Molecular autopsy of sudden unexplained deaths reveals genetic predispositions for cardiac diseases among young forensic cases

Abstract

Aims

Coronary artery disease accounts for the majority of sudden cardiac deaths (SCD) in the older population whereas cardiomyopathies and arrhythmogenic abnormalities predominate in younger SCD victims (<35 years) with a significant genetic component. The elucidation of the pathogenetic cause of death might be relevant for the prevention of further deaths within affected families. Aim of this study was to determine the portion of underlying genetic heart diseases among unexplained putative SCD cases from a large German forensic department.

Methods and results

We included 10 forensic cases of sudden unexplained death (SUD) victims aged 19–40 years, who died by SCD due to forensic autopsy. DNA was analysed by next generation panel sequencing of 174 candidate genes for channelopathies and cardiomyopathies. Cardiological examinations, genetic counselling, and subsequent genetic testing were offered to all affected families. We identified within 1 year 10 cases of SUD among 172 forensic cases. Evidence for a genetic disposition was found in 8 of 10 (80%) cases, with pathogenic mutations in 3 and variants of uncertain significance in 5 of SCD cases. Subsequent selective screening of family members revealed two additional mutation carriers.

Conclusion

The study provides strong evidence that molecular genetics improves the post mortem diagnosis of fatal genetic heart diseases among SUD victims. Molecular genetics should be integrated in forensic and pathological routine practice.

Introduction

Sudden cardiac death (SCD) refers to an unexpected death from a cardiac failure in an individual with or without pre-existing heart disease within 1 h after the onset of symptoms.¹ SCD accounts annually for about 180 000–450 000 cases in the USA and around 100 000 in Germany. The incidence rate of SCD ranges between 50 and 100 per 100 000 in the general population of industrial countries.^2,3 Coronary artery disease accounts for the majority of SCD in the elderly population whereas cardiomyopathies and arrhythmogenic abnormalities predominate in younger SCD victims <35 years with incidence rates of 0.6–6.2/100 000 patient years.^4,5 SCD in the young (<35 years, SCDY) is estimated to have a significant genetic component.⁶

Since most inherited cardiac diseases show an autosomal dominant pattern of inheritance, first-degree relatives of SCD victims have a 50% risk for the same predisposition. The elucidation of the pathogenetic risk factor for SCD might be relevant for the prevention of further deaths within affected families. Sudden cardiac death, whether as first or last sentinel event, indicates the common final endpoint of a variety of cardiac diseases. Of note, among SCD victims ∼10–15% of non-ischemic cardiomyopathies are found, whereas ∼5% are related to arrhythmogenic cardiac diseases with an isolated electrical disorder.⁷

Currently, 2–43% of putative SCDs of young people are classified as sudden unexplained death (SUD) after autopsy.^8,9 However, for different reasons only a small but not precisely known percentage of SUD cases in the young is transferred for autopsy at all.^10,11 In addition, blood or tissue samples for DNA analysis are not routinely stored in pathological and forensic practice and limit molecular analysis. Currently, there are no guidelines available for the storage of DNA derived from SCD in the German forensics and pathology departments. Although in recent studies previous methods of molecular autopsy revealed pathogenic mutations in up to 35% of SUD cases the aetiology of SCD in adults ≤40 years remains sparse.¹²

The genetics of inherited cardiac diseases is complicated due to the high number of potentially affected genes and a high proportion of family-specific mutations. At present the HGMD database accounts for >12 000 sequence variants in more than 150 genes associated with cardiac diseases (http://www.hgmd.cf.ac.uk). In addition, the knowledge of the predictive value of particular DNA sequence variants and the associated disease course is still incomplete, which hamper genetic counselling of affected families. Recently, novel sequencing technologies (NGS) enable the high throughput and ultra-deep analysis of genes, which are related to cardiac diseases. However, up to now only limited data are available on SCD victims investigated by NGS.^13–16

This study was performed to determine the portion of underlying genetic predispositions for heart diseases with NGS among unexplained putative young SCD cases and investigated in a large forensic department.

Methods

Study cohort

In a prospective study we analysed SUD victims aged 1–40 years who were transferred to the Institute of Forensic Medicine at the University Hospital Hamburg-Eppendorf within 2013. The German Federal State of Hamburg has a population of about 1.7 million people and counts on average around 18 000 deaths annually. All cases with unnatural or unknown cause of death have to be investigated in the Institute of Forensic Medicine; additionally all cases of death with intervention of rescue teams are sent to the Institute of Forensic Medicine to complete the death certificate after post mortem inspection of the body.

The stepwise forensic approach to identify SUD cases is shown in Figure 1. The inclusion criteria were (i) SUD at the age between 1 and 40 years (ii) no plausible cause of death by autopsy or post mortem inspection of the corps, respectively, (iii) negative test results by toxicological screening (barbiturates, cannabinoids, cocaine, ecstasy, ethanol, methadone, opiate and tricyclic anti-depressants), (iv) cases with structural or non-structural cardiac abnormalities, and (v) informed consent of the relatives for molecular autopsy. We also included two cases (cases #1 and #10) with cardiac infarction since they had signs of a previously manifested structural heart disease. However, we excluded SCD patients with an isolated extended acute myocardial infarction or coronary artery spasm, respectively (two cases with SCD in Figure 2).

All relatives gave informed consent to the molecular analysis of the SUD cases. Based on these criteria we included 10 SUD victims aged 19–40 years who were suspected to have died from SCD.

The study was approved by the local ethics committee of the Ruhr-University Bochum located at the Heart and Diabetes Center NRW in Bad Oeynhausen, Germany (vote No. 41/2013). The study conformed with the Declaration of Helsinki.

Clinical data

In all cases included in this study we investigated the medical history by interviewing close relatives and contacting physicians if possible. In 4 out of 10 cases pre mortem 12-lead electrocardiograms (ECG) and additional cardiological data were available from the clinical files (see Supplementary material online). We also collected data by interviews on the circumstances of death, pre-existing cardiac symptoms and syncopes, respectively. Additionally, we analysed the pedigrees of each patient for reported cardiac diseases or further SUDs within the family.

The forensic procedure

Autopsies were performed according to the guidelines for autopsy investigation of SCD.¹⁷ During external examination or autopsy 2–3 mL blood from the femoral artery was collected for isolation of genomic DNA. All samples were isolated within 48 h post mortem and stored at −20°C until further processing. Full autopsy included standardized dissection of the heart, its histological examination and toxicological screening.

Molecular genetics using next generation sequencing

DNA was isolated from whole blood and prepared for panel sequencing on the MiSeq^® System (Illumina). A total of 174 genes implicated in the pathogenesis of channelopathies and cardiomyopathies (see Supplementary material online, Table S1A) were investigated at the Heart and Diabetes Center in Bad Oeynhausen, Germany. For details of panel sequencing analysis see Supplementary material online.

Sequence variants were classified according to ACMG standards¹⁸ and compared with entries in international databases [i.e. HGMD (http://www.hgmd.cf.ac.uk.)]. Variants were examined by in silico predictive algorithms MutationTaster, PolyPhen-2, SIFT, PROVEAN, and Condel (see Supplementary material online, 5A–E).

All families were offered genetic counselling. Cardiological examination was recommended to first-degree relatives—independent of their participation in this study (s. supplements for an overview). Disease-related variants were Sanger sequenced in selected first-degree family members when appropriate after genetic counselling.

Results

Forensic and clinical findings

For forensic and clinical findings see Table 1, Figure 2, and Supplementary material online.

Genetic findings

Case #1 died aged 40 years as a driver in his truck during sleep. Autopsy revealed an enlarged heart with myocardial hypertrophy and extensive lipomatosis cordis. The heart weight was in the upper percentile of this study. His medical history revealed previous cardiological examinations with findings of a mild aortic valve insufficiency. Stress echocardiography could not be completed for fitness reasons. The patient was obese and a heavy smoker. NGS analysis revealed the heterozygous TTN variant p.E3359K which we classified as variant of unknown significance (VUS). Additionally a variant in BAG3, coding for Bcl2-associated anthanogene and associated with dilated cardiomyopathy (DCM) and myofibrillar myopathy was found. As the bioinformatics on the pathogenicity of the variant p.H248R are not consistent and the variant is not known from the literature it was classified as VUS.

Case #2 died at rest after symptoms of chest pain. He died aged 33 years witnessed by his pregnant wife from acute right heart failure. During autopsy we found extensive fat tissue on the right ventricle and left ventricular hypertrophy. We could not identify relevant genetic variants.

Case #3 died aged 19 years at rest and unwitnessed. The family rejected autopsy. The pedigree analysis revealed a family history for an unspecified heart disease and a consanguine relationship of her parents. Her father died from SCD aged 29 years. She was diagnosed aged 4 years to have a long QT-syndrome-2 (LQTS-2) and received the beta blocker Atenolol. However, this information was provided by the relatives after genotyping of case #3 during genetic counselling. In addition, sequencing data were available only for a limited number of genes. Panel analysis confirmed the heterozygous mutation KCNH2 p.N629S, which encodes a subunit of the cardiac voltage-dependent potassium channel and carries mutations in LQTS. The mutation p.N629S, which has been described with LQTS-2 in several families, is located in the pore region of the channel. In vitro analyses revealed aberrant transport and posttranslational modifications of the protein caused by this mutation. A 12 lead ECG was performed 16 months before death (see Supplementary material online) revealing a corrected QT-time (QTc) of 531 ms. We identified in the cadherin gene DSG2 the missense mutation p.Q255K. DSG2 encodes desmoglein-2, a protein of the cardiac desmosome. Mutations in DSG2 are associated with arrhythmogenic right ventricular cardiomyopathy. As there is no information on the pathogenicity of this variant from the literature and computer algorithms uniformly predict tolerability, we classified this variant as likely benign. The mother of case #3 was wildtype for both variants. A cardiac disease was excluded by echocardiography and ECG.

Case #4 died in bed after resuscitation witnessed by the partner. The patient complained from left arm pain ∼6 months before death. During autopsy enlarged left atria, enlarged hypertrophied left ventricle and mitral valve insufficiency was found. The heart weight was 590 g and the BMI was 27. Macroscopically, signs of inflammation of the endo- and/or myocardium were suspected without a coronary artery disease. Evaluation of the 12 lead ECG 5 months before death did not reveal any evidence of cardiac abnormalities. We found the mutation MYPN p.P47L, coding for myopalladin, a multifunctional protein localized at the sarcomeric Z- and I-bands and in the nucleus in cardiac and skeletal myocytes. Mutations in MYPN have previously been reported in subjects with DCM, hypertrophic cardiomyopathy (HCM), and restrictive cardiomyopathy. During the study the second of three brothers died aged 31 years from SCD in bed. He was transferred for autopsy. His heart weight was 435 g, the BMI was 20.3 and the ventricles were dilated. Myocarditis was excluded by immuno-histology. The patient did not carry the MYPN p.P46L variant. We classified the variant p.P47L therefore as likely benign (ACMG-2) since it is not characterized in the literature, not predicted as disease relevant by bioinformatics and was not found in the second deceased family member. The cardiological check up by MRI, echo, exercise tests and Holter ECGs of the parents and the remaining brother did not reveal any signs of a cardiomyopathy or arrhythmogenic disease.

The 23 year old case #5 died unwitnessed probably at rest. She was found about 12 h post mortem. The cardiac ventricles were normal but the mitral valve revealed a prolapse syndrome. During autopsy arrhythmogenic cardiac failure was suspected. The medical history revealed presyncopes ∼5 weeks before death; however, an ECG was not performed. We found the heterozygous mutation CACNA1C p.R858H, which encodes a subunit of the cardiac voltage-dependent calcium channel. The mutation has been described with LQTS-8 in several unrelated families, previously.¹⁹ In the affected patients the mutation was associated with prolonged QT intervals, syncopes, and SCD, respectively. In vitro analysis revealed an influence of the mutation on the Ca²⁺-Influx. Although the pedigree showed suspected related cases on the mother's side, genotyping of the relatives revealed that the father and her brother were also carriers. Both received an electrophysiological examination in a local university hospital. The brother had a QTc of 470 ms and the father a QTc of 471 ms. The father is treated by beta blockers whereas the brother of the deceased rejected any treatment. The mother of case #5 was also examined but was without symptoms (QTc 416 ms). The ajmaline challenge for both mutation carriers was negative. In addition to the CACNA1C mutation we identified two variants in PRDM16 and KCNE3, respectively. Mutations in PRDM16 are associated with DCM and left ventricular non-compaction cardiomyopathy. The variant p.G1042R was classified as VUS since bioinformatics tools revealed ambiguous results and no data were found in the literature. KCNE3 encodes a voltage-gated potassium channel and its mutations are listed with Brugada syndrome (BS) and familial atrial fibrillation (fAF). The variant p.D86N was classified as VUS since it is not known from the literature and the bioinformatic classification is contradictory.

The female case #6 died unwitnessed in the washroom aged 22 years. Histological examination of the myocardium revealed irregular muscle fibres especially in the right ventricle. We found in ABCC9 the heterozygous variant p.D617N. ABCC9 codes for SUR2A, the regulatory subunit of the cardiac K_ATP-channel. Mutations in ABCC9 are listed with DCM and BS. After genetic counselling further obviously not diseased family members (I:3 and II:6) were genotyped for this variant to clarify the pathogenetic impact and inheritance of the ABCC9 variant. Since we did not find evidence for a pathogenetic impact within this family we classified this variant as VUS. Additionally, we identified a variant in CACNB2 encoding the beta-subunit of the voltage-dependent calcium channel. Mutations in CACNB2 are listed with arrhythmias, BS, and ventricular fibrillation. The variant was classified as VUS since no data on the variant p.K267R in the literature were available and the bioinformatic predictions are contradictory.

Patient #7 was a mother of two children and died aged 31 years in the washroom. She was found about 4 h post mortem and had a history of relapsing syncopes. She was treated in hospital 5 months before death. A 12 lead ECG was available, however, without specific abnormalities (see Supplementary material online). Echocardiography performed 18 months before death did not reveal any structural heart disease. In accordance with the echocardiography during autopsy the heart did not show macroscopic signs of a structural disease. We found variants in DSG2, AKAP9, and SCN5A in this patient. There is no information on the pathogenicity of these variants in the scientific literature. However, DSG2 is associated with arrhythmogenic cardiomyopathies, whereas AKAP9 coding for the A-kinase anchor protein 9 and SCN5A are associated with arrhythmogenic heart disease (http://www.hgmd.cf.ac.uk). SCN5A encodes the cardiac voltage-gated sodium channel. Mutations in SCN5A are associated with DCM, BS, and LQTS.

Case #8 died aged 31 years at rest and unwitnessed. She was found about 24 h after death. During autopsy the heart weight was 550 g and the left ventricle was dilated. We identified the variant p.H1795Y in LAMA4 coding for the extracellular matrix protein laminin. The variant was classified as a VUS. Additionally, the novel variant PRDM16 p.H969R was identified, which is predicted as pathogenic by bioinformatics and but classified as VUS according to ACMG criteria.

Patient #9 died in a public indoor swimming pool after resuscitation. He was 36 years and had a BMI of 30 mainly due to intensive physical training. The suspected intake of anabolic steroids could not be confirmed by GC/LC-mass-spectrometry of the urine. During autopsy there were no clear signs of drowning as a primary cause of death. His heart weight was 450 g and his left ventricle was hypertrophied (left ventricular free wall thickness 20 mm). The patient had a history of hypertension. Due to autopsy findings the children (III-3 and III-4) of the deceased are under cardiological supervision. The panel analysis revealed two variants: The myopalladin variant MYPN p.T1137I is characterized heterogeneously as disease relevant or tolerated by bioinformatics tools. The SCN5A variant p.T427N is not listed in the literature and the bioinformatic predictions are contradictory so that it was classified as VUS. However, the variant p.E428K is listed with fAF (http://www.hgmd.cf.ac.uk).

Case #10 died unwitnessed at the age of 40 years at rest and after unsuccessful resuscitation. About 3 h before death he had relapsing syncopes. The right ventricular diameter of the free wall was 3 mm. The primary cause of death was acute myocardial infarction due to multivessel disease. However, due to structural myocardial abnormalities case #10 was included in this study. Two variants in his genome were found: DSP p.L1995S coding for desmoplakin and a cryptic splice site in the gene coding for troponin T (TNNT2). We classified DSP p.L1995S as VUS and the splice defect in TNNT2 as pathogenic. After genetic counselling the relatives II-4 and III-4 rejected genetic testing but were genotyped for scientific purposes. The mother (II-4) of case #10 was heterozygous for the TNNT2 variant, whereas his brother (III-4) was heterozygous for the DSP variant (for references of Results see Supplementary material online, 5F–R). The relatives II-4 and III-4 were examined after death of case #10 but were without clinical signs of a cardiac disease so far.

For a summary of genetic findings see Table 2 and Supplementary material online.

Discussion

Recent studies estimate that up to 75% of SCDY cases are caused by inherited heart diseases.²⁰

The yield of underlying pathogenic mutations associated with SUD in this study using NGS was 30% (3 out of 10). This represents a higher detection rate than previously reported (see Supplementary material online, 5S–X). We also found in 50% of cases (5 out of 10) variants with an unknown significance (ACMG-3) for the SCD of their carriers. They might be regarded as future candidate genes. However, their pathogenetic evidence and clinical impact could not be conclusively shown in this study.

NGS is able to identify large numbers of genes simultaneously. However, rare sequence variants suspected to be related to SCD might be detected more easily by NGS.² On the other side NGS might provide also more rare variants, which makes the interpretation of sequence data more challenging.²¹

We identified in three cases pathogenetic mutations (ACMG-5). In further five forensic cases we identified disease-related variants (ACMG-3, VUS) which might be involved in the SUD. Of note, eight of the identified variants were novel mutations influencing their pathogenetic classification. In eight cases we identified more than one variant, which complicates currently genetic counselling of the relatives.

Up to now, there are only few studies^13–16 reporting results of NGS for post mortem investigations. Brion et al.¹³ analysed mutations in 28 genes associated with arrhythmic cardiac diseases in 53 cases of SUD and sudden infant death syndrome using SOLID™ sequencing.¹³ In 38% of the cases they found putative and in 21% likely pathogenic variants. Farrugia et al.¹⁴ sequenced 23 genes associated with channelopathies in 16 cases of SUD <35 years using Ion Torrent technology. They found four likely pathogenic variants by a new variant scoring system.¹⁴ Hertz et al.¹⁵ identified in 29% of cases mutations with likely functional effects when investigating using Illumina technology 100 genes associated with cardiomyopathies or channelopathies in SUD cases aged <50 years, respectively.¹⁵ Sanchez-Morelo et al.¹⁶ found 30% pathogenic and potentially pathogenic variants probably responsible for SCD by targeted resequencing of 55 genes associated with channelopathies and cardiomyopathies of deceased younger than 55 years.¹⁶ Interestingly, the latter results correspond to our data.

In all previous NGS studies genotyping cases of SUD different variant classification systems were used and the inclusion criteria were different. In addition, the selection of genes analysed in the panel was specified to identify structural and/or arrhythmogenic cardiac diseases, respectively. Thus, these studies might not be directly comparable. We applied in this study a 174 gene panel covering all genes currently known to be associated with structural or non-structural heart diseases. We further applied the recently published ACMG classification for genetic variants (http://www.hgmd.cf.ac.uk) which allows the standardized evaluation of their current pathogenetic evidence.

Although the number of SUD cases investigated here remains limited, the cohort of this study represents all SUD cases within 1 year in the German federal state of Hamburg with 1.8 million residents. To the best of our knowledge, this is the first prospective study combining forensic autopsy, clinical data evaluation, and molecular autopsy of putative SCD victims using NGS of 174 genes with subsequent selective genetic examination of first-degree family members.

We identified in the SUD cases 14 variants, which were class 3 according to ACMG criteria. ACMG classification of these variants reflects current scientific knowledge. A re-evaluation of their pathogenetic evidence might therefore lead to changes in their classification in future. ACMG-3 variants were no used for genetic family screenings, since their pathogenetic evidence is up to now unknown. However, in some families they were used for co-segregation analysis after genetic counselling of the relatives.

Genetic testing in general is relatively expensive and might not be covered in deceased cases by health insurances in most countries. Up to now, post mortem genetics is currently practised in different countries heterogeneously and might be used for diagnostic purposes in some European states like Germany in near future.^6,15 However, it appears to be challenging to increase the number of post mortem investigations in young SUD cases even when cardiogenetic counselling is offered to their relatives.¹¹

Although this study was performed to elucidate the impact of molecular autopsy in forensic cases for scientific purposes, we recommended all relatives of the deceased cardiological examination and offered genetic counselling. However, not all relatives agreed to further cardiological examination and genetic counselling but gave consent to this study. Moreover, some of the relatives lived in distant areas of Germany which did not allow the cardiological examination in a core institution.

The awareness of the genetic burden in SUD cases is not well developed in forensic practice or pathology. Since the availability of intact genomic DNA for NGS from autopsy is still a bottle neck for the post mortem genetic analysis in SCDY, hopefully future educational programmes will improve the availability of DNA for genetic counselling of the relatives²² to realize the concept ‘mortui vivos docent’.

Limitations of the study

NGS might lead to a considerable yield of mutations with unknown pathogenetic evidence. Mutations in class ACMG-3 were not further investigated on their pathogenetic relevance, respectively. Thus, the classification of the mutations reflects current scientific knowledge assuming mono-genetic inheritance. Cardiological examination of first-degree relatives was not performed in a core facility because family members would have to travel long distances. In addition, in some families further genetic testing or recommended cardiological examination was rejected by relatives for personal reasons, respectively.

Conclusion

In a prospective study we showed that genetic testing by NGS in young SUD cases leads to a considerable yield of mutations which might allow genetic counselling of relatives and might support cardiological examinations. The combined approach of molecular autopsy and cardiological examinations might be of clinical relevance for the prevention of further SCD within families. At present the collection of DNA for post mortem analysis on mutations associated with SCD in the young is not routine practice in forensics or pathology, respectively. Since techniques are now available for high throughput sequencing and SCD has high psychosocial impact for relatives, the storage of DNA should definitely be integrated into autopsy practice.¹⁷

Supplementary material

Supplementary material is available at Europace online.

Funding

This work was supported by the Erich & Hanna Klessmann-Foundation, Gütersloh, Germany; and the Deutsche Forschungsgemeinschaft (MI 1146/2-1 to H.M.).

Conflict of interest: none declared.

Acknowledgements

First of all, we thank the families for their support! We are grateful especially for their patience and commitment despite the loss of beloved family members. We thank Désirée Gerdes, Herz- und Diabeteszentrum, Bad Oeynhausen, for her excellent technical assistance. We also thank all members of the Institut für Rechtsmedizin, Hamburg, for their kind support and all doctors in the field who supported this study by providing clinically relevant information on the deceased or their family members. We thank Benjamin Reiter for helpful discussion on the SCD victims.

References

Author notes