Disease discerning method, disease discerning device, disease discerning system, program, and recording medium

Abstract

A technique according to the present disclosure discerns the presence or absence of the affection of a cancer disease in a subject to be examined, by inputting both of acquired cancer-related small RNA data and cancer-unrelated small RNA data, and tumor marker data into a trained learning model that is generated by performing machine learning using teacher data including: a measurement data set group composed of cancer-related small RNA measurement data that is the result of measuring the expression levels of a plurality of cancer-related small RNAs, cancer-unrelated small RNA measurement data that is the result of measuring the expression levels of a plurality of cancer-unrelated small RNAs, and tumor marker measurement data that is the result of measuring the abundance of tumor markers; and information indicating the presence or absence of the affection of a cancer disease in each of the plurality of subjects.

Description

Disease discrimination method, disease discrimination device, disease discrimination system, program, and recording medium

[0001] The present disclosure relates to a disease discrimination method, a disease discrimination device, a disease discrimination system, a program, and a recording medium.

[0002] It is known that the expression levels of specific microRNAs (miRNAs) among multiple types of miRNAs increase or decrease in the blood of patients suffering from cancer diseases such as pancreatic cancer. It is also known that the presence of tumor markers such as CA19-9 (carbohydrate antigen 19-9) and CEA (carcinoembryonic antigen) increases or decreases in the blood of patients suffering from cancer diseases.

[0003] Therefore, blood is collected from a test subject to be determined whether or not the subject has cancer, and the amount of tumor markers present in the collected blood and the expression level of miRNAs that are significantly increased or decreased due to cancer are measured to determine whether or not the subject has cancer (see, for example, Non-Patent Documents 1 and 2 below).

[0004] Furthermore, Non-Patent Document 3 below discloses a method for determining the presence or absence of a disease in a subject to be examined, without using the abundance of a tumor marker, by using a trained model obtained by machine learning using the expression levels of 181 types of miRNAs that are stably expressed in collected blood as training data.

[0005] 1:Nicolai A. Schultz, Christian Dehlendorff, Benny V. Jensen, Jon K. Bjerregaard, Kaspar R. Nielsen, Stig E. Bojesen, Dan Calatayud, Svend E. Nielsen, Mette Yilmaz, Niels Henrik Hollander, Klaus K. Andersen, Julia S. Johansen, "MicroRNA Biomarkers in Whole Blood for Detection of Pancreatic Cancer," JAMA(Journal of American Medical Association), 2014 January 22 / 29, Part 311, Part 4, p.392-404 Actress 2: Kota Nakamura, Zhongxu Zhu, Souvick Roy, Eunsung Jun, Haiyong Han, Ruben M. Munoz, Satoshi Nishiwada, Geeta Sharma, Derek Cridebring, Frederic Zenhausern, Seungchan Kim, Denise J. Roe, Sourat Darabi, In-Woong Han, Douglas , Suguru Yamada , Michael J. Demeure , Carlos Becerra , Scott A. Celinski , Erkut Borazanci , Susan Tsai , Yasuhiro Kodera , Joon Oh Park , John S.Bolton, Xin Wang, Song Cheol Kim, Daniel Von Hoff, Ajay Goel, "An Exosome-based Transcriptomic Signature for Noninvasive, Early Detection of Patients With Pancreatic Ductal Adenocarcinoma: A Multicenter Cohort Study", Gastroenterology, November 2022, Vol. 163, No. 5, pp. 1252-1266. Non-patent literature 3: Masayasu Inagaki, Makoto Uchiyama, Kanae Yoshikawa-Kawabe, Masafumi Ito, Hideki Murakami, Masaharu Gunji, Makoto Minoshima, Takashi Kohnoh, Ryota Ito, Yuta Kodama, Mari Tanaka-Sakai, Atsushi Nakase1, Nozomi Goto1, Yusuke Tsushima, Shoich Mori, Masahiro Kozuka, Ryo Otomo, Mitsuharu Hirai, Masahiko Fujino, Toshihiko Yokoyama, "Comprehensive circulating microRNA profile as a supersensitive "Biomarker for early-stage lung cancer screening," Journal of Cancer Research and Clinical Oncology [online], published April 19, 2023, Internet<URL:https: / / doi.org / 10.1007 / s00432-023-04728-9> .

[0006] However, when determining whether or not a test subject has cancer, there is a problem if the test subject is determined to have cancer when he or she does not have cancer, or if the test subject is determined to not have cancer when he or she does have cancer. Therefore, a higher level of accuracy is required for determining whether or not a test subject has cancer.

[0007] The object of the present disclosure is to provide a disease discrimination method, a disease discrimination device, a disease discrimination system, a program, and a recording medium that can achieve higher accuracy of discrimination compared to when determining whether or not a test subject has cancer using small RNAs such as miRNAs whose expression levels significantly increase or decrease due to the onset of cancer and tumor markers.

[0008] A disease discrimination method according to one aspect of the present disclosure includes an acquisition step of measuring a biological sample of a test subject for determining the presence or absence of cancer, thereby acquiring small RNA data, which is the result of measuring the expression levels of each of a plurality of small RNAs in the biological sample of the test subject, and tumor marker data, which is the result of measuring the abundance, in the biological sample of the test subject, of tumor markers, which are substances other than small RNAs that increase or decrease in the body due to cancer; and cancer-associated small RNA measurement data, which are obtained by measuring biological samples collected from a plurality of subjects including subjects with cancer and subjects without the cancer, and which are the result of measuring the expression levels of a plurality of cancer-associated small RNAs whose expression levels are significantly increased or decreased in the biological sample of a patient with cancer compared to the biological sample of a subject without cancer, and cancer-unassociated small RNA measurement data, which are the result of measuring the expression levels of a plurality of cancer-unassociated small RNAs whose expression levels are not significantly increased or decreased in the biological sample of a patient with cancer compared to the biological sample of a subject without cancer. and a discrimination step of discriminating whether or not the test subject is affected by cancer by inputting both cancer-associated small RNA data and cancer-unassociated small RNA data from the small RNA data acquired in the acquisition step, and the tumor marker data acquired in the acquisition step, into a trained learning model generated by machine learning using teacher data consisting of a measurement dataset group consisting of small RNA measurement data and tumor marker measurement data resulting from measuring the amount of tumor marker present, and information indicating whether or not each of the multiple subjects is affected by cancer.

[0009] According to the technology disclosed herein, it is possible to achieve higher accuracy of discrimination compared to when determining whether or not a subject has cancer using small RNAs and tumor markers whose expression levels significantly increase or decrease due to cancer.

[0010] 9 is a diagram showing the system configuration of a disease discrimination system 20 according to an embodiment of the present disclosure. FIG. 10 is a block diagram showing the hardware configuration of a disease discrimination device 30. FIG. 11 is a block diagram showing the functional configuration of the disease discrimination device 30 realized by executing a discrimination program. FIG. 12 is a flowchart showing a method for generating a learning model 130. FIG. 13 is a diagram showing the machine learning process of the learning model 130. FIG. 14 is a flowchart showing the processing when determining the presence or absence of cancer using a learning model 130 that has undergone machine learning. FIG. 14 is a diagram showing the process of discriminating between the presence or absence of cancer. FIG. 15 is a diagram showing the grouping of miRNAs whose expression has been confirmed in blood collected from a plurality of subjects. FIG. 16 is a diagram for explaining eight patterns of data combinations of miRNA expression level data, and abundance data of tumor markers CA19-9 and CEA. FIG. 17 is a diagram showing the discrimination results obtained by discriminating between the presence or absence of cancer for each of the eight patterns of data combinations shown in FIG.

[0011] Next, embodiments of the present disclosure will be described in detail with reference to the drawings.

[0012] An example of an embodiment of the technology of the present disclosure will be described below with reference to the drawings. Note that components and processes that perform the same operations, actions, and functions are given the same reference numerals throughout the drawings, and redundant explanations may be omitted as appropriate. Each drawing is merely a schematic illustration to allow a sufficient understanding of the technology of the present disclosure. Therefore, the technology of the present disclosure is not limited to the illustrated examples. Furthermore, in this embodiment, explanations of configurations that are not directly related to the present disclosure or well-known configurations may be omitted.

[0013] FIG. 1 is a diagram showing the system configuration of a disease discrimination system 20 according to an embodiment of the present disclosure.

[0014] As shown in FIG. 1, the disease discrimination system 20 of this embodiment is composed of a next-generation sequencer 21 using NGS (Next-Generation Sequencing) technology, a tumor marker measurement device 22, and a disease discrimination device 30.

[0015] In addition to the next-generation sequencer 21, a quantitative PCR (Polymerase Chain Reaction) or a flow cytometer may also be used as a measuring device as long as it can measure the expression levels of multiple nucleic acid molecules.

[0016] The disease discrimination system 20 of this embodiment uses blood 40 collected from a test subject for which the presence or absence of cancer is to be determined, to determine whether the test subject has cancer.

[0017] Although this embodiment describes a case where blood collected from a test subject is used to determine the presence or absence of cancer, the presence or absence of cancer may also be determined using biological samples other than blood, such as body fluids, cells, extracellular vesicles, tissue fragments, etc. Here, body fluids include, for example, serum, urine, tears, saliva, sweat, semen, lymph, tissue fluid, body cavity fluid (e.g., pleural effusion, ascites, etc.), cerebrospinal fluid, amniotic fluid, vaginal fluid, nasal mucus, etc.; cells include, for example, red blood cells, white blood cells, platelets, oral swabs, etc.; extracellular vesicles include, for example, exosomes, liposomes, etc.; and tissue fragments include, for example, FFPE (Formalin Fixed Paraffin Embedded) specimens, biopsy specimens, frozen specimens, etc.

[0018] Furthermore, in this embodiment, the test subject is described as a human, but the test subject is not limited to humans, and the technology of the present disclosure can be similarly applied to cases where non-human animals other than humans are used as test subjects. Examples of non-human animals include non-human mammals (monkeys, dogs, cats, mice, rats, rabbits, cows, horses, pigs, sheep, etc.), birds (chickens, quails, etc.), etc. In other words, the technology of the present disclosure can be applied to cases where a biological sample such as blood collected from a non-human animal is used to test for the presence or absence of cancer or other diseases.

[0019] The next-generation sequencer 21 measures the expression levels of multiple small RNAs contained in blood 40 collected from a test subject. Specifically, the next-generation sequencer 21 performs an amplification process to amplify small RNAs extracted from blood 40, and then measures the relative expression level of each type of small RNA relative to other small RNAs.

[0020] Here, examples of small RNA include microRNA (hereinafter referred to as miRNA). Small RNA may also be small RNA other than miRNA (e.g., piRNA and tsRNA). The following description will be given using a case where the expression level of miRNA is measured from the blood 40 of a test subject to determine whether or not the subject is suffering from cancer.

[0021] Then, the next-generation sequencer 21 outputs the measured data on the expression levels of miRNA to the disease discrimination device 30 as miRNA data.

[0022] The tumor marker measurement device 22 also measures the amount of tumor marker present in the blood 40 of the test subject. In this embodiment, the case where the cancer disease to be tested is pancreatic cancer will be described. Therefore, the tumor marker measurement device 22 measures the amount of either or both of CA19-9 (carbohydrate antigen 19-9) and CEA (carcinoembryonic antigen) present in the blood as tumor markers.

[0023] In this embodiment, the configuration in which tumor marker data is acquired by tumor marker measurement device 22 has been described, but the present invention is not limited to this configuration. Some tumor markers can be measured in quantity by immunochromatography, which determines the amount by colorimetry using test paper or the like, without measuring with a device. Therefore, any means other than tumor marker measurement device 22 that can acquire tumor marker data may be used.

[0024] Here, a tumor marker is a substance other than small RNA such as miRNA that increases or decreases in the body due to the onset of cancer.

[0025] Then, the tumor marker measurement device 22 outputs data on the amount of the measured tumor marker present to the disease discrimination device 30 as tumor marker data.

[0026] The disease discrimination device 30 uses the miRNA data from the next-generation sequencer 21 and the tumor marker data from the tumor marker measurement device 22 to determine whether or not the subject is suffering from cancer, and displays the discrimination result.

[0027] Next, the hardware configuration of the disease discrimination apparatus 30 described above is shown in the block diagram of FIG.

[0028] The disease discrimination device 30 has the function of a computer, and as shown in Fig. 2, includes a CPU (Central Processing Unit) 31, a ROM (Read Only Memory) 32, a RAM (Random Access Memory) 33, a storage 34, an input unit 35, a display unit 36, and a communication interface (I / F) 37. The components are connected to each other via a bus 39 so as to be able to communicate with each other.

[0029] The CPU 31 (an example of a processor) is a central processing unit that executes various programs and controls each part. That is, the CPU 31 reads a program from the ROM 32 or the storage 34 and executes the program using the RAM 33 as a work area. The CPU 31 controls each of the above components and performs various arithmetic processing in accordance with the program stored in the ROM 32 or the storage 34.

[0030] The ROM 32 stores various programs and various data. The RAM 33 temporarily stores programs or data as a working area. The storage 34 is configured by an HDD (Hard Disk Drive) or an SSD (Solid State Drive) and stores various programs including the operating system and various data.

[0031] In this embodiment, for example, a discrimination program for executing a disease discrimination method for discriminating the presence or absence of cancer is recorded in the storage 34. This discrimination program may be a single program, or a group of programs configured by multiple programs or modules. The discrimination program may be recorded in the ROM 32. The ROM 32 and the storage 34 function as an example of a non-transitory recording medium.

[0032] An example of a processor is not limited to the above-mentioned CPU, which is a general-purpose processor, but may be, for example, a dedicated processor configured with a circuit designed specifically for executing a specific process. Also, an example of a processor is not limited to a single processor, but may be a processor configured by multiple processors located in physically separate locations working together.

[0033] The input unit 35 includes a pointing device such as a mouse and a keyboard, and is used to perform various inputs. The input unit 35 also receives as inputs miRNA data representing the expression levels of multiple miRNAs measured by the next-generation sequencer 21 and tumor marker data measured by the tumor marker measurement device 22.

[0034] The display unit 36 ​​is, for example, a liquid crystal display, and displays various information. The disease discrimination device 30 can present the discrimination result of the presence or absence of cancer disease, which is determined using the miRNA data and the tumor marker data, to the user through the display unit 36. The display unit 36 ​​may employ a touch panel system and function as the input unit 35 for inputting operations from the user.

[0035] The communication interface 37 is an interface for communicating with other devices, and transmits and receives data to and from external devices including the next-generation sequencer 21 and tumor marker measurement device 22 .

[0036] Next, the functional configuration of the disease discrimination device 30 realized by executing the discrimination program described above is shown in the block diagram of FIG.

[0037] As shown in FIG. 3, in the disease discrimination device 30, an acquisition unit 110, a disease presence / absence discrimination unit 120, and a learning model 130 are configured by a CPU 31 executing a discrimination program.

[0038] The acquisition unit 110 measures the blood of a test subject to determine whether or not the subject is suffering from a cancer disease, and acquires small RNA data, which is the result of measuring the expression levels of each of multiple miRNAs in the blood of the test subject, and tumor marker data, which is the result of measuring the amount of tumor markers present in the blood of the test subject.

[0039] Although the expression levels of multiple miRNAs in blood are absolute values ​​derived from the living body, it is difficult to quantify the expression levels of miRNAs in blood as absolute values ​​because the expression levels must be quantified through processes such as measurement devices and reagent processing. In other words, small RNA data is a relative representation of the expression levels of miRNAs obtained by using a next-generation sequencer 21 or data processing. However, if the expression levels of miRNAs in blood can be quantified as absolute values, the absolute values ​​of the expression levels of miRNAs in blood may be used as small RNA data.

[0040] As mentioned above, in this embodiment, the case where the cancer disease to be examined is pancreatic cancer will be described. Therefore, the acquisition unit 110 uses either one or both of CA19-9 and CEA as tumor markers, and acquires the amount of this tumor marker present in the blood as tumor marker data. If the type of cancer disease to be examined is lung cancer, stomach cancer, or the like other than pancreatic cancer, the amount of tumor marker present in the blood corresponding to the type of cancer disease to be examined will be acquired as tumor marker data.

[0041] In the disease discrimination device 30 of this embodiment, a learning model 130 for discriminating the presence or absence of cancer disease is generated in advance.

[0042] The learning model 130 is a trained learning model generated by machine learning using training data composed of a group of measurement datasets, including cancer-related miRNA measurement data, which is the result of measuring the expression levels of multiple cancer-related miRNAs whose expression levels are significantly increased or decreased in the blood of patients with cancer compared to the blood of subjects without cancer, cancer-non-related miRNA measurement data, which is the result of measuring the expression levels of multiple cancer-related miRNAs whose expression levels are not significantly increased or decreased in the blood of patients with cancer compared to the blood of subjects without cancer, and tumor marker measurement data, which is the result of measuring the abundance of tumor markers, obtained by measuring blood collected from multiple subjects including subjects with cancer and subjects without cancer.

[0043] The disease presence / absence determination unit 120 then inputs both the cancer-related miRNA data and the cancer-unrelated miRNA data from the miRNA data acquired by the acquisition unit 110, as well as the tumor marker data acquired by the acquisition unit 110, into the learning model 130, thereby determining whether the subject is suffering from cancer.

[0044] In addition, the disease presence / absence determination unit 120 uses, as multiple cancer-related miRNAs or multiple cancer-unrelated miRNAs, miRNAs that are within an arbitrary rank set from the top when the miRNAs are arranged in order of highest expression level, for example, within the top 100, among multiple miRNAs whose expression has been confirmed in the blood of any of the multiple subjects.

[0045] Then, in the disease discrimination device 30, the presence or absence of cancer in the test subject determined by the disease presence / absence determining unit 120 is displayed on the display unit 36.

[0046] Next, a method for generating the learning model 130 described above is shown in the flowchart of FIG.

[0047] First, in the collection step of step S101, blood is collected from a subject and centrifuged, and the resulting serum is dispensed into storage tubes in predetermined amounts and stored in a deep freezer at −80° C. The subjects include both healthy subjects not suffering from cancer and patients suffering from pancreatic cancer.

[0048] Next, in the measurement step of step S102, the frozen serum is removed from the deep freezer and thawed at room temperature, and then the expression levels of miRNA are measured using the next-generation sequencer 21. Also, in the measurement step of step S102, the frozen serum is removed from the deep freezer and thawed at room temperature, and then the amounts of tumor markers CA19-9 and CEA present are measured using the tumor marker measurement device 22.

[0049] Then, in the acquisition step of step S103, the disease discrimination device 30 acquires, as miRNA data, the results of measuring the expression levels of each of the multiple miRNAs in the serum of the subject measured in the measurement step of step S102. Also, in the acquisition step of step S103, the disease discrimination device 30 acquires, as tumor marker data, the results of measuring the abundance of a tumor marker in the serum of the subject measured in the measurement step of step S102.

[0050] Finally, in the learning process of step S104, the disease discrimination device 30 performs machine learning of the learning model 130 using the acquired miRNA data and tumor marker data, and information indicating whether or not each subject has cancer, as training data. The machine learning process of such a learning model 130 is shown in FIG. 5 .

[0051] As can be seen from Figure 5, in the disease discrimination system 20 of this embodiment, machine learning of the learning model 130 is performed using not only expression level data of cancer-related miRNAs whose expression levels fluctuate significantly (increase or decrease) in the blood of patients with cancer, but also expression level data of non-cancer-related miRNAs whose expression levels do not fluctuate significantly (increase or decrease) in the blood of patients with cancer.

[0052] Here, in this embodiment, the cancer disease to be tested is described as pancreatic cancer, but the type of cancer-related miRNA also varies depending on the type of cancer to be tested. Specifically, cancer-related miRNAs whose expression levels significantly fluctuate (increase or decrease) in the blood of patients with pancreatic cancer are different from cancer-related miRNAs whose expression levels significantly fluctuate (increase or decrease) in the blood of patients with lung cancer. In other words, miRNAs whose expression levels significantly fluctuate (increase or decrease) in the blood of patients with the cancer disease of the type to be tested are cancer-related miRNAs.

[0053] <Examples of learning models> Various linear and non-linear algorithms known as machine learning algorithms can be used, or multiple algorithms can be combined. For example, the following algorithms can be used:

[0054] Random forest Gradient boosting trees Extreme gradient boosting trees Light-gradient boosting machine (GBM) Neural networks Regularized regression Elastic-net regression K-Nearest neighbors Support vector machine Generalized additive model

[0055] Next, the process of determining the presence or absence of cancer using the learning model 130 that has undergone machine learning in this manner is shown in the flowchart of FIG.

[0056] First, in the collection step of step S201, blood is collected from a test subject to be determined whether or not the subject is suffering from cancer.

[0057] Next, in the measurement step of step S202, the expression level of miRNA in the blood collected from the test subject is measured by the next-generation sequencer 21. Also in the measurement step of step S202, the amounts of CA19-9 and CEA present in the blood collected from the test subject are measured by the tumor marker measurement device 22.

[0058] Then, in the acquisition step of step S203, the disease discrimination device 30 acquires, as miRNA data, the results of measuring the expression levels of multiple miRNAs in the blood of the test subject measured in the measurement step of step S202. This miRNA data includes both expression level data of cancer-related miRNAs and expression level data of miRNAs not related to cancer. Also, in the acquisition step of step S203, the disease discrimination device 30 acquires, as tumor marker data, the results of measuring the abundance of tumor markers in the blood of the test subject measured in the measurement step of step S202.

[0059] Finally, in the disease presence / absence discrimination process of step S204, the disease discrimination device 30 inputs the acquired miRNA data (both cancer-related miRNA expression level data and cancer-unrelated miRNA expression level data) and tumor marker data into the trained learning model 130, thereby obtaining a discrimination result indicating whether or not the test subject is suffering from cancer. Such a cancer presence / absence discrimination process is shown in FIG. 7 .

[0060] As can be seen from Figure 7, in the disease discrimination system 20 of this embodiment, not only expression level data of cancer-related miRNAs whose expression levels fluctuate significantly (increase or decrease) in the blood of patients with cancer, but also expression level data of non-cancer-related miRNAs whose expression levels do not fluctuate significantly (increase or decrease) in the blood of patients with cancer, and tumor marker abundance data are input into a trained learning model 130 to obtain discrimination results.

[0061] According to the disease discrimination method of this embodiment, it is possible to achieve higher discrimination accuracy compared to when determining whether or not a subject has cancer using only miRNAs and tumor markers whose expression levels significantly increase or decrease due to the onset of cancer.

[0062] The reason why the disease discrimination method of this embodiment improves the accuracy of discriminating whether or not a patient is affected by cancer will be explained below.

[0063] Even in expression level data of non-cancer-related miRNAs whose expression levels do not fluctuate significantly in the blood of patients with cancer, it is believed that some of the expression levels actually fluctuate somewhat depending on whether or not the patient has cancer. Therefore, a learning model is trained using not only cancer-related miRNA data but also non-cancer-related miRNA data, and the accuracy of discrimination in determining whether or not the patient has cancer is improved by using not only cancer-related miRNA data but also non-cancer-related miRNA data.

[0064] In addition, it is believed that some non-cancer-related miRNAs naturally contain miRNAs whose expression levels hardly fluctuate. However, when determining whether or not a patient has a cancer disease, the determination is not based on a single miRNA, but on a combination of multiple types of miRNAs. Therefore, even if a miRNA is estimated to have little fluctuation in expression level and not contribute to the determination when viewed as a single miRNA, it is believed that there are miRNAs among the non-cancer-related miRNAs that can synergistically improve the discrimination performance when combined with another miRNA. Therefore, a learning model is trained using not only cancer-related miRNA data but also non-cancer-related miRNA data, and the accuracy of the discrimination when determining whether or not a patient has a cancer disease is improved by using not only cancer-related miRNA data but also non-cancer-related miRNA data.

[0065] Next, as an example, experimental data is presented showing that by determining whether or not a patient has cancer using both cancer-related miRNA data and non-cancer-related miRNA data, as well as tumor marker data, the accuracy of determination is higher than when determining whether or not a patient has cancer using only cancer-related miRNA data and tumor marker data.

[0066] In this experiment, blood samples were collected from a total of 400 subjects, including 213 healthy individuals and 187 pancreatic cancer patients. The miRNAs whose expression was confirmed in the collected blood samples were divided into the following groups. The grouping of the miRNAs is shown in Figure 8.

[0067] (1) miRNA Group A: 230 miRNAs whose expression was confirmed in the blood collected from a total of 400 subjects, including 213 healthy individuals and 187 pancreatic cancer patients.

[0068] (2) miRNA Group B: 101 miRNAs in miRNA Group A that were confirmed to be expressed in 50 or more reads in any blood sample (corresponding to the miRNA group with the highest expression levels)

[0069] (3) miRNA Group C: 83 miRNAs (corresponding to the cancer-related miRNA group) were determined to be significantly different at p<0.05 after Bonferroni correction for multiple testing, based on a statistical significance test for the difference in expression levels between healthy subjects and pancreatic cancer patients.

[0070] (4) miRNA Group D: 30 miRNAs that belong to both miRNA Group B and miRNA Group C, are stably expressed, and have a significant difference in expression level between healthy individuals and pancreatic cancer patients (corresponding to the cancer-related miRNA group used in the above embodiment).

[0071] Then, in the disease discrimination method of the above embodiment, the presence or absence of cancer was discriminated by combining eight patterns of data as shown in Figure 9 using miRNA expression level data and abundance data of tumor markers CA19-9 and CEA.

[0072] Specifically, (1) determining the presence or absence of cancer using only the expression level data of miRNA group B, (2) determining the presence or absence of cancer using the expression level data of miRNA group B and the tumor marker CA19-9, (3) determining the presence or absence of cancer using the expression level data of miRNA group B and the tumor marker CEA, (4) determining the presence or absence of cancer using the expression level data of miRNA group B and the tumor markers CA19-9 and CEA, and (5) determining the presence or absence of cancer using the expression level data of miRNA group D. The presence or absence of cancer was determined using eight patterns of data combinations: (1) determining the presence or absence of cancer using only the expression level data; (2) determining the presence or absence of cancer using the expression level data of miRNA group D and the tumor marker CA19-9; (3) determining the presence or absence of cancer using the expression level data of miRNA group D and the tumor marker CEA; and (4) determining the presence or absence of cancer using the expression level data of miRNA group D and the tumor markers CA19-9 and CEA.

[0073] Here, miRNA group B includes both cancer-associated miRNAs and non-cancer-associated miRNAs, and miRNA group D is a group that includes only cancer-associated miRNAs.

[0074] In this experiment, blood data from 93 out of 213 healthy individuals was used for machine learning of the learning model, and the presence or absence of cancer was determined for blood data from the remaining 120 individuals.Furthermore, blood data from 83 out of 187 pancreatic cancer patients was used for machine learning of the learning model, and the presence or absence of cancer was determined for blood data from the remaining 104 individuals.

[0075] The results of determining the presence or absence of cancer for each of the eight data combinations are shown in Figure 10. Figure 10 shows the results of calculating the AUC (Area Under the Curve) value for each of the eight patterns. Here, AUC is an evaluation index for a two-class classification model. AUC takes a value ranging from 0 to 1, and a larger AUC indicates a higher classification accuracy of the two-class classification model.

[0076] 10, even when determining the presence or absence of cancer using only miRNA, pattern (1) has higher discrimination accuracy than pattern (5). In other words, it can be seen that discrimination accuracy is higher when discrimination is performed using not only cancer-associated miRNA but also non-cancer-associated miRNA.

[0077] 10, even when the tumor marker CA19-9 and / or CEA is used in combination with miRNA to determine the presence or absence of cancer, patterns (2) to (4) each have higher discrimination accuracy than patterns (6) to (8). In other words, even when a tumor marker and miRNA are used in combination, discrimination accuracy is higher when discrimination is performed using not only cancer-associated miRNA but also non-cancer-associated miRNA.

[0078] This experiment has proven that when determining whether or not a patient has cancer based on the expression level of miRNA and the amount of tumor marker present, it is possible to more accurately determine whether or not a patient has cancer by using not only cancer-related miRNA but also non-cancer-related miRNA.

[0079] The use of a next-generation sequencer generally allows stable measurement of the expression levels of approximately 300 types of miRNA. According to the disease discrimination method of the above embodiment, it is not necessary to investigate which miRNAs, among the miRNAs whose expression levels have been measured, have a significant difference in expression level depending on whether or not a patient is affected by cancer, and to exclude miRNAs whose expression level does not differ significantly depending on whether or not a patient is affected by cancer.

[0080] The disclosure of Japanese Patent Application No. 2023-184796, filed on October 27, 2023, is incorporated herein by reference in its entirety. All documents, patent applications, and technical standards mentioned herein are incorporated herein by reference to the same extent as if each individual document, patent application, and technical standard was specifically and individually indicated to be incorporated by reference.

Claims

1. An acquisition step of acquiring small RNA data, which is a result of measuring the expression levels of each of a plurality of small RNAs in a biological sample of a test subject to be examined for the presence or absence of cancer, by measuring the biological sample of the test subject, and tumor marker data, which is a result of measuring the presence amount, in the biological sample of the test subject, of tumor markers, which are substances other than small RNAs that increase or decrease in the body due to cancer; a discrimination step of discriminating whether the test subject has a cancer disease by inputting both the cancer-associated small RNA data and the non-cancer-associated small RNA data among the small RNA data acquired in the acquisition step, and the tumor marker data acquired in the acquisition step, into a trained learning model generated by machine learning using teacher data consisting of a measurement dataset group obtained by measuring biological samples collected from a plurality of subjects including a subject with a cancer disease and a subject without the cancer disease, the measurement dataset group being composed of cancer-associated small RNA measurement data, which is a result of measuring the expression levels of a plurality of cancer-associated small RNAs whose expression levels are significantly increased or decreased in biological samples from patients with a cancer disease compared to biological samples from subjects without a cancer disease, cancer-unassociated small RNA measurement data, which is a result of measuring the expression levels of a plurality of non-cancer-associated small RNAs whose expression levels are not significantly increased or decreased in biological samples from patients with a cancer disease compared to biological samples from subjects without a cancer disease, and tumor marker measurement data, which is a result of measuring the abundance of tumor markers, and information indicating the presence or absence of cancer in each of the plurality of subjects; A disease discrimination method comprising the steps of:

2. A disease discrimination method as described in claim 1, wherein in the discrimination process, small RNAs that are within a set rank from the top when arranged in order of highest expression level among multiple small RNAs whose expression has been confirmed in biological samples from any of multiple subjects are used as the multiple cancer-associated small RNAs or the multiple cancer-unassociated small RNAs.

3. The disease discrimination method according to claim 1, wherein the cancer disease is pancreatic cancer, and the tumor marker is either or both of CA19-9 and CEA.

4. The disease discrimination method according to claim 1, wherein the biological samples of the test subject and the subject are blood of the test subject and the subject.

5. The disease discrimination method according to claim 1, wherein the small RNA is a microRNA.

6. A device comprising a processor, the processor measuring a biological sample of a test subject for determining the presence or absence of cancer, and obtaining small RNA data which is a result of measuring the expression levels of each of a plurality of small RNAs in the biological sample of the test subject, and tumor marker data which is a result of measuring the presence in the biological sample of the test subject of tumor markers which are substances other than small RNAs that increase or decrease in the biological sample due to the presence of cancer, a measurement dataset group obtained by measuring biological samples collected from a plurality of subjects including a subject with cancer and a subject without cancer, the measurement dataset group being cancer-associated small RNA measurement data which is a result of measuring the expression levels of a plurality of cancer-associated small RNAs whose expression levels are significantly increased or decreased in biological samples from patients with cancer compared to biological samples from subjects without cancer, non-cancer-associated small RNA measurement data which is a result of measuring the expression levels of a plurality of non-cancer-associated small RNAs whose expression levels are not significantly increased or decreased in biological samples from patients with cancer compared to biological samples from subjects without cancer, and tumor marker measurement data which is a result of measuring the abundance of tumor markers, and a training model generated by machine learning using teacher data composed of information indicating the presence or absence of cancer in each of the plurality of subjects, the device inputting both the cancer-associated small RNA data and the non-cancer-associated small RNA data out of the acquired small RNA data, and the acquired tumor marker data, into a trained learning model, thereby determining the presence or absence of cancer in the test subject.

7. An acquisition unit that acquires small RNA data, which is the result of measuring the expression levels of each of a plurality of small RNAs in a biological sample of a test subject to be examined for the presence or absence of cancer, by measuring the biological sample of the test subject, and tumor marker data, which is the result of measuring the presence amount in the biological sample of the test subject of tumor markers, which are substances other than small RNAs that increase or decrease in the body due to cancer, in the biological sample of the test subject; a discrimination unit that discriminates whether the test subject has a cancer disease by inputting both the cancer-associated small RNA data and the non-cancer-associated small RNA data among the small RNA data acquired by the acquisition unit, and the tumor marker data acquired by the acquisition unit, into a trained learning model generated by machine learning using teacher data composed of: cancer-associated small RNA measurement data, which is a result of measuring the expression levels of multiple cancer-associated small RNAs whose expression levels are significantly increased or decreased in the biological samples of patients with cancer compared to those in the biological samples of subjects not affected by cancer, obtained by measuring biological samples collected from multiple subjects including subjects with cancer and subjects not affected by the cancer disease; cancer-unassociated small RNA measurement data, which is a result of measuring the expression levels of multiple cancer-associated small RNAs whose expression levels are not significantly increased or decreased in the biological samples of patients with cancer compared to those in the biological samples of subjects not affected by cancer, and tumor marker measurement data, which is a result of measuring the abundance of tumor markers; and information indicating the presence or absence of cancer in each of the multiple subjects; a display unit that displays the presence or absence of cancer in the test subject determined by the determination unit.

8. An acquisition step of acquiring small RNA data, which is a result of measuring the expression levels of each of a plurality of small RNAs in the biological sample of a test subject to be examined for the presence or absence of cancer, by measuring the biological sample of the test subject, and tumor marker data, which is a result of measuring the presence amount, in the biological sample of the test subject, of tumor markers, which are substances other than small RNAs that increase or decrease in the body due to the disease of cancer; a discrimination step of discriminating whether the test subject has a cancer disease by inputting both the cancer-associated small RNA data and the non-cancer-associated small RNA data among the small RNA data acquired in the acquisition step, and the tumor marker data acquired in the acquisition step, into a trained learning model generated by machine learning using teacher data consisting of a measurement dataset group obtained by measuring biological samples collected from a plurality of subjects including a subject with a cancer disease and a subject without the cancer disease, the measurement dataset group being the result of measuring the expression levels of a plurality of cancer-associated small RNAs whose expression levels are significantly increased or decreased in biological samples from patients with a cancer disease compared to biological samples from subjects without a cancer disease, cancer-unassociated small RNA measurement data being the result of measuring the expression levels of a plurality of non-cancer-associated small RNAs whose expression levels are not significantly increased or decreased in biological samples from patients with a cancer disease compared to biological samples from subjects without a cancer disease, and tumor marker measurement data being the result of measuring the abundance of a tumor marker, and information indicating the presence or absence of a cancer disease in each of the plurality of subjects; A program for causing a computer to execute the following.

9. An acquisition step of acquiring small RNA data, which is a result of measuring the expression levels of each of a plurality of small RNAs in a biological sample of a test subject to be examined for the presence or absence of cancer, by measuring the biological sample of the test subject, and tumor marker data, which is a result of measuring the presence amount, in the biological sample of the test subject, of tumor markers, which are substances other than small RNAs that increase or decrease in the body due to the disease of cancer; a discrimination step of discriminating whether the test subject has a cancer disease by inputting both the cancer-associated small RNA data and the non-cancer-associated small RNA data among the small RNA data acquired in the acquisition step, and the tumor marker data acquired in the acquisition step, into a trained learning model generated by machine learning using teacher data consisting of a measurement dataset group obtained by measuring biological samples collected from a plurality of subjects including a subject with a cancer disease and a subject without the cancer disease, the measurement dataset group being the result of measuring the expression levels of a plurality of cancer-associated small RNAs whose expression levels are significantly increased or decreased in biological samples from patients with a cancer disease compared to biological samples from subjects without a cancer disease, cancer-unassociated small RNA measurement data being the result of measuring the expression levels of a plurality of non-cancer-associated small RNAs whose expression levels are not significantly increased or decreased in biological samples from patients with a cancer disease compared to biological samples from subjects without a cancer disease, and tumor marker measurement data being the result of measuring the abundance of a tumor marker, and information indicating the presence or absence of a cancer disease in each of the plurality of subjects; A non-transitory recording medium on which a program for causing a computer to execute the above is recorded.

Images