JP2019213473A - Discrimination method of prognostic risk after pancreatic tumor extraction operation - Google Patents

Abstract

To provide prognostic discrimination methods of pancreatic tumor effective for early diagnosis of pancreatic tumor or definition of malignancy of pancreatic tumor.SOLUTION: Provided is a prognostic discrimination method of pancreatic tumor comprising: a first step of measuring the methylation status of a 5'-side untranslated region or a region including the 5'-side untranslated region and a translated region of a cancer gene and/or a cancer-suppressing gene in a specimen sample derived from a subject; and, in order to discriminate the prognosis of the subject's pancreatic tumor, a second step of substituting a measured value of the methylation status of the 5'-side untranslated region or the region including the 5'-side untranslated region and the translated region of the cancer gene and/or the cancer-suppressing gene obtained in the first step, for the discriminant of a machine learning model using, as a teacher, measurement values obtained by measuring the methylation status of the 5'-side untranslated region or the region including the 5'-side untranslated region and the translated region of the cancer gene and/or the cancer-suppressing gene in a plurality of specimen samples that are known to be a specimen sample derived from tumor tissue or a specimen sample derived from non-tumor tissue.SELECTED DRAWING: None

Description

  The present invention relates to a method for predicting the degree of malignancy or the risk of recurrence in the prognosis after extirpation of a pancreatic tumor, for example.

  Cancer deaths in Japan exceed 370,000 a year and are the leading cause of death. In addition, the number of cancer cases is increasing year by year, and is forecast to exceed 1 million in 2017. Among them, the number of pancreatic cancer cases is 39,800, about 4% of the total, which is the seventh most rare cancer. In addition, the number of deaths from pancreatic cancer in the same year is almost the same, and even when the age adjustment rate is added, the number of deaths is not decreasing for both men and women. Compared with other cancers, the 5-year relative survival rate is extremely low at 7.7% for pancreatic cancer, and it can be said that pancreatic cancer is a representative of refractory cancer (2017, calculated by National Cancer Center).

  Because pancreatic cancer has few symptoms, its discovery is delayed, and only about 20 to 30% of patients can undergo surgery. Even if resection surgery is possible, the 5-year survival rate is extremely low at about 20%, the lowest among cancers of the digestive system. However, it has recently been reported that the 5-year survival rate when a tumor is found less than 1 cm in diameter is as high as 80.4%.

  In order to improve the prognosis of pancreatic cancer, it is most important to aim for early detection and early treatment at the stage where surgery is possible. For pancreatic cancer, no clear increase was observed in tumor marker tests (CA19-9, CEA), and the rate of increase was less than half. In addition, the rate of extraction of tumors with a diameter of 1 cm or less by ultrasonography or CT examination is far from high, and is 17-70% by ultrasonography and 35-75% by contrast-enhanced CT examination.

  Therefore, using clinical specimens such as minimally invasive body fluid specimens and post-operative excised tissue specimens, the `` early diagnosis method '' or `` method for defining the degree of malignancy '' of pancreatic cancer and intraductal papillary mucinous neoplasm (IPMN) The establishment is eagerly awaited.

In recent years, abnormal expression of the mucin family of glycoproteins has been confirmed in various cancers including pancreatic cancer, and is involved in cancer malignancy in each process of cancer progression such as proliferation, invasive metastasis, resistance to anticancer drugs, and evasion of the immune mechanism. (Non-Patent Document 1). On the other hand, the present inventors have been conducting research on mucin from an early stage, and have clarified the relationship between mucin expression and malignancy in various tissues and the mechanism of the expression as follows, as follows:
(1) A detailed analysis was performed on the biological grade of pancreatic tumors and the expression of a series of mucin antigens.Pre-malignant lesions (PanIN), pancreatic cancer (PDAC), MUC1, MUC2 in intraductal papillary mucinous neoplasms (IPMN), The relationship between MUC4 mucin expression and malignancy has been clarified for the first time in the world. In addition, a novel antibody against the MUC1 cytoplasmic domain has been developed (Patent Document 1);
(2) Each mucin gene was found to control the expression of the target gene by methylation of each promoter region (Non-Patent Document 2);
(3) Development and report of a novel high-sensitivity methylation analysis method with a resolution of 0.1% exceeding the conventional DNA methylation detection limit (Patent Document 2 and Non-Patent Document 3);
(4) In the analysis by a novel methylation analysis method using actual pancreatic juice samples and tissue samples, the correlation between the malignancy and prognosis and DNA methylation was clarified (Non-patent Documents 4 and 5), and the type of pancreatic tumor This is reported as a diagnostic method (Patent Document 3).

Patent No. 5660486 Patent No.5765586 Patent No.5866739

Donald W. Kufe, Nature Reviews Cancer, 2009, 9, pp. 874-885 Norishige Yamada et al., Cancer Research, 2008, 68 (8), pp. 2708-2716 Seiya Yokoyama et al., BMC Cancer, 2012, 12:67 Seiya Yokoyama et al., PLoS One, 2014, 9 (4): e93760 Seiya Yokoyama et al., Oncotarget, 2016, 7 (27), pp. 42553-42565

  The present invention has been made in view of the above circumstances, and has as its object to provide a method for determining the prognosis of a pancreatic tumor, which is effective for early diagnosis of pancreatic tumor or definition of malignancy of pancreatic tumor.

  As a result of intensive studies to solve the above problems, the inventors have found that the methylation status of oncogenes and / or tumor suppressor genes in a specimen sample can be analyzed and that the prognosis of pancreatic tumor can be predicted by machine learning, and the present invention has been completed. I came to.

That is, the present invention includes the following.
(1) a first step of measuring the methylation status of a 5′-side untranslated region or a region containing a 5′-side untranslated region and a translated region of a cancer gene and / or a tumor suppressor gene in a specimen sample derived from a subject; In order to determine the prognosis of the subject's pancreatic tumor, the oncogene and / or cancer in a plurality of sample samples known to be a sample sample derived from tumor tissue or a sample sample derived from non-tumor tissue The discriminant of the machine learning model using the measured value obtained by measuring the methylation status of the region including the 5′-side untranslated region or the 5′-side untranslated region and the translated region of the suppressor gene as a teacher, A second step of substituting the measured value of the methylation status of the 5'-side untranslated region or the region containing the 5'-side untranslated region and the translated region of the oncogene and / or tumor suppressor gene obtained in one step. Methods for determining prognosis of pancreatic tumors.
(2) The method according to (1), wherein the machine learning model is selected from the group consisting of a support vector machine model, a neural network model, a random forest model, and a deep learning model.
(3) The method according to (1) or (2), wherein the tumor tissue and the non-tumor tissue are pancreatic tissues.
(4) (1) to (1), wherein the oncogene and / or the tumor suppressor gene is a gene set including genes encoding one or more mucin core proteins selected from the group consisting of MUC1, MUC2, and MUC4 genes. The method according to any one of 3).
(5) The method according to any one of (1) to (3), wherein the oncogene and / or the tumor suppressor gene is a gene set including the MUC1, MUC2, and MUC4 genes.
(6) The method according to any one of (1) to (3), wherein the oncogene and / or the tumor suppressor gene is a gene set consisting of the MUC1, MUC2, and MUC4 genes.
(7) The (4) to (6) of (6), wherein the 5′-side untranslated region of the MUC1 gene or the region containing the 5′-side untranslated region and the translated region is a region consisting of the nucleotide sequence shown in SEQ ID NO: 1. A method according to any one of the preceding claims.
(8) (4) to (7), wherein the 5′-side untranslated region of the MUC2 gene or the region including the 5′-side untranslated region and the translated region is a region comprising the base sequence shown in SEQ ID NO: 2. A method according to any one of the preceding claims.
(9) The region of (4) to (8), wherein the 5′-side untranslated region of the MUC4 gene or the region including the 5′-side untranslated region and the translated region is a region comprising the base sequence shown in SEQ ID NO: 3. A method according to any one of the preceding claims.
(10) The method according to any one of (1) to (9), wherein the pancreatic tumor is pancreatic cancer or intraductal papillary mucinous tumor.

  According to the present invention, the prognosis of a pancreatic tumor can be diagnosed and predicted at an early stage.

The sequence of the promoter region that regulates the expression of the human mucin core protein 1 (MUC1) gene by methylation and the sequence of the adjacent translation region (SEQ ID NO: 4) are shown. The sequence of the promoter region that regulates the expression of the human MUC1 gene by methylation and the sequence of the adjacent translation region (SEQ ID NO: 5; the DNA sequence after Bisulfite treatment, and uracil converted from unmethylated cytosine) (Displayed in thymine). The sequence of the promoter region that regulates the expression of the human mucin core protein 2 (MUC2) gene by methylation and the sequence of the adjacent translation region (SEQ ID NO: 6) are shown. The sequence of the promoter region that regulates the expression of the human MUC2 gene by methylation and the sequence of the adjacent translation region (SEQ ID NO: 7; however, this is a DNA sequence after bisulfite treatment and uracil converted from unmethylated cytosine) (Displayed in thymine). The sequence of the promoter region that regulates the expression of the human mucin core protein 4 (MUC4) gene by methylation and the sequence of the adjacent translation region (SEQ ID NO: 8) are shown. The sequence of the promoter region that regulates the expression of the human MUC4 gene by methylation and the sequence of the adjacent translation region (SEQ ID NO: 9; however, the DNA sequence after Bisulfite treatment, and uracil converted from unmethylated cytosine) (Displayed in thymine). FIG. 3 is a diagram illustrating support vector machine (SVM) model construction in the first embodiment. FIG. 14 is a diagram illustrating a result of risk discrimination by an SVM model in the second embodiment. FIG. 13 is a diagram illustrating a result of risk discrimination of a training data (50 examples) by a neural network (NNET) model in a third embodiment.

Hereinafter, the present invention will be described in detail.
The method for determining the prognosis of a pancreatic tumor according to the present invention (hereinafter, referred to as `` the method '') is a 5′-side untranslated region or 5′-side untranslated region of a cancer gene and / or a tumor suppressor gene in a specimen sample derived from a subject. A first step of measuring the methylation status of the region including the region and the translation region, and a sample sample derived from a tumor tissue or a sample sample derived from a non-tumor tissue in order to determine the prognosis of pancreatic tumor in the subject By measuring the methylation status of a region containing a 5′-side untranslated region or a 5′-side untranslated region and a translated region of a cancer gene and / or a tumor suppressor gene in a plurality of sample samples known to be The discriminant of the machine learning model using the obtained measured value as a teacher, the 5 ′ untranslated region or the 5 ′ untranslated region and the translated region of the oncogene and / or tumor suppressor gene obtained in the first step. And substituting the measured value of the methylation status of the region including In other words, the present method can be an evaluation method for determining the prognosis of a pancreatic tumor, or a method of collecting information for determining the prognosis of a pancreatic tumor.

Pancreatic tumors include, for example, pancreatic cancer and intraductal papillary mucinous neoplasm (IPMN).
The specimen sample derived from the subject includes, for example, clinical specimens such as a tissue specimen after excision of a pancreatic tumor, an excess specimen for pathological diagnosis, an extra specimen for cytology, a blood specimen, and a body fluid specimen.

  Further, as a specimen sample known to be a specimen sample derived from a tumor tissue (preferably, a pancreatic tissue (that is, pancreatic tumor tissue)), for example, a skilled pathologist may be able to obtain a macroscopic specimen from a tissue specimen after tumor excision. Pancreatic tumor patient, a sample from which the tumor was excised in step 2, tumor tissue / cells collected by needle biopsy, and formalin-fixed paraffin-embedded (FFPE) tissue sliced and only the tumor was collected using laser microdissection Examples include bile and bodily fluid samples such as pancreatic juice and blood.

  Furthermore, as a specimen sample known to be a non-tumor tissue (preferably, a non-tumor pancreatic tissue), for example, a skilled pathologist may be able to obtain a macroscopic sample from a non-tumor site from a tissue sample after tumor excision. Excised sample, non-tumor tissue / cells collected by needle biopsy, formalin-fixed paraffin-embedded (FFPE) tissue sliced, and non-tumor sample collected using laser microdissection, from non-tumor patients Bile and body fluid samples such as pancreatic juice and blood.

  As oncogenes, for example, a gene encoding a mucin core protein (MUC) (also referred to as a mucin gene), a gene encoding a Gsα protein, a gene encoding a Rab family protein, encoding CD44 involved in cancer stem cells Genes, genes encoding ligands for the immune checkpoint protein PD-1, and combinations thereof. In particular, examples of the gene encoding MUC include the MUC1 gene, the MUC2 gene, the MUC4 gene, and a combination of two or all three of them.

  On the other hand, examples of the tumor suppressor gene include a combination of genes encoding transcription factors P53 and BRCA family, a gene encoding P16 which is cyclin-dependent kinase inhibition, and a gene encoding SMAD family protein. In addition, genes encoding cadherin and catenin involved in cell adhesion, and combinations thereof.

  The target sequence for the detection of the degree of methylation (target region) is an oncogene and / or a 5′-side untranslated region of a tumor suppressor gene, an untranslated region located on the 5′-side upstream of the oncogene and / or tumor suppressor gene ( UTR), particularly a region containing a promoter region of an oncogene and / or a tumor suppressor gene. Furthermore, a region containing the 5′-side untranslated region and the translated region of the oncogene and / or the tumor suppressor gene is a 5′-untranslated region such as a promoter region and a translated region adjacent to the 5′-untranslated region (coding region). ).

  One or more (one or more) CpG sites present in a promoter region of an oncogene and / or a tumor suppressor gene or a region containing the promoter region and the translation region may be a target sequence for which the degree of methylation is to be detected. it can. Here, the CpG site means a 5′-cytosine-guanine-3 ′ (5′-CG-3 ′) dinucleotide. When two or more CpG sites are used as the target sequence, each CpG site may be used individually, or a region containing two or more CpG sites may be used as the target sequence.

  FIGS. 1 to 3 show the sequences of the promoter region and the sequences of the adjacent translation regions that regulate the expression of human MUC1, MUC2, and MUC4 genes by methylation, respectively. In FIGS. 1 to 3, (TIS) indicates a transcription initiation site (transcription initiation site), CG surrounded by a square indicates a methylated site (CpG site), CG surrounded by a thick square is human MUC The methylated sites (CpG sites) involved in gene expression are shown. In FIG. 1B, FIG. 2B, and FIG. 3B, italic T indicates uracil converted from unmethylated cytosine by bisulfite treatment in thymine. The base sequence after the transcription start site (TIS) is a translation region (coding region).

  FIG. 1A shows the sequence of a promoter region that regulates the expression of the human MUC1 gene by methylation and the sequence of an adjacent translation region (SEQ ID NO: 4). FIG. 1B shows the DNA sequence of the sequence after Bisulfite treatment (SEQ ID NO: 5), and uracil converted from unmethylated cytosine is indicated by thymine. In FIG. 1, the region between Primer 1-3 and Primer 1-4 corresponding to the second primer set described in the following Reference Example (SEQ ID NO: 1) has a CpG site (or CpG island) 172-181. (Norishige Yamada, Yukari Nishida, Hideaki Tsutsumida, Tomofumi Hamada, Masamichi Goto, Michiyo Higashi, Mitsuharu Nomoto and Suguru Yonezawa (2008) MUC1 expression is regulated by DNA methylation and histone H3-K9 modification in cancer cells.Cancer Res., 68 (8): As in 2708-16, the numbers of CpG sites are numbered sequentially from the putative promoter upstream of the human MUC1 gene (2,753 bp upstream from the transcription start point of the gene).

  FIG. 2A shows the sequence of the promoter region that regulates the expression of the human MUC2 gene by methylation and the sequence of the adjacent translation region (SEQ ID NO: 6). FIG. 2B shows the DNA sequence of the sequence after Bisulfite treatment (SEQ ID NO: 7), and uracil converted from unmethylated cytosine is indicated by thymine. In FIG. 2, the region between Primer 2-3 and Primer 2-4 corresponding to the second primer set described in the following Reference Example (SEQ ID NO: 2) has a CpG site 37-43 (Norishige Yamada, Tomofumi Hamada, Masamichi Goto, Hideaki Tsutsumida, Michiyo Higashi, Mitsuharu Nomoto and Suguru Yonezawa (2006) MUC2 expression is regulated by histone H3 modification and DNA methylation in pancreatic cancer.Int.J. Cancer, 119 (8): 1850-7 Similarly, the numbers of the CpG sites are numbered sequentially from the putative promoter upstream of the human MUC2 gene (1,989 bp upstream from the transcription start point of the gene).

  FIG. 3A shows the sequence of the promoter region that regulates the expression of the human MUC4 gene by methylation and the sequence of the adjacent translation region (SEQ ID NO: 8). FIG. 3B shows the DNA sequence of the sequence after Bisulfite treatment (SEQ ID NO: 9), and uracil converted from unmethylated cytosine is indicated by thymine. In FIG. 3, the region between Primer 4-3 and Primer 4-4 corresponding to the second primer set described in the following Reference Example (SEQ ID NO: 3) has CpG sites 108-118 (Norishige Yamada, As with Yukari Nishida, Hideaki Tsutsumida, Masamichi Goto, Michiyo Higashi, Mitsuharu Nomoto and Suguru Yonezawa (2009) Promoter CpG methylation in cancer cells contributes to regulation of MUC4.Br.J.Cancer, 100 (2): 344-51. The numbers of CpG sites are numbered sequentially from the putative promoter upstream of the human MUC4 gene (3,622 bp upstream from the transcription start point of the gene).

  In the present method, for example, the human MUC1, which is an oncogene, the MUC2 and MUC4 genes as the 5′-side untranslated region or the 5′-side untranslated region and the translated region, the sequence between the primers of the second primer set described above (that is, One or more (one or more) CpG sites present in the nucleotide sequences shown in SEQ ID NOs: 1 to 3 (particularly, methylation sites involved in the expression of the human MUC gene) are targets for which the degree of methylation is to be detected. It can be a sequence, in particular, including all methylated sites involved in the expression of the human MUC gene (CpG site), the nucleotide sequence shown in SEQ ID NOs: 1 to 3 or a region containing the nucleotide sequence, methylation Preferably, the target sequence is to be detected.

Hereinafter, each step of the method will be described.
1. First Step of Measuring Methylation Status of Cancer Gene and / or Tumor Suppressor Gene in Sample Sample Derived from Subject In this step, in the sample sample derived from the subject, the 5 ′ untranslated cancer gene and / or tumor suppressor gene The degree of methylation of the region or the region containing the 5 ′ untranslated region and the translated region is detected. Detection (or measurement) of the degree of methylation may be a conventionally known method for detecting methylation, for example, in the detection of methylated DNA in a specific region, a bisulfite (Bisulfite) by performing a base substitution reaction , Determine the sequence of the DNA. In the base substitution reaction, unmethylated cytosine reacts with sodium bisulfite to be converted to uracil. Since methylated cytosine does not react with sodium bisulfite, in principle, the methylation status of all cytosines can be detected as a difference in base. Conventional methods for detecting methylation include, for example, methylation-specific PCR (MSP) using PCR, Real-time MSP using quantitative PCR, Bisulfite-sequencing using TA cloning, and mass spectrometry. Examples include the MassARRAY method used, a pyrosequencing method for monitoring pyrophosphate release during nucleic acid uptake, and amplicon analysis using a next-generation sequencer. Further, an ICON-prove method which does not require a bisulfite reaction and a long sequence method using Pacbio have been developed. Furthermore, there is an MSE (Methylation Specific Electrophoresis) method capable of detecting a DNA methylation pattern or continuity. In the MSE method, (a) a step of subjecting DNA to bisulfite treatment, (b) subjecting the DNA after bisulfite treatment to first PCR using a first primer set corresponding to a region outside the target region Step, (c) subjecting the amplified DNA after the first PCR to the second PCR using the second primer set corresponding to the target region, (d) denaturant concentration gradient of the amplified DNA after the second PCR This is a method including a step of subjecting to gel electrophoresis (DGGE) (Japanese Patent No. 5765586 and Japanese Patent No. 5866739). Analysis using the MSE method, next-generation sequencer, and Pacbio can significantly reduce the time compared to the conventional method, Bisulfite-sequencing.

  For example, the emission intensity of the band in the photograph of the gel after DGGE in the MSE method is digitized by image processing software such as Image J, and statistical processing is performed by statistical analysis software (for example, statistical analysis software “R”). Specifically, the measured value of the methylation status of the oncogene and / or the tumor suppressor gene is defined as 100 for the measured value of the synthetic gene showing methylation positive and 0 for the measured value of the synthetic gene showing methylation negative. Is corrected by two points.

2. To determine the prognosis of the pancreatic tumor of the subject, the oncogene and / or cancer suppressor gene in a plurality of sample samples that are known to be a sample sample derived from tumor tissue or a sample sample derived from non-tumor tissue In the first step, the discriminant of the machine learning model using the measured value obtained by measuring the methylation status of the 5′-side untranslated region or the region including the 5′-side untranslated region and the translated region as a teacher, In the second step of substituting the measured value of the methylation status of the obtained oncogene and / or the region containing the 5'-side untranslated region or the 5'-side untranslated region and the translated region of the tumor suppressor gene, In the same manner as in the first step, the 5′-side of the oncogene and / or the tumor suppressor gene in a plurality of sample samples known to be a sample sample derived from a tumor tissue or a sample sample derived from a non-tumor tissue. Methylation of the region containing the translation region or the 5 'untranslated region and the translation region The situation is measured, and a discriminant of a machine learning model using the measured value as a teacher sample is created.

  Examples of the machine learning model include a support vector machine (hereinafter, referred to as “SVM”) model, a neural network (hereinafter, referred to as “NNET”) model, a random forest model, a deep learning model, and the like.

  The SVM determines a boundary plane called a hyperplane, and determines a discriminant for classifying data using the boundary plane. NNET imitates human neural circuits, constructs artificial neural networks, and performs nonlinear discriminant analysis. The random forest is a group learning algorithm using a decision tree as a weak learner, and uses a large number of decision trees learned based on training data. Deep learning uses a deep neural network of four or more layers. Using these machine learning models, a specific data item of a data set with known grouping information to be classified is used as an explanatory variable, and the grouping to be classified is used as an objective variable, and the data set is correctly classified into a known grouping. Construct a discriminant to perform As an output format, binary determination of poor prognosis or good prognosis is performed. Alternatively, the ratio / probability is calculated.

  Hereinafter, the present invention will be described in more detail with reference to Examples, but the technical scope of the present invention is not limited to these Examples.

(Reference Example) Methylation analysis of mucin gene In the following examples, in accordance with the description of Patent No. 5686739, in a tissue sample, MUC1, MUC2 and MUC4 gene promoter or expression on the translation region adjacent to the promoter. The methylation analysis of the methylated site (CpG site) involved in the DNA was performed by the MSE method.

1. MSE analysis of methylated sites (CpG sites) involved in MUC1 gene promoter expression using MSE method
Using the MSE method, the methylated site (CpG site) involved in the expression of the human MUC1 gene promoter was analyzed. FIG. 1A shows the sequence of a promoter region that regulates the expression of the human MUC1 gene by methylation and the sequence of an adjacent translation region (SEQ ID NO: 4). FIG. 1B shows the DNA sequence of the sequence after Bisulfite treatment (SEQ ID NO: 5), and uracil converted from unmethylated cytosine is indicated by thymine. In FIG. 1, a region (SEQ ID NO: 1) between Primer 1-3 and Primer 1-4 described below has CpG sites 172-181, and a region including the CpG site is used as a target region. The paper (Norishige Yamada, Yukari Nishida, Hideaki Tsutsumida, Tomofumi Hamada, Masamichi Goto, Michiyo Higashi, Mitsuharu Nomoto and Suguru Yonezawa (2008) MUC1 expression is regulated by DNA methylation and histone H3-K9 modification in cancer cells.Cancer Res. , 68 (8): 2708-16), the numbers of the CpG sites are numbered sequentially from the putative promoter upstream of the human MUC1 gene (2,753 bp upstream from the transcription start point of the gene).

DNA was extracted from the tissue sample using the DNeasy Blood & Tissue Kit (QIAGEN).
Next, the extracted DNA was subjected to Bisulfite treatment using EpiTect Bisulfite Kits (manufactured by QIAGEN).

The DNA after the bisulfite treatment was subjected to PCR using the following primers.
Primer set (base sequence in lower case is GC clamp):
Primer 1-1: 5'-AAAGGGGGAGGTTAGTTGGA-3 '(SEQ ID NO: 10);
Primer 1-2: 5'-AAACAACCCACTCCCCACCT-3 '(SEQ ID NO: 11);
Primer 1-3:
5'-cgcccgccgcgcgcggcgggcggggcgggggcacggggggAAGAGGTAGGAGGTAGGGGA-3 '(SEQ ID NO: 12);
Primer 1-4: 5'-AAAACAAAACAAATTCAAAC-3 '(SEQ ID NO: 13).

Each PCR, as shown in FIG. 1, 1 ^st PCR by using the Primer 1-1 and ^{Primer 1-2, 2 nd PCR (nested} PCR) by using the Primer 1-3 and Primer 1-4 went. As the polymerase, AmpliTaq Gold (registered trademark) Fast PCR Master Mix (manufactured by Applied Biosystem) was used. The PCR conditions and temperature settings are shown in Table 1 below.

Then, using a denaturing gradient gel DGGE gel under the conditions shown in Table 2 below, was subjected to the reaction solution after 2 ^nd PCR to DGGE. The electrophoresis conditions were as follows: electrophoresis bath temperature: 60 ° C., constant voltage: 230 V, electrophoresis time: 300 minutes. The electrophoresis tank used was a Dcode system (manufactured by BIO-RAD).

2. MSE analysis of methylated sites (CpG sites) involved in MUC2 gene promoter expression using MSE method
Using the MSE method, the methylated site (CpG site) involved in the expression of the human MUC2 gene promoter was analyzed. FIG. 2A shows the sequence of the promoter region that regulates the expression of the human MUC2 gene by methylation and the sequence of the adjacent translation region (SEQ ID NO: 6). FIG. 2B shows the DNA sequence of the sequence after Bisulfite treatment (SEQ ID NO: 7), and uracil converted from unmethylated cytosine is indicated by thymine. In FIG. 2, a region between Primer 2-3 and Primer 2-4 (SEQ ID NO: 2) described below has a CpG site 37-43, and a region including the CpG site is set as a target region. Thesis (Norishige Yamada, Tomofumi Hamada, Masamichi Goto, Hideaki Tsutsumida, Michiyo Higashi, Mitsuharu Nomoto and Suguru Yonezawa (2006) MUC2 expression is regulated by histone H3 modification and DNA methylation in pancreatic cancer.Int.J. Cancer, 119 ( 8): As in 1850-7), the numbers of CpG sites are numbered in order from the putative promoter upstream of the human MUC2 gene (1,989 bp upstream from the transcription start point of the gene).

  DNA was extracted from the tissue sample using the DNeasy Blood & Tissue Kit (manufactured by QIAGEN) in the same manner as in Section 1 above, and the extracted DNA was extracted using the EpiTect Bisulfite Kits (manufactured by QIAGEN). It was subjected to Bisulfite treatment.

The DNA after the bisulfite treatment was subjected to PCR using the following primers.
Primer set (base sequence in lower case is GC clamp):
Primer 2-1: 5'-TTTGGGGTTAGGTTTGGAAG-3 '(SEQ ID NO: 14);
Primer 2-2: 5'-ACCTTCTTCAAAATAAAACAACC-3 '(SEQ ID NO: 15);
Primer 2-3:
5'-cgcccgccgcgcgcggcgggcggggcgggggcacggggggTTTTAGAGTTTGGGTTTTAG-3 '(SEQ ID NO: 16);
Primer 2-4: 5'-TAACCTAAATACCAACACACA-3 '(SEQ ID NO: 17).

Each PCR, as shown in FIG. 2, 1 ^st PCR by using the Primer 2-1 and ^{Primer 2-2, 2 nd PCR (nested} PCR) is performed using the Primer 2-3 and Primer 2-4 Was. As the polymerase, AmpliTaq Gold (registered trademark) Fast PCR Master Mix (manufactured by Applied Biosystem) was used. Table 3 shows the PCR conditions and temperature settings.

Then, using a denaturing gradient gel DGGE gel under the conditions shown in Table 4 below, was subjected to the reaction solution after 2 ^nd PCR to DGGE. The electrophoresis conditions and the electrophoresis tank were the same as those in the first section.

3. MSE analysis of methylated sites (CpG sites) involved in MUC4 gene promoter expression using MSE method
Using the MSE method, methylated sites (CpG sites) involved in the expression of the human MUC4 gene promoter were analyzed. FIG. 3A shows the sequence of the promoter region that regulates the expression of the human MUC4 gene by methylation and the sequence of the adjacent translation region (SEQ ID NO: 8). FIG. 3B shows the DNA sequence of the sequence after Bisulfite treatment (SEQ ID NO: 9), and uracil converted from unmethylated cytosine is indicated by thymine. In FIG. 3, a region (SEQ ID NO: 3) between Primer 4-3 and Primer 4-4 described below has CpG sites 108 to 118, and a region including the CpG site is used as a target region. Thesis (Norishige Yamada, Yukari Nishida, Hideaki Tsutsumida, Masamichi Goto, Michiyo Higashi, Mitsuharu Nomoto and Suguru Yonezawa (2009) Promoter CpG methylation in cancer cells contributes to regulation of MUC4.Br.J. Cancer, 100 (2): Similarly to (344-51), the numbers of CpG sites are numbered sequentially from the putative promoter upstream of the human MUC4 gene (3,622 bp upstream from the transcription start point of the gene).

  DNA was extracted from the tissue sample using the DNeasy Blood & Tissue Kit (manufactured by QIAGEN) in the same manner as in Section 1 above, and the extracted DNA was extracted using the EpiTect Bisulfite Kits (manufactured by QIAGEN). It was subjected to Bisulfite treatment.

The DNA after the bisulfite treatment was subjected to PCR using the following primers.
Primer set (base sequence in lower case is GC clamp):
Primer 4-1: 5'-TAGTGGGGTGGGGTTGA-3 '(SEQ ID NO: 18);
Primer 4-2: 5'-AAACACCCAAAAAACCC-3 '(SEQ ID NO: 19);
Primer 4-3:
5'-cgcccgccgcgcgcggcgggcggggcgggggcacggggggAGGAGAGAAAAGGGTGATTA-3 '(SEQ ID NO: 20);
Primer 4-4: 5'-ACCCAAAAAACCCTCCTCCA-3 '(SEQ ID NO: 21).

Each PCR, as shown in FIG. 3, 1 ^st PCR by using the Primer 4-1 and ^{Primer 4-2, 2 nd PCR (nested} PCR) is performed using the Primer 4-3 and Primer 4-4 Was. As the polymerase, AmpliTaq Gold (registered trademark) Fast PCR Master Mix (manufactured by Applied Biosystem) was used. Table 5 below shows the PCR conditions and temperature settings.

Then, using a denaturing gradient gel DGGE gel under the conditions shown in Table 6, it was subjected to the reaction solution after 2 ^nd PCR to DGGE. The electrophoresis conditions and the electrophoresis tank were the same as those in the first section.

4. Statistical analysis based on emission intensity of bands in MSE gel photo
The emission intensity of the band in the photograph of the gel after DGGE in the MSE method was quantified by Image J, and statistical processing was performed by statistical analysis software “R”.

[Example 1] Construction of SVM model Materials and Methods The degree of DNA methylation of the mucin gene (gene set consisting of MUC1, MUC2 and MUC4 genes) in 84 cancerous parts and 135 non-cancerous parts including the corresponding cases was analyzed, and the prognostic information (unsatisfactory / non-good) was obtained. The threshold was set to 5 months and the data was classified into binary), and the data were used as training data. The parameters of the discriminant model were adjusted by cross-validation.

2. Results and Discussion As shown in FIG. 4, in the early stage, the group determined to be poor by the model had a significantly poor prognosis, and the hazard ratio was 14.54 in the tumor part and 9.86 in the non-tumor part at a very high angle. The group with significantly poor prognosis could be distinguished. Similarly, at the advanced stage, the group that was determined to be poor by the model had a significantly poor prognosis, and the hazard ratio was 13.97 in the tumor area and 10.44 in the non-tumor area. Was.

[Embodiment 2] Risk discrimination by SVM model Materials and Methods The results of methylation analysis of mucin gene in 20 samples of intrapancreatic papillary mucus-producing tumor (IPMN) and 29 non-tumor sites including the corresponding cases were analyzed for prognosis using the SVM model constructed in Example 1. A good / bad binary decision was made.

2. Results and Discussion Twenty cases of IPMN: intraductal papillary mucinous neoplasm (non-cancer) were analyzed as a test sample group by the SVM model constructed in Example 1, and risk prediction was performed. As shown in FIG. 5, the group determined to be poor by the model had a significantly poor prognosis, and the test group was also able to screen the significantly poor prognostic group.

[Embodiment 3] Risk discrimination by NNET model Materials and Methods Twenty-five pancreatic cancer specimens and 25 non-cancerous specimens were randomly selected, analyzed for the degree of DNA methylation of the mucin gene (gene set consisting of MUC1, MUC2 and MUC4 genes), and analyzed for prognostic information A threshold value of 5 months was classified into binary values), and a discriminator for binarizing the poor or good prognosis using a neural network was constructed. The parameters of the discriminant model were adjusted by cross-validation.

2. Results and Discussion 81 test cases of pancreatic cancer samples and 140 cases of non-cancerous samples were analyzed as test sample groups to predict prognostic risk. As shown in FIG. 6, the group determined to be good by the model in the discrimination by the single model has significantly better prognosis, and the hazard ratio is 0.363 in the tumor part, 0.356 in the non-tumor part, and 0.356 in the non-tumor part. High-risk groups could be screened for both non-tumor sites.

Claims (10)

A first step of measuring the methylation status of a region containing a 5′-side untranslated region or a 5′-side untranslated region and a translated region of a cancer gene and / or a tumor suppressor gene in a specimen sample derived from a subject,
In order to determine the prognosis of the pancreatic tumor of the subject, the oncogene and / or cancer suppression in a plurality of sample samples known to be a sample sample derived from tumor tissue or a sample sample derived from non-tumor tissue The discriminant of the machine learning model using the measurement value obtained by measuring the methylation status of the 5′-side untranslated region of the gene or the region including the 5′-side untranslated region and the translated region as a teacher, A second step of substituting the measured value of the methylation status of the region containing the 5′-side untranslated region or the 5′-side untranslated region and the translated region of the oncogene and / or the tumor suppressor gene obtained in the step,
A method for determining the prognosis of a pancreatic tumor, comprising:   The method of claim 1, wherein the machine learning model is selected from the group consisting of a support vector machine model, a neural network model, a random forest model, and a deep learning model.   3. The method according to claim 1, wherein the tumor tissue and the non-tumor tissue are pancreatic tissues.   An oncogene and / or a tumor suppressor gene is a gene set including a gene encoding one or more mucin core proteins selected from the group consisting of the MUC1 gene, the MUC2 gene, and the MUC4 gene, any one of claims 1 to 3. The method of claim 1.   The method according to any one of claims 1 to 3, wherein the oncogene and / or the tumor suppressor gene is a gene set including the MUC1, MUC2, and MUC4 genes.   The method according to any one of claims 1 to 3, wherein the oncogene and / or the tumor suppressor gene is a gene set consisting of the MUC1, MUC2, and MUC4 genes.   The region containing the 5′-side untranslated region or the 5′-side untranslated region and the translated region of the MUC1 gene is a region consisting of the nucleotide sequence shown in SEQ ID NO: 1. the method of.   The region containing the 5′-side untranslated region or the 5′-side untranslated region and the translated region of the MUC2 gene is a region consisting of the nucleotide sequence shown in SEQ ID NO: 2. the method of.   The region comprising the 5'-side untranslated region or the 5'-side untranslated region and the translated region of the MUC4 gene is a region consisting of the nucleotide sequence shown in SEQ ID NO: 3. the method of.   The method according to any one of claims 1 to 9, wherein the pancreatic tumor is pancreatic cancer or intraductal papillary mucinous tumor.