A method, apparatus, medium, and program product for predicting different stages of progression of colorectal cancer

Abstract

The application discloses a method, device, medium and program product for predicting different progress stages of colorectal cancer, which predicts different progress stages of colorectal cancer based on expression amounts of target proteins. With the aid of a computer system and device, the method can well predict whether a subject has a colorectal cancer related disease by detecting expression levels of target proteins, and can distinguish different progress stages of colorectal cancer, thereby providing a new direction and means for early treatment and effective treatment of colorectal cancer, and having practical significance for clinical guidance of colorectal cancer.

Description

Technical Field

[0001] This invention relates to the field of intelligent medical technology, and more specifically, to a method, device, medium, and program product for predicting different stages of colorectal cancer progression. Background Technology

[0002] Colorectal cancer (CRC) is the third most common type of cancer worldwide and the second leading cause of cancer death. Epidemiologically, CRC is more common in older adults, with a sharp increase in incidence after age 50. However, worryingly, the incidence of early-onset CRC (cases diagnosed in people under 50) is also rising, suggesting potential changes in environmental and genetic risk factors, as well as inadequate early screening measures.

[0003] Because colorectal cancer symptoms are nonspecific and non-invasive detection methods are lacking, the disease is often diagnosed at an advanced stage, by which time the optimal window for surgical treatment has passed. Generally, colorectal cancer progresses in adenoma-carcinoma sequence and is classified into early-stage (e.g., T1 stage) and late-stage (including T2, T3, and T4 stages). Specifically, T1 stage is further divided into superficial epithelial tumor and invasive carcinoma stages. Screening for colorectal cancer at different stages of progression would significantly improve the chances of early diagnosis and treatment. Currently, colonoscopy remains the ultimate common method for screening all colorectal cancers and is arguably the most effective single-use method for preventing colorectal cancer. However, this method still has significant limitations in detecting distal colorectal cancer, and colonoscopy is highly invasive, expensive, complex to perform, and has poor patient compliance. Therefore, a non-invasive detection method is urgently needed. Summary of the Invention

[0004] To overcome the shortcomings of the prior art, the purpose of this invention is to provide a method, device, medium, and procedure for predicting different stages of colorectal cancer progression.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] The first aspect of this invention provides a method for predicting different stages of colorectal cancer progression.

[0007] Furthermore, the method is performed by a computer, and the method includes the following steps:

[0008] Obtain expression data of target proteins from the sample to be tested, wherein the target proteins include one or more of the following: COL6A1, F2, TRIT1, ABCC2, BAD, CTSL, FGL2, MED1;

[0009] The expression data of the target protein is input into the constructed prediction model, which predicts the colorectal cancer progression stage of the subject based on the expression data of the target protein; the colorectal cancer progression stage includes: no colorectal cancer-related disease, colorectal adenoma, and colorectal cancer;

[0010] Output the prediction results.

[0011] In this invention, "not suffering from colorectal cancer-related diseases" refers to a healthy individual who does not have any colorectal cancer-related diseases such as colorectal cancer, colorectal adenoma, high-grade intraepithelial neoplasia of the colorectal region, or colorectal carcinoma in situ, and whose colonoscopy does not reveal any abnormal lesions.

[0012] In this invention, the "colorectal adenoma" is a benign tumor of the colorectal mucosal epithelium, composed of proliferating glandular epithelial cells, and belongs to the precancerous lesion stage of colorectal cancer.

[0013] In this invention, "colorectal cancer" refers to a malignant tumor that occurs in the colorectal mucosal epithelium, has broken through the basement membrane, can invade the muscular layer and serosa, and may metastasize to lymph nodes or distant sites, including the T1 stage invasive carcinoma stage and T2, T3, and T4 stages.

[0014] Furthermore, the method includes: acquiring expression data of a first target protein in the sample to be tested, inputting the expression data of the first target protein into a first prediction model, and predicting whether the subject has colorectal cancer-related diseases based on the output of the first prediction model; the first target protein includes one or two of the following: COL6A1 and F2.

[0015] In this invention, the colorectal cancer-related diseases include colorectal adenoma, high-grade intraepithelial neoplasia of the colorectum, colorectal carcinoma in situ, and colorectal cancer.

[0016] Furthermore, the method includes: acquiring expression data of a first target protein and a second target protein of the sample to be tested; firstly, inputting the expression data of the first target protein into a first prediction model; if the output of the first prediction model predicts that the subject has colorectal cancer-related diseases, then inputting the expression data of the second target protein into a second prediction model to predict whether the subject has colorectal adenoma, thereby obtaining the subject's colorectal cancer progression stage; the second target protein includes one or two of the following: TRIT1 and ABCC2.

[0017] Furthermore, the method includes: acquiring expression data of a first target protein and a third target protein of the sample to be tested; firstly, inputting the expression data of the first target protein into a first prediction model; if the output of the first prediction model predicts that the subject has colorectal cancer-related diseases, then inputting the expression data of the third target protein into a third prediction model to predict whether the subject has colorectal cancer, thereby obtaining the subject's colorectal cancer progression stage; the third target protein includes one or more of the following: BAD, CTSL, FGL2, MED1.

[0018] Furthermore, the method further includes: acquiring the expression data of the first target protein and the fourth target protein of the sample to be tested; firstly, inputting the expression data of the first target protein into the first prediction model; if the output of the first prediction model predicts that the subject has colorectal cancer-related diseases, then inputting the expression data of the fourth target protein into the fourth prediction model to predict whether the subject has high-grade intraepithelial neoplasia of the colorectal region, thereby obtaining the subject's colorectal cancer progression stage; the fourth target protein includes one or more of the following: RAMAC, NELFB, SEC61G, KCTD18, CFAP298, FMC1, NDUFA11, GNA15, CDC27, and ANKRD54.

[0019] Furthermore, the method also includes: acquiring the expression data of the first target protein and the fifth target protein of the sample to be tested; firstly, inputting the expression data of the first target protein into the first prediction model; if the output of the first prediction model predicts that the subject has colorectal cancer-related diseases, then inputting the expression data of the fifth target protein into the fifth prediction model to predict whether the subject has colorectal carcinoma in situ, thereby obtaining the subject's colorectal cancer progression stage; the fifth target protein includes one or more of the following: IGLV2-23, ATP6AP1, CASP3, ATP1B1, GTF2A2, ZNF280C, IGKV1D-13, DNAJC19, WASHC3, ME1.

[0020] In this invention, the "high-grade intraepithelial neoplasia of the colorectal region" is a severe dysplasia of the colorectal mucosal epithelium, in which the cell morphology and arrangement are close to those of cancer cells, but have not broken through the basement membrane, and belong to the highest stage of precancerous lesions.

[0021] In this invention, "colorectal carcinoma in situ" refers to the very early stage of colorectal cancer, in which cancer cells are confined to the mucosal epithelial layer, have not broken through the basement membrane, and have not metastasized to lymph nodes or distant sites, belonging to the T1 stage of superficial epithelial tumors.

[0022] Furthermore, the first target protein also includes SENP1, C8G, HIGD1A, RIPOR3, ITLN1, LOX, PTPRCAP, and MMP19.

[0023] Furthermore, the second target protein also includes GSTM2, DCAF13, PDRG1, MEAF6, SRXN1, ASRGL1, EPHB4, and BET1.

[0024] Furthermore, the third target protein also includes FGFR1, SLC9A9, ZC4H2, COL5A3, KIF21A, and FSTL1.

[0025] Furthermore, the construction steps of the first prediction model are as follows: obtaining the expression data of the first target protein, wherein the expression data of the first target protein comes from people with colorectal cancer-related diseases and people without colorectal cancer-related diseases; and inputting the expression data of the first target protein into a machine learning algorithm to construct the first prediction model.

[0026] Furthermore, the construction steps of the second prediction model are as follows: obtaining the expression data of the second target protein, which comes from people with colorectal adenoma and people with other stages of colorectal cancer progression, including high-grade intraepithelial neoplasia of the colorectal, colorectal carcinoma in situ, and colorectal cancer; inputting the expression data of the second target protein into a machine learning algorithm to construct the second prediction model.

[0027] Furthermore, the construction steps of the third prediction model are as follows: obtaining the expression data of the third target protein, which comes from people with colorectal cancer and people with other stages of colorectal cancer progression, including colorectal adenoma, high-grade intraepithelial neoplasia of the colorectal region, and colorectal carcinoma in situ; inputting the expression data of the third target protein into a machine learning algorithm to construct the third prediction model.

[0028] Furthermore, the construction steps of the fourth prediction model are as follows: obtaining the expression data of the fourth target protein, which comes from individuals with high-grade colorectal intraepithelial neoplasia and individuals with other stages of colorectal cancer progression, including colorectal adenoma, colorectal carcinoma in situ, and colorectal cancer; and inputting the expression data of the fourth target protein into a machine learning algorithm to construct the fourth prediction model.

[0029] Furthermore, the construction steps of the fifth prediction model are as follows: obtaining the expression data of the fifth target protein, which comes from people with colorectal carcinoma in situ and people with other stages of colorectal cancer progression, including colorectal adenoma, high-grade intraepithelial neoplasia of the colorectal gland, and colorectal cancer; inputting the expression data of the fifth target protein into a machine learning algorithm to construct the fifth prediction model.

[0030] Furthermore, the machine learning algorithm includes algorithmic models developed using various development tools.

[0031] Furthermore, the development tools include, but are not limited to, TensorFlow, Scikit-Learn, PyTorch, OpenNN, RapidMiner, Azure Machine Learning, Apache Mahout, Shogun, KNIME, Vertex AI, H2Oai, Anaconda, Keras, Tableau, Fast.ai, Catalyst, Amazon ML, MLJAR, Spell, and Rstudio.

[0032] Furthermore, the algorithm models include, but are not limited to, linear regression models, logistic regression models, Lasso regression models, Ridge regression models, linear discriminant analysis models, nearest neighbor models, decision tree models, perceptron models, neural network models, support vector machine models, Naive Bayes models, AdaBoost models, GBDT models, XGBoost models, LightGBM models, CatBoost models, and random forest models.

[0033] Furthermore, the first prediction model, the second prediction model, the third prediction model, the fourth prediction model, and the fifth prediction model obtain the prediction results using the following formula: , where, β0, β1,…,β p Model coefficients for each protein, x1, x2, ..., x p Expression data for each protein.

[0034] Furthermore, the first prediction model obtains prediction results using the following criteria: when the p-value is lower than the optimal classification threshold, the subject is classified as not having colorectal cancer-related diseases; when the p-value is higher than the optimal classification threshold, the subject is classified as having colorectal cancer-related diseases.

[0035] Furthermore, the second prediction model obtains prediction results using the following criteria: when the p-value is lower than the optimal classification threshold, the subject is classified as not having colorectal adenoma; when the p-value is higher than the optimal classification threshold, the subject is classified as having colorectal adenoma.

[0036] Furthermore, the third prediction model obtains prediction results using the following criteria: when the p-value is lower than the optimal classification threshold, the subject is classified as not having colorectal cancer; when the p-value is higher than the optimal classification threshold, the subject is classified as having colorectal cancer.

[0037] Furthermore, the fourth prediction model obtains prediction results using the following criteria: when the p-value is lower than the optimal classification threshold, the subject is classified as not having high-grade colorectal intraepithelial neoplasia; when the p-value is higher than the optimal classification threshold, the subject is classified as having high-grade colorectal intraepithelial neoplasia.

[0038] Furthermore, the fifth prediction model obtains prediction results using the following criteria: when the p-value is lower than the optimal classification threshold, the subject is classified as not having colorectal carcinoma in situ; when the p-value is higher than the optimal classification threshold, the subject is classified as having colorectal carcinoma in situ.

[0039] In this invention, the optimal classification threshold, also known as the best cutoff value, refers to the critical prediction probability value that best balances sensitivity and specificity in a biomarker-based classification or prediction model, thereby achieving the optimal overall classification performance of the model. This threshold is not a fixed, theoretical value, but is determined based on a specific training dataset. In some preferred embodiments of this invention, the training dataset includes 20 or more samples, such as 30, 50, 80, 100, 150, 200, 300, 500, or more.

[0040] In a specific embodiment of the present invention, the optimal classification threshold of the first prediction model is 0.5366108 in the training set and 0.7093998 in the validation set. The optimal classification threshold of the second prediction model is 0.5011554 in the training set and 0.303436 in the validation set; the optimal classification threshold of the third prediction model is 0.4025858 in the training set and 0.4482433 in the validation set; the optimal classification threshold of the fourth prediction model is 0.2027766; and the optimal classification threshold of the fifth prediction model is 0.1688771. Those skilled in the art will understand that, based on the sample distribution of the actual application population, the optimal classification threshold can be appropriately adjusted within, for example, the range of 0.4 to 0.7 to optimize classification performance in clinical practice.

[0041] A second aspect of the present invention provides a system for predicting different stages of colorectal cancer progression.

[0042] Furthermore, the system includes:

[0043] Acquisition Unit: Acquires expression data of target proteins in the sample to be tested, wherein the target proteins include one or more of the following: COL6A1, F2, TRIT1, ABCC2, BAD, CTSL, FGL2, MED1;

[0044] Processing unit: Inputs the expression data of the target protein into the constructed prediction model, which predicts the colorectal cancer progression stage of the subject based on the expression data of the target protein; the colorectal cancer progression stage includes: no colorectal cancer-related diseases, colorectal adenoma, and colorectal cancer;

[0045] Output unit: Outputs the prediction results.

[0046] A third aspect of the present invention provides a computer device.

[0047] Furthermore, the device includes: a memory and a processor; the memory is used to store a computer program; the processor executes the computer program to implement the method described in the first aspect of the present invention.

[0048] A fourth aspect of the present invention provides a computer-readable storage medium.

[0049] Furthermore, the computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in the first aspect of the present invention.

[0050] The fifth aspect of the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the method described in the first aspect of the present invention.

[0051] Advantages and beneficial effects of the present invention:

[0052] This invention has discovered non-invasive biomarkers for screening colorectal cancer at different stages and established related detection methods, which can be used to distinguish patients at different stages of colorectal cancer progression, thus helping patients to seek medical attention as early as possible for diagnosis and treatment. Attached Figure Description

[0053] Figure 1 This is a schematic diagram of the method flow provided in the first aspect of the present invention;

[0054] Figure 2 This is a schematic diagram of the system structure provided in the second aspect of the present invention;

[0055] Figure 3 This is a schematic diagram of a computer device provided in an embodiment of the present invention;

[0056] Figure 4 This is a graph showing the performance validation results of the diagnostic model for distinguishing colorectal cancer-related diseases from normal controls, as well as the validation results of its characteristic biomarkers.

[0057] Figure 5 This is a diagram showing the performance validation results and characteristic biomarker validation results of a diagnostic model for differentiating colorectal adenomas from other stages of colorectal cancer.

[0058] Figure 6 This is a graph showing the performance validation results of the diagnostic model for colorectal cancer and the validation results of its characteristic biomarkers in distinguishing colorectal cancer from other stages of colorectal cancer progression.

[0059] Figure 7 This is a graph showing the results of further performance validation and further identification of characteristic biomarkers for the plasma PRM-targeted validation model that distinguishes colorectal cancer-related diseases from normal controls.

[0060] Figure 8 This is a graph showing the results of further performance validation and further identification of characteristic biomarkers for a plasma PRM-targeted diagnostic model that distinguishes colorectal adenoma from other advanced stages of colorectal cancer.

[0061] Figure 9 This is a diagram showing the results of further performance validation and further identification of characteristic biomarkers for a plasma PRM-targeted diagnostic model that distinguishes colorectal cancer from other advanced stages of colorectal cancer.

[0062] Figure 10 This is the ROC result graph of the characteristic biomarkers (COL6A1, F2) of the colorectal cancer-related disease group vs. the normal control group after further screening in the discovery set;

[0063] Figure 11 This is the ROC result graph of the colorectal adenoma group versus other colorectal cancer-related disease groups (ABCC2, TRIT1) in the discovery set;

[0064] Figure 12 This is the ROC result graph of the colorectal cancer group compared with other colorectal cancer-related disease groups (CTSL, FGL2, BAD, MED1) in the discovery set. Detailed Implementation

[0065] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0066] In some of the processes described in the specification, claims, and accompanying drawings of this invention, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as S101, S102, etc., are merely used to distinguish different operations and do not represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0067] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0068] Figure 1 This is a schematic flowchart of a method for predicting different stages of colorectal cancer progression, provided by the first aspect of the present invention. Specifically, the method includes the following steps:

[0069] S101: Obtain the expression data of the target protein in the sample to be tested, wherein the target protein includes one or more of the following: COL6A1, F2, TRIT1, ABCC2, BAD, CTSL, FGL2, MED1;

[0070] S102: Input the expression data of the target protein into the constructed prediction model, which predicts the colorectal cancer progression stage of the subject based on the expression data of the target protein; the colorectal cancer progression stage includes: no colorectal cancer-related diseases, colorectal adenoma, and colorectal cancer;

[0071] S103: Output the prediction results.

[0072] In this invention, the term "sample" as used refers to a composition obtained from or derived from a patient / subject that contains cells and / or other molecular entities to be characterized and / or identified based on, for example, physical, biochemical, chemical, and / or physiological characteristics. For example, a sample refers to any sample derived from a patient / subject that is expected or known to contain cells and / or molecular entities to be characterized. Samples include, but are not limited to, tissue samples, primary or cultured cells or cell lines, cell cultures, cell supernatants, cell lysates, platelets, serum, plasma, vitreous fluid, lymph, synovial fluid, follicular fluid, semen, amniotic fluid, milk, whole blood, blood-derived cells, urine, cerebrospinal fluid, saliva, sputum, tears, sweat, mucus, tissue culture fluid, tissue extracts, homogenized tissue, cell extracts, and combinations thereof. In a specific embodiment of this invention, the sample is plasma.

[0073] In some embodiments of the present invention, the methods for constructing the predictive model are known to those skilled in the art and can be implemented and realized in different ways to associate the expression level of a gene marker with a certain probability or risk. Preferably, the measured concentrations of the marker and one or more other markers are mathematically combined, and the combined value is associated with the fundamental stage of colorectal cancer progression. The measured marker values ​​can be combined using any suitable existing mathematical method, and the predictive model can be constructed using machine learning algorithms. In a specific embodiment of the present invention, the predictive model is a Lasso logistic regression model.

[0074] In some implementations, after the predictive model is constructed, ROC curve analysis is used to evaluate the diagnostic efficacy of the predictive model.

[0075] In some embodiments of the present invention, the discovery set sample was sourced and grouped as follows: A total of 113 patients were included in this discovery set study cohort, comprising 25 patients with CRA (colorectal adenoma), 26 patients with HGIN (colorectal intraepithelial neoplasia), and 12 patients with CIS (colorectal carcinoma in situ) (who had not received other treatment prior to endoscopic treatment), 33 patients with CRC (colorectal cancer) (who underwent surgical resection), and 17 NC (normal controls) patients who underwent endoscopy. All patients were clinically and pathologically confirmed, and the pTNM stage of the cancer was determined according to the AJCC classification. Patients with other malignancies or imaging findings indicating distant metastasis were excluded. Basic demographic and clinical information, including sex, age, tumor size, and pathological type, was recorded.

[0076] In some embodiments, sample collection and processing are as follows: Plasma samples are collected using EDTA-K tubes prior to endoscopic treatment or surgical resection. After centrifugation at 3000 rpm for 10 minutes at room temperature, the supernatant plasma is aliquoted and stored at -80°C for subsequent analysis.

[0077] Plasma samples were processed on an automated sample processing platform (Magicomics-AP-96, Beijing Qinglian Bio-Tech Co., Ltd., China) using the Qinglian Bio-Low Abundance Protein Enrichment Magnetic Bead Kit (QLBIO MagicOmics-DMB8X, Beijing Qinglian Bio-Tech Co., Ltd., China).

[0078] In some embodiments, plasma proteomics detection was performed as follows: All samples were processed using a timsTOF HT mass spectrometer (Germany) combined with a liquid chromatography system in data-independent acquisition (DIA) mode, with a mass spectrometry scan range of m / z 300–1500 and a resolution of 60,000 (at m / z 1222). After generating the spectral library, the raw DIA data files were integrated into Spectronaut software and compared with the human UniProt database. Cystylation was set as a fixed modification, while N-terminal acetylation and methionine oxidation of proteins were set as variable modifications, allowing a maximum of two deletion cleavage sites. Identification and quantification results of proteins and peptides were output, with false discovery rates (FDR) of less than 0.01 at both protein and peptide levels.

[0079] In some embodiments, the preprocessing of plasma proteomics data related to colorectal cancer is as follows: After removing all data with values ​​less than 1, the quantitative data of all samples were standardized using the median of common proteins detectable in all samples. Only proteins detected in at least one group of samples at a rate greater than 50% were retained for subsequent analysis. Missing values ​​were imputed with the global minimum, and all data were transformed using Log2.

[0080] In some embodiments, differential expression analysis of plasma proteomics data related to colorectal cancer is performed as follows: Differential expression analysis was conducted on any two groups from the normal control group, colorectal adenoma group, high-grade colorectal intraepithelial neoplasia group, colorectal carcinoma in situ group, and colorectal cancer groups (NC, CRA, HGIN, CIS, and CRC) using limma (R software version 3.56.2). The candidate biomarker screening process is as follows:

[0081] (1) In order to obtain candidate biomarkers that show significant differences between different stages of colorectal cancer and normal groups, this disclosure first screens the intersection of significantly differentially expressed proteins in all comparison groups obtained by colorectal cancer-related disease groups vs. normal control groups with a threshold |FoldChange|≥2, P < 0.01 (i.e., colorectal adenoma group vs. normal control group ∩ colorectal high-grade intraepithelial neoplasia group vs. normal control group ∩ colorectal carcinoma in situ group vs. normal control group ∩ colorectal cancer group vs. normal control group).

[0082] (2) In order to obtain candidate biomarkers with significant differential expression specific to colorectal cancer at different stages of progression, candidate biomarkers for each stage of colorectal cancer were screened with a threshold of |FoldChange|≥2 and P < 0.05:

[0083] a. The intersection of significantly differentially expressed proteins in all comparison groups obtained from the colorectal adenoma group vs. any other colorectal cancer-related disease group (colorectal adenoma group vs. high-grade intraepithelial neoplasia of the colorectal tissue ∩ colorectal adenoma group vs. colorectal carcinoma in situ group ∩ colorectal adenoma group vs. colorectal cancer group);

[0084] b. The intersection of significantly differentially expressed proteins in all comparison groups obtained from the high-grade intraepithelial neoplasia group vs. any other colorectal cancer-related disease group (high-grade intraepithelial neoplasia group vs. colorectal adenoma group ∩ high-grade intraepithelial neoplasia group vs. colorectal carcinoma in situ group ∩ high-grade intraepithelial neoplasia group vs. colorectal cancer group);

[0085] c. The intersection of significantly differentially expressed proteins in all comparison groups obtained from the colorectal carcinoma in situ group vs. any other colorectal cancer-related disease group (colorectal carcinoma in situ group vs. colorectal adenoma group ∩ colorectal carcinoma in situ group vs. high-grade intraepithelial neoplasia of the colorectal gland group ∩ colorectal carcinoma in situ group vs. colorectal cancer group);

[0086] d. The intersection of significantly differentially expressed proteins in all comparison groups obtained from the colorectal cancer group vs. any other colorectal cancer-related disease group (colorectal cancer group vs. colorectal adenoma group ∩ colorectal cancer group vs. high-grade intraepithelial neoplasia group ∩ colorectal cancer group vs. colorectal carcinoma in situ group).

[0087] In some embodiments, the plasma biomarker feature screening and diagnostic model construction and model validation are as follows: Based on the results of the above candidate biomarkers, according to the candidate biomarkers screened for each stage of colorectal cancer disease progression, the R package glmnet (v4.1.8) is used with cv.glmnet to perform LASSO regression analysis through 10-fold cross-validation (nfolds=10), and the optimal lambda value (lambda.min) is called in each comparison group to screen out effective feature biomarkers (the feature weight coefficient of the biomarker is not 0 in the model).

[0088] Based on the importance weight scores of the feature biomarkers in each disease group, the top 10 feature biomarkers in each disease group are selected to construct the biomarker model for each disease group, specifically including:

[0089] 1) A model to distinguish patients with colorectal cancer-related diseases from normal controls, including 10 plasma protein markers (SENP1, COL6A1, F2, C8G, HIGD1A, RIPOR3, ITLN1, LOX, PTPRCAP, MMP19).

[0090] 2) A model to differentiate colorectal adenoma from other patients with advanced colorectal cancer, including 10 plasma protein markers (TRIT1, GSTM2, DCAF13, ABCC2, PDRG1, MEAF6, SRXN1, ASRGL1, EPHB4, BET1).

[0091] 3) A model to differentiate high-grade intraepithelial neoplasia of the colorectal cancer from other patients in the advanced stages of colorectal cancer, including 10 plasma protein markers (RAMAC, NELFB, SEC61G, KCTD18, CFAP298, FMC1, NDUFA11, GNA15, CDC27, ANKRD54).

[0092] 4) A model to distinguish colorectal carcinoma in situ from other patients with advanced colorectal cancer, including 10 plasma protein markers (IGLV2-23, ATP6AP1, CASP3, ATP1B1, GTF2A2, ZNF280C, IGKV1D-13, DNAJC19, WASHC3, ME1).

[0093] 5) A model to distinguish colorectal cancer from other patients at the advanced stage of colorectal cancer, including 10 plasma protein markers (FGFR1, BAD, CTSL, FGL2, SLC9A9, ZC4H2, COL5A3, KIF21A, MED1, FSTL1).

[0094] The model for each disease group was fitted using the Lasso logistic regression method, and the prediction accuracy of the model for each disease group was predicted using the predict function.

[0095] Based on the prediction models constructed for each disease group, all models were evaluated and validated using ROC analysis and confusion matrix analysis. The results of ROC analysis and confusion matrix analysis show:

[0096] 1) A model to distinguish between patients with colorectal cancer-related diseases and normal controls, AUC=1, sensitivity=100%, specificity=100% ( Figure 4 );

[0097] 2) A model to distinguish colorectal adenoma from other patients with advanced colorectal cancer, AUC=1, sensitivity=100%, specificity=100% ( Figure 5 );

[0098] 3) A model for distinguishing high-grade intraepithelial neoplasia of the colorectal region from other patients in the advanced stages of colorectal cancer, AUC=0.95, sensitivity=100%, specificity=77%;

[0099] 4) A model for distinguishing colorectal carcinoma in situ from other patients with advanced colorectal cancer, AUC=0.99, sensitivity=100%, specificity=95%;

[0100] 5) A model to distinguish colorectal cancer from other patients at the advanced stage of colorectal cancer, AUC=1, sensitivity=97%, specificity=98% ( Figure 6 ).

[0101] In some embodiments, based on the results of colorectal cancer disease models at different progression stages in the above-mentioned discovery set, PRM-targeted analysis was performed to validate three disease models with high reliability of their prediction results. The three models were: colorectal cancer disease group at different progression stages vs. normal control group; colorectal adenoma group vs. colorectal cancer-related disease group at other progression stages; and colorectal cancer group vs. colorectal cancer-related disease group at other progression stages. The specific process is as follows:

[0102] In some implementations, the validation cohort newly included 18 patients (5 healthy controls, 5 colorectal adenomas, 1 high-grade colorectal intraepithelial neoplasia, 1 colorectal carcinoma in situ, and 6 colorectal cancers), with the same inclusion criteria and plasma sample collection methods as the aforementioned discovery set. Plasma sample PRM targeting analysis was performed: using a temps-to-free (tTOF) HT device with parallel reaction monitoring (PRM) technology, peptides of candidate proteins were detected in the validation set (these peptides were unique, with no missed cleavage or oxidative modifications). The PRM data were then analyzed using SpectroDive 24.1 software.

[0103] Furthermore, trend consistency marker peptide screening was performed: Based on the above PRM analysis results, trend consistency analysis was performed on the identified peptides, and characteristic peptides consistent with the central expression trend of the plasma DIA dataset were screened according to the threshold |FoldChange| ≥ 1.2. Then, the peptide data were attributed to the protein level.

[0104] Further, biomarker feature screening, model refitting, and performance evaluation were performed on the PRM plasma dataset: Based on the proteins retained from the trend consistency biomarker peptide screening, specifically, LASSO regression analysis was conducted using the R package glmnet (v4.1.8) with cv.glmnet through 5-fold cross-validation (nfolds=5), and the optimal lambda value (lambda.min) was called to screen effective biomarkers (biomarker feature weight coefficients are not 0 in the model) for each stage of colorectal cancer progression. Specifically, this included:

[0105] 1) The model that distinguishes between the colorectal cancer-related disease group and the normal control group further identified two plasma protein markers (COL6A1, F2). Figure 7 );

[0106] 2) The model that distinguishes the colorectal adenoma group from other advanced stages of colorectal cancer further identified two plasma protein markers (TRIT1 and ABCC2). Figure 8 );

[0107] 3) The model that distinguishes colorectal cancer from other advanced stages of colorectal cancer further identified four plasma protein markers (BAD, CTSL, FGL2, MED1). Figure 9 ).

[0108] Based on the above characteristic proteins, the model was refitted, and the accuracy of the model was predicted using the predict function. ROC analysis and confusion matrix analysis were then used to further validate the plasma PRM dataset model. The results of ROC analysis and confusion matrix analysis show:

[0109] 1) A model for distinguishing between colorectal cancer-related diseases and normal control patients, with AUC=0.85, sensitivity=85%, and specificity=80%;

[0110] 2) A model for distinguishing colorectal adenomas from other patients with advanced colorectal diseases, AUC=0.98, sensitivity=100%, specificity=88%;

[0111] 3) A model to distinguish colorectal cancer from other advanced stages of colorectal cancer, AUC=1, sensitivity=100%, specificity=100%.

[0112] Furthermore, in the study, the diagnostic efficacy of COL6A1, F2, TRIT1, ABCC2, BAD, CTSL, FGL2, and MED1 was validated using PRM. The results showed that the AUC values ​​of the COL6A1 and F2 groups, the TRIT1 and ABCC2 groups, and the BAD, CTSL, FGL2, and MED1 groups were all greater than 0.7. Figures 10-12 The results indicate that COL6A1, F2, TRIT1, ABCC2, BAD, CTSL, FGL2, and MED1 all have good diagnostic efficacy in their respective models.

[0113] In some embodiments, the method includes: acquiring expression data of a first target protein in a sample to be tested, inputting the expression data of the first target protein into a first prediction model, and predicting whether the subject has colorectal cancer-related diseases based on the output of the first prediction model; the first target protein includes one or two of the following: COL6A1 and F2.

[0114] In some embodiments, the method includes: acquiring expression data of a first target protein and a second target protein of the sample to be tested; firstly, inputting the expression data of the first target protein into a first prediction model; if the output of the first prediction model predicts that the subject has colorectal cancer-related diseases, then inputting the expression data of the second target protein into a second prediction model to predict whether the subject has colorectal adenoma, thereby obtaining the subject's colorectal cancer progression stage; the second target protein includes one or two of the following: TRIT1 and ABCC2.

[0115] In some embodiments, the method includes: acquiring expression data of a first target protein and a third target protein of the sample to be tested; firstly, inputting the expression data of the first target protein into a first prediction model; if the output of the first prediction model predicts that the subject has colorectal cancer-related diseases, then inputting the expression data of the third target protein into a third prediction model to predict whether the subject has colorectal cancer, thereby obtaining the subject's colorectal cancer progression stage; the third target protein includes one or more of the following: BAD, CTSL, FGL2, MED1.

[0116] In some embodiments, the method further includes: acquiring expression data of a first target protein and a fourth target protein of the sample to be tested; firstly, inputting the expression data of the first target protein into a first prediction model; if the output of the first prediction model predicts that the subject has colorectal cancer-related diseases, then inputting the expression data of the fourth target protein into a fourth prediction model to predict whether the subject has high-grade intraepithelial neoplasia of the colorectal region, thereby obtaining the subject's colorectal cancer progression stage; the fourth target protein includes one or more of the following: RAMAC, NELFB, SEC61G, KCTD18, CFAP298, FMC1, NDUFA11, GNA15, CDC27, and ANKRD54.

[0117] In some embodiments, the method further includes: acquiring expression data of a first target protein and a fifth target protein of the sample to be tested; firstly, inputting the expression data of the first target protein into a first prediction model; if the output of the first prediction model predicts that the subject has colorectal cancer-related diseases, then inputting the expression data of the fifth target protein into a fifth prediction model to predict whether the subject has colorectal carcinoma in situ, thereby obtaining the subject's colorectal cancer progression stage; the fifth target protein includes one or more of the following: IGLV2-23, ATP6AP1, CASP3, ATP1B1, GTF2A2, ZNF280C, IGKV1D-13, DNAJC19, WASHC3, ME1.

[0118] Furthermore, the first prediction model, the second prediction model, the third prediction model, the fourth prediction model, and the fifth prediction model obtain the prediction results using the following formula: , where, β0, β1,…,β p Model coefficients for each protein, x1, x2, ..., x p Expression data for each protein.

[0119] Furthermore, the first prediction model obtains prediction results using the following criteria: when the p-value is lower than the optimal classification threshold, the subject is classified as not having colorectal cancer-related diseases; when the p-value is higher than the optimal classification threshold, the subject is classified as having colorectal cancer-related diseases.

[0120] Furthermore, the second prediction model obtains prediction results using the following criteria: when the p-value is lower than the optimal classification threshold, the subject is classified as not having colorectal adenoma; when the p-value is higher than the optimal classification threshold, the subject is classified as having colorectal adenoma.

[0121] Furthermore, the third prediction model obtains prediction results using the following criteria: when the p-value is lower than the optimal classification threshold, the subject is classified as not having colorectal cancer; when the p-value is higher than the optimal classification threshold, the subject is classified as having colorectal cancer.

[0122] Furthermore, the fourth prediction model obtains prediction results using the following criteria: when the p-value is lower than the optimal classification threshold, the subject is classified as not having high-grade colorectal intraepithelial neoplasia; when the p-value is higher than the optimal classification threshold, the subject is classified as having high-grade colorectal intraepithelial neoplasia.

[0123] Furthermore, the fifth prediction model obtains prediction results using the following criteria: when the p-value is lower than the optimal classification threshold, the subject is classified as not having colorectal carcinoma in situ; when the p-value is higher than the optimal classification threshold, the subject is classified as having colorectal carcinoma in situ.

[0124] Figure 2 This is a schematic diagram of the system structure provided by the present invention for predicting different stages of colorectal cancer progression.

[0125] The system is programmed or otherwise configured to include an acquisition unit 201, a processing unit 202, and an output unit 203;

[0126] Acquisition Unit 201: Acquires expression data of target proteins in the sample to be tested, wherein the target proteins include one or more of the following: COL6A1, F2, TRIT1, ABCC2, BAD, CTSL, FGL2, MED1;

[0127] Processing unit 202: Inputs the expression data of the target protein into the constructed prediction model, which predicts the colorectal cancer progression stage of the subject based on the expression data of the target protein; the colorectal cancer progression stage includes: no colorectal cancer-related diseases, colorectal adenoma, and colorectal cancer;

[0128] Output unit 203: Outputs the prediction results.

[0129] Figure 3 A schematic diagram of the structure of the computer device provided by the present invention.

[0130] The computer device 300 includes a processor 301 and a memory 302 coupled to the processor 301. The memory 302 stores program instructions. When the program instructions are executed by the processor 301, the processor 301 performs the method described above for predicting different stages of colorectal cancer progression.

[0131] The processor 301 can also be referred to as a CPU (Central Processing Unit). The processor 301 may be an integrated circuit chip with signal processing capabilities. The processor 301 can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor can be a microprocessor or any conventional processor.

[0132] Computer device 300 can be a mobile electronic device.

[0133] It should be understood that the systems, apparatuses, and methods described in this invention can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the couplings or direct couplings or communication connections shown or discussed may be indirect couplings or communication connections through some interfaces, apparatuses, or modules, and may be electrical, mechanical, or other forms.

[0134] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0135] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0136] The above are merely embodiments of this application and do not limit the scope of this patent application. Any equivalent structural or procedural changes made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of this application.

Claims

1. A method for predicting different stages of colorectal cancer progression, characterized in that, The method is performed by a computer and includes the following steps: The expression data of the first target protein in the sample to be tested is obtained, and the expression data of the first target protein is input into the first prediction model. Based on the output of the first prediction model, the subject is predicted to have colorectal cancer-related diseases. The first target protein is a combination of COL6A1 and F2. If the output of the first prediction model predicts that the subject has colorectal cancer-related disease, then the expression data of the second target protein is obtained, and the expression data of the second target protein is input into the second prediction model to predict whether the subject has colorectal adenoma, thus obtaining the subject's colorectal cancer progression stage; the second target protein is a combination of TRIT1 and ABCC2; Alternatively, if the output of the first prediction model predicts that the subject has colorectal cancer-related diseases, then the expression data of the third target protein is obtained, and the expression data of the third target protein is input into the third prediction model to predict whether the subject has colorectal cancer, thereby obtaining the subject's colorectal cancer progression stage; the third target protein is a combination of BAD, CTSL, FGL2 and MED1. The colorectal cancer-related diseases include colorectal adenoma, high-grade intraepithelial neoplasia of the colorectum, colorectal carcinoma in situ, and colorectal cancer.

2. The method according to claim 1, characterized in that, The steps for constructing the first prediction model are as follows: obtaining the expression data of the first target protein, which comes from people with colorectal cancer-related diseases and people without colorectal cancer-related diseases; inputting the expression data of the first target protein into a machine learning algorithm to construct the first prediction model.

3. The method according to claim 1, characterized in that, The construction steps of the second prediction model are as follows: obtain the expression data of the second target protein, which comes from people with colorectal adenoma and people with other stages of colorectal cancer progression, including high-grade intraepithelial neoplasia of the colorectal, colorectal carcinoma in situ and colorectal cancer; input the expression data of the second target protein into the machine learning algorithm to construct the second prediction model.

4. The method according to claim 1, characterized in that, The construction steps of the third prediction model are as follows: obtaining the expression data of the third target protein, which comes from people with colorectal cancer and people with other stages of colorectal cancer progression, including colorectal adenoma, high-grade intraepithelial neoplasia of the colorectal gland, and colorectal carcinoma in situ; inputting the expression data of the third target protein into a machine learning algorithm to construct the third prediction model.

5. A computer device, characterized in that, The device includes: a memory and a processor; the memory is used to store a computer program; the processor executes the computer program to implement the method of any one of claims 1-4.

6. A computer-readable storage medium, characterized in that, It stores a computer program thereon, which, when executed by a processor, implements the method as described in any one of claims 1-4.

7. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the method described in any one of claims 1-4.

Images