Histone acetylation blood biomarker for alzheimer's disease

Abstract

Methods and materials for identifying mammals (e.g., humans) as having, or being likely to have or to develop, Alzheimer's disease are provided herein. In some cases, methods and materials for stratifying mammals with Alzheimer's disease are provided herein.

Description

[0001] Attorney Docket No. 26100-0166W01

[0002] 2025-046

[0003] HISTONE ACETYLATION BLOOD BIOMARKER FOR ALZHEIMER’S DISEASE

[0004] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority from U.S. Provisional Application Serial No. 63 / 748,894, filed January 23, 2025. The disclosure of the prior application is considered part of (and is incorporated by reference in) the disclosure of this application.

[0005] TECHNICAL FIELD

[0006] This document relates to methods and materials for identifying mammals (e.g., humans) as having, or being likely to have or to develop, Alzheimer’s disease. In some cases, this document provides methods and materials for stratifying mammals with Alzheimer’s disease.

[0007] BACKGROUND

[0008] Alzheimer's Disease (AD) impacts over 55 million people worldwide. AD is a progressive neurodegenerative disease marked by memory loss and gradually declining cognitive function. AD is characterized in the brain by amyloid-P (AP) plaques and the hyperphosphorylation of tau protein, which forms neurofibrillary tangles. The preclinical phase of AD, between aggregation of these characteristic proteins and cognitive decline, can last as long as 5-10 years, as indicated by imaging technology (Jack et al., Lancet Neurol. 9: 119-128, 2010), which severely limits the utility of capturing the progression of AD through mental cognitive tests and neurological exams (Medical Tests, available at alz.org / alzheimers-dementia / diagnosis / medical_tests). The current gold standards for AD diagnosis are positron emission tomography (PET) imaging of amyloid and tau, or cerebrospinal fluid (CSF) testing of Ap and phosphorylated tau (Shaw et al., Ann. Neurol.

[0009] 65:403-413, 2009). These methods are invasive, expensive, and suboptimal for detecting early-stage AD, however.

[0010] AD remains one of the most difficult neurodegenerative disorders to diagnose early. Scalable and minimally invasive assessment of AD biomarkers, including plasma biomarkers, are needed to enable early diagnosis. Existing biomarkers, including amyloid-P, tau, and neurofilament light chain (NfL), largely reflect downstream neurodegeneration rather than the molecular changes that initiate disease progression.Attorney Docket No. 26100-0166W01

[0011] 2025-046

[0012] However, work on AD plasma biomarkers has focused on assaying plasma Ap and tau, which is predictive of cerebral Ap and tau deposition (Risacher et al., Alzheimers Dement.

[0013] 1:510-519, 2019; Palmqvist et al., JAMA Neural. 76:1060-1069, 2019; Lue et al., Neural Ther. 6:25-36, 2017; Karikari et al., Lancet Neurol. 19:422-433, 2020; Ashton et al., JAMA Neurol., 2024, doi: 10.1001 / jarnaneurol.2023.5319; Gonzalez-Ortiz et al., Mol. Neurodegener., 18:18, 2023; Song et al., BMC Med. 20:266, 2022; and Sood et al., Genome Biol. 16:185, 2015). Gene expression and epigenomic features offer a target of study to discover earlier-stage blood biomarkers for AD, potentially reflecting cell regulatory states that change detectably before the onset of AD pathology. Immune system differences can influence AD predisposition (Ramamurthy et al., bioRxiv 2020, doi:10.1101 / 2020.03.26. 010330; Gjoneska et al., Nature 518:365-369, 2015; Nott et al., Science 366:1134-1139, 2019; Huang et al., Nat. Neurosci. 20:1052-1061, 2017; and Novikova et al., Cold Spring Harbor Laboratory 694281, 2019, doi:10.1101 / 694281), especially for late onset Alzheimer’s disease (LOAD) (Heppner, et al., Nat. Rev.

[0014] Neurosci. 16:358-372, 2015). The inflammatory component of AD may be mediated by both brain cells (microglia, astrocytes) and immune cells in the peripheral blood (monocytes, neutrophils), and peripheral immune cell regulatory states may change during the preclinical phase of AD.

[0015] In peripheral blood, DNA methylation levels (Sun et al., Transl. Psychiatry 13:387, 2023; Fransquet et al., Alzheimers Dement. 14:81-103, 2018; Silva et al., Nat. Commun. 13:1-16, 2022; and Li et al., Clin. Epigenetics 13:191, 2021) and gene expression (Song et al., supra, Sood et al., supra, and Zhong et al. Alzheimers Dement. 20:2469-2484, 2024) have been evaluated as biomarkers. DNA methylation biomarkers have yielded high accuracy for AD prediction (95% CI AUC=0.93; gene expression: auROCs>0.9), but methylated DNA regions may be difficult to perturb with epigenomic drugs. Levels of gene expression in the blood have shown potential as a biomarker, but the predictions have not proven to be reliable enough for clinical testing (Lunnon et al., J. Alzheimers Dis. 33:737-753, 2013; and Lee and Lee, Sci. Rep. 10:3485, 2020).

[0016] Epigenomic features, which usually are more stable than gene expression (Johannes et al., Nat. Rev. Genet. 9:883-890, 2008), may provide the ability to cheaply and reliably predict AD in the preclinical phase, improving the efficacy of current AD treatments and informing the development (Gjoneska et al., supra, Nott et al., supra, Huang et al., supra, Novikova et al., supra, and Ramamurthy et al., Front. Mol. Neurosci. 15:948456, 2022)Attorney Docket No. 26100-0166W01

[0017] 2025-046

[0018] of new treatments (Dubois, et al., J. Alzheimers. Dis. 49:617-631, 2016; and Long and Holtzman, Cell 179:312-339, 2019).

[0019] An epigenetic modification associated with promoter and enhancer activity, histone 3 lysine 27 acetylation (H3K27ac) modification enrichment, has been linked to AD progression in mouse and human (Creyghton et al., Proc. Natl. Acad. Sci. USA 107:21931-21936, 2010). In bulk tissue measurements, AD pathology was strongly linked to differences in H3K27ac (Grubman et al., Nat. Neurosci. 22:2087-2097, 2019; and Nativio et al., Nat. Genet. 52:1024-1035, 2020). In a mouse model of neurodegeneration, H3K27ac was found to be up-regulated in immune enhancers (enhancers active in immune cell types) implicated by genome-wide association studies in the predisposition to LOAD (Gjoneska et al., Nature 518(7539):365-369, 2015). Additionally, H3K27ac is differentially regulated in microglia of the human brain through age (Ramamurthy et al., supra).

[0020] Both microglia in the brain and myeloid cells in the blood are mononuclear phagocytes, and they share a common involvement in neuropathological processes (Prinz and Priller, Nat. Rev. Neurosci. 15:300-312, 2014; Low and Ginhoux, Cell. Immunol. 330:68-78, 2018; and Werner et al., Nat. Neurosci. 23:351-362, 2020). Given these similarities, differences in H3K27ac in the blood may serve as druggable AD epigenomic blood biomarkers. Histone acetylation is easier to perturb than DNA methylation (Zhou et al., Clin. Epigenetics 14:70 (2022), and histone deacetylase (HDAC) inhibitors are commonly used in cancer treatment (Li and Seto, Cold Spring Harb. Perspect. Med. 6, 2016).

[0021] SUMMARY

[0022] This document is based, at least in part, on the identification of blood acetylation biomarker H3K27ac from the immune-enriched buffy coat fraction of blood donors as a marker for AD, on the identification of H3K27ac enrichment levels at particular genomic regions that are predictive for AD, and also on the development of a classifier that can be used to predict AD status from H3K27ac levels in blood epigenome-wide. As described herein, epigenetic dysregulation, especially aberrant histone acetylation, appears to be an early driver of AD pathology, altering gene programs that control synaptic function, neuroinflammation, and metabolic signaling. Histone acetylation serves as a key regulatory mechanism governing activity of enhancer regions of the genome that controlAttorney Docket No. 26100-0166W01

[0023] 2025-046

[0024] how and when nearby genes are expressed. Disruption of enhancer networks can set off cascades leading to neuronal dysfunction long before structural damage occurs. Despite its mechanistic importance, no clinically accessible assay has previously captured histone acetylation changes from blood, leaving a significant gap between fundamental neuroscience and practical diagnosis.

[0025] As demonstrated herein, drawing on the upregulation of H3K27ac in immune enhancers implicated in genetic predisposition to LOAD, H3K27ac was profiled in immune cells of the blood of AD patients to identify a subset of epigenomic regions in which H3K27ac enrichment was related to AD status. This work revealed that genomewide blood H3K27ac enrichment levels at 49874 genomic regions were predictive for AD on the level of other plasma biomarkers (auROC > 0.9), with a cross validation accuracy of 80%. In addition, leveraging AD genetics, H3K27ac enrichment levels at a much sparser set of 646 regions near AD risk loci captured a similar level of predictive power for AD status. Enrichment levels at these loci predicted AD status more reliably than enrichment levels at random loci or risk loci for other diseases. In other words, the classifier predicting AD status trained from blood H3K27ac levels at those 646 loci near AD genetic risk loci outperformed models trained on genetic risk loci for other diseases, suggesting that blood histone acetylation in AD occurs at genetic risk loci in a disease-associated manner. The work described herein therefore provides an opportunity for minimally-invasive early AD detection, and suggests a convergence between AD epigenetic and genetic biomarkers, even in the peripheral blood.

[0026] Having the ability to identify mammals (e.g., humans) as having early onset AD provides a unique and unrealized opportunity that allows those mammals to be properly identified and potentially treated to slow or delay progression of the disease. For example, treatment such as donanemab can be used to treat patients identified as having early onset AD using the methods and materials described herein.

[0027] In a first aspect, this document features a method for identifying a mammal as having, being likely to have, or being likely to develop AD, wherein the method includes, consists of, or consists essentially of measuring, in a blood sample from the mammal, levels of H3K27ac at a plurality of genomic regions; determining that the levels of H3K27ac are elevated as compared to reference levels of H3K27ac at the plurality of genomic regions; and identifying the mammal as having, being likely to have, or being likely to develop AD. The mammal can be a human. The plurality of genomic regions canAttorney Docket No. 26100-0166W01

[0028] 2025-046

[0029] be near AD risk loci. The plurality of genomic regions can include 500 or more genomic regions near the AD risk loci. The measuring can include chromatin immunoprecipitation followed by sequencing (ChlP-seq) of genomic DNA from the blood sample. The blood sample can be a buffy coat sample. The reference level of H3K27ac can be the level of H3K27ac in a blood sample from a mammal known not to have AD. The method can further include administering, to the mammal, a therapeutic agent for treating the AD. The therapeutic agent can include donanemab or lecanemab.

[0030] In another aspect, this document features the use of a therapeutic agent to treat a mammal identified as having, being likely to have, or being likely to develop, AD, wherein the mammal was identified as having, being likely to have, or being likely to develop AD using a method that includes, consists of, or consists essentially of measuring, in a blood sample from the mammal, levels of H3K27ac at a plurality of genomic regions; and determining that the levels of H3K27ac are elevated as compared to reference levels of H3K27ac at the plurality of genomic regions, thereby identifying the mammal as having, being likely to have, or being likely to develop AD. The mammal can be a human. The plurality of genomic regions can be near AD risk loci. The plurality of genomic regions can include 500 or more genomic regions near the AD risk loci. The therapeutic agent can include donanemab or lecanemab. The measuring can include ChlP-seq of genomic DNA from the blood sample. The blood sample can be a buffy coat sample. The reference level of H3K27ac can be the level of H3K27ac in a blood sample from a mammal known not to have AD.

[0031] In another aspect, this document features a method for using a therapeutic agent to treat a mammal identified as having, being likely to have, or being likely to develop, AD, wherein the mammal was identified as having, being likely to have, or being likely to develop AD using a method that includes, consists of, or consists essentially of measuring, in a blood sample from the mammal, levels of H3K27ac at a plurality of genomic regions; and determining that the levels of H3K27ac are elevated as compared to reference levels of H3K27ac at the plurality of genomic regions, thereby identifying the mammal as having, being likely to have, or being likely to develop AD. The mammal can be a human. The plurality of genomic regions can be near AD risk loci. The plurality of genomic regions can include 500 or more genomic regions near the AD risk loci. The therapeutic agent can include donanemab or lecanemab. The measuring can include ChlP-seq of genomic DNA from the blood sample. The blood sample can be a buffy coatAttorney Docket No. 26100-0166W01

[0032] 2025-046

[0033] sample. The reference level of H3K27ac can be the level of H3K27ac in a blood sample from a mammal known not to have AD.

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. Although methods and materials similar or equivalent to those described herein can be used to practice the invention, suitable methods and materials are described below. All publications, patent applications, patents, and other references mentioned herein are incorporated by reference in their entirety. In case of conflict, the present specification, including definitions, will control. In addition, the materials, methods, and examples are illustrative only and not intended to be limiting.

[0035] The details of one or more embodiments of the invention are set forth in the accompanying drawings and the description below. Other features, objects, and advantages of the invention will be apparent from the description and drawings, and from the claims.

[0036] DESCRIPTION OF DRAWINGS FIG. 1A is a schematic overview of the study population and sample collection for the work described herein. FIG. IB is a schematic overview of blood H3K27ac module analysis for the studies described herein. FIG. 1C is a heat map depicting activity of blood AD-up and AD-down H3K27ac modules across immune and brain tissues and cell types. FIG. ID includes a graph and a chart showing illustrative activity near APOE and surrounding genes, with SNPs annotated.

[0037] FIG. 2A includes a pair of graphs plotting cross validation Receiver Operating Characteristic (ROC) (top panel) and Precision-Recall (PRC) (bottom panel) curves for a genome-wide blood H3K27ac trained classifier. FIG. 2B is a flowchart illustrating a process for selecting disease-associated and random SNPs. FIG. 2C includes graphs plotting cross validation balanced accuracy (left panel), auROC (center panel), and auPRC (right panel) of a logistic regression model trained on active regulatory elements (AREs) residing in AD risk loci, Schizophrenia (SCZ) risk loci, Inflammatory Bowel Disease (IBD) risk loci, Type 1 diabetes risk loci (T1D), Type 2 diabetes risk loci (T2D), and 50 random sets of AREs selected from the full set of AREs, with error bars indicating SEM. In the classifiers, 646 genomic regions near risk loci for AD, SCZ, IBD, T1D, and T2D were used, along with 646 genomic regions near random loci. FIG. 2D includes aAttorney Docket No. 26100-0166W01

[0038] 2025-046

[0039] pair of graphs plotting cross validation ROC (top panel) and PRC (bottom panel) curves for the AD genetic risk informed classifier.

[0040] DETAILED DESCRIPTION

[0041] This document provides a histone acetylation- (H3K27ac-) based blood biomarker, and methods and materials for use of the biomarker to detect and / or predict AD by quantifying H3K27ac levels across enhancer regions in the mammalian (e.g., human) genome. The enhancer loci can be located near genes that modulate AD risk, including those that govern lipid metabolism, synaptic plasticity, and immune signaling. In AD, these processes are disrupted characteristically. For example, lipid metabolism shifts towards neurotoxic cholesterol transport and impaired membrane activity, synaptic plasticity declines as key neuronal genes become epigenetically silenced, and immune signaling becomes hyperactive, driving chronic neuroinflammation.

[0042] The identified enhancer regions described herein show distinct acetylation signatures tied to these pathways, providing a direct readout of their dysregulation in blood-derived nucleosomes. The methods and materials provided herein can be used to measure histone acetylation from circulating nucleosomes in plasma, effectively reading the epigenetic signature of enhancer activity without requiring neural tissue. Using ChlP-seq and complementary quantitative assays, specific acetylation patterns that reliably distinguish AD subjects from healthy controls were identified. Statistical modeling confirmed strong discriminative power between cohorts and correlated these enhancer profiles with established neuropathological data.

[0043] Unlike conventional protein or metabolite biomarkers, the platform described herein directly measures gene-regulatory activity rather than secondary effects, providing a molecular snapshot of transcriptional dysregulation that precedes amyloid and tau pathology. The technology can be readily adapted to existing sequencing or targeted-assay infrastructure, positioning it for use as a laboratory-developed test (LDT), a clinical diagnostic panel, or a companion biomarker for therapeutic trials targeting epigenetic or transcriptional pathways. Thus, the methods and materials provided herein represent the first blood-based biomarker leveraging histone acetylation for detection of AD, offering a new path toward early, mechanism-focused diagnosis. By quantifying enhancer activation signatures, it captures the epigenetic layer of disease initiation, enabling preclinical detection, patient stratification, and therapy monitoring from a minimally invasive bloodAttorney Docket No. 26100-0166W01

[0044] 2025-046

[0045] sample. Moreover, the platform’s adaptability extends its impact well beyond AD, serving as a template for epigenetic biomarker development across neurodegenerative, psychiatric, and aging-related disorders. This work narrows the boundary between molecular neuroscience and clinical application, bringing precision diagnostics one step closer to predicting, and ultimately preventing, cognitive decline.

[0046] Any appropriate mammal can be assessed for H3K27ac levels across enhancer regions as described herein. For example, humans or other primates such as monkeys can be assessed using the methods described herein. In some cases, dogs, cats, pigs, sheep, mice, rabbits, and rats can be assessed as described herein. Further, any appropriate biological sample from a mammal can be assessed for H3K27ac levels across enhancer regions as described herein. Examples of biological samples that can be assessed as described herein include, without limitation, whole blood and portions thereof (e.g., serum, plasma, or an immune-enriched buffy coat fraction). In some cases, a sample can be a serum sample. In some cases, a biological sample such as a whole blood, serum, or plasma sample can be obtained from a mammal and used to determine if the mammal has elevated H3K27ac levels across enhancer regions as described herein, which can indicate an increased likelihood (or risk) that the mammal will develop or has AD (e.g., early onset AD). An immune-enriched buffy coat fraction can be obtained using any appropriate technique(s) (e.g., using a commercially available laboratory kit to process drawn blood).

[0047] The term “elevated levels” as used herein with respect to H3K27ac levels across enhancer regions in a sample from a mammal (e.g., a human) refers to any levels that are higher than a reference level of H3K27ac across enhancer regions in a sample from a corresponding healthy mammal (e.g., a mammal of the same species that is known not to have AD). For example, elevated H3K27ac levels across enhancer regions can be levels present within a sample (e.g., a serum sample) that are at least 5 percent (e.g., at least 10 percent, at least 15 percent, at least 20 percent, at least 25 percent, at least 30 percent, at least 35 percent, at least 40 percent, at least 45 percent, at least 50 percent, at least 75 percent, or at least 100 percent) higher than the H3K27ac levels across enhancer regions in a corresponding sample from a healthy mammal.

[0048] Any appropriate method can be used to assess H3K27ac levels across enhancer regions in a sample from a mammal (e.g., a human). In some cases, H3K27ac levels can be assessed using ChlP-seq. When a mammal is determined to have elevated levels ofAttorney Docket No. 26100-0166W01

[0049] 2025-046

[0050] H3K27ac at enhancer regions (e.g., enhancer regions near AD risk loci) as described herein, the mammal can be identified as having an increased risk of developing AD, or as being likely to have AD (e.g., early onset AD).

[0051] In some cases, a method provided herein can include (a) collecting a sample (e.g., a blood sample) from a mammal (e.g., a human), (b) performing an analysis (e.g., a ChlP-seq analysis) to assess H3K27ac levels across enhancer regions in the sample, and (c) based on the results of the analysis, identifying the mammal as having an increased risk of developing AD or having AD (e.g., early onset AD), or identifying the mammal as not having an increased risk of developing or having AD (e.g., early onset AD). Any appropriate method can be used to determine whether or not a mammal is at an increased risk of developing AD or having AD. For example, a machine learning model can be used to receive data regarding measured H3K27ac levels at a plurality (e.g., 100 or more, 200 or more, 300 or more, 400 or more, 500 or more, 600 or more, or all 646) of the AD risk loci listed in TABLE 3 herein, and can then predict AD risk. In some cases, an elastic net regression machine learning model can take in data regarding measured H3K27ac levels at a plurality (e.g., 100 or more, 200 or more, 300 or more, 400 or more, 500 or more, 600 or more, or all 646) of the AD risk loci listed in TABLE 3 herein, and then can predict AD risk.

[0052] In some cases, a mammal (e.g., a human) identified as being at increased risk of developing or having AD (e.g., early onset AD) as described herein can be selected to receive one or more treatments aimed at slowing the onset and / or progression of AD. For example, a mammal identified as being at increased risk of developing or having AD based on the presence of elevated H3K27ac levels across enhancer regions (e.g., enhancer regions near AD risk loci) in a sample obtained from the mammal can be selected for treatment with one or more agents aimed at slowing the onset and / or progression of AD.

[0053] This document also provides methods for treating a mammal (e.g., a human) identified as being at increased risk of developing or having AD (e.g., early onset AD). In some cases, a mammal (e.g., a human) identified as having an increased risk of developing or having AD (e.g., early onset AD) as described herein (e.g., based, at least in part, on the presence of elevated H3K27ac levels across enhancer regions (e.g., enhancer regions near AD risk loci) in a sample obtained from the mammal) can be administered or instructed to self-administer one or more (e.g., one, two, three, four, five, or more) agents aimed at slowing the onset and / or progression of AD. When a mammal (e.g., a human)Attorney Docket No. 26100-0166W01

[0054] 2025-046

[0055] identified as being at increased risk of developing or having AD as described herein (e.g., based, at least in part, on the presence of elevated H3K27ac levels across enhancer regions in a sample obtained from the mammal) is selected for and / or administered one or more agents aimed at slowing the onset and / or progression of AD, any appropriate agent(s) can be selected and / or administered. Examples of agents that can be administered in the methods described herein include, without limitation, donanemab (Kisunla®) and / or lecanemab (Leqembi®).

[0056] When one or more agents are administered to a mammal identified as being at increased risk of developing or having AD as described herein, any appropriate amount (e.g., any appropriate dose) of an agent (e.g., donanemab and / or lecanemab) can be administered to the mammal. A therapeutically effective amount can be any amount that can slow the onset and / or progression of AD in a mammal (e.g., a human), without producing significant toxicity to the mammal. For example, a therapeutically effective amount of donanemab can be from about 300 mg to about 1500 mg (e.g. about 350 to about 1400 mg, about 700 to about 1000 mg, about 800 to about 1200 mg, about 1000 to about 1400 mg, about 350 mg, about 400 mg, about 500 mg, about 700 mg, about 800 mg, about 1000 mg, about 1050 mg, about 1200 mg, about 1400 mg, or about 1500 mg). A therapeutically effective dose of lecanemab can be from about 1 mg / kg to about 50 mg / kg (e.g., from about 1 to about 10 mg / kg, from about 10 to about 20 mg / kg, from about 20 to about 30 mg / kg, from about 30 to about 50 mg / kg, about 5 mg / kg, about 10 mg / kg, about 15 mg / kg, about 20 mg / kg, about 25 mg / kg, or about 50 mg / kg). In some cases, a therapeutically effective amount can remain constant or can be adjusted as a sliding scale or variable dose depending on the mammal’s response to the treatment. Various factors can influence the actual effective amount used for a particular application. For example, the frequency of administration, duration of treatment, use of multiple treatment agents, route of administration, and / or severity of the mammal’s condition may require an increase or decrease in the actual effective amount administered.

[0057] The frequency of administration of one or more agents can be any frequency that slows the onset and / or progression of AD in the mammal, without producing significant toxicity to the mammal. For example, the frequency of administration of an agent (e.g., donanemab) can be from about once a week to about once every two months (e.g., from about once a week to about once every other week, from about once every other week to about once a month, from about once a month to about once every other month, aboutAttorney Docket No. 26100-0166W01

[0058] 2025-046

[0059] once every two weeks, about once a month, or about once every two months). The frequency of administration can remain constant or can be variable during the duration of treatment. As with the effective amount, various factors can influence the actual frequency of administration used for a particular application. For example, the effective amount, duration of treatment, use of multiple treatments, route of administration, and severity of the mammal’s condition may require an increase or decrease in administration frequency.

[0060] An effective duration for administering one or more agents can be any duration that slows the onset and / or progression of AD in the mammal, without producing significant toxicity to the mammal. In some cases, the effective duration can vary from one month to two years or longer (e.g., from about one to about three months, from about three to about six months, from about six to about 12 months, from about 12 to about 18 months, from about 18 to about 24 months, about three months, about six months, about 12 months, about 18 months, or about 24 months). Multiple factors can influence the actual effective duration used for a particular treatment. For example, an effective duration can vary with the effective amount, frequency of administration, use of multiple treatments, route of administration, and severity of the mammal’s condition.

[0061] Administration to a mammal (e.g., a human) of one or more agents aimed at slowing the onset and / or progression of AD can result in a reduced level of amyloid-beta plaques in the mammal. The term “reduced level” as used herein with respect to a level of amyloid-beta plaques in a mammal refers to any level that is lower than the level of amyloid-beta plaques observed in that mammal prior to being treated with the one or more agents. In some cases, a reduced level of amyloid-beta plaques can be a level that is at least 5 percent (e.g., at least 10, at least 15, at least 20, at least 25, at least 35, at least 50, at least 75, or at least 90 percent) lower than the level of amyloid-beta plaques prior to the mammal being treated as described herein. In some cases, a reduced level of amyloidbeta plaques can be a level that is at least 1.5-fold (e.g., about 1.5-fold, about 2-fold, about 2.5-fold, about 3-fold, about 3.5-fold, about 4-fold, about 5-fold, or more than 5-fold) lower than the level of amyloid-beta plaques prior to being treated as described herein. It will be appreciated that levels from comparable samples are used when determining whether or not a particular level is a reduced level.

[0062] In some cases, when treating a mammal (e.g., a human) identified as being at increased risk of developing or having AD as described herein (e.g., based, at least in part,Attorney Docket No. 26100-0166W01

[0063] 2025-046

[0064] on the presence of elevated levels H3K27ac across enhancer regions (e.g., enhancer regions at AD risk loci) in a sample obtained from the mammal), the treatment can be effective to reduce the amount of amyloid-beta plaques in the mammal (e.g., in the brain of the mammal). For example, the methods and materials described herein can be used to reduce the level of amyloid-beta plaques in a mammal by, for example, 10, 20, 30, 40, 50, 60, 70, 80, 90, 95, or more than 95 percent. In some cases, the methods and materials described herein can be used to reduce the amount of amyloid-beta plaques in a mammal by, for example, at least 1.5-fold (e.g., about 1.5-fold, about 2-fold, about 2.5-fold, about 3-fold, about 3.5-fold, about 4-fold, about 5-fold, or more than 5-fold).

[0065] In some cases, when treating a mammal (e.g., a human) identified as being at increased risk of developing or having AD (e.g., early onset AD) as described herein (e.g., based, at least in part, on the presence of elevated H3K27ac levels across enhancer regions (e.g., enhancer regions at AD risk loci) in a sample obtained from the mammal), the treatment can be effective to slow the development and / or progression of AD in the mammal. For example, the methods and materials described herein can be effective to slow the progression of AD in a mammal by, for example, 10, 20, 30, 40, 50, 60, 70, 80, 90, 95, or more than 95 percent. In some cases, the methods and materials described herein can be used to slow the progression of AD in a mammal by, for example, at least 6 months (e.g., at least 6 months, at least 8 months, at least 10 months, at least 1 year, at least 1.5 years, at least 2 years, at least 2.5 years, at least 3 years, at least 4 years, or at least 5 years). In some cases, the methods and materials described herein can be used to delay the onset of AD in a mammal by, for example, at least 6 months (e.g., at least 6 months, at least 8 months, at least 10 months, at least 1 year, at least 1.5 years, at least 2 years, at least 2.5 years, at least 3 years, at least 4 years, or at least 5 years).

[0066] In some cases, when treating a mammal (e.g., a human) identified as being at increased risk of developing or having AD (e.g., early onset AD) as described herein (e.g., based, at least in part, on the presence of elevated H3K27ac levels across enhancer regions (e.g., enhancer regions at AD risk loci) in a sample obtained from the mammal), the treatment can be effective to increase the survival of the mammal. For example, the methods and materials described herein can be used to increase the survival of a mammal at increased risk of developing or having AD by, for example, 10, 20, 30, 40, 50, 60, 70, 80, 90, 95, or more than 95 percent.Attorney Docket No. 26100-0166W01

[0067] 2025-046

[0068] In some cases, a course of treatment (e.g., treatment with one or more agents such as donanemab and / or lecanemab) can be monitored in a mammal. For example, the amount amyloid-beta plaques within a mammal can be monitored over the course of treatment (e.g., with donanemab and / or lecanemab) to determine whether or not the treatment is effective and / or remains effective over time for slowing the onset and / or progression of AD. In some cases, this document provides methods and materials for assessing treatments for AD, as well as methods and materials for effectively using such treatments within mammals to slow the onset and / or progression of AD. For example, this document provides methods and materials for determining whether or not a particular treatment administered to a mammal (e.g., a human identified as being at increased risk of developing or having AD as described herein) is effective to slow the onset and / or progression of AD in that mammal.

[0069] In some cases, determining that a mammal (e.g., a human) has a reduced level of amyloid-beta plaques following administration of a treatment (e.g., donanemab and / or lecanemab), as compared to the level of amyloid-beta plaques prior to being treated, can indicate that that treatment is effective in that mammal and can be continued as is (or with one or more slight alterations). In some cases, determining that a mammal (e.g., a human) does not have a reduced level of amyloid-beta plaques following administration of a treatment (e.g., donanemab and / or lecanemab), as compared to the level of amyloid-beta plaques prior to being administered the treatment, can indicate that that treatment is not effective in that mammal and can be discontinued, altered (e.g., increased) in amount and / or frequency, or changed to a different type of treatment in that mammal.

[0070] In some cases, to assess the effectiveness of a treatment (e.g., donanemab and / or lecanemab) aimed at slowing the onset and / or progression of AD in a mammal (e.g., a human), a sample (e.g., a blood sample, a serum sample, a plasma sample, or an immune-enriched buffy coat sample) can be obtained from the mammal at any appropriate time after the mammal has been administered the treatment, and the sample can be assessed to determine the level of amyloid-beta plaques in the mammal. For example, a sample (e.g., a blood sample, a serum sample, a plasma sample, or an immune-enriched buffy coat sample) can be obtained from a mammal (e.g., a human) between about one month and about six months (e.g., about one to two months, about two to three months, about three to four months, or about four to six months) after an initial administration of a treatment aimed at slowing the onset and / or progression of AD, and the sample can be assessed toAttorney Docket No. 26100-0166W01

[0071] 2025-046

[0072] determine the level of amyloid-beta plaques within the mammal at that time. Such a determined level of amyloid-beta plaques within the mammal after being administered the treatment can be used as an indicator of whether the development and / or progression of AD within the mammal is reduced following the treatment. A reduction in the level of amyloid-beta plaques within the mammal after the mammal has been administered the treatment can indicate that the treatment was effective in that mammal. In some cases, a reduction in the level of H3K27ac within the mammal after the mammal has been administered the treatment may indicate that the treatment was effective in that mammal.

[0073] In some cases, a first sample (e.g., a first blood sample, a first serum sample, a first plasma sample, or a immune-enriched buffy coat sample) can be obtained from a mammal (e.g., a human) prior to the mammal being administered a treatment (e.g., donanemab and / or lecanemab) aimed at slowing the onset and / or progression of AD, and a second sample (e.g., a second blood sample, a second serum sample, a second plasma sample, or a second immune-enriched buffy coat sample), and optionally subsequent samples, can be obtained from the mammal after the mammal has been administered the treatment. When the level of amyloid-beta plaques in the second sample is less than (e.g., at least 1, 2, 3, 4, 5, 10, 15, 20, 25, or 50 percent less than) the level of amyloid-beta plaques in the first sample, then the treatment can be determined and / or classified as being an effective treatment for that mammal. When the level of amyloid-beta plaques in the second sample is greater than (e.g., at least 1, 2, 3, 4, 5, 10, 15, 20, 25, or 50 percent greater than) or about equal to the level of amyloid-beta plaques in the first sample, then the treatment can be determined and / or classified as not being an effective treatment for that mammal.

[0074] Exemplary Embodiments

[0075] Embodiment 1 is 1 method for identifying a mammal as having, being likely to have, or being likely to develop AD, wherein the method comprises measuring, in a blood sample from the mammal, levels of H3K27ac at a plurality of genomic regions; determining that the levels of H3K27ac are elevated as compared to reference levels of H3K27ac at the plurality of genomic regions; and identifying the mammal as having, being likely to have, or being likely to develop AD.

[0076] Embodiment 2 is the method of embodiment 1, wherein the mammal is a human.Attorney Docket No. 26100-0166W01

[0077] 2025-046

[0078] Embodiment 3 is the method of embodiment 1 or embodiment 2, wherein the plurality of genomic regions are near AD risk loci.

[0079] Embodiment 4 is the method of embodiment 3, wherein the plurality of genomic regions comprises 500 or more genomic regions near the AD risk loci.

[0080] Embodiment 5 is the method of any one of embodiments 1-4, wherein the measuring comprises ChlP-seq of genomic DNAfrom the blood sample.

[0081] Embodiment 6 is the method of any one of embodiments 1-5, wherein the blood sample is a buffy coat sample.

[0082] Embodiment 7 is the method of any one of embodiments 1-6, wherein the reference level of H3K27ac is the level of H3K27ac in a blood sample from a mammal known not to have AD.

[0083] Embodiment 8 is the method of any one of embodiments 1-7, further comprising administering, to the mammal, a therapeutic agent for treating the AD.

[0084] Embodiment 9 is the method of embodiment 8, wherein the therapeutic agent comprises donanemab or lecanemab.

[0085] Embodiment 10 is a use of a therapeutic agent to treat a mammal identified as having, being likely to have, or being likely to develop, AD, wherein the mammal was identified as having, being likely to have, or being likely to develop AD using a method comprising measuring, in a blood sample from the mammal, levels of H3K27ac at a plurality of genomic regions; and determining that the levels of H3K27ac are elevated as compared to reference levels of H3K27ac at the plurality of genomic regions, thereby identifying the mammal as having, being likely to have, or being likely to develop AD.

[0086] Embodiment 11 is the use of embodiment 10, wherein the mammal is a human. Embodiment 12 is the use of embodiment 10 or embodiment 11, wherein the plurality of genomic regions are near AD risk loci.

[0087] Embodiment 13 is the use of embodiment 12, wherein the plurality of genomic regions comprises 500 or more genomic regions near the AD risk loci.

[0088] Embodiment 14 is the use of any one of embodiments 10-13, wherein the therapeutic agent comprises donanemab or lecanemab.

[0089] Embodiment 15 is the use of any one of embodiments 10-14, wherein the measuring comprises ChlP-seq of genomic DNAfrom the blood sample.

[0090] Embodiment 16 is the use of any one of embodiments 10-15, wherein the blood sample is a buffy coat sample.Attorney Docket No. 26100-0166W01

[0091] 2025-046

[0092] Embodiment 17 is the use of any one of embodiments 10-16, wherein the reference level of H3K27ac is the level of H3K27ac in a blood sample from a mammal known not to have AD.

[0093] Embodiment 18 is a method for using a therapeutic agent to treat a mammal identified as having, being likely to have, or being likely to develop, AD, wherein the mammal was identified as having, being likely to have, or being likely to develop AD using a method comprising measuring, in a blood sample from the mammal, levels of H3K27ac at a plurality of genomic regions; and determining that the levels of H3K27ac are elevated as compared to reference levels of H3K27ac at the plurality of genomic regions, thereby identifying the mammal as having, being likely to have, or being likely to develop AD.

[0094] Embodiment 19 is the method of embodiment 18, wherein the mammal is a human.

[0095] Embodiment 20 is the method of embodiment 18 or embodiment 19, wherein the plurality of genomic regions are near AD risk loci.

[0096] Embodiment 21 is the method of embodiment 20, wherein the plurality of genomic regions comprises 500 or more genomic regions near the AD risk loci.

[0097] Embodiment 22 is the method of any one of embodiments 18-21, wherein the therapeutic agent comprises donanemab or lecanemab.

[0098] Embodiment 23 is the method of any one of embodiments 18-22, wherein the measuring comprises ChlP-seq of genomic DNAfrom the blood sample.

[0099] Embodiment 24 is the method of any one of embodiments 18-23, wherein the blood sample is a buffy coat sample.

[0100] Embodiment 25 is the method of any one of embodiments 18-24, wherein the reference level of H3K27ac is the level of H3K27ac in a blood sample from a mammal known not to have AD.

[0101] The invention will be further described in the following example, which does not limit the scope of the invention described in the claims.

[0102] EXAMPLE

[0103] Example 1 - Genetic risk informs a blood histone acetylation biomarker for Alzheimer’s disease

[0104] Methods and MaterialsAttorney Docket No. 26100-0166W01

[0105] 2025-046

[0106] Study subjects / samples and ChlP-seq'. The immune cell fraction of the blood (buffy-coat; Cottier-Fox et al., “CHAPTER 24 - Collection and processing of marrow and blood hematopoietic stem cells,” in Hematopoietic Stem Cell Transplantation in Clinical Practice, eds. Treleaven and Barrett, pp. 249-256, Churchill Livingstone, Edinburgh, 2009) was obtained from 145 subjects, including 92 AD patients and 53 control subjects (FIG. 1A and TABLE 1). Using 1 to 4 samples from each subject’s source tube (technical replicates), ChlP-seq was conducted to assess the genome-wide locations of histone acetylation associated with enhancer and promoter activity (H3K27ac). The resulting ChlP-seq reads were aligned to the genome and referred to as “enriched locations (peaks) of high H3K27ac signal” or putative “active regulatory elements” (AREs).

[0107] Quality control ofH3K27ac ChlP-seq'. Peak calling was carried out with the ENCODE pipeline (github.com / kundajelab / chipseq_pipeline; Nature 583:699-710, 2020), following work described elsewhere (Hou et al., Nat. Genet. 55:1665-1676, 2023).

[0108] Briefly, reads for each sample were mapped to human genome assembly hgl9 (bwa, vO.5.9, command: bwa aln Homo_sapiens_assemblyl9.fasta -q 5 -132 -k 2 -t $NSLOTS -o 1 -f), low quality and multiple-mapped reads were filtered (samtools, vl.3.1), and MACS2 (v2.1.1) was applied for peak calling. Samples were then filtered based on three quality control (QC) metrics: (1) Relative Strand Cross-correlation coefficient (RSC), based on the (shifted) agreement between forward-strand and reverse-strand reads, a measure of signal-to-noise enrichment ratio; (2) total reads; and (3) H3K27ac signal correlation between the samples and the samples from Roadmap epigenomes based on 405k tissue-specific AREs identified from Roadmap. Samples with RSC score > 0.6, total reads > 5M, and blood Roadmap tissues / cell types in the top three most correlated Roadmap epigenomes were retained for downstream analysis. These criteria resulted in 28 age-matched health controls and 58 AD samples.

[0109] Peak detection, ARE activity quantification, and normalization '. To get robust peaks, all reads from QC-ed AD and control samples were aggregated separately, broad peak calling (parameters: -f BEDPE -g hs -q 0.05 -broad-cutoff 0.2 — outdir , / AD broad / -n $nm.q0.05_0.2 -B — SPMR —broad) were carried out for each, and the two peak sets were merged (bedtools merge) as the final reference peak set. Only those peaks detected in Roadmap and Blueprint H3K27AC profilings from major immune cell types, including CD4+ T naive cells, CD4+ T memory cells, CD8+ T naive cells, CD8+ T memory cells,Attorney Docket No. 26100-0166W01

[0110] 2025-046

[0111] monocytes, B cells, nature killer cells, and neutrophils, were kept. A report described elsewhere (Hou et al., supra) was followed to quantify the activity of each ARE for each sample based on fragment coverage on a specific ARE divided by its length:

[0112]

[0113] where Activityi krepresents the activity of AREi in sample j, Fragjkrepresents a fragment j extended from a readj toward the 3' end to the same length as the estimated fragment length in sample k, overlap(Fragk, AREf represents the length of the overlapping part between Fragjkand ARE^ and Lengthi is the length of AREi. ARE activity was normalized for read depths of different samples, by a factor estimated for each sample based on geometry mean of reads coverage of all AREs as implemented in DESeq2 package (Love et al., Genome Biol. 15(12): 550, 2014), and then corrected for bias due to GC-content (Teng and Irizarry, Genome Res. 27:1930-1938, 2017).

[0114] Differential ChlP-seq ARE identification'. Limma voom was used to identify differential ChlP-seq AREs. First, batch effects in the data were removed using the removeBatchEffect command. Then, GC -bias-corrected and normalized mean ARE heights were modeled as a linear combination of AD status, sex, age, NSC, and RSC (the latter two being ChlP-seq quality control metrics). Fold changes and p-values were estimated based on the AD status contrast.

[0115] Module generation and annotation'. To compute patterns of activity in the blood acetylation data across different tissues in reference data sets, H3K27ac ChlP-seq samples of 59 tissues were obtained from ENCODE. Since these data were mapped to Genome Reference Consortium Human Build 38 (hg38), liftOver was used to map the AREs to hg38 as well, and -loglOP signal was computed for each tissue across the AREs (bigwigAverageOverBed). These data were binarized by a cutoff of 2 and then clustered by the k-centroids algorithm implemented in the R package flexclust using Jaccard distance as the distance metric and cluster number set at 49 (corresponding to about 1000 AREs per module on average).

[0116] Gene Ontology enrichment analysis'. To determine the biological pathways associated with genes near differentially acetylated AREs, a gene ontology (GO) enrichment analysis was conducted using Genomic Regions Enrichment of Annotations Tool (GREAT; McLean et al., Nat. Biotechnol. 28:495-501, 2010) for differentially acetylated AREs from each DESeq2 analysis. For each analysis, significant (p<0.05) AD-Attorney Docket No. 26100-0166W01

[0117] 2025-046

[0118] increased or AD-decreased AREs, or AD trait associated AREs, were used as the foreground, and the full ARE set was used as the background.

[0119] Transcription factor motif enrichment analysis'. Hypergeometric Optimization of Motif Enrichment (HOMER) was used to identify transcription factor motifs that were enriched within differentially acetylated region sets, using genomic regions as input (fmdMotifsGenome.pl command) and the full peak set as the background. Motifs from regions of 500 bp were searched for, as was recommended for H3K27ac ChlP-seq peaks.

[0120] Selection of AD genetic risk associated regions'. To test the predictive ability of AREs near disease-associated SNPs for AD diagnosis, the predictiveness of equally-sized sets of AREs residing near genetic risk loci for AD, schizophrenia (Ripke et al., Nature 511(7510):421-427, 2015), Type 1 Diabetes, and Type 2 Diabetes were assessed, with ARE sets not derived from AD genetic risk loci controlling for proximity to disease-associated SNPs in general.

[0121] To obtain ARE sets, lists of genome-wide significant SNPs for each genome-wide association study were first obtained from the GWAS Catalog. The hgl9 annotated locations of each SNP were obtained using NCBI Biomart. 50KB windows were obtained around each SNP location using bedtools slop (-b 25000). These windows were then intersected with blood AREs and each ARE was annotated with its corresponding SNP (i.e., a SNP 50 kb upstream or downstream from the ARE). AREs intersecting with multiple SNP windows were annotated to the SNP with minimum p-value. AREs not intersecting with any SNP windows were not assigned a corresponding SNP, and were filtered out to achieve the sets of trait-associated AREs. When comparing these ARE sets to multiple disease-associated ARE sets, n=646 matched the size of the smallest disease-associated ARE set (T1D).

[0122] Logistic regression model creation'. To assess the predictiveness of histone-acetylated genomic regions in blood during AD, a logistic regression model was fit to several different ARE sets. For each model, LI (LASSO) regularization was used, and cross-validation (CV) with 5 folds was performed to estimate the performance of the model. To train the model on as much data as possible, cross-validation metrics averaged across folds were reported in lieu of performance on a held-out set.

[0123] To optimize the regularization strength, the GridSearchCV python module was used to find the optimal regularization penalty coefficient from 30 options ranging from IE-4 to 1E+4 in a logarithmic scale. For each potential coefficient, a model was fit withAttorney Docket No. 26100-0166W01

[0124] 2025-046

[0125] 5-fold CV and the given regularization coefficient, and the metrics corresponding to the best-performing model were reported (i.e., with optimal coefficient). To ensure fair comparisons between models with different input ARE sets and / or shuffled labels, this hyperparameter optimization was only performed for (1) the model fit to AD genetic risk associated AREs, and (2) the model fit to all AREs.

[0126] The optimal coefficient from the AD genetic risk associated ARE model was used when fitting models to (1) non-AD disease genetic risk associated AREs, (2) AREs near randomly-selected genomic loci, and (3) AD disease genetic risk associated AREs with AD status labels shuffled. The optimal coefficient from the model fit to all AREs was used when fitting models to all AREs with AD status labels shuffled.

[0127] Validation of predictiveness of AD-associated regions'. To test whether AREs near random positions on the genome would be more predictive of AD diagnosis than other AREs, the predictiveness of 100 equally-sized sets of AREs near randomly selected genomic positions was assessed. A set of AREs associated with randomly selected background regions on the genome that were not AD-associated was created. First, 10,000 random SNPs (DBSNP 151) were sampled and 50KB windows around each position (bedtools slop -b 25000) were taken, removing windows that overlapped windows generated about AD-associated SNPs (bedtools subtract). An / / -sized random sample of the intersection of this set of windows with blood AREs was taken by randomly permuting the bedtools intersect output with the GNU utility shuf and taking the first n AREs in the permutation, and the / / -sized random sample was used as a random position-associated ARE set. When comparing these ARE sets to multiple disease-associated ARE sets, n=646 was set to match the size of the smallest disease-associated ARE set. When comparing these ARE sets to the AD-associated ARE set alone, n=1997 to match the size of the uncut AD-associated ARE set.

[0128] Results

[0129] H3K27ac ChlP-seq of subjects with and without AD'. Histone modifications associated with enhancer and promoter activity (H3K27ac) from the immune cell fraction of the blood (buffy-coat) were profiled across 145 subjects, including 92 AD and 53 control subjects (FIG. 1A; TABLE 1), each of which was associated with 1 to 4 samples from each subject’s source tube, were assessed genome-wide using chromatin immunoprecipitation followed by ChlP-seq. The ChlP-seq reads were aligned to theAttorney Docket No. 26100-0166W01

[0130] 2025-046

[0131] genome and referred to as “enriched locations (peaks) of high H3K27ac signal” or putative “active regulatory elements” (AREs). To increase sensitivity and reduce noise in the AD-specific marker identification, the union of AREs reproducible across multiple groups of AD and control subjects were taken as described above.

[0132] TABLE 1 : Numbers of male and female subjects in AD and control groups pre- and post- QC

[0133]

[0134] Blood H3K27ac ChlP-seq activity across tissues and cell types'. To distinguish the potential roles of putative gene-regulatory elements captured by the AREs, their patterns of activity were compared to H3K27ac profiles from different healthy tissues in reference data sets (FIG. IB). Their activity levels were determined by assessing the signal at each of the histone-acetylated regions found across a defined reference panel of 59 tissue and cell types based on tissue-specific H3K27ac ChlP-seq AREs (ENCODE Project Consortium, Nature 489:57-74, 2012).

[0135] A total of 49 distinct patterns of H3K27ac activity were defined across tissues in reference epigenomic datasets, which were used to group the 49,874 AREs into 49 modules. Each module was annotated by its enrichment of AD differentially acetylated member AREs as described above, and 23 modules were identified that each included a significant enrichment of AD-up AREs (which are more active in AD) or AD-down AREs (which are less active in AD) relative to background of the full ARE set. AD- down-enriched modules showed more activity in immune tissues than AD-up-enriched modules. In particular, modules 10 and 44 showed activity across all 10 immune tissues surveyed, as well as the temporal lobe (FIG. 1C). A representative graph and chart illustrating activity near APOE and surrounding genes are presented in FIG. ID, with SNPs annotated.

[0136] Blood epigenomic activity in regions associated with AD genetic risk is predictive for AD'. To assess whether genome-wide blood H3K27ac enrichment is predictive for AD, a predictive model of AD status was leveraged from blood acetylation data using allAttorney Docket No. 26100-0166W01

[0137] 2025-046

[0138] 49,874 AREs as input features. An average balanced accuracy of 80%, average auROC of 0.905, and average auPRC of 0.949 were achieved from 5 cross-validation folds (FIG. 2A). A logistic regression model with LI regularization was used to achieve this result.

[0139] To ensure that this result was not spurious, the labels of individuals with and without AD were randomized and the entire classification pipeline was re-run using all 49,874 acetylated regions in the blood H3K27ac dataset as input features. This resulted in an average of 57% balanced accuracy, with a standard deviation of 11.7% for distinguishing AD vs. non-AD across 100 randomizations.

[0140] To assess why this predictive relationship was observed, a predictive model for AD status was again leveraged based on blood H3K27ac epigenomic profiles. The input H3K27ac ARE sets in the model were restricted in the vicinity of significant (p < 5E-8) AD-associated SNPs in an AD genome-wide association study (GWAS; Jansen et al., Nat. Genet. 51(3):404-413, 2019). A diameter of 50 kb surrounding each AD GWAS locus was used to capture potential long-range gene-regulatory effects, resulting in N=1997 AD loci (TABLE 2), covering 0.085% of the genome.

[0141] TABLE 2: Number of loci used for logistic regression model creation, coverage of 50kb windows and final ARE sets

[0142]

[0143] To isolate the effect of the AD genetic risk association in the blood epigenome, these AREs were not overlapped with any blood differential AREs. Since annotations of AD status are binary, a logistic regression model was used to predict AD status from the enrichment at each ARE. This restricted set showed increased predictive accuracy for AD status (84.3% vs. 70.6% random loci) based on 5-fold cross-validation, FIG. 2B). ToAttorney Docket No. 26100-0166W01

[0144] 2025-046

[0145] ensure this result was not spurious, randomized SNP loci were used as a control, as were loci from other highly-powered genome wide association studies (GWAS) that share some risk factors and genetic risk signatures with AD (Jayaraman and Pike, Curr. Diab. Rep. 14:476, 2014), including schizophrenia (SCZ; Schizophrenia Working Group of the Psychiatric Genomics Consortium. Nature 511:421-427, 2014), Type 1 Diabetes (T1D; Farh et al., Nature 518:337-343, 2015), which has a blood basis, and Type 2 Diabetes (T2D; Scott et al., Diabetes 66:2888-2902, 2017) (FIGS. 2B-2C). The results of these analyses demonstrated that loci associated with AD predisposition provide an informative prior for the identification of an epigenomic biomarker. That is, genetic information about which regions are associated with AD can be used to identify which regions are likely to change in H3K27ac with AD. The fact that the genetic predisposition (mutations) tend to overlap the H3K27ac regions is a good indicator that the biomarker is picking up on relevant biology. In addition, the fact that blood regulatory regions near genetic risk loci show this association with AD implies a strong relationship between the pathways influencing AD predisposition and the pathways altered in the blood of AD patients.

[0146] To further assess whether this was driven by artifacts in the datasets or by the methodology, the labels of individuals with and without AD were randomized and the entire analysis pipeline was re-run using the 646 blood acetylated regions near AD genetic risk SNPs (TABLE 3) as the input features (FIG. 2D). This resulted in effectively random accuracy for distinguishing AD vs. non-AD, on average across 100 randomizations (average balanced accuracy: 50.7%, average auROC: 0.504, average auPRC: 0.698).

[0147] The studies discussed herein are the first to use histone acetylation biomarkers in the blood to predict AD status. The studies were based on the development of a robust framework that was initially used to identify candidate biomarkers for AD, and was then applied to a dataset for buffy-coat blood cells in order to drive the biomarker discovery process for AD and generate a sparse candidate set of biomarkers. The identified biomarker outperformed gold-standard CSF measures for detecting AD (Palmqvist et al., Neurology, 85: 1240-1249, 2015), and is on par with other proposed plasma measurements such as plasma p-tau217 (Ashton et al., JAMA Neurol. 81(3):255-263, 2024; Brum et al., Nat Aging 3:1079-1090, 2023; and Janelidze et al., Brain 146:1592-1601, 2023). This strong performance not only confirms the predictive nature of blood epigenomic enrichments for AD, but also represents an inexpensive and clinically viable testingAttorney Docket No. 26100-0166W01

[0148] 2025-046

[0149] option for AD. This work has significant, far-reaching applications, not only for biomarker discovery but also for potential AD treatments and cures. Potential AD treatments based on this work may include bolstering the blood immune system to better promote neurogenesis and cognition, as well as developing epigenomic drugs to target identified biomarkers. Epigenomic drugs may repress AREs that are higher in AD and / or activate AREs that are lower in AD, in order to combat disease progression.

[0150] The ability to detect, monitor, and treat AD early is crucial for effective treatment and prevention of the disease. It is well-established that AD therapies should be initiated when a person shows no symptoms or when they exhibit mild cognitive impairment (MCI) (Cummings et al., Neurol. 69:1622-1634, 2007). The work described herein is broadly significant because epigenomic measurements, including H3K27ac, can show more stability than protein levels, gene expression, or other commonly-used biomarkers. Ultimately, this may allow for earlier detection and treatment of AD, before large-scale neuron loss and clinical features emerge. For example, Ap inhibitors utilized early on during AD pathogenesis may clear amyloid in pathology, leading to downstream cognitive and behavioral improvements. The early detection of AD, before symptoms start to set in, can allow for an even earlier initiation of existing AD treatments.

[0151] Additionally, a highly predictive blood marker only greatly decreases the invasiveness and expense of AD diagnosis, but allows for widespread screening across whole populations to assign AD risk and detect the disease en masse.

[0152] The studies described herein revealed a co-occurrence (stronger: convergence) of AD genetic risk and blood epigenomic signal predictive of AD status. Given the potential functional impact of AD variants in peripheral cells as well as microglia or neurons, the mechanisms predisposing individuals to AD may be partially borne out in blood cell function (Ramamurthy et al., PLoS Comput. Biol. 20:el012356, 2024; and Novikova et al., Nat. Commun. 12:1610, 2021). New disease-modifying drugs for AD may work by decreasing neuroinflammation, and determining to what extent the inflammatory responses of blood cells contribute to neuroinflammation may provide a viable alternative path to treating AD without having to cross the blood-brain barrier. Thus, blood cell inflammatory responses in AD may inform new disease-modifying treatments for AD that are aimed at modulating the brain inflammatory response.Attorney Docket No. 26100-0166W01

[0153] 2025-046

[0154] TABLE 3: 646 blood acetylated regions near AD genetic risk SNPs

[0155]

[0156]

[0157] Attorney Docket No. 26100-0166W01

[0158] 2025-046

[0159]

[0160]

[0161] Attorney Docket No. 26100-0166W01

[0162] 2025-046

[0163]

[0164]

[0165] Attorney Docket No. 26100-0166W01

[0166] 2025-046

[0167]

[0168]

[0169] Attorney Docket No. 26100-0166W01

[0170] 2025-046

[0171]

[0172]

[0173] Attorney Docket No. 26100-0166W01

[0174] 2025-046

[0175]

[0176]

[0177] Attorney Docket No. 26100-0166W01

[0178] 2025-046

[0179]

[0180]

[0181] Attorney Docket No. 26100-0166W01

[0182] 2025-046

[0183]

[0184]

[0185] OTHER EMBODIMENTS

[0186] It is to be understood that while the invention has been described in conjunction with the detailed description thereof, the foregoing description is intended to illustrate and not limit the scope of the invention, which is defined by the scope of the appended claims. Other aspects, advantages, and modifications are within the scope of the following claims.

Claims

1. Attorney Docket No. 26100-0166W012025-046WHAT IS CLAIMED IS:

1. A method for identifying a mammal as having, being likely to have, or being likely to develop Alzheimer’s disease (AD), wherein said method comprises:measuring, in a blood sample from said mammal, levels of histone H3 lysine 27 acetylation (H3K27ac) at a plurality of genomic regions;determining that said levels of H3K27ac are elevated as compared to reference levels of H3K27ac at said plurality of genomic regions; andidentifying said mammal as having, being likely to have, or being likely to develop AD.

2. The method of claim 1, wherein said mammal is a human.

3. The method of claim 1, wherein said plurality of genomic regions are near AD risk loci.

4. The method of claim 3, wherein said plurality of genomic regions comprises 500 or more genomic regions near said AD risk loci.

5. The method of claim 1, wherein said measuring comprises chromatin immunoprecipitation followed by sequencing (ChlP-seq) of genomic DNA from said blood sample.

6. The method of claim 1, wherein said blood sample is a buffy coat sample.

7. The method of claim 1, wherein said reference level of H3K27ac is the level of H3K27ac in a blood sample from a mammal known not to have AD.

8. The method of claim 1, further comprising administering, to said mammal, a therapeutic agent for treating said AD.

9. The method of claim 8, wherein said therapeutic agent comprises donanemab or lecanemab.

10. Use of a therapeutic agent to treat a mammal identified as having, being likely to have, or being likely to develop, Alzheimer’s disease (AD), wherein said mammal was identified as having, being likely to have, or being likely to develop AD using a method comprising:Attorney Docket No. 26100-0166W012025-046measuring, in a blood sample from said mammal, levels of histone H3 lysine 27 acetylation (H3K27ac) at a plurality of genomic regions;and determining that said levels of H3K27ac are elevated as compared to reference levels of H3K27ac at said plurality of genomic regions, thereby identifying said mammal as having, being likely to have, or being likely to develop AD.

11. The use of claim 10, wherein said mammal is a human.

12. The use of claim 10, wherein said plurality of genomic regions are near AD risk loci.

13. The use of claim 12, wherein said plurality of genomic regions comprises 500 or more genomic regions near said AD risk loci.

14. The use of claimlO, wherein said therapeutic agent comprises donanemab or lecanemab.

15. The use of claim 10, wherein said measuring comprises chromatin immunoprecipitation followed by sequencing (ChlP-seq) of genomic DNA from said blood sample.

16. The use of claim 10, wherein said blood sample is a buffy coat sample.

17. The use of claim 10, wherein said reference level of H3K27ac is the level of H3K27ac in a blood sample from a mammal known not to have AD.

Images