Brain Morphology in Developmental Dyslexia and Attention Deficit Disorder/Hyperactivity George W. Hynd, EdD; Margaret Semrud-Clikeman, PhD; Alison \s=b\ This study examined the specificity of deviations in patterns of normal brain asymmetry on the magnetic resonance imaging scans of 10 dyslexic, 10 attention deficit disorder/hyperactivity (ADD/H), and 10 normal age- and sex-matched control children. Reliabilities of region of interest measurements for left and right anterior and posterior width and area, length of the bilateral insular region, and length of the bilateral planum temporale were excellent. Both the dyslexic and ADD/H children had significantly smaller right anterior-width measurements than did normal subjects. The dyslexics also had a bilaterally smaller insular region and significantly smaller left planum temporale than did the normal subjects. Seventy percent of the normal and ADD/H children had the expected left greater than right pattern of plana asymmetry, while only 10% of the dyslexic children did. The very significant increase in the incidence of plana symmetry or reversed asymmetry seems unique to dyslexia and may be related to deviations in normal patterns of corticogenesis. Although significantly more dyslexic children were left-handed than were the normal and ADD/H children, no significant relationship emerged between left-handedness, incidence of allergies or familial autoimmune disease, and variability in indexes of brain morphologic findings. (Arch Neurol. 1990;47:919-926)
Accepted for publication January 16, 1990. From the Departments of Special Education (Drs Hynd, Semrud-Clikeman, and Lorys) and Psychology (Dr Hynd), University of Georgia, Athens; Department of Neurology, Medical College of Georgia, Augusta (Dr Hynd); and Athens (Ga) Magnetic Imaging (Dr Novey and Ms Eliopulos). Reprint requests to Department of Special Education, Aderhold Hall, University of Georgia, Athens, GA 30602 (Dr Hynd).
Approximately
"^
R.
Lorys, PhD; Edward S. Novey, MD; Deborah Eliopulos, RT
3% to 6% of all
school-age children are believed to suffer from developmental dyslex¬ ia,1-2 which has been defined as a "rare but definable and diagnosable form of primary reading retardation with some
form of central
nervous
system
dysfunction."3 A diagnosis of dyslexia typically requires normal intelligence, a significant discrepancy between measured ability and reading achieve¬ ment, and the absence of other handi¬
capping conditions that may produce reading delay. While it is reasonably well docu¬ mented that some dyslexic children may variably suffer from deficits in visuoperceptual processes,4 sequenc¬ ing ability,5 phonemic segmentation,6 and automatized cognitive processing,7 it has historically been assumed that these deficits result from a disruption in underlying neurologic systems.1 Be¬ havioral evidence provides support for what has been referred to as the Wernicke-Geschwind model.8 This model implicates the involvement in reading of the bilateral posterior cortex; region of the angular gyrus; Wernicke's re¬
gion, including the superior temporal and insular regions; and Broca's area. Topographic mapping of brain electri¬ cal activity provides some support for
this model and advances evidence that the region of the supplementary motor area may be involved as well.9 How¬ ever, despite support for this neuro¬ logic model of reading, the behavioral evidence for the presumed neurologic origins of dyslexia has remained only correlative, until recently.10·11 Neuroimaging and postmortem studies comprise more direct evidence as to how neurologic substrata differ in
Downloaded From: http://archneur.jamanetwork.com/ by a Western University User on 06/08/2015
the brains of
dyslexies. Based on neurobiological theory that implicates the importance of the central language centers,1215 computed tomographic (CT)/magnetic resonance imaging (MRI) studies have provided evidence
that ties deviations in normal patterns of posterior asymmetry (left greater than right [L > R]) to the dyslexic
syndrome.1620
Both CT17 and postmortem stud¬ ies21'25 document that about 66% of normal brains are asymmetric (L > R), favoring the left planum tem¬ porale and posterior region. Other asymmetries also exist. For example, in 75% of brains the volume of the right frontal region exceeds that of the left,24 the left anterior speech region (pars opercularis and pars triangularis of the third frontal convolution) is larger than is the right,26 and the left auditory cortex27 and posterior thala¬ mus28 show similar (L > R) asymme¬ tries. Dyslexies would appear to have a higher incidence of symmetrical or re¬ versed posterior asymmetry than is found in control populations. While approximately two thirds of control brains show L > R posterior asym¬ metry,17 only 10% to 50% of dyslexic brains show this pattern.16·19·20 Cytoarchitectonic studies also support the finding of a higher-than-normal inci¬ dence of plana symmetry and docu¬ ment the presence of numerous focal dysplasias preferentially involving the left frontal, left perisylvian, and right frontal regions.29·30 The CT/MRI and postmortem studies, while few, would seem to implicate some deviation in normal patterns of neuronal migra¬ tion and maturation during the fifth to
seventh month of fetal gestation.1315·30
Neurobiological theory also predicts
that there is an interaction between deviations in normal patterns of brain morphologic appearance in dyslexia, the co-occurrence of left-handedness, and an increased incidence of allergies and familial autoimmune disease.31 While studies consistently document a significantly greater incidence of lefthandedness in populations of dys¬ lexies,1618 little direct evidence exists that ties the co-occurrence of auto¬ immune disease to left-handedness among dyslexies.32 No evidence attests to the interaction between deviations in brain morphologic findings in dys¬ lexia, the presence of autoimmune dis¬ ease, and left-handedness in dyslexia. While supportive of the neurobio¬ logical model of dyslexia, the neuroimaging and postmortem studies are characterized by serious méthodologie flaws. A critical review of these studies suggests that the neuroimaging stud¬ ies employed poorly documented dys¬ lexic populations, included question¬ able control populations, failed to in¬ clude populations with other clinical syndromes in order to document the uniqueness of deviations in patterns of brain asymmetry in dyslexies, and failed to examine the relationship be¬ tween left-handedness and autoim¬ mune disease as they relate to patterns of brain morphologic findings in dyslexia.33 The postmortem studies are similarly characterized by méthod¬
ologie problems.33
In this context then, the purpose of this study was to determine whether deviations in patterns of normal brain asymmetry characterize the brains of dyslexies. By including a clinic control population with attention deficit disorder/hyperactivity (ADD/H) with¬ out significant learning problems, the relative uniqueness of symmetry or reversed asymmetry in the region of the plana in dyslexia could be deter¬ mined. Since the neurologic basis of ADD/H presumably involves frontal dysfunction in the caudostriatal region34·35 and not the left central lan¬ guage zones, it might be expected that only the dyslexies would show differ¬ ent patterns of plana morphologic findings. Data were also obtained for the right and left anterior area and width, insular region, and posterior area and width. A second aim of this study was to improve on the diagnostic procedures employed in deriving the dyslexic and ADD/H groups such that the fre¬ quency and kind of Diagnostic and Statistical Manual of Mental Disor¬ ders, Third Edition (DSM III) codiag-
could be determined. This was viewed as important, since a recent federal reformulation of the definition of learning disabilities noted the fre¬ quent co-occurrence of learning dis¬ abilities and the attention deficit disorders.36 Finally, no neuroimaging study of patterns of brain morphologic findings in dyslexies has employed a computer program that allows for metric mea¬ surements to be made on the MRI scan itself. In using such a program, reli¬ ability of region of interest (ROI) mea¬ surements could be obtained. This was viewed as important, especially con¬ sidering the well-documented difficul¬ ties in measuring the region of the noses
planum temporale.37
SUBJECTS AND METHODS
Subjects The subjects who participated in this study were either outpatients of the Center for Clinical and Developmental Neuropsychology (CCDN) at the University of Geor¬ gia, Athens, or were normal control sub¬ jects solicited from the community. The CCDN serves as a diagnostic and referral facility for northeast Georgia, and refer¬ rals came primarily from area physicians, other clinics, or school districts. The 10 de¬ velopmental dyslexic and 10 ADD/H chil¬ dren represented consecutive referrals ex¬ cluding those patients who received other primary DSM III or Diagnostic and Statis¬ tical Manual of Mental Disorders, Revised Third Edition (DSM III-R) diagnoses or those children who were diagnosed as hav¬ ing mild mental retardation (Full Scale IQ [FSIQ] 85), a positive family history for learning problems, personal his¬ tory of difficulty learning to read, reading achievement significantly (> 20 standard score points) below FSIQ on both the Word Attack and Passage Comprehension subtests of the WRMT-R, and no behaviorally reported symptoms of hyperactivity. While some reported symptoms of inatten¬ tion and impulsivity were permitted among the dyslexies, a co-occurring diagnosis of ADD/H or ADHD excluded participation in this study. Diagnosis of ADD/H required average or better intellectual ability (FSIQ > 85); no reported family history of learning prob¬ lems; no significant deficit in reported or measured achievement; documented behav¬ ioral deficits in attention, impulsivity, and motor activity (hyperactivity) consistent with a DSM III ADD/H diagnosis; and a favorable response to stimulant medica¬ tion. Children diagnosed as normal were administered the same comprehensive evaluation procedures and were required to have average or better intellectual ability (FSIQ > 85); no reported family history of learning problems; no significant deficit in achievement; and no reported or observed medical, educational, social, or emotional difficulties. Diagnosis was based on similar proce¬ dures employed in other clinics.44 Diagnos¬ tic decisions were reached separately by two psychologists after considering all rel¬ evant historical, behavioral, and psycho¬ metric data. Diagnostic decisions were re¬ quired to reflect the above-noted criteria for inclusion in this study as well as the crite¬ ria outlined in DSMIIIor DSM III-R for the
Table
1—Subject and Psychometric Data* Developmental Dyslexies,
Attention Deficit Disorder
Normal,
Hyperactivity
xxx
(SD)
(SD)
(40.43) (8.43)
(SD)
Variable
Chronologic
age,
_
mo
Handedness, % right No. of codiagnoses No. of reported allergies/ autoimmune disease WISC-R IQs Verbal Full Scale WRMT-R (Reading Word Attack
120.60
141.20
96
96 (6.99)
(24.55) (88.00)
4/0.9
3/1.2 107.00
(11.20)
107.70 (10.89) 108.00 (8.68)
Performance
Achievement) 73.80(14.91)
Passage Comprehension
(24.07)
118.90
38
75.00(17.17)
(15.80) 110.80(13.97) 109.30(12.27)
122.80(9.96) 124.80 (15.74) 125.40 (10.91)
106.50
96.30
(16.47)
115.60
(9.97)
99.40
(10.78)
112.30
(12.34)
*
WISC-R indicates: Wechsler Intelligence Scale for Children-Revised; WRMT-R, Woodcock Reading Mastery 10 Tests-Revised. Word Attack and Passage Comprehension test scores are reported in standard scores. for all groups. =
other psychiatric disorders. Disagreement in diagnosis by the two psychologists was resolved through mutual discussion. Reli¬ abilities for this diagnostic process have been reported elsewhere44 and meet ac¬ cepted criteria for reliability in clinical
diagnosis.45
MRI Protocol
Once a child was entered into the study, sequential T, sagittal and axial MRI planes were obtained using a 0.6-T Technicare scanner (Health Images; Atlanta, Ga). A carefully selected protocol involving fifteen 7.5-mm sagittal planes (TR 690; TE 32) and eleven 5-mm axial planes (TR 500, TE 32) was used. Region of =
our knowledge. Recognizing the significant variability in brain morphologic findings and in recognition of the fact that some of the ROI may not be easily judged,3710 MRI scans were randomly selected and an inde¬ pendent measure obtained for each ROI. The initial measurements were made by an experienced radiological technologist trained specifically to identify the ROI. The reliability check was completed by a pedi¬ atrie neuropsychologist with expertise in developmental neuroanatomy. Reliability coefficients were excellent (x 0.95) across =
the 13 ROI measurements. These data reported in Table 2.
=
=
obtained.
MRI Quantification and
All MRI studies
gist
RESULTS
=
interest measurements were obtained us¬ ing the Technicare ROI measurement soft¬ ware system. Figure 1 shows how left and right anterior-width measurements were
as
are
not having
a
Reliability
by a neurolo¬ pathologic appearance.
were
read
Careful documentation of the neuroanatomic relationship between CT asymme¬ tries and measurements made post mortem reveal that the axial slice transversing the region of the planum temporale (including, in part, the supramarginal and angular gy¬ ri) is the only slice that correlates signifi¬ cantly in this regard.46 Consequently, an axial slice transversing this region was se¬ lected for obtaining the ROI measurements for the width and area of the left and right anterior (from a line drawn horizontally across the tip of the genu) and posterior (from a line drawn horizontally across the posterior tip of the splenium) regions, length of the left and right insular regions, and for total brain area. Extreme lateral sagittal slices were employed in obtaining the left and right planum temporale length measurements.
Reliability of ROI measurements ob¬ tained from MRI scans of the children's brains has not previously been reported, to
order [DSM III-R]). No significant dif¬ ferences (P > .05) existed in terms of the incidence of allergies among the subjects or in terms of the reported incidence of autoimmune disease in the families of these subjects. Overall, however, none of these measures sig¬ nificantly correlated with ROI data and were, therefore, not employed as covariates in the following analyses. Means and SD for the ROI measure¬ ments are presented in Table 2. An initial analysis revealed that there were no significant differences (P > .05) in total brain area between the dyslexic, ADD/H, and normal chil¬ dren. Therefore, it seems reasonable that any resulting differences in pat¬ terns of hemispheric asymmetry or symmetry between these groups may be associated with regional variation in neurologic development and not due to gross differences in brain morpho¬
logic findings. A 2 (left-right) X 3 (group) analysis of variance (ANOVA) indicated signif¬ icant (F [2,24] 6.00, .05) from ei¬ ther the dyslexic or normal children. It can be concluded that dyslexic children have a bilaterally shorter insular re¬ gion than do normal children. An ANOVA for between-group dif¬ ferences revealed a significant interences
=
While there were no significant (P > .05) differences in chronologic age across groups, there was a significant difference (^[2,27] 8.66, < .001) in general intellectual ability (FSIQ), with the normal children having a sig¬ =
nificantly higher
FSIQ_ (x
=
125.90)
than either the dyslexic (x 108.00) or ADD/H (x 109.30) children. Based on data from the Edinburgh Inven¬ tory,40 three dyslexic subjects were judged to be left-handed, while none of the ADD/H or normal children were identified as left-handed. A significant < .025) difference (F [2,27] 4.21, was noted between groups on the lat¬ erally quotient due to the three left¬ handers in the dyslexic group. There was also a significant differ¬ ence in the number ( 2[2] 11.10, < .004) and kind of codiagnoses be¬ tween the groups. Three dyslexies re¬ ceived DSM III codiagnoses (overanx¬ ious disorder, major depressive epi¬ sode, and ADD/WO), while seven codiagnoses were made with the ADD/ H children (three conduct disorder, one overanxious, one separation anxi¬ ety, and two oppositional defiant dis=
=
=
=
Downloaded From: http://archneur.jamanetwork.com/ by a Western University User on 06/08/2015
=
=
=
=
action between right and left length and group (F [2,24]
plana =
4.82,
.05). Tukey's pairwise compari¬ indicated that the dyslexic group differed significantly (q 5.92, < .05) from the normal group in left planum temporale length (dyslexies, 1.20 cm; normal subjects, 1.52 cm). No significant comparisons resulted when length of the right planum temporale
R patterns. The right and left measurements were con¬ sidered to differ in direction (L < R, L > R) if the magnitude of the differ¬ ence was
greater than 0.1
cm.
This
method is similar to other reports.33 As can be seen in Fig 5, a significant (x2[2] 9.60, < .008) effect resulted. In comparison with the normal and =
ADD/H children who evidenced the
typical L > R pattern (70% L > R; 30% L < R), fully 90% of the dyslexic children had either symmet¬ rical or reversed asymmetry (L < R) of plana length. Assuming that 30% L < R plana length morphology is nor¬ mal, the 90% incidence of this pattern in the dyslexic children represents a very significant departure from nor¬ more
mal patterns of differential brain mat¬ uration. When differences in posterior area and width measurements were exam¬
significant (P > .05) group, hemispheric, or interaction effects were found. Consequently, when em¬ ploying measurements of the posterior region from a line drawn horizontally across the left and right region later¬ ally from the tip of the splenium, no ined,
no
patterns of asymmetry fects
emerged.
or
group ef¬
COMMENT
Our results document the unique¬ of an increased incidence of
ness
L < R plana length symmetry/ reversed asymmetry in the brains of dyslexic children, in comparison with carefully diagnosed normal age- and sex-matched control children and a clinic control group of ADD/H chil¬ dren. The fact that there were no sig¬ nificant differences between these groups in total brain area suggests that this regional variation in brain morphologic characteristics is due to a more specific deviation in brain ontog¬ eny.
The finding that 70% of the normal and ADD/H children had L > R pat¬ terns of asymmetry in the region of the plana supports similar findings by other investigators employing CT/ MRI.17·20·33 This figure also correlates well with the postmortem findings that 65% of normal adult brains and 79% of infant brains are larger on the left side in the region of the planum temporale.21·25 The 90% incidence of L < R in the dyslexies in the length of the planum temporale is a very signif¬ icant increase over normal base rates.
Fig 1.—Magnetic resonance imaging (MRI) scan of a boy with developmental dyslexia showing reversed pattern (left greater than right) of frontal width asymmetry (top, left and right) (left frontal width, 5.4 cm; right frontal width, 5.2 cm) contrasted with a normal subject's MRI scan (bottom, left and right) showing the more typical pattern (left less than right) of frontal width asymmetry (left frontal width, 5.1 cm; right frontal width, 5.6 cm).
Downloaded From: http://archneur.jamanetwork.com/ by a Western University User on 06/08/2015
Further, our results indicate that it was primarily the significantly smaller left planum temporale that contributed to the higher incidence of L < R patterns in the dyslexies. The finding of a significantly smaller than normal left planum temporale in the
dyslexies is at variance with the bilaterally large plana reported at au¬ topsy in dyslexies.30 Two reasons may
account for this difference. Galaburda et al30 visually inspected the plana at
autopsy and concluded that
they were
bilaterally larger in area in dyslexic subjects than in normal subjects. Un¬ like the present study, no metric data supported their conclusions. Second, we only measured the length of the plana as revealed on extreme lateral
sagittal MRI scans. While there are significant difficulties in measuring the region of the plana,37 good reliabil¬ ities
were
obtained in the present
study in measuring the left (0.97 cm) and right (0.95 cm) plana length. These measurements
were
taken from the
posterior end of the sylvian fissure at the temporoparietal juncture to the
second transverse Heschl sulcus. The correlation of length measurements to plana area as reported in Galaburda and coworkers'30 autopsies is unknown. Thus, comparing our incidence of length asymmetry or symmetry with area measurements obtained at au¬ not be appropriate. Nonetheless, consistent with the conclusions of Galaburda et al,30 re-
topsy may
Table 2.—Means and SDs for ROIs in
Developmental Dyslexic, Attention Deficit Disorder/Hyperactivity, and Normal Children* Attention Deficit Developmental Dyslexies, Disorder/Hyperactive
Reliability of Measure
ROI Total area, cm2 Left frontal area, cm* Right frontal area, cm2 Left plana length, cm
Right plana length, cm length, cm Right insular length, cm
.97
Left frontal width,
.96
.95
Left insular
cm
Right frontal width, Left
(SD) (8.11) 18.44 (1.53) 19.30 (2.75) 1.20(0.41) 1.32 (0.37) 4.52 (0.88) 4.36 (0.75) 5.23 (0.20) 5.23 (0.18) 29.94 (3.66) 28.60 (4.01) 6.40 (0.27) 6.38 (0.28)
144.05 .95
cm
.93 posterior area, cm2 Right posterior area, cm2 Left posterior width, cm .93 Right posterior width, cm ROI indicates region of interest.
Normal
xxx
(SD) (15.76) 18.93(2.06) 19.71 (1.65) 1.41 (0.21) 1.20 (0.30) 4.8 (0.54) 4.64 (0.66) 5.3(0.31) 5.39 (0.30) 28.77 (3.75) 25.66 (3.49) 6.41 (0.27) 6.29 (0.41)
143.41
Fig 2.—Group and hemispheric differences in the width of the anterior region of interest measurements. L indicates left; R, right.
(SD) 144.35 (9.12)
(2.56) (2.93) 1.52 (0.35) 1.34 (0.39) 5.15 (0.58) 5.18 (0.56) 5.42 (0.26) 5.59 (0.29) 30.31 (4.37) 28.65 (4.61) 6.46 (0.13) 6.44 (0.15) 18.95
20.49
suits reported here do support the no¬ tion that the mechanism of corticogenesis, and possibly the processes associ¬ ated with the elimination of unwanted cells, is implicated in dyslexia. Since the dyslexic subjects and normal sub¬ jects did not differ significantly in right plana length (1.32 cm vs 1.34 cm, respectively) but did in terms of the left plana length (1.20 cm vs. 1.52 cm, respectively), it would seem that some deviation in corticogenesis preferen¬ tially affects the left planum temporale during development. The dispro¬ portionate clustering of focal dysplasias in the left planum temporale (11:1) in the study by Galaburda et al30 sup¬ ports this conclusion.33 The finding of a bilaterally smaller insular region in dyslexia is also of in¬ terest. While it is not unreasonable to assume some correlation between plana and insular morphologic fea¬ tures during fetal development and maturation, the left insular region has historically been associated with lan¬ guage disturbance. Marie, for example, argued that the only true language disturbance resulted from lesions af¬ fecting Wernicke's region (of which the planum temporale comprises the posterior area) and the intrahemispheric fibers connecting it to Broca's area.47 The relative importance of the insular region in dyslexia is further implicated in positron emission tomographic studies of regional cerebral glucose metabolism (rCMRglc) in which dyslexies show bilaterally de¬ creased levels of rCMRglc during read¬ ing compared with normals.48 It may be that bilaterally smaller insular re-
Fig 3.—Group and hemispheric differences in the length of the insular region of interest measurements. L indicates left; R, right.
Downloaded From: http://archneur.jamanetwork.com/ by a Western University User on 06/08/2015
100
L
90
Accepted for publication January 16, 1990. From the Departments of Special Education (Drs Hynd, Semrud-Clikeman, and Lorys) and Psychology (Dr Hynd), University of Georgia, Athens; Department of Neurology, Medical College of Georgia, Augusta (Dr Hynd); and Athens (Ga) Magnetic Imaging (Dr Novey and Ms Eliopulos). Reprint requests to Department of Special Education, Aderhold Hall, University of Georgia, Athens, GA 30602 (Dr Hynd).
Approximately
"^
R.
Lorys, PhD; Edward S. Novey, MD; Deborah Eliopulos, RT
3% to 6% of all
school-age children are believed to suffer from developmental dyslex¬ ia,1-2 which has been defined as a "rare but definable and diagnosable form of primary reading retardation with some
form of central
nervous
system
dysfunction."3 A diagnosis of dyslexia typically requires normal intelligence, a significant discrepancy between measured ability and reading achieve¬ ment, and the absence of other handi¬
capping conditions that may produce reading delay. While it is reasonably well docu¬ mented that some dyslexic children may variably suffer from deficits in visuoperceptual processes,4 sequenc¬ ing ability,5 phonemic segmentation,6 and automatized cognitive processing,7 it has historically been assumed that these deficits result from a disruption in underlying neurologic systems.1 Be¬ havioral evidence provides support for what has been referred to as the Wernicke-Geschwind model.8 This model implicates the involvement in reading of the bilateral posterior cortex; region of the angular gyrus; Wernicke's re¬
gion, including the superior temporal and insular regions; and Broca's area. Topographic mapping of brain electri¬ cal activity provides some support for
this model and advances evidence that the region of the supplementary motor area may be involved as well.9 How¬ ever, despite support for this neuro¬ logic model of reading, the behavioral evidence for the presumed neurologic origins of dyslexia has remained only correlative, until recently.10·11 Neuroimaging and postmortem studies comprise more direct evidence as to how neurologic substrata differ in
Downloaded From: http://archneur.jamanetwork.com/ by a Western University User on 06/08/2015
the brains of
dyslexies. Based on neurobiological theory that implicates the importance of the central language centers,1215 computed tomographic (CT)/magnetic resonance imaging (MRI) studies have provided evidence
that ties deviations in normal patterns of posterior asymmetry (left greater than right [L > R]) to the dyslexic
syndrome.1620
Both CT17 and postmortem stud¬ ies21'25 document that about 66% of normal brains are asymmetric (L > R), favoring the left planum tem¬ porale and posterior region. Other asymmetries also exist. For example, in 75% of brains the volume of the right frontal region exceeds that of the left,24 the left anterior speech region (pars opercularis and pars triangularis of the third frontal convolution) is larger than is the right,26 and the left auditory cortex27 and posterior thala¬ mus28 show similar (L > R) asymme¬ tries. Dyslexies would appear to have a higher incidence of symmetrical or re¬ versed posterior asymmetry than is found in control populations. While approximately two thirds of control brains show L > R posterior asym¬ metry,17 only 10% to 50% of dyslexic brains show this pattern.16·19·20 Cytoarchitectonic studies also support the finding of a higher-than-normal inci¬ dence of plana symmetry and docu¬ ment the presence of numerous focal dysplasias preferentially involving the left frontal, left perisylvian, and right frontal regions.29·30 The CT/MRI and postmortem studies, while few, would seem to implicate some deviation in normal patterns of neuronal migra¬ tion and maturation during the fifth to
seventh month of fetal gestation.1315·30
Neurobiological theory also predicts
that there is an interaction between deviations in normal patterns of brain morphologic appearance in dyslexia, the co-occurrence of left-handedness, and an increased incidence of allergies and familial autoimmune disease.31 While studies consistently document a significantly greater incidence of lefthandedness in populations of dys¬ lexies,1618 little direct evidence exists that ties the co-occurrence of auto¬ immune disease to left-handedness among dyslexies.32 No evidence attests to the interaction between deviations in brain morphologic findings in dys¬ lexia, the presence of autoimmune dis¬ ease, and left-handedness in dyslexia. While supportive of the neurobio¬ logical model of dyslexia, the neuroimaging and postmortem studies are characterized by serious méthodologie flaws. A critical review of these studies suggests that the neuroimaging stud¬ ies employed poorly documented dys¬ lexic populations, included question¬ able control populations, failed to in¬ clude populations with other clinical syndromes in order to document the uniqueness of deviations in patterns of brain asymmetry in dyslexies, and failed to examine the relationship be¬ tween left-handedness and autoim¬ mune disease as they relate to patterns of brain morphologic findings in dyslexia.33 The postmortem studies are similarly characterized by méthod¬
ologie problems.33
In this context then, the purpose of this study was to determine whether deviations in patterns of normal brain asymmetry characterize the brains of dyslexies. By including a clinic control population with attention deficit disorder/hyperactivity (ADD/H) with¬ out significant learning problems, the relative uniqueness of symmetry or reversed asymmetry in the region of the plana in dyslexia could be deter¬ mined. Since the neurologic basis of ADD/H presumably involves frontal dysfunction in the caudostriatal region34·35 and not the left central lan¬ guage zones, it might be expected that only the dyslexies would show differ¬ ent patterns of plana morphologic findings. Data were also obtained for the right and left anterior area and width, insular region, and posterior area and width. A second aim of this study was to improve on the diagnostic procedures employed in deriving the dyslexic and ADD/H groups such that the fre¬ quency and kind of Diagnostic and Statistical Manual of Mental Disor¬ ders, Third Edition (DSM III) codiag-
could be determined. This was viewed as important, since a recent federal reformulation of the definition of learning disabilities noted the fre¬ quent co-occurrence of learning dis¬ abilities and the attention deficit disorders.36 Finally, no neuroimaging study of patterns of brain morphologic findings in dyslexies has employed a computer program that allows for metric mea¬ surements to be made on the MRI scan itself. In using such a program, reli¬ ability of region of interest (ROI) mea¬ surements could be obtained. This was viewed as important, especially con¬ sidering the well-documented difficul¬ ties in measuring the region of the noses
planum temporale.37
SUBJECTS AND METHODS
Subjects The subjects who participated in this study were either outpatients of the Center for Clinical and Developmental Neuropsychology (CCDN) at the University of Geor¬ gia, Athens, or were normal control sub¬ jects solicited from the community. The CCDN serves as a diagnostic and referral facility for northeast Georgia, and refer¬ rals came primarily from area physicians, other clinics, or school districts. The 10 de¬ velopmental dyslexic and 10 ADD/H chil¬ dren represented consecutive referrals ex¬ cluding those patients who received other primary DSM III or Diagnostic and Statis¬ tical Manual of Mental Disorders, Revised Third Edition (DSM III-R) diagnoses or those children who were diagnosed as hav¬ ing mild mental retardation (Full Scale IQ [FSIQ] 85), a positive family history for learning problems, personal his¬ tory of difficulty learning to read, reading achievement significantly (> 20 standard score points) below FSIQ on both the Word Attack and Passage Comprehension subtests of the WRMT-R, and no behaviorally reported symptoms of hyperactivity. While some reported symptoms of inatten¬ tion and impulsivity were permitted among the dyslexies, a co-occurring diagnosis of ADD/H or ADHD excluded participation in this study. Diagnosis of ADD/H required average or better intellectual ability (FSIQ > 85); no reported family history of learning prob¬ lems; no significant deficit in reported or measured achievement; documented behav¬ ioral deficits in attention, impulsivity, and motor activity (hyperactivity) consistent with a DSM III ADD/H diagnosis; and a favorable response to stimulant medica¬ tion. Children diagnosed as normal were administered the same comprehensive evaluation procedures and were required to have average or better intellectual ability (FSIQ > 85); no reported family history of learning problems; no significant deficit in achievement; and no reported or observed medical, educational, social, or emotional difficulties. Diagnosis was based on similar proce¬ dures employed in other clinics.44 Diagnos¬ tic decisions were reached separately by two psychologists after considering all rel¬ evant historical, behavioral, and psycho¬ metric data. Diagnostic decisions were re¬ quired to reflect the above-noted criteria for inclusion in this study as well as the crite¬ ria outlined in DSMIIIor DSM III-R for the
Table
1—Subject and Psychometric Data* Developmental Dyslexies,
Attention Deficit Disorder
Normal,
Hyperactivity
xxx
(SD)
(SD)
(40.43) (8.43)
(SD)
Variable
Chronologic
age,
_
mo
Handedness, % right No. of codiagnoses No. of reported allergies/ autoimmune disease WISC-R IQs Verbal Full Scale WRMT-R (Reading Word Attack
120.60
141.20
96
96 (6.99)
(24.55) (88.00)
4/0.9
3/1.2 107.00
(11.20)
107.70 (10.89) 108.00 (8.68)
Performance
Achievement) 73.80(14.91)
Passage Comprehension
(24.07)
118.90
38
75.00(17.17)
(15.80) 110.80(13.97) 109.30(12.27)
122.80(9.96) 124.80 (15.74) 125.40 (10.91)
106.50
96.30
(16.47)
115.60
(9.97)
99.40
(10.78)
112.30
(12.34)
*
WISC-R indicates: Wechsler Intelligence Scale for Children-Revised; WRMT-R, Woodcock Reading Mastery 10 Tests-Revised. Word Attack and Passage Comprehension test scores are reported in standard scores. for all groups. =
other psychiatric disorders. Disagreement in diagnosis by the two psychologists was resolved through mutual discussion. Reli¬ abilities for this diagnostic process have been reported elsewhere44 and meet ac¬ cepted criteria for reliability in clinical
diagnosis.45
MRI Protocol
Once a child was entered into the study, sequential T, sagittal and axial MRI planes were obtained using a 0.6-T Technicare scanner (Health Images; Atlanta, Ga). A carefully selected protocol involving fifteen 7.5-mm sagittal planes (TR 690; TE 32) and eleven 5-mm axial planes (TR 500, TE 32) was used. Region of =
our knowledge. Recognizing the significant variability in brain morphologic findings and in recognition of the fact that some of the ROI may not be easily judged,3710 MRI scans were randomly selected and an inde¬ pendent measure obtained for each ROI. The initial measurements were made by an experienced radiological technologist trained specifically to identify the ROI. The reliability check was completed by a pedi¬ atrie neuropsychologist with expertise in developmental neuroanatomy. Reliability coefficients were excellent (x 0.95) across =
the 13 ROI measurements. These data reported in Table 2.
=
=
obtained.
MRI Quantification and
All MRI studies
gist
RESULTS
=
interest measurements were obtained us¬ ing the Technicare ROI measurement soft¬ ware system. Figure 1 shows how left and right anterior-width measurements were
as
are
not having
a
Reliability
by a neurolo¬ pathologic appearance.
were
read
Careful documentation of the neuroanatomic relationship between CT asymme¬ tries and measurements made post mortem reveal that the axial slice transversing the region of the planum temporale (including, in part, the supramarginal and angular gy¬ ri) is the only slice that correlates signifi¬ cantly in this regard.46 Consequently, an axial slice transversing this region was se¬ lected for obtaining the ROI measurements for the width and area of the left and right anterior (from a line drawn horizontally across the tip of the genu) and posterior (from a line drawn horizontally across the posterior tip of the splenium) regions, length of the left and right insular regions, and for total brain area. Extreme lateral sagittal slices were employed in obtaining the left and right planum temporale length measurements.
Reliability of ROI measurements ob¬ tained from MRI scans of the children's brains has not previously been reported, to
order [DSM III-R]). No significant dif¬ ferences (P > .05) existed in terms of the incidence of allergies among the subjects or in terms of the reported incidence of autoimmune disease in the families of these subjects. Overall, however, none of these measures sig¬ nificantly correlated with ROI data and were, therefore, not employed as covariates in the following analyses. Means and SD for the ROI measure¬ ments are presented in Table 2. An initial analysis revealed that there were no significant differences (P > .05) in total brain area between the dyslexic, ADD/H, and normal chil¬ dren. Therefore, it seems reasonable that any resulting differences in pat¬ terns of hemispheric asymmetry or symmetry between these groups may be associated with regional variation in neurologic development and not due to gross differences in brain morpho¬
logic findings. A 2 (left-right) X 3 (group) analysis of variance (ANOVA) indicated signif¬ icant (F [2,24] 6.00, .05) from ei¬ ther the dyslexic or normal children. It can be concluded that dyslexic children have a bilaterally shorter insular re¬ gion than do normal children. An ANOVA for between-group dif¬ ferences revealed a significant interences
=
While there were no significant (P > .05) differences in chronologic age across groups, there was a significant difference (^[2,27] 8.66, < .001) in general intellectual ability (FSIQ), with the normal children having a sig¬ =
nificantly higher
FSIQ_ (x
=
125.90)
than either the dyslexic (x 108.00) or ADD/H (x 109.30) children. Based on data from the Edinburgh Inven¬ tory,40 three dyslexic subjects were judged to be left-handed, while none of the ADD/H or normal children were identified as left-handed. A significant < .025) difference (F [2,27] 4.21, was noted between groups on the lat¬ erally quotient due to the three left¬ handers in the dyslexic group. There was also a significant differ¬ ence in the number ( 2[2] 11.10, < .004) and kind of codiagnoses be¬ tween the groups. Three dyslexies re¬ ceived DSM III codiagnoses (overanx¬ ious disorder, major depressive epi¬ sode, and ADD/WO), while seven codiagnoses were made with the ADD/ H children (three conduct disorder, one overanxious, one separation anxi¬ ety, and two oppositional defiant dis=
=
=
=
Downloaded From: http://archneur.jamanetwork.com/ by a Western University User on 06/08/2015
=
=
=
=
action between right and left length and group (F [2,24]
plana =
4.82,
.05). Tukey's pairwise compari¬ indicated that the dyslexic group differed significantly (q 5.92, < .05) from the normal group in left planum temporale length (dyslexies, 1.20 cm; normal subjects, 1.52 cm). No significant comparisons resulted when length of the right planum temporale
R patterns. The right and left measurements were con¬ sidered to differ in direction (L < R, L > R) if the magnitude of the differ¬ ence was
greater than 0.1
cm.
This
method is similar to other reports.33 As can be seen in Fig 5, a significant (x2[2] 9.60, < .008) effect resulted. In comparison with the normal and =
ADD/H children who evidenced the
typical L > R pattern (70% L > R; 30% L < R), fully 90% of the dyslexic children had either symmet¬ rical or reversed asymmetry (L < R) of plana length. Assuming that 30% L < R plana length morphology is nor¬ mal, the 90% incidence of this pattern in the dyslexic children represents a very significant departure from nor¬ more
mal patterns of differential brain mat¬ uration. When differences in posterior area and width measurements were exam¬
significant (P > .05) group, hemispheric, or interaction effects were found. Consequently, when em¬ ploying measurements of the posterior region from a line drawn horizontally across the left and right region later¬ ally from the tip of the splenium, no ined,
no
patterns of asymmetry fects
emerged.
or
group ef¬
COMMENT
Our results document the unique¬ of an increased incidence of
ness
L < R plana length symmetry/ reversed asymmetry in the brains of dyslexic children, in comparison with carefully diagnosed normal age- and sex-matched control children and a clinic control group of ADD/H chil¬ dren. The fact that there were no sig¬ nificant differences between these groups in total brain area suggests that this regional variation in brain morphologic characteristics is due to a more specific deviation in brain ontog¬ eny.
The finding that 70% of the normal and ADD/H children had L > R pat¬ terns of asymmetry in the region of the plana supports similar findings by other investigators employing CT/ MRI.17·20·33 This figure also correlates well with the postmortem findings that 65% of normal adult brains and 79% of infant brains are larger on the left side in the region of the planum temporale.21·25 The 90% incidence of L < R in the dyslexies in the length of the planum temporale is a very signif¬ icant increase over normal base rates.
Fig 1.—Magnetic resonance imaging (MRI) scan of a boy with developmental dyslexia showing reversed pattern (left greater than right) of frontal width asymmetry (top, left and right) (left frontal width, 5.4 cm; right frontal width, 5.2 cm) contrasted with a normal subject's MRI scan (bottom, left and right) showing the more typical pattern (left less than right) of frontal width asymmetry (left frontal width, 5.1 cm; right frontal width, 5.6 cm).
Downloaded From: http://archneur.jamanetwork.com/ by a Western University User on 06/08/2015
Further, our results indicate that it was primarily the significantly smaller left planum temporale that contributed to the higher incidence of L < R patterns in the dyslexies. The finding of a significantly smaller than normal left planum temporale in the
dyslexies is at variance with the bilaterally large plana reported at au¬ topsy in dyslexies.30 Two reasons may
account for this difference. Galaburda et al30 visually inspected the plana at
autopsy and concluded that
they were
bilaterally larger in area in dyslexic subjects than in normal subjects. Un¬ like the present study, no metric data supported their conclusions. Second, we only measured the length of the plana as revealed on extreme lateral
sagittal MRI scans. While there are significant difficulties in measuring the region of the plana,37 good reliabil¬ ities
were
obtained in the present
study in measuring the left (0.97 cm) and right (0.95 cm) plana length. These measurements
were
taken from the
posterior end of the sylvian fissure at the temporoparietal juncture to the
second transverse Heschl sulcus. The correlation of length measurements to plana area as reported in Galaburda and coworkers'30 autopsies is unknown. Thus, comparing our incidence of length asymmetry or symmetry with area measurements obtained at au¬ not be appropriate. Nonetheless, consistent with the conclusions of Galaburda et al,30 re-
topsy may
Table 2.—Means and SDs for ROIs in
Developmental Dyslexic, Attention Deficit Disorder/Hyperactivity, and Normal Children* Attention Deficit Developmental Dyslexies, Disorder/Hyperactive
Reliability of Measure
ROI Total area, cm2 Left frontal area, cm* Right frontal area, cm2 Left plana length, cm
Right plana length, cm length, cm Right insular length, cm
.97
Left frontal width,
.96
.95
Left insular
cm
Right frontal width, Left
(SD) (8.11) 18.44 (1.53) 19.30 (2.75) 1.20(0.41) 1.32 (0.37) 4.52 (0.88) 4.36 (0.75) 5.23 (0.20) 5.23 (0.18) 29.94 (3.66) 28.60 (4.01) 6.40 (0.27) 6.38 (0.28)
144.05 .95
cm
.93 posterior area, cm2 Right posterior area, cm2 Left posterior width, cm .93 Right posterior width, cm ROI indicates region of interest.
Normal
xxx
(SD) (15.76) 18.93(2.06) 19.71 (1.65) 1.41 (0.21) 1.20 (0.30) 4.8 (0.54) 4.64 (0.66) 5.3(0.31) 5.39 (0.30) 28.77 (3.75) 25.66 (3.49) 6.41 (0.27) 6.29 (0.41)
143.41
Fig 2.—Group and hemispheric differences in the width of the anterior region of interest measurements. L indicates left; R, right.
(SD) 144.35 (9.12)
(2.56) (2.93) 1.52 (0.35) 1.34 (0.39) 5.15 (0.58) 5.18 (0.56) 5.42 (0.26) 5.59 (0.29) 30.31 (4.37) 28.65 (4.61) 6.46 (0.13) 6.44 (0.15) 18.95
20.49
suits reported here do support the no¬ tion that the mechanism of corticogenesis, and possibly the processes associ¬ ated with the elimination of unwanted cells, is implicated in dyslexia. Since the dyslexic subjects and normal sub¬ jects did not differ significantly in right plana length (1.32 cm vs 1.34 cm, respectively) but did in terms of the left plana length (1.20 cm vs. 1.52 cm, respectively), it would seem that some deviation in corticogenesis preferen¬ tially affects the left planum temporale during development. The dispro¬ portionate clustering of focal dysplasias in the left planum temporale (11:1) in the study by Galaburda et al30 sup¬ ports this conclusion.33 The finding of a bilaterally smaller insular region in dyslexia is also of in¬ terest. While it is not unreasonable to assume some correlation between plana and insular morphologic fea¬ tures during fetal development and maturation, the left insular region has historically been associated with lan¬ guage disturbance. Marie, for example, argued that the only true language disturbance resulted from lesions af¬ fecting Wernicke's region (of which the planum temporale comprises the posterior area) and the intrahemispheric fibers connecting it to Broca's area.47 The relative importance of the insular region in dyslexia is further implicated in positron emission tomographic studies of regional cerebral glucose metabolism (rCMRglc) in which dyslexies show bilaterally de¬ creased levels of rCMRglc during read¬ ing compared with normals.48 It may be that bilaterally smaller insular re-
Fig 3.—Group and hemispheric differences in the length of the insular region of interest measurements. L indicates left; R, right.
Downloaded From: http://archneur.jamanetwork.com/ by a Western University User on 06/08/2015
100
L
90
