Co-Activation of ankle agonist/antagonist muscle pairs during stepping is not related to the detectable level of voluntary function after human spinal cord injury



... Abstract

J.A. Beres-Jones1, R. Joaquin1, R.U. Lukacs2 and S.J. Harkema1,2
1Department of Neurology, 2Brain Research Institute, University of California at
Los Angeles, Los Angeles, CA, USA

In many people with spinal cord injury (SCI), the tibialis anterior (TA) muscle is co-activated with the soleus (SOL) and medial gastrocnemius (MG) muscles during stepping, unlike what generally occurs in non-disabled individuals where activity in the ankle flexor alternates with activity in the ankle extensors. This co-activation of ankle agonist/antagonist muscle pairs has been attributed to loss of supraspinal influence. We tested whether there was a relationship between detectable level of voluntary function and the ankle EMG patterns in the SOL, MG and TA during stepping. We found that the SOL, MG and TA were co-activated during stepping in approximately 63% of subjects with clinically complete SCI and 67% of subjects with clinically incomplete SCI. Further, 50% of subjects with sensory but not motor function had co-active SOL, MG and TA. In subjects with some voluntary motor control, the activity generated during voluntary non-stepping oscillatory movements of the ankle differed in pattern to that generated during stepping. These data suggest that the co-contraction of ankle agonist/antagonist muscle pairs was not related to the detectable level of voluntary function after human spinal cord injury.
Supported by: NS36854, NS16333, RR00865


... Introduction

In many people with spinal cord injury (SCI) {Harkema1997}, and other neurologic injuries {knutsson1979}{Unnithan1996}, the tibialis anterior (TA) muscle is co-activated with the soleus (SOL) and medial gastrocnemius (MG) muscles during stepping, unlike what generally occurs in non-disabled individuals where activity in the ankle flexor alternates with activity in the ankle extensors at normal walking speeds {Shiavi1987}. This co-activation of ankle agonist/antagonist muscle pairs has been attributed to loss of supraspinal influence {Chen2001}. The rationale is that a primary source of inhibition of antagonistic muscles during activation of agonists occurs via reciprocal inhibition through Ia interneurons {Crone1994}. Further, regulation of these inputs occurs via supraspinal influences {Jankowska1976}{Nielson1993}.

We tested the hypothesis that co-activation among the SOL, MG and TA muscles results from a loss of supraspinal input. We tested whether there was a relationship between detectable level of voluntary function and the incidence of co-activation of the SOL/TA and MG/TA muscle pairs. If co-activation of ankle agonist/antagonist muscle pairs occurs as a result of a loss of supraspinal input, then co-activation among antagonistic muscle pairs should be seen in all subjects with clinically complete SCI, and co-activation among antagonistic muscle pairs should appear less frequently as the severity of SCI decreases. If both supraspinal regulation and Ia inhibitory Interneurons contribute to muscle recruitment patterns during locomotion, then perhaps locomotor training may modulate the Ia inhibitory system, allowing it to compensate for a loss of supraspinal influence, thereby ensuring appropriate activation patterns during stepping. GABA-ergic inhibitory systems may also be involved {Nielson2002}. These results provide further insights into the neural control of human locomotion at the level of the human spinal cord.


... Methods

We studied ## SCI subjects (Table 1). Subjects were graded according to the American Spinal Injury Association (ASIA) impairment scale {Ditunno1994}{Maynard1997}, and also received a sensory evoked potential (SEP) test. Subjects with an absent SEP and an ASIA score A (no sensory or voluntary motor function below the injury level) were considered as having a clinically complete SCI. Subjects with ASIA scores B (some sensory preservation, but no voluntary motor function below the injury level), C (some sensory and voluntary motor function, but no active movement against gravity, below the injury level) and D (some sensory and motor function, with active movement against gravity) were considered as having a clinically incomplete SCI. Further, subjects with ASIA scores A and B were considered motor-complete, while subjects with ASIA scores C and D were considered motor-incomplete.


MBWL: mean body weight load. ASAI lower limb motor score out of a possible score of 25 for each limb. Step cycle duration is mean step cycle duration (S.C. dur) plus or minus standard deviation (sd) in seconds(s). Injury level: T, thoracic; C, cervical. (--): data not available.

Each subject wore a harness connected to an overhead lift that allowed adjustment of lower limb loading. During stepping using body weight support on a treadmill (BWST) trainers provided manual assistance at the lower limbs as needed (Fig. 1). Hand placement distal to the patella facilitated extension during the stance phase, and hand placement at the popliteal fossa and proximal to the ankle facilitated foot clearance during the swing phase. A trainer positioned behind the subject aided in hip stabilization and weight shifting during stepping as needed. Non-disabled subjects stepped independently at full body weight load (BWL). SCI subjects stepped with manual assistance at the maximum load at which knee extension could be maintained during stance, which was expressed as a percentage of their BWL. Each subject stepped at a comfortable treadmill speed, which was expressed as step cycle duration.

We recorded EMG (Konigsberg) from the SOL, MG and TA muscles using bipolar surface electrodes, joint kinematics using electromagnetic sensors (SKILL technologies), and individual limb vertical ground reaction forces using insole pressure sensors (PEDAR or FSCAN). We acquired data during stepping trials as well as during attempts at voluntary oscillatory joint movements with gravity eliminated.

We quantified muscle Co-Activity among the SOL, MG and TA using a novel method that considers periods in which the muscle pairs are BOTH co-active AND co-inactive. The methodological approach utilizes an automatic and objective detection algorithm based on variance in the EMG signal (6). Briefly, points of variance change were detected in the EMG signal and the signal to noise ratio (SNR) between these ‘change-points’ was calculated (Fig. 2). Periods of muscle activation were determined using physiologically appropriate event thresholds (SNR > 2.5; activity duration > 29 ms; inactivity duration > 38 ms).


FIGURE 2. SOL, MG and TA EMG activity (A-C), signal to noise ratio (SNR, D-F), and posterior probability (G-H) during two steps from a ND subject. Bayesian change-point method via Reversible Jump Markov Chain Monte Carlo simulations identified points of change in the variation of the EMG signal. Mean SNR (dark line) and 0.025 and 0.975 quantiles (light green) of the EMG signal for each muscle from the variance of the EMG signal between each change-point are plotted (D-F). Estimated posterior probability from thresholds (SNR > 2.5; activity duration > 29 ms; inactivity duration > 38 ms) applied to the SNR to determine periods of EMG activity and inactivity (G-I).

We defined co-activity as the proportion of time that two muscles are simultaneously active AND inactive active:
CO-ACTIVITY = {(ACTIVEboth+INACTIVEboth)}/total time

We defined co-excitation as the proportion of time that two muscles are simultaneously active :
CO-EXCITATION = ACTIVEboth/ACTIVEeither.

We defined co-inhibition as the proportion of time that two muscles are simultaneously inactive:
CO-INHIBITION = INACTIVEboth/INACTIVEeither.





Figure 4 shows the co-activation value posterior density distributions for the SOL/MG, SOL/TA and MG/TA muscle pairs from a non-disabled individual. These estimates are smoothed histograms based on the 2,000 samples.


FIGURE 3. Computation of coordintation between the SOL and MG (top) and the SOL and TA (bottom). Time segments marked by: A=both muscles simultaneously active; B=either is active; C=both muscles are simultatneously inactive; D=either is inactive.




... Results


ID: identification; L/R: Left or Right; ACT: mean Co-Activation value; EXC: mean Co-Excitation value; INH: mean Co-Inhibition value; SD: standard deviation

Co-contraction AND alternation of ankle dorsi- and plantar flexors can occur in clinically complete and clinically incomplete SCI subjects during stepping



FIGURE 5. SOL, MG and TA EMG (mV) activity; hip, knee and ankle angle (o); and vertical ground reaction force (N) during 2 steps using BWST in clinically complete (ASIA A; left) and clinically incomplete (ASIA D; right) SCI subjects.

SOL, MG and TA EMG amplitude and co-activation differ during stepping and during similar voluntarily attempted ankle movements




... Conclusion

The level of voluntary control of the lower limbs is not correlated with the level of co-activation of the MG/TA, SOL/TA or SOL/MG.
1. No significant trend in co-activation values with ASIA lower limb motor score for any muscle pair (slopes all less than 0.1, r2 values all less than 1)
2. No significant trend in co-excitation values with ASIA lower limb motor score for any muscle pair (slopes all less than 0.1, r2 values all less than 1)
3. No significant trend in co-inhibition values with ASIA lower limb motor score for any muscle pair (slopes all less than 0.1, r2 values all less than 1)

Further, the muscle activation pattern observed during voluntary ankle oscillation differs from that observed during stepping. However, many studies examining co-activation of the ankle muscles measure parameters such as H-reflex and reciprocal Ia inhibition during isolated voluntary ankle movements and extrapolate the results to activation patterns seen during stepping.

However, while inhibition derived from descending pathways of the corticospinal tract is not related to co-activation of the MG/TA, SOL/TA or SOL/MG, this study does not preclude that activity in other descending pathways (e.g. rubrospinal pathways) may be involved in the modulation of EMG activity during stepping.

Further, we have previously shown that body weight loading during stepping, stepping velocity and locomotor training can impact lower limb EMG activity during stepping after human spinal cord injury. It is likely that these factors can also influence muscle co-activation patterns during stepping. Studies examining these effects would be beneficial in further understanding the neural control of human locomotion as well as aiding the enhancement of rehabilitation techniques aimed at recovery of walking.


... References

1. Chen D, Theiss RD, Ebersole K, Miller JF, Rymer WZ, Heckman CJ. Spinal interneurons that recieve input from muscle afferents are differentially modulated by dorsolateral descending systems. The American Physiological Society: 1005-1008, 2001.

2. Crone C, Nielsen J. Central control of disynaptic reciprocal inhibition. Acta Physiol Scand 152: 351-363, 1994.

3. Ditunno JF, Young W, Donovan WH, Creasey G. The international standards booklet for neurological and functional classification of spinal cord injury. Paraplegia 32: 70-80, 1994.

4. Harkema SJ, Hurley SL, Patel UK, Requejo PS, Dobkin BH, Edgerton VR. Human lumbosacral spinal cord interprets loading during stepping. J Neurophysiol 77: 797-811, 1997.

5. Jankowska E, Padel Y, Tanaka R. Disynaptic inhibition of spinal motoneurons from the motor cortex in the monkey. J Physiol 258: 467-487, 1976.

6. Johnson TD, Elashoff RM, Harkema SJ. A bayesian change-point analysis of electromyographic data: detecting muscle activation patterns and associated application. Journal of Biostatistics:2002.

7. Knutsson E, Richards CL. Different types of disturbed motor control in gait of hemiparetic patients. Brain 102: 405-430, 1979.

8. Maynard FM, Bracken MB, Creasey G, Ditunno JF, Donovan WH, Ducker TB, Garber SL, Marino RJ, Stover SL, Tator CH, Waters RL, Wilberger JE, Young W. International standards for neurological and functional classification of spinal cord injury. Spinal Cord 35: 266-274, 1997.

9. Nielsen J, Petersen NE, Ballegaard M, Biering-Sorensen F, Kiehn O. H-reflexes are less depressed following muscle stretch in spastic spinal cord injured patients than in healthy subjects. Exp Brain Res 97: 173-176, 1993.

10. Shiavi R, Bugle HJ, Limbird T. Electromyographic gait assessment, part 1: Adult EMG profiles and walking speed. Journal of Rehabilitation Research and Development 24: 13-23, 1987.

11. Unnithan VB, Dowling JJ, Frost B, Volpe BA, Bar-Or O. Cocontraction and phasic activity during GAIT in children with cerebral palsy. Electromyogr Clin Neurophysiol 36: 487-494, 1996.


... Acknowledgements

We would like to thank the subjects for their valuable contribution to this study, as well as Christie Ferreira and the members of the Human Locomotion Research Center for their contributions to this poster. We also acknowledge the collaboration of Dr. Anton Wernig, Sabina Muller and the supporting personnel of Rehabilitation Clinics, Langensteinbach, Karlsbad, Germany, for locomotor training and providing all general daily care and housing for 7 of the participants in this study.

This work was supported by NIH grants NS36854, NS16333, RR00865.