POWIC SANE Research Centre, Warneford Hospital, Oxford
Highfield Adolescent Unit, Warneford Hospital, Oxford
Department of Psychiatry, University of Oxford, Warneford Hospital, Oxford
Highfield Adolescent Unit, Warneford Hospital, Oxford
Magnetic Resonance and Image Analysis Research Centre (MARIARC), University of Liverpool, Liverpool
POWIC SANE Research Centre, University of Oxford, Warneford Hospital, Oxford
Correspondence: Professor T. J. Crow, POWIC SANE Research Centre, University Department of Psychiatry, Warneford Hospital, Oxford OX37JX, UK. E-mail: tim.crow{at}psych.ox.ac.uk
Declaration of interest None. Funding details in Acknowledgements.
|
|
|---|
Aims To investigate brain structure, asymmetry and IQ in early-onset schizophrenia.
Method Volumes of left and right cerebral hemispheres and IQ were assessed in 33 participants with early-onset DSMIV schizophrenia and 30 members of a matched, normal control group.
Results Total brain volume was significantly smaller in the group with early-onset disease (cases) relative to the control group (4.5%), especially for the left hemisphere in males (6.0%). A significant sex x diagnosis interaction in hemisphere asymmetry revealed that the female cases group had significantly reduced rightward asymmetry relative to the female control group and that the male cases tended to have reduced leftward asymmetry relative to the male control group. Decreased left hemisphere volume in males and decreased rightward hemispheric asymmetry in females correlated with reduced IQ.
Conclusions Sexually dimorphic alterations in asymmetry correlate with degree of intellectual impairment in early-onset schizophrenia.
|
|
|---|
|
View this table: [in a new window] | Table 1 Demographic characteristics of study groups of patients with early-onset schizophrenia and percentage change in ventricular, temporal lobe and total cerebral volumes in early-onset cohorts relative to matched normal controls |
|
View this table: [in a new window] | Table 2 Mean verbal, performance and full-scale IQ scores in early-onset cohorts relative to matched normal controls |
|
|
|---|
|
View this table: [in a new window] | Table 3 Demographic characteristics of 33 study participants with early-onset schizophrenia (cases) and 30 participants in a matched normal control group (controls) |
Magnetic resonance image acquisition and analysis
Magnetic resonance images were acquired using a 1.5 T Magnetom Vision whole
body system (Siemens Medical System Inc., Erlangen, Germany). One hundred and
fifty-six coronal T1-weighted images were obtained using a
three-dimensional spoiled gradient echo pulse sequence (time to repetition=34
ms, time to echo=9 ms, flip angle 30°). The field of view of the images
was 20 cm, with 1.5 mm slice thickness. The left and right temporal lobes were
optimally visualised, and their volumes best measured, on image sections
oriented perpendicularly to the long axis of the hippocampus
(Mackay et al, 1998).
These sections were obtained by reformatting oblique sections through the
acquired three-dimensional data using New Region of Interest Analysis (NRIA)
software (Brain Behavior Laboratory, University of Pennsylvania, USA) running
on an Ultra 10 Workstation (Sun Microsystems, California, USA), where the 256
x 256 x 156 acquired voxels of side 0.78 mm x 0.78 mm
x 1.5 mm were linearly interpolated to 256 x 256 x 256 cubic
voxels of side 0.78 mm. This was also a convenient sectioning direction for
volume estimation of the left and right cerebral hemisphere and lateral
ventricles.
Unbiased estimates of structure volume were obtained using the mathematically unbiased Cavalieri method of modern design stereology in combination with point counting (Roberts et al, 1994; Mackay et al, 1998), using EasyMeasure software (http://www.easymeasure.co.uk); see Roberts et al (2000). The posterior limit of the temporal lobe was defined as the point where the lateral ventricles divide into frontal and temporal horns. The cerebral hemispheres were separated from the brain-stem at the superior limit of the pons. A more detailed description of these definitions and the methodology is given in Mackay et al (1998). An inter/intrarater reliability study was carried out by three raters. Intraclass correlation coefficients were calculated (Bartko, 1966) and found to be greater than 0.9 for the lateral ventricles, and above 0.8 for temporal lobe and cerebral hemisphere. An index of asymmetry was computed by subtracting the volume of the structure in the left hemisphere (L) from the volume in the right hemisphere (R) and expressing the difference as a percentage of mean volume, i.e. (R-L)/[(R+L)/2] x 100. Temporal lobe volume, lateral ventricle volume and asymmetry measures were considered both as absolute values and as proportions of total cerebral hemisphere volume.
Intelligence assessment
Verbal, performance and full-scale IQ data were collected for 28
participants in the cases group and 30 in the control group.
Five of the original 33 participants did not complete IQ testing, and were not
considered in the analysis of IQ. Participants were tested with either the
full version of the Wechsler Intelligence Scale for Children Revised
(Wechsler, 1992) or, if they
were more than 16 years old, the Wechsler Adult Intelligence Scale
Revised (Wechsler, 1981). Whenever possible, testing was performed in one uninterrupted session.
Following testing, all sub-scale scores were transformed into age-scaled
scores to render them equivalent. Standard IQ indices were calculated.
Statistical analysis
Data were analysed using the Statistical Package for the Social Sciences,
version 10 for PC. First, one-way analysis of variance (ANOVA) was performed
in order to detect main effects and/or sex interactions in demographic
variables. Between-groups comparisons of regional volumes and IQ were
performed using the generalised linear model. Multiple analysis of variance
(MANOVA) was used to examine structural volume, verbal, performance and
full-scale IQ and sub-test differences in IQ performance.
Non-parametric chi-squared analysis was used to determine differences in the distribution of positive and negative asymmetries between groups. All correlations between demographic, treatment and illness-related variables were performed with non-parametric rho connected for multiple comparisons with the Bonferroni test. Given that the normal male brain is significantly larger than the normal female brain (e.g. Gur et al, 1999), and previous studies that show differences in asymmetry between the sexes (e.g. Bear et al, 1986), males and females were examined separately in both volumetric and IQ analyses.
|
|
|---|
|
View this table: [in a new window] | Table 4 Mean volumes (ml) of the left and right hemisphere, lateral ventricle and temporal lobe and mean asymmetry indices for each structure in the group of participants with schizophrenia (cases) and in the control group (controls). Standard deviations are shown in parentheses |
A significant main effect of sex (F=22.15, P=0.0001) indicated that males as a group had significantly larger brains than females, regardless of diagnosis. Males in the cases group demonstrated a 5.1% reduction in overall brain volume (right plus left hemisphere) relative to males in the control group (MANOVA, F=4.28, P=0.45), whereas females in the case group demonstrated a smaller, 3.5% reduction (P=0.17, NS). MANOVA of individual hemisphere volumes in males and females revealed a significant reduction in left, but not right, hemisphere volume in cases relative to controls in males (F=6.17, P=0.01) but not in females. Males also showed a trend to left temporal lobe reduction compared with male controls (P=0.08).
Overall there was no statistically significant (P<0.05)
asymmetry of the cerebral hemispheres, temporal lobes or lateral ventricles.
However, when sex was entered as an independent variable, a significant
diagnosis x sex interaction was detected in hemisphere asymmetry after
correction for overall brain size (MANOVA, F=4.39, P=0.01).
Post hoc analyses revealed that the female cases group showed
significant leftward asymmetry of cerebral hemisphere volume (t=2.28,
P=0.04), and this was significantly different from a tendency to
rightward asymmetry in the female control group (F=4.97,
P=0.03). No significant asymmetry was detected in males, where the
cases group tended towards rightward asymmetry and the control group towards
leftward asymmetry. In total, 9 out of 11 participants (82%) in the female
cases group demonstrated left greater than right asymmetry, which was present
in only 5 out of 12 (41%) of the females in the control group
(
2=4.5, P <0.04). There was no difference in the
proportion of the male cases group relative to the male control group that
showed rightward asymmetry.
IQ analysis
Mean performance IQ, verbal IQ and full-scale IQ scores are shown in
Table 5. Both males and females
in the case group showed significant impairments on all three tests relative
to controls (Table 5). Seventy
per cent of the early-onset group had IQs beneath the low average range of
performance (full-scale IQ less than 90). There was large variability in the
average verbalperformance IQ discrepancy. In the cases group the
average discrepancy was 5.14 IQ points (s.d.=14.8) compared with -0.16 in
controls (s.d.=17.5) but this was not statistically significant
(P=0.22). When the sexes were compared between groups, no significant
discrepancy was found between the male case and control groups (3.21,
s.d.=14.5, and -0.88, s.d.=19.0, respectively; P=0.47). The female
cases group demonstrated a large verbalperformance IQ discrepancy
compared with the female control group (9.22, s.d.=15.4, and 0.83, s.d.=15.8,
respectively), but this was not statistically significant (P=0.23).
When individual Wechsler sub-test scores were examined by MANOVA, significant
differences between case and control groups were found across all 11
sub-tests. The average sub-test score in the early-onset group was 7.6
(s.d.=2.0), whereas the control group average was 3 IQ points higher (10.6,
s.d.=2.4). A 3-point discrepancy in IQ sub-scale performance is indicative of
abnormality (Kaufman,
1990).
|
View this table: [in a new window] | Table 5 Mean verbal, performance and full-scale IQ scores in the group of participants with schizophrenia (cases) and the control group (controls). Standard deviations are shown in parentheses |
Correlation analysis
Age, duration of illness and current medication did not correlate with any
of the structural of IQ measures in the cases group. However, earlier age of
onset was associated with increasing ventricle volume as a proportion of total
brain volume (
= -0.35, P <0.05). There were no statistically
significant correlations between IQ and brain structure measures in the
control group as a whole, or for males and females separately. This was also
the case for the combined patient group. However, when the males and females
were examined separately, full-scale IQ was significantly correlated with left
hemisphere volume (
=0.47, P <0.05) in male cases, and verbal
IQ was positively correlated with increased rightward hemispheric asymmetry in
female cases (
=0.81, P <0.01).
|
|
|---|
Volume and IQ
In agreement with previous studies in early-onset schizophrenia (see Tables
1 and
2), this sample showed an
average 4.5% deficit in total brain volume relative to controls, which was
greater on the left than the right, and average full-scale IQ that was close
to the cut-off (80) between low average (IQ=9080) and borderline
(IQ=7966) ranges of performance. Previous studies have reported greater
severity of intellectual disturbance in earlier-onset than in later-onset
disease (Yang et al,
1995; Basso et al,
1997). Our findings corresponded particularly well with those of
Matsumoto et al
(2001), who found comparable
reductions in brain volume and full-scale IQ in a cohort of similar age.
Together, these findings suggest that alterations in gross cerebral structure
and IQ in early-onset schizophrenia are less severe than those observed in
childhood-onset disease, but greater than those observed in adult-onset
schizophrenia.
Two unexpected findings require further elucidation. First, unlike some previous studies of normal adults, we did not find significant volumeIQ correlations in our normal control sample. Most, but not all, studies find modest but significant positive correlations between overall brain volume (Willerman et al, 1991, 1992; Andreasen et al, 1993; Wickett et al, 2000) or asymmetry (Yeo et al, 1987; Reiss et al, 1996) and IQ. It is possible that a larger normal control group would have revealed comparable volumeIQ correlations. Second, the increase (average 20%) in ventricular volume in cases relative to controls failed to reach significance, although ventricle-to-brain ratio was inversely related to age at onset. Ventricular enlargement is a robust finding in adult schizophrenia (McCarley et al, 1999) and most studies of early-onset cohorts report increased (James et al, 1999; Kumra et al, 2000; Sowell et al, 2000) and/or progressive enlargement of lateral ventricular volume (Rappoport et al, 1997). Our findings indicate considerable variability in the present sample, a finding that is consistent with studies showing variability in ventricular enlargement in first-onset patients (Lieberman et al, 2001; Puri et al, 2001), particularly in the early stages of the illness (Gur et al, 1998; Puri et al, 2001).
Sex differences
Our findings provide evidence that changes in brain volume and the
relationship to IQ are, to some extent, sexually dimorphic in patients with
early-onset schizophrenia. Male, but not female, participants in the cases
group had reduced left hemisphere volume and a trend to reduced temporal lobe
volume. A significant diagnosis x sex interaction in cerebral hemisphere
asymmetry was found such that the tendency for rightward asymmetry in the
female control group was reversed to a leftward asymmetry in females with
schizophrenia. Males did not show a significant effect, although there was a
trend to reversal of the normally observed leftward pattern (i.e. in the
opposite direction to females). Sex differences in asymmetry in adults have
been previously observed in frontal, temporal and whole brain measurements
(Bilder et al, 1994;
Highley et al, 1998)
and have been shown to interact with age at onset
(Highley et al, 1998;
Maher et al, 1998; McDonald et al,
2000). Our findings suggest that sex-specific alterations in
asymmetry are present in patients with early-onset schizophrenia, consistent
with the hypothesis (Crow,
1993,
2000) that a sex-linked
determinant of asymmetry has a critical role in the aetiology of
psychosis.
Correlations between hemisphere volume and IQ measures in this cohort were sex-specific. In the male cases group, reduced left hemisphere volume was correlated with lower full-scale IQ. The reduction in rightward asymmetry in female cases relative to controls was associated with a selective reduction in verbal IQ. In effect, these structurefunction relationships were consistent with a body of evidence indicating that sex differences in IQ decline are associated with lateralised cerebral disturbance (see Kaufman, 1990, for review). A simple interpretation of the significant differences and trends in our findings is that in males with schizophrenia the deficits are associated with loss that is relatively selective to the left hemisphere, whereas in females they are associated with a loss that is relatively greater in the right hemisphere. This differs somewhat from the findings of Flaum et al (1994) who also examined brain structure and IQ in adults with schizophrenia in relation to sex. They reported that the pattern of structurefunction correlations in females with schizophrenia was similar to those of female controls, but males with schizophrenia demonstrated no significant structurefunction relationships. Further studies are required, but it is noteworthy that Flaum et al did not examine indices of cerebral asymmetry, where sexual dimorphism in healthy individuals is established (Bear et al, 1986; Barrick et al, 2001).
Neurodevelopmental antecedents of volume and IQ reduction
Although the relationship between age, sex and onset of psychosis is
complex, our findings are consistent with neurodevelopmental explanations of
schizophrenia. Adolescence is a critical stage in cerebral maturation
involving substantial volume increases in white matter relative to grey matter
(Reiss et al, 1996;
Giedd et al, 1997;
Courchesne et al,
2000), accompanied by increase in the volume of the lateral
ventricles (Giedd et al,
1996,
1997). Furthermore, these
changes are influenced by sex, as the extent of grey-matter volume reduction
is greater in males than females (Coffey
et al, 1998; Raz
et al, 1998). Our findings indicate that, in adolescents
with schizophrenia, the intellectual deficits depend on an interaction between
sex and relative hemispheric development.
|
|
|---|
LIMITATIONS
|
|
|---|
|
|
|---|
Related articles in BJP:
This article has been cited by other articles:
![]() |
S. Verma, Y. Y. Sitoh, Y.-C. L. Ho, L. Y. Poon, M. Subramaniam, Y. H. Chan, K. Sim, and S. A. Chong Hippocampal Volumes in First-Episode Psychosis J Neuropsychiatry Clin Neurosci, February 1, 2009; 21(1): 24 - 29. [Abstract] [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||