Longitudinal Working Memory Functional Dysconnectivity Reflects Heterogeneity in UHR for Psychosis

Submission No:


Submission Type:

Abstract Submission 


Siwei Liu1, Jimmy Lee2, Jesisca Tandi1, Chenhao Wang1, Joseph Lim1, New Fei Ho2, Joann Poh1, R. Alison Adcock3, Richard Keefe3, Stephen Wood4, Ranga Krishnan1, Michael Chee1, Juan Zhou1


1Duke-National University of Singapore Medical School, Singapore, Singapore, 2Institute of Mental Health, Singapore, Singapore, 3Duke University, Durham, NC, 4The University of Birmingham, Birmingham, United Kingdom


Variation in trajectories of Ultra high risk (UHR) psychosis mental state posts challenge to schizophrenia prevention or onset delay intervention. Our previous work described the heterogeneity at this prodromal stage of schizophrenia in both brain structure changes and resting-state functional network differences (Wang et al. 2016, 2017; Ho et al. 2017). Functional dysconnectivity can be one of the altered brain substrates underlying clinical symptoms (Friston et al. 2016). Lower resting-state functional connectivity (FC) within the salience network (SN) in schizophrenia was detectable at the UHR stage (Wang et al. 2016). FC between the frontoparietal network (FPN) and the SN was disrupted when network integration fell apart in the UHR state (Wang et al. 2017).
FPN and SN are important for working memory (WM), which is largely compromised in schizophrenia and lesser in UHR group (Heuvel and Fornito 2014). Our previous work (Yaakub et al. 2013) showed that WM task-related activation in the FPN and SN differed between the UHR and controls. Importantly, such differences varied with WM demands. Compared to resting-state FC, task-based FC could better predict behavioural performance (Liang et al. 2016; Ito et al. 2017). However, the WM-related FC in UHR group and its longitudinal changes are largely unknown.
To cover the gap, we sought to examine the heterogeneity in the WM task-related FC changes in a group of UHR participants over time. We expected WM-related FC would link to individual differences in clinical trajectories.


Based on the longitudinal changes of UHR state within 2 years, participants were divided into 3 groups: 42 controls, 34 UHR remitters (UHR-R) and 42 UHR non-remitters (UHR-NR). We acquired fMRI (TR/TE = 2000/30 ms, 3 x 3 x 4 mm3, 28 slices) when participants performed WM task (Yaakub et al. 2013) at different WM demands, varying from information maintenance alone (low) to requiring both maintenance and manipulation (high). We used seed-based approach (gPPI) to compare task-related FC of the FPN and the SN among groups. Voxel-wise FC with six seeds (bilateral anterior insula, parietal, and dorsal lateral prefrontal cortex, identified based on task activation) was regressed on WM demands and groups, controlling for age, gender, education, ethnicity, handedness and task accuracy. Linear mixed modelling methods were used to test longitudinal FC changes and the association between FC and clinical syndromes.


Task performance was worse at high WM demand as expected, but no significant difference was found between groups or over time. Compared to controls, higher FC between the FPN (superior parietal gyrus) and the SN (insula) at low demand was observed in the UHR-NR group at baseline (Fig. 1). Within the SN, WM-demand related FC between right insula and thalamus varied among three groups (Fig. 2): low FC at low demand and high FC at high demand in controls; high FC at low demand but low FC at high demand in the UHR-R group. In contrast, UHR-NR group had high thalamus-insula FC in both WM demands.
Moreover, longitudinal FC increase only occurred within the SN at high WM demand in the UHR-R group (Fig. 2), while other task-related baseline group differences of FC remained stable over time. Importantly, the rate of intra-SN increase of FC over time at high WM demand was associated with decline of the positive psychosis syndromes in the UHR-R group.
Supporting Image: F1v2_1000width.png
Supporting Image: F2v2_1000width.png


In support of the functional dysconnectivity hypothesis, we revealed that UHR state was accompanied by altered brain FC during WM task performance. In contrast to lower SN FC at rest, UHR state showed SN-FPN hyper-connectivity in task with low WM demand, suggesting the importance of studying UHR both at resting and in task. Importantly, intra-SN FC increase at high WM demand was linked to positive psychosis syndrome reduction in remitters, while no FC changes if UHR state persisted. We argued that task FC could reflect the clinical heterogeneity of the UHR group.

Disorders of the Nervous System:

Schizophrenia and Psychotic Disorders 1

Higher Cognitive Functions:

Executive Function

Imaging Methods:


Learning and Memory:

Working Memory 2

Modeling and Analysis Methods:

fMRI Connectivity and Network Modeling



1|2Indicates the priority used for review