WO2012046191A2

WO2012046191A2 - Identification of multi-modal associations between biomedical markers

Info

Publication number: WO2012046191A2
Application number: PCT/IB2011/054366
Authority: WO
Inventors: Nilanjana Banerjee; Angel Janevski; Sitharthan Kamalakaran; Vinay Varadan; Nevenka Dimitrova; Robert Lucito
Original assignee: Koninklijke Philips Electronics N.V.; Cold Spring Harbor Laboratory
Priority date: 2010-10-08
Filing date: 2011-10-04
Publication date: 2012-04-12
Also published as: CN103620608A; WO2012046191A3; US10340027B2; US20130196877A1; EP2625639A2

Abstract

The present invention relates to a method for identifying multi-modal associations between biomedical markers which allows for the determination of network nodes and/or high ranking network members or combinations thereof, indicative of having a diagnostic, prognostic or predictive value for a medical condition, in particular ovarian cancer. The present invention further relates to a biomedical marker or group of biomedical markers associated with a high likelihood of responsiveness of a subject to a cancer therapy, preferably a platinum based cancer therapy, wherein said biomedical marker or group of biomedical markers comprises at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 8, 19, 20 or all markers selected from PKMYT1, SKIL, RAB8A, HIRIP3, CTNNB1, NGFR, ZCCHC11, LSP1, CD200, PAX8, CYBRD1, HOXC11, TCEAL1, FZD10,FZD1, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR. Furthermore, an assay for detecting, diagnosing, graduating, monitoring or prognosticating a medical condition, or for detecting, 1 diagnosing, monitoring or prognosticating the responsiveness of a subject to a therapy against said medical condition, in particular ovarian cancer, is provided, as well as a corresponding method for classifying a subject comprising and a medical decision support system.

Description

IDENTIFICATION OF MULTI-MODAL ASSOCIATIONS BETWEEN BIOMEDICAL MARKERS

FIELD OF THE INVENTION

The present invention relates to a method for identifying multi-modal associations between biomedical markers which allows for the determination of network nodes and/or high ranking network members or combinations thereof, indicative of having a diagnostic, prognostic or predictive value for a medical condition, in particular ovarian cancer. The present invention further relates to a biomedical marker or group of biomedical markers associated with a high likelihood of responsiveness of a subject to a cancer therapy, preferably a platinum based cancer therapy, wherein said biomedical marker or group of biomedical markers comprises at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers selected from PKMYT1, SKIL, RAB8A, HIRIP3, CTNNB1, NGFR, ZCCHC11, LSP1, CD200, PAX8, CYBRD1, HOXC11, TCEAL1, FZD10, FZD1, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR. Furthermore, an assay for detecting, diagnosing, graduating, monitoring or prognosticating a medical condition, or for detecting, diagnosing, monitoring or prognosticating the responsiveness of a subject to a therapy against said medical condition, in particular ovarian cancer, is provided, as well as a corresponding method for classifying a subject and a medical decision support system.

BACKGROUND OF THE INVENTION

Ovarian cancer is the leading cause of death from gynecological malignancies mainly due to its late diagnosis, high mortality and low 5-year survival rates of only 30%. Reasons for this poor outcome include non specific presenting symptoms and identification in advanced stages of disease, mainly due to the absence of reliable screening methods for early detection. Ovarian cancer is the 6^th most common cancer world-wide with 204,000 cases and 125,000 deaths worldwide. The exact cause of developing ovarian cancer is still unknown; however, women with certain risk factors may be more likely than others to develop ovarian cancer. The top ranking factors include age, parity (like for breast cancer), personal and drug history.

Besides the correct and timely diagnosis of ovarian cancer, its treatment poses major challenges. Advanced ovarian cancer patients frequently receive carboplatinum as standard chemotherapy. Most patients initially respond to this chemotherapy. However, up to 75% of initial platinum responders recur with chemoresistant tumor within the first 2 years and eventually succumb to metastatic disease. In addition, a small fraction of ovarian cancers are primarily refractory to platinum compounds.

Thus, information about how a cancer develops through molecular events could not only be very helpful for diagnostic purposes and allow to improve the clinical outcome in patients with cancer at its earliest stage, while it is still localized and readily treatable, but also allow a clinician to predict more accurately how such a cancer is likely to respond to specific therapeutic treatments. In this way, a regimen based on knowledge of the tumor's sensitivity can be rationally designed. Hence, characterization of a cancer patient in terms of predicting treatment outcome enables the physician to make an informed decision as to a therapeutic regimen with appropriate risk and benefit trade-offs to the patient.

US Patent application 20090011049 is related to the area of cancer prognosis and therapeutics and discloses aberrant methylation patterns of particular genes in cancers. Here, the silencing of nucleic acids encoding a DNA repair or DNA damage response enzyme was used prognostically and for selecting treatments that are tailored for an individual patient. Combinations of these markers were used to provide prognostic information.

Although several genes are reported to be differentially regulated in certain cancer types and cancer therapy resistance situations, it is difficult to integrate this information in order to predict a course of events for patients suffering from or being examined for cancer, in particular ovarian cancer, and to assess early therapeutic resistance, in particular a resistance to platinum based therapeutics.

There is thus a need for an improved method for providing cancer prognostic information, as well as assays and diagnostic methods based thereon.

SUMMARY OF THE INVENTION

The present invention addresses this need and provides means and methods which allow the identification of stratifying genes based on multiple high-throughput modalities.

The above objective is in particular accomplished by a method for identifying multi-modal associations between biomedical markers comprising the steps of:

obtaining a plurality of datasets comprising data on multiple molecular profiling modalities from a plurality of primary subjects; obtaining a plurality of datasets comprising data on multiple molecular profiling modalities from a plurality of secondary subjects;

correlating the pluralities of datasets comprising data on multiple molecular profiling modalities of primary and secondary subjects;

identifying one or more stratifying biomedical markers which have different values for the primary subjects and the secondary subjects;

identifying a network and/or sub-network among the stratifying biomedical markers;

assigning a ranking score to the members of the identified network, said ranking score being based upon a network metric;

determining network nodes and/or high ranking network members or combinations thereof, indicative of having a diagnostic, prognostic or predictive value for a medical condition.

This method provides the advantage of being able to provide predictive information at an early developmental stage of a disease, e.g. a cancer disease, in particular ovarian cancer. Furthermore, it allows the assessment of a therapeutic resistance, such as a resistance to platinum based therapeutics like carboplatinum. The methodology has successfully been used to identify stratifying genes between resistant and sensitive chemotherapy patients.

In a preferred embodiment of the present invention said plurality of datasets on multiple profiling modalities from a plurality of primary and secondary subjects comprises data on methylated loci and on gene expression.

In a further preferred embodiment of the present invention the step of identifying a network and/or sub-network comprises the calculation of significance values for the stratifying biomedical markers.

In another preferred embodiment of the present invention said network metric comprises at least one element selected from the group of connectivity, adjacency, network density, network centralization, network heterogeneity, cliquishness, hub gene significance, network significance, centroid conformity, betweenness, centricity, closeness and

eccentricity.

In another preferred embodiment of the present invention said biomedical marker is a gene, a genomic locus in a coding region, a genomic locus in a non-coding region, a transcript and/or a protein. In yet another preferred embodiment of the present invention said primary subject is a healthy subject and said secondary subject is affected by a medical condition.

In a particularly preferred embodiment of the present invention, said medical condition is cancer.

In another particularly preferred embodiment of the present invention said cancer is ovarian cancer.

In a further preferred embodiment of the present invention, said predictive value is the indication of likelihood of responsiveness of a subject to a therapy comprising one or more platinum based drugs.

In a particularly preferred embodiment of the present invention said platinum based drug is carboplatinum.

In another aspect the present invention relates to a biomedical marker or group of biomedical markers associated with a high likelihood of responsiveness of a subject to a cancer therapy, preferably a platinum based cancer therapy, wherein said biomedical marker or group of biomedical markers comprises at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers selected from PKMYT1, SKIL, RAB8A, HIRIP3, CTNNB 1 , NGFR, ZCCHC11, LSP1, CD200, PAX8, CYBRD1, HOXC11, TCEAL1, FZD10, FZD1, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR, indicated in Table 1.

In a further aspect the present invention relates to an assay for detecting, diagnosing, graduating, monitoring or prognosticating a medical condition, or for detecting, diagnosing, monitoring or prognosticating the responsiveness of a subject to a therapy against said medical condition, preferably cancer, more preferably ovarian cancer, comprising at least the steps of

(a) testing in a sample obtained from a subject for the expression of a stratifying biomedical markers or group of said markers and/or of a network node and/or high ranking network member or group of said nodes or members obtained by a method as defined herein above, or as defined in the list or group of biomedical markers described herein above or below;

(b) testing in a control sample for the expression of the same marker, group of markers, network node, high ranking network member of group thereof as in (a);

(c) determining the difference in expression of markers of steps (a) and

(b); and (d) deciding on the presence or stage of a medical condition or the responsiveness of a subject to a therapy against said medical condition, preferably cancer, more preferably ovarian cancer, based on the results obtained in step (c).

In a preferred embodiment of the present invention said assay comprises the additional step of testing in a sample obtained from a subject for the methylation state and/or pattern of a stratifying biomedical marker or group of said markers and/or of a network node and/or high ranking network member or group of said nodes or members obtained by a method as defined herein above, or as defined in the list or group of biomedical markers described herein above, wherein in step (c) additionally the difference in methylation state and/or pattern is determined.

In a further aspect the present invention relates to a method for classifying a subject comprising:

(a) providing a subject's dataset comprising data on methylated loci and data on gene expression of a stratifying biomedical marker or group of said markers and/or of a network node and/or high ranking network member or group of said nodes or members obtained by a method as defined herein above, or as defined in the list or group of biomedical markers described herein above or below;

(b) accessing a database comprising database values for a stratifying biomedical marker or group of said markers and/or of a network node and/or high ranking network member or group of said nodes or members obtained by a method as defined herein above, or as defined in the list or group of biomedical markers described herein above or below; and

(c) calculating a subject's classification score based on the difference between database between the results of step (a) and (b).

In a further aspect the present invention relates to a medical decision support system comprising:

an input for providing a subject dataset comprising data on methylated loci and data on gene expression of a stratifying biomedical marker or group of said markers and/or of a network node and/or high ranking network member or group of said nodes or members obtained by a method as defined herein above, or as defined in the list or group of biomedical markers described herein above;

a computer program product for enabling a processor to carry out the method for classifying a subject comprising as define above; and

an output for outputting the subject classification score. BRIEF DESCRIPTION OF THE DRAWINGS

Fig 1 shows a flow chart of the analyses to indentify key pathways and genes in platinum resistance.

Fig 2 depicts the Wnt pathway and members of it that are stratifying genes (halos), in particular FZD1, GSK3B and CTNNB1.

Fig 3 shows a target-regulator network. The figure shows nodes, which are included if they have at least one known interaction from biological databases with the identified stratifying genes. Further indicated are CTNNB1 and CCND as hubs which are involved mainly in Wnt signaling. Further indicated is PTEN from the p53 signaling pathway,

Fig 4 depicts a methylation-expression correlation sub-network showing the nodes (red) that measure high in betweenness centrality measure. DETAILED DESCRIPTION OF EMBODIMENTS

The inventors have developed means and methods which allow the identification of stratifying genes based on multiple high-throughput modalities.

Although the present invention will be described with respect to particular embodiments, this description is not to be construed in a limiting sense.

Before describing in detail exemplary embodiments of the present invention, definitions important for understanding the present invention are given.

As used in this specification and in the appended claims, the singular forms of "a" and "an" also include the respective plurals unless the context clearly dictates otherwise.

In the context of the present invention, the terms "about" and "approximately" denote an interval of accuracy that a person skilled in the art will understand to still ensure the technical effect of the feature in question. The term typically indicates a deviation from the indicated numerical value of ±20 %, preferably ±15 %, more preferably ±10 %, and even more preferably ±5 %.

It is to be understood that the term "comprising" is not limiting. For the purposes of the present invention the term "consisting of is considered to be a preferred embodiment of the term "comprising of. If hereinafter a group is defined to comprise at least a certain number of embodiments, this is meant to also encompass a group which preferably consists of these embodiments only. Furthermore, the terms "first", "second", "third" or "(a)", "(b)", "(c)", "(d)" etc. and the like in the description and in the claims, are used for distinguishing between similar elements and not necessarily for describing a sequential or chronological order. It is to be understood that the terms so used are interchangeable under appropriate circumstances and that the embodiments of the invention described herein are capable of operation in other sequences than described or illustrated herein.

In case the terms "first", "second", "third" or "(a)", "(b)", "(c)", "(d)" etc. relate to steps of a method or use there is no time or time interval coherence between the steps, i.e. the steps may be carried out simultaneously or there may be time intervals of seconds, minutes, hours, days, weeks, months or even years between such steps, unless otherwise indicated in the application as set forth herein above or below.

It is to be understood that this invention is not limited to the particular methodology, protocols, reagents etc. described herein as these may vary. It is also to be understood that the terminology used herein is for the purpose of describing particular embodiments only, and is not intended to limit the scope of the present invention that will be limited only by the appended claims. Unless defined otherwise, all technical and scientific terms used herein have the same meanings as commonly understood by one of ordinary skill in the art.

As has been set out above, the present invention concerns in one aspect a method for identifying multi-modal associations between biomedical markers comprising the steps of:

obtaining a plurality of datasets comprising data on multiple molecular profiling modalities from a plurality of primary subjects;

obtaining a plurality of datasets comprising data on multiple molecular profiling modalities from a plurality of secondary subjects;

gene expression from a plurality of primary subjects;

assigning a ranking score to the members of the identified network, said ranking score being based upon a network metric; determining network nodes and/or high ranking network members or combinations thereof, indicative of having a diagnostic, prognostic or predictive value for a medical condition.

The term "biomedical marker" as used herein refers to a molecular, genetic, medical, biochemical, chemical, biological or physical condition associated with a subject, which may vary from one subject to another, e.g. from a subject afflicted by disease to a healthy subject.

The term "multiple molecular profiling modalities" as used herein refers to a modality associated with a molecular, genetic, medical, biochemical, chemical, biological or physical condition linked to a subject, e.g. a patient to be tested. Non- limiting examples of such modalities comprise the molecular state of a gene or genomic locus, the presence or absence or amount/level of transcripts, proteins, truncated transcripts, truncated proteins, the presence or absence or amount/level of cellular markers, the presence or absence or amount/level of surface markers, the presence or absence or amount/level of glycosylation pattern, the form of said pattern, the presence or absence of expression pattern on mR A or protein level, the form of said pattern, cell sizes, cell behavior, growth and environmental stimuli responses, motility, the presence or absence or amount/level of histological parameters, staining behavior, the presence or absence or amount/level of biochemical or chemical markers, e.g. peptides, secondary metabolites, small molecules, the presence or absence or amount/level of transcription factors, the form and/or activity of chromosomal regions or loci, as well as further modalities referring to the mentioned conditions or referring to additional conditions known to the person skilled in the art.

The term "plurality of datasets" refers to datasets comprising data on the above mentioned conditions, e.g. comprising data on profiles of one or more of the molecular, genetic, medical, biochemical, chemical, biological or physical conditions associated with a subject. A plurality of datasets may comprise at least one dataset, or more than one dataset, e.g. 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 50, 100 or more datasets. The datasets may comprise redundant or non-redundant information. The datasets may be provided in any suitable form known to the person skilled in the art, e.g. in suitable input formats for bioinformatic applications, as raw data etc.

The term "primary subjects" as used herein refers to a group of subjects, e.g. animals, in particular mammals. Preferably, a primary subject is a human being, e.g. a patient. The term may, in a specific embodiment, also refer to a sample obtained from a subject. Primary subjects are distinguished form a corresponding group of "secondary subjects" in that they can be associated with one or more of the mentioned molecular, genetic, medical, biochemical, chemical, biological or physical conditions associated with a subject which differ between the primary and secondary subjects.

The term "correlating the pluralities of datasets" as used herein means that the datasets or the comprised information is compared, e.g. between the datasets obtained from the primary and the secondary subjects and/or with datasets derivable from data repositories, from external sources, from literature values, from parallel examinations or the like.

Furthermore, the term may include the performance of statistical analyses or procedure. The term thus also means determining the influence of one marker in one modality to another marker's value in another modality. In a preferred embodiment the correlation is a statistically significant relationship or its variation between primary and secondary subjects.

The terms "stratifying biomedical markers" as used herein refers to conditions or features associated with the subjects, being derivable from or associated with the mentioned molecular, genetic, medical, biochemical, chemical, biological or physical condition used as input for the correlation step, wherein these conditions or features separate primary and secondary subjects. Thus, stratifying biomedical markers are differentiating conditions selected from the initially obtained plurality of datasets. In a typical embodiment, these biomedical markers show different values for the mentioned molecular, genetic, medical, biochemical, chemical, biological or physical condition, preferably the values are different in a statistically significant way. The term includes one differentiating marker, but also more than one such marker, e.g. a group of 2, 3, 4, 5, 6, 7, 8, 9. 10 or more markers. The stratification may be based then on comparison operations between the members of the group, e.g. be based on average values over the entire group or other statistical procedures known to the person skilled in the art.

The term "identifying a network" as used herein refers to the identification of relevant relationships between the biomedical markers. The term "relevant" means that suitable statistical procedures as known to the person skilled in the art may be employed in order to determine whether the relationship is significant. Typically, threshold values may be employed or suitable algorithms based on exclusion thresholds in order to eliminate relationships without medical or biological importance or without diagnostic or therapeutic medical value. The procedures may be repeated one or several times. Furthermore, the threshold or elimination values may be changed or varied, e.g. in dependence of the marker under considerations, the number of markers under consideration, the size of the obtained network etc. The "network" requires that every member of the network has at least one relationship or association with another member of the network. Preferably, the network members have more than one relationship or association with one or more other members of the network. Non-limiting examples of networks which may be identified are genetic or biochemical pathways, co-localized genetic markers or genetic loci, markers based on similar environmental inputs, target genes activated by transcription factors, etc.

The term "identifying a sub-network" as used herein means that within an already identified conjunction of markers or elements having relationships with other members of the network a sub-set of members is identified, which show a different type of relationship or a higher degree thereof (e.g. higher values with regard to certain conditions etc.). The term sub-network also comprises networks which only partially overlap with networks. The term also refers to more than one order of hierarchy between the networks, e.g. to sub-sub-networks etc. For example, a sub-network may be enriched in certain clinical parameters from a database, in certain pathway members, in the presence or absence of certain pathways, in genomic locations, the presence of chromosomes etc..

In the context of the present invention, the term "ranking score" refers to a score representing a numerical value. Preferably, the ranking score may be based upon a network metric. The term "network metric" refers to a measurement of performance in the network system. The term may also comprise a composite of two or more independent measures, typically in the form of a ratio. However, other combinations of measures are also possible.

One of the final steps of the method of identifying multi-modal associations between biomedical markers is the determination of network nodes. The term "network node" as used herein refers to a member of the network which shows more than a single association with other network elements.

In a preferred embodiment a network node is a multi association element or network hub. The term "network hub" means a node with a number of connections being larger than an average number of connections per node in a given network.

More preferably an important network hub is identified. The term "important network hub" as used herein refers to a hub with a number of connections being larger than an average number of connections per hub in a given network.

The method may also result in the identification of high ranking network members. The term "high-ranking network members" means that the ranking score of the members of the network is higher than the average ranking score of network members. In a specific embodiment such members may not only have one high-ranking score, but preferably 2, 3, 4, 5 or more. In yet another embodiment, the associations and/or ranking scores may be combined, e.g. a determination step may be based on two or more different informational elements such as nodes and/or hubs and/or important hubs and/or high-ranking network members or any sub-grouping thereof.

These nodes or network members may be indicative for medical conditions or have a diagnostic, prognostic or predictive value for a medical condition. The term "medical condition" as used herein refers to medical situation which differs from a healthy or normal state, e.g. a disease or a predisposition for a disease. The term "diagnostic value for a medical condition" means that one, 2, 3, 4, 5 or more values for a molecular, genetic, medical, biochemical, chemical, biological or physical condition associated with one, 2, 3, 4, 5, 6 identified nodes or network members, respectively, may upon arriving at typical thresholds, e.g. above or below conditions derived from primary subjects as defined above, be considered as indicating the presence of a medical condition, e.g. of a disease or a

predisposition for a disease. The term "prognostic value for a medical condition" refers to the prognosis of the development of a medical condition, e.g. form predisposition towards an acute illness, or from a less advanced stage to a more advanced stage, depending on one, 2, 3, 4, 5 or more values for a molecular, genetic, medical, biochemical, chemical, biological or physical condition associated with one, 2, 3, 4, 5, 6 identified nodes or network members, respectively.

The term "predictive value for a medical condition" refers to a value allowing the assessment of a medical condition or the development of said medical condition in the future, e.g. within a defined time frame of 1 to 3 weeks, 1 month, 2 month, 3 month, 4 months, 5 months, 6 months, 1, 2, 3, 4, 5, 6, 7, 10 years or more years or any other period of time. The term also includes all situations associated with said medical condition, e.g.

treatment results, responsiveness to treatments, development of resistance etc.

In a preferred embodiment of the present invention the plurality of datasets on multiple molecular profiling modalities from a plurality of primary and secondary

subsections comprises data on methylated loci and/or data on gene expression. Particularly preferred are data on methylated loci in combination with data on gene expression.

The term "methylated locus" as used herein refers to a nucleic acid derived from or present in a subject which is characterized by the presence of 5-methylcytosine ("5- mCyt") at one or a plurality of CpG dinucleotides within said nucleic acid. For example, such nucleic acid sequences may be genomic portions or loci which are derived from or which are present in a subject. These loci may, for example, comprise active or inactive portions of the genome. Preferably such loci are genes including all necessary elements for expression of the encoded genetic information, e.g. cis and trans acting elements etc. The information on methylated loci may preferably comprise information on the methylation state of said locus or genomic region or gene or any sub-fragment thereof.

In the context of the present invention the term "methylation state" means the degree of methylation present in a nucleic acid of interest. This may be expressed in absolute or relative terms, i.e. as a percentage or other numerical value or by comparison to another tissue and therein described as hypermethylated, hypomethylated or as having significantly similar or identical methylation status.

The term "hypermethylation" or "hypermethylated" as used herein refers to the average methylation state corresponding to an increased presence of 5-mCyt at one or a plurality of CpG dinucleotides within a DNA sequence of a test DNA sample, relative to the amount of 5-mCyt found at corresponding CpG dinucleotides within a normal control DNA sample.

The term "hypomethylation" as used herein refers to the average methylation state corresponding to a decreased presence of 5-mCyt at one or a plurality of CpG dinucleotides within a DNA sequence of a test DNA sample, relative to the amount of 5- mCyt found at corresponding CpG dinucleotides within a normal control DNA sample.

Thus, in a preferred embodiment the "methylated locus" may show a hypermethylation, a hypomethylation or an unmodified methylation status in comparison to a the methylation state of a primary subject.

The locus or the loci to be analysed may comprise any suitable dimension known to the person skilled in the art. For example, fragments of about 5, 10, 20, 50, or 100 nucleotides, of about 1 kbp, 2 kbp, 3 kbp, 4 kbp 5 kbp, 6 kbp, 7 kbp 10 kbp, 15 kbp, 20 kbp, 25 kbp, 30 kbp, 35 kbp, 40 kbp, 100 kbp, entire chromosomes, more than one chromosome (e.g. 2, 3, 4, 5, 6 etc. chromosomes), or the entire genome may be analysed with regard to the methylation state.

In one embodiment the epigenetic analysis of single genes (including all elements necessary for their expression, e.g. promoter, enhancer elements, open reading frame, terminator etc. or only sub-fragments thereof), or of conjunctions of genes, e.g. of pathway members, or of any combination or conjunction of genes is envisaged. Further envisaged is the epigenetic analysis of regulatory regions. The term "regulatory region" means a nucleotide sequences which affect the expression of a gene. Said regulatory regions may be located within, proximal or distal to said gene. The regulatory regions include but are not limited to constitutive promoters, tissue-specific promoters, developmental-specific promoters, inducible promoters, as well as non-coding R As (such as microR As) and the like. Promoter regulatory elements may also include certain enhancer sequence elements that control transcriptional or translational efficiency of the gene. These sequences can have various levels of binding specificity and can bind to transcription factors as well as DNA methyl-binding proteins, e.g. MeCP, Kaiso, MBD1-MBD4. The term "epigenetic" as used herein refers to the modification of biological, i.e. genetic behavior due to changes other than changes in the underlying DNA sequence. Typical, non-limiting examples of epigenetic modifcations are methylation of a genomic sector or locus, chromatin remodeling, or the interaction of DNA with RNA transcripts. Thus, besides or in addition to the analysis of methylation or methylation pattern or states also the status of chromatin and/or the presence of interaction RNA species may be determined.

The term "pathway" as used herein refers to the set of interactions occurring between a group of genes, which genes depend on each other's individual functions in order to make the aggregate function of the network available to the cell.

Methylation may be determined with any suitable means known to the person skilled in the art, for example, with an methylation assay, e.g. an assay for determining the methylation state of one or more CpG dinucleotide sequences within a sequence of DNA. Such assays may be based on the employment of methylation specific PCR or methylation specific sequencing to assess the level of DNA methylation. Details would be known to the person skilled in the art.

"Gene expression" as used herein refers to the transcription and/or translation of a gene. "Gene expression" or lack thereof may be a consequence of epigenetic

modifications of the genomic DNA associated with the marker gene and/or regulatory or promoter regions thereof etc. Genetic modifications may include SNPs, point mutations, deletions, insertions, repeat length, rearrangements, copy number variations and other polymorphisms. The analysis of either the expression levels of protein, or mRNA expression are summarized as the analysis of "expression" of the gene. The term refers to the expression of a single gene, may however also comprise the expression of a group of genes, .e.g. genes located in a pathway, genes co-localized in a genomic region, genes present on a

chromosome or in a chromosomal region etc.

In one embodiment of the invention datasets on the methylation state as described above are analysed in conjunction with one ore more other datasets on a molecular, genetic, medical, biochemical, chemical, biological or physical conditions as defined above. In another embodiment of the invention datasets on gene expression as described above may be analysed in conjunction with one or more datasets on a molecular, genetic, medical, biochemical, chemical, biological or physical conditions as defined above.

Particularly preferred is an analysis of datasets on the methylation state and datasets on gene expression. For example, data on the expression of a gene or of pathway members or of co-localized genes etc. may be correlated with the methylation state or epigenetic status of the very same elements or of neighboring elements or structures, or vice versa. Alternatively, data on the expression of a gene or of pathway members etc. may be correlated with the methylation state or epigenetic status of a different gene or a member of different pathways, e.g. of elements associated on a different level or in a different manner, or vice versa.

In a further embodiment of the invention the step of identifying a network and/or sub-network comprises the step of calculating significance values for the stratifying biomedical markers. The term "significance value" as used herein refers to any suitable statistical value which allows a statistically relevant distinction between two situations.

Preferably, the term relates to the calculation of p-values. More preferably, the significance value may be a p-value based on the Hypergeometric distribution or Fisher's exact test.

In a specific embodiment, the calculation of a significance value may be performed according to the following example. Under the assumption that there are N genes, where N would be the number of genes present in primary and secondary datasets and that M genes are linked by a different profiling modality, e.g. annotated to a specific pathway in the set of pathways, or associated by any other mean, n genes are found to be in the input list, such as comprised within the stratifying genes, for example differentially methylated, k represents the number of genes from the input list which are also annotated to the specific pathway. The probability for any given k, where k is an integer in the set of integers from 1 to n, can then be calculated according to the formula:

In a further embodiment of the present invention, the calculation of significance values for the stratifying biomedical markers may be supplemented with the performance of a suitable correction procedure. A particularly preferred procedure is the Benjamini & Hochberg False Discovery Rate (FDR) correction.

In a further embodiment of the invention the network metric to the assessed comprises at least one element selected from the group of connectivity, adjacency, network density, network centralization, network heterogeneity, cliquishness, hub gene significance, network significance, centroid significance, centroid conformity, betweenness, centricity, closeness and eccentricity.

The term "connectivity" as used herein refers to the number of network elements, e.g. genes, adjacent to a network element, e.g. gene, or that are directly linked.

The terms "adjacency" or "closeness" as used herein refers to the number of network elements, e.g. genes, which do not form neutral relationships, e.g. like or dislike others.

The term "network density" as used herein refers to the mean off-diagonal adjacency. This metric measures the overall affection among network elements, e.g. genes.

The term "network centralization" refers to topological structures of a network. For example a centralization of 1 indicates a network with star topology, whereas a centralization of 0 indicates a network where each node has the same connectivity.

The term "network heterogeneity" as used herein refers to the variance of connectivity.

The term "cliquishness" as used herein refers to a density measure of local connections.

The term "hub gene significance" as used herein refers to the association between connectivity and gene significance.

The term "network significance" as used herein refers to the average of gene significance of the genes. This is typically a measure of the average grade point average among the network elements, e.g. genes.

The term "centroid significance" or "centricity" as used herein refers to the gene significance of a suitably chosen representative node (centroid) in a the network.

The term "centroid conformity" as used herein refers to the adjacency between a network element, e.g. a gene, and a suitably chosen representative node (centroid).

The term "betweenness" as used herein refers to a high influence of a network element, e.g. gene, over the information flow between other elements, e.g. genes, in the network. The term "eccentricity" as used herein refers the accessibility of an element of the network, e.g. a gene, by all other elements, e.g. genes, in the network.

Further details as well as formula for the calculation of the corresponding metrics may be derived from suitable scientific publications known to the person skilled in the art, e.g. from Horvath and Dong, PLoS Computational Biology, 2008, 4(8), elOOOl 17, pages 1 to 27 (which is incorporated herein in its entirety), in particular from pages 3 to 7.

The metrics as mentioned above may be determined according to suitable methods known to the person skilled in the art, e.g. as described in Horvath and Dong.

Metrics may be used alone or in any combination. Preferably, the betweenness and/or the eccentricity and/or the connectivity of elements of the network, e.g. of genes, may be determined. For example, the betweenness of network elements, e.g. genes, may be used as ranking method, and the eccentricity and/or connectivity may additionally be used in order to increase the significance of the network element and/or to eliminate non-significant elements. A typical example of such a ranking is provided in Table 3.

In a preferred embodiment of the present invention a ranking score is assigned to a network element or node, e.g. a gene, according to the metric betweenness. In a non- limiting example, only network elements or nodes, e.g. genes, are considered which show a betweenness value of 0.00005 or higher, e.g. 0.0001, 0.00015, 0.0002 etc. In a further embodiment a ranking score is assigned to a network element or node, e.g. a gene, according to the metric eccentricity. In a non-limiting example, only genes are considered which show an eccentricity value of about 1.35 and higher, e.g. about 1.4, 1.45, 1.5, 1.55, 1.6, 1.65, 1.7 or higher etc. In yet another embodiment a ranking score is assigned to a network element or node, e.g. a gene, according to the metric connectivity. In a non-limiting example, only network elements or nodes, e.g. genes, are considered which show a connectivity value of about 2 or higher, e.g. about 3, 4, 5, 6, 7, 8, 9, 10, 15 or higher etc. In another embodiment of the present invention the ranking scores of betweenness, eccentricity and/or connectivity may be combined. The combination may be weighted according to any ranking list mentioned above, e.g. according to the ranking of betweenness, or according to the ranking of eccentricity, or according to the ranking of connectivity, or according to groups of two of the rankings etc.

In a further embodiment any of the other mentioned metrics, i.e. adjacency, network density, network centralization, network heterogeneity, cliquishness, hub gene significance, network significance, centroid significance, centroid conformity, centricity and closeness may also be used as primary ranking input, i.e. for the definition of a ranking score. Accordingly obtained rankings may further be combined with rankings derived from metrics such as betweenness, eccentricity and/or connectivity or any other of the above mentioned group, e.g. as secondary ranking input. A combination may be calculated according to suitable methods, e.g. be based on the average ranking position, or be based on an weighting factor, e.g. according to importance and/or significance of the metric input.

In a specific embodiment of the present invention the method for identifying multi-modal associations between biomedical markers may be carried out with the help of suitable software tools such as BingGO and/or Cytoscape. In a non-limiting example, the following steps may be carried out:

(a) importing network metrics as attributes of the nodes;

(b) selecting a subset of nodes (e.g. top ranked nodes) and edges (e.g. most correlated and anti-correlated nodes), e.g. with a threshold value of 0.5;

(c) invoking of the BingGO plug-in for Cytoscape (further details are described in Maere et al, BINGO: a Cytoscape plugin to assess overrepresentation of Gene Ontology categories in biological networks, Bio informatics, 2006, 21, 3448, which is incorporated herein in its entirety);

(d) parameters for BinGO may be set

(e) ontology: Molecular Function/Biological Process may be selected

(f) annotation (organism): Homo sapiens may be selected;

(g) statistical test : Hypergeometric test may be selected;

(h) correction : Benjamini & Hochberg False Discovery Rate (FDR) correction may be selected;

(i) significance level may be set to 0.05;

(j) testing options may be set to: Test cluster versus whole annotation; (k) the analysis, e.g. overrepresentation analysis, may be executed.

In a further preferred embodiment of the invention the biomedical marker to be assessed or analysed is a gene, a genomic locus, a transcript and/or a protein. A gene may be an entity comprising all necessary elements ensuring expression of the gene, e.g.

promoter, enhancer, open reading frame, terminator or sub-groups thereof. A genomic locus as used herein may be a genomic locus which is located in a non-coding region.

Alternatively, the genomic locus may be located in a coding-region. The genomic locus may be of a different size, e.g. encompassing a short fragment of 10 to 1000 nucleotides, or larger fragments of 2kbp, 5 kbp, 10 kbps, lOOkbp, 1 Mbp, chromosomal arms, or entire

chromosomes. The genomic locus may also be located partially in a coding region and partially in a non-coding region.

The transcript may be any form of nucleic acid derived from an active gene, e.g. an mRNA molecule, a non-spliced mRNA molecule, a truncated mRNA molecule, short fragments thereof etc. The protein may be a full-length protein, or any fragment thereof. Alternatively, an antibody or a ligand specifically binding to the protein, binding to the RNA or DNA or to the chromosomal structure is envisaged as biomedical marker to be tested.

In a further preferred embodiment the mentioned primary subject is a healthy subject. The term "healthy subject" relates to a human being not afflicted by a specific disease in comparison to a second subject, e.g. human being, with regard to the same disease. The term "healthy" thus refers to specific disease situations for which a subject shows no symptoms of disease. The term thus not necessarily means that the person is entirely free of any disease. However, also these persons are envisaged as being healthy for the purpose of the present invention.

In a further preferred embodiment said secondary subject is affected by a medical condition. A medical condition may be any disease or illness known to the person skilled in the art. Such a condition preferably results in the distinguishability of biomedical markers. Preferably the medical condition is cancer. More preferably, the medical condition is an ovarian cancer.

In a further particularly preferred embodiment of the present invention the predictive value which results from the determination of network nodes and/or high ranking network members or combinations thereof is the indication of a likelihood of responsiveness of a subject to a therapy. Such a therapy may be of any type, for instance a chemotherapy, e.g. a chemotherapy against a disease. The term "likelihood of responsiveness" as used herein refers to the probability that a subject may develop a non-responsive state towards the therapy, e.g. develops a resistance against the therapy or the given therapeutic composition. The term "chemotherapy" as used herein means the use of pharmaceutical or chemical substances to a disease, in particular to treat cancer.

In a particularly preferred embodiment the likelihood of responsiveness is a likelihood of responsiveness of a subject to a therapy comprising one or more platinum based drugs. Examples of platinum based drugs are cisplatinum and derivatives or analogs thereof, e.g. oxiplatinum, satraplatinum.

In a particularly preferred embodiment said platinum based drug is

carboplatinum. A methodology as described herein above may, hence, be used to identify network elements, e.g. genes and/or genomic loci, which allow an assessment of the likelihood to respond to a platinum based therapy, in particular to a carboplatinum based therapy, e.g. during the treatment of cancer, in particular during the treatment of ovarian cancer.

In another aspect the present invention relates to a biomedical marker or group of biomedical markers associated with the development of a disease, in particular a cancer disease, e.g. ovarian cancer, or associated with a high likelihood of responsiveness of a subject to a cancer therapy. The group of biomedical markers comprises at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers selected from PKMYTl, SKIL, RAB8A, HIRIP3, CTNNBl, NGFR, ZCCHCl l, LSPl, CD200, PAX8, CYBRDl, HOXCl l, TCEALl, FZD10, FZDl, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR, as indicated in the following Table 1 :

Table 1:

SEQ ID NO Genomic DNA

Gene Name Description nucleotide Accession No. (with sequence indicated range)

AC 000148.1

Serine threonine

PKMYT1 1 Range:

kinase

2995282..3002378

NC_000003.11

Oncogene;

SKIL 2 Range:

SMAD signaling

170075472..170110949

A RAS NC_000019.9

RAB8A oncogene; GTP 3 Range:

binding; 16222489..16244444

NC_000016.9

Histone

HIRIP3 4 Range:

repressor

30004315..30007386

NG 013302.1

CTNNBl WNT Signalling 5

Range: 5000..45997

P75; cytokine

NC 000017.10 Range:

NGFR receptor 6

47572654..47592371 interaction

NC_000001.10

ZCCHCl l Zinc finger 7 Range:

52888946..53018742

Lymphocyte NC_000011.9

LSPl specific protein; 8 Range:

signal transducer 1874199..1913492

NC 000003.11 Range:

CD200 glycoprotein 9

112051915.. 112081658

Paired box NC_000002.11

PAX8 transcription 10 Range:

factor 113973573..114036497 NC_000002.11

cytochrome b

CYBRD1 11 Range:

reductase 1

172378865..172414642

NC_000012.11

HOXC11 Homeobox CI 1 12 Range:

4366909..54370202 transcription NC 000023.10 Range:

TCEAL1 elongation factor 13

A (Sll)-like 1 102883647..102885880

NC_000012.11

frizzled homolog

FZD10 14 Range:

10

130647031..130650284

NC_000007.13

frizzled homolog

FZD1 15 Range:

1

90893782..90898131

Bardet-Biedl

NG 009416.1

BBS4 syndrome 4 16

Range: 5000..57291 protein

insulin receptor NG 008154.1

IRS2 17

substrate 2 Range: 5000..37730

NC_000005.9

T-cell leukemia

TLX3 18 Range:

homeobox 3

170736287..170739137

NC_000001.10

TSPAN2 tetraspanin 2 19 Range:

115590631..115632114

NC_000009.11 Range:

TXN thioredoxin 20

113006309..113018777

CASP8 and

NC_000002.11

FADD-like

CFLAR 21 Range:

apoptosis

201980815..202029015 regulator

In a particularly preferred embodiment of the present invention the mentioned biomedical marker or group of biomedical markers is associated with a high likelihood of responsiveness of a subject to an ovarian cancer therapy. In a further particularly preferred embodiment of the present invention the mentioned biomedical marker or group of biomedical markers is associated with a high likelihood of responsiveness of a subject to an ovarian cancer therapy comprising platinum based drugs. In yet another particularly preferred embodiment of the present invention the mentioned biomedical marker or group of biomedical markers is associated with a high likelihood of responsiveness of a subject to an ovarian cancer therapy comprising carboplatinum. In a further embodiment of the invention the method of identifying multimodal associations between biomedical markers as defined herein above may be based on datasets comprising parameters linked to the marker or group of markers defined above, e.g. at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers selected from PKMYT1, SKIL, RAB8A, HIRIP3, CTNNBl, NGFR, ZCCHCl l, LSPl, CD200,

PAX8, CYBRDl, HOXCl l, TCEAL1, FZD10, FZD1, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR, or one or more of the markers or group of markers mentioned in Tables 2, 4 or 5. Preferably, the datasets may comprise data on DNA methylation and/or gene expression.

In a further embodiment the present invention relates to a group of biomedical markers associated with the development of a disease, in particular a cancer disease, e.g. ovarian cancer, or associated with a high likelihood of responsiveness of a subject to a cancer therapy, wherein the group of biomedical markers comprises at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers selected from PKMYT1, SKIL, RAB8A, HIRIP3, CTNNBl, NGFR, ZCCHCl l, LSPl, CD200, PAX8, CYBRDl, HOXCl l, TCEAL1, FZD10, FZD1, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR, as indicated in Table 1 in combination with at least 1, 2, 3, 4, 5, 6, 7, 8, 9,10 or more of the markers mentioned in Tables 2, 4 and/or 5. Particularly preferred is a group of markers comprising least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers selected from PKMYT1, SKIL, RAB8A, HIRIP3, CTNNBl, NGFR, ZCCHCl l, LSPl, CD200, PAX8, CYBRDl, HOXCl l, TCEAL1, FZD10, FZD1, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR and 1, 2, 3, 4, 5, 6, 7, 8, 9,10 or more of the markers indicated in Table 2. In a further embodiment the group of biomedical markers comprises at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers selected from PKMYT1, SKIL, RAB8A, HIRIP3, CTNNBl, NGFR, ZCCHCl l, LSPl, CD200, PAX8, CYBRDl, HOXCl l, TCEAL1, FZD10, FZD1, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR and at least one member of one or more of the pathways indicated in Table 2, e.g. Androgen receptor, Pitx2 driven transcription regulation, Wnt signaling pathway, Gata3 and th2 cytokine gene expression, Segmentation clock, PI3K-akt, Leukocyte transendothelial migration and/or Phosphorylation of mekl by cdk5/p35. Members of the indicated pathway are known to the person skilled in the art, and/or can be derived from qualified textbooks.

The present invention envisages the markers in the form of genetic units, e.g. as genes, or in the form of expressed units, e.g. as transcripts, proteins or derivatives thereof. Furthermore, the marker may comprise secondary binding elements, such as an antibody, a binding ligand, siRNA or antisense RNA molecules specific for the marker transcript. Further included are genomic loci of the mentioned marker, e.g. the genomic DNA indicated in Table 1, or sub-fragments thereof. The marker may also comprise epigenetic modifications within the gene or genomic locus associated with the marker, e.g. methylated forms of the gene or genomic locus, hypomethylated forms of the gene or genomic locus etc.

In one embodiment of the present invention, the group of markers comprises

PKMYTl and SKIL. In a further embodiment of the present invention the group of marker comprises PKMYTl and RAB8A. In a further embodiment of the present invention the group of marker comprises PKMYTl and HIRIP3. In yet another embodiment of the present invention the group of marker comprises PKMYTl and CTNNB1. In yet another

embodiment of the present invention the group of marker comprises PKMYTl and NGFR. In yet another embodiment of the present invention the group of marker comprises PKMYTl and ZCCHC 11. In yet another embodiment of the present invention the group of marker comprises PKMYTl and LSPl . In yet another embodiment of the present invention the group of marker comprises PKMYTl and CD200. In yet another embodiment of the present invention the group of marker comprises PKMYTl and PAX8. In yet another embodiment of the present invention the group of marker comprises PKMYTl and CYBRD1. In yet another embodiment of the present invention the group of marker comprises PKMYTl and HOXC11. In yet another embodiment of the present invention the group of marker comprises PKMYTl and TCEAL1. In yet another embodiment of the present invention the group of marker comprises PKMYTl and FZD10. In yet another embodiment of the present invention the group of marker comprises PKMYTl and FZD1. In yet another embodiment of the present invention the group of marker comprises PKMYTl and BBS4. In yet another embodiment of the present invention the group of marker comprises PKMYTl and IRS2. In yet another embodiment of the present invention the group of marker comprises PKMYTl and TLX3. In yet another embodiment of the present invention the group of marker comprises PKMYTl and TSPAN2. In yet another embodiment of the present invention the group of marker comprises PKMYTl and TXN. In yet another embodiment of the present invention the group of marker comprises PKMYTl and CFLAR.

In a further embodiment of the present invention, the group of markers comprises PKMYTl and SKIL and RAB8A. In a further embodiment of the present invention the group of marker comprises PKMYTl and SKIL and HIRIP3. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and CTNNB1. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and NGFR. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and ZCCHC11. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and LSPl . In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and CD200. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and PAX8. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and CYBRD1. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and HOXC11. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and TCEAL1. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and FZD10. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and FZD1. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and BBS4. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and IRS2. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and TLX3. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and TSPAN2. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and TXN. In yet another embodiment of the present invention the group of marker comprises PKMYTl and SKIL and CFLAR.

In a further embodiment of the present invention, the group of markers comprises PKMYTl and 2, 3, 4, 5, 6, 7, 8 or more of the markers of Table 1. In a further embodiment of the present invention, the group of markers comprises SKIL and 2, 3, 4, 5, 6, 7, 8 or more of the markers of Table 1.

In yet another embodiment of the present invention, the group of markers comprises at least 1, 2, 3, 4, 5, 6, 7, 8, 9 or all of PKMYTl, SKIL, RAB8A, HIRIP3, CTNNB 1 , NGFR, ZCCHC11, LSPl, CD200 and PAX8.

In further specific embodiments the present invention relates to groups of markers as indicated in Table 4 and/or 5, e.g. markers which are overrepresentated in gene ontology categories of molecular function and/or biological processes. For instance, the present invention relates to a group of markers indicated in section "ALL" of Table 4 and/or 5. In a further embodiment, the present invention relates to a group of markers indicated in section "CENTRICITY" of Tables 4 and/or Table 5. In a further embodiment, the present invention relates to a group of markers indicated in section "CLOSENESS" of Tables 4 and/or Table 5. In a further embodiment, the present invention relates to a group of markers indicated in section "HIGHCONN" of Tables 4. In a further embodiment, the present invention relates to a group of markers indicated in section "ECCENTRICITY" of Table 5.

In further specific embodiments the present invention relates to groups comprising at least 1, 2, 3, 4, 5, 6, 7, 8, 9 or all markers of PKMYT1, SKIL, RAB8A, HIRIP3, CTNNBl, NGFR, ZCCHC11, LSPl, CD200 and PAX8 of Table 1 and at least 1, 2, 3, 4, 5 or more markers as indicated in Table 2.

In a further aspect the present invention relates to a method of diagnosis in vitro or in vivo of a medical condition, e.g. a cancer disease, preferably ovarian cancer, wherein said method is based on the determination of molecular parameters linked to the marker as defined above, e.g. a marker or group of markers comprising a at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers of Table 1. Preferably, the method of diagnosis comprises the determination of presence or absence or amount/level of an expression product (e.g. protein, transcript etc.) of one or more of the markers. In addition or alternatively, the determination of a secondary parameter such as the methylation status of the marker may be carried out. In a specific embodiment, the marker for which the expression is determined may not be identical to the marker for which a secondary parameter such as the methylation status is determined.

In a further aspect the present invention relates to a composition for in vivo or in vitro diagnosing, detecting, monitoring or prognosticating a disease, preferably a cancer disease, more preferably ovarian cancer, or for diagnosing, detecting, monitoring or prognosticating the likelihood of responsiveness of a subject to a cancer therapy, preferably the therapy against ovarian cancer, more preferably a platinum drug based therapy, even more preferably a carboplatinum based therapy, comprising a nucleic acid affinity ligand and/or a peptide affinity ligand for the expression product(s) or protein(s) of the above mentioned marker or group of markers. Such a composition may alternatively or additionally comprise an antibody against any of the above mentioned markers.

In a preferred embodiment of the present invention said nucleic acid affinity ligand or peptide affinity ligand is modified to function as an imaging contrast agent.

Further envisaged is a method of identifying a subject for eligibility for a cancer disease therapy comprising:

(a) testing in a sample obtained from subject for a parameter associated with a marker or group of markers as indicated herein above;

(b) classifying the levels of tested parameters; and (c) identifying the individual as eligible to receive a cancer disease therapy where the subject's sample is classified as having an increased expression of one or more of the above mentioned markers and/or as having a modified methylation state of one or more of the above mentioned markers.

In another aspect the present invention relates to an assay for detecting, diagnosing, graduating, monitoring or prognosticating a medical condition, preferably cancer, more preferably ovarian cancer, comprising at least the steps of

(a) testing in a sample obtained from a subject for the expression of a stratifying biomedical markers or group of said markers and/or of a network node and/or high ranking network member or group of said nodes or members wherein said members can be obtained according to the above described method; alternatively, the testing may be carried out with a marker or group of markers as defined herein above;

(c) determining the difference in expression of markers of steps (a) and

(b); and

(d) deciding on the presence or stage of medical condition or the responsiveness of a subject to a therapy against said medical condition, based on the results obtained in step (c).

In yet another aspect the present invention relates to an assay for detecting, diagnosing, graduating, monitoring or prognosticating the responsiveness of a subject to a therapy against said medical condition, preferably cancer, more preferably ovarian cancer, even more preferably the responsiveness of a subject to a platinum drug based therapy, e.g. carboplatinum, comprising at least the steps of

(c) determining the difference in expression of markers of steps (a) and

(b); and (d) deciding on the presence or stage of medical condition or the responsiveness of a subject to a therapy against said medical condition, preferably cancer, more preferably ovarian cancer, based on the results obtained in step (c).

In a preferred embodiment of the present invention, the assay as described herein above may comprises the additional step of testing in a sample obtained from a subject for the methylation state and/or pattern of a stratifying biomedical markers or group of said markers and/or of a network node and/or high ranking network member or group of said nodes or members, wherein said members can be obtained according to the above described method; alternatively, the testing may be carried out with a marker or group of markers as defined herein above; wherein in step (c) additionally the difference in methylation status and/or pattern is determined.

Alternatively an assay for detecting, diagnosing, graduating, monitoring or prognosticating a medical condition, preferably cancer, more preferably ovarian cancer as defined herein above, may comprise in step (a) the determination of a different parameter in a sample obtained from a subject. Non limiting examples of such parameters are whole genome sequences, genomic methylation pattern, the identity of methylated sections or elements, the molecular state of a gene or genomic locus, the presence or absence or amount/level of transcripts, proteins, truncated transcripts, truncated proteins, the presence or absence or amount/level of cellular markers, the presence or absence or amount/level of surface markers, the presence or absence or amount/level of glycosylation pattern, the form of said pattern, the presence or absence of expression pattern on mRNA or protein level, the form of said pattern, cell sizes, cell behavior, growth and environmental stimuli responses, motility, the presence or absence or amount/level of histological parameters, staining behavior, the presence or absence or amount/level of biochemical or chemical markers, e.g. peptides, secondary metabolites, small molecules, the presence or absence or amount/level of transcription factors, the form and/or activity of chromosomal regions or loci; and the presence or absence of further biochemical or genetic markers, e.g. the expression or methylation of genes or markers not comprised in Table 1, 2, 4 or 5, or any combination thereof.

Alternatively, an assay for detecting, diagnosing, graduating, monitoring or prognosticating the responsiveness of a subject to a therapy against said medical condition, preferably cancer, more preferably ovarian cancer, even more preferably the responsiveness of a subject to a platinum drug based therapy, e.g. carboplatinum, may comprise in step (a) the determination of a different parameter in a sample obtained from a subject. Non limiting examples of such parameters are whole genome sequences, genomic methylation pattern, the identity of methylated sections or elements, the molecular state of a gene or genomic locus, the presence or absence or amount/level of transcripts, proteins, truncated transcripts, truncated proteins, the presence or absence or amount/level of cellular markers, the presence or absence or amount/level of surface markers, the presence or absence or amount/level of glycosylation pattern, the form of said pattern, the presence or absence of expression pattern on mR A or protein level, the form of said pattern, cell sizes, cell behavior, growth and environmental stimuli responses, motility, the presence or absence or amount/level of histological parameters, staining behavior, the presence or absence or amount/level of biochemical or chemical markers, e.g. peptides, secondary metabolites, small molecules, the presence or absence or amount/level of transcription factors, the form and/or activity of chromosomal regions or loci; and the presence or absence of further biochemical or genetic markers, e.g. the expression or methylation of genes or markers not comprised in Table 1, 2, 4 or 5, or any combination thereof.

In a further specific embodiment the expression may be tested by any suitable means known to the person skilled in the art, preferably by room temperature polymerase chain reaction (RT-PCR), RNA sequencing, or gene expression detection on microarrays.

In yet another specific embodiment the methylation state or methylation pattern may be determined by using methylation specific PCR (MSP), bisulfite sequencing, the employment of microarray techniques, direct sequencing, such as, for example, implemented by Pacific Biosciences(R).

In yet another aspect the invention relates to a method for classifying a subject comprising:

(a) providing a subject's dataset comprising data on methylated loci and data on gene expression of a stratifying biomedical marker or group of said markers and/or of a network node and/or high ranking network member or group of said nodes or members wherein said members can be obtained according to the above described method;

alternatively, the dataset may be based on a marker or group of markers as defined herein above;

(b) accessing a database comprising database values for a stratifying biomedical marker or group of said markers and/or of a network node and/or high ranking network member or group of said nodes or members wherein said members can be obtained according to the above described method; alternatively, the database may be accessed for a marker or group of markers as defined herein above;

In a preferred embodiment the dataset to be provided from a subject may comprise data on methylated loci linked to the marker or group of markers defined above, e.g. at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers selected from PKMYT1, SKIL, RAB8A, HIRIP3, CTNNB1, NGFR, ZCCHC11, LSP1, CD200, PAX8, CYBRD1, HOXC11, TCEAL1, FZD10, FZD1, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR, or the markers or group of markers mentioned in Table 2, 4 or 5.

In a further specific embodiment of the present invention preferred methylated loci linked to the marker or group of markers of the present invention, or being located in the vicinity of the marker or group of markers of the present invention, in particular the markers of Table 1, or loci linked to or being located in the vicinity of said markers which are preferably consulted, analysed (e.g. via methylation detection means as defined herein) or tested in order to obtain datasets from a subject, are indicated in the following table which provides genomic coordinates linked to the markers comprised in Table 1 :

Chromosome Start End Closest Gene

¹ 52730331 52730552 ZCCHC11

¹ 52730611 52730733 ZCCHC11

¹ 52730734 52730888 ZCCHC11

¹ 52731025 52731215 ZCCHC11

¹ 52731315 52731455 ZCCHC11

¹ 52731456 52732425 ZCCHC11

¹ 115343699 115343965 TSPAN2

¹ 115344225 115344328 TSPAN2

¹ 115344465 115344664 TSPAN2

2 113750708 113750905 PAX8

2 113751138 113751290 PAX8

2 113751695 113751825 PAX8

2 113751826 113751957 PAX8

2 113751981 113752204 PAX8

2 201808587 201808903 CFLAR

2 172204489 172204729 CYBRD1 2 172204730 172204867 CYBRD1

2 172204868 172204983 CYBRD1

2 172205161 172205286 CYBRD1

2 172205287 172205410 CYBRD1

2 172205411 172205550 CYBRD1

2 201806500 201806648 CFLAR

2 201806881 201806995 CFLAR

2 201807105 201808144 CFLAR

3 41213549 41215233 CT NB1

3 41215292 41215399 CT NB1

3 41215404 41215527 CT NB1

3 41215828 41215935 CT NB1

3 41216038 41216194 CT NB1

3 113534054 113534772 CD200

3 113534878 113535030 CD200

3 171557568 171557718 SKIL

3 171557857 171557961 SKIL

3 171558082 171558200 SKIL

3 171558263 171558378 SKIL

3 171558697 171558855 SKIL

5 170666547 170667821 TLX3

5 170667889 170668074 TLX3

5 170668379 170668538 TLX3

5 170668544 170668736 TLX3

5 170668737 170668894 TLX3

5 170668895 170669053 TLX3

5 170669129 170669238 TLX3

5 170669265 170669473 TLX3

5 170669618 170669721 TLX3

5 170669860 170669985 TLX3

5 170670020 170670447 TLX3

5 170670448 170670599 TLX3 7 90537970 90538274 FZD1

7 90538398 90538569 FZD1

7 90539178 90539501 FZD1

7 90539515 90539633 FZD1

7 90539644 90539897 FZD1

7 90539959 90540209 FZD1

7 90540210 90540369 FZD1

7 90540370 90540529 FZD1

9 110096997 110097425 TXN

9 110097512 110097666 TXN

9 110097667 110097868 TXN

9 110097934 110098215 TXN

11 1848448 1848761 LSP1

11 1848762 1849063 LSP1

11 1849064 1849183 LSP1

12 52652610 52653249 HOXC11

12 52653275 52653497 HOXC11

12 52653596 52653696 HOXC11

12 52653697 52653807 HOXC11

12 52653990 52654219 HOXC11

12 52654220 52654341 HOXC11

12 52654342 52654469 HOXC11

12 52654470 52654651 HOXC11

12 52654768 52655073 HOXC11

12 52655173 52655281 HOXC11

12 129169880 129170207 FZD10

12 129170283 129170392 FZD10

12 129170525 129170816 FZD10

12 129170909 129171018 FZD10

12 129171257 129171376 FZD10

12 129171377 129171494 FZD10

12 129171997 129172097 FZD10 12 129172453 129172685 FZD10

12 129173006 129173124 FZD10

12 129173728 129173996 FZD10

13 109234815 109234915 IRS2

13 109235600 109235866 IRS2

13 109235951 109236241 IRS2

13 109236242 109236365 IRS2

13 109236896 109237128 IRS2

13 109237235 109237354 IRS2

13 109237454 109237562 IRS2

13 109237563 109237850 IRS2

13 109237890 109238485 IRS2

15 70765350 70765593 BBS4

15 70765674 70765798 BBS4

16 2969849 2969984 PKMYT1

16 2970065 2970187 PKMYT1

16 2970188 2970331 PKMYT1

16 2970620 2970773 PKMYT1

16 2970867 2971519 PKMYT1

16 29913959 29914151 HIRIP3

16 29914157 29914370 HIRIP3

16 29914439 29914684 HIRIP3

16 29914685 29914964 HIRIP3

17 44927437 44927578 NGFR

17 44927579 44927679 NGFR

17 44927680 44927795 NGFR

17 44928073 44928180 NGFR

17 44928181 44928315 NGFR

17 44928392 44928662 NGFR

19 16083047 16083266 RAB8A

19 16083371 16083493 RAB8A

19 16083525 16083845 RAB8A 19 16083874 16084520 RAB8A

23 102690660 102690795 TCEAL1

11 1831415 1831798 LSP1

In a particularly preferred embodiment the above indicated genomic regions, or sections comprising said regions, e.g. sections differing by about 10 nt to about 5 kb from the indicated regions, may be analysed for the presence of methylated nucleotides, the presence or absence of methylation pattern etc. These may preferably be correlated with corresponding expression data of one or more of the markers or group of markers indicated in Table 1. These regions may be analysed separately, or in combination, e.g. for each marker all indicated regions or a sub-set thereof may be analysed. Furthermore, any combination of regions for more than one marker may be analysed.

Alternatively, the dataset may comprise data on further parameters linked to the marker or group of markers defined above, e.g. at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers selected from PKMYT1, SKIL, RAB8A, HIRIP3, CTNNB 1 , NGFR, ZCCHC11, LSP1, CD200, PAX8, CYBRD1, HOXC11, TCEAL1, FZD10, FZD1, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR, or the markers or group of markers mentioned in Table 2, 4 or 5. Non limiting examples of such parameters are whole genome sequences, genomic methylation pattern, the identity of methylated sections or elements, the molecular state of a gene or genomic locus, the presence or absence or amount/level of transcripts, proteins, truncated transcripts, truncated proteins, the presence or absence or amount/level of cellular markers, the presence or absence or amount/level of surface markers, the presence or absence or amount/level of glycosylation pattern, the form of said pattern, the presence or absence of expression pattern on mRNA or protein level, the form of said pattern, cell sizes, cell behavior, growth and environmental stimuli responses, motility, the presence or absence or amount/level of histological parameters, staining behavior, the presence or absence or amount/level of biochemical or chemical markers, e.g. peptides, secondary metabolites, small molecules, the presence or absence or amount/level of transcription factors, the form and/or activity of chromosomal regions or loci; and the presence or absence of further biochemical or genetic markers, e.g. the expression or methylation of genes or markers not comprised in Table 1, 2, 4 or 5, or any combination thereof. In consequence, the method also may include a step of accessing a database comprising database values for the marker or group of markers defined above, e.g. at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers selected from PKMYT1, SKIL, RAB8A, HIRIP3, CTNNB1, NGFR, ZCCHC11, LSP1, CD200, PAX8, CYBRD1, HOXC11, TCEAL1, FZD10, FZD1, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR, or the markers or group of markers mentioned in Table 4 or 5. Furthermore, the method may alternatively include a step of accessing a database comprising database values with regard to further parameters linked to the marker or group of markers defined above, e.g. at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers selected from PKMYT1, SKIL, RAB8A, HIRIP3, CTNNB1, NGFR, ZCCHC11, LSP1, CD200, PAX8, CYBRD1, HOXC11, TCEAL1, FZD10, FZD1, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR, or the markers or group of markers mentioned in Table 4 or 5. Non limiting examples of such parameters are whole genome sequences, genomic

methylation pattern, the identity of methylated sections or elements, the molecular state of a gene or genomic locus, the presence or absence or amount/level of transcripts, proteins, truncated transcripts, truncated proteins, the presence or absence or amount/level of cellular markers, the presence or absence or amount/level of surface markers, the presence or absence or amount/level of glycosylation pattern, the form of said pattern, the presence or absence of expression pattern on mRNA or protein level, the form of said pattern, cell sizes, cell behavior, growth and environmental stimuli responses, motility, the presence or absence or amount/level of histological parameters, staining behavior, the presence or absence or amount/level of biochemical or chemical markers, e.g. peptides, secondary metabolites, small molecules, the presence or absence or amount/level of transcription factors, the form and/or activity of chromosomal regions or loci; and the presence or absence of further biochemical or genetic markers, e.g. the expression or methylation of genes or markers not comprised in Table 1, 2, 4 or 5, or any combination thereof.

In yet another aspect the present invention relates to a medical decision support system comprising:

an input for providing a subject dataset comprising data on methylated loci and data on gene expression of a stratifying biomedical marker or group of said markers and/or of a network node and/or high ranking network member or group of said nodes, wherein said members can be obtained according to the above described method;

alternatively, the dataset may be based on a marker or group of markers as defined herein above; a computer program product for enabling a processor to carry out the method for classifying a subject as defined above, and

an output for outputting the subject classification score.

In a preferred embodiment the dataset to be used as input may comprise data on methylated loci linked to or derived from the marker or group of markers defined above, e.g. at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers selected from PKMYT1, SKIL, RAB8A, HIRIP3, CTNNB1, NGFR, ZCCHC11, LSP1, CD200, PAX8, CYBRD1, HOXC11, TCEAL1, FZD10, FZD1, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR, or the markers or group of markers mentioned in Table 4 or 5. E.g. a subject to be tested may specifically be tested for one or more of the mentioned markers, or the group of markers as defined above.

In a specific embodiment said medical decision support system may be a molecular oncology decision making workstation. The decision making workstation may preferably be used for deciding on the initiation and/or continuation of a cancer therapy for a subject. More preferably, the decision making workstation may be used for deciding on the probability and likelihood of responsiveness to a platinum based therapy.

In a further aspect the present invention also envisages a software or computer program to be used on a decision making workstation. The software may, for example, be based on the analysis of datasets or data linked to the marker or group of markers defined above, e.g. at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers selected fromPKMYTl, SKIL, RAB8A, HIRIP3, CTNNB 1 , NGFR, ZCCHC11, LSP1, CD200, PAX8, CYBRD1, HOXC11, TCEAL1, FZD10, FZD1, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR, or the markers or group of markers mentioned in Table 2, 4 or 5. The following examples and figures are provided for illustrative purposes. It is thus understood that the example and figures are not to be construed as limiting. The skilled person in the art will clearly be able to envisage further modifications of the principles laid out herein. EXAMPLES

Example 1 - General methodology

Initially, genome-wide DNA methylation data of ovarian cancer patients was obtained. Methylation Oligonucleotide Microarray Analysis (MOMA) was used to perform genome-wide scans of CpG island methylation in normal and tumor samples according to Kamalakaran et al., Nucleic Acids Res (12): e89, 2009, which is incorporated herein in its entirety.

Patient samples were categorized in two groups: platinum-resistant patients have PFI (platinum free interval) of < 6 months (12 patients) and platinum-sensitive patients have PFI > 24 months (13 patients).

Methylation probes were filtered based on the size of target fragments and intensity to retain -190,000 probes out of the original 330,000 probes in MOMA covering 27,000 CpG islands in the human genome.

Linear models with Bayesian statistics and leave-one-out cross-validation were used to find statistically significant and robust stratifying probes/loci. A list of 749 candidate loci that stratify resistant and sensitive patients could be derived, which served as input to further pathway and network analysis. Fig. 1 summarizes the steps of the procedure.

The genes in the proximity of the candidate loci were used to identify the most significant pathways using the pathway analysis tool in GeneSpring GX11. The tool takes a list of entities (e.g. gene symbols) as an input and finds all pathways from a collection which have significant overlap with that entity list. The set of pathways used in this analysis was imported from the BioPAx, KEGG repositories (www. biopax.org;

www.genome.jp/kegg/pathway.htm). Here, overlap denotes the number of common entities between the list and the pathway. Commonness is determined via the presence of a shared identifier, i.e., Entrez Gene ID. Once the number of common entities is determined, the p- value computation for a pathway is based on the Hypergeometric method (or Fisher's exact test).

Separately, as input gene expression profiles of the genes proximal to the stratifying methylation probes using Affymetrix (H GUI 33a) were taken. Based on the two distinct measurements, a weighted methylation-expression matrix was constructed.

Methylation profiles of the unique stratifying genes were correlated to the expression profiles. The methylation-expression similarity Sy between genes i and j are defined as the absolute value of the correlation coefficient between their expression and methylation profiles according to the formula: f \ I

½^' 1 I ■ i * To obtain a threshold value to select the significant correlations, the methylation profiles were permuted 100 times and 100 methylation-expression correlation matrices were constructed.

Based on the distribution of all correlation values, a hard threshold value of 0.58 (99^th percentile) was used to identify the most highly correlated and anti-correlated connections. There were multiple probes (among the 749) representing the same gene and some probes were not present in the Affymetrix data. Ultimately, there were 263 unique genes used in the correlation matrix.

Subsequently a network graph was constructed with genes as nodes and the presence of an edge for all i and j where Sy> 0.58 was defined. A directed edge between i and j indicates the correlation of the methylation profile of i to expression of / (not expression to methylation). Network centrality measure of node betweenness was used to identify key genes. Nodes that occur on many shortest paths between other nodes have higher

betweenness than those that do not. High betweenness of a node indicates that a gene has high influence over the information flow between other genes in the network. Other network metrics computed include: connectivity (number of genes adjacent to a gene) and eccentricity (accessibility of a gene by all other genes in the network).

The network graph based on the correlation matrix was analyzed in Pajek (further details are described in Nooy et al., Exploratory Social Network Analysis with Pajek, Cambridge University Press, 2005, which is incorporated herein in its entirety), a network analysis and visualization software.

Example 2 - Assessment of overrepresentation of biological processes

Based on a network as described in Example 1, it is also possible to assess overrepresentation of biological processes or molecular functions in a selected network subset. By selecting a sub network of interconnected genes that are connected and known to be highly ranked by one of the network metrics, it is possible to determine the

overrepresentation of categories given by Gene Ontology. The steps taken to characterize the enrichment of these sub-networks involve:

Importing the network into Cytoscape (further details are described in

Shannon et al., Cytoscape: a software environment for integrated models of bio molecular interactionnetworks, Genome Research, 2003, 13(11): 2498-2504, which is incorporated herein in its entirety);

Importing the network metrics as attributes of the nodes; Selecting a subset of nodes (e.g. top ranked nodes) and edges (e.g. most correlated and anti-correlated nodes here the threshold value is 0.55)

Invoke the BingGO plug-in for Cytoscape (further details are described in Maere et al, BINGO: a Cytoscape plugin to assess overrepresentation of Gene Ontology categories in biological networks, Bio informatics, 2006, 21, 3448, which is incorporated herein in its entirety);

Set the parameters for BinGO;

Select ontology: Molecular Function/Biological Process;

Select annotation (organism): Homo sapiens;

Select statistical test : Hypergeometric test;

Select correction : Benjamini & Hochberg False Discovery Rate (FDR) correction;

Select significance level : 0.05;

Testing option : Test cluster versus whole annotation;

and

Execute overrepresentation analysis.

Example 3 - Pathways in carboplatinum sensitivity

The linear model analysis as described in Example 1 identified a set of 749 probes that differentiates between resistance and sensitivity to platinum based drugs in ovarian cancer. These were subjected to pathway analysis using GeneSpring. Pathways showing significant overlap with genes (entities) in the gene list (entity list) selected for analysis are displayed in Table 2.

Table 2: List of enriched pathways and genes

Pathways P-value

Androgen receptor 0.0016

Pitx2 driven transcription regulation 0.0042

Wnt signaling pathway 0.0135

Gata3 and th2 cytokine gene expression 0.0270

Segmentation clock 0.0297

PI3K-akt 0.0343

Leukocyte transendothelial migration 0.0441

Phosphorylation of mekl by cdk5/p35 0.0441

Gene: Function: Sequence Identifier:

GSK3B glycogen synthase kinase 3 beta (SEQ ID NO: 22)

FZDl frizzled homo log 1 (SEQ ID NO: 15)

CTNNBl WNT Signalling (SEQ ID NO: 5)

COX5B cytochrome c oxidase subunit Vb (SEQ ID NO: 23)

PXN paxilin (SEQ ID NO: 24)

POU2F1 POU class 2 homeobox 1 (SEQ ID NO: 25)

CCNE1 Cyclin El (SEQ ID NO: 26)

TMF1 TATA element modulatory factor 1 (SEQ ID NO: 27)

MAPK1 mitogen-activated protein kinase 1 (SEQ ID NO: 28)

PTEN phosphatase and tensin homo log (SEQ ID NO: 29)

NCOA3 nuclear receptor coactivator 3 (SEQ ID NO: 30)

GATA3 GATA binding protein 3 (SEQ ID NO: 31)

NFATC1 nuclear factor of activated T-cells (SEQ ID NO: 32)

PTX2 paired-like homeodomain 2 (SEQ ID NO: 33)

CCND2 cyclin D2 (SEQ ID NO: 34)

The table also highlights the genes among the pathways important in chemo sensitivity to platinum. Contributions from AR pathway, Wnt pathway and PI3K-akt pathway have been well-characterized in ovarian cancer. Methylated PITX2 has been shown to predict outcome in lymph node-negative breast cancer patients.

In Fig. 2 one of the significant pathways - the Wnt pathway is shown in more detail with the members overlapping with the established list of genes, identified in blue halo: FZD 1 , GSK3B and CTNNB 1. Methylation of another frizzle protein SFRP has been shown to promote ovarian cancer progression and chemoresistance. Suppression of CTNNBl has also been evident in many cancers. An analysis of how genes in a list are connected to each other in a target- regulator relationship based on biologically known interactions can also be revealing. Since this exercise is not focused on one particular pathway it can allow investigating the crosstalk between nodes of different pathways.

In Fig. 3, the target-regulator network based on the established list generated by GeneSpring is shown. Nodes are included in this graph if they have at least one known interaction from biological databases with a gene in our list. The hubs CTNNB1, CCND2 appear to be involved mainly in Wnt signaling pathway. PTEN from p53 signaling pathway is also represented but most interactions appear to have one or two links.

Example 4 - Network structure analysis

A weighted methylation-expression network was constructed as described above. The network centrality measure of betweenness was calculated by noting the shortest paths between all pairs of nodes (see Fig. 4). A large node in this graph corresponds to a gene that is frequently found in shortest paths between gene pairs giving it a high betweenness measure. Edges e(i,j) represent correlation (solid edges) or anti-correlation (dashed edges) between methylation profile of gene i and expression profile of gene j. Table 2 provides, inter alia, information on betweenness, eccentricity and connectivity for the central nodes in Fig. 4.

A majority of these nodes are involved in cancer-related functions or signaling pathways. Some of the central nodes include PKMYT 1 , CTNNB 1 , RAB8A and NGFR. NGFR has low connectivity but ranks higher in betweenness and eccentricity measures because it is traversed in the shortest paths of many pairs of genes. NGFR is known to act via cytokine receptor interactions and is often used as a marker (along with CA125 and p55) for ovarian cancer. Recently, expression of NGFR has also been used as a marker to measure toxicity to carboplatin. It was not identified as an enriched pathway in the GeneSpring analysis most likely because enough members of this pathway were not represented in the list or the pathway databases were incomplete. It is also possible that incorporating expression information solidified its correlation characteristics with other genes. Table 3: Identified central nodes: Annotation of central nodes with their corresponding network measure values: betweenness BTW (σ= 2.8e-5; p=0.00013), eccentricity ECC (σ=0.59 [-1.72, 1.72]; p=1.5), and connectivity CON (σ=1.07; p=1.8) Node SEQ ID NO

Description nucleotide BTW ECC CON

(Gene Name) sequence

Serine threonine

PKMYT1 1 0.00088 1.48 7

kinase

Oncogene; 0.00087

SKIL 2 1.71 4

SMAD signaling 7

A RAS

0.00086

RAB8A oncogene; GTP 3 1.48 18

6

binding;

Histone 0.00085

HIRIP3 4 1.42 5 repressor 9

0.00071

CTNNB1 WNT Signalling 5 1.66 3

2

P75; cytokine

0.00066

NGFR receptor 6 1.52 2

1

interaction

0.00057

ZCCHC11 Zinc finger 7 1.50 4

0

Lymphocyte

LSP1 specific protein; 8 0.00035 1.63 5 signal transducer

CD200 glycoprotein 9 0.0002 1.68 3

Paired box

PAX8 transcription 10 0.00019 1.62 2 factor

As can be derived from Fig. 4, RAB8A, a member of the RAS oncogene family, is highly connected. Over-expression of another member of this family, RAB25, has been associated with increased proliferation and aggressiveness in breast and ovarian cancer. Although RAB8A was not identified in the GeneSpring approach, network measures and related literature search suggest that it is likely to have an important role. A directed edge between RAB8A and NGFR implies that the methylation profile of RAB8A and the expression profile of NGFR are highly correlated.

By using the methylation-expression correlation matrix in this analysis, we were able to identify multi-modal associations that are of importance in chemo sensitivity. For example, a patient can be predicted sensitive to chemotherapy based on methylated RAB8A and under expressed NGFR. So, the combined analysis based on two data types in some sense reorders the larger list of stratifying genes by incorporation of methylation and expression data with network topological measures. Example 5 - Analysis of overrepresentation of gene ontology categories

Analysis of overrepresentation of gene ontology categories in the correlation sub-networks (interconnected genes that are connected and known to be highly ranked by one of the network metrics) identified relevant biological processes or molecular functions.

Tables 4 and 5, below, represent the results of this analysis:

INCORPORATED BY REFERENCE (RULE 20.6)

INCORPORATED BY REFERENCE (RULE 20.6) CENTRIC ΠΎ

File creaed with BI GO to)

otology: Sjjpcu n

crrrator: GO

Select d owe logy ilk :

Selected annotation nle

Discarded e idence codes :

Oven'epreserdauon

Selected statistical toss. : Myp rgeoin ine te*†

Selected correction : Benjamlni & ioehberg False Discovery Rate (FDR) correction

Selected significance level ; 0,05

Testing option : Test cluster versus whole annotation

The selected cluster :

C '4 BBS4 RABSA HOXCJ i ZCCHCU CON CYBtUi! B.VRIP3 TCi-AU

No annot tions were retrieved for die following entities:

GO-ID p-vaine corr p~v¾i¾ :¾ n X N BescripSioi*

293 1.3762E-3 2,8437E-2 1 3 7 15253 tenic-cheiate reductase activity

16723 346^'E-3 2.8437E-2 1 4 7 1 253 oxkioreductasc activity., oxidizing metal ions, NAD or NADP as acceptor

16722 3.2087EG 3..NS7C2 i 7 7 15253 oxidoreduetase activity, oxidizing meiai !orss

3702 5.7876E-3 4.4S54E-2 2 261 7 ί 5253 SNA poly-neno-e ii n seri U n Goto- .n:ti H)

INCORPORATED BY REFERENCE (RULE 20.6)

INCORPORATED BY REFERENCE (RULE 20.6) Table S: Cytoscope Bingo results on biological prooe

ALL

File crested with BiNGO to)

ontology: process

curator: 00

Selected ontology tile : jar:file:C:\Program Files\Cykr:cape v2 /i;\ ' n;AGOkrizy2- jar! GO Bi iogicst _Pfo.es*·

Selec d annotation fik : ja; : i!e:C:VPrograr.;. FlieACytoscepe v2.7 ) iugk^¾GOfori22-4j8r! H_msa ieas___deau

Discarded evidence codes :

Overrepreseoiatioa

Selected statistical test : Hypergeornetric test

Selected correction : Benjamin? & Hochberg False Discovery Rate (FDR) correction

Selected significance level : 0.05

Testing option : Test cluster versus whole annotation

The selected duster

RARRR¾AO¾ CDX;? O PNABSR^': RFXAJP CTKNB1 CDH22 S OX SUPRi SSTR4 RND3 R!A Ri AQR SPAG2 RSHLi MA K7 COASY F

PCDH2- OAS3 ?X PIN! ·/!·:>; FPA.X CRivtPl ^• ; :X i DULLARD LHX6 FBXO« RLF6 ESRRA YiKK7 TP?3BR2 RiJ si; ; OXSRi

: C DU3?¾> CNTH3 HY-pjA ^'RRR2 CD200 6NY2 ί.·Ό\Α^·.'. YiARCiCsP; TRX 1 PTEN Rf.i CONPR ^:;A N.R ΛΤΡ30 ;o'.XB€UF

GTF2B iXHBB SE V: IX;L2 H6PD CORD;: TXN CYBRD5 MOSPDJ .?NF4&7 TRi l? uxo UGOH UROS HiX>PRP3 HX1 TLX 3 P o t;.o Π3Χ4 ?ZD? ATP I A3 YRTR? CABYR LSPi PAPOLA LARGE NPLR002 ARSA ··;!.< R3A? P,BVR<; TOM TSPAR2 ΊΉΚΒ P3.03A3

ΡΚ, ΫΤ! ZiC! TCEAJJ NR3Fi COX?B f'KS IDHGG SLC2RA2.; MCOLN^■ 1-HR1P3 LO.X FBXOR.2 iDUA SNRPAi ;\;.UO6A; s 3

BALAP.^'; iiDACU ΗΒΧΪΡ ···^■ <:^■. BCAR ! TRRS2 HHR A. CL P !PA ; BR S3 COB Li. DRD: SOX? STSSlAt Τ«νίΡ4 Α2 ^ήΥί.32 SOXQ

ΟΛΖί JUNO ^••• .ίί. FGD6 GABRD XPOT OA8ARAPL2 CFLAR RA8K SAP 18 SLC25A10 SALLl T5P30 FEXGR3 ZKFRi- IMPAJ

CL0 3 CTCF ARPN TM i GPO-i PCOF2 GPC3 PAXS SOX 18 RiT3 SLC43A3 PPRMKS PToKPO XFY PRRO: CHST3 DDR TPR42

DHR3 ADFP S T DGA^'i^'f HOXCU R P! NCOA3 ye k U8B CKK3 at; ^;O NR3C? ·;;;·· ^":.: ALDH.1A2 DPP! CBHS F3C22D; S

TSC22D ;sr ::! « CBNPB ND/H ;^■';?··; R HSfisTi EHD3 NFAYCl AU'R■ CDW RACGAP i. tYRT3B t.RP DiO RPAP;

No annotations were i¾trieved for the following entities:

PNY2 XCCHCU K;DN RIARC SLI >·: <;··;· DDN CKR RPAP! ADFP SPAG7 RSFR.i DUIi.ARD SYN.i2 MG3PDS CO^": H;iD3 RL¾ i

3LG-3A3

INCORPORATED BY REFERENCE (RULE 20.6)

INCORPORATED BY REFERENCE (RULE 20.6)

INCORPORATED BY REFERENCE (RULE 20.6)

:· fsi

INCORPORATED BY REFERENCE (RULE 20.6)

INCORPORATED BY REFERENCE (RULE 20.6)

INCORPORATED BY REFERENCE (RULE 20.6)

INCORPORATED BY REFERENCE (RULE 20.6)

INCORPORATED BY REFERENCE (RULE 20.6)

INCORPORATED BY REFERENCE (RULE 20.6) 18 3 3.0G64E-3 3. 673E-2 I 14 3 13937 oeural t be closure

14020 3.0064E-3 3.0673E -2 1 E; 3 B952 prboary neural tubs ionnatiou

o

o 1679 3.6499F -3 3..067m-2 ί 1 ? 5 13957 neural tobe formation

73

~0

O 1838 3.8643E-3 3.0673 E- 2 i 1 8 3 13957 embryonic epithelial tube formation

73

> 1 39 3.8643E-73 3.0673E-2 i 1 8 3 13957 neural la morphogenesis

m

D 16358 393643 Ε··3 3.0673 E-2 1 18 3 13957 dendrite deveiopmeoi

< 2191 5 33¾643E~3 3 0675E~2 1 18 3 1 957 neural tube development

73

m 1894 4.293 ! E-3 3.2072F-2 I 20 ; 1395? ussoe ho eostasis

^"Π

m

73 16331 5.1502E-3 3.4064E42 1 24 3 1395? morphogenesis of embryonic epithelium

m

z 1 25 5.3644E-3 3.4064E-2 1 25 3 1395? regulation of lipid metabolism

O

m 30534 5.3 44E-3 3.4064 E-7! 1 25 3 13957 uduk behavior

51242 7.4287E-3 4.4926E-2 2 707 1 13957 positive regulation, of cellular physiological process m 3003 ! 7.932SE-3 4.4982E-2 1 37 3 13957 ce 11 proj ee i:i OH b sogenes i s

o

cn 8624 ¾.1 63E~3 4.4982E-2 1 33 3 13957 induction of apoptosis by extracellular signals CO.

431 19 9.2157E-3 4.8766E-2 2 789 3 1395? positive regulation of physiological process

According to the results presented in Tables 4 and 5, above the following genes were identified as highlighted, i.e. very relevant:

1) Molecular Function: (cytoscape bingo MF corr 0.55.docx; Table 4)

ALL (a special case of the analysis where all nodes with highly (anti) correlated edges are selected regardless of their network metrics values). In this group a large number of genes is highlighted, corresponding to the detection of overrepresentation of transcription regulation and DNA binding functions (see Table 4, "All").

CENTRICITY. In this group the genes CYBRD1, HOXC11, and TCEAL1 are highlighted.

CLOSENESS. In this group the genes FZD10 and FZD1 are highlighted.

HIGHCONN. In this group a large number of genes is highlighted, corresponding to the detection of overrepresentation of transcription activation, repression and receptor binding.

2) Biological Process: (cytoscape bingo BP corr 0.55.docx; Table 5)

ALL (a special case of the analysis where all nodes with highly (anti) correlated edges are selected regardless of their network metrics values). In this group a large number of genes is highlighted, corresponding to the detection of overrepresentation of regulatory and developmental processes (see Table 5, "All").

CENTRICITY. In this group gene BBS4 is highlighted.

CLOSENESS. In this group genes CTNNB1, IRS2, LSP1, NGFR, TLX3, TSPAN2, TXN are highlighted, corresponding to the detection of overrepresentation of cell motility localization, and fate.

ECCENTRICITY. In this group genes CFLAR and BBS4 are highlighted, corresponding to the detection of overrepresentation of regulation of biogenesis.

Claims

CLAIMS:

1. A method for identifying multi-modal associations between biomedical markers, comprising the steps of:

assigning a ranking score to the members of the identified network said ranking score being based upon a network metric;

2. The method of claim 1, wherein said plurality of datasets on multiple molecular profiling modalities from a plurality of primary and secondary subjects comprises data on methylated loci and on gene expression.

3. The method of claim 1, wherein said step of identifying a network and/or subnetwork comprises the calculation of significance values for the stratifying biomedical markers.

4. The method of claim 1, wherein said network metric comprises at least one element selected from the group of connectivity, adjacency, network density, network centralization, network heterogeneity, cliquishness, hub gene significance, network significance, centroid significance, centroid conformity, betweenness, centricity, closeness and eccentricity.

5. The method of claim 1, wherein said biomedical marker is a gene, a genomic locus in a coding region, a genomic locus in a non-coding region, a transcript and/or a protein.

6. The method of claim 1, wherein said primary subject is a healthy subject and said secondary subject is affected by a medical condition.

7. The method of claim 1 or 6, wherein said medical condition is cancer.

8. The method of claim 7, wherein said cancer is ovarian cancer.

9. The method of claim 8, wherein said predictive value is the indication of likelihood of responsiveness of a subject to a therapy comprising one or more platinum based drugs.

10. The method of claim 9, wherein said platinum based drug is carboplatinum.

11. A biomedical marker or group of biomedical markers associated with a high likelihood of responsiveness of a subject to a cancer therapy, preferably a platinum based cancer therapy, wherein said biomedical marker or group of biomedical markers comprises at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or all markers selected from PKMYTl, SKIL, RAB8A, HIRIP3, CTNNBl, NGFR, ZCCHCl l, LSPl, CD200, PAX8, CYBRDl, HOXCl l, TCEALl, FZD10, FZDl, BBS4, IRS2, TLX3, TSPAN2, TXN, and CFLAR, indicated in Table 1.

12. An assay for detecting, diagnosing, graduating, monitoring or prognosticating a medical condition, or for detecting, diagnosing, monitoring or prognosticating the responsiveness of a subject to a therapy against said medical condition, preferably cancer, more preferably ovarian cancer, comprising at least the steps of

(a) testing in a sample obtained from a subject for the expression of a stratifying biomedical markers or group of said markers and/or of a network node and/or high ranking network member or group of said nodes or members obtained by the method claim 1 or defined in claim 11 ;

(c) determining the difference in expression of markers of steps (a) and

(b); and

(d) deciding on the presence or stage of medical condition or the responsiveness of a subject to a therapy against said medical condition, preferably cancer, more preferably ovarian cancer, based on the results obtained in step (c).

13. The assay of claim 12, wherein said assay comprises the additional step of testing in a sample obtained from a subject for the methylation state and/or pattern of a stratifying biomedical markers or group of said markers and/or of a network node and/or high ranking network member or group of said nodes or members obtained by the method claim 1 or defined in claim 11 , wherein in step (c) additionally the difference in methylation state and/or pattern is determined.

14. A method for classifying a subject comprising:

(a) providing a subject's dataset comprising data on methylated loci and data on gene expression of a stratifying biomedical marker or group of said markers and/or of a network node and/or high ranking network member or group of said nodes or members obtained by the method claim 1 or defined in claim 11;

(b) accessing a database comprising database values for a stratifying biomedical marker or group of said markers and/or of a network node and/or high ranking network member or group of said nodes or members obtained by the method claim 1 or defined in claim 11 ;

15. A medical decision support system comprising:

an input for providing a subject dataset comprising data on methylated loci and data on gene expression of a stratifying biomedical marker or group of said markers and/or of a network node and/or high ranking network member or group of said nodes or members obtained by the method claim 1 or defined in claim 11;

a computer program product for enabling a processor to carry out the method of claim 14, and

an output for outputting the subject classification score.