Network analysis of trait and state abnormalities in depression

NeuroJmage

13, Number

6, 2001, Part 2 of 2 Parts 10

E bl@

PSYCHIATRY

Network Analysis of Trait and State Abnormalities in Depression Helen Mayberg*t, Robin Westmacott*t, Anthony McIntosh*? *Rotman Research Institute of Baycrest Centre TUniversity of Toronto Objectives. We have previously described resting-state abnormalities in regional glucose metabolism and blood flow using PET in patients with depression, and changes associated with treatment and clinical recovery (l-3). Although the relative contribution of individual regions varies as a function of clinical state, involvement of cortical (frontal F9/46/6, parietal P40), paralimbic (anterior and posterior insula, cingulate (pregenual Cg24, subgenual Cg25, posterior Cg23), hippocampus Hc) and subcortical (brainstem, cerebellum, striatum, thalamus) regions is seen across studies. Cortical deficits normalize with treatment (state effects); paralimbic and subcortical regions show a more complex state-trait pattern. Changes in these same regions are also seen with transient provoked sadness, with differences discriminating controls from depressed patients as well as fully-remitted patients from acutely-ill patients (3-4). These converging findings suggest the presence of a depression “network”. To further explore this hypothesis, multivariate partial least squares (PLS) and structural equation modeling (SEM)(5) were used to explicitly test disease-specific and state-specific functional interactions among a cortical-limbic subset of these regions, based on our previously proposed depression model (3). Methods. Resting state 150-water PET scans were performed in 3 groups of subjects (2 scans/person): acutely-ill depressed patients (n=8, medication free), euthymic, fully-recovered patients (n=9, on maintenance medication), and healthy volunteers (n=8). Following spatial and value normalization using SPM95, PLS was used to confirm differences between groups in dorsal F9, orbital Fll, Cg24, Cg25 and Hc, as previously demonstrated using univariate subtraction analyses (l-4). SEM was conducted to estimate the strength and direction of effective connections between regions as hypothesized in our depression model, informed by known anatomical and physiological pathways. ReSUkS. Task PLS confirmed significant group differences involving F9 (xyz coordinates 36,46,28), (-8,6,-12) and Cg24 (6,34,16), in a pattern comparable to that seen with univariate methods. and Cg24 from Hc, Fll, and Cg25 as a function of group. Pair-wise analyses demonstrated differences between patients 0 ...$-.p and controls (trait effects) as well as depressed and remitted $2 & patients (state effects). SEM identified significant trait and state 248 :: paths as illustrated in the figure below. Solid arrows=positive path coefficients; dotted arrows=negative coefficients. @

Fll (38,40,-12), Hc (26,-34,-8) PLS results showed segregation

Cg25 of F9

STATE .. .. ...-.E

_--

$(**&i&-

The F9-Cg24-Cg25 network appears to be a core disease abnormality, unaffected by current clinical status. In contrast, the F9-Hc-Fll network varies as a function of clinical state, with the overlap in remitted patients and controls further suggesting that this network and its components are critical targets of antidepressant treatment. PLS combined with SEM provides a complementary perspective to understanding resting-state PET abnormalities seen in patients with depression. These data provide a foundation for future studies examining mechanisms underlying treatment response and non-response and disease vulnerability. References 1. 2. 3. 4. 5.

NeuroReport 8:1057-1061, 1997 Biol Psych 48:830-843, 2000 Am J Psych 156:675-682, 1999 Neuroimage 7(S4):2901, 1998 Human Brain Mapping 2:2-22, 1994

Support:

Dana Foundation,

Stanley

Foundation,

NARSAD,

MH49553

s1071