WO2021107232A1 - Method for forming biomarker panel for diagnosing ovarian cancer and biomarker panel for diagnosing ovarian cancer - Google Patents

Abstract

The present invention relates to a method for forming a biomarker panel for diagnosing ovarian cancer, and a biomarker panel for diagnosing ovarian cancer, and forms a biomarker panel through transcript expression data analysis and machine learning analysis of ovarian cancer patients with clear differences in treatment prognosis, so as to predict the prognosis of ovarian cancer patients, and thus more appropriate treatment can be carried out for patients.

Description

Method of constructing a biomarker panel for ovarian cancer diagnosis and biomarker panel for ovarian cancer diagnosis

The present invention relates to a method of constructing a biomarker for ovarian cancer diagnosis.

70% of patients with intractable ovarian cancer (High-grade serous ovarian cancer) die mostly due to cancer. The standard treatment for chemotherapy after surgery is 25% of patients relapse within 6 months after chemotherapy, and only 31% of patients have a chance to live more than 5 years after chemotherapy, so it is necessary to develop a new treatment other than standard treatment. situation. To develop new treatments other than standard treatments, The Cancer Genome Atlas (TCGA) researcher consortium produced exome data, expression data (mRNA data) and epigenetics data for about 300 ovarian cancer patients. The patient's characteristics were identified by analysis. However, in this existing TCGA study, no patient clusters or gene expression marker lists with significant differences in prognosis were found.

Therefore, there is a need to secure a list of markers of expressed genes for predicting the prognosis of refractory ovarian cancer patients.

One aspect comprises the steps of obtaining gene expression data of a plurality of individual biomarkers from a sample isolated from the subject; And to provide a method of constructing a biomarker panel for ovarian cancer diagnosis, comprising the step of selecting a gene related to the prognosis of ovarian cancer from the data.

Another aspect is to provide a biomarker panel for ovarian cancer diagnosis constructed by the method.

Another aspect is selected from the group consisting of ARAF, ARFGAP2, IFIT1, IRF9, LMO7, MTA2, POLR2A, RAB12, RARG, RPL23P8, RPL23, RTCA, SCAP, SELENON, SERP1, SIAH2, TES2, TIMM17B, TLN1, USP19 and ZFAND6 It is to provide a biomarker panel for ovarian cancer diagnosis comprising an agent for measuring the expression level of one or more biomarkers.

Another aspect is ARAF, ARFGAP2, IFIT1, IRF9, LMO7, MTA2, POLR2A, RAB12, RARG, RPL23P8, RPL23, RTCA, SCAP, SELENON, SERP1, SIAH2, TES2, TIMM17B, and TLN1, USP19 in a sample isolated from a subject. Measuring the expression level of one or more biomarkers selected from the group consisting of; and comparing the level of the biomarker with a corresponding result of the corresponding marker in a control sample.

One aspect comprises the steps of obtaining gene expression data of a plurality of individual biomarkers from a sample isolated from the subject; And it provides a method of constructing a biomarker panel for ovarian cancer diagnosis, comprising the step of selecting a gene related to the prognosis of ovarian cancer from the data.

The method according to one embodiment includes acquiring gene expression data of a plurality of individual biomarkers from a sample isolated from a subject. The subject is a subject for diagnosing ovarian cancer, for example, a subject for predicting the likelihood of ovarian cancer, a subject for diagnosing a state of ovarian cancer, a subject for determining prognosis, prevention or treatment of ovarian cancer It means a subject for determining the dosage of the drug for use, a subject for determining a treatment method according to the progression of ovarian cancer, and the like. The subject may be a vertebrate, specifically mammals, amphibians, reptiles, birds, etc., and more specifically, may be mammals, for example, humans ( Homo sapiens ), and Koreans. can The sample may include a tissue, cell, whole blood, serum, plasma, saliva, sputum, cerebrospinal fluid, or urine sample isolated from the subject.

The data may be transcript expression data of a patient with ovarian cancer. The data are, for example, in the top 30% of those with very large variance in gene expression using the Median Absolute Deviation (MAD) to select genes related to the prognosis of refractory ovarian cancer. It may be obtained by selecting the gene. The ovarian cancer patient may be Korean.

The method according to one embodiment includes selecting a gene related to the prognosis of ovarian cancer from the data. The step may be to first select a gene having a large difference in the treatment prognosis of ovarian cancer from the data. The difference in the treatment prognosis of the ovarian cancer may be confirmed, for example, through an AUC analysis or a log-Rank test of disease-free survival. The step may include selecting a secondary gene through literature search among the primary selected genes. Specifically, the literature may be PMID (PubMed identifier): 28539603, 23160375, 27517492, 26988033, and the secondary gene can be selected by confirming whether the primary selected gene is related to a molecular biological pathway that causes or prevents cancer. have. Thereafter, a gene related to the prognosis of ovarian cancer may be finally selected from the selected secondary gene through machine learning analysis. The machine learning analysis may be performed, for example, through a random forest (RF) or decision tree model.

As used herein, the term "random forest (RF)" is a method proposed by Leo Breiman and Adel Cutler as a kind of bagging algorithm consisting of a combination of decision trees of CART. The nodes of each tree are structured so that high-dimensional data can be divided into smaller pieces of lower dimensions so that they can be quickly classified. Each of these trees completes the final classification by ensemble and voting. Trees generated by a random vector with the same probability distribution are each independently constructed, and if the number of constructed trees is taken to infinity, misclassification is generalized and converges. -Bag (Random Selection without Replacement) technique is used to make it as accurate as Adaboost, and it has strong performance against interface and noise, and has the effect of helping to converge faster than bagging and boosting.

The RF algorithm itself creates a plurality of (training data set, test data set) from the given data (for example, 50, this number can be adjusted by the user as an option) and creates a decision tree from each. This will create 50 independent decision trees. After creating 50 decision trees in this way, if you insert a test set, you will have 50 decisions (cancer/normal) per test sample (values from each decision tree), and 50 decision values. , and the final result is obtained by the majority vote. For example, in the case of Person A, if 45 decision trees are judged to be cancerous and 5 decision trees are judged to be normal, the vaverage score (ratio determined as cancer among all 50 judgments)=45/50=0.9 is calculated. do. At this time, assuming that the cut-off value, which is the criterion for distinguishing between cancer and normal, is 0.5, the average score of A is greater than 0.5, so it can be determined as "cancer".

As used herein, the term “decision tree model is a method of finding a decision rule based on the structure of a tree as one of the analysis techniques of data mining. A decision tree is a method of finding a decision rule based on a tree structure. It is a powerful and widely used analysis technique that classifies or predicts a group of interest into several subgroups.The general algorithm of a decision tree has different formation processes such as stopping rules and pruning. The rules to be

1. Separation criteria: Child nodes are determined by determining which predictor variable is used to classify the distribution of the target variable best. Purity or other classification criteria are used to determine the degree to which the distribution of the target variable is distinguished. to measure it.

2. Stop Criteria: A rule that specifies that the current node becomes a terminal node without further separation.

3. Pruning: Decision trees with too many nodes are likely to have very large prediction errors when applied to new data. Therefore, it is desirable to select a decision tree having a sub-tree structure of an appropriate size as the final model by removing inappropriate nodes from the formed decision tree.

The method according to an embodiment may further include verifying the prognosis of the selected gene using The Cancer Genome Altas (TCGA) data. As used herein, the term "TCGA (The Cancer Genome Altas) data" refers to genetic mutation data related to cancer and is a huge data set that can comprehensively explain changes at the molecular level that occur in cancer cells. used when doing Using the data, it is possible to list specific genomic or molecular level changes in cancer, as well as define a more meaningful classification system for cancer types and subtypes.

In the method, the prognosis is predicted using TCGA data using the random forest model, and the selected gene is classified into a group with a good prognosis and a group with a poor prognosis. Thereafter, the two groups classified according to the predicted prognosis can be verified by checking whether the prognosis of disease-free survival is different as predicted by the random forest model.

Another aspect provides a panel of biomarkers for diagnosing ovarian cancer constructed by the method. Another aspect is selected from the group consisting of ARAF, ARFGAP2, IFIT1, IRF9, LMO7, MTA2, POLR2A, RAB12, RARG, RPL23P8, RPL23, RTCA, SCAP, SELENON, SERP1, SIAH2, TES2, TIMM17B, TLN1, USP19 and ZFAND6 It provides a biomarker panel for ovarian cancer diagnosis, comprising an agent for measuring the expression level of one or more biomarkers. In one embodiment, the biomarker panel may include an agent for measuring the expression level of a biomarker of LMO7, RAB12, RPL23, TES, USP19, or a combination thereof. In another embodiment, the biomarker panel may include an agent for measuring the expression level of a biomarker of RPL23, USP19, or a combination thereof.

As used herein, the term "biomarker panel" is constructed using any combination of biomarkers for ovarian cancer diagnosis, and the combination may refer to the entire set, or any subset or subcombination thereof. That is, the biomarker panel may refer to one set of biomarkers, and may refer to any type of biomarker to be measured. Thus, when RPL23 is part of a panel of biomarkers, for example, RPL23 mRNA or RPL23 protein can be considered to be part of that panel. While individual biomarkers are useful as diagnostic agents, sometimes combinations of biomarkers can provide greater values than single biomarkers alone in determining a particular condition. Specifically, detecting a plurality of biomarkers in a sample may increase the sensitivity and/or specificity of the test. Thus, in one embodiment, the biomarker panel is 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 , 21 or more biomarker types. In another embodiment, the biomarker panel is comprised of a minimum number of biomarkers to generate a maximum amount of information. Thus, in various embodiments, the biomarker panel comprises 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 , consisting of more than 21 biomarker types. When a biomarker panel consists of "a set of biomarkers", no biomarkers are present other than those that make up the set. In one embodiment, the biomarker panel consists of one biomarker disclosed herein. In another embodiment, the biomarker panel consists of two biomarkers disclosed herein. In another embodiment, the biomarker panel consists of three biomarkers disclosed herein. In another embodiment, the biomarker panel consists of four biomarkers disclosed herein. In another embodiment, the biomarker panel consists of five biomarkers disclosed herein. In another embodiment, the biomarker panel consists of six biomarkers disclosed herein. In another embodiment, the biomarker panel consists of 7 biomarkers disclosed herein. In another embodiment, the biomarker panel consists of 8 biomarkers disclosed herein. In another embodiment, the biomarker panel consists of nine biomarkers disclosed herein. In another embodiment, the biomarker panel consists of 10 biomarkers disclosed herein. In another embodiment, the biomarker panel consists of 11 biomarkers disclosed herein. In another embodiment, the biomarker panel consists of 12 biomarkers disclosed herein. In another embodiment, the biomarker panel consists of 13 biomarkers disclosed herein. In another embodiment, the biomarker panel consists of 14 biomarkers disclosed herein. In another embodiment, the biomarker panel consists of 15 biomarkers disclosed herein. In another embodiment, the biomarker panel consists of 16 biomarkers disclosed herein. In another embodiment, the biomarker panel consists of 17 biomarkers disclosed herein. In another embodiment, the biomarker panel consists of 18 biomarkers disclosed herein. In another embodiment, the biomarker panel consists of 19 biomarkers disclosed herein. In another embodiment, the biomarker panel consists of 20 biomarkers disclosed herein. In another embodiment, the biomarker panel consists of 21 biomarkers disclosed herein. The biomarker of the present invention represents a statistically significant difference in the diagnosis of ovarian cancer. In one embodiment, a diagnostic test using such biomarkers alone or in combination exhibits a sensitivity and specificity of about 85% or greater, about 90% or greater, about 95% or greater, about 98% or greater, and about 100% or greater. .

The biomarker may be obtained from transcript data of cells. Details of the data are the same as described above.

The agent for measuring the expression level of the biomarker may be a primer pair, a probe, or an antisense nucleotide. Specifically, it may be an agent for measuring the mRNA level of the biomarker gene, and may be a primer pair, a probe, or an antisense nucleotide that specifically binds to the gene. In one embodiment, the biopanel may include at least two primer pairs, probes or antisense nucleotides, and each of the primer pairs, probes or antisense nucleotides is ARAF, ARFGAP2, IFIT1, IRF9, LMO7, MTA2, POLR2A. , RAB12, RARG, RPL23P8, RPL23, RTCA, SCAP, SELENON, SERP1, SIAH2, TES2, TIMM17B, TLN1, USP19 and ZFAND6.

The agent for measuring the expression level of the biomarker may be an antibody. The antibody may be a monoclonal antibody, eg, a monoclonal antibody that specifically binds to any of the above biomarkers. In one embodiment, the biopanel comprises at least 21 antibodies, each of which is ARAF, ARFGAP2, IFIT1, IRF9, LMO7, MTA2, POLR2A, RAB12, RARG, RPL23P8, RPL23, RTCA, SCAP, SELENON, SERP1 , SIAH2, TES2, TIMM17B, TLN1, USP19 and ZFAND6.

Another aspect is ARAF, ARFGAP2, IFIT1, IRF9, LMO7, MTA2, POLR2A, RAB12, RARG, RPL23P8, RPL23, RTCA, SCAP, SELENON, SERP1, SIAH2, TES2, TIMM17B, and TLN1, USP19 in a sample isolated from a subject. Measuring the expression level of one or more biomarkers selected from the group consisting of; and comparing the expression level of the biomarker with a corresponding result of the corresponding marker in a control sample.

The method according to one embodiment includes ARAF, ARFGAP2, IFIT1, IRF9, LMO7, MTA2, POLR2A, RAB12, RARG, RPL23P8, RPL23, RTCA, SCAP, SELENON, SERP1, SIAH2, TES2, TIMM17B, TLN1 in a sample isolated from an individual. , measuring the expression level of one or more biomarkers selected from the group consisting of USP19 and ZFAND6. The measurement of the biomarker level in the sample is ARAF, ARFGAP2, IFIT1, IRF9, LMO7, MTA2, POLR2A, RAB12, RARG, RPL23P8, RPL23, RTCA, SCAP, SELENON, SERP1, SIAH2, TES2, TIMM17B, TLN1, USP19 and ZFAND6 It can be carried out by measuring the mRNA level or protein level of one or more biomarker genes selected from the group consisting of. Specifically, the measurement of the mRNA level is a process of determining the presence and expression level of mRNA of genes in a sample of an individual for diagnosing cancer, and measuring the amount of mRNA. Analysis methods for this are reverse transcription polymerase reaction (RT-PCR), competitive reverse transcription polymerase reaction (Competitive RT-PCR), real-time reverse transcription polymerase reaction (Real-time RT-PCR), RNase protection assay (RPA; RNase protection) assay), Northern blotting, and a DNA chip. In addition, the protein level measurement is a process of confirming the presence and expression level of a cancer diagnostic marker protein in a sample of an individual in order to diagnose cancer. The amount of the protein can be confirmed using an antibody that specifically binds to the marker protein, and the protein expression level itself can be measured without using the antibody. The protein level measurement or comparative analysis method includes protein chip analysis, immunoassay, ligand binding assay, MALDI-TOF (Matrix Desorption/Ionization Time of Flight Mass Spectrometry) analysis, SELDI-TOF (Surface Enhanced Laser Desorption/Ionization Time of Flight Mass Spectrometry analysis, radioimmunoassay, radioimmunodiffusion method, octeroni immunodiffusion method, rocket immunoelectrophoresis, tissue immunostaining, complement fixation assay, two-dimensional electrophoresis analysis, liquid chromatography-mass spectrometry Mass Spectrometry, LC-MS), LC-MS/MS (liquid chromatography-Mass Spectrometry/Mass Spectrometry), Western blot, and ELISA (enzyme linked immunosorbent assay) are available.

The method according to one embodiment comprises comparing the expression level of the biomarker with a corresponding result of the corresponding marker in a control sample. For example, when the expression level of the biomarker is significantly increased compared to a control group, it may be determined that the prognosis of ovarian cancer is poor.

The method according to one aspect comprises the steps of analyzing transcript expression data and machine learning analysis of ovarian cancer patients with clear differences in treatment prognosis, and can construct a complex biomarker with higher diagnostic ability of ovarian cancer than a single biomarker. have. In addition, the composite biomarker constructed according to the method can provide more appropriate treatment to patients by predicting the prognosis of patients with refractory ovarian cancer.

1 shows a schematic diagram of a method for selecting a complex biomarker by constructing a model by selecting genes related to prognosis.

2A is a graph classified according to importance when constructing a random forest prediction model with 5 intractable ovarian cancer genes.

Figure 2b shows a decision tree model constructed with five intractable ovarian cancer genes.

Figure 2c shows a random forest model constructed with five intractable ovarian cancer genes.

2D is a graph showing the results of verification with TCGA data after constructing a random forest model with two genes estimated to have good predictive power for intractable ovarian cancer.

Hereinafter, preferred examples are presented to help the understanding of the present invention. However, the following examples are only provided for easier understanding of the present invention, and the contents of the present invention are not limited by the following examples.

[Example]

Example 1. Preparation of Patient Samples

This study was conducted with the approval of the Institutional Review Board of CHA Bundang Hospital. RNA samples were prepared from ovarian cancer tissues of 54 patients in their 30s and 70s who underwent ovarian cancer surgery at CHA Bundang Hospital. Specifically, RNA was extracted from ovarian cancer tissue using Trizol according to Invitrogen's protocol, and then the RNA was quantified using NanoDrop200. Thereafter, it was confirmed that there was no abnormality in the RNA using the Agilent 200 Bioanalyzer.

Characteristic	division	Number of patients (45)
age	<40	2
	[40 to 50)	12
	[50 to 60)	20 (44.4%)
	[60 ~ 70)	6
	[70 ~ 80)	5
ordnance	3rd term	40 (88.9%)
	4th	5
group	No recurrence	6
	Relapse: Platinum sensitive	27
	Relapse: Platinum resistance	6
	Refractory	6

In Table 1, most of the stages of ovarian cancer patients correspond to stage 3, but include stage 1 to 4 patients. Ages ranged from those in their 30s to 70s, with the 50s being the most common. The study analysis confirmed that the stage of the patients was highly related to the prognosis, and the expression pattern was also highly related and very different from the stage. Therefore, patients with stage 1 or 2 stage were excluded from the analysis. A total of 45 patients with refractory ovarian cancer detected at stage 3 or higher. Of the 45 patients, 6 patients had a good prognosis and 39 patients had a poor prognosis who relapsed or did not respond to treatment. In addition, the patient group was 6 patients in the No recurrence (NR) group without recurrence or recurrence. However, there were 27 patients in the Platinum sensitive (PS) group who relapsed after 6 months, 6 patients in the Platinum resistance (PR) group who had recurrence before 6 months, and 6 patients in the refractory (RF) group with very poor response. consists of names. The six patients who did not relapse had a very good prognosis, and the platinum-resistant group or the intractable group that had relapsed before 6 months had a very poor prognosis. Therefore, these groups can be divided into three groups, NR vs. PS vs. PR+RF, or two groups: NR vs. other groups, or PR+RF vs. other groups, depending on the timing of recurrence.

Example 2. Gene selection related to the prognosis of intractable ovarian cancer through expression data analysis

To select genes related to the prognosis of refractory ovarian cancer, the genes corresponding to the top 30% among those with very large variance in gene expression levels were selected using the median absolute deviation (MAD). In this case, the gene is not related to the group shown in Table 1. Gene expression analysis was performed using the mRNA sequencing pipeline of The Cancer Genome Atlas (TCGA). MapSplice v2.1.6 was used to map the transcript, and the amount of mRNA expressed in each gene was measured using RSEM-1.2.5 and NCBI hg19 RefSeq. Thereafter, by analyzing the AUC (Area Under a ROC curve) value according to the group, the AUC value was 0.75 or more and statistically significant genes were selected in the log-Rank analysis of disease-free survival and shown in Table 2 below. . At this time, the gene that predicted the group well according to the patient's prognosis was selected using the statistical ROCR package, and the survival R package was used for disease-free survival analysis.

	gene	MAD	AUC	DFSp-Value	full gene name
One	ARAF	18.8	0.88	0.038	A-Raf proto-oncogene
2	ARFGAP2	22.5	0.88	0.025	ADP ribosylation factor GTPase activating protein 2
3	IFIT1	91.6	0.79	0.039	Interferon induced protein with tetratricopeptide repeats
4	IRF9	48.2	0.82	0.010	Interferon regulatory factor 9
5	LMO7	15.1	0.88	0.006	LIM domain 7
6	MTA2	19.8	0.90	0.04	Metastasis associated 1 family member 2
7	POLR2A	23.3	0.90	0.045	RNA polymerase II subunit A
8	RAB12	13.5	0.79	0.013	RAB12, member RAS oncogene family
9	RARG	17.2	0.93	0.043	Retinoic acid receptor gamma
10	RPL23P8	135	0.85	0.014	Ribosomal protein L23 pseudogene 8
11	RPL23	769	0.86	0.042	Ribosomal protein L23
12	RTCA	11	0.79	0.047	RNA 3'-terminal phosphate cyclase
13	SCAP	14.1	0.91	0.005	SREBP chaperone
14	SELENON	20.2	0.90	0.022	Selenoprotein N
15	SERP1	48.9	0.88	0.010	Stress associated endoplasmic reticulum protein 1
16	SIAH2	21.6	0.88	8.0e-4	Siah E3 ubiqutin protein ligase 2
17	TES2	36.6	0.87	0.037	Testin LIM domain protein
18	TIMM17B	26.6	0.88	0.010	Translocase of inner mitochondrial membrane 17B
19	TLN1	28.2	0.86	0.044	Talin 1
20	USP19	12.2	0.88	0.009	Ubiquitin specific peptidase 19
21	ZFAND6	40.2	0.76	0.004	Zinc finger AN1-type containing 6

Example 3. Construction of Random Forest Model and Decision Tree Model

Among the 21 genes shown in Table 2 above, 5 genes (LMO7, RAB12, RPL23, TES and USP19) were selected in consideration of their relevance to cancer disclosed in the literature. Thereafter, a random forest model and a decision tree model, which are machine learning modeling methods, were constructed for the five genes.

2A is a graph classified according to importance when constructing a random forest prediction model with 5 intractable ovarian cancer genes. RPL23 and USP19 have the largest values in MeanDecreaseAccuracy and MeanDecreaseGini. This quantifies how much the accuracy decreases when determining the random forest prediction model excluding the genes, and the larger the number, the more important the gene corresponds to the predictive power of the model. Therefore, as shown in Fig. 2a, when the importance of genes was confirmed through the random forest model, it can be seen that RPL23 and USP19 are the most important genes in the MeanDecreaseAccuracy and MeanDecreaseGini methods.

Figure 2b shows a decision tree model constructed with five intractable ovarian cancer genes. Among the five genes, USP19 was used the most, and after dividing groups by USP19, a decision tree model was made to determine the next group using the TES gene. Therefore, it can be seen that USP19 is the most important gene among intractable ovarian cancer genes.

After constructing a random forest model for each of 5 and 2 genes, TCGA data were used for verification. In the TCGA data, there were 205 patients with stage 3 or higher. Since the clinical characteristics of the 205 patients are not clearly known from the TCGA data, the random forest model was divided into a group predicted to have a good prognosis and a group predicted to have a poor prognosis. It was verified whether it was visible or not.

Figure 2c shows a random forest model constructed with five intractable ovarian cancer genes. As shown in Figure 2c, when the random forest model constructed with 5 ovarian cancer genes was applied to the TCGA test set, 16 patients predicted to have a good prognosis and 189 patients predicted to have a poor prognosis were disease-free. It was confirmed that there was a difference in prognosis as predicted by survival.

In addition, FIG. 2D is a graph showing the results of verification with TCGA data after constructing a random forest model with two genes estimated to have good predictive power for intractable ovarian cancer. As shown in FIG. 2D , it was found that the disease-free survival prognosis of 53 patients predicted to have a good prognosis and 152 patients predicted to have a poor prognosis differed at a statistically significant level as predicted. In addition, it was confirmed that the p-value of disease-free survival was better than the random forest model made with two genes. Therefore, it can be seen that the two-gene random forest model is a suitable model for predicting the prognosis of refractory ovarian cancer.

Claims (12)

1. acquiring gene expression data of a plurality of individual biomarkers from a sample isolated from an individual; and

   A method of constructing a biomarker panel for ovarian cancer diagnosis, comprising the step of selecting a gene related to the prognosis of ovarian cancer from the data.
2. The method according to claim 1, wherein the step of selecting the gene is performed using a median absolute deviation (Median Absolute Deviation, MAD).
3. The method according to claim 1, further comprising the step of verifying the prognosis of the selected gene using TCGA (The Cancer Genome Altas) data.
4. The method according to claim 1, wherein the data is information on the expression level of individual biomarker genes.
5. The method according to claim 1, Selecting genes related to the prognosis of ovarian cancer by generating expression data information of individual biomarker genes using a random forest model and a decision tree model how to be.
6. The method according to claim 1, wherein the ovarian cancer is stage 3 or stage 4.
7. A biomarker panel for diagnosing ovarian cancer constructed by the method of claim 1.
8. At least one biologic selected from the group consisting of ARAF, ARFGAP2, IFIT1, IRF9, LMO7, MTA2, POLR2A, RAB12, RARG, RPL23P8, RPL23, RTCA, SCAP, SELENON, SERP1, SIAH2, TES2, TIMM17B, TLN1, USP19 and ZFAND6 A biomarker panel for diagnosing ovarian cancer, comprising an agent for measuring the expression level of the marker.
9. The biomarker panel of claim 8 , wherein the biomarker is obtained from transcript data of a cell.
10. The biomarker panel of claim 8, wherein the agent for measuring the expression level of the biomarker is a primer pair, Fbor, or an antisense nucleotide.
11. The biomarker panel of claim 8, wherein the agent for measuring the expression level of the biomarker is an antibody.
12. group consisting of ARAF, ARFGAP2, IFIT1, IRF9, LMO7, MTA2, POLR2A, RAB12, RARG, RPL23P8, RPL23, RTCA, SCAP, SELENON, SERP1, SIAH2, TES2, TIMM17B, TLN1, USP19 and ZFAND6 in a sample isolated from a subject Measuring the expression level of one or more biomarkers selected from; and comparing the level of the biomarker with a corresponding result of the corresponding marker in a control sample.

Images