The severity of CAA correlates with the extent of pathological changes in cerebral blood vessels. In healthy blood vessels, the Aβ peptide, along with soluble metabolites and interstitial fluid, are drained between the smooth muscle cells in the tunica media and are eliminated through a process called perivascular clearance; while in a CAA-affected brain, this mechanism is impaired, leading to Aβ aggregation, gradual destruction of smooth muscle cells, and destruction of the vessel wall, resulting in brain hemorrhage (Weller et al., 2008; Greenberg et al., 2020). In mild CAA, Aβ is restricted to the tunica media, without significant destruction of smooth muscle cells. In moderate CAA, Aβ begins to replace the tunica media, and the vessel wall becomes abnormally thickened. Severe CAA is characterized by extensive Aβ deposition, fibrinoid necrosis, microaneurysm formation, and blood leakage through compromised vessel walls (Table 1; Vonsattel et al., 1991; Biffi and Greenberg, 2011). After hypertension, CAA is the second most common cause of spontaneous intracerebral hemorrhage (ICH) in people over the age of 60 (Szidonya and Nickerson, 2023) and, when it occurs, episodes are recurrent and typically more severe than the previous one (Passero et al., 1995). Secondary to cerebral hemorrhage and infarcts, CAA can result in cognitive impairments, such as deficits in perceptual speed and episodic memory, and dementia (Chwalisz, 2021; Kozberg et al., 2021).

Cerebral amyloid angiopathy is classified as either type I or II, based on the predominant vascular compartment affected by Aβ (Thal et al., 2002). In CAA type I, Aβ accumulates mainly in cortical capillaries, as well as leptomeningeal and cortical arteries, arterioles, veins, and venules (Thal et al., 2002; Rajpoot et al., 2022). In contrast, CAA type II, the more common type in sporadic cases (Walker et al., 2024), is characterized by Aβ deposition in the walls of larger leptomeningeal and cortical vessels, sparing the capillaries. Type II is associated with increased formation of fibrils and recurrent hemorrhages, but typically not with brain atrophy or neurofibrillary tangle formation (Timmers et al., 1990; Wisniewski et al., 1991; Wattendorff et al., 1995).

Although CAA is typically considered non-inflammatory, perivascular inflammation may occur with aging due to high levels of Aβ, contributing to loss of vascular integrity and function (Miao et al., 2005; Park et al., 2014; Rosas-Hernandez et al., 2020; Chwalisz, 2021; Kozberg et al., 2021, 2022). When Aβ accumulates in the capillaries (CAA type I), the risk of inflammation increases, sometimes giving rise to a distinct inflammatory subtype called CAA-related inflammation (CAA-ri) (Grasso et al., 2021). An additional subtyping classification is CAA with and without deep perforator arteriopathy (DPA) (Goeldlin et al., 2024). DPA, like CAA, is a common form of sporadic cerebral small vessel disease that affects small arteries, veins, arterioles, venules, and capillaries in deep supratentorial structures (such as basal ganglia and thalamus) and in the brainstem (Schreiber et al., 2020; Goeldlin et al., 2023). Both conditions share several similarities, such as being more prevalent in the aging brain, being major causes of ischemic stroke and ICH, and leading to cognitive decline (Schreiber et al., 2020). However, while in CAA the alterations in brain network are more predominant in the occipital and posterior temporal lobes, in DPA those are more widely seen in the frontal and lateral temporal lobes (Schreiber et al., 2020). Similarities and differences suggest that CAA and DPA, although separate conditions, can also coexist (Schreiber et al., 2020).

The majority of CAA cases are sporadic, primarily associated with aging, and influenced by a combination of genetic susceptibility and environmental and lifestyle factors; however, familial forms of CAA do exist. Familial CAA has been linked to rare autosomal dominant mutations, many of which affect the amyloid precursor protein (APP) gene. The Dutch (E693Q) and Iowa (D694N) mutations in the APP gene result in the loss of neighboring negatively charged residues in the Aβ peptide, promoting its aggregation and likelihood for vascular deposition over parenchymal plaque formation (Van Nostrand et al., 2001; Davis et al., 2004; Baumketner et al., 2008). The Iowa mutation, associated with CAA type I (Rajpoot et al., 2022), is classified as a non-hemorrhagic variant but is highly toxic to cerebrovascular smooth muscle cells. It induces extensive structural changes and promotes perivascular inflammation, involving activation of astrocytes and microglia (Shin et al., 2002). Conversely, the Dutch mutation is associated with CAA type II (Rajpoot et al., 2022). Other APP mutations that give rise to familial CAA include the Icelandic, British, Danish, and Familial amyloidosis-Finnish mutations (Vargas-George and Dave, 2022). Additionally, the Italian (E693K) and Piedmont (L705V) mutations exhibit features of CAA; Arctic (E693G) mutation exhibits features of both CAA and AD; and Flemish (A692G) mutation exhibits vascular Aβ deposition besides parenchymal deposition (Vargas-George and Dave, 2022). In humans, CSF analysis has shown that, in individuals with heritable CAA, decreased Aβ₄₀ levels in the CSF is an early feature of CAA pathogenesis (van Etten et al., 2017), likely reflecting its increased sequestration in the cerebral vasculature rather than clearance into the CSF. Similarly, other rare hereditary forms of CAA exist due to mutations in different genes, such as CST3 (or ACYS), PRNP, ITM2B (or BRI2), GSN (or NCBI), TGFBR1, PSEN1, A1AC, MME, LRP1, ACE, variant of CAA and CR1 (Yamada, 2015; Young et al., 2021; Cozza et al., 2023). Mutations in these genes also result in Aβ accumulation, impaired perivascular clearance, and disruption of cerebral blood vessels (Cozza et al., 2023). These findings highlight the distinct pathophysiological features of familial CAA and its relevance to sporadic forms.

Different alleles of the apolipoprotein E (ApoE), a protein involved in lipid transport, fat metabolism, and brain injury repair (Liu et al., 2013), are recognized as important polygenic risk factors for both Alzheimer’s disease (AD) and CAA (DeSimone et al., 2017). The APOE ε4 allele is associated with a dose-dependent increase in vascular Aβ deposition, cognitive decline, and non-hemorrhagic MRI markers of small vessel disease, such as white matter hyperintensities and perivascular spaces. In contrast, the APOE ε2 allele is associated with a higher risk of lobar ICH and greater cortical superficial siderosis severity in CAA (DeSimone et al., 2017; Charidimou et al., 2019; Voigt et al., 2024). Additionally, the distribution of APOE alleles differs between CAA subtypes: the APOE ε4 allele is over four times more frequent in CAA type 1 (with capillary involvement) than in CAA type 2 (which spares capillaries), suggesting that ε4 may preferentially promote capillary Aβ deposition (Thal et al., 2002). Additionally, exosome sequencing studies suggest that rare genetic variants, such as SORL1, TREM2, ABCA7, and ATP8B4, may also be shared genetic risk factors for both CAA and AD (Grangeon et al., 2024).

Iatrogenic CAA (i-CAA) is a rare form of CAA that typically develops as a late consequence (2–3 decades) of neurosurgical interventions in childhood that causes prion-like spread of misfolded Aβ from cadaveric materials and instruments used in the procedure (Banerjee et al., 2022; Greenberg and Charidimou, 2023; Kaushik et al., 2023; Storti et al., 2023). Several preclinical studies have demonstrated the prion-like properties of Aβ (Kane et al., 2000; Meyer-Luehmann et al., 2006; Stöhr et al., 2012; Burwinkel et al., 2018). Three-months-old Tg2576 mice, unilaterally infused into the hippocampus with dilute supernatants of neocortical homogenates from autopsied AD patients, developed profuse Aβ plaques and vascular deposits in the injected hemisphere 5 months after the injection, showing that Aβ can be seeded in vivo (Kane et al., 2000). Similarly, young, male APP23 transgenic mice developed robust Aβ deposition in the hippocampus 4 months after the region was injected with brain homogenates from either autopsied AD patients or from aged, Aβ-laden APP23 mice (Meyer-Luehmann et al., 2006). The same strain of mice, when inoculated in the right cerebral hemisphere with brain homogenates from aged transgenic APP23 mice, sustained an increase in the bioluminescence imaging (which can monitor Aβ and PrP prion kinetics) signal in the brain approximately 261 days post inoculation, with significantly higher Aβ levels after 330 days (Stöhr et al., 2012). Additionally, 6–8-week-old APP/PS1 mice, intracerebrally injected with 10% brain homogenates derived from aged APP/PS1 mice, showed pronounced Aβ plaques and vascular deposition in the thalamus after 360 days. Similar results were observed when mice were intravenously injected with the diluted brain extracts (Burwinkel et al., 2018). Taken together, these results suggest that Aβ, through its seeding properties, can lead to development of CAA (due to Aβ deposition) and progression of CAA (due to Aβ aggregation), in addition to having a potential for propagation, spreading, and transmission. Due to its rarity, the exact prevalence of i-CAA is unknown, although Kaushik et al. (2023) reported 49 through 2023. i-CAA is associated with early onset of disease, though radiological and clinical features, including ICH, seizures, and cognitive impairment, are comparable to sporadic CAA (sCAA) (Banerjee et al., 2022). A recent study reported that CSF and plasma biomarkers for Aβ₄₀, Aβ₄₂, and total tau were similar between patients with i-CAA and sCAA; however, cognitive dysfunction and cardiovascular risk factors were more prevalent in those with sCAA than i-CAA (Pollaci et al., 2024).

Studies have shown that CAA incidence increases with age. A 1983 postmortem study (N = 84, 60 ≤ age ≤ 97) found that CAA was present in 36% of the brains of patients over the age of 60, and in 46% of those over the age of 70 (Vinters and Gilbert, 1983). Similarly, a 2021 meta-analysis of 170 studies including over 73,000 subjects studied post-mortem estimated that approximately a quarter of the study’s general population (individuals who represent the general elderly society, including those with known cerebrovascular or neurodegenerative diseases) presented moderate to severe CAA pathology (mean age = 84.9 years) (Jäkel et al., 2022). Additionally, CAA pathology was found in ∼5% of cognitively normal elderly, 20%–25% of patients with ICH, and 50%–60% of patients with lobar ICH (Jäkel et al., 2022).

Studies have also compared the incidence in men vs. women, although data remain limited compared to other conditions like AD. Between the years of 1971 and 1983 in Japan, 400 brains of individuals who were 40+ years old were age-matched and it was found that CAA was present in 18.3% of the men compared to 28% of the women (Masuda et al., 1988). However, more recent studies have not reported significant differences in CAA incidence in men vs. women (Chwalisz, 2021; Wagner et al., 2021; Koemans et al., 2023), although earlier onset and increased severity were reported in men (Shinohara et al., 2016). A study found that men with sporadic CAA experience earlier onset of symptomatic CAA-related ICH and more lobar microbleeds than women (Koemans et al., 2023). Additionally, men with familial CAA (Dutch-type) had a greater number of symptomatic ICH and a shorter interval between the first and second symptomatic ICH (Koemans et al., 2023). While the influence of sex hormones, particularly the decline in estrogen after menopause, has been hypothesized to affect amyloid processing (Xu et al., 2006; Liang et al., 2010), its direct role in CAA development remains unclear and warrants additional investigation. Notably, there is a lack of sex-stratified analyses in many studies on CAA, necessary to understand CAA-related risk and to evaluate the role of risk factors and efficacy of novel therapeutics. Moreover, data is even more limited for transgender or non-binary individuals and their risk profiles.

Sex differences in CAA have also been explored in mouse models. In Tg-SwDI mice, females have more pronounced deficits in spatial and contextual memory, a higher burden of cerebral microbleeds, lower levels of proinflammatory cytokines IL-1α, IL-2, IL-9, and INF-γ, and higher levels of monocytes/macrophages (Maniskas et al., 2021; Finger et al., 2022; Setti et al., 2022). Similarly, compared to males, 12-months old APP/PS1 female mice exhibit more severe CAA, in addition to higher parenchymal Aβ burden (especially in the hippocampus), astrocytosis and microgliosis, more microhemorrhages, higher levels of phosphorylated tau protein and proinflammatory cytokines, and greater neuronal and synaptic degeneration (Jiao et al., 2016).

Gene by sex interactions have also been observed for CAA-related outcomes. When the effect of ApoE is considered, opposite results are reported in humans and mice. In humans with mild cognitive impairment and AD, the APOE ε4 allele was associated with higher risk of cerebral cortex microbleeds in men than in women, although that was also dependent on other risk factors, such as hypertension, diabetes, and age (Shams et al., 2015; Cacciottolo et al., 2016; Finch and Shams, 2016). However, in E4FAD mice, females showed higher levels of CAA, cortical microbleeds, plaques, and soluble Aβ (Cacciottolo et al., 2016; Finch and Shams, 2016). Further investigation is necessary to explain this discrepancy.

Racial and ethnic differences in CAA and related outcomes have also been explored. An autopsy study of AD patients found similar rates of CAA in Caucasian and African American individuals; this lack of difference was confirmed in an independent sample from the National Alzheimer’s Coordinating Center database (Kamara et al., 2018). Data obtained from the Uniform Data Set suggested that being Hispanic predicted a higher likelihood of severe CAA (Ringman et al., 2014). Yet, another study reported that minority survivors of CAA-related ICH exhibited lower CAA burden as measured on MRI (Castello et al., 2021), while a meta-analysis found that APOE ε4 increased the risk for recurrent lobar ICH in White but not minority survivors of ICH (Marini et al., 2019). These findings may have been confounded by differences in survival outcomes in individuals with similar CAA burden. Some studies have reported differences in regional CAA distribution, lower prevalence and decreased severity of CAA pathology, lower proportion of CAA-related ICH, and lower prevalence of strictly lobar microbleeds in East-Asian populations compared to Western populations (Masuda et al., 1988; Ng et al., 1991; Chen et al., 2010; Yakushiji et al., 2020; De Kort et al., 2024).

There may also be interactions between race/ethnicity and genetic risk factors, like APOE alleles. For example, a meta-analysis reported that APOE ε2 and APOE ε4 were associated with lobar ICH risk in White but not Black or Hispanic participants; however, after controlling for hypertension, APOE ε4 was associated with lobar ICH risk in Hispanic but not Black participants (Marini et al., 2019). Despite these differences, current research is hampered by underrepresentation of non-white populations and reliance on autopsy or referral-based cohorts, which may not be representative. Better population-based data with diverse cohorts is necessary to understand CAA-related risk and to evaluate the role of risk factors and efficacy of novel therapeutics.

A definitive diagnosis of CAA can only be made postmortem through histopathological confirmation of Aβ deposition in cerebral vessel walls (Kuhn and Sharman, 2025). However, clinical diagnostic criteria, most notably the Boston Criteria, are used to diagnose probable or possible CAA in living patients. First introduced in 1995 and recently updated (Figure 2), the latest version (Boston Criteria v2.0), offers improved sensitivity but slightly reduced specificity compared to previous versions (Zanon Zotin et al., 2024). The criteria classify CAA into four categories: definite CAA, probable CAA with supporting pathology, probable CAA, and possible CAA (Illsley and Ramadan, 2014; Kuhn and Sharman, 2025).

Imaging techniques used to diagnose CAA include magnetic resonance imaging (MRI) and computed tomography (CT) scans, which help detect areas of cerebral hemorrhage, most often in the posterior lobar cortical and subcortical regions (Sakurai et al., 2014; Kuhn and Sharman, 2025). In cases where CAA-ri is suspected, a brain biopsy can be performed to confirm the presence of Aβ deposition (Kuhn and Sharman, 2025). Among MRI modalities, T2*-weighted gradient-recalled echo (T2*-GRE) and susceptibility weighted imaging (SWI) are particularly useful for identifying cerebral microbleeds, a hallmark of CAA. SWI is significantly more sensitive than T2*-GRE in detecting these microhemorrhages, with studies reporting that T2*-GRE can miss microhemorrhages in ∼25% of patients (Haacke et al., 2007; Charidimou et al., 2013; Sakurai et al., 2014). Although CT scans can identify larger hemorrhages, they have a lower contrast resolution compared to MRI and are considerably less sensitive for the microhemorrhages seen with SWI MRI (Sakurai et al., 2014; Azmin et al., 2015). Instead, MRI, especially using SWI sequences, remains the preferred modality for detecting the subtle hemorrhagic changes characteristic of CAA. A proposed study using high frequency serial magnetic resonance imaging to prognose CAA (NCT06128824) will attempt to visualize cortical superficial siderosis (cSS) and hemorrhagic and ischemic lesions as surrogate markers for CAA.

Tissue biopsy can be an important tool for diagnosing CAA in life. A study by Greenberg and Vonsattel (1997) examined brain specimens from individuals with confirmed CAA-related hemorrhage and found vascular Aβ deposition in 100% of the specimens with hemorrhage, suggesting high sensitivity under ideal sampling conditions. However, several issues arise from generalizing these results to clinical situations, as the characteristics of biopsied tissues (e.g., size and location) can affect detection of vascular Aβ (Greenberg and Vonsattel, 1997). The distribution of CAA pathology is patchy and segmental, meaning that not all tissue samples will capture affected vessels. This heterogeneity can lead to false negatives if vascular Aβ is absent in the sampled area, which is limited to just a small section of cortical tissue (Greenberg and Vonsattel, 1997). The prevalence of Aβ pathology increases with age, and some degree of vascular Aβ deposition is often present in older individuals; however, its presence does not necessarily indicate that CAA caused a hemorrhage, potentially leading to false positives (Greenberg and Vonsattel, 1997). Therefore, the specificity of biopsy findings is assumed to be age-dependent. In younger patients, the detection of vascular Aβ is more likely to indicate pathological CAA, whereas in older individuals, incidental Aβ deposition may reduce specificity. As such, clinical interpretation of biopsy results must consider both patient age and the inherent limitations of localized sampling (Greenberg and Vonsattel, 1997).

Amyloid-Positron Emission Tomography (PET) scans is a commonly used technique to assess Aβ levels in the brain (Verbeek et al., 2009). Amyloid PET imaging using tracers such as [¹¹C]PiB or [¹⁸F]florbetapir can detect fibrillar Aβ deposits in the brain and is commonly used to assess amyloid pathology in AD (Landau et al., 2013), while ^18FFlutemetamol can be used to help reclassify mixed arteriosclerosis + CAA into CAA pathology and arteriolosclerosis-predominant pathology (Romoli et al., 2024). While CAA can also be visualized by amyloid PET, it is not specific to CAA, as it detects total amyloid burden without distinguishing between vascular and parenchymal deposits. As a result, while a positive amyloid PET scan may suggest the presence of CAA, especially when combined with clinical criteria and MRI findings, it cannot definitively differentiate CAA from AD or confirm a diagnosis of CAA on its own.

Similar to amyloid PET imaging, cerebral spinal fluid (CSF) samples also reflect Aβ levels in the brain. It has been reported that the levels of Aβ₄₀ in the CSF are decreased in patients with CAA (Verbeek et al., 2009; Charidimou et al., 2018; Sembill et al., 2023). Although studies have suggested that the levels of Aβ₄₀ in the CSF are significantly lower in CAA compared to AD, CSF analysis alone is insufficient to fully differentiate between the two conditions (Verbeek et al., 2009; Charidimou et al., 2018; Sembill et al., 2023). A study by Sembill et al. (2023) identified a distinct CSF biomarker profile in CAA patients compared to patients with AD, mild cognitive impairment due to AD, mild cognitive impairment with unlikely AD, and healthy controls (Table 2).

In parallel with the development of blood biomarkers for AD, similar studies are in progress to prognose and diagnose CAA (NCT06960538). These biomarkers include Aβ₄₀, Aβ₄₂, total tau (t-tau), neurofilament light chain (NfL), and glial fibrillary acidic protein (GFAP), in addition to lipidomic profiling.

Cerebral amyloid angiopathy is strongly associated with AD (Figure 3; Saito et al., 2021). Both conditions involve the accumulation of Aβ peptides but differ in the site of accumulation. In AD, Aβ aggregates into parenchymal plaques primarily within the frontal and parietal association cortices, with later involvement of the hippocampus (Yoon and Jo, 2012; Greenberg et al., 2020). In contrast, CAA is marked by Aβ deposition within the vessel walls of small- to medium-sized arteries, arterioles, and capillaries of the leptomeninges and cerebral cortex (Weller et al., 1998, 2008; DeSimone et al., 2017; Chwalisz, 2021). In addition to differences in anatomical distribution, the predominant Aβ isoforms also differ. Aβ42 is the main component of parenchymal plaques in AD, whereas perivascular deposits in CAA are predominantly composed of Aβ40, although Aβ₄₂ is also present (Gatti et al., 2020; Chwalisz, 2021).

Although they share overlapping features, CAA and AD are distinct pathological entities; however, CAA pathology is present in approximately 90% of AD cases (Saito et al., 2021). In cases where they co-exist, CAA is believed to exacerbate both the pathological burden and clinical manifestations of AD. As vascular amyloid accumulates, the perivascular drainage route becomes blocked, reducing Aβ clearance and promoting its accumulation (Weller et al., 2008; Chwalisz, 2021). Routes of Aβ clearance are illustrated in Figure 4.

Recent research has highlighted the complex interplay between CAA, Aβ pathology, and tau-related neurodegeneration in AD, as well as their combined contributions to cognitive impairment. In Rabin et al. (2022) demonstrated that severe CAA pathology and high parenchymal Aβ levels were associated with increased tau deposition and accelerated cognitive decline. Godrich et al. (2024) attempted to replicate Rabin et al.’s (2022) findings using a larger dataset, confirming significant relationships with AD-related neuropathology and cognitive impairment. They reported that CAA interacts with neuritic plaques (senile plaques composed of both Aβ and immune cells) to influence neurofibrillary tangle burden and contributes to cognitive impairment by amplifying the effect of tangles on cognition, rather than acting independently (Godrich et al., 2024). More recently, Jury-Garfe et al. (2025) reported that in a CAA mouse model lacking endogenous tau (Tg-FDD/mTau^–/–), the depletion of tau reduced amyloid deposition in the vasculature, prevented vascular damage, ameliorated motor and synaptic impairments, and decreased astrocytic reactivity, suggesting that downregulation of tau in CAA can function as a potential therapeutical strategy for treating patients with parenchymal and vascular amyloid deposition. Additionally, Kang et al. (2025) recently reported that lobar cerebral microbleeds (a CAA imaging marker) are associated with phosphorylated tau 217 (p-tau217), glial fibrillary acid protein (GFAP), and neurofilament light chain (NfL), which are AD plasma biomarkers that reflect tau hyperphosphorylation, astrocytic activation, and neuronal damage, respectively. Additionally, it was found that lobar cerebral microbleeds, in conjunction with p-tau217 and GFAP, were associated with cognitive impairment in a synergistic manner (Kang et al., 2025). A recent study of CAA patients with an AD profile present significant Cornu Ammonis (CA) 2-CA3 atrophy compared to CAA patients without an AD profile, suggesting that CA2-CA3 volume can serve as a radiological marker for identifying CAA-AD overlaps (Storti et al., 2025). This emerging evidence underscores the multifaceted relationships between CAA, Aβ pathology, and tau-driven neurodegeneration in AD, highlighting the need for an integrated approach to treatment.

In addition to exacerbating AD pathology and symptoms, CAA also complicates AD treatment with the newest anti-amyloid drugs (e.g., monoclonal antibodies). In treated patients, CAA increases the risk of amyloid-related imaging abnormality (ARIA), such as brain swelling (ARIA-E) and/or bleeding (ARIA-H), that can potentially exclude them from continued treatment (Sveikata et al., 2022; Agarwal et al., 2023; Sin et al., 2023). This is hypothesized to occur because these drugs break down parenchymal plaques and increase the mobilization of Aβ aggregates to the vasculature, increasing the severity of CAA (Hampel et al., 2023). A literature review by Barakos et al. (2022) reported that the incidence of ARIA-E has varied from 0.9% to 40.6%, while that of ARIA-H has varied from 0.5% to 28.4%. However, these ranges do not differentiate between different subgroups or drugs. Clinical trials have provided insight into the percentage of patients, within the general population of the studies, who developed ARIA after receiving specific anti-amyloid drugs (Table 3). Therefore, while anti-amyloid therapies show promise for AD treatment, their effectiveness is limited by CAA-related risks such as ARIA. This underscores the need for tailored treatment strategies that account for vascular amyloid burden.

Given the high comorbidity of CAA pathology with AD (DeSimone et al., 2017; Chwalisz, 2021; Saito et al., 2021), and its contribution to conditions such as vascular cognitive impairment and stroke (Wattendorff et al., 1995), the development and use of animal models of CAA are critical for advancing our understanding of pathophysiological mechanisms and for evaluating the efficacy and safety of therapeutic interventions targeting CAA. Several mouse models that exhibit CAA exist (Table 4). In Davis et al. (2004) created a transgenic mouse model of CAA, the Tg-SwDI mouse. These mice exhibit the Dutch and Iowa mutations, as well as the Swedish K670N/M671L mutation, a double mutation that immediately precedes the aspartic acid at position 672 at the β-secretase cleavage site in the APP, causing Aβ₄₀ and Aβ₄₂ to be overproduced (Citron et al., 1994). Tg-SwDI mice show high levels of insoluble Aβ₄₀ plaques (Robison et al., 2019), the main isoform present in CAA (Chwalisz, 2021), with modest levels of Aβ₄₂. Aβ accumulation occurs primarily in the vasculature, though parenchymal accumulation is also observed, starting at ∼3 months of age with severity increasing over their lifespan (Xu et al., 2014; Paul et al., 2018; Rosas-Hernandez et al., 2020). Vascular amyloid burden is greatest in the subiculum, followed by the thalamus, but is almost absent in the cortex (Robison et al., 2019). Robust CAA-associated inflammation is exhibited in the Tg-SwDI mouse model (Fan et al., 2007a), which displays high expression of proinflammatory cytokines [e.g., tumor necrosis factor-alpha (TNF-α), interleukin-1beta (IL-1β), and interleukin-6 (IL-6)] (Robison et al., 2019). In behavior tests, these mice exhibit impaired spatial learning and memory in the Barnes maze test as early as 3 months of age (Xu et al., 2007; Robison et al., 2019, 2020).

Van Nostrand et al. (2001) have more recently created rat models carrying the familial CAA mutations, including the rTg-DI (Davis et al., 2018, 2022; Popescu et al., 2020; Zhu et al., 2020; Lee et al., 2021; Schrader et al., 2021, 2022a; 2022b,2024a; 2024b; Stanisavljevic et al., 2022; Vervuurt et al., 2022, 2024; Koundal et al., 2024) and rTg-D (Irizarry et al., 2025) models, exhibiting Type I and Type II CAA, respectively. In addition to these, other rat models exist (e.g., F344-AD, and TgAPP21), as do naturally occurring models [e.g., primates (baboon, cynomolgus monkeys, gray mouse lemur, rhesus monkeys, squirrel monkeys, and velvet monkeys), cats, and dogs] (Jäkel et al., 2017).

It is important to note that this review focuses on CAA caused by the accumulation of Aβ in cerebral vasculature, as the term “amyloid” is generic with respect to the identity of the proteins or peptides that make up amyloid aggregates. Specifically, they have “characteristic fibrillar electron microscopic appearance, typical X-ray diffraction pattern and histological staining reactions, particularly affinity for the dye Congo red with resulting green birefringence” (Westermark et al., 2007). According to the 2024 Nomenclature meeting of the International Symposium for Amyloidosis (Buxbaum et al., 2024), the terms APP and plaques, used to describe the amyloid aggregates that are associated with AD, are technically incorrect. This most recent Symposium identified 42 different proteins that are associated with amyloidosis that all meet the criterion of “APP.” Thus, the recommendation of the Symposium is that the APP from which Aβ is derived should be called Aβ precursor protein (AβPP). Additionally, the Symposium recommended avoiding the word plaque in AD, as it refers to a flat object, whereas the amyloid aggregates associated with AD are globoid in shape. However, the Symposium recognized that the near universal terminology of APP and Aβ plaques in the AD literature is unlikely to be changed due to technical inaccuracy.

While Aβ deposits have been found in non-neural tissues surrounding arterioles and microvessels, they were reported to be extracellular (Joachim et al., 1989) precluding a determination of amyloid angiopathy. Moreover, non-neural Aβ was deemed unlikely to cause clinical dysfunction akin to that of CAA.

A variant CAA arises from the formation of aggregates of transthyretin (TTR), a protein derived from the liver and choroid plexus (Brunet de Courssou et al., 2025). However, this variant of CAA is far rarer than that resulting from the accumulation of Aβ aggregates. It is associated with a naturally occurring mutation in the transthyretin gene that has a neuropathological phenotype distinct from that of CAA involving Aβ aggregates (Sekijima et al., 2016). Hereditary TTR amyloidosis affects peripheral and autonomic nervous system and heart, although effects in the kidneys, eyes, and digestive system can also be observed (Brunet de Courssou et al., 2025). In cases where TTR amyloidosis results in CAA, patients tend to present more focal central neurological symptoms, seizures, and memory complaints, compared to patients with TTR amyloidosis but no CAA (Brunet de Courssou et al., 2025).

Currently, there are no FDA-approved medications that specifically treat or prevent CAA (Kozberg et al., 2021). Management of the disease mainly revolves around reducing the risk of first-time or recurrent lobar ICH and controlling symptoms (Gatti et al., 2020; Kozberg et al., 2021). However, ongoing research is exploring strategies to lower Aβ deposition during different disease stages, such as enhancing perivascular clearance of Aβ through pharmacological and non-pharmacological interventions (Kozberg et al., 2021; Cozza et al., 2023).

The bioactive flavanonol taxifolin, also known as dihydroquercetin, has antioxidant and anti-inflammatory properties. It has been reported to disaggregate Aβ fibrils and inhibit their formation, making it a potentially safe therapeutic agent to treat CAA (Sato et al., 2013a; Saito et al., 2021). The properties of taxifolin have been demonstrated by transmission electron microscopy imaging in a mouse model of CAA (Sato et al., 2013a,b; Saito et al., 2017), and by thioflavin T fluorescence assays showing disintegration of Aβ fibrils with inhibition of new ones. This could facilitate Aβ clearance and restore cerebrovascular reactivity and spatial memory (Sato et al., 2013a; Saito et al., 2017). In wild-type mice injected with oligomeric Aβ into the hippocampus, taxifolin prevented spatial memory deficits that are typically induced by the oligomer (Wang et al., 2018). Although taxifolin has shown promising results in mouse models, its effects in humans with CAA have yet to be evaluated in clinical trials.

Cerebrolysin^®, a parenterally administered porcine brain-derived neurotrophic peptidergic drug (Plosker and Gauthier, 2009) indicated for the treatment of central nervous system (CNS) disorders in Europe and Asia², has shown positive effects in the treatment of AD in both animal models and humans. However, Cerebrolysin^® does not have FDA approval in the United States. Treatment with Cerebrolysin^® was shown to decrease Aβ deposition around the blood vessels, reduce perivascular microgliosis and astrogliosis, increase the expression of vascular fitness markers, reduce changes in the endothelial and smooth muscle cells, and preserve basal membranes and intracellular junctions in mThy1-hAPP751 Tg mice (a mouse model of AD) treated with Cerebrolysin^® for 3 months, starting at 7 or 12 months of age (Rockenstein et al., 2005). Additionally, Cerebrolysin^® significantly reduced amyloid burden and ameliorated synaptic impairments in the frontal cortex of 5-months-old mThy1-hAPP751 Tg mice (Rockenstein et al., 2002). When this strain of mice was treated with Cerebrolysin^® for 3 months, starting at 3 or 6 months of age, results showed improvement in behavioral performance, synaptic regeneration, and reduction in the proportion of neurons displaying DNA fragmentation (Rockenstein et al., 2003). In humans with advanced AD, it has been shown that treatment with Cerebrolysin^® in combination with donepezil (an acetylcholinesterase inhibitor prescribed to mitigate AD symptomatology) normalizes the levels of serum vascular endothelial growth factor (VEGF) – which increases with AD severity, playing a paradoxical role in the disease (Ali and Bracko, 2022) – while significantly improving cognitive function; however, no therapeutic effects were observed in mild to moderate AD cases (Alvarez et al., 2020).

Tramiprosate (Cerebril^®), an orally administered low molecular weight aminosulfonate compound, binds to soluble Aβ₄₂ at amino acids Lys16, Lys28, and Asp23, stabilizing it and reducing plaque formation and providing neuroprotection (Greenberg et al., 2006; Manzano et al., 2020; Cozza et al., 2023). In TgCRND8 mice treated with subcutaneous injections of 30 or 100 mg/kg/day Tramiprosate for 8–9 weeks, the drug was found to cross the BBB; reduce the % area occupied by plaques by 29% in both groups; reduce the number of plaques by 24% and 18%, respectively; reduce Aβ₄₀ plasma levels by 37% and 61%, respectively; and reduce Aβ₄₂ plasma levels by 31% and 67%, respectively (Gervais et al., 2007). Additionally, fifth generation progeny obtained from initially crossing TgCRND8 mice with C57BL/6 mice, and then alternately crossing the progeny with FVB or C57BL/6 mice up to the fifth generation, were treated for 9 weeks with 500 mg/kg/day Tramiprosate (Gervais et al., 2007). Results showed that brain levels of soluble Aβ₄₀, insoluble Aβ₄₀, soluble Aβ₄₂, and insoluble Aβ₄₀ were reduced by 30%, 31%, 25%, and 22%, respectively (Gervais et al., 2007). In phase 1 clinical trials, Tramiprosate was administered for up to 1 week with favorable pharmacokinetic properties and no major safety issues. However, minor side-effects, particularly nausea and vomiting, were shown to be dose-dependent (Greenberg et al., 2006). In phase 2 clinical trials, Tramiprosate showed no significantly beneficial effects on cerebral microbleeds, but data suggested that the drug could be safely given to CAA patients due to no major safety issues being reported (Greenberg et al., 2006; Aisen et al., 2011; Saito et al., 2021). In a study by Gauthier et al. (2009), 508 patients underwent volumetric MRI scanning to assess disease modification of the hippocampus while receiving 100 or 150 mg Tramiprosate twice daily for 78 weeks. Results showed that treatment, particularly with the 150 mg dose, slowed down hippocampal atrophy and revealed some evidence of positive effects on cognition, based on the Alzheimer’s Disease Assessment Scale-cognitive subscale (ADAS-cog) (Gauthier et al., 2009).

In cases of CAA-ri, patients respond well to immunosuppressive therapy, typically corticosteroids, although cyclophosphamide, azathioprine, mycophenolate mofetil, methotrexate, and immunoglobulin are also commonly used (Auriel et al., 2016; Meng et al., 2024). According to Kirshner and Bradshaw (2015), improvement is observed within a few weeks of treatment and occurs in approximately 72% of patients treated with immunosuppressants. Although the exact duration of treatment is yet to be determined and varies depending on the patient, a typical course of treatment consists of a high-dose steroid pulse given intravenously followed by a prolonged steroid taper of at least 6 months (Kirshner and Bradshaw, 2015; Kozberg et al., 2021). Despite the efficacy of the treatment, cases of recurrence have been reported. A retrospective cohort study by Regenhardt et al. (2020) following 48 people suffering from CAA-ri found that recurrence occurred in 26% of those who received any immunosuppressive agent, compared to 71% of those who did not receive any. Recurrence can occur at the initial inflammation site and/or additional sites (DiFrancesco et al., 2015; Kozberg et al., 2021).

A novel investigational approach to mitigate CAA and AD disease progression is the use of an RNA interference (RNA_i) therapeutic called mivelsiran (ALN-APP). The drug, which is administered intrathecally, is designed to lower Aβ levels by decreasing APP synthesis (Cohen et al., 2024). According to the manufacturer (Alnylam^® Pharmaceuticals), the drug acts by decreasing APP messenger RNA (mRNA) in the central nervous system, which reduces both APP synthesis and APP-derived cleavage products, such as Aβ³. During phase 1 clinical trials (NCT05231785), the safety, tolerability, pharmacokinetics, and pharmacodynamics of mivelsiran in patients with early-onset AD was evaluated, after single doses of 25, 50, 75 mg, or placebo were administered for 6 months. Interim results showed that 50 and 75 mg of mivelsiran are well-tolerated and produce sustainable reductions in the levels of soluble APP in the CSF, as well as Aβ₄₂ and Aβ₄₀ (Cohen et al., 2024). Phase 2 clinical trials (cAPPricorn-1/NCT06393712) will evaluate the safety, tolerability, and pharmacodynamics of mivelsiran in adult patients with sCAA and Dutch-type CAA, as well as evaluate its effects on markers of CAA progression, such as rate of new lobar cerebral microbleeds. If approved, mivelsiran will offer a new mechanism of action for drugs used in the treatment of CAA and AD, as it will focus on reducing APP synthesis and, as a consequence, Aβ isoforms, instead of focusing on Aβ peptide/plaques clearance after they have already formed, like currently approved drugs do.

It is important to note that there are certain medications that should be avoided following CAA diagnosis. The use of anticoagulant and antithrombotic agents has been generally discouraged, as these significantly increase the risk of hemorrhagic stroke in those with CAA (Banerjee et al., 2017; Sheth, 2022; Merella et al., 2023). Similarly, antiplatelet agents have been associated with an increased number of cerebral microbleeds and a higher risk of ICH recurrence in patients with CAA (Vernooij et al., 2009; Biffi et al., 2010). These agents should be avoided unless there is a compelling clinical indication for their use, such as a high risk of ischemic stroke due to blood clot formation that clearly outweighs the risk of ICH (Kozberg et al., 2021; Merella et al., 2023).

Emerging evidence suggests that dietary factors may influence the development and progression of CAA, potentially through effects on vascular health, inflammation and oxidative stress, and amyloid metabolism. In recent years, ketogenic diets (high-fat, low carbohydrate, moderate protein intake) have gained attention for their ability to promote brain health, improve cognition, and protect against other cerebrovascular and neurodegenerative diseases (Pulido-Correa et al., 2025). Pulido-Correa et al. (2025) have recently shown that consumption of a ketogenic diet increased exploratory activity and improved spatial learning and memory in male Tg-SwDI mice. These cognitive-behavioral benefits were associated with an attenuation of parenchymal amyloid pathology and increased hippocampal neurogenesis; however, cerebrovascular amyloid levels were unchanged (Pulido-Correa et al., 2025).

The Mediterranean diet has received considerable attention in recent years due to its potential neuroprotective and cardiovascular benefits (Guasch-Ferré and Willett, 2021; Laffond et al., 2023; Pant et al., 2023; Picone et al., 2024; Fekete et al., 2025). This diet is characterized by a high intake of fruits, vegetables, whole grains, legumes, nuts, and olive oil; moderate consumption of fish and poultry; low intake of red meat and processed foods; and limited consumption of dairy products and sweets. The MIND (Mediterranean-DASH Diet Intervention for Neurodegenerative Delay) diet, a hybrid of the Mediterranean and DASH diets, was specifically developed to reduce the risk of AD and cognitive decline. The MIND diet is similar to the Mediterranean diet but puts a stronger emphasis on specific brain-healthy foods like leafy greens and berries. Both diets have been consistently associated with a reduced risk of cognitive decline and dementia across numerous epidemiological and clinical studies (van den Brink et al., 2019); however, little is known about their potential for protecting against CAA, specifically. A study in Tg-SwDI mice explored the potential benefits of a diet rich in extra-virgin olive oil, a major component of the Mediterranean and MIND diets. Consumption of this diet improved cognitive-behavioral function and attenuated both vascular and parenchymal Aβ deposition, with evidence for reduced Aβ production and enhanced clearance pathways (Qosa et al., 2015). Similarly, it was found that a diet enriched with docosahexaenoic acid [an omega-3 fatty acid obtained from food (e.g., fatty fish) and fish oil supplements, that is essential for brain growth and development (Horrocks and Yeo, 1999)] reduces vascular Aβ accumulation, microhemorrhages, and inflammation in Tg2576 mice (Hur et al., 2018). A proposed clinical trial (NCT06933212) will be investigating the ability of Mediterranean diet to reduce cognitive decline in patients with CAA.

Several studies have suggested that consumption of a high-fat diet, rich in saturated and/or hydrogenated fats, as well as in simple and/or refined carbohydrates, can contribute to cognitive decline and the development and progression of AD (Kalmijn et al., 1997, 2004; Morris et al., 2003, 2006; Eskelinen et al., 2008; Kanoski and Davidson, 2011; Francis and Stevenson, 2013; Cordner and Tamashiro, 2015); however, less is known about its effects on CAA and related outcomes. In a study by Lin et al. (2016), 10 weeks of high fat diet increased cerebrovascular Aβ accumulation, elevated oxidative stress in the hippocampus, and exacerbated cognitive deficits in 5xFAD mice. Notably, high-fat diet selectively increased Aβ deposition in the microvessels of the hippocampus and in cortical arteries without affecting parenchymal Aβ levels (Lin et al., 2016). These pathological changes occurred despite minimal weight gain and only mild glucose intolerance in the high fat diet fed group of 5xFAD mice (Lin et al., 2016). Small sample sizes may have impacted the ability for these metabolic effects to reach significance; however, it was also noted that 5xFAD mice exhibited a blunted metabolic response to high fat diet compared to wild-type controls (Lin et al., 2016). It has also been reported that high fat diet-induced obesity did not increase levels of soluble or insoluble Aβ40 or Aβ42 in Tg-SwDI mice; however, this study did not distinguish between parenchymal and vascular Aβ accumulation (Zhang et al., 2013). High fat diet-fed Tg-SwDI mice also exhibited a localized increase in reactive microglia in the CA1 region of the hippocampus, as well as an increase in IL-6 expression in the striatum (Zhang et al., 2013).

Along with other environmental and genetic risk factors, consumption of a high fat diet can contribute to an increased risk for metabolic disorders, including prediabetes and type 2 diabetes mellitus (T2DM), dyslipidemia, and hypertension. Studies have suggested that metabolic disorders contribute to AD and CAA pathology. Vargas-Soria et al. (2022) reported that high fat diet-induced prediabetes accelerated amyloid accumulation in small to medium- sized vessels and increased Aβ plaque load in APPswe/PS1dE9 mice. Additionally, this study crossed APPswe/PS1dE9 with db/db mice (T2D model), resulting in T2D; findings suggest that T2DM enhanced Aβ accumulation in medium to larger- sized vessels, favoring vascular over parenchymal accumulation (Vargas-Soria et al., 2022). Additionally, both prediabetes and T2D were associated with attenuated vessel diameter, oxidative stress, and MMP activation (Vargas-Soria et al., 2022).

Several observational studies have shown an association between high cholesterol in late life and increased risk for mild cognitive impairment, AD, and vascular dementia (Kivipelto et al., 2001; Yaffe et al., 2002; Whitmer et al., 2005; Solomon et al., 2007, 2009). In humans, high cholesterol is associated with greater vascular Aβ (CAA), as well as an increase in neuritic plaques, diffuse plaques, and neurofibrillary tangles (Hawkes et al., 2015; Xu et al., 2020). However, in mice, this relationship has not yet been observed. Hohsfield et al. (2014) showed that hypercholesterolemia induced by prolonged consumption of a 5% cholesterol diet did not induce vascular amyloid accumulation in wild-type (C57BL/6J) nor 3xTg-AD mice, though intraneuronal Aβ and neuroinflammation were increased in the latter. However, measurements could not be conducted at an older age, as 3xTg-AD mice on the cholesterol diet did not survive up to 20 months (Hohsfield et al., 2014), making it impossible to know if prolonged 5% cholesterol diet would have effects similar to those seen in humans later in life.

Atherosclerosis is a chronic, progressive vascular disease characterized by the accumulation of lipids, inflammatory cells, and extracellular matrix components in medium and large arteries. This results in the formation of atherosclerotic plaques, which can lead to arterial wall thickening, loss of elasticity, and luminal narrowing. Over time, these plaques can calcify, rupture, or induce thrombosis, contributing to the development of cardiovascular events. Atherosclerosis may also contribute to AD and CAA pathology. Wang et al. (2021) showed that in 3xTg-AD mice (AD model) fed a high-fat diet, atherosclerosis induces the formation of platelet-associated fibrillar Aβ aggregates; increases CAA burden, tau pathology, and neuronal loss; impairs cerebral blood flow, and exacerbates cognitive deficits.

As defined by the World Health Organization, hypertension (also known as high blood pressure) occurs when the pressure in the blood vessels is ≥140/90 mmHg. It can be asymptomatic and is directly related to older age, genetics, overweight, sedentarism, high-salt diet, and high alcohol consumption.⁴ Hypertension can impact vascular clearance of Aβ and increase inflammation and the risk of ICH (Rius-Pérez et al., 2018; Singh et al., 2022). Studies have shown that BP management can provide protection against CAA-related ICH (Arima et al., 2010; Biffi et al., 2015), and that patients with CAA should ideally maintain their BP lower than 120/80 mmHg (Li et al., 2017; Kozberg et al., 2021). However, a study by Sin et al. (2024) analyzing data from 2,510 participants in the National Alzheimer’s Coordinating Center who had systolic BP measurements recorded before a CAA-related death, found no evidence of systolic BP being associated with CAA. Additionally, a study by Zhang et al. (2020) reported that hypertension may not be associated with worse outcomes in patients who suffered a CAA-associated ICH. In fact, hypertensive patients exhibited a 50% lower risk of death 3 months after a CAA-associated ICH compared to normotensive patients (Zhang et al., 2020). An alternative explanation provided by the authors is that prolonged use of antihypertensive medications in patients with hypertension may provide protection against CAA and its complications (Zhang et al., 2020), as was discussed previously.

Limited rodent studies have been performed investigating the role of hypertension in the progression of CAA. In 10-months-old rTg-DI rats (a rat model of CAA type I), non-pharmacological hypertension was shown to cause cognitive-behavioral deficits and the redistribution of vascular Aβ, reducing microvascular accumulation in the hippocampus and thalamus and increasing accumulation in the surface pial vessels (Stanisavljevic et al., 2022). Additionally, while hypertension increased the occurrence of vessel occlusions in the thalamus of the rTg-DI rats, it did not alter the number of cerebral microbleeds nor the extent of gliosis in the thalamus (Stanisavljevic et al., 2022). Due to conflicting results and few studies in rodent models, the relationship between BP and CAA requires further study.

Studies have established that exercise, as well as cognitively stimulating activities, can reduce the risk of dementia in humans (Cotman and Berchtold, 2002; Pope et al., 2003; Lange-Asschenfeldt and Kojda, 2008; Radak et al., 2010), healthy mice (Kempermann et al., 2002; Benaroya-Milshtein et al., 2004; Ekstrand et al., 2008; Birch et al., 2013), and other rodent models of AD (Arendash et al., 2004; Jankowsky et al., 2005; Lazarov et al., 2005; Costa et al., 2007; Hu et al., 2010; Andrade-Guerrero et al., 2023). In humans, it has been shown that exercise provides protective effects that are pathology-specific in the structure and function of aging brains (Memel et al., 2021). For example, exercise provided resilience against arteriosclerosis, atherosclerosis, and Aβ; while providing resistance against cerebrovascular dysfunction and Lewy body disease (Memel et al., 2021). Additionally, studies have shown that microinfarcts, neuronal loss, and white matter pathologies are associated with lower levels of total daily physical activity in older adults (Buchman et al., 2018).

Numerous studies have reported that exercise is neuroprotective in mouse models of AD. For example, a study by Andrade-Guerrero et al. (2023) using 13-months-old 3xTg-AD mice also showed that voluntary physical exercise reduced Aβ pathology and improved cognitive function. In a study by Maliszewska-Cyna et al. (2020), it was seen that sedentary TgCRND8 (another mouse model of AD) present lower capillary density in the cortex and higher capillary density in the hippocampus compared to non-transgenic controls. In AD brains, increased angiogenesis (formation of new blood vessels) has been observed in the hippocampus, midfrontal cortex, substantia nigra pars compacta, and locus coeruleus (Jefferies et al., 2017). Due to impaired cerebral blood flow, the state of hypoxia leads to compensatory angiogenesis through the upregulation of pro-angiogenic factors (e.g., VEGF). However, in AD brains, downstream molecules to VEGF destabilize the vessel wall of mature vessels and new vessels are not able to mature, leading to the formation of a leaky blood vessel network, which results in vascular remodeling and structural changes that compromise the integrity of the blood-brain barrier (BBB) (Jefferies et al., 2017). In the previously mentioned study by Maliszewska-Cyna et al. (2020), when the mice were exposed to 3 months of voluntary running, microvascular morphology in the affected regions were normalized and short-term spatial memory was significantly improved, despite no effects on Aβ pathology being observed. Also, in TgCRND8 mice, it was shown that the benefits of exercise extend to their offspring (Herring et al., 2012). Voluntary running during pregnancy significantly reduced Aβ plaque burden and amyloidogenic processing of APP, increased angiogenesis, improved neurovascular function, and reduced microgliosis, inflammation, and oxidative stress in transgenic offspring (Herring et al., 2012).

To date, few human studies have examined the relationship between exercise and CAA. A post-mortem analysis of 428 participants from the RUSH Memory and Aging Project reported no relationship between physical activity levels and CAA burden at autopsy (Buchman et al., 2018). However, physical activity was assessed over a relatively short duration (up to 10 days) and late in life, with 90% of this data collected within 36 months before death, potentially limiting the ability to detect long-term associations (Buchman et al., 2018). Additionally, it was found that the presence of macroinfarcts, nigral neuronal loss, and white matter pathologies are associated with lower levels of daily physical activity (Buchman et al., 2018). Of note, in the population studied, 72% were females, 23% presented the ApoE ε4 genotype, 39% had clinical dementia, and 1.06% had CAA diagnosed postmortem (Buchman et al., 2018).

In line with these findings, Robison et al. (2019) found that long-term access to a running wheel for 1, 3, or 12 h per day from 4 to 12 months of age failed to reduce CAA in mixed sex Tg-SwDI mice. However, anxiety-like behavior, increased sociability and short-term spatial memory, and decreased expression of pro-inflammatory cytokines, were observed following this exercise intervention (Robison et al., 2019). In contrast, a case study described a 55- years-old woman with CAA who experienced a temporal intracerebral hemorrhage likely during treadmill exercise (Malik et al., 2017), raising concerns that physical exertion might exacerbate cerebrovascular vulnerability in individuals with CAA. However, no detail about the level of CAA severity was provided. Together, these findings highlight the current uncertainty regarding the effects of exercise on CAA, underscoring the need for longitudinal studies to clarify whether exercise confers risk or benefit in this context.

Cognitive training has been widely studied for its potential to delay or mitigate cognitive decline and dementia, particularly in older adults. Some observational studies report that individuals who engage in lifelong cognitively stimulating activities have a lower risk of developing AD or other dementias (Schultz et al., 2015; Wu et al., 2023). Cognitive training involves structured tasks designed to improve specific cognitive domains such as memory, attention, executive function, and processing speed. Several randomized controlled trials (RCTs) (e.g., the ACTIVE trial) and meta-analyses have demonstrated that cognitive training can produce modest improvements in the trained domains that can be sustained for months to years post-training in older adults (Tennstedt and Unverzagt, 2013; Basak et al., 2020). However, RCTs in populations at risk for dementia (e.g., patients with mild cognitive impairment) have shown mixed results (Maffei et al., 2017; Orgeta et al., 2020). Cognitive training is hypothesized to protect against dementia by enhancing cognitive reserve and promoting neuroplasticity and increasing neural efficiency in trained brain regions (Park and Bischof, 2013).

Cognitive training has also been tested in animal models (Billings et al., 2007; Martinez-Coria et al., 2015; Anderson et al., 2017; Mehla et al., 2023). A study conducted in APP knock-in mice found that cognitive training at 3, 6, and 9 months of age improved cognitive and cholinergic functions and reduced Aβ load and microgliosis (Mehla et al., 2023). Similarly, a comparative study done in 3xTg-AD mice, spatially trained and tested every 3 months from 2 to 18 months of age, found training delays the decline in spatial memory and the development of AD pathology (Billings et al., 2007). However, a study by Anderson et al. (2017) found that Tg-SwDI and 5xFAD mice submitted to a high-intensity 4-months cognitive training presented little evidence of reduced Aβ pathology (including CAA) or cognitive improvement.

There are limited studies on the effects of combined interventions. A study by Robison et al. (2020) examined the effects of environmental enrichment (social and cognitive enrichment, as well as access to a running wheel for exercise) and its individual components from 4 to 8 months of age in female Tg-SwDI mice. All enrichment conditions resulted in an attenuation of vascular amyloid accumulation in the thalamus; however, cognitive-behavioral benefits varied by group, with the fully enriched environment exerting the widest range of benefits. These findings suggest that cognitive, social, and physical enrichment, particularly when combined, can be beneficial for targeting CAA and reducing the risk of cognitive decline (Robison et al., 2020).

Results from the landmark FINGER randomized controlled trial also support a multimodal interventional approach, combining cognitive training with exercise, dietary modification, social engagement, and management of vascular risk factors in at-risk older adults (Ngandu et al., 2015). However, this trial did not assess CAA specifically. In July 2025, a press release from the Alzheimer’s Association International Conference, provided updates on the U.S. Study to Protect Brain Health through Lifestyle Intervention to Reduce Risk (U.S. POINTER), a phase 3, five-site, 2-years, single-blinded randomized clinical trial of two lifestyle interventions in older adults at risk for dementia. U.S. POINTER assessed whether the results observed in the FINGER trial can be generalized to a larger, more diverse, population in the U.S. that are at risk for developing dementia and cognitive decline⁵. Participants were divided into self-guided and structured lifestyle intervention groups, which focused on increasing exercise, improving nutrition, cognitive and social challenge, and monitoring health⁶. According to the press release, participants in both groups displayed improved cognition, although those in the structured intervention group had greater improvements. Results were consistent across age, sex, ethnicity, heart health status, and ApoE ε4 genotype⁷.

Studies have reported that chronic stress and stress-related disorders increase the risk of cognitive decline and dementia (Peavy et al., 2012; Kim et al., 2023; Wallensten et al., 2023). Several studies in AD rodent models have demonstrated that stress exacerbates AD-related pathology and cognitive deficits (Carroll et al., 2011; Baglietto-Vargas et al., 2015; Stuart et al., 2017; Peterman et al., 2020; Shlomi-Loubaton et al., 2025); however, less direct evidence exists for its effects on CAA specifically. A study conducted by Huang et al. (2024) showed that chronic stress exacerbated CAA in Tg-SwDI mice, along with BBB injury and myelin loss. Chronic stress promoted vascular Aβ accumulation through neutrophil activation. Stress-induced norepinephrine enhanced STAT6 signaling, triggering NET formation and NETosis, contributing to vascular amyloid deposition. Inhibiting neutrophil recruitment or suppressing NETs formation reduced CAA severity (Huang et al., 2024). Another study using APPV717I-CT100 mice overexpressing human APP-CT100 containing the London mutation (V717I) found that chronic mobilization stress also increased vascular amyloid accumulation, accompanied by an acceleration of cognitive impairment (Jeong et al., 2006).

Sleep disorders and disturbances have been linked to poor brain health and more rapid cognitive decline (Atayde et al., 2020; Jackson et al., 2020). Sleep deprivation negatively affects attention, executive function, and short-term memory, and causes impairments in the glymphatic system, which facilitates waste removal (e.g., Aβ) that typically has enhanced activity during sleep (Durmer and Dinges, 2005; Gottesman et al., 2024).

Human and animal studies have further explored the connection between sleep disturbances and brain health. A brain autopsy of 315 patients from the Rush Memory and Aging Project showed that greater sleep fragmentation was associated with increased severity in arteriosclerosis and higher number of subcortical infarcts (Lim et al., 2016). In APP/PS1 mice with induced chronic sleep deprivation, it was shown that sleep deprivation exacerbated vascular lesions, with extensive Aβ deposits being observed, and increased expression of the lysine specific demethylase 6B (KDM6B) histone, which is associated with neuronal injury and inflammation (Yu et al., 2025). Furthermore, knockdown of the KDM6B histone ameliorated sleep deprivation-induced memory impairment, neuronal injury, and vascular lesions, and inhibited neuronal cytotoxicity of Aβ₄₂ (Yu et al., 2025).

Smoking is considered a modifiable risk factor for AD (Barnes and Yaffe, 2011; Daviglus et al., 2011; Etgen et al., 2011; Durazzo et al., 2014). It is strongly associated with oxidative stress in the brain, cleavage of Aβ via the amyloidogenic pathway, and abnormal tau phosphorylation (Durazzo et al., 2014); in addition to being a contributor to vascular dysfunction, long-term disability, and harming several organs (Archie et al., 2022). Cigarette smoking is known to impair nitrergic nerve function and nitric oxide synthesis in cerebral vascular endothelial cells, which interferes with blood flow and glucose metabolism in the brain, promoting Aβ synthesis (which further impairs blood flow), reducing intraneuronal clearance of Aβ, and potentially leading to impaired cognitive function and AD (Toda and Okamura, 2016). It has been shown that healthy, older individuals with a history of cigarette smoking, have significantly greater brain atrophy rate across several brain regions, in particular those associated with early-stage AD (Durazzo et al., 2012). Another study has reported that cigarette smoking and oxidative stress increase APP and Aβ expression in pulmonary artery smooth muscle cells, which can have implications in the association of lung function decline and cognitive decline (Karoor et al., 2022). Additionally, Aβ₄₀ and, especially, Aβ₄₂ were found to reduce the levels of contractile proteins SM22 and Calponin (Karoor et al., 2022).

Limited evidence is available regarding the relationship between smoking and CAA, specifically. A study analyzing data from 3 community-based cohort studies found that smokers were at increased risk for CAA (RR = 1.11) (Ma et al., 2025). A cohort study in participants with familial CAA (Dutch-type hereditary CAA) found no association between smoking and age at first ICH nor time of ICH recurrence (Voigt et al., 2024).

Alcohol is known for its acute effects on the brain, with temporary cognitive impairment and behavioral changes being commonly observed (Letenneur, 2004). Although sometimes classified as a risk factor for AD and other dementias, reported effects are conflicting and typically depend on the dose (Wiegmann et al., 2020). According to literature reviews, while light-to-moderate drinking (∼1–3 drinks/day) is associated with lower risk of all-cause and vascular dementia, high alcohol consumption (>14 drinks/week) is associated with increased risk of dementia, reduced brain volume, and brain damage (Letenneur, 2004; Wiegmann et al., 2020). For example, studies have shown that moderate wine consumption (<1–2 glasses/day) is associated with lower risk of AD (Orgogozo et al., 1997; Lemeshow et al., 1998). Meanwhile, a study that looked at 397 dementia cases over a mean follow-up of 23 years, found that dementia risk was higher in people who abstained from alcohol in midlife, as well as in those who consumed over 14 units of alcohol per week (Sabia et al., 2018). In contrast, data from the ALBION study showed that even light-to-moderate alcohol intake was associated with higher Aβ deposition, and Tau/Aβ and pTau/Aβ positivity in healthy individuals, compared to those who abstained from alcohol (Drouka et al., 2025). Several studies in rodent models of AD report that ethanol consumption exacerbates cognitive-behavioral deficits and amyloid pathology and disrupts metabolism and excitatory/inhibitory balance in the brain (Huang et al., 2018; Hoffman et al., 2019; Day et al., 2023). However, in line with human data, preclinical data also supports that there may be a dose-dependent relationship between alcohol consumption and AD/dementia-relevant pathology (Lundgaard et al., 2018; Kang et al., 2024).

Data on the relationship between alcohol use and CAA, specifically, are far rarer. A study analyzing data from 3 community-based cohort studies found no significant relationship between alcohol consumption and CAA (Ma et al., 2025). Additionally, a cohort study in participants with familial CAA (Dutch-type hereditary CAA) found no association between alcohol use and age at first ICH nor time of ICH recurrence (Voigt et al., 2024).

Due to studies having different definitions of what constitutes one drink; what differentiates light, from moderate, from heavy drinking; and many times not differentiating between gender; drawing conclusions about the effects of alcohol consumptions on dementia risk, or CAA more specifically, is complicated (Wiegmann et al., 2020).

Traumatic brain injury is a condition in which normal brain function is impaired by an external force. TBI increases the risk for cognitive decline and dementia and accelerates the age of onset (Shively et al., 2012; Li et al., 2016). It has been shown that patients with TBI may have accelerated onset of cognitive decline by 2+ years (Li et al., 2016) and that, although dementia risk decreases over time, it is still evident for over 30 years post trauma, especially in cases of severe TBI and of repeated TBIs. A retrospective cohort study conducted from 2005 to 2011 found that patients who were 55+ years old with moderate-to-severe TBI, or patients who were 65+ years old with mild TBI, were at increased risk of developing dementia (Gardner et al., 2014). TBI that occurs in early to midlife is associated with 2–4 times higher risk of dementia later in life, and even higher in the case of multiple TBIs (Shively et al., 2012; Ramalho and Castillo, 2015). Of note, long-term cognitive impairment following TBI episode(s) can still occur despite normal-looking MRIs and CT scans (Ramalho and Castillo, 2015).

Pathophysiological mechanisms that occur following TBI that may contribute to dementia risk are complex, and brain injuries are highly heterogeneous due to variation in force, rotation, number of injuries sustained, and other extenuating factors. Multiple case studies have reported the development of early-onset CAA in individuals <55 years of age with a history of moderate-to-severe TBI, presenting with ICH (Ehling et al., 2012; Purrucker et al., 2013; Nakayama et al., 2017; Oblak et al., 2022). However, a valid point made by Banerjee and Werring (2022) is that these cases of CAA following TBI may not be due to the injury itself, but rather result from neurosurgery performed to treat the TBI. These authors highlighted that neurosurgery, often involving cadaveric materials or spread by contaminated surgical instruments, can cause iatrogenic CAA, due to Aβ acting as a proteopathic seed.

Observational studies suggest that repeated mild TBI may also impact the development of CAA. A study of 357 contact athletes vs. a community-based cohort reported that contact sport participation was associated with more CAA (Standring et al., 2019). CAA was also associated with chronic traumatic encephalopathy (CTE), suggesting there may be some interaction between these neuropathologies (Standring et al., 2019). In a post-mortem study of retired male professional soccer players with dementia, moderate CAA was found in 4/6 and severe CAA in 1/6 cases (Ling et al., 2017). Taken together, these findings suggest that TBI may influence the initiation and/or progression of CAA. Experimental evidence is lacking in animal models of CAA; however, in AD models, TBI was shown to increase Aβ deposition, worsen cognition, delay glial activation, and enhance expression of inflammatory cytokines (Breunig et al., 2013; Kokiko-Cochran et al., 2016; Shishido et al., 2016; Zyśk et al., 2019).