Abstract
Objective: To determine lesion locations associated with the various types of aphasic disorders in patients with stroke.
Background: The anatomy of aphasia has been challenged by several recent studies. Discrepancies are likely to be due to methodologic issues.
Methods: We examined 107 patients with a standardized aphasia battery and MRI. Three examiners blinded to the clinical data rated signal abnormalities in 69 predetermined regions of interest. The statistical procedure used classification tree testing, which selected regions associated with each aphasic disorder.
Results: 1) Nonfluent aphasia depended on the presence of frontal or putaminal lesions; 2) repetition disorder on insula-external capsule lesions; 3) comprehension disorder on posterior lesions of the temporal gyri; 4) phonemic paraphasia on external capsule lesions extending either to the posterior part of the temporal lobe or to the internal capsule; 5) verbal paraphasia on temporal or caudate lesions; and 6) perseveration on caudate lesions. These analyses correctly classified 67% to 94% of patients.
Conclusions: Lesion location is the main determinant of aphasic disorders at the acute stage. Most clinical–radiologic correlations supported the classic anatomy of aphasia.
Many postmortem and radiologic studies have documented the pattern of associations between circumscribed brain lesions and aphasic syndromes. Most classic clinical–anatomic correlations have been established in small series of patients using postmortem analysis.1-4 Broca’s aphasia is mainly associated with a lesion of the left inferior frontal area, frequently extending to subcortical or insular regions.5,6 Wernicke’s aphasia is usually associated with a lesion of the posterior part of the lateral temporal areas and the underlying white matter.6 Global aphasia is related to extensive anterior–posterior lesions of the left hemisphere. Conduction aphasia is associated with lesion of the left arcuate fasciculus7,8 or of the left supramarginal gyrus,8,9 sometimes extending to the temporal cortex.7
Several studies conducted in large groups of patients have systematically examined lesion locations according to aphasia type.10-12 They have shown that an unexpectedly large proportion of aphasias deviate from classic clinical–anatomic correlations. Extreme deviations include lesion of language areas without aphasia, nonfluent aphasia with posterior lesions, or global aphasia in small and circumscribed lesions.11 Exceptions to clinical–anatomic correlations are more frequent in subcortical lesions, especially of the striatum where Broca’s, transcortical, Wernicke’s, conduction, global, or nonclassified aphasias have been reported.5,10,13-16 Several studies formally examined the associations between aphasic syndromes and lesion locations using a systematic neuroradiologic analysis. They failed to demonstrate any consistent association,10 especially in subcortical lesions,12,14 and questioned the existence of the anatomy of aphasia.12
Such discrepancies may be due to methodologic issues. The validity of brain–behavior relationship study depends on 1) the reliability of the measurement of both deficit and lesion, 2) the brain pathology,17 3) the systematic neuroradiologic evaluation of all brain regions, 4) the use of statistical analysis based on systematic covariation of deficit and lesion location, and 5) the use of nonlinear statistical procedures, at least when the deficit depends on multiple lesions.18 The reliability of the measure of the deficit is a major concern because aphasic syndromes are defined according to several criteria and hence include patients with various types and severity of language disorders.6,19 The use of aphasic disorders provides a more reliable measure. Lesion location has frequently been examined using CT scans, leading to possible underestimation of lesion extension.20 The study of brain–behavior relationship assumes that abnormality of imaging signal is associated with major dysfunction or death of previously intact neurons. This condition is met by acute brain damage such as stroke.17 The experimental design of most studies has been based on patient groupings or on the determination of the site where lesions overlap. Such designs may be misleading in situations where the same deficit depends on multiple lesions.18 These considerations support the need for a re-examination of the anatomy of aphasia. We previously proposed different modes of brain–behavior relationships and examined the relevance of statistical procedures18; this methodology was used in the current study.
Our aim was to determine lesion locations associated with aphasic disorders in stroke patients.
Patients.
From May 1994 to March 1998, 493 patients admitted to the Lille Stroke Department for an acute stroke were referred to a speech therapist for language disorders. The general organization of this department has been described elsewhere.21 A total of 386 patients were excluded for the following reasons: 1) absence of cerebral MRI owing to contraindication, early death, severe condition, or unavailability of the machine; 2) previous stroke or dementia; 3) extensive white matter abnormalities; or 4) aphasia related to pure right-brain damage. The remaining 107 patients were included (mean age 56 years; range 17 to 86; 54% men). Ninety-six patients were right-handed, five were left-handed, and six were ambidextrous. The index stroke was a cerebral infarct in 88 patients and a cerebral hemorrhage in 19 patients. Fifty-six patients had motor weakness of the preferred hand.
Methods.
Language assessment.
Language was assessed within 4 weeks after stroke when the neurologic condition became favorable. The battery was mainly based on the Montreal–Toulouse battery22 and some subtests of the Boston Diagnosis Aphasia Examination.23,24 Spontaneous fluency was rated using a three-point scale (Appendix). Criteria for classic aphasic syndromes (global, Broca, Wernicke, conduction, transcortical motor and sensory, anomic) were those of the Montreal–Toulouse battery.22 We added the two following types of language disorders: subcortical aphasia, when hypophonia was associated with the criteria of transcortical motor aphasia10,13; and word-finding difficulties, when the sole abnormality consisted of mild naming impairment. The aphasia was considered as nonclassified when the disorder did not meet criteria for any of the classic aphasic syndromes.
Neuroradiologic assessment.
MRI was performed within 3 months and usually in the first month after stroke. Three independent examiners blinded to language examination rated the signal intensity using axial and coronal slices (T2-weighted turbo spin-echo sequences) as 0 (no lesion) or 1 (lesion). Thirty-four regions of interest (ROI) in each hemisphere were determined according to the coplanar stereotaxic atlas of the human brain25 (table 1). The superior and middle temporal gyri were divided into anterior and posterior parts according to a vertical plane perpendicular to the bicommissural plane and just posterior to the Heschl gyrus. The centrum semiovale was divided into four regions, according to the anterior–posterior axis: the line joining the inferior frontal sulcus and the cingulate sulcus separated the far anterior and the middle anterior regions; the central sulcus separated the middle anterior and the middle posterior regions; and the parieto-occipital sulcus separated the middle posterior and the far posterior regions. The signal abnormality of the arcuate fasciculus was examined according to MRI criteria of Kasai et al.26 White matter and dilatation of perivascular spaces were rated separately in the hemisphere contralateral to the index stroke. Fazekas criteria27 were used to evaluate periventricular (0: no abnormality; 1: abnormalities restricted to the anterior and posterior horn; 2: nonconfluent but large abnormalities; 3: confluent abnormalities around the ventricles) and white matter abnormalities (0: no abnormality; 1: presence of some punctiform abnormalities; 2: multiple, nonpunctiform, nonconfluent abnormalities; 3: confluent abnormalities). Signal abnormalities of the basal ganglia were considered as dilatation of perivascular spaces28,29 when characterized by a diameter lower than 5 mm, an ascendant and curved orientation, and a location in the inferior part of the central nuclei. Internal capsule and brainstem abnormalities related to degeneration of the pyramidal tract were not rated.
Regions of interest
The interobserver agreement was moderate (κ = 0.55; p < 0.00001). On 7383 ROIs rated by examiners, 848 (11%) were re-examined by the three examiners until a consensus was reached.
Statistical analysis.
Two main statistical analyses were conducted. The first examined the influence of age on lesion locations. It used a correlation analysis and a group comparison between anterior and posterior lesions. Correlation analyses were performed using Pearson test and group comparison using the Kruskal-Wallis test and Mann-Whitney U test for posthoc analysis. The second selected ROI abnormalities associated with aphasic disorders. The nine dependent variables are presented in the Appendix. Independent variables were the presence of a lesion (0 = normal; 1 = lesion) in the corpus callosum and 31 left hemispheric ROIs (see table 1). ROI selection used a classification tree analysis30 according to a previously validated method.18 p Values lower than 0.05 were regarded as significant, otherwise indicated. Statistical analysis was performed using SAS31 and SIPINA32 software.
Results.
Neuroradiologic assessment.
Signal abnormality was observed in 815 ROIs (table 2). Periventricular abnormalities were found in 54 patients and white matter abnormalities in 58 patients. Two MRIs were considered normal.
Neuroradiologic findings
Influence of age on lesion locations was evaluated with the Pearson correlation test. p Values (adjusted for multiple analyses) lower than 0.002 were regarded as significant. The only significant correlations concerned age and periventricular abnormalities (R = 0.57) and age and white matter abnormalities (R = 0.53). The influence of age on the anterior–posterior distribution of lesions was assessed by grouping patients with cortical lesion restricted to precentral region (n = 5) or postcentral region (n = 26) or involving anterior–posterior regions (n = 55). Age did not significantly differ according to lesion location (precentral lesion [mean ± SD]: 66.8 ± 6.9 years; postcentral lesion: 60.9 ± 16.7 years; anterior–posterior lesion: 52.9 ± 17.6 years).
Aphasic syndromes.
The frequency of aphasic syndromes and the main lesions are reported in table 3. Pure dysarthria was observed in eight patients with various lesions. The current distribution of aphasic syndromes did not differ from that observed in our department (p > 0.8). The aphasic syndromes did not differ across stroke type (p > 0.3). Age did not significantly differ across aphasic syndromes. Language abnormalities related to pure subcortical lesions were analyzed specifically. Pure deep lesion was observed in 14 patients. Three had a subcortical aphasia related to a left thalamic lesion, extending in two cases to the internal capsule, centrum semiovale, and striatum. Two patients with a transcortical motor aphasia, two with word-finding difficulty, two with dysarthria, and one with Broca’s aphasia had a left lesion involving the centrum semiovale. One patient with anomic aphasia had a left thalamic lesion. Finally, the three patients without aphasia had left lenticular, internal capsule, and centrum semiovale lesions, extending to the caudate nucleus in one. No patient with pure deep lesion had global, transcortical sensory, or Wernicke’s aphasia.
Aphasic syndromes
These results indicated that 1) Broca’s aphasia was mainly associated with anterior and insular lesions; 2) Wernicke’s aphasia, with temporo–parietal and insular lesions; 3) global aphasia, with large antero–posterior lesions usually involving the cortex; 4) subcortical lesions were associated with subcortical, transcortical motor, and Broca’s aphasia, but never with global, transcortical sensory, or Wernicke’s aphasia; and 5) age did not significantly differ across aphasic syndromes.
Aphasic disorders.
The results are presented in table 4. Nonfluent aphasia was mainly associated with anterior cortico–subcortical lesions (figure). Repetition impairment was observed in three patients without insular lesion: one had a left lesion of Heschl gyrus and two had a left thalamic lesion. The role of damage to arcuate fasciculus and inferior parietal lobulus, frequently reported in repetition disorder, was further examined. No repetition impairment was observed in the six patients with lesion of the arcuate fasciculus restricted to its portion in the centrum semiovale. Lesion of the inferior parietal lobulus was frequently associated with insular damage and did not predict repetition disorder (Fisher’s exact test: p > 0.5 after controlling for lesion covariance). Comprehension disorder was observed in five patients without damage to the posterior part of temporal gyri or to the inferior frontal gyrus: three had a left lesion of the temporal isthmus, one had a left lesion of Heschl gyrus, and one had a left lesion of the inferior parietal lobulus.
Lesions associated with the main aphasic disorders
Figure. Lesion locations (gray areas) associated with the main aphasic disorders. Added lines represent regions of interest.
Discussion.
We found that the main determinant of aphasic disorders was the location of the lesion, with the following patterns: 1) mutism depended on fronto–putaminal lesions; 2) low fluency, on lesions of inferior frontal gyrus and putamen, or lesions of the anterior centrum semiovale extending either to the putamen or to the inferior parietal lobulus; 3) repetition disorder, on lesions of the external capsule and the posterior arm of the internal capsule; 4) oral comprehension disorder, on lesions of the posterior part of temporal gyri extending to the external capsule, or on lesions of the inferior frontal gyus; 5) impairment of picture naming and word-finding difficulty depended on a large variety of lesions involving the anterior and posterior cortex, or subcortical regions including the thalamus; 6) verbal paraphasia depended mainly on temporal or caudate lesions; 7) phonemic paraphasia, on external capsule damage extending either to the posterior part of the temporal lobe or to the internal capsule; and 8) perseveration, on lesion of the head of caudate nucleus.
These results contrast with previous studies reporting a large proportion of aphasias deviating from classic clinical–anatomic correlations,10-12 and with studies failing to demonstrate a reliable relationship between aphasic syndromes and lesion locations.12,14 We did not find any clear exception to the classic anatomy of aphasic syndromes. At the level of aphasic disorder, all models correctly classified a large proportion of patients. Discrepancies between studies are likely to be due to methodologic issues: 1) this study evaluated stroke patients using MRI and assessed the reliability of neuroradiologic analysis; 2) we assessed lesions associated with aphasic disorders whereas most previous studies examined the anatomy of aphasic syndromes; and 3) we used a previously validated methodology of clinical–anatomic correlations.18
We found that the main determinant of aphasic disorder is the lesion location. The present clinical–radiologic correlations fit classic data in aphasiology and provide additional information on the anatomy of language disorders. Mutism was associated with large frontal lesions extending to the putamen, a finding consistent with the anatomy of severe Broca’s and transcortical motor aphasias.5,6,33,34 Mutism has also been reported in frontomedial lesion,33,35 but no patients with frontomedial damage presented with mutism in the current study. Low spontaneous fluency was related to lesion of the inferior frontal gyrus, putamen, or anterior part of centrum semiovale, consistent with the anatomy of nonfluent aphasia such as Broca’s, transcortical motor, and aphasia related to lesion of the striatum and the thalamus.6 Lesion of the insula–external capsule region, which involves the arcuate fasciculus, accounted for repetition disorder. This finding is consistent with previous studies on conduction apha-sia.8 The role of damage to the supramarginalis gyrus and to the arcuate fasciculus outside the external capsule, frequently suggested in conduction aphasia,9 was not confirmed. This discrepancy may be due to the absence of patients with conduction aphasia in the current study. Oral comprehension disorder depended mainly on lesion of the posterior part of the temporal region, a finding consistent with previous results in Wernicke’s and transcortical sensory aphasias.6,36 The current test of oral comprehension also assessed syntactic comprehension, which depends on the left inferior frontal gyrus.37 This might account for the minor role of lesion of the left inferior frontal gyrus observed in the current study. The influence of damage to Heschl gyrus and temporal isthmus has already been suggested.38 The large distribution of lesions associated with word-finding difficulty including perisylvian and thalamo–striatal regions has already been reported.5,10,13,19 Verbal paraphasia mainly depended on lesions of the posterior temporalregion, of the head of the caudate nucleus, or both. Damage to the posterior temporal region is usual in Wernicke’s aphasia whereas lesion of the head of the caudate nucleus and anterior temporal cortex has not been documented in verbal paraphasia to our knowledge. Phonemic paraphasia was associated with external capsule damage extending to the posterior part of the temporal lobe or to the internal capsule. The classic role of the supramarginalis gyrus7-9 was not supported and this is probably due to the absence of conduction aphasia in the current study. The role of lesion of the head of the left caudate nucleus in perseverative errors on a picture naming test has already been suggested.39 It is consistent with the general function of the caudate in response selection.40,41
This study demonstrates close associations between aphasic disorders and specific lesions, and this supports the existence of the anatomy of aphasia. From a clinical point of view, these results support the usefulness of classic aphasiology at least in stroke, which is the leading cause of aphasia.
Appendix
Language evaluation
Spontaneous verbal fluency
0 = normal
1 = nonfluent aphasia (low rate of speech, production of a few words together, or difficulty in initiating speech)
2 = mutism or stereotyped utterances
Automatic speech
Word and sentence repetition
Oral comprehension (word and sentence–picture matching)
Picture naming test
Word-finding difficulty
Verbal paraphasia
Phonemic paraphasia
Perseveration
Acknowledgments
Supported by Ministère de l’Enseignement et de la Recherche, Research Group on Cognition in Vascular and Degenerative Disorders (grant EA 2691).
- Received July 19, 1999.
- Accepted November 13, 1999.
References
Disputes & Debates: Rapid online correspondence
NOTE: All authors' disclosures must be entered and current in our database before comments can be posted. Enter and update disclosures at http://submit.neurology.org. Exception: replies to comments concerning an article you originally authored do not require updated disclosures.
- Stay timely. Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- 200 words maximum.
- 5 references maximum. Reference 1 must be the article on which you are commenting.
- 5 authors maximum. Exception: replies can include all original authors of the article.
- Submitted comments are subject to editing and editor review prior to posting.