Plasma protein biomarker combinations and their applications and diagnostic systems for differentiating various child and adolescent psychiatric disorders

Abstract

The present invention relates to the technical field of diagnostic markers and diagnostic methods for child and adolescent mental illnesses, and specifically to a combination of plasma protein biomarkers and their applications and a diagnostic system that can be used to distinguish a variety of child and adolescent mental illnesses. The plasma protein biomarker combination includes the following proteins EFEMP1, F13A1, DCD, ADAMTSL4, GLIPR2, PIP and TGOLN2. Based on proteomics analysis and machine learning, this study evaluates the protein characteristics in the plasma of children and adolescents with depression, bipolar disorder and schizophrenia and non-psychotic healthy controls, explores possible pathogenic mechanisms, identifies and verifies potential protein biomarkers and their correlation with clinical diagnosis, and finds seven plasma proteins that can distinguish these three diseases. As objective diagnostic markers, they have high clinical application value.

Description

Technical Field

[0001] The present invention relates to the technical field of diagnostic markers and diagnostic methods for child and adolescent mental illnesses, and specifically to a plasma protein biomarker combination and its application and a diagnostic system that can be used to distinguish between a variety of child and adolescent mental illnesses. Background Art

[0002] Adolescent mental health problems are a global public health issue, and adolescent mental disorders present different symptoms and pathophysiological mechanisms compared with adults. For example, autonomic / physiological disturbances such as loss of energy, changes in weight, appetite, and sleep are common in adolescent depression, while absentmindedness / loss of interest and difficulty concentrating are more prevalent in adult depression. First-onset adolescent psychosis may be slower and more insidious, exposing them to the risk of long-term treatment delays. Previous studies have investigated the pathogenesis of various mental disorders. Molecular studies have found that major depressive disorder (MDD) is associated with changes in neurotrophic factors, proinflammatory cytokines, and dysregulation of the hypothalamic-pituitary-adrenal axis. Bipolar disorder (BD) involves disturbances in neuronal-glial plasticity, monoaminergic signaling, inflammatory homeostasis, cellular metabolic pathways, and mitochondrial function. The etiology of schizophrenia (SZ) is multifactorial, and genetic risk loci are associated with biological processes related to neurodevelopment, neuronal excitability, synaptic function, and the immune system. It is worth noting that most studies have focused on adult patients, and the etiology of adolescent mental disorders is not well studied.

[0003] MDD, BD, and SZ are prevalent, chronic major psychiatric disorders in adolescents, with overlapping features in etiology, phenotype, and genotype. Despite this, the diagnosis of adolescent psychiatric disorders is mainly based on clinical symptom criteria listed in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) without considering the underlying etiology or pathophysiological mechanisms. This approach can lead to significant symptom overlap and intra-group heterogeneity, which can cause misdiagnosis and mistreatment, especially for adolescents. Understanding the molecular nature of major psychiatric disorders in adolescents can provide valuable insights into their underlying pathophysiology. The identification of reliable biomarkers for psychiatric disorders is becoming increasingly important, which can facilitate accurate diagnosis, grouping, and personalized treatment strategies in clinical practice, as well as predict health status and guide the implementation and evaluation of medical interventions. Various omics research fields, including genomics, transcriptomics, proteomics, metabolomics, and epigenomics, have been used to identify diagnostic biomarkers for major psychiatric disorders. In particular, proteomics approaches are promising because proteins perform biological functions and closely reflect the pathophysiological processes of psychiatric disorders. Although plasma or serum proteomic studies have helped to elucidate the pathogenesis of psychiatric disorders and identify marker panels, changes in plasma proteins do not reflect the pathophysiological processes of psychiatric disorders. In adolescents with MDD, BD, or SZ, changes in plasma proteins remain undefined, and there are no reliable biomarkers to distinguish these psychiatric disorders.

[0004] Mass spectrometry (MS)-based proteomics is a powerful tool for biomarker discovery, but its full potential can only be realized when combined with advanced bioinformatics methods that can discover robust patterns. The identification of predictive protein markers, especially in peripheral blood, is crucial to improve our understanding of psychiatric disorders. Machine learning has become an important part of this quest, extracting valuable insights from the large and complex datasets generated by proteomic analysis. By integrating data from hundreds or thousands of samples or individuals, machine learning algorithms can identify meaningful biological information, protein features, and predict physiological status with great accuracy. Recent studies have shown that machine learning can distinguish major psychiatric disorders such as BD and MDD with some success. Nevertheless, the use of machine learning to discover potential plasma biomarkers to distinguish the three major psychiatric disorders (MDD, BD, and SZ) in adolescents remains an open question. Summary of the invention

[0005] The present invention is intended to provide an application of a combination of plasma protein biomarkers in the preparation of a system for diagnosing mental illness in children and adolescents, so as to solve the technical problem in the prior art of lacking biomarkers for children and adolescents that can be used to diagnose and distinguish different types of mental illnesses.

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

[0007] Application of a plasma protein biomarker combination in making a system for diagnosing mental illness in children and adolescents, wherein the plasma protein biomarker combination includes the following proteins: EFEMP1, F13A1, DCD, ADAMTSL4, GLIPR2, PIP and TGOLN2.

[0008] The present invention also provides a biomarker combination for the simultaneous diagnosis of depression, bipolar disorder and schizophrenia in children and adolescents, which is a plasma protein; it is composed of the following proteins: EFEMP1, F13A1, DCD, ADAMTSL4, GLIPR2, PIP and TGOLN2.

[0009] The present invention also provides a system for diagnosing mental illness in children and adolescents, which includes a sample pretreatment unit and a protein content detection unit; the sample pretreatment unit is used to purify a plasma sample and perform enzymatic hydrolysis on the protein in the plasma sample to obtain a polypeptide solution for quantitative analysis; the protein content detection unit is used to detect the content of each plasma protein biomarker in the polypeptide solution; the plasma protein biomarkers are composed of EFEMP1, F13A1, DCD, ADAMTSL4, GLIPR2, PIP and TGOLN2.

[0010] Furthermore, the system for diagnosing mental illnesses in children and adolescents is used to diagnose at least one mental illness in children and adolescents among depression, bipolar disorder and schizophrenia.

[0011] Furthermore, the system for diagnosing mental illness in children and adolescents is used to distinguish between children and adolescents with depression and healthy people; to distinguish between children and adolescents with bipolar disorder and healthy people; and to distinguish between children and adolescents with schizophrenia and healthy people.

[0012] Furthermore, the system for diagnosing mental illness in children and adolescents is used to distinguish between a mixed population formed by a group of children and adolescents with depression, a group of children and adolescents with bipolar disorder, a group of children and adolescents with schizophrenia, and a healthy group;

[0013] It is used to distinguish between children and adolescents with bipolar disorder, children and adolescents with depression, children and adolescents with schizophrenia, and a mixed population formed by healthy people;

[0014] Used to distinguish between children and adolescents with schizophrenia, children and adolescents with depression, children and adolescents with bipolar disorder, and a mixed population formed by healthy people.

[0015] Furthermore, the system for diagnosing mental illness in children and adolescents includes a sample preprocessing unit and a protein content detection unit.

[0016] Furthermore, the sample pre-processing unit processes blood or plasma.

[0017] Further, the protein content detection unit is used to detect the protein content of EFEMP1, F13A1, DCD, ADAMTSL4, GLIPR2, PIP and TGOLN2 in blood or plasma.

[0018] Further, the Unipot number of EFEMP1 is Q12805, the Unipot number of F13A1 is P00488, the Unipot number of DCD is P81605, the Unipot number of ADAMTSL4 is Q6UY14, the Unipot number of GLIPR2 is Q9H4G4, the Unipot number of TGOLN2 is O43493, and the Unipot number of PIP is P12273.

[0019] The principle and beneficial effects of this technical solution are:

[0020] Proteomic analysis based on parallel accumulation–serialfragmentation (PASEF) and Data Independent Acquisition (DIA) methods built on PASEF technology generated proteomic profiles in the plasma of adolescents with MDD, BD, or SZ and compared them with healthy controls (HCs). The goals of this study were to explore the potential pathogenic mechanisms of MDD, BD, and SZ, compare different proteins between adolescents and adults, study drug-protein interactions, and identify protein biomarkers and their correlation with clinical symptoms. Eleven candidate markers were screened out through the study, and then further validated in the validation cohort population, confirming seven plasma protein biomarkers: EFEMP1, F13A1, DCD, ADAMTSL4, GLIPR2, PIP, and TGOLN2. The above seven plasma protein biomarkers can be used to distinguish the three mental illnesses from healthy people, and can also be used to distinguish the three mental illnesses from each other, achieving more effective diagnosis. This technical solution discovered potential plasma biomarkers to distinguish the three major psychiatric disorders in adolescents (MDD, BD and SZ), solving major problems in the actual clinical diagnosis process. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is the visualization result of PLS-DA analysis of Example 1 for MDD, BD and SZ group patients and HCs, MDD and BD, SZ and MDD, and SZ and BD.

[0022] Figure 2 This is the visualization result of differential analysis of protein expression in plasma of MDD, BD and SZ patients in Example 1.

[0023] Figure 3 This is the research result of Example 2 on the effectiveness of using machine learning to identify the features of 11 plasma proteins and distinguish adolescent patients with MDD, BD or SZ.

[0024] Figure 4 This is the comparison of the machine learning methods established in Example 2 for distinguishing different groups and the research results on the effectiveness of distinguishing adolescent MDD, BD, SZ patients and HCs.

[0025] Figure 5 The correlation between the 11 biomarkers of Example 2 and the severity of each adolescent psychosis (showing the correlation between the 11 biomarkers and the clinical scales of MDD (a), BD (b) and SZ (c); HAMD-17, 17-item Hamilton Depression Rating Scale; HAMA-14, 14-item Hamilton Anxiety Rating Scale; YMRS, Young Mania Rating Scale; PANSS, Positive and Negative Syndrome Scale).

[0026] Figure 6 These are the experimental results of Example 2, showing that the seven plasma biomarkers identified by PRM can distinguish between MDD, BD and SZ in adolescent patients. DETAILED DESCRIPTION

[0027] The present invention is further described in detail below in conjunction with examples, but the embodiments of the present invention are not limited thereto. Unless otherwise specified, the technical means used in the following examples and experimental examples are conventional means well known to those skilled in the art, and the materials, reagents, etc. used can be obtained from commercial sources.

[0028] Example 1: Sample acquisition and proteomic detection and analysis

[0029] (1) Experimental subjects

[0030] To explore the changes in plasma proteins associated with pediatric and adolescent mental disorders, we used the discovery cohort and validation cohort (Table 1). In the discovery cohort, a total of 479 participants were included, including 146 pediatric and adolescent major depressive disorder (MDD), 112 pediatric and adolescent bipolar disorder (BD), 61 pediatric and adolescent schizophrenia (SZ), and 160 healthy controls (HCs). In the validation cohort, a total of 108 participants were recruited, including 32 adolescent MDD patients, 23 BD patients, 13 SZ patients, and 40 HCs. In the discovery cohort and validation cohort, there were no significant statistical differences in age and body mass index between HCs and patients with MDD, BD, and SZ, while there were significant statistical differences in gender distribution between BD and SZ and between BD and HC. In addition, in the discovery cohort and validation cohort, the 17-item Hamilton Depression Rating Scale (HAMD-17) and 14-item Hamilton Anxiety Rating Scale (HAMA-14) scores of participants in the MDD and BD groups were significantly higher than those of HCs. The details of the subjects are shown in Table 1.

[0031] Table 1: Statistics of basic information of subjects

[0032]

[0033] (2) Experimental methods

[0034] (2.1) Sample processing for DIA (Data-Independent Acquisition) analysis

[0035] Highly abundant proteins in plasma samples were removed using an Agilent 1290 Infinity II LC system and Multi Affinity Removal Column (Agilent Technologies, USA). Protein concentration was determined by a BCA protein assay kit (Thermo Scientific). Plasma proteins were then precipitated with trichloroacetic acid solution, reduced with 20 mM phosphine hydrochloride (TCEP), and alkylated with 40 mM iodoacetamide for 30 min in the dark. The mixture was digested with trypsin overnight at 37 °C, with a protease to sample protein ratio of 1:100 (weight ratio). The lysate was desalted using a Monospin C18 column (GLScience, Tokyo, Japan) and then vacuum centrifuged to dryness. The extracted peptides were reconstituted with 0.1% formic acid before data independent acquisition (DIA) analysis.

[0036] (2.1) Sample preparation for spectral library

[0037] Extracted peptides were prepared as described above. Peptides from each sample were pooled to obtain 100 µg of purified mixture, which was then dried by vacuum centrifugation. The dried peptides were reconstituted in Milli-Q water containing 2% acetonitrile and fractionated.

[0038] Using high pH reverse phase fractionation, peptides were separated by chromatographic column (BEH C18, 30 Å, 1.7 μm, 1 mm × 150 mm) and Waters XevoTM ACQUITY UPLC (Waters, USA) at pH 10. Eluted peptides were collected every 60 seconds for 80 minutes, and a total of 62 fractions were collected, which were then cross-condensed into 31 fractions and dried by vacuum centrifugation. The dried peptides were dissolved in 10 μl of Milli-Q water containing 0.1% formic acid, and then iRT peptides were added for chromatographic calibration.

[0039] (2.3) Liquid chromatography

[0040] Nanoflow reversed-phase chromatography was performed using a nanoElute liquid chromatography system (Bruker Daltonics). The peptides were separated at a flow rate of 300 nl / min using a nonlinear gradient starting from 2% acetonitrile in 0.1% formic acid solution, increasing to 22% acetonitrile in 0.1% formic acid solution in 8 min, then increasing to 37%, further increasing to 100% in 5 min, and finally maintaining at 100% concentration for 7 min before recalibration. A homemade column (25 cm × 75 µm, 1.5 µm C18-AQ particles) was used. Mobile phases A and B were water and acetonitrile containing 0.1% formic acid, respectively.

[0041] (2.4) Mass spectrometry

[0042] For the construction of spectral library: Liquid chromatography was connected to TimsTOFPro (Bruker) via CaptiveSpray nano-electrospray ion source. All 31 fraction samples were subjected to mass spectrometry analysis in data-dependent mode. The accumulation time and ramp time were each set to 100 ms, and mass spectra were recorded in the range of m / z 100-1700 in positive electrospray mode to generate the spectral library.

[0043] For DIA analysis: Another analysis was performed in data-independent mode, including MS1 scans from m / z 400 to m / z 1200 and 64 MS2 windows, using the DIA-PASEF acquisition scheme, covering the mass range of m / z 400 to m / z 1200. The ion mobility scan range was 0.6 to 1.6 Vs / cm scan. The collision energy is linearly related to the mobility, and the collision energy as a function of mobility was linearly increased from 59 eV at 1 / K0=1.6 Vs / cm2 to 20 eV at 1 / K0=0.6 Vs / cm2.

[0044] (2.5) Generation of spectral library and DIA data analysis

[0045] The spectral library was generated using Spectronaut version 14.2 (Biognosys) against the UNIPROT human database (only entries were reviewed) with all parameters set to default. DIA files were processed in default mode except that the correction factor for the XIC IM extraction window was set to 0.8.

[0046] (2.6) Verification using PRM

[0047] PRM detection was performed using a TimsTOF Pro2 (Bruker), and peptide precursors of selected differential protein precursors were analyzed using a Q-Exactive HF mass spectrometer (ThermoFisher Scientific, CA, USA) with a retention time window of 120 min. PRM quantification was performed using Skyline software. The differences in plasma protein abundance between healthy controls and patients with MDD, BD, and SZ were further validated.

[0048] (2.7) Statistical analysis

[0049] First, the discovered proteomics data were imputed using local weighted polynomial regression (lowess in R v.3.6.3) to calculate the local polynomial fit of the number of proteins and protein detection rate at each stage (time point) (see the reference: "Tian W, et al. Immune suppression in the early stage of COVID-19disease. Nature communications 11, 5859 (2020)."). Then, the data were normalized and transformed to logarithms using the LogNorm procedure. Least squares discriminant analysis (PLS-DA) as a supervised modeling-like method was used to assess which proteins were useful for distinguishing different groups (https: / / doi.org / 10.1002 / cem.2609). Those proteins with P < 0.05 (t-test) and fold changes > 2 or < 0.5 were considered differentially expressed proteins.

[0050] Mfuzzv.2.46.0 (https: / / www.bioconductor.org / packages / release / bioc / html / Mfuzz.html) was used to detect different subclustering models of gene expression between groups. Rv.3.6.3 was used to implement Fisher's exact test89. String version 11 was used for protein-protein interaction network analysis. Pathway enrichment analysis was performed using Fisher's exact test and the Kyoto Encyclopedia of Genes and Genomes (KEGG) metabolic pathway database. Heatmaps and network visualizations were performed using the ggplot2 package and Cytoscape v.3.5.1 implemented in the Omicsbean workbench (ggplot2: Elegant Graphics, 2009). Ingenuity Pathway analysis was performed to explore the downstream effects of the significant regulatory protein dataset. The z-score algorithm was used to predict the activation state (activated or inhibited) of a biological process. If the z-score was ≤-2, the process was predicted to be statistically significantly inhibited.

[0051] (2.8) Analysis of differentially expressed proteins

[0052] To explore changes in plasma proteins associated with adolescent mental disorders, the inventors used the mass spectrometry-based DIA-PASEF method to analyze the plasma protein expression levels of all groups in the discovery cohort. The raw data were processed by the double-bounded Bayes (DBB) estimation method and then standardized for subsequent analyses. The inventors' investigation identified 1,829 proteins in 479 samples in the discovery cohort and performed proteomic profiling. In order to intuitively describe the similarities and differences of these complex data sets, the inventors used a supervised partial least squares discriminant analysis (PLS-DA) method. The three-dimensional plots of the first three components showed a clear separation between patients and HCs, as well as the separation between MDD and BD, SZ and MDD, and SZ and BD ( Figure 1 Compared with HCs, the analysis found 20 up-regulated and 57 down-regulated differentially expressed proteins (DEPs) in MDD, 125 up-regulated and 272 down-regulated DEPs in BD, and 350 up-regulated and 95 down-regulated DEPs in SZ, with significance set at p < 0.05 and fold change > 2 or < 0.5 ( Figure 2 ad, representative data are shown in Table 2, selected representative data). Notably, 18 DEPs overlapped between patient groups compared with HC, including KDM5C, SFN, COTL1, HAPLN1, and SERPINB7 ( Figure 2 e). Among them, KDM5C is associated with X-linked syndromic mental retardation and plays a causal role in the development of major psychiatric disorders in women. SFN is expected to improve mood and cognitive function. Genomic studies have found that KRT23 is a potential candidate biomarker for MDD, HAPLN1 is a candidate biomarker for MDD, and HAPLN1 is a risk marker for depression. The CCN3 gene, which is highly expressed in the limbic region of the mammalian brain, is associated with mental mood disorders (including depression and anxiety). In addition, the inventors also compared the DEPs of MDD, BD, and SZ (representative data are shown in Table 3, excerpts of some representative data).

[0053] To focus on the potential common and specific protein features of the three major psychiatric disorders, the inventors performed a hierarchical cluster analysis and constructed a heat map using all DEPs ( Figure 2f, representative data are shown in Table 4, selected representative data). This process separated all DEPs into four different groups. Cluster 1 was characterized by increased protein levels in SZ compared with the other three groups, while cluster 2 showed decreased protein levels in BD compared with the other groups. In contrast, cluster 3 had increased protein levels in HC and MDD compared with BD and SZ, while cluster 4 had lower protein levels in HC and MDD than in BD and SZ. Pathway analysis of each cluster of DEPs (mainly focusing on cell communication, activation, adhesion, and humoral immune response) showed that although there were common altered proteins, each psychiatric disorder in adolescent patients had its own unique plasma protein signature.

[0054] Table 2: Differentially expressed proteins in depression group, bipolar disorder group, schizophrenia group and healthy group (selected representative data)

[0055]

[0056] Table 3: Differentially expressed proteins among depression group, bipolar disorder group and schizophrenia group

[0057]

[0058] Table 4: Cluster analysis of all differentially expressed proteins in depression group, bipolar disorder group, schizophrenia group and healthy group

[0059]

[0060] Example 2: Screening of biomarkers

[0061] (1) Machine Learning and Feature Selection

[0062] The inventors aimed to use machine learning techniques to identify protein features and predict physiological states based on proteomic data to better distinguish major psychiatric disorders in adolescents. Partial least squares discriminant analysis (PLS-DA), support vector machine (SVM), multilayer perceptron (MLP), logistic regression (LR) or random forest (RF) machine learning methods were used to screen protein biomarkers for distinguishing MDD, BD, SZ patients and HC. To compensate for missing values, the DBB strategy was adopted. To evaluate the performance of the model, the receiver operating characteristic (ROC) and balanced accuracy were calculated. The ROC curve was constructed using the pROC software package and the area under the curve (AUC) values ​​were calculated. The probability statistics of the ROC were calculated based on the prediction score formula derived from the PLS analysis. The ROC curve of the entire protein map can be used to compare different models based on the corresponding AUC values. The models were subjected to leave-out cross validation (LOOCV) and permutation tests, and Q 2 and R2 The permutation significance threshold was set at P < 0.05 to assess the robustness of the model and the amount of variation represented by the latent variables (LVs). To obtain the best model performance, the cross-validation error rate was also calculated, and 12 LVs were created from all observed variables determined in the PLS-DA model to obtain better predictive performance. In addition, the variable importance in projection (VIP) values ​​were analyzed, which reflect the relative importance of each protein in the projected model. A total of 11 proteins were identified by combining machine learning (ML) with bioinformatics. To verify whether the selected proteins are good classifiers that can construct classifiers that distinguish different groups, the ROC curve was quantified to evaluate their diagnostic performance.

[0063] (2) Analysis of experimental results

[0064] Machine learning was used to identify plasma biomarkers to distinguish childhood and adolescent depression from bipolar disorder and schizophrenia, to screen out potential specific biomarkers to better distinguish different adolescent mental illnesses, and to find that the symptoms of these diseases are to some extent correlated with objective clinical scales. The inventors used machine learning to identify promising protein markers and predict physiological states based on proteomic data. First, different ML methods for classification were constructed, including partial least squares discriminant analysis (PLS-DA), support vector machine (SVM), multilayer perceptron (MLP), logistic regression (LR), and random forest (RF). The performance of different models was evaluated by analyzing confusion matrices and ROC curves. The results showed that among the five machine learning methods, the PLS-DA model had the highest reliability, with an accuracy of 97% based on the plasma proteomic profiles of the HC, MDD, BD, and SZ groups, and an average AUROC value of 0.95 ( Figure 3 ab). The ROC curves drawn using the PLS-DA model also showed excellent classification ability, accurately distinguishing SZ from HC (AUC = 0.97%), BD from HC (AUC = 0.95%), and MDD from HC (AUC = 0.75%), with high specificity and sensitivity ( Figure 3 c).

[0065] To identify potential biomarkers, the inventors used a PLS-DA machine learning model based on accuracy, AUROC value, and variable importance to screen potential biomarkers ( Figure 4 a). To ensure the robustness of the model and reduce concerns about overfitting, additional evidence is provided and the model’s learning process is visualized ( Figure 4 bd). 2 The value indicates how well the model fits the training data, while Q 2 The value measures the model's ability to predict unseen data ( Figure 4b). These visualization models showed high accuracy during cross-validation, with no obvious signs of overfitting. Considering interpretability and predictability, the inventors selected 12 latent variables (LVs) as the optimal variables, which minimized the error of the prediction set and provided explanations for the model ( Figure 4 cd).

[0066] The inventors further evaluated the relative contribution of variables to classification by analyzing the importance of variables in prediction (VIP) values ​​( Figure 3 d), the higher the VIP value, the greater the contribution to the distinction between groups. By integrating bioinformatics analysis and machine learning strategies, the inventors discovered 11 protein biomarkers for distinguishing different groups, including EFEMP1 (Unipot No.: Q12805; full name: EGF-containing fibulin-like extracellular matrix protein 1), F13A1 (Unipot No.: P00488; full name: Coagulation factor XIII A chain), DCD (Unipot No.: P81605; full name: Dermcidin), ADAMTSL4 (Unipot No.: Q6UY14; full name: ADAMTS-like protein4), GLIPR2 (Unipot No.: Q9H4G4; full name: Golgi-associated plant pathogenesis-related protein 1), MRPS34 (Unipot No.: P82930; full name: Small ribosomal subunit proteinmS34), TGOLN2 (Unipot No.: O43493; full name: Trans-Golgi network integral membrane protein 2), PIP (Unipot number: P12273; full name: Prolactin-inducible protein), URB1 (Unipot number: O60287; full name: Nucleolar pre-ribosomal-associated protein 1), RCN1 (Unipot number: Q15293; full name: Reticulocalbin-1) and REG3A (Unipot number: Q06141; full name: Regenerating islet-derived protein 3-alpha) ( Figure 3e, normalized protein abundance) In previous studies, some of these biomarkers have also been identified as biomarkers for neuropsychiatric diseases, such as ADAMTS481 for Alzheimer's disease and TGOLN2 for Guillain-Barré syndrome, but these proteins have never been reported as biomarkers for MDD, BD, and SZ in children and adolescents. The ROC curves of the 11 selected biomarkers showed that the model performed well, with an AUROC value of 0.97 for SZ to distinguish from other groups (HC+BD+MDD), an AUROC value of 0.75 for MDD to distinguish from other groups (HC+BD+SZ), and an AUROC value of 0.73 for BD to distinguish from other groups (HC+SZ+MDD) ( Figure 3 f). In addition, the model using the selected 11 biomarkers performed well in distinguishing these three major psychiatric disorders in adolescents, with AUROC values ​​ranging from 0.72 to 0.88 ( Figure 4 f), also performed well in distinguishing SZ and BD from HCs, with AUROC values ​​of 0.95 and 0.87, respectively ( Figure 4 e). When analyzing the correlation between the 11 protein biomarkers and symptom severity, we found that differences in the levels of most proteins were significantly associated with the clinical features of the three major psychiatric disorders (Figure 5a-c).

[0067] (3) High-resolution targeted PRM validation of potential biomarkers

[0068] In order to evaluate the diagnostic potential of 11 potential biomarkers, the inventors conducted high-resolution targeted mass spectrometry validation based on parallel reaction monitoring (PRM) in a validation cohort consisting of 32 adolescent MDD patients, 23 BD patients, 13 SZ patients, and 40 HC patients (basic information of the subjects is shown in Table 1). PRM is a conventional mass spectrometry technology widely used in targeted protein analysis. It uses the precision of high-resolution mass spectrometers to select specific precursor ions (representing specific peptides of target proteins) and decompose them in the collision chamber, and then monitor all generated fragment ions. By measuring the intensity of these fragment ions, the amount of target protein can be inferred. TimsTOF Pro2 (Bruker) was used to detect the content of target protein in sample plasma for conventional PRM detection in the prior art. The detection process was entrusted to Yuyan Technology (Beijing) Co., Ltd. to perform conventional means of the prior art, which will not be repeated here. In order to obtain the expression data of 11 candidate biomarkers in the samples in the validation cohort and to judge the diagnostic efficacy of the 11 candidate biomarkers. The blood sample processing method refers to the previous "(2.1) Sample processing for DIA (Data-Independent Acquisition) analysis" to load the sample into the device, obtain relevant data, and calculate the protein abundance information.

[0069] The validation results showed that among the 11 biomarkers of PRM, 7 proteins were detected, including EFEMP1, F13A1, DCD, ADAMTSL4, GLIPR2, PIP and TGOLN2 (representative data are shown in Table 6, Figure 6a-g, showing protein abundance information and corresponding detection corresponding retention time and signal intensity). Among the seven proteins detected, the inventors found that the abundance of EFEMP1, F13A1 and PIP in the SZ group was significantly higher than that in the MDD group and the HCs group (Figure 6a-c). The ROC curve analysis of the diagnostic effect of the 7 proteins was performed, and the ROC curves for the 7 protein markers were constructed using the pROC software package and the area under the curve (AUROC) value was calculated. Specifically, a diagnostic model was established using binary logistic regression; the diagnostic variable (independent variable) was the protein abundance of the seven biomarkers, and the dependent variable was MDD and others (HC+BD+SZ), that is, the dependent variables were two sets of binary variables, MDD and HC+BD+SZ; and the diagnostic model was subjected to receiver operating characteristic (ROC) analysis to quantify the diagnostic performance of the biomarker combination containing seven metabolites. The dependent variable could also be two sets of binary variables, BD and others (HC+MDD+SZ), or two sets of binary variables, SZ and others (HC+MDD+BD), or two sets of binary variables, MDD and HC, or two sets of binary variables, SZ and HC, or two sets of binary variables, BD and HC, and so on. The ROC curve is used to judge the quality of classification and detection results. It is a very important and common statistical analysis method. It is a coordinate graph composed of the false positive rate (1-specificity) as the horizontal axis and the true positive rate (True positive rate, sensitivity) as the vertical axis. The curve is drawn by the different results of the test samples under different judgment criteria (thresholds). The area under the curve (AUROC) is used to indicate accuracy. The higher the AUROC value, the higher the accuracy. AUROC value > 0.7 indicates that the diagnostic model can diagnose the disease more accurately. The closer the value is to 1, the more ideal the diagnostic model is.

[0070] The ROC curves of the seven proteins detected in PRM performed well in distinguishing SZ and BD from HCs, with AUROC values ​​of 0.98 and 0.89, respectively, while the diagnostic effect on MDD from HCs was poor, with an AUROC value of 0.57 (Figure 6h). It is worth noting that the ROC curves of the seven proteins detected in PRM also performed well in distinguishing the three major mental illnesses, with AUROC values ​​of 0.65, 0.77, and 0.78 for MDD, BD, and SZ from other groups, respectively (Figure 6i). That is, among the population to be diagnosed, there are MDD patients, BD patients, SZ patients, and healthy people. Using the model of this scheme, MDD patients, BD patients, and SZ patients can be diagnosed more accurately from the above mixed population, which is more practical than simply identifying and distinguishing patients from healthy people. The marker combination composed of seven proteins in this scheme can effectively distinguish a single mental illness from other mental illnesses and healthy people. Furthermore, the expression levels of seven proteins detected in PRM were significantly correlated with the scale scores characteristic of three major psychiatric disorders (Fig. 6j-i).

[0071] Table 6: Protein abundance of 7 proteins detected in PRM

[0072]

[0073] The above is only an embodiment of the present invention, and the common knowledge such as the known specific technical solutions and / or characteristics in the solution is not described in detail here. It should be pointed out that for those skilled in the art, without departing from the technical solution of the present invention, several modifications and improvements can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. Use of a combination of plasma protein biomarkers in the preparation of a system for diagnosing mental illness in children and adolescents, characterized in that: The plasma protein biomarker combination includes the following proteins: EFEMP1 with Uniprot number Q12805, F13A1 with Uniprot number P00488, DCD with Uniprot number P81605, ADAMTSL4 with Uniprot number Q6UY14, GLIPR2 with Uniprot number Q9H4G4, PIP with Uniprot number P12273, and TGOLN2 with Uniprot number O43493; The system for diagnosing mental illness in children and adolescents is used to distinguish between a group of children and adolescents with bipolar disorder and a healthy group, or to distinguish between a group of children and adolescents with schizophrenia and a healthy group; or to distinguish between a group of children and adolescents with bipolar disorder and a group of children and adolescents with depression, a group of children and adolescents with schizophrenia, and a mixed group of healthy people, or to distinguish between a group of children and adolescents with schizophrenia and a mixed group of children and adolescents with depression, a group of children and adolescents with bipolar disorder, and a healthy group.

2. The use of the plasma protein biomarker combination according to claim 1 in making a system for diagnosing mental illness in children and adolescents, characterized in that: The system for diagnosing mental illness in children and adolescents comprises a sample preprocessing unit and a protein content detection unit.

3. The use of the plasma protein biomarker combination according to claim 2 in making a system for diagnosing mental illness in children and adolescents, characterized in that: The sample pretreatment unit processes plasma.

4. The use of the plasma protein biomarker combination according to claim 3 in making a system for diagnosing mental illness in children and adolescents, characterized in that: The protein content detection unit is used to detect the protein content of EFEMP1, F13A1, DCD, ADAMTSL4, GLIPR2, PIP and TGOLN2 in plasma.

5. A biomarker combination, characterized in that It is a plasma protein; It is composed of the following proteins: EFEMP1 with Uniprot number Q12805, F13A1 with Uniprot number P00488, DCD with Uniprot number P81605, ADAMTSL4 with Uniprot number Q6UY14, GLIPR2 with Uniprot number Q9H4G4, PIP with Uniprot number P12273, and TGOLN2 with Uniprot number O43493; The biomarker combination is used to distinguish between children and adolescents with bipolar disorder and healthy people, or to distinguish between children and adolescents with schizophrenia and healthy people, or to distinguish between children and adolescents with bipolar disorder and children and adolescents with depression, or a mixed population formed by children and adolescents with schizophrenia and healthy people, or to distinguish between children and adolescents with schizophrenia and children and adolescents with depression, or a mixed population formed by children and adolescents with bipolar disorder and healthy people.

6. A system for diagnosing mental illness in children and adolescents, characterized in that: It includes a sample pretreatment unit and a protein content detection unit; The sample pretreatment unit is used to purify the plasma sample and perform enzymatic hydrolysis on the protein in the plasma sample to obtain a polypeptide solution for quantitative analysis; the protein content detection unit is used to detect the content of each plasma protein biomarker in the polypeptide solution; the plasma protein biomarker consists of EFEMP1 with Uniprot number Q12805, F13A1 with Uniprot number P00488, DCD with Uniprot number P81605, ADAMTSL4 with Uniprot number Q6UY14, GLIPR2 with Uniprot number Q9H4G4, PIP with Uniprot number P12273 and TGOLN2 with Uniprot number O43493; The system is used to distinguish between children and adolescents with bipolar disorder and healthy people, or to distinguish between children and adolescents with schizophrenia and healthy people, or to distinguish between children and adolescents with bipolar disorder and children and adolescents with depression, or a mixed population formed by children and adolescents with schizophrenia and healthy people, or to distinguish between children and adolescents with schizophrenia and children and adolescents with depression, or a mixed population formed by children and adolescents with bipolar disorder and healthy people.

Images