Diffusion Tensor Imaging of the Superior
Longitudinal Fasciculus and Working Memory
in Recent-Onset Schizophrenia
Katherine H. Karlsgodt, Theo G.M. van Erp, Russell A. Poldrack, Carrie E. Bearden, Keith H. Nuechterlein,
and Tyrone D. Cannon
deficits in patients with schizophrenia. This study examines whether recent-onset schizophrenia is associated with anatomical changes in
to WM performance.
Methods: We applied a novel registration approach (Tract-Based Spatial Statistics [TBSS]) to diffusion tensor imaging data to examine
fractional anisotropy (FA) in the left and right SLF in 12 young adult patients with recent-onset schizophrenia and 17 matched control
Fractional anisotropy values were correlated with performance on a verbal WM task in both patient and control groups in the left but not
Conclusions: Recent-onset schizophrenia patients show deficits in frontal-parietal connections, key components of WM circuitry. More-
over, the integrity of this physiological connection predicted performance on a verbal WM task, indicating that this structural change may
white matter changes detectable in the early phases of the illness as one source of this dysfunction.
Key Words: Diffusion tensor imaging, DTI, frontal-parietal net-
works, schizophrenia, white matter, working memory
regions has been shown to be interrelated (3), indicating that
these regions are not merely coactivated but function as a
network. This coordination is likely supported by an axon
bundle known as the superior longititudinal fasciculus (SLF),
which in part connects the middle frontal gyrus/dorsolateral
prefrontal cortex (Brodmann areas [BA] 9/46) with the supramar-
ginal gyrus (BA 40) (4). Moreover, maturation of the white matter
between these regions correlates with WM capacity and func-
tional activation (5), effectively linking WM performance to
integrity of the entire circuit.
Patients with schizophrenia show particular impairments in
WM (6) and WM deficits may be a core feature of the illness, as
they influence cognitive processing (7), symptomatology (8), and
functional outcomes (9). Further, these cognitive deficits are not
widely responsive to neuroleptics. While atypical antipsychotics
are more effective in this domain than traditional antipsychotics,
impairments may persist even after treatment (10). Therefore,
gaining a full understanding of the circuitry underlying this
cognitive process is imperative.
asic animal and human imaging research have implicated
the frontal and posterior parietal cortices in working
memory (WM) (1,2). Further, cellular activity in these
Structural studies have shown frontal and parietal gray matter
changes in schizophrenia (11,12). If the complex neurocognitive
functions impaired in schizophrenia depend on regions known
to be abnormal, it seems likely that physiological connections
linking these networked regions may also be compromised.
However, the extent to which connections between cortical
regions associated with WM are altered has yet to be determined.
Investigations using traditional magnetic resonance imaging
(MRI) have shown reductions in whole brain (13) and frontal
white matter (e.g., 14) in schizophrenia. However, investigation
of specific tracts is beyond the scope of these methods due to the
necessary reduction of intersecting, multidirectional fiber tracts
into a single volumetric measurement. The advent of diffusion
tensor imaging (DTI) has allowed the assessment of the integrity
of specific white matter structures in vivo. Diffusion tensor
imaging is based on measurement of the Brownian motion of
water molecules using MRI (15). In an unobstructed medium, a
molecule has an equal likelihood of diffusing in any direction,
and the distribution of possible movements is a sphere; this is
known as isotropic diffusion. In white matter, diffusion is
restricted by axonal structure and occurs more readily parallel to
the axon than perpendicular to it. This results in a distribution of
potential movements better represented by an ellipsoid than a
sphere, which is known as anisotropic diffusion and is the basis
for DTI. An index known as fractional anisotropy (FA) charac-
terizes the eccentricity of the ellipse and is interpreted as a
measure of white matter integrity.
Previous DTI studies in schizophrenia have shown decreased
FA in frontal (16–23) and parietal lobes (22,24). However, the
subregions affected vary between studies, with some studies
finding no patient-control differences (25,26), limiting our ability
to draw conclusions from the existing body of DTI work (27).
One reason for discrepancies may be the difficulty in anatomi-
cally registering DTI data (28). Particularly in groups where tract
From the Department of Psychology (KHK, TGMV, RAP, KHN, TDC), Brain
Research Institute (RAP), Department of Psychiatry and Biobehavioral
Science (RAP, CEB, KHN, TDC), and Department of Human Genetics
(TDC), University of California, Los Angeles, California.
Address reprint requests to Katherine H. Karlsgodt, Ph.D., Department of
Received March 20, 2007; revised June 12, 2007; accepted June 14, 2007.
BIOL PSYCHIATRY 2008;63:512–518
© 2008 Society of Biological Psychiatry
shape or volume is likely to differ, rigorous registration is
imperative for voxel-based comparisons. The Tract Based Spatial
Statistics (TBSS, a part of FMRIB Software Library [FSL], Oxford,
United Kingdom) (28) registration software is designed specifi-
cally for such data. Tract Based Spatial Statistics uses nonlinear
registration to create an FA “skeleton” based on the center of all
of the tracts common to the entire group and projects the data
from the center of each subject’s tracts onto the skeleton for
group comparison. This method ensures that statistics are only
applied in regions where data exist for all subjects and maxi-
mizes the likelihood that the pooled data originate from the
center of a tract in every subject.
Our primary goal was to assess FA differences in the SLF
between patients with schizophrenia and matched control sub-
jects. Given that patients with schizophrenia show both gray
matter deficits in the frontal-parietal regions that support WM and
impaired WM performance, we hypothesized that the structural
integrity of the white matter tracts connecting these regions
would be reduced. Additionally, we conducted a secondary
analysis of the relationship between verbal WM performance and
FA in the SLF, hypothesizing that the structural integrity would
relate to performance on a verbal WM task. Given the special
importance of left hemisphere frontal and parietal regions for
verbal WM, we predicted that this relationship should be partic-
ularly evident on the left. To evaluate whether these changes are
present early in the course of illness rather than the result of
long-term disease process or treatment effects, we assessed
young adults with a recent onset of schizophrenia.
Methods and Materials
Twelve volunteers with schizophrenia and 17 healthy control
volunteers participated (see Table 1); the groups were matched
on age, gender, and years of education. All participants under-
went a verbal and written informed consent process, and partic-
ipants under the age of 18 years provided written assent, while
their parent or guardian completed written consent, as approved
by the University of California, Los Angeles (UCLA) Institutional
Review Board (IRB). Patients with schizophrenia were recruited
from the Aftercare Research Program and Adolescent Brain-
Behavior Research Clinic (ABBRC) at UCLA. Inclusion criteria for
participants with schizophrenia were a recent onset of psychotic
illness, with the beginning of the first major psychotic episode
(characterized by psychotic symptoms lasting at least 2 weeks)
occurring within the last 2 years, and a diagnosis by DSM-IV
criteria (29) for schizophrenia or schizoaffective disorder, de-
pressive subtype. Patients admitted to the research programs are
between 12 and 17 (ABBRC) or 18 and 45 years of age (After-
care); however, only participants aged 17 to 27 years old were
used in this analysis to constitute a young adult sample. Control
subjects were recruited from a community sample via advertise-
ment in newspapers and fliers. Inclusion criteria for control
subjects were age matching that of the participants with schizo-
phrenia and no evidence of any major mental disorder as
determined by the Structured Clinical Interview for DSM-IV (29),
including substance abuse and dependence. Both patients and
control subjects were excluded for known neurological disorders
(e.g., epilepsy, encephalitis), mental retardation (i.e., premorbid
intelligence quotient [IQ] not less than 70), and/or insufficient
fluency in the English language. All patients were currently being
treated with atypical antipsychotics (seven with risperidone,
three with aripiprazole, one with quetiapine, one with olanzap-
ine). Three patients were additionally being treated with temaz-
epam, trazodone, and venlafaxine, respectively.
Subjects were scanned on a 1.5T Siemens Sonata scanner
(Siemens, Erlagen, Germany) at the Ahmanson-Lovelace Brain
Mapping Center at UCLA. Head motion was restricted using foam
padding. Diffusion tensor imaging data were acquired using a
six-direction sequence with 75 contiguous 2 mm anterior com-
missure-posterior commissure (AC-PC) aligned interleaved slices
with no gap (repetition time [TR] ? 9.5 sec, echo time [TE] ? 77
msec, flip angle ? 90 degrees, matrix ? 128 ? 96, b-value ?
1000 sec/mm2, field of view [FOV] ? 256 ? 192 mm). Five
repetitions were acquired for averaging, and each scan lasted 1
minute 16 seconds for a total scan time of 6 minutes 20 seconds.
To assess verbal WM, we used a modified Sternberg item
recognition task (30). Briefly, a target set of yellow uppercase
consonants was displayed for 2 seconds, followed by a 3 second
fixation cross. Verbal WM load was manipulated by increasing
the number of consonants; loads of 3, 5, 7, and 9 target letters
were used. A green uppercase probe then appeared for 2
seconds, followed by 2 seconds of fixation before the start of the
next trial. Subjects indicated whether the probe letter matched
the letters from the target set. The task included 12 trials per load,
for a total of 48 trials. The use of multiple loads allowed us to
maximize the range of performance across subjects. To obtain a
single index of performance, we averaged percent correct across
all trials for each subject. Verbal WM behavioral data were
acquired on the same day as the structural MRI scan.
Image Processing. Images were checked for artifacts using
an automated signal-to-noise detection program that selected
scans which contained excessive levels of noise or other
artifacts. Subjects with less than three useable scans in any of
the directions were excluded from the analysis. First, the five
image acquisitions for each direction were merged into a
single four-dimensional file and then aligned to the middle
volume (representing the third acquisition) with McFlirt (FSL,
Table 1. Subject Demographics
Years of Educatione
Duration of Illnessf
Age of Onsetf
20.91 ? 3.52
20.58 ? 1.97
13.04 ? 2.77
15.00 months ? 9.94
19.54 years ? 3.55
13.61 ? 1.16
L, left; R, right.
aAge did not differ between groups: t(27) ? .321, p ? .751.
bGender did not differ between groups: ?2(1,df ? 28) ? .083, p ? .774.
cRace did not differ between groups: ?2(1, df ? 28) ? .967, p ? .915.
dHandedness did not differ between groups: ?2(1, df ? 28) ? .731, p ?
eYears of education did not differ between groups: t(27) ? ?.771, p ?
fTime of onset was unavailable for one patient.
K.H. Karlsgodt et al.
BIOL PSYCHIATRY 2008;63:512–518 513
Oxford, United Kingdom) (31) using a six-parameter registra-
tion to the mean volume. The aligned files were averaged to
create one file for each of the six directions and for the b0
image. Each of the direction files was then registered to the b0
image using a 12-parameter affine registration with a mutual
information cost function implemented in Flirt (FSL), which
served to correct for distortion due to eddy currents. The
registered images (b0 and the six direction files) were skull-
stripped using the FSL Brain Extraction Tool (BET) and the
skull-stripped files were registered to Montreal Neurological
Institute (MNI)-152 standard space again using a 12-parameter
affine registration with a mutual information cost function
implemented in Flirt (FSL). Fractional anisotropy images were
calculated using DTIFit (FMRIB Software Library’s Diffusion
Toolbox), which fits a diffusion tensor model at each voxel.
After calculation of the FA for each subject, to create a
group map we implemented a voxel-wise statistical analysis of
the FA data using TBSS (28). All subjects’ FA data were aligned
to an MNI-152 standard space FA target using the nonlinear
registration implemented by Image Registration Toolkit (IRTK,
Imperial College London; www.doc.ic.ac.uk/?dr/software). A
mean FA image was created from all the subjects in this common
space and narrowed to generate a mean FA skeleton that
represented the center (defined as the locally highest FA value)
of all tracts common to the entire group. Individual FA data were
then projected onto this skeleton by searching the area around
the skeleton in each subject’s aligned FA map in the direction
perpendicular to each tract, finding the highest local FA, and
assigning this value to the skeleton. This ensured that each
subject’s skeleton was in the group space, yet represented the
center of that subject’s own unique white matter bundles.
Regions of interest in the left and right SLF were defined on an
overlay of the mean FA skeleton on the mean FA map (Figure 1)
and verified using a DTI color map atlas (32). Each subject’s FA
skeleton was masked using the region of interest, and the
average FA was calculated for the entire region.
We assessed group differences in FA using a repeated mea-
sures analysis of variance (ANOVA) (SPSS 13, Chicago, Illinois)
with diagnosis as the between-subjects factor and FA in the right
and left SLF as the within-subjects repeated measure. This result
was subsequently decomposed using t tests of overall FA aver-
aged across each tract (33). Additionally, to enable the visualiza-
tion of particular areas within the region of interest (ROI) that
differed between groups, a voxel-wise cross-subjects analysis of
overall group differences in FA was performed using nonpara-
metric permutation testing with the FSL Randomise tool (http://
www.fmrib.ox.ac.uk/fsl/randomise/index.html). Randomise im-
plemented 5000 randomly generated permutations of the data to
perform a Monte Carlo style permutation test. Nonparametric
statistics were used because of the substantial nonnormality of
the FA measure (34); although we tested for normality across the
entire ROI, we cannot be certain that normality was maintained
at the individual voxel level. To probe differences between
groups, we examined both control-patient and patient-control
contrasts. Age was entered into this analysis as a confound
regressor to ensure that any observed effects were independent
of age-related changes. Therefore, age (after first being de-
meaned) was regressed out of the data before the permutation
tests were implemented. A 6 mm variance-smoothing kernel was
used to improve the estimation of variance being fed into the
statistical analysis. Resulting data were thresholded using a
cluster-based thresholding method (cluster threshold set at 2)
that corrected for multiple comparisons by using the null distri-
bution of the maximum cluster size (across the SLF ROIs) at p ?
To assess the relationship of FA with behavior, a robust
regression of FA predicting WM performance was performed for
each ROI (left and right SLF) in each group. Prior to the
regression, all variables (the behavioral data, left SLF FA, and
right SLF FA) were tested for normality (skew and kurtosis). Both
performance and left SLF FA were normal both across the entire
group and within both patient and control subgroups. Fractional
anisotropy in the right SLF was normally distributed across the
entire group and in the patient group; however, the control
group contained one outlier. The regression was run both with
and without this subject, and as the results remained the same,
we retained this subject in the data set. To directly evaluate the
difference between the relationship of WM and FA between
groups, a robust regression was performed to investigate the
interaction of FA and group membership, using verbal WM
percent correct as the dependent variable, and a Wald test was
used to test for significance. To account for any potential
developmental effects on FA, age was included as a factor in the
Patient and control groups did not significantly differ in mean
age [t(27) ? .321, p ? .751] or gender distribution [?2(1,df ? 28) ?
.083, p ? .774; Table 1]. On the VWM task, patients’ average
performance across loads did not differ significantly from that of
control subjects [mean percent correct ? standard deviation:
81.42% ? 12.25 vs. 81.37% ? 4.68, respectively; t(27) ? .0157,
p ? .987].
Fractional anisotropy was collapsed across the entire tract to
determine whether overall FA differed between groups. The
repeated measures ANOVA showed a significant effect of diag-
nosis [F(1,27) ? 6.168, p ? .020], a trend toward an effect of
hemisphere (left or right SLF) [F(1,27) ? 3.363, p ? .078], but no
group by hemisphere interaction [F(1,27) ? .044, p ? .706]. The
post hoc t tests indicated that FA was significantly higher in
control subjects on the left [t(27) ? ?2.37, p ? .025, two-tailed]
and on the right SLF [t(27) ? ?2.106, p ? .045, two-tailed; Figure
2]. The t tests were run with and without the single outlier;
without this subject, the effect in the right is trend-level [t(26) ?
?1.93, p ? .065, two-tailed]. There were no significant differ-
ences between patients being treated only with atypical antipsy-
Figure 1. Region of interest selection: Displayed on the mean group FA
map, the FA skeleton is in yellow and the regions of interest are in red. FA,
514 BIOL PSYCHIATRY 2008;63:512–518
K.H. Karlsgodt et al.
chotic medications and those treated with a combination of
atypical antipsychotics and additional adjunct medications.
The voxel-wise DTI analysis performed within the right and
left SLF regions of interest using cluster thresholding showed
regions within the tract in which control subjects had signifi-
cantly greater FA than patients but not the reverse (Figure 3). The
significant region was primarily located in the posterior region of
In the secondary analysis of VWM performance and FA, a
robust regression of FA predicting behavioral performance was
performed for each group, covaried for age. In both patients with
schizophrenia [F(2,9) ? 5.52 p ? .027] and control subjects
[F(2,14) ? 4.22, p ? .037], total left SLF FA (across the entire ROI)
predicted VWM performance. This association was not observed
between total right SLF FA and VWM performance in either group
[patients: F(2,9) ? .14, p ? .871, control subjects: F(2,15) ? .151,
p ? .256]. In the direct comparison of the FA-VWM relationship
between groups, on the left there was a significant effect of the
overall model [F(4,24) ? 4.61, p ? .0066] with left SLF predicting
WM performance (p ? .007). There was a significant difference
between the slope of FA and VWM between the groups (p ?
.038) as determined by a Wald test, such that patients showed a
stronger correlation than control subjects. However, overall there
was no significant main effect on the right [F(4,24) ? .51, p ?
.728) or of FA ? VWM interaction between groups (p ? .532)
(Figure 4). There was no significant effect of age on either the left
(p ? .894) or right (p ? .282) SLF.
Applying a novel registration approach (TBSS) to DTI data,
we found that, in addition to previously documented frontal-
parietal gray matter deficits, there is a decrease in the integrity of
white matter connecting these regions in patients with recent-
onset schizophrenia. This indicates that the entire circuit subserv-
ing WM, not merely its cortical subcomponents, exhibits struc-
tural deficits in schizophrenia. These results suggest that such
changes are not likely to be secondary to disease process or the
long-term effects of medication but rather are present from the
very early stages of the illness. In addition, the degree to which
the integrity of this physiological connection is impaired relates
to how well subjects are able to perform a verbal WM task,
indicating that this structural change may have functional impli-
cations, and that white matter microstructure may contribute to
verbal WM ability by determining the strength of communication
between the cortical areas involved.
Our finding that changes exist in frontal-parietal connections
even in the first episode is consistent with structural imaging
findings of gray matter volume deficits at illness onset (14,36).
Whitford et al. (12) found gray matter reductions in first-episode
patients that included both these regions. Moreno et al. (35)
Figure 2. FA differences between groups across the entire SLF ROI. FA,
fractional anisotropy; ROI, region of interest; SLF, superior longititudinal
Figure 3. Voxel-wise analysis of
group differences in the left and
right SLF (yellow ? ROI area, red ?
patients than control subjects;
trol subjects were lower than pa-
tients). FA, fractional anisotropy;
K.H. Karlsgodt et al.
BIOL PSYCHIATRY 2008;63:512–518 515
found a decrease in frontal gray matter in first-episode adoles-
cent male patients, and Narr et al. (11) found decreased gray
matter density in frontal, temporal, and parietal regions. If
cortical changes exist at the beginning of the illness, it is not
unexpected that white matter connections would be affected as
Five previous studies have investigated changes in FA in
first-episode or recent-onset patients with schizophrenia. How-
ever, like the larger body of DTI work in schizophrenia, these
findings have not been entirely conclusive. Hao et al. (18) found
decreased FA throughout the entire brain using a voxel-wise
analysis. It is possible that the differences were nonlocalized due
to systematic registration differences between groups, leading to
the comparison of imperfectly aligned white matter tracts
throughout the brain. Szeszko et al. (23) found a decrease in FA
in the left middle frontal gyrus, left posterior temporal gyrus, and
left internal capsule. Federspiel et al. (17) found a decrease in
intervoxel coherence (IC), an index of connectivity, in a number
of regions, including the anterior SLF. However, others have
found no effect in the corpus callosum (26) and hippocampus
(36) in first-episode patients. As methods become more cohesive
in this newly emerging research area, it will become possible to
determine whether the different findings were due to variation in
the effects of schizophrenia on different brain regions, discrep-
ancies in patient groups or clinical variables, or differences in
imaging and analysis methodologies.
Only a few studies have examined the functional signifi-
cance of DTI findings in schizophrenia. Existing studies have
shown that FA is associated with executive functions in the
cingulate (37) and with episodic memory (37), executive task
performance (38), and verbal learning (39) in the uncinate
fasciculus. These studies provide evidence that changes in white
matter that can be assessed with DTI are behaviorally and
clinically meaningful. However, despite widespread interest in
WM in schizophrenia, this is the first such study to assess the WM
circuitry. Our study extends the literature by showing that FA
changes in the superior longitudinal fasciculus, the major white
matter connection between prefrontal and parietal cortices,
relate to verbal WM performance. Most existing studies have
shown greater correlations in patient groups than control groups.
Possibly, when the white matter is compromised, as the majority
of studies have found in schizophrenia, further reductions in its
integrity can become more of a limiting factor, resulting in larger
effects on function in patients. Alternatively, if cortical regions
that the white matter tracts connect are themselves abnormal, as
the structural imaging literature would indicate, there may be an
additive effect of the cortical and white matter dysfunction that
leaves the patients’ performance particularly vulnerable to
smaller changes. However, it is important to consider that patient
groups tend to have higher variance than control groups, which
can limit our ability to interpret differences in correlations
between groups. However, our finding is bolstered by the
observed pattern in which the relationship was stronger on the
left than the right, replicating the neuropsychological and imag-
ing literature indicating that verbal tasks are often left lateralized.
The strength of this analysis is that it focuses on a well-
established system, in which there has been substantial research
on the cortical regions involved in both healthy control subjects
and patients with schizophrenia. Our findings show that the
patient-control differences observed in functional imaging stud-
ies are not limited to the frontal and parietal cortex but also
extend to structural connections between these regions. There-
fore, while cellular signaling differences or neurotransmitter
changes in these regions may play a role in WM deficits, the
disruption is better conceptualized as a circuit-wide deficit at not
only a functional but also a structural level. It is notoriously
difficult to improve cognitive deficits in schizophrenia with
traditional antipsychotic medications (40), which work primarily
by effecting changes at the level of neural functioning, e.g.,
dopamine signaling. However, while there is the theoretical
potential for plasticity and improvement of structural deficits,
decreases in white matter organization or in factors influencing
myelination are not currently known to improve with existing
antipsychotic treatment. Thus, adding this new dimension to our
understanding of factors contributing to WM dysfunction may
explain why some aspects of cognitive deficits are so treatment
refractory. This also indicates that future treatments that act on
molecular pathways controlling myelination and oligodendrite
function may be of particular interest with regard to treatment of
negative symptoms and cognitive impairments.
Our study was limited by a relatively small sample size. In
addition, although patients had a recent onset of psychosis with
relatively short medication histories, they were not medication-
free. Future studies in unmedicated subjects or in subjects
ascertained just prior to disease onset may help address this
issue. Further, while the TBSS approach has significant benefits
over traditional voxel-based morphometry approaches, a few
ropy; SLF, superior longititudinal fasciculus; WM, working memory.
516 BIOL PSYCHIATRY 2008;63:512–518
K.H. Karlsgodt et al.
factors limit the interpretation. First, the TBSS skeleton map is
based on FA values only, not directional information from the
diffusion tensors. Therefore, while we have created regions of
interest in the skeleton subregions in which specific fiber tracts
are predicted to be located, future tractography studies will help
ensure precisely which fiber bundles are included in these
regions. Secondly, by using FA to define the regions of the
skeleton and then assessing FA within that region, analyses may
be biased toward regions of white matter with the highest FA
values. However, while this means that changes in peripheral
regions of the bundles (that have a low FA signal) are less likely
to be assessed, this is outweighed by the benefit of analyzing
only regions in which we are confident that all subjects have
data. Our study was further limited because there was no
significant difference in performance between patients and con-
trol groups as would have been expected. One explanation for
this is that there is substantial variance within each group that
may obscure such effects. However, the regression analysis
indicates that the variance is not simply noise but can be partially
explained by individual differences in performance.
In summary, this study provides novel information regarding
reduced integrity of the primary fiber bundle connecting frontal
and parietal brain regions in recent onset of schizophrenia and its
relationship to WM performance. Prior work has been limited by
the difficulty of reliably separating out the individual tracts in the
more dorsal regions of the hemispheres using existing MRI and
DTI registration and analysis tools. The recently developed
methods used here have allowed us to be more confident in our
region of interest selection and in our ability to register and
compare regions between groups. Our methods are an improve-
ment over some of the previous approaches. Whole brain
analyses with comparisons at multiple points increase the risk of
type 1 error; however, we focused our analyses on a single
region and a single neurocognitive variable, both of which are
consistent with a priori hypotheses about patterns of brain
function in schizophrenia. Further, by combining behavior and
DTI, we can see that in patients with schizophrenia, performance
on a task known to involve a frontal-parietal network is related
not just to the integrity of the cortical regions but also of their
interconnections. The finding that verbal WM is directly related
to structural changes that are not targeted by neuroleptics may
indicate why cognitive deficits are difficult to treat and influence
how we conceptualize the complex relationship of brain struc-
ture, brain function, and treatment of cognitive deficits in schizo-
phrenia. Finally, the presence of these changes, even in the very
early stages of the disorder, furthers our understanding of the
basis of WM deficits in schizophrenia.
This research was supported by National Institutes of Health
(NIH) Grants MH65079, MH066286, GM072978, and RR021992
to TDC, 5-F31-MH068111-02 to KHK, National Institute of Mental
Health (NIMH) P50 MH066286 to KHN, and a gift to the UCLA
Foundation by Garen and Shari Staglin.
We acknowledge the help of Molly Hardt, Lara Zimmerman,
Tara Niendam, Liset Cristiano, Sabrina Lux, Malin McKinley,
and Jeff Alger, as well as the participants.
All authors have reported no known biomedical financial
interests or other potential conflicts of interest. All grant and
other financial support as well as technical and material support
has been listed.
1. Chafee MV, Goldman-Rakic PS (2000): Inactivation of parietal and pre-
frontal cortex reveals interdependence of neural activity during mem-
ory-guided saccades. J Neurophysiol 83:1550–1566.
2. Paulesu E, Frith CD, Frackowiak RS (1993): The neural correlates of the
verbal component of WM. Nature 362:342–345.
3. Chafee MV, Goldman-Rakic PS (1998): Matching patterns of activity in
WM task. J Neurophysiol 79:2919–2940.
4. Petrides M, Pandya DN (2002): Association pathways of the prefrontal
cortex and functional observations. In: Stuss D, Knight RT, editors. Prin-
5. Klingberg T (2006): Development of a superior frontal-intraparietal
network for visuo-spatial working memory. Neuropsychologia 44:
psychological dysfunction in schizophrenia. Am J Psychiatry 160:1809–
9. Green MF (1996): What are the functional consequences of neurocog-
nitive deficits in schizophrenia? Am J Psychiatry 153:321–330.
10. Mattay VS, Fera F, Tessitore A, Hariri AR, Berman KF, Das S, et al. (2006):
Neurophysiological correlates of age-related changes in WM capacity.
Mapping cortical thickness and gray matter concentration in first epi-
sode schizophrenia. Cereb Cortex 15:708–719.
12. Whitford TJ, Grieve SM, Farrow TF, Gomes L, Brennan J, Harris AW, et al.
in first-episode schizophrenia: A tensor-based morphometry study.
13. Hulshoff Pol HE, Brans RG, van Haren NE, Schnack HG, Langen M, Baare
WF, et al. (2004): Gray and white matter volume abnormalities in
monozygotic and same-gender dizygotic twins discordant for schizo-
phrenia. Biol Psychiatry 55:126–130.
14. Breier A, Buchanan RW, Elkashef A, Munson RC, Kirkpatrick B, Gellad F
(1992): Brain morphology and schizophrenia. A magnetic resonance
15. Mori S, Zhang J (2006): Principles of diffusion tensor imaging and its
applications to basic neuroscience research. Neuron 51:527–539.
schizophrenia. Neuroreport 9:425–430.
17. Federspiel A, Begre S, Kiefer C, Schroth G, Strik WK, Dierks T (2006):
Alterations of white matter connectivity in first episode schizophrenia.
18. Hao Y, Liu Z, Jiang T, Gong G, Liu H, Tan L, et al. (2006): White matter
integrity of the whole brain is disrupted in first-episode schizophrenia.
Diffusion tensor analysis in chronic schizophrenia. A preliminary study
on a high-field (3.0T) system. Eur Arch Psychiatry Clin Neurosci 255:313–
(2004): Reduced frontal white matter integrity in early-onset schizo-
phrenia: A preliminary study. Biol Psychiatry 55:1138–1145.
21. Lim KO, Hedehus M, Moseley M, de Crespigny A, Sullivan EV, Pfeffer-
baum A (1999): Compromised white matter tract integrity in schizo-
phrenia inferred from diffusion tensor imaging. Arch Gen Psychiatry
(2003): Diffusion tensor magnetic resonance imaging of disruption of
regional white matter in schizophrenia. Neuropsychobiology 47:141–
(2005): White matter abnormalities in first-episode schizophrenia or
schizoaffective disorder: A diffusion tensor imaging study. Am J Psychi-
K.H. Karlsgodt et al.
BIOL PSYCHIATRY 2008;63:512–518 517
24. Ardekani BA, Nierenberg J, Hoptman MJ, Javitt DC, Lim KO (2003): MRI Download full-text
study of white matter diffusion anisotropy in schizophrenia. Neurore-
25. Foong J, Symms MR, Barker GJ, Maier M, Miller DH, Ron MA (2002):
Investigating regional white matter in schizophrenia using diffusion
tensor imaging. Neuroreport 13:333–336.
26. Price G, Bagary MS, Cercignani M, Altmann DR, Ron MA (2005): The
27. Kanaan RA, Kim JS, Kaufmann WE, Pearlson GD, Barker GJ, McGuire PK
(2005): Diffusion tensor imaging in schizophrenia. Biol Psychiatry 58:
28. Smith SM, Jenkinson M, Johansen-Berg H, Rueckert D, Nichols TE,
of multi-subject diffusion data. Neuroimage 31:1487–1505.
30. Sternberg S (1966): High-speed scanning in human memory. Science
analysis and implementation as FSL. Neuroimage 23(suppl 1):S208–
32. Mori S, Wakana S, Nagae-Poetscher LM, van Zijl PC (2005): MRI Atlas of
33. Stata (2003): Intercooled Stata 8.1 for Macintosh. College Station, TX:
34. Jones DK, Symms MR, Cercignani M, Howard RJ (2005): The effect of
filter size on VBM analyses of DT-MRI data. Neuroimage 26:546–554.
35. Moreno D, Burdalo M, Reig S, Parellada M, Zabala A, Desco M, et al.
(2005): Structural neuroimaging in adolescents with a first psychotic
36. Begre S, Federspiel A, Kiefer C, Schroth G, Dierks T, Strik WK (2003):
Reduced hippocampal anisotropy related to anteriorization of alpha
EEG in schizophrenia. Neuroreport 14:739–742.
37. Nestor PG, Kubicki M, Gurrera RJ, Niznikiewicz M, Frumin M, McCarley
sor imaging in schizophrenia. Neuropsychology 18:629–637.
38. Kubicki M, Westin C, Maier S, Frumin M, Nestor PG, Salisbury D, et al.
(2002): Uncinate fasciculus findings in schizophrenia: A magnetic
resonance diffusion tensor imaging study. Am J Psychiatry 159:813–
39. Nakamura M, McCarley RW, Kubicki M, Dickey CC, Niznikiewicz MA,
Voglmaier MM, et al. (2005): Fronto-temporal disconnectivity in schizo-
40. Hori H, Noguchi H, Hashimoto R, Nakabayashi T, Omori M, Takahashi S,
phrenia. Schizophr Res 86:138–146.
518 BIOL PSYCHIATRY 2008;63:512–518
K.H. Karlsgodt et al.