HK1239740B - Biomarkers for rheumatoid arthritis and usage thereof - Google Patents

Description

Biomarkers for rheumatoid arthritis and their uses

CROSS-REFERENCE TO RELATED APPLICATIONS

none

Technical Field

The present invention relates to the field of biomedicine, and in particular to biomarkers and methods for predicting the risk of diseases associated with microbiota, in particular rheumatoid arthritis (RA).

Background Art

Rheumatoid arthritis (RA) is a debilitating autoimmune disease that affects tens of millions of people worldwide and increases the mortality rate of patients with its cardiovascular and other systemic complications, but the cause of RA remains unclear. Infectious pathogens have been implicated in RA. However, the characteristics and pathogenicity of RA-related pathogens are largely unclear, and the recent determination that the human body is a super-organism that hosts trillions of beneficial and harmful microorganisms has further complicated the problem. Although the use of disease-modifying antirheumatic drugs (DMARDs) has successfully alleviated the condition of many RA patients, insufficient understanding of the factors that trigger or promote the disease has hindered the development of specific and more effective treatments. Investigations into microorganisms have also revealed probiotics that prevent or alleviate RA.

It is believed that RA initiates and lurks in certain other parts of the body for several years before the onset of joint inflammation. The intestinal microbiome is a key environmental factor in human health and has a definite role in obesity, diabetes, colon cancer, etc. In addition to playing a role in nutrition and xenobiotic metabolism, the microorganisms in the terminal intestine also interact with the neuro-immune-endocrine system and blood flow to affect the entire human body. The intestinal microbiome is stably associated with a given individual, increasing its value in disease-related research. The heterogeneity of the intestinal microbiome in the population suggests that the treatment of the disease should be individualized according to the intestinal microbiome, and its role in drug activation or inactivation, immune regulation, etc. remains largely unclear. Compared with the oral microbiome, the oral microbiome is relatively under research, with the Human Microbiome Project (HMP) only collecting about 100 healthy individuals for WGS (Human Microbiome Project Consortium. A framework for human microbiomeresearch. Nature 486, 215–21 (2012), incorporated herein by reference). Despite the fact that dental and saliva samples are more readily available in outpatient settings than stool samples, metagenomic analyses of the role of the oral microbiome in disease have been lacking. It is also unknown to what extent oral and gut microbial disease markers are concordant in their identity or function.

Summary of the Invention

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

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

The evaluation and characterization of intestinal microorganisms have become a major research area for human diseases including rheumatoid arthritis (RA). In order to analyze the intestinal microbial contents of RA patients, the present inventors conducted a metagenomic association study (Metagenome-Wide Association Study, MGWAS) (Qin, J. et al. A metagenome-wide association study of gut microbiotain type 2 diabetes. Nature 490, 55–60 (2012), incorporated herein by reference) based on deep shotgun sequencing of microbial DNA from 212 individuals. The present inventors identified and confirmed the intestinal/tooth/saliva marker group (29 intestinal MLG\28 tooth MLG\19 saliva MLG) based on RA-related gene markers by random forest model. In order to intuitively assess the risk of RA disease based on these 29 intestinal MLG\28 tooth MLG\19 saliva MLG, the present inventors calculated the probability of the disease separately by random forest model based on the relative abundance spectrum of MLG markers in the training set. Our data provide insights into the characteristics of the gut/tooth/saliva metagenome associated with RA risk, serve as a paradigm for future studies investigating the pathophysiological roles of the gut/tooth/saliva metagenome in other related diseases, and offer potential applications for microbiome-based methods for assessing individuals at risk for this disease.

It is believed that the RA-associated gut microbiota (29 intestinal MLGs\28 dental MLGs\19 salivary MLGs) is valuable for increasing RA detection in the early stages for the following reasons. First, the markers of the present invention are specific and sensitive. Second, analysis of feces guarantees accuracy, safety, affordability and patient compliance. And fecal samples are transportable. Polymerase chain reaction (PCR)-based tests are comfortable and non-invasive, so people will be more likely to participate in a given screening program. Third, the markers of the present invention can also be used as a tool for treatment monitoring of RA patients to detect response to treatment.

In one aspect, a biomarker panel for predicting a disease associated with a microbiome in a subject is provided, and according to an embodiment of the present disclosure, the biomarker panel is composed of intestinal biomarkers, dental biomarkers, salivary biomarkers, or microorganisms having genomic DNA comprising at least a portion of the sequence of SEQ ID NOs: 1 to 9319, wherein

Gut biomarkers included Bifidobacterium dentium, RA-2633, Enterococcus sp., RA-781, Gordonibacter pamelaeae, RA-3396, RA-6638, RA-2441, RA-527, Clostridium sp., RA-2637, Citrobacter sp., Eubacterium sp., Citrobacter sp., RA-3215, Con-1722, Con-4360, Con-4212, Con-1261, Bifidobacterium bifidum, Klebsiella pneumoniae, Con-1423, Veillonella sp.), Con-4095, Con-4103, Con-1735, Con-1710, Con-1832, Con-1170,

Dental biomarkers include RA-10848, RA-9842, RA-9941, RA-9938, RA-10684, RA-9998, Con-7913, Con-20702, Con-11, Con-8169, Con-1708, Con-7847, Con-5233, Con-791 , Con-5566, Con-4455, Con-13169, Con-6088, Con-5554, Con-14781, Con-2466, Con-483, Con-2562, Con-4701, Con-4824, Con-5030, Con-757, Con-530, and

Salivary biomarkers include RA-27683, RA-9651, RA-13621, RA-27616, Con-6908, Con-305, Con-1559, Con-1374, Con-6746, Campylobacter rectus, Con-1141, Con-20, Streptococcus sp., Con-1238, Con-1073, Con-636, Con-1, Porphyromonas gingivalis, Lactococcus sp.,

Or a microorganism whose genomic DNA comprises at least a portion of the sequence of SEQ ID NO: 1 to 9319.

Optionally, the biomarker panel consists of at least one species listed in Table 2-2, preferably consists of at least 10%, at least 20%, at least 30%, at least 40%, at least 50%, at least 60%, at least 70%, at least 80%, at least 90%, at least 100% of the species listed in Table 2-2.

According to an embodiment of the present disclosure, the intestinal biomarker comprises at least a portion of the sequences of SEQ ID NOs: 1 to 9319 as described in Table 5.

According to an embodiment of the present disclosure, intestinal biomarkers include Bifidobacterium dentate JCVIHMP022, Prevotella copri CB7, DSM 18205, Enterococcus faecium E980, Ruminococcus obeum A2-162, Gordonibacter pamelaeae 7-10-1-bT, DSM 19378, Ruminococcus bromii L2-63, Eubacterium ventriosum ATCC 27560, Klebsiella oxytoca KCTC 1686, Clostridium asparagiforme DSM 15981, Prevotella copriCB7, DSM 18205, Citrobacter freundii 4_7_47CFAA, Eubacterium sp. 3_1_31, Citrobacter sp. 30_2, Clostridium sp. 7_2_43FAA, Roseburia intestinalis M50/1, Dialister invisus DSM 15470, Bacteroides plebeius M12, DSM 17135, Bifidobacterium bifidum S17, Klebsiella pneumoniae NTUH-K2044 NTUH-K2044), Veillonellasp. oral taxon 158F0412, Comamonas testosteroni KF-1, Klebsiella pneumoniae NTUH-K2044, Veillonella atypica ACS-134-V-Col7a, Streptococcus australis ATCC 700641, Parabacteroides merdae ATCC 43184,

Dental biomarkers included Actinomyces sp. oraltaxon 180F0310, Rothia mucilaginosa DY-18, Actinomyces graevenitzii C83, Actinomyces odontolyticus ATCC 17982, Veillonella atypica ACS-134-V-Col7a, Actinomyces sp. F0384, Actinomyces sp. oraltaxon 848F0332, Neisseria mucosa M26, ATCC 25996, Actinomyces sp. oral taxon 448F0400, Tannerella forsythensis ATCC 43037, Actinomyces sp. oral taxon 448F0400, Neisseria bacilliformis ATCC BAA-1200, Synergistetes bacterium SGP1, Lautropia mirabilis ATCC 51599, Capnocytophaga gingivalis ATCC 33624, Cardiomycobacterium ATCC Cardiobacterium hominis ATCC 15826, Capnocytophaga gingivalis ATCC 33624, Lautropia mirabilis ATCC 51599, Johnsonella ignava ATCC 51276, Propionibacterium freudenreichii shermanii CIRM-BIA1, Treponema denticola ATCC 35405, Fusobacterium sp. oral taxon 370F0437, and Cardiobacterium hominis ATCC 15826. 51599 (Lautropia mirabilis ATCC 51599), Eikenella corrodens ATCC 23834 (Eikenella corrodens ATCC 23834), Selenomonas noxia ATCC 43541 (Selenomonas noxia ATCC 43541), Porphyromonas levii DSM 23370 (Porphyromonas levii DSM 23370), Bulleidia extructa W1219,

Salivary biomarkers included Gemella haemolysans ATCC 10379, Veillonella atypica ACS-049-V-Sch6, Actinomyces odontolyticus ATCC 17982, Treponema denticola ATCC 35405, Actinomyces sp. oral taxon 448F0400, Treponema vincentii ATCC 35580, and Actinomyces sp. oral taxon 448F0400. 35580), Streptococcus australis ATCC 700641, Campylobacter rectus RM3267, CCUG 20446, Actinomyces sp. oral taxon 171F0337, Treponema denticola ATCC 35405, Streptococcus sanguinis VMC66, Actinomyces sp. oral taxon 448F0400, Actinomyces sp. oral taxon 448F0400, Neisseria truncatula ATCC BAA-1200 (Neisseria bacilliformis ATCC BAA-1200), Burkholderia mallei PRL-20 (Burkholderiamallei PRL-20), Porphyromonas gingivalis TDC60 (Porphyromonas gingivalis TDC60), Lactococcus lactis lactis KF147 (Lactococcus lactis lactis KF147).

In another aspect of the present disclosure, a biomarker panel for predicting a disease associated with a microbiome in a subject is provided. According to an embodiment of the present disclosure, the biomarker panel consists of intestinal biomarkers, dental biomarkers, and salivary biomarkers, wherein

The intestinal biomarkers include at least part of the sequences of SEQ ID NOs: 1 to 9319.

According to an embodiment of the present disclosure, the disease is rheumatoid arthritis or a related disease.

In another aspect of the present disclosure, a kit for determining the above-mentioned gene marker group is provided, comprising primers for PCR amplification and designed according to the DNA sequences listed below:

The intestinal biomarkers include at least part of the sequences of SEQ ID NOs: 1 to 9319.

In another aspect of the present disclosure, a kit for determining the above gene marker panel is provided, comprising one or more probes designed according to the genes listed below: The intestinal biomarkers include at least part of the sequences of SEQ ID NOs: 1 to 9319.

In another aspect of the present disclosure, there is provided a use of the above gene marker panel for predicting the risk of rheumatoid arthritis or related diseases in a subject to be tested, comprising:

(1) Collecting samples from the subjects to be tested;

(2) determining the relative abundance information of each biomarker of the biomarker panel according to any one of claims 1 to 5 in the sample obtained in step (1);

(3) The probability of rheumatoid arthritis is obtained by comparing the relative abundance information of each biomarker of the test subject with the training data set using a multivariate statistical model,

The probability of rheumatoid arthritis being greater than a threshold value indicates that the subject to be tested suffers from rheumatoid arthritis or a related disease or has a risk of developing rheumatoid arthritis or a related disease.

According to an embodiment of the present disclosure, a training data set is constructed using a multivariate statistical model based on the relative abundance information of each biomarker of multiple subjects with rheumatoid arthritis and multiple normal subjects. Optionally, the multivariate statistical model is a random forest model.

According to an embodiment of the present disclosure, the training data set is a matrix, wherein each row represents each biomarker of the biomarker panel according to any one of claims 1 to 5, each column represents a sample, each cell represents the relative abundance spectrum of the biomarker in the sample, and the sample disease state is a vector, wherein 1 represents rheumatoid arthritis and 0 represents the control.

According to an embodiment of the present disclosure, the relative abundance information of each of Bifidobacterium dentate, RA-2633, Enterococcus, RA-781, Gordonibacter pamelaeae, RA-3396, RA-6638, RA-2441, RA-527, Clostridium, RA-2637, Citrobacter, Eubacterium, Citrobacter, RA-3215, Con-1722, Con-4360, Con-4212, Con-1261, Bifidobacterium bifidum, Klebsiella pneumoniae, Con-1423, Veillonella, Con-4095, Con-4103, Con-1735, Con-1710, Con-1832, and Con-1170, such as Bifidobacterium dentate JCVIHMP022, Prevotella CB7, DSM 18205, Enterococcus faecium E980, Ruminococcus ovale A2-162, Gordonibacter pamelaeae 7-10-1-bT, DSM 19378, Ruminococcus brucei L2-63, Eubacterium convexum ATCC 27560, Klebsiella oxytoca KCTC 1686, Clostridium asparagiforme DSM 15981, Prevotella CB7, DSM 18205, Citrobacter freundii 4_7_47CFAA, Eubacterium 3_1_31, Citrobacter 30_2, Clostridium 7_2_43FAA, Vibrio rosenbergii M50/1, Dialister invisus DSM 15470, Bacteroides plebeius M12, DSM The relative abundance information of 17135, Bifidobacterium bifidum S17, Klebsiella pneumoniae NTUH-K2044, Veillonella oral taxon 158F0412, Comamonas testosteroni KF-1, Klebsiella pneumoniae NTUH-K2044, atypical Veillonella ACS-134-V-Col7a, Streptococcus australis ATCC700641, and Parabacteroides merdae ATCC 43184 was obtained based on the relative abundance information of SEQ ID NOs: 1 to 9319.

According to an embodiment of the present disclosure, the training data set is at least one of Table 8-1 and Table 8-2, and a probability of rheumatoid arthritis of at least 0.5 indicates that the subject to be tested suffers from rheumatoid arthritis or a related disease or is at risk of developing rheumatoid arthritis or a related disease.

In another aspect of the present disclosure, there is provided a use of the above-mentioned gene markers in preparing a kit for predicting the risk of rheumatoid arthritis or related diseases in a test subject, comprising:

(1) Collecting samples from the subjects to be tested;

(2) determining the relative abundance information of each biomarker of the biomarker panel according to any one of claims 1 to 5 in the sample obtained in step (1);

(3) The probability of rheumatoid arthritis is obtained by comparing the relative abundance information of each biomarker of the test subject with the training data set using a multivariate statistical model,

The probability of rheumatoid arthritis being greater than a threshold value indicates that the subject to be tested suffers from rheumatoid arthritis or a related disease or has a risk of developing rheumatoid arthritis or a related disease.

According to an embodiment of the present disclosure, a training data set is constructed using a multivariate statistical model based on the relative abundance information of each biomarker of multiple subjects with rheumatoid arthritis and multiple normal subjects. Optionally, the multivariate statistical model is a random forest model.

According to an embodiment of the present disclosure, the training data set is a matrix, wherein each row represents each biomarker of the biomarker panel according to any one of claims 1 to 5, each column represents a sample, each cell represents the relative abundance spectrum of the biomarker in the sample, and the sample disease state is a vector, wherein 1 represents rheumatoid arthritis and 0 represents the control.

According to an embodiment of the present disclosure, the relative abundance information of each of Bifidobacterium dentate, RA-2633, Enterococcus, RA-781, Gordonibacter pamelaeae, RA-3396, RA-6638, RA-2441, RA-527, Clostridium, RA-2637, Citrobacter, Eubacterium, Citrobacter, RA-3215, Con-1722, Con-4360, Con-4212, Con-1261, Bifidobacterium bifidum, Klebsiella pneumoniae, Con-1423, Veillonella, Con-4095, Con-4103, Con-1735, Con-1710, Con-1832, and Con-1170, such as Bifidobacterium dentate JCVIHMP022, Prevotella CB7, DSM 18205, Enterococcus faecium E980, Ruminococcus ovale A2-162, Gordonibacter pamelaeae 7-10-1-bT, DSM 19378, Ruminococcus brucei L2-63, Eubacterium convexum ATCC 27560, Klebsiella oxytoca KCTC 1686, Clostridium asparagiforme DSM 15981, Prevotella CB7, DSM 18205, Citrobacter freundii 4_7_47CFAA, Eubacterium 3_1_31, Citrobacter 30_2, Clostridium 7_2_43FAA, Vibrio rosenbergii M50/1, Dialister invisus DSM 15470, Bacteroides plebeius M12, DSM The relative abundance information of 17135, Bifidobacterium bifidum S17, Klebsiella pneumoniae NTUH-K2044, Veillonella oral taxon 158F0412, Comamonas testosteroni KF-1, Klebsiella pneumoniae NTUH-K2044, atypical Veillonella ACS-134-V-Col7a, Streptococcus australis ATCC700641, and Parabacteroides merdae ATCC 43184 was obtained based on the relative abundance information of SEQ ID NOs: 1 to 9319.

According to an embodiment of the present disclosure, the training data set is at least one of Table 8-1 and Table 8-2, and a probability of rheumatoid arthritis of at least 0.5 indicates that the subject to be tested suffers from rheumatoid arthritis or a related disease or is at risk of developing rheumatoid arthritis or a related disease.

In another aspect of the present disclosure, there is provided a method for diagnosing whether a subject has an abnormal state associated with a microbiota or is at risk of developing an abnormal state associated with a microbiota, comprising:

determining the relative abundance of the aforementioned biomarkers in a sample from the subject, and

Based on the relative abundance, it is determined whether the subject has an abnormal state associated with the microbiota or is at risk of developing an abnormal state associated with the microbiota.

According to an embodiment of the present disclosure, the method includes:

(1) Collecting samples from the subjects to be tested;

(2) determining the relative abundance information of each biomarker of the biomarker panel according to any one of claims 1 to 5 in the sample obtained in step (1);

(3) The probability of rheumatoid arthritis is obtained by comparing the relative abundance information of each biomarker of the test subject with the training data set using a multivariate statistical model,

The probability of rheumatoid arthritis being greater than a threshold value indicates that the subject to be tested suffers from rheumatoid arthritis or a related disease or has a risk of developing rheumatoid arthritis or a related disease.

According to an embodiment of the present disclosure, a training data set is constructed using a multivariate statistical model based on the relative abundance information of each biomarker of multiple subjects with rheumatoid arthritis and multiple normal subjects. Optionally, the multivariate statistical model is a random forest model.

According to an embodiment of the present disclosure, the training data set is a matrix, wherein each row represents each biomarker of the biomarker panel according to any one of claims 1 to 5, each column represents a sample, each cell represents the relative abundance spectrum of the biomarker in the sample, and the sample disease state is a vector, wherein 1 represents rheumatoid arthritis and 0 represents the control.

According to an embodiment of the present disclosure, the relative abundance information of each of Bifidobacterium dentate, RA-2633, Enterococcus, RA-781, Gordonibacter pamelaeae, RA-3396, RA-6638, RA-2441, RA-527, Clostridium, RA-2637, Citrobacter, Eubacterium, Citrobacter, RA-3215, Con-1722, Con-4360, Con-4212, Con-1261, Bifidobacterium bifidum, Klebsiella pneumoniae, Con-1423, Veillonella, Con-4095, Con-4103, Con-1735, Con-1710, Con-1832, and Con-1170, such as Bifidobacterium dentate JCVIHMP022, Prevotella CB7, DSM 18205, Enterococcus faecium E980, Ruminococcus ovale A2-162, Gordonibacter pamelaeae 7-10-1-bT, DSM 19378, Ruminococcus brucei L2-63, Eubacterium convexoventris ATCC 27560, Klebsiella oxytoca KCTC 1686, Clostridium asparagiforme DSM 15981, Prevotella CB7, DSM 18205, Citrobacter freundii 4_7_47CFAA, Eubacterium 3_1_31, Citrobacter 30_2, Clostridium 7_2_43FAA, Vibrio rosenbergii M50/1, Dialister invisus DSM 15470, Bacteroides plebeius The relative abundance information of M12, DSM17135, Bifidobacterium bifidum S17, Klebsiella pneumoniae NTUH-K2044, Veillonella oral taxon 158F0412, Comamonas testosteroni KF-1, Klebsiella pneumoniae NTUH-K2044, atypical Veillonella ACS-134-V-Col7a, Streptococcus australis ATCC700641, and Parabacteroides merdae ATCC 43184 was obtained based on the relative abundance information of SEQ ID NOs: 1 to 9319.

According to an embodiment of the present disclosure, the training data set is at least one of Table 8-1 and Table 8-2, and a probability of rheumatoid arthritis of at least 0.5 indicates that the subject to be tested suffers from rheumatoid arthritis or a related disease or is at risk of developing rheumatoid arthritis or a related disease.

BRIEF DESCRIPTION OF THE DRAWINGS

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

Figure 1. Gut or oral MLG allows classification of RA patients from healthy controls. (a, d, f) Receiver operating characteristic (ROC) curves for a training set of stool (a), teeth (d), and saliva (f) samples from untreated RA cases and unrelated normal controls (n = 157, 100, and 94 for stool, teeth, and saliva samples, respectively). Dots mark the false-positive and true-positive rates at the optimal threshold probability. (b) Classification of a stool test set consisting of 17 controls and 17 RA cases who were either related or unrelated. (c, e, g) Classification of stool (c), teeth (e), and saliva (g) samples of RA after DMARD treatment (n = 40, 38, and 24 for stool, teeth, and saliva samples, respectively). According to the European League Against Rheumatism (EULAR) criteria, a DAS28 < 2.6 indicates symptom remission. Classification results for all samples are listed in Table 12.

DETAILED DESCRIPTION

Example

The terms used herein have the meanings commonly understood by persons of ordinary skill in the art to which this invention relates. Terms such as "a," "an," and "the" are not intended to refer to only singular entities, but rather encompass general categories that are described using specific embodiments. Except as outlined in the claims, the terms herein are used to describe specific embodiments of the invention, but their usage does not limit the invention.

Implementation Method

Example 1. Identification and validation of biomarkers for assessing risk of rheumatoid arthritis

1. Materials and Methods

1.1 Sample collection and DNA extraction

The present inventors collected stool samples from a total of 212 individuals (Table 1-1, stool samples, dental plaque samples and saliva samples), including a training set (n = 157, 77 untreated RA cases and 80 healthy controls) and a test set (for related case-control pairs, n = 34, i.e., 8 related case-control pairs and 9 unrelated case-control pairs; for DMARD-treated RA patients, n = 21).

Fecal samples were collected at Peking Union Medical College Hospital, transported frozen and extracted at BGI-Shenzhen (BGI-Shenzhen) as previously described (Qin, J. et al. A metagenome-wide association study of gut microbiotain type 2 diabetes. Nature 490, 55–60 (2012), incorporated herein by reference). Dental plaque was scraped from the tooth surface using ophthalmic forceps until a volume of 3 μl was obtained. The samples were transferred to 200 μl of 1× lysis buffer containing 10 mM Tris, 1 mM EDTA, 0.5% Tween 20 and 200 μg/ml proteinase K (Fermentas) and incubated at 55°C for 2 hours. Lysis was terminated by incubation at 95°C for 10 minutes, and the samples were frozen at -80°C before transportation. DNA extraction was performed according to the protocol for fecal samples. For saliva, 100 μl of saliva was added to 100 μl of 2× lysis buffer, swabbed the posterior pharyngeal wall and added to the same tube, and the sample was lysed and extracted as for the dental samples.

RA was diagnosed at Peking Union Medical College Hospital according to the 2010 ACR/EULAR classification criteria. All phenotypic information was collected at the time of the subjects' initial visit to the hospital according to standard procedures. RA patients aged between 18 and 65 years with disease duration of at least 6 weeks, at least 1 swollen joint and 3 tender joints were recruited. Patients were excluded if they had a history of chronic serious infection, any current infection or any type of cancer. Pregnant or breastfeeding women were excluded. All patients were informed of the risk of infertility and patients who wanted to have children were excluded. Although some patients had suffered from RA for many years, they were DMARD-naive because they had not been diagnosed with RA at a local hospital before visiting Peking Union Medical College Hospital and they only took analgesics to relieve RA symptoms.

All phenotypic information was collected at the time of the subjects' initial hospital visit according to standard procedures. Only 21 of the 212 samples used to construct the gut microbial gene catalog were stool samples from DMARD-treated patients and were not analyzed in this article.

This study was approved by the institutional review boards of Peking Union Medical College Hospital and BGI-Shenzhen.

Table 1-1. Samples used for gene catalog construction

1.2 Metagenomic Sequencing and Assembly

Paired-end metagenomic sequencing (insert fragment 350 bp, sequence length 100 bp) was performed on the Illumina platform as previously described (Qin et al. 2012, supra). Sequencing reads were quality controlled and reassembled into contigs using SOAPdenovo v2.04 (Luo, R. et al. SOAPdenovo2: an empirically improved memory-efficient short-read de novo assembler. Gigascience 1, 18 (2012)., incorporated herein by reference). The average rate of host contamination was 0.37% for fecal samples, 5.55% for dental samples, and 40.85% for saliva samples.

1.3 Gene catalog construction

GeneMark v2.7d was used to predict genes in the assembled contigs. BLAT (Kent, W.J.BLAT--the BLAST-like alignment tool. Genome Res. 12, 656–64 (2002), incorporated herein by reference) was used to remove redundant genes with a threshold of 90% overlap and 95% identity (not allowing the presence of holes). This resulted in a non-redundant gene catalog of 3,800,011 genes for 212 stool samples (including 21 DMARD-treated samples) and a catalog of 3,234,997 genes for 203 oral samples (105 dental plaque samples and 98 saliva samples). The gene catalog from the stool samples was merged into an existing gut microbial reference catalog containing 4.3 million genes (Qin et al. 2012, supra) using BLAT (95% identity, 90% overlap), resulting in a final catalog of 5.9 million genes. The relative abundance of genes was determined by aligning high-quality sequencing reads to intestinal or oral reference gene catalogs using the same procedure as in the published T2D paper (Qin et al., 2012, supra).

1.4 Taxonomic annotation and abundance calculation

The predicted genes were assigned to taxonomies according to the IMG database (v400) using a previously described internal pipeline (Qin et al., 2012, supra), with 70% overlap and 65% identity assigned to phylum, 85% identity assigned to genus, and 95% identity assigned to species. The relative abundance of taxa was calculated from the relative abundance of taxa genes.

Significant differences in the relative abundance of taxa between patients and healthy controls were determined by Wilcoxon rank sum test (with p < 0.05).

1.5 Metagenomic Wage-Based Association Study (MGWAS)

For case-control comparisons of fecal microbiota, removal of genes detected in less than 6 samples (n=157) resulted in a set of 3,110,085 genes. 83,858 genes showed differences in relative abundance between controls and cases (p<0.01, Wilcoxon rank sum test, FDR=0.3285). These marker genes were clustered into MLGs based on their abundance changes in all samples (Qin et al., 2012, supra). For the construction of tooth MLGs, 209,820 marker genes were selected from 2,247,835 genes (present in at least 6 samples, n=105) (p<0.01, Wilcoxon rank sum test, FDR=0.072). For salivary MLG, the present inventors selected 206,399 marker genes (p<0.01, Wilcoxon rank sum test, FDR=0.088) from 2,404,726 genes (present in at least 6 samples, n=98).

Taxonomic assignment and abundance analysis were performed as previously described (Qin et al., 2012, supra) based on taxonomy and the relative abundance of their constituent genes. Briefly, assignment to species required that more than 90% of the genes in the MLG had more than 95% identity when aligned to the species genome, with 70% query overlap. Assignment of an MLG to a genus required that more than 80% of its genes aligned to the genome, with 85% identity in both DNA and protein sequences. The average identity to the genome calculated from all genes is shown for reference only. The MLGs were further clustered based on the Kendall correlation between their abundance in all samples regardless of case-control status, and the co-occurrence network was visualized using Cytoscape 3.0.2.

1.6 MLG-based classifier

The MLG abundance profiles of the training cohort (Tables 1-2) were 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 best set of MLG markers. The model was tested on one or more test sets and the prediction error was calculated.

For the random forest model, the "Random Forest 4.6-7 software package" packaged in R version 2.14 was used. The input was a training dataset (i.e., the relative abundance profile of the MLGs selected in the training sample), a sample disease status (the sample disease status of the training sample is a vector, 1 represents RA, 0 represents the control), and a test set (just the relative abundance profile of the MLGs selected in the test set). The inventors then used the random forest function from the random forest package in R software to construct a classifier and used the predict function to predict the test set. The output was the prediction result (probability of disease, with a threshold of 0.5, and if the probability of disease is ≥0.5, the subject is at risk of RA).

Table 1-2. Sample information of the training set (selected from the samples used for gene catalog construction in Table 1-1)

2. Results

Microbiome-based identification and validation of RA patients

To further illustrate the diagnostic or prognostic value of RA-associated microbiota, the present inventors first constructed a random forest disease classifier based on intestinal MLG. A model using 29 of the 85 intestinal MLG markers (at least 100 genes) from controls and cases gave the lowest prediction error and receiver operating characteristic (ROC) curve area (AUC) of 0.977 in the training set (n = 157) (Figure 1a, Table 2-1, Table 2-2, Table 5, Table 8-1, Table 8-2). For the test set consisting of consanguineous case-control pairs and unrelated case-control pairs (n = 34, Tables 1-3), the overall error rate was 32% (Figure 1b, Table 11) and the AUC was 0.706. Thus, the performance of the gut MLG-based model on the training set and, where applicable, the test set was comparable to or exceeded that of existing classifiers based on RA serum markers (Van der Helm-van Mil, A.H.M. Risk estimation in rheumatoid arthritis—from bench tobedside. Nat. Rev. Rheumatol. (2014). doi: 10.1038/nrrheum.2013.215, incorporated herein by reference).

Similarly, 28 MLGs selected from 171 dental MLGs (at least 100 genes) (Table 3-1, Table 3-2, Table 6, Table 9-1, Table 9-2) gave an AUC of 0.864 in the training set (Figure 1d). 19 MLGs selected from 142 salivary MLGs (at least 100 genes) (Table 4-1, Table 4-2, Table 7, Table 10-1, Table 10-2) gave an AUC of 0.898 (Figure 1f). These results indicate that fecal, dental, and salivary microbial markers are all very useful for diagnosing RA.

Furthermore, testing the gut and tooth MLG classifiers on samples from patients treated with DMARDs (Tables 1-3) still identified the majority of them as RA patients, whereas tooth samples with low disease activity (DAS28) were more often classified as healthy (Figures 1c, 1e, Table 12), suggesting that the dental microbiota faithfully reflects the effects of DMARD treatment. Furthermore, saliva samples from patients treated with DMARDs were often classified as controls, likely due to direct modulation of the saliva microbiota by DMARDs (Figure 1g, Table 12). Taken together, these results suggest that gut and oral MLG can distinguish between effective and ineffective treatments and facilitate the evaluation of therapeutic strategies.

Table 1-3 Sample information of the test set

Table 5. SEQ IDs of 29 optimal intestinal markers

MLG ID SEQ ID NO: Number of genes mlg_id:2441 1～159 159 mlg_id:4103 160～304 145 mlg_id:4212 305～709 405 mlg_id:1047 710～856 147 mlg_id:1735 857～1536 680 mlg_id:4360 1537～1646 110 mlg_id:1796 1647～1798 152 mlg_id:3396 1799～2071 273 mlg_id:2472 2072～2309 238 mlg_id:1261 2310～2991 682 mlg_id:1832 2992～3093 102 mlg_id:6638 3094～3214 121 mlg_id:1722 3215～3353 139 mlg_id:1423 3354～3455 102 mlg_id:1170 3456～3558 103 mlg_id:3215 3559～3739 181 mlg_id:4095 3740～4381 642 mlg_id:2637 4382～4754 373 mlg_id:905 4755～4885 131 mlg_id:4111 4886～6743 1858 mlg_id:1710 6744～6862 119 mlg_id:2633 6863～7113 251 mlg_id:819 7114～7425 312 mlg_id:4158 7426～7736 311 mlg_id:527 7737～7854 118 mlg_id:784 7855～8048 194 mlg_id:2473 8049～8758 710 mlg_id:781 8759～8869 111 mlg_id:5 8870～9319 450

Table 6. SEQ IDs of 28 optimal tooth markers

Table 7. SEQ IDs of 19 optimal salivary markers

MLG ID SEQ ID NO: Number of genes mlg_id:1238 1～126 126 mlg_id:1559 127～231 105 mlg_id:6908 232～360 129 mlg_id:1141 361～519 159 mlg_id:6746 520～697 178 mlg_id:1 698～5680 4983 mlg_id:27683 5681～5851 171 mlg_id:1374 5852～6032 181 mlg_id:13 6033～8482 2450 mlg_id:1073 8483～9597 1115 mlg_id:29 9598～10469 872 mlg_id:636 10470～11246 777 mlg_id:9651 11247～11383 137 mlg_id:305 11384～11485 102 mlg_id:12 11486～14228 2743 mlg_id:20 14229～16239 2011 mlg_id:2831 16240～17605 1366 mlg_id:13621 17606～18115 510 mlg_id:27616 18116～9319 123

Therefore, the present inventors have identified and validated a marker panel (29 intestinal MLGs, 28 dental MLGs, and 19 salivary MLGs) based on RA-related gene markers using a random forest model. Furthermore, the present inventors have constructed an RA classifier that assesses the risk of RA disease based on these RA-related intestinal microbiota.

Although exemplary embodiments have been shown and described, those skilled in the art should understand that the above embodiments should not be construed as limiting the present disclosure, and that changes, substitutions, and modifications may be made to the embodiments without departing from the spirit, principle, and scope of the present disclosure.

Claims (10)

1. A kit for identifying a biomarker group, the kit comprising probes and/or primers for said biomarker group, said biomarker group consisting of intestinal biomarkers including * Bifidobacterium dentium *, RA-2633, * Enterococcus* sp ., RA-781, * Gordonibacter pamelaeae *, RA-3396, RA-6638, RA-2441, RA-527, *Clostridium * sp ., RA-2637 , * Citrobacter* sp . , *Eubacterium* sp., * Citrobacter * sp . , RA-3215 , Con-1722 , Con-4360 , Con-4212 , Con-1261 , and * Bifidobacterium bifidum * . Klebsiella pneumoniae , Con-1423 , Veillonella sp . , Con-4095 , Con-4103 , Con-1735 , Con-1710 , Con-1832 , Con-1170, The gut microbiome markers consist of sequences including SEQ ID NO:1 to SEQ ID NO:9319. The nucleotide sequences of RA-2633 are shown in SEQ ID NO: 6863~7113, RA-781 in SEQ ID NO: 8759~8869, RA-3396 in SEQ ID NO: 1799~2071, RA-6638 in SEQ ID NO: 3094~3214, RA-2441 in SEQ ID NO: 1~159, RA-527 in SEQ ID NO: 7737~7854, RA-2637 in SEQ ID NO: 4382~4754, RA-3215 in SEQ ID NO: 3559~3739, Con-1722 in SEQ ID NO: 3215~3353, and Con-4360 in SEQ ID NO: 1799~2071. As shown in SEQ ID NO: 305-709, the nucleotide sequence of Con-4212 is shown in SEQ ID NO: 2310-2991, the nucleotide sequence of Con-1423 is shown in SEQ ID NO: 3354-3455, the nucleotide sequence of Con-4095 is shown in SEQ ID NO: 3740-4381, the nucleotide sequence of Con-4103 is shown in SEQ ID NO: 160-304, the nucleotide sequence of Con-1735 is shown in SEQ ID NO: 857-1536, the nucleotide sequence of Con-1710 is shown in SEQ ID NO: 6744-6862, the nucleotide sequence of Con-1832 is shown in SEQ ID NO: 2992-3093, and the nucleotide sequence of Con-1170 is shown in SEQ ID NO: 3456-3558. 2. The kit according to claim 1, wherein the intestinal biomarkers include Bifidobacterium tumefaciens JCVIHMP022, Prevotella copri CB7 , DSM 18205 , Enterococcus faecium E980, Ruminococcus obeum A2-162 , Gordonibacter pamelaeae7-10-1 -bT , DSM 19378, Ruminococcus bromii L2-63 , Eubacterium ventriosum ATCC 27560 , Klebsiella oxytoca KCTC 1686 , Clostridium asparagiforme DSM 15981 , and Citrobacter freundii . 4_7_ 47CFAA , Eubacterium sp . 3_1_31 , Citrobacter sp . 30_2 , Clostridium sp . 7_2_43FAA , Roseburia intestinalis M50/1 , Dialister invisus DSM 15470 , Bacteroides plebeius M12 , DSM 17135, Bifidobacterium bifidum S17 , Klebsiella pneumoniae NTUH-K2044 , Veillonella sp. oral taxon 158 F0412 , Comamonas testosteroni KF-1 Veillonella atypica ACS-134-V- Col7a , Streptococcus australis ATCC 700641 , Parabacteroides merdae ATCC 43184 . 3. A kit for identifying a set of biomarkers, comprising a primer set for PCR amplification and designed according to intestinal biomarkers as described in claim 1. 4. A kit for identifying a set of biomarkers, comprising one or more probe sets designed according to intestinal biomarkers as described in claim 1. 5. Use of a biomarker set in the preparation of a kit for predicting the risk of rheumatoid arthritis in a subject, said biomarker set comprising gut biomarkers including * Bifidobacterium dentium *, RA-2633, * Enterococcus* sp. , RA-781, * Gordonibacter pamelaeae *, RA-3396, RA-6638, RA-2441, RA-527, *Clostridium* sp ., RA-2637 , *Citrobacter* sp. , * Eubacterium * sp. , *Citrobacter * sp ., RA -3215 , Con-1722 , Con-4360 , Con-4212 , Con-1261 , and * Bifidobacterium bifidum *. Klebsiella pneumoniae , Con-1423 ; Veillonella sp . , Con-4095 , Con-4103 , Con-1735 , Con-1710 , Con-1832 , Con-1170, The gut microbiome markers consist of sequences including SEQ ID NO:1 ~ SEQ ID NO:9319. The nucleotide sequences of RA-2633 are shown in SEQ ID NO: 6863~7113, RA-781 in SEQ ID NO: 8759~8869, RA-3396 in SEQ ID NO: 1799~2071, RA-6638 in SEQ ID NO: 3094~3214, RA-2441 in SEQ ID NO: 1~159, RA-527 in SEQ ID NO: 7737~7854, RA-2637 in SEQ ID NO: 4382~4754, RA-3215 in SEQ ID NO: 3559~3739, Con-1722 in SEQ ID NO: 3215~3353, and Con-4360 in SEQ ID NO: 1799~2071. As shown in SEQ ID NO: 305-709, the nucleotide sequence of Con-4212 is shown in SEQ ID NO: 2310-2991, the nucleotide sequence of Con-1423 is shown in SEQ ID NO: 3354-3455, the nucleotide sequence of Con-4095 is shown in SEQ ID NO: 3740-4381, the nucleotide sequence of Con-4103 is shown in SEQ ID NO: 160-304, the nucleotide sequence of Con-1735 is shown in SEQ ID NO: 857-1536, the nucleotide sequence of Con-1710 is shown in SEQ ID NO: 6744-6862, the nucleotide sequence of Con-1832 is shown in SEQ ID NO: 2992-3093, and the nucleotide sequence of Con-1170 is shown in SEQ ID NO: 3456-3558; The uses include: (1) Collect samples from the subject to be tested; (2) Determine the relative abundance information of each biomarker in the biomarker group obtained in step (1); (3) The probability of rheumatoid arthritis is obtained by comparing the relative abundance information of each biomarker of the subject with the training dataset using a multivariate statistical model. The probability of having rheumatoid arthritis greater than a threshold indicates that the subject has rheumatoid arthritis or is at risk of developing rheumatoid arthritis. 6. The use according to claim 5, wherein the training dataset is constructed using the multivariate statistical model based on the relative abundance information of various biomarkers of multiple subjects with rheumatoid arthritis and multiple normal subjects. 7. The use according to claim 6, wherein the multivariate statistical model is a random forest model. 8. The use according to claim 6, wherein the training dataset is a matrix, wherein each row represents a biomarker of the biomarker group according to any one of claims 1 to 3, each column represents a sample, each cell represents the relative abundance spectrum of the biomarkers in the sample, and the sample disease state is a vector, wherein 1 represents rheumatoid arthritis and 0 represents control. 9. The use according to claim 8, wherein * Bifidobacterium dentium *, RA-2633, *Enterococcus * sp. , RA-781, * Gordonibacter pamelaeae *, RA-3396, RA-6638, RA-2441, RA-527, * Clostridium* sp ., RA-2637 , *Citrobacter* sp . , *Eubacterium * sp . , *Citrobacter* sp. , RA-3215 , Con-1722 , Con-4360 , Con-4212 , Con-1261 , * Bifidobacterium bifidum * , * Klebsiella pneumoniae* , Con-1423 , * Veillonella* spp. The relative abundance information of each of Veillonella sp . , Con-4095 , Con-4103 , Con-1735 , Con-1710 , Con-1832 and Con-1170 was obtained from the relative abundance information of SEQ ID NO: 1 to SEQ ID NO: 9319. 10. The use according to claim 6, wherein the training dataset is at least one of Table 8-1 and Table 8-2, and the probability of rheumatoid arthritis is at least 0.5, indicating that the subject has rheumatoid arthritis or is at risk of developing rheumatoid arthritis.