Alzheimer's disease (AD) is the most common cause of dementia and is characterized by aggregation of amyloid and tau proteins in the brain. Results from genetic studies suggest that the pathophysiology underlying AD is complex, but studying this complexity in patients remains difficult. The cerebrospinal fluid (CSF) proteome contains a large number of proteins that can reflect ongoing biological processes. Proteomics techniques can be used to measure many proteins simultaneously in individual patients and may therefore provide an opportunity to study AD disease mechanisms. Here, we review the CSF proteomics literature to identify proteins consistently associated with AD, and perform pathway analyses on these proteins to study which biological processes may be involved in the disease. We performed a literature search of studies that investigated CSF proteomic alterations related to AD. We included original research articles when they measured at least 10 proteins in (antemortem) CSF in at least 10 individuals with AD, mild cognitive impairment (MCI) or controls. We examined if proteins were consistently related to AD, defined as consistent increase or decrease in AD vs. controls across studies. Next, we used the proteins identified as input to pathway analyses using Reactome to investigate which biological processes were enriched. In total, 29 studies were included that investigated AD-related changes to the CSF proteome, including a total of 1434 individuals with AD (of whom 47.1% had a CSF biomarker profile and 9.6% a postmortem examination consistent with AD) and 1380 controls. The studies reported 1 to 138 proteins associated with AD, of which 97 proteins were reported by two or more studies. Among proteins that were measured in more than one study, 27 (27.8%) showed consistent increases, 15 (15.5%) consistent decreases and 55 (56.7%) had contrasting results. Pathway analyses showed that AD-related proteins were enriched for hemostasis, lipoprotein and extracellular matrix pathways. These results indicate that proteomic alterations in CSF associated with AD reflect involvement of various biological pathways. The frequent occurrence of inconsistent protein level changes reported by different studies suggests that additional biological and/or (pre)analytical factors may influence the CSF proteome in AD, which should be further investigated in order to improve understanding of the biological complexity underlying AD.

Background As Alzheimer’s disease (AD) pathology presents decades before dementia manifests, unbiased biomarker cut-points may more closely reflect presence of pathology than clinically defined cut-points. Currently, unbiased cerebrospinal fluid (CSF) tau cut-points are lacking. Methods We investigated CSF t-tau and p-tau cut-points across the clinical spectrum using Gaussian mixture modelling, in two independent cohorts (Amsterdam Dementia Cohort and ADNI). Results Individuals with normal cognition (NC) (total n = 1111), mild cognitive impairment (MCI) (total n = 1213) and Alzheimer’s disease dementia (AD) (total n = 1524) were included. In both cohorts, four CSF t- and p-tau distributions and three corresponding cut-points were identified. Increasingly high tau subgroups were characterized by steeper MMSE decline and higher progression risk to AD (cohort/platform-dependent HR, t-tau 1.9–21.3; p-tau 2.2–9.5). Limitations The number of subjects in some subgroups and subanalyses was small, especially in the highest tau subgroup and in tau PET analyses. Conclusions In two independent cohorts, t-tau and p-tau levels showed four subgroups. Increasingly high tau subgroups were associated with faster clinical decline, suggesting our approach may aid in more precise prognoses.

Introduction It is important to understand which biological processes change with aging, and how such changes are associated with increased Alzheimer's disease (AD) risk. We studied how cerebrospinal fluid (CSF) proteomics changed with age and tested if associations depended on amyloid status, sex, and apolipoprotein E Ɛ4 genotype. Methods We included 277 cognitively intact individuals aged 46 to 89 years from Alzheimer's Disease Neuroimaging Initiative, European Medical Information Framework for Alzheimer's Disease Multimodal Biomarker Discovery, and Metabolic Syndrome in Men. In total, 1149 proteins were measured with liquid chromatography mass spectrometry with multiple reaction monitoring/Rules‐Based Medicine, tandem mass tag mass spectrometry, and SOMAscan. We tested associations between age and protein levels in linear models and tested enrichment for Reactome pathways. Results Levels of 252 proteins increased with age independently of amyloid status. These proteins were associated with immune and signaling processes. Levels of 21 proteins decreased with older age exclusively in amyloid abnormal participants and these were enriched for extracellular matrix organization. Discussion We found amyloid‐independent and ‐dependent CSF proteome changes with older age, perhaps representing physiological aging and early AD pathology.

Alzheimer's disease (AD) is heterogenous on the molecular level. Understanding this heterogeneity is critical for AD drug development. We aimed to define AD molecular subtypes by mass spectrometry proteomics in cerebrospinal fluid (CSF). Of the 3863 proteins detected in CSF, 1058 proteins had different levels in individuals with AD (n=419) compared with controls (n=187). Cluster analyses of AD individuals on these 1058 proteins revealed five subtypes: subtype 1 was characterized by neuronal hyperplasticity; subtype 2 by innate immune activation; subtype 3 by RNA dysregulation; subtype 4 by choroid plexus dysfunction; and subtype 5 by blood-brain barrier dysfunction. Distinct genetic profiles were associated with subtypes, e.g., subtype 1 was enriched with TREM2 R47H. Subtypes also differed in brain atrophy and clinical outcomes. For example, survival was shorter in subtype 3 compared to subtype 1 (5.6 versus 8.9 years). These novel insights into AD molecular heterogeneity highlight the need for personalized medicine.

Background We previously identified four Alzheimer’s disease (AD) subgroups with increasingly higher cerebrospinal fluid (CSF) levels of tau phosphorylated at threonine 181 (p-tau). These subgroups included individuals across the cognitive spectrum, suggesting p-tau subgroups could reflect distinct biological changes in AD, rather than disease severity. Therefore, in the current study, we further investigated which potential processes may be related with p-tau subgroups, by comparing individuals on CSF markers for presynaptic structure [vesicle-associated membrane protein 2 (VAMP2)], postsynaptic structure [neurogranin (NRGN)], axonal damage [neurofilament light (NfL)], and amyloid production [beta-secretase 1 (BACE1) and amyloid-beta 1–40 (Aβ40)]. Methods We selected 348 amyloid-positive (A+) individuals (53 preclinical, 102 prodromal, 193 AD dementia) and 112 amyloid-negative (A−) cognitively normal (CN) individuals from the Amsterdam Dementia Cohort (ADC). Individuals were labeled according to their p-tau subgroup (subgroup 1: p-tau ≤ 56 pg/ml; subgroup 2: 57–96 pg/ml; subgroup 3: 97–159 pg/ml; subgroup 4: > 159 pg/ml). CSF protein levels were measured with ELISA (NRGN, BACE1, Aβ40, NfL) or single-molecule array (Simoa) (VAMP2). We tested whether protein levels differed between the p-tau subgroups within A+ individuals with linear models corrected for age and sex and whether disease stage influenced these relationships. Results Among A+ individuals, higher p-tau subgroups showed a higher percentage of AD dementia [subgroup 1: n = 41/94 (44%); subgroup 2: n = 81/147 (55%); subgroup 3: n = 59/89 (66%); subgroup 4: n = 7/11 (64%)]. Relative to controls, subgroup 1 showed reduced CSF levels of BACE1, Aβ40, and VAMP2 and higher levels of NfL. Subgroups 2 to 4 showed gradually increased CSF levels of all measured proteins, either across the first three (NfL and Aβ40) or across all subgroups (VAMP2, NRGN, BACE1). The associations did not depend on the clinical stage (interaction p-values ranging between 0.19 and 0.87). Conclusions The results suggest that biological heterogeneity in p-tau levels in AD is related to amyloid metabolism and synaptic integrity independent of clinical stage. Biomarkers reflecting amyloid metabolism and synaptic integrity may be useful outcome measures in clinical trials targeting tau pathology.

Background Memory loss is central to Alzheimer’s disease (AD), but the precise underlying mechanisms remain unclear. Post‐mortem research suggests that synaptic loss best explains symptoms. Here, we investigated in cerebrospinal fluid (CSF) whether levels of synaptic proteins are related with memory scores in older individuals with normal cognition and in preclinical AD. Method Participants were selected from ADNI and EMIF‐AD MBD if they had CSF proteomics and memory scores available. In total, we selected 85 controls (i.e., normal cognition (CN) and normal CSF amyloid‐1‐42 and tau levels; mean±sd age 69±9 years, 58% female) and 46 preclinical AD (i.e. normal cognition and abnormal CSF amyloid‐1‐42 level, normal or abnormal CSF t‐tau and p‐tau; mean±sd age 74±7 years, 50% female). Age‐, sex‐ and education‐specific norms were used to normalize memory tests (ADNI: 30‐minute delayed Rey auditory verbal learning test (ADNI), EMIF: centre‐specific delayed memory tests). We selected 37 proteins available in both cohorts that were specific for synapses according to synGO (https://www.syngoportal.org/), Z‐transformed protein levels relative to controls, and pooled these across cohorts. Protein‐memory associations were assessed with linear models; model1: including covariates education, sex and age; model2: model1 + covariate amyloid status; and model3: model2 + protein level‐amyloid status interaction. Result Memory scores were similar in the preclinical AD and control groups (P=0.52, Figure 1). Three proteins (NCSTN, FGA and NRGN) had increased levels in preclinical AD vs. controls. Across the total group, no proteins were associated with memory functioning. Repeating models including an interaction term with amyloid status showed that 9 synaptic proteins (24.3%) were associated with memory depending on amyloid status. In controls, lower protein levels related with better memory, while in preclinical AD, lower protein levels tended to be related with worse memory (Figure 2). Conclusion Of the synaptic proteins tested, associations with memory were mediated by amyloid status. The finding that lower synaptic protein levels related to better memory in controls, but to worse memory in preclinical AD suggests that synaptic protein levels may reflect different biological processes in these groups.

BackgroundWe previously identified three CSF proteomic Alzheimer’s disease (AD) subtypes: one with increased amyloid metabolism and aberrant neuronal plasticity, one with innate immune activation and one with blood‐brain barrier dysfunction. We studied their replicability in another cohort, and whether increasing the number of proteins (from 556 to 1059) would enable detecting more subtypes.MethodWe selected 419 AD individuals with abnormal CSF abeta42 and 187 controls (normal cognition and normal AD biomarkers) from Alzheimer center Amsterdam studies. With 16‐plex TMT‐MS we detected 3987 proteins in CSF, of which we clustered 1059 proteins with complete observations that were associated with AD (all p<.05). Subtypes were compared on all 2906 proteins with at least 10 observations per subgroups. Potential upstream drivers of molecular subtypes were identified with ENRICHR.ResultWe found 5 subtypes with distinct protein profiles, of which three subtypes were highly concordant with our previously observed subtypes (figure1: subtype 1a, 2a and 2b) and two were new (1b and 1c). Compared to controls, subtype 1a, 1b and 1c had high levels of neuronal plasticity proteins, and subtype 2a and 2b had low levels of plasticity proteins. Thirty‐one proteins were increased in all subtypes (figure 2). Subtype 1a individuals (n=137, 32%) also showed increased amyloid processing. Their 827 proteins associated with REST and SUZ12 as potential upstream drivers. Subtype 1b individuals (n=124, 30%) also showed increased levels of inflammation proteins. Their 988 proteins converged on SOX2 as potential upstream driver. Subtype 1c individuals (n=24,6%), also showed proteasome dysfunction. Their 517 proteins were associated with TAF1 and MYC as potential upstream drivers. Subtype 2a individuals (n=78, 19%) showed increased levels of 469 proteins, of which 45% is expressed by the choroid plexus. These converged on NFE2L2. Subtype 2b individuals (n=56, 13%) showed high levels of 649 proteins that associated with blood‐brain barrier (BBB) leakage. These proteins converged on ERG1 and ESR1 as potential upstream drivers.ConclusionIn this new dataset we replicated three AD subtypes, and detected two new subtypes. Subtypes differed on plasticity proteins, immune activation, choroid plexus function and BBB function. Each AD subtypes may require a different treatment.

BackgroundGlial fibrillary acidic protein (GFAP) is a novel Alzheimer’s Disease (AD) biomarker that associates with amyloid pathology and pathology‐related changes can be detected in different biofluids. We studied how the performance of GFAP to distinguish between controls, AD and dementia with Lewy bodies (DLB), depends on the studied biofluid (cerebrospinal fluid (CSF), serum or plasma) and the confounding effects of age or sex.MethodFrom the Amsterdam Dementia Cohort and a local repository of healthy controls we included 372 cognitively normal individuals (CN), 255 patients with AD and 120 patients with DLB (Table 1) and GFAP levels were measured in one (n=322), two (n=416) or three (n=9) biofluids. GFAP was measured using Simoa in CSF (n=473), plasma (n=257) and serum (n=451) samples. For individuals with DLB, GFAP was measured in CSF and serum only. Effects of clinical diagnosis, age and sex on GFAP levels in each body fluid were estimated with linear models. Estimated differences between diagnostic groups were corrected for age and sex, age effects were corrected for sex and diagnostic group, and sex effects were corrected for age and diagnostic group.ResultGFAP levels were increased in AD relative to controls in all biofluids (fold change (FC): 1.6(CSF), 1.9(plasma) and 1.4(serum), p‐values<0.001) and increased in AD relative to DLB (FC: 1.7(CSF) and 1.3(serum), p‐values<0.001, Figure 1). No difference was found between DLB and controls. CSF GFAP levels increased with age in all clinical groups (range of standard deviation protein level increase per year (SD/year): 0.020‐0.024), while plasma GFAP levels increased with age only in CN (SD/year: 0.032). Serum GFAP levels increased with age in CN and DLB (SD/year: 0.058(CN) and 0.074(DLB), Figure 2). CSF GFAP showed sex effects in AD only (FC in males compared to females: 1.1), whereas serum GFAP showed sex effects in all clinical groups (FC in males: 0.83(AD), 0.85(NC) and 0.72(DLB)), and plasma GFAP showed no sex effects (Figure 3A‐C).ConclusionDifferences in GFAP levels between clinical diagnoses showed the same trend for all three matrices. With older age, differences in GFAP levels between controls and AD become harder to detect.