Introduction
Obsessive–compulsive disorder (OCD) is a neuropsychiatric disorder characterized by recurrent intrusive thoughts and/or repetitive rituals called compulsions. First-line treatments are cognitive behavioral therapy (CBT) with exposure and response prevention (ERP) and pharmacotherapy, with serotonin reuptake inhibitors being effective in about 60–70% of all patients (Foa et al. Reference Foa, Liebowitz, Kozak, Davies, Campeas, Franklin, Huppert, Kjernisted, Rowan, Schmidt, Simpson and Tu2005; Abramowitz, Reference Abramowitz2006).
OCD is associated with abnormal neuronal functioning in a corticostriatal circuitry mediating inhibitory control and flexible responding. Several studies have demonstrated that patients with OCD exhibit an increased metabolism during rest conditions predominantly in the orbitofrontal cortex and in striatal areas in comparison to healthy controls (Baxter et al. Reference Baxter, Phelps, Mazziotta, Guze, Schwartz and Selin1987; Whiteside et al. Reference Whiteside, Port and Abramowitz2004). However, when cognitively challenged by decision-making or planning tasks (van den Heuvel et al. Reference van den Heuvel, Veltman, Groenewegen, Cath, van Balkom, van Hartskamp, Barkhof and van Dyck2005; Remijnse et al. Reference Remijnse, Nielen, van Balkom, Cath, van Oppen, Uylings and Veltman2006; Chamberlain et al. Reference Chamberlain, Menzies, Hampshire, Suckling, Fineberg, del Campo, Aitken, Craig, Owen, Bullmore, Robbins and Sahakian2008), these patients demonstrated a reduced responsiveness in the aforementioned areas. Current models of OCD therefore consider an impaired frontostriatal circuitry as a correlate of this disorder. Both Remijnse et al. (Reference Remijnse, Nielen, van Balkom, Cath, van Oppen, Uylings and Veltman2006) and Chamberlain et al. (Reference Chamberlain, Menzies, Hampshire, Suckling, Fineberg, del Campo, Aitken, Craig, Owen, Bullmore, Robbins and Sahakian2008) used a reversal learning (RL) paradigm to gain a better understanding of these mechanisms. RL requires flexible alteration of behavior after negative feedback and is known to be dependent on the serotonergic system (Clarke et al. Reference Clarke, Walker, Dalley, Robbins and Roberts2007) and also on the integrity of the orbitofrontal cortex and the medial striatum (Clarke et al. Reference Clarke, Robbins and Roberts2008). Remijnse, Chamberlain and co-workers demonstrated abnormally reduced activation of orbitofrontal areas on RL in OCD patients compared with healthy subjects. This supports the hypothesis of a dysfunctional frontostriatal circuitry associated with obsessive–compulsive symptoms.
To date, few studies have examined the impact of treatment in general and of psychotherapy in particular on the neurobiology of OCD (Baxter et al. Reference Baxter, Schwartz, Bergman, Szuba, Guze, Mazziotta, Alazraki, Selin, Ferng, Munford and Phelps1992; Schwartz et al. Reference Schwartz, Stoessel, Baxter, Martin and Phelps1996; Ho Pian et al. Reference Ho Pian, van Megen, Ramsey, Mandl, van Rijk, Wynne and Westenberg2005; Nakao et al. Reference Nakao, Nakagawa, Yoshiura, Nakatani, Nabeyama, Yoshizato, Kudoh, Tada, Yoshioka, Kawamoto, Togao and Kanba2005; Nabeyama et al. Reference Nabeyama, Nakagawa, Yoshiura, Nakao, Nakatani, Togao, Yoshizato, Yoshioka, Tomita and Kanba2008). Baxter et al. (Reference Baxter, Schwartz, Bergman, Szuba, Guze, Mazziotta, Alazraki, Selin, Ferng, Munford and Phelps1992) demonstrated statistically significant decreases in glucose metabolic rates in the right head of the caudate nucleus after effective treatment of OCD with either pharmacological or behavioral therapy. In a positron emission tomography (PET) study the same group replicated these findings specifically for behavioral therapy and demonstrated a normalization of pathological correlational activity between the orbital cortex and the caudate nucleus (Schwartz et al. Reference Schwartz, Stoessel, Baxter, Martin and Phelps1996). In line with these findings, Saxena et al. (Reference Saxena, Brody, Maidment, Dunkin, Colgan, Alborzian, Phelps and Baxter1999) found a reduction of primarily elevated metabolism in the orbitofrontal cortex and in the caudate nucleus under treatment with paroxetine. Using functional magnetic resonance imaging (fMRI), Nakao et al. (Reference Nakao, Nakagawa, Yoshiura, Nakatani, Nabeyama, Yoshizato, Kudoh, Tada, Yoshioka, Kawamoto, Togao and Kanba2005) demonstrated diminished activation in orbitofrontal and dorsolateral prefrontal areas in the course of treatment under symptom provocation.
RL includes planning in conjunction with cognitive flexibility and closely resembles a core cognitive dysfunction of OCD (Chamberlain et al. Reference Chamberlain, Menzies, Hampshire, Suckling, Fineberg, del Campo, Aitken, Craig, Owen, Bullmore, Robbins and Sahakian2008; Valerius et al. Reference Valerius, Lumpp, Kuelz, Freyer and Voderholzer2008). Therefore, it is possible that severely affected patients exhibit abnormalities in the neural correlates of RL and that these abnormalities will normalize under symptom reduction through psychotherapy. The only studies examining the effects of treatment on brain function in OCD to date used PET and analyzed metabolism during rest. So far no study in OCD patients has used higher cognitive tasks such as RL probing into frontostriatal circuitry in a longitudinal design. As we have recently demonstrated the high test–retest reliability of the RL task (no changes in brain activation of healthy subjects when tested twice with an interval of several weeks; Freyer et al. Reference Freyer, Valerius, Kuelz, Speck, Glauche, Hull and Voderholzer2009), we consider it suitable for a longitudinal study aiming to detect changes in the course of treatment.
In the current study we sought to investigate the impact of psychotherapy on frontostriatal responsiveness using a test challenging cognitive flexibility. Based on previous work, we expected OCD patients to show reduced orbitofrontal and striatal activation before treatment and a subsequent normalization following CBT.
Method
Subjects
Ten right-handed patients with OCD [three females, mean age 36.1 (s.d.=9.36) years] and 10 healthy controls [four females, mean age 39.6 (s.d.=10.48) years] participated in this study. All patients were in-patients being treated in the Department of Psychiatry and Psychotherapy of the University Hospital Freiburg throughout the entire study. Patients were characterized by the following compulsions and obsessions: four patients presented with washing compulsions, two with checking compulsions, one had predominantly obsessive thoughts and three presented with mixed OCD symptoms. The time interval between the onset of OCD and study participation was 11.7 (s.d.=5.6) years. Nine patients had no psychiatric co-morbidities and one patient had a past medical history of bulimic episodes. In-patient treatment duration varied from 8 to 12 weeks and was carried out by two experienced therapists specially trained in CBT with OCD patients, following a structured concept (Hohagen et al. Reference Hohagen, Winkelmann, Rasche-Ruchle, Hand, Konig, Munchau, Hiss, Geiger-Kabisch, Kappler, Schramm, Rey, Aldenhoff and Berger1998). Concomitant psychotropic medication was an exclusion criterion.
The first treatment phase comprises 6 h of assessment and treatment planning. During this phase, an individual disease model and information about the cognitive behavioral concept of OCD are provided. Thereafter, 16 treatment sessions with ERP lasting about 90 to 120 min over a period of 4 to 6 weeks were provided. The therapist accompanied and supervised the exposure sessions and assigned self-exposure practice to be completed by the patients between sessions. ERP sessions were in vivo and also included home sessions.
Measurements using the Hamilton Depression Rating Scale (HAMD; Hamilton, Reference Hamilton1960), the Yale–Brown Obsessive Compulsive Scale (YBOCS; Goodman et al. Reference Goodman, Price, Rasmussen, Mazure, Fleischmann, Hill, Heninger and Charney1989) and MRI scanning were conducted during the first week after admission before starting with the exposure phase and also immediately before discharge from hospital.
Healthy controls were recruited by personal contact and newspaper advertisements that did not specify the disorder under investigation. Controls were matched for age, sex and education. Personal or family history of a psychiatric disorder or evidence of subclinical obsessions or compulsions led to exclusion. Subjects were measured twice with an interval comparable to the treatment duration of patients.
Exclusion criteria for both patients and healthy controls were the presence of substantial neurological impairment, head injury, substance abuse, current or previous psychotic episodes, pregnancy and age >65 years. All patients were free of psychotropic medication for at least 4 weeks prior to baseline examination and throughout the whole treatment until the second MRI scan. Six patients were naïve for psychotropic medication and four patients had a medication history of selective serotonin reuptake inhibitors. The study was approved by the Ethics Committee of the University of Freiburg according to the guidelines of the Declaration of Helsinki. Informed written consent was obtained from each subject. The study was conducted in the Department of Diagnostic Radiology at the University Hospital Freiburg.
Procedures
Stimuli and task design (Fig. 1)
Details of the RL task used have been published previously (Valerius et al. Reference Valerius, Lumpp, Kuelz, Freyer and Voderholzer2008). In brief, subjects were asked to choose one of two simultaneously presented abstract patterns (a square and a triangle) in each trial by pressing a right or left button on a button box positioned on the abdomen of the subject. A feedback in the form of a green happy face or a sad red face pictogram was presented immediately after the choice, indicating whether the answer was correct or incorrect. After 10 to 15 (randomized) correct responses, the target object changed and subjects had to adapt their strategy by selecting the previously incorrect stimulus. As an additional challenge, probabilistic errors were interspersed, leading to negative feedback despite a correct response in respect of the current rule. If subjects changed their strategy after a probabilistic error, this was counted as a mistake (SCAPE=strategy change after a probabilistic error). Subjects were asked to respond as rapidly as possible without compromising accuracy.
Fig. 1. Illustration of the reversal learning task: a square and a triangle are presented. Subjects have to select one of these objects using a right-hand button press. Feedback indicated a right or wrong answer. * Indicates a wrong choice despite a correct response.
Two successive 9-min runs within one scanning session, each with 10 rule changes (hence nine reversal stages), were presented. Each discrimination phase contained between zero and four probabilistic errors. Objects appeared for a maximum of 2 s, limited by subjects' responses. Feedback was presented immediately after subject responses for 0.5 s, followed by a fixation cross for a minimum of 0.8 s, leading to an interstimulus interval of 3.3 s.
The task was programmed using ‘Presentation’ software (www.neurobehaviouralsystems.com) and was projected onto a screen behind the subject, viewed via a mirror mounted on the head coil. Before entering the scanner, subjects performed a 30-trial training session on a standard desktop PC. This was a simple probabilistic discrimination task (i.e. without reversal stages) designed to introduce the subject to the concept of a probabilistic error without the need to reverse responding.
Data acquisition
MRI was performed on a Siemens Trio 3-T scanner (Siemens AG, Germany). Functional MR images were acquired using gradient-echo echo-planar imaging (GE-EPI) sequences sensitive to blood oxygen level-dependent (BOLD) contrast [repetition time (TR)=3000 ms, echo time (TE)=30 ms, flip angle=90°, field of view (FOV)=256 mm, matrix=128×120, 32 interleaved slices, 3 mm slice thickness]. The event-related design included 400 acquisitions, of which the first five images were discarded to eliminate magnetization instability. Patients were hospitalized for CBT and performed a scanning session with two runs before and after therapy [average interval 57 (s.d.=17.8) days]. Control subjects performed two identical sessions [average interval 119 (s.d.=84.4) days]. The interval tended to be longer in the control group, but not significantly so (p=0.07). The study setting and the performing physician (T.F.) were identical in both sessions to control for possible confounds.
High-resolution anatomical images were acquired using a sagittal T1-weighted 3D-MPRAGE sequence [TR=2200 ms, inversion time (TI)=1100 ms, TE=4.91 ms, flip angle=12°, FOV=256 mm, matrix=256×256, 160 slices, voxel size=1×1×1 mm3] to exclude structural abnormalities and to aid in spatial normalization.
Data analysis and image processing
Behavioral data were analyzed in SPSS version 15.0 (www.spss.com). Parameters for task performance were reaction times and error numbers, clinical status was quantified by HAMD and YBOCS scores. Kolmogorov–Smirnov testing did not reject the assumption of normal distribution for any of the variables. Paired t tests were used to evaluate differences between the sessions. Two-sample t tests served to compare differences between patients and controls separately for each session.
Imaging data analysis was performed using SPM5 (www.fil.ion.ucl.ac.uk/spm) running under MATLAB 7.1 (www.mathworks.com). Images were corrected for slice acquisition delay, spatially co-registered to the individual T1 image and normalized onto the Montreal Neurologic Institute (MNI) Atlas using normalization parameters estimated from the anatomical image. Finally, functional data were smoothed with an isotropic 8-mm full-width at half-maximum (FWHM) Gaussian kernel.
A first-level analysis based on the general linear model (Friston et al. Reference Friston, Frith, Turner and Frackowiak1995) was performed for each subject individually. Task-related changes in fMRI signal were estimated at each voxel by modeling the onsets of the responses as stick functions with a hemodynamic response function. The time series were high-pass filtered with a cut-off at 128 s. Six movement parameters (x, y, z, pitch, roll, yaw) resulting from the image realignment were modeled as additional regressors to control for movement artifacts. Onsets co-occurred with the presentation of the feedback. We sought to model all possible event types with separate regressors to reduce heterogeneity of neuronal processing in the regressor of interest (i.e. the last reversal error). The following events were modeled: (1) first correct responses following a response shift; (2) spontaneous errors; (3) probabilistic errors, on which negative feedback was given to correct responses; (4) SCAPES; (5) reversal errors; and (6) last reversal errors, resulting in the subject shifting the response strategy. Correct responses co-occurring with positive feedback were defined as baseline trials. We did not include first correct responses in the baseline condition because we expected neuronal processing specific to successful strategy change. The last reversal error before strategy change was chosen as the critical event of interest (i.e. reflecting RL) because activation of a reversal network was assumed to specifically follow this last negative feedback. Therefore, parameter estimates reflecting activations with last reversal error compared to implicit (unmodeled) baseline were computed and constituted a statistical parametric map for each individual, which then entered a second-level analysis.
The parameter images of the first session (effect of last reversal error compared to baseline of patients and controls) were first entered into a second-level (random effects) analysis by calculating a one-sample t test to determine the overall effect of strategy change. An additional simple correlation model was used to identify brain regions showing a linear change with the YBOCS score. A two-factor ANOVA with time point (session 1 and session 2) as the within-subject factor and group (patients versus controls) as the between-subjects factor was computed to test for an interaction representing treatment-specific effects over time. A final model was set up to identify those areas co-varying with the change in the YBOCS score over the course of the treatment. To this end, we subtracted the parameter image from the first session from that of the second session separately for each patient. The same was done for the YBOCS scores. We then correlated the change in the parameter estimates with that of the YBOCS scores. Based on earlier work, we expected a clinical improvement (i.e. a reduction in the YBOCS score) to be paralleled by increasing activations in the striatum.
Effects are reported after correction for multiple comparisons using the false discovery rate (FDR) as implemented in SPM5 across the whole brain with p<0.01 at the voxel level. Based on our anatomical a priori hypotheses, we additionally report effects within the frontal cortex and the striatum with p<0.001 uncorrected (Gottfried et al. Reference Gottfried, Smith, Rugg and Dolan2004). Coordinates are given in MNI space.
Results
Behavioral data
Table 1 summarizes the clinical characteristics of the patients, and Table 2 shows the behavioral data for the patients and controls from the RL task. No significant differences overall in error rates or reaction times were found between patients and controls. However, in the patient group before therapy there was a negative correlation between the number of spontaneous errors and the YBOCS scores (YBOCS total, r=−0.78, p=0.004; YBOCS obsessions, r=−0.78, p=0.004; YBOCS compulsions, r=−0.70, p=0.013) and a negative correlation, nearing significance, between the number of SCAPES, representing unnecessary strategy changes after a probabilistic error, and the YBOCS obsessions scores (r=−0.53, p=0.06).
Table 1. Demographics and clinical data
YBOCS, Yale–Brown Obsessive Compulsive Scale; HAMD, Hamilton Depression Rating Scale.
Values are given as mean (standard deviation). p values are derived from paired t tests.
Table 2. Performance in the reversal learning (RL) task
SCAPE, Strategy change after a probabilistic error.
Mean values of event numbers and event reaction times are given with the standard deviation in parentheses.
p values of paired t tests are given for between-session differences, p values of two-sample t tests are given for between-group differences.
Following treatment, mean total YBOCS scores decreased significantly from 25.4 to 14.2 (p=0.002; one-tailed paired t test). Similarly, significant improvements were found for the YBOCS subscores for obsessions and compulsions. Using a 35% reduction of the total YBOCS scores as the criterion, seven out of 10 patients responded to CBT. The correlations between YBOCS scores and the number of spontaneous errors and the number of SCAPES were no longer detected after treatment.
Imaging data
Regions activated during strategy change (last reversal error minus baseline events) across all subjects are shown in Fig. 2. Pooled across both groups, a task-specific BOLD response was found in the bilateral insular cortex (coordinates x, y, z=33, 21, −9; T=9.82 and coordinates x, y, z=−27, 24, –6; T=8.13), dorsal medial frontal cortex (coordinates x, y, z=3, 18, 54; T=11.56), bilateral parietal cortex (coordinates x, y, z=−33, −51, 42; T=9.10 and coordinates x, y, z=45, −48, 51; T=7.32), bilateral middle frontal cortex (coordinates x, y, z=−30, 48, 18; T=6.17 and coordinates x, y, z=30, 51, 21; T=7.44) and right orbitofrontal cortex (coordinates x, y, z=24, 42, −15; T=4.63).
Fig. 2. Effect of last reversal error across both groups. Activations are rendered on a template brain in Montreal Neurologic Institute (MNI) space (p<0.01, false discovery rate-corrected).
Neither significant differences between the groups nor significant time by group interactions were found on a corrected level. However, the reanalysis with more lenient thresholds (p<0.001 uncorrected) revealed the following findings within the a priori regions. Before the start of psychotherapy, reduced activations in patients compared with controls in the bilateral orbitofrontal cortex (coordinates x, y, z=−21, 42, −12; T=3.86 and coordinates x, y, z=21, 42, −9; T=3.82) and the right putamen (24, 12, 0; T=3.42) were found (Fig. 3). No areas were found showing a linear correlation between brain activity and YBOCS scores within the patient group. After treatment, the comparison patients/controls still revealed a difference in the right orbitofrontal cortex (coordinates x, y, z=24, 42, −9; T=3.77) but not in the putamen. Testing for an interaction between group and time showed differential activations in the caudate nucleus (coordinates x, y, z=6, 6, 9; T=3.63; Fig. 4). Fig. 5 illustrates that this is explained by increasing activations over time in the patient group but not in the healthy subjects. OCD patients with greater reductions of the total YBOCS score showed smaller activation increases in the pallidal region (coordinates x, y, z=24, 3, 6; T=4.72; Fig. 6).
Fig. 3. Areas showing decreased activations with last reversal error in obsessive–compulsive disorder patients before treatment compared to controls. Activations are overlaid on a standard brain in Montreal Neurologic Institute (MNI) space (p<0.01, uncorrected).
Fig. 4. Areas showing an interaction between time (i.e. treatment in the obsessive–compulsive disorder group) and group. Activations are overlaid on a standard brain in Montreal Neurologic Institute (MNI) space (p<0.01, uncorrected).
Fig. 5. Group- and time point-specific estimates of activation in the caudate head peak voxel found in the interaction analysis. Error bars indicate 90% confidence intervals.
Fig. 6. Smaller increases in activation in the right pallidum are associated with larger improvements of the clinical presentation. Right panel: Scatterplots with 95% mean prediction interval. YBOCS, Yale–Brown Obsessive Compulsive Scale; AU, arbitrary units.
Discussion
The presented study provides the first evidence for functional neuronal changes after CBT in the basal ganglia as part of the frontostriatal network implicated in OCD using a disease-pertinent cognitive activation paradigm. To our knowledge this is the first study examining effects of psychotherapy on brain function in non-medicated, non-depressed, severely affected OCD patients in a longitudinal design with a matched group of healthy controls. However, these high methodological requirements inevitably lead to small samples, limiting the statistical power (reflected by a lack of significant effects corrected for multiple comparisons) and the data interpretation and conclusions. Nevertheless, our preliminary data stress the potential neuronal effects of successful CBT and therefore contribute to the understanding of frontostriatal pathology in OCD. We also replicate and extend findings described previously on both the behavioral and the neuronal level.
As symptoms in all patients had been stable over several years before inclusion in the study, an improvement in YBOCS scores is likely to be the effect of CBT. Treatment (i.e. CBT in patients) by group analysis showed an interaction in the caudate head. This is caused by patients showing relatively decreased activations before CBT but increased activations after receiving CBT compared to controls (see figures 4 and 5). The peak location in the frontostriatal circuitry is well in line with the proposed impairment of this circuitry in OCD (Huey et al. Reference Huey, Zahn, Krueger, Moll, Kapogiannis, Wassermann and Grafman2008; Maia et al. Reference Maia, Cooney and Peterson2008) and also with the most consistent findings as suggested by meta-analyses of neuroimaging studies of OCD (Whiteside et al. Reference Whiteside, Port and Abramowitz2004; Menzies et al. Reference Menzies, Chamberlain, Laird, Thelen, Sahakian and Bullmore2008). However, previous neuroimaging studies on CBT in OCD consistently reported hyperactivity in the right caudate before treatment that decreased to normal activity after therapy compared to control subjects (Baxter et al. Reference Baxter, Schwartz, Bergman, Szuba, Guze, Mazziotta, Alazraki, Selin, Ferng, Munford and Phelps1992; Schwartz et al. Reference Schwartz, Stoessel, Baxter, Martin and Phelps1996; Nakatani et al. Reference Nakatani, Nakgawa, Ohara, Goto, Uozumi, Iwakiri, Yamamoto, Motomura, Iikura and Yamagami2003; for a review see Linden, Reference Linden2006). It should be taken into consideration, however, that these studies investigated metabolism or cerebral blood flow in the resting state. The hypoactivity before treatment observed in our active task could be due to a lack of modulatory capacity in the caudate nucleus, meaning that patients' pathologically elevated caudate activity at baseline cannot be sufficiently upregulated when challenged through RL as it is in healthy subjects. This would explain why the difference between last reversal error and baseline before treatment was smaller in patients than in controls. Furthermore, we observed an increase in the activation difference after treatment, suggesting a recovery of modulatory capacity through CBT. Correspondingly, patients with greater clinical improvement, reflected by greater reductions in YBOCS scores, showed less increase of activation in the pallidum, suggesting regained control of caudate over pallidal activity. Taken together, these anti-parallel activity changes in the caudate and pallidum in concert with a normalization of activity in the OFC under treatment fit well into proposed models of striato-orbitofrontal pathology of OCD (Huey et al. Reference Huey, Zahn, Krueger, Moll, Kapogiannis, Wassermann and Grafman2008; Menzies et al. Reference Menzies, Chamberlain, Laird, Thelen, Sahakian and Bullmore2008). In addition, the results are in line with striato-orbitofrontal models of RL (Frank & Claus, Reference Frank and Claus2006).
The literature on neuropsychological deficits in OCD is heterogeneous, as are the behavioral results of studies testing cognitive flexibility of OCD patients with the RL task. Although Remijnse et al. (Reference Remijnse, Nielen, van Balkom, Cath, van Oppen, Uylings and Veltman2006, Reference Remijnse, Nielen, van Balkom, Hendriks, Hoogendijk, Uylings and Veltman2009) found differences in task performance, our results are in line with other work on the identical version of this task (Valerius et al. Reference Valerius, Lumpp, Kuelz, Freyer and Voderholzer2008) and on variations (Chamberlain et al. Reference Chamberlain, Menzies, Hampshire, Suckling, Fineberg, del Campo, Aitken, Craig, Owen, Bullmore, Robbins and Sahakian2008). Differences in neuronal activations observed between the groups in our study are therefore a primary effect of disease or of a compensatory neuronal reorganization.
Neuronal activations with the last reversal error across both groups (Fig. 2) are in line with previous findings (Cools et al. Reference Cools, Clark, Owen and Robbins2002; Remijnse et al. Reference Remijnse, Nielen, van Balkom, Hendriks, Hoogendijk, Uylings and Veltman2009). The between-group comparison before therapy replicates previously reported reduced activations in the orbitofrontal cortex but extends this effect to areas including the striatum (Fig. 3) (Chamberlain et al. Reference Chamberlain, Menzies, Hampshire, Suckling, Fineberg, del Campo, Aitken, Craig, Owen, Bullmore, Robbins and Sahakian2008). In previous studies (Remijnse et al. Reference Remijnse, Nielen, van Balkom, Cath, van Oppen, Uylings and Veltman2006) reduced activations in the striatum have been restricted to reward feedback or to activations with a planning task (van den Heuvel et al. Reference van den Heuvel, Veltman, Groenewegen, Cath, van Balkom, van Hartskamp, Barkhof and van Dyck2005). It is interesting to note that the activated areas in the orbitofrontal cortex in our study are more posterior than those reported previously (Chamberlain et al. Reference Chamberlain, Menzies, Hampshire, Suckling, Fineberg, del Campo, Aitken, Craig, Owen, Bullmore, Robbins and Sahakian2008). We hypothesize that differences in task design, for example the use of abstract patterns or the inclusion of baseline stimuli (for which the correct response was known in advance, as in Remijnse et al. Reference Remijnse, Nielen, van Balkom, Cath, van Oppen, Uylings and Veltman2006), are the most likely explanation.
As mentioned earlier in the discussion, this study has some limitations. Because of the small sample size we mainly report results statistically uncorrected for multiple comparisons. We are aware of potential type 1 errors and will expand the sample to confirm the reported preliminary findings. Furthermore, although we established a high test–retest reliability for the RL task in healthy subjects (Freyer et al. Reference Freyer, Valerius, Kuelz, Speck, Glauche, Hull and Voderholzer2009), we cannot definitely rule out effects of fMRI practice or time in patients. The inclusion of a subgroup of patients without therapy (e.g. on a waiting list) would therefore be desirable. Although not significantly different between patients and controls, a more rigorous control of the time interval between the first and second sessions could strengthen the results by reducing the variability.
In summary, the current study (although preliminary) provides an insight into the effect of CBT on neuronal activations in OCD patients. The present results also extend current models of neural mechanisms of psychotherapy, suggesting that cognitive therapy might act in general from a dorsolateral prefrontal cortex-dominated network (DeRubeis et al. Reference DeRubeis, Siegle and Hollon2008) to cognitively more distant targets such as the orbitofrontal cortex and the basal ganglia.
Acknowledgments
This study was supported by a grant from the German Research Foundation (DFG), Grant VO 542/5-2.
Declaration of Interest
None.
