Obsessive-compulsive disorder (OCD) is a severe neuropsychiatric disorder with prevalence rates among children and adolescents comparable to those of the adult population (1–3%) [1–3]. The symptom presentation shows a bimodal pattern, hence 50% of OCD patients report an age of onset during childhood or adolescence, respectively [1]. The cardinal features of OCD are obsessions and compulsions. Obsessions are intrusive thoughts, urges, images, or impulses, which are unpleasant and often frightening. Compulsions are repetitive, often stereotypic semi-voluntary movements or mental acts, which are performed to reduce the anxiety or the bodily reaction. OCD is a heterogeneous disorder in regard to its clinical presentation [4–6]. Varying age of onset has been suggested to point at two distinguishable subtypes characterized by differences in affected gender, global OCD symptom severity, phenotypic presentation, and comorbid conditions [7, 8]. Furthermore, OCD symptoms in children and adolescents have been clustered into three clinical dimensions [9]. The three dimensions include: 1. harm/sexual dimension, which was correlated to the occurrence of anxiety, 2. symmetry and hoarding dimension, which was most frequently seen in females and was associated with anxiety and depression, and 3. contamination and cleaning dimension, which was negatively associated with tics and attention deficit hyperkinetic disorder (ADHD) [9]. In the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-V) two specifiers (insight and comorbid tics) have been introduced which are used to clarify the OCD diagnosis [10].
The complex and heterogeneous nature of OCD has complicated efforts to identify causal mechanisms of the condition. It has been suggested that the onset and exacerbation of OCD symptoms are dependent on genetic factors [11, 12], with reports on heritability point estimates reaching 0.37 [13]. A recent study based on Swedish-born individuals confirmed that common genetic risk variants account for most of the heritable variation in OCD and suggested polygenic architecture of this complex disorder [14]. Additionally, a population-based study suggested that the increased familial risk of OCD was attributable to additive genetic factors (47%), but that non-shared environmental factors were estimated to be as important as the genetic contributions [15]. It has thus become evident that a combination of both genetic variation and environmental exposures together modify risk of developing OCD and impact its clinical trajectory, what emphasizes the need for an integrated approach to study its molecular pathophysiology [16–18]. Interestingly, both genetic and environmental risk factors were shown to impact epigenetic regulation of the genome [19, 20], suspected to also contribute to development of OCD.
Epigenetic regulation of the mammalian genome relies on DNA methylation (DNAm), posttranslational histone modifications, and non-coding RNAs. Among these epigenetic modifications DNAm, characterized by an addition of methyl group to the 5th position of cytosine base, is currently the most studied and the best understood [21–23]. DNAm has direct effect on chromatin structure, gene expression levels, alternative splicing, and genome stability. The methyl group can also be removed from DNA by a combination of oxidation reactions and base excision repair mechanisms leading to active demethylation of cytosine [24]. This plasticity and dynamic regulation of DNAm patterns across the mammalian genome allow for environmental adaptations, transient changes and long-term alterations of a cell’s transcriptomic profile [25], which may have a causative influence on common disorders. Epigenetics is thus regarded as an important mediator of environmental signals into gene regulation. Interestingly, DNAm patterns change as a function of aging and between sexes, which may aid the explanation of age-dependent liability to complex disorders, as well as differences in prevalence and disorder presentation between sexes [26, 27].
In the context of OCD, differential DNAm has been examined both by candidate-gene approaches and epigenome-wide association studies (EWASes) [21, 22, 28–33]. Candidate gene studies have consistently shown an association between DNA hypermethylation of the oxytocin receptor (OXTR) gene in adult OCD patients compared to healthy controls [29, 32, 34]. Within OCD cases, baseline hypermethylation in two target sequences, located in exon III of the OXTR gene (OXTR1 and OXTR2), correlated to OCD severity[32] and predicted impaired treatment response, which especially applied to obsessive symptoms [32, 34]. Additionally, hypermethylation at cg04523291, located in OXTR, was associated with OCD diagnosis, with impaired treatment response in OCD cases, as well as stressful life events among control individuals [32].
Also, in addition to epigenetic changes in OXTR, differential DNAm in monoamine oxidase A (MAOA) has been associated with OCD, with hypomethylation in the MAOA promotor region observed in patients diagnosed with the disorder [35]. Moreover, the serotonic neurotransmitter system has been associated with OCD, and both genetic and epigenetic variation in SLC6A4 (serotonin transporter gene) were significantly associated with OCD in pediatric patients compared to controls [28]. Another study showed that a lower baseline DNAm at the SLC6A4 promotor in OCD patients was associated with an impaired treatment response, evaluated by changes in Y-BOCS scores [31]. Additionally, DNAm levels of the brain-derived neurotrophic factor (BDNF) were reported to be significantly lower in OCD patients compared with control individuals, a result that was comparable across various peripheral tissues (saliva and blood cells) [23]. D’Addario and co-authors examined DNAm and hydroxymethylation of BDNF in adults with OCD compared to healthy controls [36]. The authors showed an increased expression of the BDNF gene in patients with OCD, as well as lower DNAm and higher hydroxymethylation levels at the gene’s canonical promotor. The authors suggested that transcriptional regulation of BDNF involves epigenetic modifications and suggested that the changes resulted from long-term pharmacotherapy [36].
We have previously examined the DNAm profile of selected genes in blood spots from i) neonates later diagnosed with OCD, and ii) blood from the same children/adolescents at the time of diagnosis along with age- and sex-matched controls [30]. The selected genes belonged to the glutaminergic, the serotonergic and dopaminergic systems, as well as neurotrophic genes and sex hormone receptor genes previously associated with OCD. The pilot study did not show any significant differential DNAm after correction for multiple testing. However, a support was shown for a DNAm difference in the gamma-aminobutyric acid (GABA) B receptor 1 in blood samples at birth and for the estrogen receptor 1 (ESR1), the myelin oligodendrocyte glycoprotein (MOG), and the brain-derived neurotrophic factor (BDNF) in blood samples at time of diagnosis. DNAm profiles of GABBR1 and MOG suggested to be associated with baseline severity and treatment effect, while DNAm profile of ESR1 was suggested associated with baseline severity. Even though, the study failed to show any significant differential DNAm after correction for multiple testing, the study pointed towards a possible regulatory role for sex hormones, factors affecting nerve myelination and function, and factors related to the immune system [28].
In addition to candidate gene approaches, several EWASes of OCD have been performed. Two EWASes have examined the differential DNAm in a pediatric population. Goodman and co-authors examined the differential DNAm in saliva from children with either ADHD or OCD [22], showing a differential DNAm in the C13orf39, C170rf54, DNAJC15, LLGL2, POLS, MAD1L1, MGC87042, PTPRN2 and SGK2 genes associated with both OCD and ADHD, as well as with the severity of the disorders. Another study in children and adolescents with OCD showed a differential DNAm in two genes PIWIL1 and MIR886. Hypermethylation of the CpGs in the PIWIL1 and MIR886 was associated with an up-regulation of expression levels of related genes which again was associated with response to cognitive behavioral therapy (CBT) [21]. In an EWAS exploring DNAm in blood samples collected from adults with OCD, differential DNAm was found in 2190 genes, where pathway analyses indicated an involvement in regulation of actin cytoskeleton, cell adhesion molecules (CAMs), actin binding and transcription regulator activity [33]. However, as the EWAS did not adjust directly for the common confounders in the methylomic data (such as tobacco smoking, technical variation in the methylation arrays, blood cell heterogeneity, etc.) a thorough replication of its findings is warranted. For example, a similar EWAS of generalized anxiety disorder and OCD, performed also in blood samples collected from a Chinese population performed on normalized DNAm data and adjusted for batch effects and blood cell proportion, but not tobacco smoking, identified much less variation in DNAm levels to be associated with the disorder [37]. The study reported differential DNAm at several single sites and genomic regions, especially highlighting epigenomic dysregulation at single CpG sites annotated to RIOK3, PEX11G, DNASE2 genes and differentially methylated regions (DMRs) annotated to CD47, USP6NL, and CYP2W1 as the most associated findings [37]. Another EWAS of OCD in an adult population found significant DNAm variation in the region of the microRNA 12136 gene (MIR12136) and in the regions of the MT-RNR2 (humanin) like 8 gene and the MT-RNR2 like 3 and MT-RNR2 like 2 genes [38].
As the abovementioned studies provided preliminary evidence on epigenetic contributions to development, severity, and treatment outcomes of OCD, more thorough studies are necessary to confirm role of epigenetic dysregulation in pathophysiology of the disorder. The aim of this project was to create a matched OCD case-control cohort to study epigenetic risk factors and lifetime epigenetic trajectories associated with OCD. This unique cohort had DNAm profiled at two developmental stages, at birth and at adolescence, as well as before and after treatment. This will allow for cross-sectional as well as longitudinal studies of epigenetic contributions to the disorder, as well as treatment response. The present paper presents a thorough description of the cohort and findings from the cross-sectional study.