Natural and Undetermined Sudden Death: Value of Post-Mortem Genetic Investigation

Background Sudden unexplained death may be the first manifestation of an unknown inherited cardiac disease. Current genetic technologies may enable the unraveling of an etiology and the identification of relatives at risk. The aim of our study was to define the etiology of natural deaths, younger than 50 years of age, and to investigate whether genetic defects associated with cardiac diseases could provide a potential etiology for the unexplained cases. Methods and Findings Our cohort included a total of 789 consecutive cases (77.19% males) <50 years old (average 38.6±12.2 years old) who died suddenly from non-violent causes. A comprehensive autopsy was performed according to current forensic guidelines. During autopsy a cause of death was identified in most cases (81.1%), mainly due to cardiac alterations (56.87%). In unexplained cases, genetic analysis of the main genes associated with sudden cardiac death was performed using Next Generation Sequencing technology. Genetic analysis was performed in suspected inherited diseases (cardiomyopathy) and in unexplained death, with identification of potentially pathogenic variants in nearly 50% and 40% of samples, respectively. Conclusions Cardiac disease is the most important cause of sudden death, especially after the age of 40. Close to 10% of cases may remain unexplained after a complete autopsy investigation. Molecular autopsy may provide an explanation for a significant part of these unexplained cases. Identification of genetic variations enables genetic counseling and undertaking of preventive measures in relatives at risk.


Introduction
Natural death defines the death primarily attributed to an illness or an internal malfunction of the body, and not directly influenced by external forces. The forensic pathologists can straightforwardly identify the cause of natural death when macroscopic investigations are conclusive [1]. However, when a macroscopic cause is not evident, the final identification of causality can become tedious and complicated. Despite comprehensive macroscopic, microscopic as well as toxicological investigation, around 5%-10% of cases will remain unexplained and will be classified as sudden unexpected deaths (SUD), often defined in the report as death from a supposed arrhythmia [2,3]. In the young population, this percentage may increase up to 30%-50% [4][5][6]. Even if the cause of death remains unanswered after a thorough forensic investigation, the legal work is usually concluded. However, from the medical standpoint, an unidentified etiology conveys dangerous clinical implications; these unexplained deaths may be caused by an inherited cardiac disease, which potentially leaves family members at risk.
In a simplistic classification, deaths caused by cardiac genetic alterations may affect two different disease groups, channelopathies and cardiomyopathies [7]. It is estimated that 10% to 25% of SUD in the adult, and up to one-third in infantile and juvenile SUD, may be explained by cardiac channelopathies [8][9][10][11]. These channelopathies include mainly Long QT syndrome (LQTS), Short QT syndrome (SQTS), Catecholaminergic Polymorphic Ventricular Tachycardia (CPVT), and Brugada syndrome (BrS) [12]. In addition, pathogenic variations in genes encoding structural proteins are responsible for cardiomyopathies (Hypertrophic Cardiomyopathy (HCM), Dilated Cardiomyopathy (DCM), and Arrhythmogenic Cardiomyopathy (AC), among others). These cardiomyopathies will usually present anatomo-morphological changes in the cardiac tissue, which can be diagnosed at autopsy [13], but recent reports have suggested that in infants they could also be potentially responsible for sudden death in the structurally normal heart [14,15].
Because SUD may be the first manifestation of an unknown inherited cardiac disease, the use of genetic testing, the so-called molecular autopsy, could be determinant in the discovery of causality, in the identification of genetic carriers in family members, and in the further adoption of preventive strategies [16,17]. Despite that current forensic guidelines recommend molecular autopsy as part of routine protocol in SUD cases, this is seldom performed [18,19]. This molecular investigation has been mainly limited to research projects, and usually constrained to the analysis of the most prevalent genes associated with channelopathies (KCNQ1, KCNH2, SCN5A and RYR2), leaving several potential candidate genes untested [20,21]. With the advent of high-throughput genetic technologies, Next Generation Sequencing (NGS), massive genetic sequencing has become available [22]. Recent reports have shown that NGS analysis could become an important asset in post-mortem examination [23][24][25][26][27]. To date, only one comprehensive study has been performed to prove the value of genetic testing in natural death [28].
In the present work we have addressed this issue by performing a prospective full epidemiological analysis of sudden death in a correlative cohort of SUD victims younger than 50 years of age, referred for forensic investigation due to out of hospital natural death. The goal was to define the etiology of natural death in the young, and to investigate whether genetic defects could contribute to this event. To perform the genetic analysis we have taken advantage of a custom-made resequencing panel. By including molecular diagnostic strategies, the ultimate goal of this work has been to develop a decision algorithm to better refine the forensic investigation, to assess the value of this powerful diagnostic tool in detecting a potential etiology, and to define which families would benefit from further clinical and genetic investigation.

Methods
Our project was initiated in 2012, in collaboration with Institut de Medicina Legal i Ciències Forenses de Catalunya (IMLCFC). The IMLCFC oversees and concentrates all SUD cases, which require forensic investigation. We have focused the project in those cases investigated by the pathologists in the Catalonia area (population of 7.5 million people).

Forensic analysis
The study was approved by the ethics committee of our Hospital, and follows the Helsinki II declaration. Our inclusion criteria were victims of sudden death, from natural cause, younger than 50 years of age. A complete autopsy examination was performed according to current international regulations [1,18]. When the macroscopic autopsy was labelled as negative, the forensic pathologists performed complete histological and toxicological investigation, and collected a blood sample for genetic investigation. We excluded those cases in which the autopsy was labelled as violent death, including death from drug overdose.

DNA sample
Genomic DNA was extracted with Chemagic MSM I from post-mortem whole blood (Chemagic human blood). DNA was checked in order to assure quality (Absorbance 260/280:260/230 should be a minimum 1.8: 2.2 respectively), and was quantified before processing with the NGS strategy. Spectrophotometric measurements were performed to assess quality ratios of absorbance; DNA concentration was determined by fluorometry (Qubit, Life Technologies). DNA integrity was assessed on a 0.8% agarose gel.

NGS sample preparation
The DNA was fragmented (Bioruptor, Diagenode). Library preparation was performed according to the manufacturer's instructions (SureSelect XT Custom 0.5-2.9Mb library, Agilent Technologies, Inc). After capture, indexed libraries were sequenced in six-sample pools per cartridge. Paired-end sequencing process was developed on MiSeq System (Illumina) using 2x150 bp reads length.

Assessment of pathogenicity
The rare variants (MAF < 1%) were classified according recent ACMG guidelines [31], following the criteria: Likely/probably benign variants (PBV): • Variants already described in any of databases, with all in-silico models predicted neutrality.
Variants of uncertain/unknown significance (VUS): • Novel variants and all in-silico models predicted neutrality or differed between predictions.
• Variants already described in any of databases, and in-silico models differed between predictions.
Likely/probably pathogenic variants (PPV): • Likely pathogenic variants reported to be disease-causing but where the author has indicated that there may be some degree of doubt, or subsequent evidence has come to light in the literature.
• Radical variants (insertions, deletions or premature stop codons) • Splice site variants between ± 5 nucleotides, and all in-silico models consulted predicted pathogenicity.
• Novel variants with all in-silico models predicting pathogenicity.
Disease causing mutations (DM): • Variants already reported to be disease-causing.

Results
This is a three-year prospective study that was started on February 2012. We have collected a total of 789 consecutive cases -609 males (77.19%) and 180 females (22.81%)-. The range of age is from 0 to 50 years of age (average 38.6±12.2 years old). The average age is 39.3±11.2 years in males and 36.2±14.9 years in females. In order to classify the cases, we have divided the cohort in groups of 10 years (0-10, 11-20, 21-30, 31-40, 41-50 years old).

Prevalence of natural death according to age
In our cohort, most cases were between 41 and 50 years of age (467 out of 789, 59.19%). A total of 190 cases (24.08%) were between 31-40 years old and 132 cases (16.73%) were below age 30. Regarding gender differences, the number of males was higher in all ranges of age, showing most differences after age 30, with males nearing 80% of cases (Fig 1).

Context in which the natural death took place
Deaths were classified depending whether they took place during stress/exercise, during sleep, or during routine daily activities. Information about the context of death was provided in 532 cases. The majority of deaths occurred during routine daily activities (376 cases, 70.68%), 98 occurred during sleep (18.42%) and 58 during exercise (10.90%). Deaths during sleep and during exercise were more common before the age of 20 (Fig 2).

Etiology of death
Concerning the cause of death, the forensic pathologist directly determined a conclusive cause of death after macroscopic evaluation (positive macroscopic autopsy) in 506 cases (64.13%), while a yet inconclusive autopsy (negative macroscopic autopsy) was reported in the remaining cases. Most macroscopically positive cases were males (383 cases, 75.69%). Regarding negative autopsy cases, most of these cases were also males (226 cases, 79.86%).

Negative macroscopic autopsy. Microscopic analyses
The macroscopic autopsy was not able to detect the cause of death in 283 cases (35.87%) and these were labeled as macroscopically negative. To identify a potential cause of death, these cases were further investigated with histological analysis. This investigation was able to refine the potential cause of death into the following subgroups: 1) Cardiac, which included coronary disease in 98 cases (34.63%) (presence of thrombus, of myocardial infarction or of severe coronary stenosis >75%), and 36 (12.72%) potentially cardiac inherited cases, by histological identification of cardiomyopathy; and, 2) unexplained cases, which included 149 (52.65%) cases with microscopic findings showing no histological alterations). Among these 149 cases, there were 23 cases (15.44%) in which death occurred before the first year of age, thus they were labeled as sudden infant death syndrome (SIDS).

Natural death according to context and final autopsy results
With further inclusion of histological analysis, 364 out of 789 cases (46%) were definitively labeled as deaths from a cardiac origin. This percentage is underrepresented, as it did not include the 149 negative cases, some of whom presumably died also from cardiac causes. Stress/exercise related death was more frequent in cardiac cases (51.72%), while deaths during daily activity were more prevalent in vascular (20.23%), as well as pulmonary (12.84%) etiologies (Fig 3).

Molecular autopsy
In addition to defining the etiology of natural death in our cohort, we wanted to assess whether the use of genetics could improve diagnosis ascertainment, and could better define which, if any, family members should undergo clinical/genetic evaluation. Thus, we focused our efforts in those forensic cases with potentially inherited disease or with negative microscopic autopsy:

Potentially inherited subgroup
This study was limited to cases previously considered macroscopically negative by the forensic pathologist but identified after histological analysis. A total of 32 samples showing histological alterations associated with cardiomyopathies (10 DCM, 19 HCM, 1 AC, and 2 fibrosis)( Fig 5) were screened by NGS method. The genetic screening identified a total of 62 rare variants in 25 out of 32 samples (78.13%). Twelve variants (19.35%) were novel. All cases carried at least one variant in genes codifying for structural proteins. However, 10 cases carried at least one additional rare variant in genes encoding proteins associated with ion channels or associated proteins. According to our classification criteria, 2 variants (3.23%) were considered PBV, 39 (62.90%) VUS, 12 (19.35%) PPV, and 9 (14.52%) DM (Table 1).

Unexplained sudden death subgroup
The analysis was performed in 119 samples with negative results after macroscopic and histological studies.
All samples were analyzed using NGS technology except in 24 samples in which the screening was limited to Sanger sequencing due to low DNA quality. The Sanger sequencing method identified at least one rare variant in 6 out of 24 (25%) samples. On the other hand, with the use of NGS technology we identified 76 out of 95 (80%) samples carrying at least one rare variant. Concerning NGS samples, 41 of them (43.2%) carried at least one rare variant in genes encoding proteins associated with ion channels, and 64 (67.3%) carried at least one rare variant in genes codifying for structural proteins. Overall, we detected 49(41.2%) samples carrying PPV and/or DM. The genetic screening identified a total of 197 rare variants, 6 detected by Sanger method and 191 by NGS method. Thirty-six (18.27%) were novel variants. Our criteria classified 16 (8.12%) as PBV, 100 (50.76%) as VUS, 60 (30.46%) as PPV and 21 (10.66%) as DM (Table 2).   Concerning ranges of age, the youngest cohort (between 0-10 years of age) included 28 samples (14 males -50%-, and 14 females -50%-). A total of 11 samples were screened by Sanger method, and 17 by NGS. The genetic screening by both technologies identified a total of 37 rare variants, 3 detected by Sanger method and 34 by NGS method. Four variants (10.81%) were novel. We detected 16 samples carrying at least one rare genetic variant (57.14%). Six samples (21.43%) carried variants in genes associated with ion channels and 12 samples (42.86%) carried variants in genes codifying for structural proteins. We identified 10 (35.71%) samples carrying PPV and/or DM. Finally, pathogenicity criteria classified 6 variants (16.22%) as PBV, 14 (37.84%) as VUS, 16 (43.24%) as PPV and 1 (2.7%) as DM.
The cohort between 11-21 years of age included 7 samples (5 males -71.43%-, and 2 females -28.57%-). A total of 2 samples were screened by Sanger method, 5 by NGS. The genetic screening identified a total of 10 rare variants by NGS method. The Sanger screening did not identify positive samples carrying rare variants. Of identified variants, 2 (20%) were novel. NGS method identified 4 samples (80%) carrying at least one rare variant, 3 (42.86%) carried variants in genes associated with ion channels, and 4 (57.14%) in genes codifying for structural proteins. Our classification criteria for all samples screened identified 2 (28.57%) samples carrying PPV and/or DM.

Discussion
In this prospective cohort we have methodically examined the etiology of natural death by performing a comprehensive investigation that includes a thorough autopsy examination and the inclusion of an extensive molecular autopsy. We have analyzed a total of 789 SD cases younger than 50 years of age. By concentrating all our cases in a same institution, a same autopsy protocol was followed, according to international forensic recommendations [1,18].
Regarding epidemiological data, our data are in concordance with previous studies [32]. Thus we observed similar results in victims' mean age (39.3 years old), gender (77.19% males), progressive increase in SD prevalence from age 0 to 50, and predominance of male deaths in all ranges of age.
The primary cause of SD was cardiac -81.1%-, similar to other cohorts. In 56.87% of cases, death was labeled from coronary artery disease (CAD), either from evidence of myocardial infarction or from the identification of severe coronary stenosis, which induced ischemiarelated arrhythmias [28]. This percentage is not much different than the one reported in other studies, which attributed nearly 40% of deaths to ischemic heart disease [33]. However, in other reports the percentage of CAD neared 80%. The difference in our percentage is probably due to the inclusion of population only younger than 50 years of age in our cohort. The percentage of CAD-related death increased with age and was highest in population above 40 years old, as it was expected. In 2015, Vassalini et al reported the presence of ischaemic alterations in 18.5% of a cohort of patients aged less than 40 years, while other studies have underlined higher incidences of coronary related SCD, with percentages ranging between 48 and 73% [34][35][36][37][38]. These discrepancies can be explained by the different diagnostic criteria used in sample selection and age cut-off.
Regarding the situation of death, most CAD-related SD occurred in population older than 40 years of age, and during daily activity or stress/exercise. In the young population, most cardiomyopathy-related cases died during stress/exercise. These results are in concordance with other published cohorts and well-known data about exercise being a significant risk factor for cardiac death.
Macroscopic autopsy was able to define the cause of death in most cases above age 30. However, in cases below age 30, a negative macroscopic autopsy was the most common scenario. After adding histological analysis, an additional 17% of cases were labeled as of cardiac origin, cardiomyopathy or coronary artery disease, increasing the total percentage of cardiac origin to 46%, with CAD totaling 28.5%. In the young population, the main cause of death was from cardiomyopathies and in younger than 10 years of age, inherited cardiac diseases were the primary cause, also similar to published reports [39][40][41][42]. Nearly 19% of our cases remained as a negative autopsy, in concordance with a recent publication of Vassalini et al [43]. In other reports this percentage ranges from 5% to 40%, probably related to the study cohort as well as autopsy protocols [33,44].
In 2015, the Swiss Society of Legal Medicine created a multidisciplinary working group, clinical and molecular geneticists together with cardiologists, in the hope of harmonizing the approach to the investigation of SCD. The key points of the recommendations were (1.) the realization of a forensic autopsy procedure for all SCD victims under 40 years of age (molecular autopsy or post-mortem genetic testing), (2.) the collection and storage of adequate samples for genetic testing, (3.) communication with the families, and (4.) a multidisciplinary approach including cardiac genetic counseling [19]. Though, despite these recommendations and the increasing availability of NGS technology, it is yet seldom performed in most forensic centers as part of the autopsy. A current matter of argue is who should pay the genetic test. In our opinion, public health system should assume the cost of these cardiac genetic analysis due to it is well established that genetic test in cases without conclusive cause of death help to identify the cause of death in a high number of cases. In addition, these genetic tests also help clinicians in identification of genetic carriers in family members, doing prevention of SCD in relatives at risk.
Similarly, according to recent cardiology guidelines [45], the use of molecular autopsy should be considered in the event of an unexplained sudden cardiac death with a suspicion of inherited disease. It remains unclear when to suspect an inherited cardiac disease in most cases, as death is often the first clinical manifestation in the families. According to these same guidelines, the identification of a pathogenic variant associated with long QT syndrome or with CPVT is diagnostic of the disease. Thus, taking these data into account, the use of molecular autopsy appears mandatory in order to attempt to provide a diagnosis, which may benefit the identification of family members at risk.
For that reason, the aim of our work was to evaluate whether molecular autopsy could increase the identification of a potential etiology of death. Thus, we performed genetic investigation in all cases classified as microscopic cardiomyopathy as well as in those that remained unexplained after macro and microscopic autopsy. Thirty-two samples were reclassified as cardiomyopathies after identification of positive histological alterations. Out of these 32 samples, 19 showed histological changes consistent with a diagnosis of HCM. These results agree with several studies that report that HCM is the most prevalent cardiomyopathy associated with SCD. After genetic analysis, 33.9% of cases carried at least one PPV or DM variant, most variants remaining as VUS.
In addition, we genetically analyzed 119 samples classified as negative autopsy cases. Genetic analysis identified 41.19% of cases carrying at least one PPV or DM variant, remaining most variants as VUS. This percentage is in concordance with other genetic studies performed with NGS panels in autopsy samples [46,47]. In post-mortem studies in which only a few genes were analyzed frequencies of detection differ between 11 to 26% [21,28,[48][49][50][51]. Our percentage is higher due to the largest number of analyzed genes. Of 119 cases classified as unexplained death, around 40% carry a potentially pathogenic variant. Concretely, in our cohort, in the population younger than 31 years of age, the percentage of potentially pathogenic rare variants was 40.4%. Some reports have established that between 10% to 25% of SUD in the adult, and up to one-third in infantile and juvenile SUD, may be explained by cardiac channelopathies [8][9][10][11]. In most of these studies, the analysis was limited to the main genes associated with channelopathies. Our higher percentage may be due to a comprehensive genetic analysis including both genes associated with channelopathies and genes associated with cardiomyopathies, recently associated with arrhythmic pathologies without any structural alteration [52,53]. In concordance with our results, recent studies performed in post-mortem samples using NGS technology showed percentages of rare variants potentially pathogenic in 30%-40% of samples analyzed [24][25][26][54][55][56][57][58].
Out of 119 samples, we identified 21 DM variants in 16 cases (13.44%). Of these, 5 variants were potentially responsible for LQT, and 2 for BrS. Recent guidelines recommend the use of post-mortem genetic testing in cases with clinical evidence suggesting a diagnosis of LQTS or CPVT [29,45]. Therefore, and according to the guidelines, a diagnosis was reached as a cause of death. In addition, 51 PPV were identified in 38 cases (31.92%). While their pathogenic role cannot be fully defined, the potential for an inherited disease makes it essential to further investigate the family members for segregation. In a recent report, Bagnall et al performed a NGS analysis in a post-mortem cohort of 490 samples died suddenly between 1-35 years old [28]. They identify nearly 35% of genetic variants classified as VUS and family segregation clarify the role of nearly 15% of cases concluding that autopsy investigation combined with genetic testing and family screening is the best way to identify a conclusive cause of death in cases died suddenly.

Overlapping diseases/genes
In our study we have identified samples classified as cardiomyopathies but carrying rare variants in genes encoding ion channels and/or associated proteins. Similarly, we have also identified samples classified as negative autopsy (potential channelopathy) but carrying rare variants in genes encoding structural proteins. Several studies have reported the potential pathophysiological mechanisms linking both entities in common genes. Thus, for example, PKP2 (encoding plakophilin-2), is the main gene associated with AC and has been reported playing a pathogenic role in BrS [52,53] despite additional studies in large cohorts should be performed to clarify this point [59]. In addition, alterations in SCN5A (encoding the sodium channel), the main gene associated with BrS, have been reported in 1-2% cases of DCM [60], and even AC [61]. In concordance with similar results from recent studies [62], this could suggest that a malignant arrhythmia could appear in early stage before a structural alteration is developed. However, further studies in larger cohorts should be performed to prove or refute this hypothesis.

Compound/Multigenic variants
In almost 50% of samples, more than one rare genetic variant was identified, even in the same gene in some cases. Most of these rare genetic variants are at present of unknown significance. However, whether they have a role in the final risk of sudden cardiac death is unknown. Their potential effect as genetic modifiers (either detrimental or protective) of the phenotype is well accepted, but larger and more comprehensive studies will be needed to obtain conclusive data.

Limitations
The first and main limitation already mentioned above is the lack of family members in order to perform a clinical-genetic segregation. The family segregation is crucial to clarify the role of identified genetic variants in each case/family. In addition, functional studies could help elucidate the pathogenic role of the variants in arrhythmogenesis but in vitro evidence of channel dysfunction associated with specific variants may not necessarily directly translate into a clinical phenotype in the complex biological environment of the human cardiovascular system. Finally, cases without any identified genetic variation could carry a defect in other genes not included in our NGS custom-panel.

Conclusions
In a prospective cohort of 789 cases of natural death, younger than 50 years of age, we show that cardiac alterations are the most common cause of death, concretely coronary artery disease. While forensic investigation can determine the cause of death in most cases, nearly 19% of cases remain unanswered after a thorough autopsy investigation. The use of NGS genetic analysis has been advocated as an important complement to the investigation of death, and the incorporation of molecular autopsy in current guidelines attest to its value, according to the experts. The molecular autopsy may help identify the cause of death in a large percentage of cases remaining as negative after autopsy. In our cases without conclusive cause of death, we identified nearly 35% of PPV and nearly 15% of DM variants therefore reaching a conclusive diagnosis according to the guidelines. In our study we show that genetic analysis should be performed when there is a suspicion for an inherited cardiac disease after macroscopic or histological analysis, and when all tests have excluded a cause of death. Of notice, in SUD victims older than 30, it is important to exclude coronary disease by histology before proceeding with molecular autopsy, as this is the most common cause of death in that population. Our data show that before age 30, despite histological analysis should be performed in order to identify any microscopic alteration, the genetic analysis should be undertaken right away because the percentage of microscopic cardiac alterations is very low. In addition, even when histology identifies any microscopic cardiac alteration, the genetic results are a helpful complement of alterations identified in order to conclude cause of death. Consequently, we have proposed a simple forensic recommendation about molecular autopsy in Sudden Death cases ( Fig  6). The evaluation of relatives should include an appropriate genetic counseling and will allow the implementation of preventive measures to the relatives at risk to prevent new cases of SCD.