Early prediction biomarker and prediction model for colorectal cancer liver metastasis
By constructing specific markers and models of exosomal RNA in plasma, the sensitivity and specificity issues of early prediction of colorectal cancer liver metastasis have been resolved, achieving highly accurate non-invasive risk assessment and providing significant early warning time and personalized treatment opportunities.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- FUDAN UNIV SHANGHAI CANCER CENT
- Filing Date
- 2025-11-20
- Publication Date
- 2026-04-17
AI Technical Summary
Existing diagnostic and predictive methods for colorectal cancer liver metastasis are insufficient in early prediction ability, prediction target sensitivity and specificity, imaging detection sensitivity is insufficient, serological markers have high false negative and false positive rates, and liquid biopsy technology has limited detection sensitivity.
Plasma exosomal RNA, including specific mRNAs and lncRNAs, was used as biomarkers. Genes were screened using regularized regression and ensemble learning algorithms, and a neural network model was constructed for prediction. The relative expression levels of exosomal RNA were used to assess the risk of colorectal cancer liver metastasis.
It significantly improves the accuracy and stability of predicting liver metastasis risk, identifies high-risk patients 18.4 months in advance, provides sufficient warning time, supports personalized intervention and treatment, reduces detection risk, and is suitable for non-invasive testing.
Smart Images
Figure CN121874342A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of biotechnology, specifically relating to an early predictive biomarker and predictive model for colorectal cancer liver metastasis. Background Technology
[0002] Colorectal cancer (CRC) is one of the leading causes of death from malignant tumors worldwide. Approximately 50% of CRC patients develop liver metastasis (CRLM) during disease progression. Liver metastasis is the most common form of distant metastasis in CRC patients and a major cause of death.
[0003] Currently, clinical assessment and monitoring of liver metastases mainly rely on imaging examinations such as CT, MRI, and PET-CT. These methods are highly accurate in detecting visible lesions, but they lack sensitivity for small lesions or early-stage metastases, often only being detected when the tumor burden has significantly increased, thus missing the opportunity for early intervention. While serological marker tests, such as CEA and CA19-9, are convenient and low-cost, their sensitivity and specificity are limited, especially in early or occult liver metastases, resulting in high false-negative and false-positive rates and a lack of organ specificity.
[0004] With the development of liquid biopsy technology, researchers have begun to explore the use of tumor-derived molecules in blood for early diagnosis and risk prediction, such as circulating tumor DNA (ctDNA) and circulating tumor cells (CTCs). ctDNA detection has proven valuable in monitoring minimal residual disease and predicting recurrence; some studies have shown that ctDNA can detect recurrence months earlier than imaging. However, the levels of ctDNA and CTCs in peripheral blood are extremely low, limiting detection sensitivity. Large sample sizes and high-sensitivity platforms are required to obtain reliable results, and these findings primarily reflect the existing disease state rather than providing early predictions of metastasis risk.
[0005] In recent years, exosomes (EVs) have attracted widespread attention as an emerging carrier for liquid biopsy. EVs are actively secreted by tumor cells and are widely present in body fluids such as blood. They can stably carry various biomolecules such as RNA, proteins, and lipids, and play an important role in intercellular communication, tumor development, and metastasis. Exosomal RNA (EV-RNA) is considered an ideal biomarker for early tumor diagnosis, prognostic assessment, and treatment monitoring due to its high source specificity, strong stability, and ease of collection. Existing studies suggest that EV-RNA is associated with the development and metastasis of various cancers, but research on predicting liver-specific metastasis in CRC is still lacking. Summary of the Invention
[0006] The purpose of this application is to provide an early predictive biomarker and predictive model for colorectal cancer liver metastasis, in order to solve the technical problems of limited early predictive ability and insufficient sensitivity and specificity of predictive targets in existing diagnostic and predictive methods for colorectal cancer liver metastasis.
[0007] To achieve the above objectives, a first aspect of this application provides an early predictive biomarker for colorectal cancer liver metastasis, characterized in that the biomarker is exosomal RNA derived from plasma, and the biomarker includes at least one of the following mRNAs and lncRNAs:
[0008] Twelve mRNAs were named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P, and SMIM20, and three lncRNAs were named AC007319.1, AL590560.1, and AC100810.3.
[0009] In one or more embodiments, it includes 12 mRNAs named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P, and SMIM20, and 3 lncRNAs named AC007319.1, AL590560.1, and AC100810.3.
[0010] In one or more embodiments, at least one of the following mRNAs and lncRNAs is included:
[0011] Two mRNAs were named ALPK3 and PALM3, and one lncRNA was named AC007319.1.
[0012] To achieve the above objectives, a second aspect of this application provides the application of early predictive biomarkers described in any of the above embodiments in predicting liver metastasis of colorectal cancer.
[0013] To achieve the above objectives, a third aspect of this application provides a method for constructing a predictive model for colorectal cancer liver metastasis, comprising:
[0014] Obtain a sample set labeled with sample tags. The sample set includes sample data from multiple colorectal cancer patients. The sample data includes the relative expression levels of each gene in exosomal RNA. The sample tags include liver metastasis, no metastasis, extrahepatic metastasis, and liver plus extrahepatic metastasis.
[0015] Differential analysis was performed on sample data with different sample labels, and genes were screened to obtain the first sample set;
[0016] A regularized regression algorithm is used to screen genes in the first sample set to obtain a second sample set;
[0017] An ensemble learning algorithm is used to filter genes in the first sample set to obtain a third sample set;
[0018] The intersection of the second sample set and the third sample set is taken to obtain the sample target set. The sample data in the sample target set includes the relative expression levels of the following genes: 12 mRNAs named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P and SMIM20, and 3 lncRNAs named AC007319.1, AL590560.1 and AC100810.3.
[0019] The target sample set is divided into a training sample set and a validation sample set. The sample data in the training sample set is used as input, and the neural network model is trained based on the sample labels.
[0020] The performance of the neural network model is tested using the sample validation set to obtain the optimal model parameters and thus a predictive model for colorectal cancer liver metastasis.
[0021] In one or more embodiments, the neural network model is an artificial neural network model, and includes an input layer, a hidden layer and an output layer, wherein the input layer includes 15 nodes, the hidden layer includes 5 nodes and the output layer includes 2 nodes.
[0022] In one or more embodiments, the differential analysis and gene screening step includes:
[0023] Screen for genes whose expression levels are significantly upregulated or downregulated in the liver metastasis group and the non-metastasis group;
[0024] Screen for genes whose expression levels are significantly upregulated or downregulated in the liver metastasis group and the extrahepatic metastasis group;
[0025] The genes with significantly upregulated or downregulated expression levels were screened in the liver plus extrahepatic metastasis group and the non-metastasis group;
[0026] Genes with different expression levels in the liver metastasis group and the liver plus extrahepatic metastasis group were removed.
[0027] In one or more embodiments, the regularized regression algorithm is specifically LASSO regression.
[0028] In one or more embodiments, the ensemble learning algorithm is specifically a random forest algorithm.
[0029] To achieve the above objectives, a fourth aspect of this application provides a method for predicting liver metastasis of colorectal cancer, comprising:
[0030] The relative expression levels of the following exosomal RNAs in the plasma of target colorectal cancer patients were obtained: 12 mRNAs named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P, and SMIM20, and 3 lncRNAs named AC007319.1, AL590560.1, and AC100810.3, and sample data were obtained.
[0031] The sample data is input into the prediction model constructed by the construction method described in any of the above embodiments to obtain the liver metastasis risk prediction results for the target colorectal cancer patient.
[0032] To achieve the above objectives, a fifth aspect of this application provides a device for predicting liver metastasis of colorectal cancer, comprising:
[0033] The acquisition module is used to obtain the relative expression levels of the following exosomal RNAs in the plasma of target colorectal cancer patients: 12 mRNAs named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P, and SMIM20, and 3 lncRNAs named AC007319.1, AL590560.1, and AC100810.3, to obtain sample data;
[0034] The prediction module is used to input the sample data into the prediction model constructed by the construction method of any one of claims 5 to 7 to obtain the prediction result of the liver metastasis risk of the target colorectal cancer patient.
[0035] To achieve the above objectives, a sixth aspect of this application provides an electronic device, comprising:
[0036] At least one processor; and
[0037] A memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform the method for predicting colorectal cancer liver metastasis as described in any of the above embodiments.
[0038] To achieve the above objectives, a seventh aspect of this application provides a machine-readable storage medium storing executable instructions that, when executed, cause the machine to perform the method for predicting colorectal cancer liver metastasis as described in any of the above embodiments.
[0039] The advantages of this application, which differ from existing technologies, are:
[0040] This application is the first to identify a group of liver metastasis-specific biomarkers in colorectal cancer exosomal RNA research. Through cross-validation with tissue RNA sequencing data and functional experiments, the role of key genes in tumor cell migration was clarified. This biomarker combination can not only be used to predict the risk of liver metastasis, but also provide potential molecular targets for subsequent targeted intervention research, thus expanding its application value.
[0041] The prediction model in this application is constructed based on biomarkers of exosomal RNA. It not only considers the expression differences of genes in different transfer states, but also combines multiple data filtering and algorithm optimization steps, which significantly improves the generalization ability and stability of the model in different datasets.
[0042] The predictive model presented in this application can identify high-risk patients for liver metastases on average 18.4 months before radiological diagnosis. This lead time far exceeds that of currently reported ctDNA panels (approximately 3.5–5 months), providing clinicians with sufficient early warning time to adjust monitoring frequency, optimize follow-up protocols, and even initiate interventional treatments earlier. This significantly extended predictive window allows patients to receive risk alerts during the asymptomatic period, which is of great significance for improving prognosis.
[0043] The predictive model of this application only requires the collection of peripheral blood samples from patients, without the need for surgical sampling or invasive biopsy, thus avoiding additional medical risks and patient suffering. Furthermore, it can be repeatedly tested multiple times during treatment and follow-up to achieve dynamic risk assessment.
[0044] The detection steps of the prediction method in this application are compatible with existing liquid biopsy technology platforms, have strong adaptability, are easy to be quickly implemented in hospital laboratories or third-party testing centers, and have broad potential for clinical promotion.
[0045] This application demonstrates the biological rationale of the predictive model by combining gene knockdown experiments, providing experimental evidence for the model's clinical credibility, and also providing potential molecular targets for subsequent targeted intervention studies, thus expanding its application value. Attached Figure Description
[0046] To more clearly illustrate the technical solutions in the embodiments of this application 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 some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0047] Figure 1 This is a flowchart illustrating one implementation method of constructing a predictive model for colorectal cancer liver metastasis according to this application.
[0048] Figure 2 This is a schematic diagram of one embodiment of the colorectal cancer liver metastasis prediction device of this application;
[0049] Figure 3 This is a schematic diagram of the structure of one embodiment of the electronic device of this application;
[0050] Figure 4 This is a diagram of exosomal RNA expression in each transition state group in Example 1 of this application;
[0051] Figure 5 This is the Venn diagram of genes screened by differential analysis in Example 2 of this application;
[0052] Figure 6 This is a data graph of the LASSO regression screening in Embodiment 2 of this application;
[0053] Figure 7 This is the gene importance ranking graph of the random forest algorithm in Embodiment 2 of this application;
[0054] Figure 8 This is a schematic diagram of the ANN model in this application;
[0055] Figure 9 This is the ROC curve of the ANN model in this application;
[0056] Figure 10 This is a data diagram of Embodiment 4 of this application;
[0057] Figure 11 This is a diagram showing the experimental results of Embodiment 5 of this application;
[0058] Figure 12 This is the Kaplan-Meier curve of Embodiment 6 of this application. Detailed Implementation
[0059] To enable those skilled in the art to better understand the technical solutions in this disclosure, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments in this disclosure, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this disclosure.
[0060] Currently, there is no exosomal RNA-based method for predicting CRC liver metastasis that has been validated in large-scale and prospective cohort studies and is available for clinical practice. Traditional prediction methods suffer from insufficient target specificity and sensitivity (circulating tumor DNA, circulating tumor cell detection, imaging detection) and limited early prediction capabilities (imaging detection).
[0061] To address the aforementioned issues, the applicant has proposed a non-invasive detection method based on exosomal RNA that can identify high-risk individuals for colorectal cancer liver metastases within a long window before radiological diagnosis, in order to guide precise monitoring and individualized intervention.
[0062] First, this application provides an early predictive biomarker for colorectal cancer liver metastasis. The biomarker is exosomal RNA derived from plasma, and includes at least one of the following mRNAs and lncRNAs:
[0063] Twelve mRNAs were named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P, and SMIM20, and three lncRNAs were named AC007319.1, AL590560.1, and AC100810.3.
[0064] Furthermore, in one embodiment, the biomarkers may include 15 mRNAs named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P, and SMIM20, and 3 lncRNAs named AC007319.1, AL590560.1, and AC100810.3.
[0065] In another embodiment, the biomarker may include at least one of the following mRNAs and lncRNAs:
[0066] Two mRNAs were named ALPK3 and PALM3, and one lncRNA was named AC007319.1.
[0067] This application also provides the application of an early predictive biomarker of any of the above embodiments in predicting liver metastasis of colorectal cancer.
[0068] This application also provides a method for constructing a predictive model for colorectal cancer liver metastasis; please refer to [link to relevant documentation]. Figure 1 , Figure 1 This is a flowchart illustrating one implementation method of constructing a predictive model for colorectal cancer liver metastasis according to this application.
[0069] like Figure 1 As shown, the construction method includes:
[0070] S100. Obtain the sample set labeled with sample tags.
[0071] The sample set includes data from multiple colorectal cancer patients. The data includes the relative expression levels of each gene in exosomal RNA, and the sample labels include liver metastasis, no metastasis, extrahepatic metastasis, and liver plus extrahepatic metastasis.
[0072] S200. Perform differential analysis on sample data with different sample labels and screen genes to obtain the first sample set.
[0073] Specifically, in one implementation, the steps of differential analysis and gene screening include:
[0074] Screen for genes whose expression levels are significantly upregulated or downregulated in the liver metastasis group and the non-metastasis group;
[0075] Screen for genes whose expression levels are significantly upregulated or downregulated in the liver metastasis group and the extrahepatic metastasis group;
[0076] We screened for genes whose expression levels were significantly upregulated or downregulated in the liver plus extrahepatic metastasis group and the non-metastasis group.
[0077] S300. Use a regularized regression algorithm to screen genes in the first sample set to obtain the second sample set.
[0078] In one implementation, the regularized regression algorithm can be specifically LASSO regression; in other implementations, other algorithms that can achieve this function can also be used, such as adaptive LASSO regression, stepwise regression, etc.
[0079] S400. Use an ensemble learning algorithm to screen genes in the first sample set to obtain the third sample set.
[0080] In one implementation, the ensemble learning algorithm may be a random forest algorithm. In other implementations, other algorithms that can achieve this function may be used, such as support vector machines, gradient boosting trees, etc.
[0081] S500. Take the intersection of the second sample set and the third sample set to obtain the target sample set.
[0082] By comparing the genes in the second and third sample sets, the intersection is selected to obtain the target sample set.
[0083] The sample data in the target set includes the relative expression levels of the following genes: 12 mRNAs named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P, and SMIM20, and 3 lncRNAs named AC007319.1, AL590560.1, and AC100810.3.
[0084] S600. Divide the target sample set into a training sample set and a validation sample set. Use the sample data in the training sample set as input and train the neural network model based on the sample labels.
[0085] In one implementation, the neural network model can be an artificial neural network (ANN) model. Specifically, the ANN model can include one input layer, one hidden layer, and one output layer. The input layer can include 15 nodes to correspond to 15 genes; the hidden layer can include 5 nodes; and the output layer can include 2 nodes, thereby outputting two prediction results: high liver metastasis risk and low liver metastasis risk.
[0086] In other embodiments, the neural network model can also be other models commonly used in the art, all of which can achieve the effects of this embodiment.
[0087] S700. Test the performance of the neural network model with a sample validation set to obtain the optimal model parameters and obtain a predictive model for colorectal cancer liver metastasis.
[0088] This application also provides a method for predicting colorectal cancer liver metastasis using the prediction model constructed above, including:
[0089] The relative expression levels of the following exosomal RNAs in the plasma of target colorectal cancer patients were obtained: 12 mRNAs named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P, and SMIM20, and 3 lncRNAs named AC007319.1, AL590560.1, and AC100810.3, and sample data were obtained.
[0090] By inputting the sample data into the prediction model constructed by the construction method of any of the above embodiments, the liver metastasis risk prediction results of the target colorectal cancer patients are obtained.
[0091] This application also provides a device for predicting liver metastasis of colorectal cancer; please refer to [link to relevant documentation]. Figure 2 , Figure 2 This is a schematic diagram of one embodiment of the colorectal cancer liver metastasis prediction device of this application.
[0092] like Figure 2 As shown, the prediction device includes an acquisition module 21 and a prediction module 22.
[0093] The acquisition module 21 is used to acquire the relative expression levels of the following exosomal RNAs in the plasma of the target colorectal cancer patient: 12 mRNAs named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P and SMIM20, and 3 lncRNAs named AC007319.1, AL590560.1 and AC100810.3, to obtain sample data;
[0094] The prediction module 22 is used to input sample data into the prediction model constructed by the construction method of any of the above embodiments to obtain the prediction result of liver metastasis risk of the target colorectal cancer patient.
[0095] The details mentioned in the above description of the method embodiments also apply to the colorectal cancer liver metastasis prediction device of the embodiments of this specification. The above-described colorectal cancer liver metastasis prediction device can be implemented in hardware, software, or a combination of hardware and software.
[0096] This application also provides an electronic device, please refer to... Figure 3 , Figure 3 This is a schematic diagram of one embodiment of the electronic device of this application. For example... Figure 3 As shown, the electronic device 30 may include at least one processor 31, a memory 32 (e.g., non-volatile memory), a RAM 33, and a communication interface 34, and the at least one processor 31, memory 32, RAM 33, and communication interface 34 are connected together via a bus 35. The at least one processor 31 executes at least one computer-readable instruction stored or encoded in the memory 32.
[0097] It should be understood that the computer-executable instructions stored in memory 32, when executed, cause at least one processor 31 to perform the above-described combinations in the various embodiments of this specification. Figure 1 The description includes various operations and functions.
[0098] In the embodiments of this specification, electronic device 30 may include, but is not limited to: personal computer, server computer, workstation, desktop computer, laptop computer, notebook computer, mobile electronic device, smartphone, tablet computer, cellular phone, personal digital assistant (PDA), handheld device, messaging device, wearable electronic device, consumer electronic device, etc.
[0099] According to one embodiment, a program product, such as a machine-readable medium, is provided. The machine-readable medium may have instructions (i.e., the elements implemented in software as described above), which, when executed by a machine, cause the machine to perform the above-described combinations of the various embodiments of this specification. Figure 1 The various operations and functions described. Specifically, a system or apparatus equipped with a readable storage medium storing software program code that implements the functions of any of the embodiments described above, and enabling the computer or processor of the system or apparatus to read and execute the instructions stored in the readable storage medium.
[0100] In this case, the program code itself, which can be read from the readable medium, can perform the functions of any of the above embodiments, and therefore the machine-readable code and the readable storage medium storing the machine-readable code constitute a part of this specification.
[0101] Examples of readable storage media include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD-RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, program code can be downloaded from a server computer or the cloud via a communication network.
[0102] The technical solution of this application will be further described in detail below with reference to specific embodiments.
[0103] Example 1: Sample Acquisition
[0104] Peripheral venous blood (10 mL) was collected from 286 colorectal cancer patients at Fudan University Cancer Hospital, Fudan University Zhongshan Hospital, and Shanghai Jiao Tong University School of Medicine Xinhua Hospital. The blood was placed in blood collection tubes containing EDTA anticoagulant and processed within 2 hours of collection. The processing procedure included: centrifugation at 800×g for 10 minutes (room temperature) to remove blood cells, followed by centrifugation at 16,000×g for 10 minutes (4℃) to remove cell debris. Subsequently, plasma exosomes were isolated using a kit, and total RNA was extracted for high-throughput sequencing. The TPM value was calculated as the expression level of transcripts.
[0105] To obtain exosomal RNA characteristics associated with liver metastasis, the applicant divided the samples into liver metastasis group (LM), no metastasis group (M0), extrahepatic metastasis only group (EM), and liver + extrahepatic metastasis group (LEM) according to the metastasis status of the patients at the time of blood collection.
[0106] Please see Figure 4 , Figure 4 This is an exosomal RNA expression map of each transition state group in Example 1 of this application, as shown below. Figure 4 As shown, there are significant differences in exosomal RNA expression among different translocation groups.
[0107] Example 2: Gene Screening
[0108] To eliminate differentially expressed genes, the applicant used DESeq2 to perform differential analyses of LM and M0, LM and EM, and LEM and EM (threshold |logFC|>1, P<0.05).
[0109] Differential analysis was used to screen for genes associated with liver metastasis, ultimately yielding 92 candidate exosomal RNAs related to liver metastasis. Specifically, differential analysis constructed four comparison groups: LM vs. EM, LM vs. M0, LEM vs. EM, and LM vs. MEM. Genes significantly upregulated or downregulated in LM vs. EM, LM vs. M0, and LEM vs. EM were intersected, while genes differentially expressed between LEM and LM were excluded to remove genes associated with extrahepatic metastasis.
[0110] Please see Figure 5 , Figure 5 This is the Venn diagram of the differentially analyzed genes screened in Example 2 of this application. Figure 5 In diagram A, the gene expression level is upregulated; in diagram B, the gene expression level downregulated is selected. Each Venn circle in the diagram represents a comparison group, and the specific correspondences are as follows: (I) LEM vs. LM, (II) LM vs. EM, (III) LEM vs. EM, (IV) LM vs. M0. Figure 5 As shown, among the 92 genes screened, 89 were upregulated genes and 3 were downregulated genes.
[0111] Furthermore, the applicant further screened 92 genes using the LASSO regression method; please refer to [link to relevant documentation]. Figure 6 , Figure 6 This is a data graph of LASSO regression screening in Example 2 of this application. In the graph, A is the regression coefficient change curve and B is the mean square error change curve. Finally, 44 important genes were screened.
[0112] Simultaneously, the applicant also screened 92 genes using a random forest algorithm; please refer to [link / reference]. Figure 7 , Figure 7 This is the gene importance ranking graph of the random forest algorithm in Embodiment 2 of this application. 21 key genes were obtained by ranking and filtering based on importance.
[0113] Furthermore, the applicant selected the intersection of 44 important genes and 21 key genes to obtain the final 15 characteristic genes, namely 12 mRNAs named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P and SMIM20, and 3 lncRNAs named AC007319.1, AL590560.1 and AC100810.3.
[0114] Example 3: Model Building
[0115] The applicant used the relative expression levels of the aforementioned 15 exosomal RNAs in the plasma of 286 patients as the target sample set for model training and validation.
[0116] The model uses an ANN model, consisting of one input layer, one hidden layer, and one output layer. Please refer to [link / reference]. Figure 8 , Figure 8 This is a schematic diagram of the ANN model in this application, as shown below. Figure 8 As shown, the input layer includes 15 nodes, each corresponding to one of the 15 exosomal RNAs; the hidden layer includes 5 nodes H1 to H5; and the output layer includes 2 nodes, corresponding to high liver metastasis risk (HLMR) and low liver metastasis risk (LLMR), respectively.
[0117] The data from 286 patients were divided into a training set (N=133) and a validation set (N=153). The model was trained using the training set and validated using the validation set. Please refer to [link to relevant documentation]. Figure 9 , Figure 9 This is the ROC curve of the ANN model in this application, where A represents the training set data and B represents the validation set data.
[0118] like Figure 9 As shown, the training set AUC of the model in this application is 0.900, 95% CI: 0.864–0.937, and the validation set AUC is 0.823, 95% CI: 0.778–0.868, demonstrating excellent predictive performance.
[0119] Example 4: Model Performance Validation
[0120] To validate the model's performance, the applicant constructed a prospective follow-up cohort of 41 colorectal cancer patients, used the model to predict the risk of liver metastasis in these 41 patients, and followed them for 3 years to test the accuracy of the predictions. Figure 10 , Figure 10 This is a data diagram of Embodiment 4 of this application.
[0121] like Figure 10 As shown, among patients predicted to have a high risk of liver metastasis, 88.9% developed liver metastasis within 3 years; among patients predicted to have a low risk of metastasis, 93.9% did not develop liver metastasis within 3 years. Therefore, the model's sensitivity was 88.9% and its specificity was 93.8%, demonstrating high accuracy and stability under different population and sample conditions.
[0122] Furthermore, the model in this application can predict the occurrence of liver metastases on average 18.4 months before radiological diagnosis. This lead time far exceeds that of currently reported ctDNA panels (approximately 3.5–5 months), providing clinicians with sufficient early warning time to adjust monitoring frequency, optimize follow-up protocols, and even initiate interventional treatments earlier. This significantly extended prediction window allows patients to receive risk alerts during the asymptomatic period, which is of great significance for improving prognosis.
[0123] Example 5: Cross-validation
[0124] To verify the biological principles of the biomarkers proposed in this application, the applicant conducted a cross-analysis of exosomal RNA and RNA sequencing data from primary tumor sites and liver metastases to screen out key genes AC007319.1, ALPK3, and PALM3 in the model. By knocking down the expression levels of these three genes, the applicant verified the role of these genes in the migration of colorectal cancer cells.
[0125] Specifically, the applicant constructed and screened experimental siRNAs based on the sequences of the three genes mentioned above. Using these siRNAs, the three genes in HCT116 cells (human colon cancer cell line) were knocked down, and Transwell migration experiments were performed. A control group (NC) was also constructed. Please refer to the experimental results for details. Figure 11 , Figure 11 This is an experimental result diagram of Example 5 of this application, where A is a cell staining diagram and B is a migration data diagram.
[0126] like Figure 11 As shown in Figures A and B, compared to the control group NC, knockdown of AC007319.1, ALPK3, and PALM3 significantly inhibited the migration ability of HCT116 cells, with PALM3 knockdown showing the most significant and statistically significant inhibitory effect. These results not only demonstrate the biological rationale of the predictive model but also provide experimental evidence for its clinical reliability. Furthermore, they offer potential molecular targets for subsequent targeted intervention studies, expanding its application value.
[0127] Example 6: Exploring the Prognostic and Predictive Value of the Model
[0128] To verify the predictive value of the model, the applicant statistically analyzed the progression-free survival (PFS) and overall survival (OS) of patients predicted as having high liver metastasis risk (HLMR) and low liver metastasis risk (LLMR) in the training and validation sets of the above embodiments, obtaining Kaplan-Meier curves, as shown below. Figure 12 , Figure 12 This is the Kaplan-Meier curve of Embodiment 6 of this application, where A is the overall survival of patients in the training set, B is the overall survival of patients in the validation set, C is the progression-free survival of patients in the validation set, and D is the progression-free survival of patients in the validation set.
[0129] like Figure 12 As shown, the high-risk liver metastasis (HLMR) group had shorter PFS and OS.
[0130] It will be apparent to those skilled in the art that this disclosure is not limited to the details of the exemplary embodiments described above, and that this disclosure can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of this disclosure is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within this disclosure. No reference numerals in the claims should be construed as limiting the scope of the claims.
[0131] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.
Claims
1. An early prediction biomarker of colorectal cancer liver metastasis, characterized in that, The biomarker is an exosomal RNA derived from plasma, and the biomarker includes at least one of the following mRNAs and lncRNAs: Twelve mRNAs were named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P, and SMIM20, and three lncRNAs were named AC007319.1, AL590560.1, and AC100810.
3.
2. The early prediction biomarker of claim 1, wherein, It includes 12 mRNAs named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P, and SMIM20, and 3 lncRNAs named AC007319.1, AL590560.1, and AC100810.
3.
3. The early prediction biomarker of claim 1, wherein, Includes at least one of the following mRNAs and lncRNAs: Two mRNAs were named ALPK3 and PALM3, and one lncRNA was named AC007319.
1.
4. The use of any one of the early predictive biomarkers according to claims 1 to 3 in predicting liver metastasis of colorectal cancer.
5. A method for constructing a prediction model of colorectal liver metastasis, characterized by, include: Obtain a sample set labeled with sample tags. The sample set includes sample data from multiple colorectal cancer patients. The sample data includes the relative expression levels of each gene in exosomal RNA. The sample tags include liver metastasis, no metastasis, extrahepatic metastasis, and liver plus extrahepatic metastasis. Differential analysis was performed on sample data with different sample labels, and genes were screened to obtain the first sample set; A regularized regression algorithm is used to screen genes in the first sample set to obtain a second sample set; An ensemble learning algorithm is used to filter genes in the first sample set to obtain a third sample set; The intersection of the second sample set and the third sample set is taken to obtain the sample target set. The sample data in the sample target set includes the relative expression levels of the following genes: 12 mRNAs named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P and SMIM20, and 3 lncRNAs named AC007319.1, AL590560.1 and AC100810.
3. The target sample set is divided into a training sample set and a validation sample set. The sample data in the training sample set is used as input, and the neural network model is trained based on the sample labels. The performance of the neural network model is tested using the sample validation set to obtain the optimal model parameters and thus a predictive model for colorectal cancer liver metastasis.
6. The construction method of claim 5, wherein, The neural network model is an artificial neural network model, and includes one input layer, one hidden layer and one output layer. The input layer includes 15 nodes, the hidden layer includes 5 nodes, and the output layer includes 2 nodes.
7. The construction method of claim 5, wherein, The differential analysis and gene screening steps include: Screen for genes whose expression levels are significantly upregulated or downregulated in the liver metastasis group and the non-metastasis group; Screen for genes whose expression levels are significantly upregulated or downregulated in the liver metastasis group and the extrahepatic metastasis group; The genes with significantly upregulated or downregulated expression levels were screened in the liver plus extrahepatic metastasis group and the non-metastasis group; Remove genes with differential expression levels between the liver metastasis group and the liver plus extrahepatic metastasis group; and / or, The regularized regression algorithm is specifically LASSO regression; and / or, The ensemble learning algorithm is specifically the random forest algorithm.
8. A method of predicting colorectal liver metastasis, characterized by, include: The relative expression levels of the following exosomal RNAs in the plasma of target colorectal cancer patients were obtained: 12 mRNAs named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P, and SMIM20, and 3 lncRNAs named AC007319.1, AL590560.1, and AC100810.3, and sample data were obtained. The sample data is input into the prediction model constructed by the construction method of any one of claims 5 to 7 to obtain the prediction results of liver metastasis risk for the target colorectal cancer patient.
9. A device for predicting colorectal liver metastasis, characterized by, include: The acquisition module is used to obtain the relative expression levels of the following exosomal RNAs in the plasma of target colorectal cancer patients: 12 mRNAs named HRC, MAP1B, ALPK3, ADAMTS5, ADAMTS4, KATNAL2, LSMEM1, PALM3, SPCS2P4, PRKAR1AP1, OR2R1P, and SMIM20, and 3 lncRNAs named AC007319.1, AL590560.1, and AC100810.3, to obtain sample data; The prediction module is used to input the sample data into the prediction model constructed by the construction method of any one of claims 5 to 7 to obtain the prediction result of the liver metastasis risk of the target colorectal cancer patient.
10. An electronic device, comprising: At least one processor; as well as A memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform the method for predicting colorectal cancer liver metastasis as described in claim 8.
11. A machine-readable storage medium storing executable instructions that, when executed, cause the machine to perform the method for predicting colorectal cancer liver metastasis as described in claim 8.