0%) primary clinical events in SARC- HCM-P/LP variants conservatively measured at 0.25%,
NEG, SARC-IND, and SARC-HCM-P/LP, respectively. both less frequent than reported in smaller cohorts (30),
The risk of heart failure was higher with SARC-HCM-P/ prior to contemporary variant classification guidelines
LP variants despite adjustment for WT (HR: 6.74; (25) and a consensus emerging on genes with definitive
95% CI: 2.43-18.7; P < 0.001) (Supplemental Table 12) evidence for disease causation in HCM (3). By examining
and also when accounting for WT, left atrial volume, lifetime risks alongside precision cardiac phenotyping,
left ventricular ejection fraction, and left ventricular we provide a critical evaluation of large-scale genomic
1106 de Marvao et al. JACC VOL. 78, NO. 11, 2021

Impact of HCM-Associated Variants on the Adult Population SEPTEMBER 14, 2021:1097–1110

F I G U R E 4 Continued

B Heart Failure

0.20
1.0
0.9 0.15
Cumulative Hazards of Event

0.8 0.10
0.7 0.05
0.6
0.00
0.5 0 10 20 30 40 50 60 70 80
0.4
0.3
0.2
0.1
0.0

0 10 20 30 40 50 60 70 80
Years
Number at risk
Genotype

SARC-NEG 157,922 157,918 157,917 157,914 157,888 157,572 127,633 76,261 8,537
SARC-IND 5,219 5,219 5,219 5,218 5,216 5,203 4,125 2,468 253
SARC-HCM-P/LP 493 493 493 492 492 483 384 222 19

0 10 20 30 40 50 60 70 80
Genotype SARC-NEG SARC-IND SARC-HCM-P/LP

screening for sarcomeric variants outside of familial dis- SARC-HCM-P/LP variants (32). SARC-HCM-P/LP vari-
ease (Central Illustration). ants were associated with a greater increase in life-
We found that SARC-HCM-P/LP variants were time risk of death and MACE in female subjects than
associated with an increased lifetime risk of adverse in male subjects. This supports findings in patients
cardiovascular events, predominantly due to heart with HCM, in whom, despite apparent reduced dis-
failure and arrhythmia, which was partially inde- ease penetrance (33), women have lower survival
pendent of left ventricular wall thickness and not regardless of genotype (34,35). In contrast, SARC-IND
restricted to individuals with cardiomyopathy or variants (which include a mixture of variants with
reaching cardiomyopathy endpoints. The fact that potential to cause HCM, DCM, or have little impact on
heart failure risk is only partially explained by cardiomyopathy risk) express a minimal phenotype in
observable phenotypic expressivity is consistent with aggregate and have a generally benign clinical course.
the profibrotic state that precedes the development of Recent study shows that the same genetic pathways
LVH in those with HCM variants (31). This finding may lead to distinct disorders through opposing ge-
highlights the potential of genomic testing for sarco- netic effects in the general population (36).
meric variants to modify cardiovascular risk assess- Adults harboring SARC-HCM-P/LP variants had an
ment even in the absence of image phenotyping. Such attenuated HCM-related phenotype characterized by
variants are also associated with greater risk in HCM increased wall thickness in an asymmetric pattern
patient cohorts, with adverse outcomes reported to involving the anteroseptum and apex, concentric
be twice as likely in patients with HCM harboring ventricular remodeling, increased left atrial volume,
JACC VOL. 78, NO. 11, 2021 de Marvao et al. 1107
SEPTEMBER 14, 2021:1097–1110 Impact of HCM-Associated Variants on the Adult Population

F I G U R E 4 Continued

C
1.76 (1.71-1.81)
Death or MACE 1.69 (1.38-2.07)
1.00 (0.92-1.09)

1.73 (1.64-1.82)
Death 1.45 (0.99-2.13)
0.9 (0.77-1.05)

1.81 (1.75-1.87)
MACE 1.83 (1.47-2.27)
1.00 (0.92-1.10)
Endpoint

2.30 (2.13-2.47)
Heart Failure 4.23 (3.07-5.83)
1.08 (0.89-1.31)

1.80 (1.73-1.87)
Arrhythmia 1.59 (1.22-2.09)
1.03 (0.93-1.14)

1.80 (1.68-1.92)
Stroke 1.54 (0.96-2.48)
0.86 (0.71-1.05)

2.71 (2.25-3.26)
Cardiac Arrest 2.52 (0.94-6.74)
0.65 (0.36-1.19)

0.5 1.0 2.0 4.0

Hazard Ratio
Sex SARC-HCM-P/LP vs SARC-NEG SARC-IND vs SARC-NEG
1108 de Marvao et al. JACC VOL. 78, NO. 11, 2021

Impact of HCM-Associated Variants on the Adult Population SEPTEMBER 14, 2021:1097–1110

C E NT R AL IL L U STR AT IO N Outcomes and Expression of Rare Variants in Hypertrophic

Cardiomyopathy–Associated Genes

de Marvao, A. et al. J Am Coll Cardiol. 2021;78(11):1097–1110.

In 200,000 adults, the prevalence of variants pathogenic or likely pathogenic for hypertrophic cardiomyopathy (SARC-HCM-P/LP) was 1 in 407, whereas the
aggregate prevalence of indeterminate sarcomeric variants was 1 in 38. The SARC-HCM-P/LP variants were associated with increased risk of death and major adverse
cardiovascular events. We found associations with hypertrophic cardiomyopathy–like imaging phenotypes although the prevalence of overt cardiomyopathy was low.
SARC-IND ¼ indeterminate variants in hypertrophic cardiomyopathy–associated sarcomere-encoding genes (rare variants that do not meet criteria for pathogenic/
likely pathogenic annotation); SARC-NEG ¼ genotype negative.

and greater trabeculation. In subjects harboring cohorts, in which even patients without family his-
SARC-HCM-P/LP variants, hypertrophy ($13 mm) was tory have a comparatively high 30% yield of sarco-
present in 18.4%, with 3 individuals presenting with mere variants (39), suggesting that patients enrolled
hypertrophy $15 mm (6.1%). Penetrance estimates into cohorts and/or referred for diagnostic
from family studies vary, but a recent study of rela- sequencing represent a skewed subset, likely
tives of HCM probands observed an HCM phenotype enriched by those with clinical risk factors (40,41).
in 37% of genotype-positive relatives at first Although the natural history of disease and family
screening (37). Although UKBB participants have a pedigrees are not known for UKBB participants with
higher degree of kinship than would be expected in a unexplained LVH, it is plausible that multifactorial
random sample (38), these results suggest that the sarcomere-negative hypertrophy is the predominant
penetrance of familial HCM is additionally driven by etiology in the community (42,43). Nonetheless,
other genetic or environmental disease–modifying increasing WT was associated with higher risk of
characteristics, shared within families but that over- death and MACE (and each individual component),
lap less commonly in the community. Conversely, which persisted despite adjustment for sex, sarco-
only 2.6% of individuals with unexplained LVH meric genotype, and systolic blood pressure. These
(WT $15 mm) harbored SARC-HCM-P/LP variants. findings illustrate the additive role of clinical and
This contrasts with the diagnostic yield in HCM genetic assessment in risk stratifying patients with
JACC VOL. 78, NO. 11, 2021 de Marvao et al. 1109
SEPTEMBER 14, 2021:1097–1110 Impact of HCM-Associated Variants on the Adult Population

unexplained LVH and managing the appropriate FUNDING SUPPORT AND AUTHOR DISCLOSURES
screening of relatives (2).
This study was supported by the Medical Research Council, UK (MC-
STUDY LIMITATIONS. UKBB is a large-cross sectional
A658-5QEB0); the National Institute for Health Research Imperial
study that is subject to selection bias, excluding College Biomedical Research Centre; the National Institute for Health
younger and potentially more severe cases (44); Research Royal Brompton Cardiovascular Biomedical Research Unit;
the British Heart Foundation (NH/17/1/32725, RG/19/6/34387, RE/18/
however, risk factor associations seem to be broadly
4/34215); Fondation Leducq (16 CVD 03); Wellcome Trust (107469/Z/
generalizable (45). The population is predominantly 15/Z, 200990/A/16/Z); the National Heart and Lung Institute Foun-
European, and further work is required to explore dation; the Royston Centre for Cardiomyopathy Research; Rosetrees
traits and outcomes in people of diverse ancestries. and CORDA (Dr Prasad); Academy of Medical Sciences (SGL015/1006;
Dr de Marvao); Mason Medical Research Trust grant (Dr de Marvao);
Although we included outcome data before enroll-
SmartHeart EPSRC Programme Grant (EP/P001009/1; Dr Bai and Dr
ment in the UKBB, we do not know the natural history Rueckert); and a Rosetrees and Stoneygate Imperial College Research
of disease in the community or what factors may in- Fellowship (Dr Whiffin). Dr Ware has consulted for MyoKardia, Inc.
fluence conversion to penetrant disease; in addition, and Foresite Labs. Dr Cook holds shares in Enleofen Bio Pte. Ltd. Dr
O’Regan has consulted for Bayer AG. All other authors have reported
co-segregation data were not available. Although
that they have no relationships relevant to the contents of this paper
there is genetic overlap between HCM and DCM, we to disclose.
found that DCM traits do not drive clinical outcomes
in individuals with SARC-IND or SARC-HCM-P/ ADDRESS FOR CORRESPONDENCE: Dr James S.
LP variants. Ware, Imperial College London, MRC London Institute
CONCLUSIONS of Medical Sciences, Hammersmith Hospital Campus,
Du Cane Road, London W12 0HS, United Kingdom.
We found that SARC-HCM-P/LP variants are present E-mail: [email protected]. OR Dr Declan P. O’Re-
in 1 in 407 adults. Although the presence of SARC- gan, Imperial College London, MRC London Institute
HCM-P/LP variants in individuals in the community of Medical Sciences, Hammersmith Hospital Campus,
was rarely associated with the degree of unexplained Du Cane Road, London W12 0HS, United Kingdom.
hypertrophy required for a diagnosis of HCM, they are E-mail: [email protected].
associated with an attenuated phenotype and an
increased risk of adverse cardiovascular events even
PERSPECTIVES
in the absence of cardiomyopathy. SARC-HCM-P/LP
variants are likely to be harbored by >18 million
COMPETENCY IN MEDICAL KNOWLEDGE: Most individuals
people worldwide, and an improved understanding of
with pathogenic sarcomeric variants do not have overt HCM, but
their clinical significance is crucially important,
a subclinical phenotype is associated with an increased risk of
especially in light of recommendations to actively
adverse cardiovascular events.
screen for such variants as secondary findings during
clinical sequencing (46). Although penetrance for
TRANSLATIONAL OUTLOOK: Further genomic analyses are
overt HCM is modest and absolute event rates are
needed to characterize the specific risks levels and types of
low, identification of these variants may enhance risk
complications associated with various inherited forms of HCM.
stratification beyond familial disease, even when
cardiomyopathy is not manifest.

REFERENCES

1. Elliott PM, Anastasakis A, Borger MA, et al. 3. Ingles J, Goldstein J, Thaxton C, et al. Evalu- Genomics (ACMG). Genet Med. 2021;23:1391–
2014 ESC guidelines on diagnosis and manage- ating the clinical validity of hypertrophic cardio- 1398.
ment of hypertrophic cardiomyopathy: the Task myopathy genes. Circ Genom Precis Med. 2019;12:
6. Green RC, Berg JS, Grody WW, et al. ACMG
Force for the Diagnosis and Management of Hy- e002460.
recommendations for reporting of incidental
pertrophic Cardiomyopathy of the European So-
findings in clinical exome and genome sequencing.
ciety of Cardiology (ESC). Eur Heart J. 2014;35: 4. Maron BJ, Maron MS, Semsarian C. Genetics of
Genet Med. 2013;15:565–574.
2733–2779. hypertrophic cardiomyopathy after 20 years:
clinical perspectives. J Am Coll Cardiol. 2012;60: 7. ACMG response to Nussbaum et al. Letter on
2. Ommen SR, Mital S, Burke MA, et al. 2020 AHA/
705–715. ACMG policy statement: the use of secondary
ACC guideline for the diagnosis and treatment of
findings recommendations for general population
patients with hypertrophic cardiomyopathy: a 5. Miller DT, Lee K, Gordon AS, et al. Recom-
screening: a policy statement of the ACMG. Genet
report of the American College of Cardiology/ mendations for reporting of secondary findings
Med. 2019;21:2838–2839.
American Heart Association Joint Committee on in clinical exome and genome sequencing,
Clinical Practice Guidelines. J Am Coll Cardiol. 2021 update: a policy statement of the 8. Semsarian C, Ingles J, Maron MS, Maron BJ.
2020;76:e159–e240. American College of Medical Genetics and New perspectives on the prevalence of
1110 de Marvao et al. JACC VOL. 78, NO. 11, 2021

Impact of HCM-Associated Variants on the Adult Population SEPTEMBER 14, 2021:1097–1110

hypertrophic cardiomyopathy. J Am Coll Cardiol. aortic stenosis are revealed by 3D model machine and clinical outcomes in hypertrophic cardiomy-
2015;65:1249–1254. learning. Eur Heart J Cardiovasc Imaging. 2020;21: opathy. Circ Genom Precis Med. 2021;14(1):
417–427. e003062.
9. Sudlow C, Gallacher J, Allen N, et al. UK Bio-
bank: an open access resource for identifying the 22. Szustakowski JD, Balasubramanian S, Sasson A, 36. Tadros R, Francis C, Xu X, et al. Shared genetic
causes of a wide range of complex diseases of et al. Advancing human genetics research and drug pathways contribute to risk of hypertrophic and
middle and old age. PLoS Med. 2015;12:e1001779. discovery through exome sequencing of the UK dilated cardiomyopathies with opposite directions
Biobank. Nat Genet. 2021;53(7):942–948. of effect. Nat Genet. 2021;53:128–134.
10. von Elm E, Altman DG, Egger M, Pocock SJ,
Gøtzsche PC, Vandenbroucke JP. The Strength- 23. Karczewski KJ, Francioli LC, Tiao G, et al. The 37. van Velzen HG, Schinkel AFL, Baart SJ, et al.
ening the Reporting of Observational Studies in mutational constraint spectrum quantified from Outcomes of contemporary family screening in
Epidemiology (STROBE) statement: guidelines for variation in 141,456 humans. Nature. 2020;581: hypertrophic cardiomyopathy. Circ Genom Precis
reporting observational studies. Lancet. 434–443. Med. 2018;11:e001896.
2007;370:1453–1457.
24. Whiffin N, Walsh R, Govind R, et al. Car- 38. Bycroft C, Freeman C, Petkova D, et al. The UK
11. Littlejohns TJ, Holliday J, Gibson LM, et al. The dioClassifier: disease- and gene-specific compu- Biobank resource with deep phenotyping and
UK Biobank imaging enhancement of 100,000 tational decision support for clinical genome genomic data. Nature. 2018;562:203–209.
participants: rationale, data collection, manage- interpretation. Genet Med. 2018;20:1246–1254.
39. Ingles J, Burns C, Bagnall RD, et al. Nonfa-
ment and future directions. Nat Commun. 2020;11:
25. Richards S, Aziz N, Bale S, et al. Standards and milial hypertrophic cardiomyopathy: prevalence,
2624.
guidelines for the interpretation of sequence var- natural history, and clinical implications. Circ Car-
12. Petersen SE, Matthews PM, Francis JM, et al. iants: a joint consensus recommendation of the diovasc Genet. 2017;10:e001620.
UK Biobank’s cardiovascular magnetic resonance American College of Medical Genetics and Geno- 40. Maron BJ, Spirito P. Impact of patient selec-
protocol. J Cardiovasc Magn Reson. 2016;18:8. mics and the Association for Molecular Pathology. tion biases on the perception of hypertrophic
Genet Med. 2015;17:405–424. cardiomyopathy and its natural history. Am J
13. Bai W, Sinclair M, Tarroni G, et al. Automated
cardiovascular magnetic resonance image analysis 26. Walsh R, Mazzarotto F, Whiffin N, et al. Cardiol. 1993;72:970–972.
with fully convolutional networks. J Cardiovasc Quantitative approaches to variant classification 41. Maron BJ, Rowin EJ, Casey SA, Maron MS. How
Magn Reson. 2018;20:65. increase the yield and precision of genetic testing hypertrophic cardiomyopathy became a contem-
in Mendelian diseases: the case of hypertrophic porary treatable genetic disease with low mortal-
14. Augusto JB, Davies RH, Bhuva AN, et al.
cardiomyopathy. Genome Med. 2019;11:5. ity: shaped by 50 years of clinical research and
Diagnosis and risk stratification in hypertrophic
cardiomyopathy using machine learning wall 27. Jordan E, Peterson L, Ai T, et al. An evidence- practice. JAMA Cardiol. 2016;1:98–105.
thickness measurement: a comparison with human based assessment of genes in dilated cardiomy- 42. Butters A, Bagnall RD, Ingles J. Revisiting the
test-retest performance. Lancet Digit Health. opathy. Circulation. 2021;144(1):7–19. diagnostic yield of hypertrophic cardiomyopathy
2020;3:e20–e28. 28. Austin PC, Lee DS, Fine JP. Introduction to the genetic testing. Circ Genom Precis Med. 2020;13:
analysis of survival data in the presence of e002930.
15. Schulz-Menger J, Bluemke DA, Bremerich J,
et al. Standardized image interpretation and post- competing risks. Circulation. 2016;133:601–609. 43. Harper AR, Goel A, Grace C, et al. Common
processing in cardiovascular magnetic resonance— 29. Petersen SE, Aung N, Sanghvi MM, et al. genetic variants and modifiable risk factors un-
2020 update: Society for Cardiovascular Magnetic Reference ranges for cardiac structure and function derpin hypertrophic cardiomyopathy susceptibility
Resonance (SCMR): Board of Trustees Task Force using cardiovascular magnetic resonance (CMR) in and expressivity. Nat Genet. 2021;53:135–142.
on Standardized Post-Processing. J Cardiovasc Caucasians from the UK Biobank population cohort. 44. Munafò MR, Tilling K, Taylor AE, Evans DM,
Magn Reson. 2020;22:19. J Cardiovasc Magn Reson. 2017;19:18. Davey Smith G. Collider scope: when selection bias
16. Biffi C, de Marvao A, Attard MI, et al. Three- 30. Bick AG, Flannick J, Ito K, et al. Burden of rare can substantially influence observed associations.
dimensional cardiovascular imaging-genetics: a sarcomere gene variants in the Framingham and Int J Epidemiol. 2018;47:226–235.
mass univariate framework. Bioinformatics. Jackson Heart Study cohorts. Am J Hum Genet. 45. Batty GD, Gale CR, Kivimäki M, Deary IJ, Bell S.
2018;34:97–103. 2012;91:513–519. Comparison of risk factor associations in UK Bio-
17. Duan J, Bello G, Schlemper J, et al. Automatic 3D 31. Ho CY, López B, Coelho-Filho OR, et al. bank against representative, general population
Bi-ventricular segmentation of cardiac images by a Myocardial fibrosis as an early manifestation of based studies with conventional response rates:
shape-refined multi-task deep learning approach. hypertrophic cardiomyopathy. N Engl J Med. prospective cohort study and individual partici-
IEEE Trans Med Imaging. 2019;38:2151–2164. 2010;363:552–563. pant meta-analysis. BMJ. 2020;368:m131.

18. Meyer HV, Dawes TJW, Serrani M, et al. Ge- 32. Ho CY, Day SM, Ashley EA, et al. Genotype and 46. Kalia SS, Adelman K, Bale SJ, et al. Recom-
netic and functional insights into the fractal lifetime burden of disease in hypertrophic cardio- mendations for reporting of secondary findings in
structure of the heart. Nature. 2020;584:589– myopathy: insights from the Sarcomeric Human clinical exome and genome sequencing, 2016 up-
594. Cardiomyopathy Registry (SHaRe). Circulation. date (ACMG SF v2.0): a policy statement of the
2018;138:1387–1398. American College of Medical Genetics and Geno-
19. de Marvao A, Dawes TJ, Shi W, et al. Pre-
mics. Genet Med. 2017;19:249–255.
cursors of hypertensive heart phenotype develop 33. Lorenzini M, Norrish G, Field E, et al. Pene-
in healthy adults: a high-resolution 3D MRI study. trance of hypertrophic cardiomyopathy in sarco-
J Am Coll Cardiol Img. 2015;8:1260–1269. mere protein mutation carriers. J Am Coll Cardiol. KEY WORDS cardiovascular magnetic
2020;76:550–559. resonance, deep learning, genetics,
20. Schafer S, de Marvao A, Adami E, et al. Titin-
truncating variants affect heart function in disease 34. Geske JB, Ong KC, Siontis KC, et al. Women hypertrophic cardiomyopathy, penetrance
cohorts and the general population. Nat Genet. with hypertrophic cardiomyopathy have worse
2017;49:46–53. survival. Eur Heart J. 2017;38:3434–3440. A PPE NDI X For supplemental figures and
21. Bhuva AN, Treibel TA, De Marvao A, et al. Sex 35. Lakdawala NK, Olivotto I, Day SM, et al. As- tables, please see the online version of this
and regional differences in myocardial plasticity in sociations between female sex, sarcomere variants paper.