Hemodynamic Correlates of Fluctuations in Neuronal Excitability: A Simultaneous Paired Associative Stimulation (PAS) and functional Near Infra-Red Spectroscopy (fNIRS) Study

Abstract

Background: The relationship between task-related hemodynamic activity and brain excitability is poorly understood in humans as it is technically challenging to combine simultaneously non-invasive brain stimulation and neuroimaging modalities. Cortical excitability corresponds to the readiness to become active and as such it may be linked to metabolic demand. Hypotheses: Cortical excitability and hemodynamic activity are positively linked so that increases in hemodynamic activity correspond to increases in excitability and vice-versa. Methods: Fluctuations of excitability and hemodynamic activity were investigated via simultaneous Transcranial Magnetic Stimulation (TMS) and functional Near Infrared Spectroscopy (fNIRS). Sixteen healthy subjects participated in a sham-controlled, pseudorandomized, counterbalanced study with PAS (PAS10/PAS25/Sham) on the right primary motor cortex (M1). The relationship between M1 excitability (Motor Evoked Potentials, MEP) and hemodynamic responses to finger tapping reconstructed via personalized fNIRS was assessed. Results: Hemodynamic activity exhibited a significant correlation with cortical excitability: increased HbO and HbR (absolute amplitude) corresponded to increased excitability and vice-versa (r=0.25; p=0.03 and r=0.16; p=0.17, respectively). The effect of PAS on excitability and hemodynamic activity showed a trend of positive correlation: correlation of MEP ratios (post-PAS/pre-PAS) with HbO and HbR ratios (r=0.19, p=0.29; r=0.18, p=0.30, respectively). Conclusions: TMS-fNIRS is a suitable technique for simultaneous investigation of excitability and hemodynamic responses and indicates a relationship between these two cortical properties. PAS effect is not limited to cortical excitability but also impacts hemodynamic processes. These findings have an impact on the application of neuromodulatory interventions in patients with neuropsychiatric disorders.

Hierarchical Bayesian Modeling of the Relationship between Task Related Hemodynamic Responses and Cortical Excitability

“…Raw fNIRS data were first preprocessed following standard recommendations (Yücel et al, 2020): a) bad channel rejections of channels exhibiting either a negative raw amplitude during the whole time course and a coefficient of variation (CV) larger than 8% (Schmitz et al, 2005;Schneider et al, 2011;Eggebrecht et al, 2012;Piper et al, 2014): b) linear regression of superficial physiological fluctuations using the average of all proximity channels (Zeff et al, 2007); c) band-pass filtering (i.e., 0.01Hz to 0.1Hz) using a 3rd order Butterworth filter (zero-phase); d) conversion in optical density changes (i.e., ∆OD) using logarithm conversion; e) ∆OD epochs extraction within a time window ranging from -10s to 30s around task onsets. Instead of the conventional process averaging extracted ∆OD epochs, we then conducted a resampling process to estimate not one but a set of 'possible' averaged ∆ODs (Cai et al, 2021b). Our rationale was to propose an evaluation preserving the intrinsic variance of averaged ∆OD related to the underlying physiological fluctuations and eventual measurement errors such as motion artifacts.…”

“…The output of the whole fNIRS data preprocessing section was a set of 80 runs (i.e., 40 sessions (16 PAS25+12 PAS10+12 sham) × 2 times) of 101 reconstructed HbO/HbR time course, for each run specified by subject (ID 1 to 16), intervention (PAS25, PAS10 or sham) and time (pre-PAS or post-PAS). Please see further details of fNIRS data processing in Appendix 1 and our previous study (Cai et al, 2021b).…”

“…Therefore, fNIRS appears as an interesting alternative offering better compatibility for simultaneous acquisition during TMS (Curtin et al, 2019). In Cai et al, 2021b, we conducted a simultaneous PAS-fNIRS study to investigate the relationship between motor task-evoked cortical hemodynamic response and M1 excitability. We opted for PAS rather than rTMS for multiple reasons.…”

“…Considering the above advantages of the Bayesian approach, in this study, we applied a Bayesian data analysis workflow (Gabry et al, 2019;Gelman et al, 2020b) on our TMS/fNIRS dataset rather than the conventional analysis conducted in Cai et al, 2021b, to further investigate the relationship between the PAS effects task-related hemodynamic responses and its effects on M1 excitability. We hypothesize that enhanced brain excitability should be associated with higher hemodynamic activity elicited by a finger tapping task, and decreased excitability should be associated with a reduced hemodynamic response to the task.…”

“…Therefore, fNIRS offers better compatibility for simultaneous acquisition during TMS [Curtin et al, 2019]. In Cai et al, 2022b, we demonstrated that simultaneous PAS-fNIRS is feasible and investigated changes in excitability and hemodynamic activity due to non-invasive neuromodulation. Among the many non-invasive brain stimulation techniques, we focused our attention on PAS, due to methodological and biological reasons.…”