-
19/8/2014 Ovid: Cerebral lateralization of language in normal
left-handed people studied by functional MRI.
http://0-ovidsp.tx.ovid.com.jabega.uma.es/sp-3.12.0b/ovidweb.cgi
1/10
[Articles]
Neurology
Nmero: Volume 52(5), 23 March 1999, pp 1038-1043
Copyright: 1999 American Academy of Neurology
Tipo de publicacin: [Articles]
ISSN: 0028-3878
Registro: 00006114-199903230-00027
Cerebral lateralization of language in normal left-handed people
studied by functional MRI
Pujol, Jess MD; Deus, Joan PhD; Losilla, Josep M. PhD;
Capdevila, Antoni MD
Informacin sobre el autorFrom the Magnetic Resonance Center of
Pedralbes (Drs. Pujol, Deus, and Capdevila), Barcelona; and the
Department of Psychobiology and Methodology (Dr. Losilla),
Autonomous University of Barcelona, Spain.
Received September 11, 1998. Accepted in final December 12,
1998.
Address correspondence and reprint requests to Dr. Jess Pujol
Magnetic Resonance Center of Pedralbes, Monestir, 3, 08034
Barcelona, Spain.
Article abstract
Objective: To use functional MRI (fMRI) to further define the
occurrence of left-hemisphere, bilateral, and
right-hemisphere language in a normal left-handed
population.
Methods: A total of 100 healthy volunteers, consisting of 50
left-handed subjects and a reference group of 50
right-handed subjects, were studied by fMRI of the frontal
cortex during silent word generation.
Results: Ninety-six percent of right-handed subjects showed fMRI
changes lateralized to the left hemisphere,
whereas 4% showed a bilateral activation pattern. In contrast,
left-hemisphere lateralization occurred in 76% of
left-handers, bilateral activation in 14%, and right-hemisphere
lateralization in the remaining 10%. The
predominance of right-hemisphere activation, however, was weak
in these cases; only a single left-handed subject
(2%) showed complete right-hemisphere lateralization.
Conclusions: Silent word generation lateralizes to the left
cerebral hemisphere in both handedness groups,
but right-hemisphere participation is frequent in normal
left-handed subjects. Exclusive right-hemisphere
activation rarely occurred in the frontal lobe region
studied.
Cerebral lateralization of language is not as well defined for
left-handed people as it is for their right-handed
counterparts.1 The intracarotid amobarbital (Wada) test had
contributed to establishing the occurrence of left-
hemisphere, bilateral, and right-hemisphere language in
left-handers.2-6 However, this procedure has been
restricted to specific patients owing to its invasiveness. These
patients showed a higher probability of abnormal
language organization due to early brain injuries and frequent
abnormal cerebral development.6 Therefore,
results from the Wada studies are only partially representative
of the normal left-handed population.
Functional MRI (fMRI) provides the opportunity to further assess
the cerebral organization of language in
normal individuals noninvasively. Studies already conducted in
right-handed subjects suggest that fMRI can
accurately estimate language lateralization if adequate
strategies are adopted.7-15 Nevertheless, fMRI
observations in the normal left-handed population are still
preliminary.9,13,14
Intrinsic verbal production has proved to be more efficient to
lateralize language than receptive language
tasks.7 Lateralized changes occur in different regions of the
frontal lobe, showing anatomic patterns that depend
on the specific word generation task used. The bulk of frontal
activations, however, regularly occur in the cortex
that surrounds the inferior frontal sulcus, including classical
Broca's area, dorsolateral prefrontal cortex
(Brodmann's areas 46 and 9), and premotor cortex.7,8,12
We examined 100 healthy volunteers with fMRI performed at this
level of the frontal lobe to evaluate the
relative participation of both hemispheres in silent word
generation in 50 left-handed subjects compared with 50
right-handed reference subjects.
Methods. Subjects. A total of 100 university students were
recruited to make up homogeneous groups of 50
right-handed and 50 left-handed subjects, each of which included
25 women and 25 men. A personal history was
recorded in each volunteer to exclude neurologic, psychiatric,
and relevant medical diseases. Particular care was
taken to rule out subjects with possible brain injuries suffered
during the perinatal age and early childhood.
Fourteen additional subjects of both genders and handedness were
also screened and examined with fMRI to
procure a reserve group to replace subjects showing eventual
poor compliance during imaging assessment.
Group assignation was carried out based on subjects'
self-reported handedness, and quantitative assessment
was determined by the 10-item Edinburgh Inventory.16 Scoring was
carried out based on Bryden's procedure,17 in
which a particular score may range from 10 (extremely
right-handed) to 50 (extremely left-handed). All subjects
reporting right-hand preference produced a score between 10 and
20 points. As expected from previous
-
19/8/2014 Ovid: Cerebral lateralization of language in normal
left-handed people studied by functional MRI.
http://0-ovidsp.tx.ovid.com.jabega.uma.es/sp-3.12.0b/ovidweb.cgi
2/10
reports,1 scoring of subjects considering themselves left-handed
showed a wider range. The handedness scores
in this group ranged from 29 to 50 points.
All subjects gave written informed consent before inclusion in
this study, which was approved by the local
Investigation Review Board. Mean age of the final whole series
was 24.6 4.8 years (24.4 4.0 years for dextrals
and 24.9 5.6 years for sinistrals). Mean age for women was 24.6
4.2 years and for men 24.6 5.5 years. Two
subject were rejected due to psychoactive substance use, and
another subject owing to the finding of
intracerebral arteriovenous malformation in the routine MRI
performed before each fMRI assessment.
Task and testing procedure. The word generation task used in the
activation protocol was standard phonetic
verbal fluency based on the Controlled Word Association Test.18
In this test, the subject is required to generate
words beginning with a designed letter. Eight different letters
(F, A, S, C, D, M, P, and R) were used in balanced
random order. Each subject performed the task using his or her
native language (Catalan or Spanish). When tested
outside the scanner, subjects produced a mean of 49.1 14.0 words
during consecutive 1-minute testing with the
letters F, A, and S. Subjects were required to articulate each
evoked word "silently and in its entirety," with only
slight tongue movements. Eyes were closed during scanning time,
and motion was minimized by using soft head
and neck holders.
Trials consisted of four 1-minute periods in which rest and
activation were alternated. In the first period
subjects were instructed to do nothing. After 1 minute, the
specific letter required for the first activation was
given through the system intercom. Attempts to generate words
beginning with this letter were performed by
subjects until the order was given to stop the task.
Subsequently, another 1-minute rest period was required
until a different second-activation letter was provided. Thus, a
functional sequence involved a two-letter trial.
Four functional sequences were obtained in each subject.
Functional MRI. A 1.5-Tesla Signa system (GE Medical Systems,
Milwaukee, WI) with standard quadrature head
coil was used. The functional sequence consisted of a spoiled
gradient recalled acquisition in a steady state
(GRASS) (repetition time/echo time/pulse angle = 73/60/30) with
a 256 64 pixel matrix, within a field of view of 24
cm, and with a section thickness of 4 mm. First-order flow
compensation gradients were used. Field homogeneity
was adjusted in each subject at the level of each functional
slice by automated shimming on the three axes. Each
image in this protocol lasted 5 seconds. The functional time
series consisted of 48 consecutive images obtained in
4 minutes.
fMRI assessment in this study was limited to the part of the
lateral frontal lobe in which large changes are
detectable with fMRI during the word generation task
used.11-15,19 This region extends around the inferior
frontal sulcus and is anterior to the precentral sulcus, as
illustrated in figure 1. Three-dimensional sequences
additionally obtained in the subjects and surface rendering
models of the brain were helpful to assist
identification of anatomic landmarks. Well-established anatomic
criteria were also used to locate the central,
precentral, and inferior frontal sulci.20,21
-
19/8/2014 Ovid: Cerebral lateralization of language in normal
left-handed people studied by functional MRI.
http://0-ovidsp.tx.ovid.com.jabega.uma.es/sp-3.12.0b/ovidweb.cgi
3/10
Figure 1. The red rectangle on the brain surface rendering
delimits the frontal region considered in this study.
Note that this region is above Broca's speech area and is
anterior to the precentral sulcus. The slices selected
around the inferior frontal sulcus are represented in green.
Two oblique-axial slices were selected to cover the region of
interest, and two functional sequences were
acquired at each location during four separate imaging runs. The
inferior frontal sulcus was used as reference for
slice selection. This reference was randomly obtained from
either the right of left frontal sulcus to minimize a
possible anatomic bias caused by subject position
asymmetries.
Image analysis. Functional sequences were analyzed using an
auxiliary workstation (SPARCstation 20; Sun
Microsystems, Mountain View, CA) and specific image analysis
software (FuncTool, GE Medical Systems, Buc,
France). Images were reconstructed in 1 256 256 pixel matrix,
and t-test images were obtained using procedures
previously described.22,23 One activation image was
reconstructed in each functional sequence with the pixel-
by-pixel calculation of t statistics resulting from the
comparison of signal intensity obtained during rest and
activation. The pooled variance estimate method of Student's
t-test was used in this procedure. The first image of
each trials was excluded from the analysis owing to its
precarious steady-state condition. To guide measurements
and to display activations, each t-test image was fused to a
corresponding anatomic image. The presence of
motion was finely detected by a researcher blinded to the
subjects' data using a cine display method. Sequences
in which detectable motion occurred were not further
considered.
Hemispheric lateralization was determined by comparing the
number of activated pixels for left and right
hemisphere in the defined prefrontal region. An activation
laterality index was computed for each functional
sequence according to the expression 100* (L - R)/(L + R), where
L was the number of pixels above the threshold
in the defined frontal region of the left hemisphere and R the
corresponding counting for the right hemisphere.
An index of +100 expresses complete lateralization to the left
hemisphere whereas an index of -100 represents an
activation entirely lateralized to the right hemisphere.
Activation changes surrounding the central sulcus were
additionally analyzed as representative of primary sensorimotor
cortex activation.
The four laterality indexes corresponding to the four activation
sequences were averaged to provide a single
score in each subject. When functional sequences were not
suitable for analysis as a result of head movement or
insufficient activation, the overall laterality index was
computed from the average of the remaining sequences.
Subjects showing three or four failing sequences were excluded
from the study and replaced by reserve
individuals.
All measurements were performed by a single researcher blind to
the handedness and gender of subjects.
Pixel-counting was repeated in 100 activation sequences in two
separate sessions to evaluate the reproducibility
of the measurements. We found and intraclass correlation
coefficient. of 0.98 for the 100 doubly computed
laterality indexes.
-
19/8/2014 Ovid: Cerebral lateralization of language in normal
left-handed people studied by functional MRI.
http://0-ovidsp.tx.ovid.com.jabega.uma.es/sp-3.12.0b/ovidweb.cgi
4/10
Activation threshold. We used a cluster size threshold in
combination with a t value threshold. On the
composite functional images, we displayed those clusters larger
than 13 pixels with t values greater than 1.9 (p