A combinatorial oligogenic basis for the phenotypic plasticity between late-onset dilated and arrhythmogenic cardiomyopathy in a single family

Regulatory approval and consent to participate

The human molecular genetic studies were approved by the institutional review board of the University of Texas Health Science Center at Houston (HSC-IMM-07-0016). The studies were in accord with the ethical principles described in the Declaration of Helsinki. The participants gave written informed consents prior to participation.

Clinical studies

A pedigree was constructed, detailed medical history was obtained, and physical examination was performed on the participating family members. The participants underwent 12-lead electrocardiography (ECG) and complete echocardiography (M-mode, 2D, and Doppler). Selected family members underwent cardiac magnetic resonance (CMR) imaging per clinical indications, as judged by their care-providing physicians.

Exome sequencing and variant annotation

Whole Exome Sequencing (WES) was performed in ten family members, including five phenotypically affected, two likely affected, and three phenotypically normal individuals on an Illumina platform. The sequence reads were aligned to the human reference genome GRCh37 (hg19). To identify the genetic variants in each exome, BAM files, containing the sequence alignment data, were preprocessed according to GATK best practices and analyzed as published with some modifications^[8,9]. The variants were called by 4 different programs, namely GATK, Platypus, VarScan2, and FreeBayes, using the BETSY system to assess consistency among the variant callers and to weigh robustness of the detection^[10]. The variants were initially filtered based on a minor allele frequency of < 5% in a European cohort from the 1000 Genomes Project as annotated by ANNOVAR, in accord with the ethnic background of the family^[11,12]. Premature truncating variants, including stop gain, frameshift, and the canonical splice site variants, were considered PVs. Nonsynonymous variants in genes known to cause cardiomyopathies and/or arrhythmias with a CADD Phred (Combined Annotation-Dependent Depletion Phred) score of > 15 were considered LPVs and included for further analysis. SIFT, PolyPhen2, Mutation Taster, and other programs were also used to predict pathogenicity.

Clinical characteristics

The clinical characteristics of five clinically affected and two likely clinically affected family members are presented in Table 1. The pedigree of the family is shown in Figure 1. Characteristics of specific family members are summarized as follows:

Figure 1. Pedigree and list of the likely pathogenic variants (LPVs) in genes known to cause cardiomyopathies. Generation and family member identifiers are listed, the latter under each symbol. Gene name and the amino acid change or splice variants are listed under each symbol, indicating presence of the variants in that exome. Square: male; Circle: female; Full circle and square: affected individuals; Open circle and square: unaffected individuals; slash through: deceased individuals. Circle and square with a dot at the center indicate carrier individuals or obligate carrier (II-1). Gray circle with white center identifies an individual with ARVC.

The proband (III-2) was a 65-year-old Caucasian female who presented several years earlier with episodes of palpitations and was found to have left bundle branch QRS morphology on a 12-lead ECG, very frequent ventricular premature beats, and short runs of nonsustained ventricular tachycardia (NSVT) on Holter monitoring [Figure 2]. She had no history of syncope, presyncope, or symptoms of heart failure. However, she had a strong family history of sudden cardiac death and heart failure. An echocardiogram and a CMR imaging showed a depressed and a normal right ventricular size and function. There was no evidence of fibrofatty infiltration of the myocardium on the CMR imaging. A high-resolution computed tomography of the chest showed no significant coronary calcification and an Agatston calcium score of zero. Over the course of subsequent years, echocardiograms showed a mildly dilated left ventricle (LV) with a reduced LVEF, as low as 30% [Figure 3A-C]. Likewise, a CMR showed LC dilatation and dysfunction and no evidence of fibro-adiposis [Figure 3D]. She was diagnosed to have primary DCM and underwent implantation of an internal cardioverter/defibrillator (ICD) and ventricular tachycardia ablation. She has remained stable without new symptoms, and her LV size and function have improved over time to normal values.

Figure 2. A 12-lead electrocardiogram in the proband (the upper three panels) and a rhythm strip (lower three panels). The electrocardiogram shows a normal sinus rhythm, probable left atrial enlargement, a normal PR interval, and an increased QRS duration with a normal axis and morphology, indicative of left bundle branch block.

Figure 3. Echocardiogram and cardiac magnetic image resonance in the proband. Panels A and B show parasternal long axis view of the ventricles in diastole (panel A) and systole (Panel B). The left ventricular end diastolic and systolic diameters were measured 5.64 and 4.47 cm, respectively, and a calculated left ventricular fractional shortening of 21%. Panel C shows myocardial tissue Doppler velocities depicting low velocities. Panel D shows cross sectional myocardial images obtained by magnetic resonance imaging, which shows enlarged ventricles and absence of fibrofatty infiltration of the myocardium.

Family member III-4 was a 54-year-old female who had a history of controlled systemic arterial hypertension. She presented a few years ago with chest pain and was found to have nonspecific T wave changes on 12-lead ECG and mild segmental hypokinesia in the septum and inferior apical walls during stress echocardiography. She had a normal right ventricular (RV) and LV size and function on multiple echocardiograms. She also underwent cardiac catheterization and coronary angiography, which did not show obstructive coronary lesions. A 24-h Holter monitor showed rare ventricular premature beats and no significant sustained or non-sustained supraventricular or ventricular arrhythmias, except for premature beats. She died suddenly at the age of 66 years. An autopsy showed four-chamber dilatation and cardiac hypertrophy with a heart weight of 496 grams and evidence of extensive fatty infiltration of the right ventricle. Microscopic examination showed fibro-fatty infiltration of the right ventricle, encompassing 50% to 80% of the right ventricular wall thickness [Figure 4]. The left ventricle was normal, except for microscopic evidence of mild hypertrophy and minimal fibrosis.

Figure 4. Histological findings in a family member with the classic ARVC. Masson trichrome stained myocardial sections, low (upper panel) and high magnification (lower panel) fields, from the right and left ventricles of family member III-4 who died suddenly. Sections from the right ventricle showing infiltration of the right ventricular wall with fibro-adipocytes, whereas the section from the left ventricle shows mild fibrosis.

Three other affected family members (II-4, III-6, III-8) were diagnosed with DCM. All were initially symptomatic for heart failure at presentation and had significantly depressed LVEF (per verbal report in II-4). Three individuals had complete or incomplete left bundle branch block, and two had atrial fibrillation. Two had an ICD implanted because of syncope and/or frequent NSVT [Table 1].

Family member IV-2 presented with palpitations and presyncope and was found to have atrioventricular nodal reentrant tachycardia (AVNRT), NSVT, and PVCs. She was treated with antiarrhythmic medications and subsequently underwent radiofrequency ablation of AVNRT and an ICD placement. Echocardiography and CMR both showed normal right and left ventricular sizes and functions. There was no evidence of fibro-fatty infiltration of the myocardium on CMR.

One family member (IV-4) presented in his 40s and had a dilated LV with a reduced LVEF of 22% on an echocardiogram. He was a heavy alcohol drinker and was considered to have alcohol-related DCM. His LVEF was gradually recovered to the low 50s upon abstinence from alcohol.

The younger generation family members (generation V) were asymptomatic at the time of initial evaluation and had normal ECGs and echocardiograms.

WES quality control metrics

Peripheral blood was collected after obtaining written consent; DNA was extracted and analyzed by short-read WES on an Illumina platform. The average number of total reads was ~45 million reads (range: 15 to 68 million reads), and the mean coverage was 33 folds (range: 9.3 to 66 folds). Detailed quality control metrics are presented in [Supplementary Table 1] and were considered in the analysis of the data in each exome.

Over > 800,000 variants that were identified across the 10 exomes. Approximately 140,000 variants had coverage of at least 20 reads, of which at least 5 reads contained the variants, compared to the reference genome. The list included ~30,500, ~21,000, and ~18,000 variants with a minor allele frequency of each variant of < 5%, < 1%, and 0.1%, respectively, in the 1000 Genomes Project dataset (1000g2015aug_eur). The latter database was used given the European ethnicity of the family. There were ~4400 non-synonymous single nucleotide variants (nsSNVs), 81 stop codon variants (10 start loss, 4 stop loss, and 67 stop gain), ~70 frameshift variants, 130 in-frame insertion, and ~730 splice region variants.

Co-segregation of the variants with the phenotype

A total of 5000 putatively functional variants, comprised of nsSNVs, stop codon, frameshift, and splice region variants and a population frequency of < 5%, were analyzed for their co-segregation with the phenotype in five affected cases, two probably affected, and three unaffected family members [Figure 1]. Variants that were called at least by 2 different algorithms in three affected individuals with the best data quality (II-4, III-2, and III-4) were included. A total of 43 variants in 39 genes co-segregated with the affected status, not including those who were probably affected [Supplementary Table 2]. Among the candidate genes that contained co-segregating variants, TTN was the only gene that is expressed at high levels in cardiac myocytes, and given its well-established role in DCM, it had the strongest biological plausibility in causing the phenotype. The ACOX3, DTX2, NBPF14 (NBPF26), PDE4DIP, and TRIOBP genes were expressed at low to medium, medium to medium to high levels in the myocardium, according to the Human Protein Atlas (https://www.proteinatlas.org/). Variants in this set of genes had low CADD Phred scores and were predicted to be benign or tolerated by several additional algorithms. Likewise, among the variants that co-segregated with the phenotype, those in the CYP2D6, PARP4, and PJVK had a CADD Phred score of > 15, but the corresponding genes were expressed at low levels in the heart.

A splicing acceptor PV (NM_133378:exon192:c.37112-1G > A) in the TTN gene, which is expected to lead to premature truncation of the TTN protein, co-segregated with the phenotype [Figure 1]. Likewise, a missense variant in the TTN gene, namely p.L3240R (NM_003319) or p.K3286R in exon 42 (NM_001267550), was predicted to be pathogenic or damaging by multiple algorithms (CADD Phred score of 18.05). The remaining variants resided in genes that are either not expressed in the heart or are expressed at low levels or are diffusely expressed in various tissues, and all had low in silico prediction of pathogenicity. All co-segregating variants and the assessment of their pathogenicity, including their expression in the heart, are listed in [Supplementary Table 2].

Both TTN variants were detected in 2 probably affected family members (IV-2 and IV-4). One member (IV-2) presenting with cardiac arrhythmias, namely paroxysmal atrial tachycardia and NSVT carried the PVs in the TTN gene. She had a normal 12-lead ECG and normal right and left ventricular size and function on echocardiogram and CMR. She had no evidence of fibrofatty infiltration of the myocardium on the CMR. Thus, the phenotype was cardiac arrhythmias in the presence of normal cardiac structure and function. Individual IV-4 was clinically suspected of having alcohol-related DCM with severely depressed LVEF, who also carried the PVs in the TTN gene. The latter serving as the susceptibility variants, as the cardiac function significantly improved upon abstinence from alcohol consumption.

Individual IV-1 had inherited the PKP2 and the TTN variants but not the DSP variant from her mother, who had ARVC. When last examined, she had right axis deviation on the ECG but had a normal right and left ventricular size and function on echocardiogram and CMR at age 36 years. She had no evidence of fibro-fatty infiltration of the myocardium on CMR.

There were four children in the family (V-1, V-2, V-3, and V-4) who were selectively screened for the TTN variants, and two were carriers. Their parents did not carry the PKP2 or the DSP variants. The two TTN variant carriers had normal ECG and echocardiogram when last examined.

Potential modifier variants in other genes known to cause cardiomyopathy and/or arrhythmias

Given the variability in the phenotypic expression of cardiomyopathy in the family members and the post-mortem diagnosis of ARVC in one family member (III-4), the exome sequencing data were analyzed for the presence of potential modifier variants, defined as those with a MAF < 0.05 and a CADD Phred score of 15 or higher in 97 genes that have been associated with hereditary cardiomyopathies or cardiac arrhythmias Supplementary Table 3. Such variants were identified in 7 genes, including the TTN gene, in the affected family members [Table 2]. None of these variants, except for the TTN variants, co-segregated with the phenotype. However, they were notable for their potential impacts on the phenotype, including the presence of two heterozygous PVs and LPVs in the DSP and the PKP2 genes, known genes for ARVC, in the family member who exhibited with the classic ARVC at autopsy [Figure 1, Figure 4, and Table 2]. This individual had the highest load of PVs and LPVs, involving 7 different genes (5 genes involved in cardiomyopathy and 2 in cardiac arrhythmias). The DSP variant was also present in two additional affected family members who also have the TTN variants and presented with classic DCM. Other notable variants were in the FHL1, OBSCN, PRKAG1, RBM20, and SGCD genes, which are implicated in hereditary cardiomyopathies^[13-15].

Given the preponderance of conduction defects and cardiac arrhythmias in the family members, the exome sequencing data were also analyzed for the presence of uncommon PVs and LPVs, and those with a CADD Phred score of > 15. The results, shown in Figure 1 and Table 3, are notable for the presence of 5 notable variants, in addition to those also involved in cardiomyopathies, in the ABCC9, AKAP9, ANK2, and SCN10A in the family members. Individual III-4, who died suddenly despite a normal cardiac function, carried LPVs in the AKAP9 and SCN10A genes, both implicated in cardiac arrhythmias^[16,17].