Dilated cardiomyopathy (DCM) is a disease of the heart muscle characterized by ventricular dilatation and impaired systolic function, in the absence of abnormal loading conditions or coronary artery disease sufficient to cause global systolic impairment.1 Clinically, DCM is manifested by signs or symptoms of heart failure (HF), ventricular and supraventricular arrhythmias, conduction system abnormalities, thromboembolism or sudden death.2 DCM is a leading cause of HF, with 30-50% of enrolled patients in many HF registries and clinical trials categorized as having non-ischemic DCM.3 DCM is the third most common cause of HF and the main reason for heart transplantation.2

The true prevalence of DCM is difficult to ascertain. Earlier estimates derived from a study performed in 1975-1984 in Olmstead County, MN, USA, indicated a prevalence of approximately 1:2700 individuals,4 which is probably underestimated, due to the less sensitive imaging techniques used at that time.5 More recently, Hershberger et al.,6 using different methods, considered the prevalence to be between 1:250 and 1:400.

Familial forms are assumed to account for 30-50% of cases, with autosomal dominant inheritance being the predominant pattern of transmission, although autosomal recessive, X-linked and mitochondrial inheritance have also been described.5,7–10 In recent years, enormous advances have been made in our knowledge of the genetic basis of DCM, especially with the advent of new sequencing technologies, particularly next-generation sequencing (NGS), which enables several genes to be sequenced simultaneously, with decreased costs and turnaround times. However, accurate molecular diagnosis and determination of genotype-phenotype correlations are hampered by genetic heterogeneity and difficulty in correctly interpreting the clinical impact of variants.

Rare genetic variants in more than 50 disease-related genes coding for proteins of the different cell components (cytoskeleton, sarcomere, nuclear membrane and ion channels) have been identified, with considerable overlap with other cardiomyopathies.6,11–14 Besides this locus heterogeneity, there is also marked allelic heterogeneity, with disease-related variants scattered throughout the causal genes.6,11 Most variants are missense, characterized by variable expressivity6,15 and low and age- and gene-dependent penetrance.6,9,15,16 The majority of disease-causing variants are private (unique to a specific family)7,8,15 and it has been suggested that 9-13% of patients carry at least two potentially disease-causing variants.10,16,17 Nevertheless, genetic diagnosis in DCM is recommended in the international guidelines, since it appears to have implications for clinical practice in genetic counseling and risk stratification.15,18

In this context, we aimed to investigate the genetic background and molecular basis of familial and idiopathic cases of DCM in Portugal and to assess possible genotype-phenotype correlations.

In this study we investigated the genetic background and molecular basis of familial and idiopathic cases of DCM in Portugal, using NGS, which enabled us to study a relatively large panel of genes, more efficiently and at lower cost, compared to previous sequencing methodologies.

Our cohort of 107 DCM patients had similar baseline characteristics to the cohort of the European INHERITANCE project (which did not include Portugal), published in 2015 by Haas et al.,10 although we had fewer symptomatic patients (49% of our patients were in NYHA class I vs. 28% of the European cohort) and a lower proportion of heart transplant recipients (10% vs. 20%).

We found rare variants in approximately one fourth of patients. This proportion is similar to that of a study by Lakdawala et al.,26 lower than the results of a recently published Finnish cohort16 (35% of patients presented with pathogenic or likely pathogenic variants), and significantly lower than the proportion of positive molecular results obtained by Haas et al.,10 in which variants causing or associated with DCM were described up to 73% of patients. This lower diagnostic yield might be related to the fact that other genes potentially related to the disease were not included,26 most importantly the titin gene (TTN), which is presumed to account for up to 14-25% of all DCM cases27 and for the majority of variants found in the INHERITANCE project10 and in the Finnish cohort,16 but may also be related to a different degree of stringency for the classification criteria of variants. These differences are even more pronounced taking into account that we considered both variants that are likely pathogenic and those of uncertain significance. We decided to do so because most of the variants identified, although predicted to be functionally significant by in silico prediction tools, were classified as of uncertain significance mainly because they were novel and originally reported in this study, and familial segregation analysis was not informative.

Although without statistical significance, we found a higher proportion of patients with rare variants in familial DCM patients, which is in line with other studies.10,26 Even so, a genetic origin is presumed in the idiopathic cases, because of the occurrence of the novo variants, incomplete penetrance, limited number of genes analyzed and small families.

The majority of variants were found in sarcomeric genes, as in other reports,10,26 most of them in MYBPC3 (10/31). MYBPC3 variants are frequent in patients with HCM, representing 30-45% of all HCM-associated variants.28 In patients with DCM, MYBPC3 variants are thought to account for approximately 2-4% of cases,6,27 a lower percentage than in our cohort. As most of these variants (8/10) had previously been described in association with HCM, one might hypothesize that the DCM phenotype represented end-stage HCM (‘burnout’ physiology). We consider this hypothesis less likely in view of the follow-up of these patients previous to the inclusion in the study, the relatively young age of presentation, and family history of DCM in all but one patient.

In our study we could not define clear genotype-phenotype correlations for patients with a rare genetic variant, or in patients with specific variants. Although TNNT2 variant carriers appear to present at a younger age, in agreement with a recent meta-analysis by Kayvanpour et al.,27 the presence of a higher left ventricular ejection fraction in these patients and in MYH7 variant carriers was not confirmed. Also, we found no statistically significant differences in gender predominance, conduction disease, ventricular or supraventricular arrhythmias or the occurrence of heart transplantation, probably due to the small number of patients with gene-specific variants. Even so, two of three patients with PLN and LMNA variants had previously received ICDs and all patients with LMNA variants underwent heart transplantation (two after enrollment in the study), in line with the observation of a higher prevalence of heart transplantation and ventricular arrhythmias in carriers of LMNA and PLN variants.10,27

The presence of more than one variant in DCM patients is in agreement with the study by Haas et al.,10 in which more than 38% of patients had two variants and 13% had three or more. However, it is not known whether the presence of two or more variants in a single individual alters the phenotypic expression. Some authors argue that some low-penetrance variants might promote a disease phenotype only in the presence of other phenomic, genomic or other epigenetic factors.6 This could explain the occurrence of DCM in the family illustrated in Figure 2A, in whom the phenotype is only expressed at a young age in the individual with two variants of uncertain significance, and the presence of just one variant is not associated with the expression of DCM. Additionally, the presence of multiple variants could be related to a more severe phenotype and clinical course, as has been described for double heterozygotes for LMNA and TTN variant carriers with respect to age of end-stage HF and heart transplantation.29 In fact, in our study, two patients with multiple variants presented a particularly unfavorable clinical scenario, with end-stage HF and subsequent heart transplantation.

Determining whether a rare genetic variant is responsible for the disease is the most critical aspect of genetic testing. Interpreting whether a genetic variant is pathogenic is increasingly complex due to the large number of rare variants in the human genome and the advent of NGS, which enables analysis of extended gene panels. In many studies, relative loose frequency criteria have been used to ascertain variant pathogenicity.10,30 In particular, as reported by Akinrinade et al.,16 the European INHERITANCE project10 reported that 73% of the patients had ‘known’, ‘likely’, or ‘potential’ disease-causing variants (46% of the patients with ‘known’ disease-causing variants in cardiomyopathy or channelopathy genes, but only 16% of those with ‘known’ DCM variants), and variant interpretation was based mainly on variants reported in HGMD, which is known to contain variants classified as pathogenic, although some are increasingly being identified as benign polymorphisms. Similarly to our study, Akinrinade et al.16 used the ExAC database,29 which includes exomes of >60 000 individuals (10 times more than the Exome Sequencing Project database used by Haas et al.10) as the reference population database. Additionally, these authors randomly selected 10 variants from Haas et al.’s study classified as potential disease-causing (classes Ia-III) variants and ascertained their frequency in the ExAC database, finding that 80% of the variants were present in over nine (allele frequency >0.015%) subjects in the ExAC database with an average of 263 carriers (allele frequency >0.44%) in the database, making these variants highly unlikely to be disease-causing. In our study, we applied the criteria in the ACMG/AMP guidelines,22 which may be rather more stringent than the criteria applied to date, possibly leading to a larger proportion of variants being categorized as of uncertain significance, in order to reduce the number of variants being reported as ‘causative’ of disease without sufficient supporting evidence for that classification.

Recently, Walsh et al.32 compared sequence data of 7855 individuals with a clinical diagnosis of cardiomyopathy, including 559 with DCM, with 60 706 reference samples from the ExAC database. In clinical cohorts, variants in TTN were the most common, with the prevalence of rare variants in MYH7, LMNA, TNNT2 and TPM1 being modest, but significantly enriched compared with ExAC. However, there was modest or no significant excess variation in other commonly DCM-associated genes, like MYBPC3, with the authors arguing against a causative role for this gene in DCM pathogenesis, which challenges some widely held conceptions10,33 and our own results. Moreover, in genes supported by an excess of pathogenic and likely pathogenic variants, even variants of uncertain significance were seen to be enriched when compared to ExAC, suggesting that clinical laboratories may be overly conservative.32

All these findings underscore the need to continue exploration of genetic mechanisms and determinants of DCM and to establish standardized and reliable criteria for defining the pathogenicity of genetic variants. This will enable accurate interpretation of genetic testing, although it should be kept in mind that as new knowledge becomes available, the classification of previously identified variants may change.

This is the first study to address the genetic background of Portuguese patients with DCM and reflects the complexity and diversity of DCM genetics.

For better interpretation of the pathogenicity of the variants found and their causative role in DCM, careful phenotypic assessment of patients and their families and molecular cascade screening are crucial. Only with further advances in diagnostic techniques and improved understanding of biological mechanisms can we gain further insight into genotype-phenotype correlations and be able to use genetics for risk stratification, as well as to develop appropriate strategies for DCM prevention and management, ultimately improving the care of patients and families with DCM.

This work was supported by the Portuguese Government's Foundation for Science and Technology [PTDC/BIM-MEC/0650/2012].

The authors have no conflicts of interest to declare.