Next-Generation Sequencing Screening Reveals Novel Genetic Variants for Dilated Cardiomyopathy in Pediatric Chinese Patients


 Dilated cardiomyopathy (DCM) is a myocardial disease characterized by bilateral or left ventricular cardiac dilation and systolic dysfunction leading to heart failure and sudden cardiac death in children. Most studies focus on the genetic alterations in DCM-related genes in adult populations; however, it remains enigmatic about the mutational landscape in pediatric DCM patients, especially in the Chinese population. We exploited the next-generation sequencing (NGS) technologies to genetically analyze 46 pediatric patients and to decipher the genotype-phenotype correlation in these patients’ clinical outcomes. Our results indicated DCM-associated pathogenic mutations in 10 genes related to the structure or function of the sarcomere, desmosomal and cytoskeletal proteins. We also identified 6 pathogenic mutations (5 novel) in the titin (TTN) gene leading to the formation of truncated TTN protein variants in 6 (13%) out of 46 patients each. Furthermore, we investigated the correlations between TTN gene mutation and clinical outcomes in these patients. Conclusion: Our data suggest that one-third of cases of pediatric DCM are caused by genetic mutations. The role of TTN variants should not be underestimated in pediatric DCM and age-dependent pathogenic penetrance of these genetic mutations needs to be considered in the case of familial DCM. Thus, NGS analysis can be applied to decode the yet unknown DCM etiological genetic factors in pediatric as well as adult patients.


Clinical Characteristics
The clinical characteristics of 46 (female-26 and male-20) pediatric patients with DCM onset are presented in Table 1 and Table 2. Almost all the patients were presented with a common respiratory syndrome like shortness of breath and cough. Notably, heart rhythm disturbances (1 atrial tachycardia, 2 premature ventricular contractions with low voltage, 1 transient junctional rhythm, 7 premature ventricular contractions, 2 left anterior fascicular blocks, 1 left bundle branch block, 3 atrial premature beats) were observed in 17 patients. It is important to note that 67.4% of patients (31/46) manifested the disease before 1 year of age with a median age of diagnosis of 6.5 months, and 84.8% (39/46) cases were diagnosed before the child's third birthday. At presentation, the mean LVEF was 28.5%, and the mean LVEDD Z-score was 7.02. The mean serum N-terminal pro-brain natriuretic peptide (NT-pro BNP) level was 14406pg/ml, much higher than the reference range (Table 3). During follow-up of median 12.5 months, 54.3% of them (25/46) were recovered from DCM and their LVEF was rescued to normal range. Because of the severity of DCM, 10 out of 46 patients died, and the average death timing range was 1-36 months (median 7 months) after the initial diagnosis. Importantly, 90% of deaths occurred within the rst year after diagnosis, and it was found to be a critical time period re ected by the signi cant morbidity and mortality. Genetic analysis of DCM showed dramatically high heterogeneity in these patients, while only 3 patients had a familial history of DCM. No signi cant sex difference was observed in LVEF, LVEDD Z-scores, age of onset, serum NT-proBNP level at diagnosis, and in follow-up time, the outcome of death or recovery (all P>0.05).

Genetic Characteristics of Mutation-Positive Group
Based on the sequencing results, the cohort was divided into two groups: the mutation-positive (MP) group (16/46) and the mutation-negative (MN) group (30/46), as shown in Table 3. Considering the stringent selection criteria and reclassi cation of DCM according to the ACMG guidelines, the genetic analysis suggested that these 16 patients could be classi ed as either 'pathogenic' or 'likely pathogenic' mutant carriers. All carriers were identi ed with only one pathogenic or likely pathogenic mutation in DCM-associated genes ( Figure 1). We identi ed 10 Table 4 summarizes the main clinical features and the details of identi ed variants in DCM patients. Excluding 3 patients with familial mutations in TTN, MYH7 and NEXN genes, 5 de novo variants were identi ed in NKX2-5, TNNI3, PRDM16, and RBM20 (n=2) genes. Among the identi ed 16 gene mutations (missense-5, non-sense-5, frameshift-4 and splice site-2), 2 same mutations were identi ed in a pair of monozygotic (MZ) twin patients, and 10 (62.5%) other mutations were novel., In total, we identi ed 6 different (5 novels) TTN truncation variants, including 3 nonsense, 1 frameshift, and 2 splice-site variants, in 13% of patients (Table 4). Furthermore, we mapped the identi ed variants to protein domains of TTN gene (Figure 3a), which showed consistent results with previous reports. Notably, 4 variants were in I-band, and the remaining 2 in A-band and M-band regions, respectively.

Genotype and Clinical Phenotype Analyses
The phenotypic severity of these patients at presentation was measured by LVEDD Z-score, LVEF, and serum proBNP level. There were no signi cant differences in distribution based on sex (P=0.202), age (P=0.23), LVEF (P=0.935), LVEDD Z-score (P=0.42), and serum proBNP level (P=0.115) between MP and MN groups. The MP group probands exhibited lower phenotypic severity (echocardiographic parameters) than MN group DCM probands, with the same LVEF of 28.5% and LVEDD Z-score of 6.4 versus 7.08, respectively. Also, we observed a trend that the onset age (median 9.5 months) in MP group was slightly higher than that in the MN group (median 6 months). Nine MP patients developed DCM before 12 months of age, and the age distribution was between 1 month and 13 years.
Despite the observation of better echocardiographic parameters and seemingly better phenotypes at presentation in MP group patients, there was no signi cant difference in the outcome (the number of deceased patients, P=0.145) and LVEF recovery patients, P=0.665) between the two groups during the follow-up of median time of 12.5 months (P=0.972). Survival curves free from death and recovery also showed similar results ( Figure 3). 37.5% (6/16) of MP DCM patients eventually suffered from heart failure related death compared to only 13% (4/30) of the MN group. Although no signi cant differences in survival free of death (P= 0.093) were documented, however, there was a trend toward a worse prognosis in our patients at advanced ages ( Figure 2a) when compared to the MN patients. The outcome of cardiac death occurred in patients with TTN truncating mutants (c.50065C>T, c.98421_98422insGG, c.37454-2A>T), and NKX2-5 (c.242delA), TNNT2 (c.422G>A), TNNI3 (c.544G>A) mutations. Notably, 5 out of 6 disease causing mutations belong to sarcomeric genes (TTN=3, MYH7,TNNT2, TNNI3).
In this study, a signi cant proportion of the patients exhibited marked improvement and better prognostic outcomes in response to heart failure treatment. In both groups, almost 50% of pediatric patients could recover (LVEF above 50%). There was a trend toward better recovery in MN patients at advanced ages ( Figure 2b).

TTN Truncation Variants in Pediatric DCM Patients
Here, we describe the clinical and genetic characteristics of 4 probands with 6 TTN mutations (as summarized in Table 4).
P2: A 12 years-old girl presented with dyspnea and fatigue. Her chest x-ray revealed enlarged heart size, and echocardiography showed LVEDD value 5.86cm and LVEF of 27%. Her brother died of severe congenital heart disease at <1 year age. Her LVEF was recovered to 58% after treatment. A novel 'likely pathogenic' nonsense heterozygous truncation mutation (c.43298T>G, p.Leu14433*) in TTN gene was con rmed in the proband and her mother by genetic analysis (Figure 4b). But her mother had no symptoms.
P4: A 12 years-old boy with DCM performed a cardiology outpatient clinical evaluation due to the sudden onset of shortened breath with abdominal pain, cough, chest tightness, and fatigue. The proband performed an electrocardiogram (ECG) showing the left anterior fascicular block. Also, he exhibited severely elevated plasmatic transaminases (AST = 275U/L, ALT = 881U/L). The abdominal CT elsewhere revealed ascites, pleural effusion. Remarkably, the plasmatic NT-proBNP level was also drastically high (960.4pg/ml) (normal values less than 125pg/ml at the proband's age). He also performed an electrocardiogram of the left anterior fascicular block and echocardiography showing LVEF of 20%, LVEDD of 7.29cm. After receiving the regimen of spironolactone 40mg/day, captopril 8.3mg/day, furosemide 40mg/day, digoxin 0.25mg/day, potassium chloride sustained release tablets 6000mg/day for 13 days, a repeated echocardiography showed a more severe dilated left ventricle (7.45cm) and a suspicion for left ventricular apex noncompaction. Three months after diagnosis of DCM, he was hospitalized again and died of ventricular tachycardia at a young age. NGS and Sanger sequencing con rmed the 'pathogenic' TTN (c.50065C>T, p.Arg16689*) variant in the proband and his mother (Figure 4b). His mother had no symptoms, and she refused to perform echocardiography. For the rst time, a pathogenic variant in TTN was associated with the DCM phenotype.
P8: A 10 years-old girl, who suffered from DCM for about 3 years and showed no response to heart failure medication, died of heart failure and pneumonia. At the onset, she had an increased serum phosphocreatine kinase (up to 2900u/L) level, and her ECG showed premature ventricular contractions. We identi ed a novel 'likely pathogenic' frameshift mutation in TTN (c.98421_98422insGG, p.I30808Pfs * 7) by NGS analysis in the proband and her father. Her father didn't show any symptoms. Involvement of skeletal myopathy was suspected in this patient because of increased phosphocreatine kinase level. The skeletal myopathy was then identi ed by muscle biopsy (failing to record the pathologic result). This suggests that this variant can cause disease in both heart and skeletal muscles.
P13: A girl diagnosed with DCM at the age of 3 months had an autosomal dominant family history of DCM. Her maternal grandfather (diagnosed at 36 and died at 52 years), the brother of the grandfather (diagnosed at 42 and died at 46 years), and their sister (diagnosed at 44 and died at 52 years) died of DCM. The proband presented with shortness of breath and atrial tachycardia. She was given a treatment of dopamine cardiotonic, furosemide, spironolactone, amiodarone, and captopril. Unfortunately, 6 months later Her mother felt shortness of breath after activity and performed echocardiography elsewhere, exhibiting LVEF=29% and LVEDD=6.8cm. After 3 years of follow-up, the proband could recover her LVEF to 62% with no signs of arrhythmia. But her mother still receives drug treatment and has not been recovered yet. By trio WES method, we identi ed a novel truncating 'likely pathogenic' mutation in TTN gene (c.105541A> T, p.K35181X) in the proband and her mother, but not in her father and 8-years old brother (Figure 3), who both had normal echocardiography parameters.

RBM20 Gene Mutation in Pediatric DCM
P6 and P7, monozygotic (MZ) twin brothers (a kinship coe cient of 0.4901), were diagnosed with DCM with LVEF 29% and 31%, respectively at the age of 12 years (the interval time of onset was 9 months). They both felt nausea, chest tightness, and shortness of breath. Both ECG showed ventricular premature beats. The MZ twin was administered with standard heart failure medication. After seven years of follow-up, the LVEF of the two patients were 56% and 55%, respectively, and remain asymptomatic. By WES screening, a 'pathogenic' de novo mutation c.2746G>A (p.E916K) in RBM20 gene was identi ed. The missense mutation was absent from controls (1000G, ESP, and ExAC-ALL), and was predicted to be 'pathogenic' by PolyPhen2, SIFT databases. The amino acid p.E916 is highly conserved across vertebrate species ( Figure 5C). We con rmed this missense mutation by Sanger sequencing in the twin patients but not in their parents ( Figure 5B). DCM is a common type of pediatric heart disease leading to poor clinical outcomes and heart transplantation [29,30]. In this cohort, with a presentation of mean LVEF of 28.5% and the mean LVEDD Z-score of 7.02, DCM presented a severe disease phenotype. 67.4% of patients (31/46) manifested DCM before 1 year of age. Genetic factors play an important role in DCM pathogenesis. However, genetic discovery and prognostic understanding are still a challenge for DCM therapy. Hence, we sought to discover the landscape of genetic variations in pediatric patients with DCM. In this study, among 46 pediatric patients with DCM, we found that 16 (34.8%) patients carried at least one pathogenic or likely pathogenic mutation in a disease-causing gene. Consistent with previous studies, mutations in sarcomere genes were detected most frequently in all these DCM patients [31]. The prevalence of pathogenic mutations seemed to be similar as in recently published cohorts [32,33]. There was no difference in clinical presentations and prognosis, no matter whether there were gene mutations or not. Mutations were identi ed frequently in childhood-onset of DCM, especially in sarcomeric genes. Importantly, TTN-related mutations were found in 10 out of 16 MP patients. TTN truncating mutations account for 12%-27% of all adult DCM cases, suggesting the importance of diagnostic sequencing [34][35][36][37][38]. However, the TTN truncating mutations have been rarely identi ed in pediatric patients previously [39][40][41]. Also, recent studies in pediatric DCM have shown similar results [14]. Notably, a genetic analysis in a cohort of 66 severe childhood cardiomyopathy, including 37 DCM cases, could not identify any TTN truncating mutations in these pediatric patients [22]. Likewise, another study involving 30 Chinese pediatric subjects with sporadic DCM pathology also could not nd any pathogenic TTN gene mutations [37]. However, in another study only one pathogenic TTN truncating variant has been identi ed in a 16 years-old boy among 36 pediatric DCM patients [38]. Obvious differently from low prevalence of TTN truncating mutants, we identi ed 6 different TTN truncating variants in 6 (13%) of 46 pediatric patients.
The prognosis of DCM cases with TTN truncating mutations has been found to be different and con icting in adult studies from moderate to severe outcomes. Notably, DCM patients with TTN mutations have shown a good response to treatment [39]. A cohort of 70 patients with end-stage DCM has shown recovery after left ventricle assist device implantation [40]. While another study has shown TTN mutation-positive patients frequently present severe cardiomyopathy and a worse 5-year prognosis [4].
The different clinical outcomes in TTN MP patients were also observed in our ndings. Three patients with TTN mutations (c.18230-1G>A, c.43298T>G, c.105541A>T) recovered after medical treatment with no further symptom, while 3 other patients harboring TTN mutations (c.50065C>T, c.98421_98422insGG, c.37454-2A>T) showed no response to treatment and died from heart failure, suggesting clinical heterogeneity. The con icting prognosis prompts that apart from the identi ed genetic factors, post-transcriptional, environmental, hormonal, and other factors could also modify the rate of disease progression. Therefore, it might be di cult to predict the clinical outcomes in pediatric DCM cases based on TTN truncating mutations alone.
It has been found that the penetrance of TTN truncating mutations can reach up to 100% by the age of 70 years [42], which leads to discordant segregation with phenotype, and the same variant has also been detected in the unaffected relatives. In some pedigrees of unaffected relatives, especially in familial DCM, a clinical follow-up with aging should be carried up. The age onset of girl P13 (3 months) with TTN truncating mutation (c.105541A>T, p.K35181X) was obviously different from her mother (23 years) and her grandfather (36 years), suggesting the possible involvement of some other factors in uencing DCM pathogenesis.
In addition, we identi ed a pathogenic RBM20 variant (p.E916K) in the twin patients (P6 and P7) with pediatric DCM. RBM20 mutations have been reportedly associated with cardiomyopathy [43]. Notably, the variant c.2746G>A (p.E916K) of RBM20 mutant identi ed in this study was located on exon11, a known hot spot for cardiomyopathy-associated mutations ( Figure 6). However, the mechanistic relation of RBM20 mutations with DCM onset is still unclear.
A de novo mutation in PRDM16 gene in an 8-months girl (P9) was identi ed to be associated with pediatric DCM. Previously known mutants of this gene are mostly missense mutations; however, a nonsense mutation leading to functional loss of PRDM16 was detected in this case. The role of PRDM16 gene mutations could be very important, warranting further studies to identify its pathogenic mechanism.

Limitations
The study suffers from the following limitations: 1) it was a single-centered, retrospective, and a small cohort study; 2) follow-up time was not long enough to estimate the long-term outcomes; 3) clinical assessment was highly recommended for family members of the probands, but only a minority of the relatives were willing to participate in this evaluation, so gene mutation carriers of a family could be underestimated in our cohort; 4) bioinformatic prediction could give us some useful information about the pathogenicity of mutants associated with DCM, however, it cannot re ect the real pathobiology of the mutants in the cardiac myocytes.

Conclusion
DCM is a genetically heterogeneous disease in children and adults. Using NGS analysis, we could detect that more than one-third of cases of DCM were caused by genetic mutations in genes related to the structure or function of sarcomeric, desmosomal and cytoskeletal proteins as well as decipher their genotype-phenotype correlations in pediatric DCM patients. Here, we discovered 6 TTN truncating mutations (5 novel) correlated with severe disease phenotypes. Furthermore, we found 16 mutations in 10 genes in 16 patients that were likely to be associated with DCM pathogenesis. Most pediatric patients were diagnosed with DCM before 1 year of age. Also, most deaths occurred within the rst year of life after diagnosis. The cardiac death occurred in patients harboring mutations in TTN (3), NKX2-5(1), TNNT2(1) genes. Hence, this study advances the genetic understanding of pediatric DCM and highlights certain genetic defects with severe clinical courses. To the best of our knowledge, this is the rst study reporting that the presence of TTN truncating mutations could be associated with the onset and development of DCM in pediatric patients. Further studies are needed to de ne the mechanisms by which pathogenic TTN variants affect outcomes in pediatric and adult patients with DCM.   Abbreviations: LVEF, left ventricular ejection fraction; LVEDD, left ventricular end diastolic diameter (Z-score, normal reference range between -2 standard deviation [SD] and +2 SD); NT-proBNP, N-terminal fragment of pro-b-natriuretic peptide. Abbreviations: LVEF, left ventricular ejection fraction; LVEDD, left ventricular end diastolic diameter (Z-score, normal reference range between -2 standard deviation [SD] and +2 SD); NT-proBNP, N-terminal pro-brain natriuretic peptide.      Family Pedigree of Patients P6 and P7 a) Pedigree analysis of probands P6 and P7 harboring DCM-associated RBM20 gene mutation. b) Sequence analysis of monozygotic twin probands and their parents. c) Alignment of RBM20 protein sequences across different vertebrates indicating mutation of highly conserved E916 amino acid residue.

Figure 6
Mutations of RBM20 Gene in the HGMD database Illustration of DCM-linked mutations in RBM20 protein as documented in the Human Gene Mutation Database. The novel missense mutation p.E916K identi ed in our NGS analysis is indicated in red color.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Coverlette.docx