Clinical Neurophysiology 124 (2013) 610618

Contents lists available at SciVerse ScienceDirect

Clinical Neurophysiology
journal homepage: www.elsevier.com/locate/clinph

MUNIX and incremental stimulation MUNE in ALS patients and control subjects
Jasna Furtula a, Birger Johnsen a, Peter Broegger Christensen b, Kirsten Pugdahl a, Carsten Bisgaard c, Mette-Kirstine Christensen d, Jens Arentsen e, Morten Frydenberg f, Anders Fuglsang-Frederiksen a,
a

Department of Clinical Neurophysiology, Aarhus University Hospital, Aarhus, Denmark Department of Neurology, Aarhus University Hospital, Aarhus, Denmark c Department of Neurology, Vejle Hospital, Vejle, Denmark d Department of Neurology, Aalborg Hospital, Aarhus University Hospital, Aalborg, Denmark e Department of Neurology, Holstebro Hospital, Holstebro, Denmark f Department of Biostatistics, School of Public Health, Aarhus University, Aarhus, Denmark
b

See Editorial, pages 433434

a r t i c l e

i n f o

h i g h l i g h t s
 MUNIX showed lower coefcient of variation than incremental stimulation MUNE.  MUNIX and incremental stimulation MUNE had similar diagnostic accuracy.  During ALS progression MUNIX showed more pronounced changes than incremental stimulation MUNE.

Article history: Available online 4 October 2012 Keywords: Motor unit number estimation Incremental stimulation MUNIX Amyotrophic lateral sclerosis Testretest reproducibility Surface-recorded motor unit potentials

a b s t r a c t
Objective: This study compares the new Motor Unit Number Estimation (MUNE) technique, MUNIX, with the more common incremental stimulation MUNE (IS-MUNE) with respect to reproducibility in healthy subjects and as potential biomarker of disease progression in patients with ALS. Methods: Thirteen ALS patients and 48 control subjects were prospectively investigated both groups were studied with MUNIX and IS-MUNE applied on the abductor digiti minimi (ADM) muscle. Additional retest was performed on 14 control subjects. Follow-up tests were carried out on 6 patients. The analysis included measures of reproducibility (Intraclass Correlation Coefcient (ICC)) and diagnostic performance (Receiver Operating Characteristic (ROC) analysis). Results: Testretest reproducibility was low to moderate for MUNIX and IS-MUNE (ICC = 0.38 and 0.56, respectively). Repeated MUNIX and IS-MUNE measurements on the same subject had a mean percentage difference (MPD) of 20% and 46%, respectively (p = 0.039). In the control group, the coefcient of variation was markedly lower for MUNIX than for IS-MUNE (26% and 44%, respectively, p < 0.0005). In ALS patients MUNIX had a notably better responsiveness in follow-up than IS-MUNE (percent change per month, 9.4 versus 5.6, p = 0.046). ROC analysis suggested similar diagnostic accuracy of both tests. Conclusions: MUNIX is a useful MUNE indicator when assessing progression of lower motor neuron affection in ALS. Furthermore, MUNIX displayed lower intrasubject variability, but no evident better diagnostic yield compared with IS-MUNE. Signicance: This study has established comparative assessment of MUNIX and IS-MUNE performance in testretest setting and as diagnostic tests on a distal muscle in ALS patients. 2012 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.

Abbreviations: ADM, abductor digiti minimi muscle; AH, abductor hallucis muscle; ALS, amyotrophic lateral sclerosis; ALSFRS-R, revised amyotrophic lateral sclerosis functional rating scale; APB, abductor pollicis brevis muscle; AUC, area under curve; BB, biceps brachii muscle; CMAP, compound muscle action potential; CV, coefcient of variation; DQEMG, decomposition-based quantitative electromyography; EDB, extensor digitorum brevis muscle; EMG, electromyography; FEV(1), forced expiratory volume in the rst second; FVC, forced vital capacity; ICC, intraclass correlation coefcient; IS-MUNE, incremental stimulation motor unit number estimation; LMN, lower motor neuron; MPD, mean percentage difference; MPS, multiple-point stimulation method; MRC, medical research council (UK); MU, motor unit; MUAP, motor unit action potential; MUNE, motor unit number estimation; MUNIX, motor unit number index; MUSIX, motor unit size index; ROC analysis, receiver operating characteristic analysis; SD, standard deviation; SEM, standard error of mean; SIP, surface interference pattern; SMUP, surface-recorded motor unit potentials; TA, tibialis anterior muscle. Corresponding author. Address: Department of Clinical Neurophysiology, Aarhus University Hospital, Noerrebrogade 44, building 10, DK-8000, Aarhus, Denmark. Tel.: +45 78463100; fax: +45 78463140. E-mail address: andefugl@rm.dk (A. Fuglsang-Frederiksen). 1388-2457/$36.00 2012 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.clinph.2012.08.023

J. Furtula et al. / Clinical Neurophysiology 124 (2013) 610618

611

1. Introduction Motor neuron loss in amyotrophic lateral sclerosis (ALS) is usually assessed indirectly by parameters such as strength, signs of chronic partial denervation in EMG (e.g. in motor unit action potential (MUAP) analysis or peak ratio interference pattern analysis), functional rating scales, spirometry test, and compound muscle action potential (CMAP) measurements. In contrast, Motor Unit Number Estimation (MUNE) reects deterioration of motor neuron function unaffected by the compensatory effects of reinnervation during disease progression. MUNE is currently the best predictor of disease progression in follow-up studies of ALS (Armon and Brandstater, 1999; Bromberg et al., 1993; Felice, 1997; Yuen and Olney, 1997) with evidence of MUNE being a better predictor of survival and time to tracheotomy than forced vital capacity or isometric muscle strength (Armon and Brandstater, 1999; de Carvalho et al., 2005). The rst MUNE method was described by McComas et al. (1971) (increment counting technique (McComas et al., 1971), later termed incremental stimulation MUNE (IS-MUNE) (Kong et al., 2009). Different methods have since been developed with the aim of making them more reliable and applicable in a clinical setting, e.g. the more recently introduced method Motor Unit Number Index (MUNIX) (Nandedkar et al., 2003, 2004). MUNIX uses the surface interference patterns (SIP) recorded during voluntary contractions to extract the average size of surface-recorded motor unit potentials (SMUP). MUNIX has not been compared with the IS-MUNE technique in the same normal subjects or ALS patients. In the present study we compared the reproducibility of MUNIX and IS-MUNE by investigating the within-subject variability and the Intraclass Correlation Coefcient (ICC) in a testretest setting on the abductor digiti minimi (ADM) muscle of normal subjects. Both techniques were in addition applied on ALS patients re-examined after 3- or 6 months in order to assess their respective responsiveness. We compared the two techniques strength as diagnostic parameters distinguishing pathologic from normal state. 2. Methods 2.1. Participant demographics 2.1.1. Control subjects Forty-eight untrained healthy subjects (25 females and 23 males) age range 2670 years, mean 44.4 years with no history of pathology in upper limb volunteered to participate in the study. Among the included healthy subjects, 14 persons (9 males, 5 females) aged 2658 years (mean 36 years) participated in the testretest study. 2.1.2. Patients ALS patients diagnosed in the period January 2007October 2010 with ability to stand, with normal cognitive function and without posture dependent respiratory deciency were eligible to participate in the trial. The patients were referred from four hospitals in two regions of Denmark from July 2008 to March 2011. All patients fullled the diagnostic criteria for ALS according to the El Escorial criteria (Brooks, 1994) and were regularly undergoing follow-up examinations by local ALS teams with 3 month intervals. At inclusion, one patient was diagnosed with denite ALS, 3 patients were in the category of probable ALS and 7 patients fullled the criteria of probable-laboratory supported ALS. Three patients were included in the category of clinically suspected ALS in order to obtain borderline changes; one patient did not progress within 2 years and was excluded from the

study. At the time of manuscript submission two patients in the category of clinically suspected ALS had progressed to probable ALS, thereby fullling the revised El Escorial criteria (Brooks et al., 2000). Thus, thirteen ALS patients were included in the study (Table 1). During follow up 3 patients changed from probable-laboratory supported to probable ALS. Six patients completed follow-up examination, which was performed 3 6 months after initial examination (mean 150 days range 99 204 days). Follow up examination was not technically feasible in two patients due to severe muscular atrophy. The remaining ve patients could not participate in the follow up examination due to their physical deterioration resulting in difculties in attending the second examination (Table 1). The clinical characteristics of the 13 patients at point of inclusion are summarized in Table 1. Eight patients were males and 5 were females, mean age was 65 years (range 5279). Disease duration from the rst symptom to inclusion ranged from 7 months to 7 years. Mean revised ALS functional rating scale (ALSFRS-R) (Cedarbaum et al., 1999) scores were documented initially for all patients together with forced vital capacity (FVC) and forced expiratory volume in one second (FEV(1)). Muscle force was evaluated using the Medical Research Council (MRC) scale. 2.2. Neurophysiological investigations The abductor digiti minimi (ADM) muscle of the right upper limb in healthy subjects was tested with MUNIX followed by IS-MUNE both techniques performed by the same investigator and with the same electrode placement. In the patient group, tests were performed on the most affected side in the same order as in control subjects. The skin was abraded and cleansed with alcohol rubbing pads. Signals were recorded by KeyPoint Classic (Medtronic, Inc.) using commercially available, disposable pre-gelled surface electrodes (Model 9013S0212; Alpine Biomed, Denmark). Percutaneous stimuli were delivered by bipolar stimulation electrodes (Model 9013L0362) over the ulnar nerve at the wrist, and signals recorded from ADM in a bellytendon montage. A gelled surface ground electrode (Model 71410-M/1; Alpine Biomed, Denmark) was placed between the recording and stimulating electrodes. Recording electrode was not moved in order to optimize the CMAP amplitude. Limb temperature was maintained between 32 and 34 C by a heating lamp. The CMAP was recorded at supramaximal stimulation and CMAP negative amplitude and area were measured. Filter settings were 2 Hz/10 kHz. 2.2.1. MUNIX The MUNIX technique is previously described by Nandedkar et al. (2003, 2004). The interference patterns were recorded at voluntary activity with gradually increasing force from slight to maximum activity in 9 steps by surface EMG (KeyPoint Classic) using a gain of 200 lV/division and lter settings of 2 Hz/4 kHz. Isometric force increments during abduction of digit V were obtained with gradual increase in resistance exerted by the examiner at the fth proximal interphalangeal joint on the supinated hand. After reaching 50% of maximal voluntary contraction, we included 30 s pauses between recordings in order to avoid fatigue (Fuglsang-Frederiksen, 2006). The degree of force increment was estimated by the resistance given by the examiner and by the amplitude and the fullness of the interference pattern. Maximum contraction was performed initially as a guideline for increments. Recordings were made within 12 s of contraction and maximum contraction was typically reached within 3 min. MUNIX was estimated on the basis of assumption of identical MU size and no superimposition at minimal force level (Nandedkar et al., 2003). The average SMUP

612

J. Furtula et al. / Clinical Neurophysiology 124 (2013) 610618

Table 1 The clinical characteristics of ALS patients at the time of inclusion. NRa 1 2 3 4 5 6 7 8 9 10 11 12 13 Age 63 62 70 60 70 65 71 79 70 65 60 69 52 El Escorial classicationb Denite Probable-lab Probable-lab Probable Probable-lab Probable-lab Probable-lab Probable Probable-lab Clinically suspectedb Probable Probable-lab Clinically suspectedb Onset type Bulbar Limb Limb Limb Limb Bulbar Limb Limb Limb Limb Limb Bulbar Limb Symptom duration (months) 35 30 23 33 17 7 16 18 84 53 60 13 26 FRS-R score 31 45 44 37 47 45 46 42 26 34 40 45 44 MRC (ADM) 4 4 4 5 4 4 4 4 4 4 4 4+ 4+ LMNc +++ +++ +++ +++ +++ ++ +++ ++ +++ + +++ ++ MRC (ADM) 4 4 4 5 4 4 4 4 4 4 4 4+ 4+ Atrophy (ADM) + + + + + + + + + + + +

Abbreviations: LMN, lower motor neuron; ADM, abductor digiti minimi muscle; MRC, medical research counsil; FRS-R, revised amyotrophic lateral sclerosis functional rating scale. a All patients received riluzole apart from patient 1 attributable to adverse effects. Three patients were treated for comorbid depression or anxiety with lithium (patient 2) and citalopram (patient 11 and 12). Patients who completed the follow-up examination are marked in bold. b El Escorial criteria from 1994 Brooks (1994) included clinically suspected ALS with LMN only in at least 2 regions. Both patients progressed to probable ALS within a year after inclusion. Applying Awaji-shima criteria de Carvalho et al. (2008) did not result in any of the patients moving to a higher diagnostic category. c LMN signs in the cervical region (): Fasciculations; Muscular atrophy; Flaccid paresis and Areexia/hyporeexi.

amplitude was expressed by Motor Unit Size Index (MUSIX) and derived by dividing the CMAP amplitude with the obtained MUNIX value (Nandedkar et al., 2010). 2.2.2. IS-MUNE IS-MUNE was performed by stimulations delivered with gradually increasing intensities until the rst motor unit on the surface EMG was evoked. The stimulus was then gradually increased such that the evoked response raised in a stepwise all-or-none manner according to recruitment of additional single motor units, as described by McComas et al. (1971). The response was regarded as stable after observing the identical response two to three times. KeyPoint Classic software performed template subtraction of successively evoked responses. The lower limit of recorded increment amplitude was 25 lV. The average SMUP area was calculated based on the average negative-peak area at 10 discrete levels. MUNE was estimated as area ratio between CMAP and average SMUP. 2.2.3. Testretest study Test and retest studies in control subjects were conducted on separate occasions (test interval ranged from 0 to 85 days, mean 10 days) with reapplication of stimulation and recording electrodes. Test and retest with IS-MUNE were conducted in continuation of MUNIX, with the same placement of electrodes. 2.3. Statistics Gender and age dependence was analysed by multiple regression analysis. Correlation between MUNIX and IS-MUNE was performed on absolute values and after being transformed to standardized z-scores. In the case of causal relation between two variables and thereby theoretically expected correlation, regression analysis was performed. Differences between test and retest values was evaluated by BlandAltman analysis (Bland and Altman, 1986) and intraclass correlation coefcient (ICC) using two-way mixed model (Shrout and Fleiss, 1979). ICC expresses reproducibility of measurements as the ratio of the inter-subject variance to the total sample variance and assumes values between 0 and 1, where 1 is synonymous with perfect reproducibility. The clinical signicance of ICC was interpreted as follows: good, ICC  0.75; moderate, 0.40  ICC < 0.75; and poor, ICC < 0.40

(Fleiss, 1981). Reproducibility was further assessed by the mean percentage difference (MPD). Percentage difference is calculated as the absolute difference between test and retest values divided by their mean value. Discrepancies in intraindividual variation between methods were tested by Pitmans test of difference in variance. We compared groups of measurements with the Wilcoxon signed rank tests for paired samples. To determine which single MUNE marker had the best discriminative potential, we compared these methods diagnostic accuracy in differentiating between patients and normal subjects by Receiver Operating Characteristic (ROC) curve analysis (Metz, 1978). The main parameter of interest is the area under the curve (AUC), for a perfect test, the area is 1 and for a test with no discriminative power, the area is 0.5. The areas under two ROC curves were compared using two linear prediction model and chi-squared test. Diagnostic thresholds were calculated by lower limit of normal and their condence intervals by binomial distribution. These estimates are, however, dependent on the distribution of patients and control subjects in the sample and 2sd as diagnostic threshold which may not always be the most optimal threshold. ROC analysis is performed to approach this problem. The change in MUNE values at follow-up was analyzed on standardized scale and values compared by Wilcoxon signed rank tests for paired samples. A p value < 0.05 was regarded as signicant. All statistical analyses were performed using STATA software version 10.0 (Statacorp LP, College Station, TX). 2.4. Ethics Before participation, each subject was informed of the purpose and potential risks of the study and gave his written voluntary consent. The study protocol was reviewed and approved by the local ethics committees and was performed according to the declaration of Helsinki. 3. Results In the control group MUNIX mean values were 176 (sd 46) and IS-MUNE mean values were 245 (sd 107). In the ALS patient group MUNIX mean values were 55 (range 2221) and IS-MUNE mean values were 78 (range 15209). In normal subjects there was no

J. Furtula et al. / Clinical Neurophysiology 124 (2013) 610618 Table 2 Summary of MUNIX and IS-MUNE testretest results in healthy subjects. N CMAP Test Retest MUNIX Test Retest IS-MUNE Test Retest 14 11.7 (2.4) 11.3 (2.3) 14 184 (38) 172 (35) 14 258 (144) 267 (135) 55.8 50.6 45.5a (41) 92 (63.6133.4) 98 (56.2171.5) 20.9 20.3 19.8a (15) 29 (20.042.0) 21 (9.547.0) 20.5 20.6 11.0 (12) 1.2 (0.841.8) 2.0 (1.33.1) Mean (SD) CV (%) MPD (%) (SD) SDr (95% CI) SDb (95% CI) ICC (95% CI)

613

0.72 (0.440.91)

0.38 (0.050.80)

0.56 (0.190.85)

Abbreviations: CV, coefcient of variation; MPD, mean percentage difference; SDr, residual standard deviation; SDb, between-subject standard deviation; ICC, intraclass correlation coefcient. a There was a signicant difference between the intraindividual variations of MUNIX and IS-MUNE: positive Pitmans variance ratio test: ratio of standard deviations = 0.3020, 95% CI, 0.16790.5432, t = 5.275, df = 12, p < 0.0005.

gender inuence on MUNIX and IS-MUNE values, and there was no change with age. The coefcient of variation was lower in MUNIX than in IS (control group, 26% and 44%, respectively). No correlation was found between MUNIX and IS-MUNE values in the control group (z-score, r = 0.090, p = 0.54) or in the patient group (z-score, r = 0.23, p = 0.45). There was no correlation between MUNE values and disease duration. MUNIX values tended to be higher in patients with higher ALSFRS-R score at inclusion, but the observed correlation was not statistically signicant (Spearmans rho = 0.50, p = 0.08). There was no correlation between ALSFRS-R score and IS-MUNE. 3.1. Testretest analysis of MUNE results and comparison between techniques Test and retest results for MUNIX and IS-MUNE are summarized in Table 2. There was a correlation between IS-MUNE testretest values (Pearson r = 0.53, p < 0.049), but not between MUNIX values (Pearson r = 0.39, p > 0.17). No systematic bias was found at test and retest by either MUNIX or IS-MUNE (t = 1.10, p > 0.29 and t = 0.25, p > 0.81). BlandAltman analysis showed that testretest differences in the values did not depend on the size of MUNE

(Spearmans rho = 0.032, p > 0.83). MUNIX had estimated residual sd (also referred as measurement error) of 29, meaning that the true value with 95% certainty lies within 29 in either direction of the predicted (tted) MUNIX value. Measurement error for ISMUNE was 92. When assessing testretest reproducibility by ICC, we found low reproducibility of MUNIX, but moderate reproducibility for IS-MUNE (0.38 versus 0.56) (Table 2). MPD for MUNIX and IS-MUNE was 20% and 46%, respectively (p = 0.039).

3.2. Relationship between MUNE, CMAP and SMUP CMAP amplitude was linearly related to MUNIX values (Fig. 1) in patients (R-squared = 0.89, p < 0.0005) and in control subjects (R-squared = 0.56, p < 0.0005), with no signicant difference in slope (p = 0.23). Six patients had moderately preserved CMAP

Fig. 1. Relationship between CMAP and MUNIX. Comparison of MUNIX and CMAP amplitude in control subjects (empty circles, n = 48) and subjects with ALS (lled circles, n = 13). Dash line and shaded area represents linear regression t in control subjects with 95% condence interval. Solid line represents linear regression t in ALS patients. Relationship between CMAP and MUNIX displayed signicant correlation in the control- as the patient group (y = 15.8x + 2.6; R-squared = 0.56, p < 0.0005 and y = 19.7x 32.7; R-squared = 0.89, p < 0.0005, respectively). There was no difference between slopes of regression lines for patients and control subjects (p = 0.23).

Fig. 2. Comparison of motor unit size index (MUSIX) and Motor Unit Number Index (MUNIX) in (A) 48 control subjects and (B) 13 subjects with ALS. (A) Negative linear relationship between MUNIX and MUSIX (y = 0.1683x + 94.244, Rsquared = 0.4065, p = 0.030). (B) Distribution of points may reect different degrees of reinnervation in patients included at different stages of the disorder. The relationship is best described by a power function: y = 791.59x0.544, Rsquared = 0.72, p < 0.001.

614

J. Furtula et al. / Clinical Neurophysiology 124 (2013) 610618

300

IS-MUNE
100 0 control patient 0 100 200 200

MUNIX

300

400

400

500

500

MUNIX values in ADM

IS-MUNE values in ADM

control

patient

Fig. 3. Comparison of surface-recorded motor unit potential (SMUP) area and incremental stimulation MUNE (IS-MUNE) in (A) 48 control subjects and (B) 13 subjects with ALS. (A) Negative linear relationship between IS-MUNE and SMUP area (y = 0.0017x + 0.962, R-squared = 0.3242, p < 0.0005). Removing a single subject with IS-MUNE of 29.7 changes slightly the regression line equation (y = 0.0011x + 0.778, R-squared = 0.3478, p < 0.0005). (B) The relationship between IS-MUNE and SMUP area in ALS patients is tted by a power function: y = 3.3187x0.441, R-squared = 0.3134, p < 0.05. Point distribution resembles pattern of motor unit size index (MUSIX) and MUNIX.

Fig. 4. MUNIX and IS-MUNE in normal subjects and subjects with ALS. The mean 2sd is shown with the means indicated by crosses and 2sd with solid lines. Using 2sd as the discriminating point, number of true positive with MUNIX was 10 sensitivity of the test 0.77. In the same setting sensitivity of IS-MUNE was 0.31. Control subject with IS-MUNE of 29 is not fullling the criteria for an outlier. Removing the value changes the mean and lower limit of normal to 250 and 42.6, respectively. Sensitivity and specicity of IS-MUNE changes to 0.38 (95% CI, 0.18 0.65) and 1 (95% CI, 0.911), respectively.

around 5 mV, but MUNIX values <30, indicating that CMAP amplitude was relatively preserved despite progressive decline in MUNIX due to large MUs. In the patient group the relation between MUSIX and MUNIX was best described as an inverse power function, i.e. the motor unit size increased with decreasing MUNIX (p < 0.001, Fig. 2B). There was a negative linear relationship between MUSIX and MUNIX in control subjects (R-squared = 0.4065, p < 0.0005) (Fig. 2A). Similar negative relationships were found between SMUP area and IS-MUNE in control subjects (linear regression, R-squared = 0.3242, p < 0.0005) (Fig. 3A) and in the patient group (power regression, R-squared = 0.3134, p < 0.05) (Fig. 3B). 3.3. Diagnostic accuracy of MUNIX and IS-MUNE With mean 2sd of control subjects used as the lower limit of normal (Fig. 4), MUNIX yielded a sensitivity of 0.77 (95% CI, 0.49 0.93), and specicity of 1 (95% CI, 0.911) (Fig. 4). IS-MUNE yielded a sensitivity of 0.31 (95% CI, 0.120.58) and a specicity of 0.98 (95% CI, 0.881). ROC analysis showed a similar ability to differentiate between patients and control subjects within techniques (AUC, MUNIX: 0.8974 versus IS-MUNE: 0.9183) (Fig. 5). Similarity was also evident in the fraction of correctly classied diseased and non-diseased subjects of both methods (MUNIX 95.08% and ISMUNE 91.80%) at their corresponding best discriminating threshold (72 and 98, respectively). 3.4. Validity of MUNE in follow-up Six patients were studied twice with an interval between rst and second study ranging from 99 to 204 days (mean 150 days).

Fig. 5. ROC analysis evaluation of the diagnostic performance of MUNIX and ISMUNE for discriminating between normal controls and ALS patients. (ROCAUC): area under the receiver operating characteristic curve. Comparison of the receiver operator characteristic (ROC) curves for the MUNIX and IS-MUNE parameters. Each contains 62 points and has some concavities. However, the areas under the curves were not signicantly different. Areas under the ROC curves are as follows: ISMUNE 0.9183 (95% CI, 0.831.0), MUNIX 0.8974 (95% CI, 0.761.0).

CMAP amplitude decreased at follow-up (Wilcoxon signed rank test; z = 2.38, p = 0.017) (Fig. 6). MUNIX decreased (p = 0.017), but IS-MUNE did not (p = 0.1159). MUNIX, expressed as percent change from baseline, declined by 9.4% (sd 4.6) per month, while CMAP declined by 7.11% (sd 3.83) per month. The responsiveness of MUNIX

J. Furtula et al. / Clinical Neurophysiology 124 (2013) 610618

615

12

10

200

150

ISMUNE
0 2 4 6 8

MUNIX

CMAP

100

50

0 0

50

100

150

200

Followup time (months)

Followup time (months)

Followup time (months)

Fig. 6. Individual values of CMAP, MUNIX and IS-MUNE of 6 patients at inclusion and during follow-up. Mean CMAP decreased from 6.6 (sd 3.6) to 4.7 (sd 3.3) during followup time (Wilcoxon mathed-pairs signed-rank test, p = 0.017). Mean MUNIX decreased from 95 (sd 86) to 56 (sd 75) during follow-up time (Wilcoxon mathed-pairs signedrank test, p = 0.017). IS-MUNE values changed from 113.9 (56.9) to 73.6 (sd 37.5), but not signicantly.

in follow-up (percent change per month) was signicantly better than of IS-MUNE (p = 0.046). There was no signicant change in MUSIX amplitude or SMUP area at follow up (p = 0.1730 and p = 0.29, respectively). There was a correlation between percent decrease in MUNIX at follow-up and corresponding increase in MUSIX (Spearman rho = 0.83, p = 0.042). The same correlation was not observed between percent decrease in IS-MUNE and corresponding increase in SMUP area (Spearman rho = 0.77, p = 0.072). 4. Discussion In the present study we investigated reproducibility, diagnostic performance and ability to assess disease progression in ALS patients with the MUNE methods MUNIX and IS-MUNE. 4.1. MUNIX and IS-MUNE ndings in control subjects MUNIX values in ADM of healthy volunteers (176 46) were similar to the earlier ndings in the literature (Table 3). IS-MUNE values of ADM (245 122) were lower than previously reported (380 79) (McComas et al., 1974), probably due to inherent methodological differences (Ballantyne and Hansen, 1974). There was no correlation between MUNIX and IS-MUNE, most likely due to inherent methodological features of both methods. Few studies have compared MUNE values obtained by two (or more) methods in the same subjects. MUNIX has previously been compared to high-density multiple-point stimulation method (MPS) only (Boekestein et al., 2012) showing no correlation in healthy subjects. MPS in healthy subjects has been compared with spike-triggered averaging (Doherty et al., 1993) and with the statistical method (Lomen-Hoerth and Olney, 2000) nding a correlation, while a recent MPS study with high-density surface electrode found no correlation with the statistical method (Blok et al., 2010) addressing variance due to alternation as the main cause. This lack of correlation between MUNE techniques can probably be explained by technical and physiological differences between them. MUNIX is based on area and power of CMAP and of interference patterns, while IS-MUNE in the present study is based on area measurements of CMAP and of increments. Additionally, IS-MUNE has

a clear physiological base assessing number of MUs, while MUNIX produces an index reecting number of MUs. Sample bias of recruited MUs at low stimulus intensities due to so-called size principle of MU recruitment (Henneman, 1957) has been emphasized as a possible error in early studies of MUNE (Brown and MilnerBrown, 1976; Milner-Brown and Brown, 1976). By contrast, surface interference patterns in MUNIX are recorded during a range of voluntary contractions, taking size principle of MU recruitment pattern into account. The average size of MUs in healthy subjects (assessed by MUSIX) increased with decreasing number of MUs in the muscle (Fig. 2A). This physiological relation between MU size and recruited MU number may narrow variance of maximal voluntary contraction force in healthy subjects. The observed coefcient of variation for MUNIX was lower (21%) than the previously published 28% (Nandedkar et al., 2010), while MPD was slightly higher (19.8%) compared to 16.8% (Nandedkar et al., 2010) and 15.3% (Ahn et al., 2010). ICC was lower (0.38) than ICC of 0.64 (Nandedkar et al., 2010) and 0.67 (Neuwirth et al., 2011). The low ICC for MUNIX in the current study can be explained by low between-subject variance relative to the within-subject variance (Shrout and Fleiss, 1979). The ICC is an index of reproducibility with xed range between 0 and 1 where the error term (within-subject variance) is weighted against the between-subject variance. Caution should be applied when interpreting ICC since a small between-subject variance (due to homogeneity of the sample or the quantitative trait of the parameter) will result in decrease of ICC. The conventional quantitative needle-detected measurements of MUAP amplitude and duration meet the same limitations in assessment of reproducibility by ICC (Ives and Doherty, 2012). Despite higher ICC, IS-MUNE exhibits a larger degree of variability from test to retest at different days. We found MPD of 46%, a large discrepancy to the recently reported study examining the intra-rater reliability with MPD of 21% (Liu et al., 2009), but more comparable to 34% initially reported by McComas et al. (1971). Experimental errors like positioning of the electrodes, the posture of the patient, time of data acquisition, and the recording equipment were negligible in present setting when comparing MUNIX and IS-MUNE. Nandedkar et al. (2010) used the initially recorded CMAP amplitude as guide to follow-up investigation. In our study, the

616

Table 3 Overview of testretest studies on MUNIX. MUNE method MUNIX Reference values Muscle Nandedkar et al. (2004) Ahn et al. (2010) ADM ADM No of subjects 33 normal subjects 62 normal subjects 22 ALS patients Nandedkar et al. (2010) Neuwirth et al. (2011)a ADM ADM 34 normal subjects 66 normal subjects Age < 60 years Age  60 years 30 normal subjects 5 PP patients 19 ALS patients 17 ALS patients MUNIX1 mean (SD) 148 (37) 142 (42) 158 (40) 163 (47) 172 (49) 150 (41) 103 (26) Intersubject CV 25 29.58 25.32 28.83 Retest Evaluation parameter r InterInterInterInterand and and and intraoperator intraoperator intraoperator intraoperator r MPD r MPD No of muscles 20 62 22 15 66 Resultsretest 0.72 J. Furtula et al. / Clinical Neurophysiology 124 (2013) 610618 0.74 and 0.86, respectively 17.5% and 15.3%, respectively 0.95 and 0.93, respectively 23.7% and 24.0%, respectively (16.8%) 0.64 0.68 and 0.67, respectively 0.7 0.94 (12%) 0.97 0.94 (18%, 12%) 0.89 0.91 (20.5%, 20%) 0.74 (19.4%) 0.86 (13.7%) ADM 14 normal subjects 184 (38) 20.65 Intraoperater ICC (MPD, median) 14 0.38 (19.8%, 14.0%) 0.56 (45.5%, 35.5%)

(MPD) ICC Inter- and intraoperator ICC r r (MPD) ICC Interoperator r (MPD, median) ICC Interoperator r (MPD, median) ICC (median percent difference)

Sandberg et al. (2011) Nandedkar et al. (2011) Boekestein et al. (2012) Compared to HD-MPS Our study Compared to IS-MUNE

TA ADM APB APB

25.24

19b 18 17

Abbreviations: PP, prior poliomyelitis; MPD, mean percentage difference: the relative difference was calculated as the absolute value of the difference between test and retest values divided by their mean value and expressed as a percentage. MPD is the mean of the obtained distribution. Although, in a single study Ahn et al. (2010) the relative difference was calculated as the mean of the absolute differences divided by the rst obtained value (also termed coefcient of variation). CV, coefcient of variation; ICC, intraclass correlation coefcient. a Results from only ADM are outlined in the table. Five to six muscles were tested in a multicentre setting biceps brachii, BB, abductor digiti minimi; ADM, abductor pollicis brevis; APB, tibialis anterior; TA, extensor digitorum brevis; EDB and abductor hallucis, AH. b Results from 10 normal subjects and 5 PP patients (4 tested bilaterally) were pooled.

J. Furtula et al. / Clinical Neurophysiology 124 (2013) 610618

617

examiner was blinded for the initially recorded CMAP amplitude and this may have introduced variation at CMAP level. The consequence of introducing previous recorded CMAP in a reproducibility study is critical when the new technique is aimed as a surrogate marker of disease progression and possible impact on prediction interval should be explored. 4.2. MUNIX and IS-MUNE ndings in ALS patients Reinnervation in ALS patients was suggested by a non-linear inverse (power) association between number of MUs and MUSIX, an indicator of the average SMUP (Fig. 2B). A similar relation was found between IS-MUNE and SMUP area (Fig. 3B). Among patients, 6 had moderately preserved CMAP around 5 mV and MRC of ADM between 4 and 5, but low MUNIX values (<30), indicating that CMAP and muscle strength were preserved due to reinnervation (Fig. 1). Similar non-linear association between MUNE and average SMUP are reported with average SMUP derived from IS-MUNE (Dantes and McComas, 1991) and MPS (Shefner et al., 2011). In ALS, where MU loss and concomitant reinnervation occur simultaneously, MUNE techniques like spike-triggered averaging technique, conventional MPS, IS-MUNE and MUNIX have displayed better reproducibility than in normal subjects (Ahn et al., 2010; Bromberg, 1993; Felice, 1995; Liu et al., 2009). Potential responses during increasing stimulation intensity are larger in patients with reduced number of motor units and underlying reinnervation, alternation would be reduced and reliability increased (Brown and Milner-Brown, 1976). MUNIX of hypothenar muscle in ALS has shown relatively high and consistent interrater reproducibility (Ahn et al., 2010; Nandedkar et al., 2011), but intrarater reproducibility assessed by MPD of 1824% (Ahn et al., 2010; Boekestein et al., 2012; Nandedkar et al., 2011) is low compared to the surprisingly low percent difference reported for IS-MUNE of 2.54.0% (Liu et al., 2009). 4.3. Diagnostic performance of MUNIX and IS-MUNE MUNIX differentiated between disease and normal status with higher sensitivity and the same specicity compared to IS-MUNE using conventional -2sd as lower limit of normal (sensitivity 77% vs. 31%, specicity 100% vs. 98%). Looking at the optimal test cutoff levels from ROC analysis, both methods had equal sensitivity of 77%; though specicity dropped to 91.8% (IS-MUNE) and 95.08 (MUNIX). Despite the high variability in IS-MUNE, its diagnostic potential was similar to that of MUNIX (AUC 0.9183 and 0.8974, respectively), thus probability of classifying patients correctly by each method was similar. There is no recorded data on time used to perform the tests. While MUNIX usually took less than 5 min to perform, IS-MUNE took at least twice as long to perform. 4.4. Potential of MUNE as biomarker of disease progression Previous studies used isometric strength, functional rating scales, spirometry parameters and MUNE in the course of disease progression in ALS and all have shown a decline with time (Dantes and McComas, 1991; de Carvalho et al., 2005; Liu et al., 2009). The rationale behind MUNIX was not to establish a diagnostic test, but rather to create a reliable test to monitor progression (Nandedkar et al., 2003). Our ndings, based on two time points for each patient, showed decline in CMAP and MUNIX, but not a signicant decline in IS-MUNE. MUNIX, expressed as percent change from baseline, declined in average by 7.48% (SEM 2.25) per month, in accordance with previously reported range of 2.7510% per month (Ahn et al., 2010; Nandedkar et al., 2010; Neuwirth et al., 2010). The absence of signicant reduction in IS-MUNE may be due to

its large variance between subjects and low reproducibility (high MPD) which increases the likelihood that the subjects ranking order would change by the second measurement. The smaller variance of MUNIX may have contributed to the detected changes of the motor unit architecture with MUNIX compared to IS-MUNE. Another reason for the inability of IS-MUNE to detect a decrease in MUNE can be the low number of patients able to attend the follow-up due to the progressive nature of the disease. A longer time period of follow-up would probably show a signicant decline in IS-MUNE. Two patients showed an unexpected increase in ISMUNE at follow-up examination, but no sign of technical error was noted. The fact that MUNIX changed signicantly and ISMUNE did not change may suggest that the techniques are measuring different properties. However, the question is whether a metric of unclear physiologic base that changes signicantly over time is better than one that may not change over a short time but has a clear physiologic base. In conclusion, MUNE is a promising tool for the assessment of diseases with motor unit loss, as it enables quantication and tracing of motor unit numbers in spite of simultaneous partly compensating collateral reinnervation. This makes it potentially useful for assessing the therapeutic benets of neuroprotective drugs in ALS. This report represents the rst study using MUNIX and IS-MUNE in comparative assessments of MUNE in healthy subjects and ALS patients. The current study demonstrates that MUNIX has better performance at follow-up than IS-MUNE. Acknowledgments This study was supported by the Lundbeck Foundation, the Danish Agency for Science, Technology and Innovation (founded by Danish Ministry of Science, Technology and Innovation) and the Dagmar Marshalls Foundation. These bodies had no role in the study design, implementation or manuscript preparation. The authors thank Per Christian Sidenius, MD and Bodil Holch Povlsen (Department of Neurology, Aarhus University Hospital) for their assistance with the admission of patients. The authors thank all patients and control subjects who participated in the study. References
Ahn SW, Kim SH, Kim JE, Kim SM, Kim SH, Park KS, et al. Reproducibility of the motor unit number index (MUNIX) in normal controls and amyotrophic lateral sclerosis patients. Muscle Nerve 2010;42:80813. Armon C, Brandstater ME. Motor unit number estimate-based rates of progression of ALS predict patient survival. Muscle Nerve 1999;22:15715. Ballantyne JP, Hansen S. A new method for the estimation of the number of motor units in a muscle. I. Control subjects and patients with myasthenia gravis. J Neurol Neurosurg Psychiatry 1974;37:90715. Blant JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:30710. Blok JH, van Dijk JP, Drenthen J, Maathuis EM, Stegeman DF. Size does matter: the inuence of motor unit potential size on statistical motor unit number estimates in healthy subjects. Clin Neurophysiol 2010;121:177280. Boekestein WA, Schelhaas HJ, van Putten MJ, Stegeman DF, Zwarts MJ, van Dijk JP. Motor unit number index (MUNIX) versus motor unit number estimation (MUNE): a direct comparison in a longitudinal study of ALS patients. Clin Neurophysiol 2012;123:16449. Bromberg MB. Motor unit estimation: reproducibility of the spike-triggered averaging technique in normal and ALS subjects. Muscle Nerve 1993;16:46671. Bromberg MB, Forshew DA, Nau KL, Bromberg J, Simmons Z, Fries TJ. Motor unit number estimation, isometric strength, and electromyographic measures in amyotrophic lateral sclerosis. Muscle Nerve 1993;16:12139. Brooks BR. El Escorial World Federation of Neurology criteria for the diagnosis of amyotrophic lateral sclerosis. Subcommittee on Motor Neuron Diseases/ Amyotrophic Lateral Sclerosis of the World Federation of Neurology Research Group on Neuromuscular Diseases and the El Escorial Clinical limits of amyotrophic lateral sclerosis workshop contributors. J Neurol Sci 1994;124(Suppl):96107. Brooks BR, Miller RG, Swash M, Munsat TL. El Escorial revisited: revised criteria for the diagnosis of amyotrophic lateral sclerosis. Amyotroph Lateral Scler Other Motor Neuron Disord 2000;1:2939.

618

J. Furtula et al. / Clinical Neurophysiology 124 (2013) 610618 McComas AJ, Fawcett PR, Campbell MJ, Sica RE. Electrophysiological estimation of the number of motor units within a human muscle. J Neurol Neurosurg Psychiatry 1971;34:12131. McComas AJ, Sica RE, Upton AR. Multiple muscle analysis of motor units in muscular dystrophy. Arch Neurol 1974;30:24951. Metz CE. Basic principles of ROC analysis. Semin Nucl Med 1978;8:28398. Milner-Brown HS, Brown WF. New methods of estimating the number of motor units in a muscle. J Neurol Neurosurg Psychiatry 1976;39:25865. Nandedkar DS, Barkhaus PE, Stalberg EV. Reproducibility of MUNIX in patients with amyotrophic lateral sclerosis. Muscle Nerve 2011;44:91922. Nandedkar SD, Barkhaus PE, Stalberg EV. Motor unit number index (MUNIX): principle, method, and ndings in healthy subjects and in patients with motor neuron disease. Muscle Nerve 2010;42:798807. Nandedkar SD, Nandedkar DS, Barkhaus PE, Stalberg EV. Motor unit number index (MUNIX): a pilot study. In: Bromberg MB, editor. Motor unit number estimation. Amsterdam: Elsevier; 2003. p. 13340. Supplements to Clinical Neurophysiology. Nandedkar SD, Nandedkar DS, Barkhaus PE, Stalberg EV. Motor unit number index (MUNIX). IEEE Trans Biomed Eng 2004;51:220911. Neuwirth C, Nandedkar S, Stalberg E, Barkhaus PE, Carvalho MD, Furtula J, et al. Motor unit number index (MUNIX): a novel neurophysiological marker for neuromuscular disorders; testretest reliability in healthy volunteers. Clin Neurophysiol 2011;122:186772. Neuwirth C, Nandedkar S, Stalberg E, Weber M. Motor unit number index (MUNIX): a novel neurophysiological technique to follow disease progression in amyotrophic lateral sclerosis. Muscle Nerve 2010;42:37984. Sandberg A, Nandedkar SD, Stlberg E. Macro electromyography and motor unit number index in the tibialis anterior muscle: differences and similarities in characterizing motor unit properties in prior polio. Muscle Nerve 2011;43:33541. Shefner JM, Watson ML, Simionescu L, Caress JB, Burns TM, Maragakis NJ, et al. Multipoint incremental motor unit number estimation as an outcome measure in ALS. Neurology 2011;77:23541. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull 1979;86:4208. Yuen EC, Olney RK. Longitudinal study of ber density and motor unit number estimate in patients with amyotrophic lateral sclerosis. Neurology 1997;49:5738.

Brown WF, Milner-Brown HS. Some electrical properties of motor units and their effects on the methods of estimating motor unit numbers. J Neurol Neurosurg Psychiatry 1976;39:24957. Cedarbaum JM, Stambler N, Malta E, Fuller C, Hilt D, Thurmond B, et al. The ALSFRSR: a revised ALS functional rating scale that incorporates assessments of respiratory function. BDNF ALS study group (Phase III). J Neurol Sci 1999;169:1321. Dantes M, McComas A. The extent and time course of motoneuron involvement in amyotrophic lateral sclerosis. Muscle Nerve 1991;14:41621. de Carvalho M, Scotto M, Lopes A, Swash M. Quantitating progression in ALS. Neurology 2005;64:17835. de Carvalho M, Dengler R, Eisen A, England JD, Kaji R, Kimura J, et al. Electrodiagnostic criteria for diagnosis of ALS. Clin Neurophysiol 2008;119:497503. Doherty TJ, Stashuk DW, Brown WF. Determinants of mean motor unit size: impact on estimates of motor unit number. Muscle Nerve 1993;16:132631. Felice KJ. A longitudinal study comparing thenar motor unit number estimates to other quantitative tests in patients with amyotrophic lateral sclerosis. Muscle Nerve 1997;20:17985. Felice KJ. Thenar motor unit number estimates using the multiple point stimulation technique: reproducibility studies in ALS patients and normal subjects. Muscle Nerve 1995;18:14126. Fleiss JL. Statistical methods for rates and proportions. New York, N.Y.: Wiley, John and Sons, Inc; 1981. p. 218. Fuglsang-Frederiksen A. The role of different EMG methods in evaluating myopathy. Clin Neurophysiol 2006;117:117389. Henneman E. Relation between size of neurons and their susceptibility to discharge. Science 1957;126:13457. Ives CT, Doherty TJ. Intra- and inter-rater reliability of motor unit number estimation and quantitative motor unit analysis in the upper trapezius. Clin Neurophysiol 2012;123:2005. Kong X, Han Z, Gozani SN. Incremental stimulation MUNE: automation with global path search and alternation pattern recognition. Suppl Clin Neurophysiol 2009;60:97103. Liu XX, Zhang J, Zheng JY, Zhang S, Xu YS, Kang DX, et al. Stratifying disease stages with different progression rates determined by electrophysiological tests in patients with amyotrophic lateral sclerosis. Muscle Nerve 2009;39:3049. Lomen-Hoerth C, Olney RK. Comparison of multiple point and statistical motor unit number estimation. Muscle Nerve 2000;23:152533.