A psychophysical study of the mechanisms of sensory recovery following nerve injury in humans.

Twenty-four subjects were studied before and up to 1 year after surgery that produced injury to a major sensory branch of the trigeminal nerve. We employed a battery of 11 psychophysical tests, in which the neural mechanisms underlying performance are understood, to study the basis of recovery following nerve injury. Immediately after nerve injury, sensation was profoundly impaired in all subjects. In the following weeks and months, the recovery of performance proceeded in an orderly fashion. Although the rates of recovery varied between subjects, the order of recovery between tasks did not. The recovery rates fell into three distinct categories. Recovery in one task, brush-stroke directional discrimination, was most rapid. Two weeks after nerve injury, 52% of subjects could discriminate brush-stroke direction; by 3 months only one subject could not perform this task. The second category comprised recovery rates for pain thresholds for noxious heat, cold and mechanical stimuli, and to preinjury performance in tasks assessing touch and vibration detection, two-point discrimination, cooling detection and subjective magnitude estimation of mechanical force. The third, slowest group included recovery rates for warming detection and grating orientation discrimination. Early recovery to preinjury performance levels in the brush-stroke direction and one-point versus two-point discrimination tasks was correlated with later recovery to near normal performance in the grating orientation task. The grating orientation task was unique in providing a measure that corresponded consistently with the subjects' reports of sensory deficits. Our psychophysical findings are consistent with neurophysiological data showing that the major primary afferent fibre classes reinnervate the skin at a similar rate. A hypothesis that accounts for the psychophysical findings in this study is that differences in recovery rates between tasks is determined largely by their relative dependencies on functional innervation density. Alternative hypotheses are considered.