Law

A Morphometric Analysis of Neuroanatomic Abnormalities in Traumatic Brain Injury

Description
A Morphometric Analysis of Neuroanatomic Abnormalities in Traumatic Brain Injury
Categories
Published
of 9
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  A Morphometric Analysisof NeuroanatomicAbnormalities inTraumatic Brain Injury Tracy D. Vannorsdall, Ph.D.Nicola G. Cascella, M.D.Vani Rao, M.D.Godfrey D. Pearlson, Ph.D.Barry Gordon, M.D., Ph.D.David J. Schretlen, Ph.D. Cognitive and structural brain abnormalities arecommon following traumatic brain injury (TBI).The authors compared cognition and brain struc-ture in 14 TBI survivors and 28 matched healthycomparison subjects. TBI survivors showedreduced cerebral volume, due mainly to whitematter changes, and poorer attention, psychomo-tor speed, and memory. Severity of white matterabnormality correlated with worse performanceon several cognitive measures that distinguishedbetween groups. Using voxel-based morphometry,regions of reduced white matter concentrationwere found throughout the cerebrum along withmore localized gray matter reductions. Findingssuggest that diffuse rather than focal aspects of TBI contribute most to cognitive outcome. (The Journal of Neuropsychiatry and ClinicalNeurosciences 2010; 22:173–181) A pproximately 1.5 million Americans sustain atraumatic brain injury (TBI) each year, 1 and 80,000to 90,000 develop long-term cognitive and neurologicaldisabilities as a result. 2 Impairments of attention, mem-ory, processing speed, and executive functioning arecommon following moderate to severe TBI. 3–5 Quanti-tative neuroimaging techniques have provided evi-dence linking these deficits to changes in brain struc-ture. While it has long been apparent that TBI ofteninvolves focal structural brain lesions related to the spe-cific injury, evidence now indicates that more general-ized abnormalities due to neuroexcitatory and meta- bolic changes also occur. 4 These are thought to occurirrespective of the specific location(s) of focal injuriesand might account for shared features of cognitive im-pairment.Most studies of TBI-related structural brain changesemploy manually traced regions of interest to quantifyatrophic changes in areas vulnerable to injury. Longi-tudinal and cross-sectional studies of moderate to se-vere TBI have found atrophy in white matter fibertracts, particularly the corpus callosum, as well as re-ductions of frontal and total brain volume and increasesin ventricular volume, likely due to white matter loss. 6 Reduced volume of the anterior and inferior frontal and Received November 20, 2008; revised April 3, 2009; accepted April 24,2009. The authors are affiliated with the Johns Hopkins UniversitySchool of Medicine in Baltimore, Maryland. Address correspondenceto David J. Schretlen, Ph.D., Johns Hopkins Hospital, 600 N. Wolfe St.,Meyer 218, Baltimore, MD 21287; dschret@jhmi.edu (e-mail).Copyright © 2010 American Psychiatric Publishing, Inc.  J Neuropsychiatry Clin Neurosci 22:2, Spring 2010  http://neuro.psychiatryonline.org  173  temporal lobes is also commonly found following mod-erate to severe TBI, likely due to abrasive contact with bony plates of the skull during the acute injury. 5 Whilethese studies are extremely informative, the regions-of-interest approach has limitations that might constrainour ability to detect localized structural effects of TBI.Specifically, this approach requires the  a priori  selectionof regions of interests, which could preclude findingatrophy in other regions. Further, regions-of-interestmethods are biased toward the selection of brain struc-tures with clearly defined boundaries over regions andstructures with unclear anatomical boundaries.More recently, automated whole brain neuroimagingmethods that permit a detailed assessment of tissue lossfollowing TBI have been developed. Voxel-based mor-phometry is one such approach. It allows for the detec-tion and localization of region-specific reductions in both gray matter and white matter concentrationsthroughout the brain. Briefly, voxel-based morphome-try entails normalizing each image to stereotactic spaceand then segmenting each voxel into gray matter, whitematter, and CSF segments based on voxel intensity. Thesegmented images are then smoothed to calculate prob-ability maps of the three tissue types within each voxel,allowing a voxel-by-voxel comparison of tissue concen-trations. 7,8 Gale and colleagues 9 used voxel-based morphometryto identify regions of reduced gray matter concentra-tions in nine TBI survivors and nine age- and sex-matched healthy adults. Participants in the TBI groupall sustained mild to severe injuries at least 1 year ear-lier. Gale et al. 9 found diffuse gray matter reductionsthroughout the brains, particularly in frontal and tem-poral cortex, subcortical regions, cerebellum, and cin-gulate gyri of TBI survivors. However, the analyseswere not adjusted for multiple comparisons, and giventhe large number of voxel-by-voxel comparisons con-ducted in voxel-based morphometry, some of the find-ings likely represented type I errors. Tomaiuolo et al. 10 compared the white matter concentrations of 19 severeTBI survivors who were 3–113 months postinjury tohealthy age- and sex-matched comparison subjects. Re-duced white matter concentrations were found in thecorpus callosum, fornix, parahippocampal gyrus, opticradiation, optic chiasm, internal capsule, and superiorfrontal gyrus. Finally, Salmond et al. 11 compared graymatter and white matter concentrations in 22 survivorsof moderate to severe TBI with those of age-, sex-, andpremorbid IQ-matched healthy comparison subjects.The TBI patients, who were all scanned at least 6months postinjury, showed reduced gray matter con-centration in the basal forebrain, hippocampal forma-tion, cerebellum, insula, thalamus, and areas of the neo-cortex. White matter concentration differences were lesspronounced and occurred in the lateral capsule andcorpus callosum.The goal of our study was to compare long-termsurvivors of TBI with carefully matched healthy adultsin terms of both cognitive and neuroanatomic abnor-malities. Based on previous research, 3–5 we hypothe-sized that the TBI group would perform more poorlythan healthy adults on tests of attention, memory, psy-chomotor speed, and executive functioning. We alsohypothesized that the TBI group would show differ-ences in global gray matter and white matter volumesrelative to matched comparison subjects, and that volu-metric measures would correlate with cognitive testperformance. Finally, we used voxel-based morphom-etry in an attempt to identify specific regional abnor-malities of gray matter and white matter tissue concen-tration in patients with TBI relative to healthy controlparticipants. METHODS Participants Sixteen TBI survivors were recruited from the commu-nity and from hospitals and rehabilitation facilities inthe Baltimore/Washington, D.C., metropolitan area.We were particularly, but not exclusively, interested inpatients who presented prominent signs of apathy. Twoindividuals did not undergo brain imaging and wereexcluded, resulting in a final sample of 14 TBI survi-vors. TBI severity was rated based on multiple factors,including Glasgow Coma Scale scores, following theusual criteria for severe (3–8), moderate (9–12), andmild (13–15) injury. All participants sustained up to 12weeks’ loss of consciousness except for two individualsfor whom the presence and duration of altered con-sciousness were unclear. However, both of these pa-tients showed dramatic personality changes, meetingDSM-IV criteria for “personality change secondary toTBI, apathetic subtype” at long-term follow-up, andshowed central or unilateral atrophy on neuroimaging.The majority of brain injuries resulted from a motorvehicle accident (n  11) or fall (n  2). All injuries weresustained at least 18 months (mean  122 months, NEUROANATOMIC ABNORMALITIES IN TBI 174  http://neuro.psychiatryonline.org  J Neuropsychiatry Clin Neurosci 22:2, Spring 2010  range  18–366 months) prior to study participation. Nopatient met criteria for a mood disorder according toDMS-IV criteria, though most met criteria for an apathysyndrome (n  10). All were free of substance abuse inthe 6 months prior to study participation.In addition, 28 healthy participants in the Johns Hop-kins Aging, Brain Imaging, and Cognition Study of nor-mal aging provided data for analyses. The individualsreported no history of head injury and were free of medical and psychiatric illnesses that can affect brainstructure or function. The comparison subjects and TBIgroups were carefully matched for age, sex, race, yearsof education, estimated premorbid IQ (NART-R), 12 andtotal intracranial volume (p  0.05 in all cases).On a single day, each participant underwent phys-ical and neurological examinations, a psychiatric in-terview, laboratory blood studies, a brain MRI scan,and detailed neurocognitive assessment. Cognitivetests included measures of psychomotor speed anddexterity (Grooved Pegboard test), 13 psychomotorprocessing speed (Trail Making test), 14 auditory di-vided attention (Brief Test of Attention), 15 letter-cuedverbal fluency (letters S and P), visual construction(Rey-Osterrieth Complex Figure Test [RCFT]), 16,17 verbal learning and recall (Hopkins Verbal LearningTest—Revised [HVLT-R]), 18 visual learning and re-call (Brief Visuospatial Memory Test—Revised[BVMT-R]), 19 and concept formation (modified Wis-consin Card Sorting Test [mWCST]). 20,21 The studywas approved by the Johns Hopkins Medicine insti-tutional review board and all subjects gave writteninformed consent. MRI Protocol and Image Analyses All participants underwent brain MRI on the same 1.5Tesla GE Signa scanner (Milwaukee, Wisconsin). Weacquired 124 contiguous 1.5 mm 3-D SPGR slices inthe coronal plane. The parameters were as follows:repetition time   35 seconds, echo time   5 seconds,flip angle  45°, and image matrix  256  256. Imageswere preprocessed with statistical parametric map-ping software (SPM2; Wellcome Department of Cog-nitive Neurology, London), using the Good et al. 8 optimized method. First, images were transformedinto standard stereotactic space, and global shapedifferences were removed with spatial normalization.This involved the application of linear and nonlineartransformations of the study images to a customizedtemplate based on the T1-weighted images of 86 rel-atively healthy normal comparison subjects (meanage   51 years, 42% male). Prior probability mapsestimated the likelihood of a given tissue type at eachvoxel, and based on this information, the imageswere segmented into volumes of gray matter, whitematter, and CSF. Images were then resliced as 1.0 mmisotropic voxels and smoothed to make the data morenormally distributed using a 12 mm full width at half maximum Gaussian kernel.Based on the segmented images, we estimated graymatter (GM), white matter (WM), and total brain vol-umes (TBV  GM  WM volumes). Total intracranial vol-ume was also calculated (TICV  GM  WM  CSF). Theratios white matter to TICV (WM/TICV) and gray mat-ter to TICV (GM/TICV) were used as proxies of whitematter- and gray matter-related atrophy (i.e., signifi-cantly decreased ratios denote atrophy). Statistical Analyses Independent sample t tests and chi-square analyseswere used to compare groups in terms of age, sex, race,education, estimated premorbid IQ, segmented brainvolumes, and cognitive test performance. Spearman’srho was used to test correlations between cognitive per-formance and measures of segmented brain volume,atrophy, and time since injury. Analyses were con-ducted using the SPSS version 15.0 for Windows. 22 Voxel-based morphometry statistical analyses wereconducted with MatLab 7.0, 23 using ANCOVA to ex-plore group differences in gray matter and white mattertissue concentrations. Although the groups did not dif-fer in age, both groups were characterized by a rela-tively wide age range (25 to 64 years old). Becausevoxel-based morphometry is highly sensitive to age-related cerebral atrophy, we treated age as a nuisancevariable for all analyses. 24 Contrasts tested the hypoth-esis that the TBI group would demonstrate areas of reduced gray and white matter concentrations relativeto the comparison group. We also tested the reverse(that TBI survivors would show increased gray andwhite matter concentrations) but hypothesized that nosignificant differences would be found. Analyses wereconducted with an absolute threshold masking of 0.1and a voxel extent threshold of 25, meaning that at least25 contiguous voxels must show the association of in-terest in order to be considered a cluster of significantlyreduced gray matter or white matter. A cluster-wisefalse discovery rate (FDR) correction of p  0.05 was VANNORSDALL  et al.  J Neuropsychiatry Clin Neurosci 22:2, Spring 2010  http://neuro.psychiatryonline.org  175  applied to limit the number of type I errors obtainedover multiple voxel-wise comparisons. RESULTS Cognitive Function As shown in Table 1, the TBI and healthy comparisongroups did not differ significantly in age, sex distri- bution, years of education, race, total intracranial vol-ume, or estimated premorbid IQ. However, TBI sur-vivors performed worse than healthy comparisonsubjects on all 12 neuropsychological measures, andsix of these differences were statistically significant.Specifically, patients with TBI required more time tocomplete measures of psychomotor speed (Trail Mak-ing test, part A: t  3.19, df   40, p  0.006; part B,t  2.63, df   40, p  0.019) and fine motor speed anddexterity (Grooved Pegboard dominant hand:t  2.60, df   40, p  0.020; nondominant hand:t  2.47, df   40, p  0.026). The TBI group also per-formed more poorly on a test of auditory dividedattention (Brief Test of Attention: t  2.93, df   40,p  0.010) and delayed verbal recall (HVLT-R recall:t  2.15, df   40, p  0.038). The groups did not differ onmeasures of letter-cued verbal fluency, design copy-ing, verbal learning, visual learning and recall, orconcept formation (p  0.05 in all cases). Time sinceinjury was not correlated with performance on any of the cognitive tests in TBI survivors (p  0.05 in allcases). Segmented Brain Volumes Time since injury was not correlated with any graymatter or white matter volumes, total brain volumes, ormeasures of atrophy (p  0.05 in all cases). As hypothe-sized, TBI survivors showed significantly smaller whitematter volumes (t  2.43, df   40, p  0.02) and lowerWM/TICV ratios (t  3.22, df   40, p  0.003) than healthycomparison participants. Overall, the TBI groupshowed a 0.73 SD reduction in white matter volumesand a 1.1 SD increase in white matter-related atrophy. TABLE 1. Comparison of Demographic Characteristics and Baseline Characteristics, Brain Volumes, and Cognition by GroupCharacteristicTBI Survivors Healthy Comparisonpn % n % Gender (male) 10 71.4 20 71.4 1.0Race (caucasian) 11 78.6 22 78.6 1.0 Mean SD Mean SD pDemographics and Baseline Characteristics Age (years) 43.5 9.7 45.5 13.2 0.62Education (years) 14.5 2.9 14.1 2.7 0.67Total intracranial volume 1540.3 162.7 1600.4 186.5 0.31Estimated premorbid IQ 103.5 10.3 105.8 10.2 0.49 Brain Volumes White matter volume 409.2 49.4 458.1 66.5 0.02*Gray matter volume 709.8 57.7 743.7 90.9 0.21Total brain volume 1119.0 95.9 1201.8 149.3 0.07White matter-related atrophy (WM/TICV) 0.27 0.02 0.29 0.02  < 0.01** Gray matter-related atrophy (GM/TICV) 0.46 0.02 0.47 0.03 0.75 Cognitive Test Performance Grooved Pegboard (dominant hand) 82.0 22.5 65.6 10.0 0.02*Grooved Pegboard (nondominant hand) 91.1 29.8 70.8 11.1 0.03*Trail Making, part A (seconds) 41.9 18.2 25.5 8.8  < 0.01** Trail Making, part B (seconds) 101.7 52.0 63.6 21.0 0.02*Brief Test of Attention 12.2 6.1 17.2 2.2  0.01** Letter fluency 22.8 9.9 27.2 7.8 0.12RCFT copy 31.2 4.7 32.1 4.3 0.51HVLT-R learning 22.6 7.1 25.6 4.4 0.10HVLT-R recall 7.0 3.6 9.0 2.3 0.04*BVMT-R learning 20.8 8.5 24.4 6.3 0.12BVMT-R recall 7.9 4.0 9.3 2.1 0.24mWCST categories 5.2 1.5 5.6 1.1 0.34*p  0.05**p  0.01IQ  intelligence quotient; WM  white matter; GM  gray matter; RCFT  Rey Complex Figure Test; HVLT-R  Hopkins Verbal Learning Test,Revised; BVMT-R  Brief Visuospatial Memory Test, Revised; mWCST  modified Wisconsin Card Sorting Test NEUROANATOMIC ABNORMALITIES IN TBI 176  http://neuro.psychiatryonline.org  J Neuropsychiatry Clin Neurosci 22:2, Spring 2010  TBI survivors also showed, on average, a 0.5 SD reduc-tion in total brain volume relative to healthy compari-son participants (t  1.88, df   40, p  0.067). The groupsdid not differ in gray matter volume or GM/TICV ra-tios (p  0.05 in all cases).In the TBI group, lower WM/TICV ratios correlatedwith worse performance on several measures of psy-chomotor speed, attention, memory and executive func-tioning. These included the Grooved Pegboard (domi-nant hand,     0.70, p  0.006; nondominant hand,    0.65, p  0.012), Brief Test of Attention (    0.61,p  0.020), word list learning on the HVLT-R (    0.58,p  0.030), visual learning and recall on the BVMT-R(    0.56, p  0.036 and     0.69, p  0.006, respectively),and categories completed on the mWCST (    0.55,p  0.041). There were no associations between GM/TICV ratio and performance on any cognitive test(p  0.05 in all cases). In the healthy comparison group,WM/TBV ratios did not correlate with performance onany cognitive measure (p  0.05 in all cases), whereaslower GM/TICV ratios correlated with worse perfor-mance on the Trail Making test, part A (    0.51,p  0.005), and delayed word recall on the HVLT-R(    0.38, p  0.045). Voxel-Based Morphometry Comparison of TissueConcentrations Voxel-based morphometry analyses showed wide-spread white matter density reductions in patientswith TBI relative to healthy comparison participants(t  2.80, df   39, p  0.05). Prominent areas of reducedwhite matter density were detected in the frontallobes, limbic region, corpus callosum, cingulate, thal-amus, parahippocampal region, and cerebellum (Ta- ble 2 and Figure 1). TBI survivors also demonstratedsignificantly reduced gray matter concentrations rel-ative to the healthy comparison group (t  3.21,df   39, p  0.05). As expected, areas of gray matterreduction were most prominent along the temporalpoles. Gray matter density reductions were also seenin central portions of the frontal, parietal, and occip-ital lobes; cingulate cortex; thalamus; and cerebellum(Table 3 and Figure 2). There were no areas in which TABLE 2. Areas of Reduced White Matter Densities in the TBI Group (N  14) Relative to Healthy Comparison Participants (N  28)X, Y, Z a Cluster Size z t p Anatomical Region 26, 15, 30 8620 4.72 5.53   0.001 Bilateral: frontal lobes, anterior and inferior prefrontal cortex,inferior frontal gyrus, sub-gyral and sub-lobar whitematter, middle frontal gyrus, limbic lobe, caudate/caudatehead region, anterior cingulate, subcallosal gyrus, andcorpus callosumLeft: sub-lobar regions (adjacent to lentiform nucleus andputamen, insular cortex)Right: superior frontal gyrus, medial frontal gyrus,dorsolateral prefrontal cortex, pre-central gyrus, cingulategyrus, dorsal and ventral anterior cingulate cortex, andsecondary motor cortex3,  9,  33 6414 4.93 5.85   0.001 Bilateral: cerebellum (anterior and posterior lobes, cerebellartonsils, culmen), brainstem (pons, midbrain, and medulla)  8, 30, 42 3250 5.21 6.32   0.001 Left: frontal lobe, dorsolateral prefrontal cortex, anteriorprefrontal cortex, medial frontal gyrus, superior frontalgyrus, cingulate gyrus, sub-gyral white matter, dorsal andventral anterior cingulate cortex, limbic lobe, secondarymotor cortex, and frontal eye fields2,  38, 8 2028 4.20 4.75 0.002 Bilateral: sub-lobar white matter, corpus callosum, limbiclobe; adjacent to thalamus/pulvinar, sub-gyral whitematter, region of the parahippocampal gyrus, cingulategyrusLeft: caudate/caudate tail region  29, 9, 30 1158 3.92 4.37 0.029 Left: frontal lobe, sub-gyral and sub-lobar white matter,caudate/caudate body region, corpus callosum, middlefrontal gyrus, limbic lobe, cingulate gyrus, anteriorcingulate, inferior frontal gyrus a Coordinates listed in Montreal Neurological Institute convention at the peak white matter voxel within each cluster for those clusters havingcluster-level p values below FDR 0.05. Only specific anatomical regions with greater than 25 voxels are reported. VANNORSDALL  et al.  J Neuropsychiatry Clin Neurosci 22:2, Spring 2010  http://neuro.psychiatryonline.org  177
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks