Blood–brain barrier permeability is associated with different neuroinflammatory profiles in Alzheimer's disease

Abstract Introduction Inflammation is an important player in Alzheimer's disease (AD), whose effects can be influenced by the blood–brain barrier (BBB). Here, we investigated the relationship between BBB permeability, indicated by cerebrospinal fluid (CSF)/plasma albumin quotient (Qalb), and CSF indexes of neuroinflammation in a cohort of biologically defined AD patients. Methods Fifty‐nine consecutive patients with mild cognitive impairment (MCI) or early AD (Mini‐Mental State Examination [MMSE] >22) underwent CSF analysis for inflammatory cytokines (interleukin [IL]‐1β, IL‐2, IL‐4, IL‐5, IL‐6, IL‐7, IL‐8, Il‐10, IL‐12, IL‐13, IL‐17, tumor necrosis factor‐α [TNF‐α], interferon‐γ [IFN‐γ], granulocyte‐monocyte colony‐stimulating factor [GM‐CSF], granulocyte colony‐stimulating factor [G‐CSF]). Using backward stepwise linear regression analysis, we explored the potential influence of each cytokine CSF level on Qalb considering age, sex, and apolipoprotein E (APOE) as covariates. Results Higher levels of IL‐4 (β = 0.356, 0.005) and IL‐8 (β = 0.249, 0.05) were associated with higher Qalb values, while macrophage inflammatory protein‐1α (MIP‐1β) (β = −0.274; p = 0.032) and TNF‐α (β = −0.248; p = 0.031) showed a significant negative association with BBB permeability. Age was also positively associated with Qalb (β = 0.283; p = 0.016). Conclusions Despite the overall integrity of the BBB, its permeability could either influence or be influenced by central neuroinflammation, reflected by CSF cytokine levels. This is in line with previous studies that showed that patients with a more intact barrier are those with more prominent neurodegeneration. Our findings suggest that different neuroinflammatory profiles can be associated with different levels of BBB permeability in AD.


INTRODUC TI ON
As the world population ages, dementia is increasingly becoming a global health burden.Among the different causes of cognitive impairment, Alzheimer's disease (AD) is the most prevalent, and great efforts in medical research are directed at finding biomarkers that could lead to an early diagnosis or hint at novel therapeutic targets.
From a pathophysiological point of view, AD is characterized by amyloidβ (Aβ) and phosphorylated tau (p-tau) accumulation in plaques and neurofibrillary tangles, respectively.These pathological changes result from the interplay of many different mechanisms and, among others, inflammation due to activation of the innate immune system has emerged as an important player, which has been linked to both beneficial and detrimental effects [1].
What are the factors skewing the effects of inflammation is still a matter of debate.Since the brain can no longer be considered an immune-privileged organ, much attention has been drawn to the blood-brain barrier (BBB) as a possible gateway regulating the passage of immune cells from the peripheral circulation into the central nervous system (CNS) [2].Even though the BBB undergoes physiological changes with aging, a leakier membrane seems to be a feature of AD, and the result of several factors challenging its integrity [3].Moreover, higher BBB permeability seems to correlate with more severe cognitive decline and a worse response to therapeutic attempts, despite no consensus having been reached on the mechanisms behind this finding [4].
With the present work, our aim was to investigate the relationship between BBB permeability, reflected by CSF/plasma albumin quotient (Qalb), and CSF inflammation, as measured by levels of inflammatory cytokines, in a cohort of biologically defined AD patients.

Subject enrolment
We enrolled 65 consecutive patients who had been referred to the UOSD Centro Demenze of "Policlinico Tor Vergata" in Rome due to memory decline, between April 2022 and December 2022.After clinical evaluation, they fulfilled criteria for either amnestic mild cognitive impairment (MCI) [5] or early dementia due to AD [6] with a Mini-Mental State Examination (MMSE) score >22.On CSF analysis, all patients belonged to the biologically defined AD continuum (i.e., CSF Aβ1-42 <600 pg/mL).
We excluded patients with recent ischemic stroke (within the last 6 months or radiological evidence of ischemic lesions), Hachinski score >4, CSF Aβ1-42 >600 pg/mL, chronic inflammatory/autoimmune conditions, signs and symptoms of ongoing infection, or finding elevated C reactive protein levels in blood.Therefore, our sample eventually included 59 patients.

Statistical analysis
All continuous variables including levels of CSF biomarkers and cytokines, age, and the calculated Qalb were expressed as means±standard deviations.
We used a backward multivariate regression analysis to explore associations between CSF cytokines levels and BBB permeability, considering age, sex, and apolipoprotein E (APOE) genotype (i.e., the presence of at least one ε4 allele) as covariates.Only cytokines with maximum values below the limit of detection of 5% were considered acceptable to be included in the analysis, thus we excluded IL-2, IL-5, IL-12, IL-13, IL-17, and IFNγ.Finally, we used Pearson's r analysis to assess correlations between cytokines associated with Qalb and AD core biomarkers.
All statistical analyses were performed with JASP © (Version 0.14 Computer Software; JASP Team 2020).Values of p < 0.05 were considered statistically significant.

RE SULTS
The study included 59 patients belonging to the AD continuum (A+T+, n = 29; A+T−, n = 30).Demographic variables and dosages of CSF cytokines levels are reported in Table 1.No association was found between TNF, IL-4, IL-8, and MIP-1β and AD core biomarkers (p > 0.05), except for a trend of positive correlation between MIP-1α and both t-tau (p = 0.058) and tau (0.088).

DISCUSS ION
Inflammation is known to be involved in the pathogenesis of AD, although its role remains unclear.It has been demonstrated that neuroinflammation can exert both detrimental and positive effects, on the one hand leading to the exacerbation of Aβ accumulation, but on the other reducing AD pathological burden [8].This dualism is further supported by different modalities of phenotypical activation of astrocytes and microglia that could produce either proinflammatory or protective cytokines, depending on disease stage or other complementary factors such as APOE [7] and BBB integrity [2].
Our study addressed the issue of the interplay between BBB permeability and inflammatory cytokine levels in the CSF.First, we found that, within our sample, Qalb mean levels were in range, meaning that the BBB was not disrupted.However, we identified that different degrees of permeability seem to impact neuroinflammation.Indeed, our analysis showed that a leakier BBB is associated with higher levels of IL-4 and IL-8, while a tighter BBB was associated with higher TNFα and MIP-1β levels.Moreover, while there was no statistically significant association between these cytokines and AD core biomarkers, we found a positive correlation trend between MIP-1α and both p-tau and t-tau CSF levels.
These results are open to different interpretations.Interestingly, both IL-4 and IL-8 have neuroprotective effects: IL-4 overexpression results in the attenuation of Aβ pathology in animal models [9], and IL-8 is a microglia-derived chemokine implicated in hypoxemicinduced neoangiogenesis [10], which could also cause alterations in BBB permeability.On the contrary, TNFα is a proinflammatory cytokine whose increased levels have been linked to the conversion from MCI to AD dementia [11] and MIP-1β promotes inflammation via chemotaxis [12].
A first hypothesis could see the BBB as an actual gatekeeper, regulating the passage of cytokines in and out of the CSF through changes in its permeability.However, in stages of the disease in which inflammation is skewed towards damage-repairing effects -reflected by higher levels of neuroprotective cytokines -a more permeable BBB could assist in the clearance of misfolded proteins.This is in line with previous studies correlating BBB breakdown to lower p-tau in the CSF [13], and with findings that soluble p-tau fully mediates amyloid-related tau aggregation in the brain [14].Thus, a tighter BBB could be associated with more intense neurodegeneration [13], likely sustained by the detrimental contribution of proinflammatory cytokines.The positive trend of correlation between MIP-1β and t-tau levels partially supports this hypothesis.However, the small sample size limited our statistical approach and did not allow us to define the causality of association between our variables.
Demographic variables and results of dosages of cerebrospinal fluid cytokine levels.Results from stepwise multivariate regression with backward entry model.