AI-based tool to assess muscles, nerves shows promise as biomarker
Data show dEMG works better than standard electromyography in SMA
A newer assessment method known as dEMB, for decomposition electromyography — which uses artificial intelligence (AI) tools to measure nerve and muscle function — strongly correlated with measures of physical function and was able to distinguish people with spinal muscular atrophy (SMA) who were able to walk from those who could not, a study reported.
The data suggested that dEMG was better at assessing SMA status and progression than standard electromyography methods, which typically are used to make an electrical recording of muscle activity.
dEMB “boasts technical ease of administration, high patient tolerability, and ability for consistent use throughout the lifespan, all of which make it both user- and patient-friendly,” the researchers wrote.
“Together these findings position dEMG as a compelling candidate biomarker for patients with SMA,” the team wrote.
The study, “Differential impact on motor unit characteristics across severities of adult spinal muscular atrophy,” was published in the Annals of Clinical and Translational Neurology.
Newer EMG type uses AI tools for analysis
SMA is characterized by the loss of motor neurons, the specialized nerve cells that connect the brain and spinal cord to muscles to control body movements. It results in muscles that do not receive the electrical signals that normally tell them to move, leading to muscle weakness and atrophy (wasting) over time.
Despite the meaningful impact of newer approved SMA treatments on motor function and survival, their effects can vary among patients and are not entirely understood.
“Thus, there is a need for informative and reliable biomarkers to assess disease status and monitor response to treatment,” the researchers wrote.
Electromyography, or EMG, is a test that’s often used in SMA patients to assess the health of motor units — which each consist of a motor neuron and all the muscle fibers it innervates, or supplies with energy. The electrical signals generated by muscle fibers belonging to motor units are called motor unit action potentials (MUAP).
dEMG is a newer type of EMG that applies AI or artificial intelligence approaches as tools to analyze signals from multiple electrodes on the skin surface. It assesses the sizes or amplitudes of action potentials and motor unit firing rates.
In this report, a team at the University of Missouri and The Ohio State University sought to determine the usefulness of dEMG as compared with standard EMG methods in people with SMA.
Their study involved a group of 28 adults with SMA, who had a mean age of 39. Among them, 11 were ambulatory, or able to walk, and 18 were non-ambulatory. All were being treated with Spinraza (nusinersen) for at least 10 months prior to the analyses.
dEMG data also were collected from eight unaffected individuals, who served as controls.
Significant differences were found in the amplitude of action potentials and in the average and peak firing rate of motor units between ambulatory and non-ambulatory patients and controls.
In fact, all three measures were the highest in SMA patients who were still able to walk, followed by controls. Non-ambulatory patients had the lowest measurements. The differences were significant between all three groups across all measures, with the notable exception of the average firing rate, which was similar for ambulatory patients and controls.
dEMB assessments shown to distinguish SMA patients able to walk
The researchers then examined if any of these dEMG parameters were associated with validated measures of motor function and strength. The team used the Revised Upper Limb Module (RULM), the patient-reported Revised SMA Functional Rating Scale (SMAFRS), and elbow flexion and elbow extension strength measures.
The results showed that all dEMG assessments correlated significantly with more than one of the clinical assessments both for patients who were able to walk and those who weren’t.
For example, the average action potential amplitude showed moderate to strong correlations with all four measures — and the greatest the amplitude, the greater the motor function and strength across the clinical measures.
The peak firing rate had moderate to low associations with all four measures, and the average firing rate had moderate associations with RULM and SMAFRS only.
By comparison, correlations between these motor function assessments and standard EMG measures were generally weaker. Also, most assessments from standard EMG and dEMG were weakly and non-significantly correlated.
[dEMG’s] correlation with measures of physical function, ability to distinguish between disease severities, and capturing of natural, real-world motor unit physiologic functioning suggest better overall utility than [standard] measurement of individual motor units for assessing disease status and progression
A clustering analysis then was used to group each dEMG parameter into a low or high cluster. The low cluster represented those with worse function (non-ambulatory), while the high cluster indicated better function (ambulatory). The proportion of patients whose ambulatory status agreed with the expected cluster placement was called the percent agreement.
Analyses showed that average amplitude of action potentials had an agreement between cluster placement and ambulatory status of 75%. However, the measure more accurately placed non-ambulatory patients on the correct cluster (88%) than ambulatory (55%) individuals. Likewise, for peak and average firing rate, the overall agreement occurred 86% of the time, with 91% for ambulatory and 82% for non-ambulatory participants.
Finally, due to the COVID-19 pandemic, repeated measures were available for only 14 study participants, with an interval of four to 22 months, or nearly two years. Regardless of the interval, the average amplitude and peak and average firing rates between measures indicated good overall test-retest reliability.
The researchers concluded that dEMG is a strong candidate for use as a biomarker in SMA.
“Its correlation with measures of physical function, ability to distinguish between disease severities, and capturing of natural, real-world motor unit physiologic functioning suggest better overall utility than [standard] measurement of individual motor units for assessing disease status and progression,” the team wrote.