Neuroscience Letters 301 (2001) 163ą166
www.elsevier.com/locate/neulet
Enlargement of the fornix in early-onset schizophrenia:
a quantitative MRI study
D.C. Daviesa,*, A.M.J. Wardellb, R. Woolseya, A.C.D. Jamesc
a
Department of Anatomy and Developmental Biology, St George's Hospital Medical School, Cranmer Terrace, Tooting,
London SW17 0RE, UK
b
Department of Child Mental Health, St. George's Hospital Medical School, London SW17 0RE, UK
c
The Warneford Hospital, Oxford OX3 7JX, UK
Received 3 December 2000; accepted 29 January 2001
Abstract
Abnormalities of temporal lobe structure and frontal lobe function occur in schizophrenia. There have been few
studies of young people with schizophrenia and little is known about temporo-frontal connectivity in the disease. There-
fore, the cross-sectional area of the body of the fornix was measured on MR images from 17 young people with
schizophrenia, nine with other serious psychiatric illnesses and eight without illness. The mean age of each group
was 16ą17 years. The mean cross-sectional fornix area in subjects with schizophrenia was signi�cantly larger than
that in subjects without illness (<40%) and psychiatric controls (<26%). There was no such signi�cant difference
between subjects without illness and psychiatric controls. The nature of the larger fornix in early-onset schizophrenia,
whether it persists and whether it occurs in schizophrenia presenting in adulthood, remain to be elucidated. q 2001
Elsevier Science Ireland Ltd. All rights reserved.
Keywords: Early-onset schizophrenia; Fornix; Hippocampal formation; Limbic system; Prefrontal cortex
Post mortem studies have revealed reductions in hippo- phenomenologically and epidemiologically continuous
campal and parahippocampal gyrus volumes in schizophre- with adult-onset schizophrenia. It is particularly attractive
nia [5,9]. Cytoarchitectural disarray and cell loss from these for investigation of the psychiatric, psychological and
structures have also been reported to occur in the disease anatomical markers of schizophrenia because (1) the struc-
[4,14,16,19,22,24]. Such abnormalities alone are unlikely to tural markers are already likely to be present, (2) early-onset
be able to account directly for the wide variety of symptoms patients may well be severely affected and (3) they present
associated with schizophrenia, many of which appear to be opportunities to follow-up cohorts throughout their lives.
prefrontal in origin. However, these hippocampal abnorm- Young people presenting with suspected schizophrenia
alities may also affect the function of brain structures with (but no other neurological or psychiatric disorder) to
which the hippocampal formation/parahippocampal gyrus Child and Adolescent Psychiatrists in the former Oxford
are connected. and South West Thames regions, were recruited as they
The fornix is the major �bre tract connecting the hippo- presented and MR images of their brains were obtained
campal formation with sub-cortical brain regions and soon afterwards. Following diagnosis ful�lling DSM-IV
provides an indirect connection between the hippocampal criteria [1], there were 17 patients (11 male, six female)
formation and the prefrontal cortex. It is a large discrete with schizophrenia (mean age 16 years 11 months ^ 5
bundle that lends itself to in vivo measurement. The asso- months SEM, range 14 years 10 monthsą20 years 5
ciation between schizophrenia and the size of the fornix was months). Although the subjects with schizophrenia had
therefore investigated in young people using magnetic reso- received neuroleptic medication before they were scanned,
nance imaging (MRI). Early-onset schizophrenia (presenta- none had received long-term treatment. Eight subjects (four
tion between 13ą18 years) is considered to be male, four female) with no history of psychiatric or neuro-
logical illness, recruited as volunteers through general prac-
* Corresponding author. Tel.: [phone]; fax: 144-20- titioners, served as normal controls. They were scanned at a
8725-3326.
mean age of 16 years 11 months ^ 7 months SEM (range 14
E-mail address: [e-mail] (D.C. Davies).
0304-3940/01/$ - see front matter q 2001 Elsevier Science Ireland Ltd. All rights reserved.
PII: S0304-3940(01)01637-8
164 D.C. Davies et al. / Neuroscience Letters 301 (2001) 163ą166
years 0 monthsą18 years 4 months). Further individuals (six effect of either age (F1;29 � 1:93, P � 0:176) or sex
male, three female) without schizophrenia or any neurolo- (F1;29 � 1:11, P � 0:301) on the mean cross-sectional area
gical abnormality, but presenting with a serious psychiatric of the body of the fornix. Breaking down the data by diag-
illness, were scanned soon after presentation. According to nosis, multiple regression analysis revealed that the mean
DSM-IV criteria, these patients were suffering from major (^SEM) cross-sectional area of the body of the fornix of
depressive disorders: with psychotic features (three), major subjects with schizophrenia (17.78 ^ 0.81 mm2) was signi�-
depressive disorder (two); bipolar I disorder (two); bipolar cantly larger (39.69%) than that of normal controls (12.73 ^
II disorder (one) and psychotic disorder-drug induced (one). 1.04 mm2; regression coef�cient, 4.88, P , 0:005). The mean
This psychiatric control group of patients had a mean age of cross-sectional area of the body of the fornix of subjects with
16 years 3 months ^ 6 months SEM (range 12 years schizophrenia was also signi�cantly larger (26.23%) than that
9 monthsą17 years 9 months) at scanning. All subjects in of psychiatric controls (14.08 ^ 1.67 mm2; regression coef�-
this study were in full-time, mainstream education prior to cient, 3.38, P , 0:036). There was no signi�cant difference
the onset of any illness, although one subject with schizo- between the mean cross-sectional area of the body of the
phrenia had an IQ of 70 and another was described as having fornix of normal control and psychiatric control subjects
mild learning dif�culties. This study received approval from (regression coef�cient, 1.50, P . 0:417). The diagnosis,
the Oxford Psychiatry Research Ethics Committee and the age at scanning, sex and cross-sectional area of the body of
St George's Healthcare Ethics Committee. All subjects and the fornix for each subject are given in Table 1.
their parents gave written informed consent. The �nding that the mean cross-sectional area of the body
Magnetic resonance images (MRI) were obtained using a
General Electric SIGNA 1.5 Tesla machine. The subject's
Table 1
chin was elevated so that the scanned volume was perpen-
The cross-sectional area (mm3) of the fornices of the subjectsa
dicular to the long axis of the temporal lobe to minimise
Subject No. Diagnosis Sex Age at Scan X-sectional Area
partial volume effects. Two initial scans were performed to
ensure correct patient orientation, i.e. the anterior part of the
1 S M 20.5 23.50
genu of the corpus callosum and the clivus followed a verti-
2 S M 18.10 19.35
cal line. Coronal volumetric T1-weighted gradient-echo
3 S M 16.3 18.55
4 S M 18.2 19.36
images (TE � 5 ms, TR � 35 ms, �eld of view 20 Ł 20
5 S F 15.1 16.79
cm and 256 Ł 256 image matrix) were employed. For the
6 S M 19.9 14.65
majority of subjects, the brain was covered by 60 slices of
7 S M 14.10 13.15
3.0 mm thickness. The brains of four subjects were covered
8 S M 17.5 20.57
by 124 slices of 1.5 mm thickness.
9 S F 16.2 11.66
10 S M 16.2 23.46
A MR scan was selected for each subject at the level of
11 S F 16.2 15.06
the mammillary bodies, where the fornices from each side of
12 S M 18.3 18.13
the brain were contiguous forming the body of the fornix,
13 S F 17.11 16.98
the long axis of which was orientated perpendicular to the
14 S M 15.1 15.34
plane of the coronal image. The cross-sectional outline of
15 S F 15.7 15.69
16 S F 15.0 20.20
the body of the fornix was traced together with the small
17 S M 16.0 19.78
area of associated septum pellucidum on a computer moni-
18 C M 17.0 15.72
tor, using a mouse-driven cursor. The upper boundary of the
19 C M 14.0 11.40
septum pellucidum was taken to follow the natural curve of
20 C F 18.3 12.14
the corpus callosum. The cross-sectional area of the body of
21 C F 18.4 10.45
22 C M 18.4 18.49
the fornix was then measured by planimetry using ANALY-
23 C F 17.2 9.63
ZEe software (Mayo Foundation) running on a Sun Sparc
24 C M 14.7 11.97
Station 2 computer. The measurement of each fornix was
25 C F 17.11 12.01
repeated three times, on each of three separate days and a
26 PC F 12.9 16.54
mean value obtained from the nine measurements. All MR
27 PC M 16.7 15.50
28 PC M 17.9 21.69
images were analysed `blind'.
29 PC F 17.2 9.56
Statistical analysis of the data was performed by means of
30 PC F 17.7 20.61
multivariate analysis of variance with the factors diagnosis,
31 PC M 16.0 7.82
age at scanning and sex. The correlation between the cross-
32 PC M 15.9 12.52
sectional area of the body of the fornix and diagnosis was
33 PC M 15.11 8.54
34 PC M 16.11 13.97
then investigated by multiple regression analysis.
Multivariate analysis of variance revealed that there was a
a
Subjects with schizophrenia (S), normal controls (C) and
signi�cant effect of diagnosis on the mean cross-sectional
psychiatric controls (PC), measured from MR scans at the level
area of the body of the fornix at the level of the mammillary of the mamillary bodies. Age at scan is given in years and
months; M, male; F, female.
bodies (F2;29 � 5:55, P � 0:009). There was no signi�cant
D.C. Davies et al. / Neuroscience Letters 301 (2001) 163ą166 165
of the fornix in young people with schizophrenia was signif- [23] and reduced size of the prefrontal cortex [25] have all
icantly larger than both those in normal control and psychia- been reported to occur in schizophrenia and these de�cien-
tric control subjects of a similar age, suggests a speci�c effect cies may stimulate compensatory innervation by hippocam-
of schizophrenia rather than a non-speci�c effect of psychia- pal efferents in the fornix.
tric illness. Drug-abuse was not a prominent feature of the The fornix does not only carry hippocampal efferents. It
subjects in this study, nor did it appear to be a factor in also contains prominent inputs from the medial septal
presenting. Although the subjects with schizophrenia had complex and the posterior hypothalamus. Thus, the possibi-
received some neuroleptic treatment before scanning, the lity exists that an increased number of axons in the fornix
fact that the subjects were scanned soon after presentation could be due to an increased number of afferents, in an
meant that this treatment was limited at the time of scanning. attempt to compensate for a dysfunctional hippocampal
Thus, it is unlikely that drug-abuse or neuroleptic treatment formation. However, since swollen neurons with granular
contributed to the enlarged cross-sectional area of the accumulations (that are presumably degenerating) have
fornices observed in patients with schizophrenia in the been reported to be present in the septal nuclei in schizo-
current study. The signal to noise ratios of both 1.5 mm phrenia [3], this would seem to be unlikely. It is also unli-
and 3.0 mm MR images were such that the outline of the kely that the greater mean cross-sectional area of the body
body of the fornix was clearly and similarly de�nable. No of the fornix observed in the current study is due to gliosis,
partial volume effects were apparent on any image measured. since there is little evidence that gliosis is a feature of schi-
Magnetic resonance imaging studies have previously zophrenia (see [18]).
indicated that cortical volume in the mesial temporal cortex The entorhinal cortex gives rise to the major cortical input
in the region of the rostral hippocampus is smaller in to the hippocampal formation. It is also a major recipient of
subjects with schizophrenia than in healthy subjects hippocampal outŻow. The results of both neuropathology
[7,8,17]. Moreover, the hippocampus itself has been [9,14] and MR imaging studies [13,21] have revealed the
reported to have a smaller volume in schizophrenia [6], parahippocampal/entorhinal area to be smaller than normal
due at least in part to a lower number of hippocampal pyra- in schizophrenia. Abnormal lamina II pre-a neuron clusters
midal neurons [14,20]. Abnormal dendritic organisation and lower neuron densities have been reported to occur in
[24] and aberrant orientation of hippocampal pyramidal the super�cial layers of the entorhinal cortex [2,19]. These
neurons [12,22] have been reported to occur in schizophre- lamina II pre-a neuron clusters were subsequently shown to
nia. However, other authors [11] failed to replicate these be poorly developed and to lie at a deeper level within the
�ndings and concluded that hippocampal outŻow through cortex than in normal control subjects [15]. These authors
CA1 to the subiculum (and thence to widespread cortical suggested that the neuronal abnormalities they described
and sub-cortical areas) is unaffected in schizophrenia. were due to arrested migration, whereby neurons generated
Cytoarchitectural abnormalities in the hippocampal late in cortical development failed to reach the super�cial
formation appear to be more pronounced in CA3 and CA4 laminae and remained in the deeper laminae. Disruption of
than in CA1 and the subicular complex [4,14]. Abnormal- layer II of the entorhinal cortex would be likely to affect
ities in CA3 and CA4 may result in alterations in the output adversely the perforant path input to the hippocampal
from these regions. The primary efferent axons from CA3 formation and pre-a neuron clusters lying at a deeper
pyramids project into the fornix, while their Schaffer collat- level than normal may also disrupt hippocampal output
erals project to CA1. Cytoarchitectural abnormalities in from the subiculum/CA1 to layer IV of the entorhinal
CA3 could result in a failure of axonal collateral branching cortex. Thus, the evidence suggests that the connections
or in collaterals not following their normal trajectories, between the entorhinal cortex and the hippocampal forma-
resulting in more axons in the fornix. CA3 projects via the tion are compromised in schizophrenia, resulting in
precommissural fornix to the lateral septal nucleus and impaired cortical input to the hippocampal formation.
efferents from the septum project to the mediodorsal thala- Furthermore, since the subicular complex projects to the
mic nucleus, which in turn projects widely to prefrontal anterior thalamic nuclear complex via the postcommissural
cortex. An abnormal input to this system could have fornix, this could result in a re-routing of some subicular
profound effects on cerebral function. Separation of the efferents (normally destined for the entorhinal cortex) to
mediodorsal thalamus from the prefrontal cortex by section- join others in the fornix, thus contributing to its larger size
ing the anterior limb of the anterior capsule in frontal lobot- in early-onset schizophrenia.
omy, results in behavioural changes (a reduction in The current results showing a larger cross-sectional area
interpersonal behaviour, emotional blunting and socially of the body of the fornix in early-onset schizophrenia are of
inappropriate behaviour) that resemble the negative symp- importance because they are derived from a well-de�ned,
toms of schizophrenia. It is also conceivable that an relatively homogeneous group of subjects. A recent study of
increased number of hippocampal efferents in the fornix post-mortem material failed to �nd any difference between
may be driven by abnormalities in its target regions. Histo- adults with schizophrenia and an age-matched comparison
logical abnormalities in the septal nuclei [3], reduced size group, in the cross-sectional area of the anterior columns of
and neuronal number in the mediodorsal thalamic nucleus the fornix [10]. There are several possible reasons for this
166 D.C. Davies et al. / Neuroscience Letters 301 (2001) 163ą166
[10] Chance, S.A., Highley, J.R., Esiri, M.M. and Crow, T.J., Fiber
discrepancy: (1) early-onset and adult-onset schizophrenia
content of the fornix in schizophrenia: lack of evidence for a
are different, (2) the larger fornix in early-onset schizophre-
primary limbic encephalopathy, Am. J. Psychiatry, 156
nia does not persist into adulthood and (3) long-term drug
(1999) 1720ą1724.
treatment/substance abuse affects the size of the fornix. All
[11] Christison, G.W., Casanova, M.F., Weinberger, D.R., Rawl-
of these possibilities require investigation and reinforce the
ings, R. and Kleinman, J.E., A quantitative investigation of
hippocampal pyramidal cell size, shape, and variability of
need to study the structural abnormalities in the brains of
orientation in schizophrenia, Arch Gen. Psychiatry, 46
populations of subjects with schizophrenia over time. The
(1989) 1027ą1032.
structural basis of the larger fornix in early-onset schizo-
[12] Conrad, A.J., Abebe, T., Austin, R., Forsythe, S. and Schei-
phrenia also requires investigation as it is not known
bel, A.B., Hippocampal pyramidal cell disarray in Schizo-
whether it is due to a larger number/size of axons or to
phrenia as a bilateral phenomenon, Arch. Gen. Psychiatry,
48 (1991) 413ą417.
some other component of the fornix. Once the histological
[13] DeLisi, L.E., Hoff, A.L., Schwartz, J.E., Shields, G.W.,
nature of the larger fornix and the population in which it
Halthore, S.N., Gupta, S.M. and Henn, F.A., Brain morphol-
occurs are known, then the functional signi�cance of the
ogy in �rst-episode schizophrenic-like psychotic patients: a
larger fornix in schizophrenia can be investigated more
quantitative magnetic resonance imaging study, Biol.
effectively.
Psychiatry, 29 (1991) 159ą175.
[14] Falkai, P. and Bogerts, B., Cell loss in the hippocampus of
schizophrenics, Eur. Arch. Psychiatry Neurol. Sci., 236
We thank Jeremy Turk for constructive criticism of the
(1986) 154ą161.
manuscript, Peter Hill for support of the project, Martin
[15] Falkai, P. and Bogerts, B., Qualitative and quantitative
Graves and Maria Schmidt for expert assistance with the
assessment of pre-alpha-cell clusters in the entorhinal
analysis of MR images. This work was supported by a
cortex of schizophrenics. A neurodevelopmental model of
schizophrenia? Schizophrenia Res., 4 (1991) 357ą360.
grant from the Oxford Regional Health Authority.
[16] Falkai, P., Bogerts, B. and Rozmek, M., Limbic pathology in
schizophrenia: the entorhinal region-a morphometric
[1] American Psychiatric Association, Diagnostic and Statisti-
study, Biol. Psychiatry, 24 (1988) 515ą521.
cal Manual of Mental Disorders, 4th, American Psychiatric
[17] Fukuzako, H., Fukuzako, T., Hashiguchi, T., Hokazono, Y.,
Press, Washington, DC, 1994, pp. 285ą286.
Takeuchi, K., Hirakawa, K., Ueyama, K., Takigawa, M.,
[2] Arnold, S.E., Hyman, B.T., Van Hoesen, G.W. and Damasio,
Kajiya, Y., Nakajo, M. and Fujimoto, T., Reduction in hippo-
A.R., Some cytoarchitectural abnormalities of the entorh-
campal volume is caused mainly by its shortening in
inal cortex in schizophrenia, Arch. Gen. Psychiatry, 48
chronic schizophrenia: assessment by MRI, Biol. Psychia-
(1991) 625ą632.
try, 39 (1996) 938ą945.
[3] Averback, P., Structural lesions of the brain in young schi-
[18] Harrison, P.J., On the neuropathology of schizophrenia and
zophrenics, Can. J. Neurol. Sci., 8 (1981) 73ą76.
its dementia: neurodevelopmental, neurodegenerative, or
[4] Benes, F.M., Sorensen, I. and Bird, E.D., Reduced neuronal
both? Neurodegeneration, 4 (1995) 1ą12.
size in posterior hippocampus of schizophrenic patients,
[19] Jacob, W. and Beckmann, H., Prenatal disturbances in the
Schizophrenia Bull., 17 (1991) 597ą608.
limbic allocortex in schizophrenics, J. Neur. Trans., 65
[5] Bogerts, B., Meertz, E. and Schonfeldt-Bausch, R., Basal
�
(1986) 303ą326.
ganglia and limbic system pathology in schizophrenia. A
[20] Jeste, D.V. and Lohr, J.B., Hippocampal pathologic �ndings
morphometric study of brain volume and shrinkage, Arch.
in schizophrenia. A morphometric study, Arch. Gen.
Gen. Psychiatry, 42 (1985) 784ą791.
Psychiatry, 46 (1989) 1019ą1024.
[6] Bogerts, B., Falkai, P., Haupts, M., Greve, B., Ernst, S.,
[21] Kawasaki, Y., Maeda, Y., Urata, K., Higashima, M., Yama-
Tapernon-Franz, U. and Heinzmann, U., Post-mortem
guchi, N., Suzuki, M., Takashima, T. and Ide, Y., A quanti-
volume measurements of the limbic system and basal
tative magnetic resonance imaging study of patients with
ganglia structures in chronic schizophrenics. Initial results
schizophrenia, Eur. Arch. Psychiatry Clin. Neurosci., 242
from a new brain collection, Schizophrenia Res., 3 (1990)
(1993) 268ą272.
295ą301.
[22] Kovelman, J.A. and Scheibel, A.B., A neurohistological
[7] Bogerts, B., Lieberman, J.A., Ashtari, M., Bilder, R.M.,
correlate of schizophrenia, Biol. Psychiatry, 19 (1984)
DeGreef, G., Lerner, G., Johns, C. and Masiar, S., Hippo-
1601ą1621.
campus-amygdala volumes and psychopathology in
[23] Pakkenberg, B., Pronounced reduction of total neuron
chronic schizophrenia, Biol. Psychiatry, 33 (1993) 236ą245.
number in mediodorsal thalamic nucleus and nucleus
[8] Breier, A., Buchanan, R.W., Elkashef, A., Munson, R.C., Kirk-
accumbens in schizophrenics, Arch. Gen. Psychiatry, 47
patrick, B. and Gellad, F., Brain morphology in schizophre-
(1990) 1023ą1028.
nia. A magnetic resonance imaging study of limbic,
[24] Scheibel, A.B. and Kovelman, J.A., Disorientation of the
prefrontal cortex, and caudate structures, Arch. Gen.
hippocampal pyramidal cell and its processes in the schizo-
Psychiatry, 49 (1992) 921ą926.
phrenic patient, Biol. Psychiatry, 16 (1981) 101ą102.
[9] Brown, R., Colter, N., Corsellis, J.A.N., Crow, T.J., Frith,
[25] Seidman, L.J., Yurgelun-Todd, D., Kremen, W.S., Woods,
C.D., Jagoe, R., Johnstone, E.C. and Marsh, L., Post-
B.T., Goldstein, J.M., Faraone, S.V. and Tsuang, M.T., Rela-
mortem evidence of structural brain changes in schizophre-
tionship of prefrontal and temporal lobe MRI measures to
nia. Differences in brain weight, temporal horn area, and
neuropsychological performance in chronic schizophrenia,
parahippocampal gyrus compared with affective disorder,
Biol. Psychiatry, 35 (1994) 235ą246.
Arch. Gen. Psychiatry, 43 (1986) 36ą42.