Multiomic markers of metabolic transitions in age-driven cognitive decline with associated intervention modalities

Abstract

The disclosure relates to compositions and methods for diagnosing, stratifying, prognosing, treating and preventing and cognitive impairment, including Alzheimer's Disease, based on the identification of disease-associated metabolic transitions.

Description

MULTIOMIC MARKERS OF METABOLIC TRANSITIONS IN AGE-DRIVEN COGNITIVE DECLINE WITH ASSOCIATED INTERVENTION MODALITIES CROSS REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to United States Provisional Patent Application No. 63 / 672,202, filed July 16, 2024, which is incorporated herein by reference in their entirety. FIELD OF INVENTION

[0002] The disclosure relates to compositions and methods for diagnosing, stratifying, prognosing, treating and preventing cognitive impairment, including Alzheimer’s Disease, based on the identification of disease-associated metabolic transitions and therapeutic targets. BACKGROUND

[0003] Alzheimer’s disease (AD) was first described by Alois Alzheimer in 1907, who identified hallmark features including amyloid plaques, neurofibrillary tangles, and lipid inclusions— structures he termed “adipose saccules” and which are now recognized as lipid droplets (LDs). Today, age and APOE genotype, particularly the APOE4 allele, remain the most robust risk factors for late-onset AD2. Genome-wide association studies (GWAS) have expanded this picture, implicating genes involved in lipid metabolism, glial clearance, and endolysosomal trafficking and processes involving processing of cellular debris such as phagocytosis, cholesterol metabolism, and innate immunity3. Yet a unified mechanistic model linking these processes to progressive neurodegeneration remains lacking.

[0004] Presently, anti-amyloid therapies with ~35% effect size equate to extending one’s dementia-free life by ~6 months81. Improved, personalized treatments are predicated on increased understanding of disease etiology. Clearly, there is a need for new methods for diagnosing and treating Alzheimer’s disease and other cognitive impairments. The present disclosure meets this need by providing methods for diagnosing, stratifying, and prognosing cognitive impairment, including Alzheimer’s Disease, based on the identification of disease- associated metabolic transitions, as well as methods for treating and preventing cognitive impairment, including Alzheimer’s Disease, by modulating specific therapeutic targets.SUMMARY OF THE INVENTION

[0005] In embodiments, the disclosure provides a method for determining whether a subject is at risk of developing mild cognitive impairment (MCI) or Alzheimer’s Disease (AD), for determining whether the subject has MCI or AD, or for characterizing the cellular cholesterol dyshomeostasis status of the subject, the method comprising determining whether one or more biological samples obtained from the subject comprises: (a) one or more genetic mutations indicative of a predisposition to ANLS attenuation and / or microglial flux capacity saturation; (b) an increased or decreased amount or concentration of one or more cerebral spinal fluid (CSF) proteins indicative of a predisposition to astrocyte-neuron lactate shuttle (ANLS) attenuation; (c) an increased or decreased amount or concentration of one or more plasma lipids indicative of a predisposition to ANLS attenuation; (d) an increased or decreased amount of one or more CSF proteins indicative of ANLS attenuation; (e) an increased or decreased amount or concentration of one or more plasma lipids indicative of ANLS attenuation; (f) an increased or decreased amount or concentration of one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation; (g) an increased or decreased amount or concentration of one or more CSF proteins indicative of microglial flux capacity saturation; and / or (h) an increased or decreased amount or concentration of one or more plasma lipids indicative of microglial flux capacity saturation, wherein: (i) the increased or decreased amount or concentration of the one or more CSF proteins of (b), (d) and / or (g); (ii) the increased or decreased amount or concentration of the one or more plasma lipids of (c), (e), (f) and / or (h); and / or (iii) the presence of the one or more genetic mutations of (a) is indicative of whether the subject is at risk of developing or has mild cognitive impairment (MCI) or Alzheimer’s Disease (AD) and / or indicates the cholesterol dyshomeostasis status of the subject, wherein the cholesterol dyshomeostasis status is selected from ANLS attenuation and / or microglial flux capacity saturation.

[0006] In embodiments, determining the increased or decreased amount or concentration of the one or more CSF proteins and / or plasma lipid markers comprises: (a) determining an amount or concentration of the one or more of the CSF proteins and / or theone or more plasma lipids; and (b) comparing the amount or concentration of the one or more CSF proteins and / or the one or more plasma lipids determined in (a) to a reference value(s) or to an amount or concentration of the one or more CSF proteins and / or the one or more plasma lipids determined in a control sample(s), wherein an increased or decreased amount or concentration of at least one of the one or more CSF proteins and / or at least one of the one or more plasma lipids in the biological sample as compared to the reference value or the amount or concentration determined in the control sample indicates the subject is at risk of developing or has MCI or AD and / or indicates the cholesterol dyshomeostasis status of the subject.

[0007] In embodiments, (a) the one or more genetic mutations indicative of a predisposition to ANLS attenuation are selected from mutations in INPP5D, PICALM, TSPOAP, TNIP1, and / or LXRA genes; (b) the one or more CSF proteins indicative of a predisposition to ANLS attenuation are selected from PLA2G7 and LPCAT2; (c) the one or more plasma lipids indicative of a predisposition to ANLS attenuation are plasma lipids indicating high DHA in phospholipid or plasmalogen form; (d) the one or more CSF proteins indicative of ANLS attenuation are selected from the CSF proteins listed in Table 1, and the group consisting of CHGA, PPIH, ARL3, EIF1B, KIFBP, WFDC10A, AKR1D1, TCEAL8, CYB5R2, NDUFA5, RAB5B, ALDH1A2, IMPACT, NREP, DSG4, XAGE2, RAB7A, CRYL1, GYG2, INPP5A, TOM1, RAC1, FGF1, SLC27A2, RAB21, LDHB, LDHC, ALDOA, MDHC, KPYM, AATC, CNTN2, and TNR; (e) the one or more plasma lipids indicative of ANLS attenuation are selected from the plasma lipids listed in Table 1, and the group consisting of CER.D19.1.18.0 and CER.D19.1.16.0; (f) the one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation are selected from the group consisting of DHA, omega-3 fatty acids, optionally in the form of phospholipids or plasmalogens; (g) the one or more CSF proteins indicative of microglial flux capacity saturation are selected from the CSF proteins listed in Table 2, and the group consisting of TAU, CD44, CH3L1, GFAP, YWHAB, YWHAZ, and GAS6; and / or (h) the one or more plasma lipids indicative of microglial flux capacity saturation are selectedfrom the plasma lipids listed in Table 2, and the group consisting of DHA / EPA acylcarnities, optionally AC.22.6 or AC.20.5, AC.13.0, AC.15.0, cholesterol esters, optionally CE.22.6, CE.20.5, CE.16.0, or CE.17.0, extra-large HDL lipoprotein cholesterol and / or cholesterol ester, optionally Nightingale assay XL.HDL.C or XL.HDL.CE.

[0008] In embodiments, (a) an increase or decrease in PLA2G7 and / or LPCAT2 is indicative of a predisposition to ANLS attenuation; (b) an increase or decrease in one or more plasma lipid indicating high DHA in phospholipid or plasmalogen form is indicative of a predisposition to ANLS attenuation; (c) an increase or decrease in CHGA, PPIH, ARL3, EIF1B, KIFBP, WFDC10A, AKR1D1, TCEAL8, CYB5R2, NDUFA5, RAB5B, ALDH1A2, IMPACT, NREP, DSG4, XAGE2, RAB7A, CRYL1, GYG2, INPP5A, TOM1, RAC1, FGF1, SLC27A2, RAB21, LDHB, LDHC, ALDOA, MDHC, KPYM, AATC, CNTN2, or TNR, is indicative of ANLS attenuation; (d) an increase or decrease in CER.D19.1.18.0 and / or CER.D19.1.16.0 is indicative of ANLS attenuation; (e) an increase or decrease in one or more omega-3 fatty acid, optionally DHA, is indicative of a predisposition to microglial flux capacity saturation; (f) an increase or decrease in one or more of TAU, CD44, CH3L1, GFAP, YWHAB, YWHAZ, and GAS6 is indicative of microglial flux capacity saturation; and / or (g) an increase or decrease in one or more of DHA / EPA acylcarnities, optionally AC.22.6 or AC.20.5, AC.13.0, AC.15.0, cholesterol esters, optionally CE.22.6, CE.20.5, CE.16.0, or CE.17.0, extra-large HDL lipoprotein cholesterol and / or cholesterol ester, optionally Nightingale assay XL.HDL.C or XL.HDL.CE, is indicative of microglial flux capacity saturation.

[0009] In embodiments, (a) an increase or decrease in PLA2G7 and / or LPCAT2 and an increase or decrease in plasma lipids indicating high DHA in phospholipid or plasmalogen form is indicative of a predisposition to ANLS attenuation; (b) an increase or decrease in one or more of RAB21, RAB7A, LDHB, LDHC, ALDOA, MDHC, KPYM, AATC, CNTN2, TNR, TAU, CD44, CH3L1, GFAP, and / or GAS6 and an increase or decrease in CER.D19.1.18.0 and / or CER.D19.1.16.0 is indicative of ANLSattenuation; (c) an increase or decrease in one or more omega-3 fatty acid, optionally DHA, is indicative of a predisposition to microglial flux capacity saturation; and / or (d) an increase or decrease in one or more of TAU, CD44, CH3L1, GFAP, and / or GAS6 and an increase or decrease in DHA / EPA acylcarnitines (optionally AC.22.6. or AC.20.5.), cholesterol esters (optionally, CE.22.6. or CE.20.5.), and / or extra-large HDL lipoprotein cholesterol and cholesterol ester (optionally XL.HDL.C or XL.HDL.CE) is indicative of microglial flux capacity saturation.

[0010] In embodiments, the one or more genetic mutation is identified by whole-genome sequencing or methylation array analysis.

[0011] In embodiments, the amount of the one or more CSF proteins and / or the one or more plasma lipids is determined by targeted mass spectrometry or aptamer-based proteomics.

[0012] In embodiments, the method further comprises determining cognitive scores and / or longitudinal biomarker trajectories of the subject.

[0013] In embodiments, the method comprises using an algorithm trained to determine whether the subject is at risk of developing or has mild cognitive impairment (MCI) or Alzheimer’s Disease (AD) and / or to determine the cholesterol dyshomeostasis status of the subject.

[0014] In embodiments, if the subject is determined to be at risk of developing MCI or AD or determined to have MCI or AD, treating the subject with a therapeutic agent for the treatment of MCI or AD.

[0015] In embodiments, the disclosure provides a method of treating MCI or Alzheimer’s disease in a subject comprising: (a) determining whether the subject has astrocyte-neuron lactate shuttle (ANLS) attenuation or microglial flux capacity saturation; and (b) administering to the subject a therapeutic treatment, wherein: (i) if the subject has a predisposition for ANLS attenuation or ANLS attenuation, the subject is treated with increased sleep, phosphatidylcholine supplementation, and / or creatine supplementation; and (ii) if the subject has a predisposition for microglial flux capacity saturation or microglial flux capacity saturation, the subject is treated with Omega3 supplementation, plasmalogen supplementation, and / or plasmalogen precursor supplementation.

[0016] In embodiments, the method further comprises serial measurements of an amount or concentration of one or more of CER.D19.1.18.0, CD44, and XL.HDL.C to monitor treatment efficacy, wherein a change in the amount or concentration is indicative of treatment efficacy.

[0017] In embodiments, the disclosure provides A method for determining whether a subject is at risk of developing mild cognitive impairment (MCI) or Alzheimer’s Disease (AD), for determining whether the subject has MCI or AD, or for characterizing the cellular cholesterol dyshomeostasis status of the subject, the method comprising: (a) determining an amount or concentration of: (i) one or more cerebral spinal fluid (CSF) proteins indicative of a predisposition to astrocyte-neuron lactate shuttle (ANLS) attenuation in a CSF sample obtained from the subject; (ii) determining an amount or concentration of one or more plasma lipids indicative of a predisposition to ANLS attenuation in a blood or plasma sample obtained from the subject; (iii) determining an amount or concentration of one or more CSF proteins indicative of ANLS attenuation in a CSF sample obtained from the subject; (iv) determining an amount or concentration of plasma lipids indicative of ANLS attenuation in a blood or plasma sample obtained from the subject; (v) determining an amount or concentration of one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation in a blood or plasma sample obtained from the subject; (vi) determining an amount or concentration of one or more CSF proteins indicative of microglial flux capacity saturation in a CSF sample obtained from the subject; and / or (vii) determining an amount or concentration of one or more plasma lipids indicative of microglial flux capacity saturation in a blood or plasma sample obtained from the subject; and (b) determining whether the subject is at risk of developing mild cognitive impairment (MCI) or Alzheimer’s Disease (AD) or has MCI or AD, or characterizing the cellular cholesterol dyshomeostasis status of the subject, by correlating the amounts determined in (a) with the risk of developing MCI or AD, the presence or absence of MCI or AD, and / or the cellular cholesterol dyshomeostasis status of the subject, wherein the correlating is performed using a classifier generated using amount or concentration data determined for the CSF proteins and / or plasma lipids determined in (a) from a plurality of CSF and / or plasma or blood samplesobtained from healthy subjects and a plurality of CSF and / or plasma or blood samples obtained from subjects with MCI and / or AD.

[0018] In embodiments, the method determines in the subject has MCI or AD.

[0019] In embodiments, the method determines whether the subject has AD associated with ANLS attenuation or microglial flux capacity saturation.

[0020] In embodiments, the method comprises determining an amount or concentration of: (a) one or more CSF proteins selected from the group consisting of: CSF proteins set forth in Table 1 or Table 2, PLA2G7, LPCAT2, RAB21, RAB7A, LDHB, LDHC, ALDOA, MDHC, KPYM, AATC, CNTN2, TNR, TAU, CD44, CH3L1, GFAP, and GAS6; and / or (b) one or more plasma lipids selected from the group consisting of: plasma lipids set forth in Table 1 or Table 2, plasma lipids indicating high DHA in phospholipid or plasmalogen form, CER.D19.1.18.0, CER.D19.1.16.0, DHA, omega-3 fatty acids, optionally in the form of phospholipids or plasmalogens, DHA / EPA acylcarnities, optionally AC.22.6 or AC.20.5, cholesterol esters, optionally CE.22.6 or CE.20.5, and extra-large HDL lipoprotein cholesterol and / or cholesterol ester, optionally Nightingale assay XL.HDL.C or XL.HDL.CE.

[0021] In embodiments, the method comprises determining the amount or concentration of at least 1, at least 2, at least 3, at least 5, or at least 10 of the CSF proteins and determining the amount or concentration of at least 2, at least 3, at least 5, or at least 10 of the serum lipids.

[0022] In embodiments, the disclosure provides a kit comprising: (a) one or more reagents for detecting one or more genetic mutations indicative of a predisposition to ANLS attenuation and / or microglial flux capacity saturation; (b) one or more reagents for detecting an amount or concentration of one or more cerebral spinal fluid (CSF) proteins indicative of a predisposition to astrocyte-neuron lactate shuttle (ANLS) attenuation; (c) one or more reagents for detecting an amount or concentration of one or more plasma lipids indicative of a predisposition to ANLS attenuation; (d) one or more reagents for detecting an amount or concentration of one or more CSF proteins indicative of ANLS attenuation; (e) one or more reagents for detecting an amount or concentration of one or more plasma lipids indicative of ANLS attenuation; (f) one or more reagents for detecting an amount or concentration of one or more plasma lipidsindicative of a predisposition to microglial flux capacity saturation; (g) one or more reagents for detecting an amount or concentration of one or more CSF proteins indicative of microglial flux capacity saturation; and / or (h) one or more reagents for detecting an amount or concentration of one or more plasma lipids indicative of microglial flux capacity saturation.

[0023] In embodiments, the kit further comprises one or more reagents for determining an amount or concentration of one or more control proteins or lipids.

[0024] In embodiments, the kit further comprises instructions for interpreting the biomarker patterns determined when using the kit, optionally to classify a subject being tested as being at risk for MCI or AD, having MCI or AD, and / or for determining whether the MCI or AD is associated with ANLS attenuation or microglial flux capacity saturation.

[0025] In embodiments, the kit further comprises a software module that can be used to classify the subject being tested by receiving biomarker measurements and providing automated classification. In embodiments, the software module comprises an algorithm or classifier, trained on biomarker data obtained from healthy subjects and / or MCI or AD patients.

[0026] In embodiments of the kit, (a) the one or more genetic mutations indicative of a predisposition to ANLS attenuation are selected from mutations in INPP5D, PICALM, TSPOAP, TNIP1, and / or LXRA genes; (b) the one or more CSF proteins indicative of a predisposition to ANLS attenuation are selected from CSF proteins set forth in Table 1, PLA2G7 and LPCAT2; (c) the one or more plasma lipids indicative of a predisposition to ANLS attenuation are selected from plasma lipids set forth in Table 1 and plasma lipids indicating high DHA in phospholipid or plasmalogen form; (d) the one or more CSF proteins indicative of ANLS attenuation are selected from CSF proteins set forth in Table 1 and the group consisting of RAB21, RAB7A, LDHB, LDHC, ALDOA, MDHC, KPYM, AATC, CNTN2, and TNR; (e) the one or more plasma lipids indicative of ANLS attenuation are selected from plasma lipids set forth in Table 1 and the group consisting of CER.D19.1.18.0 and CER.D19.1.16.0; (f) the one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation are selected from plasma lipids set forth in Table 2 and the group consisting of DHA, omega-3 fatty acids, optionally in the form of phospholipids or plasmalogens;(g) the one or more CSF proteins indicative of microglial flux capacity saturation are selected from CSF proteins set forth in Table 2 and the group consisting of TAU, CD44, CH3L1, GFAP, and GAS6; and / or (h) the one or more plasma lipids indicative of microglial flux capacity saturation are selected from plasma lipids set forth in Table 2 and the group consisting of DHA / EPA acylcarnities, optionally AC.22.6 or AC.20.5, and cholesterol esters, optionally CE.22.6 or CE.20.5, extra- large HDL lipoprotein cholesterol and / or cholesterol ester, optionally Nightingale assay XL.HDL.C or XL.HDL.CE. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] FIGs. 1A-1D provide an overview of data selection and inter-omic correlation structure. FIG. 1A provides a schematic of ADNI multiomic datasets used in the study, including plasma lipidomics, CSF mass spectrometry (MS) proteomics, CSF SomaScan proteomics, and plasma lipoprotein profiles. FIG. 1B is a diagram of the selection of 602 participants with paired plasma lipidomic and CSF proteomic data; 362 had additional CSF SomaScan data, and 601 had Nightingale lipoprotein profiles. FIG. 1C shows a distribution of clinical diagnoses across cognitive quartiles based on ADAS13 scores: Q1 = 1–7, Q2 = 8–12, Q3 = 13–18, Q4 = 19–52. CN indicates clinically normal; MCI indicates mild cognitive impairment. FIG. 1D provides heatmaps of lipid–protein correlations within each indicated cognitive quartile. Q1 included 164 subjects; Q2 included 127 subjects; Q3 included 156 subjects; and Q4 included 152 subjects. Dark shading (blue) indicates negative correlations; lighter shading (orange) indicates positive correlations. A visible shift in correlation patterns occurs between Q2 and Q3, corresponding to the cognitive transition from normal to mild impairment.

[0028] FIGs. 2A-2E show cognitive quartile organization and principal component-based quadrant analysis of lipid and protein clusters. FIG. 2A shows participants grouped into cognitive quartiles based on ADAS13 scores: Q1 = 1-7; Q2 = 8-12; Q3 = 13-18; Q4 = 19-52. Within each quartile, correlations between selected plasma lipid and CSF protein clusters are evaluated using first principal component (FPC) values. FIG.2B is a scatterplot of participants’ lipid and protein FPC scores, illustrating quadrant definitions: top-left (TL), top-right (TR), bottom-left (BL), and bottom-right (BR). This quadrant structure is used for downstream molecular and clinical comparisons. FIG. 2C is a graph showing in Quartile 1 (Q1), a comparison of white matter hyperintensity (WMH) burden between TL and BR quadrants. FIG.2D shows, in Q1, SomaScan differential expression analysis comparing Right vs Left quadrants (protein FPC axis). FIG. 2E shows, in Q1, SomaScan differential expression analysis comparing Top vs Bottom quadrants (lipid FPC axis).

[0029] FIGs. 3A-3G show lipid–protein associations linked to tau pathology and cerebrovascular changes in Quartile 3. FIG. 3A is a principal component scatterplot of participants in Q3 based on the astrogliosis protein cluster (x-axis) and the phospholipids | plasmalogens 22.6 lipid cluster (y-axis). FIG. 3B provides box-and-whisker plots comparing the top-left (TL) and bottom-right (BR) quadrants from FIG. 3A across markers of disease progression (hippocampal atrophy rate, tau PET signal), astrogliosis (CD44, CHI3L1, GFAP), beta oxidation (omega-3 and non–omega-3 acylcarnitines), and reverse cholesterol transport (omega-3 and nonomega-3 cholesterol esters). For each marker, the box on the left is TL, and the box on the right is BR. Notable proteins enriched in the BR quadrant include GFAP and GAS6, consistent with astrocyte activation (data not shown). FIG. 3C provides box-and- whisker plots comparing TL and BR quadrants from Nightingale NMR HDL lipoprotein markers, including HDL3, XL.HDL.C, and XL.HDL.CE. For each marker, the box on the left is TL, and the box on the right is BR. FIG. 3D is a principal component scatterplot of participants based on the ANLS / glycolysis protein cluster (x-axis) and the phospholipids | plasmalogens 22.6 lipid cluster (y-axis). FIG. 3E are graphs showing representative protein– lipid correlations in Q3, including LDHB–CE.22.6, SMOC1–AC.22.6, and YWHAB (14-3-3 zeta)–PE(O-16:0 / 22:6) associations. These examples illustrate disrupted coupling between lipid metabolism and astrocyte or glycolytic protein markers under metabolic stress. FIG. 4F shows a comparison of the estimated omega-3 index, calculated from DHA- and non-DHA– containing HDL particles measured by Nightingale NMR, between BR and TL quadrants. Median omega-3 index was significantly lower in the BR quadrant (p = 8.6e-7), suggesting impaired omega-3 availability may contribute to differential plasmalogen levels and astrocyte– neuron metabolic support. FIG. 3G shows SomaScan differential expression analysis comparing Right vs. Left quadrants from FIG. 3A, visualized as a volcano plot.

[0030] FIGs.4A-4D show that a ceramide-enriched lipid cluster, including CER.D19, emerges in Quartile 2 as a marker of flux differences and disrupted astrocyte–neuron metabolic support. FIG. 4A shows PCA-based quadrant analysis of the ANLS / glycolysis protein cluster (x-axis) and a ceramide-enriched lipid cluster (y-axis) in Q2. Participants in the bottom-right (BR) quadrant exhibit higher ANLS / glycolysis protein expression and lower ceramide lipid signal, suggesting a mismatch between increasing metabolic demand and reduced lipid clearance. FIG.4B is a comparison of tau PET signal and hippocampal atrophy rate between TL and BR quadrants defined in FIG. 4A. For each marker, the left box is TL, and the right box is BR. Participants in the BR quadrant show a 59% increase in tau PET signal and elevated atrophy rate. FIG. 4C provides scatter plots showing correlations between CER.D19.1.18.0 and representative ANLS / glycolysis proteins (LDHB, ALDOA, KPYM) in Q2. LDHB– CER.D19.1.18.0 is the strongest association (p = 0.0046), supporting disrupted metabolic coupling at this stage. FIG. 4D shows SomaScan differential expression analysis of CSF proteins associated with CER.D19.1.18.0 levels. Top proteins include RAB21, RAB7A, SLC27A2, RAC1, and FGF1 — involved in lipid droplet trafficking, cholesterol esterification, and oxysterol signaling. These associations were specific to Q2.

[0031] FIGs. 5A-5E demonstrate that CER.D19 marks a metabolic inflection point linked to cholesterol overload, bile acid biosynthesis, and GWAS-enriched lipid handling pathways. CER.D19 is also an indicator of beta oxidation. FIG.5A provides graphs showing correlations between CSF LDHB (a marker of astrocyte–neuron lactate shuttle activity), TNR and ALDOA and two enzymes involved in cholesterol esterification: PLA2G7 (positive correlation) and LPCAT2 (negative correlation). These associations suggest altered phosphatidylcholine (PC) processing and potential constraints on cholesterol export. FIG. 5B is a schematic illustration showing how PLA2G7 and LPCAT2 regulate the interconversion of PC and lysophosphatidylcholine (LPC), affecting cholesterol solubilization and packaging into APOE- containing lipoprotein particles for efflux. FIG. 5C provides representative AD GWAS SNPs associated with CER.D19.1.18.0 levels, including loci near NR1H3 (LXRA) and TSPOAP1, which regulate nuclear cholesterol sensing and mitochondrial cholesterol trafficking, respectively. Additional associated loci are provided in FIG. 10A and FIG. 10B. FIG. 5D is a heatmap showing correlations between CSF LDHB and selected markers across all cognitive quartiles. Variables include GFAP (astrogliosis), CER.D19.1.18.0 (cellular cholesterol load), PLA2G7 and LPCAT2 (PC metabolism), and CHGA and CNTN2 (ANLS signaling and adhesion). FIG. 5E is a summary diagram integrating lipidomic, proteomic, and genetic findings: early CER.D19 elevation coincides with metabolic stress markers (e.g., ANLS disruption), cholesterol overload, bile acid biosynthesis, and the emergence of AD risk gene expression signatures consistent with declining microglial lipid flux capacity.

[0032] FIGs.6A-6C provide a diagram of a unified hypothesis of Alzheimer’s disease through progressive collapse of microglial cholesterol flux capacity and astrocyte-neuron metabolic breakdown (FIG. 6A). A comparison of cholesterol debris flux relative to normal CSFlipoprotein turnover is cognitive normal (CN), MCI and AD is detailed (FIG. 6B). A graph showing decrease in hippocampus tissue volume and cognition at various ages is provided (FIG. 6C).

[0033] FIGs.7A-7B show Somascan proteins for Q1 (L vs. R quadrants; FIG.7A) and Enrich analysis (FIG. 7B).

[0034] FIGs. 8A-8B show Somascan proteins from Q1 (T vs B quadrants; FIG. 8A) used for EnrichR analysis (FIG. 8B).

[0035] FIGs. 9A-9B show Somascan proteins from Q2 that correlate with CER.D19.1.18.0. (T vs B quadrants; FIG. 9A) used for EnrichR analysis (FIG.9B).

[0036] FIGs.10A-10B show SNPs from AD GWAS. FIG.10A shows CER.D19.1.18.0 levels for AD GWAS SNPs found to be significant based on Kolmogorov-Smirnov statistics (168 AD summary GWAS SNPs considered), grouped by sex. FIG. 10B shows CER.D19.1.18.0 levels for SPTLC3 and NR1H3 (LXRA) found to be significant based on Kolmogorov-Smirnov statistics (168 AD summary GWAS SNPs considered), grouped by sex. DETAILED DESCRIPTION

[0037] The disclosure provides methods for analyzing and understanding Alzheimer’s disease and its progression, as well as targets for therapeutic intervention. Based on analysis of multiomic data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), as described in the accompanying Examples, the disclosure provides a unified hypothesis in which two interdependent failure modes: (i) saturation of microglial lipid flux capacity; and (ii) disruption of the astrocyte–neuron lactate shuttle (ANLS) due to excess astrocytic membrane cholesterol, drive disease progression upstream of amyloid and tau pathology. These processes appear to reinforce each other, resulting in accelerating neurodegeneration. The disclosure reframes AD as a systems-level collapse in metabolic coordination, rather than a purely linear pathological cascade. Without wishing to be bound by theory, it is postulated that this positive feedback loop (neurodegeneration → lipid debris processing → metabolic disruption → neurodegeneration) is a key destabilizing cycle responsible for disease progression in AD, reinforcing the perspective that Alzheimer’s is fundamentally a metabolic disease.

[0038] By stratifying participants by cognitive score quartiles, consistent associations linking impaired lipid clearance, metabolic stress, and genetic variants regulating cholesterol handling were identified. The disclosure provides evidence for distinct processes driving neurodegeneration in AD: (1) prior to cognitive decline, (2) early in cognitive decline but stillcognitively normal, and (3) at the threshold of MCI diagnosis. Prior to cognitive decline (ADAS13 scores 1-7), evidence of white matter hyperintensities (WMH) associated with complement and HDL proteins, independent of tau and hippocampal atrophy rates, were observed. In ADAS13 scores between 8-12, specific protein / lipid correlation networks that are associated with tau and hippocampal atrophy were observed. Early in cognitive decline (ADAS13 8-12), enzyme levels resulting in high phosphatidylcholine availability for cholesterol efflux onto CNS lipoproteins (low LPA2G7, high LPCAT2) are associated with disruption of ANLS, while the reverse is associated with high disruption. Near MCI, DHA and plasmalogens appear to be protective of lipid droplet formation and astrogliosis, presumably by decreasing lipid droplet size, and increasing their lipid debris flux capacity.

[0039] The disclosure hypothesizes that a key dynamic of AD pathophysiology is the flux capacity of microglia for processing cellular and myelin debris. Once this capacity is reached, and astrocytes are recruited to help with processing of excess debris, astrocytes' ability to provide metabolic support for neurons via ANLS is impaired, creating a negative feedback loop of neuronal death. In support of this hypothesis, evidence in ADNI for distinct processes occurring in windows of disease progression corresponding to ADAS13 scores was found. The first signal identified is a role for complement and HDL particles in cognitively normal individuals. A disruption in ANLS and a compensatory upregulation of glycolysis, in connection with elevated cellular cholesterol was also observed. Finally, increased astrogliosis and a reduction in plasma acylcarnitines and cholesterol esters, implying reduced lipid flux out of the CNS, was observed. This change in lipid and protein correlation structure is interpreted to indicate the point at which microglia reach their flux capacity for debris processing, thus recruiting and activating astrocytes, increasing astrocyte cholesterol levels, thereby further reducing neuronal metabolic support by ANLS.

[0040] The networks identified in this analysis provide evidence that progressively elevated cellular cholesterol levels, likely the result of accelerating neuronal and myelin debris, resulting in elevated cholesterol levels that are particularly harmful in astrocytes, perturbing homeostatic processes, resulting in disruption of ANLS, contributing to lipid droplet formation, and astrogliosis. The studies described herein point to actionable variations in lipids associated with slower disease progression and decreased metabolic disruption and inflammation. Abbreviations

[0041] As used in this specification and the appended claims, the singular forms “a,” “an” and “the” include plural references unless the content clearly dictates otherwise.

[0042] As used in this specification, the term “and / or” is used in this disclosure to either “and” or “or” unless indicated otherwise.

[0043] Throughout this specification, unless the context requires otherwise, the word “comprise”, or variations such as “comprises” or “comprising”, will be understood to imply the inclusion of a stated element or integer or group of elements or integers but not the exclusion of any other element or integer or group of elements or integers.

[0044] As used in this application, the terms “about” and “approximately” are used as equivalents. Any numerals used in this application with or without about / approximately are meant to cover any normal fluctuations appreciated by one of ordinary skill in the relevant art. In certain embodiments, the term “approximately” or “about” refers to a range of values that fall within 25%, 20%, 19%, 18%, 17%, 16%, 15%, 14%, 13%, 12%, 11%, 10%, 9%, 8%, 7%, 6%, 5%, 4%, 3%, 2%, 1%, or less in either direction (greater than or less than) of the stated reference value unless otherwise stated or otherwise evident from the context (except where such number would exceed 100% of a possible value).

[0045] “Decrease” or “inhibit” may refer to a decrease or inhibition of at least 5%, for example, at least 5%, at least 6%, at least 7%, at least 8%, at least 9%, at least 10%, at least 15%, at least 20%, at least 25%, at least 30%, at least 35%, at least 40%, at least 45%, at least 50%, at least 55%, at least 60%, at least 65%, at least 70%, at least 75%, at least 80%, at least 85%, at least 90%, at least 95%, at least 99% or 100%, for example, as compared to a reference or control level, e.g., in control cells or tissue.

[0046] “Increase” may refer to an increase of at least 5%, for example, at least 5%, at least 6%, at least 7%, at least 8%, at least 9%, at least 10%, at least 15%, at least 20%, at least 25%, at least 30%, at least 35%, at least 40%, at least 45v, at least 50%, at least 55%, at least 60%, at least 65%, at least 70%, at least 75%, at least 80%, at least 85%, at least 90%, at least 95%, at least 99% or at least 100%, for example, as compared to a reference level or the level in control cells or tissue. Increase also means increases of at least 1-fold, for example, at least 1-fold, at least 2-fold, at least 3-fold, at least 4-fold, at least 5-fold, at least 6-fold, at least 7-fold, at least 8-fold, at least 9-fold, at least 10-fold, at least 15-fold, at least 20-fold, at least 30-fold, at least 40-fold, at least 50-fold, at least 60-fold, at least 70-fold, at least 80-fold, at least 90-fold, at least 100-fold, at least 200-fold, at least 500-fold, at least 1000-fold or more, for example, as compared to the level of a reference or the level in control cells or tissue.

[0047] “Subjects” includes animals (e.g., mammals, swine, fish, birds, insects etc.). In some embodiments, subjects are mammals, particularly primates, especially humans. In someembodiments, subjects are livestock such as cattle, sheep, goats, cows, swine, and the like; poultry such as chickens, ducks, geese, turkeys, and the like; and domesticated animals such as dogs and cats. In some embodiments (e.g., particularly in research contexts) subjects are rodents (e.g., mice, rats, hamsters), rabbits, primates, or swine such as inbred pigs and the like. The terms “subject” and “patient” are used interchangeably herein.

[0048] An “antibody” is an immunoglobulin (Ig) molecule capable of specific binding to a target, such as a carbohydrate, polynucleotide, lipid, or polypeptide, through at least one epitope recognition site, located in the variable region of the Ig molecule. As used herein, the term encompasses not only intact polyclonal or monoclonal antibodies, but also fragments thereof, such as dAb, Fab, Fab', F(ab')2, Fv, single chain (scFv), synthetic variants thereof, naturally occurring variants, fusion proteins comprising an antibody portion with an antigen- binding fragment of the required specificity, chimeric antibodies, nanobodies, and any other modified configuration of the immunoglobulin molecule that comprises an antigen-binding site or fragment of the required specificity.

[0049] By “gene” is meant a unit of inheritance that occupies a specific locus on a chromosome and consists of transcriptional and / or translational regulatory sequences and / or a coding region and / or non-translated sequences (i.e., introns, 5′ and 3′ untranslated sequences).

[0050] “Treatment” or “treating,” as used herein, includes any desirable effect on the symptoms or pathology of a disease or condition, e.g., AD, and may include even minimal changes or improvements in one or more measurable markers of the disease or condition being treated. “Treatment” or “treating” does not necessarily indicate complete eradication or cure of the disease or condition, or associated symptoms thereof. Illustrative markers of clinical improvement will be apparent to persons skilled in the art.

[0051] Certain embodiments may employ conventional biology methods, software and systems for diagnostic purposes of the present invention. Computer software products of the invention typically include computer readable medium having computer-executable instructions for performing the logic steps of the method of the invention. The computer executable instructions may be written in a suitable computer language or combination of several languages.

[0052] The term “diagnose” or “diagnostic” or “diagnosed” includes identifying the presence or nature of a pathologic condition, such as MCI or AD, characterizing the risk of developing such a condition, and / or measuring the change (or no change) of a pathologic condition in response to therapy. Diagnostic methods may differ in their sensitivity and specificity. Incertain embodiments, the “sensitivity” of a diagnostic assay refers to the percentage of diseased subjects which test positive (percent of “true positives”). Diseased subjects not detected by the assay are typically referred to as “false negatives.” Subjects that are not diseased and which test negative in the assay may be termed “true negatives.” In certain embodiments, the “specificity” of a diagnostic assay may be defined as one (1) minus the false positive rate, where the “false positive” rate is defined as the proportion of those samples or subjects without the disease and which test positive.

[0053] The term “ANLS attenuation” indicates a stage in AD progression characterized by cholesterol-induced impairment of the astrocyte neuron lactate shuttle. ANLS attenuation may be associated with cholesterol accumulation in astrocytes and neurons, which impairs membrane dynamics, disrupts GPCR signaling, and reduces ANLS-mediated lactate delivery to neurons.

[0054] The term “microglial flux capacity saturation” indicates a stage in AD progression characterized by saturation of microglial lipid flux originating from cellular debris, myelin, amyloid plaques, or other protein aggregates. Alzheimer’s Disease

[0055] Alois Alzheimer first described the neuropathology that came to define the eponymous disease with his observations of neurofibrillary tangles, amyloid plaques, and lipid droplets82. Our modern understanding of the disease is rapidly improving, driven in part by a recognition of the increasing cost of care. Presently, anti-amyloid therapies with ~35% effect size equate to extending one’s dementia-free life by ~6 months81. Improved, personalized treatments are predicated on increased understanding of disease etiology. The two greatest risk factors for AD are age and APOE allele status2. GWAS studies have additionally identified loci associated with genes in processes involving processing of cellular debris such as phagocytosis, cholesterol metabolism, and innate immunity3.

[0056] The lipid droplets observed by Alois Alzheimer are now known to be generated primarily by microglia, containing esterified cholesterol, triglycerides and phospholipids. Lipid droplets alone are not indicative of pathology, and serve a temporary storage for lipids, including phagocytized cell debris, and are part of normal lipid homeostasis83. Cholesterol transport, synthesis, and homeostasis are tightly regulated in the brain where cholesterol can be synthesized and modified, with excess cholesterol being exported out of the CNS during sleep through reverse cholesterol transport (RCT) and / or conversion to 24-hydroxycholesterol (24OHC) or 27-hydroxycholesterol (27OHC), both of which can cross the blood brain barrierand are converted to bile acids in the liver. As cells themselves contain cholesterol, processing of dying cells and myelin turnover includes cholesterol turnover and trafficking. When microglia become overwhelmed with this task, they signal for astrocytes to assist them through astrogliosis. In both microglia and astrocytes, a key mechanism that enables off-loading excess cholesterol via lipoproteins for RCT is activation of LXR which triggers LD lipophagy, and upregulates ABCA1 and APOE mRNA expression84, 85, 9, 11. When lipid droplets become too large, they are unable to be transported to the mitochondria where CYP27A1 hydroxylates cholesterol into 27OHC, the key activator of LXR. In microglia, this results in the inability to off-load cholesterol onto lipoprotein particles11,86. In astrocytes, the same problem occurs, but the lack of 27OHC also fails to down-regulate LDLR-mediated cholesterol uptake and cholesterol biosynthesis via INSIG1 and SREBP29, 87, 88. This lack of 27OHC leads to elevated cholesterol levels across all cellular membranes, including the plasma membrane where lipid raft function and GPCR signaling are affected89 ,90, 91, 64.

[0057] GPCR signaling on astrocytes triggers lactate release in response to metabolic signaling from neurons. This signaling is part of a closed-loop system known as the astrocyte-neuron lactate shuttle (ANLS), which occurs at the nodes of Ranvier and synaptic boutons92-94. Because GPCR signaling and downstream ANLS are sensitive to astrocyte membrane cholesterol levels, multiple feedback loops (such as SREBP2 and LXR) exist to keep cholesterol levels at an optimal range89, 90, 64, 95, 96.

[0058] Many of the molecular underpinnings around lipid homeostasis, LDs and inflammation in glial cells were described in recent work by Victor et al and Haney et al97, 98. Using glial cell culture systems, they demonstrated how the overall levels of cellular cholesterol, either stored in lipid droplets, processed in the ER or golgi, or in external lipoprotein particles, are increased by APOE4. Both groups demonstrate that inflammation from fAβ and / or other TLR4 activators causes a shift towards glycolysis, away from oxidative phosphorylation of fatty acids, and an increase in ACSL1 to store the fatty acids in lipid droplets. TCW et al showed that this increase in lipid droplets and impaired cholesterol efflux could be remedied by the addition of LXR agonists11. A conclusion from these papers was that the overall increase in the amount of glial cellular lipids contributes to neurotoxicity, presumably through an increase in the amount of extracellular lipid particles that are available to neurons in the media.

[0059] While these cell culture systems have been hugely insightful for elucidating the mechanisms of lipid storage, the role of fAβ, and the role of APOE in Alzheimer’s etiology, a key limitation is that they do not recapitulate the recycling of cellular and myelin debristypically performed by microglia and astrocytes in vivo, nor do they properly model the diurnal pattern of reverse cholesterol transport in glial cells and neurons–essential to brain metabolism and homeostasis. For example, the importance of signaling pathways involved in debris processing by TREM2 have been well studied, and are known to impact cholesterol biosynthesis99. Methods

[0060] As described in the accompanying Examples, the disclosure identifies metabolic pathways and targets underlying cognitive disorders, including mild cognitive impairment (MCI) and Alzheimer’s disease (AD). Accordingly, the disclosure provides methods and compositions for diagnosing, stratifying, prognosing, treating, attenuating, and preventing cognitive impairments, including but not limited to MCI and AD. Methods disclosed herein may be used for any of these purposes.

[0061] In embodiments, the disclosure provides cerebral spinal fluid (CSF) proteins and plasma lipids for which changes in their amount or concentration is indicative of a subject being at risk for developing or having MCI or AD. These may be referred to herein as CSF protein markers or biomarkers and plasma lipid markers or biomarkers, respectively. Although referred to as “CSF proteins” and “plasma lipids,” it is understood that an amount or concentration of any of the CSF proteins and / or plasma lipids may be determined in biological samples other than CSF or plasma. CSF proteins and plasma lipids for which changes in their amount or concentration is indicative of a subject being at risk for developing or having MCI or AD include, but are not limited to, those identified herein as being associated with ANLS attenuation and those identified as being associated with microglial flux capacity saturation.

[0062] CSF proteins and serum lipids identified as markers of ANLS attenuation (or a predisposition to ANLS attenuation if indicated by *) are provided in Table 1, which indicates whether an increase (POS) or decrease (NEG) in the amount or concentration of the marker (e.g., as compared to a healthy control subject) is associated with ANLS attenuation. ANLS attenuation is a pathology, so a higher level of a POS marker indicates greater pathology. Table 1. CSF Proteins and Serum Lipids Associated with ANLS Attenuation

[0063] CSF proteins and serum lipids identified as markers of microglial flux capacity saturation (or a predisposition to microglial flux capacity saturation if indicated by *) are provided in Table 2, which indicates whether an increase (POS) or decrease (NEG) in the amount or concentration of the marker (e.g., as compared to a healthy control subject) is associated with microglial flux capacity saturation. Microglial flux capacity saturation is a pathology, so a higher level of a POS marker indicates greater pathology. Table 2. CSF Proteins and Serum Lipids Associated with Microglial Flux Capacity Saturation

[0064] Embodiments of the disclosure provide methods of determining whether a subject is at risk of MCI or AD, or has MCI or AD, by correlating the amount or concentration of CSF protein markers and / or serum lipid protein markers present in one more biological sample obtained from the subject with the presence or absence of MCI or AD. In embodiments, the method is used to determine whether the subject has MCI or AD associated with ANLS attenuation or microglial flux capacity saturation, by correlating the amount or concentration of CSF protein markers and / or serum lipid protein markers present in one more biological sample obtained from the subject with the presence of ANLS attenuation or microglial flux capacity saturation.

[0065] In embodiments of methods of the present invention, the correlating step comprises comparing the amount or concentration of the CSF proteins and / or serum lipids to normal control levels for each of the CSF proteins and / or serum lipids. In embodiments, the subject is identified as having MCI or AD if there is a difference in the amount or concentration of the CSF proteins and / or serum lipids between the subject and the normal control levels. In embodiments, the normal control level is a level in a healthy subject, while in embodiments, the normal control level is a level based on levels in a plurality of normal subjects.

[0066] In embodiments of methods of the present invention, the correlating step comprises comparing the amount or concentration of the CSF proteins and / or serum lipids to MCI or AD control levels for each of the CSF proteins and / or serum lipids. In embodiments, the subject is identified as having MCI or AD if there is not a significant difference in the amount or concentration of the CSF proteins and / or serum lipids between the subject and the MCI or AD control levels. In embodiments, the MCI or AD control level is a level in a diagnosed MCI or AD patient, while in embodiments, the MCI or AD control level is a level based on levels in a plurality of diagnosed MCI or AD patients.

[0067] In embodiments, the presence of an increased or decreased amount of one or more CSF protein and / or serum lipid disclosed herein is indicative of a subject being at risk of or having MCI or AD. As used herein, the terms “CSF protein marker” and “serum lipid marker” indicates that the CSF protein or serum lipid is present at an increased or decreased amount or concentration as compared to the amount or concentration present in a healthy subject or normal control or reference value, and that the increased or decreased amount or concentration is associated with risk of or having MCI or AD. For each marker, whether an increased or decreased amount or concentration is associated with risk of or having MCI or AD is indicated in Table 1 or Table 2 or discussed in the Examples. In embodiments, the presence of at least two, at least three, at least four, at least five, at least 10, at least 20, at least 30, or at least 40 or more CSF protein markers and / or serum lipid markers disclosed herein is indicative of a subject being at risk for or having MCI or AD.

[0068] In embodiments, the presence of an increased or decreased amount or concentration of at least one CSF protein marker and at least one serum lipid marker disclosed in any of the combinations disclosed herein is indicative of a subject having MCI or AD. In embodiments, the presence of at least two, at least three, at least four, at least five, at least 10, at least 20, at least 30, or at least 40 or more CSF proteins and at least two, at least three, at least four, at least five, at least 10, at least 20, at least 30, or at least 40 or more serum lipid markers disclosed herein is indicative of a subject having MCI or AD.

[0069] Methods for determining an amount of or a concentration of a CSF protein are known in the art and any method may be employed. Examples of such methods include but are not limited to mass spectrometry, e.g., targeted mass spectrometry, or aptamer-based proteomics. Mass spectrometry has both qualitative and quantitative uses, including quantifying the amount of a polypeptide gene product in a sample. Included are gas chromatography-mass spectrometry (GC / MS or GC-MS), liquid chromatography mass spectrometry (LC / MS or LC- MS), and ion mobility spectrometry / mass spectrometry (IMS / MS or IMMS). Accordingly, MS techniques may be used according to any of the methods provided herein to measure the levels of polypeptide gene product in a sample, and, optionally, to compare those levels to a control sample or a pre-determined value. Protein expression levels may also be determined, e.g., by immunohistochemistry using antibodies that specifically bind to protein gene products. In certain embodiments, protein expression levels are determined using a polypeptide array or an antibody array. Certain embodiments may employ standard methodologies and detectors such as western blotting and immunoprecipitation, enzyme-linked immunosorbent assays (ELISA),flow cytometry, and immunofluorescence assays (IFA), which utilize an imaging device. In embodiments, the amount or concentration of one or more CSF proteins in a CSF sample obtained from a subject is determined. In embodiments where the biological sample is obtained from a human subject, the methods are employed to detect human CSF proteins. The amino acid sequences of these proteins are known and available in the art. In embodiments, the amount or concentration of a CSF protein is determined by measuring the amount or concentration of mRNA encoding the CSF protein. mRNA may be measured by routine methods such as reverse transcription polymerase chain reaction (RT-PCR).

[0070] Methods for determining an amount of or a concentration of a serum lipid are known in the art and any method may be employed. Examples of such methods include but are not limited to targeted mass spectrometry or aptamer-based proteomics. In embodiments, the amount or concentration of one or more CSF proteins in a CSF sample obtained from a subject is determined. In embodiments where the biological sample is obtained from a human subject, the methods are employed to detect human lipids. The structure of these lipids are known and available in the art.

[0071] In embodiments of any of the methods, the one or more polypeptide or fragment thereof or the one or more lipid is detected using a detection system selected from the group consisting of: an immunohistochemistry, enzyme-linked immunosorbent assay (ELISA), radioimmunoassay (RIA), enzyme immunoassay (EIA), fluorescence immunoassay (FIA), and mass spectrometry, Nuclear Magnetic Resonance (NMR) spectroscopy.

[0072] The presence in a subject of a genetic mutation associated with a predisposition to MCI or AD may be determined by a variety of methods, including gene or mRNA sequencing of any gene(s) having mutations associated with a predisposition to MCI or AD. In embodiments, the presence of a genetic mutation is determined by whole-genome sequencing or methylation array analysis. The sequences of genes and genetic mutations associated with MCI and AD, including but not limited to those disclosed herein, are known in the art. Non-limiting examples of genetic mutations that may be determined according to the present invention include those shown in FIGs. 10A-10B.

[0073] Methods disclosed herein may be utilized on a variety of biological samples obtained from a subject. In embodiments, the biological sample is cerebral spinal fluid (CSF), blood, or serum. In embodiments, amounts or concentration of protein markers may be measured in a CSF, blood, or serum sample obtained from a subject. In embodiments, amounts or concentration of lipid markers may be measured in a CSF, serum or blood sample obtainedfrom a subject. In embodiments, amounts or concentration of protein markers may be measured in a CSF sample obtained from a subject. In embodiments, amounts or concentration of lipid markers may be measured in a serum or blood sample obtained from a subject. In embodiments, amounts or concentrations of protein markers and lipid markers may be determined in other biological samples obtained from a subject, including but not limited to, neuronal cells and monocytes. In embodiments, amounts of protein markers and lipid markers are determined in exosomes. A biological sample may be obtained using any method known to the art. In certain embodiments, a biological sample is obtained from a subject by a medical professional, e.g., at a hospital, doctor's office, testing center or laboratory. In certain embodiments, a biological sample may be obtained using a kit, which may contain a means for obtaining a sample, a means for storing the sample, and instructions for use of the kit. In some cases, the kit is provided by a molecular profiling service, which may also perform an assay on the biological sample.

[0074] In embodiments, the subject is an animal (e.g., a mammal), including but not limited to humans, non-human primates, rodents, dogs, cats, pigs, fish, and the like. In embodiments, the present methods and compositions apply to humans. In embodiments, the human is a child, an adolescent, or an adult. In embodiments, the subject has been determined to be at risk for having or is suspected of having MCI or AD, and in embodiments, the subject has been diagnosed with MCI or AD.

[0075] As used herein, the term “subject at risk for MCI or AD” refers to a subject with one or more risk factors for developing MCI or AD. Risk factors include, but are not limited to, age and genetic predisposition.

[0076] Methods and kits of the present invention may be used to determine whether a subject is at risk for MCI or AD or has MCI or AD, based on the amounts or concentrations of CSF proteins and / or serum lipids identified in Tables 1 and 2. Generally, a determination of risk for MCI or AD, or determination of MCI or AD is made when the amounts or concentrations of the CSF proteins and / or serum lipids are determined to more closely correlate to the amounts or concentrations observed in MCI or AD patients than the amounts or concentrations observed in healthy subjects. Similarly, a determination that the subject is not at risk for MCI or AD or does not have MCI or AD is generally made when the amounts or concentrations of the CSF proteins and / or serum lipids are determined to more closely correlate to the amounts or concentrations observed in healthy subjects than the amounts or concentrations observed in MCI or AD patients.

[0077] In embodiments, the disclosure includes a method of determining whether a subject is at risk of developing MCI or AD, the method comprising determining whether one or more biological samples obtained from the subject comprises: (a) one or more genetic mutations indicative of a predisposition to ANLS attenuation and / or microglial flux capacity saturation; (b) an amount or concentration of one or more cerebral spinal fluid (CSF) proteins indicative of a predisposition to astrocyte-neuron lactate shuttle (ANLS) attenuation; (c) an amount or concentration of one or more plasma lipids indicative of a predisposition to ANLS attenuation; and / or (d) an amount or concentration of one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation; wherein: (i) the presence of one or more of the genetic mutations; (ii) an increased or decreased amount or concentration of the one or more CSF proteins of (b); and / or (iii) an increased or decreased amount or concentration of the one or more plasma lipids of (c) and / or (d) is indicative that the subject is at risk of developing MCI or Alzheimer’s Disease AD, wherein the increases and / or decreases are relative to a pre-determined amount or concentration or a control or reference amount or concentration, or has changed significantly over a period of time. In embodiments, only the amount or concentration of CSF proteins or the amount or concentration of plasma lipids is determined. In embodiments, the amount or concentration of CSF proteins and the amount or concentration of plasma lipids is determined. In embodiments, the amount or concentration of CSF proteins, the amount or concentration of plasma lipids, and the presence of one or more genetic mutations is determined. In embodiments, the amount or concentration of additional CSF proteins and / or additional plasma lipids are determined, which may include, e.g., one or more CSF proteins indicative of ANLS attenuation; one or more plasma lipids indicative of ANLS attenuation; one or more CSF proteins indicative of microglial flux capacity saturation; and / or one or more plasma lipids indicative of microglial flux capacity saturation. Additionally, one or more CSF proteins and / or serum lipids known to maintain the same amount or concentration in both normal subjects and MCI or AD patients may be determined as controls.

[0078] In embodiments, the one or more genetic mutations indicative of a predisposition to ANLS attenuation are selected from mutations in INPP5D, PICALM, TSPOAP, TNIP1, and / or LXRA genes; the one or more CSF proteins indicative of a predisposition to ANLS attenuationare selected from PLA2G7 and LPCAT2; the one or more plasma lipids indicative of a predisposition to ANLS attenuation are plasma lipids indicating high DHA in phospholipid or plasmalogen form; and the one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation are selected from the group consisting of DHA, omega-3 fatty acids, optionally in the form of phospholipids or plasmalogens.

[0079] In embodiments, the disclosure includes a method of determining whether a subject has MCI or AD, the method comprising determining whether one or more biological samples obtained from the subject comprises: (a) one or more genetic mutations indicative of a predisposition to ANLS attenuation and / or microglial flux capacity saturation; (b) an increased or decreased amount of one or more CSF proteins indicative of ANLS attenuation; (c) an increased or decreased amount of plasma lipids indicative of ANLS attenuation; (d) an increased or decreased amount of one or more CSF proteins indicative of microglial flux capacity saturation; and / or (e) an increased or decreased amount of one or more plasma lipids indicative of microglial flux capacity saturation, wherein: (i) the presence of one or more of the genetic mutations; (ii) an increased or decreased amount or concentration of the one or more CSF proteins of (b) and / or (d); and / or (iii) an increased or decreased amount or concentration of the one or more plasma lipids of (c) and / or (e) is indicative that the subject has MCI or Alzheimer’s Disease AD, wherein the increases and / or decreases are relative to a pre-determined amount or concentration or a reference or control amount or concentration. In embodiments, only the amount or concentration of CSF proteins or the amount or concentration of plasma lipids is determined. In embodiments, the amount or concentration of CSF proteins and the amount or concentration of plasma lipids is determined. In embodiments, the amount or concentration of CSF proteins, the amount or concentration of plasma lipids, and the presence of one or more genetic mutations is determined. In embodiments, the amount or concentration of additional CSF proteins and / or additional plasma lipids are determined, which may include, e.g., one or more CSF proteins indicative of a predisposition to ANLS attenuation; one or more plasma lipids indicative of a predisposition to ANLS attenuation; one or more CSF proteins indicative of a predisposition to microglial flux capacity saturation; and / or one or more plasma lipids indicative of apredisposition to microglial flux capacity saturation. Additionally, one or more CSF proteins and / or serum lipids known to maintain the same amount or concentration in both normal subjects and MCI or AD patients may be determined as controls.

[0080] In embodiments, the one or more genetic mutations indicative of a predisposition to ANLS attenuation are selected from mutations in INPP5D, PICALM, TSPOAP, TNIP1, and / or LXRA genes; the one or more CSF proteins indicative of ANLS attenuation are selected from the group consisting of RAB21, RAB7A, LDHB, LDHC, ALDOA, MDHC, KPYM, AATC, CNTN2, and TNR; the one or more plasma lipids indicative of ANLS attenuation are selected from the group consisting of CER.D19.1.18.0 and CER.D19.1.16.0; the one or more CSF proteins indicative of microglial flux capacity saturation are selected from the group consisting of TAU, CD44, CH3L1, GFAP, and GAS6; and / or the one or more plasma lipids indicative of microglial flux capacity saturation are selected from the group consisting of DHA / EPA acylcarnities, optionally AC.22.6 or AC.20.5, and cholesterol esters, optionally CE.22.6 or CE.20.5, extra-large HDL lipoprotein cholesterol and / or cholesterol ester, optionally Nightingale assay XL.HDL.C or XL.HDL.CE. In embodiments, the one or more genetic mutations indicative of a predisposition to ANLS attenuation are selected from mutations in INPP5D, PICALM, TSPOAP, TNIP1, and / or LXRA genes; the one or more CSF proteins and / or one or more serum lipids indicative of ANLS attenuation are set forth in Table 1; and the one or more CSF proteins and / or serum lipids indicative of microglial flux capacity saturation are set forth in Table 2.

[0081] In embodiments, the disclosure includes a method for characterizing the cellular cholesterol dyshomeostasis status of a subject at risk of MCI or AD or having MCI or AD, the method comprising determining whether one or more biological samples obtained from the subject comprises: (a) one or more genetic mutations indicative of a predisposition to ANLS attenuation and / or microglial flux capacity saturation; (b) an increased or decreased amount of one or more cerebral spinal fluid (CSF) proteins indicative of a predisposition to astrocyte-neuron lactate shuttle (ANLS) attenuation; (c) an increased or decreased amount of one or more plasma lipids indicative of a predisposition to ANLS attenuation; (d) an increased or decreased amount of one or more CSF proteins indicative of ANLS attenuation;(e) an increased or decreased amount of plasma lipids indicative of ANLS attenuation; (f) an increased or decreased amount of one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation; (g) an increased or decreased amount of one or more CSF proteins indicative of microglial flux capacity saturation; and / or (h) an increased or decreased amount of one or more plasma lipids indicative of microglial flux capacity saturation, wherein: (i) the presence of one or more of the genetic mutations of (a) and / or the increased or decreased amount of the one or more CSF proteins and / or one or more serum lipids of (b), (c), (d) and / or (e) is indicative of ANLS attenuation; and (ii) the presence of one or more of the genetic mutations of (a) and / or the increased or decreased amount of the one or more CSF proteins and / or plasma lipids of (f), (g) and / or (h) is indicative of microglial flux capacity saturation, wherein the increases and / or decreases are relative to a pre-determined amount or concentration or a control or reference amount or concentration. In embodiments, only the amount or concentration of CSF proteins or the amount or concentration of plasma lipids is determined. In embodiments, the amount or concentration of CSF proteins and the amount or concentration of plasma lipids is determined. In embodiments, the amount or concentration of CSF proteins, the amount or concentration of plasma lipids, and the presence of one or more genetic mutations is determined. In embodiments, only the amounts or concentrations of CSF proteins and / or serum lipids indicative of ANLS attenuation and / or microglial flux capacity saturation are determined. Additionally, one or more CSF proteins and / or serum lipids known to maintain the same amount or concentration in both normal subjects and MCI or AD patients may be determined as controls.

[0082] In embodiments, the one or more genetic mutations indicative of a predisposition to ANLS attenuation are selected from mutations in INPP5D, PICALM, TSPOAP, TNIP1, and / or LXRA genes; the one or more CSF proteins indicative of a predisposition to ANLS attenuation are selected from PLA2G7 and LPCAT2; the one or more plasma lipids indicative of a predisposition to ANLS attenuation are plasma lipids indicating high DHA in phospholipid or plasmalogen form; the one or more CSF proteins indicative of ANLS attenuation are selected from the group consisting of RAB21, RAB7A, LDHB, LDHC, ALDOA, MDHC, KPYM, AATC, CNTN2, and TNR; the one or more plasma lipids indicative of ANLS attenuation are selected from the group consisting of CER.D19.1.18.0 and CER.D19.1.16.0; the one or moreplasma lipids indicative of a predisposition to microglial flux capacity saturation are selected from the group consisting of DHA, omega-3 fatty acids, optionally in the form of phospholipids or plasmalogens; the one or more CSF proteins indicative of microglial flux capacity saturation are selected from the group consisting of TAU, CD44, CH3L1, GFAP, and GAS6; and / or the one or more plasma lipids indicative of microglial flux capacity saturation are selected from the group consisting of DHA / EPA acylcarnities, optionally AC.22.6 or AC.20.5, and cholesterol esters, optionally CE.22.6 or CE.20.5, extra-large HDL lipoprotein cholesterol and / or cholesterol ester, optionally Nightingale assay XL.HDL.C or XL.HDL.CE. Plasma lipids indicating high DHA in phospholipid or plasmalogen form include, for example, docosahexaenoic acid and a special class of ether phospholipids with a vinyl-ether bond at the sn-1 position of the glycerol backbone. These are forms of omega-3 fatty acids. In embodiments, the one or more genetic mutations indicative of a predisposition to ANLS attenuation are selected from mutations in INPP5D, PICALM, TSPOAP, TNIP1, and / or LXRA genes; the one or more CSF proteins and / or one or more serum lipids indicative of ANLS attenuation are set forth in Table 1; and the one or more CSF proteins and / or serum lipids indicative of microglial flux capacity saturation are set forth in Table 2.

[0083] In embodiments, the disclosure provides a method of diagnosing and / or treating MCI or AD, comprising: (a) detecting an amount of one or more polypeptide or fragment thereof in a biological sample obtained from a subject, wherein the one or more polypeptide is set forth in Table 1 or Table 2, is YWHAB, YWHAZ, CHI3L1, CD44, GFAP, LPA2G7, LDHB, PLA2G7, CNTN2, or chromogranin A, or any combination thereof; and (b) comparing the amount of each of the one or more polypeptide or fragment thereof detected in step (a) to a reference value, a predetermined cut-off value or to an amount of each of the one or more polypeptide detected in a normal control sample, wherein an increased or reduced amount of one or more polypeptide or fragment thereof in the biological sample of (a) as compared to the reference value, the predetermined cut-off value or the amount in the normal control sample of (b) indicates the presence of MCI or AD; and (c) optionally, if the presence of MCI or AD is indicated in step (b), administering a therapeutic agent to the subject.

[0084] Additionally, one or more CSF proteins and / or serum lipids known to maintain the same amount or concentration in both normal subjects and MCI or AD patients may be determined as controls.

[0085] In embodiments, the disclosure provides a method of diagnosing and / or treating MCI or AD, comprising: (a) detecting an amount of one or more lipid in a biological sample obtained from a subject, wherein the lipid is a phospholipid or ceramide set forth in Table 1 or Table 2, or selected from the group consisting of omega3 DHA, omega3 EPA, AC.13.0, AC.15.0, CE.16.0, CE.17.0, HDL3, CER.D19.1.18.0, and CER.D19.1.16.0, or any combination thereof; and (b) comparing the amount of each of the one or more lipid detected in step (a) to a reference value, a predetermined cut-off value or to an amount of each of the one or more lipid detected in a control sample, wherein a decreased or increased amount of one or more lipid in the biological sample of (a) as compared to the reference value, the predetermined cut-off value or the amount in the control serum sample of (b) indicates the presence of MCI or AD, and (c) optionally, if the presence of MCI or AD is indicated in step (b), administering a therapeutic agent to the subject.

[0086] Additionally, one or more CSF proteins and / or serum lipids known to maintain the same amount or concentration in both normal subjects and MCI or AD patients may be determined as controls.

[0087] In embodiments, the disclosure provides a method of diagnosing and / or treating MCI or AD, comprising: (a) detecting an amount of a cluster of proteins comprising LDHB, LDHC, CD44, ALDOA, PKM, MDH1, GAA, or ENO1, or a subcombination thereof, in a biological sample obtained from a subject; and (b) comparing the amount of the cluster of proteins detected in step (a) to a reference value, a predetermined cut-off value or to an amount of the cluster of proteins detected in a control sample, wherein an increased amount of the cluster of proteins in the biological sample of (a) as compared to the predetermined cut-off value or amount in the control serum sample of (b) indicates the presence of MCI or AD; and (c) optionally, if the presence of MCI or AD is indicated in step (b), administering a therapeutic agent to the subject.

[0088] Additionally, one or more CSF proteins and / or serum lipids known to maintain the same amount or concentration in both normal subjects and MCI or AD patients may be determined as controls.

[0089] It is understood that when reference is made to “determining an amount” of a protein or lipid, the method may also be practiced be determining a concentration of the protein or lipid in a biological sample, e.g., CSF or serum.

[0090] Any of the methods disclosed herein may be similarly used to determine the prognosis of MCI or AD in a subject, for determining risk of progression of MCI or AD in a subject, monitoring progression from risk of MCI or AD to active MCI or AD in a subject, monitoring progression of MCI or AD in a subject, or for monitoring the effectiveness of a therapy or therapeutic agent used to treat MCI or AD in a subject. In embodiments, such uses may comprise practicing any of the methods disclosed herein at different time points, e.g., over a period of time or over a course of treatment, and changes in the amounts or concentrations of the CSF proteins and / or serum lipids, e.g., to more closely align with those in normal subjects or to more closely align with MCI or AD patients, is indicative of disease regression or disease progression over time.

[0091] In embodiments, the disclosure provides a method for determining the disease stage of a subject is at risk of developing or having mild cognitive impairment (MCI) or Alzheimer’s Disease (AD), comprising determining whether a biological sample(s) obtained from the subject comprises cerebral spinal fluid (CSF) protein and plasma lipid markers associated with one or more stages MCI or AD, wherein: a) a correlation between one or more CSF protein selected from ALB, APOA4, APOC1, APOC2, CP, C9, KNG1, PGLYRP2, PON1, and THRB and one or more plasmid lipid selected from hexceramides and phospholipids 18.1 indicates that subject has an ADAS13 score of 1- 7; b) a correlation between one or more CSF protein selected from ANLS / glycolysis proteins and one or more plasmid lipid selected from ceramides 16.1 / 18.0 indicates that subject has an ADAS13 score of 8-12; c) a correlation between one or more CSF protein selected from RAB21, RAB7A, SLC27A2, RAC1, FGF1, and INPP5A and CER.D19.1.18.0 indicates that subject has an ADAS13 score of 8-12; d) a correlation between one or more CSF protein selected from ANLS / glycolysis proteins(optionally selected from LDHB, LDHC, CD44, ALDOA, PKM, MDH1, GAA, and ENO1) and one or more plasmid lipid selected from phospholipids / plasmalogens 22.6 indicates that subject has an ADAS13 score of 13-18; and e) a correlation between one or more CSF protein selected from Astrogliosis proteins (optionally selected from YWHAB, YWHAZ, CHI3L1, and CD44) and one or more plasmid lipid selected from phospholipids / plasmalogens 22.6 indicates that subject has an ADAS13 score of 13-18.

[0092] Additionally, one or more CSF proteins and / or serum lipids known to maintain the same amount or concentration in both normal subjects and MCI or AD patients may be determined as controls.

[0093] In embodiments, the disclosure includes a method of monitoring MCI and / or AD in a subject, e.g., to determine whether the disease is progressing or to determine whether a treatment is slowing progression of the disease or improving the disease, the method comprising determining at two or more distinct time points: (a) an amount or concentration of one or more cerebral spinal fluid (CSF) proteins indicative of a predisposition to astrocyte-neuron lactate shuttle (ANLS) attenuation; (b) an amount or concentration of one or more plasma lipids indicative of a predisposition to ANLS attenuation; and / or (c) an amount or concentration of one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation; wherein: (i) a change in the amount or concentration of the one or more CSF proteins of (a) when measured at the two or more different time points; and / or (ii) a change in the amount or concentration of the one or more plasma lipids of (b) and / or (c) when measured at the two or more different time points is indicative that the subject is at risk of developing MCI or Alzheimer’s Disease AD, wherein the changes measured may be relative to a reference value or a control or reference amount or concentration. In embodiments, only the amount or concentration of CSF proteins or the amount or concentration of plasma lipids is determined.

[0094] In embodiments, the amount or concentration of CSF proteins and the amount or concentration of plasma lipids is determined. In embodiments, the amount or concentration of CSF proteins, the amount or concentration of plasma lipids, and the presence of one or more genetic mutations is determined. In embodiments, the amount or concentration of additional CSF proteins and / or additional plasma lipids are determined, which may include, e.g., one ormore CSF proteins indicative of ANLS attenuation; one or more plasma lipids indicative of ANLS attenuation; one or more CSF proteins indicative of microglial flux capacity saturation; and / or one or more plasma lipids indicative of microglial flux capacity saturation. Additionally, one or more CSF proteins and / or serum lipids known to maintain the same amount or concentration in both normal subjects and MCI or AD patients may be determined as controls.

[0095] In embodiments, methods of determining whether the biological sample(s) obtained from the subject comprises cerebral spinal fluid (CSF) protein and plasma lipid markers associated with one or more stages MCI or AD comprise determining an amount or concentration of the CSF proteins and / or serum lipids tested in the biological samples and comparing those amounts or concentrations to reference values or control values, e.g., using a trained algorithm or classifier. For example, a reference value may be an average amount or concentration present in the biological sample of healthy subjects, which may be determined by the classifier. In another example, a reference value may be an average amount or concentration present in the biological sample of MCI or AD patients, which may be determined by the classifier. In embodiments, the amounts or concentrations of the CSF proteins and / or the plasma lipids or present in a biological sample obtained from a patient being evaluated for MCI or AD and the amounts or concentrations of the CSF proteins and / or the plasma lipids or present in a biological sample obtained from one or more healthy control subjects are analyzed by a classifier or algorithm in the same assay. In embodiments, the classifier or algorithm is generated using amount or concentration data determined for the CSF proteins and / or plasma lipids determined in (a) from a plurality of CSF and / or plasma or blood samples obtained from healthy subjects and a plurality of CSF and / or plasma or blood samples obtained from subjects with MCI and / or AD. In embodiments, a reference value is a predetermined value.

[0096] In embodiments, the disclosure provides a method for determining whether a subject is at risk of developing mild cognitive impairment (MCI) or Alzheimer’s Disease (AD), for determining whether the subject has MCI or AD, or for characterizing the cellular cholesterol dyshomeostasis status of the subject, the method comprising: (a) determining an amount or concentration of: (i) one or more cerebral spinal fluid (CSF) proteins indicative of a predisposition to astrocyte-neuron lactate shuttle (ANLS) attenuation in a CSF sample obtained from the subject; (ii) determining an amount or concentration of one or more plasma lipids indicative of a predisposition to ANLS attenuation in a blood or plasma sample obtained from the subject;(iii) determining an amount or concentration of one or more CSF proteins indicative of ANLS attenuation in a CSF sample obtained from the subject; (iv) determining an amount or concentration of plasma lipids indicative of ANLS attenuation in a blood or plasma sample obtained from the subject; (v) determining an amount or concentration of one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation in a blood or plasma sample obtained from the subject; (vi) determining an amount or concentration of one or more CSF proteins indicative of microglial flux capacity saturation in a CSF sample obtained from the subject; and / or (vii) determining an amount or concentration of one or more plasma lipids indicative of microglial flux capacity saturation in a blood or plasma sample obtained from the subject; and (b) determining whether the subject is at risk of developing mild cognitive impairment (MCI) or Alzheimer’s Disease (AD) or has MCI or AD, or characterizing the cellular cholesterol dyshomeostasis status of the subject, by correlating the amounts determined in (a) with the risk of developing MCI or AD, the presence or absence of MCI or AD, and / or the cellular cholesterol dyshomeostasis status of the subject, wherein the correlating is performed using a trained algorithm or classifier. In embodiments, the classifier or algorithm is generated using amount or concentration data determined for the CSF proteins and / or plasma lipids determined in (a) from a plurality of CSF and / or plasma or blood samples obtained from healthy subjects and a plurality of CSF and / or plasma or blood samples obtained from subjects with MCI and / or AD. Additionally, one or more CSF proteins and / or serum lipids known to maintain the same amount or concentration in both normal subjects and MCI or AD patients may be determined as controls. Embodiments of any of the methods may also be practiced at two or more different time points, and a change in the amount or concentration of the measured CSF proteins and / or serum lipids may be correlated to MCI or AD, or to ANLS attenuation or microglial flux capacity saturation.

[0097] In embodiments, the disclosure provides a method to determine whether the subject has AD associated with ANLS attenuation or microglial flux capacity saturation. In embodiments, the method comprises determining an amount or concentration of: (a) one or more CSF proteins set forth in Table 1 or Table 2, or selected from the group consisting of: PLA2G7, LPCAT2, RAB21, RAB7A, LDHB, LDHC, ALDOA, MDHC,KPYM, AATC, CNTN2, TNR, TAU, CD44, CH3L1, GFAP, and GAS6; and / or (b) one or more plasma lipids set forth in Table 1 or Table 2, or selected from the group consisting of: plasma lipids indicating high DHA in phospholipid or plasmalogen form, CER.D19.1.18.0, CER.D19.1.16.0, DHA, omega-3 fatty acids, optionally in the form of phospholipids or plasmalogens, DHA / EPA acylcarnities, optionally AC.22.6 or AC.20.5, cholesterol esters, optionally CE.22.6 or CE.20.5, and extra-large HDL lipoprotein cholesterol and / or cholesterol ester, optionally Nightingale assay XL.HDL.C or XL.HDL.CE. In embodiments, the method comprises determining the amount or concentration of at least 1, at least 2, at least 3, at least 5, or at least 10 of the CSF proteins and determining the amount or concentration of at least 2, at least 3, at least 5, or at least 10 of the serum lipids.

[0098] Additionally, one or more CSF proteins and / or serum lipids known to maintain the same amount or concentration in both normal subjects and MCI or AD patients may be determined as controls.

[0099] In embodiments of any of the methods disclosed herein, the method may further comprise assessing one or more other markers of factors associated with a risk for MCI or AD or active MCI or AD, such as, e.g., cognitive scores and / or longitudinal biomarker trajectories. In embodiments, cognitive scores may be ADAS13 test scores. Examples of biomarkers include, but are not limited to, beta-amyloid, pTau-217, pTau-181, and neurofilament light chain (Nfl).

[0100] The determination of whether the subject has AD associated with ANLS attenuation or microglial flux capacity saturation may be based on whether the amounts or concentrations of the tested proteins and / or lipids more closely correlate to the amounts or concentrations associated with either ANLS attenuation or microglial flux capacity saturation, e.g., the amounts or concentrations determined in patients known to have either ANLS attenuation or microglial flux capacity saturation. This may be assessed using, e.g., using a trained algorithm or classifier. In embodiments, the classifier or algorithm is generated using amount or concentration data determined for the CSF proteins and / or plasma lipids determined in (a) from a plurality of CSF and / or plasma or blood samples obtained from healthy subjects and a plurality of CSF and / or plasma or blood samples obtained from subjects with MCI and / or AD associated with either ANLS attenuation or microglial flux capacity saturation. In embodiments, the classifier may further consider other markers associated with MCI or AD, such as, e.g., genetic mutations, cognitive scores, and / or longitudinal biomarkers.

[0101] In certain instances, the risk for or presence of MCI or AD in a subject is determined by comparing the amounts or concentrations of CSF proteins and / or serum lipids described herein to one or more suitable controls. A “suitable control” or “appropriate control” may include a reference value, e.g., amounts or concentrations of the CSF proteins and / or serum lipids in a healthy subject or MCI or AD patient. In certain embodiments, a “suitable control” or “appropriate control” is a predefined value. In certain embodiments, the amounts or concentrations of the CSF proteins and / or serum lipids are compared to a reference amount or concentration of the CSF proteins and / or serum lipids, which may, in certain embodiments, be a predetermined or predefined value. Reference amounts and concentrations may be determined based upon amounts or concentrations of the CSF proteins and / or serum lipids in one or more suitable control samples, which may be either from healthy subjects or MCI or AD patients.

[0102] In embodiments, an increase or a decrease in the amount or concentration of one or more CSF protein or serum lipid as compared to a control sample or value, e.g., based on a healthy subject, is indicative of a risk for MCI or AD or for having MCI or AD. In embodiments, the increase or decrease may correspond to, e.g., an increase of about 10, 20, 50, 100, 200, 500, or 1000% or a decrease of about 10, 20, 50, or 90%.

[0103] In certain instances, the risk for or presence of MCI or AD in a subject is determined by analyzing the amounts or concentrations of CSF proteins and / or serum lipids described herein using a trained algorithm or classifier. In embodiments, the classifier may further consider other markers associated with MCI or AD, such as, e.g., genetic mutations, cognitive scores, and / or longitudinal biomarkers. Analysis and correlation with MCI or AD, or correlation with ANLS attenuation or microglial flux capacity saturation, may be performed using classifiers or algorithms, e.g., algorithms designed to normalize and or improve the reliability of the data. Examples of algorithms that may be used include any of those known in the art, described herein, or described in the accompanying Example.

[0104] In embodiments, methods of the present invention may be performed using a “machine learning algorithm,” which refers to a computational-based prediction methodology, also known to persons skilled in the art as a “classifier”, which may be employed for characterizing a protein and / or lipid profile. The signals corresponding to certain amounts or concentrations, which are typically subjected to the algorithm in order to classify the protein and / or lipid profile. Supervised learning generally involves “training” a classifier to recognize the distinctions among classes, e.g., MCI or AD versus healthy, and then “testing” the accuracy ofthe classifier on an independent test set. For new, unknown samples, the classifier can be used to determine the class in which the samples belong. In particular embodiments, the correlating is performed using an algorithm, including based on amounts or concentrations of any of the CSF proteins and / or serum lipid, or any of the sets of CSF proteins and / or serum lipids described herein.

[0105] In embodiments, the classifier comprises a neural network trained on multiomic AD cohorts with known subtype classification, such as, e.g., ANLS attenuation or microglial flux capacity saturation. In embodiments, the classifier provides output indicating whether the subject is at risk for MCI or AD, whether the subject has MCI or AD, whether the MCI or AD is associated with ANLS attenuation or microglial flux capacity saturation, the predicted trajectory of disease progression, and / or a personalized treatment schedule, based on the subject’s biomarker status or dynamics.

[0106] In some implementations, unsupervised clustering of plasma lipids and CSF proteins can be performed by inputting data representative of the plasma lipids and CSF proteins (e.g., concentration data, etc.) to a clustering model, such as a k-means cluster model, a Gaussian mixture model, and / or the like. A principle component analysis (PCA) technique can be applied to the resulting clustered data to determine correlation between select plasma lipids and CSF proteins and to further determine disruption (e.g., a rate change) in lipid metabolism and glial function as cognition declines, as described further herein at least in relation to FIG. 2A. The PCA technique, in combination with the unsupervised clustering, can therefore identify (1) CDF protein trends indicative of a predisposition to ANLS attenuation and / or (2) plasma lipids that are indicative of a predisposition to microglial flux capacity saturation. At least one of the foregoing predispositions can be associated with at least one of mild cognitive impairment (MCI) and / or Alzheimer’s Disease (AD).

[0107] The foregoing correlation results produced by the unsupervised clustering and PCA technique described above can result in training data that can be used to train a classifier (e.g., a neural network based classifier, such as a softmax classifier and / or the like) to predict, e.g.,: (1) whether a subject is at risk of having MCI and / or AD (e.g., represented by a binary output of the classifier), (2) a severity level associated with MCI and / or AD (represented by a distribution output of the classifier), (3) whether the subject has AD associated with ANLS attenuation or microglial flux capacity saturation; and / or (4) the like. Alternatively or in addition, the training data can be used to train a regression model (e.g., a neural network based regression model) configured to predict, for example, (1) a probability that the subject willhave MCI and / or AD, (2) a prognosis (e.g., expressed in years) associated with MCI and / or AD, and / or (3) other predictions represented by continuous values output by the regression model.

[0108] In some implementations, the training data can include training labels that are based on other factors, such as, e.g., genetic mutations and / or cognitive decline observations in subjects. These training labels can be associated with plasma lipid data (e.g., associated with DHA, omega-3 fatty acids, etc.) and / or CSF protein data (e.g., associated with PLA2G7, LPCAT2, etc.) within the training data, such that the training data includes ground truth data. The plasma lipid data and / or the CSF protein data can be provided as input to a machine learning model (e.g., the classifier, the regression model, etc., described above), and the training labels can be used to adjust the output of the machine learning model until the machine learning model is trained to generate, based on a subject’s plasma lipid data and / or CSF protein data, prediction data associated with MCI and / or AD for that subject.

[0109] Profiling or classification using sets of CSF protein markers and / or serum lipid markers (also referred to as “biomarkers”) can be used to identify a risk for or presence of MCI or AD, or to classify an MCI or AD as associated with ANLS attenuation or microglial flux capacity saturation. Sets may be derived from analysis of amounts or concentrations in cohorts having MCI or AD. Biomarker sets associated with healthy normal subjects may also be used in the methods and compositions provided herein. As discussed above, the methods and kits disclosed herein may relate to the use of particular sets of CSF proteins and / or serum lipids, e.g., “biomarker sets,” for purposes of identification, classification, or diagnosis. The invention may also use groups of biomarker sets, herein described as “classification sets.” The amounts or concentrations of biomarkers in a set (also known as a signature) may be determined and then used to evaluate the signature of the same set of biomarkers in a biological sample, such as by a measure of similarity between the sample signature and the reference signature. In some embodiments, the method involves measuring (or obtaining) the amounts or concentrations of two or more CSF proteins and / or serum lipids that are within a biomarker set and / or within a classification set. For example, in some embodiments, a biomarker set or a classification set may contain at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 33, 35, 38, 40, 43, 45, 48, 50, 53, 58, 63, 65, 68, 100, 120, 140, 142, 145, 147, 150, 152, 157, 160, 162, 167, 175, 180, 185, 190, 195, 200, or 300 biomarkers. In some embodiments, a biomarker set or a classification set contains no more than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 33, 35, 38, 40, 43, 45, 48, 50, 53, 58, 63, 65, 68, 100, 120, 140, 142, 145, 147, 150, 152, 157, 160, 162, 167, 175, 180, 185,190, 195, 200, or 300 biomarkers. In some embodiments, a classification set contains at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, or 25 different biomarker sets. In other embodiments, a classification panel contains no more than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25 different biomarker sets.

[0110] Biomarker sets may be chosen to accommodate adequate separation of healthy from disease or suspicious amounts or concentrations of CNS proteins and / or serum lipids. Training of the classifier, i.e., algorithm, can be performed on numerous biological samples, such as at least 50, 100, 200, 300, 400, 500, 600, 700, 800, 900, 1000, 1500, 2000, 2500, 3000, 3500, or 4000 biological samples (e.g., AD patient and healthy subject samples). In some embodiments, many training / test sets are used to develop the preliminary algorithm. All results may be obtained using a support vector machine model which is trained and tested in a cross-validated mode on the samples.

[0111] A statistical evaluation of the results of the molecular profiling may provide a quantitative value or values indicative of one or more of the following: the likelihood of developing MCI or AD; the likelihood of having MCI or AD; or the likelihood of a particular subtype or status of AD, e.g., ANLS attenuation or microglial flux capacity saturation. The data may be presented directly to the physician in its most useful form to guide patient care.

[0112] The results of the molecular profiling can be statistically evaluated using a number of methods known to the art including, but not limited to: the students T test, the two sided T test, pearson rank sum analysis, hidden markov model analysis, analysis of q-q plots, principal component analysis, one way ANOVA, two way ANOVA, LIMMA and the like.

[0113] In some embodiments of the present invention, the methods disclosed herein may provide a classification, identification, or diagnosis, e.g., healthy or diseased, that is between about 85% accurate and about 99% or about 100% accurate.

[0114] In embodiments, the disclosure provides computer-implemented methods for determining whether a subject is at risk for MCI or AD, whether a subject has MCI or AD, and the subtype classification of MCI or AD, e.g., whether the MCI or AD is associated with ANLS attenuation or microglial flux capacity saturation.

[0115] In embodiments, the computer-implemented method comprises: (a) receiving input data related to one or more of the markers disclosed herein, e.g., the amount or concentration of the one or more markers in biological sample(s) obtained from the subject; (b) applying a predictive model trained to classify subjects into one or more of the following types:(i) at risk for MCI or AD; (ii) active MCI or AD; (iii) the MCI or AD associated with ANLS attenuation; and / or (iv) MCI or AD associated with microglial flux capacity saturation; and (c) outputting the predicted subtype(s) classification determined according to (b) and, optionally, associated treatment recommendations. Methods of Treatment

[0116] The disclosure provides methods for treating a subject identified as being at risk for MCI or AD and subjects identified having MCI or AD.

[0117] In embodiments, the disclosure provides a method of treating, inhibiting, or slowing the progression of MCI or AD, the method comprising determining whether a subject is at risk for MCI or AD or whether a subject has MCI or AD according to any of the methods disclosed herein, and administering a therapeutic agent to a subject, if the subject was determined to be at risk for MCI or AD or determined to have MCI or AD.

[0118] Therapeutic agents for the treatment of MCI and AD are known in the art, and any may be used. Several drugs are available to manage symptoms and potentially slow the progression of MCI or AD. These include cholinesterase inhibitors like donepezil (Aricept), rivastigmine (Exelon), and galantamine (Razadyne), which help improve cognitive function by increasing acetylcholine levels in the brain. Additionally, memantine (Namenda) is used to treat moderate to severe Alzheimer's by regulating glutamate activity. Recent advancements include the approval of lecanemab (Leqembi) and donanemab (Kisunla) for early-stage Alzheimer's, which are amyloid-targeting antibodies that may slow disease progression. Additional agents that may be used to treat MCI or AD include but are not limited to: cholinesterase inhibitors, e.g., Donepezil (Aricept), Rivastigmine (Exelon), and Galantamine (Razadyne); Memantine (Namenda); anti-amyloid antibodies, e.g., Lecanemab (Leqembi) and Donanemab (Kisunla); and Brexpiprazole (Rexulti). Additional agents that may be used include, e.g., phosphodiesterase type 4 (PDE4) and phosphodiesterase type 5 (PDE5) inhibitors, such as, e.g., sildenafil. Modes of administration and dosages for these therapeutic agents and are known in the art.

[0119] In embodiments, a subject will be treated based on their AD status, i.e., as having either ANLS attenuation or microglial flux capacity saturation. In embodiments, the disclosure provides a method of treating, inhibiting, or slowing the progression of MCI or AD, the methodcomprising determining whether a subject has MCI or AD associated with ANLS attenuation or microglial flux capacity saturation according to any of the methods disclosed herein, and administering a therapeutic agent to a subject.

[0120] In embodiments, subjects determined to have a predisposition for or active ANLS attenuation are treated with increased sleep, phosphatidylcholine supplementation, and / or creatine supplementation. In embodiments, subjects determined to have a predisposition for or active microglial flux capacity saturation are treated with Omega3 supplementation, plasmalogen supplementation, and / or plasmalogen precursor supplementation. Kits

[0121] In embodiments, the disclosure provides kits that may be used to perform any of the assays disclosed herein, e.g., kits for determining whether a subject is at risk for MCI or AD or has active MCI or AD, or for determining whether the MCI or AD is associated with ANLS attenuation or microglial flux capacity saturation.

[0122] In embodiments, a kit comprises: (a) reagents for detecting one or more genetic mutations indicative of a predisposition to ANLS attenuation and / or microglial flux capacity saturation; (b) reagents for detecting an amount or concentration of one or more cerebral spinal fluid (CSF) proteins indicative of a predisposition to astrocyte-neuron lactate shuttle (ANLS) attenuation; (c) reagents for detecting an amount or concentration of one or more plasma lipids indicative of a predisposition to ANLS attenuation; (d) reagents for detecting an amount or concentration of one or more CSF proteins indicative of ANLS attenuation; (e) reagents for detecting an amount or concentration of plasma lipids indicative of ANLS attenuation; (f) reagents for detecting an amount or concentration of one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation; (g) reagents for detecting an amount or concentration of one or more CSF proteins indicative of microglial flux capacity saturation; and / or (h) reagents for detecting an amount or concentration of one or more plasma lipids indicative of microglial flux capacity saturation.

[0123] In embodiments, the kit only comprises reagents for determining an amount or concentration of the one or more proteins (or encoding mRNA) and / or lipids. The kit may further comprise reagents for determining an amount or concentration of one or more control proteins (or encoding mRNA) or lipids, e.g., those having an amount or concentration that does not significantly vary between healthy subjects and MCI or AD patients. In embodiments, the kit may further comprise instructions for interpreting the biomarker patterns determined when using the kit, e.g. to classify a subject being tested as being at risk for MCI or AD, having MCI or AD, and / or for determining whether the MCI or AD is associated with ANLS attenuation or microglial flux capacity saturation. The kit may comprise a software module that can be used to classify the subject being tested, e.g., by receiving biomarker measurements and providing automated classification. The software module may comprises an algorithm or classifier, such as one described herein, and may have been trained on biomarker data obtained from healthy subjects and / or MCI or AD patients.

[0124] In embodiments, the one or more genetic mutations indicative of a predisposition to ANLS attenuation are selected from mutations in INPP5D, PICALM, TSPOAP, TNIP1, and / or LXRA genes; the one or more CSF proteins indicative of a predisposition to ANLS attenuation are selected the CSF proteins set forth in Table 1, PLA2G7, and LPCAT2; the one or more plasma lipids indicative of a predisposition to ANLS attenuation are selected from the plasma lipids set forth in Table 1 and plasma lipids indicating high DHA in phospholipid or plasmalogen form; the one or more CSF proteins indicative of ANLS attenuation are selected from the CSF proteins set forth in Table 1 and the group consisting of RAB21, RAB7A, LDHB, LDHC, ALDOA, MDHC, KPYM, AATC, CNTN2, and TNR; the one or more plasma lipids indicative of ANLS attenuation are selected from the plasma lipids set forth in Table 1 and the group consisting of CER.D19.1.18.0 and CER.D19.1.16.0; the one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation are selected from the plasma lipids set forth in Table 2 and the group consisting of DHA, omega-3 fatty acids, optionally in the form of phospholipids or plasmalogens; the one or more CSF proteins indicative of microglial flux capacity saturation are selected from the CSF proteins set forth in Table 2 and the group consisting of TAU, CD44, CH3L1, GFAP, and GAS6; and / or the one or more plasma lipids indicative of microglial flux capacity saturation are selected from the plasma lipids set forth in Table 2 and the group consisting of DHA / EPA acylcarnities, optionally AC.22.6 or AC.20.5, and cholesterol esters, optionally CE.22.6 or CE.20.5, extra-large HDL lipoprotein cholesterol and / or cholesterol ester, optionally Nightingale assay XL.HDL.C or XL.HDL.CE.

[0125] Examples of reagents include nucleic acids, e.g., DNA probes, that specifically bind to a genetic mutation or to an mRNA or corresponding cDNA, antibodies that bind to CSF proteins, and a Nightingale NMR lipidomics panel. EXAMPLES Example 1 MULTIOMIC EVIDENCE FOR A UNIFIED MODEL OF ALZHEIMER’S DISEASE ETIOLOGY LINKING MICROGLIAL FLUX CAPACITY AND ASTROCYTE-NEURON METABOLIC BREAKDOWN

[0126] In the healthy brain, three major mechanisms regulate cholesterol efflux. The first is reverse cholesterol transport, in which astrocyte-derived APOE-containing lipoproteins mediate cholesterol redistribution and clearance via ABCA1 and ABCG1 transporters5,6. The second is neuronal conversion of cholesterol into 24S-hydroxycholesterol (24-OHC), which crosses the blood-brain barrier (BBB) and is metabolized in the liver7,8. The third pathway, particularly active in glial cells, is the conversion of cholesterol into 27-hydroxycholesterol (27-OHC), a locally acting oxysterol that activates liver X receptor alpha (LXRA) in the CNS9,10. Activation of LXRA represses SREBP2, the master regulator of de novo cholesterol synthesis, forming a feedback loop to maintain cellular cholesterol balance11,12.

[0127] Lipid droplets act as intracellular buffers of cholesterol and are known to vary with physiological state, including diurnal rhythms13–16. However, in AD, this system becomes overwhelmed. With age and increased myelin turnover, glia are exposed to rising levels of neuronal membrane debris, challenging their capacity to traffic cholesterol efficiently. The sheer volume of excess cholesterol saturates lipoprotein particles, causing lipid droplets to enlarge and persist17. As lipid droplets grow, they lose access to mitochondria, impairing the synthesis of 27-OHC. This disrupts LXRA activation, lifting repression on SREBP2, and allowing unchecked de novo cholesterol synthesis even in an already overloaded environment11,12. This breakdown in astrocytic cholesterol homeostasis impairs their ability to metabolically support neurons and contributes to early Alzheimer’s pathogenesis18.

[0128] To investigate how these disruptions contribute to cognitive decline, we analyzed baseline multiomic data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), stratifying participants into quartiles based on ADAS13 cognitive scores18. Our approach enriched for common pathological signatures independent of strict diagnostic categories. Across these quartiles, we identified two interdependent processes that appear to drive disease progression: saturation of microglial lipid flux capacity (“microglial flux capacity saturation”),and cholesterol-induced impairment of the astrocyteneuron lactate shuttle (ANLS)(“ANLS attenuation”). These feedback loops emerge upstream of amyloid and tau pathology and show evidence of mutual reinforcement.

[0129] Our unified model integrates lipidomic, proteomic, and genetic data, linking impaired lipid handling with neuronal energetic failure. The hypothesis arose during the development of a digital twin model for personalized interventions, where failure to accurately predict biomarker trajectories pointed toward missing biological constraints—specifically, glial processing-capacity limits. Together, these findings offer a new framework for understanding late-onset AD as a systems-level breakdown of metabolic homeostasis. RESULTS Data Selection / Organization

[0130] To investigate how glial lipid processing and neuronal metabolic support vary with cognitive status, we analyzed baseline multiomic data from ADNI. We organized the ADNI data to leverage the large number of longitudinal, multiomic patient data in a way that helped to discern molecular changes that coincided with cognition. We initially selected individuals for whom both untargeted baseline plasma lipidomic data19, and targeted baseline CSF proteomic data20existed. The CSF proteomics are a set of 48 proteins chosen based on their observed differential expression between case and control among postmortem brain samples20. We also identified individuals, for whom CSF Somascan proteomics21, and / or Nightingale plasma lipoproteins were available. A venn diagram of the sample overlap for all the data modalities used is found in FIG. 1B.

[0131] Participants were stratified into quartiles based on ADAS13 cognitive scores94, providing finer resolution across the cognitive spectrum independent of clinical diagnosis (FIG. 1C). Each quartile roughly represented a different stage of progression within the corresponding multiomic data (FIG.1A). The quartiles corresponded to: Q1 = 1-7; Q2 = 8-12; Q3 = 13-18; and Q4 = 19-52, respectively representing high cognitively normal, low cognitively normal, MCI and demented.

[0132] Out of a total of 2,428 ADNI subjects (ADNI, ADNI2, ADNI GO), 1,384 patients contained longitudinal lipidomics data, 602 of whom also had baseline CSF proteomics. Individuals for whom there were plasma lipidomics, CSF proteomics and CSF Somascans totaled 362, and individuals for whom there were plasma lipidomics, CSF proteomics and Nightingale lipoproteins totaled 601.

[0133] We selected the 602 participants with paired baseline plasma lipidomic and CSF proteomic data21, including the 362 participants with additional CSF SomaScan proteomic data20and 601 with Nightingale plasma lipoprotein profiles (FIGs.1A-1B). Within this cohort, we analyzed correlations between 781 plasma lipids and 48 CSF proteins previously shown to differ between cognitively normal and AD participants22–24.

[0134] Correlation matrices were constructed separately within each cognitive quartile. Strikingly, we observed significant differences between each quartile, with visible changes and patterns in each, with blue representing negative correlations, and orange representing positive correlations (FIG.1D). Distinct shifts in lipid–protein associations emerged as cognitive scores declined, with a pronounced reversal in correlation sign (blue vs orange) between Q2 and Q3 corresponding to the cognitive transition from normal to MCI (FIG. 1D). The most visually striking change we observed was found in Q3. These results identify clear changes in correlations between the CSF proteins and plasma lipids for each of the quartiles, providing a rich, multiomic canvas to explore stage-specific changes of disease progression.

[0135] To better organize these molecular changes, we performed unsupervised clustering of plasma lipids and CSF proteins. Plasma lipids grouped into ten clusters based on headgroup and fatty acid composition, including plasmalogens, cholesterol esters, and hexosylceramides (data not shown). CSF proteins grouped into three major clusters: complement / HDLrelated proteins, hemoglobin-related proteins, and a larger cluster encompassing five subgroups: astrogliosis, ANLS / glycolysis, hypoxia, synaptic, and phagocytic markers.

[0136] Principal component analysis (PCA) was applied to each cluster to summarize major axes of variation across individuals. We calculated the correlations for the first principle components (FPC) of each protein / lipid pair. Correlations between the first principal components of lipid and protein clusters were evaluated within each quartile (FIG. 2A), revealing specific disruptions in lipid metabolism and glial function as cognition declined. Consistent with the striking visual change we saw in the individual protein / lipid correlations in Q3 (FIG.1D), we observed a wide band of significant cluster FPC correlations across multiple protein clusters with a single lipid cluster (FIG. 2A). This lipid cluster contains 32 phospholipids, 23 of which are plasmalogens (each of which have either docosahexaenoic acid (DHA) or eicosapentaenoic acid (EPA)), three cholesterol esters (two containing DHA, one containing EPA), and one sphingomyelin species. DHA and EPA are two of the three lipids also known as omega-3s, making 34 of the 35 lipids in this group omega-3s.

[0137] Surprisingly, this analysis also revealed a significant FPC cluster correlation in Q1, seemingly independent of the visual change in FIG. 1D, Q3. Leveraging additional ADNI biomarkers other than those used for the clustering, we investigated specific protein cluster / lipid cluster correlations for each quartile. Quartile 1 Analysis

[0138] In Q1, we observed early changes in the inter-omic correlation structure between plasma lipids and CSF proteins. A significant association (p = 1.2e-5) was identified between a lipid cluster enriched for hexosylceramides and phospholipids (“hexceramides | phospholipids 18.1”) and a CSF protein cluster composed of complement- and HDL-associated proteins, including ALB, APOA4, APOC1, APOC2, CP, C9, KNG1, PGLYRP2, PON1, and THRB (Fig. 2A, Q1).

[0139] To visualize variation in these correlated signals, participants were divided into quadrants based on the first principal component (FPC) values of the lipid and protein clusters (Fig. 2B). We observed differences in WMH, with the greatest difference between the top left quadrant (high hexceramides; low complement / HDL) and bottom right quadrant (low hexceramides; high complement / HDL). Comparing the bottom-right (BR) and top-left (TL) quadrants revealed a 58% higher white matter hyperintensity (WMH) burden in the BR group (Fig. 2C), characterized by lower hexceramide and higher complement / HDL signals. These differences could be attributable to both axes, as comparisons between the left and right (low vs high complement / HDL proteins) were similar, as were top vs bottom (low vs high hexceramides). The greatest difference in WMH (top left vs bottom right) was 2.00 / 3.42, or a 58% increase. Individuals with high hexceramides and low levels of complement / HDL proteins had much lower WMH values. WMH is a marker of vascular dysfunction and impaired BBB integrity, suggesting that shifts in lipid–protein coupling may precede overt cognitive symptoms25,26.

[0140] Interestingly, the top left quadrant was predominantly female (56 / 63 individuals; 0.89%)(table X). A previous report identified sex differences in ceramide (but not hexceramide) levels and WMH29,30. We thus looked to see if we observed any sex differences in WMH (supplemental figure x), and did not see any statistically significant differences, though we cannot rule out a sex-specific difference.

[0141] We next used SomaScan proteomics to compare molecular signatures between individuals in the right versus left quadrants of the plot (s (high vs low of the original CSF complement / HDL protein cluster; Fig. 2D). Using cutoffs of 0.25 Log2 fold change and p-value< 0.001, this comparison revealed 72 differentially expressed CSF proteins (FDR < 0.05), with strong enrichment for complement and coagulation pathways (data not shown). Pathway enrichment of the SomaScan proteins overlapped with those measured by mass spectrometry, suggesting that complement system activation is a convergent signal across platforms.

[0142] In a separate comparison of top versus bottom quadrants—defined by lipid cluster FPC— with cutoffs of 0.15 Log2 fold change and p-value < 0.001, we identified 49 proteins (Fig. 2E, data not shown). EnrichR identified the Alternative Complement Pathway as most significant (adj p = 3.3e-4) (data not shown). Quartile 3 Analysis

[0143] Quartile 3 (Q3) revealed striking and biologically coherent inter-omic correlation reversals, marking a major shift in homeostatic balance. Astrogliosis protein cluster

[0144] We found a strong association in Q3 between the astrogliosis protein cluster (including YWHAB, YWHAZ, CHI3L1, and CD44) and the phospholipids | plasmalogens 22.6 lipid group (FIG.3A). CHI3L1 and CD44 are receptor–ligand partners linked to neuroinflammation and disease progression27, while YWHAB and YWHAZ bind GFAP, marking activated astrocytes28. To better understand the differences and processes responsible for individuals within Q3, we plotted individuals based on their FPC for astrogliosis proteins on the x-axis, and phospholipids | plasmalogens 22.6 on the y-axis. Participants in the BR quadrant showed a 55% increase in brain TAU PET levels and a 92% increase in hippocampal atrophy rate compared to the TL quadrant (FIG. 3B). These changes were accompanied by higher GFAP and GAS6 levels in the SomaScan data, known to be regulated by LXRA29.

[0145] Supporting impaired lipid metabolism, the BR quadrant also showed increased acylcarnitines and cholesterol esters, suggesting reduced lipid oxidation and cholesterol efflux (FIGS.3B and 3C). As part of normal lipid homeostasis, fatty acids, cholesterol and other lipid species become esterified and stored in lipid droplets, with PI3K activation substantially increasing this process, and increasing the size of the lipid droplets. Large lipid droplets are less able to be transported to mitochondria by the actin cytoskeleton, where their cholesterol load is partially metabolized by CYP27A1 (found exclusively on the inner membrane of mitochondria) into 27OHC, and acyl-CoA from LD neutral lipid breakdown is metabolized to acylcarnitines. We looked for evidence of this in Q3 by comparing the lipids in the top-left quadrant of Q3 with those in the bottom-right. Of the 35 acylcarnitines in the lipidomic data, we found 7 acylcarnitines to be significantly different between the two quadrants with all ofthem trending in the expected direction, including both the DHA and EPA species (FIG. 3B). We see a similar pattern in cholesterol esters. Of the 39 cholesterol esters, we found 11 to be significantly different between the two quadrants, all of them trending in the expected direction (FIG. 3B).

[0146] To determine whether these elevated lipid species could be traced to the brain, we incorporated plasma lipoprotein data from the Nightingale NMR platform. The above quadrant-specific cholesterol ester data was echoed in Nightingale HDL markers for cholesterol and cholesterol esters in XL.HDL particles (FIG. 3C), a size class consistent with the sized observed in CNS lipoproteins from post-mortem AD patients35. Interestingly, HDL3 cholesterol levels are also elevated in the top-left quadrant (FIG.3C). Notably, the TL quadrant showed higher levels of a CNS-specific extremely large HDL particle — a lipoprotein particle class known to originate exclusively from the brain30,31. The detection of this CNS derived particle in peripheral blood provides key evidence that lipid signals observed in plasma may reflect active export from the brain. This observation supports the idea that microglial lipid processing capacity is exceeded, with reduced cholesterol and lipids trafficked out of the CNS into circulation.

[0147] Together, our observations of the bottom-right vs. top-left individuals are consistent with reduced LD processing (acylcarnitines) and cholesterol efflux (cholesterol esters) out of the brain, into the plasma, at the same time when neuronal loss is known to occur in the hippocampus (hippocampal atrophy) and temporal cortex100. Considering mass balance, with fewer lipids exiting the brain, coupled with known neuronal loss, and the known role of microglial phagocytosis, these data point to an accumulation of neuronal debris in microglia and / or astrocytes, reflective of the known accumulation of LDs in Alzheimer’s. Furthermore, individuals in the bottom right, where more neuronal death is occuring (greater tau), display even less cholesterol efflux, whereas individuals in the top left, while still progressing with the disease, display greater efflux. ANLS / glycolysis protein cluster

[0148] The strongest protein–lipid association in Q3 was between the ANLS / glycolysis protein cluster (including LDHB, LDHC, CD44, ALDOA, PKM, MDH1, GAA, and ENO1) and the lipid cluster phospholipids | plasmalogens 22.6, which includes many Docosahexaenoic acid (DHA)- and Eicosapentaenoic acid (EPA)-containing lipids (FIG. 3D). Several individual protein–lipid pairs within this quadrant showed significant associations, including LDHB and CE.22.6 (FIG. 3E). The most significant FPC protein correlation coefficient was withANLS / glycolysis at a p-value of 2.0e-3. Among the lipids and proteins that make up the FPCs, several individual correlations had significant p-values, as shown in FIG. 3F.

[0149] To better understand the differences and processes responsible for individuals within Q3, we plotted individuals based on their FPC for ANLS / glycolysis proteins on the x-axis, and phospholipids | plasmalogens 22.6 on the y-axis (FIG. 3D). Participants in the BR quadrant, characterized by higher glycolysis and lower plasmalogen signals, exhibited a 65% higher median brain TAU PET level than those in the TL quadrant (data not shown). These results are consistent with the idea that high levels of omega-3s are protective of metabolic stress, as indicated by high levels of ANLS / glycolysis proteins (and higher Tau), and support the hypothesis that impaired lipid processing results in metabolic stress as reflected in compensatory changes in glycolytic proteins.

[0150] As omega-3 are most commonly associated with brain health101, 102, and reduce microglial polarization39, researchers have created an omega-3 index to help provide a standard quantitative assessment of omega-3s in the body103. To probe whether the observed metabolic vulnerability was related to omega-3 availability, we estimated the omega-3 index using Nightingale NMR data on DHA- and non-DHA–containing HDL particles32–34. Nightingale consists of an array of lipoprotein particle size measurements and lipid compositions. The recommended omega-3 index range is between 8-11104. The median omega-3 index in the TL quadrant was within the clinically recommended range (>8.0), while the BR quadrant had a significantly lower median (6.7; p = 8.6e-7) (Fig. 3F). This finding aligns with the known protective effects of omega-3s against AD progression and highlights a mechanistic link: omega-3s and plasmalogens are critical for lipid droplet trafficking35,36.

[0151] Individuals in the lower-left quadrant had similarly low levels of Tau relative to individuals in the top-left, in spite of low omega-3 levels. To identify potential explanations for this observation, we compared the lipidomic levels of all individuals between the left and right quadrants (low vs high levels of the ANLS / glycolysis proteins). We observed higher plasmalogens and a number of plasmalogen-precursors: monoalkyl-diacylglycerols, in individuals with lower expression levels of the ANLS / glycolysis proteins (FIG. 3E). Higher plasmalogens and higher omega-3 are both known to result in lipid droplets of a smaller size105,106.

[0152] Taken together, these data reveal a coherent axis connecting decreased omega-3 and plasmalogen levels with disrupted expression of ANLS and glycolytic proteins. This dysfunction appears to originate from a buildup of neuronal membrane debris—includingmyelin fragments—that must be cleared by glia. When the burden of this cholesterol-rich material exceeds the trafficking capacity of astrocytic lipoproteins and lipid droplet systems, astrocytes become overloaded. The resulting membrane saturation impairs essential functions such as GPCR signaling and cholesterol export, disrupting astrocyte–neuron communication. In particular, ANLS becomes compromised, depriving neurons of a key metabolic support pathway.

[0153] This metabolic fragility may represent a critical inflection point that triggers early tau accumulation and neurodegeneration. Prior studies show that neurons under energetic stress activate AMP-activated protein kinase (AMPK) to compensate for ATP deficits37–39. However, in the absence of sufficient astrocytic support, AMPK activation becomes maladaptive— contributing to calcium dysregulation and tau phosphorylation40–42. This cascade promotes synaptic dysfunction and accelerates neurodegenerative progression. Quartile 2 Analysis

[0154] While Quartile 2 (Q2) initially showed some visible changes in overall inter-omic correlation structure (FIG. 1D), no strong lipid–protein cluster correlations emerged when assessed by principal component analysis (FIG. 2A, Q2). This suggests an early phase of systemic adjustments without system-wide disruption.

[0155] However, the strong astrocytic and tau pathology signals in Quartile 3 (Q3) prompted us to revisit Quartile 2 (Q2) to search for earlier signs of metabolic stress and impaired cholesterol flux — prior to measurable cognitive decline or glial activation. Although no significant lipid–protein cluster correlations emerged at the FPC level (Fig. 2A, Q2), we identified several FPC pairs with low p-values and investigated their component lipids and proteins for meaningful associations.

[0156] This analysis revealed a consistent association between the ANLS / glycolysis protein cluster and a ceramide-rich lipid cluster, specifically centered around CER.D19.1.18.0, an oddchain sphingolipid. We stratified participants by principal component scores from these two clusters (Fig. 4A), revealing a 59% increase in brain TAU PET and higher hippocampal atrophy rate in the bottom-right (BR) quadrant compared to the top-left (TL) (Fig.4B). Among individual correlations, LDHB–CER.D19.1.18.0 was the strongest (p = 0.0046), with broad correlations seen across glycolytic enzymes (Fig. 4C).

[0157] To explore the biology behind CER.D19, we performed a proteomic correlation screen using SomaScan CSF data. We identified 24 proteins significantly associated with CER.D19.1.18.0 (p < 0.003), many of which participate in lipid droplet trafficking (RAB21,RAB7A)43, cholesterol esterification (SLC27A2)44–46, and oxysterol signaling (RAC1, FGF1, INPP5A)47. These were notably Q2-specific and not observed in Q1 or Q3 (Fig. 4D; data not shown). Pathway analysis via EnrichR identified bile acid biosynthesis via 24hydroxycholesterol as the top hit (adjusted p = 6.1e-3; data not shown). The CER.D19 odd- chain fatty acid is synthesized using propionyl-CoA, a byproduct of cholesterol side-chain βoxidation during bile acid synthesis48–56.

[0158] This is significant because the conversion of cholesterol to 27-hydroxycholesterol represents one of the routes for cholesterol export from the brain, enabling it to cross the bloodbrain barrier and be metabolized into bile acids in the liver. The proteins identified here map directly to the molecular machinery required for this process, suggesting that CER.D19 levels correlate with the point where glial cells begin activating compensatory mechanisms to manage excess cholesterol through this export route.

[0159] Given this connection to bile acid metabolism, we asked whether any known AD associated risk loci might be enriched among individuals with altered CER.D19 levels. To do this, we curated a list of 101 genome-wide significant AD GWAS loci from recent meta- analyses and tested for associations with CER.D19.1.18.0. This approach revealed variants near PICALM, INPP5D, TSPOAP1, and TNIP1—all genes implicated in cholesterol trafficking, mitochondrial lipid processing, or bile acid synthesis57(Fig. 5C; data not shown). PICALM is involved in vesicular trafficking of cholesterol to the mitochondria where CYP27A1 initiates the conversion of cholesterol to bile acids and propionyl-CoA107. TNIP1 inhibits NF-KB signaling which downregulates CYP27A1108, 109. INPP5D negatively regulates PI3K / Akt signaling, increasing ABCA1and TSPOAP helps transport cholesterol from the outer to the inner mitochondrial membrane where CYP27A1 resides112. Additionally, SNP rs686548 in SPTLC3, the rate-limiting enzyme for odd-chain ceramide synthesis, was identified as the top CER.D19-associated SNP via Baker PheWeb (data not shown). Because of the known role of LXRA in lipid metabolism in the mitochondria38, we also looked to see if any LXRA-associated (NR1H3) SNPs correlated and found rs7120118 and rs3758673 to correlate with CER.D19.1.18.0.

[0160] These genetic findings reinforce the proteomic evidence, pointing to upstream regulatory steps in cholesterol clearance. Many of these genes control lipid droplet formation, vesicular trafficking, and mitochondrial cholesterol import—critical steps that precede conversion of cholesterol to 27-hydroxycholesterol for LXRA and SREBP2 signaling11,12. Their enrichment among individuals with altered CER.D19 levels suggests a geneticallyinfluenced bottleneck in cholesterol flux capacity that may predispose certain individuals to early metabolic breakdown in AD.

[0161] Having identified evidence for increased bile acid synthesis being correlated with metabolic stress in Q2, we looked for similar evidence of the other key mechanism for excess cholesterol clearance, efflux to CSF lipoproteins. We examined cholesterol handling enzymes and their relationship to ANLS / glycolysis proteins. We evaluated the lipid species between the top and bottom Q2 quadrants to characterize the lipid signature and noticed a number of lysophosphatidylcholine (LPC) lipid species as significant (data not shown). Phosphatidylcholine (PC) is a rate limiting reactant for free cholesterol esterification onto lipoproteins via ABCA1|LCAT, generating LPC in the process113, 114.

[0162] We then evaluated how the key enzymes involved in the interconversion of PC to LPC correlated with LDHB (FIG. 5B). We observed a significant positive correlation between plasma LPA2G7 and CSF LDHB (p-value = 4.04e-5) and a significant negative correlation between plasma LPCAT2 and CSF LDHB (pvalue = 3.6e-3) (FIG. 5A). LDHB, a central member of this protein cluster, serves as an indicator of neuronal metabolic stress and astrocyte–neuron metabolic coupling58,59. Notably, LDHB showed a positive correlation with CSF PLA2G7 and a negative correlation with LPCAT2 (Fig.5A), consistent with the opposing roles these enzymes play in phosphatidylcholine (PC) metabolism and cholesterol esterification60,61. These relationships suggest that dysregulated PC turnover—via increased PLA2G7 or decreased LPCAT2—could impair cholesterol packaging into APOE lipoproteins, leading to astrocytic overload and downstream metabolic stress.

[0163] Additional CSF markers associated with CER.D19 included TNR, CNTN2 and CHGA, both of which also correlated with LDHB across quartiles (Fig.5D). TNR and CNTN2 are two cell adhesion proteins that can bind directly to each other and are known to communication between astrocytes and neurons at the Nodes of Ranvier and synapses–two sites at which astrocyte-neuron lactate shuttle (ANLS) occur115, 116. CNTN2, a cell adhesion protein that localizes to nodes of Ranvier, peaked in Q2 and Q3, while CHGA, co-secreted with norepinephrine, remained stable from Q2 through Q462. These proteins are involved in astrocyte–neuron communication and reflect the integrity of GPCR-mediated signaling cascades. We found chromogranin to be highly correlated with all the glycolytic CSF proteins, including LDHB (FIGs. 5C and 5D). All of these quartile 2 multi-omic correlations are summarized in the graph shown in FIG. 5D. LDHB, a key member of the ANLS / glycolysis protein cluster, serves as an indicator of neuronal metabolic stress and astrocyte–neuronmetabolic decoupling38. Importantly, excess cholesterol in astrocytic membranes has been shown to impair GPCR signaling63,64, including pathways regulated by norepinephrine and glutamate, which are critical for triggering ANLS65. Thus, the co-variation of LDHB with CNTN2 and CHGA suggests that rising intracellular cholesterol may begin to impair signaling at precisely the point when microglial cholesterol processing begins to saturate. A summary of these integrated lipidomic, proteomic, and genomic correlations is shown in Fig. 5E.

[0164] Seeing correlations between LDHB (indicator of neuronal metabolic stress), homeostasis and astrogliosis, we wanted to observe these markers across all quartiles (FIG. 5E). We found GFAP to have the strongest correlation with LDHB in Q3, while we observed chromogranin (CgA) to correlate with LDHB earlier in Q2, continuing on through Q4. CNTN2, a marker of astrocyte-neuron communication, correlated with LDHB in Q2 and Q3, but not Q4. As with GFAP, the lack of correlation between CNTN2 and LDHB in Q4 might be due to excessive dysregulation of many signaling pathways. LPCAT2’s correlation with LDHB was strongest in Q2, be it modest, while Lp-PLA2 correlated with LDHB in all quartiles. Collectively, in Q2, we find evidence for proteins involved in ANLS / glycolysis that correlate with both cholesterol efflux and bile acid synthesis. We found contextual genetic evidence for AD GWAS variants correlating with CER.D19, which we found to be correlated with increased expression of the ANLS / glocolysis proteins. The correlations we identifiedd are consistent with elevated cellular cholesterol, likely due to the processing / trafficking of neuronal debris. These data are resonant with the dysfunctions we observed in Q3.

[0165] Taken together, these results suggest that Q2 is not a silent interval, but a biologically active phase of incipient flux imbalance, in which microglia lipid handling begins to saturate. CER.D19 appears to mark this metabolic tension point, linking excess intracellular cholesterol with early disruption of signaling-based astrocyte–neuron coordination and compensatory mitochondrial processing. This lipid accumulation appears to interfere with GPCR signaling necessary for lactate shuttling via the ANLS, leading to rising LDHB levels as neurons experience mounting metabolic stress. These changes precede overt glial activation and tau pathology, reinforcing the hypothesis that early cholesterol buildup—not depletion—initiates the breakdown of homeostatic support systems in AD. METHODS

[0166] The results disclosed here are in whole or in part based on data obtained from the AD Knowledge Portal (https: / / adknowledgeportal.org).

[0167] ADNI metabolomics data from the Baker lipidomics kit is available at the AD Knowledge Portal under https: / / adknowledgeportal.synapse.org / Explore / Studies / DetailsPage?Study=syn5592519 and https: / / adknowledgeportal.synapse.org / Explore / Studies / DetailsPage / StudyDetails?Study=syn 24989039, the full complement of clinical and demographic data for the ADNI cohorts are hosted on the LONI data sharing platform and can be requested at http: / / adni.loni.usc.edu / datasamples / access-data / . ADNI Data

[0168] All ADNI data was downloaded from https: / / adni.loni.usc.edu / . ADNI Genetics

[0169] A harmonized array dataset for ADNI has been published previously30,31. Briefly, genotype array data from ADNI1, ADNI2 / GO, ADNI Omni 2.5, and ADNI 3 were processed and imputed using the TOPMED imputation server. Variants were removed for missingness (5%), low minor allele frequency (<1%), or if out of Hardy Weinberg Equilibrium. Samples were removed for missingness (1%), relatedness (pi-hat>.25), inconsistency between reported sex and genetic sex, and for heterozygosity (>5standard deviations). Imputed data were called for variants with high imputation quality (R2>0.80). Participants that were present in more than one array were dropped from the array of lower quality. After imputation and QC, 1,559 participants and 8,028,923 variants were available for analysis. Baker Metabolomic PhWAS

[0170] https: / / metabolomics.baker.edu.au / pheweb_standard / region / Cerd191160 / 20:1277352 113173521 Clustering and Principal Component Analysis

[0171] Clustering, principal component analysis, linear regressions, and statistical analyses were performed using Wolfram Mathematica v14.0. Hippocampal Atrophy Rate Calculation

[0172] For each individual, hippocampal atrophy rate was calculated by linear regression over all available time points measures of hippocampus volume, and presented as percent per year normalized to volume at baseline. Omega-3 Index Calculation

[0173] We calculated the omega-3 index based on the description found in Schuchardt and et al Schacky et al66. Briefly, we log-transformed the O3I and / or DHA% and non-DHA% priorto fitting a regression equation. The additive model containing two terms was determined to be the final prediction model, including β-coefficients for term: eO3I = 2·629 × NMR DHA% þ 0·4673 × NMR non-DHA% − 0·1014. The correlation of the eO3I vs observed O3I values was calculated for all ADNI participants. DISCUSSION

[0174] AD has long been conceptualized as a proteinopathy, molecularly defined by extracellular amyloid-beta plaques and intracellular tau tangles. While these hallmark pathologies are correlated with disease progression and cognitive decline, there remains a disconnect between their presence and our mechanistic understanding of disease etiology. This has led to growing interest in metabolic and glial contributions to early vulnerability. In this study, we applied integrated lipidomic and proteomic profiling of CSF and matched plasma samples— stratified by cognitive performance quartiles—to uncover early molecular changes consistent with impaired cholesterol efflux. These were accompanied by genetic associations and proteomic shifts implicating microglial lipid handling, suggesting that a failure in metabolic flux capacity precedes and potentiates the downstream cascades commonly associated with AD.

[0175] Our findings converge on a model of progressive collapse in cholesterol and lipid homeostasis, driven by overwhelmed glial flux capacity. In Quartile 1 (Q1), lipid–protein correlation structures were still largely intact, with subtle shifts suggesting early adaptive changes. Notably, several of the protein and lipid clusters showing early correlation shifts are enriched for markers of vascular integrity and hemoglobin metabolism, consistent with prior reports of early BBB disruption in AD47. These associations suggest that vascular leakage may precede metabolic strain and may reflect compensatory responses to reduced cerebral perfusion.

[0176] In Quartile 2 (Q2), the absence of system-wide lipid–protein correlations initially suggested a transitional or uninformative phase. However, retrospective analysis motivated by the strong astroglial and metabolic signals observed in Q3 revealed that Q2 is in fact a critical dual inflection point. The emergence of CER.D19, an odd-chain ceramide linked to propionylCoA metabolism48–56, cholesterol esterification, and mitochondrial cholesterol trafficking, marked the beginning of detectable metabolic stress. Participants with elevated glycolytic protein expression and elevated CER.D19 levels showed higher tau burden and hippocampal atrophy, suggesting that cholesterol clearance capacity was disrupting neuronal metabolic support.

[0177] These observations were reinforced by the appearance of CNS-specific HDL markers in plasma and correlations between CER.D19 and proteins involved in bile acid biosynthesis and oxysterol metabolism. Furthermore, GWAS loci associated with CER.D19 levels included variants near several established AD risk genes: TSPOAP1, INPP5D, PICALM, TNIP1, and both NR1H3 (LXRA), and SPTLC3. These genes are involved in mitochondrial cholesterol trafficking, vesicular transport, nuclear lipid sensing, lipid droplet processing, and includes enzymes that produce CER.D19 as a byproduct of broader lipid metabolic processes67. Their enrichment supports the hypothesis that disrupted cholesterol handling is a genetically encoded vulnerability in AD and, more significantly, provides functional context for how these GWAS loci contribute to disease etiology—something that has remained elusive for many AD risk variants. Together, these data suggest that CER.D19 acts as a sentinel lipid, reflecting the point at which glial lipid flux systems begin to saturate.

[0178] By Quartile 3 (Q3), compensatory mechanisms showed widespread changes, with clear signs of metabolic and glial strain. Plasma cholesterol ester levels were inversely correlated with astrocyte-derived CNS protein markers, and participants with lower cholesterol ester signal and higher astrocytic stress showed increased tau burden. These patterns suggest that astrocytes are assisting with processing of excess cell debris, buffer excess lipids, impairing their ability to provide metabolic support for neurons via ANLS.

[0179] Our model shifts the focus of early AD pathogenesis away from amyloid and toward the failure of glial lipid processing and cholesterol clearance. Importantly, our results highlight a mechanistic explanation for the long-observed protective effects of omega-3 fatty acids: participants with higher omega-3 index showed greater preservation of plasmalogens, improved HDL profiles, and reduced tau pathology68. These effects likely stem from enhanced lipid droplet trafficking and membrane remodeling. Omega-3 polyunsaturated fatty acids, particularly DHA, contain multiple cis double bonds that introduce curvature into the lipid structure, promoting the formation of smaller, more dynamic lipid droplets69. This biophysical property facilitates mitochondrial docking and efficient cholesterol oxidation via oxysterol synthesis, thereby enhancing cholesterol efflux. In this way, omega-3s not only improve membrane composition but also mitigate cholesterol overload in astrocytes, helping preserve metabolic support to neurons during early stage disease.

[0180] Our unifying hypothesis posits that AD arises when the brain’s capacity to process, store, and export cholesterol is overwhelmed. Initially, astrocytes and microglia buffer this excess through lipid droplet formation and metabolism. However, as cholesterol-rich neuronaldebris accumulates—particularly from aging and myelin turnover—these clearance systems become saturated. Enlarged and persistent lipid droplets disrupt their dynamic interactions with mitochondria, impairing lipid transfer and mitochondrial function. This breakdown leads to mitochondrial stress, disrupted ANLS, and subsequent tau aggregation. Rather than viewing amyloid and tau as primary instigators, we propose they are downstream consequences of unresolved lipid accumulation and failed clearance mechanisms.

[0181] Our hypothesis also provides a coherent framework to reinterpret longstanding genetic observations. APOE4, the strongest genetic risk factor for late-onset AD, has a shorter half- life70 and reduced lipid-carrying capacity71compared to other isoforms, as well as altered binding affinities for LDLR and LRP171. These differences impair the recycling of cholesterol- rich APOE particles back into astrocytes, increasing the likelihood of astrocytic cholesterol overload and impairing ANLS. In contrast, APOE2 carriers show more efficient lipid recycling and reduced astrocytic burden, preserving metabolic support, likely due to their reduced affinities for LDLR and LRP172. The Christchurch mutation (APOE3 R136S), which disrupts APOE binding to heparan sulfate proteoglycans, likely reduces retention of cholesterol-laden particles in the extracellular space and inhibits their uptake, protecting astrocytes from saturation. This mechanism offers a plausible explanation for why the Christchurch variant mitigates the impact of a pathogenic PSEN1 mutation11,73–75—by preserving cholesterol flux equilibrium and delaying the onset of glial and neuronal dysfunction despite high amyloid burden.

[0182] These genetic effects align with our data, where rising CER.D19 levels are not interpreted as causal drivers of pathology, but rather as sentinel markers of disrupted cholesterol metabolism and flux capacity. Importantly, our findings intersect with known regulators of cholesterol homeostasis, including LXRA and SREBP211. Under conditions of excess lipid accumulation, particularly when lipid droplets become enlarged and fail to fuse with mitochondria, mitochondrial β-oxidation is impaired. This disruption prevents 27-OHC synthesis and feedback inhibition of SREBP2, allowing de novo cholesterol synthesis to continue unchecked—even in the setting of cholesterol overload. Julia TCW and colleagues previously demonstrated that impaired lipid droplet turnover maintains active SREBP2 signaling and sustained cholesterol production33,68. Omega-3 fatty acids may counteract this process by preventing the formation of oversized lipid droplets, thereby maintaining droplet– mitochondrial interactions and allowing appropriate downregulation of SREBP231,76,77. Thismechanism helps preserve cholesterol homeostasis and may underlie the protective metabolic effects associated with higher omega-3 index78.

[0183] Our unified hypothesis integrates these findings into a model of Alzheimer's pathophysiology, outlined in Fig. 6. The process begins when microglia phagocytose neuronal and myelin debris, loading debris-derived cholesterol onto de novo-synthesized APOE lipoproteins (Step 1). These lipoproteins enter a shared pool with astrocyte-synthesized particles, but as particle size increases due to debris loading, cholesterol efflux capacity declines (Step 2). The accumulation of cholesterol in astrocytes and neurons impairs membrane dynamics, disrupts GPCR signaling, and reduces ANLS-mediated lactate delivery to neurons (Step 3). Neurons experiencing metabolic stress activate AMPK but cannot compensate for the energetic shortfall39,79, leading to calcium dysregulation79,80, tau phosphorylation, and progressive neurodegeneration (Step 4). As microglial flux capacity saturates, astrocytes assume a phagocytic role, accelerating cholesterol accumulation and triggering astrogliosis (Step 5). The final bottleneck occurs when cholesterol processing stalls at the level of lipid droplet–mitochondrial coupling, a step genetically influenced by loci such as INPP5D, TSPOAP1, PICALM, TNIP1, SPTLC3, and NR1H3 (LXRA), all of which correlated with CER.D19.1.18.0. levels in our analysis (Step 6).

[0184] Our model reframes AD as a failure of coordinated lipid handling and metabolic support, upstream of amyloid and tau deposition. It provides a mechanistic framework to explain decades of AD genetics, proteomics, and lipidomic findings, while unifying disparate observations into a coherent system-level pathology. The root cause of vulnerability may lie not in protein aggregation per se, but in the brain's failure to remove cholesterol debris without dysregulating homeostatic processes. The appearance of CER.D19 and associated proteomic and genomic signals in Q2 provides a mechanistic entry point into this process. This research emerged from our efforts to construct digital twins of aging individuals. While our models performed well in predicting biomarker trajectories during normal cognition, their accuracy deteriorated upon reaching the mild cognitive impairment (MCI) phase. Our digital twin models consistently pointed to unexplained variability associated with cholesterol metabolism and lipid handling, particularly in glial compartments. This led us to hypothesize that a key missing axis of variation was microglial flux capacity—the ability of glial cells to manage cholesterol clearance and maintain metabolic equilibrium. The findings presented here validate that hypothesis.

[0185] Taken together, these findings suggest that AD may benefit from being reframed in terms of etiology—as a systems-level breakdown in cholesterol flux and glial metabolic capacity, rather than solely a disorder of protein aggregation. Our hypothesis offers a potentially unifying framework through which many known risk and resilience factors— including genetic variants, lipid signatures, and environmental modifiers—can be interpreted. While it does not resolve the long-standing debate over AD causation or definition, it contributes a mechanistically grounded perspective focused on the biophysical dynamics of cholesterol. By incorporating concepts of mass balance, metabolic flux, and feedback regulation, this model helps contextualize the complex and heterogeneous molecular patterns observed across patients, and can guide future efforts to refine our understanding of AD etiology, progression and personalized therapeutic strategies. *********************************************** Bibliography 1. Alzheimer, A., Stelzmann, R. A., Schnitzlein, H. N. & Murtagh, F. R. An english translation of alzheimer’s 1907 paper, “über eine eigenartige erkankung der hirnrinde.” Clin. Anat.8, 429– 431 (1995). 2. Zhao, N. et al. Alzheimer’s Risk Factors Age, APOE Genotype, and Sex Drive Distinct Molecular Pathways. Neuron 106, 727-742.e6 (2020). 3. Romero-Molina, C., Garretti, F., Andrews, S. J., Marcora, E. & Goate, A. M. Microglial efferocytosis: Diving into the Alzheimer’s disease gene pool. Neuron 110, 3513–3533 (2022). 4. Consortium, A. et al. Rare coding variants in PLCG2, ABI3, and TREM2 implicate microglialmediated innate immunity in Alzheimer’s disease. Nat. Genet. 49, 1373–1384 (2017). 5. Marchi, C. et al. ABCA1- and ABCG1-mediated cholesterol efflux capacity of cerebrospinal fluid is impaired in Alzheimer’s disease. J Lipid Res 60, 1449–1456 (2019). 6. Zhang, J. & Liu, Q. Cholesterol metabolism and homeostasis in the brain. Protein & Cell 6, (2015). 7. Sodero, A. O. 24S‐hydroxycholesterol: Cellular effects and variations in brain diseases. J. Neurochem. 157, 899–918 (2021). 8. Björkhem, I. & Meaney, S. Brain Cholesterol: Long Secret Life Behind a Barrier. Arter.,Thromb., Vasc. Biol. 24, 806–815 (2004). 9. Fu, X. et al. 27-Hydroxycholesterol Is an Endogenous Ligand for Liver X Receptor in Cholesterol-loaded Cells*. J. Biol. Chem. 276, 38378–38387 (2001). 10. Spanos, F. et al. Impaired astrocytic synaptic function by peripheral cholesterol metabolite 27-hydroxycholesterol. Front. Cell. Neurosci. 18, 1347535 (2024). 11. TCW, J. et al. Cholesterol and matrisome pathways dysregulated in astrocytes and microglia. Cell 185, 2213-2233.e25 (2022). 12. Wang, N.-Q., Sun, P.-X., Shen, Q.-Q. & Deng, M.-Y. Cholesterol Metabolism in CNS Diseases: The Potential of SREBP2 and LXR as Therapeutic Targets. Mol. Neurobiol. 62, 6283–6307 (2025). 13. Chen, J., Zhang, X., Kusumo, H., Costa, L. G. & Guizzetti, M. Cholesterol efflux is differentially regulated in neurons and astrocytes: Implications for brain cholesterol homeostasis. Biochim. Biophys. Acta (BBA) - Mol. Cell Biol. Lipids 1831, 263–275 (2013). 14. Loving, B. A. et al. Lipoprotein Lipase Regulates Microglial Lipid Droplet Accumulation. Cells 10, 198 (2021). 15. Hong, B. V. et al. Altered HDL Cholesterol Efflux Capacity and LCAT Activity is APOESpecific in Alzheimer Disease and Mild Cognitive Impairment Patients. Alzheimer’s Dement. 18, (2022). 16. Husain, M. A., Laurent, B. & Plourde, M. APOE and Alzheimer’s Disease: From Lipid Transport to Physiopathology and Therapeutics. Front Neurosci-switz 15, 630502 (2021). 17. Staurenghi, E. et al. Cholesterol Dysmetabolism in Alzheimer’s Disease: A Starring Role for Astrocytes? Antioxidants 10, 1890 (2021). 18. Mofrad, S. A., Lundervold, A. J., Vik, A. & Lundervold, A. S. Cognitive and MRI trajectories for prediction of Alzheimer’s disease. Sci. Rep. 11, 2122 (2021). 19. Huynh, K. et al. Concordant peripheral lipidome signatures in two large clinical studies of Alzheimer’s disease. Nat Commun 11, 5698 (2020). 20. Haque, R. et al. A protein panel in cerebrospinal fluid for diagnostic and predictive assessment of Alzheimer’s disease. Sci. Transl. Med. 15, eadg4122 (2023). 21. Cruchaga, C. et al. Multi-cohort cerebrospinal fluid proteomics identifies robust molecular signatures for asymptomatic and symptomatic Alzheimer’s disease. Res. Sq. rs.3.rs-3631708(2024) doi:10.21203 / rs.3.rs-3631708 / v1. 22. Li, Y. et al. Higher blood–brain barrier permeability is associated with higher white matter hyperintensities burden. J. Neurol. 264, 1474–1481 (2017). 23. Hainsworth, A. H. et al. Neuropathology of White Matter Lesions, Blood–Brain Barrier Dysfunction, and Dementia. Stroke 48, 2799–2804 (2017). 24. Gupta, N. et al. White Matter Hyperintensity-Associated Blood-Brain Barrier Disruption and Vascular Risk Factors. J. Stroke Cerebrovasc. Dis. 27, 466–471 (2018). 25. Geng, B. et al. Chitinase 3-like 1-CD44 interaction promotes metastasis and epithelial- tomesenchymal transition through β-catenin / Erk / Akt signaling in gastric cancer. J. Exp. Clin. Cancer Res. 37, 208 (2018). 26. Connolly, K. et al. Potential role of chitinase‐3‐like protein 1 (CHI3L1 / YKL‐40) in neurodegeneration and Alzheimer’s disease. Alzheimer’s Dement. 19, 9–24 (2023). 27. Satoh, J., Yamamura, T. & Arima, K. The 14-3-3 Protein ε Isoform Expressed in Reactive Astrocytes in Demyelinating Lesions of Multiple Sclerosis Binds to Vimentin and Glial Fibrillary Acidic Protein in Cultured Human Astrocytes. Am. J. Pathol. 165, 577–592 (2004). 28. Kim, S.-Y. et al. Liver X receptor and STAT1 cooperate downstream of Gas6 / Mer to induce anti-inflammatory arginase 2 expression in macrophages. Sci. Rep. 6, 29673 (2016). 29. Nelson, T. J. & Sen, A. Apolipoprotein E particle size is increased in Alzheimer’s disease. Alzheimer’s Dementia Diagnosis Assess Dis Monit 11, 10–18 (2018). 30. Schuchardt, J. P., Tintle, N., Westra, J. & Harris, W. S. Estimation and predictors of the Omega-3 Index in the UK Biobank. Br. J. Nutr. 130, 312–322 (2023). 31. Schacky, C. von. Omega-3 index in 2018 / 19. Proc. Nutr. Soc. 79, 381–387 (2020). 32. Tang, H. et al. Unexpected omega-3 activities in intracellular lipolysis and macrophage foaming revealed by fluorescence lifetime imaging. Proc. Natl. Acad. Sci. 121, e2321255121 (2024). 33. Barber, E., Sinclair, A. J. & Cameron-Smith, D. Comparative actions of omega-3 fatty acids on in-vitro lipid droplet formation. Prostaglandins, Leukot. Essent. Fat. Acids (PLEFA) 89, 359366 (2013). 34. Graham, A. Mitochondrial regulation of macrophage cholesterol homeostasis. Free Radic. Biol. Med.89, 982–992 (2015).35. Robert, S. & Tommy, N. Rab proteins implicated in lipid storage and mobilization. J. Biomed. Res. 28, 169 (2014). 36. Schroeder, B. et al. The small GTPase Rab7 as a central regulator of hepatocellular lipophagy. Hepatology 61, 1896–1907 (2015). 37. Dworak, M., McCarley, R. W., Kim, T., Kalinchuk, A. V. & Basheer, R. Sleep and Brain Energy Levels: ATP Changes during Sleep. J Neurosci 30, 9007–9016 (2010). 38. Park, J. S. et al. LDHB Deficiency Promotes Mitochondrial Dysfunction Mediated Oxidative Stress and Neurodegeneration in Adult Mouse Brain. Antioxidants 11, 261 (2022). 39. Campagno, K. E. & Mitchell, C. H. The P2X7 Receptor in Microglial Cells Modulates the Endolysosomal Axis, Autophagy, and Phagocytosis. Front. Cell. Neurosci.15, 645244 (2021). 40. Cao, L.-L., Guan, P.-P., Liang, Y.-Y., Huang, X.-S. & Wang, P. Calcium Ions Stimulate the Hyperphosphorylation of Tau by Activating Microsomal Prostaglandin E Synthase 1. Front Aging Neurosci 11, 108 (2019). 41. Pierrot, N. et al. Calcium-mediated Transient Phosphorylation of Tau and Amyloid Precursor Protein Followed by Intraneuronal Amyloid-β Accumulation*. J Biol Chem 281, 39907–39914 (2006). 42. Ghosh, A. & Giese, K. P. Calcium / calmodulin-dependent kinase II and Alzheimer’s disease. Mol Brain 8, 78 (2015). 43. Mitrofanova, A., Merscher, S. & Fornoni, A. Kidney lipid dysmetabolism and lipid droplet accumulation in chronic kidney disease. Nat. Rev. Nephrol. 19, 629–645 (2023). 44. Pacia, M. Z. et al. Rac1 regulates lipid droplets formation, nanomechanical, and nanostructural changes induced by TNF in vascular endothelium in the isolated murine aorta. Cell. Mol. Life Sci. 79, 317 (2022). 45. Jonker, J. W. et al. A PPARγ–FGF1 axis is required for adaptive adipose remodelling and metabolic homeostasis. Nature 485, 391–394 (2012). 46. Malek, M. et al. Inositol triphosphate-triggered calcium release blocks lipid exchange at endoplasmic reticulum-Golgi contact sites. Nat. Commun. 12, 2673 (2021). 47. Wongkittichote, P., Mew, N. A. & Chapman, K. A. Propionyl-CoA carboxylase – A review. Mol. Genet. Metab. 122, 145–152 (2017). 48. Ferdinandusse, S. & Houten, S. M. Peroxisomes and bile acid biosynthesis. Biochim.Biophys. Acta (BBA) - Mol. Cell Res. 1763, 1427–1440 (2006). 49. Duan, J. et al. Alzheimer’s disease risk allele of PICALM causes detrimental lipid droplets in microglia. Res. Sq. rs.3.rs-4407146 (2024) doi:10.21203 / rs.3.rs-4407146 / v1. 50. Ramirez, V. P., Gurevich, I. & Aneskievich, B. J. Emerging roles for TNIP1 in regulating post-receptor signaling. Cytokine Growth Factor Rev. 23, 109–118 (2012). 51. Pfeiffer, L. et al. DNA Methylation of Lipid-Related Genes Affects Blood Lipid Levels. Circ.: Cardiovasc. Genet. 8, 334–342 (2018). 52. Yeoh, W. J. & Krebs, P. SHIP1 and its role for innate immune regulation—Novel targets for immunotherapy. Eur. J. Immunol. 53, e2350446 (2023). 53. Smolič, T., Zorec, R. & Vardjan, N. Pathophysiology of Lipid Droplets in Neuroglia. Antioxidants 11, 22 (2021). 54. Choi, S. M. et al. Insulin Regulates Adipocyte Lipolysis via an Akt-Independent Signaling Pathway. Mol. Cell. Biol. 30, 5009–5020 (2010). 55. Biswas, L., Farhan, F., Reilly, J., Bartholomew, C. & Shu, X. TSPO Ligands Promote Cholesterol Efflux and Suppress Oxidative Stress and Inflammation in Choroidal Endothelial Cells. Int. J. Mol. Sci. 19, 3740 (2018). 56. Bréhat, J. et al. Identification of a mechanism promoting mitochondrial sterol accumulation during myocardial ischemia–reperfusion: role of TSPO and STAR. Basic Res. Cardiol. 119, 481–503 (2024). 57. Hong, C. & Tontonoz, P. Liver X receptors in lipid metabolism: opportunities for drug discovery. Nat. Rev. Drug Discov. 13, 433–444 (2014). 58. Tchoua, U., Gillard, B. K. & Pownall, H. J. HDL superphospholipidation enhances key steps in reverse cholesterol transport. Atherosclerosis 209, 430–435 (2010). 59. Cole, L. K., Vance, J. E. & Vance, D. E. Phosphatidylcholine biosynthesis and lipoprotein metabolism. Biochimica Et Biophysica Acta Bba - Mol Cell Biology Lipids 1821, 754–761 (2012). 60. Chamberlain, K. A. & Sheng, Z. Mechanisms for the maintenance and regulation of axonal energy supply. J. Neurosci. Res. 97, 897–913 (2019). 61. Hertz, L., Gibbs, M. E. & Dienel, G. A. Fluxes of lactate into, from, and among gap junctioncoupled astrocytes and their interaction with noradrenaline. Front Neurosci-switz 8,261 (2014). 62. Lange, Y. & Steck, T. L. Active membrane cholesterol as a physiological effector. Chem Phys Lipids 199, 74–93 (2016). 63. Guixà-González, R. et al. Membrane cholesterol access into a G-protein-coupled receptor. Nat Commun 8, 14505 (2017). 64. Prasanna, X., Mohole, M., Chattopadhyay, A. & Sengupta, D. Role of cholesterol-mediated effects in GPCR heterodimers. Chem Phys Lipids 227, 104852 (2020). 65. Dumitrescu, L. et al. Genetic variants and functional pathways associated with resilience to Alzheimer’s disease. Brain 143, 2561–2575 (2020). 66. Montagne, A. et al. APOE4 leads to blood–brain barrier dysfunction predicting cognitive decline. Nature 581, 71–76 (2020). 67. Hooijmans, C. R., Jong, P. C. M. P., Vries, R. B. M. de & Ritskes-Hoitinga, M. The Effects of Long-Term Omega-3 Fatty Acid Supplementation on Cognition and Alzheimer’s Pathology in Animal Models of Alzheimer’s Disease: A Systematic Review and Meta-Analysis. J. Alzheimer’s Dis. 28, 191–209 (2012). 68. Tremblay, M.-E. et al. Remodeling of lipid bodies by docosahexaenoic acid in activated microglial cells. J. Neuroinflammation 13, 116 (2016). 69. Riddell, D. R. et al. Impact of Apolipoprotein E (ApoE) Polymorphism on Brain ApoE Levels. J Neurosci 28, 11445–11453 (2008). 70. Hong, B. V. et al. High-Density Lipoprotein Changes in Alzheimer’s Disease Are APOE Genotype-Specific. Biomed 10, 1495 (2022). 71. Ruiz, J. et al. The apoE isoform binding properties of the VLDL receptor reveal marked differences from LRP and the LDL receptor. J Lipid Res 46, 1721–1731 (2005). 72. Arboleda-Velasquez, J. F. et al. Resistance to autosomal dominant Alzheimer’s disease in an APOE3 Christchurch homozygote: a case report. Nat Med 25, 1680–1683 (2019). 73. Ferris, H. A. et al. Loss of astrocyte cholesterol synthesis disrupts neuronal function and alters whole-body metabolism. Proc National Acad Sci 114, 1189–1194 (2017). 74. Madison, B. B. Srebp2: A master regulator of sterol and fatty acid synthesis 1. J. Lipid Res. 57, 333–335 (2016).75. Natunen, T. et al. Effects of NR1H3 Genetic Variation on the Expression of Liver X Receptor α and the Progression of Alzheimer’s Disease. PLoS ONE 8, e80700 (2013). 76. Schacky, C. von. Importance of EPA and DHA Blood Levels in Brain Structure and Function. Nutrients 13, 1074 (2021). 77. Coley, N. et al. Defining the Optimal Target Population for Trials of Polyunsaturated Fatty Acid Supplementation Using the Erythrocyte Omega-3 Index: A Step Towards Personalized Prevention of Cognitive Decline? J. Nutr., Heal. aging 22, 982–988 (2018). 78. Muraleedharan, R. et al. AMPK-Regulated Astrocytic Lactate Shuttle Plays a Non- CellAutonomous Role in Neuronal Survival. Cell Rep. 32, 108092 (2020). 79. Larramona-Arcas, R. et al. Sex-dependent calcium hyperactivity due to lysosomal-related dysfunction in astrocytes from APOE4 versus APOE3 gene targeted replacement mice. Mol. Neurodegener. 15, 35 (2020). 80. Borroto-Escuela, D. O. et al. Understanding the Role of Adenosine A2AR Heteroreceptor Complexes in Neurodegeneration and Neuroinflammation. Front Neurosci-switz 12, 43 (2018). 81. Sims, J.R., Zimmer, J.A., Evans, C.D., Lu, M., Ardayfio, P., Sparks, J., Wessels, A.M., Shcherbinin, S., Wang, H., Nery, E.S.M., et al. (2023). Donanemab in Early Symptomatic Alzheimer Disease. JAMA 330, 512–527. https: / / doi.org / 10.1001 / jama.2023.13239. 82. Foley, P. (2010). Lipids in Alzheimer’s disease: A century-old story. Biochim. Biophys. Acta (BBA) - Mol. Cell Biol. Lipids 1801, 750–753. 83. Ralhan, I., Chang, C.-L., Lippincott-Schwartz, J., and Ioannou, M.S. (2021). Lipid droplets in the nervous system. J. Cell Biol. 220, e202102136. 84. Abildayeva, K., Jansen, P.J., Hirsch-Reinshagen, V., Bloks, V.W., Bakker, A.H.F.,Ramaekers, F.C.S., Vente, J. de, Groen, A.K., Wellington, C.L., Kuipers, F., et al. (2006). 24(S)-Hydroxycholesterol Participates in a Liver X Receptor-controlled Pathway in Astrocytes That Regulates Apolipoprotein E-mediated Cholesterol Efflux. J Biol Chem 281, 12799–12808. https: / / doi.org / 10.1074 / jbc.m601019200.85. Matsuda, A., Nagao, K., Matsuo, M., Kioka, N., and Ueda, K. (2013). 24(S)hydroxycholesterol is actively eliminated from neuronal cells by ABCA1. J Neurochem 126, 93–101. 86. Smolič, T., Zorec, R., and Vardjan, N. (2021). Pathophysiology of Lipid Droplets inNeuroglia. Antioxidants 11, 22. https: / / doi.org / 10.3390 / antiox11010022. 87. Gamba, P., Staurenghi, E., Testa, G., Giannelli, S., Sottero, B., and Leonarduzzi, G. (2019). A Crosstalk Between Brain Cholesterol Oxidation and Glucose Metabolism in Alzheimer’s Disease. Front Neurosci-switz 13, 556. https: / / doi.org / 10.3389 / fnins.2019.00556. 88. Leoni, V., and Caccia, C. (2011). Oxysterols as biomarkers in neurodegenerative diseases. Chem Phys Lipids 164, 515–524. https: / / doi.org / 10.1016 / j.chemphyslip.2011.04.002. 89. Prasanna, X., Sengupta, D., and Chattopadhyay, A. (2016). Cholesterol-dependent Conformational Plasticity in GPCR Dimers. Sci Rep-uk 6, 31858. https: / / doi.org / 10.1038 / srep31858. 90. Guixà-González, R., Albasanz, J.L., Rodriguez-Espigares, I., Pastor, M., Sanz, F., Martí- Solano, M., Manna, M., Martinez-Seara, H., Hildebrand, P.W., Martín, M., et al. (2017). Membrane cholesterol access into a G-protein-coupled receptor. Nat Commun 8, 14505. https: / / doi.org / 10.1038 / ncomms14505. 91. McGraw, C., Yang, L., Levental, I., Lyman, E., and Skaja-Robinson, A. (2019). Membrane cholesterol depletion reduces downstream signaling activity of the adenosine A2a receptor. Biochimica Et Biophysica Acta Bba - Biomembr 1861, 760–767. 92. Horvat, A., Zorec, R., and Vardjan, N. (2021). Lactate as an Astroglial Signal Augmenting Aerobic Glycolysis and Lipid Metabolism. Front. Physiol. 12, 735532. https: / / doi.org / 10.3389 / fphys.2021.735532.93. Morland, C., Lauritzen, K.H., Puchades, M., Holm Hansen, S., Andersson, K., Gjedde,A., Attramadal, H., Storm Mathisen, J., and Bergersen, L.H. (2015). The lactate receptor,G protein coupled receptor 81 / hydroxycarboxylic acid receptor 1: Expression and action inbrain. J. Neurosci. Res. 93, 1045–1055. https: / / doi.org / 10.1002 / jnr.23593. 94. Hertz, L., Gibbs, M.E., and Dienel, G.A. (2014). Fluxes of lactate into, from, and among gap junction-coupled astrocytes and their interaction with noradrenaline. Front Neurosciswitz 8, 261. 95. Lange, Y., and Steck, T.L. (2016). Active membrane cholesterol as a physiological effector. Chem Phys Lipids 199, 74–93. https: / / doi.org / 10.1016 / j.chemphyslip.2016.02.003. 96. Javanainen, M., Enkavi, G., Guixà-Gonzaléz, R., Kulig, W., Martinez-Seara, H., Levental, I., and Vattulainen, I. (2019). Reduced level of docosahexaenoic acid shifts GPCRneuroreceptors to less ordered membrane regions. Plos Comput Biol 15, e1007033. https: / / doi.org / 10.1371 / journal.pcbi.1007033. 97. Haney, M.S., Pálovics, R., Munson, C.N., Long, C., Johansson, P.K., Yip, O., Dong, W., Rawat, E., West, E., Schlachetzki, J.C.M., et al. (2024). APOE4 / 4 is linked to damaging lipid droplets in Alzheimer’s disease microglia. Nature, 1–8. https: / / doi.org / 10.1038 / s41586-024- 07185-7. 98. Victor, M.B., Leary, N., Luna, X., Meharena, H.S., Scannail, A.N., Bozzelli, P.L., Samaan, G., Murdock, M.H., Maydell, D. von, Effenberger, A.H., et al. (2022). Lipid accumulation induced by APOE4 impairs microglial surveillance of neuronal-network activity. Cell Stem Cell 29, 1197-1212.e8. 99. Nugent, A.A., Lin, K., Lengerich, B. van, Lianoglou, S., Przybyla, L., Davis, S.S., Llapashtica, C., Wang, J., Kim, D.J., Xia, D., et al. (2020). TREM2 Regulates Microglial Cholesterol Metabolism upon Chronic Phagocytic Challenge. Neuron 105, 837-854.e9. 100. Arendt, T., Brückner, M.K., Morawski, M., Jäger, C., and Gertz, H.-J. (2015). Early neuron loss in Alzheimer’s disease: cortical or subcortical? Acta Neuropathol. Commun.3, 10. https: / / doi.org / 10.1186 / s40478-015-0187-1. 101. Kerdiles, O., Layé, S., and Calon, F. (2017). Omega-3 polyunsaturated fatty acids and brain health: Preclinical evidence for the prevention of neurodegenerative diseases. Trends Food Sci. Technol. 69, 203–213. https: / / doi.org / 10.1016 / j.tifs.2017.09.003. 102. Burckhardt, M., Herke, M., Wustmann, T., Watzke, S., Langer, G., and Fink, A. (2016).Omega 3 fatty acids for the treatment of dementia. Cochrane Database Syst. Rev. 2016,CD009002. https: / / doi.org / 10.1002 / 14651858.cd009002.pub3.102. 103. Schacky, C. von (2020). Omega-3 index in 2018 / 19. Proc. Nutr. Soc. 79, 381–387. 104. Schacky, C. von (2014). Omega-3 Index and Cardiovascular Health. Nutrients 6, 799– 814. 105. Tang, H., Liu, Z., Han, G., Geng, J., Liu, B., Zhang, R., and Zhang, Z. (2024). Unexpected omega-3 activities in intracellular lipolysis and macrophage foaming revealed by fluorescence lifetime imaging. Proc. Natl. Acad. Sci. 121, e2321255121. https: / / doi.org / 10.1073 / pnas.2321255121. 106. Barber, E., Sinclair, A.J., and Cameron-Smith, D. (2013). Comparative actions of omega-3 fatty acids on in-vitro lipid droplet formation. Prostaglandins, Leukot. Essent. Fat. Acids (PLEFA) 89, 359–366. 107. Ferdinandusse, S., and Houten, S.M. (2006). Peroxisomes and bile acid biosynthesis. Biochim. Biophys. Acta (BBA) - Mol. Cell Res.1763, 1427–1440. 108. Schacky, C. von (2020). Omega-3 index in 2018 / 19. Proc. Nutr. Soc. 79, 381–387. https: / / doi.org / 10.1017 / s0029665120006989. 109. Schuchardt, J.P., Tintle, N., Westra, J., and Harris, W.S. (2023). Estimation and predictors of the Omega-3 Index in the UK Biobank. Br. J. Nutr. 130, 312–322. 110. Schott, J.M., Crutch, S.J., Carrasquillo, M.M., Uphill, J., Shakespeare, T.J., Ryan, N.S.,Yong, K.X., Lehmann, M., Ertekin Taner, N., Graff Radford, N.R., et al. (2016). Geneticrisk factors for the posterior cortical atrophy variant of Alzheimer’s disease. Alzheimer’s Dementia 12, 862–871. https: / / doi.org / 10.1016 / j.jalz.2016.01.010. 111. Yeoh, W.J., and Krebs, P. (2023). SHIP1 and its role for innate immune regulation— Novel targets for immunotherapy. Eur. J. Immunol. 53, e2350446. 112. Suthar, S.K., Alam, M.M., Lee, J., Monga, J., Joseph, A., and Lee, S.-Y. (2021). Bioinformatic Analyses of Canonical Pathways of TSPOAP1 and its Roles in Human Diseases. Front. Mol. Biosci. 8, 667947. 113. Tchoua, U., Gillard, B.K., and Pownall, H.J. (2010). HDL superphospholipidation enhances key steps in reverse cholesterol transport. Atherosclerosis 209, 430–435. https: / / doi.org / 10.1016 / j.atherosclerosis.2009.10.002. 114. Cole, L.K., Vance, J.E., and Vance, D.E. (2012). Phosphatidylcholine biosynthesis and lipoprotein metabolism. Biochimica Et Biophysica Acta Bba - Mol Cell Biology Lipids 1821, 754–761. 115. Hertz, L., Gibbs, M.E., and Dienel, G.A. (2014). Fluxes of lactate into, from, and among gap junction-coupled astrocytes and their interaction with noradrenaline. Front Neurosciswitz 8, 261. https: / / doi.org / 10.3389 / fnins.2014.00261. 116. Chamberlain, K.A., and Sheng, Z. (2019). Mechanisms for the maintenance and regulation of axonal energy supply. J. Neurosci. Res. 97, 897–913. https: / / doi.org / 10.1002 / jnr.24411. ***********************************************

[0186] The various embodiments described herein can be combined to provide further embodiments.

[0187] Aspects of the embodiments can be modified, if necessary to employ concepts of the various patents, application, and publications to provide yet further embodiments.

[0188] These and other changes can be made to the embodiments in light of the above-detailed description. In general, in the following claims, the terms used should not be construed to limit the claims to the specific embodiments disclosed in the specification and the claims, but should be construed to include all possible embodiments along with the full scope of equivalents to which such claims are entitled. Accordingly, the claims are not limited by the disclosure.

[0189] All of the U.S. patents, U.S. patent application publications, U.S. patent applications, foreign patents, foreign patent applications and non-patent publications referred to in this specification are incorporated herein by reference, in their entireties. Paterson, Tom et al., Multiomic Evidence for a Unified Model of Alzheimer’s Disease Etiology Linking Microglial Flux Capacity and Astrocyte-Neuron Metabolic Breakdown, bioRxiv preprint doi: https: / / doi.org / 10.1101 / 2024.07.23.604835, posted May 12, 2025 is incorporated by reference in its entirety.

Claims

1. CLAIMS 1. A method for determining whether a subject is at risk of developing mild cognitive impairment (MCI) or Alzheimer’s Disease (AD), for determining whether the subject has MCI or AD, or for characterizing the cellular cholesterol dyshomeostasis status of the subject, the method comprising determining whether one or more biological samples obtained from the subject comprises: (a) one or more genetic mutations indicative of a predisposition to ANLS attenuation and / or microglial flux capacity saturation; (b) an increased or decreased amount or concentration of one or more cerebral spinal fluid (CSF) proteins indicative of a predisposition to astrocyte-neuron lactate shuttle (ANLS) attenuation; (c) an increased or decreased amount or concentration of one or more plasma lipids indicative of a predisposition to ANLS attenuation; (d) an increased or decreased amount of one or more CSF proteins indicative of ANLS attenuation; (e) an increased or decreased amount or concentration of one or more plasma lipids indicative of ANLS attenuation; (f) an increased or decreased amount or concentration of one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation; (g) an increased or decreased amount or concentration of one or more CSF proteins indicative of microglial flux capacity saturation; and / or (h) an increased or decreased amount or concentration of one or more plasma lipids indicative of microglial flux capacity saturation, wherein: (i) the increased or decreased amount or concentration of the one or more CSF proteins of (b), (d) and / or (g); (ii) the increased or decreased amount or concentration of the one or more plasma lipids of (c), (e), (f) and / or (h); and / or (iii) the presence of the one or more genetic mutations of (a) is indicative of whether the subject is at risk of developing or has mild cognitive impairment (MCI) or Alzheimer’s Disease (AD) and / or indicates the cholesterol dyshomeostasis status of the subject, wherein the cholesterol dyshomeostasis status is selected from ANLS attenuation and / or microglial flux capacity saturation.

2. The method of claim 1, wherein determining the increased or decreased amount of the one or more CSF proteins and / or plasma lipid markers comprises: (a) determining an amount or concentration of the one or more of the CSF proteins and / or the one or more plasma lipids; and (b) comparing the amount or concentration of the one or more CSF proteins and / or the one or more plasma lipids determined in (a) to a reference value(s) or to an amount or concentration of the one or more CSF proteins and / or the one or more plasma lipids determined in a control sample(s), wherein an increased or decreased amount or concentration of at least one of the one or more CSF proteins and / or at least one of the one or more plasma lipids in the biological sample as compared to the reference value or the amount or concentration determined in the control sample indicates the subject is at risk of developing or has MCI or AD and / or indicates the cholesterol dyshomeostasis status of the subject.

3. The method of claim 1 or claim 2, wherein: (a) the one or more genetic mutations indicative of a predisposition to ANLS attenuation are selected from mutations in INPP5D, PICALM, TSPOAP, TNIP1, and / or LXRA genes; (b) the one or more CSF proteins indicative of a predisposition to ANLS attenuation are selected from PLA2G7 and LPCAT2; (c) the one or more plasma lipids indicative of a predisposition to ANLS attenuation are plasma lipids indicating high DHA in phospholipid or plasmalogen form; (d) the one or more CSF proteins indicative of ANLS attenuation are selected from the group consisting of CHGA, PPIH, ARL3, EIF1B, KIFBP, WFDC10A, AKR1D1, TCEAL8, CYB5R2, NDUFA5, RAB5B, ALDH1A2, IMPACT, NREP, DSG4, XAGE2, RAB7A, CRYL1, GYG2, INPP5A, TOM1, RAC1, FGF1, SLC27A2, RAB21, LDHB, LDHC, ALDOA, MDHC, KPYM, AATC, CNTN2, and TNR; (e) the one or more plasma lipids indicative of ANLS attenuation are selected from the group consisting of CER.D19.1.18.0 and CER.D19.1.16.0; (f) the one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation are selected from the group consisting of DHA, omega-3 fatty acids, optionally in the form of phospholipids or plasmalogens;(g) the one or more CSF proteins indicative of microglial flux capacity saturation are selected from the group consisting of TAU, CD44, CH3L1, GFAP, YWHAB, YWHAZ, and GAS6; and / or (h) the one or more plasma lipids indicative of microglial flux capacity saturation are selected from the group consisting of DHA / EPA acylcarnities, optionally AC.22.6 or AC.20.5, AC.13.0, AC.15.0, cholesterol esters, optionally CE.22.6, CE.20.5, CE.16.0, or CE.17.0, extra- large HDL lipoprotein cholesterol and / or cholesterol ester, optionally Nightingale assay XL.HDL.C or XL.HDL.CE.

4. The method of claim 3, wherein: (a) an increase or decrease in PLA2G7 and / or LPCAT2 is indicative of a predisposition to ANLS attenuation; (b) an increase or decrease in one or more plasma lipid indicating high DHA in phospholipid or plasmalogen form is indicative of a predisposition to ANLS attenuation; (c) an increase or decrease in one or more of CHGA, PPIH, ARL3, EIF1B, KIFBP, WFDC10A, AKR1D1, TCEAL8, CYB5R2, NDUFA5, RAB5B, ALDH1A2, IMPACT, NREP, DSG4, XAGE2, RAB7A, CRYL1, GYG2, INPP5A, TOM1, RAC1, FGF1, SLC27A2, RAB21, LDHB, LDHC, ALDOA, MDHC, KPYM, AATC, CNTN2, and TNR is indicative of ANLS attenuation; (d) an increase or decrease in CER.D19.1.18.0 and / or CER.D19.1.16.0 is indicative of ANLS attenuation; (e) an increase or decrease in one or more omega-3 fatty acid, optionally DHA, is indicative of a predisposition to microglial flux capacity saturation; (f) an increase or decrease in one or more of TAU, CD44, CH3L1, GFAP, YWHAB, YWHAZ, and GAS6 is indicative of microglial flux capacity saturation; and / or (g) an increase or decrease in one or more of DHA / EPA acylcarnities, optionally AC.22.6 or AC.20.5, AC.13.0, AC.15.0, cholesterol esters, optionally CE.22.6, CE.20.5, CE.16.0, or CE.17.0, and extra-large HDL lipoprotein cholesterol and / or cholesterol ester, optionally Nightingale assay XL.HDL.C or XL.HDL.CE is indicative of microglial flux capacity saturation.

5. The method of claim 3, wherein: (a) an increase or decrease in PLA2G7 and / or LPCAT2 and an increase or decrease in plasma lipids indicating high DHA in phospholipid or plasmalogen form is indicative of a predisposition to ANLS attenuation; (b) an increase or decrease in RAB21, RAB7A, LDHB, LDHC, ALDOA, MDHC, KPYM, AATC, CNTN2, TNR, TAU, CD44, CH3L1, GFAP, and / or GAS6 and an increase or decrease in CER.D19.1.18.0 and / or CER.D19.1.16.0 is indicative of ANLS attenuation; (c) an increase or decrease in one or more omega-3 fatty acid, optionally DHA is indicative of a predisposition to microglial flux capacity saturation; and / or (d) an increase or decrease in TAU, CD44, CH3L1, GFAP, and / or GAS6 and an increase or decrease in DHA / EPA acylcarnitines (optionally AC.22.

6. or AC.20.5.), cholesterol esters (optionally, CE.22.

6. or CE.20.5.), and / or extra-large HDL lipoprotein cholesterol and cholesterol ester (optionally XL.HDL.C or XL.HDL.CE) is indicative of microglial flux capacity saturation.

6. The method of any one of claims 1-5, wherein the one or more genetic mutation is identified by whole-genome sequencing or methylation array analysis.

7. The method of any one of claims 1-5, wherein the amount of the one or more CSF proteins and / or the one or more plasma lipids is determined by targeted mass spectrometry or aptamer-based proteomics.

8. The method of any one of claims 1-5, wherein the method further comprises determining cognitive scores and / or longitudinal biomarker trajectories of the subject.

9. The method of any one of claims 1-8, wherein the method comprises using an algorithm trained to determine whether the subject is at risk of developing or has mild cognitive impairment (MCI) or Alzheimer’s Disease (AD) and / or to determine the cholesterol dyshomeostasis status of the subject.

10. The method of any one of claims 1-9, wherein if the subject is determined to be at risk of developing MCI or AD or determined to have MCI or AD, treating the subject with a therapeutic agent for the treatment of MCI or AD.

11. A method of treating MCI or Alzheimer’s disease in a subject comprising: (a) determining whether the subject has astrocyte-neuron lactate shuttle (ANLS) attenuation ormicroglial flux capacity saturation; and (b) administering to the subject a therapeutic treatment, wherein: (i) if the subject has a predisposition for ANLS attenuation or ANLS attenuation, the subject is treated with increased sleep, phosphatidylcholine supplementation, and / or creatine supplementation; and (ii) if the subject has a predisposition for microglial flux capacity saturation or microglial flux capacity saturation, the subject is treated with Omega3 supplementation, plasmalogen supplementation, and / or plasmalogen precursor supplementation.

12. The method of claim 10, further comprising serial measurements of an amount or concentration of one or more of CER.D19.1.18.0, CD44, and XL.HDL.C to monitor treatment efficacy, wherein a change in the amount or concentration is indicative of treatment efficacy.

13. A method for determining whether a subject is at risk of developing mild cognitive impairment (MCI) or Alzheimer’s Disease (AD), for determining whether the subject has MCI or AD, or for characterizing the cellular cholesterol dyshomeostasis status of the subject, the method comprising: (a) determining an amount or concentration of: (i) one or more cerebral spinal fluid (CSF) proteins indicative of a predisposition to astrocyte-neuron lactate shuttle (ANLS) attenuation in a CSF sample obtained from the subject; (ii) determining an amount or concentration of one or more plasma lipids indicative of a predisposition to ANLS attenuation in a blood or plasma sample obtained from the subject; (iii) determining an amount or concentration of one or more CSF proteins indicative of ANLS attenuation in a CSF sample obtained from the subject; (iv) determining an amount or concentration of plasma lipids indicative of ANLS attenuation in a blood or plasma sample obtained from the subject; (v) determining an amount or concentration of one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation in a blood or plasma sample obtained from the subject; (vi) determining an amount or concentration of one or more CSF proteins indicative of microglial flux capacity saturation in a CSF sample obtained from the subject; and / or (vii) determining an amount or concentration of one or more plasma lipids indicativeof microglial flux capacity saturation in a blood or plasma sample obtained from the subject; and (b) determining whether the subject is at risk of developing mild cognitive impairment (MCI) or Alzheimer’s Disease (AD) or has MCI or AD, or characterizing the cellular cholesterol dyshomeostasis status of the subject, by correlating the amounts determined in (a) with the risk of developing MCI or AD, the presence or absence of MCI or AD, and / or the cellular cholesterol dyshomeostasis status of the subject, wherein the correlating is performed using a classifier generated using amount or concentration data determined for the CSF proteins and / or plasma lipids determined in (a) from a plurality of CSF and / or plasma or blood samples obtained from healthy subjects and a plurality of CSF and / or plasma or blood samples obtained from subjects with MCI and / or AD.

14. The method of claim 13, wherein the method determines in the subject has MCI or AD.

15. The method of claim 13, wherein the method determines whether the subject has AD associated with ANLS attenuation or microglial flux capacity saturation.

16. The method of any one of claims 13-15, wherein the method comprises determining an amount or concentration of: (a) one or more CSF proteins selected from the group consisting of: CSF proteins set forth in Table 1 or Table 2, PLA2G7, LPCAT2, RAB21, RAB7A, LDHB, LDHC, ALDOA, MDHC, KPYM, AATC, CNTN2, TNR, TAU, CD44, CH3L1, GFAP, and GAS6; and / or (b) one or more plasma lipids selected from the group consisting of: plasma lipids set forth in Table 1 or Table 2, plasma lipids indicating high DHA in phospholipid or plasmalogen form, CER.D19.1.18.0, CER.D19.1.16.0, DHA, omega-3 fatty acids, optionally in the form of phospholipids or plasmalogens, DHA / EPA acylcarnities, optionally AC.22.6 or AC.20.5, cholesterol esters, optionally CE.22.6 or CE.20.5, and extra-large HDL lipoprotein cholesterol and / or cholesterol ester, optionally Nightingale assay XL.HDL.C or XL.HDL.CE.

17. The method of claim 16, comprising determining the amount or concentration of at least 1, at least 2, at least 3, at least 5, or at least 10 of the CSF proteins and determining the amount or concentration of at least 2, at least 3, at least 5, or at least 10 of the serum lipids.

18. A kit comprising:(a) one or more reagents for detecting one or more genetic mutations indicative of a predisposition to ANLS attenuation and / or microglial flux capacity saturation; (b) one or more reagents for detecting an amount or concentration of one or more cerebral spinal fluid (CSF) proteins indicative of a predisposition to astrocyte-neuron lactate shuttle (ANLS) attenuation; (c) one or more reagents for detecting an amount or concentration of one or more plasma lipids indicative of a predisposition to ANLS attenuation; (d) one or more reagents for detecting an amount or concentration of one or more CSF proteins indicative of ANLS attenuation; (e) one or more reagents for detecting an amount or concentration of one or more plasma lipids indicative of ANLS attenuation; (f) one or more reagents for detecting an amount or concentration of one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation; (g) one or more reagents for detecting an amount or concentration of one or more CSF proteins indicative of microglial flux capacity saturation; and / or (h) one or more reagents for detecting an amount or concentration of one or more plasma lipids indicative of microglial flux capacity saturation.

19. The kit of claim 18, further comprising one or more reagents for determining an amount or concentration of one or more control proteins or lipids.

20. The kit of claim 18 or claim 19, further comprising instructions for interpreting the biomarker patterns determined when using the kit, optionally to classify a subject being tested as being at risk for MCI or AD, having MCI or AD, and / or for determining whether the MCI or AD is associated with ANLS attenuation or microglial flux capacity saturation.

21. The kit of any one of claims 18-20, further comprising a software module that can be used to classify the subject being tested by receiving biomarker measurements and providing automated classification.

22. The kit of claim 21, wherein the software module comprises an algorithm or classifier, trained on biomarker data obtained from healthy subjects and / or MCI or AD patients.

23. The kit of any one of claims 18-22, wherein: (a) the one or more genetic mutations indicative of a predisposition to ANLS attenuation areselected from mutations in INPP5D, PICALM, TSPOAP, TNIP1, and / or LXRA genes; (b) the one or more CSF proteins indicative of a predisposition to ANLS attenuation are selected from CSF proteins set forth in Table 1, PLA2G7 and LPCAT2; (c) the one or more plasma lipids indicative of a predisposition to ANLS attenuation are selected from plasma lipids set forth in Table 1 and plasma lipids indicating high DHA in phospholipid or plasmalogen form; (d) the one or more CSF proteins indicative of ANLS attenuation are selected from CSF proteins set forth in Table 1 and the group consisting of RAB21, RAB7A, LDHB, LDHC, ALDOA, MDHC, KPYM, AATC, CNTN2, and TNR; (e) the one or more plasma lipids indicative of ANLS attenuation are selected from plasma lipids set forth in Table 1 and the group consisting of CER.D19.1.18.0 and CER.D19.1.16.0; (f) the one or more plasma lipids indicative of a predisposition to microglial flux capacity saturation are selected from plasma lipids set forth in Table 2 and the group consisting of DHA, omega-3 fatty acids, optionally in the form of phospholipids or plasmalogens; (g) the one or more CSF proteins indicative of microglial flux capacity saturation are selected from CSF proteins set forth in Table 2 and the group consisting of TAU, CD44, CH3L1, GFAP, and GAS6; and / or (h) the one or more plasma lipids indicative of microglial flux capacity saturation are selected from plasma lipids set forth in Table 2 and the group consisting of DHA / EPA acylcarnities, optionally AC.22.6 or AC.20.5, and cholesterol esters, optionally CE.22.6 or CE.20.5, extra- large HDL lipoprotein cholesterol and / or cholesterol ester, optionally Nightingale assay XL.HDL.C or XL.HDL.CE.

Images