Re: 완전척수마비 cpg 반응 자극 ' 2023 Nature

[문형철]

1. nature  
2. nature communications  
3. articles  
4. article

Rare phenomena of central rhythm and pattern generation in a case of complete spinal cord injury

Download PDF

• Article
• Open access
• Published: 06 June 2023

Rare phenomena of central rhythm and pattern generation in a case of complete spinal cord injury

• Karen Minassian, 
• Aymeric Bayart, 
• Peter Lackner, 
• Heinrich Binder, 
• Brigitta Freundl & 
• Ursula S. Hofstoetter 

Nature Communications volume 14, Article number: 3276 (2023) Cite this article

•  
•  
•  

Abstract

Lumbar central pattern generators (CPGs) control the basic rhythm and coordinate muscle activation underlying hindlimb locomotion in quadrupedal mammals. The existence and function of CPGs in humans have remained controversial. Here, we investigated a case of a male individual with complete thoracic spinal cord injury who presented with a rare form of self-sustained rhythmic spinal myoclonus in the legs and rhythmic activities induced by epidural electrical stimulation (EES). Analysis of muscle activation patterns suggested that the myoclonus tapped into spinal circuits that generate muscle spasms, rather than reflecting locomotor CPG activity as previously thought. The EES-induced patterns were fundamentally different in that they included flexor-extensor and left-right alternations, hallmarks of locomotor CPGs, and showed spontaneous errors in rhythmicity. These motor deletions, with preserved cycle frequency and period when rhythmic activity resumed, were previously reported only in animal studies and suggest a separation between rhythm generation and pattern formation. Spinal myoclonus and the EES-induced activity demonstrate that the human lumbar spinal cord contains distinct mechanisms for generating rhythmic multi-muscle patterns.

초록

요추 중앙 패턴 생성기(CPG)는 사족 포유류의 뒷다리 운동에 기초하는 기본 리듬을 제어하고 근육 활성화를 협응합니다. 인간에 CPG의 존재와 기능은 여전히 논란이 되고 있습니다. 여기에서는 완전한 흉부 척수 손상을 입은 남성 환자가 다리에서 드문 형태의 자발적인 리드미컬한 척추 근간대성 경련과 경막외 전기 자극(EES)에 의해 유발된 리드미컬한 활동을 보인 사례를 조사했습니다.

근육 활성화 패턴 분석 결과, 미오클로누스는 척추 회로에서 근육 경련을 생성하는 회로에 접근한 것으로 나타났으며, 이전에 생각되었던 운동 CPG 활동의 반영이 아니었습니다.

EES 유발 패턴은 운동 CPG의 특징인 굴곡-신전 및 좌우 교대 패턴을 포함했으며, 리듬성에서 자발적인 오류를 보여주었습니다. 리듬 활동이 재개될 때 주기 빈도와 기간이 유지된 이 운동 삭제 현상은 이전에 동물 연구에서만 보고되었으며, 리듬 생성と 패턴 형성의 분리 가능성을 시사합니다. 척추성 근육 경련과 EES 유발 활동은 인간 요추 척수 신경에 리듬적 다근육 패턴을 생성하는 독립적인 메커니즘이 존재함을 보여줍니다.

Similar content being viewed by others

Activity-dependent spinal cord neuromodulation rapidly restores trunk and leg motor functions after complete paralysis

Article 07 February 2022

Long-term physical therapy for neuropathic pain after cervical spinal cord injury and resting state electroencephalography: a case report

Article 09 April 2022

Restoring initial steps by intermittent theta burst stimulation in complete spinal cord injury patient: a case report

Article 04 August 2024

Introduction

In the motor control system for hindlimb locomotion in quadrupedal mammals, specialized neural circuits in the lumbar spinal cord represent the final network1,2. In reduced animal preparations, the neural control exerted by these circuits was shown to be necessary and sufficient to generate the basic motor patterns underlying locomotion3,4,5. This central pattern generator (CPG) capability of the lumbar spinal circuits was directly demonstrated in various animal models by the generation of so-called fictive locomotion, i.e., rhythmic motor output recorded from ventral roots or peripheral nerves with flexor-extensor and left-right alternations in the absence of descending neural inputs or peripheral feedback cues6. Fictive locomotion was expressed, inter alia, in the immobilized (neuromuscular block by curarization), acute spinal cat pretreated with L-Dopa by electrical dorsal root or dorsal column stimulation7, in the immobilized decerebrate cat by electrical stimulation of brainstem structures8, or in the isolated neonatal mouse spinal cord through the administration of excitatory amino acids and serotonin9. Such explicit evidence is sparse in non-human primates. Attempts to elicit fictive locomotion by L-Dopa and dorsal column stimulation failed in acutely spinalized and immobilized macaque monkeys10. In adult marmoset monkeys, electrical stimulation of the brainstem in the decerebrate preparation or administration of different pharmacological agents (clonidine, NMDA, serotonin) following spinalization generated various rhythmic activities that presented “component fictive patterns” with alternating as well as synchronous bursts in flexor and extensor nerves11. If combined, the component fictive patterns obtained under different experimental conditions would resemble a true fictive locomotor pattern. Yet, full fictive locomotion was not evoked by any single condition.

In humans, the experimental procedures necessary for the demonstration of fictive locomotion cannot be applied. Indirect evidence of spinally generated rhythmic activity comes from specific motor phenomena in individuals with spinal cord injury (SCI)6. First, spontaneously emerging rhythmic spinal myoclonus at a highly reproducible rate of 0.3–0.6 Hz and involving multiple lower-limb muscles was described in six individuals with chronic, clinically complete SCI12,13,14. The electromyographic patterns largely involved synchronous bursting across muscles. Despite the lack of a locomotor pattern, it was suggested that this type of self-sustained rhythmic activity was due to a partial release of a CPG12,13,14. A common denominator in the individuals expressing this type of spinal myoclonus was the presence of additional musculoskeletal pathologies below the level of the lesion, particularly of the hip12,13,14,15. A second line of indirect evidence comes from induced rhythmic activities in paralyzed leg muscles by tonic epidural electrical stimulation (EES) of the lumbar spinal cord16,17. While the most common electromyographic pattern was synchronous bursting, locomotor-like patterns with a clear reciprocal relationship between antagonistic muscle groups were detected as well18.

Assuming that locomotor CPGs exist in the human spinal cord, fundamental characteristics of their operation observed in animal experiments should also be detected in humans. Notable motor phenomena during otherwise robust fictive locomotion are so-called motor deletions, reflecting spontaneous errors in CPG operation. Motor deletions are an absence of activity in a set of synergistic motor pools during a time period when it would normally occur7,19. This failure to provide rhythmic drive can be accompanied by a failure to inactivate the antagonistic set of motoneurons, resulting in their sustained firing9.

Here, we describe rare motor phenomena all occurring in a male individual who had sustained a thoracic sensory and motor complete SCI (Supplementary Table 1) and a luxation of the left hip joint in a traffic accident. The subject presented with severe, medication-resistant muscle spasms and was referred to a clinical program for the treatment of lower-limb spasticity by EES20. The clinical assessment battery included a standardized neurological examination documented by poly-electromyographic recordings21. The same examination was conducted multiple times over a period of three months to assess the manifestations of the subject’s spasticity and, after the implantation with an epidural lead, to assess the effects of stimulation. The data analyzed and reported here were taken from these examinations. We observed muscle spasms, self-sustained spinal myoclonus, EES-induced rhythmic activity, and spontaneous motor deletions, all in the same subject. Based on the ability to directly compare these motor phenomena, we propose that the generation of spinal myoclonus is closely linked to the circuits underlying muscle spasms. Following the logic from animal studies9,11,22, the component locomotor patterns evoked by EES support the activation of the CPG, and the specific type of motor deletions detected here indicates a separation of rhythm generation and pattern formation in the human lumbar spinal cord with a flexor-dominant operation.

서론

사지 운동을 위한 후지 운동 제어 시스템에서 사족 포유류의 요추 척수 내 특수 신경 회로는 최종 네트워크를 구성합니다1,2. 축소 동물 모델에서 이러한 회로가 발휘하는 신경 제어는 운동의 기본 운동 패턴을 생성하는 데 필요하고 충분하다는 것이 입증되었습니다3,4,5. 요추 척수 회로의 이 중앙 패턴 발생기(CPG) 기능은 다양한 동물 모델에서 하강 신경 입력이나 말초 피드백 신호가 없는 상태에서 복부 신경근이나 말초 신경에서 기록된 굴곡-신전 및 좌우 교대 리듬 운동 출력인 이른바 가상의 보행 생성으로 직접 입증되었습니다6. 가상 운동은 예를 들어, 신경근 마비로 고정된(curarization에 의한 neuromuscular block) 급성 척수 절제 고양이에서 L-Dopa 사전 투여 후 전기적 등쪽 뿌리 또는 등쪽 기둥 자극을 통해7, 뇌간 구조의 전기적 자극을 받은 고정된 뇌간 절제 고양이에서8, 또는 신생아 쥐의 분리된 척수에서 흥분성 아미노산과 세로토닌 투여를 통해9 표현되었습니다. 인간을 제외한 영장류에서 이러한 명확한 증거는 드뭅니다. L-Dopa와 등쪽 기둥 자극을 통해 허구적 운동을 유발하려는 시도는 급성 척수 절제 및 고정된 마카크 원숭이에서 실패했습니다10. 성체 마모셋 원숭이에서 뇌간 전기 자극(뇌간 절제 준비 상태) 또는 척수 절제 후 다양한 약리학적 제제(클로니딘, NMDA, 세로토닌) 투여는 굴곡 및 신전 신경에서 교대 및 동기화된 발작을 보이는 “구성 요소 허구 패턴”을 포함한 다양한 리듬 활동을 생성했습니다11. 서로 다른 실험 조건 하에서 얻어진 구성 요소 허구 패턴을 결합하면 진정한 허구 운동 패턴과 유사한 모습을 보였습니다. 그러나 단일 조건으로는 완전한 허상 운동이 유발되지 않았습니다.

인간에서는 허상 운동을 입증하기 위한 실험적 절차가 적용될 수 없습니다. 척수 손상(SCI) 환자에서 척수에서 생성된 리듬 활동의 간접적 증거는 특정 운동 현상에서 관찰됩니다6. 첫째, 만성적이며 임상적으로 완전한 SCI를 가진 6명의 개인에서 0.3–0.6 Hz의 높은 재현성으로 발생하는 다중 하체 근육을 포함한 자발적 척추 리듬성 근경련이 보고되었습니다12,13,14. 전기근전도 패턴은 주로 근육 간 동기화된 폭발을 포함했습니다. 운동 패턴이 결여되었음에도 불구하고, 이 유형의 자체 유지 리듬 활동은 CPG의 부분적 해방에 기인한다는 제안이 제기되었습니다12,13,14. 이 유형의 척추 근육 경련을 나타내는 개인들의 공통된 특징은 손상 부위 아래에 추가적인 근골격계 병리, 특히 골반 부위의 병리가 존재한다는 점이었습니다12,13,14,15. 두 번째 간접적 증거는 요추 척수16,17에 대한 지속적 경막외 전기 자극(EES)으로 마비된 다리 근육에서 유도된 리듬 활동에서 비롯됩니다. 가장 흔한 전기근전도 패턴은 동기화된 발작이었지만, 대항 근육 그룹 간 명확한 상호 관계가 있는 운동 유사 패턴도 관찰되었습니다18.

인간 척수에서 운동 CPG가 존재한다고 가정할 때, 동물 실험에서 관찰된 그 작동의 기본적 특성은 인간에서도 관찰되어야 합니다. 강력한 가상의 운동 중 눈에 띄는 운동 현상은 CPG 작동의 자발적 오류 반영인 소위 운동 삭제입니다. 운동 삭제는 일반적으로 발생해야 할 시간 동안 시너지 운동 풀의 활동 부재를 의미합니다7,19. 리듬적 구동 제공의 실패는 대항 신경세포 집단의 비활성화 실패와 동반될 수 있으며, 이는 그들의 지속적 발화로 이어집니다9.

여기서는 교통사고로 인해 흉부 감각 및 운동 완전 척수 손상(SCI)을 입은 남성 환자(보충 표 1)에서 발생한 드문 운동 현상을 설명합니다. 환자는 약물 치료에 저항하는 심각한 근육 경련을 보였으며, EES20을 통한 하체 경련 치료를 위한 임상 프로그램에 소개되었습니다. 임상 평가 배터리에는 표준화된 신경학적 검사가 다중 근전도 기록으로 문서화되었습니다21. 동일한 검사는 3개월 동안 여러 번 반복되어 환자의 경련 증상을 평가했으며, 경막외 전극 삽입 후 자극의 효과를 평가하기 위해 추가로 실시되었습니다. 본 연구에서 분석 및 보고된 데이터는 이러한 검사에서 수집되었습니다. 우리는 동일한 대상에서 근육 경련, 자발적 척추 근육 경련, EES 유발 리듬 활동, 자발적 운동 삭제 현상을 관찰했습니다. 이러한 운동 현상을 직접 비교할 수 있다는 점을 바탕으로, 척추 근육 경련의 발생이 근육 경련을 조절하는 회로와 밀접하게 연관되어 있음을 제안합니다. 동물 연구의 논리9,11,22에 따라, EES로 유발된 운동 패턴의 구성 요소는 CPG의 활성화를 지원하며, 여기서 검출된 특정 유형의 운동 삭제는 인간 요추 척수에서 리듬 생성 및 패턴 형성의 분리 및 굴곡 근육 우세 작동을 나타냅니다.

Results

Self-sustained rhythmic spinal myoclonus in the paralyzed lower limbs

We observed an intriguing pattern of self-sustained rhythmic electromyographic activity in lower-limb muscles within six neurological examinations (Fig. 1). The rhythmic activities either followed brief manipulations of the lower limbs by the examiner to assess the subject’s spasticity, or occurred spontaneously between assessment segments while the subject was lying relaxed with the legs extended. This phenomenon clearly differed from ankle clonus, the only common self-sustained rhythmic muscle activity seen after chronic SCI (Supplementary Figs. 1 and 2). We identified eleven examples of such self-sustained rhythmic activities (Supplementary Table 2) that lasted for a minimum of 10 s (median duration: 18.1 s, interquartile range (IQR): 13.1–25.2 s). The longest episode had a duration of 63 s, only halted through repositioning of the lower limbs by the examiner.

Fig. 1: Self-sustained rhythmic patterns of electromyographic activity in paralyzed lower limbs recorded in the supine position.

a Spontaneous rhythmic activity that occurred after a period of 3.9 min of continuous electromyographic (EMG) recordings during which no manipulations of the lower limbs were performed by the examiner, example 9 (cf. Supplementary Table 2). b Rhythmic activity that occurred immediately following cutaneous-input evoked spasms (not shown) by right plantar stimulation with a blunt rod, example 4. c Rhythmic activity that occurred immediately following spasms (not shown) evoked by one cycle of slow passive hip and knee flexion-extension movement of the right lower limb, example 2. The EMG activities of the rhythmic phenomena are shown for the total durations they had lasted with the same time and EMG amplitude scaling in (a), (b), and (c). Body image adapted from Ipsi- and Contralateral Oligo- and Polysynaptic Reflexes in Humans Revealed by Low-Frequency Epidural Electrical Stimulation of the Lumbar Spinal Cord, Hofstoetter US, Danner SM, Freundl B, Binder H, Lackner P, Minassian K, Brain Sciences, 11, 112, 202153, MDPI, © 2021 by the authors.

Full size image

The rhythmic activities involved muscles across hip, knee, and ankle joints and both legs (cf. Supplementary Table 2). Within a given example, rhythm-cycle frequencies did not differ between muscles (Supplementary Table 3). Across examples and muscles, the mean rhythm-cycle frequency amounted to 0.35 ± 0.01 Hz, ranging from 0.18 ± 0.01 Hz to 0.50 ± 0.01 Hz. No significant interaction effect on the rhythm-cycle frequency existed between example and muscle, F(78;461) = 0.371, p = 1.000, ηp2 = 0.059. Muscle was not a significant factor for rhythm-cycle frequency, F(9;461) = 0.987, p = 0.450, ηp2 = 0.019, but example was, F(10;461) = 205.472, p < 0.001, ηp2 = 0.817.

Despite the examination days spanning a period of three months, the different geneses of the rhythmic activities, and the various rhythm-cycle frequencies, the sequence of muscle recruitment was highly robust. The electromyographic bursts appeared with a modest, yet consistent onset lag between muscles (Fig. 2a). Statistical analysis revealed no differences in the sequence of muscle recruitment between examples. No significant interaction existed between example and muscle, F(65;1) = 23.743, p = 0.162, ηp2 = 0.999. Example was not a significant factor for the sequence of muscle recruitment, F(9;1) = 95.185, p = 0.079, ηp2 = 0.999, but muscle was, F(8;1) = 1112.986, p = 0.023, ηp2 = 1.000. Across examples, activity first occurred in the left lower leg muscles, corresponding to activity in the L4, L5 (tibialis anterior) and S1, S2 (triceps surae) spinal cord segments23, followed by a spread to bilateral muscles and a complex migration of activity along the lumbar and upper sacral spinal cord (Fig. 2b).

Fig. 2: The self-sustained rhythmic activities exhibit highly robust spatiotemporal patterns of electromyographic activity in the lower limbs.

The consistent occurrence of rhythmic activities in the left rectus femoris (RF) in ten of the eleven examples (cf. Supplementary Table 2) allowed the analysis of the sequence of muscle recruitment through the relative onset lags of electromyographic (EMG) bursts. a Envelopes of the EMG activities per muscle averaged across the available rhythm cycles of examples 2, 4, and 9 (cf. Fig. 1, Supplementary Table 2), and aligned with respect to the onset of the EMG burst in the left RF (‘0’ on the x-axis). The EMG burst onsets are marked by vertical lines with the sequence given by the example numbers. b (i) Relative onset lags of rhythmic bursts occurring in the various muscles. Gray lines, individual results of the ten examples; blue line, mean values (±SE) across examples. (ii) Time course of progressive spread of the self-sustained rhythmic activity along the lumbosacral spinal cord segments based on the segmental innervations of the recruited muscles23. AD adductors, H hamstrings muscle group, L lumbar, S sacral, TA tibialis anterior, TS triceps surae muscle group. Source data are provided in the Source Data file.

Full size image

We identified these observed phenomena as a very rare form of spinal myoclonus, documented in only six other individuals with complete SCI in the literature12,13,14, based on the co-activation patterns, the range of rhythm-cycle frequencies, and the possible role of the subject’s hip pathology12,13,15 (Supplementary Fig. 3).

Muscle activation patterns do not differ between spinal myoclonus and muscle spasms

In seven examples, spinal myoclonus followed lower-limb manipulations by the examiner that elicited either seconds-long proprioceptive activation (through a single cycle of passive hip-and-knee or ankle flexion-extension movement) or cutaneous activation from the foot sole (cf. Supplementary Table 2). No examples were found following a tendon tap or a brisk manual dorsiflexion to elicit an ankle clonus (cf. Supplementary Figs. 1 and 2), suggesting that spinal myoclonus was not readily evoked by brief proprioceptive volleys to the spinal cord.

The passive movements as well as plantar stimulation always elicited muscle spasms24,25 (Supplementary Fig. 1) that had consistent electromyographic patterns when evoked repeatedly during the same neurological examination (Supplementary Fig. 4), irrespective of whether a myoclonus followed or not (Fig. 3a). The muscle spasms were characterized by a spread to multiple ipsilateral and contralateral muscles, resulting in complex muscle activation patterns outlasting the duration of the manipulation. Thus, the initially recruited spinal networks were always those underlying the generation of muscle spasms26,27. Spinal myoclonus evolved from these muscle spasms after a silent period or a period of tonic activity (Fig. 3a(i), b) and never involved muscles other than those already recruited into the spasms. Noticeably, the electromyographic patterns of the myoclonus bursts closely resembled that of the spasms, indicative of common pattern generating networks (Fig. 3c). Indeed, muscle activation patterns did not differ statistically between muscle spasms and spinal myoclonus bursts; no significant interaction effect existed between the type of activity and muscle on the respective integrated electromyographic activities, F(9;120) = 1.097, p = 0.370, ηp2 = 0.076.

Fig. 3: Spinal myoclonus patterns resemble repeatedly triggered muscle spasms.

a Muscle spasms evoked by imposed flexion-extension movement of the right lower limb (POS, knee position). The electromyographic (EMG) patterns of these muscle spasms (magenta backgrounds in (i) and (ii)) were consistent when the manipulation was repeated (see also Supplementary Fig. 4). b Spinal myoclonus evolved from the muscle spasms in response to the manipulations by the examiner. The multi-muscle EMG patterns of the myoclonus bursts (blue backgrounds) show resemblance to those of the immediately evoked spasms (magenta background); (i), example 1 (cf. Supplementary Table 2); (ii), example 5. c Polar plots show muscle activation patterns of spasms (magenta lines) and myoclonus bursts (blue lines) of all seven examples of spinal myoclonus that followed lower-limb manipulations by the examiner. Radial axes are muscles and polar coordinates are mean integrated EMG activities. Muscle activation patterns of muscle spasms and spinal myoclonus were not statistically different in any of the available examples. AD adductors, H hamstrings muscle group, L left; RF rectus femoris, R right, rep. repetition, TA tibialis anterior, TS triceps surae muscle group. Source data are provided in the Source Data file.

Full size image

The rhythmic nature of spinal myoclonus

The major difference between spinal myoclonus and muscle spasms, i.e., its rhythmic nature, compelled us to propose that spinal myoclonus bursts might be repeatedly triggered muscle spasms. The rhythmic nature of spinal myoclonus would require a sustained excitatory drive as well as self-limiting mechanisms in spasm generation. The silent periods in-between consecutive bursts could have been caused by active inhibitory mechanisms curtailing an ongoing burst and delaying the onset of the succeeding burst27 or could have been related to refractoriness or fatigue following each bursting event28. If the latter were true, the durations of silent periods should increase with the burst durations and magnitudes. Spontaneous cycle-to-cycle variations within the myoclonus episodes allowed us to study whether such relationships existed (Fig. 4a). Throughout the 94 cases of rhythmically active muscles of the eleven examples, the burst duration was not statistically correlated with the interburst duration in 89.4% (Fig. 4b(i)). The integral of the electromyographic activity of a burst was not correlated with the interburst duration in 87.3% (Fig. 4b(ii)).

Fig. 4: The repetitive nature of spinal myoclonus is not due to neural circuit refractoriness.

a (i) Exemplary recording of rhythmic electromyographic (EMG) activity of the left (L) rectus femoris (RF) indicates cycle-to-cycle variations in interburst durations (red backgrounds and horizontal bars) and burst durations (blue backgrounds), example 4 (cf. Fig. 1b, Supplementary Table 2). In this example, there was no correlation between (ii) the burst durations and the interburst durations, Pearson’s r = −0.091, p = 0.864; and (iii) the integrated EMG activity of the burst and interburst durations, r = −0.092, p = 0.863. b The relationships between (i) burst durations and interburst durations and (ii) integrated EMG activity of bursts and interburst durations were analyzed by Pearson’s product moment correlations, separately for the 94 available cases of rhythmically active muscles of the eleven spinal myoclonus examples (cf. Supplementary Table 2). White squares indicate no significant correlations, blue squares significant positive correlations, and gray squares significant negative correlations (all p < 0.05); crosses are non-rhythmically active or inactive muscles (not included in the analyses). AD adductors, H hamstrings muscle group, TA tibialis anterior, TS triceps surae muscle group. All statistical tests were two-sided.

Full size image

The excitatory drive promoting the occurrence of spinal myoclonus as well as sustaining its rhythmic behavior could have been provided by the subject’s hip pathology, as suggested previously13,14,15. Such facilitatory effect of a tonic excitatory drive was supported by observations under ongoing EES with intensities subthreshold to evoke muscle activity (Fig. 5a). Here, each repetition of passive left hip-and-knee flexion-extension movements induced spasms that evolved into spinal myoclonus. Notably, tonic stimulation with incremental, yet subthreshold intensities did not change the muscle activation pattern of spinal myoclonus (Fig. 5b), but rather advanced the onset of the first burst27 (Fig. 5c).