Application of metabolites in predicting immune reconstitution outcomes in HIV-infected patients after antiretroviral therapy

Abstract

The present invention discloses a metabolite combination and its application in predicting the results of immune reconstruction after antiretroviral treatment of HIV-infected persons. The present invention has discovered for the first time a metabolite combination and model that can accurately distinguish between immune responders and immune non-responders, which are Glycylhydroxyproline, gamma-Glutamylalanine, C16-Sphingosine-1-phosphate, dihydrouracil, deoxyeritadenine, creatinine, trans-S-(1-Propenyl)-L-cysteine, N6,N6,N6-Trimethyl-L-lysine, and glucosamine. By detecting the content of the metabolite combination in the subject, it is possible to distinguish whether the subject can restore optimal immunity after antiretroviral treatment, thereby better intervening in the treatment process of HIV-infected persons, and guiding the optimization of the treatment plan for HIV-infected persons according to the treatment response.

Description

Technical Field

[0001] The present invention relates to the field of clinical medicine, and in particular to the application of metabolite combinations in predicting immune reconstruction results of HIV-infected patients after antiretroviral therapy. Background Art

[0002] Human immunodeficiency virus (HIV) infection is one of the world's major health problems. As of 2022, there are approximately 39 million people living with HIV worldwide. HIV mainly invades the human immune system. Untreated infected people are prone to various serious infections and malignancies in the late stage of the disease, which eventually lead to death. Common HIV viruses are divided into HIV-1 and HIV-2. Currently, the main epidemic strain of HIV in my country is HIV-1. The current commonly used treatment, namely antiretroviral therapy (ART), can effectively control the replication of HIV-1 virus and restore immune function, significantly improving the life expectancy of people infected with HIV-1 (PLWH), but about 15% to 30% of PLWH still fail to restore CD4 T cell counts. These people are considered to be poorly immune reconstituted people, and their risk of acquired immunodeficiency syndrome (AIDS) and non-AIDS events is higher than that of people with good immune reconstitution.

[0003] HIV-1 infection not only leads to changes in T cell subsets, but also systemic immune activation and inflammation. Previous metabolomics studies on plasma of PLWH receiving long-term ART treatment found that compared with healthy controls (HC), HIV-1 infected persons after ART treatment had metabolic disorders such as disordered amino acid metabolism, increased phospholipid and ceramide levels. There is increasing evidence that metabolic disorders play a key role in the pathogenesis of HIV-1. However, to date, there is a lack of understanding of the comprehensive metabolic characteristics of the systemic immune effects of ART treatment, as well as the different metabolic pathways in people with good immune reconstitution compared with those with poor immune reconstitution.

[0004] Therefore, based on the above background, the present invention studies whether there are key differential metabolites in metabolic disorders between people with good immune reconstitution and those with poor immune reconstitution in HIV-1 infected people. Exploring these treatment-related metabolic differences may reveal new intervention targets to enhance immune recovery of PLWH, optimize the treatment regimen for HIV-1 infected people based on the treatment response, and provide a reference for better intervention in the treatment process of immune non-responders. Summary of the invention

[0005] In order to make up for the deficiencies of the prior art, the present invention aims to predict the results of immune reconstruction in HIV-infected individuals after receiving antiretroviral therapy.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] The first aspect of the present invention provides a metabolite combination, which includes Glycylhydroxyproline, gamma-Glutamylalanine, C16-Sphingosine-1-phosphate, dihydrouracil, deoxyeritadenine, creatinine, trans-S-(1-Propenyl)-L-cysteine, N6,N6,N6-Trimethyl-L-lysine, and glucosamine.

[0008] The second aspect of the present invention provides the use of a reagent for detecting the content of the metabolite combination described in the first aspect of the present invention in a sample to be tested in the preparation of a product for predicting the outcome of immune reconstitution in HIV-infected individuals after antiretroviral therapy.

[0009] Furthermore, the sample to be tested is a sample containing the metabolome of HIV-infected individuals receiving antiretroviral treatment.

[0010] In some embodiments, the sample to be tested includes but is not limited to serum, plasma, urine, saliva, cerebrospinal fluid, lymph fluid, amniotic fluid, follicular fluid, synovial fluid, milk, tears, semen, feces, intestinal extract, and cell or tissue extract.

[0011] In a specific embodiment of the present invention, the sample to be tested is plasma.

[0012] Furthermore, the method for detecting the content of the metabolite combination includes, but is not limited to, any one or more of mass spectrometry, nuclear magnetic resonance spectroscopy, liquid chromatography-mass spectrometry, gas chromatography-mass spectrometry, capillary electrophoresis-mass spectrometry, surface enhanced Raman spectroscopy, matrix-assisted laser desorption / ionization mass spectrometry, quantitative mass spectrometry imaging, and surface-assisted laser desorption / ionization mass spectrometry.

[0013] Furthermore, the product includes but is not limited to reagents used to detect the content of metabolite combinations and sample pretreatment reagents.

[0014] Furthermore, the sample pretreatment reagents include, but are not limited to, one or more of liquid nitrogen, anticoagulants, pancreatin, dry ice, ethanol, methanol, PBS buffer, ultrapure water, physiological saline, and tissue embedding agents.

[0015] The third aspect of the present invention provides a method for constructing a prediction model for the immune reconstruction results of HIV-infected patients after antiretroviral treatment based on the metabolite combination described in the first aspect of the present invention, which specifically includes the following steps: obtaining the content data of the metabolite combination described in the first aspect of the present invention for people with good immune reconstruction and poor immune reconstruction after antiretroviral treatment, and constructing a prediction model.

[0016] Furthermore, the method also includes verifying the effectiveness of the prediction model.

[0017] Furthermore, the method for constructing the prediction model includes but is not limited to regression analysis.

[0018] In a specific embodiment of the present invention, the method for constructing the prediction model is LASSO regression analysis.

[0019] The fourth aspect of the present invention provides a prediction model constructed based on the method described in the third aspect of the present invention, which can be used to predict the outcome of immune reconstitution in HIV-infected individuals after antiretroviral treatment.

[0020] A fifth aspect of the present invention provides a system / device, characterized in that the system / device comprises a classification unit for predicting the immune reconstitution outcome of HIV-infected individuals after receiving antiretroviral therapy;

[0021] Classification unit: using the model constructed by the metabolite combination described in the first aspect of the present invention, calculating and obtaining the classification results of good immune reconstitution and poor immune reconstitution;

[0022] Furthermore, the model constructed using the metabolite combination described in the first aspect of the present invention is a model constructed by the method described in the third aspect of the present invention.

[0023] Further, the system / device also includes an input unit and an output unit;

[0024] Input unit: used to input the content data of the metabolite combination described in the first aspect of the present invention;

[0025] Output unit: used to output classification results.

[0026] The input unit and the output unit are both connected to the classification unit in some way.

[0027] Advantages and beneficial effects of the present invention:

[0028] The present invention classifies patients according to the treatment results after ART treatment through precise metabolomics research, namely immune responders and immune non-responders. Based on the results of metabolomics and LASSO regression analysis, a metabolite combination is proposed, which can effectively predict the immune reconstruction results of HIV-1 infected patients after ART treatment, and provides a new intervention target for improving the immune recovery of HIV-1 infected patients, especially immune non-responders. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 Metabolic profiles of HC, TN, IR and INR. Figure 1A in the figure is the PLS-DA score graph of metabolites in HC, TN, IR and INR. Figure 1 B in the figure is a collection diagram of different metabolites in HC, TN, IR and INR;

[0030] Figure 2 Volcano plot of differential metabolites between IR and INR;

[0031] Figure 3 This is the correlation analysis diagram of IR and INR metabolite differences;

[0032] Figure 4 is the LASSO model and ROC curve. Figure 4 A in is the binomial deviance of the LASSO regression model for differential metabolites between IR and INR, Figure 4 B in is the LASSO regression model coefficient of the differential metabolites between IR and INR, Figure 4 The C in is the ROC curve, which shows the model performance evaluated using randomized cross-validation;

[0033] Figure 5 Figure 2 is a graph of differential metabolite contents. DETAILED DESCRIPTION

[0034] The present invention quantitatively detects the metabolomics of people with good immune reconstitution and poor immune reconstitution through precise metabolomics analysis, and analyzes the differential metabolites in the two groups through the LASSO regression model to screen out the best classification combination to distinguish the two groups. After analysis, 9 key metabolites were found, namely Glycylhydroxyproline, gamma-Glutamylalanine, C16-Sphingosine-1-phosphate, dihydrouracil, deoxyeritadenine, creatinine, trans-S-(1-Propenyl)-L-cysteine, N6,N6,N6-Trimethyl-L-lysine, glucosamine, and the classification AUC of the metabolite combination was as high as 0.980, which was significantly effective in distinguishing IR and INR.

[0035] In the present invention, the term "well-reconstituted immune system" is also called "immune responder" or "IR", and refers to HIV-infected patients who have successfully achieved long-term suppression of HIV-1 viral replication after treatment, and whose CD4 T cell levels can be restored to optimal levels.

[0036] In the present invention, the term "poor immune reconstitution" is also called "immune non-responders" or "INR", which refers to people who have successfully achieved long-term suppression of HIV-1 viral replication after treatment, but whose CD4 T cell levels have not returned to optimal levels. People with immune non-responders or poor immune reconstitution have an increased risk of opportunistic infections, malignancies and non-AIDS complications.

[0037] In the present invention, the term "metabolomics" refers to a research method that follows the research ideas of genomics and proteomics, quantitatively analyzes all metabolites in an organism, and seeks the relative relationship between metabolites and physiological and pathological changes, which is an integral part of systems biology. Metabolomics mainly studies small molecule metabolites (MW <1000) that serve as substrates and products of various metabolic pathways. The research methods of metabolomics are similar to those of proteomics, and there are usually two methods. One method is called metabolite fingerprinting, which uses liquid chromatography-mass spectrometry (LC-MS) to compare the metabolites of each blood sample to determine all the metabolites therein. In essence, metabolic fingerprinting involves comparing the mass spectrometry peaks of metabolites in different individuals, ultimately understanding the structures of different compounds, and establishing a complete set of analytical methods for identifying the characteristics of these different compounds. Another method is metabolic profiling, in which researchers assume a specific metabolic pathway and conduct more in-depth research on it.

[0038] In the present invention, the term "sample" refers to a composition from a subject, which contains cells and / or other molecular entities to be characterized and / or identified based on, for example, physical, biochemical, chemical and / or physiological characteristics. For example, a sample refers to any sample derived from a subject that is expected or known to contain the subject's metabolome. Samples include, but are not limited to, tissue samples, primary or cultured cells or cell lines, cell cultures, cell supernatants, cell lysates, platelets, serum, plasma, vitreous humor, lymphatic fluid, synovial fluid, follicular fluid, semen, amniotic fluid, milk, whole blood, blood-derived cells, urine, cerebrospinal fluid, sputum, tears, sweat, mucus, saliva, tissue culture fluid, tissue extracts, homogenized tissues, cell extracts, and combinations thereof. In a specific embodiment of the present invention, the sample is plasma.

[0039] Since the samples required for metabolomics analysis require high precision, the collection and testing of samples should minimize human contamination, degradation and modification. In order to ensure the consistency of test results, the sampling site, sampling time, tissue sample sampling size, etc. should be consistent. After the samples are collected, they should be kept as low as possible, stable in state, and clearly labeled. The collection, pretreatment, and transportation to the experiment should be as quick as possible to reduce degradation and modification.

[0040] The present invention provides a method for constructing a model for predicting the outcome of immune reconstitution in HIV-infected patients after antiretroviral therapy, using a metabolite combination of Glycylhydroxyproline, gamma-Glutamylalanine, C16-Sphingosine-1-phosphate, dihydrouracil, deoxyeritadenine, creatinine, trans-S-(1-Propenyl)-L-cysteine, N6,N6,N6-Trimethyl-L-lysine, and glucosamine to construct the model. As known to skilled artisans, the step of associating the content of the metabolite combination with a certain probability or risk can be implemented and realized in different ways. The measured values ​​of the metabolite content are mathematically combined, and the combined values ​​are associated with the fundamental problem of predicting the outcome of immune reconstitution. The measured values ​​of metabolite content can be combined by any suitable existing mathematical method, such as linear regression model, logistic regression model, LASSO regression model, RIDGE regression model, linear discriminant analysis model, nearest neighbor model, decision tree model, perceptron model, neural network model, support vector machine model, naive Bayes model, AdaBoost model, GBDT model, XGBoost model, LightGBM model, CatBoost model or random forest model, logarithmic regression model, characteristic gene linear discriminant analysis model, ShrunkenCentroids model, StepAIC model, Kth-Nearest Neighbor model, Boosting model, hidden Markov model.

[0041] In a specific embodiment of the present invention, a LASSO regression model is used.

[0042] AUC measurement is useful for comparing the accuracy of classifiers across the entire data range. Classifiers with higher AUC have higher ability to correctly classify unknowns between two target groups. ROC curves are useful for describing the performance of specific features (e.g., any metabolite content described in the present invention) when distinguishing between two populations (e.g., immune response and unresponsive individuals). ROC curves can be generated with respect to individual features, and can be generated with respect to other individual outputs, for example, a combination of two or more features can be combined with mathematical methods (e.g., addition, subtraction, multiplication, etc.) to provide a separate sum value, and the separate sum value can be plotted in the ROC curve. In addition, any combination of multiple features of a combination derived from a separate output value can be plotted in the ROC curve. These combinations of features can include tests. ROC curves are graphs of the true positive rate (sensitivity) of a test for the false positive rate (specificity) of a test.

[0043] The present invention provides a system or device programmed to implement the method of the present invention. The system may be an electronic device of a user or a computer system remotely located relative to the electronic device. The electronic device may be a mobile electronic device. In the system or device provided by the present invention, the input module, the classification module and the output module may be connected in some way and transmit data to each other, and the connection method includes but is not limited to connection through various interface connection lines, connection through the Internet, connection through a mobile storage medium to transfer data, etc.

[0044] The present invention will be further described in detail below in conjunction with the accompanying drawings and examples. The following examples are only used to illustrate the present invention and are not intended to limit the scope of the present invention. Simple improvements to the present invention made according to the essence of the present invention all fall within the scope of protection claimed in the present invention.

[0045] Example 1 Precision metabolomics and bioinformatics analysis

[0046] 1. Experimental Materials and Procedures

[0047] 1. Experimental samples

[0048] The study cohort consisted of 108 participants, including 24 TNs and 68 PLWH receiving ART (33 IRs and 35 INRs). IR was defined as a CD4+ T-cell count >350 cells / μl after ART for more than 2 years and a plasma HIV-1 viral load <20 copies / ml, and INR was defined as a CD4 T-cell count ≤350 cells / μl after ART for more than 2 years and a plasma HIV-1 viral load <20 copies / ml. Exclusion criteria included pregnancy, any abnormal blood glucose, renal, or liver function, or coinfection with hepatitis, tuberculosis, syphilis, or other infectious diseases. In addition, 16 HIV-negative HCs were recruited. The characteristics of the participants are shown in Table 1. This study was approved by the Ethics Committee of the Fifth Medical Center of the General Hospital of the Chinese People's Liberation Army (ky-2021-7-6-1) and was conducted in accordance with the principles of the Declaration of Helsinki. All participants provided written informed consent before enrollment.

[0049] Table 1. Clinical characteristics of participants at enrollment

[0050]

[0051]

[0052] Quantitative variables are expressed as median (range). Qualitative variables are expressed as percentage.

[0053] Abbreviations: HC, healthy controls; TNs, treatment-naive HIV-1-infected persons; IRs, immune responders; INRs, immune nonresponders; BMI, body mass index; NA, not available; LDL, below the limit of detection; ART, antiretroviral therapy; NRTIs, nucleoside reverse transcriptase inhibitors; NNRTIs, nonnucleoside reverse transcriptase inhibitors; PIs, protease inhibitors; T N Cell, naive T cell; T EM Cells, effector memory T cells.

[0054] a Kruskal-Wallis test; b χ2 test; c Mann-Whitney U test.

[0055] As shown in Table 1, there were no statistical differences in gender and BMI among HC, TN, IR, and INR. In addition, there were no statistical differences in ART regimen and duration between IR and INR. Consistent with previous studies, INR patients were older, had lower CD4 T cell counts, and had lower CD4 / CD8 ratios compared with IR. N The frequency of CD4 T cells decreased, whileEM Cells and CD8 T EM Compared with HC, T N HLA-DR + CD38 + CD8 T cells and PD-1 + The frequency of CD8 T cells increased significantly. Even in IR, CD8 T cell counts were significantly higher than those in HC. Therefore, ART failed to fully restore the normal immune profile of PLWH, especially in INR, and new intervention targets are needed to achieve better immune recovery.

[0056] 2. Plasma metabolomics extraction

[0057] 50 μL of plasma was mixed with 200 μL of pre-chilled methanol containing 0.28 mM phenylhydrazine, vortexed for 10 seconds, and incubated at 1500 rpm and 4°C for 30 minutes; then, the sample was kept at -20°C for 30 minutes to derivatize α-keto acids. The sample was centrifuged at 12000 rpm for 10 minutes at 4°C, and the supernatant was transferred to a new tube and spun at 4 °C. 2 The dried extracts were dried in a SpeedVac vacuum concentrator in mode O. The dried extracts were reconstituted in 2% acetonitrile in water before LC-MS analysis on an Agilent 1290II HPLC coupled to a Sciex 5600+ quadrupole TOF mass spectrometer.

[0058] 3. Quantitative analysis

[0059] Polar metabolites were separated on a Waters ACQUITY HSS-T3 column (3.0×100 mm, 1.8 μm). The MS parameters for detection were: ESI source voltage 5.5 kV in positive ion mode, -4.5 kV in negative ion mode; evaporator temperature, 500°C; drying gas (N2) pressure, 50 psi; nebulizer gas (N2) pressure, 50 psi; curtain gas (N2) pressure, 35 psi; scanning range was m / z 60-800. Information-dependent acquisition mode was used for MS / MS analysis of metabolites. The collision energy was set to (±) 35±15 eV. TF 1.7.1 software (AB Sciex, Concord, ON, Canada) was used for data acquisition and processing. All detected ions were extracted into Excel in a two-dimensional matrix format, including mass-to-charge ratio (m / z), retention time, and peak area, and isotopic peaks were filtered using MarkerView1.3 (AB Sciex, Concord, ON, Canada). PeakView 2.2 (ABSciex, Concord, ON, Canada) was used to extract MS / MS data and compare them with the Metabolite Database (AB Sciix, Concording, ON, Canada), HMDB, METLIN, and standard references to annotate ion identities.

[0060] Isotope-labeled internal standards (IS) mixture purchased from Cambridge Isotope Laboratories was spiked into the samples for metabolite quantification, including L-phenylalanine-d 8 , L-tryptophan-d 8 , L-isoleucine-d 10 , L-Leucine-d 10 , L-methionine-d 3 , L-valine-d 8 , L-proline-d 7 , L-alanine-d 7 DL-serine-d 3 DL-glutamate-d 5 , glycine-d 2 , L-aspartic acid-d 3 , L-arginine-d 7 , L-glutamine-d 5 , L-lysine-d 9 , L-histidine-d 5 , Taurine- 13 C 2 , Betaine-d 11 , urea-( 13 C, 15 N 2 ), L-sodium lactate-d 3 , N-trimethylamine oxide-d 9 , choline-d 13 , Malic acid-d 3 , citric acid-d 4 , succinate-d 4 , Fumaric acid-d 2 , hypoxanthine-d 3 , Xanthogenic acid- 15 N 2 , thymidine ( 13 C 10 ,15 N 2 ), inosine- 15 N 4 , uridine-d 2 , Methylsuccinate-d 6 , benzoic acid-d 5 , Creatine-d 3 , creatinine-d 3 Glutaric acid-d 4 , glycine-d 2 , kynurenine-d 5 , L-citrulline-d 4 , L-Threonine-( 13 C 4 , 15 N), L-tyrosine-d 7 , p-cresol sulfate-d 7 , sarcosine-d 3 , trans-4-hydroxy-L-proline-d 3 , uric acid-( 13 C; 15 N 3 ). The peak areas of endogenous metabolites were normalized to the areas of their corresponding isotope-labeled structural analogs for quantification. For endogenous metabolites without labeled structural analogs, an automatic algorithm selected the best internal standard for quantification based on the normalized minimum coefficient of variation (COV) rule. 512 metabolites were characterized and quantified using isotope-labeled internal standards. To ensure analytical accuracy, a quality control sample was injected after every ten experimental samples. Principal component analysis (PCA) showed that the quality control samples clustered together and their metabolomics were highly correlated, with Spearman correlation coefficients (r) ranging from 0.99 to 1, verifying the high stability and reproducibility of the metabolomics data.

[0061] 4. Bioinformatics and statistical analysis

[0062] The characteristics of the enrolled participants were analyzed descriptively using frequency tables of categorical variables and medians and interquartile ranges for continuous variables. Differences in sociodemographic and clinical characteristics between HC, TN, IR, and INR were assessed by chi-square tests to determine the independence of categorical variables. For continuous variables, we used the nonparametric Mann-Whitney U test for comparisons between two groups and the Kruskal-Wallis test for comparisons involving more than two groups.

[0063] Statistical analysis of metabolomics data was performed in R version 4.2.3. Differential analysis was performed using the Limma R package, with age, sex, and BMI as covariates for log2-transformed metabolomics data. P values ​​were adjusted using the Benjamini-Hochberg method. Log2 fold changes and P values ​​from the Limma model were shown as volcano plots for comparison of combined changes. Figure 2 The “UpSetR” package was used to generate a collection plot to visually represent the intersections between statistically significant metabolites (Limma P value < 0.05) in various comparison combinations. Figure 1 B. Partial least squares discriminant analysis (PLS-DA) was performed using the "ropls" R package, and the clustering of the samples was visualized. The results are shown in Figure 1 A. The least absolute shrinkage and selection operator (LASSO) with cross validation was used to select the subset of metabolites that best discriminated between INR and IR.

[0064] 5. Metabolite combination verification classification efficiency

[0065] ROC curves were plotted using the area under the curve to demonstrate the discriminative ability of the variables selected by LASSO. Networks were constructed from differentially correlated pairs (P < 0.05) and further analyzed using the R package “MEGENA” (Multiscale Embedded Gene Co-expression Network Analysis). Differential correlations were calculated using the R package DGCA, and the results are shown in Figure 3 shown.

[0066] 2. Model and Verification Results

[0067] Using the LASSO regression model with 6-fold cross-validation on all samples, plasma polar metabolites that can distinguish INR and IR were screened out. According to the area under the ROC curve (AUC), 9 polar metabolites were screened out from 60 differential metabolites (see Table 2), including Glycylhydroxyproline, gamma-Glutamylalanine, C16-Sphingosine-1-phosphate, dihydrouracil, deoxyeritadenine, creatinine, trans-S-(1-Propenyl)-L-cysteine, N6,N6,N6-Trimethyl-L-lysine, and glucosamine. The study cohort was divided into training and test sets in a ratio of 7:3. The ROC curve of the model constructed with the combination of 9 polar metabolites showed that the prediction results were extremely accurate, with an AUC of up to 0.980. The results are as follows Figure 4 The contents of these nine polar metabolites are shown in Figure 5 As shown, there are significant differences between IR and INR.

[0068] Table 2. Differential metabolites between INR and IR

[0069]

[0070]

[0071]

[0072] The description of the above embodiments is only used to understand the method and core idea of ​​the present invention. It should be pointed out that, for those skilled in the art, several improvements and modifications can be made to the present invention without departing from the principle of the present invention, and these improvements and modifications will also fall within the scope of protection of the claims of the present invention.

Claims

1. Use of a reagent for detecting the combined content of metabolites in a sample to be tested in the preparation of a product for predicting the outcome of immune reconstitution in HIV-infected individuals after antiretroviral therapy, characterized in that: The metabolite combination is a combination of Glycylhydroxyproline, gamma-Glutamylalanine, C16-Sphingosine-1-phosphate, dihydrouracil, deoxyeritadenine, creatinine, trans-S-(1-Propenyl)-L-cysteine, N6,N6,N6-Trimethyl-L-lysine, and glucosamine.

2. The use according to claim 1, characterized in that: The sample to be tested is a sample containing the metabolome of HIV-infected persons receiving antiretroviral treatment.

3. The use according to claim 1, characterized in that: The samples to be tested include serum, plasma, urine, saliva, cerebrospinal fluid, lymph fluid, amniotic fluid, follicular fluid, synovial fluid, milk, tears, semen, feces, intestinal extracts, and cell or tissue extracts.

4. The use according to claim 1, characterized in that: The sample to be tested is plasma.

5. The use according to claim 1, characterized in that: The method for detecting the content of the metabolite combination includes any one or more of mass spectrometry, nuclear magnetic resonance spectroscopy, liquid chromatography-mass spectrometry, gas chromatography-mass spectrometry, capillary electrophoresis-mass spectrometry, surface enhanced Raman spectroscopy, matrix-assisted laser desorption / ionization mass spectrometry, quantitative mass spectrometry imaging, and surface-assisted laser desorption / ionization mass spectrometry.

6. The use according to claim 1, characterized in that: The product includes reagents used to detect the content of metabolite combinations and sample pretreatment reagents.

7. The use according to claim 6, characterized in that: The sample pretreatment reagent includes one or more of liquid nitrogen, anticoagulant, pancreatin, dry ice, ethanol, methanol, PBS buffer, ultrapure water, physiological saline, and tissue embedding agent.

8. A method for constructing a prediction model for immune reconstitution outcomes of HIV-infected patients after antiretroviral therapy based on the metabolite combination of claim 1, comprising the following steps: Obtaining content data of the metabolite combination in claim 1 for people with good immune reconstitution and people with poor immune reconstitution after antiretroviral treatment, and constructing a prediction model; The method for constructing the prediction model is computational modeling; The computational modeling method is at least one of a linear regression model, a logistic regression model, a LASSO regression model, a RIDGE regression model, a linear discriminant analysis model, a nearest neighbor model, a decision tree model, a perceptron model, a neural network model, a support vector machine model, a naive Bayes model, an AdaBoost model, a GBDT model, an XGBoost model, a LightGBM model, a CatBoost model or a random forest model, a logarithmic regression model, a characteristic gene linear discriminant analysis model, a ShrunkenCentroids model, a StepAIC model, a Kth-Nearest Neighbor model, a Boosting model, and a hidden Markov model.

9. The method according to claim 8, characterized in that The method for constructing the prediction model is a LASSO regression model.

10. A system / device, characterized in that: The system / device includes a classification unit for predicting the immune reconstitution outcome of HIV-infected individuals after antiretroviral therapy; The classification unit uses the model constructed by the metabolite combination in claim 1 to calculate and obtain the classification results of good immune reconstitution and poor immune reconstitution.

11. The system / device according to claim 10, characterized in that: The model constructed using the metabolite combination in claim 1 is a model constructed according to the method described in any one of claims 8 to 9.

12. The system / device according to claim 10, characterized in that: The system / device also includes an input unit and an output unit; The input unit is used to input the content data of the metabolite combination in claim 1; The output unit is used to output the classification result; The input unit and the output unit are both connected to the classification unit in some way.

Images