Subcortical aphasia

Aphasia due to subcortical vascular events occurs ^[1-14] after either ischemic infarction or hemorrhage of the striatocapsular region, ^[5,12] thalamus, ^[7,9] or paraventricular white matter (PVWM) ^[1,12]. These numerous studies have not reached a clear consensus regarding the specific correlations between lesion site and corresponding speech and language deficits, even in symposia designed to reach consensus ^[15]. Some investigators have even questioned if there are any clinico-anatomic correlations that can be drawn between the site of lesion and the pattern of language deficits in cases of subcortical aphasia ^[2,14].

Methodologic differences among studies may be the source of difficulty in demonstrating consistent clinico-anatomic relationships with subcortical lesions. Previous studies had mixed stroke etiologies, including both subcortical ischemic and hemorrhagic cases ^[1,2,4,14]. Early mass effect in the hemorrhage cases can lead to nonspecific disturbances in global attention that can obscure clinico-anatomic relationships during the acute epoch. Only postacute testing might be relevant, and it is not always available. Many studies have assessed patients (with infarctions or hemorrhage) a single time at widely variable times postonset of illness ^[13,14]. Because aphasia often improves with time, patients are being studied at potentially varying points in recovery. Some earlier studies also have mixed anatomic regions, including examples of both thalamic and nonthalamic lesions ^[2,14]. Thalamic damage probably causes aphasia through different neuropsychologic mechanisms ^[8] than damage to the striatocapsular region. Finally, differing assessment strategies for aphasia are utilized in different centers. Description of language deficits in a strictly taxonomic fashion ^[11] versus a loosely taxonomic fashion ^[3,4,14] could make comparisons among studies difficult.

Controlling for these methodologic factors should improve the specificity of lesion site relationships with aphasia profiles following subcortical vascular events. An initial study by our group ^[10] found a consistent "core" profile among patients with left capsulostriatal infarctions. We now report our analysis of aphasia after left putaminal hemorrhage. We had two hypotheses: (1) the acute language profile of these patients would have weak anatomic correlations; (2) the late language profile would consist of several distinct profiles that are each associated with specific lesion sites. A greater variability in lesion location and aphasia profile, as compared to ischemic infarcts, can occur because hemorrhages can produce lesions beyond usual vascular territories.

Methods. Subjects. The subjects of this study are thirteen consecutive patients with CT-demonstrated left putaminal hemorrhages from whom both early and late studies could be obtained who had been admitted to Braintree Rehabilitation Hospital, Braintree, MA. There were nine men and four women with a mean age of 60.0 (SD, 16.2). All patients were right-handed except for patient 4 Table 1. None of the patients had a history of previous neurologic events. Since patients were evaluated in a rehabilitation hospital, it is likely that there was a bias toward patients having sensorimotor impairments. All patients had severe hemiparesis (except nos. 1 and 2) and severe hemisensory deficits (except nos. 1 to 5).

Clinical evaluation. Patients were evaluated twice. Early assessment was done 2 to 4 weeks following the acute illness. Late assessment was done 8 to 10 weeks postonset. Formal speech and language evaluation consisted of the Western Aphasia Battery (WAB) ^[16]. Case-by-case profiles were assembled using speech quality, fluency, repetition, naming, type of paraphasia, and comprehension.

Neuroimaging. All patients had CTs during the acute phase that documented intracerebral hemorrhage (ICH) as the stroke etiology. None of the patients had lesions elsewhere in the brain. All follow-up CTs were performed more than 2 months postonset. Furthermore, all follow-up CTs that were used for lesion localization showed no evidence of residual blood. Lesion localization was determined with a template method previously described Figure 1 ^[1].

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Schematic drawings of CTs of patients with (A) "core" language deficit, (B) persistent impairment in repetition, (C) persistent nonfluency, (D) persistent impairment in comprehension, and (E) persistent impairment in repetition and comprehension

Results. Early assessment. All 13 patients were aphasic--six fluent and seven nonfluent. All seven nonfluent patients had impaired comprehension and repetition and fit the WAB criteria of global aphasia. Only two fluent patients had impaired comprehension, one with normal repetition (by WAB criteria--transcortical sensory aphasia) and one with impaired repetition (Wernicke's aphasia). The other four patients had fluent output, ie, sentence length and grammatic, but it was terse and often delayed. These four patients had only impaired naming on the WAB (anomic aphasia). Using the area of increased density (hemorrhage) on the acute scan as the lesion site, no relationship of lesion site could be distinguished between fluent (N = 6) or nonfluent (N = 7) output or between intact (N = 5) or impaired (N = 8) comprehension.

Late assessment. Speech/language testing during the late assessment is summarized in the table. Five of the six patients who were initially fluent (nos. 1 to 5) evolved to having a very mild aphasia with minimal or no anomia but poor verbal fluency, and some had mild initiation delay. All of these patients had late lesions within the capsulostriatal and paraventricular region Figure 1, A). The only left-handed patient was in this group and had a language profile that was entirely typical. The last fluent patient (no. 6) showed improvement in comprehension but had a persistent impairment in repetition with abundant phonemic paraphasias (conduction aphasia by WAB criteria). This patient had a lesion that extended laterally into the external capsule and insula and just into the temporal white matter Figure 1, B).

Three of the initially nonfluent patients (nos. 7 to 9) improved to having fluent output. All three had persistent impairments in repetition and comprehension (Wernicke's aphasia). The other four initially nonfluent patients (nos. 10 to 13) remained nonfluent. Three (nos. 10 to 12) had persistently impaired repetition but developed good comprehension (Broca's aphasia). The fourth (no. 13) had persistent impairments in repetition and comprehension (global aphasia). All patients with prolonged impaired comprehension had lesions that extended posteriorly from the capsulostriatal area into the temporal white matter area between the posterior limit of the sylvian fissure and the temporal horn of the lateral ventricle Figure 1, C). All patients with persistently impaired fluency had postacute low-density CT lesions that included the capsulostriatal region but extended anteriorly to involve much of the deep frontal white matter Figure 1, D). The patient with both persistently impaired fluency and comprehension had a lesion that extended both anteriorly and posteriorly Figure 1, E). The pattern of recovery based on WAB aphasia taxonomic categories is presented in the table. Sixty-two percent of these patients (8 of 13) changed aphasia profiles from the early to the late epoch.

Discussion. Many prior reports on "subcortical aphasia" have classified aphasia profiles by classic syndromes and have largely failed to find a unifying taxonomy. Small lesions often produce no deficits on standard aphasia tests, and large lesions, sharing only some element of capsulostriatal damage, varied quite dramatically in language deficits. Our companion study ^[10] of nonlacunar infarctions in lenticulostriatal distributions suggested that the core deficit in language largely spares word comprehension and responsive language--repetition, oral reading, and even in the mildest case, confrontational naming. In addition, syntax and grammar for short conversational responses are spared. The deficits in this core syndrome are in "generative" language ^[17,18]--verbal fluency, sentence generation, and discourse. Damage to frontal-striatal systems ^[1] responsible for complex proceduralized output underlies this deficit profile, whether damage occurs at the dorsolateral frontal cortex, ^[19] dorsolateral striatum, ^[20] or their connections in the deep frontal white matter ^[21].

Analysis of ICH adds a layer of clinical complexity because of possible acute nonspecific effects of hemorrhage mass and uncertainty about the precise site of actual damage. This problem may account for disparate conclusions reached in prior studies of aphasia after ICH ^[22] or in studies that mixed ICH with infarction in the acute stage ^[2,4,12,14]. In the present study we could identify no distinction in lesions of the basal ganglia, internal capsule, and deep frontal or deep temporal regions, between the acute ICH sites of patients with fluent versus nonfluent output or with preserved versus impaired word comprehension.

Restudied weeks later when blood had cleared on CT and edema had abated, the lesion-language relationships were clearer. Several patients with generally (but not uniformly) smaller lesions had become essentially nonaphasic by standard measures. All had residual lesions restricted to the putamen and middle PVWM. This is the same lesion-aphasia relationship described in the cases of ischemic infarction. Had we systematically probed residual generative language capacity--verbal fluency, ^[23] sentence generation, ^[17,18] or discourse ^[24]--we might have found mild reductions.

For the patients with persistent aphasia at postacute assessment, the lesion correlations with nonfluency, impaired word comprehension, and repetition were entirely compatible with theories of the distributed anatomy of language functions ^[25]. Nonfluency was most likely to occur with damage to a large portion of the deep frontal and PVWM. This finding is generally compatible with the theory of Naeser et al ^[26] regarding specific "pathways for fluency". There are numerous pathways through this region that may play a role in maintaining fluency but none is the nonfluency pathway. These could include the subcallosal fasciculus, ^[26] the projections from the supplementary motor area to dorsolateral frontal convexity that pass through the superior PVWM, ^[27] the lateral thalamic projections to Broca's area, the premotor callosal pathways, and even the projections from posterior association cortex passing caudorostrally in the PVWM ^[28]. Infarction can occur in this pattern involving the junctional area between the lenticulostriate and deep cortical perforator territories ^[29].

Impaired word comprehension was associated with damage into the deep temporal white matter, presumably damaging thalamotemporal projections ^[1] as well as short temporal association pathways within the temporal lobe. The mechanism of impaired comprehension is presumably similar to that seen after deep temporal ICH ^[30]. Although it may occur (case 13 in Alexander et al ^[1]), this is not a common pattern of damage often seen with infarction because it would require merging of the territories of the lenticulostriate and inferior cortical division of the middle cerebral artery, and even possibly, the anterior choroidal circulation ^[31].

Impaired repetition, despite preserved fluent output and comprehension, was present in one of our patients. Based on only a single patient, our explanation is tentative, but lesion extension into the subcortical parietal white matter may be the lesion correlate. The relevant pathways could include the arcuate fasciculus, extreme capsule, and short temporoparietal association connections. This specific subinsular/parietal lesion-aphasia correlation is seen with infarction, ^[32] but in patients with primarily subcortical infarctions, it is unusual. We are not certain that an example of lenticulostriate infarction with lateral extension has been reported.

While the literature reports are confounded by these issues of mixed vascular etiologies and variable times postonset, there are several cases that are suitable for fitting to our proposal about core lesions plus extensions after ICH. In 16 reports from four studies, ^[1,2,4,14] there is good agreement in 15 cases. Four (cases 1 and 6 in Puel et al ^[14] and 17 and 18 in Alexander et al ^[1]) had a speech/language profile consistent with the core subcortical aphasia profile ^[10] and appropriately had only putaminal lesions. One (case 2-3 in Cappa et al ^[4]) had a transcortical motor aphasia with a lesion in the caudate and anterior limb of the internal capsule, but this probably represents limited generative language deficits ^[10]. One (case 9 in Puel et al ^[14]) had a global aphasia and a lesion that extended anteriorly to involve the frontal isthmus and posteriorly to involve the temporal isthmus. Four (cases 3 and 4 in Puel et al ^[14] and 15 and 16 in Alexander et al ^[1]) had a profile of Wernicke's aphasia and had evidence of a lesion that extended posteriorly to involve the temporal isthmus. Five (cases 1-1 and 2-2 in Cappa et al, ^[4] 8 and 10 in Puel et al, ^[14] and 19 in Alexander et al ^[1]) had a profile of conduction aphasia with evidence of lateral extension to involve the extreme capsule and insula. Finally, there was one case that did not fit our model, but only acute data was published. This case (5 in Puel et al ^[14]) was reported to have Broca's aphasia, but there was no anterior extension of the lesion.

Weiller et al ^[33] recently proposed that aphasia in cases with subcortical infarction is actually due to the incidental microscopic cortical ischemic damage caused by the carotid or middle cerebral obstructive lesions that produce the large subcortical infarction. There are numerous problems with this proposal. The aphasia test used by these investigators does not easily identify transcortical aphasias or the mild generative aphasia of subcortical lesions. That a purely cortical lesion--even a macroscopic one--can produce standard Broca's or Wernicke's aphasia has never been demonstrated. We are also unaware of any reports of standard Broca's or Wernicke's aphasia after diffuse hypoxic-ischemic cortical damage. Furthermore, if aphasia profiles after nonthalamic subcortical hemorrhages are similar to the profiles produced by infarctions in the same territories, the cortical explanation of Weiller et al ^[33] seems unnecessary.

A corollary result of this study is information about the natural history of recovery of aphasia after lenticulostriate ICH. Patients fluent during the early epoch are likely to have a better outcome. The worst outcome in our initially fluent patients was a persistent impairment in repetition. On the other hand, none of the patients that were nonfluent during the early epoch had a very good outcome. These patients all had persistent impairments in fluency, comprehension, or both.