Accurate interpretation of genetic variants in sudden unexpected death in infancy by trio-targeted gene-sequencing panel analysis

In sudden unexpected death in infancy cases, postmortem genetic analysis with next-generation sequencing potentially can extract candidate genes associated with sudden death. However, it is difficult to accurately interpret the clinically significant genetic variants. The study aim was to conduct trio analysis of cases of sudden unexpected death in infancy and their parents to more accurately interpret the clinically significant disease-associated gene variants associated with cause of death. From the TruSight One panel targeting 4813 genes we extracted candidate genetic variants of 66 arrhythmia-, 63 inherited metabolic disease-, 81 mitochondrial disease-, and 6 salt-losing tubulopathy-related genes in 7 cases and determined if they were de novo or parental-derived variants. Thirty-four parental-derived variants and no de novo variants were found, but none appeared to be related to the cause of death. Using trio analysis and an in silico algorithm to analyze all 4813 genes, we identified OBSCN of compound heterozygous and HCCS of hemizygous variants as new candidate genetic variants related to cause of death. Genetic analysis of these deceased infants and their living parents can provide more accurate interpretation of the clinically significant genetic variants than previously possible and help confirm the cause of death.

Sudden unexpected death in infancy (SUDI) is a term that has been variably used to refer to all cases of sudden and unexpected deaths in infancy, including sudden infant death syndrome (SIDS) 1 . In 2015, SIDS was the third leading cause of death at age 0 in Japan, with a mortality rate of 9.5 per 100,000 population 2 . However, SUDI is thought to be more common because of the difficult distinction between SIDS and accidental asphyxia or natural diseases, such as arrhythmias and inherited metabolic disease 2 . Arrhythmia, inherited metabolic disease, mitochondrial disease, and salt-losing tubulopathy are linked to sudden unexpected death (SUD) [3][4][5][6][7][8][9][10][11][12][13] . Thus, SUDI is considered to be the severest form of various related diseases. The standard methods of determining the cause of death in SUDI cases include conventional autopsy (macroscopic autopsy and detailed microscopic examination 14 ), toxicology, biochemical tests, and recently, metabolic autopsy 15 . However, in most SUDI cases, these methods do not reveal the exact cause of death, hence the urgent need to develop novel diagnostic methods.
Recently, some studies have shown that genetic analysis using next-generation sequencing (NGS) of SUDI cases was effective in diagnosing arrhythmias and inherited metabolic disease 9,10 . On the other hand, it has been reported that some of the variants in arrhythmia-related genes found in SUD cases were not associated with cause of death 16,17 . In other words, an accurate interpretation of the association between variants in disease-related genes found in SUD cases and sudden death is important in determining cause of death. However, even if variants in arrhythmia-related genes are found, it is difficult to obtain findings that can be judged as arrhythmia from macroscopic, pathological, and biochemical examinations. Most conventional NGS studies were limited to genetic analysis of sudden death cases only. Recently, familial gene analysis of cardiac sudden death cases and their parents was shown to be effective for diagnosis of cause of death [18][19][20][21][22][23][24][25] . However, most family analyses have targeted genes for arrhythmia, so it would be useful to target not only those genes but also any other www.nature.com/scientificreports/ disease-related genes. Furthermore, few studies have reported family genetic analyses with NGS that focused on SUDI. Since SUDI is considered to be the severest form of various diseases, it would be useful to analyze the genes associated not only with arrhythmia but also with other diseases that can cause sudden death. Therefore, we thought that genetic analysis of SUDI cases and their parents would lead to a more accurate interpretation of the clinically significant genetic variants associated with cause of death. The study aim was to perform trio analysis of SUD infants and their parents to interpret the clinically significant disease-associated gene variants associated with the cause of death more accurately. In addition, we used trio analysis to extract novel candidate gene variants and examined whether these extracted variants were associated with cause of death.

Results
Case collection. Among the 24 autopsied cases, 17 could not be diagnosed using conventional methods.
From the 17 undiagnosed cases, 9 patients' parents gave consent for whole-genome sequencing. One of the nine cases was diagnosed with carnitine palmitoyltransferase II deficiency by metabolic autopsy 26 . In another case, we could not extract a sufficient amount of DNA for targeted gene-sequencing panel analysis. A total of seven SUDI cases (two males, five females; age range: 3 months-2 years) were selected (Fig. 1). The case 2 parents had a consanguineous marriage (Table 1).
Target sequence of sudden death cases. On average, ~ 17.5 million total reads were produced and ~ 11.9 million reads mapped to the targeted region in each sample. The mean coverage of the coding sequence was 101.5 ± 33.7 reads, with an overall average gene level coverage ≥ 20 reads of 93.0% ± 10.7%. After the filtering steps, 1897 variants in the 4813 target genes were identified, which corresponded to an average of 271 variants per case.
Detected variants aligned to the 216 target genes. Thirty-four (18 non-synonymous and 16 synonymous) detected variants were aligned with the 216 target genes. Eleven arrhythmia-, two cardiomyopathy-, three metabolic disease-, fifteen mitochondrial disease-, and three salt-losing tubulopathy-related genes were Arrhythmia-and cardiomyopathy-related gene variants. Most cases of genetic arrhythmia have autosomal dominant (AD) inheritance, and a familial history is not always present. Sporadic cases are mostly caused by de novo variant 9 . Case 1 had Asp85Asn-KCNE1, AD inheritance of a long QT syndrome (LQTS)related gene variant. This variant was reported to be associated with LQTS and suggested as a possible cause of death in SUDI 9,27 . The same variant was found in the female infant's healthy living mother. Case 4 had Arg160Trp-MYBPC3, AD inheritance of a cardiomyopathy-related gene variant suspected to be harmful by in silico analysis. This variant reportedly could not be ruled out as a pathogenic variant 28 . The female infant's father had the same variant but did not develop cardiomyopathy. Case 5 had Gln2730Arg-AKAP9, AD inheritance of an LQTS-related gene variant. AKAP9 variants have been reported to be associated with LQTS type 11 29 , but Gln2730Arg-AKAP9 has not previously been reported to be associated with LQTS type 11. The female infant's living father had the same variant. Five of seven cases had synonymous variants that were inherited from either of the parents. In silico analysis showed that all of them were benign.
Metabolic disease-, mitochondrial disease-, and salt-losing tubulopathy-related gene variants. Most inherited metabolic disease, mitochondrial disease, and salt-losing tubulopathy have autosomal recessive (AR) or X-chromosomal recessive inheritance. Homozygous amino acid changes or at least two heterozygous amino acid changes are necessary to cause these diseases 9,30-33 .
Case 1 had Leu21 = -SDHAF2, AD inheritance of a paraganglioma-related gene variant. TraP Score of Leu21 = -SDHAF2 was 0.089, which meant that the variant was benign. In addition, her mother had the same variant. The diseases associated with the other 20 gene variants are all forms of AR inheritance. Since the variants in the cases were all heterozygous, these diseases were unlikely to have developed and led to death.
Detected de novo, homozygous, compound heterozygous, and hemizygous variants aligned with all 4813 genes in TruSight One. A total of 12 de novo variants in 10 genes, 4 homozygous variants in 4 genes, compound heterozygous variants in 4 genes, and 4 hemizygous variants in 4 genes were found in < 0.1% of the Japanese population. All de novo variants were heterozygous (Table 4, Supplementary Table 2). As of August 2021, a search conducted on PubMed using these 20 gene names and the terms "sudden, " "infant, " and "death" as keywords returned no papers related to SUDI (Fig. 2).

Discussion
First, we examined the significance of variants in disease-associated genes that could cause sudden death by performing trio analysis of SUDI cases and the parents. Arrhythmias, inherited metabolic disease, mitochondrial disease, and salt-losing tubulopathy were considered as candidate diseases that could cause SUDI since these diseases were difficult to diagnose because there are few specific findings at autopsy. Arrhythmias and inherited metabolic disease have been reported to be associated with SUDI [3][4][5][6][7][8][9][10] . In this study, we adopted genes related to arrhythmia and inherited metabolic disease that we used in a previous study as candidate genes 9 . In particular, D85N_KCNE1 has been previously reported as a genetic variant associated with LQTS. In the present study, the same variant was also found in the infant's mother, which could suggest that this variant alone was not sufficient to cause LQTS.
Mitochondrial disease is a type of inherited metabolic disease that occurs in about 1 in 5000 births 34 . There have been reports of postmortem diagnosis of mitochondrial disease 11 . Ohtake et al. reported that about 9% of those diagnosed with mitochondrial disease were cases of SUDI 12 . In addition, we adopted this gene as a candidate because most mitochondrial diseases are caused by variants in human nuclear DNA 35 .
Salt-losing tubulopathy is a group of diseases, including Bartter syndrome and Gitelman syndrome, characterized by hypokalemic metabolic alkalosis as a common feature 36 . Electrolyte abnormalities cannot be established at general autopsy. In one case, Bartter syndrome was diagnosed by postmortem genetic testing 13 , and salt-losing tubulopathy-related genes were used as candidate genes.
After the filtering step, 34 variants were found aligned with the 31 target genes. All of these variants were heterozygous. Software Implemented Fault Tolerance (SIFT), PolyPhen2, and TraP Score revealed a total of nine harmful missense variants. Genetic analysis of this case alone would have determined that these variants were likely to cause abnormal protein function and were associated with cause of death. In particular, the KCNE1, MYBPC3, and AKAP9 genes that were associated with LQTS or cardiomyopathy would have been suspicious as genes associated with cause of death because of an AD mode of inheritance 27,37,38 . These were sporadic cases without any family history of sudden death, and none of the AD-inherited arrhythmia-related gene variants in these cases were de novo. In general, if the proband had an AD genetic disease, at least some of the relatives would have the same symptoms. Therefore, these disease-associated genetic variants alone would not have caused the SUDI, as the relatives, including the parents, had not experienced SUD, obviously. These family histories and trio analysis results suggest that some of the variants in SUDI cases could not be involved in the development of the disease, although the same variants were categorized as pathological by in silico analysis or by the results of experimentation. www.nature.com/scientificreports/ Second, candidate variants causing sudden infant death were extracted for all 4813 genes in the TruSight One panel. The incidence of SIDS in families is extremely low, and most cases are sporadic 39 . It is generally believed that rare variants are associated with rare diseases 40 , and diseases that can cause SUDI are considered to be one of the most severe of rare diseases. If SUDI cases with no family history of SUD manifested genetic variants that caused the sudden death, the variants were considered to be de novo. In addition, the variants of homozygotes or compound heterozygotes in SUDI cases and heterozygotes in their parents were extracted. Similarly, the variants of X-chromosome hemizygotes in SUD in male infancy and heterozygotes in their mothers were extracted.
We found eight de novo heterozygous missense variants in seven genes, three homozygous missense variants in three genes, eleven compound heterozygous missense variants in three genes, and two X-chromosome hemizygous missense variants in two genes (Table 4). These gene names and "sudden, " "death, " and "infant" were entered into PubMed as search terms. No papers described the association between these genes and SUDI. Therefore, some of these variants might be novel variants associated with cause of SUDI. Of these genes, we focused on two as biologically plausible candidate genes potentially associated with cause of death, OBSCN and HCCS.
Obscurin, which is encoded by the OBSCN gene, has an important role in the organization of sarcomeres during myofibril formation and the regular alignment of sarcoplasmic reticulum 41 . OBSCN variants may be monogenic causes of cardiomyopathy or contribute to the disease phenotype in concert with other variants 42 . In addition, inhibition of sarcomeric activity may cause arrhythmogenesis. It has also been reported that variants in Table 3. Results of known arrhythmia-, inherited metabolic disease-, mitochondrial disease-, and saltlosing tubulopathy-related variants. MTHFR: methylenetetrahydrofolate reductase, CPT: carnitine palmitoyltransferase, AD: autosomal dominant, AR: autosomal recessive.  43 . In addition, the association of the OBSCN variant with a case of sudden cardiac death in an 8-yearold girl has also been reported 44 . Case 4 had maternal Arg1060Gln-OBSCN and paternal Ser5880Asn-OBSCN by compound heterozygotes. We speculated that these two heterozygous variants could be involved in sudden death by compound heterozygotes. Holocytochrome c-type synthase, encoded by HCCS on the X chromosome, is located on the outer surface of the inner mitochondrial membrane and catalyzes covalent attachment of heme to both Cytc and Cytc1 45 . Cytc transfers electrons from electron-transfer complex III to complex IV in the mitochondria to promote ATP production. The HCCS variant at domain IV reduced expression of HCCS and also impaired release of Cytc 46 . These suggest that HCCS plays an important role in mitochondrial function. Mitochondrial dysfunction can lead to arrhythmia and sudden death 47 . In addition, it has been found that HCCS also has an important role in apoptosis and that the HCCS variant is associated with microphthalmia with linear skin defects syndrome, which is an X-linked male-lethal disorder 45,48 . These results suggest that HCCS is an essential protein for vital function. Case 7, a 6-month-old boy, had hemizygous Ala248Ser-HCCS in domain IV. The in silico algorithms predicted that it was deleterious and probably damaging, so we speculated that this variant was associated with

variants found in seven cases
Trio analysis and frequency in Japan less than 0.1%

Four hemizygous variants in four genes on X-chromosome (1) Detected variants aligned to the target 216 genes SUD analysis only
Thirty-four heterozygous variants 7 cases ( www.nature.com/scientificreports/ the cause of death. However, there is no report about the clinical significance of Ala248Ser-HCCS, so a functional analysis study is needed. Trio analysis was thought to be useful for both evaluation of disease-related gene variants that could cause sudden death and extraction of novel candidate genes for sudden death. Trio analysis revealed that some variants in existing candidate genes for cause of death might be insufficient to cause death because these variants were inherited from either of the parents. This means that the postmortem genetic analysis of only SUDI cases could have led to misdiagnosis and was considered insufficient to determine cause of death.
We extracted genes with de novo, homozygous, compound heterozygous, and hemizygous variants and reviewed previous reports that had reported an association between these genes and SUDI. There were no reports of associations between the genes found in this study and SUDI, which suggested that OBSCN and HCCS from these extracted genes are novel candidate genes associated with death caused by gene dysfunction or diseases related to genetic variants. In a large cohort study of whole-exome sequencing for SIDS, no variants were found to be significantly more prevalent in SIDS cases than in the general population 49 . Therefore, it may be difficult to identify gene variants that are candidates for cause of death by statistical analysis alone even when comparing with a large sample of unrelated healthy subjects. If a sudden death case has an ultra-rare variant, it could be possible to determine the variant with pathological significance by examining whether the parents also have the variant.
Trio-targeted gene-sequencing panel analysis has an advantage over conventional methods for determining the cause of death. Genetic analysis of not only SUD deceased infants but also living parents can be a more exact inquiry for cause of death by SUDI. An autopsy is an investigation of the cause of death that is performed directly from a corpse. However, a genetic analysis of the SUD infants and their parents can provide more accurate interpretation of genetic variants than previously possible and help confirm a diagnosis of the cause of death. Furthermore, trio analysis can be performed in daily medical practice. It is important that the family members of those who have died suddenly should undergo standardized testing, including genetic analysis [50][51][52] . We were careful to obtain informed consent from the parents before trio analysis. Additionally, if a trio analysis reveals a hereditary disease, it is possible to refer the patient for genetic counseling. Asymptomatic siblings can be provided with important genetic information of the deceased, which may enable detection of diseases before they develop. Furthermore, diagnosed individuals can receive appropriate treatment. Trio analysis data may eventually be utilized for child death reviews.
There were several study limitations that should be considered. Targeted gene-sequencing panel analysis was performed using a panel of 4813 genes; thus, targeted gene-sequencing panel analysis of genes not included in this panel and of the whole-genome sequence was not performed. TruSight One is a 4813-gene-sequencing panel that covers a wide range of known disease-associated genes, so it is helpful in clinical diagnosis 53 . However, we believe that whole-exome analysis and whole-genome analysis are necessary to search for unknown genetic diseases. Additionally, mitochondrial genome sequencing was not performed. In the 4813 genes analyzed in this study, only limited variants, such as synonymous, missense, insertions, and deletions, were examined, and copy number polymorphisms were not analyzed. In this study, we investigated the association between the cause of death in SUD cases and arrhythmia and cardiomyopathy-related gene variants extracted by trio analysis. However, since incomplete penetrance diseases were identified in which the parents were asymptomatic and the child had a severe phenotype 54 , further genetic studies including not only the parents, but also other relatives, are necessary.
In this study, we showed that trio analysis enabled more accurate interpretation of the clinically significant gene variants in SUD infants, which could help prevent misdiagnosis and extract novel genetic variants associated with SUDI. Trio analysis is a novel method that can be useful for determining cause of death because it also analyzes genetic information from living parents or relatives as well as from deceased victims. In addition, trio analysis may enable investigations of the cause of death even in hospitals. Trio analysis can identify unknown genetic variants that may be related to cause of death and help prevent SUDI in the future.

Methods
Study design and participants. The SUD cases of infants autopsied at ≤ 2 years of age were selected between April 2013 and March 2017. Informed consent for whole-genome sequencing and trio analysis were obtained from parents. A comprehensive forensic investigation, including a thorough examination of the death scene, a review of the clinical history, and performance of an autopsy that included macroscopic and microscopic examinations and a toxicology examination, were performed in all cases.
Extraction of genomic DNA and genetic analysis. As reported previously 9 , genomic DNA of SUDI deceased infants and their living parents was isolated from blood leukocytes and buccal mucosa, respectively, by using the QIAamp DNA Blood Mini Kit (Qiagen, Tokyo, Japan) in accordance with the manufacturer's standard methods. A TruSight One sequencing panel (Illumina, San Diego, CA, USA) targeting 4813 disease-associated genes was used, and the sequencing was performed on an Illumina MiSeq (Illumina) (Supplementary Text 1).
Extraction of de novo, homozygous, compound heterozygous, hemizygous variants of SUDI. After filtering steps, to extract < 0.1% of Japanese allele variant frequencies, the data from the Genome Cohort Study of Tohoku Medical Megabank Organization (ToMMo) (http:// ijgvd. megab ank. tohoku. ac. jp) and the Human Genetic Variation Database (HGVD) in the Human Genetic Variation Browser (http:// www. genome. med. kyoto-u. ac. jp/ SnpDB/ index. html) were referred to. Trio analysis was a comparative examination of genetic variants in SUDI deceased infants and their living parents. De novo variants, found in the SUD infants but not in their parents, were extracted by trio analysis. In the SUDI cases, homozygous variants or compound heterozygous variants that were heterozygous in their parents were extracted. In addition, in boy cases, hemizygous variants in X chromosome were extracted. Of these extracted variants, allele frequencies ≥ 0.1% in the Japanese subgroup in ToMMo and HGVD were excluded. Especially for the compound heterozygous variants, both allele frequencies of < 0.1% each were adopted. Sudden infant death candidate gene names and the terms "sudden, " "infant, " and "death" were combined and searched in PubMed.
Sanger sequencing. Sanger sequencing was performed to confirm the detected candidate variants (OBSCN and HCCS), as previously described. The polymerase chain reaction primers were designed by using Primer3 version 0.4.0 (https:// bioin fo. ut. ee/ prime r3-0. 4.0). (Supplementary Table 3.) Ethical approval. Written informed consent was obtained from the parents for the use of the samples in this study, which was approved by the ethics committees of the Nagasaki University Graduate School of Medicine (20170504-3, 20200801-2). This study was performed in accordance with the Declaration of Helsinki.