Construction method of colorectal adenoma prediction model and colorectal adenoma prediction system

Abstract

According to the construction method of the colorectal adenoma prediction model and the colorectal adenoma prediction system provided by the invention, 14 kinds of colorectal adenoma related living habits and two kinds of colorectal adenoma related intestinal bacteria are found by collecting and analyzing the living habits and intestinal flora information of colorectal adenoma and healthy control people; the abundance information of the living habits and intestinal bacteria can be used as markers to jointly predict the occurrence risk of colorectal adenoma. The method is high in prediction accuracy, convenient, efficient, low in cost and suitable for wide screening in large asymptomatic people, and the problem that colorectal adenoma is difficult to screen can be solved.

Description

Technical Field

[0001] The present invention relates to the field of biomedicine, and in particular to a method for constructing a colorectal adenoma prediction model and a colorectal adenoma prediction system. Background Art

[0002] Colorectal adenoma is the most important precancerous disease of colorectal cancer, and more than 90% of colorectal cancers develop from colorectal adenoma. Considering the insidious onset and serious harm of colorectal cancer, screening, early diagnosis and early treatment of colorectal adenoma are important ways to prevent the occurrence of colorectal cancer.

[0003] According to the 2021 Consensus on Comprehensive Prevention of Colorectal Cancer in China, fecal immunochemical tests and colonoscopy are the main screening methods currently recommended in my country, and their specific screening strategies can be summarized as "one-step method" and "two-step method". Among them, the "one-step method" refers to direct screening with a colonoscope, with diagnosis and treatment carried out simultaneously, a long screening cycle, and high-quality colonoscopy is the gold standard for colorectal cancer screening. The "two-step method" aims to reduce invasive colonoscopies. Low-invasive screening tools (such as fecal immunochemical tests) are used first. When the screening result is positive, sequential colonoscopy is required to confirm the diagnosis. If you are unwilling to use colonoscopy and fecal immunochemistry, you can also consider using sigmoidoscopy, computed tomography colonography, colon capsule endoscopy, and fecal multi-target DNA for initial screening.

[0004] According to the 2021 Consensus on Comprehensive Prevention of Colorectal Cancer in China, fecal immunochemical test and colonoscopy are the main screening methods currently recommended in my country. However, the sensitivity and specificity of fecal immunochemical test are insufficient, and the cost of colonoscopy is high, which makes it impossible to screen the population on a large scale. Summary of the invention

[0005] The purpose of the present invention is to provide a method for constructing a colorectal adenoma prediction model and a colorectal adenoma prediction system.

[0006] To solve the above problems, the present invention provides a method for constructing a colorectal adenoma prediction model, comprising: collecting stool and scores of various types of living habits of asymptomatic subjects for colorectal adenoma screening; the asymptomatic subjects include healthy subjects and colorectal adenoma patients;

[0007] Based on the scores of various types of living habits of asymptomatic subjects in the sample, a univariate logistic regression analysis was performed on the various types of living habits of the subjects and colorectal adenoma to obtain the types of living habits after pre-screening;

[0008] Amplifying and quantifying the gene fragments of the target bacterial species in the DNA of the feces of each sample asymptomatic subject to obtain quantitative result information of the target bacterial species of each sample asymptomatic subject;

[0009] Based on the quantitative result information of the target bacterial species and the scores of the living habits of each species after pre-screening, a prediction model for colorectal adenoma is trained to obtain a trained prediction model for colorectal adenoma.

[0010] Furthermore, in the above method, a univariate logistic regression analysis is performed on the various types of living habits of the subjects and colorectal adenoma to obtain the types of living habits after pre-screening, including:

[0011] Univariate Logistic regression analysis was performed on the subjects' various types of living habits and colorectal adenoma, and the influencing factors with an absolute value of β greater than 0.05 and a P value less than 0.25, i.e., the types of living habits, were obtained as the types of living habits after pre-screening;

[0012] The gene fragments of the target bacteria species in the DNA of the feces of each sample asymptomatic subject are amplified and quantified to obtain quantitative result information of the target bacteria species of each sample asymptomatic subject, including:

[0013] The gene fragments of Fusobacterium nucleatum and pks+ Escherichia coli in the DNA of the feces of each sample asymptomatic subject are amplified and quantified to obtain the quantitative result information of Fusobacterium nucleatum and the quantitative result information of pks+ Escherichia coli of each sample asymptomatic subject.

[0014] Furthermore, in the above method, the gene fragments of Fusobacterium nucleatum and pks+ Escherichia coli in the DNA of the feces of each sample asymptomatic subject are amplified and quantified to obtain the quantitative result information of Fusobacterium nucleatum and the quantitative result information of pks+ Escherichia coli of each sample asymptomatic subject, including:

[0015] DNA was extracted from the feces of each asymptomatic subject using a fecal DNA extraction kit to obtain fecal genomic DNA;

[0016] By using real-time quantitative fluorescence qPCR technology and selecting 16S rRNA as the internal reference of qPCR, Fusobacterium nucleatum and pks+ Escherichia coli were amplified and the abundance was quantified to obtain the quantitative result information of Fusobacterium nucleatum and the quantitative result information of pks+ Escherichia coli.

[0017] Furthermore, in the above method, DNA is extracted from the feces of each sample asymptomatic subject using a fecal DNA extraction kit to obtain fecal genomic DNA, including:

[0018] About 100% of the stool from each sample collected from asymptomatic subjects was placed in a bead tube;

[0019] Add grinding solution to the grinding bead test tube and shake gently to fully mix the stool sample with the grinding beads and grinding solution in the grinding bead test tube;

[0020] Obtain the C1 solution, and check in advance whether there is any crystal precipitation in the C1 solution. If there is precipitate at the bottom of the bottle of the C1 solution, place the C1 solution in a water bath and heat it until the precipitate is completely dissolved;

[0021] Add the completely dissolved C1 solution to the grinding bead test tube, vortex lightly and mix thoroughly;

[0022] Place the grinding bead test tube after the C1 solution is fully mixed in a water bath and heat it in a water bath;

[0023] Fix the grinding bead test tube heated in a water bath horizontally on a vortex machine and oscillate horizontally at the maximum speed of the vortex machine. Pause once during the oscillation and let it stand to help break the cells in the feces.

[0024] Place the horizontally shaken grinding bead test tube into a high-speed centrifuge for centrifugation until the supernatant is transparent, so as to reduce impurities in the final product;

[0025] Avoid the sediment at the bottom of the grinding bead tube and use a pipette to transfer the first supernatant above the sediment to the prepared EP collection tube;

[0026] Add C2 reagent to the first supernatant in the EP collection tube, shake and mix, and then place in a refrigerator for incubation to precipitate impurities such as polysaccharides, cell debris and proteins to obtain an incubated second supernatant;

[0027] Place the EP collection tube containing the incubated second supernatant into a high-speed centrifuge for centrifugation;

[0028] Avoid the flocculent precipitate at the bottom of the EP collection tube after centrifugation, and use a pipette to transfer the second supernatant to the first sterilized EP tube;

[0029] Add C3 reagent to the second supernatant in the first sterilized EP tube, shake and mix, and then place in a refrigerator for incubation to continue to precipitate impurities such as polysaccharides, cell debris and proteins to obtain an incubated third supernatant;

[0030] Place the first sterilized EP tube containing the incubated third supernatant into a high-speed centrifuge for centrifugation;

[0031] Avoid the flocculent precipitate at the bottom of the first sterilized EP tube after centrifugation, and use a pipette to transfer about a third of the supernatant to the second sterilized EP tube;

[0032] The C4 reagent is shaken and mixed, and the C4 reagent is added to the third supernatant and lightly shaken to mix so as to bind and adsorb the DNA in the feces to obtain a mixed solution;

[0033] Add the mixed solution to the Spin Filter centrifugal column each time, centrifuge it in a high-speed centrifuge, and discard the filtrate; repeat this step until all the mixed solution is loaded, so that the DNA is selectively adsorbed on the filter membrane of the Spin Filter centrifugal column;

[0034] Add C5 reagent to the Spin Filter centrifugal column, let it stand and then centrifuge it in a high-speed centrifuge, and discard the filtrate in the collection tube of the Spin Filter centrifugal column;

[0035] Transfer the Spin Filter column to a collection tube and centrifuge it again to remove the residual C5 reagent;

[0036] Open the cap of the collection tube and let it dry;

[0037] Transfer the dried Spin Filter centrifugal column to the sterilized third EP tube, add C6 reagent directly to the center of the white filter membrane of the Spin Filter centrifugal column, and incubate at room temperature to elute and adsorb the DNA on the centrifugal column filter membrane;

[0038] The third EP tube was placed in a high-speed centrifuge for centrifugation, the Spin Filter centrifugal column was discarded, and the purified fecal genomic DNA was obtained.

[0039] Furthermore, in the above method, based on the quantitative result information of the target bacteria and the scores of the living habits of each species after pre-screening, a prediction model for colorectal adenoma is trained to obtain a trained prediction model for colorectal adenoma, including:

[0040] The quantitative result information of Fusobacterium nucleatum of all asymptomatic subjects, the quantitative result information of Escherichia coli with pks+ and the scores of various types of living habits after pre-screening are used as a data set; the data set is divided into a training set, a validation set and a test set, wherein the data in the training set, the validation set and the test set do not overlap with each other, and the training set, the validation set and the test set all include healthy subjects and colorectal adenoma test patients;

[0041] A multivariate Logistic regression model is established using the statistical software R, and the multivariate Logistic regression model is trained using the training set; the trained Logistic regression model is verified using the verification set to obtain a trained prediction model for colorectal adenoma;

[0042] The trained prediction model for colorectal adenoma is evaluated by using the test set to obtain a trained prediction model for colorectal adenoma.

[0043] Furthermore, in the above method, a multivariate logistic regression model is established using statistical software R, and the multivariate logistic regression model is trained using the training set, including:

[0044] In the cohort of the training set, multivariate logistic regression analysis was used to establish a multivariate logistic regression model for the pre-screened influencing factors, and then the redundant variables were deleted by the backward elimination method to streamline the multivariate logistic regression model;

[0045] The trained prediction model for colorectal adenoma is evaluated by using the test set to obtain a trained prediction model for colorectal adenoma, including:

[0046] The trained multivariate logistic regression model, i.e., the trained prediction model for colorectal adenoma, was evaluated using the test set, including using the pROC package of R and plotting using the ggplot2 package to obtain the trained prediction model for colorectal adenoma.

[0047] Furthermore, in the above method, the trained prediction model for colorectal adenoma retains 14 types of living habits and two types of intestinal bacteria, among which:

[0048] 14 types of lifestyle habits, including: age, gender, sedentary time, hyperlipidemia, family history, cancer history, processed meat consumption, vegetable and fruit consumption, grain consumption, sugary drink consumption, coffee consumption, smoking status, alcohol consumption, and aspirin use history;

[0049] Two intestinal bacteria, including: Fusobacterium nucleatum and pks+ Escherichia coli.

[0050] Furthermore, in the above method, the trained colorectal adenoma prediction model is as follows:

[0051] Colorectal adenoma risk index = Y / (1+Y);

[0052] Where, Y = exp(-6.42 + 0.73*X1 + 0.61*X2 + 0.56*X3 + 0.52*X4 + 0.35*X5 + 0.32*X6 + 0.3*X7 + 0.28*X8 + 0.25*X9 + 0.22*X 10 +0.21*X 11 +0.15*X 12 +0.08*X 13 -0.55*X14 -0.82*X 15 –1.79*X 16 );

[0053] X1 is the score of Fusobacterium nucleatum, X2 is the score of cancer history, X3 is the score of sugary drink consumption, X4 is pks + coli score, X5 is the score of coffee consumption, X6 is the score of gender, X7 is the score of processed meat consumption, X8 is the score of alcohol consumption, X9 is the score of hyperlipidemia, X 10 is the score of smoking status, X 11 is the score of sedentary time, X 12 For the score of family history, X 13 is the age score, X 14 is the score of grain consumption, X 15 is the score of vegetable and fruit consumption, X 16 Score the history of aspirin use.

[0054] According to another aspect of the present invention, a colorectal adenoma prediction system is provided, comprising:

[0055] A model training module, used to obtain a colorectal adenoma prediction model using any of the methods described above;

[0056] The prediction module is used to input the scores of the living habits of the person to be tested and the quantitative result information of the target bacteria species into the trained rectal adenoma prediction model to obtain the prediction results of colorectal adenoma of the person to be tested.

[0057] According to another aspect of the present invention, a computer-readable storage medium is provided, on which computer-executable instructions are stored, wherein when the computer-executable instructions are executed by a processor, the processor is caused to: execute any of the methods described above.

[0058] Compared with the existing technology, this application collected and analyzed the living habits and intestinal flora information of more than 1,800 cases of colorectal adenoma and healthy controls, and found 14 colorectal adenoma-related living habits and two colorectal adenoma-related intestinal bacteria. These living habits and intestinal bacteria abundance information can be used as markers to jointly predict the risk of colorectal adenoma. It can provide a high prediction accuracy, convenient and efficient, low cost, suitable for extensive screening in large asymptomatic populations. It can solve the problem of difficult colorectal adenoma screening. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a flow chart of a method for constructing a colorectal adenoma prediction model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0060] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0061] like Figure 1 As shown, the present invention provides a method for constructing a colorectal adenoma prediction model, comprising: steps S1 to S4.

[0062] Step S1, collecting stool and various lifestyle scores of asymptomatic subjects for colorectal adenoma screening; the asymptomatic subjects include healthy subjects and colorectal adenoma patients;

[0063] Here, since most patients with colorectal adenoma are asymptomatic in the early stage, that is, they have no abdominal pain, diarrhea, constipation, blood in the stool, abdominal distension and other discomfort symptoms. Therefore, such patients are not easy to be discovered in the early stage, and when discomfort symptoms appear, they have already developed colorectal cancer.

[0064] Healthy subjects and patients with colorectal adenoma can be examined by colonoscopy. If no colorectal adenoma is found under colonoscopy, the subjects are healthy subjects and can be labeled as healthy subjects; if colorectal adenoma is found under colonoscopy, the subjects are colorectal adenoma test patients and can be labeled as colorectal adenoma test patients. The endoscopic diagnosis and pathological diagnosis of each subject's colonoscopy can be obtained from the hospital's endoscopic report query system and pathological report query system.

[0065] Optionally, the types of living habits include:

[0066] Name, gender, age, height, weight, body mass index, education level, marital status, work status, monthly income, history of chronic diseases, medication history, cancer history, family history, surgical history, smoking and drinking history, sleep conditions, mental stress, eating habits and exercise conditions, etc.

[0067] Specifically, various types of information on the subjects' living habits can be collected, including: name, gender, age, height, weight, body mass index, education level, marital status, work status, monthly income, chronic disease history, medication history, tumor history, family history, surgery history, smoking and drinking history, sleep status, mental stress, eating habits and exercise status, etc. Among them, the mental stress of the subjects can be assessed by the stress perception scale PSS14; eating habits can be assessed by a semi-quantitative food frequency table, the food types and standards of which are formulated with reference to the "Dietary Guidelines for Chinese Residents (2016 Edition)"; and the exercise status of each subject can be quantitatively evaluated by calculating the weekly metabolic equivalents based on the International Physical Activity Scale Short Questionnaire.

[0068] Step S2, based on the scores of various types of living habits of the sample asymptomatic subjects, a univariate Logistic regression analysis is performed on the various types of living habits of the subjects and colorectal adenoma to obtain the types of living habits after pre-screening;

[0069] Preferably, a univariate logistic regression analysis is performed on the various types of living habits of the subjects and colorectal adenoma, and the influencing factors, i.e., the types of living habits, whose absolute values ​​of β values ​​are greater than 0.05 and whose P values ​​are less than 0.25 are obtained as the types of living habits after pre-screening;

[0070] Here, a logistic regression analysis was performed on each type of lifestyle and colorectal adenoma to explore which lifestyles are associated with the occurrence of colon adenoma. A β value and a P value will be obtained for each lifestyle and the occurrence of colorectal adenoma. The β value must be > 0.05 and P < 0.25, and it is considered that there is a certain correlation between this type of lifestyle and adenoma. There are a total of 23 types of such related lifestyles, specifically: the subject's age, gender, cancer history, smoking status, sweets consumption, green tea consumption, processed meat consumption, drinking status, family history, seafood and fish consumption, sugary beverage consumption, history of hyperlipidemia, coffee consumption, sedentary time, history of hypertension, number of days of light physical exercise per week, walking time, sleep time, mental stress, history of calcium channel blocker use, grain consumption, vegetable and fruit consumption, and aspirin use;

[0071] Step S3, amplifying and quantifying the gene fragments of the target bacteria in the DNA of the feces of each sample asymptomatic subject to obtain the quantitative result information of the target bacteria in each sample asymptomatic subject; preferably, the gene fragments of Fusobacterium nucleatum and pks+ Escherichia coli (Escherichia coli) in the DNA of the feces of each sample asymptomatic subject can be amplified and quantified to obtain the quantitative result information of Fusobacterium nucleatum and the quantitative result information of pks+ Escherichia coli of each sample asymptomatic subject;

[0072] Preferably, step S3 includes: step S31 and step S32.

[0073] Step S31, extracting DNA from the feces of each sample asymptomatic subject using a fecal DNA extraction kit to obtain fecal genomic DNA;

[0074] Specifically, Qiagen's PowerFecal TM DNA Isolation kit, catalog number 12830-50. In detail, step S31, including specific experimental steps for extracting fecal DNA, can be as follows:

[0075] Step S311, about 0.2-0.25 g of feces is selected from the frozen feces of each sample asymptomatic subject and placed in a 2 ml bead beating tube;

[0076] Step S312, add 750 μl of beed solution into the beed bead tube and shake slightly to fully mix the stool sample with the beed beads and beed solution in the beed bead tube;

[0077] Step S313, obtaining a C1 solution, and checking in advance whether there is any crystal precipitation in the C1 solution. If there is a precipitate at the bottom of the C1 solution, the C1 solution is placed in a water bath at a preset temperature, such as 65 degrees Celsius, and heated until the precipitate is completely dissolved;

[0078] Step S314, adding a first preset amount, such as 65 μl, of the completely dissolved C1 solution into a bead beating tube, vortexing slightly and mixing thoroughly;

[0079] Step S315, placing the bead beating tube after the C1 solution is fully mixed in a 65 degree Celsius water bath, and heating the water bath for a first preset time, such as 15 minutes;

[0080] Step S316, fix the bead beating tube after water bath heating for the first preset time horizontally on the vortex machine, and oscillate horizontally at the maximum speed of the vortex machine for a preset second time, such as 30 minutes, and pause once during the oscillation and let it stand for a preset third time, such as 3 minutes, to help rupture cells in the feces;

[0081] It is worth noting that it is necessary to ensure that the contents of each bead beating tube are complete, otherwise it is easy to cause incomplete lysis; in addition, for stool specimens with low water content or long storage time, the shaking time can be appropriately extended, but it should not exceed 1 hour, otherwise the DNA chain will easily break and produce fragments.

[0082] Step S317, placing the horizontally shaken bead bead tube into a high-speed centrifuge and centrifuging at maximum speed for a fourth preset time, such as 2 minutes, until the supernatant is transparent, so as to reduce impurities in the final product;

[0083] Step S318, avoiding the sediment at the bottom of the bead beating tube, using a pipette to carefully transfer a second preset amount, such as 400-500 μl of the first supernatant above the sediment to a prepared preset volume, such as 2 ml, of the EP collection tube;

[0084] Step S319, adding a third preset amount, such as 250 μl, of C2 reagent to the first supernatant in the EP collection tube, shaking and mixing, and then placing in a refrigerator at a preset temperature, such as 4 degrees Celsius, for incubation for a fifth preset time, such as 10 minutes, to precipitate impurities such as polysaccharides, cell debris and proteins, so as to obtain an incubated second supernatant;

[0085] Step S320, placing the EP collection tube containing the incubated second supernatant into a high-speed centrifuge and centrifuging at 15000 g for a sixth preset time, such as 2 minutes;

[0086] Step S321, avoiding the flocculent precipitate at the bottom of the EP collection tube after centrifugation, carefully transfer about a fourth preset amount, such as 500 μl, of the second supernatant into a first sterilized preset volume, such as 2 ml EP tube, using a pipette;

[0087] Step S322, adding a fifth preset amount, such as 200 μl, of C3 reagent to the second supernatant in the first sterilized EP tube, shaking and mixing, and placing in a refrigerator at a preset temperature, such as 4 degrees Celsius, for incubation for a seventh preset time, such as 10 minutes, to continue precipitating impurities such as polysaccharides, cell debris and proteins, so as to obtain an incubated third supernatant;

[0088] Step S323, placing the first sterilized EP tube containing the incubated third supernatant into a high-speed centrifuge and centrifuging at 15000 g for the eighth preset time of 2 minutes;

[0089] Step S324, avoiding the flocculent precipitate at the bottom of the first sterilized EP tube after centrifugation, carefully transfer about a sixth preset amount, such as 750 μl, of the third supernatant into a clean second sterilized 2 ml EP tube using a pipette;

[0090] Here, if there is a lot of precipitation and the supernatant is turbid, you can consider repeating steps S319 to S324 to remove impurities such as polysaccharides and proteins to increase the purity of the final product;

[0091] Step S325, shaking and mixing the C4 reagent, adding a seventh preset volume, such as 1200 μl, of the C4 reagent to the third supernatant obtained in step S324, and gently shaking and mixing for a ninth preset time, such as 5 seconds, to bind and adsorb the DNA in the feces to obtain a mixed solution;

[0092] Step S326, adding the mixed solution of the eighth preset volume, such as 650ul, to the Spin Filter centrifugal column each time, centrifuging at 15000g for 2 minutes in a high-speed centrifuge, and discarding the filtrate; repeating this step until all the mixed solution is loaded, so that the DNA is selectively adsorbed on the filter membrane of the Spin Filter centrifugal column;

[0093] Step S327, add a ninth preset volume, such as 500 μl, of C5 reagent to the Spin Filter centrifugal column, let it stand for 3 minutes, then centrifuge it at 15000 g for 2 minutes in a high-speed centrifuge, and discard the filtrate in the collection tube of the Spin Filter centrifugal column; for specimens with turbid filtrate or dark filter membrane color, repeat this step to fully remove impurities;

[0094] Step S328, transfer the Spin Filter centrifugal column to a clean collection tube, and centrifuge it again at 15000 g for 2 minutes to remove the residual C5 reagent;

[0095] Step S329, open the cover of the collection tube and dry it for 5 minutes;

[0096] Step S330, transfer the air-dried Spin Filter centrifugal column to a sterilized third 1.5 ml EP tube, add a tenth preset volume, such as 50-100 μl, of C6 reagent directly to the center of the white filter membrane of the Spin Filter centrifugal column, and incubate at room temperature for 10 minutes to elute and adsorb the DNA on the filter membrane of the spin filter centrifugal column;

[0097] Step S331, placing the third 1.5 ml EP tube into a high-speed centrifuge and centrifuging at 15000 g for 2 minutes, discarding the SpinFilter centrifugal column, and obtaining purified fecal genomic DNA;

[0098] Step S32, using real-time quantitative fluorescence qPCR technology and selecting 16S rRNA as an internal reference for qPCR, amplifying Fusobacterium nucleatum and pks+ Escherichia coli and quantifying the abundance to obtain quantitative result information of Fusobacterium nucleatum and pks+ Escherichia coli.

[0099] Here, in step S3, DNA is extracted from the stool sample using a stool DNA extraction kit, and the Fusobacterium nucleatum and pks+ Escherichia coli in the extracted stool DNA are quantified using qPCR technology.

[0100] The quantitative result information is the relative abundance value output by the qPCR (PCR) result. The second stage of qPCR will have 40 cycles. Each time qPCR runs a cycle, the DNA sequence of the target bacteria detected will be replicated and amplified by one time. The amplified DNA carries a fluorescent signal, which can be detected when it exceeds a certain intensity. Then, the number of cycles in which the sample is detected can be recorded. This number of cycles is the CT value. The negative logarithm of the difference between the CT value of the target bacteria and the CT value of the corresponding internal reference 16S rRNA is its relative abundance value.

[0101] Preferably, the cut-off point of Fusobacterium nucleatum for distinguishing healthy people from patients with colorectal adenoma is 2 to the power of -19.48. A cut-off point higher than 2 to the power of -19.48 is likely to be a patient with colorectal adenoma, and vice versa for a healthy person.

[0102] Preferably, the cut-off point for pks+ E. coli to distinguish between healthy people and patients with colorectal adenoma is 2 to the power of -23.61. Those with a higher probability than 2 to the power of -23.61 are patients with colorectal adenoma, and those with a lower probability are healthy people.

[0103] Preferably, the kit that can be used in this application is TB Green from TAKARA TM Premix ExTaq TM II reagent package, and can choose to use a 10μl reaction system, as shown in Table 1.

[0104] Table 1 TB Green Fast qRCR reaction system

[0105]

[0106]

[0107] The qPCR procedure is:

[0108] Stage 1: pre-denaturation at 95°C for 30 seconds;

[0109] The second stage—40 cycles × (denaturation at 95 °C for 5 s, annealing and extension at 60 °C for 30 s), with fluorescence collection at 60 °C.

[0110] After 40 PCR cycles, the sample was heated continuously from 55°C to 95°C, and the fluorescence signal was collected every 0.5°C.

[0111] The primer sequences for pks+ Escherichia coli (Escherichia coli) and Fusobacterium nucleatum were referenced to the primers designed by XanderW Huijsdens and M Carmen Collado, respectively, as shown in Table 2. Since 16S rRNA has few species, large content, and relatively fixed abundance, this application selected 16S rRNA as the internal reference for qPCR in this study. The specific sequence of 16S rRNA was determined through previous prior studies and our own preliminary experiments. The Primer-BLAST function in the NCBI database verified that the above primers have good specificity and can be used.

[0112] Table 2 Various primer sequences

[0113]

[0114] The primer sequences in Table 2 above were determined and synthesized by a biotechnology company. Before use, the EP tube containing the primer powder was centrifuged at 6000g for 2 minutes to deposit the primer powder floating in the EP tube to the bottom of the EP tube. According to the parameters of the primers, the corresponding amount of ddH2O or TE to be added was calculated and slowly added, fully shaken and mixed and centrifuged to ensure that the primer powder was configured into a 10μmol / L solution. The fecal genomic DNA extracted in step S31 was amplified as shown in Table 1 according to the above qPCR program. Three replicate wells were set for each sample. The number of cycles required for the fluorescence intensity in each reaction well to reach the corresponding threshold value determined the final cycle value (Cycle threshold, CT) of the target bacteria. The negative logarithm of the difference between the CT value of the target bacteria and the CT value of the corresponding internal reference 16S rRNA was calculated to obtain the relative abundance of the target bacteria.

[0115] Because the DNA content of the bacteria to be detected in my sample is very, very low, it needs to be amplified before it can be detected and quantified. Amplification is to mix all the reactants according to Table 1 first, and then implement the various temperature programs in the subsequent qPCR program, and my target DNA will be amplified one by one. The amplified new DNA is fluorescent, and it will be detected by the machine only when the fluorescence accumulates to a certain level. The first pair of primers 16SrRNA_F and 16S rRNA_R in Table 2 are internal parameters, which are relatively stable in each bacterium. There is an extra pair of internal parameters of 16S rRNA_F and 16S rRNA_R here because the qPCR method is not very stable, it may be high and low, and the variability is large. The results output by each bacterium need to subtract the internal parameters to make the results more consistent.

[0116] In Table 2, the letters F and R in the primer names represent forward and reverse, respectively. PCR primers are always a pair, and forward and reverse are equivalent to the two ends of a sequence.

[0117] Step S4, training a prediction model for colorectal adenoma based on the quantitative result information of Fusobacterium nucleatum of the asymptomatic subjects in the sample, the quantitative result information of Escherichia coli with pks+, and the scores of the living habits of various types after pre-screening, so as to obtain a trained prediction model for colorectal adenoma;

[0118] Specifically, when building the model, we already know the results of the colonoscopy report, that is, we know whether each asymptomatic subject in the sample has colorectal adenoma or is a healthy person. Our model incorporates the quantitative result information of Fusobacterium nucleatum of the asymptomatic subjects in the samples we collected, the quantitative result information of pks+ Escherichia coli, and the scores of various types of living habits after pre-screening, and trains a predictive model for whether the subject has colorectal adenoma.

[0119] Preferably, step S4 may include:

[0120] Step S41, taking the quantitative result information of Fusobacterium nucleatum of all asymptomatic subjects, the quantitative result information of Escherichia coli with pks+ and the scores of various types of living habits after pre-screening as a data set; dividing the data set into a training set, a validation set and a test set, wherein the data in the training set, the validation set and the test set do not overlap with each other, and each set in the training set, the validation set and the test set includes both healthy subjects and colorectal adenoma test patients;

[0121] Here, 60% of the data in the data set can be used as a training set, the remaining 20% ​​of the data can be used as a validation set, and the remaining 20% ​​of the data can be used as a test set; the training set includes healthy subjects and colorectal adenoma test patients; the validation set includes healthy subjects and colorectal adenoma test patients; the test set includes healthy subjects and colorectal adenoma test patients;

[0122] Step S42, using statistical software R (version 4.1.0) to establish a multivariate Logistic regression model, and training the multivariate Logistic regression model through the training set; verifying the trained Logistic regression model through the verification set to obtain a trained prediction model for colorectal adenoma;

[0123] Preferably, a multivariate logistic regression model can be established for the pre-screened influencing factors using multivariate logistic regression analysis in the cohort of the training set, and then redundant variables can be deleted by the backward elimination method to streamline the multivariate logistic regression model.

[0124] Step S43, evaluating the trained prediction model for colorectal adenoma using the test set to obtain a trained prediction model for colorectal adenoma.

[0125] Preferably, the trained multivariate logistic regression model, i.e., the trained prediction model for colorectal adenoma, can be evaluated using the test set. The specific evaluation can be achieved using the pROC package of R, and the ggplot2 package can be used to plot the trained prediction model for colorectal adenoma.

[0126] Here, the performance of the colorectal adenoma prediction model that has been trained and verified can be evaluated in the test set: specifically, it is evaluated through ROC curve analysis. Since ROC curve analysis is considered to be one of the reliable and best methods for performance characterization, the ROC curve and the area under the ROC curve are widely used to evaluate the discriminatory power of the simulated screening model.

[0127] Preferably, the trained prediction model for colorectal adenoma retains 14 types of living habits and two types of intestinal bacteria, among which:

[0128] 14 types of lifestyle habits, including: age, gender, sedentary time, hyperlipidemia, family history, cancer history, processed meat consumption, vegetable and fruit consumption, grain consumption, sugary drink consumption, coffee consumption, smoking status, alcohol consumption, and aspirin use history;

[0129] Two intestinal bacteria, including: Fusobacterium nucleatum and pks+ Escherichia coli.

[0130] Preferably, a semi-quantitative method can be used to evaluate the living habits of each subject. For example, the score of each living habit can be divided into three levels: high, medium and low.

[0131] Preferably, the amount of vegetables and fruits consumed, the amount of grains consumed, and the history of aspirin use can be considered healthy living habits. The higher the scores of these three items, the lower the risk of colorectal adenoma calculated by the model; and vice versa.

[0132] Preferably, the trained colorectal adenoma prediction model is as follows:

[0133] Colorectal adenoma risk index = Y / (1+Y);

[0134] Where, Y = exp(-6.42 + 0.73*X1 + 0.61*X2 + 0.56*X3 + 0.52*X4 + 0.35*X5 + 0.32*X6 + 0.3*X7 + 0.28*X8 + 0.25*X9 + 0.22*X 10 +0.21*X11 +0.15*X 12 +0.08*X 13 -0.55*X 14 -0.82*X 15 –1.79*X 16 );

[0135] The specific variables and their scores are shown in Figure 3, where X1 is the score of Fusobacterium nucleatum, X2 is the score of cancer history, X3 is the score of sugary drink consumption, and X4 is the score of pks + coli score, X5 is the score of coffee consumption, X6 is the score of gender, X7 is the score of processed meat consumption, X8 is the score of alcohol consumption, X9 is the score of hyperlipidemia, X 10 is the score of smoking status, X 11 is the score of sedentary time, X 12 For the score of family history, X 13 is the age score, X 14 is the score of grain consumption, X 15 is the score of vegetable and fruit consumption, X 16 Score the history of aspirin use.

[0136] Here, as shown in Table 3, the trained model retains a total of 14 lifestyle habits and two types of intestinal bacteria information: age, gender, sedentary time, hyperlipidemia, family history, cancer history, processed meat consumption, vegetable and fruit consumption, grain consumption, sugary beverage consumption, coffee consumption, smoking status, drinking status, aspirin use history, Fusobacterium nucleatum abundance, and pks+ Escherichia coli abundance.

[0137] Table 3

[0138]

[0139]

[0140]

[0141] More preferably, the final colorectal adenoma risk index is: with 0.31 as the boundary, the colorectal adenoma population is ≥ 0.31, and the healthy population is < 0.31;

[0142] Preferably, after step S4, the method further includes:

[0143] Step S5, inputting the score of the living habits of the person to be tested and the quantitative result information of the target bacteria species into the trained rectal adenoma prediction model to obtain the prediction result of colorectal adenoma of the person to be tested.

[0144] The quantitative result information of the target bacteria species of the person to be tested can be obtained by referring to step S3.

[0145] This application collected and analyzed the living habits and intestinal flora information of more than 1,800 cases of colorectal adenoma and healthy controls, and found 14 colorectal adenoma-related living habits and two colorectal adenoma-related intestinal bacteria. These living habits and intestinal bacteria abundance information can be used as markers to jointly predict the risk of colorectal adenoma. It can provide a prediction with high accuracy, convenience, efficiency, and low cost, which is suitable for extensive screening in large asymptomatic populations. It can solve the problem of difficult colorectal adenoma screening.

[0146] In addition, the present application can further develop a non-invasive colorectal adenoma prediction model kit. The kit can include specific primers for the above two colorectal adenoma-related intestinal bacteria, related PCR detection reagents, and a small program that integrates lifestyle habits and intestinal bacteria information to predict the risk of colorectal adenoma. Compared with the two existing colorectal adenoma screening technologies:

[0147] ① This kit is more accurate than the fecal immunochemical test. In our early population application, the prediction accuracy of this kit was as high as 84%, which is much higher than the 62% prediction accuracy of the fecal immunochemical test. This kit can better and quickly screen out colorectal adenoma from the population;

[0148] ② Compared with colonoscopy, this kit is more convenient, faster and less expensive. It does not require the assistance of an experienced gastroenterologist and can be used for large-scale screening among the population.

[0149] The kit of the present application is easy to use, does not require invasive operations and the assistance of professional gastroenterologists, and can quickly predict the risk of colorectal adenoma in subjects in just one hour simply by collecting the subjects' living habits and stool samples for testing.

[0150] According to another aspect of the present invention, a colorectal adenoma prediction system is provided, comprising:

[0151] A model training module, used to obtain a colorectal adenoma prediction model using any of the methods described above;

[0152] The prediction module is used to input the scores of the living habits of the person to be tested and the quantitative result information of the target bacteria species into the trained rectal adenoma prediction model to obtain the prediction results of colorectal adenoma of the person to be tested.

[0153] According to another aspect of the present invention, a computer-readable storage medium is provided, on which computer-executable instructions are stored, wherein when the computer-executable instructions are executed by a processor, the processor is caused to: execute any of the methods described above.

[0154] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0155] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0156] Obviously, those skilled in the art can make various changes and modifications to the invention without departing from the spirit and scope of the invention. Thus, if these modifications and variations of the invention fall within the scope of the claims of the invention and their equivalents, the invention is also intended to include these modifications and variations.

Claims

1. A method for constructing a colorectal adenoma prediction model, characterized in that: include: Collect stool samples from asymptomatic subjects for colorectal adenoma screening and score various lifestyle habits; The asymptomatic subjects in the sample include: healthy subjects and patients with colorectal adenoma; Based on the scores of various types of living habits of the asymptomatic subjects in the sample, a univariate logistic regression analysis was performed on the various types of living habits of the subjects and colorectal adenoma to obtain the types of living habits after pre-screening; Amplifying and quantifying the gene fragments of the target bacterial species in the DNA of the feces of each sample asymptomatic subject to obtain quantitative result information of the target bacterial species of each sample asymptomatic subject; Based on the quantitative result information of the target bacterial species and the scores of the living habits of each species after pre-screening, a prediction model for colorectal adenoma is trained to obtain a trained prediction model for colorectal adenoma.

2. The method for constructing a colorectal adenoma prediction model according to claim 1, characterized in that: Univariate Logistic regression analysis was performed on the subjects' various lifestyles and colorectal adenoma to obtain the types of lifestyles after pre-screening, including: Univariate Logistic regression analysis was performed on the various types of living habits of the subjects and colorectal adenoma, and the types of living habits with an absolute value of β greater than 0.05 and a P value less than 0.25 were obtained as the types of living habits after pre-screening; The gene fragments of the target bacteria species in the DNA of the feces of each sample asymptomatic subject are amplified and quantified to obtain quantitative result information of the target bacteria species of each sample asymptomatic subject, including: The gene fragments of Fusobacterium nucleatum and pks+ Escherichia coli in the DNA of the feces of each sample asymptomatic subject are amplified and quantified to obtain the quantitative result information of Fusobacterium nucleatum and the quantitative result information of pks+ Escherichia coli of each sample asymptomatic subject.

3. The method for constructing a colorectal adenoma prediction model according to claim 2, characterized in that: The gene fragments of Fusobacterium nucleatum and pks+ Escherichia coli in the DNA of the feces of each asymptomatic subject are amplified and quantified to obtain the quantitative result information of Fusobacterium nucleatum and pks+ Escherichia coli of each asymptomatic subject, including: DNA was extracted from the feces of each asymptomatic subject using a fecal DNA extraction kit to obtain fecal genomic DNA; By using real-time quantitative fluorescence qPCR technology and selecting 16S rRNA as the internal reference of qPCR, Fusobacterium nucleatum and pks+ Escherichia coli were amplified and the abundance was quantified to obtain the quantitative result information of Fusobacterium nucleatum and the quantitative result information of pks+ Escherichia coli.

4. The method for constructing a colorectal adenoma prediction model according to claim 3, characterized in that: The stool DNA extraction kit is used to extract DNA from the stool of each asymptomatic subject to obtain stool genomic DNA, including: About 100% of the stool from each sample collected from asymptomatic subjects was placed in a bead tube; Add grinding solution to the grinding bead test tube and shake gently to fully mix the stool sample with the grinding beads and grinding solution in the grinding bead test tube; Obtain the C1 solution, and check in advance whether there is any crystal precipitation in the C1 solution. If there is precipitate at the bottom of the bottle of the C1 solution, place the C1 solution in a water bath and heat it until the precipitate is completely dissolved; Add the completely dissolved C1 solution to the grinding bead test tube, vortex lightly and mix thoroughly; Place the grinding bead test tube after the C1 solution is fully mixed in a water bath and heat it in a water bath; Fix the grinding bead test tube heated in a water bath horizontally on a vortex machine and oscillate horizontally at the maximum speed of the vortex machine. Pause once during the oscillation and let it stand to help break the cells in the feces. Place the horizontally shaken grinding bead test tube into a high-speed centrifuge for centrifugation until the supernatant is transparent, so as to reduce impurities in the final product; Avoid the sediment at the bottom of the grinding bead tube and use a pipette to transfer the first supernatant above the sediment to the prepared EP collection tube; Add C2 reagent to the first supernatant in the EP collection tube, shake and mix, and then place in a refrigerator for incubation to precipitate impurities such as polysaccharides, cell debris and proteins to obtain an incubated second supernatant; Place the EP collection tube containing the incubated second supernatant into a centrifuge; Avoid the flocculent precipitate at the bottom of the EP collection tube after centrifugation, and use a pipette to transfer the second supernatant to the first sterilized EP tube; Add C3 reagent to the second supernatant in the first sterilized EP tube, shake and mix, and then place in a refrigerator for incubation to continue to precipitate impurities such as polysaccharides, cell debris and proteins to obtain an incubated third supernatant; Place the first sterilized EP tube containing the incubated third supernatant into a high-speed centrifuge for centrifugation; Avoid the flocculent precipitate at the bottom of the first sterilized EP tube after centrifugation, and use a pipette to transfer the third supernatant to the second sterilized EP tube; The C4 reagent is shaken and mixed, and the C4 reagent is added to the third supernatant and lightly shaken to mix so as to bind and adsorb the DNA in the feces to obtain a mixed solution; Add the mixed solution to the Spin Filter centrifugal column each time, centrifuge it in a high-speed centrifuge, and discard the filtrate; repeat this step until all the mixed solution is loaded, so that the DNA is selectively adsorbed on the filter membrane of the Spin Filter centrifugal column; Add C5 reagent to the Spin Filter centrifugal column, let it stand and then centrifuge it in a high-speed centrifuge, and discard the filtrate in the collection tube of the Spin Filter centrifugal column; Transfer the Spin Filter column to a collection tube and centrifuge it again to remove the residual C5 reagent; Open the cap of the collection tube and let it dry; Transfer the dried Spin Filter centrifugal column to the sterilized third EP tube, add C6 reagent directly to the center of the white filter membrane of the Spin Filter centrifugal column, and incubate at room temperature to elute and adsorb the DNA on the centrifugal column filter membrane; The third EP tube was placed in a centrifuge for centrifugation, the Spin Filter centrifugal column was discarded, and purified fecal genomic DNA was obtained.

5. The method for constructing a colorectal adenoma prediction model according to claim 1, characterized in that: Based on the quantitative result information of the target bacteria and the scores of the living habits of each species after pre-screening, the prediction model of colorectal adenoma is trained to obtain a trained prediction model of colorectal adenoma, including: The quantitative result information of Fusobacterium nucleatum of all asymptomatic subjects, the quantitative result information of Escherichia coli with pks+ and the scores of various types of living habits after pre-screening are used as a data set; the data set is divided into a training set, a validation set and a test set, wherein the data in the training set, the validation set and the test set do not overlap with each other, and the training set, the validation set and the test set all include healthy subjects and colorectal adenoma test patients; A multivariate Logistic regression model is established using the statistical software R, and the multivariate Logistic regression model is trained using the training set; the trained Logistic regression model is verified using the verification set to obtain a trained prediction model for colorectal adenoma; The trained prediction model for colorectal adenoma is evaluated by using the test set to obtain a trained prediction model for colorectal adenoma.

6. The method for constructing a colorectal adenoma prediction model according to claim 5, characterized in that: A multivariate logistic regression model is established using statistical software R, and the multivariate logistic regression model is trained using the training set, including: In the cohort of the training set, multivariate logistic regression analysis was used to establish a multivariate logistic regression model for the pre-screened influencing factors, and then the redundant variables were deleted by the backward elimination method to streamline the multivariate logistic regression model; The trained prediction model for colorectal adenoma is evaluated by using the test set to obtain a trained prediction model for colorectal adenoma, including: The trained multivariate logistic regression model, i.e., the trained prediction model for colorectal adenoma, was evaluated using the test set, including using the pROC package of R and plotting using the ggplot2 package to obtain the trained prediction model for colorectal adenoma.

7. The method for constructing a colorectal adenoma prediction model according to claim 1, characterized in that: The trained prediction model for colorectal adenoma retains 14 types of lifestyle habits and two types of intestinal bacteria, among which: 14 types of lifestyle habits, including: age, gender, sedentary time, hyperlipidemia, family history, cancer history, processed meat consumption, vegetable and fruit consumption, grain consumption, sugary drink consumption, coffee consumption, smoking status, alcohol consumption, and aspirin use history; Two intestinal bacteria, including: Fusobacterium nucleatum and pks+ Escherichia coli.

8. The method for constructing a colorectal adenoma prediction model according to claim 1, characterized in that: The trained colorectal adenoma prediction model is as follows: Colorectal adenoma risk index = Y / (1+Y); where Y = exp(-6.42 + 0.73*X1 + 0.61*X2 + 0.56*X3 + 0.52*X4 + 0.35*X5 + 0.32*X6 + 0.3*X7 + 0.28*X8 + 0.25*X9 + 0.22*X 10 + 0.21*X 11 + 0.15*X 12 + 0.08*X 13 - 0.55*X 14 - 0.82*X 15 – 1.79*X 16 ); X1 is the score of Fusobacterium nucleatum, X2 is the score of cancer history, X3 is the score of sugary drink consumption, X4 is pks + coli score, X5 is the score of coffee consumption, X6 is the score of gender, X7 is the score of processed meat consumption, X8 is the score of alcohol consumption, X9 is the score of hyperlipidemia, X 10 is the score of smoking status, X 11 is the score of sedentary time, X 12 For the score of family history, X 13 is the age score, X 14 is the score of grain consumption, X 15 is the score of vegetable and fruit consumption, X 16 Score the history of aspirin use.

9. A colorectal adenoma prediction system, characterized in that: include: A model training module, used to obtain a colorectal adenoma prediction model using the method according to any one of claims 1 to 8; The prediction module is used to input the scores of the living habits of the person to be tested and the quantitative result information of the target bacteria species into the trained rectal adenoma prediction model to obtain the prediction results of colorectal adenoma of the person to be tested.

10. A computer-readable storage medium having computer-executable instructions stored thereon, wherein: When the computer executable instructions are executed by a processor, the processor is caused to: perform the method according to any one of claims 1 to 8.

Images