Imbalance in multiple sclerosis: a result of slowed spinal somatosensory conduction

Abstract

Balance problems and falls are common in people with multiple sclerosis (MS) but their cause and nature are not well understood. It is known that MS affects many areas of the central nervous system that can impact postural responses to maintain balance, including the cerebellum and the spinal cord. Cerebellar balance disorders are associated with normal latencies but reduced scaling of postural responses. We therefore examined the latency and scaling of automatic postural responses, and their relationship to somatosensory evoked potentials (SSEPs), in ten people with MS and imbalance and ten age-, sex-matched, healthy controls. The latency and scaling of postural responses to backward surface translations of five different velocities and amplitudes, and the latency of spinal and supraspinal somatosensory conduction, were examined. Subjects with MS had large, but very delayed automatic postural response latencies compared to controls (161 +/- 31 ms vs. 102 +/- 21 ms, p < 0.01) and these postural response latencies correlated with the latencies of their spinal SSEPs (r = 0.73, p < 0.01). Subjects with MS also had normal or excessive scaling of postural response amplitude to perturbation velocity and amplitude. Longer latency postural responses were associated with less velocity scaling and more amplitude scaling. Balance deficits in people with MS appear to be caused by slowed spinal somatosensory conduction and not by cerebellar involvement. People with MS appear to compensate for their slowed spinal somatosensory conduction by increasing the amplitude scaling and the magnitude of their postural responses.

Figure 1

Postural response and SSEP measurement methods Figure 1A. Balance Platform. The subject stands with each foot on a force plate of the movable balance platform. The platform is translated backward, which results in forward displacement of the body over the feet. Measured responses to this support surface translation include the plantarflexion torque exerted by each foot and the latency of EMG activation of the gastrocnemius muscles, the prime movers to restore equilibrium. Figure 1B. Location of SSEPs recorded from stimulation of the posterior tibial nerve in the leg or the median nerve in the arm; potentials associated with posterior tibial nerve stimulation are shown in italics. Intervals corresponding to total afferent conduction time (ACT) and its spinal (ACTsp) and supraspinal (ACTsup) components are also shown. $$\begin{array}{l}\text{ACT}=\text{P}37-\text{N}20\\ \text{ACTsup}=\text{N}19-\text{P}13\\ \text{ACTsp}=\text{ACT}-\text{ACTsup}\end{array}$$

Figure 2

Relationships between postural response latencies and SSEP latencies. Figure 2A and 2B. Relationship between postural response latency (GAS EMG) to backward displacement and spinal SSEP conduction times (shown in Figure 2A) and supraspinal SSEP conduction times (shown in Figure 2B). The x’s on Figure 2A represent postural response latencies for subjects with absent spinal SSEPs. These subject’s values were not included in the linear regression.

Figure 3

Scaling of postural responses in MS and control subjects from this study and cerebellar subjects from our previous study (Horak et al. 1994). Figure 3A. Predictive amplitude scaling of rate of initial postural response to block presentation of increasing platform amplitudes, for MS (circles), control (squares) and cerebellar (triangles) subjects. Figure 3B. Reactive velocity scaling of rate of initial postural response to randomly presented platform velocities, for MS (circles), control (dark squares) and cerebellar (triangles) subjects. The slope values shown represent the group means and SEMs.

Figure 4

Relationships of postural response latencies and scaling Figure 4A. Relationship of postural response latency (GAS EMG) to the amplitude scaling slope for each individual MS subject (values for both legs are included) Figure 4B. Relationship of postural response latency (GAS EMG) to the velocity scaling slope for each individual MS subject (values for both legs are included).