Before SC mStim During SCmStim

Knee: 78

Knee: 78

Ankle: 102

Ankle: 102

FIGURE 4.8 Movements Evoked by SC^Stim in Awake Animals. A. An example of knee extension obtained by stimulating through a single electrode in an awake animal one month after implantation (maximum of 144 |A, 300 |s, 50°/s train). B. An example of ankle dor-siflexion elicited by stimulating through a single electrode in an awake animal one month after implantation (maximum of 200 |A, 300 |s, 50°/s train). C. An example of a whole-limb extension synergy generated when stimulating through one electrode in an awake, weight-bearing animal two months after implantation (maximum of 240 | A, 300 | s, 50°/s train). The generated torques were large enough to lift the animal's hindquarters. The examples in A, B and C are from three different animals.

With some electrodes, the characteristics of the generated movements substantially changed as the stimulus pulse amplitude was increased. Starting with single joint movements at low stimulus strengths, more intense stimulation through these electrodes (up to 240 pA) often resulted in a whole-limb extensor synergy, similar to that shown in Figure 4.8C. Though the rostral-caudal location of these electrodes varied along the lumbosacral spinal cord, the extensor synergy seemed to be the "default" synergy elicited with higher stimulus strengths. Such default synergy may reflect the existence of a dominant extensor interneuronal network within the lum-bosacral cord, the presence of which provides an advantageous extensor bias needed for postural and antigravity movement control. The default extensor synergy may also be an artifact of electrode location in the ventral portion of the cord. Given that flexor motoneuronal pools tend to have a more dorsal location in the cord than extensor pools, it is possible that these electrodes were positioned deep in the cord and close to ventral rootlets in regions activating extensor muscles.53 55 61 62 Therefore, the default synergy may be the result of stimulus spread to these rootlets. Closer examination of the relationship between the location of the implanted microwires and the resulting movement synergy is needed.

The high proportion of electrodes producing selective muscle activation in the awake cat reaffirms previous results in acute studies5455 61 62 and provides the means for designing flexible control strategies which allow for the generation of novel and functional movements. Combined with the whole limb movements seen when stimulating through one third of the electrodes, a large repertoire of stereotypical and novel movements could be generated using SCpStim.

Across all animals (n = 9), at least two thirds of the implanted electrodes (range: 67 to 100%, mean ± SD: 85.7 ± 13.2%) consistently elicited contractions in the same muscle(s) throughout the period tested. This indicates that, with suitable fixation techniques, microwires implanted in the spinal cord remain securely in place for long periods of time. For all electrodes, stimulus threshold doubled during the first ten days following implantation. This was presumably due to electrode encapsulation, a product of the physiological reaction to implanted foreign objects. Following this initial increase, stimulus thresholds remained, on average, constant throughout the duration of implantation.

SCpStim did not cause any apparent discomfort in the intact animal even when powerful muscle contractions were evoked. This finding is very important for any future clinical application of the work since it implies that pain pathways are not activated with SCpStim, thus removing a potential barrier to clinical implementation and acceptance by patients.

4.4.5 Sensory Interaction with Spinal Cord Microstimulation

Though, on average, stimulus threshold remained constant after the first ten days of microwire implantation, day-to-day variations in threshold of ± 15 pA were detected.

The oscillations in threshold were found to be dependent on the postural state and activity level of the animal. We observed that in many electrodes, thresholds for activating muscles were lowered by palpating the muscles and their synergists, by imposing joint rotation that stretched the muscles or, in some cases, by lightly touching the skin over and around them. This reduction in threshold is presumably due to the depolarization of motor and premotor neurons by sensory inputs.

The effect of afferent input on stimulus threshold was quantified. In awake animals, implanted microwires were independently activated at stimulus levels producing threshold EMG responses in one of the muscles of interest (e.g., quadriceps), and the activated muscle was stretched and shortened by manually imposed cyclical joint rotation. A gyroscope was used to measure the angular velocity of the imposed movements. At least 50 consecutive 2/s pulses were delivered through each electrode during the imposed movements, and the elicited EMG responses from all electrodes were pooled. EMG records were rectified and activity within a 20-ms window starting 10 ms following a stimulus pulse was integrated and plotted against the instantaneous angular velocity measured 2 ms prior to the stimulus pulse. The mean and standard error of integrated EMG activity within 50°/s bins were calculated and the data were curve-fitted using a power-law relationship. Figure 4.9 shows quadriceps EMG responses obtained during imposed knee flexion in an awake animal. Imposed knee flexion (lengthening of muscle) increased quadriceps EMG responses in a power-law relationship that is consistent with motoneuronal depolarization resulting largely from spindle Ia afferent input.68

4.4.6 Wire Corrosion Testing

Once the stability of implanted wires and evoked responses was established, it was necessary to find a suitable microwire for long-term implantation and SCpStim in human subjects. A suitable wire was defined as one that would maintain stable impedance for at least 10 years of use.

Accelerated in vitro corrosion testing was performed on stainless steel (SS) and platinum-iridium (Pt-Ir) wires with varying diameters and stiffness. The testing was performed on five 30-40 pm 304 SS wires, three 25 pm 90%-10% Pt-Ir wires and four 30 pm 80%-20% Pt-Ir wires. The microwires (insulated except for 30-80 pm at the tip) were placed in separate saline (0.9% NaCl) baths and a multistrand 316 SS wire in each bath served as the reference electrode. Constant-current, biphasic pulses (300-pA amplitude, 100-ps duration) were delivered at a rate of 125-500/s through each of the wires 24 hrs/day and the impedance of the wires was periodically measured.

Figure 4.10 tracks the changes in microwire impedance over time. The in vitro stimulation time is expressed in terms of anticipated human use time. This conversion was based on an anticipated human use of 4 hrs/day with stimuli delivered at a rate of 50°/s. The figure shows that four of the five SS wires corroded within the first four days of testing (i.e., within the first three months of human use). The impedance of the fifth SS and all Pt-Ir wires remained constant throughout the three months' duration of testing. These results demonstrate that the Pt-Ir microwires can survive more than the equivalent of 50 years of human stimulation. We therefore conclude that Pt-Ir may be a suitable wire for long-term SCpStim in human subjects.

FIGURE 4.9 Effect of Afferent Input on SC^Stim Threshold. Mean and standard error of integrated quadriceps EMG responses are plotted against angular velocity of imposed knee flexion (divided into 50°/s bins). Responses were obtained from an awake animal in which six spinal cord microwires activating quadriceps were independently stimulated at levels producing threshold EMG responses in the stationary limb. At least 50 consecutive 2/s pulses were delivered through each electrode. Phases of knee flexion (positive angular velocities) were associated with increased quadriceps EMG responses. The fitted curve indicates a power-law relationship with an exponent of 0.6 (r = 0.98, p < 0.0001).

FIGURE 4.9 Effect of Afferent Input on SC^Stim Threshold. Mean and standard error of integrated quadriceps EMG responses are plotted against angular velocity of imposed knee flexion (divided into 50°/s bins). Responses were obtained from an awake animal in which six spinal cord microwires activating quadriceps were independently stimulated at levels producing threshold EMG responses in the stationary limb. At least 50 consecutive 2/s pulses were delivered through each electrode. Phases of knee flexion (positive angular velocities) were associated with increased quadriceps EMG responses. The fitted curve indicates a power-law relationship with an exponent of 0.6 (r = 0.98, p < 0.0001).

4.4.7 Tissue Reaction to Implanted Wires and Long-Term Stimulation

Implanted animals were maintained for up to six months. Throughout this time, no deficits were noticed in activities of daily life. This suggests that the implanted microwires were not physically damaging the spinal cord, and the overall tissue displacement by the electrodes was minimal. Even though the animals actively played and jumped around the laboratory and housing area, the stability of evoked contractions and gross histology suggest that the electrodes remained fixed in place and caused no significant tissue damage. Therefore, it is anticipated that complications due to the fixation of electrodes in the spinal cord of paralyzed individuals who are much less physically active would be minimal. Gross examination of the spinal cord upon termination of the experiment showed mild thickening of the dura

FIGURE 4.10 Accelerated In Vitro Wire Corrosion Testing. Microwire impedance is plotted against anticipated duration of human use. Four of the five SS wires corroded within the first three months of human use. The remaining SS wire and all Pt-Ir wires survived more than 50 years of anticipated human use.

FIGURE 4.10 Accelerated In Vitro Wire Corrosion Testing. Microwire impedance is plotted against anticipated duration of human use. Four of the five SS wires corroded within the first three months of human use. The remaining SS wire and all Pt-Ir wires survived more than 50 years of anticipated human use.

mater on the dorsal surface of the cord immediately above the implanted region in two animals. In the remaining seven animals, the spinal cord looked normal. Previous studies of chronically implanted microwires performed by the Huntington Medical Research Institute in Pasadena, CA, indicated that the cellular damage in the spinal cord was limited and localized.69

4.4.8 Perspective and Conclusion

Our chronic experiments tested the feasibility of SCjuStim under the most demanding conditions: intact animals, free to perform normal bodily movements. We were particularly interested in the quality of the movements that could be elicited from within the active, non-anaesthetized spinal cord, in assessing the stability of the evoked movements over time and in determining whether the stimulation caused discomfort.

Our results demonstrated that single-joint and coordinated multi-joint movements can be obtained in the awake or deeply anaesthetized animal by stimulating the spinal cord through implanted microwires. The majority of electrodes generating extensor movements produced torques adequate for weight-bearing. The selective activation of individual muscles (such as tibialis anterior) combined with the whole-limb synergistic movements provide a range of functional movements that could be useful in restoring some mobility in paraplegia. The long-term stability of implanted wires, evoked muscle responses and stimulus thresholds affirm that the intraspinal wires remain securely in place and cause no damage to the spinal cord. The lack of any apparent discomfort experienced by the animals during stimulation of the ventral cord, even when powerful contractions were elicited, indicates that pain pathways are not activated with SCpStim in this region. The results therefore support the idea that SCpStim could be a feasible clinical tool for restoring motor function following spinal cord injury, head trauma or stroke. The suitability of Pt-Ir for long-term microstimulation suggests that human implants may be implemented sooner than originally anticipated.

4.5 DISCUSSION AND CONCLUSIONS

4.5.1 Prospects

Prospects for successful FES systems based on intraspinal implanted electrodes appear reasonable. In experiments ranging across several types of tetrapod vertebrates it seems that similar organizational and functional principles may apply in spinal cord and limb control across species, albeit with variations in detail and complexity. This chapter has described methods for gaining control of the spinal circuits that support movement and also gaining direct control of motor pools. It seems clear from the data presented that recruiting ensemble circuits is possible, obtaining "for free" some of the controls organized in the spinal cord. These ensemble circuits may be matched to the musculoskeletal complex and organized into modules that make sense in the context of tasks that frequently recur. Evolution may have constructed sets of such modules that are pre-adapted for commonly occurring tasks (locomotion, grooming, reaching). An FES control scheme that recruits natural circuit modules in the spinal cord may be better suited to motor learning by a user of FES than a system based around individual muscles alone. Issues of stability, muscle redundancy and order of recruitment might be solved at the outset by the spinal cord rather than posing problems for the user. Users of such intraspinal neuroprostheses might control stimulation patterns of recruitment in spinal cord through a more conventional interface or else via a second implant.70

On the basis of the work reviewed and presented here it seems clear that to build a successful and optimum FES intraspinal stimulating system we will need a detailed understanding of the collection of spinal reflex patterns, pattern generators and their integration. Spinal reflex systems interact with one another and will interact with FES recruited circuits. Similarly, in a partial lesion, any spinal elements which FES recruited would have to be integrated with spared descending controls and with fine fractionated movements. This type of data must be an underpinning of intraspinal stimulation research programs.

4.5.2 Future Issues and Directions

Several things have been omitted from the presentation in this chapter which bear directly on use of intraspinal stimulation and neural prostheses. Among these are in-depth discussion of integration of the intraspinal neural prostheses with other devices such as cortical implants, neuromodulating pumps or other delivery systems; integration of an FES implant with rehabilitative therapies; and the integration of FES implants with regeneration promoting transplants or gene therapies.

4.5.2.1 Integration With Other Prostheses: Cortical Recording

It seems clear to some that the ideal system of intraspinal stimulation would draw its control signals from a cortical implant (see Chapter 6), which was recording intentional and motor control activity at the "upper motoneuron" or support circuits70 (see Chapter 8). In effect this would represent a kind of neural bypass. The meshing of cortical and spinal motor representations and enabling appropriate learning at the interface will be critical to such an effort. This must depend in part on how movement and interaction with the environment are represented and controlled in these brain areas. "Primitives," "synergies" or "reflex assemblies" that are embedded in the spinal cord may embody principles of muscle recruitment that take into account the special biomechanical roles of particular muscle architectures.71 Restoring or reeducating the integration of voluntary, cortical and spinal elements and representations after regeneration or an implanted artificial stimulating/recording bridge will not be a trivial task. In part this may depend on the developmental stage. For example, although young animals develop cortical representations and learn strategies to cope with very severe disabilities,72,73 it currently appears that adult animals do not. Clearly, the more normal are the properties and controls that are implemented by an FES system, the more facile will be subsequent control and learning by the user of a cortical implant. Intraspinal FES offers the promise of a more intuitive interface for a user who is learning operation of an "FES bypass" type of system.

4.5.2.2 Cord State: Neuromodulators, Inhibition and Specificity of Stimulation

In complete spinal injuries it is very likely that intraspinal FES can be greatly augmented by appropriately restored levels of neuromodulators. Modulation by descending systems is likely to be an important aspect of normal control and motor func-tion.2'28'46'7475 Loss of some neuromodulation following injury is one of many factors responsible for the spinal cord's worsened initial state. The effects of neuromodulator and transmitter restoration have been shown in several studies.28,76 Baclofen is commonly used to reduce spinal spasticity and can be expected to form a useful adjunct to FES in the clinic. Which combinations of neuromodulation may be best to support an FES system, and more specifically an intraspinal implant, are relatively unexplored.

4.5.3 New Directions: Fiber Optic Methods in the Future?

Recently Giszter and colleagues have begun to consider the feasibility of a fiber optic system for delivering stimulation, inhibitory stimuli and neuromodulators in an intraspinal implant arrangement. In the past several years caged compounds have become routinely available. In caged compounds the bioactivity of a "caged" agent is masked by a chemical "cage" that can be removed by various means, thus allowing the agent to act in the target tissue. Light excitation using specific wavelengths including dual photon techniques (depending on the cage agent/structure) is one means of uncaging. Using light for uncaging may allow a novel type of FES device to be constructed. In an FES application we propose using fiber optic delivery of light pulses for uncaging. We believe that fiber optic release of chemically caged transmitters, modulators, and receptor agonists will provide a method of rapid multisite and precisely targeted release of excitatory, inhibitory and modulatory agents. This scheme is particularly attractive because it is likely that when the intact spinal cord is controlled by the brain there are continuously varying focal changes. Neuromodulation, excitation and inhibition are varied at different sites and in different ways depending on the motor task.77 Rapid switching of light-pulses and delivered wavelength on an implanted fiber, and rapid switching of sources among several implanted fibers in different targets may ultimately allow different patterns of uncaging of multiple compounds. Currently, as a result of communications and other applications, fiber switching technologies and fiber-optic sensors are undergoing rapid and continuous improvement. It is possible that feedback control of local neurotransmitter levels may be feasible. Even in non-FES applications (such as control of spasticity) it is conceivable that these technologies could represent a considerable advance over single intrathecal pumps for baclofen, just as the pump is an advance over oral ingestion. In an FES application of these technologies the limiting issues are likely to be (1) possible caging agent byproduct side effects,

(2) excitation wavelength side effects (heat, mutagenesis, protein degradation),

(3) the engineering of appropriate cages of appropriate wavelength selectivity, (4) the avoidance of spurious uncaging and release, and finally (5) the long term tissue compatibility and materials issues associated with implanted fibers. However, despite these concerns it is clear that fiber optic uncaging is becoming an established stimulation method78 and (at least in frogs) focal transmitter delivery at intraspinal sites is effective in recruiting patterned motor responses and "primitives."51

4.5.4 Conclusions

The prognosis for applications of intraspinal FES in the clinic at some future time seems good. Various neuroprosthetic designs are being tested, and a concerted and coherent effort is being applied to this approach to the FES problem by several independent groups.

Was this article helpful?

0 0

Post a comment