The functional significance of perinatal corpus callosumdamage

Document Sample
The functional significance of perinatal corpus callosumdamage Powered By Docstoc
					                                                                                                  Brain (2002), 125, 1782±1792


The functional signi®cance of perinatal corpus
callosum damage: an fMRI study in young adults
A. M. Santhouse,1 D. H. ffytche,1 R. J. Howard,1 S. C. R. Williams,1 A. L. Stewart,2 M. Rooney,1
J. S. Wyatt,2 L. Rifkin1 and R. M. Murray1
1Instituteof Psychiatry, King's College and 2Perinatal             Correspondence to: A. M. Santhouse, Institute of
Brain Research Group, Department of Paediatrics,                   Psychiatry, De Crespigny Park, London SE5 8AF, UK
University College London Medical School, London, UK               E-mail: a.santhouse@iop.kcl.ac.uk

Summary
We used functional MRI (fMRI) to establish the func-               with damaged corpora callosa had signi®cantly different
tional signi®cance of corpus callosum damage in young              activation patterns compared with the two control
adults who had been born very preterm. Seven subjects              groups. In the visual task, additional activity was seen
from a cohort of individuals who had been born at <33              in the right dorsolateral prefrontal cortex of the dam-
weeks gestation and who had sustained callosal damage              aged callosum group, possibly because the task was
visualized on structural MRI were compared while they              accomplished by storing information in working mem-
carried out auditory and visual tasks requiring callosal           ory. On the auditory task, a de®cit of activity was seen
transfer with nine very preterm subjects with corpora              in the right temporal lobe of the callosum group. The
callosa of normal appearance on structural MRI, and                ®ndings reveal a plasticity of function compensating for
with seven full-term controls. The very preterm subjects           early damage to the corpus callosum.

Keywords: corpus callosum; cerebral plasticity; periventricular haemorrhage; preterm birth; fMRI


Introduction
Although the mortality rate among preterm and low birth            performance subscale of the WISC-R (Wechsler, 1974) at
weight infants has fallen sharply since the introduction of        age eight (Roth et al., 1993). In adolescence, preterm
neonatal intensive care (Nishida, 1993; Battin et al., 1998;       individuals have more neurological, adjustment and reading
Richardson et al., 1998), little is known about the effects of     impairments than controls born at term (Stewart et al., 1999)
preterm birth for long-term neuropsychological function.           and have poorer educational outcome (Botting et al., 1998).
   We previously have reported the results of structural MRI          In order to investigate the functional consequences of the
examination of the brains of 72 adolescents who had been           corpus callosum abnormalities, we examined the perform-
born before 33 weeks of gestation (Stewart et al., 1999). The      ance of preterm-born young adult males with corpus callosal
most common focal structural abnormality in these individ-         damage, preterm-born young adult males without corpus
uals was thinning/atrophy of the corpus callosum, particularly     callosum damage and term-born controls on visual and
posteriorly, which was noted in 43% of preterm individuals         auditory tasks requiring callosal transfer during functional
but in only 14% of age- and sex-matched controls who had           MRI (fMRI). The two sensory modalities allowed us to
been born at full term. The corpus callosum is known to be         examine different portions of the corpus callosum. Visual
especially vulnerable to adverse consequences of premature         ®bres are carried in the splenium of the corpus callosum, and
birth such as ischaemia, haemorrhage and sepsis, because of        auditory ®bres in the posterior third of the body of the corpus
its longer myelogenetic cycle (Valk and Njiokiktjien, 1991)        callosum (Berlucchi and Aglioti, 1999). Our strategy was to
and also because of its position adjacent to the periventricular   compare brain activity during the performance of two tasks
region, one of the most common sites for haemorrhage in the        within each sensory modality, one of which required the
preterm infant (Thorburn et al., 1981; Fawer et al., 1984).        callosal transfer of information and the other of which did not,
   These abnormalities of the corpus callosum may have             and to examine differences in the pattern of activation
functional signi®cance. For example, follow-up studies of          between the three participant groups. While the tasks
preterm infants have indicated that callosal abnormalities         themselves may have non-callosal components, their design
underlie poor performance on the Kaufman Assessment                ensured that callosal transfer facilitated successful comple-
Battery for Children (K-ABC) (Kaufman and Kaufman,                 tion of the task. In the visual domain, we compared the
1983), and on the simultaneous processing scale and                activity related to a bilateral task in which stimuli were
ã Guarantors of Brain 2002
                                                                                                   Corpus callosum function        1783

                                                                         1980. These individuals became part of a longitudinal study
                                                                         of development, and underwent structural MRI scanning at
                                                                         age 14 or 15 years (Stewart et al., 1999) and the scans were
                                                                         rated independently by two neuroradiologists. From the
                                                                         radiologists' reports, those scans showing thinning or atrophy
                                                                         of the corpus callosum and no other focal brain abnormalities
                                                                         were identi®ed (i.e. no severe generalized or lobar atrophy,
Fig. 1 Positioning of the simple shapes with respect to the central      CSF shunts or other focal damage). From these subjects,
crosswire. `x' is 4° of visual angle lateral to the crosswire; `y' and   seven male right-handed individuals between the ages of 18
`z' are 1° above and below the crosswire. Shapes appear for `a' in       and 20 years were invited to take part in the study. A further
position x,x; for `b' in position x (left),z; and for `c' in position    group of nine right-handed individuals from the cohort whose
y,z.
                                                                         MRI scan was reported normal and without callosal thinning
                                                                         at age 14 or 15 years were also recruited, together with a
presented across the vertical meridian with a unilateral task in         group of seven right-handed age-matched subjects who had
which stimuli were presented in one hemi®eld. In the bilateral           been born at full term.
task, each shape is presented to a different hemisphere, the                Handedness was assessed according to the Edinburgh
one in the right visual ®eld to the left hemisphere and the one          Handedness Inventory (Old®eld, 1971). Subjects also com-
in the left visual ®eld to the right hemisphere. Agenesis of the         pleted a National Adult Reading Test (NART) (Nelson,
corpus callosum leads to a performance de®cit in this task,              1982). Handedness data for one of the subjects in the preterm
suggesting that its normal completion requires a callosal                control group, and NART data for one of the subjects in the
transfer (Karnath et al, 1991). For the unilateral task, visual          term controls were not available. All subjects included had:
callosal transfer is not required since both shapes are                  (i) no contraindications to MRI; (ii) vision unimpaired
presented to a single hemisphere.                                        without glasses; (iii) no dif®culties in hearing; (iv) no
   In the auditory domain, we compared the activity related to           neurological illness; or (v) no history of substance misuse. All
a timbre discrimination task presented to the left or right ear.         subjects gave written informed consent and the study was
The rationale for this approach relates to the ®ndings that (i)          approved by the ethical committee of The Bethlem and
monaural stimulation activates predominantly the contral-                Maudsley Hospital.
ateral hemisphere through subcortical pathways (Woldorff
et al., 1999) and (ii) timbre discrimination is processed in the
right hemisphere (Sidtis, 1980; Samson and Zatorre, 1994;
Platel et al., 1997). We hypothesized that auditory stimuli              Experiment 1: visual paradigm
presented to the left ear would cross subcortically to the right         Subjects were instructed to focus on a central crosswire on a
hemisphere and undergo timbre discrimination processing                  projection screen throughout the experiment. Stimuli were
without callosal transfer. In contrast, auditory stimuli pre-            black outlines of geometric shapes (square, rectangle, rhom-
sented to the right ear cross subcortically to the left                  boid) projected tachistoscopically onto a screen with an
hemisphere, and then require a callosal transfer back to the             exposure time of 150 ms and subtending a visual angle of 1°.
right hemisphere for timbre discrimination processing.                   Subjects were instructed to respond as to whether pairs of
   We predicted that the group with callosal thinning would              shapes were the `same' or `different', using a button press
have poorer performance on the bilateral visual task and the             with their right hand.
right ear auditory task, compared with the unilateral visual                Pairs of shapes were presented in blocks of six in 30 s
and left ear auditory tasks, and that this would be re¯ected in a        scanning epochs under the following conditions: (i) bilat-
different cortical activation pattern compared with the                  erally, 4° to either side of the central crosswire; (ii)
preterm controls with normal callosal appearance and term-               unilaterally, one located centrally 1° below the crosswire
born controls. We also predicted that we would not ®nd any               and the other 4° to the left of the central crosswire; and (iii)
differences between the group of term-born controls and the              centrally, 1° above and 1° below the crosswire (Fig. 1). The
preterm normal callosum group. The pattern of group-speci®c              central stimulus was included as part of a separate study to
differences would help elucidate the neurobiological mech-               investigate the bilateral ®eld advantage and will not be
anisms compensating for perinatal damage to the corpus                   discussed further here (Santhouse et al., 2002).
callosum.                                                                   Each block was presented four times in a pseudorandom
                                                                         order during the 6 min fMRI experiment so that a total of 72
                                                                         pairs of stimuli were seen altogether in the 12 blocks.
Methods                                                                  Response times were measured from the onset of presentation
Subjects were recruited from a cohort of infants who had been            of the stimuli, and a log was kept of response accuracy.
born before 33 weeks gestation and admitted to University                Subjects practised the paradigm for 6 min before they entered
College Hospital London Neonatal Unit between 1979 and                   the scanner.
1784      A. M. Santhouse et al.




Fig. 2 The sound stimuli consisted of the fundamental frequency of 440 Hz plus the ®rst two harmonics (top); the fundamental frequency
and three harmonics (middle); or the fundamental frequency and four harmonics (bottom).


Experiment 2: auditory paradigm                                       presented to the right ear; and (ii) presented to the left
Subjects were asked to close their eyes and judge as to               ear. Each block was presented alternately ®ve times
whether pairs of sounds were the `same' or `different'.               throughout the 5 min experiment, a total of 50 compari-
Stimuli were notes at 440 Hz, generated using Cool Edit 96            sons over the course of 5 min. Response times were
(Syntrillium Software Corporation, Ariz., USA) which                  measured from the onset of presentation of the stimuli,
differed only in their timbre (Fig. 2). Each sound lasted for         and response accuracy was measured. Subjects performed
750 ms and was separated from its paired sound by a gap of            the experiment (i) with the button press in their left hand
1000 ms.                                                              and (ii) with the button press in their right hand. Subjects
   The stimuli were presented in blocks of ®ve in 30 s                practised the paradigm for 5 min before entering the
scanning epochs under the following conditions: (i)                   scanner.
                                                                                                    Corpus callosum function           1785




Fig. 3 Structural differences between callosum and control groups. To the left of the ®gure are the areas of signi®cant differences (in red)
in white matter between callosum and control groups superimposed on a mean structural image. To the right is shown the grey matter and
CSF differences between the callosum and control groups. The graphs show the voxel intensity values in the anterior callosum (left panel)
and left superior temporal gyrus (right panel) for each subject in each of the three groups.


Analyses                                                                between-subject factor, group, with three levels. Multivariate
Behavioural data                                                        statistics were used to test signi®cance.
The data were analysed by means of SPSS for Windows
(version 8.0.2). For accuracy data, a single score of the
percentage of correct responses was included for each                   Imaging
subject. Separate analyses were conducted on accuracy and               Scan parameters
response times. For visual experiments, a two-way repeated              Functional images were acquired on a 1.5 Ta GE Neuro-
measures analysis of variance (ANOVA) was carried out with              optimized Signa LX Horizon System (General Electric,
®eld (bilateral and unilateral) as a within-subject factor and          Milwaukee, Wisc., USA), using a gradient echo planar
group as a between-subject factor, with three levels (term              sequence sensitive to blood oxygenation level-dependent
controls, preterm controls and damaged corpus callosum).                (BOLD) contrast (TR, repetition time = 3 s; TE, echo
   For auditory experiments, three-way repeated measures                time = 40 ms; ¯ip angle 90°; 64 Q 64 matrix; in-plane voxel
ANOVAs were performed, with two within-subject factors                  size 3.75 Q 3.75 mm) and 20 axial slices, 7 mm thick with a
(ear and hand) each with two levels (left or right); and one            0.7 mm interslice gap
1786       A. M. Santhouse et al.




Fig. 4 Bar chart showing response accuracy in both visual (A) and auditory (B) tasks. For visual tasks, the response accuracy is for
bilateral and unilateral tasks; for auditory accuracy, the response is for all tasks. The error bars show the standard error of the mean.

Image analysis                                                            preterm controls; and damaged callosum group versus
Structural images. High resolution sagittal images were                   preterm controls. Within-group comparisons were made for
acquired for all of the term control group, ®ve of the preterm            the visual and auditory tasks for each of the subject groups
control group and ®ve of the damaged corpus callosum group.               using a ®xed effects model, to show activations in the groups
Data for the other subjects were unavailable. For each                    separately.
group, the images were normalized using SPM99 (http://                      For the structural comparisons, separate ANOVAs were
www.®l.ion.ucl.ac.uk/spm) and a mean image generated, to                  generated for white matter, grey matter and CSF images. All
help localization of activation maxima. The normalized                    random effects structural and functional statistical parametric
structural image from each subject was segmented into white               maps were thresholded at P < 0.01, corrected for multiple
matter, grey matter and CSF images.                                       comparisons at the cluster level. Within-group ®xed effects
   Functional images. Auditory and visual experiments were                models were thresholded at P < 0.001, corrected for multiple
analysed separately. For each subject, the time series was                comparisons at the cluster level.
motion corrected (Friston et al., 1996), transformed into
standard stereotaxic space (Talairach and Tournoux, 1988)
smoothed with a 10 mm FWHM (full width half maximum)                      Results
Gaussian ®lter and high pass ®ltered using SPM99.                         There was no signi®cant difference in handedness between
Covariates were modelled with a boxcar convolved with the                 the full-term controls, preterm controls and damaged
haemodynamic response function.                                           callosum group. Mean laterality quotient scores on the
                                                                          Edinburgh Handedness Scale were, respectively, 89.7 (SD
                                                                          12.2), 89.1 (SD 13.8) and 76.3 (SD 28.7) [F(2,19) = 1.08,
Statistical inferences                                                    P = 0.358]. There were no signi®cance differences in
In order to compare the generic activations associated with               intelligence in the three groups, as measured by the NART.
each of the three groups, a two-stage random effects analysis             Mean IQ scores for the three groups were, respectively, for
was used for each hypothesis tested (Friston et al., 1999). The           the full-term controls, preterm controls and damaged
®rst stage generated subject-speci®c contrast images from the             callosum group 114 (SD 4), 109 (SD 9.5) and 104 (SD
weighted linear sum of covariate parameter estimates. For the             10.7) [F(2,19) = 2.16, P = 0.14)].
visual task, the contrast images were for bilateral versus the
unilateral stimulation, while for the auditory task the contrast
images were for left ear versus right ear stimulation. The                Structural image comparisons
second stage assessed the differences in generic activations              Statistical comparison of the white matter images in the
for each group with pairwise t test comparisons: term controls            damaged callosum group and control groups showed a
versus damaged callosum group; term controls versus                       signi®cant loss of callosal ®bres in the splenium, anterior
                                                                                               Corpus callosum function         1787

                                                                     for one of the subjects in the preterm control group. For visual
                                                                     response times, the two-way ANOVA showed a signi®cant
                                                                     main effect of group [F(2,539) = 12.6, P < 0.001], with the
                                                                     damaged callosum group signi®cantly slower on both tasks.
                                                                     Although not reaching signi®cance, Fig. 5A shows a pattern
                                                                     of response times in the damaged callosum group different
                                                                     from that of the control groups. Both preterm and term
                                                                     controls show a bilateral ®eld advantage, with the bilateral
                                                                     comparison performed faster than the unilateral one. In
                                                                     contrast, the damaged callosum group shows a bilateral ®eld
                                                                     disadvantage, with the response time to the bilateral stimulus
                                                                     being slower than that to the unilateral stimulus
                                                                     [F(2,529) = 1.2, Hotelling's trace = 0.004, P = 0.3] (Fig. 5A).
                                                                        The auditory task accuracy data showed a signi®cant group
                                                                     effect. The damaged callosum group were least accurate, with
                                                                     mean and SDs for the damaged callosum group, preterm
                                                                     controls and normal controls, respectively, being 17.8 (3.9),
                                                                     19.3 (3.8) and 20.6 (1.7) [F(2, 19) = 3.85, P = 0.039] (Fig. 4B).
                                                                     Data were not available for one of the damaged callosum
                                                                     group.
                                                                        As in the visual data, we found a signi®cant group effect in
                                                                     the response time data, with the damaged callosum group
                                                                     signi®cantly slower than the term and preterm controls
                                                                     [F(2,542) = 80, P < 0.001]. There was a signi®cant ear by
                                                                     group interaction, with the right ear response times slower
                                                                     than the left ear response times in the damaged callosum
                                                                     group, but no left ear±right ear differences in the control
                                                                     groups [F(2,542) = 3.02, Hotelling's trace = 0.011, P = 0.05)
                                                                     (Fig. 5B).
                                                                        There was also a signi®cant effect of hand, independent of
                                                                     group or ear. The direction of the effect (right hand 100 ms
                                                                     faster than left) was in the opposite direction to the expected
                                                                     superiority of left hand over right for the right hemisphere
                                                                     timbre discrimination task. It was also much longer than
                                                                     previous estimates of callosal transfer time (i.e. Clarke and
                                                                     Zaidel, 1989). We therefore concluded that the hand effect
                                                                     identi®ed was not due to callosal transfer but seemed instead
                                                                     to relate to hand dominance or task practice, as all subjects
Fig. 5 Bar charts showing mean reaction times for bilateral and      performed the left hand response experiments before the right
unilateral tasks in each of the three groups (A) and mean reaction   hand response experiments. We therefore ignored the hand
times for right and left ears in the auditory task (B).
                                                                     effect and pooled left hand and right hand experimental data
                                                                     in our fMRI analysis.
corpus callosum, genu, forceps major and forceps minor (see
Fig. 3). We also found signi®cant atrophy of the superior
temporal gyrus in both hemispheres in the callosal group, as         Differences in BOLD activation pattern between
evidenced by a loss of grey matter on the left and increase of       groups
CSF on the right. There were no signi®cant increases in white
                                                                     Visual task (bilateral presentation >unilateral
matter or grey matter in the damaged callosum group
compared with the controls.                                          presentation)
                                                                     Bilateral stimuli did not lead to signi®cantly more activation
                                                                     than unilateral stimuli in either of the control groups. In
                                                                     contrast, the same comparison led to a signi®cant difference
Performance data                                                     in activity in the right dorsolateral prefrontal cortex (BA 9/10)
There were no signi®cant differences in accuracy between the         of the damaged callosum group (Fig. 6). This difference
three subject groups in the visual tasks [F(2,19) = 0.032,           between the groups was signi®cant when we compared the
P = 0.96] (Fig. 4A). For technical reasons, data were missing        damaged callosum group and term controls (Z = 3.9; P = 0.04)
1788       A. M. Santhouse et al.

                                                                       showed activation of the right superior temporal gyrus (BA
                                                                       22). The activation in the damaged callosum group was
                                                                       signi®cantly greater than in either the term or preterm control
                                                                       groups (Z = 3.83; P = 0.05 full-term control; Z = 4.65; P = 0.04
                                                                       preterm control) (Fig. 7).
                                                                          Signi®cantly more activity was also seen in the right
                                                                       precentral gyrus (Z = 4.32, P < 0.01) and left precentral gyrus
                                                                       (Z = 3.92; P = 0.04) of the damaged callosum group compared
                                                                       with the term control group. The same regions were active in
                                                                       the comparison of the preterm control group and the damaged
                                                                       callosum group using an a priori region of interest approach
                                                                       (search volume 0.5 cm sphere at x, y, z coordinates 44, ±14,
                                                                       58, Z = 4.24, P < 0.01 for right precentral gyrus; left
                                                                       precentral gyrus search volume 0.5 cm sphere at x, y, z
                                                                       coordinates ±32, ±12, 66, Z = 2.44, P = 0.04) (Table 2). No
                                                                       signi®cant differences were seen between the full-term and
                                                                       preterm control groups, and no brain regions were signi®-
                                                                       cantly more active in the two controls groups than in the
                                                                       damaged callosum group.


                                                                       Discussion
                                                                       Our results show that preterm callosal damage affects
                                                                       behavioural performance and functional cerebral anatomy
                                                                       in early adulthood. The damaged callosum patients in our
                                                                       study could complete timbre and bilateral ®eld comparison
                                                                       tasks, although at a reduced level of performance. One
                                                                       interpretation of the residual ability is that their callosal
                                                                       damage was insuf®cient to cause a performance de®cit and
                                                                       that tasks were completed using the same neural mechanisms
                                                                       as found in normal subjects. However, the fact that we found
Fig. 6 Changes in BOLD contrast in comparison of `bilateral'           signi®cant differences in the task-related pattern of activity
versus `unilateral'. One pair of `glass brain' views (from the side    for the damaged callosum and control groups suggests that
and from above) is shown for each patient group. All the images        this was not the case. An alternative interpretation of the
have been thresholded at P < 0.001, corrected for multiple             ®ndings is that compromised callosal function has led to
comparisons at the cluster level. Only the damaged corpus
                                                                       alternative neural strategies to compensate for the perinatal
callosum group shows a difference between the two conditions,
with an area of signi®cant activation in the right dorsolateral        injury. In what follows, we examine the neural basis of this
prefrontal cortex. The difference between activation patterns in the   functional plasticity.
damaged callosum group and term controls is signi®cant and is
shown rendered onto a single brain.
                                                                       Callosal-speci®c performance de®cits
                                                                       In the visual task, the two control groups showed the normal
(Fig. 6). The same region was also signi®cant in the damaged           pattern of behavioural responses, with reaction times for
callosum±preterm control comparison using an a priori                  bilateral comparisons being faster than those for unilateral
region of interest approach (search volume 0.5 cm sphere at            comparisons (i.e. Davis and Schmit, 1971; Dimond and
x, y, z coordinates 38, 54, 24; Z = 2.8; P = 0.03) (Table 1). No       Beaumont, 1971; Merola and Liederman, 1990; Norman
regions were signi®cantly more active in the control groups            et al., 1992). In contrast, the damaged callosum group showed
than the damaged callosum group, and there were no                     the opposite pattern, with reaction times for unilateral
signi®cant differences between the control groups.                     comparisons being faster than those for bilateral ones. Our
                                                                       previous study of the cerebral activity underlying the bilateral
                                                                       ®eld advantage suggested that bilateral and unilateral com-
Auditory task (left ear presentation > right ear                       parisons were carried out by different processing mechan-
presentation)                                                          isms, the bilateral comparison requiring a callosal transfer
There were no signi®cant activations for the left ear > right          and the unilateral comparison requiring working memory
ear comparison in the term control group. In contrast, the             resources. The results presented above support this hypoth-
preterm control group and the damaged callosum group                   esis by revealing a bilateral disadvantage in subjects with a
                                                                                                     Corpus callosum function           1789

Table 1 The regions of differential group activation for the comparison bilateral > unilateral for the visual callosal
transfer task
Areas activated                                 x           y          z           Cluster size         Z score         P (corr)         BA

Callosum group > term controls
  Right dorsolateral prefrontal cortex          38          54         26          626                  3.9             0.04             9/10
Callosum group > preterm controls
  Right dorsolateral prefrontal cortex*         38          54         24           31                  2.8             0.03             9/10

The table shows x,y,z coordinates, cluster size and Brodmann area (BA) of the most signi®cant voxel in each cluster. Also shown are the
Z values and corrected P values. *Region examined with a priori hypothesis. For term and preterm control groups > callosum group and
term > preterm control, preterm > term control, there were no differences in activation.




Fig. 7 BOLD contrast in the comparison of the left versus right ear. One pair of `glass brain' views (from the side and from above) is
shown for each patient group for the left ear greater than right ear comparison (thresholding as for Fig. 6). The damaged callosum group
shows a signi®cant area of activation in the right superior temporal gyrus. This is signi®cantly different from the term control group (left
panel) and preterm control group (right panel), shown rendered onto an individual brain.




damaged corpus callosum. Our fMRI results reveal how the                   cortex in an area that previously has been associated with
damaged callosum group accomplishes the task. While                        working memory (Braver et al., 1997; Cohen et al., 1997;
normal subjects show no activity for the bilateral > unilateral            Nystrom et al., 2000; Stern et al., 2000). One possible
comparison (Fig. 6) (Santhouse et al., 2002), the damaged                  explanation for the delay in reaction time for the bilateral
corpus callosum group activates the dorsolateral prefrontal                comparison in the damaged callosum group is that shapes
1790      A. M. Santhouse et al.

Table 2 Regions of differential group activation for the timbre comparison of left ear versus right ear
Area activated                             x            y            z            Cluster size        Z score         P (corr)        BA

Callosum group > term controls
  Right superior temporal gyrus             60          ±32          16           618                 3.83             0.05           22
  Right precentral gyrus                    44          ±16          56           973                 4.32            <0.01            4
  Left precentral gyrus                    ±34          ±8           68           644                 3.92             0.04            4/6
Callosum group > preterm controls
  Right superior temporalgyrus              64          ±32          22           646                 4.65             0.04           22
  Right precentral gyrus*                   44          ±14          58            80                 4.24            <0.01
  Left precentral gyrus*                   ±32          ±12          66            24                 2.44             0.04

The table shows x,y,z coordinates, cluster size and Brodmann area (BA) of the most signi®cant voxel in each cluster. Also shown are the
Z values and corrected P values. *Region examined with a priori hypothesis. For preterm controls > term, term controls > preterm,
preterm controls > callosum and term controls > callosum, there were no differences in activation.

presented for comparison across a damaged corpus callosum                 the right ear task, even though their performance is impaired.
are held `on line' in working memory, to compensate for the               One possibility is that they use the left hemisphere, and partial
impaired callosal transfer.                                               support for this comes from the fact that we found, using an a
   In normal individuals, the discrimination of timbre takes              priori region of interest approach, signi®cantly greater
place largely in the right hemisphere (Sidtis, 1980; Samson               activation of the left superior temporal gyrus of the damaged
and Zatorre, 1994; Platel et al., 1997). The speci®c areas                corpus callosum group than the term control group for the
identi®ed in previous imaging studies include the superior                right ear greater than left ear comparison (search volume
and posterior temporal regions (Mazziotta et al., 1982) and               0.5 cm sphere at x, y, z coordinates ±60, ±32, 16; BA 22;
the superior, middle frontal and precentral gyri (Platel et al.,          Z = 3.0; P = 0.02).
1997). In theory, the right ear response time should be a few
milliseconds slower than the left ear response times, as the
right ear presentation requires a callosal transfer of signals for        Non-speci®c performance de®cits
timbre discrimination (right ear®left hemisphere®right                    We found the damaged callosum group to be slower on both
hemisphere). We did not ®nd a signi®cant difference in left               the visual and auditory tasks, regardless of whether they
and right ear reaction times in our two control groups as the             involved a callosal transfer. For the visual data, the longer
number of trials presented in an fMRI experiment are                      response times enabled the callosal group to perform with
insuf®cient to show such a small effect. In contrast, the                 accuracy equivalent to the two control groups. For the
damaged callosum group did show a signi®cant difference in                auditory data, the damaged callosum group were signi®cantly
response times to left and right ear stimuli, with the right ear          less accurate [F(2,19) = 3.85, P = 0.039] despite the delay in
response, which in normal subjects requires a callosal                    response. The ®ndings suggest that the callosal damage on
transfer, being signi®cantly slower than the left ear response.           MRI scans at age 14±15 years is associated with other more
The pattern of brain activity elicited by the auditory stimuli            subtle brain abnormalities not apparent on the structural
provides an explanation for the behavioural effect. In control            images. Another possibility is that the atrophy found in the
subjects, the right superior temporal gyrus would process                 anterior corpus callosum compromised the transfer of motor
timbre stimuli regardless of the ear of presentation, either by           signals, introducing a delay in response time. We also found
direct stimulation (left ear to right hemisphere) or through a            atrophy of the superior temporal gyrus bilaterally in the
callosal transfer (right ear®left hemisphere®corpus callo-                damaged callosum group. While this could have led to a non-
sum®right hemisphere). The result is that there is no                     speci®c de®cit in the auditory task, it could not explain the
signi®cant difference in activation of the region between                 non-speci®c de®cit found in the visual task. It is unclear why
left ear and right ear stimulation. In contrast, the damaged              this particular region is vulnerable; however, damage here
callosum group would be unable to perform an adequate                     raises an interesting possibility that it underlies language
callosal transfer from left hemisphere to the right, leading to           impairments found in preterm individuals during adolescence
an activation of the right superior temporal gyrus for left ear           (Stewart et al., 1999)
stimulation, but not for right ear stimulation (Fig. 7). We
hypothesize that it is the inability to access the timbre
discrimination area which underlies the slowing in reaction               Methodological issues
time for right ear stimulation. We also found additional left             Our structural comparisons of grey matter, white matter and
and right precentral gyrus activity in the damaged callosum               CSF images required normalization of each individual's brain
group compared with the term controls. The right precentral               to a standard template. The procedure will therefore correct
gyrus has been associated with timbre processing in previous              any overall differences in brain size between subject groups.
studies (Platel et al., 1997). An interesting question arises as          However, the warping procedure does not attempt to match
to how the damaged callosum group are still able to perform               individual anatomical structures (the ventricles or speci®c
                                                                                               Corpus callosum function         1791

gyri, for example), with the result that group differences at       callosum necessitates adoption of a different neural strategy
this anatomical level remain.                                       for forced callosal transfer tasks. The exact nature of the
   The overall performance de®cit in the damaged callosum           change is dependent on the modality of the pathways
group confounds the interpretation of the fMRI data. An             affected.
alternative interpretation of the activations is that they relate
to task dif®culty and not to differences in functional anatomy.
To protect against this possibility, our analytical strategy was
to compare differences between pairs of stimuli (e.g. bilateral     Acknowledgements
versus unilateral) across groups rather than single stimuli.        We wish to thank the subjects and the controls for giving their
   The fMRI environment is uncomfortable, has distracting           time to the study, Jan Townsend at University College
sounds and subjects are required to lie horizontally with their     London Department of Paediatrics for coordinating follow-up
movements restricted. These stimulus conditions are different       of these subjects, and Chris Andrew for technical assistance
from those of the psychophysical laboratories in which              with computer programming. This work was supported by the
reaction time experiments would normally take place.                Medical Research Council G9821480. D.H.ff. is a Wellcome
However, we do not think that the fMRI environment                  Trust Clinician Scientist Fellow.
explains the non-speci®c de®cits in the damaged callosum
group, as the other two control groups were subject to the
identical environment.                                              References
                                                                    Battin M, Ling EW, Whit®eld MF, Mackinnon M, Effer SB. Has
                                                                    the outcome for extremely low gestational age (ELGA) infants
Plasticity                                                          improved following recent advances in neonatal intensive care? Am
Functional compensation for brain injury is well described,         J Perinatol 1998; 15: 469±77.
with the greatest potential for plasticity seen in neonates         Berlucchi G, Aglioti S. Interhemispheric disconnection syndromes.
(Bower, 1990). Early animal experiments demonstrated                In: Denes C, Pizzamiglio L, editors. Handbook of clinical and
functional reorganization of the visual cortex after arti®cially    experimental neuropsychology. Hove (UK): Psychology Press;
induced visual impairment (i.e. Cynader and Mitchell, 1977;         1999. p. 635±70.
Cynader et al., 1981). Further studies showed that this             Botting N, Powls A, Cooke RW, Marlow N. Cognitive and
functional reorganization is dependent on critical time             educational outcome of very-low-birthweight children in early
periods during development (Hubel and Weisel, 1970; Ptito           adolescence. Dev Med Child Neurol 1998; 40: 652±60.
and Lepore, 1983; Sherrard et al., 1986). Similarly, in
                                                                    Bower AJ. Plasticity in the adult and neonatal central nervous
humans, the time of injury is crucial to outcomeÐthe earlier
                                                                    system. [Review]. Br J Neurosurg 1990; 4: 253±64.
the lesion is sustained, the greater the potential for improve-
ment (Lassonde and Sauerwein, 1997). Perinatal injury to the        Braver TS, Cohen JD, Nystrom LE, Jonides J, Smith EE, Noll DC.
posterior corpus callosum, one of the commonest brain               A parametric study of prefrontal cortex involvement in human
abnormalities found in individuals born very preterm                working memory. Neuroimage 1997; 5: 49±62.
(Stewart et al., 1999), occurs at a time where there is             Clarke JM, Zaidel E. Simple reaction times to lateralized light
signi®cant potential for functional reorganization of the           ¯ashes. Brain 1989; 112: 849±70.
visual and auditory ®bres that pass through the posterior
                                                                    Cohen JD, Perlstein WM, Braver TS, Nystrom LE, Noll DC,
callosum. There is evidence that the compensatory mechan-           Jonides J, et al. Temporal dynamics of brain activation during a
isms activated after damage may vary according to the               working memory task. Nature 1997; 386: 604±8.
sensory modality involved (Lassonde et al., 1990).
   In our experiment, there appeared to be different categories     Cynader M, Lepore F, Guillemot JP. Inter-hemispheric competition
                                                                    during postnatal development. Nature 1981; 290: 139±40.
of neural compensation following early callosal damage. In
the auditory system, the fMRI evidence points to a left             Cynader M, Mitchell DE. Monocular astigmatism effects on kitten
hemisphere or bilateral representation of timbre discrimin-         visual cortex development. Nature 1977; 270: 177±8.
ation, rather than the right hemisphere specialization found in     Davis R, Schmit V. Timing the transfer of information between
normals. In contrast, the activation of the dorsolateral            hemispheres in man. Acta Psychol (Amst) 1971; 35: 335±46.
prefrontal cortex in the visual experiment in the damaged
                                                                    Dimond S, Beaumont G. Use of two cerebral hemispheres to
callosal group implies the recruitment of working memory to
                                                                    increase brain capacity. Nature 1971; 232: 270±1.
compensate for impaired callosal function.
                                                                    Fawer CL, Calame A, Anderegg A. Real-time ultrasonography in
                                                                    the neonate: a systematic study of a high-risk infants population.
Conclusion                                                          Helv Paediatr Acta 1984; 39: 34±45.
In summary, we have shown that both the structure and               Friston KJ, Williams S, Howard R, Frackowiak RS, Turner R.
function of the adult brain are in¯uenced by events occurring       Movement-related effects in fMRI time-series. Magn Reson Med
perinatally. The early damage sustained by the corpus               1996; 35: 346±55.
1792       A. M. Santhouse et al.

Friston KJ, Holmes AP, Worsley KJ. How many subjects constitute         Roth SC, Baudin J, McCormick DC, Edwards AD, Townsend J,
a study? Neuroimage 1999; 10: 1±5.                                      Stewart AL, et al. Relation between ultrasound appearance of the
                                                                        brain of very preterm infants and neurodevelopmental impairment
Hubel DH, Weisel TN. The period of susceptibility to the
                                                                        at eight years. Dev Med Child Neurol 1993; 35: 755±68.
physiological effects of unilateral eye closure in kittens. J Physiol
(Lond) 1970; 206: 419±36.                                               Samson S, Zatorre RJ. Contribution of the right temporal lobe to
                                                                        musical timbre discrimination. Neuropsychologia 1994; 32: 231±
Karnath HO, Schumacher M, Wallesch CW. Limitations of
                                                                        40.
interhemispheric extracallosal transfer of visual information in
callosal agenesis. Cortex 1991; 27: 345±50.                             Santhouse AM, ffytche DH, Howard RJ, Williams SCR, Rifkin L,
Kaufman AS, Kaufman NL. K-ABC. Kaufman Assessment Battery               Murray RM. Functional imaging of the mechanisms underlying the
for Children. Circle Pines (MN): American Guidance Service; 1983.       bilateral ®eld advantage. NeuroImage 2002; in press.

Lassonde M, Sauerwein C. Neuropsychological outcome of corpus           Sherrard RM, Bower AJ, Payne JN. Innervation of the adult rat
callosotomy in children and adolescents. J Neurosurg Sci 1997; 41:      cerebellar hemisphere by ®bres from the ipsilateral inferior olive
67±73.                                                                  following unilateral neonatal pedunculotomy: an autoradiographic
                                                                        and retrograde ¯uorescent double-labelling study. Exp Brain Res
Lassonde M, Ptito M, Lepore F. Plasticity of the callosal system.       1986; 62: 411±21.
[French]. Can J Psychol 1990; 44: 166±79.
                                                                        Sidtis JJ. On the nature of the cortical function underlying right
Mazziotta JC, Phelps ME, Carson RE, Kuhl DE. Tomographic                hemisphere auditory perception. Neuropsychologia 1980; 18: 321±
mapping of human cerebral metabolism: auditory stimulation.             30.
Neurology 1982; 32: 921±37.
                                                                        Stern CE, Owen AM, Tracey I, Look RB, Rosen BR, Petrides M.
Merola JL, Liederman J. The effect of task dif®culty upon the           Activity in ventrolateral and mid-dorsolateral prefrontal cortex
extent to which performance bene®ts from between-hemisphere             during nonspatial visual working memory processing: evidence
division of inputs. Int J Neurosci 1990; 51: 35±44.                     from functional magnetic resonance imaging. Neuroimage 2000;
Nelson HE. National adult reading test (NART): Test Manual.             11: 392±9.
Windsor (UK): NFER-Nelson; 1982.                                        Stewart AL, Rifkin L, Amess AP, Kirkbride V, Townsend JP,
Nishida H. Outcome of infants born preterm, with special emphasis       Miller DH, et al. Brain structure and neurocognitive and
on extremely low birthweight infants. [Review]. Baillieres Clin         behavioural function in adolescents who were born very preterm.
Obstet Gynaecol 1993; 7: 611±31.                                        Lancet 1999; 353: 1653±7.
Norman WD, Jeeves MA, Milne A, Ludwig T. Hemispheric                    Talairach J, Tournoux P. Co-planar stereotaxic atlas of the human
interactions: the bilateral advantage and task dif®culty. Cortex        brain. Stuttgart: Thieme; 1988.
1992; 28: 623±42.
                                                                        Thorburn RJ, Lipscomb AP, Stewart AL, Reynolds EO, Hope PL,
Nystrom LE, Braver TS, Sabb FW, Delgado MR, Noll DC, Cohen              Pape KE. Prediction of death and major handicap in very preterm
JD. Working memory for letters, shapes, and locations: fMRI             infants by brain ultrasound. Lancet 1981; 1: 1119±21.
evidence against stimulus-based regional organization in human
                                                                        Valk J, Njiokiktjien C. Abnormal morphogenesis of the corpus
prefrontal cortex. Neuroimage 2000; 11: 424±46.
                                                                        callosum 1: MR imaging. In: Ramaekers G, Njiokiktjien C, editors.
Old®eld RC. The assessment and analysis of handedness: the              The child's corpus callosum. Amsterdam: Suyi; 1991. p. 266±309.
Edinburgh inventory. Neuropsychologia 1971; 9: 97±113.
                                                                        Wechsler D. Wechsler Intelligence Scale for ChildrenÐrevised.
Platel H, Price C, Baron JC, Wise R, Lambert J, Frackowiak RS,          New York: Psychological Corporation; 1974.
et al. The structural components of music perception. A functional
anatomical study. Brain 1997; 120: 229±43.                              Woldorff MG, Tempelmann C, Fell J, Tegeler C, Gaschler-
                                                                        Markefski B, Hinrichs H, et al. Lateralized auditory spatial
Ptito M, Lepore F. Interocular transfer in cats with early callosal     perception and the contralaterality of cortical processing as
transsection. Nature 1983; 301: 513±5.                                  studied with functional magnetic resonance imaging and
Richardson DK, Gray JE, Gortmaker SL, Goldmann DA, Pursley              magnetoencephalography. Hum Brain Mapp 1999; 7: 49±66.
DM, McCormick MC. Declining severity adjusted mortality:
evidence of improving neonatal intensive care. Pediatrics 1998;         Received July 19, 2001. Revised January 25, 2002.
102: 893±9.                                                             Accepted February 14, 2002