Metabolic markers for detecting autoimmune encephalitis and uses thereof

Abstract

This invention belongs to the field of biomedical technology and discloses a set of metabolic markers for detecting autoimmune encephalitis and their applications. The metabolic markers for detecting autoimmune encephalitis include 4-aminophenol, GABA, homocitrulline, retinyl phosphate, trimethyllysine, dopamine, galactose-1-phosphate, and phenylpyruvic acid. These metabolic markers exhibit high specificity, sensitivity, and accuracy in screening for autoimmune encephalitis. The combined detection of these metabolites can help achieve early warning of autoimmune encephalitis, especially antibody-negative autoimmune encephalitis.

Description

Technical Field

[0001] This invention belongs to the field of biomedical technology, specifically relating to metabolic markers for detecting autoimmune encephalitis and their applications. Background Technology

[0002] Autoimmune encephalitis is a group of rapidly progressive, immune-mediated inflammatory diseases with diverse clinical manifestations, including altered mental status, memory impairment, psychiatric symptoms, seizures, and other focal neurological deficits. The detection of autoantibodies is a crucial component of the diagnostic criteria for autoimmune encephalitis. The initial introduction of anti-NMDAR antibodies in 2007, and the discovery of other autoantibodies over the past 20 years, have fundamentally changed the perception of autoimmune encephalitis as a previously unknown and highly fatal disease. With systematic and standardized treatment after diagnosis, the prognosis of autoimmune encephalitis patients improves significantly, with markedly lower mortality and disability rates, demonstrating the substantial benefits of timely diagnosis for these patients.

[0003] However, approximately one-third to one-half of patients with autoimmune encephalitis still have undetectable autoantibodies. Diagnosis relies primarily on clinical manifestations, auxiliary examination results, and the physician's personal experience, delaying timely and effective treatment. While the clinical manifestations and auxiliary examination diagnostic criteria for these antibody-negative autoimmune encephalitis patients may have high specificity, their sensitivity is low; only about one-quarter of anti-NMDAR encephalitis patients diagnosed with antibodies meet the diagnostic criteria for antibody-negative autoimmune encephalitis. Therefore, there is an urgent clinical need to identify other biomarkers to reduce missed diagnoses and misdiagnoses.

[0004] Current research on biomarkers for autoimmune encephalitis is limited, with most studies focusing on assessing disease severity and prognosis based on clinical symptoms, imaging features, and electroencephalogram (EEG) findings. Recent studies have also explored the use of immune inflammatory molecules (C4d, CD27, YKL-40) or neurodegenerative factors (Amyloid, Tau) for the differential diagnosis of specific antibody subtypes of autoimmune encephalitis, but research on diagnostic biomarkers for antibody-negative autoimmune encephalitis is lacking. Therefore, the development of molecular markers applicable to antibody-negative autoimmune encephalitis and the development of reagent kits are of significant clinical importance. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to address the shortcomings of the existing technology by providing metabolic markers for detecting autoimmune encephalitis and their applications. These metabolic markers have high specificity, sensitivity and accuracy in screening for autoimmune encephalitis. The combined detection of these metabolites can help to achieve early warning of autoimmune encephalitis, especially antibody-negative autoimmune encephalitis.

[0006] To address the technical problems raised in this invention, this invention provides metabolic markers for detecting autoimmune encephalitis, including 4-aminophenol, aminobutyric acid, citrulline, retinoid, trimethyllysine, dopamine, galactose-1-phosphate, and phenylpyruvic acid.

[0007] In the above scheme, the autoimmune encephalitis includes antibody-negative autoimmune encephalitis and antibody-positive autoimmune encephalitis.

[0008] In the above scheme, the sample used for testing is the cerebrospinal fluid of the test subject, and the test subject is a human.

[0009] The present invention also provides the application of the above-mentioned metabolic markers in the preparation of products for detecting autoimmune encephalitis.

[0010] The present invention also provides a kit for detecting autoimmune encephalitis, the kit comprising reagents for detecting the above-mentioned metabolic markers, the reagents for detecting metabolic markers comprising standards of metabolic markers, quality control of metabolic markers, extraction solvent one, and extraction solvent two.

[0011] In the above scheme, the quality control of the metabolic marker is obtained by drying the supernatant after extraction with an extraction solvent from a normal cerebrospinal fluid specimen that has not suffered from autoimmune encephalitis.

[0012] In the above scheme, the extraction solvent is a mixed solution of acetonitrile and methanol containing an isotopic internal standard, wherein the volume ratio of acetonitrile to methanol is 1:1 to 3:1, and the concentration of the isotopic internal standard is 0.05 to 1 μmol / L.

[0013] Furthermore, the isotope internal standard is 13 C-alanine.

[0014] In the above scheme, the extraction solvent two is a mixed solution of acetonitrile and water, wherein the volume ratio of acetonitrile to water is 1:1 to 9:1.

[0015] In the above scheme, the kit may further include a mobile phase used during detection, specifically one or more of mobile phase A1, mobile phase B1, mobile phase A2, and mobile phase B2.

[0016] In the above scheme, the mobile phase A1 is a mixed solution of ammonium formate solution and acetonitrile, wherein the volume percentage of ammonium formate solution is 90%~95% and the volume percentage of acetonitrile is 5%~10%.

[0017] Furthermore, the concentration of the ammonium formate solution is 2~10 mmol / L.

[0018] In the above scheme, the mobile phase B1 is an acetonitrile solution of formic acid, wherein the mass fraction of formic acid is 0.1~0.4%.

[0019] In the above scheme, the mobile phase A2 is an ammonium acetate solution.

[0020] Furthermore, the concentration of the ammonium acetate solution is 2~10 mmol / L.

[0021] In the above scheme, the mobile phase B2 is acetonitrile or aqueous acetonitrile, wherein the mass fraction of acetonitrile is 99.5%~100%.

[0022] In the above scheme, the reagent kit is stored at a temperature of 4~20℃.

[0023] This invention also provides a method for using the above-mentioned reagent kit, comprising the following steps: (1) Extraction solvent one was added to the cerebrospinal fluid sample of the test subject for extraction. After centrifugation, the supernatant was taken and dried. Then extraction solvent two was added for redissolution. After centrifugation, the supernatant was taken to obtain the test solution of the test subject. (2) Take the standard of the metabolic marker, add extraction solvent II to redissolve, centrifuge and take the supernatant to obtain the standard solution of the metabolic marker; (3) Take the quality control sample of the metabolic marker, add extraction solvent II to redissolve, centrifuge and take the supernatant to obtain the quality control solution of the metabolic marker; (4) The test solution, standard solution and quality control solution of the test subject were separated by ultra-high performance liquid chromatography system and mass spectrometry was performed by mass spectrometry to obtain the concentration of metabolites in the cerebrospinal fluid of the test subject; among them, the quality control solution was used to monitor and evaluate the stability of the system and the reliability of the experimental data; the isotope internal standard was used to correct the matrix effect, extraction loss and instrument fluctuation; the standard was used to plot the standard curve and calculate the absolute content of each substance.

[0024] In the above scheme, the cerebrospinal fluid sample of the test subject is the supernatant of the collected cerebrospinal fluid after centrifugation. The supernatant can be used directly or frozen and thawed before use.

[0025] In the above scheme, the volume ratio of the cerebrospinal fluid sample of the test subject to the extraction solvent is 1:1 to 1:2.

[0026] In the above scheme, when using extraction solvent one for extraction, first vortex mix for 1-5 minutes, then ultrasonic mix for 20-30 minutes.

[0027] In the above scheme, the drying is vacuum drying, and the drying temperature is 30~37℃.

[0028] In the above scheme, the volume ratio of the cerebrospinal fluid sample of the test subject to the extraction solvent is 1:1 to 1:2.

[0029] In the above scheme, when using extraction solvent two for reconstitution, directly vortex mix for 1~5 minutes.

[0030] In the above scheme, the centrifugal force is 13000~15000g and the centrifugation time is 15~30min.

[0031] In the above scheme, the extraction, reconstitution, and centrifugation processes are all carried out at 3~5℃.

[0032] In the above scheme, the chromatographic columns used in the chromatography include HILIC columns and C18 silica gel packed columns. The HILIC columns use mobile phases A1 and B1, and the C18 silica gel packed columns use mobile phases A2 and B2. The elution program is gradient elution of mobile phases A and B in a certain ratio.

[0033] In the above scheme, the mass spectrometry uses an electrospray ionization source and a multiple reaction monitoring mode.

[0034] In the above scheme, the reagents used in the kit and the mobile phase used for detection, such as acetonitrile, methanol, water, ammonium formate, and ammonium acetate, are all chromatographic grade.

[0035] The above scheme also includes step (5) comparing the measured metabolite concentration with a preset reference standard.

[0036] Further, the reference standard is set as follows: the metabolite concentrations are substituted into the following formula for calculation: logit(P) = -7.9783 + 0.0018×M1 + 0.0113×M2 + 0.0234×M3 + 0.2158×M4 + 0.0038×M5 -0.1213×M6 + 0.0017×M7 + 0.1114×M8; where M1, M2, M3, M4, M5, M6, M7, and M8 are the concentrations of 4-aminophenol, GABA, homocitrulline, retinoids, trimethyllysine, dopamine, galactose-1-phosphate, and phenylpyruvic acid in the subject's cerebrospinal fluid, respectively, in ng / mL; P>0.5 is assessed as a high risk of autoimmune encephalitis.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention proposes a set of metabolic marker compositions for detecting autoimmune encephalitis by screening multiple metabolites closely related to autoimmune encephalitis. These metabolic marker compositions have high specificity, sensitivity, and accuracy in screening for autoimmune encephalitis. The combined detection of these metabolites can help achieve early warning and risk stratification of autoimmune encephalitis, especially antibody-negative autoimmune encephalitis, thus providing patients with an earlier intervention window and improving treatment response rate and prognosis.

[0038] This invention further integrates metabolic marker compositions into a kit. This kit is easy to operate, has good repeatability, and is suitable for routine clinical laboratory testing, showing promising prospects for widespread application. Using this kit, the metabolic marker compositions in the cerebrospinal fluid of subjects can be accurately quantitatively analyzed. By comparing the test results with preset reference standards, objective and reproducible quantitative data support can be provided to clinicians. This makes the auxiliary identification of antibody-negative autoimmune encephalitis no longer entirely dependent on the clinician's experience-based judgment, effectively reducing the risk of misdiagnosis or missed diagnosis due to atypical symptoms or complex medical history. Attached Figure Description

[0039] Figure 1 This is a flowchart of screening metabolic markers and constructing a reference standard model in Embodiment 1 of the present invention.

[0040] Figure 2 These are chromatograms of all quality control samples during the implementation of targeted metabolomics in Example 1 of the present invention, where a is a chromatogram of nonpolar compounds and b is a chromatogram of polar compounds.

[0041] Figure 3 This is a PCA analysis diagram of targeted metabolomics of antibody-positive autoimmune encephalitis, control group, and quality control samples from Example 1 of the present invention.

[0042] Figure 4 The diagram shows the PLS-DA analysis of cerebrospinal fluid samples from antibody-positive autoimmune encephalitis patients and control patients in the training set of Example 1 of this invention. In the diagram, a is the PLS-DA diagram and b is the permutation diagram.

[0043] Figure 5 This is a volcano plot showing the differential expression of metabolites in cerebrospinal fluid samples from antibody-positive autoimmune encephalitis patients and control group patients in Example 1 of this invention.

[0044] Figure 6 This is an upset diagram of the eight metabolic markers obtained in Example 1 of the present invention.

[0045] Figure 7 This is a correlation coefficient diagram of eight metabolic markers obtained in Example 1 of the present invention.

[0046] Figure 8 This is a comparison chart of the discriminative abilities of various machine learning models established based on eight metabolic markers in the training set according to Embodiment 1 of the present invention.

[0047] Figure 9This is a graph showing the performance of the GLM (Logistic Regression) model based on 8 metabolic markers in the training set according to Embodiment 1 of the present invention. In the graph, a is the ROC plot of the GLM model, b is the coefficient of the 8 metabolic markers in the GLM model, c is the ROC plot of the 8 metabolic markers individually, and d is the area under the ROC curve of the 8 metabolic markers.

[0048] Figure 10 This is a graph showing the performance of the GLM model (Logistic Regression) established based on 8 metabolic markers in the test set according to Embodiment 1 of the present invention. In the graph, a is the accuracy of the GLM model in the test set, b is the diagnostic efficiency of the GLM model in the test set, c is the ROC plot of each of the 8 metabolic markers, and d is the area under the ROC curve of each of the 8 metabolic markers. Detailed Implementation

[0049] To better understand the present invention, the following embodiments further illustrate the content of the present invention, but the content of the present invention is not limited to the following embodiments.

[0050] Example 1: Screening and Model Establishment of Metabolic Markers for Detecting Autoimmune Encephalitis like Figure 1 As shown, the metabolic markers of the present invention are obtained by screening using the following method: (1) 43 patients with antibody-positive autoimmune encephalitis and 43 control patients were selected as training sets, and cerebrospinal fluid samples were taken to extract metabolites. (2) Using targeted metabolomics methods, liquid chromatography was employed. Tandem mass spectrometry (LC) MS / MS technology was used to determine the content of nearly 800 major metabolites in cerebrospinal fluid; (3) The obtained data were processed using partial least squares method. Discriminant Analysis (PLS) DA), VIP analysis and t-test were used to identify metabolites with significant differences between the two groups; (4) Through modeling and analysis using various machine learning methods (CatBoost, PLS, GLM, AdaBoost, Lasso, SuperPC, LightGBM, XGBoost, Elastic_net, Ridge, Decision_tree, SVM_RFE), we obtained eight metabolite combinations that had a significant impact on all models from the differential metabolites. (5) The obtained metabolite combinations were modeled using a variety of machine learning methods, and the model with the strongest discriminative ability and applicable to clinical practice was selected. (6) Twenty patients with antibody-negative autoimmune encephalitis and 20 other patients hospitalized during the same period were selected as the test set, and cerebrospinal fluid samples were taken to extract metabolites. (7) Using metabolomics, the content of the above 8 metabolites in cerebrospinal fluid was analyzed in a targeted manner, and the discriminative ability of the constructed model was evaluated.

[0051] The more specific steps are as follows: 1. Sample inclusion In this embodiment, the following inclusion criteria were used for subjects with autoimmune encephalitis: (1) All included patients with autoimmune encephalitis met the Chinese Expert Consensus on the Diagnosis and Treatment of Autoimmune Encephalitis (2022); (2) Autoantibodies (anti-NMDAR, anti-LGI1, anti-AMPAR1, anti-AMPAR2, anti-CASPR2, anti-GABAR, anti-DPPX, anti-IgLON5, anti-GlyRα1, anti-GABAARα1, anti-GABAARβ3, anti-GABAARγ2, anti-mGluR5, anti-D2R, anti-Neurexin-3α, anti-GAD65, anti-AK5, anti-KLHL11, etc.) were detected in the patient's blood or cerebrospinal fluid, and the detected autoantibodies were related to the symptoms; (3) The patient was in the acute phase of the disease; (4) The patient was newly diagnosed and had not received immunotherapy; (5) The patient was older than 14 years.

[0052] Meanwhile, the following exclusion criteria were used for subjects with autoimmune encephalitis: (1) not meeting the 2024 Canadian Consensus Guidelines: Diagnosis and Treatment of Autoimmune Encephalitis in Adults; (2) having other organic brain diseases or a history of severe brain trauma or brain surgery, or having heart, liver, kidney diseases, diabetes and other serious physical diseases; (3) having abnormalities in routine laboratory tests (complete blood count, liver function, urinalysis); (4) female subjects who are pregnant, breastfeeding, or menstruating; (5) having a history of drug and substance abuse; (6) having significant bleeding during lumbar puncture or other factors that cause the cerebrospinal fluid specimen to not meet the testing standards.

[0053] The inclusion criteria for the control group were: patients who required lumbar puncture to rule out encephalitis or other immune diseases for clinical diagnosis, whose routine cerebrospinal fluid tests, biochemical and antibody tests were all normal, and who were finally diagnosed with primary headache, post-cold headache, somatization disorder, Hunter syndrome, peripheral neuropathy, etc., with no history of neuropsychiatric diseases, no history of drug abuse or dependence, and no systemic physical diseases; and whose routine laboratory tests showed no significant abnormalities.

[0054] 2. Cerebrospinal fluid specimen acquisition The patient lies on their side on a hard bed with their back perpendicular to the bed surface, head flexed towards the chest, and hands clasped to the knees and pressed against the abdomen, maximizing spinal kyphosis to widen the intervertebral space and facilitate puncture. The L3-L4 or L4-L5 intervertebral space is selected as the puncture point. The area around the puncture point is disinfected three times in a spiral motion with povidone-iodine from the inside out, and a sterile drape is applied. The operator wears sterile gloves, draws lidocaine, and performs infiltration anesthesia layer by layer at the puncture point. Holding the lumbar puncture needle with the bevel of the needle facing the head, the needle is inserted perpendicularly to the skin at the puncture point. As the needle passes through the skin, subcutaneous tissue, supraspinous ligament, interspinous ligament, and ligamentum flavum, there will be changes in resistance; upon breaking through the ligamentum flavum and entering the subarachnoid space, a sudden loss of resistance, a feeling of "emptiness," will be felt. The needle stylet is slowly withdrawn; cerebrospinal fluid dripping out indicates successful puncture. A manometer is connected to measure intracranial pressure, and cerebrospinal fluid is collected in a sterile specimen tube. After centrifugation, the supernatant is collected. If the supernatant is to be immediately used for subsequent metabolite extraction and detection, it can be used directly; if the supernatant is not to be used for subsequent metabolite extraction and detection, it should be transferred to a freezer at -80℃ and stored frozen. Before use, it should be taken out and thawed slowly at 4℃.

[0055] 3. Metabolite extraction Take 50 μL of sample and add 50 μL of solution containing 1 μmol / L at 4℃. 13 C-alanine was mixed in a methanol / acetonitrile solution (1:1, v / v), vortexed for 1 min, sonicated at 4°C for 30 min, centrifuged at 14000 g at 4°C for 20 min, and the supernatant was collected and dried under vacuum at 35°C. Then, 50 μL of acetonitrile aqueous solution (acetonitrile:water = 1:1, v / v) was added to reconstitute the solution, vortexed for 1 min, centrifuged at 14000 g at 4°C for 15 min, and the supernatant was collected for analysis.

[0056] 4. Chromatography-mass spectrometry analysis 1) Chromatographic conditions The samples were separated using an Agilent 1290 Infinity LC ultra-high performance liquid chromatography (UHPLC) system with a HILIC and C18 column. The HILIC column temperature was 35℃; the flow rate was 0.3 mL / min; the injection volume was 2 μL; the mobile phase A1 was 90% (v / v) 2 mmol / L ammonium formate aqueous solution + 10% (v / v) acetonitrile, and the mobile phase B1 was 0.4% (w / w) formic acid in acetonitrile solution; the gradient elution program was as follows: 0–1.0 min, 85% B1; 1.0–3.0 min, B1 linearly changed from 85% to 80%; 3.0–4.0 min, 80% B1; 4.0–6.0 min, B1 linearly changed from 80% to 70%; 6.0–10.0 min, B1 linearly changed from 70% to 50%; 10–15.5 min, B1 maintained at 50%; 15.5–15.6 min, B1 linearly changed from 50% to 85%; 15.6–23 min, B1 maintained at 85%. The C18 column temperature was 40℃; the flow rate was 0.4 mL / min; the injection volume was 2 μL; the mobile phase A2 was 5 mmol / L ammonium acetate solution, and the mobile phase B2 was 99.5% acetonitrile; the gradient elution program was as follows: 0–5 min, B2 linearly changed from 5% to 60%; 5–11 min, B2 linearly changed from 60% to 100%; 11–13 min, B2 remained at 100%; 13–13.1 min, B2 linearly changed from 100% to 5%; 13.1–16 min, B2 remained at 5%; the sample was placed in an autosampler at 4℃ throughout the analysis.

[0057] To avoid the influence of fluctuations in instrument detection signals, continuous analysis of samples is performed in a random order. Quality control samples are inserted into the sample queue to monitor and evaluate the stability of the system and the reliability of experimental data.

[0058] 2) Mass spectrometry conditions Mass spectrometry analysis was performed using an AB 6500+ QTRAP mass spectrometer (AB SCIEX). ESI source conditions were as follows: Source temperature: 580℃, Ion Source Gas1 (GS1): 45℃, Ion Source Gas2 (GS2): 60℃, Curtain Gas (CUR): 35℃, Ion Spray Voltage (IS): +4500 V or -4500 V in positive or negative modes, respectively, monitored in MRM mode.

[0059] 5. Data Processing Peak extraction was performed on the raw MRM data using Multiquant software to obtain the ratio of peak area to internal standard peak area for each substance. The content of each substance was then calculated based on the standard curve. Quality control samples were processed simultaneously with the test samples, with one quality control sample interspersed among every 10 test samples to assess the variability of the derivatization process and instrumental analysis. Quality control samples are crucial for ensuring the reproducibility, reliability, and accuracy of metabolite quantification. Metabolites in quality control samples with a coefficient of variation below 30% were considered to have reproducible test results. The extracted data underwent quality assessment before data analysis.

[0060] 6. Test Results Cerebrospinal fluid samples were collected from 43 patients with autoimmune encephalitis and 43 controls. Metabolic substances were quantified, and the chromatograms of the quality control samples were analyzed as follows: Figure 2 As shown. Figure 3 Principal component analysis (PCA) score plot, as shown, indicates that the QC samples are tightly clustered relative to other samples, demonstrating the good reproducibility of the mass spectrometry detection method in this embodiment. Here, the QC samples are equal-volume mixtures of all test samples. Partial least squares analysis was performed on the metabolite quantification data. Discriminant Analysis (PLS) DA) analysis, such as Figure 4 As shown, autoimmune encephalitis can be well distinguished from other diseases, indicating significant metabolic differences between autoimmune encephalitis patients and the control group. Figure 5 As shown in the volcano plot, combined with the t-test, taking FC>±1.2, VIP>1, and p<0.05, 23 differentially expressed metabolites were identified between the two groups of patients, including α-hydroxyisobutyric acid, 2-hydroxybutyric acid, 2-hydroxy-3-methylbutyric acid, 2-hydroxy-2-methylbutyric acid, 2-aminophenolaminobutyric acid, 4-aminophenol, aminobutyric acid, phenylacetylglutamine, 2-methylalanine, phenylpyruvic acid, butyrylcarnitine, allantoin, galactose-1-phosphate, homocitrulline, p-hydroxyphenylpropionic acid, myristate glyceride, retinyl acid, isovalerylcarnitine, valerate, dopamine, glucose-1-phosphate, trimethyllysine, acetylglutamine, and propionylcarnitine, as shown in Table 1. The sample test results indicate that the differences in the levels of these metabolites can effectively distinguish between patients with autoimmune encephalitis and the control group.

[0061] Table 1. Differential metabolites in patients with antibody-positive autoimmune encephalitis.

[0062]

[0063] Differential metabolites are not necessarily good biomarkers. This invention uses 12 commonly used machine learning methods to analyze the importance of differential metabolites, and selects 8 differential metabolites that have made significant contributions in all models for model construction. Figure 6 CatBoost uses the Categorical Boosting algorithm to automatically process categorical features and reduce overfitting through ordered boosting. The number of iterations is 30, and the loss function is logarithmic. PLS uses Partial Least Squares (PLS) and its discriminant form (PLS-DA) to solve the multicollinearity problem in high-dimensional omics data by extracting latent variables. GLM (Generalized Linear Model) connects nonnormal dependent variables with linear predictors through a link function, breaking through the assumptions of traditional linear models and using a logistic regression binomial model. AdaBoost uses a weak decision tree as the base model and builds a strong classifier by iteratively weighting misclassified samples. The number of iterations is 30. Lasso (Lasso Regression) is based on L1 regularization and achieves feature selection and regression synchronization through coefficient compression. It uses a binomial model and cross-validation with 5 folds. SuperPC (Super Principal Component Analysis) combines principal component extraction and the Cox model to optimize the principal component selection strategy for high-dimensional features. LightGBM (Lightweight Gradient Boosting Machine) uses histogram optimization and leaf-by-leaf growth strategy to balance computational efficiency and anti-overfitting ability. The minimum data for leaf nodes is 7, and it is binary. XGBoost (Extreme Gradient Boosting Tree) is based on GBDT with L1+L2 regularization, second-order Taylor expansion to optimize the loss function, supports automatic handling of missing values, η: 0.3, maximum depth 6, binary logic, number of decision trees 3. Elasticnet (Elastic Net Regression) integrates L1 and L2 regularization to address the instability of feature selection in collinear omics data, a binomial model, 5-fold cross-validation, α 0.5. Ridge (Ridge Regression) uses L2 regularization to compress collinear feature coefficients (not set to 0), solving multicollinearity, binomial model, 5-fold cross-validation. Decisiontree (Decision Tree) uses CART trees as its core, constructing the model through recursive binary search, offering strong interpretability and assessable feature importance. SVM_RFE (Support Vector Machine-Recursive Feature Elimination) combines SVM feature weights with a recursive elimination strategy to progressively select the optimal feature subset, 3-fold, 5-fold. ROC curves were generated using balanced subsampling in Monte Carlo cross-validation. In each Monte Carlo cross-validation, two-thirds of the samples were used to evaluate the importance of metabolite features, build a classification model, and validate the model on the remaining one-third of the samples. This process was repeated multiple times to calculate the performance and confidence interval for each model.

[0064] The eight metabolites are 4-aminophenol, GABA, homocitrulline, retinoid, trimethyllysine, dopamine, galactose-1-phosphate, and phenylpyruvic acid. Correlation analysis suggests that these eight metabolites are correlated, with a correlation coefficient <0.5. Figure 7 Models were constructed using various machine learning methods for eight differentially expressed metabolites, ranked from strongest to weakest as follows: CatBoost, PLS, GLM, AdaBoost, Lasso, SuperPC, LightGBM, XGBoost, Elastic_net, Ridge, Decision_tree, and SVM_RFE. Figure 8 The GLM model (Logistic regression) is the most clinically significant discriminant model, with an area under the curve of 0.849 (0.775-0.923), a sensitivity of 74.7%, and a specificity of 93.02%. Logit(P) = -7.9783 + 0.0018×M1 + 0.0113×M2 + 0.0234×M3 + 0.2158×M4 + 0.0038×M5 - 0.1213×M6 + 0.0017×M7 + 0.1114×M8 (M1, 4-aminophenol; M2, GABA; M3, citrulline; M4, retinoids; M5, trimethyllysine; M6, dopamine; M7, galactose-1-phosphate; M8, phenylpyruvic acid; unit ng / mL; P>0.5 is considered autoimmune encephalitis, otherwise autoimmune encephalitis is excluded). Figure 9 ).

[0065] Example 2: A kit for detecting autoimmune encephalitis and its detection in antibody-negative autoimmune encephalitis. A kit for detecting autoimmune encephalitis, containing: 10 μg of metabolic marker standard, 10 μg of metabolic marker quality control, and 1 mL of extraction solvent one (a 1:1 volume ratio mixture of acetonitrile and methanol, containing 1 μmol / L). 13 C-alanine), 1 mL extraction solvent 2 (a 1:1 mixture of acetonitrile and water), 10 mL mobile phase A1 (90% 2 mmol / L ammonium formate solution + 10% acetonitrile), 10 mL mobile phase B1 (0.4% formic acid in acetonitrile solution), 10 mL mobile phase A2 (5 mmol / L ammonium acetate solution), 10 mL mobile phase B2 (99.5% acetonitrile).

[0066] The preparation method of the above reagent kit is as follows: (1) Preparation of lyophilized standard powder for metabolic markers: purchased from Sigma-Aldrich, composed of 4-aminophenol, aminobutyric acid, citrulline, retinoid, trimethyllysine, dopamine, galactose-1-phosphate, and phenylpyruvic acid in a mass ratio of 400:50:20:4:36:3:200:7; (2) Preparation of quality control samples for metabolic markers: Take 200uL of normal cerebrospinal fluid sample, add 200uL of extraction solvent one, vortex mix for 1min, sonicate at 4℃ for 30min, centrifuge at 14000 g at 4℃ for 20min, and take the supernatant and vacuum dry at 37℃. (3) Preparation of extraction solvent one: Mix 0.5 mL of methanol and 0.5 mL of acetonitrile, and add 1 nmol 13 C-alanine was used as an internal standard for isotopes. (4) Preparation of extraction solvent two: Mix 0.5 mL of water and 0.5 mL of acetonitrile; (5) Preparation of mobile phase A1: Mix 9 mL of 2 mmol / L ammonium formate solution with 1 mL of acetonitrile; (6) Preparation of mobile phase B1: Add 40 mg formic acid to 9.96 g acetonitrile to prepare an acetonitrile solution containing 0.4% formic acid; (7) Preparation of mobile phase A2: Add 5 mmol of ammonium acetate to 1 L of water to prepare a 5 mmol / L ammonium acetate solution, and take 10 mL; (8) Preparation of mobile phase B2: Add 5 mL of water to 995 g of acetonitrile to prepare 99.5% acetonitrile, and take 10 mL; (9) Assemble the above reagents into the kit.

[0067] Twenty patients with antibody-negative autoimmune encephalitis and 20 patients in the control group had their cerebrospinal fluid supernatant collected. (The diagnosis of antibody-negative autoimmune encephalitis patients met the diagnostic criteria for antibody-negative autoimmune encephalitis in the 2022 Chinese Expert Consensus on the Diagnosis and Treatment of Autoimmune Encephalitis and the 2024 Canadian Consensus Guidelines: Diagnosis and Treatment of Autoimmune Encephalitis in Adults). The samples were tested using the above-mentioned kit, including the following steps: (1) Add 50 μL of extraction solvent one to 50 μL of cerebrospinal fluid supernatant of the test subject, vortex at 4℃ for 1 min, sonicate at 4℃ for 30 min, centrifuge at 14000 g at 4℃ for 20 min, take the supernatant and vacuum dry at 37℃; then add 50 μL of extraction solvent two to redissolve, vortex for 1 min, centrifuge at 14000 g at 4℃ for 15 min, take the supernatant to obtain the test solution of the test subject; (2) Take 0.2 μg, 0.4 μg, 0.6 μg, 0.8 μg and 1 μg of the metabolic marker standard respectively, add 50 μL of extraction solvent to redissolve, vortex for 1 min, centrifuge at 14000 g 4℃ for 15 min to obtain standard solutions of metabolic marker with different contents, which are used to draw standard curves; (3) Take 10 μg of the quality control of the metabolic marker, add 200 μL of extraction solvent to redissolve, vortex for 1 min, centrifuge at 14000 g at 4℃ for 15 min, take the supernatant and divide it into 4 equal parts to obtain the quality control solution of the metabolic marker, which is used for sample monitoring and evaluation of the stability of the system and the reliability of experimental data. (4) The test solution, standard solution and quality control solution of the test body were separated by ultra-high performance liquid chromatography system and mass spectrometry was performed by mass spectrometry. The analysis conditions were the same as in Example 1. The peak area and internal standard area of ​​each metabolic marker were extracted. The absolute content of each substance was calculated by combining the standard curve and the concentration of metabolites in the cerebrospinal fluid of the test body was obtained. (5) Input the concentration of metabolites in the cerebrospinal fluid of the test subjects into the diagnostic model established in Example 1 (Logit(P) = -7.9783 + 0.0018×M1 + 0.0113×M2 + 0.0234×M3 + 0.2158×M4 + 0.0038×M5 -0.1213×M6 + 0.0017×M7 + 0.1114×M8 (M1, 4-aminophenol; M2, GABA; M3, citrulline; M4, retinoid; M5, trimethyllysine; M6, dopamine; M7, galactose-1-phosphate; M8, phenylpyruvic acid; unit ng / mL), calculate the P value of each, P>0.5 is evaluated as autoimmune encephalitis, otherwise autoimmune encephalitis is excluded.

[0068] In the evaluation results, four subjects with autoimmune encephalitis and four subjects without autoimmune encephalitis were incorrectly diagnosed as belonging to the relative group, while the remaining 32 subjects were correctly diagnosed. The model obtained in this invention has an area under the curve of 0.8 in the test set, with a sensitivity of 80% and a specificity of 80%, indicating that this model can effectively distinguish the subject type. Figure 10 ).

[0069] The above embodiments are merely examples for clear illustration and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations, and any obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A metabolic marker for detecting autoimmune encephalitis, characterized in that, It includes 4-aminophenol, aminobutyric acid, citrulline, retinoic acid, trimethyllysine, dopamine, galactose-1-phosphate, and phenylpyruvic acid.

2. The metabolic marker for detecting autoimmune encephalitis according to claim 1, characterized in that, The autoimmune encephalitis includes antibody-negative autoimmune encephalitis and antibody-positive autoimmune encephalitis; the sample being tested is the cerebrospinal fluid of the test subject, and the test subject is a human.

3. The use of the metabolic marker as described in claim 1 in the preparation of products for detecting autoimmune encephalitis.

4. A reagent kit for detecting autoimmune encephalitis, characterized in that, The reagent includes a reagent for detecting the metabolic markers of claim 1; the reagent for detecting the metabolic markers includes: a standard of the metabolic marker, a quality control of the metabolic marker, an extraction solvent one, and an extraction solvent two; the extraction solvent one is a mixed solution of acetonitrile and methanol containing an isotopic internal standard, and the extraction solvent two is a mixed solution of acetonitrile and water.

5. The kit for detecting autoimmune encephalitis according to claim 4, characterized in that, The volume ratio of acetonitrile to methanol in the extraction solvent is 1:1 to 3:1, and the concentration of the isotope internal standard is 0.05 to 1 μmol / L; the isotope internal standard is... 13 C-alanine; the volume ratio of acetonitrile to water in the extraction solvent II is 1:1 to 9:

1.

6. The kit for detecting autoimmune encephalitis according to claim 4, characterized in that, The quality control of the metabolic markers is prepared by drying the supernatant after extraction with an extraction solvent from normal cerebrospinal fluid specimens that have not suffered from autoimmune encephalitis; the kit is stored at a temperature of 4~20℃.

7. The method of using the kit for detecting autoimmune encephalitis as described in claim 4, characterized in that, Includes the following steps: (1) Extraction solvent one was added to the cerebrospinal fluid sample of the test subject for extraction. After centrifugation, the supernatant was taken and dried. Then extraction solvent two was added for redissolution. After centrifugation, the supernatant was taken to obtain the test solution of the test subject. (2) Take the standard of the metabolic marker, add extraction solvent II to redissolve, centrifuge and take the supernatant to obtain the standard solution of the metabolic marker; (3) Take the quality control sample of the metabolic marker, add extraction solvent II to redissolve, centrifuge and take the supernatant to obtain the quality control solution of the metabolic marker; (4) The test solution, standard solution and quality control solution of the test subject were separated by ultra-high performance liquid chromatography system and mass spectrometry was performed by mass spectrometry to obtain the concentration of metabolites in the cerebrospinal fluid of the test subject.

8. The method of using the kit for detecting autoimmune encephalitis according to claim 7, characterized in that, The cerebrospinal fluid sample of the test subject is the supernatant of the collected cerebrospinal fluid after centrifugation; the volume ratio of the cerebrospinal fluid sample of the test subject to extraction solvent one is 1:1 to 1:2; when using extraction solvent one for extraction, first vortex mixing for 1 to 5 min, then ultrasonic mixing for 20 to 30 min; the drying is vacuum drying at a temperature of 30 to 37°C.

9. The method of using the kit for detecting autoimmune encephalitis according to claim 7, characterized in that, The volume ratio of the cerebrospinal fluid sample of the test subject to the extraction solvent II is 1:1 to 1:2; when using the extraction solvent II for reconstitution, the mixture is directly vortexed for 1 to 5 minutes; the centrifugation force is 13000 to 15000 g, and the centrifugation time is 15 to 30 minutes; the extraction, reconstitution, and centrifugation processes are all carried out at 3 to 5°C.

10. The method of using the kit for detecting autoimmune encephalitis according to claim 7, characterized in that, It also includes comparing the measured metabolite concentrations with a preset reference standard, wherein the reference standard is calculated by substituting the metabolite concentrations into the formula: logit(P) = -7.9783 + 0.0018×M1 + 0.0113×M2 + 0.0234×M3 + 0.2158×M4 + 0.0038×M5 - 0.1213×M6 + 0.0017×M7 + 0.1114×M8; where M1, M2, M3, M4, M5, M6, M7, and M8 are the concentrations of 4-aminophenol, GABA, homocitrulline, retinoids, trimethyllysine, dopamine, galactose-1-phosphate, and phenylpyruvic acid in the subject's cerebrospinal fluid, respectively, in ng / mL; P>0.5 is assessed as a high risk of autoimmune encephalitis.

Images