Exploring the neurobiology of OCD: clinical implications

Introduction

Obsessive-compulsive disorder (OCD) is a common neuropsychiatric condition, affecting 1-2% of the population globally. Despite considerable heterogeneity in the precise symptoms experienced across different patients (e.g., some patients are preoccupied with worries of contamination, whereas others obsess about symmetry), there is growing realization that common neurobiological processes may contribute to vulnerability towards OCD and its persistence. OCD is regarded as the archetypal disorder of compulsivity (i.e. a tendency towards repetitive habitual actions that a person feels a need to perform, with these tendencies having untoward functional consequences, such as detracting from overall life goals, or quality of life). Although the optimal definition of compulsivity likely depends upon perspective and context¹, within this broad clinical framework, disorders of compulsivity include OCD and related disorders such as hoarding disorder, body dysmorphic disorder, trichotillomania, skin picking disorder, and Tourette’s². Here we focus on recent advances in understanding the neurobiology of OCD, and the clinical implications of such knowledge viewed in the context of prevailing disease models.

Brain circuitry in OCD

Case vignette: Claire, a 21 year-old student, reports a five-year history of moderately severe OCD, mainly relating to taboo sexual thoughts and the repeated need to confess. Claire signs up for a research study exploring the neurobiology of OCD, in which she undertakes a clinical assessment, cognitive tests, and a structural brain scan. At the end of the session, Claire asks whether her brain scan can be used to help confirm that she has OCD. She says she has read on the Internet about research that can accurately diagnose OCD using brain scans. She asks if she can have a picture of her brain to see the changes that happen in OCD.

Structural and functional changes within the brain have long been implicated in the pathophysiology of OCD. Cortical and sub-cortical brain regions comprise a series of functionally relatively segregated circuits, that may play different roles in thoughts and behaviors³. Neuroimaging work in OCD has commonly identified structural and functional abnormalities, most typically involving the orbitofrontal cortices and basal ganglia (caudate nucleus), known as the ‘orbitofrontal circuit’ (for recent reviews see ^4–6). These findings support the classical model that OCD can be considered a disorder of maladaptive habit circuitry⁷, a model that has gained traction and evolved into more recent conceptualizations focusing on habit^8,9 and loss of top-down control by cortically-mediated inhibitory mechanisms (referred to as ‘disinhibition’)¹⁰.

In recent years, it has become apparent that OCD involves changes across a broad range of fronto-striatal loop circuits¹¹, though abnormalities of the orbitofrontal cortices and basal ganglia have commonly been reported. OCD has typically been associated with grey matter volume increases in sub-cortical structures (such as the putamen and globus pallidus), and with grey matter reductions in the cortex (especially ventral and dorsal medial cortex, and inferior frontal cortex)¹². In terms of measures of cortical thickness (a proxy for the number of neurons in a particular brain region), more widespread reductions have been typically observed, including not only in the frontal but also in the parietal and temporal parts of the brain. Another common finding reported in OCD has been reduced fractional anisotropy (a measure of fiber density, axonal diameter, and myelination in white matter) in anterior midline tracts (including parts of the corpus callosum and cingulate bundle)¹². Collectively, these neuroimaging data suggest OCD is associated distributed changes across anatomically disparate brain structures, both in terms of grey matter and white matter tracts.

One approach used more recently to explore brain structure in OCD has been to pool structural neuroimaging scans from many diverse group case-control studies, a technique referred to as ‘mega-analysis’. This approach has found that OCD is associated with smaller hippocampal volumes and larger pallidum volumes, versus controls, but failed to find any significant differences in the caudate or putamen¹³. Furthermore, OCD was associated with decreased cortical thickness in various frontal, parietal, and temporal cortical regions, versus controls¹⁴. These mega-analytic results highlight the existence of structural brain abnormalities outside of the classic orbitofrontal loop circuit.

We can also think of the neurobiological underpinnings of OCD in terms of function of distributed brain networks, whether in the so-called resting state, or during cognitive tasks. Indeed, a meta-analysis of available resting state functional connectivity neuroimaging studies in OCD identified hypo-connectivity within and across some circuits; with dysconnectivity (no particular direction of connectivity changes) in other circuits¹⁵.

Cognitive neuroimaging studies in OCD have typically focused on domains previously found to be impaired in the disorder, such as motor inhibitory control, cognitive flexibility, and executive planning. Imaging can be used to assess neurobiological underpinnings of cognitive task performance in two ways: firstly, by measuring activation in particular brain regions; and secondly, by examining functional connectivity or ‘coupling’ between such brain regions. In a meta-analysis of functional neuroimaging studies using inhibitory control tasks, patients with OCD versus controls exhibited under-activation in several brain areas (rostral and ventral anterior cingulate cortices, bilateral thalamus/caudate, right anterior insula/frontal operculum, supramarginal gyus, and orbitofrontal cortex)¹⁶.

In terms of functional connectivity, there are recent findings from case-control studies. One study found that OCD patients and their clinically asymptomatic first-degree relatives had reduced functional connectivity between anterior and posterior cortical regions during a motor inhibition task (the stop-signal task)¹⁷ (Figure 1). In another study, this time focusing on connectivity between cortex and sub-cortical regions, reduced resting state functional connectivity between the ventrolateral prefrontal cortex and dorsal caudate nucleus was linked with worse cognitive flexibility in OCD compared to controls¹⁸. Elsewhere, on a neuroimaging executive planning task, dysconnectivity was identified between cortex and basal ganglia (putamen) in both OCD patients and their clinically asymptomatic first-degree relatives, versus controls with no known family history of OCD¹⁹. Because some of these brain changes extend to first-degree family members, they may represent vulnerability markers for OCD.

Figure 1.

Example of the use of a cognitive task (the stop-signal task) during functional neuroimaging, to explore the neurobiology of OCD. Top figure: brain networks associated with undertaking an inhibitory control task. In blue are regions that activated significantly on trials requiring participants to inhibit their motor responses. In purple are regions that activated for failed inhibition trials compared to successful inhibition trials (‘error detection’). In red are regions showing significantly abnormal activation in OCD patients and their relatives, compared to controls. Interestingly, the red regions were outside the classic orbitofrontal loop. Bottom figure: hypo-connected brain pathways that were identified in OCD patients and their relatives during the inhibitory control task. It can be seen that there were many abnormally reduced connections from anterior to more posterior cortical brain areas. As such, reduced connectivity across cortical brain regions responsible for inhibiting habitual response patterns may constitute a vulnerability marker for OCD. Reproduced under Creative Commons license from Hampshire et al., Biol Psychiatry Cogn Neurosci Neuroimaging, 2019.

Collectively, what take home messages can be gleaned from the above? The imaging evidence to date suggests that OCD is associated with distributed, subtle, structural and functional brain changes involving not only the orbito-frontal loop but also other circuits. This information, including awareness of what the literature does and does not demonstrate, can be helpful when asked questions from patients, such as in the case of Claire in the vignette above. To address Claire’s question as to whether her brain scan would “show” OCD, we would explain that although brain changes have been reported when comparing groups of people with OCD to groups of people without OCD, these are ‘average’ differences and are very subtle. They cannot be seen ‘by eye’ on a person’s brain scan.

To address Claire’s other question, as to whether OCD can be diagnosed using a brain scan, the answer is no: there is no appropriately validated algorithm that can be used to diagnose OCD based on a brain scan. There have been studies using a technique called ‘multivariate pattern analysis’ to build predictive models that are capable of classifying scans (e.g. into OCD or control groups)²⁰; however, one cannot conclude from this that these algorithms would generalize to OCD at large, or to other research or clinical settings. For example, smaller studies can result in model ‘over-fit’ – a statistical issue whereby a model can apparently give astoundingly high accuracy; but this just reflects statistical fallacy and results would not generalize. Also, in one of the largest classifier studies to date, using pooled mega-analysis data and high-quality methodology, machine learning classifiers based on neuroimaging measures were found to be poor and no better than chance at identifying OCD when applied to an independent set of data²¹. There were more promising results when such techniques were applied to a subset of the data, albeit some caution is needed due to the negative overall findings in the primary analysis.

Have advances in the neurobiology of OCD affected treatments?

Case vignette: Joseph is a 28 year-old man with a ten-year history of severe OCD, with contamination obsessions and washing compulsions, extensive procrastination, and repetitive list-making/doodling. He has received appropriate treatment trials with various serotonin reuptake inhibitors, including augmentation strategies using other agents (such as low dose antipsychotic medication, and n-acetyl cysteine), and extensive cognitive behavioral therapy (CBT) using exposure response prevention (ERP). His OCD symptoms remain severe; he is incapacitated by his illness and is not able to leave the house often, work, or socialize. Following approval by an ethics board, and careful discussion of the benefits and risks, Joseph underwent a neurosurgical procedure to implant electrodes targeting the nucleus accumbens. Approximately 6 months following Deep Brain Stimulation (DBS), and continuation of pharmacotherapy and CBT, Joseph reported a significant improvement in OCD symptoms, and he was able to work again and socialize. Three years post-surgery, his symptoms remained much improved with continuation of DBS treatment.

Given the changes across distributed brain circuits described previously, an interesting question is to what extent neurobiological knowledge of OCD has changed treatment approaches for patients. Current first-line, evidence-based treatments for OCD include serotonin reuptake inhibitor medications and/or CBT with ERP. In a recent systematic review and network meta-analysis²² serotonin reuptake inhibitors showed superiority over placebo in treating adult OCD; and all the examined psychological therapies had greater efficacy than drug placebo in adult OCD. These first-line treatments for OCD have been used for >30 years and have not practically been influenced or altered by neurobiological research into OCD.

However, imaging has yielded insights into brain mechanisms by which treatments may improve OCD. There are quite a few studies now that have found that structural and functional brain changes associated with OCD symptoms normalize to some extent with successful medication treatment. Such partial normalization also occurs with psychotherapy in OCD, as revealed in a recent systematic review of the literature²³. At the same time, as noted in this review and others, there are some caveats – such as often small sample sizes, lack of suitable control conditions/groups, etc. Overall, successful treatment with medication or psychotherapy does seem to normalize at least some of the brain changes associated with OCD. This leads to the question of whether imaging could be used to direct or predict treatment response. As with diagnosing OCD at the individual patient level using a brain scan, there is no evidence that treatment can be usefully predicted at the individual subject level. Again, studies suggest that algorithms to predict treatment response using baseline scans can be built, including to predict response to psychotherapy²⁴, but these approaches have yet to be shown to generalize or be useful at the individual subject level in clinical practice.

As indicated by the case vignette of Joseph above, DBS, or other ablative techniques such as gamma ventral capsulotomy²⁵ are sometimes used in the most extreme cases of treatment-resistant OCD. These neurosurgical interventions, however, do not help everyone and when successful, may result only in reduction of OCD symptoms, not remission. This has led research into improving these interventions on the individual level based on more detailed neurobiological understanding of OCD. For example, one recent study used a clinical assessment and symptomatic provocation during functional MRI to enhance electrode placement for DBS in a small sample of patients²⁶.

Concluding remarks

Substantive advances have been made in understanding the neurobiology of OCD. We have seen that OCD is often associated with structural brain changes implicating not only the classic orbitofrontal circuit but other regions too – including relative reductions in cortical thickness across multiple regions, which may contribute to the clinical picture of disinhibition and a loss of ‘top-down control’ governing basal-ganglia driven habitual response patterns. Functional imaging has revealed hypo-activation during tasks of inhibitory control, as well as (typically) reduced functional connectivity between key brain regions, during these and other types of cognitive tasks. While some OCD related brain changes appear to normalize with successful treatment, cognitive and neural changes have also been identified in first-degree relatives of OCD patients without symptoms. This indicates that some feature may be vulnerability markers, whereas others may reflect symptoms.

Longitudinal research is needed to better delineate vulnerability versus chronicity markers in OCD, and to translate these research findings into meaningful changes in daily clinical practice. To date, first line treatments for OCD are essentially unchanged for >30 years. Neurobiological advances are useful as they can help clinicians and patients understand the illness and how treatments work when they successfully improve symptoms. With time, the hope is that predictive algorithms could be developed and validated in order to help refine diagnosis and treatment prediction at the individual patient level; but this remains a hope rather than a present day reality.