Methylation biomarker panel for diagnosing pancreatobiliary cancer

Abstract

The present invention belongs to the field of biomedicine, and particularly relates to a methylation biomarker combination for diagnosing pancreatobiliary cancer. Specifically, the methylation biomarker includes one or more of SOX17, 3-OST-2, NXPH1, SEPT9, and TERT.

Description

Technical Field

[0001] The present invention belongs to the field of biomedicine, and particularly relates to a methylation biomarker combination for diagnosing pancreatobiliary cancer. Background Art

[0002] Pancreatobiliary tract cancer includes bile tract cancer (BTC) and pancreatic cancer.

[0003] Bile tract cancer originates from cholangiocytes at different anatomical locations, such as intrahepatic, extrahepatic and gallbladder, or may directly originate from hepatocytes. Bile tract cancer includes cholangiocarcinoma (CCA), gallbladder cancer (GBC) and ampullary cancer. Cholangiocarcinoma is the second most common primary liver cancer, accounting for about 3% of all gastrointestinal tumors. Cholangiocarcinoma is further divided into intrahepatic (iCCA), perihilar (pCCA) or distal (dCCA) cholangiocarcinoma. Although bile tract cancer is relatively uncommon globally, its global incidence has increased rapidly in recent years, and it has the highest incidence in parts of East and South Asia and South America. Risk factors for bile tract cancer include primary sclerosing cholangitis (PSC), liver flukes, fibropolycystic liver diseases (such as bile duct adenomas and bile duct papillomatosis), biliary and gallbladder stones, viral hepatitis and exposure to chemical carcinogens, etc. Pancreatic cancer is one of the cancers with the highest lethality rate globally and is the seventh leading cause of cancer death worldwide due to its poor prognosis, specifically including pancreatic head cancer, pancreatic tail cancer, diffuse cancer, etc.

[0004] The diagnosis of pancreaticobiliary cancer is challenging. Due to the non-specific early symptoms or even asymptomatic manifestations, most patients are diagnosed at an advanced stage. Late diagnosis will at least lead to poor prognosis of BTC patients, and the 5-year overall survival rate is below 20%. Patients with biliary stricture and jaundice may have cholangiocarcinoma, gallbladder cancer or pancreatic cancer, and it is very difficult to distinguish malignant tumors from benign strictures (iatrogenic bile duct injury, primary sclerosing cholangitis (PSC), and choledocholithiasis). Conventionally, pancreaticobiliary cancer is diagnosed by a combination of various methods, including clinical examination, imaging, endoscopic procedures, pathological evaluation, and biochemical tests (such as CA19-9). However, these methods have some limitations. For example, CA19-9 is not applicable to patients negative for Lewis antigen (accounting for 7% of the general population), and the sensitivity and specificity of the above methods are not satisfactory. It is reported that about 15-24% of patients undergoing surgery for malignant biliary stricture are finally diagnosed as benign.

[0005] Therefore, there is an urgent need to develop a better detection method for diagnosing pancreaticobiliary cancer with high sensitivity, specificity, and safety. Summary of the Invention

[0006] Patients with pancreaticobiliary cancer usually have a poor clinical prognosis, and the 5-year overall survival rate is less than 20%. This is mainly related to late diagnosis. In addition, accurately differentiating malignant cancer from benign diseases before surgery can avoid unnecessary trauma. Therefore, there is an urgent need to develop a detection method for diagnosing malignant pancreaticobiliary cancer with high sensitivity, strong specificity, and high safety.

[0007] To achieve the above technical objectives, the present invention provides the following technical solutions:

[0008] In a first aspect, the present invention provides the use of a reagent for detecting methylation markers in a product for diagnosing pancreaticobiliary cancer, wherein the methylation markers include one or more of SOX17, 3-OST-2, NXPH1, SEPT9, and TERT.

[0009] Preferably, the methylation markers consist of SOX17, 3-OST-2, NXPH1, SEPT9, and TERT.

[0010] Preferably, the methylation markers may further include any one or more of EBF3, RASSF1, APC, EYA4, RUNX3, BNIP3, FHIT, SALL3, CCND2, FOXE1, CD1D, GSTP1, SFRP1, CDH1, hMLH1, SLIT2, CDH13, KCNK12, SLIT3, CDKN2A, MGMT, CDKN2B, NDRG4, CDO1, NPTX2, TFPI2, CLEC11, TIMP3, CNRIP1, PENK, TMEFF2 (HPP1), DAPK1, PRKCB, VIM, DCLK1, PTCHD2, ZSCAN18, DLC1, RARβ2 (RARB).

[0011] Preferably, the detection of the methylation markers refers to the detection of their methylation levels.

[0012] Preferably, the methylation markers can be detected by methods well-known in the art, specifically, for example: whole genome bisulfite sequencing (WGBS), pyrosequencing, bisulfite sequencing, methylation-specific polymerase chain reaction (methylation-specific PCR, MS-PCR), bisulfite-specific polymerase chain reaction, methylation-sensitive restriction enzyme-PCR / Southern method, combined bisulfite restriction analysis (COBRA), digital polymerase chain reaction, restriction landmark genome scanning, CpG island microarray, single nucleotide primer extension SNUPE, methylation profiling, one or more of methylation chips.

[0013] The term "pancreatobiliary cancer" as used in the present invention may also be referred to as pancreatobiliary tract cancer (Pancreatobiliary tractcance), which includes biliary tract cancer (bile tract cancer, BTC, which can also be called cholangiocarcinoma) and pancreatic cancer (pancreaticcancer); the cholangiocarcinoma includes cholangiocarcinoma (CCA), gallbladder cancer (gallbladdercancer, GBC) and ampullary cancer; the pancreatic cancer includes pancreatic head cancer, pancreatic tail cancer, and diffuse cancer.

[0014] Preferably, the non-cancer patients in the present invention are subjects who have not been diagnosed with cancer within at least 12 months. Optionally, the non-cancer patients may have the following symptoms of non-malignant tumors: gallstones, biliary obstruction, biliary stricture, pancreatic mass, pancreatic cyst, pancreatitis.

[0015] Preferably, the detection is performed on a sample derived from the subject.

[0016] Samples that can be selected in the art include bile, cells, tissues, peripheral blood, blood, serum, plasma, urine, saliva, tears, etc.; as used in the specific embodiments of the present invention, preferred samples include bile or cells and tissues taken from the biliary tract. More specifically, the cell tissues taken from the biliary tract include samples obtained by endoscopic retrograde cholangiopancreatography (ERCP) biopsy / brushing.

[0017] More preferably, the sample needs to be processed, and the processing includes the step of DNA extraction.

[0018] Preferably, the DNA extraction reagents may include any one or more of lysis buffer, binding buffer, wash buffer, and elution buffer. Lysis buffer usually consists of a protein denaturant, a detergent, a pH buffer, and a nuclease inhibitor. Binding buffer usually consists of a protein denaturant and a pH buffer. The protein denaturant is selected from one or more of guanidine isothiocyanate, guanidine hydrochloride, and urea; the detergent is selected from one or more of Tween20, IGEPAL CA-630, Triton X-100, NP-40, and SDS; the pH buffer is selected from one or more of Tris, boric acid, phosphate, MES, and HEPES; the nuclease inhibitor is selected from one or more of EDTA, EGTA, and DEPC.

[0019] Preferably, the processing may further include steps such as purification and quality inspection.

[0020] Preferably, the subject includes a suspected pancreatobiliary cancer patient.

[0021] As used herein, the term "subject" refers to any animal (e.g., a mammal), including but not limited to humans, non-human primates, rodents, etc., that will be the recipient of a particular treatment. Generally, the terms "subject" and "patient" are used interchangeably herein when referring to human subjects.

[0022] Preferably, the subject is a human.

[0023] Preferably, when jointly diagnosing using SOX17, 3-OST-2, NXPH1, SEPT9, and TERT, the optimal threshold is 0.422. It should be noted that for the threshold concentrations given above and the details below, the optimal threshold may depend on the specific measurement technique. The optimal threshold given herein relates to the measured values using the Mutation Capsule technology. If different methods are used, an analogous conversion may be required, which is within the scope of the skills of those skilled in the art.

[0024] On the other hand, the present invention provides the use of the reagent for detecting methylation markers and the reagent for detecting other markers jointly in a product for diagnosing pancreaticobiliary cancer, wherein the other markers include at least one of the following gene mutations: AKT1, KRAS, APC, NRAS, ARID1A, PIK3CA, AXIN1, PPP2R1A, BAP1, PTEN, BRAF, SMAD4, CDKN2A, TERT, TP53, EGFR, FBXW7, FGFR2, HRAS, IDH1, IDH2.

[0025] The term "marker" or "biomarker" refers to a genetic indicator that can mark changes or possible changes in the system, organ, tissue, cell, and subcellular structure or function, and can be used for disease diagnosis, disease staging, or evaluating the safety and effectiveness of new drugs and new therapies in the target population.

[0026] More specifically, the diagnosis of pancreaticobiliary cancer refers to differentiating patients with malignant tumors (pancreaticobiliary cancer) from patients without malignant tumors, and the patients without malignant tumors may have benign diseases such as gallstones, biliary obstruction, biliary stricture, pancreatic mass, pancreatic cyst, pancreatitis, etc.

[0027] In an alternative embodiment, the reagent for detecting gene mutations includes the reagent used in any of the following methods: TaqMan probe method, sequencing method, chip method, matrix-assisted laser desorption / ionization time-of-flight mass spectrometry (MALDI-TOFMS) detection, polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP), polymerase chain reaction-single strand conformation polymorphism (PCR-SSCP), allele-specific polymerase chain reaction (AS-PCR), SNaPshot method, SNPlex genotyping system, SNPStream analysis system, Sequenom genotyping system, denaturing high performance liquid chromatography (DHPLC), denaturing gradient gel electrophoresis (DGGE).

[0028] On the other hand, the present invention also provides a diagnostic system for diagnosing pancreaticobiliary cancer, and the system reports a computing device that obtains a diagnostic conclusion based on the detection results of methylation markers in a subject sample.

[0029] Preferably, the system includes:

[0030] (1) A sample collection and processing device for performing the following steps: collecting a sample from a subject and processing the sample;

[0031] (2) A methylation detection device;

[0032] (3) A calculation device for obtaining a diagnostic conclusion based on the detection results of methylation markers in the subject's sample.

[0033] Preferably, the sample includes bile, exfoliated cells in the biliary tract, and tissue samples.

[0034] Preferably, the processing includes steps such as purification, quality inspection, and DNA extraction.

[0035] On the other hand, the present invention also provides a method for diagnosing pancreaticobiliary cancer, and the method determines whether the subject has a malignant disease (pancreaticobiliary cancer) based on the detection results of methylation markers in the subject's sample.

[0036] The implementation of the method and / or system of the embodiments of the present invention may include performing or completing the selected tasks manually, automatically, or in combination thereof. Moreover, according to the actual instruments and devices of the embodiments of the method and / or system of the present invention, multiple selected tasks may be implemented by hardware, by software, or by firmware or by using an operating system in combination thereof.

[0037] The present invention has the following beneficial effects:

[0038] The technical solution provided by the present invention can obtain bile as a sample for detection by a non-invasive sampling method, which has the characteristics of high specificity and high sensitivity; it can accurately diagnose the disease at an earlier stage of the disease, enable early targeted treatment of the patient, increase the possibility of cure, and extend the survival period; at the same time, for non-cancer patients, unnecessary surgical trauma is avoided. Description of the Drawings

[0039] Figure 1 is the flowchart of the subject inclusion criteria and research involved in the present invention.

[0040] Figure 2 is the statistical chart of the basic information of the subjects involved in the present invention.

[0041] Figure 3 is the verification of the diagnostic efficacy of each diagnostic model in different datasets.

[0042] Figure 4 is the statistical analysis of the detection results of the subjects in the training cohort and the validation cohort.

[0043] Figure 5 It is the verification of the diagnostic efficacy of methylation markers.

[0044] Figure 6 It is the comparison result of the diagnostic efficacy of each diagnostic model with that of CA19-9 in different datasets.

[0045] Figure 7 It is the result statistics of the consistency of gene mutation detection results in brush biopsy samples and biopsy samples. Specific implementation manners

[0046] The present invention will be further described below in conjunction with embodiments. The following description is only for the preferred embodiments of the present invention, and does not limit the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make equivalent embodiments with equivalent changes. Any simple modification or equivalent change made to the following embodiments based on the technical essence of the present invention without departing from the content of the present invention's solution falls within the protection scope of the present invention.

[0047] Screening, Identification and Verification of Example 1, Mutant Genes and Methylated Genes

[0048] 1. Study Population and Experimental Design

[0049] The research population was selected from patients with pancreatobiliary diseases admitted to the following five hospitals from November 2018 to October 2020, with a total of 338 cases. The five hospitals are: Cancer Hospital, Chinese Academy of Medical Sciences; Dazhou Central Hospital, Sichuan Province; Beijing Chaoyang Hospital, Capital Medical University; Dongfang Hospital, Beijing University of Chinese Medicine; and Gucheng County People's Hospital, Xiangyang City, Hebei Province. 79 cases were excluded due to various reasons such as non-biliary cancer or insufficient bile DNA, and finally 259 cases were selected for molecular detection.

[0050] Among these 259 patients, 209 patients diagnosed as malignant or benign were selected as the training group (n = 104) and the validation group (n = 105), including 116 cases of malignant tumors and 93 cases of benign diseases. Malignant tumors were confirmed as biliary malignancies (pancreaticobiliary cancer) after diagnosis by ERCP (endoscopic retrograde cholangiopancreatography) biopsy / brushings (ERCP-obtained biopsies / brushings). Benign disease patients included 78 patients with cholelithiasis. No malignant lesions were found in the surgically resected specimens of these 78 patients, but chronic cholangitis was found. Among them, 15 patients were followed up for more than 12 months, and no malignant lesions were found after ERCP for cholelithiasis removal.

[0051] In addition, considering that some patients with bile tract cancer (BTC) have low sensitivity of ERCP pathology, 50 patients with negative or suspicious diagnostic results by ERCP biopsy / brushing were classified into an independent test cohort. In the test cohort, 40 patients with malignancies were confirmed during follow-up by pathological evaluation (biopsy / brushing obtained from surgically resected specimens, percutaneous needles or ERCP, n = 21), radiological imaging (n = 2), or clinical criteria (n = 17), while 10 patients with benign diseases were confirmed by surgical pathology (n = 4), radiological imaging (n = 1), or no malignancies were found after at least 12 months of follow-up (n = 5).

[0052] Patient selection and study design were as Figure 1 shown. This study was approved by the ethics review boards of the above five hospitals (ID: NCC2018JJJ-001).

[0053] 2. Sample Preparation

[0054] Bile samples were obtained from all patients. Among them, 181 patients had bile samples obtained by ERCP before treatment, and 78 patients with cholelithiasis had bile samples obtained during cholecystectomy surgery.

[0055] The obtained bile samples were centrifuged at 12,000 revolutions per minute for 10 minutes to separate the supernatant and particles. DNA was extracted from the bile samples using the TIANAMP Genomic DNA Kit (Tiangen Biotech, Beijing, China). RQ-PCR detection of the human GAPDH gene was performed using Taqman probes to determine the DNA quality.

[0056] In addition to bile samples, paired biopsy or brushing tissues obtained during ERCP from 34 and 9 patients were subjected to NGS (Next-generation sequencing technology) detection and sequencing, and genomic DNA was extracted using the QIAamp DNA Mini Kit (Qiagen, USA).

[0057] Further, the CA19-9 of the sera of most patients was detected using the Roche E601 system (Roche Diagnostics, Switzerland) by electrochemiluminescence technology.

[0058] 3. Analysis of Gene Mutations and Gene Methylation Using BileScreen

[0059] Table 1. Mutated genes and genes with methylation modifications found by the sequencing of the present invention

[0060]

[0061] Take 400 ng of DNA, fragment it by sonication, and then perform end repair. Then, digest the DNA fragments with the methylation-sensitive restriction endonuclease HHAⅠ (R0139S, New England Biolabs, MA, USA), and perform targeted gene sequencing based on amplicon-targeted capture technology through the Mutation Capsule technology. The specific technical protocol can be found in the reference Qu, C. et al. Detection of early-stage hepatocellular carcinoma in asymptomatic HBsAg-seropositive individuals by liquid biopsy. Proceedings of the National Academy of Sciences of the United States of America. 116, 6308-6312 (2019).

[0062] Briefly, process the DNA fragments digested with the restriction endonuclease HHAⅠ using the KAPA Hyper Prep kit (Roche, Switzerland). Through a series of steps, including end repair, adenylation A tailing, ligation of custom adapters, and three rounds of PCR amplification (using common sequence primers in the first round and primers containing target specificity and common sequences in the last two rounds), a sequencing library is obtained. Sequence 23 mutant genes and 44 genes with methylation modifications (Table 1).

[0063] 4. Data Processing and Mutation / Methylation Detection

[0064] Gene sequencing was performed using the Digital (UID) high-throughput sequencing platform. Briefly, a unique UID tag was added to each fragment in the sample before PCR amplification, and then the library was amplified. After sequencing, by aligning the sequences of the fragments, the repeated fragments with the same UID label were merged, while the natural repeats with different UID labels were retained, and the end coordinates of the reads were calculated based on the start coordinates and cigar information; according to the start and end coordinates of the reads, the reference sequences corresponding to the reads were truncated from the reference genome; the reads were aligned again with the hg19 genome to obtain the start and end positions of the mutations. An effective UID (EUID, Effective UID) family is defined as a UID family (Unique Identifier, UID) that contains at least two reads and at least 80% of the read types are the same. The frequency of each mutation is calculated by dividing the number of alternative EUID families by the sum of the alternative and reference families. We further manually inspected the mutations (in IGV) and used VEP (Ensembl Variant Effect Predictor) to annotate the candidate variant genes.

[0065] The detected mutations had at least four EUID families. For oncogenes with hot-spot mutations including KRAS (G12, G13, Q61, and A146), the limit of detection (LOD, limitation of detection) was set at 0.5%. For other mutant genes, including common tumor suppressor genes without hot spots, such as TP53, Smad4, and rare mutations, in order to reduce the occurrence of false positives, the LOD was set at 1%. Since there were no matched white blood cells to exclude germline mutations, the mutations detected in the samples were screened using germline and somatic mutation databases to determine the highest likelihood of germline mutations. Mutations with a frequency of ≥0.1% found in the germline mutation databases (1000AF, ESP6500 AA / EA, Exac AF) were first excluded as germline mutations. The passed mutations were further screened. For those genes with a higher frequency (≥40%), if they appeared in the COSMIC database with fewer than 10 samples, then they were likely to be germline mutations and were thus further excluded.

[0066] In terms of methylation analysis, clusters with an HHA I restriction site at the end were unmethylated sequences, and molecules with at least one HHA I restriction site and a non-restriction site at the end were methylated sequences. The methylation ratio of each base was the ratio of the number of methylated molecules to the sum of the number of methylated and non-methylated molecules.

[0067] 5. Construction of BileScreen Diagnostic Model

[0068] Malignant tumors occur when AKT1, KRAS, APC, NRAS, ARID1A, PIK3CA, AXIN1, PPP2R1A, BAP1, PTEN, BRAF, SMAD4, CDKN2A, TERT, TP53, EGFR, FBXW7, FGFR2, HRAS, IDH1, or IDH2 are mutated (as shown in Table 2 below).

[0069] Table 2. Detection Results of Mutations

[0070]

[0071]

[0072]

[0073]

[0074]

[0075]

[0076]

[0077]

[0078]

[0079]

[0080]

[0081]

[0082]

[0083]

[0084] Among the 44 methylated genes, using the training cohort (Training set), 5 methylated genes, SOX17, 3-OST-2, NXPH1, SEPT9, and TERT, were selected by stepwise penalized logistic regression for constructing a diagnostic model. The above 5 methylated gene markers were used for penalized logistic regression on the training cohort, and leave-one-out cross-validation was adopted. The performance of the model was evaluated by the area under the receiver operating characteristic curve (ROC curve), sensitivity, and specificity. The cut-off value of methylation was determined according to the Youden index of ROC analysis.

[0085] In the BileScreen model, when mutation and methylation are integrated, a positive result is obtained if either one is positive. Next, the performance of the BileScreen model was further evaluated in separate validation and test cohorts.

[0086] In addition, since in the training cohort and the validation set, most of the selected benign cases were young women with gallstones, there was a certain bias in age and gender between malignant and benign patients. To exclude the influence of this artificial sample selection on the diagnostic prediction results, 52 malignant patients and 52 benign patients matched in age and gender were assigned to the training cohort. Therefore, the age and gender distributions in the validation cohort were uneven ( Figure 2 ). However, all mutations and methylations had no obvious correlation with age and gender (correlation coefficient < 0.5, or Wilcoxon test P > 0.05).

[0087] 6. Statistical Analysis

[0088] ROC analysis (pROC package) and Wilcoxon test were used to evaluate the role of individual mutant genes or methylated modified genes in predicting disease status. The penalized Logistic regression method (glmnet package) was used to screen gene markers for the diagnostic model. In the training cohort, the ROC curve took the raw scores as input and used the Youden index alone to determine the optimal cut-off point for methylation. In addition, ROC analysis was used to compare the performance of different methods, with the "0 or 1" values determined by the corresponding cut-off points as input. Sensitivity and specificity were calculated using standard 2×2 contingency tables. All R package-related analyses were based on R software (V.3.6.3).

[0089] 7. Results

[0090] 1) Establishment of BileScreen Model Based on Gene Mutations and Methylation Modifications

[0091] Sequencing and data analysis were performed on 104 patients in the training cohort, including 52 patients with malignant tumors pathologically confirmed by ERCP and 52 patients with benign diseases. Among these 52 patients, some had their tissue sites determined not to have canceration by surgical pathology, and the other part were patients with choledocholithiasis and had been followed up for at least 12 months and were determined to have benign diseases. The clinical characteristics are summarized as Figure 2 .

[0092] Using the Mutation Capsule technology, the DNA of bile samples was analyzed, and 23 mutated genes and 44 methylated genes were detected. The most common mutated genes in cancer were TP53 (50%) and KRAS (46%). CTNNB1 and GNAS mutations were detected in both cancer and benign disease patients, so the genes were not significantly correlated with the malignant state, and CTNNB1 and GNAS mutations were excluded from the BileScreen model.

[0093] Table 3. Detection accuracy of each dataset

[0094]

[0095]

[0096] Using gene mutations to distinguish patients with cholangiocarcinoma from non-cancer (benign disease) subjects:

[0097] Mutations in AKT1, KRAS, APC, NRAS, ARID1A, PIK3CA, AXIN1, PPP2R1A, BAP1, PTEN, BRAF, SMAD4, CDKN2A, TERT, TP53, EGFR, FBXW7, FGFR2, HRAS, IDH1, IDH2, where at least one mutation detected was considered positive; the sensitivity of distinguishing patients with cholangiocarcinoma from non-cancer subjects by mutations alone was 81%, the specificity was 100%, and the AUC was 0.90 (Table 3, Figure 3 A).

[0098] Using methylation markers to distinguish patients with cholangiocarcinoma from non-cancer (benign disease) subjects:

[0099] For methylation markers, through the stepwise penalized Logistic regression method, 5 markers, SOX17, 3-OST-2, NXPH1, SEPT9, and TERT, were selected to construct a diagnostic model ( Figure 3 , Table 4). By the leave-one-out method, only methylation markers could well identify patients with cholangiocarcinoma from non-cancer cases, with a sensitivity of 88%, a specificity of 98%, and an AUC of 0.93 (Table 3, Figure 3 ). The cut-off value of the methylation score was 0.422, which produced the maximum Youden index ( Figure 5 ).

[0100] Table 4. Construction of a diagnostic model based on 5 methylation markers

[0101]

[0102] Finally, when the gene mutation and methylation markers are combined, namely BileScreen:

[0103] Positive is defined as either of them being positive, and its performance is further improved, with a sensitivity of 94%, a specificity of 98%, and an AUC of 0.96 (Table 3, Figure 3 ). BileScreen was validated in another 105 cases (validation cohort), among which 64 were malignant cases and 41 were benign cases ( Figure 4 ). BileScreen accurately predicted the disease status of 59 malignant cases and 40 non-cancer cases. BileScreen showed a sensitivity of 92% and a specificity of 98% in the validation cohort, with an AUC of 0.95 (Table 3, Figure 3 ). If only gene mutation is adopted, its sensitivity and specificity are 78% and 100% respectively. The sensitivity and specificity of only methylation marker analysis are 81% and 98% respectively (Table 3).

[0104] 2. BileScreen Model for Detecting Suspicious Malignant Tumors

[0105] We further validated BileScreen in 50 patients with unclear ERCP (test cohort) results because the ERCP diagnosis results of these 50 patients were "suspicious malignancy" or "cancer cannot be excluded". These cases were followed up for at least 12 months, and 40 out of 50 were found to be malignant. The remaining 10 cases were not found to have cancer during the follow-up period and were diagnosed as benign. Among the 40 malignant lesions, 36 were positive, and among the 10 benign lesions, 2 were positive, with a sensitivity of 90% and a specificity of 80% (Figure 3). For those patients who could not be diagnosed by ERCP, the BileScreen results were significantly correlated with the clinical results (P < 0.001, continuity-corrected chi-square test).

[0106] Only mutation or methylation can distinguish cancer and benign patients, with sensitivities of 75% and 80% respectively, specificities of 90% and 80% respectively, and AUCs of 0.83 and 0.8 respectively (Table 3, Figure 3 ).

[0107] 3. Comparison of Detection Results of CA19-9 and BileScreen Model in Each Group

[0108] In the training cohort and validation cohort, we screened out 85 and 74 patients respectively, whose serum CA19-9 data were available ( Figure 2, Table 5), a direct comparison of serum CA19-9 and BileScreen was performed on these patients. The AUCs of serum CA19-9 for differentiating benign and malignant cases in the two cohorts were 0.78 and 0.81 ( Figure 6 ), respectively. Using ≥27 U / mL as the cut-off value, the sensitivities of serum CA19-9 were 88% and 91%, and the specificities were 67% and 70%, respectively (Table 5).

[0109] Table 5. Validation results of each model and CA19-9 in different datasets

[0110]

[0111] In contrast, the sensitivities of BileScreen were 93% and 94%, and the specificities were 98% and 96%, respectively. In the entire training and validation cohorts, the sensitivities and specificities of CA19-9 were 90% and 68%, respectively, both lower than 93% and 97% of BileScreen. Therefore, BileScreen is superior to serum CA19-9 in detecting pancreatobiliary cancer, especially in terms of detection specificity.

[0112] In addition, the serum CA19-9 results of 38 patients in the test cohort were as Figure 6 shown in A. The specificity of serum CA19-9 was extremely low (14%), the AUC was 0.51, and the sensitivity was 84% (Table 4, Figure 6 ). Compared with the training and validation cohorts, the accuracy of CA19-9 in predicting patients with suspected malignancy diagnosed by ERCP was lower. In contrast, the sensitivities and specificities of BileScreen in this group were 87% and 86%, respectively.

[0113] 4. Comparison of Gene Mutation Results between Bile and ERCP Biopsy / Brush Specimens

[0114] In the test cohort, biopsy samples were obtained from 34 patients by ERCP and brush samples were obtained from 9 patients. We analyzed the gene and tissue samples for a head-to-head comparison study. Among 43 cases, 70 mutations were present in both sample types, 5 mutations were detected only in bile (brush samples), and 9 were detected only in tissue (biopsy samples) ( Figure 7 ). Therefore, 93% (70 / 75) of the mutations in bile were also detectable in tissue, and some additional mutations were found only in bile. 89% (70 / 79) of the tissue-derived mutations were detectable in bile. In addition, mutations were detected in at least one type of sample in 36 of the 43 cases, and at least one common mutation was detected between the two sample types in 34 cases (94%). Therefore, the concordance rate of the mutation status detected by bile and that detected by tissue in patients was 95% (41 / 43).

Claims

1. Use of a reagent for detecting methylation markers in the preparation of a product for diagnosing pancreatobiliary cancer, wherein the methylation markers are a combination of SOX17, 3-OST-2, NXPH1, SEPT9, and TERT.

2. The use according to claim 1, wherein the pancreatobiliary cancer includes cholangiocarcinoma and pancreatic cancer.

3. The use according to claim 2, wherein the cholangiocarcinoma includes cholangiocarcinoma, gallbladder cancer, and ampullary cancer.

4. The use according to claim 2, wherein the pancreatic cancer includes pancreatic head cancer, pancreatic tail cancer, and diffuse cancer.

5. The detection reagent for the methylation markers according to claim 1 includes reagents used in any one or more of the following methods: pyrosequencing, bisulfite sequencing, methylation-specific polymerase chain reaction, bisulfite-specific polymerase chain reaction, methylation-sensitive restriction enzyme-PCR, restriction enzyme method combined with bisulfite, digital polymerase chain reaction, restriction landmark genomic scanning, CpG island microarray, single nucleotide primer extension SNUPE, methylation profiling, methylation chip.

6. In the use according to claim 1, when jointly diagnosing using SOX17, 3-OST-2, NXPH1, SEPT9, and TERT, the optimal threshold is 0.

422.

7. A system for diagnosing pancreatobiliary cancer, the system being a computing device that obtains a diagnostic conclusion based on the detection result of the methylation markers according to claim 1 in a sample from a subject, and the system includes: (1) A sample collection and processing device for completing the following steps: collecting a sample from a subject and processing the sample; (2) A methylation detection device; (3) A computing device for obtaining a diagnostic conclusion based on the detection result of the methylation markers in the sample from a subject.

8. The system according to claim 7, wherein the pancreatobiliary cancer includes cholangiocarcinoma and pancreatic cancer.

9. The system according to claim 8, wherein the cholangiocarcinoma includes cholangiocarcinoma, gallbladder cancer, and ampullary cancer.

10. The system according to claim 8, wherein the pancreatic cancer includes pancreatic head cancer, pancreatic tail cancer, and diffuse cancer.

Images