Method for Determining Sensitivity to PARP inhibitor or genotoxic drugs based on non-functional transcripts

Abstract

The present invention relates to a method for determining sensitivity to a PARP inhibitor or a DNA damaging agent using non-functional transcripts, more specifically, a method for determining sensitivity to a PARP inhibitor or a DNA damaging agent by extracting nucleic acids from a biological sample, obtaining transcript-specific expression levels of non-functional transcripts of DNA repair-related genes, and then analyzing the transcript-specific utilization (TU) of the non-functional transcripts for each gene based on the obtained expression levels. Unlike conventional methods for determining sensitivity to a PARP inhibitor or a DNA damaging agent based on gene mutation information, the method for determining sensitivity to a PARP inhibitor or a DNA damaging agent according to the present invention utilizes information on transcripts transcribed from genes, and therefore not only can it determine sensitivity in real time, but it is also highly accurate and useful.

Description

FIELD OF THEINVENTION

[0001] The present invention relates to a method for determining sensitivity to a PARP inhibitor or a DNA damaging agent using a non-functional transcript, and more specifically, to a method for extracting nucleic acid from a biological sample, obtaining transcript-specific expression levels of non-functional transcripts of DNA repair-related genes, and then analyzing transcript-specific utilization (TU) of the non-functional transcripts for each gene based on the obtained expression levels, thereby determining sensitivity to a PARP inhibitor or a DNA damaging agent. [Background technology]

[0002] A biomarker is defined as "an indicator that can objectively measure and evaluate normal biological processes, disease progression, and drug sensitivity to treatment methods." Recently, with the development of genetic analysis technology, research on the association between specific gene mutations and specific diseases has increased, and biomarkers have been redefined as molecular and biological indicators that encompass all genes and genetic mutations, as well as the resulting differences in the expression of RNA, proteins, and metabolites.

[0003] In addition, for more effective treatment, companion diagnostic devices (CDx) that can determine the sensitivity of biomarkers are being developed to classify patient groups in order to maximize the therapeutic effects of drugs and minimize side effects.

[0004] Companion diagnosis is a diagnostic technique to predict a patient's sensitivity to a specific drug treatment in advance. To overcome the drawbacks of most existing anticancer drugs, which act on both cancer cells and normal cells and have significant side effects, targeted anticancer drugs have been developed that selectively attack specific target proteins.

[0005] However, targeted anticancer drugs are only effective in patients with the same type of cancer who have a specific target protein, so the therapeutic efficiency is very low unless patients who have the target molecule are selected.

[0006] In addition, because targeted anticancer drugs rely on cell growth and proliferation inhibition rather than cell death, there is a high possibility of resistance developing with long-term continuous drug administration. Therefore, it is necessary to analyze the target of the anticancer drug before administering the drug and select the patient group that will respond to the drug.

[0007] Roche, one of the multinational pharmaceutical companies, acquired Genentech, which developed the first breast cancer targeted anticancer drug, Herceptin, and its companion diagnostic kit, Herceptest, to launch companion diagnostic-based targeted anticancer drug therapy. Companion diagnostic kits include immunohistochemical tests such as DAKO and HercepTest to confirm the overexpression of specific proteins, FISH or CISH tests using DNA probes to confirm gene amplification of specific genes such as Ventana Medical Systems and INFORM HER-2 / NEU, and genomic techniques such as q-PCR such as Roche Diagnostics and cobas EGFR mutation test to check for the presence or absence of mutations in biomarker genes.

[0008] Meanwhile, Olaparib (AZD2281) is an anticancer drug that functions to suppress the abnormal proliferation of cancer cells, and is an inhibitor of the "PARP protein." PARP is a protein that repairs damaged DNA in cells, and plays a major role in helping cells complete DNA repair and continue to proliferate. Olaparib inhibits the proliferation of cancer cells by inhibiting the function of PARP. Olaparib is well known as a targeted treatment for ovarian and breast cancer, and is known to be an effective anticancer drug, especially for cancer patients who have genetic mutations in BRCA1 and BRCA2.

[0009] In other words, the effectiveness of anticancer drugs is greatly influenced by DNA repair ability, and since anticancer drugs have different resistance and toxicity among individuals, screening using appropriate therapeutic sensitivity markers can bring about a breakthrough in anticancer drug treatment. Recently, research on the therapeutic sensitivity of individual anticancer drugs by specific genes has been actively and continuously conducted. However, due to the complex effects of biological response-related factors to specific drugs, the diversity of therapeutic drugs and administration methods, and the difficulty of obtaining a large amount of samples, the reality is that no remarkable results have been seen yet.

[0010] Myriad Genetics has launched a product that diagnoses the presence or absence of germline BRCA1 and BRCA2 mutations as a companion diagnostic for PARP inhibitors (olaparib), talazoparib, and rucaparib. However, this product only determines the presence or absence of mutations regardless of the alleles of the BRCA1 / 2 genes, and the overall response rate (ORR, overall response rat) to PARP inhibitors is only 34%, which means that simple detection of BRCA1 / 2 germline mutations alone is not sufficient as a companion diagnostic for PARP inhibitors.

[0011] Similarly, Foundation Medicine's FoundationFocusCDxBRCA product is a companion diagnostic product that diagnoses the association between BRCA1 and BRCA2 mutations and the PARP inhibitor rucaparib, but the overall response rate (ORR) was only 53.8%, remaining a low response rate.

[0012] Thus, BRCA1 / 2 gene mutations alone can only partially predict HRD-related drug sensitivity. Therefore, attempts are being made to predict drug sensitivity by detecting the results of HRD, rather than the causes of HRD. HRD leaves various traces by failing to repair damage inflicted on the genome. In particular, genomic scar and signature 3 are well-known markers of HRD (Gulhan, DC et al., Nat. Genet. Vol. 51, pp. 912-919, 2019). Mutational signatures are used to associate genome patterns of single base mutations with specific background factors (Alexandrov, LB et al., Nature. Vol. 500, pp. 415-421, 2013). In this sense, signature 3 matches well with HRD. Genomic scars are detected by three chromosomal abnormalities, which respectively refer to telomeric allelic imbalances (NtAI), large-scale state transitions (LST) and loss of heterozygosity (LOH) (Abkevich, V. et al., Br. J. Cancer. Vol. 107, pp. 1776-1782, 2012; Popova, T. et al., Cancer Res.Vol. 72, pp. 5454-5462, 2012; Nicolai J. Birkbak. CANCER Discov. Vol. 2, 367, 2012).

[0013] Genomic HRD (gHRD) analysis is inevitably affected by the restoration of HR function. That is, the occurrence of another mutation that restores the function of BRCA1 / 2 is the cause of about half of platinum drug resistance cases occurring in ovarian cancer (Norquist, B. et al., J. Clin. Oncol. Vol. 29, pp. 3008-3015, 2011). In addition, a BRCA1 / 2-independent mechanism is known in relation to PARP inhibitor resistance (Chaudhuri, AR et al., Nature. Vol. 535, pp. 382-387, 2016). As mentioned above, even in situations where HR function is restored, the initially generated genomic scar is still detected, which is a problem. Furthermore, it is known that the technical issues in calculating the mutational signature further increase the false positive rate (Maura, F. et al., Nat. Commun. Vol. 10, 2019).

[0014] Because of these problems, a significant number of patients predicted to respond to DNA damaging drugs or PARP inhibitors do not actually do so (Watkins, JA et al., Breast Cancer Research. Vol. 16, pp. 1-11, 2014).

[0015] Therefore, new technologies are needed to understand the HRD status of patients in real time, such as methods to remove drug-resistant cases from gHRD prediction results using biomarkers that measure the functional status of HRD.

[0016] Therefore, the present inventors have made extensive efforts to develop a method for determining sensitivity to PARP inhibitors or DNA damaging drugs including platinum with high accuracy. As a result, by using transcript structure information of DNA repair-related genes to extract non-functional transcripts for each gene that are presumed to be translated into proteins that are not normally translated or that have lost their function, and then analyzing the expression levels of these transcripts, it has been found that sensitivity to PARP inhibitors or DNA damaging drugs can be determined with lower false positives and higher accuracy than HRD detection methods at the gene mutation level, thereby completing the present invention. Summary of the Invention [Problem to be solved by the invention]

[0017] An object of the present invention is to provide a method for determining sensitivity to PARP inhibitors and DNA damaging drugs using non-functional transcripts. Another object of the present invention is to provide an apparatus for determining sensitivity to PARP inhibitors and DNA damaging drugs using non-functional transcripts.

[0018] Another object of the present invention is to provide a computer readable storage medium comprising instructions adapted to be executed by a processor for determining sensitivity to PARP inhibitors and DNA damaging drugs in said method. Another object of the present invention is to provide a kit for analyzing a target RNA sequence using the above method.

[0019] To achieve the above object, the present invention provides a method for determining susceptibility to PARP inhibitors or DNA damaging drugs, comprising the steps of: a) extracting nucleic acids from a biological sample and obtaining the expression level of each transcript of a non-functional transcript of a DNA repair-related gene; b) calculating the transcript usage (TU) of the non-functional transcript of each gene based on the obtained expression level; and c) analyzing the calculated TU value and determining that a subject has susceptibility to PARP inhibitors or genotoxic drugs if the calculated TU value is equal to or greater than a reference value.

[0020] The present invention also provides an apparatus for determining susceptibility to a PARP inhibitor or a DNA damaging drug, which is utilized in the method for determining susceptibility to a PARP inhibitor or a DNA damaging drug, and a computer-readable recording medium containing instructions for carrying out the method.

[0021] The present invention also provides a targeted RNA sequence analysis (Targeted RNA-Seq) kit used in the method for determining sensitivity to a PARP inhibitor or a DNA damaging drug.

[0022] The present invention also provides a cancer treatment method comprising the step of administering a PARP inhibitor or a DNA damaging drug to a patient determined to be susceptible according to the method for determining susceptibility to a PARP inhibitor or a DNA damaging drug. [Brief description of the drawings]

[0023] FIG. 1 is an overall flow chart for determining susceptibility to PARP inhibitors or DNA damaging drugs of the present invention.

[0024] FIG. 2 is a schematic diagram of a method for calculating transcript usage (TU) according to the present invention.

[0025] FIG. 3 shows results showing that minor isoforms overexpressed in gHRD(+) samples lead to a statistically significant loss of protein function according to one embodiment of the present invention.

[0026] FIG. 4 shows results showing that genes with specific minor isoforms overexpressed in gHRD(+) samples according to one embodiment of the present invention are statistically significantly associated with DNA repair function.

[0027] FIG. 5 shows results showing that genes with minor isoforms overexpressed in gHRD(+) samples according to one embodiment of the present invention are statistically significantly associated with various DNA repair subfunctions.

[0028] 6 shows a comparison of the results of drug sensitivity prediction of breast cancer cell lines using gHRD in one embodiment of the present invention with the results of prediction using tHRD derived based on 104 transcripts of 36 breast cancer-related genes. Drug sensitivity between cell lines predicted as positive by gHRD and tHRD and cell lines determined as negative is shown as IC50 values.

[0029] Figure 7 shows a comparison of the precision and recall of prediction results of drug sensitivity of breast cancer cell lines using gHRD in one embodiment of the present invention and prediction results using tHRD derived based on 104 transcripts of 36 breast cancer-related genes. The blue dotted line shows the performance before correction with gHRD (signature 3), the solid line shows the performance after correction, and the red line shows the performance of tHRD.

[0030] FIG. 8 shows the results of calculating tHRD for 20 major transcripts out of 104 breast cancer-related transcripts discovered in one embodiment of the present invention, and comparing them with gHRD drug response prediction.

[0031] FIG. 9 shows the results of calculating tHRD for 10 major transcripts out of 104 breast cancer-related transcripts discovered in one embodiment of the present invention, and comparing them with gHRD drug response prediction.

[0032] In Figure 10, (A) and (C) are the results of predicting platinum responsiveness of ovarian cancer patients using gHRD, and (B) and (D) are the results of predicting platinum responsiveness of ovarian cancer patients using tHRD based on 89 transcripts of 25 ovarian cancer-related genes discovered in one embodiment of the present invention. (A) and (B) are the results of comparing the survival rates of patients classified as positive and negative by gHRD and tHRD, and (C) and (D) are the results of analyzing the precision and recall and receiver operating characteristics of the two methods, where the blue dotted line is gHRD (scar) and the blue solid line is gHRD (signature 3). Red indicates tHRD performance.

[0033] In Figure 11, (A) and (C) are the results of predicting platinum responsiveness of ovarian cancer patients using gHRD, and (B) and (D) are the results of predicting platinum responsiveness of ovarian cancer patients using tHRD based on the major 10 of 89 transcripts of 25 ovarian cancer-related genes discovered in one embodiment of the present invention. (A) and (B) are the results of comparing the survival rates of patients classified as positive and negative by gHRD and tHRD, and (C) and (D) are the results of analyzing the precision and recall and receiver operating characteristics of the two methods, where the blue dotted line indicates gHRD (scar), the blue solid line indicates gHRD (signature3), and the red line indicates tHRD performance.

[0034] Figure 12 compares the performance of a tHRD model trained on platinum responsiveness instead of gHRD, with the performance of predicting platinum responsiveness using gHRD. (A) and (B) are the results of comparing the survival rates after platinum treatment for patients classified as positive and negative by models trained on gHRD and tHRD, respectively, and (C) and (D) are the results of analyzing the precision and recall and receiver operating characteristics for the two methods. The blue dotted line indicates the performance of gHRD (scar), the blue solid line indicates the performance of gHRD (signature 3), and the red line indicates the performance of tHRD.

[0035] FIG. 13 shows the results of a survival analysis using the time to cancer metastasis after platinum drug treatment in which actual ovarian cancer patients were classified using the tHRD model constructed in one embodiment of the present invention, where (A) represents 12 months, (B) represents 24 months, and (C) represents the entire period.

[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Generally, the nomenclature used herein and the laboratory procedures described below are those well known and commonly employed in the art.

[0037] Terms such as first, second, A, B, etc. may be used to describe various components, but the components are not limited by the terms and are used merely to distinguish one component from another. For example, a first component may be named a second component, and similarly, a second component may be named a first component, without departing from the scope of the technology described below. The term "and / or" includes any combination of multiple related listed items or multiple related listed items.

[0038] In the terms used in this specification, the singular form should be understood to include the plural form unless the context clearly indicates otherwise, and terms such as "comprise" should be understood to mean the presence of a stated feature, number, step, operation, component, part, or combination thereof, but not to exclude the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

[0039] Before describing the drawings in detail, it should be made clear that the components in this specification are merely classified according to the main function of each component. That is, two or more components described below may be combined into one component, or one component may be divided into two or more components according to more specific functions. Furthermore, each of the components described below may additionally perform some or all of the functions of the other components in addition to its own main function, or some of the main functions of each component may be exclusively performed by the other components.

[0040] Additionally, in performing a method or method of operation, the steps of the method may be performed in an order different from that shown, unless the context clearly dictates a particular order, i.e., the steps may be performed in the same order as shown, substantially simultaneously, or in reverse order.

[0041] In the present invention, the utilization rate (TU value) of each non-functional transcript is calculated based on the expression level information of each transcript of non-functional transcripts of genes obtained from a sample, and the TU value is then analyzed to compare the obtained value with a standard value to determine sensitivity to PARP inhibitors or DNA damaging drugs. This was confirmed to enable the determination of sensitivity to PARP inhibitors or DNA damaging drugs with higher accuracy than methods that predict sensitivity based on traces appearing in genes.

[0042] That is, in one embodiment of the present invention, genomic HRD is calculated from breast cancer / ovarian cancer patient data, and the patients are classified into gHRD(+) and gHRD(-) groups. Major isoforms and multiple minor isoforms are distinguished based on a transcript structure database. Then, transcript usage (TU) values ​​for each major isoform are calculated. Major isoforms are overexpressed in gHRD(+), i.e., genes showing aberrant TU (aTU) are discovered. These are found to be mainly genes related to DNA repair. Thus, genes and their minor isoforms used to optimize drug sensitivity assessment for each cancer type are selected, and the TU values ​​of these minor isoforms are used as input data for a random forest model to train an artificial intelligence model that determines the presence or absence of HRD. A method has been developed to determine high sensitivity to PARP inhibitors or DNA damaging drugs when tHRD is positive (Figure 1).

[0043] Therefore, from one aspect, the present invention provides a method for producing a semiconductor device comprising: a) extracting nucleic acids from a biological sample and obtaining transcript-specific expression levels of non-functional transcripts of DNA repair-related genes; b) calculating the transcript usage (TU) of the non-functional transcripts of each gene based on the expression levels obtained; and c) This relates to a method for determining susceptibility to PARP inhibitors or DNA damaging drugs, which includes a step of determining that a subject has susceptibility to PARP (Poly ADP Ribose Polymerase) inhibitors or genotoxic drugs if the value obtained by analyzing the calculated TU value is equal to or greater than a reference value.

[0044] In the present invention, the nucleic acid may be characterized as being RNA, but is not limited thereto.

[0045] In the present invention, the step a) may be characterized as being carried out by a method including the steps of: ai) obtaining nucleic acid from blood, semen, vaginal cells, hair, saliva, urine, buccal cells, cancer tissue cells, FFPE samples and mixtures thereof; a-ii) removing proteins, fats, and other residues from the collected nucleic acid by salting-out method, column chromatography method, or beads method to obtain purified nucleic acid; a-iii) enriching the purified nucleic acids for DNA repair-related genes to prepare a library; a-iv) subjecting the produced library to a next-generation sequencer; and av) A step of obtaining nucleic acid sequence information (reads) using a next-generation gene sequencer.

[0046] In the present invention, the method for enriching the DNA repair-related genes may be any method known to ordinary skilled artisans, including, but not limited to, probe-based capture and primer-based amplification methods.

[0047] In the present invention, the library can be composed of cDNA, but is not limited thereto.

[0048] In the present invention, the next-generation sequencer can be used with any sequencing method known in the art. Sequencing of the nucleic acids separated by the selection method is typically performed using next-generation sequencing (NGS). Next-generation sequencing includes any sequencing method that determines the nucleotide sequence of an individual nucleic acid molecule or one of the proxies cloned to the individual nucleic acid molecule in a very similar manner (e.g., 10 5 (These molecules are sequenced simultaneously.) In one embodiment, the relative abundance of a nucleic acid species in a library can be estimated by measuring the relative occurrence of its cognate sequence in the data generated by the sequencing experiment. Next generation sequencing methods are known in the art and are described, for example, in Metzker, M. (2010) Nature Biotechnology Reviews 11:31-46, which is incorporated herein by reference.

[0049] In one embodiment, next generation sequencing is performed to determine the nucleotide sequence of individual nucleic acid molecules (e.g., the HeliScope Gene Sequencing system from Helicos BioSciences and the PacBio RS system from Pacific Biosciences). In other embodiments, sequencing, such as mass-parallel short-read sequencing, which generates more bases of sequence per sequencing unit than other sequencing methods that generate fewer but longer reads (e.g., the Solexa sequencer from Illumina Inc., San Diego, Calif.; 454 Life Sciences, Branford, Conn. and Ion Torrent), determines the nucleotide sequence of clonally extended proxies for individual nucleic acid molecules. Other methods or machines for next generation sequencing are provided by, but are not limited to, 454 Life Sciences (Branford, Connecticut), Applied Biosystems (Foster City, California, SOLiD sequencer), Helicos Biosciences Corporation (Cambridge, Massachusetts), and emulsion and microflow sequencing technology nano-infusion (e.g., GnuBio infusion).

[0050] Platforms for next generation sequencing include, but are not limited to, the Roche / 454 Genome Sequencer (GS) FLX system, the Illumina / Solexa Genome Analyzer (GA), the Life / APG Support Oligonucleotide Ligation Detection (SOLiD) system, the Polonator G.007 system, the HeliScope Gene Sequencing system from Helicos BioSciences, and the Pacific Biosciences PacBio RS system.

[0051] In the present invention, "sample", "tissue sample", "patient sample", "patient cell or tissue sample", "cancer tissue cell", "FFPE sample" or "specimen" refer to a collection of similar cells obtained from a subject or patient's tissue or circulating cells, respectively. The source of the tissue sample may be fresh, frozen and / or preserved organ, tissue sample, solid tissue from a biopsy or aspirate, blood or any blood component, bodily fluid, such as cerebrospinal fluid, amniotic fluid, peritoneal fluid or interstitial fluid, or cells from a point in the subject's pregnancy or development. The tissue sample may also contain compounds that are not naturally intermixed with the tissue in nature, such as preservatives, anticoagulants, buffers, fixatives, nutrients, antibiotics, etc. In one embodiment, the sample is prepared as a frozen sample or as a formaldehyde- or paraformaldehyde-fixed paraffin-embedded (FFPE) tissue preparation. For example, the sample may be embedded in a matrix, such as an FFPE block or a frozen sample.

[0052] In one embodiment, the sample is a tumor sample, e.g., comprises one or more precancerous or malignant cells. In certain embodiments, the sample, e.g., tumor sample, is obtained from a solid tumor, a soft tissue tumor, or a metastatic lesion. In another embodiment, the sample, e.g., tumor sample, comprises tissue or cells from a surgical resection surface. In another embodiment, the sample, e.g., tumor sample, comprises one or more circulating tumor cells (CTCs) (e.g., CTCs obtained from a blood sample).

[0053] In the present invention, the step of obtaining the expression level of non-functional transcripts for each gene can be performed using any NGS-based RNA-seq data analysis method known to ordinary technicians without any restrictions.

[0054] In the present invention, the DNA repair-related gene may include, without limitation, any gene known to be related to DNA repair, and preferably includes ABL1, ALKBH1, APEX1, APTX, ASF1A, ATM, ATP23, ATR, ATRX, ATXN3, BLM, BRCA1, BRCA2, BTG2, CCNO, CDKN2D, CEBPG, CIB1, CSNK1D, CSNK1E, DDB1, DDB2, ERCC1, ERCC2, ERCC3, ERCC4, ERCC5, ERCC6, ERCC8, EXO1, FANCA, and FAN. CC, FANCG, FEN1, GADD45A, GADD45G, GTF2H1, GTF2H4, HMGB1, HMGB1P10, HMGB2, HUS1, IGHMBP2, KAT5, LIG1, LIG3, LIG4, MLH1, MMS19, MNAT1, MPG, MRE 11, MSH2, MSH3, MSH5, MSH6, MUTYH, NBN, NHEJ1, NTHL1, OGG1, PARP1, PARP3, PMS1, PMS2, PMS2P1, PNKP, POLA1, POLD1, POLE, POLE2, POLG, POLH, POLI, P OLL, POLQ, PRKCG, RAD1, RAD17, RAD21, RAD23A, RAD23B, RAD50, RAD51, RAD51B, RAD51C, RAD52, RAD54B, RAD54L, RAD9A, RBBP8, RECQL, RECQL4, RECQL 5, REV1, RFC3, RPA1, RPAIN, RUVBL2, SETX, SMC1A, SMUG1, SOD1, SUMO1, TDG, TNP1, TP53, TP73, TREX2, UBE2A, UBE2B, UBE2N, UBE2V1, UBE2V2, UNG, UPF1 ,UVRAG,VCP,WRNIP1,XAB2,XPC,XRCC2,XRCC3,XRCC4,XRCC6,BABAM2,BRIP1,CDCA5,CHEK1,DCLRE1C,FANCB,FANCI,MGME1,MND1,MUS81,NEIL1,PAR P9, RAD51AP1, RFC4, SMARCB1, TICRR, TRIP13, UBE2T, USP47, ABRAXAS1, ASCC1, CHEK2, NSMCE4A, PARP2, RAD51AP1, RHNO1, RMI2, RPS3, TNKS1BP1, UBB,The gene may be, but is not limited to, 10 or more genes selected from the group consisting of UIMC1 and USP45.

[0055] In the present invention, when the PARP inhibitor or DNA damaging drug is applied to breast cancer, the DNA repair-related gene may be characterized as being 10 or more genes selected from the group consisting of ALKBH2, ATXN3, BABAM2, BRIP1, CDCA5, CHEK1, DCLRE1C, DDB2, ERCC1, EXO1, FANCB, FANCC, FANCI, FEN1, KAT5, MGME1, MND1, MSH5, MUS81, NEIL1, PARP3, PARP9, POLD1, RAD51, RAD51AP1, RAD54L, RFC4, RPAIN, SMARCB1, SMC1A, TICRR, TRIP13, UBE2T, UBE2V2 and USP47, but is not limited thereto.

[0056] In the present invention, when the PARP inhibitor or DNA damaging drug is applied to ovarian cancer, the DNA repair-related genes may be characterized as being 10 or more genes selected from the group consisting of ABRAXAS1, ASCC1, BLM, CHEK2, ERCC1, EXO1, GADD45A, MUTYH, NSMCE4A, PARP2, POLE2, RAD51AP1, RAD51B, RECQL4, RHNO1, RMI2, RPS3, SUMO1, TNKS1BP1, UBB, UBE2A, UIMC1, USP45, VCP and XPC, but are not limited thereto.

[0057] In the present invention, the non-functional transcript in step a) may be characterized as a minor isoform, and the minor isoform may include, but is not limited to, a transcript that is predicted to be translated into a protein that is not normally translated or has lost function, or a transcript that is not utilized for protein translation.

[0058] In the present invention, the non-functional transcripts in step a) may be characterized as being transcripts remaining after removing transcripts encoding proteins well conserved for each gene from a known database (e.g., the APRIS database) (e.g., transcripts corresponding to principal (1-5) in the APRIS database).

[0059] In the present invention, the non-functional transcripts may further include a step of selecting non-functional transcripts derived from genes with an average count per million (CPM) value of 1 or more in order to improve the accuracy of future analyses.

[0060] In the present invention, in the step b) of calculating the transcript usage (TU) of the non-functional transcript for each gene, the TU value may be a ratio-type unit in which the sum of the TPM values ​​of all transcripts occurring in one gene is used as the denominator and the TPM of each transcript is used as the numerator, and preferably, the TU value may be calculated according to the following formula 1:

number

[0061] In the present invention, after obtaining the expression level of the non-functional transcript for each gene, in order to improve the accuracy of subsequent analysis, a step of selecting transcripts from the selected genes having a TU value of 1-5% or more, preferably 3% or more.

[0062] In the present invention, the step of analyzing the calculated TU values ​​in step c) to obtain a value may be characterized by using a decision-making procedure in which the TU values ​​of non-functional transcripts overexpressed in samples known to be genomic Homologous Recombinant Deficiency (gHRD) positive or drug responsive are multiplied by a specific weighting value and summed up to derive a judgment value in a specific range, or a final judgment is made depending on whether the TU value of each non-functional transcript exceeds a specific reference value.

[0063] In the present invention, the specific weighting value can be used without limitation as long as it is a value for deriving a judgment value in a specific range, and can be characterized in that the specific weighting value has a larger value as the TU value of a non-functional transcript overexpressed in a sample known to be positive for genomic homologous recombination deficiency (gHRD) or to be drug responsive is larger.

[0064] In the present invention, the judgment value within the specific range can be used without limitation as long as it is a value that can judge whether sensitivity to a PARP inhibitor or a DNA damaging drug is positive or negative, and is preferably a value that can be normalized to a value between 0 and 1, but is not limited thereto.

[0065] In the present invention, the specific reference value can be determined based on, but is not limited to, the TU value of a non-functional transcript, preferably an overexpressed non-functional transcript, obtained from a sample known to be genomic Homologous Recombinant Deficiency (gHRD) positive or drug responsive.

[0066] For example, in the present invention, the step of analyzing the calculated TU values ​​in step c) to obtain a value may be characterized as being performed by a method including a step of multiplying the TU values ​​of non-functional transcripts corresponding to non-functional transcripts overexpressed in a sample known to be positive for genomic homologous recombination deficiency (gHRD) or to be drug responsive by a weighting value, summing the weighting values, and normalizing the summation to a range of 0 to 1.

[0067] In the present invention, the weighting value may be any value that is assigned so that the sum is between 0 and 1, and may be characterized in that the weighting value is larger as the TU value of a non-functional transcript that is overexpressed in a sample known to be positive for genomic homologous recombination deficiency (gHRD) or to be drug responsive is larger.

[0068] For example, in the present invention, the step of analyzing the calculated TU values ​​in step c) to obtain a value can be performed using a decision-making procedure in which the TU value of each obtained non-functional transcript is determined based on the TU values ​​of non-functional transcripts obtained from samples known to be genomic Homologous Recombinant Deficiency (gHRD) positive or drug responsive.

[0069] In the present invention, the step of analyzing the calculated TU value in step c) to obtain a value may be characterized by being performed using an artificial intelligence model.

[0070] In the present invention, the artificial intelligence model may be characterized by machine learning weights for the TU values ​​of non-functional transcripts of each gene derived from known genomic homologous recombination deficiency (gHRD) positive samples or drug responsiveness positive samples, and reference values ​​for judgment values ​​derived therefrom, or reference values ​​for each TU value for decision-making and a decision-making procedure structure, to construct a prediction model for the relationship between gHRD or drug responsiveness and the expression level of non-functional transcripts of each gene.

[0071] In other words, the artificial intelligence model of the present invention constructs a prediction model by machine learning a combination of weights that can be imposed on the TU values ​​of non-functional transcripts of each gene overexpressed in gHRD-positive or drug-responsive positive samples and a reference value for a judgment value consisting of the sum of the weighted TU values ​​so as to reflect various patient-specific transcript expression patterns; or, the decision-making process of comparing the TU values ​​of non-functional transcripts of each gene overexpressed in gHRD-positive or drug-responsive positive samples with the respective reference values ​​is structured so as to reflect various patient-specific transcript expression patterns, and a prediction model is constructed through machine learning.

[0072] For example, the artificial intelligence model of the present invention learns to multiply the TU values ​​of non-functional transcripts for each gene that are overexpressed in gHRD-positive or drug response-positive samples by weighted values ​​and then sums them to a value approaching 1, thereby constructing a prediction model.

[0073] In the present invention, the various patient-specific transcript expression patterns may refer to the types and patterns of transcripts overexpressed by the patient.

[0074] Therefore, when a specific combination of TU values ​​is entered as input data, the more similar the combination is to a gHRD-positive or drug-reactivity-positive sample, the more positive the decision-making or judgment value that is output. This is compared with a reference value, and if it is equal to or greater than the reference value, it is determined that the sample is sensitive to a PARP inhibitor or DNA-damaging drug.

[0075] In the present invention, the reference value can be used without limitation as long as it is a value that can determine the sensitivity of a sample to a PARP inhibitor or a DNA damaging drug, and can be preferably 0.5 to 1, more preferably 0.5 to 0.8, and most preferably 0.5, but is not limited to these.

[0076] In the present invention, the machine learning can be used without limitation as long as it is supervised learning, but is preferably characterized by being performed by any one or more methods selected from the group consisting of k-Nearest Neighbors, Linear Regression, Logistic Regression, Support Vector Machine (SVM), Decision Tree, Random Forest, and Neural Network, but is not limited thereto.

[0077] In the present invention, when the machine learning is used as a random forest, the lower decision tree analysis is trained in a direction to minimize the Gini impurity calculated by the following Equation 2:

number

[0078] In the present invention, the random forest creates multiple training data that allow overlap based on one training data, trains multiple sub-decision tree analyses based on the training data, and then calculates the average of these model predictions to derive the final predicted value.

[0079] In the present invention, when the machine learning is trained based on gHRD, it is trained to predict the HRD status using the TU values ​​of minor isoforms of genes related to DNA repair as input data, and in this case, the HRD status is classified based on gHRD.

[0080] The gHRD method is known to show positive results even when HR function has already been restored by additional mutations in BRCA1 / 2, and to have many false positives that show positive results regardless of HR function due to technical artifacts. It has been reported that the sensitivity of samples predicted by gHRD(+) to PARP inhibitors and platinum drugs is less than 50%. Therefore, it is possible to learn so that only samples in the gHRD(+) group that have a statistically significantly higher minor isoform TU transcript have functional HRD and are actually sensitive to PARP inhibitors and DNA damaging drugs. In addition, due to such low accuracy issues, as the gHRD standard value increases, false negatives that do not pass the standard value occur even though the sample actually has functional HRD and is drug sensitive, and the recall rate also decreases. Therefore, by discriminating only cases that actually show functional aTU aspects, both accuracy and recall rate can be improved, resulting in higher overall accuracy.

[0081] In the present invention, when the machine learning is trained based on drug responsiveness, the TU value of the minor isoform of the gene related to DNA repair is used as input data to train to predict responsiveness to DNA damaging drugs such as PARP inhibitors and platinum. In this case, the responsiveness to the drug is usually based on the presence or absence of cancer progression within 6 months after drug treatment. Thus, patients whose cancer progressed within 6 months were classified as resistant patients, and those whose cancer did not progress were classified as responsive patients.

[0082] In the present invention, samples determined to be HRD positive or drug responsive by the artificial intelligence model are defined as having "transcriptional homologous recombination deficiency (tHRD)."

[0083] In the present invention, the difference in TU values ​​between a group showing HRD at the genetic level (gHRD(+)) and a group not showing HRD (gHRD(-)) was statistically analyzed based on "genomic scar" and "signature 3." Relative overexpression of a specific minor isoform was defined as an aberrant TU (aTU), and overexpression in the gHRD(+) group was determined as "aTU in gHRD(+)," while overexpression in the gHRD(-) group was determined as "aTU in gHRD(-)."

[0084] Theoretically, the same gene can produce both isoforms corresponding to aTU in gHRD(+) and aTU in gHRD(-), but in practice, we found that aTU in gHRD(+) appears far more frequently in genes related to DNA repair.

[0085] The terms "genomic scar" and "signature 3" in the present invention refer to the pattern of chromosomal rearrangement or mutation that appears in cells in which the function of homologous recombination has been lost, and refer to the pattern of chromosomal rearrangement or mutation that occurs at the whole gene level, regardless of which part of the homologous recombination pathway has been lost, and can be confirmed by array-based comparative genomic hybridization (aCGH), single nucleotide polymorphism (SNP) genotyping, and next-generation sequencing (NGS).

[0086] In the present invention, the PARP inhibitor can be used without limitation as long as it is a substance that can inhibit the activity of a PARP protein. Preferably, the PARP inhibitor may be a natural compound, a synthetic compound, DNA, RNA, a peptide, an enzyme, a ligand, a cell extract, or a mammalian secretion that inhibits the activity of a PARP protein. More preferably, the PARP inhibitor is selected from the group consisting of AZD2281 (olaparib), ABT888 (veliparib), AG014699 (rucaparib), MK-4827 (niraparib), BMN-673 (talazoparib), BSI201 (iniparib), BGP15 (O-(3-piperidino- The compound may be characterized as being selected from the group consisting of, but not limited to, 2-hydroxy-1-propyl)nicotinicamidoxime, INO1001 (3-Aminobenzamide), ONO2231, nicotinamide, 3-aminobenzamide, 4-dihydro-5-[4-(1-piperidinyl)butoxy]-1(2H)-isoquinolone, benzamide, quinolone, isoquinolone, benzopyrone, cyclic benzamide, benzimidazole, indole, and phenanthridinone.

[0087] The target cancer disease of the PARP inhibitor in the present invention may be characterized as being selected from the group consisting of ACTH-producing tumors, acute lymphocytic or lymphoblastic leukemia, acute or chronic lymphoblastic leukemia, acute nonlymphocytic leukemia, bladder cancer, brain tumor, breast cancer, cervical cancer, chronic myeloid leukemia, lymphoma, endometriosis, esophageal cancer, bladder cancer, Ewing's sarcoma, tongue cancer, Hopkins lymphoma, Kaposi's sarcoma, kidney cancer, liver cancer, lung cancer, mesothelioma, multiple myeloma, neuroblastoma, non-Hopkin's lymphoma, osteosarcoma, ovarian cancer, breast cancer, prostate cancer, pancreatic cancer, colon cancer, penile cancer, retinoblastoma, skin cancer, gastric cancer, thyroid pressure, uterine cancer, testicular cancer, Wilms' tumor, and tropoblastoma.

[0088] In the present invention, the DNA damaging drug can be any drug that causes DNA transformation, such as by inducing crosslinking, double-strand breaks, or intercalation of intracellular DNA, and is preferably selected from the group consisting of bleomycin, cisplatin, carboplatin, oxaliplatin, nedaplatin, doxorubicin, etoposide, and SN38, but is not limited thereto.

[0089] In one embodiment, the method for determining sensitivity to a PARP inhibitor or a DNA damaging drug according to the present invention can include, but is not limited to, the following steps (FIG. 1). (1) Obtain RNA fragment (read) data from biological samples using massive parallel sequencing. (2) Aligning the RNA fragment data to the human reference genome (3) Filtering the sorted data according to quality (4) Selection of non-functional transcripts of DNA repair-related genes for use in the assessment (5) Obtaining the expression level of non-functional transcripts for each gene (6) Calculate the transcript usage (TU) of non-functional transcripts by gene. (7) Input the TU value into the artificial intelligence model for drug susceptibility determination to derive the output value. (8) Comparing the output value with the reference value to determine drug sensitivity

[0090] From another viewpoint, the present invention provides an apparatus for determining susceptibility to a PARP inhibitor or a DNA damaging drug, which is used in the method for determining susceptibility to a PARP inhibitor or a DNA damaging drug according to the present invention, the apparatus comprising: (1) an information acquisition unit that extracts nucleic acids from a biological sample and acquires the expression levels of each transcript of non-functional transcripts of DNA repair-related genes; (2) a calculation unit that calculates the transcript usage (TU) of non-functional transcripts for each gene based on the expression level obtained; and (3) A susceptibility assessment section that assesses susceptibility to PARP (Poly ADP Ribose Polymerase) inhibitors or DNA damaging drugs (genotoxic drugs) if the calculated TU value is analyzed and the resulting value is equal to or greater than the reference value; The present invention relates to a susceptibility determination device comprising:

[0091] In the present invention, the information acquisition unit may include, but is not limited to, a data receiving unit that receives RNA fragment (read) data obtained by a massively parallel sequence analysis method from an independently produced biological sample; a data alignment unit that aligns the received data to a reference gene; a filtering unit that performs filtering according to data quality; and an expression level acquisition unit that selects non-functional transcripts of DNA repair-related genes and acquires their expression levels.

[0092] From another aspect, the present invention provides a computer-readable recording medium used in the method for determining sensitivity to a PARP inhibitor or a DNA damaging drug according to the present invention, the medium comprising instructions configured to be executed by a processor for determining sensitivity to a PARP inhibitor or a DNA damaging drug, a) extracting nucleic acids from a biological sample and obtaining transcript-specific expression levels of non-functional transcripts of DNA repair-related genes; b) calculating the transcript usage (TU) of the non-functional transcripts for each gene based on the expression levels obtained; and c) A computer-readable recording medium including instructions configured to be executed by a processor, the instructions including a step of determining susceptibility to PARP (Poly ADP Ribose Polymerase) inhibitors or DNA damaging drugs when the calculated TU value is analyzed and the value obtained is equal to or greater than a reference value.

[0093] In another aspect, the method according to the present invention can be embodied using a computer. In one embodiment, the computer includes one or more processors coupled to a chipset. Also coupled to the chipset are memory, storage, a keyboard, a graphics adapter, a pointing device, and a network adapter. In one embodiment, the chipset's capabilities are enabled by a memory controller hub and an I / O controller hub. In another embodiment, the memory can be directly coupled to the processor instead of the chipset. The storage is any device capable of holding data, including a hard drive, a compact disk read-only memory (CD-ROM), a DVD, or other memory device. The memory is responsible for the data and instructions used by the processor. The pointing device may be a mouse, a track ball, or other type of pointing device, and is used in combination with a keyboard to send input data to the computer system. The graphics adapter displays images and other information on a display. The network adapter is connected to the computer system by a short-range or long-range communication network. The computer used in the present application is not, however, limited to the configurations described above, and may include none or additional configurations, and may be part of a Storage Area Network (SAN), and the computer of the present application may be configured to execute modules in a program for performing the method of the present application.

[0094] In the present application, a module may refer to a functional and structural combination of hardware for implementing the technical idea of ​​the present application and software for driving the hardware. For example, the module may refer to a logical unit of a certain code and a hardware resource for executing the certain code, and it is obvious to those skilled in the art that the module does not necessarily refer to physically connected code or to one type of hardware.

[0095] The method according to the present application can be embodied in hardware, firmware, or software, or a combination thereof. When embodied in software, the storage medium includes any medium that stores or transmits information in a form readable by a device such as a computer. For example, computer-readable media include Read Only Memory (ROM), Random Access Memory (RAM), magnetic disk storage media, optical storage media, flash memory devices, and other electrical, optical, or acoustic signal transmission media.

[0096] In this regard, the present application also relates to a computer readable medium including an execution module for causing a processor to execute operations including steps according to the present application described above.

[0097] From another perspective, the present invention relates to a targeted RNA-Seq kit for use in a method for determining sensitivity to a PARP inhibitor or a DNA damaging drug, the kit comprising a probe for capturing a transcript of a DNA repair-related gene; and a primer for amplifying the captured transcript. In the present invention, the transcript may be, but is not limited to, a non-functional transcript.

[0098] In the present invention, the kit may optionally include reagents necessary for performing a nucleic acid amplification reaction (e.g., polymerase chain reaction), such as a buffer, DNA polymerase, DNA polymerase cofactor, and deoxyribonucleotide-5-triphosphate (dNTP). Optionally, the kit of the present invention may include various oligonucleotide molecules, reverse transcriptase, various buffers and reagents, and an antibody that inhibits DNA polymerase activity. In addition, the optimal amount of reagents used in a particular reaction of the kit may be easily determined by one of ordinary skill in the art having learned the disclosures of this specification. Typically, the equipment of the present invention may be fabricated with separate packages or compartments containing the aforementioned components.

[0099] In one embodiment, the kit may comprise a compartmentalized carrier means for receiving a sample, a container containing reagents, a container containing probes or primers, and a container containing a probe for detecting an amplification product.

[0100] The carrier means is suitable for containing one or more containers, such as bottles, tubes, etc., each container containing an individual component used in the method of the present invention. With the present specification, a person skilled in the art can easily dispense the required formulations in the containers.

[0101] The present invention also relates to a method for treating cancer, comprising the step of administering a PARP inhibitor or a DNA damaging drug to a patient determined to be susceptible according to the method for determining susceptibility to a PARP inhibitor or a DNA damaging drug. BEST MODE FOR CARRYING OUT THEINVENTION

[0102] [Example] The present invention will be described in more detail below with reference to examples. It will be obvious to those skilled in the art that these examples are provided solely for the purpose of illustrating the present invention and that the scope of the present invention is not limited by these examples.

[0103] Example 1. Obtaining genetic and transcriptome data from breast and ovarian cancer patients We collected data for 645 breast cancer samples with available whole-exome sequencing (WES), genotyping data (Affymetrix SNP chip), and RNA sequencing (RNA-seq) data from the TCGA database and 315 ovarian cancer samples with available platinum treatment response data (https: / / portal.gdc.cancer.gov / ). For breast cancer cell lines, WES and RNA-seq data were obtained from the CCLE database (https: / / portals.broadinstitute.org / ccle).

[0104] Example 2. Calculation of genomic HRD (gHRD) The genomic scar of the samples collected in Example 1 was calculated using existing software (https: / / github.com / GerkeLab / TCGAhrd) based on Affymetrix SNP data, and the mutational signature 3 based on WES data was calculated using deconstructSigs in mSignatureDB (Po-Jung Huang. et al. mSignatureDB a database for deciphering mutational signatures in human cancers) in TCGA.

[0105] To distinguish between gHRD(+) and gHRD(-), we used the median value of genomic scar and mutational signature 3. To eliminate errors due to discrepancies between the indices (median value of genomic scar and mutational signature 3), we classified a gene as gHRD(+) if it exceeded the median of both indices, and as gHRD(-) if it was below the median.

[0106] Example 3. Calculation of Transcript Usage (TU) In TCGA and CCLE, the RNA-seq bam files were used as input to calculate the TPM (transcripts per million) of novel transcripts using a string-tied two-pass method. The process was based on the reference (Mihaela Peratea. et al. Nat Protoc. Vol. 11(9), pp.1650-1667, 2016.).

[0107] The reference was assembled using gencode v29 gtf (in the case of CCLE, gencode v19 gtf was used), and the gtfs of the individual samples were merged to produce merged.gtf, which was then used as the new reference annotation for the second transcript quantification step.

[0108] For de novo / novel transcripts that could not be annotated with existing forms, the gene from the most similarly annotated transcript was assigned to the form using gffcompare for gene annotation.

[0109] As a result, TPM values ​​were obtained for most of the transcripts expressed in each sample, and based on this, the TU for each transcript was calculated using the following formula 1 (Figure 2).

number

[0110] Example 4. Selection of minor isoforms For isoform classification, we used transcript annotations provided by the APPRIS database (Jose Manuel Rodriguez. et al. Nucleic Acids Res.Vol. 46(D1):D213-D217, 2018.), and the annotation version is 2019.02.v29.

[0111] To define minor isoforms (alternative transcripts), we used the criteria provided by APPRIS and removed all principal transcripts (1-5), which are transcripts that code for functionally and evolutionarily well-conserved proteins, as major forms. As a result, we selected a set of minor transcripts that were either alternative or not-reported.

[0112] To filter out genes with low expression levels, we performed counts per million (CPM) normalization for each gene through the HTSeq read count data for each gene, and then selected only genes with an average CPM > 1 across all samples. Only genes with transcript usage (TU) of 3% or more (TU > = 0.03) for this gene were selected for the final TU set.

[0113] Example 5. gHRD-related abnormal TU (aTU) excavation and functional analysis 5-1. Discovery of aTU As shown in Figure 2, aTUs were discovered through comparison between the gHRD(+) and gHRD(-) groups. For more precise classification, samples with genomic scar values ​​in the top 50% of gHRD(+) and samples with genomic scar values ​​in the bottom 50% of gHRD(-) were additionally selected.

[0114] The statistical significance of the difference in TU values ​​between the two groups was determined using the Mann-Whitney U test, corrected using the Benjanimi-Hochberg correction (BH correction).

[0115] A TU of a specific minor isoform that was significantly (FDR < 1%) higher in gHRD(+) was defined as aTU in gHRD(+), and conversely, a TU of a specific minor isoform that was significantly (FDR < 1%) higher in gHRD(-) was defined as aTU in gHRD(-).

[0116] 5-2. Functional analysis results We defined HR-related terms (1,033 genes) by merging the three gene ontology parent terms of DNA repair, replication, and recombination in breast cancer patient samples, and used this for GSEA (prerank) analysis. As a result, we confirmed that genes related to DNA repair were statistically clustered together, as shown in Figure 3.

[0117] Transcripts were sorted based on the U value of the Mann-Whitney U test (FDR<5%) as described in 5-1, and HR-related terms were overlapped in the matrix. A higher U value indicates that the minor isoform of the gene is more highly expressed in the gHRD(+) sample.

[0118] Method 5-1 was also applied to ovarian cancer to discover aTUs. In the case of ovarian cancer, the BH correction was not applied due to the lack of genes, and the most important aTUs were selected for each gene using the Mann-Whitney U test with P<0.05. After that, the aTUs were classified into aTU in gHRD(+) and aTU in gHRD(+) according to the direction of each gene, and it was confirmed that genes related to DNA repair were statistically densely clustered, as shown in Figure 3(B).

[0119] Example 6. Construction of a transcriptional HRD (tHRD) model 6-1. Preparing input data for model building Among the genes derived by the method of Example 5-1, genes related to DNA repair were selected, and a total of 36 genes (104 minor transcripts) were selected for breast cancer and 25 genes (89 minor transcripts) for ovarian cancer.

[0120] [Table 1] TIFF0007612016000005.tif254155TIFF0007612016000006.tif254155TIFF0007612016000007.tif167155

[0121] [Table 2] TIFF0007612016000009.tif251166TIFF0007612016000010.tif230166

[0122] For breast cancer, of the 115 transcripts, 104 that were also available in the CCLE dataset used for drug response testing were prepared as inputs to the final model, while for ovarian cancer, all transcripts were prepared as inputs.

[0123] 6-2. Building the model Random forest was used as the tHRD prediction model.

[0124] For both breast cancer and ovarian cancer, prediction models were constructed by dividing cancer into positive and negative categories based on gHRD, and in the case of ovarian cancer, a further prediction model was generated by dividing cancer into responsive and resistant categories based on platinum responsiveness.

[0125] The training set and test set were split at a ratio of 7:3, and hyperparameter tuning was performed 100 times using the random search CV package. The optimal hyperparameters were selected by measuring the average validation accuracy through 3-fold cross-validation.

[0126] We trained the model using the hyperparameters found by random search CV through the RandomForestClassifier in the sklearn.ensemble module and constructed a predictive model.

[0127] Example 7. Drug Response Analysis Breast cancer Using somatic mutation data of CCLE cell lines as input, the data were classified into 96 class types using the Rpackage deconstrucSigs. Then, using Sgenome.Hsapiens.UCSC.hg19 as a reference, the signature values ​​of each were calculated using NMF (Non-negative matrix factorization) for the previously defined Cosmic signatures.

[0128] In order to correct the incorrect assignment of signatures that may occur in the fitting method, we calculated signature 3 by fitting with the breast cancer-specific Cosmic signature set, referring to the reference (Francesco Maura. et al. Nat. cummunications, Vol. 10:2069, 2019.). Based on this, we defined cell lines that belong to the top 25% of signature 3 values ​​as gHRD(+) and the rest as gHRD(-).

[0129] For comparison, we also calculated signature 3, which was fitted to the entire cosmic signature rather than to a breast cancer-specific signature.

[0130] The difference in drug response between tHRD(+) and tHRD(-) predicted by the breast cancer random forest model was measured using 104 TU data for 36 breast cancer cell lines for which GDSC (Genomics of Drug Sensitivity in Cancer) data was available, and the IC50 values ​​for PARP inhibitors (Olaparib, Rucaparib, Veliparib, Talazoparib) and other DNA-damaging agents (Bleomycin, Cisplatin, Doxorubicin, Etoposide, SN-38) from the 2020.02 updated data provided by GDSC were used as drug sensitivity. The Mann-Whitney U test was used to compare IC50 values ​​between the two groups.

[0131] For precision recall, the signature 3 value and tHRD prediction value were varied, and the number of true positives, false positives, true negatives, and false negatives were measured based on the median IC50 value.

[0132] As a result, as shown in Figures 6 and 7, it was confirmed that the accuracy of the tHRD method was significantly superior to that of the gHRD method.

[0133] In addition, the drug response of breast cancer cell lines was predicted using the above method using only 20 major transcripts out of 104 transcripts (Table 3 below). As a result, it was confirmed that the performance was still superior to the gHRD method, as shown in Figure 8.

[0134] [Table 3]

[0135] In addition, the drug response of breast cancer cell lines was predicted using the above method using only 10 major transcripts out of 104 transcripts (Table 4 below). As a result, it was confirmed that the performance was still superior to the gHRD method, as shown in Figure 9.

[0136] [Table 4]

[0137] Ovarian cancer We used the same method as in 7-1, but to classify platinum response in TCGA ovarian cancer patients, we classified patients who received the first-line therapy presented in the reference (Victor M. Villalobos. et al. JCO Clinical Cancer Informatics, Vol.2, pp.1-16, 2018.) as platinum resistant if they progressed within 6 months, and as platinum responsive if they did not. Of the total 450 patients provided in the reference, 162 patients who had RNA-seq, genomic scar, and signature 3 all available were used for the ovarian cancer prediction model.

[0138] As a result, as shown in Figure 10, we confirmed that the accuracy of the tHRD-based method was superior to that of the gHRD method, and that the patient survival rate when classified as tHRD also showed a clear difference compared to when classified as gHRD.

[0139] In addition, the drug response of ovarian cancer patients was predicted using the above method using only 10 major transcripts out of 89 transcripts (Table 5 below). As a result, it was confirmed that the performance was equivalent to or even superior to the gHRD method, as shown in Figure 11.

[0140] [Table 5]

[0141] Furthermore, instead of training a prediction model based on gHRD as in Example 7-1, an additional prediction model was constructed based on platinum responsiveness according to the presence or absence of progression within 6 months as presented in the literature, and its performance was confirmed.

[0142] As a result, as shown in Figure 12, we confirmed that the predictive performance of the tHRD-based method was superior to that of the gHRD method, and that the patient survival rate when classified as tHRD also showed a significant difference, just like when classified as gHRD.

[0143] 7-3. Model performance verification in actual ovarian cancer patients To verify the optimal benchmark (0.4841) obtained by applying the model constructed in Example 6 (an artificial intelligence model trained to determine the presence or absence of genomic HRD by using the TU values ​​of DNA repair-related gene isoforms in TCGA-OV samples as input data for a random forest model) to RNA-seq data from ovarian cancer patients (n=27) treated with platinum chemotherapy, we predicted responsiveness and classified HRD classes using MGI DNB platform-based RNA sequencing data from independent ovarian cancer samples (n=20) treated with platinum chemotherapy.

[0144] As a result, when cancer metastasis within 6 months after drug treatment was used as the criterion, the model's prediction results showed high specificity (1.0), and when cancer metastasis within 12 months was used as the criterion, the model's prediction results showed specificity (0.78) and sensitivity (0.63) (Table 6).

[0145] [Table 6]

[0146] In addition, a survival analysis was performed over the time to progression of cancer (PFS), and it was confirmed that patients classified as HRD positive had a statistically significant lower risk of cancer metastasis within a period of less than two years when classified using the previously obtained optimal standard value (0.4841) (Figure 13).

[0147] Although the specific parts of the present invention have been described in detail above, it is clear to those skilled in the art that these specific techniques are merely preferred embodiments and do not limit the scope of the present invention. Therefore, the substantial scope of the present invention is defined by the appended claims and their equivalents. [Industrial Applicability]

[0148] The method for determining susceptibility to PARP inhibitors or DNA-damaging drugs according to the present invention differs from conventional methods for determining susceptibility to PARP inhibitors or DNA-damaging drugs based on gene mutation information in that it utilizes information on transcripts transcribed from genes, and therefore not only can it determine susceptibility in real time, but it is also highly accurate and useful.

Claims

1. a) extracting nucleic acids from a biological sample and obtaining transcript-specific expression levels of non-functional transcripts of DNA repair-related genes; b) calculating the transcript usage (TU) of the non-functional transcripts for each gene based on the expression levels obtained; and c) a method for determining in vitro susceptibility of breast cancer or ovarian cancer to a PARP inhibitor or a DNA damaging drug, comprising a step of determining that the cancer has susceptibility to a PARP (Poly ADP Ribose Polymerase) inhibitor or a DNA damaging drug when the calculated TU value is analyzed and the value is equal to or greater than a reference value, The non-functional transcript in step a) is a minor isoform, including a transcript that is predicted to be translated into a protein that is not normally translated or has lost its function, or a transcript that is not utilized for protein translation; The DNA repair-related genes include ABL1, ALKBH1, APEX1, APTX, ASF1A, ATM, ATP23, ATR, ATRX, ATXN3, BLM, BRCA1, BRCA2, BTG2, CCNO, CDKN2D, CEBPG, CIB1, CSNK1D, CSNK1E, DDB1, DDB2, ERCC1, ERCC2, ERCC3, ERCC4, ERCC5, ERCC6, ERCC8, EXO1, FANCA, FANCC, FANCG, FEN1, GADD45A, GADD45G, GTF2H1, GTF2H4, and HMGB. 1, HMGB1P10, HMGB2, HUS1, IGHMBP2, KAT5, LIG1, LIG3, LIG4, MLH1, MMS19, MNAT1, MPG, MRE11, MSH2, MSH3, MSH5, MSH6, MUTYH, NBN, NHEJ1, NTHL1, O GG1, PARP1, PARP3, PMS1, PMS2, PMS2P1, PNKP, POLA1, POLD1, POLE, POLE2, POLG, POLH, POLI, POLL, POLQ, PRKCG, RAD1, RAD17, RAD21, RAD23A, RAD23 B, RAD50, RAD51, RAD51B, RAD51C, RAD52, RAD54B, RAD54L, RAD9A, RBBP8, RECQL, RECQL4, RECQL5, REV1, RFC3, RPA1, RPAIN, RUVBL2, SETX, SMC1A, S MUG1, SOD1, SUMO1, TDG, TNP1, TP53, TP73, TREX2, UBE2A, UBE2B, UBE2N, UBE2V1, UBE2V2, UNG, UPF1, UVRAG, VCP, WRNIP1, XAB2, XPC, XRCC2, XRCC3, X ten or more genes selected from the group consisting of RCC4, XRCC6, BABAM2, BRIP1, CDCA5, CHEK1, DCLRE1C, FANCB, FANCI, MGME1, MND1, MUS81, NEIL1, PARP9, RAD51AP1, RFC4, SMARCB1, TICRR, TRIP13, UBE2T, USP47, ABRAXAS1, ASCC1, CHEK2, NSMCE4A, PARP2, RAD51AP1, RHNO1, RMI2, RPS3, TNKS1BP1, UBB, UIMC1, and USP45; The TU value is calculated as a ratio of the sum of the TPM values ​​of all transcripts occurring in a gene to the TPM of each transcript; Obtaining the value by analyzing the calculated TU in step c) is carried out using an artificial intelligence model selected from the group consisting of k-Nearest Neighbors, Linear Regression, Logistic Regression, Support Vector Machine (SVM), Decision Tree, Random Forest and Neural Network; The PARP inhibitors include AZD2281 (olaparib), ABT888 (veliparib), AG014699 (rucaparib), MK-4827 (niraparib), BMN-673 (talazoparib), BSI201 (iniparib), BGP15 (O-(3-piperidino-2-hydroxy-1-propyl)nicotinicamidoxime), INO1001 (3-aminobenzamide), ON selected from the group consisting of O2231, nicotinamide, 3-aminobenzamide, 3,4-dihydro-5-[4-(1-piperidinyl)butoxy]-1(2H)-isoquinolone, benzamide, quinolone, isoquinolone, benzopyrone, cyclic benzamide, benzimidazole, indole, and phenanthridinone; and The DNA damaging drug is selected from the group consisting of Bleomycin, Cisplatin, Carboplatin, Oxaliplatin, Nedaplatin, Doxorubicin, Etoposide and SN38; Susceptibility determination method.

2. The method for determining sensitivity to a PARP inhibitor or a DNA damaging drug according to claim 1, characterized in that the nucleic acid is RNA.

3. The method for determining sensitivity to a PARP inhibitor or a DNA damaging drug according to claim 1, wherein the step a) is carried out by a method comprising the steps of: ai) obtaining nucleic acid from blood, semen, vaginal cells, hair, saliva, urine, buccal cells, cancer tissue cells, FFPE samples and mixtures thereof; a-ii) removing proteins, fats, and other residues from the collected nucleic acid by salting-out method, column chromatography method, or beads method to obtain purified nucleic acid; a-iii) enriching the purified nucleic acids for DNA repair-related genes to prepare a library; a-iv) subjecting the produced library to a next-generation sequencer; and av) A step of obtaining nucleic acid sequence information (reads) using a next-generation gene sequencer.

4. The method for determining susceptibility to a PARP inhibitor or a DNA damaging drug according to claim 1, wherein the step of calculating the transcript usage (TU) of the non-functional transcript of each gene in the step b) is performed by the following formula 1: [0010] Here, TPM stands for transcripts per million.

5. The method for determining sensitivity to a PARP inhibitor or a DNA damaging drug according to claim 1, wherein the reference value is 0.5 to 1.

6. An apparatus for determining sensitivity to a PARP inhibitor or a DNA damaging drug, which is used in the in vitro method for determining sensitivity to a PARP inhibitor or a DNA damaging drug for breast cancer or ovarian cancer according to claim 1, (1) an information acquisition unit that extracts nucleic acids from a biological sample and acquires the expression levels of each transcript of a non-functional transcript of a DNA repair-related gene; (2) a calculation unit that calculates the transcript usage (TU) of non-functional transcripts for each gene based on the expression level obtained; and (3) A susceptibility determination section that determines susceptibility to PARP (Poly ADP Ribose Polymerase) inhibitors or DNA damaging drugs (genotoxic drugs) when the calculated TU value is analyzed and the value obtained is equal to or greater than the reference value; A susceptibility determination device comprising: The non-functional transcript is a minor isoform, including a transcript that is predicted to be translated into a protein that is not normally translated or has lost function, or a transcript that is not utilized for protein translation; The DNA repair-related genes include ABL1, ALKBH1, APEX1, APTX, ASF1A, ATM, ATP23, ATR, ATRX, ATXN3, BLM, BRCA1, BRCA2, BTG2, CCNO, CDKN2D, CEBPG, CIB1, CSNK1D, CSNK1E, DDB1, DDB2, ERCC1, ERCC2, ERCC3, ERCC4, ERCC5, ERCC6, ERCC8, EXO1, FANCA, FANCC, FANCG, FEN1, GADD45A, GADD45G, GTF2H1, GTF2H4, and HMGB. 1, HMGB1P10, HMGB2, HUS1, IGHMBP2, KAT5, LIG1, LIG3, LIG4, MLH1, MMS19, MNAT1, MPG, MRE11, MSH2, MSH3, MSH5, MSH6, MUTYH, NBN, NHEJ1, NTHL1, O GG1, PARP1, PARP3, PMS1, PMS2, PMS2P1, PNKP, POLA1, POLD1, POLE, POLE2, POLG, POLH, POLI, POLL, POLQ, PRKCG, RAD1, RAD17, RAD21, RAD23A, RAD23 B, RAD50, RAD51, RAD51B, RAD51C, RAD52, RAD54B, RAD54L, RAD9A, RBBP8, RECQL, RECQL4, RECQL5, REV1, RFC3, RPA1, RPAIN, RUVBL2, SETX, SMC1A, S MUG1, SOD1, SUMO1, TDG, TNP1, TP53, TP73, TREX2, UBE2A, UBE2B, UBE2N, UBE2V1, UBE2V2, UNG, UPF1, UVRAG, VCP, WRNIP1, XAB2, XPC, XRCC2, XRCC3, X ten or more genes selected from the group consisting of RCC4, XRCC6, BABAM2, BRIP1, CDCA5, CHEK1, DCLRE1C, FANCB, FANCI, MGME1, MND1, MUS81, NEIL1, PARP9, RAD51AP1, RFC4, SMARCB1, TICRR, TRIP13, UBE2T, USP47, ABRAXAS1, ASCC1, CHEK2, NSMCE4A, PARP2, RAD51AP1, RHNO1, RMI2, RPS3, TNKS1BP1, UBB, UIMC1, and USP45; The TU value is calculated as a ratio of the sum of the TPM values ​​of all transcripts occurring in a gene to the TPM of each transcript; Obtaining the value by analyzing the calculated TU is performed using an artificial intelligence model selected from the group consisting of k-Nearest Neighbors, Linear Regression, Logistic Regression, Support Vector Machine (SVM), Decision Tree, Random Forest and Neural Network; The PARP inhibitors include AZD2281 (olaparib), ABT888 (veliparib), AG014699 (rucaparib), MK-4827 (niraparib), BMN-673 (talazoparib), BSI201 (iniparib), BGP15 (O-(3-piperidino-2-hydroxy-1-propyl)nicotinicamidoxime), INO1001 (3-aminobenzamide), ON selected from the group consisting of O2231, nicotinamide, 3-aminobenzamide, 3,4-dihydro-5-[4-(1-piperidinyl)butoxy]-1(2H)-isoquinolone, benzamide, quinolone, isoquinolone, benzopyrone, cyclic benzamide, benzimidazole, indole, and phenanthridinone; and The DNA damaging drug is selected from the group consisting of Bleomycin, Cisplatin, Carboplatin, Oxaliplatin, Nedaplatin, Doxorubicin, Etoposide and SN38; Susceptibility determination device.

7. 2. A computer-readable recording medium for use in the in vitro method for determining sensitivity to a PARP inhibitor or a DNA damaging drug for breast cancer or ovarian cancer according to claim 1, the medium comprising instructions configured to be executed by a processor for determining sensitivity to a PARP inhibitor or a DNA damaging drug, a) extracting nucleic acids from a biological sample and obtaining transcript-specific expression levels of non-functional transcripts of DNA repair-related genes; b) calculating the transcript usage (TU) of the non-functional transcripts for each gene based on the expression levels obtained; and c) A computer-readable recording medium including instructions configured to be executed by a processor, the computer-readable recording medium including a step of determining susceptibility to PARP (Poly ADP Ribose Polymerase) inhibitors or DNA damaging drugs when the calculated TU value is analyzed and the value is equal to or greater than the reference value, The non-functional transcript in step a) is a minor isoform, including a transcript that is predicted to be translated into a protein that is not normally translated or has lost its function, or a transcript that is not utilized for protein translation; The DNA repair-related genes include ABL1, ALKBH1, APEX1, APTX, ASF1A, ATM, ATP23, ATR, ATRX, ATXN3, BLM, BRCA1, BRCA2, BTG2, CCNO, CDKN2D, CEBPG, CIB1, CSNK1D, CSNK1E, DDB1, DDB2, ERCC1, ERCC2, ERCC3, ERCC4, ERCC5, ERCC6, ERCC8, EXO1, FANCA, FANCC, FANCG, FEN1, GADD45A, GADD45G, GTF2H1, GTF2H4, and HMGB. 1, HMGB1P10, HMGB2, HUS1, IGHMBP2, KAT5, LIG1, LIG3, LIG4, MLH1, MMS19, MNAT1, MPG, MRE11, MSH2, MSH3, MSH5, MSH6, MUTYH, NBN, NHEJ1, NTHL1, O GG1, PARP1, PARP3, PMS1, PMS2, PMS2P1, PNKP, POLA1, POLD1, POLE, POLE2, POLG, POLH, POLI, POLL, POLQ, PRKCG, RAD1, RAD17, RAD21, RAD23A, RAD23 B, RAD50, RAD51, RAD51B, RAD51C, RAD52, RAD54B, RAD54L, RAD9A, RBBP8, RECQL, RECQL4, RECQL5, REV1, RFC3, RPA1, RPAIN, RUVBL2, SETX, SMC1A, S MUG1, SOD1, SUMO1, TDG, TNP1, TP53, TP73, TREX2, UBE2A, UBE2B, UBE2N, UBE2V1, UBE2V2, UNG, UPF1, UVRAG, VCP, WRNIP1, XAB2, XPC, XRCC2, XRCC3, X ten or more genes selected from the group consisting of RCC4, XRCC6, BABAM2, BRIP1, CDCA5, CHEK1, DCLRE1C, FANCB, FANCI, MGME1, MND1, MUS81, NEIL1, PARP9, RAD51AP1, RFC4, SMARCB1, TICRR, TRIP13, UBE2T, USP47, ABRAXAS1, ASCC1, CHEK2, NSMCE4A, PARP2, RAD51AP1, RHNO1, RMI2, RPS3, TNKS1BP1, UBB, UIMC1, and USP45; The TU value is calculated as a ratio of the sum of the TPM values ​​of all transcripts occurring in a gene to the TPM of each transcript; Obtaining the value by analyzing the calculated TU in step c) is carried out using an artificial intelligence model selected from the group consisting of k-Nearest Neighbors, Linear Regression, Logistic Regression, Support Vector Machine (SVM), Decision Tree, Random Forest and Neural Network; The PARP inhibitors include AZD2281 (olaparib), ABT888 (veliparib), AG014699 (rucaparib), MK-4827 (niraparib), BMN-673 (talazoparib), BSI201 (iniparib), BGP15 (O-(3-piperidino-2-hydroxy-1-propyl)nicotinicamidoxime), INO1001 (3-aminobenzamide), ON selected from the group consisting of O2231, nicotinamide, 3-aminobenzamide, 3,4-dihydro-5-[4-(1-piperidinyl)butoxy]-1(2H)-isoquinolone, benzamide, quinolone, isoquinolone, benzopyrone, cyclic benzamide, benzimidazole, indole, and phenanthridinone; and The DNA damaging drug is selected from the group consisting of Bleomycin, Cisplatin, Carboplatin, Oxaliplatin, Nedaplatin, Doxorubicin, Etoposide and SN38; Recording medium.

8. A targeted RNA sequence analysis (Targeted RNA-Seq) kit for use in the in vitro sensitivity determination method for breast cancer or ovarian cancer to a PARP inhibitor or a DNA damaging drug according to claim 1, comprising: A targeted RNA sequence analysis kit comprising: a probe for capturing a transcript of a DNA repair-related gene; and a primer for amplifying the captured transcript, The DNA repair-related genes include ABL1, ALKBH1, APEX1, APTX, ASF1A, ATM, ATP23, ATR, ATRX, ATXN3, BLM, BRCA1, BRCA2, BTG2, CCNO, CDKN2D, CEBPG, CIB1, CSNK1D, CSNK1E, DDB1, DDB2, ERCC1, ERCC2, ERCC3, ERCC4, ERCC5, ERCC6, ERCC8, EXO1, FANCA, FANCC, FANCG, FEN1, GADD45A, GADD45G, GTF2H1, GTF2H4, and HMGB. 1, HMGB1P10, HMGB2, HUS1, IGHMBP2, KAT5, LIG1, LIG3, LIG4, MLH1, MMS19, MNAT1, MPG, MRE11, MSH2, MSH3, MSH5, MSH6, MUTYH, NBN, NHEJ1, NTHL1, O GG1, PARP1, PARP3, PMS1, PMS2, PMS2P1, PNKP, POLA1, POLD1, POLE, POLE2, POLG, POLH, POLI, POLL, POLQ, PRKCG, RAD1, RAD17, RAD21, RAD23A, RAD23 B, RAD50, RAD51, RAD51B, RAD51C, RAD52, RAD54B, RAD54L, RAD9A, RBBP8, RECQL, RECQL4, RECQL5, REV1, RFC3, RPA1, RPAIN, RUVBL2, SETX, SMC1A, S MUG1, SOD1, SUMO1, TDG, TNP1, TP53, TP73, TREX2, UBE2A, UBE2B, UBE2N, UBE2V1, UBE2V2, UNG, UPF1, UVRAG, VCP, WRNIP1, XAB2, XPC, XRCC2, XRCC3, X ten or more genes selected from the group consisting of RCC4, XRCC6, BABAM2, BRIP1, CDCA5, CHEK1, DCLRE1C, FANCB, FANCI, MGME1, MND1, MUS81, NEIL1, PARP9, RAD51AP1, RFC4, SMARCB1, TICRR, TRIP13, UBE2T, USP47, ABRAXAS1, ASCC1, CHEK2, NSMCE4A, PARP2, RAD51AP1, RHNO1, RMI2, RPS3, TNKS1BP1, UBB, UIMC1, and USP45; kit.

Images