Consistent with our vascular hypothesis, the patients with PD experienced higher serum vascular adhesion molecule levels, as well as high expression levels of angiogenic miRNAs, suggesting the occurrence of vascular inflammation and related neuroinflammation. Furthermore, we identified gray matter atrophy throughout much of the cortex, including in anatomical locations that typically appear to be particularly sensitive to the effects of inflammation, such as the cerebellum, bilateral posterior cingulate gyrus, left parahippocampus, and left temporal gyrus [28]. We further demonstrated, for the first time, that gray matter volume atrophy in the parahippocampus correlated with VCAM-1 levels and disease severity scores. As for disease progression, VCAM-1 levels were also found to be positively correlated with disease severity. The correlations of the vascular inflammatory parameters with brain gray matter atrophy in the dominant side of the parahippocampus and temporal lobe may mirror the relationship between vascular inflammatory factors and neuroinflammation-related brain atrophy during the neurodegenerative process in PD.
Volumetric MRI measures of whole-brain and regional atrophy provide indirect assessments of neuronal loss, and correlations of such measures with clinical disease progression have been proposed as useful radiological tool of neurodegeneration in PD [29–31]. Data from our previous MRI studies suggest a spatial pattern of regional volume loss in PD that closely parallels the respiratory dysfunction, perceptual impairment, and systemic oxidative stress seen in PD [10, 32]. The PD signature of regional atrophy observed in this study was characterized by predominant involvement of the left parahippocampus as well as the bilateral posterior cingulate gyrus, fusiform, left temporal gyrus, and cerebellum, with gray matter volume changes in these regions also being correlated with disease severity levels (as measured by UPDRS II, III, and 176 scores). These atrophic areas include key components of mental processing (the superior temporal gyrus)[33], memory formation (the hippocampus), high level visual processing (the fusiform cortex) [34], and social cognition and affective processing (the cerebellum)[35]. A previous study revealed that poor performance on visuoperceptual tests was significantly associated with gray matter decreases in the fusiform, the parahippocampus, and the middle occipital gyrus[36], while another study reported that decreased cortical thickness was also related to cognitive deterioration (including in the storage of prior experiences, integration of external perceptions, and semantic processing)[37]. Gray matter atrophy has been widely studied in previous PD research studies, but what factors cause brain cortex atrophy? Recent research revealed that the progression of cognitive impairment in PD patients occurs through two steps: grey matter hypo-metabolism and atrophy[38]. The hypo-metabolism may due to lower cerebral profusion and related to vascular factors. Lower perfusion in PD and PD dementia patients has been demonstrated by using arterial spin labeling (ASL) MRI[22]. In the brain parenchyma, neurovascular changes may interact with the neurodegenerative process in idiopathic PD, with markers of neurovascular status including white matter changes and cerebral blood flow [7]. Gray matter volume atrophy is closely associated with underlying white matter changes and may be correlated with vascular factors like neoangiogenesis and BBB dysfunction [39, 40] .
4.2. VCAM-1 and miRNA profiles as vascular inflammation indicators in PD patients
According to Fig. 2, disease progression and brain gray matter volume atrophy are the combined end results of peripheral inflammation and BBB dysfunction and their subsequent sequelae. Peripheral inflammation and the aging process can cause BBB dysfunction and cause inflammatory factors to infiltrate into neurons and adjacent supporting tissue. Inflammation also causes vascular damage that changes microvessel integrity, resulting in microvascular hypoperfusion and inducing neoangiogenesis, which in turn result in macro-brain structural changes like WM hypertrophy and, finally, neurodegeneration and clinical disease progression [6, 41].
In this study, the significantly higher levels of -1-1, miRNA-29a and miRNA-22 in the PD patients were important clues suggesting that underlying vascular factors contribute to neuroinflammation. In brain tissues, dysfunction of the BBB accompanied by microglia activation and neutrophil infiltration leads to the loss of dopaminergic neurons caused by programmed cell death [13][42, 43]. Under conditions of inflammation, the microglial release of proinflammatory cytokines, including tumor necrosis factor-alpha (TNF-α), interleukin (IL)-1β, and transforming growth factor-β, acts on the endothelium of BBB cells to stimulate the upregulation of VCAM-1 and ICAM‐1, causing the transmigration of leukocytes and related subtypes from the blood vessels into the CNS, promoting neuroinflammation[44, 45]. Cell-bound VCAM-1 allows human brain microvascular endothelium to control immune cell trafficking across the BBB. It is upregulated in the inflammatory-active brain lesions of patients with multiple sclerosis (MS)[46], neurodegenerative diseases like AD[18] and atherosclerosis[47]. Not only is it correlated with the proinflammatory cytokines involved in BBB dysfunction, but according to one in vitro study, soluble VCAM-1 directly impairs the integrity of human brain endothelial barriers, causing increased permeability that compromises the barrier function by inducing intracellular signaling by integrin alpha-4[15]. Meanwhile, in normal elderly healthy individuals, elevated plasma VCAM-1 is associated with impaired cerebrovascular function and mobility impairments [48]. An association between VCAM-1 and CO2-related cerebral vasoreactivity, as well as gray matter atrophy, has also been reported in diabetic subjects [49], while elevated VCAM-1 is also associated with rather large effects on white matter hyperintensities[50]. In this study, the levels of soluble VCAM-1 were positively correlated with the clinical PD disease severity scores. Therefore, VCAM-1 may serve an important role in BBB dysfunction and as a clinical biomarker of PD and its clinical consequences.
BBB dysfunction may also be associated with angiogenesis[51]. In the brain, immature vessels likely lack the full characteristics of the BBB, including the development of tight junctions, the recruitment of pericytes, and the formation of a glial limitans. Thus, angiogenesis may compromise the function of the BBB, which could contribute in turn to ongoing neuroinflammation by allowing peripheral molecules and immune cells to access the brain parenchyma. Indeed, increased CSF levels of angiogenesis markers have been seen in PD patients and associated with BBB permeability, white matter lesions, and cerebral microbleeding [52]. In this study, the angiogenesis-related miRNA-29a and miRNA-22 were up-regulated in the PD patients, suggesting an underlying and ongoing angiogenesis process. MiRNA-22 regulates inflammation and angiogenesis by targeting VE-cadherin[53] and miRNA-29a can modulate the angiogenic properties of human endothelial cells [19] and be upregulated by proinflammatory cytokine transforming growth factor β (TGF-β) [54].The miRNA is also involved in neuron growth[55] and atherosclerosis formation[56].An in vitro study found that miRNA-29a overexpression promoted the formation of new blood vessels, while miR-29a suppression completely blocked TGF-β1-stimulated angiogenesis [54]. TGF-β1 expression was increased in striatal neurons and in activated microglia on the lesion side of a 6-hydroxydopamine-induced PD mouse model in one in vivo study [57] and was elevated in the CSF fluid of PD patients in an in vitro study, indicating that it acts as a neuroprotective factor in injured brains. Another previous study revealed the down-regulation of miR29a in treatment-naïve PD patients [58], but up-regulation in L-dopa-treated PD patients has also been observed[20]. In another neurodegenerative disease, AD, miRNA-29a was found to be up-regulated in blood-CSF samples[59] but decreased in brain tissue [60]In this study, the up-regulation of miR-29a may have been due to the combination of hypoxia-related neoangiogenesis mechanisms and L-dopa treatment-related angiogenesis[61]. Our previous study revealed that the hypoxia-like status in obstructive sleep apnea may cause alterations to regional cerebral blood flow [62]. Hypoxia–ischemia can induce profound non-preferential and immediate mobilization of leucocytes into the circulation, along with a significant influx of neutrophils into the brain with subsequent neuroinflammation [63]. Another factor of hypoxia in the microstructures of the brain is the change in neurovascular unit (NVU) integrity caused by L-dopa treatment that in turn causes regional cerebral metabolism and blood flow changes, which are also related to the angiogenesis and increased BBB permeability induced by VEGF regulation, and is reflected in the clinical result of l-dopa-induced dyskinesia. Our previous study using MRI arterial spin labeling also showed vascular changes in PD patients receiving acute L-dopa treatment. Hypoperfusion in the occipital, parietal, and cerebellar regions, as well as extensive neocortical hypoperfusion, in PD dementia patients has also been noted. In this study, the upregulated expression of miRNA-29a was positively correlated with cerebellar atrophy and whole brain gray matter atrophy. As for disease progression under L-dopa treatment, evidence suggests that miRNA-29a expression level changes play a role in the microvascular remodeling process, brain morphology, and vascular flow changes.
Partial correlations among disease severity, specific brain atrophy, and vascular markers
The PD signature of regional atrophy in this study was characterized by the predominant involvement of the left parahippocampus, bilateral posterior cingulate gyrus, fusiform, left temporal gyrus, and cerebellum, in addition to being correlated with disease severity, findings which were also supported by our previous studies [64, 65]. After controlling for age and sex, volume changes in the left parahippocampus were weakly to moderately correlate with VCAM-1 expression levels. The parahippocampus is a key component of the human ventral temporal cortex, which has the function of visual categorization and recognition[66], is part of the salience network that functions in memory retrieval [67], and is part of the default-mode network involved in memory encoding [68]. Our previous study demonstrated that mesial temporal network degeneration interacts with systemic oxidative stress and cognitive impairment in PD patients [65]. In parahippocampal gyrus brain tissue, the accumulation of alpha-synclein is directly related to the level of beta-amyloid and the Braak tangle stage and can predict cognitive status in PD dementia patients[69]. Such accumulation of misfolded protein may trigger the expression of endothelial and insulin receptor signaling pathway genes for angiogenesis in neurodegenerative diseases like AD [70] and be related to changes in the expression of hypoxia and angiogenesis-related genes[71]. Vascular degeneration in PD appears to be the result of endothelial cell degeneration, even as the capillary basement membrane is retained. Increased string vessel formation suggests a role for vascular hypoperfusion in the progression of PD, while leakage of the BBB may be associated with astrocytosis due to inflammatory stimulation, which appears to be a pathological change secondary to the initial lesion in PD[72]. As shown in Table 3 of this study, VCAM-1 expression was moderately correlated with all the disease severity scores and weakly to moderately correlated with whole brain atrophy and left parahippocampal gyrus atrophy, supporting the role of VCAM-1 as a vascular marker in PD and providing further evidence of the interactions among neuroinflammation, neurovascular unit integrity, BBB dysfunction, and subsequent PD disease progression. For this reason, the combination of vascular markers with brain gray matter atrophy can integrate more components of PD pathogenesis.
Limitations
The interpretation of the findings presented here must be tempered by some of the limitations of the present study. First, the patients who participated were recruited from a single tertiary center and so may not be representative of all PD populations. Second, this study was a cross-sectional and small-sized study, such that readers should interpret the results with caution. Further causal relationships among grey matter volume changes, clinical disease severity, and vascular inflammatory markers may need to be delineated by future longitudinal studies. In addition, the measurement of only a few vascular inflammatory biomarkers cannot be considered a valid tool for exploring the multifaceted, complex BBB dysfunction and angiogenesis imbalance in PD. The micro-environment interactions in the BBB region and brain parenchyma due to CNS and systemic oxidative stress can be affected by individual genetic variations and physical exercise, and these interactions were not well evaluated in the present study. In addition, recent research suggests that network-based rather than regional anatomic involvements contribute to the neurodegeneration and disease severity seen in PD, such that assessing structural or functional connectivity between brain regions associated with neurovascular changes would clarify the complex neural network involved in vascular factor and BBB dysfunction.