The objective of this study was to describe and quantify changes in both the microvascular media tunica and ECM components and to compare these changes between the AD and non-AD groups. We found a higher proportion of sclerotic arterioles and collagenic venules in patients with ADNC, whereas only the severity of collagenized venules was notable. ECM protein evaluations showed that the expression of Col IV and fibronectin was lower and that of agrin was higher in AD cases that may represent microvascular dysfunction, which was independent of age. Our findings suggest that not only is the media tunica damaged but also the BM-related ECM is affected with ADNC in the microvasculature of the brain parenchyma, and the venules are more diseased.
Microvascular media tunica pathology
Notably, in the current study, the severity of collagenized venules was more significantly correlated with ADNC, which is the most important finding concerning the potential role of venules in AD progression seldom known from previous histopathological studies. This is partly because small vessels have seldom been differentiated into arterioles or venules in previous studies. With accumulating knowledge of the glymphatic system, there is increasing speculation that structural alterations causing venules and veins draining might also contribute to the pathology of AD, while amyloid might be eliminated from the brain along the peri-venous spaces in the glymphatic system[28]. In a study of amyloid precursor protein and presenilin 1 rats, venous collagenosis was also observed in the leptomeningeal, cortical, hippocampal, and cerebellar areas[8]. Notably, the involved veins and venules stained positive for Aβ and tau, indicating the concurrent deposition of collagen and amyloid. However, direct evidence in humans is lacking. Magnetic resonance imaging (MRI)-visible deep medullary venules showed a higher tortuosity in patients with early AD in a 7T imaging study[29], and our previous MRI data also suggested a significant correlation between fewer deep medullary venules and brain atrophy[30], further supporting the potential role of venules in age-related neurodegenerative process. Histopathologically, studies on venular impairment in AD brains are not as plentiful as corresponding studies on arterioles, with only Keith et al. reporting the presence of venous collagenosis in AD individuals[31]. However, most previous studies suggest that periarteriolar spaces are involved in ADNC[32]. In our study, the lack of a significant association between white matter/cortex/hippocampus arteriolosclerosis and severe ADNC might be owing to the limited sample size and paucity of extensive arteriolosclerosis. However, we found that the venules were more diseased in the brain parenchyma with ADNC. Our results highlight the importance of studying the long-neglected brain venular system to understand the roles of arterioles and venules in AD progression.
Basement Membrane-related extracellular matrix protein changes
Endothelial cells are major components of the BBB, which are covered with BM and are surrounded by other ECM proteins in the neurovascular unit. Since some portions of Aβ are eliminated along the neurovascular unit and across the BBB, disrupting these pathways may result in exacerbated Aβ accumulation in the brain. Thus, BBB dysfunction has been demonstrated in AD in various studies and is emerging as a vital pathological trait[33]. Col IV, laminin, and fibronectin are the main and stable components of BM-related ECM, and their changes are well characterized in terms of vasculature lesions. Clinically, HSPG expression is increased in atherosclerotic and inflammatory processes in patients with AD compared to that in normal controls[34]. We hypothesize that not only microvascular media tunica damage but also microvascular functions affected by ECM component changes may cause and, consequently, might also contribute to AD pathogenesis.
Variable results for changes in Col IV, laminin, and fibronectin have been reported in previous studies. Some studies reported that cerebral microvessels of subclinical AD and AD patients had significantly higher levels of Col IV compared to those of controls[18, 35]. In contrast (and similar to our findings), other studies also reported that Col I and III increased, but Col IV decreased in microvessels of AD patients relative to microvessels of healthy controls[13, 15]. Although laminin is the most abundant non-collagenous protein in the BM of the central nervous system, it has scarcely been explored, and only Damodarasamy et al. reported lower laminin levels in tissues with high ADNC [13]. Recent immunohistopathological studies have also demonstrated that staining of cellular fibronectin (the adhesive protein that connects cells with collagen fibers in the ECM) is highly variable[18].
Perlecan, the largest ECM-associated HSPG, was found in the choroid plexus, subendothelial BM, and the BM surrounding the brain[36]. Recent pathological studies have suggested that perlecan and other HSPG are increased in AD, which correlates with increased Aβ staining, indicating a potential contribution to ADNC[18, 37]. However, earlier investigations found no significant difference in perlecan mRNA levels in hippocampal tissue from patients with AD when compared to that of age-matched controls[38]. Our data also supported that perlecan was relatively similar between AD cases and controls, with and at extremely low levels. Largely consistent with conclusions from other postmortem studies[39, 40], we reported higher levels of agrin in AD cases. Preclinical studies have suggested that agrin is important for maintaining BBB composition by altering aquaporin 4 levels, which could influence Aβ homeostasis in AD[41]. It is speculated that these inconsistent findings are related to different subjects, specific types of mouse models, and phases of AD pathologies. Therefore, a better understanding of the relationship between ECM proteins in the BM and AD at different stages is required, which will facilitate the active investigation of therapeutics across the BBB.
To the best of our knowledge, this is the first human pathological study to systematically investigate the relationship between different cerebral microvascular types and ADNC. However, this study has some limitations. The main limitation of our study is the small size of the cohort owing to the difficulty in acquiring brain tissues. Second, as in common pathological studies, the selected samples in our observational study cannot represent the whole-brain status, and we cannot rule out a casual observation. Third, limited demographic and clinical (e.g., vascular risk factors) data were available, and our findings cannot be adjusted for other underlying factors, except for age and sex.