Liver cirrhosis specific intestinal flora marker composition and application thereof in noninvasive diagnosis
By detecting specific gut microbiota changes in fecal samples and using machine learning models, the problem of insufficient sensitivity and specificity of existing non-invasive diagnostic methods in early-stage cirrhosis has been solved, achieving non-invasive and accurate screening and diagnosis of cirrhosis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- THE FIRST AFFILIATED HOSPITAL OF BENGBU MEDICAL COLLEGE
- Filing Date
- 2025-12-31
- Publication Date
- 2026-04-17
AI Technical Summary
Existing non-invasive diagnostic methods, such as serum biochemical tests and imaging examinations, lack sufficient sensitivity and specificity in the early diagnosis of cirrhosis. While liver biopsy is accurate, it is highly invasive, limiting its application in large-scale population screening and dynamic monitoring.
By detecting the relative abundance changes of Clostridium, Bacteroidetes, Firmicutes, Veillonella, Streptococcus, Klebsiella, and Prevotella in fecal samples, and combining machine learning models, a combination of gut microbiota biomarkers is used for non-invasive diagnosis, incorporating prior biological knowledge to improve the robustness of the diagnostic model.
It provides a non-invasive screening and diagnostic approach for liver cirrhosis, improving the ability to identify early lesions and patient compliance, avoiding the limitations of traditional methods, and enhancing the accessibility and accuracy of diagnosis.
Smart Images
Figure CN121874336A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of medical technology, specifically relating to a composition of liver cirrhosis-specific intestinal flora markers and its application in non-invasive diagnosis. Background Technology
[0002] Currently, the clinical diagnosis and assessment of cirrhosis mainly rely on serum biochemical tests (such as ALT and AST levels), imaging examinations (such as ultrasound, CT, and FibroScan), and liver biopsy. While serological markers and imaging are non-invasive or minimally invasive and easily repeatable, their sensitivity for identifying early-stage cirrhosis is limited, and their specificity is low, making it difficult to accurately reflect the early progression of the disease. Although liver biopsy is considered the "gold standard" for diagnosis, directly assessing the degree of liver fibrosis and structural changes, its invasive nature carries risks such as bleeding and infection. Furthermore, it is limited by sampling errors and the operator's experience, making it unsuitable as a routine screening and dynamic monitoring method.
[0003] However, non-invasive methods based on blood and imaging are insufficient in terms of sensitivity and specificity for early diagnosis, which may lead to missed diagnoses; while liver biopsy has high accuracy, its invasiveness, high cost and potential risks greatly limit its clinical application in large-scale population screening, early diagnosis and dynamic monitoring of disease course. Summary of the Invention
[0004] The purpose of this invention is to provide a liver cirrhosis-specific gut microbiota marker composition and its application in non-invasive diagnosis, so as to solve the problems mentioned in the background art.
[0005] To achieve the above objectives, the present invention provides the following technical solution:
[0006] A liver cirrhosis-specific gut microbiota marker composition, the composition comprising a reagent for detecting changes in the relative abundance of at least one gut microbiota selected from:
[0007] Clostridium, Bacteroidetes, Firmicutes, Veillonella, Streptococcus, Klebsiella, and Prevotella.
[0008] Preferably, the composition is used to detect relative abundance changes of at least one of the following markers:
[0009] The relative abundance of Clostridium was downregulated;
[0010] The relative abundance of Bacteroidetes was downregulated;
[0011] The relative abundance of Firmicutes was downregulated;
[0012] The relative abundance of Veillonella was upregulated;
[0013] The relative abundance of Streptococcus was upregulated;
[0014] The relative abundance of Klebsiella spp. was upregulated;
[0015] The relative abundance of Prevostia was downregulated.
[0016] Preferably, the reagent comprises one or more of the following: primer pairs capable of specifically amplifying the variable region (V3-V4 region) of the 16S rRNA gene of the gut microbiota, probes specifically binding to the microbial DNA or RNA, or antibodies capable of specifically recognizing the microbial surface antigens.
[0017] A kit for noninvasive diagnosis of liver cirrhosis, the kit comprising the gut microbiota marker composition described in any one of the above claims.
[0018] Preferably, the kit further includes one or more of the following components:
[0019] Nucleic acid extraction reagents;
[0020] PCR amplification reagents;
[0021] Hybridization detection reagents;
[0022] Sequencing library preparation reagents;
[0023] Standard or reference product.
[0024] A non-invasive method for diagnosing liver cirrhosis, the method comprising the following steps:
[0025] (a) Obtaining gut microbiota information from fecal samples of subjects;
[0026] (b) Using the kit described above, the relative abundance of at least one gut microbiota marker of claim 1 in the sample is detected;
[0027] (c) Compare the relative abundance of the biomarkers detected in step (b) with the reference abundance in the healthy control group;
[0028] (d) Based on the comparison results, determine whether the subject has cirrhosis or is at risk of developing cirrhosis.
[0029] Preferably, the method for obtaining gut microbiota information in step (a) includes metagenomic sequencing, 16S rRNA gene sequencing, quantitative PCR, gene chip, or fluorescence in situ hybridization.
[0030] Preferably, the determination in step (d) is made through a machine learning model.
[0031] The use of a gut microbiota biomarker composition in the preparation of a formulation or kit for noninvasive diagnosis of liver cirrhosis, wherein the gut microbiota biomarker composition is any one of the compositions described above.
[0032] Preferably, the non-invasive diagnosis includes screening, auxiliary diagnosis, or risk assessment for cirrhosis.
[0033] Compared with the prior art, the beneficial effects of the present invention are:
[0034] (1) By detecting the synergistic change patterns of specific bacterial groups such as Clostridium, Bacteroidetes, Firmicutes, Prevotella, Veillonella, Streptococcus and Klebsiella in fecal samples, a non-invasive solution is provided for the screening and diagnosis of cirrhosis. This avoids the limitations of traditional liver biopsy, which is invasive and imaging examinations are not sensitive to early lesions, and improves the accessibility of diagnosis and patient compliance.
[0035] (2) By incorporating known ecological interactions among microbial communities, such as cooperation and competition, into the loss function of the machine learning model in the form of a penalty term, biological prior knowledge is transformed into constraints of the model, thereby guiding the training process to be more in line with biological logic and improving the robustness and generalization ability of the diagnostic model. Attached Figure Description
[0036] Figure 1 This is a flowchart of the present invention. Detailed Implementation
[0037] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0038] Example 1:
[0039] Please see Figure 1 As shown, a liver cirrhosis-specific gut microbiota marker composition is presented. The composition comprises a reagent for detecting changes in the relative abundance of at least one gut microbiota, selected from:
[0040] Clostridium, Bacteroidetes, Firmicutes, Veillonella, Streptococcus, Klebsiella, and Prevotella.
[0041] In one embodiment of the invention, the composition is used to detect changes in the relative abundance of at least one of the following markers:
[0042] The relative abundance of Clostridium was downregulated;
[0043] The relative abundance of Bacteroidetes was downregulated;
[0044] The relative abundance of Firmicutes was downregulated;
[0045] The relative abundance of Veillonella was upregulated;
[0046] The relative abundance of Streptococcus was upregulated;
[0047] The relative abundance of Klebsiella spp. was upregulated;
[0048] The relative abundance of Prevostia was downregulated.
[0049] In one embodiment of the present invention, the reagent includes one or more of the following: primer pairs capable of specifically amplifying the variable region (V3-V4 region) of the 16S rRNA gene of the gut microbiota, probes specifically binding to microbial DNA or RNA, or antibodies capable of specifically recognizing microbial surface antigens.
[0050] A kit for non-invasive diagnosis of liver cirrhosis, the kit comprising a composition of gut microbiota markers of any of the above-mentioned types.
[0051] In one embodiment of the present invention, the kit further includes one or more of the following components:
[0052] Nucleic acid extraction reagents;
[0053] PCR amplification reagents;
[0054] Hybridization detection reagents;
[0055] Sequencing library preparation reagents;
[0056] Standard or reference product.
[0057] Specifically, nucleic acid extraction reagents may include lysis buffers (such as buffers containing SDS and EDTA), proteinase K, binding beads (such as silica membrane columns or magnetic beads), and elution buffers (such as TE buffer) for extracting high-quality DNA from fecal samples.
[0058] PCR amplification reagents may include Taq polymerase, dNTPs, buffer (containing magnesium ions), and primer mixture;
[0059] Hybridization detection reagents may include hybridization buffers (such as SSC buffer), washing buffers (such as SDS solution), and detection substrates (such as chemiluminescent substrates).
[0060] Sequencing library preparation reagents may include transposases (for tagging), adaptors, and index primers for library preparation on Illumina or PacBio platforms;
[0061] Standards or controls may include microbial DNA of known abundance (such as simulated community samples, such as the ZymoBIOMICS Microbial Community Standard) or positive / negative control samples.
[0062] A non-invasive method for diagnosing liver cirrhosis, comprising the following steps:
[0063] (a) Obtaining gut microbiota information from fecal samples of subjects;
[0064] (b) Using the kit, the relative abundance of at least one gut microbiota marker of claim 1 in a sample;
[0065] (c) Compare the relative abundance of the biomarkers detected in step (b) with the reference abundance in the healthy control group;
[0066] (d) Based on the comparison results, determine whether the subject has cirrhosis or is at risk of developing cirrhosis.
[0067] Specifically, in step (a), fecal samples should be collected using sterile containers and immediately frozen at -80°C until DNA extraction to avoid changes in the microbial community composition;
[0068] Information about the gut microbiota can be obtained through various methods, such as metagenomic sequencing (using the Illumina NovaSeq platform), 16S rRNA gene sequencing, quantitative PCR, gene chips, or fluorescence in situ hybridization (using species-specific probes).
[0069] In step (b), relative abundance can be determined by calculating the proportion of each marker's sequence readings to the total sequence readings, or by converting the Ct value of quantitative PCR to relative abundance;
[0070] In step (c), the reference abundance of the healthy control group can be obtained from a public database or a locally established database of healthy people (including individuals matched for age and sex).
[0071] In step (d), the determination can be based on a predefined threshold or scoring system. For example, if the abundance change direction of multiple biomarkers is consistent with the expectation (e.g., more than 3 biomarkers are consistent), then it is determined to be a high risk of cirrhosis; or a risk score can be calculated using a logistic regression model, and if the score is higher than 0.5, it is determined to be positive.
[0072] In one embodiment of the present invention, the method for obtaining gut microbiota information in step (a) includes metagenomic sequencing, 16S rRNA gene sequencing, quantitative PCR, gene chip or fluorescence in situ hybridization.
[0073] In one embodiment of the present invention, the determination in step (d) is performed using a machine learning model.
[0074] The use of a gut microbiota marker composition in the preparation of a formulation or kit for non-invasive diagnosis of liver cirrhosis, wherein the gut microbiota marker composition is any one of the above-mentioned compositions.
[0075] Based on the application of the above-mentioned liver cirrhosis-specific gut microbiota marker composition, non-invasive diagnosis includes screening, auxiliary diagnosis, or risk assessment of liver cirrhosis.
[0076] Example 2:
[0077] refer to Figure 1 As shown, a microbial interaction penalty term reflecting the known ecological relationships among gut microbiota is introduced to incorporate prior biological knowledge into the model, thereby training a diagnostic model. The specific steps include:
[0078] 1. Data Preparation and Feature Engineering
[0079] Metagenomic data were collected from 1200 human fecal samples, including 600 patients with cirrhosis (Child-Pugh A) and 600 healthy controls, which were randomly divided into training, validation and test sets in a ratio of 8:1:1.
[0080] Based on sequencing data, the relative abundance of all target bacterial groups (such as Clostridium, Veillonella, etc.) in each sample is calculated to form a feature vector. ;
[0081] Based on public databases such as KEGG and MetaCyc, as well as published literature, a prior knowledge matrix is constructed. Matrix quantification of the ecological relationships between different bacterial communities:
[0082] When microorganisms and When the relationship is known to be cooperative (e.g., mutual support), ;
[0083] When microorganisms and When the relationship is known to be competitive (e.g., competing for the same nutrients), ;
[0084] When the relationship is unknown or neutral ;
[0085] 2. Innovative Algorithm Model and Loss Function Design
[0086] Based on a multilayer perceptron (MLP) architecture, a custom loss function is used. It consists of two parts.
[0087] ;
[0088] It is the standard cross-entropy loss, used to measure the difference between the model's predicted label and the true label, ensuring classification accuracy;
[0089] This is a penalty term for microbial interactions;
[0090] It is a hyperparameter used to balance the weights of the two terms;
[0091] Interaction penalty term The specific formula is as follows:
[0092] ;
[0093] As can be seen from the above, the core lies in the computational model. For the sample Predicted probability of liver cirrhosis category Relative to input features (i.e., microbial abundance) and The partial derivative (i.e., gradient) of the microorganism; this gradient can be interpreted as the "contribution sensitivity" of the microorganism to the risk of cirrhosis.
[0094] when, (Cooperation): Two cooperating bacteria have similar "contribution directions" to the model decision. If their contribution sensitivity differs greatly (the difference in gradient is large), the penalty term will increase to punish this behavior that violates known biological laws.
[0095] when (Competition): Two competing bacteria contribute to the model decision in different directions. If their contribution sensitivities are similar (the difference in gradient is small), the penalty term will also increase.
[0096] By minimizing During the learning process, the guided model spontaneously learns decision rules consistent with known microbial ecological relationships.
[0097] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A liver cirrhosis-specific gut microbiota marker composition, characterized in that, The composition comprises a reagent for detecting changes in the relative abundance of at least one gut microbiota, wherein the gut microbiota is selected from: Clostridium, Bacteroidetes, Firmicutes, Veillonella, Streptococcus, Klebsiella, and Prevotella.
2. The liver cirrhosis-specific intestinal flora marker composition according to claim 1, characterized in that: The composition is used to detect changes in the relative abundance of at least one of the following markers: The relative abundance of Clostridium was downregulated; The relative abundance of Bacteroidetes was downregulated; The relative abundance of Firmicutes was downregulated; The relative abundance of Veillonella was upregulated; The relative abundance of Streptococcus was upregulated; The relative abundance of Klebsiella spp. was upregulated; The relative abundance of Prevostia was downregulated.
3. The liver cirrhosis-specific intestinal flora marker composition according to claim 1, characterized in that: The reagents include one or more of the following: primer pairs capable of specifically amplifying the variable region (V3-V4 region) of the 16S rRNA gene of the gut microbiota, probes specifically binding to the microbial DNA or RNA, or antibodies specifically recognizing the microbial surface antigens.
4. A reagent kit for non-invasive diagnosis of liver cirrhosis, characterized in that, The kit comprises the gut microbiota marker composition according to any one of claims 1 to 3.
5. The kit for non-invasive diagnosis of liver cirrhosis according to claim 4, characterized in that: The kit also includes one or more of the following components: Nucleic acid extraction reagents; PCR amplification reagents; Hybridization detection reagents; Sequencing library preparation reagents; Standard or reference product.
6. A non-invasive method for diagnosing liver cirrhosis, characterized in that, The method includes the following steps: (a) Obtaining gut microbiota information from fecal samples of subjects; (b) Using the kit described above, the relative abundance of at least one gut microbiota marker of claim 1 in the sample is detected; (c) Compare the relative abundance of the biomarkers detected in step (b) with the reference abundance in the healthy control group; (d) Based on the comparison results, determine whether the subject has cirrhosis or is at risk of developing cirrhosis.
7. The method for non-invasive diagnosis of liver cirrhosis according to claim 6, characterized in that: The methods for obtaining gut microbiota information described in step (a) include metagenomic sequencing, 16S rRNA gene sequencing, quantitative PCR, gene chip, or fluorescence in situ hybridization.
8. The method for non-invasive diagnosis of liver cirrhosis according to claim 6, characterized in that: The judgment in step (d) is made through a machine learning model.
9. The use of an intestinal flora marker composition in the preparation of a formulation or kit for noninvasive diagnosis of liver cirrhosis, characterized in that, The gut microbiota marker composition is the composition according to any one of claims 1 to 3.