Recently, Saur, Lange, .... Darren S. Kadis1,2, Elizabeth N. Kerr2, James T. Rutka3,4,. O. Carter ...... Darren S. Kadis, M.A. 1, 2, Mary Lou Smith, Ph.D., C. Psych.
DEVELOPMENTAL PLASTICITY OF LANGUAGE REPRESENTATION IN HEALTHY SUBJECTS AND CHILDREN WITH MEDICALLY INTRACTABLE EPILEPSY
by
Darren S. Kadis
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Psychology University of Toronto © Copyright by Darren S. Kadis 2010
ii Developmental Plasticity of Language Representation in Healthy Subjects and Children with Medically Intractable Epilepsy Darren S. Kadis Doctor of Philosophy Department of Psychology University of Toronto 2010
Abstract This thesis includes four studies designed to improve the ability to predict and assess language representation in healthy children and/or individuals with neurological disorders arising in childhood. In the first study, the role of pathology type on interhemispheric plasticity of language was determined by comparing lateralization in children with developmental, acquired, and tumour pathologies. Findings from 105 consecutive intracarotid sodium amobarbital procedures were retrospectively analyzed, revealing no lateralization differences between pathology groups. In the second study, a novel verb generation paradigm and magnetoencephalography (MEG) were used to determine the spatial-temporal characteristics of language expression in healthy subjects (n = 12) and children with neurological disorders (n = 4). Time-frequency and differential beamformer analyses revealed low-beta event-related desynchronization (ERD) in the left inferior frontal lobe for verb generation. The paradigm was well-tolerated by all subjects.
iii The third study involved assessment of expressive language lateralization in 25 healthy subjects, aged 5-18 years, using two novel MEG paradigms: covert picture naming and verb generation. Novel analyses permitted objective quantification of ERD lateralization on an individual basis. For both tasks, left lateralization of frontal lobe ERD tended to increase with advancing age. Findings suggest that adult-typical left lateralization emerges from an early bilateral language network in normal development. In the fourth study, frontal lobe ERD lateralization for naming and verb generation was characterized in 14 children and adolescents with neurological disorders. Novel analyses permitted objective assessment of individual scans at multiple contrast time windows. In several cases, rapid hemispheric shifts in predominant frontal lobe ERD were observed through the response period. On an individual basis, the assessment protocol showed promise for future use in a presurgical context. These studies serve to advance the understanding of normal paediatric language representation, and improve the ability to predict and assess language lateralization in individuals who have experienced early neurological insults.
iv
Acknowledgements I have much gratitude for my primary supervisor and professional mentor, Dr Mary Lou Smith. Since my initiation into clinical research as an undergraduate, through my final days as a doctoral candidate, Mary Lou has served as a model of academic excellence to which I continue to strive. I am very grateful for all of the guidance provided by Dr Mary Pat McAndrews and Dr Margot Taylor, who along with Mary Lou, comprised my expert doctoral supervisory committee. Thank you Mary Lou, Mary Pat, and Margot. Beyond collaboration, Dr Elizabeth Pang served as an unofficial fourth supervisor / advisor for my MEG studies, and was particularly involved in the clinical implementation and evaluation of our MEG protocols. Travis Mills was extremely helpful with programming and running our MEG paradigms, and was absolutely integral to the development of our novel MEG analyses. Thank you Liz and Travis. The clinical team serving children with refractory epilepsy at the Hospital for Sick Children is fantastic, and I very much appreciate them entrusting me with their patients and data. In addition to Dr Mary Lou Smith and Dr Elizabeth Pang, I gratefully acknowledge all the other clinicians I’ve worked with during my graduate studies. In particular, I thank my doctoral research collaborators, Irene Elliott, Dr Elizabeth Kerr, Dr Lucyna Lach (McGill), Dr William Logan, Dr Ayako Ochi, Dr Hiroshi Otsubo, Dr James Rutka, Dr Carter Snead, and Dr Shelly Weiss. I am very lucky to have met and worked with many of the patients and their families involved in our research studies. Those interactions were the perks in my clinical research. I received funding through a Doctoral Research Award granted by the Canadian Institute of Health Research in collaboration with Epilepsy Canada, and a Doctoral Research Scholarship awarded from the Ontario Student Opportunity Trust Fund through the Hospital for Sick Children. My studies were funded by a research grant from the Behavioural Neurology Group at the University of Toronto. I am fortunate to have great friends both inside and outside of the Hospital. I have immense admiration for my parents, Robert and Vivienne (Bob and Viv), who have gracefully supported me at all times. My sister, Andrea (to me, always Annie), has encouraged me in the way only a close sibling can: her jabs to my kidneys were mild and tempered with the occasional pat on the back. My partner, Arijana, has demonstrated Zen master-like patience in these last years. Whether she likes it or not, she remains my sounding-board, my confidant, and my best friend.
v
Table of Contents Abstract
ii
Acknowledgements
iv
Table of Contents
v
List of Tables
x
List of Figures
xi
Chapter 1: Rationale
1
Chapter 2: Language
3
2.1 Language Acquisition
Chapter 3: Language Representation 3.1 Historical context, seminal works, canonical language regions 3.2 Extra-canonical language regions
4
8 8 16
3.2.1 The basal ganglia and language
17
3.2.2 The cerebellum and language
20
3.2.3 Thalamus
23
3.2.4 Right hemisphere
26
Chapter 4: Classification of language disturbances (aphasiology)
30
Chapter 5: Plasticity
34
5.1 Paediatric advantage
34
5.2 Interhemispheric plasticity
34
5.3 Intrahemispheric plasticity
39
5.4 Mechanism of plasticity
39
vi
Chapter 6: Pathology Type vs Language Lateralization
42
6.1 Study rationale
42
6.2 Abstract
45
6.3 Introduction
46
6.4 Methods
49
6.4.1 Participants
49
6.4.2 Assessment of language laterality
50
6.4.3 Analyses
52
6.5 Results
52
6.6 Discussion
54
6.7 Acknowledgements
59
Chapter 7: Magnetoencephalography
64
7.1 Neuroimaging with MEG
64
7.2 MEG recording
64
7.3 MEG analyses
67
Chapter 8: Expressive Language Mapping in MEG
71
8.1 Study rationale
71
8.2 Abstract
74
8.3 Introduction
75
8.4 Methods
79
8.4.1 Participants
79
8.4.2 Expressive language paradigm
79
8.4.3 Stimulus presentation
80
8.4.4 Anatomical MRI acquisition and co-registration
81
8.4.5 Time-frequency response analyses
82
8.5 Results 8.5.1 Virtual sensor time-frequency response analyses
83 83
vii 8.5.2 Differential beamformer analyses
83
8.5.3 Individual control subjects
84
8.5.4 Control group
84
8.5.5 Clinical case series
84
8.6 Discussion
86
8.7 Acknowledgements
91
Chapter 9: Characterizing Expressive Language Representation Across Childhood
95
9.1 Study rationale
95
9.2 Abstract
98
9.3 Introduction
99
9.4 Method
103
9.4.1 Participants
103
9.4.2 Expressive language tasks, stimuli and presentation
103
9.4.2.1 Covert naming to confrontation
104
9.4.2.2 Covert verb generation to confrontation
105
9.4.3 MEG data acquisition
105
9.4.4 Anatomical MRI acquisition and coregistration
106
9.4.5 Differential beamformer analyses with bootstrap-derived thresholds
106
9.4.6 Region of interest
108
9.4.7 Analyses of ERD for naming and verb generation
108
9.4.7.1 Laterality index for ERD power
108
9.4.7.2 Total and hemispheric extent of ERD
109
9.5 Results 9.5.1 Naming
110 110
9.5.1.1 Laterality index for ERD power for naming
110
9.5.1.2 Extent of ERD for naming
110
9.5.1.3 Comparison of LIERD and LIVOX for naming
111
viii 9.5.2 Verb generation
111
9.5.2.1 Laterality index for ERD power for verb generation
111
9.5.2.2 Extent of ERD for verb generation
111
9.5.2.3 Comparison of LIERD and LIVOX for verb generation
112
9.5.3 Naming versus verb generation lateralization
112
9.6 Discussion
112
9.7 Acknowledgements
118
9.8 Figure captions
121
Chapter 10: Expressive Language MEG in Paediatric Neurosurgical Candidates
125
10.1 Introduction
125
10.2 Method
126
10.2.1 Participants
126
10.2.2 Expressive language tasks
128
10.2.3 Analyses
129
10.3 Results
130
10.3.1 Naming and verb generation within the patient group
131
10.3.2 Concordance of naming and verb generation
131
10.3.3 Effect of seizure side on lateralization
131
10.3.4 Age at seizure onset and lateralization
132
10.3.5 Effect of handedness
133
10.3.6 Concordance with IAP
133
10.3.7 Concordance with fMRI
134
10.3.8 Concordance with receptive language MEG
135
10.3.9 Individual case studies
135
10.4 Discussion
156
ix
Chapter 11: General Summary
162
11.1 Review Objectives
162
11.2 Summary of findings
163
11.3 Implications
165
11.4 Future Directions
169
References
171
x
List of Tables Chapter 6 Table 6.1 Group demographic and seizure-related characteristics
60
Table 6.2 Possible antecedents for individuals in the acquired pathology group
61
Chapter 9 Table 9.1 Participant demographic, performance, and neuromagnetic findings for naming
119
Table 9.2 Participant performance and neuromagnetic findings for verb generation
120
Chapter 10 Table 10.1 Demographic and seizure related data, with performance and lateralization findings for all patients in the study
161
xi
List of Figures Chapter 6 Figure 6.1 Language laterality by pathology group – individuals with left hemisphere seizures only
62
Figure 6.2 Language laterality by pathology group – individuals with left hemisphere seizures and onset before 6 years of age
63
Chapter 8 Figure 8.1 Example of target stimulus and fixation stimulus
92
Figure 8.2 Averaged time-frequency response plot for virtual sensors placed at the superior portion of the left frontal operculum, control subjects only
93
Figure 8.3 Event-related synchrony and desynchrony for control subject verb generation, findings from differential beamformer analyses
94
Chapter 9 Figure 9.1 Example of test and inter-trial stimuli
122
Figure 9.2 Scatterplot with linear trendline for LIERD versus age at assessment for the naming task
123
Figure 9.3 Scatterplot with linear trendline for LIERD versus age at assessment for the verb generation task
124
1
Chapter 1: Rationale The purpose of this thesis is to address a number of recently identified problems related to language mapping in children with and without neurological insult. The research questions I address have clinical relevance; the choice of studies and designs reflect an interest of serving patients. In particular, this work is intended to ultimately benefit children undergoing investigations to determine surgical candidacy for treatment of medically intractable epilepsy. In epilepsy surgery, two principal goals compete to establish candidacy and surgical approach: seizure amelioration, accomplished through full resection of seizure-generating tissue, and preservation or promotion of psychological/neurological function, realized through sparing of eloquent tissue. The studies presented in this thesis are designed to help us predict or assess language lateralization and localization in childhood. To advance our understanding of how children’s brains represent language, and to explore the relationship between brain injury and functional plasticity, we study both healthy children and those with chronic neurological conditions. It is generally agreed that the paediatric brain is more plastic for language than the adult brain; however, no consensus exists on the cause(s) or mechanism(s) of plasticity. In our research involving the intracarotid sodium amobarbital procedure (IAP), we assess the effects of pathology type on language lateralization in children with intractable epilepsy. The primary goal of the study was to identify and/or rule out predictors (causes) of typical and atypical language lateralization, which immediately benefits paediatric neurosurgical candidates.
2 Our appreciation of the mechanism(s) for paediatric language plasticity is relatively poor. Two competing theories have been proposed to explain atypical lateralization (e.g., Rasmussen & Milner, 1977) and atypical localization (e.g., Kadis, Iida, Kerr, et al., 2007) observed in the context of early insult. The first theory posits that children’s language networks look much like adult language networks. According to this theory, adult-atypical representation following early insult occurs through shifts or recruitment of neural tissue that is not typically involved in the control of language. The second theory holds that children’s language networks are relatively bilateral and extensive, transitioning to focal and left lateralized representation as part of the normal developmental trajectory. The second theory explains adult-atypical representation following early insult as a function of dynamic representation across childhood. Each theory has received support through neuroimaging research in recent years. Our language studies using magnetoencephalography (MEG) were designed to help us characterize language representation across early development, and address questions regarding the mechanism for language plasticity following early neurological insult. In the chapters that follow, I review the unique features of human communication, the history of aphasia study and language mapping, and the literature on plasticity of language representation and neuroimaging with MEG. The general introduction is intended to provide context for the four research studies presented in this dissertation.
3
Chapter 2: Language A consensus on the definition of language remains elusive, although most theorists seem to agree that human language differs from communications observed in other animals, in important ways. Many species communicate through sound and gesture, and there is evidence that various animal groups encode and transmit information, symbolically. Honey bees effectively transmit spatial maps through a gestural ‘waggle’ or ‘dance’ (see, von Frisch, 1974), and vervet monkeys exhibit distinct ‘alarm calls’ to indicate the presence of different types of predators, each eliciting distinct and appropriate responses from conspecifics (Struhsaker, 1968, in Arcadi, 2005). These language-like behaviours, each critical to survival, are indeed evidence of symbolic communication in other animal groups; however, distinctions from human language, remain. Compared to humans, the repertoires of animals are relatively small, and the manner of delivery and receipt of each communication is relatively fixed (see Arcadi, 2005). An English user can call upon nearly 500,000 words to form an expression; most of the words are arbitrarily related to the objects, concepts, or functions, they represent. In many cases, a word may have multiple meanings which may be modulated by the context in which it is being used. Words are transmitted verbally, gesturally (colloquially, or through standardized signing), and through writing. Most importantly, human language is rule-based - mastery of the rules of grammar permits generation of an infinite number of novel word combinations, making human language uniquely complex and powerful among animal systems of communication. At the core of human language is verbal communication. All human groups participate in spoken discourse as the dominant mode of transmission in person-to-person interactions.
4 Linguists study spoken language by breaking down and analyzing sentences and words at the constituent level. The most basic unit of speech is the phoneme, or smallest unit of sound in a given word. Each language employs its own set of phonemes; the average English speaker uses approximately 40 of these elements, regularly. Phonemes are purely sound units, which don’t necessarily convey meaning. The smallest unit of sound that conveys meaning is called a morpheme. A phoneme may also be a morpheme, depending on the context. Consider a pluralizing terminal ‘s’ sound applied to a concrete noun (e.g., as “chair” becomes “chairs”): the ‘s’ serves as both a basic unit of sound and conveyer of meaning. These elements are the building blocks for words. A lexicon is the complete collection of words in a particular language, and a personal lexicon is all the words known to an individual; as previously noted, there are nearly half a million words in the complete English lexicon. The lexical units can be combined to form sentences, based on the rules of grammar, or syntax. Adjustments in pitch, volume, and speed of delivery are collectively termed prosody. Prosody is used to provide emphasis in speech, and promotes fluidity by providing word segmentation cues for the listener (in English, initial syllables are typically stressed, allowing the speaker to deliver words in rapid succession while still being understood). Skilled speakers integrate all these elements to convey arguments and narratives (stories) in discourse, the free-form give-and-take verbal exchange typically observed between two or more individuals. 2.1 Language Acquisition Humans acquire language early in life in a progressive, predictable pattern; typical language development is approximately time-locked across cultures (see Kuhl, 2004), with some degree of variability between individuals. By one year of age, children generate
5 predominantly language-specific speech sounds, prefer to listen to their ‘native’ languages, and have often begun to produce their first words. In the months that follow their first birthdays, children experience considerable growth in their vocabularies, begin to form sentences, and ultimately engage in discourse by 3 years of age. Indeed, the complexities of language are learned rapidly in the first years. The study of early language learning has revealed a necessary interaction of biological and environmental endowments for successful language development. From the first moments following birth, humans can distinguish speech from non-speech sounds (Vouloumanos & Werker, 2007), and by two months of age, prefer speech sounds over synthesized analogues (Vouloumanos & Werker, 2004). Unlike adults, newborns and young infants are able to discriminate among phonemes from all languages (Eimas, Siqueland, Jusczyk, & Vigorito, 1971; Streeter, 1976; Trehub, 1976; Werker & Tees, 1984), suggesting an early biological (perceptual) bias for language. This generalized ability to discriminate phonemes of all languages does not persist, however. Researchers have observed marked decreases in cross-language phonemic discrimination by 6 months of age, and by 12 months, phonemic discrimination is specialized to the language or languages to which the child has been regularly exposed (Kuhl, Stevens, Hayashi, et al., 2006). The selective ‘tuning’ to a native language or languages is believed to be critical for normal language development, as it provides a necessary foundation for word segmentation (see Werker, 2003). With attunement for sounds of their native language, infants begin to discover highfrequency phonemic pairings and sequences through statistical learning processes (e.g., Saffran, Aslin, & Newport, 1996), promoting the formation of word prototypes. In most
6 continuous speech (e.g., discourse), the breaks between words are only implied – silences are equally likely to occur within words as between words (see Werker, 2003; Kuhl, 2004). A familiarity with prototypical pairings helps infants identify particular sound combinations as word units. Infants also use prosodic cues to establish prototypes used in word segmentation: in spoken English, the initial syllable of a word typically receives emphasis, informing the listener of the breaks between words (this strong-weak syllabic pattern is known as trochaic, and is contrasted by weak-strong patterns of some other languages, which are known as iambic). Infants prefer words conforming to the emphatic pattern of their native language by 9 months of age (Jusczyk, Cutler, Redanz, 1993). Between the ages of 7 months and 10 months, infants segment words based on statistical information about phonemic pairing and prosodic regularities; from 10 months on, infants begin to segment atypical word forms, as well (Jusczyk, 2002; Vihman, Nakai, DePaolis, & Hallé, 2004). Speech segmentation serves as a necessary precursor for vocabulary development (Graf Estes, Evans, Alibali, & Saffran, 2007; see also, Brent & Siskind, 2001). Infants must be able recognize sound sequences as word candidates (i.e., based on conformity to prototypical speech patterns), before associating those sounds with meanings. Graf Estes et al. tested 17 month old infants on object-label learning tasks, and found that ‘words’ conforming to statistical expectations (i.e., containing high-frequency phonemic pairings and prosodic regularities) were easier to pair with objects than nonconforming words. Indeed, infants are most likely to associate meaning to words that are easiest to segment from fluent speech (Graf Estes et al., 2007; see also, Swingley, 2005, 2007). In this way,
7 infants build their vocabularies, acquiring knowledge of word-meaning associations in parallel to developmental changes in word segmentation ability. As children continue to develop their vocabularies, they begin to appreciate the pairings of words used by others, and attempt simple sentence formation of their own (e.g., a child may pair a well-learned verb and noun to produce, “throw ball”). Interactions with adult speakers, who provide examples of sentences and typically ‘correct’ (syntactically) the speech of the infant, are imperative to successful language learning (Chang, Dell, & Bock, 2006; see also, Kuhl, 2004). The impact of social interaction may be difficult to quantify, although its significance to normal language acquisition has been well established. Discourse-like turn-taking in adult-infant verbal exchange promotes language-like vocalizations from children as young as three months old (Bloom, Russell, & Wassenberg, 1987), and maternal attention and feedback (e.g., gaze, facial expression, contact, etc.) to infant babbling determines the quality of vocalizations in older infants (Goldstein, King, & West, 2003). Adult-infant interactions help shape the development of grammar, which infants acquire, implicitly; by two years of age, toddlers are able to appropriately distinguish transitive (object-dependent) and intransitive (object-independent) forms of novel verbs (Naigles, Bavin, & Smith, 2005), without receiving explicit training in complex syntax.
8
Chapter 3: Language Representation 3.1 Historical context, seminal works, canonical language regions Most of what is known about language representation in the brain has been learned from studying individuals who have experienced discreet cerebral insults, typically in adulthood, resulting in selective language impairments (acquired aphasias). The French physician Jean-Baptiste Bouillaud (1796-1881), who retrospectively analyzed the case files of his contemporaries, was the first to argue from clinical observation that expressive language functions are supported exclusively by the frontal lobes (Bouillaud, 1825). The observations came at a time when equipotentiality and functional localization were each considered viable in explaining the relationship between brain and behaviour. Bouillaud’s findings were consistent with arguments for frontal lobe control of motor functioning already championed by practitioners of phrenology, including Franz Gall (1758-1828), Bouillaud’s mentor. In his 1825 report to the Royal Academy of Medicine, Bouillaud personally attacked the famous equipotentialist, Jean Pierre Flourens, for ignoring “a mass of facts” demonstrating both neural control of musculature and consistent focal representation (see, Bouillaud, 1825). Bouillaud’s fervent argument for functional localization seriously challenged supporters of equipotentility, inspiring rigorous study of brain-behaviour relationships, nurturing the growth of behavioural neuroscience. Interestingly, Bouillaud has recently been criticized for selecting cases fitting his expectations and ignoring a substantial set of contradictory cases where language disturbance and frontal lobe injury did not co-occur; Bouillaud is also criticized for failing to report on the relatively obvious relationship between language and laterality – left
9 hemisphere insult was documented in essentially all cases of language disturbance available to Bouillaud (Luzzatti & Whitaker, 2001). Consistent establishment of left hemisphere language dominance remains one of the strongest arguments against the equipotential brain. Perhaps the most well-known example of a case study advancing our understanding our language localization is that of Pierre Paul Broca’s (1824-1880) patient, Leborgne (see, Broca, 1861a; 1861b). Leborgne was admitted to the Bicêtre Hospital in Paris at age 30, having lost the ability to speak only a couple of months prior. Leborgne had suffered from epilepsy for a number of years (age at onset and semiology are unknown), but had led a relatively normal life with gainful employment up until the time of his admittance. When he arrived at Bicêtre, the hospital staff noted that Leborgne appeared healthy and intelligent and clearly understood the speech of others. His non-verbal responses to questions and directions were always appropriate, and he was reportedly adept at conveying information through gesticulation. Leborgne’s attempts at speech were mostly unsuccessful, however; his utterances were limited to, “tan, tan”. Because of his principally unvaried vocalizations, Leborgne became known as patient ‘Tan’. Interestingly, Leborgne would occasionally become frustrated with other patients or with staff who could not interpret his gestures; his frustration tended to develop into rage, and he would spout the single brief curse, “Sacre nom de Dieu”, demonstrating that ‘automatic’ speech was still possible, that his ability to articulate remained intact (i.e., the expressive language impairment was not secondary to an apraxia), and that the impairment was limited to the specific domain of voluntary speech production. Broca reports that Leborgne remained stable for ten years, suffering only from “aphemia” (a term introduced to describe Leborgne’s acquired poverty of speech; we now use the term
10 “aphasia” in its place). At age 40, Leborgne began to develop motor impairments of the right arm and leg, which progressed for another 11 years, ultimately rendering him bedridden. The patient died at age 51, after receiving surgical treatment for a gangrenous right leg. At autopsy, Broca recovered Leborgne’s brain and fixed it in alcohol for gross examination (it remains at the Musée Dupuytren in Paris, today). Broca noted a cystic lesion, large enough to accommodate a chicken’s egg, in the posterior inferior region of the left frontal lobe. The progressive nature of Leborgne’s conditions suggested a gradual growth of the cyst over a period of more than 20 years. The tissue around the lesion was soft, not compressed, suggesting to Broca a gradual cortical atrophy related to cystic infiltration. Although diffuse abnormalities were also observed throughout the left hemisphere (see also, Castaigne Lhermitte, Signoret, & Abelanet, 1980; Signoret, Castaigne, Lhermitte, Abelanet, & Lavorel, 1984; Dronkers, Plaisant, Iba-Zizen, & Cabanis, 2007), Broca attributed Leborgne’s condition to the cyst, and argued that the center of mass of the lesion would indicate its point of origin and the focal insult responsible for Leborgne’s initial clinical presentation with loss of speech (Broca did not comment on the source of Leborgne’s epilepsy). Broca concluded that the third convolution of the left inferior frontal gyrus (pars opercularis) is the seat of the faculty of articulated language. Broca cautioned that the initial focus may have also involved the second convolution of the inferior frontal gyrus (pars triangularis), but felt that the third convolution was probably most important to speech. Over time, this left posterior inferior frontal region, corresponding primarily to Brodmann area 44 (the operculum corresponds to area 44, pars triangularis corresponds to area 45), has become known as Broca’s area.
11 In the months that followed his initial investigations of patient Leborgne, Broca encountered additional patients suffering from expressive aphasias; in all cases, post mortem analyses revealed lesions involving the third convolution of the inferior frontal gyrus of the left hemisphere (Broca, 1861b; 1865). These later observations supported the proposed role of the left operculum in voluntary speech, firmly establishing ‘Broca’s area’ as responsible for expressive language functioning. Individuals with acquired expressive language deficits, often termed ‘Broca’s aphasias’, may experience various degrees of difficulty with naming, word finding, grammar, and general language production (e.g., speech loses fluidity and production becomes onerous). A little more than a decade after Broca published his findings on expressive language localization in the brain, the German physician, Carl Wernicke (1848-1905), wrote about a group of patients with language disturbances distinct from those previously documented by Broca (Wernicke, 1874). Whereas Broca’s patients could not speak, or produced little speech with great encumbrance, Wernicke’s patients readily produced fluid (wellarticulated, rapid, prosodic) speech without any apparent distress. Though produced in abundance, the speech of Wernicke’s patients lacked meaning and/or coherence; in addition, these patients were unable to understand the speech of others (Wernicke, 1874; see also Geschwind, 1964; 1970). Wernicke noted that his patients tended to make word substitutions, appeared to randomly string words together, and/or used nonsense words in place of actual words when speaking. His patients frequently responded inappropriately to the verbal questioning of examiners and seemed unable to follow simple verbal directions speech comprehension, or receptive language, was lost. Post-mortem analyses of the patient’s brains revealed lesions substantially caudal to Broca’s area, in the posterior
12 superior region of the left temporal lobe; this posterior language region is now known as Wernicke’s area. Wernicke was less specific and consistent than Broca in his description of the neuroanatomical substrate for the aphasias he had documented. In his 1874 publication, Wernicke described the posterior language region as involving only the first (caudal) convolution of the superior temporal gyrus; in the diagram that accompanied his description, Wernicke identified the language area by a single point, presumably marking the centre of the language region, in the middle of the superior temporal gyrus (a printer’s error placed this central point on a drawing of the right hemisphere; Bogen & Bogen, 1976). Several years later, Wernicke depicted the language region as considerably more extensive, including both the first temporal gyrus and superior aspects of the middle temporal gyrus of the left hemisphere (Wernicke, 1881). In the decades that followed, others continued to include surrounding neuroanatomy in their descriptions of Wernicke’s area, typically incorporating the angular and supramarginal gyri, and occasionally including inferior temporal cortex as well inferior Rolandic regions (for an excellent account of the variability in published localizations of Wernicke’s area over the course of a century, see Bogen & Bogen, 1976). The tendency seems to have been to expand the boundaries of Wernicke’s area to include extrafrontal cortical regions whose stimulation or ablation resulted in some form of language disturbance. However, the expansion of neuroanatomical delineations was not met with equal development of aphasia nomenclature. Although clinicians had documented distinct language functions associated with various regions of the temporal and parietal lobes (including naming and reading; for an early review of the evidence, see Penfield & Roberts, 1957), the term Wernicke’s
13 aphasia still referred to the specific aphasic presentation of Wernicke’s earliest patients: fluent nonsensical speech with comprehension deficits. Perhaps in an effort to inject parsimony into the growing field of aphasiology, Norman Geschwind (1926-1984) depicted Wernicke’s area as only a small portion of the posterior superior temporal gyrus (e.g., Geschwind, 1970; 1972), in a manner consistent with Wernicke’s initial description (1874). Geschwind knowingly excluded the inferior temporal lobe which was thought to specifically support object naming (the region was once referred to as Mills’ naming centre or Nielsen’s language formulation area; Penfield & Roberts, 1957) and the angular and supramarginal gyri, generally believed to be important for reading (collectively, the left angular and supramarginal gyri comprise Dejerine’s area; see Dejerine, 1892; Geschwind; 1965), probably because these functions were not considered central to a differential diagnosis of Wernicke’s aphasia. Geschwind’s narrow, focal depiction has largely persevered in the literature; most researchers identify Wernicke’s area as a small region, centered over the posterior superior temporal lobe. The boundaries of Wernicke’s area remain undefined, and the inconsistencies in the earliest literature essentially guarantee variability among researchers today. In addition to identification of a posterior language region involved in word selection and comprehension, Wernicke made major contributions to cognitive neuroscience, particularly in the area of aphasiology, by offering a neuroanatomical model and theoretical framework for language processing. Based on his and colleagues’ clinical observations, Wernicke proposed that normal discourse is made possible through two distinct neural regions and processes: an anterior region (Broca’s area) is responsible for the transformation of mental language into fluent speech, and a posterior region (Wernicke’s area) is responsible for
14 associating meaning with spoken language (Wernicke, 1874; Geschwind, 1970). Injuries to the anterior region prevent the coding of ideas into spoken words and sentences, while comprehension remains intact. Injuries to the posterior region prevent comprehension of others’ speech and interfere with the ability produce meaningful speech, because ideas and concepts cannot be associated with spoken words. According to Wernicke’s theory, spoken words are associated with ideas or concepts; the roles of the anterior and posterior language regions are to code mental language into and out of speech forms, respectively. The anterior-posterior dichotomy of the model fit well with neuroanatomical research carried out around the time of Wernicke’s writing. The motor speech cortex, Broca’s area, was confined to the posterior frontal lobe; in 1870, Fritsch and Hitzig had reported on their experiments involving electrical stimulation of the brains of dogs, finding exclusive frontal lobe control of voluntary-type motor functions. Similarly, Wernicke’s receptive language area, which transformed or interpreted speech, was localized to the superior posterior temporal lobe; in 1855, Heschl had described the role of the neighboring transverse temporal gyrus in primary auditory processing. The localization of language regions in close proximity to primary motor and auditory cortices suggested a logic and efficiency in cerebral organization that remains attractive to clinicians and researchers seeking simple models of representation and processing. Wernicke also noted that in normal discourse, an individual has to engage in both receptive and expressive language processes in a coordinated manner. He reasoned that the two primary language regions were connected by a subcortical loop (the cortical perisylvian regions that separated the frontal and posterior temporal lobes were already shown to support other processes, so the connection was necessarily subcortical), what we now
15 recognize as the superior longitudinal fasciculus, or the arcuate fasciculus. The subcortical loop was thought to bridge the two language processing centers, allowing mental language representations to be transmitted accurately to the expressive language cortex of the frontal lobe; indeed, the arcuate fasciculus is composed of a large bundle of axons connecting the frontal, parietal, and temporal lobes, with relatively greater density and extent in the left hemisphere near canonical Wernicke’s area (see Nucifora, Verma, Melhem, Gem, & Gur, 2005; Powell, Parker, Alexander, et al., 2006). Wernicke had not documented language disturbances resulting from injury to the subcortical structures, but did predict deficits that would follow from injury to the subcortical tract. Disconnection of the left inferior frontal and posterior temporal cortices should spare both speech comprehension and fluency, but interfere with the transmission of ideational representations constructed in the posterior temporal lobe to the expressive language cortex of the inferior frontal lobe, resulting in a conspicuous impairment in word and sentence repetition. A few years after Wernicke postulated that subcortical tract lesions could lead to aphasias by way of interrupted transmission within the normal language network, Ludwig Lichtheim documented the first case of “conduction aphasia” in an individual with injury of the left arcuate fasciculus (1885). Lichtheim’s patient could speak fluently, yet produced the sort of paraphrasic errors (substitution of related words for target words) seen in individuals with posterior superior temporal lesions. Lichtheim’s patient could be easily differentiated from an individual with Wernicke’s aphasia, as conduction aphasia spares speech comprehension. Lichtheim documented the predicted specific impairment for word and sentence repetition; difficulty with speech repetition is
16 pronounced compared to other features of conduction aphasia, and remains the diagnostic hallmark of the disorder (see Geschwind, 1970). Wernicke’s two-component model could explain the language deficits observed in individuals with known discreet cerebral insults, and importantly, was useful in predicting the localization of lesions based on the patients’ presentations. So impressive was the predictive utility of the model – exemplified in the prediction of conduction aphasia – that Wernicke’s model remained principally unchanged and largely unchallenged for nearly a century. Although findings from stimulation mapping in the 1950s (Penfield & Roberts, 1957) and more recently lesional and neuroimaging studies have demonstrated the importance of other brain structures, Wernicke’s original two-component model remains a core component of neuropsychology, neurology, and speech and language pathology curricula. Wernicke’s ideas were well-championed by Lichtheim, and survived in the more recent research and writing of Norman Geschwind – as such, the two-component model is sometimes referred to as the Wernicke-Lichtheim-Geschwind model of language. 3.2 Extra-canonical language regions Since the earliest language writings of language localization (e.g., those of Bouillaud, Broca and Wernicke), scientists have challenged the notion that the brain supports language exclusively within discreet cortical regions. John Hughlings Jackson, a proponent of theories of functional localization, was probably the first to argue that subcortical regions significantly contribute to normal language processes. Hughlings Jackson (1866) suggested that lesion proximity to the basal ganglia could be used to predict the nature of language disturbance: proximal and encroaching lesions would affect articulation, and lesions distal
17 to the basal ganglia would result in word selection errors. The popular WernickeLichtheim-Geschwind model fails to characterize the contributions of distributed regions of the brain; we now appreciate that language is not processed in a neatly-parceled twocomponent fashion, and have begun to understand the relative and important contributions of ‘extra-canonical’ language regions. 3.2.1 The basal ganglia and language That the basal ganglia play a role in language processes should not be surprising, as the majority of these structures (caudate nucleus, putamen, globus pallidus, claustrum, subthalamic nucleus, nucleus accumbens, and the amygdala) are known to form circuits involved in the control of motor behaviour, including oral-facial movement necessary for fluent and well articulated speech. It is not clear whether the basal ganglia’s contribution to language is limited to motor effects, however. A number of recent lesion, invasive treatment,and neuroimaging studies suggest that the constituent structures of the basal ganglia may each play specific roles in language processing, independent of the motor aspects of speech production. The functional importance of the basal ganglia is classically inferred from study of populations with known acquired basal ganglia insults or damage to systems that modulate basal ganglia activity, particularly individuals with Parkinson’s or Huntington’s disease. Parkinson’s disease results from damage to the dopamine producing cells of the substantia nigra; the dopaminergic projections of the substantia nigra modulate activity of the basal ganglia structures, which project to the thalami and motor cortices to promote voluntary and suppress involuntary movements. Degeneration of the substantia nigra is associated
18 with both akinesia (lack of movement) and dyskinesia (involuntary, excessive movement), depending on how the basal ganglia are affected. In Huntington’s disease, progressive degeneration of the striatum (the caudate and putamen) leads to various dyskinesias; individuals may present with tremors or even jerky (choreiform) or writhing motor behaviours (athetoid). In both Parkinson’s and Huntington’s, speech is often slurred, prosody may be lost, and fluency is reduced. The effects of basal ganglia insult on language function are typically interpreted as secondary to motor dysfunction (e.g., Radanovic & Scaff, 2003), although recent studies are challenging that interpretation. Gil Robles, Gatignol, Capelle, et al. (2005) studied individuals undergoing stimulation mapping for resection of cortico-subortical tumours of the left hemisphere, and found that stimulation of the left putamen, but not the caudate, produced dysarthria. Stimulations of the left caudate produced perseverations, which the authors interpreted as evidence of nonmotor contribution of basal ganglia to language. Others have documented dysprosodic yet well-articulated speech in individuals with bilateral or right putaminal infarcts (Van Lanker Sidtis, Pachana, Cummings, & Sidtis, 2006; see also, Riecker, Kassubek, Gröschel, et al. 2006), suggesting that the subcortical structures contribute to language beyond the motor domain. There appear to be contrasting effects of left putamen stimulation versus bilateral or right putaminal infarcts; at this time, the role of the left putamen in articulation and prosody cannot be conclusively determined. Parsons, Rogers, Braaten et al. (2006) conducted a meta-analysis of the cognitive effects of subthalamic nucleus deep brain stimulation (DBS) used to treat Parkinson’s disorder. The researchers documented consistent small negative effects of DBS on executive functioning and verbal memory, and consistent moderate and negative effects of DBS on verbal
19 fluency. Others have shown that the declines in verbal fluency appear to be relatively robust and long-lasting, while non-linguistic cognitive declines are relatively mild and transient in subthalamic DBS (Zangaglia, Pacchetti, Pasotti, et al., 2009). Findings indicate a specific role for the subthalamic nuclei in expressive language processing. Narayana, Jacks, Robin et al. (2009) used PET to show that ‘unintended’ activation of the premotor cortex occurs with subthalmic DBS; virtual lesions with TMS to the activated premotor cortical region were shown to impair language expression, suggeting that negative effects of subthalamic DBS are moderated by basal ganglia connectivity to motor regions of the cortex. Some researchers have suggested that the basal ganglia structures contribute to language function in an indirect manner; the effects of lesions on language result from either shared vasculature or structural connectivity. Hillis, Barker, Wityk et al. (2004) found that variability in language dysfunction following left caudate infarct could be explained by corresponding left cortical hypoperfusion, rather than left caudate integrity. Specifically, the degree of left anterior perisylvian hypoperfusion was associated with nonfluent aphasic symptoms, and the degree of left posterior perisylvian hypoperfusion was associated with symptoms of fluent aphasias (see also, Nadeau & Crosson, 1997). Hillis et al. suggest that the effects may be due to large vessel narrowing (stenosis) associated with the deep infarct, rather than the direct effects of lesions to the caudate. Henry, Berman, Nagarajan et al. (2004) used DTI to demonstrate structural connectivity between the putamen and inferior frontal cortical sites deemed language-positive in an individual who underwent cortical stimulation mapping. Henry et al. found extensive connectivity among the cortical sites, and between cortical sites and subcortical nuclei. Deep lesions may involve cortical-
20 cortical or cortical-subcortical pathways independent of the basal ganglia or cortical damage; effects of disconnection should be distinguished from deep nucleic and cortical contributions. Further, Henry et al. found that the putamen connected to sites related to speech arrest and oral-motor control, supporting the role for the left putamen in modulation of speech fluency and articulation; motor effects of basal ganglia on language cannot be ruled out. In cases of both transient and chronic dysphasia following basal ganglia insult, we might consider cortical-cortical disconnection or subcortical-cortical diaschesis as the mechanism for dysfunction; direct effects of basal ganglia lesions on language, independent of motor effects, have not been conclusively demonstrated (see Nadeau & Crosson, 1997). 3.2.2 The cerebellum and language The effects of cerebellar injury on balance and fine motor control (timing/planning and coordination) have been studied for nearly two hundred years (beginning with experiments by the famous vivisector, Francois Magendie; in Miura, 1923; Luciani, 1891, in Manni & Petrosini, 1997; for an excellent review of cerebellar control of adaptive coordination, see, Thach, Goodkin, & Keating, 1992). Although scientists have long proposed a relationship between cerebellar integrity and speech production, the effects of cerebellar lesions on language have long been interpreted as secondary to oral-motor deficits (see Miura, 1923). In recent years, researchers have inferred cerebellar contributions to normal language processes, arguing for effects independent of motor functioning, primarily through studies of individuals with cerebellar lesions (e.g., Molinari, Leggio, & Silveri, 1997). In the 1980s, cases of ‘cerebellar mutism’ (a non-fluent aphasia, with intense poverty of speech and later dysarthria) were documented in children who underwent surgery for
21 tumours of the posterior fossa. These aphasic presentations were considered secondary to motor coordination or planning/initiation deficits. In most cases of cerebellar mutism, symptoms resolve within several weeks of surgery, possibly due to a settling of edema related to the resection (van Dongen, Catsman-Berrevoets, & van Mourik, 1994), although persistent symptoms have been reported at one year post-surgery and beyond (Turgut, 2008; see also De Smet, Baillieux, Catsman-Berrevoets et al., 2007; Huber, Bradley, Spiegler, & Dennis, 2006). Return of speech is typically interpreted as recovery from surgical aggravation, not as a function of plasticity, probably due to the onset at resection (i.e., not due to presence of tumour) and relatively transient nature of the condition. However, functional reorganization in recovery has not been ruled out, as investigations into mechanisms for speech return have not been conducted. The incidence and impact of mutism and dysarthria in children undergoing infratentorial surgery is not entirely clear, although recent retrospective studies are helping to establish risk and prevalence, and course for recovery. Most research in this area relies on small case series methods, and findings are frequently inconclusive (e.g., Frank, Schoch, Hein-Kropp, et al., 2007; see discussion of Frank, Schoch, Ricther, et al., 2007). Ozimek, Richter, HeinKropp et al. (2004) suggest that two types of mutism may follow from cerebellar resections: the first type involves the dentate nuclei (deep within the cerebellar hemispheres) and results in dysarthria, and the second type involves the cerebellar vermis, with language disturbance co-occurring with significant behavioural disturbances. This proposed differentiation is based on investigations of only 4 subjects, and requires further investigation. In both dentate and vermal tumour resections described, mutism may be secondary to motor and behavioural abnormalities. Turgut (2008) has studied the largest
22 group of children who underwent surgery for resection of cerebellar tumours (n = 450 in a multi-centre study), and reports that nearly a quarter met diagnostic criteria for mutism, post-surgically. Brainstem infiltration was the only factor identified as increasing risk of post-operative aphasia. De Smet, Baillieux, Catsman-Berrevoets et al. (2007) conducted a meta analysis on cases of cerebellar mutism reported from the 1970s through 2006, and found that 98.8% of those with confirmed mutism went on to experience dysarthria of varying severity and duration, with a significant number of individuals experiencing persistent symptoms at several years post surgery. Collectively, findings suggest that children undergoing surgical resection of cerebellar tumours are at significant risk for both transient and persistent speech deficits, although the mechanism for cerebellar mutism remains unclear. Aside from the identification of brainstem infiltration as a risk factor for post operative aphasia (Turgut, 2008), other attempts at assessing specific brain-behaviour relationships in children undergoing surgical resection for cerebellar tumours have not revealed any significant patterns (Richter et al., 2005; see also Richter, Aslan, Gerwig et al., 2007). The failure to isolate focal cerebellar regions and their contribution to language in children with tumours does not necessarily indicate a lack of discrete organization for language in the cerebellum. We know that the cerebrum is plastic for language representation, particularly in children (e.g., Rassmussen & Milner, 1977); if the cerebellum is developmentally plastic for language, it may be difficult to establish normal functional localization though study of individuals with chronic lesions (i.e., tumours). Studies of individuals with acute cerebellar injuries have revealed moderately focal contributions to language processes. In adults experiencing cerebellar infarcts, lesions of
23 the right cerebellar hemisphere are associated with impaired verbal fluency and agrammatic speech, but not lesions of the left cerebellar hemisphere (Silveri et al, 1994; Richter, Gerwig, Aslan, et al., 2007). Similarly, children and adolescents with acute right cerebellar lesions experience reduced fluency in speech, although the dysfunction appears relatively moderate in early injuries (Frank, et al., 2007). Since the majority of efferent fibers from the cerebellar hemispheres decussate to innervate the contralateral midbrain (specifically, the contralateral red nucleus, thalamus, and cerebral cortex (including the inferior frontal gyrus; Booth, Wood, Lu, et al., 2007)), the effects of deep right cerebellar insult may be understood in terms of impoverished or altered input to the dominant left cerebrum. The cerebellar afferent and efferent projections are classically understood as responsible for coordination of motor behaviour (via pyramidal and extra-pyramidal pathways), so reduced verbal fluency and agrammaticism can be understood as the result of oral-motor apraxia or planning/sequencing deficits. However, direct effects on language processes cannot be ruled out, as researchers have documented changes across a wide range of cognitive and psychological domains in individuals with acquired cerebellar lesions (see, Schmahmann & Sherman, 1997; 1998). These widespread cognitive and psychological changes are difficult to explain in terms of motor dysfunction, suggesting a more direct role for the cerebellum in high cognitive processing (Baillieux, De Smet, Paquier, et al., 2008). 3.2.3 Thalamus Penfield and Roberts (1959) were the first to argue from extensive surgical and stimulation mapping experience that the thalami are involved in language processes and other higher cognitive functions. Penfield and Roberts observed the extensive connectivity of the thalamus and basal ganglia with cortical regions known to contribute to normal language
24 functioning, and proposed that the deep structures were important for both receptive (sensory) and expressive (motor) aspects of language. Specifically, the pulvinar (caudal thalamus) was observed to connect with the superior posterior temporal and parietal cortices, and the anterior thalamic nuclei to frontal cortices involved in oral-facial movements and speech formulation (i.e., Broca’s area). The thalamus was thought to form a crucial node in the relay of both sensory and motor information; this is the classical view of thalamic function and its involvement in higher cognitive functioning, which remains relatively intact today. The study of individuals undergoing neurosurgery for treatment of movement disorders has supported Penfield and Roberts findings of thalamic importance for language functioning. Bell (1968) studied 50 patients with Parkinson’s disease who were treated with strategic ventral lateral thalamic lesions, and observed various language disturbances following surgery. Interestingly, left (but not right) thalamic lesions were associated with both dysarthria and naming difficulties, suggesting a lateralized effect. Poor articulation can be explained as a symptom of motor dysfunction; however, the observation of anomia following left ventral lateral thalamic lesions indicates a unilateral effect on language, independent of motor function. In addition to anomia, researchers have reported deficits in speech fluency, comprehension, and word repetition following left thalamotomy (Bruce, Foote, Rosenbek, et al., 2004). The lateralized thalamic effect is not limited to the ventral lateral nuclei. Ojemann, Fedio, and Van Buren (1968) studied individuals undergoing thalamotomy with pre-resective stimulation mapping for treatment of various movement disorders. Twelve subjects received pulvinar stimulation without reporting evoked sensory phenomena. Within this
25 group, stimulation of the left pulvinar produced anomia with otherwise-intact (fluent) speech in right-handed subjects (n = 5), whereas right pulvinar stimulation failed to elicit anomia in any right-hander (n = 6). One left-handed subject receiving right pulvinar stimulation demonstrated anomic speech, suggesting that the thalamus is lateralized for language, with potential for contralateral representation in left-handers (which may represent intrahemispheric plasticity for thalamic contributions to language). The thalamic stroke literature generally supports a role for the left thalamus in language processing. Individuals with left hemorrhagic infarcts experience the full gamut of fluent and nonfluent aphasic symptoms, including paraphrasias, jargon, anomia, and even stuttering (e.g., Cappa & Vignolo, 1979; Mohr, Walter, & Duncan, 1975; in Bhatnagar & Mandybur, 2005; Van Borsel, van der Made, & Santens, 2003; Radanovic & Scaff, 2003). The majority of studies involve adult subjects, but similar aphasic symptoms have been observed in paediatric populations with thalamic injury, as well (Nass, Boyce, Leventhal, et al., 2000; Gout, Seibel, Rouvière, et al., 2005). In a similar vein as for basal ganglia structures, the effects of thalamic integrity on speech and language are often interpreted as secondary to motor or sensory dysfunction, or as indirect effects on a classic language network. Researchers have argued that thalamic injury results in dysregulation of cortical activity in language-relevant regions; language deficits follow from dysregulation of canonical perisylvian regions (Metz-Lutz, Namer, Gounot, et al., 2000; Crosson, 1999; Nadeau & Crosson, 1997). Others have argued that language deficits are secondary to memory or attentional impairments that have also been documented in individuals with thalamic insult (Radonovic & Scaff, 2003; see also Crosson, 1999). The mechanisms for thalamic aphasias remain unclear; the thalami may
26 have multiple influences on normal language functioning by way of extensive connectivity with motor and sensory systems and cortical regions known to be involved in expressive and receptive language processes. 3.2.4 Right hemisphere The incidence of aphasia following right hemisphere insult in both right and left handed adults is very low. Penfield and Roberts (1956) reported only one case of right injury leading to dysphasia in a right handed adult in 10 years of vigorous neurological and language study; the single case was documented as transient (see also, Youngjohn, 1986). Various estimates of aphasia following right hemisphere injury have all indicated a relative infrequency compared to aphasia following left hemisphere insult (e.g., Ettlinger, Jackson, & Zangwill, 1955; Archibald & Wepman, 1968; Faglia, Rottoli, & Vignilo, 1990). Where language disturbances have been documented following right hemisphere insult, the deficits are typically subtle (mild word recognition deficits, dysprosodic speech) and frequently transient (e.g., Marcie, Hécaen, Dubois, & Angelrgues, 1965; Archibald & Wepman, 1968; see also Shapiro & Danly, 1985). Investigations into the contribution of the right hemisphere in the normal language processing of healthy individuals have generally failed to demonstrate a significant role. Thiel, Habedank, Winhuisen, et al. (2005) used repetitive transcranial magnetic stimulation (rTMS) in healthy individuals, and adults with left hemisphere brain tumours, to determine the role of the left and right hemisphere in verb generation to spoken words. Stimulations of the left inferior frontal region disrupted verb generation in both the patient and control groups; these disruptions included non-responding, inaccurate verb generation, and
27 increased latency to generate verbs. Right hemisphere stimulations failed to elicit speech arrest or errors in either group, suggesting a non-essential role for the right hemisphere in verb generation. Interestingly, latencies to respond were increased following both left and right hemisphere stimulations, with the patient group producing verbs at greater latency than the control group with right rTMS. Although the right hemisphere stimulations failed to arrest speech or produce errors in all cases, the degree of patient bilaterality, confirmed by PET, predicted the extent to which latencies were affected (see also Winhuisen, Thiel, Schumacher, et al., 2005). Findings from rTMS suggest that the right hemisphere contributes only subtly to language in healthy individuals; in adults with left hemisphere injury, right hemisphere activations for language on neuroimaging may represent nonessential cerebral involvement (see also, Lazar, Marshall, Pile-Spellman, et al., 2000). Alternatively, error rates and changes in response latencies following right hemisphere rTMS may provide more sensitive measures of right hemisphere contributions to language processes than speech arrest (for a recent review, see Devlin & Watkins, 2007). It is well established that individuals with early injury to the left hemisphere (particularly injury involving perisylvian cortex) frequently experience interhemispheric reorganization, where language shifts from the left to right hemisphere to varying extents (e.g., Rasmussen & Milner, 1977); the opportunity for language to reorganize interhemispherically is limited by age at insult (Brazdil, Zakopcan, Kuba., 2003; Duncan, Moss, Bandy, et al., 1997; Helmstaedter Kurthen, Linke, & Elger, 1997; Muller, Rothermel, Behen, et al., 1998, 1999; Pataraia et al., 2004; Rasmussen & Milner, 1977; Saltzman-Benaiah, Scott, & Smith, 2003; Satz, Strauss, Wada, & Orsini, 1988; Springer, Binder, Hammeke, et al., 1999). The quality of language function in individuals experiencing interhemispheric reorganization
28 also appears to vary as a function of age at insult. Bates, Reilly, Wulfeck, et al. (2001) demonstrated that individuals with congenital lesions of the left versus right hemisphere are functionally equivalent across a wide range of standardized language tests. Congenital insults of either hemisphere permit language to develop in an essentially normal fashion (presumably in the healthy hemisphere), although subtle long-term effects can be detected in both groups when discourse is carefully analyzed (Bates et al., 2001; Reilly, Bates, & Marchman, 1998; Vargha-Khadem, O’Gorman, & Watters, 1985). Stark and McGregor (1997) compared language functioning in two girls, each receiving hemispherectomy around age 3 years for treatment of seizures related to Rasmussen’s encephalitis. The patient with left hemispherectomy had more word finding, syntactic production, and comprehension deficits than the patient with right hemispherectomy, suggesting some degree of early language specialization of the left hemisphere, and that functional recovery may be limited even in early childhood (see also Vanlanker-Sidtis, 2004). Peru, Moro, Tellini, and Tassinari (2006) studied an individual who experienced left haemorrhage and aphasia at age 11. The patient was assessed for language function at age 12, 13, and 14 years. In spite of the relatively late age at insult, the subject demonstrated right hemisphere language representation with the dichotic listening task (a test of receptive language); however, at three years post injury, language recovery was subtotal. Interestingly, recovery of receptive language exceeded that of expressive language function. It is unclear whether findings indicate a relative propensity for interhemispheric plasticity of receptive language, as supported by functional testing and findings from the dichotic listening task, or whether the young patient was unique for pre-injury language representation (i.e., relatively greater right hemisphere involvement for receptive language function). Others have studied
29 language representation in individuals experiencing left hemisphere insults in adulthood, and have found that recovery of function is related to peri-lesional left hemisphere fMRI activations, rather than right hemisphere involvement (e.g., Lazar, Marshall, Pile-Spellman, 2000). Among individuals experiencing interhemispheric reorganization, right hemisphere representation typically involves areas homotopic to canonical left hemisphere language areas. For example, Staudt, Lidzba, Grodd, et al. (2002) used fMRI and a novel word-chain generation paradigm to establish language representation in individuals with extensive preor perinatal left hemisphere insult resulting in right hemiparesis. All subjects demonstrated right hemisphere dominance with fMRI; when localization of activations were compared to those of control subjects, the researchers found that right hemisphere activations in the patient group were indeed homotopic to the left hemisphere activations in controls. This finding may represent interhemispheric connectivity among homotopic regions by way of commissural fibres. Early language experience may involve both the left and right hemisphere, with typical functional shifts toward or establishment within the left hemisphere. In cases of interhemispheric reorganization, language has either shifted contralaterally or established into the right hemisphere, resulting in atypical lateralization.
30
Chapter 4: Classification of Language Disturbances (Aphasiology) In the late 1800s, Broca introduced the term aphemia (“without speech”) to describe the impoverished language expression of his patients; today, the general term aphasia is used in its place, and may refer to any of a wide range of acquired language deficits. The first aphasia classifications were limited to Broca’s aphasia and Wernicke’s aphasia, after the physicians who first characterized brain-language relationships, and conduction aphasia, the disorder predicted by Wernicke (1874) and later documented by Lichtheim (1885). Under Wernicke’s model, language was processed exclusively in the inferior frontal and superior posterior temporal region of the left hemisphere; a subcortical connecting structure, the arcuate fasciculus, directly transferred information originating in the posterior region to the anterior region. Wernicke’s simple model predicted only three discreet aphasias, each resulting from insult to one of the cortical sites or the subcortical tract, and a handful of aphasia combinations resulting from insults spanning multiple regions of the brain’s language pathway. The classification fit a simple expressivereceptive (or motor-sensory) dichotomy, consistent with the 19th century understanding of localization of brain functions (i.e., frontal lobes controlled motor functioning, the posterior brain processed sensory information). This expressive-receptive language dichotomy remains in popular use today, with the terms motor aphasia and sensory aphasia occasionally used in place of expressive aphasia and receptive aphasia, respectively. However, the implied anterior-posterior dichotomy is now recognized as inaccurate: posterior lesions can lead to expressive language deficits, and anterior lesions can indeed lead to difficulty with comprehension or language reception (Albert & Helm-Estabrooks,
31 1988a; see also Poeck, Jerschensteiner, & Hartje, 1972; Willmes & Poeck, 1993). Further, we now differentiate aphasias from motor (apraxia) and sensory (auditory, visual) processing deficits that may lead to language disturbances, suggesting an inappropriateness of the expressive-receptive (and motor-sensory) aphasia terminology. In the 1960s and 1970s, Norman Geschwind, perhaps the most important proponent of Wernicke’s model, introduced a new dichotomy in aphasia classification: fluent versus nonfluent aphasias (Geschwind, 1965, 1971). This new classification referred only to the quality of speech produced by an individual with an acquired language deficit. Classic presentations of Wernicke’s aphasia typify the disorders of the fluent aphasias category: although comprehension is affected, speech remains effortless, prosodic, typically grammatical, and abundant. Broca’s aphasia typifies disorders of the nonfluent aphasia category: comprehension is spared, but speech is effortful, lacks prosody and grammar, and is often poorly articulated. As with the expressive-receptive dichotomy, the categorization along the dimension of fluency was thought to largely correspond to an anterior-posterior division in the brain, although Geschwind was aware that some presentations challenge the association (e.g., posterior temporal lesions may produce anomic aphasias, where individuals cannot name objects efficiently or appropriately, and word-finding deficits; the resulting disorder from a posterior lesion could be classified as nonfluent). Unlike the expressive-receptive terminology, there is no implication that motor or sensory deficits underlie the language disturbances in classifications along the dimension of fluency. Since Geschwind’s proposal of a fluent-nonfluent dichotomy, additional aphasia classification systems have emerged to differentiate the variety of language disorders that have been documented. In general, the tendency was to increase the use of discreet labels
32 for the various aphasias observed. In 1976, Bogen and Bogen wrote, “there are almost as many classifications of aphasia as aphasiologists” (p. 834). Advances in neuroimaging and neurolinguistic research over the past 30 years have likely complicated the taxonomy further. In the 1980s, language researchers often distinguished cortical from subcortical aphasias, introducing unique categories for disorders based on lesion location and presentation (see Albert & Helm-Estabrooks, 1988a, 1988b). Among the cortical aphasias were Broca’s and Wernicke’s aphasia, transcortical motor and transcortical sensory aphasias, and anomic aphasia. Individuals with transcortical motor aphasia demonstrate laborious, agrammatical speech, with intact comprehension; unlike those with Broca’s aphasia, individuals with transcortical motor aphasia demonstrate intact repetition. In transcortical sensory aphasia, speech is fluent and comprehension is impaired; unlike individuals with Wernicke’s aphasia, repetition remains intact. In anomic aphasia, individuals have difficulty with object naming. Subcortical aphasias include those language disturbances that result from injury to the basal ganglia or thalamus, and are typically associated with decreased, laborious speech. In recent years, functional neuroimaging has shown widespread involvement of multiple brain regions for all language tasks, suggesting that brain-language relationships are not mapped on a one-to-one basis, in contrast to the classic Wernicke-Lichtheim-Geschwind model (Hillis, 2007; see also Binder, 1997). Rather, the brain may represent language in a ‘distributed network’ fashion, where functions are supported by a set of collectively engaged neural regions, rather than discreet brain region supporting discreet language processes. These findings suggest that aphasia syndromes are best understood as resultant of dysfunctional language networks, rather than focal deficits from focal injuries; the classic aphasia labels mask the individual
33 variations in network functioning, and therefore fail to perfectly characterize the individual presentations or predict lesion locations. However, with advances in network analyses of functional neuroimaging data, and with increasing research attention on commonalities in language processing deficits and network insults, we are beginning to understand how diffuse regions support seemingly disparate language tasks (Hillis, 2007).
34
Chapter 5: Plasticity 5.1 Paediatric advantage The differential effects of unilateral cerebral insult on language function in adults and children are well established. Left hemisphere inferior frontal and posterior temporal lesions in adulthood typically render individuals permanently dysphasic or aphasic. In contrast, left or right hemisphere lesions occurring prenatally or in infancy usually result in little or no permanent language disturbance (Bates et al., 2001; Reilly, Bates, & Marchman, 1998; cp Vargha-Khadem, Watters, & O’Gorman, 1985). The sparing of language function in children with unilateral cerebral insults is believed to reflect a relative plasticity afforded to the developing paediatric brain. 5.2 Interhemispheric plasticity Language plasticity has long been understood as an interhemispheric effect – in the context of left hemisphere insult, the right hemisphere assumes, to varying degrees, the roles typical of the left hemisphere in language control. Indeed, individuals with left hemispherectomy in the first few years of life retain or develop essentially normal gross language functions (e.g., Devlin, Cross, Harkness, et al., 2003; Liégeois, Cross, Polkey et al., 2008; although subtle deficits can be detected, e.g., Stark & McGregor, 1997; Vanlancker-Sidtis, 2004). Preservation of language in early left hemispherectormy reveals the potential for exclusive right hemisphere language representation, and massive interhemispheric plasticity in the developing brain. Interestingly, several research groups have reported relatively favorable outcomes for receptive language but not expressive language in the context of left
35 hemisphere injury, suggesting a greater propensity for interhemispheric reorganization of the posterior language network (e.g., Boatman, Freeman, Vining, et al., 1999; Peru et al., 2006; cp Stark, Bleile, Brant, Freeman, & Vining, 1995); however, the mechanism for the receptive language advantage has not been well studied. Interhemispheric plasticity and its developmental limits have been most well characterized using the intracarotid amobarbital procedure (IAP; also known as the Wada test, after its founder, Juhn A. Wada) (Wada, 1949; 1997). The procedure involves brief unilateral anesthetization of the anterior and lateral cerebrum (anterior and middle cerebral artery territories), during which time the patient attempts verbal responding to examiner questions, and typically names and describes picture and object stimuli. In this way, both comprehension and language production can be assessed. If only left or right hemisphere anesthetization produces a transient dysphasia or aphasia, the patient is deemed to possess unilateral language dominance. If both left and right, or neither left nor right hemisphere injections affect language functioning, the patient is deemed to possess bilateral language representation. Some groups employ standardized testing and compute laterality indices to quantify hemispheric contributions. IAP in children requires flexibility in test administration, precluding the use of standard test batteries and laterality index computation. Findings from IAP research support a theory of relative plasticity in early development. Individuals with early onset epilepsy of the left hemisphere are significantly more likely to demonstrate atypical language lateralization than those with late neural insult of comparable extent (Rasmussen & Milner, 1977; Satz, Strauss, Wada, & Orsini, 1988; Helmstaedter, Kurthen, Linke, & Elger, 1997, Brazdil, Zakopcan, Kuba, et al., 2003;
36 Saltzman-Benaiah, Scott, & Smith, 2003). In early childhood, language either develops in or shifts over to the right hemisphere, establishing atypical language representation. Largescale studies (e.g., Rasmussen & Milner, 1977) demonstrate limits in language interhemispheric plasticity; collectively, findings suggest a rapidly decreasing potential for language to reorganize from the left to the right hemisphere after the age of 5 or 6 years. Functional neuroimaging techniques are useful for assessing regions of the brain that are statistically related (correlated) to the task being investigated, and may inform us of limits to interhemispheric reorganization. One of the major criticisms in use of functional neuroimaging in presurgical investigation is that the techniques cannot distinguish regions that are involved in a particular cognitive processes or behaviour from regions necessary for execution. For this reason, neuroimaging findings should only be considered in surgical planning after validation of the protocol against invasive techniques, such as IAP (for laterality), stimulation mapping (for localization), and existing lesion literature. Indeed, many centers have now abandoned invasive and moderately invasive language mapping procedures in favor of validated functional neuroimaging techniques (primarily fMRI and MEG for lateralization), which are associated with essentially zero morbidity and relatively low costs (for a recent international survey and discussion on the fate of the IAP, see Baxendale, Thompson, & Duncan, 2008a, 2008b, and associated commentaries; also Baxendale 2009). The neuroimaging data relating age of seizure onset to language laterality generally support IAP findings. Researchers employing PET have reported increased right hemisphere activity/metabolism for language tasks in individuals with seizure onset in early childhood (Duncan et al., 1997; Muller, et al., 1998, 1999). Increased atypical language lateralization
37 following early insult has also been documented using fMRI (e.g., Springer et al., 1999). In their recent large-scale study, Gaillard, Berl, Moore, et al. (2007) assessed language lateralization in 102 children and adults with left hemisphere epileptogenesis, using a battery of expressive and receptive language tasks. Patients ranged in age from 4 to 55 years, with seizure onset between 6 months and 43 years. Those with first seizures before age 6 years were much more likely (43%) to demonstrate atypical language lateralization than those with later seizure onset (19.7%). The rate of atypical language representation decreased with increasing age at onset: atypical language lateralization was observed in 24% of patients with first seizures between 7 and 15 years, and only 14% of subjects with onset after 16 years of age. Recently, researchers have shown increased atypical receptive language lateralization in early seizure onset using a word recognition task in MEG (Pataraia et al., 2004). Some have argued that factors such as duration of seizure disorder (Yuan, Szaflarski, Schmithors, et al., 2006) and patterns of seizure spread (Janszky, Jokeit, Heinemann, et al., 2003) better predict language lateralization than age at seizure onset. Liégeois, Conelly, Cross, et al. (2004) found that age at onset of chronic seizures failed to predict language lateralization; however, 9/10 of Liégeois et al.’s subjects experienced first seizures before age 6 years (of which, 2 had febrile seizures in infancy), suggesting they lacked the variability to truly assess the role of early seizures in prediction of language lateralization. Interestingly, the literature on adult plasticity for language representation shows that the right hemisphere becomes ‘active’ following both early and late left hemisphere insult. Activations typically occur in regions homologous to canonical Broca’s and Wernicke’s areas (Staudt et al., 2002; Vanlanker-Sidtis, 2004). Some researchers have documented
38 recovery from aphasia with de novo right hemisphere activations (e.g., Raboyeau, De Boiessezon, Marie, et al., 2008), while others have observed increased right hemisphere activations following left hemisphere injury, in the absence of intact or recovered language (Theil et al., 2001; Blank, Bird, Turkheimer, & Wise, 2003). Vikingstad et al. (2000) used fMRI to study language lateralization in adults with left hemisphere congenital arteriovenous malformations (AVM), adults who had recently experienced left hemisphere stroke, and age-matched controls. Each of the individuals with AVM demonstrated right or bilateral language representation, whereas control subjects were left hemisphere dominant. Similarly, individuals in the stroke group demonstrated shifts toward right hemisphere activation for language tasks; however, shifts occurred to a lesser degree than those with AVM, and individuals with stroke experienced persistent aphasias of variable severity, in spite of right hemisphere dominance on fMRI. Whereas all of the subjects with AVM had intact language, eventual recovery from aphasia in the stroke group was associated with shifts of language back to the left hemisphere. This finding may seem confounded by the possibility that individuals with less severe insults may be more likely to experience both recovery and return of ‘typical’ patterns of brain activity. However, there is a growing body of research suggesting that left hemisphere perilesional activation better predicts recovery of language than contralateral activations (Cao, Vikingstad, George, Johnson, & Welch, 1999; Rosen Petersen, Linenweber, et al., 2000; Fernandez, Cardebat, Demonet, et al., 2004; Meinzer, Flaisch, Breitenstein, et al., 2008). Recently, Saur, Lange, Baumgaertner, et al. (2006) demonstrated a trajectory for recovery from aphasia following left hemisphere insult in adulthood: initial increases in right hemisphere activations are
39 followed by left hemisphere perilesional activations in those experiencing the greatest recovery of language. 5.3 Intrahemispheric plasticity Recently, we studied intrahemispheric plasticity of language using extraoperative stimulation mapping (ESM) in children with early onset epilepsy of the left hemisphere. Using a novel coregistration technique, whereby ESM data are plotted in standard stereotactic space, we demonstrated that expressive language sites may reorganize within the frontal lobe of the left hemisphere (intrahemispheric reorganization). We observed expressive language sites (i.e., those necessary for generating free speech and/or overt naming) at regions anterior and superior to canonical Broca’s area; intrahemispheric displacement of language occurred to an extent not previously described. Findings suggest that in addition to the well-documented contralateral reorganization observed following early left hemisphere injury, language may also reorganize within the classically ‘dominant’ hemisphere (Kadis, Iida, Kerr, et al., 2007). These findings are consistent with the adult neuroimaging literature, showing left hemisphere perilesional activation in recovery from acquired aphasias in adulthood. 5.4 Mechanism of plasticity Although it is generally agreed that the paediatric brain is more plastic for language than adult brain, no consensus exists on the mechanism(s) of the plasticity. Two competing theories have been proposed to explain atypical lateralization (e.g., Rasmussen & Milner, 1977) and atypical localization (e.g., Kadis et al., 2007) observed in the context of early insult:
40 1) ‘immature’ language networks look much like adult networks; in cases where language representation is adult-atypical, function has shifted and brain regions not typically involved have been recruited to support language; 2) ‘immature’ language networks are extensive and bilateral; language establishes into non-canonical (adult-atypical) regions following early insult because extensive networks lead focal networks in the normal developmental trajectory; Indeed, each theory has received some support through neuroimaging research. Gaillard, Sachs, Whitnah, et al. (2003) employed fMRI to assess children (aged 7 to 14 years) and adults engaging in a semantic fluency task, and failed to observe group differences in localization or extent of activation for the language task (cp, Gaillard, Hertz-Pannier, Mott, et al., 2000). Wood, Harvey, Wellard, et al. (2004) compared asymmetry and extent of activations in children and adults completing a verb generation and orthographic lexical retrieval (fluency) task in fMRI. Although children (aged 6 to 15 years) demonstrated a higher rate of atypical lateralization (15% of children, compared to only 6% of adults), the difference was not statistically significant. Extents of activations in each hemisphere were similar for children and adults. These findings suggest that atypical representation in the context of early injury reflects shifts from canonical to contralateral or perilesional regions. In contrast, Holland, Plante, Weber Byars et al. (2001) used fMRI to assess healthy children aged 7-18 years participating in a verb generation paradigm, and found that left hemisphere lateralization increased with age. In their recent study, Brown, Lugar, Coalson, et al. (2005) used fMRI to assess cortical activity in healthy participants aged 7-32 years during
41 three performance-matched overt word generation tasks, and observed relatively widespread and bilateral representation in children whereas adults demonstrated language representation focused in frontal and parietal regions of the left hemisphere. Using MEG, Ressel, Wilke, Lidzba, Lutzenberger, and Krageloh-Mann (2008) studied hemispheric differences in 7-16 year old children completing an overt verb generation and vowel identification task, and found that left hemisphere lateralization increased with age. These findings support a theory of extensive and bilateral language representation in childhood. Unfortunately, given the paucity of research on developmental changes in language representation, and the variability of tasks and analyses employed, it is difficult to describe immature networks and normal developmental changes. The developmental trajectory for language representation has not been comprehensively assessed/described using a battery of tasks that are performance-matched for all participants.
42
Chapter 6: Pathology Type vs Language Lateralization 6.1 Study rationale Researchers have examined the predictive utility of a number of demographic, psychological, and neurological variables on language lateralization in patient populations. Among those variables, side of injury and age at insult are typically regarded as most predictive (see Chapter 5). There is little agreement on the role of pathology type on prediction of language lateralization. Duchowny, Jayakar, Harvey, et al. (1996) found that acquired lesions encroaching on canonical language areas (of the left hemisphere) resulted in atypical (bilateral or right hemisphere) lateralization, yet developmental lesions were not associated with interhemispheric reorganization. In contrast, others have observed atypical lateralization in the context of developmental pathologies (e.g., Vikingstad, Cao, Thomas, et al., 2000; Briellmann, Labate, Harvey, et al., 2006). The predictive utility of knowing pathology type has not been systematically analyzed. Our experience at the Hospital for Sick Children permitted a large-scale retrospective analysis of the relation between pathology type and hemispheric lateralization of language, as determined by IAP. Our findings were published in Epilepsia (Kadis, Kerr, Rutka, et al., 2009), and are included, as published, in this dissertation. Epilepsia’s published editorial policies indicate that authors retain the right to use published material in other personal compilations or publications: “After an article has been accepted, authors may share or print electronic copies of the article (accepted and revised to
43 address peer review) with colleagues, and may use the material in personal compilations, other publications of his/her own work, and for educational/research purposes” (International League Against Epilepsy, 2009, Public Access of Accepted/Published Articles section). The definitive version of the article, “Pathology type does not predict language lateralization in children with medically intractable epilepsy”, is available at http://www3.interscience.wiley.com/journal/117957420/home.
44
Running head: PATHOLOGY AND LANGUAGE DOMINANCE
Pathology Type Does Not Predict Language Lateralization in Children with Medically Intractable Epilepsy
Darren S. Kadis1,2, Elizabeth N. Kerr2, James T. Rutka3,4, O. Carter Snead III5,6, Shelly K. Weiss5,6, Mary Lou Smith1,2,5
1 2 3
Department of Psychology, University of Toronto
Department of Psychology, Hospital for Sick Children
Department of Surgery, Division of Neurosurgery, University of Toronto 4
5
Division of Neurosurgery, Hospital for Sick Children
Department of Pediatrics, Division of Neurology, University of Toronto 6
Division of Neurology, Hospital for Sick Children
45 6.2 Abstract Purpose: We examined potential differences in the effects of pathology type on language lateralization in paediatric epilepsy. Methods: We examined findings from intracarotid sodium amobarbital procedure (IAP/Wada) in a large consecutive sample of children with refractory epilepsy. Subjects were assigned to one of three pathology groups: developmental (n = 28), acquired (n = 26), and tumour (n = 20); groups were compared for language lateralization. Results: Rates of atypical language lateralization did not differ across groups. Greater than half of the subjects with left hemisphere insults and seizure onset before 6 years of age had atypical language lateralization, independent of pathology type. Discussion: Atypical language lateralization may occur in the context of developmental, acquired, and/or tumour pathology.
46
6.3 Introduction It is well established that language representation is more plastic in children than adults. Children demonstrate better recovery of language function than adults experiencing insults of comparable extent (Vargha-Khadem et al., 1985; Reilly et al., 1998; Bates et al., 2001). Individuals with early left hemisphere lesions are more likely to express atypical (right or bilateral) language representation than those with insults occurring later in life, probably explaining the paediatric advantage for language recovery following cerebral insult. The role of age at insult on language plasticity has been well documented in several modalities (intracarotid sodium amobarbital procedure (IAP, also known as the Wada test): Brazdil et al., 2003; Helmstaedter et al., 1997, Rasmussen & Milner, 1977; Saltzman-Benaiah et al., 2003; Satz et al., 1988; positron emission tomography (PET): Duncan et al., 1997; Muller et al., 1998, 1999; functional magnetic resonance imaging (fMRI): Springer et al., 1999; and magnetoencephalography (MEG): Pataraia et al., 2004). However, age at insult incompletely predicts language lateralization in adults with epilepsy. Some have suggested that duration of seizure disorder (Yuan et al., 2006) and patterns of seizure spread (Janszky et al., 2003) better predict language lateralization; these and other factors affecting language function and reorganization are not well understood (see also, Liégeois et al., 2004). Duchowny et al. (1996) were the first to report on differential effects of developmental versus acquired pathology on language laterality in children with epilepsy. In their study, Duchowny et al. inferred language dominance from unilateral extraoperative stimulation mapping (ESM) and found that only individuals with acquired insults experienced before
47 age 5 years demonstrated interhemispheric language reorganization. Individuals with insults acquired later in life, and those with developmental pathologies, had typically lateralized (left hemisphere) language representation. In their sample, Duchowny et al. observed developmental lesions that abutted and encroached upon language sites, suggesting that the developmental lesions were insufficient to displace language. Using ESM, Duchowny et al. were only able to assess single hemispheres for language, precluding detection of bilateral language representation. Further, because the electrode grids employed were limited in extent (typically covering only portions of the frontal or temporal lobes), the authors were unable to detect mixed dominance (a form of atypical lateralization, where expressive and receptive language are contralaterally represented) or intrahemispheric reorganization (see Kadis et al., 2007). Therefore, the failure to interrupt language with ESM need not indicate contralateral dominance, but may reflect incomplete assessment of the surgical hemisphere. Ideally, any study of the role of pathology on language dominance would employ methods that assess both hemispheres, comprehensively. IAP is still the “gold standard” for determination of language dominance, although non-invasive neuroimaging techniques are gaining popularity as protocols become validated. Saltzman-Benaiah et al. (2003) explored predictors of language dominance in children with intractable epilepsy using IAP, and found that 7 out of 15 children with identifiable left-hemisphere developmental lesions demonstrated atypical language lateralization. Although the sample was too small to formally compare individuals with acquired versus developmental pathologies, their findings demonstrate that atypical language lateralization can occur in the context of developmental pathology.
48 Vikingstad et al. (2000) used fMRI to study language lateralization in a small sample (n = 5) of adults with left hemisphere arteriovenous malformations (AVM) known to arise in early development. Lateralization findings were contrasted against those from a group of individuals who had experienced left hemisphere stroke in adulthood, and healthy control subjects. Each of the individuals with AVM demonstrated right or bilateral language representation, whereas control subjects were left hemisphere dominant (see also, Lazar et al., 2000). Those in the stroke group also demonstrated shifts toward right hemisphere activation for language tasks, although to a lesser degree than those with AVM, and with persistent aphasias of variable severity. Whereas all of the subjects with AVM had intact language, recovery from aphasia in the stroke participants was associated with shifts of language back to the left hemisphere. Recently, Briellmann et al. (2006) used fMRI to compare language lateralization in healthy controls with that of adolescents and adults with developmental (n = 16) versus acquired lesions (hippocampal sclerosis; n = 25) restricted to the temporal lobe. Atypical language was observed at an increased rate among the patient groups, but laterality indices did not differ between groups with developmental versus acquired lesions. The findings suggested that temporal lobe insults are associated with increased atypical language lateralization, but that the nature of the temporal lobe pathology does not predict hemispheric language dominance. Other fMRI studies have found that atypical language lateralization was not associated with the nature of the lesion; however, these studies varied in their classification of lesions as developmental versus acquired (Anderson et al., 2006; Gaillard et al., 2007). In the present study, we assessed language lateralization from a large consecutive sample of children who underwent IAP for language lateralization as part of a presurgical workup for
49 treatment of refractory epilepsy. Although others have documented atypical dominance in small groups with relatively homogenous insults, this is the first study to investigate the role of pathology in a large paediatric sample. The extensive retrospective series provides the necessary power to determine whether the nature of pathology is useful for prediction of language lateralization. 6.4 Methods 6.4.1 Participants We retrospectively analyzed findings from 105 consecutive cases of children with localization-related epilepsy who underwent IAP for language lateralization at the Hospital for Sick Children (Toronto, ON, Canada) between 1982 and 2007. All subjects included in our analyses had unilateral seizure foci and medically intractable epilepsy; subjects were investigated for candidacy for surgical treatment (described previously, Minassian et al., 1999; Snead, 2001). The review of patient files and subsequent analyses were approved by the hospital’s research ethics board. For inclusion in our analyses, subjects were required to have pathologies verified through histological study of resected tissue, and at least one conclusive language investigation. One subject received multiple subpial transections without resection of tissue and histopathological analysis; a classification of cortical dysplasia was determined through review of the patient’s MRI, and the subject was included in our study. Of the full pool, 9 subjects were excluded due to insufficient abnormality for determination of pathology, 12 were excluded for mixed pathology (e.g., cortical dysplasia with mesial temporal sclerosis), and 2 were excluded due to bilateral epileptogenesis. An additional 8 subjects were
50 excluded because language investigations were inconclusive due to poor cooperation or obtundation after injection. In the remaining 74 cases, abnormalities were classified as “developmental” in 28 subjects, and “acquired” in 26 subjects. A third group with cerebral tumours were included (“tumour” group, n = 20); the ontogeny of these insults remains unknown. The developmental pathologies were cortical dysplasia (n = 18), Sturge-Weber disease (n = 4), congenital porencephalic cyst (n = 2), cavernous haemangioma (n = 2), tuberous sclerosis (n = 1), and AVM (n = 1). Acquired pathologies were mesial temporal sclerosis (n = 13), gliosis (n = 6), encephalitis (n = 5, of which 2 were Rasmussen’s), and infarct (n = 2). Subjects in the tumour group had astrocytomas (n = 7), oligodendroglioma (n = 1), mixed gliomas (n = 2), unspecified gliomas (n = 3), gangliogliomas (n = 3), and dysembryoplastic neuroepithelial tumours (DNETs; n = 4). Patient charts were consulted to determine demographic and seizure-related data, summarized in Table 6.1. Handedness was established through parent and patient report of hand preference, and from observation during assessments. Antecedents to insult for individuals in the acquired pathology group are presented in Table 6.2. 6.4.2 Assessment of language laterality All subjects were studied using IAP for language lateralization. To promote simplicity and power in analyses, we dichotomized language lateralization as “typical” (left hemisphere dominance) or “atypical” (bilateral representation or right hemisphere dominance). The IAP protocol used at the Hospital for Sick Children has been documented previously (Fernandes & Smith, 2000; Saltzman-Benaiah et al., 2003), and is described only briefly, below.
51 Participants underwent baseline language assessment prior to IAP. At baseline assessment, subjects were asked to name pictures and objects, spell, read, recite days of the week or the alphabet, and count. Items and procedures successfully completed at baseline were included in the IAP assessment battery; in this way, the IAP assessments were tailored to each subject’s ability. To assess the specialization of each hemisphere, sodium amobarbital (Amytal) was administered unilaterally (at 1.5 mg/kg by body weight) via subjects’ internal carotid arteries. Ongoing EEG was consulted to verify drug effects; paralysis of limbs contralateral to the barbiturate injections confirmed cerebral perfusion. Language testing commenced immediately following observation of slow waves over the perfused hemisphere. Errors relative to baseline and/or speech arrest during barbiturate perfusion were indicative of language representation in the tested hemisphere. After clearance of the drug (verified by absence of slow waves on EEG), items associated with errors were retested; performance at retesting confirmed that errors reflected language interference associated with the barbiturate administration. Following retesting, the catheter was repositioned into the contralateral intracarotid artery to test language function of the other hemisphere. All subjects received both left and right hemisphere injections. If only left-sided or only right-sided injections produced errors or speech arrest, the individual was deemed to possess lateralized speech. If language was conclusively interrupted following injections to both hemispheres, or if neither left- nor right-sided injections produced errors or speech arrest, the participant was deemed to possess bilateral language.
52 6.4.3 Analyses We compared groups across the following demographic, cognitive, and seizure-related variables: sex, handedness, Verbal IQ (VIQ) and Performance IQ (PIQ), seizure side, seizure site, age at seizure onset, and age at assessment. Category-frequency data were analyzed using chi-square (χ2) goodness-of-fit tests; IQ and age data were analyzed using one-way nalyses of variance (ANOVAs). We adopted an alpha level of .05 for these analyses. We assessed the effects of age at seizure onset (< 6 years vs 6 years or older) on language laterality for the whole group and for the subset with left hemisphere epilepsies, regardless of pathology group membership, using χ2 goodness-of-fit tests. We compared language lateralization across groups using χ2 goodness-of-fit analyses. Subsets were contrasted to control for effect of seizure side and age at seizure onset. We adopted an alpha level of .05 for each analysis; exact probabilities were computed where expected cell counts were less than 5. We did not adjust for multiple comparisons, as the conflicting findings in the literature suggested that any effect of pathology on language laterality would be small. 6.5 Results Groups were equivalent in terms of sex, seizure side, seizure site, age at assessment, and VIQ and PIQ, p > 0.05. We observed significant group differences in handedness, with the vast majority of left-handers belonging to the developmental pathology group, χ 2 = 10.98, df = 4, exact p < 0.05. Groups also significantly differed in age at seizure onset, with
53 individuals in the developmental and acquired pathology groups experiencing first seizures at a younger age than individuals in the tumour group, F(2, 70) = 3.22, mean squared error (MSE) = 18.73, p < 0.05. Within the whole sample, 35 subjects had seizure onset before 6 years of age, and 38 subjects had seizure onset at age 6 years or later; age at seizure onset was not reliably documented for one subject in the tumour group. Atypical language lateralization was significantly more frequent in subjects with early seizure onset (51.4%, compared to 15.8% in the later seizure onset group), χ 2 = 10.49, df = 1, p < 0.05. Among subjects with left hemisphere epilepsies, 27 had seizure onset before 6 years of age, and 23 had seizure onset at age 6 years or later. Atypical language lateralization was more likely in those with left hemisphere seizure onset before 6 years of age (62.9%) than those with later onset left hemisphere epilepsies (26.0%), χ 2 = 6.8, df = 1, p < 0.05. Individuals with atypical language lateralization had lower VIQ [mean (M) = 81.2, standard deviation (SD) = 19.9] and PIQ (M = 81.0, SD = 17.9) than those with typical language lateralization (VIQ: M = 89.2, SD = 17.4; PIQ: M = 92.5, SD = 17.1). For PIQ scores, the difference was statistically significant, t(70) = 2.5, p < .05. Cognitive differences are predicted by language lateralization, not laterality of seizure foci, as individuals with left hemisphere seizures and typical language lateralization (n = 25) had higher VIQ (M = 89.4) and PIQ (M = 92.2) than those with left hemisphere seizures and atypical language lateralization (n = 18; VIQ: M = 80.7; PIQ: M = 80.9). Atypical language lateralization was documented in 11 subjects with developmental pathologies (39.2%), 9 subjects with acquired pathologies (34.6%), and 4 subjects with
54 tumours (20.0%). The difference was not statistically significant, χ 2 = 2.07, df = 2, p > .05. Only 1 of 25 subjects with right hemisphere seizures had atypical (bilateral) language representation at IAP testing; this subject was right-handed, and had right frontal cortical dysplasia. Frequency of typical and atypical language lateralization for subjects in each pathology group is depicted in Figure 6.1. Fifty subjects had left hemisphere seizures (19 with developmental pathologies, 18 with acquired pathologies, and 13 with tumours). Within this subset, atypical language lateralization was documented in 10 subjects with developmental lesions (52.6%), 9 subjects with acquired pathologies (50.0%), and 4 subjects with tumours (30.8%). The difference was not statistically significant, χ 2 = 1.67, df = 2, p > .05. The frequency of typical and atypical language lateralization as related to age at seizure onset and side of seizure focus is shown in Figure 6.2. Twenty-seven subjects had left hemisphere epilepsy with seizure onset before 6 years of age (14 with developmental pathologies, 11 with acquired pathologies, 2 with tumours). Within these subsets, atypical language lateralization was documented in 8 subjects with developmental lesions (57.1%), 7 subjects with acquired pathologies (63.6%), and both subjects with tumours (100%). The difference was not statistically significant, χ 2 = 1.38, df = 2, exact p > .05. 6.6 Discussion We observed increased atypical language lateralization in our clinical sample (in 46% of subjects with left hemisphere seizures) compared to healthy paediatric (Balsamo et al., 2002) and adult populations (Knecht et al., 2000; see also Szaflarski et al., 2006). This
55 finding is consistent with previous reports of increased plasticity for language in children with intractable epilepsy (e.g., Yuan et al., 2006). The low rate (single case) of atypical language representation in subjects with right hemisphere seizures is consistent with normal population estimates of language lateralization (e.g., Knecht et al., 2000). As has been previously shown (Billingsley & Smith, 2000; Loring et al., 1999; Gleissner et al., 2003), IQ scores were lower in individuals with atypical language lateralization than those with typical language dominance. Findings suggest that interhemispheric plasticity comes at a cost to cognitive functioning. Generalized impairment may reflect reduced neural resources, or right hemisphere crowding. We failed to observe an effect of pathology on language lateralization. Although atypical language lateralization was observed among subjects with developmental lesions, acquired lesions, and tumours, the proportion of cases within each group was not significantly different. When we restricted our analyses to include only individuals with left hemisphere pathology associated with seizure onset before 6 years of age, we again observed a high degree of atypical language lateralization (63.0% across pathology groups), but failed to detect group effects. In our case series, atypical language lateralization was more likely than typical language lateralization among individuals with left hemisphere insults related to early seizure onset; however, pathology did not predict hemispheric language dominance. Left handedness was more common in children with developmental pathologies, suggesting that hand-motor preference may shift as a function of age at insult. Woods et al. (1988) commented that it is difficult to determine whether there is a causative association between
56 handedness and language representation, or whether left handedness is secondary to lesions likely to produce atypical lateralization. This is the first large scale study to demonstrate that the nature of pathology does not determine language lateralization in children and adolescents with intractable epilepsy. Others have explored the effects of pathology on language laterality in relatively small and homogenous samples (e.g., Vikingstad et al, 2000; Briellmann et al., 2006); here, we were able to assess the relationship across a broad range of insults that lead to seizure disorders in childhood. Findings are particularly relevant for centres offering surgery for paediatric patients, as children are likely to present with seizure disorders associated with varied (often extratemporal) pathologies compared to adults (Holmes, 1993; see also, Snead, 2001). One of the challenges we encountered with this study was with the assignment of pathologies into developmental and acquired categories. We can be relatively certain that lesions categorized as “developmental” were present at birth, as the time courses for emergence of cortical dysplasias and the vascular abnormalities, the predominant pathologies in the developmental group, are well understood. However, we can be less certain that individuals with lesions categorized as “acquired” were entirely free of significant pathologies at birth. Although it is generally accepted that hippocampal and/or mesial temporal scleroses are acquired in nature, the etiology for these pathologies are not yet fully understood (see Briellman et al., 2006). Similarly, it is not known when a tumour presenting in childhood began its development.
57 For the current analyses, we classified language lateralization as “typical” versus “atypical”; this dichotomy is commonly used in studies examining factors associated with interhemispheric reorganization (Anderson et al., 2006; Gaillard et al., 2007; SaltzmanBenaiah et al., 2003). The dichotomy promotes simplicity and maximizes power in contingency table analyses. However, information pertaining to the degree of lateralization of language is sacrificed. It is possible that type of pathology affects degree of lateralization, such that one group experiences interhemispheric reorganization to a greater extent than another (i.e., more right hemisphere involvement in one pathology group compared to another). We were not able to describe the extent of lesions in our sample due to the differences in MRI scanning over the course of the ascertainment of subjects (1982 to present). The location of the lesion may be important in influencing the type and degree of language reorganization. However, lesion encroachment on classical language areas in children has not always confirmed this assumption. DeVos et al. (1995) used IAP to study language lateralization in 12 children with left perisylvian tumours and seizure onset before 6 years, and observed typical language lateralization in 83% of subjects. A subset underwent electrocortical stimulation mapping, which confirmed language representation in close proximity to the tumours. Similarly, Liégois et al. (2004) assessed language representation in children using fMRI, and found that lesions that were proximal to or encroaching on Broca’s area were not associated with interhemispheric reorganization. Interestingly, children with lesions distal to Broca’s area were more likely to demonstrate atypical language lateralization, suggesting that location of lesions does not predict reorganization in an obvious manner.
58 The use of IAP in assessing language representation is limiting, as the protocol does not permit characterization of reorganization beyond simple hemispheric participation. We have previously documented intrahemispheric reorganization of language from canonical to atypical regions within the dominant hemisphere using extensive electrode grid arrays and stimulation mapping (Kadis et al., 2007); because IAP involves broad perfusions, the technique will correctly identify hemispheric contributions to language, but cannot characterize subtle intrahemispheric shifts in representation. Others have demonstrated that functional MRI allows for greater evaluation of adaptive variants of language representation (Berl et al., 2006). Future studies employing whole head functional neuroimaging may demonstrate differential effects of seizures on anterior (expressive) versus posterior (receptive) language regions, while remaining sensitive to intrahemispheric reorganization.
59
6.7 Acknowledgements This research was supported, in part, by a Studentship to DSK through the Ontario Student Opportunity Trust Fund – Hospital for Sick Children Foundation Student Scholarship Program, and a Doctoral Research Award to DSK through the Canadian Institutes of Health Research (CIHR) in partnership with Epilepsy Canada. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. We have no conflicts of interest to disclose.
60 Table 6.1 Group demographic and seizure-related characteristics.
a
Developmental (n = 28)
Acquired (n = 26)
Tumour (n = 20)
n
n
n
Sex male female
9 19
13 13
13 7
Handedness * left right mixed
12 16 0
4 22 0
2 17 1
Side of Seizure Focus left right
19 9
18 8
13 7
Site of Seizure Focus frontal temporal parietal occipital multilobar
3 11 1 1 12
3 17 0 1 5
4 14 1 0 1
Age at assessment, years Mean (SD) Range
14.8 (3.0) 7.3-18.2
13.8 (3.5) 6.3-17.4
13.0 (2.5) 7.6-18.0
Age at seizure onset, years a ** Mean (SD) Range
5.3 (4.6) 0.1-16.0
5.4 (4.2) 0.0-14.5
8.4 (4.0) 1.3-15.3
Intellectual functioning b VIQ PIQ
84.8 (19.4) 87.3 (18.4)
83.8 (15.6) 86.0 (15.6)
95.4 (18.9) 96.8 (19.5)
Age at seizure onset was not reliably documented for one subject with a tumour, statistics based on n = 19.
b
A subset of patients underwent neuropsychological assessment. Standardized verbal intelligence (VIQ), and performance intelligence (PIQ) scores are presented for 25 subjects with developmental pathologies, 25 with acquired pathologies, and 16 subjects from the tumour group. * Group differences observed at χ2 testing, exact p < 0.05 ** Group differences observed with ANOVA, p < 0.05
61 Table 6.2 Possible antecedents for individuals in the acquired pathology group ID
Pathology
Antecedents / conditions
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20
gliosis encephalitis MTS MTS gliosis infarct encephalitis MTS infarct MTS MTS MTS MTS encephalitis MTS gliosis gliosis MTS gliosis encephalitis
21
MTS
22 23 24 25 26
gliosis MTS MTS MTS encephalitis
none noted encephalitis at time of seizure onset none noted febrile illness (influenza) at time of seizure onset none noted tetralogy of Fallot, spina bifida encephalitis at time of seizure onset febrile seizures at 5 years none noted none noted febrile seizures at 0.5 years encephalitis at time of seizure onset none noted encephalitis at 5 years none noted none noted none noted none noted none noted encephalitis at time of seizure onset presumed viral encephalitis at time of seizure onset none noted none noted none noted none noted encephalitis at time of seizure onset
Seizure side
Onset (years)
Language laterality
left left right left left left left left left left left right right left left right right left right left
12.0 3.5 14.5 0.9 8.0 1.0 6.0 6.0 0.0 6.0 0.5 4.0 5.75 12.0 4.0 6.0 7.0 3.5 9.0 1.0
left bilateral left bilateral right right left left left left left left left left right left left right left bilateral
right
8.2
left
left left left left right
14.4 1.5 0.7 1.0 1.0
right left right left left
62
Figure 6.1
Stacked bars indicate group frequencies of typical and atypical language lateralization in subjects with left hemisphere seizures (n = 54). Yellow bars represent frequency of typical language lateralization; orange bars represent frequency of atypical language lateralization.
63 Figure 6.2
Stacked bars indicate group frequencies of typical and atypical language lateralization in subjects with left hemisphere seizure onset before 6 years of age (n = 28). Yellow bars represent frequency of typical language lateralization; orange bars represent frequency of atypical language lateralization.
64
Chapter 7: Magnetoencephalography 7.1 Neuroimaging with MEG In recent years, magnetoencephalography (MEG) has emerged as a valuable tool for functional neuroimaging of the brain. Clinicians and researchers have successfully used MEG to localize and characterize pathological tissue in neurological disease, and to assess patterns of neuronal firing associated with various behavioural and cognitive processes in both patient and control populations. Unlike more traditional functional neuroimaging techniques (i.e., PET and fMRI), MEG boasts excellent (sub-millisecond) temporal resolution, permitting relatively direct assessment of neuronal firing through measurement of rapidly-changing neuromagnetic fields. Recent advances in computing and increased availability of MEG systems have promoted sophisticated collection and data analyses, securing MEG’s place in the future of both clinical and non-clinical functional neuroimaging. 7.2 MEG recording MEG involves detection, recording, and analysis of neuromagnetic signals outside the skull. Detection is made possible through the use of superconducting quantum interference devices (SQUIDs), extremely sensitive magnetometers that transmit electrical current in response to varying magnetic fields. In order to superconduct (a quantum mechanical phenomenon, where a wire possesses no electrical resistance and no internal magnetic field), SQUIDs must be maintained at extremely low temperatures, typically around 4°K in liquid helium baths. Because neuromagnetic signals are extremely weak – on the order of
65 1/108 the strength of the earth’s static field – MEG recording must be carried out in rooms that have been carefully shielded from external magnetic sources. As with EEG, the MEG signal is primarily attributed to graded post-synaptic potentials (PSPs) of cortical neurons and their dendrites (Sato, Balish, & Muratore, 1991; Lopes da Silva, 1991). Although action potentials (APs) are associated with higher voltage than PSPs, axonal conduction is unlikely to drive the signal in either EEG or MEG. APs are fast compared to PSPs, and involve relatively small surface regions at given moment; the currents associated with an action potential are said to be “quadrupolar” (two dipolar currents per propagating fiber) with proximal opposing currents localized near the point of depolarization. Due to the rapidly changing focal source, even neighbouring fibers are unlikely to be temporally coordinated, and therefore are likely to cancel when summated in typical EEG and MEG recordings. In contrast, PSPs are sustained over several milliseconds, involve greater spatial areas (dendrite and soma), and are associated with currents that are dipolar in nature (single current vector). Cortical PSPs are likely to coordinate temporally in neighboring neurons of similar-orientation, which promotes positive summation in typical EEG and MEG recordings. Unlike standard EEGs, which involve measurement of cerebral electrical potentials at the scalp, MEG involves measurement of magnetic fields (associated with the cerebral electrical potentials) outside the head. In typical EEG recordings, electrocortical signals must pass through multiple tissues, including the dura, skull, and scalp, each with their own conductivity (resistance, capacitance, and inductance), before detection with electrodes. Because electrocortical signals must pass through intermediate materials known to produce attenuation and distortion, scalp EEG recordings are subject to non-trivial filtering and
66 spatial smearing. In contrast, neuromagnetic signals pass freely through the dura, skull, and scalp – the MEG signal is relatively unfiltered and free of spatial smearing from intermediate materials. MEG is unique among functional neuroimaging modalities in its sensitivity to primary neuronal currents. PET may be used to identify metabolic activity (e.g., FDG tracer) or regional cerebral blood flow (e.g., 15O or H215O tracers) across the brain, assays of neural integrity and process-related activation. Similarly fMRI provides a measure of processrelated blood oxygenation across the brain, which reflects local neuronal activity. Both PET and fMRI measure changes at the order of seconds and greater, too slow to detect neuronal discharges (order of several milliseconds for PSPs). While EEG and MEG recordings are limited in temporal resolution only by their sampling rate, the two modalities do differ in terms of signal source. EEG involves measurement of electrical potentials at the scalp, which may be driven by neuronal sources of any orientation located anywhere within the brain, with preference for radially oriented (perpendicular to the surface of the skull and scalp) cortical sources (reviewed in Sato et al., 1991; see also, Lopes da Silva, 1991). The EEG recorded at the scalp is necessarily a measure of electrical summation from diffuse fields, and for this reason, is said to primarily measure extracellular volume potentials, not primary currents. In contrast, MEG is relatively insensitive to radiallyoriented and deep sources, which typically correspond to magnetic fields contained within the skull. However, tangential currents, such as those produced by post synaptic pyramidal cells within in cortical sulci (approximately two thirds of cortical neurons are contained within the sulci), are associated with magnetic fields that pass perpendicularly through the skull. These extracranial magnetic field changes are well-suited for detection with today’s
67 MEG systems which involve a dense array of magnetometers arranged proximally to the skull’s surface. MEG preferentially measures cortical primary currents (intracellular), rather than the volume potentials. The nature and magnitude of neuromagnetic signals of interest must be taken into consideration when establishing an MEG acquisition paradigm. In pathological states, such as epilepsy, paroxysmal activity may have sufficiently strong current densities that a single spike may be detected in continuous recording. Similarly, rhythmic background activity may be sufficiently synchronous that raw, unaveraged data can be correctly analyzed for source (Hämäläinen, 1992, reviewed in Hämäläinen & Hari, 2002). In contrast, most studies of sensory, motor, or higher cognitive processes require averaging over multiple trials for identification of neuromagnetic responses. When responses are averaged over many trials, consistent spatial-temporal changes will summate; however, uncoordinated (presumably unrelated to the process under investigation) neuromagnetic changes will tend to cancel over multiple trials. Thus, averaging can be used to improve the signal-to-noise ratio, provided that responses over trials are spatially and temporally consistent. 7.3 MEG analyses Data in MEG can be analyzed in several ways. In evoked response paradigms, traditional analyses involve dipole modeling for determination of probable sources. The dipole source (also called the equivalent current dipole or ECD) is a vector quantity, indicating strength and direction of sources for a magnetic field observed at a particular moment. Because evoked responses are relatively time-locked (responses occur at a consistent period following some known event, such as stimulus presentation or button pressing) and phase-
68 locked (consistent positivity/negativity at relevant sensors), averages over trials produce smooth waveforms that may be well described in terms of a dipolar source or sources. Dipoles are fit to a spherical model of the brain/head, or to a 3D structural volume of the subject’s head, using an iterative least-squares statistical approach. Unfortunately, ECD modeling results in non-unique solutions to observed magnetic fields; indeed, there are an infinite number of possible sources that can explain a particular MEG recording. (This situation, where the recorded magnetic field represents an outcome and the neural source is unknown, is mathematically/logically known as the inverse problem.) To improve the dipole model and attempt to solve the inverse problem, a priori constraints are placed on ECD localization. Some constraints are straightforward and easily justified – e.g., the source must originate from within the headspace. Others, such as number and general location of the dipole(s), may be based on theoretical assumptions, and may be poorly justified and potentially misleading. A modeled dipole can be statistically analyzed for fit to observed recordings; clinicians and researchers may require that modeled dipoles pass a predetermined statistical threshold for inclusion in their studies. A relatively new set of analytic techniques known as beamforming – processes that spatially segment or filter the brain into discrete units (voxels) for measurement of local changes in power over time – have improved our ability to assess neural responses that cannot be easily analyzed using traditional ECD modeling (see Hillebrand & Barnes, 2005; Cheyne & Vrba, 2006). Induced responses, as opposed to evoked responses, fail to exhibit time-locking and/or phase-locking to known events (e.g., stimulus presentation or buttonpressing), and will therefore tend to cancel when averaged over trials. Unlike ECD, beamformer techniques are well-suited to analyzing induced responses, as time-course data
69 for each spatial unit is provided by the beamformer, which can then be analyzed over extended periods rather than at a single latency (as required for ECD modeling) (see Hillebrand & Barnes, 2005). The beamformer analyses can provide frequency and power information for each voxel for a given time window, or compare frequency and power changes between one time window (active) and another (baseline). In this way, paradigms where multiple response states are compared (i.e., those used in fMRI) can be translated for use in MEG. Key to the beamformer process is the establishment of weights matrices - sets of values that indicate the relative contribution of each sensor for each voxel in the potential source space. These weights are determined automatically by algorithms that take into account source location, orientation and covariance, and permit spatially discrete prolonged recording across the source space. The primary assumption under beamforming procedures is that each distinct cortical area produces a unique activation profile over time. When discrete cerebral units display highly correlated activation profiles, beamforming tends to identify a single source between actual proximal sources or no source at all (cancellation) when actual sources are distal to each other (Van Veen, van Drongelen, Yuchtman, & Suzuki, 1997). For this reason, beamformer analyses are most appropriate for characterizing induced responses, and least suited for evoked studies, particularly where distal sources tend to activate in a highly coordinated manner (see Hillebrand & Barnes, 2005). Several research groups have proposed methods to control or eliminate errors arising from beamforming of highly correlated distal sources; these approaches typically require a priori knowledge of the spatially distinct sources (see, Herdman, Wollbrink, Chau, et al., 2003; Brookes, Stevenson, Barnes, et al., 2007; Quraan & Cheyne, in press). In MEG studies involving
70 self-paced responding (e.g., generative language tasks), differential beamforming techniques may be used to characterize the induced response. Because distal sources do not correlate over the extended periods used in differential contrasts, differential beamforming is particularly suited to characterizing the neuromagnetic profile of language expression.
71
Chapter 8: Expressive Language Mapping in MEG 8.1 Study rationale In recent years, researchers have established the feasibility of mapping receptive language in healthy and clinical populations using MEG (for a review, see Salmelin, 2007). In the most widely-accepted receptive language paradigm, subjects learn a list of words prior to MEG scanning, and then perform an auditory word recognition task (listen to old and new words, indicating which words were on the initial list) while in the scanner. Listening to spoken words is associated with a well-coordinated evoked (time- and phase-locked) signal in the posterior temporal lobe; modeling the evoked signal permits determination of hemispheric involvement in receptive language processing. Unfortunately, there are no widely-accepted protocols for mapping expressive language in MEG, and none of the published experimental paradigms employ stimuli and response sets that are appropriate for use with young children or low functioning patients. Unlike receptive language, expressive language involves intentional response generation, which is not necessarily time-locked to stimulus presentation. With expressive language, the neuromagnetic signal is not well coordinated and cannot be analyzed using traditional (equivalent current dipole) analyses. Rather, the associated induced neuromagnetic response must be characterized using novel analyses which do not rely on traditional signal averaging techniques for response isolation. In the current study, we set out to determine the feasibility of mapping expressive language using picture stimuli that are appropriate for use with children as young as 5 years of age,
72 with single word covert responding. The neuromagnetic response was assessed using beamforming techniques. We first isolated the target signal using “virtual sensor” analyses, then characterized the extent and location of the expressive language response at both group (control studies) and individual (patient) levels using differential analyses. Findings were accepted for publication in a special MEG applications issue of Down Syndrome Quarterly (Kadis, Smith, Mills, & Pang, 2008), and are included, as published, below.
73
Running Head: Expressive language mapping in children using MEG
MEG localization of expressive language cortex in healthy children: Application to paediatric clinical populations
Darren S. Kadis, M.A. 1, 2, Mary Lou Smith, Ph.D., C. Psych. 1, 2, Travis Mills, B.Sc. 3, Elizabeth W. Pang, Ph.D. 4, 5
1 2 3
Department of Psychology, The University of Toronto
Department of Psychology, The Hospital for Sick Children
Department of Diagnostic Imaging, The Hospital for Sick Children 4
Division of Neurology, The Hospital for Sick Children
5
Department of Paediatrics, The University of Toronto
74
8.2 Abstract There is interest in using MEG to map the spatial-temporal characteristics of the language network. While receptive language in MEG is well studied, protocols for lateralizing and localizing expressive language are not as well developed; currently, there are no standard acquisition and analysis parameters. In this study, we introduce a paradigm designed to map expressive language in children and adolescents. Subjects generate verbs associated with pictures of familiar objects, while neuromagnetic signals are recorded. A group of 12 healthy controls (aged 13-18 years) and a small clinical case series (n = 4; aged 10-13 years) were assessed. Differential beamformer analyses identified task-related low beta desynchronization in left inferior frontal cortex for all subjects. We propose that the protocol may have broader applicability to other paediatric clinical groups.
75 8.3 Introduction Language is a vital aspect of human life and any impingement on the language system is difficult, if not devastating, for an individual. These impingements can be congenital or acquired; that is, they can occur early in life and impact language acquisition or development, or, they can occur later in life as a result of injury or disease. Whatever the aetiology, the inability to develop or maintain a functional language network has significant negative impact on an individual’s quality of life. Because of the importance of language, there has been great interest in understanding the neural components, connections and functions of the human language network. Early lesion studies (Broca, 1861; Wernicke, 1874) and invasive cortical stimulation studies (Penfield & Roberts, 1959) on human patients (for a review, see Engel, et al., 2005), and more recently, the development of non-invasive functional neuroimaging, primarily fMRI and PET, have revealed the primary brain regions involved in language processing (for a review, see Price, 2000). A relative newcomer to the neuroimaging field, whole-head magnetoencephalography (MEG) has emerged as a valuable technique for the non-invasive investigation of neuronal activity in healthy and clinical populations. Over the last decade, the number of studies using MEG to examine language function has increased dramatically (for a review, see Salmelin, 2007). MEG has been used to study a number of language processes, including basic speech perception, reading, and speech production (review, see Salmelin, 2007). While many groups examined language function in healthy adults, there is also a growing literature on speech perception in healthy infants (e.g., Sambeth, et al., 2006, Draganova, et al., 2007).
76 The majority of language research using MEG has been generated from presurgical mapping of adults (for reviews, see Wheless, et al., 2004; Mäkelä, et al., 2006) and children (for reviews, see Otsubo & Snead, 2001; Duchowny, 2007) with refractory epilepsy. MEG is used to lateralize and localize both epileptogenic (seizure-generating) and eloquent tissue; comprehensive mapping maximizes likelihood of seizure reduction through full resection of epileptogenic tissue, while minimizing morbidity through preservation of eloquent cortex. MEG has also been used to study language processes in individuals with dyslexia (e.g., Trauzettel-Klosinski, et al., 2006; for a review, see Sarkari, et al., 2002), autism (e.g., Flagg, et al., 2005; Kasai, et al., 2005; Oram Cardy, et al., 2005), schizophrenia (e.g.; Yamasue, et al., 2004), and brain injury (e.g., Breier, BillingsleyMarshall, Pataraia et al., 2006; Breier, Maher, Novak, et al., 2006). Although language is a complex multi-faceted phenomenon, it can be modeled as a simple two component system: receptive language localized primarily in Wernicke’s area and expressive language in Broca’s area (Geschwind, 1970). Receptive language has been well studied in MEG (for a review, see Papanicolaou, et al., 2006) because language reception can be time-locked to stimulus presentation, and this allows for efficient analysis with basic source localization techniques, such as equivalent current dipoles. A well-used clinical paradigm for receptive language mapping involves listening to a list of previously-learned words while performing a recognition task in the MEG (Breier, et al., 1998; 1999). Consistent with the Geschwind (1970) model of language, in healthy individuals, the sources obtained from this task localize to left posterior superior temporal gyrus (Wernicke’s area) (Breier, et al., 1998; 1999; Papanicolaou, et al., 2006). This technique has been clinically validated against the intracarotid sodium amobarbital
77 procedure (IAP, also known as the Wada test; Wada & Rasmussen, 1960) for lateralization (Papanicolaou, et al., 2004), direct cortical stimulation (Simos, et al., 1999), and fMRI for lateralization and localization of receptive language (Billingsley-Marshall, et al., 2007). As further validation, this task was applied in a group of healthy children (Mohamed, et al., 2008) and analysed with synthetic aperture magnetometry (SAM; Robinson & Vrba, 1999; Vrba & Robinson, 2001). In contrast to dipole analysis, SAM analysis belongs to the family of beamforming, or spatial filtering, techniques that look at changes in magnetic field power across the brain; beamforming is relatively unbiased compared to dipole analysis, as it requires no a priori assumptions about source number or location. In the group of healthy children, consistent with the dipole analyses studies, power decreases were observed in Wernicke’s area (Mohamed, et al., 2008). While the receptive language protocol is well used, an obvious limitation is that it does not engage anterior language regions involved in expressive language processes, and therefore incompletely characterizes the language network. Development of an effective expressive language mapping protocol is relatively challenging, due to the generative nature of expressive language. Active participation from the subject is required; furthermore, there are varied generation strategies that can be employed, and thus it is difficult to consistently obtain either phase- or time-locked responses. Until the development of spatial filtering techniques, such as various beamformer analyses (Van Veen, et al., 1997; Robinson & Vrba, 1999; Gross, et al., 2001; Sekihara, et al., 2001; Vrba & Robinson, 2001; Cheyne, et al., 2006), it was very difficult to obtain appropriate source detection and appropriately characterize expressive language results.
78 Early studies in healthy adult subjects localized expressive language cortex using dipole source analysis and either verb generation (Eulitz, et al., 1994), picture naming (Salmelin, et al., 1994), word completion (Dhond., et al., 2001), or picture categorization (Vihla, et al., 2006) tasks. While these studies localized Broca’s area (left inferior frontal lobe), one frequently identified issue was the inability to obtain good source separation from possibly simultaneously active frontal, temporal, parietal and occipital sources. This is a limitation of dipole source analysis. More recently, beamforming techniques have been applied to expressive language tasks. A study using both verb generation and vowel identification tasks lateralized expressive language (Ressel, et al., 2006); another study using a word generation task (Yamamoto, et al., 2006) reported oscillatory changes in the 8-25 Hz frequency band, and a reading task (Herdman, et al., 2007) identified early activation of inferior frontal areas, probably reflecting preparation for language production. While these few studies have been reported, the literature is sparse and no studies have yet been done in children. In the current study, we introduce a protocol for mapping expressive language in children and adolescents. Subjects silently generate ‘action words’ upon confrontation with pictures of items familiar to young children. Neuromagnetic changes associated with participation in this covert verb generation task are characterized in terms of latency, frequency, and neural location. We present findings from a healthy paediatric control group and a clinical series of children with neurological insult resulting in a seizure disorder and/or language disturbance. Our objective is to demonstrate the utility of this task in a group of healthy children and also to demonstrate its applicability to a small clinical population.
79 8.4 Methods 8.4.1 Participants Sixteen children were tested in this study. Twelve healthy adolescents (5 male, 7 female; mean age = 16.5 years, range = 13-18 years) served as control subjects. Control subjects were in age-appropriate grades and performing successfully in school. They were free of neurological history, learning disability, and language disturbance, as reported by parents. Scores on the Edinburgh Handedness Inventory (Oldfield, 1971) indicated that eleven control subjects were right-handed, one was ambidextrous. Four patients (1 male, 3 females; mean age = 11.25 years; range = 10-13 years) were referred by Neurology or Neurosurgery for participation in this study because there was concern that their neurological disorders affected language. Three patients had a history of intractable seizure disorders; one subject had a history of intraventricular haemorrhage associated with an arteriovenous malformation. All subjects participated on a voluntary basis. Parent consent, and child consent or assent, were obtained in all cases. Subjects received a small gift for their participation. Recruitment, MRI and MEG scanning, and analyses were carried out at the Hospital for Sick Children (Toronto, Ontario, Canada). This study was approved by the Hospital’s research ethics board. 8.4.2 Expressive language paradigm To maximize viability of use across childhood and among patient populations, we developed an expressive language paradigm that can be completed by children as young as
80 5 years of age. Subjects were asked to silently generate an ‘action word’ as quickly as possible when a picture of a familiar object was presented. All subjects were trained on the full task prior to MEG scanning; we required overt responding during the training stage, to assess compliance and accuracy. The stimuli used in training were different from those used in the MEG task. Subjects were also asked to minimize eye movements by looking at a fixation cross during the interstimulus period and to remain still for the duration of testing. 8.4.3 Stimulus presentation A number of standardized psychological batteries and normative studies were consulted to generate a list of 80 items familiar to young children, to which they could generate a verb associated with the item. We obtained exemplary colour digital images of the items and placed each in a small white box background. These photographic images served as our target stimuli (see Figure 8.1 for an example). Interstimulus images were developed by phase scrambling the stimulus images. A fixation cross was superimposed on each scrambled image, to serve as a focal point during the interstimulus period (Figure 8.1). In addition, to ensure attention during the scanning period, we included a vigilance trial which occurred with 15% probability. The vigilance trial consisted of a photo of a hand clicking a computer mouse; subjects were asked to press a button, as quickly as possible, upon presentation of this image. Stimuli were back-projected via mirrors on the ceiling to a screen located 65 cm from the subject’s nose. All images were presented foveally and the white box background subtended 5º of arc. Each target stimulus was presented for 500 ms in random order,
81 without repetition. This was followed immediately by an interstimulus image which was presented with a randomly jittered interstimulus interval between 1500 to 2500 ms. This was immediately followed by the next target image so that no blank screen was ever seen. On a vigilance trial, the vigilance image replaced the stimulus image and remained on the screen for up to 2000 ms, or until the button was pressed. Stimulus delivery was controlled by Presentation software (Neurobehavioral Systems, Albany, CA) and a photo-diode in the MEG room detected the projected stimuli and directly sent a trigger to the MEG acquisition system. The paradigm required less than 4 minutes of MEG scanning for completion. MEG Data Acquisition Prior to scanning, each subject was required to remove all metal. Fiducial markers were placed on the subject’s nasion, and left and right pre-auricular points. Subjects were tested supine in a magnetically shielded room at the Hospital for Sick Children. Neuromagnetic activity was recorded at 625 samples/sec, and a bandpass of DC-100 Hz, using a wholehead 151-channel MEG system (VSM Medtech, Coquitlam, BC, Canada). Subjects were required to remain as still as possible for the duration of the testing; in all cases, subject compliance was confirmed with recorded head motion of less than 0.5 cm during data acquisition. 8.4.4 Anatomical MRI acquisition and co-registration For all subjects, we acquired 3D-SPGR T1 MR images at 1.5T (Signa Advantage System, GE Medical, WI). Patients typically underwent further MRI studies as part of their clinical investigations. Prior to the MRI acquisition, the MEG fiducial markers were replaced with MRI contrast-sensitive markers. In cases where MEG and MRI scans occurred on separate
82 visits, digital photographs were used to ensure consistent placement. Each subject’s MEG study was coregistered to their structural MRI. Segmentations were performed on the MRI volumes using BrainSuite2 (Sandor, et al., 1997; Shattuck, et al., 2001; 2002; Dogdas, et al., 2005;;); the inner skull morphology was used to develop multiple sphere models for later beamformer analyses of MEG data. 8.4.5 Time-frequency response analyses To characterize neural activity associated with the verb generation task, we placed 12 ‘virtual sensors’ (Cheyne, et al., 2006) bilaterally along the posterior inferior frontal gyrus (pars triangularis, pars opercularis), and along the middle frontal gyrus at the cortical surface in each subject. In the left hemisphere, these sites are known to support expressive language (i.e., Broca’s area and surrounding cortex); in subjects with atypical language lateralization, the homotopic cortex of the right hemisphere is believed to support expressive language (e.g., Jabbour, et al., 2005). As a first pass, the virtual sensors were used to assess power changes over a broad frequency spectrum (0-100 Hz), related to the onset of each target stimulus in comparison to a 300 ms pre-stimulus baseline period. The neuromagnetic activity localized to each virtual sensor site was visualized in timefrequency response (TFR) plots and assessed across control subjects. Virtual sensor sites with consistent oscillatory changes were identified, and averaged across subjects. We analyzed the average TFR plots to determine optimal frequency and time windows for significant task-related changes. Findings were used to direct all subsequent differential beamformer analyses for both control and clinical subjects.
83 To characterize effects in the control group, MEG and MRI data were converted to Analyze format, spatially normalized and averaged using SPM2 routines (Friston, 2003). Patient data were not pooled; case findings are presented. 8.5 Results 8.5.1 Virtual sensor time-frequency response analyses In each control subject, we observed beta desynchronization at virtual sensors placed on the inferior frontal and middle frontal gyri of the left hemisphere. We noted substantial variability of neuromagnetic activity across virtual sensor locations, and across subjects. The most consistent oscillatory changes occurred at the superior aspect of left pars opercularis, canonical Broca’s area. We performed a permutation test (2000 permutations) on the averaged control subject TFR plots established at the superior aspect of the operculum, and observed significant low-beta desynchronization (at approximately 13-23 Hz) from 200-900 ms following the onset of target image presentation (p < 0.005, uncorrected) (see Figure 8.2). The low-beta desynchrony remained significant at 450- 650 ms with more conservative permutation testing (p < 0.001, uncorrected). 8.5.2 Differential beamformer analyses The findings from the averaged control subject time-frequency response analyses were used to direct subsequent differential beamformer analyses for both control subjects and patients. The differential beamformer parameters were as follows: baseline period = -200 to 0 ms (i.e., the 200 ms period immediately preceding target presentation); active period = 450 to 650 ms post stimulus; and, frequency band = low-beta (13-23 Hz).
84 8.5.3 Individual control subjects On a case-by-case basis, using the above parameters, we observed the expected beta desynchrony over the left inferior frontal lobe. In some individuals, beta synchronization was seen over approximately homologous areas of the right hemisphere. Also, in all cases, bilateral occipital desynchronization was observed. 8.5.4 Control group The group analyses revealed beta desynchronization over the posterior inferior frontal gyrus (canonical Broca’s area) and neighboring precentral gyrus. Beta synchronization was observed over the right middle frontal gyrus. Neuromagnetic changes over the left and right hemispheres are presented in Figure 8.3. 8.5.5 Clinical case series Case 1. Patient A is a 13 year old right handed female with refractory epilepsy since 1 year of age, secondary to encephalitis. At age 6 years, the patient underwent pre-operative workup for subdural grid placement. EEG and MEG revealed interictal spikes arising from the left temporal, parietal, and frontal regions. MRI revealed extensive left hemisphere abnormality and PET showed hypermetabolism in the left temporoparietal region. Both fMRI and IAP/Wada for language lateralization were inconclusive. Electrocorticography from left hemisphere subdural grid and temporal depth electrodes indicated extensive ictal involvement of the mesial temporal, temporal, frontal and parietal cortices. Extraoperative stimulation mapping revealed two language naming sites, one at the left posterior middle temporal gyrus and another at the second convolution of the left inferior frontal gyrus (pars
85 triangularis). A temporal lobectomy with cortical excision over the posterior portion of the superior temporal gyrus and multiple subpial transections over the inferior parietal and posterior inferior frontal gyrus were performed. In the immediate postsurgical period, the patient experienced only moderate seizure relief, with intact language. The seizures continued and at age 13, the patient was re-investigated for candidacy of repeat surgery. As part of her MEG work-up, the patient completed the verb generation paradigm. Beta desynchrony was observed in left inferior frontal and middle frontal gyri, anterior and superior to canonical Broca’s area with no desynchrony over the superior portion of the left frontal operculum. This finding is mildly atypical although entirely consistent with the extraoperative stimulation mapping results. Also consistent with the extraoperative stimulation results, we observed beta desynchronization in the left middle posterior temporal cortex. Finally, in the right hemisphere, we observed beta synchronization, similar to what was seen in the control participants. Case 2. Patient B is a 10 year old right handed male with refractory seizures since 8 years of age. During his seizures, the patient may vocalize, although the expression does not resemble speech. Interictal EEG suggested left central epileptogenesis while MEG revealed spikes in left central and temporal regions, and bilateral parasagittal regions. PET revealed hypoperfusion of Rolandic and perisylvian cortical areas although MRI was reported normal. Our standard receptive language protocol (Chuang, et al., 2006) indicated left posterior temporal localization. The expressive language paradigm demonstrated desynchronization over the left inferior frontal lobe, suggesting typical lateralization and localization.
86 Case 3. Patient C is an 11 year old right-handed female with refractory seizures since 3 years of age. During her seizures, the patient experiences a brief behavioural arrest, followed by abnormal bilateral hand posturing accompanied by screaming. Prolonged video EEG recordings were unable to lateralize the seizure onset; however, MEG revealed a cluster of interictal spikes in the right frontal lobe and PET reported hypometabolism of the right inferior frontal region. MRI was reported normal. The expressive language task demonstrated beta desynchronization over the superior portion of the left frontal operculum, suggesting typical lateralization and localization of expressive language. Case 4. Patient D is an 11 year old female with a recent history of intraventricular hemorrhage secondary to an arteriovenous malformation. Cerebral angiogram revealed a deep left frontal AVM with predominant involvement of left anterior cerebral artery; a complex of other arteries, veins and right ACA were also implicated. Neuropsychological assessment of speech and language revealed some deficits in the initiation and maintenance of discourse, although gross expressive and receptive language remained intact. Our MEG expressive language task demonstrated beta desynchronization over the left inferior frontal cortex, suggesting typical lateralization and localization of expressive language. 8.6 Discussion The combination of a covert verb generation task and differential beamformer analyses proved useful for mapping cortical regions responsible for expressive language processing
87 on both a group level for the controls, and at an individual level for both control and clinical subjects. Although others have demonstrated the utility of MEG for assessing receptive language lateralization and localization in adults and children, this is the first study, to our knowledge, to characterize expressive language representation in healthy paediatric subjects using MEG. Furthermore, we report on a small clinical paediatric case series where we demonstrated how findings from control subject studies can be used to direct analyses in cases where there is concern that language representation may be atypical due to developmental plasticity. In the paediatric control subjects, current findings demonstrate typical lateralization and localization of expressive language. Specifically, the covert verb generation task induced low beta event-related desynchrony (ERD; see Pfurtscheller & Aranibar, 1977) over the superior portion of pars opercularis, canonical Broca’s area. The low beta band desynchrony is consistent with an MEG study of word generation in adults (Yamamoto, et al., 2006). The parameters of the active time window are consistent with a MEG reading task in adults (Herdman, et al., 2007) and MEG models of the clinical neurophysiology of language (Salmelin, 2007). ERD has been interpreted as an interruption of synchronous resting activity reflecting taskrelated engagement (Pfurtscheller & Lopes da Silva, 1999) and our observation of ERD in canonical regions suggests that the healthy adolescents in this study have lateralized and localized expressive language in an adult-typical manner. This is consistent with fMRI studies comparing adult and paediatric language representation (e.g., Gaillard Balsamo, Ibrahim et al., 2003; Gaillard, Sachs, Whitnah, et al., 2003; for a review, see Sachs & Gaillard, 2003).
88 Interestingly, in the averaged control group, we observed ERS over the right frontal region. Event-related synchrony (ERS) is thought to reflect task deactivation or inhibition (Pfurtscheller & Lopes da Silva, 1999); these findings may reflect feedback/inhibition loops in the language network whereby activation of left inferior frontal areas inhibits the approximately homologous right hemisphere areas. In the motor system, there is evidence of transcallosal inhibitory connections between primary motor cortices to prevent mirror movements (Cheyne & Weinberg, 1989; Kristeva, et al., 1991). For the language network, further studies need to be completed to elucidate the significance of these synchronous oscillatory changes. In addition to inferior frontal neuromagnetic effects, we observed bilateral occipital ERD in all cases, believed to represent primary visual processing of the target images. Although we controlled for contrast and luminance differences, we still observed substantial occipital ERD. In our differential analyses the ‘active period’ is proximal to target presentation onset, whereas the ‘baseline period’ is relatively distal to the interstimulus image presentation onset; the target image is therefore relatively novel during the active period, while subjects may have habituated to the interstimulus images by the baseline period. Furthermore, the targets are meaningful and require responding, whereas the interstimulus images are necessarily less engaging. In the paediatric clinical case series, our identification of left inferior frontal ERD in all of the patients confirms the suitability of this task for application in clinical populations. When these patients were referred to us, there was initial concern that their disease might have disrupted or displaced expressive language. However, our testing showed that in 4 of 4 cases, expressive language was lateralized to the left hemisphere, and in 3 of 4 cases
89 (Patients B, C & D), expressive language localized to canonical Broca’s area. In the case with atypical localization (Patient A), the expressive language task in MEG identified locations consistent with language-positive sites obtained by extraoperative stimulation mapping prior to epilepsy surgery. This case serves as direct confirmation of the ability of our task to localize expressive language. The mean age of our clinical group is slightly younger than the mean of the control group; however, we successfully obtained the expected lateralizations and localizations in both groups. There are several design features of the task that contributed to this success. Our covert verb generation task was designed to be appropriate for use in young children and individuals who have limited language ability. By selecting high-frequency (i.e., familiar) stimuli and a simple response set (i.e., think of an action word), we maximized the likelihood of adequate performance and applicability across individuals with potentially differing language sophistication. The use of a covert task with an embedded vigilance component allowed us to minimize jaw and head movements, which is a critical consideration when testing children. The vigilance task served a dual purpose of maintaining interest in the child while allowing us to monitor attention to task. Finally, incorporating an overt training component allowed us to ensure compliance and monitor performance. While the stimuli used in this task were very simple, it still provided clear localizations even in the older adolescents. It is not clear whether task difficulty modulates engagement of language cortex across early development (Brown, et al., 2005; Szaflarski, et al., 2006); however, future studies of potential developmental changes for expressive language
90 processing and representation may benefit by using a basic and common set of stimuli across ages. There is still a need to validate this task for lateralization (i.e., against fMRI and the ‘gold standard’ IAP/Wada test) and localization (i.e., against fMRI and direct cortical stimulation mapping) in these and other clinical groups. As well, greater numbers of healthy adults, adolescents, and younger children need to be tested. However, the evidence that we have presented here suggests that we have developed a task capable of localizing expressive language in children and adolescents, and this task has applicability to clinical populations where language function may be compromised. Our finding of both ERD in canonical language areas and ERS in approximately homologous right hemisphere regions points to the possibility that this task and analysis protocol may tap into different components of the complex and anatomically-widespread language network; thereby, permitting improved characterization of expressive language processing and ultimately promising improved understanding of language as a system.
91
8.7 Acknowledgements The authors would like to thank Sonya Bells, Marion Malone, and Frank Wang for assistance with data acquisition. DSK is supported through a Studentship, in part, by the Ontario Student Opportunity Trust Fund - Hospital for Sick Children Foundation Student Scholarship Program, and through a Doctoral Research Award from the Canadian Institutes of Health Research (CIHR) in partnership with Epilepsy Canada.
92
Figure 8.1
Example of a target stimulus and an interstimulation fixation stimulus.
93 Figure 8.2 The image part with relationship ID rId12 was not found in the file.
Averaged time-frequency plot (TFR) for control subjects from virtual sensors placed at the superior portion of left pars opercularis (canonical Broca’s area). Significant low-beta (13-23 Hz) desynchronization was observed at 200-900 ms (p < 0.005, uncorrected). This ERD remained significant in the 450-650 ms window with more conservative permutation testing (p < 0.001, uncorrected).
94
Figure 8.3
(right panel) Averaged data from the healthy control subjects demonstrating beta desynchrony (blue) over left posterior inferior frontal gyrus (canonical Broca’s area) and neighbouring precentral gyrus. (left panel) Averaged data from the healthy control subjects demonstrating beta synchronization (pink) over the right middle frontal gyrus.
95
Chapter 9: Characterizing Expressive Language Representation Across Childhood 9.1 Study rationale It is generally accepted that the paediatric brain is relatively plastic for language compared to the adult brain; however, no consensus exists on the mechanism(s) for the paediatric advantage. Two competing theories have been proposed to explain atypical lateralization (e.g., Rasmussen & Milner, 1977) and localization (e.g., Kadis et al., 2007) observed in the context of early neurological insult: 1) children’s language networks look much like adult networks; adult-atypical representation following early injury occurs through recruitment of neural tissue not typically involved in the control of language; 2) children’s language networks are relatively bilateral and extensive, becoming gradually lateralized and focal through childhood and/or adolescence; adult-atypical representation occurs through focalization away from the cerebral insult. In the current study, we attempt to address questions regarding plasticity in the context of early injury through characterization of the normal developmental trajectory for expressive language representation. In Kadis et al. (2008; see Chapter 8), we established the feasibility of using a brief verb generation task in MEG to characterize expressive language representation in healthy children and a small patient population. At a group level, healthy controls demonstrated consistent low-beta ERD in the left inferior frontal lobe, as expected. For individual patient scans, we adopted parameters deemed optimal in control group analyses, and succeeded in lateralizing and localizing ERD through visual inspection. The lack of a reliable objective thresholding procedure precluded meaningful subject-wise comparisons of control data.
96 In the current study, we employ two MEG language tasks, naming and verb generation, to characterize expressive language representation in a group of healthy subjects aged 5 to 18 years. In order to comment on changes from early childhood through young adulthood, we developed a novel analytic protocol that permitted tailored yet objective analyses of single subject data. The protocol facilitates quantification of lateralization and extent of ERD on an individual basis; these measures are examined for relationship with subject age. Findings have been summarized in the following manuscript.
97 Running head: EXPRESSIVE LANGUAGE DEVELOPMENTAL TRAJECTORY
Characterizing the Normal Developmental Trajectory of Expressive Language Lateralization using Magnetoencephalography
Darren S. Kadis1, 2, Elizabeth W. Pang3, 4, Travis Mills5 Margot J. Taylor1, 5, Mary Pat McAndrews1, 6, Mary Lou Smith1, 2, 3
1 2
Department of Psychology, Hospital for Sick Children 3
4
Department of Psychology, University of Toronto
Division of Neurology, Hospital for Sick Children
Department of Pediatrics, Division of Neurology, University of Toronto 5
6
Diagnostic Imaging, Hospital for Sick Children
Brain Imaging & Behaviour Systems, Toronto Western Research Institute
98
9.2 Abstract Purpose: To characterize the developmental trajectory for expressive language representation and test competing explanations for the relative plasticity of language representation in childhood. Method: We studied 25 healthy children and adolescents (aged 5-19 years) participating in covert picture naming and verb generation tasks in MEG. Extent and lateralization of neuromagnetic responses were quantified using a novel thresholding procedure for differential beamformer analyses. Results: Extent of cortex supporting naming and verb generation was highly variable across subjects, and did not correlate with age. However, left hemisphere lateralization tended to increase for both naming and verb generation across the developmental period; this observation was supported by a significant correlation between verb generation laterality index scores and age. Discussion: Findings suggest that adult-typical left hemisphere lateralization emerges from an early bilateral language network. Early bilaterality may explain the paediatric advantage for interhemispheric plasticity of language.
99 9.3 Introduction Rasmussen and Milner (1977) were among the first to demonstrate convincingly the developmental limits for plasticity of language representation. In their study of language lateralization in epilepsy surgery candidates (assessing laterality using the intracarotid sodium amobarbital procedure, or IAP; Wada, 1949; 1997), Rasmussen and Milner found that atypical (bilateral or right hemisphere) language representation was most common in individuals with seizure onset in the first few years of life. Those with adult-onset epilepsies were less likely to demonstrate atypical lateralization, and were more likely to experience language deficits (see also, Branch, Milner, & Rasmussen, 1964). The relative plasticity of language representation in childhood has been well documented using IAP (Rasmussen & Milner, 1977; Satz, Strauss, Wada, & Orsini, 1988; Helmstaedter, Kurthen, Linke, & Elger, 1997, Brazdil, Zakopcan, Kuba, et al., 2003; Saltzman-Benaiah, Scott, & Smith, 2003; Kadis, Kerr, Rutka, et al., 2009), PET (Duncan, Moss, Bandy, et al., 1997; Muller, Rothermel, Behen, et al., 1998; Muller, Rothemel, Behen, et al., 1999), fMRI (Springer, Binder, Hammeke, et al., 1999), and recently magnetoencephalography (MEG) (Pataraia, Simos, Castillo, et al., 2004). Despite the converging results from these studies, little is known about the mechanisms underlying the paediatric advantage in language neuroplasticity. Two competing theories have been proposed to explain how atypical language representation establishes following early injury: (1) ‘immature’ language networks look much like adult networks; in cases where language representation is adult-atypical, function has reorganized and brain regions not typically involved have been recruited to support language; the paediatric brain has a relative propensity to recruit extra-canonical neural
100 resources for language processing, possibly due to non-commitment of those regions; (2) ‘immature’ language networks are extensive and bilateral; language establishes into noncanonical (adult-atypical) regions following early insult as diffuse networks precede focal networks in the normal developmental trajectory. Indeed, each theory has received support through functional neuroimaging of healthy subjects across childhood and adolescence. Gaillard, Sachs, Whitnah, et al. (2003) employed fMRI to compare healthy children, aged 7 to 14 years, with adults while engaging in a semantic fluency task. The researchers failed to observe differences in location or extent of activations between the child and adult groups (cp, Gaillard, Hertz-Pannier, Mott, et al., 2000). Similarly, Wood, Harvey, Wellard, et al. (2004) compared asymmetry and extent of activations in children and adults completing a verb generation and orthographic lexical retrieval (fluency) task in fMRI. Although children (aged 6 to 15 years) demonstrated a higher rate of atypical lateralization (15% of children, compared to only 6% of adults), the difference was not statistically significant, and the localization of activations was comparable. These findings suggest that children’s language networks are similar to those of adults. Therefore, atypical representation in the context of early injury reflects shifts or reorganization from canonical to contralateral or perilesional regions; atypical representation is de novo, following early neurological insult. In contrast, Holland, Plante, Weber Byars et al. (2001) used fMRI to assess healthy children aged 7-18 years participating in a verb generation paradigm, and found that left hemisphere lateralization increased with age. Brown, Lugar, Coalson, et al. (2005) used fMRI to assess cortical activity in healthy participants aged 7-32 years during three performance-matched overt word generation tasks, and observed relatively widespread and bilateral
101 representation in children, whereas adults demonstrated language representation focused in frontal and parietal regions of the left hemisphere. Using MEG, Ressel, Wilke, Lidzba, Lutzenberger, and Krageloh-Mann (2008) studied hemispheric differences in 7-16 year old children completing an overt verb generation and vowel identification task, and found that left hemisphere lateralization increased with age. These findings support a theory of extensive and bilateral language representation in childhood. Atypical representation following early neurological insult is facilitated by normal widespread paediatric language networks that fail to develop into adult-typical focal networks due to an interrupted or altered developmental trajectory. The inconsistent findings regarding changes in language lateralization across childhood may be explained, in part, by varied task selection. Language is not a unitary function; unique cognitive demands are likely to be associated with different profiles of neural engagement. There is some evidence for distinct patterns of lateralization of expressive versus receptive language in healthy children and adults (Szaflarski, Holland, Schmithorst, & Byars, 2006; compared to expressive language, receptive language functions are relatively plastic in childhood, see, Boatman, Freeman, Vining, et al., 1999). Differences in the literature may reflect the variable engagement of expressive or receptive components of the language network. Ideally, studies addressing the developmental trajectory of language representation should focus on well defined, distinct language processes. To minimize differences associated with effort, performance should be matched or nearly matched across study participants. In order to test the two competing theories of the development of language representation, we recently developed two MEG expressive language paradigms for use with healthy
102 subjects and children with neurological insults (one of the paradigms, verb generation, was introduced in Kadis, Smith, Mills, & Pang, 2008). We preferred MEG over other neuroimaging modalities (e.g., PET, fMRI). MEG is non-invasive, and in our experience, less intimidating than fMRI for use with young children. The MEG scanner is completely silent, the MEG dewar encompasses the head only, and fast recording of fleeting neuromagnetic signals permits use of brief paradigms, thus minimizing the demand for prolonged periods without motion. The two tasks we developed, covert naming and covert verb generation, involve developmentally appropriate picture stimuli, do not require participants to read, and can be administered in any language. Covert responding minimizes movement and muscle artifact, thus maximizing signal-to-noise. In healthy children and adults, expressive language responses have been characterized by low-beta event-related desynchrony (ERD) of neuromagnetic activity in the left inferior frontal lobe (Kadis et al., 2008; see also, Ressel et al., 2008). In this study, we employed our covert naming and verb generation tasks to assess expressive language representation across childhood. The simplicity of each task facilitated use with children as young as 5 years of age, permitting characterization of representation around the generally accepted age-limit for interhemispheric plasticity (see Rasmussen & Milner, 1977). We found consistent ERD for naming and verb generation across subjects, although latency for maximal ERD varied on an individual basis. To objectively assess ERD laterality, we developed an individual thresholding protocol using the novel application of bootstrap statistics in differential beamformer analyses. To isolate the expressive components of naming and verb generation, we analyzed neuromagnetic changes occurring within an extensive frontal lobe region-of-interest (ROI). The power
103 and spatial extent of ERD within the frontal lobes were separately analyzed; total ERD and number of surviving voxels in the left versus right hemisphere were summarized and assessed for relationship to participant age. 9.4 Method 9.4.1 Participants Twenty-five children and adolescents (15 male, ranging in age from 5.2 to 18.9 years, mean age 12.9 years) participated in this study. Subjects were recruited from the community, and were free of any history of neurological disorder, learning disability, and language disturbance. Twenty-four subjects completed the Edinburgh Handedness Inventory (Oldfield, 1971); scores indicated that 22 were right handed, 2 had mixed handedness. The 1 subject that did not complete the inventory reported right hand dominance. Subjects over 18 years of age provided written consent; parents provided written consent for subjects less than 18 years of age, and the children assented to participate in the study. Subjects received a small gift for their participation. MEG and MRI scanning and analyses were carried out at the Hospital for Sick Children (Toronto, ON, Canada). The study was approved by the Hospital's research ethics board. 9.4.2 Expressive language tasks, stimuli and presentation The development and implementation of our verb generation paradigm has been previously documented (Kadis et al., 2008); the naming paradigm was established simultaneously, in exactly the same manner. Stimuli and procedures are described only briefly, below.
104 Based on a number of standardized language batteries and normative studies, we established a 160-item pool of objects whose names and/or usage are familiar to typically developing 5 year old children. From the pool of items, 80 were selected for the naming task, 80 for verb generation. We obtained exemplary colour digital photographs of each object for presentation on a plain white background. These images served as our test stimuli. Inter-trial fixation stimuli were phase-scrambled colour images with a superimposed central black fixation cross. Examples of test and fixation stimuli are presented in Figure 9.1. To promote vigilance during the scanning period, we also included a picture of a hand clicking a computer mouse; subjects were asked to quickly button-press upon presentation of this stimulus. Prior to MEG scanning, subjects were trained on overt versions of the naming and verb generation tasks (using a separate set of training stimuli). Once compliance was established, subjects were instructed to begin responding covertly. We proceeded with scanning only after we determined that the subject was familiar with and able to comply with the task requirements. Stimuli were back-projected to a screen fixed in front of the opening of the MEG dewar, approximately 65 cm from each subject's eyes. The use of small, 12 cm square images, promoted foveal viewing – images were contained within 2-3º of the center of the visual field. Stimuli were delivered using Presentation software (Neurobehavioral Systems, Albany, CA); a photo-diode in the MEG room detected projected stimuli and directly triggered the MEG acquisition system for accurate trial epoching. 9.4.2.1 Covert naming to confrontation
105 Subjects viewed alternating test and fixation stimuli. Test images were presented for 500 ms in random order, without repetition. For each test image, subjects were asked to think of the name of the object pictured, as quickly as possible. Fixation stimuli were presented for 1500-2500 ms (duration randomly jittered). Subjects were instructed to simply focus on the central cross. Vigilance trials appeared in place of test stimuli at a 15% probability of occurrence, and remained on screen for 2000 ms or until the subject button-pressed. The task required less than 4 minutes of MEG scanning for completion. Following the scans, response accuracy was assessed in children aged 10 years and younger by repeating the task with overt responding. Older children were not assessed for accuracy, as the normative data predict ceiling or near-ceiling performance beyond 10 years of age. 9.4.2.2 Covert verb generation to confrontation The verb generation task was presented in exactly the same manner as the naming task. Subjects were asked to covertly generate “action words” corresponding to test stimuli. To assess accuracy, children aged 10 years and younger repeated the verb generation task with overt responding following the MEG scan. 9.4.3 MEG data acquisition Subjects were required to remove all metal prior to scanning. Fiducial markers were placed at the nasion and left and right pre-auricular points. All subjects were tested in the supine position in a magnetically shielded room which houses the MEG sensors. Neuromagnetic activity was recorded at 625 samples per second, at DC-100 Hz bandpass, using a 151channel whole-head MEG system (VSM Medtech, Coquitlam, BC, Canada). Subjects were
106 asked to remain as still as possible for the duration of testing; in all cases, compliance was confirmed with recorded head motion of 5 mm or less over the scanning period. 9.4.4 Anatomical MRI acquisition and coregistration MEG fiducials were replaced with MRI contrast-sensitive markers for coregistration of functional and structural data. Subjects underwent 3D-SPGR T1 MR imaging of the whole head at 1.5T (Signa Advantage System, GE Medical, WI). The 3D volume was automatically tissue segmented using BrainSuite2 (Dogdas, Shattuck, & Leahy, 2005; Sandor, & Leahy, 1997; Shattuck & Leahy, 2002; Shattuck, Sandor-Leahy, Schaper, Rottenberg, & Leahy, 2001) to establish inner skull morphology. A mask of each subject's inner skull was used to develop multiple sphere models for beamforming analyses. 9.4.5 Differential beamformer analyses with bootstrap-derived thresholds Neuromagnetic activity associated with naming and verb generation was assessed using differential beamformer analyses (see Robinson & Vrba, 1999; Sekihara, Nagarajan, Poeppel, Marantz, & Miyashita, 2001; Van Veen, van Drongelen, Yuchtman, & Suzuki, 1997; Vrba & Robinson, 2001). Beamforming is a spatial filtering technique that permits characterization of oscillatory changes throughout the brain. The differential approach involves direct comparison of an active and a baseline period over a focused frequency range. Previous investigations revealed largely consistent ERD in the left inferior and middle frontal gyri for naming and verb generation between 13 and 23 Hz, corresponding to the low-beta band (Kadis et al., 2008; see also Ressel et al., 2008). This ERD may be characteristic of a variety of language processes, including reading of single words (e.g., Hirata, Kato, Taniguchi, et al., 2004). Group low-beta ERD occurred between 200-800 ms
107 following stimulus presentation; however, individuals typically demonstrated brief (approximately 200-400 ms in duration) ERD at latencies that varied from subject-tosubject. We necessarily focus on brief periods in differential analyses, as contrasts with lengthy windows tend to include non-relevant neuromagnetic changes, potentially masking target signals. To optimize individual analyses in the current study, we computed differential beamformer analyses for 13-23 Hz activity during four overlapping active windows for each subject: 300-500 ms, 400-600 ms, 500-700 ms, and 600-800 ms following the onset of test stimulus presentation. Active windows were contrasted against a single baseline window consisting of the 200 ms period immediately preceding test stimulus presentation. The sliding window approach permits individual tailoring of analyses while maintaining objectivity and power at a single subject level. Thresholding of individual data must be sufficiently flexible in order to accommodate individual variability in signal strength and location, yet be objectively determined so as to remain meaningful in between-subject comparisons. The beamformer relies on multiple trials for establishment of a reliable covariance matrix, necessary for accurate source analyses, precluding direct statistical comparison of active and baseline windows on a trial-to-trial basis. Therefore, to objectively assess the reliability of observed ERD using all available trials for each of the naming and verb generation runs, we developed a bootstrap statistical procedure, whereby observed data are randomly sampled with replacement to establish possible alternate data sets (pseudo runs), collectively providing distributions of voxel-wise neuromagnetic changes. In the current implementation, we established 99 pseudo runs of 80 trials each (for every naming and verb generation study). The observed
108 ERD per voxel was then assessed across runs (actual and pseudo) using non-parametric inferential statistics at an a priori chosen alpha of p < 0.01, uncorrected. 9.4.6 Region of interest The neural substrate for expressive language in both healthy and neurological populations is confined to the frontal lobes (see Penfield & Roberts, 1959; Ojemann, 1979; Ojemann, Ojemann, Lettich, & Berger, 1989; Kadis, Iida, Kerr, et al., 2007). In preliminary analyses for both naming and verb generation, we observed expected frontal lobe ERD surviving the bootstrap procedure; we also observed a strong posterior signal, focused over the primary visual cortex, reflecting visual processing of the picture stimuli. To quantify frontal lobe involvement in naming and verb generation, and to control for distal posterior signals, we restricted analyses to a probabilistic volume of the human frontal lobe (developed by the International Consortium for Brain Mapping, made publicly available through the University of California’s Laboratory of Neuro Imaging at http://www.loni.ucla.edu). Individual scans were automatically spatially normalized using SPM2 routines (Friston, 2003), then trimmed to exclude extra-frontal ERD. 9.4.7 Analyses of ERD for naming and verb generation 9.4.7.1 Laterality index for ERD power To determine the relative power of ERD within the left versus right frontal lobe for naming and verb generation, we computed laterality indices (LIERD) for all surviving data across the four active-baseline contrast windows. Total left (ERDL) versus total right (ERDR) hemisphere desynchrony was compared, as follows:
109 LIERD = (ERDL – ERDR) ÷ (ERDL + ERDR) LIERD scores range in value from +1 (completely left) to -1 (completely right). Scores around 0 indicate bilateral contributions. Only frontal lobe ERD surviving the bootstrap procedure was considered in computation of LIERD. 9.4.7.2 Total and hemispheric extent of ERD To assess the spatial extent of ERD, we summed the number of voxels surviving the bootstrapping procedure across all four contrast windows. A greater number of surviving voxels indicates relatively widespread source, reflecting widespread expressive language representation. We computed laterality indices (LIVOX) for total number of surviving voxels within the left (VOXL) and right (VOXR) frontal lobes, as follows: LIVOX = (VOXL – VOXR) ÷ (VOXL + VOXR) LIVOX scores range in value from +1 to -1, with scores around 0 representing an equal number of surviving voxels in the left and right frontal lobes. This particular laterality index does not take into account the power of ERD observed at each surviving voxel, but serves to compare extent of left versus right frontal lobe contributing to naming and verb generation. Only frontal lobe ERD surviving the bootstrap procedure was considered in computation of LIVOX. We assessed the correlation of LIERD and LIVOX within naming and verb generation to determine the uniqueness of each as a measure of lateralization. We also compared LI scores for naming and verb generation to determine the consistency of ERD lateralization across expressive language tasks.
110 9.5 Results All participants correctly button-pressed upon presentation of vigilance trials in naming and verb generation paradigms. 9.5.1 Naming Twenty-three subjects completed the covert naming task. In 4 of the oldest subjects (age 18.0-18.9), we failed to observe low-beta ERD surviving the bootstrap procedure, in either hemisphere. These participants were excluded from further laterality assessment on this task. Post-scan testing in children aged 10 years and younger indicated near perfect naming accuracy (scores ranging from 98.8% to 100%, M = 99.9%). Individual demographic, performance, and neuromagnetic findings are presented in Table 9.1. 9.5.1.1 Laterality index for ERD power for naming There was a tendency for naming LIERD scores to increase with age, suggesting that the left frontal lobe becomes increasingly dominant for naming across childhood (Figure 9.2). However, the correlation between LIERD scores for naming and age was not significant, r = 0.24, n = 19, p > 0.05. 9.5.1.2 Extent of ERD for naming There was considerable variability in the number of surviving voxels observed for each subject (M = 191.6, SD = 3071). In 4 of the 5 oldest subjects, we failed to observe ERD surviving the bootstrap procedure. In subjects with surviving data, number of voxels did not correlate with age at assessment, r = -0.11, n = 19, p > 0.05.
111 LIVOX scores ranged from -1 to +1 (M = 0.25, SD = 0.74, n = 19), and did not significantly correlate with subject age, r = 0.25, p > 0.05, n = 19. 9.5.1.3 Comparison of LIERD and LIVOX for naming Naming laterality indices built on ERD power versus number of surviving voxels were virtually identical (see Table 9.1). The LIERD and LIVOX scores for naming were perfectly correlated, r = 1.00, p < 0.05, n = 19. 9.5.2 Verb generation Twenty-four subjects completed the covert verb generation task. In two subjects, aged 9 and 18 years, we failed to observe low-beta ERD surviving the bootstrap procedure. Postscan verb generation accuracy ranged from 70% to 100% (M = 88%) in children aged 10 years and younger. Individual performance measures for the youngest participants and neuromagnetic findings for the whole group are presented in Table 9.2. 9.5.2.1 Laterality index for ERD power for verb generation LIERD scores for verb generation significantly increased with age, r = 0.40, p < 0.05, n = 22 (see Figure 9.3). 9.5.2.2 Extent of ERD for verb generation The number of surviving voxels varied considerably across subjects (M = 188.2, SD = 231.9). Among those with surviving data, number of voxels did not correlate with subject age, r = 0.19, n = 22, p > 0.05.
112 LIVOX scores for verb generation significantly correlated with subject age, r = 0.42, n = 22, p < 0.05. 9.5.2.3 Comparison of LIERD and LIVOX for verb generation Laterality indices for ERD power versus surviving voxel count were virtually identical (see Table 9.2 values for each subject). The LIERD and LIVOX for verb generation were perfectly correlated, r = 1.00, n = 22, p < 0.05. 9.5.3 Naming versus verb generation lateralization Total power of surviving ERD for naming and verb generation were significantly correlated, r = 0.58, n = 22, p < 0.05; total number of surviving voxels for naming and verb generation were also significantly correlated, r = 0.56, n = 22, p < 0.05. Since LIERD and LIVOX scores were found to be redundant in both naming and verb generation analyses, we only report LIERD scores for task-wise comparisons of lateralization. Seventeen subjects had ERD surviving the bootstrap procedure for both naming and verb generation. Naming and verb generation LIERD scores were not significantly correlated, r = 0.18, n = 17, p > 0.05. Indeed, in five cases, predominant ERD for naming was observed contralateral to predominant ERD for verb generation. 9.6 Discussion In the current study, we characterized changes in expressive language representation from childhood through adolescence. In studying the normal developmental trajectory, we
113 hoped to advance our understanding of how atypical language representation establishes in the context of early injury. To our knowledge, this is the first study to use age-appropriate stimuli and tailored objective individual analyses of neuromagnetic data to assess language representation in healthy children. We observed differences in naming and verb generation ERD lateralizations across development. For naming, we failed to observe a significant correlation between participant age and ERD lateralization. In contrast, we observed a significant increase in left lateralization with advancing age for verb generation. Importantly, the regression plots for the two tasks suggest a common developmental trajectory characterized by increasing left lateralization across childhood (Figure 9.2 and Figure 9.3). The observed developmental trajectory is consistent with recent neuroimaging studies showing differences in language lateralization of children and adults (Holland et al., 2001; Brown et al., 2005; Ressel et al., 2008; see also, Szaflarski et al., 2006). Collectively, findings suggest that right or bilateral language representation following early left hemisphere injury is facilitated by a bilateral paediatric language network that precedes adult-typical left hemisphere lateralization in the normal developmental trajectory. This explanation is supported by an extensive clinical literature indicating a decreasing potential for interhemispheric plasticity beginning around age 5 or 6 years (Rasmussen & Milner, 1977; Satz, Strauss, Wada, & Orsini, 1988; Helmstaedter, Kurthen, Linke, & Elger, 1997, Brazdil, Zakopcan, Kuba, et al., 2003; Saltzman-Benaiah, Scott, & Smith, 2003; Kadis, Kerr, Rutka, et al., 2009, Duncan, Moss, Bandy, et al., 1997; Muller, Rothermel, Behen, et al., 1998; Muller, Rothemel, Behen, et al., 1999, Springer, Binder, Hammeke, et al., 1999; Pataraia, Simos, Castillo, et al., 2004).
114 Interestingly, naming and verb generation LIERD scores were not significantly correlated, indicating a non-equivalence of these paradigms for expressive language lateralization. Similar task-related lateralization differences have been previously documented. Benson, FitzGerald, LeSueur et al. (1999) used picture naming, single word reading, and verb generation paradigms in fMRI to determine language lateralization a group of healthy adults and neurosurgical candidates considered at-risk for post-operative aphasia. In healthy subjects, Benson et al. observed poor concordance between lateralizations derived from each task. Naming and single word reading had poor test-retest reliability, while verb generation lateralizations were stable over time. Lateralizations obtained from verb generation fMRI were concordant with the IAP and stimulation mapping findings in the patient group, indicating probable task superiority for use in non-invasive language lateralizations. In the current study, we were unable to assess the reliability of naming and verb generation lateralizations over time. We are currently using MEG to study naming and verb generation in a group of children with medically intractable epilepsy, several of whom have undergone IAP and stimulation mapping to determine surgical candidacy. Comparison of MEG expressive language protocols with findings from conventional invasive assessments will allow us to determine the clinical utility of the current protocols for use in presurgical investigations. We failed to observe any age-related changes in extent of ERD for naming or verb generation. Inter-subject variability in surviving voxel counts limits power to detect effects of development on extent of expressive language cortex. Age-related changes in overall voxel count are minimal in comparison to changes in ERD lateralization over development. Future larger-scale study may provide better characterization of age related changes in
115 extent of ERD. Ideally, assessment will include counts of surviving voxels both within and outside of canonical Broca’s area, in order to comment on possible developmental changes in patterns of representation within each frontal lobe. Interestingly, laterality indices built on power (LIERD) and voxel counts (LIVOX) were perfectly correlated within each of the language paradigms. This was surprising, as power and extent measurements were included to provide distinct information about the surviving data for each subject. Across subjects, total power and total surviving voxels were significantly but imperfectly correlated. The redundancy of the lateralization indices suggests that within each subject, left versus right hemisphere differences in ERD strength are perfectly reflected in the hemispheric distribution of surviving voxels. Post-scan testing indicated that children as young as 5 years of age could accurately complete the picture naming task. Near-perfect accuracy for naming in the youngest participants guarantees performance-matching across the children and adolescents tested. The verb generation task proved to be slightly more challenging for participants, with a mean accuracy of 88% at post-scan testing in children aged 10 years and younger. Errors tended to result from omission of responses, rather than inappropriate generation of verbs. Informal testing revealed that young participants could correctly generate verbs when time limits for responding were eliminated, suggesting that a reduced rate of presentation may be appropriate for use with young children or populations with developmental delays or cognitive deficits. Interestingly, we observed reliable ERD in a majority of the participants who demonstrated imperfect accuracy for verb generation, suggesting that ceiling performance may not be
116 necessary for assessment of language representation in MEG. It is not known whether failed attempts at generating verbs is equivalent to successful verb generation in terms of neuromagnetic signal; increased effort associated with difficult items may be associated with a distinct neuromagnetic profile, which could accentuate, mask, or attenuate the target signal. In the future, researchers may circumvent questions of signal equivalence for successes and failures by pre-screening participants to establish a tailored stimulus set, or removing stimuli associated with errors prior to conducting any source analyses. The prescreening approach may be preferable, as alternate stimuli may be chosen to maintain set size and promote target signal in source analyses. Since our expressive language paradigms involve picture stimuli and simple one-word covert responding, they can be easily implemented for use with subjects speaking any language, without demand for literacy. The tasks were designed to be as easy to complete as possible, with a focus on engaging expressive language processes. We have successfully used these tasks to assess expressive language representation in clinical populations (Kadis et al., 2008), and children as young as 5 years of age. However, children younger than five years of age may have difficulty completing the tasks, particularly verb generation (as indicated by performance in our 5-10 year old subjects), and may present additional challenges for successful MEG scanning. Children, more so than adults, tend to move during the scan, often in response to stimulus presentation, introducing task-related noise. Covert responding is helpful in maintaining stillness, although we have observed silent mouthing and subtle orofacial muscle movements while scanning young children, resulting in small head movements and muscle artefact. In general, signal-to-noise is relatively low in very small children - the MEG dewar is optimized for adult-sized heads, so source-to-
117 sensor distance is increased with smaller head circumference. Biological noise is more prevalent in MEG scans of small children, due to the proximity of children’s cardiovascular and respiratory organs to the MEG dewar. Newer MEG systems that can continuously record head location will permit subjects to make small movements during the scan period, and facilitate overt responding in expressive language paradigms. We hope that advances in MEG technology will permit the future study of language representation in children younger than 5 years of age, a period characterized by massive potential for plasticity of language representation.
118
9.7 Acknowledgements This research was supported, in part, by a Studentship to DSK through the Ontario Student Opportunity Trust Fund – Hospital for Sick Children Foundation Student Scholarship Program, and a Doctoral Research Award to DSK through the Canadian Institutes of Health Research (CIHR) in partnership with Epilepsy Canada. We have no conflicts of interest to disclose.
119
Table 9.1 Participant demographic, performance, and neuromagnetic findings for naming Subject
Age
Sex
Hand
Naming Accuracy
Naming Total ERD
Naming LIERD
Naming Total voxels
Naming LIVOX
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
5.17 5.57 7.13 7.36 7.37 7.71 8.08 9.13 10.64 11.07 11.30 12.37 13.52 14.43 14.65 15.25 15.96 17.52 17.62 17.87 18.00 18.16 18.24 18.43 18.92
M M M F M F M F M M M M M F F M M M F M F M F F F
R R R R R R R R B R R R B R R R R R R R R R R R R
100.00% 100.00% 100.00% 100.00% 98.75% 100.00% N/A 100.00% 100.00% N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A
-278.44 -32.35 -60.95 -2.53 -124.36 -50.16 /NA -12.77 -797.76 -18.93 -132.36 N/A -259.20 -60.08 -221.87 -103.22 -2.01 -14.71 -2.25 -14.62 0.00 0.00 0.00 -253.40 0.00
-0.51 -0.61 -1.00 1.00 0.91 0.71 N/A -0.85 0.16 0.97 0.04 N/A 0.93 0.86 1.00 -0.43 1.00 -0.83 1.00 0.23 ---0.16 --
643 85 102 8 214 90 N/A 25 1322 41 181 N/A 484 115 421 155 4 26 4 39 0 0 0 448 0
-0.52 -0.62 -1.00 1.00 0.90 0.71 N/A -0.84 0.15 0.95 0.06 N/A 0.93 0.86 1.00 -0.38 1.00 -0.77 1.00 0.28 ---0.13 --
120
Table 9.2 Participant performance and neuromagnetic findings for verb generation Subject
Verb Accuracy
Verb Total ERD
Verb Gen LIERD
Verb L+R voxels
Verb LIVOX
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
70% 70% 100% N/A 94% 100% 75% 93% 100% N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A
-37.76 -20.08 -104.55 N/A -81.87 -12.78 -9.27 0.00 -405.61 -109.76 -29.99 -126.91 -62.19 -127.81 -546.39 -112.58 -218.09 -1.10 -38.29 -21.80 -1.36 -207.30 0.00 -431.04 -1.48
0.30 -1.00 0.86 N/A 0.61 0.59 0.61 -0.68 0.01 0.87 -0.96 0.46 0.86 0.97 -0.62 0.99 1.00 0.92 -0.11 1.00 0.92 -0.46 1.00
88 42 139 N/A 165 26 24 0 620 206 56 257 132 227 765 183 326 3 77 57 3 328 0 789 4
0.34 -1.00 0.81 N/A -0.67 0.54 0.58 -0.62 -0.08 0.82 -0.96 0.44 0.83 0.96 -0.55 0.99 1.00 0.92 -0.16 1.00 0.96 -0.46 1.00
121
9.8 Figure caption Figure 9.1. Example of test and inter-trial stimuli. Figure 9.2. Scatterplot with linear trendline (p > 0.05) for LIERD versus age at assessment for the naming task. Figure 9.3. Scatterplot with linear trendline (p < 0.05) for LIERD versus age at assessment for the verb generation task.
122 Figure 9.1
123 Figure 9.2
124 Figure 9.3
125 Chapter 10: Expressive Language MEG in Paediatric Neurosurgical Candidates 10.1 Introduction Surgical programs in epilepsy have long relied on invasive procedures such IAP and electrocortical stimulation mapping for identification of brain regions supporting gross language functions. These techniques involve the temporary disruption of normal nervous functions and concurrent behavioural assessments, an approach that permits characterization of brain-behaviour dependencies on a patient-by-patient basis. Although IAP and stimulation mapping remain the ‘gold standards’ for presurgical language lateralization and localization, respectively, many centers offering epilepsy surgery have begun to move away from invasive techniques in favor of newer noninvasive functional neuroimaging protocols (Baxendale, Thompson, & Duncan, 2008a; 2008b; Baxendale, 2009). Both fMRI and MEG have the advantage of extremely low or zero associated morbidity, relatively low cost, with the possibility of lateralizing and localizing languagerelevant cortex within a single scan session and assessment modality. With MEG in particular, there is the potential to map both epileptogenic and eloquent tissue in a single modality – comprehensive functional maps developed in a common space are particularly attractive in surgical planning. Unlike the traditional invasive procedures, neuroimaging techniques can be used in research involving healthy populations, permitting studies of normative representation and the direct comparison of control and patient findings. Due to the correlative nature of neuroimaging, new protocols must be validated against IAP and stimulation mapping to establish similarity or functional equivalence. Once validated, neuroimaging protocols have the potential to serve as surrogates for the invasive procedures
126 in clinical contexts (see Baxendale, Thompson, & Duncan, 2008b, and commentary from international survey participants in the same issue). We have previously documented our experience with MEG expressive language mapping in healthy adolescents (n = 12) and a small patient group (n = 4) (Kadis, et al., 2008). In controls and patients, we observed consistent low-beta event-related desynchrony (ERD) typically focused over the left inferior frontal gyrus, canonical Broca’s area. In the current study, we assess ERD lateralization for naming and verb generation in a larger and relatively varied patient group (n = 14). Individual patient data are assessed using the novel analytic protocol involving bootstrap statistical thresholding for differential beamformer analyses, as described in Chapter 9. Findings are contrasted along several dimensions known to predict language lateralization, and compared against IAP, and fMRI, and other MEG measures of language lateralization, where available. 10.2 Method 10.2.1 Participants Fourteen patients (aged 9.0 to 15.8 years; 6 male) with significant neurological disorders participated in this study. At the time of our investigations, 13 subjects were being considered for neurosurgical treatment of medically intractable epilepsy; one subject (Subject #4) had a history of intraventricular hemorrhage secondary to an arteriovenous malformation (AVM), which warranted neurosurgical consideration. All subjects were referred for expressive language lateralization using MEG to determine whether they would be at risk for post-operative aphasia.
127 In all cases, parents consented to their children’s participation, and the children provided assent to participate. Patients received a small gift for their participation. The study was approved by the research ethics board at the Hospital for Sick Children (Toronto, ON, Canada). Patients had previously undergone a subset of the following clinical investigations: EEG, prolonged video-EEG, inter-ictal MEG, CT, MRI, PET, and neuropsychological evaluation. Those who had previous surgeries may have received chronic electrode grid implants for electrocorticography and electrocortical stimulation mapping. Two patients underwent IAP for language lateralization (we have previously documented the IAP protocol used at the Hospital for Sick Children; Fernandes & Smith, 2000; Saltzman-Benaiah et al., 2003, Kadis et al., 2009). Eight patients underwent fMRI for language lateralization and localization (using a battery of expressive and receptive tasks), and two subjects had MEG for receptive language lateralization (using an auditory word recognition task, as in Simos, Breier, Zouridakis, & Papanicolaou, 1998; see also, Chuang et al., 2006). Demographic and medical data are presented in Table 10.1. Brief case histories and summaries from our expressive language MEG mapping for all patients are presented following the group summaries of findings. Subjects #1, #2, #3, and #4 correspond to the patients presented in Kadis et al. (2008; see Chapter 8). At that time, we described ERD patterns for each patient completing the verb generation task. Here, we employ objective measures to quantify lateralization of frontal lobe ERD for both naming and verb generation in a significantly larger patient group.
128 10.2.2 Expressive language tasks The MEG stimulus presentation and acquisition procedures have been documented previously (Kadis et al., 2008; see Chapters 8 and 9), and are described only briefly, below. Subjects were trained on the naming and verb generation tasks prior to MEG scanning. Subjects viewed pictures of everyday objects (500 ms) and were asked to quickly and covertly generate a name for the object (naming task, 80 objects) or an action word associated with the object (verb generation task, 80 objects). Between test stimuli, subjects viewed scrambled colour images (1500 ms), which served as baselines in differential analyses. Subjects also button-clicked when viewing a picture of a hand clicking a computer mouse, appearing in place of test stimuli at a 15% probability; this trial was used to promote and monitor vigilance in participants. Recording took place in a magnetically shielded room, using a 151-channel whole-head MEG system (VSM Medtech, Coquitlam, BC, Canada). For a majority of participants, we assessed accuracy of naming and verb generation to confrontation after scanning, regardless of patient age. In some cases, patient fatigue and/or time constraints associated with other clinical investigations precluded formal assessment of accuracy – in these instances, we confirmed participant ability from performance during training. Structural MRIs were obtained as part of the each patient’s neurological work-up, typically at 1.5T, following the MEG scans. Fiducial markers placed at the nasion and pre- auricular points facilitated coregistration of the MEG and MRI scans.
129 10.2.3 Analyses Neuromagnetic data were analyzed using our novel tailored yet objective protocol, based on sliding differential windows and thresholds derived from bootstrap statistics (see Chapter 9). In brief, we computed differential beamformer analyses for 13-23 Hz activity during four overlapping active windows for each subject: 300-500 ms, 400-600 ms, 500-700 ms, and 600-800 ms following the onset of test stimulus presentation. Each active window was contrasted against a single baseline window consisting of the 200 ms period immediately preceding test stimulus presentation. Differential changes were analyzed using a bootstrap statistical procedure, whereby observed data are randomly sampled with replacement to establish possible alternate data sets (pseudo runs), collectively providing distributions of voxel-wise neuromagnetic changes. In the current implementation, we established 99 pseudo runs of 80 trials each (for each naming and verb generation study). The observed ERD per voxel was then assessed across runs (actual and pseudo) using non-parametric inferential statistics at an a priori chosen alpha of p < 0.01, uncorrected. To facilitate quantification of frontal lobe involvement in naming and verb generation, and to control for distal posterior signals associated with object viewing, we restricted analyses to a probabilistic volume of the human frontal lobe. To quantify the relative power of ERD within the left versus right frontal lobe for naming and verb generation, we computed laterality indices (LI) for surviving voxels at each differential window and across the four active-baseline contrasts (summary LI). For each LI, left (ERDL) versus right (ERDR) hemisphere desynchrony was compared, as follows: LI = (ERDL – ERDR) ÷ (ERDL + ERDR)
130 LI scores range in value from +1 (completely left) to -1 (completely right). Scores around 0 indicate bilateral contributions. For this study, we have adopted the following convention for LI classification: left (typical) lateralization for scores ranging from 0.20 to 1.00, bilateral (atypical) from -0.20 +0.20, and right (atypical) lateralization from -1.00 to -0.20 (as in Gaillard, Balsamo, Xu, et al., 2002). Some patients demonstrated complex patterns of frontal lobe ERD across differential windows. In several cases, we observed shifts in lateralization (left, bilateral, right) across differential windows for a single language task. Plots of ERD lateralization at each differential window are presented following each case description. Summary LIs for each patient’s naming and verb generation study are presented in Table 10.1. We assessed the relationship between summary LI scores and several demographic and medical variables known to predict language lateralization, including seizure side and age at seizure onset for subjects with epilepsy, and handedness. Expressive language MEG findings are compared with IAP and fMRI studies for language, MEG studies of receptive language, where available. 10.3 Results Below, we present group findings in relation to a number of variables known to predict language lateralization. The heterogeneity of the patient group (seizure type, aetiologies, abilities, previous investigations, etc.), along with the sometimes complex patterns of ERD lateralization observed for naming and verb generation, requires that we also consider findings on a patient-by-patient basis. Following the group summaries, we present individual case histories along with corresponding naming and verb generation study findings.
131 10.3.1 Naming and verb generation within the patient group All subjects completed the naming task. We failed to observe voxels surviving the bootstrap thresholding procedure for one subject (Subject #3). Among the 13 subjects with reliable ERD for naming, LI values ranged from -1.00 to +1.00, with a mean of 0.38. LIs for naming indicated left dominance in 8 subjects, bilateral representation in 2 subjects, and right dominance in 3 subjects. All subjects attempted the verb generation task. The paradigm was deemed too difficult for two subjects; Subject #5 correctly generated verbs for only 25% of stimuli, and Subject #11 failed to produce verbs, instead naming items or not responding, during the training session. In one other subject (Subject #6) we failed to observe voxels surviving the bootstrap procedure. In the remaining subjects, LI values ranged from -0.29 to +1.00, with a mean of 0.50. LIs for verb generation indicated left hemisphere dominance in 8 subjects, bilateral representation in 1 subject, and right hemisphere dominance in 2 subjects. 10.3.2 Concordance of naming and verb generation Naming and verb generation LI scores were not significantly correlated, r = 0.26, n = 10, p > 0.05. Similarly, dichotomous classifications of typical versus atypical lateralization were not significantly correlated across expressive language tasks, r = 0.22, n = 10, p > 0.05. 10.3.3 Effect of seizure side on lateralization In the 13 subjects with epilepsy, the epileptogenic hemisphere was determined by scalp EEG, prolonged video-EEG, and/or interictal MEG. Seizure side (left, bilateral, right) did not predict lateralization (typical versus atypical) for naming, χ2 = 0.21, df = 2, n = 12, exact p >
132 0.05. Left hemisphere naming dominance was observed in 3 of 5 subjects with left hemisphere epilepsy, 2 of 4 subjects with bilateral epilepsy, and 2 of 3 subjects with right hemisphere epilepsy. The single subject with the left parasaggital AVM (Subject #4) had left lateralized naming. Seizure side did not predict verb generation lateralization, χ2 = 2.86, df = 2, n = 10, exact p > 0.05. Left hemisphere dominance for verb generation was observed in all 4 subjects with left hemisphere epilepsy, 2 of 4 subjects with bilateral epilepsy, and 1 of 2 subjects with right hemisphere seizures. The subject with left AVM (Subject #4) had left lateralized verb generation. 10.3.4 Age at seizure onset and lateralization The effects of early versus late onset left hemisphere epilepsy on language lateralization are well documented. Individuals with first seizures before 5 or 6 years of age are more likely to demonstrate atypical lateralization than those with later onset (e.g., Rasmussen & Milner, 1977). Of the 13 subjects with epilepsy, 7 had seizure onset before age 6 years. Only 4 of the 13 subjects had left hemisphere epilepsy, 2 of which experienced seizure onset before age 6 years. Early versus late seizure onset did not predict typicality of naming lateralization, χ2 = 0.3, df = 1, n = 12, exact p > 0.05. Among those with left hemisphere seizures, 1 of 2 with early onset, and 1 of 2 with late onset had typical naming lateralization.
133 Lateralization for verb generation did not differ for subjects with early versus late seizure onset, χ2 = 0.5, df = 1, n = 10, exact p > 0.05. All 4 subjects with left hemisphere epilepsy (2 with early onset, 2 with late onset) had typical lateralization for verb generation. 10.3.5 Effect of handedness Right or bilateral language lateralization is more common in healthy bimanual and left handed subjects (Knecht et al., 2000). In patient samples, similar increases in atypical lateralization associated with sinistrality may reflect a common response to left hemisphere pathology, although it remains difficult to parcel causation (Woods et al., 1988). We were unable to formally assess lateralization in relation to handedness. In the current sample, 13 of 14 subjects were right handed. The single left handed subject (Subject #11, with bilateral infarcts and left lateralized frontal-temporal-parietal epilepsy) had a moderate hemi-paresis secondary to multiple perinatal infarcts; this subject demonstrated left lateralization for naming, but was unable to complete the verb generation task. 10.3.6 Concordance with IAP Three subjects (Subjects #1, #10, and #11) had undergone IAP for language lateralization. Subject #1 had undergone a total of three IAP studies, the first two were inconclusive and the third suggested bilateral representation. ERD for both MEG naming and MEG verb generation was left lateralized. The presence of language positive tissue in the left hemisphere was confirmed by electrocortical stimulation mapping, completed prior to resective surgery (see Kadis et al., 2008).
134 Subject #10 had right hemisphere lateralization on IAP. Our MEG naming study suggested right lateralization, although the MEG verb generation study suggested left lateralization. Subject #11 underwent IAP that indicated bilateral representation of language, with relatively greater right hemisphere involvement. The IAP study was complicated due to obtundation with injections to both hemispheres. Our naming MEG suggested left lateralization, with accuracy of 80%. The subject was unable to complete the MEG verb generation task. 10.3.7 Concordance with fMRI Eight subjects had fMRI for language lateralization. The fMRI studies involved a battery of expressive (e.g., picture naming, word fluency) and receptive language tasks (e.g., listening to stories) of varying difficulty. Findings were analyzed through visual inspection of whole head activations, which are demonstrably equivalent to statistical thresholding techniques (e.g., Gaillard, Balsamo, Grandin, et al., 2002). Our MEG findings were frequently discordant with the fMRI studies. Four subjects (Subjects #5, #8, #12, #14) had left lateralized language in fMRI. Of these, two had typical lateralization with MEG naming. Three of the four had reliable ERD for verb generation. Of these three, two had typical lateralization for MEG verb generation. Three subjects (Subjects #1, 7, 10) had bilateral language representation in fMRI. Of these, 2 had typical lateralization for MEG naming, 1 had atypical (right) lateralization for naming. Similarly, 2 had typical lateralization and 1 had atypical (right) lateralization for verb generation.
135 The single subject (Subject #11) with right hemisphere lateralization in fMRI had typical lateralization for naming in MEG. 10.3.8 Concordance with receptive language MEG Two subjects had previous receptive language MEG studies. Both were reported to have atypical receptive language lateralization. Subject #7 had bilateral receptive language lateralization. We observed left lateralization for MEG naming and right lateralization for MEG verb generation for this subject. Subject #12 had right hemisphere receptive language lateralization. We observed left hemisphere lateralization for both naming and verb generation in MEG. 10.3.9 Individual case studies In the figures that follow each patient history, we present ERD lateralization (LI) and relative power (as a percent of the maximal power window) values for each differential contrast corresponding to the patient’s naming and verb generation study. LI scores range in value from +1 (completely left) to -1 (completely right), and are represented as dots in the figures. Scores around 0 indicate bilateral contributions; the dotted lines surrounding the abscissa indicate a conventional value used to identify bilateral language representation. Relative power at each differential window is represented by the blue columns in the figures. Subject #1. Subject is a 12.8 year old right-handed (by parent report and neurologist observation) female with seizures beginning at 13 months, secondary to encephalitis. MRI showed extensive left hemisphere abnormality. PET revealed hypermetabolism in the left
136 temporoparietal region. EEG and MEG revealed scattered perisylvian interictal spike discharge. At age 6 years, the subject underwent fMRI and IAP for language mapping. Both studies were inconclusive. At age 7 years, she received an extensive left hemisphere subdural electrode grid implant, with mesial temporal depth electrodes, prior to resective surgery. Electrocorticography revealed ictal involvement of frontal, temporal, and parietal cortices, and mesial temporal discharge. Stimulation mapping revealed two languagepositive sites: one in the inferior frontal lobe (pars triangularis), and another over the middle temporal gyrus. She received a temporal lobectomy with cortical excision of the posterior superior temporal gyrus and multiple subpial transections over the inferior frontal and inferior parietal regions. Seizures continued following surgery. Repeat MEG revealed residual discharge along the spared regions of the left superior temporal gyrus. The subject underwent repeat IAP at age 11, which was inconclusive for language lateralization, and again at age 12, which indicated possible bilateral speech. A repeat fMRI suggested bilateral representation, also. MEG indicated left hemisphere representation of receptive language. The subject experiences some general cognitive delays, with relative weakness in language processing, verbal memory, and executive functioning, and is in a modified school program. She participated in both the naming and verb generation MEG tasks. During training, she demonstrated no difficulty with either task. Performance was not formally assessed following the scan.
137
We observed bilateral ERD surviving the bootstrap statistical procedure, in both paradigms. The differential windows with maximal ERD power indicated left frontal lateralization for both naming and verb generation, consistent with the observation of a left inferior frontal language site at stimulation mapping. Although the early differential window for both naming and verb generation indicates rightward asymmetry of surviving ERD, the total power of beta ERD in this first window was relatively low, suggesting predominant left lateralization.
138 Subject #2. The subject is a 10 year old, right-handed (parent report) male with a history of complex partial seizures since 8 years of age. His seizures are characterized by staring and smiling, lasting approximately 20 seconds. The seizures may also involve right arm flexion, with or without vocalizations, but never with speech. In the immediate post-ictal period, speech is absent for up to 1 minute. Seizures occur in clusters of several per day, with approximately one cluster per week. MRIs established at age 8 and 10 years were reported as normal. At age 8 years, EEG revealed epileptiform discharges over the left frontal region. Prolonged video-EEG revealed interictal spiking occurred over the left frontal-central region, with probable ictogenesis near the midline. The subject had normal developmental milestones, and does well in school. His parents report that he is slightly uncoordinated, but has intact fine motor skills. The subject completed both the naming and verb generation to confrontation. Accuracy was not assessed, but the subject reported no difficulty with either paradigm. We observed ERD surviving the bootstrap statistical procedure for both naming and verb generation.
139
The LIs indicate left-hemisphere expressive language lateralization. Electrographic studies and seizure semiology (e.g., right arm flexion) suggest focal epilepsy, likely involving the left frontal lobe. Ictal involvement of expressive language cortical areas is suggested by the subject’s ictal and post-ictal speech arrest. Subject #3. The subject is an 11.9 year old right-handed (EHI = 100) female with a history of seizures involving laughter, screaming, and bilateral arm extension, since 3 years of age. CT and MRI were reported as normal. PET showed bilateral frontal hypermetabolism and left anterior temporal hypometabolism. Prolonged video-EEG showed right frontal slowing, without spike activity. MEG showed diffuse bilateral frontal spikes, with greater involvement of the left hemisphere. The subject had normal early developmental milestones, and is doing well in mainstream school. She participated in the verb generation MEG paradigm. Accuracy was not formally assessed.
140
We observed only one differential window with surviving ERD. Findings indicate left frontal lateralization for verb generation. Subject #4. The subject is an 11.8 year old right-handed (EHI = 100%) female with a recent history of intraventricular hemorrhage secondary to an arteriovenous malformation (AVM). The event resulted in a coma, requiring intubation. Cerebral angiogram revealed an extensive parasaggital (primarily anterior cerebral artery) AVM, with cystic encephalomalacia of the left parietal lobe. She has a documented history of expressive and receptive language difficulties. She participated in both naming and verb generation to confrontation. Post-scan assessment revealed 100% naming accuracy, 90% verb generation accuracy.
141
The maximal and comparable ERD differential windows showed predominantly left hemisphere involvement for naming and verb generation, suggesting typical lateralization of expressive language. Subject # 5. The subject is a 9.0 year old right-handed (EHI = 78.9) male with a history of simple partial seizures since 5.8 years of age. Seizures begin with a sensation of sharp pain in the left eye, and may progress to involve left arm and leg stiffening. MRIs conducted at ages 6, 7, and 8 were reported as normal. The subject has undergone prolonged video-EEG monitoring three times. Clinical seizures were observed during the first two admissions;
142 however, muscle artifact associated with seizures prevented localization of seizure onset. At the third admission, seizure onset was localized to the right fronto-central region. Interictal epileptiform discharges were observed over the same region. In contrast, interictal MEG conducted with the subject under general anesthetic revealed scatter spike sources over the left hemisphere. He experienced normal developmental milestones, and is currently one year behind his cohort in a mainstream school. He has relative strengths in nonverbal abilities, relative weakness in language skills, and decreased right hand fine motor speed. An fMRI for language indicated left hemisphere dominance. The subject participated in both of our MEG expressive language tasks. Post-scan testing showed 100% accuracy on naming, but only 25% accuracy for verb generation. The subject indicated experiencing considerable difficulty with the verb generation task. We report only on naming, below. The image part with relationship ID rId28 was not found in the file.
We observed ERD surviving the bootstrap procedure for only one differential window. The LI indicated left hemisphere representation of expressive language, consistent with fMRI findings.
143 Subject #6. The subject is a 9.3 year old right-handed (EHI) score = 75) female with panhypopituitarism. She experienced her first seizure at 3 days, due to hypoglycemia, and began hormone replacement therapy shortly after. The subject remained seizure free until age 4 years, when she experienced a single generalized tonic-clonic seizure. At age 6 years, she was diagnosed with dilated cardiomyopathy. She continues to experience seizures that are characterized by behavioural arrest, staring, and mouth automatisms, lasting up to 2 minutes. She also experiences occasional drop attacks. Seizures occur 4-5 times per week. Prolonged video-EEG revealed bilateral interictal perisylvian spikes, with predominant ictogenesis localized to the left temporal lobe. Interictal MEG revealed bilateral rolandic and temporal spiking. MRI was significant, indicating the empty pituitary sella, as well as increased abnormal signal in the left temporal pole with some blurring of the grey-white matter border of cortex near the amygdala. She is delayed in most areas of cognition, with pronounced verbal skills deficits. The subject completed the naming task with some degree of difficulty, but was unable to complete verb generation. Post-scan overt testing revealed 82.5% naming accuracy.
144 We observed ERD surviving the bootstrap procedure at all differential time windows for naming to confrontation. The subject’s maximal ERD window reveals mostly left-lateralized frontal lobe representation for naming. A second similarly-powered differential window supports left hemisphere lateralization. Subject #7. Subject is a 10.2 year old right-handed (EHI = 90) female with a history of complex partial seizures since age 5 years. MRI showed cortical malformation in the right posterior temporal and occipital regions, multiple right periventricular heterotopias, and cerebellar dysgenesis. EEG showed epileptiform discharges over the posterior temporal, parietal, and occipital regions of the right hemisphere. She had normal developmental milestones, receives average grades in school, but requires additional help for math and language courses. An fMRI for language showed bilateral activations, with atypical right occipital activations for all language paradigms. A receptive language MEG showed bilateral representation, with possible right-sided predominance. The subject participated in both our naming and verb generation tasks. At post-scan assessment, she named 90% of objects, correctly. Accuracy for verb generation was only 70%.
145
The subject had ERD surviving the bootstrap in both hemispheres. For naming, all LIs indicated left hemisphere lateralization. In contrast, the two peak LIs for verb generation were in opposite directions, with the maximal ERD differential window showing right lateralization. Collectively, findings suggest bilateral frontal representation of expressive language, consistent with fMRI and receptive language MEG studies. We are somewhat cautious, however, of the verb generation lateralization findings, due to the relatively low accuracy observed at post-scan assessment. Subject #8. The subject is a 12.1 year old right-handed (EHI = 100; with left hand hemiparesis) female with a history of seizures since 2 years of age. First seizures were generalized tonic-clonic. At around age 4 or 5 years, the seizure-type became more lateralized, and currently involves initial stiffening and occasional jerking of the left-arm and leg. During her seizures, she is unable to speak, but retains awareness. She has a small left hand and stiffness in her left arm. MRIs done when the subject was 5 and 8 years of age were reported as normal. An MRI done at 11 years of age was reported as normal, with a possible right frontal abnormality, consistent with a focal cortical dysplasia in the supplementary motor area. Prolonged video-EEG showed right fronto-central initiation of
146 brief tonic seizures. MEG revealed primary interictal discharges at the right supplementarymotor area, with atypical discharges in the left and right premotor regions. The subject showed left hemisphere lateralization for language in fMRI. Early milestones were normal. She is currently an average student in a mainstream school. Formal neuropsychological evaluation showed a somewhat below average cognitive profile, with specific deficits on visual-spatial processing and weakness and reduced fine motor skills of the left-hand. The subject participated in both of the expressive language MEG paradigms. Post-scan testing revealed 100% naming and 90% verb generation accuracy.
147 The subject showed a complex pattern of lateralization on the expressive language MEG tasks. For naming, both leftward and rightward LIs were observed, with the maximal ERD differential window indicating left hemisphere lateralization. For verb generation, we again observed a mix of leftward and rightward LIs, with the maximal ERD window indicating right hemisphere lateralization. Collectively, findings suggest bilateral representation of expressive language. Subject #9. The subject is a 12.4 year old right-handed (EHI = 100) female with a history of daily seizures since 3 years of age. Her seizures are characterized by an initial drop, followed by shaking of the right leg and then right arm. She has variable loss of speech during seizures. MRI revealed cortical thickening in the left peri-rolandic area, suggesting cortical dysplasia or a low grade glioma. Prolonged video-EEG showed left central, parietal, and midline interictal epileptiform discharge; the ictal onset was not lateralized. Interictal MEG showed a cluster of spike sources over the left motor and sensory cortex. She experienced normal developmental milestones. She is an average student, with some difficulties in math. She has no motor deficits outside of her seizures. The subject participated in both of the expressive language MEG tasks. Post-scan assessment revealed 100% accuracy for both naming and verb generation.
148
The subject demonstrated a complex pattern of lateralization across differential windows for both expressive language tasks. For naming, the LI for the maximal ERD window showed right hemisphere lateralization, although a window with comparable ERD showed left lateralization. Similarly, for verb generation, we observed both left and right lateralized LI scores, with the maximal ERD window indicating left lateralization. Findings suggest that both hemispheres support expressive language. Subject #10. The subject is a 14.5 year old right-handed (EHI=87.5) male with a history of seizures since 8 years of age. Typical seizures involve staring and motor arrest. Multiple
149 MRIs showed left mesial temporal abnormalities involving the hippocampus and parahippocampal gyrus. Abnormalities were consistent with DNET or low grade glioma. Prolonged video-EEG revealed generalized abnormalities, with left-sided predominance. Interictal MEG showed polyspikes over the left inferior frontal, superior and middle temporal, and occipital regions. The subject participated in fMRI and IAP for language. Findings on the fMRI battery were not clear; one task suggested exclusive right hemisphere language representation. The IAP suggested right hemisphere speech dominance. The subject has a history of learning disability, with relatively strong visuo-spatial reasoning and weak language skills and executive functions. At age 12.5, he received a left temporal resection, which revealed oligodendrocytosis in the hippocampal region. Seizures recurred at 10 days following surgery. The subject participated in both of our expressive language MEG tasks. He was fidgety for the naming task which was completed first, but compliant for verb generation. Post-scan assessment revealed 100% accuracy in both tasks.
150
We observed significant ERD in both hemispheres for naming and verb generation. LIs shifted from left to right over the analysis windows for naming to confrontation. The maximal ERD window for naming suggested exclusive right hemisphere representation. In contrast, the LIs for verb generation indicate leftward, although not exclusive left hemisphere, dominance. Collectively, the scans suggest a complex lateralization profile, involving bilateral frontal representation of expressive language. Subject #11. The subject is a 14.6 year old left-handed (EHI = -100; subject has right hemiparesis) male with seizures since 6 years of age. During seizures he is unresponsive and falls to the ground. Postictally, he is disoriented and agitated. The subject has a healthy twin sister. He was second to arrive, required resuscitation at birth, and suffered bilateral infarcts involving the left middle cerebral artery territory and the right parieto-occipital region. MRI shows the bilateral infarcts, as well as hypoplasia of the left hemisphere. He has a 2 cm arachnoid cyst in the left temporal lobe. Prolonged video-EEG revealed a left temporal ictal onset. Interictal MEG showed three clusters of spike sources in the frontal, parietal, and occipital lobes, surrounding the temporal lobe cyst. An fMRI for language suggested right hemisphere lateralization. The subject experienced a complicated IAP, with some speech
151 disturbance following both left and right injections; major speech representation is in the right hemisphere. Compared to his twin sister, he was significantly delayed with achievement of early milestones. He has full scale IQ in the extremely low range, with both verbal and non-verbal deficits. The subject completed the picture naming task, but was unable to complete verb generation. Post-scan assessment revealed 80% accuracy on naming.
We observed ERD surviving the bootstrap in both hemispheres. LI scores ranged from left lateralized to bilateral. The maximal ERD differential window showed naming lateralized to the left hemisphere. Subject #12. The subject is a 14.8 year old right-handed (EHI = 100) female with a 5 month history of seizures. Her first seizures were complex-partial, lasting up to 10 minutes, with left hand and leg twitching. Her current seizures involve staring and bowel incontinence. MRI indicated a right temporal lobe low-grade glioma. Prolonged video-EEG monitoring showed seizures originating in the right temporal lobe, with bilateral interictal discharges. Interictal MEG showed right temporal discharges at the border of the tumour, and bilateral
152 Rolandic spiking. The bilateral Rolandic discharges did not have clinical correlates, and were determined to likely reflect irritation rather than epileptogenesis. An fMRI for language revealed left hemisphere lateralization. Receptive language MEG was right lateralized. The subject had normal developmental milestones. Formal neuropsychological evaluation revealed perceptual reasoning and visual memory in the high-average range. Verbal skills were average. Attention and processing speed were compromised. The subject participated in our naming and verb generation MEG paradigms. At post-scan testing, she correctly named 100%, and correctly generated verbs to 91.25% of objects.
153 We observed ERD surviving the bootstrap in both hemispheres for naming, and only the left hemisphere for verb generation. The LI for the maximal ERD differential window for naming indicated left lateralization. All LIs for verb generation indicated left lateralization. Collectively, findings indicate left frontal lateralization for expressive language. Subject #13. The subject is a 15.7 year old right-handed (EHI = 89) male, with a history of generalized tonic-clonic seizures since age 14 years. Seizures occur approximately once per month. Initial CT and structural MRI were both reported as normal. Repeat MRI revealed a mass along the left facial nerve, likely representing either an incidental neurofibroma or fibroma. Prolonged video-EEG and interictal MEG revealed bilateral frontal epileptiform discharges. The subject experienced normal early developmental milestones. Parents report that he has experienced memory problems since infancy, and is disorganized. The subject participated in both the naming and verb generation MEG tasks. At post-scan assessment he correctly named and generated verbs to all items (100% accuracy, both tasks).
154
We observed significant ERD in both hemispheres, for both naming and verb generation. LIs for naming were observed around the bilateral range. The LI score for the maximal ERD window in verb generation was near the bilateral range, with greater right than left hemisphere involvement. Other differential windows with similar ERD power showed mixed LIs. Collectively, findings suggest bilateral representation of expressive language. Subject #14. The subject is a 15.8 year old right-handed (EHI = 97) male with a history of seizures since 12 years of age. The subject experiences multiple seizure types. Typical seizures begin with staring, spinning, and falling, and then progress into generalized tonicclonic seizures lasting less than 1 minute, with 30 minute post-ictal fatigue. He also experiences bilateral arm flexion with loss of speech, and reports ‘déjà vu’ experiences for up to 5 minutes, daily. Multiple MRIs have been reported as normal, or normal with possible right mesial temporal sclerosis. A subtle increased FLAIR signal was observed at the most posterior aspect of the right hippocampus. Prolonged video-EEG suggested right frontotemporal ictal onset, with frequent interictal temporal and frontal discharge. Interictal MEG showed a cluster of spikes in the right fronto-temporal region. The subject underwent invasive mapping with an electrode grid and strips, which revealed right mesial temporal
155 onset with rapid secondary generalization. His early milestones were normal. He demonstrated some decline in school performance since onset of seizures. At neuropsychological assessment, he demonstrated compromise of verbal and nonverbal functioning. An fMRI for language indicated left hemisphere dominance. The subject participated in both of the expressive language MEG tasks. Accuracy was not formally assessed, but he demonstrated competence at training, and reported no difficulty with naming or verb generation following scanning.
156 The subject demonstrated exclusively right lateralized ERD for naming to confrontation. For verb generation, the subject demonstrated a mix of LI scores, with the maximal ERD window showing bilateral representation. Findings suggest bilateral language representation of expressive language. 10.4 Discussion In this study, we used two developmentally appropriate MEG tasks to study expressive language lateralization in a small group of subjects with neurological abnormalities. Neuromagnetic data were analyzed using a novel thresholding procedure involving bootstrap statistics for differential beamformer analyses. The tasks and MEG scanning were welltolerated by subjects, and the analyses permitted quantification of hemispheric involvement within and across paradigms, yielding extensive lateralization data on a patient-by-patient basis. At the group level, we failed to observe differences in MEG naming or verb generation lateralization as effects of seizure side, age at seizure onset, or handedness. The lack of significant differences among subgroups may reflect the small, heterogeneous sample in the current study. However, some observations ran counter to effects described in the literature, and deserve mention. We observed typical left hemisphere lateralization for verb generation in all 4 subjects with left hemisphere seizures, and only 1 of 3 subjects with right hemisphere seizures, counter to our expectations. The trend toward increased atypical lateralization for verb generation with right hemisphere insult persisted even after accounting for age at seizure onset. It is important to note that side of seizure focus may not accurately reflect the extent or lateralization of an insult; summary findings can be misleading in small case series
157 analyses. For example, Subject #11 had extensive bilateral insults (perinatal middle cerebral artery infarcts), with exclusive left hemisphere epileptogenesis. Although left hemisphere epilepsy is associated with increased atypical lateralization in the literature, the implicit assumption is that the right hemisphere language homologues remain intact, which is not always the case. We observed moderate concordance between expressive language MEG and IAP findings, and relatively poor concordance between the MEG findings and fMRI language and MEG receptive language lateralizations. The two expressive language studies tended to indicate left lateralization in subjects with multimodal assessments, whereas atypical lateralization was more common in IAP, fMRI, and receptive language MEG studies. The discord may reflect, in part, the different language tasks used and the associated relative engagement of brain regions supporting receptive language. In paediatric patients, there is some evidence for increased interhemispheric plasticity for receptive compared to expressive language (Boatman, Freeman, Vining, et al., 1999). In both IAP and fMRI, a combination of expressive and receptive language tasks is used to determine lateralization, and the receptive language MEG protocol is designed to identify relative involvement of the left versus right posterior superior temporal lobe, thus biasing these modalities toward right hemisphere involvement in the current paediatric case series. Our MEG expressive language protocols, by design, provide lateralization information for anterior language cortex only. As in our previous MEG study of expressive language lateralization in healthy children and adolescents (Chapter 9), laterality indices for naming and verb generation were not significantly correlated. Others have documented similar task-related lateralization differences. Benson et al. (1999) observed poor concordance for picture naming, word
158 reading, and verb generation lateralizations in healthy subjects studied in fMRI. Lateralizations derived from verb generation were most reliable at test-retest. In adult neurosurgical candidates, Benson et al. found that verb generation lateralizations were highly concordant with IAP findings. Findings suggest that verb generation may be superior to naming for assessment of language lateralization in fMRI, although the relative validity has not yet been determined for the MEG naming and verb generation tasks used in the current study. However, in one case (Subject #10), we know that lateralization for the MEG naming task, but not the MEG verb generation task, was concordant with lateralization determined through IAP. In the future, we hope to assess validity of the two expressive language MEG paradigms by comparing findings against the ‘gold standard’ IAP for lateralization. Ideally, subjects would have clearly lateralized epilepsy with well-defined unilateral insults. At IAP testing, special attention to types of errors made may differentiate expressive versus receptive language lateralization. Only if agreement between modalities is high on a subject-by-subject basis should we consider naming and verb generation in MEG as appropriate surrogates for the invasive procedures. The individual case descriptions and lateralization findings highlight the important potential clinical application of a valid expressive language mapping protocols for MEG. To be useful in a paediatric presurgical context, paradigms must involve developmentally appropriate stimuli and responses, findings must be reliable and quantifiable, and the protocol must have power at the single subject level. In the current study, the vast majority of patients were able to complete both naming and verb generation tasks and had frontal lobe ERD surviving the bootstrap statistical thresholding procedure. Thus, the two expressive language MEG
159 protocols show promise for future clinical implementation, although validation is still required. The most conspicuous finding in individual analyses was the occasional fluctuating ERD lateralization observed across differential contrast windows, within single naming or verb generation studies. Summary LIs were typically concordant with ERD lateralization in the predominant window; however, occasional discordant lateralization of subordinate windows suggested complex patterns of hemispheric involvement across the response/analysis period. Choosing a single active time window for all differential analyses would have lead to incorrect categorizations of frontal lobe language lateralization, and should be avoided in future studies. Shifts in ERD lateralization across windows necessitate a tailored approach involving multiple differential contrasts for unbiased characterization of the primary neuromagnetic response. In the current study, we were unable to assess the role of naming and verb generation accuracy on lateralization findings. Most subjects were able to complete the tasks, although some reported that verb generation was relatively challenging. Subsets scored below perfect on naming and verb generation and two subjects found verb generation too difficult to complete. Given the small sample, we were unable to formally assess the relationship between performance and lateralization findings, or between performance and surviving ERD signal. Future large-scale studies will permit the assessment of the role of response accuracy.
160 Collectively, findings indicate that the two expressive language MEG paradigms and novel analyses are promising for use with children with epilepsy/neurological disorders, although further clinical study is necessary prior to implementation in presurgical workups.
161 Table 10.1. Demographic and seizure related data, with performance and lateralization findings for all patients in the study.
Subject
Sex
Hand
Age
Seizure Onset
Seizure Side
Seizure Site
Naming Accuracy
Naming LI
Verb Gen Accuracy
01 02 03 04 05 06 07 08 09 10 11 12 13 14
F M F F M F F F F M M F M M
R R R R R R R R R R L R R R
12.8 10.6 11.9 11.8 9.0 9.3 10.2 12.1 12.4 14.5 14.6 14.8 15.7 15.8
1 8 3 * 5 4 5 2 3 8 6 14 14 12
L L B * B B R R L L L L B R
FTP F FT * F FTP TO F FP T FTP T F FT
nfa nfa n/a 100 100 83 90 100
0.92 0.95 n/a 0.94 1.00 0.48 0.89 0.10 -0.21 -0.54 0.52 0.85 0.09 -1.00
nfa nfa nfa 90 25 nfa 70 90 100 100 n/a 91 100 nfa
100 100 80 100 100 nfa
Verb Gen LI 0.84 1.00 1.00 0.75 ---0.25 -0.07 0.42 0.63 n/a 1.00 -0.29 0.44
Hand = handedness, determined by Edinburgh Handedness Inventory score for subjects #3-14; Sex, M = male, F = female; Age = age at assessment, in years; Seizure Onset = age at seizure onset, in years; Seizure Side, L = left hemisphere epileptogenesis, B = bilateral independent discharge, R = right hemisphere epileptogenesis; Seizure Site, F = frontal lobe, T = temporal lobe, P = parietal lobe, O = occipital lobe; Naming Accuracy is percent of items correctly named at post-scan assessment, nfa = not formally assessed; Naming LI = summary lateralization index for naming; Verb Gen Accuracy is percent of items subject correctly generated verbs to at post-scan assessment, nfa = not formally assessed; Verb Gen LI = summary lateralization index for verb generation, -- indicates no beta ERD surviving the bootstrap thresholding procedure * Subject #4 has a left hemisphere fronto-parietal AVM, and does not have a seizure disorder
162 Chapter 11: General Summary 11.1 Review objectives The primary objectives for this thesis were to improve our ability to predict and assess language representation in healthy children and individuals with neurological disorders arising in childhood. Across cultures, language functions are acquired from infancy through adolescence in a highly regular manner; these functions are ultimately represented in a conventional adult language network in the brain. In studying healthy children, we hoped to advance our understanding of how the canonical language network is established during normal development. We also studied individuals with early neurological insult, to characterize plasticity and determine predictors of language representation in the context of injury. In particular, we studied children with intractable seizure disorders who are candidates for neurosurgical treatments. For these children, accurate assessment of epileptogenic and eloquent cortex is essential for maximizing seizure control and minimizing surgical morbidity. We also studied how adult-atypical language representation establishes in the context of early brain insult. Decades of research have demonstrated the relative plasticity of language representation in children versus adults with neurological abnormalities, although the mechanism for the paediatric advantage has not been adequately characterized. By comparing findings from language mapping of healthy children at different ages, and also children and adolescents with intractable epilepsy, we were able to comment on how atypical representation may come about in the context of an early neurological insult.
163 11.2 Summary of findings We used multiple techniques to investigate predictors of language representation in patient and control groups. In Chapter 6 (Kadis et al., 2009), we described findings from our retrospective IAP study. Duchowny et al. (1996) had previously argued that early acquired left hemisphere lesions, but not developmental lesions, result in interhemispheric shifts in language representation. Our extensive experience with IAP at the Hospital for Sick Children permitted large-scale analyses of possible effects of pathology type. We observed equivalent rates of typical and atypical language lateralization in individuals with developmental lesions, acquired lesions, and tumours, indicating that pathology type does not predict interhemispheric plasticity in children with epilepsy. Our lateralization findings were consistent with previous accounts of increased atypical lateralization in individuals with left hemisphere seizures beginning before age 6 years, and support the widely held notion that the potential for interhemispheric reorganization of language decreases significantly during childhood. Seizure side and age at seizure onset, but not pathology type, remain the best predictors of language lateralization. In Chapter 8, we introduced of a novel MEG protocol for mapping expressive language in children and adolescents. We studied a group of healthy adolescents (n = 12) and a small group of adolescents with neurological disorders (n = 4), to characterize the neuromagnetic correlates of verb generation. Control group time-frequency response analyses with virtual sensors placed over the left and right lateral frontal cortices revealed consistent left hemisphere low-beta ERD at approximately 450-650 ms following presentation of picture stimuli. These findings were used to guide differential beamformer analyses, which showed consistent low-beta ERD over canonical Broca’s area and occasional low-beta ERS over the
164 right hemisphere for control subjects completing the verb generation task. In the small patient group, expressive language MEG findings were consistent with other studies of language representation (e.g., cortical stimulation mapping), and showed promise for use in presurgical investigations. In Chapter 9, we studied expressive language lateralization in 25 healthy subjects, aged 5 though 18 years, using naming and verb generation MEG paradigms. The inclusion of young children maximized our ability to detect changes in lateralization that may explain the decreasing potential for interhemispheric plasticity after about age 6 years. To characterize the developmental trajectory, we developed a statistical protocol that permitted tailored yet objective analyses of individual scans. We were able to quantify left versus right frontal lobe low-beta ERD, determined to reflect the expressive language processes (the occasional contralateral ERS observed in Kadis et al., 2008, described in Chapter 8, was not reliable). For naming, we did not observe significant changes in ERD lateralization across development, although regression analyses suggested a modest trend toward increased left hemisphere lateralization with age. For verb generation, we observed a significant increase in left hemisphere lateralization from age 5 though 18, with the regression line indicating a relatively bilateral immature language network that shifts toward left lateralization as part of the normal developmental trajectory. In Chapter 10, we described findings from MEG expressive language mapping in a small (n = 14) paediatric patient group; all subjects had neurological disorders, and all but one had intractable epilepsy. The thresholding and quantification procedures introduced in Chapter 9 permitted objective study of individual patient scans, which is necessary for ultimate clinical implementation. On an individual basis, the assessment protocol showed promise for future
165 use in a presurgical context. A majority of subjects were able to complete both naming and verb generation tasks and had ERD surviving the bootstrap procedure, indicating that lateralizations were based on reliable neuromagnetic changes. We plotted lateralization at four differential windows for each scan, in some cases revealing complex lateralizations that involve rapid shifts in ERD lateralization through the naming or verb generation response period. Group findings were relatively difficult to interpret, due to the small sample size and varied patient profiles. A small subset underwent additional IAP, fMRI, or receptive language MEG studies for language lateralization. In these subjects, we observed poor concordance with our expressive language MEG findings. The discordance may reflect differences in task selection and associated language network engagement. 11.3 Implications Our IAP study (Kadis et al., 2009) has immediate application in the evaluation of surgical candidacy and planning surgical interventions for children with intractable epilepsy. We observed equal rates of typical and atypical language lateralization across pathology groups in large scale, indicating that pathology type cannot be used to predict language dominance. In establishing predictive models of patient lateralization, whether for implementation in formal statistical analyses or informal assessments, the identification and exclusion of nonpredictors (i.e., pathology type) is as important as the consideration of relevant variables, such as seizure side and age at seizure onset. Contrary to previous reports (e.g., Duchowny et al., 1996), our study indicates that the presence of developmental pathologies does not rule out bilateral or right hemisphere language representation in children with medically intractable epilepsy.
166 Our MEG studies with healthy children reveal normal changes in expressive language lateralization through childhood. Differences in representation of naming versus verb generation may explain conflicting reports on lateralization across development in the literature. Informal descriptive evaluations suggest a common trend toward increasing left lateralization of frontal lobe ERD with advancing age for both tasks, although findings from formal inferential statistical analyses indicate task-related differences. Seemingly related paradigms such as naming and verb generation to confrontation with picture stimuli, both expressive language tasks, may involve discreet neuroanatomy (see Mätzig, Druks, Masterson, & Vigliocco, 2009) with associated non-negligible differences in frontal lobe lateralization across development (i.e., naming is known to involve diffuse regions, and lateralization for naming may not be well characterized in frontal lobe ROI analyses; see also, Benson et al. 1999). The task-related differences may be partially explained by incomplete assessment of expressive language areas, due to the exclusive use of picture stimuli in each of our MEG paradigms. Within naming paradigms alone, researchers have documented different patterns of neuroanatomical involvement and lesion sensitivity based on the modality of stimulus presentation. Malow, Blaxton, Sato, et al. (1996) studied sites for naming errors in adults with epilepsy who were undergoing stimulation mapping using pictures and auditory object descriptions; Malow et al. observed distinct sites related to the type of stimulus used (see also, Hamberger, Goodman, Perrine, & Tamny, 2001). Similar stimulus-modality effects have been documented using fMRI (e.g., Miller, Finney, Meador, & Loring, 2009) and MEG (Vihla, Laine, & Salmelin, 2006), and there is some evidence that auditory naming is more sensitive to anterior temporal lesions than visual object naming (Hamberger & Seidel, 2009; Hamberger & Seidel, 2003; see however, Price, Winterburn,
167 Giraud, et al., 2003). In summarizing the literature on paediatric language representation, researchers must remain sensitive to the multifaceted nature of language, acknowledging that discreet processes may be differentially represented in the brain, and that the choice of test stimulus modality may affect the generalizability of findings. Our MEG studies have important clinical implications. We demonstrated the feasibility of assessing expressive language representation with naming and verb generation tasks in MEG, in a sizeable paediatric clinical group. Outside of our studies, we found no published reports of expressive language MEG protocols that are appropriate for use with low-functioning paediatric populations (e.g., children with epilepsy). We designed the naming and verb generation tasks to be brief in administration, engaging, and easy to complete, with response sets that are familiar to young children. Differences in LIs for naming and verb generation demonstrate that discrete language processes often have contrasting lateralization patterns, even within a single subject. Although it is difficult to interpret the discordant naming and verb generation lateralizations at this stage, we feel that it is advantageous to assess language in a comprehensive manner, using multiple paradigms whenever patient ability permits. Our naming and verb generation paradigms, together, may provide a better picture of expressive language lateralization than either task considered independently (see, however, Benson et al., 1999). With larger-scale reliability analyses and validation against IAP, we could formally assess the unique utility of naming and verb generation in MEG. Because each task requires only 4-5 minutes in the MEG scanner, it would be relatively easy to augment these studies with additional language mapping protocols in clinical implementations, to establish an optimal noninvasive clinical battery.
168 The normal developmental changes in frontal lobe lateralization for verb generation (and the similar trend observed for naming) speak to the probable mechanism for establishment of atypical language lateralization following early left hemisphere insult. Our findings show that expressive language is represented differently in the brains of children and adults. Normal development involves relatively bilateral frontal representation of expressive language in early childhood, with shifts toward left frontal lateralization through adolescence. We argue that language fails to shift leftward in the context of early left hemisphere insult; rather, the right hemisphere retains support of critical language functions, resulting in bilateral or right lateralization in adulthood. Accordingly, bilateral language representation seems to serve as a prerequisite for adult-atypical lateralization. Therefore, the potential for bilateral or right hemisphere representation following left hemisphere injury necessarily decreases through childhood, as left lateralization establishes as part of the normal developmental trajectory. If the observed increases in left lateralization are indeed mechanisms for a decreasing potential for interhemispheric plasticity in childhood, then centers offering neurosurgery for children would be well advised to increase their willingness to operate in or near canonical language regions when patients are very young. Neurosurgeons are typically hesitant to operate in the left (adult-dominant) hemisphere when the epileptogenic zone clearly encroaches on perisylvian language regions, for fear of producing lasting aphasias. However, by offering surgery to infant patients, the neurosurgical team may be able to capitalize on the relatively bilateral (plastic) language network of early childhood. Surgical treatments for epilepsy are typically only offered after patients have demonstrated intractability, meaning multiple anticonvulsants at therapeutic doses have failed to control
169 seizures over a prolonged period of time (years). Waiting to establish intractability prior to offering neurosurgery may come at a cost to the patient, in that opportunity for interhemispheric reorganization of language decreases with age. Modified criteria for the establishment of intractability and an aggressive surgical approach in early childhood may be warranted. It is important to note, however, that relative plasticity of early childhood may be specific to the perisylvian language network, without extension to other neural regions or cognitive domains. Anderson, Spencer-Smith, Leventer et al. (2009) recently studied general cognitive and executive functioning outcomes in individuals with various neurological insults occurring at different times in childhood, and observed non-linear changes in age-related vulnerability. Poor outcomes were associated with insults occurring before 2 years of age and between 7 and 9 years of age, with relative protection for insults occurring between 3 and 6 years of age. Those with prenatal or perinatal insults were at greatest risk for general cognitive and executive dysfunction, directly contrasting the literature on language outcomes in children with congenital cerebral insults (Reilly, Bates, & Marchman, 1998; Bates et al., 2001). 11.4 Future directions Research into prediction of language representation will continue to benefit neurological populations. Centers with extensive experience in IAP and stimulation mapping have the greatest potential to rapidly advance our understanding of how the brain represents language in the context of injury. The continued improvement of predictive models of language representation will facilitate informed surgical decision making, and particularly benefit patients of centers with limited investigative resources.
170 In assessing control and patient participants, we found that naming and verb generation were differentially represented across development, and in some cases, differentially lateralized within single subjects. Although both tasks were intended to help us identify expressive language cortex, it is clear that we were studying discrete language processes, fractions of normal language expression. In the future, particularly with the implementation of newer MEG systems that permit continuous monitoring of head location, we may be able to use paradigms involving overt responding. Ideally, we would not restrict responses to single word production, but rather promote natural responding by engaging the subject in dialogue. In doing so, we may be able to approximate normal language expression and ultimately study the neural representation of free discourse (see Chapter 2). This would provide a comprehensive picture of the brain’s representation of language, which would improve our ability to predict and prevent post-operative dysphasias that can occur even when canonical language regions are spared in neurological accidents neurosurgical interventions (see Chapters 3 and 4). In our current studies, the generative nature of naming and verb generation and epoching based on stimulus presentation resulted in induced, rather than evoked neuromagnetic responses. The induced signal was characterized with differential beamformer analyses. Differential beamforming retains excellent spatial resolution but does not fully capitalize on the excellent millisecond temporal resolution possible with MEG recordings. In the future, overt responses may be used to epoch data from continuous MEG recordings, which may reveal evoked characteristics of language expression. If the precise timing of expressive language responding could be determined, we could develop sophisticated network analyses which reveal the temporal dynamics of the distributed language system, thus capitalizing on
171 the excellent spatial and temporal resolution of MEG. Ultimately, we hope to understand the functional importance of each component of the language network, and appreciate how discrete regions interact to permit complex language processing.
172 References
Albert, M. L., & Helm-Estabrooks, N. (1988a). Diagnosis and treatment of aphasia. part I. The Journal of the American Medical Association, 259(7), 1043-1047. Albert, M. L., & Helm-Estabrooks, N. (1988b). Diagnosis and treatment of aphasia. part II. The Journal of the American Medical Association, 259(8), 1205-1210. Anderson, D. P., Harvey, A. S., Saling, M. M., Anderson, V., Kean, M., Abbott, D. F., et al. (2006). FMRI lateralization of expressive language in children with cerebral lesions. Epilepsia, 47(6), 998-1008. Anderson, V., Spencer-Smith, M., Leventer, R., Coleman, L., Anderson, P., Williams, J., et al. (2009). Childhood brain insult: Can age at insult help us predict outcome? Brain, 132(Pt 1), 45-56. Arcadi, A. C. (2005). Language evolution: What do chimpanzees have to say? Current Biology, 15(21), R884-6. Archibald, Y. M., & Wepman, J. M. (1968). Language disturbance and nonverbal cognitive performance in eight patients following injury to the right hemisphere. Brain, 91(1), 117-130. Baillieux, H., De Smet, H. J., Paquier, P. F., De Deyn, P. P., & Marien, P. (2008). Cerebellar neurocognition: Insights into the bottom of the brain. Clinical Neurology and Neurosurgery, 110(8), 763-773.
173 Balsamo, L. M., Xu, B., Grandin, C. B., Petrella, J. R., Braniecki, S. H., Elliott, T. K., et al. (2002). A functional magnetic resonance imaging study of left hemisphere language dominance in children. Archives of Neurology, 59(7), 1168-1174. Bates, E., Reilly, J., Wulfeck, B., Dronkers, N., Opie, M., Fenson, J., et al. (2001). Differential effects of unilateral lesions on language production in children and adults. Brain Lang, 79(2), 223-265. Baxendale, S. (2009). The Wada test. Current Opinion in Neurology, 22(2), 185-189. Baxendale, S., Thompson, P. J., & Duncan, J. S. (2008a). Evidence-based practice: A reevaluation of the intracarotid amobarbital procedure (Wada test). Archives of Neurology, 65(6), 841-845. Baxendale, S., Thompson, P. J., & Duncan, J. S. (2008b). The role of the Wada test in the surgical treatment of temporal lobe epilepsy: An international survey. Epilepsia, 49(4), 715-20; discussion 720-5. Bell, D. S. (1968). Speech functions of the thalamus inferred from the effects of thalamotomy. Brain, 91(4), 619-638. Benson, R. R., FitzGerald, D. B., LeSueur, L. L., Kennedy, D. N., Kwong, K. K., Buchbinder, B. R., et al. (1999). Language dominance determined by whole brain functional MRI in patients with brain lesions. Neurology, 52(4), 798-809. Berl, M. M., Vaidya, C. J., & Gaillard, W. D. (2006). Functional imaging of developmental and adaptive changes in neurocognition. NeuroImage, 30(3), 679-691.
174 Bhatnagar, S. C., & Mandybur, G. T. (2005). Effects of intralaminar thalamic stimulation on language functions. Brain and Language, 92(1), 1-11. Billingsley-Marshall, R. L., Clear, T., Mencl, W. E., Simos, P. G., Swank, P. R., Men, D., et al. (2007). A comparison of functional MRI and magnetoencephalography for receptive language mapping. Journal of Neuroscience Methods, 161(2), 306-313. Billingsley, R. & Smith, M. L. (2000). Intelligence profiles in children and adolescents with left temporal lobe epilepsy: Relationship to language laterality. Brain and Cognition, 43, 44-49. Binder, J. R. (1997). Neuroanatomy of language processing studied with functional MRI. Clinical Neuroscience, 4(2), 87-94. Blank, S. C., Bird, H., Turkheimer, F., & Wise, R. J. (2003). Speech production after stroke: The role of the right pars opercularis. Annals of Neurology, 54(3), 310-320. Bloom, K., Russell, A., & Wassenberg, K. (1987). Turn taking affects the quality of infant vocalizations. Journal of Child Language, 14(2), 211-227. Boatman, 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(4), 579-586. Bogen, J. E., & Bogen, G. M. (1976). Wernicke's region--where is it? Annals of the New York Academy of Sciences, 280, 834-843.
175 Booth, J. R., Wood, L., Lu, D., Houk, J. C., & Bitan, T. (2007). The role of the basal ganglia and cerebellum in language processing. Brain Research, 1133(1), 136-144. Bouillaud, J. B. (1825). Recherches cliniques propres à démontrer que la perte de la parole correspond à la lésion des lobules antérieurs du cerveau, et à confirmer l’opinion de M. gall, sur le siège de l’organe du langage articulé. Archives Générales De Médecine, 8, 25-45. Branch, C., Milner, B., & Rasmussen, T. (1964). Intracarotid sodium amytal for the lateralization of cerebral speech dominance; observations in 123 patients. Journal of Neurosurgery, 21, 399-405. Brazdil, M., Zakopcan, J., Kuba, R., Fanfrdlova, Z., & Rektor, I. (2003). Atypical hemispheric language dominance in left temporal lobe epilepsy as a result of the reorganization of language functions. Epilepsy & Behavior, 4(4), 414-419. Breier, J. I., Billingsley-Marshall, R., Pataraia, E., Castillo, E. M., & Papanicolaou, A. C. (2006). Magnetoencephalographic studies of language reorganization after cerebral insult. Arch Phys Med Rehabil, 87(12), 77-83. Breier, J. I., Maher, L. M., Novak, B., & Papanicolaou, A. C. (2006). Functional imaging before and after constraint-induced language therapy for aphasia using magnetoencephalography. Neurocase, 12(6), 322-331.
176 Breier, J. I., Simos, P. G., Zouridakis, G., & Papanicolaou, A. C. (1998). Relative timing of neuronal activity in distinct temporal lobe areas during a recognition memory task for words. Journal of Clinical and Experimental Neuropsychology, 20(6), 782-790. Breier, J. I., Simos, P. G., Zouridakis, G., & Papanicolaou, A. C. (1999). Lateralization of cerebral activation in auditory verbal and non-verbal memory tasks using magnetoencephalography. Brain Topography, 12(2), 89-97. Brent, M. R., & Siskind, J. M. (2001). The role of exposure to isolated words in early vocabulary development. Cognition, 81(2), B33-44. Briellmann, R. S., Labate, A., Harvey, A. S., Saling, M. M., Sveller, C., Lillywhite, L., et al. (2006). Is language lateralization in temporal lobe epilepsy patients related to the nature of the epileptogenic lesion? Epilepsia, 47(5), 916-920. Broca, P. P. (1861a). Perte de la parole, ramollissement chronique et destruction partielle du lobe antérieur gauche du cerveau. Bulletin De La Société Anthropologique, 2, 235-238. Broca, P. P. (1861b). Remarques sur le siége de la faculté du langage articulé, suivies d'une observation d'aphémie (perte de la parole). Bulletin De La Société Anthropologique, 6, 330-357. Broca, P. P. (1865). Sur le siege de la faculté du language articulé. Bulletins De La Société Anthropologie De Paris, 6, 377-393.
177 Brookes, M. J., Stevenson, C. M., Barnes, G. R., Hillebrand, A., Simpson, M. I., Francis, S. T., &Morris, P. G. (2007). Beamformer reconstruction of correlated sources using a modified source model. Neuroimage, 34(4), 1454-1465. Brown, T. T., Lugar, H. M., Coalson, R. S., Miezin, F. M., Petersen, S. E., & Schlaggar, B. L. (2005). Developmental changes in human cerebral functional organization for word generation. Cerebral Cortex, 15(3), 275-290. Bruce, B. B., Foote, K. D., Rosenbek, J., Sapienza, C., Romrell, J., Crucian, G., et al. (2004). Aphasia and thalamotomy: Important issues. Stereotactic and Functional Neurosurgery, 82(4), 186-190. Cao, Y., Vikingstad, E. M., George, K. P., Johnson, A. F., & Welch, K. M. (1999). Cortical language activation in stroke patients recovering from aphasia with functional MRI. Stroke, 30(11), 2331-2340. Cappa, S. F., & Vignolo, L. A. (1979). "Transcortical" features of aphasia following left thalamic hemorrhage. Cortex, 15(1), 121-130. Castaigne, P., Lhermitte, F., Signoret, J. L., & Abelanet, R. (1980). Description and scanographic study of leborgne's brain. broca's discovery. [Description et etude scannographique du cerveau de Leborgne. La decouverte de Broca] Revue Neurologique, 136(10), 563-583. Chang, F., Dell, G. S., & Bock, K. (2006). Becoming syntactic. Psychological Review, 113(2), 234-272.
178 Cheyne, D., Bakhtazad, L., & Gaetz, W. (2006). Spatiotemporal mapping of cortical activity accompanying voluntary movements using an event-related beamforming approach. Human Brain Mapping, 27(3), 213-229. Cheyne, D., Bostan, A. C., Gaetz, W., & Pang, E. W. (2007). Event-related beamforming: A robust method for presurgical functional mapping using MEG. Clinical Neurophysiology, 118(8), 1691-1704. Cheyne, D., & Vrba, J. (2006). Biomagnetism. Encyclopedia of medical devices and instrumentation (2nd ed., pp. 230-255) . Cheyne, D., & Weinberg, H. (1989). Neuromagnetic fields accompanying unilateral finger movements: Pre-movement and movement-evoked fields. Experimental Brain Research, 78(3), 604-612. Chuang, N. A., Otsubo, H., Pang, E. W., & Chuang, S. H. (2006). Pediatric magnetoencephalography and magnetic source imaging. Neuroimaging Clinics of North America, 16(1), 193-210, ix-x. Crosson, B. (1999). Subcortical mechanisms in language: Lexical-semantic mechanisms and the thalamus. Brain and Cognition, 40(2), 414-438. De Smet, H. J., Baillieux, H., Catsman-Berrevoets, C., De Deyn, P. P., Marien, P., & Paquier, P. F. (2007). Postoperative motor speech production in children with the syndrome of 'cerebellar' mutism and subsequent dysarthria: A critical review of the literature. European Journal of Paediatric Neurology, 11(4), 193-207.
179 Dejerine, J. (1892). Contribution à l'étude anatomo-pathologique et clinique des différentes variétés de cécité verbale. Mémoires De La Société De Biologie, 4, 61-90. Devlin, 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 childhood and adolescence. Brain, 126, 556-566. Devlin, J. T., & Watkins, K. E. (2007). Stimulating language: Insights from TMS. Brain, 130(Pt 3), 610-622. DeVos, K. J., Wyllie, E., Geckler, C., Kotagal, P., & Comair, Y. (1995). Language dominance in patients with early childhood tumors near left hemisphere language areas. Neurology, 45(2), 349-356. Dhond, R. P., Buckner, R. L., Dale, A. M., Marinkovic, K., & Halgren, E. (2001). Spatiotemporal maps of brain activity underlying word generation and their modification during repetition priming. The Journal of Neuroscience, 21(10), 35643571. Dogdas, B., Shattuck, D. W., & Leahy, R. M. (2005). Segmentation of skull and scalp in 3-D human MRI using mathematical morphology. Human Brain Mapping, 26(4), 273-285. Draganova, R., Eswaran, H., Murphy, P., Lowery, C., & Preissl, H. (2007). Serial magnetoencephalographic study of fetal and newborn auditory discriminative evoked responses. Early Human Development, 83(3), 199-207.
180 Dronkers, N. F., Plaisant, O., Iba-Zizen, M. T., & Cabanis, E. A. (2007). Paul Broca's historic cases: High resolution MR imaging of the brains of leborgne and lelong. Brain, 130(Pt 5), 1432-1441. Duchowny, M. (2007). Language localization, the developing brain and childhood epilepsy: Back to the future. Journal of the International Neuropsychological Society, 13(3), 501504. Duchowny, M., Jayakar, P., Harvey, A. S., Resnick, T., Alvarez, L., Dean, P., et al. (1996). Language cortex representation: Effects of developmental versus acquired pathology. Annals of Neurology, 40(1), 31-38. Duncan, J. D., Moss, S. D., Bandy, D. J., Manwaring, K., Kaplan, A. M., Reiman, E. M., et al. (1997). Use of positron emission tomography for presurgical localization of eloquent brain areas in children with seizures. Pediatric Neurosurgery, 26(3), 144-156. Eimas, P. D., Siqueland, E. R., Jusczyk, P., & Vigorito, J. (1971). Speech perception in infants. Science, 171(968), 303-306. Engel, A. K., Moll, C. K., Fried, I., & Ojemann, G. A. (2005). Invasive recordings from the human brain: Clinical insights and beyond. Nature Reviews, Neuroscience, 6(1), 35-47. Ettlinger, G., Jackson, C. V., & Zangwill, O. L. (1955). Dysphasia following right temporal lobectomy in a right-handed man. Journal of Neurology, Neurosurgery, and Psychiatry, 18(3), 214-217.
181 Eulitz, C., Elbert, T., Bartenstein, P., Weiller, C., Muller, S. P., & Pantev, C. (1994). Comparison of magnetic and metabolic brain activity during a verb generation task. Neuroreport, 6(1), 97-100. Faglia, L., Rottoli, M. R., & Vignolo, L. A. (1990). Aphasia due to lesions confined to the right hemisphere in right handed patients: A review of the literature including the Italian cases. Italian Journal of Neurological Sciences, 11(2), 131-144. Fernandes, M. A., & Smith, M. L. (2000). Comparing the fused dichotic words test and the intracarotid amobarbital procedure in children with epilepsy. Neuropsychologia, 38(9), 1216-1228. Fernandez, B., Cardebat, D., Demonet, J. F., Joseph, P. A., Mazaux, J. M., Barat, M., et al. (2004). Functional MRI follow-up study of language processes in healthy subjects and during recovery in a case of aphasia. Stroke, 35(9), 2171-2176. Flagg, E. J., Cardy, J. E., Roberts, W., & Roberts, T. P. (2005). Language lateralization development in children with autism: Insights from the late field magnetoencephalogram. Neuroscience Letters, 386(2), 82-87. Frank, B., Schoch, B., Hein-Kropp, C., Dimitrova, A., Hovel, M., Ziegler, W., et al. (2007). Verb generation in children and adolescents with acute cerebellar lesions. Neuropsychologia, 45(5), 977-988.
182 Frank, B., Schoch, B., Richter, S., Frings, M., Karnath, H. O., & Timmann, D. (2007). Cerebellar lesion studies of cognitive function in children and adolescents - limitations and negative findings. Cerebellum, 6(3), 242-253. Friston, K. (2003). Introduction: Experimental design and statistical parameter mapping. In R. Frackowiak (Ed.), Human brain function (2nd ed.). Fritsch, G. T., & Hitzig, E. (1870). Ueber die elektrische erregbarkeit des grosshirns. Arch. Anat. Physiol., 37, 300-332. Gaillard, W. D., Balsamo, L. M., Ibrahim, Z., Sachs, B. C., & Xu, B. (2003). fMRI identifies regional specialization of neural networks for reading in young children. Neurology, 60(1), 94-100. Gaillard, W. D., Berl, M. M., Moore, E. N., Ritzl, E. K., Rosenberger, L. R., Weinstein, S. L., et al. (2007). Atypical language in lesional and nonlesional complex partial epilepsy. Neurology, 69(18), 1761-1771. Gaillard, W. D., Hertz-Pannier, L., Mott, S. H., Barnett, A. S., LeBihan, D., & Theodore, W. H. (2000). Functional anatomy of cognitive development: FMRI of verbal fluency in children and adults. Neurology, 54(1), 180-185. Gaillard, W. D., Sachs, B. C., Whitnah, J. R., Ahmad, Z., Balsamo, L. M., Petrella, J. R., et al. (2003). Developmental aspects of language processing: FMRI of verbal fluency in children and adults. Hum Brain Mapp, 18(3), 176-185.
183 Geschwind, N. (1964). Non-aphasic disorders of speech. International Journal of Neurology, 4(3), 207-214. Geschwind, N. (1965). Disconnexion syndromes in animals and man. I. Brain, 88(2), 237294. Geschwind, N. (1970). The organization of language and the brain. Science, 170(961), 940944. Geschwind, N. (1971). Current concepts: Aphasia. The New England Journal of Medicine, 284(12), 654-656. Geschwind, N. (1972). Language and the brain. Scientific American, 226(4), 76-83. Gil Robles, S., Gatignol, P., Capelle, L., Mitchell, M. C., & Duffau, H. (2005). The role of dominant striatum in language: A study using intraoperative electrical stimulations. Journal of Neurology, Neurosurgery, and Psychiatry, 76(7), 940-946. Gleissner, U., Kurthen, M., Sassen, R., Kuczaty, S., Elger, C. E., Linke, D. B., et al. (2003). Clinical and neuropsychological characteristics of pediatric epilepsy patients with atypical language dominance. Epilepsy & Behavior, 4(6), 746-752. Goldstein, M. H., King, A. P., & West, M. J. (2003). Social interaction shapes babbling: Testing parallels between birdsong and speech. Proceedings of the National Academy of Sciences of the United States of America, 100(13), 8030-8035.
184 Gout, A., Seibel, N., Rouviere, C., Husson, B., Hermans, B., Laporte, N., et al. (2005). Aphasia owing to subcortical brain infarcts in childhood. Journal of Child Neurology, 20(12), 1003-1008. Graf Estes, K., Evans, J. L., Alibali, M. W., & Saffran, J. R. (2007). Can infants map meaning to newly segmented words? statistical segmentation and word learning. Psychological Science, 18(3), 254-260. Gross, J., Kujala, J., Hamalainen, M., Timmermann, L., Schnitzler, A., & Salmelin, R. (2001). Dynamic imaging of coherent sources: Studying neural interactions in the human brain. Proceedings of the National Academy of Sciences of the United States of America, 98(2), 694-699. Hämäläinen, M., & Hari, R. (2002). Magnetoencephalographic characterization of dynamic brain activation: Basic principles and methods of data collection and source analysis. In A. W. Toga, & J. C. Mazziotta (Eds.), Brain Mapping: The Methods (pp. 227-254). New York: Academic Press. Hämäläinen, M. S. (1992). Magnetoencephalography: A tool for functional brain imaging. Brain Topography, 5(2), 95-102. Hamberger, M. J., Goodman, R. R., Perrine, K., & Tamny, T. (2001). Anatomic dissociation of auditory and visual naming in the lateral temporal cortex. Neurology, 56(1), 56-61.
185 Hamberger, M. J., & Seidel, W. T. (2003). Auditory and visual naming tests: normative and patient data for accuracy, response time, and tip-of-the-tongue. Journal of the International Neuropsychological Society, 9(3), 479-489. Hamberger, M. J., & Seidel, W. T. (2009). Localization of cortical dysfunction based on auditory and visual naming performance. Journal of the International Neuropsychological Society, 15(4), 429-535. Helmstaedter, C., Kurthen, M., Linke, D. B., & Elger, C. E. (1997). Patterns of language dominance in focal left and right hemisphere epilepsies: Relation to MRI findings, EEG, sex, and age at onset of epilepsy. Brain and Cognition, 33(2), 135-150. Henry, R. G., Berman, J. I., Nagarajan, S. S., Mukherjee, P., & Berger, M. S. (2004). Subcortical pathways serving cortical language sites: Initial experience with diffusion tensor imaging fiber tracking combined with intraoperative language mapping. NeuroImage, 21(2), 616-622. Herdman, A. T., Pang, E. W., Ressel, V., Gaetz, W., & Cheyne, D. (2007). Task-related modulation of early cortical responses during language production: An event-related synthetic aperture magnetometry study. Cerebral Cortex, 17(11), 2536-2543. Herdman, A. T., Wollbrink, A., Chau, W., Ishii, R., Ross, B., & Pantev, C. (2003). Determination of activation areas in the human auditory cortex by means of synthetic aperture magnetometry. NeuroImage, 20(2), 995-1005.
186 Hillebrand, A., & Barnes, G. R. (2005). Beamformer analysis of MEG data. International Review of Neurobiology, 68, 149-171. Hillis, A. E. (2007). Aphasia: Progress in the last quarter of a century. Neurology, 69(2), 200213. Hillis, A. E., Barker, P. B., Wityk, R. J., Aldrich, E. M., Restrepo, L., Breese, E. L., et al. (2004). Variability in subcortical aphasia is due to variable sites of cortical hypoperfusion. Brain and Language, 89(3), 524-530. Hirata, M., Kato, A., Taniguchi, M., Saitoh, Y., Ninomiya, H., Ihara, A., et al. (2004). Determination of language dominance with synthetic aperture magnetometry: Comparison with the Wada test. NeuroImage, 23(1), 46-53. Holland, S. K., Plante, E., Weber Byars, A., Strawsburg, R. H., Schmithorst, V. J., & Ball, W. S.,Jr. (2001). Normal fMRI brain activation patterns in children performing a verb generation task. NeuroImage, 14(4), 837-843. Holmes, G. L. (1993). Surgery for intractable seizures in infancy and early childhood. Neurology, 43(11 Suppl. 5), S28-37. Huber, J. F., Bradley, K., Spiegler, B., & Dennis, M. (2007). Long-term neuromotor speech deficits in survivors of childhood posterior fossa tumors: Effects of tumor type, radiation, age at diagnosis, and survival years. Journal of Child Neurology, 22(7), 848854.
187 International League Against Epilepsy (ILAE). (2009). Instructions for authors. Epilepsia, 50, July, 2009. Jabbour, R. A., Hempel, A., Gates, J. R., Zhang, W., & Risse, G. L. (2005). Right hemisphere language mapping in patients with bilateral language. Epilepsy & Behavior, 6(4), 587-592. Jackson, J. H. (1866). Notes on the physiology and pathology of language. Medical Times and Gazette, 1, 659. Janszky, J., Jokeit, H., Heinemann, D., Schulz, R., Woermann, F. G., & Ebner, A. (2003). Epileptic activity influences the speech organization in medial temporal lobe epilepsy. Brain, 126, 2043-2051. Jusczyk, P. W. (2002). Some critical developments in acquiring native language sound organization during the first year. The Annals of Otology, Rhinology & Laryngology, 189(Suppl.), 11-15. Jusczyk, P. W.,Cutler, A., & Redanz, N. J. (1993). Infants’ preference for the predominant stress patterns of English words. Child Development, 64(3), 675-687.. Kadis, D. S., Iida, K., Kerr, E. N., Logan, W. J., McAndrews, M. P., Ochi, A., Otsubo, H., Rutka, J. T., Snead, O. C. 3rd., Weiss, S. K., & Smith, M. L. (2007). Intrahemispheric reorganization of language in children with medically intractable epilepsy of the left hemisphere. Journal of the International Neuropsychological Society, 13(3), 505-516.
188 Kadis, D. S., Kerr, E. N., Rutka, J. T., Snead, O. C. 3rd., Weiss, S. K., & Smith, M. L. (2009). Pathology type does not predict language lateralization in children with medically intractable epilepsy. Epilepsia, 50(6), 1498-1504. Kadis, D. S., Smith, M. L., Mills, T., & Pang, E. W. (2008). Expressive language mapping in children using MEG; MEG localization of expressive language cortex in healthy children: Appliation to paediatric clinical populations. Down Syndrome Quarterly, 10(2), 5-12. Kasai, K., Hashimoto, O., Kawakubo, Y., Yumoto, M., Kamio, S., Itoh, K., et al. (2005). Delayed automatic detection of change in speech sounds in adults with autism: A magnetoencephalographic study. Clinical Neurophysiology, 116(7), 1655-1664. Knecht, S., Deppe, M., Drager, B., Bobe, L., Lohmann, H., Ringelstein, E., et al. (2000). Language lateralization in healthy right-handers. Brain, 123 ( Pt 1), 74-81. Kristeva, R., Cheyne, D., & Deecke, L. (1991). Neuromagnetic fields accompanying unilateral and bilateral voluntary movements: Topography and analysis of cortical sources. Electroencephalography and Clinical Neurophysiology, 81(4), 284-298. Kuhl, P. K. (2004). Early language acquisition: Cracking the speech code. Nature Reviews, Neuroscience, 5(11), 831-843. Kuhl, P. K., Stevens, E., Hayashi, A., Deguchi, T., Kiritani, S., & Iverson, P. (2006). Infants show a facilitation effect for native language phonetic perception between 6 and 12 months. Developmental Science, 9(2), F13-F21.
189 Lazar, R. M., Marshall, R. S., Pile-Spellman, J., Duong, H. C., Mohr, J. P., Young, W. L., et al. (2000). Interhemispheric transfer of language in patients with left frontal cerebral arteriovenous malformation. Neuropsychologia, 38(10), 1325-1332. Lichtheim, L. (1885). On aphasia. Brain, 7, 433-484. Liegeois, F., Connelly, A., Cross, J. H., Boyd, S. G., Gadian, D. G., Vargha-Khadem, F., et al. (2004). Language reorganization in children with early-onset lesions of the left hemisphere: An fMRI study. Brain, 127, 1229-1236. Liegeois, F., Cross, J. H., Polkey, C., Harkness, W., & Vargha-Khadem, F. (2008). Language after hemispherectomy in childhood: Contributions from memory and intelligence. Neuropsychologia, 46(13), 3101-3107. Lopes da Silva, F. (1991). Neural mechanisms underlying brain waves: From neural membranes to networks. Electroencephalography and Clinical Neurophysiology, 79(2), 81-93. Loring, D. W., Strauss, E., Hermann, B. P., Perrine, K., Trenerry, M. R., Barr, W. B., et al. (1999). Effects of anomalous language representation on neuropsychological performance in temporal lobe epilepsy. Neurology, 53(2), 260-264. Luciani, L. (1891). Il cervelletto:Nuovi studi di fisiologia normale e patologica. Le Monnier. Luzzatti, C., & Whitaker, H. (2001). Jean-baptiste bouillaud, claude-francois lallemand, and the role of the frontal lobe: Location and mislocation of language in the early 19th century. Archives of Neurology, 58(7), 1157-1162.
190 Makela, J. P., Forss, N., Jaaskelainen, J., Kirveskari, E., Korvenoja, A., & Paetau, R. (2006). Magnetoencephalography in neurosurgery. Neurosurgery, 59(3), 493-510; discussion 510-1. Malow, B. A., Blaxton, T. A., Sato, S., Bookheimer, S. Y., Kufta, C. V., Figlozzi, C. M., & Theodore, W. H. (1996). Cortical stimulation elicits regional distinctions in auditory and visual naming. Epilepsia, 37(3), 245-252. Manni, E., & Petrosini, L. (1997). Luciani's work on the cerebellum a century later. Trends in Neurosciences, 20(3), 112-116. Marcie, P., Hécaen, H., Dubois, J., & Angelergues, R. (1965). Les réalisations du langage chez les malades atteints de lésions de l'hémisphere droit. Neuropsychologia, 3(3), 217245. Mätzig, S., Druks, J., Masterson, J., & Vigliocco, G. (2009). Noun and verb differences in picture naming: past studies and new evidence. Cortex, 45(6), 738-758. Meinzer, M., Flaisch, T., Breitenstein, C., Wienbruch, C., Elbert, T., & Rockstroh, B. (2008). Functional re-recruitment of dysfunctional brain areas predicts language recovery in chronic aphasia. NeuroImage, 39(4), 2038-2046. Metz-Lutz, M. N., Namer, I. J., Gounot, D., Kleitz, C., Armspach, J. P., & Kehrli, P. (2000). Language functional neuro-imaging changes following focal left thalamic infarction. Neuroreport, 11(13), 2907-2912.
191 Miller, K. M., Finney, G. R., Meador, K. J., & Loring, D. W. (2009). Auditory responsive naming versus visual confrontation naming in dementia. Clinical Neuropsychology, 23, 1-16. Minassian, B. A., Otsubo, H., Weiss, S., Elliott, I., Rutka, J. T., & Snead, O. C. 3rd. (1999). Magnetoencephalographic localization in pediatric epilepsy surgery: Comparison with invasive intracranial electroencephalography. Annals of Neurology, 46(4), 627-633. Miura, K. (1923). On the function of the cerebellum. Science, 58(1506), 361-362. Mohamed, I. S., Cheyne, D., Gaetz, W. C., Otsubo, H., Logan, W. J., Snead, O. C. 3rd., et al. (2008). Spatiotemporal patterns of oscillatory brain activity during auditory word recognition in children: A synthetic aperture magnetometry study. International Journal of Psychophysiology, 68(2), 141-148. Mohr, J. O., Walter, W. C., & Duncan, G. W. (1975). Thalamic hemorrhage and aphasia. Brain and Language, 2, 17-31. Molinari, M., Leggio, M. G., & Silveri, M. C. (1997). Verbal fluency and agrammatism. International Review of Neurobiology, 41, 325-339. Muller, R. A., Rothermel, R. D., Behen, M. E., Muzik, O., Chakraborty, P. K., & Chugani, H. T. (1999). Language organization in patients with early and late left-hemisphere lesion: A PET study. Neuropsychologia, 37(5), 545-557.
192 Muller, R. A., Rothermel, R. D., Behen, M. E., Muzik, O., Mangner, T. J., Chakraborty, P. K., et al. (1998). Brain organization of language after early unilateral lesion: A PET study. Brain and Language, 62(3), 422-451. Nadeau, S. E., & Crosson, B. (1997). Subcortical aphasia. Brain and Language, 58(3), 355402. Naigles, L. R., Bavin, E. L., & Smith, M. A. (2005). Toddlers recognize verbs in novel situations and sentences. Developmental Science, 8(5), 424-431. Narajana, S., Jacks, A., Robin, D.A., Poizner, H., Zhang, W., Franklin, C., Liotti, M., Vogel, D., & Fox, P. T. (2009) A noninvasive imaging approach to understanding speech changes following deep brain stimulation in Parkinson’s disease. American Journal of Speech Language Pathology, 18(2), 146-161. Nass, R., Boyce, L., Leventhal, F., Levine, B., Allen, J., Maxfield, C., et al. (2000). Acquired aphasia in children after surgical resection of left-thalamic tumors. Developmental Medicine and Child Neurology, 42(9), 580-590. Nucifora, P. G., Verma, R., Melhem, E. R., Gur, R. E., & Gur, R. C. (2005). Leftward asymmetry in relative fiber density of the arcuate fasciculus. Neuroreport, 16(8), 791794. Ojemann, G. A. (1979). Individual variability in cortical localization of language. Journal of Neurosurgery, 50(2), 164-169.
193 Ojemann, G. A., Fedio, P., & Van Buren, J. M. (1968). Anomia from pulvinar and subcortical parietal stimulation. Brain, 91(1), 99-116. Ojemann, G., Ojemann, J., Lettich, E., & Berger, M. (1989). Cortical language localization in left, dominant hemisphere: An electrical stimulation mapping investigation in 117 patients. Journal of Neurosurgery, 71(3), 316-326. Oldfield, R. C. (1971). The assessment and analysis of handedness: The edinburgh inventory. Neuropsychologia, 9(1), 97-113. Oram Cardy, J. E., Flagg, E. J., Roberts, W., & Roberts, T. P. (2005). Delayed mismatch field for speech and non-speech sounds in children with autism. Neuroreport, 16(5), 521-525. Otsubo, H., & Snead, O. C. 3rd. (2001). Magnetoencephalography and magnetic source imaging in children. Journal of Child Neurology, 16(4), 227-235. Ozimek, A., Richter, S., Hein-Kropp, C., Schoch, B., Gorissen, B., Kaiser, O., et al. (2004). Cerebellar mutism--report of four cases. Journal of Neurology, 251(8), 963-972. Papanicolaou, A. C., Pazo-Alvarez, P., Castillo, E. M., Billingsley-Marshall, R. L., Breier, J. I., Swank, P. R., et al. (2006). Functional neuroimaging with MEG: Normative language profiles. NeuroImage, 33(1), 326-342. Papanicolaou, A. C., Simos, P. G., Castillo, E. M., Breier, J. I., Sarkari, S., Pataraia, E., et al. (2004). Magnetocephalography: A noninvasive alternative to the wada procedure. J Neurosurg, 100(5), 867-876.
194 Parsons, T. D., Rogers, S. A., Braaten, A. J., Woods, S. P., & Tröster, A. I. (2006). Cognitive sequelae of subthalamic nucleus deep brain stimulation in Parkinson’s diseas: a metaanalysis. Lancet Neurology, 5(7), 578-588. Pataraia, E., Simos, P. G., Castillo, E. M., Billingsley-Marshall, R. L., McGregor, A. L., Breier, J. I., et al. (2004). Reorganization of language-specific cortex in patients with lesions or mesial temporal epilepsy. Neurology, 63(10), 1825-1832. Penfield, W., & Roberts, L. (1959). Speech and brain-mechanisms. New Jersey: Princeton University Press. Peru, A., Moro, V., Tellini, P., & Tassinari, G. (2006). Suggestive evidence for an involvement of the right hemisphere in the recovery from childhood aphasia: A 3-year follow-up case study. Neurocase, 12(3), 179-190. Pfurtscheller, G., & Aranibar, A. (1977). Event-related cortical desynchronization detected by power measurements of scalp EEG. Electroencephalography and Clinical Neurophysiology, 42(6), 817-826. Pfurtscheller, G., & Lopes da Silva, F. H. (1999). Event-related EEG/MEG synchronization and desynchronization: Basic principles. Clinical Neurophysiology, 110(11), 1842-1857. Poeck, K., Kerschensteiner, M., & Hartje, W. (1972). A quantitative study on language understanding in fluent and nonfluent aphasia. Cortex, 8(3), 299-304.
195 Powell, H. W., Parker, G. J., Alexander, D. C., Symms, M. R., Boulby, P. A., WheelerKingshott, C. A., et al. (2006). Hemispheric asymmetries in language-related pathways: A combined functional MRI and tractography study. NeuroImage, 32(1), 388-399. Price, C. J. (2000). The anatomy of language: Contributions from functional neuroimaging. Journal of Anatomy, 197 Pt 3, 335-359. Price, C. J., Winterburn, D., Giraud, A. L., Moore, C. J., & Noppeney, U. (2003). Cortical localisation of the visual and auditory word form areas: a reconsideration of the evidence. Brain & Language, 86(2), 272-286. Quraan, M. A., & Cheyne, D. (in press). Reconstruction of correlated brain activity with adaptive spatial filters in MEG. Neuroimage. Raboyeau, G., De Boissezon, X., Marie, N., Balduyck, S., Puel, M., Bezy, C., et al. (2008). Right hemisphere activation in recovery from aphasia: Lesion effect or function recruitment? Neurology, 70(4), 290-298. Radanovic, M., & Scaff, M. (2003). Speech and language disturbances due to subcortical lesions. Brain and Language, 84(3), 337-352. Rasmussen, T., & Milner, B. (1977). The role of early left-brain injury in determining lateralization of cerebral speech functions. Ann N Y Acad Sci, 299, 355-369. Reilly, J. S., Bates, E. A., & Marchman, V. A. (1998). Narrative discourse in children with early focal brain injury. Brain and Language, 61(3), 335-375.
196 Ressel, V., Wilke, M., Lidzba, K., Lutzenberger, W., & Krageloh-Mann, I. (2008). Increases in language lateralization in normal children as observed using magnetoencephalography. Brain and Language, 106(3), 167-176. Richter, S., Aslan, B., Gerwig, M., Wilhelm, H., Kramer, S., Todica, O., et al. (2007). Patients with chronic focal cerebellar lesions show no cognitive abnormalities in a bedside test. Neurocase : Case Studies in Neuropsychology, Neuropsychiatry, and Behavioural Neurology, 13(1), 25-36. Richter, S., Gerwig, M., Aslan, B., Wilhelm, H., Schoch, B., Dimitrova, A., et al. (2007). Cognitive functions in patients with MR-defined chronic focal cerebellar lesions. Journal of Neurology, 254(9), 1193-1203. Richter, S., Schoch, B., Kaiser, O., Groetschel, H., Hein-Kropp, C., Maschke, M., et al. (2005). Children and adolescents with chronic cerebellar lesions show no clinically relevant signs of aphasia or neglect. Journal of Neurophysiology, 94(6), 4108-4120. Riecker, A., Kassubek, J., Groschel, K., Grodd, W., & Ackermann, H. (2006). The cerebral control of speech tempo: Opposite relationship between speaking rate and BOLD signal changes at striatal and cerebellar structures. NeuroImage, 29(1), 46-53. Robinson, S. E., & Vrba, J. (1999). Functional neuroimaging by synthetic aperture magnetometry (SAM). Sendai: Tohoku University Press.
197 Rosen, H. J., Petersen, S. E., Linenweber, M. R., Snyder, A. Z., White, D. A., Chapman, L., et al. (2000). Neural correlates of recovery from aphasia after damage to left inferior frontal cortex. Neurology, 55(12), 1883-1894. Sachs, B. C., & Gaillard, W. D. (2003). Organization of language networks in children: Functional magnetic resonance imaging studies. Current Neurology and Neuroscience Reports, 3(2), 157-162. Saffran, J. R., Aslin, R. N., & Newport, E. L. (1996). Statistical learning by 8-month-old infants. Science, 274(5294), 1926-1928. Salmelin, R. (2007). Clinical neurophysiology of language: The MEG approach. Clinical Neurophysiology, 118(2), 237-254. Salmelin, R., Hari, R., Lounasmaa, O. V., & Sams, M. (1994). Dynamics of brain activation during picture naming. Nature, 368(6470), 463-465. Saltzman-Benaiah, J., Scott, K., & Smith, M. L. (2003). Factors associated with atypical speech representation in children with intractable epilepsy. Neuropsychologia, 41(14), 1967-1974. Sambeth, A., Huotilainen, M., Kushnerenko, E., Fellman, V., & Pihko, E. (2006). Newborns discriminate novel from harmonic sounds: A study using magnetoencephalography. Clinical Neurophysiology, 117(3), 496-503. Sandor, S., & Leahy, R. (1997). Surface-based labeling of cortical anatomy using a deformable atlas. IEEE Transactions on Medical Imaging, 16(1), 41-54.
198 Sarkari, S., Simos, P. G., Fletcher, J. M., Castillo, E. M., Breier, J. I., & Papanicolaou, A. C. (2002). Contributions of magnetic source imaging to the understanding of dyslexia. Seminars in Pediatric Neurology, 9(3), 229-238. Sato, S., Balish, M., & Muratore, R. (1991). Principles of magnetoencephalography. Journal of Clinical Neurophysiology, 8(2), 144-156. Satz, P., Strauss, E., Wada, J., & Orsini, D. L. (1988). Some correlates of intra- and interhemispheric speech organization after left focal brain injury. Neuropsychologia, 26(2), 345-350. Saur, D., Lange, R., Baumgaertner, A., Schraknepper, V., Willmes, K., Rijntjes, M., et al. (2006). Dynamics of language reorganization after stroke. Brain, 129(Pt 6), 1371-1384. Schmahmann, J. D., & Sherman, J. C. (1997). Cerebellar cognitive affective syndrome. International Review of Neurobiology, 41, 433-440. Schmahmann, J. D., & Sherman, J. C. (1998). The cerebellar cognitive affective syndrome. Brain, 121, 561-579. Sekihara, K., Nagarajan, S. S., Poeppel, D., Marantz, A., & Miyashita, Y. (2001). Reconstructing spatio-temporal activities of neural sources using an MEG vector beamformer technique. IEEE Transactions on Bio-Medical Engineering, 48(7), 760771. Shapiro, B. E., & Danly, M. (1985). The role of the right hemisphere in the control of speech prosody in propositional and affective contexts. Brain and Language, 25(1), 19-36.
199 Shattuck, D. W., & Leahy, R. M. (2002). BrainSuite: An automated cortical surface identification tool. Medical Image Analysis, 6(2), 129-142. Shattuck, D. W., Sandor-Leahy, S. R., Schaper, K. A., Rottenberg, D. A., & Leahy, R. M. (2001). Magnetic resonance image tissue classification using a partial volume model. NeuroImage, 13(5), 856-876. Signoret, J. L., Castaigne, P., Lhermitte, F., Abelanet, R., & Lavorel, P. (1984). Rediscovery of leborgne's brain: Anatomical description with CT scan. Brain and Language, 22(2), 303-319. Silveri, M. C., Leggio, M. G., & Molinari, M. (1994). The cerebellum contributes to linguistic production: A case of agrammatic speech following a right cerebellar lesion. Neurology, 44(11), 2047-2050. Simos, P. G., Breier, J. I., Zouridakis, G., & Papanicolaou, A. C. (1998). Assessment of functional cerebral laterality for language using magnetoencephalography. Journal of Clinical Neurophysiology, 15(4), 364-372. Simos, P. G., Papanicolaou, A. C., Breier, J. I., Wheless, J. W., Constantinou, J. E., Gormley, W. B., et al. (1999). Localization of language-specific cortex by using magnetic source imaging and electrical stimulation mapping. Journal of Neurosurgery, 91(5), 787-796. Snead, O. C. 3rd. (2001). Surgical treatment of medically refractory epilepsy in childhood. Brain & Development, 23(4), 199-207.
200 Springer, J. A., Binder, J. R., Hammeke, T. A., Swanson, S. J., Frost, J. A., Bellgowan, P. S., et al. (1999). Language dominance in neurologically normal and epilepsy subjects: A functional MRI study. Brain, 122 (Pt 11), 2033-2046. Stark, R. E., Bleile, K., Brandt, J., Freeman, J., & Vining, E. P. (1995). Speech-language outcomes of hemispherectomy in children and young adults. Brain and Language, 51(3), 406-421. Stark, R. E., & McGregor, K. K. (1997). Follow-up study of a right- and a lefthemispherectomized child: Implications for localization and impairment of language in children. Brain and Language, 60(2), 222-242. Staudt, M., Lidzba, K., Grodd, W., Wildgruber, D., Erb, M., & Krageloh-Mann, I. (2002). Right-hemispheric organization of language following early left-sided brain lesions: Functional MRI topography. NeuroImage, 16(4), 954-967. Streeter, L. A. (1976). Language perception of 2-month-old infants shows effects of both innate mechanisms and experience. Nature, 259(5538), 39-41. Struhsaker, T. T. (1968). Auditory communication among vervet monkeys (cercopithecus aethiops). In S. A. Altmann (Ed.), Social Communication Among Primates (pp. 281324). Chicago: University of Chicago Press. Swingley, D. (2005). Statistical clustering and the contents of the infant vocabulary. Cognitive Psychology, 50(1), 86-132.
201 Swingley, D. (2007). Lexical exposure and word-form encoding in 1.5-year-olds. Developmental Psychology, 43(2), 454-464. Szaflarski, J. P., Holland, S. K., Schmithorst, V. J., & Byars, A. W. (2006). fMRI study of language lateralization in children and adults. Human Brain Mapping, 27(3), 202-212. Szaflarski, J. P., Schmithorst, V. J., Altaye, M., Byars, A. W., Ret, J., Plante, E., et al. (2006). A longitudinal functional magnetic resonance imaging study of language development in children 5 to 11 years old. Annals of Neurology, 59(5), 796-807. Thach, W. T., Goodkin, H. P., & Keating, J. G. (1992). The cerebellum and the adaptive coordination of movement. Annual Review of Neuroscience, 15, 403-442. Thiel, A., Habedank, B., Winhuisen, L., Herholz, K., Kessler, J., Haupt, W. F., et al. (2005). Essential language function of the right hemisphere in brain tumor patients. Annals of Neurology, 57(1), 128-131. Thiel, A., Herholz, K., Koyuncu, A., Ghaemi, M., Kracht, L. W., Habedank, B., et al. (2001). Plasticity of language networks in patients with brain tumors: A positron emission tomography activation study. Annals of Neurology, 50(5), 620-629. Trauzettel-Klosinski, S., Durrwachter, U., Klosinski, G., & Braun, C. (2006). Cortical activation during word reading and picture naming in dyslexic and non-readingimpaired children. Clinical Neurophysiology, 117(5), 1085-1097. Trehub, S. E. (1976). The discrimination of foreign speech contrasts by infants and adults. Child Development, 47, 466-472.
202 Turgut, M. (2008). Cerebellar mutism. Journal of Neurosurgery, Pediatrics, 1(3), 262. Van Borsel, J., van der Made, S., & Santens, P. (2003). Thalamic stuttering: A distinct clinical entity? Brain and Language, 85(2), 185-189. van Dongen, H. R., Catsman-Berrevoets, C. E., & van Mourik, M. (1994). The syndrome of 'cerebellar' mutism and subsequent dysarthria. Neurology, 44(11), 2040-2046. Van Lancker Sidtis, D., Pachana, N., Cummings, J. L., & Sidtis, J. J. (2006). Dysprosodic speech following basal ganglia insult: Toward a conceptual framework for the study of the cerebral representation of prosody. Brain and Language, 97(2), 135-153. Van Veen, B. D., van Drongelen, W., Yuchtman, M., & Suzuki, A. (1997). Localization of brain electrical activity via linearly constrained minimum variance spatial filtering. IEEE Transactions on Bio-Medical Engineering, 44(9), 867-880. Vanlancker-Sidtis, D. (2004). When only the right hemisphere is left: Studies in language and communication. Brain and Language, 91(2), 199-211. Vargha-Khadem, F., Watters, G. V., & O'Gorman, A. M. (1985). Development of speech and language following bilateral frontal lesions. Brain and Language, 25(1), 167-183. Vargha-Khadem, F., O'Gorman, A. M., & Watters, G. V. (1985). Aphasia and handedness in relation to hemispheric side, age at injury and severity of cerebral lesion during childhood. Brain, 108 ( Pt 3)(Pt 3), 677-696.
203 Vargha-Khadem, F., Watters, G. V., & O'Gorman, A. M. (1985). Development of speech and language following bilateral frontal lesions. Brain and Language, 25(1), 167-183. Vihla, M., Laine, M., & Salmelin, R. (2006). Cortical dynamics of visual/semantic vs. phonological analysis in picture confrontation. NeuroImage, 33(2), 732-738. Vihman, M. M., Nakai, S., DePaolis, R. A., & Hallé, P. (2004). The role of accentual pattern in early lexical representation. Journal of Memory and Language, 50, 336-353. Vikingstad, E. M., Cao, Y., Thomas, A. J., Johnson, A. F., Malik, G. M., & Welch, K. M. (2000). Language hemispheric dominance in patients with congenital lesions of eloquent brain. Neurosurgery, 47(3), 562-570. von Frisch, K. (1974). Decoding the language of the bee. Science (New York, N.Y.), 185(4152), 663-668. Vouloumanos, A., & Werker, J. F. (2004). Tuned to the signal: The privileged status of speech for young infants. Developmental Science, 7(3), 270-276. Vouloumanos, A., & Werker, J. F. (2007). Listening to language at birth: Evidence for a bias for speech in neonates. Developmental Science, 10(2), 159-164. Vrba, J., & Robinson, S. E. (2001). Signal processing in magnetoencephalography. Methods, 25(2), 249-271. Wada, J. A. (1949). Igaku to seibutsuqaku No. 14)
204 Wada, J. A. (1997). Clinical experimental observations of carotid artery injections of sodium amytal. Brain and Cognition, 33(1), 11-13. Wada, J. A., & Rassmussen, T. (1960). Intracarotid injection of sodium amytal for the lateralization of cerebral speech dominance. Journal of Neurosurgery, 17, 266-282. Werker, J. F. (2003). Baby steps to learning language. The Journal of Pediatrics, 143(4 Suppl), S62-9. Werker, J. F., & Tees, R. C. (1984). Phonemic and phonetic factors in adult cross-language speech perception. The Journal of the Acoustical Society of America, 75(6), 1866-1878. Wernicke, C. (1874). Der aphasische symptomencomplex. Breslau: Cohn and Weigert. Wernicke, C. (1881). Lehrbuch der gehirnkrankheiten fur aerzte und studirende. Kassel: Theodor Fischer. Wheless, J. W., Castillo, E., Maggio, V., Kim, H. L., Breier, J. I., Simos, P. G., et al. (2004). Magnetoencephalography (MEG) and magnetic source imaging (MSI). The Neurologist, 10(3), 138-153. Willmes, K., & Poeck, K. (1993). To what extent can aphasic syndromes be localized? Brain, 116 ( Pt 6)(Pt 6), 1527-1540. Winhuisen, L., Thiel, A., Schumacher, B., Kessler, J., Rudolf, J., Haupt, W. F., et al. (2005). Role of the contralateral inferior frontal gyrus in recovery of language function in
205 poststroke aphasia: A combined repetitive transcranial magnetic stimulation and positron emission tomography study. Stroke, 36(8), 1759-1763. Wood, 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(6), 1035-1044. Woods, R. P., Dodrill, C. B., & Ojemann, G. A. (1988). Brain injury, handedness, and speech lateralization in a series of amobarbital studies. Annals of Neurology, 23(5), 510518. Yamamoto, M., Ukai, S., Shinosaki, K., Ishii, R., Kawaguchi, S., Ogawa, A., et al. (2006). Spatially filtered magnetoencephalographic analysis of cortical oscillatory changes in basic brain rhythms during the japanese 'shiritori' word generation task. Neuropsychobiology, 53(4), 215-222. Yamasue, H., Yamada, H., Yumoto, M., Kamio, S., Kudo, N., Uetsuki, M., et al. (2004). Abnormal association between reduced magnetic mismatch field to speech sounds and smaller left planum temporale volume in schizophrenia. NeuroImage, 22(2), 720-727. Youngjohn, J. R. (1986). Rapid recovery from crossed aphasia in a dextral. Archives of Clinical Neuropsychology, 1(2), 165-174. Yuan, W., Szaflarski, J. P., Schmithorst, V. J., Schapiro, M., Byars, A. W., Strawsburg, R. H., et al. (2006). fMRI shows atypical language lateralization in pediatric epilepsy patients. Epilepsia, 47(3), 593-600.
206 Zangaglia, R., Pacchetti, C., Pasotti, C., Mancini, F., Servello, D., Sinforiani, E., Cristina, S., Sassi, M., & Nappi, G. (2009). Deep brain stimulation and cognitive functions in Parkinson’s disease: A three year controlled study. Movement Disorders, 24(11), 16211628.
