Hyperactive error responses and altered connectivity in ventromedial and frontoinsular cortices in obsessive-compulsive disorder

Abstract

Background: Patients with obsessive-compulsive disorder (OCD) show abnormal functioning in ventral frontal brain regions involved in emotional/motivational processes, including anterior insula/frontal operculum (aI/fO) and ventromedial frontal cortex (VMPFC). While OCD has been associated with an increased neural response to errors, the influence of motivational factors on this effect remains poorly understood.

Methods: To investigate the contribution of motivational factors to error processing in OCD and to examine functional connectivity between regions involved in the error response, functional magnetic resonance imaging data were measured in 39 OCD patients (20 unmedicated, 19 medicated) and 38 control subjects (20 unmedicated, 18 medicated) during an error-eliciting interference task where motivational context was varied using monetary incentives (null, loss, and gain).

Results: Across all errors, OCD patients showed reduced deactivation of VMPFC and greater activation in left aI/FO compared with control subjects. For errors specifically resulting in a loss, patients further hyperactivated VMPFC, as well as right aI/FO. Independent of activity associated with task events, OCD patients showed greater functional connectivity between VMPFC and regions of bilateral aI/FO and right thalamus.

Conclusions: Obsessive-compulsive disorder patients show greater activation in neural regions associated with emotion and valuation when making errors, which could be related to altered intrinsic functional connectivity between brain networks. These results highlight the importance of emotional/motivational responses to mistakes in OCD and point to the need for further study of network interactions in the disorder.

Figure 1. Diagram of incentive flanker task

Prior to participation, subjects learned to associate two letters with a left button press (in this example, “S” and “K”) and two different letters with a right button press (in this example, “H” and “C”). During the task, subjects pressed the left or right button based on the identity of a target letter placed in the second, third, or fourth position in a string of five letters (all examples in figure show target in third position). The target letter was always different from flanking letters; “low” interference trials were those where target and flankers designated the same button press (middle row), while “high” interference trials were those where target and flankers designated opposite button presses (examples shown in top and bottom rows). Letter stimuli were presented on-screen for 300 ms, followed by a blank screen until a response was made or until response deadline was achieved. Feedback was presented immediately following response, indicating to the subject whether they made a correct response, an error of commission, or were not within the response deadline (i.e., omission error). Duration of feedback was varied based on the individual subject's RT such that total time between presentation of letter stimuli and end of feedback was 1500 ms. Feedback was followed by a 2000 ms blank inter-trial interval (ITI). Prior to letter stimuli, subjects received cues on each trial indicating whether an error (correct response) would result in a loss of money (failure to lose money) (LOSS trials), a failure to gain money (gain of money) (GAIN trials), or no change in money (NULL trials). The amount of money each trial was worth was real and presented on-screen (10 or 50 cents for loss or gain trials, 0 cents for null trials). In order to be able to decouple the BOLD signal associated with the response/feedback from that elicited by cues, cues were jittered between 1500 and 9500 ms in increments of 500 ms.

Figure 2. Errors > correct trials in OCD patients and control subjects

A) Whole-brain activations and deactivations (warm colors: activations, cool colors: deactivations; displayed at p < .005 with 20 contiguous voxels) for OCD patients and controls; B) Activity in VMPFC (x = 0, y = 51, z = -15, k = 51, z = 3.35) and left aI/fO (x = -33, y = 30, z = -3, k = 20, z = 3.2) was greater in OCD patients than control subjects; C) Post-hoc ANOVAs examining effects of medication status on activity in regions showing group difference revealed a trend toward an interaction between diagnosis and medication in VMPFC (F (1,73) = 3.5, p = .067). For aI/fO, there was a main effect of medication (F (1,73) = 6.5, p = .01). Color bars represent t scores

Figure 3. Loss > null errors in OCD patients and control subjects

A) Whole-brain activations (displayed at p < .005 with 20 contiguous voxels) for OCD patients and controls, showing activity in VMPFC (x = -9, y = 36, z = -24, k = 73, z = 3.91) and right aI/fO (x = 45, y = 24, z = -3, k = 34, z = 3.21) for the loss > null error contrast in OCD patients but not in controls; B) Group comparisons revealed greater activity in VMPFC (x = 0, y = 27, z = -21, k = 52, z = 3.65) and right aI/fO (x = 45, y = 24, z = -3, k = 21, z = 3.4) in OCD patients. There were no areas of deactivation for either group, and no regions where controls showed more activations than OCD patients; C) Post-hoc ANOVAs examining effects of medication status on activity in regions showing group differences found no main effects of medication (VMPFC: p = 0.81, aI/fO: p = 0.94) and no interactions between medication and diagnosis (VMPFC: p = 0.49, aI/fO: p = 0.83). Color bars represent t scores.

Figure 4

Functional connectivity with VMPFC. A) OCD patients showed greater positive connectivity than controls between VMPFC seed and right aI/fO (x = 39, y = 9, z = 3, k = 134, z = 3.63), left aI/fO (x = -27, y = 12, z = -15, k = 38, z = 4.09), and right thalamus (x = 15, y = -18, z = 18, k = 95, z = 3.27). Numbers above axial slices represent z-coordinates. Color bar represents t scores; B) Post-hoc ANOVAs examining effects of medication status on activity in regions showing group differences found no main effects of medication. There was a significant medication-by-diagnosis interaction in right aI/fO (F (1,73) = 4.4, p = .04), and no interactions in left aI/fO or right thalamus. Values on y-axes represent parameter estimates; C) Positive correlations between VMPFC-right aI/fO connectivity values and evoked activity in VMPFC for errors > corrects contrast (r = 0.32, p = .05) and right aI/fO for loss > null errors contrast (r = 0.42, p = .007). There was also a correlation (not shown) between VMPFC-right thalamus connectivity and right aI/fO activity for loss > null error contrast (r = 0.31, p = .05). Values represent parameter estimates.