Therapeutic effect prediction marker for anti-depression treatment and application thereof

Abstract

The invention provides a marker for predicting the curative effect of anti-depression treatment. The marker is selected from one or more of plasma cytokines OPN, PF4, CD163, Syndecan-1, ACE-2, a leptin receptor (leptin R) and BMPR-1A (bone morphogenetic protein receptor-1A). The invention also provides an application of the plasma cell factor in preparation of a product for predicting the curative effect of the anti-depression treatment, and the plasma cell factor is selected from one or more of OPN, PF4, CD163, Syndecan-1, ACE-2, leptin R and BMPR-1A, and has relatively high MDD curative effect prediction efficiency.

Description

Technical Field

[0001] The present invention belongs to the field of biomedicine technology, and particularly relates to the treatment of depression, and specifically to an efficacy prediction marker for antidepressant treatment and its application. Background Art

[0002] Major depressive disorder (MDD) is a type of mental illness with high morbidity, high recurrence, high disability and high suicide rate that seriously affects the quality of life of patients. The 2019 China Mental Health Survey (CMHS) showed that the lifetime prevalence of mood disorders in China is 6.8%, and the lifetime prevalence of MDD is 3.4%, which is the second largest mental illness in China after anxiety disorders, causing a serious disease burden. However, there is still a lack of objective and effective predictive indicators for the efficacy of antidepressant treatment, so the treatment of MDD still relies heavily on the doctor's personal clinical experience or assessment scale. Therefore, finding highly sensitive and specific biomarkers to optimize the diagnosis and treatment of depression is an urgent problem that needs to be solved today.

[0003] In recent years, a large number of studies have shown that immune-inflammatory mechanisms may be related to antidepressant treatment. Certain cytokines can participate in cell signal transduction, immune response, and response to internal and external factors such as inflammation. It has been reported that cytokines related to the pathogenesis of MDD may also be related to the efficacy of antidepressant treatment, and some cytokines can predict the efficacy of antidepressant drugs. For example, MDD patients with better response to antidepressant drugs have lower baseline IL-8 levels than those with poor response. In addition, increased levels of IL-1β and BDNF in plasma have also been confirmed as potential predictors of patients with refractory depression.

[0004] Despite the above findings, there are still many limitations in current studies, such as small sample size, testing only one or individual cytokines, lack of longitudinal studies, low reproducibility, etc. Therefore, the development of clinically relevant biomarkers remains crucial.

[0005] As a widely used artificial intelligence technology, machine learning has been widely used in the automatic analysis of complex data across various biomedical fields, and has obvious advantages in interpreting omics data, identifying biomarkers, and building prediction models for precision medicine. However, the research on using machine learning to analyze cytokine data in MDD to establish efficacy prediction models is still quite insufficient. The present invention explores the potential of plasma cytokine expression profiles in predicting efficacy biomarkers for antidepressant treatment. Summary of the invention

[0006] The purpose of the present invention is to overcome at least one disadvantage of the above-mentioned prior art and to provide a marker for predicting the efficacy of antidepressant treatment and its application.

[0007] In order to solve the above technical problems, the first aspect of the present invention provides a marker for predicting the efficacy of antidepressant treatment, the main feature of which is that it is selected from one or more of plasma cytokines OPN, PF4, CD163, Syndecan-1, ACE-2, leptin R (leptin Receptor) and BMPR-1A.

[0008] The second aspect of the present invention provides the use of plasma cytokines in the preparation of a product for predicting the efficacy of antidepressant treatment, the main feature of which is that the plasma cytokines are selected from one or more of OPN, PF4, CD163, Syndecan-1, ACE-2, leptinR and BMPR-1A.

[0009] The product includes a kit containing reagents for detecting the expression level of the plasma cytokines or a high-throughput detection technology product, such as a protein antibody chip.

[0010] The plasma cytokines consist of OPN, PF4, CD163, Syndecan-1, ACE-2, leptin R and BMPR-1A.

[0011] The product includes a prediction model, which adopts the formula: response score = CD163 expression level × 0.74 + BMPR-IA expression level × 0.56 + Leptin R expression level × 0.33 + OPN expression level × 0.33 + PF4 expression level × 0.10 + ACE-2 expression level × 0.10 + Syndecan-1 expression level × 0.04.

[0012] The efficacy prediction marker for antidepressant treatment of the present invention and the application thereof have high efficacy in predicting the efficacy of MDD. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 A is a graph showing the accuracy results of different machine learning models tested through self-service resampling.

[0014] Figure 1 B is the area under the curve (AUC) result graph of different machine learning models tested by self-service resampling.

[0015] Figure 1 C is the AUC result graph of the random forest (RF) model, which performs better than other algorithms and obtains the highest average AUC score.

[0016] Figure 1D is a graph showing the correlation between the response score calculated using the RF model and the percentage reduction in the MADRS score. DETAILED DESCRIPTION

[0017] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further described below in conjunction with specific embodiments.

[0018] The present invention integrates the omics data of cytokines and uses machine learning methods to systematically explore their efficacy prediction value in antidepressant treatment. Specifically, protein antibody chips are used to detect peripheral plasma of 440 depression patients and healthy controls to identify characteristic cytokines; multiple machine learning algorithms are used to identify key cytokines, and an antidepressant efficacy prediction model is constructed, and the model is internally validated. The present invention provides a more objective biological indicator for efficacy prediction of antidepressant treatment, and no related research reports or invention patents have been seen before.

[0019] Study subjects

[0020] The iMore study is a prospective observational cohort study that included patients with depression and followed up the patients for 8 weeks. Symptom changes were assessed using scales such as the Montgomery-Asberg Depression Rating Scale (MADRS). Peripheral blood was drawn from patients at baseline, and multidimensional indicators including cytokines and immune cells were detected. It aims to build a diagnostic model for depression and a model for efficacy prediction. This study has been approved by the Research Ethics Committee of Shanghai Mental Health Center, China, with the ethics approval number 2020-87. Written informed consent was obtained from all participants before the study. In addition, it has been registered in the Clinical Trial Registry with the identifier NCT04518592. The specific study design and details can be found in Integrated Module of Multidimensional Omics for Peripheral Biomarkers (iMORE) in patients with major depressive disorder: rationale and design of a prospective multicentre cohort study.

[0021] The patients with depression included in this study were patients enrolled in iMore from December 2020 to June 2022 and completed 440 cytokine tests. A total of 84 patients with depression were included. Inclusion criteria for patients with depression: adult patients aged 18 to 65 years who met the diagnostic criteria for major depressive disorder (MDD) in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5).

[0022] The inclusion criteria are as follows:

[0023] 1. Adults diagnosed with major depressive disorder (MDD) experiencing a first episode or relapse.

[0024] 2. Baseline MADRS total score ≥24.

[0025] 3. Baseline Hamilton Depression Rating Scale 17 (HAMD-17) total score ≥ 20.

[0026] 4. Provide informed consent voluntarily.

[0027] Exclusion criteria mainly include:

[0028] 1. Current presence of any Axis I disorder other than major depressive disorder (MDD).

[0029] 2. There is a serious risk of suicide or the score on suicidal ideation (MADRS item 10) is greater than 3.

[0030] 3. The presence of serious physical illnesses, such as cerebrovascular disease, respiratory disease, blood disease, liver and kidney disease, endocrine disease, nervous system disease and other systemic diseases.

[0031] 4. Pregnancy or lactation.

[0032] Since one patient was excluded due to extremely low cytokine test results, 83 patients completed cytokine testing at baseline. Among them, 49 patients with depression completed 8 weeks of follow-up. The construction of the efficacy model in the present invention is based on these 49 patients with depression.

[0033] Cytokine detection method:

[0034] The evaluation of 440 cytokines was performed using a sandwich approach based on antibody array technology (HumanCytokine Antibody Array, RayBiotech). Specifically, 100 μL of plasma was diluted 1:2 and added to the chip embedded with 440 primary antibodies and incubated overnight at 4°C. After a comprehensive washing step, biotinylated antibodies were added, incubated for 2 hours, and washed again. Subsequently, 80 μL of Cy3-conjugated streptavidin was dispensed and incubated for 1 hour in the dark at room temperature. After another wash, the fluorescence signal data specific to Cy3 was captured by the InnoScan 300 microarray scanner (Innopsys).

[0035] Example 1 MDD treatment effect prediction model

[0036] Development set research subjects: the above 49 patients with depression.

[0037] Test set research object: Use the bootstrap method and resample 100 times as the test set.

[0038] Blood samples collected at baseline from 49 patients with depression in the development set were tested for 440 cytokines based on antibody microarray chip technology. The MADRS scale score reduction rate before and after treatment exceeded 50% to evaluate whether they were in the effective treatment group and the ineffective treatment group. Among the final 49 patients, 31 patients were in the effective treatment group and 18 patients were in the ineffective treatment group. The efficacy prediction model of the present invention was obtained based on the control of the effective treatment group and the ineffective treatment group. Since this study is a natural observational study, there is no intervention in the type of medication used by the patients, and the medication treatment of all patients is determined or confirmed by clinicians based on the patient's condition.

[0039] Identification of cytokines associated with response to antidepressant medication

[0040] Univariate logistic regression was applied to identify cytokines associated with antidepressant treatment response, with the dependent variables being treatment response and treatment ineffectiveness. Univariate logistic regression results less than 0.05 were considered to be associated with treatment response, and seven cytokines were ultimately identified. Specifically, it was found that increased levels of OPN, PF4, CD163, and Syndecan-1 could predict treatment response, and decreased levels of ACE-2, leptin R, and BMPR-1A could predict treatment response. ROC curve analysis showed that the top seven factors associated with treatment response, OPN, PF4, CD163, Syndecan-1, ACE-2, leptin R, and BMPR-1A, had an AUC greater than 0.7 for each factor, and when leptin R was used as a single factor, it had the highest predictive value for treatment efficacy, with an AUC of 0.786.

[0041] Constructing the optimal prediction model for MDD treatment effect

[0042] Based on the development set samples and the 7 cytokines related to antidepressant efficacy identified above, eight classic machine learning algorithms were further used to construct the optimal prediction model for MDD treatment efficacy, including: elastic network (Enet), LASSO, ridge regression, gradient boosting machine (GBM), random forest (RF), supervised principal component analysis (SuperPC), extreme gradient boosting (XGboost) and support vector machine (SVM). The bootsrap method was used to perform 100 self-help resampling of all samples to create a test set as internal validation, and the accuracy, area under the curve (AUC), recall rate and F1 score of each model were calculated to determine the optimal efficacy prediction model.

[0043] Figure 1 A shows the accuracy results when testing different machine learning models through bootstrap resampling; Figure 1 B shows the corresponding area under the curve (AUC) results. The model built based on XGboost showed obvious overfitting and was therefore excluded from subsequent analysis.

[0044] like Figure 1 C and Figure 1 As shown in D, the optimal prediction model comes from the random forest algorithm. The average AUC score of the RF model is the highest, which is 0.944. Its model is specifically (reaction score = CD163 expression × 0.74 + BMPR-IA expression × 0.56 + Leptin R expression × 0.33 + OPN expression × 0.33 + PF4 expression × 0.10 + ACE-2 expression × 0.10 + Syndecan-1 expression × 0.04). Therefore, the present invention has a higher predictive value.

[0045] In this specification, the present invention has been described with reference to specific embodiments thereof. However, it is apparent that various modifications and variations may be made without departing from the spirit and scope of the present invention. Therefore, the description should be regarded as illustrative rather than restrictive.

Claims

1. A marker for predicting the efficacy of antidepressant treatment, characterized in that: One or more selected from the group consisting of plasma cytokines OPN, PF4, CD163, Syndecan-1, ACE-2, leptin R and BMPR-1A.

2. Use of plasma cytokines in the preparation of a product for predicting the efficacy of antidepressant treatment, characterized in that: The plasma cytokine is selected from one or more of OPN, PF4, CD163, Syndecan-1, ACE-2, leptin R and BMPR-1A.

3. The use according to claim 2, characterized in that: The product includes a kit or a chip containing reagents for detecting the expression level of the plasma cytokines.

4. The use according to claim 2, characterized in that: The plasma cytokines consist of OPN, PF4, CD163, Syndecan-1, ACE-2, leptin R and BMPR-1A.

5. The use according to claim 2, characterized in that: The product includes a prediction model, which adopts the formula: response score = CD163 expression level × 0.74 + BMPR-IA expression level × 0.56 + Leptin R expression level × 0.33 + OPN expression level × 0.33 + PF4 expression level × 0.10 + ACE-2 expression level × 0.10 + Syndecan-1 expression level × 0.04.

Images