A rheumatoid arthritis diagnosis product, system and application thereof

Abstract

The present invention discloses a rheumatoid arthritis diagnostic product, system and application thereof, and relates to the technical field of rheumatoid arthritis. Diagnostic markers include: calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine, anti-cyclic citrullinated peptide antibodies and lysine. The present invention provides biomarkers that can accurately diagnose rheumatoid arthritis RA, including calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine, anti-cyclic citrullinated peptide antibodies (anti-CCP) and lysine, a total of 5 indicators. These 5 indicators have the technical advantages of few indicators and excellent diagnostic performance. The diagnostic model constructed based on the above 5 indicators has good sensitivity, specificity and accuracy. It can be used for the diagnosis of rheumatoid arthritis.

Description

Technical Field

[0001] The present invention relates to the technical field of rheumatoid arthritis, and in particular to a rheumatoid arthritis diagnosis product, system and application thereof. Background Art

[0002] Rheumatoid arthritis (RA) is an autoimmune disease that causes symmetrical and multi-site joint damage. The factors and exact causes are still unclear, but genetic and environmental factors may affect the risk and severity of disease progression. Compared with other chronic diseases, RA patients have a poor quality of life due to its disability. If treatment is insufficient or delayed, 80% and 40% of RA patients will suffer from joint dislocations and joint damage, respectively, and will be unable to work. Even the most effective treatments cannot reverse the joint damage of RA patients, and their physical function can never return to normal. The best strategy for RA patients is to relieve clinical symptoms such as inflammation, joint swelling, and inflammation-related indicators. Early diagnosis and treatment of RA can prevent the progression of joint damage in 90% of early RA patients. Therefore, early diagnosis, that is, timely identification and confirmation of disease activity in RA patients, is crucial to prevent or reduce joint damage and disability and improve prognosis.

[0003] At present, the clinical diagnosis of RA mainly relies on clinical symptoms and laboratory indicators. Clinical symptoms include pain and swelling of the metacarpophalangeal joints, metatarsophalangeal joints or both joints; morning stiffness of the finger joints lasting 30 minutes or more; laboratory indicators include autoantibodies, C-reactive protein levels and erythrocyte sedimentation rate. Commonly used autoantibody indicators are RF and anti-CCP antibodies. Although 50%-80% of RA patients are antibody-positive, many RA patients are negative for RF and anti-CCP. RF is not specific to RA and may be present in patients with other diseases, such as hepatitis C and healthy elderly people. The diagnostic accuracy of these tests is not satisfactory. The limited clinical experience of non-rheumatologists in evaluating these patients, especially in primary care, and the fact that mild signs and symptoms often overlap with other forms of inflammatory polyarthritis, pose a major obstacle to accurate diagnosis.

[0004] Increased demand for energy and biosynthesis leads to metabolic disorders in RA patients and promotes chronic inflammation of damaged tissues. RA patients experience metabolic disorders in the body, and these changes may be closely related to the pathogenesis and disease progression of RA. Amino acids (AA), lipids, and carbohydrates are the most important metabolites in the body. AA not only plays an important role in energy supply and conversion, but also regulates immunity, acid-base balance, cell signal transduction, etc. to maintain the body's homeostasis. Amino acids in the human body are dynamic and change according to different physiological or pathological conditions; therefore, serum AA levels are potential biomarkers in disease diagnosis or treatment monitoring. Blood samples are considered to be a good resource for obtaining information on the disease status of RA patients. The integration of multiple metabolites may have considerable sensitivity and specificity in predicting RA.

[0005] Machine learning methods can combine the results of statistical and medical studies, improve statistical power through larger sample sizes, and improve the accuracy of results by reducing standard errors. Therefore, this approach can produce more accurate estimates of effect sizes. Although global advances in metabolomics have provided some theoretical insights into the pathogenesis and diagnosis of RA, the results are contradictory due to differences in influencing factors. Although several metabolic pathways have been associated with RA patients, the exact trends remain uncertain.

[0006] In view of this, the present invention is proposed. Summary of the invention

[0007] The object of the present invention is to provide a rheumatoid arthritis diagnosis product, system and application thereof to solve the above-mentioned technical problems.

[0008] The present invention is achieved in that:

[0009] In a first aspect, the present invention provides an application of a reagent for detecting diagnostic markers in the preparation of a diagnostic product for rheumatoid arthritis, wherein the diagnostic markers include: calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine, anti-cyclic citrullinated peptide antibodies (anti-CCP) and lysine.

[0010] In a second aspect, the present invention also provides a rheumatoid arthritis diagnostic reagent, which includes reagents for detecting calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine, anti-cyclic citrullinated peptide antibodies (anti-CCP) and lysine levels.

[0011] In a third aspect, the present invention also provides a rheumatoid arthritis diagnostic kit, which comprises the above-mentioned rheumatoid arthritis diagnostic reagent.

[0012] In a fourth aspect, the present invention also provides the use of diagnostic markers in constructing and / or evaluating a rheumatoid arthritis diagnostic model, wherein the diagnostic markers include: calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine, anti-cyclic citrullinated peptide antibodies (anti-CCP) and lysine.

[0013] In a fifth aspect, the present invention also provides a system for diagnosing rheumatoid arthritis, which comprises: (1) a unit for determining a diagnostic marker combination characteristic value in a sample; (2) a unit for comparing the diagnostic marker combination characteristic value in the sample; (3) a data storage medium, wherein the diagnostic marker combination characteristic value is the content or concentration of the diagnostic marker; the diagnostic markers include: calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine, anti-cyclic citrullinated peptide antibodies (anti-CCP) and lysine.

[0014] The present invention has the following beneficial effects:

[0015] The present invention provides biomarkers for accurate diagnosis of rheumatoid arthritis RA, including calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine, anti-cyclic citrullinated peptide antibodies (anti-CCP) and lysine, a total of 5 indicators. These 5 indicators have the technical advantages of few indicators and excellent diagnostic performance. The diagnostic model constructed based on the above 5 indicators has good sensitivity, specificity and accuracy. It can be used for the diagnosis of rheumatoid arthritis. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 Score plots for the supervised orthogonal partial least squares discriminant analysis (OPLS-DA) model for the control group (blue) and the rheumatoid arthritis group (pink);

[0018] Figure 2 (a) Calcium (Ca), arginine / ornithine (Arg / Orn), anti-cyclic citrullinated peptide (CCP), lysine (Lys), and phenylalanine / tyrosine (Phe / Tyr) contents in rheumatoid arthritis (RA) and control groups; (b) Shapley additive interpretation (SHAP) model of the five indicators of the V5 model;

[0019] Figure 3(a) Receiver operating characteristic (ROC) curve of the model obtained from the training set of five different indicators selected from the rheumatoid arthritis (RA) and control groups; (b) Receiver operating characteristic (ROC) curve of the model obtained from the validation set of five different indicators selected from the RA group and the control group; (c) Receiver operating characteristic (ROC) curve of the model obtained from the training set of eight different indicators selected from the RA group and the control group; (d) Receiver operating characteristic (ROC) curve of the model obtained from the validation set of eight different indicators selected from the RA group and the control group; (e) Receiver operating characteristic (ROC) curve of the model obtained from the training set of 19 different indicators selected from the RA and control groups; (f) Receiver operating characteristic (ROC) curve of the model obtained from the validation set of 19 different indicators selected from the RA group and the control group;

[0020] Figure 4 (a) Recursive feature elimination and cross validation (RFECV) - anomaly detection algorithm (ADA) index screening diagram; (b) non-trend correspondence analysis diagram of V5 and V8 models; (c) model receiver operating characteristic (ROC) curves obtained by V5 and V8 models: ESR, anti-CCP, ASO, RF, CRP, IgG, IgA, IgM in rheumatoid arthritis (RA) and control groups; (d) correlation diagram of amino acids and clinical indicators of RA;

[0021] Figure 5 It is a correlation network diagram between the amino acid indicators used for modeling;

[0022] Figure 6 The changes of metabolites between the control and RA groups were analyzed for Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways;

[0023] Figure 7 Permutation test model of OPLS-DA model, control group (blue) and RA group (pink). DETAILED DESCRIPTION

[0024] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the technical scheme in the embodiments of the present invention will be described clearly and completely below. If the specific conditions are not specified in the embodiments, they are carried out according to conventional conditions or conditions recommended by the manufacturer. If the manufacturer of the reagents or instruments used is not specified, they are all conventional products that can be purchased commercially.

[0025] In order to evaluate the correlation between metabolite combinations and rheumatoid arthritis, the present invention collects samples from RA patients and healthy people who are known not to have RA, comprehensively analyzes the metabolomics of the samples, screens metabolites with significant differences in content between the two groups, and uses three methods to screen the indicators required to establish a diagnostic model by combining 21 AAs and >100 clinical indicators, thereby finding a marker combination suitable for the diagnosis of rheumatoid arthritis. Through extensive and in-depth research, the present invention has discovered markers related to RA, calcium, the ratio of arginine to ornithine, and lysine. Compared with healthy people who do not have RA, the content of the above marker combination in RA patients is significantly reduced. Compared with healthy people who do not have RA, the ratio of phenylalanine to tyrosine and the level of anti-cyclic citrullinated peptide antibodies (anti-CCP) are significantly increased.

[0026] It will be appreciated by those skilled in the art that any method known in the art may be used to determine the level of a metabolite or marker, such as chromatography, spectroscopy, mass spectrometry, or a combination thereof, and mass spectrometry is particularly preferred. Chromatography may include GC, LC, HPLC, and UHPLC; spectroscopy may include UV / Vis, IR, and NMR; mass spectrometer / spectroscopy may include ESI-QqQ, ESI-QqTOF, MALDI-QqQ, MALDI-QqTOF, and MALDI-TOF-TOF. Among them, GC is gas chromatography, CE is capillary electrophoresis, LC is liquid chromatography, HPLC is high performance liquid chromatography, UHPLC is ultra-high performance liquid chromatography, UV-Vis is ultraviolet visible, IR is infrared, NIR is near infrared, NMR is nuclear magnetic resonance, ESI is electron spray ionization, MALDI is matrix-assisted laser desorption / ionization, TOF is time of flight, and APCI is atmospheric pressure chemical ionization.

[0027] The term "area under the curve" or "AUC" refers to the area under the curve of the receiver operating characteristic (ROC) curve, both of which are well known in the field. The AUC measurement is useful for comparing the accuracy of classifiers across the entire data range. A classifiers with higher AUC have a higher ability to correctly classify unknowns between two target groups (e.g., RA samples and healthy group samples without RA). The ROC curve is useful for describing the performance of a specific feature (e.g., any biomarker described in the present invention and / or any item of additional biomedical information) when distinguishing between two populations (e.g., individuals with RA and healthy groups without RA). Typically, feature data is selected across the entire population (e.g., cases and controls) in ascending order based on the value of a single feature. Then, for each value of the feature, the true positive and false positive rates of the data are calculated. The true positive rate is determined by counting the number of cases with a value higher than the feature and dividing by the total number of cases. The false positive rate is determined by counting the number of controls with a value higher than the feature and dividing by the total number of controls. Although the definition refers to situations where a feature is elevated in cases compared to controls, the definition also applies to situations where a feature is low in cases compared to controls (in which case samples below the value of the feature will be counted). 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 mathematically combined (e.g., added, subtracted, multiplied, etc.) to provide a single sum value, and the single sum value can be plotted in an ROC curve. In addition, any combination of multiple features where the combination is derived from individual output values ​​can be plotted in an ROC curve.

[0028] The term "data storage medium" includes data storage media based on a single physical entity such as a CD, CD-ROM, hard disk, optical storage medium or magnetic disk, or a cloud disk. In addition, the term also includes data storage media composed of physically separate entities, which are operatively connected to each other in a manner that provides the above-mentioned data collection, preferably in a suitable manner for query search.

[0029] The "system" in the present invention relates to different tools that are operatively connected to each other. The tools may be implemented in a single device or may be physically separate devices that are operatively connected to each other. The tool for comparing metabolite characteristic values ​​preferably operates based on an algorithm for comparison. The data storage medium preferably comprises the above-mentioned data set or database, wherein each group of stored data sets indicates RA or its susceptibility. Therefore, the system of the present invention allows identification of whether the data set stored in the data storage medium contains a test data set. As a result, the system of the present invention can be used as a diagnostic tool for diagnosing RA or its susceptibility. In a preferred embodiment of the system, a tool for determining the characteristic value of a sample metabolite is included.

[0030] In a first aspect, the present invention provides an application of a reagent for detecting diagnostic markers in the preparation of a diagnostic product for rheumatoid arthritis, wherein the diagnostic markers include: calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine, anti-cyclic citrullinated peptide antibodies (anti-CCP) and lysine.

[0031] The present invention uses liquid chromatography-mass spectrometry (LC-MS) to detect the concentration changes of 21 AA (amino acids) in the serum of RA patients. In addition, by combining 21 AA and >100 clinical indicators, 5 indicators required for establishing a diagnostic model are screened out, and a variety of machine learning methods (such as random forest, K nearest neighbor, logistic regression, Gaussian Bayes, support vector machine and AdaBoost) can be used according to the screened indicators to establish and verify the diagnostic model of RA disease. When diagnosing rheumatoid arthritis, the concentration or content of the diagnostic marker of the sample to be tested is input, and the diagnostic result is output based on the diagnostic model.

[0032] Compared with 8 rheumatoid-specific clinical indicators (ESR, anti-CCP, ASO, RF, CRP, IgG, IgA and IgM), the 5 diagnostic markers screened out in the present invention have higher diagnostic sensitivity, specificity and accuracy.

[0033] In the present invention, a variety of methods were used to screen important indicators of RA, and the results showed that Arg, Phe and Lys were among the most important indicators of RA. In addition, they were associated with various clinical indicators of RA, indicating that they may play an important role in the occurrence and development of RA. Arginine is a non-essential basic amino acid that usually relies on its own synthesis in the body. However, under pathological conditions, arginine synthesis may be limited, which reduces the translation and activity of cell cycle proteins and other proteins, thereby inhibiting T cell activation. Arg is the only substrate for NO synthesis, which affects immunity and exerts a regulatory effect by regulating NO. The role of NO in immune cells includes activating T lymphocytes, natural killer cells and peripheral blood monocytes, regulating the synthesis and release of related cytokines, such as TNF-α and IL-1β, and promoting related immune responses, thereby exacerbating the inflammatory response of RA. As reported in previous studies, as synovial inflammation in RA patients increases, arginine levels gradually decrease, so arginine levels can be used as an indirect basis for assessing the degree of inflammation. Given the versatility of arginine and its importance in health and disease, arginine metabolism is very likely to be involved in the molecular mechanism of RA pathogenesis.

[0034] Phenylalanine is an essential amino acid that is converted to tyrosine by downstream phenylalanine under normal aerobic conditions. Studies based on serum biochemical analysis have shown that there is a correlation between inflammation and reduced phenylalanine conversion, and the potential mechanism of this change is due to reduced phenyl-4-hydroxylase activity caused by oxidative stress under immune activation. Inflammation significantly increases serum Phe levels (hyperphenylalaninemia), which is also observed under infectious inflammatory conditions, and reduced serum Phe is a characteristic of RA patients. This suggests that there is a negative metabolic conversion during infection-induced inflammation. However, research on another indicator in RA, Lys, is still limited. The present invention shows that lysine may play an important role in the occurrence and development of RA through correlation analysis between lysine, RF and Ca.

[0035] In a preferred embodiment of the present invention, the application includes: detecting the level of a diagnostic marker in a sample to be tested. The level refers to the content or concentration of the diagnostic marker.

[0036] In a preferred embodiment of the present invention, targeted or non-targeted chromatography, spectroscopy, mass spectrometry, ELISA or immunoturbidimetry is used to detect the content of anti-CCP in the sample to be tested; targeted or non-targeted chromatography, spectroscopy, mass spectrometry or a combination thereof is used to detect calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine and the level of lysine.

[0037] In an optional embodiment, the anti-CCP content in the test sample is detected by ELISA or immunoturbidimetry, or the anti-CCP content in the test sample is obtained by LC-MS.

[0038] In a preferred embodiment of the present invention, the sample to be tested is blood, plasma or serum.

[0039] In a preferred embodiment of the present invention, the rheumatoid arthritis diagnostic product is a kit.

[0040] In a preferred embodiment of the present invention, the reagent for detecting the diagnostic marker is selected from the reagent for detecting the anti-CCP level, the reagent for detecting calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine and the level of lysine, and the reagent is selected from the reagent for chromatography, spectroscopy, mass spectrometry, ELISA or immunoturbidimetry. For example, the reagent for detecting the anti-CCP level includes but is not limited to: CCP protein, detection antibody with a marker.

[0041] In a preferred embodiment of the present invention, the reagent for detecting the diagnostic marker is selected from the group consisting of LC-MS reagents for detecting calcium, arginine to ornithine ratio, phenylalanine to tyrosine ratio, anti-cyclic citrullinated peptide antibody (anti-CCP) and lysine level.

[0042] In a second aspect, the present invention also provides a rheumatoid arthritis diagnostic reagent, which includes reagents for detecting calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine, anti-cyclic citrullinated peptide antibodies (anti-CCP) and lysine levels.

[0043] In a preferred embodiment of the present invention, the sample is blood, plasma or serum.

[0044] In a third aspect, the present invention also provides a rheumatoid arthritis diagnostic kit, which comprises the above-mentioned rheumatoid arthritis diagnostic reagent.

[0045] In a fourth aspect, the present invention also provides the use of diagnostic markers in constructing and / or evaluating a diagnostic model for rheumatoid arthritis, wherein the diagnostic markers include: calcium, a ratio of arginine to ornithine, a ratio of phenylalanine to tyrosine, anti-cyclic citrullinated peptide antibodies (anti-CCP), and lysine;

[0046] Preferably, any one of the following machine learning classification algorithms is used to construct a rheumatoid arthritis diagnostic model: random forest, K nearest neighbor, logistic regression, Gaussian Bayes, support vector machine and AdaBoost.

[0047] In a fifth aspect, the present invention also provides a system or device for diagnosing rheumatoid arthritis, which comprises: (1) a unit for determining a diagnostic marker combination characteristic value in a sample; (2) a unit for comparing the diagnostic marker combination characteristic value in a sample; (3) a data storage medium, wherein the diagnostic marker combination characteristic value is the content or concentration of the diagnostic marker; the diagnostic markers include: calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine, anti-cyclic citrullinated peptide antibodies (anti-CCP) and lysine.

[0048] The features and performance of the present invention are further described in detail below in conjunction with the embodiments.

[0049] The present invention uses liquid chromatography-mass spectrometry (LC-MS) to detect the concentration changes of 21 AAs in the serum of RA patients. In addition, by combining 21 AAs and >100 clinical indicators, three methods, recursive feature elimination and cross-validation anomaly detection algorithm (RFECV-ADA), recursive feature elimination and logistic regression (RFE-LR), and mutual information method (MI), are used to screen the indicators required for establishing a diagnostic model, and six machine learning methods are used to establish and verify the diagnostic model of RA disease based on the screened indicators. Finally, by examining the correlation between AAs and clinical indicators of RA disease, the present invention also provides insights into the development of RA metabolic mechanisms and diagnosis.

[0050] Example 1

[0051] This embodiment provides a method for constructing an RA diagnostic model.

[0052] 1. Materials and Methods

[0053] 1.1. Participants

[0054] Patients included in this study were treated at Xijing Hospital (Xi'an, China) from July 1, 2023 to December 2, 2024 and were divided into two groups: RA (n = 207) and control group (n = 152). All RA patients met the 2010 and 1987 criteria of the American College of Rheumatology. The included RA patients included those who had not received any steroids, immunosuppressive drugs, or antibiotics in the past three months. Participants in the control group were healthy and had no diseases. Exclusion criteria were: (1) receiving immunosuppressive therapy or having a history of immunosuppressive therapy, and (2) receiving hormone therapy or having a history of hormone therapy.

[0055] 1.2. Blood preparation

[0056] After an overnight fast, venous blood was collected intravenously by experienced nurses and processed according to the requirements of AA, biochemistry, immunology, and routine blood tests.

[0057] 1.3 Data Collection

[0058] All tests are completed in the clinical laboratory of Xijing Hospital, the data are uploaded and stored in the LIS system, and the basic information of the patients is counted and entered by trained personnel. All test results are quality controlled before, during and after the test to ensure quality and accuracy.

[0059] 1.4 LC-MS / MS Materials and Equipment

[0060] HPLC-MS / MS was used to analyze 21 amino acids [alanine (Ala), arginine (Arg), aspartic acid (Asp), citrulline (Cit), cysteine ​​(Cys), glutamine (Gln), glutamic acid (Glu), glycine (gly), homocysteine ​​(Hcy), histidine (His), leucine (Leu), lysine (Lys), methionine (Met), ornithine (Orn), phenylalanine (Phe), proline (Pro), serine (Ser), threonine (Thr), tryptophan (Trp), tyrosine (Try), and valine (Val)]. Standard AA was labeled (Panel A, Cambridge Isotope Laboratories) and conjugated with 15N13C Gly, d4 Ala, d4C13 Arg, d3Asp, d2 Cit, d3 Glu, d3 Leu, d3 Met, d2 Orn, 13C6-Phe, 13C6-Tyr, and d8-Val (Cambridge Isotope Laboratories). All analyses were performed using LC-MS / MS.

[0061] 1.5 LC-MS / MS sample pretreatment

[0062] To detect the 21 AAs, 50 μL of serum was dropped onto a blank blood collection filter paper and allowed to penetrate completely. The 3.5 mm dried serum filter paper was removed with a stick and then extracted with 100 μL of isotope standard extracts of known concentration for 15 min. The extraction solution was dried under nitrogen. A hydrochloric acid n-butanol solution (3 mol / L) was prepared using a volume ratio of n-butanol to acetyl chloride of 9:1. Then, 60 μL of hydrochloric acid n-butanol was added to the dried sample and the derivatization reaction was carried out at 65 °C for 20 min. The derivatization solution was freeze-dried in N2 and then a 100-μL ACN solution was added as the detection solution. The mass spectra of the 21 substances tested on the serum chip were analyzed using a series of mass spectrometers and their concentrations were calculated.

[0063] 1.6 Mass spectrometry analysis

[0064] To detect 21 AAs, 100% ACN was used as the HPLC eluent, and the optimized HPLC parameters are listed in Table 1. The injection volume was 20 μL. Two experiments were performed in positive mode within one mass spectrometry cycle, and the optimized MS / MS parameters are listed in Table 2. In experiment 1, the scan type was as follows: neutral loss, loss at 102.00 Da, start at 140.00 Da, stop at 280.00 Da, CEP start at 13.07, stop at 17.46. In experiment 2, the scan type was multiple reaction monitoring mode, and the optimized parameters are shown in Table 3.

[0065] 2 Statistical analysis

[0066] Table 1: Gradient conditions of liquid chromatography mobile phase for detection of 21 amino acids

[0067]

[0068] Table 2: Mass spectrometry parameters for detecting 21 amino acids

[0069]

[0070] Table 3: MS parameters for the multiple reaction monitoring mode for the detection of 21 amino acids

[0071]

[0072] MS data were analyzed using Analyst 1.6.2 and ChemView software (version 1.6.1) (AB Sciex, Darmstadt, Germany). SPSS software (version 23.0; IBM, Armonk, NY, USA) was used for between-group analysis of quantitative data, using Fisher's exact test for categorical variables and Mann-Whitney U or t-test for continuous variables. Normally distributed quantitative data were expressed as mean ± SD and compared using one-way analysis of variance with P values ​​(statistical significance was set at P < 0.05). Graphical data were generated using GraphPad Prism (version 5; GraphPad Software, San Diego, CA, USA) and R version 3.6.2 (R Project for Statistical Computing). Orthogonal partial least squares discriminant analysis (OPLS-DA) models were used to analyze differences between the overall data of the RA group and the control group. Binary logistic regression analysis was used to evaluate whether AA and its ratio were risk factors for RA. Receiver operating characteristic (ROC) curves were used to evaluate and compare the diagnostic performance of the models. Kyoto Encyclopedia of Genes and Genomes (KEGG) was used to illustrate the differential biological functions and pathways enriched by the differentially expressed metabolites.

[0073] 2.1. Establishment of diagnostic model

[0074] In order to establish an effective diagnostic model, 40 indicators with the highest contribution values ​​were screened using RFECV-ADA, RFE-LR, and MI methods, and common indicators were considered as potential indicators for model building. In order to enhance and verify the effectiveness of the RA diagnostic model, six machine learning classification algorithms (random forest, K nearest neighbor, logistic regression, Gaussian Bayes, support vector machine, and AdaBoost) were used to build the model. The sensitivity, specificity, accuracy, positive predictive value, negative predictive value, area under the curve (AUC), and ROC curve of these models were evaluated. To ensure the stability of all algorithms, the data were cross-validated 10 times, 9 of which were used for training and 1 for testing.

[0075] result:

[0076] (1) Patient characteristics: By analyzing the basic information of the participants, including height, weight, blood pressure and living habits, as well as the expression levels and differences of all indicators, there were no significant differences in basic information between the RA group and the control group. Specifically, the RA group and the control group were similar in age, gender and BMI, with no significant differences; there were no significant differences in blood type, smoking and alcohol consumption; however, 7.48% of RA patients had hypertension, 29.41% of the control group had exercise habits, while patients in the RA group had no regular exercise habits, which may be related to joint pain in RA patients. Compared with the control group, there were significant differences in 19 AAs and their ratios in the RA group (14 indicators were significantly reduced, and 5 indicators were significantly increased), and there were significant differences in 73 clinical indicators.

[0077] (2) Distribution of measurement indicators in each group

[0078] The OPLS-DA model was used to analyze and compare the overall differences of all indicators between the two groups ( Figure 1 ), and use the permutation test to verify the validity of the model. Figure 7 As shown in Figure 2, the model effectively distinguished the RA group from the control group (R2Y=0.362, Q2=-0.364). Figure 1 and Figure 7 As shown in the figure, there were overall differences in all indicators between the RA group and the control group. The OPLS-DA model can effectively distinguish the RA group from the control group.

[0079] (3) Metabolic pathway

[0080] The metabolism of RA patients changes significantly with the disease. Differential metabolic pathways corresponding to significantly different metabolites were identified through enrichment and pathway analysis (predictor variable importance > 1.0). Pathway analysis comparing KEGG differences between women with RA and control groups showed that metabolic pathways such as arginine biosynthesis; glutathione metabolism; nitrogen metabolism; valine, leucine, and isoleucine biosynthesis; and arginine and proline metabolism were the most differentiated pathways ( Figure 6 ).

[0081] Metabolite-metabolite interaction networks highlight the potential functional relationships between various annotated metabolites. Chemical associations of metabolite networks were extracted. Figure 5 , L-glutamate was the most contributing metabolite, while L-lysine, L-cysteine, L-glutamine, and L-arginine had the highest mediators. All five metabolites significantly contributed to the abnormal metabolic pathways in RA patients.

[0082] (4) Performance evaluation of candidate indicators based on classification algorithms.

[0083] According to the difference in the expression levels of AA and clinical indicators, three methods (RFECV-ADA, RFE-LR, and MI) were used to cross-screen the top 40 indicators with the largest differences, and potential biomarkers for RA and control groups were extracted for modeling. Five markers [calcium (Ca), arginine / ornithine, anti-CCP, lysine, and phenylalanine / tyrosine] were selected as candidate markers to establish a diagnostic model for RA (V5) (p < 0.0001) ( Figure 2 (a)). The Shapley Additive Explanation (SHAP) model of the five indicators refers to Figure 2 As shown in (b), Ca contributed the most to the model, followed by arginine, lysyl, anti-CCP-, and phenylalanine / tyrosine. The effectiveness of the diagnostic model based on the five candidate indicators was examined using six classifier algorithms. The ROC, sensitivity, specificity, and accuracy of the V5 model obtained in the training set were >0.971, 89.50%, 94.28%, and 91.97%, respectively ( Figure 3 (a).

[0084] The ROC curve, sensitivity, specificity, and accuracy of the V5 model validated using the test set were >0.937, 84.89%, 89.78%, and 87.47%, respectively ( Figure 3 (b)

[0085] Example 2

[0086] This embodiment provides a method for diagnosing RA, which specifically includes the following steps:

[0087] (1) LC-MS was performed on 35 serum samples according to the method of Example 1 to obtain the levels of calcium (Ca), arginine / ornithine, anti-CCP, lysine, and phenylalanine / tyrosine;

[0088] (2) Then, the parameters of step (1) are substituted into the diagnostic model of RA (V5) constructed in Example 1, and the results are outputted using the logistic regression model as an example.

[0089] RA was diagnosed according to the following criteria.

[0090] The cut-off value of the RA (V5) diagnostic model was 0.50. A value > 0.50 was diagnosed as RA, and a value < 0.50 was diagnosed as non-RA (healthy controls).

[0091]

[0092] Comparative Example 1

[0093] This comparative study established an effective diagnostic model by screening AA and common clinical indicators without the need for rheumatoid-specific indicators. The top 40 indicators with the largest differences were cross-screened using three methods: RFECV-ADA, RFE-LR, and MI, and potential biomarkers in the RA group and control group were extracted for modeling. 19 indicators (Arg / Orn, Val / Phe, Val, Orn / Cit, Arg, Cys, Gln, Glu, Glucose, Glu / Cit, Orn, EC, Cit / Arg, Leu, SG, UBG, Pip, G, and HGB) were selected as candidates for establishing the RA diagnostic model (V19) (p<0.0001) ( Figure 4 (a).

[0094] Six classification algorithms were used to test the effectiveness of the diagnostic V19 model established using 19 candidate indicators. The ROC, sensitivity, specificity, and accuracy obtained in the V19 model established using the training set were >0.986, 93.94%, 94.30%, and 95.03%, respectively ( Figure 3 The ROC curve, sensitivity, specificity, and accuracy of the V19 model validated using the test set were >0.936, 84.89%, 89.78%, and 87.47%, respectively ( Figure 3 (middle f).

[0095] Comparative Example 2

[0096] Eight rheumatoid-specific clinical indicators (ESR, anti-CCP, ASO, RF, CRP, IgG, IgA, and IgM) were used to establish a diagnostic model (V8). Six classification algorithms were used to examine and validate the effectiveness of the diagnostic V8 model established using these indicators. The ROC, sensitivity, specificity, and accuracy of the V8 model established using the training set were >0.896, 87.21%, 72.84%, and 81.73%, respectively ( Figure 3 The ROC, sensitivity, specificity, and accuracy of the V8 model validated using the test set were >0.838, 80.67%, 72.67%, and 77.55%, respectively ( Figure 3 (d)

[0097] Experimental Example 1

[0098] The optimal number of metrics required to build a diagnostic model for RA was examined using rfecv_ada; Figure 4 As shown in Figure a, the optimal number of five indicators contributes the most to the construction of the model. After more than five indicators, the cross-validation score of the constructed model does not increase further. Therefore, this experiment focuses on the diagnostic models with 5 (V5) and 8 (V8) indicators, as shown in Figure 2. Figure 4The decision curve analysis results of V5 and V8 show that when the high risk threshold is > 0.2, the net return rate of the V5 diagnostic model is higher than that of the V8 diagnostic model ( Figure 4 (b) Comparison of ROC of V5, V8, ESR, anti-CCP, ASO, RF, CRP, IgG, IgA and IgM ( Figure 4 The curve in middle c) shows that the AUCs of V5, V8 and the clinically used rheumatoid-specific single marker are greater than 0.995, 0.964 and 0.502, respectively.

[0099] Experimental Example 2

[0100] AA and its ratio serve as independent predictors of RA.

[0101] To further explore the potential value of AA and its ratio in RA, binary logistic regression modeling was performed to identify indicators for predicting the occurrence of RA (Table 4). 16 amino acids and their ratios (Arg, Cys, Val, Lys, Tyr, Arg / Orn, Val / Phe, Leu / Phe, Tyr / Leu, Asp, Glu / Cit, Orn / Cit, Gly / Ala, Cit / Arg, Met / Phe, and Met / Leu) were identified as risk factors for RA. Elevated levels of Arg, Cys, Val, Lys, Tyr, Arg / Orn, Val / Phe, Leu / Phe, and Tyr / Leu were associated with an increased risk of RA. In addition, elevated levels of Asp, Glu / Cit, Orn / Cit, Gly / Ala, Cit / Arg, Met / Phe, and Met / Leu were negative risk factors for the development of RA.

[0102] Table 4: Logistic regression model to assess risk factors for RA

[0103]

[0104] Experimental Example 3

[0105] Correlation of AA with clinical manifestations of the disease.

[0106] ESR, anti-CCP, ASO, RF, CRP, IgG, IgA and IgM levels are closely related to the occurrence and development of RA. Pearson correlation coefficient and P value were used to evaluate the relationship between AA and RA.

[0107] The correlation between relevant clinical indicators was used to further explore the role of AA in the onset and progression of RA ( Figure 4 (d)

[0108] Pearson correlation coefficient analysis showed that in the RA disease group, seven AAs and their ratios were significantly correlated with ESR (Val / Phe: 0.368**, 0; Asp: 0.317**, 0.001; Phe / Tyr: 0.333**, 0.001, Phe: 0.309**, 0.002; Leu / Ala: 0.234*, 0.021; Leu / Phe: -0.262**, 0.009; Met / Phe: -0.317**, 0.001). Three amino acids and their ratios were significantly correlated with anti-CCP activity (Val / Phe: -0.232*, 0.025; Asp: 0.216*, 0.037; Gly / Ala: 0.248*, 0.016). Five amino acids and their ratios were significantly correlated with RF (Cys: 0.243*, 0.016; Asp: 0.206*, 0.042; Lys: 0.230*, 0.023; Phe / Tyr: 0.248*, 0.014; Phe: 0.245*, 0.015). Six amino acids and their ratios were significantly correlated with CRP (Val / Phe: -0.308*, 0.025; Asp: 0.232*, 0.025, Gly / Ala: 0.216*, 0.037). (Valine / Phe: -0.303**, 0.002; Aspartic acid: 0.376**, 0; Phe: 0.247*, 0.014; Ser: 0.203*, 0.045; Leucine / Alanine: 0.285**, 0.004; Methionine / Phe: -2.252*, 0.012). Two AAs and their ratios were significantly correlated with IgG (Arg / Orn: -0.302**, 0.007; His: 0.245*, 0.029). These three AAs and their ratios were significantly correlated with IgA (Cys: 0.344**, 0.002; Asp: 0.223*, 0.047; His: 0.287**, 0.01). Two AA and their ratios were significantly correlated with IgM (Pip: -0.221*, 0.049; Thr: -0.239*, 0.033). Eight amino acids and their ratios were significantly correlated with calcium (Cys: -0.422**, 0.005; Orn / Cit: -0.413**, 0.007; Lys: -0.327*, 0.035; Cit: 0.449**, 0.001. Cit / Arg: 0.510**, 0.001; Cit / Phrase: 0.412**, 0.007; Gln / Cit: -0.380*, 0.013; Thr / Cit: -0.358*, 0.02).

[0109] In summary, based on the important role of AA in the occurrence and development of RA, the present invention uses LC-MS technology to target 21 AAs, further explore the important position of AA in RA patients, and combine it with a variety of machine learning methods to establish and compare the effectiveness of several diagnostic models to further develop the clinical diagnosis of RA. In order to explain which model is more suitable for the diagnosis of RA and provide the possibility for direct clinical diagnosis, the present invention uses serological indicators. Three diagnostic models were established: the diagnostic model established by the V5 method uses AA and RA-specific serological markers, with a total of five indicators; 6 RA-specific serological markers were used in the diagnostic model established by the V8 method; 19 common clinical indicators, including AA and other non-RA-specific indicators, were used to develop the V19 diagnostic model.

[0110] Among them, V19 of Comparative Example 1 showed the highest diagnostic efficiency. Although all commonly used clinical indicators were selected, the inclusion of too many indicators would bring economic burden to patients. However, it uses AA and other common clinical indicators to provide doctors with early warning of RA before the diagnosis of RA is suspected. V5 in Example 1 uses the least indicators but provides excellent diagnostic performance; therefore, this diagnostic method is more conducive to the clinical diagnosis of RA. V8 of Comparative Example 2 uses six RA-specific serological markers to simulate RA. Although the diagnostic efficiency of this model is not as good as that of V19 and V5, it improves the diagnostic efficiency of traditional single indicators and increases the ROC of a single indicator from 0.502 to >0.838. This model can be used in hospitals that do not have the conditions to detect AA.

[0111] The RA diagnostic model established by the present invention has a good diagnostic effect, and different hospitals can adopt appropriate diagnostic models according to their actual conditions. However, V5 is still the preferred RA diagnostic model because it uses the least indicators and has excellent diagnostic performance.

[0112] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. Use of a diagnostic marker in the preparation of a diagnostic product for rheumatoid arthritis, characterized in that: The diagnostic markers include: calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine, anti-cyclic citrullinated peptide antibodies and lysine; the sample to be tested is blood, plasma or serum.

2. The use according to claim 1, characterized in that: The applications include: The content of anti-cyclic citrullinated peptide antibodies in the sample to be tested is detected by targeted or non-targeted chromatography, spectroscopy, mass spectrometry, ELISA or immunoturbidimetry; the ratio of calcium, arginine to ornithine, the ratio of phenylalanine to tyrosine and the level of lysine are detected by targeted or non-targeted chromatography, spectroscopy, mass spectrometry or a combination thereof.

3. The use according to claim 2, characterized in that: The rheumatoid arthritis diagnostic product is a kit.

4. The use according to claim 2, characterized in that: The reagents for detecting diagnostic markers are selected from reagents for detecting anti-cyclic citrullinated peptide antibody levels, calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine and the level of lysine, and the reagents are selected from reagents for chromatography, spectroscopy, mass spectrometry, ELISA or immunoturbidimetry.

5. The use according to claim 4, characterized in that: The reagents for detecting diagnostic markers are selected from reagents for detecting calcium, arginine to ornithine ratio, phenylalanine to tyrosine ratio, anti-cyclic citrullinated peptide antibodies and lysine levels by LC-MS.

6. Use of a diagnostic marker in constructing and / or evaluating a diagnostic model for rheumatoid arthritis, characterized in that: The diagnostic markers include: calcium, arginine to ornithine ratio, phenylalanine to tyrosine ratio, anti-cyclic citrullinated peptide antibodies, and lysine; The rheumatoid arthritis diagnostic model was constructed using any of the following machine learning classification algorithms: random forest, K nearest neighbor, logistic regression, Gaussian Bayes, support vector machine, and AdaBoost.

7. A system for diagnosing rheumatoid arthritis, characterized in that: It comprises: (1) a unit for determining a diagnostic marker combination characteristic value in a sample; (2) a unit for comparing the diagnostic marker combination characteristic value in a sample; and (3) a data storage medium, wherein the diagnostic marker combination characteristic value is the content or concentration of the diagnostic marker; the diagnostic markers include: calcium, the ratio of arginine to ornithine, the ratio of phenylalanine to tyrosine, anti-cyclic citrullinated peptide antibodies, and lysine.

Images