Lm

FWa

b

c

d

a

ARRAA

KLNCHB

1

(nDtnfitjha

0d

Neuropsychologia 46 (2008) 3101–3107

Contents lists available at ScienceDirect

Neuropsychologia

journa l homepage: www.e lsev ier .com/ locate /neuropsychologia

anguage after hemispherectomy in childhood: Contributions fromemory and intelligence

rédérique Liégeoisa,∗, J. Helen Crossb,c, Charles Polkeyd,illiam Harknessb,c, Faraneh Vargha-Khadema

Developmental Cognitive Neuroscience Unit, UCL Institute of Child Health, 30 Guilford Street, London WC1N 1EH UKNeurosciences Unit, UCL Institute of Child Health, London, UKGreat Ormond Street Hospital for Children NHS Trust, London, UKNeurosurgery, King’s College Hospital, London, UK

r t i c l e i n f o

rticle history:eceived 19 September 2007eceived in revised form 27 June 2008ccepted 1 July 2008vailable online 9 July 2008

eywords:anguage processingeurosurgeryhild developmentemispheric differencesrain plasticity

a b s t r a c t

After hemispherectomy (removal or disconnection of an entire cerebral hemisphere) in childhood fortreatment of intractable epilepsy, gross speech and language functions are often rescued. Whether morecomplex functions, such as syntactic processing, are selectively impaired, remains controversial. Here wepresent a cross-sectional study of expressive and receptive language functions in 30 patients who haveundergone hemispherectomy (17 left, 13 right). The patients had developed epilepsy-induced pathol-ogy either during the pre/perinatal period (19 cases), or after a period of normal development (11cases; onset range = 20 months to 10 years). The patients were assessed at least 1 year post surgeryon tests of receptive vocabulary, expressive and receptive grammar, instruction comprehension, andsemantic association. Measures of verbal and nonverbal intelligence, short-term verbal memory, andworking memory were also obtained. Principal component analysis revealed that two core componentscould be extracted from the five language measures, one reflecting receptive vocabulary abilities, andthe other a composite of the other four measures. Regression analyses revealed that the best predic-tor of the composite language score was the short-term verbal memory score with some contributionfrom verbal intelligence, while the only predictor of receptive vocabulary was verbal intelligence. Theresults suggest that during childhood the lone left and right hemispheres have a similar potential fordeveloping an adequate level of receptive vocabulary. However, congenital pathology affecting either

hemisphere, and postnatal damage to the left hemisphere result in substantial language deficits that

ation

acr

is1Va

are reflected also in limitgence.

. Introduction

A growing body of evidence from functional imaging studiese.g. Schapiro et al., 2004; Wood et al., 2004) and transcra-ial Doppler sonography (e.g. Lohmann, Drager, Muller-Ehrenberg,eppe, & Knecht, 2005) suggests that speech and language func-

ions are predominantly represented in the left hemisphere inormally developing children, just as in adults. Based on suchndings, the expectation is that left hemisphere lesions sus-

ained in childhood interfere with speech and language functionust as they do in adults. Yet, compared to adult-onset leftemisphere lesions, injuries acquired early in life lead to remark-bly good language outcome (Bates et al., 2001). It is widely

∗ Corresponding author. Tel.: +44 20 7905 2728; fax: +44 20 7905 2616.E-mail address: f.liegeois@ich.ucl.ac.uk (F. Liégeois).

Ce&likia

028-3932/$ – see front matter © 2008 Elsevier Ltd. All rights reserved.oi:10.1016/j.neuropsychologia.2008.07.001

s in short-term verbal memory, and, to a lesser extent, in verbal intelli-

© 2008 Elsevier Ltd. All rights reserved.

ssumed that this difference in outcome is due to increasederebral plasticity and the capacity of the immature brain toeorganize.

The most striking instance of cerebral reorganization is seenn children who have undergone removal of the left hemi-phere for the relief of intractable epilepsy (Vargha-Khadem et al.,997; Vargha-Khadem, Isaacs, Papaleoudi, Polkey, & Wilson, 1991;argha-Khadem & Polkey, 1992; Vargha-Khadem & Mishkin, 1997),nd whose everyday language functions are spared (Basser, 1962;urtiss, de Bode, & Mathern, 2001; Dennis & Kohn, 1975; Devlint al., 2003; Pulsifer et al., 2004; Stark, Bleile, Brandt, Freeman,Vining, 1995; Vanlancker-Sidtis, 2004). Whether more complex

anguage functions, such as syntactic processing, are selectivelympaired (e.g. Dennis & Kohn, 1975), or whether their level is ineeping with the patients’ lowered intelligence and memory capac-ties (Mariotti, Iuvone, Torrioli, & Silveri, 1998; Ogden, 1988; seelso, Bishop, 1983), is still under debate.

3 ycholo

sacia&tataibr1csis

hdhseirsaio

aopsieiiAYhto

rw3puopil

2

TP

P

LLLLLLLLLLLLLLLLLRRRRRRRRRRRRR

NbC

102 F. Liégeois et al. / Neurops

Studies of language outcome after hemispherectomy have raisedeveral issues that need clarification. First, hemispherectomy beingrare surgical procedure, large group studies on language out-

ome have been rarely reported. The majority of patients describedn the literature suffered from developmental or pre/perinatallycquired pathologies (e.g. Beardworth & Adams, 1988; Dennis

Kohn, 1975; Dennis & Whitaker, 1976; Verity et al., 1982)hus making it difficult to examine the effects of congenital vs.cquired onset of pathology (see Curtiss et al., 2001, for an excep-ion). Specifically, studies of patients with pathology acquiredfter birth and later in childhood (for example as a result ofnfarct or Rasmussen’s encephalitis) have involved small num-ers (Vargha-Khadem et al., 1991), and the range of age at injuryeported was limited (Boatman et al., 1999; Gott, 1973; Stark et al.,995; Stark & McGregor, 1997). A relatively large study, directlyomparing the language competencies of left and right hemi-pherectomy patients with postnatal and pre/perinatal pathology,s therefore required and serves as the primary goal of the presenttudy.

A second issue is that despite different language measures beingighly correlated, they have been traditionally examined indepen-ently. In most large-scale studies, one or two language functionsave been studied (single word knowledge in Pulsifer et al., 2004;poken grammar in Curtiss et al., 2001; Curtiss & de Bode, 2003;veryday communication in Devlin et al., 2003), preventing thenvestigation of inter-correlated deficits. The second goal of this

eport, therefore, was to extract core language components fromeveral standardized measures of language function and use theses dependent variables. This data reduction step enabled us todentify the predictors of outcome for distinct core componentsf language functions.

2

hp

able 1atient details

atient Sex Side of hemispherectomy Pathology

1 F Left Cortical dysplasia2 M Left Sturge–Weber synd3 F Left Sturge–Weber synd4 M Left Epilepsy of unknow5 M Left CVA6 M Left CVA7 M Left CVA8 F Left CVAa

9 M Left CVA10 M Left CVA11 M Left CVA12 F Left CVA13 F Left CVA14 M Left Rasmussen’s encep15 F Left Rasmussen’s encep16 F Left Rasmussen’s encep17 M Left Rasmussen’s encep1 M Right Cyst2 M Right HEMIMG3 M Right Sturge–Weber synd4 M Right Hemimegalencepha5 M Right Pachygyria6 Right CVA7 M Right Hydrocephalus8 M Right CVAb

9 M Right CVA10 M Right Rasmussen’s encep11 F Right Rasmussen’s encep12 M Right Rasmussen’s encep13 F Right Rasmussen’s encep

ote: Abbreviations—F, female; M, male; CVA, cerebro-vascular accident of the middle ceefore onset of pathology. Cases L2 and L14 have previously been reported in Vargha-Khases L16 and L17 lost expressive communication as a consequence of the pathology buta Probably caused by multiple amniocenteses.b Following meningitis.

gia 46 (2008) 3101–3107

Finally, in hemispherectomy patients both verbal intelligencend verbal memory can be severely impaired (as often seen pre-peratively, Devlin et al., 2003; Pulsifer et al., 2004). Despite theirotential intimate relationship to language abilities, the relation-hip between them has not been consistently taken into accountn studies on language outcome (but see Ogden, 1988, and Mariottit al., 1998, for exceptions). A final goal therefore was to exam-ne the question of hemispheric differences in language abilitiesn relation to verbal intelligence and short-term verbal memory.dmittedly, these cognitive functions rely on similar processes.et, despite their interlinked nature, it is possible to assess theypothesis of a selective language impairment over and abovehe restrictions in verbal intelligence and short-term verbal mem-ry.

In the present study, standardized measures of expressive andeceptive language (semantic association, receptive vocabulary, asell as sentence production and comprehension), were collected in0 hemispherectomy patients (19 with pre/perinatal and 11 withostnatal pathology) and reduced to “core” language componentssing a principal component analysis (PCA). The effects of onsetf pathology (Pre/perinatal vs. Postnatal) and hemispheric side ofathology (Left vs. Right) on the core language components were

nvestigated, together with the roles played by memory and intel-ectual functions on language outcome.

. Methods

.1. Participants

The group was selected from a pool of children and adolescents who underwentemispherectomy for relief from intractable epilepsy at Great Ormond Street Hos-ital for Children, or at King’s College Hospitals, London, U.K. Patients were selected

Onset Of pathology Age at onset of habitual seizures

Pre/perinatal 2;0rome Pre/perinatal 0;1rome Pre/perinatal 0;1n origin Pre/perinatal 0;3

Pre/perinatal 0;1Pre/perinatal 5;6Pre/perinatal 0;1Pre/perinatal 0;7Pre/perinatal 6;0Pre/perinatal 0;1Pre/perinatal 2;0Pre/perinatal 7;0Postnatal 1;8

halitis Postnatal 3;6halitis Postnatal 6;6halitis Postnatal 6;6halitis Postnatal 10;0

Pre/perinatal 4;6Pre/perinatal 0;1

rome Pre/perinatal 0;3ly Pre/perinatal 0;7

Pre/perinatal 2;9Pre/perinatal 2;6Pre/perinatal 0;1Postnatal 7;0 (CVA at 0;6)Postnatal 3;0

halitis Postnatal 4;6halitis Postnatal 6;0halitis Postnatal 7;0halitis Postnatal 8;0

rebral artery; Age in years; months. Note that only L17 and R13 were left-handedadem et al. (1997) and Jackson, Connelly, Gordon, and Gadian (1994), respectively.had regained fluent conversational skills at the time of assessment.

F. Liégeois et al. / Neuropsychologia 46 (2008) 3101–3107 3103

Table 2Descriptive data for the left (Left-X) and right (Right-X) hemispherectomy groups

Left-X group Right-X group

Mean (years;months) S.D. (months) Range (years;months) Mean (years;months) S.D. (months) Range (years;months)

Age at onset of symptoms 1;7 37 Birth to 10 2;5 36 Birth to 8;0A to 10A to 15;E to 14;A 1 to 24

a(d

tt(goeeoa1curc

tS

2

r

2

bvfasrua

2

(

(i

(

(

2

aatpsos

3

3

3

TN

F

I

S

VER

R

SI

ge at onset of habitual seizures 3;0 38 0;1ge at operation 9;10 42 3;2lapsed time since surgery 5;9 49 1;0ge at testing 15;6 48 9;1

ccording to the following criteria: (i) age above 6 years at the time of testing;ii) seizure free status; (iii) absence of severe behavioural problems or attentionifficulty precluding neuropsychological assessment.

The final sample (see Table 1 for details) consisted of 30 patients who consentedo participate in the study. In 19 patients, the pathology underlying the epilep-ic disorder had originated in the pre/perinatal period, and was either of acquiredinfarct), developmental (e.g. cortical dysplasia, hemimegalencephaly), or of pro-ressive (Sturge–Weber syndrome) nature. In these patients, the diagnostic featuref hemiplegia was typically detected within the first 6 months of life with thepilepsy commencing months or years later. In the remaining 11 patients, intractablepilepsy and/or hemiplegia occurred during the first decade of life after a periodf normal development. The aetiology in these cases was either a middle cerebralrtery infarct (n = 3) or Rasmussen’s encephalitis (n = 8; Bien et al., 2005; Rasmussen,978). In these 11 cases, age at onset of pathology was coincident with the first clini-al signs of the disorder (i.e. hemiparesis, seizures, or both). Seventeen patients hadndergone left hemispherectomy (Left-X, patients L1 to L17) and 13 had undergoneight hemispherectomy (Right-X, patients R1 to R13; Table 1). Table 2 presents thelinical characteristics of the left and right hemispherectomy groups.

Informed consent was obtained from all parents/guardians of patients and fromhose patients aged over 18, in accordance with guidelines of the Great Ormondtreet Hospital for Children/Institute of Child Health Research Ethics Committee.

.2. Neuropsychological assessment

Minimum scores, means and standard deviations of the tests used for the neu-opsychological assessment are given in Table 3.

.2.1. Intelligence and memoryThe age-appropriate Wechsler scales of intelligence were administered and Ver-

al and Performance IQ scores were computed for each participant. Short-termerbal memory (STVM) was assessed using the Recalling Sentences (RS) subtestrom the Clinical Evaluation of Language Fundamentals (CELF), which measures theccuracy of sentence repetition. Although the RS subtest is not a conventional mea-ure of verbal memory, it was chosen because it provides a standardized measure ofecall of meaningful verbal material. Verbal working memory (WM) was assessedsing the Digit Span (DS) subtest from the intelligence scales, which measures theccuracy of repeating strings of digits both forward and backward.

.2.2. Language

(i) Expressive grammarThe Formulated Sentences (FS) subtest of the CELF requires the generation of

a sentence using a target word spoken by the experimenter. Grammatical andsemantic errors are noted.

bago5

able 3europsychological tests

unction Test/subtest

ntelligence Wechlser test of intelligence-UKversion (WISC-III-R, Wechsler, 1WAIS-III, Wechsler, 1998)

hort-term verbal memory (STVM) Recalling Sentences (RS) from Cversion (Revised; Semel, Wiig, &Secord, 1987)

erbal working memory (WM) Language measures Digit Span (DS) from intelligencxpressive grammar Formulated Sentences (FS) fromeceptive vocabulary British Picture Vocabulary Scale

Dunn, Whetton, & Pintilie, 1982eceptive grammar TROG (test for reception of gram

Bishop, 1982)emantic association Word Classes (WC) from CELFnstruction comprehension Concepts and Directions (CD) fr

CELF

* Can be used for adults.

3;6 34 0;1 to 8;07 9;7 58 0;4 to 15;65 5;7 39 0;11 to 11;10;3 15;2 58 7;11 to 21;1

ii) Receptive vocabularyThe British Picture Vocabulary Scale (BPVS) assesses the ability to select a

picture from four alternatives corresponding to the word pronounced by theexperimenter.

ii) Receptive grammarThe Test for Reception of Grammar (TROG) evaluates the comprehension of

various grammatical structures (e.g. singular/plural noun inflection, compara-tive/absolute, reversible passive, etc.). The task is to select one out of four picturescorresponding to a spoken sentence. The test consists of 20 blocks (each contain-ing four trials), the criterion for “passing” a block being correct picture matchingon all items.

iv) Semantic associationThe Word Classes (WC) subtest of the CELF requires the participant to select,

out of three or four words spoken by the experimenter, the two that “go togetherbest” (e.g. “girl, hat, boy”).

v) Instruction comprehensionThe Concept and Directions (CD) subtest of the CELF requires following

instructions of increasing length and difficulty given by the experimenter topoint to geometric figures of varying shapes and colours.

.2.3. Missing dataData collection was not completed for two patients (L3 and R8) on the CELF,

nd not available for two others (L6 and R5) on DS. Minimum scaled scores weressigned if patients failed to pass the pretest items, or failed to complete a sub-est because of attentional demands. For example, patients L4 and R5 failed theretest items of all the CELF subtests, and were therefore given the minimum scaledcore. The short attention span of one patient (R1) precluded cooperation through-ut the CELF administration, and resulted in minimum scaled scores on discontinuedubtests.

. Results

.1. Descriptive statistics

.1.1. Clinical characteristicsMann–Whitney tests revealed no significant differences

etween Left-X and Right-X groups on any of the descriptive vari-bles listed in Table 2 (p > 0.30 in all cases). The hemispherectomyroups were further divided into subgroups on the basis of age atnset of pathology (pre/perinatal: 12 Left-X; 7 Right-X; postnatal:Left-X; 6 Right-X).

Normative data:mean score (S.D.)

Minimum score Norms (up toyears;months)

992; or100 (15) 45 89

ELF, UK 10 (3) 3 16;11

e scale 10 (3) 1 89CELF 10(3) 3 16;11(BPVS,

)100 (15) 41 15;08

mar, 100 (15) 55 12;11*

10 (3) 3 16;11om 10 (3) 3 16;11

3104 F. Liégeois et al. / Neuropsycholo

Table 4Intelligence scores in the four hemispherectomy groups

Test Mean score (S.D.)

Left-X group Right-X group

Pre/perinatal Postnatal Pre/perinatal Postnatal

Verbal intelligence, VIQ 70 (13) 64 (14) 58 (12) 83 (12)N

Sa

po(ps

3

wgTo(cR(

3

aoOtpti

ts

3

mww(pmouv

rzs

Lo

3

ymp

3

mi

3

wpapptcpp

3

fa

3

adap

rcpinteraction effect). In contrast, partialling out the effect of VIQ still

TC

T

CWFTB

N

onverbal intelligence, PIQ 62 (12) 67 (15) 50 (5) 77 (13)

cores within the average range (below 1.5 S.D. of the mean of normative data orbove) are highlighted in bold.

Although examining intellectual outcome was not the mainurpose of the study, analyses of variance with hemispheric sidef removal (Side, Left-X vs. Right-X) and Onset of pathologypre/perinatal vs. postnatal) used as independent variables wereerformed on both VIQ and PIQ measures to describe the profileseen in each of the four groups (Table 4).

.1.2. Verbal IQ (VIQ)The Right-X group with postnatal pathology performed best

ith a mean score within the low average range, while the otherroups’ scores were within the extremely low range (70 or below,able 4). Statistical analyses showed no significant main effectf Side or Age at Onset, but their interaction was significantF = 9.89, p = 0.004). Post hoc tests (Mann–Whitney with Bonferroniorrection—familywise p = 0.025) revealed a trend for postnatalight-X patients to perform better than postnatal Left-X patientsU = 4.50, p = 0.052).

.1.3. Performance IQ (PIQ)The pattern of results differed from that for verbal IQ, with

ll groups performing below the average range, with mean scoresf 77 or below (Table 4). There was a significant main effect ofnset of pathology (F = 12.82, p = 0.001) and a significant interac-

ion between the effects of Side and Onset of pathology (F = 5.79,= 0.02). Overall, patients with postnatal onset performed better

han those with pre/perinatal onset, but this difference was signif-cant only in the Right-X group (U = 0.00, p = 0.001).

In order to compare results between tests, standard scores wereransformed into z scores (relative to age-appropriate norms) inubsequent analyses.

.2. Data reduction

As shown in the correlation matrix (Table 5), all languageeasures were highly correlated, indicating that factor analysisould be appropriate. Kaiser–Meyer–Olkin measure of adequacyas above 0.6 (0.792), Bartlett’s test of sphericity was significant

approximate �2 = 101.48, df = 10, p < 0.001), and measures of sam-

ling adequacy were all above 0.5, indicating adequacy of theethod. A principal component analysis performed on the z scores

btained from the five language measures (no rotation, eigenval-es > 0) revealed that two Components accounted for 86.6% of theariance. The first component (eigenvalue = 4.462), subsequently

rs

tr

able 5orrelations (Spearman’s test) between language measures (one-tailed � coefficients; *p

est Instruction comprehension, CD Semantic association, WC Express

D / 0.75* 0.87*C 0.75* / 0.79*

S 0.87* 0.79* /ROG 0.76* 0.76* 0.67*PVS 0.54* 0.68* 0.71*

ote: see text for test acronyms.

gia 46 (2008) 3101–3107

eferred to as LANG, reflected a composite of CD, TROG, WC, and FSscores. The second component (eigenvalue = 0.73) was the BPVS zcore, subsequently referred to as VOCAB.

Subsequent analyses used only the two component z scores (i.e.ANG and VOCAB) as dependent measures to examine languageutcome in the four hemispherectomy subgroups.

.3. Language outcome

Results from the VOCAB and LANG components were anal-sed separately. It is noteworthy that on both measures theost impaired scores were obtained by the Right-X group with

re/perinatal pathology (Table 6).

.3.1. Receptive vocabulary component (VOCAB)In all groups except the Right-X group with pre/perinatal onset,

ean scores were within or just below the average range. No signif-cant main or interaction effects (p > 0.20 in all cases) were detected.

.3.2. Language composite measure (LANG)Only the Right-X group with postnatal pathology performed

ithin the average range. Significant main effects of Onset ofathology (F = 7.68, p = 0.01) and Side (F = 4.36, p = 0.047), as wells their interaction (F = 15.22, p = 0.001), were detected. Overall,atients with postnatal onset performed better than those withre/perinatal onset, and the Right-X group performed better thanhe Left-X group. Post hoc tests (Mann–Whitney with Bonferroniorrection—familywise p = 0.025) revealed that Right-X patientserformed better than Left-X patients but only when onset ofathology was postnatal (U = 3.0, p = 0.03).

.4. Language outcome in relation to intelligence and memory

Two multiple linear regression analyses with three independentactors (VIQ, WM and STVM) were carried out separately for LANGnd VOCAB outcome.

.4.1. Predictors of the composite language score (LANG)Taken together, the three factors accounted for 92% of the vari-

nce in LANG scores. Examination of the T values revealed that WMid not contribute to LANG outcome (p > 0.50), while VIQ (p = 0.033)nd STVM (p < 0.001) did. It is concluded that STVM is the bestredictor of LANG outcome (see correlation plot in Fig. 1A).

Consistent with the above finding, analyses of covarianceevealed that main and interaction effects seen for LANG out-ome were no longer significant once the effect of STVM had beenartialled out (p > 0.40 for the main effects and p > 0.12 for the

esulted in a significant effect of Side (p = 0.04), and trends towardsignificant interaction (p = 0.06) and effect of Age at onset (p = 0.07).

As illustrated in Fig. 1, all patients except one with STVM withinhe normal range also obtained language scores within the normalange, suggesting that the two measures are highly interlinked.

< 0.005)

ive grammar, FS Receptive grammar, TROG Receptive vocabulary, BPVS

0.76* 0.54*0.76* 0.68*0.67* 0.71*/ 0.51*0.51* /

F. Liégeois et al. / Neuropsychologia 46 (2008) 3101–3107 3105

Table 6Summary of results from language, short-term verbal memory and working memory tests in each group

Function Mean z score (S.D.)

Left-X group Right-X group

Pre/perinatal Postnatal Pre/perinatal Postnatal

Receptive Vocabulary, VOCAB −1.46 (2.0) −1.67 (1.2) −2.42 (−0.9) −1.16 (1.2)Composite language score, LANG −1.67 (0.7) −2.0 (0.3) −2.19 (0.5) −0.52 (1.2)WS

S w 1.5

3

apv

4

e

FbmovD

oqcgta

orking memory, Digit Span −1.82 (0.7)hort-term verbal memory, Recalling Sentences −1.88 (0.43)

ee Section 2 for detailed description of tests. Scores within the average range (belo

.4.2. Predictors of the receptive vocabulary score (VOCAB)Taken together, the three variables accounted for 80% of the vari-

nce in VOCAB scores, but only VIQ contributed significantly to therediction of VOCAB scores (p = 0.001; p > 0.30 for the two otherariables; see correlation plot in Fig. 1B).

. Discussion

The primary goal of the present study was to investigate theffects of hemispheric side of injury and age at onset of pathol-

ig. 1. Correlations between outcome measures and best predictors. (A) Correlationetween short-term verbal memory and composite language (LANG) scores (Spear-an’s one-tailed � coefficient = 0.82, p < 0.001). Dotted lines indicate lower limit

f average range. (B) Correlation between verbal intelligence (VIQ) and receptiveocabulary (VOCAB) scores (Spearman’s one-tailed � coefficient = 0.83, p < 0.001).otted lines indicate lower limit of average range.

vpsppaslht

4

aggtgutipttl(clb2ad2

ooicJoVaolsl

−2.07 (0.9) −2.39 (0.7) −0.94 (0.4)−1.93 (0.28) −2.0 (0.75) −0.40 (1.87)

S.D. of the mean of normative data or above) are highlighted in bold.

gy (pre/perinatal vs. postnatal) on language outcome. The mainuestion was whether the isolated left and right hemispheres haveomparable abilities to subserve language function. A secondaryoal was to determine whether language outcome was relatedo the outcome of other cognitive abilities such as intelligence,nd memory. In the domains of language, verbal memory, anderbal and nonverbal intelligence, the best outcome was seen inatients with right hemisphere pathology of postnatal onset whocored within the average range. In contrast, those with right-sidedathology of pre/perinatal origin performed worse than the otheratient groups (more than two S.D.s below the normal mean) inll domains. Together, these results suggest that the right hemi-phere makes an important contribution to the development ofanguage and cognition during infancy, but once these functionsave emerged, their subsequent development is less dependent onhe integrity of that cerebral hemisphere.

.1. Intelligence after hemispherectomy

In the groups with pre/perinatal pathology, we failed to detectny effect of side of surgery for either verbal or nonverbal intelli-ence. For both left and right hemispherectomy patients from thisroup, however, mean verbal and nonverbal IQ scores were morehan 30 points below normal. The fact that most patients with con-enital pathology scored well below the normal range suggests thatnilateral damage of early onset does not necessarily result in bet-er outcome compared to that acquired later in childhood. This isn contrast to expectations based on the notion of increasing brainlasticity with decreasing age at insult, but consistent with the ideahat there is a cost associated with brain injury sustained beforehe emergence and early development of cognitive abilities. It isikely that within the group of pre/perinatal pathology, aetiologyprogressive, developmental, acquired, etc.) also contributes to out-ome (see Curtiss et al., 2001, for a similar conclusion on spokenanguage). For instance, developmental pathologies are known toe associated with unfavourable cognitive outcome (Devlin et al.,003), while reports of favourable language outcome – that is age,ppropriate language skills relative to neurologically intact chil-ren – have mainly focused on cases with infarcts (e.g. Bates et al.,001).

In contrast to the profile of patients with pre/perinatal pathol-gy, those with postnatal pathology showed the adverse effectsf left-sided damage on verbal intelligence. Verbal IQ was morempaired in the left than in the right hemispherectomy group,onsistent with the adult pattern of lateralization (Warrington,ames, & Maciejewski, 1986) and other reports on small groupsf children with hemispherectomy (e.g. Stark & McGregor, 1997;argha-Khadem, Isaacs, Papaleoudi, Polkey, & Wilson, 1991). It

ppears therefore that after a period of typical development hasccurred, the isolated right hemisphere cannot rescue verbal intel-ectual abilities as efficiently as the isolated left hemisphere. Aimilar pattern has been observed in children with focal unilateralesions and hemiplegia, although in the latter groups intellectual

3 ycholo

fV

(irtbmalbvcvsi

4

i(wsTotoa12o3tbb2

ipvsts(

4

4

agtpmopcsftsat

tpilgpm

4

rtgaDvsbllg“poiiGmhlfm

4

iabibupfotfcoitesa

Iaghr

106 F. Liégeois et al. / Neurops

unctions were relatively spared (Rankin & Vargha-Khadem, 2007;argha-Khadem, Isaacs, Watkins, & Mishkin, 2000).

Contrary to the pattern seen in adults with parietal damageWarrington et al., 1986), all four groups showed impairmentn nonverbal intelligence, with no worse outcome in those withight hemispherectomy. Paradoxically, in the present report it washe group with postnatal right-sided pathology that achieved theest score (23 points below the normal mean). The absence of aore striking impairment in nonverbal abilities after postnatally

cquired pathology and right hemispherectomy suggests that theeft hemisphere can subserve the visuospatial functions measuredy the Performance IQ test. A similar absence of laterality effect forisuospatial functions has been reported in 71 hemispherectomyases, regardless of aetiology (Pulsifer et al., 2004). Since examiningerbal abilities was the main purpose of the study, no other mea-ures of visuospatial functions were obtained. This issue should benvestigated in the future.

.2. Predictors of receptive vocabulary after hemispherectomy

Our data suggest that a single hemisphere is capable of subserv-ng an adequate level of vocabulary knowledge for everyday usesee below). In particular, in the left hemispherectomy group, thoseith pre/perinatal and those with postnatal pathology achieved

imilar scores (around 1.5 S.D. or 22 points below the normal mean).his relative preservation of vocabulary knowledge compared tother language skills in all groups except one (pre/perinatal injuryo the right hemisphere) is consistent with results from numerousther studies on hemispherectomy patients with acquired as wells congenital pathology (Boatman et al., 1999; Dennis & Whitaker,976; Ogden, 1988, 1996; Stark et al., 1995; Vargha-Khadem et al.,001; Zaidel, 1977). Again, the group with pre/perinatal pathol-gy to the right hemisphere performed the worst (more than0 points below the normal mean), indicating vocabulary limita-ions in line with restrictions in verbal intelligence. The associationetween single-word knowledge and general intelligence has alsoeen observed in a larger study than the current one (Pulsifer et al.,004), but verbal intelligence was not examined separately.

The sole predictor of receptive vocabulary scores was verbalntelligence, suggesting that both may rely on similar cognitiverocesses. The most parsimonious explanation is that receptiveocabulary and verbal IQ mainly assess the retrieval of consolidatedemantic knowledge, which might be more bilaterally representedhan other aspects of language and memory. Our data suggest hemi-pheric equipotentiality for vocabulary knowledge in childhoodconsistent with Mariotti et al., 1998).

.3. Predictors of language after hemispherectomy

.3.1. Language and verbal intelligenceAmong the groups, the highest level of performance was

chieved by patients with right hemispherectomy of postnatal ori-in, and it is indeed only in the group with postnatal pathologyhat the effect of side of hemispherectomy was evident. The sameattern of results has been previously reported for spoken gram-ar (Curtiss et al., 2001), in a study where a similar classification

f timing of pathology was used (i.e. developmental vs. acquiredathology). The effect of lowered intelligence on language out-ome has previously been addressed in a single-case study. In theirtudy of an adult left hemispherectomy patient who had suffered

rom Sturge–Weber syndrome, Mariotti et al. (1998) argued thathe patient was mildly impaired in language function, but to theame extent as IQ-matched controls. This was also consistent withreport by Bishop (1983) who showed that children without a his-

ory of brain damage but with intellectual abilities in the low range

apb

i

gia 46 (2008) 3101–3107

end to respond randomly on tests requiring complex languagerocessing. Our results suggest that although verbal intelligence

s related to language outcome (as measured by our compositeanguage score), it is not its best predictor. Group differences in lan-uage outcome were still present once the effect of VIQ had beenartialled out. A better predictor turned out to be short-term verbalemory capacity—as measured by sentence repetition (see below).

.3.2. Language and memoryLanguage deficits in the present patient group were highly cor-

elated with short-term verbal memory deficits, as evidenced inhe right pre/perinatal as well as the left hemisphere pathologyroups. The close relationship between short-term verbal memorynd language has previously been highlighted in an early study byennis and Kohn (1975), who had rejected the hypothesis that lowerbal IQ was the predictor of poor syntactic abilities, and insteaduggested that the “processing of auditory-verbal material” maye one of the components that explains the difference between

eft and right hemispherectomy patients. A similar hypothesis wasater put forward by Zaidel (1977) who concluded that the lin-uistic deficit displayed by the isolated right hemisphere couldreflect a limitation in short-term sequential verbal memory andresumably in covert verbal rehearsal [. . .]” (p. 11). The data fromur patient groups are consistent with these hypotheses. Interest-ngly, deficits in short-term verbal memory have also been reportedn children with Specific language Impairment (SLI, Archibald &athercole, 2006), who do not suffer from gross neurological andotor deficits. Our data suggest similarly that, in patients who

ave undergone hemispherectomy, short-term verbal memory andanguage functions overlap considerably. This may be due to bothunctions relying on similar processes involving the retrieval of

eaningful verbal sequences.

.4. Other factors influencing functional outcome

As in most clinical populations, there was a large variabilityn outcome, with language and cognitive scores ranging from theverage to the exceptionally low range. This variability is proba-ly due to a combination of factors such as individual differences

n disease progression, response to treatment of epilepsy, premor-id cognitive abilities, socioeconomic, and genetic influences. Wesed type of pathology as a categorical classifier (pre/perinatal vs.ostnatal) and obtained indications of age at onset of pathologyrom this binary distinction. However, it is noted that age at onsetf chronic seizures was significantly later in the postnatal than inhe pre/perinatal group and this factor alone could have accountedor some of the variability in language outcome. Age at onset ofhronic epilepsy can in itself be indicative of the severity and typef underlying brain disease, independent of hemispheric side ofnvolvement. Indeed, it is likely that factors other than the onset ofhe initial pathology, hemispheric side of removal, and severity ofpilepsy, account for variability in outcome in patients with hemi-pherectomy (see Curtiss & De Bode, 1999 on the importance ofetiology).

Aetiology may be another important contributor to outcome.n this series, the majority of children with pre/perinatal diseaseffecting the right hemisphere suffered from nonvascular patholo-ies. This may have led to an overrepresentation of disorders thatave more pervasive and widespread effects, and are known toesult in least favourable medical outcome (Kossoff et al., 2003)

nd global learning difficulties (Devlin et al., 2003). The small sam-le size precluded further investigations of these issues that shoulde addressed through multi-centre studies.

Finally, it should be noted that language outcome observed heres not necessarily a direct effect of surgery per se, as numerous

ycholo

rph

5

ttfbpbdnl

A

tuaB

R

A

B

B

B

B

BB

B

C

C

C

D

D

D

D

G

J

K

L

M

O

O

P

R

R

S

S

S

S

V

V

V

V

V

V

V

W

W

W

F. Liégeois et al. / Neurops

eports have shown that language and cognitive dysfunctions areresent pre-operatively, but may remain stable or improve afteremispherectomy (e.g. Devlin et al., 2003; Pulsifer et al., 2004).

. Conclusions

Our data suggest that within the limits of verbal intelligence,here is hemispheric equipotentiality during childhood for seman-ic knowledge, and within the limits of STVM, for languageunctions. Short-term verbal memory and language capacity isest afforded by right hemispherectomy patients with postnatalathology. Thus, relatively effective verbal memory functions cane subserved by the left hemisphere if it has had a period of normalevelopment, at least until birth. In all other cases, the prognosis isot favourable, and our hypothesis is that this is what largely limits

anguage functions.

cknowledgements

This research was supported by the Epilepsy Research Founda-ion. We thank Dr. Martin King for statistical advice. This work wasndertaken at GOSH/UCL Institute of Child Health who receivedproportion of funding from the Department of Health’s NIHR

iomedical Research Centres funding scheme.

eferences

rchibald, L. M. D., & Gathercole, S. E. (2006). Short-term and working memory inspecific language impairment. International Journal of Language and Communi-cation Disorders, 41, 675–693.

asser, L. S. (1962). Hemiplegia of early onset and the faculty of speech with specialreference to the effect of hemispherectomy. Brain, 85, 427–460.

ates, E., Reilly, J., Wulfeck, B., Dronkers, N., Opie, M., Fenson, J., et al. (2001). Differ-ential effects of unilateral lesions on language production in children and adults.Brain and Language, 79, 223–265.

eardworth, E. D., & Adams, B. T. (1988). Modified hemispherectomy for epilepsy:Early results in 10 Cases. British Journal of Neurosurgery, 2, 73–84.

ien, C. G., Granata, T., Antozzi, C., Cross, J. H., Dulac, O., Kurthen, M., et al. (2005).Pathogenesis, diagnosis and treatment of Rasmussen encephalitis: A Europeanconsensus statement. Brain, 128, 454–471.

ishop, D. V. M. (1982). Test for reception of grammar. Abingdon, Oxon, UK: MRC.ishop, D. V. M. (1983). Linguistic impairment after left hemidecortication for infan-

tile hemiplegia: A reappraisal. Quarterly Journal of Experimental Psychology, 35A,199–207.

oatman, D., Freeman, J., Vining, E., Pulsifer, M., Miglioretti, D., Minahan, R., et al.(1999). Language recovery after left hemispherectomy in children with late-onset seizures. Annals of Neurology, 46, 579–586.

urtiss, S., & de Bode. (1999). Age and etiology as predictors of language outcomefollowing hemispherectomy. Developmental Neuroscience, 21, 174–181.

urtiss, S., & de Bode, S. (2003). How normal is grammatical development in the righthemisphere following hemispherectomy? The root infinitive stage and beyond.Brain and Language, 86, 193–206.

urtiss, S., de Bode, S., & Mathern, G. W. (2001). Spoken language outcomes afterhemispherectomy: Factoring in etiology. Brain and Language, 79, 379–396.

ennis, M., & Kohn, B. (1975). Comprehension of syntax in infantile hemiplegics aftercerebral hemidecortication: Left hemisphere superiority. Brain and Language, 2,472–482.

ennis, M., & Whitaker, H. A. (1976). Language acquisition following hemidecorti-cation: Linguistic superiority of the left over the right hemisphere. Brain and

Language, 2, 404–433.

evlin, A. M., Cross, J. H., Harkness, W., Chong, W. K., Harding, B., Vargha-Khadem, F.,et al. (2003). Clinical outcomes of hemispherectomy for epilepsy in childhoodand adolescence. Brain, 126, 556–566.

unn, L. M., Whetton, C., & Pintilie, D. (1982). British Picture Vocabulary Scale. Wind-sor, Berkshire, UK: NFER-Nelson.

W

Z

gia 46 (2008) 3101–3107 3107

ott, P. S. (1973). Cognitive abilities following right and left hemispherectomy. Cortex,9, 266–274.

ackson, G. D., Connelly, A., Cross, J. H., Gordon, I., & Gadian, D. G. (1994).Functional magnetic resonance imaging of focal seizures. Neurology, 44,850–856.

ossoff, E. H., Vining, E. P., Pillas, D. J., Pyzik, P. L., Avellino, A. M., Carson, B. S., et al.(2003). Hemispherectomy for intractable unihemispheric epilepsy etiology vs.outcome. Neurology, 61, 887–890.

ohmann, H., Drager, B., Muller-Ehrenberg, S., Deppe, M., & Knecht, S. (2005).Language lateralization in young children assessed by functional transcranialDoppler sonography. Neuroimage, 24, 780–790.

ariotti, P., Iuvone, L., Torrioli, M. G., & Silveri, M. C. (1998). Linguistic andnon-linguistic abilities in a patient with early left hemispherectomy. Neuropsy-chologia, 36, 1303–1312.

gden, J. A. (1988). Language and memory functions after long recovery periods inleft-hemispherectomized subjects. Neuropsychologia, 26, 645–659.

gden, J. A. (1996). Phonological dyslexia and phonological dysgraphia followingleft and right hemispherectomy. Neuropsychologia, 34, 905–918.

ulsifer, M. B., Brandt, J., Salorio, C. F., Vining, E. P., Carson, B. S., & Freeman, J. M.(2004). The cognitive outcome of hemispherectomy in 71 children. Epilepsia, 45,243–254.

ankin, P., & Vargha-Khadem, F. (2007). Neuropsychological evaluation—children.In J. Engel & T. A. Pedley (Eds.), Epilepsy: A comprehensive textbook. London:Lippincott Williams & Wilkins.

asmussen, T. (1978). Further observations on the syndrome of chronic encephalitisand epilepsy. Applied Neurophysiology, 41, 1–12.

chapiro, M. B., Schmithorst, V. J., Wilke, M., Byars, A. W., Strawsburg, R. H., & Holland,S. K. (2004). BOLD fMRI signal increases with age in selected brain regions inchildren. Neuroreport, 15, 2575–2578.

emel, E., Wiig, E. H., & Secord, W. (1987). Clinical Evaluation of LanguageFundamentals—Revised (CELF-RUK) . Sidcup, Kent, UK: The Psychological Corpo-ration.

tark, R. E., Bleile, K., Brandt, J., Freeman, J., & Vining, E. (1995). Speech–languageoutcomes of hemispherectomy in children and young adults. Brain and Language,51, 406–421.

tark, R. E., & McGregor, K. K. (1997). Follow-up study of a right- and a left-hemispherectomized child: Implications for localization and impairment oflanguage in children. Brain and Language, 60, 222–242.

anlancker-Sidtis, D. (2004). When only the right hemisphere is left: Studies inlanguage and communication. Brain and Language, 91, 199–211.

argha-Khadem, F., Carr, L. J., Isaacs, E., Brett, E., Adams, C., & Mishkin, M. (1997).Onset of speech after left hemispherectomy in a nine-year-old boy. Brain, 120,159–182.

argha-Khadem, F., Isaacs, E. B., Papaleoudi, H., Polkey, C. E., & Wilson, J. (1991).Development of language in six hemispherectomized patients. Brain, 114,473–495.

argha-Khadem, F., Isaacs, E., Watkins, K. E., & Mishkin, M. (2000). Ontogenetic spe-cialization of hemispheric function. In J. M. Oxbury, C. E. Polkey, & M. Duchowny(Eds.), Intractable Focal Epilepsy (pp. 405–418). London, UK: W.B. Saunders.

argha-Khadem, F., & Mishkin, M. (1997). Speech and language outcome after hemi-spherectomy in childhood. In I. Tuxhorn, H. Holtausen, & G. K. Boenigk (Eds.),Paediatric epilepsy syndromes and their surgical treatment (pp. 774–784). London,UK: John Libbey & Company Ltd. Medical Books.

argha-Khadem, F., & Polkey, C. E. (1992). A review of cognitive outcome afterhemidecortication in humans. In F. D. Rose & M. H. Johnson (Eds.), Recoveryfrom Brain Damage. Reflections and Directions (pp. 137–148). London, UK: PlenumPress.

erity, C. M., Strauss, E. H., Moyes, P. D., Wada, J. A., Dunn, H. G., & Lapointe, J. S. (1982).Long-term follow-up after cerebral hemispherectomy: Neurophysiologic, radi-ologic, and psychological findings. Neurology, 32, 629–639.

arrington, E. K., James, M., & Maciejewski, C. (1986). The WAIS as a lateralizingand localizing diagnostic instrument: A study of 656 patients with unilateralcerebral lesions. Neuropsychologia, 24, 223–239.

echsler, D. (1992). Wechsler Intelligence Scale for children (3rd ed.). UK (WISC III).Sidcup, Kent, UK: The Psychological Corporation.

echsler, D. (1998). Wechsler Adult Intelligence Scale (3rd ed.). UK (WAIS III). Sidcup,Kent, UK: The Psychological Corporation.

ood, A. G., Harvey, A. S., Wellard, R. M., Abbott, D. F., Anderson, V., Kean, M.,et al. (2004). Language cortex activation in normal children. Neurology, 63,1035–1044.

aidel, E. (1977). Unilateral auditory language comprehension on the token test fol-lowing cerebral commissurotomy and hemispherectomy. Neuropsychologia, 15,1–18.