Dispersion of mean consecutive differences in single‐fiber electromyography increases diagnostic sensitivity for myasthenia gravis

Abstract

Introduction: In this study we aimed to investigate the dispersion of mean consecutive difference (MCD) of concentric needle jitter studies of patients with myasthenia gravis (MG) and its effect on diagnostic sensitivity for MG.Methods: One hundred fifty-three patients, including 76 patients with MG and 77 controls with possible MG who later received another diagnosis, underwent stimulated concentric needle jitter studies of the frontalis muscle. MCD mean, standard deviation (SD), and coefficient of variation (CV) were calculated. Diagnostic sensitivity and specificity were determined using receiver operating characteristic (ROC) analyses.Results: MG patients showed a significantly greater MCD mean (MG: control, 26.3 μs; 13.5 μs [median]; P < .0001), MCD SD (MG: control, 12.8 μs; 5.1 μs [median]; P < .0001), and MCD CV (MG: control, 46.1; 37.5 [median]; P < .001) than those without MG. An ROC curve of SD showed a large area under the curve (0.88), and a cut-off value of 7.2 μs, which was calculated by maximum Youden index, exhibited high diagnostic sensitivity (86%) for MG. Combined MCD mean, outliers, and SD criteria showed higher sensitivity (88%) than conventional criteria alone (82%), at the expense of lower specificity. Five MG patients with normal MCD mean and abnormal MCD SD had only ocular symptoms.Discussion: The dispersion of MCD as measured by MCD SD greater than 7.2 μs is significantly increased in patients with MG and may be a useful measure of abnormal jitter in the diagnosis of MG, especially for identifying patients with mild disease.

The dispersion of individual jitter values is a sensitive diagnostic indicator of voluntary SFEMG

“…We found that the optimal cutoffs for SD according to Youden statistics were 13.5 microseconds for the OO, 12.6 microseconds for the FRO, and 11.8 microseconds for the EDC, whereas the areas under the receiver-operating characteristic curves were 0.89, 0.87, and 0.89 respectively. Similar to the results of Kojima et al, 5 these values were characterized by high sensitivity and specificity, which was comparable to those resulting when using the number of outliers as a criterion (Table 1). Abnormal jitter is not specific to MG, but to disturbed neuromuscular transmission of any cause.…”

“…Our data underscore the clinical utility of the method introduced by Kojima et al, 5 suggesting that the variability of signal transmission at the motor endplate inherent in MG can be reflected in the dispersion of individual jitter values in a single muscle SFEMG study. Its concordance with the established measure of jitter dispersion, namely the number of outliers in a single SFEMG session, is high but not perfect (Table 2).…”

“…Of practical value is the relatively high diagnostic specificity of SD, which outperforms that of the conventional criteria. Moreover, it should be noted that our optimal cut-off values were well above 10 microseconds for all three muscles, that is, markedly greater than the 7.2-microsecond cutoff of Kojima et al 5 This discrepancy may partly arise from methodological differences between voluntary and stimulated SFEMG, whereas the latter generally exhibits smaller SD values for MCD and outliers, at least for FRO and EDC. 3 The combination of SD with the conventional diagnostic criteria (by an inclusive logical "or") did not increase classification sensitivity for the diagnosis of MG.…”

“…The dispersion of individual jitter values appears to be particularly promising in this direction. In their recent article, 5 Kojima et al showed that the standard deviation (SD) of individual jitter values was significantly greater in myasthenia gravis (MG) patients than in control subjects. The authors employed stimulated SFEMG with a concentric needle and found that a cut-off value of 7.2 microseconds can separate myasthenic from normal frontalis (FRO) muscles with a sensitivity of 86%.…”

“…Herein we retrospectively assessed the sensitivity and specificity of the SD method in two muscles not assessed in the previous study, the orbicularis oculi (OO) and the extensor digitorum communis (EDC), using data on volitional rather than stimulated SFEMG from a previous study. 6 Dispersion was calculated by means of the SD of the 20 action potential pairs, according to the method of Kojima et al 5 examined by the electromyographer (in the raster and the superimposed mode) to assure that each spike complied with established criteria for potential waveforms. 3 SD values were compared with conventional SFEMG criteria for abnormality from our laboratory.…”