Biomarker combinations for alzheimer's disease in apo e epsilon 3 / epsilon 3 genotype individuals

Abstract

The present application relates to the application of biomarker combination in the preparation of Alzheimer's disease detection kit for APOE epsilon 3 / epsilon 3 genotype individual, the biomarker is selected from at least one of ANGPTL1, PON2, PTN, SCGB3A1, PBLD, MORC3, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, GSTA3, EIF2AK3, CTSD, FOLH1, realize the high accuracy, high specificity in vitro diagnosis of Alzheimer's disease in the population.

Description

Technical Field

[0001] This invention relates to the field of diagnostic and assessment technology for mental disorders, and more particularly to a combination of biomarkers for Alzheimer's disease in individuals with the APOE ε3 / ε3 genotype. Background Technology

[0002] Currently, the clinical diagnosis and early diagnosis of Alzheimer's disease (AD) mainly rely on the following types of technologies, all of which have significant limitations when applied to large-scale population screening or early intervention: (1) Neuroimaging examinations (such as Aβ-PET, Tau-PET, structural MRI) High cost and poor accessibility: PET scan equipment is expensive, with extremely high costs per scan, and it must be performed in large medical centers, making it difficult to use as a routine screening or dynamic monitoring method. Periodicity and radiation exposure: PET scans involve radiation and should not be repeated in a short period of time, making them unsuitable for frequent use in assessing disease progression or monitoring treatment effectiveness.

[0003] (2) Detection of cerebrospinal fluid biomarkers (such as Aβ42, p-Tau, t-Tau) Invasive procedure and high risk: Obtaining cerebrospinal fluid requires lumbar puncture, which is an invasive procedure with low patient compliance and risks of complications such as infection and headache, making it completely unsuitable for early screening of asymptomatic individuals. Complex sample processing: The collection, transportation, and storage of cerebrospinal fluid require extremely stringent conditions, the standardization of procedures is low, and test results vary greatly between different centers, affecting comparability.

[0004] (3) Reported universal blood biomarkers Insufficient sensitivity and specificity: In the early or very early stages of the disease, the concentration of AD-related proteins in the blood changes only slightly and is greatly affected by individual differences and other systemic diseases. This results in the diagnostic accuracy (AUC value) of a single or a few biomarkers often being at a critical level of 0.7-0.8, which cannot meet the strict requirements for clinical diagnosis. Lack of population specificity: Most existing biomarkers are based on mixed genotype populations, and their diagnostic cutoff values ​​may not be applicable to all genetic background subgroups, leading to unstable application efficacy in different populations.

[0005] Existing technologies have the following key, under-recognized limitations when applied to the specific population of APOE ε3 / ε3 genotype carriers (who constitute the majority of the global population) for diagnosis: (1) Diagnostic "blind spots" and insufficient sensitivity Currently, mainstream AD biomarkers (such as Aβ and Tau-related proteins) show the most significant changes in APOE ε4 carriers, leading to an implicit bias in clinical research and diagnostic standards towards this population. Conversely, in patients with the APOE ε3 / ε3 genotype, the concentration patterns of these universal biomarkers may be atypical, smaller in magnitude, or appear later, resulting in a significant decrease in sensitivity and an increase in missed diagnosis rates when using existing universal standards for diagnosis. This makes APOE ε3 / ε3 individuals, who constitute a considerable proportion of AD patients, a "blind spot" or "grey area" for early diagnosis.

[0006] (2) The heterogeneity of pathological mechanisms has not been considered. The pathogenesis and progression of AD in the APOE ε3 / ε3 genotype may be independent of or different from the typical APOE ε4-driven amyloid deposition pathway. Current technologies (such as Aβ-targeted PET or detection) fail to cover or reflect the pathophysiological processes specific to the APOE ε3 / ε3 population (such as synaptic dysfunction, lipid metabolism disorders, and specific neuroinflammatory pathways). Therefore, relying on existing biomarkers cannot accurately capture the disease essence in this population, resulting in low biological specificity in diagnosis.

[0007] (3) Lack of dedicated diagnostic tools and standards Currently, there are no rigorously validated AD diagnostic kits or biomarker combinations specifically designed for the APOE ε3 / ε3 population on the market or in existing patents. Directly applying general standards is essentially "forcing a square peg into a round hole," and cannot achieve accurate and early identification of this population. Summary of the Invention

[0008] To address the aforementioned problems in the prior art, the present invention provides a combination of biomarkers for Alzheimer's disease in individuals with the APOE ε3 / ε3 genotype.

[0009] To achieve the above objectives, the present invention provides the application of a combination of biomarkers in the preparation of a detection kit for Alzheimer's disease in individuals with the APOE ε3 / ε3 genotype. The main feature is that the biomarkers are selected from at least one of ANGPTL1, PON2, PTN, SCGB3A1, PBLD, MORC3, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, GSTA3, EIF2AK3, CTSD, and FOLH1.

[0010] The specific abbreviations are as follows: ANGPTL1 (Angiopoietin-like 1): Angiopoietin-like protein 1; PON2 (Paraoxonase 2): Paraoxonase 2; PTN (Pleiotrophin): a multifunctional growth factor; SCGB3A1 (Secretoglobin family 3A member 1): Member 1 of the secretoglobin family 3A; PBLD (Phenazine biosynthesis-like protein domain-containing protein): A phenazine biosynthesis-like protein domain-containing protein. MORC3 (MORC family CW-type zinc finger protein 3): MORC family CW-type zinc finger protein 3; EIF5A (Eukaryotic translation initiation factor 5A): Eukaryotic translation initiation factor 5A; FAM3D (Family with sequence similarity 3 member D): Sequence similarity family 3 member D; PIKFYVE (Phosphoinositide kinase, FYVE-type zinc finger containing): Phosphoinositide kinase, containing FYVE-type zinc finger; CKB (Creatine kinase B-type): Creatine kinase type B; HGF (Hepatocyte growth factor): Hepatocyte growth factor; TREH (Trehalase): Trehalase; GSTA3 (Glutathione S-transferase A3): Glutathione S-transferase A3; EIF2AK3 (Eukaryotic translation initiation factor 2-alpha kinase 3): Eukaryotic translation initiation factor 2α kinase 3; CTSD (Cathepsin D): Cathepsin D; FOLH1 (Folate hydrolase 1): Folic acid hydrolase 1; Preferably, the biomarker is selected from at least two of ANGPTL1, PON2, PTN, SCGB3A1, PBLD, MORC3, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, GSTA3, EIF2AK3, CTSD, and FOLH1.

[0011] Preferably, the biomarker is a combination of PTN, PBLD, FAM3D, CKB, HGF, GSTA3, EIF2AK3, and FOLH1; The biomarkers mentioned are a combination of PON2, PTN, PBLD, FAM3D, CKB, HGF, GSTA3, EIF2AK3, and FOLH1; The biomarkers mentioned are a combination of ANGPTL1, PTN, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, and FOLH1; The biomarkers mentioned are a combination of ANGPTL1, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, HGF, TREH, and FOLH1; The biomarkers mentioned are a combination of ANGPTL1, PON2, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, CKB, HGF, and TREH; The biomarkers mentioned are a combination of ANGPTL1, PON2, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, HGF, TREH, GSTA3, and FOLH1; The biomarkers mentioned are a combination of ANGPTL1, SCGB3A1, PBLD, EIF5A, FAM3D, and HGF; The biomarkers mentioned are a combination of ANGPTL1, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, GSTA3, and FOLH1; The biomarkers mentioned are combinations of ANGPTL1, PON2, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, GSTA3, and FOLH1; or, The biomarkers mentioned are a combination of ANGPTL1, PTN, SCGB3A1, PBLD, FAM3D, CKB, HGF, GSTA3, EIF2AK3, and FOLH1.

[0012] Ideally, the test sample should be plasma.

[0013] Preferably, the kit further includes a genotyping reagent for determining APOE genotypes as ε3 / ε3 homozygous.

[0014] Preferably, Alzheimer's disease in individuals with the APOE ε3 / ε3 genotype is diagnosed by comparing the levels of detected biomarker combinations with preset cutoff values.

[0015] Preferably, the reagent kit includes calibrators and quality control samples, wherein the quality control samples include: APOE ε3 / ε3 genotype individuals with positive matrix quality control for Alzheimer's disease; APOE ε3 / ε3 genotype individuals with Alzheimer's disease negative matrix quality control.

[0016] The biomarker combination provided by this invention for Alzheimer's disease in individuals with the APOE ε3 / ε3 genotype enables high-precision and high-specificity in vitro diagnosis of Alzheimer's disease in a specific population (APOE ε3 / ε3) through proteomics discovery. Attached Figure Description

[0017] Figure 1 This image shows the results of population-specific differentially expressed proteins for APOE ε3 / ε3 genotypes. Figure 1 A represents the APOE ε3 / ε3 genotype differential protein volcano plot; Figure 1 B is a volcano plot of differentially expressed proteins in APOE3 / 4 genotype populations; Figure 1 C represents the differential protein volcano plot of the APOE ε4 / 4 genotype population; Figure 1 D. Statistical bar chart of differentially expressed proteins in three genotype populations; Figure 1 E is the upset plot of differentially expressed proteins in the three genotype populations. Sixteen unique differentially expressed proteins were screened out that appeared only in the APOE ε3 / ε3 genotype: ANGPTL1, PON2, PTN, SCGB3A1, PBLD, MORC3, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, GSTA3, EIF2AK3, CTSD, and FOLH1.

[0018] Figure 2 To create a GO biological process enrichment map for the 16 APOE ε3 / ε3 genotype-specific differentially expressed proteins obtained through screening.

[0019] Figure 3 The NPX expression values ​​of 16 APOE ε3 / ε3 genotype-specific differentially expressed proteins were obtained through screening.

[0020] Figure 4 The graph shows the AUC comparison results for the top ten protein combinations. Among them, Clinical only: only clinical variables (gender, age, BMI, and years of education); Protein only: only protein combination; Protein+clinical: clinical variables combined with protein combination.

[0021] Figure 5 The reagent kit provided for this invention. Detailed Implementation

[0022] To more clearly describe the technical content of the present invention, the following description is provided in conjunction with specific embodiments.

[0023] Unless otherwise specified, the experimental methods used in the following examples are conventional methods; unless otherwise specified, the reagents and materials used in the following examples are commercially available.

[0024] The combination of protein biomarkers provided by this invention, including at least two of ANGPTL1, PON2, PTN, SCGB3A1, PBLD, MORC3, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, GSTA3, EIF2AK3, CTSD, and FOLH1, has excellent diagnostic ability for Alzheimer's disease in the specific population with the APOE ε3 / ε3 genotype.

[0025] like Figure 5 As shown, based on the above-described combination of biomarkers, the present invention provides a kit for in vitro detection. This kit contains reagents for the specific detection of the at least two protein biomarkers, specifically including: Specific detection reagents: For each selected protein biomarker (such as ANGPTL1, PON2, etc.), the kit provides corresponding specific capture and detection reagents. These typically consist of validated monoclonal antibody pairs, including a capture antibody coated on a solid-phase support and a detection antibody labeled with an enzyme or fluorescent marker.

[0026] Solid-phase carriers: such as 96-well microplates or suspended magnetic beads.

[0027] Detection system reagents include signal generation and amplification reagents, such as horseradish peroxidase (HRP)-labeled streptavidin, corresponding chemiluminescent or fluorescent substrates.

[0028] Reference materials and quality control materials: Calibrator: A solution containing a known concentration of the recombinant target protein, used to plot a standard curve.

[0029] Quality control products: APOE ε3 / ε3 AD positive matrix quality control: A mixed matrix containing moderate levels of the target protein, mimicking the patient's condition. APOE ε3 / ε3 HC negative matrix quality control: A healthy control (HC) matrix confirming the absence of the target protein or extremely low levels. Specific steps: Donor screening and raw material preparation: Fasting venous blood was collected from clinically diagnosed AD patients with the APOE ε3 / ε3 genotype and HC individuals. Plasma was separated according to matrix requirements, then inactivated at 56℃ and centrifuged to remove precipitates, yielding AD-positive and HC-negative matrix control raw material solutions. The concentration of target protein markers (such as ANGPTL1, PON2, etc.) was determined using ELISA. AD-positive raw material solutions were diluted or combined with HC-negative matrix to adjust to the expected levels, while HC-negative raw material solutions were ensured to have marker content below the detection limit. After adding BSA, protease inhibitors, preservatives, and cryoprotectants, the mixture was aliquoted and stored at -20℃ or -70℃.

[0030] Quality control product performance verification and standardized management: Perform uniformity testing, stability monitoring (long-term, short-term, and freeze-thaw tests) and consistency verification between three consecutive batches on the repackaged quality control products. Set the acceptable range at ±2SD or ±15%~20% of the mean of at least 20 independent measurements. Label the products with the name, batch number, expiration date, and storage conditions.

[0031] Instructions for use: This kit is only applicable to or specifically optimized for individuals with the APOE ε3 / ε3 genotype. Clearly specify the source of quality control materials, usage procedures, and diagnostic criteria. Provide specific diagnostic cutoff values ​​based on data from this population.

[0032] Specifically, this kit is an enzyme-linked immunosorbent assay (ELISA) kit based on a double-antibody sandwich method, and its specific components are shown in Table 1 below: The method for detection using the above-mentioned kit includes the following steps: Sample and Genotype Confirmation: Obtain a peripheral blood plasma sample from the individual to be tested. The individual's APOE genotype must be confirmed as ε3 / ε3 homozygous beforehand or concurrently using a commercial genotyping kit. This step is essential to ensure diagnostic accuracy.

[0033] Parallel or sequential detection: Add samples to the kit for parallel detection of at least two selected protein biomarkers (e.g., ANGPTL1 and PON2). A typical workflow is: sample loading and incubation (antigen-antibody binding) - washing - addition of labeled detection antibody for secondary incubation and washing - addition of substrate for color development / luminescence. A typical three-step ELISA workflow can be completed in approximately 3 hours. Sample loading and incubation: Add 100 μL of the test sample (must be from an APOE ε3 / ε3 individual) or calibrator / control sample to the corresponding antibody-coated wells. Incubate at 37°C for 60 minutes. Wash 5 times. Add detection antibody: Add 100 μL of a mixture of biotinylated detection antibody working solution to each well. Incubate at 37°C for 45 minutes. Wash 5 times. Add enzyme conjugate and substrate: Add 100 μL of streptavidin-HRP working solution to each well. Incubate at 37°C for 30 minutes. Wash 5 times. Add equal volumes (100 μL) of substrate solutions A and B to each well sequentially, and incubate at room temperature in the dark for 10 minutes. Detection and calculation: Add 50 μL of stop solution to each well, and immediately read the relative luminescence unit (RLU) value of each well on a chemiluminescence microplate reader. Plot the calibrator concentration on the x-axis and the RLU value on the y-axis, fit a four-parameter logistic curve, and calculate the concentrations of the three proteins in each sample.

[0034] Signal reading and concentration conversion: Use the corresponding instruments to read the signal values ​​of each detection well, and calculate the absolute or relative concentration of each target protein in the sample according to their respective standard curves.

[0035] A dedicated model was used for diagnostic interpretation: the concentration values ​​of each protein were substituted into a diagnostic mathematical model specifically developed for the APOE ε3 / ε3 population. This model was trained by the applicant based on data from the initial 86 samples; detailed steps are described in the model construction section.

[0036] For the kit of the present invention, the detection method is not limited to ELISA, but can be replaced by chemiluminescent immunoassay, fluorescent immunochromatographic test strips or a multiplex detection platform based on liquid phase chip (Luminex); the solid phase carrier is not limited to microplates, but can be magnetic beads, nanospheres or biochips; the signal marker is not limited to HRP-chemiluminescence system, but can be alkaline phosphatase, luciferin, time-resolved fluorescence, etc.

[0037] Example 1 Sample cohorts and proteomic data acquisition Sample source and grouping: Peripheral blood plasma samples were collected from 86 APOE ε3 / ε3 individuals, including 47 patients with clinically diagnosed Alzheimer's disease (AD group) and 39 healthy controls (HC group).

[0038] Peripheral blood plasma samples were collected from 87 individuals with APOE ε3 / ε4, including 45 cases in the AD group and 42 cases in the HC group.

[0039] Peripheral blood plasma samples were collected from 143 APOE ε4 / ε4 homozygous individuals, including 71 cases in the AD group and 72 cases in the HC group.

[0040] The APOE genotype of all samples was confirmed by real-time quantitative PCR: First, peripheral blood genomic DNA was extracted from the subjects, and universal primers and allele-specific probes were designed for the two SNP sites of the APOE gene, rs429358 (388T>C) and rs7412 (526C>T). (For the rs429358 site, FAM-labeled wild-type T probe and VIC-labeled mutant C probe were used; for the rs7412 site, FAM-labeled wild-type C probe and VIC-labeled mutant T probe were used.) The DNA template to be tested was prepared into a 25 μL reaction system with each primer, probe, and PCR premix. The following program was run on the real-time quantitative PCR instrument: 95℃ pre-denaturation for 10 min, followed by 40 cycles of 95℃ denaturation for 15 s and 60℃ annealing / extension for 60 s (with fluorescence signal collected). After amplification, the genotype was determined based on the amplification curve.

[0041] Proteomics analysis: Olink Explore protein assays, utilizing Proximity Extension Assay (PEA) technology and the Illumina NGS sequencing platform, performed proteomics analysis on all samples. Relative quantitative data for 3072 proteins in each sample were obtained.

[0042] Differential protein screening and biomarker combination determination Data preprocessing and statistical analysis: Quantitative data were standardized and normalized. Then, unpaired t-tests (or Mann-Whitney U tests) were used to compare the expression levels of each protein between the AD and HC groups. A significance threshold of p < 0.05 was set, and the Benjamini-Hochberg method was used for multiple test correction to control the false detection rate (FDR < 0.05).

[0043] Filtering results By comparing the proteomic data of the three genotypes APOE ε3 / ε3, APOE ε3 / ε4, and APOE ε4 / ε4, proteins showing significant differences (FDR<0.05) in the APOE ε3 / ε4 and APOE ε4 / ε4 genotypes were removed, and differentially expressed proteins were retained only in the APOE ε3 / ε3 genotype.

[0044] After the above exclusionary screening, such as Figure 1As shown, a total of 16 unique differentially expressed proteins were obtained that appeared only in the APOE ε3 / ε3 genotype: ANGPTL1, PON2, PTN, SCGB3A1, PBLD, MORC3, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, GSTA3, EIF2AK3, CTSD, and FOLH1. These proteins showed no significant differential expression or opposite expression trends in both the APOE ε3 / ε4 and APOE ε4 / ε4 genotypes, and therefore can serve as APOE ε3 / ε3 genotype-specific protein markers.

[0045] The NPX expression values ​​of 16 APOE ε3 / ε3 genotype-specific differentially expressed proteins—ANGPTL1, PON2, PTN, SCGB3A1, PBLD, MORC3, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, GSTA3, EIF2AK3, CTSD, and FOLH1—are as follows: Figure 3 As shown.

[0046] GO biological process enrichment analysis was performed on the 16 APOE ε3 / ε3 genotype-specific differentially expressed proteins obtained from screening. The results are as follows: Figure 2 As shown, these proteins are significantly enriched in multiple pathways. Specifically: Peptidyl-serine phosphorylation: Gene ratio 0.125, enrichment fraction 1.65; Myelination: Gene ratio 0.150, enrichment fraction 1.63; Cellular modified amino acid metabolic process: Gene ratio 0.175, enrichment fraction 1.61; Exogenous peptide antigen processing and presentation: Gene ratio 0.200, enrichment fraction 1.59.

[0047] The above results indicate that the 16 differentially expressed proteins specific to the APOE ε3 / ε3 genotype are mainly involved in biological processes such as protein phosphorylation modification, myelination, amino acid-derived metabolism, and immune presentation of exogenous antigens, suggesting that these pathways may be closely related to their genotype-specific functional regulation.

[0048] Model building Based on the 16 unique differentially expressed proteins of the APOE ε3 / ε3 genotype obtained above, a systematic evaluation was conducted on all possible non-empty protein combinations (a total of 2^16−1 = 65,535). Using the ability to distinguish the target population as the indicator, an L2 regularized logistic regression model was used for each protein combination, with the outcome variable being disease status: HC=0, AD=1. The model outputs the predicted probability of AD, and an ROC curve and AUC are calculated based on this predicted probability.

[0049] To evaluate the model's generalization ability, stratified 5-fold cross-validation was performed (random seed = 20260624), and the specific process is as follows: The 86 APOE ε3 / ε3 samples were stratified according to HC / AD status; While maintaining a relatively consistent HC / AD ratio, the sample was divided into 5 folds; Four folds are used as the training set each time, and the remaining 1 fold is used as the test set; Fit an L2-regularized logistic regression model to the training set; Use the trained model to predict the AD probability of test set samples; Repeat 5 times to ensure that each sample receives a predicted probability once; Summarize all predicted probabilities and calculate the overall cross-validation ROC curve and AUC.

[0050] To avoid data leakage, all data preprocessing steps were performed only on the training set. Continuous variables, including protein NPX values, age, BMI, and years of education, had their means and standard deviations calculated within each training fold, and were standardized using the training set parameters for the corresponding test fold. Information from the test fold was not used in the standardized parameter estimation. For each training fold, the model was fitted on the training set, and then the predicted probability of AD was output for the test set. The predicted probabilities of all test folds were combined to calculate the ROC curve and AUC for each repeated cross-validation. The top 100 AUC combinations are shown in Table 2 below.

[0051] Specifically, the top ten protein combinations with the highest AUC were selected (referred to as combination 1 (PTN, PBLD, FAM3D, CKB, HGF, GSTA3, EIF2AK3, FOLH1), combination 2 (PON2, PTN, PBLD, FAM3D, CKB, HGF, GSTA3, EIF2AK3, FOLH1), combination 3 (ANGPTL1, PTN, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, FOLH1), combination 4 (ANGPTL1, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, HGF, TREH, FOLH1), combination 5 (ANGPTL1, PON2, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH), and combination 6 (PON2, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH). 6 (ANGPTL1, PON2, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, HGF, TREH, GSTA3, FOLH1), combination 7 (ANGPTL1 , SCGB3A1, PBLD, EIF5A, FAM3D, HGF), combination 8 (ANGPTL1, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, CKB, H GF, TREH, GSTA3, FOLH1), combination 9 (ANGPTL1, PON2, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TR EH, GSTA3, FOLH1), combination 10 (ANGPTL1, PTN, SCGB3A1, PBLD, FAM3D, CKB, HGF, GSTA3, EIF2AK3, FOLH1).

[0052] For these ten protein combinations, the above modeling method was used to compare the following three cases: (1) Only clinical variables were included (gender was included in the model as a binary variable. Age, BMI and years of education were included in the model as continuous variables); (2) Only the above ten protein combinations were used (i.e., ten individual protein combinations); (3) Protein combinations were combined with clinical variables (i.e., protein combination + clinical variables).

[0053] The specific process of the above analysis is as follows: 1. Definition of outcome variables: This analysis was conducted only in the APOE ε3 / ε3 population to compare HC and AD.

[0054] Let the disease state of the i-th sample be: The model aims to predict, based on protein NPX values ​​and / or clinical variables: That is, the probability that the sample belongs to AD.

[0055] 2. Predictor variable matrix For the i-th sample, let its predictor variable vector be: Where p is the number of predictor variables in the model.

[0056] 2.1 Individual Models for Clinical Variables: 2.2 Protein Combination Individual Model: 2.3 Protein Combination + Clinical Variable Model: 3. L2 Regularized Logistic Regression Model For the i-th sample, the model is in the form of: Equivalent land: in, This represents protein NPX values ​​and / or clinical variables. The intercept is... is the regression coefficient.

[0057] The model parameters are obtained by minimizing the negative log-likelihood function with an L2 penalty term: in, The regularization strength parameter is set as follows in this analysis: Intercept term Not subject to punishment.

[0058] In 5-fold cross-validation, the model is fitted only in the training fold and used to predict the probability that each sample in the corresponding test fold belongs to AD. The predicted probabilities of all test folds are combined and used to plot the ROC curve and calculate the cross-validation AUC.

[0059] like Figure 4As shown, the cross-validation AUC of the top ten protein individual models ranged from 0.950 to 0.953, significantly higher than the 0.678 of the clinical variable individual models, indicating that the differentially expressed protein combinations selected in this invention possess excellent discriminative ability. Both the protein combination model and the protein combination combined with clinical variable model significantly outperformed the clinical variable individual models. Compared to the protein combination individual models, adding gender, age, BMI, and years of education did not further significantly improve the AUC. These results demonstrate that the optimal protein combination selected from the 16 differentially expressed proteins can achieve high-performance discrimination of APOE ε3 / ε3 genotype-related phenotypes, and has potential application value.

[0060] In practical clinical applications, the diagnostic process of this invention can be two-stage: The first phase involves rapid, low-cost APOE genotyping screening for all patients.

[0061] In the second phase, only individuals identified as having APOE ε3 / ε3 underwent blood protein biomarker detection and risk assessment using the specialized kit of this invention. This stratified diagnostic strategy is the preferred approach for achieving precision medicine and maximizing cost-effectiveness.

[0062] In this specification, the invention has been described with reference to specific embodiments thereof. However, it will be apparent that various modifications and variations can be made without departing from the spirit and scope of the invention. Therefore, the specification and drawings should be considered illustrative rather than restrictive.

Claims

1. The application of a combination of biomarkers in the preparation of a diagnostic kit for Alzheimer's disease in individuals with the APOE ε3 / ε3 genotype, characterized in that, The biomarkers are selected from at least one of ANGPTL1, PON2, PTN, SCGB3A1, PBLD, MORC3, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, GSTA3, EIF2AK3, CTSD, and FOLH1.

2. The application according to claim 1, characterized in that, The biomarkers are selected from at least two of ANGPTL1, PON2, PTN, SCGB3A1, PBLD, MORC3, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, GSTA3, EIF2AK3, CTSD, and FOLH1.

3. The application according to claim 1, characterized in that, The biomarkers mentioned are a combination of PTN, PBLD, FAM3D, CKB, HGF, GSTA3, EIF2AK3, and FOLH1; The biomarkers mentioned are a combination of PON2, PTN, PBLD, FAM3D, CKB, HGF, GSTA3, EIF2AK3, and FOLH1; The biomarkers mentioned are a combination of ANGPTL1, PTN, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, and FOLH1; The biomarkers mentioned are a combination of ANGPTL1, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, HGF, TREH, and FOLH1; The biomarkers mentioned are a combination of ANGPTL1, PON2, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, CKB, HGF, and TREH; The biomarkers mentioned are a combination of ANGPTL1, PON2, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, HGF, TREH, GSTA3, and FOLH1; The biomarkers mentioned are a combination of ANGPTL1, SCGB3A1, PBLD, EIF5A, FAM3D, and HGF; The biomarkers mentioned are a combination of ANGPTL1, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, GSTA3, and FOLH1; The biomarkers mentioned are combinations of ANGPTL1, PON2, SCGB3A1, PBLD, EIF5A, FAM3D, PIKFYVE, CKB, HGF, TREH, GSTA3, and FOLH1; or, The biomarkers mentioned are a combination of ANGPTL1, PTN, SCGB3A1, PBLD, FAM3D, CKB, HGF, GSTA3, EIF2AK3, and FOLH1.

4. The application according to claim 1, characterized in that, The sample tested was plasma.

5. The application according to claim 1, characterized in that, The kit also includes a genotyping reagent for determining APOE genotypes that are homozygous for ε3 / ε3.

6. The application according to claim 1, characterized in that, Alzheimer's disease in individuals with the APOE ε3 / ε3 genotype can be diagnosed by comparing the levels of detected biomarker combinations with preset cutoff values.

7. The application according to claim 1, characterized in that, The kit includes calibrators and quality control samples, wherein the quality control samples include: APOE ε3 / ε3 genotype individuals with positive matrix quality control for Alzheimer's disease; APOE ε3 / ε3 genotype individuals with Alzheimer's disease negative matrix quality control.

Images