TITLE

Abstract

Stimulation of the anterior nerve roots to assist paraplegics in functional standing has been investigated in two subjects. Both have achieved hands supported standing, but function has been limited primarily by excessive hip flexion. By mapping motor responses to root stimulation we have demonstrated the difficulty in producing distinct moments, such as hip extension that may improve the posture. It is therefore tentatively concluded that root stimulation may not be the best approach to achieve functional standing.

Background

Electrical stimulation of paralysed lower limb muscles in complete spinal cord lesion paraplegics has been successful in providing exercise or assisting in useful functions, such as standing for physiological and psychosocial benefits. Predominantly this work has used surface stimulation, but some paraplegics regard such systems as unaesthetic and it is limited in specificity and access to muscle responses and hence lower limb movements. Stimulation by implant may address this.

Reviewing the expected advantages and disadvantages in stimulation by peripheral nerve and nerve root electrodes, Rushton [1] suggested that root stimulation for leg control, implemented by similar techniques established in the sacral anterior root stimulator implant (SARSI) for bladder control, may be feasible. It offered the advantage of exposing all lower motor neurons for placement of electrodes and therefore increased access to muscle responses. The main disadvantages were the risk of root damage during surgery and the expected muscle co-activation, arising from the anatomical innervation, producing mixed movements.

From this, the lumbosacral anterior root stimulator implant (LARSI) project was started in 1992, to measure lower limb muscle responses during nerve root stimulation in complete spinal cord lesion paraplegics and to investigate whether this could be applied to assist in functional standing. This paper examines whether standing by this method is a practical option.

Methods

Subjects (T3-T12, complete traumatic spinal cord lesion) are selected [2], before following three months of muscle retraining using surface stimulation. Progression to implant is dependent upon demonstrating safe standing at home by surface stimulation [3].

The surgical procedure to implant LARSI requires identification of each of the L2-S2 anterior roots, left and right, by anatomy and by stimulation, at the cauda equina before placement in one of the 12 intradural electrodes. There is no rhizotomy of posterior roots, unlike that in the conventional SARSI procedure.

Post-implant testing is primarily through palpation and using the multi-moment chair [4] to measure joint moments in both lower limbs simultaneously during stimulation. These are used to identify threshold and maximal levels from single root stimulation and to select appropriate stimulation patterns for muscle retraining. Further, by mapping stimulation levels to muscle responses and investigating co-activation effects, combinations of roots to provide useful function, such as standing, are developed and tried.

Results

Recruitment: The databases of two U.K. spinal injuries units were screened for potential subjects. From a population of 522 traumatic complete lesion paraplegics, 115 (22%) were considered physically suitable from their medical records and contacted. 50 showed interest and after a clinical assessment 28 were recruited. 10 achieved hands-supported standing at home by surface stimulation. 7 subjects were considered for implant, but 5 declined because of the potential risks (loss of sensation, reflex erection and bladder control) associated with surgery. Only 2 have been implanted; subject 1 (female, T9 from a RTA in 1990, implanted 1994), and subject 2 (male, T9/T10 from a fall in 1989, implanted 1996) who also had a SARSI implanted at the same time.

Subject 1: Root damage at surgery was limited with responses from 11 roots available. Joint moments measured in the multi-moment chair have demonstrated the difficulties in producing distinct movements, either from single or combined root stimulation [5]; knee extension accompanied by hip flexion, hip extension by plantarflexion and ankle inversion. Plantarflexion and ankle inversion is invariably highly correlated, as are hip internal rotation and knee extension.

Subject 2: Nerve damage was experienced in 7 roots during surgery (these were significantly larger in diameter than the first subject) but have since recovered. Unfortunately he had intermittent problems with the implant, that required a replacement (mid 1999) following changes to the manufacturing methods. Following the repair, all 12 roots were available and muscle retraining was restarted. Unfortunately he now presents with no response from L2R (possibly a cable break - under investigation), resulting in right knee extension being weak (grade 2). Selective stimulation is required to provide adequate left knee extension (grade 4) with minimal inversion and plantarflexion. On a few occasions he has achieved standing in the clinic, but this hasn’t been followed up extensively because of the implant problems and the subject being unavailable for testing for 6 months each year.

Assisted standing: The results we present are from subject 1, though subject 2’s motor responses during stimulation and few stands to date do not suggest that there would be any significant differences. Subject 1 has been standing since 1995 and at home since 1996. Standing is open loop using hand supports, with different patterns used in ‘sit-to-stand’ and ‘standing’ to accommodate the dependence of moments on joint angles. For ‘standing’, roots/levels have been selected to produce favourable ratios in knee extension to hip flexion (L4L, L3/5R) and to ‘correct’ for unwanted moments (S1L; plantarflexor, hip abductor, weak hip extensor). Moments, as measured in the multi-moment chair during stimulation with our optimum pattern for ‘standing’ while the subject is extended, are shown in figure 1 (he, hip extension; ke, knee extension; pf, plantarflexion; her, hip external rotation; hab, hip abduction; inv, ankle inversion; aab, ankle abduction).

Fig.1; Joint moments while stimulating ‘standing’ roots

(L4/S1L, 2-7s; L3/5R, 9-13s; L4/S1L L3/5R, 15-19s)

However, the stands are not functional because of poor posture: excessive hip flexion (~35^o) and internal rotation, especially on the right, lordosis, plantarflexion (~8^o) and ankle inversion. Significant forces (50% body weight) are taken through the arms [6] and the stands are limited to 2 minutes. Selective stimulation and posture in retraining has improved responses in standing, reducing hip adduction, flexion and ankle inversion, without compromising knee extension, but the improvements have not been sufficient. Pre-LARSI, this subject stood upright by surface stimulation. She also has no determining contractures, attaining upright posture while standing passively in an Oswestry frame and able to lie ‘fully’ prone/supine. The posture is therefore attributed to root stimulation. Co-activation in muscle responses is evident, figure 1, but this does not necessarily demonstrate the determinants or the relative contribution of muscles.

Excessive plantarflexor moments may contribute to poor posture, but in figure 1 these are shown to be small. EMGs measured during stimulation have shown that weight bearing (support reactions) does have an effect on gastrocnemius activity, but these responses are only weak. In addition, though reflex activity has been observed during passive ankle movement, the stiffness remains unchanged. The ankle effects are observed as the subject accepts increased forces through their arms as they try to extend their hips. As a result, they transfer between standing on flatfoot to forefoot, with the ankle inverting as it is inclined to do.

EMG studies have shown that of the roots used in standing, L4L and L5R activate both the hip flexors rectus femoris and iliacus [5]. Non M-wave activity effected by weight bearing has been observed in these muscles, but this is weak and therefore unlikely to be significant. Using local anaesthetic on iliacus did improve the posture slightly, but when applied to rectus femoris resulted in a significant reduction in knee extension, with relatively little change in hip flexion. Measurements of the hip flexion moment in standing [6] and in the multi-moment chair, with and without stimulation, figure 2, show a dependence on hip angle.

Fig.2; Hip extension moments with hip angle

The flexion moment is seen to increase as the hips are extended. Since in standing the hips are flexed at ~35^o, where the hip moments are close to ‘neutral’, large extension moments at the hip would be required to attain ‘full extension’. Muscle co-activation does not allow hip extensor muscles to be stimulated further and use of the hand supports to extend the hips, since these provide a closed-link system, has not been successful.

Discussion

Our data suggests that using medical records, 22% of paraplegics with a traumatic complete lesion may be suitable for FES-standing. However, since 22/50 of those assessed were unsuitable, this figure is probably closer to 12%, similar to published figures of 5-12% [7]. This work has further emphasised the importance of careful subject selection, and the role that pre-op surface stimulation plays to assess compliance and identify unwanted effects in exercising or standing. Progression to implant has been lower than expected, discouraged by the results to date not showing functional standing. In innovative projects, such as LARSI, it is evident that whereas some subjects get involved at an early stage, others will wait until the results have been proven.

Motor root responses have been extensively mapped in two subjects, demonstrating muscle co-activation from single root stimulation. It was postulated that this could be addressed by stimulating combinations of roots, but this has proved difficult and distinct movements have not always been possible. Standing by LARSI has been demonstrated, but significant difficulties have been experienced in attaining a good posture. Reducing hip flexion remains essential to improve the quality and function of standing. Extensive tests suggest that it cannot be corrected by root stimulation alone and it may only be achievable through additional medical or surgical interventions. We therefore propose to inject the psoas muscle with botulinum toxin to reduce hip flexion. This may allow us to reduce the contribution of L5R, hence reducing the contribution from rectus femoris, and compensate for the loss of knee extension by increasing L3R.

Comparing to standing by surface stimulation, it is our opinion that at the present time LARSI has not demonstrated significant improvements in function. We accept that only 2 subjects have been tested, but taken with the low recruitment rate and comments made by the paraplegics the clinical viability of LARSI is questioned. We therefore suggest that root stimulation may not be the best approach to achieve functional standing. However, the improvement in general fitness and muscle integrity of both subjects and the demonstration of recreational cycling [8] suggests further investigation. One approach being considered is to modify LARSI into a multi-modal implant to stimulate nerve roots for muscle retraining, health benefits, tissue viability and bladder/bowel function.

Acknowledgements: The authors would like to thank all the subjects for their time. The project was supported by grants from the Medical Research Council (1992-1997) and the Wellcome Trust (1998-2001).

References

[1] Rushton DN. Choice of nerve roots for multichannel leg controller implant. In: Popovic (ed), Advances in External Control of Human Extremities X. Belgrade, Nauka, pp. 98-108, 1990.

[2] Rushton DN et al. Selecting candidates for a lower limb stimulator implant programme - a patient centred method. Spinal Cord, 36: 303-309, 1998.

[3] Wood DE et al. Experience in using knee angles as part of a closed loop algorithm to control FES assisted paraplegic standing. 6^th Int. Workshop on FES, Vienna, Austria, pp. 137-140, 1998.

[4] Wood DE et al. Apparatus to simultaneously measure fourteen isometric leg joint moments - Part 2: multi-moment chair system. Medical and Biological Engineering and Computing, 37(2): 148-154, 1999.

[5] Rushton DN et al. LARSI: How to obtain favourable muscle contractions? 2nd Ann. Conf. of IFESS, pp. 163-164, Vancouver, Canada, 1997.

[6] Norton JA et al. The determinants of posture in paraplegics standing using LARSI. 5^th Ann. Conf. of IFESS, pp. 478-481, Aalborg, Denmark, 2000.

[7] Jaeger RJ et al. Estimating the user population of a simple electrical stimulation system for paraplegics. Paraplegia, 28: 505-511, 1990.

[8] Perkins TA et al. Development of paraplegic leg powered cycling with LARSI. 4^th Ann. Conf. of IFESS, pp. 139-142, Sendai, Japan, 1999.