Screening method of metabolic marker for distinguishing non-colorectal cancer disease and colorectal cancer and application of screening method

Abstract

The invention discloses a screening method of metabolic markers for distinguishing non-colorectal cancer diseases and colorectal cancer and application of the screening method. After a urine sample is treated, high-flux gas chromatography-mass spectrometry (GC-MS) derivatization polar metabolite analysis and liquid chromatography-mass spectrometry (LC-MS) polar metabolite detection are respectively carried out. The method comprises the following steps: sequentially carrying out homogenization treatment on obtained metabonomics data, screening 14 metabolic markers from the metabonomics data by adopting a univariate and multivariate statistical analysis method, and selecting 10, 8 and 4 metabolic markers from the 14 metabolic markers, so that the method also has a relatively good distinguishing effect on non-colorectal cancer diseases and colorectal cancer diseases. The invention discloses a non-invasive, high-accuracy and non-invasive urine metabolite marker combination, which improves the diagnosis and evaluation capability of early patients with colorectal cancer so as to realize early discovery, early diagnosis and early intervention clinically, thereby reducing the morbidity and mortality of the patients with colorectal cancer.

Description

Technical Field

[0001] The present invention relates to the field of metabolic marker analysis and detection, and in particular to a screening method for metabolic markers for distinguishing non-colorectal cancer diseases from colorectal cancer and an application thereof. Background Art

[0002] Colorectal cancer (CRC) refers to cancer originating from the colorectal mucosal epithelium. It is also one of the most common malignant tumors in the world, with an incidence and mortality rate ranking third and second in the world, respectively. In my country, the number of cases and mortality rates of colorectal cancer rank second and fourth among all malignant tumors, respectively. According to the latest cancer data released by the National Cancer Center in 2022, there are about 1.9 million new cases of colorectal cancer and about 900,000 deaths each year in the world. Among them, the number of cases of colorectal cancer in China is about 520,000, accounting for 27.4% of the world; the number of deaths is about 240,000, accounting for 26.7% of the world. China has become the country with the most new cases and deaths of colorectal cancer in the world. This disease has seriously endangered the health of Chinese residents and has also brought heavy burdens to society and individuals. Due to the low rate of early diagnosis and treatment of colorectal cancer patients in my country, many patients are already in the late stage when they are discovered. Therefore, early diagnosis and early treatment are the key to improving the prognosis of colorectal cancer, reducing the burden of the disease, and increasing the survival of patients.

[0003] At present, common screening methods for colorectal cancer mainly include imaging examinations, fecal occult blood tests and carcinoembryonic antigen, as well as endoscopic biopsy and other examination methods. The former, as a non-invasive screening method, has low accuracy and specificity, and is not sensitive enough for the early diagnosis of colorectal cancer; and the latter, although endoscopic biopsy is the gold standard for the diagnosis of colorectal cancer, is invasive and will cause physical discomfort to patients, and has low patient acceptance and high costs. Endoscopic biopsy may cause other complications. Since the above methods all have certain limitations, they are not suitable for early diagnosis and screening of large populations. Therefore, the development of a non-invasive, non-invasive, and highly accurate early screening method is of great significance to improving the treatment effect of colorectal cancer and improving patient prognosis.

[0004] The occurrence and development of colorectal cancer is a complex process mediated by multiple genes and various signaling pathways. Generally, colorectal cancer evolves from benign lesions such as precancerous diseases (colorectal polyps, inflammatory bowel disease) and precancerous lesions (such as colorectal adenomas). Among them, colorectal polyps are divided into adenomatous polyps and non-adenomatous polyps according to their pathological properties. Adenomatous polyps are precancerous lesions of colorectal cancer, and their symptoms include abdominal pain, diarrhea, constipation, blood in the stool, and anemia. Some patients have no obvious discomfort clinically, and they are diagnosed with cancer at the first diagnosis; colorectal adenoma is the most common precancerous lesion of colorectal cancer, and the incidence rate increases with age, and it develops into cancer through the classic "adenoma-adenocarcinoma" pathway. Colorectal adenomas are divided into tubular adenomas, tubulovillous adenomas, and villous adenomas. Among them, villous adenomas have the highest cancerization rate and tubular adenomas have the lowest. Therefore, if a method can be established to diagnose benign colorectal diseases and colorectal cancer early and perform timely intervention and surgical treatment, it will not only curb the incidence of colorectal cancer but also improve the survival rate of patients.

[0005] Metabolomics is a technology that has emerged after genomics, transcriptomics and proteomics. It can reflect the physiological and pathological conditions of the body and the differences between individuals by qualitatively and quantitatively analyzing small molecule metabolites with a relative molecular weight of less than 1000 Da in biological samples (blood, urine, feces and tissues). Metabolites are downstream products of the genome, transcriptome and proteome, which are closer to the biological phenotype and can therefore depict the state of life in a more real-time manner. The levels of gene and protein expression can be amplified by metabolites, and the detection of metabolites is easier to find key markers that cause disease. During tumor development, changes in metabolites may occur earlier than the appearance of their specific morphology or characteristics. Therefore, metabolomics provides a new perspective that can reveal metabolic abnormalities during tumor occurrence and development. Previous studies on colorectal cancer have mainly focused on the comparison of healthy and colorectal cancer patients based on plasma metabolomics, while there have been few studies on the differences between non-colorectal cancer diseases and colorectal cancer patients. Compared with blood, urine samples are easier to obtain, and urine sample collection is convenient, fast, non-invasive and low-cost. Therefore, developing a method based on a combination of urine markers to distinguish non-colorectal cancer diseases from colorectal cancer will help to determine the patient's pathological status at an early stage and help patients choose the best treatment plan, thereby alleviating their physical and psychological suffering. Summary of the invention

[0006] This study aims to study the small molecule metabolites in urine samples of patients with non-colorectal cancer diseases and colorectal cancer based on non-targeted metabolomics technology, to find a set of disease-specific urine marker combinations to build a diagnostic model for non-colorectal cancer diseases and colorectal cancer, and to provide effective treatment strategies and guidance for the early diagnosis and early treatment of colorectal cancer in clinical practice.

[0007] In order to solve the above technical problems, the present invention adopts the following technical solutions: The present invention discloses a method for screening metabolic markers for distinguishing non-colorectal cancer diseases from colorectal cancer, comprising the following steps: 1) The urine sample was divided into two parts, one for polar metabolite derivatization and the other for polar metabolite extraction; 2) The urine samples treated with polar metabolites were subjected to high-throughput gas chromatography-mass spectrometry (GC-MS) derivatization polar metabolite analysis, and the urine samples extracted with polar metabolites were subjected to liquid chromatography-mass spectrometry (LC-MS) polar metabolite detection; 3) Preprocessing GC-MS mass spectrometry data and LC-MS mass spectrometry data respectively; 4) After parsing the raw data obtained in step 3), the spectral information of the compound parent ion and secondary fragment ion is obtained, and compared with the commonly used metabolite database to obtain metabolomics data; 5) Perform log-transformation and normalization autoencoder (NormAE) homogenization on the metabolomics data; 6) Univariate and multivariate statistical analysis methods were used to screen out metabolites with significant differences in the urine of patients with non-colorectal cancer diseases and colorectal cancer diseases.

[0008] Preferably, the screening method is for non-disease diagnosis purposes.

[0009] Preferably, the polar metabolite derivatization treatment includes: thawing the urine sample, centrifuging, adding urease, incubating at 37°C for 15-45min, cooling on ice, adding cold methanol and vortexing, ultrasonicating in an ice bath, centrifuging to obtain the supernatant, vacuum concentrating and drying, oximating with a methoxyammonium salt pyridine solution, and adding MSTFA (N-methyltrimethylsilyltrifluoroacetamide) for derivatization; the polar metabolite extraction includes: thawing the urine sample, vortexing, adding a mixture of methyl tert-butyl ether and methanol, vortexing, ultrasonicating in an ice bath, adding a mixture of methanol and water, vortexing, centrifuging, taking the aqueous phase from the lower layer into a centrifuge tube, adding methanol thereto, vortexing, incubating at -20°C for 1-2 hours, centrifuging at 4°C for 10-15 min, taking the supernatant into an LC tube, and spinning dry in a vacuum concentrator; adding LC-MS grade water for reconstitution, ultrasonic-assisted treatment, and then centrifuging at room temperature to obtain.

[0010] Preferably, the GC-MS detection conditions are: high-purity helium is used as the carrier gas with a flow rate of 1 mL / min; the gas chromatography temperature program is: initial temperature 50°C, maintained for 2 min, continuously heated to 330°C at 15°C / min, maintained for 2 min; the main parameters of the injection port are: 280°C, split injection (10:1); the main parameters of the mass spectrometer are: electron ionization source (EI), temperature 250°C, electron energy 70 eV; scanning range 50-500 amu, detection voltage 1.2 kV.

[0011] Preferably, the LC-MS detection conditions are as follows: mobile phase A is an aqueous solution containing 0.1% formic acid; mobile phase B is an acetonitrile solution containing 0.1% formic acid, and the flow rate is 0.4 mL / min; the separation elution gradient is as follows: 0 min, 1% mobile phase B; 13 min, 70% mobile phase B, 13.01 min, 99% mobile phase B; 18 min, 99% mobile phase B; 18.01 min, 1% mobile phase B; 22 min, 1% mobile phase B; the sample injection volume is 3 μL, and the autosampler temperature is 10°C; data is collected using positive and negative ionization modes, and the specific parameters are as follows: AGC is 3E+6, Maximum IT is 200 milliseconds, the first full scan (Full scan) resolution is 70,000 FWHM (@200 m / z), and in the second scan mode (FullMS / dd-MS2), the resolution of the secondary mass spectrometry is 1.75 10,000, the quadrupole window was 1.5 m / z, the AGC was 1E+5, the maximum ion injection time was 50 ms, the HCD relative collision energy was 30 eV, the ion spray voltage was 3,500 V in positive mode and 3,000 V in negative mode, the nebulizer was 20 psi, the sheath gas temperature was 400 °C, and the sheath gas flow rate was 10 L / min.

[0012] Preferably, the univariate statistical analysis method is univariate ROC curve analysis.

[0013] Preferably, the multivariate statistical analysis method is based on multivariate statistical analysis in an OPLS-DA model.

[0014] Preferably, the urine samples are urine samples from patients with non-colorectal cancer diseases and patients with colorectal cancer.

[0015] The invention discloses metabolic markers for distinguishing non-colorectal cancer diseases from colorectal cancer and applications thereof.

[0016] The present invention discloses markers obtained by screening the screening method, and the markers include: 3-hydroxyhippuric acid, p-hydroxyphenylpropionic acid, phenylacetylglutamine, and N-cinnamoylglycine.

[0017] Preferably, the markers include: 3-hydroxyhippuric acid, 3-(3-hydroxyphenyl)-3-hydroxypropionic acid, hippuric acid, 4-hydroxyhippuric acid, p-hydroxyphenylpropionic acid, 3-(3-hydroxyphenyl)propionic acid, phenylacetylglutamine and N-cinnamoylglycine.

[0018] Preferably, the markers further include: p-cresol glucuronide and indoxyl sulfate.

[0019] Preferably, the markers further include: 2-amino-2-methyl-1,3-propanediol, p-cresol, L-glutamyl-L-leucine, and L-methionyl-L-aspartic acid.

[0020] The invention discloses the use of the marker in preparing a reagent for distinguishing non-colorectal cancer diseases from colorectal cancer.

[0021] The invention discloses a kit for distinguishing non-colorectal cancer diseases from colorectal cancer, and the kit comprises the marker.

[0022] In response to the problems in the current diagnosis and screening of colorectal cancer, the present invention has developed a non-invasive, highly accurate, and non-invasive combination of urine metabolite markers. By analyzing the metabolites with significant differences in urine samples of patients with non-colorectal cancer diseases and colorectal cancer patients, a group of biomarkers was screened out to construct a diagnostic method for distinguishing non-colorectal cancer diseases from colorectal cancer. This study aims to improve the diagnostic evaluation capabilities of early colorectal cancer patients, so as to achieve early detection, early diagnosis, and early intervention in the clinic, thereby reducing the morbidity and mortality of colorectal cancer patients.

[0023] Compared with the prior art, the present invention establishes a convenient, fast and non-invasive sampling method, which is suitable for distinguishing non-colorectal cancer diseases from colorectal cancer patients. A set of biomarker combinations related to non-colorectal cancer diseases / colorectal cancer is provided by non-targeted metabolomics technology. Through external validation data sets, it is confirmed that the model has excellent performance in predictive stability and can effectively distinguish non-colorectal cancer diseases from colorectal cancer patients. The present invention aims to establish a predictive model for the early diagnosis of colorectal cancer, which provides important basis and guidance for the clinical diagnosis of early colorectal cancer. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 . Multivariate ROC curve analysis of the modeling group to distinguish between the non-colorectal cancer disease group and the colorectal cancer group based on the combination of 14 metabolite markers.

[0025] Figure 2 . Multivariate ROC curve analysis of the validation set to distinguish between the non-colorectal cancer disease group and the colorectal cancer group based on a combination of 14 metabolite markers.

[0026] Figure 3 . Multivariate ROC curve analysis of the validation set to distinguish between the non-colorectal cancer disease group and the colorectal cancer group based on a combination of 10 metabolite markers.

[0027] Figure 4 . Multivariate ROC curve analysis of the validation group to distinguish between the non-colorectal cancer disease group and the colorectal cancer group based on the combination of 8 metabolite markers.

[0028] Figure 5 . Multivariate ROC curve analysis of the validation group to distinguish between the non-colorectal cancer disease group and the colorectal cancer group based on the combination of four metabolite markers. DETAILED DESCRIPTION

[0029] The specific implementation of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the present invention, and is not used to limit the present invention.

[0030] Example 1 Detection object and sample collection 1. Subject information: 1) The inclusion and exclusion criteria for patients with non-colorectal cancer diseases are as follows: Inclusion criteria: (1) male or female aged ≥18 years; (2) subjects read and fully understood the patient instructions and signed the informed consent form; (3) non-colorectal cancer disease group: confirmed by biopsy / postoperative pathology or clinically diagnosed as benign colorectal disease through comprehensive evaluation by clinicians, including but not limited to colorectal polyps, colorectal adenomas, etc.

[0031] Exclusion criteria: (1) pregnancy or lactation; (2) emergency or rescue; (3) history of blood transfusion within 7 days before sampling; (4) people who have received organ transplantation or non-autologous (allogeneic) bone marrow or stem cell transplantation; (5) history of malignant tumor or any anti-tumor treatment before sampling.

[0032] 2) The inclusion and exclusion criteria for patients with colorectal cancer are as follows: Inclusion criteria: (1) male or female aged ≥18 years; (2) subjects read and fully understood the patient instructions and signed the informed consent; (3) Colorectal cancer group: patients diagnosed with colorectal cancer by biopsy / postoperative pathology or through comprehensive evaluation by clinicians.

[0033] Exclusion criteria: (1) pregnancy or lactation; (2) emergency or requiring rescue; (3) history of blood transfusion within 7 days before sampling; (4) people who have received organ transplantation or non-autologous (allogeneic) bone marrow or stem cell transplantation; (5) history of malignant tumor or any anti-tumor treatment before sampling; (6) concurrent multiple primary malignant tumors.

[0034] 2. Sample collection and separation: Urine samples were collected in the early morning on an empty stomach, and the mid-morning urine supernatant samples were collected. Centrifugation conditions: 1600 g, 10 min, 4°C. Dispense into centrifuge tubes and store in liquid nitrogen tanks or -80°C refrigerators.

[0035] 3. Clinical information collection: Information collection must comply with the principle of informed consent and protect the privacy of the subjects. The collected data include but are not limited to: demographics, lifestyle, medical history, biochemical results, molecular level detection, imaging examination results, pathological examination results, cancer staging, surgery / chemoradiotherapy, follow-up data, etc.

[0036] Example 2 Urine sample extraction and metabolite detection 1. Urine sample extraction: (1) Derivatization of polar metabolites: Thawed urine samples were centrifuged at 4°C for 1 min, 40 µL of supernatant was transferred to a GC extraction tube, 50 µL of urease (1.5 mg / mL) was added, and then the samples were mixed in a constant temperature mixer. The samples were incubated at 37°C and 500 rpm for 30 min. After cooling on ice for 1 min, 500 µL of cold methanol was added and vortexed for 10 s, and then ultrasonicated on ice for 10 min. The samples were centrifuged at 12700 rpm at 4°C for 5 min, and 120 µL of supernatant was transferred to the corresponding GC tube, and then dried in a vacuum concentrator (Speed-Vac). The dried GC tube samples were oximated with methoxyammonium salt pyridine solution, and then derivatized with MSTFA for GC-MS derivatization analysis of polar metabolites. (2) Polar metabolite extraction: Thawed urine samples were vortexed for 30 seconds. Then, 40 μL of the sample was transferred to an LC extraction tube, 700 μL of a mixture of methyl tert-butyl ether and methanol (methyl tert-butyl ether: methanol = 3:1, v / v) was added, vortexed, and ultrasonicated on ice for 15 min; 350 μL of a mixture of methanol and water (methanol: water = 3:1, v / v) was added, vortexed, and centrifuged at 12700 rpm for 5 min. Take 300 μL of the aqueous phase from the lower layer into a centrifuge tube, add 900 μL of methanol, vortex to mix, and incubate at -20°C for 1 hour to precipitate the protein; then, centrifuge at 12700 rpm for 10 min at 4°C, take 1000 μL of the supernatant and transfer it to the corresponding LC tube, and then spin dry it in a vacuum concentrator (Speed-Vac); add 200 μL of water (LC-MS grade) to the dried LC tube sample for re-dissolution, ultrasonic-assisted treatment for 5 min, and then centrifuge the centrifuge tube at room temperature for 5 min (12000 rpm). Transfer the supernatant to a 2 mL glass injection bottle for LC-MS polar metabolite detection; 2. High-throughput gas chromatography-mass spectrometry (GC-MS) detection: - GC-MS derivatization polar substance analysis: After derivatization, the samples were injected into a gas chromatography system (7890B, Agilent) equipped with an Rxi®-5Sil MS column (Restek 30m * 0.25mm*0.25μm) and a time-of-flight mass spectrometry system (PEGASUS-BT, Leco) for detection. High-purity helium was used as the carrier gas with a flow rate of 1mL / min. The gas chromatography temperature program was: initial temperature 50℃, maintained for 2 min, and then continuously heated to 330℃ at 15℃ / min and maintained for 2 min. The main parameters of the injection port were: 280℃, split injection (10:1). The main parameters of the mass spectrometry were: electron ionization source (EI), temperature 250℃, electron energy 70eV; scanning range 50-500 amu, detection voltage 1.2 kV.

[0037] 3. UHPLC-MS detection: (1) LC-MS polar metabolite analysis: Polar metabolite analysis used a Waters ACQUTTYUPLC ® HSST3 column (1.8 µm 2.1*100 mm) for small molecule separation at a column temperature of 40 °C. The liquid chromatography and mass spectrometry used an ACQUITY UPLC I-Class liquid chromatography system (Waters) and a Q-Exactive mass spectrometry system (Thermo Fisher Scientific), respectively. The mobile phase composition was as follows: mobile phase A was an aqueous solution containing 0.1% formic acid; mobile phase B was an acetonitrile solution containing 0.1% formic acid, with a flow rate of 0.4 mL / min. The separation elution gradient was as follows: 0 min, 1% mobile phase B; 13 min, 70% mobile phase B, 13.01 min, 99% mobile phase B; 18 min, 99% mobile phase B; 18.01 min, 1% mobile phase B; 22 min, 1% mobile phase B. The sample injection volume was 3 μL, and the autosampler temperature was 10°C; (2) LC-MS mass spectrometry parameters: Full scan and data-dependent acquisition (DDA) were used to obtain primary (MS1) and secondary mass spectrometry (MS2) spectra for polar metabolites. The full scan mass spectrum range was 100–1500 Da. The secondary scan mode (Full MS / dd-MS2) scan range was 100−310, 300-710, and 700−1500 Da. The MS instrument used an Orbitrap high-resolution mass spectrometer (Thermo Fisher) equipped with an electrospray ionization source (ESI), and data were collected in positive and negative ionization modes. The specific parameters are as follows: AGC is 3E+6, Maximum IT is 200 ms, the first-level full scan (Full scan) resolution is 70,000 FWHM (@200 m / z), in the second-level scan mode (Full MS / dd-MS2), the resolution of the secondary mass spectrometer is 17,500, the quadrupole window is 1.5 m / z, the AGC is 1E+5, the maximum ion injection time is 50 ms, the HCD relative collision energy is 30 eV, the ion spray voltage is 3,500 V in positive mode and 3,000 V in negative mode, the nebulizer is 20 psi, the sheath gas temperature is 400°C, and the sheath gas flow rate is 10 L / min.

[0038] Example 3 Metabolomics data preprocessing and metabolite identification 1. Metabolomics data preprocessing: (1) GC-MS mass spectrometry data preprocessing: Leco ChromaTOF (version 5.4) was used to obtain raw data from GC-MS. Peak extraction and peak alignment were performed using Target Search in the Bioconductor software package in R language (version 4.0.3). The retention time of fatty acid methyl esters (FAME) was converted into retention index (RI) and annotated using the Fiehn database. The final identification of metabolites was confirmed by manual comparison of chromatograms.

[0039] (2) LC-MS mass spectrometry data preprocessing: The metabolomics data preprocessing steps include peak extraction, peak alignment, baseline correction, peak filtering, isotope peak removal, missing value filling, etc. First, use professional software to extract detection peaks and align peaks from the mass spectrometry raw data, then apply baseline correction to remove noise and retain the original signal peak; the raw data is converted into central discrete data, and at the same time, the isotope peaks in the mass spectrometry data are removed; then, the peaks in a single sample are aligned with the retention times of the peaks in the chromatograms of other samples, the isotope peaks and low-intensity adducts in the same analyte are removed, and the missing values ​​of characteristic peaks > 50% are eliminated; the random forest chain equation (MICE Forest) is used to fill the missing values ​​to obtain the final mass spectrometry matrix data. In order to reduce the batch effect between samples and the difference in metabolite concentrations between samples and make the data distribution more symmetrical, logarithmic transformation (Log-transformation) and normalization autoencoder (NormalizationAutoencoder, NormAE) are used for homogenization processing; 2. Identification of metabolites: According to the original data, the spectral information of the compound parent ion and secondary fragment ion is obtained after the software is parsed, such as the mass-to-charge ratio (m / z) of the primary mass spectrum and the fragments of the secondary ion. The metabolites are qualitatively identified by matching the spectral information of the primary and secondary metabolites in the database; commonly used metabolite databases include HMDB (www.hmdb.ca), PubChem (https: / / pubchem.ncbi.nlm.nih.gov), MassBank (http: / / www.massbank.jp), and MassBank of North America (https: / / massbank.us); based on the metabolites identified in the relevant databases, the metabolites are finally verified according to the retention time, MS1, and MS2 mass spectrometry information when the standard is separated under the same chromatographic column and mass spectrometry conditions. The standard for metabolite identification is that the retention time is within 0.1min difference and the molecular weight error of the metabolite is less than 10ppm; Example 4 Diagnostic method for distinguishing non-colorectal cancer diseases from colorectal cancer based on urine marker combination 1. Subject information: A total of 350 urine samples were collected from two medical centers. Among them, there were 160 patients with non-colorectal cancer diseases and 190 patients with colorectal cancer. The modeling group samples and the validation group samples are different samples. The specific distribution is as follows: there are 263 subjects in the modeling group, including 120 patients with non-colorectal cancer diseases and 143 patients with colorectal cancer; there are 87 subjects in the validation group, including 40 patients with non-colorectal cancer diseases and 47 patients with colorectal cancer (Table 1).

[0040] Table 1. Sample information

[0041] 2. Biomarker combination screening: (1) Data standardization: The metabolomics data were subjected to log-transformation and NormAE normalization processing in sequence, aiming to scale the raw data to a specific range, eliminate the order of magnitude differences between different samples and the batch effects between different samples, and obtain standardized data.

[0042] (2) Screening of urine biomarker combinations: Univariate and multivariate statistical analysis methods were used to screen out metabolites with significant differences in urine between patients with non-colorectal cancer and colorectal cancer. Specifically, potential biomarkers were screened using a combination of univariate ROC curve analysis and the VIP value in the orthogonal partial least squares discriminant analysis (OPLS-DA) model.

[0043] 1) Univariate ROC analysis is a method to examine the impact of a single metabolic molecule on the performance of a classification model, and to explore the evaluation of a single feature on the diagnostic performance of the ROC model. Univariate ROC curve analysis was performed based on non-colorectal cancer disease and colorectal cancer group samples in the modeling group to obtain the area under the ROC curve (AUC) values ​​of all single metabolic molecules, and the univariate ROC curve AUC value > 0.65 was used as a screening condition; 2) Based on multivariate statistical analysis in the OPLS-DA model, the VIP values ​​of all variables contributing to the model were obtained. The VIP value was used to measure the influence and explanatory power of the content of each metabolite component on the classification and discrimination of samples, and to assist in the screening of metabolic markers. Therefore, based on the modeling group data, an OPLS-DA model was constructed between the non-colorectal cancer disease and colorectal cancer groups to obtain the VIP value of each metabolite contributing to the model, and VIP>1.8 was used as the screening condition; 3) The intersection was taken based on the conditions of univariate AUC>0.65 and VIP>1.8. Finally, 14 metabolome features were screened out. The 14 marker compositions were 3-(3-hydroxyphenyl)-3-hydroxypropionic acid, 4-hydroxyhippuric acid, N-cinnamoylglycine, hippuric acid, p-cresol glucuronide, p-hydroxyphenylpropionic acid, 2-amino-2-methyl-1,3-propanediol, p-cresol, 3-(3-hydroxyphenyl)propionic acid, indoxyl sulfate, phenylacetylglutamine, L-glutamyl-L-leucine, L-methionyl-L-aspartic acid, and 3-hydroxyhippuric acid (Table 2). Subsequently, a machine learning method was used to construct a diagnostic model for non-colorectal cancer diseases and colorectal cancer.

[0044] Table 2. Combination of 14 urine markers to distinguish non-colorectal cancer disease group from colorectal cancer group Serial number Urine markers-English Urine Markers-Chinese HMDB ID AUC VIP 1 3-(3-Hydroxyphenyl)-3-hydroxypropionic acid 3-(3-Hydroxyphenyl)-3-hydroxypropionic acid HMDB0002643 0.766 2.113 2 4-Hydroxyhippuric acid 4-Hydroxyhippuric acid HMDB0013678 0.759 2.229 3 N-Cinnamoylglycine N-Cinnamyl Glycine HMDB0011621 0.738 1.964 4 Hippuric acid Hippuric acid HMDB0000714 0.762 2.178 5 p-Cresol glucuronide p-Cresol glucuronide HMDB0011686 0.732 2.195 6 3-(4-Hydroxyphenyl)propionic acid Para-hydroxyphenylpropionic acid HMDB0002199 0.757 2.222 7 2-Amino-2-methylpropane-1,3-diol 2-Amino-2-methyl-1,3-propanediol HMDB0244975 0.674 1.887 8 p-Cresol p-cresol HMDB0001858 0.691 1.829 9 3-(3-Hydroxyphenyl)propionic acid 3-(3-Hydroxyphenyl)propionic acid HMDB0000375 0.744 2.139 10 Indoxyl sulfate Indoxyl sulfate HMDB0000682 0.712 2.126 11 Phenylacetylglutamine Phenylacetylglutamine HMDB0006344 0.743 2.297 12 L-Glutamyl-L-leucine L-Glutamyl-L-Leucine HMDB0028823 0.689 1.801 13 L-Methionyl-L-aspartic acid L-Methionyl-L-Aspartic Acid HMDB0028969 0.692 1.812 14 3-Hydroxyhippuric acid 3-Hydroxyhippuric acid HMDB0006116 0.779 2.212 3. Build a diagnostic model to distinguish non-colorectal cancer diseases from colorectal cancer In order to evaluate the diagnostic effect of the 14 selected metabolic marker combinations in distinguishing non-colorectal cancer diseases from colorectal cancer, a multivariate ROC analysis was performed based on the above 14 urine marker combinations. Specifically, 3 / 4 samples were randomly selected from the modeling group sample data as training sets (training), and the remaining 1 / 4 samples were used as test sets (test). Among them, the training set was used to build and train the machine learning classification model, and the remaining samples were used as test sets to verify the discriminative ability of the trained model. The SVM was used to randomly iterate 1000 times, and the diagnostic model for distinguishing non-colorectal cancer diseases from colorectal cancer was constructed by statistically calculating the average value of the final model accuracy. The evaluation and verification indicators of the model are sensitivity, specificity, accuracy, precision and area under the curve (AUC).

[0045] The sensitivity is: ; Specificity is: ; The accuracy is: ; The accuracy is: ; Generally speaking, the ROC curve is a method to study the relationship between model sensitivity and specificity, with sensitivity as the ordinate and 1-specificity as the abscissa. The evaluation basis is to compare the AUC value of the area under the curve. When AUC is greater than 0.5, the closer it is to 1, the better the model performance and the better the judgment effect; if it is less than 0.5, it means that the performance of the model is close to random guessing, with poor discrimination ability and poor accuracy. In addition to common parameters such as the receiver operating curve (ROC) and the area under the curve (AUC), the ROC classification prediction model also includes sensitivity, specificity, accuracy, and precision.

[0046] in: TP (True positive): The number of samples that are actually positive examples but are correctly predicted as positive examples; TN (Ture Negative): The number of samples that are actually negative examples but are correctly predicted as negative examples; FP (False Positive): The number of samples that are actually negative examples but are mistakenly predicted as positive examples; FN (False Negative): The number of samples that are actually positive examples but are mistakenly predicted as negative examples; The results showed that AUC = 0.901 (specificity: 90.0%, sensitivity: 75.0%, precision: 90.0%, accuracy: 81.8%), indicating that the diagnostic model for distinguishing non-colorectal cancer diseases from colorectal cancer based on the combination of 14 metabolic markers has a high diagnostic ability. Figure 1 .

[0047] 4. External Validation of the Diagnostic Model for Non-Colorectal Cancer Diseases and Colorectal Cancer In order to further verify the predictive effect of the non-colorectal cancer disease and colorectal cancer diagnostic model constructed based on the 14 metabolite markers, the validation group data set was used as an unknown sample and put into the non-colorectal cancer disease and colorectal cancer diagnostic model constructed using the modeling group to evaluate the independent validation effect of the model on the unknown data set other than the modeling group data set. As shown in Table 3, the evaluation model for predicting whether an individual has colorectal cancer (critical value = 0.5358) based on 14 biomarkers was detected in 38 of the 47 colorectal cancer patients, and 9 were classified as non-colorectal cancer disease groups; in 40 non-colorectal cancer disease patients, 36 were detected, and 4 were classified as colorectal cancer groups; based on these values, we constructed a confusion matrix. According to the sensitivity, specificity, precision and accuracy calculation method, the calculation results show that AUC = 0.894 (specificity: 90.0%, sensitivity: 80.9%, precision: 90.5%, accuracy: 85.1%). The test results are shown in Figure 2 .

[0048] This indicates that the non-colorectal cancer disease and colorectal cancer diagnostic model established by the screened 14 urine metabolic marker combination has good predictive ability.

[0049] Table 3. Confusion matrix of the diagnostic model for distinguishing non-colorectal cancer diseases from colorectal cancer Colorectal cancer Non-colorectal cancer diseases Colorectal cancer, N=47 38 (TP) 9 (FN) Non-colorectal cancer diseases, N=40 4 (FP) 36 (TN) 5. The diagnostic results of different metabolic marker combinations in distinguishing non-colorectal cancer diseases from colorectal cancer Based on metabolomics analysis, 14 urine metabolite marker combinations were screened and obtained. From these 14 metabolite marker combinations, 10, 8 and 4 metabolite markers were selected for combination, respectively. Using the SVM method, a diagnostic model for distinguishing non-colorectal cancer diseases from colorectal cancer was constructed in the modeling group samples. Then, the validation group samples, as unknown samples, were placed in the above diagnostic model to verify the effectiveness of different metabolite markers in distinguishing non-colorectal cancer diseases from colorectal cancer diseases.

[0050] - A multivariate ROC curve analysis was performed using a combination of 10 metabolite markers, 3-hydroxyhippuric acid, 3-(3-hydroxyphenyl)-3-hydroxypropionic acid, hippuric acid, 4-hydroxyhippuric acid, p-hydroxyphenylpropionic acid, 3-(3-hydroxyphenyl)propionic acid, phenylacetylglutamine, N-cinnamoylglycine, p-cresol glucuronide, and indoxyl sulfate. The external validation results showed that AUC=0.883 (specificity: 85.0%, sensitivity: 78.7%, precision: 86.1%, accuracy: 81.6%). This shows that the prediction model established based on the combination of 10 metabolite markers also has a good diagnostic effect. The test results are shown in Figure 3 .

[0051] - A multivariate ROC curve analysis was performed using a combination of 8 metabolite markers, 3-hydroxyhippuric acid, 3-(3-hydroxyphenyl)-3-hydroxypropionic acid, hippuric acid, 4-hydroxyhippuric acid, p-hydroxyphenylpropionic acid, 3-(3-hydroxyphenyl)propionic acid, phenylacetylglutamine, and N-cinnamoylglycine. The external validation results showed that AUC=0.888 (specificity: 85.0%, sensitivity: 74.5%, precision: 85.4%, accuracy: 79.3%). This shows that the prediction model established based on the combination of 8 metabolite markers also has a good diagnostic effect. The test results are shown in Figure 4 .

[0052] - A multivariate ROC curve analysis was performed using a combination of four metabolite markers, 3-hydroxyhippuric acid, p-hydroxyphenylpropionic acid, phenylacetylglutamine, and N-cinnamoylglycine. The external validation results showed that AUC = 0.832 (specificity: 80.0%, sensitivity: 72.3%, precision: 81.0%, accuracy: 75.9%). This indicates that the prediction model based on the combination of four metabolite markers also has a good diagnostic effect. The test results are shown in Figure 5 .

[0053] The present invention illustrates the process method of the present invention through the above-mentioned embodiments, but the present invention is not limited to the above-mentioned process steps, that is, it does not mean that the present invention must rely on the above-mentioned process steps to be implemented. Those skilled in the art should understand that any improvement of the present invention, equivalent replacement of the raw materials selected by the present invention, addition of auxiliary components, selection of specific methods, etc., all fall within the protection scope and disclosure scope of the present invention.

Claims

1. A method for screening metabolic markers for distinguishing non-colorectal cancer diseases from colorectal cancer, characterized in that: The steps include: 1) The urine sample was divided into two parts, one for polar metabolite derivatization and the other for polar metabolite extraction; 2) The urine samples treated with polar metabolites were subjected to high-throughput gas chromatography-mass spectrometry (GC-MS) derivatization polar metabolite analysis, and the urine samples with polar metabolites extracted were subjected to liquid chromatography-mass spectrometry (LC-MS) polar metabolite detection; 3) Preprocessing GC-MS mass spectrometry data and LC-MS mass spectrometry data respectively; 4) After parsing the raw data obtained in step 3), the spectral information of the compound parent ion and secondary fragment ion is obtained, and compared with the commonly used metabolite database to obtain metabolomics data; 5) Perform log-transformation and normalization autoencoder (NormAE) homogenization on the metabolomics data; 6) Univariate and multivariate statistical analysis methods were used to screen out metabolites with significant differences in the urine of patients with non-colorectal cancer diseases and colorectal cancer diseases.

2. The screening method according to claim 1, characterized in that The GC-MS detection conditions are as follows: high-purity helium is used as carrier gas with a flow rate of 1 mL / min; the gas chromatography temperature program is as follows: initial temperature 50°C, maintained for 2 min, continuously heated to 330°C at 15°C / min, maintained for 2 min; the main parameters of the injection port are: 280°C, split injection (10:1); the main parameters of the mass spectrometer are: electron ionization source (EI), temperature 250°C, electron energy 70 eV; scanning range 50-500 amu, detection voltage 1.2 kV.

3. The screening method according to claim 1, characterized in that The LC-MS detection conditions are as follows: mobile phase A is an aqueous solution containing 0.1% formic acid; mobile phase B is an acetonitrile solution containing 0.1% formic acid, and the flow rate is 0.4 mL / min; the separation elution gradient is as follows: 0 min, 1% mobile phase B; 13 min, 70% mobile phase B, 13.01 min, 99% mobile phase B; 18 min, 99% mobile phase B; 18.01 min, 1% mobile phase B; 22 min, 1% mobile phase B; sample injection volume was 3 μL, and the autosampler temperature was 10°C. Data were collected in both positive and negative ionization modes with the following parameters: automatic gain control (AGC) was 3E+6, Maximum IT was 200 ms, and the first-level full scan resolution was 70,000 FWHM (@200 m / z). In the second-level scan mode (Full MS / dd-MS2), the resolution of the secondary mass spectrometer was 17,500, the quadrupole window was 1.5 m / z, the AGC was 1E+5, the maximum ion injection time was 50 ms, the relative collision energy (HCD) was 30 eV, the ion spray voltage was 3,500 V in positive mode and 3,000 V in negative mode, the nebulizer was 20 psi, the sheath gas temperature was 400°C, and the sheath gas flow rate was 10 L / min.

4. The screening method according to claim 1, characterized in that The univariate statistical analysis method is univariate ROC curve analysis; the multivariate statistical analysis method is multivariate statistical analysis based on OPLS-DA model.

5. The marker obtained by screening according to any one of claims 1 to 4, characterized in that: The markers include: 3-hydroxyhippuric acid, p-hydroxyphenylpropionic acid, phenylacetylglutamine, and N-cinnamoylglycine.

6. The marker obtained by screening according to any one of claims 1 to 4, characterized in that: The markers include: 3-hydroxyhippuric acid, 3-(3-hydroxyphenyl)-3-hydroxypropionic acid, hippuric acid, 4-hydroxyhippuric acid, p-hydroxyphenylpropionic acid, 3-(3-hydroxyphenyl)propionic acid, phenylacetylglutamine and N-cinnamoylglycine.

7. The marker according to claim 6, characterized in that The markers also include: p-cresol glucuronide and indoxyl sulfate.

8. The marker according to claim 7, characterized in that The markers also include: 2-amino-2-methyl-1,3-propanediol, p-cresol, L-glutamyl-L-leucine, and L-methionyl-L-aspartic acid.

9. Use of the marker according to any one of claims 5 to 8 in the preparation of a reagent for distinguishing non-colorectal cancer diseases from colorectal cancer.

10. A kit for distinguishing non-colorectal cancer diseases from colorectal cancer, characterized in that: The kit comprises the marker according to any one of claims 5 to 8.

Images