Retrosplenial cortex connectivity in schizophrenia

Abstract

In this paper, we build on our previous analysis [Bluhm, R.L., Miller, J., Lanius, R.A., Osuch, E.A., Boksman, K., Neufeld, R.W.J., et al., 2007 Spontaneous low-frequency fluctuations in the BOLD signal in schizophrenic patients: anomalies in the default network. Schizophrenia Bulletin 33, 1004–1012] of resting state connectivity in schizophrenia by examining alterations in connectivity of the retrosplenial cortex. We have previously demonstrated altered connectivity of the posterior cingulate/precuneus, particularly with other regions of the “default network” (which includes the medial prefrontal cortex and bilateral lateral parietal cortex). It was hypothesized that the retrosplenial cortex would show aberrant patterns of connectivity with regions of the default network and regions associated with memory. Patients with schizophrenia (N = 17) and healthy controls (N = 17) underwent a 5.5-min resting functional magnetic resonance imaging scan. Lower correlations were observed in patients with schizophrenia than in healthy controls between the retrosplenial cortex and both the temporal lobe and regions of the default network. In patients with schizophrenia, activity in the retrosplenial cortex correlated negatively with activity in bilateral anterior cingulate gyrus/medial prefrontal cortex (BA 32/10), despite the fact that these regions, as part of the default network, were expected to show positive correlations in activity. Connectivity of the retrosplenial cortex was greater in patients with more positive symptoms with areas previously associated with hallucinations, particularly the left superior temporal gyrus. These results suggest that spontaneous activity in the retrosplenial cortex during rest is altered in patients with schizophrenia. These alterations may help to explain alterations in self-oriented processing in this patient population.

Introduction

Interest has recently increased in examining patterns of connectivity between brain areas that occur during rest, that is, that appear to be independent of the performance of specific tasks that require activity of the areas of interest. Early research in this field demonstrated that there are low-frequency (~ 0.1 Hz) oscillations in resting state activity between functionally related areas such as motor (Biswal et al., 1995) or visual or language (Cordes et al., 2000) areas. More recently, a number of studies have demonstrated resting state functional connectivity among a group of regions previously implicated in the so-called “default state,” which is said to occur when an individual is resting and engaging in stimulus-independent thought. Key areas of this “default network” are the medial prefrontal cortex (mPFC), posterior cingulate cortex (PCC)/precuneus and bilateral lateral parietal regions (Fox et al., 2005, Fransson, 2005). These areas are not only positively correlated with each other at rest; they are also negatively correlated with a second network of areas implicated in a variety of cognitive tasks including the dorsolateral prefrontal cortex, premotor areas and the intraparietal sulcus (Fox et al., 2005, Fransson, 2005). This second network has been called the “task-positive” network (Fox et al., 2005). Despite the fact that the function of the default network remains unclear, and that there are doubts as to the utility of resting state analyses (e.g. Morcom and Fletcher, 2007a, Morcom and Fletcher, 2007b), there is a rapidly growing literature on the default network in both healthy adults and a variety of psychiatric conditions (Buckner et al., 2008).

Alterations in the activity of the default and/or task-positive networks may explain deficits observed both clinically and experimentally in schizophrenia. First, schizophrenic patients perform poorly on various cognitive tasks and show abnormal patterns of activation in brain regions associated with the task-positive network (Cornblatt and Keilp, 1994, Cirillo and Seidman, 2003). Second, the areas of the default network, particularly the mPFC and PCC, have been associated with self-reflection (Northoff and Bermpohl, 2004) and the default state has been described as one of self-monitoring (Gusnard et al., 2001). Key features of schizophrenia, such as auditory hallucinations and paranoid ideation, can be understood as alterations in self-perception (Flashman, 2004).

There also is emerging evidence of functional disconnectivity in the brain in schizophrenia, both between widespread, diverse areas of cortex and subcortical structures (Liang et al., 2006), and within the default network (Bluhm et al., 2007, Zhou et al., 2007). Two studies examining activity of the default network during a low-level cognitive task have demonstrated that patients with schizophrenia have more pronounced task-induced deactivations in structures in the default network (Harrison et al., 2007) and that the activity in the default network in schizophrenia differs in both spatial extent and temporal frequency from that of healthy control subjects (Garrity et al., 2007). With regard to alterations in temporal frequency in the default network, studies of resting state connectivity in healthy individuals have generally found that correlations between brain regions occur at very low frequencies, specifically less than 0.1 Hz. Garrity et al. (2007) found that there were differences between healthy control subjects and patients with schizophrenia in a frequency range from 0.03 to 0.24 Hz. In addition to these studies of activity in the default network during a cognitive task, our previous study (Bluhm et al., 2007) examined low-frequency (< 0.1 Hz) connectivity within the default network during a resting state. This study used a “seed region” analysis based on methods described in previous work (Fransson, 2005), and found significantly less correlation in low-frequency oscillations between a PCC/precuneus seed region and other areas in the default network (Bluhm et al., 2007). Connectivity in the default network in patients with schizophrenia has also been shown to vary with positive (Garrity et al., 2007, Bluhm et al., 2007) and negative (Bluhm et al., 2007) symptoms.

Although studies have begun to examine differences in connectivity of the various nodes of the default network in healthy adults (Margulies et al., 2007, Uddin et al., 2009), to date no studies have been published that examine connectivity of different nodes of the default network in patients with schizophrenia. We present here a follow-up study to our investigation of altered PCC/precuneus connectivity in schizophrenia; the focus of this article is on the retrosplenial cortex.

The retrosplenial cortex has been implicated in the pathophysiology of schizophrenia in both human studies and animal models. It has been shown to exhibit abnormal activity in a semantic memory task (Tendolkar et al., 2004) and in memory encoding (Hofer et al., 2003) in patients with schizophrenia. Grey matter volume deficits in the retrosplenial and posterior cingulate cortices have also been observed (Mitelman et al., 2005). Finally, the retrosplenial cortex is one of the first regions to show degenerative changes in phencyclidine animal models of psychosis (Olney and Farber, 1995). Destruction of pyramidal neurons in retrosplenial cortex disrupts the brain circuit that connects the hippocampus and parahippocampal gyrus and the thalamus in other animal models (Sharp et al., 2001).

The retrosplenial cortex is also of interest in the context of the default network. Previous studies using a seed region to investigate connectivity of the posterior cingulate and other areas of the default network (Fox et al., 2005, Fransson, 2005, Bluhm et al., 2007) have focused on the posterior cingulate/precuneus. Vogt et al. (2006) have shown, however, that although there are strong interconnections within the subdivisions of the posterior cingulate gyrus, there are differences in the connectivity of the dorsal posterior cingulate, the ventral posterior cingulate and the retrosplenial cortex (Brodmann areas 29 and 30, forming the ventral bank of the posterior cingulate gyrus) (Vogt et al., 2006). Although the retrosplenial cortex has been found to correlate with the PCC, thus suggesting that it may be part of the default network, it is also known to be involved in memory (Ranganath et al., 2005) including autobiographical memory and spatial memory (Iaria et al., 2007), and in making self-referential judgments (Johnson et al., 2005). A recent meta-analysis (Spreng et al., 2009) further suggests that there may be a common set of brain areas, which includes the retrosplenial cortex, that underlies the default mode, autobiographical memory, spatial navigation, imagining oneself in the future and theory of mind; this common set of brain regions includes the retrosplenial cortex.

Thus, the retrosplenial cortex appears to be a good candidate for investigating altered connectivity of the default network in schizophrenia. Given previous findings, using this data set, of abnormal connectivity in the default network of schizophrenic patients (Bluhm et al., 2007) and the wider frequency range in which these correlations have been observed in this patient population (Garrity et al., 2007, Calhoun et al., 2008), we hypothesized that schizophrenic patients, in comparison to healthy controls, would show altered connectivity between the retrosplenial cortex and other regions of the default network within an extended range of resting state oscillations (< 0.3 Hz). In particular, we expected patients with schizophrenia to show less positive correlation between the retrosplenial cortex and default network regions and less negative correlation with task-positive regions. We further hypothesized that, in patients with schizophrenia, connectivity of the retrosplenial cortex would correlate with the extent of positive and of negative symptoms in brain regions previously implicated in the etiology of schizophrenic symptoms, including the temporal lobes, frontal lobes and striatum (Taylor et al., 2005, Garrity et al., 2007, Bluhm et al., 2007).