Rovira et al. Translational Psychiatry (2020)10:199 https://doi.org/10.1038/s41398-020-0860-4 Translational Psychiatry ARTICLE Open Access Epigenome-wide association study of attention- deficit/hyperactivity disorder in adults Paula Rovira1,2, Cristina Sánchez-Mora 1,2,3,4, Mireia Pagerols1,2, Vanesa Richarte2,3,5, Montserrat Corrales2,3,5, Christian Fadeuilhe2,3,5, Laura Vilar-Ribó 1,2, Lorena Arribas1,2, Gemma Shireby6, Eilis Hannon 6, Jonathan Mill 6 , Miquel Casas2,3,5, Josep Antoni Ramos-Quiroga 1,2,3,5, María Soler Artigas 1,2,3,4 and Marta Ribasés 1,2,3,4 Abstract Attention-deficit/hyperactivity disorder (ADHD) is a highly heritable neurodevelopmental disorder that often persists into adulthood. There is growing evidence that epigenetic dysregulation participates in ADHD. Given that only a limited number of epigenome-wide association studies (EWASs) of ADHD have been conducted so far and they have mainly focused on pediatric and population-based samples, we performed an EWAS in a clinical sample of adults with ADHD. We report one CpG site and four regions differentially methylated between patients and controls, which are located in or near genes previously involved in autoimmune diseases, cancer or neuroticism. Our sensitivity analyses indicate that smoking status is not responsible for these results and that polygenic risk burden for ADHD does not greatly impact the signatures identified. Additionally, we show an overlap of our EWAS findings with genetic signatures previously described for ADHD and with epigenetic signatures for smoking behavior and maternal smoking. These findings support a role of DNA methylation in ADHD and emphasize the need for additional efforts in larger samples to clarify the role of epigenetic mechanisms on ADHD across the lifespan. 1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; Introduction emerged as a plausible mechanism by which environ- Attention-deficit/hyperactivity disorder (ADHD) is a mental exposures can lead to long-lasting alterations, such common neurodevelopmental disorder characterized by as variation in brain structure or neuronal circuits, found age-inappropriate levels of inattention, impulsivity and in psychiatric disorders5–7. There is growing evidence that hyperactivity1. ADHD is a disabling condition in child- epigenetic dysregulation is a feature of ADHD6,8–11, hood and adolescence which often persists into adult- depression12, autism13–16, schizophrenia17,18 and bipolar hood, interfering with the quality of social, academic, or disorder19. occupational functioning2,3. Studies of DNA methylation profiles in ADHD have been ADHD is a multifactorial disorder with an estimated conducted using peripheral blood, cord blood, buccal heritability of 76%. Twenty-two percent of its phenotypic samples or saliva6,9–11,20–28. Candidate gene studies have variance is explained by common genetic variants1,4 and revealed differential methylation patterns in genes involved the proportion of variance still to be explained might be, in the dopaminergic, serotoninergic and neurotrophic to some extent, accounted for by gene by environment systems, including SLC6A4, DRD4, COMT, ANKK1, interactions. In this context, epigenetic processes have BDNF, or NGFR, associated with ADHD symptomatology and severity23–28. Seven epigenome-wide association stu- dies (EWASs) on ADHD have been run to date, with Correspondence: María Soler Artigas (maria.soler@vhir.org) or Marta Ribasés (marta.ribases@vhir.org) sample sizes ranging from 54 subjects for clinical samples21 1 Psychiatric Genetics Unit, Group of Psychiatry, Mental Health and Addiction, to 4,689 individuals in a meta-analysis considering ADHD Vall d’Hebron Research Institute (VHIR), Universitat Autònoma de Barcelona, symptomatology in general population9, yielding non- Barcelona, Spain 2 Department of Psychiatry, Hospital Universitari Vall d’Hebron, Barcelona, Spain overlapping findings across them6,9–11,20–22. There is Full list of author information is available at the end of the article © The Author(s) 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Rovira et al. Translational Psychiatry (2020)10:199 Page 2 of 12 limited research on adults using this approach, given that Materials and methods most of the EWASs have focused on pediatric sam- Participants and clinical assessment ples6,10,11,20,22. To the best of our knowledge, only two The clinical sample consisted of 103 ADHD subjects studies evaluated methylome-wide patterns on adults9,21. that were referred to an ADHD program from primary One identified methylation changes associated with ADHD care centers and adult community mental health services. symptomatology that did not remain significant when All subjects were evaluated and recruited prospectively results were meta-analyzed across cohorts9. The second from a restricted geographic area of Catalonia (Spain) in a one found hypermethylated regions in genes involved in specialized out-patient program for Adult ADHD and by fatty acid metabolism and fatty acid oxidation pathways a single clinical group at Hospital Universitari Vall associated with ADHD persistence when compared to d’Hebron of Barcelona (Spain). remittance21. In the childhood period, Wilmot et al. ana- The clinical assessment consisted of structured inter- lyzed a population cohort of school-age boys and found views and self-reported questionnaires in two different lower methylation levels at the VIPR2 gene in ADHD steps: (i) assessment of ADHD diagnosis based on subjects compared to their age- and sex- matched con- symptomatology using the Conner’s Adult ADHD trols10, results that were recently replicated in the largest Diagnostic Interview for DSM-IV (CAADID) by a psy- EWAS on ADHD in children conducted so far22. In a chiatrist and, (ii) assessment of the severity of ADHD similar aged population cohort, Walton et al. investigated symptoms, the levels of impairment and the presence of ADHD symptom trajectories from birth to adolescence comorbid disorders by a psychologist to increase the and pointed to epigenetic marks in genes related to neural diagnostic accuracy and reduce the likelihood of mis- tube development and peroxisomal mechanisms as candi- diagnosis with the Conners ADHD Rating Scale dates to be involved in the different ADHD symptom (CAARS), the ADHD Rating Scale (ADHD-RS), the trajectories across time6. In the most recent EWAS eval- Clinical Global Impression (CGI), the Wender Utah uating ADHD symptoms in population-based cohorts, Rating Scale (WURS), the Sheehan Disability Inventory aberrant methylation patterns at birth in different regions, (SDS), and the Structured Clinical Interview for DSM-IV lying in the ERC2 and CREB5 genes among others, were Axis I and II Disorders (SCID-I and SCID-II). After- associated with later ADHD symptoms in childhood or wards, the psychiatrist and psychologist integrate the adolescence11. And finally, the latest and largest EWAS clinical information and self-reports for the valid conducted in a clinical sample of children with ADHD assessment of symptoms and impairments. In case of supported the association between ADHD polygenic risk discordance between different raters of ADHD symp- and DNA methylation patterns at the GART and SON toms or inconsistencies between reporters in responses genes22. to items measuring similar symptoms, the clinician- Recent evidence supports a large genetic overlap between identified symptoms on the CAADID prevailed. Exclu- ADHD in children and adults29, but little is known about sion criteria were IQ < 70; lifelong and current history of the co-occurrence between the epigenetic signatures mood, psychotic, anxiety, substance abuse, and DSM-IV characterizing both groups of age. In addition, although axis II disorders; pervasive developmental disorders; a various studies report shared genetics between ADHD and history or the current presence of a condition or illness, several psychiatric and behavioral traits4,29, this overlap has including neurologic, metabolic, cardiac, liver, kidney, or not been assessed yet using epigenome-wide data. respiratory disease; a chronic medication of any kind; Whereas most previous studies considered pediatric birth weight≤1.5 kg; and other neurological or systemic clinical samples or adult population-based cohorts with disorders that might explain ADHD symptoms. For measures of ADHD symptoms, we report an EWAS on a more detailed information on clinical assessment see clinical sample of adults with ADHD. With these data we Sánchez-Mora et al.30. (i) assessed DNA methylation signatures for ADHD in Data pertaining to exposure to 17 stressful life events adults through an EWAS in peripheral blood mono- (six gestational and 11 postnatal) were collected retro- nuclear cells, (ii) tested whether either polygenic risk spectively with the CAADID Part I31 and were available burden for ADHD or smoking status had an impact on from 98 subjects with ADHD. No information was those DNA methylation signatures, (iii) examined whe- available from controls. Specifically, this questionnaire ther exposure to stressful life events had an effect on these includes: premature birth, illegal drug abuse during methylation patterns in ADHD subjects and (iv) explored pregnancy, maternal smoking, prenatal exposure to drugs, the overlap between these findings and results from pre- maternal health problems during pregnancy, other pro- vious meta-analyses of genome-wide association studies blems during maternal pregnancy, exposure to heavy (GWAS-MA) on clinical ADHD or ADHD symptoms in metals, malnutrition, financial stress and/or poverty, population-based samples, and EWAS on ADHD symp- extreme familial stress, neglect, familiar violence, emo- toms or exposure to stressful life events. tional and physical maltreatment, sexual abuse, death or Rovira et al. Translational Psychiatry (2020)10:199 Page 3 of 12 separation from a loved one, and other trauma in child- calculate the estimated age of participants according to hood or adolescence. their DNA methylation data, which correlated with their The control sample consisted of 100 unrelated healthy reported age (ρ = 0.82, SE = 0.04, P < 2.00E−16). Poorly blood donors matched by sex and ethnicity with the performing probes or samples were removed using the clinical group. Individuals with ADHD symptomatology wateRmelon R package (version 0.9.9;35). The exclusion were excluded retrospectively under the following criteria: criteria for the probes included detection P-values >0.05 (1) having been diagnosed with ADHD previously or (2) for >1% of the samples and a beadcount <3 for >5% of the answering positively to the lifetime presence of the fol- samples. Probes that were cross-reactive, present in sexual lowing ADHD symptoms: (a) often has trouble in keeping chromosomes or that contained polymorphisms were also attention on tasks, (b) usually loses things needed for excluded from the study36,37. Samples with >1% of probes tasks, (c) often fidgets with hands or feet or squirms in with a detection P-value >0.01 were also removed. Probes seat, and (d) often gets up from seat when remaining in that passed the quality control filters were quantile nor- seat is expected. malized with the dasen function of the wateRmelon All subjects reported European ancestry, which was R package. confirmed through principal component analysis (PCA) using genetic data. The study was approved by the Clinical Bioinformatic and statistical analyses Research Ethics Committee (CREC) of Hospital Uni- PCA of methylation values was conducted using the versitari Vall d’Hebron, all methods were performed in prcomp function of the stats R package, first separately for accordance to the relevant guidelines and regulations and each batch and then across all batches. Within batch, written informed consent was obtained from all subjects non-biological experimental variation (Sentrix Position before inclusion into the study. and chip ID) of normalized methylation values was tested for association with the Principal Component loadings DNA isolation, quantification, and genome-wide DNA (PCs). Chip ID was associated with the first PC (PC1) in methylation assays all three batches, which accounted for the 99% of the Peripheral blood mononuclear cells (PBMCs) of variation of samples. We therefore adjusted the beta patients with ADHD and controls were isolated using the values with the ComBat function of the SVA R package38 Ficoll density gradient method, and DNA was extracted for this variable. The effect of batch and sex on adjusted using the QIAamp DNA Mini Kit DNA Purification fol- methylation values of probes present in the three batches lowing manufacturer’s instructions (Qiagen, Hilden, after quality control (n = 744,227) was tested for asso- Germany). The quality of the samples was checked by ciation with the PCs estimated in the overall sample. NanoDrop® ND-1000 (Thermo Fisher Scientific, MA) Evidence of clustering according to batch was visually and by PicoGreen® (Thermo Fisher Scientific, MA). detected and statistically confirmed with a significant Genome-wide DNA methylation was assessed with the association of PC1 with batch (P-value < 2.20E−16). Illumina Infinium MethylationEPIC BeadChip Kit (EPIC Given that detailed smoking information was not array) (Illumina, San Diego, CA, USA) following sodium available for each individual, an individual smoking score bisulfite treatment of genomic DNA. (continuous measure) was generated based on DNA methylation sites known to be associated with current DNA methylation analysis based on ADHD diagnosis smoking using a method developed by Elliot and collea- Data preprocessing and normalization gues39. To account for methylation differences between The 203 samples included in this study were assayed in cell types, we estimated the cell-type composition using three batches, which were preprocessed and normalized the estimateCellCounts function of the FlowSorted. separately. Raw signal intensities of each probe were Blood.450k R package40. extracted using the Illumina Genome Studio software Probe-wise differential methylation analysis was per- (https://support.illumina.com) and were imported into formed using the lmFit function of the limma R pack- the R software (3.6.0 version; https://www.R-project.org) age41. Each CpG site was tested individually in a linear using the minfiData 0.2 package32. The bisulfite conver- regression model with normalized, corrected beta values sion control probes and the 59 single nucleotide poly- as the dependent variable and ADHD status as indepen- morphism (SNP) probes of the EPIC array were used to dent predictor, including covariates for sex, age, batch, calculate the bisulfite conversion reaction efficiency and smoking score and cell-type composition. Age was to confirm the absence of sample contamination, included as covariate in all the analysis, since it was sig- respectively. Sex was confirmed for all samples using the nificantly different between cases and controls. Multiple getSex function of the minfi R package33. The Horvath testing corrections were applied using false discovery rate Epigenetic Clock algorithm34 implemented by the agep (FDR) with a cut-off of 5%42. The qqman R package was function of the wateRmelon R package was used to used to generate the Manhattan plot. Rovira et al. Translational Psychiatry (2020)10:199 Page 4 of 12 The post-hoc power analysis in our sample calculated ADHD PRSs for each individual were generated with with the EPIC array online tool (https://epigenetics.essex. PRSice2 (https://choishingwan.github.io/PRSice/) includ- ac.uk/shiny/EPICDNAmPowerCalcs/)43 using the default ing sex and the first five PCs as covariates in the model. significance threshold (P-value < 9.42E−08) showed that To set an empirical threshold for the best-fit PRS, 1,000 6.12% of sites had > 90% power to detect a mean permutations were run. Information about the pre- methylation difference of 1%. imputation quality control at individual and SNP level At the differentially methylated CpG site, we tested the for the 195 individuals in the target sample and about association between DNA methylation and the exposure the phasing and imputation software used is described to at least one stressful life event, and to each stressful life elsewhere29. The European ancestry panel of the 1000 event separately using the lmFit function of the limma R Genomes Project was considered as reference for package. As 17 stressful life events were tested, Bonferroni the imputation (ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/) correction was set at P < 2.94E−03. We also tested the and best guess genotypes were filtered by excluding var- correlation between the number of stressful life events iants with MAF < 0.05, missing rate>0.01, Hardy- (sum of overall stressful life events and also separated in Weinberg Equilibrium (P < 1.00E−06). Ambiguous pre- and post-natal periods) and DNA methylation levels strand and multiallelic variants were removed and inde- using Spearman’s correlation. pendent SNPs (obtained using the clumping parameters To identify differentially methylated regions (DMRs), p1 = 1, p2 = 1, r2 = 0.2, kb=250 in PLINK1.946) present we used the Python module comb-p44 to group spatially in all individuals were included (n = 37,527). correlated CpG sites with a seed of P-value < 0.01 and 500 base pairs (bp) as the maximum distance. DMR P-values Enrichment analyses were corrected for multiple testing using the Šidák cor- We assessed whether probes in different categories: (i) rection45 and significant regions were defined as those showing a statistically significant proportion of methyla- with at least two probes and an adjusted P-value < 0.05. tion variance explained by additive genetic effects as DMRs were mapped to genes using the interface provided reported by Zeng et al.47; (ii) probes identified in previous by the minfi R package or the UCSC Genome Browser to EWASs on exposure to adverse live events48–50; (iii) identify the closest gene when no genes were mapped to a probes identified in previous EWASs on ADHD21,22 or region (https://genome.ucsc.edu/cgi-bin/hgGateway). ADHD symptoms6,9 or (iv) probes located in ADHD- Sensitivity analyses were conducted with the same associated loci identified through GWAS4,29,51 showed, on parameters described above for the probe-wise and average, a stronger association with adult ADHD than regional analyses excluding smoking score as covariate in other methylation sites by regressing our EWAS test the model. statistics (Zscore) on each CpG category as described by van Dongen et al. 9: DNA methylation analysis based on ADHD diagnosis controlling for ADHD polygenic burden Bioinformatic and statistical analyses jZscore j ¼ Intercept þ βcategory x  category x; ADHD polygenic burden was inferred using a Polygenic Risk Score (PRS) built in a subset of 195 individuals with where |Zscore| represents the absolute value of the Zscore genotype data available, from three different genotyping from our EWAS on adult ADHD, category x represents waves (Illumina HumanOmni1-Quad BeadChip (n = 3), whether a CpG belongs or not to a specific category and Illumina HumanOmni2.5-8 BeadChip (n = 29) and Infi- βcategory x represents the effect estimate for that category. nium™ Global Screening Array-24 v2.0 (n = 163) (Illu- A CpG was assigned to a category if it was associated to mina, San Diego, CA, USA), using summary statistics of the phenotype of interest according to the P-value the largest GWAS-MA performed to date on ADHD4, thresholds shown in Supplementary Table 1 [excel file]. with different P-value thresholds ((PT) < 1e−04, 5e−04, For GWAS, we considered CpG sites within windows of 0.001, 0.005, 0.01, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 1). None of 10 kb, 100 kb, and 1 Mb around significant variants the samples used in this study were included in this (Supplementary Table 1 [excel file]). For each enrichment GWAS-MA4, and thus did not contribute to defining the test, bootstrap standard errors were computed with 2,000 variants included in the PRS. bootstraps using the “simpleboot” R package. Bonferroni In this subset of 195 individuals, sensitivity analyses for correction was applied for multiple comparison correc- the differentially methylated sites and regions were con- tion (PBootstrap < 3.85E−03; accounting for the 13 analyses ducted with the same parameters used in the original conducted). EWAS but including the PRS explaining the most var- We also tested for enrichment of regulatory domains, iance (Nagelkerke’s R2) as an additional covariate to ontological categories and pathways, using CpG sites with control for ADHD polygenic risk burden. P-value<1.00E−05 in our results. For the enrichment Rovira et al. Translational Psychiatry (2020)10:199 Page 5 of 12 analysis of regulatory domains, ontological categories and P = 1.19E−07, P.adj=0.07) and the region in chromosome pathways, probes were annotated with the Illumina 11 were no longer significant and the other regions Human EPIC array annotation R package (“IlluminaHu- remained significant (Table 2). manMethylationEPICanno.ilm10b2.hg19”). The enrich- We subsequently tested whether the polygenic risk ment analyses for transcription factor binding sites burden for ADHD had an effect on the DNA methylation (TFBS) and DNase I hypersensitive sites (DHS) from the signatures. After constructing PRSs at different P-value ENCODE project52 were performed using a two-sided thresholds from the largest GWAS-MA on ADHD in Fisher’s 2×2 exact test. The enrichment analyses for GO children and adults4, the PRS explaining the most var- terms and KEGG, Reactome or Biocarta pathways were iance in our sample was found for PT = 0.001 (NSNPs = assessed using the gsameth function of the missMethyl R 490, R2 = 0.052, Pperm = 0.029), and was significantly package53. Gene sets denoting canonical pathways were higher in ADHD patients than controls (P = 3.10E−03; downloaded from MSigDB (http://www.broadinstitute. Supplementary Fig. 2). After adding it as a covariate to the org/gsea/msigdb), which integrates Kyoto Encyclopedia model fitted for the EWAS, we found that the cg07143296 of Genes and Genomes (KEGG) (http://www.genome.jp/ CpG site (logFC = 0.066, P = 1.60E−08, P.adj=0.012) and kegg/), BioCarta (http://www.biocarta.com/), Reactome three of the four genomic regions identified remained (https://reactome.org/) and Gene Ontology (GO) (http:// significant (Table 2). www.geneontology.org/) resources. We then tested whether CpG sites whose methylomic The datasets for this article are not publicly available variation is mainly explained by additive genetic effects because of limitations in ethical approvals and the sum- showed, on average, a stronger association with adult mary data will be available upon request. ADHD than other methylation sites included in the array, and found a significant enrichment of signal for adult Results ADHD among them (PBootstrap = 2.39E−04). In addition, Our sample consisted of 103 cases and 100 controls when we assessed the overlap between genetic and epi- after quality control. The distribution of sexes was not genetic signatures of ADHD, we found suggestive evi- significantly different between groups (χ2 = 2.60, P = dence of overlap between our EWAS results and probes 0.11), with 56% and 45% of cases and controls being male, annotated to ADHD-associated loci in the largest GWAS respectively. Age of participants was significantly different meta-analyses on ADHD across the lifespan or GWAS- between cases and controls (P = 3.61E−04), with a mean MA on ADHD symptoms in children (PBootstrap = 6.75E age of 31.90 (SD = 11.45) years in cases and of 37.25 (SD −03 and PBootstrap = 1.36E−02, respectively), but not with = 9.47) years in controls. In the case group, 35% of par- results of previous GWAS-MA on ADHD conducted ticipants experienced no stressful life events, 35% were separately in adults or children (Supplementary Table 1 exposed to at least one prenatal stressful life event and [excel file]). We also considered CpG sites differentially 54% were exposed to at least one of them after birth methylated in previous EWAS on individuals exposed to (Supplementary Table 2; Supplementary Fig. 1). adverse life events, on clinical ADHD or on ADHD We identified one differentially methylated CpG site, symptoms and found that CpG sites previously associated cg07143296, in the EWAS (P.adj = 0.033; Fig. 1a, b; Table with current vs never smoking and with maternal smok- 1; EWAS inflation factor λ = 0.67). This CpG lies 77 bp ing showed a highly significant enrichment of signal for upstream the PCNXL3 gene and was hypermethylated in adult ADHD (PBootstrap = 9.03E−18 and PBootstrap = 4.62E patients, with a mean difference of 0.2% between groups −14, respectively) (Supplementary Table 1 [excel file]). (Table 1, Fig. 2). When evaluating the effect of prenatal However, no overlap was detected with findings of pre- and postnatal stressful life events on the methylation vious EWASs on ADHD, on ADHD symptoms and on patterns of ADHD subjects at this CpG site, we found no physical/emotional neglect or abuse (Supplementary significant differences in the methylation levels between Table 1 [excel file]). individuals with ADHD exposed to stressful life events When we focused on the top 15 differentially methy- compared to those not exposed. The combined analysis of lated CpG sites (P < 1.00E−05) in our EWAS, we found multiple correlated CpG sites showed evidence of asso- no enrichment of regulatory domains (TFBS and DHS) ciation between ADHD and methylation levels in four from the ENCODE project52 nor ontological categories or genomic regions (P.adj < 0.02), with the most significant pathways from GO terms, KEGG, Reactome or Biocarta one spanning six CpG sites and located in the DENND2D (Supplementary Table 3 [excel file]). gene (P.adj = 2.52E−07; Table 2). The smoking score was not significantly different between cases and controls Discussion (mean score in cases = −2.42, mean in controls = −3.34, To the best of our knowledge, this is the first study P = 0.05). When we excluded it from the fitted model as evaluating DNA methylation signatures in a clinical a sensitivity analysis, cg07143296 (logFC = 0.0059, sample of adults with ADHD and testing whether Rovira et al. Translational Psychiatry (2020)10:199 Page 6 of 12 Fig. 1 Results of the epigenome-wide association study. a Manhattan plot. Horizontal line indicates 5% FDR significance threshold (P- value=6.72E−08). b Quantile-quantile plot. smoking status, polygenic risk burden for ADHD or anxiety-like behavior, circadian rhythmicity and drug exposure to stressful life events had an impact on the addiction60–62, ADK, whose deficiency may result in methylation signatures identified. altered dopaminergic function, attentional impairment, Methylation differences were found in regions that and learning impairments63,64, or LAT, whose genetic include genes related to cancer and pulmonary function variation has been associated with educational (DENND2D)54,55, neuroticism and regulation of histone attainment65. acetylation dynamics (PWWP2B)56,57 or regulation of The lack of overlap between our EWAS results and immune signaling (UBASH3A)58. We also identified a those from previous EWASs on ADHD in child- CpG site (cg07143296) significantly hypermethylated in hood6,10,11,20,22 is in line with the fact that genome-wide ADHD, located close to PCNXL3, a gene related to DNA methylation is highly age dependent34. Contrary to autoimmune diseases59. Although not achieving sig- some risk factors stably involved in ADHD throughout nificance after multiple comparison correction, CpG sites the lifespan, DNA methylation is developmental-stage in ADHD-related genes were found among the top ten specific and hence the patterns contributing to ADHD signals of the EWAS, including CREM, which has been susceptibility may differ over time. The absence of overlap previously associated with impulsivity, hyperactivity, between our results and findings from previous EWASs Table 1 Top 10 ADHD-associated differentially methylated CpG sites. Probe ID Location % logFC P P.adj Gene DHS DHS location Overlap transcription factor binding site cg07143296 11: 65383707 0.24 0.0062 4.41E−08 0.033 PCNXL3 Yes TSS200 POLR2A | SIN3A | RBBP5 | EGR1 | YY1 | MAX | ZBTB7A | MYC (+77 bp) cg05041517 1: 1021029 0.24 0.0105 1.53E−07 0.057 C1orf159 No – – cg01174734 10: 35485053 0.83 0.0152 4.54E−07 0.113 CREM Yes 1st Exon; Body; 5’UTR POLR2A | TCF7L2 | TCF12 | STAT5A | TBP | PML | NFATC1 | FOXM1 | BCL3 | EP300 | MXI1 | STAT3 | KAP1 | ATF3 | REST | GTF2F1 | JUND | MTA3 | NFIC | CEBPB | GABPA | MAX | RFX5 | TRIM28 | ATF2 | RELA | E2F1 | TAF1 | CBX3 | USF1 | RUNX3 | BCLAF1 | CREB1 | MAZ | CTCF | E2F4 | SMC3 | RAD21 | ZNF143 cg25889770 1: 161068063 0.48 0.0073 1.40E−06 0.242 KLHDC9 Yes TSS200 NFIC | PML | FOXM1 | SETDB1 | TFAP2C | POLR2A | RCOR1 | TBP | RFX5 | SIN3A | MXI1 | MAX | (+88 bp) SREBP1 | HSF1 | TAF1 | SIN3AK20 | RELA | RUNX3 | YY1 | CTCF | PAX5 | POU2F2 | ZNF143 | ELF1 | BCL3 | ATF3 | SMC3 Rovira et al. Translational Psychiatry (2020)10:199 cg15705054 1: 154927709 0.36 0.0063 1.89E−06 0.242 PBXIP1 Yes 5’UTR POLR2A | GATA1 | CBX3 | MAZ | MAX | SIN3AK20 | GATA2 | EP300 | SIN3A | RCOR1 | EGR1 | GABPA | TAL1 | MXI1 | RUNX3 | TCF3 | TBP | SPI1 | RFX5 | TCF12 | ELF1 | TBL1XR1 | FOXA1 | PAX5 cg05770546 10: 76004016 2.17 0.0295 1.96E−06 0.242 ADK Yes Body CTCF | ZNF143 | RAD21 | SMC3 cg05529890 16: 28997459 0.93 0.0124 3.40E−06 0.349 LAT Yes Exon boundary; Body – cg23874234 6: 57087441 0.81 0.0129 3.75E−06 0.349 RAB23 Yes TSS1500 POLR2A | PHF8 | TAF1 | TCF7L2 | E2F1 | MBD4 | TBP (−0.34 kb) cg16449840 12: 54718606 0.28 0.0058 5.00E−06 0.385 COPZ1 (+0.27 Yes TSS1500 HDAC8 | MYC | UBTF | E2F1 | ELK1 | MBD4 | RCOR1 | FOXM1 | RXRA | CBX3 | GTF2F1 | FOSL2 | kb) TBP | NR2F2 | JUND | BHLHE40 | IRF1 | CTCF | STAT3 | EP300 | MAFF | ZNF143 | MAX | MAZ | STAT1 | REST | GATA3 | ARID3A | SMC3 | RAD21 | RUNX3 | MAFK | MXI1 | GATA2 | USF2 | ATF3 | SIN3AK20 | USF1 | CTCFL | NFYB | GATA1 | RFX5 | E2F4 | ELK4 | ATF1 cg24479820 1: 2388073 0.91 0.0147 5.62E−06 0.385 PLCH2 (10.83 kb) No – – Location (chromosome: base pair), DNA methylation change between groups (%), log fold change estimates (logFC), P-values (P) and adjusted P-values (P.adj) are shown for each CpG site. Page 7 of 12 Rovira et al. Translational Psychiatry (2020)10:199 Page 8 of 12 mind that we used an estimated smoking score that might be a less accurate tool than clinical data, it has been postulated as a valid marker for current tobacco exposure13,39. We also report preliminary data supporting overlap between epigenetic signatures of ADHD and smoking- related traits or behaviors. Enrichment of top-ranking CpGs from previous EWASs on smoking behavior49 or maternal smoking50 was obtained. In addition, methyla- tion differences were identified in regions lying in or near genes (such as DENND2D or PWWP2B) related to phe- notypes where tobacco exposure is a key risk factor66–68, and maternal smoking, which increases risk of ADHD in the offspring69–71, was the most frequently prenatal stressful life event reported by participants with ADHD. To note, sixty-five percent of individuals with ADHD reported having been exposed to stressful life events, a circumstance that has been associated with the persis- tence of the disorder into adulthood72. Extreme familial stress was found among the most frequently reported Fig. 2 CpG‐specific DNA methylation levels. Boxplot showing the postnatal exposures in individuals with ADHD, which is levels of DNA methylation in cases and controls at cg07143296. not surprising given that the presence of ADHD has been associated to varying degrees of disturbances in family and marital functioning73–75. However, no effect of on ADHD in the adulthood period9,21 could be ascribed to stressful live events on DNA methylation patterns was differences in the characteristics of the samples and on the found in ADHD subjects. Given that our study lacked data array used (clinical vs population-based samples and EPIC on exposure to stressful live events in controls, larger vs Infinium Human Methylation 450K array9), to random studies including cases and controls are needed to variation and limited statistical power or, as previously understand the impact of environmental factors on DNA suggested by Meijer et al.21, to the fact that the epigenetic methylation patterns associated with ADHD. effects identified may not be those with the strongest The results of the present study should be interpreted in effect sizes on the phenotype21. the context of several limitations. First, the limited sample Results on the relationship between genetic and epige- size of the present EWAS, which should be viewed as a netic signatures in ADHD were not conclusive. We found pilot study whose findings await further replication. Sec- enrichment of signal for adult ADHD in CpGs whose ond, our study design allowed the assessment of methy- methylation variance is mainly explained by additive lation patterns in a restricted clinical sample of genetic effects47 and suggestive evidence of enrichment in medication-naïve subjects with no comorbid disorders. loci described in the largest GWAS-MA on ADHD4 and This design may have facilitated the identification of novel on ADHD symptoms51. However, no evidence was found epigenetic signatures, which may not have been possible for overlap between our EWAS results and loci from using a broader recruitment strategy. However, given that smaller GWAS-MAs on ADHD28 or for a substantial patients under medication and/or with lifetime comor- effect of the polygenic burden for ADHD on the methy- bidities were excluded and this group accounts for a not lation patterns identified. These inconsistent results negligible proportion of the overall ADHD group, further should be interpreted in the context of the limited sta- studies in larger samples including cases and controls tistical power of the EWAS and warrant further meeting common inclusion criteria, more relaxed in investigation. terms of medication or comorbid disorders, will be Our EWAS findings do not seem to be driven by an required to clarify whether the results obtained could be effect of current smoking since they were significant when generalized to a more realistic clinical situation. Third, the we adjusted the model for it. When excluding smoking low inflation factor obtained indicates that the distribu- status from the model, we did not detect an effect of tion of effect sizes in the present EWAS were not driven methylation on ADHD through smoking for cg07143296 by systematic biases but also suggests that our study had or for the region in chromosome 11 but we cannot rule limited statistical power and that the data may have been out a mediating effect for the remaining regions as their overcorrected, which may have prevented us from signal becomes more significant. Although bearing in detecting methylation signatures with small effect sizes. Table 2 ADHD-associated differentially methylated regions identified using comb-p. Chr Coordinates Gene annotation CpGs Covariates in the model Min P P P.adj 1 111743038- 111743411 DENND2D cg18924738, cg19269039, cg20317872, cg23184711, Sex, age, batch, smoking score and cell-type 8.48E−05 1.26E−10 2.52E−07 cg00619207, cg19268695 composition Sex, age, batch, smoking score, cell-type 5.19E−06 7.25E−12 1.45E−08 composition and PRS 111743200- 111743411 cg18924738, cg19269039, cg20317872, cg23184711, Sex, age, batch and cell-type composition 7.73E−05 3.49E−10 1.23E−06 cg00619207, cg19268695 10 134231487- 134231549 PWWP2B (-0.12 kb) cg09287328, cg03860038, cg23249922 Sex, age, batch, smoking score and cell-type 7.61E−03 1.50E−07 1.80E−03 composition Sex, age, batch, smoking score, cell-type 4.61E−03 8.49E−08 1.02E−03 composition and PRS Rovira et al. Translational Psychiatry (2020)10:199 Sex, age, batch and cell-type composition 9.41E−04 1.52E−08 1.82E−04 11 69241075- 69241093 LOC102724265 / AK094674 cg22203628, cg13647725 Sex, age, batch, smoking score and cell-type 7.61E−03 1.18E−07 4.86E−03 composition Sex, age, batch, smoking score, cell-type 2.28E−03 1.84E−08 7.60E−04 composition and PRS Sex, age, batch and cell-type composition 1.92E−03 6.18E−08 2.55E−03 21 43823797- 43823863 UBASH3A ( + 0.11 kb) cg20272209, cg27280688, cg10690747, cg13578652 Sex, age, batch, smoking score and cell-type 1.53E−03 1.84E−06 0.02 composition Sex, age, batch, smoking score, cell-type 4.61E−03 7.84E−06 0.09 composition and PRS 43823749- 43823863 cg20272209, cg27280688, cg10690747, cg13578652 Sex, age, batch and cell-type composition 2.52E−04 2.03E−09 1.33E−05 For each region, its coordinates (chromosome and start-end bp), CpGs included, minimum P-value in the region (Min P), non-corrected DMR P-value (P) and corrected P-value (P.adj) are shown along with the gene annotation obtained from the Illumina Human EPIC array annotation file and from UCSC when a gene was not annotated in a region. Results including polygenic risk score (PRS) for ADHD or excluding smoking score as covariate are also shown. Page 9 of 12 Rovira et al. Translational Psychiatry (2020)10:199 Page 10 of 12 And fourth, peripheral tissues used generally as proxies Universitat Autònoma de Barcelona, Barcelona, Spain. 6University of Exeter have limited utility for inferring variation in the brain76, Medical School, University of Exeter, Exeter, UK although these novel signatures identified in blood might Conflict of interest be used as biomarkers for the disorder. Paula Rovira, Dr. Sánchez-Mora, Dr. Pagerols, Laura Vilar-Ribó, Lorena Arribas, In summary, we conducted the largest study assessing Gemma Shireby, Dr. Hannon, Prof. Mill, Dr. Soler Artigas and Dr. Ribasés reported no biomedical financial interests or potential conflicts of interest. DNA methylation signatures in a clinical sample of adult Vanesa Richarte has served on the speakers for Eli Lilly, Rubio and Shire in the patients with ADHD. Our results suggest that ADHD last 5 years. She has received travel awards from Eli Lilly and Co. and Shire for polygenic risk burden or current smoking status do not participating in psychiatric meetings. The ADHD Program has received unrestricted educational and research support from Eli Lilly and Co., Janssen- change substantially the methylomic variation between Cilag, Shire, Rovi, Psious and Laboratorios Rubió in the past 2 years. Montserrat cases and controls, suggest an overlap between epigenetic Corrales received travel awards for taking part in psychiatric meetings from signatures of ADHD and smoking-related traits, and point Shire. Christian Fadeuilhe received fees to give talks for Shire, Ferrer, Italfarmaco and Otsuka in the last 5 years. He also received travel awards (air to an overlap between genetic and epigenetic signatures in tickets +hotel) for taking part in psychiatric meetings from Janssen-Cilag, ADHD. These results emphasize the need of additional Rubió, Shire, Lundbeck, Otsuka and Ferrer. Prof. Casas has received travel grants efforts in larger samples and the inclusion of stressful life and research support from Eli Lilly and Co., Janssen-Cilag, Shire and Lundbeck and served as consultant for Eli Lilly and Co., Janssen-Cilag, Shire and events in future studies to clarify the role of epigenetic Lundbeck. Prof. Ramos-Quiroga was on the speakers’ bureau and/or acted as mechanisms and environmental risk factors on ADHD consultant for Eli-Lilly, Janssen-Cilag, Novartis, Shire, Lundbeck, Almirall, across the lifespan. Braingaze, Sincrolab, Medice, and Rubió in the last 5 years. He also received travel awards (air tickets + hotel) for taking part in psychiatric meetings from Acknowledgements Janssen-Cilag, Rubió, Shire, Medice and Eli- Lilly. The Department of Psychiatry Over the course of this investigation, P.R. is a recipient of a predoctoral chaired by him received unrestricted educational and research support from fellowship from the Agència de Gestió d’Ajuts Universitaris i de Recerca the following companies in the last 5 years: Eli-Lilly, Lundbeck, Janssen- Cilag, (AGAUR), Generalitat de Catalunya, Spain (2016FI_B 00899). C.S.M. is a recipient Actelion, Shire, Ferrer, Oryzon, Roche, Psious, and Rubió. of a Sara Borrell contract from the Instituto de Salud Carlos III, Ministerio de Economía, Industria y Competitividad, Spain (CD15/00199). M.P has been a recipient of a predoctoral fellowship from the Vall d’Hebron Research Institute Publisher’s note (PRED-VHIR-2013). L.V. is a recipient of a predoctoral fellowship from the Springer Nature remains neutral with regard to jurisdictional claims in Instituto de Salud Carlos III, Spain (FI18/00285). M.S.A. was a recipient of a published maps and institutional affiliations. contract from the Biomedical Network Research Center on Mental Health (CIBERSAM), Madrid, Spain, and now is a recipient of a Juan de la Cierva Supplementary Information accompanies this paper at (https://doi.org/ Incorporación contract from the Ministry of Science, Innovation and 10.1038/s41398-020-0860-4). Universities, Spain (IJC2018-035346-I). M.R. is a recipient of a Miguel de Servet contract from the Instituto de Salud Carlos III, Spain (CP09/00119 and CPII15/ Received: 13 November 2019 Revised: 5 May 2020 Accepted: 15 May 2020 00023). The research leading to these results has received funding from the European Union H2020 Programme (H2020/2014-2020) under grant agreements no. 667302 (CoCA) and no. 728018 (Eat2beNICE), from the Instituto de Salud Carlos III (PI15/01789, PI16/01505, PI17/00289, PI18/01788, PI19/00721, and P19/01224), from the Pla estratègic de recerca i innovació en References salut (PERIS); Generalitat de Catalunya (MENTAL-Cat; SLT006/17/287) and from 1. Faraone, S. V. et al. Attention-deficit/hyperactivity disorder. Nat. Rev. Dis. Prim. the Agència de Gestió d’Ajuts Universitaris i de Recerca-AGAUR, Generalitat de 1, 15020 (2015). Catalunya (2017SGR1461) and cofinanced by the European Regional 2. Franke, B. et al. Live fast, die young? A review on the developmental trajec- Development Fund (ERDF) and by “la Marató de TV3” (092330/31),. The work tories of ADHD across the lifespan. Eur. Neuropsychopharmacol.: J. Eur. Coll. was also supported by the ECNP Network ‘ADHD across the Lifespan’ (https:// Neuropsychopharmacol. 28, 1059–1088 (2018). www.ecnp.eu/research-innovation/ECNP-networks/List-ECNP-Networks/). This 3. Kupper, T. et al. The negative impact of attention-deficit/hyperactivity disorder paper reflects only the authors’ views, and the European Union is not liable for on occupational health in adults and adolescents. Int. Arch. Occup. Environ. any use that may be made of the information contained therein. The authors Health 85, 837–847 (2012). are grateful to patients and controls who kindly participated in this research. 4. Demontis, D. et al. Discovery of the first genome-wide significant risk loci for The Microarray-based DNA methylation profiling was performed by the attention deficit/hyperactivity disorder. Nat. Genet. 51, 63–75 (2019). Spanish “Centro Nacional de Genotipado” (CEGEN‐ISCIII) (www.cegen.org). We 5. Rutten, B. P. & Mill, J. Epigenetic mediation of environmental influences in thank Dr. Walton for providing the list of 783 probes with P < 1.00E−03 in their major psychotic disorders. Schizophrenia Bull. 35, 1045–1056 (2009). EWAS on childhood ADHD symptoms. We thank Dr. Meijer for providing the 6. Walton, E. et al. Epigenetic profiling of ADHD symptoms trajectories: a pro- lists of 723 and 658 probes with P < 1.00E−03 in their EWAS on persistent spective, methylome-wide study. Mol. Psychiatry 22, 250–256 (2017). ADHD versus healthy controls and in their EWAS on persistent versus remittent 7. Mill, J. & Petronis, A. Pre- and peri-natal environmental risks for attention-deficit ADHD, respectively. We thank Dr. Mooney for providing the list of 679 probes hyperactivity disorder (ADHD): the potential role of epigenetic processes in with P < 1.00E−03 in their EWAS on childhood ADHD. We thank Dr. Zeng and mediating susceptibility. J. Child Psychol. Psychiatry Allied Discip. 49, 1020–1030 Dr. Haley for providing the list of probes of the Illumina EPIC array with a (2008). significant proportion of methylation variance explained by additive genetic 8. Hamza, M. et al. Epigenetics and ADHD: toward an integrative approach of effects. the disorder pathogenesis. J. Atten. Disord. 23, 655–664 (2019). 9. van Dongen, J. et al. Epigenome-wide association study of ADHD symptoms Author details in adults. Biol. Psychiatry https://doi.org/10.1016/j.biopsych.2019.02.016 (2019). 1 Psychiatric Genetics Unit, Group of Psychiatry, Mental Health and Addiction, 10. Wilmot, B. et al. Methylomic analysis of salivary DNA in childhood ADHD Vall d’Hebron Research Institute (VHIR), Universitat Autònoma de Barcelona, identifies altered DNA methylation in VIPR2. J. Child Psychol. Psychiatry Allied Barcelona, Spain. 2Department of Psychiatry, Hospital Universitari Vall d’Hebron, Discip. 57, 152–160 (2016). Barcelona, Spain. 3Biomedical Network Research Centre on Mental Health 11. Neumann, A. et al. Association between DNA methylation and ADHD (CIBERSAM), Instituto de Salud Carlos III, Madrid, Spain. 4Department of symptoms from birth to school age: a prospective meta-analysis. Preprint at Genetics, Microbiology, and Statistics, Faculty of Biology, University of https://www.biorxiv.org/content/10.1101/ (2019). Barcelona, Catalonia, Spain. 5Department of Psychiatry and Legal Medicine, Rovira et al. Translational Psychiatry (2020)10:199 Page 11 of 12 12. Li, M. et al. EWAS Atlas: a curated knowledgebase of epigenome-wide asso- 39. Elliott, H. R. et al. Differences in smoking associated DNA methylation patterns ciation studies. Nucleic Acids Res. 47(D1), D983–D988 (2019). in South Asians and Europeans. Clin. Epigenetics 6, 4 (2014). 13. Hannon, E. et al. Elevated polygenic burden for autism is associated with 40. Jaffe, A. E. FlowSorted.Blood.450k: Illumina HumanMethylation data on sorted differential DNA methylation at birth. Genome Med. 10, 19 (2018). blood cell populations. R package version 1.22.0. Bioconductor. https://doi.org/ 14. Loke, Y. J., Hannan, A. J. & Craig, J. M. The role of epigenetic change in autism 10.18129/B9.bioc.FlowSorted.Blood.450k (2019). spectrum disorders. Front. Neurol. 6, 107 (2015). 41. Ritchie, M. E. et al. limma powers differential expression analyses for RNA- 15. Sun, W. et al. Histone acetylome-wide association study of autism spectrum sequencing and microarray studies. Nucleic Acids Res. 43, e47 (2015). disorder. Cell 167, 1385–1397 e1311 (2016). 42. Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: a practical 16. Tremblay, M. W. & Jiang, Y. H. DNA methylation and susceptibility to autism and powerful approach to multiple testing. J. R. Stat. Soc.: Ser. B (Methodol.) 57, spectrum disorder. Annu. Rev. Med. 70, 151–166 (2019). 289–300 (1995). 17. Pries, L. K., Guloksuz, S. & Kenis, G. DNA methylation in schizophrenia. Adv. Exp. 43. Mansell, G. et al. Guidance for DNA methylation studies: statistical insights Med. Biol. 978, 211–236 (2017). from the Illumina EPIC array. BMC genomics 20, 366 (2019). 18. Viana, J. et al. Schizophrenia-associated methylomic variation: molecular sig- 44. Pedersen, B. S., Schwartz, D. A., Yang, I. V. & Kechris, K. J. Comb-p: software for natures of disease and polygenic risk burden across multiple brain regions. combining, analyzing, grouping and correcting spatially correlated P-values. Hum. Mol. Genet. 26, 210–225 (2017). Bioinformatics 28, 2986–2988 (2012). 19. Ludwig, B. & Dwivedi, Y. Dissecting bipolar disorder complexity through 45. Šidák, Z. K. Rectangular confidence regions for the means of multivariate epigenomic approach. Mol. psychiatry 21, 1490–1498 (2016). normal distributions. J. Am. Stat. Assoc. 62, 626–633 (1967). 20. Gervin, K., Nordeng, H., Ystrom, E., Reichborn-Kjennerud, T. & Lyle, R. 46. Purcell, S. et al. PLINK: a tool set for whole-genome association and Long-term prenatal exposure to paracetamol is associated with DNA population-based linkage analyses. Am. J. Hum. Genet. 81, 559–575 (2007). methylation differences in children diagnosed with ADHD. Clin. Epige- 47. Zeng, Y. et al. Parent of origin genetic effects on methylation in humans are netics 9, 77 (2017). common and influence complex trait variation. Nat. Commun. 10, 1383 (2019). 21. Meijer et al. Genome-wide DNA methylation patterns in persistent attention- 48. Cecil, C. A. et al. Epigenetic signatures of childhood abuse and neglect: deficit/hyperactivity disorder and in association with impulsive and callous implications for psychiatric vulnerability. J. Psychiatr. Res. 83, 184–194 (2016). traits. Front. Genet 11, 16 (2020). 49. Joehanes, R. et al. Epigenetic signatures of cigarette smoking. Circulation 22. Mooney, M. A. et al. Large epigenome-wide association study of childhood Cardiovascular Genet. 9, 436–447 (2016). ADHD identifies peripheral DNA methylation associated with disease and 50. Joubert, B. R. et al. DNA methylation in newborns and maternal smoking in polygenic risk burden. Transl. Psychiatry 10, 8 (2020). pregnancy: genome-wide consortium meta-analysis. Am. J. Hum. Genet. 98, 23. Heinrich, H. et al. Attention, cognitive control and motivation in ADHD: linking 680–696 (2016). event-related brain potentials and DNA methylation patterns in boys at early 51. Middeldorp, C. M. et al. A genome-wide association meta-analysis of atten- school age. Sci. Rep. 7, 3823 (2017). tion-deficit/hyperactivity disorder symptoms in population-based pediatric 24. Sengupta, S. M., Smith, A. K., Grizenko, N. & Joober, R. Locus-specific DNA cohorts. J. Am. Acad. Child Adolesc. Psychiatry 55, 896–905 e896 (2016). methylation changes and phenotypic variability in children with attention- 52. The ENCODE Project Consortium. An integrated encyclopedia of DNA ele- deficit hyperactivity disorder. Psychiatry Res. 256, 298–304 (2017). ments in the human genome. Nature 489, 57–74 (2012). 25. van Mil, N. H. et al. DNA methylation profiles at birth and child ADHD 53. Phipson, B., Maksimovic, J. & Oshlack, A. missMethyl: an R package for ana- symptoms. J. Psychiatr. Res. 49, 51–59 (2014). lyzing data from Illumina’s HumanMethylation450 platform. Bioinformatics 32, 26. Xu, Y. et al. Multiple epigenetic factors predict the attention deficit/hyper- 286–288 (2016). activity disorder among the Chinese Han children. J. Psychiatr. Res. 64, 40–50 54. Kanda, T., Ishii, K., Kawaguchi, H., Kitajima, K. & Takenaka, D. High signal intensity (2015). in the dentate nucleus and globus pallidus on unenhanced T1-weighted MR 27. Dadds, M. R., Schollar-Root, O., Lenroot, R., Moul, C. & Hawes, D. J. Epigenetic images: relationship with increasing cumulative dose of a gadolinium-based regulation of the DRD4 gene and dimensions of attention-deficit/hyperactivity contrast material. Radiology 270, 834–841 (2014). disorder in children. Eur. Child Adolesc. Psychiatry 25, 1081–1089 (2016). 55. Shrine, N. et al. New genetic signals for lung function highlight pathways and 28. Park, S. et al. Associations between serotonin transporter gene (SLC6A4) chronic obstructive pulmonary disease associations across multiple ancestries. methylation and clinical characteristics and cortical thickness in children with Nat. Genet. 51, 481–493 (2019). ADHD. Psychological Med. 45, 3009–3017 (2015). 56. Okbay, A. et al. Genetic variants associated with subjective well-being, 29. Rovira, P. et al. Shared genetic background between children and adults with depressive symptoms, and neuroticism identified through genome-wide attention deficit/hyperactivity disorder. Neuropsychopharmacology 1–10 analyses. Nat. Genet. 48, 624–633 (2016). (2020). 57. Zhang, T. et al. A variant NuRD complex containing PWWP2A/B excludes 30. Sanchez-Mora, C. et al. Epigenetic signature for attention-deficit/hyperactivity MBD2/3 to regulate transcription at active genes. Nat. Commun. 9, 3798 disorder: identification of miR-26b-5p, miR-185-5p, and miR-191-5p as (2018). potential biomarkers in peripheral blood mononuclear cells. Neuropsycho- 58. Ge, Y., Paisie, T. K., Newman, J. R. B., McIntyre, L. M. & Concannon, P. UBASH3A pharmacology 44, 890–897 (2019). mediates risk for type 1 diabetes through inhibition of T-cell receptor-induced 31. Epstein, J. N., Johnson, D. E. & Conners, C. K. Conners Adult ADHD Diagnostic NF-kappaB signaling. Diabetes 66, 2033–2043 (2017). Interview for DSM-IV. (Multi-Health Systems, North Tonawanda, NY, 1999). 59. Sun, C. et al. High-density genotyping of immune-related loci identifies new 32. Hansen, K. D., Aryee, M. & Timp, W. minfiData: example data for the Illumina SLE risk variants in individuals with Asian ancestry. Nat. Genet. 48, 323–330 Methylation 450k array. R package version 0.24.0. Bioconductor. https://doi. (2016). org/10.18129/B9.bioc.minfiData (2017). 60. Maldonado, R., Smadja, C., Mazucchelli, C. & Sassone-Corsi, P. Altered emo- 33. Aryee, M. J. et al. Minfi: a flexible and comprehensive bioconductor package tional and locomotor responses in mice deficient in the transcription factor for the analysis of Infinium DNA methylation microarrays. Bioinformatics 30, CREM. Proc. Natl Acad. Sci. 96, 14094–14099 (1999). 1363–1369 (2014). 61. Lahti, T. A. & Partonen, T. CREM mutations and ADHD symptoms. Med. 34. Horvath, S. DNA methylation age of human tissues and cell types. Genome Hypotheses 72, 544–545 (2009). Biol. 14, R115 (2013). 62. Miller, M. L. et al. Ventral striatal regulation of CREM mediates impulsive action 35. Pidsley, R. et al. A data-driven approach to preprocessing Illumina 450K and drug addiction vulnerability. Mol. Psychiatry 23, 1328–1335 (2018). methylation array data. BMC genomics 14, 293 (2013). 63. Boison, D. & Aronica, E. Comorbidities in neurology: is adenosine the common 36. Chen, Y. A. et al. Discovery of cross-reactive probes and polymorphic CpGs in link? Neuropharmacology 97, 18–34 (2015). the Illumina Infinium HumanMethylation450 microarray. Epigenetics 8, 64. Sandau, U. S. et al. Adenosine kinase deficiency in the brain results in mala- 203–209 (2013). daptive synaptic plasticity. J. Neurosci. 36, 12117–12128 (2016). 37. McCartney. et al. Identification of polymorphic and off-target probe binding 65. Lee, J. J. et al. Gene discovery and polygenic prediction from a genome-wide sites on the Illumina Infinium MethylationEPIC BeadChip. Genomics Data 9, association study of educational attainment in 1.1 million individuals. Nat. 22–24 (2016). Genet. 50, 1112–1121 (2018). 38. Johnson, W. E., Li, C. & Rabinovic, A. Adjusting batch effects in microarray 66. Ling, B. et al. Suppression of non-small cell lung cancer proliferation and expression data using empirical Bayes methods. Biostatistics 8, 118–127 (2007). tumorigenicity by DENND2D. Lung Cancer 79, 104–110 (2013). Rovira et al. Translational Psychiatry (2020)10:199 Page 12 of 12 67. Zhang, T. et al. Downregulation of miR-522 suppresses proliferation and large Swedish population-based study of twins. J. Atten. Disord. 16, metastasis of non-small cell lung cancer cells by directly targeting DENN/ 13–22 (2012). MADD domain containing 2D. Sci. Rep. 6, 19346 (2016). 73. Eakin, L. et al. The marital and family functioning of adults with ADHD and 68. Park, S. L. et al. Mercapturic acids derived from the toxicants acrolein and their spouses. J. Atten. Disord. 8, 1–10 (2004). crotonaldehyde in the urine of cigarette smokers from five ethnic groups with 74. Harpin, V. A. The effect of ADHD on the life of an individual, their family, and differing risks for lung cancer. PLoS ONE 10, e0124841 (2015). community from preschool to adult life. Arch. Dis. Child 90(Suppl 1), i2–i7 69. Joelsson, P. et al. Prenatal smoking exposure and neuropsychiatric comor- (2005). bidity of ADHD: a finnish nationwide population-based cohort study. BMC 75. Schermerhorn, A. C. et al. Offspring ADHD as a risk factor for parental marital psychiatry 16, 306 (2016). problems: controls for genetic and environmental confounds. Twin Res. Hum. 70. Huang, L. et al. Maternal smoking and attention-deficit/hyperactivity disorder Genet. 15, 700–713 (2012). in offspring: a meta-analysis. Pediatrics 141, e20172465 (2018). 76. Hannon, E., Lunnon, K., Schalkwyk, L. & Mill, J. Interindividual methylomic 71. He, C. W., Liao, C. P. & Pan, C. L. Wnt signalling in the development of axon, variation across blood, cortex, and cerebellum: implications for epigenetic dendrites and synapses. Open Biol. 8, 180116 (2018). studies of neurological and neuropsychiatric phenotypes. Epigenetics 10, 72. Friedrichs, B., Igl, W., Larsson, H. & Larsson, J. O. Coexisting psychiatric 1024–1032 (2015). problems and stressful life events in adults with symptoms of ADHD-a