doi:10.1093/brain/awh414
Brain (2005), 128, 1377–1385
Reading skills after left anterior temporal lobe
resection: an fMRI study
Uta Noppeney,1 Cathy J. Price,1 John S. Duncan2 and Matthias J. Koepp2
1
Wellcome Department of Imaging Neuroscience and
Department of Clinical and Experimental Epilepsy,
Institute of Neurology, London, UK
2
Correspondence to: U. Noppeney, Wellcome Department
of Imaging Neuroscience, University College London,
12 Queen Square, London WC1N 3BG, UK
E-mail: [email protected]
Summary
Maintaining language functions after left hemisphere
lesions has been associated with compensatory right hemisphere activation. It remains unclear whether recruitment
of right hemisphere regions necessarily provides an effective mechanism to compensate for language deficits. To
investigate the compensatory mechanisms that mediate
good reading skills in patients after left anterior temporal
lobe resection for mesial temporal lobe epilepsy (mTLE),
we tested for the effect of their reading ability on the
regional fMRI (functional MRI) signal elicited by sentence
reading. Sixteen control subjects and 16 patients participated in the study. In the activation condition, they silently
read nine-word sentences, and in the baseline condition
they viewed nine-word sentences after all the letters were
transformed into false fonts. Reading ability in controls
and patients significantly (P < 0.05, corrected) predicted
activations in a left hemisphere middle temporal region
that was part of the normal sentence reading system.
In addition, reading ability in patients, but not controls,
significantly predicted activation in the right inferior
frontal sulcus, right hippocampus and right inferior
temporal sulcus. Right inferior frontal activation was only
observed in the patients. In contrast, right hippocampal
and inferior temporal activation was observed in all controls and in patients whose reading ability was within the
normal range, indicating the importance of these regions
for efficient encoding during normal sentence reading.
We conclude that proficient reading skills following left
anterior temporal lobe resection for mTLE rely on two
mechanisms: (i) integrating regions from the normal system (i.e. the left middle temporal, right hippocampus and
anterior superior temporal sulcus); and (ii) recruiting
right hemisphere regions (i.e. the right inferior frontal
sulcus) that are not activated in control subjects.
Keywords: sentence comprehension; functional MRI; left anterior temporal lobe resection; temporal lobe epilepsy
Abbreviations: fMRI = functional MRI; HCV = hippocampal volume; mTLE = mesial temporal lobe epilepsy
Received September 16, 2004. Revised December 16, 2004. Accepted January 4, 2005. Advance Access publication
February 16, 2005
Introduction
Contralateral, possibly homologous cortical areas are thought
to compensate for language deficits after left hemisphere
lesions (Wernicke, 1886; Kertesz, 1988; Weiller et al., 1995;
Thulborn et al., 1999; Gold and Kertesz, 2000; Blasi et al.,
2002). It remains controversial whether recruitment of right
hemisphere regions necessarily provides an effective mechanism to compensate for language deficits. A necessary role
for right hemisphere regions in language processing was
initially proposed on the basis of patients with left hemisphere
lesions who showed residual language impairment after additional transient (sodium amobarbital test, Kinsbourne, 1971)
or permanent (e.g. stroke, Basso et al., 1989) right hemisphere
lesions. Functional imaging studies that have investigated the
effect of language abilities on regional brain activation have
yielded conflicting results: while language recovery in
patients with acute aphasia has been suggested to depend
on restored activation in unaffected left hemisphere regions
(Karbe et al., 1998; Heiss et al., 1999), chronic post-stroke
aphasia (Weiller et al., 1995; Thulborn et al., 2000; Blasi et al.,
2002) or slowly evolving brain lesions (Thiel et al., 2001)
have been shown to benefit from inter-hemisphere reorganization. These discrepancies suggest that lesion location and
extension as well as onset and time course of the pathological
process determine in part which reorganization pattern is
beneficial for language processing.
Functional imaging studies of language processing in
chronic epilepsy patients with a left hemisphere focus have
provided some evidence for a preoperative right hemispheric
# The Author (2005). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: [email protected]
1378
U. Noppeney et al.
activation shift (Billingsley et al., 2001; Carpentier et al.,
2001; Gaillard et al., 2002; Rutten et al., 2002; Adcock
et al., 2003; Sabsevitz et al., 2003). However, these preoperative functional MRI (fMRI) studies are difficult to interpret because of residual responses in the left anterior temporal
pole (Rutten et al., 2002) and influences of ictal or even post-/
interictal epileptic activity, that may falsely lateralize language function to the right hemisphere (Gaillard et al.,
1995; Bellgowan et al., 1998; Jayakar et al., 2002). Moreover, the majority of studies primarily used global brain
activation lateralization indices as possible predictors of postoperative outcome of language functions. For instance, less
lateralized fMRI activations on a preoperative semantic
decision task were associated with less confrontation naming
deficits postoperatively (Adcock et al., 2003). Although these
results suggest that higher preoperative right hemisphere
activation predicts good postoperative naming, they do not
indicate which parts of the language system have been reorganized or whether the reorganization occurs within the normal language system.
In the present study, we investigated the neuronal systems
that mediate good reading performance in patients who had
undergone left anterior temporal lobe resection for the treatment of mesial temporal lobe epilepsy (mTLE). Factors
which usually affect studies in chronic epilepsy patients
(i.e. interictal, ictal and post-ictal epileptic activity and the
presence of usually more than two antiepileptic drugs given in
high doses) were excluded by including only postoperative
patients who had been seizure free for >2 years and were
either on low doses of a single antiepileptic drug or on no
antiepileptic drugs at all.
The neural systems that mediate good reading performance
postoperatively were identified by including a composite
reading ability score as a predictor in our general linear
model. In this way, we were able to determine brain areas
where activation increases with reading ability, i.e. areas
where increased activation is associated with high reading
ability. This performance correlational approach overcomes
some of the limitations involved in the direct comparison of
reading activation in different subject groups (e.g. the comparison of reading activation in pre versus postoperative
patients or patients versus controls). For example, it capitalizes on the range of language abilities within the patient group
rather than allowing performance differences to confound true
differences between patients and controls or between patients
pre- and postoperatively. Nevertheless, we also included a
group of age- and gender-matched control subjects in order
to dissociate compensatory mechanisms in the patients that
occur within the normal reading system from those that occur
in areas that are not observed in control subjects.
Thus, the present study aimed to characterize the neural
systems underlying successful behavioural compensation and
high reading skills in postoperative patients.
In the activation conditions, the participants silently
read sentences for meaning. In the baseline conditions,
they attentively viewed false fonts. This low-level baseline
allowed us to identify the entire sentence reading activation
from visual attention and orthographic processing to semantic
integration at the sentential level. We then assessed whether
reading ability could predict activation in the regions engaged
by sentence reading in patients.
Subjects and methods
Subjects
Sixteen control subjects (five males; mean age 34 years; range
20–53) and 16 patients after left anterior temporal lobe resection
(six males; mean age 36 years; range: 29–53) participated in the
fMRI study and the associated behavioural tests. Each of the patients
had undergone surgical resection of the left anterior temporal lobe as
a therapeutic procedure for refractory epilepsy (duration of epilepsy,
mean 22.1 years, range 7–46; age of onset, mean 7.6 years, range
1–20; length of postoperative period, mean 5.9 years; range 2–9; see
Appendix for individual data). The duration and onset of epilepsy
were highly correlated (correlation coefficient 0.67). Prior to surgery,
all patients had a definite MRI diagnosis of unilateral left-sided
hippocampal sclerosis according to accepted qualitative and quantitative MRI criteria with unilaterally reduced hippocampal volumes
corrected for intracranial volume, abnormal hippocampal volume
asymmetry indices and increased hippocampal T2 times (Woermann
et al., 1998). In all patients, the MRI findings of hippocampal sclerosis were histologically verified in the excised mesial temporal
structures (hippocampus, part of the amygdala and anterior 3 cm
of lateral temporal neocortex). All patients were seizure free for at
least 2 years (range 2–9) following surgery. At the time of fMRI
scanning, patients were on monotherapy with carbamazepine (n = 9),
sodium valproate (n = 1), topiramate (n = 1) or phenobarbitone
(n = 1); the remaining four patients had stopped taking antiepileptic
drugs. Out of a large epilepsy surgery programme spanning 1994–
2000 with 100 left anterior temporal lobe resections for mTLE, we
were only able to recruit a total of 16 patients who would fit our
inclusion criteria and were willing to participate: seizure freedom
>2 years, significant reduction of antiepileptic drugs or free of antiepileptic drugs, able to tolerate an fMRI session and sufficient reading abilities (see Appendix for individual clinical data).
All participants were right-handed English native speakers (except
for one left-handed English native patient). Written informed
consent was obtained in all cases according to the Declaration of
Helsinki. The approvals of the Joint Ethics Committee of The
Institute of Neurology and The National Hospital for Neurology
and Neurosurgery were obtained.
Behavioural testing (before scanning)
All subjects were assessed for their reading age (Schornell test) and
verbal IQ (National Adult Reading Test; NART). Their reading
fluency was assessed by measuring the passage time (mean = 30 s,
SD = 8) and the number of errors (accuracy) for a standardized text
paragraph. These reading tests (NART, reading age, reading fluency
and reading accuracy) were selected out of a large preoperative
routine neuropsychological test programme, so as to match preoperative reading ability. In addition, the digit span was measured to
provide insight into subjects’ attention span and ensure that they
were able to attend during the experiment.
Before the actual scanning, subjects were given a training session
where they read similar nine-word sentences in the same rapid visual
presentation mode as in the scanner. To ensure that subjects were
Reading after left anterior temporal lobe resection
able to read sentences, they were asked to read them out aloud and to
explain their semantic content.
Table 1 Behavioural data
Experimental design
Reading age
Sentence comprehension
(proportion correct)
NART
Reading fluency (s)
Reading accuracy
(no. of errors)
Digit span (scaled)
In the activation condition, subjects silently read nine-word
sentences for meaning. Altogether they read 240 novel sentences
covering a range of different syntactic structures and semantic
content. No explicit task was required during scanning in order to
avoid inducing additional executive processes (see below for evaluation of sentence comprehension and recognition). In the baseline
condition, subjects attentively viewed nine-word sentences after
all the letters were transformed into false fonts. This baseline
controlled for visual input, but not lexical, semantic or syntactic
content. The paradigm therefore maximized our chances of seeing
reading-related activation at any level of the reading system. Blocks
of six sentences were interleaved with blocks of five false font
sentences. The order of conditions was counterbalanced across
sessions and subjects.
Sentences and false fonts were presented one word at a time at a
fixed rate [word duration 300 ms, word stimulus onset asynchrony
350 ms, sentence word stimulus onset asynchrony 4350 ms; block
length for (i) six sentences, 26.1 s; (ii) five false font sentences,
21.7 s]. This serial presentation mode was used to control for visual
input and eye movements, and to equate the reading pace of control
subjects and patients. Continuous eye monitoring was performed to
ensure that subjects were attending to the stimuli.
Behavioural testing (after scanning)
Recognition memory test
To ensure that subjects had processed the sentences during the fMRI
experiment, they were given a surprise recognition test. In other
words, subjects were informed about the recognition test only
after scanning in order not to confound the activation results by
intentional encoding components. All 240 of the ‘old’ sentences
that subjects had read in the scanner were re-presented together
with 240 ‘new’ sentences. The test required subjects to indicate
by a two-choice key press (i) whether or not they had read the
sentence in the experiment (old/new judgement); and (ii) whether
or not they were confident about their decision (confidence judgement). Subjects were only asked to be as accurate as possible;
reaction time was not emphasized. Two patients and one control
subject did not perform all parts of the recognition memory test; one
control subject and one patient did not participate.
Word by word sentence reading and
comprehension test
Using the rapid visual presentation mode, subjects read 52
novel sentences. Thirty of these sentences were followed by a question that reformulated the preceding sentence. The questions were
presented as a whole on the screen. Subjects had to indicate their
answer by a two-choice key press. The sentences and questions
covered a range of different syntactic structures related to different
degrees of comprehension difficulty (see Appendix for example
sentences).
Reading ability measure
Five behavioural measures were acquired from patients and controls: (i) reading fluency; (ii) reading accuracy; (iii) reading age;
Controls
17.7 (1.8) n = 15
0.8 (0.07) n = 14
112.7 (9.1) n = 15
33.9 (4.9)
0.6 (0.9)
12.0 (3.3)
1379
Patients
16.4 (2.4)
0.8 (0.1)
105.8 (10.4)
37.4 (12.0)
0.8 (1.5)
11.3 (3.6)
Values are across-volunteer means (SD).
(iv) sentence comprehension; and (v) verbal IQ. As these performance scores were highly correlated with one another (pairwise
correlations >0.5), we were not able to dissociate their effects on
brain activation. To obtain a single reading measure, the five performance scores for control subjects and patients were therefore
entered into a factor analysis (as implemented in SPSS; missing
values were replaced by the group mean; see Table 1 for group
means and Appendix for individual data). This factor analysis
yielded only one main (i.e. eigenvalue >1) component that explained
63.7% of the variance in the data. The subject-specific scores for this
first factor thus represent reading ability and were used as predictors
in the subsequent fMRI analysis.
fMRI
A 2 T Siemens Vision system was used to acquire both T1 anatomical volume images and T2*-weighted axial echoplanar images with
blood oxygenation level-dependent (BOLD) contrast [gradient echo,
Cartesian k-space sampling, echo time (TE) = 40 ms, repetition time
(TR) = 2.9 s, slices acquired sequentially in the descending direction,
64 3 64 matrix, 3 3 3 3 3 mm3 voxels spatial resolution, 1.2 mm
interslice gap, 1.8 mm slice thickness, 38 slices covering nearly the
whole brain]. To avoid Nyquist ghost artefacts, a generalized reconstruction algorithm was used for data processing (Josephs et al.,
2000). There were two sessions with a total of 350 volume images
per session. The first six volumes were discarded to allow for T1
equilibration effects.
Data analysis
The data were analysed with statistical parametric mapping (using
SPM99 software from the Wellcome Department of Imaging Neuroscience, London; http//www.fil.ion.ucl.ac.uk/spm). Scans from each
subject were realigned using the first as a reference, spatially normalized (Friston et al., 1995) into standard space (Talairach and Tournoux
1988), resampled to 3 3 3 3 3 mm3 voxels and spatially smoothed with
a Gaussian kernel of 8 mm full width at half-maximum (FWHM). The
time series in each voxel was highpass filtered to 1/100 Hz. The fMRI
experiment was modelled in an event-related fashion with regressors
entered into the design matrix after convolving each event-related
stick function with a canonical haemodynamic response function
and its first temporal derivative. Each stick function encoded the
onset of a single sentence. Other covariates of no interest were the
realignment parameters (to account for motion artefacts).
Condition-specific effects for each subject were estimated according to the general linear model (Friston et al., 1995). The contrast
image (sentence reading relative to baseline) for each subject was
1380
U. Noppeney et al.
then entered into a second level analysis of variance (ANOVA),
which modelled the group effect (i.e. control subjects or patients)
on the contrast of interest. The mean corrected reading scores were
entered as covariates separately for control subjects and patients.
The mean corrected duration of epilepsy was entered as a covariate
for the patients. Inferences were made at the second level to emulate
a random effects analysis and enable inferences at the population
level (Friston et al., 1999). Given that our design was balanced, this
two-stage procedure is mathematically identical to a random/mixed
effects analysis.
At the second level, we tested for the following.
Sentence effects in control subjects
Sentences > baseline in control subjects.
Sentence effects in patients
Sentences > baseline in patients.
Sentence effects that increased with reading ability
for both patients and controls
Sentence-evoked activations that increased with reading ability in
both patients and controls were identified with a conjunction of
(i) positive correlation with reading ability in patients; (ii) positive
correlation with reading ability in controls; (iii) the sentence effect
in patients (contrast 1); and (iv) the sentence effect in controls
(contrast 2).
Sentence effects that increased with reading ability
for patients
Sentence-evoked activations that increased with reading ability only
in patients were identified with a conjunction of (i) a positive correlation with reading ability in patients; and (ii) the sentence effect in
patients (i.e. contrast 2). Furthermore, we only report regions that
also showed a significant interaction (i.e. correlation with reading
ability in patients > correlation with reading ability in controls) at
0.01 uncorrected.
Sentence effects that increased with reading ability
for controls
Sentence-evoked activations that increased with reading ability only
in controls were identified with a conjunction of (i) a positive
correlation with reading ability in controls; and (ii) the sentence
effect in controls (i.e. contrast 1). Furthermore, we only report
regions that also showed a significant interaction (i.e. correlation
with reading ability in controls > correlation with reading ability in
patients) at 0.01 uncorrected.
Unless otherwise stated, the main effects are reported at P < 0.05
(corrected for multiple comparisons in the entire brain volume).
Results
Behavioural data
Sixteen control subjects and 16 patients after left anterior
temporal lobe resection were included in this study.
The summary performance scores (group mean 6 SD) of
control subjects and patients are presented in Table 1 for
the following behavioural parameters: (i) sentence comprehension skills (=word by word sentence reading and comprehension test, see Subjects and methods); (ii) NART-IQ;
(iii) reading accuracy; (iv) reading age; (v) reading fluency;
(iv) working memory span; and (vii) sentence recognition
score (see below). The individual performance scores are
presented in the Appendix.
The composite reading scores derived from the factor
analysis were not significantly different across groups (P >
0.05; patients, mean = 3.2, SD = 1.2; controls, mean = 3.2,
SD = 0.6).
Recognition memory test
After the fMRI experiment, subjects performed a surprise
sentence recognition test (see Subjects and methods,
Table 2). Accuracy of confident and non-confident recognition (Snodgrass and Corwin, 1988) was indexed by the discrimination measure Pr (probability hit – probability false
alarm). The discrimination rates were significantly greater
than zero for both controls [Pr = 0.48, t(13) = 7.2] and patients
[Pr = 0.39, t(12) = 7.2, P < 0.001, two-tailed] and they were
not significantly different across groups [t(25) = 1.0, P > 0.05].
Summing over the sure/not sure decision, controls categorized 76% of the new sentences as new, and 74% of the old
sentences as old, while patients categorized 72% of new
sentences as new and 67% of old sentences as old.
Table 2 Recognition performance for new and old
sentences that were read during the fMRI study
Word
type
Old
C
P
New
C
P
Fig. 1 Sentence activations for control subjects (red) and
patients (green) are rendered on a normalized brain. Height
threshold: P < 0.05 corrected. Extent threshold: >3 voxels.
Recognition judgement (proportion of responses)
Sure old
Unsure old
Sure new
Unsure new
0.57 (0.18)
0.56 (0.20)
0.17 (0.16)
0.16 (0.12)
0.14 (0.10)
0.14 (0.12)
0.12 (0.08)
0.15 (0.15)
0.09 (0.12)
0.16 (0.15)
0.15 (0.17)
0.16 (0.13)
0.52 (0.25)
0.40 (0.24)
0.24 (0.16)
0.27 (0.20)
Values are across-volunteer means (SD); C = control subjects;
P = patients.
Reading after left anterior temporal lobe resection
1381
Fig. 2 Left: effect of reading ability (i) common to control subjects and patients (row 1); and (ii) in patients (rows 2–4) on coronal
slices of a structural image created by averaging the normalized images from 16 patients. Height threshold: P < 0.001 uncorrected. Middle:
parameter estimates for (1) sentence reading relative to baseline for patients (2) sentence reading relative to baseline for controls
(3) correlation with reading ability for patients and (4) correlation with reading ability for controls. The parameter estimates reflect the
contributions of each of these explanatory variables to the experimental variance. Right: scatter plots depict the correlation between
individuals’ regional fMRI signal and reading ability (open circles, control subjects; filled circles, patients). The ordinate represents the
fMRI signal and the abscissa represents reading ability.
Functional imaging data
Sentence effects in control subjects
In control subjects, reading sentences activated the left
inferior frontal, middle temporal/superior temporal sulcus and
fusiform gyri, the left precentral sulcus, bilateral temporal
poles and right parahippocampal gyrus. In addition, activation was noted in the left occipital cortex (ramus descendens)
and right cerebellum.
1382
U. Noppeney et al.
Table 3 Activation affected by reading ability
Coordinates
x
In patients and controls
Left middle temporal
gyrus/superior temporal
sulcus
In patients only
Right inferior frontal
sulcus
Right hippocampus
Right inferior temporal
sulcus
y
Z-score
z
63
15
18
5.6 (tmin = 2.5)
45
27
24
5.0 (tmin = 3.6)
24
48
12
36
24
12
5.8 (tmin = 4.5)
5.1 (tmin = 3.7)
Sentence effects in patients
In patients, reading sentences activated the right temporal
pole. Activation in the middle temporal gyri, the precentral
sulci and the inferior frontal gyri was bilateral. In addition,
activation was noted in the right fusiform gyrus and
cerebellum.
Sentence effects that increase with reading
ability for patients and controls
In both patients and control subjects, sentence-evoked activation was positively predicted by reading ability only in the left
middle temporal gyrus/superior temporal sulcus.
Sentence effects that increase with reading
ability in patients only
Two different activation patterns were observed. First, sentence activation in the right inferior frontal sulcus was present
and positively predicted by reading ability only in patients,
while no sentence activation was observed for the control
subjects. Secondly, sentence activation in the right hippocampus and right inferior temporal sulcus was also positively
predicted by patient reading ability. However, in these two
regions, activation for sentence reading relative to baseline
was observed in control subjects and in patients whose reading ability was within the normal range, while poor reading
performance in patients was associated with low activation
relative to baseline.
Sentence effects that increase with reading
ability in controls only
No significant effects were observed.
We obtained equivalent regional activations when the lefthanded patient was excluded in an analysis limited to the
15 right-handed patients. In addition, this analysis revealed
sentence activation in the right superior temporal sulcus
(coordinates: x = 54 y = 48 z = 15) increasing with reading
ability in patients only.
Discussion
The current study investigated the neural systems that mediate the maintenance of reading functions in patients after left
anterior temporal lobe resection. In the patients, reading ability significantly predicted activation in left temporal regions
that form part of the normal sentence comprehension system
and a right frontal region that was not activated for sentence
reading in control subjects. Moreover, the patients showed an
effect of reading ability in the right hippocampus and the right
superior temporal sulcus, indicating their importance for efficient encoding during language processing following left
temporal lobe resection.
Three distinct types of activation patterns could be
distinguished in the regions that contributed to successful
maintenance of sentence comprehension.
First, sentence activation in a left middle temporal area was
present and increasing with reading ability both in controls
and in patients. Consistent with previous studies of sentence
reading, this area might therefore be engaged in semantic
integration at the sentential level (Vandenberghe et al.,
2002) in both normals and patients.
Secondly, sentence activation in the right prefrontal region
was only present and increasing with reading ability in the
patients. Thus, high reading ability in postoperative patients
is also mediated via recruitment of a right hemisphere region
that is not activated in the control subjects. Possibly, the right
prefrontal region might be engaged in working memory processes or particular syntactic operations like its homotope left
hemisphere region in control subjects (Just et al., 1996;
Caplan et al., 1998, 1999). It might also function as an articulatory control system. Clearly, future studies are needed to
characterize the functional role of this area following left
temporal lobe resection.
Thirdly, the right hippocampus and superior temporal sulcus showed an effect of reading ability only in the patient
group and were activated relative to baseline only in the
control subjects and in the patients whose reading ability
was within the normal range. An activation pattern of this
sort might simply reflect differences in task performance
between patients with high and low reading skills. For
instance, given the hippocampal role in incidental encoding,
subjects with high reading skills might be more engaged in
encoding the sentences. Alternatively, our results suggest
that proficient reading skills in the absence of the left
anterior hippocampus rely on the intactness of the right
hippocampus. Consequently, reading ability might be
reduced in those patients who have additional neuronal dysfunction in the right hippocampus not evident on MRI. Most
autopsy studies of hippocampal sclerosis from patients with
long-standing TLE have revealed that the degree of hippocampal cell loss is usually asymmetrical, but present, to
some degree, on both sides in the majority of patients
(Babb and Brown, 1987). As many as 25% of patients
with unilateral hippocampal sclerosis had symmetrical damage associated with endfolium sclerosis often noted in the
Reading after left anterior temporal lobe resection
contralateral hippocampus (Margerison and Corsellis, 1966).
Endfolium sclerosis is difficult to detect using MRI (Van
Paesschen et al., 1995). We have excluded severe hippocampal sclerosis of the contralateral hippocampus by T2 and
HCV measures, but lesser degrees (e.g. endfolium sclerosis)
may be present and not found on MRI. We previously reported evidence for contralateral benzodiazepine receptor
abnormalities in about one-third of patients in whom high
resolution MRI only revealed unilateral hippocampal sclerosis (Koepp et al., 1997). Our current fMRI findings might
reflect the functional correlate of these subtle structural
abnormalities leading to functional disturbances despite
seizure freedom.
Taken collectively, our results suggest that effective language maintenance in postoperative patients depends on two
mechanisms: (i) integrating regions from the normal system
(i.e. the left middle temporal, right hippocampus and anterior
superior temporal sulcus); and (ii) recruiting right hemisphere
regions (right prefrontal area) that are not activated in control
subjects.
While our study emphasizes the importance of right hemispheric activations for language maintenance after left
anterior temporal lobe resection, it cannot determine whether
these plastic changes occurred pre- or postoperatively. Thus,
inter-hemisphere reorganization has been emphasized previously by fMRI studies that demonstrated a right hemisphere
shift of language-related activation as indexed by decreased
lateralization indices in about a quarter of preoperative
patients with mTLE relative to control subjects (Billingsley
et al., 2001; Carpentier et al., 2001; Gaillard et al., 2002;
Rutten et al., 2002; Adcock et al., 2003). However, some of
these studies directly compared language activation between
control subjects and patients with lower reading abilities, so
that activations reflecting compensatory mechanisms could
not be dissociated from task performance confounds. In our
study, the association of high reading ability after left temporal lobectomy with right hemisphere activation identifies
inter-hemisphere reorganization as an important neural mechanism for behavioural language compensation in mTLE
patients after left temporal lobectomy.
Moreover, by selecting only patients in whom the epileptogenic area had been completely removed as indicated by
postoperative seizure freedom, the current study allows us to
eliminate the effect of ongoing epileptic activity on language
reorganization. There is some evidence that metabolism and
blood flow are uncoupled during the interictal state (Breier
et al., 1997) in patients with localization-related epilepsies
(Gaillard et al., 1995). It is conceivable that frequent seizure
activity can cause severe focal uncoupling that compromised
the BOLD effect, leading to false activation and language
lateralization (Jayakar et al., 2002). Furthermore, atypical,
right-sided or bilateral speech representation in left mTLE
recently has been associated with higher spiking frequency,
suggesting that the epileptic activity itself, i.e. interictal
discharges and seizure spread, may influence speech reorganization (Janszky et al., 2003).
1383
In conclusion, the activation results in patients after successful anterior temporal lobe resection demonstrate that left
and right hemisphere regions are involved in effective language maintenance. These findings will inform future longitudinal fMRI studies that examine patients before and after
anterior temporal lobe resection and thus provide further
insights into the causal factors and time course of the functional reorganization in patients with long-standing mTLE.
Acknowledgements
We wish to thank P. Thompson and S. Baxendale for supervising the routine neuropsychological testing, and the radiographers at the Functional Imaging Laboratory for their
assistance in collecting the data. U.N. and C.P. were supported by the Wellcome Trust, and M.K. by the National Society
for Epilepsy.
References
Adcock JE, Wise RG, Oxbury JM, Oxbury SM, Matthews PM. Quantitative
fMRI assessment of the differences in lateralization of language-related
brain activation in patients with temporal lobe epilepsy. Neuroimage 2003;
18: 423–38.
Babb TL, Brown WJ. Pathological findings in epilepsy. In: Engel JR, editor.
Surgical treatment of epilepsies. New York: Raven Press; 1987. p. 511–40.
Basso A, Gardelli M, Grassi MP, Mariotti M. The role of the right hemisphere
in recovery from aphasia. Two case studies. Cortex 1989; 25: 555–66.
Bellgowan PS, Binder JR, Swanson SJ, et al. Side of seizure focus predicts left
medial temporal lobe activation during verbal encoding. Neurology 1998;
51: 479–84.
Billingsley RL, McAndrews MP, Crawley AP, Mikulis DJ. Functional MRI of
phonological and semantic processing in temporal lobe epilepsy. Brain
2001; 124: 1218–27.
Blasi V, Young AC, Tansy AP, Petersen SE, Snyder AZ, Corbetta M. Word
retrieval learning modulates right frontal cortex in patients with left frontal
damage. Neuron 2002; 36: 159–70.
Breier JI, Mullani NA, Thomas AB, et al. Effects of duration of epilepsy on
the uncoupling of metabolism and blood flow in complex partial seizures.
Neurology 1997; 48: 1047–53.
Caplan D, Alpert N, Waters G. Effects of syntactic structure and propositional
number on patterns of regional cerebral blood flow. J Cogn Neurosci 1998;
10: 541–52.
Caplan D, Alpert N, Waters G. PET studies of syntactic processing with
auditory sentence presentation. Neuroimage 1999; 9: 343–51.
Carpentier A, Pugh KR, Westerveld M, et al. Functional MRI of language
processing: dependence on input modality and temporal lobe epilepsy.
Epilepsia 2001; 42: 1241–54.
Friston KJ, Holmes A, Worsley KJ, Poline JB, Frith CD, Frackowiak R.
Statistical parametric mapping: a general linear approach. Hum Brain
Mapp 1995; 2: 189–210.
Friston KJ, Holmes AP, Price CJ, Buchel C, Worsley KJ. Multisubject fMRI
studies and conjunction analyses. Neuroimage 1999; 10: 385–96.
Gaillard WD, Fazilat S, White S, et al. Interictal metabolism and blood flow
are uncoupled in temporal lobe cortex of patients with complex partial
epilepsy. Neurology 1995; 45: 1841–1847.
Gaillard WD, Balsamo L, Xu B, et al. Language dominance in partial epilepsy
patients identified with an fMRI reading task. Neurology 2002; 59: 256–65.
Gold BT, Kertesz A. Right hemisphere semantic processing of visual words in
an aphasic patient: an fMRI study. Brain Lang 2000; 73: 456–65.
Heiss WD, Kessler J, Thiel A, Ghaemi M, Karbe H. Differential capacity of
left and right hemispheric areas for compensation of poststroke aphasia.
Ann Neurol 1999; 45: 430–8.
1384
U. Noppeney et al.
Janszky J, Jokeit H, Heinemann D, Schulz R, Woermann FG, Ebner A.
Epileptic activity influences the speech organisation in medial temporal
lobe epilepsy. Brain 2003; 126: 2043–51.
Jayakar P, Bernal B, Santiago ML, Altman N. False lateralization of language
cortex on functional MRI after a cluster of focal seizures. Neurology 2002;
58: 490–2.
Josephs O, Deichmann R, Turner R. Trajectory measurement and generalized
reconstruction in rectilinear EPI. ISMRM Meeting; 2000 p. 1517.
Just MA, Carpenter PA, Keller TA, Eddy WF, Thulborn KR. Brain activation
modulated by sentence comprehension. Science 1996; 274: 114–6.
Karbe H, Thiel A, Weber-Luxenburger G, Herholz K, Kessler J, Heiss WD.
Brain plasticity in poststroke aphasia: what is the contribution of the right
hemisphere? Brain Lang 1998; 64: 215–30.
Kertesz A. What do we learn from recovery from aphasia? Adv Neurol 1988;
47: 277–92.
Kinsbourne M. The minor cerebral hemisphere as a source of aphasic speech.
Arch Neurol 1971; 25: 302–6.
Koepp MJ, Labbe C, Richardson MP, et al. Regional hippocampal [11C]flumazenil PET in temporal lobe epilepsy with unilateral and bilateral hippocampal sclerosis. Brain 1997; 120: 1865–76.
Margerison JH, Corsellis JA. Epilepsy and the temporal lobes. A clinical,
electroencephalographic and neuropathological study of the brain in
epilepsy, with particular reference to the temporal lobes. Brain 1966;
89: 499–530.
Muller RA, Rothermel RD, Behen ME, Muzik O, Chakraborty PK, Chugani
HT. Language organization in patients with early and late left-hemisphere
lesion: a PET study. Neuropsychologia 1999; 37: 545–57.
Muller RA, Rothermel RD, Behen ME, et al. Brain organization of language
after early unilateral lesion: a PET study. Brain Lang 1998; 62: 422–51.
Mummery CJ, Patterson K, Wise RJ, Vandenbergh R, Price CJ, Hodges JR.
Disrupted temporal lobe connections in semantic dementia. Brain 1999;
122: 61–73.
Rutten GJ, Ramsey NF, van Rijen PC, Noordmans HJ, van Veelen CW.
Development of a functional magnetic resonance imaging protocol for
intraoperative localization of critical temporoparietal language areas.
Ann Neurol 2002; 51: 350–60.
Sabsevitz DS, Swanson SJ, Hammeke TA, et al. Use of preoperative functional neuroimaging to predict language deficits from epilepsy surgery.
Neurology 2003; 60: 1788–92.
Scott SK, Blank CC, Rosen S, Wise RJ. Identification of a pathway for
intelligible speech in the left temporal lobe. Brain 2000; 123: 2400–6.
Snodgrass JG, Corwin J. Pragmatics of measuring recognition memory:
applications to dementia and amnesia. J Exp Psychol 1988; 117: 34–50.
Staudt M, Lidzba K, Grodd W, Wildgruber D, Erb M, Krageloh-Mann I.
Right-hemispheric organization of language following early left-sided
brain lesions: functional MRI topography. Neuroimage 2002; 16: 954–67.
Talairach J, Tournoux P. Co-planar stereotaxic atlas of the human brain.
Stuttgart: Thieme; 1988.
Thiel A, Herholz K, Koyuncu A, et al. Plasticity of language networks in
patients with brain tumors: a positron emission tomography activation
study. Ann Neurol 2001; 50: 620–9.
Thulborn KR, Carpenter PA, Just MA. Plasticity of language-related brain
function during recovery from stroke. Stroke 1999; 30: 749–54.
Van Paesschen W, Sisodiya S, Connelly A, Duncan JS, Free SL, Raymond
AA. Quantitative hippocampal MRI and intractable temporal lobe epilepsy.
Neurology 1995; 45: 2233–40.
Vandenberghe R, Nobre AC, Price CJ. The response of left temporal cortex to
sentences. J Cogn Neursosci 2002; 14: 550–60.
Weiller C, Isensee C, Rijntjes M, et al. Recovery from Wernicke’s aphasia:
a positron emission tomographic study. Ann Neurol 1995; 37:
723–32.
Wernicke, C. Die neueren Arbeiten ueber Aphasie. Fortschr Med 1886; 4:
371–7.
Woermann FG, Barker GJ, Birnie KD, Meencke HJ, Duncan JS. Regional
changes in hippocampal T2 relaxation and volume: a quantitative magnetic
resonance imaging study of hippocampal sclerosis. J Neurol Neurosurg
Psychiatry 1998; 65: 656–64.
Appendix
Example sentences of the word by word sentence
comprehension test
The poet joined the culture club for career reasons.
The poet became a member of the club to further his career?
The ill horse was given medicine by the doctor.
The horse received medical treatment?
After shooting the old policeman, the criminal ran away.
The policeman shot the criminal?
As the scientist won an award, friends congratulated him.
The colleagues congratulated the scientist for his award?
The explosion injured six workers during their lunch break.
Six labourers were injured in the explosion?
The foreign minister apologized for not attending the conference.
The minister turned up at the meeting?
The white whale attracted many visitors in the zoo.
Many people came to see the white whale?
While the tulips survived the winter, the daffodils died.
The daffodils survived the winter?
The plumber removed the dirt in the water pipe.
The water pipe was cleaned by the workman?
The dutiful sexton lit the candles in the chapel.
The chapel was illuminated by candles?
Reading after left anterior temporal lobe resection
1385
Individual behavioural data
No.
Patients
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Controls
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Sentence comprehension
(proportion correct)
Reading
age
NART
Fluency
Accuracy
(no. of errors)
Digit span
scaled
Recognition
discrimination
0.9
0.53
0.67
0.9
0.9
0.63
0.73
0.7
0.87
0.7
0.9
0.57
0.87
0.73
0.93
0.73
18.1
16
18.1
15.6
19.5
13.2
16
19
14.4
11.1
18.6
13.5
18.6
16
17.7
17.3
114
102
115
102
120
98
111
121
95
100
115
85
101
95
117
102
25
49
30
27
22
70
35
36
42
45
32
52
31
30
36
37
0
0
0
1
0
1
0
0
1
4
0
5
0
1
0
0
12
7
18
10
15
7
10
14
10
7
10
10
8
11
18
14
0.27
0.09
0.38
0.52
0.42
0.22
0.49
0.56
0.14
0.39
0.21
0.31
0.82
0.4
0.54
0.51
0.87
0.83
–
0.8
0.9
0.8
–
0.83
0.7
0.77
0.9
0.77
0.9
0.87
0.93
0.93
15.6
16.4
18.6
14.6
–
19
19.5
16.4
18.6
14.1
19.5
18.6
19
17.3
19
18.6
111
106
118
102
–
107
120
108
110
90
124
122
121
116
117
119
30
27
35
35
32
36
40
30
38
44
33
40
30
35
29
28
0
1
3
2
1
0
1
0
1
1
0
0
0
0
0
0
14
12
18
9
15
9
17
10
8
8
15
8
14
10
12
13
0.61
0.53
0.39
0.78
0.61
0.18
0.47
0.44
0.45
0.14
0.77
0.33
0.38
0.32
0.63
0.48
Individual clinical data: patients
Patient behavioural data pre- and postoperatively
No.
Age (years)
Onset
Duration
Years postop
Design learning
Figure retention
WADA
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
34
28
34
35
32
35
29
42
36
36
46
29
25
53
45
30
13
5
11
1
20
1
15
14
9
4
6
1
3
2
4
12
12
19
17
25
7
26
10
24
19
28
34
21
20
46
35
10
9
4
6
9
5
8
4
4
8
4
6
7
2
5
6
8
Pre
Post
Pre
Post
34
39
45
31
35
34
37
36
36
30
24
35
38
30
36
40
33 6 10
(normal range)
42
41
45
32
39
34
41
37
29
31
21
33
44
31
42
40
100
100
100
100
100
85
100
100
84
100
100
92
96
100
92
100
84–100
(normal range)
100
100
91
89
100
95
96
100
88
100
100
89
100
100
98
100
L
No
No
L
L
L
L
L
No
L
L
No
No
R
L
L