Reducing motor evoked potential amplitude variability through normalization.

Abstract

BackgroundTranscranial Magnetic Stimulation (TMS) is used for in vivo assessment of human motor cortical excitability, with application of TMS pulses over the motor cortex resulting in muscle responses that can be recorded with electromyography (EMG) as Motor Evoked Potentials (MEPs). These have been widely explored as potential biomarkers for neuropsychiatric disorders but methodological heterogeneity in acquisition, and inherent high variability, have led to constraints in reproducibility. Normalization, consisting in scaling the signal of interest to a known and repeatable measurement, reduces variability and is standard practice for between-subject comparisons of EMG. The effect of normalization on variability of MEP amplitude has not yet been explored and was assessed here using several methods.MethodsThree maximal voluntary isometric contractions (MVICs) and 40 MEPs were collected from the right hand in healthy volunteers, with a retest session conducted 4 to 8 weeks later. MEP amplitude was normalized using either external references (MVICs) or internal references (extreme MEPs). Iterative re-sampling of 30 normalized MEPs per subject was repeated 5,000 times to define, for each normalization method, distributions for between-subject coefficients of variation (CV) of the mean MEP amplitude. Intra-class correlation coefficients (ICC) were used to assess the impact of normalization on test–retest stability of MEP amplitude measurements.ResultsIn the absence of normalization, MEPs collected from the right hand of 47 healthy volunteers were within reported values regarding between-subject variability (95% confidence intervals for the CV: [1.0567,1.0577]) and showed good temporal stability (ICC = 0.77). Internal reference normalization substantially reduced between-subject variability, by values of up to 64%, while external reference normalization had no impact or increased between-subject variability. Normalization with the smallest references reduced test–retest stability, with use of the largest references resulting in slight reduction or improvement of ICCs. Internal reference normalization using the largest MEPs was found to be robust to several sensitivity analyses.ConclusionInternal, but not external, reference normalization reduces between-subject variability of MEP amplitude, and has a minimal impact on within-subject variability when conducted with the largest references. Additional research is necessary to further validate these normalization methods toward potential use of MEPs as biomarkers of neuropsychiatric disorders.