Luiz R T da Silva, Maria S G Rocha, Arnaldo F Neto, André K Takahata, Slawomir Nasuto, Fabio Godinho, Bradley Voytek, Diogo C Soriano
Decomposing STN-LFPs into periodic and aperiodic components reveals distinct phenotype- and movement-specific neural dynamics in patients with PD. The differential modulation of the aperiodic exponent and knee frequency suggests that TD and PIGD may engage distinct circuit-level responses during movement, potentially involving differences in excitation/inhibition-related dynamics and intrinsic neural timescales.
OBJECTIVE: Traditional analyses of subthalamic nucleus local field potentials (STN-LFPs) may obscure relevant dynamics by conflating periodic and aperiodic activity. We investigated whether spectral parameterization reveals phenotype- and movement-dependent neural signatures in Parkinson's disease (PD).
METHODS: Intraoperative STN-LFPs (35 STN: 20 PIGD, 15 TD) were recorded across rest and movement conditions during deep brain stimulation surgery in twenty-two PD patients (10 tremor-dominant [TD], 12 postural instability and gait disorder [PIGD]). Power spectral densities were parameterized into aperiodic-adjusted and non-adjusted bandpowers (alpha, low-beta, and high-beta), and aperiodic features (offset, knee frequency, and exponent). Linear Mixed-Effects models assessed interactions between phenotype and condition. Multivariate logistic regression classified phenotypes, and Spearman's rank correlations evaluated pairwise relationships between each UPDRS subscores (rigidity, tremor, and bradykinesia) and each spectral feature.
RESULTS: TD patients exhibited significant suppression in aperiodic-adjusted low-beta power during movement. Aperiodic features revealed significant phenotype-dependent modulation, marked by significant knee frequency enhancement in PIGD during movement (p = 0.002) and significant phenotype differences in movement-related exponent modulation (p < 0.01). Combining periodic and aperiodic movement-related features in logistic regression yielded the highest classification performance. Furthermore, aperiodic spectral features did not significantly correlate with the severity of PD symptoms.
CONCLUSIONS: Decomposing STN-LFPs into periodic and aperiodic components reveals distinct phenotype- and movement-specific neural dynamics in patients with PD. The differential modulation of the aperiodic exponent and knee frequency suggests that TD and PIGD may engage distinct circuit-level responses during movement, potentially involving differences in excitation/inhibition-related dynamics and intrinsic neural timescales.