Previous regional cerebral blood flow (rCBF)
studies on patients with unipolar major depressive dis-
order (MDD) have analysed clusters of voxels or single
regions and yielded conflicting results, showing either
higher or lower rCBF in MDD as compared to normal
controls (CTR). The aim of this study was to assess
rCBF distribution changes in 68 MDD patients, investi-
gating the data set with both volume of interest (VOI)
analysis and principal component analysis (PCA). The
rCBF distribution in 68 MDD and 66 CTR, at rest,
was compared. Technetium-99m d,l-hexamethylpropylene
amine oxime single-photon emission tomography was
performed and the uptake in 27 VOIs, bilaterally, was
assessed using a standardising brain atlas. Data were
then grouped into factors by means of PCA performed
on rCBF of all 134 subjects and based on all 54 VOIs.
VOI analysis showed a significant group × VOI × hemi-
sphere interaction (P<0.001). rCBF in eight VOIs (in the
prefrontal, temporal, occipital and central structures) dif-
fered significantly between groups at the P<0.05 level.
PCA identified 11 anatomo-functional regions that inter-
acted with groups (P<0.001). As compared to CTR,
MDD rCBF was relatively higher in right associative
temporo-parietal-occipital cortex (P<0.01) and bilateral-
ly in prefrontal (P<0.005) and frontal cortex (P<0.025),
anterior temporal cortex and central structures (P<0.05
and P<0.001 respectively). Higher rCBF in a selected
group of MDD as compared to CTR at rest was found
using PCA in five clusters of regions sharing close
anatomical and functional relationships. At the single VOI
level, all eight regions showing group differences were
included in such clusters. PCA is a data-driven method
for recasting VOIs to be used for group evaluation and
comparison. The appearance of significant differences
absent at the VOI level emphasises the value of analys-
ing the relationships among brain regions for the investi-
gation of psychiatric disease.
