CSF Aβ40 Levels Do Not Correlate with the Clinical Manifestations of Alzheimer’s Disease

Alzheimer’s disease (AD) is the most frequent cause of dementia, affecting around 50 million people worldwide, with a prevalence expected to triplicate by 2050 [1]. Even though the mechanism underlying AD is still poorly understood, the “amyloid cascade” continues to be the most widely accepted hypothesis. Thus, the amyloid precursor protein is sequentially cleaved in the amyloidogenic pathway to amyloid-β peptides of various lengths. The longest ones, Aβ₄₂ and Aβ₄₀, are prone to aggregation in hypothetical conditions of overproduction or reduced clearance [2, 3]. The quantification of these peptides in cerebrospinal fluid (CSF) is nowadays an essential tool as they are reliable biomarkers for AD, even providing the basis for the definition of the disease itself, according to the latest diagnostic criteria [4]. Decreased Aβ₄₂ in CSF has been previously correlated with higher amyloid deposits in pathological [5] and in vivo studies [6, 7], even predicting the progression to dementia from asymptomatic or mild cognitive impairment stages [8, 9]. However, no correlation has been found between CSF concentration of Aβ₄₂ and cognitive symptoms [6, 8‒11]. Less is known about Aβ₄₀, apart from that a decreased CSF Aβ₄₂/Aβ₄₀ ratio is a better indicator of amyloid deposits than Aβ₄₂ alone [12‒14]. Therefore, the role of Aβ₄₀ has been limited to adjusting for the interindividual variability in the total production of amyloid-β peptides [14]. With the present study, we aimed to evaluate the clinical and prognostic significance of the CSF levels of Aβ₄₀.

We retrospectively analyzed the electronic clinical records of the patients with suspected cognitive impairment evaluated at Hospital Universitario La Paz, Madrid, Spain, from April 2019 to May 2022. The inclusion criteria were (1) diagnosis of the AD continuum by the CSF biomarker profile according to the National Institute of Ageing-Alzheimer’s Association (NIA-AA) research framework [4]; (2) age above 18 years old; (3) a neuroimaging study showing no signs suggestive of a non-neurodegenerative etiology; (4) laboratory tests with folate and B12 within normal range, normal thyroid, renal, and hepatic function; (5) negative syphilis and VIH serologies; and (6) at least one evaluation at the Cognitive Disorders Clinic. Patients with a progression to the dementia stage according to the Global Deterioration Scale (GDS) [15] in less than 6 months after the beginning of the symptoms were excluded, as well as those presenting signs in either the neuroimaging study or laboratory tests indicative of a non-neurodegenerative etiology. The study was approved by the Ethics Committee at Hospital Universitario La Paz on April 28th, 2022 (PI-5239). All data were managed in fulfillment of the Declaration of Helsinki.

Every participant underwent a complete neurological examination on the first visit and a neuropsychological assessment with the Montreal Cognitive Assessment (MoCA) [16] or the Mini-Mental Status Examination (MMSE) [17] tests on the first visit and every 6 months while the follow-up lasted. Neurology consultants specialized in cognitive disorders performed neurological examinations and neuropsychological tests.

CSF samples were obtained by lumbar puncture and centrifuged for 10 min at 2,000 g and 20°C. The concentration of Aβ₄₀, Aβ₄₂, phosphorylated Tau (p-Tau), and total Tau were measured by chemiluminescence enzyme immunoassay on Lumipulse (Fujirebio^®). A diagnosis of the AD continuum was established if the Aβ₄₂/Aβ₄₀ ratio <0.068 pg/mL. The threshold for p-Tau positivity was 59 pg/mL. Participants were classified according to the concentration of Aβ₄₀ as follows: hyposecretors (Aβ₄₀ <7,755 pg/mL), normosecretors (Aβ₄₀ 7,755–16,715 pg/mL), or hypersecretors (Aβ₄₀ >16,715 pg/mL). The threshold values correspond to those provided by the manufacturer for diagnostic purposes [18, 19].

Demographic data including sex, age, history of hypertension, diabetes mellitus, and dyslipidemia, as well as a family history of dementia, were collected. To assess the presentation and progression of cognitive symptoms in the participants, we recorded the initial symptoms and estimated time of onset as perceived by the patient or caregivers, the clinical phenotype according to the International Working Group criteria [20], the score on MoCA test, and GDS stage in the first consultation and during the follow-up. Seven patients underwent MMSE at some point during the follow-up, so the results were translated to MoCA scores using conversion tables [21].

First, a raw analysis of the three subgroups established according to Aβ₄₀ levels was performed. We also completed a stratified analysis based on p-Tau levels. Differences in qualitative variables were assessed with the χ² test, and Fisher’s exact test was applied when a value was inferior to five. The Kolmogorov-Smirnov test was used to determine if the quantitative variables were normally distributed. Differences in quantitative values were assessed with Student’s t test or ANOVA when parametric analyses were deemed and with Mann-Whitney’s U test or Kruskal-Wallis’ test for nonparametric analyses. A Kaplan-Meier estimator was used to ascertain differences between groups in the progression to mild and moderate dementia stages (GDS ≥4 and GDS ≥5, respectively) [15], and changes in the MoCA score during the follow-up were standardized by dividing the difference by the interval between tests. Correlation among CSF biomarkers and the aforementioned clinical variables were assessed with Pearson’s and Spearman’s coefficients for normally and non-normally distributed variables, respectively. A logistic regression analysis was also performed to test the influence of CSF biomarkers on the phenotypical variants and progression of AD. SPSS 28 (IBM, Armonk, NY, USA) was used for the statistical analysis.

The inclusion criteria were met by 76 subjects, with a mean follow-up of 1.79 years (standard deviation [SD] = 1.01). The flowchart for patient selection is depicted in Figure 1. Of the 76 participants, 22 were classified as hyposecretors, 47 as normosecretors, and 7 as hypersecretors. Table 1 presents the concentration of CSF biomarkers for each subgroup. The demographic and clinical data for both raw and stratified analyses are shown in Table 2. No significant differences were found among groups regarding age, initial MoCA score, or initial GDS stage. The distribution of positive p-Tau status varied significantly across the subgroups, being more frequent in the normo- and hypersecretor categories (p = 0.003). No differences in demographic or clinical variables were found in the stratified analysis either. A positive correlation was found between Aβ₄₀ and p-Tau levels (ρ = 0.605, p < 0.001), Aβ₄₀, and total-Tau (ρ = 0.594, p < 0.001) and between Aβ₄₀ and Aβ₄₂ (ρ = 0.742, p < 0.001).

Biomarker concentrations (pg/mL) were not significantly different between the amnestic and non-amnestic subgroups: Aβ₄₀ 10,140.500, interquartile range (IQR) = 7,409 versus 9,489, IQR = 4,024 (p = 0.682); Aβ₄₂ 377, IQR = 224 versus 385.500, IQR = 292 (p = 0.284); Aβ₄₂/Aβ₄₀ 0.042, SD = 0.01 versus 0.046, SD = 0.01 (p = 0.063); p-Tau 111.842, SD = 70.72 versus 102.044, SD = 51.53 (p = 0.265); and total-Tau 695.230, SD = 391.02 versus 646.570, SD = 312.99 (p = 0.288). Amnestic syndrome was the most frequent IWG-2 phenotype, being found in 48 (63.2%) participants, followed by frontal variant in 21 (27.6%) participants. The distribution of AD phenotypes did not differ among subgroups both in the raw and in the stratified analyses nor did changes in the MoCA score (Table 3).

There were no significant differences among subgroups in the time of progression to dementia (Table 3; Fig. 2). No correlation was found between Aβ₄₀ levels and changes in the MoCA score (ρ = 0.185, p = 0.141), Aβ₄₀ and time of progression to mild dementia (ρ = 0.045, p = 0.700), or between that of biomarker and time of progression to moderate dementia (r = −0.003, p = 0.979).

In the logistic regression analysis, collinearity was found for total Tau and p-Tau (variance inflation factor = 12.688 and 12.506, respectively), so the former was removed from the model. Aβ₄₀, Aβ₄₂, p-Tau, and age were included as predictors. No significant association was found between the presence of a non-amnestic phenotype and Aβ₄₀ (p = 0.134), p-Tau (p = 0.175), or age (p = 0.051). Aβ₄₂ levels were positively associated with presenting a non-amnestic AD variant (p = 0.036, OR = 1.007, 95% CI = 1.000–1.013). The model showed a significant fit (χ² = 10.348, p = 0.035). Classification precision, specificity, and sensitivity were 72%, 89.6%, and 40.7%, respectively. Regarding progress to the mild dementia stage, no predictors showed significant associations: Aβ₄₀ (p = 0.439), Aβ₄₂ (p = 0.485), p-Tau (p = 0.387), or age (p = 0.632). No significant interactions were found for Aβ₄₀ (p = 0.672), Aβ₄₂ (p = 0.563), p-Tau (p = 0.744), or age (p = 0.839) and progress to moderate dementia.

In our study, the concentration of CSF Aβ₄₀ in AD patients did not correlate with the initial symptoms of the disease, the intensity of the cognitive decline as measured by the MoCA test, or the progression to the dementia stage. CSF biomarkers for AD have played a critical role in paving the road toward standardizing a biological definition of the disease and have homogenized its diagnostic and research criteria [20, 22]. The value of CSF biomarkers as a diagnostic tool for neurodegenerative diseases has been further stated with their proven ability to differentiate among parkinsonian syndromes [23]. The additional measurement of Aβ₄₀ has been demonstrated to reflect amyloid positron emission tomography status better than Aβ₄₂ alone [12] and to provide clinicians with a more accurate AD diagnosis [13] as it normalizes Aβ₄₂ concentration to total amyloid-β peptides. Studies on Aβ₄₀ as an independent biomarker and its possible relationship with the clinical or pathological aspects of AD have been otherwise scarce. In an anatomopathological series, no significant difference was found in its concentration between patients with AD and subjects with other neurodegenerative dementias [24]. Very recently, Verde et al. [25] reported preliminary data regarding the role of Aβ₄₀ and p-Tau levels in differentiating amnestic from atypical AD phenotypes [25], pointing toward possible slightly different biochemical and neuropathological changes underlying non-amnestic forms. However, our study did not replicate their results. We did observe a positive correlation between Aβ₄₀ and p-Tau, and between Aβ₄₀ and total-Tau, in concordance with previous observations [26], that reinforces the hypothesized modulation of Tau pathology by amyloid-β peptides. Although the exact link between these two pathological hallmarks remains to be elucidated, mouse models suggest that early accumulation of Aβ peptides induces a progressive conformation of Tau seeds into larger aggregates [27]. In prospective studies, CSF p-Tau and total-Tau have shown an increase in amyloid-positive individuals, with Aβ biomarkers somehow preceding Tau pathology but failing to show a steady rise themselves [11, 28]. This finding might suggest that higher levels of the former respond to disease progression and neuronal damage (Tau has proved to be a reliable indicator of disease progression and cognitive decline [29, 30]), though without a clinical correlation as measured by the tools employed in our study. Thus, amyloid pathology might be an initial and necessary step that subsequently drives disease progression in AD patients. However, as Aβ deposits precede by decades the emergence of symptoms [1], the latest stages of AD could become independent of amyloid pathology and be more related to unspecific mechanisms of neurodegeneration in which Tau pathology might play a leading role [31]. Further association with biomarkers for neurodegeneration such as neurofilament light protein [32], a set of data not available from our study population, would support this hypothesis. CSF Aβ₄₀ levels reliably reflect total Aβ load [26] and therefore show interindividual variability, but if AD progresses and Tau pathology becomes the dominant process, it is unlikely that they serve such function in normalizing p-Tau levels even though they were mechanistically related.

CSF biomarkers have been widely considered of particular interest in the search for a better understanding of the elusive mechanisms underlying AD as its interaction with the extracellular space of the encephalon would plausibly reflect the neuropathological changes occurring [33]. However, research on the clinical correlation of amyloid-β peptides has not yielded remarkable results. The concentration of CSF Aβ₄₂ has been inversely correlated with amyloid plaque load both in anatomopathological [5, 34] and amyloid-positron emission tomography imaging series [7, 35]. The remarkable stability of that peptide throughout the disease progression has prevented allocating Aβ₄₂ a defined pathological role [11, 28]. Nonetheless, soluble Aβ oligomers appear to be the toxic form of amyloid instead of plaques themselves [36], a fact that might account for the absence of a clinical connotation of Aβ₄₂ levels. In that sense, we found no apparent effect of Aβ₄₀ in the progression of cognitive decline or the clinical presentation of the disease either. The finding of a considerable proportion of cognitively normal subjects showing Aβ deposition has contributed to the reframing of the role of amyloid in AD pathogenesis [37]. The high frequency of concomitant pathology according to recent anatomopathological data further indicates a considerably higher complexity of the mechanisms involved [38]. Regarding the potential role of Aβ₄₂ as a predictor of a non-amnestic AD variant observed in our study, the OR value is not sufficient to be deemed clinically meaningful. Future studies should evaluate the predictive value of Aβ biomarkers for atypical variants as it was not the primary aim of our work.

Despite the retrospective design of our study and its subsequent limitations, the lack of a significant association between Aβ₄₀ and the phenotype or course of the disease but with a positive correlation with p-Tau and total-Tau requires consideration. It suggests a prominent role of Aβ₄₀ in neurodegeneration, though not a determinant of clinical manifestations or prognosis. An IWG-2 phenotype different from amnestic syndrome was reported for a third of the participants. Atypical variants of AD are estimated to represent around a third of the patients with an early onset of the disease, while they would only account for 6% of late-onset cases [39]. Although the measurement of CSF biomarkers has steadily become a standard diagnostic procedure for patients presenting to our Cognitive Disorders Clinic, the proportion of participants with non-amnestic AD phenotypes might be overrepresented in our sample as it is a helpful tool for an adequate distinction among the different neurodegenerative dementias in the case of an atypical presentation. Further studies with a prospective design are deemed to establish the role of Aβ₄₀ within the neurodegenerative process and how its interindividual variability might influence other constituents of AD.

In the present study, no significant differences were found in the clinical manifestations or progression of the disease in AD patients according to their Aβ₄₀ concentration. Aβ₄₀ positively correlated with p-Tau and total Tau concentrations, supporting their probable interaction in AD pathophysiology.

This study protocol was reviewed and approved by the Ethics Committee at Hospital Universitario La Paz on April 28th, 2022; approval number PI-5239. The study obtained a waiver of informed consent from the Ethics Committee at Hospital Universitario La Paz.

The authors have no conflicts of interest to declare.

No funding was received.

J.G.C. contributed to study design, manuscript writing, and data acquisition. H.M.S. and O.R.F. contributed to laboratory analysis and manuscript revision. M.H.B. and S.S.L. contributed to data acquisition and manuscript revision. A.F.G. contributed to manuscript revision and critical reading. A.M.M. contributed to study design and manuscript revision.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.