Dyslexia-specific brain activation profile becomes normal following successful remedial training P.G. Simos, PhD J.M. Fletcher, PhD E. Bergman, MD J.I. Breier, PhD B.R. Foorman, PhD E.M. Castillo, PhD R.N. Davis, MA M. Fitzgerald, BA and A.C. Papanicolaou, PhD Abstract���Objectives: To examine changes in the spatiotemporal brain activation profiles associated with successful completion of an intensive intervention program in individual dyslexic children. Methods: The authors obtained magnetic source imaging scans during a pseudoword reading task from eight children (7 to 17 years old) before and after 80 hours of intensive remedial instruction. All children were initially diagnosed with dyslexia, marked by severe difficulties in word recognition and phonologic processing. Eight children who never experienced reading problems were also tested on two occasions separated by a 2-month interval. Results: Before intervention, all children with dyslexia showed distinctly aberrant activation profiles featuring little or no activation of the posterior portion of the superior temporal gyrus (STGp), an area normally involved in phonologic processing, and increased activation of the corresponding right hemisphere area. After intervention that produced significant improvement in reading skills, activity in the left STGp increased by several orders of magnitude in every participant. No systematic changes were obtained in the activation profiles of the children without dyslexia as a function of time. Conclusions: These findings suggest that the deficit in functional brain organization underlying dyslexia can be reversed after sufficiently intense intervention lasting as little as 2 months, and are consistent with current proposals that reading difficulties in many children represent a variation of normal development that can be altered by intensive intervention. NEUROLOGY 2002 58:1203���1213 Dyslexia, a persistent difficulty in acquiring word reading skills, affects a significant proportion of school-aged children and is a serious contributor to academic failure. Dyslexia seems to have a neuro- logic basis, but the precise nature of this impairment is not fully understood. There is agreement among researchers that the core problem in dyslexia is re- lated to a functional impairment within the brain mechanism specialized for language, specifically in the component responsible for phonologic analysis. Reading acquisition in children requires the develop- ment of an appreciation for the segmental nature of speech, a skill known as phonemic awareness. Once the child realizes that spoken words are composed of smaller segments (the phonemes), he or she can learn to treat written words as multisegment units and grasp the correspondence between letters (or let- ter complexes) and phonemes. This concept is known as the alphabetic principle and lies at the heart of teaching programs that focus primarily on the devel- opment of phonemic decoding skills. Many studies have shown that children with dyslexia have poor phonemic awareness skills.1-3 These deficiencies lead to the poor development of word recognition skills.4 Thus, measures of phonemic processing predict later reading achievement5-7 and can be reliably used to identify children with dyslexia.2,3 With the emergence of functional imaging meth- ods that allow for the detection, localization, and quantification of brain activity associated with cogni- tive function, it is possible to assess systematically the putative brain mechanisms underlying dyslexia. Functional brain imaging is well suited for the study of activation profiles peculiar to dyslexia. Functional brain imaging is noninvasive and can be used repeatedly with both dyslexic children and normal control participants in the context of explor- atory studies. Two functional imaging methods have been used with children: fMRI and, most recently, magnetic source imaging (MSI). Like PET, which is used only with adults because it involves injection of radioactive isotopes, fMRI captures blood flow and See also page 1139 From the Vivian L. Smith Center for Neurologic Research, Department of Neurosurgery (Drs. Simos, Breier, Castillo, and Papanicolaou, and R. Davis and M. Fitzgerald), and Department of Pediatrics (Drs. Fletcher and Foorman) University of Texas Health Science Center���Houston Texas Reading Institute (Dr. Bergman), Houston, TX and Department of Psychology (R. Davis), University of Houston, TX. Supported in part by grants from NSF (REC-9979968), National Institute of Neurological Disorders and Stroke (NS37941���01), and NICHD (DA10715) to A.C.P. Received July 27, 2001. Accepted in final form January 27, 2002. Address correspondence and reprint requests to Dr. Panagiotis G. Simos, Department of Neurosurgery, University of Texas���Houston Health Science Center, 6431 Fannin, Suite 7.152, Houston, TX 77030 e-mail: psimos@uth.tmc.edu Copyright �� 2002 by AAN Enterprises, Inc. 1203

local metabolic changes contingent on the differential degree of activation of various brain structures during the performance of specific tasks. MSI, conversely, pro- vides a real-time, spatiotemporal map of brain activity by directly measuring electrical currents in neuronal aggregates during task performance. Because dyslexia is not typically associated with structural brain lesions,8 its mechanism could be a deviant form of functional organization of the brain structures that subserve reading-related functions. Such functional aberrations would be expected to ap- pear as brain activation profiles specific to dyslexic individuals, and therefore different from those of nondyslexic individuals engaged in the same language-processing tasks. Previous research using all three imaging modalities suggest that engage- ment in tasks that require phonologic decoding (such as reading of pseudowords) is associated with in- creased activation in some areas. These include the posterior portion of the superior temporal, the angu- lar, and supramarginal gyri (henceforth collectively referred to as the temporoparietal region), and also the inferior frontal lobe, primarily in the left hemisphere.9-11 In addition, word and pseudoword reading tasks engage areas on the basal surface of the temporal lobe in the vicinity of the lingual and the fusiform gyri12-14 to a greater extent than when nonlinguistic visual stimuli are used. Despite several inconsistencies among studies with respect to the engagement of a particular area in reading,15 the overall consensus is that a network of areas are in- volved in word recognition, each of which may be differentially activated depending on specific task demands.9 PET studies comparing activation profiles of adults with dyslexia to those of nonimpaired readers have found reduced blood flow in the left temporopa- rietal area during performance of reading and pho- nologic processing tasks16-18 but normal activation in the left inferior frontal areas.18,19 In addition, the asymmetry of activity favoring the left hemisphere, usually observed in normal readers during reading tasks, has been found to be significantly attenuated in adults with dyslexia.20 These results are consistent with data from fMRI studies, in which nonimpaired readers as a group demonstrated an incremental acti- vation in temporoparietal areas with increasing de- mands for phonologic analysis. In contrast, impaired readers did not demonstrate this pattern.21,22 In addi- tion, the latter group showed reversed (right left) hemispheric asymmetries of activation in posterior temporal regions when compared with the group of nonimpaired readers. Two main conclusions emerge from these studies. First, left hemisphere temporoparietal areas in chil- dren and adults with dyslexia fail to show the activa- tion seen in nonimpaired readers during engagement in tasks that pose substantial demands for phono- logic analysis. Second, dyslexic readers may rely on the engagement of both inferior frontal (in adult dys- lexics) and right hemisphere temporoparietal areas to a greater extent than nonimpaired readers. Recently, we used MSI to demonstrate the exis- tence of a distinct spatiotemporal profile of brain activation associated with word and pseudoword reading that reliably differentiated between individ- ual children with and without dyslexia. The validity of this procedure for obtaining spatiotemporal pro- files of activation during complex cognitive tasks has been established in clinical studies in which MSI- derived maps demonstrated excellent concordance when compared with invasive brain maps, including electrocortical stimulation mapping and the Wada procedure.23-27 The aberrant profile in children with dyslexia fea- tures predominant activation of the right posterior superior temporal gyrus (STGp), and the right infe- rior parietal region (angular and supramarginal gyri). In contrast, most normal readers display pre- dominant activation of the left STGp and the left inferior parietal region. This activity typically occurs between 300 and 800 ms after stimulus onset (some- times persisting up to approximately 1200 ms), and is preceded by activation of the left lingual and fusi- form gyri, predominantly in the left hemisphere. When the stimuli to be read are meaningful, late activity (300 to 800 ms) is also found in the middle temporal gyrus and mesial temporal cortex in both groups. Although these differences between good and poor readers are found in the context of both word28 and pseudoword29 reading tasks, they are more likely related to the engagement of neurophysiologic pro- cesses involved in phonologic decoding. We have shown previously that decoding is severely disrupted by direct electrical stimulation of the left STGp30 and that facility in phonologic decoding is a major predic- tor of success in reading acquisition.5-7 This region displays normal levels of activity during performance on simple word recognition tasks presented in the auditory modality.28 The discrepancy in the activa- tion profiles between the auditory and the printed word processing tasks points to a functional disrup- tion in the brain circuit that supports reading rather than a functional deficit restricted to the temporopa- rietal region. However, performance of more complex phonologic processing tasks may reveal differences between dyslexic and nonimpaired children with re- spect to the degree of activation of this area.31 Although dyslexia is a chronic reading disorder that may persist into adulthood,32 recent studies have demonstrated that it can be remediated with relatively short periods of intensive remedial instruc- tion.33,34 In agreement with the notion that phonologic processing difficulties form the core characteristic of the most common type of dyslexia, instructional pro- grams that focus primarily on the development of pho- nemic awareness and decoding skills produce the best outcomes. The apparent contradiction between the neurologic hypothesis of dyslexia and the ���malleability��� of the phenotypic profile of the disorder introduce the following possibilities: 1) intervention may be associ- 1204 NEUROLOGY 58 April (2 of 2) 2002