The Graphemic/Motor Frontal Area Exner s Area Revisited

The Graphemic/Motor Frontal Area Exner s Area Revisited Franck-Emmanuel Roux, MD, PhD, 1,2 Olivier Dufor, PhD, 1 Carlo Giussani, MD, 1,2 Yannick Wamain, MSc, 3 Louisa Draper, MB, BS, MA, 2 Marieke Longcamp, PhD 3 and Jean-François Démonet, MD, PhD 1,2 Objective: In 1881, Exner first described a graphic motor image center in the middle frontal gyrus. Current psycholinguistic models of handwriting involve the conversion of abstract, orthographic representations into motor representations before a sequence of appropriate hand movements is produced. Direct cortical stimulation and functional magnetic resonance imaging (fmri) were used to study the human frontal areas involved in writing. Methods: Cortical electrical stimulation mapping was used intraoperatively in 12 patients during the removal of brain tumors to identify the areas involved in oral language (sentence reading and naming) and writing, and to spare them during surgery. The fmri activation experiment involved 12 right-handed and 12 left-handed healthy volunteers using word dictation (without visual control) and 2 control tasks. Results: Direct cortical electrical stimulation of restricted areas rostral to the primary motor hand area (Brodmann area [BA] 6) impaired handwriting in 6 patients, without disturbing hand movements or oral language tasks. In 6 other patients, stimulation of lower frontal regions showed deficits combining handwriting with other language tasks. fmri also revealed selective activation during word handwriting in left versus right BA6 depending on handedness. This area was anatomically matched to those areas that affected handwriting on electrical stimulation. Interpretation: An area in middle frontal gyrus (BA6) that we have termed the graphemic/motor frontal area supports bridging between orthography and motor programs specific to handwriting. Ann Neurol 2009;66:537 545 Agraphia, the loss or impairment of the ability to produce writing, is a neurological condition known to result from damage to the left parietal and frontal lobes. 1 Agraphia is a broad term; symptoms are diverse and may be observed in isolation, or associated with oral language impairments. 2 Although the role of the left inferior parietal cortex in writing has been demonstrated, 3,4 controversies persist regarding the existence of a more specific handwriting site in the frontal lobe. 1,5,6 Indeed, since the theory of a handwriting center was formulated by Sigmund Exner 7 in 1881 ( graphic motor image center, located in the foot of the middle frontal gyrus), only a few cases have been published to support this hypothesis. 8 10 To identify the frontal cortical areas involved in language and to spare them during surgery, cortical electrical stimulation mapping was used intraoperatively in 12 patients during the removal of brain tumors. Direct electrostimulation gave us the opportunity to study the neural basis of handwriting and other oral language tasks. This mapping procedure has become a standard clinical practice for the resection of brain tumors in language-related regions, allowing the surgeon to identify and spare functional areas. Stimulationinduced impairment of language function indicates that the area beneath the electrode is involved in the function in question, for example, picture naming or reading tasks. 11 13 We then assessed whether results of an fmri activation experiment involving healthy volunteers (12 right-handed, 12 left-handed) and handwriting tasks would parallel findings from electrical stimulation mapping. Material and Methods In this study, 12 patients underwent surgical resection of brain tumors in our Neurosurgery Department (Table 1). All patients were French speakers and fulfilled the following criteria: 1) had no language deficit preoperatively; and 2) had a lesion located in the left frontal region, or the homologous right-sided territory in 1 left-hander. All the patients and their families gave their informed consent to study their language areas by direct brain mapping. Handedness was as- From 1 INSERM, Imagerie cérébrale et handicaps neurologiques, University of Toulouse, 2 Pole Neurosciences, Centres Hospitalo- Universitaires, and 3 LAPMA, Paul Sabatier University, Toulouse, France. Address correspondence to Dr Roux, INSERM 825 et Service de Neurochirurgie, Hôpital Purpan, F-31059 Toulouse. France. E-mail: franck_emmanuel.roux@yahoo.fr Potential conflict of interest: Nothing to report. Received Sep 18, 2008, and in revised form Jun 16, 2009. Accepted for publication Jul 10, 2009. Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ ana.21804 2009 American Neurological Association 537

Table 1. Demographics and Topography of Explored Brain Regions in the 12 Patients Patient Gender/Age, y/occupation/handedness (Edinburgh Score) Frontal Brain Regions Tested by Direct Stimulation a C.V. M/25/student/LH ( 100) b Superior and middle right frontal gyri D.O. M/40/computer engineer/rh ( 79) Superior and middle left frontal gyri V.I.C. F/41/teacher/RH ( 92) Superior and middle left frontal gyri P.V. M/46/teacher/RH ( 100) Superior and middle left frontal gyri C.J.B. M/19/student/RH ( 69) Superior and middle left frontal gyri C.P. M/44/engineer/RH ( 75) Superior and middle left frontal gyri B.D. M/35/shopkeeper/RH ( 100) Middle and inferior left frontal gyri G.C. F/49/actress/RH ( 100) Middle and inferior left frontal gyri R.J. F/42/housewife/RH ( 82) Superior and middle left frontal gyri A.L. F/ 39/shopkeeper/RH ( 100) Middle and inferior left frontal gyri P.E. M/33/engineer/LH ( 78) Superior and middle left frontal gyri J.G. F/32/driver/RH ( 67%) Middle and inferior left frontal gyri a All patients also had their precentral gyrus studied. b Family of left-handers (mother and uncle also left-handed). M male; LH left-handed; RH right-handed; F female. sessed in all subjects using the Edinburgh Inventory. 14 Preand postoperative language-standardized tests were given for visual naming 15 and written and oral understanding, oral fluency, reading, dictation, repetition, written transcription, and object handling. 16 Writing was primarily evaluated by copying (isolated letters, words, and numbers) and shortsentence dictation tasks. Numbers of errors and handwriting calligraphy were scored and compared pre- and postoperatively. Cortical Mapping Procedure The cortex was directly stimulated using the bipolar electrode of the Nimbus cortical stimulator (1mm-wide electrodes separated by 6mm: Newmedic, Toulouse, France). The current amplitude was started at 2mA, and progressively increased by 1mA, using biphasic square wave pulses of 1 millisecond at 60Hz. Stimulation was guided by a neuronavigational system, with 3-dimensional reconstructions of the brain (Stealth Station, Sofamor Danek, Surgical Navigation Technologies, Broomfield, CO) to localize the Rolandic sulcus and the precentral or frontal gyrus. Care was taken to avoid electrical diffusion and afterdischarges by stimulating under the level of stimulation generating afterdischarges. Therefore, intraoperative cortical stimulation was used to localize the areas of the functional cortex after determination of the afterdischarge threshold by electrocorticography. Following identification of the primary motor areas (corresponding to the hand, face, or foot) with direct electrostimulation, language tasks, including handwriting, were then used to map premotor areas. The number of cortical stimulation sites varied from patient to patient depending on the size of craniotomy. The sites of cortical stimulation were about 1cm apart. The patients performed 2 initial oral language tasks: a visual naming task using drawings of various objects, and a reading task using unrelated and unrehearsed simple sentences. 13 These 2 tasks were used as control tasks. The patients were then asked to perform a handwriting task; another set of simple sentences was dictated, which the patient was asked to write with their preferred hand. Each stimulation site was tested systematically with these 3 tasks. Sites showing no reproducible language interference were not included in this study. For the writing task, the dictation of a sentence to be written was commenced, and then direct stimulation was applied while the patient was writing. An assistant alerted the surgeon if any performance impairments were induced by stimulation. To be validated as a language site, each stimulated site was tested at least 3 times for each task. When a functional site was found, it was marked by a sterile ticket of 0.25cm 2. Intraoperative photographs of the brain were taken showing the validated sites, and the whole procedure was systematically recorded on video. Although this mapping process aims to spare language areas during tumor removal, occasionally it was necessary to resect a functional site for oncological reasons. Functional Magnetic Resonance Imaging Procedure Functional magnetic resonance imaging (fmri) was used in 12 right-handed (mean age, 22 years; standard deviation [SD], 3.88; handedness, 63%; SD, 19.18) and 12 lefthanded volunteers (mean age, 25 years; SD, 5.05; handedness, 69%; SD, 18.91) with no neurological diseases. fmri block-design sessions were performed to study language function, using word dictation (without visual control) and repetition tasks, and alternating activation and resting blocks. Two control tasks were used to match the sensory input and motor output of either language tasks while being devoid of linguistic content; on hearing GO, instruction subjects either drew circles or repeated the syllable /pra/. In each group of participants, the contrast dictation repetition exclusively 538 Annals of Neurology Vol 66 No 4 October 2009

Fig 1. Direct brain mapping findings. Twelve patients were studied (mean age, 37 years; range, 19 49 years, 7 men, 5 women; 10 right-handed, 2 left-handed). For each patient, the stimulation sites are shown on a photograph of the surgical field and on a 3-dimensional reconstruction of brain volume (oblique view, azimuth 270 in all cases except patient C.V. 90, elevation 45 ). The upper line shows results from the 6 patients who showed pure agraphia symptoms during stimulation. These sites are represented by blue dots. White dots represent hand motor sites (ie, sites where stimulation elicits hand contractions). Green dots represent sites where stimulation elicited ocular deviations (frontal eye field). The lower line summarizes findings in the other 6 patients, who had agraphia associated with impairments of oral language tasks: either reading aloud and naming (purple dots) or naming only (yellow dots). masked by conditions Circle and Syllable (threshold p 0.001) was used to isolate the brain substrates of word handwriting relative to word repetition. In addition, null conjunction analyses were conducted in each group to reveal activation common to handwriting with the preferred hand and the nonpreferred hand for the dictation repetition contrast. Acquisitions were performed on a Siemens (Erlangen, Germany) Magnetom Vision (1.5 Tesla) at the Neuroradiology Department of Toulouse Purpan Hospital. Functional scans were acquired after sagittal localization images with a single-shot echo-planar gradient-echo pulse sequence (repetition time 3,500 milliseconds, echo time 60 milliseconds, flip angle 90, field of view 240 mm). The 16 axial slices covering the whole brain were aligned with the anterior commissure posterior commissure line and acquired following an interleaved mode in a 4 4 matrix 64 64 with a resolution of 3.75 3.75mm in plane and 5mm between planes in each volume. Stimuli were frequent words delivered binaurally by a personal computer using Presentation (Neurobehavioral Systems, Albany, CA) software during silent periods of the imager with Resonance Technology (Northridge, CA) headphones (4 blocks each involving 10 items). Words were presented in a different order for the 2 tasks. fmri images were analyzed with Matlab 7 (Math- Works, Natick, MA) and the Statistical Parametric Mapping software SPM5 (Wellcome Department of Imaging Neuroscience, London, UK, www.fil.ion.ucl.ac.uk). Results Figure 1 summarizes the results of language mapping by direct stimulation obtained in the 12 patients. A specific effect on handwriting was observed for stimulation of restricted cortical areas located rostrally to the left-sided primary motor hand area in 5 right-handed (D.O., C.I.V., P.V., C.J.B., C.P.) and one left-handed (C.V.) patients. After mapping the primary motor areas, including those controlling the right hand, mapping of language areas was then performed in the middle and superior frontal gyri. Stimulation of restricted areas in the middle frontal gyrus (for patients D.O., C.I.V., P.V. and C.P.) and in the superior frontal gyrus (for patient C.J.B.) resulted in interference with writing, but did not have any effect on other language tests (picture naming, sentence reading). No motor hand contractions or ocular movements were seen during stimulation in premotor areas inducing these writing interferences. Pure agraphia symptoms induced by stimulation were characterized by poor grapheme production, slow writing, or writing arrest. No electrical diffusion was detected with electrocorticography. For oncologic reasons, these writing areas had to be partially removed in patients D.O. and P.V. Interestingly, these 2 patients had significant handwriting difficulties postoperatively (Fig 2). Patient C.V. (left-handed) had a small low-grade tumor (oligodendroglioma, grade II for World Health Organization II) located in the posterior part of the right superior frontal gyrus. Hand contraction was seen during stimulation of the right precentral gyrus. Lefthand writing was impaired while stimulating a premotor site in the junction of the superior and middle frontal gyri. The patient stopped writing under stimulation and resumed immediately when stimulation was Roux et al: Graphemic/Motor Frontal Area 539

Fig 2. Evolution of handwriting before and after operation. This figure demonstrates the handwriting impairment seen in 2 patients (P.V. and D.O.) in whom the location of the tumor meant that resection necessitated partial removal of the writing area. On the left, examples of their handwriting preoperatively, and on the right, postoperative handwriting after partial removal of the writing area. removed. This area was spared during tumor resection. No picture naming, reading interferences, or hand contraction were identified in this premotor area. In 6 other patients, stimulation of the premotor cortex elicited combined language impairments. In 5 patients (B.D., R.J., G.C., P.E., J.G.), writing disturbances were associated with reading and/or naming impairments during stimulation of restricted areas located in lower parts of the left premotor cortex, including BAs 44/45 (Broca s area) as well as in the left superior frontal gyrus in 1 case (P.E., left-hander). In most of these cases, patients stopped writing sharply on stimulation, although perseverations were seen in 1 case (J.G.). Once stimulation was removed, patients resumed writing after a few seconds. Finally, in only 1 patient (A.L.) out of 12, writing and reading were unaffected during the whole stimulation session, although 2 sites of oral naming interference were identified in the middle frontal gyrus. Figure 3A shows stimulation sites positioned in the standard normalized Montreal Neurological Institute (MNI) space (listed in Table 2). Pure agraphia sites (blue dots) were located in a premotor region corresponding to BA6 and close to the superior frontal sulcus. Sites where stimulations induced combined language symptoms (purple dots) or disorders of oral language tasks only (yellow dots) were located in lower portions of this region (with the exception of PE, a left-handed patient in whom stimulation in the superior frontal gyrus affected the 3 tasks). Figure 3B shows, for the masked dictation repetition fmri contrast in right-handed healthy participants, highly significant (false discovery rate [FDR] 0.05) activation in a premotor region close to the area determined by the cortical stimulation, that is, the left superior frontal sulcus (BA6). In left-handers, the same analysis revealed activation in the very same area. In both groups, activations were also seen in the hand motor cortex homolateral to the writing hand, as well as, depending on the group, in a few other premotor and parietal clusters. Figure 3C shows that, in right-handers, conjunct effects of handwriting (dictation repetition) with preferred and with nonpreferred hand were colocalized, again in the left superior frontal sulcus (BA6), at a conservative threshold (FDR 0.05). Results were less significant in left-handers (threshold p 0.001 uncorrected). Whereas relative to results shown in 3B, the conjunction in right-handers strongly concurred to indicate a left-sided dominance of BA6 engagement, in left-handers, effects were less marked and shifted to the right hemisphere, that is, the driving hemisphere for hand movements. In Figure 4, based on single-subject analyses among left-handers, we identified a subject (M.S.) showing an exclusively right-sided premotor activation very close to the stimulation site found in C.V. (distance 7.5mm), the left-handed surgical case operated on in the right hemisphere. 540 Annals of Neurology Vol 66 No 4 October 2009

Fig 3. Normalized stimulation sites and functional magnetic resonance imaging (fmri) results. (A) Three-dimensional (3D) brain oblique view (azimuth 270, elevation 45 ) showing stimulation sites positioned in the standard normalized Montreal Neurological Institute (MNI) space. Each 3D brain volume was normalized in the MNI space, and parameters were used to obtain normalized coordinates from stimulation site locations, which were perioperatively visualized and positioned on 3D original images provided by a neuronavigation software (Medtronic, Minneapolis, MN). Each subject corresponds to a group of dots linked together by white segments. Color codes for dots are as detailed in Figure 1. Please note that stimulation sites from 11 patients are grouped together on the left hemisphere. Only patient C.V., who is left-handed, was operated on in the upper part of the right hemisphere. (B) 3D brain oblique views (azimuth 270, elevation 45 ) and transverse section (z 46) showing fmri results of word dictation repetition contrast masked by Circle and Syllable tasks in right-handers (RH) and left-handers (LH). In right-handed subjects (hot colors), the activation cluster specific to word handwriting lays in the F1/F2 sulcus (Brodmann area 6) bilaterally with peaks at Talairach coordinates x, y, z 26, 6, 42 (z 3.49) on the left, and x, y, z 26, 2, 46 (Z 3.42 ) on the right (false discovery rate p 0.05). In left-handed subjects (cold colors), a similar cluster was found in the same left F1/F2 (BA6) region with a peak at Talairach coordinates x, y, z 26, 2, 46 (z 3.05) (false discovery rate p 0.05). (C) 3D brain oblique views (left, azimuth 270, elevation 45, right, azimuth 90, elevation 45 ) displaying results of null conjunctions between preferred and nonpreferred handwriting dictation repetition contrasts in right-handers (left, yellow-red) and in left-handers (right, pink-purple).

Table 2. Intraoperative Talairach Coordinates of Sites Showing Interference on Stimulation in the Explored Group of 12 Patients Type of Interference Subject Talairach Coordinates Anatomical Area/BA x y z Hand A.L. 36 17 65 Precentral gyrus/ba6 movements A.L. 45 17 60 Postcentral gyrus/ba3 C.V. 17 a 14 71 Superior frontal gyrus/ba6 C.V. 23 a 6 67 Superior frontal gyrus/ba6 D.O. 41 16 62 Precentral gyrus/ba6 V.I.C. 43 18 62 Precentral gyrus/ba4 V.I.C. 38 20 66 Precentral gyrus/ba6-4 Barycenter 34.78 15.56 64.60 Precentral gyrus/ba6 Writing C.J.B. 23 0 68 Superior frontal gyrus/ba6 C.V. 13 a 1 68 Superior frontal gyrus/ba6 D.O. 37 15 54 Middle frontal gyrus/ba8-6 D.O. 37 16 54 Middle frontal gyrus/ba8-6 V.I.C. 43 2 60 Precentral gyrus/ba6 V.P. 38 4 54 Middle frontal gyrus/ba6 C.P. 35 10 57 Middle frontal gyrus/ba6 Barycenter 32.24 5.35 59.47 Middle frontal gyrus/ba6 Reading & A.L. 38 2 58 Middle frontal gyrus/ba6 naming A.L. 38 14 53 Middle frontal gyrus/ba6 B.D. 50 14 49 Middle frontal gyrus/ba6 G.C. 64 24 22 Inferior frontal gyrus/ba45 P.E. 40 15 50 Middle frontal gyrus/ba6 Barycenter 46.16 14.02 46.21 Middle frontal gyrus/ba6 Writing, B.D. 57 13 39 Middle frontal gyrus/ba8 reading, & naming G.C. 66 21 24 Inferior frontal gyrus/ba45 P.E. 29 2 64 Superior frontal gyrus/ba6 R.J. 49 7 49 Middle frontal gyrus/ba6 J.G. 59 16 9 Inferior frontal gyrus/ba44 Barycenter 52.19 11.76 37.11 Middle frontal gyrus/ba9 Eye C.J.B. 25 1 64 Superior frontal gyrus/ba6 movements R.J. 49 8 49 Middle frontal gyrus/ba6 V.P. 41 0 49 Middle frontal gyrus/ba6 Barycenter 38.61 3.35 54.17 Middle frontal gyrus/ba6 a C.V. (left-handed patient) coordinates along the x axis have been shifted into the left hemisphere to calculate the barycenter. BA Brodmann area. Discussion Frontal lobe lesions very rarely produce pure agraphia symptoms (writing impairment with preserved oral language). More often, agraphia seen in frontal lobe lesions is an aspect of nonfluent aphasia syndromes featuring letter omissions and substitutions, anomia, and agrammatic symptoms. 2,17 However, some results of fmri or positron emission tomography experiments in healthy subjects showed specific activation in the anterior part of the superior parietal lobe and the posterior part of the superior and middle frontal gyri during handwriting tasks 5,6,18 or observation of handwriting stimuli. 19 Here, the converging results from 2 different experiments and methods (fmri and electrical cortical stimulation) demonstrate that an area located close to the superior frontal sulcus in BA6, anterior to the hand primary motor area, is selectively involved in the handwritten production of words. However, further experiments are needed to confirm such a selectivity relative 542 Annals of Neurology Vol 66 No 4 October 2009

Fig 4. Among left-handed subjects, 1 case (M.S.) was isolated to show the variability of activation localization in this group and its closeness (M.S. activation peak: Talairach coordinates x, y, z 11, 7, 64; z 3.62) to the cortical stimulation site (x, y, z 13, 1, 68) recorded in the left-handed surgical case C.V. (Euclidian distance 7.5mm). to other functions, for example, calculation. 20 In addition, our results showed no handwriting-specific effects in 6 patients, and these negative findings might be the consequence of a functional reorganization induced by the presence of tumor. The specificity of stimulation effects in the 6 patients showing pure agraphia symptoms during dictation tasks is based on the absence of impairment of both elementary motor function and the 2 oral language tasks. Coordinates of pure agraphia sites and the activated cluster in the left BA6 in healthy right-handers overlapped; Euclidian distance between the barycenter of stimulation sites and activation cluster peak was on average 18mm, the difference being mainly accounted for by the location of the activation peaks in the depth of the superior frontal sulcus (averaged x coordinate 25mm) while electrical stimuli were applied on the cortical surface (x coordinate 32mm). In addition, fmri signal increase was selective to word handwriting, because dictation was contrasted with a repetition task involving the same words, and exclusive masks (involving circle tracing and syllable repetition, respectively) were used to exclude the influence of changes in cortical activity relating to elementary sensorimotor processes. These convergent results led us to term this area of the superior frontal sulcus the graphemic/motor frontal area (GMFA). This area might stand for the neural counterpart of the interface between the orthographic, abstract representations of words, which involve sequences of graphemes, and motor programs that make it possible to further transcode such graphemes and groups of graphemes into allographic letter strings. The allographic stage of handwriting results in a number of specifications concerning graphomotor shapes; not only does it preserve the orthographic content to be produced, but it also involves graphic styles (eg, lower versus upper case, cursive versus script) and fine-grained dynamic characteristics that make each subject s handwritten production unique. Such a grapheme/motor transcoding stage has long been supposed in cognitive models of spelling 21 and more precisely specified in the model of Van Galen for handwriting. 22 While supporting this transcoding process, it is likely that the GMFA further sends information downward to hand-specific premotor/motor areas in charge of allographic motor specifications, ultimately resulting in the command of appropriate finger and hand movements. Even the primary motor cortex seems to play a role in complex movement execution. 23 Partial removal of GMFA can affect easiness and fluidity of handwriting as shown in our 2 patients, D.O. and P.V., in whom surgical intervention resulted in handwriting that was not only slowed, but also showed profound and persistent changes in the subject s script style. Interestingly, fmri experiment revealed that the handwriting-related activation in BA6 was actually bilateral in our right-handed subjects (although slightly preponderant in the left hemisphere). Previously, it has been suggested that complex movements performed with the right hand by right-handers elicit marked inhibition effects in the premotor/motor areas of the right hemisphere due to a transcallosal modulation exerted by the active left premotor area; the reciprocal phenomenon for movements performed by the left hand seemed to be less pronounced. 24 One interpretation of these findings could be that the GMFA territory is bilateral in right-handers. This opinion is in accord with studies showing bilateral synchronization of neural activities in motor regions of both hemispheres during unimanual self-paced finger movements. 25 However, repetitive transcranial magnetic stimulation experiments have suggested that stimulation of the motor cortex in 1 hemisphere may elicit transcallosal inhibitory effects in the homologous contralateral region, 26,27 especially in right-handed subjects undergoing stimulation of the left motor cortex. 28 Here, since fmri blood oxygen level-dependent signal changes cannot distinguish between the effects of excitatory and inhibitory neural activities, one could speculate alternatively that the activation observed in the right BA6 in our group of right-handed subjects relates to inhibitory activity sent from the left BA6 to the homologous right-sided region. Conjunctions between handwriting conditions involving the preferred and nonpreferred hands, as well as the results obtained in left-handed subjects, shed additional light on hemispheric dominance effects in these premotor areas. In right-handed subjects, the conjunction showed that in the left GMFA the same amount of activation was elicited for writing with the nonpreferred hand as with the preferred one. This result strongly reinforces our assumption that this restricted portion of the left superior premotor cortex has a special role at the interface between orthographic and Roux et al: Graphemic/Motor Frontal Area 543

graphomotor codes during written language production. Moreover, this result suggests a marked decrease of activation in the right GMFA in the nonpreferred hand condition; this region, which was strongly activated in the preferred hand condition, did not show any effect in the conjunction. In our group of lefthanders, word dictation against repetition revealed significant activation in a restricted area overlapping with, and at the same (stringent) threshold as, that found in the right-handers. This finding suggests that GMFA is located in the left hemisphere in most left-handed subjects, a majority of whom present with a left-sided hemispheric dominance for language. However, the overall pattern of results in the left-hander group brought up more distributed and complex and less marked local changes as well as more interindividual variability 29 than in the right-hander group. The preferred/nonpreferred conjunction showed effects that were less significant and right-sided. Effects of complex motor tasks in the premotor cortex homolateral to the moving hand have been suggested to be less marked in left-handers than in right-handers. 30 The absence of effect in conjunction analysis in the left GMFA suggests that its role in our left-hander group might be less straightforward than in the right-handers; reciprocally, its right-sided homolog might have a more prominent function. This hypothesis is reinforced by individual findings implicating the right GMFA in some of our left-handed subjects explored either with fmri (Fig 4) or with electrical stimulation (patient C.V., a lefthanded subject and the only one in our surgical group operated on in the right hemisphere). Surgical results also contributed to show across-subject variability and atypical language substrates in left-handed subjects, as the other left-handed subject in our 12 patients, P.E., showed errors in handwriting, reading, and naming on stimulation of the left superior frontal gyrus. Cortical stimulation affecting naming and/or reading are represented by yellow dots in Figures 1 and 3A. The location of these sites tends to be more ventral than handwriting-specific sites; this finding might receive a somatotopic interpretation. Indeed, language tasks involving spoken output were linked to premotor areas (z coordinate of the barycenter of the yellow dots 46.21) located in front of the mouth area in the primary motor strip in the same way as GMFA (z coordinate of the barycenter of the blue dots 59.47) was located in front of the hand area. Finally, results from our surgical study also showed that some stimulation sites affected all language tasks, whether involving oral or written output (Figs 1 and 3A, purple dots). Apart from patient P.E., this effect was observed in sites that were the most ventrally located in the explored portions of the frontal cortex; these sites were localized in BAs 44 and 45 (Broca s area) in 3 out of 5 cases. This finding is in accordance with neuropsychological observations of stroke-induced agraphia showing that impairments of higher-order control of language processing relate to lesions of BA44/45, whereas motor deficits affecting letter shape (eg, lower/upper case errors) corresponded to damage to BA6. 31 Although the role of BA44/45 in processing abstract language representations including orthographic ones is well known, this differential topography of effects further suggests that dorsally located regions in upper premotor cortex (BA6) harbor the interface between language representations and motor programs from which allographic handwriting movements originate. Results from direct cortical stimulation and fmri concur to show that a subpart of the upper premotor cortex, the GMFA, transforms orthographic code in graphic traces unique to each individual although legible for anyone. The confirmation of Exner s proposal raises a number of issues, however, regarding the precise functions and topography of the GMFA along with its functional relationships with other areas involved in writing such as the parietal cortex and the primary motor cortex,which were implicated in our experiment as well as in previous studies. Further studies using transcranial magnetic stimulation and effective connectivity in neuroimaging data should help to elucidate whether, in most subjects, a left hemispheric GMFA inhibits its right-sided homolog. Lefthandedness also yields an interesting case as to whether orthographic code versus its graphomotor counterparts are transferred via the corpus callosum from the dominant left hemisphere to the right-sided premotor/motor areas. Finally and crucially, there should be further investigation into how a portion of the hand-related premotor cortex can specialize in programming language-related graphomotor movements through the general process of learning to read and write. References 1. Roeltgen DP. Agraphia. In: Heilman KM, Valenstein E, eds. Clinical Neuropsychology. 4th ed. New York, NY: Oxford University Press; 2003:126 145. 2. Benson DF, Cummings JL. Agraphia. In Frederiks JAM, ed. Handbook of Clinical Neurology. Amsterdam, The Netherlands: Elsevier; 1985:457 472. 3. Basso A, Taborelli A, Vignolo LA. Dissociated disorders of speaking and writing in aphasia. J Neurol Neurosurg Psychiatry 1978;41:556 563. 4. Roeltgen DP, Heilman KM. Lexical agraphia: further support for the two-system hypothesis of linguistic agraphia. Brain 1984;107:811 827. 5. Katanoda K, Yoshikawa K, Sugishita M. A functional MRI study on the neural substrates for writing. Hum Brain Mapp 2001;13:34 42. 6. Sugihara G, Kaminaga T, Sugishita M. Interindividual uniformity and variety of the Writing center: a functional MRI study. Neuroimage 2006;32:1837 1849. 544 Annals of Neurology Vol 66 No 4 October 2009

7. Exner S. Untersuchungen über die lokalisation der Functionen in der Grosshirnrinde des Menschen. Vienna, Austria: Wilhelm Braunmüller; 1881. 8. Gordinier HC. A case of brain tumor at the base of the second left frontal convolution. Am J Med Sci 1899;117:526 535. 9. Mahoudeau D, David M, Lecoeur J. Un nouveau cas d agraphie sans aphasie révélatrice d une tumeur métastatique du pied de la deuxième circonvolution frontale gauche. Rev Neurol (Paris) 1951;84:159 161. 10. Rapcsak SZ, Arthur SA, Rubens AB. Lexical agraphia from focal lesion of the precentral gyrus. Neurology 1988;38: 1119 1123. 11. Ojemann G, Ojemann J, Lettich E, Berger M. Cortical language localization in left, dominant hemisphere: an electrical stimulation mapping investigation in 117 patients. J Neurosurg 1989;71:316 326. 12. Lubrano V, Roux FE, Demonet JF. Writing-specific sites in frontal areas: a cortical stimulation study. J Neurosurg 2004; 101:787 798. 13. Roux FE, Lubrano V, Lauwers-Cances V, et al. Intra-operative mapping of cortical areas involved in reading in mono- and bilingual patients. Brain 2004;127:1796 1810. 14. Oldfield R. The assessment and analysis of handedness: the Edinburgh Inventory. Neuropsychologia 1971;9:97 113. 15. Deloche G, Hannequin D. Test de dénomination orale de 80 images: DO 80. Paris, France: ECPA; 1997. 16. Nespoulous JL, Joanette Y, Roch-Lecours A, et al. Protocole Montreal-Toulouse d examen linguistique de l aphasie (MT86). Isbergues, France: L Ortho-Edition. 1992 17. Marcie P, Hecaen H. Agraphia: writing disorders associated with unilateral cortical lesions. In: Heilman KM, Valenstein E, eds. Clinical Neuropsychology. 1st ed. New York, NY: Oxford University Press; 1979. 18. Tokunaga H, Nishikawa T, Ikejiri Y, et al. Different neural substrates for Kanji and Kana writing: a PET study. Neuroreport 1999;10:3315 3319. 19. Longcamp M, Anton JL, Roth M, et al. Visual presentation of single letters activates a premotor area involved in writing. Neuroimage 2003;19:1492 1500. 20. Roux FE, Boukhatem L, Draper L, Sacko O, Démonet JF. Cortical calculation localization using electrostimulation. J Neurosurg 2009;110:1291 1299. 21. Ellis AW, Young AW, Anderson C. Modes of word recognition in the left and right cerebral hemispheres. Brain Lang 1988;35: 254 273. 22. Van Galen GP. Handwriting: issues for a psychomotor theory. Hum Mov Sci 1991;10:165 191. 23. Gerloff C, Corwell B, Chen R, et al. The role of the human motor cortex in the control of complex and simple finger mouvement sequences. Brain 1998;121:1695 1709. 24. Ziemann U, Hallett M. Hemispheric asymmetry of ipsilateral motor cortex activation during unimanual motor tasks: further evidence for motor dominance. Clin Neurophysiol 2001;112: 107 113. 25. Gerloff C, Richard J, Hadley J, et al. Functional coupling and regional activation of human cortical motor areas during simple, internally paced and externally paced finger movements. Brain 1998;121:1513 1531. 26. Ferbert A, Priori A, Rothwell JC, et al. Interhemispheric inhibition of the human motor cortex. J Physiol 1992;453: 525 546. 27. Wassermann EM, Wedegaertner FR, Ziemann U, et al. Crossed reduction of human motor cortex excitability by 1-Hz transcranial magnetic stimulation. Neurosci Lett 1998;250:141 144. 28. Gilio F, Rizzo V, Siebner HR, et al. Effects on the right motor hand-area excitability produced by low-frequency rtms over human contralateral homologous cortex. J Physiol 2003;551: 563 573. 29. Josse G, Herve PY, Crivello F, et al. Hemispheric specialization for language: brain volume matters. Brain Res 2006;1068: 184 193. 30. Verstynen T, Diedrichsen J, Albert N, et al. Ipsilateral motor cortex activity during unimanual hand movements relates to task complexity. J Neurophysiol 2005;93:1209 1222. 31. Hillis AE, Chang S, Breese E, et al. The crucial role of posterior frontal regions in modality specific components of the spelling process. Neurocase 2004;10:175 187. Roux et al: Graphemic/Motor Frontal Area 545