WO2023068220A1 - 予測方法及びバイオマーカー - Google Patents

予測方法及びバイオマーカー Download PDF

Info

Publication number
WO2023068220A1
WO2023068220A1 PCT/JP2022/038566 JP2022038566W WO2023068220A1 WO 2023068220 A1 WO2023068220 A1 WO 2023068220A1 JP 2022038566 W JP2022038566 W JP 2022038566W WO 2023068220 A1 WO2023068220 A1 WO 2023068220A1
Authority
WO
WIPO (PCT)
Prior art keywords
mir
hsa
prediction
mirna
mirnas
Prior art date
Application number
PCT/JP2022/038566
Other languages
English (en)
French (fr)
Inventor
イファン ジャン,
雄 藤田
悌 後藤
Original Assignee
株式会社Preferred Networks
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 株式会社Preferred Networks filed Critical 株式会社Preferred Networks
Publication of WO2023068220A1 publication Critical patent/WO2023068220A1/ja

Links

Images

Classifications

    • CCHEMISTRY; METALLURGY
    • C12BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12MAPPARATUS FOR ENZYMOLOGY OR MICROBIOLOGY; APPARATUS FOR CULTURING MICROORGANISMS FOR PRODUCING BIOMASS, FOR GROWING CELLS OR FOR OBTAINING FERMENTATION OR METABOLIC PRODUCTS, i.e. BIOREACTORS OR FERMENTERS
    • C12M1/00Apparatus for enzymology or microbiology
    • CCHEMISTRY; METALLURGY
    • C12BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12MAPPARATUS FOR ENZYMOLOGY OR MICROBIOLOGY; APPARATUS FOR CULTURING MICROORGANISMS FOR PRODUCING BIOMASS, FOR GROWING CELLS OR FOR OBTAINING FERMENTATION OR METABOLIC PRODUCTS, i.e. BIOREACTORS OR FERMENTERS
    • C12M1/00Apparatus for enzymology or microbiology
    • C12M1/34Measuring or testing with condition measuring or sensing means, e.g. colony counters
    • CCHEMISTRY; METALLURGY
    • C12BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12NMICROORGANISMS OR ENZYMES; COMPOSITIONS THEREOF; PROPAGATING, PRESERVING, OR MAINTAINING MICROORGANISMS; MUTATION OR GENETIC ENGINEERING; CULTURE MEDIA
    • C12N15/00Mutation or genetic engineering; DNA or RNA concerning genetic engineering, vectors, e.g. plasmids, or their isolation, preparation or purification; Use of hosts therefor
    • C12N15/09Recombinant DNA-technology
    • C12N15/11DNA or RNA fragments; Modified forms thereof; Non-coding nucleic acids having a biological activity
    • C12N15/113Non-coding nucleic acids modulating the expression of genes, e.g. antisense oligonucleotides; Antisense DNA or RNA; Triplex- forming oligonucleotides; Catalytic nucleic acids, e.g. ribozymes; Nucleic acids used in co-suppression or gene silencing
    • CCHEMISTRY; METALLURGY
    • C12BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12QMEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q1/00Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions
    • C12Q1/68Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions involving nucleic acids
    • C12Q1/6876Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes
    • C12Q1/6883Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes for diseases caused by alterations of genetic material

Definitions

  • the present disclosure relates to prediction methods and biomarkers.
  • miRNA may be used for cancer diagnosis or treatment.
  • the patient's reaction may be observed after administering the drug to the patient.
  • the patient's PFS progression Free Survival
  • Such patient response and PFS are useful tools as criteria for predicting drug efficacy.
  • the present disclosure provides a prediction method, a prediction device, and a biomarker for predicting drug effects from values based on predetermined miRNAs derived from subject samples.
  • a prediction method of the present disclosure includes a prediction step of predicting the effect of a drug on a disease based on the value for a given miRNA derived from a sample of interest.
  • the prediction result obtained by the prediction method can assist doctors' diagnosis and the like.
  • Biomarkers for use in predicting the effect of agents on the diseases of the present disclosure include hsa-miR-429, hsa-miR-200b-3p, hsa-miR-200a-3p, hsa-miR-141-3p or hsa- It contains at least one or more miRNAs selected from miR-100-5p.
  • FIG. 1 is a conceptual diagram showing a prediction device according to a first embodiment
  • FIG. FIG. 2 is a conceptual diagram showing generation of a trained model used in the prediction device of FIG. 1
  • FIG. 4 is a block diagram showing the configuration of a training device that generates trained models
  • 4 is a flow chart showing selection of miRNAs to be used in the prediction device according to the first embodiment. Selected miRNAs used for PFS prediction are shown. Graph showing the score of each miRNA.
  • FIG. 3 is a distribution diagram showing the effect of hsa-miR-429.
  • FIG. 2 is a distribution diagram showing the effect of hsa-miR-200b-3p.
  • FIG. 2 is a distribution diagram showing the effect of hsa-miR-200a-3p.
  • FIG. 1 is a conceptual diagram showing a prediction device according to a first embodiment
  • FIG. 2 is a conceptual diagram showing generation of a trained model used in the prediction device of FIG. 1
  • FIG. 2 is a distribution diagram showing the effect of hsa-miR-100-5p.
  • FIG. 2 is a distribution diagram showing the effect of hsa-miR-141-3p.
  • 1 is a block diagram showing the configuration of a prediction device according to a first embodiment;
  • FIG. 6 is a flowchart for explaining prediction processing in the prediction device according to the first embodiment;
  • It is a block diagram which shows the hardware constitutions of a training apparatus and a prediction apparatus. It is each value which shows the prediction result by the prediction apparatus which concerns on Example 1.
  • FIG. FIG. 10 is each value showing each prediction result using one type of miRNA by the prediction device according to Example 2.
  • FIG. FIG. 10 shows values showing prediction results using a plurality of miRNAs by the prediction device according to Example 2.
  • FIG. 10 shows values showing respective prediction results using four kinds of miRNAs by the prediction device according to Example 2.
  • FIG. 10 shows values showing respective prediction results using three kinds of miRNAs by the prediction device according to Example 2.
  • FIG. 10 shows values showing prediction results using two kinds of miRNAs by the prediction device according to Example 2.
  • FIG. It is a conceptual diagram which shows the prediction apparatus which concerns on 2nd Embodiment.
  • FIG. 19 is a conceptual diagram showing generation of a trained model used in the prediction device of FIG. 18; It is each value which shows the prediction result using the expression level of miRNA by the prediction apparatus based on 2nd Embodiment. It is each value which shows the prediction result using the expression level of miRNA and specific clinical information by the prediction apparatus based on 2nd Embodiment.
  • FIG. 9 is a flowchart for explaining prediction processing in a prediction device according to the second embodiment
  • FIG. 11 is a conceptual diagram showing a prediction device according to a third embodiment
  • It is a figure which shows about the identification method of a super responder, a mild responder, and a non-responder.
  • FIG. 24 is a conceptual diagram showing generation of a trained model used in the prediction device of FIG. 23
  • FIG. 10 is a graph showing p-values regarding miRNA expression levels and clinical information obtained by a prediction device according to the third embodiment.
  • FIG. It is each value which shows the prediction result using the expression level of miRNA and specific clinical information by the prediction apparatus based on 2nd Embodiment.
  • 10 is a flowchart for explaining prediction processing in a prediction device according to the third embodiment;
  • a prediction method, a prediction device, and a biomarker according to embodiments will be described below with reference to the drawings.
  • the prediction method and prediction device according to the embodiments predict "effect of drug” based on "predetermined value derived from target sample”.
  • the biomarker according to the embodiment is a type of "predetermined value derived from a target sample” used in the prediction method, prediction device, and computer program.
  • the same components are denoted by the same reference numerals, and the description thereof is omitted.
  • miRNA is microRNA.
  • microRNA expression level is measured, for example, by extracting RNA from a sample and quantifying or semi-quantifying the expression level of microRNA contained in the extracted RNA.
  • a commercially available RNA extraction kit can be used to extract RNA from the sample. It can be performed using a commercially available RNA extraction kit according to the procedure of the attached instruction manual. That is, the expression level of microRNA can be obtained by in vitro measurement.
  • microRNA expression level can be measured, for example, by commonly used methods such as DNA microarray, RNA microarray, quantitative PCR, and next-generation sequencer.
  • the measured microRNA expression level may be an absolute expression level or a relative expression level.
  • a “DNA microarray” is a base material immobilized with DNA probes having a base sequence complementary to at least part of the microRNA to be measured.
  • the extracted RNA is fluorescently labeled using a commercially available kit, and has a sequence complementary to the microRNA arranged on the DNA microarray. After hybridizing with a DNA probe, the expression level of microRNA can be measured by measuring the fluorescence intensity.
  • RNA microarray is a base material immobilized with RNA probes having a base sequence complementary to at least part of the microRNA to be measured.
  • the extracted RNA is fluorescently labeled using a commercially available kit, and has a sequence complementary to the microRNA arranged on the RNA microarray. After hybridizing with an RNA probe, the expression level of microRNA can be measured by measuring the fluorescence intensity.
  • RNA expression level measured by quantitative PCR may be, for example, an absolute expression level such as the copy number, or a relative expression level normalized to the copy number.
  • next-generation sequencer When measuring the expression level of microRNA using a "next-generation sequencer", for example, cDNA is prepared from the extracted RNA, amplicons are obtained by PCR, and amplicon sequencing is performed using the next-generation sequencer. do. Low-quality reads and the like are eliminated from the obtained sequencing data by quality control, and the expression level is measured by the number of reads having the desired base sequence.
  • the next-generation sequencer is a term that is contrasted with a capillary sequencer that uses the Sanger method.
  • Next-generation sequencers use sequencing principles such as synthetic sequencing, pyrosequencing, and ligase reaction sequencing. Amplicon sequencing by next-generation sequencers can be performed according to the manufacturer's recommended protocol.
  • Measurement of the expression level of microRNA includes not only measuring the expression level of mature microRNA, but also measuring the expression level of the transcription product of the gene encoding the microRNA to be measured or its processing product.
  • the primary transcript of a gene encoding a microRNA is called an early transcript (pri-miRNA) and has a stem-loop hairpin structure.
  • the pri-miRNA is converted into a mature miRNA precursor (pre-miRNA) having a stem-loop structure by an RNase III-like enzyme called Drosha.
  • the pre-miRNA is transported out of the cell nucleus and spliced by an enzyme called Dicer into a 20 to 25-mer double-stranded mature miRNA.
  • transcripts of genes encoding microRNAs or their processing products include pri-miRNAs, pre-miRNAs and mature miRNAs.
  • the expression level of the double-stranded mature miRNA can be measured not only by measuring the expression level of the single-stranded mature miRNA, but also by measuring the expression level of the single-stranded RNA that forms a complementary pair with it. may be included.
  • the microRNA having the same biological function as the microRNA such as homologs such as homologs and orthologs, mutants such as genetic polymorphisms, or Measuring the expression level of these derivatives may be included.
  • Nucleotide sequences of transcription products of genes encoding microRNAs to be measured, homologues, mutants or derivatives of microRNAs can be identified by referring to, for example, miRBase (http://www.mirbase.org/). can be done.
  • Subjects are primates including humans and chimpanzees, pet animals including dogs and cats, livestock animals including cattle, horses, sheep and goats, and mammals including rodents including mice and rats. you can A subject may be a human or a non-human mammal. In the following, a subject means, as an example, a subject for whom PFS is predicted, and is synonymous with a patient.
  • the origin of the "sample” may be, for example, bodily fluids such as blood to be detected, cerebrospinal fluid, lymph, ascites, and interstitial fluid; secretions such as urine, saliva, tears, and sweat.
  • the source of the sample is preferably at least one selected from the group consisting of blood, urine, cerebrospinal fluid, lymph, tear and ascites, more preferably blood, and serum or plasma prepared from blood.
  • PFS Progression Free Survival
  • the "drug effect” will be explained using the result of whether the patient is a “responder” or a “non-responder” as an example, but is not limited to this.
  • “responder” means a person who can obtain the effect of the drug.
  • a “non-responder” is someone who does not respond to a drug.
  • Whether a subject is a responder or a non-responder is a criterion determined by a physician at a given time after a subject has been treated with a drug.
  • a systemic evaluation including a CT scan may be performed at the end of four courses of anticancer drugs to determine the clinical efficacy of the drugs. When dosing is repeated thereafter, systemic assessments may be performed every 4 courses to determine efficacy.
  • the prediction method according to the first embodiment for example, using the expression level of miRNA derived from a sample of a patient who developed a certain disease, information on the period of stable condition after treatment of the patient, for example, PFS By making predictions, it is possible to predict the effect of a particular drug.
  • the prediction method according to the second embodiment uses the expression level of miRNA derived from a sample of a patient who has developed a certain disease to predict whether the patient is a responder or a non-responder. This makes it possible to predict the effects of specific drugs.
  • biomarkers according to the first and second embodiments can use specific miRNAs to predict the effects of drugs.
  • the prediction device 1 uses the target miRNA expression level as input data and uses a trained model M to predict the target PFS. That is, an example of predicting PFS will be described here as a value for predicting the effect of a drug.
  • a trained model M used by the prediction device 1 is generated by the training device 2 as shown in FIG. Therefore, first, learning, which is the generation of a trained model used for PFS prediction, will be described. After that, biomarkers according to embodiments will be described, and then prediction of PFS according to embodiments will be described. Specifically, an example of predicting the effect of nivolumab on lung cancer patients will be described.
  • the training device 2 executes learning processing using a training data set composed of real data of a plurality of patients to generate a trained model M.
  • the training device 2 uses the patient's "expression level of a specific miRNA” as an explanatory variable and the patient's "PFS" as the objective variable. Learn by learning.
  • This “miRNA expression level” is the miRNA expression level measured by a general method from serum obtained from patients administered with nivolumab. Specifically, patients receiving 240 mg of nivolumab every 2 weeks or 480 mg of nivolumab every 4 weeks will be targeted, and blood will be collected within 1 month after the start of nivolumab administration. The duration of nivolumab treatment is the same as PFS. Expression levels of miRNAs were measured by a next-generation sequencer.
  • the training device 2 includes a data acquisition unit 21 that acquires a training data set 23 used for learning, and a learning unit 22 that performs learning using the acquired training data set 23.
  • the training device 2 also stores the acquired training data set 23, the program P1 used for executing the learning process in the training device 2, and the trained model M generated by the learning unit 22 in the storage device.
  • the learning unit 22 for example, statistical techniques such as random forest, neural network, or decision tree can be used. In this case, as an example, a random forest is used and learning processing is performed by 5-fold cross-validation, but other machine learning techniques may be used.
  • the training device 2 can learn the relationship between the expression levels of different combinations of miRNAs and the PFS. For example, the training device 2 can generate a trained model M5 that has learned the relationship between the expression levels of five types of miRNAs and PFS. In addition, the training device 2 can generate a trained model M12 that has learned the relationship between the expression levels of certain 12 types of miRNAs and PFS. In addition, the training device 2 can generate a trained model M x that has learned the relationship between the expression level of an arbitrary type x of miRNA and the PFS as training data. It should be noted that the trained model M is simply used when no particular distinction is made.
  • the type of miRNA whose expression level is used to generate the learned model M can be selected using, for example, an analysis method such as analysis of variance as described below using the flowchart in FIG.
  • the data for analysis used in the analysis are data in which the expression levels of miRNAs obtained from a plurality of subjects who developed a certain disease are associated with PFS, and both are actually measured values.
  • the description is given assuming that the miRNA selection process is executed in the training device 2.
  • the present invention is not limited to this, miRNA selection is executed in another device, and the training device 2 performs It may be one that executes a learning process.
  • the training device 2 first initializes the value of N for each miRNA type (S01). Specifically, the value of N is set to zero.
  • the training device 2 divides the targets from the analysis data into a "long group” or a "short group” for each set threshold value (S02).
  • the threshold is an example in which the unit of PFS is "month”. For example, when the threshold is "3", the "long group” is the “group with a PFS of 3 months or more”, and the “short group” is , “group with PFS less than 3 months".
  • eight patterns of 3, 6, 9, 12, 15, 18, 21 and 24 are sequentially set for the threshold will be described. Therefore, the threshold is initially set to "3", and the PFS is divided into a "group of 3 months or more" and a "group of less than 3 months". Then, in subsequent processing, the threshold is set to "6".
  • an example using the eight patterns of threshold values described above will be described, but the above-described processing and order are examples.
  • the training device 2 removes subjects whose miRNA expression levels are outliers for each miRNA type from each of the long groups and short groups divided in step S02 (S03). Specifically, the training device 2 identifies the median value of the miRNA expression level for each miRNA type. Then, for each target miRNA expression level, outliers within a predetermined range (for example, “median absolute deviation>3”) about the median value are excluded.
  • a predetermined range for example, “median absolute deviation>3”
  • the training device 2 calculates the F value for each type of miRNA using the expression level of miRNA in the long group and the expression level of miRNA in the short group remaining in step S03 (S04).
  • the F value can be calculated using the within-group variance as the denominator and the between-group variance as the numerator.
  • the training device 2 uses the F value calculated in step S04 to calculate the p-value for each miRNA type (S05).
  • step S05 When the p-value calculated in step S05 is less than a predetermined value (for example, p-value ⁇ 0.05) (YES in S06), the training device 2 determines that this miRNA is effective in this threshold pattern. Increment the value of N for miRNA (S07). The value of N incremented here is stored in the storage device as the value of N for each miRNA.
  • a predetermined value for example, p-value ⁇ 0.05
  • the training device 2 determines whether or not there are unprocessed thresholds that have not been used for grouping in S02 (S08). If there is an unprocessed threshold value (YES in S08), the process returns to step S01, initializes the value of N, and repeats the processes of steps S01 to S06.
  • the training device 2 refers to the N value of each miRNA and selects the miRNA with the highest N value. (S09). Then, the miRNA expression level selected here becomes input data to the trained model M used in the prediction device 1 .
  • Fig. 5 is an example of miRNAs selected in groups with N values of 8 or more, 4 or more, and 3 or more.
  • the number in parentheses after the type of miRNA is the value of N for that miRNA.
  • the top 5 miRNAs enclosed by solid lines all had an N value of 8, and all of the 8 patterns had a p value of ⁇ 0.05.
  • the top 12 miRNAs enclosed by dashed lines have N values of 8 to 4, and p values ⁇ 0.05 for at least 4 patterns.
  • the top 19 miRNAs enclosed by the dashed-dotted line have N values between 8 and 3, and p values ⁇ 0.05 for at least 3 patterns.
  • miRNAs with N of 2 are hsa-miR-671-5p, hsa-miR-6087, hsa-miR-574-3p, hsa-miR-483-3p, hsa-miR-454-3p, hsa-miR-382-5p, hsa-miR-320d, hsa-miR-320b, hsa-miR-30b-5p, hsa-miR-3065-5p, hsa-miR-224-5p, hsa-miR-223- 3p, hsa-miR-191-5p, hsa-miR-146b-5p, hsa-miR-135b-5p, hsa-miR
  • miRNAs with N set to 1 are hsa-miR-92b-3p, hsa-miR-652-3p, hsa-miR-590-3p, hsa-miR-423-3p, hsa-miR-378c, hsa -miR-339-5p, hsa-miR-335-5p, hsa-miR-335-3p, hsa-miR-320a, hsa-miR-30e-3p, hsa-miR-29a-3p, hsa-miR-26b -5p, hsa-miR-221-3p, hsa-miR-215-5p, hsa-miR-2110, hsa-miR-206, hsa-miR-196a-5p, hsa-miR-190a-5p, hsa-miR
  • Fig. 6 shows a graph of scores obtained for each miRNA. Specifically, the graph shown in FIG. 6 is a graph obtained by scoring the total value of F values in eight patterns for each miRNA.
  • the top five types shown in FIG. 6 are the same as the selected top five types in FIG.
  • All of the 8 patterns of thresholds set in step S02 have p values ⁇ 0.05 (N ⁇ 8), and are selected as being useful for PFS prediction.
  • these five miRNAs are expected to be useful as biomarkers.
  • FIG. 7A is the distribution of PFS when hsa-miR-429 is used as a biomarker.
  • Each is a distribution when a certain period is set as a threshold T, and subjects are divided into a "long group (long)” or a "short group (short)" for each threshold T. From the upper left, the distributions are shown when the threshold T is set to 3 months, 6 months, 9 months, 12 months, 18 months, and 24 months.
  • the example shown in Figure 7B is the distribution of PFS when hsa-miR-200b-3p is used as a biomarker.
  • FIG. 7B also shows the distribution when the PFS is divided into a "long group” or a "short group” with a threshold T of every 3 months, 6 months, 9 months, 12 months, 18 months, and 24 months.
  • the example shown in FIG. 7C is also the distribution of PFS when hsa-miR-200a-3p is used as a biomarker.
  • the example shown in FIG. 7D also shows a similar PFS distribution when hsa-miR-100-5p is used as a biomarker.
  • FIG. 7E also shows a similar PFS distribution when hsa-miR-141-3p is used as a biomarker.
  • a highly accurate biomarker can be said to be a highly accurate biomarker if the same distribution trend appears in the "long group” and "short group” in each threshold pattern. Therefore, for the top five miRNAs shown in the examples of FIGS. 7A to 7E, the distributions of the “long group” and the “short group” appeared in the same way in the pattern of each threshold value, so it is considered to be a highly accurate biomarker. Conceivable.
  • biomarker As noted above, specific miRNAs can be biomarkers. Specifically, it can be a biomarker used to predict the effect of a drug on a disease.
  • diseases include lung cancer, colon cancer, stomach cancer, pancreatic cancer, liver cancer, breast cancer, kidney cancer, head and neck cancer, skin cancer, malignant lymphoma cancer, esophageal cancer, bladder cancer, ureter cancer, and immune response diseases. is mentioned.
  • it can be a biomarker used to predict the effect of a drug on lung cancer.
  • an anti-cancer drug which is an example of a drug
  • Immune checkpoint inhibitors include anti-PD-1 antibodies, anti-PD-L1 antibodies, anti-CTLA-4 antibodies and the like.
  • anti-PD-1 antibodies include nivolumab, pembrolizumab, spartalizumab, semiplimab, and the like.
  • anti-PD-L1 antibodies include avelumab, atezolizumab, durvalumab and the like.
  • the specific miRNA is at least It is one miRNA.
  • specific miRNAs are hsa-miR-429, hsa-miR-200b-3p, hsa-miR-200a-3p, hsa-miR-141-3p, hsa-miR-100-5p, hsa-miR-3679 from the group consisting of -5p, hsa-miR-200c-3p, hsa-miR-181a-5p, hsa-miR-4516, hsa-miR-21-5p, hsa-miR-452-5p and hsa-miR-320e It may be at least one miRNA selected.
  • miRNAs are hsa-miR-429, hsa-miR-200b-3p, hsa-miR-200a-3p, hsa-miR-141-3p, hsa-miR-100-5p, hsa-miR-3679 -5p, hsa-miR-200c-3p, hsa-miR-181a-5p, hsa-miR-4516, hsa-miR-21-5p, hsa-miR-452-5p, hsa-miR-320e, hsa-miR -99a-5p, hsa-miR-98-5p, hsa-miR-4429, hsa-miR-378a-3p, hsa-miR-22-3p, hsa-miR-1301-3p and hsa-miR-1246 It may
  • the specific miRNA is at least one of hsa-miR-429, hsa-miR-200b-3p, hsa-miR-200a-3p, hsa-miR-141-3p or hsa-miR-100-5p, and , hsa-miR-3679-5p, hsa-miR-200c-3p, hsa-miR-181a-5p, hsa-miR-4516, hsa-miR-21-5p, hsa-miR-452-5p and hsa-miR It may be one or more miRNAs selected from the group consisting of -320e.
  • the specific miRNA comprises at least one of hsa-miR-429, hsa-miR-200b-3p, hsa-miR-200a-3p, hsa-miR-141-3p or hsa-miR-100-5p, and hsa-miR-3679-5p, hsa-miR-200c-3p, hsa-miR-181a-5p, hsa-miR-4516, hsa-miR-21-5p, hsa-miR-452-5p and hsa-miR- It may be one or more miRNAs selected from the group consisting of 320e.
  • a specific miRNA from the group consisting of hsa-miR-100-5p and also hsa-miR-429, hsa-miR-200b-3p, hsa-miR-200a-3p and hsa-miR-141-3p It may be one or more miRNAs selected.
  • the specific miRNA is from the group consisting of hsa-miR-100-5p and hsa-miR-200a-3p and also hsa-miR-429, hsa-miR-200b-3p and hsa-miR-141-3p It may be one or more miRNAs selected.
  • the specific miRNA is from the group consisting of hsa-miR-100-5p, hsa-miR-200a-3p and hsa-miR-141-3p, and also hsa-miR-429 and hsa-miR-200b-3p It may be one or more miRNAs selected.
  • 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, or 19 miRNAs can be used as biomarkers. can.
  • miRNA sequences are as follows.
  • hsa-miR-429 hsa-miR-200b-3p, hsa-miR-200a-3p, hsa-miR-141-3p and hsa-miR-100-5p
  • the miRNAs described above can be used as biomarkers for judging the efficacy of nivolumab for patients with lung cancer.
  • 1, 2, 3, 4 or 5 miRNAs selected from the above 5 types can be used as biomarkers.
  • At least one or more miRNAs can be used as biomarkers for judging the efficacy of nivolumab for patients who have developed lung cancer.
  • 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, or 12 miRNAs selected from the above 12 species can be used as biomarkers.
  • hsa-miR-429 hsa-miR-200b-3p, hsa-miR-200a-3p, hsa-miR-141-3p, hsa-miR-100-5p, hsa-miR-3679-5p, hsa- miR-200c-3p, hsa-miR-181a-5p, hsa-miR-4516, hsa-miR-21-5p, hsa-miR-452-5p, hsa-miR-320e, hsa-miR-99a-5p, Selected from 19 groups of hsa-miR-98-5p, hsa-miR-4429, hsa-miR-378a-3p, hsa-miR-22-3p, hsa-miR-1301-3p and hsa-miR-1246 At least one or
  • the prediction device 1 described above with reference to FIG. 1 is used. Specifically, using the trained model M, the expression level of miRNA measured from the target sample is used as input data, and the PFS of the target is predicted. As shown in FIG. 8, the prediction device 1 includes an acquisition unit 11 that acquires expression level data 14 indicating the expression level of a specific miRNA measured from blood collected from a subject, and learns from the acquired expression level data 14. The prediction unit 12 predicts the PFS using the finished model M1, and the output unit 13 outputs the predicted PFS. The prediction device 1 also stores the expression level data 14, the program P for executing the prediction device 1, and the learned model M1 in the storage device.
  • the timing of collecting the sample is a predetermined timing before the start of treatment. Specifically, it is any timing from one month before the start of drug administration to the start of drug administration.
  • step S11 the expression level of specific miRNA from the blood collected in step S11 is measured (S12).
  • the processing is executed in the prediction device 1.
  • the dashed line indicates the processing executed by the prediction device 1 .
  • the acquisition unit 11 acquires the expression level of the specific miRNA measured in step S12 (S13).
  • the prediction unit 12 uses the miRNA expression levels obtained in step S13 to predict PFS using the trained model M (S14).
  • the output unit 13 outputs the PFS predicted in step S14 (S15).
  • the prediction device 1 can predict the PFS of a target using the expression level of a specific miRNA of the target. Therefore, for example, the prediction result by the prediction device 1 can assist a doctor's diagnosis.
  • each device in the above-described embodiment may be configured by hardware, CPU (Central Processing Unit), GPU (Graphics Processing Unit), etc. It may be configured by information processing of software (program).
  • software information processing software that realizes at least part of the functions of each device in the above-described embodiments can be transferred to a flexible disk, CD-ROM (Compact Disc-Read Only Memory), USB (Universal Serial Bus) memory or other non-temporary storage medium (non-temporary computer-readable medium) and read into a computer to execute software information processing.
  • the software may be downloaded via a communication network.
  • information processing may be performed by hardware by implementing software in circuits such as ASIC (Application Specific Integrated Circuit) and FPGA (Field Programmable Gate Array).
  • the type of storage medium that stores the software is not limited.
  • the storage medium is not limited to removable media such as magnetic disks and optical disks, and may be fixed storage media such as hard disks and memories.
  • the storage medium may be provided inside the computer (main storage device, auxiliary storage device, etc.) or may be provided outside the computer.
  • FIG. 10 is a block diagram showing an example of the hardware configuration of each device (prediction device 1 or training device 2) in the above-described embodiment.
  • Each device includes, for example, a processor 71, a main storage device 72 (memory), an auxiliary storage device 73 (memory), a network interface 74, and a device interface 75, which are connected via a bus 76.
  • a processor 71 for example, a main storage device 72 (memory), an auxiliary storage device 73 (memory), a network interface 74, and a device interface 75, which are connected via a bus 76.
  • a computer 7 implemented as
  • each device (prediction device 1 or training device 2) in the above-described embodiment is a system that realizes functions by one or more computers executing commands stored in one or more storage devices. may be configured. Further, the information transmitted from the terminal may be processed by one or more computers provided on the cloud, and the processing result may be transmitted to the terminal.
  • each device in the above-described embodiments is executed in parallel using one or more processors or using multiple computers via a network. good too. Also, various operations may be distributed to a plurality of operation cores in the processor and executed in parallel. Also, part or all of the processing, means, etc. of the present disclosure may be executed by at least one of a processor and a storage device provided on a cloud capable of communicating with the computer 7 via a network. Thus, each device in the above-described embodiments may be in the form of parallel computing by one or more computers.
  • the processor 71 may be an electronic circuit (processing circuit, processing circuitry, CPU, GPU, FPGA, ASIC, etc.) that performs at least computer control or computation.
  • Processor 71 may also be a general-purpose processor, a dedicated processing circuit designed to perform a particular operation, or a semiconductor device containing both a general-purpose processor and dedicated processing circuitry.
  • the processor 71 may include an optical circuit, or may include an arithmetic function based on quantum computing.
  • the processor 71 can perform arithmetic processing based on the data and software (programs) input from each device, etc. of the internal configuration of the computer 7, and output the arithmetic result and control signal to each device, etc.
  • the processor 71 may control each component of the computer 7 by executing the OS (Operating System) of the computer 7, applications, and the like.
  • Each device (prediction device 1 or training device 2) in the above-described embodiments may be realized by one or more processors 71.
  • the processor 71 may refer to one or more electronic circuits arranged on one chip, or one or more electronic circuits arranged on two or more chips or two or more devices. You can point When multiple electronic circuits are used, each electronic circuit may communicate by wire or wirelessly.
  • the main storage device 72 is a storage device that stores commands executed by the processor 71 and various types of data.
  • the auxiliary storage device 73 is a storage device other than the main storage device 72 .
  • These storage devices mean any electronic components capable of storing electronic information, and may be semiconductor memories.
  • the semiconductor memory may be either volatile memory or non-volatile memory.
  • a storage device for storing various data in each device (prediction device 1 or training device 2) in the above-described embodiments may be realized by the main storage device 72 or the auxiliary storage device 73, and is built into the processor 71. It may be realized by an internal memory.
  • the storage unit 102 in the above-described embodiment may be realized by the main storage device 72 or the auxiliary storage device 73.
  • a plurality of processors may be connected (coupled) to one storage device (memory), or a single processor may be connected.
  • a plurality of storage devices (memories) may be connected (coupled) to one processor.
  • Each device (prediction device 1 or training device 2) in the above-described embodiments consists of at least one storage device (memory) and a plurality of processors connected (coupled) to this at least one storage device (memory).
  • at least one processor among the plurality of processors may include a configuration that is connected (coupled) to at least one storage device (memory).
  • this configuration may be realized by storage devices (memory) and processors included in a plurality of computers.
  • a configuration in which a storage device (memory) is integrated with a processor for example, a cache memory including an L1 cache and an L2 cache
  • a configuration in which a storage device (memory) is integrated with a processor for example, a cache memory including an L1 cache and an L2
  • the network interface 74 is an interface for connecting to the communication network 8 wirelessly or by wire. As for the network interface 74, an appropriate interface such as one conforming to existing communication standards may be used. The network interface 74 may exchange information with the external device 9A connected via the communication network 8 .
  • the communication network 8 may be any one of WAN (Wide Area Network), LAN (Local Area Network), PAN (Personal Area Network), etc., or a combination of them. It is sufficient if information can be exchanged between them. Examples of WANs include the Internet, examples of LANs include IEEE 802.11 and Ethernet (registered trademark), and examples of PANs include Bluetooth (registered trademark) and NFC (Near Field Communication).
  • the device interface 75 is an interface such as USB that directly connects with the external device 9B.
  • the external device 9A is a device connected to the computer 7 via a network.
  • the external device 9B is a device that is directly connected to the computer 7. FIG.
  • the external device 9A or the external device 9B may be an input device.
  • the input device is, for example, a device such as a camera, microphone, motion capture, various sensors, keyboard, mouse, touch panel, etc., and provides the computer 7 with acquired information.
  • a device such as a personal computer, a tablet terminal, a smartphone, or the like that includes an input unit, a memory, and a processor may be used.
  • the external device 9A or the external device 9B may be an output device as an example.
  • the output device may be, for example, a display device such as LCD (Liquid Crystal Display), CRT (Cathode Ray Tube), PDP (Plasma Display Panel), organic EL (Electro Luminescence) panel, etc., and outputs audio etc. It may be a speaker or the like.
  • a display device such as LCD (Liquid Crystal Display), CRT (Cathode Ray Tube), PDP (Plasma Display Panel), organic EL (Electro Luminescence) panel, etc.
  • It may be a speaker or the like.
  • it may be a device such as a personal computer, a tablet terminal, or a smart phone that includes an output unit, a memory, and a processor.
  • the external device 9A or the external device 9B may be a storage device (memory).
  • the external device 9A may be a network storage or the like, and the external device 9B may be a storage such as an HDD.
  • the external device 9A or the external device 9B may be a device having the functions of some of the components of each device (prediction device 1 or training device 2) in the above-described embodiments. That is, the computer 7 may transmit or receive part or all of the processing results of the external device 9A or the external device 9B.
  • the expression "at least one (one) of a, b and c" or "at least one (one) of a, b or c" includes any of a, b, c, a-b, ac, b-c or a-b-c. Also, multiple instances of any element may be included, such as a-a, a-b-b, a-a-b-b-c-c, and so on. It also includes the addition of other elements than the listed elements (a, b and c), such as having d such as a-b-c-d.
  • connection and “coupled” when used, they refer to direct connection/coupling, indirect connection/coupling , electrically connected/coupled, communicatively connected/coupled, operatively connected/coupled, physically connected/coupled, etc. intended as a term.
  • the term should be interpreted appropriately according to the context in which the term is used, but any form of connection/bonding that is not intentionally or naturally excluded is not included in the term. should be interpreted restrictively.
  • the physical structure of element A is such that it is capable of performing operation B configuration, including that a permanent or temporary setting/configuration of element A is configured/set to actually perform action B good.
  • element A is a general-purpose processor
  • the processor has a hardware configuration capable of executing operation B, and operation B is performed by setting a permanent or temporary program (instruction). It just needs to be configured to actually run.
  • the element A is a dedicated processor or a dedicated arithmetic circuit, etc., regardless of whether or not control instructions and data are actually attached, the circuit structure of the processor actually executes the operation B. It just needs to be implemented.
  • each piece of hardware may work together to perform the predetermined processing, or a part of the hardware may perform the predetermined processing. You may do all of Also, some hardware may perform a part of the predetermined processing, and another hardware may perform the rest of the predetermined processing.
  • expressions such as "one or more hardware performs the first process, and the one or more hardware performs the second process" (including similar expressions ) is used, the hardware that performs the first process and the hardware that performs the second process may be the same or different. In other words, the hardware that performs the first process and the hardware that performs the second process may be included in the one or more pieces of hardware.
  • the hardware may include electronic circuits, devices including electronic circuits, and the like.
  • each of the plurality of storage devices may store only part of the data. , may store the entire data. Further, a configuration may be included in which some of the plurality of storage devices store data.
  • FIG. 11 shows the accuracy of prediction results in the prediction device 1 using the trained model M.
  • PFS is information on the duration of post-treatment stable state when each patient who developed lung cancer received nivolumab.
  • the amount of nivolumab administered to each patient was the same period as PFS, and the administration period was 240 mg every 2 weeks or 480 mg every 4 weeks. is the same period as PFS, and the number of patients covered is 213.
  • PFS is information on the duration of post-treatment stable state when each patient who developed lung cancer received nivolumab.
  • the amount of nivolumab administered to each patient was the same period as PFS, and the administration period was 240 mg every 2 weeks or 480 mg every 4 weeks. is the same period as PFS, and the number of patients covered is 213.
  • FIG. 1 is information on the duration of post-treatment stable state when each patient who developed lung cancer received nivolumab.
  • the amount of nivolumab administered to each patient was the same period
  • Model M 19 with expression level of miRNA as input model M 36 with expression level of top 36 miRNAs as input, model M 61 with expression level of top 61 miRNAs as input, expression of 2576 miRNAs
  • Figure 2 shows the accuracy of prediction results of model M2576 with quantity as input.
  • the model M12 in which the expression levels of the top 12 miRNAs are used as input, has the highest accuracy in the mean square error.
  • Example 2 In Example 1 shown in FIG. 11, the accuracy of PFS prediction results using the expression levels of the top 12 miRNAs was the highest. In Example 2, among these top 12 miRNAs, miRNAs that affect the improvement of the accuracy of PFS prediction were searched.
  • FIG. 12 shows the expression of one miRNA (hsa-miR-429, hsa-miR-200b-3p, hsa-miR-200a-3p, hsa-miR-141-3p and hsa-miR-100-5p).
  • Mean squared error (mse) Mean squared error (mse), coefficient of determination (R 2 ), Spearman's correlation coefficient (spearman_r), Spearman's P-value (spearman_p), Pearson's correlation coefficient (pearson_r) for PFS, which is the prediction result using quantity ), indicating Pearson's P-value (pearson_p).
  • FIG. 13 shows the top two miRNAs (hsa-miR-141-3p and hsa-miR-429), the top three miRNAs (hsa-miR-141-3p, hsa-miR-429 and hsa-miR -200a-3p), top 4 miRNAs (hsa-miR-141-3p, hsa-miR-429, hsa-miR-200a-3p and hsa-miR-200b-3p) and top 5 miRNAs (hsa -miR-141-3p, hsa-miR-429, hsa-miR-200a-3p, hsa-miR-200b-3p and hsa-miR-100-5p) , mean squared error (mse), coefficient of determination (R 2 ), Spearman's correlation coefficient (spearman_r), Spearman's P-value (spearman_
  • Figure 14 shows the mean square error (mse), coefficient of determination ( R2 ), and Spearman's correlation coefficient when predicting PFS with the expression levels of four types of miRNA, excluding one type of miRNA from the top five types.
  • pearman_r Spearman's P-value
  • pearman_p Pearson's correlation coefficient
  • pearson_r Pearson's P-value
  • pearson_p Pearson's P-value
  • Figure 15 shows four species (hsa-miR-200b-3p, hsa-miR-200a-3p, hsa-miR-141-3p and hsa-miR-100-5p ), the mean square error (mse), coefficient of determination (R 2 ), Spearman's correlation coefficient (spearman_r), and Spearman's P-value (spearman_p), Pearson's correlation coefficient (pearson_r), and Pearson's P-value (pearson_p) are shown.
  • the mean square error mse
  • coefficient of determination R 2
  • Spearman's correlation coefficient spearman_r
  • Spearman's P-value spearman_p
  • Pearson's correlation coefficient pearson_r
  • Pearson's P-value pearson_p
  • Figure 16 shows the mean square error (mse) when PFS is predicted by the expression level of two miRNAs, excluding one type of miRNA from the three types with high PFS prediction accuracy in Figure 15.
  • Coefficient (R 2 ) Spearman's correlation coefficient (spearman_r), Spearman's P-value (spearman_p), Pearson's correlation coefficient (pearson_r), Pearson's P-value (pearson_p) are shown.
  • the hsa-miR- It can be said that the mean squared error (124.974067) of the PFS prediction accuracy using the expression levels of two types of miRNAs, 100-5p and hsa-miR-200a-3p, is the highest.
  • the mean square error 123.751727, 124.724570
  • ⁇ Modification 1> In the prediction method described above, an example of prediction using the trained model M was explained.
  • the effect of the drug may be predicted according to whether the expression level is larger or smaller than each cutoff value. For example, for five miRNAs hsa-miR-429, hsa-miR-200b-3p, hsa-miR-200a-3p, hsa-miR-141-3p and hsa-miR-100-5p, each Set expression level cutoff values T 1 to T 5 . Then, the expression level of each miRNA is compared with each cutoff value.
  • the effect of the drug on the subject may be obtained.
  • T 1 to T 12 may be set.
  • the expression level of each miRNA is compared with each cutoff value.
  • the conditions were selected from among the expression levels of hsa-miR-429, hsa-miR-200b-3p, hsa-miR-200a-3p, hsa-miR-141-3p and hsa-miR-100-5p.
  • the cutoff value T of the miRNA expression level of hsa-miR-100-5p, hsa-miR-429, hsa-miR-200b-3p, hsa-miR-200a-3p and hsa-miR-141-3p A cutoff value T for the expression level of one or more miRNAs selected from the group consisting of may be set.
  • the expression level of these two or more miRNAs measured from the subject sample is compared with the cutoff value of the expression level of each miRNA, and if the expression level of each miRNA satisfies the conditions, the effect of the drug is obtained for the subject. It may be possible.
  • a cutoff value T for the expression level of one or more miRNAs selected from the group consisting of 3p may be set.
  • the expression level of these two or more miRNAs measured from the subject sample is compared with the cutoff value of the expression level of each miRNA, and if the expression level of each miRNA satisfies the conditions, the effect of the drug is obtained for the subject. It may be possible.
  • cutoff values T for the expression levels of two types of miRNAs hsa-miR-100-5p and hsa-miR-200a-3p, and hsa-miR-429, hsa-miR-200b-3p and hsa-
  • a cutoff value T for the expression level of one or more miRNAs selected from the group consisting of miR-141-3p may be set. The expression level of these three or more miRNAs measured from the subject sample is compared with the cutoff value of the expression level of each miRNA. It may be possible.
  • a cutoff value T for the expression level of one or more specific miRNAs may be set.
  • the expression level of one or more miRNAs measured from the sample of interest is compared with the cutoff value of the expression level of each miRNA, and if the expression level of each miRNA satisfies the conditions, the effect of the drug is obtained on the subject.
  • it is not necessarily required that the expression level of all miRNAs is equal to or higher than each cutoff value or equal to or lower than the cutoff value. It may be one that satisfies a condition such that the expression level of a predetermined number of miRNAs is equal to or higher than the cutoff value or equal to or lower than the cutoff value.
  • the prediction device 1A uses the expression level of target miRNA as input data, and determines whether the target is a responder or a non-responder using a trained model M2. Predict. That is, here, an example of predicting whether the target is a responder or a non-responder will be described as a value for predicting the effect of a drug. Specifically, an example of predicting the effect of nivolumab on lung cancer patients will be described.
  • the input data includes clinical information of the subject as well as miRNA expression levels. Therefore, the trained model M2 used by the prediction device 1A is, as shown in FIG. is learned by the training device 2A.
  • the training device 2A can use, for example, statistical methods such as random forests, neural networks, or decision trees.
  • learning processing was performed by XgBoost using a decision tree, but other machine learning techniques may be used.
  • the trained model M2 this time was evaluated using 5-fold cross-validation.
  • the clinical information includes, for example, the subject's "sex”, “smoking history”, “gene mutation”, “age”, “PD-L1”, “histopathological classification”, and the like.
  • the configuration of the training device 2A is the same as that of the training device 2 described above with reference to FIG. 3, the description using the drawings is omitted.
  • the configuration of the prediction device 1A is the same as that of the prediction device 1 described above with reference to FIG.
  • the learned model M2 differs depending on the data used for learning.
  • the training data set includes the expression level of a specific miRNA and does not include clinical information
  • the trained model M2 is a responder or a non-responder with the expression level of the specific miRNA as input data. to predict.
  • Clinical information is stored in the storage device of the prediction device 1A.
  • the training device 2A and the prediction device 1A may be implemented by the computer 7 having the hardware configuration described above with reference to FIG.
  • each clinical information is converted into binary data and used.
  • gender is treated as binary data of 0 and 1 indicating whether a person is male or female.
  • smoking history was treated as binary data of 0 and 1 indicating whether or not the subject was a smoker.
  • Gene mutation was treated as binary data of 0 or 1 indicating whether or not there was a gene mutation.
  • a threshold age for example, 75 years old was set, and binary data according to whether or not the age was above the threshold was handled.
  • PD-L1 was treated with four categories of data: "1. NE (data not available)", "2. Negative", "3. Positive” and "4. Strong positive”.
  • the selection of the type of miRNA whose expression level is used in the prediction device 1A according to the second embodiment can also be made in the same manner as described in the first embodiment.
  • the prediction device 1A according to the second embodiment uses the 61 types of miRNAs (miRNAs shown in FIG. 6) with the highest scores in the first embodiment.
  • Fig. 19 shows the results of predicting whether a subject is a responder or a non-responder using the expression levels of 61 miRNAs of the subject.
  • the 61 types of miRNAs are hsa-miR-141-3p, hsa-miR-429, hsa-miR-200a-3p, hsa-miR-200b-3p, hsa-miR-100-5p, hsa-miR-200c -3p, hsa-miR-3679-5p, hsa-miR-135b-5p, hsa-miR-181a-5p, hsa-miR-21-5p, hsa-miR-4516, hsa-miR-320e, hsa-miR -452-5p, hsa-miR-4429, hsa-miR-99a-5p, hsa-m
  • responders In the example shown in FIG. 19, responders, non-responders, and each weighted average prediction result accuracy (Accuracy), precision (Precision), specificity (sensitivity), negative predictive value ( Negative Predictive Value (NPV)), F value (F1-Score), number of samples (support) and AUC (Area Under the Curve).
  • FIG. 20 shows whether the subject is a responder or a non-responder using clinical information such as age, sex, and smoking history in addition to the expression levels of 61 miRNAs similar to those in FIG. This is the expected result.
  • the example shown in FIG. 20 shows the accuracy, precision, specificity, sensitivity, negative predictive value (NPV), F value, support, and AUC of responders, non-responders, and their respective weighted averages. Comparing the accuracy shown in FIG. 20 with the accuracy shown in FIG. 19, when clinical information of age, sex, and smoking history is used, the accuracy is improved compared to when clinical information is not used. I understand.
  • the example shown in FIG. 21 is the accuracy of prediction results using other clinical information in addition to miRNA expression levels and clinical information such as sex, age, and smoking history.
  • the accuracy is 0.782 and 0.787, which decreased compared to the accuracy of 0.792 when age, sex, and smoking history were used as clinical information for miRNA expression levels.
  • the specific miRNAs described above can be used as biomarkers.
  • drugs for lung cancer, colon cancer, stomach cancer, pancreatic cancer, lung cancer, breast cancer, kidney cancer, head and neck cancer, skin cancer, malignant lymphoma cancer, esophageal cancer, etc. can be a biomarker used for prediction of efficacy.
  • it can be a biomarker used to predict the effect of a drug on lung cancer.
  • the drug can be an immune checkpoint inhibitor.
  • miRNAs are hsa-miR-141-3p, hsa-miR-429, hsa-miR-200a-3p, hsa-miR-200b-3p, hsa-miR-100-5p, hsa-miR-200c-3p , hsa-miR-3679-5p, hsa-miR-135b-5p, hsa-miR-181a-5p, hsa-miR-21-5p, hsa-miR-4516, hsa-miR-320e, hsa-miR-452 -5p, hsa-miR-4429, hsa-miR-99a-5p, hsa-miR-22-3p, hsa-miR-1246, hsa-miR-98-5p, hsa-miR-1301-3p, hsa
  • miRNA sequences are as follows.
  • Prediction processing using the prediction device 1A will be described using the flowchart shown in FIG.
  • the learned model M2 uses the expression level of a specific miRNA and clinical information as input data. Therefore, the prediction process uses the miRNA expression level measured from the target sample using the trained model M2 and the clinical information of the target as input data, and predicts whether the target is a responder or a non-responder.
  • the target's blood is collected (S21).
  • the timing of sample collection is, for example, from one month before the start of drug administration to the start of drug administration, as in the case described above in the first embodiment.
  • the process is executed in the prediction device 1A.
  • the dashed line portion is the processing executed in the prediction device 1A.
  • the acquisition unit 11 of the prediction device 1A acquires the specific miRNA expression level measured in step S22 and the clinical information of the target (S23).
  • Subject clinical information may be entered by a user, for example, via an input device.
  • the prediction unit 12 of the prediction device 1A uses the miRNA expression level and clinical information acquired in step S23 to predict whether the target is a responder or a non-responder using the trained model M2 (S24 ).
  • the output unit 13 of the prediction device 1A outputs a prediction result indicating whether the target predicted in step S24 is a responder or a non-responder (S25).
  • the prediction device 1A can predict whether the target is a responder or a non-responder using the miRNA of a specific target.
  • the prediction accuracy is improved when specific clinical information of the subject is added to the prediction of whether the subject is a responder or a non-responder. Therefore, for example, the prediction result by the prediction device 1A can assist the doctor's judgment.
  • the prediction device 1B according to the third embodiment uses the target miRNA expression level as input data, and the target is a super responder, a mild responder, or a non-responder using a trained model M3. Predict which of That is, while the prediction device 1A according to the second embodiment described above performs two-class classification, the prediction device 1B according to the third embodiment performs three-class classification using the same input data. is implemented. Specifically, the prediction device 1B according to the second embodiment classifies the target as to whether the drug is effective or not effective. The prediction device 1A classifies the target as to whether a high drug effect can be obtained, a low drug effect can be obtained, or no drug effect.
  • the input data includes clinical information of the subject along with miRNA expression levels.
  • super responders are classifications (response) determined according to the period of PFS of the subject and the effect of cancer treatment of the subject, as shown in FIG. Specifically, the effects of cancer treatment are classified into four categories: CR (complete response), PR (partial response), PD (progressive disease), and SD (stable disease). be done. In one example shown in FIG.
  • a Super Responder is a subject "with treatment efficacy classified as CR or PR AND a PFS greater than or equal to a first threshold T1 .”
  • Mild Responders are "subjects whose therapeutic effect is classified as CR or PR and whose PFS is less than the first threshold T 1 ", and "subjects whose therapeutic effect is classified as SD. and a PFS that is equal to or greater than the second threshold value T2 .
  • a Non Responder is a subject "with treatment efficacy classified as SD or PD AND PFS less than the second threshold T2 .”
  • the trained model M3 used by the prediction device 1B uses the miRNA expression level and clinical information as explanatory variables to determine whether the model is a super responder, a mild responder, or a non-responder.
  • the training device 2B learns the relationship of the data set used as the objective variable. Similar to the examples described above, the training device 2B may also utilize statistical techniques such as random forests, neural networks or decision trees. This time, a trained model M3 that has been trained by a random forest will be described as an example. Also, this trained model M3 was evaluated using 5-fold cross-validation.
  • a doctor determines whether the therapeutic effect of the subject is CR, PR, PD or SD.
  • Super responders, mild responders and non-responders are determined according to PFS and response to treatment.
  • the clinical information used as input data is the same as the above example, for example, the subject's "sex”, “smoking history”, “gene mutation”, “age”, “PD-L1”, “histopathological Classification”, etc.
  • the learned model M3 differs depending on the data used for learning. As an example, when the training data set includes the expression level of a specific miRNA and specific clinical information, the trained model M3 also uses the expression level of the specific miRNA and specific clinical information as input data for super responders, miles Predict whether it is a responder or a non-responder. In addition to the expression level data, clinical information is stored in the storage device of the prediction device 1B. Also, the training device 2B and the prediction device 1B may be implemented by a computer having the hardware configuration described above with reference to FIG.
  • FIG. 26 shows the p-values obtained for 45 miRNAs and 3 clinical information selected when the threshold PFS is "20 months" and the p-value cutoff value is "0.05". shows a graph of In FIG. 26, the horizontal axis is the type of either miRNA or clinical information, and the vertical axis indicates the p-value for each miRNA or clinical information. In FIG. 26, clinical information is marked with *.
  • the three selected clinical information are PD-L1, smoking rate and gene mutation.
  • the selected 45 miRNAs are hsa-miR-21-5p, hsa-miR-429, hsa-miR-3679-5p, hsa-miR-100-5p, hsa-miR-141-3p, hsa - miR-200c-3p, hsa-miR-200a-3p, hsa-miR-2110, hsa-miR-130a-3p, hsa-miR-378a-3p, hsa-miR-98-5p, hsa-miR-203a , hsa-miR-382-5p, hsa-miR-4532, hsa-miR-664a-5p, hsa-miR-145-3p, hsa-miR-483-3p, hs
  • Figure 27 shows whether the subject is a super responder, a mild responder or a non-responder using the expression levels of 45 miRNAs of the subject and PD-L1, smoking rate and gene mutation as clinical information. This is the expected result.
  • the example shown in FIG. 27 shows the precision (Precision), recall (Recall), F value (F1-score), and number of samples (Support) for super responders, mild responders, and non-responders.
  • the number of samples is the number of subjects used as a data set for learning. Also, in this case, the accuracy of the trained model M3 was 0.62.
  • the specific miRNAs described above can be used as biomarkers.
  • drugs for lung cancer, colon cancer, stomach cancer, pancreatic cancer, lung cancer, breast cancer, kidney cancer, head and neck cancer, skin cancer, malignant lymphoma cancer, esophageal cancer, etc. can be a biomarker used for prediction of efficacy.
  • it can be a biomarker used to predict the effect of a drug on lung cancer.
  • the drug can be an immune checkpoint inhibitor.
  • miRNAs are hsa-miR-21-5p, hsa-miR-429, hsa-miR-3679-5p, hsa-miR-100-5p, hsa-miR-141-3p, hsa-miR-200c-3p , hsa-miR-200a-3p, hsa-miR-2110, hsa-miR-130a-3p, hsa-miR-378a-3p, hsa-miR-98-5p, hsa-miR-203a, hsa-miR-382 -5p, hsa-miR-4532, hsa-miR-664a-5p, hsa-miR-145-3p, hsa-miR-483-3p, hsa-miR-181a-5p, hsa-miR-1301-3p, h
  • the specific miRNAs are hsa-miR-21-5p, hsa-miR-429, hsa-miR-141-3p, hsa-miR-200c-3p, hsa-miR-200a-3p, hsa- miR-3679-5p, hsa-miR-100-5p, hsa-miR-130a-3p, hsa-miR-2110, hsa-miR-98-5p, hsa-miR-378a-3p, hsa-let-7f- 5p, hsa-miR-424-5p, hsa-miR-181a-5p, hsa-miR-382-5p, hsa-miR-22-3p, hsa-miR-29c-5p, hsa-miR-483-3p, hsa-miR-320e, hsa
  • the learned model M3 uses the expression level of a specific miRNA and clinical information as input data. Therefore, the prediction process uses the miRNA expression level measured from the target sample using the trained model M3 and the clinical information of the target as input data, and whether the target is a super responder, a mild responder, or a non-responder predict there will be
  • the target's blood is collected (S31).
  • the timing of sample collection is, for example, from one month before the start of drug administration to the start of drug administration, as in the case described above in the first embodiment.
  • the processing is executed in the prediction device 1B.
  • the dashed line indicates the processing executed by the prediction device 1B.
  • the acquisition unit 11 of the prediction device 1B acquires the specific miRNA expression level measured in step S32 and the clinical information of the target (S33).
  • Subject clinical information may be entered by a user, for example, via an input device.
  • the prediction unit 12 of the prediction device 1B uses the miRNA expression level and clinical information acquired in step S33 to determine whether the target is a super responder, a mild responder, or a non-responder according to the trained model M2. is predicted (S34).
  • the output unit 13 of the prediction device 1B outputs a prediction result indicating whether the target predicted in step S34 is a super responder, a mild responder, or a non-responder (S35).
  • the prediction device 1B can predict whether the target is a super responder, a mild responder, or a non-responder using the miRNA of a specific target. In addition to predicting whether a subject is a super responder, mild responder, or non-responder, along with the miRNA expression level, the prediction accuracy will be improved when the subject's specific clinical information is added. I understand. Therefore, for example, the prediction result by the prediction device 1B can assist the doctor's judgment.
  • the program update system, control system, mobile object, program update method, and program described in all claims of this disclosure are realized by hardware resources such as processors, memories, and cooperation with programs.
  • the prediction method, prediction device, computer program, and biomarkers of the present disclosure are useful, for example, in determining the efficacy of nivolumab for patients who have developed lung cancer.
  • prediction device 11 acquisition unit 12 prediction unit 13 expression level data 13 output unit 2 training device 21 data acquisition unit 22 learning unit 23 training data set

Landscapes

  • Life Sciences & Earth Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Organic Chemistry (AREA)
  • Genetics & Genomics (AREA)
  • Wood Science & Technology (AREA)
  • Zoology (AREA)
  • Biotechnology (AREA)
  • Biomedical Technology (AREA)
  • General Engineering & Computer Science (AREA)
  • Microbiology (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Proteomics, Peptides & Aminoacids (AREA)
  • Molecular Biology (AREA)
  • Sustainable Development (AREA)
  • Medicinal Chemistry (AREA)
  • Physics & Mathematics (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Immunology (AREA)
  • Plant Pathology (AREA)
  • Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)

Abstract

対象の薬剤の効果を予測する。予測方法は、対象の試料に由来する所定のmiRNAに関する値に基づいて疾患に対する薬剤の効果を予測する予測ステップを含む。

Description

予測方法及びバイオマーカー
 本開示は、予測方法及びバイオマーカーに関する。
 近年、miRNAを利用した様々な研究が進められている。がんの診断又は治療に、miRNAを利用することがある。その他、食道がんの患者での薬効予測にmiRNAを利用する技術もある。
 例えば、薬剤が患者に効果があったか否かの判断として、患者に薬剤を投与後にその患者の反応を観察することがある。また、その患者のPFS (無増悪生存期間:Progression Free Survival)を用いることもできる。このような患者の反応やPFSは、薬効の予測に用いる基準として有効な手段となる。
特開2020-80773号公報
 本開示は、対象の試料に由来する所定のmiRNAに基づく値から薬剤の効果を予測する予測方法、予測装置及びバイオマーカーを提供する。
 本開示の予測方法は、対象の試料に由来する所定のmiRNAに関する値に基づいて疾患に対する薬剤の効果を予測する予測ステップを含む。予測方法により得られた予測結果は、医師の診断等を補助することができる。
 本開示の疾患に対する薬剤の効果の予測に用いるためのバイオマーカーは、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p又はhsa-miR-100-5pから選択された少なくとも1以上のmiRNAを含む。
 これらの概括的かつ特定の態様は、システム、方法、及びコンピュータプログラム、並びに、それらの組み合わせにより、実現されてもよい。
第1の実施形態に係る予測装置を示す概念図である。 図1の予測装置で用いる学習済みモデルの生成を示す概念図である。 学習済みモデルを生成する訓練装置の構成を示すブロック図である。 第1の実施形態に係る予測装置で利用するmiRNAの選択を示すフローチャートである。 選択されたPFSの予測に利用するmiRNAを示す。 各miRNAのスコアを示すグラフである。 hsa-miR-429の効果を示す分布図である。 hsa-miR-200b-3pの効果を示す分布図である。 hsa-miR-200a-3pの効果を示す分布図である。 hsa-miR-100-5pの効果を示す分布図である。 hsa-miR-141-3pの効果を示す分布図である。 第1の実施形態に係る予測装置の構成を示すブロック図である。 第1の実施形態に係る予測装置における予測の処理を説明するフローチャートである。 訓練装置及び予測装置のハードウェア構成を示すブロック図である。 実施例1に係る予測装置による予測結果を示す各値である。 実施例2に係る予測装置による1種のmiRNAを使用した各予測結果を示す各値である。 実施例2に係る予測装置による複数のmiRNAを使用した予測結果を示す各値である。 実施例2に係る予測装置による4種のmiRNAを使用した各予測結果を示す各値である。 実施例2に係る予測装置による3種のmiRNAを使用した各予測結果を示す各値である。 実施例2に係る予測装置による2種のmiRNAを使用した各予測結果を示す各値である。 第2の実施形態に係る予測装置を示す概念図である。 図18の予測装置で用いる学習済みモデルの生成を示す概念図である。 第2の実施形態に係る予測装置によるmiRNAの発現量を用いた予測結果を示す各値である。 第2の実施形態に係る予測装置によるmiRNAの発現量及び特定の臨床情報を用いた予測結果を示す各値である。 第2の実施形態に係る予測装置によるmiRNAの発現量及び各臨床情報を用いた予測結果である正確度である。 第2の実施形態に係る予測装置における予測の処理を説明するフローチャートである。 第3の実施形態に係る予測装置を示す概念図である。 スーパーレスポンダー、マイルドレスポンダー及びノンレスポンダーの特定方法について示す図である。 図23の予測装置で用いる学習済みモデルの生成を示す概念図である。 第3の実施形態に係る予測装置によるmiRNAの発現量及び臨床情報に関するp値を示すグラフである。 第2の実施形態に係る予測装置によるmiRNAの発現量及び特定の臨床情報を用いた予測結果を示す各値である。 第3の実施形態に係る予測装置における予測の処理を説明するフローチャートである。
[実施形態]
 以下に、図面を参照して実施形態に係る予測方法、予測装置及びバイオマーカーについて説明する。実施形態に係る予測方法及び予測装置は、「対象の試料に由来する所定の値」に基づいて、「薬剤の効果」を予測する。また、実施形態に係るバイオマーカーは、予測方法、予測装置及びコンピュータプログラムで利用する「対象の試料に由来する所定の値」の種別である。なお、以下の説明では、同一の構成について、同一の符号を付して説明を省略する。
 以下の説明において、「対象の試料に由来する所定の値」は、「対象のmiRNAの発現量」を例として説明する。また、miRNAは、マイクロRNAである。
 「マイクロRNAの発現量」は、例えば、試料からRNAを抽出し、抽出されるRNAに含まれるマイクロRNAの発現量を定量、又は半定量することで測定される。試料からのRNAの抽出には、市販のRNA抽出キットを用いることができる。市販のRNA抽出キットを用いて、付属の取り扱い説明書の手順に準じて行うことができる。すなわち、マイクロRNAの発現量は、インビトロによる測定で得ることができる。
 「マイクロRNAの発現量」は、例えば、DNAマイクロアレイ、RNAマイクロアレイ、定量PCR、次世代シーケンサー等の通常用いられる方法によって測定することができる。測定されるマイクロRNAの発現量は、絶対発現量であってもよく、相対発現量であってもよい。
 「DNAマイクロアレイ」は、測定対象となるマイクロRNAの少なくとも一部と相補的な塩基配列を有するDNAプローブを基材上に固定化したものである。DNAマイクロアレイを用いて特定マイクロRNAの発現量を測定する場合、例えば、抽出されたRNAを、市販のキットを用いて蛍光標識し、DNAマイクロアレイ上に配置されたマイクロRNAと相補的な配列を有するDNAプローブとハイブリダイズさせた後、蛍光強度を測定することで、マイクロRNAの発現量を測定することができる。
 「RNAマイクロアレイ」は、測定対象となるマイクロRNAの少なくとも一部と相補的な塩基配列を有するRNAプローブを基材上に固定化したものである。RNAマイクロアレイを用いて特定マイクロRNAの発現量を測定する場合、例えば、抽出されたRNAを、市販のキットを用いて蛍光標識し、RNAマイクロアレイ上に配置されたマイクロRNAと相補的な配列を有するRNAプローブとハイブリダイズさせた後、蛍光強度を測定することで、マイクロRNAの発現量を測定することができる。
 「定量PCR」には、リアルタイムPCR(RT-PCR)、デジタルPCR等の公知の手法を適用することができる。定量PCRでマイクロRNAの発現量を測定する場合、例えば、抽出されたRNAからcDNAを調製し、cDNAを測定対象となるマイクロRNAに特異的なプライマーセットを用いて定量PCRにより増幅することで、マイクロRNAの発現量を測定することができる。定量PCRで測定されるマイクロRNAの発現量は、例えば、コピー数のような絶対発現量であってよく、コピー数を規格化した相対発現量であってもよい。
 「次世代シーケンサー」を用いてマイクロRNAの発現量を測定する場合、例えば、抽出されたRNAからcDNAを調製し、PCRによってアンプリコンを得て、次世代シーケンサーを用いてアンプリコンシーケンシングを実施する。得られたシーケンシングデータから、クオリティコントロールにより、低クオリティリードなどを削除した上で、所望の塩基配列を有するリード数によって発現量が測定される。ここで次世代シーケンサーとは、サンガー法を用いるキャピラリーシーケンサーと対比される用語である。次世代シーケンサーでは、合成シーケンシング法、パイロシーケンシング法、リガーゼ反応シーケンシング法等のシーケンシング原理が用いられる。次世代シーケンサーによるアンプリコンシーケンシングは、メーカー推奨プロトコルに従って実施することができる。
 マイクロRNAの発現量の測定には、成熟マイクロRNAの発現量を測定することのみならず、測定対象のマイクロRNAをコードする遺伝子の転写産物又はそのプロセシング産物の発現量を測定することが含まれてもよい。一般にマイクロRNAをコードする遺伝子の一次転写産物は初期転写産物(pri-miRNA)と呼ばれ、ステムループのヘアピン構造を有する。pri-miRNAはRNaseIII様のDroshaと呼ばれる酵素によってステムループ構造をもつ成熟miRNAの前駆体(pre-miRNA)に変換される。pre-miRNAは細胞核の外に輸送され、Dicerと呼ばれる酵素のスプライシングによって、20から25塩基長の2本鎖の成熟miRNAとなる。したがって、マイクロRNAをコードする遺伝子の転写産物又はそのプロセッシング産物は、pri-miRNA、pre-miRNA及び成熟miRNAを包含する。なお、2本鎖の成熟miRNAの発現量の測定には、1本鎖の成熟miRNAの発現量を測定することのみならず、それと相補対を形成する1本鎖RNAの発現量を測定することが含まれてよい。また、測定対象となるマイクロRNAの発現量の測定には、当該マイクロRNAと生物学的な機能が同等であるマイクロRNA、例えばホモログ、オルソログ等の同族体、遺伝子多型などの変異体、又はこれらの誘導体の発現量を測定することが含まれてもよい。測定対象のマイクロRNAをコードする遺伝子の転写産物の塩基配列、マイクロRNAの同族体、変異体又は誘導体は、例えば、miRBase(http://www.mirbase.org/)を参照して同定することができる。
 「対象」は、ヒト、チンパンジー等を含む霊長類、イヌ、ネコ等のペット動物、ウシ、ウマ、ヒツジ、ヤギ等を含む家畜動物、マウス、ラット等を含む齧歯類を含む哺乳動物であってよい。対象はヒトであってもよく、非ヒト哺乳動物であってもよい。以下において対象は、一例として、PFSが予測される被検者を意味し、患者と同義である。
 「試料」の由来は、例えば、検出対象の血液、脳脊髄液、リンパ液、腹水、組織液等の体液;尿、唾液、涙液、汗等の分泌物であってよい。試料の由来として好ましくは血液、尿、脳脊髄液、リンパ液、涙液及び腹水からなる群から選択される少なくとも1種であり、より好ましくは血液であり、血液から調製される血清又は血漿であってもよい。
 「薬剤の効果」は、「PFS(無増悪生存期間:Progression Free Survival)」により表される値を例として説明するが、これに限定されるものではなく、他の指標を用いて測定してもよい。ここで、「PFS」とは、ある疾患を発症した対象が薬剤を用いて治療後、病状が進行せず安定した状態である期間を意味する。
 または、「薬剤の効果」は、「レスポンダー」であるか、「ノンレスポンダー」であるかの結果を例として説明するが、これに限定されるものではない。ここで、「レスポンダー」は、薬剤の効果が得られる者をいう。「ノンレスポンダー」は、薬剤の効果が得られない者をいう。レスポンダー又はノンレスポンダーのいずれであるかは、対象が薬剤を用いて治療後の所定時点で、医師によって決定される基準である。一例として、抗がん剤開始4コース終了時にCT検査含めて全身評価を行い、薬剤の臨床的効果を決定してもよい。その後も投与が繰り返さる際、4コース毎に全身評価を実施し、効果を決定してもよい。
 第1の実施形態に係る予測方法は、例えば、ある疾患を発症した患者の試料に由来するmiRNAの発現量を用いて、その患者の治療後の安定した状態の期間に関する情報、例えば、PFSを予測することにより、特定の薬剤の効果を予測することができる。
 また、第2の実施形態に係る予測方法は、例えば、ある疾患を発症した患者の試料に由来するmiRNAの発現量を用いて、その患者がレスポンダー又はノンレスポンダーのいずれであるかを予測することにより、特定の薬剤の効果を予測することができる。
 さらに、第1及び第2の実施形態に係るバイオマーカーは、特定のmiRNAを薬剤の効果の予測に利用することができる。
〈第1の実施形態〉
 ここで、図1に示すように、第1の実施形態に係る予測装置1は、対象のmiRNAの発現量を入力データとし、学習済みモデルMを用いて対象のPFSを予測する。すなわち、ここでは、薬剤の効果を予測する値として、PFSを予測する例で説明する。予測装置1が用いる学習済みモデルMは、図2に示すように、訓練装置2によって生成される。したがって、まず、PFSの予測に用いる学習済みモデルの生成である学習について説明する。その後、実施形態に係るバイオマーカーについて説明し、続いて、実施形態に係るPFSの予測について説明する。具体的には、ニボルマブの肺癌患者への効果予測をする一例を用いて説明する。
〈学習処理〉
 図2に示すように、係る訓練装置2は、複数の患者の実データで構成される訓練データセットを用いて学習処理を実行し、学習済みモデルMを生成する。具体的には、訓練装置2は、患者の「特定のmiRNAの発現量」を説明変数とし、その患者の「PFS」を目的変数とする訓練データを複数組含む訓練データセットの関係を、機械学習により学習する。この「miRNAの発現量」は、ニボルマブを投与された患者から得られた血清から、一般的な方法により測定されたmiRNAの発現量である。具体的には、「240mgのニボルマブを2週間毎に投与する患者」または「480mgのニボルマブを4週間毎に投与する患者」を対象とし、採血は、ニボルマブの投与開始から1ヶ月以内に行う。ニボルマブの投与期間は、PFSと同じ期間である。miRNAの発現量は、次世代シーケンサーによって測定された。
 図3に示すように、訓練装置2は、学習に用いる訓練データセット23を取得するデータ取得部21と、取得した訓練データセット23により学習を実行する学習部22とを含む。また、訓練装置2は、記憶装置において、取得した訓練データセット23と、訓練装置2における学習処理の実行に用いられるプログラムP1と、学習部22により生成された学習済みモデルMとを記憶する。学習部22においては、例えば、ランダムフォレスト、ニューラルネットワーク又は決定木等の統計学的手法を利用することができる。なお、今回は、一例として、ランダムフォレストを利用し、5分割交差検証により学習処理を行ったが、他の機械学習の手法を用いてもよい。
 ここで、訓練装置2は、異なる複数の組み合わせのmiRNAの発現量と、PFSとの関係をそれぞれ学習することができる。例えば、訓練装置2は、ある5種のmiRNAの発現量と、PFSとの関係を学習した学習済みモデルM5を生成することができる。また、訓練装置2は、ある12種のmiRNAの発現量と、PFSとの関係を学習した学習済みモデルM12を生成することができる。その他、訓練装置2は、任意の種類xのmiRNAの発現量と、PFSとを訓練データとし、これらmiRNAの発現量とPFSとの関係を学習した学習済みモデルMxを生成することができる。なお、特に区別しない場合には、単に学習済みモデルMとする。
《miRNAの選択》
 ここで学習済みモデルMの生成に発現量を利用するmiRNAの種類は、例えば、以下で図4のフローチャートを用いて説明するような分散分析等の分析方法を用いて選択することができる。分析で用いられる分析用のデータは、ある疾患を発症した複数の対象から取得されたmiRNAの発現量と、PFSとが関連付けられるデータであって、いずれも実測値である。なお、ここでは、訓練装置2においてmiRNAの選択の処理を実行するものとして説明するが、これに限定されず、他の装置においてmiRNAの選択が実行され、訓練装置2ではその選択結果に基づいて学習処理を実行するものであってもよい。
 図4に示すように、まず、訓練装置2は、各miRNAの種別について、Nの値を初期化する(S01)。具体的には、Nの値を0に設定する。
 次に、訓練装置2は、分析用のデータから対象を、設定した閾値毎に、PFSが『長いグループ』又は『短いグループ』に分ける(S02)。閾値は、PFSの単位を「月」とした例であり、例えば、閾値が「3」である場合、『長いグループ』は、「PFSが3ヶ月以上のグループ」であり、『短いグループ』は、「PFSが3ヶ月未満のグループ」となる。ここでは、閾値を、3,6,9,12,15,18,21及び24の8パターンについて順番に行う例で説明する。したがって、初めに閾値「3」に設定されており、PFSが「3カ月以上のグループ」と「3カ月未満のグループ」とに分ける。そして、その後の処理で、閾値「6」に設定する。なお、ここでは、上述の8パターンの閾値を用いた例で説明するが、上述の処理及び順序は一例である。
 続いて、訓練装置2は、ステップS02で分けた長いグループ及び短いグループのそれぞれから、miRNAの種別毎に、miRNAの発現量が外れ値となる対象を除く(S03)。具体的には、訓練装置2は、miRNAの種別毎に、miRNAの発現量の中央値を特定する。そして、各対象のmiRNAの発現量について、中央値について所定の範囲(例えば、「中央絶対偏差>3」)となる外れ値を除く。
 次に、訓練装置2は、miRNAの種別毎に、ステップS03で残った対象の長いグループのmiRNAの発現量と、短いグループのmiRNAの発現量を用いて、F値を算出する(S04)。F値は、群内分散を分母、群間分散を分子として算出することができる。
 続いて、訓練装置2は、ステップS04で算出されたF値を用いて、miRNAの種別毎に、p値を算出する(S05)。
 訓練装置2は、ステップS05で算出したp値が所定値未満(例えば、p値<0.05)である場合(S06でYES)、この閾値のパターンにおいて、このmiRNAが有効であるとして、このmiRNAについてNの値をインクリメントする(S07)。ここでインクリメントされたNの値は、各miRNAのNの値として記憶装置で記憶される。
 続いて、訓練装置2は、S02でグループ分けに使用されていない未処理の閾値があるか否かを判定する(S08)。未処理の閾値がある場合(S08でYES)、ステップS01に戻り、Nの値を初期化してステップS01乃至06の処理を繰り返す。
 全ての閾値でグループを設定して、ステップS01乃至S06の処理が終了した場合(S08でNO)、訓練装置2は、各miRNAのNの値を参照し、Nの値が上位のmiRNAを選択する(S09)。そして、ここで選択されたmiRNAの発現量が、予測装置1で使用する学習済みモデルMへの入力データとなる。
 ここで、上位のmiRNAの決定方法としては、様々な方法があるが、図5は、Nの値が8以上のグループ、4以上のグループ、3以上のグループで選択するmiRNAの一例である。なお、図5において、miRNAの種別の後ろに括弧で付される番号は、そのmiRNAについてのNの値である。例えば、実線で囲んだ上位5種のmiRNAは、全てNの値が8であり、8パターンの全てについてp値<0.05となったものである。また、破線で囲んだ上位12種のmiRNAは、Nの値が8~4であり、少なくとも4パターンについてp値<0.05となったものである。さらに、一点鎖線で囲んだ上位19種のmiRNAは、Nの値が8~3であり、少なくとも3パターンについてp値<0.05となったものである。
 なお、その他、これら19種のmiRNAを含め、合計で2576種のmiRNAについて同様にステップS01~S09の処理を実行してp値を求めた。その結果、Nが2となったmiRNAは、hsa-miR-671-5p,hsa-miR-6087,hsa-miR-574-3p,hsa-miR-483-3p,hsa-miR-454-3p,hsa-miR-382-5p,hsa-miR-320d,hsa-miR-320b,hsa-miR-30b-5p,hsa-miR-3065-5p,hsa-miR-224-5p,hsa-miR-223-3p,hsa-miR-191-5p,hsa-miR-146b-5p,hsa-miR-135b-5p,hsa-miR-125b-5p,hsa-miR-101-3pの17種であった。また、Nが1となったmiRNAは、hsa-miR-92b-3p,hsa-miR-652-3p,hsa-miR-590-3p,hsa-miR-423-3p,hsa-miR-378c,hsa-miR-339-5p,hsa-miR-335-5p,hsa-miR-335-3p,hsa-miR-320a,hsa-miR-30e-3p,hsa-miR-29a-3p,hsa-miR-26b-5p,hsa-miR-221-3p,hsa-miR-215-5p,hsa-miR-2110,hsa-miR-206,hsa-miR-196a-5p,hsa-miR-190a-5p,hsa-miR-183-5p,hsa-miR-145-3p,hsa-miR-130a-3p,hsa-miR-10b-5p,hsa-miR-106b-5p,hsa-miR-1,hsa-let-7f-5pの25種であった。なお、それ以外の2515種のmiRNAについては、各パターンでp値が所定値未満とならなかった。
 図6は、各miRNAについて得られたスコアのグラフを示す。具体的には、図6に示すグラフは、各miRNAについて、8パターンでのF値の合計値をスコアとしたグラフである。図6に示す上位の5種は、選択された図5の上位の5種と同一である。
 ここで、例えば、上位5種のmiRNAであるhsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p及びhsa-miR-100-5pについては、ステップS02で設定された8パターンの閾値について、全てp値<0.05となったものであり(N≧8)、PFSの予測に有用であるとして選択される。また、これら5種のmiRNAは、バイオマーカーに有用であると予想される。
 上位12種であるhsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p,hsa-miR-100-5p,hsa-miR-3679-5p,hsa-miR-200c-3p,hsa-miR-181a-5p,hsa-miR-4516,hsa-miR-21-5p,hsa-miR-452-5p及びhsa-miR-320eについては、4パターン以上の閾値について、p値<0.05となったものであり(N≧4)、PFSの予測に利用可能であるとして選択される。また、これら12種のmiRNAは、バイオマーカーに有用であると予想される。
 上位19種であるhsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p,hsa-miR-100-5p,hsa-miR-3679-5p,hsa-miR-200c-3p,hsa-miR-181a-5p,hsa-miR-4516,hsa-miR-21-5p,hsa-miR-452-5p,hsa-miR-320e,hsa-miR-99a-5p,hsa-miR-98-5p,hsa-miR-4429,hsa-miR-378a-3p,hsa-miR-22-3p,hsa-miR-1301-3p及びhsa-miR-1246についても、3種以上の閾値について、p値<0.05となったものであり(N≧3)、PFSの予測に利用可能であるとして選択される。また、これら12種のmiRNAはバイオマーカーに有用であると予想される。
 また、これは、各miRNAをバイオマーカーとした場合のPFSの分布をみても明らかである。図7Aに示す例は、hsa-miR-429をバイオマーカーとした場合のPFSの分布である。それぞれ、ある期間を閾値Tとし、対象を、この閾値T毎にPFSが『長いグループ(long)』又は『短いグループ(short)』に分けた場合の分布である。左上から、閾値Tが3カ月、6カ月、9カ月、12カ月、18カ月、24カ月と設定された場合の分布を示す。
 図7Bに示す例は、hsa-miR-200b-3pをバイオマーカーとした場合のPFSの分布である。図7Bにおいても、閾値Tが3カ月、6カ月、9カ月、12カ月、18カ月、24カ月毎の、PFSが『長いグループ』又は『短いグループ』に分けた場合の分布である。図7Cに示す例も、同様に、hsa-miR-200a-3pをバイオマーカーとした場合のPFSの分布である。また、図7Dに示す例も、hsa-miR-100-5pをバイオマーカーとした場合の同様のPFSの分布である。さらに、図7Eに示す例も、同様に、hsa-miR-141-3pをバイオマーカーとした場合の同様のPFSの分布である。
 各閾値のパターンで『長いグループ』と『短いグループ』とで分布の傾向が同様に表れると精度の高いバイオマーカーといえる。したがって、図7A乃至図7Eの例に示す上位5種のmiRNAについては、各閾値のパターンで『長いグループ』と『短いグループ』との分布が同様に表れたため、精度の高いバイオマーカーであると考えられる。
〈バイオマーカー〉
 上述したように、特定のmiRNAをバイオマーカーとすることができる。具体的には、疾患に対する薬剤の効果の予測に用いるバイオマーカーとすることができる。
 疾患の一例としては、肺癌、大腸癌、胃癌、膵臓癌、肝臓癌、乳癌、腎臓癌、頭頸部癌、皮膚癌、悪性リンパ腫癌、食道癌、膀胱癌、尿管癌、免疫反応に対する疾患等が挙げられる。好ましくは、肺癌に対する薬剤の効果の予測に用いるバイオマーカーとすることができる。
 薬剤としては、例えば、抗がん剤等が挙げられる。好ましくは、薬剤の一例である抗がん剤は、免疫チェックポイント阻害剤とすることができる。免疫チェックポイント阻害剤としては、抗PD-1抗体、抗PD-L1抗体、抗CTLA-4抗体等が挙げられる。例えば、抗PD-1抗体としては、ニボルマブ、ペムブロリズマブ、スパルタリズマブ、セミプリマブ等が挙げられる。また例えば、抗PD-L1抗体としては、アベルマブ、アテゾリズマブ、デュルバルマブ等が挙げられる。
 特定のmiRNAは、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p及びhsa-miR-100-5pからなる群から選択される少なくとも1つのmiRNAである。
 または、特定のmiRNAは、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p,hsa-miR-100-5p,hsa-miR-3679-5p,hsa-miR-200c-3p,hsa-miR-181a-5p,hsa-miR-4516,hsa-miR-21-5p,hsa-miR-452-5p及びhsa-miR-320eからなる群から選択される少なくとも1つのmiRNAであってもよい。
 または、特定のmiRNAは、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p,hsa-miR-100-5p,hsa-miR-3679-5p,hsa-miR-200c-3p,hsa-miR-181a-5p,hsa-miR-4516,hsa-miR-21-5p,hsa-miR-452-5p,hsa-miR-320e,hsa-miR-99a-5p,hsa-miR-98-5p,hsa-miR-4429,hsa-miR-378a-3p,hsa-miR-22-3p,hsa-miR-1301-3p及びhsa-miR-1246からなる群から選択される少なくとも1つのmiRNAであってもよい。
 または、特定のmiRNAは、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p又はhsa-miR-100-5pの少なくとも1つと、さらに、hsa-miR-3679-5p,hsa-miR-200c-3p,hsa-miR-181a-5p,hsa-miR-4516,hsa-miR-21-5p,hsa-miR-452-5p及びhsa-miR-320eからなる群から選択される1以上のmiRNAであってもよい。
 または、特定のmiRNAは、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p又はhsa-miR-100-5pの少なくとも1つと、さらにhsa-miR-3679-5p,hsa-miR-200c-3p,hsa-miR-181a-5p,hsa-miR-4516,hsa-miR-21-5p,hsa-miR-452-5p及びhsa-miR-320eからなる群から選択される1以上のmiRNAであってもよい。
 または、特定のmiRNAは、hsa-miR-100-5p と、さらにhsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p及びhsa-miR-141-3pからなる群から選択される1以上のmiRNAであってもよい。
 または、特定のmiRNAは、hsa-miR-100-5p 及びhsa-miR-200a-3pと、さらにhsa-miR-429,hsa-miR-200b-3p及びhsa-miR-141-3pからなる群から選択される1以上のmiRNAであってもよい。
 または、特定のmiRNAは、hsa-miR-100-5p ,hsa-miR-200a-3p及び hsa-miR-141-3pと、さらにhsa-miR-429及びhsa-miR-200b-3pからなる群から選択される1以上のmiRNAであってもよい。
 例えば、1、2、3、4、5、6、7、8、9、10、11、12、13、14、15、16、17、18、または19種類のmiRNAをバイオマーカーとして用いることができる。
 miRNAの配列としては、以下を一例とする。
Figure JPOXMLDOC01-appb-T000001
 例えば、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p及びhsa-miR-100-5pの5種の群から選択された少なくとも1以上のmiRNAを、肺癌を発症した患者に対するニボルマブの薬効の判定のバイオマーカーとすることができる。例えば、上記の5種から選択された1,2,3,4又は5種のmiRNAをバイオマーカーとすることができる。
 また例えば、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p,hsa-miR-100-5p,hsa-miR-3679-5p,hsa-miR-200c-3p,hsa-miR-181a-5p,hsa-miR-4516,hsa-miR-21-5p,hsa-miR-452-5p及びhsa-miR-320eの12種の群から選択された少なくとも1以上のmiRNAを、肺癌を発症した患者に対するニボルマブの薬効の判定のバイオマーカーとすることができる。例えば、上記の12種から選択された1,2,3,4,5,6,7,8,9,10,11又は12種のmiRNAをバイオマーカーとすることができる。
 さらに、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p,hsa-miR-100-5p,hsa-miR-3679-5p,hsa-miR-200c-3p,hsa-miR-181a-5p,hsa-miR-4516,hsa-miR-21-5p,hsa-miR-452-5p,hsa-miR-320e,hsa-miR-99a-5p,hsa-miR-98-5p,hsa-miR-4429,hsa-miR-378a-3p,hsa-miR-22-3p,hsa-miR-1301-3p及びhsa-miR-1246の19種の群から選択された少なくとも1以上のmiRNAを、肺癌を発症した患者に対するニボルマブの薬効の判定のバイオマーカーとすることができる。例えば、上記の19種から選択された1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18又は19種のmiRNAをバイオマーカーとすることができる。
〈予測処理〉
 予測処理においては、図1に上述した予測装置1を用いる。具体的には、学習済みモデルMを用いて、対象の試料から計測されたmiRNAの発現量を入力データとし、対象のPFSを予測する。図8に示すように、予測装置1は、対象から採取された血液から計測された特定のmiRNAの発現量を示す発現量データ14を取得する取得部11と、取得した発現量データ14から学習済みモデルM1を用いてPFSを予測する予測部12と、予測したPFSを出力処理する出力部13とを備える。また、予測装置1は、記憶装置に、発現量データ14と、予測装置1を実行するプログラムPと、学習済みモデルM1とを記憶する。
 図9のフローチャートを用いて、予測処理について説明する。まず、対象の血液を採取する(S11)。ここで、試料の採取のタイミングは、治療開始前の所定のタイミングである。具体的には、薬剤の投与開始の1ヶ月前から、投与開始までのいずれかのタイミングである。
 続いて、ステップS11で採取された血液からの特定のmiRNAの発現量を計測する(S12)。
 その後、予測装置1において処理が実行される。図9のフローチャートにおいて、破線部が予測装置1において実行される処理である。具体的には、取得部11は、ステップS12で計測された特定のmiRNAの発現量を取得する(S13)。
 続いて、予測部12は、ステップS13で取得したmiRNAの発現量を用いて、学習済みモデルMによりPFSを予測する(S14)。
 また、出力部13は、ステップS14で予測したPFSを出力する(S15)。
 このように、予測装置1は、対象の特定のmiRNAの発現量を用いて、その対象のPFSを予測することができる。したがって、例えば、予測装置1による予測結果は、医師の診断等を補助することができる。
 前述した実施形態における各装置(予測装置1又は訓練装置)の一部又は全部は、ハードウェアで構成されていてもよいし、CPU(Central Processing Unit)、GPU(Graphics Processing Unit)等が実行するソフトウェア(プログラム)の情報処理で構成されてもよい。ソフトウェアの情報処理で構成される場合には、前述した実施形態における各装置の少なくとも一部の機能を実現するソフトウェアを、フレキシブルディスク、CD-ROM(Compact Disc-Read Only Memory)、USB(Universal Serial Bus)メモリ等の非一時的な記憶媒体(非一時的なコンピュータ可読媒体)に収納し、コンピュータに読み込ませることにより、ソフトウェアの情報処理を実行してもよい。また、通信ネットワークを介して当該ソフトウェアがダウンロードされてもよい。さらに、ソフトウェアがASIC(Application Specific Integrated Circuit)、FPGA(Field Programmable Gate Array)等の回路に実装されることにより、情報処理がハードウェアにより実行されてもよい。
 ソフトウェアを収納する記憶媒体の種類は限定されるものではない。記憶媒体は、磁気ディスク、光ディスク等の着脱可能なものに限定されず、ハードディスク、メモリ等の固定型の記憶媒体であってもよい。また、記憶媒体は、コンピュータ内部に備えられてもよいし(主記憶装置、補助記憶装置等)、コンピュータ外部に備えられてもよい。
 図10は、前述した実施形態における各装置(予測装置1又は訓練装置2)のハードウェア構成の一例を示すブロック図である。各装置は、一例として、プロセッサ71と、主記憶装置72(メモリ)と、補助記憶装置73(メモリ)と、ネットワークインタフェース74と、デバイスインタフェース75と、を備え、これらがバス76を介して接続されたコンピュータ7として実現されてもよい。
 図10のコンピュータ7は、各構成要素を一つ備えているが、同じ構成要素を複数備えていてもよい。また、図10では、1台のコンピュータ7が示されているが、ソフトウェアが複数台のコンピュータにインストールされて、当該複数台のコンピュータそれぞれがソフトウェアの同一の又は異なる一部の処理を実行してもよい。この場合、コンピュータそれぞれがネットワークインタフェース74等を介して通信して処理を実行する分散コンピューティングの形態であってもよい。つまり、前述した実施形態における各装置(予測装置1又は訓練装置2)は、1又は複数の記憶装置に記憶された命令を1台又は複数台のコンピュータが実行することで機能を実現するシステムとして構成されてもよい。また、端末から送信された情報をクラウド上に設けられた1台又は複数台のコンピュータで処理し、この処理結果を端末に送信するような構成であってもよい。
 前述した実施形態における各装置(予測装置1又は訓練装置2)の各種演算は、1又は複数のプロセッサを用いて、又は、ネットワークを介した複数台のコンピュータを用いて、並列処理で実行されてもよい。また、各種演算が、プロセッサ内に複数ある演算コアに振り分けられて、並列処理で実行されてもよい。また、本開示の処理、手段等の一部又は全部は、ネットワークを介してコンピュータ7と通信可能なクラウド上に設けられたプロセッサ及び記憶装置の少なくとも一方により実行されてもよい。このように、前述した実施形態における各装置は、1台又は複数台のコンピュータによる並列コンピューティングの形態であってもよい。
プロセッサ71は、少なくともコンピュータの制御又は演算のいずれかを行う電子回路(処理回路、Processing circuit、Processing circuitry、CPU、GPU、FPGA、ASIC等)であってもよい。また、プロセッサ71は、汎用プロセッサ、特定の演算を実行するために設計された専用の処理回路又は汎用プロセッサと専用の処理回路との両方を含む半導体装置のいずれであってもよい。また、プロセッサ71は、光回路を含むものであってもよいし、量子コンピューティングに基づく演算機能を含むものであってもよい。
 プロセッサ71は、コンピュータ7の内部構成の各装置等から入力されたデータやソフトウェア(プログラム)に基づいて演算処理を行い、演算結果や制御信号を各装置等に出力することができる。プロセッサ71は、コンピュータ7のOS(Operating System)や、アプリケーション等を実行することにより、コンピュータ7を構成する各構成要素を制御してもよい。
 前述した実施形態における各装置(予測装置1又は訓練装置2)は、1又は複数のプロセッサ71により実現されてもよい。ここで、プロセッサ71は、1チップ上に配置された1又は複数の電子回路を指してもよいし、2つ以上のチップあるいは2つ以上のデバイス上に配置された1又は複数の電子回路を指してもよい。複数の電子回路を用いる場合、各電子回路は有線又は無線により通信してもよい。
 主記憶装置72は、プロセッサ71が実行する命令及び各種データ等を記憶する記憶装置であり、主記憶装置72に記憶された情報がプロセッサ71により読み出される。補助記憶装置73は、主記憶装置72以外の記憶装置である。なお、これらの記憶装置は、電子情報を格納可能な任意の電子部品を意味するものとし、半導体のメモリでもよい。半導体のメモリは、揮発性メモリ、不揮発性メモリのいずれでもよい。前述した実施形態における各装置(予測装置1又は訓練装置2)において各種データを保存するための記憶装置は、主記憶装置72又は補助記憶装置73により実現されてもよく、プロセッサ71に内蔵される内蔵メモリにより実現されてもよい。例えば、前述した実施形態における記憶部102は、主記憶装置72又は補助記憶装置73により実現されてもよい。
 記憶装置(メモリ)1つに対して、複数のプロセッサが接続(結合)されてもよいし、単数のプロセッサが接続されてもよい。プロセッサ1つに対して、複数の記憶装置(メモリ)が接続(結合)されてもよい。前述した実施形態における各装置(予測装置1又は訓練装置2)が、少なくとも1つの記憶装置(メモリ)とこの少なくとも1つの記憶装置(メモリ)に接続(結合)される複数のプロセッサで構成される場合、複数のプロセッサのうち少なくとも1つのプロセッサが、少なくとも1つの記憶装置(メモリ)に接続(結合)される構成を含んでもよい。また、複数台のコンピュータに含まれる記憶装置(メモリ))とプロセッサによって、この構成が実現されてもよい。さらに、記憶装置(メモリ)がプロセッサと一体になっている構成(例えば、L1キャッシュ、L2キャッシュを含むキャッシュメモリ)を含んでもよい。
 ネットワークインタフェース74は、無線又は有線により、通信ネットワーク8に接続するためのインタフェースである。ネットワークインタフェース74は、既存の通信規格に適合したもの等、適切なインタフェースを用いればよい。ネットワークインタフェース74により、通信ネットワーク8を介して接続された外部装置9Aと情報のやり取りが行われてもよい。なお、通信ネットワーク8は、WAN(Wide Area Network)、LAN(Local Area Network)、PAN(Personal Area Network)等の何れか、又は、それらの組み合わせであってよく、コンピュータ7と外部装置9Aとの間で情報のやり取りが行われるものであればよい。WANの一例としてインターネット等があり、LANの一例としてIEEE802.11やイーサネット(登録商標)等があり、PANの一例としてBluetooth(登録商標)やNFC(Near Field Communication)等がある。
 デバイスインタフェース75は、外部装置9Bと直接接続するUSB等のインタフェースである。
 外部装置9Aはコンピュータ7とネットワークを介して接続されている装置である。外部装置9Bはコンピュータ7と直接接続されている装置である。
 外部装置9A又は外部装置9Bは、一例として、入力装置であってもよい。入力装置は、例えば、カメラ、マイクロフォン、モーションキャプチャ、各種センサ、キーボード、マウス、タッチパネル等のデバイスであり、取得した情報をコンピュータ7に与える。また、パーソナルコンピュータ、タブレット端末、スマートフォン等の入力部とメモリとプロセッサを備えるデバイスであってもよい。
 また、外部装置9A又は外部装置9Bは、一例として、出力装置でもよい。出力装置は、例えば、LCD(Liquid Crystal Display)、CRT(Cathode Ray Tube)、PDP(Plasma Display Panel)、有機EL(Electro Luminescence)パネル等の表示装置であってもよいし、音声等を出力するスピーカ等であってもよい。また、パーソナルコンピュータ、タブレット端末、スマートフォン等の出力部とメモリとプロセッサを備えるデバイスであってもよい。
 また、外部装置9Aまた外部装置9Bは、記憶装置(メモリ)であってもよい。例えば、外部装置9Aはネットワークストレージ等であってもよく、外部装置9BはHDD等のストレージであってもよい。
 また、外部装置9A又は外部装置9Bは、前述した実施形態における各装置(予測装置1、又は訓練装置2)の構成要素の一部の機能を有する装置でもよい。つまり、コンピュータ7は、外部装置9A又は外部装置9Bの処理結果の一部又は全部を送信又は受信してもよい。
 本明細書(請求項を含む)において、「a、b及びcの少なくとも1つ(一方)」又は「a、b又はcの少なくとも1つ(一方)」の表現(同様な表現を含む)が用いられる場合は、a、b、c、a-b、a-c、b-c又はa-b-cのいずれかを含む。また、a-a、a-b-b、a-a-b-b-c-c等のように、いずれかの要素について複数のインスタンスを含んでもよい。さらに、a-b-c-dのようにdを有する等、列挙された要素(a、b及びc)以外の他の要素を加えることも含む。
本明細書(請求項を含む)において、「データを入力として/を用いて/データに基づいて/に従って/に応じて」等の表現(同様な表現を含む)が用いられる場合は、特に断りがない場合、データそのものを用いる場合や、データに何らかの処理を行ったもの(例えば、ノイズ加算したもの、正規化したもの、データから抽出した特徴量、データの中間表現等)を用いる場合を含む。また、「データを入力として/を用いて/データに基づいて/に従って/に応じて」何らかの結果が得られる旨が記載されている場合(同様な表現を含む)、特に断りがない場合、当該データのみに基づいて当該結果が得られる場合や、当該データ以外の他のデータ、要因、条件及び/又は状態にも影響を受けて当該結果が得られる場合を含む。また、「データを出力する」旨が記載されている場合(同様な表現を含む)、特に断りがない場合、データそのものを出力として用いる場合や、データに何らかの処理を行ったもの(例えば、ノイズ加算したもの、正規化したもの、データから抽出した特徴量、各種データの中間表現等)を出力として用いる場合を含む。
 本明細書(請求項を含む)において、「接続される(connected)」及び「結合される(coupled)」との用語が用いられる場合は、直接的な接続/結合、間接的な接続/結合、電気的(electrically)な接続/結合、通信的(communicatively)な接続/結合、機能的(operatively)な接続/結合、物理的(physically)な接続/結合等のいずれをも含む非限定的な用語として意図される。当該用語は、当該用語が用いられた文脈に応じて適宜解釈されるべきであるが、意図的に或いは当然に排除されるのではない接続/結合形態は、当該用語に含まれるものして非限定的に解釈されるべきである。
 本明細書(請求項を含む)において、「AがBするよう構成される(A configured to B)」との表現が用いられる場合は、要素Aの物理的構造が、動作Bを実行可能な構成を有するとともに、要素Aの恒常的(permanent)又は一時的(temporary)な設定(setting/configuration)が、動作Bを実際に実行するように設定(configured/set)されていることを含んでよい。例えば、要素Aが汎用プロセッサである場合、当該プロセッサが動作Bを実行可能なハードウェア構成を有するとともに、恒常的(permanent)又は一時的(temporary)なプログラム(命令)の設定により、動作Bを実際に実行するように設定(configured)されていればよい。また、要素Aが専用プロセッサ又は専用演算回路等である場合、制御用命令及びデータが実際に付属しているか否かとは無関係に、当該プロセッサの回路的構造が動作Bを実際に実行するように構築(implemented)されていればよい。
 本明細書(請求項を含む)において、含有又は所有を意味する用語(例えば、「含む(comprising/including)」及び有する「(having)等)」が用いられる場合は、当該用語の目的語により示される対象物以外の物を含有又は所有する場合を含む、open-endedな用語として意図される。これらの含有又は所有を意味する用語の目的語が数量を指定しない又は単数を示唆する表現(a又はanを冠詞とする表現)である場合は、当該表現は特定の数に限定されないものとして解釈されるべきである。
 本明細書(請求項を含む)において、ある箇所において「1つ又は複数(one or more)」又は「少なくとも1つ(at least one)」等の表現が用いられ、他の箇所において数量を指定しない又は単数を示唆する表現(a又はanを冠詞とする表現)が用いられているとしても、後者の表現が「1つ」を意味することを意図しない。一般に、数量を指定しない又は単数を示唆する表現(a又はanを冠詞とする表現)は、必ずしも特定の数に限定されないものとして解釈されるべきである。
 本明細書において、ある実施例の有する特定の構成について特定の効果(advantage/result)が得られる旨が記載されている場合、別段の理由がない限り、当該構成を有する他の1つ又は複数の実施例についても当該効果が得られると理解されるべきである。但し当該効果の有無は、一般に種々の要因、条件、状態等に依存し、当該構成により必ず当該効果が得られるものではないと理解されるべきである。当該効果は、種々の要因、条件、状態等が満たされたときに実施例に記載の当該構成により得られるものに過ぎず、当該構成又は類似の構成を規定したクレームに係る発明において、当該効果が必ずしも得られるものではない。
 本明細書(請求項を含む)において、複数のハードウェアが所定の処理を行う場合、各ハードウェアが協働して所定の処理を行ってもよいし、一部のハードウェアが所定の処理の全てを行ってもよい。また、一部のハードウェアが所定の処理の一部を行い、別のハードウェアが所定の処理の残りを行ってもよい。本明細書(請求項を含む)において、「1又は複数のハードウェアが第1の処理を行い、前記1又は複数のハードウェアが第2の処理を行う」等の表現(同様な表現を含む)が用いられている場合、第1の処理を行うハードウェアと第2の処理を行うハードウェアは同じものであってもよいし、異なるものであってもよい。つまり、第1の処理を行うハードウェア及び第2の処理を行うハードウェアが、前記1又は複数のハードウェアに含まれていればよい。なお、ハードウェアは、電子回路、電子回路を含む装置等を含んでよい。
本明細書(請求項を含む)において、複数の記憶装置(メモリ)がデータの記憶を行う場合、複数の記憶装置のうち個々の記憶装置は、データの一部のみを記憶してもよいし、データの全体を記憶してもよい。また、複数の記憶装置のうち一部の記憶装置がデータを記憶する構成を含んでもよい。
〈実施例1〉
 図11に、学習済みモデルMを用いた予測装置1における予測結果の精度を示す。図11に示す例は、肺癌を発症した複数の患者に、それぞれニボルマブを投与した場合の治療後の安定した状態の期間に関する情報であるPFSを対象とする。各患者に投与したニボルマブの量は、PFSと同じ期間であり、投与期間は、2週間毎に240mg又は4週間毎に480mgであり、ニボルマブの投与は効果が表れる間継続されるため投与から採血の期間は、PFSと同じ期間であり、対象とした患者の数は、213人である。図11では、具体的には上述した方法で選択した上位5種のmiRNAの発現量を入力とするモデルM5、上位12種のmiRNAの発現量を入力とするモデルM12、上位19種のmiRNAの発現量を入力とするモデルM19、上位36種のmiRNAの発現量を入力とするモデルM36、上位61種のmiRNAの発現量を入力とするモデルM61、2576種のmiRNAの発現量を入力とするモデルM2576の予測結果の精度を示す。図11では、実測値である複数の患者の実際のPFSと、予測装置1で各患者のmiRNAの発現量を用いた予測結果のPFSとについての、平均二乗誤差(mse)、決定係数(R2)、スピアマンの相関係数(spearman_r)、スピアマンのP値(spearman_p)、ピアソンの相関係数(pearson_r)、ピアソンのP値(pearson_p)を示す。図11に示す例では、平均二乗誤差において、上位12種のmiRNAの発現量を入力とするモデルM12が最も精度が高いという結果が得られた。
〈実施例2〉
 図11に示した実施例1では、上位12種のmiRNAの発現量を用いたPFSの予測結果の精度が最も高い結果となった。実施例2では、この上位12種のmiRNAの中から、PFSの予測の精度の向上に影響を与えるmiRNAを探索した。
 図12は、それぞれ1種のmiRNA(hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p及びhsa-miR-100-5p)の発現量を用いた予測結果であるPFSについての、平均二乗誤差(mse)、決定係数(R2)、スピアマンの相関係数(spearman_r)、スピアマンのP値(spearman_p)、ピアソンの相関係数(pearson_r)、ピアソンのP値(pearson_p)を示す。
 また、図13は、上位2種のmiRNA(hsa-miR-141-3p及びhsa-miR-429)、上位3種のmiRNA(hsa-miR-141-3p,hsa-miR-429及びhsa-miR-200a-3p)、上位4種のmiRNA(hsa-miR-141-3p,hsa-miR-429,hsa-miR-200a-3p及びhsa-miR-200b-3p)及び上位5種のmiRNA(hsa-miR-141-3p,hsa-miR-429,hsa-miR-200a-3p,hsa-miR-200b-3p及びhsa-miR-100-5p)の発現量を用いた予測結果であるPFSについての、平均二乗誤差(mse)、決定係数(R2)、スピアマンの相関係数(spearman_r)、スピアマンのP値(spearman_p)、ピアソンの相関係数(pearson_r)、ピアソンのP値(pearson_p)を示す。なお、上位5種に使用したmiRNAは、図11の5種と同一であるが、得られた各値が同一ではないのは、ランダム性を有する学習方法を用いて再度モデルを学習しているためである。
 図14は、上位5種から、1種類ずつmiRNAを除いた4種のmiRNAの発現量でそれぞれPFSを予測した場合の平均二乗誤差(mse)、決定係数(R2)、スピアマンの相関係数(spearman_r)、スピアマンのP値(spearman_p)、ピアソンの相関係数(pearson_r)、ピアソンのP値(pearson_p)を示す。図12に示す例では、上位5種のmiRNAからhsa-miR-429を除いた、hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p及びhsa-miR-100-5pの4種のmiRNAの発現量を用いたPFSの予測精度についての平均二乗誤差(121.195953)が最も低く、予測精度が高いといえる。これは、図11に示す5種のmiRNAの発現量を用いた場合の平均二乗誤差(123.751727)、及び、図13に示す5種のmiRNAの発現量を用いた場合の平均二乗誤差(124.724570)と比較しても予測精度が高いことがわかる。
 図15は、図14でPFSの予測精度が高いとされた4種(hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p及びhsa-miR-100-5p)のmiRNAから、1種類ずつmiRNAを除いた3種のmiRNAの発現量でそれぞれPFSを予測した場合の平均二乗誤差(mse)、決定係数(R2)、スピアマンの相関係数(spearman_r)、スピアマンのP値(spearman_p)、ピアソンの相関係数(pearson_r)、ピアソンのP値(pearson_p)を示す。図15に示す例では、hsa-miR-100-5p ,hsa-miR-200b-3p,hsa-miR-200a-3p及びhsa-miR-141-3pの4種のmiRNAからhsa-miR-200b-3pを除いたhsa-miR-100-5p,hsa-miR-200a-3p及びhsa-miR-141-3pの3種のmiRNAの発現量を用いたPFSの予測精度についての平均二乗誤差(122.106030)が最も低く、予測精度が高いといえる。これも、図11に示す5種のmiRNAの発現量を用いた場合の平均二乗誤差(123.751727)、及び、図13に示す5種のmiRNAの発現量を用いた場合の平均二乗誤差(124.724570)と比較しても予測精度が高いことがわかる。
 図16は、図15でPFSの予測精度が高いとされた3種から、1種類ずつmiRNAを除いた2種のmiRNAの発現量でそれぞれPFSを予測した場合の平均二乗誤差(mse)、決定係数(R2)、スピアマンの相関係数(spearman_r)、スピアマンのP値(spearman_p)、ピアソンの相関係数(pearson_r)、ピアソンのP値(pearson_p)を示す。図16に示す例では、hsa-miR-100-5p,hsa-miR-200a-3p及びhsa-miR-141-3pの3種のmiRNAからhsa-miR-141-3pを除いたhsa-miR-100-5p及びhsa-miR-200a-3pの2種のmiRNAの発現量を用いたPFSの予測精度についての平均二乗誤差(124.974067)が最も高いといえる。この場合、5種のmiRNAの発現量を用いた場合の平均二乗誤差(123.751727, 124.724570)と比較すると、若干精度は劣るが、予測精度は、同程度に達している。
 また、図12乃至16に示す平均二乗誤差を比較すると、少なくともhsa-miR-100-5p又はhsa-miR-200a-3pのいずれかのmiRNAの発現量を利用した場合、利用しない場合と比較して、予測精度が高いことがわかる。
〈変形例1〉
 上述した予測方法では、学習済みモデルMを用いて予測する例で説明したが、各miRNAの発現量に対してそれぞれ所定のカットオフ値を予め設定し、対象の試料から測定された各miRNAの発現量が、各カットオフ値より大きい又は小さい等に応じて、薬剤の効果を予測するようにしてもよい。例えば、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p及びhsa-miR-100-5pの5種類のmiRNAに対して、それぞれの発現量のカットオフ値T1~T5を設定する。そして、各miRNAの発現量を、それぞれのカットオフ値と比較する。その結果、例えば、所定数以上(例えば、1以上、2以上、3以上、4以上又は全て)のmiRNAの発現量で条件を満たす場合、対象に対して薬剤の効果が得られるとしてもよい。
 その他、例えば、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p,hsa-miR-100-5p,hsa-miR-3679-5p,hsa-miR-200c-3p,hsa-miR-181a-5p,hsa-miR-4516,hsa-miR-21-5p,hsa-miR-452-5p及びhsa-miR-320eの発現量それぞれのカットオフ値T1~T12を設定してもよい。そして、各miRNAの発現量を、それぞれのカットオフ値と比較する。その結果、例えば、所定数以上のmiRNAの発現量で条件を満たす場合、対象に対して薬剤の効果が得られるとしてもよい。またこのとき、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p及びhsa-miR-100-5pの発現量の中から、条件を満たすものが所定数以上であり、かつ、hsa-miR-3679-5p,hsa-miR-200c-3p,hsa-miR-181a-5p,hsa-miR-4516,hsa-miR-21-5p,hsa-miR-452-5p及びhsa-miR-320eの発現量の中から条件を満たすものが所定数以上である場合、対象に対して薬剤の効果が得られるとしてもよい。
 例えば、hsa-miR-100-5pのmiRNAの発現量のカットオフ値Tと、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p及びhsa-miR-141-3pからなる群から選択された1種以上のmiRNAの発現量のカットオフ値Tを設定してもよい。対象の試料から測定されたこれら2以上のmiRNAの発現量を、各miRNAの発現量のカットオフ値と比較し、各miRNAの発現量で条件を満たす場合、対象に対して薬剤の効果が得られるとしてもよい。
 また例えば、hsa-miR-200a-3pのmiRNAの発現量のカットオフ値Tと、hsa-miR-100-5p,hsa-miR-429,hsa-miR-200b-3p及びhsa-miR-141-3pからなる群から選択された1種以上のmiRNAの発現量のカットオフ値Tを設定してもよい。対象の試料から測定されたこれら2以上のmiRNAの発現量を、各miRNAの発現量のカットオフ値と比較し、各miRNAの発現量で条件を満たす場合、対象に対して薬剤の効果が得られるとしてもよい。
 その他、hsa-miR-100-5p及びhsa-miR-200a-3pの2種のmiRNAの発現量のそれぞれのカットオフ値Tと、hsa-miR-429,hsa-miR-200b-3p及びhsa-miR-141-3pからなる群から選択された1種以上のmiRNAの発現量のカットオフ値Tを設定してもよい。対象の試料から測定されたこれら3以上のmiRNAの発現量を、各miRNAの発現量のカットオフ値と比較し、各miRNAの発現量で条件を満たす場合、対象に対して薬剤の効果が得られるとしてもよい。
 その他、特定の1以上のmiRNAの発現量のカットオフ値Tをそれぞれ設定してもよい。対象の試料から測定された1以上のmiRNAの発現量を、各miRNAの発現量のカットオフ値と比較し、各miRNAの発現量で条件を満たす場合、対象に対して薬剤の効果が得られるとしてもよい。なお、ここで、特定のmiRNAが複数ある場合、全てのmiRNAの発現量が各カットオフ値以上である又はカットオフ値以下である等の条件を満たすことが必ずしも要件とされる必要はなく、所定の数のmiRNAの発現量がカットオフ値以上である又はカットオフ値以下である等の条件を満たすものであってもよい。
〈第2の実施形態〉
 図17に示すように、第2の実施形態に係る予測装置1Aは、対象のmiRNAの発現量を入力データとし、学習済みモデルM2を用いて対象がレスポンダーであるかノンレスポンダーであるかを予測する。すなわち、ここでは、薬剤の効果を予測する値として、対象がレスポンダー又はノンレスポンダーのいずれであるかを予測する例で説明する。具体的には、ニボルマブの肺癌患者への効果予測をする例を用いて説明する。図17に示す例では、入力データは、miRNAの発現量とともに、対象の臨床情報を含む。したがって、予測装置1Aが利用する学習済みモデルM2は、図18に示すように、miRNAの発現量及び臨床情報を説明変数とし、レスポンダーであるかノンレスポンダーであるかを目的変数とするデータセットの関係を、訓練装置2Aにより学習されたものである。訓練装置2Aは、例えば、例えば、ランダムフォレスト、ニューラルネットワーク又は決定木等の統計学的手法を利用することができる。なお、今回は、一例として、決定木を利用するXgBoostにより学習処理を行ったが、他の機械学習の手法を用いてもよい。また、今回の学習済みモデルM2は、5分割交差検証を用いて評価されたものである。
 ここで、レスポンダーであるかノンレスポンダーであるかについては、例えば、医師により判断された結果である。また、臨床情報は、例えば、対象の『性別』、『喫煙歴』、『遺伝子変異』、『年齢』、『PD-L1』、『病理組織学的な分類』等である。
《学習処理》
 訓練装置2Aの構成については、図3を用いて上述した訓練装置2と同様であるため、図示を用いた説明を省略する。また、予測装置1Aの構成についても、図8を用いて上述した予測装置1と同様であるため、図示を用いた説明を省略する。ここで、学習に利用されたデータに応じて、学習済みモデルM2が異なるものとなる。一例として、訓練データセットが特定のmiRNAの発現量を含み、臨床情報を含まない場合、学習済みモデルM2は、その特定のmiRNAの発現量を入力データとしてレスポンダーであるかノンレスポンダーであるかを予測する。なお、予測装置1Aの記憶装置においては、臨床情報を記憶する。また、訓練装置2A及び予測装置1Aも、図10を用いて上述したハードウェア構成のコンピュータ7により実現されてもよい。
 実施形態2に係る予測装置1Aで用いる学習済みモデルM2の学習では、各臨床情報をバイナリデータに変換して利用した。例えば、性別は、男性であるか女性であるかを0と1とのバイナリデータとして扱った。また、喫煙歴は、喫煙者であるか否かを0と1とのバイナリデータとして扱った。遺伝子変異は、遺伝子変異があるか否を0か1かのバイナリデータとして扱った。さらに、年齢は、閾値となる年齢(例えば、75歳)を設け、閾値以上であるか否かに応じたバイナリデータで扱った。また、PD-L1は、、「1.NE(データが利用できない場合)」、「2.ネガティブ」、「3.ポジティブ」及び「4.ストロングポジティブ」の4つのカテゴリデータで扱った。具体的には、ネガティブ:0≦ x <1、ポジティブ:1≦ x <50、ストロングポジティブ:50 ≦ xである。病理組織学的な分類については、、「1.Sq」、「2.Non-sq」、「3.Other」の3つのカテゴリデータで扱った。Sqには組織学= Sqが含まれ、Non-sqには組織学= Ad、NSCLC、NSCLC-NOS、Ad-sq、Ad(ROS1)が含まれる。Otherには、残りのすべてのタイプの組織学が含まれる。これらの臨床情報は、例えば、問診の結果や医師の判断に基づいて得られた情報である。なお、各臨床情報を連続値として扱ってモデルを学習してもよい。
《miRNAの選択》
 第2の実施形態に係る予測装置1Aで発現量を利用するmiRNAの種類の選択も第1の実施形態で説明したのと同様の方法で選択することができる。第2の実施形態に係る予測装置1Aでは、第1の実施形態においてスコアが上位61種のmiRNA(図6に示したmiRNA)を利用する。
 図19は、対象の61種のmiRNAの発現量を用いて、対象がレスポンダー又はノンレスポンダーのいずれであるかを予測した結果である。61種類のmiRNAとは、hsa-miR-141-3p、hsa-miR-429、hsa-miR-200a-3p、hsa-miR-200b-3p、hsa-miR-100-5p、hsa-miR-200c-3p、hsa-miR-3679-5p、hsa-miR-135b-5p、hsa-miR-181a-5p、hsa-miR-21-5p、hsa-miR-4516、hsa-miR-320e、hsa-miR-452-5p、hsa-miR-4429、hsa-miR-99a-5p、hsa-miR-22-3p、hsa-miR-1246、hsa-miR-98-5p、hsa-miR-1301-3p、hsa-miR-378a-3p、hsa-miR-6087、hsa-miR-671-5p、hsa-miR-146b-5p、hsa-miR-224-5p、hsa-miR-3065-5p、hsa-miR-125b-5p、hsa-miR-382-5p、hsa-miR-223-3p、hsa-miR-320b、hsa-miR-483-3p、hsa-miR-454-3p、hsa-miR-320d、hsa-miR-101-3p、hsa-miR-30b-5p、hsa-miR-191-5p、hsa-miR-574-3p、hsa-miR-423-3p、hsa-miR-215-5p、hsa-miR-196a-5p、hsa-miR-335-3p、hsa-miR-206、hsa-miR-339-5p、hsa-miR-320a、hsa-let-7f-5p、hsa-miR-221-3p、hsa-miR-30e-3p、hsa-miR-183-5p、hsa-miR-378c、hsa-miR-92b-3p、hsa-miR-145-3p、hsa-miR-29a-3p、hsa-miR-652-3p、hsa-miR-2110、hsa-miR-10b-5p、hsa-miR-26b-5p、hsa-miR-335-5p、hsa-miR-130a-3p、hsa-miR-106b-5p、hsa-miR-1、hsa-miR-590-3p及びhsa-miR-190a-5pである。
 図19に示す例では、レスポンダー、ノンレスポンダー、及び、それぞれの加重平均の予測結果の正確度(Accuracy)、精度(Precision)、特異度(specificity)、感度 (sensitivity)、陰性的中率(Negative Predictive Value(NPV))、F値(F1-Score)、サンプル数(support)及びAUC(Area Under the Curve)を示す。
 また、図20は、図19と同様の61種のmiRNAの発現量に加え、臨床情報である年齢、性別、及び、喫煙歴を用いて、対象がレスポンダー 又はノンレスポンダーのいずれであるかを予測した結果である。図20に示す例では、レスポンダー、ノンレスポンダー、及び、それぞれの加重平均の正確度、精度、特異度、感度、陰性的中率(NPV)、F値、支持度、及び、AUCを示す。図20に示す正確度を図19に示す正確度と比較すると、年齢、性別、及び、喫煙歴の臨床情報を用いた場合、臨床情報を用いていない場合と比較して、正確度が向上したことが分かる。
 なお、図21に示す例は、miRNAの発現量と、臨床情報である性別、年齢及び喫煙歴に加え、他の臨床情報を用いた予測結果の正確度である。これによると、miRNAの発現量と、臨床情報として性別、年齢及び喫煙歴に加え、PD-L1を追加した場合、及び、病理組織学的な分類を追加した場合には、正確度はそれぞれ、0.782及び0.787となり、miRNAの発現量に臨床情報として年齢、性別、及び、喫煙歴を用いた場合の正確度0.792と比較して、低下した。しかしながら、miRNAの発現量と、臨床情報として年齢、性別、及び、喫煙歴に加え、遺伝子変異を追加した場合、病理組織学的な分類及び遺伝子変異を追加した場合、病理組織学的な分類、PD-L1及び遺伝子変異を追加した場合、PD-L1及び遺伝子変異を追加した場合には、正確度はそれぞれ0.796となり、向上した。
《バイオマーカー》
 したがって、例えば、上述した特定のmiRNAをバイオマーカーとすることができる。具体的には、第1の実施形態の場合と同様に、肺癌、大腸癌、胃癌、膵臓癌、肺癌、乳癌、腎臓癌、頭頸部癌、皮膚癌、悪性リンパ腫癌、食道癌等に対する薬剤の効果の予測に用いるバイオマーカーとすることができる。好ましくは、肺癌に対する薬剤の効果の予測に用いるバイオマーカーとすることができる。また、第1の実施形態の場合と同様に、薬剤は、免疫チェックポイント阻害剤とすることができる。
 特定のmiRNAは、hsa-miR-141-3p、hsa-miR-429、hsa-miR-200a-3p、hsa-miR-200b-3p、hsa-miR-100-5p、hsa-miR-200c-3p、hsa-miR-3679-5p、hsa-miR-135b-5p、hsa-miR-181a-5p、hsa-miR-21-5p、hsa-miR-4516、hsa-miR-320e、hsa-miR-452-5p、hsa-miR-4429、hsa-miR-99a-5p、hsa-miR-22-3p、hsa-miR-1246、hsa-miR-98-5p、hsa-miR-1301-3p、hsa-miR-378a-3p、hsa-miR-6087、hsa-miR-671-5p、hsa-miR-146b-5p、hsa-miR-224-5p、hsa-miR-3065-5p、hsa-miR-125b-5p、hsa-miR-382-5p、hsa-miR-223-3p、hsa-miR-320b、hsa-miR-483-3p、hsa-miR-454-3p、hsa-miR-320d、hsa-miR-101-3p、hsa-miR-30b-5p、hsa-miR-191-5p、hsa-miR-574-3p、hsa-miR-423-3p、hsa-miR-215-5p、hsa-miR-196a-5p、hsa-miR-335-3p、hsa-miR-206、hsa-miR-339-5p、hsa-miR-320a、hsa-let-7f-5p、hsa-miR-221-3p、hsa-miR-30e-3p、hsa-miR-183-5p、hsa-miR-378c、hsa-miR-92b-3p、hsa-miR-145-3p、hsa-miR-29a-3p、hsa-miR-652-3p、hsa-miR-2110、hsa-miR-10b-5p、hsa-miR-26b-5p、hsa-miR-335-5p、hsa-miR-130a-3p、hsa-miR-106b-5p、hsa-miR-1、hsa-miR-590-3p及びhsa-miR-190a-5pの61種の群から選択された少なくとも1以上のmiRNAである。
 miRNAの配列としては、以下を一例とする。
Figure JPOXMLDOC01-appb-T000002
Figure JPOXMLDOC01-appb-T000003
《予測処理》
 予測装置1Aを用いた予測処理について、図22に示すフローチャートを用いて説明する。ここでは、学習済みモデルM2が、特定のmiRNAの発現量及び臨床情報を入力データとする場合について説明する。したがって、予測処理は、学習済みモデルM2を用いて対象の試料から計測されたmiRNAの発現量及び対象の臨床情報を入力データとし、対象がレスポンダー又はノンレスポンダーのいずれであるかを予測する。
 まず、予測装置1Aにおける処理の前に、対象の血液を採取する(S21)。ここで、試料の採取のタイミングは、例えば、第1の実施形態で上述した場合と同様に、薬剤の投与開始の1ヶ月前から、投与開始までの間である。
 続いて、ステップS21で採取された血液からの特定のmiRNAの発現量を計測する(S22)。
 その後、予測装置1Aにおいて処理が実行される。図22のフローチャートにおいて、破線部分が、予測装置1Aにおいて実行される処理である。具体的には、予測装置1Aの取得部11は、ステップS22で計測された特定のmiRNAの発現量、及び、対象の臨床情報を取得する(S23)。対象の臨床情報は、例えば、入力装置を介してユーザによって入力され得る。
 続いて、予測装置1Aの予測部12は、ステップS23で取得したmiRNAの発現量及び臨床情報を用いて、学習済みモデルM2により対象がレスポンダー又はノンレスポンダーのいずれであるかを予測する(S24)。
 また、予測装置1Aの出力部13は、ステップS24で予測した対象がレスポンダーであるか又はノンレスポンダーであるかの予測結果を出力する(S25)。
 このように、予測装置1Aは、特定の対象のmiRNAを用いて、その対象がレスポンダー又はノンレスポンダーのいずれであるかを予測することができる。また、miRNAの発現量とともに、対象がレスポンダー又はノンレスポンダーのいずれであるかの予測に、その対象の特定の臨床情報を用いて追加した場合、予測精度が向上することがわかる。したがって、例えば、予測装置1Aによる予測結果は、医師の判断等を補助することができる。
〈第3の実施形態〉
 図23に示すように、第3の実施形態に係る予測装置1Bは、対象のmiRNAの発現量を入力データとし、学習済みモデルM3を用いて対象がスーパーレスポンダー、マイルドレスポンダー又はノンレスポンダーのいずれであるかを予測する。すなわち、上述した第2の実施形態に係る予測装置1Aは、2クラス分類を実施するものであるのに対し、第3の実施形態に係る予測装置1Bは、同じ入力データを用いて3クラス分類を実施するものである。具体的には、第2の実施形態に係る予測装置1Bは、対象が、薬剤の効果があるか、又は、薬剤の効果がないかを分類するものであるが、第3の実施形態に係る予測装置1Aは、対象が、薬剤の効果が高く得られるか、薬剤の効果が低く得られるか、又は、薬剤の効果がないかを分類するものである。図23に示す例では、入力データは、miRNAの発現量とともに、対象の臨床情報を含む。
 ここで、スーパーレスポンダー、マイルドレスポンダー及びノンレスポンダーは、図24に示すように、対象のPFSの期間及び対象の癌治療の効果に応じて定められる分類(Response)である。具体的には、癌治療の効果は、CR(complete response:完全奏効)、PR(partial response:部分奏効)、PD(progressive disease:進行)、SD(stable disease:安定)の4つのカテゴリに分類される。図24に示す一例では、スーパーレスポンダー(Super Responder)は、「治療の効果がCR又はPRと分類され、かつ、PFSが第1の閾値T1以上」である対象である。また、マイルドレスポンダー(Mild Responder)は、「治療の効果がCR又はPRと分類され、かつ、PFSが第1の閾値T1未満」である対象、及び、「治療の効果がSDと分類され、かつ、PFSが第2の閾値T2以上」である対象である。さらに、ノンレスポンダー(Non Responder)は、「治療の効果がSD又はPDと分類され、かつ、PFSが第2の閾値T2未満」である対象である。なお、各閾値T1,T2は任意に設定することができる。また、T1=T2であってもよい。図24に示す例では、第1の閾値T1及び第2の閾値T2ともに、『20』の例である。
 したがって、予測装置1Bが利用する学習済みモデルM3は、図25に示すように、miRNAの発現量及び臨床情報を説明変数とし、スーパーレスポンダー、マイルドレスポンダー又はノンレスポンダーの何れであるかを目的変数とするデータセットの関係を、訓練装置2Bにより学習されたものである。上述した例と同様に、訓練装置2Bも、例えば、ランダムフォレスト、ニューラルネットワーク又は決定木等の統計学的手法を利用することができる。今回は、ランダムフォレストにより学習を行った学習済みモデルM3を例として説明する。また、今回の学習済みモデルM3は、5分割交差検証を用いて評価されたものである。
 ここで、対象の治療の効果がCR、PR、PD又はSDのいずれであるかについては、例えば、医師により判断される。PFS及び治療の効果(Response)に応じて、スーパーレスポンダー、マイルドレスポンダー及びノンレスポンダーの何れであるかが決定される。また入力データとして用いる臨床情報は、上述した例と同様に、例えば、対象の『性別』、『喫煙歴』、『遺伝子変異』、『年齢』、『PD-L1』、『病理組織学的な分類』等である。
《学習処理》
 訓練装置2Bの構成については、図3を用いて上述した訓練装置2と同様であるため、図示を用いた説明を省略する。また、予測装置1Bの構成についても図8を用いて上述した予測装置1と同様であるため、図示を用いた説明を省略する。ここで、学習に利用されたデータに応じて、学習済みモデルM3が異なるものとなる。一例として、訓練データセットが、特定のmiRNAの発現量及び特定の臨床情報を含む場合、学習済みモデルM3も、その特定のmiRNAの発現量及び特定の臨床情報を入力データとしてスーパーレスポンダー、マイルドレスポンダー又はノンレスポンダーの何れであるかを予測する。なお、予測装置1Bの記憶装置においては、予測装置1Bの記憶装置においては、発現量データに加え、臨床情報を記憶する。また、訓練装置2B及び予測装置1Bも、図10を用いて上述したハードウェア構成のコンピュータにより実現されてもよい。
《miRNAの選択》
・39種の例
 第3の実施形態に係る予測装置1Bで発現量を利用するmiRNAの種類の選択も第1実施形態で説明したのと同様の方法で選択することができる。図26は、閾値であるPFSを「20カ月」とし、p値のカットオフ値を「0.05」とした場合に選択された45種のmiRNA及び3種の臨床情報について得られたp値のグラフを示す。図26において、横軸は、miRNA又は臨床情報のいずれかの種別であり、縦軸は、各miRNA又は臨床情報に関するp値を示す。図26において、臨床情報には、*を付している。
 図26に示すように、選択された3種の臨床情報は、PD-L1、喫煙率及び遺伝子変異である。また、選択された45種のmiRNAは、hsa-miR-21-5p、hsa-miR-429、hsa-miR-3679-5p、hsa-miR-100-5p、hsa-miR-141-3p、hsa-miR-200c-3p、hsa-miR-200a-3p、hsa-miR-2110、hsa-miR-130a-3p、hsa-miR-378a-3p、hsa-miR-98-5p、hsa-miR-203a、hsa-miR-382-5p、hsa-miR-4532、hsa-miR-664a-5p、hsa-miR-145-3p、hsa-miR-483-3p、hsa-miR-181a-5p、hsa-miR-1301-3p、hsa-miR-320e、hsa-let-7f-5p、hsa-miR-424-5p、hsa-miR-320b、hsa-miR-101-3p、hsa-miR-378c、hsa-miR-22-3p、hsa-let-7a-5p、hsa-miR-1246、hsa-miR-320d、hsa-miR-32-5p、hsa-miR-628-3p、hsa-miR-190a-5p、hsa-miR-20b-5p、hsa-miR-224-5p、hsa-miR-139-5p、hsa-miR-584-5p、hsa-miR-369-5p、hsa-miR-148a-3p、hsa-miR-320c、hsa-miR-3605-3p、hsa-miR-95-3p、hsa-miR-205-5p、hsa-miR-145-5p、hsa-miR-625-3p、hsa-miR-150-5pである。図26の結果によれば、図5で選択された上位5種のmiRNAのうちの4種であるhsa-miR-429、hsa-miR-100-5p、hsa-miR-141-3p及びhsa-miR-200a-3pは、p値が低く、ここでも高い精度が得られるバイオマーカーとして選択される結果となっている。
 図27は、対象の45種のmiRNAの発現量と、臨床情報としてPD-L1、喫煙率及び遺伝子変異を用いて、対象がスーパーレスポンダー、マイルドレスポンダー又はノンレスポンダーの何れであるかを予測した結果である。
 図27に示す例では、スーパーレスポンダー、マイルドレスポンダー及びノンレスポンダーの精度(Precision)、再現率(Recall)、F値(F1-score)及びサンプル数(Support)を示す。ここで、サンプル数は、学習のデータセットとした対象者の数である。また、この場合において、学習済みモデルM3の正確度(Accuracy)は、0.62であった。
《バイオマーカー》
 したがって、例えば、上述した特定のmiRNAをバイオマーカーとすることができる。具体的には、第1の実施形態の場合と同様に、肺癌、大腸癌、胃癌、膵臓癌、肺癌、乳癌、腎臓癌、頭頸部癌、皮膚癌、悪性リンパ腫癌、食道癌等に対する薬剤の効果の予測に用いるバイオマーカーとすることができる。好ましくは、肺癌に対する薬剤の効果の予測に用いるバイオマーカーとすることができる。また、第1の実施形態の場合と同様に、薬剤は、免疫チェックポイント阻害剤とすることができる。
 特定のmiRNAは、hsa-miR-21-5p、hsa-miR-429、hsa-miR-3679-5p、hsa-miR-100-5p、hsa-miR-141-3p、hsa-miR-200c-3p、hsa-miR-200a-3p、hsa-miR-2110、hsa-miR-130a-3p、hsa-miR-378a-3p、hsa-miR-98-5p、hsa-miR-203a、hsa-miR-382-5p、hsa-miR-4532、hsa-miR-664a-5p、hsa-miR-145-3p、hsa-miR-483-3p、hsa-miR-181a-5p、hsa-miR-1301-3p、hsa-miR-320e、hsa-let-7f-5p、hsa-miR-424-5p、hsa-miR-320b、hsa-miR-101-3p、hsa-miR-378c、hsa-miR-22-3p、hsa-let-7a-5p、hsa-miR-1246、hsa-miR-320d、hsa-miR-32-5p、hsa-miR-628-3p、hsa-miR-190a-5p、hsa-miR-20b-5p、hsa-miR-224-5p、hsa-miR-139-5p、hsa-miR-584-5p、hsa-miR-369-5p、hsa-miR-148a-3p、hsa-miR-320c、hsa-miR-3605-3p、hsa-miR-95-3p、hsa-miR-205-5p、hsa-miR-145-5p、hsa-miR-625-3p、hsa-miR-150-5pの45種の群から選択された少なくとも1以上のmiRNAである。
 または、特定のmiRNAは、miRNAは、hsa-miR-21-5p、hsa-miR-429、hsa-miR-141-3p、hsa-miR-200c-3p、hsa-miR-200a-3p、hsa-miR-3679-5p、hsa-miR-100-5p、hsa-miR-130a-3p、hsa-miR-2110、hsa-miR-98-5p、hsa-miR-378a-3p、hsa-let-7f-5p、hsa-miR-424-5p、hsa-miR-181a-5p、hsa-miR-382-5p、hsa-miR-22-3p、hsa-miR-29c-5p、hsa-miR-483-3p、hsa-miR-320e、hsa-miR-4532、hsa-miR-32-5p、hsa-miR-369-5p、hsa-miR-1246、hsa-miR-101-3p、hsa-miR-378c、hsa-miR-320b、hsa-miR-320d、hsa-miR-1301-3p、hsa-miR-28-3p、hsa-miR-148a-3p、hsa-miR-363-3p、hsa-miR-205-5p、hsa-miR-125b-5p、hsa-miR-664a-5p、hsa-miR-20b-5p、hsa-miR-584-5p、hsa-miR-145-5p、hsa-miR-320c、hsa-miR-95-3pの39種の群から選択された少なくとも1以上のmiRNAである。
《予想処理》
 予測装置1Bを用いた予測処理について、図28に示すフローチャートを用いて説明する。学習済みモデルM3は、特定のmiRNAの発現量及び臨床情報を入力データとする。したがって、予測処理は、学習済みモデルM3を用いて対象の試料から計測されたmiRNAの発現量及び対象の臨床情報を入力データとし、対象がスーパーレスポンダー、マイルドレスポンダー又はノンレスポンダーのいずれであるかを予測する。
 まず、予測装置1Bにおける処理の前に、対象の血液を採取する(S31)。ここで、試料の採取のタイミングは、例えば、第1の実施形態で上述した場合と同様に、薬剤の投与開始の1ヶ月前から、投与開始までの間である。
 続いて、ステップS31で採取された血液からの特定のmiRNAの発現量を計測する(S32)。
 その後、予測装置1Bにおいて処理が実行される。図28のフローチャートにおいて、破線部分が、予測装置1Bにおいて実行される処理である。具体的には、予測装置1Bの取得部11は、ステップS32で計測された特定のmiRNAの発現量、及び、対象の臨床情報を取得する(S33)。対象の臨床情報は、例えば、入力装置を介してユーザによって入力され得る。
 続いて、予測装置1Bの予測部12は、ステップS33で取得したmiRNAの発現量及び臨床情報を用いて、学習済みモデルM2により対象がスーパーレスポンダー、マイルドレスポンダー又はノンレスポンダーのいずれであるかを予測する(S34)。
 また、予測装置1Bの出力部13は、ステップS34で予測した対象がスーパーレスポンダー、マイルドレスポンダー又はノンレスポンダーのいずれであるかの予測結果を出力する(S35)。
 このように、予測装置1Bは、特定の対象のmiRNAを用いて、その対象がスーパーレスポンダー、マイルドレスポンダー又はノンレスポンダーのいずれであるかを予測することができる。また、miRNAの発現量とともに、対象がスーパーレスポンダー、マイルドレスポンダー又はノンレスポンダーのいずれであるかの予測に、その対象の特定の臨床情報を用いて追加した場合、予測精度が向上することがわかる。したがって、例えば、予測装置1Bによる予測結果は、医師の判断等を補助することができる。
 本開示の全請求項に記載のプログラム更新システム、制御システム、移動体、プログラム更新方法及びプログラムは、ハードウェア資源、例えば、プロセッサ、メモリ、及びプログラムとの協働などによって、実現される。
以上、本開示の実施形態について詳述したが、本開示は上記した個々の実施形態に限定されるものではない。特許請求の範囲に規定された内容及びその均等物から導き出される本発明の概念的な思想と趣旨を逸脱しない範囲において、種々の追加、変更、置き換え、部分的削除等が可能である。例えば、前述した実施形態において、数値又は数式を説明に用いている場合、これらは例示的な目的で示されたものであり、本開示の範囲を限定するものではない。また、実施形態で示した各動作の順序も例示的なものであり、本開示の範囲を限定するものではない。
 本開示の予測方法、予測装置、コンピュータプログラム及びバイオマーカーは、例えば、肺癌を発症した患者に対するニボルマブの薬効の判定に有用である。
1      予測装置
11      取得部
12      予測部
13      発現量データ
13      出力部
2      訓練装置
21      データ取得部
22      学習部
23      訓練データセット

Claims (33)

  1.  対象の試料に由来する所定のmiRNAに関する値に基づいて、前記対象が罹患する疾患に対する薬剤の効果を予測する予測ステップ、
     を含む予測方法。
  2.  前記miRNAに関する値は、前記対象の血液に基づいて測定されるmiRNAの発現量である
     請求項1に記載の予測方法。
  3.  前記疾患は、肺癌である
    請求項1又は2に記載の予測方法。
  4.  前記薬剤は、免疫チェックポイント阻害剤である
    請求項1乃至3のいずれか1に記載の予測方法。
  5.  前記薬剤は、ニボルマブである
    請求項4に記載の予測方法。
  6.  前記miRNAに関する値として、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p及びhsa-miR-100-5pからなる群から選択された少なくとも1以上のmiRNAの発現量を取得する取得ステップを含む
     請求項4又は5に記載の予測方法。
  7.  前記miRNAに関する値として、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p,hsa-miR-100-5p,hsa-miR-3679-5p,hsa-miR-200c-3p,hsa-miR-181a-5p,hsa-miR-4516,hsa-miR-21-5p,hsa-miR-452-5p及びhsa-miR-320eからなる群からから選択された少なくとも1以上のmiRNAの発現量を取得する取得ステップを含む
     請求項4又は5に記載の予測方法。
  8.  前記miRNAに関する値として、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p,hsa-miR-100-5p,hsa-miR-3679-5p,hsa-miR-200c-3p,hsa-miR-181a-5p,hsa-miR-4516,hsa-miR-21-5p,hsa-miR-452-5p,hsa-miR-320e,hsa-miR-99a-5p,hsa-miR-98-5p,hsa-miR-4429,hsa-miR-378a-3p,hsa-miR-22-3p,hsa-miR-1301-3p及びhsa-miR-1246からなる群から選択された少なくとも1以上のmiRNAの発現量を取得する取得ステップを含む
     請求項4又は5に記載の予測方法。
  9.  前記miRNAに関する値として、hsa-miR-100-5pを含む少なくとも1以上のmiRNAの発現量を取得する取得ステップを含む
     請求項4乃至8のいずれか1に記載の予測方法。
  10.  前記miRNAに関する値として、hsa-miR-200a-3pを含む少なくとも1以上のmiRNAの発現量を取得する取得ステップを含む
     請求項4乃至8のいずれか1に記載の予測方法。
  11.  前記miRNAに関する値として、hsa-miR-100-5p及びhsa-miR-200a-3pを含む少なくとも2以上のmiRNAの発現量を取得する取得ステップを含む
     請求項4乃至10のいずれか1に記載の予測方法。
  12. 前記予測ステップは、前記miRNAに関する値に基づいて、PFS(Progression Free Survival)を予測することを含む
     請求項1乃至11のいずれか1に記載の予測方法。
  13. 前記予測ステップは、前記miRNAに関する値に基づいて、前記対象が少なくともレスポンダー又はノンレスポンダーのいずれであるかを予測することを含む
     請求項1乃至11のいずれか1に記載の予測方法。
  14.  前記miRNAに関する値に加え、前記対象の臨床情報に基づいて、前記対象が少なくともレスポンダー又はノンレスポンダーのいずれであるかを予測する
     請求項13に記載の予測方法。
  15.  前記予測ステップは、学習済みモデルを用いて前記薬剤の効果を特定する値を予測することを含む
    請求項1乃至14のいずれか1に記載の予測方法。
  16.  対象の試料に由来する所定のmiRNAに関する値を取得する取得部と、
     前記所定のmiRNAに関する値に基づいて、前記対象が罹患する疾患に対する薬剤の効果を予測する予測部と、
     を備える予測装置。
  17.  前記miRNAに関する値は、前記対象の血液に基づいて測定されるmiRNAの発現量である
     請求項16に記載の予測装置。
  18.  前記疾患は、肺癌である
     請求項16又は17に記載の予測装置。
  19.  前記薬剤は、免疫チェックポイント阻害剤である
     請求項16乃至18のいずれか1に記載の予測装置。
  20.  前記薬剤は、ニボルマブである
     請求項19に記載の予測装置。
  21.  前記取得部では、前記miRNAに関する値として、hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p及びhsa-miR-100-5pからなる群から選択された少なくとも1以上のmiRNAの発現量を取得する
     請求項19又は20に記載の予測装置。
  22.  前記予測部は、前記薬剤の効果を予測する値として、PFS(Progression Free Survival)を予測する
     請求項16乃至21のいずれか1に記載の予測装置。
  23.  前記予測部は、前記薬剤の効果を予測する値として、前記対象が少なくともレスポンダー又はノンレスポンダーのいずれであるかを予測する
     請求項16乃至21のいずれか1に記載の予測装置。
  24.  前記取得部は、前記所定のmiRNAに関する値とともに、前記対象の臨床情報を取得し、
     前記予測部は、前記miRNAに関する値に加え、前記対象の臨床情報に基づいて、前記対象が少なくともレスポンダー又はノンレスポンダーのいずれであるかを予測する
     請求項23に記載の予測装置。
  25.  前記予測部は、学習済みモデルを用いて前記薬剤の効果を特定する値を予測する
     請求項16乃至24のいずれか1に記載の予測装置。
  26. 1以上のmiRNAを含む、肺癌に対する免疫チェックポイント阻害剤の効果の予測に用いるためのバイオマーカー。
  27.  hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p及びhsa-miR-100-5pからなる群から選択された少なくとも1以上のmiRNAを含む、肺癌に対する免疫チェックポイント阻害剤の効果の予測に用いるためのバイオマーカー。
  28. hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p,hsa-miR-100-5p,hsa-miR-3679-5p,hsa-miR-200c-3p,hsa-miR-181a-5p,hsa-miR-4516,hsa-miR-21-5p,hsa-miR-452-5p及びhsa-miR-320eからなる群から選択された少なくとも1以上のmiRNAを含む、肺癌に対する免疫チェックポイント阻害剤の効果の予測に用いるためのバイオマーカー。
  29. hsa-miR-429,hsa-miR-200b-3p,hsa-miR-200a-3p,hsa-miR-141-3p,sa-miR-100-5p,hsa-miR-3679-5p,hsa-miR-200c-3p,hsa-miR-181a-5p,hsa-miR-4516,hsa-miR-21-5p,hsa-miR-452-5p,hsa-miR-320e,hsa-miR-99a-5p,hsa-miR-98-5p,hsa-miR-4429,hsa-miR-378a-3p,hsa-miR-22-3p,hsa-miR-1301-3p及びhsa-miR-1246からなる群から選択された少なくとも1以上のmiRNAを含む、肺癌に対する免疫チェックポイント阻害剤の効果の予測に用いるためのバイオマーカー。
  30.  hsa-miR-100-5pを含む少なくとも1以上のmiRNAを含む、肺癌に対する免疫チェックポイント阻害剤の効果の予測に用いるためのバイオマーカー。
  31.  hsa-miR-200a-3pを含む少なくとも1以上のmiRNAを含む、疾患に対する免疫チェックポイント阻害剤の効果の予測に用いるためのバイオマーカー。
  32.  hsa-miR-100-5p及びhsa-miR-200a-3pを含む少なくとも2以上のmiRNAを含む、肺癌に対する免疫チェックポイント阻害剤の効果の予測に用いるためのバイオマーカー。
  33.  前記薬剤は、ニボルマブである
    請求項26乃至32のいずれか1に記載のバイオマーカー。
PCT/JP2022/038566 2021-10-18 2022-10-17 予測方法及びバイオマーカー WO2023068220A1 (ja)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2021170231 2021-10-18
JP2021-170231 2021-10-18

Publications (1)

Publication Number Publication Date
WO2023068220A1 true WO2023068220A1 (ja) 2023-04-27

Family

ID=86059219

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2022/038566 WO2023068220A1 (ja) 2021-10-18 2022-10-17 予測方法及びバイオマーカー

Country Status (1)

Country Link
WO (1) WO2023068220A1 (ja)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110117565A1 (en) * 2008-12-15 2011-05-19 Micromedmark Biotech Co., Ltd. Serum or plasma microrna as biomarkers for non-small cell lung cancer
WO2011080315A1 (en) * 2009-12-30 2011-07-07 Febit Holding Gmbh Mirna fingerprint in the diagnosis of prostate cancer
WO2011113030A2 (en) * 2010-03-11 2011-09-15 H.Lee Moffitt Cancer Center & Research Institute Human cancer micro-rna expression profiles predictive of chemo-response
US20120035068A1 (en) * 2007-12-20 2012-02-09 Celgene Corporation Use of mirco-rna as a biomarker of immunomodulatory drug activity
JP2018521976A (ja) * 2015-05-29 2018-08-09 ジェネンテック, インコーポレイテッド 癌のための治療方法及び診断方法
JP2019536805A (ja) * 2016-12-05 2019-12-19 ジー1、セラピューティクス、インコーポレイテッドG1 Therapeutics, Inc. 化学療法計画中の免疫応答の保護

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120035068A1 (en) * 2007-12-20 2012-02-09 Celgene Corporation Use of mirco-rna as a biomarker of immunomodulatory drug activity
US20110117565A1 (en) * 2008-12-15 2011-05-19 Micromedmark Biotech Co., Ltd. Serum or plasma microrna as biomarkers for non-small cell lung cancer
WO2011080315A1 (en) * 2009-12-30 2011-07-07 Febit Holding Gmbh Mirna fingerprint in the diagnosis of prostate cancer
WO2011113030A2 (en) * 2010-03-11 2011-09-15 H.Lee Moffitt Cancer Center & Research Institute Human cancer micro-rna expression profiles predictive of chemo-response
JP2018521976A (ja) * 2015-05-29 2018-08-09 ジェネンテック, インコーポレイテッド 癌のための治療方法及び診断方法
JP2019536805A (ja) * 2016-12-05 2019-12-19 ジー1、セラピューティクス、インコーポレイテッドG1 Therapeutics, Inc. 化学療法計画中の免疫応答の保護

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
CHO YOON, MI-KYUNG SONG, TAE SUNG KIM & JAE-CHUN RYU: "Identification of miRNA alterations as biomarkers of particulate matter (PM) 2.5 in two model systems", TOXICOLOGY AND ENVIRONMENTAL HEALTH SCIENCES, SPECIAL ISSUE: ABSTRACTS OF THE 10TH INTERNATIONAL CONFERENCE ON ENVIRONMENTAL HEALTH SCIENCE, vol. 9, no. 4, 1 December 2017 (2017-12-01), pages S34, XP093059537 *
HE QINGFANG, FANG YIRONG, LU FENG, PAN JIN, WANG LIXIN, GONG WEIWEI, FEI FANGRONG, CUI JINGJIE, ZHONG JIEMING, HU RUYING, LIANG MI: "Analysis of differential expression profile of miRNA in peripheral blood of patients with lung cancer", JOURNAL OF CLINICAL LABORATORY, vol. 33, no. 9, 1 November 2019 (2019-11-01), US , XP093059518, ISSN: 0887-8013, DOI: 10.1002/jcla.23003 *
HSU YA-LING, HUNG JEN-YU, CHEN FENG-WEI, CHANG KUO-FENG, CHANG WEI-AN, TSAI YING-MING, CHONG INN-WEN, KUO PO-LIN: "Identification of novel gene expression signature in lung adenocarcinoma by using next-generation sequencing data and bioinformatics analysis", ONCOTARGET, vol. 8, no. 62, 1 January 2017 (2017-01-01), pages 104831 - 104854, XP093059520 *
JU LIXIA, HAN MINGQUAN, LI XUEFEI, ZHAO CHAO: "MicroRNA Signature of Lung Adenocarcinoma with EGFR Exon 19 Deletion", JOURNAL OF CANCER, vol. 8, no. 7, 1 January 2017 (2017-01-01), AU , pages 1311 - 1318, XP093059522, ISSN: 1837-9664, DOI: 10.7150/jca.17817 *

Similar Documents

Publication Publication Date Title
JP6415712B2 (ja) 標的遺伝子発現の数学的モデル化を用いるTGF−β細胞シグナル伝達経路活性の評価
EP3692537B1 (en) Assessment of notch cellular signaling pathway activity using mathematical modelling of target gene expression
EP3729439B1 (en) Assessment of mapk-ap 1 cellular signaling pathway activity using mathematical modelling of target gene expression
EP2438188B1 (en) miRNA FINGERPRINT IN THE DIAGNOSIS OF DISEASES
JP6931125B2 (ja) 標的遺伝子発現の数学的モデル化を使用する、jak−stat1/2細胞シグナル伝達経路活性の評価
US20220305075A1 (en) Methods and compositions for detecting and treating endometriosis
CN112951327B (zh) 药物敏感预测方法、电子设备及计算机可读存储介质
TW201636429A (zh) 用於檢測川崎病的方法及套組
WO2020210487A1 (en) Systems and methods for nutrigenomics and nutrigenetic analysis
US20230175058A1 (en) Methods and systems for abnormality detection in the patterns of nucleic acids
US20230348980A1 (en) Systems and methods of detecting a risk of alzheimer's disease using a circulating-free mrna profiling assay
JP2020072741A (ja) 個人の喫煙ステータスを予測するためのシステムおよび方法
JP2023524016A (ja) 結腸細胞増殖性障害を特定するためのrnaマーカと方法
WO2023068220A1 (ja) 予測方法及びバイオマーカー
CN114627970A (zh) 结肠腺癌的焦亡相关lncRNA预后模型及其构建方法和应用
WO2018210338A1 (en) Methods for detecting malignant colon conditions
CN111172285A (zh) 用于胰腺癌早期诊断和/或预后监测的miRNA组及其应用
Ivanov et al. 4565 Sex-Specific Differences in the Genomic Landscape of Pediatric and Adult Glioblastoma
WO2023230321A1 (en) Machine learning systems and methods for gene set enrichment analysis and scoring
CN117854740A (zh) 一种结直肠癌患者多西他赛敏感性预测方法及系统
JP2024074960A (ja) 臓器健康および疾患をモニタリングするための方法およびシステム
RU2021132994A (ru) Эпиэйджинг: новая экосистема для управления здоровым старением
Vijay et al. Statistical Techniques for Breast Cancer Classification: A Review
KR20190118240A (ko) 통증 판별용 마커

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22883524

Country of ref document: EP

Kind code of ref document: A1