Fronto-cingulate effective connectivity in major depression: A study with fMRI and dynamic causal modeling

Abstract

Functional imaging studies are indicating disrupted error monitoring and executive control in a fronto-cingulate network in major depression. However, univariate statistical analyses allow only for a limited assessment of directed neuronal interactions. Therefore, the present study used dynamic causal modeling (DCM) of a fronto-cingulate network to re-analyze the data from a preceding fMRI study in 16 drug-free patients with major depression and 16 healthy controls using the Stroop Color-Word Test (Wagner et al., 2006).

In both groups, a significant reciprocal interregional connectivity was found in a cognitive control network including prefrontal cortex (PFC) and dorsal anterior cingulate cortex (ACC). With regard to intrinsic connections we detected a significant difference for dorsal to rostral ACC connectivity between depressive patients and controls in terms of higher connectivity in patients. Additionally, a task by group interaction was observed for the bilinear interaction signaling enhanced task-related input from the dorsal to rostral ACC in subjects with depression. This could be related to the inability of patients to down-regulate rostral ACC activation as observed in the previous univariate analysis.

The correlation between interference scores and intrinsic connections from dorsal ACC to dorsolateral PFC (DLPFC) was significant for both groups together, but no significant group differences in correlations could be detected. Thus, the observed relationship between control functions of the dorsal ACC exerted over DLPFC and interference scores appears to be valid in both patients with depression and controls. The findings are consistent with current models of a differential involvement of the fronto-cingulate system in the pathophysiology of major depression.

Introduction

Depression is a serious and mostly incapacitating disorder which is characterized by a range of symptoms affecting both emotional and cognitive domains. These symptoms have been shown to be associated with changes in cerebral activation in a mainly fronto-cingulate network. In a recent study we tested the hypothesis of a disorder-related disruption in prefrontal and anterior cingulate functions by employing the Stroop Color-Word Test (SCWT) (Stroop, 1935) in a sample of medication-free patients with unipolar depression (Wagner et al., 2006). The SCWT demands a number of different cognitive subprocesses like error monitoring, response selection and suppression of inadequate responses (Banich et al., 2001, Cohen et al., 1990). We found that behavioral performance and dorsal ACC activation of depressed patients was comparable to healthy controls, but observed a relative hyperactivity in the rostral part of ACC, in the left ventrolateral (VLPFC) during the incongruent conditions and additionally in the left dorsolateral prefrontal cortex (DLPFC) during the interference condition. The rostral ACC activity and left DLPFC activity were correlated with the Stroop interference effect.

The ACC has been subdivided into a dorsal cognitive (Bush et al., 1998) and rostral affective (Etkin et al., 2006, Whalen et al., 1998) division. According to this functional specialization, the affective subdivision of the ACC has connections to orbitofrontal cortex, amygdala, hippocampus and periaqueductal gray whereas the cognitive subdivision is interconnected with the DLPFC, parietal cortex, premotor, and supplementary motor areas (Devinsky et al., 1995, Vogt et al., 2005, Vogt et al., 2006). The rostral ACC plays a critical role for the integration of affective and cognitive functions (Mayberg, 2000, Wagner et al., 2006). Moreover, the affective division has been shown to correlate with autonomic responses during execution of cognitive functions of the Stroop task (Critchley et al., 2005). Hence, the findings of this preceding study indicating a reduced ability to down-regulate the rostral part of the ACC in patients with depression suggest an overactive affective system within which the rostral ACC is known to play a central role. The result is moreover related to previous reports of a dynamic interplay between these cingulate subdivisions in healthy subjects (Drevets and Raichle, 1998), showing a relative deactivation in dorsal ACC during emotional paradigms and a relative deactivation in rostral ACC during cognitive tasks (Whalen et al., 1998). Given convergent evidence that the ACC operates in close interaction with the PFC (Brown et al., 1999) this recent finding indicates a disruption within this fronto-cingulate system in major depression.

Although the exact role and functional interplay of PFC and the ACC during the Stroop task is still a matter of debate, in healthy subjects most studies demonstrated close fronto-cingulate interactions (Gehring and Knight, 2000, Koski and Paus, 2000, Leung et al., 2000, Mead et al., 2002, Pardo et al., 1990, Peterson et al., 1999). However, the precise temporal sequence and directionality of regional involvement has not entirely been clarified. Different models with regard to the functional subdivision between ACC and DLPFC have been developed. According to a widely accepted model, the ACC monitors conflict and performance thus signaling when adjustments in control are needed whereas the DLPFC exerts cognitive control (Cohen et al., 2000, Kerns et al., 2004, MacDonald et al., 2000). Notably, an involvement of the ACC appears to decrease after periods of initial task performance relative to increased activity in DLPFC (Erickson et al., 2004). This finding suggests a functional parcellation between these two regions in implementing cognitive control (Cohen et al., 2000).

Although abundant literature exists with regard to functional segregation of different brain areas involved in cognitive control processes, information about interactions between defined areas has been acquired only indirectly. Therefore, in order to further examine concerted network operations, the analysis of cerebral activity has more recently moved to the search for connectivity within activated dynamic and transient networks (Durstewitz and Deco, 2008). The notion of functional and effective connectivity was originally defined in the context of electrophysiological studies. Functional connectivity was conceptualized as the observed correlations of different neurophysiological events, whereas the notion of effective connectivity refers explicitly to the influence that one neural system exerts over another (Friston, 1994). Dynamic causal modeling (DCM) has been introduced for addressing effective connectivity (Friston et al., 2003). Etkin et al. (2006) performed a psychophysiological interaction analysis of the Stroop interference effect and partially included also a DCM analysis. They demonstrated that emotional conflict is resolved through top-down inhibition of amygdala activity by the rostral cingulate cortex. These findings are supporting the role of the affective subdivision proposed by previous authors.

The present study aimed to explicitly model effective connectivity within the network subserving the Stroop test interference effect in healthy controls and patients with unipolar major depression. The analysis was performed using DCM (Penny et al., 2004b) and the selected model aimed to describe propagated information processing via primary sensory visual input to higher prefrontal and midfrontal regions. DCM is based on a set of first order differential equations and allows modeling dynamic aspects of neuronal interaction. This dynamic causal modeling strategy represents an advantage as compared to other linear modeling approaches based on instantaneous relationships such as Psychophysiological Interactions (PPI) or Structural Equation Modeling (SEM). DCM is particularly suitable to model event-related paradigms such as the Stroop task where a clear driving sensory input can be defined (Schlösser et al., 2006, Schlösser et al., 2007).

The effect of different levels of conflict and required cognitive control associated with incongruent and congruent Stroop tasks was modeled as bilinear interactions. Parameter estimation is based on a Bayesian approach by means of an expectation maximization algorithm (Penny et al., 2003).

Previous findings have suggested a functional dissociation of dorsal and rostral ACC. We hypothesized that this interaction is disrupted in major depression. Following the findings from our previous univariate analysis, we expected higher effective connectivity between the rostral and the dorsal portion of the ACC in patients with major depression relative to healthy controls signaling enhanced involvement of the rostral, affective division of the ACC. In addition, we examined effective connectivity between the PFC (DLPFC and VLPFC) and the dorsal ACC in both groups and analyzed the data in an exploratory fashion.

Since we were primarily interested in task-related group differences in effective connectivity, the model was restricted to the left hemisphere where the main findings with regard to group differences or task × group interaction effects were evident in the previous univariate analysis. This focus was also consistent with a previous review of Stroop studies suggesting involvement of the ACC and left PFC (Cabeza and Nyberg, 2000).