Microbial marker for predicting risk of brain glioma and application thereof

By using microbial markers and computational models such as kiwi bacteria in the risk assessment of brain glioma, the problem of not using salivary bacteria microbiome model in the existing technology is solved, and a new risk assessment tool for brain glioma is provided to improve the Predictive accuracy and clinical application value.

CN118957066AInactive Publication Date: 2024-11-15CANCER INST & HOSPITAL CHINESE ACADEMY OF MEDICAL SCI

Patent Information

Application Number
CN202411052522.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-01
Publication Date
2024-11-15
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing risk assessment of brain gliomas mainly relies on genetic markers, and fails to effectively utilize salivary microbiome models, resulting in challenges in predicting and treating brain gliomas, especially eradication of invasive tumor cells and drug resistance issues. .

Method used

A set of microbial markers, including Miracid, Bacteroides, Carbon Dioxide Fibrophila, Cili and K. Heliobacteria, were tested for their abundance through digital imaging, protein immunity, microarrays, nucleic acid sequencing and other technologies, and evaluated by computational models. Risk score, judge the high-risk group of cerebral glioma.

Benefits of technology

Filling the gap in the salivary microbiome model in the risk assessment of glioma in brain glioma, revealing the potential link between salivary microorganisms and brain glioma risks, providing new tools for the prevention and management of brain gliomas, improving Predictive accuracy and clinical application value.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure BDA0004976512110000061
    Figure BDA0004976512110000061
  • Figure BDA0004976512110000071
    Figure BDA0004976512110000071
  • Figure BDA0004976512110000111
    Figure BDA0004976512110000111
Patent Text Reader

Abstract

The present disclosure provides a microbial marker for predicting the risk of brain glioma, and a computational model based on the microbial marker. The microbial marker fills the blank of brain glioma risk assessment by using a saliva flora microbiome model, and the developed risk scoring model provides a new tool for the risk assessment of the brain glioma. The model is based on the analysis of saliva microbiome, reveals the potential relation between specific microorganisms and brain glioma risk, and provides support for the prevention and management of brain glioma.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of biomedicine technology, and in particular relates to a microbial marker for predicting the risk of brain glioma and an application thereof. Background Art

[0002] Gliomas are the most common primary tumors in the central nervous system. They are not only highly invasive, but also have a very poor prognosis. Although neurosurgery can remove part of the tumor, surgery is often unable to completely eradicate invasive tumor cells. In addition, the limited number of surgeries and the increased resistance of tumor cells to treatment have severely restricted the effective treatment of gliomas. The remaining tumor cells are difficult to effectively control with existing treatment methods, resulting in a poor prognosis for gliomas and a low overall survival rate. Even after standard resection, radiotherapy, and chemotherapy, only about 5.5% of glioma patients survive for more than 5 years.

[0003] The human oral saliva microbiota includes more than 1,000 different microorganisms, which is the second most complex microbiota in the human body and is associated with systemic diseases and the occurrence of various cancers. Oral saliva microorganisms may induce tumors through various mechanisms. First, dysbiosis of the oral saliva microbiota can lead to chronic inflammation, induce cell proliferation and mutation, and secondly, dysbiosis of the oral saliva microbiota may promote tumorigenesis by affecting the metabolic pathways of host cells, affecting cell proliferation and inhibiting apoptosis. In addition, dysbiosis of the oral saliva microbiota will produce some carcinogens, inducing malignant transformation of cells.

[0004] However, currently, glioma risk assessment mostly uses genetic markers, and there are no reports on the use of salivary flora microbiome models for glioma risk assessment. Summary of the invention

[0005] In order to solve at least one of the above problems, the present disclosure provides a group of microbial markers for predicting the risk of glioma and its application. Using the microbial markers provided by the present disclosure, the risk of glioma can be assessed.

[0006] According to a first aspect of the present disclosure, a microbial marker for predicting the risk of brain glioma is provided, wherein the microbial marker includes one or more of the following: Atopobium, Bacteroides, Capnocytophaga, Leptotrichia and Cardiobacterium.

[0007] According to a second aspect of the present disclosure, a detection kit for predicting the risk of developing brain glioma is provided, comprising: a reagent for detecting the microbial marker described in the first aspect.

[0008] In some embodiments, the reagents include reagents for detecting the abundance of the microbial markers in a sample by one or more of digital imaging technology, protein immunoassay technology, microarray technology, dye technology, nucleic acid sequencing technology, nucleic acid hybridization technology, nucleic acid amplification technology, chromatography technology, and mass spectrometry technology.

[0009] In some embodiments, the nucleic acid sequencing technology includes one or more of the following: 16S rRNA sequencing, whole genome sequencing, targeted sequencing, PCR-pyrosequencing and second-generation high-throughput sequencing.

[0010] In some embodiments, the reagent comprises a primer, a probe, an antisense oligonucleotide, an aptamer, or an antibody specific for the microbial marker.

[0011] In some embodiments, the kit includes reagents, test strips or chips.

[0012] According to a third aspect of the present disclosure, a method for evaluating the risk of glioma is provided, the method comprising: using a computational model to calculate a risk score (RiskScore) of a subject, wherein the risk score (RiskScore) = g_Atopobium×0.61733369+g_Bacteroides×2.93668876+g_Capnocytophaga×(-1.79325740)+g_Leptotrichia×(-0.07011431)+g_Cardiobact erium×(-0.64509608); wherein, the g_Atopobium, g_Bacteroides, g_Capnocytophaga, g_Leptotrichia, and g_Cardiobacterium represent the abundance of the microbial markers; the evaluation standard is: using the median of the risk score as the critical value, when the risk score ≥ the median, it is judged to be a high-risk group for glioma, and when the risk score < the median, it is judged to be a low-risk group for glioma, and the higher the score, the higher the risk of glioma for the subject.

[0013] In some embodiments, the abundance of the microbial marker is derived from the test results of a test sample from a subject.

[0014] In some embodiments, the sample to be tested includes a body fluid sample.

[0015] In some embodiments, the body fluid sample includes: saliva, whole blood, serum, plasma, milk, urine, lumbar or ventricular CSF, lymph fluid, prostatic fluid, semen, sputum, feces, tears, tumor cells, alveolar lavage fluid, sputum, pus, nasopharyngeal swab, oral swab, cerebrospinal fluid, pleural effusion, amniotic fluid, peritoneal fluid, aqueous humor, vitreous humor, vaginal discharge and their processed products.

[0016] According to a fourth aspect of the present disclosure, a system for evaluating the risk of glioma is provided, the system comprising:

[0017] (1) a data input module, used to obtain the abundance of the microbial markers in the test sample of the subject;

[0018] (2) a model calculation module, based on the data input in step (1), using a calculation model, risk score (RiskScore) = g_Atopobium × 0.61733369 + g_Bacteroides × 2.93668876 + g_Capnocytophaga × (-1.79325740) + g_Leptotrichia × (-0.07011431) + g_Cardiobacterium × (-0.64509608) for calculation; wherein, g_Atopobium, g_Bacteroides, g_Capnocytophaga, g_Leptotrichia, g_Cardiobacterium represent the abundance of the microbial markers;

[0019] (3) The result output module uses the median risk score as the critical value. When the risk score is ≥ the median, the subject is judged to be in the high-risk group for glioma. When the risk score is < the median, the subject is judged to be in the low-risk group for glioma. The higher the score, the higher the risk of glioma for the subject.

[0020] In some embodiments, the brain glioma patient data can be from a database known in the art, or from clinical samples.

[0021] In some embodiments, the sample to be tested includes a body fluid sample.

[0022] In some embodiments, the body fluid sample includes: saliva, whole blood, serum, plasma, milk, urine, lumbar or ventricular CSF, lymph fluid, prostatic fluid, semen, sputum, feces, tears, tumor cells, alveolar lavage fluid, sputum, pus, nasopharyngeal swab, oral swab, cerebrospinal fluid, pleural effusion, amniotic fluid, peritoneal fluid, aqueous humor, vitreous humor, vaginal discharge and their processed products.

[0023] According to a fifth aspect of the present disclosure, provided is the application of the microbial marker, the kit, the evaluation method or the evaluation system for non-disease diagnosis or treatment purposes in the prediction, early screening or diagnosis of brain glioma.

[0024] According to a sixth aspect of the present disclosure, provided is a use of the microbial marker described in the first aspect in preparing a kit for evaluating the likelihood of occurrence of brain glioma.

[0025] Beneficial effects:

[0026] The microbial markers for predicting the risk of glioma provided by the present disclosure, as well as the computational model based on the microbial markers, fill the gap in the use of the salivary flora microbiome model for glioma risk assessment, and demonstrate how the microbial composition in saliva may be associated with the risk of glioma. The risk scoring model described in the present disclosure provides a new tool for glioma risk assessment. The model is based on the analysis of the salivary microbiome, revealing the potential connection between specific microorganisms and glioma risk. The model can play a role in future clinical applications and provide support for the prevention and management of glioma. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 The microbial dilution curve of the saliva sample is shown, with the horizontal axis representing the amount of randomly sampled sequencing data and the vertical axis representing the number of observed OUTs.

[0028] Figure 2 The abundance structure diagram of the top 15 microbial phyla classification levels of the saliva flora of brain glioma patients and healthy controls at different levels is shown.

[0029] Figure 3 The abundance structure diagram of the top 15 microbial class classification levels of the saliva flora of brain glioma patients and healthy controls at different levels is shown.

[0030] Figure 4 The abundance structure diagram of the top 15 microbial order classification levels of the saliva flora of brain glioma patients and healthy controls at different levels is shown.

[0031] Figure 5 The abundance structure diagram of the top 15 microbial taxa in the saliva flora of brain glioma patients and healthy controls at different levels is shown.

[0032] Figure 6 The abundance structure diagram of the top 15 microbial genera in the saliva flora of brain glioma patients and healthy controls at different levels is shown.

[0033] Figure 7The abundance structure diagram of the top 15 microbial species in the saliva flora of brain glioma patients and healthy controls at different levels is shown.

[0034] Figure 8 The statistical display of characteristic microbial LDA score of LEfSe analysis results is shown.

[0035] Fig. 9 The figure shows the LEfSe analysis of microbial evolutionary branch diagram. Different circle layers radiating from the inside to the outside represent classification levels, and each node represents a species classification at that level. The higher the abundance of the species, the larger the node.

[0036] Fig.10 The ASV volcano plot of the genus-level differences in salivary microbiota between glioma patients and healthy controls is shown.

[0037] Fig.11 The network diagram analysis of the genus-level correlation of salivary microorganisms between patients with brain glioma and healthy controls is shown.

[0038] Fig.12 The figure shows the RF feature screening. The left side is the MeanDecreaseGini plot, which is used to evaluate the contribution of each species to the classification efficiency of the model. The larger the MeanDecreaseGini value, the better the classification. The right side is the relative richness difference box plot, which is used to show the richness variation between different categories.

[0039] Fig.13 The left figure shows the regression analysis based on LASSO cross-validation tuning parameter screening, and the right figure shows the optimal parameter (lambda) for drawing a vertical line.

[0040] Fig.14 The characteristic bacterial communities screened by the two algorithms are shown, and the overlapping parts are retained as the optimized characteristic bacterial communities, which are the intersections of the feature selections of the machine learning algorithms.

[0041] Fig.15 The AUC results of the validation set and training set model scores are shown, and the model exhibits good prediction efficiency. DETAILED DESCRIPTION

[0042] The construction of this model included saliva samples from 23 patients with high-grade glioma and 24 healthy controls. The salivary flora data were sequenced using paired-end Illumina sequencing. Based on the SRA raw data, the format was converted to Fastq format, and the quality of the obtained valid sequences was evaluated to obtain high-quality sequences that can be used for subsequent analysis. The QIIME2 software was used to call the denoise-paired method integrated in the DADA2 plug-in to perform denoising, merging, de-chimerizing and de-duplication processing on the paired-end sequences and generate an amplicon sequence variant (ASV) table. In order to obtain the species classification information corresponding to each ASV, the Greengenes2 reference database (http: / / greengenes.secondgenome.com / ) was used to count the community composition of each sample at each classification level, including phylum, class, order, family, genus, and species, and the distribution of community structure components was calculated. The LEfSe analysis and drawing were used to intuitively display the different species at all levels between groups. The TMM normalization preprocessing provided by the R software package edgeR was used, and the quasi-likelihood The F-test algorithm was used for differential analysis. The differential flora was screened and analyzed with a P value of <0.05. The samples were randomly divided into a training set and a validation set at a ratio of 1:1 (Table 1). Feature selection was performed in the training set according to two methods. The first method used the random forest model algorithm using the microeco package to select the characteristic flora. The second method used the glmnet package provided by the R language to obtain the model flora and model coefficients from the differential bacteria. The characteristic flora screened under the two algorithms were compared, and the overlapping parts were retained as the optimized characteristic flora. The SVM method was sampled in the validation set to construct a disease diagnosis classifier based on the final model flora. The ROC curve method in pROC (version 1.12.1) was used to evaluate the model performance of the diagnostic model.

[0043] definition

[0044] Unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as commonly used in the field to which the present invention belongs. For the purpose of interpreting this specification, the following definitions will apply, and where appropriate, terms used in the singular will also include the plural form, and vice versa.

[0045] As used herein, the articles "a," "an," and "an" include plural referents unless the context clearly dictates otherwise.

[0046] The expression "about" as used herein is understood by one of ordinary skill in the art and varies within a certain range depending on the context in which it is used. If one of ordinary skill in the art is not aware of the use of the term based on the context in which it is used, "about" will mean up to plus or minus 10% of the specified value.

[0047] The term "risk scoring model" used in this article is used to predict or evaluate the likelihood or risk of a specific event, and to estimate the risk of a disease based on a series of predictive factors. Risk scoring models usually collect and analyze large amounts of data, and use various statistical and machine learning techniques to identify which factors are associated with specific health outcomes and assign corresponding weights to these factors. When constructing a risk scoring model, potential risk factors are first identified, and then the strength of the association between these factors and disease risk is evaluated through methods such as regression analysis. These association strengths are converted into specific numerical weights to form a weighted sum scoring formula. Individuals can obtain a risk score by calculating the product of their own risk factors and the corresponding weights. The score represents the relative risk of developing a specific disease. Through accurate risk assessment, risk scoring models play a key role in improving the efficiency of early diagnosis and treatment of diseases, optimizing resource allocation, and promoting personalized medicine.

[0048] The term "ROC curve" used in this article: the full name is "Receiver Operating Characteristic Curve", which is a graphical tool used to evaluate the performance of binary classification problem models. It shows the diagnostic ability of the model by plotting the true positive rate (TPR) and false positive rate (FPR) at different thresholds. The horizontal axis of the ROC curve is the false positive rate (FPR), and the vertical axis is the true positive rate (TPR). The closer the curve is to the upper left corner, the better the performance of the model, that is, it has higher sensitivity and specificity. The area under the curve is called AUC (Area under the curve). The value of AUC is between 0 and 1, which is used to quantify the overall discriminative ability of the model. The closer the AUC value is to 1, the better the performance of the model. The ROC curve provides an intuitive framework to evaluate the performance of different models and can be used to compare models without determining the optimal threshold.

[0049] The term "amplicon sequence variant (ASV)" used in this article is used to analyze genetic diversity in microbial communities, providing a more detailed and precise way to identify and distinguish microbial sequences. The ASV identification process involves high-throughput sequencing of specific regions of DNA in the sample, and then accurately identifying single nucleotide variations, even very small sequence differences. The core advantage of ASV technology is that it can capture microbial diversity with almost complete sequence resolution, and can accurately track dynamic changes and small variations in microbial communities. The use of ASV improves the accuracy and resolution of microbial ecological research, and analyzes changes in microbial diversity and community structure at a more detailed level.

[0050] The term "LASSO-logistic regression" used in this article: It combines the LASSO method and the logistic regression model to deal with classification problems. LASSO aims to promote sparse solutions by adding regularization terms to the model, that is, to make some of the coefficient estimates of the model zero. This helps to reduce the complexity of the data while maintaining the accuracy of the model and achieve feature selection. The purpose of applying LASSO to logistic regression, namely LASSO-logistic regression, is to automatically select features through regularization techniques when dealing with classification problems while avoiding overfitting. The LASSO-logistic regression method is particularly suitable for situations where the number of features is much larger than the number of samples. It can effectively identify and retain those features that contribute most to model predictions, while ignoring other unimportant features.

[0051] The term "LEfSe analysis" used in this article: LEfSe (Linear discriminant analysis EffectSize) analysis is a statistical method for high-dimensional data, mainly used in microbiome research. By identifying features with significant differences between two or more biological conditions and considering the relative abundance of the features, LEfSe analysis combines linear discriminant analysis (LDA) and effect size measurement to evaluate which features (such as specific microorganisms) can best explain the differences between two or more conditions. It mainly includes multiple comparisons, using non-parametric statistical tests to perform multiple comparisons of features between groups to identify features that change significantly between different groups; effect size measurement, for each significant feature, an effect size is calculated to evaluate its importance in changing between different groups; linear discriminant analysis (LDA), for each feature with significant differences, LDA is performed to estimate the effect size of each feature for distinguishing differences between groups. LEfSe analysis provides a powerful tool for understanding the microbial dynamics and functions of complex biological systems, and identifies microbial composition components that are significantly enriched or lacking under specific conditions.

[0052] The term "Random Forest (RF) algorithm" used in this article is an ensemble learning method, mainly used for classification, regression and other tasks. It works by building multiple decision trees during training and outputs the average prediction value of the patterns of multiple decision trees as the final prediction result. The main features of the random forest algorithm include: 1. Ensemble learning method. Random forest is a kind of ensemble learning method. It improves the accuracy and stability of prediction by building multiple decision trees; 2. Strong anti-overfitting ability: Due to the use of the average prediction value of multiple decision trees, random forest can effectively reduce overfitting and improve the generalization ability of the model; 3. Easy to parallelize: Each decision tree in the random forest is independently constructed, so the algorithm is easy to parallelize and speed up the training process. The random forest algorithm is efficient, accurate and easy to use. Its core lies in combining the prediction capabilities of multiple decision trees to improve the accuracy of the overall model.

[0053] Examples and drawings are provided below to help understand the present invention. However, it should be understood that these examples and drawings are only used to illustrate the present invention, but do not constitute any limitation. The actual protection scope of the present invention is set forth in the claims. It should be noted that for those skilled in the art, several variations and improvements made on the basis of this invention conception all belong to the protection scope of the present invention. The reagents used without indicating the manufacturer are all conventional products that can be obtained commercially.

[0054] Example

[0055] Example 1: Sequencing data processing and dilution curve analysis

[0056] Saliva samples from 23 patients with high-grade glioma and 24 healthy controls were obtained from the NCBI database (sample information is shown in Table 1).

[0057] Table 1. Sample information

[0058]

[0059]

[0060] In the table, HC represents healthy subjects, and HGG represents patients with brain glioma.

[0061] The saliva flora data was sequenced using paired-end Illumina sequencing to generate Sequence Read Archive (SRA) raw data, and the SRA raw data was analyzed as follows:

[0062] 1. Data filtering and quality assessment: Use the Fastq-dump command included in the Fastq-dump software to convert the SRA raw data format into the Fastq format, perform quality assessment on the obtained valid sequences, and then screen out high-quality sequences suitable for subsequent analysis.

[0063] 2. Generate ASV table: Use QIIME2 software and DADA2 plug-in to process the high-quality sequences obtained above. This method uses QIIME2 software package and DADA2 plug-in to process double-end sequencing data. DADA2 is one of the main methods for denoising and identifying amplicon sequence variations (ASVs) in microbial community analysis. It uses the denoise-paired method to denoise the sequences, merge double-end sequences, remove chimeras and repeat sequences, and other steps, which promotes the transition from operational taxonomic unit (OTU) clustering based on 97% similarity to a more accurate amplicon sequence variation (ASV) analysis method. The main methods include: 1. Initialize the QIIME2 environment and activate the corresponding version; 2. Use DADA2's denoise-paired method to preprocess the input double-end sequence data, including denoising, sequence merging, removing chimeras and deduplication, and save the processing results as ASVs in the .qza format.

[0064] 3. Dilution curve analysis: Using dilution curve analysis, a certain number of individuals are randomly selected from the sample, and the number of species represented by these individuals is counted. A dilution curve is constructed based on the number of individuals selected and the number of species represented. The dilution curve is used to compare the species richness in samples with different sequencing data volumes, evaluate the sequencing data volume of the sample, and determine the composition and diversity level of the microbial community. If the species in a sample have been fully sampled, the curve will tend to be stable, indicating that increasing the sampling volume will no longer significantly increase the discovery of new species. The dilution curve analysis method provides a means of quantifying and comparing species diversity in microbial samples, and accurately determines the relationship between the number of species in the sample and the number of individuals selected ( Figure 1 ). By drawing the dilution curve and comparing the dilution curves of different samples, the differences in species diversity between samples can be intuitively displayed, the rationality of the sequencing data volume can be evaluated, and the richness of species in the samples can be indirectly reflected. The evaluation and quality control results of the saliva sample microbial dilution curve show that the curve tends to be flat, indicating that as the amount of data extracted increases, the number of OUTs detected no longer increases, indicating that the amount of sequencing data this time is reasonable.

[0065] Example 2. Taxonomic analysis and community structure component diagram

[0066] In order to map each amplicon sequence variation ASV to its corresponding species classification information, the Greengenes2 reference database (version Release 13.8, http: / / greengenes.secondgenome.com) was first selected as the reference source for species annotation. A naive Bayes classifier was trained based on the Greengenes2 database using the Fit-classifier-naive-bayes method in the q2-feature-classifier plug-in. This step generates a naive Bayes classification annotation file for subsequent classification annotation by aligning the ASV sequence with the known species sequence in the database. Next, based on the classification annotation file generated in the previous step, the classify-sklearn method in the Q2-feature-classifier plug-in is applied to determine the types of microorganisms present in the saliva flora samples, quantify the number of sequences of each microorganism in the saliva flora samples, that is, its relative abundance, effectively observe the changes in the community structure of the saliva flora samples at different classification levels, compare the community structures of multiple samples, and use taxonomic analysis methods to reveal the community composition characteristics of different groups or samples at multiple biological classification levels. The analysis results of the community structure are displayed through community bar charts, showing the types of microorganisms contained in each sample at a specific taxonomic level, showing the relative abundance of each microorganism in the sample, that is, their proportion in the community. Taxonomic analysis is used to reveal the microbial classification comparison of saliva flora samples at each classification level, revealing the species diversity and relative abundance within the community, and the relationship between species. We constructed species abundance tables at different taxonomic levels from phylum to species. The abundance tables not only include all levels of biological classification, phylum, class, order, family, genus, and species, but also reflect the relative abundance of each species (specifically each genus) of the salivary flora in different samples of glioma patients and healthy people. The abundance diagram corresponding to the abundance table is shown in the figure below. Figure 2 to Figure 7 shown.

[0067] Example 3. LEfSe analysis

[0068] LEfSe (Linear discriminant analysis effect size) analysis was used, which is a statistical method for comparing two or more groups of samples to identify species with significant differences in abundance. LEfSe analysis visually displays the differential species between different groups at all levels by drawing charts. The analysis process first performed a Kruskal-Wallis rank sum test to detect all characteristic species, and then compared the differences in species abundance between different groups to identify species with significant differences. Subsequently, a Wilcoxon rank sum test was performed to determine the species with significant differences. All subspecies of these species were further tested to see if they belonged to the same classification level to ensure the consistency of the differences. Finally, a linear discriminant analysis (LDA) was performed to accurately identify the statistically and biologically significant differential bacterial genera between the disease group and the control group. The analysis results are shown in the figure. Figure 8 and Fig. 9 As shown, bacterial genera with significant differences between glioma patients and healthy controls were revealed.

[0069] Example 4. Analysis of Differential Bacterial Species

[0070] Advanced biostatistical methods based on R language were used to process and analyze the raw amplicon sequence variation ASV data collected at the Genus level. The edgeR package in R language was used to perform data preprocessing steps and apply the TMM standardization method, which adjusts the systematic deviation in the data by reducing the influence of extreme values, thereby improving the accuracy and reliability of subsequent analysis. In order to identify ASVs that show significant differences between different conditions or treatment groups, the quasi-likelihood F-test method provided by the edgeR package was used to evaluate the differences in ASV expression through a sophisticated mathematical model, taking into account the discreteness of the data and the potential variation between samples, and providing detailed information on the differential expression of ASVs, including the estimated size and direction, and whether these differences are statistically significant.

[0071] After preprocessing, a volcano plot was drawn according to the P value < 0.05 screening standard ( Fig.10), intuitively displaying the significant differences in gene expression data, and identifying ASVs with significant differences at the genus level through analysis of ASVs. The identification of these ASVs revealed the changes in the salivary flora between patients with brain glioma and healthy controls, including 10 significantly upregulated genera and 7 significantly downregulated genera. The 10 significantly upregulated genera included g_Alloscardovia; g_Scardovia; g_Atopobium; g_Slackia; g_Bacteroides; g_Staphylococcus; g_Lact obacillus; g_Shuttleworthia; g_Acinetobacter; g_Acholeplasma; 7 significantly downregulated bacterial genera included g_Paludibacter; g_Porphyromonas; g_Capnocytophaga; g_Schwartzia; g_Selenomonas; g_Leptotrichia; g_Cardiobacterium. The differential salivary flora analysis provides a valuable research basis for understanding the development of brain glioma disease.

[0072] Example 5. Correlation network analysis

[0073] Correlation network analysis constructs a correlation network diagram by calculating the correlation between species. The NetworkX network analysis toolkit is used to calculate the node degree distribution, node connectivity and other properties of the network. Key species are found based on network properties to reflect the interaction between species in the sample. The top 50 species in terms of total abundance at the classification level are selected, and the Spearman correlation coefficient is used. The absolute value of the correlation coefficient is ≥0.5, and the significant P value is <0.05. The node size indicates the abundance of the species, and different colors indicate different species. The color of the connecting line indicates positive and negative correlation, red indicates positive correlation, and green indicates negative correlation. The thickness of the line indicates the size of the correlation coefficient. The thicker the line, the higher the correlation between the salivary flora species (specifically each bacterial genera); the more lines, the closer the connection between the salivary flora species and other species ( Fig.11 ).

[0074] from Fig.11It can be seen that Network analysis was used to detect the correlation between salivary microorganisms in healthy people and patients with glioma, and to analyze the interactions between dominant species. The analysis results showed that the dominant genera in healthy people were g_Prevotella_7, g_Prevotella, g_Streptococcus, g_Porphyromonas and g_Veillonella, while the dominant genera in patients with glioma were g_Haemophilus, g_Gemella, g_Neisseria, g_Prevotella_7, g_Granulicatella, g_Porphyromonas, g_Actinomyces, g_Veillonella and g_Prevotella.

[0075] Example 6. Screening of characteristic bacterial flora

[0076] The original ASV data were randomly divided into two equal parts including training set and validation set based on Genus level, and two different methods were adopted for feature selection in the training set. The first method was to use the microeco package (https: / / github.com / ChiLiubio / microeco) with version 1.2.0 to apply the random forest model algorithm to select the characteristic flora. The second method was to perform feature selection through the glmnet package (version 4.1-7, https: / / glmnet.stanford.edu) of R language, in which 10-fold cross validation and a logistic regression model (LASSO-logistic regression model) with the Family parameter set to "binomial" were used to identify flora with significant differences and their model coefficients. After completing the above steps, the characteristic flora screened by the two methods were compared, and their intersection was retained as the characteristic flora selected for final optimization.

[0077] LASSO-logistic regression and random forest RF algorithms were used to obtain their own characteristics. The LASSO-logistic regression method successfully identified 8 key characteristic bacterial genera from the complex microbiome data. In addition, the random forest RF algorithm was used to identify 12 characteristic bacterial genera by constructing multiple decision trees and combining their prediction results. In order to further refine and confirm the key components of these characteristic bacterial genera, the Venn diagram analysis was used on this basis to extract the intersection of the features and differential bacterial genera screened by the above methods, and 5 bacterial genera were determined as the subsequent model bacterial community, and the model was constructed by weighted summation ( Fig.12 , Fig.13 and Fig.14 ).

[0078] LASSO regression dimensionality reduction was performed on the relevant characteristic microbial markers, and a risk score model was constructed. This process mainly relied on the R package glmnet (v4.0-2). In order to build a more accurate regression model, lambda screening was first performed using cross-validation, and then the model corresponding to lamdba.min was selected. The abundance matrix of the relevant characteristic microbial markers in the model was further extracted, and the risk score of each sample was calculated based on the following formula:

[0079]

[0080] Where abundance represents the abundance of the corresponding microbial marker, β represents the regression coefficient (coef) of the corresponding marker in the LASSO regression result, RScore represents the sum of the abundance of the significantly correlated marker in each sample multiplied by the coef of the corresponding marker, i represents the sample, and j represents the microbial marker.

[0081] Substituting the five bacterial genera into the above formula, we obtained a scoring model for predicting the risk of brain glioma: risk score = g_Atopobium×0.61733369+g_Bacteroides×2.93668876+g_Capnocytophaga×(-1.79325740)+g_Leptotrichia×(-0.07011431)+g_Cardiobacterium×(-0.64509608). The median of the risk score of the validation set was used as the critical value. When the risk score was ≥ the median, it was judged as a high-risk group for glioma. When the risk score was < the median, it was judged as a low-risk group for glioma. The higher the score, the higher the risk of glioma.

[0082] Example 7. Validation and performance evaluation

[0083] Based on the five bacterial genera obtained by the above-mentioned feature screening, the diagnostic risk models were constructed using the training set and validation set samples described in Table 1, respectively, aiming to predict the risk of glioma by analyzing the composition of the saliva flora. A method based on support vector machine (SVM) was adopted to construct a diagnostic classifier for a specific disease. This process involves applying the SVM algorithm on the training data set, and the specific kernel function used is the Sigmoid kernel. In order to optimize the generalization ability of the model and ensure the accuracy of the evaluation results, in order to comprehensively evaluate the performance of the constructed diagnostic model, we further used the receiver operating characteristic (ROC) curve analysis method provided in the pROC software package (version 1.12.1, https: / / cran.r-project.org / web / packages / pROC / index.html). In order to evaluate the reliability and stability of the constructed diagnostic model, the model feature scores of the training set and the validation set were calculated to obtain the comprehensive area under the curve AUC value. By calculating the true positive rate and false positive rate of the model at different thresholds, the ability of the model to distinguish different health states is reflected, such as Fig.15 As shown in the figure, the AUC of the training set is 0.869, and the AUC of the validation set is 0.608. The research results show that the AUC value is greater than 0.5, indicating that the model has good diagnostic performance.

[0084] The technical solution of the present invention is not limited to the above-mentioned specific embodiments. All technical variations made according to the technical solution of the present invention fall within the protection scope of the present invention.

Claims

1. A microbial marker for predicting the risk of brain glioma, wherein the microbial marker comprises one or more of the following: Atopobium, Bacteroides, Capnocytophaga, Leptotrichia and Cardiobacterium.

2. A test kit for predicting the risk of brain glioma, including: A reagent for detecting the microbial marker according to claim 1.

3. The kit according to claim 2, characterized in that The reagents include reagents for detecting the abundance of the microbial markers in a sample by one or more of digital imaging technology, protein immunoassay technology, microarray technology, dye technology, nucleic acid sequencing technology, nucleic acid hybridization technology, nucleic acid amplification technology, chromatography technology, and mass spectrometry technology; Preferably, the nucleic acid sequencing technology includes one or more of the following: 16S rRNA sequencing, whole genome sequencing, targeted sequencing, PCR-pyrosequencing and second-generation high-throughput sequencing; Preferably, the reagent comprises a primer, a probe, an antisense oligonucleotide, an aptamer or an antibody specific for the microbial marker; Preferably, the kit comprises reagents, test paper or chip.

4. A method for evaluating the risk of brain glioma, the method comprising: The risk score of the subject is calculated by using a calculation model, wherein the risk score=g_Atopobium×0.61733369+g_Bacteroides×2.93668876+g_Capnocytophaga×(-1.79325740)+g_Leptotrichia×(-0.07011431)+g_Cardiobacterium×(-0.64509608), wherein the g_Atopobium, g_Bacteroides, g_Capnocytophaga, g_Leptotrichia, and g_Cardiobacterium represent the abundance of the microbial markers; The median of the risk score was used as the critical value. When the risk score was ≥ the median, the subject was judged to be in the high-risk group for glioma. When the risk score was < the median, the subject was judged to be in the low-risk group for glioma. The higher the score, the higher the risk of glioma for the subject.

5. The evaluation method according to claim 4, characterized in that: The abundance of the microbial marker comes from the test results of the test sample of the subject; Preferably, the sample to be tested includes a body fluid sample; Preferably, the body fluid sample includes: saliva, whole blood, serum, plasma, milk, urine, lumbar or ventricular CSF, lymph fluid, prostatic fluid, semen, sputum, feces, tears, tumor cells, alveolar lavage fluid, sputum, pus, nasopharyngeal swab, oral swab, cerebrospinal fluid, pleural effusion, amniotic fluid, peritoneal fluid, aqueous humor, vitreous humor, vaginal discharge and their processed products.

6. A system for evaluating the risk of brain glioma, the system comprising: (1) a data acquisition module, used to obtain the abundance of the microbial markers in the test sample of the subject; (2) a calculation module, based on the data obtained in step (1), using a calculation model, risk score = g_Atopobium × 0.61733369 + g_Bacteroides × 2.93668876 + g_Capnocytophaga × (-1.79325740) + g_Leptotrichia × (-0.07011431) + g_Cardiobacterium × (-0.64509608) for calculation; wherein, the g_Atopobium, g_Bacteroides, g_Capnocytophaga, g_Leptotrichia, and g_Cardiobacterium represent the abundance of the microbial markers; (3) The result output module uses the median risk score as the critical value. When the risk score is ≥ the median, the subject is judged to be in the high-risk group for glioma. When the risk score is < the median, the subject is judged to be in the low-risk group for glioma. The higher the score, the higher the risk of glioma for the subject.

7. The evaluation system according to claim 6, characterized in that: The sample to be tested includes a body fluid sample.

8. The evaluation system according to claim 7, characterized in that: The body fluid samples include: saliva, whole blood, serum, plasma, milk, urine, lumbar or ventricular CSF, lymph fluid, prostatic fluid, semen, sputum, feces, tears, tumor cells, alveolar lavage fluid, sputum, pus, nasopharyngeal swab, oral swab, cerebrospinal fluid, pleural effusion, amniotic fluid, peritoneal fluid, aqueous humor, vitreous humor, vaginal discharge and their processed products.

9. Use of the microbial marker according to claim 1, the kit according to claim 2 or 3, the evaluation method according to claim 4 or 5, or the evaluation system according to any one of claims 6 to 8 for non-disease diagnosis or treatment purposes in early screening, prediction, detection or prognosis of brain glioma.

10. Use of the microbial marker according to claim 1 in the preparation of a kit for evaluating the possibility of occurrence of brain glioma.

Citation Information

Patent Citations

  • Application of group of biomarkers in glioma prognosis evaluation kit or evaluation model and construction method of prognosis evaluation model

    CN115993456A

  • Marker for predicting glioma prognosis and application thereof

    CN116769914A

  • Methods of identifying cancer-associated microbial biomarkers

    WO2023034618A1

Cited By

  • Brain glioma saliva microbial marker, product and application thereof

    CN120310913A

  • Meningioma saliva microbial marker, product and application thereof

    CN120310914A

  • Brain glioma intestinal microbial marker, product and application thereof

    CN120400346A