Defining the natural history of Alzheimer’s disease by longitudinal cerebrospinal fluid proteomics.

Defining the natural history of Alzheimer’s disease by longitudinal cerebrospinal fluid proteomics. | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Brief Communication Defining the natural history of Alzheimer’s disease by longitudinal cerebrospinal fluid proteomics. Betty Tijms, Diederick de Leeuw, Calvin Trieu, Martí Jimenéz-Mausbach, and 13 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8544834/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Defining the molecular natural history of Alzheimer’s disease (AD) is essential for earlier diagnosis and effective therapeutical interventions. Using longitudinal cerebrospinal fluid proteomics from 191 individuals across the AD spectrum, with up to 15 years of follow-up, we identified protein changes that preceded and tracked with development of amyloid-β (Aβ) and/or p-tau181 abnormality over 2.8–6.1 years. Early alterations included SMOC1 and YWHAZ and were enriched for glycolytic, synaptic, and axonal guidance pathways, which remained consistently altered across advancing disease stages. After Aβ and p-tau181 abnormality, 185 proteins progressively decreased with disease worsening, of which a subset of synaptic and axonal proteins including NPTX2, EPHA10 also tracked cognitive decline. Together, our findings can support clinical trial design through accurate effect size estimates, and enable biomarker guided therapeutic targeting across the full pre-amyloid-to-dementia disease continuum. Health sciences/Neurology/Neurological disorders/Neurodegenerative diseases/Alzheimer's disease Health sciences/Biomarkers/Diagnostic markers Figures Figure 1 Main text Over 57 million people worldwide have dementia, which is caused by Alzheimer’s disease (AD) in 70% of the cases. AD is biologically defined by amyloid-β (Aβ) plaques and tau tangles, yet the temporal sequence of molecular changes preceding and following these hallmark pathologies remains unclear. Identifying such molecular changes is difficult, as the brain is largely inaccessible for detailed molecular investigations in patients. Cerebrospinal fluid (CSF), which is in close contact with the brain, provides an accessible window into disease-related molecular processes and has been instrumental in establishing Aβ and tau as core AD biomarkers 1 . Proteomic techniques allow simultaneous measurement of thousands of additional proteins in single CSF samples, leading to the identification of consistent alterations in AD relative to controls 2–8 . Longitudinal CSF studies using targeted techniques have demonstrated that it is possible to find early changes related to Aβ and tau. 9–11 However, whether additional CSF protein changes occur before Aβ and tau abnormalities, and how these evolve across the disease continuum remains unclear. Here we measured longitudinal CSF untargeted proteomics to characterise temporally ordered molecular changes across the AD spectrum. We analysed 465 serial CSF samples from 191 individuals, including cognitively intact participants, individuals with mild cognitive impairment, and patients with AD dementia, with up to 15 years follow-up (average 3.4±2.3 years; on average 2.4±0.7 serial samples; table; methods). In parallel, we measured Aβ42/40, phosphorylated tau at threonine-181 (ptau181) and total tau (ttau) to anchor proteomic changes to established biomarkers of AD pathology. This design enabled us to identify protein alterations that preceded, coincided with, or followed the development of amyloid and tau abnormalities. Among 83 cognitively intact individuals with normal Aβ42/40(A-) and ptau181(T-) markers at baseline, a subset of 11(13.3%) individuals developed abnormal Aβ42/40 levels over 2.8±1.5 years (figure a; supplementary table 1), defining a ‘pre-amyloid’ stage. In these individuals, CSF ttau and ptau181 levels also increased (figure b), reaching abnormal levels with a time lag of 2.6-3.3 years relative to Aβ42/40 abnormality (figure b, c; supplemental figure). After that time, ptau181 levels in the pre-amyloid group were at the same height of cognitively intact A+T+ individuals. In contrast, tau markers increased much slower in the group of cognitively intact individuals with A+T- at baseline relative to the pre-amyloid individuals and were projected to develop abnormal ptau181 levels after ~11.9±10.3 years (supplementary figure). This finding corroborates our previous observations that cognitively intact A+T- individuals may represent a biological subtype of AD 2,4 , rather than a ‘pre-tau’ stage. Indeed, comparisons on the CSF proteome also suggested specific alterations in cognitively intact A+T- (supplemental table 1). Accordingly, we took the pre-amyloid, cognitively intact A+T+, mild cognitive impairment A+T+ and dementia A+T+ groups as representing the typical AD disease course for the remainder of this study. We estimated that progression to each subsequent disease stage took approximately 5-6 years, consistent with previous estimates 12 . Thus, we defined a ‘concatenated disease time’, in which we estimated for each stage the average trajectory of CSF protein level changes over 6-year time windows: starting in the pre-amyloid stage, followed by intact cognition with A+T+, mild cognitive impairment with A+T+ and AD dementia A+T+. Using this framework, we examined longitudinal CSF proteomic changes relative to the emergence of amyloid and tau abnormalities and across subsequent disease stages. We assessed longitudinal CSF changes for 1506 of the 3203 proteins detected, selected for having >400 observations across all 465 samples. At baseline, CSF levels of 459 (30.5%) proteins differed in any of the groups relative to controls, and 615 (40.8%) proteins had significant changes over time in at least one of the groups. Full statistics for all proteins and groups are presented in supplementary table 1. We identified a set of 55 proteins with longitudinal CSF changes that emerged during the pre-amyloid stage and persisted across disease progression. Early alterations included increasing levels of SMOC1 and YWHAZ and other proteins that were enriched for glycolytic processes (figure d, e), as well as decreasing levels of proteins including synaptic protein NPTX2 and interneuron marker SST (figure e). Together, the 43 increasing proteins were enriched for the upstream transcription factor PBX3 that has DNA binding activity and is involved in nervous system development 13 (supplementary table 3). Several of these proteins have been implicated in cross-sectional CSF mass spectrometry studies 4,6,14 , targeted longitudinal studies 10 , and in cross-sectionally modelled disease-time estimates in autosomal dominant AD 15 and Down syndrome 16 . While most proteins in our study had consistent timings of change as previously estimated for SMOC1 and YWHAZ 15 , others, NPTX2 and NPTXR, occurred earlier in our longitudinal measurements. Our longitudinal analyses demonstrate that these alterations arise years before conventional biomarker positivity in sporadic AD and remain stable across advancing disease stages. The persistent directional changes along the AD disease course suggest sustained engagement of metabolic and synaptic processes early in AD pathogenesis. We then set out to understand the ordering of proteomic changes, and identified distinct proteomic signatures that predicted subsequent change in Aβ42/40 and/or ptau181 levels over time (‘upstream proteins’), or changed alongside Aβ42/40 and/or ptau181, or had longitudinal changes predicted by baseline Aβ42/40 and/or ptau181 levels (‘downstream proteins’; figure f). Higher baseline levels of proteins that were enriched for humoral immune response that predicted steeper decreases in Aβ42/40 (figure h), and this included UCHL1, which plays a role in clearing misfolded proteins 17 . Baseline levels of SMOC1 and ALDOA on the other hand predicted subsequent increases in ptau181 levels, together with another 109 proteins that were enriched for synapse organisation, axon development, glycolytic processes, heparan sulfate proteoglycan metabolism, and gliogenesis, of which most started changing in later disease stages. Longitudinal changes of 5 proteins, including NUTF2 (a nuclear transport factor), occurred alongside increasing ptau181 levels. Following the development of Aβ42/40 and ptau181 abnormality, we observed a distinct pattern of proteomic change characterised by progressive decreases in CSF protein levels with disease worsening. In total, 185 proteins declined across later disease stages and were enriched for synapse organisation, axonal development and cell adhesion (figure k, l). This group included proteins VGF and NRXN1, of which lower CSF levels in dementia compared to controls have been reported in previous cross-sectional 2,15,16 and longitudinal 10 mass spectrometry studies. We also observed increasing levels of ACHE in the dementia stage. ACHE breaks down acetylcholine in the synapse and might indicate that dysregulated cholinergic signalling is a late event, consistent with a previous postmortem study. 18 These findings indicate that synaptic and axonal dysfunction emerge downstream of Aβ42/40 and ptau181 abnormality and exacerbates with clinical progression. To assess the clinical relevance of these proteomic changes, we examined their relationship with cognitive decline. To increase statistical power, we combined control and pre-amyloid group into A- and tested decline on the memory delayed recall scores, as this test is sensitive to very early cognitive alterations 11 . Baseline levels of 112 proteins predicted steeper decline on delayed recall, including YWHAZ (figure j) and C3. These proteins were enriched for extracellular matrix organisation and innate immune activation (figure h). Of note, none of proteins with consistent changes over concatenated disease time changed simultaneously with delayed recall scores. In later disease stages, we tested global cognitive decline measured by the Mini-Mental State Examination (MMSE) across all A+T+ participants. Higher baseline levels of 9 proteins including YWHAZ and PGAM1, and lower levels of 56 proteins including SCG2 and NPTX2 predicted steeper decline on the MMSE. These proteins were enriched for cell adhesion, neuron and glia development (figure o). Of these several synaptic and axonal proteins, including NPTX2 and EPHA10, tracked longitudinally with worsening cognitive performance measured by the MMSE. These associations suggest that late-stage proteomic changes reflect neurobiological processes that are closely linked to cognitive deterioration and may serve as proxy markers of disease progression. A limitation of this study is that within each stage samples sizes are modest, reflecting the challenges in obtaining repeated lumbar punctures. This primarily reduced power for disease stage-specific analyses with cognitive decline, which is inherently heterogenous in AD 19 . Nevertheless, our total sample size is large, and our results recapitulate previous CSF proteomic studies that used cross-sectional designs 2–8 or modelled disease-time in autosomal dominant AD and Down syndrome 15,16 , supporting the validity and robustness of our proteomic approach. Together, these results define the molecular natural history of sporadic AD across ~25 years, spanning from pre-amyloid stages to dementia, using serial CSF proteomics. We identified early biomarker candidates beyond Aβ and tau that change years before established biomarker abnormality and delineate downstream proteomic signatures associated with synaptic dysfunction and cognitive decline. By providing temporally ordered proteomic signatures and quantified effect size estimates, this work informs clinical trial design, supports biomarker-guided patient stratification, and highlights a window for earlier therapeutic intervention in sporadic AD. Methods Study participants We selected 246 individuals from the Amsterdam Dementia Cohort 20 or the EMIF-AD preclinAD study 21 when they had longitudinally collected repeated CSF samples (in total 628 samples) available in our biobank and were on their first visit cognitively intact (CI) with normal Aβ42/40 (A-) and ptau181 levels (T-), or at least A+ based on CSF (n=190) or PET (n=1). The present analyses were performed on proteomic results from n=191 with 465 samples, excluding small groups or related individuals (3 MCI A+T-, 3 Dementia A+T- and 3 NC A-T+, 2 MCI and 2 dementia individuals who upon remeasuring Aβ42/40 had an A- status, and n=42 co-twins). MCI and AD type dementia diagnoses were determined at our memory clinic during multidisciplinary consensus meetings based on international consensus guidelines 20 . Most individuals had repeated cognitive testing around the time of CSF sampling 20,21 . We used the MMSE score as a measure of global cognitive performance in individuals with A+. In the A- group we used the delayed recall score of the Dutch Auditory Verbal Learning Task (AVLT, possible score range 1-15), as we previously found this measure to be most sensitive to cognitive decline in very early stages of AD 11,22 . All participants provided written informed consent to use their biomaterial and clinical data for scientific research. All studies were approved by the medical ethics committee of the Amsterdam UMC in the Netherlands, as well as the Western Norway regional committee for medical and health research ethics. CSF collection and targeted measurements CSF samples were obtained by lumbar puncture between the L3/L4, L4/L5 or L5/S1 intervertebral space with a 25-gauge needle and syringe and collected in polypropylene tubes. For all participants CSF sample collection, processing and biobank storage at the Alzheimer center biobank at the department of Laboratory Medicine was performed according to international guidelines 23 . The 628 samples were randomised over 96-well plates, keeping repeated samples from the same individual next to each other and randomising on Aβ, tau and clinical status within blocks of 14 samples (see next section on TMT measurements). We used random sampling as implemented in R to determine the layouts and with 100 μl CSF in each well and stored at −80 °C. Then Aβ 1-42, Aβ 1-40, phosphorylated tau at threonine 181 and total tau were measured in each sample in singlicate using the fully automated CLEIA on the LUMIPULSE® G System (LUMIPULSE® G600II, REF: 703380; Fujirebio Diagnostics, Inc.) according to the manufacturer’s instructions at the Neurochemistry Laboratory at the Amsterdam University Medical Center, the Netherlands. Part of those data were reported before 24 . Technicians were blinded to diagnosis. Five individuals had too low CSF volume at baseline to remeasure Aβ as well as performing mass spec, for which we prioritized mass spec and used their historical measures based on Innotest (n=1), Euroimmune (n=3) or amyloid PET (n=1) to determine baseline A status. We used for Aβ42/40 a cutoff of 61.5 pg/ml to determine tau abnormality (unpublished based on same cohort as 25 ). Finally, a few samples had ttau levels >2000 pg/ml, which was above the upper limit of detection, and we imputed these values based on individuals ptau181 levels that correlated strongly (r=.68 before and r=.92 after removal of technical outliers, p-value <0.001). TMT mass spec measurements Proteomic analyses were performed using tandem mass tag (TMT) mass spectrometry as previously described². CSF protein concentrations were measured by BCA assay, and 25 μg of protein per sample was aliquoted into 96-well Lo-Bind plates, frozen on dry ice, lyophilized, and stored at −80 °C. Proteins were solubilized in 8 M urea/20 mM methylamine, reduced with 10 mM DTT, alkylated with 25 mM iodoacetamide, and quenched with DTT. After dilution to 1 M urea, samples were digested overnight at 37 °C with trypsin. Peptides were acidified, desalted, and lyophilized. Control samples were resuspended for LC-MS analysis; all others were resuspended in HEPES buffer for TMT labeling. A total of 644 samples were labeled across 46 experiments (14 samples + 2 references per set), combined, desalted, and fractionated by high-pH reverse-phase HPLC. Ten fractions per set were collected, lyophilized, resuspended, and peptide concentrations were measured using Nanodrop before LC–MS/MS analysis, and 500 ng of peptides was injected for LC-MS/MS analysis LC–MS/MS The acquisition of experiments was randomized to avoid bias. The peptide mixture was measured on an Ultimate 3000 RSLC system (Thermo Fisher Scientific) coupled to an Orbitrap Exploris 480 mass spectrometer with an EASY-Spray nano-ESI source. Peptides were trapped on a PepMap precolumn (2 cm × 75 μm, 3 μm C18) and separated on a 25 cm analytical column (PepMap RSLC, 2 μm C18) using a biphasic ACN gradient (solvent A: 0.1% formic acid in water; solvent B: 100% ACN) at 250 nl/min. The gradient was: 5% B for trapping (5 min), 5–7% B over 0.5 min, 7–25% B over 76.5 min, 25–38% B over 15 min, 38–85% B over 3 min, followed by 7 min at 85% B and 10 min at 5% B for column conditioning. MS acquisition was performed in FAIMS-enabled data-dependent mode with two compensation voltages (−50 V and −70 V). Full MS scans were acquired at 60,000 resolution (400–1,600 m/z), AGC target 3 × 10⁶, followed by HCD fragmentation (NCE 32%) and MS/MS scans at 45,000 resolution. Dynamic exclusion was set to 30 s. Raw files were processed in Proteome Discoverer 2.5 using Sequest HT against Swiss-Prot (July 2023). Search parameters included 10 ppm precursor tolerance, 0.02 Da fragment tolerance, static TMTpro (+304.207 Da on N-terminus and K), carbamidomethyl (C), and dynamic methionine oxidation. Up to two missed cleavages were allowed, minimum peptide length 6. PSM FDR was controlled at 1% (strict) using target-decoy. Reporter ion quantification was intensity-based with coisolation threshold 50 and S/N ≥10. Postprocessing TMT mass spec data Technical deviations may have influenced protein abundance across the TMT experiments. Before the statistical analyses, we normalized protein abundance according to the internal reference scaling normalization procedure 100 for TMT proteomics data that use the common pool reference channels to normalize values between plex experiments 26 , adapted to scale according to the median instead of the total sum to reduce the influence of outliers. Briefly, the first step in this two-step approach normalized the grand total protein intensities for each of the 14 channels within an experiment to match these to the two reference channels. In the second step, a correction factor was calculated based on common pooled internal standards to normalize reporter ion intensities of proteins between TMT experiments. Next, protein values were log 2-transformed and then scaled according to the mean and standard deviation of the baseline measurement in the control group, so that positive and negative values indicate higher and lower than normal. For all proteins, we report gene names to aid comparisons with other AD subtyping literature using either proteomics or RNA-seq data. Upon visual quality control, some experiments had low levels for specific proteins, which are flagged in the results file (supplemental table 1). Statistical analyses We compared groups on descriptive characteristics with chi-squared tests for discrete variables (sex and APOE e4 genotype), and with linear regression models for continuous outcomes. Next, we tested for each protein changes over time using linear mixed models with repeated protein levels as outcome measure, time and group as main predictors as well as an interaction terms of time X group, age and sex as covariates. Not all models converged when including both random slopes and intercepts, and so for reasons of parsimony ran these models including random intercepts only. Differences between group at baseline, group specific slopes and group differences between slopes were extracted from each model with the emmeans package (v1.11.1). Next, we performed generalised linear mixed-effects models (glmm) using a Markov Chain Monte Carlo (MCMC) approach as implemented in the MCMCglmm package (v2.36) 27 to test which proteins changed together with Aβ42/40, ptau181 and ttau, as well as with delayed memory recall test and MMSE scores over time. We used uninformed priors, an effective sample size of 1000, thinning of 100, and a burn-in of 15000 (resulting in 115000 iterations per protein). A strength of this approach is that it can simultaneously model changes over time in multiple outcomes (e.g., Aβ42/40 levels or neuropsychological test scores together with other CSF protein levels). Thus, with this model correlations of the random effects (i.e., intercepts with each other, intercepts of one variable with slope of the other variable, and between slopes of both variable) can be easily inferred from the sampled posterior distributions, which is not possible for a standard linear mixed model that gives single parameter estimates based on likelihood maximization. For the main analyses, our aim was to identify patterns of coherent changes over time across concatenated disease time. As such, we used a liberal p value threshold of p<0.05 to determine significance as well as consistent changes over different disease stages. Please note that all results are reported in the supplement. We used the Gene Ontology database (GO; release 2025-10-10) through Panther v19.0 to study if proteins with consistent changes were related to specific biological processes and report associated pathways when pFDR<0.05, and ENRICHR to identify potential upstream transcription factors associated with CSF proteomic changes with pFDR<0.05. All analyses were performed in R v4.5.0 “How About a Twenty-Six”. Declarations Acknowledgements BMT received support for this work from ZonMW VIDI (#09150171910068), the European Union (ERC, DecipherAD, #101171721). Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Council. Neither the European Union nor the granting authority can be held responsible for them, TAP-dementia (www.tap-dementia.nl) that is funded by ZonMW (no. 10510032120003), and the Amsterdam Cohort Hub, as part of the Sector Plan 'Accelerating Health' of the Dutch Ministry of Education, Culture and Science, and Novo Nordisk. Mass spectrometry-based proteomic analyses were performed by the Proteomics Unit at the University of Bergen (PROBE; KFW, OM, FB). This facility is a member of the National Network of Advanced Proteomics Infrastructure (NAPI), which is funded by the Reseearch Council of Norway (INFRASTRUKTUR-program project number: 295910). Research of AH is funded by Stichting Alzheimer Nederland (WE.06-2021-06), the Alzheimer Drug Discovery Foundation (CANTATE project), Stichting Steun Alzheimer Centrum Amsterdam and the Davos Alzheimer Collaborative. She is a recepient of the TAP dementia fund (the Dutch Research Counsil, ZonMW, no. 10510032120003). Research of FD is funded by ZonMW, Alzheimer Nederland (WE.03-2018-10), Stichting Steun Alzheimercentrum Amsterdam, and Davos Alzheimer Collaborative (AccDx program). CT receives research support by the European Commission (MarieCurie International Training Network, Grant Agreement No. 860197(MIRIADE) and No. 101119596 (TAME), Innovative Medicines Initia-tives 3TR (Horizon 2020, grant 831434) EPND (IMI 2 Joint Under-taking (JU), grant 101034344), and JPND (bPRIDE, CCAD), EuropeanPartnership on Metrology, co-financed by the European Union’s Hori-zon Europe Research and Innovation Programme and by the Participating States ((22HLT07 NEuroBioStand), Horizon Europe (PREDICTFTD,101156175, CCAD), CANTATE project funded by the Alzheimer Drug Discovery Foundation, Alzheimer’s Association (grant SG-22-856131-SABB NEXT), Michael J. Fox Foundation,Health Holland, the Dutch Research Council (ZonMW), AlzheimerDrug Discovery Foundation, Selfridges Group Foundation, AlzheimerNetherlands. C.T. is recipient of ABOARD, which is a public-privatepartnership receiving funding from ZonMW (No. 73305095007) and Health Holland, Topsector Life Sciences & Health (PPP-allowance;No. LSHM20106). C.T. is recipient of TAP-dementia, a ZonMw fundedproject (No. 10510032120003) in the context of the Dutch NationalDementia Strategy. PJV received support from the European Federation of Pharmaceutical Industries and Associations Innovative Medicines Initiative Joint Undertaking (European Medical Information Framework grant 115372,), Stichting Dioraphte, Alzheimer Nederland (WE. 09-2016-10). The project has received funding from the Innovative Medicines Initiative Joint Undertaking under grant agreement (115952). This Joint Undertaking receives the support from the European Union’s Horizon 2020 research and innovation program and European Federation Pharmaceutical Industries and Associations. Research of Alzheimer Centre Amsterdam is part of the neurodegeneration research program of Amsterdam Neuroscience. Alzheimer Centre Amsterdam is supported by the Stichting Alzheimer Nederland and Stichting Steun Alzheimercentrum Amsterdam. COI M.J. is an employee and minor shareholder of Novo Nordisk A/S. AH was part of the educational advisory board of Eli Lilly and the advisory board of the Brain Research Center. CET has research contracts with Acumen, ADx Neurosciences, AC-Immune, Alamar, Aribio, Axon Neurosciences, Beckman-Coulter, BioConnect, Bioorchestra, Brainstorm Therapeutics, C2N diagnostics, Celgene, Cognition Therapeutics, EIP Pharma, Eisai, Eli Lilly, Fujirebio, Instant Nano Biosensors, Merck, Muna, Nitrase Therapeutics, Novo Nordisk, Olink, PeopleBio, Quanterix, Roche, Sysmex, Toyama, Vaccinex, Vivoryon. She is editor in chief of Alzheimer Research and Therapy, and serves on editorial boards of Molecular Neurodegeneration, Alzheimer’s & Dementia, Neurology: Neuroimmunology & Neuroinflammation, Medidact Neurologie/Springer, and is committee member to define guidelines for Cognitive disturbances, and one for acute Neurology in the Netherlands. She has consultancy/speaker contracts for Aribio, Biogen, Beckman-Coulter, Cognition Therapeutics, Danaher, Eisai, Eli Lilly, Janssen, Merck, Neurogen Biomarking, Nordic Biosciences, Novo Nordisk, Novartis, Olink, Quanterix, Roche, Sanofi and Veravas. Data availability Raw proteomic and clinical data is available upon reasonable request for the purposes of replication. References Jack, C. R. et al. Revised criteria for diagnosis and staging of Alzheimer’s disease: Alzheimer’s Association Workgroup. Alzheimer’s Dement. 20 , 5143–5169 (2024). Tijms, B. M. et al. Cerebrospinal fluid proteomics in patients with Alzheimer’s disease reveals five molecular subtypes with distinct genetic risk profiles. Nat. Aging 33–47 (2024) doi:10.1038/s43587-023-00550-7. Tijms, B. M. et al. Pathophysiological subtypes of Alzheimer’s disease based on cerebrospinal fluid proteomics. Brain 143 , 3776–3792 (2020). Visser, P. J. et al. Cerebrospinal fluid tau levels are associated with abnormal neuronal plasticity markers in Alzheimer’s disease. Mol Neurodegener 17 , (2022). Johnson, E. C. B. et al. Large-scale proteomic analysis of Alzheimer’s disease brain and cerebrospinal fluid reveals early changes in energy metabolism associated with microglia and astrocyte activation. Nature Medicine 1–31 (2020) doi:10.1038/s41591-020-0815-6. Higginbotham, L. et al. Integrated proteomics reveals brain-based cerebrospinal fluid biomarkers in asymptomatic and symptomatic Alzheimer’s disease. Sci Adv 6 , eaaz9360 (2020). Geus, M. B. de et al. Mass spectrometry in cerebrospinal fluid uncovers association of glycolysis biomarkers with Alzheimer’s disease in a large clinical sample. Sci. Rep. 13 , 22406 (2023). Park, S. A. et al. SWATH-MS analysis of cerebrospinal fluid to generate a robust battery of biomarkers for Alzheimer’s disease. Scientific Reports 1–10 (2020) doi:10.1038/s41598-020-64461-y. Wildsmith, K. R. et al. Identification of longitudinally dynamic biomarkers in Alzheimer’s disease cerebrospinal fluid by targeted proteomics. Molecular Neurodegeneration 9 , 22 (2014). Libiger, O. et al. Longitudinal CSF proteomics identifies NPTX2 as a prognostic biomarker of Alzheimer’s disease. Alzheimer’s Dementia (2021) doi:10.1002/alz.12353. Tijms, B. M. et al. Pre-amyloid stage of Alzheimer’s disease in cognitively normal individuals. Annals of Clinical and Translational Neurology 5 , 1037–1047 (2018). Vermunt, L. et al. Duration of preclinical, prodromal, and dementia stages of Alzheimer’s disease in relation to age, sex, and APOE genotype. Alzheimer’s & Dementia 1–11 (2019) doi:10.1016/j.jalz.2019.04.001. Dammer, E. B. et al. Proteomic analysis of Alzheimer’s disease cerebrospinal fluid reveals alterations associated with APOE ε4 and atomoxetine treatment. Sci. Transl. Med. 16 , eadn3504 (2024). Modeste, E. S. et al. Quantitative proteomics of cerebrospinal fluid from African Americans and Caucasians reveals shared and divergent changes in Alzheimer’s disease. Mol. Neurodegener. 18 , 48 (2023). Johnson, E. C. B. et al. Cerebrospinal fluid proteomics define the natural history of autosomal dominant Alzheimer’s disease. Nat. Med. 1–10 (2023) doi:10.1038/s41591-023-02476-4. Montoliu-Gaya, L. et al. Proteomic analysis of Down syndrome cerebrospinal fluid compared to late-onset and autosomal dominant Alzheimer´s disease. Nat. Commun. 16 , 6003 (2025). Mi, Z. & Graham, S. H. Role of UCHL1 in the pathogenesis of neurodegenerative diseases and brain injury. Ageing Res. Rev. 86 , 101856 (2023). Davis, K. L. et al. Cholinergic Markers in Elderly Patients With Early Signs of Alzheimer Disease. JAMA 281 , 1401–1406 (1999). Jutten, R. J. et al. Finding Treatment Effects in Alzheimer Trials in the Face of Disease Progression Heterogeneity. Neurology 96 , e2673–e2684 (2021). Flier, W. M. van der & Scheltens, P. Amsterdam Dementia Cohort: Performing Research to Optimize Care. J. Alzheimer’s Dis. 62 , 1091–1111 (2018). Konijnenberg, E. et al. The EMIF-AD PreclinAD study: study design and baseline cohort overview. Alzheimer’s Research & Therapy 10 , S85 (2018). Donohue, M. C. et al. Association Between Elevated Brain Amyloid and Subsequent Cognitive Decline Among Cognitively Normal Persons. JAMA : the journal of the American Medical Association 317 , 2305–2316 (2017). Campo, M. del et al. Recommendations to standardize preanalytical confounding factors in Alzheimer’s and Parkinson’s disease cerebrospinal fluid biomarkers: an update. Biomarkers in Medicine 6 , 419–430 (2012). Leeuw, D. M. de et al. Cerebrospinal Fluid Amyloid and Tau Biomarker Changes Across the Alzheimer Disease Clinical Spectrum. JAMA Netw. Open 8 , e2519919 (2025). Willemse, E. A. J. et al. Comparing CSF amyloid‐beta biomarker ratios for two automated immunoassays, Elecsys and Lumipulse, with amyloid PET status. Alzheimer’s Dementia Diagnosis Assess Dis Monit 13 , e12182 (2021). Plubell, D. L. et al. Extended Multiplexing of Tandem Mass Tags (TMT) Labeling Reveals Age and High Fat Diet Specific Proteome Changes in Mouse Epididymal Adipose Tissue*. Mol Cell Proteomics 16 , 873–890 (2017). MCMCglmm Course Notes Jarrod Hadfield ( [email protected] ). 1–141 (2019). Tables Table 1 Participant characteristics at the time of first CSF sampling. Controls A-T- (n=72) Pre-amyloid A-T- (n=11) Cognitively intact A+T- (n=11) Cognitively intact A+T+ (n=20) MCI A+T+ (n=27) Dementia A+T+ (n=50) Characteristic n missing Age in years 0 62.96 ± 7.82 63.73 ± 7.98 61 ± 6.6 70.42 ± 8.47 a,b,c 66 .9± 7.3 a,c 65.5 ± 7.84 d Women 0 30 (41.7%) 7 (63.6%) 5 (45.5%) 15 (75%) 10 (37.0%) d 26 (52.0%) Education level 6 5.14 ± 1.59 4.73 ± 1.68 5.64 ± 1.36 4.79 ± 1.93 5.25 ± 1.4 5.24 ± 1.18 >=1 APOE e4 3 19 (26.76%) 8 (72.73%) a 8 (72.73%) a 13 (65%) a 22 (84.6%) a 33 (67.35%) a MMSE 1 28.73 ± 1.55 28.91 ± 1.45 28.64 ± 1.03 28.45 ± 1.61 25.6 ± 2.5 a,b,c,d 22.5 ± 4.21 a,b,c,d,e AVLT Delayed recall 25 8.62 ± 2.97 10 ± 3 8.27 ± 2.72 7.16 ± 2.14 a,b 2.5 ± 2.0 a,b,c,d 2.39 ± 2.66 a,b,c,d Aβ42/40 ratio 5 0.1 ± 0.01 0.08 ± 0.01 a 0.06 ± 0.01 a,b 0.04 ± 0.01 a,b,c 0.04 ± 0.01 a,b,c 0.04 ± 0.01 a,b,c Ptau181 pg/ml 5 36.53 ± 11.4 45.76 ± 12.16 40.96 ± 12.96 101.83 ± 35.97 a,b,c 125.8 ± 43.4 a,b,c,d 134.76 ± 52.29 a,b,c,d Ttau pg/ml 5 330.08 ± 107.47 387.44 ± 107.84 308.27 ± 93.34 721.95 ± 350.82 a,b,c 786.0 ± 237.3 a,b,c 853.69 ± 306.21 a,b,c,d Number of repeated CSF samples 0 2.68 ± 0.82 2.82 ± 0.87 2.36 ± 0.67 2.65 ± 0.81 2.2 ± 0.4 a,b,d 2.08 ± 0.27 a,b,d Averge time between CSF sampling 0 2.61 ± 1.24 3.47 ± 2.63 3.41 ± 2.46 2.35 ± 0.89 b,c 2.4 ± 1.6 b,c 1.61 ± 0.91 a,b,c,d,e Maximum years repeated CSF sampling 0 4.18 ± 2.18 5.69 ± 3.75 a 4.22 ± 2.47 3.78 ± 1.88 b 2.7 ± 1.8 a,b,c 1.68 ± 0.89 a,b,c,d,e A- is normal amyloid beta 1-42/1-40 (Aβ42/40) levels, A+ abnormal Aβ42/40 levels, T- is normal tau phosphorylated at threonine 181 (ptau181) levels, T+ is abnormal ptau181 levels, MCI is mild cognitive impairment, APOE is apolipoprotein, MMSE is mini mental state examination, AVLT is Auditory Verbal Learning Test, CSF is cerebrospinal fluid. Groups were compared using pairwise linear regression tests for continuous variables, or by chi2 tests for categorical measures a differs from controls with p<0.05, b differs from pre-amyloid with p<.05, c differs from cognitively intact A+T- with p<0.05, d differs from cognitively intact A+T+ with p<.05, e differs from mild cognitive impairment A+T+ with p<.05. Not enough volume to re-measure amyloid, ptau181, and ttau for n=5: for these individuals their amyloid and/or ptau status at first visit was determined based on amyloid PET for n=1 and on historical Euroimmune measures for n=4. Continuous measures are reported in mean ± standard deviation, discrete variables in n (%). Proteomic results for the cognitively intact A+T- group are reported in supplemental table 1. Additional Declarations Yes there is potential Competing Interest. BMT received funding from Novo Nordisk for this study, all competing interest can be found in the manuscript file. Supplementary Files supplementcombi1.pdf Supplementary figure Person trajectories and estimated group changes in CSF Aβ42/40 (top); ptau181(middle) and total tau (bottom) levels for controls, pre-amyloid, cognitively intact (CI) with abnormal Aβ42/4 (A+) and normal ptau181 (T-), CI with A+ and abnormal ptau181 (T+), mild cognitive impairment (MCI) with A+T+ and dementia with A+T+, * is slope significant different from 0 with p <0.05, a indicates that slope differs from controls with p <0.05. Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8544834","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Brief Communication","associatedPublications":[],"authors":[{"id":574776391,"identity":"23abb0a7-34ee-43a1-9d68-ebb3425de0d5","order_by":0,"name":"Betty Tijms","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA6klEQVRIiWNgGAWjYBACAyA+IMHAIAflM8MkLPBoYQZrMQayGRuQtEjg1QICiQ1EazFn4D94wDKnLr1/2vHnDz7usZbXnZGd+IFxB24tlg1Ah0luO5w743aOYeOMZ+mG227kbpZgPIPHYQfAWg7kNtzOYWzmOXCYEahlGwNjG0Etdenyt9MfNv85cNieWC3MCQa3EwybGQ4cTiSoxbKZ2QDkF8ONQL/M7DmQnrztzNvNEol4tJizNz7+DHSYvNzt9Acffhywtt12PHfjh49tNji1gCKCGdPEBNwawIDxAwEFo2AUjIJRMMIBAKOGWvPPyhU/AAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0002-2612-1797","institution":"Amsterdam UMC, location VUmc","correspondingAuthor":true,"prefix":"","firstName":"Betty","middleName":"","lastName":"Tijms","suffix":""},{"id":574776392,"identity":"b35a6b6e-8683-4aa8-b044-cbf0c0ea88c3","order_by":1,"name":"Diederick de Leeuw","email":"","orcid":"","institution":"Amsterdam UMC location VUmc","correspondingAuthor":false,"prefix":"","firstName":"Diederick","middleName":"","lastName":"de Leeuw","suffix":""},{"id":574776393,"identity":"92573e07-ffc1-4b18-bdcf-403b72b3ac6d","order_by":2,"name":"Calvin Trieu","email":"","orcid":"","institution":"Amsterdam University Medical Centers, Location VUmc","correspondingAuthor":false,"prefix":"","firstName":"Calvin","middleName":"","lastName":"Trieu","suffix":""},{"id":574776394,"identity":"a8cb89a3-c040-48cf-8cf0-7bd24c554691","order_by":3,"name":"Martí Jimenéz-Mausbach","email":"","orcid":"","institution":"Novo Nordisk A/S","correspondingAuthor":false,"prefix":"","firstName":"Martí","middleName":"","lastName":"Jimenéz-Mausbach","suffix":""},{"id":574776395,"identity":"6a84bb08-f40a-4483-9962-7d99597edc13","order_by":4,"name":"Katarina Fritz-Wallace","email":"","orcid":"https://orcid.org/0000-0002-2317-9387","institution":"University of Bergen","correspondingAuthor":false,"prefix":"","firstName":"Katarina","middleName":"","lastName":"Fritz-Wallace","suffix":""},{"id":574776396,"identity":"571fd7d8-acf4-4f9a-8c1c-9d1538118fce","order_by":5,"name":"Olav Mjaavatten","email":"","orcid":"","institution":"University of Bergen","correspondingAuthor":false,"prefix":"","firstName":"Olav","middleName":"","lastName":"Mjaavatten","suffix":""},{"id":574776397,"identity":"95623375-6aef-41c7-aec6-73a5cc9ebe99","order_by":6,"name":"Elena-Raluca Bludjea","email":"","orcid":"","institution":"Amsterdam UMC location VUmc","correspondingAuthor":false,"prefix":"","firstName":"Elena-Raluca","middleName":"","lastName":"Bludjea","suffix":""},{"id":574776398,"identity":"b4fafec2-4368-4628-89cd-46132de2c316","order_by":7,"name":"Roos Jutten","email":"","orcid":"https://orcid.org/0000-0002-3712-4790","institution":"Amsterdam UMC location VUmc","correspondingAuthor":false,"prefix":"","firstName":"Roos","middleName":"","lastName":"Jutten","suffix":""},{"id":574776399,"identity":"72922649-29fe-47a2-a85f-12d54611decd","order_by":8,"name":"Argonde van Harten","email":"","orcid":"","institution":"Alzheimer Center Amsterdam, Neurology, Vrije Universiteit Amsterdam, Amsterdam UMC location VUmc, Amsterdam, The Netherlands.","correspondingAuthor":false,"prefix":"","firstName":"Argonde","middleName":"van","lastName":"Harten","suffix":""},{"id":574776400,"identity":"ab474247-27f1-4d8a-8080-f4f3386dbf34","order_by":9,"name":"Flora Duits","email":"","orcid":"","institution":"Alzheimer Center Amsterdam, Department of Neurology, Amsterdam University Medical Centers, Vrije Universiteit, Amsterdam Neuroscience, Amsterdam, The Netherlands","correspondingAuthor":false,"prefix":"","firstName":"Flora","middleName":"","lastName":"Duits","suffix":""},{"id":574776401,"identity":"8e668861-5a49-4c55-8b12-d06b0ac29e66","order_by":10,"name":"Anouk den Braber","email":"","orcid":"","institution":"Amsterdam UMC, location VUmc","correspondingAuthor":false,"prefix":"","firstName":"Anouk","middleName":"den","lastName":"Braber","suffix":""},{"id":574776402,"identity":"5cb8b7b3-02ae-4c12-9955-154d7dd4fbaa","order_by":11,"name":"Henne Holstege","email":"","orcid":"https://orcid.org/0000-0002-7688-3087","institution":"Amsterdam UMC Location VUmc","correspondingAuthor":false,"prefix":"","firstName":"Henne","middleName":"","lastName":"Holstege","suffix":""},{"id":574776403,"identity":"e79f252b-da56-48ca-aa88-98795dbff640","order_by":12,"name":"Marissa Zwan","email":"","orcid":"","institution":"Alzheimer Center Amsterdam, Department of Neurology, Amsterdam Neuroscience, Vrije Universiteit Amsterdam, Amsterdam UMC, Amsterdam","correspondingAuthor":false,"prefix":"","firstName":"Marissa","middleName":"","lastName":"Zwan","suffix":""},{"id":574776404,"identity":"2e3cd6d5-d741-4728-bb5e-9227021d5913","order_by":13,"name":"Everard Vijverberg","email":"","orcid":"","institution":"Vrije Universiteit Amsterdam, Amsterdam UMC","correspondingAuthor":false,"prefix":"","firstName":"Everard","middleName":"","lastName":"Vijverberg","suffix":""},{"id":574776405,"identity":"6cd62f48-0b64-4f1d-b498-90435ef9f01a","order_by":14,"name":"Frode Berven","email":"","orcid":"","institution":"Proteomics Unit, Department of Biomedicine, University of Bergen","correspondingAuthor":false,"prefix":"","firstName":"Frode","middleName":"","lastName":"Berven","suffix":""},{"id":574776406,"identity":"119b91c7-b10c-4d54-8d9a-997058038dc7","order_by":15,"name":"Charlotte Teunissen","email":"","orcid":"https://orcid.org/0000-0002-4061-0837","institution":"Neurochemistry Laboratory and Biobank","correspondingAuthor":false,"prefix":"","firstName":"Charlotte","middleName":"","lastName":"Teunissen","suffix":""},{"id":574776407,"identity":"8f543e9b-f924-4c4a-9ceb-79fa54e98650","order_by":16,"name":"Pieter Jelle Visser","email":"","orcid":"https://orcid.org/0000-0001-8008-9727","institution":"AmsterdamUMC location VUmc","correspondingAuthor":false,"prefix":"","firstName":"Pieter","middleName":"Jelle","lastName":"Visser","suffix":""}],"badges":[],"createdAt":"2026-01-07 19:45:21","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8544834/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8544834/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":100355033,"identity":"1fed6de4-e28d-494f-8753-562fb7195545","added_by":"auto","created_at":"2026-01-16 05:31:03","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1589963,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003ea\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Person trajectories and estimated group changes in CSF Aβ42/40 and \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eb\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e ptau181 levels for controls, pre-amyloid, cognitively intact (CI) with abnormal Aβ42/40 (A+) and abnormal ptau181 (T+), mild cognitive impairment (MCI) with A+T+ and dementia with A+T+. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003ec\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Modelled Aβ42/40 (inverted such that higher indicates more abnormal to ease comparisons with ptau181) and ptau181 over concatenated disease time. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003ed \u003c/strong\u003e\u003c/em\u003e\u003cem\u003eand\u003c/em\u003e\u003cem\u003e\u003cstrong\u003e e \u003c/strong\u003e\u003c/em\u003e\u003cem\u003eproteins that respectively increase or decrease from the pre-amyloid stage into the dementia according to concatenated disease time. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003ef\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e log2(p)*-1 values of associations for proteins of which baseline levels predicted changes over time in CSF Aβ42/40 and/or ptau181 levels over time (i.e., upstream), proteins of which changes over time were predicted by baseline CSF Aβ42/40 or ptau181 levels (downstream), and proteins of which changes in CSF levels correlated with changes in Aβ42/40 and/or ptau181 levels over time.\u003c/em\u003e\u003cem\u003e\u003cstrong\u003e g \u003c/strong\u003e\u003c/em\u003e\u003cem\u003eselected gene ontology (GO) biological processes associated with early and later increasing proteins or decreasing proteins.\u003c/em\u003e\u003cem\u003e\u003cstrong\u003e h \u003c/strong\u003e\u003c/em\u003e\u003cem\u003eselected gene ontology (GO) biological processes associated with proteins that had upstream associations with Aβ42/40 or ptau181, or with subsequent decline over time on delayed memory recall scores. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003ei\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e person trajectories and estimated group changes on the Auditory Verbal Learning Test (AVLT) delayed recall scores. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003ej\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e participant slopes of delayed recall scores over time (y axis) predicted by baseline levels of YWHAZ (x axis) across the control (grey) and pre-amyloid (green) groups. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003ek\u003c/strong\u003e\u003c/em\u003e\u003cem\u003eand \u003c/em\u003e\u003cem\u003e\u003cstrong\u003el\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e protein changes over concatenated disease time that started in the CI A+T+ stage or later. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003em\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e person specific trajectories on the mini-mental state examination (MMSE). \u003c/em\u003e\u003cem\u003e\u003cstrong\u003en\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e estimated person specific slopes on the MMSE (y axis) as predicted by person specific slopes in NPTX2 (y axis) across the total A+ group (yellow indicates CI A+T+, brown MCI A+T+, and red dementia A+T+). \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eo\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e GO biological processes associated with proteins that predicted subsequent changes over time in MMSE scores.\u003c/em\u003e\u003cem\u003e\u003cstrong\u003e \u003c/strong\u003e\u003c/em\u003e\u003cem\u003eIn all trajectory plots * indicates that slope significant different from 0 with p \u0026lt;0.05, a indicates that slope differs from controls with p \u0026lt;0.05, see supplemental table 2 and 3 for all statistical details of all 1511 proteins and cognitive scores tested.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Combifigure.png","url":"https://assets-eu.researchsquare.com/files/rs-8544834/v1/185f8b9127b4fbf04344edba.png"},{"id":101943499,"identity":"2f2bd843-7ed5-4b63-b2fc-6c0a0b44ec5d","added_by":"auto","created_at":"2026-02-05 09:42:06","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2122104,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8544834/v1/ea92a5b0-675d-4353-bba8-20ea40d866f3.pdf"},{"id":100355035,"identity":"b5055843-bc89-4905-a56a-dbaaf2f708fd","added_by":"auto","created_at":"2026-01-16 05:31:03","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":4200183,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eSupplementary figure\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Person trajectories and estimated group changes in CSF Aβ42/40 (top); ptau181(middle) and total tau (bottom) levels for controls, pre-amyloid, cognitively intact (CI) with abnormal Aβ42/4 (A+) and normal ptau181 (T-), CI with A+ and abnormal ptau181 (T+), mild cognitive impairment (MCI) with A+T+ and dementia with A+T+, * is slope significant different from 0 with p \u0026lt;0.05, a indicates that slope differs from controls with p \u0026lt;0.05.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"supplementcombi1.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8544834/v1/3b6f834392c180c20afaeffa.pdf"}],"financialInterests":"\u003cb\u003eYes\u003c/b\u003e there is potential Competing Interest.\nBMT received funding from Novo Nordisk for this study, all competing interest can be found in the manuscript file.","formattedTitle":"Defining the natural history of Alzheimer’s disease by longitudinal cerebrospinal fluid proteomics.","fulltext":[{"header":"Main text ","content":"\u003cp\u003eOver 57 million people worldwide have dementia, which is caused by Alzheimer’s disease (AD) in 70% of the cases. AD is biologically defined by amyloid-β (Aβ) plaques and tau tangles, yet the temporal sequence of molecular changes preceding and following these hallmark pathologies remains unclear. Identifying such molecular changes is difficult, as the brain is largely inaccessible for detailed molecular investigations in patients. Cerebrospinal fluid (CSF), which is in close contact with the brain, provides an accessible window into disease-related molecular processes and has been instrumental in establishing Aβ and tau as core AD biomarkers\u003csup\u003e1\u003c/sup\u003e. Proteomic techniques allow simultaneous measurement of thousands of additional proteins in single CSF samples, leading to the identification of consistent alterations in AD relative to controls\u003csup\u003e2–8\u003c/sup\u003e. Longitudinal CSF studies using targeted techniques have demonstrated that it is possible to find early changes related to Aβ and tau.\u003csup\u003e9–11\u003c/sup\u003e However, whether additional CSF protein changes occur before Aβ and tau abnormalities, and how these evolve across the disease continuum remains unclear.\u003c/p\u003e\n\u003cp\u003eHere we measured longitudinal CSF untargeted proteomics to characterise temporally ordered molecular changes across the AD spectrum. We analysed 465 serial CSF samples from 191 individuals, including cognitively intact participants, individuals with mild cognitive impairment, and patients with AD dementia, with up to 15 years follow-up (average 3.4±2.3 years; on average 2.4±0.7 serial samples; table; methods). In parallel, we measured Aβ42/40, phosphorylated tau at threonine-181 (ptau181) and total tau (ttau) to anchor proteomic changes to established biomarkers of AD pathology. This design enabled us to identify protein alterations that preceded, coincided with, or followed the development of amyloid and tau abnormalities.\u003c/p\u003e\n\u003cp\u003eAmong 83 cognitively intact individuals with normal Aβ42/40(A-) and ptau181(T-) markers at baseline, a subset of 11(13.3%) individuals developed abnormal Aβ42/40 levels over 2.8±1.5 years (figure a; supplementary table 1), defining a ‘pre-amyloid’ stage. In these individuals, CSF ttau and ptau181 levels also increased (figure b), reaching abnormal levels with a time lag of 2.6-3.3 years relative to Aβ42/40 abnormality (figure b, c; supplemental figure). After that time, ptau181 levels in the pre-amyloid group were at the same height of cognitively intact A+T+ individuals. In contrast, tau markers increased much slower in the group of cognitively intact individuals with A+T- at baseline relative to the pre-amyloid individuals and were projected to develop abnormal ptau181 levels after ~11.9±10.3 years (supplementary figure). This finding corroborates our previous observations\u0026nbsp;that cognitively intact A+T- individuals may represent a biological subtype of AD\u003csup\u003e2,4\u003c/sup\u003e, rather than a ‘pre-tau’ stage. Indeed, comparisons on the CSF proteome also suggested specific alterations in cognitively intact A+T- (supplemental table 1). Accordingly, we took the pre-amyloid, cognitively intact A+T+, mild cognitive impairment A+T+ and dementia A+T+ groups as representing the typical AD disease course for the remainder of this study. We estimated that progression to each subsequent disease stage took approximately 5-6 years, consistent with previous estimates\u003csup\u003e12\u003c/sup\u003e. Thus, we defined a ‘concatenated disease time’, in which we estimated for each stage the average trajectory of CSF protein level changes over 6-year time windows: starting in the pre-amyloid stage, followed by intact cognition with A+T+, mild cognitive impairment with A+T+ and AD dementia A+T+. Using this framework, we examined longitudinal CSF proteomic changes relative to the emergence of amyloid and tau abnormalities and across subsequent disease stages.\u003c/p\u003e\n\u003cp\u003eWe assessed longitudinal CSF changes for 1506 of the 3203 proteins detected, selected for having \u0026gt;400 observations across all 465 samples. At baseline, CSF levels of 459 (30.5%) proteins differed in any of the groups relative to controls, and 615 (40.8%) proteins had significant changes over time in at least one of the groups. Full statistics for all proteins and groups are presented in supplementary table 1.\u003c/p\u003e\n\u003cp\u003eWe identified a set of 55 proteins with longitudinal CSF changes that emerged during the pre-amyloid stage and persisted across disease progression. Early alterations included increasing levels of SMOC1 and YWHAZ and other proteins that were enriched for glycolytic processes (figure d, e), as well as decreasing levels of proteins including synaptic protein NPTX2 and interneuron marker SST (figure e). Together, the 43 increasing proteins were enriched for the upstream transcription factor PBX3 that has DNA binding activity and is involved in nervous system development\u003csup\u003e13\u003c/sup\u003e (supplementary table 3). Several of these proteins have been implicated in cross-sectional CSF mass spectrometry studies\u003csup\u003e4,6,14\u003c/sup\u003e, targeted longitudinal studies\u003csup\u003e10\u003c/sup\u003e, and in cross-sectionally modelled disease-time estimates in autosomal dominant AD\u003csup\u003e15\u003c/sup\u003e and Down syndrome\u003csup\u003e16\u003c/sup\u003e. While most proteins in our study had consistent timings of change as previously estimated for SMOC1 and YWHAZ\u003csup\u003e15\u003c/sup\u003e , others, NPTX2 and NPTXR, occurred earlier in our longitudinal measurements. Our longitudinal analyses demonstrate that these alterations arise years before conventional biomarker positivity in sporadic AD and remain stable across advancing disease stages. The persistent directional changes along the AD disease course suggest sustained engagement of metabolic and synaptic processes early in AD pathogenesis.\u003c/p\u003e\n\u003cp\u003eWe then set out to understand the ordering of proteomic changes, and identified distinct proteomic signatures that predicted subsequent change in Aβ42/40 and/or ptau181 levels over time (‘upstream proteins’), or changed alongside Aβ42/40 and/or ptau181, or had longitudinal changes predicted by baseline Aβ42/40 and/or ptau181 levels (‘downstream proteins’; figure f). Higher baseline levels of proteins that were enriched for humoral immune response that predicted steeper decreases in Aβ42/40 (figure h), and this included UCHL1, which plays a role in clearing misfolded proteins\u003csup\u003e17\u003c/sup\u003e. Baseline levels of SMOC1 and ALDOA on the other hand predicted subsequent increases in ptau181 levels, together with another 109 proteins that were enriched for synapse organisation, axon development, glycolytic processes, heparan sulfate proteoglycan metabolism, and gliogenesis, of which most started changing in later disease stages. Longitudinal changes of 5 proteins, including NUTF2 (a nuclear transport factor), occurred alongside increasing ptau181 levels.\u003c/p\u003e\n\u003cp\u003eFollowing the development of Aβ42/40 and ptau181 abnormality, we observed a distinct pattern of proteomic change characterised by progressive decreases in CSF protein levels with disease worsening. In total, 185 proteins declined across later disease stages and were enriched for synapse organisation, axonal development and cell adhesion (figure k, l). This group included proteins VGF and NRXN1, of which lower CSF levels in dementia compared to controls have been reported in previous cross-sectional\u003csup\u003e2,15,16\u003c/sup\u003e and longitudinal\u003csup\u003e10\u003c/sup\u003e mass spectrometry studies. We also observed increasing levels of ACHE in the dementia stage. \u0026nbsp;ACHE breaks down acetylcholine in the synapse and might indicate that dysregulated cholinergic signalling is a late event, consistent with a previous postmortem study.\u003csup\u003e18\u003c/sup\u003e These findings indicate that synaptic and axonal dysfunction emerge downstream of Aβ42/40 and ptau181 abnormality and exacerbates with clinical progression.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo assess the clinical relevance of these proteomic changes, we examined their relationship with cognitive decline. To increase statistical power, we combined control and pre-amyloid group into A- and tested decline on the memory delayed recall scores, as this test is sensitive to very early cognitive alterations\u003csup\u003e11\u003c/sup\u003e. Baseline levels of 112 proteins predicted steeper decline on delayed recall, including YWHAZ (figure j) and C3. These proteins were enriched for extracellular matrix organisation and innate immune activation (figure h). Of note, none of proteins with consistent changes over concatenated disease time changed simultaneously with delayed recall scores. In later disease stages, we tested global cognitive decline measured by the Mini-Mental State Examination (MMSE) across all A+T+ participants. Higher baseline levels of 9 proteins including YWHAZ and PGAM1, and lower levels of 56 proteins including SCG2 and NPTX2 predicted steeper decline on the MMSE. These proteins were enriched for cell adhesion, neuron and glia development (figure o). Of these several synaptic and axonal proteins, including NPTX2 and EPHA10, tracked longitudinally with worsening cognitive performance measured by the MMSE. These associations suggest that late-stage proteomic changes reflect neurobiological processes that are closely linked to cognitive deterioration and may serve as proxy markers of disease progression.\u003c/p\u003e\n\u003cp\u003eA limitation of this study is that within each stage samples sizes are modest, reflecting the challenges in obtaining repeated lumbar punctures. This primarily reduced power for disease stage-specific analyses with cognitive decline, which is inherently heterogenous in AD\u003csup\u003e19\u003c/sup\u003e . Nevertheless, our total sample size is large, and our results recapitulate previous CSF proteomic studies that used cross-sectional designs\u003csup\u003e2–8\u003c/sup\u003e or modelled disease-time in autosomal dominant AD and Down syndrome\u003csup\u003e15,16\u003c/sup\u003e, supporting the validity and robustness of our proteomic approach.\u003c/p\u003e\n\u003cp\u003eTogether, these results define the molecular natural history of sporadic AD across ~25 years, spanning from pre-amyloid stages to dementia, using serial CSF proteomics. We identified early biomarker candidates beyond Aβ and tau that change years before established biomarker abnormality and delineate downstream proteomic signatures associated with synaptic dysfunction and cognitive decline. By providing temporally ordered proteomic signatures and quantified effect size estimates, this work informs clinical trial design, supports biomarker-guided patient stratification, and highlights a window for earlier therapeutic intervention in sporadic AD.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cem\u003eStudy participants\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eWe selected 246 individuals from the Amsterdam Dementia Cohort\u003csup\u003e20\u003c/sup\u003e or the EMIF-AD preclinAD study\u003csup\u003e21\u003c/sup\u003e when they had longitudinally collected repeated CSF samples (in total 628 samples) available in our biobank and were on their first visit cognitively intact (CI) with normal Aβ42/40 (A-) and ptau181 levels (T-), or at least A+ based on CSF (n=190) or PET (n=1). The present analyses were performed on proteomic results from n=191 with 465 samples, excluding small groups or related individuals (3 MCI A+T-, 3 Dementia A+T- and 3 NC A-T+, 2 MCI and 2 dementia individuals who upon remeasuring Aβ42/40 had an A- status, and n=42 co-twins). MCI and AD type dementia diagnoses were determined at our memory clinic during multidisciplinary consensus meetings based on international consensus guidelines\u003csup\u003e20\u003c/sup\u003e. Most individuals had repeated cognitive testing around the time of CSF sampling\u003csup\u003e20,21\u003c/sup\u003e. We used the MMSE score as a measure of global cognitive performance in individuals with A+. In the A- group we used the delayed recall score of the Dutch Auditory Verbal Learning Task (AVLT, possible score range 1-15), as we previously found this measure to be most sensitive to cognitive decline in very early stages of AD\u003csup\u003e11,22\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eAll participants provided written informed consent to use their biomaterial and clinical data for scientific research. All studies were approved by the medical ethics committee of the Amsterdam UMC in the Netherlands, as well as the Western Norway regional committee for medical and health research ethics.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCSF collection and targeted measurements\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eCSF samples were obtained by lumbar puncture between the L3/L4, L4/L5 or L5/S1 intervertebral space with a 25-gauge needle and syringe and collected in polypropylene tubes. For all participants CSF sample collection, processing and biobank storage at the Alzheimer center biobank at the department of Laboratory Medicine was performed according to international guidelines\u003csup\u003e23\u003c/sup\u003e. The 628 samples were randomised over 96-well plates, keeping repeated samples from the same individual next to each other and randomising on Aβ, tau and clinical status within blocks of 14 samples (see next section on TMT measurements). We used random sampling as implemented in R to determine the layouts and with 100 μl CSF in each well and stored at −80 °C. Then Aβ 1-42, Aβ 1-40, phosphorylated tau at threonine 181 and total tau were measured in each sample in singlicate using the fully automated CLEIA on the LUMIPULSE® G System (LUMIPULSE® G600II, REF: 703380; Fujirebio Diagnostics, Inc.) according to the manufacturer’s instructions at the Neurochemistry Laboratory at the Amsterdam University Medical Center, the Netherlands. Part of those data were reported before\u003csup\u003e24\u003c/sup\u003e. Technicians were blinded to diagnosis. Five individuals had too low CSF volume at baseline to remeasure Aβ as well as performing mass spec, for which we prioritized mass spec and used their historical measures based on Innotest (n=1), Euroimmune (n=3) or amyloid PET (n=1) to determine baseline A status. We used\u0026nbsp;for Aβ42/40 a cutoff of \u0026lt;0.071 pg/ml to determine amyloid abnormality\u003csup\u003e25\u003c/sup\u003e , \u0026nbsp;and for ptau181 \u0026gt;61.5 pg/ml to determine tau abnormality (unpublished based on same cohort as\u003csup\u003e25\u003c/sup\u003e). Finally, a few samples had ttau levels \u0026gt;2000 pg/ml, which was above the upper limit of detection, and we imputed these values based on individuals ptau181 levels that correlated strongly (r=.68 before and r=.92 after removal of technical outliers, p-value \u0026lt;0.001).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eTMT mass spec measurements\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eProteomic analyses were performed using tandem mass tag (TMT) mass spectrometry as previously described². CSF protein concentrations were measured by BCA assay, and 25 μg of protein per sample was aliquoted into 96-well Lo-Bind plates, frozen on dry ice, lyophilized, and stored at −80 °C. Proteins were solubilized in 8 M urea/20 mM methylamine, reduced with 10 mM DTT, alkylated with 25 mM iodoacetamide, and quenched with DTT. After dilution to 1 M urea, samples were digested overnight at 37 °C with trypsin. Peptides were acidified, desalted, and lyophilized. Control samples were resuspended for LC-MS analysis; all others were resuspended in HEPES buffer for TMT labeling. A total of 644 samples were labeled across 46 experiments (14 samples + 2 references per set), combined, desalted, and fractionated by high-pH reverse-phase HPLC. Ten fractions per set were collected, lyophilized, resuspended, and peptide concentrations were measured using Nanodrop before LC–MS/MS analysis, and 500 ng of peptides was injected for LC-MS/MS analysis\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eLC–MS/MS\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe acquisition of experiments was randomized to avoid bias. The peptide mixture was measured on an Ultimate 3000 RSLC system (Thermo Fisher Scientific) coupled to an Orbitrap Exploris 480 mass spectrometer with an EASY-Spray nano-ESI source. Peptides were trapped on a PepMap precolumn (2 cm × 75 μm, 3 μm C18) and separated on a 25 cm analytical column (PepMap RSLC, 2 μm C18) using a biphasic ACN gradient (solvent A: 0.1% formic acid in water; solvent B: 100% ACN) at 250 nl/min. The gradient was: 5% B for trapping (5 min), 5–7% B over 0.5 min, 7–25% B over 76.5 min, 25–38% B over 15 min, 38–85% B over 3 min, followed by 7 min at 85% B and 10 min at 5% B for column conditioning. MS acquisition was performed in FAIMS-enabled data-dependent mode with two compensation voltages (−50 V and −70 V). Full MS scans were acquired at 60,000 resolution (400–1,600 m/z), AGC target 3 × 10⁶, followed by HCD fragmentation (NCE 32%) and MS/MS scans at 45,000 resolution. Dynamic exclusion was set to 30 s. Raw files were processed in Proteome Discoverer 2.5 using Sequest HT against Swiss-Prot (July 2023). Search parameters included 10 ppm precursor tolerance, 0.02 Da fragment tolerance, static TMTpro (+304.207 Da on N-terminus and K), carbamidomethyl (C), and dynamic methionine oxidation. Up to two missed cleavages were allowed, minimum peptide length 6. PSM FDR was controlled at 1% (strict) using target-decoy. Reporter ion quantification was intensity-based with coisolation threshold 50 and S/N ≥10.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003ePostprocessing TMT mass spec data\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eTechnical deviations may have influenced protein abundance across the TMT experiments. Before the statistical analyses, we normalized protein abundance according to the internal reference scaling normalization procedure 100 for TMT proteomics data that use the common pool reference channels to normalize values between plex experiments\u003csup\u003e26\u003c/sup\u003e, adapted to scale according to the median instead of the total sum to reduce the influence of outliers. Briefly, the first step in this two-step approach normalized the grand total protein intensities for each of the 14 channels within an experiment to match these to the two reference channels. In the second step, a correction factor was calculated based on common pooled internal standards to normalize reporter ion intensities of proteins between TMT experiments. Next, protein values were log 2-transformed and then scaled according to the mean and standard deviation of the baseline measurement in the control group, so that positive and negative values indicate higher and lower than normal. For all proteins, we report gene names to aid comparisons with other AD subtyping literature using either proteomics or RNA-seq data. Upon visual quality control, some experiments had low levels for specific proteins, which are flagged in the results file (supplemental table 1).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eStatistical analyses\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eWe compared groups on descriptive characteristics with chi-squared tests for discrete variables (sex and APOE e4 genotype), and with linear regression models for continuous outcomes. Next, we tested for each protein changes over time using linear mixed models with repeated protein levels as outcome measure, time and group as main predictors as well as an interaction terms of time X group, age and sex as covariates. Not all models converged when including both random slopes and intercepts, and so for reasons of parsimony ran these models including random intercepts only. Differences between group at baseline, group specific slopes and group differences between slopes were extracted from each model with the emmeans package (v1.11.1). Next, we performed generalised linear mixed-effects models (glmm) using a Markov Chain Monte Carlo (MCMC) approach as implemented in the MCMCglmm package (v2.36)\u003csup\u003e27\u003c/sup\u003e to test which proteins changed together with Aβ42/40, ptau181 and ttau, as well as with delayed memory recall test and MMSE scores over time. We used uninformed priors, an effective sample size of 1000, thinning of 100, and a burn-in of 15000 (resulting in 115000 iterations per protein). A strength of this approach is that it can simultaneously model changes over time in multiple outcomes (e.g., Aβ42/40 levels or neuropsychological test scores together with other CSF protein levels). Thus, with this model correlations of the random effects (i.e., intercepts with each other, intercepts of one variable with slope of the other variable, and between slopes of both variable) can be easily inferred from the sampled posterior distributions, which is not possible for a standard linear mixed model that gives single parameter estimates based on likelihood maximization. For the main analyses, our aim was to identify patterns of coherent changes over time across concatenated disease time. As such, we used a liberal p value threshold of p\u0026lt;0.05 to determine significance as well as consistent changes over different disease stages. Please note that all results are reported in the supplement. We used the Gene Ontology database (GO; release 2025-10-10) through Panther v19.0 to study if proteins with consistent changes were related to specific biological processes and report associated pathways when pFDR\u0026lt;0.05, and ENRICHR to identify potential upstream transcription factors associated with CSF proteomic changes with pFDR\u0026lt;0.05. All analyses were performed in R v4.5.0 “How About a Twenty-Six”.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003eAcknowledgements\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBMT received support for this work from ZonMW VIDI (#09150171910068),\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ethe European Union (ERC, DecipherAD, #101171721). Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Council. Neither the European Union nor the granting authority can be held responsible for them, TAP-dementia (www.tap-dementia.nl) that is funded by ZonMW (no. 10510032120003), and the Amsterdam Cohort Hub, as part of the Sector Plan 'Accelerating Health' of the Dutch Ministry of Education, Culture and Science, and Novo Nordisk.\u003c/p\u003e\n\u003cp\u003eMass spectrometry-based proteomic analyses were performed by the Proteomics Unit at the University of Bergen (PROBE; KFW, OM, FB). This facility is a member of the National Network of Advanced Proteomics Infrastructure (NAPI), which is funded by the Reseearch Council of Norway (INFRASTRUKTUR-program project number: 295910).\u003c/p\u003e\n\u003cp\u003eResearch of AH is funded by Stichting Alzheimer Nederland (WE.06-2021-06), the Alzheimer Drug Discovery Foundation (CANTATE project), Stichting Steun Alzheimer Centrum Amsterdam and the Davos Alzheimer Collaborative. She is a recepient of the TAP dementia fund (the Dutch Research Counsil, ZonMW, no. 10510032120003).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eResearch of FD is funded by ZonMW, Alzheimer Nederland (WE.03-2018-10), Stichting Steun Alzheimercentrum Amsterdam, and Davos Alzheimer Collaborative (AccDx program).\u003c/p\u003e\n\u003cp\u003eCT receives research support by the European Commission (MarieCurie International Training Network, Grant Agreement No. 860197(MIRIADE) and No. 101119596 (TAME), Innovative Medicines Initia-tives 3TR (Horizon 2020, grant 831434) EPND (IMI 2 Joint Under-taking (JU), grant 101034344), and JPND (bPRIDE, CCAD), EuropeanPartnership on Metrology, co-financed by the European Union’s Hori-zon Europe Research and Innovation Programme and by the Participating States ((22HLT07 NEuroBioStand), Horizon Europe (PREDICTFTD,101156175, CCAD), CANTATE project funded by the Alzheimer Drug Discovery Foundation, Alzheimer’s Association (grant SG-22-856131-SABB NEXT), Michael J. Fox Foundation,Health Holland, the Dutch Research Council (ZonMW), AlzheimerDrug Discovery Foundation, Selfridges Group Foundation, AlzheimerNetherlands. C.T. is recipient of ABOARD, which is a public-privatepartnership receiving funding from ZonMW (No. 73305095007) and Health Holland, Topsector Life Sciences \u0026amp; Health (PPP-allowance;No. LSHM20106). C.T. is recipient of TAP-dementia, a ZonMw fundedproject (No. 10510032120003) in the context of the Dutch NationalDementia Strategy.\u003c/p\u003e\n\u003cp\u003ePJV received support from the European Federation of Pharmaceutical Industries and Associations Innovative Medicines Initiative Joint Undertaking (European Medical Information Framework grant 115372,), Stichting Dioraphte, Alzheimer Nederland (WE. 09-2016-10). The project has received funding from the Innovative Medicines Initiative Joint Undertaking under grant agreement (115952). This Joint Undertaking receives the support from the European Union’s Horizon 2020 research and innovation program and European Federation Pharmaceutical Industries and Associations.\u003c/p\u003e\n\u003cp\u003eResearch of Alzheimer Centre Amsterdam is part of the neurodegeneration research program of Amsterdam Neuroscience. Alzheimer Centre Amsterdam is supported by the Stichting Alzheimer Nederland and Stichting Steun Alzheimercentrum Amsterdam.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCOI\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eM.J. is an employee and minor shareholder of Novo Nordisk A/S.\u003c/p\u003e\n\u003cp\u003eAH was part of the educational advisory board of Eli Lilly and the advisory board of the Brain Research Center.\u003c/p\u003e\n\u003cp\u003eCET has research contracts with Acumen, ADx Neurosciences, AC-Immune, Alamar, Aribio, Axon Neurosciences, Beckman-Coulter, BioConnect, Bioorchestra, Brainstorm Therapeutics, C2N diagnostics, Celgene, Cognition Therapeutics, EIP Pharma, Eisai, Eli Lilly, Fujirebio, Instant Nano Biosensors, Merck, Muna, Nitrase Therapeutics, Novo Nordisk, Olink, PeopleBio, Quanterix, Roche, Sysmex, Toyama, Vaccinex, Vivoryon.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eShe is editor in chief of Alzheimer Research and Therapy, and serves on editorial boards of Molecular Neurodegeneration, Alzheimer’s \u0026amp; Dementia, Neurology: Neuroimmunology \u0026amp; Neuroinflammation, Medidact Neurologie/Springer, and is committee member to define guidelines for Cognitive disturbances, and one for acute Neurology in the Netherlands. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eShe has consultancy/speaker contracts for Aribio, Biogen, Beckman-Coulter, Cognition Therapeutics, Danaher, Eisai, Eli Lilly, Janssen, Merck, Neurogen Biomarking, Nordic Biosciences, Novo Nordisk, Novartis, Olink, Quanterix, Roche, Sanofi and Veravas.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eData availability\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRaw proteomic and clinical data is available upon reasonable request for the purposes of replication.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eJack, C. R. \u003cem\u003eet al.\u003c/em\u003e Revised criteria for diagnosis and staging of Alzheimer\u0026rsquo;s disease: Alzheimer\u0026rsquo;s Association Workgroup. \u003cem\u003eAlzheimer\u0026rsquo;s Dement.\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 5143\u0026ndash;5169 (2024).\u003c/li\u003e\n\u003cli\u003eTijms, B. M. \u003cem\u003eet al.\u003c/em\u003e Cerebrospinal fluid proteomics in patients with Alzheimer\u0026rsquo;s disease reveals five molecular subtypes with distinct genetic risk profiles. \u003cem\u003eNat. Aging\u003c/em\u003e 33\u0026ndash;47 (2024) doi:10.1038/s43587-023-00550-7.\u003c/li\u003e\n\u003cli\u003eTijms, B. M. \u003cem\u003eet al.\u003c/em\u003e Pathophysiological subtypes of Alzheimer\u0026rsquo;s disease based on cerebrospinal fluid proteomics. \u003cem\u003eBrain\u003c/em\u003e \u003cstrong\u003e143\u003c/strong\u003e, 3776\u0026ndash;3792 (2020).\u003c/li\u003e\n\u003cli\u003eVisser, P. J. \u003cem\u003eet al.\u003c/em\u003e Cerebrospinal fluid tau levels are associated with abnormal neuronal plasticity markers in Alzheimer\u0026rsquo;s disease. \u003cem\u003eMol Neurodegener\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, (2022).\u003c/li\u003e\n\u003cli\u003eJohnson, E. C. B. \u003cem\u003eet al.\u003c/em\u003e Large-scale proteomic analysis of Alzheimer\u0026rsquo;s disease brain and cerebrospinal fluid reveals early changes in energy metabolism associated with microglia and astrocyte activation. \u003cem\u003eNature Medicine\u003c/em\u003e 1\u0026ndash;31 (2020) doi:10.1038/s41591-020-0815-6.\u003c/li\u003e\n\u003cli\u003eHigginbotham, L. \u003cem\u003eet al.\u003c/em\u003e Integrated proteomics reveals brain-based cerebrospinal fluid biomarkers in asymptomatic and symptomatic Alzheimer\u0026rsquo;s disease. \u003cem\u003eSci Adv\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, eaaz9360 (2020).\u003c/li\u003e\n\u003cli\u003eGeus, M. B. de \u003cem\u003eet al.\u003c/em\u003e Mass spectrometry in cerebrospinal fluid uncovers association of glycolysis biomarkers with Alzheimer\u0026rsquo;s disease in a large clinical sample. \u003cem\u003eSci. Rep.\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 22406 (2023).\u003c/li\u003e\n\u003cli\u003ePark, S. A. \u003cem\u003eet al.\u003c/em\u003e SWATH-MS analysis of cerebrospinal fluid to generate a robust battery of biomarkers for Alzheimer\u0026rsquo;s disease. \u003cem\u003eScientific Reports\u003c/em\u003e 1\u0026ndash;10 (2020) doi:10.1038/s41598-020-64461-y.\u003c/li\u003e\n\u003cli\u003eWildsmith, K. R. \u003cem\u003eet al.\u003c/em\u003e Identification of longitudinally dynamic biomarkers in Alzheimer\u0026rsquo;s disease cerebrospinal fluid by targeted proteomics. \u003cem\u003eMolecular Neurodegeneration\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 22 (2014).\u003c/li\u003e\n\u003cli\u003eLibiger, O. \u003cem\u003eet al.\u003c/em\u003e Longitudinal CSF proteomics identifies NPTX2 as a prognostic biomarker of Alzheimer\u0026rsquo;s disease. \u003cem\u003eAlzheimer\u0026rsquo;s Dementia\u003c/em\u003e (2021) doi:10.1002/alz.12353.\u003c/li\u003e\n\u003cli\u003eTijms, B. M. \u003cem\u003eet al.\u003c/em\u003e Pre-amyloid stage of Alzheimer\u0026rsquo;s disease in cognitively normal individuals. \u003cem\u003eAnnals of Clinical and Translational Neurology\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 1037\u0026ndash;1047 (2018).\u003c/li\u003e\n\u003cli\u003eVermunt, L. \u003cem\u003eet al.\u003c/em\u003e Duration of preclinical, prodromal, and dementia stages of Alzheimer\u0026rsquo;s disease in relation to age, sex, and APOE genotype. \u003cem\u003eAlzheimer\u0026rsquo;s \u0026amp; Dementia\u003c/em\u003e 1\u0026ndash;11 (2019) doi:10.1016/j.jalz.2019.04.001.\u003c/li\u003e\n\u003cli\u003eDammer, E. B. \u003cem\u003eet al.\u003c/em\u003e Proteomic analysis of Alzheimer\u0026rsquo;s disease cerebrospinal fluid reveals alterations associated with APOE \u0026epsilon;4 and atomoxetine treatment. \u003cem\u003eSci. Transl. Med.\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, eadn3504 (2024).\u003c/li\u003e\n\u003cli\u003eModeste, E. S. \u003cem\u003eet al.\u003c/em\u003e Quantitative proteomics of cerebrospinal fluid from African Americans and Caucasians reveals shared and divergent changes in Alzheimer\u0026rsquo;s disease. \u003cem\u003eMol. Neurodegener.\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 48 (2023).\u003c/li\u003e\n\u003cli\u003eJohnson, E. C. B. \u003cem\u003eet al.\u003c/em\u003e Cerebrospinal fluid proteomics define the natural history of autosomal dominant Alzheimer\u0026rsquo;s disease. \u003cem\u003eNat. Med.\u003c/em\u003e 1\u0026ndash;10 (2023) doi:10.1038/s41591-023-02476-4.\u003c/li\u003e\n\u003cli\u003eMontoliu-Gaya, L. \u003cem\u003eet al.\u003c/em\u003e Proteomic analysis of Down syndrome cerebrospinal fluid compared to late-onset and autosomal dominant Alzheimer\u0026acute;s disease. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 6003 (2025).\u003c/li\u003e\n\u003cli\u003eMi, Z. \u0026amp; Graham, S. H. Role of UCHL1 in the pathogenesis of neurodegenerative diseases and brain injury. \u003cem\u003eAgeing Res. Rev.\u003c/em\u003e \u003cstrong\u003e86\u003c/strong\u003e, 101856 (2023).\u003c/li\u003e\n\u003cli\u003eDavis, K. L. \u003cem\u003eet al.\u003c/em\u003e Cholinergic Markers in Elderly Patients With Early Signs of Alzheimer Disease. \u003cem\u003eJAMA\u003c/em\u003e \u003cstrong\u003e281\u003c/strong\u003e, 1401\u0026ndash;1406 (1999).\u003c/li\u003e\n\u003cli\u003eJutten, R. J. \u003cem\u003eet al.\u003c/em\u003e Finding Treatment Effects in Alzheimer Trials in the Face of Disease Progression Heterogeneity. \u003cem\u003eNeurology\u003c/em\u003e \u003cstrong\u003e96\u003c/strong\u003e, e2673\u0026ndash;e2684 (2021).\u003c/li\u003e\n\u003cli\u003eFlier, W. M. van der \u0026amp; Scheltens, P. Amsterdam Dementia Cohort: Performing Research to Optimize Care. \u003cem\u003eJ. Alzheimer\u0026rsquo;s Dis.\u003c/em\u003e \u003cstrong\u003e62\u003c/strong\u003e, 1091\u0026ndash;1111 (2018).\u003c/li\u003e\n\u003cli\u003eKonijnenberg, E. \u003cem\u003eet al.\u003c/em\u003e The EMIF-AD PreclinAD study: study design and baseline cohort overview. \u003cem\u003eAlzheimer\u0026rsquo;s Research \u0026amp; Therapy\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, S85 (2018).\u003c/li\u003e\n\u003cli\u003eDonohue, M. C. \u003cem\u003eet al.\u003c/em\u003e Association Between Elevated Brain Amyloid and Subsequent Cognitive Decline Among Cognitively Normal Persons. \u003cem\u003eJAMA : the journal of the American Medical Association\u003c/em\u003e \u003cstrong\u003e317\u003c/strong\u003e, 2305\u0026ndash;2316 (2017).\u003c/li\u003e\n\u003cli\u003eCampo, M. del \u003cem\u003eet al.\u003c/em\u003e Recommendations to standardize preanalytical confounding factors in Alzheimer\u0026rsquo;s and Parkinson\u0026rsquo;s disease cerebrospinal fluid biomarkers: an update. \u003cem\u003eBiomarkers in Medicine\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 419\u0026ndash;430 (2012).\u003c/li\u003e\n\u003cli\u003eLeeuw, D. M. de \u003cem\u003eet al.\u003c/em\u003e Cerebrospinal Fluid Amyloid and Tau Biomarker Changes Across the Alzheimer Disease Clinical Spectrum. \u003cem\u003eJAMA Netw. Open\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, e2519919 (2025).\u003c/li\u003e\n\u003cli\u003eWillemse, E. A. J. \u003cem\u003eet al.\u003c/em\u003e Comparing CSF amyloid‐beta biomarker ratios for two automated immunoassays, Elecsys and Lumipulse, with amyloid PET status. \u003cem\u003eAlzheimer\u0026rsquo;s Dementia Diagnosis Assess Dis Monit\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, e12182 (2021).\u003c/li\u003e\n\u003cli\u003ePlubell, D. L. \u003cem\u003eet al.\u003c/em\u003e Extended Multiplexing of Tandem Mass Tags (TMT) Labeling Reveals Age and High Fat Diet Specific Proteome Changes in Mouse Epididymal Adipose Tissue*. \u003cem\u003eMol Cell Proteomics\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 873\u0026ndash;890 (2017).\u003c/li\u003e\n\u003cli\u003eMCMCglmm Course Notes Jarrod Hadfield ([email protected]). 1\u0026ndash;141 (2019).\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1 Participant characteristics at the time of first CSF sampling.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"745\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 2.323%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003eControls\u003c/p\u003e\n \u003cp\u003eA-T-\u003cbr\u003e\u0026nbsp;(n=72)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003ePre-amyloid\u003c/p\u003e\n \u003cp\u003eA-T-\u003cbr\u003e\u0026nbsp;(n=11)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.3727%;\"\u003e\n \u003cp\u003eCognitively intact A+T-\u003cbr\u003e\u0026nbsp; (n=11)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003eCognitively intact A+T+\u003cbr\u003e\u0026nbsp; (n=20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.4182%;\"\u003e\n \u003cp\u003eMCI\u003c/p\u003e\n \u003cp\u003eA+T+\u003cbr\u003e\u0026nbsp; (n=27)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.6004%;\"\u003e\n \u003cp\u003eDementia\u003c/p\u003e\n \u003cp\u003eA+T+\u003cbr\u003e\u0026nbsp; (n=50)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003eCharacteristic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 2.323%;\"\u003e\n \u003cp\u003en missing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.3727%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.4182%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.6004%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003eAge in years\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 2.323%;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e62.96 \u0026plusmn; 7.82\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e63.73 \u0026plusmn; 7.98\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.3727%;\"\u003e\n \u003cp\u003e61 \u0026plusmn; 6.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003e70.42 \u0026plusmn; 8.47\u003csup\u003ea,b,c\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.4182%;\"\u003e\n \u003cp\u003e66 .9\u0026plusmn; 7.3\u003csup\u003ea,c\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.6004%;\"\u003e\n \u003cp\u003e65.5 \u0026plusmn; 7.84\u003csup\u003ed\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003eWomen\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 2.323%;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e30 (41.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e7 (63.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.3727%;\"\u003e\n \u003cp\u003e5 (45.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003e15 (75%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.4182%;\"\u003e\n \u003cp\u003e10 (37.0%)\u003csup\u003ed\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.6004%;\"\u003e\n \u003cp\u003e26 (52.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003eEducation level\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 2.323%;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e5.14 \u0026plusmn; 1.59\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e4.73 \u0026plusmn; 1.68\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.3727%;\"\u003e\n \u003cp\u003e5.64 \u0026plusmn; 1.36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003e4.79 \u0026plusmn; 1.93\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.4182%;\"\u003e\n \u003cp\u003e5.25 \u0026plusmn; 1.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.6004%;\"\u003e\n \u003cp\u003e5.24 \u0026plusmn; 1.18\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003e\u0026gt;=1 APOE e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 2.323%;\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e19 (26.76%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e8 (72.73%)\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.3727%;\"\u003e\n \u003cp\u003e8 (72.73%)\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003e13 (65%)\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.4182%;\"\u003e\n \u003cp\u003e22 (84.6%)\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.6004%;\"\u003e\n \u003cp\u003e33 (67.35%)\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003eMMSE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 2.323%;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e28.73 \u0026plusmn; 1.55\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e28.91 \u0026plusmn; 1.45\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.3727%;\"\u003e\n \u003cp\u003e28.64 \u0026plusmn; 1.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003e28.45 \u0026plusmn; 1.61\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.4182%;\"\u003e\n \u003cp\u003e25.6 \u0026plusmn; 2.5\u003csup\u003ea,b,c,d\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.6004%;\"\u003e\n \u003cp\u003e22.5 \u0026plusmn; 4.21\u003csup\u003ea,b,c,d,e\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003eAVLT Delayed recall\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 2.323%;\"\u003e\n \u003cp\u003e25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e8.62 \u0026plusmn; 2.97\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e10 \u0026plusmn; 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.3727%;\"\u003e\n \u003cp\u003e8.27 \u0026plusmn; 2.72\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003e7.16 \u0026plusmn; 2.14\u003csup\u003ea,b\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.4182%;\"\u003e\n \u003cp\u003e2.5 \u0026plusmn; 2.0\u003csup\u003ea,b,c,d\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.6004%;\"\u003e\n \u003cp\u003e2.39 \u0026plusmn; 2.66\u003csup\u003ea,b,c,d\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003eA\u0026beta;42/40 ratio\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 2.323%;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e0.1 \u0026plusmn; 0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e0.08 \u0026plusmn; 0.01\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.3727%;\"\u003e\n \u003cp\u003e0.06 \u0026plusmn; 0.01\u003csup\u003ea,b\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003e0.04 \u0026plusmn; 0.01\u003csup\u003ea,b,c\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.4182%;\"\u003e\n \u003cp\u003e0.04 \u0026plusmn; 0.01\u003csup\u003ea,b,c\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.6004%;\"\u003e\n \u003cp\u003e0.04 \u0026plusmn; 0.01\u003csup\u003ea,b,c\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003ePtau181 pg/ml\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 2.323%;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e36.53 \u0026plusmn; 11.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e45.76 \u0026plusmn; 12.16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.3727%;\"\u003e\n \u003cp\u003e40.96 \u0026plusmn; 12.96\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003e101.83 \u0026plusmn; 35.97\u003csup\u003ea,b,c\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.4182%;\"\u003e\n \u003cp\u003e125.8 \u0026plusmn; 43.4\u003csup\u003ea,b,c,d\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.6004%;\"\u003e\n \u003cp\u003e134.76 \u0026plusmn; 52.29\u003csup\u003ea,b,c,d\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003eTtau pg/ml\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 2.323%;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e330.08 \u0026plusmn; 107.47\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e387.44 \u0026plusmn; 107.84\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.3727%;\"\u003e\n \u003cp\u003e308.27 \u0026plusmn; 93.34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003e721.95 \u0026plusmn; 350.82\u003csup\u003ea,b,c\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.4182%;\"\u003e\n \u003cp\u003e786.0 \u0026plusmn; 237.3\u003csup\u003ea,b,c\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.6004%;\"\u003e\n \u003cp\u003e853.69 \u0026plusmn; 306.21\u003csup\u003ea,b,c,d\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003eNumber of repeated CSF samples\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 2.323%;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e2.68 \u0026plusmn; 0.82\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e2.82 \u0026plusmn; 0.87\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.3727%;\"\u003e\n \u003cp\u003e2.36 \u0026plusmn; 0.67\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003e2.65 \u0026plusmn; 0.81\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.4182%;\"\u003e\n \u003cp\u003e2.2 \u0026plusmn; 0.4\u003csup\u003ea,b,d\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.6004%;\"\u003e\n \u003cp\u003e2.08 \u0026plusmn; 0.27\u003csup\u003ea,b,d\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003eAverge time between CSF sampling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 2.323%;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e2.61 \u0026plusmn; 1.24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e3.47 \u0026plusmn; 2.63\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.3727%;\"\u003e\n \u003cp\u003e3.41 \u0026plusmn; 2.46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003e2.35 \u0026plusmn; 0.89\u003csup\u003eb,c\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.4182%;\"\u003e\n \u003cp\u003e2.4 \u0026plusmn; 1.6\u003csup\u003eb,c\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.6004%;\"\u003e\n \u003cp\u003e1.61 \u0026plusmn; 0.91\u003csup\u003ea,b,c,d,e\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003eMaximum years repeated CSF sampling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 2.323%;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e4.18 \u0026plusmn; 2.18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.9172%;\"\u003e\n \u003cp\u003e5.69 \u0026plusmn; 3.75 a\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.3727%;\"\u003e\n \u003cp\u003e4.22 \u0026plusmn; 2.47\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 4.7826%;\"\u003e\n \u003cp\u003e3.78 \u0026plusmn; 1.88\u003csup\u003eb\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.4182%;\"\u003e\n \u003cp\u003e2.7 \u0026plusmn; 1.8\u003csup\u003ea,b,c\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.6004%;\"\u003e\n \u003cp\u003e1.68 \u0026plusmn; 0.89\u003csup\u003ea,b,c,d,e\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"10\" valign=\"bottom\" style=\"width: 33.8884%;\"\u003e\n \u003cp\u003eA- is normal amyloid beta 1-42/1-40 (A\u0026beta;42/40) levels, A+ abnormal A\u0026beta;42/40 levels, T- is normal tau phosphorylated at threonine 181 (ptau181) levels, T+ is abnormal ptau181 levels, MCI is mild cognitive impairment, APOE is apolipoprotein, MMSE is mini mental state examination, AVLT is Auditory Verbal Learning Test, CSF is cerebrospinal fluid.\u003c/p\u003e\n \u003cp\u003eGroups were compared using pairwise linear regression tests for continuous variables, or by chi2 tests for categorical measures\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"10\" valign=\"bottom\" style=\"width: 33.8884%;\"\u003e\n \u003cp\u003ea differs from controls with p\u0026lt;0.05, b differs from pre-amyloid with p\u0026lt;.05, c differs from cognitively intact A+T- with p\u0026lt;0.05, d differs from cognitively intact A+T+ with p\u0026lt;.05, e differs from mild cognitive impairment A+T+ \u0026nbsp;with p\u0026lt;.05.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"10\" valign=\"bottom\" style=\"width: 33.8884%;\"\u003e\n \u003cp\u003eNot enough volume to re-measure amyloid, ptau181, and ttau for n=5: for these individuals their amyloid and/or ptau status at first visit was determined based on amyloid PET for n=1 and on historical Euroimmune measures for n=4.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"7\" valign=\"bottom\" style=\"width: 25.8262%;\"\u003e\n \u003cp\u003eContinuous measures are reported in mean \u0026plusmn; standard deviation, discrete variables in n (%).\u003c/p\u003e\n \u003cp\u003eProteomic results for the cognitively intact A+T- group are reported in supplemental table 1.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 4.646%;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 3.4162%;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8544834/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8544834/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Defining the molecular natural history of Alzheimer’s disease (AD) is essential for earlier diagnosis and effective therapeutical interventions. Using longitudinal cerebrospinal fluid proteomics from 191 individuals across the AD spectrum, with up to 15 years of follow-up, we identified protein changes that preceded and tracked with development of amyloid-β (Aβ) and/or p-tau181 abnormality over 2.8–6.1 years. Early alterations included SMOC1 and YWHAZ and were enriched for glycolytic, synaptic, and axonal guidance pathways, which remained consistently altered across advancing disease stages. After Aβ and p-tau181 abnormality, 185 proteins progressively decreased with disease worsening, of which a subset of synaptic and axonal proteins including NPTX2, EPHA10 also tracked cognitive decline. Together, our findings can support clinical trial design through accurate effect size estimates, and enable biomarker guided therapeutic targeting across the full pre-amyloid-to-dementia disease continuum.","manuscriptTitle":"Defining the natural history of Alzheimer’s disease by longitudinal cerebrospinal fluid proteomics.","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-16 05:30:58","doi":"10.21203/rs.3.rs-8544834/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-medicine","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"nm","sideBox":"Learn more about [Nature Medicine](http://www.nature.com/nm/)","snPcode":"","submissionUrl":"","title":"Nature Medicine","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Research","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"94e36e08-30f5-48a4-9b51-b6a75f3105df","owner":[],"postedDate":"January 16th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":61163823,"name":"Health sciences/Neurology/Neurological disorders/Neurodegenerative diseases/Alzheimer's disease"},{"id":61163824,"name":"Health sciences/Biomarkers/Diagnostic markers"}],"tags":[],"updatedAt":"2026-05-08T16:52:50+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-16 05:30:58","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8544834","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8544834","identity":"rs-8544834","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}