WO2011144738A1

WO2011144738A1 - Critical gene targets for cytotoxic therapy

Info

Publication number: WO2011144738A1
Application number: PCT/EP2011/058259
Authority: WO
Inventors: Arno Lukas; Johannes Soellner; Andreas Bernthaler; Bernd Mayer; Raul Fechete; Paul Perco; Andreas Heinzel
Original assignee: Emergentec Biodevelopment Gmbh
Priority date: 2010-05-21
Filing date: 2011-05-20
Publication date: 2011-11-24

Abstract

The invention relates to a method of identifying critical gene targets by means of their identity as functional synthetic lethal partners to downmodulated or functionally compromised genes A to predict the response to a targeted cytotoxic therapy in a patient, comprising: a) providing an array of genomic sequences or expression products in synthetic lethal relationship, b) providing - data of downmodulated genes A in a patient's sample, and - data of expressed genes B in said patient's sample in synthetic-lethal relationship to at least one of said genes A, as identified using said array, wherein the suppression of genes B is otherwise not lethal in a normal cell, to obtain an individualized data set, c) repeating step b) to obtain a repertoire of individualized data sets relevant for a specific patient population, d) applying said repertoire to an algorithm, wherein the result is the identification of genes B' in said patient population which are in a synthetic multi-lethal relationship to at least two of said genes A ande) producing an analytic carrier from said genes B'.

Description

Critical gene targets for cytotoxic therapy

The present invention relates to a method of identifying critical gene targets to predict the response to a targeted cytotoxic therapy in a patient and an analytical carrier comprising such critical gene targets.

Cancer is among the leading causes of mortality in the industrialized parts of the world. Some 2.9 million cases are diagnosed annually in Europe, with 1 .7 million deaths attributed to this class of diseases. The efficacy of anti-cancer therapies is still limited, in a first place due to heterogeneity of cancer limiting efficacy of drugs per se, and second if a drug shows initial efficacy by the emergence of drug resistance.

Improvements in experimental molecular biology in the realm of the "Omics" revolution have significantly contributed to the understanding of molecular processes involved in cancer (Lowe and Lin, Carcinogenesis 2000, 21 :485-95). The term Omics summarizes a broad spectrum of techniques which measure cell-wide activation of genes, respective proteins, metabolites, etc. Omics allows the quantitative assessment of many thousand cellular components in parallel, and hence provides a system-wide landscape of cellular components being specific for a tumor cell. These developments have triggered the emergence of a new discipline, Systems Biology (Call et al., Lancet Oncol 2008, 9: 1002-1 1 ), aimed at integrating all available data on cellular features for deriving a descriptive (or even quantitative) understanding of cellular events. A prototypic representation of multi-level information integration is interaction networks.

From all these advancements the available data basis for describing molecular processes specific for cancer cells has enormously increased. However, what unfortunately became clear is the significant molecular variance of cancer when comparing different organs affected, but also when analyzing a particular type of cancer, or even speculating that each cancer is to some degree patient specific.

Apparently, cancer cells have a multitude of strategies for escaping natural clearance mechanisms. Aberrant cells should inherently follow apoptosis, but take routes to halt this process (Cho W (Ed.), An Omics Perspective on Cancer Research. 2010,

Springer, NY.; Khalil and Hill, Curr Opin Oncol 2005, 17: 44-8); cancer cells are at least to some extent identified as such by the patient's immune system, but exhibit immunomodulatory capabilities and other methods for escape (Rapberger et al., BMC Syst Biol 2008, 2:2.); tumors modulate their local environment e.g. triggering nutrition supply via angiogenesis (Michaelis et al., Mol Cancer 2009, 8:80); cancer cells when attacked with e.g. chemotherapeutics develop resistance mechanisms (Michaelis et al., Int J Oncol 2006, 28: 439-446). Obviously, cancer has to be seen as a highly complex system in a dynamic interplay of cell internal survival routes under the selective pressure of cell internal and external death pathways, under the overall control of the body's immune system. Complex systems in a broad definition involve a large number of elements whose interaction and respective dynamics generates unexpected (emergent) properties. Cancer is a prominent representative of this class: A well-balanced machinery of cell survival and cell death looses corrective measures, resulting in growth of tumor mass and eventually the development of resistance to therapy. Major advancements have been made in managing this disease, and an impressive body of knowledge on key molecular processes characteristic for cancer has been assembled so far. But still, putting the pieces together for reaching an understanding of cancer, being the prerequisite for rational and targeted therapy, remains a major challenge.

Malignant neoplasms, also commonly called cancers, form a group of diseases characterised by uncontrolled growth of cells in higher organisms. Division of cells, invasion of adjacent tissues and metastasis are thereby the three prime indicators for prognosis of disease development.

While many cell types can give rise to cancerous growth, malignant neoplasms can be crudely classified in a number of ways. Based on phenotype, distinction in solid neoplasms (tumours) and non-solid cancers such as leukemias is possible. Other than that, the source tissue or cell-type giving rise to carcinogenesis or the organ where growth occurs in (in case of metastasis) is used to classify cancers.

Carcinoma: Malignant tumors derived from epithelial cells. This group

represents the most common cancers, including the common forms of breast, prostate, lung and colon cancer.

Sarcoma: Malignant tumors derived from connective tissue, or mesenchymal cells.

Lymphoma and leukemia: Malignancies derived from hematopoietic (blood- forming) cells.

Germ cell tumor: Tumors derived from totipotent cells. In adults most often found in the testicle and ovary.

Blastic tumor or blastoma: A tumor (usually malignant) which resembles an immature or embryonic tissue. These tumors are mostly seen in children.

Aside of cancers also other malignancies suffer from declining efficacy of therapies through development of biological resistance. Among those infectious diseases are paramount as they contribute significantly to mortality and particularly morbidity rates, particularly in developing or non-industrialised parts of the world. In this regard, Malaria, Tuberculosis and HIV are also known as the three big killers.

There is a great need for robust cytotoxic therapies avoiding or reducing development of resistances.

Cytotoxic therapy usually is directed against a specific target. Among the targeted immunotherapies there is for instance, passive immunotherapy, e.g. using monoclonal antibodies, or active immunotherapy employing vaccines. Effectivity or applicability of any particular therapy can vary greatly between types of cancer or based on presentation in individual patients. As a consequence differentiation between patients may be beneficial in many cases; however molecular biomarkers to make adequate decisions are often not available yet. One of the exceptions is the Her2/neu monoclonal antibody therapy (marketed as Herceptin and Trastuzumab) which can greatly benefit from pre-emptive diagnostic tests for the expression status of the HER- 2 protein as the factor is found in only approximately 25-30% of breast cancer patients at sufficient levels. Also, within a particular type of therapy differentiation by cancer type is usually necessary. For example, the choice of chemotherapeutic agent can be influenced by gender of the patient, principal effectivity on certain cancerous cells and, related to that, rate of degradation of drugs in cancerous tissues (Takemura et al., Gan To Kagaku Ryoho. 2004 Jul;31 (7):1053-6.). These factors may vary between cancer types as well as between patients afflicted with the same class of cancer.

Targeted chemotherapy often relies on cytostatic (cell proliferation inhibiting) chemicals which are toxic or detrimental for dividing or metabolically highly active cells and may therefore exhibit severe side-effects on tissues exhibiting naturally high proliferation rates. For drug examples, plant alcaloids, such as the Vinca alkaloids Vincristine and Vinblastine, inhibit microtubule function and therefore prevent chromosomal seggregation leading to mitotic arrest. These drugs are effective or otherwise indicated only in a limited number of neoplastic disorders, however, specifically neuroblastoma, lymphomas and leukemias in the case of Vincristine. Similarly, immunotherapies such as the activation of innate and adaptive immune responses by interferon therapy are applicable only against a limited spectrum of neoplastic disorders, always additionally including a potential patient bias. Many more

chemotherapeutics are in use and implicated for certain cancers in individual or combination therapies.

In contrast to classical chemotherapy, rational drug design and screening procedures have enabled certain targeted therapies by influencing activity of factors of specific relevance in cancers, sometimes allowing improved efficacy and side effect profiles as in the case of Imatinib in chronic myelogenous leukemia and a number of other cancers. Targeted therapies are thereby defined by a specific, defined molecular target commonly found in a class of cancer which is amenable to drugs. In contrast to standard chemotherapies it is possible to focus on molecular changes elementary to cancerogenesis or progression which is the basis for the mentioned improvement in effect / side effect profiles (whereas e.g. inference with microtubule formation by vincristine affects most cells of the body). While these developments are cause to hope, they suffer from some of the same problems as standard chemotherapy, namely the development of resistance against the drug stemming from mutations/alterations after onset of therapy or alternatively from genetic/epigenetic heterogeneity within treated tumours to begin with. Alternatively, only a certain subset of cancers may initially feature the drug target in adequate concentrations, as described for Herceptin before. Heterogeneity within cancerous populations of one individual, including cancer stem-cells, and between patients is the primary biological basis for reduced drug efficacy, followed by drug resistance. To add robustness (avoidance of disease recurrence) to therapy therefore has to be a central goal in drug screening.

A major shortcoming of cancer therapies and certain other areas of therapy including but not limited to infectious diseases and potentially disorders related to somatic mutations is the development of resistance to those therapies. In the case of cancer and infectious diseases such resistance is mediated by development of pathogen or cancer cell populations generally resulting in diseased cells, tissues or individuals not or less affected by therapy when compared to the original population of cells, tissues or patients. This variability and potential to experience deterioration of therapy efficacy seemingly inherent to many drugged biological systems can be seen both between individuals (for example when comparing between neoplastic

tissues/growths/isolates stemming from at least two individuals) as well as within individuals (for example genetically/epigenetically/phenotypically different cell populations within one neoplastic tissue/growth/isolate). Therapies are therefore only of limited value if they cannot avoid recurrence of a disease in a form resistant to the original therapy or provide only limited efficacy in a substantial fraction of patients. Unfortunately this is a common situation when treating infectious diseases and cancer and potentially other diseases.

The concepts of synthetic lethality and synthetic sickness refer to the lethal or sickening effect of elimitation of function of two specific genes or molecules encoded by these genes (for example by genetic knock-outs, natural or induced genetic or epigenetic mutation or alteration, small interfering RNAs or interfering drug effects) on a single or multicellular organism, if elimination of such function of either of the two genes is not lethal (Conde-Pueyo et al., BMC Syst Biol. 2009 Dec 16;3:1 16).

The concepts of synthetic lethality and synthetic sickness have been described before in detail, particularly in the context of yeasts as experimental screening options are more straightforward there when compared to, for instance, mammalian cell lines (US2004/0121324A1 ).

Recently, comprehensive high-throughput data of synthetically lethal genetic double knock-outs have been published for the yeast Saccharomyces cerevisiae (Koh JL et al., Nucleic Acids Res. 2010 Jan;38(Database issue):D502-7. Epub 2009 Oct 30.).

Igelhart J.D. et al. (New England J. Med.2009, 361 , 2, 189-191 ) describe the synthetic lethal relationship between BRCA1 and BRCA2 with PARP1 .

The described approach describes utilization of a synthetic lethal relationship for a targeted cancer therapy but it is not patient specific and not built upon patent specific data of expressed genes.

Inoue A. et al. (J. Mol. Endocrinol., 2002, 29, 2, 175-192) disclose the analysis of cancer samples of individual patients based on estrogen-responsive expression profiling. Specifically, they build a custom cDNA array equipped with probes (cDNAs) for

138/204 genes which had previously been shown to be differentially regulated upon estrogen treatment plus 27 for internal control.

Conde-Pueyo N et al. (BMC Systems Biology, 2009, 3:116, 1 -15) describe human synthetic lethal inference as potential anti-cancer target gene detection.

A review of a microarry based method for detection of synthetic lethal relationships among genes, applied to yeast is described by Tucker et al. (Nat Genet. 2003 Nov;

35(3):204-5).

Baffa et al. (J Pathol. 2009 Oct;219(2):214-21 ) disclose an application of a custom microRNA chip for detection of microRNAs specifically relevant for metastasis in a number of tumours.

Briefly and more formally, a gene or gene product associated function A, also termed primary gene or gene A is termed synthetically lethal with respect to a gene or gene product associated function B, also termed secondary gene or gene B, if removal or inhibition of both functionalities is associated with a lethal effect, i.e. cell death or substantial decrease in viability, while inhibition of either one single functionality does not significantly impact cell viability.

It is the objective of the present invention to provide the means to identify drug targets and biomarkers for diagnosis and therapy of diseases, specifically where cytotoxic therapies are characterized by a reduced likelihood or later onset of resistance to therapy through emergence of drug resistant phenotypes and an improved efficacy in terms of broader patient coverage.

The objective is solved by the claimed subject matter.

Thus, the present invention relates to a method of identifying critical gene targets, specifically by means of their identity, as functional synthetic lethal partners to downmodulated or functionally compromised genes A to predict the response to a targeted cytotoxic therapy in a patient, comprising:

a) providing an array of genomic sequences or expression products in synthetic lethal relationship,

b) providing

- data of downmodulated genes A in a patient's sample, and

- data of expressed genes B in said patient's sample in synthetic-lethal relationship to at least one of said genes A, as identified using said array, wherein the suppression of genes B is otherwise not lethal in a normal cell, to obtain an individualized data set,

c) repeating step b) to obtain a repertoire of individualized data sets relevant for a specific patient population,

d) applying said repertoire to an algorithm, wherein the result is the identification of genes B^' in said patient population which are in a synthetic multi-lethal relationship to at least two of said genes A and

e) producing an analytic carrier from said genes B^'.More precisely in the context of the current invention, the functionality of products or expression level of a gene A is reduced below a level where a knockout of the associated gene B can be expected to activate the synthetic lethal effect. As such a gene A is defined as the partner of a synthetic lethal pair which is depleted or otherwise functionally compromised while functionalities encoded in the gene B are present. As only present functionalities such as proteins can be drug targets, gene B products provide the targets for cytotoxic therapies. Of note, the presence of a gene B implies the existence of a gene A, otherwise there is no synthetic lethal relationship. The terms "gene A" and "gene B" inherently define the role of the respective gene within the synthetic lethal relationship.

A subset of identified genes B are genes B', which are defined as those genes B with relationship to more than one potential gene A. This is the nature of synthetic multi-lethality in comparison to classical synthetic lethality. It is important to stress the potential association, as a particular gene B (the drug target) may be associated with two or more potential genes A (Ai, A₂ A_n) by a previously assigned synthetic lethal interaction. In a particular patient sample any number (between none, specifically one and all) of the potential genes A will be realized as gene A and therefore their functionality will be missing, while said gene B will always be the drug target for initiating the lethal effect as long as at least one of the potential genes A is realized, i.e.functionally missing.. Which of the potential Ai .... A_n are realized in a particular patient sample as gene A depends on the sample. However having at least two genes which can take the role of a gene A for a particular gene B is one of the very unique and most crucial aspects of this invention, as this provides robustness to the cytotoxic effect. Classic synthetic lethality considers only single synthetic lethal pairs and possibly how well a particular pair, i.e. by presence of B and lack of A, is conserved across diseased tissues such as cancers.

It is important that knowledge on the precise biological context, for example such as involved pathways), of a gene is not necessary to make use of synthetic lethal interactions, which intrinsically define the criticality of the genes. Such information can be useful to better understand the biological basis for the observed effect, but the synthetic lethality defines the criticality for the cytotoxic effect. Moreover, members of synthetic lethal pairs may stem from disjunct biological pathways. The algorithm for analysing said repertoire solves four distinct aspects. First, expression of genes or the presence of mutants have to be determined. This has been solved multiply and is well established based on a number of input data types.

Tusher et al. (Proc Natl Acad Sci U S A. 2001 Apr 24;98(9):51 16-21 ) describe significance analysis of differentially expressed genes including correction for False Discovery Rates.

Quackenbush (N Engl J Med. 2006 Jun 8;354(23):2463-72) reviews the use of DNA microarrays for tumor classification based on expression profiles.

Commonly used methods for microarray analysis are reviewed by Alison et al. (Nat Rev Genet. 2006 Jan;7(1 ):55-65) while Muhlberger et al. summarize current analysis methodology across several Omics tracks utilizing open-source tools (Methods Mol Biol. 201 1 ;719:379-97).

Goh et al. review approaches and potential pitfalls of analysing matched normal/tumor pairs (PLoS One. 201 1 Mar 18;6(3):e17810).

Based on these analyses combined with a dataset of synthetic lethal interactions pairs of genes among known synthetic lethal relationship to obtain genes B and associated genes A can be determined. In a further step synthetic multi-lethal combinations can be derived from gene A - B pairs by selecting those genes B^* which are associated with multiple genes A. The optimal degree of multiplicity, i.e. how many genes A should at least and at most be associated with a gene B, is thereby

aparameter which can be adjusted according to the specific needs. At this level data analysis is still operating on individual samples.

In a further step only those gene A - B combinations are retained which are specifically found in diseased samples but not in control samples. That is implicitly the case for technologies such as DNA microarrays where a comparison of healthy versus diseased is used to determine differential regulation of genes. For full genome sequencing and comparable technologies optimally diseased and normal tissue are sequenced and differences are compared, commonly in the context of annotated genes. Alternatively a general background can be assumed, but that is inferior to working on paired data obtained from the same patient/donor. This third step is important to make sure that synthetic lethal effects do not take place in multiple tissues rather than occuring in diseased cells/tissues.

In a further step, the algorithm data between individual samples are compared where the essential element is to identify genes B^* which are expressed and functionally active in as many samples as possible to achieve maximum sample coverage. The essential element is that while the gene B or gene B^* has to be constant, genes A can vary for each sample and can be entirely disjunct.

In all samples the synthetic lethal effect may be initiated by perturbing the function of gene B. ln a preferred embodiment the analytic carrier is produced by integrating said genes B^' together with said genes A being in a synthetic multi-lethal relationship to genes B^'.

In an alternative embodiment the analytic carrier is produced by further integrating said genes B together with said genes A in a synthetic lethal relationship to genes B.lt is preferred that the method further comprises the validation of said analytic carrier, wherein a drug targeting one of the genes B^' has been proven to be effective in a patient of said patient population.

Specifically, those genes B^' are identified, which are higher than a threshold, optionally afflicted with a ranking, reflecting the degree of multi-lethality and/or coverage of said patient population, such that the most critical genes B^'are identified for the specific patient population.

According to a preferred example, said array is based on genomic sequence data obtained by orthology mapping of genomic sequences from an eukaryotic species, which could be human or other than human. Further preferred is the use of an array of eukaryotic genomic sequences or respective expression products, which is derived from eukaryotic species, such as vertebrates, including mammalian specifically human; fungi, including yeast; and insect.

Preferably said genes B or B^'are of human origin.

In a preferred embodiment the individualised data set comprises data from samples of healthy and diseased cells of a patient. This would specifically provide for comparative analysis.

The analytical carrier is preferably further used for determining a patient profile, such as a pattern of genetic sequences or their expression products specific for said patient, e.g. for determining the appropriate cytotoxic therapy for said patient or the stratification of a patient population in clinical therapy, e.g. to increase the likelihood of drug response in said patient population.

The analytical carrier according to the invention essentially consists of critical target genes B^', such as determined according to the above described method, and optionally the downmodulated genes A in a synthetic multi-lethal relationship to each gene target B^', optionally together with genes A in a synthetic lethal relationship to genes B, as a library that displays said genes as nucleotide sequences and/or expression products encoded by the sequences.

The carrier according to the invention specifically comprises genes or expression products, which are differentially tagged or located in spatial distinct compartments.

Specifically the carrier according to the invention is a microcarrier in a storage- stable form. The carrier is preferably dedicated or relevant to a patient population suffering from a disease selected from the group consisting of neoplastic disorders and infectious diseases and other disorders where drug based therapy is afflicted with cytotoxic effects, in particular cancer, including colon, lung, breast and prostate cancer. Among infectious diseases only those can be targeted by the invention where cytotoxic effect on infected patient cells can be expected to be beneficial. Classically anti- infectious therapies target microorganisms, for example bacterial infections by antibiotics, affecting bacterial metabolic processes directly. Drugs developed according to the presented invention act by exerting a cytotoxic effect upon infected host cells. The present invention is therefore specifically applied to intracellular pathogens where killing of infected host cells is the aim, more specifically to chronically infecting pathogens, even more specifically to the causative agents of Tuberculosis, Malaria and Human Immunodeficiency Syndrome. Other infectious agents of explicit interest include the causative agents of Lyme Borreliosis, Toxoplasmosis and Mycoplasma infections. The listed diseases have in common that they are subject to development of resistance, have proven very resilient to immunological therapies such as vaccines and perturb the cellular machinery of infected cells. This perturbation can be utilized by the method of the present invention to eliminate specifically host/patient cells infected by the pathogen in question, for example infected macrophages in the case of

Tuberculosis/Mycobacterium tuberculosis, infected hepatocytes in the case of

Malaria/Plasmodium sp., infected PBMCs (Peripheral Blood Mononulcear Cells) and specifically Dendritic cells, Macrophages and T-helper cells in the case of HIV/AIDS.

In accordance with the present invention a method of producing a drug for targeted cytotoxic therapy comprises the following steps:

a) providing an analytical carrier as described above,

b) identifying a critical target gene B^' for a specific patient population,

c) obtaining a drug by drug discovery or design, which drug targets said gene B^', said functionality or expression products, and

d) manufacturing a pharmaceutically acceptable formulation of said drug.

According to another aspect of the present invention a method of identifying a patient eligible to targeted cytotoxic therapy comprises the following steps:

a) analyzing a patient sample to provide analytical results for one or more genes,

b) providing an analytic carrier as described above, and

c) matching said analytical results to predict the patient's response to said therapy.

According to a further aspect of the present invention a method of identifying a patient specific cytotoxic therapy, comprising a) analyzing a patient sample to provide analytical results for one or more genes,

b) providing an analytic carrier as described above, and

c) matching said analytical results to identify a gene B^' target and/ or targeting drug specifically relevant for said patient.

Specifically the cytotoxic therapy is provided for the therapy of infectious disease, specifically infections with intracellular pathogens, or neoplastic disease, in particular solid tumor disease.

The term "gene A" as used herein shall refer to genes or genetic sequences that are downmodulated in diseased cells when compared to normal cells. Gene A is in a synthetic-lethal relationship with a gene B. Both, gene A and B, together are referred to as synthetic-lethal gene pair. Typically the presence of the gene A or the level or function of gene A expression is reduced by 3 fold, preferably more than 3 fold. The term shall also refer to a series of genes A, called, Ai , A₂, A₃, ... A_n, A,, each of them being different genes A, being in synthetic lethal relationship with the same gene B.

The term "gene B" as used according to the invention refers to critical genes or genetic sequences, which provide a target or hub of therapy to kill a diseased cell. While the level or function of gene B typically is in the normal range in both, normal or healthy and diseased cells, it becomes a critical target of action when being attacked by a respective cell-killing therapy, if one or more of genes A are downmodulated in a diseased cell. Thus, the high selectivity of an anti-cellular therapy targeting gene B, to target diseased cells only is imminent through the relationship of the synthetic lethal gene pair, thereby reducing possible side-effects of such therapy.

The term "critical" with respect to gene B in the context of the present invention refers to the relevance of a gene B as a gene target that can be successfully

addressed in cytotoxic therapy to treat diseased cells with a high probability.

Depending on the algorithm used the most critical genes B may e.g. be selected according to the coverage of human samples or individual data sets considered. A preferred coverage for a critical gene B target is at least 30%, preferably at least 40%, preferably at least 50%, preferably at least 60%, preferably at least 70%, preferably at least 80%, even at least 90% of a repertoire of individualized data sets, which preferably comprise at least 10 different samples, preferably at least 30, even preferably at least 60 different samples.

According to a different criterion the level of criticalty is determined by the degree of synthetic multi-lethal relationship, meaning the number of different genes A found downmodulated and in a synthetic multi-lethal relationship to the same gene B^'. The higher the number of such different genes A, the more useful would be the respective gene B^' as a cell killing target. In the determination of such criticalty preferred multi-lethal degrees would be at least two genes A forming a triple with gene B, more preferably at least 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25 different genes A. It is, however, preferred that those genes B^' are excluded which have ubiquitary lethal- relationships with more than 500 different genes A. Typically a threshold is determined to select only the most critical genes B^'. Above the threshold value a ranking within the most critical genes B^' is usually established.

A suitable criterium for determining the criticalty of genes B virtually is the ranking according to the coverage of effected cases where high coverage indicates criticalty. A further virtual second line criterium is the number of identified down- modulated genes A per case where high numbers indicate criticalty.

A specific test for determining the criticalty and/or validating a gene B employs the comparisons of samples of the same species or individual, which have (i) absent, downmodulated or otherwise non-functional gene(s) A and gene B, and (ii) functional gene(s) A in the normal range and absent, downmodulated or otherwise non-functional gene B. Viability of the cells is tested. Concept is that down-regulation of gene B only has no or significantly less effect on cell viability, whereas joint down-regulation of gene(s) A and B results in synthetic lethality for the cell population.

As a preferred result at least triples of genes on the output side are obtained, i.e. one B and at least two A. For further enrichment duples, triples, quadruples and generally multiples may be used. This can assure that the next best screening technique or preprocessing to convert duples into triples does not circumvent the algorithm.

The term "analytical carrier" as used according to the invention refers to objects of technical nature comprising gene information on a support matrix allowing for the analysis of individual genes or gene groups, their functionality and/or expression products (expression analysis) for the purpose of providing targets relevant to specific patient populations or individual patients to develop specific therapies. Analytical carriers are typically used in the form of libraries comprising a repertoire of genetic sequences, in the literal form or as nucleotide sequences that are translated into products of expression or display. DNA analytical carriers can be used to detect DNA (as in comparative genomic hybridization), or detect RNA (most commonly as cDNA after reverse transcription) that may or may not be translated into proteins.

Analytical carriers are primarily used as a tool to provide researchers with the capability to perform genetic tests needed to extract medical information from

advances in genomics and proteomics, since many genetic tests may be accomplished in parallel. This is particularly useful for the drug discovery process or to determine which genes exist in a sample, e.g. by detecting specific pieces of mRNA. Specific applications would employ the full human genomic information, e.g. from individuals, on an analytical carrier, to determine relevant gene targets using a specific algorithm. ln many cases, DNA or oligonucleotide microcarriers are fabricated, e.g. on a solid carrier. In standard microarrays, the genetic sequences are attached via surface engineering as a collection of orderly microscopic "spots", called features, each with a specific probe attached to a solid surface, such as glass, plastic or silicon biochip (commonly known gene chip, genome chip, DNA chip or gene array). Thousands of them can be placed in known locations on a single DNA microarray. The alternative bead array is a collection of microscopic polystyrene beads, each with a specific probe and a specific tag.

Examples of analytical carriers are DNA microarrays are sold as GeneChip™ (Affymetrix, CA, USA) and BeadChip™ (lllumina, CA, USA), Roche NimbleGen DNA microarray products (Roche, Switzerland), DNA microarrays sold as TagMan™ Array (Applied Biosystems, CA, USA), or Agilent Human Gene Expression Microarrays (Agilent, CA, USA).

Preferred analytical carriers include those targets among the most critical target genes B, which have been validated by respective drugs in the clinics. Besides, a series of new potentially relevant targets are revealed by the method according to the invention, which provides for an analytic carrier highly relevant for selecting therapies for individual patients or patient populations or indications. The analytical carrier, which is provided for preparing a respective profile from an individual, is specifically useful for selecting patients amenable for specific targeting therapies, in particular for clinical trials investigating the responsiveness to a certain drug (stratification). The

personalized profile may also be prepared for diagnostic purposes to compare with a reference gene profile and to determine similarities to the patient sample to identify the stage or sub-type of disease. Tissue samples from primary diseased cells, e.g. primary tumor cells, or from distant diseased cells, e.g. metastases, may serve to prepare a highly differentiated profile. Monitoring during a certain therapy on a patient-by-patient basis is feasible when a series of gene profiles is prepared in the course of therapy. Samples from patients, typically mammalians including human patients and probands, usually are derived from healthy or diseased tissue, or from body fluids, e.g. blood or urine. Preferably samples from human patients are taken according to the invention.

The gene targets B displayed on lists or other carriers, optionally together with the respective genes A, can be provided as a tool, e.g. to design drugs targeting said genes, respective expression products or functionalities. A specific utilization is related to the provision of candidates for drug development, specifically to provide a target in screening assays, functional or binding assays, to provide ligands specifically interacting with one of the most critical genes B.

A focussed analytical carrier according to the invention would preferably be enriched in the most critical target genes B, optionally together with or paired with the respective genes A. In a preferred embodiment the analytical carrier is a tool bearing a collection of genes essentially consisting of the genes B, optionally together with the respective genes A^'and further optionally genes B and genes A in a synthetic lethal relationship to genes B. The term "essentially consisting of as used herein refers to a percentage of genes or genetic sequences present at the carrier site, which is at least 50%, preferably at least 60%, preferably at least 70%, preferably at least 80%, preferably at least 90%, up to 100%. In specific cases it is preferred that all of the genes B located on the carrier have been identified according to the invention and thus are considered the most critical genes B. The eventual presence of respectively paired genes A is not taken into consideration when determining the share of genes B within a carrier comprising a gene array.

Thus, the present invention is based on coverage of gene pairs of synthetic lethal relationship in individual patient samples, encompassing personalized medicine approaches. Building on the concept of synthetic lethality the paradigm is expanded to synthetic multi-lethality to identify drug target candidates characterized and ranked by relevance in a majority fraction of patients.

With the tools according to the invention cytotoxic therapies at risk of resistance are particularly addressed, including acquired chemoresistance in cancer therapy. This phenomenon resembles the loss of efficacy of a given drug in the course of treatment, clearly representing a major obstacle for successful cancer therapy.

By the present invention the loop is closed starting from Complex Systems Science methods involving general characteristics of stability and robustness and finally to therapy target identification and experimental verification. A computational workflow is proposed for identification of novel targets, e.g. for cancer therapy, by combining experimental base data, subsequent Complex Systems analysis followed by selection of key features potentially resembling an Achilles heel of cancer cells.

For the purpose of the present invention, the function or activity of a gene in context of the present invention refers to the gene or functional activity of products encoded by the gene including RNA molecules and proteins including chemical modifications of these, such as post-transcriptionally modified RNAs and post- translationally modified proteins.

For the purpose of the invention synthetic lethality and synthetic sickness are used equivalently, particularly because distinctions can be difficult and subjective to make and also because concepts described here pertain to both.

By the present invention the concept of synthetic lethality is non-trivially expanded by considering not only duples (pairs including one gene A) of genes and their products but also triples (one gene B and two genes A), quadruples (one gene B and three genes A) and generally multiple gene functionalities, the removal of which results in death or substantial morbidity of a specific biological sample, cells, cell-lines or entire multicellular organisms. This expansion is herein called "synthetic multi- lethality".

Thereby a surprisingly simple ranking of critical gene targets was obtained, among them validated targets and most interestingly new targets of cytotoxic therapy. The presence of well-known targets in the panel of the most critical gene targets, also called hubs, validated the approach. Thereby the relevance of the new targets relevant for the patient population or indication is considered to be validated as well. New targets are understood as known genes having hitherto unknown functionality and relevance in the specific patient population or disease indication. For example, well- known targets of specific cancer therapy are also considered new and can be repositioned, if they turn out to be among the most critical gene targets of a patient population or indication, which relationship was previously unknown before.

While synthetic lethality was usually used to describe the behaviour of single cell-lines or entire species without resolution to individuals, the present invention focusses on the behaviour of single cell populations, tissues or individuals while expansion to entire species or groups of species is possible. This expanded concept of synthetic multi-lethality describes situations where removal of gene and/or gene product functionality of a specific duple, triple, quadruple or any larger individualised set S of genes leads to death or substantial morbidity in an individual single or multicellular organism or biological sample at least under certain external conditions. This means removal of gene functionality of any sub-set of genes including gene B and at least one gene A taken from such a set S has a comparable effect in this organism or biological sample. As naturally all genes are likely part of synthetic multi-lethal gene- sets (at least removal of function of all genes in a genome can be expected to be lethal to an organism in question) the present invention specifically refers to the concept of "minimal sets of synthetic multi-lethality". This means that the lethal effect must be attributable to varying combinations of determined genes selected from a genome rather than individual genes or any possible subsets of genes.

To formalise this concept, an individual data set referring to an organism or biological sample presents a set of genes Ai, A₂, A₃, .A_n, which are down-modulated or functionally compromised and are part of a set S of genes together with a functionally non-compromised gene B, where compromising the function of B should result in a lethal effect or should lead to substantial morbidity due to pairwise given synthetic lethal relationships to genes Ai, A₂, A₃, .A_n.

The synthetic multi-lethality data are successfully utilized for the selection of potential drug targets or biomarkers characterised by broad coverage of individual data sets, which comprise gene pairs in synthetic lethal relationship. This enables the robust and successful application of a certain drug therapy or biomarker to a major fraction of cell populations or individuals of an organisms characterized by a specific disease or disease state. Ranking of the most critical gene targets B may be explicitly determined by the broad coverage of diseased individuals or organisms, tissues or biological samples, as well as the reduction of the likelihood to develop resistance against a particular therapy, including but not limited to cases of disease recurrence, reflected by the degree of synthetic multi-lethality.

Classical synthetic lethality or synthetic sickness screens have previously been used to identify potential drug candidates specific to certain diseased cells. These cells were considered amenable to drugs if a synthetically lethal partner A of a direct or indirect drug target B is functionally inactivated in those cells, for example by down- regulation, mutation, deletion or other mechanisms, however, present or otherwise functionally available in healthy/uninfected cells. Drug target candidates identified in such a way are not less likely to be subject to the development of resistance than drug target candidates identified by other means, for example high-throughput small- molecule screens. Multiple mono-lethal relationships to different genes A might be effective in a patient sample targeting a selected gene B. Therefore it is rather likely to preserve sensitivity to the drug even if single mono-lethal relationships to gene partner A, are bypassed by resistance mechanisms in a patient. In the same sense the efficacy can be increased in terms of patient coverage. Variabilitiy of down-modulated genes A in different patient samples is taken into consideration by the approach.

Based on the concept of synthetic multi-lethality a workflow has been developed to select drug targets characterized by increased probability of therapy efficacy across individuals and reduced risk of development of resistance within individuals as well as reduced risk of disease recurrence in the case at least of, but not limited to, neoplastic disorders, including solid tumor disease, and infectious disease.

Specifically in a first step down-regulated, mutated or otherwise functionally compromised genes A or gene products are identified. Such identification can be conducted by several means including, but not limited to, transcriptomics experiments indicating significantly down-regulated genes, nucleotide sequencing indicating mutated genes or a multitude of other experimental approaches leading to the identification of gene function partially or completely missing or otherwise

compromised in particular diseased individuals. Specifically this step is performed sequentially and repeatedly for multiple individuals or biological samples, for example by analyzing transcriptomics data from multiple individuals or cell-lines associated with or suffering from a particular disease or disease condition. Synthetic multi-lethality genes Ai, A_n can be envisioned where at least one B^' common to all these Ai, A_n has to be present in as many individuals/biological samples as possible. In each iteration new combinations of genes Ai ..A_n, (primary, functionally compromised) and B^' (secondary, functionally non-compromised gene) can be identified. Based on this iterated sampling technique genes B^', e.g.gene products not functionally inactivated or down-regulated, can be identified which are associated by synthetic multi-lethality with one or several different genes A (gene products functionally inactivated or down- regulated,). More precisely, a specific gene taken from set Ai ... A_n, which is associated with a specific gene B may or may not be down-regulated in a specific patient, tissue or otherwise sample of organic matter associated with a particular disease or disease state as long as it is down-regulated or functionally inactivated in at least one of these samples. Elements from Ai ..A_n may but need not be shared between individual samples or individuals while B must be present in at least the substantial majority of analysed samples. A preferred aim is to identify sets of multiple genes, where those sets contain at least three elements (a gene triple) where at least one is a potential drug target or biomarker B^' functionally present in as many analyzed individuals/tissue samples as feasibly possible. This iterative aspect for enriching genes B^' optimally generally available in diseased organisms/cells associated with a set of genes Ai, A_n of size of at least two highlights the potential to identify drug targets or biomarkers, which will be characterized by enhanced patient coverage or therapy stability where therapy stability, is defined as the reduced likelihood or delayed onset of resistance against therapy compared to standard therapies.

The described approach is mainly geared towards identification of drug targets and ultimately drug therapies or targets of diagnostic methods applicable in a multitude of patients with improved therapy stability and reduced likelihood of disease recurrence after end of drug therapy. The same approach is well suited as a basis for

individualised medicine as well. Individual patients can be profiled for the presence of genes B^' in a synthetic multi-lethal relationship with sets of downmodulated genes A. While such combinations may be infrequent in most patients, certain individuals, particularly but not exclusively in neoplastic disorders can show particular B^' and Ai,

A_n combinations which make this particular patient suitable for a particular therapy based on drug targets encoded by B^'. While this approach is related to testing for the presence of Her2/neu for the targeted therapy with herceptin our approach adds stability to therapy for a particular patient, which would otherwise not be present or subject to chance or luck. Other patients not featuring as many missing functionalities in potential genes A should be more likely to experience remissions or inefficacy of therapy. On the other hand they may exhibit other B^' and Ai, A_n combinations allowing selection of other accepted therapies for efficient therapy based on molecular profiling of available functions of potential genes B^' and missing functions of potential genes A.

The so identified hubs can be addressed in specific therapeutic settings. Drug targets can be drugged directly and indirectly, the essential part is that function is removed or altered. Exemplary hubs can be directly drugged using entries as provided in DrugBank. Independent whether broad coverage of individuals and patient populations or optimized solutions for individuals is the aim, drug targets are of relevance only when they can be drugged. More generally, therapies (independent of their nature) need to lead to a reduction of available gene B^' encoded functionality for synthetic multi- lethality to take effect. According to a preferred embodiment drugs are identified, which remove or reduce functionality encoded by one or several genes B^'. Such reduction can be accomplished by selecting drugs, for example RNAi therapies, which specifically reduce the levels of gene B^' encoded RNAs. Similar effects can be achieved by therapies affecting the transcription regulatory mechanisms of gene B^', thus indirectly leading to removal or substantial reduction of B^' encoded functionality. Such reductions can be achieved by targeting molecular pathways leading to down- regulation of the relevant gene or genes. Small molecule drugs can be used to directly inhibit functionalities encoded by genes B^' or lead to activation or repression of regulators of gene B^' encoded protein post-translational modifications. If functional inhibition is not possible, in certain cases the presence of gene B^' products can be utilized by other means, if expression levels are increased compared to healthy tissues. In case of antibody accessible proteins (usually cell-surface or transmembrane proteins) the availability of the encoded protein and decreased likelihood of proteins loss due to functional limitations in the cancerous tissue make them amenable to immunotherapies, such as antibody (usually monoclonal antibody) based therapies. It does at this level make no difference whether antibody mode of action is conferred by antibody functionalization (for example pro-drug activating enzymes) or by immunological mechanisms such as antibody dependent cytotoxicity (ADCC).

Alternatively, functions of molecules encoded by genes B^' can be used to convert prodrugs into active drugs, if encoded functionalities comprise an adequate enzymatic functionality or go along with presence of a suitable enzymatic functionality more downstream in a pysiological pathway in a tightly associated fashion. The nature of genes of class B^' make the latter two therapeutic approaches equally less likely to therapy escape, therapy inefficacy or recurrence of disease after end of therapy to those approaches described before, however, here the availability of gene B encoded functionality in healthy tissues may be taken into consideration.

Potential drug targets or biomarkers B^' may either already be subject to known drug/target interactions or may be subject to identification of novel drugs or re-use of known drugs.

Anticipated drugs include, but are not limited to, small molecule drugs or small interfering RNAs (RNAi). Potential drug targets B and/or B^'can be combined for designing therapies, as long as they are in no synthetic multi-lethal relationship to each other. Such combination therapies should then be assessed based on samples of individuals to determine the potential increase in patient/sample coverage by complennentary coverage or increase of potential therapy robustness by coverage of samples by multiple drug targets.

In many cases drugs affecting the presence of gene B^' encoded functionality will already be known, allowing for re-positioning of drugs for new indications. In certain cases, such as antineoplastic agents, the applicability for certain tumours may be new or untested and the presented invention can be used to propose the most suitable therapy from a set of drugs. Drug databanks such as DrugBank (Wishart et. al, Nucleic Acids Res. 2008 Jan;36 (Database issue): D901 -6) and Stich2 (Kuhn et al., Nucleic Acids Res. 2010 Jan;38 (Database issue): D552-6. Epub 2009 Nov 6) play important roles in the identification of drugs for re-positioning as they allow

identificiation of drugs associated with molecular drug targets identical or associated with identified products of genes B^'.

The present invention is described in further detail in the following examples, which are not in any way intended to limit the scope of the invention as claimed.

Examples: Identifying critical gene targets

The examples describe the application of the method for identifying critical gene targets to predict the response to a targeted cytotoxic therapy in a patient for three different neoplastic disorders of high relevance in terms of incidence and number of associated deaths. These are female ductal breast carcinoma, non small-cell lung adenocarcinoma and colon adenocarcinoma. For each of these cancer types the method was applied separately. To identify critical gene targets B^' specific for a cancer type the following stepwise procedure was applied.

In the absence of a comprehensive set of experimentally determined synthetic lethal relationships between human genes an alternative way was chosen, taking respective available information for yeast genes and mapping these genes to ortholog human genes.

Therefore the publicly available data set of synthetic lethality relationships between yeast genes (DRYGIN: a database of quantitative genetic interaction networks in yeast. Koh JL, Ding H, Costanzo M, Baryshnikova A, Toufighi K, Bader GD, Myers CL, Andrews BJ, Boone C. Nucleic Acids Res. 2010 Jan;38 (Database issue): D502-7. Epub 2009 Oct 30) was filtered for significant synthetic lethal interactions according to specifications proposed by the authors. This set of genes was mapped to ortholog (homology between organisms) human genes. The aim of this step was gaining information of synthetic lethal relations between genes in homo sapiens (human) based on the data given for yeast. To do so mapping information was obtained from five providers: Homologene (content downloaded on 2010/02/03: URL: http://www.ncbi.nlm.nih.gov/homologene; reference: Database resources of the

National Center for Biotechnology Information. Wheeler DL, Barrett T, Benson DA, Bryant SH, Canese K, Chetvernin V, Church DM, DiCuccio M, Edgar R, Federhen S, Geer LY, Kapustin Y, Khovayko O, Landsman D, Lipman DJ, Madden TL, Maglott DR, Ostell J, Miller V, Pruitt KD, Schuler GD, Sequeira E, Sherry ST, Sirotkin K, Souvorov A, Starchenko G, Tatusov RL, Tatusova TA, Wagner L, Yaschenko E. Nucleic Acids Res. 2007 Jan;35 (Database issue): D5-12. Epub 2006 Dec 14.), Inparanoid (content downloaded on 2010/02/03; URL: http://inparanoid.sbc.su.se/cqi-bin/index.cqi;

reference: InParanoid 6: eukarvotic ortholoq clusters with inparaloqs. Berqlund AC, Siolund E, Ostlund G, Sonnhammer EL. Nucleic Acids Res. 2008 Jan;36 (Database issue): D263-6. Epub 2007 Nov 30), RoundUp (content downloaded on 2010/02/03; URL: http://roundup.hms.harvard.edu/site/index.php; reference: Roundup: a multi- genome repository of orthologs and evolutionary distances. Deluca TF, Wu IH, Pu J, Monaghan T, Peshkin L, Singh S, Wall DP. Bioinformatics. 2006 Aug 15;22(16):2044- 6. Epub 2006 Jun 15.), OMA (content downloaded on 2010/02/03; URL:

http://www.omabrowser.org/cgi-bin/gateway.pl; reference: Algorithm of OMA for large- scale orthology inference. Roth AC, Gonnet GH, Dessimoz C. BMC Bioinformatics. 2008 Dec 4;9:518. Erratum in: BMC Bioinformatics.2009;10. doi:10.1 186/1471 -2105- 10-220) and ENSEMBL biomart (content downloaded on 2010/02/22; URL:

http://www.ensembl.org/index.html; reference: Ensembl 2007. Hubbard TJ, Aken BL, Beal K, Ballester B, Caccamo M, Chen Y, Clarke L, Coates G, Cunningham F, Cutts T, Down T, Dyer SC, Fitzgerald S, Fernandez-Banet J, Graf S, Haider S, Hammond M, Herrero J, Holland R, Howe K, Howe K, Johnson N, Kahari A, Keefe D, Kokocinski F, Kulesha E, Lawson D, Longden I, Melsopp C, Megy K, Meidl P, Ouverdin B, Parker A, Prlic A, Rice S, Rios D, Schuster M, Sealy I, Severin J, Slater G, Smedley D, Spudich G, Trevanion S, Vilella A, Vogel J, White S, Wood M, Cox T, Curwen V, Durbin R, Fernandez-Suarez XM, Flicek P, Kasprzyk A, Proctor G, Searle S, Smith J, Ureta-Vidal A, Birney E. Nucleic Acids Res. 2007 Jan;35 (Database issue): D610-7. Epub 2006 Dec 5).

While this process entailed several technical steps primarily because of different sequence ID spaces both the raw synthetic lethality dataset provided for yeast and the generated orthology dataset could be mapped to ensembl gene IDs, thus also removing ambiguities due to protein isoforms mapped to in component databases. The mapping was done assinging human ensemble gene IDs to yeast ensemble gene IDs leading to a human gene set for which synthetic lethal relations are extrapolated as given in the table below. Minimum number of 1 2 3 4 5 sources supporting

ortholog mapping

Number of ortholog human

3.714 1 .631 1 .055 662 340 genes

Number of synthetic lethal,

204.124 31 .068 13.144 5.408 1 .478 ortholog human gene pairs

Tablel : Overview on numbers of human genes and their synthetic lethal relations as resulting from ortholog mapping of yeast to human data sets. Given are the numbers of distinct human genes ortholog to at least one yeast gene which takes place in a synthetic sick interaction with orthology reported in minimum N databases and the numbers of distinct extrapolated, synthetic lethal human gene pairs (mapped by orthology with support of minimum N data sources).

The set union of the human genes mappable from yeast, corresponding genes supported by at least one orthology source were used for further analysis.

To determine expression of genes B and downmodulation of genes A in patient samples as well as expression of genes A in healthy tissue samples of a particular cancer type, publicly available transcriptomics raw data from primary tumor tissue and of samples from healthy tissue of the same tissue type were identified in a literature and database search. The sources for these data are listed in Table 2, Table 3 and Table 4. Transcriptomics raw data from studies as listed in Table 2, Table 3 and Table 4 were pre-processed to assess data quality. Therefore outlier detection was done based on hierarchical clustering and detection call distribution of the quantile-quantile normalized data set. Sample duplicates and outliers were excluded to receive the data pools subsequently used for the drawing procedure.

Study Number of cases Number of controls

Landi et al. 2008 58 49

Su et al. 2007 27 29

Yap et al. 2005 49 9

Shedden et al. 2008 321 0

Total 455 87 Table 2: Selected studies providing transchptomics raw data gained using HG- U133A microarrays (Affi metrics, CA, USA) from human samples comprising primary tumor tissue and respective normal tissue. Number of samples from non small-cell lung adenocarcinoma (cases) and healthy tissue (controls) used for further analysis are given (outlier adjusted). References are: Gene expression signature of cigarette smoking and its role in lung adenocarcinoma development and survival. Landi MT, Dracheva T, Rotunno M, Figueroa JD, Liu H, Dasgupta A, Mann FE, Fukuoka J, Hames M, Bergen AW, Murphy SE, Yang P, Pesatori AC, Consonni D, Bertazzi PA, Wacholder S, Shih JH, Caporaso NE, Jen J. PLoS One. 2008 Feb 20;3(2):e1651 . Selection of DDX5 as a novel internal control for Q-RT-PCR from microarray data using a block bootstrap re-sampling scheme. Su LJ, Chang CW, Wu YC, Chen KC, Lin CJ, Liang SC, Lin CH, Whang-Peng J, Hsu SL, Chen CH, Huang CY. BMC Genomics. 2007 Jun 1 ;8:140. Conserved transcription factor binding sites of cancer markers derived from primary lung adenocarcinoma microarrays. Yap YL, Lam DC, Luc G, Zhang XW, Hernandez D, Gras R, Wang E, Chiu SW, Chung LP, Lam WK, Smith DK, Minna JD, Danchin A, Wong MP. Nucleic Acids Res. 2005 Jan 14;33(1 ):409-21 . Gene expression-based survival prediction in lung adenocarcinoma: a multi-site, blinded validation study. Director's Challenge Consortium for the Molecular Classification of Lung Adenocarcinoma, Shedden K, Taylor JM, Enkemann SA, Tsao MS, Yeatman TJ, Gerald WL, Eschrich S, Jurisica I, Giordano TJ, Misek DE, Chang AC, Zhu CQ, Strumpf D, Hanash S, Shepherd FA, Ding K, Seymour L, Naoki K, Pennell N, Weir B, Verhaak R, Ladd-Acosta C, Golub T, Gruidl M, Sharma A, Szoke J, Zakowski M, Rusch V, Kris M, Viale A, Motoi N, Travis W, Conley B, Seshan VE, Meyerson M, Kuick R, Dobbin KK, Lively T, Jacobson JW, Beer DG. Nat Med. 2008 Aug;14(8):822- 7. Epub 2008 Jul 20.

Table 3: Selected studies providing transchptomics raw data gained using HG- U133 Plus 2.0 microarrays (Affimetrics, CA, USA) from human samples comprising primary tumor tissue and respective normal tissue. Number of samples from female ductal breast carcinoma (cases) and healthy tissue (controls) used for further analysis are given (outlier adjusted). References are: X chromosomal abnormalities in basal- like human breast cancer. Richardson AL, Wang ZC, De Nicolo A, Lu X, Brown M, Miron A, Liao X, Iglehart JD, Livingston DM, Ganesan S. Cancer Cell. 2006

Feb;9(2):121 -32. Proliferative genes dominate malignancy-risk gene signature in histologically-normal breast tissue. Chen DT, Nasir A, Culhane A, Venkataramu C, Fulp W, Rubio R, Wang T, Agrawal D, McCarthy SM, Gruidl M, Bloom G, Anderson T, White J, Quackenbush J, Yeatman T. Breast Cancer Res Treat. 2010 Jan;1 19(2):335- 46. Epub 2009 Mar 6.

Table 4: Selected studies providing transcriptomics raw data gained using HG- U133 Plus 2.0 microarrays (Affimetrics, CA, USA) from human samples comprising primary tumor tissue and respective normal tissue. Number of samples from colon adenocarcinoma (cases) and healthy tissue (controls) used for further analysis are given (outlier adjusted). References are: Transcriptome profile of human colorectal adenomas. Sabates-Bellver J, Van der Flier LG, de Palo M, Cattaneo E, Maake C, Rehrauer H, Laczko E, Kurowski MA, Bujnicki JM, Menigatti M, Luz J, Ranalli TV, Gomes V, Pastorelli A, Faggiani R, Anti M, Jiricny J, Clevers H, Marra G. Mol Cancer Res. 2007 Dec;5(12):1263-75. Metastasis-Associated Gene Expression Changes Predict Poor Outcomes in Patients with Dukes Stage B and C Colorectal Cancer. Jorissen RN, Gibbs P, Christie M, Prakash S, Lipton L, Desai J, Kerr D, Aaltonen LA, Arango D, Kruh0ffer M, Orntoft TF, Andersen CL, Gruidl M, Kamath VP, Eschrich S, Yeatman TJ, Sieber OM. Clin Cancer Res. 2009 Dec 15;15(24):7642-7651 . Epub. Evaluation of microarray preprocessing algorithms based on concordance with RT- PCR in clinical samples. Gyorffy B, Molnar B, Lage H, Szallasi Z, Eklund AC. PLoS One. 2009 May 21 ;4(5):e5645. Genome-wide gene expression analysis of mucosal colonic biopsies and isolated colonocytes suggests a continuous inflammatory state in the lamina propria of patients with quiescent ulcerative colitis. Bjerrum JT, Hansen M, Olsen J, Nielsen OH. Inflamm Bowel Dis. 2009 Oct 15. In the given example, downmodulation is determined (using data from gene expression micro-arrays) by measuring lowered concentration of gene expression products (messenger RNA). For each patient's sample each gene is statistically tested for being downmodulated in the sense of observing a significantly lowered

concentration of the expression product in the patient's sample compared to the normal tissue sample.

While described transcriptomics raw data pre-processing led to reasonably homogeneous data sets an iterative drawing process from these cancer type specific data sets as given in Table 2, Table 3 and Table 4 was applied to be able to estimate stability of results, consider varying numbers of case and control samples for the different cancer types and reduce biases. Each single drawing step was done by randomly selecting 30 tumor tissue samples for the cases group and randomly selecting 30 healthy samples for the control group from the cancer type specific sample set. This drawing is iterated 20 times for each cancer type leading to the respective result statistics. For each stratified random sampling, respective raw data were separately quantile-quantile normalized. To estimate non specific contributions from the analysis procedure and the data sets, separate runs with inverted sample assignments to the analysis groups were performed. This means 30 tumor samples were drawn for the control group and 30 healthy samples were drawn for the case group in each iterative run.

In each of these drawing steps a ranked list of critical genes B is derived based on a defined procedure. To consider a gene B expressed in a patient's sample or a gene A expressed in a healthy tissue sample, the MAS5 detection calls for features on Affymetix microarray platforms were used. This means, a valid intensity value from a microarray experiment exists for a feature called "present". However, no valid intensity value could be measured due to high mismatch intensities or zero concentration of respective RNA for features called "absent". A gene might be represented on a microarray by multiple features. It was defined to consider a gene as expressed in case all of its features are called present in a sample measurement. Otherwise the gene is called absent. An intensity value for a gene is derived as mean from respective feature intensities in a sample measurement.

In a single drawing step defining 30 cases and 30 controls each gene of the ortholog human gene set is checked for each case to be

a) downmodulated compared to the 30 controls and / or

b) expressed

Based on a) and b) for each gene of the ortholog human gene set a coverage value is calculated applying the information of synthetic letal relations between genes. The coverage represents the fraction of cases where a synthetic lethal effect is expected when repressing the respective gene B. A case is considered as affected if the gene B is expressed according to b) and at least one gene A exists according to a) for the case. For a gene A additionally the condition to be present in controls is essential. By repressing a gene B, different cases can be affected due to the same or different genes A. A gene might be a gene B and / or contribute as a gene A to the coverage of another gene B for a particular case depending on the synthetic lethal relationships to other genes.

Targeting of gene B^' or gene B^' products shall only lead to reduced viability in tumor cells due to the multi-synthetic lethal effect. Therefore if function encoded by gene B is depleted or missing (due to therapeutic intervention) only cells with missing or depleted functions encoded by genes A in multi-synthetic lethal relation with gene B^' (diseased cells) are affected. This specificity of therapeutic intervention to diseased cells is secured by the need of genes A to be expressed in healthy tissue.

Performing 20 drawing steps standard deviation for the coverage value can be calculated. The coverage of each gene was corrected by substraction of the

background coverage value gained by inverting cases and controls. The corrected coverage is used for sorting where highest coverage is prefered. As result of this analysis process a ranked list of critical genes B^' is derived. Additionally to each gene B^' a list of genes A was derived indicating the percentage the gene A contributes to the coverage of gene B^'.

Potential gene targets B^' were ranked by the coverage of cases (maximum coverage of patient samples indicating optimum expected stability of therapy). This ranking scheme has been applied for generation of general target lists specific for a particular cancer and enriches for encoded functions putatively less likely to be lost during cancer development. These data were further enriched by drugability data present in the STICH 2 database (STITCH 2: an interaction network database for small molecules and proteins. Kuhn M, Szklarczyk D, Franceschini A, Campillos M, von Mering C, Jensen LJ, Beyer A, Bork P. Nucleic Acids Res. 2010 Jan;38 (Database issue): D552-6. Epub 2009 Nov 6.) allowing identification of already known drugs targeting identified, putative drug targets B and thus readily paving the way for drug repositioning in context of our newly identified genes B^' and their products.

As an important note, selection of drug targets and ranking of these for individual patients is done differently, basically by selecting those genes B^' or combinations of genes B^' covered by a substantial multitude of genes A. Genes B^' with hundreds of As may also be discouraged, however, as this multitude may indicate increased levels of toxicity upon removal of function. Manual analysis of gene- associated function and, for example presence and dimension of paralogous clusters among genes A can help substantially in selecting targets. Independent whether drug targets for individuals or populations of patients are to be selected, rationalization of gene function has been proven useful in many cases. Table 5, Table 6 and Table 7 are extracts of gene B lists specific for the three cancer types ranked by coverage. Other genes B for which currently no marketed direct repressor therapy exists have not been included into the Tables. Genes B^' are listed with descending coverage values from top (rank 1 ) to the bottom (rank n) where the rank indicates the position of the gene in the list relative to other genes. Only genes B^' are extracted from these lists having an entry for inhibiting chemicals in STITCH 2 database investigated or discussed for the treatment of the specific cancer type. The tables hold the following information indicated by the column headers: Gene Symbol, ENSG: Ensembl Gene ID, Rank: position in the list sorted by coverage, Cvrg %: coverage value in % indicting the fraction of cases affected when targeting gene B^' as mean value derived from sample drawing, Stdv Cvrg %: standard deviation of the coverage value in % derived from sample drawing. Expr. change %: change of medium expression intensity of gene B in Patient's samples in % relative to medium expression intensity in controls, Drug: drugs targeting the expression product of gene B according to the STITCH 2 database.

Table 5: Extract of genes B from the ranked result list derived from processing data specific for colon adenocarcinoma samples.

Dihydrofolate reductase (gene symbol: DHFR) is known as a target for the approved drugs Pemetrexed, Trimetrexate and Methotrexate. These where clinically investigated in colon cancer patients as antineoplastic agents showing activity in particular in combination therapy. These drugs are antifolates, which impair the function of folic acids. Antifolates are used in cancer chemotherapy.

Aldehyde dehydrogenase 1 family, member A1 (gene symbol: ALDH1A1 ) is known as a target for Tretinoin, a naturally occurring derivative of vitamin A (retinol). The group of retinoids such as tretinoin shows antineoplastic activity and is used in the treatment of acute promyelocytic leukemia. However in-vitro studies have shown modulating affects also for colon cancer cells.

Histone deacetylase 1 , 3 and 8 (gene symbol: HDAC1 , HDAC3 and HDAC8) are known as targets for Vorinostat, which is an antineoplastic agent approved for cutaneous T-cell lymphoma. It inhibits the encymatic activity of histone deacetylases Class I and II. It has shown anti cancer activity in colon tumor cells and is currently under clinical investigation for colorectal cancer patients.

Topoisomerase (DNA) I (gene symbol: TOP1 ) is known as target for Irinotecan which is an antineoplastic enzyme inhibitor primarily used as part of the front line treatment of metastatic colorectal cancer. Irinotecan prevents religation of the DNA strand by binding to topoisomerase l-DNA complex, and causes double-strand DNA breakage and cell death.

Table 6: Extract of genes B from the ranked result list derived from processing data specific for non small-cell lung adenocarcinoma samples.

Tubulin, alpha 1 c (gene symbol: TUBA1 C) is known as target for Epothilone B which is a 16-membered macrolide with antineoplastic effects. It inhibits microtubule function and is investigated for use/tretment in lung cancer and other neoplasms.

Polymerase (DNA directed), alpha 2 (70kD subunit) (gene symbol: POLA2) is known as target for the non-classical alkylating agent Dacarbazine. It is usually used in conjunction with other drugs as an antineoplastic second-line therapy. Dacarbazine when used with other chemotherapeutic agents has shown activity in treatment of non- small cell lung cancer.

Gene ENSG Rank Cvrg Stdv Expr. Drug Symbol % Cvrg change

% %

CYP51A1 ENSG00000001630 1 77 8 42 Letrozole

CCNA2 ENSG00000145386 18 55 9 68 LY2931 1 1 ,

DENSPM

SLC25A5 ENSG00000005022 196 18 12 35 Clodronate Gene ENSG Rank Cvrg Stdv Expr. Drug

Symbol % Cvrg change

% %

SLC25A6 ENSG00000169100 209 17 1 1 -24 Clodronate

Table 7: Extract of genes B from the ranked result list derived from processing data specific for female ductal breast carcinoma samples.

Cytochrome P450, family 51 , subfamily A, polypeptide 1 (gene symbol:

CYP51A1 ) is known to be affected in activity by Letrozole which is a non-steroidal aromatoase inhibitor. CYP51A1 is involved in cholesterol biosynthesis. Letrozole is used as adjuvant treatment of hormonally-responsive breast cancer.

Solute carrier family 25 (mitochondrial carrier; adenine nucleotide translocator), member 5 (gene symbol: SLC25A5) and solute carrier family 25 (mitochondrial carrier; adenine nucleotide translocator), member 6 (gene symbol: SLC25A6) are known targets for Clodronate which is a (non-nitrogenous) bisphosphonate affecting calcium metabolism. Antineoplastic activity of clodronate was clinically investigated as adjuvant treatment in metastatic breast cancer patients.

Cycline A2 (gene symbol: CCNA2) is known to be repressed by LY2931 1 1 and DENSPM. LY2931 1 1 is known to be a leukotriene B4 antagonist, a 5-lipoxygenase inhibitor and a peroxisome proliferator-activated receptor (PPAR)-gamma agonist with cytotoxic properties in cell lines. It shows synergistic activity with the active metabolite of capecitabine in two breast cancer cell lines. DENSPM (N1 , N1 1 -Diethylnorspermine) induce programmed cell death in breast cancer models and it was clinically

investigated for treating metastatic breast cancer patients.

Selection of drugs based on proposed target molecule lists

The described example results in the selection and ranking of molecules the functional activity of which should be removed or impaired to produce a synthetic multi- lethal effect in a specific diseased tissue. For many of these small molecule drugs have been experimentally implicated to directly (for example through irreversible binding to the active site of the gene B encoded molecule) or indirectly (through effective down-regulation of gene B transcription) effect impairment of gene B encoded function. The discussed example has therefore been extended to include small molecule data contained in the publicly available database STITCH 2. Chemicals or drugs indicated by the database to have repressive effect on particular genes B were associated with these based on ENSG.

Claims

1 . A method of identifying critical gene targets as functional synthetic lethal partners to downmodulated or functionally compromised genes A to predict the response to a targeted cytotoxic therapy in a patient, comprising:

b) providing

- data of downmodulated genes A in a patient's sample, and

- data of expressed genes B in said patient's sample in synthetic-lethal relationship to at least one of said genes A, as identified using said array, wherein the suppression of genes B is otherwise not lethal in a normal cell, to obtain an

individualized data set,

e) producing an analytic carrier from said genes B^'.

2. Method according to claim 1 , wherein said analytic carrier further comprises said genes A in a synthetic-multi-lethal relationship to gene B^'.

3. Method according to claim 1 or 2, wherein said analytic carrier further comprises said genes B and optionally genes A in a synthetic lethal relationship to genes B.

4. Method according to claim 1 or 2, further comprising the validation of said analytic carrier, wherein a drug targeting one of the genes B^' has been proven to be effective in a patient of said patient population.

5. Method according to claim 1 to 4, wherein said genes B^' are identified, which are higher than a threshold, optionally afflicted with a ranking, reflecting the degree of multi-lethality and/or coverage of said patient population.

6. Method according to any of claims 1 to 5, wherein said array is based on genomic sequence data obtained by orthology mapping of genomic sequences from an eukaryotic species.

7. Method according to any of claims 1 to 6, wherein said genes B and B^' are of human origin.

8. Method according to any of claims 1 to 7, wherein said analytical carrier is further used for determining a patient profile.

9. Analytical carrier essentially consisting of critical target genes B^' and optionally down-modulated genes A in a synthetic multi-lethal relationship to each gene target B^', as a library that displays said genes as nucleotide sequences and/or expression products encoded by the sequences.

10. Carrier according to claim 9, wherein said genes or expression products are differentially tagged or located in spatial distinct compartments.

1 1 . Carrier according to any of claims 9 to 10, which is a microcarrier in a storage-stable form.

12. Carrier according to any of claims 9 to 1 1 , which is relevant to a patient population suffering from a disease selected from the group consisting of neoplastic disorders and infectious diseases and other disorders where drug based therapy is afflicted with cytotoxic effects, in particular cancer, including colon, lung, breast and prostate cancer.

13. Method of producing a drug for targeted cytotoxic therapy, comprising a) providing an analytical carrier according to any of claims 9 to 12,

b) identifying a critical target gene B^' for a specific patient population, c) obtaining a drug by drug discovery or design, which drug targets said critical gene target, said functionality or expression products, and

d) manufacturing a pharmaceutically acceptable formulation of said drug.

14. A method of identifying a patient eligible to targeted cytotoxic therapy, comprising

b) providing an analytic carrier according to any of claims 9 to 12, and c) matching said analytical results to predict the patient's response to said therapy.

15. A method of identifying a patient specific cytotoxic therapy, comprising a) analyzing a patient sample to provide analytical results for one or more genes,

b) providing an analytic carrier according to any of claims 9 to 12, and c) matching said analytical results to identify a gene target B^' and/ or a targeting drug specifically relevant for said patient.

16. Method according to any of claims 13 to 15, wherein said cytotoxic therapy is provided for the therapy of infectious disease or neoplastic disease.