Model for prognosis prediction of colorectal cancer and application thereof

Abstract

The invention discloses a model for prognosis prediction of colorectal cancer and application of the model, the model comprises biomarkers ARL6IP4, ATP2A1, CRYAB, ELFN2, MAP2, MAT1A, ORC1 and POU4F1, the prognosis of a subject can be accurately predicted according to the model, the model provides a reliable prognosis prediction marker for evaluation of a patient with colorectal cancer, and the model can be used for predicting the prognosis of the patient with colorectal cancer. The prognosis prediction capability of the colorectal cancer patient is improved, and the clinical application prospect is wide.

Description

Technical Field

[0001] The present invention belongs to the field of biomedicine and relates to a model for predicting disease prognosis, and in particular, to a model for predicting the prognosis of colorectal cancer and its application. Background Art

[0002] Colorectal cancer (CRC) is the most common malignancy of the digestive system and the second leading cause of cancer-related death worldwide. Despite significant progress in current treatments, the prognosis of CRC patients remains bleak. Colorectal cancer is highly heterogeneous and complex, with a variety of differences in its clinical manifestations and prognosis. According to the description of the American Joint Committee on Cancer, the main prognostic indicator for colorectal cancer is still the tumor-node-metastasis (TNM) staging system. However, there are also significant differences in clinical outcomes among colorectal cancer patients with the same TNM stage. Therefore, there is an urgent need to find more effective means and potential biomarkers to predict the prognosis of colorectal cancer.

[0003] Liquid-liquid phase separation (LLPS) is a phenomenon caused by weak interactions between biomolecules, leading to the formation of membraneless organelles, which are also called biomolecular condensates. Biomolecular condensates are the basis for the spatiotemporal coordination of intracellular biological activities and play a vital role in many life activities in organisms, including gene expression regulation, signal transduction, and regulation of cellular metabolism. The disruption of this precise spatiotemporal regulation may lead to devastating pathological consequences, such as tumors. LLPS plays a vital role in the occurrence and development of tumors. A study reported that SHP2 gene mutation directly confers its phase separation ability to promote MAPK activation, thereby promoting tumorigenesis. Other studies have found that DDX21 phase separation condensates target and activate MCM5, thereby promoting the activation of epithelial-mesenchymal transition (EMT), which contributes to the metastasis of CRC. More and more studies have begun to construct prognostic models for different cancer types based on LLPS-related genes (LRGs). However, a risk model associated with LRGs has not yet been established in CRC. Summary of the invention

[0004] In order to make up for the deficiencies of the prior art, the purpose of the present invention is to provide a risk scoring model associated with LRGs for predicting the prognosis of colorectal cancer.

[0005] In order to achieve the above object, the present invention adopts the following technical solution: The first aspect of the present invention provides a biomarker combination for predicting the prognosis of colorectal cancer.

[0006] Furthermore, the biomarker combination includes ARL6IP4, ATP2A1, CRYAB, ELFN2, MAP2, MAT1A, ORC1 and POU4F1.

[0007] In the present invention, the term "biomarker" means a compound, preferably a gene, which is differentially present (i.e., increased or decreased) in a biological sample from a subject or a group of subjects with a first phenotype (e.g., with a disease) compared to a biological sample from a subject or a group of subjects with a second phenotype (e.g., without a disease). The term "biomarker" generally refers to the presence / concentration / content of a gene or the presence / concentration / content of two or more genes. In the present invention, the biomarkers include ARL6IP4, ATP2A1, CRYAB, ELFN2, MAP2, MAT1A, ORC1 and POU4F1. In the present invention, the biomarkers include genes and proteins encoded therein and homologs, mutations, and isoforms thereof. The term encompasses full-length, unprocessed markers, as well as markers in any form derived from processing in cells. The term encompasses naturally occurring variants (e.g., splice variants or allelic variants) of markers.

[0008] The second aspect of the present invention provides the use of a reagent for detecting the expression level of the biomarker combination described in the first aspect of the present invention in a sample in the preparation of a product for colorectal cancer prediction.

[0009] Furthermore, the reagents include reagents for detecting the expression amount of the biomarker combination in a sample through nucleic acid sequencing technology, nucleic acid hybridization technology, chromatography technology, mass spectrometry technology, digital imaging technology, protein immunoassay technology, dye technology, and / or second-generation sequencing technology.

[0010] Furthermore, the reagents include reagents for detecting the mRNA expression level of the biomarker combination and / or reagents for detecting the protein expression level of the biomarker combination.

[0011] Preferably, the reagent comprises a reagent for detecting the level of cDNA complementary to mRNA transcribed from the biomarker combination.

[0012] Preferably, the reagent is selected from a primer or a probe.

[0013] Furthermore, the reagents include reagents for detecting the level of polypeptides or proteins encoded by the biomarker combination.

[0014] Preferably, the agent is selected from an antibody, an antibody fragment or an affinity protein.

[0015] Furthermore, the primers described in the present invention can be prepared by chemical synthesis methods well known to those skilled in the art, and can be appropriately designed by referring to known information using methods well known to those skilled in the art, and prepared by chemical synthesis.

[0016] Further, the probe described in the present invention can be prepared by chemical synthesis, appropriately designed by referring to known information using methods known to those skilled in the art, and prepared by chemical synthesis, or can be prepared by preparing a gene containing a desired nucleic acid sequence from a biological material and amplifying it using primers designed to amplify the desired nucleic acid sequence.

[0017] Further, the probe hybridized with the nucleic acid sequence of the gene can be DNA, RNA, DNA-RNA chimera, PNA or other derivatives. The length of the probe is not limited, as long as it completes specific hybridization and specifically binds to the target nucleotide sequence, any length is acceptable. The length of the probe can be as short as 25, 20, 15, 13 or 10 bases. Similarly, the length of the probe can be as long as 60, 80, 100, 150, 300 base pairs or longer, or even the entire gene. Since different probe lengths have different effects on hybridization efficiency and signal specificity, the length of the probe is usually at least 14 base pairs, and the longest is generally no more than 30 base pairs. The length complementary to the target nucleotide sequence is best with 15-25 base pairs. The probe self-complementary sequence is preferably less than 4 base pairs to avoid affecting the hybridization efficiency.

[0018] The third aspect of the present invention provides a product for predicting the prognosis of colorectal cancer.

[0019] Furthermore, the product includes a reagent for detecting the expression level of the biomarker combination described in the first aspect of the present invention.

[0020] Furthermore, the reagents include reagents for detecting the mRNA expression level of the biomarker combination and / or reagents for detecting the protein expression level of the biomarker combination.

[0021] Preferably, the reagent is selected from a primer, a probe, an antibody, an antibody fragment, or an affinity protein.

[0022] Preferably, the product further comprises a total RNA extraction reagent, a reverse transcription reagent, and / or a second-generation sequencing reagent.

[0023] Furthermore, the expression level of the biomarker combination described in the present invention can be detected by measurement methods known in the art, including (but not limited to) methods for detecting the amount of RNA transcripts of genes in the biomarker combination or the amount of polypeptides encoded by genes in the biomarker combination.

[0024] Preferably, the RNA transcript of the gene can be converted into a cDNA complementary thereto by methods known in the art, and the amount of the RNA transcript can be obtained by measuring the amount of the complementary cDNA. The amount of the RNA transcript of the gene or the cDNA complementary thereto can be standardized for the amount of total RNA or total cDNA in the sample or for the amount of RNA transcripts of a group of housekeeping genes or the cDNA complementary thereto.

[0025] Preferably, the RNA transcript can be detected and quantified by methods such as hybridization, amplification or sequencing, including (but not limited to) hybridization of RNA transcripts with probes or primers, and methods for detecting the amount of RNA transcripts or their corresponding cDNA products by various quantitative PCR techniques or sequencing techniques based on polymerase chain reaction (PCR). The quantitative PCR techniques include (but not limited to) fluorescent quantitative PCR, real-time PCR or semi-quantitative PCR techniques. The sequencing techniques include (but not limited to) Sanger sequencing, second-generation sequencing, third-generation sequencing and single-cell sequencing, etc.

[0026] A fourth aspect of the present invention provides a risk scoring model for predicting the prognosis of colorectal cancer.

[0027] Furthermore, the risk scoring model includes the following biomarker combination: ARL6IP4, ATP2A1, CRYAB, ELFN2, MAP2, MAT1A, ORC1 and POU4F1.

[0028] Preferably, the risk scoring model uses the following regression equation to calculate the risk score: Risk Score = , where n is the number of genes used to predict prognosis, Exp i For each gene expression level, Coef i is the regression coefficient of each gene; when the risk score is lower than the median, it is a low risk, and when the risk score is higher than the median, it is a high risk.

[0029] A fifth aspect of the present invention provides a device or system for predicting the prognosis of colorectal cancer.

[0030] Further, the device or system comprises: Data acquisition module: used to obtain the biomarker combination expression spectrum data described in the first aspect of the present invention in the sample of the subject to be tested; Prediction module: used for providing the biomarker combination expression spectrum data obtained by the data acquisition module as input data to a trained prediction model, wherein the prediction model is trained to predict the prognosis of the subject based on the biomarker combination expression spectrum data of the subject; Prediction result output module: used to obtain the output result of the prediction model in the prediction module to obtain the prognosis prediction result of the subject.

[0031] Preferably, the prediction model is a Cox regression model.

[0032] More preferably, the Cox regression model is a LASSO Cox regression model.

[0033] Most preferably, the prediction model is the risk scoring model described in the fourth aspect of the present invention.

[0034] A sixth aspect of the present invention provides a computer device.

[0035] Furthermore, the computer device includes a memory and a processor, the memory stores a program, and the processor implements the following method when executing the program: Acquiring the biomarker combination expression profile data of the first aspect of the present invention in the sample of the test subject; Providing the biomarker combination expression profile data as input data to a trained prediction model; Output the prognosis prediction result of the subject to be tested.

[0036] Preferably, the prediction model is a Cox regression model.

[0037] More preferably, the Cox regression model is a LASSO Cox regression model; Most preferably, the prediction model is the risk scoring model described in the fourth aspect of the present invention.

[0038] Furthermore, the computer device of the present invention includes (but is not limited to) any personal computer, server or other terminal that can interact with the user through a keyboard, touchpad or voice control device. The computing device herein may also include a mobile terminal, which includes (but is not limited to) any electronic device that can interact with the user through a keyboard, touchpad or voice control device, such as a tablet computer, a smart phone, a personal digital assistant (PDA), a smart wearable device or other terminal. The network in which the computing device is located includes (but is not limited to) the Internet, a wide area network, a metropolitan area network, a local area network, and a virtual private network (VPN).

[0039] Further, the memory of the present invention includes non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. The code of the operating system is stored thereon. For example, the memory also stores codes or instructions, and by running these codes or instructions, the risk scoring model for colorectal cancer prognosis prediction provided by the embodiments disclosed herein can be implemented. Volatile memory may include random access memory (RAM) or external cache memory.

[0040] Furthermore, the computer device of the present invention may include a processor, a memory, an external interface, a display, and an input device connected via a system bus. The processor is used to provide computing and control capabilities. The display of the computer device may be a liquid crystal display or an electronic ink display, and the input device may be a touch layer covered on the display, or may be a key, trackball, or touchpad provided on the housing of the computing device, or may be an external keyboard, touchpad, or mouse, etc.

[0041] The processor may include one or more microprocessors or digital processors. The processor may call program codes stored in the memory to execute related functions. The processor is also called a central processing unit (CPU), which may be a large-scale integrated circuit, a computing core (Core) and a control core (Control Unit).

[0042] A seventh aspect of the present invention provides a computer-readable storage medium.

[0043] Furthermore, the computer-readable storage medium includes a stored computer program.

[0044] Among them, when the computer program is running, it controls the computer-readable storage medium to implement the method described in the sixth aspect of the present invention.

[0045] An eighth aspect of the present invention provides any one of the following applications.

[0046] Furthermore, the application includes: 1) Application of the biomarker combination described in the first aspect of the present invention in establishing a prognostic risk scoring model for colorectal cancer; 2) Application of the risk scoring model described in the fourth aspect of the present invention in predicting the sensitivity of colorectal cancer patients to chemotherapy drugs; wherein, when the risk score is low risk, the colorectal cancer patient is sensitive to chemotherapy drugs.

[0047] Preferably, the chemotherapy drugs include oxaliplatin, 5-fluorouracil (5-FU), irinotecan, dabrafenib, lapatinib, and ulixertinib.

[0048] A ninth aspect of the present invention provides the use of a reagent for detecting the expression level of ARL6IP4 in a sample in the preparation of a product for diagnosing colorectal cancer.

[0049] Furthermore, the product includes a chip, a test kit, a test paper or a high-throughput sequencing platform.

[0050] In the present invention, the kit comprises a reagent for detecting the expression level of ARL6IP4 in a sample. For example, the kit can be an RT-PCR kit, a DNA chip kit, an ELISA kit, a protein chip kit, a rapid kit, or an MRM (multiple reaction monitoring) kit.

[0051] For example, the diagnostic kit may further comprise elements necessary for reverse transcription polymerase chain reaction. The RT-PCR kit comprises a pair of primers specific for the gene encoding the ARL6IP4 protein. Each primer is a nucleotide having specificity for the nucleic acid sequence of the gene, and its length may be about 7 to 50 bp, more particularly about 10 to 39 bp. In addition, the kit may further comprise primers specific for the nucleic acid sequence of the control gene. In addition, the RT-PCR kit may comprise a test tube or suitable vessel, a reaction buffer (different pH values ​​and magnesium concentrations), deoxynucleotides (dNTPs), enzymes (e.g., Taq polymerase and reverse transcriptase), deoxyribonuclease inhibitors, ribonuclease inhibitors, DEPC-water, and sterile water.

[0052] In addition, the diagnostic kit of the present disclosure may include elements necessary for operating a DNA chip. The DNA chip kit may include a substrate bound to a gene or cDNA or an oligonucleotide equivalent to a fragment thereof, and reagents, agents and enzymes for constructing a fluorescently labeled probe. In addition, the substrate may include a control gene or cDNA or an oligonucleotide equivalent to a fragment thereof.

[0053] In some embodiments, the kit of the present disclosure may include elements necessary for performing ELISA. The ELISA kit may include antibodies specific for proteins. The antibodies have high selectivity and affinity for marker proteins, no cross-reactivity with other proteins, and may be monoclonal antibodies, polyclonal antibodies, or recombinant antibodies. In addition, the ELISA kit may include antibodies specific for control proteins. In addition, the ELISA kit may further include reagents capable of detecting the bound antibodies, for example, labeled second antibodies, chromophores, enzymes (e.g., conjugated with antibodies), and substrates thereof, or substances capable of binding the antibodies.

[0054] In addition, the present invention also provides a method for constructing a risk scoring model for colorectal cancer prognosis prediction according to the fourth aspect of the present invention, the construction method comprising the following steps: 1) Candidate gene analysis: Screening LLPS-related genes (LRGs) in the cancer expression profile database; 2) Model building: Using the LRGs obtained in step 1) of the LASSO Cox analysis, the genes constituting the prognostic gene signature were finally determined to construct the model, and the genes included ARL6IP4, ATP2A1, CRYAB, ELFN2, MAP2, MAT1A, ORC1, and POU4F1.

[0055] Furthermore, the construction method also includes verifying the effectiveness of the model.

[0056] Compared with the prior art, the present invention has the following advantages and beneficial effects: The present invention provides a novel model for predicting the prognosis of colorectal cancer, and the stability and effectiveness of the model are confirmed by a validation data set of colorectal cancer patients. The model provides reliable biomarkers for the prognosis evaluation of colorectal cancer patients, improves the ability to predict the prognosis of colorectal cancer patients, can effectively identify colorectal cancer patients with high-risk prognosis, and can perform early monitoring and effective intervention in clinical practice, thereby reducing the incidence and mortality of adverse prognosis in colorectal cancer patients and improving the prognosis of colorectal cancer patients. The model has broad prospects for clinical application. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 Design a schematic diagram for this study; Figure 2Figure 1 is a diagram of differential expression analysis and functional annotation of LRGs; Figure A is a Venn diagram showing the intersection of genes identified in the DrLLPS database and TCGA-CRC; Figure B is a heat map of LRGs expression changes in CRC based on the adjusted p-value <0.05, |log2FC|>1; Figure C is a PCA analysis of differentially expressed LRGs; Figure D is a GO enrichment analysis of differentially expressed LRGs in terms of biological process (BP), cellular component (CC) and molecular function (MF); Figure E is a KEGG enrichment analysis of differentially expressed LRGs; Figure 3 To construct the prognostic risk model diagram associated with LLPS; Figure A is the univariate Cox regression analysis of LRGs in TCGA-CRC; Figures B and C determine 16 prognostic genes by LASSO regression and the optimal parameters obtained by screening for prognostic differential LRGs; Figure D performs multivariate Cox regression analysis on the prognostic genes obtained from the LASSO regression analysis; Figure EG, PCA shows the distribution of low risk and high risk among all genes, LRGs, and risk genes; Figure 4 Figure 1 is the prognostic performance and validation correlation diagram of LRGs risk signature; Figure AC is the KM curve showing the OS of two risk groups in TCGA-CRC in the training (A), test (B) and whole (C) cohorts; Figure DF is the risk score, survival status (red dots indicate death, blue dots indicate survival) and gene expression distribution of model genes in the training (D), test (E) and whole (F) cohorts; Figure GI is the ROC curve used to evaluate the sensitivity and specificity of the 1-year, 3-year and 5-year survival rates of the risk score in the training (G), test (H) and whole (I) cohorts; Figure J is the KM curve showing the OS of the two risk groups in GSE39582; Figure K is the KM curve showing the RFS of the two risk groups in GSE39582; Figure L is the KM curve showing the prognosis of tumor stage I-II subgroups; Figure 5 Figure 1 is a graph showing the correlation between ROC curve analysis, Nomogram construction, and risk model of clinical characteristics; Figure A is the prognostic ROC curve of risk score and other clinical factors; Figure B is a nomogram developed to predict the 1-, 3-, and 5-year survival rates of colorectal cancer patients; Figure C is a calibration curve used to evaluate the accuracy of the nomogram; Figure D is a C-index curve used to evaluate the consistency index of the nomogram; Figure E is a heat map showing the relationship between the expression levels of 8 genes and various clinical pathological characteristics; Figure 6Figure 1 is a correlation diagram between risk characteristics and immune cell infiltration; A is a differential analysis diagram of immune cell infiltration between two risk groups; B is a correlation analysis diagram between infiltrating immune cell types and risk scores; C is a bar graph of the distribution of infiltrating immune cells between high and low risk groups; D is a correlation analysis diagram between infiltrating immune cell types and eight LRGs; Figure 7 Figure 1 is a graph related to immune characteristics and drug sensitivity analysis; A is a graph showing the difference in matrix, immune and ESTIMATE scores between the two risk groups; B is a graph showing the difference in immune function between the two risk groups; CE is a graph showing the comparison of TIDE scores, rejection scores and dysfunction scores between the two risk groups; FK is a graph showing the difference in drug sensitivity between the two groups of patients to Oxaliplatin (F), 5-FU (G), Irinotecan (H), Dabrafenib (I), Lapatinib (J) and Ulixertinib (K); Figure 8 It is a correlation diagram between risk characteristics and TMB; A and B show the top 15 mutant genes in the high-risk group and low-risk group, respectively; C is the KM analysis diagram of the high TMB group and the low TMB group; D is the KM curve under the combined influence of risk score and TMB; Fig. 9 The figure shows the expression level of ARL6IP4 in colorectal cancer and its phase separation ability; A shows the mRNA level of ARL6IP4 in colorectal cancer and normal tissues from the TCGA database; B shows the comparison of ARL6IP4 mRNA levels in 10 pairs of colorectal cancer tissues and normal tissues; C shows the three-dimensional reconstruction image of HCT116 living cells transfected with GFP-ARL6IP4 plasmid obtained by confocal laser scanning microscopy; D shows the image (upper) and quantitative analysis (lower) of GFP-ARL6IP4 FRAP, data are expressed as Mean±SE, n=3 droplets; E shows the fusion of two GFP-ARL6IP4 droplets to form a larger droplet; Fig.10 : is the ROC curve diagram of ARL6IP4, wherein A is the ROC curve diagram of ARL6IP4 in the database, and B is the ROC curve diagram of ARL6IP4 in clinical samples. DETAILED DESCRIPTION

[0058] The present invention is further described below in conjunction with specific embodiments, which are only used to explain the present invention and cannot be understood as limiting the present invention. It can be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and purpose of the present invention, and the scope of the present invention is defined by the claims and their equivalents. The experimental methods used in the following embodiments are conventional methods unless otherwise specified; the reagents, biological materials, etc. used in the following embodiments, unless otherwise specified, can all be obtained from commercial sources.

[0059] Example 1 Establishment of a risk scoring model for colorectal cancer prognosis prediction 1. Experimental Methods 1. Experimental data source: RNA sequencing data, corresponding clinicopathological characteristics, and mutation data of CRC were retrieved from the TCGA website (https: / / portal.gdc.cancer.gov / ), which contains 571 tumor samples and 44 adjacent normal samples.

[0060] The GSE39582 dataset containing 566 CRC samples was downloaded from the GEO website (https: / / www.ncbi.nlm.nih.gov / geo / ) for the validation cohort. The list of LRGs was selected from DrLLPS (http: / / llps.biocuckoo.cn / ), an online database containing 150 scaffold proteins that are drivers of LLPS, 987 regulators that help regulate LLPS, and 8148 potential target proteins that may affect the formation of membrane-bound compartments. 3611 LRGs identified in Homo sapiens were screened for subsequent studies.

[0061] 2. Differential expression analysis and functional annotation of LRGs: A Venn diagram was used to display the intersection genes between TCGA-CRC and LRGs. The limma package was used to screen the differentially expressed genes between colorectal cancer and adjacent normal samples, with |log2FC|>1 and false discovery rate (FDR)<0.05 as filtering criteria. Heat map analysis was generated using ggplot2 and pheatmap packages. Principal component analysis (PCA) was used to visualize the distribution of normal and tumor samples. The potential functions and pathways of LRGs were explored using GO and KEGG.

[0062] 3. Construction and validation of the LRGs risk model: The expression data were log2 transformed for subsequent analysis; the R package “sva” was used to adjust the batch effect between TCGA-CRC and GSE39582. By excluding genes that were not present in both the GEO and TCGA databases, we screened 773 differentially expressed LRGs. Patients were randomly divided into a training cohort and a test cohort in a 1:1 ratio. Then, we evaluated whether there were statistically significant differences in the clinical characteristics of the two groups of patients (including age, sex, tumor stage, T, N, and M). In the training cohort, univariate Cox regression was used to select LRGs with prognostic value. In the univariate Cox regression analysis, p < 0.01 was considered statistically significant. To avoid overfitting, we used the R package “glmnet” to perform the least absolute shrinkage and selection operator (LASSO) Cox regression, followed by multivariate Cox regression for further screening of prognostic LRGs. Finally, we constructed a new prognostic signature using 8 genes. We calculated the risk score for each sample using the following formula: Risk score = . All patients were divided into low-risk and high-risk groups according to the median risk score calculated from the training cohort. To further evaluate the feasibility, GSE39582 was used as an external validation cohort, and the risk score calculation method was consistent with that in the training group. In addition, principal component analysis (PCA) was used to study the distribution of risk scores. To determine the prognosis of the two groups of patients, Kaplan-Meier (KM) analysis was performed using the R packages “survival” and “survminer”. Receiver operating characteristic (ROC) analysis was used to verify the predictive specificity and sensitivity of the risk model, and the area under curves (AUC) was calculated.

[0063] 4. Correlation between LRGs risk model and clinical factors: In order to evaluate the prognostic value of the risk score, univariate and multivariate Cox regression analysis was used to screen independent risk factors that were significantly associated with survival (p<0.05). The ROC curve was used to evaluate the sensitivity and specificity of the risk score and clinical factors through the "timeROC" package. In the process of constructing the nomogram, we used the "survival" and "rms" packages, and included gender, risk score, age, and tumor stage as variables in the analysis. We drew calibration curves to evaluate the consistency between the actual and predicted probabilities of 1-year, 3-year, and 5-year overall survival (OS). The C-index was used to evaluate the predictive ability of the nomogram.

[0064] 2. Experimental Results 1. Differential expression analysis and functional annotation of LRGs Our research flow chart is as follows Figure 1 As shown. The genes in the TCGA-CRC cohort were intersected with the LRGs in the DrLLPS database to obtain 3530 LRGs ( Figure 2 A). Then, in the TCGA-CRC cohort, a total of 3526 genes with expression data were screened for subsequent differential expression analysis between colorectal cancer samples and normal samples. Based on the thresholds of FDR < 0.05 and |log2FC| > 1, we identified 828 differentially expressed LRGs, of which 299 were downregulated and 529 were upregulated. The heatmaps show the top 50 upregulated and downregulated differentially expressed LRGs, respectively ( Figure 2 B). Based on these differentially expressed LRGs, PCA showed significant differences between normal samples and colorectal cancer samples ( Figure 2 C). We performed GO and KEGG pathway annotation analysis to reveal the potential biological functions of differentially expressed LRGs in CRC. GO analysis showed that these differentially expressed LRGs were related to the processes of LLPS, such as the assembly of membrane-free organelles ( Figure 2 D). KEGG analysis showed that these differentially expressed LRGs were related to tumor-related pathways, such as cell cycle, DNA replication, HIF-1 signaling pathway, platinum resistance, p53 signaling pathway, etc. ( Figure 2 E).

[0065] 2. Construction and validation of a predictive risk model based on LRGs The CRC samples in the TCGA database were randomly divided into a training group (50%, n=271 cases) and a validation group (50%, n=271 cases). In addition, the clinical characteristics between the two groups were compared, and the results showed no difference. Next, in order to construct a new prognostic signature, we performed univariate Cox regression analysis in the training cohort to preliminarily screen genes associated with prognosis. 43 genes were identified as prognostic genes for colorectal cancer, of which 10 genes were considered protective genes with HR<1, and the remaining 33 genes were considered risk genes with HR>1 ( Figure 3 A). In addition, we perform LASSO Cox regression to reduce overfitting ( Figure 3 B, C), followed by multivariate Cox regression analysis ( Figure 3 D). We obtained 8 genes (ARL6IP4, ATP2A1, CRYAB, ELFN2, MAP2, MAT1A, ORC1, and POU4F1) associated with OS in colorectal cancer patients and used them to construct a prognostic signature. Then, the risk score for each colorectal cancer patient was calculated according to the following algorithm: Risk score = , Coef i and Exp iRepresent the coefficient and expression level associated with each risk gene, respectively. Therefore, we can calculate the risk score of each sample by applying the above formula. According to the median risk score of the training cohort, the patients were divided into high-risk group and low-risk group. We analyzed the differential expression of 8 prognostic signature genes in the high-risk group and the low-risk group. PCA showed that all genes in TCGA-CRC of the patients ( Figure 3 E), LRGs ( Figure 3 F) and risk genes ( Figure 3 G). It is particularly noteworthy that colorectal cancer patients are distributed in significantly different directions according to different risk genes, which suggests that risk signatures may effectively distinguish high-risk and low-risk groups. KM analysis showed that the OS ( Figure 4 A) and PFS were significantly lower than those in the low-risk group. The distribution of risk scores and survival status showed that in the training cohort, as the risk score increased, the mortality rate also increased. The heat map shows the expression of 8 genes in the high-risk group and the low-risk group ( Figure 4 D). We then used the same formula to calculate a risk score for each patient in the validation cohort and for the entire cohort, which was used to verify the validity of this risk signature. The results were similar to those observed in the training group ( Figure 4 B, C, E, F). In addition, the ROC curve was used to evaluate the predictive ability of the model. In the training cohort, the AUCs at 1, 3, and 5 years were 0.863, 0.840, and 0.752, respectively ( Figure 4 G). For the validation cohort, the AUCs at 1, 3, and 5 years were 0.669, 0.676, and 0.648, respectively ( Figure 4 H), in the entire cohort, the AUCs were 0.748, 0.749, and 0.700, respectively ( Figure 4 I). These results highlight the superior predictive accuracy of our model. In addition, we introduced the GEO dataset (GSE39582) as an external validation cohort. The KM curve of GSE39582 showed that the high-risk group had a worse prognosis than the low-risk group, which was consistent with the prognostic results observed in the TCGA cohort ( Figure 4 JK). In addition, to evaluate the effectiveness of the LRGs risk model in predicting the prognosis of patients with colorectal cancer, we divided the entire patients into different subgroups according to clinical characteristics, including tumor stage I-II or tumor stage III-IV, male or female, young (≤65 years old) or old (>~65 years old), N0 or N1-N2, M0 or M1. These results confirmed the effectiveness and robustness of the risk model in predicting the survival outcomes of patients with colorectal cancer.

[0066] 3. Independent prognostic significance and clinical characteristics of the risk model: Univariate Cox regression analysis showed that risk score, age, tumor stage, T, N, and M stages were important prognostic factors affecting OS in patients with colorectal cancer. In addition, multivariate analysis showed that the risk score was an independent prognostic indicator for OS (HR 1.030, 95%CI 1.019-1.041, p<0.001). In addition, ROC analysis showed that the AUC of the risk score was 0.749, which was superior to tumor stage (AUC=0.720) in predicting markers ( Figure 5 A). We created a prognostic nomogram incorporating sex, risk score, age, and tumor stage ( Figure 5 B). The calibration curves show that the predictions of the nomogram for 1-year, 3-year, and 5-year OS probabilities are very consistent with the ideal predictions ( Figure 5 C). The C-index value of the nomogram was significantly higher than that of other clinical characteristics ( Figure 5 D). These results indicate that the nomogram is effective in predicting patient prognosis and can be used as a tool for clinical decision making. In addition, the risk scores grouped by tumor stage, T stage, N stage, and M stage were also different. The heat map shows the correlation between the expression levels of the eight LRGs in the TCGA database and the clinicopathological characteristics of CRC patients ( Figure 5 E).

[0067] Example 2 Relationship between risk score, immune characteristics and drug sensitivity Experimental methods 1. Analysis of immune characteristics and tumor mutation load: The CIBERSORT algorithm was used to evaluate the differences in 22 immune cell types between high-risk and low-risk groups. In addition, correlation analysis was used to explore the relationship between immune cell types and characteristic genes. The "estimate" R package was used to calculate matrix, immune, and estimate scores. We then compared the differences between the two risk groups. We used the "maftools" package to draw a waterfall chart to show the mutation landscape in our samples. We used the "survival" package to explore the relationship between TMB, risk score, and patient survival.

[0068] 2. Immunotherapy response and drug sensitivity: The Tumor Immune Dysfunction and Exclusion (TIDE) algorithm (http: / / tide.dfci.harvard.edu / ) was used to evaluate the responsiveness of the two groups of patients to immune checkpoint blockade (ICB) therapy. The "oncoppredict" package in R was used to analyze the difference in drug sensitivity between the two groups. Detailed data were obtained from the Genomics of Cancer Drug Sensitivity (GDSC) database (https: / / www.cancerrxgene.org / ).

[0069] 2. Experimental Results 1. Relationship between risk score and immune cell infiltration It is widely recognized that the tumor immune microenvironment plays a key role in regulating cancer progression, clinical outcomes, and treatment responses. To explore the correlation between immune cell signatures and risk scores, we first evaluated the relative scores of 22 tumor-infiltrating immune cells in each CRC sample by the CIBERSORT algorithm. We observed more infiltration of memory B cells and T regulatory cells (Tregs) in the high-risk group compared with the low-risk group, but fewer CD8+ T cells, M1 macrophages, and resting dendritic cells in the high-risk group ( Figure 6 A). Spearman correlation analysis showed that the risk score was negatively correlated with resting dendritic cells, M1 macrophages, and CD8+ T cells, but positively correlated with memory B cells and Tregs ( Figure 6 B). The composition of each typical immune cell Figure 6 C. In addition, Figure 6 D shows that these eight genes are all associated with immune cell infiltration, suggesting that these genes may serve as potential targets for CRC immunotherapy.

[0070] 2. Relationship between risk characteristics and immune characteristics and drug sensitivity The ESTIMATE algorithm showed that the interstitial score and ESTIMATE score of the high-risk group were higher than those of the low-risk group, while there was no significant difference in the immune score between the two groups ( Figure 7 A), indicating that stromal cell infiltration increased with the increase of risk score. The difference analysis of immune-related functions showed that the APC co-inhibition score, cytolytic activity score, pro-inflammatory score, MHC class I score and T cell co-inhibition score were lower in the high-risk group ( Figure 7 B). In addition, the TIDE score in the high-risk group was higher than that in the low-risk group ( Figure 7 CE). In addition, we investigated the differences in drug sensitivity between the two risk groups. In terms of chemotherapy drugs used for colorectal cancer, oxaliplatin, 5-FU, and irinotecan were more effective in the low-risk group than in the high-risk group ( Figure 6 FH). For targeted therapy of colorectal cancer, the low-risk group is more sensitive to dabrafenib, lapatinib, and ulixertinib than the high-risk group ( Figure 6 In addition, differences in sensitivity to an additional 101 drugs were observed between the two risk groups.

[0071] 3. Correlation between risk characteristics and TMB We analyzed the mutation of CRC in high-risk and low-risk groups. Maftools analysis results showed the top 15 high-frequency mutation genes in high-risk and low-risk groups ( Figure 8 A, B). We found that the high-risk group had a higher mutation rate compared with the low-risk group. In particular, the most commonly mutated gene was APC. Previous studies have shown that high TMB predicts a poor prognosis in various types of cancer. When evaluating the prognostic value of TMB, we observed that there was no statistical difference in survival expectancy between patients in the high-TMB group and the low-TMB group, but the high-TMB group tended to have a worse prognosis ( Figure 8 C). Next, colorectal cancer patients were divided into four groups according to TMB and risk scores for survival analysis, including low TMB + low risk group, low TMB + high risk group, high TMB + low risk group and high TMB + high risk group. The results showed that the survival outcomes of patients in the low TMB + low risk group were better than those in the other three groups ( Figure 8 D).

[0072] Example 3 Expression level and phase separation ability of ARL6IP4 in colorectal cancer 1. Experimental Methods 1. Clinical specimens, RNA extraction, and qRT-PCR: Ten pairs of colorectal cancer tissues and corresponding adjacent normal tissues (ANTs) were collected from colorectal cancer patients who underwent radical resection or palliative resection in the Department of Gastrointestinal Surgery, Affiliated Hospital of Xuzhou Medical University as a cohort, and consent was obtained from all patients. Total RNA was extracted from the cohort using RNA Isolater Total RNA Extraction Reagent (Vazyme, Nanjing, China). 1000 ng of RNA was reverse transcribed using the SweScript RT I First-Strand cDNA Synthesis Kit (Servicebio, Wuhan, China). cDNA analysis was performed by real-time quantitative polymerase chain reaction (qRT-PCR) using ChamQ SYBR qPCR Master Mix (Vazyme, Nanjing, China) on a LightCycler 96 system (Roche, Switzerland). GAPDH was used as an endogenous control. Primer information: ARL6IP4 F: TGACCAAGGAGGAGTGGGAT, ARL6IP4 R: RCCTCTAGGACCTCGCCATCT.

[0073] 2. Phase separation experiment of colorectal cancer cells: HCT116 cells were cultured in glass-bottomed culture dishes and subsequently transfected with ARL6IP4-GFP plasmid. GFP-labeled spots were observed using a confocal laser scanning microscope (CLSM, Leica STELLARIS 5, Germany). Fluorescence recovery after photobleaching (FRAP) was performed using a full-power 488 nm laser. Time-lapse images were captured after photobleaching to monitor the recovery of fluorescence intensity every second for a total duration of 40 seconds. GraphPadPrism software was used to quantitatively analyze and graphically represent the data.

[0074] 2. Experimental Results To investigate whether the genes that constitute the risk signature of this study are phase-separated, we selected ARL6IP4, which had the highest coefficient in multivariate Cox regression analysis, for experimental validation. First, upregulation of ARL6IP4 was confirmed in the TCGA-CRC cohort ( Fig. 9 A). In addition, we evaluated the expression level of ARL6IP4 in 10 pairs of fresh-frozen tissue specimens using quantitative qRT-PCR. The results showed that ARL6IP4 was significantly upregulated in most CRC tissues compared with ANTs ( Fig. 9 B). Subsequently, after ectopic transfection of ARL6IP4-GFP plasmid, we used 3D confocal microscopy to observe that ARL6IP4 formed droplet-like structures in living CRC cells ( Fig. 9 C). To assess the mobility of Arl6IP4 condensates, we performed FRAP experiments. The results showed that fluorescence recovery in Arl6IP4 condensates is rapid, occurring within approximately 40 seconds after photobleaching ( Fig. 9 D). In addition, through time-lapse imaging, we observed that the punctate structures exhibited fusion behavior in living CRC cells ( Fig. 9 E). In conclusion, our results indicate that Arl6IP4 is upregulated and phase separated in CRC.

[0075] Furthermore, we performed ROC analysis on tumors (n=647) and normal tissues (n=51) in the TCGA database to analyze the diagnostic value of ARL6IP4 in colorectal cancer patients and found that its AUC was 0.839 ( Fig.10 A); Further, the ROC analysis of clinical cohort tumors (n=40) and normal tissues (n=40) was performed to verify the diagnostic value of ARL6IP4 in colorectal cancer patients, and it was found that its AUC was 0.734 ( Fig.10 B). Taken together, these results suggest that ARL6IP4 has good diagnostic efficacy.

[0076] The description of the above embodiments is only used to understand the method and core idea of ​​the present invention. It should be pointed out that, for those skilled in the art, several improvements and modifications can be made to the present invention without departing from the principle of the present invention, and these improvements and modifications will also fall within the scope of protection of the claims of the present invention.

Claims

1. A biomarker combination for predicting the prognosis of colorectal cancer, characterized in that: The biomarker combination includes ARL6IP4, ATP2A1, CRYAB, ELFN2, MAP2, MAT1A, ORC1 and POU4F1.

2. Use of a reagent for detecting the expression of the biomarker combination described in claim 1 in a sample in the preparation of a product for predicting colorectal cancer.

3. The use according to claim 2, characterized in that: The reagents include reagents for detecting the expression amount of the biomarker combination in a sample by nucleic acid sequencing technology, nucleic acid hybridization technology, chromatography technology, mass spectrometry technology, digital imaging technology, protein immunoassay technology, dye technology, and / or next-generation sequencing technology; Preferably, the reagent includes a reagent for detecting the mRNA expression amount of the biomarker combination and / or a reagent for detecting the protein expression amount of the biomarker combination; Preferably, the reagents include reagents for detecting the level of cDNA complementary to mRNA transcribed from the biomarker combination; Preferably, the reagent is selected from a primer or a probe; Preferably, the reagent comprises a reagent for detecting the level of a polypeptide or protein encoded by the biomarker combination; Preferably, the agent is selected from an antibody, an antibody fragment or an affinity protein.

4. A product for predicting the prognosis of colorectal cancer, characterized in that: The product includes a reagent for detecting the expression amount of the biomarker combination of claim 1; Preferably, the reagent includes a reagent for detecting the mRNA expression amount of the biomarker combination and / or a reagent for detecting the protein expression amount of the biomarker combination; Preferably, the reagent is selected from a primer, a probe, an antibody, an antibody fragment, or an affinity protein; Preferably, the product further comprises a total RNA extraction reagent, a reverse transcription reagent, and / or a second-generation sequencing reagent.

5. A risk scoring model for colorectal cancer prognosis prediction, characterized in that: The risk scoring model includes the following biomarker combination: ARL6IP4, ATP2A1, CRYAB, ELFN2, MAP2, MAT1A, ORC1, and POU4F1; Preferably, the risk scoring model uses the following regression equation to calculate the risk score: Risk Score = , where n is the number of genes used to predict prognosis, Expi is the expression level of each gene, and Coefi is the regression coefficient of each gene; when the risk score is lower than the median, it is a low risk, and when the risk score is higher than the median, it is a high risk.

6. A device or system for predicting the prognosis of colorectal cancer, characterized in that: The device or system comprises: Data acquisition module: used to obtain the biomarker combination expression spectrum data of claim 1 in the sample of the subject to be tested; Prediction module: used for providing the biomarker combination expression profile data obtained by the data acquisition module as input data to a trained prediction model, wherein the prediction model is trained to predict the prognosis of the subject based on the biomarker combination expression profile data of the subject; Prediction result output module: used to obtain the output result of the prediction model in the prediction module to obtain the prognosis prediction result of the subject; Preferably, the prediction model is a Cox regression model; More preferably, the Cox regression model is a LASSO Cox regression model; Most preferably, the prediction model is the risk scoring model described in claim 5.

7. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores a program, and the processor implements the following method when executing the program: Obtaining the biomarker combination expression profile data of claim 1 in the sample of the subject to be tested; Providing the biomarker combination expression profile data as input data to a trained prediction model; Outputting the prognosis prediction result of the subject to be tested; Preferably, the prediction model is a Cox regression model; More preferably, the Cox regression model is a LASSO Cox regression model; Most preferably, the prediction model is the risk scoring model described in claim 5.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program; Wherein, when the computer program is running, it controls the computer-readable storage medium to implement the method of claim 7.

9. Any of the following applications, characterized in that: The applications include: 1) Use of the biomarker combination of claim 1 in establishing a prognostic risk scoring model for colorectal cancer; 2) Use of the risk scoring model of claim 5 in predicting the sensitivity of colorectal cancer patients to chemotherapy drugs; wherein, when the risk score is low risk, the colorectal cancer patient is sensitive to chemotherapy drugs; Preferably, the chemotherapy drugs include Oxaliplatin, 5-FU, Irinotecan, Dabrafenib, Lapatinib, and Ulixertinib.

10. Use of a reagent for detecting the expression level of ARL6IP4 in a sample in the preparation of a product for diagnosing colorectal cancer.

Images