Examining iTBS-induced effects within a computational model of resting-state EEG activity integrated with calcium-dependent plasticity.

Schematic of the methodology and neurophysiological model processes. A) Corticothalamic column underlying resting-state EEG activity and a B) 4-population corticothalamic neural population model of its circuitry, comprised of cortical excitatory pyramidal (e) and inhibitory interneuron (i) populations, and thalamic excitatory relay (s) and inhibitory reticular (r) nuclei populations. C) Intermittent theta-burst stimulation (iTBS) protocols deliver voltage-spiking stimuli to excitatory cortical populations. Cyan and purple timeseries depict pre- and post-stimulation resting-state activity, respectively, altered by plasticity-induced circuit modifications. D) Flowchart of Ca2+ scheme wherein post-synaptic depolarization and glutaminergic binding induce metaplastic NMDARs to open, enabling the influx of calcium to drive plasticity effects, with moderate and large concentrations driving LTD and LTP, respectively. Connection weight changes are immediately translated through signalled plasticity (a physiological analog of immediate, though functionally unrealized cell states) before manifesting as fully expressed functional modifications via altered ν values, following a signal transduction delay. In conjunction, NMDAR conductance is altered based on the prior weight dynamics, thereby mediating the next instance of calcium influx subsequent plasticity effect. The signaled (; pink) and expressed (ν; purple) synaptic weight over pre, active, and post-canonical iTBS are shown for the excitatory-excitatory (ee) connection. Yellow and blue windows depict when modulated connection weight is relatively stronger and weaker than initial efficacy, respectively.

Induced broadband power modulation is enhanced by IAF-iTBS alignment.

A) Heatmap of iTBS-induced broadband power suppression (pre - post AUC Δ) across parameter space (brighter = stronger suppression), with B) samples of power spectra prior to (cyan) and following (purple) stimulation, showing greater degrees of power modulation near alpha harmonic frequencies; green marks the canonical iTBS protocol. C) Further simulations with slowing/accelerating the alpha-band frequency show that IAF-iTBS resonance leads to proportional shifts in the inter-burst frequency of the most effective protocols. D) Broadband power change as a function of ARD (top-left), pulses-per-burst (bottom-left), inter-burst frequency (top-right), and pulse rate (bottom-right). Together these results support the premise that natural circuit frequency predicts the degree of iTBS-induced broadband power modulation.

Selective synaptic modifications underlie broadband power modulation and demarcate protocol efficacy from circuit-wide plasticity effect.

A) Heatmap of circuit-mean connection weight changes revealed a confined region of enhanced modification across parameter space distinct from broadband power suppression. B) Connection modification patterns for a sample of protocols illustrate unique synaptic plasticity effects from generalized circuit weight changes. C) Circuit-mean modifications adhere to a Gaussian distribution over protocol pulse rate, highlighting an optimal range within which plasticity is induced. Moreover, D) group-level weight modification and broadband power modulation are not statistically associated, implying a necessity to induce selective plasticity effects to compel functionally efficacious changes. E) Selective modifications are driven further by resonant effects, which visibly enhance inhibitory-afferent LTP across circuit connections.

Induced calcium concentrations corroborate a two-pronged effect on spectra via inhibitory LTP and excitatory LTD.

Heatmaps of circuit-mean A) LTP and B) LTD calcium volumes, as well as C) LTP and D) LTD volumes for each corticothalamic connection. Parameter changes induced unique patterns of calcium release, with protocol efficacy being defined by an amplification of LTP calcium globally and LTD calcium within cortical-excitatory afferent connections. These unique calcium effects are demonstrated across subset of different protocols, including standard (lime green) and model-optimal (dark green) iTBS. E) Circuit-mean LTP calcium volume decrease was strongly predicted by pulse rate increase and strongly predictive of circuit-mean connection weight change, while F) LTD volume decrease was only moderately predicted by pulse rate increase and not predictive of weight change.

iTBS efficacy is robustly captured by corticothalamic dampening and intrathalamic enhancements of circuit loop gains.

A) Corticothalamic connection gains may be mathematically collapsed into the circuit’s three major loop gains: X (Cortical), Y (Corticothalamic), and Z (Intrathalamic). B) Post-stimulation circuit distributions in XYZ loop gain space show a robust trajectory in relation to broadband power change, with canonical (green) and a subset of novel iTBS protocols induce different displacements in space, corresponding to unique neurophysiological modifications. Point colors reflect broadband power modulation (as seen in Fig. 2A). C) Broadband power changes are shown to be a robust product of decreases and increases to Y and Z loop gains, respectively, implying downstream modifications bear the most functional importance on broadband power changes. These trends are further amplified by D) alpha resonance distance, which also manifests downstream to more strongly displace Y and Z circuit loop gains.

MLR coefficients between ARD and broadband power modulation over unique corticothalamic alpha frequencies.

Bolded values represent statistical significance following multiple comparison corrections.

MLR coefficients for iTBS-induced modifications to corticothalamic connection weights (Δν), predictive of broadband power modulation and predicted by ARD.

Bolded values represent statistical significance following multiple comparison corrections.

MLR coefficients for induced LTD & LTP calcium volumes, predictive of broadband power modulation and predicted by ARD.

Bolded values represent statistical significance following multiple comparison corrections.

MLR coefficients of corticothalamic loop gains change, predictive of broadband power modulation and predicted by ARD.

Bolded values represent statistical significance following multiple comparison corrections.

Corticothalamic neural population with metaplasticity model parameters.

Heatmap of post-iTBS connection weights for each connection within the corticothalamic circuit.

Purple denotes weight enhancements, while blue denotes reductions.

Scatterplot distributions of post-stimulation X, Y, and Z loop gains across pulses-per-burst, inter-burst frequency, and pulse rate iTBS parameters.

Relationship between iTBS-induced changes of alpha and broadband power, connection weights, and loop gains.

Illustration of circuit-mean synaptic weight change across pulse rate.

Circuit-mean synaptic weight change over pulse rates, featuring canonical iTBS (lime), model-optimal iTBS (dark green), and HF-rTMS (black).

Method for standardizing protocol intensity.

Scaling is based on the proportional stimulation intensity of canonical iTBS.

Primary analyses conducted with protocols using a fixed pulse amplitude.

rTMS-induced effects on broadband power modulation, Circuit-mean and individual synaptic weight change, LTD and LTP calcium levels, and X, Y, and Z loop gains.

Comparison of active-stimulation NMDAR conductance variance between scaled- (purple) and fixed- (green) amplitude iTBS protocols.

A) Relationship between protocol pulse rate and standard deviation of NMDAR conductance. Relationship between NMDAR engagement and B) broadband power modulation, C) circuit-mean synaptic weight change, and D) plasticity-inducing calcium volumes.