1. Application of Network Theory to Characterize
Functional Brain Networks in the At-Risk Mental State
The International Conference for Network Science
Chicago, USA.
June 2012
Presenter: L.D. Lord

2. At-Risk Mental State (ARMS)
• Psychiatric diagnosis
• “Attenuated” psychotic symptoms
• Negative Symptoms: - Cognitive / executive deficits
- Anhedonia
- Decline in social function
• Positive Symptoms: - Delusions (attenuated)
- Hallucinations (attenuated)
• Risk of developing Schizophrenia ~ 400x

3. ARMS & Transition to Psychosis
~30% = convert ARMS-NT
to SZ
ARMT-T
?

4. Anticipating the transition
to psychosis
Can we use Neuroimaging
as a diagnostic tool to
identify “true” prodromals?
Is there a biomarker for the transition to psychosis?
ARMS-Transition subjects VS ARMS-NON-Transition VS Controls

5. Why use Graph Theory to study the ARMS
and schizophrenia [SZ] (i)
- SZ is NOT a focal pathology
- SZ involves disruptions in the large-scale dynamics
of brain networks (i.e. functional integration)
Adapted from Bullmore & Sporns
(2012). Nature Neuroscience.

6. Why use Graph Theory to study the ARMS
and schizophrenia [SZ] (ii)
Healthy Schizophrenia
Degree Degree
Schizophrenia
Controls
Cortical Hub Regions:
Metabolically Costly , but Modular organization changes
maximize efficieny: Altered in SZ
Adapted from Fornito et al (2012). Neuroimage
Adapted from Buckner et al (2009). J Neurosci

7. The Anterior Cingulate Cortex (ACC):
(the present study)
MRI (Saggital view) 3D Rendition
ACC anomalies in SZ & ARMS subjects:
- Loss of grey matter (Pantelis et al. 2003a);
-Abnormal folding ((Fornito et al. 2008)
-)
- Neurochemical Imbalances (Stone et al. 2012)

8. 12 2
9 9
1 1
3 3
11 19
16
7
12 2
14 13
4 5
15
6 17
8

9. 11
8
2 13
10
14
15
9 7
12 18
1
19
6 3
17
4 5 16

10. Building Edges: fMRI and the BOLD signal
Deoxy-Hb
Oxy-Hb
O2 pressure
gradient
Experimental Task = Verbal Fluency Task

11. activity A activity B
activity A
“Functional Connectivity”
activity B
activity A
activity B
association matrix network representation
(partial correlations)
Adapted from: Bassett et al. (2008). J. Neurosci

12. Building Edges: Specifics
-Partial correlations: Limit the possible contributions to
pairwise correlation by third-parties
- Partial corr coef for node pair  Fisher Z  Averaged across subjects
- Avg. Fisher Z  p-values (for statistical significance)
- Edge under α threshold (p < 0.05) considered significant
(i.e. inclusion in the graph)
- Multiple comparison correction (FDR)

13. Regional Network Metrics of Interest
(i)
Degree-Centrality (DC):
Total # of connections
(ii)
Farness-Centrality (FC):
Avg. shortest path length between a given node and all the other nodes
constituting the network.
(iii)
Betweenness-Centrality (BC)
Frequency at which a node is visited when information is transferred along the
shortest routes between any given pair of nodes in the system

14. Controls
-Symmetry ρ = 39%
-Density
-ACC (BC)
ARMS NON-CONVERT
ρ = 40% ARMS CONVERTED
ρ = 33%

15. Betweenness-Centrality:

16. Degree-Centrality:
Btw-groups comparisons
achieved via permutation-test

17. fMRI contrasts: Activation data
ARMS-T > Controls
ARMS-T > ARMS-NT
* No ACC differences in activation!
Adapted from: Allen et al. (2012). Schiz Bull

18. Summary:
• Differences in standard graph metrics indicate a
reduced contribution of ACC to information routing in
executive networks in ARMS-T subjects.
• These ACC abnormalities differentiate ARMS-T subjects
from both controls and ARMS subjects who do not
develop psychosis (ARMS-NT).
• In ARMS-T patients, ACC abnormalities are present 8.5
months prior to the onset of psychosis
• ACC regional metric alterations are independent of:
age, clinical presentation, IQ, medication use

19. Challenges and Outstanding Questions (i)
• The Etiology: Stable Brain differences
(i.e. neurodevelopmental) vs. dynamic
changes leading up to SZ
• The Diagnosis: Bringing network analyses to the level of
individual patients – other imaging modalities (EEG), other
metrics.

20. Challenges and Outstanding Questions (ii)
• The Neurophysiology: Regional Alterations in
neurotransmission affecting network
parameters?
(preservation of activation
vs. disrupted connectivity)
• The Intervention: Once ARMS is properly diagnosed,
how do we prevent transition to psychosis
?

21. Acknowledgements
Martinos Center for Imperial College London King’s College London
Biomedical Imaging Experimental Medicine Institute of Psychiatry
(Boston, USA) (London, UK) (London, UK)
S. Hyde F.E. Turkheimer (now @ King’s) P. Allen
P. Expert (now @ King’s) O. Howes
R. Lambiotte (now @ Namur) M. Broome
P. Fusar-Poli
I. Valli
P. McGuire
Contact info: LD.LORD13@gmail.com