As part of our ongoing coverage of the 2013 International Congress on Schizophrenia Research (ICOSR), held 21-25 April in Colorado Springs, Colorado, we bring you session summaries from some of the Young Investigator Travel Award winners. For this report, we thank Jared Young of the University of California, San Diego.
10 June 2013. On the afternoon of Tuesday, 23 April, the ICOSR featured a series of short talks on "Cognitive Neuroscience Perspectives on Memory Network Dysfunction in Schizophrenia and Clinical Risk States." The primary theme of the session was the examination of brain network connectivity with fMRI BOLD activity during memory tasks. Patients with schizophrenia—medicated, and unmedicated—as well as first-episode patients (FES), and those with ultra-high risk for schizophrenia (UHR), were compared with healthy comparison subjects (HCS). Correlations with fMRI data and connectivity were also presented. On the whole, tasks were chosen and designed based on deficits that patients consistently exhibit outside of the scanner, although most presenters reported comparable test performances in the scanner between patients and HCS. The numerous accounts of differences in activity and connectivity between populations were the focus of this series of presentations.
Memory is a very broad topic. Thus, some studies focused on item encoding or retrieval, comparing these with recall, while others focused on connectivity during resting state, habituation, or learned irrelevance. Each of these aspects of memory requires distinct neural substrates, although there is overlap. The hippocampus was the main structure discussed in the session, although associations with areas such as the prefrontal cortex and visual association areas were presented. It appeared that while regional activation of patients differed in these tasks, not all differences were equal. In particular, activation during familiarity encoding was often strong in patients, while activation during recall tasks was weaker. Ultimately, the neurochemicals of these altered neural substrates during task performances can be investigated, perhaps providing targets for treatment development.
The first presentation, by Martin Lapage of McGill University, Montreal, Canada, began by asking, What are the neurochemical correlates of poor self-reflection (overconfidence) in schizophrenia patients? Lapage noted that verbal memory scores from the MATRICS cognitive test battery correlated well with such poor reflection and self-certainty deficits (Buchy et al., 2010; Lepage et al., 2008). Thus, in the current studies, Lapage and colleagues adapted the virtual town task (King et al,. 2002) to assess memory in FES patients and HCS. This task used a mixture of avatars, places, and objects, the association among which the subject was required to remember, then fMRI scans were conducted during recognition performance. Describing what proved to be a consistent theme throughout the rest of the session, the FES and HCS groups did not differ in overall performance of the task, but differences in fMRI activation were noted. Specifically, contrasts of fMRI activation between source and object recognition trials revealed greater ventrolateral prefrontal cortex (PFC) activity for avatars compared with objects in HCS only. FES patients did not exhibit such activation differences between depictions of people and with objects. There was evidence that right ventrolateral PFC activation during source recognition was modulated by self-reflectiveness. Thus, Lapage suggested these data support the premise that insight relies on accurate memory. During the question period, however, Lapage acknowledged that the data presented were not so much about deficits, per se, as about possible differences in cognitive mechanisms between FES and HCS.
Lisa Williams of Vanderbilt University, Nashville, Tennessee, used fMRI measures in response to repeated stimuli (activation habituation) to determine whether there may be differences between people with chronic schizophrenia compared with HCS. Specifically, activation of the hippocampus to repeated stimuli was examined in response to faces. Holt and colleagues (Holt et al., 2005) had previously demonstrated reduced habituation in schizophrenia patients compared with HCS when presenting faces in 40 s-blocks. Currently, 120 s-blocks were utilized and a target detection task was included to ensure that all subjects were attending to the task. Cognitive demands were kept as low as possible, however, in order to minimize potential confounds. The lack of differences in performance suggested all subjects were attending to the stimuli similarly. Williams showed that while HCS exhibited habituation, schizophrenia patients did not. An audience member suggested that this lack of habituation could have reflected the non-BOLD hippocampal response of the patients in the first place. Williams acknowledged this possibility and said that further analyses would assess habituation differences.
Matthijs Bossong of the Institute of Psychiatry, London, U.K., then discussed a study of novelty salience processing activation in the hippocampal-striatal midbrain circuit of people at UHR for psychosis. This study was conducted in part due to the possible link of this circuit and novelty salience to psychoses (Lodge and Grace, 2011). An oddball task was used whereby 73 percent were standard images, with 9 percent each of neutral oddballs, target stimuli, and non-targets; again, no performance differences were described in the populations compared, as UHR did not differ from HCS on accuracy or reaction time for detecting oddball stimuli. Bossong focused on the fMRI data, wherein right parahippocampal gyrus activation was lower in UHR subjects versus HCS in neutral and novel stimuli due to a lack of increase in UHR. Interestingly, UHR subjects had greater right caudate nucleus reactivity to novel stimuli, but less to neutral stimuli. Finally, functional connectivity between the ventral tegmental area and caudate was observed in controls but not in UHR subjects. Such alterations in specific regions of interest likely resulted in altered circuitry overall.
In the fourth talk, New-Fei Ho of Massachusetts General Hospital in Boston presented data investigating resting-state connectivity. Previously, Zhou and colleagues (Zhou et al., 2008) described altered hippocampal activity in the resting state of patients with schizophrenia. In this study, Ho and colleagues were primarily interested in the connectivity between among regions within the hippocampus (CA1, CA3, CA4, and dentate gyrus [DG]), comparing schizophrenia patients and HCS. It was observed that the CA3 region had great connectivity to the mesolimbic region, while CA1 connectivity was sparser, with some thalamocortical connection. DG, CA3, and CA1 had overlapping regions, while the orbitofrontal cortex and amygdala had common connectivity to DG, CA1, and CA3. Altered connectivity was observed in schizophrenia patients, a finding that was replicated in another testing cohort. Ho suggested that reversing such connectivity changes (e.g., reduced left DG coupling with cingulate cortex BA32, 24, and inferior frontal gyrus) could provide an intermediate target for treatment development. It is as yet unknown if these altered connections have functional significance, however, or if they are compensatory in nature.
In the next talk, Nina Kragulijac of the University of Alabama, Birmingham, also examined connectivity in hippocampal resting state in schizophrenia patients. Specifically, however, Kragulijac and colleagues were interested in whether connectivity alterations in schizophrenia patients related to elevated glutamate levels in the hippocampus and medial PFC compared with HCS. Examination of the hippocampal resting state of unmedicated schizophrenia patients and HCS revealed disconnectivity of several regions. No correlation between altered glutamate levels and connectivity strength was observed, however. It remains unclear what the functional significance of this poor connectivity in schizophrenia patients might be, but these data do not implicate a role for altered glutamate levels in underlying these differences.
Tanusree Das, University of Texas Southwestern, Dallas, started her talk by showing a useful schematic detailing the separation of hippocampal structures that could relate to relational memory (Tamminga et al., 2012). Das and colleagues hypothesized that decreased mossy fiber innervation may relate to increased sensitivity of the CA3 region (increased BOLD during relational memory tasks), leading to false associations. To test this hypothesis, they examined fMRI BOLD activity of medicated patients, unmedicated patients, and HCS performing a relational memory transitive inference task (Preston et al., 2004). Medicated schizophrenia patients exhibited lower BOLD intensity of activation during AB, BC, and DE recall. Increased DG/CA3 regions of activation were seen in unmedicated patients, supporting the original hypothesis. Das postulated that erroneous memories could lead to psychosis despite that no deficits in actual performance were observed.
In the seventh talk of the session, Chair John Ragland of the University of California, Davis, described his laboratory's interest in memories embedded in context. At first, Ragland described Eichenbaum's findings (Eichenbaum et al., 2007) that binding items within context occurs, in which the perirhinal cortex represents the items, the parahippocampal region represents the context, and the hippocampus links these two aspects. It has been demonstrated by Libby and colleagues (Libby et al., 2012) that these processes are affected in schizophrenia patients. The fact that performance in HCS improves over time, however, led Ragland and colleagues to suggest that there may be distinct aspects to recollection deficits in schizophrenia patients (Ragland et al., 2012).
Ragland presented current data using the task from Haskins and colleagues (Haskins et al., 2008), whereby encoding and retrieval were examined, with confidence ratings acquired to bin fMRI data. HCS exhibited greater activation in the PFC than did schizophrenia patients. Importantly, the perirhinal cortical activation correlated with familiarity, consistent with prior studies, and this effect was particularly strong in schizophrenia patients. These data support the premise that less efficient item-specific memory processing may be a compensatory mechanism in schizophrenia patients. Future studies exploring relational processes, particularly related to the PFC, will be conducted for treatment development.
In the next talk, Kristen Haut of Yale University, New Haven, Connecticut, also examined connectivity using fMRI, this time during a memory encoding and retrieval task. The premise of this work was to differentiate connectivity in recognition versus recall tasks, as schizophrenia patients exhibit greater deficits in the latter. These studies were conducted in UHR subjects versus HCS. Although fMRI data during scanner measurement were presented, they were collected outside of the scanner, and significant deficits in episodic memory were observed in the UHR group compared with HCS. Connectivity of the medial temporal lobe and visual association areas was more diffuse in UHR subjects. In the recognition part, greater connectivity was observed in UHR subjects. These data support previous evidence from this session that schizophrenia subjects, and even those at risk of developing schizophrenia, exhibit additional connectivity when familiarity can be used in tasks. This additional connectivity may be compensation for greater difficulty in recall tasks, or indeed may be the cause of difficulty of these patients in recall tasks.
In the ninth talk of the session, Janardhanan Narayanaswamy of the National Institute of Mental Health and Neurosciences, Bangalore, India, examined learned irrelevance in siblings of schizophrenia patients compared with HCS. Interestingly, Narayanaswamy presented data on slower reaction times in siblings of schizophrenia patients compared with HCS. Deficient bilateral hippocampal activation was also observed in siblings of patients relative to HCS. These data support the hypothesis that disrupted learned irrelevance occurs in subjects with a genetic risk for schizophrenia and that it is associated with reduced hippocampal activation. Thus, learned irrelevance may be a useful endophenotype for schizophrenia. What remains unclear, however, is why these siblings do not suffer from schizophrenia.
In the final talk, Sarah McEwan of the University of California, Los Angeles, presented data from the North American Prodrome Longitudinal Study (NAPLS). NAPLS is an eight-site consortium with, among other data, fMRI analyses for psychosis on UHR youths. Of 329 subjects (aged 12-33), a small percentage converted to schizophrenia, but most had not. McEwan described fMRI evidence generated during a paired associate learning task that pointed toward reduced cerebellar activity during encoding being an early indicator of imminent conversion to psychosis. Furthermore, compensatory activity in the medial temporal lobe may attenuate conversion to psychosis. Future studies will focus on frontal brain regions linked to successful encoding.—Jared W. Young.