HK1240281B - Biomarkers for colorectal cancer related diseases - Google Patents

Description

Biomarkers of colorectal cancer-related diseases

CROSS-REFERENCE TO RELATED APPLICATIONS

none

field

The present invention relates to biomarkers and methods for predicting the risk of microbial-associated diseases, particularly colorectal cancer and advanced adenomas in the colorectum.

background

Colorectal cancer (CRC) is one of the top three most frequently diagnosed cancers in the world and is the main cause of cancer death. Its incidence rate is higher in more developed countries, but in historically low-risk areas such as East Asia, Spain and Eastern Europe, due to the so-called Western lifestyle, the incidence rate is rapidly rising. In the development of colorectal cancer, genetic changes have accumulated for many years, generally involving the loss of tumor suppressor gene adenomatous polyposis coli gene (APC), and subsequently occurring in KRAS, PIK3CA and TP53 activation or inactivation mutations (Brenner, H., Kloor, M. & Pox, C.P. Colorectal cancer. Lancet 383, 1490-502 (2014), incorporated herein by reference). Although most CRC cases are sporadic, abnormal adenomas usually occur before their occurrence, and the abnormal adenomas can progress to malignant forms, which is called adenoma-cancer sequence. Early diagnosis of colorectal adenoma and colorectal cancer not only helps to prevent death, but also helps to reduce the cost of surgical intervention.

CRC is one of the most studied diseases associated with the gut microbiota. However, the causal relationship of the disease is usually studied by administering a cocktail of antibiotics, which eliminates the gut microbiota without knowing the exact microbial strains and genes that are at work. Fusobacterium has been detected in colorectal cancer compared to normal colon tissue and has been found to be enriched in adenomas. Fusobacterium nucleatum, a periodontal disease pathogen, has been found to promote bone marrow infiltration of intestinal tumors in Apc ^Min/+ mice and is associated with upregulation of pro-inflammatory genes such as Ptgs2 (COX-2), Scyb1 (IL8), Il6, Tnf (TNFα), and Mmp3 in mice and humans (Kostic, AD et al., Fusobacterium nucleatum potentiates intestinal tumorigenesis and modulates the tumor-immune microenvironment. Cell Host Microbe 14, 207-215 (2013), incorporated herein by reference). However, it is not yet clear whether more bacteria or archaea can serve as markers or contributors to the etiology of colorectal cancer.

Current CRC detection, such as flexible sigmoidoscopy and colonoscopy, is invasive, and the patient may feel uncomfortable or unhappy during the detection process and intestinal preparation. The interaction between intestinal microbiome and immune system plays an important role in many diseases of the intestine and intestine (Cho, I. & Blaser, M.J. The human microbiome: at the interface of health and disease. Nature Rev. Genet. 13, 260-270 (2012), incorporated herein by reference). Intestinal microbiome analysis of fecal DNA has the potential to be used as non-invasive detection to find specific biomarkers that can be used as screening tools for early diagnosis of CRC patients, thereby obtaining longer survival time and better quality of life.

summary

The embodiments of the present disclosure are intended to solve at least one problem existing in the prior art to at least some extent.

The present invention is based on the following findings of the inventors:

The assessment and characterization of intestinal microbiota has become the main research field of human diseases (including colorectal cancer). The present inventors carried out metagenomic whole genome shotgun sequencing for the first time for stool samples from healthy controls, colorectal cancer and adenoma patients. In order to analyze the intestinal microbial content in colorectal cancer and adenoma patients, the present inventors carried out metagenomic association analysis (Metagenome-Wide Association Study) (MGWAS) scheme (Qin, J. et al. Ametagenome-wide association study of gut microbiota in type 2diabetes.Nature 490,55-60 (2012), incorporated herein by reference). In order to compare the fecal microbial communities of healthy controls, progressive adenoma groups and cancer patient groups, the genes whose relative abundance showed significant differences between any two groups were identified (p < 0.05, Kruskal-Wallis test). These marker genes were then clustered into MLGs (metagenomic linkage groups) (Qin et al., 2012, supra) based on their abundance changes in all samples, and the inventors identified the MLG characteristics of these tumors. The inventors then identified and validated 15 MLGs for early and non-invasive diagnosis of colorectal cancer, and 10 MLGs for early and non-invasive diagnosis of colorectal adenomas. In order to exploit the potential of these gut microbiota-based CRC classifications, the inventors calculated the probability of disease using random forest models based on 15 MLGs and 10 MLGs, respectively. The inventors' data provide insightful insights into the characterization of gut metagenomes associated with CRC risk, and also provide a paradigm for future studies of the role of gut metagenomes in the pathophysiology of other related conditions, while also revealing the potential use of gut-microbiota-based methods for assessing individuals at risk of such conditions.

It is believed that the above-mentioned 15 MLG and 10 MLG are of great value for improving the early detection of CRC for the following reasons. First, the markers of the present invention are more specific and sensitive than conventional markers. Second, stool analysis has accuracy, safety, economic affordability and patient compliance. Stool samples are transportable. Compared with colonoscopy that requires intestinal preparation, the present invention relates to a comfortable and non-invasive in vitro method, so it is easier for people to participate in a given screening program. Third, the markers of the present invention can also be used as a treatment monitoring tool for CRC patients to detect the response to treatment.

Thus, in a first aspect, the present invention provides a biomarker panel for predicting or diagnosing a disease associated with a microbiota in a subject or determining whether a subject is at risk of developing said disease.

In a second aspect, the present invention provides a kit for predicting or diagnosing a disease associated with a microbiota in a subject, or determining whether a subject has a risk of developing the disease, comprising reagents for determining the level or amount of each biomarker of the biomarker panel of the present invention in a sample.

In a third aspect, the present invention provides the use of a reagent for determining the level or amount of each biomarker of the biomarker panel of the present invention in the preparation of a kit for predicting or diagnosing a disease associated with a microbiome in a subject or for determining whether a subject has a risk of developing the disease.

In a fourth aspect, the present invention provides a method for predicting or diagnosing a disease associated with a microbiota in a subject or determining whether a subject is at risk of developing the disease.

BRIEF DESCRIPTION OF THE DRAWINGS

Various aspects and advantages of the present disclosure will become apparent and readily understood from the following description taken in conjunction with the accompanying drawings, in which:

Figure 1 shows that intestinal MLGs are able to classify colorectal cancer samples from healthy control samples. (a) Distribution of errors in 5 10-fold cross validations of random forest classification of cancer as the number of MLGs increases. The model was trained using the relative abundance of MLGs (>100 genes) in control and cancer samples (n=55 and 41). The black curve represents the average of 5 validations (gray line). The black straight line indicates the number of MLGs in the optimal set (15 MLGs) (Table 2-1, Table 2-2). Even when age and BMI factors are taken into account together with MLGs, the same MLGs are still screened. (b) Box-and-whisker plot of the probability of cancer in the cross-validation training set according to the model in (a). (c) Receiver operating curve (ROC) of the training set. At a cut-off of 0.5, the AUC was 98.34% and the 95% confidence interval (CI) was 96.29-100%. (d) Classification results of the test set consisting of 8 control samples (black squares), 47 advanced adenoma samples (open circles), and 5 cancer samples (solid black circles). (e) ROC of the test set. At a critical value of 0.5, the AUC was 96%, and the 95% CI was 87.88-100%. If the probability of cancer is ≥ 0.5, the subject is at risk of colorectal cancer. The results in Figure 1 indicate that the above 15 MLGs can be used as biomarkers for diagnosing colorectal cancer and/or determining the risk of colorectal cancer with high sensitivity and specificity.

Figure 2 shows that intestinal MLG can classify advanced adenoma samples from healthy control samples. (a) Distribution of errors in 5 10-fold cross validations of random forest classification of advanced adenoma as the number of MLG increases. The model was trained using the relative abundance of MLG (>100 genes) in the control group and advanced adenoma samples (n=55 and 42). The black curve represents the average of 5 validations (gray line). The black straight line indicates the number of MLGs in the optimal set (10 MLGs) (Table 6-1, Table 6-2, Table 7). Even when age and BMI factors are considered together with MLG, the same MLGs are still screened. (b) Box-and-whisker plot of the probability of advanced adenoma in the cross-validation training set according to the model in (a). (c) Receiver operating curve (ROC) of the training set. At a critical value of 0.5, the AUC was 87.38% and the 95% confidence interval (CI) was 80.21-94.55%. (d) Classification results of the test set consisting of 15 control samples (open circles) and 15 advanced adenoma samples (solid black circles). (e) ROC of the test set. When the optimal cutoff value was 0.4572, the AUC was 90.67%, the true positive rate (TPR) was 1, and the false positive rate (FPR) was 0.2667. If the probability of colorectal adenoma is ≥ 0.4572 (optimal cutoff value), the subject is at risk of colorectal adenoma. The results in Figure 2 show that the above 10 MLGs can be used as biomarkers for diagnosing advanced adenoma and/or determining the risk of advanced adenoma with high sensitivity and specificity.

Details

The terms used herein have the meanings commonly understood by those skilled in the art in the art to which the present invention relates. However, in order to better understand the present invention, the definitions and explanations of the relevant terms are as follows.

Terms such as "a," "an," and "the" are not intended to refer to only a singular entity but also include a species of which specific examples may be used.

According to the present invention, the term "biomarker" (also referred to as "biological marker") refers to a measurable indicator of a biological state or condition of a subject. Such biomarkers can be any substance in a subject, such as nucleic acid markers (e.g., DNA), protein markers, cytokine markers, chemokine markers, carbohydrate markers, antigen markers, antibody markers, species markers (species/genus markers) and functional markers (KO/OG markers), etc., as long as they are related to a specific biological state or condition (e.g., disease) of the subject. Biomarkers are usually measured and evaluated to detect normal biological processes, pathological processes, or pharmacological responses to therapeutic interventions, and biomarkers are useful in many scientific fields.

According to the present invention, the term "biomarker panel" refers to a group of biomarkers (ie, a combination of two or more biomarkers).

According to the present invention, the term "microbiota-related disease" refers to a disease associated with an imbalance of the microbiota in the intestine. For example, the disease may be caused, induced, or exacerbated by an imbalance of the microbiota in the intestine. Such a disease may be an advanced adenoma of the colorectal gland or a colorectal malignancy/cancer.

According to the present invention, the term "subject" refers to an animal, particularly a mammal, such as a primate, preferably a human.

According to the present invention, the expression "colorectal cancer" has the same meaning as "colorectal carcinoma".

According to the present invention, the expressions "advanced adenoma" and "advanced adenoma of the colorectum" have the same meaning as "advanced adenoma in colorectal cancer".

According to the present invention, the expressions "cutoff" and "cut-off" have the same meaning and refer to a predicted cutoff value. The predicted cutoff value can be obtained by routine experimentation (e.g., by parallel measurement of the relative abundance of biomarkers in samples from subjects with known physiological states).

According to the present invention, the term "MLG" is defined as a group of genetic material in a metagenome that is physically likely to be connected to form a unit rather than independently distributed (see, Qin, J. et al. Ametagenome-wide association study of gut microbiota in type 2 diabetes. Nature 490, 55-60 (2012), the entire contents of which are incorporated herein by reference). MLGs eliminate the need to fully determine the specific microbial species present in a metagenome, which is very important because there are still a large number of unknown organisms and there is frequent lateral gene transfer (LGT) between bacteria. In the present invention, MLGs refer to a group of genes with consistent abundance levels and taxonomic assignments.

According to the present invention, the term "specific fragment" of an MLG is a fragment of an MLG that is unique to that MLG. Conventional methods can be used to determine whether a fragment is unique to the MLG from which it is derived. For example, the sequence of the fragment can be entered into a public database (such as GenBank) and a BLAST program can be performed. If the fragment is only present in one species in the database (in which case, this MLG will represent or correspond to that species), or if there is no homolog with at least 90% identity (such as 95% identity) to the fragment in the database (in which case, the MLG will refer to an unknown species), then the fragment can be considered unique. As discussed above, an MLG generally refers to a specific microbial species (known or unknown), so a "specific fragment" of an MLG can also be considered to be a unique genomic fragment of a specific microbial species (i.e., the fragment is only present in a specific microbial species).

According to the present invention, the term "identity" refers to the degree of match between two polypeptides or between two nucleic acids. When two sequences being compared have the same base or amino acid monomer subunit at a certain site (for example, adenine at a certain site in each of the two DNA molecules, or lysine at a certain site in each of the two polypeptides), the two molecules are identical at that site. The percent identity between two sequences is a function of the number of matching sites shared by the two sequences divided by the total number of sites compared × 100. For example, if 6 out of 10 sites in two sequences match, the two sequences have 60% identity. For example, the DNA sequences: CTGACT and CAGGTT have 50% identity (3 out of 6 sites match). Typically, the comparison of two sequences is performed in a manner that produces maximum identity. This alignment can be performed using a computer program based on the method of Needleman et al. (J. Mol. Biol. 48:443-453, 1970), such as the Align program (DNAstar, Inc.).

According to the present invention, the expression "reagent for determining the level or amount of a biomarker" refers to a reagent that can be used to quantify or measure the level or amount of a biomarker in a sample. Based on the sequence of the biomarker provided by the present invention, such reagents can be easily designed or obtained by conventional methods well known in the art. For example, such reagents include, but are not limited to, PCR primers that can be used to quantify or measure the level or amount of a biomarker by, for example, real-time PCR; probes that can be used to quantify or measure the level or amount of a biomarker by, for example, quantitative Southern blotting; microarrays (e.g., gene chips) that can be used to quantify or measure the level or amount of a biomarker. In addition, as known in the art, second-generation sequencing methods or third-generation sequencing methods can also be used to quantify or measure the level or amount of a biomarker. Therefore, such reagents can also be commercially available reagents for performing second-generation sequencing methods or third-generation sequencing methods.

According to the present invention, the expression "capable of specifically amplifying" a primer of a specific nucleic acid or a specific sequence means that when used for amplification (e.g., PCR amplification), the primer specifically anneals to the specific nucleic acid or sequence and produces a unique amplification product (i.e., does not anneal to other nucleic acids or sequences or produce other by-products).

According to the present invention, the expression "a probe capable of specifically hybridizing to a specific nucleic acid or a specific sequence" means that when used for hybridization or detection under stringent conditions, the probe specifically anneals to and hybridizes with the specific nucleic acid or sequence, but does not anneal to or hybridize with other nucleic acids or sequences.

It is common knowledge for those skilled in the art to design primers or probes based on specific sequences (such as specific MLG or specific fragments thereof). For example, such common knowledge can be found in various textbooks (see, for example, J. Sambrook et al., Molecular Cloning: Laboratory Manual, 2nd ed., Cold Spring Harbor Laboratory Press, 1989; F. M. Ausubel et al., Short Protocols in Molecular Biology, 3rd ed., John Wiley & Sons, Inc.); and in many papers, such as Buck et al. (1999), Lowe et al. (1990), etc.

According to the present invention, the term "second-generation sequencing method" refers to a new generation of DNA sequencing methods developed in recent years, including, for example, Illumina GA, Roche 454, and ABI Solid; and is different from traditional sequencing methods (such as Sanger sequencing). Second-generation sequencing methods differ from traditional sequencing methods (such as Sanger sequencing) in that second-generation sequencing methods analyze DNA sequences through sequencing-by-synthesis. Second-generation sequencing methods have the following advantages: 1) low cost, which is 1% of the cost of traditional sequencing methods; 2) high throughput, capable of sequencing multiple samples simultaneously, and a single Solexa sequencing run can generate approximately 50 billion (50G) bases of data; 3) high accuracy (greater than 98.4%), effectively solving the problem of multiple repeat sequence reads. On the other hand, when the number of sequences to be sequenced is predetermined, high sequencing throughput increases the sequencing depth of the sequences (for example, each sequence can be sequenced multiple times), thereby ensuring the credibility of the sequencing results.

According to the present invention, the term "third generation sequencing method" refers to a recently developed new generation of single molecule sequencing technology. Third generation sequencing technology offers advantages over current sequencing technologies, including (i) higher throughput; (ii) shorter turnaround time (e.g., sequencing a metazoan genome with high coverage in minutes); (iii) longer sequencing lengths to enhance de novo assembly and enable direct detection of haplotypes and even whole chromosome phasing; (iv) higher consistent accuracy to enable rare variant detection; (v) small amounts of starting material (theoretically, only a single molecule is required for sequencing); and (vi) low cost, with achieving high coverage sequencing of the human genome for less than $100 becoming a reasonable goal for society. For more details on third-generation sequencing methods, see, for example, Eric E. Schadt et al., A window into third-generation sequencing, Human Molecular Genetics, 2010, Vol. 19, Review Issue 2, R227-R240, incorporated herein by reference.

According to the present invention, the term "relative abundance" has a conventional meaning known in the art and can be calculated by methods known in the art. For example, the relative abundance of a gene (i.e., a biomarker) or MLG can be measured or calculated by the method disclosed in Qin, J. et al., A metagenome-wide association study of gut microbiota in type 2 diabetes. Nature 490, 55-60 (2012) (incorporated herein by reference).

Those skilled in the art will understand that the above term definitions are provided for a better understanding of the present invention, but the above term definitions are not intended to limit the present invention, except as outlined in the claims.

In a first aspect, the present invention provides a biomarker panel for predicting or diagnosing a disease associated with a microbiota in a subject or determining whether a subject has a risk of developing the disease, comprising the following biomarkers (all biomarkers are listed in Table 7):

(1) MLG 317 or one or more specific fragments thereof, wherein MLG 317 consists of SEQ ID NOs: 933-1052;

(2) MLG 3770 or one or more specific fragments thereof, wherein MLG 3770 consists of SEQ ID NOs: 1053-1281; and

(3) MLG 3840 or one or more specific fragments thereof, wherein MLG 3840 consists of SEQ ID NOs: 238-639;

Optionally, the biomarker panel further comprises one or more of the following biomarkers:

(4) MLG 665 or one or more specific fragments thereof, wherein MLG 665 consists of SEQ ID NOs: 640-932;

(5) MLG 721 or one or more specific fragments thereof, wherein MLG 721 consists of SEQ ID NOs: 120-237;

(6) MLG 1738 or one or more specific fragments thereof, wherein MLG 1738 consists of SEQ ID NOs: 1471-2436;

(7) MLG 1340 or one or more specific fragments thereof, wherein MLG 1340 consists of SEQ ID NOs: 2893-3067;

(8) MLG 5954 or one or more specific fragments thereof, wherein MLG 5954 consists of SEQ ID NOs: 1-119;

(9) MLG 711 or one or more specific fragments thereof, wherein MLG 711 consists of SEQ ID NOs: 1282-1470; and

(10) MLG 4668 or one or more specific fragments thereof, wherein MLG 4668 consists of SEQ ID NOs: 2437-2892.

In a preferred embodiment, the biomarker panel of the present invention comprises the biomarkers as defined in (1)-(8).

As known to those skilled in the art, a specific fragment may be of any length, so long as such fragment is unique to the MLG from which it is derived or the species represented by the MLG (i.e., the fragment is not present in other MLGs or other species). However, for convenience, the length of the specific fragment may be at least 30 bp, or at least 40 bp, or at least 50 bp, or at least 60 bp, or at least 70 bp, or at least 80 bp, or at least 90 bp, or at least 100 bp, or at least 150 bp, or at least 200 bp, or at least 250 bp, or at least 300 bp, or at least 350 bp, or at least 400 bp, or at least 450 bp, or at least 500 bp, or at least 600 bp, or at least 700 bp, or at least 800 bp, or at least 900 bp, or at least 1000 bp, or at least 1500 bp, or at least 2000 bp.

For example, in preferred embodiments, the biomarker panels of the present invention may also be characterized by any one or more of the following:

(1) the one or more specific fragments of MLG 5954 are selected from SEQ ID NOs: 1-119 or any combination thereof;

(2) the one or more specific fragments of MLG 721 are selected from SEQ ID NOs: 120-237 or any combination thereof;

(3) the one or more specific fragments of MLG 3840 are selected from SEQ ID NOs: 238-639 or any combination thereof;

(4) the one or more specific fragments of MLG 665 are selected from SEQ ID NOs: 640-932 or any combination thereof;

(5) the one or more specific fragments of MLG 317 are selected from SEQ ID NOs: 933-1052 or any combination thereof;

(6) the one or more specific fragments of MLG 3770 are selected from SEQ ID NOs: 1053-1281 or any combination thereof;

(7) the one or more specific fragments of MLG 711 are selected from SEQ ID NOs: 1282-1470 or any combination thereof;

(8) the one or more specific fragments of MLG1738 are selected from SEQ ID NOs: 1471-2436 or any combination thereof;

(9) the one or more specific fragments of MLG 4668 are selected from SEQ ID NOs: 2437-2892 or any combination thereof; and

(10) The one or more specific fragments of MLG1340 are selected from SEQ ID NOs: 2893-3067 or any combination thereof.

In a preferred embodiment, the disease is advanced adenoma in the colorectum.

In preferred embodiments, the subject is a mammal, such as a primate, preferably a human.

In a preferred embodiment, the biomarker panels of the present invention are used to distinguish patients with advanced adenomas from healthy subjects.

In a second aspect, the present invention provides a kit for predicting or diagnosing a disease associated with a microbiota in a subject, or determining whether a subject is at risk of developing the disease, comprising reagents for determining the level or amount of each biomarker of a biomarker panel according to the present invention in a sample.

In a preferred embodiment, the reagents used to determine the level or amount of each biomarker of the biomarker panel are selected from the group consisting of:

(a) A primer set comprising:

(a1) one or more primers capable of specifically amplifying MLG 317 or one or more specific fragments thereof, wherein MLG 317 consists of SEQ ID NOs: 933-1052;

(a2) one or more primers capable of specifically amplifying MLG 3770 or one or more specific fragments thereof, said MLG 3770 consisting of SEQ ID NOs: 1053-1281; and

(a3) one or more primers capable of specifically amplifying MLG 3840 or one or more specific fragments thereof, wherein MLG 3840 consists of SEQ ID NOs: 238-639;

Optionally, the primer set further comprises one or more of the following primers:

(a4) one or more primers capable of specifically amplifying MLG 665 or one or more specific fragments thereof, wherein MLG 665 consists of SEQ ID NOs: 640-932;

(a5) one or more primers capable of specifically amplifying MLG 721 or one or more specific fragments thereof, wherein MLG 721 consists of SEQ ID NOs: 120-237;

(a6) one or more primers capable of specifically amplifying MLG 1738 or one or more specific fragments thereof, wherein MLG 1738 consists of SEQ ID NOs: 1471-2436;

(a7) one or more primers capable of specifically amplifying MLG 1340 or one or more specific fragments thereof, wherein MLG 1340 consists of SEQ ID NOs: 2893-3067;

(a8) one or more primers capable of specifically amplifying MLG 5954 or one or more specific fragments thereof, wherein MLG 5954 consists of SEQ ID NOs: 1-119;

(a9) one or more primers capable of specifically amplifying MLG 711 or one or more specific fragments thereof, wherein MLG 711 consists of SEQ ID NOs: 1282-1470; and

(a10) one or more primers capable of specifically amplifying MLG 4668 or one or more specific fragments thereof, wherein MLG 4668 consists of SEQ ID NOs: 2437-2892;

(b) a probe set comprising:

(b1) one or more probes capable of specifically hybridizing to MLG 317 or one or more specific fragments thereof, wherein MLG 317 consists of SEQ ID NOs: 933-1052;

(b2) one or more probes capable of specifically hybridizing to MLG 3770 or one or more specific fragments thereof, said MLG 3770 consisting of SEQ ID NOs: 1053-1281; and

(b3) one or more probes capable of specifically hybridizing to MLG 3840 or one or more specific fragments thereof, wherein MLG 3840 consists of SEQ ID NOs: 238-639;

Optionally, the probe set further comprises one or more of the following probes:

(b4) one or more probes capable of specifically hybridizing to MLG 665 or one or more specific fragments thereof, wherein MLG 665 consists of SEQ ID NOs: 640-932;

(b5) one or more probes capable of specifically hybridizing to MLG 721 or one or more specific fragments thereof, wherein MLG 721 consists of SEQ ID NOs: 120-237;

(b6) one or more probes capable of specifically hybridizing to MLG 1738 or one or more specific fragments thereof, wherein MLG 1738 consists of SEQ ID NOs: 1471-2436;

(b7) one or more probes capable of specifically hybridizing to MLG 1340 or one or more specific fragments thereof, wherein MLG 1340 consists of SEQ ID NOs: 2893-3067;

(b8) one or more probes capable of specifically hybridizing to MLG 5954 or one or more specific fragments thereof, wherein MLG 5954 consists of SEQ ID NOs: 1-119;

(b9) one or more probes capable of specifically hybridizing to MLG 711 or one or more specific fragments thereof, wherein MLG 711 consists of SEQ ID NOs: 1282-1470; and

(b10) one or more probes capable of specifically hybridizing to MLG 4668 or one or more specific fragments thereof, wherein MLG 4668 consists of SEQ ID NOs: 2437-2892;

(c) a microarray comprising the primer set of (a) and/or the probe set of (b);

(d) reagents for performing a second generation sequencing method or a third generation sequencing method; and

(e) Any combination of (a)-(d).

In a preferred embodiment, the primer set comprises primers as defined in (a1) to (a8).

In a preferred embodiment, the probe set comprises probes as defined in (b1) to (b8).

In a preferred embodiment, the kit predicts or diagnoses a disease associated with a microbiota in a subject, or determines whether the subject has a risk of developing the disease, by a method comprising the following steps:

(1) using the kit to determine the level or amount of each biomarker of the biomarker panel according to the present invention in a sample from the subject;

(2) calculating the probability of the disease by comparing the level or amount of each biomarker in the sample with a training data set using a multivariate statistical model (such as a random forest model);

When the probability of the disease is greater than a critical value, it indicates that the subject suffers from the disease or has a risk of developing the disease.

In a preferred embodiment, the training dataset comprises data on the level of each biomarker or the amount thereof in a plurality of subjects suffering from the disease and a plurality of healthy subjects.

In a preferred embodiment, the training data set comprises the data in Table 8, and when the probability is greater than a critical value of 0.4572, it indicates that the subject has the disease or has a risk of developing the disease.

In preferred embodiments, the subject is a mammal, such as a primate, preferably a human.

In a preferred embodiment, the sample is a stool sample.

In a preferred embodiment, the level of each biomarker or the amount thereof is the relative abundance of each biomarker in the sample.

In a preferred embodiment, the disease is advanced adenoma in the colorectum.

In a preferred embodiment, the kit further comprises additional reagents, such as reagents for processing the sample (e.g., sterile water), reagents for performing PCR amplification (e.g., polymerase, dNTPs, and amplification buffer), and reagents for performing hybridization (such as labeling buffer, hybridization buffer, and wash buffer).

In a third aspect, the present invention provides the use of a reagent for determining the level or amount of each biomarker of a biomarker panel according to the present invention in the preparation of a kit for predicting or diagnosing a disease associated with a microbiome in a subject or determining whether the subject has a risk of developing the disease.

In a preferred embodiment, the reagents for determining the level or amount of each biomarker of the biomarker panel are as defined above.

In a preferred embodiment, the kit predicts or diagnoses a disease associated with a microbiota in a subject, or determines whether the subject has a risk of developing the disease, by a method comprising the following steps:

(1) using the kit to determine the level or amount of each biomarker of the biomarker panel according to the present invention in a sample;

(2) calculating the probability of the disease by comparing the level or amount of each biomarker in the sample with a training data set using a multivariate statistical model (such as a random forest model);

When the probability of the disease is greater than a critical value, it indicates that the subject suffers from the disease or has a risk of developing the disease.

In a preferred embodiment, the training dataset comprises data on the level of each biomarker or the amount thereof in a plurality of subjects suffering from the disease and a plurality of healthy subjects.

In a preferred embodiment, the training data set comprises the data in Table 8, and when the probability of the disease is greater than a critical value of 0.4572, it indicates that the subject has the disease or has a risk of developing the disease.

In preferred embodiments, the subject is a mammal, such as a primate, preferably a human.

In a preferred embodiment, the sample is a stool sample.

In a preferred embodiment, the level of each biomarker or the amount thereof is the relative abundance of each biomarker in the sample.

In a preferred embodiment, the disease is advanced adenoma in the colorectum.

In a preferred embodiment, the kit further comprises additional reagents, such as reagents for handling the sample (e.g., sterile water), reagents for performing PCR amplification (e.g., polymerase, dNTPs, and amplification buffer), and reagents for performing hybridization (such as labeling buffer, hybridization buffer, and wash buffer).

In a fourth aspect, the present invention provides a method for predicting or diagnosing a disease associated with a microbiota in a subject or determining whether the subject has a risk of developing the disease, comprising the steps of:

(1) determining the level or amount of each biomarker of the biomarker panel according to any one of claims 1 to 6 in a sample from the subject;

(2) calculating the probability of the disease by comparing the level or amount of each biomarker in the sample with a training dataset using a multivariate statistical model (such as a random forest model);

When the probability of the disease is greater than a critical value, it indicates that the subject suffers from the disease or has a risk of developing the disease.

In a preferred embodiment, the training dataset comprises data on the level of each biomarker or the amount thereof in a plurality of subjects suffering from the disease and a plurality of healthy subjects.

In a preferred embodiment, the training data set comprises the data in Table 8, and when the probability of the disease is greater than a critical value of 0.4572, it indicates that the subject has the disease or has a risk of developing the disease.

In a preferred embodiment, a kit as defined above or a reagent as defined above is used in step (1).

In preferred embodiments, the subject is a mammal, such as a primate, preferably a human.

In a preferred embodiment, the sample is a stool sample.

In a preferred embodiment, the level of each biomarker or the amount thereof is the relative abundance of each biomarker in the sample.

In a preferred embodiment, the disease is advanced adenoma in the colorectum.

In a preferred embodiment, the method is performed in vitro.

The present invention is further illustrated in the following non-limiting examples. Unless otherwise indicated, parts and percentages are by weight and degrees are in degrees Celsius. All reagents used are commercially available. It will be apparent to one of ordinary skill in the art that these examples, while representing preferred embodiments of the present invention, are provided by way of illustration only.

Example

Example 1. Identification and validation of biomarkers for assessing CRC-related disease risk

1. Sample Collection and Sequencing

1.1 Subjects and Patients

The study was conducted in participants of a health screening program according to national screening recommendations for CRC (Stadlmayr, A. et al. Nonalcoholic fatty liver disease: an independent risk factor for colorectal neoplasia. J Intern Med 270, 41-49 (2011), incorporated herein by reference) and in patients suspected of having CRC who underwent a colonoscopy as part of a clinical workup at the Department of Internal Medicine of Oberndorf Hospital (teaching hospital of the Paracelsus Medical University of Salzburg, Austria) between 2010 and 2012. The study was approved by the local ethics committee (Ethikkommission des Landes Salzburg, approval number 415-E/1262/2-2010), and informed consent was obtained from all participants.

A laxative (containing polyethylene glycol 59.0 g, sodium sulfate 5.68 g, sodium bicarbonate 1.68 g, NaCl 1.46 g, and potassium chloride 0.74 g; Norgine, Marburg, Germany) was used for bowel preparation before colonoscopy. Based on a combined analysis of macroscopic and histological findings, colonoscopy findings were classified as tubular adenoma, advanced adenoma (i.e., villous or tubulovillous features, size ≥1 cm or high-grade dysplasia), or carcinoma (Bond, J.H. Polyp guideline: diagnosis, treatment, and surveillance for patients with colorectal polyps ACG Colorectal Polyp Guideline. Am. J. Gastroenterol. 95, 3053-3063 (2000), Winawer SJ & AG., Z. The advanced adenoma as the primary target of screening. Gastrointest Endosc Clin N Am 12, 1-9 (2002), incorporated herein by reference). Lesions were classified according to location (ie, right colon (including cecum, ascending colon, and transverse colon), left colon (from splenic flexure to sigmoid colon), and rectum alone).

The initial analysis included data from 147 Caucasian individuals aged 45 to 86 years, including 57 healthy controls (24 men, 33 women), 44 patients with advanced adenoma (22 women, 22 men), and 46 patients with cancer (18 men, 28 women) (Table 1-1). An additional 9 samples (6 healthy controls, 3 advanced adenoma samples, Table 1-1) were also used in the test set of the MLG-based cancer classifier (Figure 1d). To date, no studies have explored the given topic in a comparable manner; therefore, a formal power analysis for sample size calculation could not be performed. However, based on previous studies of patient fecal microbiota using 16S and metagenomic shotgun sequencing, this was a reasonable sample size. Subjects were stratified by sex, age, and body mass index (BMI) to make the three groups (control group, advanced adenoma group, and cancer group) comparable on these variables. In the advanced adenoma group, 14 cases had lesions located in the right colon (including the cecum, ascending colon, and transverse colon), 15 cases had lesions located in the left colon (from the splenic flexure to the sigmoid colon), and 15 cases had lesions located in the rectum. In the carcinoma group, 8 cases had lesions located in the right colon, 11 cases had lesions located in the left colon, and 27 cases had lesions located in the rectum. Colorectal cancer is classified by the American Joint Committee on Cancer (AJCC) TNM staging system (Greene, F.L. Current TNM staging of colorectal cancer. Lancet. Oncol. 8, 572-3 (2007), incorporated herein by reference).

Table 1-1: Clinical data of all 156 samples

1.2 Stool samples

Fresh stool samples were collected from all patients and subjects. The samples were mechanically homogenized with a sterile spatula, and four aliquots were collected using a Sarstedt stool sampling system (Sarstedt, Nümbrecht, Germany). Each aliquot contained 1 g of stool and was placed in a sterile 12-ml cryovial. The stool aliquots were then stored in a −20°C home freezer and transported to the laboratory in cold packs within 48 hours of collection, where they were immediately stored at −80°C. None of the patients and subjects had received probiotics or antibiotics within the previous 3 months.

1.3 DNA extraction

Fecal samples were thawed on ice and DNA was extracted using the Qiagen QIAamp DNA Stool Mini Kit (Qiagen) according to the manufacturer's instructions. The extracts were treated with DNase-free RNase to eliminate RNA contamination. DNA quantity was determined using a NanoDrop spectrophotometer, a Qubit fluorometer (using the Quant-iT™ dsDNA BR Assay Kit), and gel electrophoresis.

1.4 Metagenomic sequencing and gene catalog construction

Paired-end metagenomic sequencing was performed on the Illumina platform (insert size 350 bp, read length 100 bp), and the sequencing reads were quality controlled and assembled de novo into contigs as previously described (Qin et al., 2012, supra) using SOAPdenovo v2.04 (default parameters were used except for -K 51-M3-F-u) (Luo, R. et al. SOAPdenovo2: an empirically improved memory-efficient short-read de novo assembler. Gigascience 1, 18 (2012), incorporated herein by reference). High-quality sequencing reads were assembled de novo (average of 5 GB per sample) and the identified genes were compiled into a 3.5M non-redundant gene set, which resulted in an average of 76.3% of reads per sample being matched.

Gene prediction was performed on the assembled contigs using GeneMark v2.7d. Redundant genes were removed using BLAT (Kent, W.J.BLAT--the BLAST-like alignment tool. Genome Res. 12, 656-64 (2002), incorporated herein by reference), with 90% overlap and 95% identity (no gaps allowed) as critical values. The relative abundance of genes was determined by comparing high-quality sequencing reads with the gene catalog using the same program as Qin et al. 2012 (supra).

The predicted genes were assigned taxonomically based on the IMG database (v400) using an in-house pipeline previously described (Qin et al. 2012, supra) using 80% overlap and 65% identity, with a top 10% score (BLASTN v2.2.24, -e 0.01-b 100-K 1-F T-m 8). The cutoffs for assignment to phylum were 65% identity, 85% identity for genus, and 95% identity for species. If multiple hits were present, a cutoff of ≥50% identity was used for the taxon in question.

2. Metagenomic Wage-Based Association Study (MGWAS)

In order to compare the fecal microbiota of healthy controls, advanced adenomas and cancer patients, genes whose relative abundance showed significant differences between any two of the above groups were identified (Benjamin-Hochberg q value <0.1, Kruskal-Wallis test). These marker genes were then clustered into MLGs based on their abundance changes in all three groups of samples, which enabled the identification of the microbial species characteristics of each group (Qin et al., 2012, supra). Nine of the 147 samples contained more than 20% of the genus Escherichia (Escherichia) (2 controls, 2 adenomas, 5 cancer samples), and then the samples were used only in the test set for the MLG-based cancer classifier (Figure 1d). Another 9 samples (6 healthy controls, 3 advanced adenoma samples, Table 1-1) were also used in the test set for the above classifier.

As previously described, MLGs were assigned taxonomically and characterized based on the relative abundance of their constituent genes (Qin et al. 2012, supra). Briefly, assignment of an MLG to a species requires that >90% of the genes in the MLG be mapped to the genome of the species with >95% identity and >70% query overlap. Assignment of an MLG to a genus requires that >80% of the genes in the MLG be mapped to the genome of the genus with at least 85% identity in both DNA and protein sequences.

To explore the characteristics of the gut microbiome in healthy or tumor samples, the present inventors identified 130,715 genes that showed significant differences in abundance between any two of the three groups (Kruskal-Wallis test, Benjamin-Hochberg q value <0.1). With the exception of serum ferritin and red meat intake, no phenotype other than tumor status showed significant differences between controls, adenomas, and cancer patients (p < 0.05, Kruskal-Wallis test, Tables 1-2). Compared with healthy and advanced adenoma samples, 58.9% of the gene markers were significantly elevated in cancer samples, indicating that they are specific for colorectal cancer; another 24.3% of the genes were significantly more abundant in cancer samples than in control samples, but had intermediate levels in advanced adenoma samples. Among the genes with a downward trend, 5388 genes (4.1% of the total) were significantly decreased in cancer samples compared with healthy and advanced adenoma samples; the abundance of 2601 genes (2.0% of the total) was significantly lower in cancer samples than in control samples, and had moderate levels in advanced adenoma samples. These genes enriched in control samples, rather than those enriched in adenoma or cancer samples, were more frequently matched to the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. The difference in the number of increasing and decreasing genes suggests that the increase in pathogenic bacteria is more pronounced than the decrease in beneficial bacteria during cancer development. Based on the co-variation of the abundance of each gene in all samples, the significantly different genes were clustered into 126 MLGs, which enabled the identification of the microbial species characteristics of each group (Qin et al., 2012, ibid.).

3. MLG-based classification of colorectal cancer or adenoma

In order to evaluate the diagnostic value of the fecal microbiome of colorectal cancer, the inventors constructed a random forest classifier that can detect cancer samples. The MLG abundance profile of the training cohort (set) was used to train a random forest model (R.2.14, randomForest4.6-7 software package) (Liaw, Andy & Wiener, Matthew. Classification and Regression by randomForest, R News (2002), Vol. 2/3, p. 18, incorporated herein by reference) to select the optimal set of MLG markers. The model was tested on the test set and the prediction error was determined. Regarding the random forest model, by using the "randomForest4.6-7 package" in R vision 2.14, the training data set (i.e., the relative abundance profile of the MLG selected in the training sample), the sample disease status (the sample disease status of the training sample is a vector, 1 represents a case, 0 represents a control) and the test data set (i.e., the relative abundance profile of the MLG selected in the test set) was input. The inventors then used the random Forest function of the randomForest package in R software to construct a classification, and used the prediction function to predict the test set. The output is the prediction result (probability of being sick; the critical value refers to the optimal critical value, if the probability of the disease ≥ the optimal critical value, the subject is at risk of the disease).

The random forest model (R 3.0.2, randomForest4.6-7 package) was subjected to 10-fold cross validation using the MLG abundance profiles of control, advanced adenoma or cancer samples. The cross validation error curves obtained from 5 10-fold cross validations (each curve is the average of 10 test sets) were averaged, and the minimum error in the average curve plus the standard deviation at that point was used as the critical value. All sets of MLG markers with an error less than the critical value (≤50) were listed, and the set with the least number of MLGs was selected as the optimal set. The probability of adenoma or cancer was calculated using the MLG set, and ROC (R 3.0.2, pROC3 package) was plotted. The model was further tested on the test set, and the prediction error was determined.

The optimal selection of 15 MLG markers (Table 2-1, Table 2-2) was obtained by performing 5 repetitions of 10-fold cross validation (i.e., 50 tests) on a training set consisting of 55 controls and 41 cancer samples (Table 1-1). In short, 5 repetitions of 10-fold cross validation (i.e., 50 tests) were performed on a training set consisting of 55 controls and 41 cancer samples. In each test, the random forest test ranked the importance of each MLG. The inventors selected the top 15 important MLGs and ranked the mlgs according to the number of occurrences. The above top 15 MLGs were used to construct a classifier. Table 5 lists the ranking of the importance of MLGs.

Our results showed that the 15 MLGs performed well in the training set, with an area under the receiver operating curve (AUC) of 98.34% (cutoff = 0.5, Figures 1a, 1b, 1c, Tables 3-1, 3-2, 4, and 5). The classification error in the test set (8 control samples, 47 advanced adenoma samples, and 5 cancer samples) was low, with an AUC of 96% (advanced adenomas were considered non-cancerous, cutoff = 0.5, Figures 1d, 1e, Table 4), consistent with their predominantly benign nature. Among the MLG markers were mlg-75 and mlg-84, which are likely oral anaerobes. The former showed a high odds ratio for adenoma (Table 2-2), suggesting an early role in pathogenesis. Other MLG markers included Bacteroides massiliensis, mlg-2985, mlg-121, and 10 additional taxonomically undefined MLGs (Table 2-2). Thus, the MLGs selected by the cancer classifier displayed important features of the gut microbiome that contribute to disease progression in adenomas and carcinomas and have great potential for early and noninvasive diagnosis of these tumors.

In addition, the results in Table 5 show that for the combination of the top 2 important MLGs (MLG 5045 and MLG 121), the AUC is 0.91751663; for the combination of the top 3 important MLGs (MLG5045, MLG 121, and MLG 75), the AUC is 0.970731707; for the combination of the top 4 MLGs, the AUC is 0.959645233; for the combination of the top 5 MLGs, the AUC is 0.975609756; for the combination of the top 6 MLGs, the AUC is 0.978713969; for the combination of the top 7 MLGs, the AUC is 0.980044346; for the combination of the top 8 MLGs, the AUC is 0.985365854; for the combination of the top 9 MLGs, the AUC is 0.961000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 The UC was 0.984035477; for the first 10 MLGs, the AUC was 0.981818182; for the first 11 MLGs, the AUC was 0.980931264; for the first 12 MLGs, the AUC was 0.979157428; for the first 13 MLGs, the AUC was 0.987583149; for the first 14 MLGs, the AUC was 0.986696231; and for the first 15 MLGs, the AUC was 0.983370288. These results indicate that MLG 5045 and MLG121 are the most important biomarkers among the 15 MLGs, and that the combination of MLG 5045 and MLG121 is sufficient for diagnosing colorectal cancer and/or determining the risk of colorectal cancer with high sensitivity and specificity (AUC = 0.91751663); the incorporation of other MLG biomarkers from the 15 MLGs can improve the sensitivity and specificity of diagnosis or prediction to a certain extent. In particular, these results also indicate that the combination of the first 13 MLGs can be considered the optimal biomarker panel, which has the best sensitivity and specificity for the diagnosis or prediction of colorectal cancer (AUC = 0.987583149).

These results fully support that the MLGs identified herein (particularly MLG 5045 and MLG121, optionally in combination with one or more additional MLGs among these 15 MLGs) can be used as biomarkers for diagnosing colorectal cancer and/or determining the risk of developing colorectal cancer with high sensitivity and specificity.

The present inventors further directed investigation into the utility of intestinal MLG for identifying adenomas, which are more difficult to screen for than colorectal cancers but are important for early intervention.

Similarly, 5 repetitions of 10-fold cross validation (i.e., 50 tests) were performed on a training set consisting of 55 controls and 42 adenoma samples. In each test, the random forest test ranked the importance of each MLG. The inventors selected the top 10 important MLGs and ranked them according to the number of occurrences. These top 10 MLGs were used to construct the classifier. The ranking of the importance of the MLGs is listed in Table 10.

After five repetitions of 10-fold cross-validation, the random forest model selected 10 MLGs (Table 6-1, Table 6-2, Figure 2a) that allowed for optimal classification of the training set (55 controls and 42 advanced adenomas, Tables 8 and 9, Figure 2b), with an AUC of 0.8738 (critical value = 0.5, Figure 2c). In the test set (consisting of 15 controls and 15 advanced adenomas, new samples that were not used), all advanced adenoma samples were correctly classified (critical value = 0.4572, Figures 2d, 2e, Table 9). Therefore, fecal MLG provides a new opportunity for the noninvasive detection of colorectal advanced adenomas.

In addition, the results in Table 10 show that for the combination of the top 2 important MLGs (MLG 317 and MLG3770), the AUC is 0.782251082; for the top 3 important MLGs (MLG 317, MLG 3770 and MLG 3840), the AUC was 0.805194805; for the combination of the first 4 MLGs, the AUC was 0.773160173; for the combination of the first 5 MLGs, the AUC was 0.795238095; for the combination of the first 6 MLGs, the AUC was 0.780952381; for the combination of the first 7 MLGs, the AUC was 0.895670996; for the combination of the first 8 MLGs, the AUC was 0.896536797; for the combination of the first 9 MLGs, the AUC was 0.884848485; for the combination of the first 10 MLGs, the AUC was 0.873809524. These results indicate that MLG 317, MLG3770, and MLG 3840 are the most important biomarkers among the 10 MLGs, and that the combination of MLG 317, MLG 3770, and MLG 3840 is sufficient for diagnosing colorectal advanced adenoma and/or determining the risk of developing colorectal advanced adenoma with high sensitivity and high specificity (AUC = 0.805194805); and that the incorporation of other MLG biomarkers among the 10 MLGs can improve the sensitivity and specificity of diagnosis or prediction to a certain extent. In particular, these results also indicate that the combination of the first 8 MLGs can be considered the optimal biomarker group, which has the best sensitivity and specificity for the diagnosis or prediction of colorectal advanced adenoma (AUC = 0.896536797).

These results fully support that the MLGs identified herein (particularly MLG 317, MLG 3770 and MLG 3840, optionally in combination with one or more additional MLGs among the above 10 MLGs) can be used as biomarkers for diagnosing colorectal advanced adenoma and/or determining the risk of developing colorectal advanced adenoma with high sensitivity and specificity.

Therefore, the present inventors identified and validated 15 MLGs for the early and non-invasive diagnosis of colorectal cancer and 10 MLGs for the early and non-invasive diagnosis of colorectal adenoma using a random forest model based on relevant gene markers. The present inventors also constructed a method for assessing the risk of colorectal cancer and adenoma based on these relevant intestinal microbiota.

Claims (16)

1. A set of biomarkers for predicting or diagnosing progressive adenomas in the colorectum or determining whether a subject is at risk of developing progressive adenomas in the colorectum, comprising the following biomarkers: (1) MLG 317, wherein MLG 317 is composed of the nucleic acids shown in SEQ ID NO: 933-1052; (2) MLG 3770, wherein MLG 3770 is composed of the nucleic acids shown in SEQ ID NO: 1053-1281; and (3) MLG 3840, wherein MLG 3840 is composed of the nucleic acid shown in SEQ ID NO: 238-639; The biomarker set is used to distinguish patients with progressive adenomas from healthy subjects; the subjects are human beings. 2. The biomarker set of claim 1, wherein the biomarker set further comprises one or more of the following biomarkers: (4) MLG 665, wherein MLG 665 is composed of the nucleic acid shown in SEQ ID NO: 640-932; (5) MLG 721, wherein MLG 721 is composed of the nucleic acids shown in SEQ ID NO: 120-237; (6) MLG 1738, wherein MLG 1738 is composed of the nucleic acids shown in SEQ ID NO: 1471-2436; (7) MLG 1340, wherein MLG 1340 is composed of the nucleic acid shown in SEQ ID NO: 2893-3067; (8) MLG 5954, wherein MLG 5954 is composed of the nucleic acids shown in SEQ ID NO: 1-119; (9) MLG 711, wherein MLG 711 is composed of the nucleic acids shown in SEQ ID NO: 1282-1470; and (10) MLG 4668, wherein MLG 4668 is composed of the nucleic acid shown in SEQ ID NO: 2437-2892. 3. The biomarker set of claim 2, wherein the biomarker set comprises biomarkers as defined in (1)-(8). 4. A kit for predicting or diagnosing progressive adenoma in the colorectum in a subject, or for determining whether a subject is at risk of developing progressive adenoma in the colorectum, comprising reagents for determining the level in a sample of each of the biomarkers in the group of biomarkers according to any one of claims 1-3; The subject is a human being; the sample is a fecal sample. 5. The kit of claim 4, wherein the reagent is selected from: (a) Primer set, which includes: (a1) Primers capable of specifically amplifying MLG 317, wherein MLG 317 is composed of the nucleic acid shown in SEQ ID NO: 933-1052; (a2) Primers capable of specifically amplifying MLG 3770, wherein MLG 3770 is composed of the nucleic acid shown in SEQ ID NO: 1053-1281; and (a3) Primers capable of specifically amplifying MLG 3840, wherein MLG 3840 is composed of the nucleic acid shown in SEQ ID NO: 238-639; (b) A probe array comprising: (b1) A probe capable of specifically hybridizing with MLG 317, wherein MLG 317 is composed of the nucleic acid shown in SEQ ID NO: 933-1052; (b2) A probe capable of specifically hybridizing with MLG 3770, said MLG 3770 being composed of the nucleic acids shown in SEQ ID NO: 1053-1281; and (b3) A probe capable of specifically hybridizing with MLG 3840, wherein MLG 3840 is composed of the nucleic acid shown in SEQ ID NO: 238-639; (c) A microarray containing the primer set of (a) and/or the probe set of (b); (d) Reagents for performing second-generation or third-generation sequencing methods; and Any combination of (e)(a)-(d). 6. The kit of claim 5, wherein the primer set further comprises one or more of the following primers: (a4) Primers capable of specifically amplifying MLG 665, wherein MLG 665 is composed of the nucleic acid shown in SEQ ID NO: 640-932; (a5) Primers capable of specifically amplifying MLG 721, wherein MLG 721 is composed of the nucleic acid shown in SEQ ID NO: 120-237; (a6) Primers capable of specifically amplifying MLG 1738, wherein MLG 1738 is composed of the nucleic acid shown in SEQ ID NO: 1471-2436; (a7) Primers capable of specifically amplifying MLG 1340, wherein MLG 1340 is composed of the nucleic acid shown in SEQ ID NO: 2893-3067; (a8) Primers capable of specifically amplifying MLG 5954, wherein MLG 5954 is composed of the nucleic acid shown in SEQ ID NO: 1-119; (a9) Primers capable of specifically amplifying MLG 711, wherein MLG 711 is composed of the nucleic acid shown in SEQ ID NO: 1282-1470; and (a10) Primers capable of specifically amplifying MLG 4668, which is composed of the nucleic acid shown in SEQ ID NO: 2437-2892. 7. The kit of claim 6, wherein the primer set comprises primers as defined in (a1)-(a8). 8. The kit of claim 5, wherein the probe set further comprises one or more of the following probes: (b4) A probe capable of specifically hybridizing with MLG 665, wherein MLG 665 is composed of the nucleic acid shown in SEQ ID NO: 640-932; (b5) A probe capable of specifically hybridizing with MLG 721, wherein MLG 721 is composed of the nucleic acid shown in SEQ ID NO: 120-237; (b6) A probe capable of specifically hybridizing with MLG 1738, wherein MLG 1738 is composed of the nucleic acid shown in SEQ ID NO: 1471-2436; (b7) A probe capable of specifically hybridizing with MLG 1340, wherein MLG 1340 is composed of the nucleic acid shown in SEQ ID NO: 2893-3067; (b8) A probe capable of specifically hybridizing with MLG 5954, wherein MLG 5954 is composed of the nucleic acid shown in SEQ ID NO: 1-119; (b9) A probe capable of specifically hybridizing with MLG 711, said MLG 711 being composed of the nucleic acids shown in SEQ ID NO: 1282-1470; and (b10) A probe capable of specifically hybridizing with MLG 4668, which is composed of the nucleic acid shown in SEQ ID NO: 2437-2892. 9. The kit of claim 8, wherein the probe set comprises probes as defined in (b1)-(b8). 10. The kit of claim 4, wherein the kit further comprises additional reagents selected from reagents for processing the sample, reagents for performing PCR amplification, and reagents for performing hybridization. 11. The kit of claim 10, wherein the reagent for treating the sample is sterile water. 12. The kit of claim 10, wherein the reagents used for PCR amplification are selected from polymerase, dNTPs, and amplification buffer. 13. The kit of claim 10, wherein the reagent for hybridization is selected from labeling buffer, hybridization buffer, and washing buffer. 14. Use of reagents for determining the level of each biomarker in the biomarker group of any one of claims 1-3 in the preparation of a kit for predicting or diagnosing progressive adenoma in the colorectum or determining whether a subject is at risk of developing progressive adenoma in the colorectum; The subjects mentioned are human beings. 15. The use as claimed in claim 14, wherein the reagent is as defined in any one of claims 5-9. 16. The use according to claim 14 or 15, wherein the kit predicts or diagnoses progressive adenoma in the colorectal region in the subject, or determines whether the subject is at risk of developing progressive adenoma in the colorectal region, by means of a method comprising the steps of: (1) Using the kit, determine the level of each biomarker in the group of biomarkers according to any one of claims 1-3 in a sample from the subject, wherein the level of each biomarker is the relative abundance of each biomarker in the sample; (2) The probability of progressive adenoma in the colorectal region is calculated by comparing the level of each biomarker in the sample with a training dataset using a multivariate statistical model; the multivariate statistical model is a random forest model; the training dataset contains data on the level of each biomarker for multiple subjects with progressive adenoma in the colorectal region and multiple healthy subjects; When the probability of progressive adenoma in the colorectal region is greater than a critical value, it indicates that the subject has progressive adenoma in the colorectal region or is at risk of developing progressive adenoma in the colorectal region. The sample in question is a fecal sample.