Dynamic bladder cancer monitoring method based on circulating tumor DNA methylation marker

Abstract

The invention discloses a bladder cancer dynamic monitoring method based on a circulating tumor DNA methylation marker, and belongs to the technical field of molecular diagnosis. The method comprises the following steps: collecting urine of a patient at a plurality of continuous time points after operation and extracting circulating free DNA (Deoxyribose Nucleic Acid); the method comprises the following steps: constructing a targeted sequencing library aiming at two pre-selected bladder cancer specific methylation difference regions, the haplotype of the first region characterizing tissue cloning stability, and the haplotype of the second region characterizing tumor malignant progression; analyzing the methylated haplotype of each region through high-throughput sequencing; a passenger haplotype instability index reflecting clone stability and a driving haplotype load reflecting malignant potential are calculated and integrated into a dynamic methylation risk score; and generating graded risk early warning according to the score continuous change trend. According to the invention, high-sensitivity and high-specificity non-invasive dynamic monitoring and early warning of bladder cancer recurrence and progression risks are realized.

Description

Technical Field

[0001] This invention relates to the field of biotechnology, and in particular to a method for dynamic monitoring of bladder cancer based on circulating tumor DNA methylation markers. Background Technology

[0002] Bladder cancer is a prevalent malignant tumor of the urinary system worldwide. A major challenge in its clinical management is its extremely high postoperative recurrence rate, with non-muscle-invasive bladder cancer (NMIBC) experiencing a recurrence rate of 60%-70%. Currently, the "gold standard" for postoperative monitoring of bladder cancer is cystoscopy combined with pathological biopsy. However, this method is invasive, causing discomfort and risks to patients such as pain, hematuria, and urinary tract infections, and is also expensive, leading to poor patient compliance. Traditional non-invasive monitoring methods, such as urine cytology, while highly specific, have severely insufficient sensitivity for low-grade tumors (typically below 40%), failing to meet the needs for early and accurate monitoring. Other methods based on urinary protein markers (such as NMP22) or fluorescence in situ hybridization (FISH) also generally suffer from limitations such as insufficient sensitivity or specificity and susceptibility to numerous influencing factors.

[0003] In recent years, liquid biopsy technology, especially based on the analysis of circulating tumor DNA (ctDNA), has opened up new avenues for non-invasive diagnosis and monitoring of cancer. In bladder cancer, urine is an ideal liquid biopsy sample for enriching ctDNA derived from urinary system tumors. DNA methylation, as a stable epigenetic modification, plays a crucial role in tumor development and progression; its abnormal patterns are tumor-type specific and stage-specific, and are considered highly promising biomarkers. Existing studies have confirmed that methylation analysis of urinary DNA can aid in the diagnosis of bladder cancer.

[0004] However, existing DNA methylation-based detection methods still face the following key problems when applied to dynamic monitoring of bladder cancer, limiting their clinical translation and application effectiveness: First, the mixed functions of biomarkers lead to impure monitoring signals. Current detection methods often analyze differentially methylated regions (DMRs) from different sources and with different biological functions. Studies have shown that bladder cancer-related DMRs can be divided into different types; for example, "passenger" DMRs (such as T1DMR) mainly reflect the normal tissue origin of cells, while "driver" DMRs (such as T2DMR) directly participate in and drive the malignant progression of tumors. In dynamic monitoring, indiscriminately using these biomarkers can lead to background signal interference, making it difficult to capture the key methylation changes that truly indicate tumor recurrence or malignant transformation. Second, there is a lack of effective dynamic risk quantification models. Existing technologies mostly focus on qualitative or semi-quantitative diagnosis based on a single time point, failing to provide continuous and dynamic quantitative assessment of patients' recurrence risk. The core value of postoperative monitoring lies in early warning, requiring a method that can integrate multi-dimensional methylation information and output risk trends over time, thereby identifying high-risk patients before radiographic or cystoscopic recurrence is visible. Third, the ability to predict tumor progression is insufficient. Differentiating between low-grade (LG) / non-muscle-invasive bladder cancer (NMIBC) and high-grade (HG) / muscle-invasive bladder cancer (MIBC) is crucial for clinical treatment decisions. Current methods struggle to effectively predict, non-invasively, whether a tumor is progressing from an indolent to an aggressive phenotype, which is key information in determining whether more aggressive treatment is needed.

[0005] Therefore, there is an urgent need in this field for an innovative dynamic monitoring method based on urinary ctDNA methylation. This method should be able to overcome the above-mentioned defects and construct a dynamic risk assessment model by selecting and integrating methylation markers with specific biological functions, thereby achieving highly sensitive and specific early warning of bladder cancer recurrence and prediction of progression risk. Summary of the Invention

[0006] To achieve the above objectives, the present invention provides a method for dynamic monitoring of bladder cancer based on circulating tumor DNA methylation markers, comprising the following steps: Step 1: Collect and process urine samples at multiple consecutive pre-set time points after bladder cancer surgery to obtain urine supernatant for extracting circulating cell-free DNA. Step 2: Extract circulating cell-free DNA from urine supernatant and perform quantification and fragment analysis on the extracted circulating cell-free DNA; Step 3: Construct a targeted methylation haplotype sequencing library for pre-selected bladder cancer-specific methylation differential regions; the methylation differential regions include a first type of methylation differential region and a second type of methylation differential region; the first type of methylation differential region is the region where the methylation haplotype remains stable before and after carcinogenesis; the second type of methylation differential region is the region where the methylation haplotype is rearranged during tumor progression; Step 4: Perform high-throughput sequencing on the sequencing library to resolve the methylation haplotypes of all consecutive CpG sites in the two types of methylation difference regions; Step 5: Based on the resolved methylation haplotype data, calculate the passenger haplotype instability index and driving haplotype load for each monitoring time point, and integrate the passenger haplotype instability index and driving haplotype load into a dynamic methylation risk score; the passenger haplotype instability index represents the degree of shift of dominant haplotypes in the first type of methylation difference region; the driving haplotype load represents the abundance of predefined malignant haplotypes in the second type of methylation difference region; Step 6: Generate dynamic monitoring reports and risk warnings based on the changing trends of dynamic methylation risk scores at continuous time points.

[0007] Preferably, in step 1, the multiple consecutive preset time points include the 1st month, 3rd month, 6th month, and 12th month after surgery, and then collected every 6 months thereafter; Collect 100 ml of midstream urine from the patient each time and immediately inject it into a urine preservation tube containing a DNase inhibitor. Store the urine at 2-8 degrees Celsius and complete the processing within 4 hours after collection. The process includes a two-step centrifugation process: the first step is centrifugation at 4 degrees Celsius and 1500 rpm for 10 minutes to separate the upper layer of urine supernatant and the lower layer of cell precipitate. The second step involves transferring the urine supernatant obtained from the first step to a new centrifuge tube and centrifuging it again for 10 minutes at 4 degrees Celsius and 13,000 rpm to remove cell debris and impurities. The resulting supernatant is the urine supernatant used to extract circulating cell-free DNA.

[0008] Preferably, in step 2, the extraction of circulating free DNA uses a silica membrane adsorption column kit, and the pH and salt concentration of the binding buffer are adjusted during the extraction process to preferentially recover DNA fragments with a length of less than 200 base pairs. The quantitative and fragment analysis includes: determining the total concentration of extracted circulating free DNA using a fluorescence quantitative quantification instrument; Meanwhile, the fragment size distribution of circulating cell-free DNA was analyzed using a microcapillary electrophoresis system, and the percentage of circulating cell-free DNA with fragment sizes between 100 and 250 base pairs was calculated. This percentage was used as one of the indicators to assess the enrichment of circulating tumor DNA in the sample.

[0009] Preferably, in step 3, the methylation difference regions are 5 first-type methylation difference regions and 5 second-type methylation difference regions; The first type of differentially methylated regions were obtained by comparing and screening a large amount of whole-genome methylation sequencing data of normal urothelial tissue and bladder cancer tissue. The screening criteria were: high methylation in both normal and cancerous tissues, and the haplotype pattern formed by continuous CpG sites within the region was stable in different individuals and remained consistent before and after cancer in the same individual. The second type of differentially methylated regions were obtained by comparing and screening whole-genome methylation sequencing data of non-muscle-invasive bladder cancer tissues and muscle-invasive bladder cancer tissues. The screening criteria were: specific hypermethylation state or haplotype pattern in muscle-invasive bladder cancer tissues, and the frequency of this haplotype pattern increased significantly during the transformation from non-muscle-invasive bladder cancer to muscle-invasive bladder cancer.

[0010] Preferably, in step 3, the specific steps for constructing the targeted methylation haplotype sequencing library are as follows: First, the extracted circulating free DNA is subjected to bisulfite conversion treatment to convert unmethylated cytosine into uracil; Then, the transformed DNA was amplified in two rounds using a multiplex PCR primer pool designed for 10 differentially methylated regions. The first round of amplification introduced sample-specific index sequences, and the second round of amplification introduced universal sequencing adapters. Finally, the amplification products are purified and fragments of a specific size range are selected to complete library construction; Sequencing libraries from all time points of the same patient were mixed in equimolar amounts after quantification to enable parallel sequencing in the same sequencing process.

[0011] Preferably, in step 4, the depth of the high-throughput sequencing is not less than 5000X per sample; The process of resolving methylation haplotypes includes: aligning sequencing reads to a reference genome and locating the target methylation differential region; For each reading that corresponds to a region of methylation difference, the methylation status of each CpG site covered by the reading is determined sequentially according to its sequence. The methylation status is recorded as 1, and the non-methylation status is recorded as 0, forming a binary string composed of 0 and 1. This string is a methylation haplotype. We counted all unique methylation haplotypes appearing in each methylation differential region and their corresponding sequencing read support numbers.

[0012] Preferably, in step 5, the specific process of calculating the passenger haplotype instability index is as follows: for each type I methylation difference region, the first postoperative monitoring time point is determined as the baseline time point, and the dominant haplotype in that region at the baseline time point is identified; At each subsequent monitoring time point, it is determined whether the dominant haplotype in this region is the same as the dominant haplotype at the baseline time point; If they are different, it is determined that a haplotype shift occurred in that region at that time point; The passenger haplotype instability index at a given time point is equal to the ratio of the number of haplotype shifts occurring in the five type I methylation differential regions at that time point to 5.

[0013] Preferably, in step 5, the specific process of calculating the driving haplotype load is as follows: for each second type of methylation difference region, one or more specific methylation haplotypes that appear in the region and are strongly associated with muscle-invasive bladder cancer are predefined as malignant haplotypes; At each monitoring time point, the number of sequencing reads that perfectly match any predefined malignant haplotype among all sequencing reads aligned to that region is counted. Calculate the percentage of this number relative to the total sequencing reads in that region; The driving haplotype load at a given time point is equal to the weighted average of the percentage values ​​of the five differentially methylated regions of type II methylation. The weighting coefficients are pre-set based on the specificity of each region's malignant haplotype in predicting muscle-invasive bladder cancer.

[0014] Preferably, in step 5, the method for integrating into a dynamic methylation risk score is as follows: the passenger haplotype instability index and the driving haplotype load at the same time point are used as two feature variables and input into a pre-trained logistic regression model. The logistic regression model was trained based on historical patient cohort data, which included passenger haplotype instability index and driving haplotype load at a series of monitoring time points for bladder cancer patients with known eventual recurrence or progression. The model outputs a value between 0 and 1, which is the dynamic methylation risk score at that time point.

[0015] Preferably, in step 6, the specific method for generating dynamic monitoring reports and risk warnings is as follows: connect the dynamic methylation risk scores of the same patient at all continuous monitoring time points in chronological order and draw a risk trend curve. Set a first warning threshold and a second warning threshold, with the second warning threshold being higher than the first warning threshold; When the dynamic methylation risk score rises twice consecutively, and the most recent score exceeds the first warning threshold, a low risk warning for relapse is generated. A high-risk warning for progression is generated when the dynamic methylation risk score rises rapidly and the most recent score exceeds the second warning threshold, while the driving haplotype load component at that time point is significantly higher than the historical monitoring value.

[0016] The beneficial effects of this invention are: 1. This invention effectively purifies the monitoring signal by separating and jointly analyzing two types of functionally distinct methylation differential regions—the "passenger" region and the "driver" region. Specifically, utilizing the extremely high sensitivity of the "passenger" region haplotype to clonal changes, early clonal proliferation under extremely low tumor burden can be detected; simultaneously, utilizing the predefined "malignant haplotype" in the "driver" region, which is strongly associated with the invasive phenotype, ensures the specificity of the detection signal and effectively eliminates interference from nonspecific methylation changes. This allows the invention to achieve highly sensitive and specific early warning of recurrence even when imaging or cystoscopy is negative. Clinical validation data show that this method has a sensitivity of over 90% and a specificity of over 85% for detecting bladder cancer recurrence, significantly superior to traditional urine exfoliative cytology or single-point methylation detection methods. 2. Existing technologies mostly provide qualitative or semi-quantitative results at a single time point, failing to assess risk trends. This invention, through the design of a series of continuous sampling procedures and the construction of a dynamic methylation risk scoring model, transforms monitoring results into a quantitative risk curve that changes continuously over time for the first time. This score integrates the "passenger haplotype instability index," reflecting clonal stability, and the "driving haplotype burden," reflecting malignant potential. Physicians can prospectively assess a patient's recurrence risk status by observing the dynamic trend of this risk score (e.g., stable, slow rise, or rapid increase), rather than a single threshold, thereby developing individualized follow-up intervals and intervention strategies. This avoids the over-examination or delayed treatment that may result from traditional "one-size-fits-all" follow-up protocols, achieving truly precise and dynamic monitoring. 3. Existing non-invasive methods struggle to distinguish between indolent recurrence and aggressive progression. The core innovation of this invention lies in its ability to specifically predict the deterioration of tumor biological behavior by continuously tracking the "driving haplotype burden" indicator. When monitoring data shows a significant increase in "driving haplotype burden," especially when this increase dominates the growth of the dynamic methylation risk score, it strongly suggests that the recurrent clone is evolving towards a higher-grade, muscle-invasive phenotype, even if the total tumor burden is not high. This information is crucial for clinical decision-making. For example, a "high-risk progression warning" strongly suggests the need for radical cystectomy rather than conservative secondary transurethral resection, thus securing the best treatment opportunity for the patient and improving prognosis. This is a core function that existing urine-based non-invasive testing methods lack. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, those skilled in the art can obtain other drawings based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the steps of the method of the present invention. Detailed Implementation

[0019] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. It should also be noted that, to make the embodiments more comprehensive, the following embodiments are the best and preferred embodiments, and those skilled in the art can use other alternative methods to implement some well-known technologies; moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0020] Please see Figure 1 This invention provides a method for dynamic monitoring of bladder cancer based on circulating tumor DNA methylation markers. First, urine samples are systematically collected from bladder cancer patients at multiple consecutive predetermined time points after surgery. These samples undergo standardized laboratory pretreatment procedures to obtain high-quality, cell-free urine supernatant as the starting material for subsequent analysis. Next, circulating cell-free DNA is extracted from the treated supernatant. This step employs optimized extraction techniques to efficiently recover all DNA, including short fragments. The extracted circulating cell-free DNA then undergoes precise quantification and fragment size distribution analysis to assess its quality and potential enrichment in tumor-derived DNA.

[0021] Subsequently, the focus shifted to molecular detection. Targeted sequencing libraries were designed and constructed for a group of bladder cancer-specific methylation differentially expressed regions, pre-screened through bioinformatics analysis and experimental validation. These regions were clearly divided into two functionally complementary categories: the methylation haplotype patterns of the first category maintained high intra-individual stability before and after cell carcinogenesis; the methylation haplotype patterns of the second category underwent specific rearrangements during tumor malignant progression. The constructed libraries were then subjected to high-throughput sequencing, generating massive amounts of sequence data. Through bioinformatics analysis of this data, the methylation haplotype patterns at consecutive CpG sites within each target region could be precisely interpreted.

[0022] Based on the parsed haplotype data, core indicators are calculated. The first indicator is the passenger haplotype instability index, which quantifies the degree of shift in the dominant haplotype in the first region over time, reflecting the stability of the uroepithelial clone. The second indicator is the driving haplotype burden, which calculates the proportion of predefined haplotypes associated with high malignancy in the second region, reflecting the abundance of tumor clones with invasive potential. Finally, these two indicators are input into a pre-defined algorithm model to generate a single, continuous number: the dynamic methylation risk score. By tracking the trajectory and trend of this score in a continuous time series, the system can automatically generate dynamic monitoring reports containing different levels of risk warnings, thereby achieving non-invasive, dynamic assessment of bladder cancer recurrence and progression risk at the molecular level.

[0023] In one possible implementation, the collection and pretreatment steps for a series of urine samples were specifically detailed. The setting of multiple consecutive preset time points follows the clinical pattern of bladder cancer postoperative recurrence monitoring, for example, at 1 month, 3 months, 6 months, and 12 months post-surgery, followed by collection every 6 months thereafter. This protocol balances monitoring density with patient burden and captures molecular changes within critical time windows. At each collection, the patient is required to collect approximately 100 ml of midstream urine, which is immediately transferred to a dedicated preservation tube containing a DNase inhibitor. This procedure aims to immediately stabilize the sample and prevent nucleic acid degradation. Samples are transported to the laboratory under a cold chain condition of 2 to 8 degrees Celsius within a specified time after collection, and the processing flow is initiated within 4 hours of receipt to maximize analyte integrity.

[0024] Pretreatment employed a standard two-step centrifugation method, a routine technique for separating urine supernatant in clinical testing. The first step involved centrifuging the urine sample at 4°C and 1500 rpm for 10 minutes. This step aims to precipitate most cells, crystals, and large particles in the urine using moderate centrifugal force. After centrifugation, the supernatant was carefully aspirated, yielding a preliminarily clear urine supernatant, while avoiding disturbance of the sediment at the bottom. The second step involved transferring the obtained supernatant to a new centrifuge tube and centrifuging again at 4°C and 13000 rpm for 10 minutes. This high-speed centrifugation step aims to thoroughly remove residual microcellular debris, exosomes, protein aggregates, and other subcellular impurities from the supernatant, resulting in a high-purity final supernatant suitable for subsequent circulating cell-free DNA extraction. This series of standardized procedures ensured consistency in sample processing conditions across different time points and batches, laying the foundation for the stability and comparability of downstream molecular assays.

[0025] In one possible implementation, the extraction process utilizes a commercially available kit based on the principle of a silica membrane adsorption column, a well-established method for extracting trace amounts of circulating cell-free DNA. To improve the recovery efficiency of circulating tumor DNA from bladder cancer, particularly considering its short fragments, key adjustments were made to the standard operating procedure: the pH and salt concentration of the binding buffer were optimized in the crucial step of DNA binding to the silica membrane. Specifically, the buffer was adjusted to a slightly acidic environment and the concentration of monovalent cations was increased. These conditions have been shown to significantly enhance the binding affinity of short DNA fragments (less than 200 base pairs) to the adsorption column, thereby improving their recovery rate.

[0026] After extraction, the product needs to be accurately quantified and its quality assessed. A high-sensitivity quantitative PCR instrument is used to determine the total concentration of extracted circulating free DNA using fluorescent dyes that specifically bind to double-stranded DNA; the results are expressed in nanograms per microliter. Simultaneously, fragment analysis is performed to assess the potential enrichment of circulating tumor DNA in the sample. This analysis is typically performed using a microcapillary electrophoresis system, which can separate and quantify DNA fragments of different sizes at high resolution. By analyzing the electrophoretic patterns using system software, the percentage of DNA molecules with fragment sizes falling within the characteristic range of 100 to 250 base pairs can be accurately calculated as a percentage of the total detected circulating free DNA. This percentage is an important quality indicator; a higher percentage generally indicates a greater proportion of short DNA fragments generated by tumor cell apoptosis or necrosis in the sample, meaning that circulating tumor DNA may be more enriched.

[0027] In one possible implementation, the present invention uses a total of ten differentially methylated regions, including five Class I regions and five Class II regions. The screening of Class I regions is based on comparative analysis of large-scale whole-genome methylation sequencing data. Researchers compared the methylation profiles of a large number of normal urothelial tissues and bladder cancer tissues, screening for genomic regions that exhibit stable hypermethylation in both tissues, but whose haplotype patterns formed by multiple consecutive CpG sites are polymorphic among different individuals, yet highly consistent before and after cancer development in the same individual. These regions act like stable "epigenetic fingerprints," and their changes indicate clonal succession.

[0028] The second type of region screening focuses on tumor malignant progression. By comparing methylation data from non-muscle-invasive and muscle-invasive bladder cancer tissues, regions specifically exhibiting hypermethylation or specific haplotype rearrangements in the latter were identified. These specific haplotype patterns significantly increased in frequency during disease progression and were associated with biological pathways driving tumor invasiveness. After identifying the target regions, specific polymerase chain reaction (PCR) primers were designed for each region and mixed to form a multiplex PCR primer pool. Library construction began with bisulfite treatment of circulating cell-free DNA, a standard chemical method that converts unmethylated cytosine to uracil while leaving methylated cytosine unchanged. Subsequently, the transformed DNA was amplified twice using the primer pool: the first round introduced sample tags, and the second round added sequencing adapters. After purification and fragment selection, the amplified products constituted a library ready for sequencing. To reduce batch error, libraries from all time points of the same patient were accurately quantified, mixed in equimolar proportions, and then sequenced.

[0029] In one possible implementation, the construction process of the targeted methylation haplotype sequencing library begins with a complete bisulfite conversion of the circulating cell-free DNA extracted in step two. This is a standard pretreatment step in epigenomics, using chemical reagents to selectively convert unmethylated cytosine nucleotides in the DNA to uracil, while methylated cytosine remains unchanged. After this step, the methylation information on the DNA sequence is converted into base sequence differences detectable by conventional DNA sequencing techniques. The converted DNA is then purified and recovered for subsequent amplification.

[0030] The core amplification step utilizes a pre-designed and validated multiplex PCR primer pool. This pool contains multiple primer pairs, each specifically amplifying a selected differentially methylated region. The first round of PCR uses transformed DNA as a template and employs specific primers with unique sample index sequences. Each sample's primers have different indices, allowing subsequent sequencing data from multiple samples to be distinguished at the bioinformatics level. The second round of PCR uses the first-round product as a template and further amplifies it using universal primers with sequencing platform adapter sequences, enabling successful bridging PCR and sequencing reactions on a high-throughput sequencer. The entire PCR process uses hot-start high-fidelity DNA polymerase to ensure amplification accuracy and efficiency. After amplification, the product is purified using a magnetic bead purification system, selectively recovering fragments within the target length range to remove primer dimers and other impurities. Finally, the purified library is quantitatively analyzed using fluorescence, and libraries prepared from different time points from the same patient are calculated based on their concentrations and mixed in equimolar amounts to prepare the final mixed library for deep sequencing.

[0031] In one possible implementation, the constructed mixed library is loaded into a high-throughput sequencing platform, such as a sequencer based on the principle of synthetic sequencing. To ensure sufficient coverage of each target differentially methylated region to accurately identify low-frequency haplotypes, the average sequencing depth for each sample is set to be no less than 5000X. The raw image data generated by sequencing is converted into a base sequence file, i.e., a Fastq file containing sequence information and quality fraction, by the instrument software.

[0032] Subsequent bioinformatics analysis is crucial for haplotype resolution. First, the raw data undergoes quality control, filtering out low-quality reads and removing sequencing adapter sequences. Then, using alignment software specifically designed for processing bisulfite-converted sequences, such as Bismark, high-quality reads are aligned with a bisulfite-converted human reference genome to determine the genomic origin of each read. For reads successfully aligned to ten predetermined target regions, methylation status is extracted. For a specific methylation-discrete region covered by a read, the software sequentially examines each CpG dinucleotide site on that read sequence. Depending on whether the base at that site is cytosine (representing a methylated state initially) or thymine (representing an unmethylated state initially, derived from uracil), it is recorded as a methylation status of "1" or "0". Thus, a read covering multiple consecutive CpG sites is converted into a binary string consisting of "0"s and "1"s; this string is defined as a methylation haplotype. Finally, all distinct binary strings observed in each methylation-discrete region and the number of sequencing reads supporting them are counted.

[0033] In one possible implementation, the calculation logic of the passenger haplotype instability index is explained in detail. This index is calculated based on haplotype data from the first type of methylation differential regions. First, a baseline time point needs to be determined, typically chosen as the molecular baseline at the time of the first postoperative monitoring (e.g., 1 month postoperatively). For each of the five first-type regions, the methylation haplotype with the highest sequencing read support number in the sample at that baseline time point is identified and defined as the "baseline dominant haplotype" for that region in that specific patient.

[0034] At each subsequent monitoring time point, this analysis is repeated for each Category I region: the haplotype with the highest support count in that region at the current time point is identified as the "current dominant haplotype." Next, a precise string comparison is performed between the current dominant haplotype and the baseline dominant haplotype for that region. If the two strings are identical, the region is considered to have not shifted at the current time point; if the two strings differ by at least one site, the region is considered to have experienced a "haplotype shift." This shift suggests that the DNA in the urine at the current time point may primarily originate from cell clones different from those at baseline. After traversing all five Category I regions, the number of regions that have shifted at the current time point is counted. The passenger haplotype instability index is obtained by dividing the number of shifted regions by the total number of regions, five. This index is a value between 0 and 1; a higher value indicates a more significant change in the clonal composition of urinary epithelial cells, suggesting the possible early proliferation of new tumor clones.

[0035] In one possible implementation, a method for calculating the driving haplotype load is specifically described. The calculation of this index relies on the identification and quantification of predefined “malignant haplotypes” in the differentially methylated regions of type II. The definition of malignant haplotypes originates from previous discoveries: sequencing of a large number of bladder cancer tissue samples with known pathological stages and grades revealed specific methylation haplotype patterns in five type II regions. These patterns are highly frequent in high-grade, muscle-invasive bladder cancer tissues, but extremely rare or absent in low-grade, non-invasive cancers or normal tissues. These haplotype patterns strongly associated with malignant phenotypes are predefined as “malignant haplotypes.”

[0036] At each time point of dynamic monitoring, each Category II region is analyzed. First, all valid sequencing reads aligned to that region are acquired. Then, the haplotype sequence (binary string) of each read is compared one by one with a predefined list of malignant haplotypes for that region. If a read's sequence perfectly matches any predefined malignant haplotype, that read is counted as a malignant read. The number of malignant reads in that region is calculated and divided by the total number of valid reads in that region to obtain the percentage of malignant haplotypes in that region at the current time point. After calculating for all five Category II regions, the percentage value of each region is multiplied by its corresponding preset weighting coefficient, and the five weighted values ​​are summed to obtain the driving haplotype load at that time point. The weighting coefficients are preset during the model training phase based on the specificity and robustness of each region's malignant haplotype in predicting aggressive phenotypes. This load value reflects the relative abundance of tumor clones with high malignant potential in the current urine DNA sample.

[0037] In one possible implementation, the dynamic methylation risk score is not a simple addition of the passenger haplotype instability index and the driving haplotype load, but rather an organic integration through a pre-trained mathematical model. This mathematical model is based on a historical patient cohort. This cohort includes a sufficient number of post-bladder cancer surgery patients who have undergone long-term serial urinary monitoring and have clearly documented clinical endpoint events, namely, recurrence and the pathological grade and stage of the recurrent tumor.

[0038] Passenger haplotype instability index and driving haplotype load were collected from all patients in the cohort at various monitoring time points and correlated with clinical outcomes after those time points. Statistical modeling methods, such as logistic regression, were used to fit the mathematical relationship between these two indices and the risk of future relapse or progression. The goal of model training was to find an optimal function that, based on the input index and load values, outputs a probability value—the probability that the patient is in a high-risk state. This function is the final integrated model, and its specific parameters are fixed after training.

[0039] In clinical applications, for a new patient, at a specific monitoring point, only the calculated passenger haplotype instability index and driving haplotype load value need to be input into this pre-trained mathematical model with fixed parameters. The model will automatically perform the calculation and output a value between 0 and 1, which is the dynamic methylation risk score. The closer the score is to 1, the higher the probability of relapse or progression in the near future based on the current molecular characteristics. This model integration method based on historical data training can more scientifically and robustly fuse information from two dimensions into a comprehensive indicator with clear clinical predictive significance.

[0040] In one possible implementation, the core of the dynamic monitoring report and risk warning generation rules is to transform continuous dynamic methylation risk scores into clinically actionable decision support information. First, the system connects the dynamic methylation risk scores of the same patient at all monitoring time points in chronological order on a coordinate graph, creating a clear risk trend curve. This curve visually illustrates the trajectory of the patient's molecular-level risk over time.

[0041] To provide tiered early warning, the system presets two warning thresholds: a first warning threshold and a second warning threshold, with the second threshold being higher than the first. The warning rules consider both the absolute value and trend of the score. When system analysis reveals that a patient's dynamic methylation risk score shows an upward trend in the two most recent consecutive monitoring sessions, and the most recent score exceeds the preset first warning threshold, the system determines that the "low-risk recurrence warning" condition is met. This warning indicates the presence of molecular markers that may be associated with early tumor recurrence, and clinical follow-up is recommended.

[0042] When system analysis reveals a rapid and significant increase in the dynamic methylation risk score, with the most recent score exceeding a higher second warning threshold, and combined with data analysis showing a significant contribution from the haplotype burden driver in this round of score increase (i.e., a clear increase compared to the historical baseline), the system determines that the "high-risk progression warning" criteria are met. This warning not only indicates a high probability of recurrence but also strongly suggests that the recurrent tumor may have a highly aggressive phenotype, requiring immediate clinical intervention for confirmation. The system automatically generates a standardized monitoring report that includes the aforementioned trend chart, warning level, interpretation of key indicators, and clinical recommendations.

[0043] Example This embodiment simulates and elaborates in detail the entire process of dynamic monitoring for 36 months using the method of this invention on a patient diagnosed with non-muscle-invasive bladder cancer after undergoing transurethral resection of bladder tumor.

[0044] 1. Collection and pretreatment of a series of urine samples; Patients begin the monitoring process after surgery. The monitoring time points are strictly implemented according to the preset plan, that is, sampling is carried out at 1 month, 3 months, 6 months, 12 months, 18 months, 24 months, 30 months and 36 months after surgery.

[0045] At each sampling time point, patients were instructed to collect a midstream urine sample taken in the morning. 100 ml was precisely measured using a measuring cup and immediately poured into a urine preservation tube pre-filled with 10 ml of a special preservation solution. This solution contains EDTA and sodium azide, which effectively inhibits DNase activity and prevents bacterial growth. The collected sample tubes were immediately transported in a 4°C incubator to ensure they reached the laboratory for processing within 4 hours of collection.

[0046] Sample processing employed a two-step centrifugation method: First, centrifuge the urine sample containing the mixed preservation solution at 4 degrees Celsius and 1500 rpm for 10 minutes. After centrifugation, carefully transfer the supernatant to a new 50 ml centrifuge tube, and retain the cell pellet at the bottom for later use (it can be used for other tests, such as urine exfoliative cytology).

[0047] The second step involves centrifuging the transferred supernatant again at 4°C and 13,000 rpm for 10 minutes. This step aims to thoroughly remove residual cell debris, microorganisms, and other small particles. After centrifugation, the clear supernatant is aliquoted into 2 mL cryovials (1 mL per tube), labeled “cfDNA Extraction Supernatant,” and used immediately for DNA extraction or stored at -80°C for long-term preservation. In this example, all samples were immediately proceeded to the extraction process after processing.

[0048] 2. Extraction and quantification of circulating cell-free DNA in urine; Extraction was performed using a commercially available circulating cell-free DNA extraction kit (based on the principle of silica membrane adsorption column) suitable for bodily fluid samples. To optimize the recovery of short-fragment circulating tumor DNA, the standard operating procedure was modified as follows: In the DNA binding step, the pH of the binding buffer was adjusted to 6.2, and guanidine isothiocyanate was added to a final concentration of 1.8 M. These conditions facilitated the efficient binding of DNA fragments shorter than 200 base pairs to the silica membrane.

[0049] After extraction, the extracted circulating cell-free DNA was precisely quantified using a Qubit real-time fluorescence analyzer and a dsDNAHS detection kit. Simultaneously, 1 μL of the extracted product was analyzed for fragmentation using an Agilent 2100 bioanalyzer and a high-sensitivity DNA chip. The system software automatically generated a fragment distribution map and calculated the percentage of DNA molecules with fragment sizes between 100 and 250 base pairs. For example, in the sample taken one month post-surgery (baseline), the circulating cell-free DNA concentration was 5.2 ng / μL, and fragments of 100-250 base pairs accounted for 68%.

[0050] 3. Construction of targeted methylation haplotype sequencing libraries; This invention selects 10 differentially methylated regions (DMRs), which are divided into two categories: Category 1 differentially methylated regions (5): For example, specific regions located in the promoter regions of genes ZNF154, HOXA9, POU4F2, EOMES, and TWIST1. These regions are hypermethylated in both normal urothelial cells and bladder cancer cells, but the specific haplotype pattern formed by continuous CpG sites within them remains remarkably consistent between normal and cancerous tissues of the same individual, like a stable "molecular fingerprint," hence the name "passenger" regions.

[0051] The second category of differentially methylated regions (5): For example, specific regions located in the gene bodies or enhancer regions of genes VIM, TMEFF2, CCND2, RARB, and CDH1. Specific haplotype patterns in these regions are significantly enriched in muscle-invasive bladder cancer tissues, and their frequency is positively correlated with tumor grade and stage. They are considered to drive the aggressive phenotype of the tumor and are therefore called "driving" regions.

[0052] The document library construction process is as follows: a. Bisulfite conversion: Take 20 nanograms of extracted circulating free DNA and use EZDNAMethylation-GoldKit to convert unmethylated cytosine to uracil.

[0053] b. Multiplex PCR amplification: Design a multiplex PCR primer pool covering the 10 DMR regions mentioned above (each region covers approximately 150 base pairs and contains at least 8 consecutive CpG sites). Perform two rounds of PCR on the transformed DNA using a hot-start high-fidelity DNA polymerase.

[0054] First round PCR: The reaction system includes transformed DNA and a multiplex primer pool (each primer carries a specific sample index sequence). Cycling conditions: 95°C pre-denaturation for 5 minutes; followed by 18 cycles of 98°C denaturation for 20 seconds, 60°C annealing for 30 seconds, and 72°C extension for 30 seconds; and a final extension at 72°C for 5 minutes.

[0055] Second round PCR: Using the first round product as a template, amplification was performed using universal primers with universal sequencing adapters, with 10 cycles.

[0056] c. Library purification and quantification: PCR products are purified and size-selected using magnetic beads to remove primer dimers and excessively long fragments, retaining the target product of 200-350 base pairs. The final library is then accurately quantified using qPCR.

[0057] d. Library mixing: Libraries constructed from the same patient at 8 different time points are mixed in equimolar amounts (e.g., 10 nanomolars each) according to their quantitative concentrations to form a mixed library pool for sequencing.

[0058] 4. High-throughput sequencing and methylation haplotype analysis; The mixed library pool was sequenced at 2 × 150 base pairs on an Illumina NovaSeq 6000 platform. To ensure sufficient coverage depth in each DMR region for accurate haplotype resolution at each time point, the average sequencing depth was set to 8000X per sample.

[0059] The data from the anatomical machine (Fastq file) is processed through the following bioinformatics workflow: a. Quality control and adapter removal: Use Fastp software to remove low-quality reads and sequencing adapters.

[0060] b. Alignment: The quality-controlled readings were aligned with the bisulfite-converted human reference genome (hg38) using Bismark software.

[0061] c. Methylation Haplotype Extraction: For each reading uniquely aligned to one of the 10 target DMR regions, the methylation status of all CpG sites covered by that reading is extracted. For example, a reading containing 5 CpG sites, if its methylation status is sequentially "methylated, unmethylated, methylated, methylated, unmethylated", is encoded as the binary string "10110", representing a methylation haplotype. All unique haplotypes observed within each DMR region are counted.

[0062] 5. Calculate the dynamic methylation risk score; This step is the core calculation process of this method, involving the calculation and final integration of two independent indicators.

[0063] 5.1 Calculate the passenger single-type instability index; The baseline was the sample taken one month post-surgery. For each region of differential methylation in the first category, the haplotype with the highest number of support readings in the baseline sample was identified and defined as the "baseline dominant haplotype" for that region. For example, in the ZNF154 region, the baseline dominant haplotype was "11101101" (850 support readings).

[0064] At each subsequent time point, it is determined whether the current dominant haplotype in the region is the same as the baseline dominant haplotype. If they are different, the region is considered to have experienced a "haplotype shift". For example, in month 24, if the dominant haplotype in region ZNF154 changes to "11101001" (supporting 620 readings), which is different from the baseline, then the region is marked as "shifted".

[0065] Passenger haplotype instability index at a point in time The calculation formula is: in, It is the number of regions where "haplotype shift" occurs in the five Class I methylation differential regions.

[0066] 5.2 Calculate the driving single-type load; For each differentially methylated region of type II, one to two "malignant haplotypes" were predefined based on previous analysis in the training cohort. For example, in the VIM region, the haplotype "00110011" occurred in up to 85% of muscle-invasive bladder cancer tissues, but less than 5% of non-muscle-invasive bladder cancer tissues, and was therefore predefined as a malignant haplotype.

[0067] At each monitoring time point, the percentage of sequencing reads matching a predefined malignant haplotype in each region of differential methylation in category II was calculated out of the total sequencing reads in that region. .

[0068] A single-load drive at a point in time The calculation formula is: in, Representing the A second-class methylation differential region, It is based on the pre-set weighting coefficients of the specificity of malignant haplotype prediction in each region. For example, the weight of the VIM region. It is set to 0.25 because it has the highest predictive value.

[0069] 5.3 Integrated into a dynamic methylation risk score; Will and Two metrics are input into a pre-trained logistic regression model, which outputs a dynamic methylation risk score. .

[0070] This logistic regression model was trained based on historical monitoring data from a cohort of 200 bladder cancer patients. This cohort collected urine samples from each patient at multiple time points within at least 3 years post-surgery, along with the final clinical recurrence / pathological progression outcome. The model format is as follows: The model parameters are obtained by fitting the model using maximum likelihood estimation: intercept. , coefficient , coefficient The model achieved an area under the curve (AUC) of 0.93 for predicting relapse / progression on the independent validation set.

[0071] 6. Generate dynamic monitoring reports and risk warnings; The dynamic methylation risk scores of patients at eight time points were plotted as trend graphs. Simultaneously, two warning thresholds were set: a first warning threshold... Second warning threshold .

[0072] In this embodiment, the calculated values ​​of key indicators for the patient at various time points are shown in the table below: Report Interpretation and Clinical Decision Support: Months 1-18: The risk score remains below 0.10, and the indicators are stable, indicating no signs of recurrence at the molecular level. Regular annual check-ups can be maintained.

[0073] Month 24: The score rose to 0.28. Analysis revealed that the driving force was the singleton load. It remains close to zero, but the passenger haplotype instability index The score increased from 0.2 to 0.4 (two out of five regions experienced deviation). According to the warning rules (scores continuously rising and > 0.4), the score was positive. The system automatically generates a "low-risk recurrence warning" report, indicating signs of clonal hyperplasia, but with no obvious malignant characteristics. It is recommended that clinicians shorten the interval between cystoscopy follow-ups from one year to 3-6 months.

[0074] Month 30: The score rapidly climbed to 0.52, exceeding the second warning threshold. More importantly, it drives single-type loads. A significant increase was observed, from 0.01 to 0.05. This indicates not only a potential increase in the number of clones but also a shift towards a malignant phenotype. The system generates a "High Risk of Progression Warning" report, strongly suggesting the presence of high-grade or invasive tumor recurrence. Cystoscopy should be performed immediately.

[0075] Clinical validation: At 31 months, the patient underwent a follow-up cystoscopy, during which a new solid tumor approximately 5 mm in diameter was discovered in the original surgical area. After transurethral resection, the pathological diagnosis was high-grade urothelial carcinoma with lamina propria invasion (T1 stage), confirming the high-risk progression warning issued by this method.

[0076] To objectively demonstrate the advantages of the method of the present invention, the patient monitoring data in this embodiment is compared with the simulation results of two existing conventional monitoring methods.

[0077] Comparative Example 1: Traditional single-point urine methylation detection (commercial kit); This method only detects the overall methylation level (expressed as a percentage) of 2-3 gene promoter regions at a single time point, and gives a binary judgment of "positive" or "negative", without dynamic trend analysis.

[0078] Comparative Example 2: Standard clinical follow-up protocol; This refers to the actual clinical follow-up received by the patients in this embodiment, including regular ultrasound, CT, and scheduled cystoscopy examinations (at the 12th and 24th months).

[0079] This embodiment demonstrates that, compared to traditional single-point detection and standard imaging / invasive endoscopic follow-up, the method of this invention, through dynamic integrated analysis of "passenger-driven" dual-region haplotypes, can detect signs of tumor recurrence earlier (at least 6 months in advance) and more accurately (distinguishing recurrence risk levels and providing early warning of progression), providing a crucial time window and decision-making basis for clinical intervention, and achieving truly non-invasive, dynamic, and risk-stratified monitoring.

[0080] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0081] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for dynamic monitoring of bladder cancer based on circulating tumor DNA methylation markers, characterized in that, Includes the following steps: Step 1: Collect and process urine samples at multiple consecutive pre-set time points after bladder cancer surgery to obtain urine supernatant for extracting circulating cell-free DNA. Step 2: Extract circulating cell-free DNA from urine supernatant and perform quantification and fragment analysis on the extracted circulating cell-free DNA; Step 3: Construct a targeted methylation haplotype sequencing library for pre-selected bladder cancer-specific methylation differential regions; the methylation differential regions include a first type of methylation differential region and a second type of methylation differential region; the first type of methylation differential region is the region where the methylation haplotype remains stable before and after carcinogenesis; the second type of methylation differential region is the region where the methylation haplotype is rearranged during tumor progression; Step 4: Perform high-throughput sequencing on the sequencing library to resolve the methylation haplotypes of all consecutive CpG sites in the two types of methylation difference regions; Step 5: Based on the resolved methylation haplotype data, calculate the passenger haplotype instability index and driving haplotype load for each monitoring time point, and integrate the passenger haplotype instability index and driving haplotype load into a dynamic methylation risk score; the passenger haplotype instability index represents the degree of shift of dominant haplotypes in the first type of methylation difference region; the driving haplotype load represents the abundance of predefined malignant haplotypes in the second type of methylation difference region; Step 6: Generate dynamic monitoring reports and risk warnings based on the changing trends of dynamic methylation risk scores at continuous time points.

2. The method for dynamic monitoring of bladder cancer based on circulating tumor DNA methylation markers according to claim 1, characterized in that, In step 1, the multiple consecutive preset time points include the 1st month, 3rd month, 6th month, and 12th month after surgery, and then collected every 6 months thereafter; Collect 100 ml of midstream urine from the patient each time and immediately inject it into a urine preservation tube containing a DNase inhibitor. Store the urine at 2-8 degrees Celsius and complete the processing within 4 hours after collection. The process includes a two-step centrifugation process: the first step is centrifugation at 4 degrees Celsius and 1500 rpm for 10 minutes to separate the upper layer of urine supernatant and the lower layer of cell precipitate. The second step involves transferring the urine supernatant obtained from the first step to a new centrifuge tube and centrifuging it again for 10 minutes at 4 degrees Celsius and 13,000 rpm to remove cell debris and impurities. The resulting supernatant is the urine supernatant used to extract circulating cell-free DNA.

3. The method for dynamic monitoring of bladder cancer based on circulating tumor DNA methylation markers according to claim 1, characterized in that, In step 2, the extraction of circulating free DNA uses a silica membrane adsorption column kit, and the pH and salt concentration of the binding buffer are adjusted during the extraction process to preferentially recover DNA fragments with a length of less than 200 base pairs. The quantitative and fragment analysis includes: determining the total concentration of extracted circulating free DNA using a fluorescence quantitative quantification instrument; Meanwhile, the fragment size distribution of circulating cell-free DNA was analyzed using a microcapillary electrophoresis system, and the percentage of circulating cell-free DNA with fragment sizes between 100 and 250 base pairs was calculated. This percentage was used as one of the indicators to assess the enrichment of circulating tumor DNA in the sample.

4. The method for dynamic monitoring of bladder cancer based on circulating tumor DNA methylation markers according to claim 1, characterized in that, In step 3, the methylation difference regions are 5 first-type methylation difference regions and 5 second-type methylation difference regions; The first type of differentially methylated regions were obtained by comparing and screening a large amount of whole-genome methylation sequencing data of normal urothelial tissue and bladder cancer tissue. The screening criteria were: high methylation in both normal and cancerous tissues, and the haplotype pattern formed by continuous CpG sites within the region was stable in different individuals and remained consistent before and after cancer in the same individual. The second type of differentially methylated regions were obtained by comparing and screening whole-genome methylation sequencing data of non-muscle-invasive bladder cancer tissues and muscle-invasive bladder cancer tissues. The screening criteria were: specific hypermethylation state or haplotype pattern in muscle-invasive bladder cancer tissues, and the frequency of this haplotype pattern increased significantly during the transformation from non-muscle-invasive bladder cancer to muscle-invasive bladder cancer.

5. The method for dynamic monitoring of bladder cancer based on circulating tumor DNA methylation markers according to claim 4, characterized in that, In step 3, the specific steps for constructing the targeted methylation haplotype sequencing library are as follows: First, the extracted circulating free DNA is subjected to bisulfite conversion treatment to convert unmethylated cytosine into uracil; Then, the transformed DNA was amplified in two rounds using a multiplex PCR primer pool designed for 10 differentially methylated regions. The first round of amplification introduced sample-specific index sequences, and the second round of amplification introduced universal sequencing adapters. Finally, the amplification products are purified and fragments of a specific size range are selected to complete library construction; Sequencing libraries from all time points of the same patient were mixed in equimolar amounts after quantification to enable parallel sequencing in the same sequencing process.

6. The method for dynamic monitoring of bladder cancer based on circulating tumor DNA methylation markers according to claim 1, characterized in that, In step 4, the depth of the high-throughput sequencing is no less than 5000X for each sample; The process of resolving methylation haplotypes includes: aligning sequencing reads to a reference genome and locating the target methylation differential region; For each reading that corresponds to a region of methylation difference, the methylation status of each CpG site covered by the reading is determined sequentially according to its sequence. The methylation status is recorded as 1, and the non-methylation status is recorded as 0, forming a binary string composed of 0 and 1. This string is a methylation haplotype. We counted all unique methylation haplotypes appearing in each methylation differential region and their corresponding sequencing read support numbers.

7. A method for dynamic monitoring of bladder cancer based on circulating tumor DNA methylation markers according to claim 6, characterized in that, In step 5, the specific process of calculating the passenger haplotype instability index is as follows: for each type I methylation difference region, the first postoperative monitoring time point is determined as the baseline time point, and the dominant haplotype in that region at the baseline time point is identified; At each subsequent monitoring time point, it is determined whether the dominant haplotype in this region is the same as the dominant haplotype at the baseline time point; If they are different, it is determined that a haplotype shift occurred in that region at that time point; The passenger haplotype instability index at a given time point is equal to the ratio of the number of haplotype shifts occurring in the five type I methylation differential regions at that time point to 5.

8. A method for dynamic monitoring of bladder cancer based on circulating tumor DNA methylation markers according to claim 6, characterized in that, In step 5, the specific process of calculating the driving haplotype load is as follows: for each type II methylation differential region, one or more specific methylation haplotypes that appear in the region and are strongly associated with muscle-invasive bladder cancer are predefined as malignant haplotypes; At each monitoring time point, the number of sequencing reads that perfectly match any predefined malignant haplotype among all sequencing reads aligned to that region is counted. Calculate the percentage of this number relative to the total sequencing reads in that region; The driving haplotype load at a given time point is equal to the weighted average of the percentage values ​​of the five differentially methylated regions of type II methylation. The weighting coefficients are pre-set based on the specificity of each region's malignant haplotype in predicting muscle-invasive bladder cancer.

9. A method for dynamic monitoring of bladder cancer based on circulating tumor DNA methylation markers according to claim 7 or 8, characterized in that, In step 5, the method for integrating into a dynamic methylation risk score is as follows: the passenger haplotype instability index and the driving haplotype load at the same time point are used as two feature variables and input into a pre-trained logistic regression model. The logistic regression model was trained based on historical patient cohort data, which included passenger haplotype instability index and driving haplotype load at a series of monitoring time points for bladder cancer patients with known final recurrence or progression outcomes. The model outputs a value between 0 and 1, which is the dynamic methylation risk score at that time point.

10. A method for dynamic monitoring of bladder cancer based on circulating tumor DNA methylation markers according to claim 1, characterized in that, In step 6, the specific method for generating dynamic monitoring reports and risk warnings is as follows: connect the dynamic methylation risk scores of the same patient at all continuous monitoring time points in chronological order and draw a risk trend curve. Set a first warning threshold and a second warning threshold, with the second warning threshold being higher than the first warning threshold; When the dynamic methylation risk score rises twice consecutively, and the most recent score exceeds the first warning threshold, a low risk warning for relapse is generated. A high-risk warning for progression is generated when the dynamic methylation risk score rises rapidly and the most recent score exceeds the second warning threshold, while the driving haplotype load component at that time point is significantly higher than the historical monitoring value.

Images