Results

Ten patients were identified; all were right handed. Table 1displays demographic data and the stroke aetiology subtype classification.11 Eight of 10 patients met criteria for probable or possible cardioembolic stroke and only one patient had no clear aetiology. One patient had an infarction after a subarachnoid hemorrhage in the left middle cerebral artery (MCA). The initial language assessment was completed at a mean time of 17 days after stroke. All patients manifested severe global language impairment, as assessed on the WAB. Table 2 presents the WAB and BNT scores at the time of the initial language assessment.

A cluster analysis using the acute language data (table 2) showed three distinct subgroups, which were differentiated based on severity of aphasia as represented by the AQ (χ²=7.90, p<0.02) and fluency (χ² =9.0, p<0.01). Cluster 1 included the most severely aphasic patients (mean AQ=2.3), characterised by dense non-fluent speech with severe speech initiation deficit, marked comprehension impairment, no repetition, and grossly defective naming. Cluster 2 (mean AQ=34.9) patients disclosed dense non-fluent spontaneous speech, moderate comprehension impairment, minimal capacity for naming monosyllabic high frequency nouns, and limited repetition. Cluster 3 (mean AQ=13.9) showed low level dysfluent speech, marked comprehension impairment, minimal or no repetition, and severely defective naming characterised by perseverative monosyllabic utterances. Table 3 presents the AQ and fluency scores of the 10 cases, according to the established clusters.

Cluster 1 and cluster 2 differed significantly with respect to fluency (U=0; p<0.05), repetition (U=0; p<0.05), and naming on both the WAB (U=0; p<0.05) and the BNT (U=0; p<0.05). Cluster 2 and cluster 3 were significantly different on the same variables as cluster 1 and cluster 2: fluency (U=0; p<0.05), repetition (U=8; p<0.05), WAB naming (U=8; p<0.05), and BNT (U=8; p<0.05). Cluster 1 and cluster 3 differed significantly for severity of aphasia (AQ) (U=0; p<0.05) and fluency (U=0; p<0.05).

Follow up language assessment completed 10–12 weeks after stroke disclosed that the three clusters established during the acute stage were characterised by differential patterns of aphasic evolvement. Patients in cluster 1 showed minimal or no change in all language domains and remained globally aphasic. Cluster 2 patients, initially characterised by the relative preservation of repetition, evolved into TCM type aphasic patients, with relative preservation of repetition, as well as notable recovery of comprehension and naming. Patients in cluster 3 showed variable language resolution, but were generally characterised by increased narrative fluency and improved auditory comprehension, with inconsistent changes in repetition and naming. All patients in cluster 3 evolved into a Wernicke’s type aphasia profile, based on WAB criteria. Table 4 provides the follow up raw language scores from the WAB on individual patients within three clusters.

Composite lesion analysis showed that patients in cluster 1, who presented with the most severe global aphasia initially, and showed minimal or no significant recovery of language functions during the first 3 months after stroke, were characterised by common lesioning of the left superior temporal gyrus, primarily involving territory of the posterior temporal and middle temporal branches of the MCA (fig A). Patients with acute GAWH who evolved into TCM type aphasic patients (cluster 2) had common lesioning of the left inferior frontal gyrus and adjacent subcortical white matter, primarily involving the territory of the precentral, central, lenticulostriate, and insular branches of the MCA (fig B). Patients with acute GAWH who evolved into Wernicke’s type aphasia were characterised by lesioning of the left precentral and postcentral gyri, involving the central and posterior parietal vascular territories (fig C).

Discussion

GAWH is a striking clinical phenomenon because of the unusual dissociation between language and motor functions. Because global aphasia typically results from large perisylvian lesions, adjacent motor areas are usually affected. The absence of hemiparesis suggests that the cortical representation of motor and language function may not always follow classic descriptions, or that only unusual lesions may cause this syndrome.

Although patients with acute GAWH are similar on neurological examination, they are heterogeneous with respect to early aphasia profile, language recovery, and lesion profile. Our findings disclose three distinct subtypes of GAWH, based on acute language profiles and aphasic evolvement during the first 3 months after stroke. One subtype (persistent GAWH) presents with extremely dense global aphasia with virtually no expressive or receptive language function and is characterised by negligible recovery of language. A second subtype (GAWH-TCM) presents with acute global aphasia, characterised by limited expressive and receptive language capacity, specifically involving low level repetition and naming, and evolves into TCM type aphasia. A third subtype (GAWH-Wernicke) with moderately severe global aphasia during the acute stage evolves into Wernicke’s type aphasia. Although we differentiated three subtypes of GAWH, it is important to note that classification based on a cluster analysis does not necessarily denote anatomically distinct or functionally homogeneous subtypes.

Our anatomical findings are consistent with the collective findings of previous case reports which hold that various lesion constellations, including both single and multiple lesions, can produce GAWH. Early reports of GAWH suggested that multiple lesions which involved anterior and posterior language areas but spared the motor cortex were necessary to cause this syndrome, and that it was highly indicative of embolic aetiology. Van Horn and Hawes2 reported three patients with GAWH, all of whom had two lesions involving the left posterior frontal area and the left temporoparietal region. They concluded that GAWH is a sign of embolic encephalopathy, caused by two discrete lesions. Tranel et al 3described three cases, again due to double lesions involving the left frontal operculum and the left parietotemporal region. They also concluded that GAWH is produced by two discrete lesions of the left middle cerebral artery and is an indicator of embolic encephalopathy. However, later reports questioned both the necessity for multiple lesions and the role of embolism in GAWH. Ferro4 reported two patients, each of whom had a single large infarct of the left MCA involving both Broca’s and Wernicke’s areas, but which spared the posterior limb of the internal capsule. Legatt et al 5 reported six non-hemiparetic patients with global aphasia, including one with a single infarct of the anterior branches of the left MCA, two with subarachnoid hemorrhage, and one with multiple metastases six months after presentation. Subsequently, Deleval et al 6 reported two patients with GAWH caused by single prerolandic infarcts. Bogousslavsky7 described a case of GAWH secondary to a single left temporoparietal infarct and noted that two infarcts are not necessary to produce GAWH, and that the syndrome does not necessarily imply a double lesion or an embolic aetiology.

Nagaratnam et al 9 reported 12 patients described as having GAWH. However, it is unclear how several of their patients met Boston diagnostic aphasia examination (BDAE) criteria for global aphasia because they were only mildly impaired on nearly all language functions assessed, according to scores on the aphasia severity rating scale. They reported that four cases had severe global language impairment, but “the remainder had moderate to mild abnormalities.” This series seems to describe a different population from the other reports in the literature and from our own.

In addition to the different lesion profiles of the three GAWH clusters, many of our individual patients were similar to previously reported patients. Case 1 (table 1) had a posterior lesion involving the posterior temporal region without anterior involvement which was similar to a case reported by Bogousslavsky.7 In addition to the finding that GAWH may result from a single temporal lesion, these patients support the notion that fluency may be seriously disrupted by posterior lesions, in the absence of anterior involvement. Conversely, two patients in the present sample (case 5 from cluster 3; case 8 from cluster 2) had anterior lesions with no posterior involvement. These patients were similar to the two patients described by Deleval et al 6 with respect to lesion site and language recovery. They provide further evidence of the notion that receptive language functions may be severely disrupted by anterior lesions, exclusively.

Such anatomical diversity is well documented in case series of the more common global aphasia with hemiparesis. Although global aphasia is commonly associated with large left perisylvian lesions involving both the anterior and posterior language areas, lesion analyses of groups of patients with global aphasia have consistently shown that the lesion constellations responsible for global language impairment are extremely variable. In a series of 37 patients with global language impairment, Vignolo et al 16 found that only 22 (59%) of their patients had large left perisylvian lesions involving both Broca’s and Wernicke’s areas. Eight (22%) had anterior lesions sparing Wernicke’s area, three (8%) had posterior lesions sparing Broca’s area, and four (11%) had deep lesions of the insula and lenticular nucleus. Similarly, Scarpa et al 17 reported that only 53% of a series of 46 patients with global language impairment showed the classic anterior-posterior lesion profile, whereas 32% of their patients manifested global aphasia secondary to a deep lesion. Chronic global language impairment has also been documented after a single lesion of the left thalamus.18

Our data in conjunction with the other cases reviewed (table 5) suggest that whereas cardioembolism is the single most common aetiology of GAWH, an equal number of cases result from other conditions. The clinical evaluation of some of the prior cases reported as embolic actually showed occlusion of the internal carotid or of a cortical branch on evaluation; these cases are currently classified as large artery atherosclerotic strokes in the TOAST scheme. Although some such cases may represent stroke due to artery to artery embolisation or occult cardioembolism, other cases in this category may be atherogenic or thrombotic and thus are stroke causes which many neurologists might not currently treat with anticoagulation. Additionally, almost 20% of the cases in table 5 were due to conditions such as intracerebral haemorrhage, subarachnoid haemorrhage, or metastasis in which anticoagulation is clearly contraindicated. Thus although the finding of GAWH is suggestive of an embolic stroke process, it is clearly not pathognomonic of embolism and these patients will require thorough evaluation for the cause of their cerebral infarction.

By contrast with previous reports,3 6 the patients in this study were not uniformly characterised by extensive recovery of language function. Moreover, the degree of recovery was related to the subtype of GAWH. Patients with persistent GAWH (cluster 1) showed minimal or no recovery 3 months after onset and continued to manifest severe dense global aphasia. Conversely, patients with GAWH-TCM (cluster 2) disclosed notable recovery in most language domains. Finally, GAWH-Wernicke (cluster 3) patients disclosed significant language recovery, but only in selective language domains (fluency, comprehension). Clearly, patients with acute GAWH do not invariably experience significant language recovery. Our findings also support the utility of detailed early language assessment for this population. In our study, the first neuropsychological evaluation identified findings which were associated with those patients with residual function who went on to better long term recoveries. Such residual function can also be exploited in current rehabilitation interventions and may increase functional communication with care givers.