Biomarker for gastric cancer assessment and application thereof

Abstract

The invention discloses a biomarker for gastric cancer assessment and application thereof, and belongs to the technical field of biological medicine. The biomarker comprises at least one of the following florae: a Metuquise flora, an MEblack flora, an MEpin flora, an MEblue flora, an MEylow flora, an MEtan flora, and an MEgreenylow flora. The invention further discloses a method for preparing the biomarker. The biomarker comprises at least one of the following florae: the Metuquise flora, the MEblack flora, the MEpin flora, the MEblue flora, the MEylow flora, the MEtan flora, and the MEgreenylow flora. The biomarker is screened from fecal intestinal flora, the biomarker is developed on the basis of a microorganism sequencing sequence, dependence of traditional microorganism related inventions on a database is avoided, and the influence of annotation of microorganism classification status on a result is reduced. The method provided by the invention can realize self-test of people at home, reduces the detection threshold, improves the detection convenience and efficiency, can be widely applied to noninvasive GC early screening, and has great application potential.

Description

Technical Field

[0001] The present invention relates to the field of biomedicine technology, and in particular to a biomarker for gastric cancer assessment and application thereof. Background Art

[0002] Data released by the World Health Organization (WHO) in 2024 showed that cancer is one of the leading causes of death among the global population. The incidence of gastric cancer (GC) ranks fifth among all malignant tumors, second only to lung cancer, breast cancer, colorectal cancer, and prostate cancer, with a prevalence of nearly 1 million (about 4.8%). The data found that there were nearly 10 million new deaths from malignant tumors, among which GC ranked fifth. A study published in 2021 found that there are large differences in the epidemiological trends of GC in different regions. The incidence and mortality rates of GC are highest in East Asia, while the incidence rates are lower in North America and Northern Europe. GC has become a serious public health issue and has received widespread attention.

[0003] Surgery is one of the important treatments for GC. At present, except for a few special cases, such as huge primary lesions and extensive metastasis to abdominal organs, if the patient's condition permits, laparotomy should be performed to remove the primary lesion and relieve GC symptoms. According to statistics including more than 10,000 GC surgeries, the surgical rate is about 81.8% and the total resection rate is about 49.7%. In recent years, the resection rate has increased to about 75%. In addition, radiotherapy, chemotherapy, traditional Chinese medicine treatment, immunotherapy, etc. have also made important contributions to saving patients' lives and improving their quality of life.

[0004] However, even with the continuous improvement of medical technology, the mortality rate of GC has not decreased significantly. This requires scientists to not only continue to explore treatment methods, but also continue to work hard on early detection, early diagnosis, and early treatment of GC.

[0005] The symptoms of GC and precancerous lesions are hidden and non-specific, so early GC is difficult to detect. Studies have shown that usually less than 10% of GC can be diagnosed early. According to the depth of cancer tissue infiltration, it is divided into early GC and advanced GC (mid- and late-stage GC). At present, the commonly used methods for clinical diagnosis of GC are mainly medical history inquiry, routine examination, endoscopic examination, pathological biopsy and protein-related tumor marker detection, among which the gold standard for diagnosis is pathological biopsy.

[0006] However, the methods currently used for large-scale population GC screening all have limitations. First, pathological biopsy is an invasive examination, which is usually completed at the same time as surgery. Puncture biopsy has the risk of causing tumor cell metastasis and is therefore not suitable for GC screening. Secondly, it is difficult for imaging examinations to detect small tumors. Not only that, there are also problems such as low detection efficiency and high cost, which make it difficult to use for large-scale population screening. Protein biomarkers such as alpha-fetoprotein have the disadvantage of low sensitivity. Finally, the recently approved methylation-related detection kits use blood as the test sample, which makes it impossible for people to purchase the kits online, collect specimens at home, and complete the test by mailing them back to the testing agency, raising the threshold for testing.

[0007] Therefore, the discovery of biomarkers that are non-invasive, have good auxiliary diagnostic capabilities, high detection efficiency, and can be collected by themselves is of great significance for early screening of GC, thereby improving patients' survival time and quality of life.

[0008] A large number of studies have shown that the gut microbiota (GM) is a forgotten organ of the human body. Its genome is the second genome of humans, and its function affects human health. GM plays a vital role in the occurrence and development of GC in multiple dimensions such as physiology, nutrition, and immunology. GM affects the health of the stomach in many ways, such as by regulating the immune system and affecting metabolites. Usually, feces is an important specimen for studying GM, and the subjects can collect it by themselves. By studying the composition of GM, it was found that non-invasive markers related to GC auxiliary diagnosis have biological significance.

[0009] In summary, the discovery of a non-invasive GM biomarker related to GC with high sensitivity and specificity for early screening and auxiliary diagnosis has great significance and value. Summary of the invention

[0010] In view of the fact that there is no non-invasive biomarker of GM related to GC in the above-mentioned prior art, the present invention provides a biomarker for gastric cancer assessment, which uses intestinal flora as a marker, can achieve non-invasive screening, and is used for early screening and auxiliary diagnosis of GC.

[0011] In one aspect, the present invention discloses a biomarker for gastric cancer assessment, wherein the biomarker comprises at least one of the following flora:

[0012] Meturquoise flora: a flora having one of the characteristics of the sequences shown in SEQ ID NO: 1 to SEQ ID NO: 25;

[0013] MEblack bacterial colony: a bacterial colony having one of the characteristics of the sequences shown in SEQ ID NO: 26-SEQ ID NO: 33;

[0014] MEpink bacterial colony: a bacterial colony having one of the characteristics of the sequences shown in SEQ ID NO: 34 to SEQ ID NO: 41;

[0015] MEblue bacterial colony: a bacterial colony having one of the characteristics of the sequence shown in SEQ ID NO: 60-SEQ ID NO: 72;

[0016] MEyellow flora: a flora having one of the characteristics of the sequences shown in SEQ ID NO: 79 to SEQ ID NO: 89;

[0017] MEtan bacterial group: a bacterial group having one of the characteristics of the sequence shown in SEQ ID NO: 90-SEQ ID NO: 95;

[0018] MEgreenyellow bacterial community: a bacterial community having one of the characteristics of the sequences shown in SEQ ID NO: 105-SEQ ID NO: 110.

[0019] In some of these protocols, the biomarkers also include the following flora:

[0020] MEmagenta bacterial community: a bacterial community having one of the characteristics of the sequence shown in SEQ ID NO: 42-SEQ ID NO: 48;

[0021] MEgreen bacterial colony: a bacterial colony having one of the characteristics of the sequence shown in SEQ ID NO: 49-SEQ ID NO: 59;

[0022] MEpurple bacterial colony: a bacterial colony having one of the characteristics of the sequence shown in SEQ ID NO: 73-SEQ ID NO: 78;

[0023] MEred bacterial community: a bacterial community having one of the characteristics of the sequences shown in SEQ ID NO: 96-SEQ ID NO: 104;

[0024] MEbrown flora: a flora having one of the characteristics of the sequences shown in SEQ ID NO: 111-SEQ ID NO: 124.

[0025] On the other hand, the present invention also discloses the use of the above-mentioned biomarker for gastric cancer assessment in the preparation of reagents and / or equipment for diagnosing and / or treating gastric cancer. Preferably, the biomarker is intestinal flora, and more preferably, the biomarker is present in feces.

[0026] In some of these schemes, an increased relative abundance of MEblack flora, MEturquoise flora, and / or MEyellow flora in the feces of the subject to be evaluated is used as an indication for high risk assessment of gastric cancer, and a decreased relative abundance of MEblue flora, MEgreenyellow flora, MEpink flora, and / or MEtan flora in the feces of the subject to be evaluated is used as an indication for high risk assessment of gastric cancer.

[0027] On the other hand, the present invention also discloses a kit for gastric cancer assessment, comprising reagents for detecting the relative abundance of the above-mentioned biomarkers.

[0028] On the other hand, the present invention also discloses a system for gastric cancer assessment, comprising:

[0029] A data acquisition module, used to obtain the relative abundance of the above-mentioned biomarkers in the feces of the subject to be evaluated;

[0030] A data analysis module is used to analyze the relative abundance changes of the above biomarkers, and an increase in the relative abundance of the MEblack flora, MEturquoise flora, and / or MEyellow flora, and a decrease in the relative abundance of the MEblue flora, MEgreenyellow flora, MEpink flora, and / or MEtan flora are used as high-risk indicators for gastric cancer;

[0031] The result output module is used to output the above analysis results.

[0032] In some of the schemes, in the data analysis module, the morbidity index GCi is used as an indicator for evaluating the risk of gastric cancer. When the morbidity index GCi is higher than a threshold, it is evaluated as a high risk of gastric cancer.

[0033] The morbidity index GCi is calculated according to the following formula:

[0034]

[0035] in, ,

[0036] In the formula, Including the relative abundance of Meturquoise flora, the relative abundance of MEblack flora, the relative abundance of MEpink flora, the relative abundance of MEmagenta flora, the relative abundance of MEgreen flora, the relative abundance of MEblue flora, the relative abundance of MEpurple flora, the relative abundance of MEyellow flora, the relative abundance of MEtan flora, the relative abundance of MEred flora, the relative abundance of MEgreenyellow flora, the relative abundance of MEbrown flora, the age of the subject to be evaluated and a fixed intercept of 1, For each The weight value of the contribution to the incidence index GCi, T represents and Transpose two one-dimensional matrices, e is a natural constant.

[0037] It can be understood that according to common knowledge in the art, transposing means to transform the previous matrix, and Multiply them according to their corresponding relationship and then add them, such as:

[0038]

[0039] This means that and According to their corresponding relationship, The relative abundance of the feature is multiplied by the corresponding index 2.882620, and then all the features are multiplied by the corresponding index and added together.

[0040] In some of the schemes, the threshold is obtained by collecting a number of samples in advance to form a GC group and a non-GC control group, respectively, calculating the incidence index GCi corresponding to each sample, and calculating the threshold by the Youden coefficient.

[0041] In some of the schemes, the weight value is obtained by pre-collecting a number of samples to form a GC group and a non-GC control group, respectively, calculating the corresponding incidence index of each sample, and training through a logistic regression method.

[0042] In some of the schemes, and / or, the relative abundance value is the number of sequences normalized after sequencing. Preferably, the relative abundance value is calculated by the following method: randomly selecting M sequences of a single sample, which contain N sequences of the bacterial community, and taking N / M as the relative abundance of the bacterial community; more preferably, repeating the above operation at least twice, and taking the average value of N / M calculated at least twice as the relative abundance of the bacterial community.

[0043] It is understandable that the number of sample sequences extracted can be selected according to the sample situation, sequencing depth, etc. For example, if 10,000 sequences (M=10,000) are selected, the number of sample sequences should be greater than 10,000, otherwise the sample data should be eliminated. The repeated sampling operation is to ensure the stability and accuracy of the sampling and avoid the impact of accidental deviations on the results. The specific number of times can be selected according to the actual situation and requirements, such as 1,000 consecutive samplings.

[0044] For example, 10,000 sequences after denoising and splicing of a single sample can be randomly sampled, and the sample can be continuously sampled 1,000 times. The average integer of the bacterial community in all the samples is the relative abundance of the bacterial community in the sample.

[0045] In some of the schemes, when the subject to be evaluated is a male, the threshold is 0.377-0.527, preferably 0.385, and when the subject to be evaluated is a female, the threshold is 0.487;

[0046] When the subject to be evaluated is male, MEmagenta flora, MEyellow flora, Meturquoise flora, MEblack flora, age of the subject to be evaluated, MEred flora, MEgreen flora, MEtan flora, MEpurple flora, MEblue flora, MEbrown flora, MEpink flora, MEgreenyellow flora, and fixed intercept 1 correspond to They are 2.882620±0.288, 2.812127±0.281, 2.484835±0.248, 0.403535±0.040, 0.078187±0.008, 0.054065±0.005, 0.009098±0.0001, -0.000644±0.00006, -0.011147±0.001, -0.053151±0.005, -0.142920±0.014, -0.250427±0.025, -0.893172±0.089 and -4.667147±0.466, respectively; when the subject to be evaluated is a female, The corresponding values ​​of MEturquoise flora, MEblack flora, MEpink flora, MEmagenta flora, age of the subject to be evaluated, MEgreen flora, MEblue flora, MEpurple flora, MEyellow flora, MEtan flora, MEred flora, MEgreenyellow flora, MEbrown flora, and fixed intercept 1 They are 0.753863±0.075, 0.623466±0.062, 0.398748±0.040, 0.226382±0.023, 0.053385±0.005, 0.049135±0.005, -0.018421±0.002, -0.054549±0.005, -0.068974±0.007, -0.115541±0.011, -0.178647±0.018, -0.286749±0.029, -0.309868±0.031, and -1.893126±0.189.

[0047] On the other hand, the present invention also discloses a method for evaluating gastric cancer for non-diagnostic and therapeutic purposes, comprising the following steps:

[0048] 1) Obtaining the relative abundance of the above biomarkers in the feces of the subject to be evaluated;

[0049] 2) Analyze the changes in the relative abundance of the above biomarkers to determine the gastric cancer risk of the subject to be evaluated.

[0050] On the basis of being in accordance with the common sense in the art, the above-mentioned preferred conditions can be arbitrarily combined to obtain the preferred embodiments of the present invention.

[0051] The reagents and raw materials used in the present invention are commercially available.

[0052] The positive and progressive effects of the present invention are:

[0053] The biomarker for gastric cancer assessment of the present invention is screened from the fecal intestinal flora, and can be used to screen GC using fecal samples, and peripheral blood that can be collected by personnel with blood collection qualifications is not required. In addition, the present invention develops biomarkers based on microbial sequencing sequences, avoids the dependence of traditional microbial-related inventions on databases, and reduces the impact of the classification status of annotated microorganisms on the results. The present invention can realize self-testing at home for the population, lowers the detection threshold, improves the convenience and efficiency of detection, and at the same time, the product detection sensitivity and specificity are no obvious deficiencies compared with other similar inventions. It can realize self-testing at home for the population, lowers the detection threshold, improves the convenience and efficiency of detection, and at the same time, the optimal sensitivity and specificity of the product for male gastric cancer patients can reach 100% and 91.7%, and the optimal sensitivity and specificity of the product for female gastric cancer patients can reach 90% and 85.7%. It can be widely used in non-invasive GC early screening and has great application potential. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 OPLS-DA schematic diagram of the characteristic changes in relative abundance of intestinal flora in GC patients.

[0055] Figure 2 Schematic diagrams of 12 MEs were obtained to establish the correlation coefficients between OTUs that contributed most to the OPLS-DA model.

[0056] Figure 3 The characteristic changes in the relative abundance of ME composed of 12 bacterial communities in the HC group and the GC group.

[0057] Figure 4 It is the diagnostic ability of the microbial index calculated by the bacterial disorder index NB-GDI method.

[0058] Figure 5The ability of mGCi to distinguish GC from healthy male people, male gastritis patients, and male colorectal cancer patients.

[0059] Figure 6 The ability of fGCi to distinguish female GC population from healthy female population, female gastritis patients, and female colorectal cancer patients.

[0060] Figure 7 Figure 2 is the ROC curve of mGCi and fGCi in the validation set. DETAILED DESCRIPTION

[0061] The present invention is further described below by way of examples, but the present invention is not limited to the scope of the examples. The reagents and equipment whose sources are not specified in the following examples are all conventionally commercially available; the experimental methods in the following examples without specific conditions are selected according to conventional methods and conditions, or according to the product instructions.

[0062] Example 1

[0063] In this embodiment, biomarkers based on bacterial flora are screened.

[0064] 1. Data Source and Processing

[0065] 61 healthy controls (HC), 54 gastritis patients (CG), 29 colorectal cancer patients (CRC) and 83 gastric cancer patients (GC) from the study by Zhang et al. (Zhang et al., Applied Microbiology and Biotechnology, 2021, 105:803-814) were used as the training set population for this work. The data comes from the NCBI public database (https: / / www.ncbi.nlm.nih.gov / bioproject / ?term=PRJNA639644). The bioinformatics processing of the data was carried out in the same way as described in the literature, i.e., the bacterial 16S rRNA gene V4 data from the feces of the training set population was obtained using primers 515F (5'-GTGCCAGCMGCCGCGGTAA-3', SEQ ID NO. 125) and 806R (5'-GGACTACHVGGGTWTCTAAT-3', SEQ ID NO. 126). The sequencing platform was Illumina HiSeq 4000 (2×150 bp), and the primers were removed from the original sequencing sequence using QIIME (Quantitative Insights Into Microbial Ecology, v1.9.1) to perform quality control on the sequence. The sequences were assembled using FLASH software, and the representative sequences of OTUs (Operational Taxonomic Units) were obtained using VSEARCH (v2.4.4). The specific methods are described in the study of Zhang et al.

[0066] Finally, 12,630,269 sequences were obtained from the training population, with an average of 55,886.1 sequences per sample. After rarefaction curve analysis, the data volume of all samples reached a plateau, indicating that the depth of sequencing of each sample can represent the diversity of bacteria in the sample. The data was homogenized. Specifically, 10,000 sequences were randomly selected from the denoised and spliced ​​sequences of a single sample, and the samples were continuously selected for 1,000 times. The average integer of the bacterial community in all the extractions was the relative abundance of the bacterial community in the sample. A total of 2,943 relative abundance values ​​of OTUs were obtained for subsequent analysis, where each OTU represents a set obtained by grouping according to sequence similarity. Sequences with high sequence similarity are divided into one OTU, and all sequences in the same OTU have at least one sequence among several characteristic sequences in the OTU.

[0067] 2. Data Analysis

[0068] In order to identify the bacteria with characteristic changes in relative abundance in the intestine of GC patients, the OPLS-DA analysis model was established using simca software with default parameters to compare the HC and GC, CRC and GC, and CG and GC groups, respectively.

[0069] Figure 1 The following is an OPLS-DA diagram of the characteristic changes in the relative abundance of intestinal flora in GC patients, where (A) is the OPLS-DA model between HC, CRC, CG and GC; (B) is the OPLS-DA model between HC and GC; (C) is the OPLS-DA model between CRC and GC; (D) is the OPLS-DA model between CG and GC. There were 61 cases in HC, 54 cases in CG, 29 cases in CRC, and 83 cases in GC. The results showed that compared with the HC, CRC and CG groups, the overall structure of the intestinal flora in the GC group changed significantly.

[0070] The orthogonal partial least squares discriminant analysis (OPLS-DA) model between the groups was established using R language (v4.3.1). The OTUs with the highest contribution to the differences were obtained through the OPLS-DA model. Then, considering the intersection of each group, 126 OTUs were finally selected from 2,943 OTUs for subsequent analysis.

[0071] The OTUs were correlated with each other, and the correlation (R value) of the relative abundance of each OTU was calculated using R language (v4.3.1), which was defined as an R value matrix to obtain a correlation coefficient matrix (Spearman's R) with 126 rows and 126 columns. The WGCNA data package in R language (v4.3.1) was used to perform dimensionality reduction analysis on these OTUs, and a 1-R conversion was performed. The new matrix obtained was used as the distance matrix, and then the 1-R matrix was combined with the Ward.D2 algorithm to establish a clustering tree (with 1-R as the distance). When the distance was greater than the threshold (set to 1), the OTUs would be assigned to other MEs. There were no less than 5 OTUs in the same ME, and the sum of the relative abundances of the OTUs in the ME was taken as the relative abundance of the ME. Finally, the 126 OTUs were clustered into 12 MEs (Module Eigengenes, bacterial module).

[0072] The above 12 MEs include: Meturquoise flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 1- SEQ ID NO: 25), MEblack flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 26- SEQ ID NO: 33), MEpink flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 34- SEQ ID NO: 41), MEblue flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 60- SEQ ID NO: 72), MEyellow flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 79- SEQ ID NO: 89), MEtan flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 90- SEQ ID NO: 95), MEgreenyellow flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 105- SEQ ID NO: 110), MEmagenta flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 114- SEQ ID NO: 115), MEorange flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 116- SEQ ID NO: 117), MEorange flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 118- SEQ ID NO: 119), MEorange flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 120- SEQ ID NO: 121), MEorange flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 122- SEQ ID NO: 123), MEorange flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 124- SEQ ID NO: 125), MEorange flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 126- SEQ ID NO: NO: 42- SEQ ID NO: 48), MEgreen flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 49- SEQ ID NO: 59), MEpurple flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 73- SEQ ID NO: 78), MEred flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 96-SEQ ID NO: 104), and MEbrown flora (a flora having one of the characteristics of the sequences shown in SEQ ID NO: 111- SEQID NO: 124).

[0073] It is understandable that the above bacterial community names have no special meaning and are only for the convenience of distinction. Each bacterial community is distinguished by its sequence characteristics.

[0074] Figure 2 In order to establish 12 MEs using the correlation coefficients between 126 OTUs that contributed most to the OPLS-DA model differences of HC, CRC, CG, and GC, (A) divided the 126 OTUs into 12 MEs and named them based on the Spearman correlation coefficients between OTUs using the WGCNA package in R language; (B) shows the correlation between each ME.

[0075] Furthermore, this example explored the changes in 12 MEs in GC relative to HC. In SAS software (v9.4), the Mann-Whitney test was used to analyze the differences between HC and GC in the 12 MEs in the training set population. P < 0.05 indicated a significant difference. There were 61 cases of HC and 83 cases of GC. The results are shown in Figure 3 As shown, Figure 3 Characteristic changes in the relative abundance of MEs composed of 12 bacterial communities in the GC group compared with the HC group.

[0076] The results showed that the relative abundance of MEblack, MEturquoise, and MEyellow were significantly increased in the GC group, while that of MEblue, MEgreenyellow, MEpink, and MEtan were significantly decreased in GC.

[0077] Example 2

[0078] This example uses the biomarkers of Example 1 to establish an analytical model for gastric cancer assessment.

[0079] 1. Establishment of analysis and evaluation model.

[0080] 1. Model building concept

[0081] The inventors first used the method for calculating the bacterial disorder index NB-GDI in their previous research (see CN116721696A) to analyze and calculate the above 12 bacterial flora as potential microbial markers for the diagnosis of gastric cancer. The results are as follows: Figure 4 As shown, Figure 4 The diagnostic ability of the microbial index calculated by the bacterial disorder index NB-GDI method.

[0082] The results showed that the ROC AUC of NB-GDI as a diagnostic indicator was only 0.726, and both sensitivity and specificity were not ideal.

[0083] Then, in order to improve the diagnostic ability, the inventors comprehensively considered the characteristics of clinical data and combined it with microbial information. In line with the medical theory that the occurrence and development of diseases may be driven by multiple dimensions, the evaluation model was adjusted. Through machine learning, combined with the relative abundance of 12 MEs and the age of the volunteers, the weight of MEs with significant differences between GC and HC was high, and the weight of MEs with no significant differences was low. The GC incidence index mGCi for men and the index fGCi for women were calculated respectively.

[0084] 2. Model building

[0085] The following model was established to calculate the morbidity index GCi

[0086]

[0087] in, ,

[0088] Considering the differences between men and women, a mathematical model is used to calculate the GC incidence index (mGCi or fGCi) for men and women, respectively, where z(feature) is z(male) and z(female) for men and women, β(feature) is β(male) and β(female) for men and women, and X(feature) is X(male) and X(female) for men and women, respectively. The details are as follows:

[0089] The male GC incidence index mGCi is calculated by the following formula:

[0090]

[0091] in:

[0092]

[0093] Parameters in the mGCi formula are the eigenvalues ​​of the male subjects The feature is obtained by linear combination; the feature is composed of the relative abundance of 12 MEs, the age of the individual, and an intercept with a fixed value of 1. The weight of the linear combination of the features is recorded as , which means that each feature pair The contribution of this vector is obtained by training it through logistic regression on the training set.

[0094] Specifically, the synthetic minority class oversampling (SMOTE) method is used to resample the minority class to balance the data set, and then the data is divided into the model training set and the model validation set in a ratio of 7:3, and a 10× cross-validation strategy is adopted. represents the linear score of all features in this individual, that is, The transpose of ) multiply the inner product of each eigenvalue , and then add the sum. It is then transformed into a value between 0 and 1 through the Sigmoid function, which is the mGCi index.

[0095] The female GC incidence index fGCi was calculated by the following formula:

[0096]

[0097] in:

[0098]

[0099] Parameters in the fGCi formula The eigenvalues ​​of the female subjects The feature is obtained by linear combination; the feature is composed of the relative abundance of 12 MEs, the age of the individual, and an intercept with a fixed value of 1. The weight of the linear combination of the features is recorded as , which means that each feature pair The contribution of this vector is obtained by training it through logistic regression on the training set.

[0100] Specifically, the synthetic minority class oversampling (SMOTE) method is used to resample the minority class to balance the data set, and then the data set is divided into the model training set and the model validation set in a ratio of 7:3, and a 10× cross-validation strategy is adopted. represents the linear score of all features in this individual, that is, The transpose of ) multiply the inner product of each eigenvalue , and then add the sum. It is then transformed into a value between 0 and 1 through the Sigmoid function, which is the fGCi index.

[0101] 3. Model Validation

[0102] For the validation set population, stool samples were collected using the same method, and the same experiments and data analysis were performed. The ME in the validation set was established based on the ME of the characteristic changes in the training set population, and the mGCi and fGCi of the validation set were calculated.

[0103] The ROC curve was used to demonstrate the potential diagnostic ability of mGCi and fGCi. In SAS software (v9.4), the ROC model of mGCi and fGCi was established, the area under the curve (AUC) was calculated, and the Youden coefficient was used to calculate the model threshold, as well as the corresponding sensitivity and specificity under the threshold.

[0104] Figure 5 mGCi has the ability to distinguish GC from male healthy people, male gastritis patients, and male colorectal cancer patients. Among them, (A) is the ROC analysis of the HC group and the GC group; (B) is the CRC group and the GC group; (C) is the CG group and the GC group. AUC represents the area under the curve. There were 32 male HCs, 24 male CGs, 16 male CRCs, and 55 male GCs.

[0105] The results showed that the AUC of the ROC model constructed by male HC and male GC mGCi was 0.8932, and the sensitivity and specificity were 75% and 90.9% when the threshold was 0.377; the AUC of the mGCi constructed by male CRC and male GC was 0.844, and the sensitivity and specificity were 78.2% and 100% when the threshold was 0.527; the AUC of the mGCi constructed by male CG and male GC was 0.869, and the sensitivity and specificity were 78.2% and 100% when the threshold was 0.527, indicating that mGCi has a good auxiliary diagnostic ability for male GC.

[0106] Figure 6 fGCi has the ability to distinguish female GC from healthy females, female gastritis patients, and female colorectal cancer patients. Among them, (A) is the ROC analysis of the HC group and the GC group. (B) is the CRC group and the GC group. (C) is the CG group and the GC group. AUC represents the area under the curve. There were 28 female HCs, 24 female CGs, 16 female CRCs, and 26 female GCs.

[0107] The results showed that the ROC model AUC of the fGCi constructed by female HC and female GC was 0.946, and the sensitivity and specificity were 78.6% and 100% respectively when the threshold was 0.403; the AUC of the fGCi constructed by female CRC and female GC was 0.933, and the sensitivity and specificity were 84.6.2% and 100% respectively when the threshold was 0.518; the AUC of the fGCi constructed by female CG and female GC was 0.844, and the sensitivity and specificity were 78.2% and 100% respectively when the threshold was 0.527, indicating that fGCi has a good auxiliary diagnostic ability for female GC.

[0108] The above experimental results show that this embodiment uses mGCi or fGCi to perform GC risk assessment on males or females respectively, and has good diagnostic ability.

[0109] Example 3

[0110] In this example, 30 healthy people (Validation of Healthy Control, vHC) and 90 GC patients (Validation of Gastric Cancer, vGC) were included as the validation set population, and this validation set population was derived from the NCBI public database (https: / / www.ncbi.nlm.nih.gov / bioproject / ?term=PRJNA817689). Fecal sample collection, detection, and data analysis were the same as those described in the study by Zhang et al. in Example 1. A total of 12,873,446 sequences were detected in the validation set population samples, with an average of 107,278.7 sequences per sample. After dilution curve analysis, the sequencing amount of the validation set samples reached the plateau period, and after data normalization, a total of 2,139 OTUs were obtained.

[0111] In the validation set, MEs (i.e., 12 MEs, each of which contains a strain of at least one sequence fragment in the common bacterial community) whose representative sequences have 100% similarity to the OTUs with characteristic changes in the ME composition in the training set in Example 1 were selected, and the mGCi and fGCi of each sample in this validation set were calculated using the formulas for mGCi and fGCi in Example 2.

[0112] The results are as follows Figure 7 As shown, Figure 7 The ROC curves of mGCi and fGCi in the validation set population, AUC represents the area under the line. (A) is the ROC analysis of the male HC group and the GC group, and (B) is the ROC analysis of the female HC group and the GC group. There were 20 cases in the male vHC group, 10 cases in the female vHC group, 69 cases in the male vGC group, and 21 cases in the female vGC group.

[0113] The results showed that after the corresponding ROC analysis of male members in the vHC group and vGC group (n = 20, 69), the AUC reached 0.914, and the sensitivity and specificity were 90% and 87% at a threshold of 0.385; after the corresponding ROC analysis of female members in the vHC group and vGC group (n = 10, 21), the AUC reached 0.833, and the sensitivity and specificity were 90% and 71.4% at a threshold of 0.487. These results indicate that both mGCi and fGCi have good potential for diagnosing GC.

[0114] Example 4

[0115] In this example, 28 healthy subjects (Validation of Healthy Control, vHC2), including 14 males and 12 females, and 14 GC patients (Validation of Gastric Cancer, vGC2), including 7 males and 7 females, were enrolled as the second validation set population. The validation set population was from a tertiary hospital in Shanghai, and the stool sample collection, detection, and data analysis were the same as those described in Example 3.

[0116] The results showed that when the threshold of mGCi was 0.385, the sensitivity and specificity of mGCi in distinguishing vHC2 from vGC2 were 100% and 91.7%, respectively; when the threshold of fGCi was 0.487, the sensitivity and specificity were 85.7% and 85.7%, respectively.

Claims

1. A biomarker for gastric cancer assessment, characterized in that: The biomarkers include at least one of the following bacterial groups: Meturquoise flora: a flora having one of the characteristics of the sequences shown in SEQ ID NO: 1 to SEQ ID NO: 25; MEblack bacterial colony: a bacterial colony having one of the characteristics of the sequences shown in SEQ ID NO: 26-SEQ ID NO: 33; MEpink bacterial colony: a bacterial colony having one of the characteristics of the sequences shown in SEQ ID NO: 34 to SEQ ID NO: 41; MEblue bacterial colony: a bacterial colony having one of the characteristics of the sequence shown in SEQ ID NO: 60-SEQ ID NO: 72; MEyellow flora: a flora having one of the characteristics of the sequences shown in SEQ ID NO: 79 to SEQ ID NO: 89; MEtan bacterial group: a bacterial group having one of the characteristics of the sequence shown in SEQ ID NO: 90-SEQ ID NO: 95; MEgreenyellow bacterial community: a bacterial community having one of the characteristics of the sequences shown in SEQ ID NO: 105-SEQ ID NO:

110.

2. The biomarker for gastric cancer assessment according to claim 1, characterized in that The biomarkers also include the following flora: MEmagenta bacterial community: a bacterial community having one of the characteristics of the sequence shown in SEQ ID NO: 42-SEQ ID NO: 48; MEgreen bacterial colony: a bacterial colony having one of the characteristics of the sequence shown in SEQ ID NO: 49-SEQ ID NO: 59; MEpurple bacterial colony: a bacterial colony having one of the characteristics of the sequence shown in SEQ ID NO: 73-SEQ ID NO: 78; MEred bacterial community: a bacterial community having one of the characteristics of the sequences shown in SEQ ID NO: 96-SEQ ID NO: 104; MEbrown flora: a flora having one of the characteristics of the sequences shown in SEQ ID NO: 111-SEQ ID NO:

124.

3. Use of the biomarker for gastric cancer assessment according to any one of claims 1 or 2 in the preparation of reagents and / or equipment for diagnosing and / or treating gastric cancer, preferably, the biomarker is intestinal flora, and more preferably, the biomarker is present in feces.

4. The use according to any one of claims 1 or 2, characterized in that: An increase in the relative abundance of MEblack flora, MEturquoise flora, and / or MEyellow flora in the feces of the subject to be evaluated is an indication for high risk assessment of gastric cancer, and a decrease in the relative abundance of MEblue flora, MEgreenyellow flora, MEpink flora, and / or MEtan flora in the feces of the subject to be evaluated is an indication for high risk assessment of gastric cancer.

5. A kit for gastric cancer assessment, characterized in that: Comprising reagents for detecting the relative abundance of the biomarkers of claim 1.

6. A system for gastric cancer assessment, characterized in that: include: A data acquisition module, used to obtain the relative abundance of the biomarker according to claim 2 in the feces of the subject to be evaluated; A data analysis module is used to analyze the relative abundance changes of the above biomarkers, and an increase in the relative abundance of the MEblack flora, MEturquoise flora, and / or MEyellow flora, and a decrease in the relative abundance of the MEblue flora, MEgreenyellow flora, MEpink flora, and / or MEtan flora are used as high-risk indicators for gastric cancer; The result output module is used to output the above analysis results.

7. The system for evaluating gastric cancer according to claim 6, characterized in that: In the data analysis module, the morbidity index GCi is used as an indicator for evaluating the risk of gastric cancer. When the morbidity index GCi is higher than a threshold, it is evaluated as a high risk of gastric cancer. The morbidity index GCi is calculated according to the following formula: in, , In the formula, Including the relative abundance of Meturquoise flora, the relative abundance of MEblack flora, the relative abundance of MEpink flora, the relative abundance of MEmagenta flora, the relative abundance of MEgreen flora, the relative abundance of MEblue flora, the relative abundance of MEpurple flora, the relative abundance of MEyellow flora, the relative abundance of MEtan flora, the relative abundance of MEred flora, the relative abundance of MEgreenyellow flora, the relative abundance of MEbrown flora, the age of the subject to be evaluated and a fixed intercept of 1, For each The weight value of the contribution to the incidence index GCi, T represents and Transpose two one-dimensional matrices, e is a natural constant.

8. The system for evaluating gastric cancer according to claim 7, characterized in that: The threshold is obtained by collecting several samples in advance to form a GC group and a non-GC control group, respectively calculating the morbidity index GCi corresponding to each sample, and calculating the threshold by the Youden coefficient; And / or, the weight value is obtained by pre-collecting a number of samples to form a GC group and a non-GC control group, respectively, calculating the corresponding disease index of each sample, and training by a logistic regression method; And / or, the relative abundance value is the number of sequences normalized after sequencing. Preferably, the relative abundance value is calculated by the following method: randomly selecting M sequences of a single sample, which contain N sequences of the bacterial community, and taking N / M as the relative abundance of the bacterial community; more preferably, repeating the above operation at least twice, and taking the average value of N / M calculated at least twice as the relative abundance of the bacterial community.

9. The system for evaluating gastric cancer according to claim 8, characterized in that: When the subject to be evaluated is a male, the threshold is 0.377-0.527, preferably 0.385; when the subject to be evaluated is a female, the threshold is 0.487; When the subject to be evaluated is male, MEmagenta flora, MEyellow flora, Meturquoise flora, MEblack flora, age of the subject to be evaluated, MEred flora, MEgreen flora, MEtan flora, MEpurple flora, MEblue flora, MEbrown flora, MEpink flora, MEgreenyellow flora, and fixed intercept 1 correspond to They are 2.882620±0.288, 2.812127±0.281, 2.484835±0.248, 0.403535±0.040, 0.078187±0.008, 0.054065±0.005, 0.009098±0.0001, -0.000644±0.00006, -0.011147±0.001, -0.053151±0.005, -0.142920±0.014, -0.250427±0.025, -0.893172±0.089 and -4.667147±0.466, respectively; when the subject to be evaluated is a female, The corresponding values ​​of MEturquoise flora, MEblack flora, MEpink flora, MEmagenta flora, age of the subject to be evaluated, MEgreen flora, MEblue flora, MEpurple flora, MEyellow flora, MEtan flora, MEred flora, MEgreenyellow flora, MEbrown flora, and fixed intercept 1 They are 0.753863±0.075, 0.623466±0.062, 0.398748±0.040, 0.226382±0.023, 0.053385±0.005, 0.049135±0.005, -0.018421±0.002, -0.054549±0.005, -0.068974±0.007, -0.115541±0.011, -0.178647±0.018, -0.286749±0.029, -0.309868±0.031, and -1.893126±0.

189.

10. A method for evaluating gastric cancer for non-diagnostic and therapeutic purposes, characterized in that: The following steps are involved: 1) obtaining the relative abundance of the biomarker according to claim 1 in the feces of the subject to be evaluated; 2) Analyze the changes in the relative abundance of the above biomarkers to determine the gastric cancer risk of the subject to be evaluated.

Images