Objective Sudden unexpected death in epilepsy (SUDEP) is a leading cause of epilepsy-related mortality in young adults. It has been suggested that SUDEP may kill over 20 000 people with epilepsy in China yearly. The aetiology of SUDEP is unclear. Little is known about candidate genes for SUDEP in people of Chinese origin as most studies have ascertained this in Caucasians. No candidate genes for SUDEP in Chinese people have been identified.
Methods We performed whole exome sequencing (WES) in DNA samples collected from five incident cases of SUDEP identified in a large epilepsy cohort in rural China. We filtered rare variants identified from these cases as well as screened for SUDEP, epilepsy, heart disease or respiratory disease-related genes from previous published reports and compared them with publicly available data, living epilepsy controls and ethnicity-match non-epilepsy controls, to identify potential candidate genes for SUDEP.
Results After the filtering process, the five cases carried 168 qualified mutations in 167 genes. Among these genetic anomalies, we identified rare variants in SCN5A (1/5:20% in our cases), KIF6 (1/5:20% in our cases) and TBX18 (1/5:20% in our cases) which were absent in 330 living epilepsy control alleles from the same original cohort and 320 ethnicity-match non-epilepsy control alleles.
Conclusions These three genes were previously related to heart disease, providing support to the hypothesis that underlying heart disorder may be a driver of SUDEP risk.
Statistics from Altmetric.com
Sudden unexpected death in epilepsy (SUDEP) is a leading cause of epilepsy-related mortality in young adults in Western Countries.1 The incidence of all suspected (probable and possible) SUDEP in China was reported at 2.34 (95% CI 1.36 to 3.77) per 1000 person-years, suggesting that yearly over 20 000 people die as result of SUDEP.2
The mechanism of SUDEP remains unclear. Proposed mechanisms include seizure-related respiratory, cardiac and autonomic dysfunction, suggesting heterogeneity.3 Increased genome-wide polygenic burden in SUDEP cases provided some evidence for genetic susceptibility.4 Studies in models and humans suggest that some specific genetic background may predispose individuals to potentially fatal cardiorespiratory dysfunctions.5 To date, several genetic features mainly involving ion channels coexpressed in the heart and brain have been described mainly in Caucasians. No candidate genes for SUDEP in Chinese people have been identified.
Previous reports on candidate genes for SUDEP were mainly from retrospective cases studies6–9 or from follow-up study with small sample-size.10 11 Adequately powered cohort studies are needed to corroborate these results and identify new genetic candidates. The low incidence of SUDEP and the time gap between the collection of DNA samples and the occurrence of SUDEP has made it difficult to accomplish such studies.
We performed whole exome sequencing (WES) in DNA samples collected from five probable SUDEP cases identified in a large community-based cohort in China. We filtered rare variants identified from our cases as well as screened for SUDEP, epilepsy, heart disease or respiratory disease-related genes from previous published reports and compared them with publicly available data, living epilepsy controls and ethnicity-match non-epilepsy controls, to identify potential candidate genes for SUDEP in the Chinese population.
Material and methods
Study cohort, follow-up and confirmation of SUDEP cases
Between January 2010 and December 2011, 1562 (median age 38 years; 58% males) individuals with epilepsy from rural areas in the Chinese provinces of Henan, Shanxi and Ningxia took part in the project ‘Validation of Clinical Assessment Tools for Population Genetic Studies of Epilepsy’. Fifty-six per cent of the participants had focal seizures. Sixty per cent were on antiseizure medication (ASM) monotherapy and only 20% were in 1-year remission when enrolled.
Two follow-up visits were conducted, the first during March 2013 to October 2014, and the second between October 2015 and March 2016. For those who had died, death certificates were collected from the appropriate office. A specifically designed Verbal Autopsy Questionnaire was used2 and the full datasets were assessed by a multidisciplinary expert panel. Fifteen death cases were attributed to SUDEP and 13 of them were attributed as probable SUDEP. Detailed report on methods and attribution of death was previously published.2
Blood draw and DNA extraction at baseline
At baseline, 2 mL peripheral blood of each participant was collected in EDTA anticoagulant vacutainer and all samples were sent by cold chain logistics to Huashan Hospital, Shanghai, China. Genomic DNA was extracted from whole blood. Extracted DNA samples were stored at −80°C until use.
Selection of cases for sequencing
Inclusion criteria: (1) death previously ascertained as probable SUDEP;2 (2) age at death between 15 and 39 years; (3) DNA sample available and having passed an integrity assessment (on a 0.8% agarose gel).
There were 13 cases of probable SUDEP of whom seven died between ages of 15–39 years but only 5 DNA samples passed the integrity test.
WES and bioinformatics analysis
The DNA library was constructed by fragmenting the genomic DNA; a Qubit 2.0 Fluorometer was used to determine the concentration of the library; the pooled capture library was quantified by Qubit (Invitrogen) and Bioanalyzer (Agilent) and sequenced on an Illumina HiSeq 2500 using a paired end, 150 nucleotides in length run mode. All sequencing processes were controlled by data collection software according to the IlluminaX User Guide. We aligned the paired-end reads to the reference human genome (hg38) using the third-party software BWA (Burrows–Wheeler Alignment, V.5.9). The average mapping ratio was as high as 99%. The Flagstat tool was used to assess the mapping information. We then analysed the distribution of each sample’s reads in the target region and the enrichment of reads in the genome. Single nucleotide variations (SNVs) were then processed using the GATK UnifiedGenotyper (GenomeAnalysisTK- 3.1–1). Last, we annotated the mutations using ANNOVAR software (GenomeAnalysisTK-3.1–1).
Qualified variants were identified through the following filtering process: (1) low-quality reads were removed (marked ‘Pass’ in filter and Genotype Quality >20); (2) SNVs located out of exonic regions and splicing sites were removed; (3) synonymous SNVs were removed; (4) common SNVs (mutation frequency >0.01 in 1000 genomes project were removed; (5) SNVs marked other than ‘damaging’ in SIFT (http://sift.jcvi.org/) and PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2/) prediction were removed. Variants which passed the filtering process were regard as qualified variants. The genes which those qualified variants were in were regarded as qualified genes. Function of each qualified genes was checked to screen candidate genes for SUDEP, including genes previously reported to be related to epilepsy, cardiac diseases, respiratory diseases and SUDEP (ANK2, DEPDC5, KCNE1, KCNE2, KCNH2, KCNQ1, RYR2, CACNA1A, SCN1A, SCN1B, SCN2A and SCN5A).7 10 12–14 Potential candidate variants were further genotyped in 330 living control with epilepsy alleles (age and gender matched from the original cohort) and 320 ethnicity-match healthy control alleles by restriction digestion. The baseline clinical characteristics of epilepsy controls and cases for WES are provided in online supplementary file 1. The flowchart is provided in figure 1.
Standard protocol approvals, registrations and patient consents
The original study was approved by the joint Chinese University of Hong Kong-New Territories East Cluster Research Ethics Committee and the institutional review board of the Beijing Neurosurgical Institute in China. The follow-up exercise and genetic analysis were approved by the Medical Ethics Committee of Fudan University affiliated Huashan Hospital, Shanghai, China. Written informed consent had been obtained from all participants of the original study or when applicable assent was obtained from legally acceptable guardians.
The raw data that support the findings of this study are available from the corresponding authors on reasonable request.
Baseline characteristics of the five cases are summarised in table 1. Three of them were male. All of them had convulsive seizures in the year of death. The median age of onset of epilepsy was 14 years. Four of them had had EEG recording and two had had CT/MRI scan.
Detailed death-related information is shown in table 2. The average age of death was 25.6 years. None was in remission in the year prior to death. Three of the deaths were witnessed and two died after a witnessed seizure.
We obtained more than 86% uniquely mapped reads and more than 99% good matched reads. The average sequencing depth in the exome region was approximately more than 120× (table 3). Target SNV distribution and target SNV function are presented in online supplementary file 2. In summary, we obtained high quality WES data.
After the filtering process, five subjects carried 168 qualified mutations in 167 genes. Among these genetic anomalies, two affected genes, SCN5A and CACNA1A, were previously related to SUDEP. We also identified six variants in genes (NEB, SCN9A, GUF1, TLR4, TRPM2 and PLA2G6) associated with epilepsy and another three variants in genes (KIF6, TBX18, CYSLTR2) associated with asthma, arrhythmia or heart disease. Detailed information of these candidate genes and their function is summarised in table 4.
Subject #1 suffered from convulsive seizures. Prior to treatment with valproic acid, he had 3–5 seizures a month, which dropped to about one a month on starting treatment. He carried qualified variants in CYSLTR2 and KIF6. The rare R64W mutation in KIF6 identified in this subject was absent in 330 living epilepsy controls and 320 ethnicity-match non-epilepsy controls. The T410C variant in CYSLTR2 in this subject was absent in epilepsy controls and only one such variant was found in non-epilepsy control alleles.
Subject #2 had frequent convulsive seizures despite effective treatment with several ASMs. She had a febrile seizure aged 2 years and her sister has epilepsy. She carried rare mutations in SCN9A. The T2132C variant in SCN9A was found in 1 of 330 epilepsy control alleles and two of 320 non-epilepsy control alleles.
Subject #3 had frequent focal seizures of temporal origin with occasional progression to bilateral tonic-clonic seizure. She died during a seizure at age 28. She carried a rare R1139W mutation in SCN5A, which was located in a topological domain (https://www.uniprot.org), classified as a variant of uncertain significance and probably related to Long QT syndrome 3 or Brugada syndrome in the ClinVar database (https://www.ncbi.nlm.nih.gov/clinvar/). Her ECG showed a QT of 0.39 s and a QTc of 0.4 s. This variant in SCN5A was absent in epilepsy controls and in non-epilepsy controls. She also carried qualified variants in CACNA1A, GUF1.
Subject #4 had up to 10 focal seizures with progression to bilateral tonic clonic seizures a month which continued despite treatment with carbamazepine and phenobarbital. He had mild learning disability and carried variants in TBX18 and NEB. The G821A variant in TBX18 in this subject was absent in 330 living epilepsy controls and 320 ethnicity-match non-epilepsy controls. The T13661C variant in NEB was identified in two of the epilepsy control alleles and three of the non-epilepsy control alleles.
Subject #5 had untreated convulsive seizures. He carried qualified variant in PLA2G6. This C2255G variant was identified in three of the epilepsy control alleles and two of non-epilepsy control alleles.
We used WES to identify potential candidate gene in incident SUDEP cases and we found a new, rare, harmful variant in SCN5A, absent in epilepsy and non-epilepsy controls. Other variants in SCN5A were previously reported in SUDEP cases,7 12 and potentially predisposing to malignant cardiac arrhythmia.15–17 We also identified two rare qualified variants in KIF6 and TBX16 (previously associated with heart diseases), also absent in epilepsy and non-epilepsy controls. These findings highlighted the role of potential heart problem in Chinese victims of SUDEP.
Genetic are said to play a key role in SUDEP. A study involving the genetic testing in a large cohort of individuals with epilepsy and cardiac conduction disorders detected putative pathogenic disease-causing mutations in genes encoding cardiac ion channel in 24% of unrelated individuals with epilepsy.18 Another study investigated the long-term outcome of 17 infants with migrating focal seizures with KCNT1 mutations. They found that an extracerebral involvement, like arteriovenous fistula, dilated cardiomyopathy, may be involved in the high prevalence of SUDEP in these cases.19 To date, several studies had attempted to identify candidate genes from postmortem DNA samples. These studies identified several potential candidate variants in genes, coding for proteins with key function on cell excitability and electrophysiology of the heart. Retrospective studies may have selection bias which may impact external validity. Genetic investigation in SUDEP cases from large cohort studies would help in verifying previous results and identifying new potential candidate variants. The incidence of SUDEP is 1.2/1000 person-years,20 implying that only very large DNA sample library can support genetic studies in incident SUDEP cases. Our DNA samples were collected prospectively at baseline in a large community-based cohort in China seven to 8 years ago and thus devoid of major selection bias.
In the follow-up of the cohort, we identified 15 sudden death cases that fulfil the criteria for SUDEP (13 probable cases and 2 possible cases). We selected SUDEP cases who died in the age of 15–39, as this is the group at the highest risk.20 All had convulsive seizures and were not in remission, which are also markers of high risk.
In a UK study, 40% of subjects who died suddenly and might otherwise have had a diagnosis of SUDEP, were shown at postmortem to have probable cardiovascular causes of death.21 We did not have postmortem data but young age at death and events surrounding death decreased the possibility of cardiovascular or other underlying causes. With the multidisciplinary panel, paying particular attention to those in whom the reported or certified cause of death was cardiac, cerebral vascular, unknown or sudden, we were able to identify ‘high-quality’ probable SUDEP cases for WES.
The SCN5A gene encodes the alpha subunit of the main cardiac sodium channel Nav1.5. SCN5A variants have been causatively associated with Brugada syndrome, long QT syndrome, cardiac conduction system dysfunction, dilated cardiomyopathy and so on.22 Some variants in SCN5A have been related to sudden death.16 23 24 Ala572Asp, Pro1090Leu, Pro2006Ala variants in SCN5A 7 and a missense mutation R523C12 were found in SUDEP cases previously.
We also identified a new rare missense mutation in SCN5A in a young female. This missense mutation R1193W is rare, and the frequency of this variant was 0.04% in 1000 Genomes Project, and it was absent in epilepsy control alleles from the same original cohort and non-epilepsy control alleles. It was predicted to be pathogenic by the use of the prediction programme PolyPhen and SIFT and has not been previously reported. This variant may be related to Brugada syndrome, Long QT syndrome 3 in the ClinVar database. The resting ECG record of this individual showed that her QT was 0.39 s and QTc was 0.4 s which was within the normal range, but it does not exclude potential causality. People with some mutations in SCN5A are at higher risk of sudden death, even those with a normal QT interval.25
Only few genetic case series studies in SUDEP were reported previously and table 5 summarised potential rare variants identified. We also identified two new rare qualified variants, which was absent in epilepsy control alleles from the same original cohort and non-epilepsy control alleles, in genes that had not yet been reported in SUDEP cases. We identified rare variants in TBX18 and KIF6 in two of our cases. TBX18 is attractive target for biopace making as it is important in sinoatrial node development and so has the potential to have a broad effect on cardiac tissue phenotype.26 TBX18 had been associated with the generation of cardiac pacemaker cells and sick sinus syndrome.27 28 TBX18 could have an important role in restoring pacemaker function in human sick sinus syndrome (the most common bradycardia in humans and may increase the risk of sudden death).28 KIF6 is ubiquitously expressed in coronary arteries and other vascular tissue and previously related with coronary heart disease. Several other variations in KIF6 were reported to be related to coronary epicardial endothelial dysfunction in males in a case-control study.29 These variants in SCN5A, TBX18 and KIF6 suggest that heart disorder may play a key role in the mechanism of some SUDEP cases.
We also identified rare qualified variants in CACNA1A, NEB, SCN9A, GUF1, TLR4, TRPM2 and PLA2G6 which were previously reported in people with epilepsy. CACNA1A is calcium voltage-gated channel gene mainly expressed in the brain. A study of 14 SUDEP cases using a panel target resequencing found that 3 of them had variants in CACNA1A.10 SCN9A is a sodium voltage-gated channel gene which had a wide expression. A rare SCN9A mutation was previously associated with febrile seizures plus and Dravet syndrome.30 A mutation in NEB was previously reported in a Korean family with intellectual disability, epilepsy and early-childhood-onset generalised muscle weakness.31 Mutation in PLA2G6 was reported in a Chinese pedigree with familial cortical myoclonic tremor with epilepsy.32 We identified a variant in CYSLTR2. Rare variant in CYSLTR2 was previously reported in people with asthma.33 One report identified asthma as a risk factor of SUDEP.34
Other new variants we identified were in genes (NEB, GUF1, TLR4, TRPM2 and PLA2G6) which have not been previously be associated with SUDEP. Further studies of these variants may provide new insights into other potential mechanism of SUDEP in people of Chinese descent.
Our study has limitations. First, postmortem examinations are rare in China. We used several methods to improve the diagnosis of SUDEP without postmortem examinations as previously described,2 and ascertainment bias may still have existed. We selected five SUDEP cases who were at a high risk of SUDEP for WES to ensure a better quality. Probable SUDEP cases or other no-definite SUDEP cases were also included for testing and analysis in previous reports.6 7 9–11 13 35 Second, we only screened candidate variants previously associated with SUDEP, epilepsy, cardiac diseases and respiratory diseases as these are thought to be the main players in SUDEP pathophysiology.1 As a result, we may have ignored some variants whose function was unknown or could induce sudden death via mechanisms other than heart or respiratory disorder. Third, the function and pathogenicity of the candidate variants we identified were still unknown and we could only predicate the function of the variants from previous publications, SIFT and PolyPhen scores and the ClinVar database. Further studies were needed to work out the effect of these variants on the function of genes. Fourth, we did not genotype these variants in the surviving family members and in patients with heart disease. We cannot investigate whether the variants identified in our study were de novo variants, and we cannot assess the contribution of these variants in heart disease and whether they were unique in SUDEP cases. We could try to genotype these variants in their relatives and patients with heart diseases in further studies. Last, our sample size was very small; however, the number of incident SUDEP cases are unlikely to be large in a community-based study.
SUDEP is the most tragic outcome of epilepsy. We investigated the genetic background in SUDEP cases in a Chinese cohort. We identified a rare variant in SCN5A which may have a role in the occurrence of SUDEP. We had also identified several new potential candidate genes for SUDEP. Our result could help to form a better understanding of genetic deficit and how they contribute to SUDEP, especially in Chinese population where the large data gap exists.
The authors are grateful to Professor Sanjay Sisodiya for critically reviewing the manuscript.
Contributors YG contributed with concept and design, analysis and interpretation of data, drafting of the manuscript, revision of the manuscript and critical revision of the manuscript for important intellectual content. DD contributed with DNA sample collection, obtaining the necessary research grants, concept and design, drafting of the manuscript, revision of the manuscript and critical revision of the manuscript for important intellectual content and study supervisor. GZ contributed with DNA sample collection, interpretation of data and revision of manuscript. PK contributed with DNA sample collection, interpretation of data and critical revision of the manuscript for important intellectual content. WW contributed with DNA sample collection, interpretation of data and revision of manuscript. ZH contributed with concept and design, study supervising, drafting of the manuscript, revision of the manuscript and critical revision of the manuscript for important intellectual content. JWS contributed with DNA sample collection, obtaining the necessary research grants, concept and design, drafting of the manuscript, revision of the manuscript and critical revision of the manuscript for important intellectual content.
Funding This study was funded by Shanghai Municipal Science and Technology Major Project (2018SHZDZX01), Key Research Project of the Chinese Ministry of Science and Technology (2016YFC0904400), a NIH/NINDS grant (1R21NS069223-01) and National Natural Science Foundation of China (81271443). JWS is based at UCLH/UCL Biomedical Research Centre, which receives a proportion of funding from the UK Department of Health's NIHR Research Centres funding scheme. He receives support from the Dr. Marvin Weil Epilepsy Research Fund and UK Epilepsy Society. PK is supported by a Medical Research Future Fund Practitioner Fellowship.
Competing interests JWS has received research funding from Eisai, and UCB, research support and personal fees from UCB, GW and Zogenix outside the submitted work. All other authors have no disclosures to report.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available on reasonable request.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.