Exploratory DNA methylation analysis in post-mortem heart tissue of sudden unexplained death Authors: Charlotte Sutter, Cordula Haas, Peter K. Bode, Jacqueline Neubauer, Jeppe Dyrberg Andersen Published: 2024 ABSTRACT BackgroundSudden unexplained death (SUD) is a devastating event in the young. Despite efforts to identify causal genetic variants, many cases remain unexplained after genetic screening. This study aimed to investigate an alternative potential contributor to SUD by studying the human methylome using the MethylationEPIC v2.0 BeadChip kit in heart tissue from SUD cases. The genome-wide methylation results of the SUD cases were compared to the results of a control cohort. The SUD cases were divided into three main groups based on their autopsy reports, heart morphology and histopathology (primaryN: macroscopically and histologically normal heart; primaryCM: macroscopically or histologically abnormal heart, suspected cardiomyopathies; and secondary: myocardial damage due to other underlying conditions). The main focus of this study was to identify differentially methylated regions (DMRs) between the case groups and the control cohort.ResultsWe identified DMRs for both the primaryN and primaryCM groups, whereas the secondary group yielded no such results. In the primaryN cases, the corresponding genes for each DMR led to the identification of genes with common biological pathways. Some were associated with heart morphology (e.g. heart outflow tract morphogenesis or trabecular morphogenesis), but the majority belonged to more general cellular regulatory pathways (e.g. transcription coactivator activity, long non-coding RNAs, etc.). Although no common pathways were found for the primaryCM group, some common regulatory molecular functions were identified, such as p53 binding and transcription coactivator activity.ConclusionsOur study is the first to investigate the whole human methylome in heart tissue of SUD cases. We propose that there are observable differences in the methylation patterns of the case groups that may have contributed to SUD. Still, further studies are required to improve our understanding of the impact of methylation levels on SUD risk and to pinpoint methylation-based screening opportunities for SUD relatives. Background Sudden cardiac death (SCD) is a devastating event, especially in young people. The first symptom of SCD is often death itself, leaving neither the affected person nor their relatives time to prepare for these extraordinary circumstances [1, 2]. By definition, a person has died of SCD if death occurred less than one hour after onset of symptoms and was witnessed. If unwitnessed, SCD is assumed if the person has been seen alive and well less than 24 h before death [2–5]. One subgroup of SCD is termed sudden unexplained death (SUD) [2], which is commonly defined as deaths occurring in people older than 1 year. In these cases, potentially inherited cardiac conditions are often suspected to be the main cause of death [6, 7]. SUD cases may be termed sudden arrhythmic death syndrome (SADS) if they have a negative pathological and toxicological assessment [2]. While sudden death in infants until the age of 1 year is classified as sudden infant death syndrome (SIDS) [8], sudden death in older people (> 50 years of age) often occurs due to chronic degenerative diseases like coronary artery disease, heart failure or valvular defects [6]. Especially in young persons who have died suddenly and unexpectedly, it has been recommended by the International Heart Rhythm Society [7] to perform a thorough forensic investigation. This includes a revision of the medical history of the deceased as well as a complete autopsy with a macroscopic, histopathological and toxicological assessment. Finally, a molecular autopsy is recommended for all SUD cases [7, 9]. There are already many studies that have investigated genetic variants in SUD. Since the first molecular autopsy in 1999 by Ackerman et al. [10], inherited cardiac conditions like cardiac channelopathies including long QT syndrome (LQTS), short QT syndrome (SQTS), Brugada syndrome (BrS), catecholaminergic polymorphic ventricular tachycardia (CPVT) or cardiomyopathic conditions including hypertrophic cardiomyopathy (HCM), dilated cardiomyopathy (DCM) and arrhythmogenic cardiomyopathy (ACM) are being investigated as the underlying cause of SCD, SADS and SUD [11]. Generally, the pathophysiology of SUD is complex [12], and many genes have been associated with underlying cardiac diseases [13]. The genes most commonly implicated in the development of cardiac channelopathies include KCNH2, KCNJ2, KCNQ1, RYR2 and SCN5A [11, 13–15]. For many channelopathies like LQTS, it has further been reported that not all genetic variants lead to disease, and incomplete penetrance is a common feature [16]. For cardiomyopathic conditions, variants in the genes MYH7, JPH2, TNNI3, TTN, LMNA, NEXN, PKP2, DSG2, DSP, TMEM43, MYBPC3, JUP and others have been often observed as having functional effects on cardiac proteins [11–14, 17]. Other commonly described genes are the two transcription factors TBX5 and GATA4 that have been implicated in SUD [18]. They are responsible for cardiovascular development [19] and if mutated can cause congenital heart defects and progressive cardiac conduction disorders [20]. Despite the many genetic variants that have been identified as SCD and SUD contributors, many cases still present no clear genetic causality [11, 12, 14, 17–21]. Therefore, a few studies have investigated gene regulation (DNA methylation) and gene expression (RNA) [22–25] to find potential risk markers and elucidate the complex etiology of SUD. So far, most of the DNA methylation studies have focused on targeted regions or specific conditions (e.g. arrhythmias or sudden death in epilepsy (SUDEP)). One study identified differential methylation of the ABCA1 promoter region between a case cohort of SCD victims and a control cohort [23]. Here, they targeted the ABCA1 gene promoter with a methyl-specific PCR. Other studies tried to identify differential methylation to either predict the phenotype severity of arrhythmic conditions [24] or the development of DCM [25] but failed in doing so. Being able to explain potential underlying causes for the sudden death of a beloved person can help relatives find closure for their loss and motivate them to seek counselling in case of a genetic familial inheritance. However, as molecular autopsies of the deceased often remain negative, it is crucial to expand the knowledge on potential genetic or epigenetic causes that contribute to sudden death events. At the Zurich Institute of Forensic Medicine (ZIFM), University of Zurich, Switzerland, approximately 10 sudden death cases are reported annually without any explanation of the exact cause of death. To our knowledge, no study has previously investigated the entire human methylome in SUD cases. Our study, therefore, aims to fill this gap. We will explore the human methylome using the InfiniumTM MethylationEPIC v2.0 BeadChip kit in three different SUD groups and compare the results to that of a control cohort. We will also describe genes and biological pathways of the identified differentially methylated regions (DMRs) between SUD and control samples. Methods Study cohort The study cohort comprised a total of 54 unrelated individuals in the SUD cohort and 20 unrelated individuals in the control cohort who all died between the age of one and 62 years. Left ventricle heart tissue was collected during autopsies at the ZIFM between 2013 and 2022. The heart tissue sections were shock-frozen in liquid nitrogen and then stored at −80\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^\circ$$\end{document}∘C. Despite the general definition of SUD cases being younger than 50 years of age, we included individuals up to 62 years of age in this study if there had been no other explanation for the sudden death event [26]. All individuals were examined according to a standardized procedure, including a complete autopsy, death scene investigation and toxicological and histopathological screening. The following characteristics were extracted from autopsy reports for all deceased persons: sex, age, post-mortem interval, body mass index (BMI), body height, body weight, heart weight, symptoms prior to death, circumstances at death, reanimation, alcohol status, illicit substance consumption, medical history, microbiological findings and family history. All cases were categorized into one of the following groups according to their case history and histopathological and morphological appearance of the heart: (1) primary normal condition (primaryN), (2) primary cardiomyopathy condition (primaryCM), (3) secondary condition (secondary) and (4) control samples. The primaryN group (n = 26) comprised deceased persons whose heart appeared both macroscopically and histologically inconspicuous, and arrhythmia was the most likely underlying cause of death. Potentially, nine cases in the primaryN group might be classified as sudden unexpected death in epilepsy (SUDEP) because they were either diagnosed with epilepsy prior to death or SUDEP was assumed to be the most likely cause of death in the absence of a terminal seizure. The primaryCM group (n = 18) included cases with a macroscopically and histologically abnormal heart, most of which were likely to be cardiomyopathies. All cases where a prior disease (e.g. hypertension, old infarction scars, past myocarditis, etc.) led to a heart condition are summarized in the secondary group (n = 10). Finally, the control group (n = 20) consisted of suddenly deceased individuals that did not present with any heart-related condition nor SUD, but died because of other non-violent causes (including drowning, illicit substance overdose and accidents). For the categorization into the three SUD groups, a second expert opinion was obtained. Following the recommendations in [27], the initial interpretation of the SUD cases by the physicians at the ZIFM was re-examined by a cardiopathologist, who re-evaluated the histological specimen in combination with the macroscopic photographs and case history. In cases where no histological slides were available (n = 12), the initial autopsy report and macroscopic photographs were reviewed by the cardiopathologist. A summary of the above-mentioned characteristics for the entire study cohort can be found in Supplementary Table 1. For statistical comparison of the metadata among the groups, a student’s t test was used for continuous variables and a fisher’s exact test for categorical variables. DNA extraction, bisulfite conversion and genome-wide methylation assessment DNA from approximately 10 mg left ventricle heart tissue was extracted with the QIAcube instrument (Qiagen, Hilden, Germany) according to the QIAmp® DNA Investigator kit (Qiagen). Based on a study that predicted the amount of usable chip probes from DNA degradation [28], samples of sufficient quantity and quality were selected before proceeding with the bisulfite treatment. Up to 250 ng DNA was bisulfite converted with the EZ DNA MethylationTM Kit (Zymo Research, Irvine, CA, USA) following the manufacturer‘s instructions for use with the Infinium® Methylation Assay (Illumina, San Diego CA, USA). The human methylome of the samples was assessed with the InfiniumTM MethylationEPIC v2.0 BeadChip kit (Illumina) on an iScan instrument (Illumina). This current advancement of the original MethylationEPIC BeadChip kit (Illumina) targets more than 935,000 CpG sites across the human genome. All samples were run in singlets in three batches. Data analysis Data quality control Analysis of the raw data was primarily performed with the R package SeSaMe [29]. The data quality was checked by assessing the signal background, the mean signal intensity, the number of NAs and the bisulfite conversion efficiency. All samples fulfilled the required quality criteria (see Supplementary Fig. 1). A principal component analysis (PCA) was performed on the M-values with the R function prcomp [30]. Factors with potentially confounding effects on principle components (PCs) were identified with the plomics package [31] by testing for an association between independent and dependant variables through linear regression. In the first five PCs, factors showing a statistically significant impact (p value < 0.05) on the results were identified. Consequently, all downstream analyses included a correction for all these factors. Differentially methylated regions DMRs were determined with the DMR function from the SeSaMe package, including a function argument for the previously identified confounding factors. The enrichGO function from the clusterProfiler package [32] was used for gene ontology (GO) analyses. Prediction of the biological age For the prediction of the biological age, Horvath’s clock [33] was applied to each sample using the R package methylclock [34]. Normal distribution of the predictions was checked with the Shapiro–Wilk normality test, and the prediction accuracies between the case and control groups were compared with the Kruskal–Wallis rank sum test. Exome analysis For most of the case samples of this study, exome data were available from previous studies [21, 35, 36]. A summary of these results can be found in the Supplementary Table 2. Results Study cohort From the 74 deceased persons included in this study, 20 were control samples (n female = 6, n male = 14, mean age female = 29.4 years, mean age male = 25.5 years), 26 were classified as primaryN (n female = 6, n male = 20, mean age female = 29.3 years, mean age male = 30.7 years), 18 as primaryCM (n female = 8, n male = 10, mean age female = 32.2 years, mean age male = 32.8 years) and 10 as secondary (n female = 3, n male = 7, mean age female = 34.9 years, mean age male = 40.6 years). Across all groups, 7 individuals were above the age of 50 (n primaryN group = 1, n primaryCM group = 3, n secondary group = 2, n control group = 1). A summary of all comparisons can be found in Supplementary Table 1. Statistical comparisons between each case group and the control group were performed on all available metadata to investigate possible confounding variables caused by the study design. The comparison between the primaryN and the control group showed one statistically significant difference, namely in illicit substance consumption, as more primaryN individuals had not consumed illicit substances (primaryN: n = 22/26, control: n = 7/20, p value = 0.023). Also, the comparison between the primaryCM and the control group yielded a statistically significant difference in the consumption of illicit substances. More controls than primaryCM individuals had consumed amphetamines, cannabis, benzodiazepines, cocaine and opiates (primaryCM: n = 0/18, control: n = 6/20, p value = 0.010), all of which are likely QT interval prolonging according to www.crediblemeds.org, www.brugadadrugs.com and personal communication. Additionally, the heart weight was statistically significantly higher in the primaryCM group than in the control group (primaryCM: mean (SD) = 420 g (154.08 g), control: mean (SD) = 306.5 g (108.25 g), p value = 0.014). When comparing the secondary and the control group, a statistically significantly older mean age was observed for the secondary group (secondary: mean (SD) = 38.92 years (11.51 years), control: mean (SD) = 26.65 years (11.98 years), p value = 0.014) as well as a statistically significantly shorter post-mortem interval (mean secondary (SD) = 22 h (9.36 h), mean control (SD) = 36.9 h (27.52 h), p value = 0.038) and a statistically significantly longer storage time of the tissue (secondary: mean (SD) = 8.2 years (1.81 years), control: mean (SD) = 4.3 years (2.15 years), p value = 0.00004). Finally, no one in the secondary group died during normal daily activities (e.g. working in an office) compared to five people in the control group (p value = 0.026), and heart weight was larger in the secondary group compared to the control group (secondary: mean (SD) = 434 g (135.42 g), control: mean (SD) = 306.5 g (108.25 g), p value = 0.020). Age prediction A common assessment in methylome studies is the prediction of the biological age based on DNA methylation changes at age-associated CpG sites [37]. Therefore, a prediction of the biological age based on Horvath’s clock [33] was performed on all samples (see Supplementary Fig. 2). No statistically significant differences in the prediction accuracy were observed among the three case groups and the control group (p value = 0.13). Additionally, all three case groups were considered together and compared to the control group, which yielded a non-significant result in the prediction accuracy (p value = 0.063). Principle component analysis An initial investigation of potentially confounding factors revealed twelve statistically significant correlations (p value < 0.05) with PCs: sex, age, post-mortem interval, body weight, BMI, batch, sample position on the array chip, sample storage time, reanimation, medical history, event at death and alcohol consumption (see Supplementary Fig. 3). Based on these results, a correction for the confounding variables was carried out in the subsequent analysis for DMRs. Case/control annotation was statistically significantly correlated (p value < 0.05) with PC4, PC6, PC7 and PC9 within the first ten PCs. PCA plots were generated for all combinations of these PCs with PC6 versus PC9 yielding the clearest clustering of the case and control groups (see Supplementary Fig. 3). PrimaryN case group Differentially methylated regions DMRs between the primaryN and the control groups were identified by filtering for statistically significant regions (p value < 0.05) with an absolute difference in the region’s beta values of at least 0.1 (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta$$\end{document}Δ beta = 0.1). All investigated regions are also visualized in a Volcano plot (see Supplementary Fig. 4). For the primaryN group, 605 DMRs were found. Of those, 366 were statistically significantly hypomethylated (p value < 0.05), while 239 were statistically significantly hypermethylated (p value < 0.05). All DMRs were associated with a total of 524 genes (see Supplementary Table 3). Gene ontology GO analysis of biological pathways revealed primarily cell and organ development-related pathways (Fig. 1). Notably, the pathway with the lowest false discovery rate (FDR) value was the heart outflow tract morphogenesis. Other pathways with low FDRs included organ growth, cell fate commitment, anterior/posterior pattern specification, collagen fibril organization and artery development. The GO analysis of molecular functions or cellular components did not identify any common functions or components for the statistically significant genes of the primaryN cases.Fig. 1Gene ontology (GO) term analysis of biological pathways for the genes associated with the differentially methylated regions (DMRs) of the primaryN group. FDR, False discovery rate Gene ontology (GO) term analysis of biological pathways for the genes associated with the differentially methylated regions (DMRs) of the primaryN group. FDR, False discovery rate Genes The genes associated with the 20 DMRs with the lowest p values were investigated in more detail. All 20 genes are listed in Table 1, and a heatmap with all CpG sites corresponding to these 20 DMRs can be found in Supplementary Fig. 5. A GO term analysis investigating common biological pathways, molecular functions or cellular components of these genes did not yield any results. Some of the 20 genes are involved in regulating gene expression (long non-coding RNAs, a microRNA and a homeobox gene) or displaying enzymatic activities (a NADH dehydrogenase component, a RING finger domain component and a methyltransferase).Table 1Genes associated with the 20 differentially methylated regions (DMRs) with the lowest p values of the primaryN casesGene nameProtein functionADAMTS4/NDUFS2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{1}$$\end{document}1Produces Alzheimer disease amyloid marker APP669-711/see below [38]ARHGAP25Rac GTPase-activating protein that regulates phagocytosis in human neutrophilic granulocytes [39]CUTALP/PSMD5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{1}$$\end{document}1Long non-coding RNA [40]/proteasome inhibition caused by TNF-\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upalpha$$\end{document}α NF-\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varkappa$$\end{document}ϰB in higher eukaryotes [41]ENSG00000264456/RNF135\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{1}$$\end{document}1Long non-coding RNA/promotes the proliferation of human glioblastoma cells via the ERK pathway [42]ENSG00000272729/HAGLR/HOXD3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{1}$$\end{document}1Long non-coding RNA/long non-coding RNA of potential tumour suppressive function in clear cell renal cell carcinoma (HAGLR) [43]ENSG00000282221/LAD1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{1}$$\end{document}1Long non-coding RNA/anchoring filament protein, may lead to autoimmune diseases [44]FAAP20Ubiquitin-binding Fanconi anaemia nuclear core-complex protein [45]FGF12Fibroblast growth factor that is a modulator for neuronal network activity [46]HOXD3Member of the homeobox family of genes, for transcriptional regulation of morphogenesis in all multicellular organisms [47]MIR4687/STIM1/STIM1-AS1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{1}$$\end{document}1MicroRNA/component of store-operated Ca2+ channel function [48]/STIM1 antisense RNA 1MTLNMitochondrial peptide that enhances function of respiratory complex I [49]NDUFS2NADH:ubiquinone oxidoreductase subunit S4 of OXPHOS complex I [50]OTOGNoncollagenous protein specific to the inner ear, associated with mild-to-moderate nonsyndromic hearing loss [51]PRDM8/PRDM8-AS1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{1}$$\end{document}1PR domain zinc finger protein 8, part of histone methyltransferases [52]/PRDM8 antisense RNARFTN1Facilitates gastric cancer progression by modulating AKT/p38 signalling pathways [53]RP1Might regulate retinal oxygen levels, mutations associated with retinitis pigmentosa [54]SLC16A3H+ symporter in lactate transport, highly expressed in metastatic tumours [55]TMEM88Suppressor of the Wnt/\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upbeta$$\end{document}β-catenin signalling pathway [56]UQCRC1Core component of mitochondrial respiratory chain complex III [57]XAF1Proapoptotic tumour suppressor [58]\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1The underlying DMR belongs to all of these genes Table 1Genes associated with the 20 differentially methylated regions (DMRs) with the lowest p values of the primaryN casesGene nameProtein functionADAMTS4/NDUFS2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{1}$$\end{document}1Produces Alzheimer disease amyloid marker APP669-711/see below [38]ARHGAP25Rac GTPase-activating protein that regulates phagocytosis in human neutrophilic granulocytes [39]CUTALP/PSMD5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{1}$$\end{document}1Long non-coding RNA [40]/proteasome inhibition caused by TNF-\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upalpha$$\end{document}α NF-\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varkappa$$\end{document}ϰB in higher eukaryotes [41]ENSG00000264456/RNF135\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{1}$$\end{document}1Long non-coding RNA/promotes the proliferation of human glioblastoma cells via the ERK pathway [42]ENSG00000272729/HAGLR/HOXD3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{1}$$\end{document}1Long non-coding RNA/long non-coding RNA of potential tumour suppressive function in clear cell renal cell carcinoma (HAGLR) [43]ENSG00000282221/LAD1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{1}$$\end{document}1Long non-coding RNA/anchoring filament protein, may lead to autoimmune diseases [44]FAAP20Ubiquitin-binding Fanconi anaemia nuclear core-complex protein [45]FGF12Fibroblast growth factor that is a modulator for neuronal network activity [46]HOXD3Member of the homeobox family of genes, for transcriptional regulation of morphogenesis in all multicellular organisms [47]MIR4687/STIM1/STIM1-AS1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{1}$$\end{document}1MicroRNA/component of store-operated Ca2+ channel function [48]/STIM1 antisense RNA 1MTLNMitochondrial peptide that enhances function of respiratory complex I [49]NDUFS2NADH:ubiquinone oxidoreductase subunit S4 of OXPHOS complex I [50]OTOGNoncollagenous protein specific to the inner ear, associated with mild-to-moderate nonsyndromic hearing loss [51]PRDM8/PRDM8-AS1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{1}$$\end{document}1PR domain zinc finger protein 8, part of histone methyltransferases [52]/PRDM8 antisense RNARFTN1Facilitates gastric cancer progression by modulating AKT/p38 signalling pathways [53]RP1Might regulate retinal oxygen levels, mutations associated with retinitis pigmentosa [54]SLC16A3H+ symporter in lactate transport, highly expressed in metastatic tumours [55]TMEM88Suppressor of the Wnt/\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upbeta$$\end{document}β-catenin signalling pathway [56]UQCRC1Core component of mitochondrial respiratory chain complex III [57]XAF1Proapoptotic tumour suppressor [58]\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1The underlying DMR belongs to all of these genes Genes associated with the 20 differentially methylated regions (DMRs) with the lowest p values of the primaryN cases \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1The underlying DMR belongs to all of these genes PrimaryCM case group Differentially methylated regions A total of 63 DMRs were identified for the primaryCM group in comparison to the controls. All inspected regions were also visualized in a Volcano plot (see Supplementary Fig. 4). Of these 63 identified DMRs, 32 were hypomethylated and 31 were hypermethylated. These DMRs were associated with 58 genes (see Supplementary Table 3). Gene ontology The GO analysis did not reveal any biological pathways or cellular components for the DMR associated genes. However, four molecular functions common to several of those genes were found, namely p53 binding, lysine-acetylated histone binding, acetylation-dependent protein binding and transcription coactivator activity (see Fig. 2).Fig. 2Gene ontology (GO) term analysis of molecular functions for the genes associated with the differentially methylated regions (DMRs) of the primaryCM group Gene ontology (GO) term analysis of molecular functions for the genes associated with the differentially methylated regions (DMRs) of the primaryCM group Genes The 20 DMRs with the lowest p values were associated with 19 genes (Table 2). One DMR was not associated with any gene. GO analysis revealed an enrichment for genes associated with the epigenetic regulation of gene expression. A heatmap of all CpG sites belonging to the DMRs associated with a gene can be found in Supplementary Fig. 5.Table 2Genes associated with the 20 differentially methylated regions (DMRs) with the lowest p values of the primaryCM casesGene nameProtein functionBLCAP/NNAT\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1Tumour suppressor gene, associated with bladder cancer [59]/regulation of cellular molecular signalling from governing of free cytosolic Ca2+ to insulin secretion under pathological conditions [60]\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^2$$\end{document}2BRSK1AMPK-related subfamily of protein kinase, mainly characterized in the brain for its role in neuronal polarization [61]ENSG00000244459Long non-coding RNAENSG00000256341/MACROD1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1Long non-coding RNA/involved in MARylation, having hydrolyse activity [62]ENSG00000261363/LINC02181\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1Long non-coding RNA/long non-coding RNAENSG00000261888Long non-coding RNAENSG00000262560/HYPK/SERF2/SERINC4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1Long non-coding RNA/interaction partner of Huntingtin, associated with the NatA complex [63]/promotes aggregation of A \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upbeta$$\end{document}β in human neuroblastoma cells overexpressing amyloid precursor protein [64]/part of the serine incorporator family, has antiviral activities [65]ENSG00000284779/IGF2/IGF2-AS/INS-IGF2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1Long non-coding RNA/insulin-like growth factor, involved in the regulation of cell proliferation, growth, migration, differentiation and survival [66]/IGF2 antisense RNA/INS-IGF2 read-through transcript proteinHOXC4Up-regulates NF-\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varkappa$$\end{document}ϰB signalling and promotes the cell proliferation, drives hematopoiesis [67]KCNQ1/KCNQ1OT1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1Voltage-gated potassium channel, major driver of LQTS [68]/oncogenic long non-coding RNA [69]LINC02940Long non-coding RNAMYBPHMyosin-binding protein, potential association with HCM [70]POU6F1POU domain, class 6, transcription factor 1 involved in developmental processes, implicated in the suppression of lung adenocarcinoma suppression [71]SEMA6CSemaphorin family member, involved in cancer promoting-signalling pathways sustaining cell viability and self-renewal [72]SMARCA4Regulates gene expression through higher-order chromatin organization in epithelial cells [73]STK11Tumour suppressor gene, AMPK regulation of mTOR and HIF-1-\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upalpha$$\end{document}α [74]WNT6Member of the WNT gene family, mediates various cellular functions [75]ZIC4Zinc-finger protein, implicated in cerebellar development in mice [76]ZIC4-AS1ZIC4 antisense RNA 11The underlying DMR belongs to all of these genes2Complex interaction between NNAT and BLCAP with NNAT being imprinted while BLCAP is not Table 2Genes associated with the 20 differentially methylated regions (DMRs) with the lowest p values of the primaryCM casesGene nameProtein functionBLCAP/NNAT\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1Tumour suppressor gene, associated with bladder cancer [59]/regulation of cellular molecular signalling from governing of free cytosolic Ca2+ to insulin secretion under pathological conditions [60]\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^2$$\end{document}2BRSK1AMPK-related subfamily of protein kinase, mainly characterized in the brain for its role in neuronal polarization [61]ENSG00000244459Long non-coding RNAENSG00000256341/MACROD1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1Long non-coding RNA/involved in MARylation, having hydrolyse activity [62]ENSG00000261363/LINC02181\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1Long non-coding RNA/long non-coding RNAENSG00000261888Long non-coding RNAENSG00000262560/HYPK/SERF2/SERINC4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1Long non-coding RNA/interaction partner of Huntingtin, associated with the NatA complex [63]/promotes aggregation of A \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upbeta$$\end{document}β in human neuroblastoma cells overexpressing amyloid precursor protein [64]/part of the serine incorporator family, has antiviral activities [65]ENSG00000284779/IGF2/IGF2-AS/INS-IGF2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1Long non-coding RNA/insulin-like growth factor, involved in the regulation of cell proliferation, growth, migration, differentiation and survival [66]/IGF2 antisense RNA/INS-IGF2 read-through transcript proteinHOXC4Up-regulates NF-\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varkappa$$\end{document}ϰB signalling and promotes the cell proliferation, drives hematopoiesis [67]KCNQ1/KCNQ1OT1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^1$$\end{document}1Voltage-gated potassium channel, major driver of LQTS [68]/oncogenic long non-coding RNA [69]LINC02940Long non-coding RNAMYBPHMyosin-binding protein, potential association with HCM [70]POU6F1POU domain, class 6, transcription factor 1 involved in developmental processes, implicated in the suppression of lung adenocarcinoma suppression [71]SEMA6CSemaphorin family member, involved in cancer promoting-signalling pathways sustaining cell viability and self-renewal [72]SMARCA4Regulates gene expression through higher-order chromatin organization in epithelial cells [73]STK11Tumour suppressor gene, AMPK regulation of mTOR and HIF-1-\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upalpha$$\end{document}α [74]WNT6Member of the WNT gene family, mediates various cellular functions [75]ZIC4Zinc-finger protein, implicated in cerebellar development in mice [76]ZIC4-AS1ZIC4 antisense RNA 11The underlying DMR belongs to all of these genes2Complex interaction between NNAT and BLCAP with NNAT being imprinted while BLCAP is not Genes associated with the 20 differentially methylated regions (DMRs) with the lowest p values of the primaryCM cases 1The underlying DMR belongs to all of these genes 2Complex interaction between NNAT and BLCAP with NNAT being imprinted while BLCAP is not Secondary case group We did not find any statistically significant DMRs for the comparison of the secondary group and the control cohort. Exome analysis The previously published exome results were compared to the methylation results by investigating the beta values in samples with a pathogenic or likely pathogenic variant in contrast to all other samples. The beta values in CpG sites of all genes with (likely) pathogenic variants were visually inspected in boxplots with the exception of LZTR1 and CALR3, for which no CpG sites are covered in the InfiniumTM MethylationEPIC v2.0 BeadChip kit (see Supplementary Fig. 6). Samples with (likely) pathogenic variants were not found to have any significantly different beta values across the genes affected by the variants. Discussion In this study, we compared the human methylome in post-mortem ventricular tissue of SUD cases to a suddenly deceased non-SUD control cohort. We divided the SUD cases into three groups: (1) primaryN, whose heart appeared morphologically and histopathologically normal, suspected to be primarily arrhythmias; (2) primaryCM, whose heart was morphologically enlarged and histopathologically conspicuous, compatible with a primary cardiomyopathy; and (3) secondary, whose heart failure was suspected to result from a prior condition (e.g. hypertension, past myocarditis, etc.). With this subdivision into three groups, we strived to obtain meaningful results that are least possibly affected by the heterogeneity of our study cohort. Additionally, this should also ensure to have big enough groups to perform our analysis. As part of our grouping criteria, we included the heart weight as an indicator for a normally weighted or enlarged heart. For all SUD cases, the heart weight was measured after the opening of the heart chambers and the removing of blood and clots. Though we are aware that this might lead to flawed heart weights in comparison to measurements before chamber opening [77], all cases are consistently measured in the same way ensuring a comparability within our cohort. The distribution of the metadata among the three case groups and the control group was statistically significantly different for the consumption of illicit substances in the primaryN group, for the heart weight and the consumption of illicit substances in the primaryCM group and for age, post-mortem interval, storage time, heart weight and the event at death in the secondary group. The smaller number of persons consuming illicit substances in the primaryN and primaryCM group compared to the control group is explained because the control cohort explicitly includes sudden death cases due to illicit substance overdoses. Though this could potentially affect methylation patterns, we are assuming that it does not do so in heart tissue and should therefore be irrelevant for this study. An additional finding in the primaryCM group was the statistically significantly larger heart weight, which may be explained by the definition of this case group as it included all cases with a morphologically enlarged heart. In the secondary group, both the greater heart weight and the older age can be explained by the definition of this group. As it includes cases whose cardiac dysfunction (both in normally weighted and enlarged hearts) is due to a previous morbidity or accumulating lifestyle factors, it is logical that such cases tend to be found in older individuals. The other statistically significant differences between this case group and the control group (shorter post-mortem interval, longer storage time and smaller number of deaths that occurred during normal daily activities like working in an office) are assumed to be accidental findings. We adjusted for the statistically significantly different factors in our analysis. In contrast to many other SUD studies, this study also included a small number of cases over the age of 50 years (case groups n = 6, control group n = 1). Since these individuals did not exhibit unexpected or significantly different methylation patterns compared to the other cases within their respective groups, it is reasonable to assume that excluding these cases would not have significantly impacted the results. Given that these individuals experienced an unexpected death similar to all individuals under 50, their inclusion in the analysis was deemed appropriate for this study. The investigation of the methylome found that the primaryN group yielded the most promising results out of all three case groups. Of the 524 genes associated with DMRs, some were directly involved in heart-related biological pathways, such as the outflow tract morphogenesis, which is crucial for blood supply to the arteries during heart development and in the adult heart [78], the artery development and trabecular morphogenesis. Although neither of these pathways are directly related to arrhythmic cardiac conditions, which are suspected to be the cause of death in this case group, we hypothesize that even general abnormalities in heart morphogenesis could have contributed to SUD through an arrhythmic condition. All other observed pathways were not directly related to heart functions. This may indicate that some differential methylation can be observed in heart-related pathways, although the majority of DMRs are found in genes associated with more general functions like organ growth, cell fate commitment, mesoderm development, mesenchyme development and connective tissue development. These findings suggest that differential methylation does not affect just a single heart-associated pathway but most likely affects a combination of different biological pathways whose complex interactions may have contributed to a fatal heart condition. In the primaryCM group, no common biological pathways were identified. This could be explained by the relatively small number of genes associated with DMRs (n = 58). In addition, this case group includes a variety of suspected cardiomyopathies like HCM or DCM which might make it difficult to identify common pathways. Notably, four common molecular functions were identified (p53 binding, lysine-acetylated histone binding, acetylation-dependent protein binding and transcription coactivator activity), all of which are related to gene regulation. We speculate that changes in such regulatory functions may have contributed to the malfunction of the heart by adding to its abnormal growth. For instance, p53 has been shown to contribute to the development of cardiovascular diseases via mitochondrial dysfunction and was reported to be elevated in DCM or diabetic cardiomyopathy [79]. Furthermore, histone acetylation inhibition was reported to prevent ventricular hypertrophy development in rats which could indicate that differential expression in such pathways might contribute to an altered heart morphology as well [80]. Amongst the genes associated with the DMRs with the lowest p values in the primaryN group, it was striking that one gene, FGF12, had already been shown to be associated with cardiac arrhythmia and early onset of epilepsy in mice [81]. FGF12 is involved in multiple biological processes including embryonic development, cell growth, morphogenesis, tissue repair, tumour growth and invasion [46]. Genetic variants have also been implicated in developmental and epileptic encephalitis as well as autism spectrum disorder. Although the previously reported association with SUD was due to pathogenic variants of FGF12, a deregulation of this gene caused by changes in its methylation pattern may contribute to its pathogenic effect. Another gene with reported heart specific functions found in the primaryN group was TMEM88. TMEM88 encodes the transmembrane protein 88 that is a suppressor of the Wnt/\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upbeta$$\end{document}β-catenin signalling pathway, which regulates cardiovascular progenitor cell specification [56]. It has been shown that the knockdown of TMEM88 results in a cell fate shift towards endothelial rather than cardiomyocyte development. Therefore, its deregulation through differential methylation may also have a detrimental effect on the heart. Of the other genes associated with the top 20 DMRs, several genes are involved in tumourigenesis [42, 43, 53, 55, 58], others contribute to different diseases like Alzheimer’s [38], retinitis pigmentosa [54], hearing loss [51], leucocyte adhesion deficiency [82] or autoimmune diseases [44]. As there was no clear common denominator for most of these genes, further studies will be required to elucidate their role in SUD. Additionally, it might be beneficial to investigate other tissues than heart, especially in suspected SUDEP cases where an investigation of brain tissue may provide further insights. The exclusion of all suspected SUDEP cases might have led to subtle differences in the overall results, though predicting the extent of this impact is challenging. It is possible that the variation observed in the methylation patterns would have been slightly reduced, even though this study focused on heart rather than brain tissue. Nonetheless, the genes associated with the lowest p values for this group did not include genes related to epileptic conditions, suggesting that SUDEP cases were unlikely to be a key factor in the observed outcomes, and their exclusion would likely not have significantly altered the results for this group. Amongst the genes associated with the DMRs with the lowest p values in the primaryCM group, MYBPH was previously reported to be associated with cardiac disorders, including HCM [70]; however, further studies are needed to confirm these findings. Some of the other significant genes include one gene involved in cancer development [72] and genes with regulatory functions like chromatin structure regulation [73], cell proliferation [67] or long non-coding RNAs. It is striking that some of the 20 DMRs with the lowest p values in both primaryN and primaryCM group are driven by only one case being significantly differentially methylated than the other cases of that group and the entire control cohort. While these DMRs are not representative for the whole case group, finding differential methylation in one case could still be of great importance for the respective case. An example of this is SUD062 in the primary CM group, where we identified KCNQ1 as differentially methylated. Pathogenic variation in KCNQ1 cause LQTS [11] and KCNQ1 is a recommended target for genetic testing in SUD cases [83]. Although the primaryCM group does not include LQTS, recent studies have suggested that cardiomyopathies and channelopathies might not be as genetically distinctive as previously assumed [84, 85]. In the same case, WNT6 was also identified to be differentially methylated. WNT6 is a member of the WNT gene family. It mediates various cellular functions, including some in heart tissue. Upon heart injury, the WNT signalling is inhibited to prevent further cardiac damage [75]. This causes an increased development of cardiac progenitor cells and an overall inhibited proliferation, both of which contribute to a faster healing of the heart. In the context of SUD, differential expression of this gene might cause an impaired response to cardiac stress adding to the ultimate heart failure. Our overall results for the secondary group suggest no differential methylation between this case group and the control group. This is most likely due to the heterogeneity of the case group that includes SUD cases with a variety of prior conditions such as past myocarditis, chemotherapy, renal insufficiency, hypertension, previous mental disease and old infarction scars on the heart. Furthermore, this study only investigated DNA methylation patterns in heart tissue although the underlying primary condition in this case group is not always found in the heart. An investigation of a different tissue, like brain or kidney, in the secondary group might reveal more significant results regarding DNA methylation patterns. As differential methylation could be strongly influenced by imprinting, we compared the genes associated with DMRs from all case groups to a list of 150 previously reported imprinted genes [86]. Six of these were present in our results (KCNQ1, KCNQ1DN, WT1, BLCAP, NNAT and DLGAP2); however, their imprinting patterns were reported to be elsewhere than in heart tissue [60, 87–90]. Therefore, we suggest that imprinting did not affect the results of our study. In methylome studies, it is common practice to predict the biological age based on DNA methylation changes at age-associated CpG sites [37]. An overall under- or overestimation of age in the case groups of this study could be a general indicator for a more or less exhausted methylome. We predicted biological age with Horvath’s original pan-tissue epigenetic clock based on 353 age-dependent CpG sites [33]. We found no statistically significant differences among the case and control groups. With the relatively limited sample size, the power to detect small differences was not big, and therefore, it cannot be excluded that a biological age difference exists nonetheless between the case and control groups. A bigger sample size would be needed to confirm this. Taken together, our findings suggest a moderate influence of differential methylation patterns on some biological pathways and genes associated with the human heart. This aligns with other studies that have found methylation differences in heart diseases such as various forms of cardiomyopathies including DCM [91, 92] and HCM [92]. In addition, some studies have reported an overall moderate increase in global methylation levels in some cardiac diseases like coronary heart disease, acute coronary syndrome, cardiomyopathies or heart valve issues, while other diseases like atherosclerosis showed a global hypomethylation [93]. This correlates well with our findings, as the primaryN and primaryCM group both showed a moderate global hypermethylation, while the secondary group showed a moderate hypomethylation compared to the control samples. Pathogenic or likely pathogenic variants were not found to affect the methylation levels in the respective samples suggesting that any methylation differences observed in this study should not be influenced by genetic alterations. However, a more detailed analysis of the exome-methylome interaction in these cases might reveal more insights in future studies. Conclusion In this study, we aimed to show for the first time that whole DNA methylation analysis of SUD cases can be a potential addition to existing genetic investigations. We find that there are DMRs between cases with and without morphological abnormalities of the heart and a control group. We even identified some individual cases whose methylation patterns at some DMRs were significantly different from the control cohort and the remaining cases of the respective case group. Therefore, our study shows that DNA methylation may be an additional contributor to the development of SUD and a potential future target for therapeutic interventions in relatives. In addition, the investigation of other omics approaches could also greatly contribute to a better understanding of SUD in general. Supplementary Information Supplementary Material 1.Supplementary Material 2.Supplementary Material 3. Supplementary Material 1. Supplementary Material 2. Supplementary Material 3.