Progressive Microstructural Alterations in Subcortical Nuclei in Parkinson's Disease: A Diffusion Magnetic Resonance Imaging Study.

Xueqin Bai,Cheng Zhou,Tao Guo,Xiaojun Guan,Jingjing Wu,Xiaocao Liu,Ting Gao,Luyan Gu,Min Xuan,Quanquan Gu,Peiyu Huang,Zhe Song,Yaping Yan,Jiali Pu,Baorong Zhang,Xiaojun Xu,Minming Zhang
DOI: https://doi.org/10.1016/j.parkreldis.2021.06.003
IF: 4.402
2021-01-01
Parkinsonism & Related Disorders
Abstract:Objectives: To explore the microstructural alterations in subcortical nuclei in Parkinson's disease (PD) at different stages with diffusion kurtosis imaging (DKI) and tensor imaging and to test the performance of diffusion metrics in identifying PD. Methods: 108 PD patients (64 patients in early-stage PD group (EPD) and 44 patients in moderate-late-stage PD group (MLPD)) and 64 healthy controls (HC) were included. Tensor and kurtosis metrics in the subcortical nuclei were compared. Partial correlation was used to correlate the diffusion metrics and Unified Parkinson's Disease Rating Scale part-III (UPDRS-III) score. Logistic regression and receiver operating characteristic analysis were applied to test the diagnostic performance of the diffusion metrics. Results: Compared with HC, both EPD and MLPD patients showed higher fractional anisotropy and axial diffusivity, lower mean kurtosis (MK) and axial kurtosis in substantia nigra, lower MK and radial kurtosis (RK) in globus pallidus (GP) and thalamus (all p < 0.05). Compared with EPD, MLPD patients showed lower MK and RK in GP and thalamus (all p < 0.05). MK and RK in GP and thalamus were negatively correlated with UPDRS-III score (all p < 0.01). The logistic regression model combining kurtosis and tensor metrics showed the best performance in diagnosing PD, EPD, and MLPD (areas under curve were 0.817, 0.769, and 0.914, respectively). Conclusions: PD has progressive microstructural alterations in the subcortical nuclei. DKI is sensitive to detect microstructural alterations in GP and thalamus during PD progression. Combining kurtosis and tensor metrics can achieve a good performance in diagnosing PD.
What problem does this paper attempt to address?