C46 TEST‐RETEST AND INTER‐EXAMINER RELIABILITY OF AUTOMATED MOTOR UNIT NUMBER ESTIMATION (MUNE): COMPARISON OF MULTIPLE POINT STIMULATION AND STATISTICAL METHOD
Naito Y, Asahi M, Kuzuhara S
Department of Neurology, Mie University, Tsu, Mie, Japan
E‐mail address for correspondence: [email protected]
Background: Motor unit number estimation (MUNE) is a useful tool to assess the disease progression of ALS and other motor neuron diseases; however, it has not become popular in Japan because of its complicated procedures. Recently, two automated systems of common techniques for MUNE became available for clinical use – the multiple point stimulation (MPS) and the statistical method. This study compares these two automated techniques for MUNE to determine which is more reproducible.
Objectives: Fourteen normal controls aged 23–45 years were measured. Two neurologists skilled in electromyography participated in this study as the examiners.
Methods: Surface recorded motor unit action potentials (SMUPs) of the right hypothenar muscle (ADM) were measured on normal controls. For multiple point stimulation, 10 different threshold SMUPs were recorded and averaged automatically with the MPS MUNE program on the NIHON KOHDEN MEB‐9104. For the statistical method, mean SMUP amplitude was measured at several stimulus levels, typically spanning >40% of CMAP amplitude range, with the statistical MUNE program on the Nicolet Viking IV. Both techniques were performed by two examiners twice, and results averaged, electrodes changed, and all recording repeated, totalling eight times for each muscle.
Results: By the statistical method, the estimated motor unit number for normal controls was 135±19 (mean +/− standard deviation) for examiner 1 and 131±25 for examiner 2. There was no statistical significance in these numbers, and inter‐examiner correlation coefficient was 0.857 (p<0.05) for statistical method. Test‐retest correlation coefficient was 0.772 (p<0.05) for examiner 1 and 0.857 (p<0.05) for examiner 2.
By the MPS method, the estimated motor unit number for normal controls was 194±41 for examiner 1 and 183±53 for examiner 2. There was no statistical significance in these numbers, and the inter‐examiner correlation coefficient was 0.687 (p<0.05) for the MPS method. Test‐retest correlation coefficient was 0.736 (p<0.05) for examiner 1 and 0.739 (p<0.05) for examiner 2. Inter‐method correlation coefficient was 0.567 (p<0.05) for examiner 1 and 0.389 (ns) for examiner 2.
Discussion: The statistical method showed significantly smaller number than the MPS method by both of the examiners, and the inter‐method correlation coefficient was not significant. While the MPS method depends on discrimination of each SMUP, the procedure is not fully automated and needs more sophisticated skills.
Conclusions: Both the statistical method and the automated MPS method were reproducible, and easily used in clinical practice. The estimated motor unit numbers are not always compatible between the two methods, and mixed use is not recommended.
C47 THERE IS EVIDENCE FOR WHOLE BRAIN ATROPHY IN PRIMARY LATERAL SCLEROSIS AND IT APPEARS TO BE RELATED TO GREY MATTER ATROPHY COMPARED TO WHITE MATTER ATROPHY
Tartaglia MC, Findlater K, Rowe A, Lee D, Strong MJ
London Health Sciences Centre, London, ON, Canada
E‐mail address for correspondence: [email protected]
Background: Primary lateral sclerosis (PLS) is an idiopathic, neurodegenerative disorder of the upper motor neurons. The pathophysiology of PLS is still unknown. A number of small magnetic resonance studies have observed cortical atrophy, most notably in the cortical and subcortical portions of the precentral area. Atrophy of the motor cortex has been considered to be highly suggestive of PLS (1); however, not all studies have reported atrophy (2). This discrepancy may be due to the fact that most studies that have assessed atrophy have done so in a qualitative manner and thus only severe atrophy becomes apparent (3). In addition, it is unknown whether the atrophy is confined to the grey matter (GM), white matter (WM) or both.
Objectives: The aim of this study was to assess whether we could observe whole brain atrophy in patients with PLS and then to quantify atrophy and assess whether this was related to the loss of GM, of WM or both.
Methods: Patients with a clinically definite diagnosis of PLS were recruited from the neuromuscular clinic. High resolution T1‐weighted images were acquired in patients with PLS and controls. SIENAX, a package for the estimation of atrophy at a single time‐point, was used to calculate GM and WM volumes as well as whole brain volume (WBV). Mann‐Whitney tests were used to compare the two groups for differences in age, GM and WM volume and WBV.
Results: Eight patients and three controls have been scanned to date. There was no significant difference between the two groups' age: patients' mean age was 62.8±10.7 years and controls' mean age was 55.67±16.5 (p = 0.306) years. There was a trend towards the mean WBV being smaller in patients than controls: 1.42±0.09 vs. 1.51±0.09, respectively (p = 0.133). Mean GM was smaller in patients than in controls, 0.73±0.07 vs. 0.80±0.06, respectively (p = 0.194). WM volume was similar in patients and controls, 0.68±0.04 vs. 0.71±0.4, respectively (p = 0.414).
Conclusions: Our preliminary results suggest that there is whole brain atrophy in PLS and it is more closely related to GM atrophy than WM. These findings suggest that PLS pathology may be more closely related to cortical disease than the corticospinal tracts. WBV and more specifically GM volumes may help differentiate this disease from others.
References
- Pringle CE, et al. Brain 1992;115:495–520.
- Younger DS, et al. Arch Neurol. 1988;45:1304–7.
- Kuipers-Upmeijer J, et al. J Neurol Neurosurg Psychiatry 2001;71:615–20.
C48 VOLUMETRIC PROTON SPECTROSCOPIC IMAGING IN AMYOTROPHIC LATERAL SCLEROSIS
Sharma K, Govindaraju V, Bowen B, Domenig C, Steele J, Bradley W, Maudsley A
Miller School of Medicine, University of Miami, Miami, Florida, USA
E‐mail address for correspondence: [email protected]
Background: Previous MRS studies used either a single‐voxel or a 2D‐spectroscopic imaging (2D‐SI) technique to detect cerebral metabolites in ALS patients. These techniques obtain metabolic information from a single voxel or multiple voxels in a slice. To understand the pathophysiology of the disease process, it is important to evaluate metabolic changes in the whole brain. This is the first study to use proton 3D‐SI technique in ALS patients.
Objective: To assess the feasibility of a volumetric MRSI technique in detecting biomarkers for the upper motor neuron (UMN) dysfunction.
Methods: Eight patients with definite, sporadic ALS (54±11 years of age) and six controls (48±10 years) were scanned on a 3T MR scanner using a 3D‐EPSI sequence (FOV: 280×280×180 mm3, 50×50×18 phase encodes, slab thickness of 135 mm, TR/TE = 1710/70 ms, and 26 minutes acquisition time). Severity of the disease was assessed using ALSFRS and vital capacity, and UMN dysfunction was evaluated by finger tap in 10 seconds and foot tap in 10 seconds. Spectroscopic data were processed using software developed in‐house. Data from three ALS patients and one control were excluded due to poor quality. Data from eight regions of interest (precentral gyrus, corona radiata, internal capsule, midbrain and medulla) along the corticospinal tract were sampled to quantitate N‐acetyl aspartate (NAA), creatine (Cr) and choline (Cho). Metabolite data were normalized. A two‐tailed t‐test was used and a p‐value of <0.05 was considered significant.
Results: Important findings were: the normalized [NAA] was lower in patients with ALS compared to controls on both sides at the precentral gyrus (right: 28%, p = 0.04; left: 20%, p = 0.06), corona radiata (right: 22%, p = 0.05; left: 16%, p = 0.04) and internal capsule (right: 21%, p = 0.06; left: 18%, p = 0.005). The concentration of NAA did not reach a significant level at the left precentral gyrus and the right internal capsule but showed a trend towards it. There were correlations between the concentration of NAA at various levels along the motor pathway and the clinical measurements of UMN dysfunction (r = 0.4–0.9). There were correlations between the ALSFRS and disease duration (r = 0.4–0.5), vital capacity (r = 0.4–0.5) and concentration of NAA (r = 0.4–0.9).
Conclusions: The observed decrease in concentration of NAA and a trend for correlation with UMN dysfunction support previous findings that NAA may be used as a biomarker for UMN dysfunction. Our findings suggest that the whole‐brain SI technique is well suited to measure the cerebral metabolite changes along the motor pathway in ALS patients.
Acknowledgement: Funding was provided by the Stanley Glaser Foundation Award.
C49 A COMPARISON OF CEREBRAL HAEMODYNAMIC PARAMETERS IN PATIENTS WITH ALS AND PLS ACCOMPANIED BY CHANGES IN COGNITIVE IMPAIRMENT
Murphy MJ1, Grace GM2, Strong MJ2, Tartaglia MC2, Orange JB3, Kozak RI4, Chen X1, Rowe A2, Findlater K2, Lee T‐Y1
1Imaging Research Laboratories, Robarts Research Institute, London, ON, 2Department of Clinical Neurological Sciences, 3School of Communication Sciences and Disorders, University of Western Ontario, London ON, 4Department of Radiology, St Joseph's Health Centre, London, ON, Canada
E‐mail address for correspondence: [email protected]
Background: ALS and PLS are rare neurodegenerative diseases of the motor system. Increasingly, frontotemporal syndromes associated with cognitive impairment (CI) have been observed in ALS, often associated with concomitant decreases in brain perfusion. Although similar features have been observed in PLS, there is limited information contrasting ALS and PLS with respect to haemodynamic patterns in the presence or absence of CI.
Objectives: 1) To contrast cerebral haemodynamics in ALS and PLS patients with and without CI, and 2) to examine the relationship between CI and alterations in cerebral blood flow (CBF), cerebral blood volume (CBV), and mean transit time (MTT).
Methods: The patient groups consisted of 12 ALS patients (seven men, five women) aged 40 to 63 years (mean, 51.4±8.9 years), 18 PLS patients (nine men, nine women) aged 44 to 72 years (mean, 59.1±8.7 years), and six spousal controls (five men, one woman) aged 34 to 63 years (mean, 52.3±10.8 years). All patients and controls underwent NT followed by a computed tomography (CT) perfusion head scan to measure CBF, CBV, and MTT. Patients were stratified into two groups: 1) cognitively‐intact PLS and ALS patients (PLS and ALS; defined as those patients having either one or no abnormal scores on NT testing) and 2) PLS and ALS patients with cognitive impairment (PLSci and ALSci; defined as those patients having two or more abnormal test scores). Abnormality of a test score was determined as a score that fell more than 2.0 SD below the mean. Stratified groups were compared with each other using a one‐way ANOVA with post hoc Tukey tests to determine differences in haemodynamic parameters based on segmented regions of the brain corresponding to the four cortical lobes and deep grey matter structures. Results for the study were considered significant for p<0.05.
Results: One of 12 ALS and four of 18 PLS patients were considered cognitively impaired based on NT. Since the ALSci group only had one member, statistical analysis could not be performed with this group. The ALS and PLS groups were statistically similar in all regions for haemodynamic parameters with the exception of CBV in the frontal and temporal grey matter, where PLS had an elevated CBV. The PLSci group had significantly increased MTT in all regions of the grey and white matter compared to the control and ALS groups. Furthermore, the PLSci group had significantly increased CBV in all grey matter regions (except the parietal lobe) compared with the controls and in the frontal and temporal region compared with the ALS group. Although the ALS group was statistically similar to controls, there was a global trend for increased MTT and CBV in the ALS group.
Conclusion: PLS and ALS were statistically similar, both demonstrating increased trends of CBV and MTT in all regions compared to controls. The PLSci group displayed the greatest deficits in cerebral haemodynamics, namely MTT.
Acknowledgement: This research was funded by ALSA.
C50 THE PERFORMANCE OF CLINICAL OUTCOME MEASURES IN A PHASE II TRIAL OF GLATIRAMER ACETATE IN ALS
Gordon PH, Montes J, Diamond B, Doorish C, Rowland LP, Mitsumoto H
Columbia University, New York, NY, USA
E‐mail address for correspondence: [email protected]
Background: A standard set of clinical outcome measures, including the ALS Functional Rating Scale (ALSFRS‐R), manual muscle testing (MMT), forced vital capacity (FVC) and quality of life scales (QOL), are currently applied in phase III trials in ALS. While test performance has been analysed extensively in large, long trials, how they perform in shorter trials with fewer subjects has not been well studied. As the number of neuroprotective agents increases, there is a growing need for small efficient phase II trials to screen drugs and to select the correct dose for efficacy trials. Which outcomes are best suited for small early phase trials is not yet known.
Objective: To assess the performance of clinical outcome measures from a short‐duration small sample size phase II randomized controlled trial.
Methods: Data were analysed on the performance of the ALSFRS‐R, MMT, FVC and QOL from a phase II trial of glatiramer acetate in 30 patients. Outcomes were obtained at baseline and at months 1–6. The rate of change in the ALSFRS‐R at baseline (DeltaFS) was assessed to determine feasibility for stratification in future trials. Statistical comparisons were made using Pearson correlation coefficients, Kaplan‐Meier life table curves, and Cox proportional hazards models.
Results: Correlations were found between ALSFRS‐R/ MMT (r = 0.60, p = 0.0005), ALSFRS‐R/QOL (r = 0.50, p = 0.0044), and QOL/FVC (r = 0.42, p = 0.0197) at baseline. The slope of ALSFRS‐R correlated with slopes of MMT (r = 0.73, p<0.0001), FVC (r = 0.89, p <0.0001), and QOL (r = 0.58, p = 0.0007). The slope of FVC correlated with slopes of MMT (r = 0.69, p<0.0001) and QOL (r = 0.63, p = 0.0002). Baseline ALSFRS‐R (p = 0.0490) and FVC (p = 0.0005) predicted survival, while MMT did not (p = 0.3987). Slopes of ALSRS‐R (p = 0.0028) and FVC (p = 0.0030) were also found to predict survival. When dichotomized by baseline median scores, the DeltaFS (p = 0.0235, median = 0.5) and ALSFRS‐R (p = 0.0037, median = 33) were associated with survival. The DeltaFS, when broken down by tertile, was also associated with survival (p = 0.0063).
Conclusions: Standard clinical outcomes performed reliably in a short‐duration small sample size trial. The ALSFRS‐R baseline score and slope predicted survival. MMT and FVC, while reliable, were less clearly predictable in rates of decline and were less strongly predictive of survival. The ALSFRS‐R, because of its simplicity and reliability may be best suited for small clinical trials, and the rate of decline taken at baseline visit (DeltaFS) could be used to stratify patients by rate of progression. The efficiency of early phase clinical trials might be improved by using only one or a few highly reliable and meaningful outcome measures, such as the ALSFRS‐R.
C51 IDENTIFYING RESPONDERS TO INCREASE POWER AND SENSITIVITY IN ALS CLINICAL TRIALS
Moore DH, Miller RG
California Pacific Medical Center, San Francisco, CA, USA
E‐mail address for correspondence: [email protected]
Background: Variability in patient rates of decline supports the hypothesis that ALS is a heterogeneous disease. This suggests that it may be unrealistic to expect that a single agent will be uniformly effective for all patients. It would be desirable to define a statistical test that can detect responses in a subset of patients while other patients show no response.
Objective: To compare power of an analytic plan designed to detect a subset of responding patients with two conventional comparisons based on responses of all patients.
Methods: We used a linear mixed effects model to estimate pre‐ and post‐treatment changes in ALSFRS slope for each patient in two phase III negative trials of gabapentin and TCH346. These results were used to set a cut‐off point for defining response to treatment based on the change in slope. A test for treatment effect is based on counting the number of responders (i.e. those with changes greater than the cut‐off point) in treated compared to placebo arms of a clinical trial. Sample sizes for this test (Responder Test) were compared with those required for two conventional tests: a t‐test, comparing mean responses, and a non‐parametric test (Wilcoxon test) for all patients. Sample sizes for the Responder and Wilcoxon tests were obtained from tables in a publication describing rank tests designed for small proportions of responders (1).
Results: Estimates for initial slope and change in slope of ALSFRS after four months in the clinical trials were similar. One hundred and sixty‐six patients from a gabapentin trial (where treatment had no effect) and 110 placebo patients from a TCH346 trial had six or more monthly ALSFRS scores. Mean slopes during the initial four months were: 0.75(SD 0.78) units/month for gabapentin and −0.74(1.06) for TCH346 placebo. Mean slope changes after the first four months were −0.04(1.22) and −0.14(1.18), respectively. Subsets of patients in the gabapentin trial (22%) and in the TCH346 trial (23%) were fast progressors with initial slopes greater than −1.2 units/month. A slope change of 1 SD in either study would represent a slope reduction of ∼1.2 ALSFRS units/month, which might be more readily seen in the fast progressor subset. Slope changes of this magnitude or greater were seen in 13% of gabapentin and in 11% of TCH346 patients. Sample sizes for change of 1 or 2 SD and different fractions of responders are compared below:
Sample sizes (per arm) required for 80% power:
Conclusions: A phase III clinical trial examining responders with 1–2 unit reduction in ALSFRS slope could result in smaller sample size and more efficient and sensitive trials.
References
- Johnson RA, Verrill S, Moore DH II. Two-sample rank tests for detecting changes that occur in a small proportion of the treated population. Biometrics 1987;43:641–55.