Metabolic marker combination for colorectal cancer t staging evaluation and detection and staging discrimination method thereof

Abstract

This invention provides a combination of metabolic biomarkers for T-staging assessment of colorectal cancer, comprising four groups of metabolite combinations specifically distinguishing different T stages of colorectal cancer, each group containing 10 endogenous small molecule metabolites. This invention also provides detection methods and applications for the above-mentioned metabolic biomarker combinations. Furthermore, this invention provides a method for T-staging colorectal cancer for non-diagnostic and therapeutic purposes, as well as a combination of metabolic biomarkers related to the evolution of T-staging and tumor invasion depth in colorectal cancer. The advantages of this invention are: by combining iEESI-MS detection with multivariate statistics and machine learning, specific metabolic biomarkers related to T-staging of colorectal cancer are screened, a T-staging discrimination model is constructed, and objective, accurate, and rapid assessment of T-staging of colorectal cancer is achieved, providing molecular support for preoperative individualized treatment.

Description

Technical Field

[0001] This invention relates to the fields of biomedicine and analytical detection technology, and in particular to a combination of metabolic biomarkers for T-staging assessment of colorectal cancer and its detection and staging method. Background Technology

[0002] Colorectal cancer is a common malignant tumor of the digestive tract in clinical practice. Exploring its progression mechanism and early diagnostic methods is a core research direction in the field of precision medicine. The T stage of colorectal cancer, which represents the depth of tumor invasion, is the "gold standard" for determining surgical approach and prognostic assessment. Early-stage (T1-T2 N0M0) patients can usually undergo local resection, while locally advanced (T3-T4 or N+) patients are recommended to undergo neoadjuvant chemoradiotherapy before surgery. The deeper the T stage, the more layers of the intestinal wall the tumor penetrates, the higher the risk of local recurrence, and the worse the patient's prognosis. However, current T staging assessment methods mainly rely on imaging examinations (CT, MRI) and postoperative pathological anatomy analysis, which can only reflect the depth of tumor invasion and cannot reveal the intrinsic "molecular driving force" that drives the tumor to gradually invade from T1 to T4. This blind spot seriously restricts the accuracy of preoperative individualized treatment decisions.

[0003] Abnormal tumor metabolism is one of the important markers of cancer. Metabolic molecules not only provide energy sources for tumor cells, but also directly participate in signal transduction and epigenetic regulation, thereby driving local tumor invasion. Therefore, systematically analyzing the dynamic changes in metabolic characteristics during the evolution of T staging in colorectal cancer, identifying metabolic molecular markers that can reflect tumor invasive potential, and revealing the molecular mechanisms that promote invasion are of great significance for achieving accurate preoperative T staging and personalized treatment.

[0004] In recent years, metabolomics research in colorectal cancer has made groundbreaking progress. Studies have moved beyond simply describing metabolite profiles to elucidating the functional mechanisms of metabolites. Numerous studies have used metabolomics technologies to screen for combinations of metabolites with diagnostic or prognostic value, and their clinical application potential has been confirmed in independent validation cohorts. Existing studies have systematically depicted the dynamic changes in metabolites during the progression from normal to adenoma to colorectal cancer, and directly demonstrated the functional significance of metabolic molecules driving colorectal cancer progression through the complete mechanism chain of receptor binding and signaling pathway activation. However, existing research still has significant limitations. It often focuses on single metabolites or single pathways, lacking a systematic and dynamic analysis of the metabolic characteristics of all stages from T1 to T4, failing to clarify the synergistic effects and dynamic changes of various differentially expressed metabolites, and unable to fully reveal the process by which metabolic reprogramming drives the continuous evolution of T stages.

[0005] The metabolic characteristics of tumors at different T stages are unique. Previous studies have found that T3 stage shows an increase in triglycerides and acetate compared to T1-T2 stages, while T4 stage is accompanied by a decrease in lipids, acetate, and succinate. This indicates that the evolution of tumor T stage is accompanied by metabolic phenotypic remodeling, suggesting that differentially metabolized molecules have the potential function of driving tumor invasion depth. However, this study only proposed metabolic differences in T stage, and failed to screen for combinations of T stage metabolic biomarkers with high specificity and sensitivity, nor did it establish a precise T stage stratification model based on metabolic characteristics. It also failed to clarify the dynamic changes and synergistic effects of key differentially metabolites such as triglycerides, acetate, and succinate in the evolution of T stage, and could not fully reveal the core process by which metabolic reprogramming drives the continuous evolution of T stage from low to high invasion. Overall, existing research mostly focuses on comparing tumors with normal tissues, or statically comparing metastatic and non-metastatic samples. Few studies focus on the dynamic changes in metabolism during the continuous evolution of local invasion of the primary lesion (T1-T4), lacking dynamic analysis of the continuous evolution of the depth of local invasion. Furthermore, current research on the metabolic characteristics of T stages is mostly descriptive, only reporting differences in metabolites, without developing specific combinations of metabolic markers that can be used for clinical staging, nor establishing standardized and scalable staging assessment methods.

[0006] Modern mass spectrometry technology has provided new insights into the study of metabolic characteristics in colorectal cancer. Traditional mass spectrometry techniques such as gas chromatography-mass spectrometry (GC-MS) and liquid chromatography-mass spectrometry (LC-MS) have played an important role in detecting related metabolites such as fatty acids, amino acids, and phosphatidylcholine. However, existing technologies still have many limitations. In terms of sample pretreatment, traditional mass spectrometry techniques are cumbersome, requiring multiple extraction and derivatization operations, which can easily lead to sample loss and introduce experimental errors, affecting the accuracy and reliability of detection results. At the same time, they are difficult to meet the needs of in-depth research on low-abundance metabolites. More importantly, traditional metabolomics studies often rely on tissue homogenization, which loses the spatial information of the sample and cannot accurately locate key areas of tumor invasion. The differentially expressed metabolic molecules in these key areas may be the molecular driving force that determines whether the tumor can break through the basement membrane and invade deeper tissues.

[0007] Therefore, there is an urgent need for an analytical method that can resolve the metabolic characteristics within tissues to meet the practical needs of research on the metabolic characteristics of T-stage colorectal cancer. Summary of the Invention

[0008] The technical problem to be solved by this invention is to provide a combination of metabolic biomarkers for T-staging assessment of colorectal cancer and its detection and staging method. It uses internal extraction electrospray ionization mass spectrometry (iEESI-MS) detection technology combined with multivariate statistical analysis and machine learning algorithms to analyze the metabolic profile of different T-staging of colorectal cancer, screen out specific metabolic biomarker combinations related to tumor invasion depth, establish a standardized iEESI-MS detection method and construct a T-staging model, so as to achieve objective, accurate and rapid assessment of T-staging of colorectal cancer, make up for the deficiencies of existing technologies, provide molecular support for preoperative individualized treatment, and fill the application gap of iEESI-MS in this field.

[0009] The present invention solves the above-mentioned technical problems by adopting the following technical solutions: A combination of metabolic biomarkers for assessing T-stage of colorectal cancer, comprising four groups of metabolite combinations specifically distinguishing different T-stages of colorectal cancer, each group containing 10 endogenous small molecule metabolites: The combination of metabolic markers used to distinguish normal tissue from Tis / T1 phase is: lysophosphatidylinositol 20:4, cyclic phosphatidylinositol phosphate 16:0, uric acid, palmitoleic acid, oxidized phosphatidylcholine O-32:2, inosine, sphingomyelin 42:2, sphingomyelin 42:3, acetylcholine, and lysophosphatidylcholine P-18:0. The combination of metabolic markers used to distinguish between Tis / T1 and T2 phases is as follows: phosphatidylcholine 36:0, acetylcholine, sphingomyelin 40:1, phosphatidylcholine 31:0 hydrogen adduct, phosphatidylcholine hydrogen adduct, phosphatidylcholine 30:0 sodium adduct, cyclic phospholysinic acid 16:0, lysophosphatidylinositol 20:4, taurine, and inosine; The combination of metabolic markers used to distinguish between T2 and T3 phases is as follows: oxidized phosphatidylcholine O-32:2, phosphatidylcholine hydrogen adduct, phosphatidylcholine potassium adduct, oxidized phosphatidylcholine O-38:5, phosphatidylcholine 36:0, sphingomyelin 42:3, sphingomyelin 42:2, sphingomyelin 40:1, phosphatidylcholine 30:0, and lysophosphatidylcholine 16:0. The combination of metabolic markers used to distinguish between T3 and T4 phases is: sphingomyelin 42:3, phosphatidylcholine 36:0, inosine, phosphatidylcholine 30:0, phosphatidylserine 36:1, phosphatidylcholine 33:0, phosphatidylcholine 31:0, cyclic phospholysinic acid 16:0, phosphatidylinositol 38:3, and cyclic phosphatidylinositol.

[0010] An iEESI-MS detection method for the above-mentioned combination of metabolic markers, wherein the method is a non-diagnostic and non-therapeutic in vitro tissue metabolic profiling detection method, comprising the following steps: (1) Sample processing: Collect ex vivo tissue samples related to colorectal cancer and corresponding ex vivo normal tissue samples adjacent to the cancer. After ex vivo, the samples are quickly frozen in liquid nitrogen and stored. Samples are taken directly during testing. (2) iEESI-MS detection: The iEESI-MS ion source and mass spectrometer were used together. The sampling needle was inserted into the ex vivo tissue sample, and the extraction solvent was injected to extract tissue metabolites. Metabolic spectrum data were collected simultaneously. (3) Data preprocessing and biomarker identification: Peak identification, peak alignment, missing value filtering and normalization are performed on the collected raw mass spectrometry data, and the identification of each metabolic molecule in the combination of metabolic biomarkers is completed by combining the metabolite database.

[0011] As one of the preferred embodiments of the present invention, in step (2), the extraction solvent is a mixed solution of methanol and water in a volume ratio of 95:5, which is injected into the tissue at a flow rate of 2 μL / min using a syringe pump.

[0012] As one of the preferred embodiments of the present invention, in step (2), the ionization voltage detected by iEESI-MS is ±4.5kV, the temperature of the ion transfer tube is 250℃, the injection volume is 0.5~1.0mg of tissue, and the tissue sample is a direct sample taken from fresh frozen tissue.

[0013] As one of the preferred embodiments of the present invention, in step (2), the iEESI-MS detection is performed in positive and negative ion mode, and the mass scan range is [missing information]. m / z 100–900, with a single sample testing time not exceeding 5 minutes.

[0014] The use of a combination of the above-mentioned metabolic markers in the preparation of reagents or kits for colorectal cancer research or T-staging analysis of ex vivo samples.

[0015] A method for T-staging of colorectal cancer for non-diagnostic and therapeutic purposes uses the above-mentioned combination of metabolic markers as staging features. The method uses the above-mentioned detection method to obtain the expression level data of the metabolic markers in the sample to be tested, and inputs the data into a pre-constructed machine learning staging model to distinguish different T stages of colorectal cancer.

[0016] As one of the preferred embodiments of the present invention, the machine learning algorithm is a support vector machine, and the staging discrimination model includes four sub-models that respectively distinguish between normal tissue and Tis / T1 stage, Tis / T1 stage and T2 stage, T2 stage and T3 stage, and T3 stage and T4 stage.

[0017] As one of the preferred embodiments of the present invention, the expression level of the metabolic markers fluctuates regularly with the evolution of colorectal cancer T stages from Tis / T1, T2, T3 to T4, and the staging discrimination model determines the local invasion depth and stages of the tumor based on this specific expression change trend.

[0018] A combination of metabolic markers associated with the progression of T stage and tumor invasion depth in colorectal cancer, wherein the metabolic markers include: inosine, sphingomyelin 36:1, cyclic inositol phosphate, cyclic lysophosphatidylcholine phosphate 16:0, sphingomyelin 42:2, phosphatidylserine 36:1, sphingomyelin 40:1, phosphatidylcholine 37:4, lysophosphatidylcholine P-18:0 / 0:0, lysophosphatidylinositol 20:4, phosphatidylcholine 29:0, phosphatidylglycerol 34:1, phosphatidylcholine 33:1, phosphatidylcholine 32:2, phosphatidylcholine 38:2, phosphatidylcholine 30:0, and phosphatidylglycerol 32:0.

[0019] The advantages of this invention compared to the prior art are: (1) This invention provides a four-group stratified core metabolic marker combination specifically for staging. Each group of markers has significant characteristic expression differences at the corresponding T stage, which can effectively distinguish normal tissue from Tis / T1, Tis / T1 from T2, T2 from T3, and T3 from T4 samples. It has high specificity and is highly consistent with clinical staging. Its dynamic change pattern is highly matched with the clinical evolution path of T1→T2→T3→T4, providing a stable and reliable molecular indicator system for refined metabolic analysis of T staging of colorectal cancer. (2) This invention also clarifies the combination of trend metabolic markers related to T stage evolution and tumor invasion depth, which can clearly reflect the tumor progression pattern from a metabolic perspective, fill the gap in the existing technology’s understanding of the metabolic mechanism of tumor progression, and provide important molecular evidence for elucidating the intrinsic metabolic characteristics of colorectal cancer invasion and development. (3) This invention uses iEESI-MS ionization mass spectrometry technology, which does not require tissue homogenization and can directly extract and analyze tissue metabolites from isolated tissues, preserving the metabolic information of the sample and laying the foundation for subsequent metabolic studies of the tumor-normal tissue junction area; at the same time, it significantly simplifies the operation process, eliminating the need for complex pretreatment such as grinding, extraction, and derivatization, and the detection time for a single sample is short and the analytical throughput is high, making it suitable for rapid detection of large batches of isolated samples; (4) This invention systematically optimizes key parameters such as extraction solvent system, ionization voltage, ion transmission tube temperature, scanning mode and scanning range to ensure stable and efficient detection of core metabolic markers. The detection signal is clear and reproducible. Only 0.5-1.0 mg of trace tissue is needed to complete the analysis, which is suitable for early small tumors or biopsy samples and overcomes the dependence of traditional metabolomics on large sample volumes. (5) The present invention constructs a machine learning staging discrimination model based on the combination of core biomarkers, which can realize objective, standardized and reproducible evaluation of T staging of colorectal cancer. The model has been verified by training set and independent test set, and the accuracy, sensitivity and specificity are excellent, which significantly improves the reliability and scientificity of staging analysis. (6) The overall technical solution of this invention is geared towards in vitro sample analysis and scientific research for non-diagnostic and therapeutic purposes. The application scenarios are clearly defined and standardized. It deeply integrates iEESI-MS technology with T staging assessment of colorectal cancer, expands the application scope of iEESI-MS in clinical detection and related research of gastrointestinal tumors, and can provide a reference technical solution for the screening and detection of staging markers for other tumors. It has good scientific research promotion value and application prospects. Attached Figure Description

[0020] Figure 1 This is an OPLS-DA model score map of the differential metabolic characteristics of colorectal cancer at different T stages in Example 1 (in the figure, A is the OPLS-DA score map of normal tissue and Tis / T1 stage; B is the OPLS-DA score map of Tis / T1 and T2 stages; C is the OPLS-DA score map of T2 / T3 stage; D is the OPLS-DA score map of T3 / T4 stage). Figure 2 This is a graph showing the OPLS-DA model parameters for screening the differential metabolic characteristics of different T stages of colorectal cancer in Example 1 (Figure A shows the OPLS-DA model parameters for normal tissue and Tis / T1 stage; Figure B shows the OPLS-DA model parameters for Tis / T1 and T2 stages; Figure C shows the OPLS-DA model parameters for T2 / T3 stage; Figure D shows the OPLS-DA model parameters for T3 / T4 stage). Figure 3 This is a STEM trend analysis and cluster heatmap of metabolic characteristics of colorectal cancer at different T stages in Example 1 (Figure A shows the trend of 8 metabolites in Module 1 changing with T stage; Figure B shows the trend of 10 metabolites in Module 2 changing with T stage; Figure C shows the cluster heatmap of 8 metabolites in Module 1; Figure D shows the cluster heatmap of 10 metabolites in Module 2). Figure 4 The ROC curves constructed based on the SVM algorithm in Example 2 are used to distinguish T-stage feature combinations (in the figure, Figure A is the ROC curve distinguishing normal tissue and Tis / T1 stage; Figure B is the ROC curve distinguishing Tis / T1 and T2 stage; Figure C is the ROC curve distinguishing T2 / T3 stage; Figure D is the ROC curve distinguishing T3 / T4 stage). Figure 5The figures are ROC curves of the four SVM discriminant models validated on the independent test set in Example 2 (Figure A is the ROC curve that distinguishes between normal tissue and Tis / T1 stage; Figure B is the ROC curve that distinguishes between Tis / T1 and T2 stage; Figure C is the ROC curve that distinguishes between T2 and T3 stage; Figure D is the ROC curve that distinguishes between T3 and T4 stage). Detailed Implementation

[0021] The embodiments of the present invention are described in detail below. These embodiments are implemented based on the technical solution of the present invention, and provide detailed implementation methods and specific operation processes. However, the scope of protection of the present invention is not limited to the following embodiments.

[0022] The relevant terms in this invention are defined as follows: T staging: refers to the depth of tumor invasion. According to the AJCC 8th edition TNM staging system, it is divided into Tis (carcinoma in situ), T1 (tumor invades the submucosa), T2 (tumor invades the muscularis propria), T3 (tumor penetrates the muscularis propria and enters the subserosal layer), and T4 (tumor penetrates the visceral peritoneum or directly invades other organs). iEESI-MS: Internal extractive electrosprayionization mass spectrometry is a direct mass spectrometry technique that allows for the direct analysis of tissue samples without sample pretreatment. Metabolic markers: These are endogenous small molecule metabolites that are significantly associated with T stage of colorectal cancer, including lipids, amino acids, and energy metabolism intermediates. Discriminant model: refers to a mathematical model based on machine learning algorithms that uses the expression level of a combination of metabolic biomarkers as a feature to determine the T stage of colorectal cancer; ROC curve: refers to the receiver operating characteristic curve, used to evaluate the discriminative power of a diagnostic model; AUC: Area under the ROC curve, used to measure the diagnostic accuracy of a model.

[0023] Meanwhile, the English abbreviations and Chinese names of all metabolic markers in this invention are expressed in accordance with the universal nomenclature of lipid metabolomics. The naming of compounds such as phospholipids, sphingolipids, and lysins follows the standards of the International Lipid Classification and Nomenclature Committee (ILCN). The English abbreviations and corresponding Chinese names are consistent with the conventional expressions in the field and the annotations of mainstream mass spectrometry databases, ensuring that the names of each metabolic compound are standardized, uniform, and unambiguous, which is convenient for those skilled in the art to understand and implement.

[0024] Unless otherwise specified, the reagents and experimental methods used in this invention are all conventional reagents and methods in the field and will not be described in detail here.

[0025] Example 1: Preliminary screening and trend analysis of differentially expressed metabolites in different T stages of colorectal cancer: 1. Experimental Sample Collection Tumor tissue samples were collected from surgically removed patients with pathologically confirmed colorectal cancer and classified into Tis, T1, T2, T3, and T4 stages, totaling 105 cases. Among them, there were 18 cases of Tis / T1 stage, 24 cases of T2 stage, 51 cases of T3 stage, and 12 cases of T4 stage. At the same time, adjacent normal tissue (≥5cm from the tumor edge) was collected from each patient as a control. All samples were from patients at the First Hospital of Jilin University who had not undergone preoperative radiotherapy or chemotherapy. The tissues were flash-frozen in liquid nitrogen within 30 minutes after excision and stored at -80℃ for later use.

[0026] Given that both Tis and T1 stages are considered early-stage colorectal cancer in clinical practice, and the sample size of T1 stage is limited, Tis and T1 stage samples were combined into an early infiltration group (Tis / T1) for metabolic characteristic difference analysis of staging evolution with T2, T3, and T4 stage samples. This grouping method conforms to the clinical staging evolution pattern and can ensure the statistical power of inter-group difference analysis.

[0027] 2. iEESI-MS detection (1) Take out the frozen tissue sample from the -80℃ freezer. No thawing is required. Use a sterile sampling needle to take 0.5-1.0 mg of fresh frozen tissue for later use.

[0028] (2) Connect the iEESI-MS ion source to the mass spectrometer and adjust the instrument parameters as follows: ionization voltage ±4.5kV, ion transmission tube temperature 250℃, detection mode positive and negative ion mode, mass scan range m / z 100-900; Insert the sampling needle into the prepared tissue sample, and inject an extraction solvent of methanol and water at a volume ratio of 95:5 at a flow rate of 2 μL / min using a syringe pump to extract tissue metabolites. Simultaneously collect raw metabolic spectrum data, and control the detection time of a single sample within 5 minutes.

[0029] The iEESI-MS ion source was a self-made laboratory source. Its working principle is as follows: a quartz capillary acts as a "channel" to the mass spectrometer, with one end connected to a micro-sampling needle and the other end inserted parallel to the tissue sample. An extraction solvent is injected into the tissue sample via a syringe pump to selectively extract chemicals and acquire mass spectrometry data. The mass spectrometer used is an Orbitrap Fusion. TM Tribrid TM mass spectrometer (Thermo Scientific, San Jose, CA, USA).

[0030] 3. Mass spectrometry data preprocessing Raw mass spectrometry data, including mass-to-charge ratio and ion intensity, were acquired using Xcalibur software. Peak identification and alignment were performed using MetaboAnalystR software. Specifically, the raw iEESI-MS mass spectrometry data were preprocessed using MetaboAnalystR with the following steps: peak integrity check, missing value handling, and standardization. Characteristic peaks with more than 50% missing values ​​were filtered out. Data normalization was performed using logarithmic transformation (base 10) and Pareto scaling to generate a metabolite-sample intensity matrix. Metabolites were then preliminarily identified using metabolite databases such as HMDB for subsequent analysis.

[0031] 4. Screening and trend analysis of differential metabolites (1) Multivariate statistical methods such as PCA and OPLS-DA were used to analyze the differences in metabolite profiles among different T stages. An OPLS-DA model was initially established, and the T-test was used to compare the differences in metabolite expression levels among colorectal cancer tissue samples in the Tis / T1, T2, T3, and T4 groups. Differential metabolites in different T stages of colorectal cancer were screened using VIP>1 and P<0.05 as criteria.

[0032] Figure 1 This is a score map of the OPLS-DA model showing the differences in metabolic characteristics at different T stages of colorectal cancer. Figure 2 OPLS-DA model parameter plot for screening differential metabolic characteristics at different T stages of colorectal cancer. Figure 1 It can be seen that there are significant differences in normal tissue, Tis / T1, T2, T3, and T4 stages. Figure 2 It can be seen that the preliminary model for distinguishing between normal tissue and Tis / T1 has a prediction accuracy of Q. 2 The explanatory power R of the model is 0.763. 2 Y is 0.955; the model prediction accuracy Q used to distinguish Tis / T1 and T2 is... 2 The explanatory power R of the model is 0.699. 2 Y is 0.982; the model prediction accuracy Q used to distinguish between T2 and T3 is... 2 The explanatory power R of the model is 0.518. 2 Y is 0.91; the model prediction accuracy Q used to distinguish between T3 and T4 is... 2 The explanatory power R of the model is 0.69. 2 Y is 0.951, and none of the above models are overfitting.

[0033] (2) Trend clustering analysis of differentially expressed metabolites was performed using the STEM algorithm. The samples were analyzed in the order of “N”-“T1”-“T2”-“T3”-“T4”. All data were filtered out by a mathematical model to remove data with insignificant differences in expression over time, resulting in 50 clustering modules. Based on the STEM clustering results (-1 indicates insignificance), modules with a p-value less than 0.05 after correction by the False Discovery Rate method were selected as significant modules. Two significant modules were selected, and trend charts and cluster heatmaps of the significant modules were plotted (metabolites showing continuous upregulation, continuous downregulation, or key stage transitional expression changes).

[0034] Figure 3 STEM trend analysis and cluster heatmap results of corresponding T-stage metabolic characteristics. Figure 3 It can be seen that inosine, SM 36:1; O2, Inositol cyclic phosphate, CPA (16:0), SM 42:2; O2, PS (36:1), SM 40:1; O2, PC (37:4), LysoPC (P-18:0 / 0:0), LysoPI (20:4), PC (29:0), PG (34:1), PC (33:1), PC (32:2), PC (38:2), PC (30:0), and PG (32:0) were significantly upregulated in T3 and T4 stages and were significantly correlated with the depth of local tumor invasion.

[0035] Example 2: Determination of the core metabolic biomarker combination for T staging of colorectal cancer and construction and validation of the discriminant model: 1. Screening and Combination of Core Biomarkers One hundred and five colorectal cancer tissue samples were divided into a training set (70 cases) and a validation set (35 cases) in a two-thirds ratio. Based on the differential metabolite analysis results (Example 1), support vector machines (SVM) were used in MetaboAnalyst 6.0 software for feature selection and combination to identify the core metabolites that contributed most to the T-staging of colorectal cancer. A combination of metabolic biomarkers for colorectal cancer T-staging assessment was constructed, which can specifically distinguish between Tis / T1, T2, T3, and T4 stages of colorectal cancer. The classification method was linear support vector machine, and the feature selection method was the built-in support vector machine in MetaboAnalyst 6.0. The remaining one-third of the samples served as the validation set (35 cases) for validating the feature combination.

[0036] The ROC results for each model are as follows: Figure 4As shown in the figure. The results showed that the four SVM discriminant models constructed based on 10 metabolic biomarkers were used to distinguish normal tissue from Tis / T1. The metabolic biomarker combinations used to distinguish normal tissue from Tis / T1 were LysoPI (20:4), CPA (16:0), Uricacid, Palmitoleic acid, PC (O-32:2), Inosine, SM 42:2; O2, SM 42:3; O2, Acetylcholine, and LysoPC (P-18:0 / 0:0). The AUC value reached 0.984, the sensitivity was 77.27%, and the specificity was 99.56%. The metabolic biomarker combinations used to distinguish Tis / T1 from T2 were PC (36:0), Acetylcholine, SM 40:1; O2, PC (31:0)+H + Phosphorylcholine + H + PC(30:0)+Na + The combination of CPA (16:0), LysoPI (20:4), Taurine, and Inosine showed an AUC of 0.844, a sensitivity of 77.78%, and a specificity of 83.33%. The metabolic marker combination distinguishing between T2 and T3 was PC (O-32:2), Phosphorylcholine+H... + Phosphorylcholine + K + The combination of O2, PC (O-38:5), PC (36:0), SM 42:3; O2, SM 42:2; O2, SM 40:1; O2, PC (30:0) and LysoPC (16:0) had an AUC of 0.816, a sensitivity of 97.54%, and a specificity of 38.18%. The combination of metabolic markers used to distinguish between T3 and T4 was SM 42:3; O2, PC (36:0), Inosine, PC (30:0), PS (36:1), PC (33:0), PC (31:0), CPA (16:0), PI (38:3) and Inositolcyclic phosphate, with an AUC of 0.81, a sensitivity of 23.26%, and a specificity of 98.36%.

[0037] 2. Model Evaluation Tissue samples from 80 newly diagnosed colorectal cancer patients were included as a new test set. Among them, there were 5 patients with Tis / T1 stage, 9 patients with T2 stage, 57 patients with T3 stage, and 11 patients with T4 stage. All of them were patients from Jilin University Hospital who had not undergone preoperative radiotherapy and chemotherapy between 2019 and 2022. The samples were independent of the training set / validation set and there were no statistically significant differences in baseline characteristics (P>0.05).

[0038] For this independent test set, metabolic profile detection was performed strictly according to the iEESI-MS detection procedure and sample processing method in Example 1, and data of each metabolite in the above four core metabolic biomarker combinations were obtained respectively. Subsequently, the corresponding metabolite data of the independent test set were substituted into the four SVM staging discriminant sub-models (normal tissue vs Tis / T1, Tis / T1 vs T2, T2 vs T3, T3 vs T4) constructed from the training set. The model automatically output the staging probability of the sample, plotted the ROC curve, and calculated the AUC value, sensitivity and specificity to complete the independent verification of the metabolic biomarker combination and discriminant model of the present invention.

[0039] The results are as follows Figure 5 As shown in the results, the constructed model for distinguishing normal tissue from Tis / T1 achieved an AUC of 0.986, sensitivity of 98.63%, and specificity of 88.89% in independent test set validation; the constructed model for distinguishing Tis / T1 and T2 achieved an AUC of 0.784, sensitivity of 87.88%, and specificity of 60.0% in independent test set validation; the constructed model for distinguishing T2 and T3 achieved an AUC of 0.872, sensitivity of 92.59%, and specificity of 55.0% in independent test set validation; and the constructed model for distinguishing T3 and T4 achieved an AUC of 0.930, sensitivity of 70.0%, and specificity of 90.0% in independent test set validation.

[0040] Therefore, the four sets of core metabolic biomarkers for colorectal cancer T staging constructed in this invention and the corresponding SVM discriminant models all exhibited good discriminant efficacy in the independent sample test set. The overall model is stable and reliable, and can effectively achieve objective and accurate assessment of different T stages of colorectal cancer. In summary, this invention utilizes iEESI-MS mass spectrometry detection technology combined with multivariate statistical analysis, STEM trend analysis, and support vector machine models to first screen and obtain differentially expressed metabolites and trend-related metabolites for different T stages of colorectal cancer. Then, through feature selection and model optimization, four sets of core metabolic biomarker combinations that can specifically distinguish different T stages were finally determined, and a colorectal cancer T stage discrimination model with high accuracy and good stability was constructed.

[0041] The biomarker combination used to distinguish normal tissue from Tis / T1 phase consists of lysophosphatidylinositol 20:4 (LysoPI (20:4)), cyclic phosphatidylinositol phosphate 16:0 (CPA (16:0)), uric acid, palmitoleic acid, oxyphosphatidylcholine O-32:2 (PC (O-32:2)), inosine, sphingomyelin 42:2 (SM 42:2;O2), sphingomyelin 42:3 (SM 42:3;O2), acetylcholine, and lysophosphatidylcholine P-18:0 (LysoPC (P-18:0 / 0:0)). The training set AUC reached 0.984, the independent test set AUC reached 0.986, the sensitivity was 98.63%, and the specificity was 88.89%. The biomarker combination used to differentiate between Tis / T1 and T2 phases consists of phosphatidylcholine 36:0 (PC (36:0)), acetylcholine, sphingomyelin 40:1 (SM 40:1;O2), and phosphatidylcholine 31:0 hydrogen adduct (PC (31:0) + H+). + Phosphorylcholine (PC(30:0) + Na) ion adduct + The assay consisted of cyclic phosphatidic acid (CPA (16:0)), lysophosphatidylinositol (LysoPI (20:4)), taurine, and inosine. The training set AUC reached 0.844, the independent test set AUC reached 0.784, the sensitivity was 87.88%, and the specificity was 60.0%. The biomarker combination used to distinguish between T2 and T3 consists of oxidized phosphatidylcholine O-32:2 (PC (O-32:2)) and phosphocholine hydrogen adduct (Phosphorylcholine+H). + Phosphorylcholine + potassium ion adduct (Phosphorylcholine + K) + The assay consisted of oxidized phosphatidylcholine O-38:5 (PC (O-38:5)), phosphatidylcholine 36:0 (PC (36:0)), sphingomyelin 42:3 (SM 42:3;O2), sphingomyelin 42:2 (SM 42:2;O2), sphingomyelin 40:1 (SM 40:1;O2), phosphatidylcholine 30:0 (PC (30:0)), and lysophosphatidylcholine 16:0 (LysoPC (16:0)). The training set AUC reached 0.816, the independent test set AUC reached 0.872, the sensitivity was 92.59%, and the specificity was 55.0%. The biomarker combination used to distinguish between T3 and T4 phases consists of sphingomyelin 42:3 (SM 42:3; O2), phosphatidylcholine 36:0 (PC (36:0)), inosine, phosphatidylcholine 30:0 (PC (30:0)), phosphatidylserine 36:1 (PS (36:1)), phosphatidylcholine 33:0 (PC (33:0)), phosphatidylcholine 31:0 (PC (31:0)), cyclic phosphatidic acid 16:0 (CPA (16:0)), phosphatidylinositol 38:3 (PI (38:3)), and inositol cyclic phosphate. The training set AUC reached 0.81, the independent test set AUC reached 0.930, the sensitivity was 70.0%, and the specificity was 90.0%.

[0042] The above-mentioned biomarker combination and discrimination model have high discrimination accuracy and good stability, and can effectively distinguish and accurately assess the Tis / T1, T2, T3 and T4 stages of colorectal cancer, providing stable and reliable metabolic characteristics and technical support for rapid in vitro assessment of the depth of colorectal cancer tumor invasion.

[0043] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A combination of metabolic biomarkers for T-staging assessment of colorectal cancer, characterized in that, The biomarker combination comprises four groups of metabolite combinations specifically used to differentiate different T stages of colorectal cancer, each group containing 10 endogenous small molecule metabolites: The combination of metabolic markers used to distinguish normal tissue from Tis / T1 phase is: lysophosphatidylinositol 20:4, cyclic phosphatidylinositol phosphate 16:0, uric acid, palmitoleic acid, oxidized phosphatidylcholine O-32:2, inosine, sphingomyelin 42:2, sphingomyelin 42:3, acetylcholine, and lysophosphatidylcholine P-18:

0. The combination of metabolic markers used to distinguish between Tis / T1 and T2 phases is as follows: phosphatidylcholine 36:0, acetylcholine, sphingomyelin 40:1, phosphatidylcholine 31:0 hydrogen adduct, phosphatidylcholine hydrogen adduct, phosphatidylcholine 30:0 sodium adduct, cyclic phospholysinic acid 16:0, lysophosphatidylinositol 20:4, taurine, and inosine; The combination of metabolic markers used to distinguish between T2 and T3 phases is as follows: oxidized phosphatidylcholine O-32:2, phosphatidylcholine hydrogen adduct, phosphatidylcholine potassium adduct, oxidized phosphatidylcholine O-38:5, phosphatidylcholine 36:0, sphingomyelin 42:3, sphingomyelin 42:2, sphingomyelin 40:1, phosphatidylcholine 30:0, and lysophosphatidylcholine 16:

0. The combination of metabolic markers used to distinguish between T3 and T4 phases is: sphingomyelin 42:3, phosphatidylcholine 36:0, inosine, phosphatidylcholine 30:0, phosphatidylserine 36:1, phosphatidylcholine 33:0, phosphatidylcholine 31:0, cyclic phospholysinic acid 16:0, phosphatidylinositol 38:3, and cyclic phosphatidylinositol.

2. An iEESI-MS detection method for a combination of metabolic biomarkers as described in claim 1, characterized in that, The method described is a non-diagnostic / therapeutic in vitro tissue metabolic profiling detection method, comprising the following steps: (1) Sample processing: Collect ex vivo tissue samples related to colorectal cancer and corresponding ex vivo normal tissue samples adjacent to the cancer. After ex vivo, the samples are quickly frozen in liquid nitrogen and stored. Samples are taken directly during testing. (2) iEESI-MS detection: The iEESI-MS ion source and mass spectrometer were used together. The sampling needle was inserted into the ex vivo tissue sample, and the extraction solvent was injected to extract tissue metabolites. Metabolic spectrum data were collected simultaneously. (3) Data preprocessing and biomarker identification: Peak identification, peak alignment, missing value filtering and normalization are performed on the collected raw mass spectrometry data, and the identification of each metabolic molecule in the combination of metabolic biomarkers is completed by combining the metabolite database.

3. The detection method according to claim 2, characterized in that, In step (2), the extraction solvent is a mixture of methanol and water in a volume ratio of 95:5, which is injected into the tissue at a flow rate of 2 μL / min using a syringe pump.

4. The detection method according to claim 2, characterized in that, In step (2), the ionization voltage detected by iEESI-MS is ±4.5kV, the temperature of the ion transfer tube is 250℃, the injection volume is 0.5~1.0mg of tissue, and the tissue sample is a direct sample taken from fresh frozen tissue.

5. The detection method according to claim 2, characterized in that, In step (2), iEESI-MS detection is performed in positive and negative ion modes, and the mass scan range is [missing information]. m / z 100–900, with a single sample testing time not exceeding 5 minutes.

6. The use of a combination of metabolic markers as described in claim 1 in the preparation of reagents or kits for colorectal cancer research or in vitro sample T-staging analysis.

7. A method for T-staging of colorectal cancer for non-diagnostic and therapeutic purposes, characterized in that, Using the combination of metabolic biomarkers described in claim 1 as staging features, the expression level data of the metabolic biomarkers in the sample to be tested are obtained by any of the detection methods described in claims 2-5, and the data is input into a pre-constructed machine learning staging model, thereby realizing the differentiation of different T stages of colorectal cancer.

8. The discrimination method according to claim 7, characterized in that, The machine learning algorithm is a support vector machine, and the staging discrimination model includes four sub-models that respectively distinguish between normal tissue and Tis / T1 stage, Tis / T1 stage and T2 stage, T2 stage and T3 stage, and T3 stage and T4 stage.

9. The discrimination method according to claim 8, characterized in that, The expression levels of the metabolic markers fluctuate regularly with the progression of colorectal cancer T stages from Tis / T1, T2, T3 to T4. The staging model uses this specific expression change trend to determine the depth of local tumor invasion and stage the tumor.

10. A combination of metabolic markers associated with the progression of T stage and tumor invasion depth in colorectal cancer, characterized in that, The metabolic markers include: inosine, sphingomyelin 36:1, cyclic inositol phosphate, cyclic lysophosphatidylcholine phosphate 16:0, sphingomyelin 42:2, phosphatidylserine 36:1, sphingomyelin 40:1, phosphatidylcholine 37:4, lysophosphatidylcholine P-18:0 / 0:0, lysophosphatidylinositol 20:4, phosphatidylcholine 29:0, phosphatidylglycerol 34:1, phosphatidylcholine 33:1, phosphatidylcholine 32:2, phosphatidylcholine 38:2, phosphatidylcholine 30:0, and phosphatidylglycerol 32:0.

Images