Center for Engineering, Modeling and Applied Social Sciences, Federal University of ABC, São Bernardo do Campo, Brazil; Brazilian Institute of Neuroscience and Neurotechnology, Campinas, São Paulo, Brazil; Department of Cosmic Rays and Chronology, Institute of Physics, University of Campinas, Campinas, Brazil.
Center for Engineering, Modeling and Applied Social Sciences, Federal University of ABC, São Bernardo do Campo, Brazil; Brazilian Institute of Neuroscience and Neurotechnology, Campinas, São Paulo, Brazil.
Clin Neurophysiol. 2022 Aug;140:45-58. doi: 10.1016/j.clinph.2022.05.013. Epub 2022 Jun 7.
Parkinson's disease (PD) patients may be categorized into tremor-dominant (TD) and postural-instability and gait disorder (PIGD) motor phenotypes, but the dynamical aspects of subthalamic nucleus local field potentials (STN-LFP) and the neural correlates of this phenotypical classification remain unclear.
35 STN-LFP (20 PIGD and 15 TD) were investigated through continuous wavelet transform and machine-learning-based methods. The beta oscillation - the main band associated with motor impairment in PD - dynamics was characterized through beta burst parameters across phenotypes and burst intervals under specific proposed criteria for optimal burst threshold definition.
Low-frequency (13-22 Hz) beta burst probability was the best predictor for PD phenotypes (75% accuracy). PIGD patients presented higher average burst duration (p = 0.018), while TD patients exhibited higher burst probability (p = 0.014). Categorization into shorter and longer than 400 ms bursts led to significant interaction between burst length categories and the phenotypes (p < 0.050) as revealed by mixed-effects models. Long burst durations and short bursts probability positively correlated, respectively, with rigidity-bradykinesia (p = 0.029) and tremor (p = 0.038) scores.
Subthalamic low-frequency beta bursts differed between TD and PIGD phenotypes and correlated with motor symptoms.
These findings improve the PD phenotypes' electrophysiological characterization and may define new criteria for adaptive deep brain stimulation.
帕金森病（PD）患者可分为震颤为主（TD）和姿势不稳伴步态障碍（PIGD）运动表型，但丘脑底核局部场电位（STN-LFP）的动力学特征以及这种表型分类的神经相关性尚不清楚。
通过连续小波变换和基于机器学习的方法研究了 35 个 STN-LFP（20 个 PIGD 和 15 个 TD）。通过跨表型的β爆发参数和在特定提出的最佳爆发阈值定义标准下的爆发间隔，对与运动障碍相关的主要频段β振荡的动力学特征进行了特征描述。
低频（13-22 Hz）β爆发概率是 PD 表型的最佳预测因子（准确率为 75%）。PIGD 患者的平均爆发持续时间较长（p=0.018），而 TD 患者的爆发概率较高（p=0.014）。根据混合效应模型，将爆发时长分为较短和较长两组，两组间的爆发时长和表型之间存在显著的交互作用（p<0.050）。长爆发持续时间和短爆发概率分别与僵硬-运动迟缓（p=0.029）和震颤（p=0.038）评分呈正相关。
TD 和 PIGD 表型之间的 STN 低频β爆发存在差异，并与运动症状相关。
这些发现提高了 PD 表型的电生理特征描述，并可能为适应性脑深部刺激定义新的标准。
