Title: Diffusion-Tensor Imaging: Executive Function in Subcortical Ischemic Vascular Disease and Mild Cognitive Impairment
1Diffusion-Tensor Imaging Executive Function in
Subcortical Ischemic Vascular Disease and Mild
Cognitive Impairment
- Stephen Correia, Ph.D.
- Dementia Research Fellow,
- Neuropsychology
3 February 2005
2Diffusion-Tensor Imaging
- MRI technique for in-vivo characterization of 3D
white matter microstructure. - Measures magnitude and direction of water
diffusion in biological tissue in 3D. - More sensitive to white matter changes than
conventional MRI sequences.
3DTI Basics Water Diffusion
Isotropic
Anisotropic
Adapted from Beaulieu (2002). NMR in Biomed
15435-455
4DTI Scalar Parameters
- Trace The magnitude of diffusion in a voxel.
- Fractional Anisotropy (FA) The extent to which
diffusion is directionally restricted.
5DTI Scalar Maps
T2-weighted
FA map
Trace map
T2 FA Moseley et al. (2002) Brain
Cognition 50396-413. Trace Molko et al. (2001)
Stroke 32(9) 2049-54
6DTI Basics White Matter Integrity
- Lesioned white matter
- Increased diffusion (increased Trace)
- Decreased anisotropy (decreased FA)
- Etiology of DTI changes unclear
- Axon loss membrane breakdown
- Demyelination
- Gliosis/inflammation
7Why study frontal systems in SIVD MCI?
- Frontal-subcortical circuit disruption in SIVD
- Executive and behavioral impairment
- Functional decline and dementia.
- Executive impairment in MCI (CDR0.5) may hasten
conversion to dementia. - Use DTI and cognitive testing as independent
probes of frontal systems integrity and function
to identify a subset of SIVD and MCI patients at
greater risk for conversion to dementia.
Almkvist (2000), The matter of white matter,
81-95. Albert et al. (2001), JINS 7(5)
631-5. Chen et al (2000), Neurology 55(12)
1847-53.
8Patient Groups
- SIVD Subcortical ischemic vascular disease
- Subcortical hyperintensities (SH) on T2-weighted
MRI lacunar infarction on T1 - Gliosis and demyelination due to underlying small
vessel disease.
70-year-old normal
52-year-old CADASIL
65-year-old CADASIL
9Patient Groups
- CADASIL Cerebral autosomal dominant
arteriopathy with subcortical infarction and
leukoencephalopathy. - Inherited form of SIVD.
- Effect on brain parenchyma same as in SIVD.
- Relatively pure form of SIVD, excellent model.
- MCI
- Petersen criteria for amnestic MCI
- Normal controls
10Subcortical Ischemic Vascular Disease (SIVD)
- Subcortical hyperintensities (SH) on T2-weighted
MRI lacunar infarction - Gliosis and demyelination due to underlying small
vessel disease.
70-year-old normal
52-year-old CADASIL
65-year-old CADASIL
11Impact of SIVD
- Mental flexibility
- Speed of processing
- Complex attention
- Executive functioning working memory
- Apathy
12Prior Studies of DTI
- DTI in Aging
- Anterior posterior gradient of DTI changes.
(e.g., Pfefferbaum, 2000) - Correlations w/executive function. (e.g.
OSullivan, 2001, Madden 2004) - DTI in SIVD
- DTI abnormalities in normal appearing white
matter (NAWM) - DTI in NAWM more strongly correlated w/executive
function than DTI in SH. (OSullivan 2004) - DTI in MCI
- Little progression of anterior-posterior gradient
found in normal aging (Head, 2004) - DTI changes in regions expected for AD (left CSO,
temporal lobes, left HC) (Fellgeibel, 2004) - Association of DTI w/cognitive function not well
studied.
13Objectives
- To assess white matter integrity in patients with
SIVD vs. MCI vs. normal controls using DTI. - To determine the association between DTI
parameters in white matter and attention/executive
function and processing speed.
14Hypotheses
- Increased FA and decreased Trace in SIVD MCI
vs. NC. - FA and Trace in NAWM will correlate significantly
with performance on tests of attention/executive
function and psychomotor processing speed.
15Method
- Subjects recruited from Butler Hospital Memory
Aging Program _at_ Brown - NC recruited from family members of patients
- MRI done generally within 2 months of cognitive
testing.
16Key inclusion criteria SIVD
- n 9 (4 CADASIL)
- Identified mainly on radiological grounds for
protocol different than that of the MCI subjects. - Greater than expected SH for age on a visual
rating scale (Vataja et al., 2003 Eur J Neurol
10, 625-31) - Cognitive complaint
- Consensus diagnosis of SIVD or genetically
confirmed CADASIL - MMSE 24
- Global CDR 0.5
- ADL normal or only slightly impaired
- Excluded diagnosis of probable or possible AD
17Key inclusion criteria MCI
- n 9
- Documented memory complaint
- MMSE 24
- Global CDR 0.5
- ADL normal or only slightly impaired
- 1.5 SD below age-corrected mean on HVLT-R
delayed recall or retained - Excluded diagnosis of probable or possible AD
18Key inclusion criteria NC
- Absence of significant memory complaint
- MMSE within normal limits
- CDR 0
- ADL normal
- Normal memory function for age
19DTI Acquisition
- Siemens Symphony 1.5T
- 3 acquisitions with offset in slice direction by
0.0mm, 1.7 mm and 3.4 mm, 5mm thick slices - 0.1mm inter-slice spacing, 30 slices per
acquisition - matrix 128 mm x128 mm FOV 21.7cm x 21.7cm,
in-plane sample spacing was 0.85 mm - TR7200, TE156
- b values (0, 500, 1000 mm2/s) or (0, 1000 mm2/s)
- 12 non-collinear directions,
- The first three datasets were interleaved and
zero-filled in the slice direction to form a
fourth dataset with resulting inter-slice
distance of 0.85 mm. - FA and Trace maps derived.
20Additional MRI Acquisitions
- 3D T1 volume (MPRAGE) for volumetric analysis
- 3 interleaved FLAIR acquisitions concatenated
into a pseudo 3D volume for assessment of SH
volume - Voxel dimensions on MPRAGE pseudo FLAIR match
DTI.
21Image Analysis
- Describing DTI parameters in NAWM, SH, and in
anterior and posterior white matter. - Analyze AVW 5.0, 6.0 (Mayo Clinic)
- ROI 5 x 5 square voxels
- Periventricular white matter
- Centrum semiovale
- ROIs were placed on T2-weighted images (b0)
images and transferred to FA and Trace maps for
measurement - Recorded location as NAWM vs. SH anterior vs.
posterior
22DTI in SIVD ROI Placement
23Image Analysis
- Parenchymal volume estimation
- Performed on MPRAGE sequences
- Voxel estimation tool in Analyze following skull
stripping. - SH volume
- Performed on pseudo-3D FLAIR images
- SH thresholding following skull stripping
w/operator correction - Sum of all voxels with intensity levels within SH
threshold range
24Cognitive Tests
- DRS I/P
- SDMT
- TMT A B
- COWAT (FAS)
25Results Demographics
Variable NC (n8) SIVD (n9) MCI (n9) Overall p
Age _at_ scan (yrs) 68.014.8 58.610.7a 76.78.4a .011
Education (yrs) 12.63.4 14.03.3 13.72.9 ns
MMSE 29.01.6 28.61.2 27.31.5 ns
Female 50.0 66.7 55.6 ns
- SIVD group younger than MCI
- All subsequent group analyses covaried for age _at_
scan
26Results Parenchymal SH volumes
Variable NC (n8) SIVD (n9) MCI (n9) Overall p
Parenchymal (cm3) 1099.996.28 1184.2 191.1 1103.7153.8 ns
SH/parench .002.002a .033.025a,b .009.009b .003
- No significant differences across groups on
estimated parenchymal volume. - SIVD had higher ratio of SH to parenchymal
volume than NC or MCI
27Results SH FA Trace
(mm2/s x 10-3)
- No group differences in SH or Trace in regions
of SH
28Results NAWM FA Trace
p.126
(mm2/s x 10-3)
- SIVD had lower FA vs. NC and higher trace vs. MCI
29Results NAWM Anterior/Posterior
30Results DTI Cognition
- SH
- SDMT with SH/parenchymal ratio (r .45, p .02)
- SDMT with FA in SH (r -.61, p .01)
- NAWM
- SDMT with NAWM FA (r -.42, p lt .04)
- SDMT with anterior NAWM FA (r -.46, p lt .02)
- SDMT with NAWM Trace (r .40, p lt .05)
- No other significant correlations between
tests/DTI variables
31Conclusions
- Consistent w/previous results showing DTI changes
in NAWM in SIVD. - NC and MCI were similar on DTI
- SIVD may alter the age-related gradient of
anterior to posterior DTI changes. - Processing speed associated with DTI parameters
in both NAWM and SH. - DTI may provide method for describing
differential effect of disorders on white matter
and detect associations between NAWM and
cognitive function.
32Limitations
- Small n
- SIVD group younger than MCI
- SIVD group radiographically characterized.
- Limited range of cognitive deficits
- Correlation analyses exploratory
- ROI analysis not capture DTI differences in other
regions
33Future Directions
- Additional data collection underway
- Assess differential impact of CADASIL vs. SIVD
- Differential contribution of SH volume vs. DTI in
these groups. - Assess DTI correlation with experimental working
memory measures.
34Tractography
Superior view color fiber maps
Lateral view color fiber maps
Zhang Laidlaw http//csdl.computer.org/comp/pro
ceedings/vis/2004/8788/00/87880028p.pdf.
35Acknowledgments
- Stephen Salloway
- Paul Malloy
- David Laidlaw
- Song Zhang
- Thea Brennan-Krohn
- Erin Schlicting
- Jerome Sanes
- Lynn Fanella
36Support
- NIA AG020498-02
- Alzheimers Association NIRG-03-6195
- Start-MH Grant
- NIMH K08MH01487W
- The Human Brain Project (NIBIB NIMH)
- Ittleson Fund at Brown
- P20 NCRR15578-01
- Center for Translational Brain Research at Brown.
37THANK YOU
38Results DTI Cognition
- Whole Sample (partial correlations controlled for
age) - SDMT with FA in SH (r .54, p .03)
- Bivariate group analyses
- SIVD
- DRS I/P with FA in NAWM FA (r .706, p lt .03)
- DRS I/P with Trace in ant. NAWM (r -.685, p lt
.04) - MCI
- TMT-B with FA in ant. NAWM FA (r .-681, p lt
.04) - COWAT with SH/parench ratio (r -.792, p lt .01)
39Results DTI Cognition
- NC
- SDMT with FA in NAWM FA (r .-760, p lt .03)
- SDMT with FA in ant. NAWM FA (r .-750, p lt .03)
40Research Focus
- Frontal Systems Disruption
- ?
- Changes in Executive Cognition and Behavior
- ?
- Functional Disability/Conversion to Dementia
41Results FA Trace in Genu Splenium
(mm2/s x 10-3)
- No group differences in SH or Trace in regions
of SH
42Results FA Trace in Temporal Lobe White Matter
(mm2/s x 10-3)
- FA Lower in SIVD and MCI groups vs. NC
bilaterally. - Trace Higher in SIVD vs. NC on right and
higher in SIVD than both NC and MCI on left.
43Image Analysis
44Image Analysis
- Temporal lobe white matter rectangular ROI (6 x
3) in left right temporal stem - 10 consecutive coronal slices starting at the
mamillary bodies and proceeding posteriorly. - Corpus callosum square (3 x 3) ROI in left
right genu and splenium on 5 consecutive slices. - Placed directly on FA or Trace maps
- No classification of SH vs. NAWM
45Image Analysis
- Analyze AVW 5.0, 6.0 (Mayo Clinic)
- Periventricular white matter 3 ROIs (5 x 5)
around each horn, 2 axial slices. - Centrum semiovale Up to 5 ROIs (5 x 5) in each
hemisphere in NAWM and SH, 2 axial slices - ROIs were placed on b0 images and transferred to
FA and Trace maps for measurement - FLAIR and MPRAGE used for guidance.
- Recorded location as NAWM vs. SH anterior vs.
posterior
46Results DTI Cognition
- Age _at_ scan
- SDMT (r -.45, plt.05)
- TMT-B (r .58, plt.05)
- All HVLT-R (r .41 - .69, plt.05)
- Education none
- Parenchymal volume
- SDMT (r -.42, plt.05)
- TMT-B (r -.53, plt.05)
- DRS-Memory (r .43, plt.05)
- SH ratio None
47Results NAWM
Variable Variable Variable Controls (n8) SIVD (n9) MCI (n9) Overall p
FA FA FA FA
NAWM NAWM .438.045e .438.045e .408.035e .409.030 .043
Trace (mm2/s x 10-3) Trace (mm2/s x 10-3) Trace (mm2/s x 10-3) Trace (mm2/s x 10-3)
NAWM .761.045c .761.045c .761.045c .824.052c,d .797.047d .013
- SIVD had lower NAWM FA than NC
- SIVD had higher NAWM Trace than either NC or MCI
- No differences between NC and MCI in NAWM FA or
Trace - No group differences in FA or Trace in regions
of SH (not shown)
48Results Corpus Callosum
Variable Controls (n8) SIVD (n9) MCI (n9) Overall p
FA
Genu .813.041 .720.170 .769.067 ns
Splenium .811.085 .807.097 .815.082 ns
Trace (mm2/s x 10-3) Trace (mm2/s x 10-3)
Genu .730.043 .893.259 .935.517 ns
Splenium .675.046 .747.053 .692.076 ns
- No group differences in FA or Trace in genu or
splenium.
49Results Temporal Lobe White Matter
Variable Controls (n8) SIVD (n9) MCI (n9) Overall p
FA
TLWM R .574.052a,b .502.053a .527.039b .027
TLWM L .551.056c,d .445.051c .487.039d .001
Trace (mm2/s x 10-3) Trace (mm2/s x 10-3)
TLWM R .768.061e .855.062e .796.050 .034
TLWM L .785.070f .920.076f,g .823.070g .001
- FA Lower in SIVD and MCI groups vs. NC
bilaterally. - Trace Higher in SIVD vs. NC on right and
higher in SIVD than both NC and MCI on left.
50Image Analysis
Skull stripping and parenchymal volume estimation
51Results Attention/Executive
Variable Controls (n8) SIVD (n9) MCI (n9) Overall p
SDMT 44.009.04 45.0020.52 32.3313.90 ns
TMT-A 29.257.13a 40.3815.65 53.5616.70a .015
TMT-B 88.3849.50 101.1345.02 180.6778.99 .057
COWA 40.2514.14 35.5012.62 31.4411.22 ns
DRS I/P 35.002.88 35.882.80 33.332.96 ns
- TMT-A NC better than MCI trend (p .058) for
NC better than SIVD - TMT-B NC better than MCI
52Results Attention/Executive
p.126
- SIVD intermediate on all measures except DRS I/P
- TMT-A NC better than SIVD and MCI MCI and SIVD
not different. - TMT-B NC better than MCI no other pair-wise
differences
(mm2/s x 10-3)
53Results Memory
p.126
- SIVD intermediate on all measures
- MCI significantly worse than NC on all measures
- MCI significantly worse than SIVD on all
measures except HVLT-R Total Learning
(mm2/s x 10-3)
54Results Attention/Executive (T scores)
Variable Controls (n8) SIVD (n9) MCI (n9) Overall p
SDMT 47.007.75 51.8813.49 54.444.00 ns
TMT-A 42.508.33a,b 58.0013.81a 63.2216.35b .013
TMT-B 46.7511.89c 58.2515.12 72.119.68c .013
COWA 46.3715.41 52.7514.63 56.1114.63 ns
DRS I/P 51.639.53 48.759.45 57.339.84 ns
- TMT-A NC better than SIVD and MCI MCI and
SIVD not different. - TMT-B NC better than MCI no other pair-wise
differences
55Results Memory
Variable Controls (n8) SIVD (n9) MCI (n9) Overall p
HVLT-R
Total 24.884.64a 24.633.25 14.444.50a .003
Delay 9.502.07b 7.503.12b 2.332.18b .031
retn 97.3817.82 74.1324.36 34.4430.54 ns
Discrm 10.381.69 9.882.64 6.222.19 ns
DRS-Mem 24.001.31 23.501.31 21.892.80 ns
- HVLT-R Total Learning MCI lower than NC or
SIVD - HVLT-R Delayed Recall Significant differences
between all pairs
56Results Memory (T scores)
Variable Controls (n8) SIVD (n9) MCI (n9) Overall p
HVLT-R
Total 53.008.26a 58.117.37 74.2211.95a lt.001
Delay 49.758.83b 62.2212.87b 78.2211.22b lt.001
retn 45.889.20c 61.5613.27c 80.0019.22c lt.001
Discrm 53.7515.09d 58.5617.02e 76.8916.57d,e .017
DRS-Mem 46.756.11f 49.115.71g 61.5613.22f,g .005
- HVLT-R Total Learning NC better than MCI
- HVLT-R Delayed Recall Significant differences
between all pairs - HVLT-R retained Significant differences
between all pairs - HVLT-R Discrim NC better than MCI, SIVD better
than MCI - DRS-Memory NC and SIVD both better than MCI
57Results Overall FA Trace
(mm2/s x 10-3)