Three-gene biomarkers for colorectal cancer, detection reagents and their applications
By jointly detecting the three gene biomarkers for colorectal cancer, ARRDC4, CBX3, and MMP1, a risk scoring model was constructed, which solved the problems of insufficient sensitivity and specificity of existing colorectal cancer screening methods and achieved efficient colorectal cancer diagnosis and risk assessment.
Patent Information
- Application Number
- CN202611595332.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-15
- Publication Date
- 2026-11-17
AI Technical Summary
Existing colorectal cancer screening methods rely on highly invasive colonoscopy and non-invasive but limited-sensitivity fecal immunochemical testing. Furthermore, molecular diagnostic methods for single genes lack sufficient sensitivity and specificity, making it difficult to meet the needs of large-scale population screening.
The combined detection of three gene biomarkers for colorectal cancer, ARRDC4, CBX3 and MMP1, was used. The mRNA expression level was detected by specific primer pairs, and a risk scoring model was constructed for the diagnosis and risk assessment of colorectal cancer.
It achieves highly sensitive and specific colorectal cancer diagnosis, reduces missed diagnoses and unnecessary colonoscopies, provides personalized risk assessment, and is applicable to multiple detection platforms.
Smart Images

Figure CN123108152A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of molecular diagnostics, and more specifically to three-gene biomarkers for colorectal cancer and their applications. Background Technology
[0002] Colorectal cancer, also known as colorectal malignant tumors, is a malignant tumor originating from the epithelial cells of the colon or rectum. The 5-year survival rate for colorectal cancer confined to the intestinal wall can exceed 90%. [1] However, the 5-year survival rate drops to below approximately 15% after distant metastasis. Therefore, effective diagnosis and intervention before tumor development and metastasis are key strategies for reducing the disease burden, improving patient survival rates, and enhancing quality of life.
[0003] Currently, colorectal cancer screening mainly relies on colonoscopy and fecal immunochemical testing (FIT). Although colonoscopy is considered the gold standard for diagnosis, its invasiveness, high cost, and limited patient compliance make it difficult to meet the needs of large-scale population screening. While non-invasive methods such as FIT are simple to perform, their sensitivity for precancerous lesions and tumors remains limited. Previous studies have reported that CEA has limited value in detecting colorectal cancer and can be elevated in individuals with inflammatory bowel disease, liver disease, pancreatitis, and smokers. [2] A systematic review and meta-analysis showed that, at a reference concentration of 5 mg / L, CEA had a pooled sensitivity of 59% (range 33%–83%) and a pooled specificity of 89% (range 58%–97%) for colorectal cancer. [3] .
[0004] Furthermore, existing molecular diagnostic methods are mostly based on differential expression of single genes. However, the development and progression of colorectal cancer involves abnormalities in multiple signaling pathways, and changes in the expression of a single gene may be affected by various factors, leading to insufficient diagnostic sensitivity or specificity.
[0005] Therefore, there is an urgent need to develop a non-invasive, highly sensitive, and highly specific diagnostic method for colorectal cancer based on the joint detection of multiple functionally complementary genes. Summary of the Invention
[0006] This invention provides three gene biomarkers for colorectal cancer and their applications.
[0007] To solve the above-mentioned technical problems, this application adopts the following technical solution:
[0008] Three gene biomarkers for colorectal cancer, including ARRDC4, CBX3, and MMP1.
[0009] A second aspect of the present invention aims to provide any of the following applications of the above-mentioned biomarkers: A. Application in the preparation of products for the diagnosis of colorectal cancer; B. Application in the preparation of products for colorectal cancer screening; C. Application in the preparation of products for assessing the risk of colorectal cancer.
[0010] A third aspect of the present invention aims to provide any of the following applications of a detection reagent for detecting the expression levels of the aforementioned biomarkers: A. Application in the preparation of products for the diagnosis of colorectal cancer; B. Application in the preparation of products for colorectal cancer screening; C. Application in the preparation of products for assessing the risk of colorectal cancer.
[0011] Preferably, the reagent includes primer pairs that specifically detect ARRDC4 mRNA, CBX3 mRNA, and MMP1 mRNA.
[0012] Preferably, the primer pairs for specifically detecting ARRDC4 mRNA have nucleotide sequences as shown in SEQ ID NO:1 and SEQ ID NO:2, the primer pairs for specifically detecting CBX3 mRNA have nucleotide sequences as shown in SEQ ID NO:3 and SEQ ID NO:4, and the primer pairs for specifically detecting MMP1 mRNA have nucleotide sequences as shown in SEQ ID NO:5 and SEQ ID NO:6.
[0013] Preferably, the reagent further includes a primer pair for detecting the internal reference gene GAPDH.
[0014] The fourth aspect of this invention aims to provide an intelligent prediction system for colorectal cancer based on the joint detection of three genes: ARRDC4, CBX3, and MMP1, comprising: (1) Data acquisition module: used to acquire the mRNA expression levels of ARRDC4, CBX3 and MMP1 genes in subject samples; (2) Risk Score Calculation Module: Calculates the risk score according to the following formula: Risk Score =β0+β1×E_ARRDC4+β2×E_CBX3+β3×E_MMP1 Where E_ARRDC4, E_CBX3, and E_MMP1 represent the normalized expression levels of the ARRDC4, CBX3, and MMP1 genes, respectively; β0 is the intercept, and β1, β2, and β3 are the regression coefficients corresponding to each gene. (3) Risk assessment module: The calculated risk score is compared with the preset cutoff value. If the risk score is ≥ the preset cutoff value, the subject is judged to be at high risk of colorectal cancer; if the risk score is < the preset cutoff value, the subject is judged to be at low risk.
[0015] Preferably, the regression coefficients in the risk scoring formula are obtained by performing Firth penalized logistic regression fitting on the training set queue (GSE21510), specifically: β0= -22.6508, β1= -0.3505, β2= 2.5706, β3= 0.5162; That is: Risk Score = -22.6508 - 0.3505 × E_ARRDC4 + 2.5706 × E_CBX3 +0.5162 × E_MMP1.
[0016] Preferably, the preset cutoff value is -4.5154; when the subject's risk score is ≥-4.5154, they are judged to be at high risk of colorectal cancer; when the risk score is <-4.5154, they are judged to be at low risk.
[0017] Preferably, the subject sample is a tissue sample, blood sample, serum sample, plasma sample, fecal sample, or intestinal exfoliated cell sample.
[0018] A fifth aspect of the present invention aims to provide a kit for the diagnosis, screening or risk assessment of colorectal cancer, the kit comprising primer pairs for the specific detection of ARRDC4 mRNA, CBX3 mRNA and MMP1 mRNA.
[0019] Compared with the prior art, the present invention has the following beneficial effects: This invention identifies a key gene combination in colorectal cancer—the downregulation of the tumor suppressor gene ARRDC4 and the upregulation of the oncogenes CBX3 and MMP1—and establishes a three-gene joint detection system. ARRDC4, as a tumor suppressor gene, is specifically lowly expressed in colorectal cancer; CBX3, as an epigenetic regulator, is highly expressed in colorectal cancer; and MMP1, as a matrix metalloproteinase, is highly expressed in colorectal cancer. These three genes are involved in different functional pathways, and joint detection can achieve functional complementarity, improving the sensitivity and specificity of diagnosis. Specifically, this is manifested in the following aspects: High sensitivity: Based on the three-gene combined risk scoring model, the sensitivity reached 95.26% in the independent validation set, which can effectively detect colorectal cancer patients and reduce missed diagnoses.
[0020] High specificity: The specificity reaches 100% in the independent validation set, which can accurately exclude non-cancerous individuals and reduce unnecessary colonoscopy examinations.
[0021] Functional complementarity: The combined detection of low expression of tumor suppressor gene ARRDC4 and high expression of oncogenes CBX3 and MMP1 involves different molecular pathways and has complementary functions, avoiding false positives or false negatives that may occur with single gene detection.
[0022] Convenient sample acquisition: This invention enables the joint detection of three marker genes based on clinical tissue samples, providing new detection indicators for molecular auxiliary diagnosis of colorectal cancer.
[0023] Quantification: Personalized risk assessment is achieved through continuous risk scoring, which facilitates clinical decision-making.
[0024] Multi-platform compatibility: Detection can be achieved through various technical platforms such as qRT-PCR and RNA-seq. Attached Figure Description
[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0026] Figure 1 is a flowchart of the present invention.
[0027] Figure 2 The results of differentially expressed gene screening and LASSO regression feature selection for colorectal cancer are shown. The top of the figure is a volcano plot of differentially expressed genes, with red representing upregulated genes, blue representing downregulated genes, and gray representing genes with no significant difference. The bottom of the figure is a LASSO regression coefficient path plot, with the horizontal axis representing log(λ) and the vertical axis representing the regression coefficient.
[0028] Figure 3 This figure shows the results of random forest gene importance assessment and core gene intersection screening. The top part is a bar chart of random forest gene importance, sorted by the decrease in average Gini coefficient; the bottom part is a Venn diagram of genes screened by random forest and LASSO, with the intersection genes identified as ARRDC4, CBX3, and MMP1.
[0029] Figure 4The figures show the ROC curve analysis results of the gene diagnostic model. In the figure, A represents the ROC curves of ARRDC4, CBX3, and MMP1 distinguishing colorectal cancer tissue from normal tissue in the discovery cohort (GSE21510), and B represents the ROC curves of ARRDC4, CBX3, and MMP1 distinguishing colorectal cancer tissue from normal tissue in the independent validation cohort (GSE39582). The area under the curve (AUC) and 95% confidence interval (95% CI) for each gene are labeled in the figure.
[0030] Figure 5 The expression of three marker genes in colorectal cancer (CE) tissue and adjacent normal tissue (PE) of patients is shown. P<0.001 , P<0.0001 (paired t-test). Detailed Implementation
[0031] 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.
[0032] See the attached flowchart for a schematic diagram of the invention. Figure 1 Unless otherwise specified, the experimental methods used in the following examples are conventional methods, performed according to the techniques or conditions described in the literature in this field or according to the product instructions. Unless otherwise specified, the materials and reagents used in the following examples are commercially available.
[0033] Example 1. Screening of colorectal cancer-specific marker genes ARRDC4, CBX3, and MMP1 The GSE21510 dataset was downloaded from the GEO database. This dataset contains gene expression profiles of colorectal cancer tumor tissues and their paired adjacent normal tissues. Differentially expressed genes (DEGs) were analyzed using the limma package (Linear Models for Microarray Data). Using a threshold of |log2 fold change| > 1 and an adjusted p-value < 0.05, 1,795 DEGs were identified, of which 897 genes were significantly upregulated in CRC tissues and 898 genes were significantly downregulated.
[0034] Further weighted gene co-expression network analysis (WGCNA) was used to identify the gene modules most significantly associated with the CRC phenotype. Specific parameter settings were as follows: soft threshold power β = 5 (to achieve a scale-free topology fit index R² of 0.9), network type signed, TOM type signed, minimum module size 30 genes, and module merge threshold (mergeCutHeight) of 0.25 (modules with a feature gene (ME) dissimilarity < 0.25 were merged). Key modules were screened based on the Pearson correlation coefficient between module feature genes (ME) and the CRC phenotype. Hub genes were screened using thresholds of |module membership (MM)| > 0.8 and |gene significance (GS)| > 0.1. The intersection of the selected hub genes with the aforementioned 1,795 DEGs was calculated, ultimately yielding 472 candidate hub genes. Two complementary machine learning algorithms, LASSO regression and random forest, were used for feature selection of the candidate genes. LASSO regression uses the glmnet package, with family=binomial and alpha=1. The optimal penalty parameter λ is determined through 10-fold cross-validation. The optimal λ value is selected based on the minimum average cross-validation bias (lambda.min), and 10 genes with non-zero regression coefficients are selected. Random forest uses the randomForest package, with the number of decision trees (ntree) set to 500. The tuneRF function searches for the optimal mtry value (i.e., the number of variables randomly selected at each split node) within the range of 1 to 10. The optimal parameter is determined based on the lowest out-of-bag (OOB) error rate. Genes with the highest importance are selected by sorting according to the decrease in average Gini coefficient. The intersection of the two algorithms identifies three common candidate genes: ARRDC4, CBX3, and MMP1 (see appendix). Figure 2 and attached Figure 3 The NCBI Gene IDs for the three genes are: ARRDC4 (Gene ID: 91947), CBX3 (Gene ID: 11335), and MMP1 (Gene ID: 4312).
[0035] Example 2. Diagnostic efficacy assessment of three specific marker genes for colorectal cancer Using the GSE21510 dataset as the discovery queue and the GSE39582 dataset as the independent validation queue, expression profile data were downloaded from the GEO database. The raw expression values were transformed by log2 and then normalized to quantiles using the normalizeBetweenArrays function of the limma package. If the gene was located in a row and the sample was located in a column, the matrix was transposed to a sample × gene matrix. ARRDC4, CBX3, and MMP1 were matched using regular expressions to remove samples with missing target gene expression. Based on the dataset grouping information, adjacent normal tissue was defined as 0 and colorectal cancer tumor tissue as 1. ROC analysis was performed using the pROC package (version 1.18.0) in R software (version 4.3.1). The roc function was called for each of the three genes, with the core parameters set as response=Group, predictor=Exp_gene, levels=c(0,1), direction="auto", ci=TRUE, ci.method="DeLong", and smooth=FALSE. The AUC and its 95% confidence interval were calculated using the DeLong method. ROC curves were plotted with 1-specificity on the x-axis and sensitivity on the y-axis.
[0036] The results show (see appendix) Figure 4 In the discovery cohort, the AUCs for ARRDC4, CBX3, and MMP1 reached 0.996, 1.000, and 0.996, respectively; in the independent validation cohort, the AUCs were 0.958 (95% CI: 0.926–0.982), 0.932 (95% CI: 0.900–0.960), and 0.911 (95% CI: 0.874–0.944), respectively. The AUCs in the validation cohort were slightly lower than those in the discovery cohort, but all remained above 0.91, indicating that all three genes possess good diagnostic accuracy and stability, and that any single gene could serve as a candidate biomarker with clinical translational potential.
[0037] Example 3. Validation of the expression of three marker genes in clinical colorectal cancer patient samples This study was approved by the Ethics Committee of the Affiliated Hospital of Hebei University (Ethics Approval No.: HDFYLL-KY-2026-193), and all patients signed informed consent forms. Paired tumor epithelial tissue (CE) and adjacent normal epithelial tissue (PE) samples were collected from 74 patients with primary colorectal cancer at the Affiliated Hospital of Hebei University, totaling 74 pairs. Total RNA was extracted, cDNA was synthesized using SweScript SuperMix, and qRT-PCR was performed using SYBR Green Mix (primers are shown in Table 1). GAPDH was used as an internal reference gene, and its relative expression level was calculated using the 2^(-ΔΔCt) method.
[0038] Table 1
[0039] The results show (see appendix) Figure 5 Compared with adjacent normal tissue, ARRDC4 expression was significantly downregulated in CRC tumor tissue (P<0.0001), while CBX3 and MMP1 expression was significantly upregulated (P<0.05, P<0.01). ARRDC4 showed a downregulated trend in 75.7% of patients, while CBX3 and MMP1 showed an upregulated trend in 70.3% and 74.3% of patients, respectively, and the expression characteristics were consistent in most patients.
[0040] This embodiment validated the differential expression patterns of three genes at the clinical sample level, confirming that they can serve as key characteristic genes for the diagnosis of colorectal cancer, and providing clinical evidence for the auxiliary diagnosis of resectable CRC.
[0041] Example 4. Construction and validation of a risk scoring model based on the joint detection of ARRDC4, CBX3 and MMP1 three genes. 1. Model Building Based on the expression data of three genes, ARRDC4, CBX3, and MMP1, in the training set cohort (GSE21510, containing 148 samples), and using colorectal cancer pathological diagnosis as the gold standard (normal tissue = 0, tumor tissue = 1), a multi-gene joint diagnostic model was constructed using Firth penalized logistic regression. Firth penalized regression is suitable for handling small samples or completely segregated problems and can provide stable parameter estimates.
[0042] The formula for the optimal model obtained by fitting is: Risk Score = -22.6508 - 0.3505 × E_ARRDC4 + 2.5706 × E_CBX3 +0.5162 × E_MMP1 Wherein, E_ARRDC4, E_CBX3, and E_MMP1 represent the normalized mRNA expression levels of the corresponding genes. This formula shows that ARRDC4 expression is negatively correlated with the risk score (negative coefficient), meaning low ARRDC4 expression increases the risk of colorectal cancer; while CBX3 and MMP1 expression are positively correlated with the risk score (positive coefficient), meaning high expression of both increases the risk of colorectal cancer. These results are completely consistent with the expression trends of the three genes in the differential expression analysis of Example 1 and the clinical sample validation of Example 3, demonstrating the functional complementary advantages of combined detection of tumor suppressor genes and oncogenes.
[0043] 2. Model Diagnostic Efficacy Evaluation The model's discriminative ability was evaluated using a training set cohort (GSE21510) and an independent external validation set cohort (GSE39582, containing 585 samples). The AUC was calculated using ROC curve analysis.
[0044] The results showed that the AUC on the training set was 1.000 (95% CI: 1.000–1.000), and the AUC on the validation set was 0.984 (95% CI: 0.972–0.996). The model maintained high diagnostic accuracy on independent external data, indicating that the three-gene joint risk scoring model constructed in this invention has good generalization ability and stable diagnostic performance. Compared with the AUC of single genes (0.911–0.958) in Example 2, the AUC of the joint model (0.984) was significantly improved, demonstrating that multi-gene joint detection may be superior to single-gene detection.
[0045] 3. Determination of risk stratification cutoff values and validation of clinical performance. In the validation set (GSE39582), a sensitivity-first strategy was adopted to determine the optimal risk stratification cutoff value, that is, the threshold point with the highest specificity was selected while ensuring that the model sensitivity is not lower than 85%.
[0046] ROC curve analysis revealed an optimal cutoff value of -4.5154. At this cutoff value, the model's clinical performance is as follows: The sensitivity was 95.26% (422 / 443), the specificity was 100% (19 / 19), the accuracy was 95.45%, the positive predictive value (PPV) was 100%, and the negative predictive value (NPV) was 47.50%.
[0047] When -4.5154 was used as the judgment threshold, the model achieved a detection sensitivity of 95.26% for colorectal cancer, and no normal samples were misclassified as positive (specificity 100%), demonstrating excellent clinical application potential.
[0048] 4. Comparison with existing clinical biomarkers The efficacy of the three-gene combined diagnostic model constructed in this invention was compared with that of CEA, a commonly used serum biomarker for colorectal cancer. Previous meta-analyses have shown that CEA has a sensitivity of only 49% and a specificity of 53% in the diagnosis of colorectal cancer, indicating limited diagnostic efficacy. [4] A systematic review reported that the sensitivity of fecal CEA ranged from 50.0% to 85.7%, the specificity from 73.0% to 100.0%, and the AUC ranged from 0.704 to 0.831. [5]The three-gene combined model of this invention achieved a sensitivity of 95.26% and a specificity of 100% in the validation set. Compared with the diagnostic efficacy of CEA reported in previous literature, this model showed higher values in terms of sensitivity and specificity, suggesting that it has better diagnostic potential than CEA.
[0049] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.
[0050] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
[0051] References cited in this invention [1] Siegel RL, Miller KD, Jemal A. Cancer statistics, 2020. CA CancerJ Clin. 2020 Jan;70(1):7-30. doi: 10.3322 / caac.21590. Epub 2020 Jan 8. PMID:31912902. [2] Duffy MJ. Carcinoembryonic antigen as a marker for colorectalcancer: is it clinically useful? Clin Chem. 2001 Apr;47(4):624-30. PMID:11274010. [3] Wang R, Wang Q, Li P. Significance of carcinoembryonic antigen detection in the early diagnosis of colorectal cancer: A systematic review and meta-analysis. World J Gastrointest Surg. 2023 Dec 27;15(12):2907-2918.PMID: 38222002. [4] Jorboniyan S, Fakhar HBZ, Akbari ME, Izadi N, Rangraz S.Investigation of the expression level of methylated septin 9 gene andserological carcinoembryonic antigen to diagnose colorectal cancer. A meta-analysis study. Prz Gastroenterol. 2024;16(4):368-379. PMID: 39810876. [5] Li X, Stassen L, Schrotz-King P, Zhao Z, Cardoso R, Raut JR,Bhardwaj M, Brenner H. Potential of Fecal Carcinoembryonic Antigen forNoninvasive Detection of Colorectal Cancer: A Systematic Review. Cancers(Basel). 2023 Nov 30;15(23):5656. PMID: 38067360.
Claims
1. A three-gene biomarker for colorectal cancer, characterized in that, The biomarkers include ARRDC4, CBX3, and MMP1.
2. Any of the following applications of the biomarker of claim 1: A. Application in the preparation of products for the diagnosis of colorectal cancer; B. Application in the preparation of products for colorectal cancer screening; C. Application in the preparation of products for assessing the risk of colorectal cancer.
3. Any of the following applications of the detection reagent for detecting the expression level of the biomarker according to claim 1: A. Application in the preparation of products for the diagnosis of colorectal cancer; B. Application in the preparation of products for colorectal cancer screening; C. Application in the preparation of products for assessing the risk of colorectal cancer.
4. The application according to claim 3, characterized in that, The reagents include primer pairs that specifically detect ARRDC4 mRNA, CBX3 mRNA, and MMP1 mRNA.
5. The application according to claim 4, characterized in that, The primer pairs for specifically detecting ARRDC4 mRNA have nucleotide sequences as shown in SEQ ID NO:1 and SEQ ID NO:2, the primer pairs for specifically detecting CBX3 mRNA have nucleotide sequences as shown in SEQ ID NO:3 and SEQ ID NO:4, and the primer pairs for specifically detecting MMP1 mRNA have nucleotide sequences as shown in SEQ ID NO:5 and SEQ ID NO:
6.
6. A colorectal cancer intelligent prediction system based on the joint detection of three genes, ARRDC4, CBX3, and MMP1, characterized in that, include: (1) Data acquisition module: used to acquire the mRNA expression levels of ARRDC4, CBX3 and MMP1 genes in subject samples; (2) Risk Score Calculation Module: Calculates the risk score according to the following formula: Risk Score =β0+β1×E_ARRDC4+β2×E_CBX3+β3×E_MMP1 Where E_ARRDC4, E_CBX3, and E_MMP1 represent the normalized expression levels of the ARRDC4, CBX3, and MMP1 genes, respectively; β0 is the intercept, and β1, β2, and β3 are the regression coefficients corresponding to each gene. (3) Risk assessment module: The calculated risk score is compared with the preset cutoff value. If the risk score is ≥ the preset cutoff value, the subject is judged to be at high risk of colorectal cancer; if the risk score is < the preset cutoff value, the subject is judged to be at low risk.
7. The intelligent prediction system according to claim 6, characterized in that, The regression coefficients in the risk scoring formula are obtained by fitting a Firth-penalized logistic regression to the training set queue (GSE21510), specifically: β0= -22.6508,β1= -0.3505,β2= 2.5706,β3= 0.5162; That is: Risk Score = -22.6508 - 0.3505 × E_ARRDC4 + 2.5706 × E_CBX3 + 0.5162 × E_MMP1.
8. The intelligent prediction system according to claim 6, characterized in that, The preset cutoff value is -4.5154; when the subject's risk score is ≥-4.5154, they are considered to be at high risk of colorectal cancer; when the risk score is <-4.5154, they are considered to be at low risk.
9. The system according to claim 6, characterized in that, The subject samples were tissue samples.
10. A reagent kit, characterized in that, The kit is used for the diagnosis, screening, or risk assessment of colorectal cancer, and includes primer pairs that specifically detect ARRDC4 mRNA, CBX3 mRNA, and MMP1 mRNA.
