INTERNATIONAL JOURNAL OF LEPROSY^ Volume 63, Number 3

Printed in the U.S.A.

Electrophysiological Correlates of Hanseniasis'Angarai G. Ramakrishnan and Thaiyar M. Srinivasan 2

Clinical neurophysiological studies sig-nificantly help in the diagnosis and assess-ment of various neuropathies. Hanseniasis(leprosy) primarily affects the peripheralnervous system and secondarily involvesskin and certain other tissues ( 8 ). The nervesmay be involved at any level, from the pe-ripheral cutaneous nerve twigs to the dorsalroot ganglia. The basis of the lesion in theperipheral nerves is the bacillus' neurotrop-ism ( 2 ). Leprous infection entails both seg-mental demyelination and axonal degen-eration (Srivivas, K. and Ramanujam, K.Hansen's disease and the neurologist. An-nual Meeting of the Neurological Society ofIndia, Vishakapatnam, 1981). Demyelin-ating neuropathies cause reduction in themaximum conduction velocities, and ax-onal polyneuropathies result in reduction ofsensory or motor response amplitudes (9 ).Thus, in the case of leprosy, one expectsreduction in the amplitude of the compoundnerve action potentials (CNAP)* in addi-tion to slowed conduction.

' Received for publication on 7 March 1994; ac-cepted for publication in revised form on 2 June 1995.

A. G. Ramakrishnan, Ph.D., Assistant Professor,Department of Electrical Engineering, Indian Instituteof Science, Bangalore 560012, India. T. M. Srinivasan,Ph.D., Gladys McGarey Foundation, Scottsdale, Ari-zona 85252-3421, U.S.A.

* List of symbols and abbreviations: SNAP = sen-sory nerve action potential; CNAP = compound nerveaction potential; EMG = electromyography; NCV =nerve conduction velocity; NCT = nerve conductiontime; S = motor threshold of stimulation; S.D. = stan-dard deviation; N/S = not significant; N3 = absolutelatency of first major negative peak in sensory nerveaction potential recorded from the third digit; P4 =absolute latency of positive peak following first majornegative peak in the SNAP recorded from the thirddigit; N5 = absolute latency of first major negative peakin the CNAP recorded from the elbow; P6 = absolutelatency of positive peak following first major negativepeak in the CNAP recorded from the elbow; N9 =absolute latency of first major negative peak in theCNAP recorded from the Erb's point; P I 1 = absolutelatency of positive peak following first major negativepeak in the CNAP recorded from the Erb's point; V,= conduction velocity of palm segment of median nerve;V,„ = conduction velocity of forearm segment of me-dian nerve; V, = conduction velocity of arm segment

Reduction in motor nerve conduction ve-locity (NCV) ( 1,4,6, 10. 13, 14 ) and changes inelectromyography (EMG) (I 7, 12 ) in leprosyhave been reported by many in the litera-ture. However, studies on sensory conduc-tion have been few (3,5, '°). Since it is knownthat the sensory nerves are the first to beaffected in leprosy ( 2 ), it is natural to lookfor correlates in sensory nerve conductionparameters. Thus, it is surprising that mostearlier electrophysiological studies in lep-rosy have concentrated on the conductionvelocity in motor fibers and EMG.

The work reported here involved thestudy, comparison and classification of pe-ripheral sensory nerve action potentials(SNAPs) and compound nerve action po-tentials (CNAPs) obtained by stimulationof the median nerves of normal individualsand leprosy patients. The purpose of thestudy was primarily to examine peripheralsensory nerve conduction in both clinicallynormal and abnormal nerves of patients inorder to identify the time domain param-eters which are significantly affected and tolook for early electrophysiological changesin clinically normal nerves. A comprehen-sive approach is made in observing all rel-evant variables, namely sensory nerve con-duction velocity (NCV), motor threshold ofstimulation (S), peak-to-peak amplitudes,and the absolute and relative latencies ofthe dominant peaks of the CNAPs. In ad-dition to the comparison of the mean andvariance values of the various variables, itis very useful to obtain an objective measureof effectiveness for each parameter in dif-ferentiating normal from abnormal poten-tials. This discriminating power of different

of median nerve; A, = peak-to-peak amplitude of theSNAP recorded from the third digit; A, = peak-to-peak amplitude of the CNAP recorded from the elbow,medial to brachial artery; Ah = peak-to-peak amplitudeof the CNAP recorded from the Erb's point (brachialplexus); D, = average discriminant score for class I(normal subjects); D2 = average discriminant score forclass 2 (clinically affected nerves of patients); D* =critical score used for classifying any data into one ofthe groups; D 2 = Mahalabobis' generalized distancebetween the two groups.

395

Active - 2.5 onposterior to Cz,7 cm from midline

Supraclavicularfossa (hidden)

Elbow recordingpoints

Ground(dorsal side offorearm or hand)

Median nerve-stimulation points

Recording points(third digit)

Reference - Fz

396^ International Journal of Leprosy^ 1995

parameters was obtained by a multivariateanalysis and verified by means of correla-tion with clinical observations.

Since all leprosy is neuropathy ( 2 ), andalso because the major motivation behindthis study is to look for early neurophysi-ological indications before the appearanceof clinical symptoms, the data were delib-erately not grouped as those from tuber-culoid, lepromatous, or other forms of lep-rosy. This approach has a certain validitysince there is some involvement of periph-eral nerves in every clinical form of leprosy.Thus, the general significance and the com-mon trends and extent of neuropathy couldbe brought out, as one can see clearly fromthe results of this study. Further, the authorsfeel that if the data were classified into dif-ferent clinical groups, the sample size of eachgroup would become too small to have anystatistical significance. Thus, the results ofthis study must be interpreted as generalevaluations of neuropathic abnormalities inleprosy, rather than that of any particularclinical form of leprosy.

MATERIALS AND METHODSAn IBM PC/XT compatible computer has

been converted into a dedicated system foracquisition, analysis, and classification ofevoked potentials by the addition of appro-priate hardware and software. A completelyprogrammable, two-channel, instrumenta-tion amplifier has been designed and usedto pick up the biopotentials. Employing thissystem, the median nerves of the subjectswere stimulated percutaneously at the wrist,between the tendons flexor carpi radialis andthe palmaris longus at the distal crease. Re-sponses were recorded from two ortho-dromal locations and one antidromal site.Figure 1 illustrates the locations of the stim-ulating and recording electrodes. Responseswere recorded from a) the palmar side ofthird digit, b) just proximal to the elbowcrease, medial to the brachial artery and c)the supraclavicular fossa over the brachialplexus at Erb's point. In addition, somato-sensory evoked potentials were recordedfrom the contralateral cortex. The groundelectrode was always positioned somewherebetween the stimulus and the recording sites.The electrode-skin interface impedance wasalways kept below 10 Kohms. The Erb'spoint and the cortical potentials were ac-

C z

FIG. I. Stimulation and recording sites used in thestudy.

quired with the midline forehead (FZ) asreference.

The study involved 25 normal subjectsand 21 patients, both groups reasonablymatched by age and sex. The normal sub-jects were drawn from a healthy population,and the patients were from the Central Lep-rosy Teaching and Research Institute,Chenglepet, South India. Of the normal in-dividuals, 23 of them were in the age group19-31 with the mean age of 24.9 years anda mean height of 163.9 cm. None of themhad a history of neurological problems. Thestimulus intensity was increased from a verylow value and maintained at a value to effecta minimal thumb twitch. Square wave puls-es of 100 At scc duration were applied at arate of 1 per second to identify the stimulusstrength required. The intensity and thestimulus electrode position were adjusted toeffect a minimal thumb twitch at minimalvoltage. The stimulus threshold having beenidentified, averaging was started at 2 stimuliper second. With such low voltages, repet-itive stimulation at 2 per second is not pain-ful or uncomfortable.

The recording was done with an inter-stimulus interval of 510 msec, which en-sures that the supply frequency (50 Hz)noise, if present, is out of phase in succes-sive sweeps so that it cancels itself. A filterpassband of 10-3000 Hz (3 dB) was usedfor all the recordings. Whenever excessiveartifact was present, that particular sweep

—25— N5

— 20 —

t.-15

a CSNELB1

—40— ̂ 3

—30 —

a

CSNDIGI—10

NCT10—

20—

6^110^12^1 14^16^16 20TIME IN 14SEC

b

so(a)

2 4P4

-a-

0

5

10 (b)

0^1 61 1 1 110 12 14 16 la 20TN( IN LISEC

4NCT

4^6

C119

CSNEJt8 I

^

NCT I^PI I

4^6^a^10^12^14^16^16^201THEN LISEC

—11 —

— 6—

—

1 -

3(a)

63, 3^Ramakrishnan and Srinivasans Electrophysiological Correlates^397

FIG. 2. Set of potentials recorded from a normalsubject following median nerve stimulation at the wrist.a = SNAP from the third digit; b = CNAP from theelbow; c = EP from the Erb's point.

of data was automatically rejected by theprogram. The sweep of the averaging com-puter was started immediately after the de-livery of the stimulus with zero delay. Six-teen sweeps were averaged in the case ofnerve conduction potentials at the digit andthe elbow and 64 sweeps in the case of Erb'spoint potential. Activity was recorded for20 msec at the rate of 16.66 KHz (samplesat 60 msec equal intervals). Potentials ateach point were recorded twice by perform-ing two complete trials to ensure the cor-

rectness and repeatability of the recordedwaveforms. For the normal subjects, re-sponses were obtained only from any oneof the upper extremities. The recordingsfrom normal subjects were done in the air-conditioned laboratory where the temper-ature was maintained nearly constant at27°C.

For patient data collection, the entireevoked potential averaging system wasmoved to the leprosy hospital. Twenty-oneinpatients were studied to investigate theextent of neurological damage. Sixteen ofthem were in the age group of 20-33 (mean25.1 years, mean height 160.0 cm). Most ofthem had a disease history of 2 years ormore (mean 5.7 years, range 1 to 15 years),and a number of them had been taking drugsrecommended by the hospital for a mini-mum period of 1 year (mean 3.4 years, range0.3 to 10 years). During the collection of thedata from patients on various days, the roomtemperature varied between 29°C and 32°C.None of the patients studied had fever orany other health problem which affectedtheir body temperature. Many of the pa-tients were affected only unilaterally. How-ever, responses were obtained from bothupper extremities in order to electrophysi-°logically assess the clinically normal me-dian nerves of the patients.

In order to explain the nomenclature used,a set of potentials recorded from a normalsubject is presented in Figure 2. Here, anupright deflection indicates a negative po-tential and a downward deflection indicatesa positive potential. By convention, theevoked potential peaks are named by theirpolarities and latencies. Thus, the negativebrachial plexus peak which occurs around9 to 10 msec is known usually as N9 andthe following positive wave is referred to asP 1 1 (Fig. 2c). Although there are no suchcommonly accepted terms for the periph-eral responses, the same convention has beenused by the authors in naming these poten-tials. Thus, the first major negative peak inthe SNAP from the third digit (Fig. 2a),whose latency was found to be around 3msec for normal individuals, has beentermed N3, "N" denoting the negative po-larity of the peak and "3" indicating theapproximate latency at which it normallyoccurs. Similarly, the major positive peakwhich follows N3 in the digit response has

398^ International Journal of Leprosy^ 1995

been named as "P4," showing that the av-erage normal latency of this peak (Fig. 2a)is around 4 msec. In an analogous manner,the major negative peak and the followingpositive peak in the responses from the el-bow have been designated "N5" and "P6,"respectively (Fig. 2b). The x axis in all ofthe figures (Figs. 2a to 2c) is time in msecreckoned from the instant of stimulation.The x scale is the same for all three poten-tials and is equal to 2 msec per division.The y axis is the absolute amplitude of thewaveforms in microvolts. The nerve con-duction times (NCT), as measured in thisstudy, also have been indicated on thewaveforms.

The patient data have been divided intotwo groups; one consisting ofdata from thoseupper limbs with clinical evidence of me-dian nerve abnormality in the form of en-largement or tenderness or pain or associ-ated muscle weakness, and the other fromclinically normal upper limbs. Thus, it ispossible that data from one hand of a pa-tient was added to group(i) and that fromthe other hand was added to group(ii), de-pending upon the clinical symptoms of theindividual median nerves. This distinctionwas made in order to observe whether thereare any definite changes in the response ofthe nerve prior to the clinical manifestationof dysfunction.

The motor threshold of stimulation (S)was noted in each case. The neuroaveragersystem developed has a provision to marka peak and the following valley, and thento read the difference in amplitude and la-tency. This facility enables measurement ofthe peak-to-peak amplitude and duration(interpeak latency between the main nega-tive peak and the following positive peak)of a potential. The absolute latencies of themain negative peaks (N3, N5 and N9) of allthree responses were read from the respec-tive waveforms on the screen and noted.Also measured were the interpeak (relative)latencies (P4—N3, P6—N5, P11—N9) of theresponse peaks. Normally, the duration ofthe nerve action potential is measured fromthe initial deflection of the baseline to theintersection of the descending phase and thebaseline ( 9 ). However, in this study, the dif-ference between the latencies of the negativeand positive peaks was noted, since this canbe measured with better accuracy and will

roughly correspond to half the duration ofthe potential as defined above. Further,changes in the interval between the positiveand negative peaks is considered to reflectthe dispersion of fast- and slow-conductingimpulses ( 9 ) and, thus, could be studied forany possible differences between normalsubjects and patients. In addition, the peak-to-valley amplitudes of the compound nerveaction potentials at the finger (Ad = P4—N3)and elbow (A, = P6—N5) were measured andtabulated. Similarly, A b (P11—N9), the peak-to-peak amplitude of the waveform record-ed at Erb's point, was measured and tabu-lated. The nerve impulse arrival times (nerveconduction times, NCT) at the three pe-ripheral recording sites were measured upto the initial positive deflection. The actuallengths of the nerve segments between theelectrode sites were measured in each pa-tient, and the NCVs were computed by di-viding these distances by the respective seg-mental conduction times. In case of the armsegment, the length of the nerve involvedis the difference between the distances ofthe two recording sites, namely, elbow andErb's point, from the wrist. Referring to Fig-ure 1, this segmental length can be identifiedas (Le, — La). This length is divided by thecorresponding segmental conduction timeto arrive at the value of NCV (V,). Thisconduction time is the difference in arrivaltimes of the propagating nerve impulse atthe brachial plexus (NCT, arm) and elbow(NCT, forearm). The statistical significanceof the deviation of patient values from thenormal subjects was determined for each ofthe nerve conduction parameters.

Electrophysiological data were correlatedto the clinical assessment of the patients.From the patient files maintained by thehospital, the clinical evaluations of the pa-tients regarding the sensory and the motordamage in the upper limbs were recorded.Areas of partial or total loss of sensation,paresis, paralysis or anesthekinesia, loca-tions of active and healed patches, handclawing, if any, details regarding decom-pression of any nerve by surgical slitting ofthe nerve sheath, etc., were also noted, witha view to correlate the electrophysiologicaldata with the clinical observations. As partof another ongoing study in the hospital,these sensory and motor evaluations of thepatients were being carried out once a month

63, 3^Ramakrishnan and Srinivasan: Electrophysiological Correlates^399

and the data updated. Thus, it was worth-while to look at the relationship betweenthese clinical findings and the electrophys-iological parameters measured.

In clinical practice, it may be very incon-venient to deal with so many different mea-sured variables, especially when there is anoverlap in the range of values for normalsubjects and patients in some of the param-eters. Hence, it is ideal to obtain an inte-grated measure or index based on these pa-rameters, depending on the magnitude ofwhich categorization of a nerve as normalor abnormal will be possible. Discriminantanalysis is the appropriate statistical tool touse here ( 1 s) since this technique is normallyused to classify an observation into one oftwo or more mutually exclusive and ex-haustive groups on the basis of a set of pre-dictor variables. The analysis also evaluatesthe effectiveness of the various parametersin distinguishing the abnormal from thenormal nerve potentials. In the presentstudy, the interest lies mainly in classifyinga nerve into one of two distinct groups fromthe values of a set of numerical variables,namely, the measured nerve conductionvariables.

The approach in discriminant analysis isas follows. After dividing the sample intotwo or more distinct groups, data are col-lected on the values of the variables for allthe individuals belonging to each group. Thediscriminant analysis then derives a linearcombination of these parameters which"best" discriminates between these groupswith least error. In the case of a dichotomy,the problem is to ascertain appropriateweights for the series of variables yieldingmaximum linear separation in the two con-trasted groups. The particular simple formof linear discriminant function allows a clearinterpretation of the effect of each of thepredictor variables. Thus, the main objec-tives in pursuing discriminant analysis inthe work reported here are: a) to test wheth-er significant differences exist between nor-mal subjects and patients; b) to determinewhich variables account most for such in-tergroup differences; c) to find a linear com-bination of the predictor variables that rep-resents the two groups by maximizing theseparation between the groups and mini-mizing the separation within the groups; d)to establish procedures for assigning new

sets of data to one of the groups, assumingthat they come from one of the a priori de-fined groups (namely, normal individualsand leprosy patients).

If the analysis uses data on n observationsto calculate the discriminant function andthen classifies the same n observations, thenthe results will be biased. There will be morecorrect classifications than the discriminantfunction is capable of delivering under morerealistic conditions. To avoid this, it is ad-visable to fit a function to part of the dataand then to use this function to classify theremaining sets of data. Hence, of the 25normal data collected, 18 were used to de-termine the discriminant function, reserv-ing the rest to test the efficacy of the ob-tained classifier. To have a clear demarca-tion between the normal and patient data,only the data from the clinically (mediannerve) affected upper limbs were employedin arriving at the discriminant function.Thus, of the 42 sets of upper limb responsesfrom 21 patients, only 17 were included inthe procedure for estimation of the coeffi-cients of the linear discriminant function.

RESULTSTable 1 lists the means, standard devia-

tions and the spread of the data of the stim-ulus threshold and the absolute latencies ofthe responses for both normal subjects andpatients. Some of the responses could notbe recorded from three upper limbs of twopatients. Hence, the data from those limbswere not included in either group for com-puting the statistics. It is observed that boththe sensory and motor thresholds of the me-dian nerve are appreciably higher in pa-tients. The increase in the stimulus thresh-old is statistically significant (p < 0.005) inboth groups of patient data. A look at themean values of the absolute latencies re-veals that it is only the distal latency N3which is considerably prolonged in the pa-tients (p < 0.001 in both groups). The in-crease in N5 is not at all significant. Thechanges in N9 latencies, although significantin the patient group 1 (p < 0.005), are notsignificant in the other group (clinically un-affected).

The means and variance of the fall times(interpeak latencies) of the potentials aregiven in Table 2. It is observed that therelative latencies of the potentials are sig-

400^ International Journal of Leprosy^ 1995

TABLE 1. Stimulus thresholds and absolute latencies of normal subjects and leprosypat ients. a

Age^Height^ N3^N5^N9(yrs.)^

(cm)^(V)^

(msec)^

(msec)^(msec)

Mean^25S.D.^3Range^19 to 31

Mean^25S.D.^4Range^20 to 33p Value

MeanS.D.Rangep Value

Normal subjects

^

164^34^2.72^4.79

^

9^9^0.23^0.29149 to 180^20 to 57^2.10 to 3.24^4.26 to 5.58

Patients [median nerve clinically affected limbs (17)]

^

160^

59^

3.79^

5.07

^

9^

32^

0.88^

0.64149 to 185

^25 to 164

^2.64 to 5.70

^3.48 to 6.00

0.005^

0.001^

N/SPatients [median nerve clinically unaffected limbs (22)]

^

50^3.58^4.92

^

17^0.63^0.4921 to 86^2.88 to 5.52^4.20 to 6.30

^

0.001^0.001^N/S

9.660.66

8.53 to 11.07

10.731.46

8.80 to 14.130.005

9.900.80

8.40 to 11.73N/S

See list of symbols and abbreviations on page 000.

nificantly different only in the case of thedigital responses (P4-N3) and that, too, onlyfor the clinically affected nerves of the pa-tients. There is no appreciable difference be-tween the normal subjects and the two pa-tient groups with respect to the mean valuesof the fall times of the potentials at any ofthe other recording sites, but the spread ofvalues is large for the patients.

Table 3 lists the nerve condition times(NCT) and the segmental conduction ve-locities of the normal subjects and patients.Table 4 shows the sample statistics of the

peak-to-peak amplitudes for each of thethree responses for all the three sets of data(normal subjects, clinically abnormal limbs,and clinically normal limbs) and, also, thesignificance of deviation of each group ofpatient data from normal data. It is seenthat the reduction in amplitudes of CNAPsare very highly significant, p < 0.001 at allrecording locations for the clinically affectedpatient nerves. Further, even in the case ofclinically unaffected limbs, the two distalamplitudes are significantly reduced at thesame level (p < 0.001). However, NCVs are

TABLE 2. Relative latencies of response peaks of normal subjects and hanseniasis pa-tients.a

P4-N3 P6-N5 P11-N9

(msec) (msec) msec)

Normal subjectsMean 0.99 1.08 1.56S.D. 0.13 0.20 0.44Range 0.72 to 1.32 0.78 to 1.56 0.53 to 2.33

Patients [clinically affected nerves (17)]Mean 0.76 1.25S.D. 0.22 0.67 1.77Range 0.36 to 1.14 0.54 to 3.48 0.77p Value 0.001 N/S 0.33 to 3.13 N/S

Patients [clinically normal nerves (22)]Mean 1.02 1.18 1.45S.D. 0.24 0.40 0.57Range 0.54 to 1.74 0.66 to 2.58 0.33 to 2.67p Value N/S N/S N/S

Sec list of symbols and abbreviations on page 000.

63, 3^Ramakrishnan and Srinivasan: Electrophvsiological Correlates^401

TABLE 3. NCT and NCV of normal subjects and patients.a

Nerve conduction times Segmental cond. velocities

Palm Forearm Arm Palm Forearm Arm

msec (m sec ) (msec) (V,„ m/s) (V,, m/s) Va, m/s)

MeanS.D.Range

1.970.22

1.44 to 2.34

Normal subjects

^

3.92^8.14^67.0

^

0.29^0.58^7.33.24 to 4.50^7.20 to 9.33^50.9 to 80.2

Patients [clinically affected median nerves (17)]

75.54.7

66.7 to 86.4

70.56.9

56.9 to 83.1

Mean 3.04 4.19 9.34 52.0 68.1 62.2S.D. 0.84 0.53 1.35 10.9 7.5 12.6Range 2.22 to 5.10 3.30 to 5.28 7.93 to 13.20 28.8 to 67.6 58.0 to 90.9 36.3 to 91.6p Value 0.001 N/S 0.001 0.001 0.001 0.02

Patients [clinically unaffected median nerves (22)]Mean 2.75 4.03 8.48 57.7 66.4 65.7S.D. 0.63 0.37 0.83 10.9 4.5 7.4Range 2.10 to 4.26 3.48 to 4.86 7.13 to 10.80 35.0 to 74.1 58.0 to 73.2 53.0 to 80.6p Value 0.001 N/S N/S 0.005 0.001 0.05

See list of symbols and abbreviations on page 000.

reduced at this level of significance (p <0.001) in only 3 of the 6 patient groups.

Nonetheless, there is noticeable reductionin the NCVs of all three segments for bothsets of patient nerves. Even in the clinicallynormal set, the decrease is highly significant(p < 0.005) for the palm segmental velocity(V0), very highly significant (p < 0.001) forthe forearm segmental velocity (Vfa ), andsignificant (p < 0.05) for the arm segment(V a ). Table 4 clearly shows that the mag-nitudes of all three peripheral potentials are

considerably lower in patients than in thenormal subjects. The mean peak-to-peakamplitudes of the responses at the digit, el-bow and Erb's point in the clinically ab-normal group are 8.5%, 24.6% and 55.0%,respectively, of the corresponding values forthe normal subjects (p < 0.001 in each case).The values for the clinically unaffected groupare 27.2% for A, and 60.1% for A, and,again, both are very highly significant (p <0.001). These results are pictorially repre-sented in Figure 3.

TABLE 4. Amplitudes of normal subjects' and patient responses.a

P4-N3 P6-N5 P11-N9

(Ad , AV) (A,, AV) (A,„ AV)Normal subjects

Mean 49.96 17.13 7.26S.D. 20.56 5.12 2.62Range 18.49 to 91.04 8.42 to 28.87 3.15 to 12.50

Patients [clinically affected median nerves (17)]Mean 4.24 4.21 3.99S.D. 4.28 3.19 2.26Range 0.70 to 14.97 0.38 to 11.65 0.85 to 11.48p Value 0.001 0.001 0.001

Patients [clinically normal median nerves (22)]Mean 13.63 10.30 6.35S.D. 9.14 5.36 3.94Range 0.79 to 34.72 2.36 to 19.43 2.22 to 19.12p Value 0.001 0.001 N/S

See list of symbols and abbreviations on page 000.

I 80

— 70 -50

- 60

Cl.E

Patients: Normal nerves [1]^Affected nerves 1:11

80

V,^Ad— -70

Normals w100

90

j Ae

-40

-30 1—

-20

-10^0>

>

13 60.L

50U,7

4 0in

30

20

Vfa

402^ International Journal of Leprosy^ 1995

FIG. 3. Mean (±S.D.) values of nerve conduction parameters of both normal subjects and leprosy patients.A, = digit amplitude (P4—N3); = elbow amplitude (P6—N5); A, = Erb's amplitude (P1 l—N9); S = stimulusthreshold; V, = NCV of palm segment; V, = NCV of forearm; V. = NCV of arm.

The patient response waveforms have anumber of small peaks spread out over time,indicating different arrival times in differentnerve fibers demyelinated or otherwisedamaged to different degrees. Thus, thewaveshapes of all three peripheral poten-tials in the patients appreciably differ fromthose of controls.

While applying the discriminant analysisto begin with, 13 parameters were consid-ered, namely, the stimulus threshold, thefour fall times, the central conduction time(which is the interpeak latency between N9and the latency of the cortical potential), thethree segmental nerve conduction veloci-ties, and the four peak-to-peak amplitudesof the responses. The discriminant functionobtained was highly effective in discrimi-nating between the two groups. The sepa-ration between the groups is significant atthe 0.05% level (p < 0.0005), denoting thatthe two groups are very significantly dis-tinct. However, an inspection of the resultsshowed that the discriminating power of five

of the parameters is 15%, while that ofthe others is

See list of symbols and abbreviations on page 000.

Parameter

S

A,

Parameter

SV PA.,

Disc. power

(% of )

15.1810.2915.0130.8528.68

Disc. power

(% of D 2 )10.9019.0170.09

63, 3^Ramakrishnan and Srinivasan: Electrophysiological Correlates^403

TABLE 5. Discriminating powers of pe-^TABLE 6. Significance of distal nerve con-ripheral conduction parameters (number of duction predictors (number of variances =variances = 5)."^ 3). a

Sec list of symbols and abbreviations on page 000.

clinically abnormal nerves. Only 4 of the 25sets of data from clinically normal nervesof patients are classified as normal, whilethe other 21 sets of responses from limbswith clinically confirmed involvement ofonly the ulnar nerve are declared abnormalwith respect to the median nerve also. Allfive predictor variables have good discrim-inating power, contributing 10% to 30% tothe mean separation between the groups.This result quantitatively shows that the twodistal responses sufficiently identify an ab-normal median nerve. In other words, thediscriminating powers of the NCV in thearm segment as well as the amplitude of theErb's potential are poor compared to thoseof the distal potentials.

However, from the point of view ofa fieldsurvey, it would be ideal if classification (asnormal or abnormal) is possible based onthe response obtained from a single record-ing site. Hence, in the final iteration, theeffectiveness of a classifier involving onlythree parameters was studied. These are thedigital response amplitude (A d), the palmsegmental conduction velocity (V0), and thestimulus strength (S). The discriminatingpowers of each variable obtained in this thirditeration are tabulated in Table 6. Table 7shows the value of the discriminant func-tion for each of the data and the resultantclassification. A "b" in the table precedingthe subject ID No. identifies the trainingdata (data which were utilized to determinethe discriminant function). D, and D, de-note, respectively, the mean scores of thetraining data from class 1 (normal group)and class II (affected nerves of patients). D*is the critical score (obtained here as [D, +D2]/2) used to classify any unknown data

into one of the above two contrasted classes.The intergroup separation based on onlythese variables is, again, very highly signif-icant (p < 0.0005). The false-negative clas-sification of the control subject N7 is dueto the unusually high stimulus threshold of57 volts.

Let us consider each of the four sets ofdata from clinically unaffected patient limbsclassified as normal. In the case of the pa-tient P5, the stimulus threshold is very lowsince he is a young boy (age 15, lower thanthe control group). Also, his right hand isnormal with normal values of NCVs (V, =65.3 m/sec, Vra = 69.3 m/sec) and ampli-tudes (A d = 22.0 /IV, A, = 15.8 kiV) andwith no clinical symptoms. Hence, the clas-sification of P5-R as normal is justified. Pa-tient P13 has been fully cured of the diseaseand thus is rightly classified. This patienthad BT (borderline tuberculoid) leprosy, hadbeen given dapsone (DDS) monotherapy ina local hospital, later MDT (multidrug ther-apy— paucibacillary) for 6 months and thenwas treated with 50 mg DDS—on the whole,treated for a period of 10 years. He hadminimal abduction deformity in the rightlittle finger and was operated on for surgicalcorrection of the flexion deformity of thelittle finger at the metacarpophalangeal jointlevel. Lastly, patient P20 also has a perfectlynormal right hand, both by the absence ofsymptoms and by the values of the variousvariables. Now, looking at the clinicallynormal data classified as abnormal, P4, forexample, has ulnar nerve involvement inboth hands and median nerve involvementin the right hand only. The left arm digitpotential has a very low amplitude of 2.0AN, suggesting severe neuropathy of the leftmedian nerve. The NCV of the distal seg-ment (palm) also has a very low value of

404^ International Journal of Leprosy^ 1995

35.2 m/sec. The classifier, although trainedwith data from clinically abnormal nerves,is able to correctly classify affected nervessuch as this one which show no clinical dys-function. This shows the efficacy of the clas-sifier and also proves that electrophysiolog-ical studies can reveal neural abnormalitiesearlier than the clinical manifestation of thedisease.

Thus the 3-predictor linear classifier isfound to be capable of identifying properlyboth the normal and abnormal hands of thepatients. The false-negative classification canbe avoided by biasing the critical score val-ue. This, however, may entail a few false-positive classifications and, thus, may workcontrary to interest. However, 24 out of 25normal subjects (96%) have been rightlyclassified, thus proving the usefulness of thediscriminant function.

Clinical correlation. The clinical assess-ments from the patients' records were cor-related with the neurophysiological data ob-tained. Three cases are illustrated here asexamples. Patient PIO has near total loss ofsensation in the distal segment up to a dis-tance of 10 cm proximal to the wrist of bothhands. Assessment of muscle power showsthat the right median nerve is normal andthere is lower median nerve paresis in theleft hand. The fact that there is a total lossof sensation in the left hand is not reflectedin the conduction velocity of the palm seg-ment (65.3 m/sec), which is a perfectly nor-mal value; whereas the amplitudes of theperipheral potentials are very low and thevalues are marginally higher in the case ofthe right hand (left: Ad = 1.2 1V, A, = 1.5AV, right: Ad = 1.3 AV, A, = 2.5 AV) and,thus, correlate well with the clinical obser-vations. Patient P13 has nearly been, curedof the disease and has some residual lowerulnar motor weakness in the right handwhereas the median nerve innervated mo-tor and sensory activities are fully normalin the left upper limb. For this patient, theright arm distal conduction velocity (54.9m/sec) is in the abnormal range. The am-plitudes of both distal peripheral potentials(Ad : 34.7, 26.9 AV; A,: 19.4, 12.3 AV) aretotally in the normal range. Considering theother case, Patient P20 has no sensationsdistal to the wrist in the left hand while themotor activity is normal. In the right arm,although the median nerve is thickened,

TABLE 7. Classification ofdata by the dis-criminant . fiinction of 3 variables.

(I), = 20.97 1), = 6.63 D* = 13.80)

Subj. D Gr.' Subj. D Gr.'

NI 15.07 I hP6R 3.63 1N2 14.12 1 P6L 12.64 1

6 1\13 28.37 1 P712 4.50 1"N4 24.65 1 hP7L 5.11 1hN5 18.05 1 hP8R 7.67 IhN6 18.60 1 hP8L 5.42 1hN7 12.06 "P9R 12.44 1

hN8 16.88 1 "P9L 8.09 2hN9 19.98 1 hPIOR 4.72 1hN10 20.37 1 hPIOL 8.52 IN11 21.21 1 PI IR 6.49N12 16.74 1 PI IL 5.17 I

"N13 29.61 1 hP12R 8.24 1N14 17.78 1 hP12L 3.30

hN15 19.09 1 PI 312 15.44 1hN16 20.76 1 Pl3L 14.07 1N17 21.23 1 hP14R 10.32 2N18 14.38 1 P14L 13.29

hN19 17.66 1 1315R 12.69 2'N 2 0 22.47 1 P15L 11.24 2"N2I 24.90 1 P16R 6.56 2"NT/ 22.07 1 P16L 7.19 2N23 25.60 1 P17R 7.90 2

hN24 26.79 1 P17L 11.90 2"N25 17.40 1 hP18R 11.33P 1 R 11.28 2 hP18L 7.73 2PI L 12.85 2 P19R -5.42 2

hP2R 7.14 2 P19L -6.43P2L 9.69 2 P2OR 14.10 1

''P3R 10.06 2 hP2OL 0.78 2hP3L 6.91 2 P21R 9.40 2P4 R -7.24 2 P21L 8.74P4 L 2.03P5R 14.24 1

hP5L 3.75 1

1 = Normal; 2 = abnormal.h Data utilized in arriving at the discriminant coef-

ficients.Underlining denotes measurements in normals be-

low critical score or measurement in patients abovecritical score.

there is neither sensory loss nor any motorweakness. The right arm digital responseamplitude of 25.7 AV is less than the min-imum normal value (mean - S.D.) of 31.5AV and higher than the maximum in thepatient range (mean + S.D.) of 18.7 AV; thevalue of the left arm potential is only 2.1AV. A similar difference is seen with elbowvalues. On the other hand, the left arm palmNCV (58.8 m/sec), although less than theright side, is almost a normal value and givesa misleading picture if one goes by it.

63, 3^Ramakrishnan and Srinivasan: Electrophysiological Correlates^405

DISCUSSION

Every leprosy patient suffers from pe-ripheral nerve involvement as a result ofthe disease. This may vary from the in-volvement of intradermal nerves in a cu-taneous patch to a major lesion in the nervetrunk. Thus, there is no non-neural leprosy.Despite the magnitude of the problem ofleprosy incidence in the world today andthe fact that about 20% of the patients sufferfrom major sensory and motor neurologicaldeficits, extensive electrophysiological stud-ies in leprosy have been few. Further, rel-atively more work has been done on motorconduction than on sensory conduction.Since sensory loss precedes motor deficit inalmost all the patients, with the idea of earlydetection in mind, only sensory conductionstudies have been performed in this study.These data were correlated with the clinicaldata of the patients. The principal time do-main variables that demarcate the healthyresponses from the pathological ones havebeen extracted by subjecting the data to dis-criminant analysis.

It is observed that in most of the Hansen'sdisease patients, the ulnar nerve is the firstto be affected. Because of fear, shame andsocial stigma associated with this disease,few people come to leprosy centers in theearly stages of the disease. Thus, even in theoutpatient departments, almost all the pa-tients reporting have advanced ulnar nerveinvolvement. This observation of a higherpercentage and/or degree of involvement ofthe ulnar nerves is in general agreement withthe findings of earlier researchers (1,10,13,16).The chief purpose of the work reported herehas been to study the early changes in theaffected nerves. So, the ulnar nerve couldnot be considered because, as explainedabove, it was difficult to get patients withminimal involvement of ulnar nerves. Thus,the median nerve was selected for study inthe patients. This was done to find out thechanges occurring early during the infectionof the median nerve. Most of the patientschosen had little or no clinical involvementof the median nerve in at least one side.

Unlike the amplitudes and the NCVs, thechanges in the absolute and relative laten-cies of the peaks in the various responsesare not statistically significant, except forthe palm segment. Thus, the latencies are

not very useful in reflecting the abnormal-ities of patient responses. Trying to relatethe nerve conduction velocities with theclinical findings has not revealed any sig-nificant correlation; whereas there is nearlyperfect correlation between the amplitudesof the compound nerve action potentials andthe motor or sensory deficits observed clin-ically. There is an appreciable reduction inthe mean value of the sensory NCVs of thetwo distal segments (forearm and palm) al-though correlation with the clinical symp-toms could not always be established. Thenormal range of palm NCV is 59.7 m/secto 74.3 m/sec. The ranges of patient valuesare 41.1 m/sec to 62.9 m/sec for the clini-cally affected group and 46.8 m/sec to 68.6m/scc for the other group. Thus there is anoverlap in the ranges and, hence, discrim-ination based upon NCV alone cannot beerror-free, although the reduction in bothdistal NCVs is highly significant. However,in the case of the amplitudes of the digitalsensory nerve action potentials, there is aclear margin of around 21 AV and 7 AVbetween the normal range and the two ab-normal ranges. Again, there is a margin ofaround 5 AV between the ranges of elbowamplitudes between controls and the clini-cally affected group. This reduction in am-plitudes is in spite of the higher stimulusstrengths employed due to the increasedthresholds of the patients. Hence, it appearsthat the amplitudes of the distal peripheralpotentials are more reliable indicators ofleprous neuropathy than the sensory NCVs.These detailed studies revealing the relativesignificance of the amplitudes over the NCVsas also the confirmation by correlation withclinical features have not been reported sofar and, hence, might reflect an importantcharacteristic of leprous neuropathy. Mostof the earlier research has tried to relateEMG and motor NCV and, rarely, sensoryNCV with the clinical dysfunction. Thus, apotentially useful fact has probably beenmissed. It seems that the electrodiagnosticapplication of NCVs in leprous neuritisshould be limited to the relatively advancedcases.

The overall reduction consistently seen inthe amplitudes of the peripheral potentialsimplies a considerable reduction in thenumber of active, fast-conducting sensorynerve fibers. The significance values as well

406^ International Journal of Leprosy^ 1995

as the ranges of the parameters as shown inFigure 3 underline the importance of thechanges in the two distal potentials, even inthe clinically normal group.

The discriminant analysis confirms thatthe nerve action potentials recorded fromthe median nerve at the elbow and thirddigit show discernable abnormalities muchearlier to the clinical manifestation in theform of sensory or motor deficit. The resultsclearly show that a classifier with only threevariables is fully capable of distinguishingbetween the normal and abnormal mediannerve responses. This discriminant functionis given by,

D = —0.063S + 0.1817V, + 0.2081Ad

where S is the motor threshold of stimula-tion, V p is the NCV of the palm segment,and A, is the amplitude of the digit poten-tial.

The functional derangement of nerves isrevealed by nerve conduction studies beforethe appearance of clinical signs. The reduc-tion in the amplitude of responses in clin-ically unaffected nerves may indicate an ear-ly stage of nerve involvement, thus being ofsome predictive value. With this finding,therapy can be directed to the nerve earlyenough to prevent possible development ofany disability. After further confirmativeinvestigations, the study of distal conduc-tion of the upper limbs may possibly be usedto screen a population exposed to Myco-bacterium leprae. However, one is not surewhether similar observations would be madein the case of ulnar nerves. Thus, it 'wouldbe very informative if the studies of theabove nature were carried out on ulnarnerves of patients in the early stages of lep-rosy. If such studies also show similar re-sults in ulnar nerves then, in a field situationfor a preliminary screening of the exposedpopulation, it might suffice to take only thedigit response to the median or ulnar nervestimulation at the wrist and to note the stim-ulus threshold. If there is an order of mag-nitude difference between the left and rightside, or if any of the values is much belowthe normal range, then such cases can bereferred to the leprosy centers for smear tests,biopsy, etc., to establish the presence or ab-sence of infection by Al. leprae. The scoreof the discriminant function may also be

used as the deciding factor. A quick decisioncan be made and the whole testing and es-timating procedure may take only 15 min.

Such a quantitative electrophysiologicalassessment of sensory nerves could possiblybecome a tool for diagnosis as well as formonitoring both the progress of the diseaseand the effects of treatment. It could alsoperhaps be used to help evaluate and com-pare the various methods of treatment.However, the occurrence of one false-neg-ative classification indicates that further im-provement of the performance of the clas-sifier is perhaps required to pave the wayfor wider use of this method under fieldconditions. The authors hope that by in-corporating these methods for clinical eval-uation of the suspected population it mayeventually be possible to diagnose the neu-ropathy earlier than hitherto possible, lead-ing to better management and ultimate con-trol of the disease.

It is significant that impairment of NCVwas not always found even with patientshaving symptoms of neural involvement. Inother neuropathies, neurologists normallyrely more upon NCV than amplitudes, sincevariabilities in amplitudes could be causedby factors not physiological in nature. But,in the case of hanseniasis, since the meanperipheral SNAP amplitudes are reduced bya factor of 10 and the elbow CNAP ampli-tudes by a factor of more than 4, it may bepreferable to use the neural response am-plitudes rather than the NCVs for clinicalevaluation.

SUMMARYSensory nerve action potentials (SNAPs)

and compound nerve action potentials(CNAPs) were recorded from 25 normalsubjects and 21 hanseniasis patients follow-ing electrical stimulation of the mediannerve at the wrist. The various nerve con-duction parameters from the affected nervesof the patients were compared with thosefrom the clinically normal nerves of patientsas well as data from healthy individuals.Analysis of the data and clinical correlationstudies indicate the suitability of ampli-tudes of the SNAPs and CNAPs rather thanthe nerve conduction velocities in bettercharacterizing the neuropathy of the pa-tients. Significantly reduced amplitudes of

63, 3^Ramakrishnan and Srinivasan: Electrophysiological Correlates^407

responses from clinically unaffected nervesof patients indicate an early stage of neu-ropathy, thus being of predictive value. Fur-ther, a discriminant classifier, trained ondata from clinically affected nerves of pa-tients, classified most of the data from clin-ically unaffected nerves of patients as ab-normal. This indicates that clinical neuro-physiological studies can reveal leprousneuropathy much before it becomes clini-cally evident by means of sensory or motorloss. A discriminant score involving onlythe parameters of motor threshold, ampli-tude of digit potential and palm nerve con-duction velocity is able to classify almostall of the normal and abnormal responses.The authors hope that further confirmativestudies might ultimately lead to the use ofthe study of distal sensory conduction in theupper limbs in possible screening of a pop-ulation exposed to Mycobacterium leprae.On the other hand, misclassification of anormal person occurred and suggests thatfurther refinement of the methods is nec-essary in order to facilitate wider use of themethods under field conditions.

RESUMENDespues de Ia estimulaciOn electrica del nervio me-

dian° de la muneca se registraron los potenciales deacciOn nerviosa sensorial y mixta (SNAPs y CNAPs)en 25 sujetos sanos y en 21 pacientes hansenianos. Losdiferentes parâmetros de conducciOn nerviosa de losnervios afectados en los pacientes se compararon conaquellos de los nervios no afectados y con los mismospardmetros en los sujetos sanos. El andlisis de los datosy los estudios de correlaciOn clinica indican que lamediciOn de los potenciales de acci6n nerviosa es unmejor pardmetro que la mediciOn de las velocidactesde conducciOn nerviosa para establecer y caracterizarla neuropatia de los pacientes. Las respuestas de am-plitud significativamente reducida en los nervios cli-nicamente no afectados son indicativas de una neu-ropatia incipiente y por lo tanto tienen valor predic-tivo. Usando esta tecnica, un examinador experimen-tado, clasific6 a la mayoria de los nervios clinicamentesanos como anormales. Esto indica que los estudiosneurofisiolOgicos puedan revelar Ia neuropatia leprosamucho antes de que esta liege a ser clinicamente evi-dente. Un registro discriminativo que comprenda sololos pardmetros del umbral motor, Ia amplitud del po-tencial digital, y la velocidad de conduciOn del nerviopalmar, es suficiente para clasificar a casi todas lasrespuestas normales y anormales. Los autores esperanque los estudios corroborativos posteriores puedan fi-nalmente conducir al empleo de estas têcnicas paraestablecer el nivel de conducciOn sensorial distal en las

extremidades superiores, en programas de exploraciOnde las poblaciones expuestas a Mycobacterium leprac.Por otro lado, tambien occurriO la clasificaciOn errOneade una persona normal; esto indica que es necesariohacer un refinamiento de los metodos para que estosse puedan aplicar bajo condiciones de campo.

RÉSUMÉ

Les potentiels d'action nerveux sensoriels et les po-tenticls d'action nerveux mixtes ont etc enregistres chez25 sujets normaux et 21 patients hanseniens apresstimulation êlectrique du nerf median au poignet. LesdifThrents paramêtres de la conduction nerveuse desnerfs affectés des patients ont etc compares avec cruxdes nerfs cliniquement normaux des patients ainsiqu'avec les donnees provenant des individus en bonnesante. L'analyse des donnees et les etudes de correlationclinique indiquent que c'est l'amplitude des potentielsd'action nerveux, plutOt que la vitesse de conductionnerveuse, qui caracterise le micux la neuropathie despatients. Des amplitudes de rêponses reduites de ma-niere significative de nerfs cliniquement non atteintschez des patients indiquent un stage precoce de neu-ropathie, ct ont done une valcur predictive. De plus,une classification discriminante, realisee a partir dedonnees provenant de nerfs cliniquement atteints depatients, a class& comme anormaux Ia plupart des nerfscliniquement indemnes des patients. Ceci indique quedes etudes neurophysiologiques cliniques peuvent re-veler une neuropathie lepreuse longtemps avant qu'ellene devienne evidente cliniquement suite a une pertesensorielle ou motrice. Un score discriminant compre-nant seulement les parametres de seuil motcur, l'am-plitude du potentiel au doigt et la vitesse de conductionnerveuse palmaire permet de classer presque toutes lesreponses normales et anormales. Les auteurs esperentque d'autres etudes de confirmation pourront finale-ment conduire a ]'utilisation de l'êtude de la conduc-tion sensorielle distale dans les membres superieurspour un depistage de Ia population exposé au Myco-bacterium leprae. D'autre part, une mauvaise classifi-cation d'une personne normale survint, ce qui suggerequ'un raffinement de la mêthode est encore necessaireafin de faciliter une utilisation plus large de Ia methodedans les conditions de terrain.

Acknowledgment. The work reported here was com-pleted when the authors were with the Indian Instituteof Technology, Madras. The authors thank the De-partment of Electronics, Government of India, for pro-viding part of the funds for carrying out these studies.The authors gratefully acknowledge the help of Dr. K.S. Rao, former Deputy Director of the Central LeprosyTeaching and Research Institute (CLTRI), Chenglepet,South India, for providing access to the patients andtheir functional assessment data, and Mr. Anand Sa-tyadoss, physiotherapist, CLTRI, for his help. Thanksarc also due to Dr. S. S. K. Ayyar, Consultant Neu-rologist, Vijaya Hospital, Madras, for his suggestions.

408^ International Journal of Leprosy^ 1995

Finally, the authors are grateful to the reviewers fortheir constructive comments.

9.

REFERENCES1. ANTIA, N. H., PANDYA, S. S. and DASTUR, D. K. 10.

Nerves in the arm in leprosy. 1. Clinical, electro-diagnostic and operative aspects. Int. J. Lepr. 38(1970) 12-29.

2. CHAROSKY, C. B., GATTI, J. C. and CARDAMA, J. 11.E. Neuropathies in Hansen's disease. (Editorial)Int. J. Lcpr. 51 (1983) 576-586.

3. DASH, M. S. A study of the conduction velocity 12.of sensory fibers of the ulnar nerve in leprosy. Int.J. Lepr. 35 (1968) 460-469.

4. DHAND, U. K., KUMAR, B., DHAND, R., CHOPRA, 13.J. S. and KAUR, S. Phrenic nerve conduction inleprosy. Int. J. Lepr. 56 (1988) 389-393.

5. GOURIE-DEVI, M. Greater auricular nerve con- 14.duction in leprosy. Indian J. Lepr. 56 (1984) 182-190.

6. FlAcia-r-r, E. R., SHIPLEY, D. E. and LIVENGOOD,R. Motor nerve conduction velocity studies of 15.the ulnar nerve in patients with leprosy. Int. J.Lepr. 36 (1968) 282-287.

7. HARIDASAN, G., SANGHVI, S. H., JOSHI, V. M., 16.PANDYA, S. S. and DESAI, A. D. Quantitative elec-tromyography using automatic analysis. J. Neurol.Sci. 48 (1980) 353-365.

8. JOPLING, W. H. Handbook of Leprosy. 3rd edn.

London: William Heinemann Medical Books Ltd.,1984, p. 1.KIMURA, J. Electrodiagnosis in Diseases of Nerveand Afuscle: Principles and Practice. Philadelphia:F. A. Davis, 1983, pp. 74-76.MCLEOD, J. G., HARGRAVE, J. C., WALSH, J. C.,BOOTH, G. C., GYE, R. S. and BARRON, S. A. Nerveconduction studies in leprosy. Int. J. Lepr. 43 (1975)21-31.RAMAKRISHNAN, A. G. Event related potentialmonitoring and analysis in health and disease,Ph.D. dissertation, I. I. T., Madras, India, 1989.SEBILLE, A. and GRAY, F. Electromyographic re-cording and muscle biopsy in lepromatous lep-rosy. J. Neurol. Sci. 40 (1979) 3-10.Sow, A. S., KANDHARI, K. C. and SINGH, N. Mo-tor nerve conduction studies in leprosy. Int. J.Dermatol. 10 (1971) 151-155.SWIFT, T. R., HAcxErr, E. R., SHIPLEY, D. E. andMINER, K. M. The peroneal and tibial nerves inlepromatous leprosy; clinical and electrophysio-logic observations. Int. J. Lepr. 41 (1973) 25-34.TABACHNICK, B. C. and FIDELL, L. S. Using Mul-tivariate Statistics. New York: Harper & Row,1983, p. 9.VERGHESE, M., ITTIMANI, K. V., SATYANARAYAN,K. R., MATHAI, R. and BHAKTHAVIZIAM, C. Astudy of the conduction velocity of the motor fi-bers of ulnar and median nerves in leprosy. Int.J. Lepr. 38 (1970) 271-277.

Page 1Page 2Page 3Page 4Page 5Page 6Page 7Page 8Page 9Page 10Page 11Page 12Page 13Page 14