3'ournal of Neurology, Neurosurgery, and Psychiatry 1993;56:393-399

Changes in sensation after nerve injury or

amputation: the role of central factors

S Braune, W Schady

AbstractDynamic changes in somatosensory cor-tical maps are known to occur in experi-mental animals subjected to peripheralnerve transection or amputation. Tostudy the sensory effects of central ner-vous system adaptation to temporary orpermanent loss of input from a part ofthe hand, multimodality quantitativesensory tests were carried out in 11patients with complete traumatic divi-sion and repair of the median or ulnarnerves and in six patients who hadundergone amputation of one or moredigits. As expected, vibration, two pointdiscrimination, and tactile thresholdswere raised in the territory of the injurednerve in a graded fashion, sensitivitybeing poorest in the patients with themost recent injuries. Surprisingly, locali-sation was better in the tips than at thebase of the hypoaesthetic fingers, sug-gesting a central attentional gradient.Stimulus-response curves conformed toa power function whose exponent washigher in denervated than in normalskin. Changes in psychophysical func-tions were also discernible in the intacthand. There was no hyperaesthesia in theterritory of the nerve adjacent to theinjured one or in the stump in the case ofamputees. Central factors contribute tothe sensory changes seen after nerveinjury, but the functional effects of thecortical reorganisation that follows par-tial deafferentation are more subtle thana simple heightening of sensitivity in thesurrounding skin.

(7 Neurol Neurosurg Psychiatry 1993;56:393-399)

Department ofNeurology,Manchester RoyalInfirmary,Manchester M13 9WLS BrauneW SchadyCorrespondence toDr W Schady, Departmentof Neurology, ManchesterRoyal Infirmary, OxfordRoad, Manchester,M13 9WL, UKReceived 6 January 1992and in revised form3 July 1992.

Accepted 29 July 1992

The body surface is represented somatotopi-cally in the mammalian somatosensory cor-

tex. It has been shown, however, that corticalmaps in rodents and monkeys can changeduring adult life in response to deafferenta-tion. Within weeks or months of upper limbnerve transection the region of cortex servingthe area of denervated skin, which initiallyfalls silent, is taken over by inputs from otherparts of the hand (for a review see Wall').

There is a wealth of information on thesensory events that follow nerve injury andregeneration in humans. Much less is knownabout the relative contribution of peripheraland central factors to functional recovery ofsensation. It is self evident that unless the

skin is reinnervated it will remain anaesthetic.As nerve fibres grow back to the skin from thesite of injury, possibly with some contributionfrom collateral sprouting,2 the patient's senso-ry experience will be determined by the num-ber and type of axons that establish functionalconnections with cutaneous receptors, bychanges in the encoding properties of regen-erated fibres, and by the brain's response tothe initial sensory deprivation and subsequentreturn of afferent activity.

If a similar reorganisation in cortical mapsoccurs in humans as in animals after partialdeafferentation, information from newly rein-nervated cutaneous receptors might not beprocessed normally because their original cor-tical representation had been "taken over" byinputs arising elsewhere. In addition, cuta-neous sensibility in the area surrounding theanaesthetic skin might be altered by itsincreased representation in the somatosensorycortex. The principal aim of our study was toassess the functional sensory effects of centralnervous system adaptation to temporary(nerve injury) or permanent (amputation)loss of input from a part of the hand.

Materials and methodsPATIENTSPatients who had suffered a traumatic upperlimb nerve transection or finger amputationand had been treated in the department ofplastic surgery at Withington Hospital wereapproached for participation in this study.Their informed consent was obtained inaccordance with the Declaration of Helsinki.

Nerve injuryWe studied 1 1 patients (6 men) with 12microsurgically repaired nerve injuries. Theirmean age was 35 years (range 17-73 years).The median nerve was affected in sixpatients, the ulnar nerve in five, and in onecase both nerves had been divided. All lesionswere at the wrist, with the exception of onepatient whose ulnar nerve had been damagedin the distal third of the forearm. The left sidewas affected in seven patients and the right infive, the damaged limb being the dominantone in half.

In nine patients the injury was caused byaccidental cuts from glass, one was due todamage by industrial machinery, and one wasself inflicted with a knife. Surgical exposureconfirmed total transection of the nerve inquestion in all cases. Epineurial repair wascarried out in nine instances and fascicular

393 on O

ctober 8, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.56.4.393 on 1 April 1993. D

ownloaded from

Braune, Schady

repair in three, performed within 24 hours ofinjury except for one patient whose nerve wasnot repaired until two months later. Themedian interval between nerve injury andtesting for the study was 10 (5-66) months.Patients were divided into those whose injurywas less than a year old (8 nerves; median 9-3months) and those whose injury was morethan a year old (4 nerves; median 17-8months). None of them had received formalsensory rehabilitation.

AmputationIn addition, we studied six patients who hadundergone amputation of one (4 cases) ortwo digits (2 cases). They were all male, witha mean age of 35 (17-53) years. Three of thesix fingers tested had been completely ampu-tated; in the other three the proximal phalanxremained. Machine accidents were responsi-ble in four patients. In the two others surgerywas performed electively, namely for removalof a lump in one and for a chronic pain syn-drome after a crush injury in the other. Thedominant hand was affected in four patients,the non-dominant hand in two. During theoperation the surgeon had stretched the digi-tal nerve endings and thermocoagulated thedistal 1-2 cm to prevent neuroma formation.The median interval between amputation andtesting was 7-3 (5 6-58) months.

MethodsA history of the nerve injury and its conse-quences was obtained in accordance with aset protocol. The affected limb was examinedin detail and the area of sensory deficit wascarefully mapped. Additional informationabout the initial clinical picture, the degree ofnerve division, the operative technique, andthe postoperative course was gained from thepatient's notes.

Multimodal quantitative sensory tests wereundertaken with the patient comfortably seat-ed and the tested hand resting on a sandbag.The right hand was always tested first. Threeareas were defined and tested separately, cor-responding in nerve injury patients to maxi-mally affected, partly numb, and unaffectedskin.Area A-In patients with nerve injuries the

area of maximal numbness consisted of thepulp of the index or the little finger, depend-ing on whether the median or the ulnar nervehad been divided. For amputees area A waslocated on the stump.Area B-In patients with nerve injuries the

area of partial numbness tested was on thevolar aspect of the proximal phalanx of thesame finger as for area A or, if the maximalsensory deficit extended to the base of thedigit, on the thenar or hypothenar eminence.In amputees area B was at the base of theaffected finger or, if digital amputation wascomplete, on the thenar or hypothenar emi-nence.Area C-For patients with nerve injuries

the area of unaffected skin tested was on thepulp of the little finger for median nervelesions and on the pulp of the index for ulnar

nerve lesions. For amputees area C wasalways on the pulp of the little finger, as allamputated digits were in the territory of themedian nerve.

In both groups of patients all tests wererepeated at the same sites in the unaffectedhand (areas CA, CB, and CC), so that eachpatient served as his or her own control.Some tests included additional locations, asdescribed below.

Thermal and pain thresholdsWarmth, cold, heat pain, and cold painthresholds were determined by the Marstockmethod.3 A Peltier thermode with a surfacearea of 12-5 cm2 was applied on the thenar orhypothenar eminence, depending on theaffected nerve. A weight of 240 g was placedon the thermode to ensure good contact. Thebaseline temperature was set at 310 C, whichthe subjects perceived as thermally indiffer-ent. The stimulating surface was thenwarmed 10 C a second until the subjects firstperceived warmth, when they pressed aswitch, thus returning the thermode tempera-ture to baseline. The procedure was repeateda minimum of four times with intervals ofabout 10 seconds between them. The ther-mode was then cooled and the same sequencewas used to obtain a cold threshold.

Thermal pain thresholds were measuredwith the same equipment, using a rate ofchange of 20 C/s for heat pain and 30 C/s forcold pain. Subjects were instructed to pressthe switch when they felt the thermode tohave become painfully hot or cold, dependingon the direction of the current. Five valueswere usually obtained for each thermalthreshold and the mean of the last three wascalculated.

Vibration thresholdsA vibrameter (Somedic AB) was used to mea-sure vibration thresholds in the pulp of thelittle and index fingers. The device delivers asine wave at 100 Hz through a hand heldprobe applied with a constant force, whichwas monitored by the investigator. Theamplitude of the sine wave was slowlyincreased until vibration was first perceived; itwas then reduced to obtain the vibration dis-appearance threshold. The mean of threeperception and disappearance thresholds wasused as a measure of sensitivity to vibration ateach testing site.

Tactile thresholdsTactile thresholds were obtained with a set ofSemmes-Weinstein monofilaments, a variantof von Frey hairs. They were recalibrated inour laboratory, the maximum force exertedbeing given in millinewtons (0-06-230 mN).The filaments were applied perpendicularlyto the skin in areas A, B, and C. The thresh-old was considered as the force required toobtain a correct response in four out of fivetrials.

Stimulus-response curvesThe relation between subjective sensory mag-nitude and stimulus intensity was studied fur-

394 on O

ctober 8, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.56.4.393 on 1 April 1993. D

ownloaded from

Changes in sensation after nerve injury or amputation: the role of centralfactors

ther by obtaining stimulus-response curves.4When the tactile threshold in area A had beenestablished seven supra threshold filamentsspanning the remaining range of the set wereselected. The middle filament was used as thereference. Pairs of stimuli were delivered byprodding the skin with the reference, followedabout a second later by another filament ran-domly chosen from among the seven original-ly selected. The second stimulus could thusbe weaker, stronger, or identical to the first.The subject was asked to rate on an openscale the intensity of the second stimulus inrelation to the reference, which was arbitrarilyrated as 10. Each filament was tested at leastfour times and the mean value was used toplot the subject's intensity rating against thestimulus force. The procedure was carriedout in areas A and B and at correspondingsites in the unaffected hand (areas CA andCB).

Two point discriminationTwo point discrimination was measured bymeans of calipers with blunt points applied ina longitudinal direction. The threshold wasconsidered to be the smallest distancebetween the points which the patient felt astwo stimuli in two out of three trials.

LocognosiaThe patient's ability to localise tactile stimuliwas tested with the technique introduced byNoordenbos.5 Subjects were given red gogglesto wear. While their eyes were closed thehand was touched briefly but firmly with ared felt-tip pen 1 mm in diameter at the tip.Having opened their eyes, but still wearingred goggles, subjects were unable to see thered dot on the skin. They were asked to markwith a black felt tip pen the spot where theythought they had been touched. The distancebetween the red and black marks was mea-sured and the direction of the error was docu-mented. The test was carried out on thedistal, middle, and proximal phalanges of theindex, ring, and little fingers (twice in the ringfinger-namely, along its ulnar and medianaspects) and on the thenar and hypothenareminences of both hands.

Data analysisSensory thresholds were not normally distrib-uted, and non-parametric statistics were thusused throughout. Comparisons were madebetween the median test values at the samesite on affected and non-affected hands andbetween several locations on the same limb.The Mann-Whitney U test was used forunpaired data and Wilcoxon's signed ranktest for paired data, with the SPSS-X statisti-cal package.

ResultsPATIENTS WITH NERVE INJURYClinical stateAll patients had regained some degree of feel-ing in the territory of the affected nerve. ATinel sign could be evoked over the scar in

seven patients. Only one patient had sponta-neous shooting pain from the site of injury tothe fingers supplied by the damaged nerve.Dysaethesia on gentle touch or spontaneousparaesthesia in the hypoaesthetic area, orboth, was recorded in seven patients. Motordeficits in the affected hand ranged fromMRC grades 2-3 in five cases to grades 3-4in three cases and grades 4-5 in four cases.With regard to autonomic function, one halfof the patients reported higher sensitivity tocold, leading to stiffness, occasional pain, anda colder feeling than in the unaffected hand.-Sweating was subjectively increased in theterritory of the damaged nerve in two cases.

Nerve conduction studiesOrthodromic motor and sensory nerve con-duction studies were performed in 11 of the12 injured nerves under investigation. Motoraction potentials were obtainable in all butone, but sensory potentials could be recordedin only three. By comparison with the valuesobtained in the contralateral unaffectednerves, the mean distal latency of the motoraction potentials was delayed by a factor of1-82, the mean distal and proximal ampli-tudes were reduced to 26% and 30%, respec-tively, and the mean nerve conductionvelocity in the forearm segment was slowed to82%. The mean distal latency of the threesensory action potentials obtainable wasdelayed by a factor of 1-4, while the meanamplitudes were only 24% of the values onthe contralateral side.

Sensory thresholdsIn the patient group as a whole, thermalthresholds and pain sensitivity were notaltered at the sites tested (the thenar emi-nence for median nerve injuries and thehypothenar area for ulnar lesions), whichwere hypoaesthetic for tactile stimuli. Themedian difference in thresholds between rightand left hands was <10 C for warmth, cold,and heat pain, and 1.60 C for cold pain. Thesubgroup of patients whose injury was lessthan 12 months old were hypersensitive topainful thermal stimuli on the affected side:the median difference with the healthy sidewas -2-1° C for heat pain (range -4-1 to- 10 C; p < 0-05) and -5.3° C for cold pain(- 6-6 to - 3-3° C; p = 0 05).

Vibration thresholds were higher in area A(the zone of maximal numbness) than at thecorresponding site CA in the opposite hand(median 1i1 gm in area A and 0 57 gm inarea CA; p = 0-033). Tactile sensitivityshowed the greatest alteration. The mediantactile thresholds at the fingertip (area A) andat the base of the finger (area B) were 18-6and 2-2 mN, respectively, both of which weresignificantly higher than the median values athomologous sites CA and CB in the soundlimb (table 1). The highest values wereobserved in patients whose nerve injurieswere recent (table 2).Two point discrimination was not record-

able from area A in seven patients and fromarea B in four patients. In the remainder,

395 on O

ctober 8, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.56.4.393 on 1 April 1993. D

ownloaded from

Braune, Schady

Table 1 Comparison between the affected and the non-affected limb for tactile and vibration thresholds, two pointdiscrimination and locognosia in 11 patients with 12 nerve injuries

Affected Median (range) Control Median (range)Modality site threshold site threshold p Value

Tactile threshold (mN) Area A 18-6 (1-9-146) Area CA 0-06 (0-06-1 9) 0-002Area B 2-2 (0-06-146) Area CB 0-06 (0 06-0 7) 0 011Area C 0-06 (0 06-054) Area CC 0 3 (0 06-0 54) NS

Vibration threshold (um) Area A 1-1 (0 08-29 4) Area CA 0 57 (0-17-5-1) 0 033Area C 0 47 (0-04-5-1) Area CC 0-31 (0-15-9 0) NS

Two point discrimination (mm) Area A* 17-0 (10-25) Area CA 4-0 (2-9) 0 043Area Bt 15-5 (6-42) Area CB 7-5 (4-14) 0-035Area C 5 5 (2-13) Area CC 4 0 (2-17) NS

Locognosia (mm) DP 5 0 (3-9) CDP 4 0 (1-10) NSMP 9-5 (4-29) CMP 6 0 (2-11) 0 032PP 11-5 (3-31) CPP 5 0 (0-12) 0 023T 8-0 (6-13) CT 6-0 (3-13) NS

DP = distal phalanges, MP = middle phalanges, PP = proximal phalanges, T = turnaround hypothena eminences.*Not recordable in 7 patients.tNot recordable in 4 patients.

median discrimination was significantlyincreased in sites A and B by comparisonwith their counterparts in the unaffected hand(table 1). For two point discrimination, as forvibration and tactile thresholds, the medianright-left difference for area C was zero.The ability to localise tactile stimuli (locog-

nosia) was impaired in the middle and proxi-mal phalanges of the fingers supplied by theinjured nerve. The largest error in localisationwas 31 mm. Misreferrals tended to be in alongitudinal direction and did not extend tothe cutaneous territory of a neighbouringnerve. There were no differences between thevalues obtained on the medial and lateralaspects of the ring finger. Surprisingly, locog-nosia was virtually normal in the tips of thefingers supplied by the injured nerve. Themedian difference between affected and unaf-fected sides at this location was 1 mm, with anarrower range of values than at any of theother sites tested (fig 1). Misreferrals were ofthe same order in the two subgroups ofpatients for the distal and middle phalanges.For the proximal phalanx locognosia was bet-ter in those whose repairs had been carriedout more than 12 months earlier (table 2).

Stimulus-response curvesThe stimulus-response curves resulting fromplotting the patient's subjective rating of theintensity of a range of tactile stimuli againststimulus force were best described as a power

Table 2 Differences in threshold between affected and unaffected limb for patients whosenerve injuries were less or more than a year old

Median (range) difference in threshold

Site Injury InjuryModality compared <12 months old >12 months old

Tactile threshold (mN) A-CA 49-9 (1-8-144) 11-3 (1-8-19)B-CB 8-9 (0-147) 1-1 (-0-5-6)C-CC 0 (-0-5-0) 0 (-05-0-6)

Vibration threshold (gim) A-CA 3-7 (-0-2-24) -0-3 (- 1-3-3)C-CC - 0-1 (- 3-9-0-5) 0 (-0-6-0-3)

Two point discrimination (mm) A-CA* 21-5 (21-22) 10-5 (8-13) + 2B-CBt 1 (0-35) 5 (- 1-21) + 1C-CC 0 (-42) 1 (0-4)

Locognosia (mm) DP-CDP 1 (- 2-8) 2 (- 1-4)MP-CMP 8-5 (-2-23) 8 (3-8)PP-CPP 10-5 (-9-26) 5 (-2-8)T-CT 2 (-7-6) 3 (-4-8)

DP = distal phalanges, MP = middle phalanges, PP = proximal phalanges, T = thenar andhypothenar eminences.*Not recordable in 5 patients.tNot recordable in 3 patients.

function conforming to a straight line on log-log coordinates (fig 2). The slope of this line(the exponent of the power function) was sig-nificantly steeper in the pulp of the fingerssupplied by the injured nerve (area A) than atthe same location in the other hand (areaCA) (median 068 v 038; p = 0 003). Forarea B the difference in slopes betweeninjured and healthy hands was less but stillsignificant (039 v 027; p = 0045).The slopes of the stimulus-response curves

at the tip and base of the fingers in the nor-mal hand also differed, being significantlysteeper at CA (038) than CB (027; p =0005). The median slopes tended to belower in the group of patients whose injurieswere more than 12 months old than thoseunder 12 months (A, 039 v 074); B, 040 v062; CA, 028 v 044; CB, 026 v 034), butthese differences were not statistically signifi-cant. In a group of seven age matched con-trols there were no differences between theslopes of stimulus-response curves at thesame locations (median slopes 041 at A, 040at CA, 044 at B, and 037 at CB).

PATIENTS WITH AMPUTATED FINGERSSensitivity in the stump (A) or the adjacentskin (B) was not significantly different fromsensitivity in the corresponding sites in theother hand (CA, CB), though small numbersmade meaningful statistical analysis difficult.Acuity for any of these tests was notimproved in the skin of the stump. Indeed,thresholds tended to be somewhat higher inthe affected hand (median tactile thresholds062 mN in A v 006 mN in CA and 06 mNin B v 03 mN in CB; median two point dis-crimination 10-5 mm in A v 4-5 mm in CAand 9*0 mm in B v 7 0 mm in CB).Locognosia was moderately impaired in thepalm a few centimetres proximal to the stump(location B; median 7-5 mm v 3 mm in theintact limb). Vibration, thermal, and painthresholds were similar in both hands.The slopes of the stimulus-response curves

obtained in the same way as for patients withnerve injuries did not show any side to sidedifferences. The slopes in areas A (049) andCA (048) were steeper than those in B(032) and CB (030), but these differencesdid not reach statistical significance.

396 on O

ctober 8, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.56.4.393 on 1 April 1993. D

ownloaded from

Changes in sensation after nerve injury or amputation: the role of centralfactors

gm

4 Vibration

1I

A C

25 -

20 -

15 -

105-

0-

mmLocognosia

C

Figure 1 Median thresholds and quartiles for touch, vibration, two point discriminaand locognosia at different sites in the hand after peripheral nerve regeneration.

DiscussionWe studied 11 patients with 12 complete,microsurgically repaired median or ulnarnerve transections. The interval betweennerve injury and testing of sensibility aver-

aged 10 months, so it can be assumed thatsome regeneration had occurred. To observethe effect of permanent deprivation of sensoryinput we also studied six patients who hadundergone amputation of one or more fin-gers. Standard methods were used to assess

sensibility.67 All techniques of quantitativesensory testing have the inherent problem of

Figure 2 Stimulus-response function in a

patient with a right ulnarnerve lesion. The force ofthe tactile stimulus isplotted against the subject'smean rating of intensity inseveral trials. Circlescorrespond to valuesobtainedfrom the pulp ofthe righ littlefinger (siteA) and crosses correspondto values from the intactleft little finger (site CA).

2-

0)

c

cccc

0-j

1-

0

0

I 22

Log Force ( mN )

intrasubject and intersubject variability, so itwas reassuring to find in our patients that forhealthy skin the median right-left differencein touch, two point discrimination, and vibra-tion thresholds was exactly zero. There wasthus no effect on these results of dominanceor order of testing.Some of the findings in area A (area of

maximal numbness) and area B (partialnumbness) in our patients with nerve injuriescan be ascribed to peripheral factors. Asexpected, touch was impaired in areas A andB in a graded fashion. Similar proximo-distalgradation in mechanical thresholds afternerve suture has been noted by manyauthors.8-'0 It is of interest that sensitivity wasbetter in the group of patients whose lesionswere over a year old, indicating ongoingregeneration of nerve fibres and establish-ment of adequate connections with cutaneousreceptors.' 1-13

Vibration thresholds were also raised inarea A, though less so than tactile thresholds,probably because of propagation of the stimu-lus outside the hypoaesthetic zone. Paciniancorpuscles, which are responsible for encod-ing high frequency vibration, are known tohave very low firing thresholds and largereceptive fields.'4 Deep afferents (for instance,those in the interossei in the case of mediannerve lesions) may also have been activatedby the vibratory stimulus. It is unlikely thatreinnervated cutaneous Pacinian corpusclescontributed much to vibration sensitivity inour patients, as few if any units of these cor-puscles make successful connections withreceptors of this class,'5 though there may besome cross reinnervation of Pacinian recep-tors by sensory axons previously specialiseddifferently.'6 In a recently published case offorearm amputation and replantation, wherepropagation of the stimulus to the intact skincannot have occurred, vibration sensitivitywas the only submodality to be totally absentthree and a halfyears after the accident.'7

Thermal thresholds were not altered in ourpatients, perhaps because they were tested inthe palm of the hand 5-66 months after nerverepair, by which time large numbers of regen-erating unmyelinated and small myelinatedfibres should have covered the short distancebetween the site of injury and the thenar orhypothenar eminence.'8 However, hyperalge-sia persisted in these areas during the first 12months, a phenomenon described by Headand Sherren in 1905'9 and subsequently con-firmed by others.8202' It has variously beenthought to be due to central release frominhibitory influences,22 lowered thresholds inregenerating polymodal nociceptors,'8 orectopic discharges generated in the injurednerve.23Two point discrimination is one of the old-

est tests of cutaneous sensibility. It has thedrawback of being static, whereas tactilerecognition usually involves exploration bythe moving finger. Nevertheless, defectivetwo point discrimination is a useful pointerto imperfect reinnervation at a time whenordinary threshold tests may have normal-

mN100, Touch

11

0.A1 B

A B C

397

mm25 2PD201510 -T

5 i I

0

T

on October 8, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.56.4.393 on 1 A

pril 1993. Dow

nloaded from

Braune, Schady

ised,"24-26 in our patients with nerve repair. Itwas unrecordable in over half (understand-ably so since most injuries were relativelyrecent) and was impaired at sites A and B inthe remainder. We also found somewhatdefective discrimination and touch in thestump of patients who had undergone ampu-tation of a finger. This is probably explainedby distortion of the normal anatomical rela-tionships between the skin flaps used to coverthe stump, though central factors may also beinvolved.

Localisation of touch (locognosia) is amore functional task than threshold tests andmay be a better predictor of outcome afternerve injury than two point discrimination.27Tactile stimuli are misreferred long afternerve fibre regeneration has occurred, possi-bly because of misdirection of outgrowingaxons, with persisting abnormalities in thetopographic organisation of the central ner-vous system.28-3' It was thus unexpected tofind normal locognosia in the fingertips of ournerve injured patients, although elsewhere inthe finger localisation was significantlyimpaired. Dyck et al reported normal primarylocalisation when testing the fingers ofpatients several years after median nerveinjury, but the perceptual territory wasincreased.32 Other authors have not testedseveral areas of skin systematically with punc-tate stimuli. The fact that in our patients thefingertips were the first to recover normallocalisational capacity, when they should bethe least endowed with reinnervatedmechanoreceptors, indicates that central fac-tors are at play. For the purpose of localisinga tactile stimulus, preferential attention ispaid to information from the fingertips, how-ever scanty, because of its greater relevance inactive touch. It has been shown that in thefingertips of healthy subjects input from a sin-gle mechanoreceptive unit may suffice foraccurate point localisation.33We did not observe any misreferral of tac-

tile sensations from the healthy skin to thecutaneous territory of the injured nerve, forinstance in the ring finger. This might havebeen expected if collateral sprouting of nervefibres from adjacent normally innervated skinhad occurred. However, such sprouting isrestricted in experimental animals to highthreshold afferents, which mediate perceptionof pain and temperature.'03435 Our thermalstimulator could be used only in the palmand was too large to carry out thermal locali-sation tests.The subject's rating of the intensity of vari-

ous tactile stimuli is a psychophysical func-tion that depends on the recruitment ofsensory units but probably also on centralfactors. Our finldings indicate that temporaryor permanent deprivation of peripheral inputcauses the power function to become steep-er-that is, sensory magnitude grows morerapidly with increasing stimulus force: stimu-lus-response curves in our patients weresteeper in the hypoaesthetic fingers (areas Aand B) than in the healthy skin of the oppo-site hand (areas CA and CB), and they tend-

ed to be steeper in the first 12 months afterinjury than later on, when presumably inner-vation was richer. Similar findings werereported by Franzen and Lindblom.36 It is asif the central nervous system were trying tocompensate for the loss of peripheral input byassigning higher intensity ratings tosuprathreshold stimuli. Our normal controlsshowed no difference in the slope of the stim-ulus-response function between tip and baseof the fingers.

Another finding that cannot be explainedon a peripheral basis is the steeper slope ofthe psychophysical intensity function in thefingertip than the finger base in the normalhand of our nerve injured patients. In addi-tion, our data in amputees showed a trendtowards steeper slopes both in the stump andat the corresponding contralateral location, inthe absence of tactile threshold changes. Thismay indicate a general resetting of theresponsiveness of the brain to suprathresholdtactile cutaneous stimuli, even those from theother side of the body, when there is loss ofinput from one finger.

Nevertheless, the brain does not respondby heightening the sensitivity of the skin tojust noticeable stimuli in the territory of thenerve adjacent to the injured one or in thestump in amputees. Multimodality thresholdsin these areas were not significantly differentfrom those at corresponding sites in the intactlimb. This is in contrast with Haber's findingof improved tactile acuity in stumps of aboveelbow amputations.37 The different amount ofamputated tissue may account for the differ-ence between Haber's and our results.

In conclusion, our findings support the roleof central factors in certain judgments ofintensity and localisation of a cutaneous tac-tile stimulus after sensory deprivation. Thecortical reorganisation that follows digitamputation or nerve section in monkeys3' 38might be anticipated to cause hypersensitivityin the adjacent skin, owing to the improved"grain" of its somatosensory representation.This is probably too simplistic a view of thecentral nervous system's adaptation to loss ofafferent input. Subtle changes taking place atcortical or subcortical level may be aimed atmaking the best use of the sparse or distortedinput from the partially deafferented area ofskin. Dynamic changes in the central nervoussystem after nerve repair and regenerationmay form the basis for successful sensoryrehabilitation in such patients.39

This study was supported by a grant from the Hanns SeidelStiftung. We thank Mr S Watson for allowing us to studypatients under his care and Dr R Schmidt and Professor H ETorebjork for help in the design of the study protocol.

1 Wall JT. Variable organization in cortical maps of the skinas an indication of the lifelong adaptive capacities of cir-cuits in the mammalian brain. TINS 1988j11:549-57.

2 Inbal R, Rousso M, Ashur H, Wall PD, Devor M.Collateral sprouting in skin and sensory recovery afternerve injury in man. Pain 1988;28:141-54.

3 Fruhstorfer H, Lindblom U, Schmidt WG. Method forquantitative estimation of thermal thresholds inpatients. 7 Neurol Neurosurg Psychiatry 1976;39: 1071-5.

4 Stevens SS. Sensory power functions and neural events.In: Loewenstein WR, ed. Handbook of sensory physiologyVol 1. Berlin: Springer, 1971: 226-42.

398 on O

ctober 8, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.56.4.393 on 1 April 1993. D

ownloaded from

Changes in sensation after nerve injury or amputation: the role of centralfactors

5 Noordenbos W. The sensory stimulus and the verbaliza-tion of the response: the pain problem. In: Somjen JJ,ed. Neurophysiology studied in man. Amsterdam:Excerpta Medica, 1972:207-14.

6 Lindblom U. Quantitative testing of sensibility includingpain. In: Stalberg E, Young RR, eds. Clinical neurophysi-ology London: Butterworth, 1981:168-90.

7 Mitchell S, Schady W. Somatic sensation in the elderly.In: Thomas PK, ed. Peripheral nerve changes in theelderly. Chichester: Wiley, 1988:157-64.

8 Trotter W, Davies HM. Experimental studies in the inner-vation of the skin. 7 Physiol 1909;38: 134-246.

9 Seddon HJ. Surgical disorders of the peripheral nerves. 2nded. Edinburgh: Churchill Livingstone, 1975.

10 Kingery WS, Vallin JA. The development of chronicmechanical hyperalgesia, autotomy and collateralsprouting following sciatic nerve section in rat. Pain1989;38:321-32.

11 Burgess PR, Horch KW. Specific regeneration of cuta-neous fibers in the cat. J7 Neurophysiol 1973;36:101-14.

12 Horch K. Guidance of regrowing sensory axons after cuta-neous nerve lesions in the cat. Neurophysiol1979;42: 1437-49.

13 Terzis J K, Dykes R W. Reinnervation of glabrous skin inbaboons: properties of cutaneous mechanoreceptorssubsequent to nerve transection. Neurophysiol1980;44:1214-25.

14 Johansson R S. Tactile sensibility in the human hand:receptive field characteristics of mechanoreceptive unitsin the glabrous skin area. Physiol 1978;281:101-23.

15 Mackel R, Kunesch E, Waldhor F, Struppler A.Reinnervation of mechanoreceptors in the humanglabrous skin following peripheral nerve repair. BrainRes 1983;268:49-65.

16 Schiff J, Loewenstein W R. Development of a receptor ona foreign nerve fiber in a pacinian corpuscle. Science1972;177:712-5.

17 Krarup C, Upton J, Creager M A. Nerve regeneration andreinnervation after limb amputation and replantation:clinical and physiological findings. Muscle Nerve1990;13:291-304.

18 Shea V K, Perl E R. Regeneration of cutaneous afferentunmyelinated (C) fibers after transection. Neurophysiol1985;54:502-12.

19 Head H, Sherren J. The consequences of injury to theperipheral nerves in man. Brain 1905;28: 116-338.

20 Sunderland S, Kelly M. The painful sequelae of injuries toperipheral nerves. AustN ZJ Surg 1948;18:75-118.

21 Denny-Brown D. The release of deep pain by nerveinjury. Brain 1965;88:725-38.

22 Boring E G. Cutaneous sensation after nerve division.Q Exp Physiol 1916;10: 1-95.

23 Ochoa J. Pain in local nerve lesions. In: Culp W J, OchoaJ, eds. Abnormal nerves and muscles as impulse generators.New York and Oxford: Oxford University Press, 1982:568-87.

24 Onne L. Recovery of sensibility and sudomotor activity inthe hand after nerve suture. Acta Chir Scand1962;300(suppl): 1-69.

25 Kankaanpaa U, Bakalim G. Peripheral nerve injuries ofthe upper extremity. Sensory return of 137 neurorrha-phies. Acta Orthop Scand 1976;47:41-5.

26 Wynn Parry C B, Salter M. Sensory reeducation aftermedian nerve lesions. Hand 1976;8:250-7.

27 Marsh D. The validation of measures of outcome follow-ing suture of divided peripheral nerves supplying thehand. Hand Surg [Br] 1990;15:25-34.

28 Hawkins G L. Faulty sensory localization in nerve regen-eration. An index of functional recovery followingsuture. Neurosurg 1948;5:1 1-8.

29 Hallin R G, Wiesenfeld Z, Lindblom U. Neuro-physiological studies on patients with sutured mediannerves: faulty sensory localization after nerve regenera-tion and its physiological correlates. Exp Neurol1981;73:90-106.

30 Mackel R, Brink E E, Wittkowsky G. Properties ofcutaneous mechanosensitive afferents during the earlystages of regeneration in man. Brain Res 1985;329:49-69.

31 Wall J T, Kaas J H, Sur M, Nelson R J, Felleman D J,Merzenich M M. Functional reorganization insomatosensory cortical areas 3b and 1 of adult monkeysafter median nerve repair: possible relationships tosensory recovery in humans. Y Neuroscience 1986;6:218-32.

32 Dyck P J, Lambert E H, Wood M B, Linscheid R L.Assessment of nerve regeneration and adaptation aftermedian nerve reconnection and digital neurovascularflap transfer. Neurology 1988;38:1586-91.

33 Schady W, Torebiork H E, Ochoa J L. Cerebral localiza-tion function from the input of single mechanoreceptiveunits in man. Acta Physiol Scand 1983;119:277-85.

34 Devor M, Schonfeld D, Seltzer Z, Wall P D. Two modesof cutaneous reinnervation following peripheral nerveinjury. Comp Neur 1979;185:211-20.

35 Brenan A. Collateral reinnervation of skin by C fibers fol-lowing nerve injury in the rat. Brain Res 1986;385:152-5.

36 Franzen 0, Lindblom U. Tactile intensity functions inpatients with sutured peripheral nerve. In: Zotterman,ed. Sensory functions of the skin in primates. Oxford:Pergamon, 1976: 113-8.

37 Haber W B. Effects of loss of limb on sensory function.JPsychol 1955;40:115-23.

38 Merzenich M M, Nelson R J, Stryker M P, Cynader M S,Schoppmann A, Zook J M. Somatosensory cortical mapchanges following digit amputation in adult monkeys.Comp Neurol 1984;224:591-605.

39 Dellon A L, Curtis R M, Egerton M T. Reeducation ofsensation in the hand after nerve injury and repair.Plastic Reconstr Surg 1974;53:297-305.

399 on O

ctober 8, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.56.4.393 on 1 April 1993. D

ownloaded from