988

The location of chronic poststroke brain damage is typically based on T1-weighted, T2-weighted, or fluid attenuated

inversion recovery MRI sequences.1–3 These are principled techniques providing information about the extent of the cor-tical and subcortical lesions. However, it is important to note that these methods may fail to detect the full extent of white matter injury. It is well known from animal and human stud-ies that cortical and subcortical strokes are associated with white matter injury and anterograde and retrograde Wallerian degeneration.4–6 Therefore, cortical disconnections as a result of white matter injury may occur in areas not included in the frank lesion but still account for behavioral impairment.

Aphasia is an impairment of language processing, com-monly observed after a dominant hemisphere stroke.7 The emergence of aphasia after a stroke is traditionally associated with structural damage affecting cortical regions related to language functions. Classically, deficits in speech production are expected to occur after lesions affecting pars opercularis and pars triangularis in the left inferior frontal gyrus (now referred to as Broca area, involving the cytoarchitectonic

Brodmann areas [BAs] 44 and 45). Conversely, deficits in speech comprehension are typically associated with dam-age affecting portions of the temporal lobe, especially BA 22—Wernicke area.8

It is a caveat that much of what we know about brain–-language relationships has been derived from lesion studies that may have underappreciated the extent of cortical dis-connection. In fact, circuit disconnections could account for ≤26% of exceptions to the classical radiological–clinical cor-relations in aphasia.9

Newer imaging methodologies now enable the comprehen-sive mapping of all neural connections in the brain at medium and large scale (the brain connectome).10–15 The purpose of this study was to examine 2 independent but complementary forms of cortical damage—necrosis and disconnection—in relation to global aphasia measures and naming impairments in patients with aphasia. For this purpose, we performed a comprehensive assessment of aphasia severity and nam-ing impairments in a series of subjects with chronic aphasia caused by a left hemisphere stroke. We also assessed necrotic

Background and Purpose—Studies assessing the relationship between chronic poststroke language impairment (aphasia) and ischemic brain damage usually rely on measuring the extent of brain necrosis observed on MRI. Nonetheless, clinical observation suggests that patients can exhibit deficits that are more severe than what would be expected based on lesion location and size. This phenomenon is commonly explained as being the result of cortical disconnection. To understand whether disconnection contributes to clinical symptoms, we assessed the relationship between language impairments and structural brain connectivity (the connectome) in patients with chronic aphasia after a stroke.

Methods—Thirty-nine patients with chronic aphasia underwent language assessment and MRI scanning. Relying on MRI data, we reconstructed the individual connectome from T1-weighted and diffusion tensor imaging. Deterministic fiber tractography was used to assess connectivity between each possible pair of cortical Brodmann areas. Multiple linear regression analyses were performed to evaluate the relationship between language performance and cortical necrosis and cortical disconnection.

Results—We observed that structural disconnection of Brodmann area 45 (spared by the necrotic tissue) was independently associated with naming performance, controlling for the extent of Brodmann area 45 necrosis (F=4.62; P<0.01; necrosis: β=0.43; P=0.03; disconnection β=1.21; P<0.001).

Conclusions—We suggest that cortical disconnection, as measured by the structural connectome, is an independent predictor of naming impairment in patients with chronic aphasia. The full extent of clinically relevant brain damage after an ischemic stroke may be underappreciated by visual inspection of cortical necrosis alone. (Stroke. 2014;45:988-993.)

Key Words: aphasia ◼ connectome ◼ diffusion tensor imaging ◼ stroke

Assessing the Clinical Effect of Residual Cortical Disconnection After Ischemic Strokes

Leonardo Bonilha, MD, PhD; Chris Rorden, PhD; Julius Fridriksson, PhD

Received November 16, 2013; final revision received January 31, 2014; accepted February 5, 2014.From the Department of Neurology, Medical University of South Carolina, Charleston (L.B.); and Departments of Psychology (C.R.) and Communication

Sciences and Disorders (J.F.), University of South Carolina, Columbia.The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.

113.004137/-/DC1.Correspondence to Leonardo Bonilha, MD, PhD, Division of Neurology, Comprehensive Epilepsy Center, Medical University of South Carolina,

Charleston, SC. E-mail bonilha@musc.edu© 2014 American Heart Association, Inc.

Stroke is available at http://stroke.ahajournals.org DOI: 10.1161/STROKEAHA.113.004137

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

Bonilha et al Cortical Connectivity and Aphasia Severity 989

tissue damage and impaired neural connectivity from these subjects using high-resolution MRI. Furthermore, we intro-duce new connectome-mapping techniques to accurately assess the anatomy of cortical and subcortical lesions in sub-jects with stroke.

MethodsSubjectsWe studied 39 individuals (mean age, 62.7±12.8 years; 22 men; 16 women) who had a left hemisphere ischemic stroke ≥6 months before enrolling in this study (Table I in the online-only Data Supplement). All subjects enrolled in this study were right-handed before the stroke.

All subjects were recruited from the local community. They did not have a history or imaging evidence of other previous strokes and had no history of other neurological illnesses. None of the subjects had a history of intellectual or learning disabilities. All subjects were able to ambulate either with no assistance or with the use of a cane. Patients with a history of learning disability, with a history of uncon-trolled seizures, or with more severe forms of motor impairment (ie, not able to ambulate) were excluded from this study.

All subjects signed an informed consent to participate in this study. Subjects were enrolled at the Medical University of South Carolina or at the University of South Carolina. The Institutional Review Boards at University of South Carolina and at Medical University of South Carolina approved this study.

Language TestingAll subjects underwent comprehensive behavioral testing at the time of enrollment in the study. Testing was performed by a speech pa-thologist and included the Western Aphasia Battery16 (WAB) and the Philadelphia Naming Test17 (PNT).

The main purpose of language testing was to provide an assess-ment of (1) aphasia severity and (2) naming performance. Aphasia severity was determined using the WAB aphasia quotient, a measure aphasia severity ranging from 0 to 100 (>93.8 indicates language abilities within normal limits).

Conversely, naming performance was determined based on PNT performance. The PNT is a computer-based assessment of naming in

patients with aphasia17 and includes 175 pictures representing mid- and high-frequency nouns from a word frequency list compiled by Francis et al.18

All subjects underwent a formal evaluation of speech apraxia based on the Subtest Six on the Apraxia Battery for Adults–Second Edition,19 to ensure that apraxia of speech was not a significant con-founder accounting for naming difficulties.

MRI AcquisitionAll subjects underwent MRI scanning at University of South Carolina or at Medical University of South Carolina. At both sites, MRI scan-ning was performed using the same type of MRI scanner (ie, a 3T Siemens Trio equipped with a 12-channel head coil). The MRI scan-ning protocol is described in detail in the online-only Data Supplement.

Preprocessing of MRI DataTo ensure adequate quantification of necrotic lesion burden and cor-tical connectivity, we optimized connectome-mapping methodology to evaluate structural brain properties while preserving anatomic authenticity in spite of the spatial distortions caused by the stroke lesion. MRI data preprocessing for connectome mapping involve multiple sequential steps,10 namely (1) segmentation of the cerebral cortex into multiple anatomically defined regions of interest (ROIs) and spatial registration of T1-weighted images; (2) transformation of white matter maps and cortical ROIs onto diffusion tensor imaging space; (3) fiber tracking and connectome reconstruction; and (4) cor-tical connectivity assessment. These steps can be challenging when applied to brains with large anatomic distortions (because of mass effects or pathological atrophy) caused by structural lesions (ie, ne-crotic tissue from the stroke).

Connectome methodology and necrotic lesion assessment are ex-plained in the Methods section in the online-only Data Supplement and in Figures 1 and 2.

It should be noted that this method does not take into consideration physiological asymmetries between hemispheres. Our group previ-ously assessed the degree of physiological asymmetries in the connec-tome in healthy older adults.20 None of the ROIs tested in this study are associated with a significant asymmetry in healthy subjects, except for a part of BA 45 (ie, the opercularis component of the inferior frontal

Figure 1. Image preprocessing steps used during the construction of the connectome. The top row demonstrates T1-weighted images from a repre-sentative subject. Necrotic tissue can be observed in the left hemisphere, affecting the insula and precentral dorsolateral regions. The middle row demonstrates the probabilistic maps of gray (cyan) and white matter (white) and their relationship with the necrotic lesion mask (in red, manually drawn on T2 images). The bottom row demonstrates the probabilistic map of gray matter segmented in accordance with the Brodmann area Atlas. Note the exclusion of the necrotic tissue from the segmented cortical map, as demonstrated by the 3-dimensional reconstructions on the third column.

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

990 Stroke April 2014

gyrus), which demonstrates a leftward asymmetry. These findings are in accordance with previous literature.21 Thus, left-sided fiber reduc-tions are not likely to represent a physiological pattern.

Statistical AnalysesWe performed multiple linear regression analyses with the perfor-mance on naming measures defined as the dependent variable, with the following independent variables: (1) percentage of damage to each BA area, (2) percentage number of fibers of each BA area. Each BA area was analyzed separately. The explanatory power of the re-sulting regression model was determined as R2 (proportional reduc-tion in error) with the explanatory factors of interest entered into the regression analysis using a stepwise approach. The level of statistical significance was set at P<0.05, adjusted through Bonferroni correc-tion based on the number of regressions investigated.

ResultsLanguage-Aphasia Severity and Naming Performance

Aphasia SeverityThe mean aphasia quotient for the group of subjects was 57.94 (SD, 28.27). Performances on WAB subtests were as follows (mean±SD): comprehension=7.67±1.96; fluency=5.23±3.32; speech repetition=5.25±3.71; information content=5.69±3.20. Aphasia types were distributed as follows: 38% (15/39) anomic aphasia; 38% (15/39) Broca aphasia; 5% (2/39) Wernicke aphasia; 10% (4/39) with conduction aphasia; 8% (3/39) with global aphasia. Naming performance: subjects correctly named an average of 51±33% of the PNT items (Table I in the in the online-only Data Supplement).

Necrotic Lesion LocationAll subjects exhibited a cortical/subcortical lesion in the left hemisphere, broadly distributed within the vascular territory of the left middle cerebral artery. A considerable variability in lesion location was observed within the middle cerebral artery territory. The insula and the subcortical aspect of the fron-tal operculum were the locations of maximal lesion overlap (Figure I in the online-only Data Supplement).

ConnectomeThe average connectome from all subjects is demonstrated in Figure 3. The relative reduction in connectivity for the BAs in the left hemisphere can be observed by the relative pro-portion of connectivity weight per BA in the left hemisphere when compared with the right hemisphere (Figure II in the online-only Data Supplement). This analysis excluded recipro-cal connections to the region itself. Figure III in the online-only

Data Supplement demonstrates the extent of necrosis of each BA. Interestingly, comparing the data from Figures II and III in the online-only Data Supplement, reduction in connectiv-ity was observed in regions with a higher percentage of dam-age, but it was not a perfect match, indicating that connectivity damage and cortical necrosis are complementary factors. This relationship can be appreciated in Figure IV in the online-only Data Supplement, demonstrating the correlation between necrosis and connectivity, for each ROI.

Overall Necrosis Extent and Language ImpairmentWe did not observe a relationship between overall lesion size and aphasia quotient on the WAB (r=−0.16; P=0.33), the WAB naming subscore (r=0.04; P=0.84), and correctly named items on the PNT (r=−0.07; P=0.67).

Regional Necrosis and Connectivity and Language ImpairmentWe focused our analyses on general aphasia measures (assessed through the WAB): (1) comprehension of sequential commands and (2) speech fluency; and naming performance (measured by the PNT): (1) percentage of correctly named items. We assessed brain ROIs typically associated with speech and language, namely BA 22, 37, and 45.

Figure 2. Deterministic tractography results from a representative subject (same subject as in Figure 1). Left side of the figure shows the tracts with color indicating the mean direction of each fiber. Note the relative reduction in tracked fibers in the left frontal lobe (highlighted by the dark rectangle) when compared with the right side. The resulting connectome from this subject is demonstrated on the right side of the figure. The scale bar represents: log(number of fibers). BA indicates Brodmann area.

Figure 3. The average connectome from all subjects. The scale bar represents: log(number of fibers).

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

Bonilha et al Cortical Connectivity and Aphasia Severity 991

We observed a significant relationship between number of correctly named items on the PNT and a model composed of BA 45 necrotic damage and BA 45 disconnection (F=4.62, P<0.01; necrosis: β=0.43, P=0.03; disconnection: β=1.21, P<0.001; Figure 4). We also observed a significant relation-ship between comprehension and a model composed by BA 22 necrosis and BA 22 disconnection (F=3.05, P=0.047; necrosis: β=−0.35, P=0.43; disconnection: β=47; P=0.01). However, in this model, necrosis was not independently asso-ciated with comprehension. Finally, we also observed signifi-cant relationship between fluency and a model composed by BA 45 necrosis and BA 45 disconnection (F=4.62, P<0.01; necrosis: β=2.89, P=0.13; disconnection: β=7.52, P=0.025). Similarly, in this model, necrosis alone was not an indepen-dent predictor of fluency (Figure 4).

Interestingly, aphasia severity (aphasia quotient) was not associated with a model composed by necrosis and disconnec-tion in BA 45 (P=0.34), BA 22 (P=0.8), or BA 37 (P=0.84), suggesting that specific language impairments are distinctly associated with regional necrosis and disconnection.

DiscussionIn this study, we evaluated the relationships between cortical necrosis and cortical disconnection with global aphasia mea-sures and naming performance. We introduced new imaging methodology to assess connectome abnormalities in patients with chronic stroke, and we demonstrated that cortical discon-nection and cortical necrosis are coexisting phenomena. We observed that correct naming is dependent upon the preserva-tion of cortical integrity and the preservation of cortical con-nectivity of BA 45. It is important to note that this finding does

not suggest that disconnection and necrosis in other cortical regions does not contribute to naming impairment. Rather, our study focused on a small set of cortical regions, including BA 45, to clearly demonstrate how disconnection contributes to behavioral impairment.

These findings are relevant to the understanding of the mechanisms underlying stroke-related impairments because they underscore 2 important factors: (1) cortical disconnection is an independent form of damage, which may not be read-ily appreciated by measurement of cortical necrosis and (2) cortical disconnection is a complementary factor that explains behavioral deficits, such as naming impairment.

Cortical Disconnection as an Invisible Form of DamageThe full extent of structural injury is usually appreciated only weeks or months after the ischemic event through structural MRI.22 Even though it is possible to define the extension of the necrotic lesions into subcortical regions, it is difficult to quantify, based on visual inspection alone, the magnitude of white matter reduction, particularly as it relates to fibers that connect cortical regions remote from the stroke site.

In this study, we demonstrated that cortical regions in the hemisphere affected by the stroke exhibit a reduction in con-nectivity when compared with their homologue counterparts in the nonaffected hemisphere. This observation suggests that cortical areas that are apparently intact during visual inspec-tion of MRI images exhibit a reduction in structural connec-tivity. The degree of connectivity reduction cannot be inferred based on the location of the necrosis alone. This fact is perhaps best explained in Figure 5 where preservation of connectivity of the inferior frontal gyrus may occur for some subjects with anterior supra-Sylvian necrosis but is absent for other subjects with a relatively similar extent of damage.

To assess the extent of disconnection in different cortical regions, we relied on diffusion MRI tractography. Although MRI tractography has the advantage of providing in vivo information about structural brain integrity, it is not without technical limitations.23 In fact, structural connectome mapping is a novel methodological approach in neuroscience and neu-roimaging, with growing popularity and applicability.11,12 In this study, we introduce new technical improvements to enable connectome mapping in subjects with stroke. Nevertheless, the results should be interpreted in the context of limitations of MRI tractography.

Cortical Disconnection and Naming PerformanceWe demonstrated that naming performance is associated with preservation of cortical integrity and cortical connectivity of BA 45. Importantly, cortical connectivity predicts naming per-formance when controlling for the volume of necrosis.

Our group has demonstrated that loss of white matter fibers supporting the dominant inferior frontal gyrus can lead to severe aphasia, even if the region seems to have been spared by the stroke.24 Our current study provides further evidence, suggesting that preserved cortical connectivity is an indepen-dent predictor of accurate naming performance in subjects with chronic aphasia. Furthermore, disconnection may play

Figure 4. Damage and disconnection of regions in the left hemi-sphere were independent predictors of naming performance. The x axis represents the percentage of fibers in the left hemisphere Brodmann area (BA) when compared with the homologous right hemisphere BA. The y axis represents the residual error from the correlation between the percentage of cortical damage to the left BA and the behavioral performance. Top plot, the relationship between BA 22 and auditory comprehension of sequential com-mands (comp); middle plot, between BA 45 and number of items correctly named on the Philadelphia Naming Test (correct); and bottom plot, between BA 45 and fluency. Below each plot, we demonstrate the statistical results for the multiple linear regres-sion model when BA cortical damage and BA cortical disconnec-tion were defined as independent variables and the behavioral measure as the dependent variable. The model F is shown along with β and P values associated with cortical disconnection.

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

992 Stroke April 2014

a role in aphasia recovery. Our group has demonstrated that naming recovery as a result of therapy is associated with functional modulation of the inferior frontal cortex,25 lead-ing to an increase in the number of correctly named items. However, functional modulation of the frontal region does not occur in all subjects with a partially preserved frontal cortex. Subjects who do not achieve functional recruitment fail to improve, and a possible reason underlying suboptimal frontal recruitment is the lack of structural connectivity. This theory was not tested in the study, but it could be addressed by future research.

The results described in this article highlight the impor-tance of subcortical lesions in patients with aphasia. Kreisler et al26 demonstrated that basal nuclei and subcortical lesions were frequently associated with almost all forms of language impairment. Cortical disconnection as a result of specific sub-cortical lesions may play a significant role in aphasia.27 In fact, unmeasured disconnection can account for discrepancies about the crucial anatomy supporting naming. Specifically, because lesion studies rely on the manual delineation of areas of frank necrosis, they may underappreciate damage as a result of remote cortical disconnection. Expressive deficits are commonly observed after strokes affecting the dominant inferior frontal gyrus (involving BA 44 and 45)28 However, naming involves a network of peri-Sylvian structures, and naming impairment may occur as a result of lesions affect-ing several aspects of the cortical language network. DeLeon et al29 demonstrated that several regions are essential for dis-tinct processes underlying naming. Specifically, anomia may arise from combined dysfunction involving the left anterior, inferior and posterior middle/superior temporal cortex, pos-terior inferior frontal, and inferior parietal cortex.3,29 Baldo et al30 demonstrated that damage to the left midposterior mid-dle temporal gyrus prevents the retrieval of names associated with objects (ie, lexical-semantic retrieval). These findings

complement observations from Walker et al31 and Schwartz et al32 demonstrating that semantic errors during naming are associated with damage to the left anterior and mid middle temporal gyrus, whereas phonological errors are associated with lack of integrity of the left supramarginal gyrus and infe-rior frontal cortex.33 However, note that the results of Dronkers et al28 and Schwartz et al do not completely agree with the findings from Hillis et al. Importantly, disagreement among studies may be related to 2 issues. First, naming involves mul-tiple cognitive domains, such as visual/auditory perception and processing, semantic decision, lexical retrieval, phono-logical encoding, and speech articulation.3 Second, it does not seem that factors, such as apraxia of speech, can account for our findings because it is typically associated with damage to BA 44 or premotor cortex and not BA 45.34–36 Finally, the loca-tion of damage or cortical dysfunction may be underestimated by current methods used to define the frank lesion. As such, more pervasive network damage can underlie a more salient lesion but be underappreciated by brain mapping techniques, and localization fails to define a crucial region or network.

In summary, we suggest that correct naming in subjects with chronic aphasia is dependent on the preservation of corti-cal integrity and the preservation of cortical connectivity of BA 45. We also suggest that structural evaluation of brain damage in relationship with language impairment after stroke can be improved by measuring the subcortical injury and its remote ramifications into distant cortical disconnection. This connectivity-based approach can improve the understand-ing of the mechanisms, leading to language and behavioral impairments after stroke. It may also provide evidence for degree of structural integrity supporting recovery.

Sources of FundingThis publication was supported by National Institute of Deafness and Communication Disorders (NIDCD), grant number DC009571. It was

Figure 5. This diagram demonstrates surface brain and white matter fiber anatomies for 5 representative patients to illustrate the variability in cortical lesion location and the degree of naming impairment. Subjects are demonstrated per column. The first row shows the reconstruction of the cortical surface from T1-weighted images (the stroke lesion is apparent as the area of liquefactive necrosis). The second, third, and fourth rows demonstrate reconstructed whole brain deterministic white matter tracts (viewed from the left side, right side and above, respectively). Note that that sub-jects 8 and 12 (S8 and S12, respectively) demonstrated lesions encompassing the inferior frontal gyrus; however, their performance in confrontation naming was the relatively good (the percentage of correctly named items on the PNT was, respectively, 91% [aphasia quotient {AQ}, 96], and 90% [AQ=21]). In turn, subjects 23, 24, and 25 (S23, S24, and S25, respectively) exhibited profound difficulty in naming (4% [AQ=23], 5% [AQ=31], 5%

[AQ=71]), albeit with variable necrotic lesion anatomy. Importantly, left inferior frontal fibers (highlighted by the dark rectangle) were rela-tively preserved in patients with good naming performance (S8 and S12), but largely absent in patients with poor performance (S23, S24, and S25), suggesting a common mechanism underlying naming and structural integrity.

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

Bonilha et al Cortical Connectivity and Aphasia Severity 993

also supported by the South Carolina Clinical & Translational Research (SCTR) Institute, with an academic home at the Medical University of South Carolina, through NIH Grant Number UL1 TR000062.

DisclosuresNone.

References 1. Rorden C, Karnath HO, Bonilha L. Improving lesion-symptom mapping.

J Cogn Neurosci. 2007;19:1081–1088. 2. Hillis AE. Aphasia: progress in the last quarter of a century. Neurology.

2007;69:200–213. 3. Newhart M, Ken L, Kleinman JT, Heidler-Gary J, Hillis AE. Neural

networks essential for naming and word comprehension. Cogn Behav Neurol. 2007;20:25–30.

4. Thomalla G, Glauche V, Weiller C, Röther J. Time course of wallerian degeneration after ischaemic stroke revealed by diffusion tensor imag-ing. J Neurol Neurosurg Psychiatry. 2005;76:266–268.

5. Hervé D, Molko N, Pappata S, Buffon F, LeBihan D, Bousser MG, et al. Longitudinal thalamic diffusion changes after middle cerebral artery infarcts. J Neurol Neurosurg Psychiatry. 2005;76:200–205.

6. Lama S, Qiao M, Kirton A, Sun S, Cheng E, Foniok T, et al. Imaging cor-ticospinal degeneration in neonatal rats with unilateral cerebral infarc-tion. Exp Neurol. 2011;228:192–199.

7. Wade DT, Hewer RL, David RM, Enderby PM. Aphasia after stroke: natural history and associated deficits. J Neurol Neurosurg Psychiatry. 1986;49:11–16.

8. Damasio AR. Aphasia. N Engl J Med. 1992;326:531–539. 9. Croquelois A, Bogousslavsky J. Stroke aphasia: 1,500 consecutive cases.

Cerebrovasc Dis. 2011;31:392–399. 10. Daducci A, Gerhard S, Griffa A, Lemkaddem A, Cammoun L, Gigandet

X, et al. The connectome mapper: an open-source processing pipeline to map connectomes with MRI. PLoS One. 2012;7:e48121.

11. Hagmann P, Cammoun L, Gigandet X, Gerhard S, Grant PE, Wedeen V, et al. MR connectomics: principles and challenges. J Neurosci Methods. 2010;194:34–45.

12. Hagmann P, Cammoun L, Gigandet X, Meuli R, Honey CJ, Wedeen VJ, et al. Mapping the structural core of human cerebral cortex. PLoS Biol. 2008;6:e159.

13. Sporns O. The human connectome: origins and challenges. Neuroimage. 2013;80:53–61.

14. Sporns O. The human connectome: a complex network. Ann N Y Acad Sci. 2011;1224:109–125.

15. Sporns O, Tononi G, Kötter R. The human connectome: A structural description of the human brain. PLoS Comput Biol. 2005;1:e42.

16. Kertesz A. The Western Aphasia Battery—Revised. New York: Grune & Stratton; 2007.

17. Roach A, Schwartz MF, Martin N, Grewal RS, Brecher A. The phila-delphia naming test: scoring and rationale. Clinical Aphasiology. 1996;24:121–133.

18. Francis WN, Kucera H, Mackie AW. Frequency Analysis of English Usage: Lexicon and Grammar. Boston: Houghton Mifflin; 1982.

19. Dabul B. Apraxia Battery for Adults. Examiner’s Manual. Austin, TX: Pro-Ed; 2000.

20. Bonilha L, Nesland T, Rorden C, Fridriksson J. Asymmetry of the structural brain connectome in healthy older adults. Front Psychiatry. 2014;4:186.

21. Keller SS, Crow T, Foundas A, Amunts K, Roberts N. Broca’s area: nomenclature, anatomy, typology and asymmetry. Brain Lang. 2009;109:29–48.

22. Caplan LR. Caplan’s Stroke: a Clinical Approach. Philadelphia: Elsevier/Saunders; 2009.

23. Gigandet X, Griffa A, Kober T, Daducci A, Gilbert G, Connelly A, et al. A connectome-based comparison of diffusion MRI schemes. PLoS One. 2013;8:e75061.

24. Fridriksson J, Bonilha L, Rorden C. Severe Broca’s aphasia without Broca’s area damage. Behav Neurol. 2007;18:237–238.

25. Fridriksson J, Richardson JD, Fillmore P, Cai B. Left hemisphere plastic-ity and aphasia recovery. Neuroimage. 2012;60:854–863.

26. Kreisler A, Godefroy O, Delmaire C, Debachy B, Leclercq M, Pruvo JP, et al. The anatomy of aphasia revisited. Neurology. 2000;54:1117–1123.

27. Bonilha L, Fridriksson J. Subcortical damage and white matter discon-nection associated with non-fluent speech. Brain. 2009;132(pt 6):e108.

28. Dronkers NF, Plaisant O, Iba-Zizen MT, Cabanis EA. Paul Broca’s his-toric cases: high resolution MR imaging of the brains of Leborgne and Lelong. Brain. 2007;130(pt 5):1432–1441.

29. DeLeon J, Gottesman RF, Kleinman JT, Newhart M, Davis C, Heidler-Gary J, et al. Neural regions essential for distinct cognitive pro-cesses underlying picture naming. Brain. 2007;130(pt 5):1408–1422.

30. Baldo JV, Arévalo A, Patterson JP, Dronkers NF. Grey and white matter correlates of picture naming: evidence from a voxel-based lesion analy-sis of the Boston Naming Test. Cortex. 2013;49:658–667.

31. Walker GM, Schwartz MF, Kimberg DY, Faseyitan O, Brecher A, Dell GS, et al. Support for anterior temporal involvement in semantic error production in aphasia: new evidence from VLSM. Brain Lang. 2011;117:110–122.

32. Schwartz MF, Kimberg DY, Walker GM, Faseyitan O, Brecher A, Dell GS, et al. Anterior temporal involvement in semantic word retrieval: voxel-based lesion-symptom mapping evidence from aphasia. Brain. 2009;132(pt 12):3411–3427.

33. Schwartz MF, Faseyitan O, Kim J, Coslett HB. The dorsal stream con-tribution to phonological retrieval in object naming. Brain. 2012;135(pt 12):3799–3814.

34. Whitwell JL, Duffy JR, Strand EA, Xia R, Mandrekar J, Machulda MM, et al. Distinct regional anatomic and functional correlates of neurode-generative apraxia of speech and aphasia: an MRI and FDG-PET study. Brain Lang. 2013;125:245–252.

35. Josephs KA, Duffy JR, Strand EA, Machulda MM, Senjem ML, Master AV, et al. Characterizing a neurodegenerative syndrome: primary pro-gressive apraxia of speech. Brain. 2012;135(pt 5):1522–1536.

36. Richardson JD, Fillmore P, Rorden C, Lapointe LL, Fridriksson J. Re-establishing Broca’s initial findings. Brain Lang. 2012;123:125–130.

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

Leonardo Bonilha, Chris Rorden and Julius FridrikssonAssessing the Clinical Effect of Residual Cortical Disconnection After Ischemic Strokes

Print ISSN: 0039-2499. Online ISSN: 1524-4628 Copyright © 2014 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Stroke doi: 10.1161/STROKEAHA.113.004137

2014;45:988-993; originally published online March 11, 2014;Stroke. 

http://stroke.ahajournals.org/content/45/4/988World Wide Web at:

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

http://stroke.ahajournals.org/content/suppl/2016/04/03/STROKEAHA.113.004137.DC2.html http://stroke.ahajournals.org/content/suppl/2014/03/11/STROKEAHA.113.004137.DC1.html

Data Supplement (unedited) at:

  http://stroke.ahajournals.org//subscriptions/

is online at: Stroke Information about subscribing to Subscriptions: 

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

  document. Permissions and Rights Question and Answer process is available in the

Request Permissions in the middle column of the Web page under Services. Further information about thisOnce the online version of the published article for which permission is being requested is located, click

can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.Strokein Requests for permissions to reproduce figures, tables, or portions of articles originally publishedPermissions:

by guest on Novem

ber 4, 2016http://stroke.ahajournals.org/

Dow

nloaded from

SUPPLEMENTAL MATERIAL: Methods: 1) MRI acquisition parameters

A) T1-weighted images (3D MP-RAGE, TR = 2250ms, TE = 4.15ms, 256×256matrix,

256×256mm FOV, parallel imaging GRAPPA=2, 80 reference lines, TA=377s);

B) Diffusion EPI scan (30-directions with b=1000 s/mm2 and b=2000 s/mm2, TR =

6100ms, TE = 101ms, 82×82 matrix, 222x222mm FOV, parallel imaging GRAPPA=2,

80 45 contiguous 2.7mm axial slices, TA=390s).

2) Cortical segmentation and spatial registration of T1 weighted images.

We employed an extension of the software Statistical Parametric Mapping (SPM)

entitled “Clinical Toolbox” developed by our group1. The Clinical Toolbox utilizes a cost-

function approach2 to normalize the brain into the standard stereotaxic space (MNI

space). This step employs a manually-defined map of the stroke lesion, drawn by one of

the authors (Bonilha) to weigh tissue influence on normalization. Notably, the Clinical

Toolbox exploits SPM’s unified normalization-segmentation subroutines to yield

probabilistic gray and white matter tissue maps3. These were essential to guide

subsequent connectivity assessment steps. Linear and non-linear normalization

parameters obtained from the normalization of the brain to standard space were

inverted and applied to a BA ROI Atlas in standard space (distributed with MRIcro4).

The probabilistic map of gray matter in native T1 space was segmented into a map of

cortical BA ROIs. Importantly, the areas originally involved in the hand-drawn lesion

were excluded from this step in order to eliminate the erroneous segmentation of the

necrotic tissue into viable cortical regions. These steps are demonstrated in Figure 1.

3) Transforming white matter map and cortical ROIs onto DTI space.

To improve registration between T1 and DTI spaces, the native volumetric T2

weighted image was linearly co-registered onto the native T1 image using mutual

information algorithms in SPM. This step generated a T2 weighted image matched to

the native T1 image (the “registered T2 image”). The registered T2 image was linearly

co-registered onto the B0 image using FMRIB's Linear Image Registration Tool (FLIRT).

The transformation matrices were then applied to the map of segmented cortical ROIs

and to the white matter probabilistic tissue map, yielding cortical ROIs and white matter

maps in DTI space.

4) Fiber tracking and connectome reconstruction.

Whole brain tractography was reconstructed in DTI space using the software

Diffusion Toolkit (DTK)5. Acquisition geometry and gradients were updated using the

tool dcm2nii, as part of the software MRIcron. Tractography was reconstructed using a

maximal angle threshold of 45 degrees, in accordance with the Fiber Assignment by

Continuous Tracking (FACT) algorithm6. Two restriction maps were applied. The first

one was a dynamic contrast range based on the diffusion-weighted image (DWI), as

part of default DTK parameters. Importantly, the second restriction mask was the

probabilistic white matter map transformed into DTI space. By applying a white matter

restriction mask, we ensured that important quality benchmarks were met, namely: 1)

tractography did not attempt to reconstruct fibers in the location of the necrotic lesion,

where random diffusion can lead to erroneous fiber tracking and misrepresentation of

connectivity; 2) the termination boundaries overlapped and extended into the cortical

ROI map, thereby ensuring an accurate end point calculation for each white matter

fiber. Once cortical ROIs and white matter maps were registered into DTI space, each

and every white matter fiber was evaluated regarding its origin and termination. For

each possible pair of BAs, the number of fibers connecting the pair was computed and

recorded in a connectivity matrix. These steps were performed through in-house scripts

written in MATLAB.

5) Cortical connectivity.

In order to assess connectivity, we computed the weighted sum of all

connections to four cortical areas that are typically implicated in language processing:

BAs 22, 37, 44, and 45. Certainly, other regions play a role in language processing.

However, we chose to focus on these four areas to highlight the potential importance

of disconnection in functional impairment. Finally, to assess fiber reduction, the

percentage of fibers compared with the homologous BA in the right hemisphere is

calculated (yielding the “percentage number of fibers”, as referred to in the sections

below). These steps are performed through in-house scripts written in MATLAB.

Importantly, this step aimed to eliminate the effect of bias caused by subjects having

been scanned in different scanners or with different DTI sequences. Since each

patient’s connectivity measures were adjusted based on the patient’s contralateral

hemisphere, the final connectivity measure was normalized based on each subject’s

imaging properties and therefore comparable across all subjects.

6) Necrotic lesion burden.

The percentage of damage to each cortical ROI is calculated by overlaying the

hand drawn lesion map onto the parcellated cortical map in native T1 space, where the

percentage of damage to each BA is calculated. These steps are performed through in-

house scripts written in MATLAB.

Online Table I- Demographic and language information. Language performance was assessed through the Western Aphasia Battery (WAB) and Philadelphia Naming Test (PNT). Legends: RH=right handed; LH=left handed; M=male, F=female; W=white; AA= African-American; SD=standard deviation; AQ= aphasia quotient (WAB); Comp=comprehension (WAB); Flu=fluency (WAB); Repet=repetition (WAB); NamingSub=Naming Subscore (WAB); Naming=fraction of correctly named items (PNT).

Subject Handedness Age Gender Race Aphasia

Type AQ Comp Flu Repet NamingSub (WAB)

Naming (PNT)

1 RH 77 M W Anomic 87.8 8 9 9.2 8.7 0.65

2 RH 71 M W Broca 62.7 9.35 4 3.4 7.6 0.32

3 RH 69 M W Broca 87.2 8.9 9 8.2 8.5 0.73

4 RH 38 M W Broca 41.8 7.5 4 2.8 3.6 0.42

5 RH 67 M W Anomic 77.8 9 4 9.4 8.5 0.5

6 RH 50 M AA Anomic 80.8 9.8 8 8.3 8.3 0.4

7 RH 48 F W Anomic 79 9 6 8.2 8.6 0.06

8 RH 65 M W Anomic 96 10 9 9.2 10 0.91

9 RH 39 F W Broca 32 8 1 2.5 2.7 0.04

10 RH 58 M B Anomic 83.2 9.3 7 9 9.3 0.78

11 RH 77 M W Broca 21 4 1 2 1

12 RH 68 F B Broca 21 10 1 0 0 0.9

13 RH 64 M B Conduction 80 9 9 6.1 7 0.54

14 RH 75 F W Anomic 92 9 9 9.6 9 0.83

15 RH 62 M W Anomic 86 10 8 9.4 .

16 RH 70 M W Broca 51 7 2 7.6 3.7 0.51

17 RH 49 F W Broca 43 7 4 1.2 2.5 0.63

18 RH 86 F W Anomic 68.7 7.25 7 7.4 6.7 0.38

19 RH 57 M W Wernicke 31 6 8 0.6 0.8 0.84

20 RH 60 F W Anomic 86 9 8 9.6 8.6 0.82

21 RH 57 F W Anomic 95 9 10 9.6 9.2 0.85

22 RH 63 M W Anomic 92.1 9.85 9 9.8 8.4 0.81

23 RH 60 F W Global 23 6 0 1.7 0.04

24 RH 52 F W Broca 31 7 2 2.6 2.9 0.05

25 RH 57 M W Broca 71 7 5 7.4 8.5 0.05

26 RH 82 F W Conduction 70 7 7 6 6.6 0.37

27 RH 47 M W Broca 26 5 1 1.4 .

28 RH 62 M W Broca 48 6 4 5.6 1.9 0.08

29 RH 60 M W Wernicke 31 5 5 0.3 0.6 0.92

30 RH 62 F W Global 17 6 1 0.5 1 0.88

31 RH 76 M W Wernicke 31 4 8 0.8 0.6 0.01

32 RH 80 F W Global 25 5 1 1.5 .

33 RH 43 F W Anomic 82 8 8 8.2 9.1 0.87

34 RH 71 M W Broca 26 10 0 0.9 2 0.01

35 RH 77 M W Anomic 95 10 9 10 9.5 0.93

36 RH 37 M B Global 6 3 0 0 0.38

37 RH 71 F W Wernicke 47 6 5 2.4 0.23

38 RH 48 M W Broca 44 8 2 2.4 4.3 0.43

39 RH 78 F W Anomic 90 10 9 10 . Average (±SD) .

62±12.7 . . .

57.9±27.9

7.7±1.9

5.2±3.3

5.3±3.7 5.8±3.4 0.51±0.33

Online Figure I- The spatial distribution of necrotic lesion masks (transformed into

standard MNI stereotaxic space) from all subjects included in this study is

demonstrated by overlaying the lesion masks onto a T1-weighted atlas. Voxels

are color-coded in accordance with the number of individuals with a lesion

affecting each specific that specific voxel.

Online Figure II- The relative number of fibers in representative BA ROIs in the left

hemisphere, compared with the homologue BA in the right hemisphere (0= no

remaining fibers, 1= left hemisphere fibers are at least equal in number to right

hemisphere fibers). The box plots demonstrated the distributions across all

subjects per each ROI.

Online Figure III- The degree of sparing from necrosis of representative BA ROIs

(ranging from 1- the ROI was completely spared, to 0 – completely included by

the necrosis). The box plots demonstrated the distributions across all subjects

per each ROI.

Online Figure IV- The relationship between necrosis and disconnection for

representative Bas. The scatter plots demonstrate percentage of necrosis (y-

axis, 0= no necrosis; 1= complete necrosis) and percentage of remaining fibers

compared with the right hemisphere (0= no remaining fibers, 1= left hemisphere

fibers are at least equal in number to right hemisphere fibers). The correlation

coefficient for each ROI is also demonstrated underlying each scatter plot.

subjects per each ROI.

References 1. Rorden C, Bonilha L, Fridriksson J, Bender B, Karnath HO. Age-specific ct and

mri templates for spatial normalization. Neuroimage. 2012;61:957-965 2. Brett M, Leff AP, Rorden C, Ashburner J. Spatial normalization of brain images

with focal lesions using cost function masking. Neuroimage. 2001;14:486-500

3. Ashburner J, Friston KJ. Unified segmentation. Neuroimage. 2005;26:839-851 4. Rorden C, Brett M. Stereotaxic display of brain lesions. Behav Neurol.

2000;12:191-200 5. Wang R, Benner T, Sorensen AG, Wedeen VJ. Diffusion toolkit: A software

package for diffusion imaging data processing and tractography. Proc. Intl. Soc. Mag. Reson. Med. 2007 abstract #3720

6. Mori S, van Zijl PC. Fiber tracking: Principles and strategies - a technical review. NMR in biomedicine. 2002;15:468-480

iris

22 Stroke 日本語版 Vol. 9, No. 2

虚血性脳卒中後に残存する皮質離断の臨床的影響に関する評価

Assessing the Clinical Effect of Residual Cortical Disconnection After Ischemic Strokes

Leonardo Bonilha, MD, PhD 1 ; Chris Rorden, PhD 2 ; Julius Fridriksson, PhD 3 1 Department of Neurology, Medical University of South Carolina, Charleston; 2 Departments of Psychology; and 3 Communication Sciences and Disorders, University of South Carolina, Columbia

背景および目的： 脳卒中後の慢性言語障害（失語症）と虚血性脳損傷の関連性に関する研究では通常，MRI による脳壊死した範囲の測定に基づいて評価している。しかし，患者に発現する障害は，病変の位置や大きさに基づく予測よりも重度な可能性があることが臨床所見から示唆される。この現象は一般に，皮質離断の結果であるとして説明されている。皮質離断が臨床症状の原因であるか否かを確認するため，本研究では脳卒中後に慢性失語症を発症した患者において，言語障害と脳の構造的接続性［コネクトーム（connectome）］との関連性を評価した。 方法： 39 例の慢性失語症患者に言語評価とMRI を実施した。MRI データに基づいて，T1強調画像および拡散テンソル画像（DTI）から個々のコネクトームを再構築した。神経線維の決定論的トラクトグラフィー（deterministic

tractography）を用いて，ブロードマンの大脳皮質分類において可能性のある各組み合わせ間での接続性を評価した。多重線形回帰分析を行い，言語機能と，皮質壊死および皮質離断との関連性を評価した。 結果： 壊死の範囲で調整すると，ブロードマン 45 野の構造的な離断（壊死組織を免れた）が呼称機能に独立して関連していることが観察された（ F ＝ 4.62， p ＜ 0.01，壊死： β ＝ 0.43， p ＝ 0.03，離断： β ＝ 1.21， p ＜ 0.001）。 結論： 本研究によって，構造的コネクトームにより測定された皮質離断は慢性失語症患者の呼称障害の独立予測因子であることが示唆された。虚血性脳卒中後の臨床的に重要な脳障害の全体像は，皮質壊死の視覚的検査のみでは十分に評価されていない可能性がある。

Stroke 2014; 45: 988-993

Abstract

図 5

典型的な慢性失語症患者5例の脳表面および白質線維の解剖学的構造を示し，患者により異なる皮質病変部位と呼称障害の重症度を示す。各被験者の画像を縦一列に示す。最上列は，T1強調画像による皮質表面の再構築を示す（脳卒中病変は液化壊死の領域として明示されている）。2，3，4列は再構築された全脳の決定論的白質路を示す（上から順に左側，右側，上部からの表示）。注目すべきは，被験者8（S8）と12（S12）において下前頭回を含む病変が認められたが，対面呼称テストの結果は比較的良好であったことである［各患者のテストの正解率は，S8は91%｛失語指数（AQ） ＝ 96｝，S12は 90%（AQ ＝ 21）］。一方，被験者23（S23），24（S24），25（S25）では，壊死病変の解剖学的形態は様々であったが深刻な呼称障害を呈していた［S23は 4%（AQ ＝ 23），S24は 5%（AQ ＝ 31），S25は 5%（AQ ＝ 71）］。重要なことは，呼称機能が良好な患者（S8，S12）では左下前頭回の神経線維（黒い長方形で囲んだ部分）が比較的維持されていたが，機能が不良な患者（S23，S24，S25）では大きく欠損していたことである。これらの所見から，呼称機能と構造的健全性の根底には共通の機序があることが示唆される。

Stroke-J-v9-i2-Ab.indd 22Stroke-J-v9-i2-Ab.indd 22 8/7/2014 3:49:10 PM8/7/2014 3:49:10 PM