KR20230149789A - Method of providing information for diagnosis of ovarian cancer - Google Patents

Abstract

The present invention relates to a method of providing information for diagnosing ovarian cancer. In particular, in order to provide information necessary for diagnosing ovarian cancer, the CA-125 level and the relative abundance of Acinetobacter obtained from the subject's sample are measured, and the probability of ovarian cancer is diagnosed using these as variables.

Description

{Method of providing information for diagnosis of ovarian cancer}

The present invention relates to a method of providing information for diagnosing ovarian cancer.

Ovarian cancer is one of the most lethal gynecological cancers worldwide. In the United States, the number of cancer deaths due to ovarian cancer in 2019 is estimated at 13,980 (4.9% of female cancer deaths), and in Korea, the incidence of ovarian cancer is gradually increasing. However, because there is no effective diagnostic method for ovarian cancer, most ovarian cancers are diagnosed at a stage when treatment is difficult, resulting in high recurrence and mortality rates.

Distinguishing between ovarian cancer and benign ovarian tumors is an important issue for determining treatment plans, including surgical approaches. Currently available detection tools for ovarian cancer include blood cancer antigen 125 (CA-125) levels, ultrasonography, computed tomography (CT) scans, and magnetic resonance imaging (MRI). However, considering the differences in their diagnostic performance and accuracy, there is still a need for improvement in ovarian cancer diagnosis methods.

Recent microbiome studies have shown that significant differences in microbiome composition are associated with oral, esophageal, pancreatic, and colorectal cancers. Advances in sequencing technologies for microbial genomes are expanding microbiome data and expanding our understanding of microbiome-host interactions, particularly studies of extracellular vesicles secreted by microorganisms and detectable in body fluids. It is being done. However, the relationship between blood microbial extracellular vesicles and ovarian cancer has not yet been investigated.

Korean Patent No. 1576053

The present invention provides a method of providing information for diagnosing ovarian cancer using metagenomic analysis.

The present invention provides a diagnostic device for diagnosing ovarian cancer using metagenomic analysis.

1. Determining the relative abundance of Acinetobacter in a sample of the subject; and

A method of providing information for diagnosing ovarian cancer, including calculating the probability that the subject has ovarian cancer according to Equation 1 below.

[Equation 1]

Y1 = X1 * patient age (age) + X2 * Acinetobacter relative abundance.

(X1 is the weight for the patient's age in Equation 1, X2 is the weight for the relative abundance of Acinetobacter in Equation 1, and the sum of X1 and X2 is between 0 and 1.)

2. The method of providing information for diagnosing ovarian cancer in 1 above, wherein X1 is 0.015 to 0.045 and X2 is 0.35 to 0.65.

3. Determining the relative abundance of CA-125 and Acinetobacter in a sample from the subject; and

A method of providing information for diagnosing ovarian cancer, including calculating the probability that the subject has ovarian cancer according to Equation 2 below.

[Equation 2]

Y2 = X3 * Patient age + X4 * Sample CA-125 level + X5 * Acinetobacter relative abundance.

(X3 is the weight for the patient's age in Equation 2, X4 is the weight for the relative abundance of Acinetobacter in Equation 2, X5 is the weight for the sample CA-125 in Equation 2, The sum of X5 is between 0 and 1.)

4. The method of providing information for diagnosing ovarian cancer according to 3 above, wherein X3 is 0.010 to 0.030, X4 is 0.003 to 0.0055, and X5 is 0.45 to 0.65.

5. The method of any one of 1 to 4 above, wherein the step of measuring the relative abundance of Acinetobacter in the blood of the subject is,

A method of providing information for diagnosing ovarian cancer, which is performed by analyzing DNA contained in extracellular vesicles isolated from the blood.

6. Measuring CA-125 levels in a sample from the subject; and

A method of providing information for diagnosing ovarian cancer, including calculating the probability that the subject has ovarian cancer according to Equation 3 below.

[Equation 3]

Y3 = X6 * subject age + X7 * sample CA-125 level.

(X6 is the weight for patient age in Equation 3, X7 is the weight for sample CA-125 in Equation 3, and the sum of X6 and X7 is between 0 and 1.)

7. The method of providing information for diagnosing ovarian cancer in item 6 above, wherein X6 is 0.010 to 0.035 and X7 is 0.002 to 0.005.

8. An ovarian cancer diagnostic device including an arithmetic unit that calculates the probability that a subject has ovarian cancer according to Equation 2 below.

[Equation 2]

Y2 = X3 * Patient age + X4 * Sample CA-125 level + X5 * Acinetobacter relative abundance.

(X3 is the weight for the patient's age in Equation 2, X4 is the weight for the relative abundance of Acinetobacter in Equation 2, X5 is the weight for the sample CA-125 in Equation 2, The sum of X5 is between 0 and 1.)

9. The method of providing information for diagnosing ovarian cancer according to 8 above, wherein X3 is 0.010 to 0.030, X4 is 0.003 to 0.0055, and X5 is 0.45 to 0.65.

Through the present invention, it is possible to efficiently distinguish between benign and malignant ovarian tumors and can be used for diagnosing ovarian cancer.

Figure 1 is an overall study design flow chart.
Figure 2 presents the landscape of metagenomic profiles in all patients. Figure 2 (A) shows the structure at the phylum-level. Figure 2 (B) shows the organization at the genus-level. Red and green horizontal bars below the plot represent patients with ovarian cancer and patients with benign ovarian tumors, respectively.
Figure 3 shows the results of comparing α-diversity at the genus level between the two groups. Figure 3(A) is the training set, and Figure 3(B) is the test set.
Figure 4 shows a multidimensional plot at the genus level. Figure 4(A) shows the training set, and Figure 4(B) shows the test set.
Figure 5 shows the overlap of overall biomarkers among eight statistical methods of microbiome abundance at the genus level.
Figure 6 shows the results of performance comparison between diagnostic models for distinguishing ovarian cancer from benign ovarian tumors.

The present invention relates to a method for providing information for diagnosing ovarian cancer, comprising measuring the relative abundance of Acinetobacter in a sample of a subject and calculating the probability that the subject has ovarian cancer according to Equation 1 below: It's about.

[Equation 1]

Y1 = X1 * patient age (age) + X2 * Acinetobacter relative abundance.

In the present invention, Acinetobacter is a gram-negative bacterium that is part of the microbiome in a subject and can be used as an indicator for cancer diagnosis. For example, it can be used as an indicator of ovarian cancer (OC).

In the present invention, extracellular vesicles secreted by Acinetobacter contained in the subject sample can induce cytotoxicity of epithelial cells and host inflammatory response.

In this specification, “subject” refers to all animals such as rats, mice, livestock, etc., including humans, that have developed or may develop diabetes. As a specific example, it may be a mammal, including humans.

In the present invention, the relative abundance of Acinetobacter can be measured using a sample of an object. For example, this may include the subject's blood, saliva, or body hair.

In the present invention, the relative abundance of Acinetobacter may be a ratio calculated using a centered log ratio transformation (CLR) transformation.

In the present invention, the relative abundance of Acinetobacter may be “count value of Acinetobacter in the blood of the subject/count value of the entire genus.”

In the present invention, the step of measuring the relative abundance of Acinetobacter in a sample of the subject may be performed by analyzing DNA contained in extracellular vesicles isolated from the sample. For example, the relative abundance of Acinetobacter can be measured by analyzing the DNA of extracellular vesicles isolated from the subject's blood.

In the present invention, the step of calculating the probability that a subject has ovarian cancer can be calculated using a mathematical formula according to variables. For example, when patient age and Acinetobacter relative abundance are used as variables, it can be calculated using Equation 1 above.

In the present invention, the patient's age was calculated as full age.

In the present invention, each variable in Equation 1 can be calculated as a value multiplied by each weight, depending on the weight it occupies in Equation 1. And the sum of the values of each weight may be between 0 and 1 (eg, greater than 0 and less than or equal to 1).

For example, X1 may be a weight for the patient's age in Equation 1, and X2 may be a weight for the relative abundance of Acinetobacter in Equation 1.

In the present invention, X1 may be 0.010 to 0.050, for example, 0.015 to 0.045, or 0.020 to 0.040.

In the present invention, X2 may be 0.30 to 0.70, for example, 0.35 to 0.65, or 0.40 to 0.60.

The present invention provides information for diagnosing ovarian cancer, including measuring the relative abundance of CA-125 and Acinetobacter in a sample of the subject and calculating the probability that the subject has ovarian cancer according to Equation 2 below. It's about method.

[Equation 2]

Y2 = X3 * Patient age + X4 * Blood CA-125 level + X5 * Acinetobacter relative abundance.

In the present invention, CA-125 (Cancer antigen 125) refers to cancer antigen 125 and is one of the tumor markers. For example, it can be used as an indicator of ovarian cancer, uterine cancer, etc.

In the present invention, CA-125 levels can be measured using a sample from a subject. For example, this may include the subject's blood, saliva, or body hair.

In the present invention, the step of calculating the probability that a subject has ovarian cancer can be calculated using a mathematical formula according to variables. For example, when patient age, CA-125 level, and Acinetobacter relative abundance are used as variables, it can be calculated using Equation 2 above.

In the present invention, each variable in Equation 2 can be calculated as a value multiplied by each weight according to the weight occupied in Equation 2. And the sum of the values of each weight may be between 0 and 1 (eg, greater than 0 and less than or equal to 1).

For example, X3 may be the weight for patient age in Equation 2, X4 may be the weight for Acinetobacter relative abundance in Equation 2, and there is.

In the present invention, X3 may be 0.001 to 0.050, for example, 0.005 to 0.040, or 0.010 to 0.030.

In the present invention, X4 may be 0.001 to 0.010, for example, 0.002 to 0.007, 0.003 to 0.0055.

In the present invention, X5 may be 0.30 to 0.80, for example, 0.35 to 0.75, or 0.45 to 0.65.

The present invention relates to a method of providing information for diagnosing ovarian cancer, including measuring CA-125 levels in the blood of a subject and calculating the probability that the subject has ovarian cancer according to Equation 3 below.

[Equation 3]

Y3 = X6 * subject age + X7 * blood CA-125 level.

In the present invention, the step of calculating the probability that a subject has ovarian cancer can be calculated using a mathematical formula according to variables. For example, when patient age and CA-125 levels are used as variables, it can be calculated using Equation 3 above.

In the present invention, each variable in Equation 3 can be calculated as a value multiplied by each weight, depending on the weight it occupies in Equation 3. And the sum of the values of each weight may be between 0 and 1 (eg, greater than 0 and less than or equal to 1).

For example, X6 may be the weight for patient age in Equation 3, and X7 may be the weight for blood CA-125 in Equation 3.

In the present invention, X6 may be 0.001 to 0.10, for example, 0.001 to 0.050, or 0.010 to 0.035.

In the present invention, X7 may be 0.001 to 0.010, for example, 0.0015 to 0.007, or 0.002 to 0.005.

The present invention relates to an ovarian cancer diagnostic device, including a calculation unit that calculates the probability that a subject has ovarian cancer according to Equation 2 below.

[Equation 2]

Y2 = X3 * Patient age + X4 * Blood CA-125 level + X5 * Acinetobacter relative abundance.

(X3 is the weight for the patient's age in Equation 2, X4 is the weight for the relative abundance of Acinetobacter in Equation 2, X5 is the weight for the blood CA-125 in Equation 2, The sum of X5 is between 0 and 1 (e.g., greater than 0 and less than or equal to 1).

In the present invention, the diagnostic device is not limited as long as it includes an operation unit that calculates by introducing mathematical equations.

For other details, please refer to the information provided in the information provision method above.

Hereinafter, the present invention will be described in detail using examples and experimental examples to specifically explain the present invention.

1. Materials and Methods

1.1. Study Population

Biological samples were collected from patients scheduled to undergo surgery for an adnexal mass. The patient's blood samples and cancer tissue were collected the day before and at the time of surgery, respectively, and stored in the Seoul National University Human Biobank.

Relevant patients were identified and frozen blood samples were obtained from the Human Biobank. Inclusion criteria for the study population were as follows:

(1) 18 years of age or older; (2) underwent surgery for an adnexal mass between June 2012 and February 2018; and (3) pathologically diagnosed as epithelial OC or benign ovarian tumor.

However, patients with the following conditions were excluded.

(A) Diagnosed with a malignancy other than OC concurrently or prior to surgery; (B) received preoperative adjuvant chemotherapy or targeted therapy; (C) Corresponds to borderline ovarian tumors; and (D) have severe comorbidities such as end-stage renal disease, uncontrolled diabetes mellitus, or long-term corticosteroid use.

A total of 166 OC patients and 76 benign ovarian tumor patients were selected based on the above criteria. Medical records were reviewed to collect baseline characteristics, including age at diagnosis, BMI, comorbidities, and initial blood CA-125 level. We also reviewed the pathology results of all patients and collected information on FIGO stage for the study group. Afterwards, the clinicopathological characteristics of the OC group and the benign ovarian tumor group were compared. Metagenomic profiling was performed on frozen blood samples from patients according to the procedures described below.

1.2. EV isolation and DNA extraction from blood samples

Extracellular vesicles were isolated from blood samples using a differential centrifugation method. Blood samples were centrifuged at 3000 rpm for 15 min at 4°C, and 100 μL of supernatant was mixed with 1 PBS, pH 7.4 (ML 008-01, Welgene, Korea). Suspended particles were settled by centrifugation at 10,000 After centrifugation, the supernatant was sterilized through a 0.22-μm filter to completely remove bacteria and foreign substances.

To extract DNA from the membrane of extracellular vesicles, EVs isolated from blood in the previous step were boiled at 100°C for 40 minutes. The supernatant was collected after centrifugation at 13,000 rpm for 30 minutes at 4°C to remove remaining suspended particles and debris. DNA from EVs was extracted using a DNA isolation kit according to standard protocols (PowerSoil DNA Isolation Kit, MO BIO, Carlsbad, CA, USA). The DNA of EVs in each sample was quantified using the QIAxpert system (QIAGEN, Hilden, Germany).

1.3. Microbial Metagenomic Analysis

For 16S rDNA gene-based metagenome analysis, bacterial genomic DNA was subjected to primers specific for the V3-V4 hypervariable regions of the 16S rDNA gene, 16S_V3_f (5'-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG-3', SEQ ID NO: 1) and 16S_V4_r ( It was amplified using 5'-GTCTCGTGGGCTCGGAGATGTGTATAAGAG ACAGGACTACHVGGGTATCTAATCC-3', SEQ ID NO: 2).

Libraries were prepared using PCR products according to the MiSeq system guide (Illumina, San Diego, CA, USA) and quantified using QIAxpert (QIAGEN, Hilden, Germany). Each amplicon was then quantified, set to equimolar ratio, pooled, and sequenced on a MiSeq (Illumina, San Diego, CA, USA) according to the manufacturer's recommendations.

1.4. Analysis of Microbial Composition in the Microbiota

Paired-end reads matching adapter sequences were trimmed by Cutadapt (version 1.1.6). The resulting FASTQ files containing paired-end reads were merged with CASPER and then quality filtered using the Phred (Q) score-based criteria described by Bokulich. After merging, reads shorter than 300bp were removed. To identify chimeric sequences, a reference-based chimera detection step was performed using VSEARCH against the Greengenes database.

Next, sequence reads were clustered into operational taxonomic units (OTUs) using CD-HIT with a de novo clustering algorithm, under a threshold of 97% sequence similarity.

Representative sequences of OTUs were finally classified using the Greengenes database (version 13.8) with UCLUST (parallel_assign_taxonomy_uclust.py script in QIIME (version 1.9.1) under default parameters). The Chao index, which estimates the abundance of individual taxa, was estimated to measure the diversity of each sample.

1.5. Development of the Jindal model for ovarian cancer (OC)

To construct and validate the diagnostic model for OC, samples from each group were randomly divided into training set and test set at a ratio of 2:1, considering the ratio of OC and benign ovarian tumors in a total of 242 samples. The values of each training set and test set were converted to a median log ratio. Discovery of microbiome biomarkers and construction of diagnostic models were performed on the training set (n = 161), and validation of the newly developed diagnostic model was performed on the test set (n = 81).

If the zero rate was greater than 99%, the genus was filtered. Eight widely used statistical methods (Wilcoxon, Metastats, EdgeR, DESeq2, ZIG, ZIBSeq, ANCOM, and CLR) with filtered count data to identify specific microbiome biomarkers differentially distributed between OC and benign ovarian tumor groups. Meta-genome analysis was performed using Perm.

Because the algorithm was developed based on abundant data, abundant OTUs were used. We compared the list of significant microbiome biomarkers identified by each statistical method to determine the extent to which overlapping biomarkers were possible, since each method provides a different list of microbiome biomarkers and most overlapping biomarkers are expected to be highly plausible biomarkers. A marker has been selected.

We combined microbiome biomarkers with patient age and blood CA-125 levels to construct several diagnostic models to identify OC in benign ovarian tumors, and these models were validated on a test set. To evaluate the diagnostic performance of the developed models, the sensitivity, specificity, and AUC of each model were calculated.

1.6. statistical analysis

Statistical analysis was performed to evaluate differences in clinicopathological characteristics between the two groups. Student's t-test and Mann-Whitney U test were used to compare continuous variables, and Pearson's chi-square test and Fisher's exact test were used to compare categorical variables. The Shannon index was calculated to measure the α-diversity of microbiota.

A summary of the eight statistical methods applied to metagenome analysis is as follows:

(1) The Wilcoxon rank sum test is a nonparametric type of 2-sample t-test that uses rank sum for observations.

(2) Metastats compares the number of samples per group and the number of taxaons. When the number of taxa was greater than the number of samples, Welch's t-test statistics were applied. Otherwise, Fisher's exact test was used.

(3) EdgeR and (4) DESeq2 methods are commonly used for RNA sequence data analysis. Application of these methods is frequently attempted as metagenome data is extracted from 16S rDNA. Both methods are negative binomial models. However, the difference between the two methods is that EdgeR uses a trimmed mean of M-values normalization, while DESeq2 uses a relative log expression normalization.

(5) ZIG uses a lognormal mixture model for the number of taxa considering the sparsity of the OTU table. To overcome high false-positive rates, experimental Bayes shrinkage of parameter estimates was adopted.

(6) ZIBSeq uses a beta mixing model for relative abundance. Relative abundance after sum scaling normalization was performed due to the large number of zeros and resulting in a skewed distribution.

(7) ANCOM was used to compare the relative abundance of OTUs, and the Wilcoxon rank sum test was used for comparison of two groups after log-ratio transformation of all pairwise taxa. The Kruskal-Wallis test was used for comparison of three groups, and the Freidman test was used for comparison of repeated data.

(8) CLR Perm uses count data to fit a logistic model after the centered log-ratio transformation, thereby relaxing the sum into a single constraint of relative abundance. A permutation test was adopted to reduce the false discovery rate (FDR).

R statistical software (version 3.4.4; R Foundation for Statistical Computing, Vienna, Austria; ISBN 3-900051-07-0; http://www.R-project.org) was used for statistical analysis. A two-sided p-value of less than 0.05 was considered statistically significant.

2. Results

2.1. Characteristics of the study population

The overall study design is shown in Figure 1. Table 1 below shows the clinicopathological characteristics of all patients. Patients in the OC group were significantly older than those in the benign ovarian tumor group (mean patient in OC group: 53.6 years and mean in benign ovarian tumor group: 49.4 years; p = 0.041), but had lower body mass index (BMI), menopausal status, and Other characteristics, such as comorbidities, were similar. Even after a 2:1 random distribution of patients into training and test sets, OC patients were still older than patients with benign ovarian tumors in the training set, whereas patients were similar in age in the test set. No differences in BMI, menopausal status, and comorbidities were observed between the OC and benign ovarian tumor groups in both the training and testing sets.

(BMI: Body Mass Index, CA-125: Cancer antigen 125, FIGO: International Federation of Gynecology and Obstetrics, SD: Standard deviation.)

In the training set, blood CA-125 levels were significantly higher in OC patients (median: 331.1 IU/mL in the OC patient group and median: 22.3 IU/mL in the benign ovarian tumor patient group; p < 0.001) compared to 110 OC patients. Of these, 39 (35.5%) and 71 (64.5%) were diagnosed with International Federation of Gynecology and Obstetrics (FIGO) stages I-II and III-IV, respectively. The most common histologic type was high-grade serous carcinoma. -grade serous carcinoma) was observed in 54.5% of OC patients. Among 51 patients with benign ovarian tumors, mucinous cystadenoma was the most common pathological diagnosis at 47.1%, followed by serous cystadenoma. This was followed by 15.7%.

In the test set, the OC group showed significantly higher blood CA-125 levels compared to the benign ovarian tumor group (median for patients in the OC group: 432.3 IU/mL and median for patients in the benign ovarian tumor group: 20.6 IU/mL; p < 0.001). FIGO stage I-II disease was observed in 41.1% of OC patients. The most common histological types in the OC and benign ovarian tumor groups were high-grade serous carcinoma (50.0%) and serous cystoma (28.0%), respectively.

2.2. Comparison of Metagenomic Profiles between the Two Groups

Figure 2 shows the metagenomic profile of all patients. Figure 2(A) shows all 31 phylum-levels detected in the OC and benign ovarian tumor groups. In genus-level organization, a total of 587 genera were detected in all patients. Among them, 110 significantly differently distributed genera identified by at least two statistical methods showed their relative abundances in Figure 2(B). Here, the genus Acinetobacter showed high relative abundance in both OC and benign ovarian tumor groups. In the training set, 107 of 110 (97.3%) patients with ovarian cancer had Acinetobacter, and 50 of 51 (98.0%) patients with benign ovarian tumors had Acinetobacter. In the test set, Acinetobacter was found in 98.2% (55/56) and 100.0% (25/25) of the OC and benign ovarian tumor groups, respectively.

When comparing genus-level α-diversity, the Shannon index was significantly higher for the training set (median: 3.294 vs. 3.263, p = 0.270) and test set (median, 3.210 vs. 3.238; p = 0.810) (Figure 3). reference).

Clustering was analyzed at the genus level using multidimensional plots to compare β-diversity. However, these plots did not reveal distinct clustering between the OC group and the benign ovarian tumor group in the training and testing sets (see Figure 4).

2.3. Developing an ovarian cancer diagnostic model from a training set

Statistically significant genus-level microbiome biomarkers were identified that differed between the OC group and the benign ovarian tumor group through metagenomic analysis using various statistical methods: Wilcoxon test, Metastats, EdgeR, DESeq2, zero-inflated. Gaussian mix model (ZIG), zero-inflated beta regression (ZIBSeq), analysis of composition of microbiomes (ANCOM), centered log-ratio transformation, and permutation logistic regression model (CLR Perm) were 1, 98, 3, 8, and 447, respectively. , 56, 1 and 2 biomarkers were identified with adjusted q values using FDR (≤0.05). Table 2 presents the top 10 microbiome biomarkers identified by each statistical method, in order of overlap.

Gene Wilcoxon Metastats EdgeR DESeq2 ZIG ZIBSeq ANCOM CLR Perm Acinetobacter <0.001 0.008 0.093 0.043 <0.001 <0.001 Acinetobacter <0.001 Isoptericola 0.841 <0.001 0.487 One 0.002 <0.001 Not detected 0.855 Terrisporobacter 0.841 0.023 0.600 One <0.001 <0.001 Not detected 0.944 SM1A02 0.989 0.008 0.528 One <0.001 0.002 Not detected 0.935 Candidatus Alysiosphaera 0.841 <0.001 0.476 One 0.005 <0.001 Not detected 0.901 Ralstonia 0.841 <0.001 0.462 One <0.001 0.005 Not detected 0.913 Hydrogenophaga 0.771 <0.001 0.872 One <0.001 0.027 Not detected 0.809 Pseudorhodoferax 0.921 0.024 0.811 One <0.001 0.007 Not detected 0.779 Bryobacter 0.841 0.023 0.420 One 0.007 0.999 Not detected 0.849 Varibaculum 0.841 0.013 0.600 One 0.599 <0.001 Not detected 0.416

(Table 2 shows q value.)

Next, we examined the overlap of these genus-level microbiome biomarkers among eight statistical methods (see Figure 5). A total of 486 biomarkers were confirmed to be significantly differentially distributed by at least one statistical method. Of these, 110 and 9 markers overlapped in at least 2 and 3 statistical methods, respectively. Acinetobacter was the only common genus identified by overlapping seven statistical analysis methods. In particular, Acinetobacter was significantly more abundant in the OC group than in the benign ovarian tumor group (median (interquartile range): 0.084 (0.037-0.222) vs. 0.033 (0.008-0.075); Wilcoxon rank sum test, p <0.001) . Therefore, we selected Acinetobacter as the most likely genus-level microbial biomarker to distinguish OC from benign ovarian tumors.

By combining the relative abundance of Acinetobacter and patients' clinicopathological variables, several diagnostic models were constructed to distinguish OC from benign ovarian tumors. The model equation and its weights are as follows.

(1) Patient age + Acinetobacter relative abundance

[Equation 4]

y = 0.0282 * patient age + 0.4757 * Acinetobacter relative abundance

2) Patient age + Acinetobacter relative abundance + CA-125

[Equation 5]

y = 0.0209 * patient age + 0.0043 * CA-125 + 0.5299 * Acinetobacter relative abundance

3) Patient age + CA-125

[Equation 6]

y = 0.0225 * patient age + 0.003 * CA-125

A model consisting of patient age, blood CA-125 level and relative abundance of Acinetobacter (see (2) above) gave 86.4% sensitivity and 78.4% specificity (see Table 3). In addition, model (2) above had an AUC of 0.898, which was superior to other models.

2.4. Validation of ovarian cancer diagnostic model

The developed diagnostic model was validated on a test set. Among various models, model (2) above, which consists of patient age, initial blood CA-125 level, and relative Acinetobacter relative abundance, showed the best diagnostic performance in distinguishing OC from benign ovarian tumors as follows: sensitivity 82.1% ; Specificity 68.0%; and AUC 0.846 (Table 3 and Figure 6).

<110> SEOUL NATIONAL UNIVERSITY R&DB FOUNDATION SEOUL NATIONAL UNIVERSITY HOSPITAL <120> Metagenomic biomarker for diagnosis of ovarian cancer <130> 20P11038 <160> 2 <170> KoPatentIn 3.0 <210> 1 <211> 50 <212> DNA <213> Artificial Sequence <220> <223> artificial primer <400> 1 tcgtcggcag cgtcagatgt gtataagaga cagcctacgg gnggcwgcag 50 <210> 2 <211> 55 <212> DNA <213> Artificial Sequence <220> <223> artificial primer <400> 2 gtctcgtggg ctcggagatg tgtataagag acaggactac hvgggtatct aatcc 55

Claims (2)

measuring CA-125 levels in a sample from the subject; and
A method of providing information for diagnosing ovarian cancer, including calculating the probability that the subject has ovarian cancer according to Equation 3 below.
[Equation 3]
Y3 = X6 * subject age + X7 * sample CA-125 level.
(X6 is the weight for patient age in Equation 3, X7 is the weight for sample CA-125 in Equation 3, and the sum of X6 and X7 is between 0 and 1.)
The method of claim 1, wherein X6 is 0.010 to 0.035 and X7 is 0.002 to 0.005.