A four-gene melanoma prognosis risk assessment model based on ribosis typing screening and a construction method and application thereof

Abstract

The application discloses a four-gene melanoma prognosis risk evaluation model based on RiboSis typing screening and a construction method and application thereof. The four-gene melanoma prognosis risk evaluation model based on RiboSis typing screening is characterized in that the model takes the expression levels of gene markers GBP2, APOBEC3G, HCST and CA14 as input variables, and calculates a risk score according to the following formula: RiskScore=-0.25856323*Z(GBP2)-0.12074882*Z(APOBEC3G)-0.03239219*Z(HCST)+0.02166196*Z(CA14); and melanoma patients are divided into a high-risk group or a low-risk group according to the risk score. The model has the advantages of few genes, convenient detection, good prognosis risk stratification value and clinical transformation potential.

Description

Technical Field

[0001] This invention belongs to the fields of tumor molecular markers, bioinformatics analysis and medical decision support technology, specifically involving a four-gene melanoma prognostic risk assessment model based on RiboSis typing and screening, its construction method and application. Background Technology

[0002] Melanoma is a highly malignant skin tumor characterized by its aggressiveness, high risk of metastasis, and significant prognostic variability. In recent years, the development of immune checkpoint inhibitors and targeted therapies has improved the survival outcomes of some melanoma patients, but significant differences in treatment benefits and overall survival still exist among different patients. Therefore, establishing a molecular model that can accurately assess the prognostic risk of melanoma patients is of great significance for patient stratification, treatment strategy selection, and individualized follow-up.

[0003] Currently, prognostic assessment for melanoma patients mainly relies on traditional clinicopathological indicators such as clinical stage, pathological features, and metastatic status. However, traditional indicators are insufficient to fully reflect the molecular heterogeneity within the tumor, differences in the immune microenvironment, and variations in treatment response. With the development of high-throughput sequencing technology and bioinformatics methods, constructing multi-gene prognostic models based on transcriptome expression profiling has become an important direction for tumor prognostic assessment.

[0004] While existing melanoma prognostic models can stratify risk to some extent, they still have the following shortcomings: First, some models include a large number of genes, resulting in high testing costs and hindering clinical translation; second, some models are built solely based on bulk transcriptome data, lacking support from single-cell-level features; third, some models lack cross-cohort validation or auxiliary assessment of immunotherapy response; and fourth, some screening strategies fail to fully integrate specific molecular biological processes for candidate gene extraction, leaving room for improvement in model stability and interpretability.

[0005] Ribosome biogenesis (RiboSis) is closely related to processes such as tumor cell proliferation, protein synthesis, metabolic remodeling, and stress response. RiboSis typing, based on a set of genes related to ribosome biogenesis, can reflect the specific molecular state of melanoma samples. It should be noted that the final model genes screened based on RiboSis typing are not necessarily limited to core genes of ribosome biogenesis. Its technical significance lies in using RiboSis typing as an entry point for candidate gene screening, combining single-cell characteristics and multi-cohort data to screen gene combinations with stable prognostic value.

[0006] Therefore, there is an urgent need to provide a molecular model that has a small number of genes, is easy to detect, can be validated across cohorts, and can be used for prognostic risk assessment of melanoma and auxiliary assessment of immunotherapy response. Summary of the Invention

[0007] Objectives of the Invention: The first objective of this invention is to provide a four-gene melanoma prognostic risk assessment model based on RiboSis typing; the second objective is to provide a method for constructing the risk assessment model; and the third objective is to provide an application of the risk assessment model. This invention addresses the problems of existing melanoma prognostic assessment models, such as a large number of genes, a single screening source, insufficient model stability, and difficulties in clinical translation.

[0008] This invention combines bulk transcriptome RiboSis typing, single-cell RiboSis activity characteristics, multi-cohort common gene screening, and Cox / LASSO-Cox regression analysis to screen four prognostic-related genes: GBP2, APOBEC3G, HCST, and CA14. A risk scoring model is then constructed for prognostic risk stratification, survival probability prediction, and auxiliary assessment of immunotherapy response in melanoma patients.

[0009] Technical Solution: The present invention discloses a four-gene melanoma prognostic risk assessment model based on RiboSis typing, characterized in that the model uses the expression levels of gene markers GBP2, APOBEC3G, HCST, and CA14 as input variables, and calculates the risk score according to the following formula:

[0010] RiskScore = -0.25856323 × Z(GBP2) - 0.12074882 × Z(APOBEC3G) -0.03239219×Z(HCST)+0.02166196×Z(CA14);

[0011] Wherein, Z(GBP2), Z(APOBEC3G), Z(HCST), and Z(CA14) represent the Z-score standardized expression values ​​of GBP2, APOBEC3G, HCST, and CA14, respectively. Z-score standardization refers to converting the original expression level of the corresponding gene in the test sample or test sample cohort into a standardized expression level. The calculation formula is: Z = (X - μ) / σ, where X represents the expression level of the corresponding gene in the test sample, μ represents the mean expression of the corresponding gene in the test sample cohort, training cohort, or reference cohort, and σ represents the standard deviation of the expression of the corresponding gene in the test sample cohort, training cohort, or reference cohort. After Z-score standardization, the differences in expression levels between different genes and different detection platforms are converted into a standardized deviation relative to the reference distribution, thus facilitating the inclusion of the four genes GBP2, APOBEC3G, HCST, and CA14 in the same risk scoring formula for calculation.

[0012] Furthermore, melanoma patients are divided into high-risk and low-risk groups based on the RiskScore; when the RiskScore is greater than a preset threshold, the patient is divided into the high-risk group; when the RiskScore is less than or equal to the preset threshold, the patient is divided into the low-risk group, where the preset threshold is 0.

[0013] Furthermore, the gene markers GBP2, APOBEC3G, HCST, and CA14 are a combination of melanoma prognosis-related model genes obtained through the RiboSis typing screening process; wherein, RiboSis refers to genes related to ribosome biosynthesis.

[0014] Furthermore, the RiboSis typing screening includes: calculating the RiboSis score of melanoma samples based on the set of genes related to ribosome biosynthesis, typing the melanoma samples according to the RiboSis score, and screening for GBP2, APOBEC3G, HCST and CA14 by combining single-cell RiboSis activity characteristics, screening for common genes in multiple cohorts, univariate Cox regression analysis and LASSO-Cox regression analysis.

[0015] Furthermore, the prognostic risk assessment includes overall survival risk assessment, 1-year overall survival probability prediction, 3-year overall survival probability prediction, and 5-year overall survival probability prediction for melanoma patients.

[0016] The method for constructing a four-gene melanoma prognostic risk assessment model based on RiboSis typing and screening, as described in this invention, includes the following steps:

[0017] S1. Obtain bulk transcriptome expression data, clinical survival data, and single-cell RNA sequencing data from melanoma patients;

[0018] S2. Calculate the RiboSisscore of bulk melanoma samples based on the ribosome biosynthesis-related gene set, and divide the melanoma samples into High_RiboSis group and Low_RiboSis group according to the RiboSis score.

[0019] S3. Differential expression analysis was performed on the High_RiboSis group and the Low_RiboSis group to obtain differentially expressed genes of RiboSis subtype in the bulk transcriptome.

[0020] S4. Calculate the single-cell RiboSis activity score based on single-cell RNA sequencing data, and screen characteristic genes related to single-cell RiboSis activity.

[0021] S5. Intersection screening is performed on the differentially expressed genes of RiboSis subtypes in the bulk transcriptome, the characteristic genes related to RiboSis activity in single cells, and the common genes of multiple melanoma cohorts to obtain candidate genes.

[0022] S6. Perform univariate Cox regression analysis and LASSO-Cox regression analysis on the candidate genes to screen out GBP2, APOBEC3G, HCST and CA14;

[0023] S7. Based on the standardized expression levels and corresponding regression coefficients of GBP2, APOBEC3G, HCST and CA14, a prognostic risk assessment model for melanoma was constructed.

[0024] The present invention relates to the application of the RiboSis-based four-gene melanoma prognostic risk assessment model in a prepared melanoma patient prognostic risk assessment product. The product includes: reagents for detecting the expression levels of gene markers GBP2, APOBEC3G, HCST, and CA14, and instructions for using the risk assessment model. The instructions state: obtain the expression levels of GBP2, APOBEC3G, HCST, and CA14; standardize the expression levels using Z-scores; calculate the melanoma patient prognostic risk score (RiskScore) according to the formula of the RiboSis-based four-gene melanoma prognostic risk assessment model; and use 0 as a preset threshold. When the RiskScore is greater than 0, the patient is classified as a high-risk group; when the RiskScore is less than or equal to 0, the patient is classified as a low-risk group.

[0025] Preferably, the product for prognostic risk assessment of melanoma patients includes reagent kits, chips, test strips, high-throughput sequencing platforms, real-time quantitative PCR detection reagents, nucleic acid probe combinations, protein detection products, or computer-aided analysis systems.

[0026] The reagents for detecting the expression levels of the gene markers GBP2, APOBEC3G, HCST, and CA14 include: primers, probes, sequencing adapters, amplification reagents, hybridization probes, nucleic acid chip probes, or sequencing library construction reagents for detecting the mRNA expression levels of GBP2, APOBEC3G, HCST, and CA14, or antibodies, peptides, aptamers, compounds, protein chip reagents, or immunoassay compositions for detecting the corresponding protein expression levels of GBP2, APOBEC3G, HCST, and CA14.

[0027] The in vitro diagnostic or non-diagnostic melanoma patient prognostic risk assessment system based on the RiboSis subtyping screening four-gene melanoma prognostic risk assessment model, as described in this invention, includes the following modules:

[0028] The expression data acquisition module is used to acquire the expression levels of GBP2, APOBEC3G, HCST and CA14 in the sample to be tested;

[0029] The standardization module is used to perform Z-score standardization on the expression levels of GBP2, APOBEC3G, HCST and CA14 according to Z=(X-μ) / σ, where X is the expression level of the corresponding gene, μ is the mean expression of the corresponding gene in the test sample cohort, training cohort or reference cohort, and σ is the standard deviation of the expression of the corresponding gene in the test sample cohort, training cohort or reference cohort.

[0030] The risk score calculation module is used to calculate the RiskScore according to the following formula:

[0031] RiskScore=-0.25856323×Z(GBP2)-0.12074882×Z(APOBEC3G)-0.03239219×Z(HCST) +0.02166196×Z(CA14); where Z(GBP2), Z(APOBEC3G), Z(HCST) and Z(CA14) represent the Z-score normalized expression values ​​of GBP2, APOBEC3G, HCST and CA14, respectively.

[0032] The risk stratification module is used to classify patients with melanoma under test into a high-risk group when the RiskScore is greater than 0, and into a low-risk group when the RiskScore is less than or equal to 0.

[0033] The results output module is used to output the prognostic risk assessment results for melanoma patients.

[0034] The present invention relates to a prognostic risk assessment model for melanoma based on RiboSis typing, or a method for constructing such a model, or a prognostic risk assessment product for melanoma patients prepared using the RiboSis-based four-gene melanoma prognostic risk assessment model, or the application of such an in vitro diagnostic or non-diagnostic melanoma patient prognostic risk assessment system in the preparation of an auxiliary assessment product for melanoma immunotherapy response.

[0035] Furthermore, the immunotherapy includes anti-PD-1 therapy, anti-PD-L1 therapy, anti-CTLA-4 therapy, or a combination thereof.

[0036] This invention provides the application of a reagent for detecting the expression level of gene markers in the preparation of products for prognostic risk assessment of melanoma patients, wherein the gene markers include GBP2, APOBEC3G, HCST and CA14.

[0037] The kit for prognostic risk assessment of melanoma patients described in this invention includes reagents for detecting the expression levels of GBP2, APOBEC3G, HCST and CA14, as well as instructions for use. The instructions for use describe how to calculate the prognostic risk score of melanoma patients based on the expression levels of GBP2, APOBEC3G, HCST and CA14, and how to stratify the prognostic risk of patients based on the prognostic risk score.

[0038] Preferably, the prognostic risk assessment includes overall survival risk assessment, prognostic risk stratification, 1-year overall survival probability prediction, 3-year overall survival probability prediction, and 5-year overall survival probability prediction for melanoma patients.

[0039] Preferably, the expression level of the gene marker can be obtained by RNA sequencing, real-time quantitative PCR, reverse transcription PCR, digital PCR, nucleic acid microarray, in situ hybridization, NanoString, Western blotting, enzyme-linked immunosorbent assay (ELISA), immunohistochemical staining, immunofluorescence staining, mass spectrometry, or protein microarray detection.

[0040] The design principle of this invention lies in the fact that melanoma exhibits significant molecular heterogeneity, and traditional clinical staging or single transcriptome differential analysis is insufficient to fully reflect prognostic differences among patients. This invention uses the RiboSis classification as its upstream biological basis, combining TCGA melanoma bulk transcriptome, single-cell transcriptome, and multiple external bulk expression cohorts to screen stable candidate genes associated with RiboSis classification, malignant cell status, and overall patient survival. Furthermore, through univariate Cox regression and LASSO-Cox regression compression models, a four-gene prognostic risk assessment model consisting of GBP2, APOBEC3G, HCST, and CA14 is finally constructed.

[0041] Based on the above design principles, this invention further transforms the candidate gene screening results related to RiboSis genotyping into a computable, stratified, and productizable four-gene risk assessment scheme. This scheme uses the standardized expression values ​​of GBP2, APOBEC3G, HCST, and CA14 as inputs, employs a fixed-coefficient RiskScore formula as the calculation rule, and outputs high-risk or low-risk group results with a uniform preset threshold of 0, thus forming a complete technical path from gene expression detection and risk score calculation to prognostic risk stratification.

[0042] The core of this invention lies in identifying a four-gene biomarker combination consisting of GBP2, APOBEC3G, HCST, and CA14, and establishing a fixed-coefficient RiskScore calculation formula and a risk stratification standard with a threshold of 0. This model uses the Z-score-normalized expression values ​​of the four genes to calculate a prognostic risk score for melanoma patients. A RiskScore greater than 0 indicates a high-risk group, while a RiskScore less than or equal to 0 indicates a low-risk group.

[0043] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0044] 1. Fewer model genes, making detection and translation easier: The model of this invention contains only four genes: GBP2, APOBEC3G, HCST and CA14. Compared with prognostic models containing more genes, the detection cost is lower, the data processing is simpler, and it is more suitable for subsequent reagent kit development and clinical auxiliary evaluation applications.

[0045] 2. The screening strategy is more stable and hierarchical: This invention uses RiboSis typing as the entry point for candidate gene screening, and combines bulk transcriptomics, single-cell RiboSis activity characteristics and multi-cohort common gene screening to avoid relying on a single data source for modeling, thereby improving the stability of candidate gene screening.

[0046] 3. Effective prognostic risk stratification for melanoma: The RiskScore constructed in this invention can divide melanoma patients into high-risk and low-risk groups. Patients in the high-risk group have poorer overall survival, while patients in the low-risk group have better overall survival, demonstrating good prognostic risk stratification capabilities.

[0047] 4. It has value in assisting the assessment of immunotherapy response: The model of this invention can not only be used for overall survival risk assessment, but also for assisting the assessment of melanoma immunotherapy response, providing a reference for the formulation of individualized treatment strategies for patients.

[0048] 5. Applicable to a variety of product forms: This invention can be prepared as a reagent kit, chip, detection product or computer-aided analysis system, and is suitable for various detection platforms such as RNA sequencing, qRT-PCR, digital PCR, nucleic acid chip, and immunohistochemistry. Attached Figure Description

[0049] Figure 1The following are the candidate gene screening results for the melanoma prognostic risk assessment model based on RiboSis classification in the examples; where A is the survival analysis results of TCGA melanoma patients divided into high and low RiboSis groups according to RiboSis score; B is the t-SNE distribution results of RiboSis-High and RiboSis-Low cells in single-cell transcriptome data; and C is the intersection screening results of single-cell related genes, differentially expressed genes related to RiboSis classification in the TCGA cohort, and common genes in multiple bulk cohorts.

[0050] Figure 2 The results of the LASSO-Cox regression screening of the four-gene model in the example are shown; where A is the LASSO regression coefficient path diagram; B is the LASSO-Cox regression cross-validation curve and the optimal model gene screening results.

[0051] Figure 3 The following are the prognostic validation results of the four-gene RiskScore in the TCGA melanoma training cohort in the examples; where A is the survival analysis results of patients in the high-risk and low-risk groups; and B is the ROC curve of the four-gene RiskScore predicting the 1-year, 3-year, and 5-year overall survival.

[0052] Figure 4 The following are the prognostic validation results of the four-gene RiskScore in the external validation cohort in the example; where A, B, and C are the survival analysis results of patients in the high-risk and low-risk groups in the GSE65904, GSE54467, and GSE19234 cohorts, respectively; and D is the AUC comparison result of the four-gene RiskScore predicting 1-year, 3-year, and 5-year overall survival in the TCGA cohort and the external validation cohort.

[0053] Figure 5 The following are the validation results of the application of the four-gene RiskScore in the assessment of immunotherapy response in the examples; where A is the distribution of the proportion of immunotherapy responders and non-responders in different risk groups; and B is the ROC curve of the four-gene RiskScore distinguishing between immunotherapy responders and non-responders. Detailed Implementation

[0054] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings. Obviously, the described embodiments are only for illustrating the present invention and are not intended to limit the scope of protection of the present invention. Other implementation methods obtained by those skilled in the art based on the embodiments of the present invention without creative effort should all fall within the scope of protection of the present invention.

[0055] Unless otherwise stated, TCGA in this invention refers to The Cancer Genome Atlas database, GEO refers to the Gene Expression Omnibus database, TCGA-SKCM refers to the cutaneous melanoma cohort in the TCGA database; bulk transcriptome data refers to transcriptome expression data derived from whole tissue samples; single-cell transcriptome data refers to gene expression data obtained at single-cell resolution.

[0056] In this invention, RiboSis refers to a scoring, activity characterization, or typing system constructed based on a set of genes related to ribosome biosynthesis. The RiboSis score is a sample score calculated based on the RiboSis-related gene set at the bulk transcriptome sample level, used to characterize the RiboSis-related activity level in the overall tissue sample. The High_RiboSis group and Low_RiboSis group refer to the high RiboSis score group and low RiboSis score group obtained after typing or grouping melanoma samples according to the RiboSis score at the bulk transcriptome sample level.

[0057] The single-cell RiboSis activity score refers to a cell score calculated based on the RiboSis-related gene set at the single-cell transcriptome level, used to characterize the RiboSis-related activity level in a single cell. In one specific embodiment, the single-cell RiboSis activity score is calculated using the AUCell method. The RiboSis-High and RiboSis-Low cell subpopulations refer to the high-RiboSis-activity cell subpopulation and low-RiboSis-activity cell subpopulation obtained by grouping single cells according to the single-cell RiboSis activity score.

[0058] RiskScore refers to a continuous risk score calculated based on the Z-score standardized expression values ​​of GBP2, APOBEC3G, HCST, and CA14 and fixed coefficients; Risk stratification refers to classifying samples with RiskScore>0 into a high-risk group and samples with RiskScore≤0 into a low-risk group, with RiskScore=0 as a preset threshold.

[0059] Example 1: RiboSis typing and candidate gene screening

[0060] This embodiment illustrates the method for screening candidate genes based on RiboSis typing and single-cell RiboSis activity characteristics in this invention.

[0061] 1.1 Data Source

[0062] Bulk transcriptome expression data and corresponding clinical survival information were obtained from melanoma patients. Bulk transcriptome data refers to transcriptome expression data derived from whole tumor tissue samples. This bulk transcriptome data included TCGA-SKCM cohort data from the TCGA database (https: / / portal.gdc.cancer.gov / ) and multiple external melanoma validation cohorts from the GEO database (https: / / www.ncbi.nlm.nih.gov / geo / ), including GSE65904, GSE54467, GSE19234, and the MAGE-A3 immunotherapy cohort GSE35640. The expression matrices were preprocessed, and samples with complete overall survival time and survival status information were retained. Single-cell RNA sequencing data were obtained from the publicly published melanoma single-cell transcriptome dataset GSE215120, and the ribosome biosynthesis-related gene set was derived from the ribosome biosynthesis-related functional gene set reported in the literature (PMID: 38343327).

[0063] 1.2 RiboSis ssGSEA Score Calculation

[0064] Based on the pre-organized RiboSis gene set (derived from the set of ribosome biosynthesis-related functional genes reported in the literature (PMID: 38343327) mentioned in 1.1), the RiboSis score of each melanoma sample in the TCGA-SKCM cohort was calculated using the single-sample gene set enrichment analysis ssGSEA method.

[0065] 1.3 K-means clustering classification

[0066] K-means clustering was performed on the TCGA-SKCM melanoma samples based on the RiboSis score, dividing patients into High_RiboSis and Low_RiboSis groups. The TCGA-SKCM cohort, derived from the TCGA database, serves as the bulk feature selection cohort and model training set in this invention.

[0067] 1.4 Survival Analysis

[0068] Kaplan-Meier survival analysis was used to assess the difference in overall survival between the High_RiboSis group and the Low_RiboSis group, and the log-rank test was used to compare the survival curves of the two groups.

[0069] Figure 1The results showed that the overall survival curve of patients in the High_RiboSis group was lower than that of the Low_RiboSis group, indicating that patients in the High_RiboSis group had poorer overall survival. The log-rank test showed that the difference in overall survival between the two groups was statistically significant (P=0.0028). These results demonstrate that the RiboSis subtyping based on the RiboSis ssGSEA score is significantly associated with overall survival in melanoma patients and can serve as a molecular subtyping basis for subsequent candidate gene screening.

[0070] 1.5 Bulk Data Differential Expression

[0071] Differential expression analysis was performed on the High_RiboSis and Low_RiboSis groups using the limma method to identify bulk differentially expressed genes associated with RiboSis genotyping. The selection criteria for differentially expressed genes were: adj.P.Val < 0.05 and |logFC| > 1.

[0072] 1.6 Constructing a single-cell RiboSis feature set

[0073] To identify key molecular features associated with RiboSis activity at the single-cell level, after obtaining melanoma single-cell RNA sequencing data, the single-cell expression matrix was first preprocessed and clustered for annotation. The RiboSis gene set was then matched with the detection genes in the single-cell expression matrix to construct a feature gene set for subsequent scoring.

[0074] The RiboSis activity of each cell was scored using the AUCell method. Figure 1 As shown in Figure B, based on the AUCell score of each cell, single cells were divided into RiboSis-High and RiboSis-Low cell subpopulations. The FindMarkers function in the Seurat package was used to perform differential expression analysis on the RiboSis-High and RiboSis-Low cell subpopulations to identify molecular characteristics under different RiboSis activity states. The differential expression analysis employed the Wilcoxon rank-sum test, combined with screening based on expression ratio and fold change. After multiple validation corrections, genes meeting the condition FDR < 0.05 were defined as significantly differentially expressed genes.

[0075] Based on the single-cell AUCell score, Spearman rank correlation analysis was performed on the genome-wide expression levels and single-cell RiboSis activity scores. After the P-values ​​were corrected using the BH method, significantly correlated genes were screened and sorted by correlation coefficient. The top 150 genes with the most significant positive correlation to single-cell RiboSis activity scores were extracted as the RiboSis continuously correlated gene set.

[0076] By combining the differentially expressed genes with related genes, a single-cell RiboSis feature set was constructed. This feature set integrates grouped differential information and continuous covariation information related to RiboSis activity, and can more comprehensively reflect the molecular characteristics of RiboSis at the single-cell level.

[0077] 1.7 Candidate Gene Construction

[0078] By intersecting and screening differentially expressed genes from bulk transcriptome RiboSis subtypes, single-cell RiboSis feature sets, and common expressed genes from multiple melanoma cohorts, a set of candidate genes for subsequent prognostic modeling was obtained.

[0079] like Figure 1 As shown in C, through the above screening process, 32 candidate genes related to melanoma prognosis were finally selected through intersection screening, providing a foundation for the subsequent construction of a stable prognostic risk model.

[0080] Example 2: Construction of a four-gene prognostic risk assessment model

[0081] This embodiment illustrates the construction process of the four-gene melanoma prognostic risk assessment model of the present invention.

[0082] Based on the candidate gene set obtained in Example 1, and combined with the overall survival information of melanoma patients, univariate Cox regression analysis was performed on the candidate genes to screen prognostic candidate genes that were significantly associated with overall survival. Further LASSO-Cox regression analysis was used to reduce the dimensionality of the candidate genes, resulting in four genes ultimately used to construct the model: GBP2, APOBEC3G, HCST, and CA14.

[0083] The four genes are human genes, including GBP2 (guanylate binding protein 2, NCBI Gene ID 2634), APOBEC3G (apolipoprotein B mRNA editing enzyme catalytic subunit 3G, NCBI Gene ID 60489), HCST (hematopoietic cell signal transducer, NCBI Gene ID 10870), and CA14 (carbonic anhydrase 14, NCBI Gene ID 23632). The nucleic acid and transcript sequences of these genes can be obtained from the NCBI Gene database or the RefSeq database. The preferred expression level described in this invention is the mRNA expression level of the above genes. This invention is not limited to a specific detection platform or specific transcripts, as long as the expression levels of GBP2, APOBEC3G, HCST, and CA14 can be obtained and used for RiskScore calculation.

[0084] As attached Figure 2 As shown, LASSO-Cox regression analysis ultimately identified GBP2, APOBEC3G, HCST, and CA14 as input variables for the melanoma prognostic risk assessment model. Based on the standardized expression values ​​of the four genes and their corresponding regression coefficients, the following risk scoring model was established:

[0085] RiskScore=-0.25856323×Z(GBP2)-0.12074882×Z(APOBEC3G)-0.03239219×Z(HCST)+0.02166196 ×Z(CA14); where Z(GBP2), Z(APOBEC3G), Z(HCST) and Z(CA14) represent the normalized expression values ​​of GBP2, APOBEC3G, HCST and CA14, respectively.

[0086] In a preferred embodiment, the standardized expression value is the Z-score standardized expression value. Specifically, for an expression matrix containing multiple melanoma samples to be tested, the mean and standard deviation of the expression values ​​of each gene in GBP2, APOBEC3G, HCST, and CA14 in the same sample cohort are calculated, and then standardized according to the following formula:

[0087] Z(Gene) = (X(Gene) - μ(Gene)) / σ(Gene); where Gene is GBP2, APOBEC3G, HCST, or CA14; X(Gene) represents the expression value of the corresponding gene in the test sample; μ(Gene) represents the mean expression value of the gene in the same test sample cohort; and σ(Gene) represents the standard deviation of the gene expression in the same test sample cohort. When the standard deviation of a gene expression in the same test sample cohort is 0 or a missing value is generated after standardization, the Z-score value corresponding to that gene is set to 0. After Z-score standardization, the obtained Z(GBP2), Z(APOBEC3G), Z(HCST), and Z(CA14) are used to calculate the RiskScore according to the above formula.

[0088] Example 3: Prognostic validation of the four-gene model in the TCGA training cohort

[0089] This embodiment is used to illustrate the prognostic assessment capability of the four-gene RiskScore of the present invention in the TCGA melanoma training cohort.

[0090] The RiskScore for each patient in the TCGA melanoma cohort was calculated according to the formula described in Example 2, and patients were divided into high-risk and low-risk groups based on a preset threshold of 0. Subsequently, overall survival analysis was performed on patients in the high-risk and low-risk groups, and the predictive ability of the RiskScore for overall patient survival was evaluated.

[0091] like Figure 3 As shown in Figure A, patients with TCGA-SKCM melanoma can be divided into high-risk and low-risk groups based on the four-gene RiskScore. Kaplan-Meier survival analysis results showed that the overall survival curve of patients in the high-risk group was lower than that of patients in the low-risk group, indicating that the overall survival of patients in the high-risk group was worse. The log-rank test results showed that the difference in overall survival between the two groups was statistically significant (P < 0.0001).

[0092] Furthermore, time-dependent ROC curves were used to evaluate the predictive ability of RiskScore for overall survival at different time points. For example... Figure 3 As shown in B, the four-gene model of this invention has the ability to predict the 1-year, 3-year and 5-year overall survival of TCGA-SKCM melanoma patients, with AUCs of 0.654, 0.674 and 0.686 for predicting 1-year, 3-year and 5-year overall survival, respectively.

[0093] Example 4: Validation of the four-gene model in an external independent melanoma cohort

[0094] This embodiment is used to illustrate the prognostic predictive ability and cross-cohort stability of the four-gene RiskScore of the present invention in an external independent melanoma cohort.

[0095] The four-gene risk scoring model described in Example 2 was applied to independent external validation cohorts of melanoma, including GSE65904, GSE54467, and GSE19234. The RiskScore was calculated for each patient, and patients were divided into high-risk and low-risk groups based on their RiskScore. Subsequently, overall survival analysis was performed on patients in different risk groups, and time-dependent ROC curves were used to evaluate the model's predictive ability for overall survival in different cohorts.

[0096] like Figure 4 As shown in Figure A, in the GSE65904 cohort, melanoma patients were divided into high-risk and low-risk groups based on the four-gene RiskScore. Kaplan-Meier survival analysis showed that the overall survival curve of patients in the high-risk group was lower than that of patients in the low-risk group, and the log-rank test showed that the difference in overall survival between the two groups was statistically significant (P=0.0062).

[0097] like Figure 4 As shown in Figure B, in the GSE54467 cohort, the overall survival curve for patients in the high-risk group was also lower than that for patients in the low-risk group, suggesting a poorer prognosis for patients in the high-risk group. The log-rank test results showed that the difference in overall survival between the two groups was statistically significant (P=0.0086).

[0098] like Figure 4 As shown in C, in the GSE19234 cohort, the four-gene RiskScore was also able to distinguish patients with different prognostic risks. The overall survival of patients in the high-risk group was lower than that of patients in the low-risk group. The log-rank test results showed that the difference in overall survival between the two groups was statistically significant (P=0.0015).

[0099] Furthermore, such as Figure 4 As shown in Figure D, the time-dependent AUC was used to evaluate the predictive ability of the four-gene RiskScore for overall survival in different cohorts. The results showed that in cohorts GSE19234, GSE54467, and GSE65904, the RiskScore demonstrated predictive ability for overall survival at different time points. Specifically, in cohort GSE19234, the predictive AUCs for 1-year and 3-year overall survival were approximately 0.810 and 0.967, respectively; in cohort GSE54467, the predictive AUCs for 1-year, 3-year, and 5-year overall survival were approximately 0.718, 0.679, and 0.670, respectively; and in cohort GSE65904, the predictive AUCs for 1-year, 3-year, and 5-year overall survival were approximately 0.669, 0.688, and 0.624, respectively.

[0100] The above results demonstrate that the four-gene RiskScore of this invention can stratify the overall survival of patients in multiple external independent melanoma cohorts and has predictive ability for overall survival at different time points, indicating that the model has good cross-cohort stability and generalization ability.

[0101] Example 5: Validation of the application of the four-gene model in an immunotherapy cohort

[0102] This embodiment illustrates the application value of the four-gene RiskScore of the present invention in a melanoma immunotherapy cohort, especially its ability to assist in the evaluation of immunotherapy response.

[0103] The four-gene risk scoring model described in Example 2 was applied to the melanoma immunotherapy-related cohort GSE35640. This cohort is an MAGE-A3-related immunotherapy cohort. In this example, samples were grouped according to their treatment response status recorded in the cohort. Samples with a treatment response status marked as R were defined as immunotherapy responders; samples with a treatment response status marked as NR were defined as immunotherapy non-responders; and samples with a treatment response status marked as NE were not included in the immunotherapy response analysis because their efficacy was not evaluable.

[0104] After obtaining the expression levels of GBP2, APOBEC3G, HCST, and CA14 in the GSE35640 cohort, the expression levels of these genes were Z-score normalized, and the RiskScore for each patient was calculated according to the formula described in Example 2. Using 0 as a preset threshold, patients were divided into high-risk and low-risk groups based on their RiskScore, where RiskScore > 0 was defined as the high-risk group, and RiskScore ≤ 0 was defined as the low-risk group.

[0105] Subsequently, the proportions of Responders and Non-responders in the high-risk and low-risk groups were compared, and ROC curves were used to evaluate the ability of RiskScore to distinguish the response status of immunotherapy.

[0106] like Figure 5 As shown in Figure A, the proportion of responders was higher in the low-risk group than in the high-risk group, while the proportion of non-responders was higher in the high-risk group. This result suggests that different risk groups based on the four-gene RiskScore are associated with immunotherapy response status, and patients in the low-risk group may have a higher proportion of immunotherapy responses.

[0107] Furthermore, such as Figure 5As shown in Figure B, the ability of the RiskScore to distinguish the response status of immunotherapy was evaluated using ROC curves. The results showed that the area under the ROC curve (AUC) was 0.659. This result indicates that the four-gene RiskScore of this invention has a certain ability to distinguish the response status of melanoma patients to immunotherapy and can be used as a reference indicator for auxiliary assessment of melanoma immunotherapy response.

[0108] Therefore, this embodiment shows that the four-gene RiskScore of the present invention can not only be used for overall survival risk assessment and prognostic risk stratification of melanoma patients, but also for auxiliary assessment of the likelihood of response to immunotherapy, but its results do not replace clinical efficacy evaluation alone.

[0109] Example 6: Reagent Kit and Computer-Aided Analysis System

[0110] This embodiment illustrates the application of the four-gene model of the present invention in reagent kits and computer-aided analysis systems.

[0111] This invention provides a kit for prognostic risk assessment of melanoma patients. The kit includes reagents for detecting the expression levels of GBP2, APOBEC3G, HCST and CA14, as well as instructions for use.

[0112] The reagents used to detect the expression levels of GBP2, APOBEC3G, HCST, and CA14 can be reagents for detecting their mRNA expression levels or reagents for detecting their corresponding protein expression levels. Preferably, the detection methods include RNA sequencing, real-time quantitative PCR, digital PCR, nucleic acid microarray, NanoString, immunohistochemical staining, or protein microarray detection.

[0113] The instruction manual describes the following: Obtain the expression levels of GBP2, APOBEC3G, HCST, and CA14 in the melanoma sample to be tested; standardize the above expression levels; calculate the RiskScore according to the following formula:

[0114] RiskScore=-0.25856323×Z(GBP2)-0.12074882×Z(APOBEC3G)-0.03239219×Z(HCST)+0.02166196×Z(CA14);

[0115] The study stratified melanoma patients based on their prognostic risk using the RiskScore. The RiskScore is a continuous risk score calculated using Z-score normalized expression values ​​of four genes and fixed coefficients; a threshold of 0 was used, with RiskScore > 0 indicating a high-risk group and RiskScore ≤ 0 indicating a low-risk group. The output included the RiskScore value and the high / low-risk stratification results, and could be used for overall survival risk assessment and auxiliary evaluation of immunotherapy response.

[0116] This invention provides a prognostic risk assessment system for melanoma patients for in vitro diagnostic or non-diagnostic purposes, including an expression data acquisition module, a standardization module, a risk score calculation module, a risk stratification module, and a result output module.

[0117] The expression data acquisition module is used to obtain the expression levels of GBP2, APOBEC3G, HCST, and CA14 in the melanoma sample to be tested. These expression levels can be mRNA expression levels obtained through RNA sequencing, gene chips, real-time quantitative PCR, digital PCR, NanoString, or other transcriptional level detection methods.

[0118] The standardization module is used to perform Z-score standardization on the above expression levels according to Z=(X-μ) / σ, where X represents the expression level of the corresponding gene in the test sample, and μ and σ represent the mean and standard deviation of the expression of the corresponding gene in the test sample queue, training queue, or reference queue, respectively.

[0119] The risk score calculation module is used to calculate the RiskScore according to the four-gene formula of this invention.

[0120] The risk stratification module is used to classify patients into high-risk or low-risk groups based on the RiskScore and a preset threshold of 0. Specifically, when the RiskScore is greater than 0, the patient is classified as a high-risk group; when the RiskScore is less than or equal to 0, the patient is classified as a low-risk group.

[0121] The results output module is used to output the prognostic risk assessment results for melanoma patients. The prognostic risk assessment results include RiskScore values, stratified results of high-risk or low-risk groups, and optional overall survival risk indicators, follow-up priority indicators, or treatment response auxiliary assessment indicators.

[0122] In one optional implementation, the immunotherapy response auxiliary assessment prompt is generated based on the RiskScore value or high / low risk stratification results. When the RiskScore ≤ 0, the patient is classified into the low-risk group, and the system can output an auxiliary prompt indicating a higher probability of immunotherapy response; when the RiskScore > 0, the patient is classified into the high-risk group, and the system can output an auxiliary prompt indicating a relatively lower probability of immunotherapy response. The immunotherapy response auxiliary assessment prompt is only intended as an auxiliary reference for treatment decision-making and does not replace clinical efficacy evaluation or the physician's comprehensive judgment.

[0123] The aforementioned kits and computer-aided analysis systems can be used for overall survival risk assessment, survival probability prediction, prognostic risk stratification, and auxiliary assessment of immunotherapy response in melanoma patients.

[0124] The four-gene melanoma prognostic risk assessment model based on RiboSis typing provided by this invention only requires detecting the expression levels of four genes: GBP2, APOBEC3G, HCST, and CA14. It can then calculate the RiskScore and stratify the prognostic risk of melanoma patients. This model has a small number of genes, flexible detection methods, and a simple calculation process, and is compatible with various detection platforms such as RNA sequencing, qRT-PCR, digital PCR, nucleic acid microarrays, NanoString, and immunohistochemical staining.

[0125] This invention can be further developed into a detection kit, chip, molecular diagnostic auxiliary product, or computer-aided analysis system for overall survival risk assessment, survival probability prediction, prognostic stratification management, and auxiliary assessment of immunotherapy response in melanoma patients, and has good clinical translation potential and industrial application value.

[0126] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, any modifications, equivalent substitutions, or improvements made to the technical solutions of the present invention without departing from the spirit and principles thereof should be included within the scope of protection of the present invention.

Claims

1. A prognostic risk assessment model for four-gene melanoma based on RiboSis typing, characterized in that, The model uses the expression levels of the gene markers GBP2, APOBEC3G, HCST, and CA14 as input variables and calculates a risk score according to the following formula: RiskScore = -0.25856323×Z(GBP2) - 0.12074882×Z(APOBEC3G) - 0.03239219×Z(HCST) + 0.02166196×Z(CA14); Wherein, Z(GBP2), Z(APOBEC3G), Z(HCST), and Z(CA14) represent the Z-score standardized expression values ​​of GBP2, APOBEC3G, HCST, and CA14, respectively; the Z-score standardized expression values ​​are calculated according to the formula Z=(X-μ) / σ, where X is the expression level of the corresponding gene, μ is the mean expression of the corresponding gene in the test sample cohort, training cohort, or reference cohort, and σ is the standard deviation of the expression of the corresponding gene in the test sample cohort, training cohort, or reference cohort; the RiskScore divides melanoma patients into high-risk and low-risk groups; when the RiskScore is greater than a preset threshold, the patient is classified into the high-risk group; when the RiskScore is less than or equal to the preset threshold, the patient is classified into the low-risk group, and the preset threshold is 0.

2. The four-gene melanoma prognostic risk assessment model based on RiboSis typing and screening according to claim 1, characterized in that, The RiboSis typing screening includes: calculating the RiboSis score of melanoma samples based on the set of genes related to ribosome biosynthesis, typing the melanoma samples according to the RiboSis score, and screening for GBP2, APOBEC3G, HCST and CA14 by combining single-cell RiboSis activity characteristics, screening for common genes in multiple cohorts, univariate Cox regression analysis and LASSO-Cox regression analysis.

3. The prognostic risk assessment model for four-gene melanoma based on RiboSis typing according to claim 1, characterized in that, The prognostic risk assessment includes overall survival risk assessment, 1-year overall survival probability prediction, 3-year overall survival probability prediction, and 5-year overall survival probability prediction for melanoma patients.

4. A method for constructing a four-gene melanoma prognostic risk assessment model based on RiboSis typing as described in claim 1, characterized in that, Includes the following steps: S1. Obtain bulk transcriptome expression data, clinical survival data, and single-cell RNA sequencing data from melanoma patients; S2. Calculate the RiboSis score of bulk melanoma samples based on the ribosome biosynthesis-related gene set, and divide the melanoma samples into High_RiboSis group and Low_RiboSis group according to the RiboSis score. S3. Differential expression analysis was performed on the High_RiboSis group and the Low_RiboSis group to obtain differentially expressed genes of RiboSis subtype in the bulk transcriptome. S4. Calculate the single-cell RiboSis activity score based on single-cell RNA sequencing data, and screen characteristic genes related to single-cell RiboSis activity. S5. Intersection screening is performed on the differentially expressed genes of RiboSis subtypes in the bulk transcriptome, the characteristic genes related to RiboSis activity in single cells, and the common genes of multiple melanoma cohorts to obtain candidate genes. S6. Perform univariate Cox regression analysis and LASSO-Cox regression analysis on the candidate genes to screen out GBP2, APOBEC3G, HCST and CA14; S7. Based on the standardized expression levels and corresponding regression coefficients of GBP2, APOBEC3G, HCST and CA14, a prognostic risk assessment model for melanoma was constructed.

5. The application of a four-gene melanoma prognostic risk assessment model based on RiboSis typing and screening as described in claim 1 in the preparation of a melanoma patient prognostic risk assessment product, characterized in that, The product includes: reagents for detecting the expression levels of gene markers GBP2, APOBEC3G, HCST, and CA14, and instructions for using the risk assessment model; the instructions state: obtain the expression levels of GBP2, APOBEC3G, HCST, and CA14, standardize the expression levels using Z-score, calculate the prognostic risk score RiskScore for melanoma patients according to the formula in claim 1, and use 0 as a preset threshold. When the RiskScore is greater than 0, the patient is classified into a high-risk group; when the RiskScore is less than or equal to 0, the patient is classified into a low-risk group.

6. The application according to claim 5, characterized in that, The products used for prognostic risk assessment of melanoma patients include reagent kits, chips, test strips, high-throughput sequencing platforms, real-time quantitative PCR detection reagents, nucleic acid probe combinations, protein detection products, or computer-aided analysis systems.

7. The application according to claim 5, characterized in that, The reagents for detecting the expression levels of the gene markers GBP2, APOBEC3G, HCST, and CA14 include: primers, probes, sequencing adapters, amplification reagents, hybridization probes, nucleic acid chip probes, or sequencing library construction reagents for detecting the mRNA expression levels of GBP2, APOBEC3G, HCST, and CA14; or antibodies, peptides, aptamers, compounds, protein chip reagents, or immunoassay compositions for detecting the corresponding protein expression levels of GBP2, APOBEC3G, HCST, and CA14. The expression levels of the gene markers are obtained by RNA sequencing, real-time quantitative PCR, reverse transcription PCR, digital PCR, nucleic acid chip, in situ hybridization, NanoString, Western blotting, enzyme-linked immunosorbent assay (ELISA), immunohistochemical staining, mass spectrometry, or protein chip detection.

8. A prognostic risk assessment system for melanoma patients for in vitro diagnostic or non-diagnostic purposes, comprising the four-gene melanoma prognostic risk assessment model based on RiboSis typing as described in claim 1, characterized in that, Includes the following modules: The expression data acquisition module is used to acquire the expression levels of GBP2, APOBEC3G, HCST and CA14 in the sample to be tested; The standardization module is used to perform Z-score standardization on the expression levels of GBP2, APOBEC3G, HCST and CA14 according to Z=(X-μ) / σ, where X is the expression level of the corresponding gene, μ is the mean expression of the corresponding gene in the test sample cohort, training cohort or reference cohort, and σ is the standard deviation of the expression of the corresponding gene in the test sample cohort, training cohort or reference cohort. The risk score calculation module is used to calculate the RiskScore according to the following formula: RiskScore = -0.25856323×Z(GBP2) - 0.12074882×Z(APOBEC3G) - 0.03239219×Z(HCST) + 0.02166196×Z(CA14); where Z(GBP2), Z(APOBEC3G), Z(HCST), and Z(CA14) represent the Z-score standardized values ​​of GBP2, APOBEC3G, HCST, and CA14, respectively. The risk stratification module is used to classify patients with melanoma under test into a high-risk group when the RiskScore is greater than 0, and into a low-risk group when the RiskScore is less than or equal to 0. The results output module is used to output the prognostic risk assessment results for melanoma patients.

9. The application of a melanoma patient prognostic risk assessment product prepared using the four-gene melanoma prognostic risk assessment model based on RiboSis typing and screening as described in claim 1, or the melanoma patient prognostic risk assessment system for in vitro diagnostic or non-diagnostic purposes as described in claim 8, in the preparation of an auxiliary assessment product for melanoma immunotherapy response.

10. The application according to claim 9, characterized in that, The immunotherapy includes anti-PD-1 therapy, anti-PD-L1 therapy, anti-CTLA-4 therapy, or a combination thereof.

11. The application of a reagent for detecting gene marker expression levels in the preparation of products for prognostic risk assessment of melanoma patients, characterized in that, The genetic markers include GBP2, APOBEC3G, HCST, and CA14.

12. The application according to claim 11, characterized in that, The gene markers are combinations of prognostic-related genes obtained through the RiboSis typing screening process; wherein, RiboSis refers to genes related to ribosome biosynthesis.

13. A kit for prognostic risk assessment in melanoma patients, characterized in that, The kit includes reagents for detecting the expression levels of GBP2, APOBEC3G, HCST and CA14, as well as instructions for use. The instructions for use describe how to calculate the prognostic risk score of melanoma patients based on the expression levels of GBP2, APOBEC3G, HCST and CA14, and how to stratify patients according to the prognostic risk score.

Images