DOI: 10.1212/01.wnl.0000286365.41070.54 2007;69;2146-2154 Neurology

Hackney, D. Alsop, S. Wong and C. A. Walsh B. S. Chang, T. Katzir, T. Liu, K. Corriveau, M. Barzillai, K. A. Apse, A. Bodell, D.

malformationA structural basis for reading fluency: White matter defects in a genetic brain

This information is current as of February 7, 2008

http://www.neurology.org/cgi/content/full/69/23/2146located on the World Wide Web at:

The online version of this article, along with updated information and services, is

All rights reserved. Print ISSN: 0028-3878. Online ISSN: 1526-632X. since 1951, it is now a weekly with 48 issues per year. Copyright © 2007 by AAN Enterprises, Inc.

® is the official journal of the American Academy of Neurology. Published continuouslyNeurology

at Harvard University on February 7, 2008 www.neurology.orgDownloaded from

A structural basis for reading fluencyWhite matter defects in a genetic brain malformation

B.S. Chang, MD*T. Katzir, PhD*T. Liu, PhDK. Corriveau, MEdM. Barzillai, MEdK.A. Apse, ScMA. Bodell, MSD. Hackney, MDD. Alsop, PhDS. Wong, PhDC.A. Walsh, MD, PhD

ABSTRACT

Background: Multiple lines of evidence have suggested that developmental dyslexia may be asso-ciated with abnormalities of neuronal migration or axonal connectivity. In patients with periven-tricular nodular heterotopia—a rare genetic brain malformation characterized by misplacednodules of gray matter along the lateral ventricles—a specific and unexpected reading disability ispresent, despite normal intelligence. We sought to investigate the cognitive and structural brainbases of this phenomenon.

Methods: Ten adult subjects with heterotopia, 10 with dyslexia, and 10 normal controls wereevaluated, using a battery of neuropsychometric measures. White matter integrity and fiber tractorganization were examined in six heterotopia subjects, using diffusion tensor imaging methods.

Results: Subjects with heterotopia and those with developmental dyslexia shared a common be-havioral profile, with specific deficits in reading fluency. Individuals with dyslexia seemed to havea more prominent phonological impairment than heterotopia subjects. Periventricular nodular het-erotopia was associated with specific, focal disruptions in white matter microstructure and orga-nization in the vicinity of gray matter nodules. The degree of white matter integrity correlated withreading fluency in this population.

Conclusions: We demonstrate that a genetic disorder of gray matter heterotopia shares behav-ioral characteristics with developmental dyslexia, and that focal white matter defects in this dis-order may serve as the structural brain basis of this phenomenon. Our findings represent apotential model for the use of developmental brain malformations in the investigation of abnormalcognitive function. Neurology® 2007;69:2146–2154

GLOSSARYDTI � diffusion tensor imaging; FA � fractional anisotropy; PNH � periventricular nodular heterotopia; ROI � region ofinterest.

Dyslexia is one of the most common learning problems in the general population, affect-ing 5% to 17% of children.1 It has typically been defined as a specific, unexpected impair-ment in reading despite adequate intelligence and educational exposure. The core deficitin most individuals with dyslexia seems to be one affecting phonological processing, orthe ability to manipulate the sound segments of words.2 However, many individuals withdyslexia have fluency-based reading problems,3 including problems with rapid letter,word, and sentence reading and the ability to perform other rapid naming tasks, and theconcurrence of both phonological problems and fluency difficulty may be of particularsignificance.4

Multiple functional neuroimaging studies in dyslexic and normal readers have impli-cated a complex brain network for reading that seems to involve mostly the perisylvian

*These authors contributed equally to this work.

From Beth Israel Deaconess Medical Center and Harvard Medical School (B.S.C., K.A.A., A.B., D.H., D.A.), Boston, MA; University ofHaifa (T.K.), Haifa, Israel; Brigham and Women’s Hospital and Harvard Center for Neurodegeneration and Repair (T.L., S.W.), Boston;Harvard Graduate School of Education (K.C.), Cambridge, MA; Center for Reading and Language Research (M.B.), Tufts University,Medford, MA; and Children’s Hospital Boston (C.A.W.), Harvard Medical School, Howard Hughes Medical Institute.

B.S.C. was supported by the National Institute of Neurological Disorders and Stroke of the NIH K23 grant NS049159). B.S.C. and T.K. weresupported by an interfaculty initiative grant from the Mind-Brain-Behavior program of Harvard University. C.A.W. is an Investigator of theHoward Hughes Medical Institute.

Disclosure: The authors report no conflicts of interest.

Address correspondence andreprint requests to Dr. BernardS. Chang, ComprehensiveEpilepsy Center, KS-457, BethIsrael Deaconess MedicalCenter, 330 Brookline Ave,Boston, MA 02215bchang@bidmc.harvard.edu

2146 Copyright © 2007 by AAN Enterprises, Inc. at Harvard University on February 7, 2008 www.neurology.orgDownloaded from

region of the left hemisphere.5,6 However,anatomic neuroimaging studies haveshown mixed results when gray matterstructures in this and other brain regions indyslexic readers are compared with thoseof normal readers.7,8 Instead, several linesof evidence suggest that a functional dis-connection of relevant cortical regions maybe a potential basis for reading difficulties.

For example, postmortem histologic ex-amination of brains from dyslexic individ-uals has demonstrated the presence ofectopic neurons in the molecular layer ofcerebral cortex,9 raising the possibility offaulty neuronal migration and subsequentdisruptions in axonal connectivity. Tworecently identified dyslexia susceptibility

genes, DCDC2 and ROBO1, encode pro-teins that may be involved in axonal path-finding or neuronal migration.10,11

Abnormal patterns of cortical activationseen during PET scans have also suggestedthe possibility of connectivity problems indyslexia.12,13 Finally, diffusion tensor imag-ing (DTI), a noninvasive, MRI-basedmethod that allows for analysis of whitematter microstructure and visualization offiber tracts, has demonstrated a correlationbetween white matter integrity and wordreading ability in children and adults.14-17

We have previously shown that patientswith the genetic brain malformation ofperiventricular nodular heterotopia (PNH)have impaired reading despite normal in-telligence, attention, working memory,and educational exposure.18 PNH, whichcan be associated with mutations in theFLNA gene, is characterized by misplacednodules of gray matter that line the lateralventricles of the brain bilaterally (figure1),19 representing neurons that fail to mi-grate properly from the ventricular zoneduring fetal brain development. Patientstypically present with seizures in adoles-cence, but they generally have no grossneurologic disability,20 despite this remark-able alteration in gray matter architecture.

Here we show that whereas patientswith PNH have very distinctive neuroana-tomic abnormalities, they share commonbehavioral features with individuals withdevelopmental dyslexia, in particular aprominent impairment in reading fluency.Moreover, we demonstrate that this read-ing fluency deficit is associated with whitematter defects adjacent to the periventricu-lar nodules, suggesting a structural corre-late for this particular feature of dyslexia.

METHODS Subjects. Ten PNH subjects were recruitedfrom direct clinician referrals and from a database of sub-jects who had participated in a genetic study of the condi-tion. Subjects were included if they were 18 years or olderand had an MRI diagnosis of PNH based on the presence ofat least one periventricular nodule of gray matter signal in-tensity. All had been diagnosed with PNH after clinicallypresenting with seizures; none was diagnosed or selectedbased on cognitive impairment of any type. Six of these sub-jects had MRI with DTI performed in this study; they didnot differ prominently from the four others in demographic

Figure 1 Brain imaging of subjects with periventricular nodular heterotopia

(A) Axial (left) and coronal (right) T1-weighted anatomic brain images from a subject withperiventricular nodular heterotopia (PNH) demonstrate the presence of misplaced, or hetero-topic, gray matter nodules (black arrows) located bilaterally along the lateral walls of thelateral ventricles, at times forming a confluent mass along the ventricular wall and at timesprotruding into the ventricular lumen. (B) Automated segmentation analysis of gray and whitematter by fractional anisotropy in another PNH subject allows for three-dimensional visual-ization of confluent periventricular heterotopic gray matter (green).

Neurology 69 December 4, 2007 2147 at Harvard University on February 7, 2008 www.neurology.orgDownloaded from

features or heterotopia characteristics, and although theyhad slightly higher intelligence and rapid naming scores,these differences were not significant. Five of the subjects inthis report had results of other behavioral testing describedpreviously.18

Ten dyslexic subjects 18 years or older were recruitedthrough a learning disability center associated with a localuniversity. All spoke English as a first language, were physi-cally healthy, and had no history of psychiatric or neurologicdisorders. All reported a history of reading difficulty thathad been identified in childhood and had required specialliteracy instruction. For confirmation of a persisting readingproblem in this population, subjects were tested using theneuropsychometric battery described below. Those who (1)performed more than 1 SD below the population mean onany measure of word reading (Word ID or Word Attackfrom the Woodcock Reading Mastery Tests–Revised, SightWord Efficiency or Phonetic Decoding Efficiency from theTest of Word Reading Efficiency) and (2) performed morethan 1 SD below their own full-scale IQ on any measure ofword reading, were classified as dyslexic and included in thestudy (see Niogi and McCandliss17 for discussion of similarcriteria).

Ten normal reader subjects 18 years or older were re-cruited through local universities. All spoke English as a firstlanguage, were physically healthy, and had no history of psy-chiatric or neurologic disorders. All denied any history ofreading problems or learning disabilities and were tested us-ing the neuropsychometric battery described below. Thosewho had both verbal and performance IQ scores within thenormal range and who performed within 1 SD of the popula-tion mean on all word reading measures listed above wereclassified as normal readers and included in the study.

All subjects gave informed consent. The study was car-ried out according to research protocols approved by the in-stitutional review board of the Beth Israel DeaconessMedical Center, Boston.

Behavioral testing. Intelligence was measured using theWechsler Adult Intelligence Scale or the Wechsler Abbrevi-ated Scale of Intelligence, which yield verbal, performance,and full-scale standard scores.21,22 A wide range of compo-nent reading skills, word reading skills, and connected textreading skills were assessed. Phonological processing wasevaluated using the Comprehensive Test of PhonologicalProcessing.23 Three tasks of the Rapid Automatized Naming/Rapid Alternating Stimulus test previously linked to dyslexiain older age groups (letters, digits, objects) wereadministered.24-26 Word-level reading was assessed with un-timed real-word and nonword decoding tests (Word ID andWord Attack subtests of Woodcock ReadingMastery Tests–Revised)27 and timed, rapid real-word and nonword decod-ing tests (Sight Word Efficiency and Phonetic DecodingEfficiency subtests of the Test ofWord Reading Efficiency).28

Connected-text reading was assessed using the Gray OralReading Test Third Edition29 and the Passage Comprehen-sion subtest of the Woodcock Reading MasteryTests–Revised.

Neuroimaging. Brain MRI with DTI was performed usinga 3-T GE VH/1 scanner with the product head coil. Ana-tomic images were acquired using a T1-weighted, three-dimensional, magnetization-prepared, rapid-acquisition,gradient-echo (MPRAGE) volume acquisition (TE1 � MIN,TI � 400 milliseconds, flip � 10, FOV � 240 mm, matrix size

256 � 256 voxels, slice thickness � 1.5 mm, no skip). DTIwas performed using a diffusion-weighted, single-shot, spin-echo, echo-planar, imaging sequence (TE1 � MIN, TR �

10000 milliseconds, FOV � 240 mm, matrix size 256 � 256voxels, slice thickness � 2.0 mm, no skip, NEX � 1). Ab-value of 1000 was used, and 15 noncollinear directionswere calculated. The diffusion-weighted sequence was per-formed three times for each subject during a single session.

Regions of interest (ROIs) for fractional anisotropy (FA)analysis of white matter were drawn in two ways on ana-tomic images and FA maps using MRIcro software,30 by twoof the investigators working together. For hemispheric whitematter FA calculation, multiple ROIs were hand drawn bi-laterally in the centrum semiovale, corona radiata, and inter-nal capsule for each subject, carefully excluding gray matterin periventricular nodules as well as regions of cortex andnormal deep gray matter. Hemispheric white matter FA wasthen calculated for each subject by averaging the mean FAsin these ROIs. For an analysis of the effect of heterotopicnodules on overlying focal white matter FA, multiple ROIswere hand drawn for each subject, with half overlying one ormore periventricular nodules and the other half in the ho-mologous contralateral region of white matter not overlyinga nodule. Mean FA from all ROIs overlying nodules wasthen compared with that from all ROIs not overlyingnodules.

Fiber tractography was performed with DTIStudio soft-ware available from Johns Hopkins University, based on thefiber assignment by continuous tracking method.31 Fibertracking was performed from all the voxels within the brain(using a brute-force approach) with both orthograde and ret-rograde initiation along the direction of the principal eigen-vector in each voxel. Propagation in each fiber tract wasterminated if a voxel with FA � 0.4 was reached or if theangle of tract became � 70 degrees during tracking. Resultsthat penetrated manually segmented ROIs were then visual-ized, either in three-dimensional space or in two dimensionscoregistered onto anatomic images.

Statistical analysis. Differences in behavioral scoresamong PNH subjects, dyslexic subjects, and normal readerson the various cognitive and reading tasks were analyzed forsignificance, using a univariate analysis of variance with posthoc comparisons. Pearson correlation coefficients were cal-culated for the relationships between FA measures in PNHsubjects and various behavioral scores. A paired Student ttest was used to compare FA values between ROIs overlyingperiventricular nodules in PNH subjects and homologouscontralateral ROIs not overlying nodules. A significancethreshold of p � 0.05 was used for all comparisons.

RESULTS Phonological, rapid naming, and read-ing skills. Thirty adult subjects from three groups(10 with PNH, 10 with dyslexia, and 10 normalreaders) were evaluated, using a battery of neuro-psychometric measures. Normal readers had ahigher male:female ratio and were younger on av-erage, but these differences were not significant.Verbal, performance, and full-scale IQ scoreswere closely comparable among the three groups(table 1).

Component reading skills, including phono-logical processing and rapid automatized naming,

2148 Neurology 69 December 4, 2007 at Harvard University on February 7, 2008 www.neurology.orgDownloaded from

were assessed. Subjects with PNH and subjectswith dyslexia demonstrated worse phonemic

awareness (as measured by Elision) than normalreaders, and this difference was significant forsubjects with dyslexia (figure 2A). Subjects withPNH and subjects with dyslexia shared a com-mon profile on measures of rapid automatizednaming, demonstrating significantly lower scoresthan normal readers on the letters and digitssubtests; a similar, though nonsignificant, trendwas found on the objects subtest (figure 2B).

At the word reading level, no difference amongthe three groups was found in untimed real-wordand nonword reading tasks (Word ID and WordAttack; figure 2C), whereas subjects with PNHand subjects with dyslexia were significantlyworse than normal readers on measures of timed,rapid real-word and nonword reading (SightWord Efficiency and Phonetic Decoding Efficien-cy; figure 2D).

Passage reading of connected text (Gray OralReading Test) demonstrated a trend in PNH sub-jects and dyslexic subjects toward poor accuracy,rate, and fluency, though this was not significant.Passage comprehension as evaluated by two inde-pendent measures (Gray Oral Reading Test,Woodcock) was intact for PNH subjects and dys-lexic subjects.

White matter integrity and relationship to readingability. Because the reading disability in PNHseemed to be characterized primarily by difficul-ties with fluency and rapid naming, we hypothe-sized that variations in hemispheric white mattermicrostructure in PNH might serve as the neuro-anatomic substrate for this behavioral finding,because acquired white matter abnormalitieshave been associated with reaction time and pro-cessing speed deficits in other populations,32,33

and there is accumulating evidence of white mat-ter microstructural changes in patients withdyslexia.14-17

Figure 2 Performance of subjects with periventricular nodular heterotopia,subjects with dyslexia, and normal readers on word reading andreading-related tasks

(A) Subjects with periventricular nodular heterotopia (PNH) and subjects with dyslexia per-formed worse than normal readers on the Elision subtest of the Comprehensive Test of Pho-nological Processing, a measure of phonemic awareness; this difference was significant forsubjects with dyslexia. (B) Subjects with PNH and subjects with dyslexia showed a similarprofile on tests of rapid automatized naming. Both groups were significantly worse than nor-mal readers on the letters and digits rapid automatized naming subtests. Both groupsshowed a trend toward worse performance on the objects subtest as well, though this wasnot significant. A discrepancy was seen in the performance of PNH subjects and dyslexicsubjects on tasks of word reading, depending on whether the tasks were presented in atime-sensitive manner. Reading of real words and nonwords in an untimed way (C; Word IDand Word Attack subtests of Woodcock Reading Mastery Test–Revised) was performed wellby PNH subjects and dyslexic subjects when compared with normal readers. In contrast,reading of real words and nonwords in a setting in which subjects were asked to perform thetasks as quickly as possible and scaled scores were time sensitive (D; Sight Word Efficiencyand Phonetic Decoding Efficiency subtests of Test of Word Reading Efficiency) revealed thatPNH subjects and dyslexic subjects performed significantly worse than normal readers. As-terisks indicate significance.

Table 1 Characteristics of subjects with periventricular nodular heterotopia, subjects with dyslexia, and normalreaders

Periventricular nodularheterotopia Dyslexic readers Normal readers Statistical comparison

Age, mean (range) 35.1 (19–48) 34.6 (22–47) 25.5 (19–46) F�3.256; p�0.0541

Sex, M:F 2:8 3:7 7:3 p � 0.0698

Handedness, R:L 8:0 6:1 7:1 p � 0.4667

Verbal IQ, mean (SD) 104.0 (15.9) 105.7 (9.6) 108.8 (7.8) F�0.4715;p�0.6291

Performance IQ, mean (SD) 100.5 (7.7) 105.2 (10.4) 109.1 (9.8) F�2.105; p�0.1415

Full-scale IQ, mean (SD) 103.3 (10.9) 105.0 (8.9) 110.4 (6.6) F�1.697; p�0.2022

Univariate analysis of variance was used to analyze differences in mean age, verbal IQ, performance IQ, and full-scale IQamong the three groups. Fisher exact test was used to analyze differences in sex and handedness proportion among the threegroups; in these cases, p values presented are for the comparison between the two extreme proportions. Handedness wasnot recorded for all subjects.

Neurology 69 December 4, 2007 2149 at Harvard University on February 7, 2008 www.neurology.orgDownloaded from

Six PNH subjects underwent magneticresonance-based DTI, and hemispheric whitematter FA, a global measure of white matter in-tegrity derived from the directional diffusivity ofwater, was analyzed by calculating a compositescore across multiple ROIs hand drawn bilater-ally for each subject to exclude the periventricularnodules as well as cortical gray matter and nor-mal deep gray matter. Hemispheric white matterFA was significantly lower in subjects with lowerscores on one measure of reading fluency, rapidnaming of digits (figure 3), and a similar trendwas seen with another fluency measure, rapidnaming of letters (table 2). No relationship was

seen with measures of phonological processing orword reading. FA in the left superior corona ra-diata and left posterior limb of the internal cap-sule, two specific white matter fiber bundlespreviously linked to word reading skills in dys-lexic and normal readers,15,17 showed no relation-ship with measures of reading fluency, wordreading, or phonological processing in PNHsubjects.

The focal effect of gray matter heterotopia on whitematter microstructure. To explore the specific ef-fect of heterotopic gray matter on the integrity ofsubcortical white matter, FA was calculated inROIs drawn in areas of white matter directlyoverlying periventricular heterotopic nodules insubjects with PNH; comparison was then madewith FA from homologous contralateral ROIsthat did not overlie nodules. Asymmetric nodulelocation thus allowed for the use of each subjectas his own control; in this way, normalizing(warping) PNH images for comparison with con-trol subjects was avoided, because that may haveintroduced unexpected distortions caused by het-erotopic gray matter. Mean FAwas lower (0.43 vs0.48; p � 0.03) in the ROIs over nodules than inthe homologous contralateral ROIs not over nod-ules, suggesting focal and specific alterations inwhite matter microstructure over deep hetero-topic gray matter.

Cortico-cortical fiber tract organization. BecauseFA analysis suggested focal white matter alter-ations over nodules, we sought to explore theseregions using fiber tracking methods. Three-dimensional fiber tracts in regions overlying graymatter heterotopia were visualized, using tractog-raphy software. Fiber tracts traveling through theadjacent white matter from and to other regionsof cortex were seen to deviate around the periven-

Figure 3 Correlation of reading fluency withhemispheric white matter integrityin periventricular nodularheterotopia

Scatterplot showing the relationship between performanceon the rapid automatized naming (RAN) digits subtest (scaledscore) and hemispheric white matter fractional anisotropy(FA) for six subjects with periventricular nodular heterotopia.As described in Methods, hemispheric white matter FA wascalculated by averaging the mean FAs across multiple re-gions of interest drawn bilaterally to exclude normal corticaland deep gray matter and heterotopic gray matter nodules.Pearson correlation analysis demonstrated that the relation-ship depicted was significant.

Table 2 Relationship of white matter fractional anisotropy to behavioral measures in periventricular nodularheterotopia

Hemisphericwhite matter FA

Left superiorcorona radiata FA

Left posterior limbof internal capsule FA

r p r p r p

RAN (letters) 0.78 0.07 0.20 0.71 �0.04 0.94

RAN (digits) 0.91* 0.01 0.34 0.51 0.20 0.70

Real word reading (Word ID) 0.28 0.60 �0.29 0.57 �0.63 0.18

Nonword reading (Word Attack) 0.58 0.22 �0.02 0.97 �0.36 0.48

Elision 0.61 0.20 0.08 0.89 �0.25 0.63

Pearson correlation coefficient ( r ) for relationship between FA and behavioral scaled score is indicated in each cell. Two-tailed p values are presented.*Significant.FA � fractional anisotropy; RAN � rapid automatized naming.

2150 Neurology 69 December 4, 2007 at Harvard University on February 7, 2008 www.neurology.orgDownloaded from

tricular nodules, but in general, no fiber tractswere seen emanating from or projecting into thenodules themselves. Strikingly, more superficialsegments of subcortical white matter overlyinggray matter nodules seemed to have a consistentpaucity of well-organized fiber tracts.

Coregistration of visualized fiber tracts ontotwo-dimensional anatomic MRI images for mul-tiple PNH subjects (figures 4 and 5) confirmed thespecificity of white matter disorganization withregard to heterotopic nodule location. Subjectswith posterior heterotopic nodules demonstrateddisorganized white matter over the nodules butnormally organized frontal white matter (figure 4,bottom row), whereas subjects with frontal heter-otopic nodules demonstrated disorganized whitematter anteriorly but normally organized poste-rior white matter (figure 5).

DISCUSSION Dyslexia is a complex, heteroge-neous disorder that affects the capacity of thebrain to acquire one of the most fundamentalskills of modern human communication, the abil-ity to read. We have used the anatomic and be-

havioral study of a genetic brain malformationcharacterized by neuronal migration failure to ex-plore one specific aspect of reading disability—namely, reading fluency. Our work takesadvantage of an unusual opportunity to study thestructural basis of cognitive function: the associa-tion of a well-defined developmental brain mal-formation with a specific behavioral profile.

Most studies of dyslexia have demonstratedthe centrality of a phonological processing defi-cit.34 Although systematic research on the role ofphonological processes in reading failure and in-tervention has proven highly predictive for manyindividuals, the heterogeneity of reading disabili-ties and the complexity of reading breakdownsuggest that a unidimensional view of readingmay be too simplistic. In many subjects with dys-lexia, additional deficits, including sensorimotorproblems or naming speed deficits, are alsopresent.35 Individuals who have a double deficit—both phonological difficulties and naming speedimpairment—may benefit from remediationmethods that are not solely phonologicallybased,36 and reading fluency in general tends notto improve with remediation as much as readingaccuracy does.37

Behavioral profiles can also change over time. Inadults with dyslexia, for example, fundamentalword decoding skills may become quite accurate af-ter years of educational experience and readingpractice, whereas fluency-based deficits can remainand often serve as the solemarkers of reading break-down in this “compensated” population.38 Indeed,the dyslexic adults in our study seem to demonstratea compensated behavioral profile, in which single-word decoding skills are normal but both phonolog-ical skills and naming speed remain impaired whencompared with normal reader controls.

The form of reading disability associated withPNH seems to be mainly fluency based, althoughthere was also a trend toward worse phonologicalskills in this population compared with normalreaders. Other studies have found that a relativelysmall subset of dyslexic individuals as a wholehave pure fluency problems without significantphonological deficits. These individuals do nottypically display decoding problems or a generaldecrease in speed of processing; rather, they showisolated difficulty with the fluent processing ofletters, words, and connected text.4 The behav-ioral profile of our reading-disabled PNH sub-jects is most similar to that of those individuals.

There has been increasing neuroimaging evi-dence that abnormalities in cortico-cortical con-nectivity may be important in many individuals

Figure 4 Specificity of white matter fiber tract disorganization: Normal controland subject with posterior periventricular nodular heterotopia

Images in each row comprise an axial, T1-weighted, structural MRI image (left), visualizedfiber tracts in the right frontal subcortical white matter and genu of the corpus callosum(middle), and visualized fiber tracts in the left posterior white matter and splenium of thecorpus callosum (right). Images in the top row, from a normal control subject without brainmalformation, demonstrate that normal frontal and posterior white matter contains callosal,projection, and intrahemispheric cortico-cortical fibers in organized bundles. Images in thebottom row, from a patient with bilateral posterior periventricular nodular heterotopia, dem-onstrate that the left posterior white matter is abnormally organized: fiber tracts deviatearound the periventricular nodules, and there is a distinct paucity of organized fiber tracts inthe gyri overlying the nodules (white arrowheads). Right frontal white matter appears nor-mally organized. Fiber tracts are depicted only unilaterally in each image, to improve figureclarity.

Neurology 69 December 4, 2007 2151 at Harvard University on February 7, 2008 www.neurology.orgDownloaded from

with dyslexia. A decrease in FA, a measure ofwhite matter integrity, has been reported in thetemporoparietal white matter in a group of adultswith poor reading, and FA in a segment of lefttemporoparietal white matter in particular corre-lates with the degree of word reading ability inadult normal readers and dyslexic readers.14 Simi-lar relationships have been observed between FAin several proximate left hemisphere white matterregions and the reading scores of young children,including normal readers and also those withreading disability.15-17

Isolated fluency deficits, in particular, stronglysuggest the possibility of a functional disconnec-tion of cortical regions involved in reading, be-cause abnormalities in fiber tracts connectingthese areas might be expected to lead to a diffi-culty with the rapid processing and integration ofserial stimuli, whereas the elemental phonologicaland orthographical skills necessary for readingaccuracy might remain relatively unaffected. In-deed, some investigators have found that rapidnaming, a timed measure of automaticity of infor-mation retrieval, correlates with FA, and theyhave suggested that time-sensitive processing may

be facilitated by increased connectivity.16 In addi-tion, faster learning of novel speech sounds is as-sociated with denser parietal white matterstructure, especially in the left hemisphere,39

whereas lower FA in specific brain regions seemsto be associated with slower response times.33 Fi-nally, detailed anatomic studies of dyslexic sub-jects have suggested that abnormalities in afrontal-cerebellar circuit may be particularly rele-vant to those individuals with rapid naming prob-lems and the double-deficit subtype of readingdisability.40 If this is the case, then disruptions ofcortico-cortical fiber tracts along many differentpoints within the cerebral white matter might beexpected to lead to a common behavioral profileof reading fluency impairment.

Our findings strongly support the hypothesisthat disconnection of cortical regions plays a crit-ical role in reading fluency. We had previouslyshown that the degree of reading impairment inPNH subjects is worse in those whose nodules areanatomically more widespread, rather than re-stricted in distribution.18 Our current findingssuggest an explanation: it may be the degree towhich long cortico-cortical fiber tracts are af-fected along their paths that influences the sub-jects’ reading performance, rather than thespecific locations of gray matter nodules them-selves. Indeed, whereas hemispheric white matterFA correlated strongly with performance on read-ing fluency tasks, we found no such correlationusing FA in the left superior corona radiata orposterior limb of the internal capsule—specificwhite matter regions that have been associatedwith word reading ability in other studies.15,17 Al-though our ability to detect such a correlationmay have been limited by sample size and a re-stricted FA range in PNH, these findings suggestthat fluency deficits may arise from more wide-spread white matter disruptions than decodingdeficits, which may localize more precisely to spe-cific regions in the left hemisphere.

There are several limitations to our work. Thesample size is small, although PNH is a rare mal-formation, and our findings are consistent acrosssubjects. A larger number of subjects in a futurestudy might allow for a more refined analysis ofthe effect of nodule location on behavioral mea-sures. Of necessity, our results are only correla-tive in nature. Although it is highly plausible thatthe abnormalities of white matter connectivity inPNH lead to the observed reading fluency prob-lems, it is conceivable that no causal relationshipexists or that years of difficulty with reading leadto structural changes within the cerebral white

Figure 5 Specificity of white matter fiber tract disorganization: Subjects withfrontal periventricular nodular heterotopia

Images in each row comprise an axial, T1-weighted, structural MRI image (left), visualizedfiber tracts in the right frontal subcortical white matter and genu of the corpus callosum(middle), and visualized fiber tracts in the left posterior white matter and splenium of thecorpus callosum (right). Images in the top row, from a subject with a single, large, right frontalperiventricular nodule, demonstrate that the right frontal white matter is abnormally orga-nized: fiber tracts deviate around the large periventricular nodule, and there is a distinctpaucity of organized fiber tracts in the gyri overlying the nodules (white arrowheads). Imagesin the bottom row, from a subject with bilateral frontal heterotopia, demonstrate a similardisorganization of right frontal white matter. In both subjects, left posterior white matterappears normally organized. Fiber tracts are depicted only unilaterally in each image, to im-prove figure clarity.

2152 Neurology 69 December 4, 2007 at Harvard University on February 7, 2008 www.neurology.orgDownloaded from

matter. A longitudinal study of young childrenwith PNH might help to address some of theseissues, although the rarity of the condition andthe usual delay in diagnosis until at least adoles-cence, after the development of seizures, wouldsignificantly limit subject recruitment. All of ourPNH subjects have had seizures, but the lack ofcorrelation between reading impairment and se-verity of epilepsy,18 and the presence of a correla-tion between decreasing reading fluency scoresand decreasing white matter integrity, suggestthat it is the effect of the malformation itself,rather than epilepsy or anticonvulsant use, thatleads to our behavioral findings. Finally, tractog-raphy methods do not allow us to identify the spe-cific anatomic fates of white matter fiber tracts,only their degree of directional organization. Al-though disorganized or circuitous fiber pathwayscould theoretically maintain cortico-cortical con-nectivity, it is reasonable to assume that suchpathways would not be optimally efficient andcould result in functional impairment. Functionalconnectivity might be addressed using functionalMRI to examine patterns of gray matter coactiva-tion during reading-related tasks.

Our findings do raise the possibility that simi-lar brain malformations characterized by neuro-nal migration failure and the presence ofheterotopic gray matter might share common be-havioral features with PNH. However, most suchdisorders are associated with a significant degreeof global cognitive impairment, unlike PNH,making detailed behavioral assessment morechallenging, and potentially limiting the ability toobtain high-quality DTI. There are several poten-tial mechanisms by which neuronal migrationfailure might lead to the observed white mattermicrostructural changes, including disruptions inradial glial cell function or the formation of aber-rant fiber tracts emanating from affected overly-ing cortex, but the investigation of these is likelyto require molecular and microscopic techniquesin animal model systems rather than in vivo hu-man neuroimaging.

Malformations of cortical development in-clude not just disorders of gray matter heteroto-pia but many other conditions in which thenormal process of brain development is disruptedduring embryonic and fetal life.41 Some of theseresult from abnormalities in the proliferation ofneuronal progenitor cells, resulting in too manyor too few mature neurons; others, like PNH, re-sult from abnormalities in the migration of neu-rons from proliferative zones to their ultimatedestinations in the cerebral cortex; and yet others

result from abnormalities in cerebral cortical or-ganization. As advances are made in our under-standing of the molecular and genetic bases ofmany such malformations, it becomes increas-ingly important to understand the cognitive andfunctional consequences of cortical maldevelop-ment, especially as these disorders are being clini-cally diagnosed with increasing frequencybecause of the widespread use of brain MRI.42

Historically, much of what we know aboutbrain–behavior relationships has come from thestudy of acquired lesions affecting the maturebrain. More recently, functional neuroimaginghas provided the capability to study the anatomicbasis of physiologic behavior noninvasively innormal subjects. Our work, and that of others,43,44

supports the longstanding conception that devel-opmental brain malformations can serve as an al-ternative model for our understanding of brain–behavior relationships, one that may beparticularly relevant to the study of neurodevel-opmental disabilities such as dyslexia.

ACKNOWLEDGMENTThe authors thank all the subjects who participated, without whomthis study could not have been completed.

Received May 2, 2007. Accepted in final form July 20, 2007.

REFERENCES1. Shaywitz SE. Current concepts: dyslexia. N Engl J Med

1998;338:307–312.2. Habib M. The neurological basis of developmental

dyslexia: an overview and working hypothesis. Brain2000;123:2373–2399.

3. Katzir T, Wolf M, O’Brien B, et al. Reading fluency:the whole is more than the parts. Ann Dyslexia 2006;56:51–83.

4. Wolf M, Bowers PG. The double-deficit hypothesis forthe developmental dyslexias. J Educ Psychol 1999;91:1–24.

5. Demonet JF, Taylor MJ, Chaix Y. Developmental dys-lexia. Lancet 2004;363:1451–1460.

6. Pugh KR, Mencl WE, Jenner AR, et al. Functional neu-roimaging studies of reading and reading disability (de-velopmental dyslexia). Ment Retard Dev Disabil ResRev 2000;6:207–213.

7. Leonard C, EckertM, Lombardino L, et al. Anatomicalrisk factors for phonological dyslexia. Cereb Cortex2001;11:148–157.

8. Eckert M. Neuroanatomical markers for dyslexia: a re-view of dyslexia structural imaging studies. Neurosci-entist 2004;10:362–371.

9. Galaburda AM, Sherman GF, Rosen GD, Aboitiz F,Geschwind N. Developmental dyslexia: four consecu-tive patients with cortical anomalies. Ann Neurol 1985;18:222–233.

10. Hannula-Jouppi K, Kaminen-Ahola N, Taipale M, etal. The axon guidance receptor gene ROBO1 is a can-didate gene for developmental dyslexia. PLoS Genet2005;1:e50.

Neurology 69 December 4, 2007 2153 at Harvard University on February 7, 2008 www.neurology.orgDownloaded from

11. Meng H, Smith SD, Hager K, et al. DCDC2 is associ-ated with reading disability and modulates neuronaldevelopment in the brain. Proc Natl Acad Sci USA2005;102:17053–17058.

12. Horwitz B, Rumsey JM, Donohue BC. Functional con-nectivity of the angular gyrus in normal reading anddyslexia. Proc Natl Acad Sci USA 1998;95:8939–8944.

13. Paulesu E, Frith U, Snowling M, et al. Is developmentaldyslexia a disconnection syndrome? Evidence fromPET scanning. Brain 1996;119:143–157.

14. Klingberg T, Hedehus M, Temple E, et al. Microstruc-ture of temporo-parietal white matter as a basis forreading ability: evidence from diffusion tensor mag-netic resonance imaging. Neuron 2000;25:493–500.

15. Beaulieu C, Plewes C, Paulson LA, et al. Imaging brainconnectivity in children with diverse reading ability.NeuroImage 2005;25:1266–1271.

16. Deutsch GK, Dougherty RF, Bammer R, Siok WT, Gab-rieli JD, Wandell B. Children’s reading performance iscorrelated with white matter structure measured by diffu-sion tensor imaging. Cortex 2005;41:354–363.

17. Niogi SN, McCandliss BD. Left lateralized white mat-ter microstructure accounts for individual differencesin reading ability and disability. Neuropsychologia2006;44:2178–2188.

18. Chang BS, Ly J, Appignani B, et al. Reading impairmentin the neuronalmigration disorder of periventricular nod-ular heterotopia. Neurology 2005;64:799–803.

19. Fox JW, Lamperti ED, Eksioglu YZ, et al. Mutationsin filamin 1 prevent migration of cerebral cortical neu-rons in human periventricular heterotopia. Neuron1998;21:1315–1325.

20. d’Orsi G, Tinuper P, Bisulli F, et al. Clinical featuresand long term outcome of epilepsy in periventricularnodular heterotopia. Simple compared with plusforms. J Neurol Neurosurg Psychiatry 2004;75:873–878.

21. Wechsler D. Wechsler Adult Intelligence Scale—3rdEdition (WAIS-3). San Antonio, TX: Harcourt Assess-ment Inc; 1997.

22. Wechsler D.Wechsler Abbreviated Scale of Intelligence(WASI). San Antonio, TX: Harcourt Assessment Inc;1999.

23. Wagner RK, Torgesen JK, Rashotte CA. Comprehen-sive Test of Phonological Processing. Austin, TX:Pro-Ed Inc; 1999.

24. Wolf M, Denckla MB. Rapid Automatized Namingand Rapid Alternating Stimulus Tests (RAN/RAS).Austin, TX: Pro-Ed Inc; 2005.

25. Cardoso-Martins C, Pennington BF. The relationship be-tween phoneme awareness and rapid serial naming skillsand literacy acquisition: the role of developmental periodand reading ability. Sci Stud Read 2004;8:27–52.

26. Semrud-Clikeman M, Guy K, Griffin JD. Rapid nam-ing deficits in children and adolescents with readingdisabilities and attention deficit hyperactivity disor-ders. Brain Lang 2000;74:70–83.

27. Woodcock R. Woodcock Reading Mastery Tests—Revised. Circle Pines, MN: American Guidance Ser-vice; 1987.

28. Torgesen JK, Wagner RK, Rashotte CA. Test of WordReading Efficiency (TOWRE). Austin, TX: Pro-Ed Inc;2001.

29. Wiederholt J, Bryant B. Gray Oral Reading Test(GORT-3). 3rd ed. Odessa, FL: Psychological Assess-ment Resources; 1992.

30. Rorden C, Brett M. Stereotaxic display of brain le-sions. Behav Neurol 2000;12:191–200.

31. Jiang H, van Zijl PC, Kim J, Pearlson GD, Mori S.DtiStudio: resource program for diffusion tensor com-putation and fiber bundle tracking. Comput MethodsPrograms Biomed 2006;81:106–116.

32. Felmingham KL, Baguley IJ, Green AM. Effects of dif-fuse axonal injury on speed of information processingfollowing severe traumatic brain injury. Neuropsy-chology 2004;18:564–571.

33. Madden DJ, Whiting WL, Huettel SA, White LE, Mac-Fall JR, Provenzale JM. Diffusion tensor imaging ofadult age differences in cerebral white matter: relationto response time. NeuroImage 2004;21:1174–1181.

34. Wagner RK, Torgesen JK. The nature of phonologicalprocessing and its causal role in the acquisition of read-ing skills. Psychol Bull 1987;101:192–212.

35. Ramus F, Rosen S, Dakin SC, et al. Theories of devel-opmental dyslexia: insights from a multiple case studyof dyslexic adults. Brain 2003;126:841–865.

36. Wolf M,Miller L, Donnelly K. Retrieval, automaticity,vocabulary elaboration, orthography (RAVE-O): acomprehensive fluency-based reading intervention pro-gram. J Learn Disabil 2000;33:322–324.

37. Alexander AW, Slinger-Constant A. Current status oftreatments for dyslexia: critical review. J Child Neurol2004;19:744–758.

38. Shaywitz SE, Shaywitz BA, Fulbright RK, et al. Neuralsystems for compensation and persistence: young adultoutcome of childhood reading disability. Biol Psychia-try 2003;54:25–33.

39. Golestani N, Paus T, Zatorre RJ. Anatomical corre-lates of learning novel speech sounds. Neuron 2002;35:997–1010.

40. Eckert MA, Leonard CM, Richards TL, Aylward EH,Thomson J, Berninger VW. Anatomical correlates ofdyslexia: frontal and cerebellar findings. Brain 2003;126:482–494.

41. Barkovich AJ, Kuzniecky RI, Jackson GD, Guerrini R,Dobyns WB. A developmental and genetic classifica-tion for malformations of cortical development. Neu-rology 2005;65:1873–1887.

42. Sisodiya SM. Malformations of cortical development:burdens and insights from important causes of epi-lepsy. Lancet Neurol 2004;3:29–38.

43. Jansen AC, Leonard G, Bastos AC, et al. Cognitivefunctioning in bilateral perisylvian polymicrogyria(BPP): clinical and radiological correlations. EpilepsyBehav 2005;6:393–404.

44. Janszky J, Ebner A, Kruse B, et al. Functional organiza-tion of the brain with malformations of cortical devel-opment. Ann Neurol 2003;53:759–767.

2154 Neurology 69 December 4, 2007 at Harvard University on February 7, 2008 www.neurology.orgDownloaded from

DOI: 10.1212/01.wnl.0000286365.41070.54 2007;69;2146-2154 Neurology

Hackney, D. Alsop, S. Wong and C. A. Walsh B. S. Chang, T. Katzir, T. Liu, K. Corriveau, M. Barzillai, K. A. Apse, A. Bodell, D.

malformationA structural basis for reading fluency: White matter defects in a genetic brain

This information is current as of February 7, 2008

& ServicesUpdated Information

http://www.neurology.org/cgi/content/full/69/23/2146including high-resolution figures, can be found at:

Subspecialty Collections

http://www.neurology.org/cgi/collection/all_genetics All Genetics

http://www.neurology.org/cgi/collection/developmental_disorders Developmental disorders sment

http://www.neurology.org/cgi/collection/neuropsychological_asses Neuropsychological assessment

http://www.neurology.org/cgi/collection/dyslexia Dyslexia http://www.neurology.org/cgi/collection/mri

MRIfollowing collection(s): This article, along with others on similar topics, appears in the

Permissions & Licensing

http://www.neurology.org/misc/Permissions.shtmlor in its entirety can be found online at: Information about reproducing this article in parts (figures, tables)

Reprints http://www.neurology.org/misc/reprints.shtml

Information about ordering reprints can be found online:

at Harvard University on February 7, 2008 www.neurology.orgDownloaded from