Marker combination for predicting activity state of rheumatoid arthritis disease, model construction method and application

Abstract

The invention belongs to the technical field of biological medicines, and particularly relates to a marker combination for predicting the activity state of rheumatoid arthritis diseases, a model construction method and application. The marker combination is composed of PLA2G2A and CRP (C-reactive protein). The marker combination for predicting the activity state of the rheumatoid arthritis disease overcomes the defects that existing diagnostic indexes are tedious in operation and large in subjective human factor interference, has the advantages of being easy and convenient to operate and more objective in result, and has good clinical application value. The invention further provides a PLA2G2A and CRP combined regression diagnosis model, the PLA2G2A and CRP combined regression diagnosis model can be used for identifying the disease activity state of a rheumatoid arthritis patient, the sensitivity is 85%, the specificity is 85%, when the diagnosis model value is larger than 5.667, the patient is in the rheumatoid arthritis activity period, and if the diagnosis model value is smaller than 5.667, the patient is in the rheumatoid arthritis remission period.

Description

Technical Field

[0001] The present invention belongs to the field of biomedical technology, and particularly relates to a biomarker combination for predicting the disease activity status of rheumatoid arthritis, a model construction method and applications thereof. Background Art

[0002] Rheumatoid arthritis (RA) is a chronic inflammatory autoimmune disease of the joints that persists locally. The main pathological manifestation is the abnormal hyperplasia of the synovium of the joints, which invades cartilage and bone tissues, leading to joint destruction. It can cause multiple organs of the body to be involved. The clinical manifestations of RA are diverse, and the pathogenesis and clinical manifestations among individuals are not exactly the same. Common clinical manifestations include morning stiffness, pain, joint swelling and joint deformity, and systemic symptoms such as weight loss, low fever and fatigue can also occur, as well as complications such as osteoporosis, subcutaneous nodules, interstitial lung disease, pericarditis, pleurisy and rheumatoid vasculitis.

[0003] The repeated activity of rheumatoid arthritis is the fundamental cause of the continuous progression of the disease to joint destruction. Its treatment mainly focuses on achieving the standard treatment for evaluating the inflammatory reaction activity of rheumatoid arthritis, that is, corresponding drugs are given according to whether the patient is active and the degree of activity. The efficacy evaluation of drug use is also based on the standard treatment, that is, the clinical efficacy of the drug is evaluated according to whether the patient's disease is active. It can be seen that the disease activity status of rheumatoid arthritis plays an important role in the occurrence, development and treatment of rheumatoid arthritis.

[0004] Currently, the Disease Activity score (DAS) used to evaluate the disease activity status of rheumatoid arthritis is calculated by a formula based on the swelling and tenderness of 28 joints in the whole body of rheumatoid arthritis patients, combined with the clinical inflammatory indicators C-reactive protein and erythrocyte sedimentation rate (ESR). Affected by factors such as the patient's self-pain evaluation index and different pain thresholds of the patient, there are great individual differences. Its key parameters C-reactive protein or erythrocyte sedimentation rate are non-specific inflammatory indicators and are also affected by other inflammatory reactions.

[0005] Chinese Patent Publication No. CN118460696A discloses a rheumatoid arthritis early screening kit using HERV:MLT2B2 as a biomarker. By quantitatively analyzing the expression level of HERV:MLT2B2 in peripheral blood mononuclear cells (PBMCs), it is found that the expression of the endogenous retrovirus HERV:MLT2B2 is significantly up-regulated in the PBMCs of rheumatoid arthritis patients. At the same time, it is found that the expression of HERV:MLT2B2 is closely related to disease indicators such as the activity of rheumatoid arthritis patients, and the ROC curve predicts that HERV:MLT2B2 can be used for the diagnosis of early RA.

[0006] Another Chinese invention patent publication number CN118624916A discloses a biomarker for the diagnosis of rheumatoid arthritis and its application. The biomarker includes HIF-1α and / or SLC35F5; using the HIF-1α / SLC35F5 signaling pathway, the expression of both HIF-1α / SLC35F5 pathways in fibroblast-like synoviocytes (RAFLS) of synovial tissues of RA patients is significantly up-regulated compared with that of normal people. Both HIF-1α and SLC35F5 in RAFLS cells can be used as biomarkers for evaluating early diagnosis and disease activity of RA, and can also be used as potential new therapeutic targets.

[0007] However, the above-mentioned invention patent only provides markers for the early screening of rheumatoid arthritis. At present, there is no specific clinical objective evaluation index for the disease activity of rheumatoid arthritis for clinical use. Summary of the Invention

[0008] In view of the deficiencies of the prior art, the present invention provides a marker combination for predicting the disease activity status of rheumatoid arthritis, a model construction method and an application.

[0009] In order to achieve the above object of the present invention, the specific technical solution adopted by the present invention is as follows:

[0010] A marker combination for predicting the disease activity status of rheumatoid arthritis, the marker combination is composed of PLA2G2A and CRP.

[0011] Preferably, the activity status is the active stage and the remission stage of rheumatoid arthritis.

[0012] The present invention also provides the application of the above-mentioned marker combination in constructing a model for predicting the disease activity status of rheumatoid arthritis.

[0013] The present invention also provides a model for predicting the disease activity status of rheumatoid arthritis, which is constructed by the above-mentioned marker combination.

[0014] The present invention also provides a method for constructing the above-mentioned model for predicting the disease activity status of rheumatoid arthritis, including the following steps:

[0015] (1) Collect peripheral blood samples of healthy subjects, patients with active rheumatoid arthritis and remission stage, and perform proteomic analysis to obtain significantly different proteins;

[0016] (2) Randomly divide the peripheral blood samples into a training set and a test set in equal proportions, and use the random forest machine learning method to perform discriminant analysis on the significantly different proteins to obtain single indicators of the markers;

[0017] (3) Verify the protein expression levels of individual indicators, evaluate the diagnostic efficacy of the analysis equations obtained for individual indicators and combined indicators, determine the optimal equation, and construct a prediction model for the disease activity status of rheumatoid arthritis.

[0018] Preferably, the markers described in step (2) are PLA2G2A and CRP.

[0019] Preferably, the evaluation process in step (3) includes: using the standardized CRP and PLA2G2A protein concentration values as independent variables, and the active stage of rheumatoid arthritis and the remission stage of rheumatoid arthritis as dependent variables, and establishing three models respectively by logistic regression analysis: only CRP, only PLA2G2A, and the combination of both, and obtaining the AUC value, the optimal threshold, sensitivity, and specificity by using the ROC curve analysis method.

[0020] Preferably, the prediction model for the disease activity status of rheumatoid arthritis in step (3) is X = 0.506 + 1.831×A + 1.617×B;

[0021] where X is the model value, A is the standardized value of the CRP protein concentration in the subject's plasma, and B is the standardized value of the PLA2G2A protein concentration in the subject's plasma.

[0022] Preferably, when X is greater than 5.667, the subject is in the active stage of rheumatoid arthritis; when X is less than 5.667, the subject is in the non-active stage of rheumatoid arthritis.

[0023] The present invention also provides the application of the above prediction model for the disease activity status of rheumatoid arthritis or the prediction model for the disease activity status of rheumatoid arthritis constructed by the above construction method in the preparation of products for predicting the disease activity status of rheumatoid arthritis.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] (1) The present invention provides a combination of markers for predicting the disease activity status of rheumatoid arthritis, which is composed of PLA2G2A and CRP, overcomes the defects of cumbersome operation and large interference of subjective human factors in current diagnostic indicators, has the advantages of simple operation and more objective results, and has good clinical application value.

[0026] (2) The present invention also provides a combined regression diagnostic model of PLA2G2A and CRP, which can be used to identify the disease activity status of rheumatoid arthritis patients. Its sensitivity is 85%, and the specificity is 85%. When the diagnostic model value is greater than 5.667, the patient is in the active stage of rheumatoid arthritis. If it is less than 5.667, the patient is in the remission stage of rheumatoid arthritis, providing new ideas and means for the clinical accurate judgment of the disease activity status and treatment medication of RA patients. Description of the Drawings

[0027] Figure 1 Results of differential protein identification and analysis of peripheral blood proteome in the three groups of ACT, REM, and Ctrl.

[0028] Figure 2 Differential protein expression levels between the ACT group and the REM group.

[0029] Figure 3 Differential protein expression levels between the ACT group and the Ctrl group.

[0030] Figure 4 Differential protein expression levels between the REM group and the Ctrl group.

[0031] Figure 5 Results of the diagnostic efficacy of significantly differential proteins analyzed by random forest.

[0032] Figure 6 Comparison of the relative abundances of PLA2G2A protein in peripheral blood among the three groups of ACT, REM, and Ctrl (compared with the Ctrl group, **** indicates p < 0.0001).

[0033] Figure 7 Comparison of the relative abundances of PLA2G2A protein in peripheral blood between the ACT and REM groups (compared with the ACT group, ** indicates p < 0.01).

[0034] Figure 8 Comparison of the relative abundances of CRP protein in peripheral blood between the ACT and REM groups (compared with the ACT group, ** indicates p < 0.01).

[0035] Figure 9 ROC curve of the combined analysis of PLA2G2A and CRP. Detailed Implementation Modes

[0036] The present invention will be further described in detail below in combination with specific embodiments. The following embodiments are not used to limit the present invention, but only to illustrate the present invention. The experimental methods used in the following embodiments, unless otherwise specified, and the experimental methods without specific conditions noted in the embodiments are usually in accordance with conventional conditions. The materials, reagents, etc. used in the following embodiments, unless otherwise specified, can be obtained from commercial channels.

[0037] Example

[0038] 1. Research subjects: 80 patients with active RA (ACT), 80 patients with remission RA (REM), and 40 healthy individuals (Ctrl). The basic characteristics of the blood proteome detection population are shown in Table 1. Plasma from RA patients and healthy individuals was obtained from outpatients at Shanghai Guanghua Hospital of Integrated Traditional Chinese and Western Medicine. All enrolled individuals signed informed consent forms. The disease activity of RA was defined by the Disease Activity Score 28 (DAS28).

[0039] Table 1 Basic characteristics of the detection population

[0040] Item Ctrl (N = 40) REM (N = 80) ACT (N = 80) pvalue Age 54.83(31-75) 56.23(35-74) 58.75(37-75) 0.127 Gender (female / male) 29，11 66，14 66，14 2.039 BMI 22.85±2.61 23.08±2.96 22.83±3.50 0.858 Fasting blood glucose 5.42(4.33-8.14) 5.38(3.91-10.40) 5.36(4.16-8.92) 0.664 Total cholesterol 5.37±1.03 4.96±1.12 4.88±1.01 0.053 Triglyceride 1.34±0.63 1.38±0.65 1.38±1.52 0.260

[0041] Note: The p value is the difference in the comparison between groups.

[0042] 2. Research methods:

[0043] (1) Peripheral blood was collected from patients with active RA (ACT), patients with remission RA (REM), and healthy individuals (Ctrl) respectively for proteomic detection and analysis.

[0044] Inclusion criteria: Meeting the 1987 ACR diagnostic criteria or the 2010 ACR / EULAR classification criteria, and meeting the DAS28-ESR scoring criteria; both RF and CCP antibodies are positive; aged between 18 and 65 years old; gender is not limited; patients who have signed the informed consent form and voluntarily participated in the study.

[0045] Exclusion criteria: Complicated with severe primary diseases in other systems such as cardiovascular and cerebrovascular, respiratory, digestive, and tumor; patients with other autoimmune diseases; patients with severe mental illnesses; patients considered by other researchers as not suitable for inclusion in the disease; patients currently participating in other clinical trials.

[0046] Data collection: General information, past medical history, drug allergy history, BMI, blood glucose, triglyceride, total cholesterol, CRP, ESR, and DAS28-ESR and other indicators of patients were collected; 3 mL of peripheral whole blood was collected from each enrolled person.

[0047] Sample processing: Whole blood was collected, left to stand at 4°C for 2 h, then centrifuged at 2500 rpm for 10 min, the plasma was collected, aliquoted into 1.5 mL EPP tubes, and stored in a -80°C refrigerator for future testing.

[0048] Detection method: Plasma from ACT, REM, and Ctrl patients meeting the criteria, 500 μL for each case, was transported at low temperature and uniformly sent to Shanghai Zhongke New Life Technology Co., Ltd. for protein detection and identification using the company's deep blood proteome detection platform.

[0049] (2) The obtained significantly differentially expressed proteins were identified and analyzed using the random forest machine learning method to obtain valuable discriminant proteins.

[0050] Random forest machine learning method: First, data preprocessing is performed, including loading data, handling missing values, encoding categorical variables, and dividing the number of ACT, REM, and Ctrl patients into training and validation sets at a ratio of 1:1; subsequently, a model is constructed by integrating multiple decision trees. Each tree randomly selects a subset of features during training and uses Bootstrap sampling (sampling with replacement) to generate samples to increase diversity and reduce the risk of overfitting; the decision tree grows independently until it reaches the preset depth or the minimum number of samples and then stops. The final model prediction result is determined by the voting (classification) or mean (regression) of all trees; then, the model performance is evaluated using metrics such as accuracy and F1 value, and key variables (CRP) are identified through feature importance analysis; finally, the hyperparameters are optimized through grid search or cross-validation to balance the model complexity and generalization ability, completing the entire process.

[0051] According to the random forest machine learning method, 22 valuable discriminant proteins were obtained, and the top 3 proteins were selected for further verification.

[0052] (3) The peripheral blood protein expression levels of ACT patients and REM patients were verified using the ELISA method.

[0053] Sample source and grouping: The levels of CRP and PLA2G2A in the peripheral blood of ACT and REM rheumatoid arthritis patients were detected by the ELISA method. Another 40 samples were reselected (from active and remission patients with matched gender, age, blood glucose, and blood lipid levels) to verify the peripheral blood protein expression levels, that is, 20 cases in each of the ACT group and the REM group.

[0054] Materials: The CRP detection kit (ml106583) was purchased from Enzyme-linked Biotechnology Co., Ltd.; the PLA2G2A detection kit (ELH-PLA2G2A) was purchased from RayBiotech Biotechnology Company.

[0055] Detection method: Refer to the kit operation instructions to detect the levels of CRP and PLA2G2A in the peripheral blood.

[0056] (4) The diagnostic efficacy of single indicators and combined analysis equations was evaluated using the ROC curve analysis method for the detection results to determine the optimal equation.

[0057] The ROC curve analysis method is as follows:

[0058] A. Data preprocessing: Perform Z-score normalization on CRP and PLA2G2A, that is, calculate the mean and standard deviation of each variable, and then perform the operation of (measured value - mean) ÷ standard deviation on each data point.

[0059] B. Perform binary logistic regression analysis: Use the normalized CRP and PLA2G2A as independent variables, and the active stage and remission stage of rheumatoid arthritis as dependent variables (which need to be converted to 0 and 1, active stage of rheumatoid arthritis = 1, remission stage of rheumatoid arthritis = 0). Three models are established respectively by logistic regression analysis: CRP only, PLA2G2A only, and the combination of both. The regression coefficient is measured for each model, and the corresponding predicted probability is calculated.

[0060] C. ROC curve analysis: Calculate the AUC value for each model, find the optimal threshold, and obtain the sensitivity and specificity.

[0061] D. Determination of the Cutoff value: Through ROC curve analysis, maximize the Youden index (sensitivity + specificity - 1), and obtain the Cutoff value after cleaning the data.

[0062] 3. Research results:

[0063] (1) The results of proteomics data analysis are as Figures 1 - 4 shown. The peripheral blood proteins in the ACT, REM, and Ctrl groups can be significantly distinguished ( Figure 1 ). A total of 506 significantly different proteins were obtained by comparing the ACT group with the REM group ( Figure 2 ), a total of 572 significantly different proteins were obtained by comparing the ACT group with the Ctrl group ( Figure 3 ), and a total of 87 significantly different proteins were obtained by comparing the REM group with the Ctrl group ( Figure 4 ).

[0064] (2) The results of the diagnostic efficacy of significantly different proteins analyzed by random forest are as Figure 5 shown, showing that the CRP, SSA2, and PLA2G2A proteins ranked among the top three significantly discriminant proteins respectively. However, since the SSA2 protein is mainly used for the diagnosis of Sjogren's syndrome and lupus syndrome diseases and has low specificity in RA diseases, it is discarded, and only the CRP and PLA2G2A proteins are selected as screening indicators for the next verification.

[0065] The comparison of the relative abundances of PLA2G2A protein in the plasma of each group is as Figure 6 shown. The results show that the expression level of PLA2G2A protein in the ACT group is significantly higher than that in the REM group and significantly higher than that in the Ctrl group, and the difference is statistically significant (p < 0.0001).

[0066] (3) The basic characteristics of the ELISA-verified population are shown in Table 2. The ELISA test results further verified the analysis results of the random forest. The expression level of PLA2G2A in the peripheral blood of the ACT group was significantly higher than that in the REM group (see Figure 7 ); the expression level of CRP in the peripheral blood of the ACT group was significantly higher than that in the REM group (see Figure 8 ).

[0067] (4) The ROC curve is as shown in Figure 9 . The results showed that the combined regression model of PLA2G2A and CRP could be used to distinguish whether RA patients were in the ACT or REM stage. The ACU of its combined diagnosis was 0.8825, indicating that the combination of PLA2G2A and CRP in plasma had good value for the diagnosis of rheumatoid arthritis; the sensitivity was 85%, and the specificity was 85%, with good specificity and high sensitivity; the Cutoff value was 5.667, that is, if the diagnostic model value was greater than 5.667, the patient was in the active stage of rheumatoid arthritis, and if it was less than 5.667, the patient was in the remission stage of rheumatoid arthritis.

[0068] The combined regression model of PLA2G2A and CRP is: X = 0.506 + 1.831×A + 1.617×B,

[0069] where X is the model value, A is the standardized value of the CRP concentration in the plasma of the subject, and B is the standardized value of the PLA2G2A concentration in the plasma of the subject.

[0070] The standardization formula is as follows:

[0071] CRP_std = (measured CRP value - 1.89) ÷ 2.19;

[0072] PLA2G2A_std = (measured PLA2G2A value - 6.51) ÷ 6.19.

[0073] Table 2 Basic characteristics of the ELISA-verified population

[0074] Item REM (N = 20) ACT (N = 20) pvalue Age 51.5±8.95 57.45±9.85 0.0528 Gender (female / male) 18 / 2 16 / 4 0.6614 BMI 23.55±2.80 22.67±3.38 0.3750 Fasting blood glucose 5.30±1.01 5.36(4.16-8.92) 0.3399 Total cholesterol 4.94±0.964 4.69±0.93 0.4124 Triglyceride 1.41±0.77 1.93±2.88 0.4405 CRP 2.65±4.29 27.66±35.47 0.0033 ESR 12.95±7.48 48.45±28.18 <0.0001 DAS28 - ESR 1.76±0.47 5.25±1.02 <0.0001

[0075] Note: The p value is the difference in the comparison between groups.

[0076] Comparative Example 1

[0077] The difference from the example is only that sPLA2-IIA is substituted for PLA2G2A in the combined regression model of PLA2G2A and CRP to construct a prediction model for the disease activity status of rheumatoid arthritis.

[0078] Comparative Example 2

[0079] The difference from the embodiment is only that CRP in the combined regression model of PLA2G2A and CRP is replaced by SSA2 to construct a prediction model for the disease activity status of rheumatoid arthritis.

[0080] Comparative Example 3

[0081] The difference from the embodiment is only that PLA2G2A in the combined regression model of PLA2G2A and CRP is replaced by anti-cyclic citrullinated peptide (ACPA) antibody to construct a prediction model for the disease activity status of rheumatoid arthritis.

[0082] Comparative Example 4

[0083] The difference from the embodiment is only that CPR in the combined regression model of PLA2G2A and CRP is replaced by anti-PTX3 antibody to construct a prediction model for the disease activity status of rheumatoid arthritis.

[0084] Comparative Example 5

[0085] The difference from the embodiment is only that the random forest model is replaced by a generalized linear model (GLM), and the constructed model cannot predict the disease activity status of rheumatoid arthritis.

[0086] The sensitivity and specificity of the prediction models for the disease activity status of rheumatoid arthritis in Comparative Examples 1-4 were evaluated respectively, and the results are shown in Table 3.

[0087] Table 3 Performance indicators of different prediction models for the disease activity status of rheumatoid arthritis

[0088] Number AUC value Sensitivity % Specificity % Comparative Example 1 0.6275 78 76 Comparative Example 2 0.5670 58 55 Comparative Example 3 0.7194 71 73 Comparative Example 4 0.5923 67 64

[0089] The above detailed description is a specific description of one feasible embodiment of the present invention, and this embodiment is not intended to limit the patent scope of the present invention. Any equivalent implementation or modification without departing from the present invention shall be included within the scope of the technical solution of the present invention.

Claims

1. A marker combination for predicting the disease activity status of rheumatoid arthritis, characterized in that, The biomarker combination consists of PLA2G2A and CRP.

2. The marker combination according to claim 1, wherein The activity states are the active stage and the remission stage of rheumatoid arthritis.

3. Use of a biomarker combination as claimed in claim 1 or 2 in constructing a prediction model for the activity state of rheumatoid arthritis.

4. A prediction model for the disease activity status of rheumatoid arthritis, characterized in that, Constructed by the biomarker combination as claimed in claim 1 or 2.

5. A method for constructing a prediction model for the disease activity status of rheumatoid arthritis as described in claim 4, characterized in that, Comprising the following steps: (1) Collect peripheral blood samples from healthy subjects, patients in the active stage and remission stage of rheumatoid arthritis, and perform proteomic analysis to obtain significantly different proteins. (2) Randomly divide the peripheral blood samples into a training set and a test set in equal proportions, and use the random forest machine learning method to perform discriminant analysis on the significantly different proteins to obtain single indicators of the biomarkers. (3) Verify the protein expression levels of the single indicators, evaluate the diagnostic efficacy of the analysis equations obtained for the single indicators and combined indicators, determine the optimal equation, and construct a prediction model for the activity state of rheumatoid arthritis.

6. The construction method according to claim 5, wherein The biomarkers in step (2) are PLA2G2A and CRP.

7. The construction method according to claim 6, characterized in that The evaluation process in step (3) includes: taking the standardized CRP and PLA2G2A protein concentration values as independent variables, and the active stage and remission stage of rheumatoid arthritis as dependent variables, and using logistic regression analysis to establish three models respectively: only CRP, only PLA2G2A, and the combination of the two, and obtaining the AUC value, optimal threshold, sensitivity and specificity by using the ROC curve analysis method.

8. The construction method according to claim 7, wherein The prediction model for the activity state of rheumatoid arthritis in step (3) is X = 0.506 + 1.831×A + 1.617×B; where X is the model value, A is the standardized value of the CRP protein concentration in the plasma of the subject, and B is the standardized value of the PLA2G2A protein concentration in the plasma of the subject.

9. The construction method according to claim 8, characterized in that If X is greater than 5.667, the subject is in the active stage of rheumatoid arthritis; if X is less than 5.667, the subject is in the remission stage of rheumatoid arthritis.

10. Use of a prediction model for the activity state of rheumatoid arthritis as claimed in claim 4 or a prediction model for the activity state of rheumatoid arthritis constructed by the construction method as claimed in any one of claims 5 - 9 in the preparation of a product for predicting the activity state of rheumatoid arthritis.

Images