Regional white matter hyperintensity burden in automated segmentation distinguishes late-life depressed subjects from comparison subjects matched for vascular risk factors

Abstract

Objective: Segmented brain white matter hyperintensities were compared between subjects with late-life depression and age-matched subjects with similar vascular risk factor scores. Correlations between neuropsychological performance and whole brain-segmented white matter hyperintensities and white and gray matter volumes were also examined.

Method: Eighty-three subjects with late-life depression and 32 comparison subjects underwent physical examination, psychiatric evaluation, neuropsychological testing, vascular risk factor assessment, and brain magnetic resonance imaging (MRI). Automated segmentation methods were used to compare the total brain and regional white matter hyperintensity burden between depressed patients and comparison subjects.

Results: Depressed patients and comparison subjects did not differ in demographic variables, including vascular risk factor, or whole brain-segmented volumes. However, depressed subjects had seven regions of greater white matter hyperintensities located in the following white matter tracts: the superior longitudinal fasciculus, fronto-occipital fasciculus, uncinate fasciculus, extreme capsule, and inferior longitudinal fasciculus. These white matter tracts underlie brain regions associated with cognitive and emotional function. In depressed patients but not comparison subjects, volumes of three of these regions correlated with executive function; whole brain white matter hyperintensities correlated with executive function; whole brain white matter correlated with episodic memory, processing speed, and executive function; and whole brain gray matter correlated with processing speed.

Conclusions: These findings support the hypothesis that the strategic location of white matter hyperintensities may be critical in late-life depression. Further, the correlation of neuropsychological deficits with the volumes of whole brain white matter hyperintensities and gray and white matter in depressed subjects but not comparison subjects supports the hypothesis of an interaction between these structural brain components and depressed status.

Conflict of interest statement

Dr. Sheline serves as a consultant for Wyeth Pharmaceuticals, Cyberonics, and Eli Lilly; she also serves on the speaker’s bureau of Eli Lilly. Dr. Mintun serves as a consultant for Elan Pharmaceuticals, Hoffman-LaRoche, Avid Radiopharmaceuticals, and Eisai. Dr. Mc-Kinstry serves as a speaker for BioMarin Pharmaceutical. Drs. Price, Barch, Epstein, Wilkins, and Snyder and Mr. Vaishnavi, Mr. Couture, and Dr. Schechtman report no competing interests.

Figures

FIGURE 1

Automated Segmentation of White Matter Hyperintensitiesa a The first row shows T2-weighted images. The second row shows T1-weighted images. The third row shows images following automated segmentation. Of note are the following findings indicated by arrows in row 3: 1) minimal periventricular hyperintensity; 2) the caudate appears similar to white matter hyperintensity on the T2-weighted image but is correctly segmented as gray matter by the segmentation algorithm; 3) small white matter hyperintensity is not clearly identifiable on a T2-weighted image; 4) area of gray matter, from gyral folding in the temporal lobe that appears as white matter hyperintensity on T1- and T2-weighted images but is correctly segmented as gray matter; 5) large periventricular hyperintensity; 6) area potentially appearing as white matter hyperintensity that is actually correctly segmented as gray matter; 7) large periventricular hyperintensity; 8) white matter hyperintensity; and 9) white matter hyperintensity.

FIGURE 2

Differences Between Depressed Patients and Comparison Subjects in Regional White Matter Hyperintensity Volumesa a The two top panels display the averaged difference maps between depressed and comparison subjects in oblique right and left views, respectively. The semitransparent display allows the lesions to be seen three-dimensionally, underlying the surface Brodmann area. Selected Brodmann areas are indicated in the upper three-dimensional brain representations as well as the bottom individual panels (A–D), which are four horizontal slices through the brain that indicate the z axis level of the panels. They are as follows: A) right superior longitudinal fasciculus 1/fronto-occipital fasciculus (blue) as defined by Petrides and Pandya (45) and Schmahmann and Pandya (28), underlying right Brodmann’s areas 6, 8, and 9 (extending from z=42–58); B) right superior longitudinal fasciculus 2 and 3 (purple) (26, 43), underlying the right dorsolateral prefrontal cortex (Brodmann’s area 9/46) (extending from z=26–38), a second lesion in the right superior longitudinal fasciculus 2 and 3 (green), underlying right Brodmann’s area 9/32 (extending from z=26–38), and the left superior longitudinal fasciculus 2 and 3 (red) in white matter underlying Brodmann’s area 46/32 (extending from z=18–26); C) left uncinate fasciculus/frontal operculum (yellow) in white matter underlying Brodmann’s areas 44/6 and 9 (extending from z=10–22) and the right extreme capsule (pink) in white matter underlying the insula (extending from z=6–14); and D) inferior longitudinal fasciculus (orange) in white matter adjacent to the right parahippocampus (extending from z=−14 to −6).

FIGURE 3

White Matter Hyperintensity Comparison by Region in Depressed Patients and Comparison Subjects a Regions correspond with the lesion locations shown in Figure 2: right superior longitudinal fasciculus division 1/fronto-occipital fasciculus [blue]; right superior longitudinal fasciculus divisions 2 and 3 [purple]; second region of right superior longitudinal fasciculus divisions 2 and 3 [green]; left superior longitudinal fasciculus divisions 2 and 3 [yellow]; left uncinate fasciculus/frontal operculum [red]; right extreme capsule [pink]; and right inferior longitudinal fasciculus [orange]). *p