A METHOD TO DETERMINE AGENTS FOR PERSONALIZED USE

Abstract

The present invention relates to a method for identifying one or more compounds specifically binding to a target structure of a given diseased tissue in an individual, said method comprises the determination of the binding affinity of a number of compounds to the one or more docking spaces of a mutated gene identified in the individual and identifying one or more compounds specifically binding to the mutated protein. Further, the present invention relates to a computer program comprising instructions which cause the computer to carry out several steps of the method.

Claims

1. A method for identifying one or more compounds specifically binding to a target structure of a given diseased tissue, comprising: (i) identifying a mutated gene in the transcriptome of said diseased tissue and identifying at least one mutation comprised in said mutated gene; (ii) providing a three-dimensional (3D) structure of a wild-type or homolog protein expressed by a wild-type or homolog gene corresponding to the mutated gene identified in step (i); (iii) determining a 3D structure of a mutated protein which is the expression product of the mutated gene identified in step (i) or one or more docking spaces thereof, comprising: (a) adapting the amino acid sequence of the 3D structure of the wild-type or homolog protein of step (ii) to the expression product of the mutated gene identified in step (i) and defining one or more docking spaces of the obtained 3D structure of mutated protein, or defining one or more docking spaces of the 3D structure of the wild-type or homolog protein of step (ii) and adapting the amino acid sequence of said one or more docking spaces to the expression product of the mutated gene identified in step (i); (iv) providing 3D structures of a selection of compounds and fitting each 3D structure of each compound with the one or more docking spaces of step (iii); (v) determining the binding affinity of each compound to the one or more docking spaces; and (vi) identifying one or more compounds specifically binding to the mutated protein.

2. The method of claim 1, wherein the step (i) of identifying a mutated gene and the at least one mutation comprises: (a) providing a sample from the diseased tissue containing mRNA; (b) optionally isolating and/or purifying the mRNA; (c) optionally generating cDNA from the mRNA by a polymerase chain reaction; and (d) identifying at least one mutation by means of at least one step selected from the group consisting of: sequencing the mRNA and/or the cDNA; hybridizing the mRNA and/or the cDNA with a chip containing a variety of single-stranded nucleotides embracing mutated and non-mutated sequences; and conducting a polymerase chain reaction with a number of primers including those specific for a particular mutation.

3. The method of claim 1, wherein the diseased tissue is a neoplasm.

4. The method of claim 1, wherein the mutated gene, the mutated protein, or a combination thereof is associated with the onset or progression of a neoplasm.

5. The method of claim 1, wherein the selection of compounds used in step (iv) comprises at least five compounds.

6. The method of claim 1, wherein: the 3D structure of the wild-type or homolog protein of step (ii) is a crystal structure, a 3D NMR structure or a calculated hypothetical three-dimensional structure and is, optionally, obtained from a structure database; and/or the mutation is a point mutation and the mutated protein differs from the non-mutated protein by a single amino acid moiety only and each docking space embraces the different single amino acid moiety.

7. The method of claim 1, wherein at least steps (ii)-(v) are conducted in a computer-assisted manner.

8. The method of claim 1, wherein at least one of the compounds of which 3D structures are provided in step (iv) is characterized by one or more of the properties selected from the group consisting of: the compound has a molecular weight of not more than 1000 Da, the compound is not approved as an antineoplastic agent, the compound is has known pharmacokinetic properties, and the compound is approved for one or more pharmaceutical purposes other than antineoplastic activity.

9. The method of claim 1, wherein step (v) of determining the binding affinity of each compound to the one or more docking spaces comprises: (a) generating a 3D grid box of each docking space of the mutated protein and of each compound, wherein each grid box comprises grid points defined in all three dimensions that provide pieces of information selected from the group consisting of charges, partial charges, the ability to form hydrogen bonds, the ability to form pi-pi-electron interactions, and the ability to form van-der-Waals forces; (b) fitting each 3D structure of a compound with the one or more docking spaces in a manner that the 3D structure of the compound can rotate and scans over each docking space; (c) determining the binding energy between each compound and each docking space at each grid point and calculating binding affinity for each compound at each 3D orientation with each docking space; and (d) determining the lowest binding affinity for each compound-protein interaction.

10. The method of claim 1, wherein the method further comprises the following steps: defining one or more docking spaces of the structure of the wild-type or homolog protein of step (ii) each corresponding to the respective docking spaces of the structure of the mutated protein of step (iii); fitting the compounds with these one or more docking spaces; determining the lowest binding energy of each compound to these one or more docking spaces and thereby determining the binding affinity; comparing the binding affinity of each compound to the docking spaces of the mutated and of the wild-type or homolog compound; and identifying one or more compounds having a higher binding affinity to the docking space of the wild-type or homolog protein than to the corresponding docking space of the mutated protein.

11. The method of claim 1, wherein determining the binding affinity of each compound to the one or more docking spaces includes using Lamarckian Genetic Algorithm.

12. The method of claim 1, wherein a docking space embraces the whole protein, the surface of the whole protein optionally including one or more potential binding pockets or only the surrounding area of the pharmacophore binding site.

13. The method of claim 1, wherein the diseased tissue is compared with comparable healthy tissue.

14. The method of claim 13, wherein the comparable healthy tissue is obtained from the same individual as the diseased tissue.

15. The method of claim 13, wherein the comparable healthy tissue is obtained from another individual of the same species.

16. The method of claim 1, wherein the diseased tissue bears one or more genetic variations selected from the group consisting of one or more mutations, one or more different alleles, one or more polymorphisms, or combinations of two or more thereof, in comparison to corresponding healthy tissue.

17. The method of claim 1, wherein the diseased tissue bears one or more mutations associated with the disease state of the diseased tissue in comparison to corresponding healthy tissue.

18. The method of claim 13, wherein the comparison between the diseased tissue with comparable healthy tissue is comparing the specific binding of the one or more compounds to one or more target structures of a given diseased tissue with the binding of said one or more compounds to target structures which are the counterparts in healthy tissue of the one or more target structures of the given diseased tissue.

19. The method of claim 1, wherein said method further comprises the step (vii) of determining toxicological and pharmacologic properties of the compounds identified in step (vi) from one or more databases and identifying a compound of comparably low toxicity and, optionally, high pharmacologic activity in antineoplastic treatment.

20. The method of claim 19, wherein said method is a method for identifying an antineoplastic agent which has antineoplastic activity against the neoplasm, wherein said antineoplastic agent is or comprises one or more compounds identified in any of steps (vi) or (vii).

21. The method of claim 1, wherein the compounds of the selection of compounds are approved for one or more pharmaceutical purposes.

22. The method of claim 21, wherein the compounds of the selection of compounds are approved for one or more pharmaceutical purposes other than antineoplastic activity and are not approved as antineoplastic agents.

23. A pharmaceutical composition comprising one or more compounds identified in any of steps (vi) or (vii) of claim 19 and a pharmaceutically acceptable carrier.

24. A method for treating a neoplasm in an individual, comprising administering a compound identified in any of steps (vi) or (vii) of claim 19.

25. A computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out at least steps (iv) and (v) of the method of claim 1.

26. A storage device comprising, stored thereon, the computer program of claim 25.

Description

EXAMPLES

[0153] Materials and Methods

[0154] 1 Isolation of RNA from Tumor Cells or Tissue Derived from the Patient.

[0155] The test sample was taken up directly after removal into RNA stabilization solution. Total RNA isolation was performed with a column-based extraction procedure to obtain pure RNA without DNA digestion. The quality of the RNA was proven and the threshold set was an RNA integrity score of 6.8 or higher. To exclude ribosomal RNA sequences from further analysis, the RNA was hybridized with eukaryotic ribosomal RNA biotin-labeled oligonucleotide probes to deplete ribosomal RNA from total RNA. For the preparation of poly A+ RNA, streptavidin-coated magnetic beads coupled with oligo-dT were used. Five micrograms total RNA were mixed with beads and RNA purification beads and incubated. After incubation for 5-10 min, the beads were pelleted on a magnetic stand and the supernatant can be discarded. After washing the beads with washing buffer, the beads were resuspended in elution buffer to elute RNA from the beads. Then, a binding process with binding buffer took place again. The RNA bead mix was eluted again and the RNA was fragmented by heat treatment at 65° C. for 5-10 min. the elution and prime mix contains hexamers with random sequences and reverse transcriptase and was used to start cDNA synthesis from the RNA templates the supernatant was transferred to the master mix and put into a PCR plate with the barcode sequence. If the thermal cycling was finished, the RNA strand was removed and substituted by a second cDNA strand. Using specific beads, double-stranded cDNA was separated from RNA and the reaction mix. Overhanging strand ends from fragmentation was finally digested by 3′-5′ exonuclease to blunt ends. 5′ overhangs were filled to blunt ends by polymerase.

[0156] This method was applicable for cells, solid tissues, blood and other body fluids. Total RNA quality and quantity was evaluated by a microfluidics-based platform. After loading, the sample migrates through micro-channels to electrophoretically separate the sample components. The fluorescent probe intercalates into RNA strands and the fluorescence was recorded. The example does not only refer to coding mRNA, but also for non-coding RNA and epigenetically changed DNA sequences, as well as proteins, peptides, lipids, and all other metabolic chemical substances.

[0157] 2. Determination of a Mutational Profile and Transcript Abundance by RNA-Sequencing

[0158] End-repaired, A-tailed and adaptor-ligated cDNA was PCR-amplified by 10 cycles. The library was sequenced in paired-end mode (2×100 bp) using commercial RNA sequencing systems. The resulting sequences were aligned to a reference genome. Discrepencies concerning point mutations, deletions amplifications insertions etc. were recorded. Normalized RNA expressions were quantified using the RPKM measure. RPKM values for transcripts and the ratios of transcripts were taken into consideration to calculate the overall RPKM value for each gene.

[0159] 3. Examination whether Three-Dimensional Protein Crystal Structures are Available

[0160] Several proteins encoded by genes found to be mutated in from the comparison of the mutational profile obtained from RNA sequencing with protein crystal structures of the correspondingly affected proteins. In cases, where isoforms or splicing variants of target proteins are available, several homology models were prepared in parallel. Information on alterations in helices, disulfide bridges secondary loop structures, distortions of beta-sheets etc. may change the protein conformation and therefore may alter the binding properties of drugs. In selected cases, sequence alignments of proteins from different species were performed, because the interspecies comparison can give information of interest about commonly conserved and unique sequence motifs, key amino acid positions in the pharmacophore domains, identical location of helix bending residues etc. Furthermore, co-crystallization of target proteins with other binding proteins, small molecules, antibodies, peptides etc. was also considered, since they may not only stabilize the protein of interest, but also change its conformation from an inactive in to an active state and vice versa.

[0161] Furthermore, electrostatic potential maps were calculated to determine hot spots of electron density that may interfere with binding properties of affected amino acid residues. This information may be of interest to find the most appropriate small molecule inhibitor drugs.

[0162] 4. Generation of Mutation-Specific Protein Homology Models that Resemble the Mutated Genes in Individual Tumors.

[0163] The method described herein was conducted on a high performance computer running on Linux etc. to meet the requirements of the multi-stage process of protein modeling. For several calculations, the supercomputer MOGON II (Mainz, Germany) was used. Either the crystallography-based structures of human target proteins or corresponding crystal structures from other species were used for homology modeling. Internet-based databases for protein crystal structures were searched for the availability of three-dimensional structures that could be used as templates to create models of patient-specific mutations. In cases, where crystal structures of human proteins were not available, corresponding protein structures from other species may serve a template to generate human protein homology models. Homology modeling was based on the creation of three-dimensional models of proteins with known amino acid sequence, but unknown crystal structure. A precondition for homology modeling was the existence of a crystal structure of a related protein. With an available crystal structure (e.g. a wild-type protein), the sequence of the known (wild-type) protein can be aligned to the protein with the still unknown 3D-structure (e.g. the mutant counterpart of the wild-type protein).

[0164] Based on the known crystal structure of the wild-type protein, a hypothetical 3D-structure of the corresponding mutant protein can be calculated. This homology model was as better as more conserved are the amino acid sequences of known and unknown proteins. As a first step, the protein sequence was downloaded from a corresponding website, (e.g. UniProt) in FASTA format. Then, the known 3D structure of the related protein, which should serve as template, was downloaded and both protein sequences were compared using BLAST (Basic Local Alignment Search Tool) and ClustalW2. Crystal structures or homology models of wild-type proteins were then modified by insertion of the amino acid exchanges delineated from RNA-sequencing of the specific patient tumor. The subsequent homology models of mutated proteins were created using the alignment file with appropriate alignment programs.

[0165] The Swiss-MODEL structure assessment tool was then used to select the best homology model for molecular docking. Model evaluation was done with the help several tools (Anolea, GROMOS, QMEAN, DFIRE etc.). In a cellular environment. proteins existed in a hydrated form. Therefore, hydrogens were added to Asn and Gln residues.

[0166] 5. Bioinformatic Screening of a Library of FDA-Approved Drugs that Preferentially Bind with High Affinity to these Mutated Proteins.

[0167] A high performance Linux-based computer cluster was desirable for running virtual drug screening campaigns in sufficiently short time to deliver results to the decision-making physicians A library of FDA-approved drugs (>1500 compounds) was used to investigate the binding of drug to the mutation-specific protein homology models by means of specific virtual drug screening programs. These FDA-approved drugs do not only contain anticancer drugs, but drugs that were used for all kinds of diseases. The idea was that drugs frequently do not act in a mono-specific manner, but have broader activity spectra. Therefore, drugs for a specific disease indication may also inhibit related mutated proteins as in cancer. These inhibitory drugs were identified by bioinformatic calculation of drug-protein binding affinities. With this approach, approved drugs could be used off-label to treat individual tumors according to their individual mutations. This was the main concept of the present drug repurposing invention. Several algorithms to identify the best binding drugs with independent techniques were used. As an example, the 10 top-ranked out of >1500 FDA-approved drugs with highest affinities was selected. Homology-modeled mutant patient-specific proteins were set as rigid receptor molecules.

[0168] The prepared output files indicated information on atomic partial changes, torsion degrees of freedom and different atom types was added, e.g. aliphatic and aromatic carbon or polar atoms forming hydrogen bonds such as in PDQT format. In cases, where target proteins contain known pharmacophore sites, grids around selected amino acid residues of that pharmacophore were defined to calculate drug binding (defined docking approach). In those cases, where no drug-binding site of a target protein was known, interaction energies for the whole protein (blind docking approach) were first calculated. The region showing with the highest binding affinity were then be used to set a grid and a defined docking followed as a second step. The grid box was constructed to define docking spaces.

[0169] The dimensions of the grid box was set around the entire protein (blind docking approach) or around defined pharmacophore sites (defined docking approach) in a manner that the ligand could freely move and rotate in the docking space. The grid box consists of for instance 126 grid points in all three dimensions (X, Y and Z axes) separated by a distance of for instance 1 between each one. Energies at each grid point were then evaluated for each atom type present in the ligand, and the values were used to predict the energy of a particular ligand configuration. Three independent docking calculations were conducted, with 25,000,000 energy evaluations and 250 runs by using the Lamarckian Genetic Algorithm. The corresponding binding energies and the number of conformations in each cluster were attained from the docking log files (dig). The corresponding lowest binding energies (LBE) were obtained from the docking log files (dig), and mean values ±SD were calculated.

[0170] The docking results were visualized to prove the correct binding of the drugs to the relevant drug-binding sites of the mutated tumor proteins. By using databases and computer algorithms, the identified drug candidates were examined for their toxicity profile and their potential interaction with other potentially co-medicated drugs. To prove the specificity of the identified candidate drugs for a given mutated target protein, the binding of this drug to both the mutated and the wild-type protein models was performed. If more models are available (splice variants, proteins from other species), they were also be included in the docking procedure to obtain the best possible information about binding of this drug to the target protein.

[0171] To set up molecular docking, the data were first copied into the corresponding folder of the ligand docking program. Before doing so, two-dimensional chemical structures were converted to three-dimensional ones using appropriate software programs. The energy of the compound was minimized and the new structure saved as mol file. For subsequent molecular docking, the files of the ligands were prepared in pdbqt format, the ones of the target proteins in gpf, glg, and dpf file format. Then, the script for running the docking was prepared. Each calculation has a maximum runtime of five days (=7200 min). Each calculation was started using the script. The results of the running jobs were saved in the directory of the ligand. After finalizing the jobs, the results can be copied to personal computers. For docking campaigns of more than 64 ligands, a node-long script was used. Furthermore, it was taken into account that a drug that has been identified to bind to a given target protein found in the tumor genome of a patient might not only bind to this protein but also to several others. Binding to off-target proteins may be a reason for non-specific side effects in normal tissues. For this reason, web-server based algorithms for drug target identification were used. With this strategy, it could be estimated whether or not an identified drug candidate binds specifically to the corresponding target protein. The virtual drug screening procedure described herein was mainly based on rigid docking approaches, i.e. conformational changes during binding of a drug to its target protein were not considered. For this reason, flexible docking techniques were also considered to be included in this screening program (e.g. Molecular Dynamics simulations). In selected cases, the results obtained by this virtual screening process were experimentally verified. Using recombinant proteins, the binding of promising drug candidates was investigated by appropriate techniques such as microscale thermopheresis, surface plasmon resonance spectroscopy, isothermal calorimetry etc.

[0172] 6. Inspection of Scientific Literature Databases, whether the Top-Ranked Drugs have been Described to be Cytotoxic Towards Cancer Cells.

[0173] In many cases, drugs approved for diseases other than cancer have been described in the literature to exert also cytotoxic activity against tumor cells. These published data may serve as confirmation that the drugs identified by the present technical procedure may indeed be able to kill cancer cells. Common databases were screened such as PubMed, Scopus, SciFinder, Google Scholar etc. and other professional data mining tools and software supported the high-throughput screening of published literature.

[0174] 7. Decision Making of the Attending Physician which Drug can be Chosen to Treat Individual Tumors with Specific Gene Mutations.

[0175] The information obtained then served as a support for decision-making by physicians, tumor boards, other decision makers or automated devices like informational simulators of biological processes on the basis of the overall of clinical, laboratory and other information available on the individual patient as well as on criteria like availability, toxicity profile, side effects and drug interaction risk to serve as a generator of drug candidates for individual precision medicine in oncology and other fields.

[0176] Results

[0177] Biopsy material of a liver metastasis of a breast carcinoma has been obtained from a 50-year old patient. The patient had received various chemotherapies over more than a decade and showed extended metastases. The tumor was progressive and not responsive to the current chemotherapies anymore while tumor marker Ca15.3 rose to 22,230 units/ml. PDL Antibody therapy (Keytruda, 100 mg) did not change tumor markers in controls. The responsible tumor board recommended NAP-Paclitaxel with little hope that this would substantially change the course of the disease.

[0178] To gain further options of therapy, biopsy of liver metastasis was performed and test results were obtained as shown below. Irinotecan was identified as a candidate for treatment according to the test results and infused in two-weeks intervals according to the standard protocols. After that, CA15.3 went down to 1513 units and malignant ascites was markedly reduced. After half a year, the patient showed stable disease and clinical well-being. She took a two week vacation to France and feels well.

[0179] The complete transcriptome with >20,000 mRNA species was sequenced. RNA sequencing of the presented patient showed a total number of 47,562 mutations.

[0180] A database of 2483 proteins that are described in the literature as being cancer-related was prepared.

[0181] 611 RNA mutations in the presented patient led to amino acid changes in cancer-related proteins of this database.

[0182] From these 2483 proteins, 85 DNA repair proteins were excluded, because mutated DNA repair function cannot be pharmacologically regained.

[0183] From the remaining 2398 proteins, the 561 amino acid mutations were distributed among the proteins as follows:

[0184] 253 proteins with 1 amino acid mutation;

[0185] 69 proteins with 2 amino acid mutations;

[0186] 18 proteins with 3 amino acid mutations;

[0187] 19 proteins with 4 or more amino acid mutations.

[0188] Of the affected 359 proteins, 12 three-dimensional crystal structures were available. As more and more crystal structures of the human proteome were determined, the number of testable proteins increased over time. This means that the power of identification of effective repurposing drugs increased with increasing knowledge about the availability of three-dimensional protein structures.

[0189] The wild-type sequences of these 12 proteins were used, included the mutations and prepared three-dimensional homology models of these mutated proteins. Ten of the mutated proteins carries each one amino acid change. Two further proteins carried two amino acid mutations:

TABLE-US-00002 Mutated genes Type of mutation Amino acid exchange ABCA1 missense variant V825I ABCB1 missense variant S893A AKR1C3 missense variant E77G ANKRD27 missense variant P761G ANXA5 missense variant I81T APOBEC3B missense variant K146T ATP7B missense variant S406A und V1140A CASP8 missense variant K344H HLA-B missense variant H140S und I218V NQO1 missense variant P187S PARP1 missense variant V762A TLR1 missense variant N248S

[0190] All 12 homology models were subjected to virtual drug screening with >1500 FDA-approved drugs. This screening campaign resulted in 12 drug ranking lists. Each the top 10 drugs of all 12 drug ranking lists were inspected and searched for those drugs which appeared in more than one of these lists:

TABLE-US-00003 How often appearing in the 12 top 10 lists Remarks 9 Conivaptan Against low sodium levels 7 Venetoclax antineoplastic agent; BCL-2 inhibitor 6 Nilotinib tyrosine kinase inhibitor against chronic myeloid leukaemia 4 Vumon DNA topoisomerase type II inhibitor 4 Ponatinib tyrosine kinase inhibitor against chronic myeloid leukaemia and acute lymphoblastic leukemia 4 Eltrombopag against thrombocytopenia with chronic immune (idiopathic) thrombocytopenic purpura 3 Irinotecan DNA topoisomerase type 1 inhibitor 2 Olaparib PARP inhibitor, against breast cancer 2 Sirolimus mTOR inhibitor

[0191] As all of these drugs bind with high affinity to more than one mutated protein the identified drugs have a multi-specific target specificity. It can be expected that they were more active than mono-specific drugs that bind only to one single target. This approach was also applied by us for other conditions (cancers with one specific driver mutation, mutation-mediated inherited and somatic genetic diseases).

[0192] REFERENCES

[0193] Aronson J K. Old drugs—new uses. Br. J Clin. Pharmacol. 2007;64:563-565.

[0194] Ashburn T T, Thor K B. Drug repositioning: identifying and developing new uses for existing drugs. Nat. Rev. Drug Discov. 2004;3:673-683.

[0195] Dakshanamurthy S, Issa N T, Assefnia S, Seshasayee A, Peters O J, Madhavan S, Uren A, Brown M L, Byers S W. Predicting new indications for approved drugs using a proteochemometric method. J Med Chem. 2012;55(15):6832-6848.

[0196] Efferth T, Saeed M E M, Mirghani E, Alim A, Yassin Z, Saeed E, Khalid H E, Daak S. Integration of phytochemicals and phytotherapy into cancer precision medicine. Oncotarget 2017;8:50284-50304.

[0197] Hientz K, Mohr A, Bhakta-Guha D, Efferth T. The role of p53 in cancer drug resistance and targeted chemotherapy. Oncotarget 2017;8:8921-8946.

[0198] Kadioglu O, Saeed M, Kuete V, Greten H J, Efferth T. Oridonin Targets Multiple Drug-Resistant Tumor Cells as Determined by in Silico and in Vitro Analyses. Frontier in Pharmacology 2018;9:355. doi: 10.3389/fphar.2018.00355.

[0199] Kola I, Landis J. Can the pharmaceutical industry reduce attrition rates? Nat. Rev. Drug Discov. 2004;3:711-715.

[0200] Marin J J G, Lozano E, Herraez E, Asensio M, Di Giacomo S, Romero MR, Briz O, Serrano M A, Efferth T, Macias R I R. Chemoresistance and chemosensitization in cholangiocarcinoma. Biochim. Biophys. Acta 2018;1864:1444-1453.

[0201] Mbaveng A T, Kuete V, Efferth T. Potential of Central, Eastern and Western Africa Medicinal Plants for Cancer Therapy: Spotlight on Resistant Cells and Molecular Targets. Front. Pharmacol. 2017;8:343

[0202] Moehler T. Clinical experience with thalidomide and lenalidomide in multiple myeloma. Curr. Cancer Drug Targets 2012;12:372-390.

[0203] Mullard A. FDA drug approvals. Nat. Rev. Drug Discov. 2012; 11:91-94.

[0204] Schmidt F, Efferth T. Tumor Heterogeneity, Single-Cell Sequencing, and Drug Resistance. Pharmaceuticals (Basel) 2016;9.pii: E33.

[0205] Vargesson N. Thalidomide-induced teratogenesis: history and mechanisms. Birth Def. Res. Part C, Embryo Today: Rev. 2015;105:140-156.

[0206] Volm M, Efferth T. Prediction of Cancer Drug Resistance and Implications for Personalized Medicine. Front. Oncol. 2015;5:282.

[0207] Walther Z, Sklar J. Molecular tumor profiling for prediction of response to anticancer therapies. Cancer J. 2011;17:71-79.

[0208] WO 2003/057173.

[0209] WO 2001/035316. [0210] The present listing of claims replaces all prior listings or versions of claims in the present application.