A method for predicting recurrence or metastasis of neuroblastoma based on USP44 methylation level

Abstract

The application belongs to the technical field of biological medicine, and discloses application of a substance for detecting a methylation level of USP44 in preparation of a product for diagnosing or predicting recurrence and / or metastasis risk of neuroblastoma, wherein a methylation site of the USP44 is cg27100916, and the substance is used in combination with clinical information when diagnosing or predicting the recurrence and / or metastasis risk of neuroblastoma; the clinical information comprises at least one of basic information, clinical stages and MYCN gene amplification states of a patient; and the clinical stages comprise INSS stages and / or COG stages. The application provides a prediction model capable of early and accurate prediction of the recurrence and / or metastasis risk of neuroblastoma, significantly improves the accuracy of stratification of the recurrence and / or metastasis risk of neuroblastoma, and effectively supplements prediction information and prediction limitations of traditional clinical indexes.

Description

Technical Field

[0001] This invention belongs to the field of biomedical technology, specifically relating to a method for predicting recurrence or metastasis of neuroblastoma based on USP44 methylation levels. Background Technology

[0002] Tumorigenesis and development are the result of joint regulation by genetics and epigenetics, with DNA methylation being the most common epigenetic alteration. DNA methylation changes can be detected in the early stages of tumor progression. Recent studies have reported that methylation alterations in a group of genes can be used for early tumor diagnosis, tumor prognosis prediction, and disease progression monitoring. In gastric cancer, liver cancer, prostate cancer, and bladder cancer, literature has reported that molecular tags composed of methylation alterations can predict patient recurrence and metastasis.

[0003] Currently, clinical monitoring and early warning of neuroblastoma recurrence and metastasis still rely on tumor markers such as urinary VMA / HVA, serum NSE, and LDH levels, as well as imaging and pathological examinations such as 123I-MIBG whole-body scintigraphy, MRI, and bone marrow biopsy. These traditional detection methods all have certain limitations: a positive diagnosis can only be made after the tumor lesion has formed, the tumor burden has reached a certain volume, or tumor markers have risen abnormally. They are unable to provide early warning before recurrence and metastasis occur, exhibiting a certain degree of lag. Currently, the COG clinical risk stratification system integrates key indicators such as age, INSS stage, and MYCN amplification status, but it lacks specific, highly sensitive molecular markers for precise quantitative assessment of recurrence and / or metastasis risk, making it difficult to achieve individualized, prospective risk prediction.

[0004] USP44, as an important deubiquitinizing enzyme, plays a key regulatory role in cell cycle regulation, chromatin remodeling, and tumorigenesis and development. However, the correlation between its abnormal methylation level and the occurrence, progression, recurrence, and / or metastasis of neuroblastoma is still unclear, and there is no predictive model for the risk of recurrence and metastasis of neuroblastoma based on the methylation status of USP44. Summary of the Invention

[0005] This invention aims to overcome the shortcomings of existing technologies, such as delayed early warning and insufficient molecular subtyping, and to solve the following key technical problems: (1) Screening and identifying USP44 specific methylation sites that are highly associated with recurrence and / or metastasis of neuroblastoma from public databases; (2) Establishing a stable, sensitive, and standardized USP44 methylation detection method; (3) Integrating clinical and pathological parameters such as patient age, INSS stage, and MYCN amplification status to construct a predictive model that can predict the risk of recurrence and / or metastasis of neuroblastoma early and accurately, thereby enabling early identification and stratified management of patients with high recurrence and / or metastasis risk before the appearance of lesions visible on tumor imaging, and providing molecular basis for the development of individualized intervention strategies in clinical practice.

[0006] The first aspect of the present invention aims to provide the use of a substance for detecting USP44 methylation levels in the preparation of products for diagnosing or predicting the risk of recurrence and / or metastasis of neuroblastoma.

[0007] A second aspect of the present invention is to provide a detection kit.

[0008] The third aspect of this invention aims to provide a method for constructing a model for predicting the risk of recurrence and / or metastasis in neuroblastoma.

[0009] The fourth aspect of this invention aims to provide a model running module for assessing the risk of recurrence and / or metastasis in neuroblastoma.

[0010] The fifth aspect of this invention aims to provide a predictive system for assessing the risk of recurrence and / or metastasis in neuroblastoma.

[0011] The sixth aspect of this invention is to provide an electronic device.

[0012] To achieve the above objectives, the technical solution adopted by the present invention is as follows: In a first aspect, the invention provides the use of a substance for detecting the methylation level of USP44 in the preparation of a product for diagnosing or predicting the risk of recurrence and / or metastasis of neuroblastoma, wherein the methylation site of USP44 is cg27100916. The substance is used in conjunction with clinical information when diagnosing or predicting the risk of recurrence and / or metastasis of neuroblastoma; The clinical information includes at least one of the patient's basic information, clinical stage, and MYCN gene amplification status. The clinical staging includes INSS staging and / or COG staging.

[0013] In some embodiments of the present invention, the patient's basic information includes age.

[0014] In some embodiments of the present invention, the substance used to detect the methylation level of USP44 is selected from reagents used in one or more of the following methods: PCR detection technology based on bisulfite conversion, DNA sequencing methods, methylation-sensitive restriction endonuclease analysis, chip-based methylation mapping analysis technology, and mass spectrometry detection technology.

[0015] In some embodiments of the present invention, the PCR detection technique based on bisulfite conversion includes at least one of methylation-specific PCR and methylation-sensitive high-resolution melting curve method.

[0016] In some embodiments of the present invention, the methylation-specific PCR is real-time quantitative methylation-specific PCR.

[0017] In some embodiments of the present invention, the DNA sequencing method includes at least one of bisulfite sequencing, whole-genome methylation sequencing, and simplified methylation sequencing.

[0018] In some embodiments of the present invention, the mass spectrometry detection technology includes flight mass spectrometry.

[0019] In some embodiments of the present invention, the substance used to detect the methylation level of USP44 includes at least one of nucleic acid molecules, specific antibodies, chemical modification reagents, and methylation-related enzymes.

[0020] In some embodiments of the present invention, the substance used to detect USP44 methylation is a nucleic acid molecule.

[0021] In some embodiments of the present invention, the substance for detecting USP44 methylation levels comprises a primer set and a probe; the primer set comprises an upstream primer and a downstream primer: The nucleotide sequence of the upstream primer is 5'-TTTTGAAGTTATAAATCGGGC-3'; The nucleotide sequence of the downstream primer is 5'-CGATACAAATAAAAACAAACTCAA-3'; The nucleotide sequence of the probe is 5'-TTATTATGTAGTTATTGTTTTTGTTTCGAG-3'.

[0022] In some embodiments of the present invention, the nucleotide sequence of the probe is modified with a fluorescent reporter group and a fluorescent quencher group at both ends, respectively.

[0023] In some embodiments of the present invention, the fluorescent reporter group is modified at the 5' end of the probe, and the fluorescent quencher group is modified at the 3' end of the probe.

[0024] In some embodiments of the present invention, the fluorescent reporter group is selected from any one of FAM, HEX, TET, JOE, NED, VIC, CY3, CY5, ROX or TAMRA.

[0025] In some embodiments of the present invention, the fluorescence quenching group is selected from any one of TAMRA, MGB, BHQ, Dabcyl, or Eclipse.

[0026] In some embodiments of the present invention, the fluorescent reporter group FAM is modified at the 5' end of the probe.

[0027] In some embodiments of the present invention, the fluorescence quenching group BHQ is modified at the 3' end of the probe.

[0028] A second aspect of the present invention provides a detection kit comprising the substance described in the first aspect of the present invention for detecting USP44 methylation levels.

[0029] In some embodiments of the present invention, the test kit further includes a positive control, a negative control, and a reaction solution.

[0030] In some embodiments of the present invention, the detection kit further includes reagents for sulfite conversion of genomic DNA.

[0031] In some embodiments of the present invention, the reagent for sulfite conversion of genomic DNA includes bisulfite.

[0032] Bisulfite converts unmethylated cytosine on DNA into uracil.

[0033] In some embodiments of the present invention, the bisulfite is a bisulfite.

[0034] In some embodiments of the present invention, the bisulfite includes at least one of calcium bisulfite, sodium bisulfite, potassium bisulfite, and ammonium bisulfite.

[0035] A third aspect of the present invention provides a method for constructing a model for predicting the risk of recurrence and / or metastasis of neuroblastoma, comprising the following steps: Data on USP44 methylation in the sample to be tested were collected, and clinical information was integrated to construct a dataset; the USP44 methylation site was cg27100916; the clinical information included at least one of the patient's basic information, clinical stage, and MYCN gene amplification status. The clinical staging includes INSS staging and / or COG staging; Models were built on the dataset to obtain a model for predicting the risk of recurrence and / or metastasis of neuroblastoma; The construction method includes at least one of SVM, LASSO, and ElasticNet.

[0036] In some embodiments of the present invention, the patient's basic information includes age.

[0037] A fourth aspect of the present invention provides a model running module for assessing the risk of recurrence and / or metastasis in neuroblastoma; the model running module includes a computing component; the computing component calculates data on USP44 methylation and clinical information in the subject's sample; The data on USP44 methylation includes the USP44 methylation level; The clinical information includes at least one of the patient's basic information, clinical stage, and MYCN gene amplification status. The clinical staging includes INSS staging and / or COG staging.

[0038] In some embodiments of the present invention, the patient's basic information includes age.

[0039] A fifth aspect of the invention provides a predictive system for assessing the risk of recurrence and / or metastasis in neuroblastoma, said predictive system comprising at least one of 1) to 3): 1) Data input module; 2) Model execution module; 3) Result output module; The model running module includes the model running module for assessing the risk of recurrence and / or metastasis of neuroblastoma as described in the fourth aspect of this invention.

[0040] In some embodiments of the present invention, the data input module includes a sample genomic DNA extraction module, a genomic DNA sulfite conversion module, and a methylation level detection module.

[0041] In some embodiments of the present invention, the sample includes at least one of blood, tissue, cells, cerebrospinal fluid, urine, saliva, and semen; the sample is an ex vivo biological sample derived from a mammal.

[0042] In some embodiments of the present invention, the blood sample includes at least one of plasma and serum.

[0043] In some embodiments of the present invention, the tissue sample includes at least one of a fresh tissue sample and a paraffin tissue sample.

[0044] In some embodiments of the present invention, the methylation level detection module includes: mixing DNA obtained from the genomic DNA sulfite conversion module with the primer set and probe described in the first aspect of the present invention, and performing PCR amplification detection.

[0045] In some embodiments of the present invention, the PCR amplification program is as follows: 93-96℃, 4-6 min; 94-96℃, 14-16 s, 60-62℃, 29-32 s, 18-22 cycles; 93-96℃, 9-11 s, 58-62℃, 30 s, 30-32 cycles, fluorescence collection.

[0046] In some embodiments of the present invention, the detection channel for collecting fluorescence is a FAM signal channel and a VIC signal channel.

[0047] In some embodiments of the present invention, the method for assessing the methylation level of the methylation level detection module specifically includes: (1) Quality control: Negative controls: The FAM detection channel has no S-shaped amplification curve, the VIC detection channel has an S-shaped amplification curve, and the Ct value is ≤45; Positive control: The FAM detection channel shows an S-shaped amplification curve, and the Ct value is ≤45; All of the above requirements must be met simultaneously in the same experiment; otherwise, the experiment is invalid and must be repeated.

[0048] (2) Interpretation of experimental test results: When the sample amplification curve is a typical S-type amplification curve and both the negative and positive controls are under control, the results are interpreted as follows: if USP44 (FAM channel) Ct≤40 and β-actin reference (VIC channel) Ct≤45, it is judged as USP44 methylation positive. If USP44 (FAM channel) Ct>40 or has no Ct value and β-actin reference (VIC channel) Ct≤45, it is judged as USP44 methylation negative.

[0049] A sixth aspect of the present invention provides an electronic device including a storage device, a processor, and a computer program stored in the storage device and executable on the processor, wherein the computer program stored in the storage device and executable on the processor includes the prediction system described in the fifth aspect of the present invention.

[0050] The beneficial effects of this invention are: This invention screens and identifies the USP44-specific methylation site cg27100916, which is highly associated with recurrence and / or metastasis of neuroblastoma, and provides a method for detecting USP44 methylation sites, including the use of a specific primer set and probe. By integrating clinicopathological parameters such as patient age, INSS stage, and MYCN amplification status, this invention constructs a predictive model that can predict the risk of recurrence and / or metastasis of neuroblastoma early and accurately, significantly improving the accuracy of risk stratification for recurrence and / or metastasis of neuroblastoma, and supplementing the predictive limitations of traditional clinical indicators, effectively adding predictive information. The USP44 methylation site detection method of this invention is simple and easy to perform, and has good prospects for clinical translational application, providing new evidence for the treatment and follow-up of recurrence and / or metastasis of neuroblastoma. Attached Figure Description

[0051] The present invention will be further described below with reference to the accompanying drawings and embodiments, wherein: Figure 1 This document presents the screening results and prognostic value analysis of key CpG methylation sites in the USP44 gene. A shows the importance ranking of CpG methylation sites screened using ElasticNet and LASSO machine learning algorithms, with color gradients representing site ranking (red indicating higher importance, blue indicating lower importance). B is a violin plot showing the methylation level distribution of cg07783282 (left) and cg27100916 (right) in patients in the low / intermediate-risk group (n=49) and high-risk group (n=56) of the GSE73515 dataset. C shows the Kaplan-Meier curves (top) and OS (bottom) for cg27100916 methylation status (positive vs. negative) in the cohort of this invention (n=83), along with hazard ratios (HR), 95% confidence intervals, and Log-rank tests. P Value; D represents the Kaplan-Meier curves of event-free survival (top figure) and overall survival (bottom figure) stratified by cg07783282 methylation status (positive vs. negative) in the cohort of this invention (n=83), with hazard ratio (HR), 95% confidence interval, and Log-rank test indicated. P value.

[0052] Figure 2The results show the performance and variable importance analysis of different machine learning models. A is a bar chart comparing the predicted AUC values ​​of SVM, ElasticNet, and LASSO models for OS (red) and EFS (blue); B is the average importance (%) of each variable for OS (red) and EFS (blue) in all models, sorted by importance; C is a heatmap showing the importance (%) of each variable (Age, INSS, COG, MYCN, cg27100916) for OS (left) and EFS (right) in different models (LASSO, ElasticNet, SVM) with COG risk stratification included; D is a heatmap showing the importance (%) of each variable (Age, INSS, MYCN, cg27100916) for OS (left) and EFS (right) in different models after excluding COG risk stratification.

[0053] Figure 3 The results show the performance comparison of the models under different variable combinations. A is a bar chart displaying the average AUC values ​​of the five variable combination models (COG+cg27100916, COG only, INSS+Age+MYCN+cg27100916, INSS+Age+MYCN, and cg27100916 only) for overall survival (OS, red) and event-free survival (EFS, blue). B is the corresponding ROC curve, comparing the predictive performance of the five variable combination models for OS (left) and EFS (right). Detailed Implementation

[0054] The following will describe the concept and technical effects of the present invention clearly and completely with reference to embodiments, so as to fully understand the purpose, features and effects of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are all within the scope of protection of the present invention.

[0055] Unless otherwise specified in the examples, the procedures should be performed under standard conditions or conditions recommended by the manufacturer. Reagents or instruments whose manufacturers are not specified are all commercially available products.

[0056] Terminology Explanation: MYCN amplification is one of the most important molecular genetic abnormalities in neuroblastoma, and it is closely related to tumor invasiveness, treatment response, and prognosis.

[0057] CpG island: CpG island, Methylated CpG dinucleotide.

[0058] COG Risk Classification: Children's Oncology Group (COG) Neuroblastoma Risk Classification.

[0059] INSS: International Neuroblastoma Staging System.

[0060] The clinical information of 83 patients with neuroblastoma in this embodiment is shown in Table 1, and the summary table of clinical variables is shown in Table 2. Among them, the case numbers are the wax-coated volumes of the Women and Children's Medical Center Affiliated to Guangzhou Medical University (83 valid cases from NB1-NB86); age group: <18 months / ≥18 months (the conventional threshold for COG risk stratification of neuroblastoma is 547 days); INSS staging: International Neuroblastoma Staging System, original stages 1 / 2 / 3 / 4; INSS grouping: Early (stages 1-2) / Advanced (stages 3-4); COG risk stratification: Children's Oncology Group risk stratification: High risk / Non-high risk (Low+Intermediate); MYCN amplification: Amplified (≥4 copies) / Non-amplified; cg27100916: cg27100916 site in the promoter region of the USP44 gene, Positive = methylation positive, Negative = methylation negative; OS events: Overall Survival: 1 = death, 0 = survival at last follow-up; OS time (days): the interval from initial diagnosis to death / last follow-up; EFS event: Event-Free Survival: 1 = relapse / progression / death, 0 = no event; EFS time (days): the interval from initial diagnosis to the first event or last follow-up.

[0061] Sample inclusion and exclusion process: 86 cases were included (tumor tissue from neuroblastoma patients at the Women and Children's Medical Center Affiliated to Guangzhou Medical University) → 3 cases with incomplete OS / EFS follow-up were excluded → 83 cases entered Kaplan-Meier survival analysis → 5 cases with failed cg27100916 detection or missing clinical variables were further excluded → 78 cases entered machine learning modeling and AUC assessment.

[0062] Table 1. Clinical information of 83 patients with neuroblastoma

[0063] Table 2 Summary of Clinical Variables

[0064] The features and performance of the present invention will be further described in detail below with reference to embodiments.

[0065] Example 1: Screening of USP44 methylation sites based on public databases Methylome and clinical parameters were obtained from the neuroblastoma public database GSE73515. Bioinformatics analysis and machine learning algorithms were used to screen for USP44-specific methylation sites that are significantly associated with neuroblastoma recurrence and / or metastasis, and to identify core detection sites to provide targets for subsequent validation.

[0066] 1. Data Source Download methylation data and complete clinical case information of neuroblastoma patients from the neuroblastoma public database GSE73515, including patient age, sex, COG risk stratification, MYCN amplification, etc.

[0067] 2. Data Preprocessing The methylation and gene expression data were standardized, batch-effects corrected, and missing values ​​were imputed to obtain a valid dataset that could be used for subsequent analysis.

[0068] 3. Machine learning method for screening USP44 differential methylation sites The inventors used glmnet to fit a penalized regression model for the 15 CpG islands in USP44 with COG risk stratification as the outcome event. This model included two methods: LASSO (alpha=1) and ElasticNet (alpha=0.5). The 15 CpG sites were assigned importance rankings (lower rankings indicate higher importance) using specific methods.

[0069] Both machine learning methods prioritized CpG sites with the highest common ranking for downstream evaluation, as shown in the results. Figure 1 As shown in Figure A, the screening results showed that two USP44 CpG sites contributed the most to the COG risk of neuroblastoma: cg27100916 and cg07783282, which can be used as target sites for subsequent detection and model construction.

[0070] 4. Verification Since the dataset lacks events with and without survival, the inventors analyzed the distribution characteristics of the two methylation sites cg27100916 and cg07783282 in each COG group.

[0071] The results are as follows Figure 1 As shown in Figure B, in the high-risk neuroblastoma group (high), USP44 cg27100916 and USP44 cg07783282 showed significantly high methylation levels, suggesting that high methylation levels of USP44 are a factor contributing to poor prognosis in patients.

[0072] Example 2: Establishment of a method for detecting the methylation level of the USP44 gene DNA samples were extracted from tumor tissues of 83 patients with neuroblastoma (Guangzhou Medical University Affiliated Women and Children's Medical Center). The DNA was modified using bisulfite treatment to convert unmethylated cytosine to uracil, while methylated cytosine remained unchanged. Methylation-specific PCR (MSP) was then used to amplify the USP44 target region, followed by electrophoresis / quantitative fluorescence detection to quantitatively assess the methylation level of the USP44 gene promoter or key regions, establishing a stable and reproducible molecular detection method. Specific experimental steps included: 1. Genomic DNA extraction from samples DNA extraction from tumor tissue samples was performed using a paraffin tissue DNA extraction kit (one-step extraction type (FFPE genomic DNA high sensitivity), Kaishuo Biotechnology (Xiamen) Co., Ltd., RC1102).

[0073] 2. Bisulfite Conversion Approximately 2-3g of the genomic DNA obtained above was added to the bisulfite conversion reaction system, and the bisulfite conversion experiment was performed according to the instructions of the bisulfite conversion kit (Zymo Research, D5006).

[0074] USP44 cg27100916 methylation site sequence (original sequence): AACATTAAAAAAAAGAAAAGAAAAAAGCCTCCTTCCCTGCCCACCTTCAT CAATGATGTCCCCAAACTTACCTATTAAAACCTCAAAATCAGTCTTTCCC CAGGTCGGCCTTGGATAACTCAGAGGTCAGTCCCTGCAGGATCGCCCAG TTTCCATCAGGGCAAATGCTGATCTCCTAAGTGAAAAGTCCTGGCGGCTG CCTCCAGTGCCCCCCTGCAAATCATTACAAATCAACCTCTTTCAGGTTTG GCTTAAACTCTCACCTCCTGGAAACTGCTTTCCTGAAGTCACAAACCGGG CGTGCAGATCCCCATCCCCCACCCCCAGCAGTGCATTATTATGCAGTCAC TGCTTTTGTTCCGAGTTCTGAGCCTGCTTCTATTTGCATCGTAACCTCTT TGGGGGCCGGGACTGTGCACTTGCTCCCTTTCTATACCTCTCCACAGTTC CTAGTACCGGCCAAAAACACTGCTGAAATGCCTAATGAATCAACCAGAAG TTCCTTAGCATAATATAGGAGCGTGGAAGTGATGTTTAGGGGGGCTTCCC TTGCTTCCCTGATCCCCAGATGGAGAGAAATGGCAATTTGCCCTTCCAAG G (SEQ ID NO: 1).

[0075] Sequence of USP44 cg27100916 methylation site after sulfite treatment: AATATTAAAAAAAAGAAAAGAAAAAAGTTTTTTTTTTTGTTTATTTTTAT TAATGATGTTTTTAAATTTATTTATTAAAATTTTAAAATTAGTTTTTTTT TAGGTCGGTTTTTGGATAATTTAGAGGTTAGTTTTTGTAGGATCGTTTAG TTTTTATTAGGGTAAATGTTGATTTTTTAAGTGAAAAGTTTTGGCGGTTG TTTTTAGTGTTTTTTTGTAAATTATTATAAATTAATTTTTTTTAGGTTTG GTTTAAATTTTTATTTTTTGGAAATTGTTTTTTTGAAGTTATAAATCGGG CGTGTAGATTTTTATTTTTTATTTTTAGTAGTGTATTATTATGTAGTTAT TGTTTTTGTTTCGAGTTTTGAGTTTGTTTTTATTTGTATCGTAATTTTTT TGGGGGTCGGGATTGTGTATTTGTTTTTTTTTTATATTTTTTTAGTTT TTAGTATCGGTTAAAAATATTGTTGAAATGTTTAATGAATTAATTAGAAG TTTTTTAGTATAATATAGGAGCGTGGAAGGTGATGTTTAGGGGGGTTTTTT TTGTTTTTTTGATTTTTAGATGGAGAGAAATGGTAATTTGTTTTTTTAAG G (SEQ ID NO:2).

[0076] 3. Methylation-specific PCR (MSP) amplification Specific primers were designed based on the promoter region and key CpG island sequences of the USP44 gene. The PCR amplification system (Takara, RR390A) was prepared according to Table 3, and the PCR reaction was carried out according to the amplification procedure in Table 4.

[0077] USP44 cg27100916 primers and probes: Upstream primer F: 5'-TTTTGAAGTTATAAATCGGGC-3' (SEQ ID NO:3); Downstream primer R: 5'-CGATACAAATAAAAACAAACTCAA-3' (SEQ ID NO:4); Probe: 5'-TTATTATGTAGTTATTGTTTTTGTTTCGAG-3' (SEQ ID NO:5); The probe has a fluorescent group FAM attached to its 5' end and a quenching group BHQ attached to its 3' end.

[0078] β-actin upstream primer F: 5'-TGGTGATGGAGGAGGTTTAG-3' (SEQ ID NO:6); β-actin downstream primer R: 5'-CCTTAAAAATTACAAAAACCACAACC-3' (SEQ ID NO:7); β-actin probe: 5'-AGTTTTTTGGATTGTGAATTTGTGT-3' (SEQ ID NO:8); The probe has a fluorescent group VIC attached to its 5' end and a quenching group BHQ attached to its 3' end.

[0079] Table 3. PCR amplification system

[0080] Table 4 PCR Amplification Program

[0081] 4. Methylation level assessment The methylation level was assessed using the Ct value calculation method, with methylation-positive and negative controls used as standard curves to quantitatively evaluate the degree of methylation in the promoter region of the USP44 gene.

[0082] The criteria for interpreting USP44 methylation levels are as follows: (1) Experimental testing quality control: Negative control: The FAM detection channel showed no S-shaped amplification curve; the VIC detection channel showed an S-shaped amplification curve, and the Ct value was ≤45. Positive control (ZymoD5014 Human Methylated Non-Methylated (WGA) DNA Set, derived from HCT116 DKO cell genomic DNA): The FAM detection channel showed an S-type amplification curve and a Ct value ≤45.

[0083] All of the above requirements must be met simultaneously in the same experiment; otherwise, the experiment is invalid and must be repeated.

[0084] (2) Interpretation of experimental test results: When the sample amplification curve is a typical S-type amplification curve and both the negative and positive controls are under control, the results are interpreted as follows: if USP44 (FAM channel) Ct≤40 and β-actin reference (VIC channel) Ct≤45, it is judged as USP44 methylation positive. If USP44 (FAM channel) Ct>40 or has no Ct value and β-actin reference (VIC channel) Ct≤45, it is judged as USP44 methylation negative; If neither of the above two conditions is met, the test is deemed invalid and must be repeated.

[0085] Example 3: Construction of a multifactorial model for predicting the risk of relapse and / or metastasis In clinical samples of neuroblastoma (tumor tissue samples from 83 neuroblastoma patients used in Example 2, all from the Women and Children's Medical Center Affiliated to Guangzhou Medical University; the USP44 cg27100916 methylation level was measured by the MSP detection method established in Example 2), the correlation between USP44 methylation level and recurrence and metastasis was verified. At the same time, key clinicopathological indicators such as age, INSS (clinical stage), and MYCN amplification status were included. The above parameters were integrated through statistical models (Cox regression or machine learning model) to establish a quantifiable recurrence and / or metastasis risk prediction model, so as to achieve stratified assessment of the recurrence and / or metastasis risk of patients.

[0086] 1. Screening for effective USP44-specific methylation sites to be included in the risk model For the USP44-specific methylation sites cg27100916 and cg7783282 screened from the above public database, the methylation level in paraffin tissue samples was detected using the detection method of Example 2.

[0087] Patients were divided into negative and positive groups. Survival analysis revealed that cg27100916 had a significant impact on event-free survival (EFS): when cg27100916 was highly methylated, the event-free survival rate was significantly reduced. Figure 1 In the top panel (C), the overall survival (OS) rate showed a downward trend. Figure 1 (C, down panel); while cg7783282 had no effect on patient survival ( Figure 1 (D). Therefore, in the subsequent risk model construction, only the USP44-specific methylation site cg27100916 was included.

[0088] 2. Comparison of the advantages of common machine learning methods We used three machine learning methods—SVM, LASSO, and ElasticNet—to evaluate the impact of the risk model integrating clinical information on overall survival and event-free survival in cg27100916.

[0089] The inventors used AUC values ​​to evaluate the predictive capabilities of three machine learning methods for OS and EFS. A bar chart compares the AUC values ​​of SVM, ElasticNet, and LASSO models for OS (red) and EFS (blue). Figure 2(A); Simultaneously, the importance of cg27100916 and clinical information such as MYCN amplification, Age, INSS (clinical stage), and COG (clinical risk stratification, including MYCN amplification, Age, INSS, and other chromosomal variations 1q, etc.) for predicting OS and EFS was also assessed. Results are as follows: Figure 2 As shown, the average importance (%) of each variable to OS (red) and EFS (blue) in all models, sorted by importance ( Figure 2 (B); The heatmap shows the importance (%) of each variable (Age, INSS, COG, MYCN, cg27100916) to OS (left) and EFS (right) in different models (LASSO, ElasticNet, SVM) with COG risk stratification. Figure 2 (C); The heatmap shows the importance (%) of each variable (Age, INSS, MYCN, cg27100916) to OS (left) and EFS (right) in different models after excluding COG risk stratification. Figure 2 (D).

[0090] The specific experimental steps and results are as follows: (1) Enrollment and Exclusion. 86 patients with neuroblastoma (Guangzhou Medical University Affiliated Women and Children's Medical Center) were enrolled; 3 patients with incomplete OS / EFS follow-up were excluded, resulting in 83 patients for Kaplan-Meier survival analysis. Figure 1 C, Figure 1 (D, n=83); based on this, 5 samples with failed cg27100916 testing or missing clinical variables such as Age / INSS / MYCN / COG were further excluded, and finally 78 cases were included in machine learning modeling and AUC assessment ( Figure 2 , Figure 3 (n=78); of which 25 were OS events (32.1%) and 36 were EFS events (46.2%).

[0091] (2) Variable coding. All input variables were coded as binary variables according to their clinical significance: Age (<18 months=0, ≥18 months=1); INSS (Early=0, Advanced=1); COG (Nonhigh=0, High=1); MYCN (Nonamp=0, Amp=1); cg27100916 (negative=0, positive=1).

[0092] (3) Machine learning algorithms and parameters. The following three algorithms were implemented using Python 3 and scikit-learn: LASSO - L1 regularized logistic regression (penalty=l1, solver=SAGA, regularization parameter α=0.05, i.e. C=20, maximum iterations 20000); ElasticNet - L1+L2 hybrid regularized logistic regression (penalty=elasticnet, solver=SAGA, α=0.1, i.e. C=10, L1 ratio=0.5, maximum iterations 20000); SVM - radial basis function support vector machine (kernel=RBF, C=1.0, probability=True).

[0093] (4) Cross-validation and standardization. Five-fold hierarchical cross-validation (StratifiedKFold, shuffle=True, random_state=42) is adopted to ensure that the OS / EFS event ratio is consistent in each fold; within each fold, StandardScaler is first fitted on the training set and then applied to the test set to avoid data leakage.

[0094] (5) AUC Calculation and Aggregation. The AUC (roc_auc_score) is calculated on each fold of the test set based on the model output probability; the mean and standard deviation of the 5-fold AUC are used as the cross-validation AUC for a single algorithm. Figure 2 The height of column A and the error bars); then the sample-level prediction probabilities of the three algorithms, SVM, LASSO, and ElasticNet, are averaged sample by sample with equal weights to form the out-of-fold prediction of the model group, and the ROC curve is plotted. Figure 3 (Middle B) Figure 3 The “average AUC” shown in A is the average of the AUC mean values ​​of the three algorithms’ 5-fold cross-validation.

[0095] (6) Quantification of variable importance. For linear models (LASSO, ElasticNet), the absolute value of the standardized coefficients is used as the importance; for SVM, permutation importance (the decrease in AUC after multiple random shuffling of a single variable) is used; the importance of all variables within the same model is normalized to a percentage (total = 100%) for cross-algorithm comparison. The results are shown in […]. Figure 2 B (average of 3 algorithms) Figure 2 C (including the complete model of COG), Figure 2 D (model excluding COG).

[0096] All the code above uses a fixed random seed of 42, and the results can be completely reproduced.

[0097] 3. The predictive ability of the risk model for EFS and OS The risk prediction model construction process of this invention specifically includes the following 6 steps: (1) Candidate variable determination. The methylation candidate sites were derived from Example 1—with COG risk stratification as the outcome, the USP44 15 CpG sites in the neuroblastoma public database GSE73515 were fitted with glmnet penalized regression models using LASSO (alpha=1) and ElasticNet (alpha=0.5), respectively, and the sites with the highest ranking in both methods were selected; after survival analysis (Example 3, Section 1), only cg27100916 was finally included in the risk model; the four clinical variables recognized in NB clinical practice were included—Age, INSS, MYCN, and COG.

[0098] (2) Variable coding. The five candidate variables are uniformly coded in a binary classification (see Section 2 (2) of this embodiment).

[0099] (3) Model training. For each variable combination (COG+cg27100916, COG only, INSS+Age+MYCN+cg27100916, INSS+Age+MYCN, cg27100916 only, a total of 5 groups), LASSO, ElasticNet, and SVM were used for model fitting respectively; during the training process, 5-fold stratified cross-validation was used, and each fold was first fitted to the StandardScaler on the training subset and then applied to the test subset to prevent data leakage; each sample obtained the predicted probability of "belonging to the event group" under the 3 algorithms. .

[0100] (4) Risk score output. The final risk score is the ensemble of the predicted probabilities of the three algorithms, with a value range of [0,1]. The larger the value, the higher the risk of recurrence and / or metastasis. If risk stratification is required, the high / low risk groups are divided by taking the threshold with the highest value of Youden's J index in the OOF prediction.

[0101] (5) Quantification of variable importance. For linear models (LASSO, ElasticNet), the absolute value of the standardized coefficients is taken; for SVM, permutation importance is used; importance within the same model is normalized to a percentage (total = 100%), facilitating cross-algorithm comparisons. Figure 1 China A Figure 2 B, Figure 2 C, Figure 2 (D).

[0102] (6) Model performance comparison. Steps 3-5 were completed for the above 5 groups of variables respectively, and the mean AUC of 5-fold cross-validation (average of 3 algorithms) was used as the average AUC of the model group to quantify the predictive ability of each group and the AUC increment (ΔAUC) brought by adding cg27100916, which was used to demonstrate the incremental predictive value of cg27100916 methylation level for traditional clinical variables.

[0103] The entire process is implemented using Python 3 and the scikit-learn library, with a fixed random seed of 42. The model parameters and data are described in Section 2 of this embodiment.

[0104] The predictive power of five variable combination models for event-free survival (EFS) and overall survival (OS) was evaluated by calculating the AUC. The five groups were COG+cg27100916, COG only, INSS+Age+MYCN+cg27100916, INSS+Age+MYCN, and cg27100916 only (cg27100916 was used as a single input to verify the independent predictive power of this methylation site).

[0105] The results are as follows Figure 3 As shown, the average AUC values ​​of OS and EFS for the COG+cg27100916 group were 0.732 and 0.755, respectively; for the COG only group, they were 0.679 and 0.721, respectively; for the INSS+Age+MYCN+cg27100916 group, they were 0.662 and 0.805, respectively; for the INSS+Age+MYCN group, they were 0.604 and 0.789, respectively; while the average AUC values ​​of OS and EFS for the cg27100916 only group (using only cg27100916 as a single input) were only 0.564 and 0.622, respectively. Figure 3 (A) The ROC curve results are as follows: Figure 3 As shown in Figure B. In summary: the COG+cg27100916 group showed the best OS prediction performance, while the INSS+Age+MYCN+cg27100916 group showed the best EFS prediction performance. However, the AUC value of cg27100916 as a standalone input was significantly lower than any combination containing clinical variables, indicating that cg27100916 does not have independent prognostic predictive ability. Its clinical value lies in the incremental predictive value after being combined with COG risk stratification or clinical variables such as age / INSS / MYCN. This further confirms the necessity and rationality of the present invention in jointly modeling cg27100916 methylation with clinicopathological parameters.

[0106] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments, and various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention. Furthermore, the embodiments of the present invention and the features thereof can be combined with each other unless otherwise specified.

Claims

1. The use of a substance for detecting the methylation level of USP44 in the preparation of products for diagnosing or predicting the risk of recurrence and / or metastasis of neuroblastoma, wherein the methylation site of USP44 is cg27100916; The substance is used in conjunction with clinical information when diagnosing or predicting the risk of recurrence and / or metastasis of neuroblastoma; The clinical information includes at least one of the patient's basic information, clinical stage, and MYCN gene amplification status. The clinical staging includes INSS staging and / or COG staging.

2. The application according to claim 1, characterized in that, The substance used to detect the methylation level of USP44 includes a primer set and a probe; The primer set includes an upstream primer and a downstream primer: The nucleotide sequence of the upstream primer is 5'-TTTTGAAGTTATAAATCGGGC-3'; The nucleotide sequence of the downstream primer is 5'-CGATACAAATAAAAACAAACTCAA-3'; The nucleotide sequence of the probe is 5'-TTATTATGTAGTTATTGTTTTTGTTTCGAG-3'; Preferably, the nucleotide sequence of the probe is modified with a fluorescent reporter group and a fluorescent quencher group at both ends.

3. The application according to claim 2, characterized in that, The fluorescent reporter group is modified at the 5' end of the probe, and the fluorescent quencher group is modified at the 3' end of the probe.

4. A test kit, characterized in that, The test kit includes the substance for detecting USP44 methylation levels as described in claim 1 or 2.

5. The detection kit according to claim 4, characterized in that, The test kit also includes positive control, negative control and reaction solution.

6. The detection kit according to claim 5, characterized in that, The test kit also includes reagents for sulfite conversion of genomic DNA.

7. A method for constructing a model to predict the risk of recurrence and / or metastasis in neuroblastoma, characterized in that, Includes the following steps: Data on USP44 methylation in the sample to be tested were collected, and clinical information was integrated to construct a dataset; the USP44 methylation site was cg27100916; the clinical information included at least one of the patient's basic information, clinical stage, and MYCN gene amplification status. The clinical staging includes INSS staging and / or COG staging; Models were built on the dataset to obtain a model for predicting the risk of recurrence and / or metastasis of neuroblastoma; The construction method includes at least one of SVM, LASSO, and ElasticNet.

8. A model execution module for assessing the risk of recurrence and / or metastasis in neuroblastoma, characterized in that: The model operation module includes a computing component; the computing component calculates data on USP44 methylation and clinical information in the subject's test sample; The data on USP44 methylation includes the USP44 methylation level; The clinical information includes at least one of the patient's basic information, clinical stage, and MYCN gene amplification status. The clinical staging includes INSS staging and / or COG staging.

9. A predictive system for assessing the risk of recurrence and / or metastasis in neuroblastoma, characterized in that, The prediction system includes at least one of 1) to 3): 1) Data input module; 2) Model execution module; 3) Result output module; The model running module includes the model running module for assessing the risk of recurrence and / or metastasis of neuroblastoma as described in claim 8.

10. An electronic device comprising a storage device, a processor, and a computer program stored in the storage device and executable on the processor, characterized in that: The computer program stored in the memory and capable of running on the processor includes the prediction system of claim 9.

Images