tremor, the cogwheelphenomenon parkinson's disease · the term "cogwheel phenomenon"...

Journal of Neurology, Neurosurgery, and Psychiatry, 1981, 44, 534-546

Tremor, the cogwheel phenomenon and clonus inParkinson's diseaseL J FINDLEY, M A GRESTY, AND G M HALMAGYI*

From tile Medical Research Council Hearing and Balance Unit, Institute of Neurology,National Hospital, Queen Square, London and the Department of Clinical Neurophysiology,*Royal Prince Alfred Hospital, Camperdown NSW, Australia

SUMMARY Resting and postural tremor, intention and action tremor, clonus and the cog-wheel phenomenon in Parkinson's disease have been characterised in terms of frequency contentusing spectral analysis. Typical resting tremor ranged in peak frequency from 4 to 5 3 Hz withtremor in each individual varying only by 0-2 to 0 3 Hz. The peak frequency of postural tremorranged between 6 and 6'2 Hz. Intention tremor appeared to be an exaggeration of posturaltremor. Clonus evoked by active or passive stretch at the wrist had a frequency of 6 Hz andappeared to be a continuation of postural tremor. The cogwheel phenomenon was found atfrequencies between 6 and 6-5 Hz and between 7-5 to 9 Hz. Action tremor was indistinguishablefrom the cogwheel phenomenon. Some patients had either a symptomatic resting tremor with aconcurrent 6 Hz component of smaller amplitude or a symptomatic postural tremor with a

4-5 Hz component of smaller amplitude. These combinations would produce two peaks in thepower spectrum. When this occurred EMG studies showed that individual muscles had twotypes of rhythmical activation suggesting that the tremors have separate mechanisms. Likewisesome patients had a symptomatic 6 Hz tremor on posture with a second peak at 8-10 Hz in thephysiological band. Therefore, the 6 Hz postural tremor is not an exaggeration of physiologicaltremor. On the basis of wave form and frequency similarities postural tremor, the low frequencytype of active or passive cogwheeling, intention tremor and clonus possibly involve a common

spinal mechanism. Higher frequency cogwheel phenomenon and action tremor may be an

exaggeration of physiological tremor. More than 80% of patients with Parkinson's diseasemanifest tremors at both 4-5 Hz and 6 Hz. This combination would appear to be the strongestobjective criterion for the diagnosis of basal ganglia disease.

There are three forms of tremulous movement inParkinson's disease, the resting tremor, action andpostural tremor, and the cogwheel phenomenon.Parkinson' re-emphasised that resting tremor

was a dominant feature in this disease, since whenthere have been numerous descriptions of itscharacteristics.2-4 Frequently commencing infingers and thumb on one side and later involvinglips, tongue and the ipsilateral ankle. It is usuallydistal, and is seen with the limb at rest ceasingbefore or with the onset of voluntary movement.As the disease progresses, however, it may bepresent throughout action and can be maintained

Addressfor reprint requests: DrMA Gresty,MRC HearingandBalanceUnit, The National Hospital, Queen Square, London, WCI- 3BG.Accepted 12 February 1981

in posture. Resting tremor is reported between 4and 7 Hz, and results from the reciprocal contrac-tion of antagonistic muscle groups.5 The origin ofthis rhythmical involuntary activity is still notclear,5 6 but degeneration of the nigro-striatal path-way and depletion of dopamine in the striatum arethe most constant central features required for thesymptoms of Parkinson's disease to occur.7Lewy8 discussed the appearance of a static (that

is postural) and locomotorial action tremor inparalysis agitans. De Jong9 distinguished actionfrom intention tremor, and his recordings showeda tremor throughout movement which was con-tinuous with the tremor that occurred when thelimb was maintained in posture. Lance et al'0found marked action tremor in 17 of 33 patients

534

Protected by copyright.

on April 17, 2021 by guest.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.44.6.534 on 1 June 1981. Dow

nloaded from

http://jnnp.bmj.com/

Tremor, the cogwheel phenomenon and clonus in Parkinson's disease

with Parkinson's disease who had symptomaticresting tremor, and in six out of seven patientswithout resting tremor. The frequency range ofaction tremor was 7 to 12 Hz and it was shownthat this tremor arose through co-contraction ofactive muscles, and was thus distinct from theresting tremor. Lance'0 and his colleagues thoughtthat this tremor was an exaggerated physiologicaltremor independent of the stretch reflex arc orgamma efferent activity. In addition to the com-mon tremors, there is recent evidence in Parkin-son's disease that there can be an associatedintention tremor, that is, an exaggeration of tremorbrought out at the termination of movement, suchas during the finger to nose test. This was inter-preted as being due to exaggeration of posturaltremor.19

Rigidity as a feature of the disease is notmentioned in Parkinson's original monograph.'The term "cogwheel phenomenon" has been at-tributed to Camillo Negro, who first described it in1901,1" though the first description in Englishliterature is that of Moyer in 1911.12 Lance'0elicited the cogwheel phenomenon during passivemovements of the hand at the wrist joint at fre-quencies ranging more often between 6 and 12 Hzthan from 3 to 6 Hz. He concluded that eitherthe resting or the action tremor was responsiblefor what was interpreted to be a segmentation of ahyperactive stretch response to give the typical"cogwheel effect". This concept of the origin ofthe "cogwheel phenomenon" was supported bySelby,2 though Denny-Brown3 suggested that bothrigidity and tremor could be attributed to thesame underlying neuronal mechanism. The latterarises because this mechanism is unstable. Theorigin of rigidity also is not certain, particularly asto whether the final common pathway is via thegamma efferent system,13 14 the alpha motor-neurones,15 or both'6 17; a summary of currentviews is given by Evarts.18

Finally, there are occasional reports that sus-tained passive stretch in extension or flexion ofthe wrists or fingers in some patients produces apowerful regular tremor which we find is clinicallyindistinguishable from clonus and feels as if it is acontinuation of the cogwheel phenomenon pro-duced during the dynamic phase of muscle stretch.This has been recognised in all stages in thenatural history of the disease.3The present survey of the rhythmical involun-

tary movements of Parkinson's disease has utilisedtechniques of wave form analysis in an attemptto make a more precise and simple classificationof the diverse phenomena involved. The primary

concern has been a phenomenological analysis ofdistal tremor as transduced by movement detectorsattached externally to the limb. Surface electro-myograms were performed in order to relate thewave forms of tremor to muscle activity. Onlymovements in flexion and extension at the wristhave been considered; and because the amplitudeof the tremors in Parkinson's disease fluctuate somarkedly, attention has been given only to whethertremor was symptomatic or not.

Subjects and methods

Patients with idiopathic Parkinson's disease (mean age64+8-1 (ISD) years, and mean duration of symptoms5 9±4.2 (ISD) years) were studied in the MRC Hear-ing and Balance Unit at the National Hospital forNervous Diseases, Queen Square. The first 40 of thesepatients were included in the survey of resting andpostural tremor, the results of which are presented inthe frequency histograms in fig 1. More than 20 ad-ditional patients were investigated during the prepara-tion of the manuscript. Thirty patients were takinglevodopa in a combined tablet with a peripheral de-carboxylase inhibitor carbidopa (Sinemet); the averagedaily dose was 370 mg (range 110 to 825 mg). Six ofthe levodopa group and three of the other patientswere taking anticholinergic drugs. Drug treated patientswere all studied in the morning, having discontinuedtheir medication on the previous evening. An initialpilot study of patients with Parkinson's disease con-firmed that levodopa had no effect on the frequencyof tremor and, in keeping with the findings of others,did not always attenuate the amplitude.20 Manypatients were studied on several occasions.The subject was seated in a chair with broad flat

arms, each covered with a sheet of foam rubber todampen artefacts produced by the arm resting on a"pulse." Tremor at rest was measured with thesubject's forearms lying pronated along the arms ofthe chair with the hands dangling freely over theedges. Tremor in posture was measured with thehands elevated to a horizontal position, gently stretch-ing out beyond the arms of the chair with the sub-ject's forearms still supported in their length by thearmchair. Cogwheel phenomena were evoked passivelyand actively. In the active situation the subject wasseated as above with hands extending beyond thearms of the chair; to evoke cogwheeling the wristswere flexed and extended throughout their range ofmovement as rhythmically as possible, each cycletaking typically two to three seconds. The examinerpalpated each wrist gently during this manoeuvre.Passive cogwheeling was evoked by the examinerflexing and extending the subject's hand whilst gentlypalpating the subject's wrist. Clonus was evoked eitherby the examiner manipulating the wrist in flexion /extension or supination/pronation in order to inducestretch, or by the subject pressing in flexion with hisfingers on to a strain gauge mounted beyond the edge

535



http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from


L J Findley, M A Gresty, and G M Halmagyi

n

10

5

Tremors of Porkinson's diseoseResting tremors

- Symptomotic- Asymptomatic

............ Sub-peaks

n

4 5 6 7 8 9Hz

Postural tremorsclonus a

Hz

of the arm of the chair and thus inducing tendonstretch.Tremulous and vibratory movements were trans-

duced using linear accelerometers and the Schottkybarrier photodetector. In transducing tremor theaccelerometer was mounted on the dorsum of the handwith its sensitive axis in the plane of flexion/extensionor pronation/supination. A target light for theSchottky barrier photodetector was attached to thebody of the accelerometer. Accelerometers used werelinear piezo-resistive devices with a flat frequencyresponse from DC to at least 200 Hz and sensitive toat least 1/lOOOth of g. All recordings were calibratedabsolutely with respect to gravity. The Schottkyphotodetector was used to transduce limb displace-ment. The detector was mounted in the focal planeof a camera which was positioned to view the limbfrom a suitable angle, in this case viewing the lightplaced on the dorsum of the hand from the side. Thedetector gives a read out in x and y co-ordinates ofthe target light.19 Recordings were presented on astorage oscilloscope and a y/t electrostatic ink jetrecorder. Surface electromyograms were taken usingsilver/silver chloride cup electrodes of 4 mm diameter,placed 2 cm apart. Accelerometric signals were pro-cessed "on line" using a digital computer to calculatepower spectra. They were shaped using a high passfilter with a time constant of 1 second and a low passfilter with a fall off of 6 dB per octave having an

Fig 1 Frequency histograms of resting and posturaltremors in the individual limbs of 40 patients withidiopathic Parkinson's disease. Upper histogram:typical gross resting tremor characteristic of thedisease, occupying frequency band 3-8 to 5 6 Hz.Above 5,8 Hz are faster tremors of smaller amplitude,more akin in appearance to the typical posturaltremor of Parkinsonism although present with limbin a resting position. Lower histogram: posturaltremor at frequencies below 56 Hz had the visiblecharacteristics of the typical resting tremor althoughthe limb was in posture, whereas those above 58 Hzhad the appearance of the typical fine fast tremor ofParkinsonism. The envelope of asymptomatic tremorfrequencies closely follows the distribution ofsymptomatic tremors suggesting analysis of theasymptomatic tremor may have diagnostic value.Sub peaks on the power spectra which were notharmonically related to the dominant pptk frequencyare plotted in the histograms and these are alsopredominantly distributed around the modal valuesfor the frequencies of symptomatic tremor. Insertsin the lower histogram show the coincidence offrequencies cogwheeling, clonus and typicalsymptomatic postural tremor. The bars indicate theabsolute range of frequency values obtained forcogwheeling and rippling.

elbow at 16 Hz, then digitised with a 10 ms samplingrate, windowed with a hamming function andsubjected to a 1024 point Fourier decomposition fromwhich auto spectra and power spectra from DC to25 Hz and a dynamic range of 48 dB were calculated.

In Parkinson's disease the typical resting tremorin a limb is 4 Hz and when the limb lifts into posturea tremor at 6 Hz becomes manifest; does this meanthat the tremor generating mechanism has speededup? We will argue that the two frequencies of tremorare generated by different mechanisms, but one canonly argue this on the basis of spectral analysis if strictcriteria are met. Firstly one must show that tremorat the two frequencies can occur together producingtwo harmonically unrelated peaks in the powerspectrum. Secondly, when this occurs it must be dueto the fact that the tremors run concurrently and donot alternate. As tremor of the hand involves severalmuscles, each of which may be supporting a differenttremor frequency and the accelerometer transducesthe net sum of these as if they were a single-mechanism, the presence of harmonically unrelatedpeaks in the power spectrum of a tremor is not itselfsufficient to make the conclusion that separate physio-logical mechanisms are involved. Accordingly incases of compound tremor we examined the activityin single muscles by electromyography to establish thatthe patterns of electrical activity also showed thatmore than one type of periodic activation was present.

536



http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from



Results

RESTING AND POSTURAL TREMORS

Amplitude of tremorTremor amplitude was simply characterised ascausing symptoms or not. In terms of amplitude,in the case of a 6 Hz tremor frequency, a peakacceleration of 0 01 to 0 1 g (g=981 cm s-2) weremoderately symptomatic. Acceleration amplitudesabove this range caused severe symptoms and ac-celerations below this range were not symptomatic.

Frequency analysis of tremorEach tremor was characterised in terms of itsdominant frequency and also in terms of har-monically unrelated peaks which were within8 dBs of the dominant peak in amplitude.Asymptomatic tremors were characterised onlyin terms of the dominant frequency present. Thefrequency parameters obtained in this way wereplotted in histogram form presented in fig 1. Thecolumns of the histogram were selected to be0-2 Hz wide. This resolution was chosen because,both within recording sessions and from session to

001012 -

rms J

-- 6T0 Hz 25 --

session, the frequency of a symptomatic tremorwas never seen to vary more than about 0 3 Hz.In fact, we have found the constant frequency ofpathological tremors in individual subjects to bea useful criterion to distinguish between normaland pathological tremor.

Resting tremorsAs illustrated in fig 1, the majority of symptomaticresting tremors in Parkinson's disease have peakfrequencies which lie between 4 and 5 3 Hz. Asmaller grouping lies between 5 8 and 6-8 Hz. Thelower frequency group is further divided betweena concentration around 4 to 46 Hz and a con-centration from 4-8 to 5 4 Hz. There is a distinctdiscontinuity in the tremor spectrum between 5-6and 5-8 Hz. The addition of asymptomatic andsubsidiary tremor frequencies to the histogramonly accentuates the frequency distribution al-ready determined by the symptomatic tremors.The tremors with frequencies from 4 to just above5 Hz are those which were clinically identifiable asthe classic resting tremors of Parkinson's disease(fig 2). The discontinuity in the histogram corres-

Hz 2!

] 0'19 0 0 1 9OigResting tremor Postural tremor

uzp_j nv-wu

Hz 25 0 Hz 25

]0019V1\p\vAA~'xf1O Q0ig ooig

4-5 + 6Hz tremor is 6Hz+ physiological tremor

Fig 2 A uto-spectra of resting and postural tremors of the hand in Parkinson'sdiseasc. The top left spectrum shows analysis of a fairly pure resting tremor.The tremor has a dominant frequency between 4 and 5 Hz, with 2nd, 3rd,4th and 5th harmonics. The top right spectrum is of pure postural tremor witha dominant frequency of 6 to 6 5 Hz. The bottom left spectrum is of a restingtremor with dominant frequencies at 4 and S Hz and 6 to 7 Hz. Theconcurrence of these two frequency components which-are-not harmonicallyrelated indicates that there are two coincident tremors. The bottom rightspectrum is of a very fine postural tremor, with two peaks, one at 6 Hz, theother at a higher frequency about 7 5 Hz which, by similar argument as before,indicates two separate tremor mechanisms. In this case the 6 Hz is probably thepathological postural tremor of the disease, whereas the higher frequencycomponent is the patient's normal physiological tremor.

0.002 -

ra.rnn i ~6

537



http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from



ponds to the division between what are, on appear-ance, typically resting and typically posturaltremors. That is, in some patients the typicalpostural tremor may also be evident in the restingposition.

Postural tremorsThe frequencies of postural tremors are pre-dominantly grouped around 6 to 6'2 Hz and theserepresent the fine typical postural tremor of thedisease (fig 2). The incidences of tremor around4 to 5 Hz reflect the observation that occasionallythe resting tremor of the disease may continuewhen the limb is raised in posture.Amongst our patients we observed a typical

symptomatic resting tremor in 31 limbs at rest andin 10 limbs on posture. Conversely the typical higherfrequency symptomatic postural tremor was ob-served in 25 limbs on posture and in 10 at rest. Wemay conclude from the comparable incidences ofthe two types of tremor in this disease that theyare almost equally characteristic of the condition.

Differentiation of resting, postural andphysiological tremor in power spectra and EMGAnalysis of a tremor commonly produced a powerspectrum with more than one peak, the peaksbeing harmonically unrelated. A resting tremorat 4-5 Hz sometimes also contained a frequencycomponent at 6 Hz; alternatively a posturaltremor might consist of components at both 6 Hzand 8-9 Hz (fig 2). This finding suggests that morethan one frequency of tremor was being supportedin the limb. EMG studies running concurrentlywith the tremor analysis revealed that the multipletremors could also arise from one muscle at thesame time and evidence for this is presented infig 3. The recordings were taken over the bellyof the extensor digitorum in the right forearmof one patient with the hand in rest and posture.In each condition there were at least two types ofEMG activity present. At rest there was a largeamplitude double spike packet at about 4 Hz witha smaller, more ragged higher frequency burst ofactivity which averaged 6 Hz. In posture therewere less regular large amplitude single spikedpackets with a single large spike at 6*7 Hz to-gether with a much smaller burst of activity atabout 8 to 9 Hz, which was interpreted as being"normal" physiological activity.Although interpretation of the EMG activity was

difficult on occasions, mixed types of activity werefound in individual wrist and hand muscles ofpatients who had tremors which decomposed into

150 ms

120ms

Posture 687Hz8.4Hz

MaxISI l7ms

4+ 42+I

250 ms

Ext

4HzResting tremor 6Hz

Fig 3 Surface EMG recordings over the belly ofextensor digitorium in right forearm of a patientwith idiopathic Parkinson's disease. Tremor waveforms were recorded with an accelerometer mountedon the dorsum of the hand. In posture there arelarge bursts of activity characterised by a dominantsingle spike at an average frequency of 6-7 Hztogether with smaller bursts of activity at an averagefrequency of 8 to 9 Hz. Spectral analysis of thetremor wave form also revealed similar distinctfrequencies. A t rest there are large double spikedpackets of activity at 4 Hz superimposed uponbackground bursts of activity at about 6 Hz. Whenbursts of different tremor activities are superimposedin this way it can be difficult to estimate individualfrequencies in which case a useful parameter isthe maximum inter-silent interval (Max ISI) whichcorresponds to the highest frequency tremor present.These results demonstrate that both at rest and inposture separate myogenic tremor mechanisms mayco-exist.

more than one dominant frequency. This findingsupports the hypothesis that the resting, sympto-matic postural and physiological tremors areseparate entities, generated by different mechan-isms. In the analysis of postural tremors, fivelimbs with symptomatic tremor and two withasymptomatic tremor had significant peaks at both6 and 8 to 9 Hz. Seven limbs with symptomatictremor and five with asymptomatic tremor hadsignificant peaks at 4 to 5 and 6 Hz. Two limbswith symptomatic tremor had peaks on the powerspectrum at all three frequencies. Amongst rest-ing tremors; one limb had symptomatic and fivehad asymptomatic tremor with peaks at 6 and8-9 Hz, four had symptomatic and six asympto-matic tremor with peaks at 4-5 and 6 Hz and four

W-"%-4

538

4.



http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from



limbs, all with symptomatic tremor, had peaks atall three frequencies.

A symptomatic tremorsThe distribution of peak frequency of asympto-matic tremor was similar to that of thesymptomatic tremors in many patients (fig 1),implying that tremor analysis could help in thedetection of early Parkinsonism in patients wherethere was no symptomatic tremor.

Incidence of symptomatic postural and restingtremorsRegardless of the dominant frequency oftremor, amongst the 40 patients whose results arepresented in fig 1, 44 limbs had "typical" sympto-matic resting tremor, and 44 limbs had "typical"symptomatic postural tremor. Thirteen limbs hadsymptomatic tremors which occurred with asimilar appearance in both rest and posture; thetremors were both of the 6 Hz postural type infour, and a further four were of the 4-5 Hz

A Cogwheeling

Extension LLt 1 cycle 2cycles

iFlexion

resting type; the remaining five showed both typesof tremor symptomatically. The concurrence ofthe classic symptomatic resting tremor ofParkinsonism with symptomatic postural tremorin the same limb was rare. Far more common wasthat one particular form of tremor, the 4-5 Hztype or the 6 Hz type, predominated in "appear-ance," and on spectral analysis was found to havea second frequency present corresponding to theother type of tremor.

COGWHEELING PHENOMENONWe adopted the following strict criteria which arecord of cogwheeling had to fulfil in order to besubject to frequency measurements. Firstly theexaminer had to "feel" the jerks of cogwheelrigidity occurring in a regular run, and at thesame time identify the corresponding parts of themovement recording on the oscilloscope or inkjet recorder. Secondly, this part of the recordthen was examined for similar wave forms whichrepeat without interruption, thus establishing

B 6Hz Tremor and cogwheelingcogwheeling

133ms

C Cogwheeling and 12Hz tremor

cogwheeling

% 82ms(12-13Hz)

1s

Fig 4 Examples of the most typical wave forms of cogwheel phenomenon.The recordings were taken with an accelerometer mounted on the dorsumof the hand, which was passively moved in flexion and extension. Upperright "A" illustrates a "sawtooth" shaped wave form; the rapid downstroke indicates that the strongest contraction is in the flexors. Later in thetrace, when two cycles of cogwheeling appear with similar wave forms, thephenomenon has the appearance of a transient tremor and can be measuredin terms of frequency (6 Hz). The upper right hand traces "B" illustrate asinusoidal cogwheel wave form which is well developed for several cycles.In this patient, throughout all posture and movement, a 6 Hz tremorpersisted. The cogwheel phenomenon appears to be a continuation andexaggeration of the tremor. The traces illustrated in "C" show a strongbeat of cogwheeling at 6-7 Hz occurring with a much faster transient tremorat 12-13 Hz, which may represent an apparent frequency doubling of thebasic tremor through enhancement of the second harmonic.

F

539



http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from


540

periodicity. Frequency measurements then weremade from the repetition rate of the wave form.Frequencies of individual samples of cogwheelingwere based on a mean taken over as many cyclesas possible. Cogwheeling which occurred at aharmonic of a lower frequency present in therecord was classified with the fundamental (fig 4C).Applying these criteria, we have encounteredwave forms which correspond to the clinicalappreciation of cogwheeling at two distinct fre-quencies, 6 Hz and 8 to 9 Hz, without overlap.

Cogwheel phenomenon at 6 HzFifteen of our patients cogwheeling which hadperiodic wave forms with frequencies between 6and 7 Hz. The mean frequency was 6-22 (SD)±0.137) Hz, indicating that the frequency distribu-tion was biased towards 6 Hz, reminiscent ofpostural symptomatic tremor. At times the waveform of cogwheeling appeared to be sinusoidal,much like an exaggeration of ongoing posturaltremor (at around 6 Hz) (fig 4B, C). At othertimes the wave form was more sawtooth in shape(fig 4A).

Cogwheel phenomenon at 8 to 9 Hz-"rippling"Eighteen of our patients exhibited cogwheelingwhich ranged in frequency between 7-5 and 9 Hz,with a mean of 8'6 (SD±0*7) Hz (fig 5). This typeof resistance to passive movement felt smootherand more regular than lower frequency cogwheel-ing, and invariably had a sinusoidal wave form;"rippling" describes fairly well how it felt. Cog-wheeling at 6 Hz and rippling at 8-9 Hz were not

Rippling

ext

Clonus

a a

1s


mutually exclusive phenomena; both occurred infour limbs in our patients in the same recordingsession. Cogwheeling or rippling evoked passivelydiffered little from that brought out actively bythe patient himself.

COGWHEEL PHENOMENON AND CLONUSIn six of our original survey of 40 patients, and in10 others examined subsequently, we found thatduring manipulations to evoke cogwheel phenom-ena, sustained stretching of the wrist evokeda vibration which was clinically indistinguishablefrom clonus. Moreover, on the occasions whenwe were able to record symptomatic posturaltremor, cogwheeling with regular wave form, andclonus in the same limb during a single recordingsession, all three phenomena were at the samefrequency with similar wave forms. Thesesimilarities are illustrated in the recordings of fig 6which were taken from a 56-year-old patient witha year long history of idiopathic Parkinson'sdisease. The resting tremor of his right hand hada frequency of 4-5 Hz. Postural tremor of the handwas at 6*3 Hz. There was also the cogwheelphenomena and an action tremor present at fre-quencies between 6 and 6-3 Hz, while wrist ex-tension produced clonus at 6-1 Hz. The cogwheelphenomena and clonus tended to be well de-veloped in patients with a symptomatic posturaltremor, and in all but one such patient cog-wheeling was at the same frequency as thetremor.

Patients with Parkinsonism and clonus of theupper limb did not have "pyramidal" signs, butwe did examine clonus in other diseases. Thepatients included a 17-year-old male with cervicalcord transection due to trauma, a 50-year-oldmale with cervical myelopathy, two adult males

Fig 5 "Rippling." An example of high frequencycogwheeling recorded from a patient with idiopathicParkinson's disease. The transient tremor is typicallysinusoidal. The lower traces are of clonus in extensionand pronation at the wrist recorded from a patientwith an extensive brain stem angioma. The clonus inextension-flexion is similar in wave form and frequencyto the rippling and the clonus in supination-pronationis similar to that of "sawtooth" shaped cogwheelingas illustrated in fig 3. The arrows indicate the pointat which the hand was manipulated by the examiner.These records were obtained using an accelerometermounted on the dorsum of the hand with itssensitive axis in the plane of movement. The samerelative gains are maintained throughout.



http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from



Passive cogwheeling Active cogwheeling

I I r-.

Postural tremor

~~~~~~- - clonus

ForcePressure

Resting tremor

Fig 6 Recordings of cogwheeling, active cogwheeling (action tremor),resting tremor, postural tremor and clonus taken from a patient withidiopathic Parkinson's disease. The records were all taken in the same

recording session within a few minutes of each other. The traces of theupper left show acceleration and displacement records of passivecogwheeling evoked with a frequency of 6 Hz. A few seconds later thepatient was asked to move his own hand in flexion and extension, as in therecords of the upper right, evoking a more sustained tremor at 6 Hz withsimilar wave form. The middle traces are of accelerometric recordings of thehand tremor in posture which goes into clonus as the patient presses down(hand being extended) on a force platform. The arrows indicate the pointat which he begins to press on the platform. The frequency of clonus is6 Hz. In the lowest trace the hand is at rest and develops a typical 4-3 Hzresting tremor. A constant relative gain on the accelerometer traces wasmaintained throughout.

2 4 6 8 10 14 18 22Hz

A

isExaggerated physiologicaltremorincortfcal inforct

B

2 4 6 8 10 14Hz

L.

Clanus upper imb in cervici

DB-66 -

Fig 7 Raw data records and_ corresponding power spectra

54- of a fine 9 Hz sinusoidaltremor, presumed to be anexaggeration of physiological

42- tremor in cortical infarct andof clonus at 6 Hz in a patientwith a cervical cord lesion. The

L....LL.I raw data records are of18 22 comparable amplitudes. The

arrows in "A " indicate the.~jEMIIiuI|uhl sections of the raw data record

lWIIIEI1RW which have a measurable'IIUI!' frequency corresponding to the

finely tuned peak of the powerIs spectrum.-al myelopothy

541

K



http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from


542

with a cerebral cortical infarcts and a 27-year-oldmale vwith a brain stem angioma. The first andlast patients had clonus on passive supination andpronation, and on flexion and extension of thewrist; the remaining patients had clonus on bothpassive stretch and during maintained arm posture.As with cogwheel phenomenon, we found twofrequenicies at which clonus occurred. The patientwith a cervical myelopathy had clonus at 6 Hz;the patient with a cortical infarct also had clonussomewhat less well sustained at 9 Hz (fig 7). Thepatient with an angioma had transient clonus inboth supination and pronation and flexion at9-2 Hz (fig 5), and the patient with a cord sectionhad transient clonus in pronation at 9 Hz. Thusthere was a parallel between the two forms of cog-wheelinig at 6 Hz and 8 to 9 Hz, and the twofrequencies at which we have demonstrated clonus.The greatest differences between the cogwheelphenomenon and clonus was that clonus almostinvariably had a more regular wave form and wasoften a more powerful tremor lasting longer (figs5 and 6).

Postural tremor of the hand in normal subjectsThe possibility that symptomatic postural tremorof the hand in people with Parkinson's diseaserepresented an exaggeration of normal physio-logical tremor led us to survey the characteristicsof postural tremor in normal adults. Ninety-fivenormal drug-free subjects were examined toexclude any significant neurological lesions andeach individual was subjected to six spectralanalyses of tremor of each hand in posture.Normal postural tremor of the hand in subjectsof all ages was often broadly tuned with severalpeaks of similar amplitude making characterisationin terms of a single frequency difficult. Moreoverthe overall shape of the spectra derived changedfrequently from one tremor sample to the next.In order to characterise the variability of tremorfrequency the following measurements wereadopted. For the six recordings from each handthe lowest tremor frequencies were identified.These were then combined and averaged over allhands and subjects in successive 10 year agegroups. The same was done separately for thehighest peak frequencies found. The means andstandard deviations thus calculated were plottedas a function of age group (fig 8). Variability andrange of dominant tremor frequency did notalter with age, although there was an overalldecline in frequency in the oldest age group. Therewas no evidence of a convergence on 6 Hz tremorwith age. These findings are in contrast to those


Hz10

9

81

7

61

45 -

4

I 4

{ 4 {{

+I I

15-20Age ranges

21-30 31-40 41-50 51-60 61-70 71-80

Fig 8 Means and standard deviations of the highestand lowest peak frequencies of tremor observedduring six recordings of each limb on hands of95 normal subjects plotted as a function of agegroupings. The data were averaged over both handsand all subjects in each age group. There is notendency for the variability of tremor frequency tochange with age and no tendency for convergenceon a 6 Hz tremor similar to the postural tremor ofParkinson's disease. There is a trend towardsdecreasing tremor frequencies with the most advancedage group.

in patients with Parkinson's disease in whom thepostural tremor was relatively invariant in fre-quency and most commonly at 6 Hz. Under cer-tain circumstances normal subjects may showtremor characteristics similar to those of basalganglia disease (fig 9). In fig 9A, an example of abroadly tuned postural hand tremor in a normalsubject is presented. Although one could say thatthe dominant tremor frequency lay between 8and 10 Hz, there are obviously many more com-ponents of significance. The spectrum of figure 8Bwas derived from the same subject after perform-ing vigorous isometric exercises with his hand andthen holding it in posture whilst forcibly contract-ing his forearm muscles. These manoeuvres pro-duced a more finely tuned tremor with a singledominant frequency at 6 Hz; there is also adistinguishable sub-peak at 4 Hz. The combinationof fatigue and rigidity in this normal subjectinduced a pattern of tremor similar in frequencyto the pathological tremors of basal gangliadisease.



http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from



., m.I I ...III .I aI. i.

2 4 6 8 10 14 18 22Hz

m I I IIIIIIII11 1 I l~I m1a Inia II

2 4 6 8 10 14 18 22Hz

A

7sFig 9 Raw data records and corresponding power spectra of tremor of thehand in posture recorded from a normal male subject aged 34 yr. "A "-Thepower spectrum of the tremor of normal posture is an example of a broadlytuned tremor. Although the peak value is clearly 10 Hz there are significantcomponents between 7 and 12 Hz making it difficult to characterise thetremor in terms of a single frequency. (The high power content at 2 Hzprobably represents postural adjustment.) "B"-The subject has maintainedisometric co-contraction of the hand muscles for several minutes. As shownin the raw records, the tremor has become more regular in appearance andthe power spectrum shows a finely tuned peak at 6 Hz. There is also a

significant subsidiary peak at 4-5 Hz. Postural adjustments at lowfrequencies have increased in amplitude and there is probably an increasedcardioballistic component. The raw data records are to scale.

Discussion

Different types of tremor in Parkinson's diseaseThe apparently diverse tremors of Parkinson'sdisease can be grouped together into three fre-quency bands, which for convenience will bereferred to as 4-5 Hz tremor, 6 Hz tremor, and8-9 Hz tremor. Tremors in these three fre-quency bands are separate entities as shown byspectral analysis and EMG studies in individualmuscles.The classical resting tremor of Parkinson's

disease had a fundamental frequency between 4and 5-3 Hz. The actual frequency in any indi-vidual patient varied only by 0-2 to 03 Hz. Onoccasioins, a typical tremor at rest persisted withsimilar amplitude and wave form when the armwas lifted into posture, then the frequency re-mained unchanged.

Several other Parkinsonian tremors could be

grouped in the 6 Hz frequency band. The finepostural tremor, in appearance quite differentfrom the resting, tremor which can extend intoposture, had a typical frequency between 6 and6-2 Hz. Intention tremor was an exaggeration ofthe postural tremor. A low frequency type ofcogwheel phenomena occurred with a typicalfrequency of 6-2 Hz. A low frequency type ofaction tremor which we refer to as active cog-wheeling, occurred at a frequency indistinguishablefrom that of the passive cogwheel phenomenon.Finally, in some patients a sustained stretch atclinical feel of clonus which was also at 6 Hz.

In the highest frequency band a few patientshad postural tremors in the higher frequencytype of cogwheel phenomena, and action tremor.Some also exhibited postural tremor between7 6 and 94 Hz. All patients who had activeand passive "cogwheeling" at 6 Hz also hadactive and passive "cogwheeling" at 8 to 9 Hz,

543



http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from


544

which to avoid confusion and to communicate theclinical feel of the phenomenon we refer to as

"rippling."Gordon Holmes in 1904 commented on the

similarity between the resting tremor of Parkin-son's disease and the resting, postural and intentiontremors of cerebello-rubal lesions.21 On the basisthat they shared common mesencephalic structuresand on the criterion of frequency, Holmes sug-

gested that they shared a common mechanism.In our experience using spectral analysis to analysecerebello-rubral tremors, we have found it im-possible to distinguish them from Parkinsonianresting tremor on the basis of frequency alone.This is true not only for the upper limb, but alsofor the head.19 20The tremor provided by sustained stretch in

Parkinson's disease gives a clue to the possibleorigin of the 6 Hz tremor complex. It is indis-tinguishable from clonus as felt in patients withcervical cord lesions, and we see no reason why itshould be referred to in any other terms. Yet it isfound in patients with Parkinson's disease withoutclinical evidence of "pyramidal" lesion. Thepossible similarities between the tremor producedby sustained stretch in Parkinson's disease andclonus in "pyramidal" lesions was first com-

mented upon by de Jong,9 and Denny-Brown3asserted that "sudden stretch of the finger flexorswill then, as also in the earlier stage of the disease,set up a series of the characteristic beats oftremor," a description which could be used torefer to the evocation of clonus.22 The mechanismresponsible for the generation of the rhythm ofclonus lies most probably in the spinal cord;22there is no apparent reason why the tremor ofsustained stretch in Parkinson's disease should not

be due to the same process. On the basis ofsimilarities of wave form and frequency, we believethat the other 6 Hz tremors of the disease alsoinvolve the same mechanism.

rremor in the region of 8-9 Hz occurs in a

number of diseases. We have observed sympto-

matic postural tremor in this frequency bandfollowing cortical infarcts, with the tremor becom-ing asymptomatic with clinical recovery. In one

case of mesencephalic angioma we observed a

clonus tremor produced by sustained fingerextension at a frequency of 9 Hz which was similarto the high frequency active rippling in Parkinson'sdisease. This frequency band is typical of normalphysiological tremor, which is determinedprimarily by the minimum recruitment rate ofmotor units.23 Tremors in this frequency range

may be caused by an exaggeration of normal


physiological mechanisms. In the case of the highfrequency clonus and rippling there must, inaddition, be an unusually high degree ofsynchronisation of motor units.

The identity of tremors at 6 Hz in Parkinson'sdiseasePostural tremor at its most typical, intentiontremor, low frequency active and passive cog-wheeling and clonic tremor on sustained stretchin patients with Parkinson's disease all have fre-quencies close to 6 Hz. This is only one featurewhich might suggest that they share a commonmechanism. In addition, when several of thesetremors occur in the same patient (for examplefig 5), they have similar wave forms, and one canblend into another without alteration inrhythmicity. There is less similarity in wave formsbetween patients; however, typical wave formswhich can be taken by cogwheel phenomena(figs 3, 5), one of which is sinusoidal whilst theother consists of a fast stroke in one directionfollowed by a slower return, are also characteristicof clonus. With respect to power and harmonicdistortion, the 6 Hz tremors lie on a continuum.Clonus is the most powerful of the tremors andhas the most fixed frequency, whereas posturaltremor is usually the weakest and has greatestharmonic distortion. Between these extremes, indescending order of power, lie action tremor,cogwheel phenomena and intention tremor. Thuswe have two main differences separating thesetremors, one being the way in which they areevoked and the other the degree of synchronisationand extent of recruitment and grouping of motorunits. These several apparently diverse tremulousphenomena may have a common mechanism deter-mining their rhythmicity. The tremogenic activityin each case is triggered in different ways; inclonus and cogwheel phenomena by peripheralinput consequent upon stretch; and in action, in-tention and postural tremor by descending nervousactivity either directly or as a consequence ofstretch. Nevertheless once established the tremorsare similar, differing largely only in magnitude andharmonic distortion.

Clinical utility of tremor analysis in Parkinson'sdiseaseThe common frequencies of symptomatic tremorin Parkinson's disease are equally 4-5 Hz and 6 Hz.When occurring symptomatically they are more orless mutually exclusive. However, when asympto-matic tremor is taken into account, 75% of ourpatients had either a symptomatic resting tremor



http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from



which on spectral analysis also had a smalleramplitude 6 Hz tremor either concurrent with theresting tremor or occurring in posture; or asymptomatic typical postural tremor at 6 Hz witha smaller asymptomatic 4-5 Hz tremor component.Such a combination of tremors at these twodistinct frequencies in the same limb we havefound only in basal ganglia disease, and accord-ingly consider this feature of the tremors of thedisease to be pathognomonic. In support of thisassertion, we have examined, using the sametechniques, more than 25 patients with multiplesclerosis in which tremor was a symptom, sixpatients with peripheral neuropathy associatedwith tremor, four patients with tremor of"hysterical" origin, and eight patients with cere-brovascular disease or cerebral neoplasm produc-ing tremor. Tremors were found at a variety offrequencies, but in these cases we have not foundthe combination of 4-5 and 6 Hz tremors.

In elderly people presenting with tremor, thefirm diagnosis of Parkinson's disease can be diffi-cult and here the correct diagnosis is particularlyimportant because of the possible serious sideeffects of anti-Parkinsonian drugs. If on spectralanalysis the combination of 4-5 and 6 Hz tremoris found (as occurred in 85% of our patients) thenstrong support is given to the diagnosis ofParkinson's disease. Conversely, when the diagnosisof basal ganglia disease is difficult, the inability todemonstrate two unrelated tremor frequencies inthe same limb suggests alternative diagnoses. Aclinical example of this situation is in the dif-ferential diagnosis of essential tremor occurringin elderly patients demonstrating a positiveFroment sign;24 that is cogwheel phenomenadetected in a limb only when the opposite limb ismoved voluntarily. Although this sign was origin-ally described as pathognomonic of Parkinson'sdisease, it is a frequent feature of uncomplicatedessential tremor, thus making the distinction some-times difficult.25

Relationship between normal and pathologicaltremorCertain features of tremor in normal people arerelevant to the analysis of tremors of Parkinson'sdisease. Our finding, in agreement with Marsdenet a126 is that tremors in the right and left handsare frequently quite dissimilar. Tremor in theasymptomatic hand of a hemi-Parkinsonian patientcannot be used as representative of what thepremorbid tremor would have been in thesymptomatic limb.

It has been suggested by Lance et al10 that the

action tremor of Parkinson's disease representsan exaggeration of normal physiological tremor.As the postural tremor, intention tremor andcogwheel phenomena of the disease are alsosimilar in frequency, the question may be put asto whether they are all consequences of heightenednormal physiological mechanisms. As shown,normal physiological tremor is a complex phenom-enon, often composed of several frequencies ofcomparable significance. There is a tendency innormal tremor for a modal frequency of about6 Hz with other modal frequencies between 7 and9 Hz. The 6 Hz tremors of Parkinson's diseasemay well represent an exaggeration of theparticular component of normal tremor whichcontributes the content at 6 Hz, but it would besimplistic and misleading to assert that they arean exaggeration of physiological tremor as if itwere a unitary mechanism.Further evidence that normal tremor has

various component mechanisms is that undervarious conditions one or another aspect may wellbe heightened and made symptomatic. Thus, forexample, tremor in the unstressed limb of anormal subject may be broadly tuned with signifi-cant components between 4 and 10 Hz, whereasshould the limb be stressed a more sharply tunedtremor at 6 Hz can appear (fig 9). Under pro-longed stress of co-contraction a violent tremordevelops: this, for example, is the commonphenomenon of tremor in the limb of the fatiguedand stressed rock climber; if occurring inpyramidal lesions it would be referred to as clonusand in Parkinson's disease "the tremor of sus-tained stretch." We believe that it is the sametremor in each case, the differences being onlyin the manner and circumstances of itsevocation.

We thank the physicians at the National Hospital,Queen Square and at St Mary's Hospital, Pad-dington, for allowing us to study the patientsunder their care.

References

1 Parkinson J. An essay on the Shaking Palsy.London: Whitingham Rowland 1817.

2 Selby G. Parkinson's disease. In Vinken PJ,Bruyn GW, eds. Handbook of Clinical NeurologyVol 6, Diseases of Basal Ganglia. Amsterdam:North Holland 1968: 173-211.

3 Denny-Brown D. Clinical symptomatology ofdiseases of the basal ganglia. In: Vinken PJ,Bruyn GW, eds. Handbook of Clinical NeurologyVol 6, Diseases of Basal Ganglia. Amsterdam:North Holland 1968: 133-172.

545



http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from


546

4 Struppler A, Erbel F, Velho F. Overview on thepathophysiology of Parkinsonian and otherpathological tremors. In: Desmedt JE, ed.Physiological Tremor, Pathological Tremor andClonus, Vol. 5, Progress in Clinical Neuro-physiology. Basel: Karger 1978: 114-128.

5 Rondot P, Bathien N. Pathophysiology ofParkinsonian tremor. In: Desmedt JE, ed.Physiological Tremor, Pathological Tremor andClonus Vol. 5, Progress in Clinical Neuro-physiology. Basel: Karger 1978: 138-149.

6 Aronson N. Neurophysiology of tremor.J Neurosurg 1966; 24:207-9.

7 Ehringer H, Hornykiewicz 0. Verteilung vonNoradrenalin und Dopamine (3-hydroxytyramin)im Gehirn des Menschen und ihr Verhalten beiErkrankungen des extrapyramidalen. KlinWeitschr 1960; 38:1236-9.

8 Lewy EF. Die Lehre vom Tonus und derBewegung. Heidelberg: Springer 1923.

10 Lance JW, Schwarb RS, Peterson EA. Actiontremor and the cogwheel phenomenon inParkinson's disease. Brain 1963; 86:95-110.

11 Negro C. Le Phenomena "de la Roue Dentee".Encephale 1928; 23:203-4.

12 Moyer H. A new diagnostic sign in paralysisagitans, the cogwheel resistance of the extremities.JAMA 191 1; 57:2125.

13 Matthews PBC, Rushworth G. Effect of Procaineon Stretch Reflex. J Physiol 1957; 135:245-64.

14 Matthews PBC, Rushworth G. The relativeSensitivity of Muscle Nerve Fibres to Procaine.J Physiol 1957; 135:65-9.

15 Denny-Brown D. Clinical symptomatology ofdiseases of the basal ganglia. In: Vinken PJ,Bruyn GW, eds. Handbook of Clinical NeurologyVol 6, Diseases of Basal Ganglia. Amsterdam:North Holland 1968: 133-72.

16 Stern J, Ward AA, Jr. The relationship of alphaand gamma motor systems to the efficiency of


the surgical therapy of Parkinsonism. J Neurosurg1963; 20:185-7.

17 Jasper HH. Discussion on the anatomical andphysiological aspects of Parkinson's disease. JNeurosurg 1966; 24:235-6.

18 Evarts EV. Neurophysiological mechanisms inParkinson's disease. In: Birkmayer W, Horny-kiewicz 0, eds. Advances in Parkinsonism. Basel:Roche 1976: 37-51.

19 Findley LJ, Gresty MA, Halmagyi GM. A noveltechnique for recording arm movements anda survey of common abnormalities in nervousdiseases. Arch Neurol 1980; 38:42.

20 Boshes B. Further insights into Parkinsoniantremor. In: Birkmayer W, Hornykiewcz 0, eds.Advances in Parkinsonism. Basel: Roche 1976:303-18.

21 Holmes G. On certain tremors in organic cere-bral lesions. BraSn 1904; 27:360-75.

22 Dimitrijevic MR, Nathan PW, Sherwood AM.Clonus: The role of central mechanisms. J Neurol,Neurosurg and Psychiatry 1980; 43:321-32.

23 Marsden CD. The mechanisms of physiologicaltremor and their significance for pathologicaltremors. In: Desmedt JA, ed. PhysiologicalTremor, Pathological Tremor and Clonus Vol. 5,Progress in Neurophysiology. Basel: Karger1978: 1-16.

24 Froment J, Gardere H. La rigidite et la rue dent6eparkinsonienne s'efacent au repos. Rev Neurol1926; 1:52-3.

25 Salisachs P. Dos Signos Clinicos no Conocidosdel Tremblor Essencial. Med Clin (Barcelona)1978; 70:120-1.

26 Marsden CD, Meadows JC, Lange GGW, WatsonRS. The relation between physiological handtremor of the two hands in healthy subjects.Electroencephalogr Clin Neurophysiol 1969; 27:179-85.



http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from


tremor, the cogwheelphenomenon parkinson's disease · the term "cogwheel phenomenon"...

Documents