Application of reagent for detecting biomarker in preparation of sepsis early diagnosis product

Abstract

The invention relates to the field of biomedicine, and discloses application of a reagent for detecting a biomarker in preparation of a product for early diagnosis or prognosis of sepsis. And the biomarker is any one or a combination of more of TLR5, HMGB2, C19orf59, TXK and DGKA. The ability and reliability of the five biomarkers for distinguishing healthy individuals and sepsis patients in two queues are verified through an ROC curve and a PR curve, and candidate biomarkers are further verified through a CLP mouse model.

Description

Technical Field

[0001] The present invention relates to the field of biomedicine. Specifically, the present invention provides a use of a reagent for detecting biomarkers in preparing a product for early diagnosis of sepsis. Background Art

[0002] Sepsis is a systemic inflammatory response syndrome caused by infection. If not promptly diagnosed and treated, it can lead to multi-organ failure and be life-threatening. Timely and accurate diagnosis is crucial for early intervention and improving the prognosis of patients with sepsis. In recent years, progress has been made in sepsis diagnostic research. Conventional biomarkers, including C-reactive protein (CRP), procalcitonin (PCT), and white blood cell count, have been demonstrated to be useful for assessing infection severity and guiding medication. Furthermore, imaging modalities such as ultrasound, CT, and magnetic resonance imaging can localize foci of infection, providing evidence for the etiology of sepsis. However, current diagnostic methods for sepsis remain challenging due to poor sensitivity, specificity, and efficiency. The development of more specific biomarkers to assist in sepsis detection in routine clinical care is urgently needed. Summary of the Invention

[0003] Based on the above technical problems, the present invention provides a use of a reagent for detecting biomarkers in the preparation of an early diagnosis product for sepsis, thereby providing a new strategy for the early diagnosis of sepsis with good diagnostic performance.

[0004] This study obtained gene expression data from sepsis patients from the GEO database (GSE134347), standardized the data to remove batch effects, and ensured data quality. Differential expression analysis was then performed between the sepsis patients and a normal control group to identify genes with significant differential expression. A co-expression network of transcripts was constructed using the WGCNA method to identify key gene modules associated with sepsis characteristics. Combining differentially expressed gene analysis with WGCNA, and taking the intersection of the two, a total of 277 genes were identified. LASSO regression and RF machine learning algorithms were then used to identify five potential biomarkers: TLR5, HMGB2, C19orf59, TXK, and DGKA.

[0005] The specific technical solutions provided by the present invention are as follows: In a first aspect, the present invention provides a use of a reagent for detecting a biomarker in preparing a product for early diagnosis or prognosis of sepsis, wherein the biomarker is any one or a combination of several of TLR5, HMGB2, C19orf59, TXK and DGKA.

[0006] As a preferred embodiment of the present invention, the biomarker is any one or a combination of several of TLR5, HMGB2 and C19orf59.

[0007] As a preferred embodiment of the present invention, down-regulation of TLR5, HMGB2 or C19orf59 expression and up-regulation of TXK or DGKA expression in the detected sample indicate a good prognosis for the sepsis patient.

[0008] As a preferred embodiment of the present invention, the test sample is blood.

[0009] As a preferred embodiment of the present invention, the product is a detection reagent, a test kit, a test paper, a probe or a chip.

[0010] In a second aspect, the present invention provides a use of a reagent for detecting a biomarker in screening drugs for preventing or treating sepsis, wherein the biomarker is any one or a combination of several of TLR5, HMGB2, C19orf59, TXK and DGKA.

[0011] As a preferred embodiment of the present invention, the biomarker is any one or a combination of several of TLR5, HMGB2 and C19orf59.

[0012] In a third aspect, the present invention provides a method for screening a drug for preventing or treating sepsis, comprising the following steps: A sepsis model is induced by drug administration, and the expression levels of the biomarkers described in claim 1 are detected and compared before and after administration. When the expression of TLR5, HMGB2 or C19orf59 in the test sample is downregulated and the expression of TXK or DGKA is upregulated, it indicates that the administered drug is the target drug.

[0013] As a preferred embodiment of the present invention, the biomarker is any one of TLR5, HMGB2 and C19orf59.

[0014] The present invention obtained differentially expressed genes from the GEO database (GSE134347) in patients with sepsis and performed WGCNA analysis to identify key modules and genes associated with sepsis characteristics. The intersection of the differentially expressed genes and the WGCNA hub genes was included. Subsequently, LASSO regression and RF machine learning algorithms were used to screen candidate biomarkers for sepsis: TLR5, HMGB2, C19orf59, TXK, and DGKA. Furthermore, support vector machine-reflective factor analysis (SVM-RFE) and process-like linear regression analysis (PLS-DA) analyses demonstrated that these five biomarkers may be key genes for sepsis. Receiver operating characteristic (ROC) and predictor performance (PR) curves were used to validate the ability and reliability of these five biomarkers in differentiating healthy individuals from patients with sepsis in two cohorts. The present invention further established a CLP mouse model to validate the candidate biomarkers, ultimately obtaining... BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 Figure 2 shows RNA-seq differential expression analysis. A is a heatmap of the correlation between sepsis and healthy samples. Red represents a strong correlation, and white represents a weak correlation. B is a PCA plot of sepsis and healthy samples. C shows a volcano plot of differentially expressed genes (DEGs) between the sepsis and healthy groups. Blue dots indicate downregulated genes, red dots indicate upregulated genes, and gray dots indicate genes with no significant changes.

[0015] Figure 2 Figure 1: WGCNA screening of hub genes. Figure A shows the network fitting index at different soft threshold powers. Figure B shows the average connectivity of the WGCNA network. Figure C shows the gene hierarchical clustering diagram and module visualization. Figure D shows the correlation analysis between the module and sepsis. Each small unit contains the corresponding correlation coefficient and P E is a blue module scatter plot showing the relationship between the gene significance (GS) value of the gene and the module membership (MM) value of the gene. F is a brown module scatter plot showing the relationship between GS and MM.

[0016] Figure 3 Figure 3 is the enrichment analysis of core genes. A Venn diagram shows the intersection of DEGs obtained by RNA-seq and genes screened by WGCNA hub genes. B heat map shows the expression levels of 277 overlapping genes. C Figure 3 KEGG enrichment analysis of 277 genes to identify enriched pathways related to sepsis. D Figure 3 GO enrichment analysis of 277 genes to identify enriched GO terms related to sepsis. The top 10 terms were ranked according to statistical significance ( P <0.05) for annotation.

[0017] Figure 4Figure 1 shows the identification of candidate biomarkers in the training cohort. Figure A shows the LASSO coefficient spectrum for the 14 variables. Figure B shows the optimal value of the parameter λ in the LASSO regression. Figure C shows the variable importance ranked by mean decrease in accuracy and Gini index. Figure D shows the heat map of the expression levels of the five biomarkers TLR5, HMGB2, C19orf59, TXK, and DGKA in the training cohort. Figure E shows the confusion matrix of the five biomarkers in the SVM model. The numbers in the heat map cells indicate the number of correctly predicted samples and the ratio of true to predicted cases. Figure F shows the PLS-DA analysis based on the expression profiles of the five biomarkers in the training cohort. The PLS-DA is plotted along the first two principal component axes (PC1 and PC2).

[0018] Figure 5 The expression levels of the five candidate genes in the external validation set and training set are shown in Figure 2. Box plots A to C show the expression levels of the candidate genes in the GSE134347, GSE236713, and GSE185263 datasets. P <0.05,** P <0.01,*** P <0.001, **** P <0.0001 compared with the control group. Statistical significance was assessed by Wilcoxon test.

[0019] Figure 6 Five candidate biomarkers were validated in the GSE236713 dataset (control group, n = 30; sepsis group, n = 324). A heatmap shows the expression levels of the five biomarkers in the GSE236713 dataset. B PLS-DA analysis based on the expression profiles of the five biomarkers in the GSE236713 dataset. PLS-DA plots were performed along the first two principal component axes (PC1 and PC2). C shows the confusion matrix of the five biomarkers in the RF model. The numbers in the heatmap cells indicate the proportion of true versus predicted cases. D shows the confusion matrix of the five biomarkers in the SVM model. The numbers in the heatmap cells indicate the proportion of true versus predicted cases. E shows the receiver operating characteristic (ROC) curves for the five biomarkers in the GSE236713 dataset. F shows the price-performance ratio (PR) curves for the five biomarkers in the GSE236713 dataset.

[0020] Figure 7Five candidate biomarkers were validated in the GSE185263 dataset (control group, n = 44; sepsis group, n = 348). A heatmap shows the expression levels of the five biomarkers in the GSE185263 dataset. B PLS-DA analysis based on the expression profiles of the five biomarkers in the GSE185263 dataset. The PLS-DA plot is plotted along the first two principal component axes (PC1 and PC2). C shows the confusion matrix of the five biomarkers in the RF model. The numbers in the heatmap cells indicate the ratio of true cases to predicted cases. D shows the confusion matrix of the five biomarkers in the SVM model. The numbers in the heatmap cells indicate the ratio of true cases to predicted cases. E shows the receiver operating characteristic (ROC) curves for the five biomarkers in the GSE185263 dataset. F shows the price-performance ratio (PR) curves for the five candidate biomarkers in the GSE185263 dataset.

[0021] Figure 8 To establish a septic injury mouse model using CLP and to examine related indicators. A: Mild CLP injury model. B: Sepsis scores in the sham, 8-hour CLP, and 24-hour CLP groups. C: Rectal temperatures in the sham, 8-hour CLP, and 24-hour CLP groups. D: Statistical graph of routine blood test parameters in each group. n = 8, * P <0.05,** P <0.01,*** P <0.001, **** P <0.0001, vs Sham group; ns indicates no significant difference.

[0022] Figure 9 qRT-PCR analysis of candidate biomarkers. A-D Detection of mRNA expression levels of MCEMP1 (C19orf59), HMGB2, TLR5, and DGKA in the sham group, CLP 8h group, and CLP 24h group. n = 8, * P <0.05,** P <0.01,*** P <0.001, **** P <0.0001, vs Sham group; ns indicates no significant difference. Specific implementation plan The present invention implements the following method, combining gene expression data with machine learning algorithms to screen for sepsis biomarkers. In addition to genetic data analysis, various experimental platforms, statistical methods, and machine learning algorithms are employed to ensure high model accuracy.

[0023] Unless otherwise specified, the scientific and technical terms used herein are understood according to the knowledge of ordinary technicians in the relevant fields.

[0024] The test materials used in the examples of the present invention are all conventional test materials in the art and can be purchased through commercial channels.

[0025] Example 1 Gene expression data acquisition and preprocessing A search using the terms "sepsis" and "blood" in the NCBI Gene Expression Omnibus (GEO, https: / / www.ncbi.nlm.nih.gov / geo / ) yielded three microarray datasets: GSE134347, GSE236713, and GSE185263. GSE134347 was designated as the training set, including gene expression profiles from healthy individuals and patients with sepsis. The GSE236713 and GSE185263 datasets served as external validation sets to confirm the robustness of the study results. Table 1 lists detailed information for each dataset, including the microarray platform used, data type, and sample information.

[0026] Table 1 Descriptive information of the GEO dataset GSE134347 microarray data were background corrected using Robust Multi-Array Average (RMA) and then normalized using quantile normalization. Probes were annotated, and probes that did not match any gene symbols were excluded. For genes with multiple probes, the average expression level was used as the final expression level.

[0027] Example 2 Identification of differentially expressed genes plan: The sample correlation heat map was generated using the "gplots" package, and the principal component analysis (PCA) was performed using the "ggplot2" package in R software. Subsequently, DEGs were screened using the limma R package, with the screening threshold set to |log2FC|>1. P <0.05. Finally, the ggplot2 package was used to draw a volcano plot to visualize the expression patterns of these DEGs.

[0028] result: Sample correlation analysis and PCA analysis showed that the sepsis group was significantly separated from the healthy group ( Figure 1A and B in the middle). Subsequently, the expression levels of differentially expressed genes (DEGs) were displayed using a volcano plot, and it was found that 315 genes were upregulated and 340 genes were downregulated ( Figure 1 These genes were used in subsequent analyses.

[0029] Example 3 Weighted gene co-expression network analysis (WGCNA) plan: A gene co-expression network was constructed using WGCNA to correlate the gene network with clinical features. First, the 'sft$powerEstimate' function was used to determine the optimal soft threshold of β = 13 and R² = 0.84. Based on this threshold, the similarity matrix was converted into an adjacency matrix and finally into a topological overlap matrix. Next, a hierarchical clustering approach was used to identify distinct gene modules, each containing at least 50 genes. Subsequently, the Pearson correlation coefficient was calculated to identify the modules most associated with sepsis (key modules). Furthermore, the gene-score (GS) and gene-mass ratio (MM) values ​​were calculated for genes in the brown and blue modules. GS was used to assess the relationship between genes and sepsis features, while MM defined the correlation between module signature genes and gene expression profiles. Finally, hub genes within the modules were selected based on the criteria of |GS| > 0.8 and |MM| > 0.8.

[0030] To further explore the biological functions and potential pathways of sepsis-related core genes, we performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis using the clusterProfiler R package. GO categories include biological process (BP), molecular function (MF), and cellular component (CC). The results of these functional enrichment analyses were then visualized using bubble plots and bar charts.

[0031] result: We constructed a gene co-expression network using WGCNA and obtained modules and hub genes associated with sepsis in the training dataset. After quality control, a normalized count matrix containing 25,243 genes was obtained. The optimal soft threshold power β = 13 (R 2 = 0.84) established a scale-free network ( Figure 2 A total of 18 modules were identified by hierarchical clustering method ( Figure 2 Middle C).

[0032] Subsequently, the correlation between each module and sepsis was calculated, and the hub genes within the module were further identified. Figure 2As shown in Figure D, the brown module had the strongest positive correlation with sepsis (r = 0.73, p = 1e-41), while the blue module had the highest negative correlation (r = -0.89, p = 1e-81).

[0033] In addition, the GS and MM values ​​of genes in the brown and blue modules were calculated. Finally, 124 and 658 hub genes with |GS|>0.8 and |MM|>0.8 were identified in the brown and blue modules, respectively ( Figure 2 Given their significant association with sepsis, these genes were prioritized for further investigation.

[0034] Based on the intersection of 655 DEGs and 782 WGCNA hub genes, a total of 277 genes were screened ( Figure 3 A), Heat map visualizing the differential expression levels of these 277 genes between healthy individuals and sepsis patients ( Figure 3 Middle B).

[0035] The potential functions and pathways of these core genes were annotated through KEGG pathway and GO functional enrichment analysis. The results of KEGG pathway enrichment analysis showed that the core genes were mainly involved in the T cell receptor signaling pathway, the PD-1 checkpoint pathway in cancer, and the differentiation of Th1 and Th2 cells ( Figure 3 In addition, the enriched BPs include NAD(P) + Nucleoside enzyme activity, NAD + Nucleosidase activity and phosphatase binding. The enriched CC terms included specific granules, cell tail edge, and pseudopodia. Finally, the enriched MF terms included positive regulation of leukocyte cell-cell adhesion, positive regulation of leukocyte activation, and immune response regulation signaling pathways ( Figure 3 Middle D). This further demonstrates the role of these core genes in the immune and inflammatory processes of sepsis.

[0036] Example 4 Machine learning model building plan: To identify diagnostic biomarkers associated with sepsis, LASSO and RF analyses were performed using the glmnet and Random Forest R packages, respectively. Through parameter optimization, an alpha value of 1 was set to construct a LASSO regression model. Furthermore, the "mtry" parameter and the optimal number of trees were determined based on the out-of-bag error estimate in the RF analysis. Subsequently, 5-fold cross-validation predictions were performed for the three machine learning methods: LASSO, RF, and SVM-RFE.

[0037] result: LASSO and RF analyses were performed on the GSE134347 dataset to identify biomarkers associated with sepsis. Specifically, the optimal lambda value corresponding to the minimum mean square error (MSE) was determined by LASSO regression as the penalty coefficient for selecting 14 feature genes ( Figure 4 (A, B in Chinese).

[0038] Subsequently, the RF model was constructed using these 14 genes, and their importance was evaluated and ranked. It is worth noting that the top five genes ranked according to the "Mean Decrease Accuracy" and "Mean Decrease Gini" values ​​are assumed to be potential biomarkers: TLR5, HMGB2, C19orf59, TXK, and DGKA ( Figure 4 Middle C).

[0039] Figure 4 The expression levels of these five biomarkers are visualized as a heat map in Figure D. Compared with healthy individuals, TLR5, HMGB2, and C19orf59 were upregulated, while TXK and DGKA were downregulated in patients with sepsis.

[0040] The SVM-RFE model was used to further validate the discriminative ability of these five biomarkers, demonstrating their ability to effectively distinguish sepsis patients from healthy individuals (classification accuracy = 100%) ( Figure 4 Middle E).

[0041] Similarly, PLS-DA also showed good classification performance in the training cohort ( Figure 4 (F), further strengthening the robustness of these five biomarkers in the diagnosis of sepsis.

[0042] Example 5 Biomarker validation 7. Dataset Verification plan: To further validate the accuracy of the potential biomarkers, two external validation cohorts, GSE236713 (comprising 30 healthy samples and 324 sepsis samples) and GSE185263 (comprising 44 healthy samples and 348 sepsis samples), were downloaded to evaluate the diagnostic performance of the selected biomarkers. Four different methods were used to assess the predictive value of each potential biomarker: supportive machine-recurrent factor analysis (SVM-RFE), partial linear serotonin-induced diagnoses (PLS-DA), receiver operating characteristic (ROC) curves, and predictor-response (PR) curves. In the SVM-RFE analysis, the e1071 R package was used to assess the classification performance of the biomarkers, and a linear kernel function was applied to fit the model. In addition, the classification performance of PLS-DA was evaluated using the Wukong online website (omicsolution.com / wkomics / main / ). The pROC and pRROC R packages were used to generate ROC and PR curves, respectively, and the area under the curve (AUC) values ​​were calculated.

[0043] result: First, the expression levels of TLR5, HMGB2, C19orf59, TXK, and DGKA were analyzed based on these two datasets. Compared with healthy individuals, the expression levels of TLR5, HMGB2, and C19orf59 in patients with sepsis were significantly increased, while the expression levels of TXK and DGKA were significantly decreased (p < 0.05) ( Figure 5 ).

[0044] Likewise, the heatmap shows the significant differences in expression of these biomarkers between healthy and sepsis samples ( Figure 6 China A and Figure 7 Middle A).

[0045] Furthermore, the results of PLS-DA further emphasized the ability of these five biomarkers to discriminate between healthy individuals and patients with sepsis ( Figure 6 Middle B and Figure 7 Middle B).

[0046] After constructing the RF and SVM-RFE models based on the GSE236713 dataset, the prediction accuracy of these biomarkers reached 97.18% and 97.46%, respectively ( Figure 6 (C, D).

[0047] For the GSE185263 dataset, the prediction accuracy is 96.17% and 94.39% respectively ( Figure 7 (C, D).

[0048] Furthermore, to more rigorously evaluate the performance indicators, ROC and PR curves were used, and the results showed that these five biomarkers had high area under the curve (AUC) values ​​(AUC>0.7) ( Figure 6 E, F and Figure 7 (E, F in the middle).

[0049] The findings suggest that these five potential genes have important application prospects as key biomarkers for sepsis diagnosis, providing a promising avenue for early and accurate detection.

[0050] 2. Experimental Verification plan: A mouse CLP model was established. Before surgery, all mice were fasted for 8 hours and had free access to water. Using 2% isoflurane inhalation anesthesia, a 1-2 cm abdominal incision was made along the midline of the abdomen to expose the cecum and adjacent intestines. A 4-0 nylon suture was used to tightly ligate the terminal third of the cecum, followed by a single puncture of the cecal wall. A small amount of feces was squeezed through the puncture site to ensure patency, and then it was returned to the abdominal cavity. Sterile 6-0 silk sutures were used to suture the peritoneum, fascia, abdominal muscles, and skin incisions. The sham operation group underwent the same procedures as the surgical group, except that the cecum was not ligated and punctured.

[0051] A fully automatic hematology analyzer (Jinrui Technology Co., Ltd., KT6200VET) was used to detect routine blood parameters, including white blood cells (WBC), lymphocytes (LYM), intermediate cells (MID), granulocytes (GRA), platelets (PLT), and red blood cells (RBC).

[0052] Total RNA was extracted from mouse whole blood samples using Trizol reagent (Takara Bio Inc., Kusatsu, Japan) and subsequently reverse-transcribed into cDNA using HiFi Script gDNA Removal RT Master Mix (Cat. No. CW2020M, Jiangsu Kangwei Biotechnology Co., Ltd.). mRNA levels were quantified using SYBR (Cat. No. CW3008M, Jiangsu Kangwei Biotechnology Co., Ltd.). Primer sequences are listed in Table 2.

[0053] Table 2 Detailed information of PCR primer sequences result: To evaluate whether biomarkers exhibit similar effects in a mouse model of sepsis, we established a CLP (cecal ligation and puncture) mouse injury model, which consists of two steps: cecal ligation and cecal puncture ( Figure 8 (A) The sepsis scores and rectal temperatures of mice were measured 8 or 24 hours after CLP. Figure 8 As shown in B and C, CLP injury leads to a decrease in rectal temperature and an increase in sepsis score. In addition, CLP injury also leads to a decrease in blood routine indicators such as WBC, LYM, MID, GRA, and PLT levels, while an increase in RBC levels ( Figure 8Fig. 3B. Fig. 3C. Fig. 3D. It is shown that the CLP mouse model is successfully constructed.

[0054] qRT-PCR results confirmed that MCEMP1 (the homologous gene of C19orf59 in mice), HMGB2 and TLR5 were consistently up-regulated in the CLP group compared with the sham operation group Figure 9 Fig. 3B. Fig. 3C. Fig. 3D. It is shown that the CLP mouse model is successfully constructed. Figure 9 Fig. 3B. Fig. 3C. Fig. 3D. It is shown that the CLP mouse model is successfully constructed.

[0055] The present application screens key genes related to sepsis from multiple dimensions by combining WGCNA and machine learning technology, and confirms the reliability of the key genes as biomarkers of sepsis through data set verification and animal experiments. The method can provide strong support for early diagnosis of sepsis, improve the accuracy and timeliness of diagnosis, and reduce the mortality of sepsis.

[0056] The technical features of the above-described embodiments can be combined arbitrarily. To make the description concise, all possible combinations of the technical features in the above-described embodiments are not described, but as long as the combinations of the technical features do not exist contradictions, they should be considered as the scope of the present disclosure.

[0057] The above-described embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, several modifications and improvements can be made, which are all within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. Use of a reagent for detecting biomarkers in the preparation of a product for early diagnosis or prognosis of sepsis, characterized in that: The biomarker is any one or a combination of several of TLR5, HMGB2, C19orf59, TXK and DGKA.

2. The use according to claim 1, characterized in that The biomarker is any one or a combination of TLR5, HMGB2 and C19orf59.

3. The use according to claim 1, characterized in that Down-regulation of TLR5, HMGB2, or C19orf59 expression and up-regulation of TXK or DGKA expression in the detected samples indicate a good prognosis for patients with sepsis.

4. The use according to claim 3, characterized in that The test sample is blood.

5. The use according to claim 1, characterized in that The product is a detection reagent, a test kit, a test paper, a probe or a chip.

6. Use of a reagent for detecting a biomarker in screening drugs for preventing or treating sepsis, characterized in that: The biomarker is any one or a combination of several of TLR5, HMGB2, C19orf59, TXK and DGKA.

7. A method for screening drugs for preventing or treating sepsis, characterized in that: The following steps are involved: A sepsis model is induced by drug administration, and the expression levels of the biomarkers described in claim 1 are detected and compared before and after administration. When the expression of TLR5, HMGB2 or C19orf59 in the test sample is downregulated and the expression of TXK or DGKA is upregulated, it indicates that the administered drug is the target drug.

8. The method according to claim 7, characterized in that The biomarker is any one or a combination of TLR5, HMGB2 and C19orf59.

Images