Biomarker combination for evaluating prognosis of prostate cancer and application of biomarker combination
By constructing a biomarker combination SRGS based on aging-related genes, the problem of accurately predicting the risk of biochemical recurrence in prostate cancer patients has been solved, achieving efficient prognostic assessment and providing a more precise basis for treatment planning.
Patent Information
- Application Number
- CN202511004764.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-21
- Publication Date
- 2025-10-28
AI Technical Summary
Current technologies struggle to accurately predict the risk of biochemical recurrence in prostate cancer patients, especially those with high-risk disease characteristics. The lack of effective prognostic biomarkers and assessment models makes it difficult to predict the clinical course and to develop precise treatment plans.
Age-related genes were identified using weighted gene co-expression network analysis (WGCNA). A biomarker ensemble (SRGS) consisting of BMP6, CBX2, CCNE1, CCNE2, CHEK1, LMNB1, PCNA, POLD1, and UBE2C was constructed. A highly stable and accurate assessment model was developed using machine learning algorithms to evaluate the prognosis of prostate cancer patients.
This biomarker combination demonstrated strong predictive power across multiple datasets, effectively predicting the risk of biochemical recurrence in prostate cancer patients and assessing their prognosis. It exhibits high accuracy and stability, outperforming existing technologies.
Smart Images
Figure CN120843676A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of biomedical technology and relates to a combination of biomarkers for assessing the prognosis of prostate cancer and their application. Background Technology
[0002] Prostate cancer (PCa) is a common malignant tumor among men worldwide. Traditionally, PCa has been considered a slow-growing malignant tumor. However, during the course of the disease, some patients develop elevated prostate-specific antigen (PSA) levels and biochemical recurrence (BCR), and these patients have a significantly increased risk of metastasis or further progression.
[0003] On the one hand, the unique characteristic of slow progression of prostate cancer (PCa) makes its clinical course difficult to predict; on the other hand, existing research shows that biomarkers for predicting biomarkers for ... Summary of the Invention
[0004] The main objective of this invention is to provide a combination of biomarkers for assessing the prognosis of prostate cancer and their application, in order to solve at least one of the above-mentioned technical problems.
[0005] According to a first aspect of the invention, a biomarker combination for assessing the prognosis of prostate cancer is provided, comprising BMP6 (Bone Morphogenetic Protein 6), CBX2 (Chromobox Protein Homolog 2), CCNE1 (Cyclin E1), CCNE2, CHEK1 (Checkpoint Kinase 1), LMNB1 (Lamin B1), PCNA (Proliferating Cell Nuclear Antigen), POLD1 (DNA Polymerase Delta Catalytic Subunit), and UBE2C (Ubiquitin-Conjugating Enzyme E2C).
[0006] This invention first identifies senescence-related genes associated with prostate cancer prognosis through weighted gene co-expression network analysis (WGCNA). Then, based on the identified senescence-related genes, a set of 9 senescence-related gene signatures (SRGS) biomarkers with high stability and accuracy, which can be used to assess prostate cancer prognosis, was developed using 107 machine learning algorithms.
[0007] The biomarker combination provided by this invention has demonstrated strong predictive ability on the TCGA-PRAD training dataset and six independent validation datasets (Stockholm, Cambridge, CancerMap, GSE54460, Taylor, and an internal clinical dataset), and can be effectively used to predict the risk of BCR in prostate cancer patients and assess the prognosis of prostate cancer patients.
[0008] The biomarker combination provided by this invention can be used to prepare products for assessing the prognosis of prostate cancer.
[0009] According to a second aspect of the invention, the use of a product for detecting the expression levels of various genes in SRGS in a sample is provided in the preparation of a product for assessing the prognosis of prostate cancer.
[0010] In some implementations, the sample can be tumor tissue or peripheral blood. Specifically, peripheral blood can be collected when a patient is first diagnosed with PCa and / or tumor tissue can be collected after tumor resection to detect the expression levels of various genes in SRGS in peripheral blood or tumor tissue in order to assess the patient's prognosis.
[0011] In some implementations, the product for detecting the expression levels of each gene in the SRGS sample can be any reagent, kit, chip, and / or instrument known in the art that can quantitatively detect the expression levels of each gene in the SRGS sample.
[0012] In some implementations, the product for detecting the expression levels of each gene in the SRGS in the sample can be a reagent, kit, chip, and / or instrument suitable for detecting the expression levels of each lncRNA and mRNA in the SRGS in the sample by at least one of the following methods: nucleic acid sequencing, nucleic acid hybridization, or nucleic acid amplification.
[0013] In some implementations, the product for detecting the expression levels of each gene in the SRGS in the sample can be a reagent, kit, chip, and / or instrument suitable for detecting the expression levels of each gene in the SRGS in the sample by at least one of the following methods: high-throughput sequencing, in situ hybridization, gene chip, or quantitative real-time PCR.
[0014] In some implementations, the product used to detect the expression levels of each gene in the SRGS sample can be a real-time quantitative PCR kit for detecting the expression levels of each gene in the SRGS sample.
[0015] In this invention, products for assessing the prognosis of prostate cancer include, but are not limited to, reagents, kits, chips, detection systems, or instruments for assessing the prognosis of prostate cancer patients. Attached Figure Description
[0016] Figure 1 Kaplan-Meier curves for BCR of PCa patients based on SRGS prediction in the TCGA-PRAD (n=498), Stockholm (n=92), Cambridge (n=111), CancerMap (n=132), GSE54460 (n=100) and Taylor (n=140) datasets; Figure 2 ROC curves and area under the curve (AUC) for predicting BCR in PCa patients in different datasets based on SRGS. Figure 3 AUC values of time-dependent ROC curves for 1-year, 3-year, and 5-year BCR in PCa patients from different datasets based on SRGS prediction; Figure 4 C-index of SRGS compared to other clinical factors in different datasets; Figure 5 Univariate Cox regression analysis of SRGS with other biomarkers in different datasets; Figure 6 For different datasets, the average C-index of SRGS is compared with that of other biomarkers. The error bars represent the median and 95% confidence interval (CI). Figure 7 Kaplan-Meier survival curves for BCR in PCa patients based on SRGS prediction in the SYSU-PRAD (n=90) dataset; Figure 8 In the SYSU-PRAD dataset, SRGS was used to evaluate the time-dependent ROC curves of BCR at 1, 2 and 3 years for PCa patients; Figures 9-10The following are univariate and multivariate Cox regression analyses of SRGS and other clinical factors in the SYSU-PRAD dataset; Figure 11 Comparison of SRGS with other clinical factors in the SYSU-PRAD dataset using C-index. Detailed Implementation
[0017] The present invention will be further described in detail below with reference to the embodiments. The embodiments are for illustrative purposes only and do not limit the invention in any way. Unless otherwise specified, the raw materials and reagents used in the embodiments are conventional products that can be obtained commercially; experimental methods that do not specify specific conditions in the embodiments are generally performed under conventional conditions in the art or according to the conditions recommended by the manufacturer.
[0018] Example 1 In this embodiment, standard statistical tests were used to analyze clinical data, including the Log-rank test, Kruskal-Wallis test, Wilcoxon test, and Student t-test. Spearman correlation was used to analyze continuously distributed variables. All statistical tests were two-tailed. P <0.05 indicates significance. To correct for multiple tests, adjustments were made based on the Benjamini-Hochberg FDR. P Values. Biochemical relapse-free survival was described using Kaplan-Meier plots (Log-rank test), with the optimal cutoff point determined using the "survminer" package. All clinical data were analyzed using R (version 4.2.2), expressed as mean and standard deviation (SD) of at least three independent replicates, with statistical analysis performed using GraphPad Prism (version 9.0.0).
[0019] 1. Data and Processing PCa datasets containing BCR-free survival time and patient clinical information were retrieved from six public databases (TCGA-PRAD, Stockholm, Cambridge, CancerMap, GSE54460, and Taylor, downloaded on January 12, 2023), resulting in six datasets (correspondingly named TCGA-PRAD, Stockholm, Cambridge, CancerMap, GSE54460, and Taylor) for constructing and validating the biomarker ensemble of the present invention.
[0020] All data are publicly available through the TCGA Research Network Portal (https: / / portal.gdc.cancer.gov / ) and GEO (http: / / www.ncbi.nlm.nih.gov / geo / ). The GTEX dataset can be downloaded from https: / / xenabrowser.net / .
[0021] To verify the effectiveness of SRGS in a clinical setting, this invention collected frozen PCa tissues from 90 clinical PCa patients, along with corresponding clinical information such as biochemical recurrence-free survival time. This internal clinical dataset was named SYSU-PRAD. Specifically, the PCa tissues came from patients who underwent radical prostatectomy without preoperative chemotherapy, radiotherapy, or other treatment interventions. All specimens were pathologically evaluated by professional pathologists. Single-cell RNA sequencing was used to analyze the expression levels of corresponding genes in the tumor tissues of each PCa patient.
[0022] The clinicopathological information of the seven datasets is shown in Table 1.
[0023] Transcriptomic and proteomic analyses were performed by Shanghai Zhongke New Life Biotechnology Co., Ltd. Single-cell RNA sequencing was performed using a 10× Chromium single-cell platform (Lianchuan Biotechnology Co., Ltd., Hangzhou, China). Quantile normalization, subsequent data processing, and bioinformatics analysis of all data were performed using R software (version 4.2.2) and the corresponding R packages.
[0024] Table 17 shows the clinicopathological information of the datasets.
[0025] 2. Construction and prognostic assessment capability verification of the SRGS model (1) Preliminary verification of the construction and prognostic assessment capabilities of the SRGS model S1. Through WGCNA final identification and screening, 38 aging-related genes, including BMP6, CBX2, CCNE1, CCNE2, CHEK1, LMNB1, PCNA, POLD1, and UBE2C, were selected as candidate biomarkers. Using the above 38 candidate biomarkers, 107 prediction models were developed in the TCGA-PRAD dataset based on 10 machine learning methods within the LOOCV framework.
[0026] The 10 machine learning methods are: Cox partial least squares regression (plsRcox), survival support vector machine (Survival-SVM), generalized boosted regression modeling (GBM), random survival forest (RSF), least absolute shrinkage and selection operator (Lasso), step Cox regression (stepCox), ridge regression, CoxBoost, elastic network (Enet), and supervised principal components (SuperPC).
[0027] S2. To ensure consistency, all datasets were standardized before being input into the machine learning models. All models trained on the TCGA-PRAD dataset were externally validated using the same parameters on five independent datasets (Stockholm, Cambridge, CancerMap, GSE54460, and Taylor). The Harrell concordance index (C-index) for each model was calculated, and the model with the highest average C-index was selected as the optimal model. The optimal model combined SuperPC and LASSO regression, with an average C-index of 0.695. In the TCGA-PRAD dataset, the optimal λ value for LASSO regression was determined using the minimum partial likelihood bias within the LOOCV framework.
[0028] The final result was a biomarker group SRGS consisting of nine aging-related genes (BMP6, CBX2, CCNE1, CCNE2, CHEK1, LMNB1, PCNA, POLD1, and UBE2C).
[0029] S3, SRGS evaluation models First, SuperPC modeling was performed on the log2-transformed TCGA data (including data from nine genes: BMP6, CBX2, CCNE1, CCNE2, CHEK1, LMNB1, PCNA, POLD1, and UBE2C): the superpc.train function was used to construct the SuperPC model (based on supervised principal component analysis). Then, 10-fold cross-validation was performed using superpc.cv to select the optimal threshold and number of principal components (n.components), resulting in the trained SuperPC model, which is the SRGS evaluation model.
[0030] The expression levels of BMP6, CBX2, CCNE1, CCNE2, CHEK1, LMNB1, PCNA, POLD1, and UBE2C were obtained from the patient's tumor tissue or peripheral blood. After the expression level data were transformed by log2, the projection score of the expression level data of the nine genes in SRGS onto the above-trained SuperPC model vector was calculated using the superpc.predict function in the superpc package (i.e., the combined gene score after dimensionality reduction). This score is the patient's SRGS score.
[0031] The survminer package is used to determine the optimal threshold (cut-off value) for the SRGS score. The patient's SRGS score is compared with the cut-off value, and the patient is divided into a high-risk group and a low-risk group based on the comparison results. If the SRGS score is greater than the cut-off value, the patient is judged as high-risk; if the SRGS score is less than or equal to the threshold, the patient is judged as low-risk.
[0032] like Figure 1 As shown, in the TCGA-PRAD training set and five independent validation datasets (Stockholm, Cambridge, CancerMap, GSE54460, and Taylor), the BCR outcome of patients in the high-risk SRGS score group was significantly worse than that of patients in the low-risk SRGS score group (all...). P <0.05).
[0033] This demonstrates that by using the biomarker combination provided by this invention, calculating the patient's SRGS score, and then comparing the patient's SRGS score with a threshold (cut-off value), the prognosis of prostate cancer patients can be assessed. If the SRGS score is greater than the threshold, it suggests that the patient's biochemical recurrence-free survival (BCR-free survival, BFS) may be relatively short, the risk of BCR is high, and the prognosis is poor; if the SRGS score is less than or equal to the threshold, it suggests that the patient's risk of BCR is low, and the prognosis is good.
[0034] (2) Evaluate the prognostic assessment capability of SRGS through ROC analysis. The prognostic ability of SRGS was further evaluated based on the area under the curve (AUC) values of the six datasets, and the results are as follows: Figure 2 As shown, the values are TCGA-PRAD (0.709), Stockholm (0.693), Cambridge (0.68), CancerMap (0.714), GSE54460 (0.6551) and Taylor (0.747), respectively.
[0035] Figure 3 The AUC values of the time-dependent ROC curves for 1-year, 3-year, and 5-year BCR in PCa patients based on SRGS prediction are shown in Table 2.
[0036] Figure 2 , Figure 3 The results show that SRGS performs consistently across all datasets and has strong prognostic performance.
[0037] Table 2. AUC values of SRGS at 1 year, 3 years, and 5 years.
[0038] (3) Comparison of the predictive power of SRGS and other clinical factors on the prognosis of PCa This invention compares the predictive power of SRGS with other clinical factors on the prognosis of PCa using the C-index. Figure 4 As shown, in all six datasets, the SRGS provided by this invention has the highest C-index relative to other clinical factors (including Gleason score and T stage) in predicting BFS, demonstrating that the SRGS provided by this invention has superior predictive accuracy and stability.
[0039] (4) Comparison of the predictive ability of SRGS with other biomarkers for PCa prognosis This invention compared the predictive ability of SRGS with 60 biomarkers / biomarker combinations disclosed in the prior art (see Table 3) for PCa prognosis using the C-index, and the results are as follows: Figures 5-6 As shown.
[0040] Univariate Cox regression analysis of the above biomarkers across all datasets showed that SRGS was significantly associated with the prognosis of PCa. Figure 5 This indicates that it has good robustness.
[0041] Furthermore, SRGS outperformed other biomarkers in C-index across all datasets. Unlike many other evaluation models, SRGS, simplified to a nine-gene biomarker combination via LASSO regression and SuperPC, demonstrated better extrapolation potential, reduced the risk of overfitting, and achieved consistently strong predictive performance across multiple datasets. Figure 6 ).
[0042] Table 3. Relevant information for 60 biomarkers
[0043] (5) Validation in the clinical internal dataset (SYSU-PRAD) Using the assessment model of “S3, SRGS assessment model” in “(1) Construction of SRGS model and preliminary verification of prognostic assessment ability”, patients in the SYSU-PRAD dataset were divided into high-risk group and low-risk group.
[0044] like Figure 7 As shown, Kaplan-Meier survival analysis results indicate that patients with high SRGS scores had significantly worse BCR outcomes compared to patients with low SRGS scores. P <0.0001).
[0045] like Figure 8 As shown, the time-dependent ROC curve analysis results indicate that the AUC values of SRGS for assessing the 1-year, 2-year, and 3-year BCR of patients were 0.6969, 0.7559, and 0.7869, respectively, indicating that SRGS has high accuracy and specificity in assessing the prognosis of PCa patients, especially in assessing the BCR outcome of PCa patients.
[0046] like Figures 9-10 As shown, based on univariate and multivariate Cox analyses, SRGS was found to be an independent predictor of BCR in PCa.
[0047] Furthermore, the predictive power of SRGS and other clinical factors on PCa prognosis was compared. Figure 11 As shown, in the SYSU-PRAD dataset, the SRGS provided by this invention has the highest C-index and high predictive accuracy in PCa prognostic assessment compared to other clinical factors (including age, T stage, N stage, Gleason score, and PSA).
[0048] The above descriptions are merely some embodiments of the present invention. Those skilled in the art can make various modifications and improvements without departing from the inventive concept of the present invention, and these all fall within the scope of protection of the present invention.
Claims
1. A combination of biomarkers for assessing the prognosis of prostate cancer, characterized in that, The biomarker combination consists of BMP6, CBX2, CCNE1, CCNE2, CHEK1, LMNB1, PCNA, POLD1, and UBE2C.
2. The use of the biomarker combination according to claim 1 in the preparation of products for assessing the prognosis of prostate cancer.
3. The use of the product for detecting the expression levels of each biomarker in a sample in the biomarker combination according to claim 1 in the preparation of a product for assessing the prognosis of prostate cancer.
4. The application according to claim 3, characterized in that, The sample is tumor tissue or peripheral blood.
5. The application according to claim 3 or 4, characterized in that, The product for detecting the expression level of each biomarker in the biomarker combination according to claim 1 in a sample is a reagent, kit, chip, and / or instrument suitable for detecting the expression level of each biomarker in the biomarker combination according to claim 1 in a sample by at least one of the following methods: nucleic acid sequencing, nucleic acid hybridization, or nucleic acid amplification.
6. The application according to claim 5, characterized in that, The product for detecting the expression level of each biomarker in the biomarker combination according to claim 1 in a sample is a reagent, kit, chip, and / or instrument suitable for detecting the expression level of each biomarker in the biomarker combination according to claim 1 in a sample by at least one of the following methods: high-throughput sequencing, in situ hybridization, gene chip, or quantitative real-time PCR.