Obsessive–compulsive disorder (OCD) is a symptomatologically heterogeneous psychiatric disorder with a worldwide lifetime prevalence ranging from 2 to 3% that often leads to chronic disability (Abramowitz et al., 2009). The primary clinical manifestations of this disorder include recurrent, ruminative and anxiety-inducing thoughts (obsessions) and/or repetitively ritualized mental acts or behaviors (compulsions). Existing treatment methods entailing pharmacotherapy and cognitive behavioral therapy (CBT) generally only result in partial symptom improvements, with roughly one third of cases continuing to exhibit severe chronic illness that is resistant to treatment (Bloch et al., 2006, 2012; Mallet et al., 2008; Greenberg et al., 2010; Grant, 2014; Odlaug et al., 2014). Neurosurgical intervention has demonstrated substantial symptom alleviation in such cases (Jung et al., 2014). It typically involves either an ablative procedure or reversible neuromodulation using deep brain stimulation (DBS) at locations within the cortico-striato-thalamo-cortical (CSTC) circuitry, from which effects spread throughout the brain (Greenberg et al., 2010; Jung et al., 2014; Pepper et al., 2015).

The CSTC model of OCD is currently the prevailing model in terms of the neural and pathophysiological underpinnings of the disorder (Pauls et al., 2014), and is supported by decades of animal research that have led to findings of a strong association between OCD-like behavior and aberrant functioning of fronto-striatal neural circuitry (Fineberg et al., 2011). The functionally distinct loops comprising this circuitry originate in specific territories of the frontal cortex, project to target areas in the striatum and pass through the basal ganglia to the thalamus, with recurrent projections back to their corresponding frontal cortical regions of origin (Milad and Rauch, 2012; Nakao et al., 2014). Research in humans has been heavily informed by findings in animals, and brain imaging studies have been extensively employed in the investigation and refinement of the CSTC model (Whiteside et al., 2004; Norman et al., 2016). Although this area of research has yielded variable results in terms of specific regions and the direction of dysfunction (increased vs. decreased activity) in different studies, abnormal activity in the fronto-striatal regions is consistently reported. Resting-state functional magnetic resonance imaging (fMRI) studies most often find increased activity in the orbitofrontal cortex (OFC) and striatum, while symptom provocation paradigms most often find increases in the OFC, caudate and anterior cingulate cortex (ACC) (Saxena et al., 2001; Schlösser et al., 2010; Nakao et al., 2014).

Given the aberrant functioning of the OFC in OCD, it is unsurprising that diminished performance on decision-making tasks is commonly reported in affected individuals. Decision-making ability is a multifaceted higher order function with significant real-world implications that relies on intact functioning of the prefrontal cortex for representing and integrating relevant information. The Iowa Gambling Task (IGT), an extensively used decision-making task paradigm, requires the participant to track the stimulus-reward association by distinguishing between the stimuli with an immediate large reward but a long-term negative consequence and the stimuli with an immediate small reward but a long-term positive consequence. Compared to similar neuropsychological tasks, such as the Wisconsin Card Sorting Test (WCST), the IGT is sensitive to damage and/or dysfunction across a wider range of brain areas, including anterior and posterior regions of the orbitofrontal and ventromedial prefrontal cortex, as well as the dorsolateral prefrontal cortex (Bechara et al., 1998, 2001).

Although a number of studies have investigated the relationship between OCD and decision-making ability using IGT, their findings have not been consistent. Cavedini et al. (2002) demonstrated that subjects with OCD performed more poorly on the IGT than subjects with panic disorder and healthy controls (Cavedini et al., 2002). Furthermore, they found poor neuropsychological task performance to be a predictor of poor outcome to pharmacological treatments. Cavallaro et al. (2003) also found that OCD patients performed significantly worse on the IGT than both control subjects and schizophrenia patients (Cavallaro et al., 2003). However, Nielen et al. (2002) and Lawrence et al. (2006) found no differences between the performance of OCD patients and control subjects in the IGT (Nielen et al., 2002; Lawrence et al., 2006). Interestingly, Zhang et al. (2015) recently found that medicated, non-medicated and remitted OCD patients all performed significantly worse on the IGT than comparison subjects (Zhang et al., 2015)^. These inconsistencies may in part be due to the heterogeneity of OCD, the methodological differences, and the sampling constraints in some of the studies.

These inconsistent results indicate that the links between OCD, decision-making and altered prefrontal functions remain unclear. Therefore, studies investigating the effects of surgical intervention in the brain circuitry on OCD and decision-making could shed light on the puzzle (Brown et al., 2016). Surprisingly, there has been only one study to date investigating the effects of surgical intervention in OCD on decision-making as assessed by the IGT (Csigo et al., 2010). Csigo and colleagues monitored five refractory OCD patients that underwent anterior capsulotomy (AC), and five non-operated refractory OCD patients over a 2-year period. Their results indicated improvements in IGT performance in the former group at the 1- and 2-year follow-ups (but not at the 1-month or 6-month follow-ups) with no improvements in the non-operated group. Although these results suggest that AC may lead to improved decision-making in OCD, the study was done with a very limited sample size. It is important to understand the difference between short-term and long-term AC effects on decision making, because it would help us to interpret the circuitry affected by AC.

We therefore aimed to conduct a broad and reliable assessment of decision making under risk as measured by the IGT before and after AC at different time points in a large sample of OCD patients as compared to healthy individuals, while also examining the relationship between task performance and scores on a range of clinical measures. We found that significant improvements in the IGT performance after AC happened only in the patient group that had received the surgery for more than 1 year, suggesting a long-term plasticity in the brain circuitry was responsible for such improvements. Furthermore, the improvements in IGT performance could not be explained by their psychiatrical rating scale scores, which showed improvements shortly after AC.

The results of the present study converge with extensive research in the field of OCD demonstrating impaired decision making associated with the disorder, particularly in refractory OCD patients, and further suggest that AC results in improvement across obsessive-compulsive, depressive and anxious clinical symptoms as well as decision-making ability under ambiguity. Our findings indicate that improved decision making occurs over time as a result of the normalization of aberrant CSTC activity, rather than as an instantaneous result of the neurosurgical lesion, therefore possibly constituting further evidence for the role of the CSTC circuitry in decision-making ability. We found no evidence for an association between clinical symptoms and decision making, suggesting that different neural substrates are at play.

Anatomically, the prefrontal cortex (PFC) is essential for the neural processes that underlie decision making and its sub processes, including planning, inductive reasoning, reward processing and manipulating complex information (Cservenka and Nagel, 2012; Rebola et al., 2012). Studies often differentiate three sub-regions of the PFC: the dorsolateral prefrontal cortex (DLPFC), the orbitofrontal cortex (OFC) and the anterior cingulate cortex (ACC) (Fuster, 2001). The OFC occupies the ventral region of the PFC and can further be divided into the ventromedial and ventrolateral portions, with the ventromedial OFC most frequently implicated in OCD (Bechara et al., 1994). This area appears critical in integrating affective information relayed from the other limbic areas and directly signaling certain reward types. In environments where rewarding options may constantly change, the OFC plays an important role in adaptive behavior (Elliott et al., 2000). It maintains a representation of reward history and modifies former reward representations that may have become disadvantageous. Together with frequently reported neuroimaging findings of aberrant OFC activity in OCD patients, these lines of evidence lend support to an association between reward mechanisms and symptomatology of the disorder. OFC impairments are often acknowledged as the most supported account for the inadequate behavioral choices and inflexibility of behavioral change typically expressed in OCD, although the underlying circuitry is likely far more complex.

Interestingly, patients with lesions in the OFC show deficits in their IGT performance similar to OCD patients (Bechara et al., 1994). More generally, recent research has shown improvements in verbal memory, visual memory, visuospatial skills and executive function in treatment-refractory OCD following AC as compared to both a non-surgical treatment group and healthy controls (Gong et al., 2017). Our study therefore provides further evidence that abnormal OFC functionality may be a contributing factor of the OCD behavior. We postulate that AC may enhance OFC activity via its disinhibition, though the mechanism by which this occurs is not immediately obvious. One of the major pathways in the anterior internal capsule carries projections from the thalamus to the frontal cortex. However, these projections are excitatory. Corticopontine projections also pass through the internal capsule, although they are unlikely to affect the OFC. It is therefore likely that capsulotomy works by disrupting the basal ganglia’s internal connections between the caudate and lentiform nucleus. Modifying these connections may affect the output of the basal ganglia, leading to the disinhibition of the OFC. This indirect modulation could also explain why behavioral improvement is mostly observed only in the long run (Pepper et al., 2015).

We found that the AC improved OCD symptoms as indicated by clinical scores, which is consistent with previous reports (D’Astous et al., 2013). In addition, our observed outcome of improved decision-making ability occurring over time independently of symptom alleviation is in line with that of Csigo et al. (2010), and is indicative of possible differences in the neural substrates underlying clinical symptoms and neuropsychological difficulties in OCD. The objective of neurosurgical techniques is to influence the connections between cortical areas (e.g., orbitofrontal cortex and cingulum), the basal ganglia (particularly the caudate nucleus) and the medial dorsal thalamic nucleus. Our findings indicate that IGT performance is not directly influenced by the instantaneous changes caused by the disruption of these connections. The critical role of receptor function in the neural circuitry mediating behavioral control evokes the possibility that decision making as measured by the IGT task is modulated by dopamine or serotonin signaling. Neurochemical alterations of the serotonergic and dopamine systems are assumed to be involved in the mediation of OCD symptoms (Denys et al., 2004). Serotonin (5-HT) has received the most attention in OCD, as enhancement of 5-HT neurotransmission results in a therapeutic effect; potent 5-HT re-uptake inhibitors are effective treatments for OCD (Huang and Williams, 2007). However, there is evidence that dopamine receptor levels modulate impulsivity and effortful decision making in mice (Simon et al., 2013). Furthermore, striatal dopamine D2/D3-receptor availability was demonstrated to have an inverse relationship with response inhibition speed on the Stop-Signal Task (a measure of impulsive behavior) and a positive relationship with inhibition-related fMRI activation in the fronto-striatal neural circuitry in healthy individuals (Ghahremani et al., 2012). Further research is needed to determine whether long-term rebalancing of neurotransmitter levels after AC contributes to changes in decision-making ability and other cognitive facets, such as impulsivity, that are commonly impaired in OCD.

Various accounts consider neuropsychological impairment an intermediate phenomenon linking aberrant brain activity with clinical symptomatology (Zhang et al., 2015). Moreover, unaffected relatives of individuals with OCD consistently exhibit comparable deficits in decision making to probands (Viswanath et al., 2009; Cavedini et al., 2010). There is evidence that behavioral impairment in both is associated with gray matter abnormalities in regions including the orbitofrontal cortex, indicating a heritable component (Menzies et al., 2007). However, it remains unclear whether this component is a primary feature of disease circuitry or a compensatory mechanism for other primary features. Future studies aimed at improved characterization of the neural substrates of neuropsychological and clinical characteristics before and after surgical intervention can help to elucidate the complex overlap and divergence between these disease characteristics.

Finally, our sample was comprised of individuals with severe OCD that is resistant to conventional treatment methods. Studies have indicated that individuals with refractory OCD exhibit a higher degree of neuropsychological deficits than treatment responsive individuals in terms of attention, set-shifting ability, linguistic performance, decision making, impulsivity (Csigo et al., 2010). Interestingly, there is also evidence that poor performance on the IGT is predictive of poor pharmacological treatment outcomes (Cavedini et al., 2002). However, it is important to bear in mind that dysfunction within any of the other neural systems that feed into the OFC/vmPFC system can lead to similar decision-making impairments, particularly when the disorders are marked by recognized heterogeneity in neural and symptomatological manifestation.

The current study corroborates previous findings supporting the effectiveness of AC for the treatment of refractory OCD. Our results revealed that long-term changes after AC in the brain circuit may be revealed by the patients’ improvements in IGT. These changes were not correlated to the psychiatric rating scales, which started to show improvements in the patients shortly after AC. Future studies will be needed to demonstrate the neural circuitry underlying both categories of change and how they are affected by AC across time.

TY, BS, and SZ conceived of the project. TY and YC designed and supervised the experiments. YC and TY analyzed the data. CZ, ST, KZ, HJ, HG, YX, TW, BS, and SZ collected the data. CZ, ST, TW, BS, and SZ performed the surgery. CZ, YC, KZ, and TY wrote the paper with inputs from all authors.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.