The functional anatomy of a hysterical paralysis

Abstract

The concept of a conversion disorder (such as hysterical paralysis) has always been controversial (Ron, M.A. (1996). Somatization and conversion disorders. In: B.S. Fogel, R.S. Schiffer & S.M. Rao (Eds.), Neuropsychiatry. Williams and Wilkins, Baltimore, MD). Although the diagnosis is recognised by current psychiatric taxonomies, many physicians still regard such disorders either as feigned or as failure to find the responsible organic cause for the patient's symptoms. We report a woman with left sided paralysis (and without somatosensory loss) in whom no organic disease or structural lesion could be found. By contrast, psychological trauma was associated with the onset and recurrent exacerbation of her hemiparalysis. We recorded brain activity when the patient prepared to move and tried to move her paralysed (left) leg and when she prepared to move and did move her good (right) leg. Preparing to move or moving her good leg, and also preparing to move her paralysed leg, activated motor and/or premotor areas previously described with movement preparation and execution. The attempt to move the paralysed leg failed to activate right primary motor cortex. Instead, the right orbito-frontal and right anterior cingulate cortex were significantly activated. We suggest that these two areas inhibit prefrontal (willed) effects on the right primary motor cortex when the patient tries to move her left leg.

Introduction

We studied a right-handed woman of 45 who meets the Diagnostic and Statistical Manual of Mental Disorders (DSM IV) criteria for conversion disorder. Her main symptom is left sided paralysis of two-and-a-half years' standing. No movement of the left leg has been observed over this period, although there is now limited and effortful movement of the left shoulder to command. Psychological stress and trauma were associated with the initiation and recurrent exacerbation of her condition in which she becomes more generally unresponsive. The patient's previous medical history was remarkable for episodes of depression, some of which were accompanied by mutism and lower limb weakness (also of presumed hysterical origin). Neuropsychological testing was normal (verbal intelligence, attention, language, visuo-spatial cognition, visual recognition, buccofacial praxis and praxis of the right hand and arm) except for mild impairments in visual and verbal recall. The patient did not meet any of the DSM IV criteria for malingering.

The neurological examination is entirely normal except for the paralysis of the left arm and leg (with intact tendon reflexes). Investigations for organic causes of the hemiparalysis have been extensive. Cranial computerised tomography (CCT), repeated magnetic resonance imaging (MRI) of the brain and spinal cord, blood tests, electro-cardiogram, doppler ultrasound of the carotid arteries, and central motor conduction studies including transcranial magnetic stimulation, were all normal.

Prior studies of electrical stimulation in both humans and other animals have revealed negative motor areas which can inhibit spontaneous movements (the so-called `arrest reaction'). These negative areas include the anterior cingulate and orbito-frontal cortex (Kaada, 1960; Lüders et al., 1995). A recent positron emission tomography study of normal subjects (Paus et al., 1993) has also shown that anterior cingulate cortex plays a crucial role in the suppression of inappropriate motor responses. That anterior cingulate and orbito-frontal regions have inhibitory functions is also suggested by their activation during the Stroop task. Functional imaging studies have implicated these areas in the suppression of competing responses to colour names presented in incongruent colours (Bench et al., 1993; Carter et al., 1995; George et al., 1994; Pardo et al., 1990). Our patient, then, is an appropriate subject in whom to test the hypothesis that these areas are implicated in hysterical paralysis: She would appear to be a pure case of `hysterical neurosis, conversion type' (Ron, 1996), and one furthermore who can act as her own control due to the clearcut distinction between the motor capacity of her left (bad) and right (good) legs.

We accordingly measured cerebral activity, indexed by changes in regional cerebral blood flow (rCBF) in a control (no movement — 6 repeats) and four experimental conditions in a single session. During each rCBF measurement we also took continuous surface electromyographic (EMG) recordings from the legs. In the four experimental conditions (3 repeats each) the patient was required to prepare to lift her good right (a) or her bad left (b) leg on each beat of a metronome (0.5 Hz), but not to actually move it unless touched on that leg (that touch was given only before or after the actual scanning period); in addition, the patient was asked to lift her good right leg (c) or to lift her bad left leg (d) on each beat of the metronome. She was instructed to try as hard as possible to lift each leg. Both legs were strapped down so that physical movement of the right leg was restricted. This was done to control for the absence of movement when the patient tried to move her paralyzed leg.

The areas that were activated (relative to the baseline, P<0.001) when the patient moved her right leg (against restraint) were as expected (Fink et al., 1997). They included the dorsolateral prefrontal cortex bilaterally (DLPFC) and left lateral premotor areas, left primary sensori-motor cortex (S1 and M1), bilateral secondary somatosensory areas (inferior parietal cortex), and the vermis and the cerebellar hemispheres bilaterally. As predicted (Stephan et al., 1995), preparation to move the right leg (relative to the baseline, P<0.001) activated a subset of these areas, including DLPFC bilaterally, right lateral premotor and bilateral inferior parietal cortex and the vermis and cerebellar hemispheres, but not the left primary sensori-motor cortex. Activations specific to movements of the right (good) leg relative to attempts to move the left (bad) leg are shown in Table 1 and Fig. 1.

Preparation to move the (bad) left leg activated the left lateral premotor cortex and the cerebellar hemispheres bilaterally (relative to baseline, P<0.001), indicating the patient's `readiness' to move the paralyzed leg upon touch. This finding is evidence against the idea that the hemiparalysis is feigned. Attempting to move the paralyzed leg led to activations (relative to baseline, P<0.001) of movement-related areas including left DLPFC and the cerebellar hemispheres bilaterally, which is again evidence against faking. No activation of right premotor areas or of right primary sensori-motor cortex was observed (consistent with the lack of any indication of muscle excitation in her bad left leg on the surface EMG). Instead, as predicted, the right anterior cingulate was significantly activated during this condition (and this condition alone). Cingulate activation was confirmed (P<0.001) when the attempt to move the left leg was compared with preparing to move the left leg. In addition, this latter comparison revealed significant activation of right orbito-frontal cortex. Activations specific to attempted movement of the left leg relative to movement of the right leg are shown in Table 1 and Fig. 2. The right anterior cingulate and the right orbito-frontal cortex (each with a distinct local maximum) are the only areas that are differentially activated in this comparison. They are regions that have been previously implicated in action, emotion, and motor inhibition (Devinsky et al., 1995).

Charcot (1889)conjectured that hysterical paralysis implicated what he called `dynamic or functional lesions' in the cortical motor zone opposite the paralysis. These functional lesions were regarded as akin to localised oedema, anaemia or active hyperaemia `of which no trace is found after death' (Freud, 1893).

In partial support of Charcot's position we have found a localised failure to increase rCBF in right motor cortex. But this failure only emerges when the patient tries to move her left leg. Unlike the case of either stroke or Charcot's putative `functional lesion', our patient shows no significant asymmetry of rCBF in the motor cortices at rest (as indicated by the baseline scans). Furthermore, the activations when she prepares to move her paralysed leg and her good leg are similar. It is only when she tries to move the left (bad) leg, that significant activation of the right anterior cingulate and right orbito-frontal cortex is found. Furthermore, downstream activation of cerebellum suggests that the lack of motor cortical activation is selective and specific. We propose that, in our patient, anterior cingulate and orbito-frontal cortex actively inhibit movement of the left leg despite DLPFC activation and downstream activation in the cerebellum. More speculatively, it is possible that orbito-frontal cortex is the distal source of the unconscious inhibition (Fuster, 1981) while the anterior cingulate (which mediates between emotion and action) is the proximal instrument that `disconnects' premotor/prefrontal areas from primary motor cortex. This functional disconnection hypothesis is further supported by the fact that premotor frontal regions are activated to a similar extent when the patient moves her right leg and tries to move her left leg. Since there is no evidence of on-going structural or functional pathology, we conjecture that it is the will to move that triggers the hemiparalysis via the pathological activation of orbito-frontal and cingulate cortex.

It is unlikely that the unilateral activation of the right anterior cingulate in our study is due to an adverse emotional state of the patient when instructed to move her (bad) left leg. She showed no behavioural indications of anxiety or distress during scanning. Likewise, we did not observe significantly increased activation of limbic structures specifically concerned with potent emotions (e.g. the amygdaloid nuclear complex) during attempted raising of the bad left leg. It is also unlikely that the activation of right anterior cingulate is due to the patient employing motor imagery during her attempts to lift her left leg. If this were so, one would expect bilateral activation of many areas in addition to the cingulate, including, for example, supplementary motor area (SMA) and more lateral premotor areas (Decety et al., 1994; Stephan et al., 1995). Furthermore, activation of orbito-frontal cortex has not been described in previous studies of motor imagery (Crammond, 1997).

We conclude, then, with Vogt et al. (1992), that anterior cingulate provides `a meeting place for interactions between cognitive and motivational processes, particularly in relation to the generation of motor output.' Hysteria, however, is a protean disorder and one would not necessarily expect that our current neurobiologic findings will generalize to hysterical blindness, deafness, or somatosensory loss. Nonetheless, the conceptual pattern of our results is consistent with the case report of Tiihonen et al. (1995)in which psychogenic paraesthesia was associated with `simultaneous activation of frontal inhibitory areas and inhibition of the somatosensory cortex.'