Effect of Vascular Risk Factors on Blood-Brain Barrier and Cerebrospinal Fluid Biomarkers Along the Alzheimer’s Disease Continuum: A Retrospective Observational Study

INTRODUCTION

The blood-brain barrier (BBB) is an integral part of the Neurovascular Unit, which is a structure of the central nervous system constituted by vascular endothelial cells, basement membrane, pericytes, astrocytes, microglia, and neurons [1]. Under physiological conditions, the BBB maintains a regulated passage of metabolites and molecules between the systemic circulation and the cerebral parenchyma. This regulation ensures a suitable brain environment for the proper functioning and survival of neurons and glial cells. Indeed, there is growing evidence that BBB alterations might support neurode generation in Alzheimer’s disease (AD), favoring neuro inflammation and the accumulation of pathological proteins [2, 3]. Neuropathological data reported that amyloid deposition around cerebral vessels is a primary mechanism responsible for vascular-derived brain damage and BBB impairment in AD [4, 5]. Furthermore, the presence of vascular alterations is a well-recognized early characteristic of AD [6] and an association between vascular reactivity and amyloid pathology has also been recently observed [7]. Alterations in BBB permeability could be partially related to the detrimental effects of amyloid-β (Aβ) peptides on pericytes [8, 9]. Moreover, several biochemical processes, e.g., endothelin release, increased synthesis of vascular growth factors, oxidative stress, and disregulation of nitric oxide synthase enzyme, concur to BBB breakdown and to neurodegeneration in AD [2, 10].

In addition, individuals carrying the Apo lipoprotein E (APOE) ɛ4 allele, the main risk factor associated with late-onset AD [11, 12], exhibit accelerated loss of cerebrovascular integrity, thinning of the micro vascular basement membrane, enhanced pericytes degeneration, and BBB alterations [13–15]. Interestingly, along with the influence of vascular risk factors (VRF), such as hypertension, hypercholesterolemia, hypertriglyceridemia, and diabetes mellitus (DM), the aging process is suggested to affect BBB integrity, especially in the later stages of AD [10]. However, it is important to note that amyloid-related changes are supposed to begin 15–20 years before cognitive symptoms appear [16, 17], in a time window when individuals are typically less exposed to common VRF. Therefore, it would be useful to investigate whether BBB dysfunction begins in the preclinical stage of the disease, closely related to the occurrence of amyloid deposition, or later, during the clinical phase of AD, when VRF exert their detrimental effects. To assess BBB integrity in vivo, the cerebrospinal fluid (CSF)/plasma albumin ratio (QAlb) is commonly used [18]. As albumin is exclusively synthesized in the liver, its concentration in the CSF derives only from the blood, transported by passive diffusion across the BBB. Consequently, QAlb is considered a valuable surrogate index of BBB permeability. Despite the amount of scientific research on this topic, the findings on BBB dysfunction in AD are not clear and often conflicting [19, 20]. It is important to note that actual impairment of the BBB appears to be limited to a small number of patients [21–23]. Given this context, the aim of the present study was to investigate BBB alterations in the AD continuum (ADc) using the Amyloid (A) and p-tau (T) status, according to the NIA-AA research framework [24]. This classification allows researchers to evaluate AD pathology from a biological perspective while excluding other potential diagnoses, such as vascular or mixed dementia. Additionally, we examined the relationship between BBB permeability and CSF AD biomarkers, taking into account APOE genotype and the presence of coexisting VRF.

MATERIALS AND METHODS

Patients’ recruitment

In this retrospective observational study, we analyzed a cohort of 308 patients who attended the Centro Demenze of the “Tor Vergata” Hospital of Rome, between 2016 and 2020. All patients underwent a complete clinical investigation for diagnostic purposes, including medical history, neurological examination, neuropsychological assessment, and a complete blood screening with APOE genotyping. Patients were classified as APOE4 if having one or two copies of the ɛ4 allele (ɛ3/ɛ4 and ɛ4/ɛ4), while APOE3 if homozygous for the ɛ3 allele (ɛ3/ɛ3). Patients also underwent magnetic resonance imaging or fluorodeoxygluose positron emission tomography and lumbar puncture for CSF analysis. 211 patients fulfilled the clinical criteria for mild cognitive impairment or mild AD [25, 26]. Exclusion criteria were: 1) co-presence of other neurodegenerative diseases2) clinically manifest acute stroke in the previous 6 months showing a Hachinsky scale score > 4, 3) recent head trauma,4) neuro-oncological or neurosurgical history, 5) inflammatory/infectious disease of the central nervous system. Eventually, 131 patients were selected for this study (Fig. 1 for the Flow Chart).

OPEN IN VIEWER

Fig. 1

Patients’ selection flowchart summarizing the enrollment procedures.

The control group consisted of 24 healthy subjects who were matched for age and sex. These individuals were assessed for headache at the Tor Vergata Hospital Emergency Department between November 2016 and August 2020. The CSF samples were collected in accordance with standard hospital practice. The control subjects did not carry a diagnosis of primary neurological disorder other than headache, were free of cognitive complains or psychiatric illnesses and did not show any of the exclusion criteria aforementioned for AD patients. The presence of VRF such as hypertension, hypercholesterolemia, hypertriglyceridemia, and diabetes mellitus was recorded for patients and controls.

This study was approved by the ethics committee of Tor Vergata Hospital and performed according to the ethical principles of the Declaration of Helsinki.

RESULTS

BBB, APOE genotype, and CSF AD biomarkers

To examine BBB permeability in ADc patients based on APOE genotype, we conducted an ANCOVA test, using age and sex as covariates, comparing APOE4 and APOE3 groups. Our analysis did not reveal any significant difference in QAlb levels between the two groups (p = 0.990). Furthermore, there was no significant difference observed in the frequency of pathological QAlb levels between the APOE4 and APOE3 groups (χ² = 0.02, p = 0.889). To further explore the relationship between APOE genotype, AT status and BBB permeability, we performed an ANCOVA test, still using age and sex as covariates, stratifying patients based on both APOE genotype and AT status. Our analysis revealed a significant difference in QAlb levels between the A + T+ APOE4 and A + T–APOE4 subgroups (post-hoc analysis, Bonferroni corrected, p = 0.014). However, no significant differences were observed in the other comparisons (Fig. 2). Examining the frequency of pathological BBB, we found that the A + T+ APOE4 subgroup had the lowest percentage (3.1%), while the A + T–APOE4 the highest (25%). Although the difference in percentage between these two groups was found to be statistically significant (χ² = 6.335, p = 0.012), the comparison among all subgroups and controls did not highlight a significance difference (χ² = 7.288, p = 0.121). (Complete statistical values can be found in Table 1).

OPEN IN VIEWER

Fig. 2

Qalb levels among controls and patients groups according to APOE genotype and AT status, ^*p < 0.05.

OPEN IN VIEWER

Table 2

Multiple linear regression analysis in APOE4 and APOE3 patients

	APOE4		APOE3
Qalb	β (std)	p	β (std)	p
Sex	–0.324	0.004 *	–0.271	0.026 *
Hypertension	0.229	0.038 *	0.039	0.750
Hypercholesterolemia	–0.230	0.038 *	–0.129	0.289
CSF p-tau	–0.341	0.002 *	–0.080	0.513
	R2 = 0.272		R2 = 0.059

CSF, cerebrospinal fluid; APOE4, patients carrying ɛ4/ɛ4 or ɛ3/ɛ4 genotypes; APOE3, patients carrying ɛ3/ɛ3.

DISCUSSION

To our knowledge, very few studies focused on the evaluation of BBB integrity along AD continuum categories [28, 29]. In this study, we aimed to investigate BBB permeability in ADc patients, stratified by the AT status, as defined by the AT (N) research framework [24] and APOE genotype.

Initially, we compared QAlb values of the entire group of ADc patients with controls. Our results confirm previous findings that BBB breakdown, indicated by QAlb > 9, is not evident in the early stages of AD [19]. However, it should be noted that several other studies have reported a relative increase of BBB permeability in AD patients compared to age-matched healthy controls [20, 23], leading to ongoing debates and inconclusive results. These conflicting findings between studies may be attributed to several factors. First, there can be clinical overlap between different dementia subtypes, particularly in studies lacking specific AD-related biomarkers. Additionally, the methodologies used to assess BBB integrity vary among studies, which can contribute to differences in results [30]. Furthermore, the stage of the disease at which BBB permeability is evaluated and the age of disease onset (early-onset versus late-onset AD) could also account for variability in results [22]. Indeed, it has been suggested that BBB alterations in preclinical stages of AD may contribute to disease development, promoting the accumulation of amyloid aggregates [31, 32]. In the present study, we found that BBB is not impaired in early clinical phases of AD and it is conceivable to suppose that the loss of BBB selectivity occurs in later stages, coinciding with the presence of aging effects and VRF, both favoring vascular dysfunction and neuroinflammation [33]. In our study, when stratifying AD patients according to the AT classification, we found significantly higher QAlb values in the A + T–than in the A + T+ subgroup. This finding is consistent with a recent CSF proteomic study by Tijms et al., which also identified a cluster of AD patients characterized by BBB dysfunction and normal levels of p-tau [28]. The A + T–subgroup, classified as “AD pathological changes”, but not as full-blown AD, showed nonetheless neuropsychological deficits and cognitive decline [24]. It is noteworthy that a recent neuropathological study demonstrated that the majority of A + T–patients, despite having low p-tau levels in the CSF, exhibited autopsy-confirmed AD pathology, showing no significant differences in amyloid deposition and fibrillary tangles compared with A + T+ patients [34]. Interestingly, our results showed an inverse correlation between QAlb and p-tau levels in ADc patients. As it has been demonstrated, highly permeable BBB could facilitate the drainage of tau peptides [35], then our data led us to speculate that the higher BBB permeability in A + T–patients might have contributed to their low p-tau levels in the CSF.

Considerable literature supports the central role of the APOE genotype in determining the efficiency of BBB permeability, again with conflicting results. Some pathological studies reported BBB breakdown in AD brain samples of patients expressing APOE ɛ4 allele [13, 15], while several in vivo studies did not show significant differences in BBB permeability depending on APOE genotypes [36, 37]. In line with these findings, our analysis did not reveal a significant difference in QAlb values between APOE4 and APOE3 patients. However, when we further stratified the APOE groups according to the AT status, we observed that A + T–APOE4 patients exhibited higher QAlb values, significantly different respect to the A + T+ APOE4 patients, which, in contrast, showed lower Qalb values. These results are intriguing and provide support for the existence of a distinct pathophysiological pattern in APOE4 carriers, depending on CSF p-tau levels. Indeed, A + T–APOE4 patients have been associated with peculiar elevated neuroinflammation [38], preserved synaptic plasticity [39] and, as demonstrated in the present study, high BBB permeability. Furthermore, APOE4 genotype has been associated not only with worse neurodegeneration in tauopathies [40], but also with the coexistence of different proteinopathies [41]. The most frequently pathological depositions associated with amyloid pathology, beyond tau aggregates, are α-synuclein and TAR DNA-binding Protein 43 (TDP-43) [42]. In particular, synucleinopathies showed higher BBB permeability compared with healthy subjects and other neurodegenerative diseases [43, 44]. Thus, it is useful not to exclude that the accumulation of other pathological proteins (i.e., α-synuclein) could have influenced the BBB permeability in our cohort, especially in A + T–APOE4 patients. Despite this, it is important to note that the patients recruited in our study showed a clinical profile concordant with a cognitive impairment due to AD [25, 26], without neurological signs referable to other neurodegenerative diseases. Unfortunately, to date there are no sensitive biomarkers for the identification of the concomitant presence of other proteinopathies.

Finally, we conducted multivariate regression analyses separately in the APOE4 and APOE3 patient groups, to explore the potential impact of BBB permeability on CSF AD biomarkers, considering also several factors that could influence BBB integrity, such as clinical features, demographic characteristics and VRF [10]. Our results revealed an inverse association between p-tau levels and QAlb in APOE4 patients, suggesting that, increased BBB permeability is associated with lower levels of p-tau in the CSF. This finding indicates a potential link between BBB dysfunction and the clearance of tau peptides. Interestingly, this association was not observed in APOE3 patients, suggesting a possible different relationship between BBB permeability and p-tau levels based on APOE genotype. Furthermore, in APOE4 patients, QAlb showed associations with sex, hypertension, and hypercholesterolemia. The association between sex and QAlb, with higher values observed in men compared to women, is consistent with previous reports [45, 46]. The exact reason for this association is not fully understood, but it may be related to hormonal differences, with estrogens playing a role in maintaining BBB integrity [45]. Notably, hypertension and hypercholesterolemia were found to be associated with QAlb specifically in APOE4 carriers. Experimental evidence suggests that these VRF can contribute to BBB disruption and cognitive deficits [10, 48]. Chronic hypertension, in particular, is a well-established risk factor for dementia, leading to macro- and micro infarcts that could impair BBB function and contribute to cognitive decline. Diabetes mellitus, particularly type II, which is known to have detrimental effects on the endothelium through hyperglycemia, did not show associations with BBB changes in our cohort of AD patients, not even among APOE4 carriers who typically exhibit more pronounced neurodegeneration in the presence of diabetes [49, 50].

Overall, these findings highlight the potential interaction between APOE genotype, VRF, and BBB integrity in AD. A clear-cut BBB breakdown was not retrievable, even in the presence of VRF and cognitive decline, at least in the early stages of the disease. However, APOE4 carriers might be more susceptible to the detrimental effects played by hypertension and hypercholesterolemia on the BBB and possibly cognitive outcomes. Further research is needed to elucidate the underlying mechanisms and clinical implications of these associations.