Association of APOE genotype and cerebrospinal fluid Aβ and tau biomarkers with cognitive and motor phenotype in amyotrophic lateral sclerosis

Abstract Objective Little is known about amyotrophic lateral sclerosis (ALS)‐nonspecific cognitive deficits – most notably memory disturbance – and their biological underpinnings. We investigated the associations of the Alzheimer's disease (AD) genetic risk factor APOE and cerebrospinal fluid (CSF) biomarkers Aβ and tau proteins with cognitive and motor phenotype in ALS. Methods APOE haplotype was determined in 281 ALS patients; for 105 of these, CSF levels of Aβ42, Aβ40, total tau (T‐tau), and phosphorylated tau (P‐tau181) were quantified by chemiluminescence enzyme immunoassay (CLEIA). The Edinburgh Cognitive and Behavioural ALS Screen (ECAS) was employed to evaluate the neuropsychological phenotype. Results APOE‐E4 allele was associated with worse ECAS memory score (median, 14.0 in carriers vs. 16.0 in non‐carriers) and lower CSF Aβ42 (−0.8 vs. 0.1, log‐transformed values) and Aβ42/40 ratio (−0.1 vs. 0.3). Some 37.1% of ALS patients showed low Aβ42 levels, possibly reflecting cerebral Aβ deposition. While lower Aβ42/40 correlated with lower memory score (β = 0.20), Aβ42 positively correlated with both ALS‐specific (β = 0.24) and ALS‐nonspecific (β = 0.24) scores. Although Aβ42/40 negatively correlated with T‐tau (β = −0.29) and P‐tau181 (β = −0.33), we found an unexpected positive association of Aβ42 and Aβ40 with both tau proteins. Regarding motor phenotype, lower levels of Aβ species were associated with lower motor neuron (LMN) signs (Aβ40: β = 0.34; Aβ42: β = 0.22). Conclusions APOE haplotype and CSF Aβ biomarkers are associated with cognitive deficits in ALS and particularly with memory impairment. This might partly reflect AD‐like pathophysiological processes, but additional ALS‐specific mechanisms could be involved.


INTRODUC TI ON
Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disorder characterized by loss of upper (UMNs) and lower motor neurons (LMNs) and leading to progressive paralysis of voluntary muscles [1].
Increasing evidence indicates that, beyond motor symptoms, almost 50% of ALS patients show neuropsychological impairment, mainly concerning language, verbal fluency, and executive functions [2].However, little is known about ALS-nonspecific cognitive deficits -particularly memory disturbance -and their biological underpinnings.Since memory impairment represents the distinctive feature of Alzheimer's disease (AD) [3], some studies have investigated biological AD hallmarks in relation to cognitive symptoms in ALS patients and the possible involvement of Aβ and tau proteins in ALS pathophysiology [4].Moreover, the acknowledged co-occurrence of proteinopathy in neurodegenerative disorders demonstrated the importance of mixed pathology as an underrated but key element to unveil complexity behind neurodegeneration [5].There is evidence for a potential role of amyloid precursor protein (APP) in the cellular response to axonal damage, with increased immunoreactivity of this protein in the perikarya of anterior horn cells suggesting an early protective effect [6,7].However, intracellular Aβ deposition might also be a late deleterious event leading to oxidative stress, activation of proapoptotic pathways [8], and TDP-43 accumulation [9].Concerning the other AD protein hallmarks, evaluation of the diagnostic potential of cerebrospinal fluid (CSF) total (T-tau) and phosphorylated tau (P-tau 181 ) has provided conflicting results [10,11]; nevertheless, T-tau might serve as prognostic biomarker reflecting the entity of motor neuron (MN) degeneration [12] similarly to serum levels of neurofilament light chain (NFL).Recent studies have reported increased plasma phosphorylated P-tau 181 levels in ALS patients which were proposed as a novel marker specific to LMN degeneration [13][14][15].Finally, a major risk factor for AD is represented by the E4 allele of the APOE gene.Whereas a pathogenic role of E4 in frontotemporal dementia (FTD) is still a matter of debate [16], with a recent article pointing out an unexpectedly increased risk in carriers of the APOE-E2 allele [17], inconsistent data are available for ALS [18].Indeed, while some evidence suggests a deleterious role of the E4 allele in ALS pathogenesis [19], other works failed to confirm an influence of APOE on clinical phenotype [20,21].
In this work, we investigated the potential association of APOE haplotype, as well as CSF Aβ and tau biomarkers, with motor and cognitive/behavioral features in ALS.Specifically, we explored the incidence of amyloid and tau pathology in a deep-phenotyped ALS cohort to estimate the occurrence of co-pathology and to verify whether AD-related mechanisms may be involved in ALS pathogenesis.

Study cohort
A cohort of 281 Italian inpatients, diagnosed with ALS according to the El Escorial revised criteria [22], was recruited at IRCCS Istituto Auxologico Italiano between 2014 and 2022.
The following demographic and clinical data were collected: sex; age at onset; disease duration; survival; family history of ALS; motor phenotype (classic, bulbar, respiratory, flail arm, flail leg, UMN-predominant [UMN-p], primary lateral sclerosis [PLS], progressive muscular atrophy [PMA]) [23]; revised ALS Functional Rating Scale (ALSFRS-R) scores at evaluation and disease progression rate (ΔFS) [24,25]; and presence of oculomotor abnormalities as previously described [26].Motor impairment was assessed in all patients using the following scoring systems: the Penn Upper Motor Neuron Score (PUMNS) to account for UMN regional involvement [27] and a modified version of the Lower Motor Neuron Score to account for LMN signs as previously described [28,29].Spinal LMN involvement was also measured using the Medical Research Council Conclusions: APOE haplotype and CSF Aβ biomarkers are associated with cognitive deficits in ALS and particularly with memory impairment.This might partly reflect AD-like pathophysiological processes, but additional ALS-specific mechanisms could be involved.

K E Y W O R D S
Alzheimer's disease, amyloid beta, amyotrophic lateral sclerosis, APOE genotype, tau proteins | 3 of 13 APOE GENOTYPE AND CSF AΒ AND TAU IN ALS version was used to perform an extensive evaluation of both cognitive and behavioral profile of the study population [30].Behavioral symptoms were further investigated with the Frontal Behavior Inventoy (FBI) [31].Detailed descriptions of neuropsychological scores are provided in Supplementary Methods.

APOE haplotype analysis
APOE haplotype was determined by imputing rs7412 and rs429358 from previously generated genotyping data [32] or by direct sequencing of APOE exon 4. A full description of the methodology is reported in the Supplementary Methods.
The cohort was subdivided into two groups according to the presence of at least one E4 allele.Considering the putative protective role of the E2 allele against AD and the indetermined significance of the E2|E4 genotype, patients carrying this specific genotype were excluded from analyses [33].defined by P-tau 181 and T-tau levels ≥56.5 pg/mL and ≥ 404 pg/mL, respectively [34].NFL measurement was performed on the Simoa SR-X platform (Quanterix, Lexington, MA, USA) as previously described [35].

Statistical analysis
Analyses were performed with statistical software R version 4.1.1.
Descriptive statistics were reported as means and standard deviations for quantitative variables or frequencies (%) for categorical ones.Paired sample t-test was used to compare demographic and clinical features of the whole cohort versus the CSF subcohort.Log transformation was applied to all CSF and serum biomarkers values to obtain a normal distribution, and derived measures were used in the regression analyses.Dependent and independent variables were standardized prior to regression analyses to achieve standardized beta values.Linear regression was employed for modeling the association of CSF biomarkers with motor and cognitive variables of interest, indexes of disease progression, as well as serum levels of NFL, and to investigate differences in the distribution of these variables in APOE-E4 carriers and non-carriers.Accordingly, separate comparisons between patients stratified according to amyloid and tau status were also performed.Binary logistic regression was used to assess if CSF biomarkers predicted presence or absence of cognitive impairment in different ECAS subdomains.Chi squared test was employed to compare the distribution of APOE genotypes in cognitively impaired and unimpaired patients.Age at evaluation was introduced as a covariate when appropriate.A sensitivity analysis was performed to assess if the association of APOE status with ECAS scores retained significance after covariation for variables known to be associated with cognitive impairment (gender, ΔFS, region of onset, and C9orf72 expansion).Survival analysis was performed with Kaplan-Meier curves and log-rank test was used to compare survival curves across groups.Values of p < 0.05 were considered statistically significant.
Conversely, the E4 allele was not associated with ALS-specific cognitive domains, nor with behavioral symptoms assessed using both ECAS and FBI.Finally, no differences were observed regarding age at onset, site of onset, survival, and motor features.

Association of CSF biomarkers with motor features
Conversely, no association of T-tau and P-tau 181 with motor phenotype and burden of UMN and LMN signs was observed, nor were CSF biomarkers associated with survival.When analyzing relationships between APOE and CSF biomarkers, we found associations partly explainable according to an AD-like Although a major ALS pathomechanism is represented by accumulation of TDP-43 protein within MNs [37], other, less investigated biological processes might be at play.Indeed, neuropathological studies identified Aβ pathology in up to one-half of autopsied ALS cases [38,39].Although this could be partly explained by the frequent occurrence of Aβ pathology in the elderly, in our cohort we found an unexpected prevalence of patients with low Aβ 42 levels, namely twice as high as that reported for cognitively unimpaired individuals of similar age by a meta-analysis [40], thus possibly reflecting increased cortical Aβ burden of potential pathogenic relevance.

DISCUSS ION
Interestingly, lower Aβ 42/40 was associated with worse memory by an AD co-pathology that could be more frequent than generally considered [39,41].
However, APOE could also contribute to cognitive abnormalities in ALS via different pathways, as suggested by the association between the E4 allele, TDP-43 pathology, and hippocampal sclerosis reported in a large neuropathological study [42].
Furthermore, a comprehensive analysis of our findings suggests that strictly applying the same biological paradigms used in AD to ALS might not be totally appropriate, with the risk of overlooking relevant biological clues.Indeed, contrary to AD, in our ALS of axonal damage across different neurological conditions, including traumatic brain injury [44].However, it is unclear whether the putative increase in APP following neuronal damage can be explained by impairment of cell structures involved in APP metabolism such as the Golgi apparatus, which is disrupted in ALS MNs [45], or rather represents a protective mechanism against glutamate excitotoxicity or proteasomal stress [46].Regardless, the finding that lower Aβ 42 and Aβ 40 levels correlated with LMN involvement is consistent with the hypothesis that APP and its fragments might be part of a protective system whose deficiency accelerates disease progression [7] 6); and (5) CSF T-tau levels may represent a biomarker of disease progression in ALS.
The collateral finding that higher levels of tau proteins are associated with higher verbal fluency scores remains unclear.However, considering that the strongest association with fluency score was observed for Aβ 40 levels and not for tau proteins, we cannot exclude that Aβ 40 increase following early neuronal damage may elicit a protective response which initially succeeds in preserving cognitive functions before being overwhelmed by the disease.
Our study has some limitations.First, data were derived from a referral centre which is more susceptible to biases, such as longer median survival, compared with a registry population.Secondly, ECAS is tailored to assess primarily ALS-specific cognitive deficits, and may not represent the ideal tool to assess the presence of subclinical memory deficit or other cognitive features typical of AD.Moreover, although the measurement of CSF AD biomarkers complements the characterization of cognitive and motor symptoms in our ALS cohort, it does not allow a thorough exploration of the biological pathways through which Aβ metabolism might contribute to ALS pathophysiology.In particular, it is not possible to establish whether low CSF levels of Aβ 42 are associated with extracellular amyloid plaque deposition as described in AD or with intracellular Aβ accumulation as suggested by a previous neuropathological study [8].Moreover, sAPPα and sAPPβ were not investigated.The lack of neuroimaging data prevented us from investigating whether amyloid and tau pathology were associated with atrophy of brain regions usually involved in AD, namely hippocampus and precuneus.
Finally, despite the high concordance between CSF Aβ 42/40 levels and amyloid-positron emission tomography (PET) markers [48,49], amyloid-PET derived data would be useful to have an additional confirmation of our findings and to further investigate this topic also in relation to spatial and regional distribution.
Our findings suggest that CSF levels of Aβ and tau proteins might be associated with cognitive and motor features of ALS patients.
Moreover, albeit preliminary, our results indicate that Aβ species might play a more important role in ALS pathogenesis than previously thought.Further studies, using PET-derived data or directly neuropathological findings, are thus warranted to investigate the biological significance of these proteins in the pathological processes leading to ALS-related neurodegeneration.
scores, while Aβ 42 inversely correlated with scores in both ALSspecific and -nonspecific domains.Furthermore, the presence of at least one APOE-E4 allele was associated with lower values of both Aβ 42/40 and Aβ 42 and with more severe cognitive impairment in ALSnonspecific domains, particularly memory, supporting the role of APOE as major genetic determinant of cognitive impairment through Aβ-dependent mechanisms.Interestingly, association of APOE haplotypes with ALS-nonspecific domain of ECAS remained significant after covariation for variables influencing the cognitive profile.These findings are consistent with dynamics partly recapitulating AD pathophysiological processes, with Aβ deposition triggering tau accumulation as indicated by the inverse association between lower Aβ 42/40 and tau protein levels.Therefore, Aβ pathology might contribute to cognitive impairment in ALS through AD-like mechanisms.Morevoer, it cannot be completely ruled out that the observed effect of amyloid species on cognitive impairment in ALS could be driven F I G U R E 4 Simple dispersion with adjustment curve illustrating correlations of Aβ 40 and Aβ 42 with clinical indexes of lower motor neuron (LMN) impairment, namely LMN (a-c) and Medical Research Council (MRC) scores (b-d), respectively.
cohort we found a positive correlation between CSF levels of Aβ and tau proteins.This only applied to single Aβ species and not to the Aβ 42/40 ratio.Therefore, it could be hypothesized that during the disease process, neuronal damage -reflected by increased T-tau, which in turn is associated with the index of disease progression ΔFS as well as serum levels of NFL -leads to an indirect increase of APP, and subsequently of Aβ 42 and Aβ 40 species, as a result of impaired axoplasmic transport or reactively enhanced APP synthesis [6, 43].In fact, APP has been reported as a marker F I G U R E 5 Simple dispersion with adjustment curve illustrating significant positive correlation between cerebrospinal fluid (CSF) T-tau levels and disease progression rate (ΔFS).F I G U R E 6 Illustration of proposed biological interplay between classic amyotrophic lateral sclerosis (ALS)-related mechanisms of damage and Aβ pathways.Neuronal damage caused by ALS (1) resulting in TDP-43 accumulation (2) is associated with progressive increase of T-tau protein (3).These cellular changes lead to an indirect increase of amyloid precursor protein (APP) (protein dysmetabolism or reactive increase) (4).The APP increase may subsequently cause intracellular amyloid accumulation or extracellular amyloid plaque deposition (5), circularly triggering further neuronal damage(6).Image created with BioRender.Adopted from "Structural overview of an animal cell template" by BioRe nder.com (2022).Retrieved from https:// app.biore nder.com/ biore nder-templ ates.