فیزیولوژی هذیان حرکتی در ناتوانی در ادراک بیماری برای همی پلژی: پیامدها برای مدل های فعلی آگاهی حرکتی

Abstract Right brain damaged patients sometimes deny that their left arm is paralysed or even claim to have just moved it. This condition is known as anosognosia for hemiplegia (AHP). Here, we used fMRI to study patients with and without AHP during the execution of a motor task. We found that the delusional belief of having moved was preceded by brain activation of the cortical regions that are implicated in motor control in the left intact hemisphere and in the spared motor regions of the right hemisphere; patients without anosognosia did not present with the same degree of activation. We conclude that the false belief of movement is associated with a combination of strategically placed brain lesions and the preceding residual neural activity of the fronto-parietal motor network. These findings provide evidence that the activity of motor cortices contributes to our beliefs about the state of our motor system.

1. Introduction We are normally aware that the body that we inhabit is our own; we are aware of the state of our motor system and the sense of being (rather than not being) the cause of an act (Jeannerod, 2006); these are all crucial aspects of the sense of being ‘us’. Neurological and psychiatric disorders can provoke pathological experiences or beliefs1 concerning these apparently obvious feelings (see review in Prigatano, 2010). The study of these conditions has provided important information for theories of motor control and consciousness (Bisiach, 1988, Bottini et al., 2010, Frith et al., 2000 and Jeannerod, 1997). One relevant case is that of right brain-damaged patients with left hemiplegia who deny their motor deficit, even when repeatedly questioned by the examiner; this denial is a condition called anosognosia for hemiplegia (AHP) and was first described by Babinski (1914). Several interpretations of AHP have been offered (see review in Bisiach, 1995 and Vallar and Ronchi, 2006). These range from the psychodynamic hypotheses of the denial of illness (Weinstein & Kahn, 1955) to a more anatomically informed hypothesis, which postulates a disconnection between the right-sided motor regions and the left-sided language brain regions that are in charge of verbally reporting the patient’s feelings (Gazzaniga, 1989 and Geschwind, 1965). More recently, when summarising a large body of evidence (Adair et al., 1997 and Gold et al., 1994), Heilman, 1991 and Heilman et al., 1998 framed anosognosia in a “feed-forward” theory of motor control, that explains the deficit as a specific failure to formulate an intention to move. According to this model, motor plans/intentions are constantly compared with the somatosensory consequences of actions. In normal conditions, the motor intentional system activates simultaneously the motor system and a representation of how the body position will change after the execution of the movement (body representation or comparator; for a detailed description of the model see Fig. 1 in Heilman et al., 1998, p. 1907). When patients with hemiplegia, who are aware of their motor impairment, intend to perform a movement, the monitor-comparator system detects the discrepancy between the expected movement and the failed performance (see Fig. 1 in Heilman et al., 1998) and thus the patients recognise their paralysis. The authors proposed that anosognosic patients may have lost their motor intention: if patients have no intention and thus no expectancy to move the plegic limb, the intact comparator has no material with which to notice any mismatch. As a consequence, patients do not recognise their motor disability (Heilman, 1991 and Heilman et al., 1998). However, this theory fails to explain the more productive aspect of AHP (Bottini et al., 2009), namely the delusional belief of having moved an otherwise paralysed limb. Frith et al. (2000) proposed a different interpretation of AHP, in the context of a more articulated model of motor control (Wolpert and Ghahramani, 2000 and Wolpert et al., 1995), that may explain the productive aspects of the syndrome. According to the model, under normal circumstances, motor commands are generated when there is a discrepancy between the actual and the desired state (i.e. motor representation of the desired goal) of the motor system: a set of motor controllers select the appropriate motor commands, and the motor predictors estimate the sensory consequences of the intended motor act (forward model). This prediction (based on the efferent copy of the programmed movement) is compared to the sensory feedback related to the actual execution of the movement by a motor-comparator system. Lesion plot of the AHP patients compared with non-AHP patients. (A) Overlays of ... Fig. 1. Lesion plot of the AHP patients compared with non-AHP patients. (A) Overlays of the lesion plot of the five patients with AHP. The number of overlapping lesions is illustrated using different colours ranging from red (lesion in two patients) to white (lesion in five patients). (B) Overlays of the lesion plot of the six patients without AHP. The number of overlapping lesions is illustrated using different colours ranging for red (lesion in two patients) to white (lesion in six patients). (C) Subtraction analysis: the patients showing AHP (n = 5) minus the patients without AHP (n = 6). Regions frequently damaged in patients with AHP but spared in non-AHP patients are illustrated with warm colours, from dark red to white. Only the regions that were damaged at least the 40% more in the AHP group than in the non-AHP group are reported. The cold colours, from dark to light blue illustrate regions more frequently damaged in non-AHP patients than in AHP patients. Only the regions that were damaged at least the 40% more in the non-AHP group than in the AHP group patients are reported. The MNI z coordinates of each transverse section are given. Figure options In hemiplegic patients without AHP, the comparator detects a mismatch between the intended movement and the failure of the performance due to the paralysis. In patients with AHP, the representation of the desired and the predicted position of the limb would be preserved with a seemingly normal experience of having initiated a movement. According to Frith et al. (2000), this experience is grounded on the representation of the predicted consequences of the movement rather than upon the actual sensory feedback (Fourneret & Jeannerod, 1998). Frith et al. (2000, pp.1781) also propose that the false belief of having moved occurs because the information derived from sensory feedback, concerning actual limb position, is “not available […] or neglected” (Frith et al., 2000, p. 1781). This leads to a failure to detect the discrepancy between the estimated and the actual consequences of the programmed movement. In addition, AHP patients would fail to update the operations of the predictors, hence they would fail to learn that the action did not occur (Frith et al., 2000). In a similar vein, Berti, Spinazzola, Pia, and Rabuffetti (2007) proposed that the brain lesion of AHP should directly affect a neural comparator to cause “the failure of a motor monitoring component that does not detect the mismatch between a desired action and the actual status of the sensorimotor system in face of an intact capacity of programming movements and forming sensorimotor predictions” ( Berti et al., 2007, p. 163). Clearly, the aforementioned theories make contrasting neurophysiological predictions on the functional status of the brains of AHP patients. Heilman’s hypothesis (1991) predicts a global damage of the “intentional-preparatory” premotor system; conversely, the other two interpretations (Berti et al., 2007 and Frith et al., 2000) are consistent with the idea of a distributed anatomical system for motor control with a possible sparing, in AHP patients, of some specific components of the motor system that are involved in the stage of motor planning and simulation of the state of the motor system (Berti et al., 2007 and Frith et al., 2000). The third theory may imply the additional assumption that “the brain activity leading to the construction of a conscious intention of action is normal” (Berti et al., 2007, p. 163) and that the deficit is restricted to a putative comparator system. There is some behavioural evidence to support the idea of preserved motor intentionality in AHP patients. Berti et al. (2007) found that a right-brain-damaged patient with AHP had electromyoghraphic activity of the left upper trapezius2 following the instruction to produce a reaching movement with the left upper limb. Additional evidence of residual intentionality in AHP patients, which predominates over sensory feedback and its connection with the delusions of having moved, comes from the work of Fotopoulou et al. (2008); they showed that the delusional component of AHP for a prosthetic limb is present when patients are asked to move that limb rather than when they are merely asked to observe the prosthetic arm being moved by an examiner. Recently, Garbarini et al. (2012) have shown that the request to move the paralysed hand has consequences for the actual movement of the unaffected hand. They asked their patients to execute a ‘bimanual’ motor task involving drawing lines with the right hand and, simultaneously, drawing circles with their left paralysed hand. Although no overt motion was present in the paralysed left hand, a bimanual coupling effect, comparable to that observed in healthy subjects, was found in the AHP patients: the trajectories of the intact hand were influenced by the requested movement of the paralysed hand, with the intact hand has a tendency to produce an oval trajectory. Interestingly, the same effect was not found in hemiplegic patients without AHP or in motor-neglect patients. Similar observations have been made by Pia et al. (2012) who showed that there was a preserved bimanual temporal coupling effect in AHP patients using the reaching task that was originally described by Kelso, Southard, and Goodman (1979). Anatomical investigations have the potential to provide information that can be used to test current neuropsychological models including those of anosognosia for hemiplegia. Early studies have associated AHP with the parietal lobe lesions that are typical of spatial neglect (Critchley, 1953). However, these were usually single-case studies that lacked control groups of hemiplegic patients with neglect and without anosognosia (see review in Pia, Neppi-Modona, Ricci, & Berti, 2004). More recent studies, in which the more rigorous approach of having specific control groups was used, suggest that AHP is best explained by lesions of the lateral premotor cortex, the sensory-motor primary cortex and the insula (Berti et al., 2005, Fotopoulou et al., 2010, Karnath et al., 2005 and Vocat et al., 2010). These studies demonstrated that the brain damage of the motor regions was not complete, but spared, for example, the supplementary motor cortex (SMA) and the pre-SMA (Berti et al., 2005). These regions, together with undamaged regions of the motor and premotor cortex, may contribute to generate motor plans and monitor the consequences of those plans. This lesional ‘defective’ evidence may explain the lack of awareness of the motor deficit. Nevertheless, it has little explanatory value when attempting to explain the situation when a patient claims to have moved his left hand, as requested by the examiner, in spite of the fact that no movement has actually been performed (Ramachandran, 1996). An active neural representation should be needed for a positive – or productive – symptom of this type (Bottini et al., 2009). The aforementioned behavioural evidence (Berti et al., 2007, Fotopoulou et al., 2008, Garbarini et al., 2012 and Pia et al., 2012) suggests that activity in spared motor regions should be present in conjunction with the false experience of movement that is reported by anosognosic patients. The explicit demonstration of the functional brain patterns preceding this non-veridical experience of movement is the evidence that we sought in the experiments presented in this paper. Is the false belief of movement accompanied by residual activity inside the motor cortices when patients are asked to move a paralysed limb, or is this belief associated with the activity of only the higher order cortices? Is such activity, if any, occurring in the spared regions of the same insulted hemisphere or does the undamaged hemisphere ‘accompany’ the false belief of movement? What is the difference in the fMRI patterns of AHP patients and patients who have the same level of hemiplegia and spatial neglect but who do not have AHP? How does brain activity in AHP and non-AHP paralysed patients compare with that of healthy controls? These were the main questions that we addressed in this paper. Using fMRI, we studied the neural correlates of motor execution or attained motor execution (for the paralysed left limb) in right-brain-damaged (RBD) patients with hemiplegia and normal awareness of their deficit and in patients with AHP. We focused on the motor activity of the upper limb, with particular emphasis on finger movements. The motor finger opposition task was interleaved with times when subjects commented upon whether they had been able to actually move the right or left fingers. The same fMRI study was also performed in 24 healthy controls whose fMRI activation patterns, excluding the areas involved by lesions in the patients, served to constrain the fMRI analyses. With this protocol, we expected to identify the neural systems that are normally implicated in motion and motor planning and its alterations in patients with AHP.

3. Results All of the patients presented severe peripersonal neglect and complete left hemiplegia (Table 1). Five patients presented AHP according to Bisiach’s scale. The three patients with AHP included in the fMRI study (P1–P3, Table 1) also claimed that they had performed bimanual actions (to raise both arms in the air as if holding a tray and clapping hands) with both hands. The remaining six non-anosognosic patients (P6–P11) did not present the slightest sign of AHP for hemiplegia and were fully aware of their motor deficit. 3.1. Evaluation of anosognosia during the fMRI experiment During the fMRI experiment, patients were alternatively asked to relax or to move either their right or their left hand and to make judgements about their performance. Of course, no overt movement was expected for the left paralysed hand, nor was it observed. Yet, this experimental condition mimicked the typical examination of motor functions when the false belief of movement was elicited. No spurious right-sided movements were observed following the instruction to move the paralysed hand. The non-AHP patients (P6–P8) performed the tasks flawlessly, moved their right hand as requested, admitted their left motor deficit and recognised their spared right hand motor function (no error were observed). Conversely, the three AHP patients included in the fMRI experiment (P1–P3) claimed to have moved their paralysed left hand even if no such action occurred (100% of the responses indicated the presence of a non-veridical experience of movement). No errors were produced when assessing the resting state periods or the normally functioning right hand. 3.2. Lesion-mapping results The comparison of the lesion distribution of AHP patients (n = 5) with that of patients without AHP (n = 6) revealed anatomical patterns similar to those in previous studies ( Berti et al., 2005, Fotopoulou et al., 2010, Karnath et al., 2005 and Vocat et al., 2010). The lesions of the AHP patients overlapped in the right basal ganglia and thalamus, in the right ventral premotor cortices (BA6 and 44) and in the insula ( Fig. 1A). A subtraction analysis of the lesion patterns ( Rorden & Karnath, 2004) showed that the lesion of these areas was at least the 50% more frequent in the AHP patients. Conversely, the same differential proportion was observed in the precentral gyrus (BA2 and 3), the supramarginal gyrus and the dorsal premotor cortex when comparing the non-AHP with the AHP patients. The result of the subtraction analysis between the AHP and non-AHP patients is illustrated in Fig. 1C. 3.3. fMRI results 3.3.1. Patients without anosognosia During the movement of the non-paralysed limb, the non-AHP patients showed the activation of the left precentral (BA4/6) and left postcentral gyrus (BA3), the left middle cingulum, the left superior temporal lobule, the left parietal lobule and left supramarginal gyrus, the left paracentral lobule, the left precuneus and the right cerebellum (Table 2). The activation of this motor and premotor network, when moving the non-paralysed limb, was confirmed by a conjunction analysis of the independent individual effects (Table 2) and by the exploration of the individual fMRI data (Supplementary Table 4). When the non-AHP patients were invited to move their left paralysed hand, no voxel survived a corrected statistical threshold.7 This group-level observation was confirmed by the exploration of the individual fMRI data (Supplementary Table 4). 3.3.2. Patients with anosognosia The AHP patients showed, for the movement of the non-paralysed limb (Table 2), activation of the left precentral gyrus (BA4/6), the left inferior parietal lobule and supramarginal gyrus, the left superior parietal gyrus and superior temporal gyrus and the left rolandic operculum. We also found the activation of the left putamen. In the right hemisphere, we found the activation of the inferior frontal gyrus, the superior temporal pole and the cerebellum. A bilateral activation of SMA was also found. More notably, when the AHP patients were asked to move their left paralysed hand, we found the activation of the left precentral (BA4/6) and postcentral gyrus (BA3), the left superior and inferior parietal lobe (BA7 and BA40), the left precuneus, and the left superior and inferior frontal gyrus. We also found activation of the right paracentral lobule of the right middle cingulum (Table 2). The left thalamus and the putamen were also activated. A bilateral activation of the SMA was also found. This pattern was confirmed by a conjunction analysis of the independent individual effects (Table 2) and by the exploration of the individual fMRI data (Supplementary Tables 1–3). 3.3.3. Comparison between AHP and non-AHP patients The comparison of the fMRI patterns between the two groups showed significant differences for the “move the left hand” condition, as indicated by the highly significant group by hand interaction effects (see Fig. 2); notably, these regions included a number of motor-related brain areas, including the SMA bilaterally, the left precentral (BA4 an BA6) and postcentral (BA3) gyri, the left inferior parietal lobule, the left precuneus, the left paracentral lobule (BA6), and the left middle portion of the cingulate gyrus and the rolandic operculum (the full list of regions is given in Table 3). We also found activation in the right superior temporal pole. Brain regions significantly more active in all of the AHP patients (AHP+) than ... Fig. 2. Brain regions significantly more active in all of the AHP patients (AHP+) than in non-AHP patients (AHP−) when requested to “move” their left paralysed hand (A and B). The bar graph (C) describes the fMRI response in the right SMA; this region was significantly more active in the AHP patients when “moving” the left hand (yellow bars) and was less active in both groups of the patients when they were truly moving the right hand (brown bars). All of the six patients had left hemiplegia, and no movement occurred for the “move the left hand” condition. Each pair of bars corresponds to one of the six patients (three AHP on the left, and three non-AHP on the right). Brain activations are visualised on a standard MNI template (Montreal Neurological Institute, MNI) thresholded at p < .001 uncorrected. Figure options Conversely, the comparison between the AHP and the non-AHP patients for the movement of the right hand did not show any difference in the brain regions involved in motor preparation or execution. The non-AHP patients did not show larger activations than the AHP patients for the “move the right hand condition”, but showed a larger activation in the left cerebellum for the “move the left hand condition” (stereotactic coordinates x = −14; y = −58; z = −32. Z score: 4.2; FWE p = .04). The lack of a corrected signal in the simple “move the left hand” effect for the non-AHP invites to treat this finding with some caution (see footnote 7). 3.3.4. Comparison between the movement condition of the two hands in AHP patients Finally, for the AHP patients, we compared the activation patterns during movement of the right hand and the “movement” of the left hand. The movement of the right hand was associated with a greater activation of the right cerebellum (stereotactic coordinates: x = 18; y = −52; z = −22; Z score: uncorrected p = .003; small volume corrected p = .017). No significant effects were observed in the opposite comparison at the multiple comparison corrected threshold. 3.3.5. Classification of the activation patterns observed in AHP patients compared with those of healthy participants We defined the level of congruity between the patients’ patterns of brain activation for a given effect and those of the healthy controls. To this end, we calculated the level of overlap between the cortical activation patterns of the control subjects while moving or imagining to move the left and the right hand, and the brain activation of the AHP patients (main effect FWE .05) in response to the command to move the paralysed or the unaffected hand. As expected, we found a large degree of congruity between the activation patterns observed in the AHP patients and the regions activated in healthy control subjects for the movement of the right hand. More interestingly, when patients were asked to move their paralysed hand (the left hand), the activation showed a substantial amount of overlap with regions activated in the controls when moving the right or the left hand, i.e., in the left posterior parietal cortex and sensory-motor cortex (Fig. 3). Classification of the activation patterns observed in the AHP patients, compared ... Fig. 3. Classification of the activation patterns observed in the AHP patients, compared with those of healthy subjects. In this contingency table, each square illustrates the level of congruity between the patients’ patterns of brain activation for a given effect and those of the healthy controls. Green and blue colours illustrate the brain regions that were activated in the healthy subjects by the motor and the motor imagery tasks, respectively. The cortical activation patterns during the motor task in the patients with AHP are illustrated in red colour. Brain activations are visualised on a normal brain. For example, in the upper left square, congruency is illustrated for the activation patterns of the controls while moving (in green) or imagining moving (in blue) their left hand and for the AHP patients in response to the command to move their paralysed hand (in red).