30. Urenjak J, Williams SR, Gadian DG, et al. Proton nuclear 32. Nitsch RM, Blusztajn JK, Pittas AG, et al. Evidence for a magnetic resonance spectroscopy unambiguously identifies membrane defect in Alzheimer disease brain. Proc Natl Acad different neural cell types. J Neurosci 1993;13:981–989. Sci USA 1992;89:1671–1675. 31. Stokes CE, Hawthorne JN. Reduced phosphoinositide concen- 33. Young LT, Kish SJ, Li PP, et al. Decreased brain [3H]inositol trations in anterior temporal cortex of Alzheimer-diseased 1,4,5-triphosphate binding in Alzheimer’s disease. Neurosci brains. J Neurochem 1987;48:1018–1021. Lett 1988;94:198–202.

Evidence for cortical “disconnection” as a mechanism of age-related cognitive decline

M. O’Sullivan, MRCP; D.K. Jones, PhD; P.E. Summers, PhD; R.G. Morris PhD; S.C.R. Williams, PhD; and H.S. Markus, DM

Article abstract—Background: Normal aging is accompanied by a decline of cognitive abilities, and executive skills may be affected selectively, but the underlying mechanisms remain obscure and preventive strategies are lacking. It has been suggested that cortical “disconnection” due to the loss of fibers may play an important role. But, to date, there has been no direct demonstration of structural disconnection in humans in vivo. Methods: The authors used diffusion tensor MRI to look for evidence of ultrastructural changes in cerebral white matter in a group of 20 elderly volunteers with normal conventional MRI scans, and a group of 10 younger controls. The older group also underwent neuropsycho- logical assessment. Results: Diffusional anisotropy, a marker of white matter tract integrity, was reduced in the white matter of older subjects and fell linearly with increasing age in the older group. Mean diffusivity was higher in the older group and increased with age. These changes were maximal in anterior white matter. In the older group, anterior mean diffusivity correlated with executive function assessed by the Trail Making Test. Conclusions: These findings provide direct evidence that white matter tract disruption occurs in normal aging and would be consistent with the cortical disconnection hypothesis of age-related cognitive decline. Maximal changes in anterior white matter provide a plausible structural basis for selective loss of executive functions. In addition to providing new information about the biological basis of cognitive abilities, diffusion tensor MRI may be a sensitive tool for assessing interventions aimed at preventing cognitive decline. NEUROLOGY 2001;57:632–638

A progressive decline of cognitive functions is a rec- Indirect evidence for loss of functional “connectiv- ognized corollary of increasing age. Despite the clear ity” with age has been provided by a recent PET importance of these changes, the underlying mecha- study,5 where functional relationships between the nisms remain obscure. Early studies describing loss left dorsolateral prefrontal cortex and right hip- of neocortical neurons have been largely refuted by pocampus were examined during two cognitive tasks, the application of more accurate techniques.1 An- both activating a distributed processing network other possible mechanism is “cortical disconnection,” that included these nodes. In older subjects, both which leads to a loss of the functional integration of regions retained the capacity to become activated, neurocognitive networks. The concept of disconnec- but the functional relationships between them tion was first proposed by Geschwind in 19652 and changed with age, as did overall patterns of cognitive has gained credence as modern theories of cognitive activation. These findings suggest that age-related function have recognized the importance of large- decline in the performance of these tasks might re- scale, distributed cortical processing networks.3 More flect changes in functional integration rather than recently, coordinated activity of large distributed net- dysfunction of specific gray matter areas. works has been inferred from analysis of fMRI data.4 A plausible anatomic substrate for functional dis-

From the Division of Clinical Neuroscience (Dr. O’Sullivan and Prof. Markus), St. George’s Hospital Medical School, London; and the Departments of Neuroimaging (Dr. Summers and Prof. Williams), Neuropsychology (Dr. Morris), and Old Age Psychiatry (Dr. Jones), Institute of Psychiatry, London, United Kingdom. Supported in part by the Stroke Association of the United Kingdom. Also supported by a Clinical Research Fellowship from St. George’s Hospital Medical School, United Kingdom (M.O’S.), by a grant from the Joint Research Committee of King’s College Hospital (P.E.S.), and by a grant from the Wellcome Trust (D.K.J.). Received November 13, 2000. Accepted in final form April 7, 2001. Address correspondence and reprint requests to Dr. Mike O’Sullivan, Department of Clinical Neuroscience, St. George’s Hospital Medical School, Cranmer Terrace, London SW17 0RE, United Kingdom; e-mail: [email protected]

632 Copyright © 2001 by AAN Enterprises, Inc. Figure 1. Fractional anisotropy images and selection of regions of interest (ROI). Images are shown at similar slice positions in a subject aged 71 years (left) and one aged 29 years (right). High signal intensity (bright- ness) reflects higher fractional aniso- tropy (FA). The genu and splenium of the can be seen clearly as areas of high anisotropy connecting the hemispheres. A whole white matter ROI is shown on the left image; ante- rior, middle, and posterior ROI are shown on the right image. Overall, sig- nal intensity in white matter is lower for the patient on the left, consistent with lower FA, and in turn lower tract integrity, in the older subject. connection is disruption of the white matter tracts search Ethics Committee of King’s College Hospital and all that link the components of distributed neurocogni- subjects gave informed, written consent. tive networks, or “structural disconnection.” A post- MRI. Diffusion tensor imaging was performed on a mortem study has provided evidence that normal 1.5-T G.E. Signa MR scanner (General Electric, Milwau- aging is accompanied by a loss of white matter fibers, kee, WI). An imaging sequence optimized for measurement particularly small, myelinated fibers.6 However, to of the diffusion tensor in white matter was used14 and the date, there has been no direct, in vivo demonstration of diffusion tensor elements calculated at each voxel. The structural disconnection in humans. diffusion tensor provides a three-dimensional representa- The recent development of diffusion tensor MRI tion of water diffusion. When diffusion is equal in all direc- has provided, for the first time, a noninvasive tions, it is described as isotropic, and when diffusion is method for delineating anatomy of white matter greater in some directions than others as anisotropic. FA can be considered as the proportion of the whole tensor pathways7 and subsequent assessment of pathologic that is anisotropic. Mean diffusivity is a directionally aver- tract disruption. In white matter tracts, water mo- aged measure of diffusion, with the effect of anisotropy tion is restricted by axonal membranes and myelin, removed. Measures of FA and mean diffusivity were de- and water moves more freely along the direction of a rived using equations previously published by Basser and tract than perpendicular to it. This directionality of Pierpaoli.15 The maximum strength of diffusion-encoding diffusion can be quantified by fractional anisotropy gradients was 22 mT·mϪ1. Fifteen to 18 near-axial slices (FA). FA varies from zero in tissue where diffusion is were obtained for each patient, providing near whole-brain equal in all directions to 1 where diffusion is entirely coverage. Diffusion tensor imaging took 4 to 6 minutes to unidirectional. Increasing anisotropy is a likely complete. Axial T1-weighted, T2-weighted, and fluid- marker of white matter tract integrity: anisotropy attenuated inversion recovery (FLAIR) images also were correlates with histologic markers of myelination,8 performed on all patients in the older group to exclude and there is accumulating evidence that changes in clinically silent brain disease. FA are of functional significance.9-13 Neuropsychological testing. Subjects in the older group In this study, we applied diffusion tensor MRI in a also underwent neuropsychological assessment. Two pa- group of elderly volunteers and younger subjects in tients in this group declined testing, so neuropsychological an attempt to provide a direct in vivo demonstration scores were available for 18 of the older group. Tests in- of white matter tract disruption with normal aging, cluded the Wisconsin Card Sorting Test, Reitan Trail Mak- and to correlate such changes with cognitive ing Test, Verbal Fluency, and the Mini-Mental State dysfunction. Examination. Premorbid intelligence was assessed using the National Adult Reading Test–Revised (NART-R) pro- Methods. Subjects. Twenty healthy elderly volunteers nunciation test. (age range, 56 to 85 years) were recruited from both a Image analysis. For each slice, a region of interest cohort of volunteers from the Clinical Age Research Unit (ROI) containing the white matter of one hemisphere was at King’s College Hospital and a separate cohort selected selected using a semiautomated approach (figure 1) using randomly from general practitioner lists in Bromley, South the Dispim display software (David Plummer, University London, United Kingdom. All were assessed by a clinician College London, United Kingdom). By applying an FA (M.O.’S.) and underwent full neurologic examination. threshold of 0.1, peripheral isotropic tissue and tissue with Ten younger volunteers (age range, 23 to 37 years) also very low anisotropy, including the skull, CSF, and cortical were recruited from general staff at King’s College London. gray matter, were removed from each slice. This was done No subject had a history of current or previous neurologic by placing a cursor near the gray matter/white matter or psychiatric illness. Approval was obtained from the Re- interface. The Contour function of the Dispim software is

August (2 of 2) 2001 NEUROLOGY 57 633 then used to draw automatically around all the tissue with FA above 0.1, starting from the point nearest to the cursor. ROI were edited by hand to remove errors produced by magnetic susceptibility artifacts. The corresponding T2- weighted images were used to confirm that these artifacts were outside the brain. The resulting gaps in white matter outline were filled in by hand to correspond visually with the boundary generated by the Contour function. For each ROI, a median value for FA and a mean value for the trace of the diffusion tensor were obtained. Median values were used for FA to reduce the effect of outliers. Each ROI was subsequently divided into anterior, middle, and posterior areas using the posterior margin of the genu and the ante- rior margin of the splenium of the corpus callosum as anatomic markers (see figure 1). For slices rostral to the level of these structures, an imaginary coronal plane pro- jecting vertically from each of these markers was used to define these areas. Values from six slices at and above the level of the lateral ventricles were combined to form ante- rior, middle, and posterior volumes of interest, and mean values of FA and mean diffusivity were calculated for each. Intrarater reliability. A reproducibility analysis was performed to assess the reliability of selection and hand- editing of ROI. Five subjects from each group were selected at random by a colleague with no knowledge of subject details or the purpose of the study. All subject information was removed from the computer image files, and they were randomized and renamed with the letters A to J. A second set of whole white matter ROI was then obtained using the same methods.

Results. General findings. A single member of the older group was found to have diffuse hyperintensities of white matter on T2-weighted MRI and was excluded from subsequent analysis. This left 19 subjects in the older Figure 2. Values of white matter fractional anisotropy (A) group (17 with neuropsychological test data), mean age and mean diffusivity (B) plotted against age for 29 Ϯ 72 8 years (range, 56 to 85 years), and 10 subjects in the healthy subjects. Significant linear correlations were Ϯ younger group, mean age 30 5 years (range, 23 to 37 found in the older group of subjects (Ͼ55 years) but not years). There was no difference in sex frequency between the younger group. The shape of the relationship suggests ϭ groups (p 0.67). a possible age threshold for white matter deterioration. Comparison of elderly and young subject groups. Whole white matter FA declined with age, whereas mean diffusivity increased; figure 2 plots values for all subjects in both groups. Table 1 shows the results of group compar- cm3). The corresponding coefficients of variation were 1.1% isons for volumes of interest derived separately from ante- for mean diffusivity, 2.2% for FA, and 2.7% for white mat- rior, middle, and posterior ROI. Reduction of anisotropy ter volume. was not uniform throughout the brain. FA of anterior Correlations of diffusion tensor–derived parameters white matter was reduced by 9.9% in the older subjects, with age in older group. Table 2 shows results of linear whereas there was no significant reduction in FA of poste- regression analysis examining the correlations between rior white matter in older individuals. Middle white mat- age and diffusion tensor parameters in the older group; ter had an intermediate but significant FA reduction of there was no evidence of age correlations in the younger 6.0%. Similarly, mean diffusivity was increased by 13.5% group. Significant negative correlations between age and in anterior white matter of older subjects, 12.1% in middle FA were demonstrated in anterior and middle white mat- white matter, and 8.7% in posterior white matter. These ter regions, but no significant relationship was observed regional differences in reduction of FA are depicted graph- for posterior white matter. Corresponding positive correla- ically in figure 3. tions between age and mean diffusivity also were seen in Intrarater reliability. Repeat, blinded analysis in a these regions. Volume of anterior white matter was seen to randomly selected subgroup of subjects revealed excellent decline significantly with age. However, after covarying for reproducibility of all measures. The within-subject SD pro- both volume and sex, significant independent relationships vide a practical index of measurement error16: these were of age with FA and mean diffusivity of anterior white 10 ϫ 10Ϫ6mm2·sϪ1 for mean diffusivity (mean for all mea- matter remained. FA of anterior white matter was both surements 851 ϫ 10Ϫ6mm2·sϪ1), 0.005 for FA (mean lower in the sixth decade and declined more rapidly with 0.234), and 9.9 cm3 for white matter volume (mean 368.7 advancing age than FA of middle or posterior regions.

634 NEUROLOGY 57 August (2 of 2) 2001 Table 1 Comparisons of mean (SD) diffusion tensor–derived parameters in elderly and young subject groups

Region of interest Elderly (n ϭ 19) Young (n ϭ 10) p Value p Value*

Total white matter Volume, cm3 357 (38) 393 (35) 0.021 0.008 FA 0.225 (0.012) 0.238 (0.007) 0.001 0.028 Diffusivity, 10Ϫ6mm2 ⅐ SϪ1 912 (58) 814 (23) Ͻ0.001 Ͻ0.001 Anterior white matter Volume, cm3 100 (15) 112 (17) 0.074 0.049 FA 0.199 (0.012) 0.221 (0.007) Ͻ0.001 Ͻ0.001 Diffusivity, 10Ϫ6mm2 ⅐ SϪ1 942 (64) 830 (27) Ͻ0.001 Ͻ0.001 Middle white matter Volume, cm3 104 (20) 114 (22) 0.24 0.194 FA 0.256 (0.019) 0.272 (0.015) 0.016 0.067 Diffusivity, 10Ϫ6mm2 ⅐ SϪ1 881 (65) 786 (17) Ͻ0.001 Ͻ0.001 Posterior white matter Volume, cm3 159 (25) 174 (17) 0.067 0.052 FA 0.234 (0.013) 0.239 (0.012) 0.267 0.510 Diffusivity, 10Ϫ6mm2 ⅐ SϪ1 896 (50) 824 (24) Ͻ0.001 0.002 p Values are shown before (p) and after (p*) covarying for sex and white matter volume (atrophy). FA ϭ fractional anisotropy.

Effects of sex. Four of the 10 younger patients and 6 of (0.007) in women and 0.238 (0.007) in men (p ϭ 0.88). the 19 elderly patients were women. In both younger and Similarly, no evidence of sex differences was found in FA older groups, there was a trend toward lower total white or mean diffusivity in the older subjects. matter volume in women: for the younger group, mean Neuropsychological correlates of diffusion tensor param- total volume (SD) was 368 (34) cm3 for women vs 408 (27) eters in older group. Neuropsychological test scores were cm3 for men (p ϭ 0.069); for the older group the corre- available for 17 of the 19 subjects in the older group. Re- sponding values were 338 (7) cm3 versus 365 (43) cm3 (p ϭ itan Trail Making B-A, a measure of attentional set- 0.045). Correcting for the difference in sex frequency did shifting and executive function, correlated with mean increase the significance of volume differences between diffusivity of anterior regions. Meanwhile, corrected Verbal groups (table 1). For the younger group, mean diffusivity Fluency scores correlated with FA of middle white matter. (SD) was 810 (18) ϫ 10Ϫ6mm2·sϪ1 in women and 816(27) ϫ These correlations were independent of age, sex, and pre- 10Ϫ6mm2·sϪ1 in men (p ϭ 0.69), and mean FA was 0.238 morbid IQ. Correlation coefficients are shown in table 3.

Discussion. Using diffusion tensor imaging, we have demonstrated a reduction of FA and an in- crease in mean diffusivity of water diffusion in the cerebral white matter of a group of healthy, elderly volunteers compared with a group of younger sub- jects. These changes were maximal in frontal re- gions. In older subjects, a decline of FA and increase of mean diffusivity were observed with increasing age, again most marked in frontal regions. Mean diffusivity of anterior white matter correlated with executive function, as assessed by the Reitan Trail Making Test. These findings are likely to reflect al- terations in white matter ultrastructure accompany- ing aging that are not apparent using conventional imaging techniques, yet are of functional signifi- Figure 3. Comparisons of fractional anisotropy between cance. This is consistent with the hypothesis that elderly (gray bars) and young (black bars) groups in ante- “disconnection” of cortical areas and functional dis- rior, middle, and posterior white matter. Loss of anisot- ruption of large-scale neurocognitive networks occur ropy with age was not uniform throughout the brain: the as part of the aging process. largest proportional loss was seen in anterior regions, and Age-related cognitive decline is an important so- no significant loss was identified in posterior white matter cioeconomic issue, and understanding the mecha- *p Ͻ 0.05, **p Ͻ 0.01. nisms has clinical implications, both in selecting August (2 of 2) 2001 NEUROLOGY 57 635 Table 2 Correlations between diffusion tensor–derived Table 3 Correlations of mean diffusivity and fractional parameters and age in the older subject group, before and after anisotropy from anterior, middle, and posterior regions with covarying for sex and white matter volume neuropsychological test scores

Covarying for Trail Making Before covarying volume/sex B-A Verbal Fluency

p p Region of interest rr* rr* Region of interest Pearson’s r Value Pearson’s r Value Anterior MD 0.57† 0.61† Ϫ0.04 0.44 Total white matter Anterior FA Ϫ0.41 Ϫ0.09 0.22 0.04 Ϫ Volume 0.39 0.095 Middle MD 0.53† 0.12 Ϫ0.08 0.05 ؊ ؊ FA 0.60 0.007 0.57 0.017 Middle FA Ϫ0.09 0.34 0.69‡ 0.61† Diffusivity 0.69 0.001 0.68 0.003 Posterior MD 0.44 0.55 0.27 0.36 Anterior white matter Posterior FA Ϫ0.32 0.04 0.37 0.28 Volume Ϫ0.55 0.016 Correlation coefficients are shown before (r) and after (r*) cova- FA ؊0.65 0.002 ؊0.53 0.029 rying for age, sex, white matter volume, premorbid IQ (National Diffusivity 0.68 0.001 0.65 0.005 Adult Reading Test–Revised), and general intellectual impair- Middle white matter ment (Mini-Mental State Examination Score). p Values are not corrected for multiple comparisons. Volume Ϫ0.31 0.194 † p Ͻ 0.05. FA Ϫ0.48 0.039 Ϫ0.42 0.090 ‡ p Ͻ 0.01. Diffusivity 0.74 <0.001 0.84 <0.001 FA ϭ fractional anisotropy; MD ϭ mean diffusivity. Posterior white matter Volume Ϫ0.11 0.652 FA Ϫ0.26 0.276 Ϫ0.22 0.400 rameters derived from diffusion tensor imaging correlate with white matter ultrastructure and are of Diffusivity 0.38 0.108 0.42 0.090 functional importance. Diffusional anisotropy corre- Correlations significant at the p Ͻ 0.05 level, after covarying, are lates with histologic markers of myelination8 and in- shown in boldface type. creasing anisotropy accompanies maturation of 18 FA ϭ fractional anisotropy. white matter tracts in children. Loss of anisotropy has been demonstrated in a number of conditions in which disruption of white matter tracts is believed to promising candidate therapies and in choosing surro- occur, including cerebral small vessel disease, MS, gate markers for their evaluation. If, as we propose, motor neuron disease, and schizophrenia.9-12 The cerebral disconnection is a fundamental mechanism functional importance of loss of FA has been con- of cognitive dysfunction, diffusion tensor imaging firmed by the findings of a correlation between an- will allow evaluation of new therapies in smaller isotropy of left temperoparietal white matter and numbers of subjects without the difficulties associ- reading ability,13 and a correlation between aniso- ated with repeated neuropsychological testing. tropy of corticospinal tracts and disease severity Our findings provide a link between evidence of scores in motor neuron disease.11 accompanying aging and ev- The finding of a selective decline of anterior FA idence of age-related anatomic and biochemical with age is intriguing because selective impairment changes in cerebral white matter. A number of find- of executive functions with age has been described ings from a previous PET cognitive activation study both in humans19,20 and nonhuman primates,21 and support the conclusion that disruption of functional functional imaging studies suggest that frontal corti- relationships between cortical regions, or disconnec- cal areas, particularly the dorsolateral prefrontal tion, underlies age-related changes in performance.5 cortex, are involved in performance of these cognitive Although the conclusions remain speculative, the au- tasks.22 We hypothesize that the dependence of age- thors argue that the capacity of the prefrontal cortex related changes in prefrontal activation on cognitive to become activated is not attenuated by age, but context, in the PET study described previously, is rather the coordination of the whole neural response due to selective disruption of specific white matter is impaired, implying a loss of functional connectiv- processing streams, leading to functional disconnec- ity. However, functional imaging techniques provide tion of some cognitive networks, but not others. only indirect evidence of cerebral disconnection, and An important limitation of this study, which ap- depend on the assumption of a linear relationship plies to all diffusion tensor imaging studies at between neural activity and hemodynamic response present, is that functionally important fiber tracts (vasoneuronal coupling)17 that may not hold in el- are likely to occupy only a proportion of white matter derly subjects. ROI. Selection of ROI is likely to be crude compared Diffusion tensor imaging is free of such assump- with the complex functional anatomy of cognitive tions, and accumulating evidence suggests that pa- processes, and it is not possible to assess the relative 636 NEUROLOGY 57 August (2 of 2) 2001 contributions of major white matter tracts such as imaging will be a valuable tool for exploring these the superior longitudinal fasciculus, which pass mechanisms in vivo. Although our neuropsychologi- through several ROI, and shorter association fibers. cal findings should be interpreted cautiously, they The greater magnitude of changes in anterior re- should stimulate further work on the role of struc- gions could reflect either preferential involvement of tural changes of white matter in age-related cogni- shorter fibers, or that affected long tracts occupy a tive decline. greater proportion of the anterior ROI. Ultimately, it Postmortem histologic evidence does point to a may be possible to track relevant fiber projections loss of myelinated white matter fibers in normal ag- using mathematical modeling techniques. Although ing, which is consistent with the FA changes de- there is growing interest in this field, at present scribed here.6 Furthermore, age-related activation of these methods can be applied only to major projec- microglia has been demonstrated in the brains of tions such as the pyramidal tracts and corpus callo- older, normal people,26 and microglial cell inclusions, sum, and their reliability is unproven.23,24 Despite thought to reflect phagocytosis of myelin, have been the limitations of sampling large white matter areas, observed. A diffuse age-related increase in levels of this method is sufficient to suggest that changes in inducible nitric oxide synthase also has been ob- white matter ultrastructure are not uniform, and served in white matter of old rhesus monkeys,27 sug- paves the way for a more detailed analysis of the gesting that nitric oxide could mediate age-related anatomy of white matter aging. white matter injury. These findings hint at possible The sample size for this study was chosen to allow mechanisms whereby age-related loss of connectivity correlation of diffusion tensor imaging parameters may occur. with age, and is insufficient for a detailed analysis of We hypothesize that reduction in integrity of neuropsychological correlations. However, the corre- white matter tracts is an important mechanism of lations seen with the Reitan Trail Making and Ver- age-related cognitive decline and that executive dys- bal Fluency tests are plausible, being consistent with function in elderly subjects may be a form of cerebral current knowledge of the functional anatomy of disconnection syndrome. Our findings, together with these tasks: Trail Making is an executive test of at- those from functional imaging and pathologic stud- tentional set shifting sensitive to frontal dysfunction, ies, can be combined in a comprehensive “disconnec- whereas the middle volume of interest would have tion” theory of cognitive changes in normal aging. included the , a major white mat- Diffusion tensor imaging provides, for the first time, ter tract providing input to Broca’s area. These cor- a direct method for assessing cerebral connectivity in relations also were robust when covariates were vivo. This technique can be performed on many of considered, including possible effects of age, premor- the MRI scanners in current clinical use, and may bid intelligence, and atrophy. The inconsistency be- provide a tool for the evaluation of emerging thera- tween correlations with FA and mean diffusivity pies to prevent cognitive decline in the elderly. (with FA correlating most strongly with Verbal Flu- ency and mean diffusivity correlating most strongly with Trail Making) does not necessarily cast doubt Acknowledgment on these findings. Although these measures often are The authors thank Dr. Mark Horsfield of the Division of Medical Physics, University of Leicester for providing software for analysis correlated, they may provide different information of diffusion tensor images; Professor Steve Jackson for allowing about white matter structure, and it is tempting to them to study subjects from the Clinical Age Research Unit Data- speculate that Verbal Fluency involves highly or- base (King’s College Hospital); and Emma Ouldred for help with dered parallel fiber tracts, where changes in FA may recruiting these subjects. be more sensitive to pathologic change. However, a more important factor is likely to be measurement References error of FA, which is greater in tissue of lower an- 1. Morrison JH, Hof PR. Life and death of neurons in the aging 25 isotropy, making correlations with FA more diffi- brain. Science 1997;278:412–419. cult to detect in anterior white matter. 2. Geschwind N. Disconnection syndromes in animals and man. Cognitive function also is influenced by several Brain 1965;88:237–294. 3. Mesulam M. Large-scale neurocognitive networks and distrib- factors that cannot be accounted for in this study. uted processing for attention, language, and memory. Ann Although changes in white matter volume due to Neurol 1990;28:597–613. atrophy have been considered, no account can be 4. Bullmore ET, Rabe-Hesketh S, Morris RG, et al. Functional taken of changes in gray matter volume dispropor- magnetic resonance image analysis of a large-scale neurocog- nitive network. Neuroimage 1996;4:16–33. tionate to white matter atrophy, and we cannot ex- 5. Esposito G, Kirkby BS, Van Horn JD, et al. Context- clude the possibility that some degeneration of white dependent, neural system-specific neurophysiological concom- matter fibers occurs secondary to specific gray mat- itants of ageing: mapping PET correlates during cognitive activation. Brain 1999;122:963–979. ter disease. Further careful clinicopathologic studies 6. Tang Y, Nyengaard JR, Pakkenberg B, et al. Age-induced are required to investigate these issues. Functional white matter changes in the human brain: a stereological reorganization of cortex also may influence cognitive investigation. Neurobiol Aging 1997;18:609–615. performance. Nevertheless, reorganization of large, 7. Makris N, Worth AJ, Sorensen AG, et al. Morphometry of in vivo human white matter association pathways with diffusion- distributed cognitive networks is likely to depend on weighted magnetic resonance imaging. Ann Neurol 1997;42: integrity of white matter tracts, and diffusion tensor 951–962. August (2 of 2) 2001 NEUROLOGY 57 637 8. Wimberger DM, Roberts TP, Barkovich AJ, et al. Identifica- 18. Klingberg T, Vaidya CJ, Gabrieli JDE, et al. Myelination and tion of “premyelination” by diffusion-weighted MRI. J Comput organization of the frontal white matter in children: a diffu- Assist Tomogr 1995;19:28–33. sion tensor MRI study. Neuroreport 1999;10:2817–2821. 9. Jones DK, Lythgoe D, Horsfield MA, et al. Characterisation of 19. Belleville S, Peretz I, Malenfant D. Examination of the white matter damage in ischaemic leukoaraiosis with diffusion working memory components in normal aging and in de- tensor magnetic resonance imaging. Stroke 1999;30:393–397. mentia of the Alzheimer type. Neuropsychologia 1996;34: 10. Werring DJ, Clark CA, Barker GJ, et al. Diffusion tensor 195–207. imaging of lesions and normal-appearing white matter in mul- 20. Daigneault S, Braun CM. Working memory and the Self-Ordered tiple sclerosis. Neurology 1999;52:1626–1632. Pointing Task: further evidence of early prefrontal decline in nor- 11. Ellis CM, Simmons A, Jones DK, et al. Diffusion tensor MRI mal aging. J Clin Exp Neuropsychol 1993;15:881–895. assesses corticospinal tract damage in ALS. Neurology 1999; 21. Peters A. Age-related changes in oligodendrocytes in monkey 53:1051–1058. . J Comp Neurol 1996;371:153–163. 12. Lim KO, Hedehus M, Moseley M, et al. Compromised white 22. Berman KF, Ostrem JL, Randolph C, et al. Physiological acti- matter tract integrity in schizophrenia inferred from diffusion vation of a cortical network during performance of the Wiscon- tensor imaging. Arch Gen Psychiatry 1999;56:367–374. sin Card Sorting Test: a positron emission tomography study. 13. Klingberg T, Hedehus M, Temple E, et al. Microstructure of Neuropsychologia 1995;33:1027–1046. temporo-parietal white matter as a basis for reading ability: 23. Basser PJ, Pajevic S, Pierpaoli C, et al. In vivo fiber tractography evidence from diffusion tensor magnetic resonance imaging. using DT-MRI data. Magn Reson Med 2000;44:625–632. Neuron 2000;25:493–500. 24. Mori S, Crain BJ, Chacko VP, et al. Three-dimensional track- 14. Jones DK, Horsfield MA, Simmons A. Optimal strategies for ing of axonal projections in the brain by magnetic resonance measuring diffusion in anisotropic systems by magnetic reso- imaging. Ann Neurol 1999;45:265–269. nance imaging. Magn Reson Med 1999;42:515–525. 25. Pierpaoli C, Basser PJ. Toward a quantitative assessment of 15. Basser PJ, Pierpaoli C. Microstructural and physiological fea- diffusion anisotropy. Magn Reson Med 1996;36:893–906. tures of tissue elucidated by quantitative-diffusion-tensor 26. Sheng JG, Mrak RE, Griffin WS. Enlarged and phagocytic, MRI. J Magn Reson 1996;111:209–219. but not primed, interleukin-1 (IL-1) alpha-immunoreactive 16. Bland JM. An introduction to medical statistics, 3rd ed. Ox- microglia increase with age in the normal human brain. Acta ford: Oxford University Press, 2000. Neuropathol (Berl) 1998;95:229–234. 17. Taylor JG, Krause B, Shah NJ, et al. On the relation between 27. Sloane JA, Hollander W, Moss MB, et al. Increased microglial brain images and brain neural networks. Hum Brain Map activation and protein nitration in white matter of the aging 2000;9:165–182. monkey. Neurobiol Aging 1999;20:395–405.

Activate your online subscription At www.neurology.org, subscribers can now access Neurology online offers: the full text of the current issue of Neurology and • Access to journal content in both Adobe Acrobat back issues to 1999. Select the “Login instructions” PDF or HTML formats link that is provided on the Help screen. Here you • Links to PubMed will be guided through a step-by-step activation • Extensive search capabilities process. • Complete online Information for Authors • Examinations on designated articles for CME credit • Access to in-depth supplementary scientific data

638 NEUROLOGY 57 August (2 of 2) 2001