Background and Purpose Previous studies of lesion localization in denial of hemiplegia have often been confounded by factors such as the cerebral etiology, which aspects of the disorder are denied, and whether delusions are present. Our investigation focuses on denial of hemiplegia, without concomitant delusions, after cerebrovascular accident (CVA).
Methods The CT scans of 30 patients with denial of hemiplegia after acute stroke were compared with those from 10 CVA patients with hemiplegia and visuospatial neglect but no denial. Lesion sites were detailed using the templates of Damasio and Damasio and, for the deep structures, those of Talairach and Tournoux.
Results Analysis of the CT scans demonstrated that 26 of 30 denial patients had unilateral right-sided lesions and that this group showed a significantly higher incidence of lesions in deep white matter and the basal ganglia.
Conclusions The results are discussed in relation to recent ideas regarding the role of the basal ganglia and subcortical circuits in movement and executive function.
The localization of the cerebral pathology in cases of denial of hemiplegia, first described by Pick1 in 1898 and termed “anosognosia” by Babinski,2 has resulted in various sites being proposed as the anatomic correlate underlying the disorder. Early descriptions of the probable cerebral localization, reviewed by Gerstmann,3 have clearly been conflicting. Both Potzl4 and Barkman5 conclude that lesions of the right parietal lobe, right optic thalamus, or right thalamoparietal radiation are the necessary prerequisites, yet Muller6 and Pineas7 report autopsied cases in which these areas remained intact. Roth8 describes the autopsy findings in a patient with anosognosia for left hemiparesis that showed an extensive right hemisphere spongioblastoma multiforme destroying the cortex and extensive underlying white matter; the thalamus was not studied in detail.
Weinstein and Kahn,9 using the term anosognosia to describe denial of any defect or illness, report the lesion location (confirmed either at operation or autopsy) in 22 patients with brain tumor. For the 3 patients who denied their hemiplegia, the areas involved were the thalamus, basal ganglia, and corpus callosum in the first and the right temporal lobe in the second; the third patient had a cholesteatoma of the lateral ventricles. Hier et al,10 in a study of 41 patients with unilateral right hemisphere stroke, classified anosognosia as a failure to acknowledge either the stroke or subsequent deficits. They found that denial of illness showed a significant correlation with the extent of injury (measured by CT) to frontal, parietal, and temporal lobes, as well as deep structures. Anosognosia seemed to depend on disruption of both parietal structures and connections outside the immediate parietal region.
Bisiach et al,11 investigating denial of motor impairment and denial of visual field defects in patients with right hemisphere vascular or neoplastic pathology, found no detectable difference in the lesion sites (assessed by CT scan) between the two conditions, even though the disorders were dissociable. Anosognosia was associated with both superficial and deep lesions and with damage confined to deep structures. For lesions affecting the cortex, the inferoposterior parietal region seemed of particular importance, implying that damage to neural structures outside the respective primary cortical area is a necessary factor. With damage to deep structures, the thalamus and/or lenticular nucleus seemed the most likely anatomic correlate. The findings of Bisiach et al that cortical or deep structure lesions may be associated with anosognosia confirm the results of two earlier studies. Heilman and Valenstein12 described 10 patients with denial of hemiparesis or denial of illness whose mixed pathologies (vascular or tumor) were shown by CT scan to involve the right inferior parietal lobe in 9 cases and the left frontal lobe in 1. Left frontal cortical and subcortical infarct associated with denial of illness has also been described by Dronkers and Knight,13 and Watson and Heilman14 report unilateral right thalamic hemorrhage in 3 patients with denial of weakness or lack of concern for left hemiparesis.
More recently, Starkstein et al15 have studied the CT evidence of lesion size and location, as well as the amount of cortical and subcortical atrophy, in a group of patients with anosognosia for either motor or visual field deficits after acute stroke. Patients with denial showed a significantly higher frequency of right hemisphere lesions, mainly involving the superior temporal and inferior parietal cortex, basal ganglia, and thalamus, as well as significantly more bilateral cortical atrophy in the area of the frontal horns, lateral ventricle, and the third ventricles. The authors conclude that the findings support Nielsen's16 suggestion that anosognosia may be caused by thalamic lesions or isolation of the thalamus from the parietal cortex and that disruption of frontal subcortical or diencephalic structures may be an important predisposing factor.
In a detailed study comparing two groups of right hemisphere stroke patients, with or without persistent denial of hemiplegia, Levine et al17 describe the 6 subjects with persistent anosognosia as all having large lesions with damage to either the central gyri or the underlying corona radiata. In nearly all cases, the extensive damage to the corona radiata interrupted projections to and from wide areas of the frontal and parietal convexities anterior and posterior to the central gyri. The lesions of the second group of patients, who either never had denial or whose anosognosia for hemiplegia lasted less than a few days, varied considerably in size, with 2 such patients having very extensive lesions that were as large or larger than any of those of the 6 persistent anosognosia cases. Although a large lesion may therefore be necessary for persistent denial of hemiplegia, it is not sufficient to cause it. The authors conclude that although all the patients with persistent denial had large right hemisphere strokes, no single discrete region other than the sensorimotor pathways had to be damaged to cause it.
Levine et al17 also mention that delusions regarding the paralyzed arm were common, with at least 2 patients in the persistent-denial group believing the limb not to be their own. Previous studies, such as those of Ives and Nielsen18 and Nielsen,16 have argued that there should be a distinction between these two types of awareness, ie, denial of hemiplegia (whereby patients believe a paralyzed limb is functioning normally) and delusions regarding a body part (when ownership of the limb is denied). The study by Ives and Nielsen18 of 2 autopsy cases indicates that the delusional disturbance can be caused by focal lesions, and they claim that the disorder is quite different from denial of hemiplegia. Nielsen,16 comparing the anatomic substrates of the two phenomena, concludes that delusions of absence of a limb is produced by a lesion of the thalamoparietal peduncle, whereas an intrathalamic lesion or one isolating the thalamus from the frontal, parietal, and temporal cortexes results in unawareness of hemiplegia.
Clearly, previous studies have been confounded by factors such as different pathologies (vascular and neoplastic), the type of anosognosia (unawareness or denial of hemiplegia and/or denial of visual defects, denial of stroke deficits, or denial of any illness), and the addition of a delusional disorder concomitant with the denial of hemiplegia. We therefore undertook our study to investigate the locus of pathology in stroke patients with denial of hemiplegia but without any delusional disorder. Thirty such patients were examined in the immediate poststroke period, and no subject showed a delusional disturbance even though this was particularly sought. We report the neuroradiological findings in these patients compared with those in similar cerebrovascular accident (CVA) patients with visuospatial neglect and hemiplegia but no denial.
Subjects and Methods
The admission details of all patients admitted to the John Radcliffe Hospital or to the Department of Neurology, Radcliffe Infirmary (Oxford, UK), during the 47-month period of study were screened by a neurologist (S.E.). Those patients diagnosed as having had a definite acute CVA and who underwent CT were seen. Exclusion criteria consisted of previous CVA, other ongoing neurological disorder, psychiatric illness, dementia or confusional state, and timing of the CT scan at <24 hours or >3 months after the insult.19 20 Initial interviews were dependent on a sufficient level of patient communication; those with severe expressive or receptive dysphasia or reduced consciousness level were also excluded. Of the patients who could be interviewed, 34 demonstrated some degree of denial of hemiplegia. Four of these patients were excluded because 3 had CT scans carried out at <24 hours and 1 at >3 months after their CVA. (This included 1 case of denial of right-sided hemiplegia who was excluded because her CT scan had been undertaken <24 hours after the insult.)
Thirty patients with denial of hemiplegia therefore made up the study group, comprising 18 men and 12 women with a mean±SD age of 74.1±8.9 years (range, 48 to 85 years). Twenty-eight exhibited concomitant visuospatial neglect and 2 did not. The comparison group, for whom the same inclusion criteria applied, consisted of 10 CVA patients with hemiplegia and left-sided visuospatial neglect but no denial of their paralysis (4 women and 6 men; mean±SD age, 73±9.3 years [range, 56 to 84 years]). For clarity, this latter group of patients will be termed throughout the text as “neglect” patients, and those with denial of hemiplegia (who may or may not also have had neglect) will be termed the “denial” patients. For the denial group, the mean±SD time between the CVA and CT scan was 6.03±6.1 days; for the neglect patients, it was 2.6±1.8 days. The timing of CT scan was determined by each patient's admitting general physician on the basis of clinical grounds alone. The investigation was approved by the Central Oxford Research Ethics Committee, and all patients gave their informed consent.
CT was performed with an IGE Sytec 3000 scanner with 10-mm contiguous sections. Lesions were traced onto templates as described by Damasio and Damasio21 in their Fig A14. The templates of axial sections were analyzed according to the presence or absence of infarcted tissue within the 41 brain areas listed in Figs 1⇓ and 2. Subsequent analysis of lesions of the deep structures was performed according to the method of Talairach and Tournoux.22
Inclusion Criteria and Classification of Denial
The degree of denial in hemiplegia can range from acknowledgment of the paresis but not its implications to continuous obstinate denial, whereby the patient totally refuses to accept any problem even after it has been repeatedly pointed out by the examiner. A classification of the level of denial observed in each patient was therefore necessary. Interviews regarding the degree of denial were audiotaped and subsequently scored independently by both experimenters; the correlation between the ratings on 155 taped interviews was 0.83 (Spearman's ρ). We found that the following subdivisions allowed accurate identification and definition: 0, no denial; 1, accepts diagnosis and disabilities but demonstrates a failure to integrate knowledge of illness; 2, accepts diagnosis but denies aspects of disability; 3, does not fully accept diagnosis or disability but accepts that something is wrong; and 4, total denial (denies that anything related to health is wrong).
In the subsequent text, the level of denial is provided for the time of the initial neurological examination. The mean±SD time between stroke and assessment of denial was 4.7±4.8 days (range, <1 to 17 days). Whenever possible, patients had their grade of denial monitored on a daily basis, but this was not always feasible because, as would be expected in a study focusing on the clinical course that immediately follows CVA, some patients became either comatose, too ill, or died. The summary below indicates the number of patients with changes in the denial rating and patient outcome over time (excluding 6 who were lost to follow-up). Denial was improved in a total of 13 patients: in 4 at <1 week, in 6 at 1 week to 1 month, in 1 at >1 month, and in 2 at 2 months. Denial was the same in a total of 5 patients: in 2 at 1 to 2 weeks, in 2 at 2 to 3 weeks, and in 1 at >1 month. A total of 6 patients died or were too ill to assess: 1 at 3 days, and 5 at 1 to 3 weeks.
Tables 1⇓ and 2⇓ give the patient details for the hemiplegia-denial and visuospatial-neglect groups, respectively. Because the size of lesion was large in many of our patients, it was not possible to produce a meaningful composite CT diagram to illustrate possible common areas of cortical damage.
Type of Lesion
In patients with denial of hemiplegia, there were 3 cases of hemorrhage and 27 cases of infarction; in the neglect group, there were 3 cases of hemorrhage and 7 of infarction.
Side of Lesion
In the patients with denial, 26 had unilateral right-sided lesions and 4 showed bilateral involvement. Of those with left-sided damage, one patient had grade 1 denial, 2 had grade 2, and 1 had grade 4. In the visuospatial-neglect group, there were 9 patients with right-sided lesions and 1 with bilateral involvement.
Cerebral Areas Involved
There are a total of 14 frontal (F01 to F14), 11 temporal (T01, T03 to T12), 6 parietal (P01 to P06), and 7 occipital areas (O01 to O07) depicted on the Damasio and Damasio21 templates. Subcortical areas include the deep white matter, thalamus, and basal ganglia, giving a total of 41 separate brain sites.
Using the detailed scan evaluation from the templates of axial sections, each patient's scan was studied, and the lesioned sites were plotted according to the presence or absence of infarcted tissue within these 41 brain areas. The numbers of patients in each group whose CT showed a lesion in any one of the 41 sites are represented in the histograms (Figs 1 and 2⇑⇓) for the right and left hemispheres, respectively. There were no significant differences between the two groups in the number of lesion sites in any of the cortical areas analyzed. Both groups showed a high percentage of damage in areas F08 (denial 57%, neglect 70%), F09 (denial 70%, neglect 50%), and P05 (denial 77%, neglect 70%). When the temporal, parietal, and occipital areas were combined, patients with visuospatial neglect tended to have more extensive lesions overall than those with denial; the mean number of areas involved was 8.9 and 5.8 for the two groups, respectively (significant at P=.031, χ2=4.67).
With regard to the basal ganglia, there was a significant difference in the number of patients with denial (70%) who had lesions in this area compared with those with visuospatial neglect (30%; P=.025, χ2=5). Similarly, the number of patients in the denial group (83%) with deep white matter involvement proved significantly different from that of such patients in the neglect group (60%; P=.031, χ2=4.67).
Because the Damasio and Damasio templates21 do not analyze the deep structures in detail, the stereotactic map of Talairach and Tournoux22 was superimposed on this area for each patient's scan. Lesions were identified as involving the right and/or left basal ganglia (subdivided into the caudate and lenticular nuclei) or the thalamus, insula, and deep white matter (subdivided into internal capsule, external capsule, and corona radiata). Additional statistical analyses were carried out between the two groups to compare right-sided and bilateral deep-structure lesions. The only comparison that proved significant was when both right and left sides of the basal ganglia were considered together (P=.035, χ2=4.4). Twenty-five patients (83.3%) with denial of hemiplegia showed a lesion in this area compared with 5 (50%) with visuospatial neglect. Separate analyses of the caudate nuclei showed involvement in 20 denial patients and 4 neglect patients; analysis of the lenticular nuclei indicated lesions in this area in 22 denial and 4 neglect patients. Neither analysis proved significant.
To ascertain whether any areas of infarction were correlated with the severity of denial, the 30 patients with denial of hemiplegia were divided into two groups (grade 1+grade 2 and grade 3+grade 4) depending on the grade of denial at the time of the initial neurological assessment; the total numbers of patients in each group were 11 and 19, respectively. There was no obvious difference in the number of lesions in cortical areas either between the two graded groups of denial or when either group was compared separately with the neglect patients. However, using the Talairach and Tournoux22 templates for the deep structures, the patients in the grade 1+2 group showed a higher incidence of lesions in the basal ganglia (10 of 11 cases, 91%) compared with either the denial grade 3+4 group (15 of 19 cases, 79%) or the neglect patients (5 of 10, 50%).
Analysis of the CT scan data revealed that the lesions were right sided in both patient groups, with the exception of 4 patients with denial and 1 with neglect who showed bilateral involvement. The predominance of right hemisphere CVA in our investigation is at least partly attributable to the inherent difficulties in assessing denial because of the frequent presence of severe receptive and/or expressive language disorders in patients with left hemisphere lesions. Twenty-six of the 30 denial patients had unilateral right-sided lesions without involvement of the left hemisphere. (The location of the lesion in our 1 subject with denial of right-sided hemiplegia, who was excluded from the study because of the timing of her CT, was confined to deep structures [ie, the left basal ganglia, thalamus, and internal capsule], an area quite different from that described by Dronkers and Knight13 in their patient with denial of right-sided weakness, whose lesion on CT was found to be an infarct of the dorsolateral left prefrontal cortex extending both superiorly and posteriorly).
Both groups had lesions in the frontal areas, particularly F08 and F09, which correspond to the premotor, Rolandic, and paraventricular regions. Such a finding is not surprising because hemiplegia was a prerequisite for inclusion into the study. Although there was no difference between the groups when separate cortical areas (frontal, temporal, parietal, and occipital) were compared, when these sites were combined, patients with neglect showed more extensive lesions overall than those with denial. A larger area of cerebral damage compared with other CVA patients is therefore not essential for denial of hemiplegia.
On the Damasio and Damasio21 template analysis, deep white matter involvement was a significant finding in our patients with denial of hemiplegia (83%) compared with the neglect cases (60%). Bisiach et al11 similarly report damage confined to deep structures in their investigation of patients with denial of motor impairments and denial of field defects, and they conclude that the thalamus and/or the lenticular nucleus seem the most likely anatomic correlate. The thalamus was also implicated in the CT study of patients with anosognosia for motor or visual deficits by Starkstein et al.15 However, on our more detailed analysis of deep structures using the Talairach and Tournoux22 templates, neither comparison of the thalamus nor of the lenticular nucleus indicated a significant difference between the two groups. In fact, hemiplegia-denial patients had less thalamic damage (40%) than the neglect cases (60%). Levine et al17 describe damage to the underlying corona radiata in their 6 patients, but again, using the Talairach and Tournoux22 analysis, we found no difference in damage to the corona radiata between the two groups.
In both of our neuroanatomic analyses,21 22 the basal ganglia proved to be significantly more involved in patients with denial of hemiplegia than in those with neglect but no denial. This region has previously been reported as involved in denial of hemiplegia by Damasio and Damasio21 (1 patient) and Weinstein and Kahn9 (1 patient), as well as in denial of either motor or visual deficits by Starkstein et al.15 Of interest is the finding that patients with less severe denial of hemiplegia showed a higher incidence of basal ganglia involvement than those graded as more severe. This result is similar to that reported by Starkstein et al15 of a higher frequency of lesions involving the basal ganglia in patients classified as having moderate anosognosia as opposed to mild or severe. Using the Talairach and Tournoux22 templates, 80% of our hemiplegia-denial patients had involvement of the right basal ganglia, with this figure increasing to 83% when both right and left basal ganglia were considered.
The functions of the basal ganglia are still poorly understood,23 but they are thought to play a key role in the preparation and execution of normal movement generated in the cerebral cortex. Patients with basal ganglia dysfunction not only have disturbances of movement but also show alterations in intellectual function, personality, and mood.24 In a recent study25 investigating neuropsychological deficits in denial, a group of 20 patients with denial of hemiplegia after acute CVA were compared with 10 subjects with visuospatial neglect without denial. There was no evidence from neuropsychological testing that the denial patients showed additional changes in intellectual ability, frontal lobe function, or personality or a combined loss of sensation with reduced cognition. Cummings24 describes five frontal-subcortical circuits, with the putamen being primarily involved in the motor circuit and the caudate playing a critical role in the prefrontal circuits that mediate emotional and executive function; a deficit in the latter reflects involvement of the dorsolateral prefrontal circuit as it projects through the basal ganglia. Dysfunction of a circuit structure may produce symptoms by altering its effect on distant structures within the circuit. Marsden and Obeso26 also suggest that subcircuits possibly exist within the basal ganglia that are concerned with the regulation of thought, planning, and emotion; if this is the case, the loss of a contribution from this region could result in inflexibility of mental and motor response. It is an inflexibility of the response to the lack of movement in a paralyzed limb that characterizes denial of hemiplegia.
It should be noted that the evaluation of CTs in our patients in whom the mean time after ictus to scan was <1 week may have underestimated the area of infarction, particularly if compared with MRI; with exclusion of CT scans performed at <24 hours after the event, major areas of infarction would have been visualized.
In conclusion, our study involved a large homogenous group of patients, all with the same etiology (CVA) and all examined by CT scan within days after stroke. The denial patients had denial specifically of their hemiplegia, and all were without delusional disturbance of body parts. When the essential comparison is made with other CVA patients with hemiplegia and visuospatial neglect but no denial, detailed CT scan analyses implicate significant involvement of the deep white matter and basal ganglia, with lesions being almost exclusively right-sided.
This study was supported by grant G-9024591 from the MRC Centre for Brain and Behaviour, Oxford, UK (Dr Small).
- Received April 16, 1996.
- Revision received July 2, 1996.
- Accepted August 25, 1996.
- Copyright © 1997 by American Heart Association
Pick A. Beitrage zur Pathologic und Pathologische Anatomic des Central Nervensystems mit Bermerkungen zur Normalen Anatomic Desselben. Berlin, Germany: Karger; 1898:168-185.
Babinski J. Contribution a l'etude des troubles mentaux dans l'hemiplegie organique cerebrale (anosognosie). Rev Neurol. 1914;27:845-847.
Potzl O. Ueber storungen der selbstwahrnehmung bei linksseitiger hemiplegie. Neurol Psychiat.. 1925;93:117-168.
Barkman A. De l'anosognosie dans l'hemiplegie cerebrale: contribution clinique a l'etude de ce symptome. Acta Med Scand.. 1925;62:213-254.
Muller F. Ueber storungen der sensibilitat bei erkrankungen des gehirns. In: von Volkmann R, ed. Sammiung Klinische Vortrage. Leipzig, Germany: Breitkopf & Hartel; 1905; ser 14:394-395.
Pineas H. Der manger an krankheitsbewusstsein und seine variationen als symptom organischer erkrankungen. Nervenarzt.. 1926;16:238-248.
Roth M. Disorders of the body image caused by lesions of the right parietal lobe. Brain.. 1949;72:89-111.
Hier D, Mondlock J, Caplan R. Behavioral abnormalities after right hemisphere stroke. Neurology.. 1983;33:337-344.
Watson R, Heilman K. Thalamic neglect. Neurology.. 1979;29:690-694.
Starkstein S, Federoff P, Price T, Leiguarda R, Robinson R. Anosognosia in patients with cerebrovascular lesions: a study of causative factors. Stroke.. 1992;10:1446-1453.
Nielsen J. Disturbances of the body scheme: their physiologic mechanism. Bull L A Neurol Soc.. 1938;3:127-135.
Levine DN, Calvanio R, Rinn WE. The pathogenesis of anosognosia for hemiplegia. Neurology.. 1991;41:1770-1781.
Ives E, Nielsen J. Disturbance of body scheme: delusion of absence of part of body in two cases with autopsy verification of lesion. Bull L A Neurol Sci.. 1937;2:120-125.
Damasio H, Damasio A. Lesion Analysis in Neuropsychology. New York, NY: Oxford University Press; 1989.
Talairach J, Tournoux P. Co-Planar Stereotaxic Atlas of the Human Brain. New York, NY: Thieme Medical Publishers Inc; 1988.
Oppenheimer DR, Esiri MM. Diseases of the basal ganglia, cerebellum and motor neurones. In: Hume-Adams J, Duchen LW, eds. Greenfield's Neuropathology. Baltimore, Md: Edward Arnold; 1992; chap 15:988-1045.
Marsden CD, Obeso JA. The functions of the basal ganglia and the paradox of stereotaxic surgery in Parkinson's disease. Brain.. 1994;117:87-89.