Basal Ganglia Volumes inChildren With Attention-Deficit
Hyperactivity Disorder Elizabeth H. Aylward, PhD; Allan L. Reiss, MD; Mark J. Reader, PhD;
Harvey S. Singer, MD; Jan E. Brown, RN; Martha B. Denckla, MD
Received Jan 11, 1995. Received revised April 4, 1995. Accepted for publi-cation May 17, 1995.From the Division of Psychiatric Neuro-Imaging (Dr Aylward), the Neu-roimaging Research Center (Dr Reiss) and the Department of Develop-mental Cognitive Neurology (Drs Reader and Denckla), Kennedy-KriegerInstitute, and the Departments of Psychiatry and Behavioral Sciences(Drs Aylward, Reiss, Reader, and Denckla), Neurology (Dr Singer, MsBrown, and Dr Denckla), and Pediatrics (Dr Singer and Ms Brown),School of Medicine, Johns Hopkins University, Baltimore, MD.
Address correspondence to Dr Elizabeth H. Aylward, Johns Hopkins Uni-versity School of Medicine, Division of Psychiatric Neuro-Imaging,Department of Psychiatry, 600 N Wolfe StreetMeyer 3-166, Baltimore,MD 21205.
Previous research has demonstrated volume reduction of the left globus pallidus in children with the codiagnoses ofTourette syndrome and attention-deficit hyperactivity disorder (ADHD), in comparison with children who haveTourette syndrome alone and with normal controls. The purpose of this study was to determine whether children withADHD alone also had volume reduction of the globus pallidus or other basal ganglia structures. Subjects were 10 boyswith ADHD, 16 boys with Tourette syndrome and ADHD, and 11 normal control boys. Groups were matched for age.Boys with ADHD were individually matched for age, handedness, and IQ to 10 of the 16 boys with Tourette syndromeand ADHD. Volumes of caudate, putamen, and globus pallidus were measured and corrected for brain volume. The boyswith ADHD had significantly smaller left globus pallidus volume and total globus pallidus volume (corrected for brainvolume) than the normal controls. The Tourette syndrome plus ADHD group did not differ from the ADHD group onany of the measures. We conclude that small globus pallidus volume, particularly on the left side, is associated withADHD. (J Child Neurol 1996; 11: 112-115).
Our group recently reported abnormalities in volumetricmeasures of globus pallidus in children with a codiagno-sis of Tourette syndrome and attention deficit hyperactiv-ity disorder (ADHD).1 Specifically, children with bothTourette syndrome and ADHD showed a significantlysmaller volume of the left globus pallidus than childrenwith Tourette syndrome only or normal control children.Analysis of lenticular asymmetry also showed more right-sided predominance in the Tourette syndrome plusADHD group compared with controls, and with a trend inthe same direction for the Tourette syndrome-only groupcompared with controls. The purpose of this follow-upstudy was to determine whether the basal ganglia anom-
alies found in children with Tourette syndrome plusADHD might be more directly associated with the ADHDdiagnosis rather than with Tourette syndrome.
SUBJECTS AND METHODS
SubjectsSubjects included 10 boys with ADHD, 16 boys with Tourette
syndrome plus ADHD, and 11 boys with no known developmen-tal disorders. Procedures were fully explained to all subjects,and informed consent was obtained from the subjects and their
parents. Both boys with ADHD and those with Tourette syn-drome plus ADHD met Diagnostic and Statistical Manual ofMental Disorders, 3rd edition revised, criteria for the diagnosisof ADHD, as judged by interview with experienced clinicians.Tourette syndrome was diagnosed using criteria defined by theTourette Syndrome Classification Group.2 The subjects withTourette syndrome plus ADHD included all of the boys withTourette syndrome plus ADHD in the sample described previ-
ously by Singer et al,1 plus one additional subject who wasenrolled subsequent to publication of the Singer et al study. Thecontrol subjects were healthy boys who were recruited fromoutside the hospital. These were 11 of the 14 boys included inthe sample described previously by Singer et al,1 with the three
youngest boys in the previous sample omitted in order to ensure
at UNIVERSITY OF WATERLOO on December 16, 2014jcn.sagepub.comDownloaded from
that groups matched on age. They had no personal history ofADHD or other neurologic disorder, as assessed by interview
involving Diagnostic and Statistical Manual of Mental Disorders,3rd edition revised, criteria. Mean age (t standard deviation[SD]) for the boys with ADHD was 11.26 (t 1.62) years, for the
boys with Tourette syndrome plus ADHD was 11.32 (t 1.46)years, and for the control subjects was 10.71 ( 1.98) years.There were no significant differences among the three groups(ADHD, Tourette syndrome plus ADHD, control) on age (F = 0.49;df = 2,34; P = .62). All of the subjects were right-handed. At thetime of the scan, 11 of the 16 boys with Tourette syndrome plusADHD were taking medication: three were taking clonidine, onewas taking methylphenidate, six were taking desipramine, andone was taking pimozide. All of the boys with ADHD were takingmethylphenidate at the time of the scan.
Ten boys with ADHD were selected to individually match10 of the boys with Tourette syndrome plus ADHD. Therefore, a
matched-pairs sample was identified, consisting of the 10 boyswith ADHD and 10 of the 16 boys with Tourette syndrome plusADHD. The subjects were individually matched on age (within1.5 years) and IQ (within 10 points). Mean age (t SD) for the boyswith ADHD was 11.26 (t 1.62) years and for the matched boyswith Tourette syndrome plus ADHD was 11.37 (t 1.26) years.Mean IQ (t SD) for the boys with ADHD was 104.20 (t 17.32) andfor the matched boys with Tourette syndrome plus ADHD was103.6 (t 12.17). The matched-pairs sample did not differ onmean age (t = 0.45, P = .66) or IQ (t = 0.27, P = .79). No attemptwas made to individually match the control subjects, because IQdata for them were unavailable.
Magnetic Resonance Imaging ScansProcedures for obtaining and measuring scans were identical tothose described by Singer et al.l Briefly, measurements were
performed on contiguous 3-mm inversion-recovery axial scansthat were angled parallel to the anterior commissure-posteriorcommissure line, using a TR of 3000 ms, TE of 20 ms, and Tr of500 ms. Images were obtained from the level of the fourth ven-tricle to above the corpus callosum.
Measurements included left and right volumes of caudate,putamen, and globus pallidus. For each region, area measure-ments were summed across all slices in which the region was
observed, and this sum was multiplied by slice thickness to yieldan approximate volume. Because the inversion-recovery imagedata set did not extend through the entire brain, a &dquo;representa-tive&dquo; brain volume was calculated by summing the areas of thefive largest brain slices in the image data set and multiplying byslice thickness.
Scans were measured by three raters (E.H.A., A.L.R., and
B.S.), according to specific rules for selecting slices and identify-ing structure boundaries (see Singer et all). Interrater reliability,based on measurement of at least 10 scans per rater, was
assessed by intraclass correlation coefficients. The average ofthe three possible correlational pairs was .95 for caudate, .86 for
putamen, and .78 for globus pallidus. All measurements were
performed blind to diagnosis. Because scans from the ADHD
sample were obtained and measured after those from the
Tourette syndrome plus ADHD and control groups, ratersrerated some of the previous scans to ensure continued reliabil-
ity and maintain blindness to diagnosis.
Statistical AnalysesFor the three-group comparisons (ADHD, Tourette syndrome plusADHD, and controls), differences for volumetric measures wereassessed using analysis of covariance, with brain volume as thecovariate. Post hoc comparisons were performed using the Scheffemethod. Group differences in asymmetry for each structure wereassessed using repeated measures analysis of variance, with side
(right versus left) as the repeated measure. In order to compareour results with those of previous investigators, we also performedanalyses of variance using an asymmetry index, (right - left)/(light+ left), for each structure as the dependent variable, as well as x2analyses, whereby each subject was categorized as having left
greater than right asymmetry, right greater than left asymmetry, or
right equal to left volume for each structure. Subjects were classi-fied as right greater than left if their asymmetry score (right vol-ume minus left volume) was 1 SD or more above the mean, as left
greater than right if their asymmetry score was 1 SD or morebelow the mean, and right equals left if their asymmetry score wasless than 1 SD above or below the mean of all asymmetry scores.
For the matched-pairs analyses, all brain region volumeswere corrected for overall brain volume by calculating ratios
(region volume divided by representative brain volume).Because of the nonnormal distribution of these ratio measures,
nonparametric analyses (Wilcoxon signed-ranks tests) were per-formed. Nonparametric analyses comparing the two matched
groups on asymmetry indices ([right - left]/[right + left]) werealso performed for each structure.
All P values represent two-tailed comparisons.
Representative brain volumes and region measures foreach group are displayed in Table 1. Three-group analy-ses, using analysis of covariance with brain as a covariate,revealed significant group differences for total globus pal-lidus volume (F = 3.79; df = 2,33; P = .03) and left globuspallidus volume (F = 4.58; df = 2,33; P = .02). Post hocanalysis revealed that it was the ADHD group that differedfrom the control group, while the Tourette syndrome plusADHD group differed from neither of the others.
Groups did not differ on any of the other region mea-sures or on asymmetry indices for any structure, asassessed by a group x side interaction in the repeated-measures analyses of variance (using right and left sidesas the repeated measures). When an asymmetry index wasused ([right - left]/[right + left]), the difference betweenthe Tourette syndrome plus ADHD group and controlswas almost significant (P = .058) for lenticular nuclei(putamen plus globus pallidus) but not for caudate, withthe Tourette syndrome plus ADHD group demonstratingright greater than left asymmetry and the control groupdemonstrating the reverse pattern. The x2 analyses, com-paring the three groups on percentage of subjects havingright greater than left asymmetry, left greater than rightasymmetry or right equal to left volume, revealed no sig-nificant group differences for any structure.
Data from the matched-pairs sample (Tourette syn-drome plus ADHD and ADHD) were analyzed separately.
at UNIVERSITY OF WATERLOO on December 16, 2014jcn.sagepub.comDownloaded from
Wilcoxon signed-rank tests indicated no significant differ-ences between the matched Tourette syndrome plusADHD and ADHD groups on any of the basal ganglia orventricular ratio measures and no differences on asym-
metry indices for any structure. z
Results of this study suggest that small globus pallidusvolume, particularly on the left side, is associated withADHD. Because the ADHD group did not differ from theTourette syndrome plus ADHD group but did differ fromcontrols, we conclude that previous findings of globuspallidus anomalies in the Tourette syndrome plus ADHDgroupl may have been related more to their ADHD diag-nosis than to Tourette syndrome per se.
In contrast to our previous study, however, we didnot find significantly smaller globus pallidus volume forthe Tourette syndrome plus ADHD group in comparisonto the control group. Although the same measurementswere used as in the previous study, the groups in the cur-rent study were somewhat smaller and included onlymales. As can be observed from the data in Table 1, theboys with Tourette syndrome plus ADHD had smaller leftglobus pallidus and smaller total globus pallidus than thecontrols, but this difference did not reach statistical sig-nificance in the current study. The magnitude of thegroup difference (8.5% for left globus pallidus and 5.5%for total globus pallidus) was, however, approximatelythe same as in our previous study.
Our study did not find any significant group differ-ences on asymmetry of structures, using repeated mea-sures analyses of variance with left and right structurevolumes used as the repeated measures. In order to com-
pare our results with those of Singer et al,1 we alsoassessed group differences on an asymmetry index, (right- left)/(right + left), for each structure. Consistent with thefmding of Singer et al, the Tourette syndrome plus ADHDgroup again showed more rightward asymmetry of thelenticular nuclei (putamen plus globus pallidus) than didthe control group, although this finding did not quite
reach significance (P = .058). (When Singer et al restrictedtheir analysis to males only, the group difference forlenticular asymmetry was also only marginally significant,with a P of .07.) The ADHD group did not differ from con-trols or from the Tourette syndrome plus ADHD group onasymmetry of lenticular nuclei. This finding suggests thatanomalous lenticular asymmetry may be associated withTourette syndrome alone and not with ADHD. This wouldbe consistent with the finding of Singer et al of a trend inthe direction of more right-sided predominance of lenticu-lar nuclei in subjects with Tourette syndrome who did nothave ADHD, as well as a significant finding for those sub-jects with Tourette syndrome who did have ADHD.
Unlike Hynd et all and Castellanos et al,4 we did notfind any differences between subjects with ADHD andcontrols on asymmetry of caudate nucleus. Using areameasures of caudate in one axial slice, Hynd et alreported that right greater than left asymmetry of caudatearea was more common in ADHD subjects than controlsand that this asymmetry was due primarily to smaller leftcaudate in the subjects with ADHD than in the controls.Using volume measurements similar to ours, Castellanoset a14 found that control subjects had a significant rightgreater than left caudate asymmetry, whereas subjectswith ADHD demonstrated no such asymmetry. We foundno main effect for caudate asymmetry in the repeated-measures analysis of variance (using right and left cau-date volumes as the repeated measures) and no group xasymmetry interaction. Examination of group means, in
fact, suggests that both the ADHD and Tourette syn-drome plus ADHD groups had slightly larger left caudatethan right caudate mean volumes, in contrast to the con-trol subjects, whose mean right and left caudate volumeswere the same. In order to be consistent with the methodof analysis used by Hynd et a1,3 we also categorized sub-jects as having right greater than left asymmetry, leftgreater than right asymmetry, and right equal to left vol-ume for each structure. x2...