Abnormalities in connectivity of white-matter tracts
in patients with familial and non-familial
schizophrenia
Q. Wang
1
, W. Deng
1
, C. Huang
1
,M.Li
1
,X.Ma
1
, Y. Wang
1
, L. Jiang
1
, S. Lui
2
, X. Huang
2
, S. E. Chua
3
,
C. Cheung
3
, G. M. McAlonan
3
, P. C. Sham
3
, R. M. Murray
4
, D. A. Collier
5
, Q. Gong
2
* and T. Li
1
,
4
,
5
*
1
The Psychiatric Laboratory and Department of Psychiatry, State Key Laboratory of Biotherapy, West China Hospital, Sichuan University,
Chengdu, China
2
Huaxi MR Research Centre, Department of Radiology, West China Hospital, Sichuan University, Chengdu, China
3
Department of Psychiatry, The University of Hong Kong, Pokfulam, S.A.R. China
4
Department of Psychological Medicine and Psychiatry, Institute of Psychiatry, King’s College London, UK
5
MRC SGDP Centre, Institute of Psychiatry, King’s College London, UK
Background. Abnormalities in the connectivity of white-matter (WM) tracts in schizophrenia are supported by
evidence from post-mortem investigations, functional and structural magnetic resonance imaging (MRI) and diffusion
tensor imaging (DTI). The aims of this study were to explore the microstructural changes in first-episode
schizophrenia in a Han Chinese population and to investigate whether a family history of psychiatric disorder is
related to the severity of WM tract integrity abnormalities in these patients.
Method. T1-weighted MR and DT images were collected in 68 patients with first-episode schizophrenia [22 with a
positive family history (PFH) and 46 with a negative family history (NFH)] and 100 healthy controls. Voxel-based
analysis was performed and WM integrity was quantified by fractional anisotropy (FA). Cluster- and voxel-level
analyses were performed by using two-sample t tests between patients and controls and/or using a full factorial
model with one factor and three levels among the three sample groups (patients with PFH or NFH, and controls), as
appropriate.
Results. FA deficits were observed in the patient group, especially in the left temporal lobe and right corpus
callosum. This effect was more severe in the non-familial schizophrenia than in the familial schizophrenia subgroup.
Conclusions. Overall, these findings support the hypothesis that loss of WM integrity may be an important
pathophysiological feature of schizophrenia, with particular implications for brain dysmaturation in non-familial and
familial schizophrenia.
Received 1 May 2010; Revised 25 October 2010 ; Accepted 15 November 2010 ; First published online 16 December 2010
Key words: DTI, family history, first episode, fractional anisotropy, schizophrenia, voxel-based analysis.
Introduction
Schizophrenia is a complex brain disorder with puta-
tive disconnectivity between multiple brain regions
(Friston & Frith, 1995). Several lines of evidence have
supported the presence of abnormal white-matter
(WM) tract integrity in patients with schizophrenia.
Post-mortem investigations have shown neuronal
deficits related to glial elements, which may lead to
abnormalities in myelination and synaptic integrity
(Hof et al. 2003; Harrison & Weinberger, 2005).
Structural brain imaging research pertaining to
diffusion tensor imaging (DTI) studies has also
suggested that deficits in WM tracts exist in patients
with schizophrenia (Kubicki et al. 2002a, 2005, 2007 ;
Kanaan et al. 2005 ; Karlsgodt et al. 2008; Lee et al. 2009;
Phillips et al. 2009). Furthermore, functional brain
imaging studies have suggested that patients with
schizophrenia may show impaired fronto-temporal
functional connectivity (Friston, 1998; Lawrie et al.
2002; Liu et al. 2008 ; Kim et al. 2009). Although these
are promising findings, the evidence supporting
abnormal WM tract integrity in patients with schizo-
phrenia remains equivocal (Assaf & Pasternak, 2008).
There are several possible reasons for this. First,
studies used different standards for data acquisition
and data analysis and different methods for data
* Address for correspondence : Professors Tao Li or Qiyong Gong,
28 Dian Xin Nan Road, West China Hospital, Chengdu, Sichuan,
610041, P.R. China.
(Email : xuntao26@hotmail.com)
Psychological Medicine (2011), 41, 1691–1700. f Cambridge University Press 2010
doi:10.1017/S0033291710002412
ORIGINAL ARTICLE
post-processing. Second, small sample size is a limi-
tation in most of the studies. Third, some confounding
factors, such as patient age, course of illness and
medication use, may not have been controlled
adequately. Therefore, definitive pathophysiological
evidence of abnormal WM tract integrity in schizo-
phrenia has yet to be confirmed.
DTI is a useful tool for assessing WM structural
integrity and connectivity in vivo because it yields a
series of quantitative measures such as fractional
anisotropy (FA), which reflects the integrity of WM
tracts. FA reduction has been found in many parts of
the principal WM bundles in schizophrenia, with di-
verse results across studies (Kubicki et al. 2002b, 2003;
Ardekani et al. 2003 ; Sun et al. 2003; Wang et al. 2004).
Genetic factors also play a substantial role in the
pathogenesis of schizophrenia. Some of the brain
morphological deficits seen in schizophrenia are
thought to be related to genetic risk (Bartley et al. 1997;
Callicott & Weinberger, 1999 ; Rijsdijk et al. 2005), and
the heritability of total brain volume has been esti-
mated to be up to 94% (Bartley et al. 1997; Rijsdijk et al.
2005). Twin studies in schizophrenia have reported
that the change in grey matter may be heritable (Baare
et al. 2001; Koolschijn et al. 2008). Previous investiga-
tions have suggested that the non-psychotic relatives
of patients with schizophrenia have brain volume
deficits that are similar to those seen in the patients
themselves, albeit less severe (Gogtay et al. 2003 ;
Job et al. 2003, 2005, 2006; McDonald et al. 2006 ; Ho,
2007). Studies of twins discordant for schizophrenia
have found liability-related decreases of grey matter
volume (Baare et al. 2001; Cannon et al. 2002; Hulshoff
Pol et al. 2004, 2006b) and increases of WM volume
(Hulshoff Pol et al. 2006a, b).
However, the role of family history is still uncertain.
For example, in a study using a large sample size,
Honea et al. (2008) found that brain regional differences
may be a weak intermediate phenotype for schizo-
phrenia. Wood et al. (2005) found that a family history
of schizophrenia was not associated with a greater de-
gree of structural brain abnormalities in an ultra-high-
risk group. DeLisi et al. (1988) found that conventional
volumetric quantification of ventricular space failed to
reveal differences between high-risk participants and
controls; however, by using apparent diffusion coef-
ficients (ADCs) to assess the ventricular space change,
they found a significant difference between the partici-
pants at high risk and the controls.
In the present study, we aimed to explore the
microstructural changes in first-episode schizophrenia
in a Han Chinese population, and to investigate
whether a family history of psychiatric disorders is
related to the severity of WM tract integrity abnor-
malities in such patients.
Method
Participants
We recruited 68 participants with schizophrenia from
in-patient and out-patient psychiatric units in West
China Hospital, Sichuan University. The patients were
assessed by one of two qualified psychiatrists (W.D.
and M.L.) shortly after presentation in their first
episode of psychotic illness to the mental health
services. The psychiatric history of each patient was
reviewed to exclude those with a previous history of
any major psychiatric disorder, including psychotic,
affective and schizo-affective disorders, head trauma,
drug abuse and neurological disorders. Diagnosis
was made according to DSM-IV criteria. All partici-
pants were interviewed using the Structured Clinical
Interview for the DSM-IV (SCID-P for patients and
SCID-NP for controls). All patients were followed
up for at least 6 months to confirm the diagnosis,
especially those participants who were initially diag-
nosed as schizophreniform psychosis. Thirty out of
68 patients were neuroleptic-naive at the time of
magnetic resonance imaging (MRI) scanning, and
the remaining 38 [23 with a negative family history
(NFH) and 15 with a positive family history (PFH) of
schizophrenia] had been minimally treated with anti-
psychotics such as risperidone or olanzapine at low
dosage (ranging from 25 to 75 mg of chlorpromazine
daily dose equivalents) for a brief duration of less than
3 days prior to MRI scanning.
Patients also underwent further clinical evaluation,
which included symptoms on the Positive and
Negative Syndrome Scale (PANSS; Kay et al. 1987)
and global functioning on the Global Assessment
of Functioning (GAF; Hall, 1995). Healthy controls
(n=100) were recruited from the same districts as
the cases ; these individuals were excluded if their
first-degree relatives suffered from any mental illness.
The age of the patients and controls at examination
was less than 45 years, and all participants were right-
handed.
Family psychiatric history was obtained by inter-
viewing the patients and both parents, and also other
first-degree relatives (e.g. sibling and offspring) where
possible, who provided information on family history
in details during the clinical interview. This study
adopted the definition of family history as described
by Xu et al. (2008). Patients with PFH were defined as
having at least one relative with schizophrenia in their
first- or second-degree relatives ; otherwise, they were
defined as patients with NFH. Of the 68 patients
studied, 22 were found to have PFH. Within the PFH
group, four had first-degree relatives (two mothers,
one father and one sister) and the other 18 had second-
degree relatives with a history of schizophrenia.
1692 Q. Wang et al.
All participants were Han Chinese and provided
written informed consent. This study was approved
by the Institutional Review Broad of West China
Hospital, Sichuan University.
MRI scans
All participants underwent MRI scanning in the
Department of Radiology at West China Hospital
using a Signa 3.0 T scanner (GE Medical Systems,
USA) with an eight-channel phase array head coil.
In our MRI unit, our usual practice to ensure quality
assurance of MRI images is based on an in-house
protocol based on Jia et al. (2010), in which we use
phantoms to measure signal-to-noise ratio (SNR) and
image uniformity on a daily basis, noting the voltage
of the transmit radiofrequency amplifier, etc. (Firbank
et al. 2000). High-resolution DTI data were acquired by
using a single-shot spin echo planar imaging (EPI) se-
quence [repetition time/echo time (TR/TE)=10 000/
70.8 ms, 3-mm axial slices with no gap, matrix=
256r256 (0.94 mmr0.94 mmr3 mm), field of view
(FOV)=24 cm
2
, acquisition time=5 min 40 s]. The DTI
sequence used in this protocol included 15 diffusion
gradient directions [b=1000 s/mm
2
, number of ex-
citations (NEX)=2] and one volume without diffusion
weighting (b=0, NEX=2) for 42 slices throughout the
whole brain. Anatomical three-dimensional spoiled
gradient (3D-SPGR) T1 data were also acquired for
registration purposes [TR/TE=8.5/3.4 ms, 1-mm ax-
ial slices, matrix=512r512, FOV=24 cm
2
, inversion
time (TI) =400 ms, NEX=1]. All scans were reviewed
by an experienced neuroradiologist to exclude obvious
gross abnormalities.
Image processing
Images were processed and analysed with SPM5 soft-
ware (www.fil.ion.ucl.ac.uk/spm/software/spm5/).
FA maps were generated from each participant’s DTI
scan using the freely available DTIstudio software
(http://cmrm.med.jhmi.edu/). Prior to FA calcu-
lation, the DTI scans were realigned using the built-in
function in DTIstudio so that each DTI image (b=0s/
mm
2
) was corrected for motion. Two patients with
schizophrenia and one normal control were excluded
in this study because of head and body motion. All
3D-SPGR images were corrected for inhomogeneity,
normalized, and segmented using an integrated gen-
erative model (unified segmentation; Ashburner &
Friston, 2005) with default parameters. The DTI data-
set was registered with the anatomical T1 by mutual-
information co-registration between the b=0 image
and the T1 image. The normalization parameter of
the T1 image was used to normalize the FA map to
standard space. The normalized FA maps were re-
sliced to 2 mmr2mmr2 mm and smoothed with a
6-mm full-width at half-maximum (FWHM) isotropic
Gaussian kernel (Kubicki et al. 2005; Winterer et al.
2008). An explicit mask for statistical analysis was
created by averaging the WM mask of all subjects and
threshold at 0.2 (SPM Masking Toolbox ; Ridgway et al.
2009).
Statistical analyses
Pearson’s x
2
test, Student’s t test and analysis of vari-
ance (ANOVA) were used to compare the distribution
and differences of categorical and continuous data
respectively. Cluster- and voxel-level analyses were
performed by using a full factorial model of two-
sample t tests between the patients and controls
and/or a full factorial model of one factor and three
levels statistical comparisons among the three sample
groups (PFH, NFH and control), as appropriate. Age,
gender and educational attendance were included as
covariates. Cluster-level significance probability was
set at p<0.05 (uncorrected) with cluster voxels >50 in
two-sample t tests (Cheung et al. 2010). In this ex-
ploratory analysis, voxel-level significance probability
was set at p<0.001 (uncorrected) with cluster voxels
>50 (Ke et al. 2009) in a full factorial model of one
factor and three levels statistical comparisons among
the three sample groups. For comparison among the
three groups, the results in the group mapping analy-
sis were saved to files and imported into the MarsBar
toolbox and the mean values of FA of each region were
extracted to be calculated for each subject (Cheung
et al. 2008 ; Jia et al. 2010). These values were further
analysed using the Statistical Package for Social
Sciences (SPSS) for Windows version 13.0 (SPSS Inc.,
USA). Bonferroni correction was applied for multiple
comparisons and the level of statistical significance
was set at p<0.05 [12 comparisons (three groupsr
four voxels) were set].
Results
Demographic characteristics
The demographic characteristics of the participants
are shown in Table 1. There were no significant dif-
ferences in mean age (t=x1.14, p=0.256, range=
15–45 years), gender distribution (Pearson’s x
2
=0.395,
p<0.53) or mean educational attendance (t=x 1.578,
p=0.116) between the patients and controls. Signifi-
cant differences in educational attainment were found
among the patients in the PFH, NFH and control
groups (F=4.30, p<0. 015); however, no significant
differences were observed for the mean age (F=0.650,
Connectivity of white-matter tracts in schizophrenia 1693
p<0.52) and gender ratio (Pearson’s x
2
=3.42, p<0.18)
among these groups. In the subgroup comparisons,
differences in educational attainment were found be-
tween the control and PFH groups, and between the
PFH and NFH groups. Both control and NFH groups
had significantly higher educational attainment than
the PFH group (Table 1).
Comparison between the patients and controls
Figure 1 and Table 2 show that, compared with the
controls, FA in the patients was lower in the right
cerebral sublobar extranuclear WM of the corpus
callosum (RCC), left cerebral sublobar extranuclear
WM of the corpus callosum (LCC) and left temporal
lobe WM (LT) when stringent cluster-level analysis
was adopted. The voxel- and cluster-level maps were
very similar in the regions showing lower FA in the
patients, although the voxel-level maps also showed
additional regions with increased FA in the patients,
but generally with very small spatial extent (<50).
This difference did not exist in the cluster-level maps.
Comparison between the PFH, NFH and control
groups
Figure 2 and Table 3 show that, by using analysis of
covariance (ANCOVA) in SPM5, we found significant
differences in the FA values among the three groups
in the RCC, LT, right cerebral parietal lobe precuneus
(RPP), and left cerebral occipital lobe precuneus
(LOP). The mean FA values from all participants
were then extracted in these four regions and com-
pared among the PFH, NFP and control groups by
ANCOVA with SPSS. Significant differences in these
regions were confirmed after a Bonferroni multiple
correction test (12 comparisons). In the between-group
comparison, the mean FA values in both the PFH and
NFH groups were lower in all four regions when
compared with the control group, and the NFH group
had lower FA values in the four regions when com-
pared with the PFH group. In addition, there were no
significant associations between the duration of illness
or the GAF score, on the one hand, and FA values of
the four regions, on the other, in patients with schizo-
phrenia (p>0.05). We also did not find any significant
difference in FA values in the four brain regions be-
tween the patients who were neuroleptic-naive and
those who were on medication (p>0.05).
Discussion
In line with previous studies, this study supports the
finding that patients with first-episode schizophrenia
show callosal (Price et al. 2005) and temporal WM ab-
normalities (Lim et al. 1999; Wang et al. 2003, 2004)
in the form of FA deficits compared with controls.
Table 1. Demographic profile of participants [values are mean (S.D.)]
Controls
Cases
Total
PFH group
(drug)
NFH group
(drug)
n 100 68 22/15 46/23
Age (years) 25.58 (8.07) 24.13 (7.96) 24.00 (8.91) 24.20 (7.74)
Educational attainment (years) 12.8 (3.35) 11.99 (3.18) 10.59 (3.61) 12.65 (2.75)
Age range (years) 15–45 16–45 16–45 15–44
Gender (M/F) 52/48 32/36 7/15 25/21
Age at onset (years) 23.15 (8.12) 23.42 (7.86)
Duration of illness (months) 9.15 (16.46) 8.73 (14.58)
PANSS-P 26.83 (6.03) 25.79 (6.58)
PANSS-N 19.70 (8.30) 19.53 (8.21)
PANSS-G 49.91 (9.65) 50.06 (10.85)
PANSS-T 96.43 (14.36) 95.38 (19.98)
GAF 27.87 (7.02) 26.81 (10.55)
PFH, Positive family history ; NFH, negative family history ; M, male ; F, female ; PANSS, Positive and Negative Syndrome
Scale ; PANSS-P, subscales for positive symptoms; PANSS-N, subscales for negative symptoms ; PANSS-G, subscales for
general psychopathological symptoms; PANSS-T, total score of PANSS; GAF, Global Assessment of Functioning.
For cases and controls, there were no differences in age [t(166)=x1.14, p<0.256], educational attendance [t(166)=x1.578,
p<0.123] or gender [x
2
(1, df=167)=0.395, p<0.530]. For control, PFH and NFH groups, there were no differences in age
[F(2)=0.650, p<0.523] but there was a difference in educational attendance [F(2)=4.303, p<0.015]. Post-hoc educational
attendance differed between the control and PFH groups (p<0.004), and between PFH and NFH groups (p< 0.015), but not
between the control and NFH groups (p<0.798). Gender did not differ among the groups [x
2
(3, df= 166)=1.37, p<0.50].
1694 Q. Wang et al.
Moreover, the three-group ANOVA revealed that the
most striking FA deficits in the PFH and NFH patients
were located in the right cerebral sublobar extra-
nuclear WM, that is the posterior portion of the corpus
callosum (RCC), as compared with controls. Taken
together with the RPP, LT and LOP FA deficits, our
findings indicate the presence of widespread temporo-
parieto-occipital WM disconnectivity in the disorder.
Temporal and callosal WM abnormalities have been
widely reported in schizophrenia (Lim et al . 1999;
Shenton et al. 2001 ; Steel et al. 2001 ; Minami et al. 2003;
Wang et al. 2003, 2004; Kanaan et al. 2005 ; Lee et al.
2009; Phillips et al . 2009). Our results are also partly in
line with the findings of Schulze et al. (2003) and Wood
et al. (2005); the latter speculated that the mechanisms
resulting in gross morphological anomalies in the
hippocampus and anterior cingulate cortex in psy-
chosis were mainly due to environmental factors
rather than genetic loading. Cumulative evidence
from FA studies supports the involvement of the
corpus callosum in the pathology of inter-hemispheric
connectivity (Price et al. 2005 ; Cheung et al. 2008 ;
Rotarska-Jagiela et al. 2008 ; Gasparotti et al. 2009) as
the principal WM tract in the brain with homotypic
connections to the contralateral cortex (Bloom &
Hynd, 2005; Gasparotti et al. 2009). In addition, several
studies have reported grey matter reductions of the
lateral and medial temporal cortices in schizophrenia
(Narr et al. 2004; Goldman et al. 2008; Honea et al.
2008). The temporal grey matter deficits were pre-
sumed to be related to auditory sensory (Rabinowicz
et al. 2000) and neurocognitive impairment in patients
with schizophrenia (DeLisi, 2001; Harrison, 2004;
Green et al. 2005; Yamada et al. 2007). This study is
in line with previous findings that indicated that
temporal neocortical networks deficits may be related
to schizophrenia (Phillips et al. 2009). Our finding of
reduced FA in WM adjacent to the right precuneus
concurs with that reported in the first DTI study
of subjects with first-episode schizophrenia (Cheung
0.6
0.55
0.5
0.45
0.4
FA value
LT RCC LCC
White matter tracts
Schizophrenia Controls
Left
4
3
2
1
–22 20 25
Right
(a)
(b)
Fig. 1. Comparison between cases and controls : in the right cerebral sublobar extranuclear white matter (WM) of the corpus
callosum (RCC), left cerebral temporal lobe subgyral WM (LT) and left cerebral sublobar extranuclear WM of the corpus
callosum (LCC) regions, different fractional anisotropy (FA) values were obtained between the patients and normal controls
at the uncorrected cluster level (p <0.05).
Table 2. Comparison between cases with schizophrenia
and controls
Region (MNI coordinates) K
p (cluster
level)
LT (x45, x16, x22) 395 0.0216
RCC (18, x 45, 20) 916 0.0012
LCC (x6, x 14, 25) 308 0.0388
LT, Left cerebral temporal lobe subgyral white matter
(WM); RCC, right cerebral sublobar extranuclear WM of
corpus callosum ; LCC, left cerebral sublobar extranuclear
WM of corpus callosum ; K, number of voxels; MNI,
Montreal Neurological Institute.
Connectivity of white-matter tracts in schizophrenia 1695
![Table 1. Demographic profile of participants [values are mean (S.D.)]](/figures/table-1-demographic-profile-of-participants-values-are-mean-dqze9x65.webp)