clonus: role of mechanisms - jnnp.bmj.com · clonus: the role ofcentral mechanisms ... three kinds...

Journal of Neurology, Neurosurgery, and Psychiatry, 1980, 43, 321-332

Clonus: the role of central mechanismsM R DIMITRIJEVIC, P W NATHAN,* AND A M SHERWOOD

From the Institute for Clinical Neurophysiology, School of Medicine, Ljubljana, Yugoslavia; The NationalHospitalfor Nervous Diseases, Queen Square, London, England; and Department of Clinical Neurophysiology,Texas Institute.for Rehabilitation and Research, Baylor College of Medicine, Houston, Texas, USA

SUMMARY The peripheral and central components in sustained clonus were investigated.The excitability of the motoneurons responding to maintained stretch by clonus was examined bytendon taps, trains of vibratory stimuli and by H-reflex afferent volleys. Every burst of clonicdischarge of the motoneurons was shown to be followed by a refractory period, which was followedby a shorter excitatory period. It was concluded that the motoneurons responding clonically to a

continuous stretch cannot respond until their excitability has been regained after the refractoryperiod. Attempts to change the rate of clonus in various ways failed to do so. Whether motoneuronsof clonic muscles tend to respond maximally to other la volleys at the rate of clonus was examinedby applying repeated taps to the tendon at rates from 1 to 15 Hz. There was a maximal responseat the rate of clonus. Inputs other than those induced by stretch cause clonus; examples of cutaneou3inputs causing it are given.

The following facts are known about clonus. Theusual rates are 5 to 8 Hz; in any muscle in onepatient, the rate tends to remain the same. Likeother reflexes induced by stretch, clonus occurswith an appropriate degree of increased muscletone. When a muscle is exceedingly hypertonic,it does not occur. Clonus is usually present inneurological disorders with abnormally increasedtendon jerks, although one sees patients withonly slightly increased jerks and very easilyobtained clonus. Any mechanism or pharma-cological substance that suppresses the excessivejerks and muscle hypertonia tends also tosuppress clonus. Clonus arises only when thereis a lesion involving the great majority of fibresof the lateral corticospinal tract. This statementis based on personal observations of a large seriesof patients having operations on the spinal cordwith histological examination of the centralnervous system (to be published).One important question is whether clonus is

due to a repeated activation of muscle stretchreceptors or depends also on a spinal organisation

Addresses for reprint requests: Dr M R Dimitrijevic. TIRR. TexasMedical Center, Houston, Texas 77030, USA and DrPW Nathan, The National Hospital for Nervous Diseases,Queen Square, London WCIN 3BG.

Accepted 24 October 1979

* Member of the external scientific staff, Medical Research Council,Great Britain.

that includes the repetitively activated stretchreflex. The evidence presented here favours theview that the underlying control mechanism ofsustained clonus is a central governor ofrhythmicity, activated and maintained by Iasynchronised volleys.Two groups of workers studied clonus in man

by recording the activity of single spindle afferentsin peripheral nerves by means of fine electrodesput into the nerves.' 2 In Szumski et al's work,'the clonus consisted of a few beats of clonusfollowing a tendon tap in a flexor of the wrist.This unsustained clonus could be prolonged byJendrassik's manoeuvre. They concluded that inclonus the spindles were abnormally sensitive andthat the motoneurons particularly important formanifesting clonus were the dynamic fusimotorneurons. On the other hand, Hagbarth et al2 whostudied sustained clonus, found no hyperactivityof the spindles nor were they respondingexcessively to stretch. They showed that in clonusthere was no alpha-gamma co-activation; therewere no impulses coming in from the spindlesduring the contraction phase of the muscle; theyoccurred only during the muscle's relaxation.

Sustained clonus can occur in neurologicallyhealthy people, as has been shown in the workof Gottlieb and Agarwal. It can be facilitatedby pharmacological substances that increasespindle discharge in stretched muscles. Struppleret a14 obtained it by injecting succinylcholine

321

Protected by copyright.

on August 29, 2019 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.43.4.321 on 1 April 1980. D

ownloaded from

http://jnnp.bmj.com/

M R Dimitrijevic, P W Nathan, and A M Sherwood

intravenously and Marsden, Meadows andHodgson5 obtained it by injecting adrenalin intra-arterially. They also showed that it could beobtained in normal subjects by vibrating thebelly and the tendon of the triceps surae"particularly if the evoked tension was great".The emphasis in recent studies has been on the

investigation of the role of the peripheralmechanisms of the stretch reflex. The exceptionwas Walsh6 who showed the importance ofcentral components of the reflex in man. Thepurpose of the present investigation was toelucidate the mechanisms of sustained clonus andto find out the roles of peripheral and centralcomponents contributing to the discharge ofmotoneurons in clonus.

Materials

This study was carried out on eight healthysubjects and 18 patients. Eleven patients hadtraumatic division of the spinal cord at variouslevels, some clinically complete, others partial,and seven had hemiplegia due to vascular lesionsin the internal capsule. The paraplegic patientswith so-called clinical complete lesions could notmove any part of the body below the lesion andthere was no sensibility and no thermo-regulatorysweating below the lesion. It should beemphasised that in some of these patients withcomplete lesions as judged on clinical criteria,physiological investigation has shown evidence ofthe influence of the brain on segmental reflexes.This has been described elsewhere.7

Methods

The majority of experiments were done on thetriceps surae; some were also done on thequadriceps, tibialis aliterior, adductors, andbiceps femoris and semitendinosus muscles.Clonus was usually ev6ked in the manner inwhich it is done in clinical neurology. The footor patella was pulled up or down and a constantforce was maintained. In some experiments thelimb was placed with its lateral side downwardson a board, powdered with boric acid to reducefriction between it and the board. At other timesit was on its anterior surface over a couch, withthe foot protruding beyond the end of the couch.

In one group of experiments, attempts weremade to slow the rate of clonus in four ways.One way was to reduce the temperature of thelimb by leaving it in ice-water. Another way wasto put viscous damping on the movements of thefoot by putting it in a stiff jelly, 5% w/v Carbopol

934. The viscosity of this solution was Brookfield16,0-0 (cps) 20 rpm. The aim was to put dragon the foot and thus slow down the movement.The third way was to add a mass to the footperforming clonus. The fourth way was for oneexperimenter to induce clonus while the otheropposed the movement by pulling the foot inthe opposite direction, thus adding force to theantagonists of the movement.

Clonus was recorded by electromyography(EMG), by electrogoniometry, and by a pressuretransducer between the foot and a rigid support-ing surface. Tendon taps were performed bymeans of an ADI vibrator, model AV-6; it wasapplied at various rates up to 14 Hz. It was alsoused to apply vibration to tendons at rates upto 200 Hz. The head displacement, driven by anelectromagnetic coil, was up to 3 mm, dependingon the force on the head.For nerve stimulation, electrical stimuli, in the

form of constant voltage square waves of variableamplitude and pulse-width, were applied throughdamp, cotton-covered stainless steel electrodes orthrough Beckman silver-silver chloride electrodes.For H-reflex responses, electrodes were placedover the tibial nerve in the popliteal fossa. Asquare wave stimulus of 1 ms duration was used,to favour afferent rather than motor fibres; thestimulus strength was also adjusted to obtain anH-wave and avoid an M-wave. For electrlicnoxious stimuli, 20 ms trains of 0-2 ms voltagepulses repeated at 2000 Hz. were applied viaCopland-Davies clip-on electrodes.The movement of the muscle was recorded by

several methods. The joint angle was measuredby an electrogoniometer. Another method ofrecording the movement of the foot utilised anair bladder connected to a Statham pressuretransducer; this was connected to a signal con-ditioning device so that this pressure could berecorded with other mechanical and electricalevents. The air bladder was between theexaminer's hand and the plantar surface of thepatient's foot. Recording electrodes were silver-silver chloride surface electrodes 1 cm diameter,4 cm apart.

For computation of the average EMGamplitude, a voltage-to-frequency convertertechnique was used to digitalise the EMG signalafter full-wave rectification. Digital integrationfor an appropriate response interval (usually20 ms) was followed by averaging more than 100responses. The resultant data points thuscomputed for each set were normalised to amaximum set value of 100.

Tn one series of experiments, the EMG was

322



http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


Clonus: the role of central mechanisms

displayed on an oscilloscope by triggering thesweep by the onset of the clonus burst.Electrical stimuli could then be initiated after anadjustable delay. Due to the variability of theshape and amplitude of the EMG discharge,there was a jitter of 5 to 10 ms in this delay.However, this was less than 10% of the totalinterval between two discharges.

Results

Phases of central excitability during clonusThe first question asked was whether the silentperiod between two beats of clonus was due tounloading the spindles or to a central refractoryperiod. In these experiments clonus wasmaintained by continual stretch of the tricepssurae while the excitability of the motoneuronsperforming clonus was examined by putting inthree kinds of test stimuli with different delaysafter the preceding clonic burst. These addedstimuli were afferent volleys elicited by tendontaps, H-reflex stimulation, and vibration of thetendon at 80 Hz with 3 mm displacement. Theexperiments were done on a patient wi,thhemiplegia following a vascular lesion of theinternal capsule and on 3 patients with lesions ofthe cervical cord. The sequence of mechanicaland electrical events during clonus is shown infiR I. It is an excerpt of a continuous recordingof sustained clonus maintained by constantstretch on the foot. The times between clonicbursts have been numbered so that they can bereferred to. This numbering is mainly based onthe recordings in man of Hagbarth, Wallin andLofstedt8 and Hagbarth et al.2 Particularlyrelevant are their recordings of ankle clonus,shown in fig 4 of their second paper.

Point 1 marks the beginning of plantar flexionin response to the contraction of the tricepssurae. Between points 1 and 2, the musclespindles will be unloaded and their afferents willbe silent. At point 2, the limit of plantar flexionis reached. The foot is maintained in plantarflexion between 2 and 3. After point 3, the footdorsiflexes in response to the steady forcemaintained by the examiner. Between points 1and 3. the muscle is contracting. The spindlescease to fire after the next point I is reached. Inthe case illustrated, in which ankle clonus was ata rate of 6 Hz, the duration of muscle relaxationlasted about 65 ms. Among all patients, therange of this period was 50 to 75 ms. In thiscase, the subsequent EMG burst lasts about 50ms; the range among all patients was 40 to 70 ms.The silent period in this case was 120 ms; the

323

2 3 4

1 1 [1Mech '11Z [

EMGOn

0 50 100 150 200

Fig 1 Illustration of two clonic EMG bursts duringongoing clonus. Top channel: mechanogram= footmovement I mm Hg indicates I mm Hg pressure inair bladder (see Methods). Bottom channel: EMG insoleus. See text for meaning of numbers. Time inms in all figures.

range among all patients was 90 to 130 ms.In the case of the gastrocnemius, it was shown

by Dimitrijevic and Nathan9 that the latency ofjerks was the same in complete cord lesions as innormals. The latency was 30±3 ms (shown intheir fig 2). The motoneurons will have receivedthe impulses due to this stimulus 15 ms after thetap, afferent and efferent nerve fibres havingabout the same conduction velocities. With theH-reflex afferent volley, the latency of theresponse will be a few ms shorter, as the stimulusis delivered to the nerve in the popliteal fossa.These predominantly Ia afferent stimuli weretriggered by the onset of the clonic burst; theywere timed to occur at incremental 10 msintervals after the onset of the previous EMGburst. They were put into every altern,ate cycleof clonus, in order to minimise any prolongedeffects from the stimulus lasting into the silentperiod after the next clonic burst.

Testing with tendon tapThe experiment shown in fig 2 was made in apatient with a partial lesion of the cord at the6th cervical segment. The top trace shows theAchilles tendon jerk; ankle clonus was notinduced. The arrow shows the time of deliveryof the tap. A response occurred first when thetap was given 90 ms after the onset of theprevious burst. At 110 ms, the entire EMG burstwas replaced by the T-wave. Thus this tap firedthe entire motoneuron pool and no motoneuronswere available for the clonic burst. This responseto tapping started a little earlier than the clonicburst would have donie. At 130 and 150 ms, largeresponses were obtained, occurring during the



http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from



EMG ]

Vibr lum ,,,,40ms 1Orms

EMG X -

Vibr ......

60ms 130ms

]0 40 80 120 160 2002mV

Fig 2 Availability of motor units to a tap on tendoAchillis applied at times shown by arrows. Top trace

is the ankle jerk alone. All traces are threesuperimposed records.

clonic burst. Taps given after 150 ms elicited noresponse; nor did they add to the amount nor

the duration of the EMG response. This phase ofirresponsiveness continued into the early phase ofirresponsiveness after the next clonic burst.Thus the tendon tap produced a response when

it was applied while the muscle was not con-

tracting; when it was applied before this, no

response was obtained. After the clonic burst,there was a period of 90-100 ms during which themotoneurons were not activated by interposed Ia

afferents. But large responses to tap did occur inthe interval from 100 to 150 ms after the onsetof the previous EMG discharge. This was aconsistent finding in two spinal cord injurypatients and one hemiplegic.

Testing with a train of vibratory stimuliIn the second group of these experiments, ashort tetanic train of vibration was used. Thetrain was too short to induce a tonic vibrationreflex and so it did not induce a sustained

5mV1 EMGJVibr

9Oms 140mslOOms

Fig 3 A vailability of motor units to vibration ontendo Achillis. The times shown are the times afterthe onset of the previous clonic EMG burst.

stretch reflex. The reasons for using vilbration arethat it acts as a continuous conditioningstimulus. The central state could be examinedcontinuously by noting which stimulus of thetrain elicited a phasic response. The spindlereceptors are more sensitive to vibration than totaps on the tendon. Further, vibration excitesthe spindles irrespective of the state of con-traction or relaxation of the muscle.

In a patient with spastic hemiparesis followinga vascular lesion in the internal capsule,vibration was applied to the Achilles tendon as atrain of 14 impulses lasting 60 ms at intervals of10 ms after the onset of the previous clonicburst. As can be seen from fig 3, no response wasobtained until the interval between the beginningof the previous clonic burst and the onset ofvibration was 90 ms. A large phasic responseoccurred when this interval was 110, 130, and140 ms. If one assumes that the latency to thevibratory stimulus was of the same order asthe latency of a jerk-around 30 ms&-theresponse to the vibration followed the 7thstimulus at a 90 ms interval, the 2nd stimulus at a110 ms interval, and the first stimulus at the 130and 140 ms intervals. Thus no phasic response tovibration was obtained for 90 to 100 ms after theonset of the previous clonic burst. Themotoneurons could be activated only after 90ms, and many could be activated at 130 and140 ms.

Testing with H-reflex aflerent volleysIn the third group of these experiments, anH-reflex afferent volley was used. Thiselectrical stimulus to the tibial nerve also by-passes the input from the spindles by avoiding thecontraction and relaxation of the muscle. With

324

t



http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from



2 mV 0 40 80 120 160 200

Fig 4 A vailability of motor units to H-reflexafferent volley. All traces are three superimposedrecords.

this stimulus, the motoneurons receive a more

synchronised volley than with a tendon tap andthey receive it a few ms earlier. The pulseduration and strength of the stimulus was

adjusted to obtain a maximal H-wave without an

M-wave. On some occasions an M-wave didoccur. If it was small, the trace was used, but ifit was large, it was discarded. The resulits of a

typical experiment performed on the samepatient as was the subject of fig 3 are shown infig 4. In the first two traces, this stimulus beinggiven 30 and 50 ms after the onset of theprevious EMG burst, the H-wave is seen to bevery small. The response was bigger at 90 ms andreached a peak with a stimulus delay of 60 ms.At 160 ms it came within the next clonic burstand at 170 ms at the end of that burst. Neither ofthese volleys prolonged nor increased the burst.It is concluded that the excitation of Ia afferentfibres cannot activate motor units in this exampleof clonus for 90 to 100 ms after the onset of theprevious clonic burst nor at 160 to 170 ms afterthe burst.

Results of testing with three kinds of Ia afferentvolleysThis group of experiments shows that everyclonic burst is followed by a refractory period,followed by an excitatory period. The timecourse varied in accordance with the rate ofclonus. The testing of these periods with H-reflex afferents shows that the refractory periodis not due to withdrawal of afferent impulsesfrom the s;pindles.The curves obtained show that when the

motoneurons supplying the triceps surae muscleare discharging clonically, they are refractory toIa afferent stimuli immediately after the clonicburst and remain so for about 100 ms; they thenbecome excitable for about 60 ms.

It is evident that these centrally organisedcyclical changes of excitability are the basis ofclonus and that they determine its rate. In clonus,the duration of the periods will not be quite asconstant as the periods following synchronisedinputs, such as those of tendon taps or H-reflexafferents. For the stretch of the muscle is moregradual and so the response of motor units isless synchronised. The duration of the clonicburst was usually around 60 ms for the tricepssurae and that of the tendon tap 15 ms. It canbe inferred that the last activation of musclespindles before they are unloaded occursapproximately 30 ms prior to the onset of thelast motor unit potential; for there is a prolongedEMG discharge in response to stretch-respondingafferents.

Attempts to change the rate of clonusOne of the features of clonus is the constancyof its rate in any muscle. In five of the patients,sustained ankle clonus was recorded at varioustimes over a period of two years. In anyindividual patient, the clonus rate of a muscledid not vary by more than + 15°% of its averagerate; this was less than 0-7 Hz deviation from theaverage rate of 5 3 Hz in the most extreme case.These slight changes in rate occurred randomlyin time; there was no tendency for the rate ofdecrease or increase over a period of monthsor years.The question arises whether the rate of clonus

can be changed to values outside this narrowband since, as was shown above, a tap appliedwith the appropriate delay during clonus couldinduce the motor units to fire earlier than theywould otherwise have done. By shortening thesilent period following each EMG discharge.and by increasing the synchronicity of thedischarge, the rate of clonus could perhaps be

325



http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


Al R Dimitrijevic, P W Nathan, and A M Sherwood

increased. Experiments were done in which theafferent volley from the tap on the tendonarrived at the motoneurons at the beginning ofthe time when they were discharging in the clonicburst. None of these trials changed the rate ofclonus by more than the spontaneous variabilityin this rate, mentioned above.As the rate of clonus could be changed only

to this limited degree by the addition of Ia

afferent volleys, experiments were performed totry to change the rate in other ways that actedvia peripheral factors. The four ways, describedin Methods, of placing the limb in ice-water, ofadding a mass to the foot, of moving the footthrough a viscous solution, and of the experi-menter opposing the movement of the footcarrying out clonus, sometimes changed theamplitude of clonus. But the clonus did notcontinue at a slower rate; it ceased to occur.

Just as it stopped, for two or three beats the rateof clonus was slowed by a maximum of I Hz.It appears, then that the constancy and im-mutability of the rate of clonus in any muscledepend on central organisation.

Tendon reflexes repeated at various rates inclonic musclesThe question was asked whether in clonicallyresponding muscles a maximal number ofmotoneurons respond to tendon taps deliveredat the rate of clonus. The response of a muscleto a repetitive tapping on its tendon depends on

the peripheral factor of the state of the spindlereceptors during contraction and relaxation ofthe muscle and on the central factor of cyclic

CaWW 50

0az

0

0 2 4 6 8 10Rate of tapping (Hz)

Fig 5 The curve is the averaged amplitude of 100response3, normalised to the peak value at a tappingrate of I Hz, shown at each rate of tapping on

tendo Achillis. Representative successive sets of foursuperimposed traces are given above the curve, at

tapping rates of 1, 4, 6, 7 and 10 Hz.

excitability at the time when the volley from thetap is received. To produce repetitive tendonjerks, it was necessary to decrease the tensionapplied continuously to the muscle to below thethreshold for inducing sustained clonus. In fact,the weight of the foot provided adequate tensionfor obtaining jerks without inducing clonus.Taps were delivered to the tendo Achillis atvarious rates in three patients with spinal cordlesions, two patients with hemiplegia due tovascular lesions of the internal capsule and threenormal subjects. The response to 100 consecutivetaps was integrated and averaged, starting 30 msafter the tap and continuing for 20 ms; the firstfew responses of every series were omitted, asthe response pattern was not immediatelystabilised. The data obtained in a patient withan almost complete lesion at the 5th thoracicsegment are shown in fig 5. The amplitude ofthe EMG response, normalised to the amplitudeat 1 Hz and named 100 is shown at each tappingrate. Four sequential sample responses fromwhich the curve was obtained are shown aboveit. Maximal responses occurred at two repetitionrates: 1-2 Hz and 5-6 Hz. The rate of clonuswas 5-6 Hz in this patient.

This result was typical of all patients: themaximal EMG response occurred when the rateof tapping was at the rate of clonus. A maximalnumber of motoneurons responded to taps givenat a rate of 1 Hz; taps not shown in the figuregiven at slower repetition rates also produced amaximal number of responding units. With tapsgiven at 1 Hz or less, each tap is treated as anisolated event; most central activity from theprevious tap will have subsided before the nexttap is given. All motoneurons did not respondto the taps given at rates of 3, 4, 7, 8, 9 and10 Hz. In all cases a maximal number of motorunits responded to tapping at rates of 5 and 6 Hz,rates similar to the rate of clonus of that musclein that subject. This suggests that the rate ofclonus is a manifestation of a periodicity ofexcitation affecting motoneurons responding tostretch.With tapping rates above that of clonus, two

kinds of response were seen: a poorly syn-chronised low amplitude response to eachstimulus; or a large synchronised response toalternate taps. There is a 40-50% response ata tapping rate of 10 Hz, which is twice the rateof tapping of the 95-lCO% response at 5 Hz.That peripheral factors were not playing a roleis seen by examination of the maximum respon-siveness possible from a single motor unit. Fig 6shows single motor unit responses to tapping rates

326



http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from



IL-~

=~~~3

I4ms

0 2 4 6 8

Tendon tap frequency (Hz)10 12 14 16

Fig 6 Response of single motor unit of tricepssurae to taps repetitively applied at rates of 1-16 Hz.Rates shown in abscissa, percentage of responses to100 taps in ordinate. Inset in right upper corner;samples of response of single muscle fibres to tappingat 5 and at 8 Hz.

AEMG -

EMG -4 ----- 4^1 : [0 5mV

Mech r [lmmHg

E F

EMG

Mech-0ih lW 4lt 1 t 1 It t t I t

. . . I

250ms

of I to 16 Hz. Samples of jittering response are

shown at a tapping rate of 5 Hz and of a responseto alternate taps at 8 Hz. While it was possibleto excite a single unit by tapping at rates up tothe clonus rate, it was impossible to elicit morethan a 50% response beyond that rate, indepen-dent of the amount of contraction of the muscle.This can be understood in terms of the centrallyinduced refractory period. The failure of thewhole motor unit population to respond at lowerrates (fig 5) was due to induced afterdischarges,as described below.The response of motor units in spastic patients

was examined in more detail. In these patients,a single tap often produced a first response,followed by two bursts of discharge of the motorunits. A typical experiment was carried out ona patient with hemiplegia. The foot was plantar-flexed sufficiently to prevent sustained ankleclonus. The effect of tapping at different rates onthe tendo Achillis is shown in fig 7. The EMGis shown on the top line and the movement ofthe muscle by the mechanogram trace on thebottom line. The rate of clonus in this musclewas 5*5 Hz. The first few responses at eachtapping rate were not recorded; each recordshown is a single example taken from a seriesof taps. The rate of tapping was 2 in a, 3-3 in b,6 in c, 8 in d, 10 in e and 12-5 in f. Fig 7A showsa first large synchronised EMG response,followed by a second poorly synchronisedresponse of variable amplitude, and a thirdresponse of lower amplitude with less syn-chronisation. Each of these responses is a

Fig 7 Response of the triceps surae to tap ontendo Achillis at various rates (see text) in a patientwho responded to a tap by 3 EMG bursts. Arrowsshow time of taps. Mechanogram records musclemovement. I mm Hg means I mm Hg pressure inair bladder.

muscular contraction, as is shown by themechanogram. The early response follows thetendon tap with the expected delay of 30 ms;the later responses follow the tap with the samelatencies as that of successive beats of clonus.The proposed interpretation of these events isthat the second response is a poorly synchronisedphasic stretch reflex induced by the stretchwhen the muscle relaxes at the end of the firstresponse, and the third response is the same,being induced by relaxation following the secondresponse. These responses are occurring at therate of clonus, 5-5 Hz. These responses aresmaller than the first response, being less syn-chronised. They are less synchronised as thestimulus invoking them is less synchronised thanthat of a tap.The largest response occurs when the tap is

delivered 30 ms before or at the same time as the2nd or 3rd responses to a previous tap; this isseen in fig 7C and E. With a tapping rate of 3-3Hz (fig 7B), one tap occurred in the period be-tween two delayed responses and the next tapoccurred after the second delayed response. Theresult was that every alternate tap produced alarge response and the other alternate tapsproduced a small first response wiith a minimal

50

-10

-o

-L

327

t

Br - *4

t



http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


M R Dimitrijevic, P W Nathan, and A M Sherwoodmovement and very small delayed responses. It isto be noted that when the 2nd and 4th and sub-s1quent alternate taps were delivered, the musclewas relaxing and the spindles were available forstretching. It is apparent that the delayedresponses were also inducing refractory periods,similar to those following the first response.

It is clear that the amplitude of the response totapping depends on the time at which the tap isdelivered in relation to the periods of refrac-toriness and excitability occurring at the time.There is no response to a tendon tap givenwithin 100 to 120 ms of a previous discharge,whether that discharge is the 1st, 2nd or 3rdresponse to a tap. When the volley from the tapreaches the motoneurons between 120 and 160ms after any of these previous discharges, itcauses a synchronised response. Inputs received

during the refractory period following evenpoorly synchronised discharges of motoneuronscan have no effect, whereas inputs receivedduring the excitatory periods facilitate the dis-charge that is about to occur. When there ispoor synchronisation of the discharge of motorunits in the 2nd and 3rd responses to a tap, thenew input finds a longer excitatory period duringwhich the response can be facilitated. The inputfrom tapping does not alter the periodicity andre-set the cycle. Tapping cannot activate thesystem to respond at rates other than the clonicrate.Why the whole population of motoneurons

failed to respond at tapping rates of 3 and 4 Hzin the experiment illustrated in fig 5, could beaccounted for also on central factors.These experiments with activating spindle

A

Tib ant

Ext dig long

Soleus

Gastroc

.1 1. L. thl , wilUl,aI JIf 1 .1J aJ(iEl ..

et e e~~~~~~~~~~~~~~~~~~,,-M+ +-t- '.JA I

Gonio ankle

Accel tendon

B

Tib ant lo I

Ext dig long 0 ,

Soleus 150,V

Gastroc

Gonio ankle

Fig 8 (a) Spray to dorsum ofhemiplegic foot. (b) Spray tounaffected foot. Time in topchannel, I s; this channel israised during spraying.Penultimate channel indicatesmovement of foot, downwardtrace showing dorsiflexion offoot. Last channel showsmovement of tendo Achillis.

Accel tendon

I

328



http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from



afferents at various repetition rates show that theresponse to such an input is dependent on thetime when the input arrives in relation to therefractory and excitatory periods. The responseof the total muscle also depends on the proportionof the motoneurons that is available.

Cutaneous stimuli causing clonusMuscles liable to show clonus discharge clonicallynot only with Ia afferent volleys but also with anyexteroceptive stimulus of sudden onset andsufficient intensity. Although clonus is usuallyand easily induced by stretch of the muscle, it isalso induced by other stimuli. The cutaneousstimulation most likely to induce and maintainclonus are noxious stimuli and sudden cold.Merely fixing the clip-on electrodes onto the skinoften induced clonus; in some of our patientswith spastic paraparesis, this stimulus was soeffective in maintaining clonus that it was im-possible to use these electrodes, clonus continuingas long as the points were in the epidermis.An example of clonus being induced by a

cutanteous cold stimulus is illustrated in fig 8.The patients had a right hemiplegia due to avascular lesion in the internal capsule. Figure 8shows clonus induced by spraying the dorsum ofthe right foot with ethyl chloride, and 8 b thatinduced by spraying the dorsum of the left un-affected foot. The duration of spraying is shownby the elevation of the time-maker. This cuta-neous stimulus to either foot caused clonus in aflexor response. The flexion came on 0 5 s afterspraying the ipsilateral foot and 1 0 s after spray-ing the contralateral unaffected foot. Theseexamples show that stimuli that induce a poly-synaptic flexor or extensor reflex tend to induceclonus. On some occasions the clonus occurredwithout any movement of the limb preceding theclonus. A variety of stimuli can elicit clonus whenthe motoneurons are abnormally excitable. Invery spastic patients, a continuous noxiouscutaneous stimulus can cause sustained clonus ifthe muscle is under some tension. Experimentswere done in which clonus was maintained aslong as a crocodile clip was attached to the foot;it caused clonus of the triceps surae and often ofthe tibialis anterior as well, and tonic contractionof the other muscles of the leg.

In instances where clonus was induced andmaintained by tens-ion on the muscle, the ampli-tude of the beats would sometimes decline andfade out. Then, cutaneous volleys induced byscratching the skin over the muscle would oftenprovide sufficient input to the spinal cord torestore the amplitude of the clonus.

329

Discussion

Whether clonus is entirely due to the peripheralrepetitive activation of the spindle stretch recep-tors or depends also on a spinal governormechanism created by the cyclic activation ofthe stretch reflex is the main question we haveinvestigated.When single taps are given to the Achilles

tendon of spastic patients, the duration of thereflex contraction and relaxation of the musclesis around 300 ms. If clonus were simply arepetitive stretch reflex with no regulation ofrate, it could occur at a rate of 3 cycles a second,given time for a complete cycle of contractionand relaxation of the triceps surae; or it mightoccur at any rate up to 14 Hz, when tetanusoccurs. In fact, clonus, as is well known, occurswithin a narrow band of frequencies from 5 to8 Hz. This fact alone shows that central regula-tory factors are operating.

If the rhythm of clonus is determined by theavailability of the spindle receptors during thephases of contraction and relaxation of themuscle, it should be possible to change the rateof clonus by exciting these receptors at differentrates. And if clonus is due to the modulation ofa continuous stretch on the muscle by the con-traction and relaxation of the muscle itself, thenthe rate of clonus would be affected by the massof the part being moved, by muscular viscosity,and by the force applied to the part. We wereun,able to change the rate of clonus; nor was thisachieved by Walsh." nor by Sherwood andDimitrijevic'0 using other methods. We also didnot change the rate of clonus by altering theforce applied to the part.The investigatioins reported here show that

between two bursts of clonus there is a refractorystate within the spinal segments followed by anexcitatory state. The motoneurons cannot respondto continuing tension put on the muscle untilthey have become adequately excitable follow-ing the refractory period. This periodicity can bemodified only during the very brief time whenthe refractory period is changing to the excitatoryperiod; Ia inputs can then influence the period.As the duration of the EMG burst in sustained

clonus depends on the duration of spindleactivity, the question arises whether this durationcan be prolonged by extending the period ofactivation of the spindles. We have found that itcannot be, since delivering additional stimuli inthe form of tendon taps, vibration or electricaldepolarisation of the Ia fibres during the periodimmediately following the excitable phase does



http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from



not result in any activation of motor unitsinvolved in the clonus activity. This demonstratesthe complete unresponsiveness of these pathwaysduring this time.The sudden decline in excitability after the end

of the clonus EMG burst and the long durationof the refractory period, together with the rapidrecovery of excitability, suggest the presence ofan active inhibitory process. Brune and Schenck"examined the silent period between two clonicbursts by putting in random H-reflex volleys.Some of their H-waves were true H-waves andothers had an intensity to cause large M-waves.They also found a loss of responsiveness tostimulation between the EMG bursts. They con-cluded that the cessation of motoneuron activityat the beginning of the silent period was due torefractoriness of the motoneurons after firing,combineld with Renshaw inhibition, and that therest of the period was due to withdrawal ofexcitation coming from the spindle afferents.Struppler, Burg and Erbel12 have concluded thatthese factors include not only muscle spindleunloading, but also autogenetic inhibition byGolgi afferents, recurrent inhibition by Renshawcells, and the refractory phase of the moto-neurons.An experiment demonstrating the importance

of spinal mechanisms in clonus was performed byWalsh.13 To a foot performing clonus, he addeda periodic force. Modulation of this appliedforce was at frequencies of 5 and 5 5 Hz whilethe clonus rate was 6Hz. Beat frequencies cor-responding to the difference in rate of clonus andthe force modulation frequency (0 5 and 1 Hz)occurred. He concluded6"13 from this and fromother evidence that the rhythm of clonus is"centrally tuned".Other investigators who favoured the central

excitatory state as being the essential factor inclonus are Wachholder and Altenburger14 andBrune and Schenck." Wachholder and Alten-burger 14 showed that the latency of the first beatof the clonus was the same as that of the stretchreflex; they claimed that this time relation wasnot maintained during sustained clonus. Theythere,fore concluded that clonus is started bystretch and that the rhythmical discharge ismaintained by central factors. We found that thelatency of the EMG response was correct through-out sustaine,d clonus for each burst to be aresponse to Ia afferent volleys.The features of clonus that have led us to put

most weight on central factors were the findingsthat a maximal number of motoneurons respondto repetii ive tendon taps at the rate of clonus of

that muscle, and that the mechanical activity ofthe contraction of the muscle adjusted itself tothat rate. Furthermore, the rate of clonus isrelative,ly constant, and the direction of the smallshifts in rate is peripherally controlled. We haveshown that an important mechanism in this rateregulation is the periodic repetition of excitatoryand refractory states corresponding to the cyclicactivation of clonus.That clonus depends essentially on factors in

the periphery and minimally on central factors isthe view of investigators to be discussed now.Bernhard and Therman"5 studied the tricepssurae of the cat in a decerebate state. Decereba-tion causes a particular kind of hyperexcitabilityof stretch reflexes. Using passive movements ofthe limbs and muscles, they showed that the con-tinuous muscle contraction could be changedinto a rhythmical contraction. When a tonicallycontracting extensor muscle was passively flex-ed,the contraction became rhythmically intermittent:there were bursts of EMG actively separated bysilent periods. When the new position of the limbwas taken up, the tonic activity recurred. It wasthe changing proprioceptive input from themuscle itself, from its synergists or its antagonists,that changed the tonic activity into bursts ofdischarges separated by silent periods. Singlemotor units were then discharging within limitedfrequency bands. They repeated the observationsof Denny-Brown'6 that the application of passivetension of the muscle will start rhythmicalactivity of a single unit in the extensor muscleof a decerebrated cat. This work then shows thatin the decerebrate cat proprioceptive input fromthe limb can start rhythmical discharges of motorunits while the limb is being moved.A characteristic of clonus is the synchrony of

the motor discharge. It may not be marked forthe first three or four beats of clonus, then itbecomes established. The synchrony of dischargeoccurs although peripheral factors contributing toclonus, such as the in,put from the spindles, thegeometry of the muscle and the rate of relaxa-tion of the parts of the muscle, are far frombeing synchronised. This shows that the reflex isrigidly controlled in time and also in spatialextent in the motor unit pool. The discrepancybetween the scattered peripheral factors and thesynchronised motor unit response indicates thatcentral mechanisms are playing a major role.

It appears to be that when an adequate volleyof impulses reaches the motoneurons liable torespond clonically, it fires them synchronously;and it is the synchrony of firing that starts clonus.This synchronous response occurring in the same

330



http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from



rhythm determines the rate of clonus. Thesynchronous firing of these motoneurons is alsostarted by a tendon tap, and so one tendon tapcan initiate several bursts of clonus.The question remains unanswered why certain

motoneurons tend to fire synchronously ratherthan asynchronously in response to a continuousinput. Whatever the cause, there is no absolutedifference in the organisation of these moto-neurons, for they often respond tonically; andmotoneurons of other muscles that habituallyrespond tonically can be entrained to respondclonically when the whole limb is being rhy.h-mically moved in a clonic rhythm.We conclude that the hypothesis that clonus

is essentially a peripheral phenomenon, being arepeated stretch reflex, is inadequate. The evi-dence favours the hypothesis that it relies on acentral, spinal generator or controller. Thisgenerator should be conceived as a transitoryfunctional organisation, made up of the followingelements: the movement of the muscle and partsof the limb, the proprioceptive volleys from thelimb, the segmental reflex activity which isinfluenced by peripheral, propriospinal and supra-segmental mechanisms, when these are present.The features of this transitory functional organi-sation include:(a) cyclic, regular activation at a fixed phase by

Ia afferent volleys;(b) activation of motoneurons and coactivation of

convergent interneurons;(c) functional isolation of segmentally organised

repetitive stretch reflex activity from the dis-ruptive influence of extraneous volleys;

(d) maintenance of the motoneuron pool excita-bility;

(e) the resultant excitatory-refractory cycles whichsustain and regulate the oscillatory movements.

These features are consistent with the idea of aspinal governor, rather than a peripheralmechanical controller. While the peripheral inputis essential for the reactivation of the cyclicbursts, the spinal mechanism tightly controls theoverall activity. The periods of refractoriness thatare the cause of the intermittent discharge ofclonus must be due to inhibition of the moto-neurons and/or interneurons. The period ofrefractoriness is too long to be due to Renshawinhibition; and Veale, Rees and Mark'7 haveshown that this early recurrent inhibition isarbsent in spasticity in man. What does cause thisinhibition of certain motoneurons and relatedinterneurons remains unknown. We have notpresented any evidence on this matter, as in thispaper we are concerned only with the relative

331

importance of peripheral and central mechanismsin the genesis of clonus.As clonus is due to mechanisms within the

local segments of the spinal cord in spasticity, itmight be expected to see some indications of itspresence in normal subjects. This is the case, aswe mentioned in the introduction; and it iscommon experience that when one sits and putsthe correct amount of tension on the leg, aclonic movement of the leg can easily be inducedand maintained. Clonus in normal subjects hasbeen studied by Gottlieb and Agarwal.3. Thisclonus had the same characteristics as clonus inpatients, showed the same relatively narrow bandof frequencies based on the same timing of EMGbursts, and the same relative independence ofloading on the limb. To obtain clonus in normalsubjects, apart from the methods mentionedbefore of injecting substances that increasespindle discharge, increased muscle tension hasto be maintained, and other inputs to the moto-neurons such as those necessary for voluntarymovements have to be prevented. Gottlieb andAgarwal induced flexion and extension of thefoot at the ankle joint by a foot-plate connectedto a motor, which oscillated the foot at repetitionrates between 5 and 8 Hz. With these rates ofdriven oscillation, there were "large excursions ofthe foot with very small applied torques andstrong synchronous activation of the soleusmuscle and, sometimes, of the anterior tibialmuscle". And what was more interesting, theyoften found that when the motor was turned off,clonus continued for more than a minute withoutany tension applied to the foot; it could havecontinued longer but the subjects then stopped it.When there are lesions dividing corticospinal

tract fibres, these segmental circuits, present inthe normal, become more independent and lessinfluenced by volleys of impulses fro-m the brain.In the aibsence of supraspinal inputs, the firingthreshold of motoneurons becomes low. Whenclonically reacting motoneurons are facilitatedby Ia affergent inputs, they will respond clonicallyto any input. We often found, as have so manyother investigators, that clonus could be inducedby the non-specific descending facilitation causedby Jendrassik's manoeuvre. Further, clonus couldbe induced by inputs in flexor reflex afferents. Itis also true that stimuli activating these pathwayscan stop clonus. Which event occurs depends,among other factors, on the position of thestimulus invoking flexion in relation to themuscle responding clonically.

In conclusion, it is now clear why the questionof the relative importance of the central and the

D



http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from



peripheral mechanisms in defining clonus hasbeen difficult to answer. As we have shown, therigidly determined cyclic excitatory-inhibitoryphases of the segmental organisation determinethe rate of clonus. Yet excitation for themaintenance of these oscillations must beprovided by properly timed muscle spindleafferent volley. The interaction of the musclecontraction-relaxation cycle with the excitatory-inhibitory cycle of the repetitive stretch reflexautomatically adjusts the reflex contractionduration to the value determined by the segmentalgovernor.

We thank our colleagues from the departmentsof neurology, neurosurgery and orthopaedicsurgery at the University Hospital of Ljubljana,of the National Hospital for Nervous Diseases,Queen Square, London, and of the Texas Insti-tute for Rehabilitation and Research, Houston,for referring their patients to us for our study.This wo>rk was made possible through thefinancial support of the Boris Kidric Fund ofSlovenia, the Medical Research Council of GreatBritain, a grant from the Bob and Vivian SmithFoundation of Houston, and RehabilitationService Administration Grants 13-P-59275/6 and16-P-56813/6.

References

1 Szumski AJ, Burg D, Struppler A, Velho F.Activity of muscle spindles during muscle twitchand clonus in normal and spastic human sub-jects. Electroencephalogr Clin Neurophysiol 1974;7:589-97.

2 Hagbarth KE, Wallin G, Lofstedt L, AquiloniusSM. Muscle spindle activity in alternating tremorof Parkinsonism and in clonus. J Neurol Neuro-surg Psychiatry 1975; 38:636-41.

3 Gottlieb GL, Agarwal GC. Physiological clonusin man. Exp'Neurol 1977; 54:616-21.

4 Struppler A, Schulte FJ, Scheininger R, Kukku

M. Eine elektromyographische Untersuchung beiSpastik und Rigor. Deutsch Z Nervenheilk 1961;183:134-47.

5 Marsden CD, Meadows JD, Hodgson HJ. Obser-vations on the reflex response to muscle vibrationand its voluntary control. Brain 1969; 92:829-46.

6 Walsh EG. Clonus: beats provoked by the appli-cation of a rhythmic force. J Neurol NeurosurgPsychiatry 1976; 39:266-74.

7 Dimitrijevic MR, Spencer WA, Trontelj JV,Dimitrijevic M. Reflex effects of vibration inpatients with spinal cord lesions. Neurology1977; 27:1078-86.

8 Hagbarth KE, Wallin G, Lofstedt L. Musclespindle activity in man during voluntary fastalternating movement. J Neurol NeurosurgPsychiatry 1975; 38:625-35.

9 Dimitrijevic MR, Nathan PW. Studies of spas-ticity in man: 2, Analysis of stretch reflexes inspasticity. Brain 1967; 90:333-58.

10 Sherwood AM, Dimitrijevic MR. Clonus: a sim-plistic model of a complex system. In: Proc 29 AConf Eng Med Biol 1976; 18:278.

1 1 Brune HF, Schenck E. NeurophysiologischeUntersuchungen uber den Klonus bei Spastikern.Arch Psychiatr Nervenkr 1960; 201:65-80.

12 Struppler A, Burg D, Erbel F. The unloadingreflex under normal and pathological conditionsin man. In: Desmedt JE, ed. New Developmentsin Electromyography and Clinical Neurophysi-ology, vol 3. Basel: Karger, 1973; 603-17.

13 Walsh EG. Ankle clonus-an autonomous centralpacemaker? J Physiol 1971; 212:38-9.

14 Wachholder K, Altenburger H. ExperimentelleUntersuchungen zur Entstehung des Fussklonus.Deutsch Z Nervenheilk 1925; 84:117-21.

15 Bernhard CG, Therman PO. Rhythmical activityof motor units in myotatic reflexes. Acta PhysiolScand (suppl) 1947; 14:1-14.

16 Denny-Brown D. On the nature of posturalreflexes. Proc R Soc [Biotl 1929; 104:252-301.

17 Veale JL, Rees S, Mark RF. Renshaw cellactivity in normal and spastic man. In: DesmedtJE, ed. New Developments in Electromyographyand Clinical Neurophysiology, vol 3. Basel:Karger, 1973; 523-37.

332



http://jnnp.bmj.com

/J N

eurol Neurosurg P


ownloaded from


clonus: role of mechanisms - jnnp.bmj.com · clonus: the role ofcentral mechanisms ... three kinds...

Documents