WO2009015799A2 - Detection of platinum resistance - Google Patents

Abstract

The invention relates to a method for classifying a patient as platinum-sensitive or platinum-resistant, the use of the expression profile of a subgroup of genes for classifying a patient as platinum-sensitive or platinum-resistant, and a method for identifying informative genes in order to be able to classify a patient as platinum-sensitive or platinum-resistant.

Description

 Detection of platinum resistance

The present invention relates to a method of classifying a patient as platinum-sensitive or platinum-resistant, to using the expression profile of a subset of genes to classify a patient as platinum-sensitive or platinum-resistant, and to a method of identifying informative genes to classify a patient as platinum-sensitive or platinum-resistant ,

Targeted drug treatment of tumor diseases remains one of the greatest challenges facing modern drug discovery despite intensive research activities. For example, the most important group of therapeutically used antitumoral substances continues to be the so-called Cytostatic agents that act in different ways toxic to the body's own cells and so inhibit cell growth and division. Important cytostatics are based on the ingredient platinum, such as the Cisplatin ^® , Eloxatin ^® (Oxaliplatin) or Carboplat ^® (Carboplatin).

The main areas of application for platinum-based cytostatic agents are ovarian carcinoma, testicular, bronchial, bladder and cervical carcinomas and squamous cell carcinomas of the head and neck. Platinum-based cytostatic drugs are usually given as an infusion and often used in combination with other chemotherapeutic agents.

Ovarian cancer, or ovarian cancer, is one of the most common malignancies in women. The incidence is currently 23 per one hundred thousand. In 2000, approximately 9,671 women were diagnosed with ovarian cancer in Germany for the first time; Source: Robert Koch Institute Berlin. In the USA, about 20,000 new cases were expected in 2006, ie about 3% of all tumors in women, and about 15,300 deaths; Source: American Cancer Society: Cancer Facts and Figures 2006, Atlanta, GA: American Cancer Society 2006.

Approximately 65% of ovarian cancers are detected only in advanced stages, leading to a poor prognosis.

The standard surgical treatment of ovarian cancer is hysterectomy with adnexectomy, omentectomy, pelvic and para-aortic lymphadenectomy.

With a few exceptions, all patients receive platinum-based chemotherapy, eg with carboplatin and another non-platinum-based chemotherapy drug, such as paclitaxel (Taxol®). Despite this treatment, 20-30% of patients experience no clinical response due to platinum resistance, and the majority of women treated relapse.

To date, there are no histopathological parameters indicating platinum resistance. The only test that can predict such platinum resistance is the ATP chemosensitivity assay. However, a minimum number of tumor cells is necessary for the performance of this test, which often can not be obtained with small tumors. The ATP chemosensitivity assay is very complicated to carry out and difficult to integrate into routine diagnostics. Furthermore, this assay is less specific, the specificity is about 40%.

Against this background, the provision of a method by which platinum resistance can be detected is of great interest. Indeed, the administration of platinum-based chemotherapeutics leads to a variety of side effects that burden the body. Noteworthy here are dose-limiting kidney damage, which shows up just 2 weeks after the start of therapy. Frequently, hearing loss occurs with respect to higher frequencies, especially in children. In addition, repeated neuropathies with paresthesia, convulsions and loss of motor functions are observed with repeated administration. Occasionally, the occurrence of anaphylactoid reactions is detected.

In a reliable prediction of platinum resistance these strains of the organism could be avoided and early, for example, platinum-free chemotherapeutic agents are administered.

De Smet et al. (2006), Predicting the clinical behavior of ovarian cancer from gene expression profiles, Int. J. Gynecol. Cancer 16 (Suppl. 1), pages 147-151, describes that platinum-resistant and platinum-sensitive patients suffering from an ovarian carcinoma have a different gene expression pattern. Specifically, the authors identified 500 different genes that are platinum-resistant Women are expressed differently than in platinum-sensitive women. The authors established two reference groups, one representing gene expression in platinum-resistant women, and one reference group representing gene expression in platinum-sensitive women. The possibility of classifying patients as platinum-sensitive or platinum-sensitive prior to the initiation of chemotherapy because of similarities to one or the other reference group is considered. A disadvantage of the De Smet et al. proposed approach is that the predictive certainty of platinum resistance is very low. The authors have been inaccurate in compiling the reference groups. For example, in the evaluation of the reference groups, two tumor samples were used which were obtained from patients who were treated with the alkylanthene cyclophosphamide and a platinum-containing cytostatic instead of a combination of a platinum-containing cytostatic and paclitaxel. The group of tumor samples used to evaluate the reference group of platinum-resistant women contained two samples taken during chemotherapy. However, it was not possible to predict exactly whether these two patients were at least partially sensitive to platinum-based chemotherapy. In addition, the authors provide only 500 differentially expressed genes, which they claim to contain false positive genes and false negative genes are missing in this list. Thus, the list is not a true "predictor set," ie, a smaller number of genes that allow reliable prediction. These disadvantages lead to a classification accuracy of only 76.92%.

Helleman et al., (2006), Molecular profiling of platinum resistant ovarian cancer, Int. J. Cancer 118, pages 193-1971, describe a method of classifying patients suffering from ovarian carcinoma into a platinum-resistant and a platinum-sensitive group. This classification should be based on the different expression of a predictor set consisting of nine identified genes. The authors have established different reference groups, with one patient classified into one or the other of these reference groups on the basis of her expression profile of the nine identified genes and thus classified as platinum-sensitive or platinum-sensitive. sistent. However, the authors have put together the reference groups inaccurate and contrary to accepted guidelines, eg the German Society for Gynecology and Obstetrics. In establishing the reference groups, the tumor's response to purine chemotherapy was classified using a method that determined the so-called clinical response. Depending on the size of the tumor foci following therapy, as determined by imaging techniques such as ultrasound, the response rates will become complete response, partial response, stable disease, and disease progression ("progression") divided. In the reference group of platinum-resistant women ("non-responders"), the authors finally summarized 13 patients with progression and one patient who relapsed three months after surgical removal. The remaining patients are platinum-sensitive ("responder", n = 82), although they respond differently to platinum therapy. The authors themselves write that a resistant subpopulation could be present among the tumors that were obtained from patients with a partial response. In patients without disease progression ("no progression") different time intervals are measured. In other words, the establishment of the reference groups was not rigorously pursued by the authors and generally accepted guidelines for the classification of platinum resistance and platinum sensitivity were ignored. The sensitivity of Helleman et al. Therefore, known method is only 89% and the specificity is only 59%.

Against this background, it is an object of the present invention to provide a method for classifying a patient as platinum-sensitive or platinum-resistant, which largely avoids the disadvantages of the prior art. In particular, such a classification method should be provided which has a high accuracy. This object is achieved by a method comprising the following steps:

1. Providing a biological sample of a patient,

2. determination of the gene expression profile of a subset of genes from the biological sample,

3. Comparison of the gene expression profile a) determined in step (2) with a first reference gene expression profile of the subgroup of genes, wherein the first reference gene expression profile was obtained from platinum-sensitive patients, and b) with a second reference gene expression profile of the subgroup of genes, wherein the second reference gene expression profile is platinum resis - tenten patients was won,

4. Classification of the patient a) as platinum-sensitive when the gene expression profile of the patient is more similar to the first reference gene expression profile, or b) as platinum-resistant when the gene expression profile of the patient is more similar to the second reference gene expression profile, the subset of genes from at least one Gene selected from the group consisting of: PPADPClA (NM_001030059.1), EGFR (NM_005228.3), NKD2 (NM_033120.2), GPM6B (NM_001001994.1), AMMECR1 (NM_001025580.1), LTB4R (NM_181657 .1), BX112399; TUSC1 (NM_001004125.1), MUC15 (NM_145650.2), HMFN0839 (NM_032717.3), SCAP1 (NM_003726.2), XM_498568, CARD4 (NM_006092.1), FTCD (NM_206965.1), XM_371586, LOC400464 (NM_001013670. 1), COMP (NM_000095.2), LOC440686 (NM_001025303.1), COL5A1 (NM_000093.2), HIST1H2BD (NM_138720.1), F2R (NM_001992.2), KIAA1875 (NM_032529.1), HLA-DMB (NM_002118 .3), FOXPl (AK025793), APOCl (NM_001645.3), WBSCR27 (NM_152559.2), CATSPER1 (NM_053054.2), JOSD2 (NM_138334.1), BCAR3 (BX106566), BGN (NM_001711.3), HOXB7 (NM_004502.2), FXYD3 (NM_005971.2), NIPAI (NM_144599.3), GSR (NM_000637.2), CSTI (NM_001898.2), LAT2 (NM_022040.2), HLA-DQBl (NM_002123.2), LSAMP (NM_002338.2), LOC652670 (XM_942247.1), LOC642412 (XM_925931.1), NEK8 (NM_178170.2), CD40 (NM_001250.3), SAMDII (NM_152486.2), MSC ( NM_005098.2), CR607098, PLCXD1 (NM_018390.1), RASELLA (NM_206827.1), COL3A1 (NM_000090.2), FLJ13391 (NM_032181.1), COLI Al (NM_000088.2), ZNF539 (NM_203282.1), REMl (NM_014012.4), EDG3 (NM_005226.2), CSPG2 (NM_004385.2), TYRPl (NM_000550.1).

"Platinum-sensitive" means according to the invention in accordance with the guidelines of the German Society of Gynecology and Obstetrics, that a tumor patient or a tumor patient with such a platinum-containing substance, for example. A cytostatic agent such as cisplatin or carboplatin, is treatable that it within six months no relapse (recurrence) occurs after completion of therapy. By contrast, "platinum-resistant" means according to the invention that a patient with a platinum-containing substance is not appropriately treatable and that a recurrence occurs within six months after completion of the chemotherapy.

A biological sample of a patient according to the invention contains representative genetic material of the patient to be classified biological tissue, for example in the form of DNA and / or RNA, which allows the determination of a gene expression profile. Therefore, a biological sample comprises a biological cell, a tissue or tissue part, an organ or organ part, whereby the sample can be provided in liquid or solid form, also in cryopreserved form. Typical biological material originates from a tumor, such as an ovarian carcinoma, which can be obtained by a surgical procedure in the patient to be classified.

Gene expression profile according to the invention is understood to mean the entirety of the transcriptional activity of a multiplicity of genes of an individual, for example of the patient to be classified. Gene expression profiles can be determined by techniques known to those skilled in the art, for example with the aid of DNA microarrays which are equipped with a representative number of different cDNA molecules, to which the genetic material of the individual, for example with a Marker provided mRNA molecules or cDNA molecules generated therefrom, can hybridize. The hybridization activity indicates which genes are expressed in the individual. Furthermore, a statement about the strength of the expression is possible. The compilation of the hybridization activities of all investigated molecules or transcripts makes it possible to produce a gene expression profile of the patient.

"Similarity" with a reference gene expression profile according to the invention means that the transcriptional activity of one or more genes from the specified subgroup corresponds more closely to the transcriptional activity of the corresponding gene or genes from one reference group than to the transcriptional activity of the corresponding gene or genes from the other reference group. This similarity according to the invention can be determined by means of mathematical methods known in the art. This takes place, for example, via so-called "support vector machines" (SVM). An SVM fits into a mathematical space a multidimensional hyperplane that serves as a separation plane between two different groups or classes. The gene expression profile determined for a patient can be represented by a mathematical vector which occupies a specific position in the mathematical space. Depending on the location of the vector in the multidimensional space with respect to the hyperplane of SVM, that is, depending on whether the vector attributable to a patient is on either side of the hyperplane, the gene expression profile of the patient is either similar to the reference gene expression profile from the platinum-sensitive patients (1st class) or to the reference gene expression profile from the platinum-resistant patients (2nd class). An overview of the use of SVM can be found, for example, in Burges CJC (1998), A tutorial on support vector machines for pattern recognition, Data Mining and Knowledge Discovery 2 (2): pages 121-167, and in Perez-Diez et al , (2005), Microarray for Cancer Diagnosis and Classification, Microarray Technology and Cancer Gene Profiling, pages 1-12. The content of the above publications is incorporated herein by reference. The genes which form part of the subgroup are identified according to the invention by the international gene symbol of the HUGO Gene Nomenclature Committee (HGNC), if known, and the accession number of the GenBank. Access to the GenBank is, for example, via the website of the National Center for Biotechnology Information (NCBI) at http://www.ncbi.nlm.nih.gov/ possible.

The problem underlying the invention is hereby completely solved. The inventors have each analyzed twelve platinum-sensitive and platinum-resistant ovarian carcinoma samples and were able to detect 102 different genes or transcripts which show a different expression profile in platinum-sensitive patients than in platinum-resistant patients. From these genes, in turn, the inventors were able to identify 55 different genes which are particularly suitable for classifying a patient to be examined, for example a patient suffering from an ovarian carcinoma, as platinum-sensitive or platinum-resistant. This group of genes is also called a "gene predictor set". The inventors have tested the method according to the invention on 18 different cryopreserved tumor samples from patients for whom a classification into platinum-sensitive and platinum-resistant has already been carried out using classical clinical parameters, so that it was already known whether platinum resistance or platinum sensitivity existed. The result was a 100% correct prediction of the platinum resistance of the samples. The method according to the invention is therefore particularly reliable.

Against this background, another object of the present invention is the use of the gene expression profile of a subset of genes for classifying a patient as platinum-sensitive or platinum-resistant, wherein the subset of genes consists of at least one gene selected from the group of genes described above.

It is preferred in the classification method according to the invention and the use according to the invention, if the subgroup of genes from at least 5, preferably from at least 10, more preferably from at least 20, more preferably at least 30, more preferably at least 40, more preferably at least 50 and most preferably all 55 genes selected from the group identified above.

This measure has the advantage that the accuracy of the classification of a patient in platinum-sensitive or platinum-resistant is further increased. Thus, a highly reliable classification is already made possible if the gene expression profile of at least 5 genes is analyzed and compared with the reference expression profiles, but the reliability is further increased the more genes are examined from the subgroup.

In the classification method of the present invention, it is preferred that the first reference gene expression profile be obtained from those patients who do not relapse six months after completing platinum-based chemotherapy (platinum-sensitive patients), or if the second reference expression profile was obtained from those patients within six Suffer recurrence months after completion of platinum-based chemotherapy (platinum-resistant patients).

This measure has the advantage that reference gene expression profiles are established, which are based on generally accepted guidelines, such as those issued by the German Society for Gynecology and Obstetrics e.V. to assess whether a patient is platinum-sensitive or platinum-resistant. This measure additionally ensures that a correct classification of the patient to be examined takes place.

It is preferred in the classification method according to the invention if the number of platinum-sensitive patients substantially corresponds to the number of platinum-resistant patients. This measure ensures that meaningful and above all comparable reference gene expression profiles can be established. The classification of the patient to be examined is increased in its accuracy again. In contrast, for example, the reference expression profile used by Helleman et al. (Supra) is not balanced and the associated classification method is inaccurate. The authors examined 19 platinum-sensitive patients, but only 5 platinum-resistant patients.

In the method according to the invention, it is preferred if the biological sample has tumor tissue, preferably derived from an ovarian carcinoma.

This measure has the advantage that the classification method according to the invention enables the conditions for a targeted therapy of a cancer patient, preferably a patient suffering from an ovarian carcinoma. Thus, early detection of platinum resistance avoids the administration of toxic platinum and the associated burden on the patient and makes a decision for alternative therapy.

It is preferred in the classification method according to the invention, when the biological sample has at least 50% tumor tissue.

This measure has the advantage that a particularly good and reliable classification of the patient to be examined is ensured. Thus, gene expression activity is detected in tumor tissue in particular, which makes it possible to distinguish between platinum-sensitive patients and platinum-resistant patients.

It is preferred in the classification method according to the invention, when the tumor tissue is provided in cryopreserved form.

With this measure, the inventive method is modified such that its implementation is not immediately after the removal of the biological sample has to be done. Rather, the removed biological sample can be frozen and thus preserved and carried out the implementation of the method according to the invention in a special laboratory equipped for this purpose, to which the preserved sample is überfuhrt. The inventors have found that cryopreserved tissue is as suitable for classification as "fresh" biological tissue taken just prior to processing.

In the classification method according to the invention, it is preferred if steps (3) and (4) are carried out using a support vector machine (SVM).

This measure has the advantage that the comparison of the determined gene expression profile of the biological sample with the reference gene expression profiles and the classification of the patient by means of an established and accurate mathematical method. A "support vector machine" (SVM) represents a classifier. An SVM subdivides a set of objects, for example genes or patients to be examined, into two classes so that the widest possible area remains free of objects around the class boundary. The SVM is a mathematical pattern recognition technique that is implemented in computer programs, and the basis for building an SVM is a set of training objects, each of which is known to which class it belongs, for example, a set of patients who Each object is represented by a vector in a vector space, and the task of the SVM is to fit into this space a multidimensional hyperplane that acts as the interface and the training objects divides into two classes The distance of the vectors closest to the hyperplane to the hypereb ene is maximized. This wide, empty room should later ensure that even objects that do not correspond exactly to the training objects are classified as reliably as possible. Using the SVM, a class prediction can then be made on the basis of the expression profile of at least one of the 55 genes identified by the inventors, by means of which a vector, which represents the gene expression profile of a patient being examined, on one or the other side of the hyperplane is arranged, depending on to which reference gene expression profile greater similarities exist. This makes it possible to classify a patient in platinum-sensitive or platinum-resistant. The mathematical method of SVM is described, for example, in Burges CJ. C. (supra) and Perez-Diaz et al. (Supra). The content of these publications is incorporated by reference into the description.

It is preferred if step (2) of the classification method according to the invention comprises the following steps:

2.1 isolation of total RNA,

2.2 Transcription of RNA into Complementary DNA (cDNA),

2.3 / M vitro transcription of the cDNA to obtain RNA transcripts,

2.4 Hybridization of the RNA transcripts to a biochip having at least those nucleic acid molecules which are hybridizable with RNA transcripts of the genes of the subgroup, and

2.5 Determination of the expression profile based on the hybridization activity of the RNA transcripts of the genes of the subgroup to the biochip.

This procedure has the advantage that measures which in themselves belong to the routine of a molecular biologist can be taken, can be carried out in molecular biological laboratories without great effort and ensure a reliable determination of the gene expression profile of the biological sample. A correct classification of the patient is thus ensured.

According to the invention, a preferred biochip is the "Human-6 v2 Expression BeadChip" from Illumina.

Thus, a biochip is used, with which the expression of more than 48,000 transcripts or genes of humans can be examined. The chip is characterized by its high sensitivity, selectivity and measurement precision. Furthermore, only small amounts of biological material or RNA required to perform gene expression analysis. It is understood that other biochips are also suitable which have such nucleic acid molecules with which the transcripts of at least one gene, preferably of all 55 identified genes of the subgroup, are hybridisable.

A further subject matter of the present invention relates to a method for identifying informative genes in order to be able to classify a patient as platinum-sensitive or platinum-resistant, which comprises the following steps:

1. Provision of biological samples a) from platinum-sensitive patients and b) from platinum-resistant patients,

2. Determination of gene expression profiles a) from the biological sample from the platinum-sensitive patients and b) from the biological sample from the platinum-resistant patients,

3. Identification of genes differentially expressed in the platinum-sensitive patient than in the platinum-resistant patient to obtain informative genes ("gene predictor set"), the biological sample being obtained from such platinum-sensitive patients 6 months after completion of a platelet-sensitive patient platinum-based chemotherapy did not relapse, and the biological sample was obtained from those platinum-resistant patients who relapsed within 6 months of completing platinum-based chemotherapy.

This method differs from the identification methods known in the art in that only those patients were used to obtain informative genes, which are classified as platinum-sensitive or platinum-resistant according to generally accepted guidelines. A corresponding classification is, for example, also by the German Society of Gynecology and Obstetrics eV in the form of a "recommendation for diagnostics and diagnostics Therapy of malignant ovarian tumors ", as of September 2006, see the website at http://www.dggg.de/leitlinien2006/pdf-2006/2-onko-gyn/2/2- 5a-ovarialkarzinom-kurz-pdf.

In the case of the identification method according to the invention, it is preferred if the number of platinum-sensitive patients essentially corresponds to the number of platinum-resistant patients.

This approach has the advantage that the gene expression profiles from the platinum-sensitive patients and the platinum-resistant patients are balanced and representative, and the reliable identification of "real" informative genes is made possible by contrast to that described by Helleman et al (supra) The researchers selected a platinum-resistant, or platinum-sensitive, sample of 5 platinum-resistant and 19 platinum-sensitive patients and identified gene expression profiles, which identified nine informative genes may not be suitable for reliable classification of patients.

In the identification method according to the invention, it is preferred if the biological samples have tumor tissue, preferably derived from an ovarian carcinoma.

The inventors have found that the identification method according to the invention can identify particularly well those informative genes which enable a distinction to be made between platinum-sensitive ovarian carcinoma patients and platinum-resistant ovarian carcinoma patients. The identification method according to the invention is therefore particularly suitable for the classification of tumor patients.

It is preferred if the biological samples have at least 50% tumor tissue, which can be provided preferably also in cryopreserved form. This measure ensures that reliable informative genes are identifiable. Thus, such genes are suitable as "gene predictors", which are expressed differently in tumors of platinum-sensitive and platinum-resistant patients. Therefore, using only those biological samples that have at least 50% tumor tissue reliably leads to the identification of informative genes.

The inventors have found that not only "fresh" tumor tissue is suitable for the identification of informative genes by the method of the invention, but equally cryopreserved tissue. The method can therefore be carried out in any molecular biological laboratory, which is not necessarily in close proximity to a clinic in which the patient to be examined, the tissue sample is removed. This can rather be frozen and preserved and sent to the examination laboratory.

It is preferred if step (3) is carried out using a support vector machine (SVM).

This measure has the advantage that an established and reliable mathematical method is used by means of which informative genes can be obtained. Reference is made to the above statements on SVM in connection with the classification method according to the invention, which apply equally here.

In the identification method according to the invention it is preferred if step (2) comprises the following steps:

2.1 isolation of the total RNA from the biological sample from the platinum-sensitive patients and from the biological sample from the platinum-resistant patients,

2.2 Transcription of the isolated RNA into complementary DNA (cDNA), 2.3 / nv / rro transcription of the cDNA to obtain RNA transcripts,

2.4 hybridizing the RNA transcripts to a biochip having hybridizable nucleic acid molecules representing the human genome, and

2.5 Determination of the expression profiles of the biological samples based on the hybridization activity of the RNA transcripts to the biochip.

This procedure has the advantage that the measures are carried out which enable a reliable determination of gene expression profiles from the biological samples using established molecular biological methods. Reference is made to the above statements in connection with the classification method according to the invention, which apply equally here.

It is preferred if the biochip used is the "Human 6 v2 Expression BeadChip" from Illumina.

This measure has the advantage that such a biochip is used with which the expression of more than 48,000 transcripts can be investigated. The chip is characterized by its high sensitivity, selectivity and measurement precision. Furthermore, only small amounts of biological material or isolated total RNA are required to perform a gene expression analysis. It is understood that other comparable biochips are also suitable.

It is understood that the features mentioned above and those yet to be explained below can be used not only in the particular combination given, but also in other combinations or in isolation, without departing from the scope of the present invention.

The present invention will now be explained in more detail with reference to exemplary embodiments, which are purely illustrative and do not limit the scope of the present invention. Reference is made to the accompanying drawings, in which: Fig. 1 shows the RNA expression profiles of cisplatin-resistant and

-sensitive ovarian cancer in a microarray analysis on the "Human 6 v2 Expression BeadChip" (Illumina);

Fig. 2 shows the result of a "principle componenf" analysis (PCA) of

Gene expression data with the total data set (A) and the data of the "gene predictor set" consisting of 55 genes (B). Each point corresponds to a data set (array). Resistant samples are white, sensitive black.

Exemplary embodiments

1. Identification of 55 informative genes

12 Ovarian cancer patients who did not relapse 6 months after completion of platinum-based chemotherapy (platinum-sensitive patients) and 12 ovarian cancer patients who relapsed within 6 months of platinum-based chemotherapy (platinum-resistant patients) were sampled from an ovarian cancer by an experienced surgeon , The tissue samples were cryopreserved with an ischemia time of less than 10 minutes.

Of the cryopreserved tissues, sections of 10 μm each were made. These were characterized histopathologically in such a way that only those samples were selected which contained a tumor surface fraction of> 50%.

From the tissue sections, the total RNA was isolated, transcribed into complementary DNA (cDNA). The cDNA was transcribed in vitro, whereby the RNA transcripts were labeled with a detectable fluorescent marker. The RNA transcripts were hybridized with a "Human 6 v2 Expression BeadChip" from Illumina. The result of the subsequent microarray analysis is shown in FIG.

Significance analyzes identified 102 transcripts that are differentially expressed in cisplatin-resistant and cisplatin-sensitive ovarian cancers. The presented hierarchical cluster analysis already allows a very good separation of the cisplatin-resistant (white) from the cisplatin-sensitive samples (black). Each row corresponds to a gene or transcript, each column corresponds to an array. The gray tone indicates the expression value. The next step is a "class prediction" analysis (CPA) via "support vector-machines" (SVM). From the 102 differentially expressed transcripts, a "gene pre-dictor set" of 55 genes could be determined by cross-validation ("leave-one-out"). These 55 informative genes are shown in Table 1 below.

50 ILMN. _11548 0,553 C0L1A1 NM..000088.2

50 ILMN. _11548 0,553 C0L1A1 NM..000088.2

51 ILMN. _14608 0,55 ZNF539 NM. .203282.1

52 ILMN. _5270 0.55 REM1 NM. .014012.4

53 ILMN. J8256 0.54 EDG3 NM. _005226.2

54 ILMN. _25778 0,533 CSPG2 NM. _004385.2

55 ILMN. _21174 0,529 TYRP1 NM. _000550.1

Table 1: Informative genes ("gene predictor set")

2. Verification of the "gene predictor sets" containing the 55 genes

The predictive power of the gene predictor set was verified from 18 samples derived from cisplatin-resistant ovarian cancer patients. The result is shown in Table 2.

Table 2: Predictive force of the gene predictor set; Classification accuracy; 1: cisplatin sensitive, 0: cisplatin resistant

It can be seen from Table 2 that the "gene predictor set" of the 55 genes allows a 100% correct prediction of cisplatin resistance or cis-platinum sensitivity.

FIG. 2 shows the result of a principal component analysis ("PCA") of the gene expression data with the overall data set (A) and the data of the "gene predictor set" from the 55 identified genes (B). Each point corresponds to a data set (array). Resistant samples are white, sensitive black. The PCA with the unfiltered data set does not allow a clear distinction between cisplatin-sensitive and resistant patients, whereas the 55 genes of the "gene predictor set" allow a clear separation of the two groups of patents; see. Partial picture (B).

3. Conclusion

For the first time, the inventors provide a method by which a classification of patients into platinum-sensitive and platinum-resistant can be reliably performed. At the heart of this process is the provision of a gene predictor set consisting of 55 differentially expressed genes. With the help of this method, for example, ovarian carcinoma patients can be identified who benefit from a purine-containing therapy. On the other hand, platinum-resistant patients can be excluded and alternative chemotherapy can be chosen immediately to avoid unnecessary toxicities. The inventors are also for the first time providing a reliable method to identify informative genes by which a classification of patients in platinum-sensitive and platinum-resistant is made possible.

Claims

Patent claims 1. Method for classifying a patient as platinum sensitive or platinum resistant, comprising the following steps: 1. Providing a biological sample from a patient, 2. Determination of the gene expression profile of a subset of genes from the biological sample, 3. Comparison of the gene expression profile determined in step (2) a) with a first reference gene expression profile of the subgroup of genes, the first reference gene expression profile being obtained from platinum-sensitive patients, and b) with a second reference gene expression profile of the subgroup of genes, the second reference gene expression profile being obtained from platinum-resistant patients patients were recruited, 4. Classify the patient a) as platinum sensitive if the patient's gene expression profile has greater similarities with the first reference gene expression profile, or b) as platinum resistant if the patient's gene expression profile has greater similarities with the second reference gene expression profile, characterized in that the subset of genes consists of at least one gene selected from the group consisting of: PPAPDClA (NM_001030059.1), EGFR (NM_005228.3), NKD2 (NM_033120.2), GPM6B (NM_001001994.1) , AMMECRl (NM_001025580.1), LTB4R (NM_181657.1), BXl 12399; TUSCl (NM_001004125.1), MUC15 (NM_145650.2) _; HMFN0839 (NM_032717.3), SCAPl (NM_003726.2), XM_498568, CARD4 (NM_006092.1), FTCD (NM_206965.1), 0686 (NM_001025303. 1), COL5A1 (NM_000093.2), HIST1H2BD (NM_138720.1), F2R (NM_001992.2), KIAA1875 (NMJB2529.1), HLA-DMB (NM_002118.3), FOXPl (AK025793), APOCl (NM_001645.3 ), WBSCR27 (NM_152559.2), CATSPERl (NM_053054.2), JOSD2 (NM_138334.1), BCAR3 (BX106566), BGN (NM_001711.3), HOXB7 (NM_004502.2), FXYD3 (NM_005971.2), NIPAl (NM_144599.3), GSR (NM_000637.2), CSTl (NM_001898.2), LAT2 (NM_022040.2), HLA-DQBl (NM_002123.2), LSAMP (NM_002338.2), LOC652670 (XM_942247.1), LOC642412 (XM_925931.1), NEK8 (NM_178170.2), CD40 (NM_001250.3), SAMDIl (NM_152486.2), MSC (NM_005098.2), CR607098, PLCXDl (NM_018390.1), RASLIlA (NM_206827.1) , COL3A1 (NM_000090.2), FLJ13391 (NM_032181.1), COLlAl (NM_000088.2), ZNF539 (NM_203282.1), REMl (NM_014012.4), EDG3 (NM_005226.2), CSPG2 (NM_004385.2), TYRPl (NM_OOO55O.l). 2. The method according to claim 1, characterized in that the subgroup of genes consists of at least 5, preferably of at least 10, more preferably of at least 20, more preferably of at least 30, more preferably of at least 40, more preferably of at least 50 and most preferably consists of all 55 genes selected from the group identified in claim 1. 3. The method according to claim 1 or 2, characterized in that the first reference gene expression profile was obtained from patients who do not suffer a recurrence 6 months after completion of platinum-based chemotherapy (platinum-sensitive patients). 4. Method according to one of the preceding claims, characterized in that the second reference gene expression profile is obtained from such patients who suffered a recurrence within 6 months of completing platinum-based chemotherapy (platinum-resistant patients). 5. The method according to claim 3 or 4, characterized in that the number of platinum-sensitive patients essentially corresponds to the number of platinum-resistant patients. 6. Method according to one of the preceding claims, characterized in that the biological sample comprises tumor tissue, preferably originating from an ovarian carcinoma. 7. The method according to claim 6, characterized in that the biological sample has at least 50% tumor tissue. 8. The method according to claim 6 or 7, characterized in that the tumor tissue is provided in cryopreserved form. 9. Method according to one of the preceding claims, characterized in that steps (3) and (4) are carried out using a “support vector machine (SVM)”. 10. Method according to one of the preceding claims, characterized in that step (2) comprises the following steps: 2.1 Isolation of total RNA, 2.2 Rewriting the RNA into complementary DNA (cDNA), 2.3 In virro transcription of cDNA to obtain RNA transcripts, 2.4 Hybridization of the RNA transcripts to a biochip that has at least those nucleic acid molecules that can be hybridized with RNA transcripts of the genes of the subgroup, and 2.5 Determination of the expression profile based on the hybridization activity of the RNA transcripts of the genes of the subgroup to the biochip. 11. The method according to claim 10, characterized in that the “Human-6 v2 Expression BeadChip” from Illumina is used as the biochip. 12. Use of the gene expression profile of a subset of genes to classify a patient as platinum sensitive or platinum resistant, characterized in that the subset of genes consists of at least one gene selected from the group consisting of: PPAPDClA (NM_001030059.1), EGFR ( NM_005228.3), NKD2 (NM_033120.2), GPM6B (NM_001001994.1), AMMECRl (NM_001025580.1), LTB4R (NM_181657.1), BXl 12399; TUSCl (NM_001004125.1), MUC15 (NM_145650.2), HMFN0839 (NM_032717.3), SCAPl (NM_003726.2), XM_498568, CARD4 (NM_006092.1), FTCD (NM_206965.1), 64 (NM_001013670. 1), COMP (NM_000095.2), LOC440686 (NMJ)01025303.1), COL5A1 (NM_000093.2), HIST1H2BD (NM_138720.1), F2R (NM_001992.2), KIAA1875 (NM_032529.1), HLA-DMB (NM_002118 .3), FOXPl (AK025793), APOCl (NM_001645.3), WBSCR27 (NM_152559.2), CATSPERl (NM_053054.2), JOSD2 (NM_138334.1), BCAR3 (BX106566), BGN (NMJ)Ol 711.3), HOXB7 (NM_004502.2), FXYD3 (NM_005971.2), NIPAl (NM_144599.3), GSR (NM_000637.2), CSTl (NM_001898.2), LAT2 (NM_022040.2), HLA-DQBl (NM_002123.2) , LSAMP (NM_002338.2), LOC652670 (XM_942247.1), LOC642412 (XM_925931.1), NEK8 (NM_178170.2), CD40 (NMJ)Ol 250.3), SAMDIl (NMJ 52486.2), MSC (NMJ)05098.2), CR607098, PLCXDl (NMJ)18390.1), RASLIlA (NM_206827.1), COL3A1 (NM_000090.2), FLJ13391 (NMJ)32181.1), COLlAl (NMJ)00088.2), ZNF539 (NM_203282.1), REMl (NMJH4012.4 ) , EDG3 (NMJ)05226.2), CSPG2 (NMJ)04385.2), TYRPl (NMJ)00550.1). 13. Use according to claim 12, characterized in that the subgroup of genes consists of at least 5, preferably of at least 10, more preferably of at least 20, more preferably of at least 30, more preferably consists of at least 40, more preferably at least 50 and most preferably all 55 genes selected from the group identified in claim 12. 14. Method for identifying informative genes to classify a patient as platinum sensitive or platinum resistant, comprising the following steps: 1. Provision of biological samples a) from platinum-sensitive patients and b) from platinum-resistant patients, 2. Determination of the gene expression profiles a) from the biological sample from the platinum-sensitive patients and b) from the biological sample from the platinum-resistant patients, 3. Identification of genes that are expressed differently in the platinum-sensitive patients than in the platinum-resistant patients, in order to obtain informative genes (“gene predictor set”), characterized in that the biological sample was obtained from those platinum-sensitive patients who did not suffer a recurrence 6 months after completion of platinum-based chemotherapy, and the ^biological sample was obtained from those platinum-resistant patients who suffered a recurrence within 6 months after completion of platinum-based chemotherapy. 15. The method according to claim 14, characterized in that the number of platinum-sensitive patients essentially corresponds to the number of platinum-resistant patients. 16. The method according to claim 14 or 15, characterized in that the biological samples have tumor tissue, preferably originating from an ovarian carcinoma. 17. The method according to claim 16, characterized in that the biological samples have at least 50% tumor tissue. 18. The method according to claim 16 or 17, characterized in that the tumor tissues are provided in cryopreserved form. 19. The method according to any one of claims 14 to 18, characterized in that step (3) takes place using a “support vector machine (SVM)”. 20. The method according to any one of claims 14 to 19, characterized in that step (2) comprises the following steps: 2.1 Isolation of total RNA from the biological sample from the platinum-sensitive patients and from the biological sample from the platinum-resistant patients, 2.2 Rewriting the isolated RNA into complementary DNA (cDNA), 2.3 //7-vzrro transcription of cDNA to obtain RNA transcripts, 2.4 Hybridization of the RNA transcripts to a biochip that has hybridizable nucleic acid molecules representing the human genome, and 2.5 Determination of the expression profiles of the biological samples based on the hybridization activity of the RNA transcripts to the biochip. 21. The method according to claim 20, characterized in that the “Human-6 v2 Expression BeadChip” from Illumina is used as the biochip.