Cognitive and brain cytokine profile of non-demented individuals with cerebral amyloid-beta deposition

Mounting evidence from human and animal studies indicates that neuroinflammation plays an important role in the development and progression of Alzheimer’s disease (AD). Specifically, epidemiological studies revealed that cognitively healthy individuals under long-term administration of anti-inflammatories have a reduced risk of developing symptomatic AD [1, 2]. The benefit of early anti-inflammatory treatment suggests that a detrimental neuroinflammatory process is active at preclinical AD stages [1, 3, 4]. Indeed, studies in transgenic models of amyloid pathology have demonstrated the presence of a disease-aggravating neuroinflammatory process concomitant with the intraneuronal accumulation of amyloid-beta (Aβ) oligomers [5‐9]. While an incipient inflammatory process has been well characterized in AD animal models and has been associated with the abnormal deposition of Aβ, evidence from human studies remains scarce.

In the general population, it has been estimated that about one-third of older non-cognitively impaired individuals (NCI) display brain AD neuropathology [10‐13]. Given that aging is the major risk factor for AD and that AD has a long, pre-symptomatic phase where neuropathology starts to ‘silently’ build-up; it is then expected that a proportion of NCI individuals with AD neuropathology should be at the earliest, asymptomatic stages of AD.

Towards elucidating whether incipient AD pathology is associated with inflammatory marker expression and cognitive decline, we analyzed cognitive changes and pro-inflammatory cytokine expression in the temporal, parietal, and frontal cortices of a clinically and neuropathologically characterized cohort of non-demented elderly individuals stratified according to the presence or absence of cerebral Aβ.

Our results indicate that non-demented individuals harboring amyloid pathology display impairments in perceptual speed accompanied by an increase in key cerebral pro-inflammatory markers across different brain regions, strengthening the evidence for an early dysregulation of inflammatory markers at presumable preclinical stages of AD.

Demographic and neuropathological characteristics of elderly individuals aged 66 to 79 years with or without Aβ deposition are illustrated in Table 1 and Additional files 1, 2 and 3, Supplemental Tables 1-3. Representative images of Aβ and tau histopathology are illustrated in Fig. 1 and in Additional file 4, Supplemental Figure 1. No differences in global cognition, age, sex, years of education, tau pathology, and post-mortem interval were found between groups. The APOE e4 allele was enriched in cognitively normal individuals with global Aβ deposition (p = 0.044). Individuals with cerebral Aβ met a neuropathological diagnosis of AD according to the modified NIA-Reagan (p = 0.007) and CERAD (p = 0.0003) criteria. Cognitive decline and brain inflammatory protein expression were assessed in elderly individuals with no symptomatic manifestation of AD.

In this study, we show that non-demented individuals with cortical Aβ deposition display cognitive impairments in specific cognitive domains, such as perceptual speed, suggesting that such cognitive domain might be disrupted at early stages in the AD continuum. This is in line with previous studies reporting the occurrence of cognitive impairments in NCI individuals with Aβ pathology [17, 18]. Our studies also underscore that global cognitive scores might mask subtle cognitive domain-specific deficits. Furthermore, key inflammatory cytokines were differentially elevated across brain regions from elderly individuals with Aβ deposition (Fig. 7), reinforcing the notion of an early, preclinical neuroinflammatory process in the continuum of the AD pathology [1, 3, 4].

Notably, non-demented individuals with Aβ pathology had an overall increase in IL-1β in the three brain regions analyzed. Thus, raising the possibility that IL-1β might be one of the earliest inflammatory markers associated with the accumulation of Aβ neuropathology in older adults. IL-1β overexpression might contribute to potentiate AD neuropathology [19‐21]. Early involvement of IL-1β in the AD continuum is also suggested by its presence in diffuse amyloid plaques in AD brains [22] and increased expression of caspase-1, which promotes IL-1β maturation, at prodromal and clinical stages of AD [23]. Moreover, IL-1β upregulation has been reported before the establishment of an overt AD pathology in Down syndrome [24, 25]. Increased neuronal IL-1β expression is present in a transgenic rat model of amyloidosis before amyloid plaque deposition [5] .

IL-1β expression may lead to the translation of IL-6 [26]. In our study, IL-6 levels were likewise increased in the temporal cortex of non-demented individuals with Aβ pathology. Similar to IL-1β, IL-6 is also present in diffuse amyloid plaques in AD brains [27] and can enhance AD pathology and cognitive deficits in animal models [28‐30]. Importantly, genetic variations of the IL-6 and IL-1β genes have been associated with an increased risk of AD [31]. Of note, although IL-1β and IL-6 expression is often accompanied by increased TNF-α levels [32], we did not observe higher TNF-α levels in brains with Aβ deposition. In addition, the relatively low levels of TNF-α measured in this study, might mask subtle expression differences between brains with and without Aβ deposition. Therefore, more sensitive assays are needed to confirm whether TNF-α becomes dysregulated at early AD stages.

The chemokine eotaxin-3 was also elevated in the temporal cortex of AD-asymptomatic individuals with Aβ pathology. Increased eotaxin-3 levels have been reported in the CSF of individuals with AD dementia and can distinguish AD from other neurodegenerative diseases [33‐35]. Moreover, eotaxin-3 levels are elevated before the establishment of overt AD pathology in the brains of people with Down syndrome [25]. It has also been reported that eotaxin-3 levels correlate with CSF tau pathology in AD [33]. Interestingly, in our study, eotaxin-3 was only upregulated in the temporal cortex of NCI individuals with Aβ pathology, and such region displayed higher levels of tau pathology than the parietal and frontal cortices. Therefore, higher eotaxin-3 levels might reflect an immune response to pathological tau in elders with amyloid deposition.

Increased levels of MCP-1 were found in the parietal cortex of individuals with Aβ. MCP-1 is a potent monocyte chemoattractant, and its upregulation in transgenic models of AD was suggested to contribute to peripheral monocyte infiltration into the brain [36, 37]. MCP-1 upregulation has been reported in AD brains and body fluids [38, 39]. Notably, MCP-1 neuronal levels are upregulated before the development of amyloid plaques in a transgenic rat model of the AD-like pathology and in Aβ-burdened neurons in the human cerebral cortex [9], supporting its early involvement at incipient stages of amyloid accumulation.

A trend towards higher levels of IL-1β, IL-6, and MCP-1 levels was observed in the frontal cortex of Aβ+ individuals. This suggests that altered inflammatory protein expression in the frontal cortex might occur after its exacerbation in the temporal and parietal cortices. It is also possible that the small sample size posed a limitation to detect significant differences in this cortical region. Therefore, these analyses remain exploratory and await to be confirmed in a larger and independent dataset.

Our study also showed that although amyloid levels did not significantly differ among brain regions, the temporal cortex displayed higher tau deposition and inflammatory cytokine expression than the parietal and frontal cortices. Therefore, it might be possible that the combined tau and amyloid pathologies make the temporal cortex a more vulnerable region to increased inflammatory protein expression.

In accordance with our results, cognitively normal individuals with CSF Aβ pathological values displayed an upregulation of CSF inflammatory and cerebrovascular markers compared with NCI individuals with normal CSF Aβ values [40]. In addition, astroglial and microglial activation are already present before the onset of dementia in populations at genetic risk of AD [25, 41]. Importantly, AD-resilient brains display decreased microglia and astroglia activation and anti-inflammatory and pro-resolution cytokine signature [42, 43]. Moreover, “SuperAgers” brains also display lower levels of microglial activation, comparable to those found in young individuals [44]. Besides the expression of inflammatory cytokines, other pathological events have been reported to occur in the brains of cognitively normal individuals harboring AD neuropathology such as synaptic deficits [45], oxidative stress [45], growth factor dysregulation [46], and epigenetic modifications [47]. Further experiments are needed to understand whether these and the previously mentioned inflammatory changes are also present in brains with or without Aβ pathology.

In this study, none of the upregulated cytokines in the temporal, parietal, or frontal cortices was associated with the rate of cognitive decline. It is important to note that longitudinal cognitive measures and cytokine brain expression analyses are disconnected in the time of assessment. Therefore, this type of analyses would have more informative value if cytokine expression was measured in cerebrospinal fluid followed by post-mortem assessment of cytokine brain expression at the time of death.

Surprisingly, our results also show that some cytokines that display similar expression levels in brains with and without Aβ deposition, correlate either positively or negatively with the rate of cognitive decline in specific cognitive domains. While higher cytokine expression (IL-1α, IL-13, and MDC) in the temporal cortex was associated with worsening of cognitive function, the opposite was true for cytokine levels in the parietal (MIP-1β and TARC) and frontal (IL-16 and IP-10) cortices. These observations might reflect the transient nature of cytokine expression and its relationship with AD progression [48]. Moreover, the negative associations between cytokine expression in the temporal cortex and cognitive deficits might be associated with the increased tau pathology that was observed in this region. However, these complex exploratory observations warrant further comprehensive longitudinal investigations to elucidate whether cytokine expression exerts transient effects on cognition at different stages of the AD continuum.

This study has several limitations that must be considered. First, although post-mortem pathological analysis of AD-asymptomatic individuals indicated the presence of pathology meeting AD criteria, we cannot ascertain whether these individuals would have progressed through the AD continuum had they lived longer. However, the faster cognitive decline in elderly individuals with Aβ deposition, the higher proportion of APOE e4 individuals in this population, and the neuropathological diagnosis of AD according to modified NIA-Reagan and CERAD criteria, would suggest that these individuals might have indeed been at the earliest stages of AD. Second, to study preclinical stages of AD and to avoid confounding factors (such as the presence of Super-Agers), we conducted our investigations in NCI individuals under age 80. Such criteria limited the number of samples from the ROS and ROSMAP cohorts available for analysis. Although the groups in our study were sex-balanced, the restricted sample size was not sufficient to investigate the effects of sex on cytokine expression and cognition. Moreover, the narrow sample analyzed in this study could have limited the detection of subtle cognitive deficiencies in other cognitive domains. Ideally, this study should be confirmed in a larger sample size, when available. Finally, we have segregated our population by the extent of extracellular Aβ deposition. Besides being a highly selected sample, we acknowledge that there is not a clear “cutoff” value between what represents healthy and AD-related Aβ deposition. Moreover, to fully understand the earliest inflammatory changes associated with preclinical AD, it would be important to investigate the neuroinflammatory response associated with early intraneuronal Aβ accumulation. Equally important would be to measure soluble Aβ levels in brains with and without Aβ deposition to define the Aβ concentration threshold associated with cytokine dysregulation.

Our findings provide a link between amyloid deposition, key brain inflammatory cytokines, and cognitive decline in non-demented elderly individuals with amyloid accumulation. Our study also suggests that stratification by amyloid positivity should be considered when evaluating anti-inflammatory approaches for AD. The investigation of cognitively healthy individuals with AD neuropathology offers the possibility of identifying the earliest neuropathological changes associated with preclinical AD, allowing for the identification of novel therapeutic targets and biomarker candidates.