This study is the first to use the AFQ analysis method on simultaneous PET/MRI imaging to reveal the main white matter microstructure changes and brain glucose metabolism distribution in PD patients and their interrelationships, thereby providing a basis for PD's microstructural and functional connectivity.
The results of this study indicate that compared to HC, PD patients have significantly higher MD and AD values in the right CST. Increased MD suggests disrupted integrity of nerve fiber bundles, while increased AD reflects axonal damage. Therefore, this study reveals that the integrity of the right CST is impaired in PD patients, which is consistent with previous findings(Pimer, et al.,2023;Yang, et al.,2022). However, some studies have suggested structural changes in the CST (i.e., increased FA) in PD patients(Atkinson-Clement, et al.,2017;Chen, et al.,2018), which is inconsistent with our results. Other studies(Lu, et al.,2016) found no differences in DTI metrics of the CST between PD patients and controls. Variations in results may be due to differences in the disease severity of PD subjects included in the studies. Early-stage PD might show no or compensatory changes in the CST, whereas later stages show impaired integrity. The study of Guimarães et al.(Guimaraes, et al.,2018) supports this view, showing increased AD and RD values in severe PD compared to mild PD and HC.
However, further studies are needed to determine when CST remains unchanged, when compensatory changes occur, or whether it is potentially affected by other pathophysiological mechanisms. The discrepancies might also be due to different analysis methods. Previous studies often used ROI-based analysis, VBM, TBSS, or other methods. Our study uses the fully automated DTI fiber tracking technique AFQ(Yeatman, et al.,2012), which eliminates errors from manually drawing ROIs and avoids averaging across entire fiber bundles, which may mask critical information. In this study, using the AFQ technique for node-by-node comparisons, it was found that in PD patients, MD and AD values of the subsegments (nodes 67–100) of the CST, located near the corona radiata and cerebral peduncle, were significantly increased compared to HC. This indicates white matter microstructural changes in this region, pinpointing the precise location of fiber bundle damage. This technique has been used in studies of brain white matter microstructural changes in diseases such as Alzheimer's, epilepsy, Wilson's disease, and systemic lupus erythematosus(Chen, et al.,2020;Keller, et al.,2017;Wu, et al.,2023;Zhang, et al.,2024). However, there have been few studies on PD.
Interestingly, we found that in PD patients, the H-Y stage and UPDRS-III scores were positively correlated with the MD and AD values of the damaged subsegments of the right CST, further supporting the notion that the integrity of the CST is related to disease severity. Additionally, we found that in PD patients, MMSE and MoCA scores were negatively correlated with the MD values of the same portion of the right CST, indicating that as the integrity of the right CST deteriorates, cognitive deficits in patients become more pronounced. Our results suggest that the CST, as the primary motor pathway transmitting information from the brain to the spinal cord, primarily governing voluntary movements of skeletal muscles, is also associated with cognitive decline in PD patients when its integrity is compromised, which is consistent with the conclusions of Sang (Sang, et al.,2022).
Using AFQ technology for node-by-node comparison, we identified localized microstructural changes in fiber tracts that were not detectable over the entire tract. Compared to HC, PD patients exhibited increased MD, AD, and RD values in segments of bilateral ThR, increased MD values in segments of the right CST, increased MD and RD values in segments of the right IFOF, and decreased FA value and increased RD value in segments of the left UF. ThR connects white matter fibers that connect the frontal cortex to the thalamus and basal ganglia, and its disruption may lead to memory impairment(Niida, et al.,2018). IFOF connects the occipital and frontal lobes, including fibers linking the frontal lobe to the posterior parietal and temporal lobes(Taoka T,2006). Our study showed that the most severe IFOF changes occurred in the anterior and posterior segments, potentially explaining cognitive and motor deficits in PD. UF connects the orbitofrontal cortex to the anterior temporal lobe and plays a crucial role in memory. However, no significant correlation was found between clinical scores and these fiber tracts, likely due to our small sample size and the lack of detailed clinical assessments in specific cognitive domains (such as five cognitive areas(Han, et al.,2021): attention and working memory, working memory, executive function, language, memory, and visuospatial function), which requires further research to confirm.
The simultaneous acquisition of 18F-FDG PET images allowed us to analyze SUVr in regions of interest based on the Automated Anatomical Labeling (AAL) atlas. Results showed increased glucose metabolism in the right paramedian lobule and bilateral putamen in PD compared to HC, while the right calcarine cortex exhibited decreased metabolism. This aligns with previous findings of brain metabolic changes in PD, where relative hypermetabolism is seen in the striatum (particularly the putamen), thalamus, pons, cerebellum, and motor cortex, and hypometabolism is observed in the premotor, supplementary motor, and posterior parietal areas(Ma, et al.,2007). The PD-related pattern (PDRP), established through scaled subprofile model (SSM)/principal component analysis (PCA) and 18F-FDG PET imaging, corroborates these metabolic changes(Eidelberg,2009;Woo, et al.,2017). This metabolic pattern likely reflects dysfunction in the cortical-striatal-pallidal-thalamic-cortical (CSPTC) motor circuit(DeLong MR,2007). This was confirmed in patients with PD undergoing subthalamic nucleus deep brain stimulation (DBS)(Lin, et al.,2008), where PDRP expression values obtained by preoperative 18F-FDG PET imaging significantly correlated with spontaneous discharge frequencies in the subthalamic nucleus, suggesting PDRP is linked to neuronal activity at critical nodes in the basal ganglia output pathway. Thus, PDRP explains the widespread functional abnormalities in CSPTC and related neural pathways in PD.
However, we have found in our clinical work that not all PD patients strictly follow the described metabolic pattern, likely due to variations in disease severity and clinical subtypes. Several studies have validated our hypotheses. For instance, the Parkinson's Disease Cognitive-Related Pattern (PDCP) is characterized by decreased metabolism in the medial frontal and parietal lobes and increased cerebellum, vermis, and dentate gyrus (Ko, et al.,2017). The Parkinson's Disease Tremor-Related Pattern (PDTP) shows varying degrees of increased glucose metabolism in the cerebellum and caudate/putamen(Mure, et al.,2011;Song, et al.,2023). However, more research is needed to confirm this. This study used a univariate model and could not investigate the relationship between glucose metabolism and clinical indices. Applying covariance analysis could quantify PDRP expression and study its relationship with clinical indices or predictive capabilities(Meles, et al.,2017).
This study observed significant damage in the right CST near the cerebral cortex, along with increased glucose metabolism in the adjacent right paramedian lobule (PML). Correlation analysis revealed that the mean MD and AD values of the damaged segments of the right CST were positively correlated with the FDG SUVr of the right PML. We hypothesize that the PML increases metabolic activity to compensate for the loss of structural connectivity in the CST, indicating that increased PML metabolism may be a compensatory mechanism(Kordys, et al.,2017). Additionally, due to the impairment of the CST and other motor pathways, the brain may reconfigure other neural networks to maintain essential motor control(Chu, et al.,2023;Lizarraga, et al.,2024), which could also lead to increased metabolic activity in the PML. By simultaneous PET/MRI imaging, we linked changes in white matter microstructure with glucose metabolism distribution and their interrelationships, revealing potential pathophysiological mechanisms of PD.
4.1 Limitations
Firstly, the sample size of this study is relatively small, and it is a single-center study, which may introduce bias. Secondly, we should have performed detailed subgroup analyses of PD patients, limiting the depth of analysis regarding disease severity and different clinical subtypes. Consequently, the pathophysiological study of different clinical manifestations and disease severity must be more comprehensive. Lastly, although this study is a multimodal neuroimaging study, it did not cover more imaging methods to identify the relationships between different brain pathological features. Future research will aim to recruit more subjects to increase sample size, incorporate multicenter studies, and utilize additional neuroimaging techniques for subgroup analysis of different clinical presentations (including motor and non-motor symptoms) and severities to gain a deeper understanding of PD's pathophysiology and provide comprehensive explanations for disease symptoms and surgical treatment principles, aiding in the development of more effective treatments.