The main pathological hallmark of CAA is the progressive deposition of Aβ40 and Aβ42 peptides within the wall of the cortical and leptomeningeal small vessels (arterioles, capillaries, and venules) [2]. Aβ40 fragment is the main isoform in vascular amyloid, and Aβ42 fragment is more frequently found in parenchymal plaques; however, Aβ42 is needed to initiate deposition in vessels both in animal models and in humans with CAA [14,15]. The Aβ40:42 ratio of cortical capillaries in tissue specimens is similar to that of parenchymal plaques and different from the one measured in arterial or venular CAA [14,16]. The Aβ42 fragment may then be the initial seed of vascular Aβ, and it deposits later into parenchymal plaques when the disease progresses. Aβ40 fragments are smaller and soluble, and the impairment of the perivascular clearance facilitates their deposition along arteries and veins. This different ratio of amyloid fragments refers to the histopathological definition of type I and type II CAA, respectively, with and without capillary involvement [14]. Vascular Aβ deposition affects the integrity and function of small vessels and, in particular, increases resistance and decreases blood flow, facilitating both microinfarctions and microhemorrhages [17]. This has been demonstrated in the arteriolar and capillary side of CAA and AD but is still neglected on the venular side. Moreover, a dysfunction in the neurovascular unit (NVU) and perivascular transport route has been proposed in CAA and AD, but the venular side of this dysfunction has rarely been addressed.

The recently revised concept of neurovasculome [18] represents an evolution of the NVU theory, including the entire extracranial (aortic arch to base of the skull) and intracranial vasculature and associated cells in the skull, brain, and meninges. Limiting the analysis to the venous side, the extracranial veins have a well-defined role in regulating cerebrospinal fluid (CSF) homeostasis and clearance [19]. Within this larger frame, the NVU is the specific neurovascular association occurring at each segment of the neurovasculome [18]. Therefore, the neurovasculome is constituted by multiple NVUs, and they may differ in cell composition depending on the specific vascular segment [20]. These differences were explored in murine models ranging from the pial arterioles to the pial venules through the penetrating arterial branches, capillaries, and ascending venules. The architecture and function of the NVU were investigated mainly in the neocortex, distinguishing the following sequential zones: pial artery/arteriole, penetrating arteriole, arteriole-capillary transition, capillary bed, post-capillary venules, ascending venule and pial vein/venule [18]. Capillaries form a dense tridimensional network throughout the parenchyma and account for >90% of the total length of the vessels in mice [21]. They join into a larger vessel, transition between capillaries and veins and drain into ascending venules, followed by pial venules. Cell types differ in different zones: the arteriolar side contains arterial endothelial cells and arteriolar smooth muscle cells (aSΜC); the venular side contains venous endothelial cells and venous smooth muscle cells (vSΜC). The same cells have different features in arteries and veins. The basement membrane on the vessel and glial sides limits the perivascular space, containing several other cell types, such as fibroblasts and macrophages. Larger arterioles have a more evident perivascular space than smaller arterioles due to the fusion of the glial and pial basement membrane layers as the caliber of the vessel decreases. Pericytes are represented in capillaries and post-capillary veins. Iuxtavascular microglia and oligodendrocyte precursor cells are present from the arteriolar-capillary transition zone to the capillary-venular transition zone.

NVU regulates cerebral blood flow (CBF) and waste clearance. Cerebrovascular autoregulation is an adaptive process that stabilizes CBF with changing arterial pressure, the main determinant of cerebral perfusion pressure [22]. The arteriolar side of the NVU works as the main site of autoregulation due to the myogenic response of the aSΜCs for guaranteeing the neurovascular coupling. The functions of the NVU are based on the BBB integrity [23], avoiding the paracellular passage of water-soluble molecules through specialized tight junctions among the endothelial cells and allowing molecule-specific exchange through specific influx and efflux transporters. In this way, the BBB assures a restricted permeability [24,25,26,27,28], leaving transcytosis to the post-capillary venules [29], which might represent a point-of-least resistance. Brain Aβ homeostasis is maintained through the endothelial regulation of influx and efflux transporters [5,19]. The waste clearance from the brain parenchyma follows several routes among the BBB, dural lymphatic vessels and along the perivascular spaces [19]. Perivascular clearance received increasing attention because of its proposed role in the pathophysiology of SVD and dementia, particularly amyloid-related brain diseases, such as CAA and AD. According to the glymphatic model, soluble molecules enter from the periarteriolar spaces, where CSF mixes with interstitial fluid (ISF), and drain along perivenular spaces in the same direction as blood flow [30,31].

On the contrary, the intramural periarterial drainage model suggests that solutes within the interstitial fluid drain along the basement membranes of the same arterioles in the opposite direction of blood flow [32,33]. An alternative model proposes that CSF and interstitial fluid mix at the pial surface and soluble waste products are cleared without unidirectional flow [34]. The driving forces moving solutes along the different routes are a crucial issue, and arterial pulsations [35,36,37] and vasomotion [38,39,40,41] were proposed for this role. The translation from animal models to humans is not immediate, and preclinical studies investigating perivascular fluid flow often rely on invasive techniques (including cranial window surgeries and injection of tracers) and the use of anesthesia. Both of them may affect cerebral homeostasis and hemodynamics [41].

Among the mechanisms involved in the clearance of small solutes, such as Aβ, the perivascular drainage pathway is the main route in CAA and AD. The perivascular drainage pathway is the movement of small solutes along the ISF and into the perivascular spaces or Virchow-Robin spaces [42]. The two proposed perivascular drainage pathways for Aβ are the intramural periarterial drainage and the glymphatic clearance [43,44]. The intramural periarterial drainage pathway involves the bulk flow of ISF into the basement membranes of capillaries and arteries [45]. In CAA, amyloid fragments usually deposit along SΜs of the tunica media and basement membranes of the tunica adventitia of artery walls, which is the site of the periarterial drainage. This suggests that poor clearance of this drainage pathway contributes to vascular Aβ deposition and CAA. The intramural periarterial drainage pathway does not involve venular function. In addition, ApoE ε4 has also been associated with reduced perivascular drainage of Aβ along the periarterial pathway [19]. The glymphatic clearance pathway involves the periarterial influx of CSF entering the brain, which combines with ISF and moves along the venous perivascular spaces of large cerebral veins [33,44]. Soluble Aβ is one of the solutes that drain along the glymphatic clearance pathway, and recent reports demonstrate this perivenous drainage to be impaired in AD [43,46].

Although both arteries and veins are involved in Aβ drainage, there has been controversy regarding the distribution of Aβ within different vessel walls. Pathological studies demonstrated that CAA leads to amyloid deposition along arteries and capillaries, with minimal involvement of veins in AD [47,48]. The main limitation of the human pathological data is the lack of consistent methods used to distinguish between arterioles and venules [49], partially due to brain tissue processing. Therefore, the venous-related pathophysiology should not be neglected in AD. A detailed knowledge of the angioarchitecture of the cerebral cortex may help to understand the role of arteries and veins in these diseases.