Introduction

The immediate and dramatic improvements seen seconds after deep brain stimulation (DBS) in a patient with essential tremor or Parkinson disease, combined with the rapid return of symptoms following DBS shutoff, suggest that its mechanism of action involves modulation of existing neuronal circuitry as opposed to the formation of new permanent neuronal connections. Thus, the term neuronal plasticity , traditionally referring to changes in neuronal structure and function that outlast the stimulation period, would not be applicable ( ). Changes in the computational network state, such as disinhibition, rather than sprouting of new axons, can best explain the time course of such changes. In contrast, the delay in symptomatic improvement seen in other forms of neuromodulation therapy ( ) is more consistent with other mechanisms of action, namely subacute to chronic formation of new neural networks via mechanisms other than simple computational state changes, which then could be characterized as neuronal plasticity. It is likely, however, that multiple mechanisms of neuromodulation are at play in states of chronic therapy. These mechanisms include representational plasticity, which can be defined as the modification of central nervous system topographic maps (for example, the homuncular mapping in primary somatosensory and motor cortices) owing to changes in neuronal input. This chapter will review what is known about neuronal plasticity, discuss the evidence of plasticity in a number of disease states in which neuromodulation is a potential therapy, and review the evidence of the connection between neuromodulation therapy and neuronal plasticity.

Topographic Organization of the Central Nervous System: Historical Overview

In , Penfield and Boldrey reported data obtained via electrical stimulation of the frontoparietal cortex in neurosurgical patients, and described orderly maps of the body present in the precentral motor cortex as well as the postcentral sensory cortex. The well-known homunculi of the precentral and postcentral motor and sensory cortices were further elaborated upon by Penfield and Rasmussen 13 years after the initial reports ( ) ( Fig. 9.1 ).

Somatosensory homunculus.

Mogilner, A.Y., 1993. Functional Organization and Plasticity of the Human Primary Somatosensory Cortex. New York University, New York.

For two decades subsequently, the consensus was that these maps were static, determined at birth or perhaps during an early critical period of development, and did not significantly change during an individual’s lifetime. In the early 1980s, pioneering work by Merzenich and others demonstrated that, in contrast to previous belief, these maps maintain the ability to reorganize in response to a variety of both peripheral and central perturbations, termed cortical plasticity . Following the transection of the median nerve in owl and squirrel monkeys, the somatosensory cortical territory previously responsive to median nerve inputs became almost immediately responsive to inputs from the uninjured radial and ulnar nerve afferents ( ). Ultimately, over a period of months, an entirely new topographic map emerged, with extensive representations of the radial and ulnar nerve territory of the hand appearing in areas previously responsive to median nerve inputs. Similarly, following amputation of one or two digits in monkeys, the cortical territory originally responsive to these digits became completely occupied by representations of the adjacent skin territories, including the adjacent digits, palmar pads, and digit stumps ( ). Similarly, when digits of monkeys were surgically fused, mapping of the somatosensory cortical area in the subsequent months revealed that the normal sharp discontinuity between the individual digit representations was abolished ( ).

Following upper extremity deafferentation, the upper limb area of the cortex ultimately became responsive to stimulation of the lower part of the face, an intracortical distance of approximately 10 mm ( ). The spatial extent of such large-scale reorganization suggested that multiple mechanisms account for these changes, including:

• 1.
  

  simple computational changes in the relative weights of excitatory and inhibitory inputs to a predefined neural matrix,

  
  
• 2.
  

  axonal sprouting,

  
  
• 3.
  

  changes in synaptic size, number, and morphology,

  
  
• 4
  

  .reorganization at the subcortical level ( ).

The first direct evidence of similar plasticity occurring in humans was demonstrated in 1993 by the author and colleagues, via magnetoencephalography (MEG), a noninvasive method of brain mapping utilizing recordings and localizations of the weak magnetic fields produced by neural activity ( ) ( Fig. 9.2 ). The primary somatosensory cortex representation of the hand was mapped in two adults with syndactyly, both prior to and following surgical separation of the fused digits. Prior to surgery, cortical maps were abnormal, demonstrating shrunken hand representations without the usual somatotopy. Within weeks after digit separation, cortical reorganization was noted to occur, spanning distances of approximately 1 cm, and resulting in appropriate somatotopic representations ( Fig. 9.3 ).

Somatotopic organization of the hand area of the primary somatosensory cortex in normal adults as demonstrated by magnetoencephalography (MEG). (A and B) MEG dipole source locations for the digits of the hand in one normal adult projected onto a three-dimensional MRI reconstruction of the subject’s brain, with the color key shown in (C). (D) Two-dimensional plot of the aforementioned dipole locations showing the average ( symbols ) and standard errors (surrounding gray ovals ) of localization for each finger in the y,z (coronal) plane. (E) Composite intersubject map of hand area in nine subjects obtained by repetitive least-squares minimization of same-finger distances for all fingers between subjects, noting a distance over 1 cm between the cortical representation of the thumb and the little finger.

Cortical plasticity following partial surgical reconstruction of complex syndactyly in an adult. Small finger was separated from central ones, which remained together. Digit 2 was congenitally absent. (A) Three-dimensional MRI reconstruction of the patient’s brain, with the cortex anterior to the postcentral gyrus removed to visualize dipole sources. (B) Presurgical map with dipole locations obtained for the thumb and middle and little fingers; the hand is shown on the right with the color-coded key. Coronal (top) and sagittal (bottom) graphs of dipole locations (mean + SEM) are shown for these three digits studied. Maps show significant overlap of digit locations and a reduced inferior–superior extent of the hand area compared to normative data. (C) Hand map following surgical separation of digit 5. Data shown were obtained 7 days after surgery. Coronal (top) and parasagittal (bottom) views illustrate that the fingers have attained distinct cortical locations. (D) Plot of middle–little finger distance over time. The patient was studied before surgery five times over a period of 6 months and 1, 4, and 6 weeks after surgery. The thick vertical bar indicates the date of surgery (September 3). The cortical interfinger distance increased by 2.9 + 1 mm ( P < .001) following surgery, and this increase was observable as soon as 1 week after surgery.

Subsequent functional imaging studies utilizing MEG, and later functional MRI (fMRI), demonstrated evidence of plasticity in patients with a variety of both central and peripheral nervous system injuries, including amputation, spinal cord injury, peripheral nerve injury, and stroke ( ). Reorganization has been demonstrated to occur during the rehabilitation period following neurologic injury ( ). Analogous plastic changes have been demonstrated to occur following short-term motor skill learning ( ), and trained musicians show enlarged representations both of the hand used to play the instrument and in the tonotopic map of the auditory cortex compared with controls ( ).

Neuronal Plasticity in Disease States

A variety of chronic conditions treated with neuromodulation are associated with neuronal plasticity at multiple levels along the neuraxis. Whether this plasticity represents the generator of the pathology or is merely an epiphenomenon remains unclear; nonetheless these findings must be taken into account when postulating mechanisms of action of neuromodulation therapies.