Introduction

Obsessive-compulsive disorder (OCD) is a common neuropsychiatric disorder consisting of unwanted thoughts (obsessions) and/or repetitive behaviors (compulsions) (Pauls et al. 2014). The lifetime prevalence of OCD is reported to be 1.6–2.3 % (Kessler et al. 2005; Ruscio et al. 2010), and it has been considered one of the leading causes of life disturbance (Michaud et al. 2006). Substantial numbers of neuroimaging studies have demonstrated the presence of brain abnormalities in OCD (Menzies et al. 2008; Nakao et al. 2014; Peng et al. 2012; Piras et al. 2013, 2015; Radua and Mataix-Cols 2009). The most widely accepted model of OCD proposes that abnormalities of the fronto-striatal circuit, involving the frontal cortex, anterior cingulate cortex (ACC), striatum, and thalamus, play an important role in its pathophysiology (Cummings 1993; Graybiel and Rauch 2000; Menzies et al. 2008; Milad and Rauch 2012; Saxena et al. 1998). Furthermore, recent evidence has implicated abnormalities in additional brain regions involving the angular and supramarginal gyri, parietal lobe, insula, occipital lobe, and cerebellum in OCD patients (Menzies et al. 2008; Nishida et al. 2011; Piras et al. 2015; Song et al. 2011). These findings indicate that the pathophysiology of OCD involves a widespread neural network.

OCD is also well known as a clinically heterogeneous disorder that would be better understood as a dimensional disorder consisting of multiple overlapping obsessive-compulsive (OC) symptom dimensions (Mataix-Cols et al. 2005). Additionally, it is speculated that one of the reasons for the inconsistency of neuroimaging findings of OCD (Piras et al. 2015) is its clinical heterogeneity. Several studies have used voxel-based morphometry (VBM) to investigate the association between symptom dimensions and gray and white matter volume in OCD (Alvarenga et al. 2012; Gilbert et al. 2008; van den Heuvel et al. 2009; Lázaro et al. 2014a, 2014b; Pujol et al. 2004; Valente et al. 2005), but their results were inconsistent in respect to the relationship between OC symptom dimensions and gray and white matter volumes. From a clinical point of view, OCD patients have tended to present plural OC symptom dimensions in conjunction with their particular severity profile. This means that it is necessary to consider all OC symptoms together when accumulating data for OCD research.

Brain alterations due to the coexistence of depression symptoms and age at onset were reported in some studies in OCD patients. Cardoner et al. (2007) reported that lifetime major depressive disorder contributed to gray matter volume alterations in OCD patients. Christian et al. (2008) reported that OCD patients showed significantly larger gray matter volume in the left thalamus and also that OCD patients without major depression showed larger gray matter volume in the bilateral thalamus and left orbitofrontal cortex compared with healthy controls. As for brain abnormality associated with different age at onset, Rosso et al. (2014) suggested that age at onset may be a moderator of some of the white matter changes in pediatric OCD, and Busatto et al. (2001) found differences in regional cerebral blood flow between early onset OCD patients and late onset OCD patients. More recently, in terms of the presence of comorbid autistic spectrum disorders (ASD) in OCD, Cath et al. (2008) reported phenomenological overlapping between comorbid ASD and pure OCD in autistic phenomena by using Autism-Spectrum Quotient (AQ) (one of the screening tools of ASD; Baron-Cohen et al. 2001) subscales. Also, a previous study reported that ASD patients showed structural brain alterations in the lateral occipital lobe, pericentral region, medial temporal lobe, basal ganglia, and proximate to the right parietal operculum (Nickl-Jockschat et al. 2012). Based on these reports, in a previous study from our team, Kobayashi et al. (2015) investigated the correlations between AQ scores and regional gray matter volumes in patients with OCD. They found a positive correlation between AQ scores and regional gray matter volume in the left dorsolateral prefrontal cortex (DLPFC) and amygdala.

Considering these findings, we hypothesized that gray and white matter volumes in specific brain regions contributed to each of the obsessive-compulsive symptom dimensions. To investigate the relationship between gray and white matter volume and symptom dimensions of OCD while taking into account the effects of ASD traits, we applied all severity scores of OC dimensions of OCI-R simultaneously and AQ scores as covariates using VBM.

Methods

Subjects

Patients were recruited from outpatients of Chiba University Hospital, Japan. All patients were diagnosed as OCD by a trained interviewer using the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I) (Spitzer et al. 1997). We employed the Yale-Brown Obsessive-Compulsive Scale (Y-BOCS) (Goodman et al. 1989) for assessment of OC symptom severity. We also employed OCI-R to determine the profiles of the six OC symptom dimensions of each patient (washing, checking, ordering, obsessing, hoarding, and neutralizing; the severity of each dimension was assessed on a scale of 0 to 12). Patients with a Y-BOCS score of 16 or higher and a total intelligence quotient (IQ) of 80 or higher as assessed by Wechsler Adult Intelligence Scale (WAIS-III) (Wechsler 1997) were included. The patients ranged in age from 18 to 48 years. Exclusion criteria were neurological disorders, schizophrenia and related disorders including delusional disorder or psychotic disorders, substance dependencies, organic brain diseases, and severe physical diseases. Handedness of the participants was determined by Edinburgh Handedness Inventory (Oldfield 1971). We also measured ASD traits of patients using the AQ scale (Wakabayashi et al. 2006; Wakabayashi et al. 2004). The Institutional Research and Ethics Committee of the Graduate School of Medicine, Chiba University, approved the study (No. 1330), and written informed consent for the study was obtained from each subject before the assessments began. The trial was registered as UMIN000008765.

MRI acquisition

All subjects underwent T1-weighted MRI by scanner equipped with a 32-channel phased-array head coil (Discovery MR750 3.0 T; GE Healthcare). Images were collected by 3D fast spoiled gradient-echo (FSPGR) sequence (echo time: 3.164 ms; repetition time: 8.124 ms; flip angle: 15°; acquisition matrix: 256 × 256; slice thickness: 1 mm; field of view: 25.6 × 25.6 cm2; number of excitations: 1; bandwidth: 31.25 kHz; inversion time: 420 ms; acceleration factor: 2).

MRI data processing

We processed T1-weighted MR images using Statistical Parametric Mapping 8 (SPM8, Wellcome Institute of Neurology, University College London, UK) running under MATLAB R2013a (The MathWorks Inc., Natick, MA, USA). The VBM8 toolbox (http://dbm.neuro.uni-jena.de/vbm/), which is an extension of the unified segmentation model consisting of spatial normalization, bias field correction, and tissue segmentation (Ashburner and Friston 2005), was used for preprocessing the images. Registration to the stereotactic space of the Montreal Neurological Institute (MNI) consisted of linear affine transformation and nonlinear deformation using high-dimensional Diffeomorphic Anatomical Registration through Exponential Lie Algebra (DARTEL) normalization (Ashburner 2007). The normalized and segmented images were modulated by applying a nonlinear deformation, which allows comparison of absolute amounts of tissue corrected for individual differences in brain size (Klein et al. 2009). Finally, bias-corrected, modulated, and warped tissue maps were smoothed with an 8-mm full width at half maximum Gaussian kernel. Voxel resolution of smoothed images was 1.5 × 1.5 × 1.5 mm.

Statistical analysis

Statistical analysis was also performed with SPM8, which implemented a general linear model. We investigated correlations between the score of each OC symptom dimension and the gray and white matter volume by multiple regression analysis. First, we performed whole-brain analysis to determine the correlation between OCI-R scores and gray and white matter volume in OCD patients. The significance level was initially set at p < 0.001 uncorrected for multiple comparisons. After that, each dimension analysis was reported as significant if the cluster size survived at q < 0.05, false discovery rate (FDR) corrected for multiple comparisons at the cluster level, and the cluster size was obtained after applying gray matter or white matter mask. We treated all of the remaining OCI-R scores as covariates for obtaining specific brain alterations in each symptom dimension. Further, gender, age, and AQ scores were used as covariates for all analyses to control for potentially confounding variables. The anatomic location of each resulting cluster was determined using the MRI Atlas (Oishi et al. 2010).

Results

Forty-four patients with OCD were initially entered into this study, but eleven were then excluded on the basis of the exclusion criteria. Thus, thirty-three patients were analyzed. Their detailed demographics are summarized in Table 1. We investigated the correlations between respective obsessive-compulsive symptom dimension scores and gray and white matter volumes in OCD patients in consideration of the effects of ASD traits. There was a significant negative correlation between washing scores and regional brain volumes in the right thalamus (peak MNI coordinates: x, 3; y, −18; z, 12; cluster size, 527; Fig. 1 and Table 2). Further, there was a significant negative correlation between hoarding scores and regional brain volumes in the left angular gyrus white matter (peak MNI coordinates: x, −42; y, −45; z, 33; cluster size, 1761; Fig. 2 and Table 2). As mentioned in the Introduction, comorbid depressive symptoms (as measured using the BDI) and age at OCD symptom onset may differentially affects brain morphology. Then, we analyzed the data using the BDI scores as a nuisance covariate in gray matter. We found a negative correlation between washing dimension score and the volume of the left superior temporal gyrus, left thalamus, and the left postcentral gyrus in addition to the right thalamus, the same result as before using BDI scores as a nuisance covariate (Fig. S1 and Table S1). In terms of white matter, only cluster size and peak MNI coordinates were changed from 1761 (x, −42; y, −45; z, 33) to 1229 (x, −42; y, −43; z, 31) in angular gyrus white matter, the same region as before using the BDI scores. In addition, we analyzed the data using onset as a nuisance covariate in gray matter. As a result, only cluster size was changed from 527 to 493 in the right thalamus.

Table 1 Clinical characteristics of patients with OCD
Full size table
Fig. 1
figure 1

Correlations between washing symptom dimension scores and right thalamus volumes in OCD patients. The negative correlation between regional gray matter volumes in the right thalamus and the washing dimension scores. Results are shown at q < 0.05, false discovery rate (FDR) corrected for multiple comparisons at the cluster level. Color bar shows t-value. Covariates are the remaining five dimension scores, gender, age, and AQ scores. L; left

Full size image
Table 2 Negative correlations between OCI-R variables and brain volumes in OCD patients
Full size table
Fig. 2
figure 2

Correlations between hoarding symptom dimension scores and the left angular gyrus white matter volumes in OCD patients. The negative correlation between regional white matter volumes in the left angular gyrus and the hoarding dimension scores. Results are shown at q < 0.05, false discovery rate (FDR) corrected for multiple comparisons at the cluster level. Color bar shows t-value. Covariates are the remaining five dimension scores, gender, age, and AQ scores. L; left

Full size image

Discussion

To our knowledge, this is the first study to investigate the relationship between OC symptom dimensions and brain morphometry while considering the influence of age, gender, other OC symptom dimensions and ASD traits. Focusing on symptom dimensions to investigate the clinical heterogeneity of OCD in this study, we found significant correlations between washing scores and gray matter volumes in the right thalamus (Fig. 1 and Table 2) and between hoarding scores and volumes in the left angular gyrus white matter (Fig. 2 and Table 2). As far as we know, no studies had taken into consideration the influence of the severity scores of the other OC symptom dimensions using OCI-R when investigating brain regions with gray and white matter volumes showing significant correlations with each symptom dimension score in a whole-brain analysis. The thalamus was identified as one of the important regions in the fronto-striatal circuit and is one node within a network implicated in the pathophysiology of OCD (Menzies et al. 2008; Saxena et al. 1998). The pathway projects from the cortex to the thalamus via the striatum and back to the cortex (Alexander et al. 1986). Dysfunction in this circuit has been considered to represent the pathophysiology of this disorder. Previous neuroimaging studies focusing on the OC symptom dimensions also investigated the relationship between this circuit and symptom dimensions. In the washing symptom dimension, several neuroimaging studies have reported these brain abnormalities. Lázaro et al. (2014a) found that fractional anisotropy was significantly decreased in the thalamus in patients with a predominant contamination/washing dimension in a diffusion tensor imaging study. They discovered abnormalities in the microstructure of white matter in a putative limbic-cortical-striatal-thalamic circuit. Also, a functional MRI (fMRI) study found decreased activation in the thalamus in patients with OCD compared to controls when contamination-relevant stimuli were provoked (Gilbert et al. 2009). A perfusion-weighted imaging study in unmedicated OCD patients with contamination/washing symptom suggested that regional cerebral blood flow was significantly increased after symptom-provocation task in the thalamus (Chen et al. 2004). However, there have been no morphometric studies to show the relationship between washing symptom dimension and thalamus. Therefore, our investigation provided additional evidence for the relationship between washing symptom dimension and thalamus volume, and this was in accordance with previous DTI, fMRI, and perfusion-weighted imaging studies. Those studies suggested that OCD patients with washing/contamination symptom have functional and morphometric alterations in the thalamus. Previously, three research groups investigated the relationship between washing dimension and brain regions using VBM. Those studies did not report the thalamus, but various other brain regions related to washing dimension. Okada et al. (2015) reported that washing dimension scores were negatively correlated with gray matter volume in the right insula and positively correlated with gray matter volume in the right cerebellar tonsil. Van den Heuvel et al. (2009) showed significant negative correlation in the bilateral caudate. Gilbert et al. (2008) reported significant negative correlation in right Brodmann area 6. Okada et al. (2015) and van den Heuvel et al. (2009) employed different assessment scales, the symptom checklist of the Yale-Brown Obsessive-Compulsive Scale (Goodman et al. 1989), and Dimensional Yale-Brown Obsessive-Compulsive Scale (Rosario-Campos et al. 2006) and Padua Inventory-Revised (Sanavio 1988; Van Oppen et al. 1995) assessment scale, respectively. Gilbert et al. (2008) conducted multivariate regression analysis including the brain region volumes that were identified in the comparison between OCD patients and healthy controls. Those methodological differences might have led to the distinctions in the results.

In clinical scenarios, OCD patients with contamination/washing symptoms harbor obsessions that they might have touched objects even when they have simply glimpsed the objects. These obsessions increase and induce compulsions such as washing hands repeatedly. Rauch et al. advocated that the obsessions represent failures in filtering at the level of the thalamus, attributed to deficient modulation of the cortico-striato-thalamic collateral pathway (Rauch et al. 2002). They claimed that compulsion is induced when the information that is normally processed efficiently outside of the conscious domain instead finds access to the explicit processing system because of striatal dysfunction. As a result, striato-thalamic modulation might have occurred via performance such as ritualized thoughts or behaviors that activate the adjacent, intact striato-thalamic network. This compensatory function is used to explain the reason for the numerous repetitive behaviors required until the precipitating obsessions settle down. To view our result in terms of this theory, the decreasing thalamic volume in OCD patients with washing dimension might be associated with the normal striato-thalamic network function against cognitive intrusion such as related fear of contamination. Further, the prefrontal cortex was indicated as the region being involved with emotional behaviors such as disgust (Lawrence et al. 2007). In addition, significantly higher fractional anisotropy in bilateral prefrontal white matter was shown in patients with a predominant contamination/cleaning symptom dimension (Ha et al. 2009). Therefore, we could also expect alteration of the prefrontal cortex in OCD patients who had the washing dimension. The reason why we found a correlation with the washing dimension score in the thalamus but not in the prefrontal cortex might have been caused by excluding the effect of other dimensions by treating the remaining five dimension scores as nuisance covariates. Also, our small sample size could have induced these results. Although our study did not identify the thalamus as the region most responsible for the washing dimension, the thalamus might be involved in the compulsion, such as repetitive hand washing. Then, we also analyzed the data with the addition of the BDI scores (n = 28 out of 33) as a nuisance covariate. In gray matter analysis, we found a negative correlation between washing dimension scores and the volume of the left superior temporal gyrus, left thalamus, and left postcentral gyrus in addition to the right thalamus (Fig. S1 and Table S1). This result, namely, that more negative correlations in thalamus volume were shown by excluding the effect of depression, supported the previous studies (Kong et al. 2014; Peng et al. 2016). They reported increased gray matter volume in the left (Kong et al. 2014) or bilateral (Peng et al. 2016) thalamus in medication-naive major depression disorder compared with HC. Furthermore, an additional result of the involvement of the superior temporal gyrus in the pathophysiology of OCD was consistent with previous reports (Choi et al. 2006; Nakamae et al. 2012; Tang et al. 2015). Nakamae et al. (2012) reported reduced cortical thickness, and Choi et al. (2006) and Tang et al. (2015) reported significantly smaller gray matter volume in the superior temporal gyrus compared with healthy controls. Although our sample size was small, the result suggested that the involvement of thalamus in the washing symptom dimension in OCD patients exists with or without the comorbidity of depression.

On the other hand, there has been no other report regarding angular gyrus white matter showing a significant negative correlation with hoarding dimension scores. Piras et al. (2015), however, reported in a meta-analysis of DTI studies that altered anatomical connectivity between frontal and parieto-occipital associative cortices might be related to the pathophysiology of OCD. Our result with angular gyrus white matter could support this notion. With regard to gray matter studies, Rauch et al. (1994) showed involvement of the angular gyrus with a symptom-provocation PET study of OCD. Also, Valente et al. (2005) found decreased right angular and supramarginal gyrus volumes in OCD patients. From those studies, Menzies et al. (2008) indicated the existence of extra brain regions involved in OCD in addition to the fronto-striatal circuit. They suggested that parietal lobe dysfunction particularly within the angular gyrus could be related to the cognitive impairment in OCD. According to symptomatology, hoarding symptom showed various cognitive deficits. Then, the parietal lobe including the angular gyrus might contribute to the cognitive deficit events of hoarding symptom such as the excessive acquisition of and inability to discard objects (Tolin et al. 2012). As for hoarding dimension, Gilbert et al. (2008) indicated that gray matter volume in left Brodmann area 6 decreased with hoarding scores, and Mataix-Cols et al. (2004) reported significantly greater activation in the left precentral gyrus and right orbitofrontal cortex compared with healthy controls. Based on these studies, involvement of the fronto-striatal circuit in hoarding dimensions was actually preconceived. Nevertheless, our result showed the correlation between hoarding dimension scores and brain volumes in angular gyrus white matter. This result might also have been influenced by our nuisance covariates and small sample size.

There are several limitations to our study. First, the patient sample may not have been large enough to allow us to detect robust correlations between symptom dimension scores and gray and white matter volumes in brain regions. This might be one of the reasons for the correlations being detected in only one dimension each in gray and white matter. Second, 81 % of our participants were taking medications. Previous studies have reported medication effects on brain structures including the thalamus in OCD (Gilbert et al. 2000; Hoexter et al. 2012; Valente et al. 2005). Finally, our sample included OCD patients with comorbid diseases. Future studies focusing on OC symptom dimensions with larger numbers of treatment-naive subjects and less comorbidity might elucidate the pathophysiology of OCD.

In conclusion, we found significant correlations between washing dimension scores and gray matter volumes in the right thalamus and between hoarding dimension and brain volume in left angular gyrus white matter while treating the remaining five dimension scores, gender, age, and AQ scores as nuisance covariates. Although our results were preliminary, we found a particular relation between a symptom dimension and the respective brain region. While searching for the dysfunction of neural circuits, it is important to observe regions playing a major role in each symptom dimension for a better understanding of the pathophysiology of this disabling disorder.