Method for predicting likelihood of response of human subject having tumor

Abstract

Disclosed herein a method for predicting the likelihood of a response of a human subject having a tumor to a therapy comprising an FGFR inhibitor.

Description

METHOD FOR PREDICTING LIKELIHOOD OF RESPONSE OF HUMAN SUBJECT HAVING TUMOR

[0001] The invention provides gene expression information useful for predicting whether cancer patients are likely to have a treatment response to a therapy comprising an FGFR inhibitor.

[0002] Tasurgratinib is known as an inhibitor against fibroblast growth factor receptors (FGFR) 1, 2 and 3, and has been reported to have a cell growth inhibitory effect against stomach cancer, lung cancer, bladder cancer and endometrial cancer (PTL 1). The compound has also been reported to have a high therapeutic effect for bile duct cancer (PTL 2), breast cancer (PTL 3) and hepatocellular carcinoma (PTL 4). Known pharmacologically acceptable salts of the compound include succinic acid salts and maleic acid salts (PTL 5). Tasurgratinib 1.5 succinate has been approved as "TASFYGO"(Registered Trademark) in Japan for the treatment of patients with unresectable biliary tract cancer with FGFR2 gene fusions or rearrangements that progressed after cancer chemotherapy.

[0003] Unfortunately, most anti-tumor treatments are associated with undesirable side effects, such as profound nausea, vomiting, or severe fatigue. Also, while anti-tumor treatments have been successful, they do not produce significant clinical responses in all patients who receive them, resulting in undesirable side effects, delays, and costs associated with ineffective treatment. Therefore, biomarkers that can be used to predict the response of a subject to an antitumor agent, prior to administration thereof are greatly needed.

[0004] US 2014 / 0235614 A1US 2018 / 0015079 A1US 2018 / 0303817 A1US 2020 / 0375970 A1US 2017 / 0217935 A1

[0005] An object of the present disclosure is to provide gene expression information useful for predicting whether cancer patients are likely to have a treatment response to a therapy comprising an FGFR inhibitor.

[0006] The present disclosure provides the following inventions: (1) A method for predicting the likelihood of a response of a human subject having a tumor to a therapy comprising an FGFR inhibitor, comprising: (a) measuring a level of an RNA transcript of selected genes in a biological sample obtained from the human subject, wherein the selected genes comprise at least one gene selected from the group consisting of ADAM28, AQP3, ATP2C2, BNIP3L, CCDC60, CYP4X1, DUSP4, DUSP6, ETV4, ETV5, FGFR3, IL20RA, LAD1, LRIG1, MUC2, MUC6, NMUR2, PAK5, PDE10A, RHEX, S100A4, SPRED1, SPRY2, VSIG2 and ZG16B; (b) normalizing the level of the RNA transcript of each of the genes to obtain a normalized gene expression level for each gene; (c) calculating the signature score based on the mRNA expression level of the selected genes; and (d) predicting the likelihood of the response based on the signature score. (2) The method of (1), wherein the signature score is calculated based on the following equation: FGF Signature-25 score = (log2(ADAM28)+log2((AQP3)+log2(ATP2C2)+log2(BNIP3L)+log2(CCDC60)+log2(CYP4X1)+log2(DUSP4)+log2(DUSP6)+log2(ETV4)+log2(ETV5)+log2(FGFR3)+log2(IL20RA)+log2(LAD1)+log2(LRIG1)+log2(MUC2)+log2(MUC6)+log2(NMUR2)+log2(PAK5)+log2(PDE10A)+log2(RHEX)+log2(S100A4)+log2(SPRED1)+log2(SPRY2)+log2(VSIG2)+log2(ZG16B)) / 25. (3) The method of (1), wherein the selected genes are DUSP4, DUSP6, ETV4, FGFR3, IL20RA, PAK5 and SPRY2, and the signature score is calculated based on the following equation: FGF Signature-7 score = (log2(DUSP4)+log2(DUSP6)+log2(ETV4)+log2(FGFR3)+log2(IL20RA)+log2(PAK5)+log2(SPRY2)) / 7. (4) A method for predicting the likelihood of a response of a human subject having a tumor to a therapy comprising an FGFR inhibitor, comprising: (a) measuring a level of an RNA transcript of selected genes in a biological sample obtained from the human subject, wherein the selected genes are DUSP4, DUSP6, ETV4, FGFR3, IL20RA, PAK5 and SPRY2; (b) normalizing the level of the RNA transcript of each of the genes to obtain a normalized gene expression level for each gene; (c) calculating the signature score based on the following equation: FGF Signature-7 score = (log2(DUSP4)+log2(DUSP6)+log2(ETV4)+log2(FGFR3)+log2(IL20RA)+log2(PAK5)+log2(SPRY2)) / 7; and (d) predicting the likelihood of the response based on the signature score. (5) A method for predicting the likelihood of a response of a human subject having a tumor to a therapy comprising an FGFR inhibitor, comprising: (a) measuring a level of an RNA transcript of selected genes in a biological sample obtained from the human subject, wherein the selected genes are CCND1, CHAD, DUSP4, DUSP6, EFNA1, EPHA2, ETV4, FGF10, GADD45G, IL20RA, KAT2B, MLLT3, PPP2CB, PRKAA2, PTPRR, SMAD4, TNFRSF10B, TNFRSF10D, TNFSF10 and XPA; (b) normalizing the level of the RNA transcript of each of the genes to obtain a normalized gene expression level for each gene; (c) calculating the signature score based on the following equation: FGF Signature-20 score = (log2(CCND1)+log2(CHAD)+log2(DUSP4)+log2(DUSP6)+log2(EFNA1)+log2(EPHA2)+log2(ETV4)+log2(FGF10)+log2(GADD45G)+log2(IL20RA)+log2(KAT2B)+log2(MLLT3)+log2(PPP2CB)+log2(PRKAA2)+log2(PTPRR)+log2(SMAD4)+log2(TNFRSF10B)+log2(TNFRSF10D)+log2(TNFSF10)+log2(XPA)) / 20; and (d) predicting the likelihood of the response based on the signature score. (6) A method for predicting the likelihood of a response of a human subject having a tumor to a therapy comprising an FGFR inhibitor, comprising: (a) measuring a level of an RNA transcript of selected genes in a biological sample obtained from the human subject, wherein the selected genes are BMPR1B, CCND1, DUSP4, ETV4, EFNA1, FGF10, GADD45G, HSPA1A, IRAK3, MAP2K4, MUTYH, NTHL1, PLCE1, SMAD3, TNFRSF10C and TNFRSF10D; (b) normalizing the level of the RNA transcript of each of the genes to obtain a normalized gene expression level for each gene; (c) calculating the signature score based on the following equation: FGF Signature-16 score = (log2(BMPR1B)+log2(CCND1)+log2(DUSP4)+log2(EFNA1)+log2(ETV4)+log2(FGF10)+log2(GADD45G)+log2(HSPA1A)+log2(IRAK3)+log2(MAP2K4)+log2(MUTYH)+log2(NTHL1)+log2(PLCE1)+log2(SMAD3)+log2(TNFRSF10C)+log2(TNFRSF10D)) / 16; and (d) predicting the likelihood of the response based on the signature score. (7) The method of any one of (1) to (6), wherein the biological sample is selected from the group consisting of a blood sample, a serum sample, a plasma sample, and a tumor sample, preferably the biological sample is a tumor sample. (8) The method of any one of (1) to (7), wherein the FGFR inhibitor is 5-((2-(4-(1-(2-hydroxyethyl)piperidin-4-yl)benzamide)pyridin-4-yl)oxy)-6-(2-methoxyethoxy)-N-methyl-1H-indole-1-carboxamide represented by Formula (I): or a pharmacologically acceptable salt thereof. (9) The method of (8), wherein the pharmacologically acceptable salt is 1.5 succinate. (10) The method of any one of (1) to (9), wherein the tumor is breast cancer. (11) The method of (10), wherein the breast cancer is metastatic breast cancer, locally advanced breast cancer, recurrent breast cancer, or unresectable breast cancer. (12) The method of (10) or (11), wherein the breast cancer is hormone receptor-positive. (13) The method of (10), wherein the breast cancer is hormone receptor-positive and HER2-negative metastatic breast cancer.

[0007] Another aspect of the disclosure is a method for treating a tumor in a human subject in need thereof. In particular the present disclosure also provides the following inventions: (1A) A method for treating a tumor in a human subject in need thereof comprising: (a) measuring a level of an RNA transcript of selected genes in a biological sample obtained from the human subject, wherein the selected genes comprise at least one gene selected from the group consisting of ADAM28, AQP3, ATP2C2, BNIP3L, CCDC60, CYP4X1, DUSP4, DUSP6, ETV4, ETV5, FGFR3, IL20RA, LAD1, LRIG1, MUC2, MUC6, NMUR2, PAK5, PDE10A, RHEX, S100A4, SPRED1, SPRY2, VSIG2 and ZG16B; (b) normalizing the level of the RNA transcript of each of the genes to obtain a normalized gene expression level for each gene; (c) calculating the signature score based on the mRNA expression level of the selected genes; (d) predicting the likelihood of the response based on the signature score; and (e) administering the FGFR inhibitor to the human subject who is predicted as having a response to a therapy comprising an FGFR inhibitor. (2A) The method of (1A), wherein the signature score is calculated based on the following equation: FGF Signature-25 score = (log2(ADAM28)+log2((AQP3)+log2(ATP2C2)+log2(BNIP3L)+log2(CCDC60)+log2(CYP4X1)+log2(DUSP4)+log2(DUSP6)+log2(ETV4)+log2(ETV5)+log2(FGFR3)+log2(IL20RA)+log2(LAD1)+log2(LRIG1)+log2(MUC2)+log2(MUC6)+log2(NMUR2)+log2(PAK5)+log2(PDE10A)+log2(RHEX)+log2(S100A4)+log2(SPRED1)+log2(SPRY2)+log2(VSIG2)+log2(ZG16B)) / 25. (3A) The method of (1A), wherein the selected genes are DUSP4, DUSP6, ETV4, FGFR3, IL20RA, PAK5 and SPRY2, and the signature score is calculated based on the following equation: FGF Signature-7 score = (log2(DUSP4)+log2(DUSP6)+log2(ETV4)+log2(FGFR3)+log2(IL20RA)+log2(PAK5)+log2(SPRY2)) / 7. (4A) A method for treating a tumor in a human subject in need thereof comprising: (a) measuring a level of an RNA transcript of selected genes in a biological sample obtained from the human subject, wherein the selected genes are DUSP4, DUSP6, ETV4, FGFR3, IL20RA, PAK5 and SPRY2; (b) normalizing the level of the RNA transcript of each of the genes to obtain a normalized gene expression level for each gene; (c) calculating the signature score based on the following equation: FGF Signature-7 score = (log2(DUSP4)+log2(DUSP6)+log2(ETV4)+log2(FGFR3)+log2(IL20RA)+log2(PAK5)+log2(SPRY2)) / 7; (d) predicting the likelihood of the response based on the signature score; and (e) administering the FGFR inhibitor to the human subject who is predicted as having a response to a therapy comprising an FGFR inhibitor. (5A) A method for treating a tumor in a human subject in need thereof comprising: (a) measuring a level of an RNA transcript of selected genes in a biological sample obtained from the human subject, wherein the selected genes are CCND1, CHAD, DUSP4, DUSP6, EFNA1, EPHA2, ETV4, FGF10, GADD45G, IL20RA, KAT2B, MLLT3, PPP2CB, PRKAA2, PTPRR, SMAD4, TNFRSF10B, TNFRSF10D, TNFSF10 and XPA; (b) normalizing the level of the RNA transcript of each of the genes to obtain a normalized gene expression level for each gene; (c) calculating the signature score based on the following equation: FGF Signature-20 score = (log2(CCND1)+log2(CHAD)+log2(DUSP4)+log2(DUSP6)+log2(EFNA1)+log2(EPHA2)+log2(ETV4)+log2(FGF10)+log2(GADD45G)+log2(IL20RA)+log2(KAT2B)+log2(MLLT3)+log2(PPP2CB)+log2(PRKAA2)+log2(PTPRR)+log2(SMAD4)+log2(TNFRSF10B)+log2(TNFRSF10D)+log2(TNFSF10)+log2(XPA)) / 20; (d) predicting the likelihood of the response based on the signature score; and (e) administering the FGFR inhibitor to the human subject who is predicted as having a response to a therapy comprising an FGFR inhibitor. (6A) A method for treating a tumor in a human subject in need thereof comprising: (a) measuring a level of an RNA transcript of selected genes in a biological sample obtained from the human subject, wherein the selected genes are BMPR1B, CCND1, DUSP4, ETV4, EFNA1, FGF10, GADD45G, HSPA1A, IRAK3, MAP2K4, MUTYH, NTHL1, PLCE1, SMAD3, TNFRSF10C and TNFRSF10D; (b) normalizing the level of the RNA transcript of each of the genes to obtain a normalized gene expression level for each gene; (c) calculating the signature score based on the following equation: FGF Signature-16 score = (log2(BMPR1B)+log2(CCND1)+log2(DUSP4)+log2(EFNA1)+log2(ETV4)+log2(FGF10)+log2(GADD45G)+log2(HSPA1A)+log2(IRAK3)+log2(MAP2K4)+log2(MUTYH)+log2(NTHL1)+log2(PLCE1)+log2(SMAD3)+log2(TNFRSF10C)+log2(TNFRSF10D)) / 16; (d) predicting the likelihood of the response based on the signature score; and (e) administering the FGFR inhibitor to the human subject who is predicted as having a response to a therapy comprising an FGFR inhibitor. (7A) The method of any one of (1A) to (6A), wherein the biological sample is selected from the group consisting of a blood sample, a serum sample, a plasma sample, and a tumor sample, preferably the biological sample is a tumor sample. (8A) The method of any one of (1A) to (7A), wherein the FGFR inhibitor is 5-((2-(4-(1-(2-hydroxyethyl)piperidin-4-yl)benzamide)pyridin-4-yl)oxy)-6-(2-methoxyethoxy)-N-methyl-1H-indole-1-carboxamide represented by Formula (I): or a pharmacologically acceptable salt thereof. (9A) The method of (8A), wherein the pharmacologically acceptable salt is 1.5 succinate. (10A) The method of any one of (1A) to (9A), wherein the tumor is breast cancer. (11A) The method of (10A), wherein the breast cancer is metastatic breast cancer, locally advanced breast cancer, recurrent breast cancer, or unresectable breast cancer. (12A) The method of (10A) or (11A), wherein the breast cancer is hormone receptor-positive. (13A) The method of (10A), wherein the breast cancer is hormone receptor-positive and HER2-negative metastatic breast cancer.

[0008] Fig. 1 shows the analysis of 47 genes in Example 1. According to the manual curation of all 47 genes in the FGF signal pathway in cancer, we have narrowed down 25 genes to finalize the FGF signature. AMP: FGFR1-4 amplifications (but do not have FGFR1-4 mutation / fusions, do not have driver mutations); Driver: oncogenic driver mutations (genes that are not FGFR1-4); FGFL: FGF Ligand; FGFL high: high ligand expression (but do not have FGFR1-4 mutations / fusion; do not have driver mutations; do not have FGFR amplifications); Fusion: gene fusions in FGFR1-4; MUT: gain-of-function mutations in FGFR1-4 (mutations in FGFR1-4 are always exclusive from FGFR1-4 fusions); Others: the rest of tumors not in the above groups; BCLA: Bladder Urothelial Carcinoma.Fig. 2 shows the Box-Whisker plot of FGF Signature-7 by CR / PR and SD / PD in Example 2. CR: Complete Response; PR: Partial Response; SD: Stable Disease; PD: Progressive Disease.Fig. 3 shows the ROC curve for CR / PR vs. SD / PD based on FGF Signature-7 in Example 2. AUC=0.808.Fig. 4 shows the cutoff determination with ORR for FGF Signature-7 in Example 2.Fig. 5 shows the dichotomized PFS analysis with cutoff = -0.918 for FGF Signature-7 in Example 2.Fig. 6 shows the dichotomized PFS analysis with cutoff = -1.175 for FGF Signature-7 in Example 2.Fig. 7 shows the Box-Whisker plot of FGF Signature-20 by CR / PR and SD / PD in Example 3.Fig. 8 shows he ROC curve for CR / PR vs. SD / PD based on FGF Signature-20 in Example 3. AUC=0.862.Fig. 9 shows the cutoff determination with ORR for FGF Signature-20 in Example 3.Fig. 10 shows the dichotomized PFS analysis with cutoff= -0.058 for FGF Signature-20 in Example 3.Fig. 11 shows the dichotomized PFS analysis with cutoff= -0.412 for FGF Signature-20 in Example 3.Fig. 12 shows the Box-Whisker plot of FGF Signature-16 by CR / PR and SD / PD in Example 4.Fig. 13 shows the ROC curve for CR / PR vs. SD / PD based on FGF Signature-16 in Example 4. AUC = 0.850Fig. 14 shows the cutoff determination with ORR for FGF Signature-16 in Example 4.Fig. 15 shows the dichotomized PFS analysis with cutoff= -0.148 for FGF Signature-16 in Example 4.Fig. 16 shows the dichotomized PFS analysis with cutoff= -0.530 for FGF Signature-16 in Example 4. NE: Not Evaluable; CI: Confidence Interval.

[0009] Various terms relating to aspects of the description are used throughout the specification and claims. Such terms are to be given their ordinary meaning in the art unless otherwise indicated. Other specifically defined terms are to be construed in a manner consistent with the definitions provided herein.

[0010] As used in this specification and the appended claims, the singular forms "a", "an", and "the" include plural referents unless the content clearly dictates otherwise.

[0011] The term "FGFR inhibitor" refers to a compound that inhibits tyrosine kinase of a fibroblast growth factor receptor (FGFR). The FGFR inhibitor may have an effect of inhibiting a kinase other than FGFR tyrosine kinase. The FGFR inhibitor is not particularly limited as long as it is capable of inhibiting FGFR tyrosine kinase, and examples thereof include Tasurgratinib, Alofanib, Anlotinib, Brivanib, Danusertib, Derazantinib, Dovitinib, Erdafitinib, Ferulic Acid, Fexagratinib, Futibatinib, Gambogenic Acid, Gunagratinib, HMPL-453, Infigratinib, Lirafugratinib, Lucitanib, Masitinib, Narazaciclib, Nintedanib, Pazopanib, Pemigatinib, Ponatinib, Resigratinib, Rogaratinib, Sulfatinib, Tinengotinib, Zoligratinib, and pharmacologically acceptable salts thereof.

[0012] The term "pharmacologically acceptable salt" is not particularly restricted as to the type of salt. Examples of such salts include, but are not limited to, inorganic acid addition salt such as hydrochloric acid salt, sulfuric acid salt, carbonic acid salt, bicarbonate salt, hydrobromic acid salt and hydriodic acid salt; organic carboxylic acid addition salt such as succinic acid salt, acetic acid salt, maleic acid salt, lactic acid salt, tartaric acid salt and trifluoroacetic acid salt; organic sulfonic acid addition salt such as methanesulfonic acid salt, hydroxymethanesulfonic acid salt, hydroxyethanesulfonic acid salt, benzenesulfonic acid salt, toluenesulfonic acid salt and taurine salt; amine addition salt such as trimethylamine salt, triethylamine salt, pyridine salt, procaine salt, picoline salt, dicyclohexylamine salt, N,N'-dibenzylethylenediamine salt, N-methylglucamine salt, diethanolamine salt, triethanolamine salt, tris(hydroxymethylamino)methane salt and phenethylbenzylamine salt; and amino acid addition salt such as arginine salt, lysine salt, serine salt, glycine salt, aspartic acid salt and glutamic acid salt.

[0013] The term "Tasurgratinib" refers to 5-((2-(4-(1-(2-hydroxyethyl)piperidin-4-yl)benzamide)pyridin-4-yl)oxy)-6-(2-methoxyethoxy)-N-methyl-1H-indole-1-carboxamide represented by Formula (I): or a pharmacologically acceptable salt thereof. An example of a pharmacologically acceptable salt of Tasurgratinib is Tasurgratinib 1.5 succinate. Tasurgratinib 1.5 succinate is also referred to as E7090.

[0014] The term "therapy comprising an FGFR inhibitor" means any therapy comprising an administration of an FGFR inhibitor.

[0015] Non-limiting examples of response are: a decrease in tumor size, a decrease in metastasis of a tumor, a prolonged period of progression-free survival, or a prolonged period of overall survival after treatment. To a human subject who is predicted as response to a therapy, the therapy is recommendable as a preferable treatment.

[0016] In certain embodiments, a subject is determined to have response to a therapy comprising an FGFR inhibitor, if the subject shows a partial response following treatment with the therapy. "Partial Response" means at least 30% decrease in the sum of the longest diameter (LD) of target lesions, taking as reference the baseline summed LD. In some embodiments, a subject is determined to have response to a therapy comprising an FGFR inhibitor, if the subject shows tumor shrinkage post-treatment with the therapy. "% of maximum tumor shrinkage" (MTS) means percent change of sum of diameters of target lesions, taking as reference the baseline sum diameters. In other embodiments, a subject is determined to have response to a therapy comprising an FGFR inhibitor, if the subject shows overall survival. "Overall Survival" (OS) refers to the time from randomization until death from any cause. "Randomization" means randomization of a patient into a test group or a control group when therapy plan for a patient is determined. In some embodiments, a subject is determined to have response to a therapy comprising an FGFR inhibitor, if the subject shows both prolonged period of overall survival and tumor shrinkage. In other embodiments, a subject is determined to have response to a therapy comprising an FGFR inhibitor, if the subject shows prolonged period of progression-free survival. "Progression-Free Survival" (PFS) refers to the time from the date of randomization to the date of first documentation of disease progression or death, whichever occurs first. In some embodiments, a subject is determined to have response to a therapy comprising an FGFR inhibitor, if the subject shows both prolonged period of progression free survival and tumor shrinkage. In some embodiments, a subject is determined to have response to a therapy comprising an FGFR inhibitor, if the subject shows prolonged period of overall survival, prolonged period of progression-free survival and tumor shrinkage.

[0017] A variety of suitable methods can be employed to measuring a level of an RNA transcript. For example, mRNA expression can be determined using RNA-sequencing, nCounter (NanoString Technologies(Registered Trademark)) systems, Northern blot or dot blot analysis, reverse transcriptase-PCR (RT-PCR; e.g., quantitative RT-PCR), in situ hybridization (e.g., quantitative in situ hybridization) or nucleic acid array (e.g., oligonucleotide arrays or gene chips) analysis. For example, a level of an RNA transcript was measured using nCounter systems (NanoString Technologies(Registered Trademark)), which uses molecular “barcodes” and microscopic imaging to detect and count up to several hundred unique transcripts in one hybridization reaction. Details of such methods are described below and in, e.g., Sambrook et al., Molecular Cloning: A Laboratory Manual Second Edition vol. 1, 2 and 3. Cold Spring Harbor Laboratory Press: Cold Spring Harbor, New York, USA, Nov. 1989; Gibson et al. (1996) Genome Res., 6(10):995-1001; and Zhang et al. (2005) Environ. Sci. Technol., 39(8):2777-2785; U.S. Publication No. 2004086915; European Patent No. 0543942; and U.S. Patent No. 7,101,663; the disclosures of each of which are incorporated herein by reference in their entirety.

[0018] In one embodiment, a level of an RNA transcript in a biological sample can be determined using nCounter systems (NanoString Technologies(Registered Trademark)). Details of such systems are described in US2003 / 0013091, US2007 / 0166708, US2010 / 0015607, US2010 / 0261026, US2010 / 0262374, US2010 / 01 12710, US2010 / 0047924, US2014 / 0371088, US201 1 / 0086774 and WO2017 / 015099.

[0019] In another embodiment, a level of an RNA transcript in a biological sample can be determined using nucleic acid (or oligonucleotide) arrays (e.g., an array described below under "Arrays and Kits"). For example, isolated mRNA from a biological sample can be amplified using RT-PCR with, e.g., random hexamer or oligo(dT)-primer mediated first strand synthesis. The amplicons can be fragmented into shorter segments. The RT-PCR step can be used to detectably-label the amplicons, or, optionally, the amplicons can be detectably-labeled subsequent to the RT-PCR step. For example, the detectable-label can be enzymatically (e.g., by nick-translation or kinase such as T4 polynucleotide kinase) or chemically conjugated to the amplicons using any of a variety of suitable techniques (see, e.g., Sambrook et al., supra). The detectably-labeled-amplicons are then contacted with a plurality of polynucleotide probe sets, each set containing one or more of a polynucleotide (e.g., an oligonucleotide) probe specific for (and capable of binding to) a corresponding amplicon, and where the plurality contains many probe sets each corresponding to a different amplicon.

[0020] In another embodiment, a level of an RNA transcript in a biological sample can be determined by isolating total mRNA from the biological sample (see, e.g., Sambrook et al. (supra) and U.S. Patent No. 6,812,341) and subjecting the isolated mRNA to agarose gel electrophoresis to separate the mRNA by size. The size-separated mRNAs are then transferred (e.g., by diffusion) to a solid support such as a nitrocellulose membrane. A level of an RNA transcript in the biological sample can then be determined using one or more detectably-labeled-polynucleotide probes, complementary to the mRNA sequence of interest, which bind to and thus render detectable their corresponding mRNA. Detectable-labels include, e.g., fluorescent (e.g., umbelliferone, fluorescein, fluorescein isothiocyanate, rhodamine, dichlorotriazinylamine fluorescein, dansyl chloride, allophycocyanin (APC), or phycoerythrin), luminescent (e.g., europium, terbium, QdotTMnanoparticles supplied by the Quantum Dot Corporation, Palo Alto, CA), radiological (e.g., 1251, 1311, 35S, 32P, 33P, or 3H), and enzymatic (horseradish peroxidase, alkaline phosphatase, beta-galactosidase, or acetylcholinesterase) labels.

[0021] Generally, the probe sets are bound to a solid support and the position of each probe set is predetermined on the solid support. The binding of a detectably-labeled amplicon to a corresponding probe of a probe set indicates the level of the target RNA transcript in the biological sample. Additional methods for measuring a level of an RNA transcript using nucleic acid arrays are described in, e.g., U.S. Patent Nos. 5,445,934; 6,027,880; 6,057,100; 6,156,501; 6,261,776; and 6,576,424; the disclosures of each of which are incorporated herein by reference in their entirety.

[0022] The term "normalized" with regard to a gene expression level refers to the level of the transcript relative to the mean levels of transcripts of a set of reference genes. The reference genes frequently used to normalize gene expression are mRNAs for the housekeeping genes such as ACAD9, AGK, AMMECR1L, C10orf76, CC2D1B, CNOT10, CNOT4, COG7, DDX50, DHX16, DNAJC14, EDC3, EIF2B4, ERCC3, FCF1, FTSJ2, GPATCH3, HDAC3, MRPS5, MTMR14, NOL7, NUBP1, PIAS1, PIK3R4, PRPF38A, RBM45, SAP130, SF3A3, SLC4A1AP, TLK2, TMUB2, TRIM39, TTC31, USP39, VPS33B, ZC3H14, ZKSCAN5, ZNF143, ZNF346, and ZNF384.

[0023] Suitable biological samples for the methods described herein include any biological fluid, cell, tissue, or fraction thereof, which contains RNA transcripts to be measured. A biological sample can be, for example, a specimen obtained from a human subject or can be derived from such a subject. For example, a sample can be a tissue section obtained by biopsy, archived tumor tissue, or cells that are placed in or adapted to tissue culture. A biological sample can also be a biological fluid such as blood, plasma, serum, or such a sample absorbed onto a substrate (e.g., glass, polymer, paper). A biological sample can also include a tumor sample. In specific embodiments, the biological sample is a tumor cell or a tumor tissue obtained from a region of the subject containing a tumor. For example, the biological sample may be a breast cancer sample. A biological sample can be further fractionated, if desired, to a fraction containing particular cell types. For example, a blood sample can be fractionated into serum or into fractions containing particular types of blood cells such as red blood cells or white blood cells (leukocytes). If desired, a sample can be a combination of samples from a subject such as a combination of a tissue and fluid sample.

[0024] Any suitable methods for obtaining the biological samples can be employed, although exemplary methods include, e.g., phlebotomy, fine needle aspirate biopsy procedure. Samples can also be collected, e.g., by microdissection (e.g., laser capture microdissection (LCM) or laser microdissection (LMD)).

[0025] Methods for obtaining and / or storing samples that preserve the activity or integrity of molecules (e.g., nucleic acids or proteins) in the sample are well known to those skilled in the art. For example, a biological sample can be further contacted with one or more additional agents such as buffers and / or inhibitors, including one or more of nuclease, protease, and phosphatase inhibitors, which preserve or minimize changes in the molecules (e.g., nucleic acids or proteins) in the sample. Such inhibitors include, for example, chelators such as ethylenediamine tetraacetic acid (EDTA), ethylene glycol bis(P-aminoethyl ether) N,N,N1,Nl-tetraacetic acid (EGTA), protease inhibitors such as phenylmethylsulfonyl fluoride (PMSF), aprotinin, leupeptin, antipain, and the like, and phosphatase inhibitors such as phosphate, sodium fluoride, vanadate, and the like. Suitable buffers and conditions for isolating molecules are well known to those skilled in the art and can be varied depending, for example, on the type of molecule in the sample to be characterized (see, for example, Ausubel et al. Current Protocols in Molecular Biology (Supplement 47), John Wiley & Sons, New York (1999); Harlow and Lane, Antibodies: A Laboratory Manual (Cold Spring Harbor Laboratory Press (1988); Harlow and Lane, Using Antibodies: A Laboratory Manual, Cold Spring Harbor Press (1999); Tietz Textbook of Clinical Chemistry, 3rd ed. Burtis and Ashwood, eds. W.B. Saunders, Philadelphia, (1999)). A sample also can be processed to eliminate or minimize the presence of interfering substances. For example, a biological sample can be fractionated or purified to remove one or more materials that are not of interest. Methods of fractionating or purifying a biological sample include, but are not limited to, chromatographic methods such as liquid chromatography, ion-exchange chromatography, size-exclusion chromatography, or affinity chromatography. For use in the methods described herein, a sample can be in a variety of physical states. For example, a sample can be a liquid or solid, can be dissolved or suspended in a liquid, can be in an emulsion or gel, or can be absorbed onto a material.

[0026] The type of tumor of a human subject is not particularly restricted, and may be breast cancer, stomach cancer, non-small-cell lung cancer, bladder cancer, endometrial cancer, hepatocellular carcinoma, bile duct cancer, melanoma, esophageal cancer, colorectal cancer, renal cell carcinoma, head and neck cancer, pleural mesothelioma or Hodgkin's lymphoma. In one embodiment, the tumor is breast cancer. The breast cancer may be metastatic breast cancer, locally advanced breast cancer, recurrent breast cancer, or unresectable breast cancer. The breast cancer may be hormone receptor-positive. The breast cancer may be hormone receptor-positive and HER2-negative metastatic breast cancer.

[0027] In an embodiment, the signature score is calculated based on the following equation: FGF Signature-25 score = (log2(ADAM28)+log2((AQP3)+log2(ATP2C2)+log2(BNIP3L)+log2(CCDC60)+log2(CYP4X1)+log2(DUSP4)+log2(DUSP6)+log2(ETV4)+log2(ETV5)+log2(FGFR3)+log2(IL20RA)+log2(LAD1)+log2(LRIG1)+log2(MUC2)+log2(MUC6)+log2(NMUR2)+log2(PAK5)+log2(PDE10A)+log2(RHEX)+log2(S100A4)+log2(SPRED1)+log2(SPRY2)+log2(VSIG2)+log2(ZG16B)) / 25. In another embodiments, the signature score is calculated based on the following equation: FGF Signature-7 score = (log2(DUSP4)+log2(DUSP6)+log2(ETV4)+log2(FGFR3)+log2(IL20RA)+log2(PAK5)+log2(SPRY2)) / 7. In another embodiments, the signature score is calculated based on the following equation: FGF Signature-20 score = (log2(CCND1)+log2(CHAD)+log2(DUSP4)+log2(DUSP6)+log2(EFNA1)+log2(EPHA2)+log2(ETV4)+log2(FGF10)+log2(GADD45G)+log2(IL20RA)+log2(KAT2B)+log2(MLLT3)+log2(PPP2CB)+log2(PRKAA2)+log2(PTPRR)+log2(SMAD4)+log2(TNFRSF10B)+log2(TNFRSF10D)+log2(TNFSF10)+log2(XPA)) / 20. In another embodiments, the signature score is calculated based on the following equation: FGF Signature-16 score = (log2(BMPR1B)+log2(CCND1)+log2(DUSP4)+log2(EFNA1)+log2(ETV4)+log2(FGF10)+log2(GADD45G)+log2(HSPA1A)+log2(IRAK3)+log2(MAP2K4)+log2(MUTYH)+log2(NTHL1)+log2(PLCE1)+log2(SMAD3)+log2(TNFRSF10C)+log2(TNFRSF10D)) / 16. log2(ADAM28) refers to logarithm to base 2-transformed normalized expression level of ADAM28, and this applies similarly to other genes.

[0028] In some embodiments, the signature score is compared to the control. The control is a pre-established cut-off value. In one embodiment, the pre-established cut-off value is a signature score that is determined based on receiver operating characteristic (ROC) analysis or percentile analysis predicting tumor response with a higher positive predictive value compared to no cut-off, and wherein a signature score equal to or below the pre-established cut-off value is a low signature score and a value higher than the pre-established cut-off value is a high signature score. The tumor response is an objective response rate (ORR) or % of maximum tumor shrinkage. In another embodiment, the pre-established cut-off value is a signature score at which the ORR difference was approximately 30% between subjects with signature scores above and below Cutoff was set as the Cutoff2, and the signature score at which the Cutoff difference peaked at ORR > 30% was set as the Cutoff1, and wherein a signature score equal to or below the pre-established cut-off value is a low signature score and a value higher than the pre-established cut-off value is a high signature score.

[0029] In some embodiments, as described above, the methods described herein can involve, calculating a signature score of a biological sample obtained from a human subject having a tumor, wherein the signature score, compared to a control, predicts that the human subject is in need of (or can benefit from) a treatment comprising an FGFR inhibitor. In certain embodiments, when the signature score in a biological sample from a subject having a tumor is higher than the control, the subject is identified as in need of a therapy comprising an FGFR inhibitor. In this context, the term “control” includes a biological sample (e.g., from the same tumor tissue) obtained from a human subject who is not in need of a therapy comprising an FGFR inhibitor. Such subject who is not in need of a therapy comprising an FGFR inhibitor may include a human subject who is predicted to respond to the therapy but such response to the therapy is not significantly better than a predicted response to a therapy with other drugs. The term “control” also includes a biological sample (e.g., from the same tumor tissue) obtained in the past from a human subject who is known to be not in need of a therapy comprising an FGFR inhibitor and used as a reference for future comparisons to test samples taken from human subjects for which necessity for the therapy is to be predicted.

[0030] In some embodiments, a “positive control” may be used instead of a “control.” The “positive control” signature score of a biological may alternatively be pre-established by an analysis of one or more human subjects that have been identified as in need of a therapy comprising an FGFR inhibitor.

[0031] In certain embodiments, the “control” is a pre-determined cut-off value.

[0032] In some embodiments, the methods described herein include determining if the signature score falls above or below a predetermined cut-off value.

[0033] In accordance with the methods described herein, a reference signature score is identified as a cut-off value, above or below of which is predictive of having response to a therapy comprising an FGFR inhibitor. Some cut-off values are not absolute in that clinical correlations can still remain significant over a range of values on either side of the cutoff; however, it is possible to select an optimal cut-off value signature score for a particular sample type. Cut-off values determined for use in the methods described herein can be compared with, e.g., published ranges of signature score but can be individualized to the methodology used and patient population. It is understood that improvements in optimal cut-off values could be determined depending on the sophistication of statistical methods used and on the number and source of samples used to determine reference level values for the different sample types. Therefore, established cut-off values can be adjusted up or down, on the basis of periodic re-evaluations or changes in methodology or population distribution.

[0034] The pre-established cut-off value can be a signature score that is determined based on receiver operating characteristic (ROC) analysis. Consider the situation where there are two groups of patients and by using an established standard technique one group is known to have a response to a therapy comprising an FGFR inhibitor, and the other is known to not having a response to a therapy comprising an FGFR inhibitor. A measurement using a biological sample from all members of the two groups is used to test for the response to a therapy comprising an FGFR inhibitor. The test will find some, but not all, subjects that have a response to a therapy comprising an FGFR inhibitor. The ratio of the subjects having a response to a therapy found by the test to the total number of the subjects having a response to the therapy (known by the established standard technique) is the true positive rate (also known as sensitivity). The test will find some, but not all, the human subjects not having a response to a therapy comprising an FGFR inhibitor. The ratio of the subjects not having a response to the therapy found by the test to the total number of the subjects not having a response to the therapy (known by the established standard technique) is the true negative rate (also known as specificity). It is created by plotting the fraction of true positives out of the positives versus the fraction of false positives out of the negatives, at various threshold settings.

[0035] In one embodiment, the signature score is determined based on cut-off value predicting tumor response with a positive predictive value, wherein a signature score equal to or below the pre-established cut-off value is a low signature score and a value higher than the pre-established cut-off value is a high signature score. The positive predictive value is the proportion of positive test results that are true positives; it reflects the probability that a positive test reflects the underlying condition being tested for. Methods of constructing ROC curves and determining positive predictive values are well known in the art. In certain embodiments, tumor response is an objective response rate (ORR), a clinical benefit rate (CBR) or % of maximum tumor shrinkage (MTS).

[0036] In another embodiment, the pre-established cut-off value is a signature score at which the ORR difference was approximately 30% between subjects with signature scores above and below Cutoff was set as the Cutoff2, and the signature score at which the Cutoff difference peaked at ORR > 30% was set as the Cutoff1, and wherein a signature score equal to or below the pre-established cut-off value is a low signature score and a value higher than the pre-established cut-off value is a high signature score.

[0037] In all of these embodiments, a signature score equal to or below the pre-established cut-off value is a low signature score and a value higher than the pre-established cut-off value is a high signature score. Examples

[0038] Example 1: TCGA derived FGF signature development (FGF Signature-25) Using RNA-sequencing data from the public cancer genome database, The Cancer Genome Atlas (TCGA), and literature curation, 25 genes were identified as highly expressed in tumor specimens with FGFR gene abnormalities targeted by FDA-approved FGFR inhibitors. With this 25 genes FGF Signature, FGF / FGFR signal activation in E7090-J081-102 (NCT04572295; 102 Study) clinical tumor samples was evaluated.

[0039] Four tumor types were selected in the TCGA dataset to develop an FGF gene signature. Bladder cancer and cholangiocarcinoma were selected as FGFR inhibitors have been approved by regulatory agencies for these 2 indications in tumors harboring FGFR genetic alterations including activating gene mutations or gene fusions. In addition, lung squamous cell carcinoma and endometrial cancer with relatively frequent FGFR mutation or fusion, 5% and 12% respectively, were also selected to increase the statistical power of the analysis.

[0040] In each of the above-selected 4 tumor types, differential gene expression analysis was performed to identify genes expressed at higher level in FGFR-altered (defined as activating gene mutation or gene fusion based on OncoKB annotation “Oncogenic” or “Likely Oncogenic”) tumors than tumors in the control group. The control group included tumors not harboring FGFR mutations or fusions, not harboring FGFR amplifications, not harboring other well-characterized oncogenic driver mutations (KRAS, NRAS, HRAS, PIK3CA in bladder cancers; KRAS, BRAF, PIK3CA, IDH1, IDH2 in cholangiocarcinoma; EGFR, KRAS, ALK, ROS1, RET, MET, BRAF, NTRK in lung squamous cell carcinoma; POLE, PIK3CA, KRAS, CTNNB1 in endometrial cancers), and not expressing high level FGF ligands (defined as the top quartile of total FGF1-23 expression in each tumor type). For each of the 4 tumor types, two-sample Wilcoxon rank-sum test was performed for differential expression analysis between FGFR-altered tumors and tumors in the control group. P-values were adjusted by the Benjamini-Hochberg procedure. Finally, a meta-analysis was performed to combine the results of the 4 tumor types into a single overall statistical test using Fisher’s method, and a meta-analysis p-value was computed for each gene.

[0041] To select FGF signature genes, the following criteria were determined: the adjusted meta-analysis p-value < 0.05, fold changes of gene expression in FGFR-altered tumors when compared with the control group were greater than 1.5 in at least 3 of the 4 analyzed tumor types. From the above differential gene expression analysis, 47 genes met these criteria. This initial 47-gene FGF signature was further refined as follows (see Fig. 1). Of the 47 genes, additional filtering was applied to select genes that met predefined empirical p-value threshold in 3 of the 4 tumor types (p<0.01 for bladder cancer, p<0.2 for cholangiocarcinoma, p<0.05 for lung squamous cell carcinoma, and p<0.01 for endometrial cancer). This additional filtering resulted in 13 genes (ADAM28, AQP3, ATP2C2, CCDC60, CYP4X1, DUSP6, FGFR3, IL20RA, NMUR2, PDE10A, RHEX, VSIG2, ZG16B) that demonstrated consistent overexpression with relatively large fold change in multiple tumor types. Inclusion of these 13 genes in the final 25-gene signature was mainly based on a data-driven approach.

[0042] The criteria for selecting the initial 47 genes were then relaxed to: adjusted meta-analysis p-value < 0.05, fold changes of gene expression in FGFR-altered tumors versus the control group were greater than 1.2 in at least 3 of the 4 analyzed tumor types. Based on these criteria, 138 genes were identified. After excluding the 13 genes that were included in the final 25-gene signature as described above, the remaining 125 genes were subject to literature search. Of these 125 genes, 8 genes (BNIP3L, LAD1, LRIG1, MUC2, MUC6, S100A4, SPRED1, SPRY2) were confirmed as FGF pathway downstream genes, and their expression were upregulated by FGF pathway activation in preclinical studies. During literature review, 4 additional genes (PAK7 (synonym of PAK5), ETV4, ETV5, DUSP4) were identified as FGF pathway downstream genes consistently reported in multiple papers using different preclinical model systems. Manual inspection of these 4 genes showed that they were upregulated in FGFR-altered tumors in 2 of the 4 analyzed TCGA tumor types. These 12 genes (8 plus 4) were combined with the above 13 genes to constitute the final 25-gene FGF signature (see table 1). Inclusion of the 12 genes was based on a combined data-driven and biology-driven approach.

[0043]

[0044] Examples 2-4: 102 study (Hormone Receptor (HR)+, HER2- breast cancer) biomarker analysis:

[0045] Example 2. TCGA derived FGF signature-7 vs. clinical outcome Since only 7 genes are included in the nCounter Pan-cancer Panel (DUSP4, DUSP6, FGFR3, ETV4, IL20RA, PAK7 (synonym of PAK5), SPRY2) among the genes constituting the above TCGA-derived FGF signature (FGF Signature-25), the FGF signature for nCounter dataset was prepared. The correlation analysis between the algorithm with 7 genes and the clinical efficacy in the 102 Study (Best Overall Response (BOR), Progression-Free Survival (PFS)) suggested that FGF signature can predict the therapeutic effect of FGFR inhibitors, including E7090, in HR+, HER2- breast cancer. The Area Under Curve (AUC) of the ROC curve was 0.808 (see Fig. 3). In addition, the score at which the overall response rate (ORR) difference was approximately 30% between subjects with signature scores above and below Cutoff was set as the Cutoff2, and the score at which the Cutoff difference peaked at ORR > 30% was set as the Cutoff1 (Fig. 4: Cutoff determination with ORR for FGF Signature-7). In the dichotomized PFS analysis using cutoff values of -0.918 (see Fig. 5) and -1.175 (see Fig. 6), the hazard ratios (HRs) were 0.382 and 0.524, respectively.

[0046]

[0047] Example 3. New FGF signature from clinical trial sample's RNA expression data (FGF signature-20) A new clinical FGF signature for predicting clinical outcomes was developed from the 102 study nCounter dataset. We tried to identify genes with significant correlation with BOR (logistic regression) and found high baseline expressions of 20 significant genes (CCND1, CHAD, DUSP4, DUSP6, EFNA1, EPHA2, ETV4, FGF10, GADD45G, IL20RA, KAT2B, MLLT3, PPP2CB, PRKAA2, PTPRR, SMAD4, TNFRSF10B, TNFRSF10D, TNFSF10, XPA) correlates with clinical response with p-value <0.05. Then, we established a new clinical FGF signature score with the average log2expression for 20 genes(see Fig. 7). The correlation analysis between the new algorithm with 20 genes and the clinical efficacy in the 102 study (BOR, % of maximum tumor shrinkage (MTS)) indicated better predictive power as expected, AUC=0.862 with ROC curve (see Fig. 8). In addition, the score at which the ORR difference was approximately 30% between subjects with signature scores above and below Cutoff was set as the Cutoff2, and the score at which the Cutoff difference peaked at ORR > 30% was set as the Cutoff1 (Fig. 9: Cutoff determination with ORR for FGF Signature-20). In the dichotomized PFS analysis using cutoff values of -0.058 (see Fig. 10) and -0.412 (see Fig. 11), the HRs were 0.138 and 0.506, respectively.

[0048]

[0049] Example 4. Develop new FGF signatures using 102 Study Pharmacodynamics marker data (FGF signature-16) The genes with significant correlation with BOR and significantly decreased by E7090 treatment were identified. Correlation between high baseline expression and BOR (logistic regression, p<0.2) was observed in 80 genes. Also, a significant decrease in Cycle 3 Day 1 vs. baseline (paired Wilcox test, p<0.2) was observed in 74 genes, and the overlap of high baseline expression and a significant decrease was 16 in genes (BMPR1B, CCND1, DUSP4, ETV4, EFNA1, FGF10, GADD45G, HSPA1A, IRAK3, MAP2K4, MUTYH, NTHL1, SMAD3, TNFRSF10C, TNFRSF10D, PLCE1) (see Fig. 12). The correlation analysis of this 16 genes signature and clinical outcome resulted in AUC=0.850 with ROC curve (see Fig. 13). In addition, the score at which the ORR difference was approximately 30% between subjects with signature scores above and below Cutoff was set as the Cutoff2, and the score at which the Cutoff difference peaked at ORR > 30% was set as the Cutoff1 (Fig. 14: Cutoff determination with ORR for FGF Signature-16). In the dichotomized analysis of median PFS using Cutoff values of -0.148 (see Fig. 15) and -0.530 (see Fig. 16), the HRs were 0.278 and 0.484, respectively.

[0050]

Claims

1. A method for predicting the likelihood of a response of a human subject having a tumor to a therapy comprising an FGFR inhibitor, comprising: (a) measuring a level of an RNA transcript of selected genes in a biological sample obtained from the human subject, wherein the selected genes comprise at least one gene selected from the group consisting of ADAM28, AQP3, ATP2C2, BNIP3L, CCDC60, CYP4X1, DUSP4, DUSP6, ETV4, ETV5, FGFR3, IL20RA, LAD1, LRIG1, MUC2, MUC6, NMUR2, PAK5, PDE10A, RHEX, S100A4, SPRED1, SPRY2, VSIG2 and ZG16B; (b) normalizing the level of the RNA transcript of each of the genes to obtain a normalized gene expression level for each gene; (c) calculating the signature score based on the mRNA expression level of the selected genes; and (d) predicting the likelihood of the response based on the signature score.

2. The method of claim 1, wherein the signature score is calculated based on the following equation: FGF Signature-25 score = (log2(ADAM28)+log2((AQP3)+log2(ATP2C2)+log2(BNIP3L)+log2(CCDC60)+log2(CYP4X1)+log2(DUSP4)+log2(DUSP6)+log2(ETV4)+log2(ETV5)+log2(FGFR3)+log2(IL20RA)+log2(LAD1)+log2(LRIG1)+log2(MUC2)+log2(MUC6)+log2(NMUR2)+log2(PAK5)+log2(PDE10A)+log2(RHEX)+log2(S100A4)+log2(SPRED1)+log2(SPRY2)+log2(VSIG2)+log2(ZG16B)) / 25.

3. The method of claim 1, wherein the selected genes are DUSP4, DUSP6, ETV4, FGFR3, IL20RA, PAK5 and SPRY2, and the signature score is calculated based on the following equation: FGF Signature-7 score = (log2(DUSP4)+log2(DUSP6)+log2(ETV4)+log2(FGFR3)+log2(IL20RA)+log2(PAK5)+log2(SPRY2)) / 7.

4. A method for predicting the likelihood of a response of a human subject having a tumor to a therapy comprising an FGFR inhibitor, comprising: (a) measuring a level of an RNA transcript of selected genes in a biological sample obtained from the human subject, wherein the selected genes are DUSP4, DUSP6, ETV4, FGFR3, IL20RA, PAK5 and SPRY2 in a biological sample obtained from the human subject; (b) normalizing the level of the RNA transcript of each of the genes to obtain a normalized gene expression level for each gene; (c) calculating the signature score based on the following equation: FGF Signature-7 score = (log2(DUSP4)+log2(DUSP6)+log2(ETV4)+log2(FGFR3)+log2(IL20RA)+log2(PAK5)+log2(SPRY2)) / 7; and (d) predicting the likelihood of the response based on the signature score.

5. A method for predicting the likelihood of a response of a human subject having a tumor to a therapy comprising an FGFR inhibitor, comprising: (a) measuring a level of an RNA transcript of selected genes in a biological sample obtained from the human subject, wherein the selected genes are CCND1, CHAD, DUSP4, DUSP6, EFNA1, EPHA2, ETV4, FGF10, GADD45G, IL20RA, KAT2B, MLLT3, PPP2CB, PRKAA2, PTPRR, SMAD4, TNFRSF10B, TNFRSF10D, TNFSF10 and XPA in a biological sample obtained from the human subject; (b) normalizing the level of the RNA transcript of each of the genes to obtain a normalized gene expression level for each gene; (c) calculating the signature score based on the following equation: FGF Signature-20 score = (log2(CCND1)+log2(CHAD)+log2(DUSP4)+log2(DUSP6)+log2(EFNA1)+log2(EPHA2)+log2(ETV4)+log2(FGF10)+log2(GADD45G)+log2(IL20RA)+log2(KAT2B)+log2(MLLT3)+log2(PPP2CB)+log2(PRKAA2)+log2(PTPRR)+log2(SMAD4)+log2(TNFRSF10B)+log2(TNFRSF10D)+log2(TNFSF10)+log2(XPA)) / 20; and (d) predicting the likelihood of the response based on the signature score.

6. A method for predicting the likelihood of a response of a human subject having a tumor to a therapy comprising an FGFR inhibitor, comprising: (a) measuring a level of an RNA transcript of selected genes in a biological sample obtained from the human subject, wherein the selected genes are BMPR1B, CCND1, DUSP4, ETV4, EFNA1, FGF10, GADD45G, HSPA1A, IRAK3, MAP2K4, MUTYH, NTHL1, PLCE1, SMAD3, TNFRSF10C and TNFRSF10D in a biological sample obtained from the human subject; (b) normalizing the level of the RNA transcript of each of the genes to obtain a normalized gene expression level for each gene; (c) calculating the signature score based on the following equation: FGF Signature-16 score = (log2(BMPR1B)+log2(CCND1)+log2(DUSP4)+log2(EFNA1)+log2(ETV4)+log2(FGF10)+log2(GADD45G)+log2(HSPA1A)+log2(IRAK3)+log2(MAP2K4)+log2(MUTYH)+log2(NTHL1)+log2(PLCE1)+log2(SMAD3)+log2(TNFRSF10C)+log2(TNFRSF10D)) / 16; and (d) predicting the likelihood of the response based on the signature score.

7. The method of any one of claims 1 to 6, wherein the biological sample is a tumor sample.

8. The method of any one of claims 1 to 7, wherein the FGFR inhibitor is 5-((2-(4-(1-(2-hydroxyethyl)piperidin-4-yl)benzamide)pyridin-4-yl)oxy)-6-(2-methoxyethoxy)-N-methyl-1H-indole-1-carboxamide represented by Formula (I): or a pharmacologically acceptable salt thereof.

9. The method of claim 8, wherein the pharmacologically acceptable salt is 1.5 succinate.

10. The method of any one of claims 1 to 9, wherein the tumor is breast cancer.

11. The method of claim 10, wherein the breast cancer is metastatic breast cancer, locally advanced breast cancer, recurrent breast cancer, or unresectable breast cancer.

12. The method of claim 10 or 11, wherein the breast cancer is hormone receptor-positive.

13. The method of claim 10, wherein the breast cancer is hormone receptor-positive and HER2-negative metastatic breast cancer.

Images