Metabolic marker for diagnosing Alzheimer's disease and mild cognitive impairment and application thereof

Abstract

The invention provides a metabolic marker for diagnosing Alzheimer's disease and mild cognitive impairment and application of the metabolic marker, and belongs to the technical field of disease diagnosis markers. According to the method, metabolic markers are obtained through metabolite difference analysis and screening between samples of subjects with normal cognition and other types of cognitive impairment diseases and samples of subjects with Alzheimer's disease and mild cognitive impairment diseases, and modeling and verification are carried out through a machine learning algorithm; a high-sensitivity and high-specificity Alzheimer's disease and mild cognitive impairment diagnosis model is constructed, and the method has important significance in early screening, early diagnosis and early intervention of Alzheimer's disease.

Description

Technical Field

[0001] This invention belongs to the field of disease diagnostic biomarker technology, specifically relating to a metabolic biomarker for diagnosing Alzheimer's disease and mild cognitive impairment and its application. Background Technology

[0002] Alzheimer's disease (AD) is a common age-related neurodegenerative disease that can lead to the loss of daily living abilities in older adults. AD has an insidious onset and progresses slowly. Memory and cognitive impairment, as well as loss of language and behavioral abilities, are its main clinical manifestations. As the disease worsens, it is often accompanied by severe amnesia and motor dysfunction, ultimately leading to death. Most AD patients are diagnosed at an irreversible stage of dementia. Therefore, early diagnosis is crucial for the treatment of AD.

[0003] Currently, diagnostic methods for Alzheimer's disease (AD) have some limitations. Because clinical symptoms of AD typically appear only in later stages of disease progression, accurate diagnosis before symptoms manifest is difficult. Furthermore, diagnosis relies heavily on clinical assessment and lacks objective biomarkers as diagnostic criteria. Additionally, lesions on imaging are often only revealed when the disease is relatively severe, hindering early diagnosis. Moreover, the current lack of AD-specific biomarkers makes accurate diagnosis in the early stages of the disease challenging. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a metabolic biomarker for diagnosing Alzheimer's disease and mild cognitive impairment. The metabolic biomarker can accurately distinguish between cognitively normal individuals and other types of cognitive impairment diseases, as well as Alzheimer's disease and mild cognitive impairment, providing an important basis for early screening, early diagnosis and early intervention of Alzheimer's disease.

[0005] This invention provides a metabolic biomarker comprising at least five metabolites: ribose, asparagine, 2-hydroxyvalerate, glucose, ascorbic acid, 6-aminocaproic acid, nicotinamide, phosphatidylinositol 38:4, phosphatidylcholine 40:6, phosphatidylethanolamine 38:5, phosphatidylglycerol 34:1, ceramide t41:0 (t16:0 / 25:0), hexosylceramide d41:1 (d18:1 / 23:0), triglycerides 55:5, and triglycerides 57:3.

[0006] Preferred metabolites include the following: ribose, glucose, 6-aminocaproic acid, nicotinamide, and phosphatidylinositol 38:4.

[0007] Preferred metabolites include the following: ribose, glucose, 6-aminocaproic acid, nicotinamide, phosphatidylinositol 38:4, phosphatidylethanolamine 38:5, phosphatidylglycerol 34:1 and hexosylceramide d41:1 (d18:1 / 23:0).

[0008] Preferred metabolites include the following: ribose, asparagine, glucose, 6-aminocaproic acid, nicotinamide, phosphatidylinositol 38:4, phosphatidylethanolamine 38:5, phosphatidylglycerol 34:1, ceramide t41:0 (t16:0 / 25:0), hexosylceramide d41:1 (d18:1 / 23:0), and triglycerides 55:5.

[0009] This invention provides an application of the aforementioned metabolic biomarkers in constructing predictive models for Alzheimer's disease and mild cognitive impairment.

[0010] Preferably, the prediction model is modeled using machine learning support vector machines.

[0011] This invention provides an Alzheimer's disease and mild cognitive impairment prediction system, comprising modules with the following connection states;

[0012] The data acquisition module is used to acquire the detection data of the metabolic biomarkers described in the above technical solution in the sample to be tested;

[0013] The data analysis module is used to analyze the metabolic biomarker detection data collected by the data acquisition module in the diagnostic model constructed in the application described in the above technical solution to obtain prediction results;

[0014] A data output module is used to output the prediction results obtained by the data analysis module to the terminal for display.

[0015] Preferably, the method for analyzing and predicting using the diagnostic model involves taking the measurement results of the metabolic markers in the sample to be tested as input data and inputting them into the diagnostic model. When the output result is ≥ a threshold, it is judged as Alzheimer's disease and mild cognitive impairment; when the output result is < a threshold, it is judged as cognitively normal and other types of cognitive impairment diseases.

[0016] The threshold is 0.4223.

[0017] The present invention provides an Alzheimer's disease and mild cognitive impairment prediction device, which is equipped with the Alzheimer's disease and mild cognitive impairment prediction system described in the above technical solution.

[0018] This invention provides the application of a reagent for detecting the aforementioned metabolic markers in the preparation of a kit for differentiating between cognitively normal individuals and other types of cognitive impairment diseases, as well as Alzheimer's disease and mild cognitive impairment.

[0019] This invention provides a metabolic biomarker comprising at least five metabolites: ribose, asparagine, 2-hydroxyvalerate, glucose, ascorbic acid, 6-aminocaproic acid, nicotinamide, phosphatidylinositol 38:4, phosphatidylcholine 40:6, phosphatidylethanolamine 38:5, phosphatidylglycerol 34:1, ceramide t41:0 (t16:0 / 25:0), hexosylceramide d41:1 (d18:1 / 23:0), triglycerides 55:5, and triglycerides 57:3. This invention, based on the screening results of the aforementioned metabolic biomarkers, uses machine learning algorithms to model and distinguish between cognitively normal individuals and other types of cognitive impairment diseases with Alzheimer's disease and mild cognitive impairment. Results show that the constructed predictive model has an AUC value above 0.8, a sensitivity above 0.75, and a specificity above 0.739. This indicates that the aforementioned metabolic biomarkers can accurately distinguish between cognitively normal individuals and other types of cognitive impairment diseases with Alzheimer's disease and mild cognitive impairment, thus providing a reliable basis for early clinical diagnosis of Alzheimer's disease and offering a feasible new strategy for early diagnosis. Furthermore, the detection of these metabolic biomarkers is entirely non-invasive, with simple and convenient sample collection and operation, making it particularly suitable for large-scale Alzheimer's disease risk screening, especially valuable in remote areas with limited medical resources. Attached Figure Description

[0020] Figure 1 The results of multivariate ROC curve analysis for 15 important biomarkers that distinguish HC+SCD+VCI vs AD+MCI in the modeling group;

[0021] Figure 2 To validate the results of multivariate ROC curve analysis of 15 important biomarkers that distinguish HC+SCD+VCI vs AD+MCI in the group. Detailed Implementation

[0022] This invention provides a metabolic biomarker comprising at least five metabolites: ribose, asparagine, 2-hydroxyvalerate, glucose, ascorbic acid, 6-aminocaproic acid, nicotinamide, phosphatidylinositol 38:4, phosphatidylcholine 40:6, phosphatidylethanolamine 38:5, phosphatidylglycerol 34:1, ceramide t41:0 (t16:0 / 25:0), hexosylceramide d41:1 (d18:1 / 23:0), triglycerides 55:5, and triglycerides 57:3.

[0023] In this invention, the metabolic biomarkers are obtained by screening through metabolite difference analysis between samples of subjects with normal cognition and other types of cognitive impairment and samples of subjects with Alzheimer's disease and mild cognitive impairment. The biomarkers are modeled and verified through machine learning algorithms to show that they can accurately distinguish between subjects with normal cognition and other types of cognitive impairment and subjects with Alzheimer's disease and mild cognitive impairment, thereby achieving early diagnosis of Alzheimer's disease and mild cognitive impairment.

[0024] In this invention, the metabolic markers include technical solutions containing 6, 7, 8, 9, 10, 11, 12, 13, 14, and 15 of the aforementioned metabolites. This invention does not impose any particular limitation on the number or types of metabolites; any composition of metabolites is acceptable.

[0025] In this invention, the metabolic markers preferably include the following metabolites: ribose, glucose, 6-aminocaproic acid, nicotinamide, and phosphatidylinositol 38:4. The metabolic markers preferably include the following metabolites: ribose, glucose, 6-aminocaproic acid, nicotinamide, phosphatidylinositol 38:4, phosphatidylethanolamine 38:5, phosphatidylglycerol 34:1, and hexosylceramide d41:1 (d18:1 / 23:0). The metabolic markers preferably include the following metabolites: ribose, asparagine, glucose, 6-aminocaproic acid, nicotinamide, phosphatidylinositol 38:4, phosphatidylethanolamine 38:5, phosphatidylglycerol 34:1, ceramide t41:0 (t16:0 / 25:0), hexosylceramide d41:1 (d18:1 / 23:0), and triglycerides 55:5.

[0026] In this embodiment of the invention, models were built based on 5, 8, 11, and 15 metabolites, respectively, to verify the impact of different numbers of metabolites on the accuracy of diagnosing Alzheimer's disease and mild cognitive impairment. The results showed that the prediction models constructed by the above four schemes all had AUC values ​​above 0.8, sensitivity above 0.75, and specificity above 0.739, indicating that modeling with 5 or more metabolites has high predictive accuracy. At the same time, the results of the embodiments showed that with the increase of the number of metabolic biomarkers, the AUC, sensitivity, and specificity of the constructed prediction models were improved to a certain extent.

[0027] This invention provides an application of the aforementioned metabolic biomarkers in constructing predictive models for Alzheimer's disease and mild cognitive impairment.

[0028] In this invention, the prediction model is preferably modeled using a machine learning support vector machine. During modeling, the machine learning support vector machine (SVM) is preferably used for 1000 random loop iterations.

[0029] In this invention, when constructing the prediction model, normal cognition and other types of cognitive impairment diseases are used as control groups, and Alzheimer's disease and mild cognitive impairment are used as disease groups. Three-quarters of the sample size of the control group and the disease group are used as the training set, and the remaining one-quarter of the sample size is used as the test set to verify the constructed prediction model.

[0030] This invention provides an Alzheimer's disease and mild cognitive impairment prediction system, comprising modules with the following connection states;

[0031] The data acquisition module is used to acquire the detection data of the metabolic biomarkers described in the above technical solution in the sample to be tested;

[0032] The data analysis module is used to analyze the metabolic biomarker detection data collected by the data acquisition module in the diagnostic model constructed in the application described in the above technical solution to obtain prediction results;

[0033] A data output module is used to output the prediction results obtained by the data analysis module to the terminal for display.

[0034] This invention does not impose any special restrictions on the connection state; any module connection method known in the art can be used, such as electrical connection.

[0035] In this invention, the method for analyzing and predicting using the diagnostic model preferably uses the measurement results of the metabolic markers in the sample to be tested as input data into the diagnostic model. When the output result is ≥ a threshold, it is judged as Alzheimer's disease and mild cognitive impairment; when the output result is < a threshold, it is judged as normal cognition and other types of cognitive impairment diseases; the threshold is 0.4223.

[0036] The present invention provides an Alzheimer's disease and mild cognitive impairment prediction device, which is equipped with the Alzheimer's disease and mild cognitive impairment prediction system described in the above technical solution.

[0037] In this invention, the Alzheimer's disease and mild cognitive impairment prediction device preferably further includes a metabolite detection device and / or a terminal storage or display device.

[0038] This invention provides the application of a reagent for detecting the aforementioned metabolic markers in the preparation of a kit for differentiating between cognitively normal individuals and other types of cognitive impairment diseases, as well as Alzheimer's disease and mild cognitive impairment.

[0039] In this invention, the reagents include metabolite extraction reagents and / or detection reagents. The extraction reagents preferably include at least one of the following solutions: a mixture of methyl tert-butyl ether and methanol, an aqueous methanol solution, an isopropanol-acetonitrile solution, or ice-cold methanol. The volume ratio of methyl tert-butyl ether to methanol in the methyl tert-butyl ether and methanol mixture is preferably 3:1. The volume ratio of acetonitrile to isopropanol in the isopropanol-acetonitrile solution is preferably 3:1. The volume ratio of methanol to water in the aqueous methanol solution is 3:1. The detection reagents preferably include an aqueous solution of 0.1% acetic acid (volume percentage) and 0.1% ammonium acetate (mass percentage), an acetonitrile-isopropanol mixture containing 0.1% acetic acid (volume percentage) and 0.1% ammonium acetate (mass percentage), an aqueous solution containing 0.1% formic acid (volume percentage), and an acetonitrile solution containing 0.1% formic acid (volume percentage).

[0040] The following detailed description, in conjunction with embodiments, illustrates a metabolic biomarker for diagnosing Alzheimer's disease and mild cognitive impairment, and its application thereof. However, these descriptions should not be construed as limiting the scope of protection of this invention.

[0041] Example 1

[0042] A method for screening metabolic biomarkers for predicting Alzheimer's disease and mild cognitive impairment.

[0043] 1. Subject Information

[0044] 1) Inclusion criteria:

[0045] Participants must meet all of the following inclusion criteria to be eligible to participate in this experiment:

[0046] (1) Males or females aged 18 years or older;

[0047] (2) Read and fully understand the information, sign the informed consent form, and be able to provide a blood sample for metabolomics testing;

[0048] (3) The diagnosis of cognitive impairment meets the corresponding clinical diagnostic criteria, as follows:

[0049] a. AD patients meet the 2011 National Institute on Aging and Alzheimer's Association (NIA-AA) diagnostic criteria for "probable AD dementia";

[0050] b. Patients with mild cognitive impairment (MCI) must meet the 2004 Petersen diagnostic criteria for amnesic MCI;

[0051] c. Patients with Subjective Cognitive Decline (SCD) must meet the SCD concept proposed by the Subjective Cognitive Decline Initiative (SCD-I) in 2014 and meet the SCD characteristics supplemented by SCD-I in 2020.

[0052] d. Patients with vascular cognitive impairment (VCI) must meet the diagnostic criteria for vascular cognitive impairment published by the Vascular Behavioral and Cognitive Disorders International Association (Vas-Cog) in 2014;

[0053] 2) Exclusion criteria:

[0054] Subjects who meet any of the following exclusion criteria are ineligible to participate in this experiment:

[0055] (1) During pregnancy or lactation;

[0056] (2) Failure to cooperate fully with the scale due to various reasons, such as severe visual and auditory impairment, or severe mental and behavioral abnormalities;

[0057] (3) There are contraindications to cranial MRI, such as after plate placement, claustrophobia, etc.; or the patient is unable to cooperate with the imaging examination due to agitation, etc.

[0058] (4) Previous central nervous system damage leading to cognitive decline: such as severe craniocerebral trauma, primary central nervous system tumors, etc.

[0059] (5) Emergency room visit or resuscitation required;

[0060] (6) History of blood transfusion within 7 days prior to sampling;

[0061] (7) People who have received organ transplants or have previously received non-autologous (allogeneic) bone marrow or stem cell transplants;

[0062] (8) Comorbid primary tumor.

[0063] 3) Subject information

[0064] In this embodiment, plasma samples were collected from 172 subjects at two medical centers. These subjects were divided into two groups: a cognitively normal and other types of cognitive impairment (HC+SCD+VCI) group (n=103) and an Alzheimer's disease and mild cognitive impairment (AD+MCI) group (n=69). The HC+SCD+VCI group specifically included 33 healthy controls (HC), 23 subjective cognitive decline (SCD), and 47 vascular cognitive impairment and other cognitive disorders (VCI). The AD+MCI group specifically included 60 Alzheimer's disease (AD) and 9 mild cognitive impairment (MCI) subjects. The plasma samples used for the modeling group were from 80 people with normal cognition and other types of cognitive impairment (HC+SCD+VCI) and 53 people with Alzheimer's disease and mild cognitive impairment (AD+MCI); the plasma samples used for the validation group were from 23 people with normal cognition and other types of cognitive impairment (HC+SCD+VCI) and 16 people with Alzheimer's disease and mild cognitive impairment (AD+MCI) (Table 1).

[0065] Table 1 Subject Information

[0066]

[0067]

[0068] 2. Plasma metabolite detection

[0069] 1) Test reagents:

[0070] Methanol, acetonitrile, water, acetic acid, methyl tert-butyl ether of mass spectrometry grade, and formic acid of chromatographic (HPLC) grade were all purchased from Sigma-Aldrich, USA.

[0071] 2) Sample preparation:

[0072] Take 100 μL of plasma and place it in 1000 μL of pre-cooled solution (methyl tert-butyl ether: methanol, volume ratio 3:1). Vortex to mix the extracted blood sample and obtain the sample extract. Add 500 μL of solution (methanol: water, volume ratio 3:1) to the sample extract, sonicate, let stand, vortex and centrifuge to separate the layers. The upper layer is the organic phase and the lower layer is the aqueous phase.

[0073] Organic phase processing: After the sample is separated into layers, take 500 μL of the upper organic phase into a centrifuge tube, dry it, add 200 μL of solution (acetonitrile:isopropanol, volume ratio 3:1), and incubate at room temperature for 15 minutes; after incubation, vortex the centrifuge tube to mix, sonicate for 5 minutes, and then centrifuge at room temperature for 5 minutes (12000 rpm); take 180 μL of supernatant from the centrifuge tube into a 2 mL glass vial, which is the organic phase test solution, and perform LC-MS detection.

[0074] Aqueous phase preparation: After sample separation, transfer the lower 400 μL aqueous phase to a centrifuge tube and add 1100 μL of ice-cold methanol to precipitate proteins. After protein precipitation, centrifuge the tube and transfer 1000 μL of the supernatant to a new centrifuge tube, then dry overnight. Add 200 μL of water to the dried centrifuge tube and incubate at room temperature for 15 minutes. After incubation, vortex the mixture, sonicate for 5 minutes, and then centrifuge at room temperature for 5 minutes (12000 rpm). Transfer 180 μL of the supernatant from the centrifuge tube to a 2 mL glass vial as the aqueous phase test solution. Analyze using LC-MS.

[0075] 3) Detection of small molecule metabolites:

[0076] Organic phase using Waters ACQUTTY BEH C8 1.7μm 2.1×100mm column, with WatersACQUTTY water phase. Small molecule separation was performed using an HSS T3 1.8μm 2.1×100mm column; both liquid chromatography and mass spectrometry used an ACQUITYUPLC I-Class liquid chromatography system (Waters) and a Q-Exactive mass spectrometry system (Thermo Fisher Scientific).

[0077] The mobile phase parameters are as follows:

[0078] Organic phase test solution mobile phase parameters - Mobile phase A is an aqueous solution containing 0.1% acetic acid and 0.1% ammonium acetate; Mobile phase B is an acetonitrile-isopropanol (7:3 v / v) solution containing 0.1% acetic acid and 0.1% ammonium acetate. The separation elution gradient is as follows: 0-12 minutes is 55%-89% mobile phase B, 12-19.5 minutes is 100% mobile phase B.

[0079] The mobile phase parameters of the aqueous test solution are as follows: Mobile phase A is an aqueous solution containing 0.1% formic acid; Mobile phase B is an acetonitrile solution containing 0.1% formic acid. The separation and elution gradient is as follows: 0-13 minutes is 1%-70% mobile phase B, and 13-18 minutes is 99% mobile phase B.

[0080] The mass spectrometry parameters are as follows:

[0081] Mass spectrometry data were acquired using Full MS and Full MS / dd-MS2 (each with both positive and negative modes). The parameters used by QExactive were as follows: Full MS mode had a resolution of 70,000, a scan range of 100-1500 m / z, an Automatic Gain Control (AGC) of 3E+6, and a Maximum IT of 200 ms; in Full MS / dd-MS2 mode, the resolution of the secondary mass spectrometer was 17,500, the quadrupole window was 1.5 m / z, the AGC was 1E+5, the maximum ion implantation time was 50 ms, and the Higher Energy Collisional Dissociation (HCD) was 30 eV.

[0082] 3. Metabolomics data preprocessing and metabolite identification

[0083] 1) Metabolomics data processing:

[0084] (1) Extract peaks from the RAW format file of the mass spectrometer and convert it into a FeatureXML format file to reduce the dimensionality of the original mass spectrometry data and improve the signal-to-noise ratio; (2) Use the peak alignment algorithm of OpenMS software to correct and align the retention time of the extracted peak format data between samples, thereby converting the mass spectrometry data into a data matrix; (3) Match and filter the isotope peaks in the data matrix obtained in step 2, and replace abnormal data (0, negative values, background noise, etc.) with missing values; (4) Remove the characteristic peaks with a detection rate of <80% from all the characteristic peaks obtained in step 3, fill the median value of the characteristic peak with the characteristic peaks with a detection rate of >80%, and add 5% random noise (following a standard normal distribution); (5) In order to reduce the difference in metabolite concentration between samples and make the data distribution more symmetrical, use NormalizationAutoencoder (NormAE) to perform normalization processing to remove systematic errors such as batch effects.

[0085] 2) Identification of metabolites:

[0086] After analyzing the raw data using software, the spectral information of the primary precursor ion (MS1) and secondary fragment ion (MS2) of the compound is obtained. This information, such as the mass-to-charge ratio (m / z) of the primary mass spectrometer and the fragment ion data, is matched with the spectral information of primary and secondary metabolites in public databases to qualitatively identify the metabolites. Commonly used metabolite databases include the Human Metabolite Database (HMDB, www.hmdb.ca), the Metabolomics Database (Metlin, metlin.scripps.edu), the Mass Spectrometry Database (www.massbank.jp), and the Lipid Map Database (Lipidmap, www.lipidmaps.org). Metabolites identified based on these databases are then finally validated using retention times, MS1, and MS2 mass spectrometry data obtained from separation of standards under the same chromatographic column and mass spectrometry conditions. The criteria for metabolite identification are a retention time difference within 0.1 min and a theoretical and measured molecular weight difference of less than 10 ppm.

[0087] 4. Data Analysis

[0088] 1) Biomarker screening

[0089] Metabolite detection was performed on the above samples, and a total of 465 metabolites were obtained after annotation. LASSO (Least Absolute Shrinkage and Selection Operator) regression analysis was performed on the data of the modeling group. The average error corresponding to each regularization parameter alpha was calculated using 5-fold cross-validation. The optimal alpha with the smallest error was found to be 0.0473. Metabolites with non-zero regression coefficients in their corresponding models were retained. Finally, 15 differential metabolites (Table 2) were selected as important metabolic markers to distinguish between the HC+SCD+VCI group and the AD+MCI group.

[0090] Table 2. 15 Important Metabolic Markers for Differentiating HC+SCD+VCI from AD+MCI

[0091] serial number Logo - English Logo - Chinese HMDBID 1 Ribose Ribose HMDB0000283 2 Asparagine Asparagine HMDB0251512 3 2-Hydroxyvalericacid 2-Hydroxyvalerate HMDB0001863 4 Glucose glucose HMDB0000122 5 Ascorbate ascorbic acid HMDB0000044 6 6-Aminohexanoic acid 6-Aminohexanoic acid HMDB0001901 7 Nicotinamide Niacinamide HMDB0001406 8 PI38:4 Phosphatidylinositol 38:4 HMDB0009815 9 PC40:6 Phosphatidylcholine 40:6 HMDB0008057 10 PE38:5 Phosphatidylethanolamine 38:5 HMDB0009036 11 PG34:1 Phosphatidylglycerol 34:1 HMDB0010574 12 Cert41:0(t16:0 / 25:0) Ceramide t41:0 (t16:0 / 25:0) - 13 Hex1Cerd41:1(d18:1 / 23:0) Hexosylceramide d41:1 (d18:1 / 23:0) HMDB0341524 14 TAG55:5 Triglycerides 55:5 - 15 TAG57:3 Triglycerides 57:3 -

[0092] Example 2

[0093] Methods for constructing diagnostic models to differentiate between cognitively normal individuals and other types of cognitive impairment (HC+SCD+VCI) with Alzheimer's disease and mild cognitive impairment (AD+MCI).

[0094] To verify the discriminative effect of the 15 metabolic biomarkers screened in Example 1 in distinguishing between HC+SCD+VCI and AD+MCI, a model was built using the 15 biomarkers and multivariate ROC curve analysis was performed. Three-quarters of the sample data from the HC+SCD+VCI and AD+MCI groups in the modeling group were randomly selected as the training set and one-quarter as the test set for validation. The model was then iterated 1000 times using a support vector machine (SVM) machine learning method. By statistically analyzing the average accuracy of the final model, a diagnostic model for distinguishing between HC+SCD+VCI and AD+MCI was constructed.

[0095] ROC curves are a method for studying the relationship between model sensitivity and specificity. Sensitivity is plotted on the ordinate, and specificity on the x-axis. The evaluation criterion is the area under the curve (AUC). An AUC greater than 0.5, and closer to 1, indicates better model performance and better discrimination. An AUC less than 0.5 indicates poor model accuracy. ROC classification prediction models, in addition to common parameters such as the receiver operating characteristic (ROC) curve and AUC, also include sensitivity and specificity.

[0096] Sensitivity calculation is shown in Formula I:

[0097]

[0098] Specificity is calculated using Formula II:

[0099]

[0100] Among them, TP (True Positive): True positive, the number of samples that are actually positive but were correctly predicted as positive;

[0101] TN (True Negative): The number of samples that are actually negative but were correctly predicted as negative.

[0102] FP (False Positive): The number of samples that are actually negative but are incorrectly predicted as positive.

[0103] FN (False Negative): The number of samples that are actually positive but are incorrectly predicted as negative.

[0104] The results are as follows Figure 1 As shown, the diagnostic model constructed using 15 metabolic biomarkers had an AUC of 0.969 (sensitivity = 0.923, specificity = 0.850). This result indicates that the constructed diagnostic model has high discriminative power.

[0105] To further validate the diagnostic model built based on the modeling group data, validation group data was used to validate the model. Multivariate ROC curve analysis was performed to evaluate the model's independent validation performance on unknown datasets outside the modeling group dataset. After the validation group samples were placed into the model constructed by the modeling group, the probability value was output for each sample based on the detection data of 15 important metabolic biomarkers distinguishing between HC+SCD+VCI and AD+MCI. Using the probability value of each sample as the discrimination threshold, a confusion matrix (including true positive, true negative, false positive, and false negative) was obtained. Sensitivity and specificity could be calculated using formulas. A point could be marked on the ROC analysis graph with sensitivity on the ordinate and 1-specificity on the abscissa. Similarly, when the probability value of each sample was used as the discrimination threshold, multiple different points were obtained in the ROC analysis graph. Connecting these points would produce an ROC curve. Figure 2 Among them, the point with the best sensitivity and specificity is selected, and the discrimination threshold at this time is 0.4223.

[0106] As shown in Table 3, the confusion matrix results indicate that in the diagnostic model constructed based on the 15 metabolic biomarkers, a discrimination threshold of 0.4223 was used. When the output result was ≥ the threshold, the diagnosis was Alzheimer's disease and mild cognitive impairment; when the output result was < the threshold, the diagnosis was normal cognition and other types of cognitive impairment. Among the 23 subjects with normal cognition and other types of cognitive impairment, 18 were correctly diagnosed as having normal cognition and other types of cognitive impairment, and 5 were misdiagnosed as having Alzheimer's disease and mild cognitive impairment. Among the 16 subjects with Alzheimer's disease and mild cognitive impairment, 13 were correctly diagnosed, and 3 were misdiagnosed as having normal cognition and other types of cognitive impairment. The ROC analysis results of the diagnostic model in the validation group are as follows: Figure 2 As shown, sensitivity and specificity were calculated based on the confusion matrix results, with an AUC of 0.837 (sensitivity = 0.813, specificity = 0.783). These results indicate that the established diagnostic model for distinguishing between cognitively normal individuals and other types of cognitive impairment with Alzheimer's disease and mild cognitive impairment also demonstrates good discriminative performance in the validation group.

[0107] Table 3. Confusion matrix between AD+MCI and HC+SCD+VCI diagnostic models

[0108] Types of diseases AD+MCI HC+SCD+VCI AD+MCI, N=16 13(TP) 3(FN) HC+SCD+VCI, N=23 5(FP) 18(TN)

[0109] Example 3

[0110] Modeling was performed using the following 11 metabolic biomarkers: ribose, asparagine, glucose, 6-aminocaproic acid, nicotinamide, phosphatidylinositol 38:4, phosphatidylethanolamine 38:5, phosphatidylglycerol 34:1, ceramide t41:0 (t16:0 / 25:0), hexosylceramide d41:1 (d18:1 / 23:0), and triglycerides 55:5. The predictive model was constructed according to the method in Example 2, and the same validation set samples were used for validation.

[0111] The results showed that the predictive model constructed using 11 metabolic biomarkers had an AUC of 0.915, with a sensitivity of 0.923 and a specificity of 0.800.

[0112] The validation group samples were used to verify the above prediction model and multivariate ROC curve analysis was performed. The results showed that the prediction model constructed with the above 11 metabolic biomarkers had an AUC of 0.840, with a sensitivity of 0.875 and a specificity of 0.739.

[0113] Example 4

[0114] Modeling was performed using the following eight metabolic biomarkers: ribose, glucose, 6-aminocaproic acid, nicotinamide, phosphatidylinositol 38:4, phosphatidylethanolamine 38:5, phosphatidylglycerol 34:1, and hexosylceramide d41:1 (d18:1 / 23:0); the predictive model was constructed according to the method in Example 2, and the same validation set samples were used for validation.

[0115] The results showed that when modeling using 8 metabolic biomarkers, the AUC was 0.918, with sensitivity of 0.929 and specificity of 0.850.

[0116] The validation group samples were used to verify the above prediction model and multivariate ROC curve analysis was performed. The prediction model constructed with the above 8 metabolic biomarkers had an AUC of 0.810, with a sensitivity of 0.813 and a specificity of 0.783.

[0117] Example 5

[0118] Diagnostic models for different combinations of metabolic biomarkers were modeled using the following five metabolic biomarkers: ribose, glucose, 6-aminocaproic acid, nicotinamide, and phosphatidylinositol 38:4. Predictive models were constructed according to the method in Example 2, and validated using the same validation set samples.

[0119] The results showed that when using five metabolic biomarkers for modeling, the AUC was 0.873, with sensitivity of 0.923 and specificity of 0.750, all of which had stable discriminative ability.

[0120] The validation group samples were used to verify the above prediction model and multivariate ROC curve analysis was performed. AUC = 0.802, where sensitivity = 0.750 and specificity = 0.739.

[0121] The above results show that the prediction models constructed using 5, 8, 11, and 15 metabolic biomarkers in this invention all have good discrimination effects.

[0122] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A metabolic marker characterized in that, includes at least 5 metabolites of ribose, asparagine, 2-hydroxyvalerate, glucose, ascorbate, 6-aminohexanoate, nicotinamide, phosphatidylinositol 38:4, phosphatidylcholine 40:6, phosphatidylethanolamine 38:5, phosphatidylglycerol 34:1, ceramide t41:0 (t16:0 / 25:0), hexosaccharide ceramide d41:1 (d18:1 / 23:0), triglyceride 55:5, and triglyceride 57:

3.

2. The metabolic marker of claim 1, wherein, includes the following metabolites: ribose, glucose, 6-aminohexanoate, nicotinamide, and phosphatidylinositol 38:

4.

3. The metabolic marker of claim 1, wherein, includes the following metabolites: ribose, glucose, 6-aminohexanoate, nicotinamide, phosphatidylinositol 38:4, phosphatidylethanolamine 38:5, phosphatidylglycerol 34:1, and hexosaccharide ceramide d41:1 (d18:1 / 23:0).

4. The metabolic marker of claim 1, wherein, includes the following metabolites: ribose, asparagine, glucose, 6-aminohexanoate, nicotinamide, phosphatidylinositol 38:4, phosphatidylethanolamine 38:5, phosphatidylglycerol 34:1, ceramide t41:0 (t16:0 / 25:0), hexosaccharide ceramide d41:1 (d18:1 / 23:0), and triglyceride 55:

5.

5. Use of the metabolite markers of any one of claims 1-4 in constructing a prediction model for Alzheimer's disease and mild cognitive impairment.

6. Use according to claim 5, characterized in that, The prediction model is modeled using a machine learning support vector machine.

7. An Alzheimer's disease and mild cognitive impairment prediction system, characterized by, The module includes the following connection states: a data acquisition module for acquiring detection data of the metabolite markers of any one of claims 1-4 in the sample to be tested; a data analysis module for analyzing the detection data of the metabolite markers acquired by the data acquisition module in the diagnostic model constructed in the use of claim 5 or 6 to obtain a prediction result; and a data output module for outputting the prediction result obtained by the data analysis module to a terminal for display.

8. The Alzheimer's disease and mild cognitive impairment prediction system of claim 7, wherein, The method for analyzing and predicting by the diagnostic model uses the determination results of the metabolite markers in the sample to be tested as input data, inputs the diagnostic model, and when the output result is ≥ threshold value, it is judged as Alzheimer's disease and mild cognitive impairment; when the output result is < threshold value, it is judged as normal cognition and other types of cognitive impairment diseases. The threshold value is 0.4223.

9. An Alzheimer's disease and mild cognitive impairment prediction device, characterized by, The prediction system for Alzheimer's disease and mild cognitive impairment of claim 7 or 8.

10. Use of a reagent for detecting the metabolite markers of any one of claims 1-4 in the preparation of a kit for distinguishing between normal cognition and other types of cognitive impairment diseases and Alzheimer's disease and mild cognitive impairment.

Images