FIELD OF THE INVENTION

The present invention relates to the field of medicine, in particular of oncology. It relates to a new method to classify patients suffering from a cancer for therapeutic intervention.

BACKGROUND OF THE INVENTION

Epithelial ovarian cancer is the most lethal form of gynaeco logic malignancy. Because of the insidious onset of the disease and the lack of reliable screening tests, patients are often diagnosed with advanced disease. The established clinical prognosis factors are based on disease at diagnosis, including the extent of residual disease post-surgery, histological subtype, tumor grade and stage. Although many patients respond initially to standard combinations of surgery and chemotherapy, most of them will develop recurrence and eventually succumb to their disease.

In recent years, several studies have analyzed large-scale transcription profiling to identify differentially expressed genes in ovarian cancer according to tumor status, histological subtypes and metastatic spread. Yet, the molecular biology of ovarian cancer is still not completely understood, making difficult the development of more effective therapies Ovarian cancers may be of different sub-types with different pathological features and outcomes. Due to these variations, the appropriate therapy for each of this sub-type may differ. Considerable efforts have been made in order to find markers that could be used to classify ovarian cancer. In this aim, the use of microRNAs has been extensively studied. In particular, miR-200 family members have been shown to accumulate in ovarian cancer (Iorio et al, 2007; Nam et al, 2008; Hu et al, 2009; Bendoraite et al, 2010). However, the correlation between the expression of miR-200s and of the prognosis remains uncertain. Indeed, in these studies, it has been shown that high expression of miR-200 could be linked to poor prognosis (Nam et al, 2008) or to good prognosis (Hu et al, 2009).

Accordingly, there is still a strong need to provide reliable markers that could be used to stratify patients for therapeutic intervention. SUMMARY OF THE INVENTION

The inventors demonstrate that expression levels of GSR and MYL9 genes can be used as a dual signature to classify a tumor, to predict the clinical outcome of the patient, to determine if a patient will be a good responder to a treatment, and to propose to the patient the most appropriate treatment according to the classification of its tumor.

Accordingly, in a first aspect, the present invention concerns a method for predicting clinical outcome of a patient affected with a cancer, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a high expression level of GSR gene and a low expression level of MYL9 gene being indicative of a longer progression free survival and/or an increased patient survival.

The present invention also concerns a method for selecting a patient affected with a cancer for surgical debulking and/or a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, or for determining whether a patient affected with a cancer is susceptible to benefit from surgical debulking and/or a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a high expression level of GSR gene and a low expression level of MYL9 gene being indicative that said patient is susceptible to benefit from surgical debulking and/or a dose- intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells.

The present invention further concerns a method for selecting a patient affected with a cancer for neo-adjuvant chemotherapy before surgical debulking, or for determining whether a patient affected with a cancer is susceptible to benefit from neo-adjuvant chemotherapy before surgical debulking, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a low expression level of GSR gene and a high expression level of MYL9 gene being indicative that said patient is susceptible to benefit from neo-adjuvant chemotherapy before surgical debulking.

The expression level of each gene may be determined (i) by measuring the quantity of mR A, preferably by quantitative or semi-quantitative RT-PCR, by real time quantitative or semi-quantitative RT-PCR, Nanostring technology, sequencing based approaches or by transcriptome approaches, and/or (ii) by measuring the quantity or the activity of encoded protein, preferably by immunochemistry, semi-quantitative Western-Blot or protein or antibody arrays.

The methods of the invention may further comprise comparing the expression levels of the genes to a reference expression level. Before to be compared with the reference expression level, the expression levels of GSR and MYL9 genes may be normalized using the expression level of an endogenous control gene having a stable expression in different cancer samples, such as RPLPO, HPRT1, GAPDH and B2M genes.

The expression levels of GSR and MYL9 genes may also be compared with the mean expression level of GSR and MYL9 genes among a population of randomly selected cancer samples.

The methods of the invention may further comprise determining the expression level of one or several additional genes selected from the group consisting of DNAJC10, PIK3R3, PRDXl, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKR1A1, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHX1, STIP1, IGF1, TIMP2, FNl, TNFRSFllB, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COLlAl, LY96, FGFR2, MYH9 and PDGFRB, preferably selected from the group consisting of AKR1A1, TIMP2, NQOl, PRKCZ and FNl, more preferably selected from the group consisting of AKR1A1, TIMP2, NQOl and FNl, and even more preferably selected from the group consisting ofAKRIAl and TIMP2.

Optionally, the expression level of one or several additional genes selected from the group consisting of CPT1B, PARK7, SUCLG2, MT-C02, CAT, COX5A, IDH3A, CLYBL, ATP6V0E1, NDUFB4, SDHB, NDUFA7, UQCR11, NDUFA6, AC02, FH, TRAK1, UQCRQ, NDUFA1, PPA2, ATP5J2, NDUFA11, CASP9, UQCRC1, NDUFS1, NDUFS7, CASP8, NDUFA8, COX5B, COX7A2, GPD2, ATP5D, OGDH, ATP 50, COX8A, IDH3G, NDUFA2, APHIA, ATP5I, XDH, COX7B, ATP5G3, NDUFA5, CYCS, AIFMl, ATP7A, NDUFAFl, MT- ND6, DID, IDH2, PDHA1, ATP6V0B, COX4I1, NDUFB3, COX17, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYH11, IGF1R, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, IL1RAP, IGFBP3, VCAM1, ACAA1, ACAD8, ACADM, ACADSB, AKR7A2, AKR7A3, ALDH6A1, AUH, BCKDHA, CBR1, CLPP, DNAJC3, DNAJC4, DNAJC8, ECHS1, FTH1, GCLC, GCLM, GSTK1, HADH, HIBADH HMGCL, HMGCS1, HSD17B4, IVD, KRAS, MCCC2, MGST2, NQ02, PIK3CB, PPIB and VCP, may also be determined.

The methods of the invention may also comprise assessing at least one another prognosis markers such as histological sub-type, tumor grade, tumor stage, p53 status, BRCAl/2 status, mitotic index, tumor size or extent of residual disease after surgery, or expression of proliferation markers.

The chemotherapy inducing acute oxidative stress in cancer cells may comprise an antineoplastic agent inducing the accumulation of ROS, preferably selected from the group consisting of piperlongumine and its derivatives, lanperisone and its derivatives, anthracyclines, taxanes, alkylating agents, epipodophyllotoxins, camptothecins and cytotoxic antibiotics.

Alternatively, the chemotherapy inducing acute oxidative stress in cancer cells may comprise a compound having the ability to induce or increase oxidative stress, preferably a compound selected from the group consisting of tolperisone; artemisinin; darinaparsin; motexafm gadolinium; menadione; shikonin; paracetamol; acetylsalicylic acid; geldanamycin; 3,7-diaminophenothiazinium redox dyes; disulfiram; polysulfide-based anticancer drugs; diallyldisulfide and diallyltrisulfide; isothiocyanate organosulfur agents; electrophilic Michael acceptors; superoxide dismutase inhibitors; superoxide dismutase mimetics; compounds disturbing the glutathione redox-system; compounds disturbing the thioredoxin system; compounds inhibiting NQOl function; compounds inhibiting APE-Refl function; 2- deoxyglucose; 3-bromopyruvate; dichloroacetate; redox inactive vitamine E analogues; 3,3'- diindolylmethane; Bz-423; erastin and RSL5; and their derivatives and analogues, and any combination thereof.

Preferably, radiotherapy is nanoparticle enhanced radiotherapy.

In another aspect, the present invention concerns a kit comprising detection means selected from the group consisting of a pair of primers, a probe and an antibody specific to GSR gene and MYL9 gene, and optionally, a leaflet providing guidelines to use such a kit.

The present invention also concerns the use of this kit for predicting clinical outcome and/or for selecting a patient affected with a cancer for surgical debulking and/or a dose- intensive chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells and/or for selecting a patient affected with a cancer for neo-adjuvant chemotherapy before surgical debulking.

Preferably, the cancer is selected from the group consisting of ovarian cancer, cervical cancer, vulvar cancer, vaginal cancer, prostate cancer, lung cancer, pancreas cancer, colorectal cancer and leukemia. More preferably, the cancer is ovarian cancer.

In a further aspect, the present invention concerns an compound inducing the accumulation of ROS, preferably an antineoplastic agent selected from the group consisting of piperlongumine and its derivatives, lanperisone and its derivatives, anthracyclines, taxanes, alkylating agents, epipodophyllotoxins, camptothecins and cytotoxic antibiotics, for use in the treatment of a patient affected with a cancer, preferably an ovarian cancer, and having high expression level of GSR gene and low expression level of MYL9 gene, said expression levels being determined in a cancer sample from said patient. Preferably, the antineoplastic agent is selected from the group consisting of piperlongumine and its derivatives, lanperisone and its derivatives and epipodophyllotoxins.

Preferably, the patient is a patient in whom expression levels of GSR and MYL9 genes have been determined in a cancer sample from said patient. BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1: Expression of miR-200 microRNAs is stimulated by oxidative stress. Fig .la: Unsupervised hierarchical clustering of miRNA microarray expression data from mouse fibroblasts following H ₂ 0 ₂ exposure (hours, h). miRNAs were selected based on differential expression (P < 0.05 Illumina custom error model) at early (0.5h-lh), mid (2h-4h) or late (8h-24h) time points, compared to untreated. Values are means of normalized expression from 2 independent experiments. Up-and down-regulated miRNAs upon stress are shown in red and blue, respectively. Fig. lb: qRT-PCR of miR-200 family members upon time-course of H ₂ 0 ₂ treatment in mouse cells (immortalized fibroblasts; CT26, colon carcinoma; NMuMG, mammary gland epithelial cells; K-ras and H-ras transformed fibroblasts) and human cell lines (MD A-MB-435 S , melanoma cells; 293T, kidney cells, MDA-MB-436 and BT-549, breast adenocarcinoma and SKOV3, ovarian adenocarcinoma). Data are means of fold changes (normalized to untreated, expressed in log2) ± sd.

Figure 2: miR-200 family members are regulated by oxidative stress. Fig. 2a: Sequence alignment of miR-200 family members in human (left) and mouse (right) species. The miR-141/200a seed sequence differs from miR-200b/200c/429 one's, suggesting that each subset can recognize different targets. Fig. 2b: Chromosomal organization of the miR- 200 family members in mouse (mmu) and human (hsa) species. Fig. 2c: qRT-PCR cycle threshold (Ct) of miR-200s family members in several human cell lines. MDA-MB-468, MCF7 and IGROV cells express high basal levels of all tested miR-200s. Fig. 2d: Ct values of miR-200s in MDA-MB-468 (upper panel) and MCF7 (lower panel) cell lines under untreated conditions (-) or treated (+) with H ₂ 0 ₂ for 3 hours. Fig. 2e: Representative graph of miR-141, miR-200c, JUN (c-Jun) and HOI (Heme oxygenase) expression levels in mouse fibroblasts during time-course stimulation with H ₂ 0 ₂ , PMA, tBHQ and Serum. Jun and HOI were used as internal positive controls. PMA: Phorbol Myristate Acetate; tBHQ: tertiary butylhydroquinone.

Figure 3: Balance of p38/JNK pathways by miR-141/200a. Fig. 3a: Schema of ROS-signaling pathway in miR-141-overexpressing mouse fibroblasts. Micro-array data were analyzed using Ingenuity IPA software. Significantly up-regulated mRNAs (Ras and JNK1/2) and down-regulated mRNAs (p38 MAPK) are shown (P < 0.05, Illumina custom error model with FDR). Fold changes are normalized to control miRNA-overexpressing cells. Fig. 3b: MAPK14 mRNA level evaluated by microarray (top) or qRT-PCR (bottom) in mouse fibroblasts, 3 days post-transfection with transfection reagent (Trans), control miRNA (Control) or specific miRNA, as indicated. Data are means of fold changes (normalized to control) ± sem. *P < 0.05; **P < 0.01; ***P < 0.001 (Student's t-test). Fig. 3c: Western blots showing MAPK-related proteins in mouse fibroblasts transfected with specific miRNA, as noted, under basal conditions (-) or after H ₂ 0 ₂ exposure (hours, h). Fig. 3d: Western blots showing p38a protein in mouse (fibroblasts, NMuMG) and human (MDA-MB-435S, SKOV3) cell lines, following miRNA transfection, as indicated. Fig. 3e: Western blot showing p38a protein levels in diverse human cancer cell lines. A cell line belongs to the high or low class of miR141/200a if their miR-141/200a content is respectively, above or below, the average miR-141/200a level from all analyzed cell lines. Using this rule, cells belonging to the class of "high miR-141/200a" express at least one of the miR at a 5 fold higher level than their average levels. Fig. 3f: Correlation plot extracted from NCI-60 Agilent microarray datasets showing that MAPK14 mRNA level is inversely correlated with endogenous miR- 200a levels (R = -0.29; P = 0.023 Pearson's test).

Figure 4: miR-141/200a modulate MAPK pathways response to stress. Fig. 4a: Western blots showing MK 3 (MAP2K3), MK 6 (MAP2K6), MK 4 (MAP2K4) and MKK7 (MAP2K7) proteins and their phosphorylated isoforms in miR-141 or miR-200a overexpressing cells under basal conditions (-) or upon H ₂ 0 ₂ exposure (time in hours, h). GAPDH is used as internal loading control. Fig. 4b: Bar graphs showing mRNA levels of genes involved in MAPK or JNK pathways, quantified by qRT-PCR, in miR-141 or miR- 200a-overexpressing cells, under basal conditions or upon H ₂ 0 ₂ exposure (time in hours, h). Expression values are represented as means of fold changes (normalized to untreated and control miRNA) ± sem. *P < 0.05; **P < 0.01; ***P < 0.001 (Student's t-test).

Figure 5: miR-200a and miR-141 target p38a through its 3'UTR. Fig. 5a: Western blots from showing total and phosphorylated isoforms of p38a and JNK in miR-141 or miR- 200a-expressing fibroblasts either untreated (-) or incubated with MG132. E-cadherin and β- catenin are used as internal controls. Fig. 5b: Schematic representation and relative normalized luciferase (Firefly/Renilla) activity of MAPK14- or ZEB2-3'UTR-luciferase constructs transfected with miR-200s, as indicated. Fig. 5c: Sequence alignment of putative miR-200a or miR-141 binding sites in the 3'UTR of human (hsa) and mouse (mmu) MAPK14 genes. Fig. 5d,e: Schematic representation of human (d) or mouse (e) MAPK14-3'UTR deletion mutants and relative normalized luciferase activity upon miR-200s transfection. Fig. 5f: Normalized luciferase activity of wt or mutated MAPK14-reporter constructs after transfection with miRNA-specific antisense oligomers (inhibitors) in MDA-MB-468 cells (high endogenous miR141/200a). Fig. 5g: Normalized luciferase activity of the indicated reporter constructs after transfection of 293T cells (low endogenous miR-200s levels) with each miR in presence of miRNA inhibitors. Values are presented as relative fold changes of Firefly/Renilla activity ratio (normalized to control) ± sem. *P < 0.05; **P < 0.01; ***P < 0.001 (t-test).

Figure 6: miR-141/200a enhance tumorigenesis in mouse models. Fig. 6a: Left, representative dishes and individual colonies from K-ras-trans formed fibroblasts stably transfected with a control miRNA (ras + control) or miR-141 (ras + miR-141). Middle, relative number of visible colonies in soft agar formed from control and miR-141- overexpressing clones. Right, percentage (%) of visible colonies ranged according to their size: Small < 2.2 mm2, 2.2 mm2 < Medium < 4.5 mm2 and Big > 4.5 mm2. Fig. 6b: Tumor volumes 9 days after xenograft of control or miR- 141-overexpressing ras-trans formed fibroblasts. Values are means ± sem. n = 10 in each group. Fig. 6c: Tumor volumes 20 days after xenograft of Control-, miR-141- or miR-200a-overexpressing ovarian cancer cells, n > 17 in each group. Fig. 6d: Left, Kaplan-Meir tumor free survival curves from Control-, miR- 141- or miR-200aoverexpressing mice (Log Rank-test). Right, tumor growth curves (Student's t-test). *P < 0.05; **P < 0.01; ***P < 0.001; ns: not significant (Student's t-test).

Figure 7: Main patient characteristics and clinicopathological features of ovarian cancers in the Curie Cohort. Tumor samples were obtained from a cohort of consecutive ovarian carcinoma patients, treated at the Institut Curie between 1989 and 2005. For each patient, before chemotherapy, a surgical tumor specimen was taken for pathological analysis and tumor tissue cryopreservation. The median patient's age was 57,8 years (with a range of 31-86 years). Ovarian carcinomas were classified according to the World Health Organization histological classification of gynecological tumors. Pathological analysis showed that tumors corresponded to 82 serous, 8 mucinous, 8 endometrioid, 6 clear cell carcinomas, 2 carcinosarcomas and 1 malignant Brenner tumor. 100 cases were classified as high histological grade (grade 2 and 3) and 7 as low grade (grade 1). Clinical staging showed that 31 cases (29%) were considered as early stages (FIGO I-IIc) and 76 cases (71%) as advanced stage (III/IV). Patients were treated with a combination of surgery and chemotherapy, the latter including alkylating or alkylating- like agents +/- taxane as first line treatment in most cases. Figure 8: Clinical characteristics of patients in the Curie cohort and features of corresponding tumor samples. Abbreviation: NA : not analyzed. Ovarian cancer patients have been treated with taxanes and/or alkylating agents. Treatments combine mainly Taxol and cis-platin or carboplatine (referred to as Taxanes + Alkylating- like agents), on one hand, 5 or 5-FU, holoxan (ifosfamid) and cis-platine (including anti-metabolic and Alkylating and Alkylating- like agents, referred to as Alkylating), on the other.

Figure 9: miR-200a and stress response predict good prognosis in ovarian cancer patients. Fig. 9a,b: Scatter diagrams showing p38a IHC score relative to MAPK14 mRNA (a) and miR-200a levels (b) in ovarian tumors (Spearman test). Fig. 9c: Representative views

10 of p38a immunostaining in tumors with low or high miRNA-200a levels. Fig. 9d: Cellular pathways (from Ingenuity IPA software) positively or negatively-correlated with miR-200a levels in ovarian tumors. Unsupervised comparative analysis was done using Fisher Exact test. P values were adjusted using Benjamini-Hochberg (B-H) multiple testing correction. Fig. 9e: Unsupervised hierarchical clustering of "stress" and "fibrosis" signatures from microarray

15 ovarian cancer dataset of Institut Curie. Each row represents a tumor and each column a gene.

Red and blue colors indicate gene expression levels in a single tumor above and below the mean, respectively. Color saturation indicates the magnitude of deviation from the mean. Genes cluster together according to the "stress" and "fibrosis" signatures (bottom of the matrix). The dendrogram of samples (left of the matrix) allows classification of patients

20 according to the signatures. Fig. 9f : Kaplan-Meier of progression-free (PFS) and overall (OS) survival curves, according to the signatures (n = 51 "stress"; n = 56 "fibrosis") or to the average of miR-200a and miR-141 expression rates (n = 46 high; n = 45 low miR). Fig.9g: Unsupervised hierarchical clustering of the signatures from AOCS microarray dataset. Fig.9h: Kaplan-Meier of PFS and OS curves of AOCS patients according to the "stress" (n =

25 158) and "fibrosis" (n =128) signatures. Log-rank test was used for Kaplan-Meier. Scale bars = 50 um.

Figure 10: List of ovarian cancers - quantitative analysis of miR-200 family members and p38 IHC Table showing quantitative data of p38a protein levels using H Score, defined as the product of staining intensity and percentage of positive cells in each 30 analyzed ovarian tumor. miR-200a, miR-141 and MAPK14 mRNA levels have been normalized to referent miRs or housekeeping genes (AACt), as indicated in Method section, and are centered to the mean (log2 of fold change relative to the mean value). In this experiment, both IHC staining and miRs expression values were obtained from 50 patients, were available. Low, Medium and High classes have been defined according to the level of miR-200a in each tumor reported to the mean and in order to provide sample sets of similar size (n=17 in each group). Classes have thus been defined as follows (AACt of miR-200a): low (-2.8; -0.4), medium (-0.4; +0.4), high (+0.4;+2.8).

Figure 11: List of genes composing the "stress" and "fibrosis" signatures. For each gene are indicated the gene symbol, probe set number, gene name, gene function, coefficient of correlation with miR-200a and the corresponding P values (determined by Pearson's test - see Method section). Genes, correlated with miR-200a and found in the Ingenuity "Nrf2-mediated oxidative stress response" canonical pathway, compose the "stress" signature. Genes, anti-correlated with miR-200a and found in the Ingenuity "Hepatic fibrosis" canonical pathway, compose the "fibrosis" signature.

Figure 12: Characteristics of patients and tumor samples associated with the "stress" and "fibrosis" signatures. Association of "stress" and "fibrosis" signatures with clinical characteristics and remission status of patients from Institut Curie and AOCS cohorts, when available. The biological response was evaluated by the CA12.5 after the first cure of treatment. Clinical response was evaluated according to tumor mass evolution by monitoring patients throughout the chemotherapeutic treatment and was considered as incomplete in case of no or partial response to treatment. Debulking status was defined as optimal for tumor residues <1 cm after resection, and suboptimal for >1 cm. Data are means of log values ± sem. P values are from Chi-square (grade, stage, clinical response), Fischer Exact (debulking status) or Student's t (CA12.5) tests, ns: not significant.

Figure 13: Univariate and multivariate analyses showing predictive value of "stress" and "fibrosis" profiles adjusted for standard clinical prognostic factors. Univariate or multivariate Cox proportional hazards regression was conducted with SPSS 19.0 software using the enter method. Respective associated hazard ratios (HR) and P values are indicated.

Figure 14: miR-141/200a enhance response to chemotherapeutic reagents. Fig.

14a: Histograms showing percentage of apoptotic cells (annexinV+/DAPI-) of SKOV3 cells trans fected with miR-141, miR-200a or siRNA-p38, with or without treatment, as indicated. Fig. 14b: Growth curves in days (d) of Control-, miR-141- or miR200a-overexpressing tumors under untreated conditions (solid lines) or upon paclitaxel treatment (dashed lines). Fig. 14c: Percentage of tumor growth inhibition (TGI) after 8 days of paclitaxel treatment, n > 11 tumors in each category for each condition (treated or untreated). Fig. 14d: Percentage of TGI per tumor after 8 days of paclitaxel treatment. Data are means ± sem. *P < 0.05; **P < 0.01; ***P < 0.001 (Student's t-test).

Figure 15: Kaplan-Meier of progression-free survival (PFS) and overall survival (OS) curves, according to the dual signature comprising genes listed in Tables 1 and 2, in the Curie cohort. Log-rank test was used for Kaplan-Meier.

Figure 16: Kaplan-Meier of progression-free survival (PFS) and overall survival (OS) curves, according to the dual signature comprising genes listed in Tables 1 and 2, in the AOCS cohort. Log-rank test was used for Kaplan-Meier.

Figure 17: Kaplan-Meier of progression-free survival (PFS) and overall survival (OS) curves, according to the dual signature comprising GSR and MYL9 genes in both the Curie and the AOCS cohort and Kaplan-Meier of overall survival (OS) curves, according to the dual signature comprising GSR and MYL9 genes in the Duke Cancer Institute cohort. Log-rank test was used for Kaplan-Meier.

Figure 18: Kaplan-Meier of progression-free survival (PFS) and overall survival (OS) curves, according to the dual signature comprising genes listed in Figure 11 without GSR and MYL9 genes, in both the Curie and the AOCS cohort and Kaplan-Meier of overall survival (OS) curves, according to the dual signature comprising genes listed in Figure 11 without GSR and MYL9 genes, in the Duke Cancer Institute cohort. Log-rank test was used for Kaplan-Meier.

Figure 19 : Kaplan-Meier of progression-free survival (PFS) and overall survival

(OS) curves, according to miR-141 expression levels in the Curie cohort. Log-rank test was used for Kaplan-Meier.

Figure 20 : Kaplan-Meier of progression-free survival (PFS) and overall survival (OS) curves, according to miR-200 expression levels in the Curie cohort. Log-rank test was used for Kaplan-Meier.

Figure 21 : Kaplan-Meier of progression-free survival (PFS) and overall survival (OS) curves, according to the average of miR-141 and miR-200a expression values in the Curie cohort. Log-rank test was used for Kaplan-Meier.

Figure 22: Kaplan-Meier of overall survival (OS) curves, according to the dual signature comprising genes listed in Tables 1 and 2, in the Duke Cancer Institute cohort. Log- rank test was used for Kaplan-Meier.

Figure 23: The performance of a binary classifier of 6 genes (GSR, MYL9, AKR1A1, TIMP2, NQOl and FN1 genes) and a binary classifier of 4 genes (GSR, MYL9, AKR1A1, TIMP2 genes) was then evaluated by Receiver Operating Characteristic (ROC) analysis. ROC curves performed by plotting the true positive rate (sensitivity) versus the false positive rate (1 - specificity) at various threshold settings of the functions.

DETAILED DESCRIPTION OF THE INVENTION

By analyzing a large set of human ovarian adenocarcinomas, the inventors have revealed that tumors accumulating miR-200a/141 exhibit not only low p38a protein level but also a dual signature with a high expression of "Stress" genes and low expression of "Fibrosis" genes, which is correlated with longer progression-free interval and overall survival of patients. They have further noted that high expression of miR-200a/141 enhances sensitivity to paclitaxel treatment, a chemotherapeutic drug known to increase the accumulation of reactive oxygen species (ROS).

On this basis, the inventors have thus herein identified a miR-200a-dependent dual gene signature involved in oxidative stress and fibrosis which allows the stratification of patients into two groups: a first group thereafter referred to as "stress" pattern (high expression of "stress" genes and low expression of "fibrosis" genes) and a second group thereafter referred to as "fibrosis" pattern (high expression of "fibrosis" genes and low expression of "stress" genes). Patients of the first group have longer progression-free interval and overall survival and are more sensitive to a treatment, in particular to a treatment inducing an acute oxidative stress, than those of the second group.

Interestingly, the inventors have further demonstrated that two genes of this signature, namely GSR and MYL9 (whose expression is positively and negatively- correlated with miR- 200a, respectively), are sufficient and essential to stratify the patients in these two groups.

It is important to note that this patient stratification cannot be performed only based on the miR200a/141 expression due to the uncertain correlation between the expression of miR- 200s and the prognosis (Nam et al, 2008; Hu et al, 2009). This uncertainty was further illustrated herein. Indeed, in the experimental section, the inventors have shown that stratification of patients could not be performed based on the expression rate of either miR- 141 or miR-200a due to non-significant p-values (Figures 19 and 20). In order to avoid potential variations due to inaccuracies in miRNA measurements, the inventors have also considered the average of miR- 141 and miR-200a expression values to define prognosis. However, results were not statistically significant (Figure 21, p-values: 0.29 and 0.053). These results demonstrate that the expression rates of miR- 141 or miR-200a could not be used as satisfying predictive markers, thereby explaining why controversial results were previously published on such issue. Definitions

The methods of the invention as disclosed herein, may be in vivo, ex vivo or in vitro methods. Preferably, the methods of the invention are in vitro methods.

As used herein, the term "subject" or "patient" refers to an animal, preferably to a mammal, even more preferably to a human, including adult, child and human at the prenatal stage. However, the term "subject" can also refer to non-human animals, in particular mammals such as dogs, cats, horses, cows, pigs, sheeps and non-human primates, among others, that are in need of treatment.

The term "cancer" or "tumor", as used herein, refers to the presence of cells possessing characteristics typical of cancer-causing cells, such as uncontrolled proliferation, immortality, metastatic potential, rapid growth and proliferation rate, and certain characteristic morphological features. This term refers to any type of malignancy (primary or metastases). Typical cancers are solid or hematopoietic cancers such as ovarian, cervix uteri, endometrium, vulva, vagina, breast, stomach, oesophageal, sarcoma, bladder, rectum, colon, lung or ORL cancers, paediatric tumours (neuroblastoma, glioblastoma multiforme), lymphoma, leukaemia, myeloma, seminoma, Hodgkin and malignant hemopathies. Preferably, the term "cancer" refers to solid cancers. In a particular embodiment, the cancer is selected from the group consisting of gynaecologic cancer such as ovarian cancer, cervical cancer, endometrial cancer, vulval cancer and vaginal cancer, as well as other type of cancers such as lung cancer, prostate cancer, pancreas cancer, colorectal cancer or leukemia. Preferably, the cancer is gynaecologic cancer. In a preferred embodiment, the cancer is ovarian cancer, more particularly ovarian adenocarcinoma (also named surface epithelial- stromal tumor). The term "ovarian adenocarcinoma" includes serous tumors, endometrioid tumors, mucinous cystadenocarcinomas, clear-cell ovarian tumors and Brenner tumors.

The term "surgical debulking", as used herein, refers to as surgical removal of part of a malignant tumour which cannot be completely excised, in order to make subsequent therapy with drugs, radiation or other adjunctive measures more effective. As used herein, the term "optimal debulking" refers to a surgery resulting in tumor residue of less than 1 cm in diameter after resection. The term "partial debulking" refers to a surgery resulting in tumor residue of more than 1 cm in diameter after resection.

As used herein, the term "treatment", "treat" or "treating" refers to any act intended to ameliorate the health status of patients such as therapy, prevention, prophylaxis and retardation of the disease. In certain embodiments, such term refers to the amelioration or eradication of a disease or symptoms associated with a disease. In other embodiments, this term refers to minimizing the spread or worsening of the disease resulting from the administration of one or more therapeutic agents to a subject with such a disease. This term refers to the treatment at any stage of the disease. In particular, it can be an adjuvant therapy (chemo- or radiotherapy after surgery) or a neo-adjuvant therapy (chemo- or radiotherapy before surgery). In particular, the term "to treat a cancer" or "treating a cancer" means reversing, alleviating, inhibiting the progress of, or preventing, either partially or completely, the growth of tumors, tumor metastases, or other cancer-causing or neoplastic cells in a patient.

The term "sample", as used herein, means any sample containing cells derived from a subject, preferably a sample which contains nucleic acids. Examples of such samples include fluids such as blood, plasma, saliva, urine and seminal fluid samples as well as biopsies, organs, tissues, cell samples or cancer associated ascite fluids. The sample may be treated prior to its use. The term "sample" may also refer to any sample containing free circulating nucleic acids. The term "cancer sample" refers to any sample comprising tumor cells derived from a patient, preferably a sample which comprises nucleic acids. Preferably, the sample contains only tumor cells (i.e., no normal or healthy cell). The term "cancer sample" may also refer to any sample comprising free circulating nucleic acids from tumor cells. Preferably, the sample contains only nucleic acids from tumor cells.

As used herein, the term "good prognosis" refers to a longer progression free survival and/or an increased patient survival. On the other hand, the term "poor prognosis" refers to an early disease progression and/or decreased patient survival.

As used herein, the term "progression free survival" refers to the time interval between the date of diagnosis and the first confirmed sign of disease recurrence. The term "patient survival" refers to the time interval between the date of diagnosis and the date of death.

By "good responder" is intended a patient who shows a good therapeutic benefit of the treatment, that is to say a longer disease free survival, a longer overall survival, a decreased metastasis occurrence, a decreased tumor growth and/or a tumor regression in comparison to a population of patients suffering from the same cancer and having the same treatment.

Alternatively, by "poor responder" is intended a patient who shows a weak therapeutic benefit of the treatment, that is to say a shorter disease free survival, a shorter overall survival, an increased metastasis occurrence and/or an increased tumor growth in comparison to a population of patients suffering from the same cancer and having the same treatment. The inventors have herein identified a dual gene expression signature exhibiting predictive value regarding the clinical outcome of the patient. They have demonstrated that high expression level of GSR gene involved in the NRF2-mediated oxidative stress response, associated with low expression level of MYL9 gene involved in hepatic fibrosis, are correlated with good prognosis. On the contrary, under-expression of GSR gene and over-expression of MYL9 gene, are associated with poor prognosis.

Accordingly, in a first aspect, the present invention discloses a method for predicting clinical outcome of a patient affected with a cancer, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a high expression level of GSR gene and a low expression level of MYL9 gene being indicative of a good prognosis, i.e. a longer progression free survival and/or an increased patient survival.

The present invention also discloses a method for predicting clinical outcome of a patient affected with a cancer, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a low expression level of GSR gene and a high expression level of MYL9 gene being indicative of a poor prognosis, i.e. early disease progression and/or decreased patient survival.

GSR gene (Gene ID: 2936) encodes glutathione reductase, an enzyme which reduces oxidized glutathione disulfide to the sulfhydryl form and is a central enzyme of cellular antioxidant defense. Multiple alternatively spliced transcript variants encoding different iso forms have been found (Genbank accessions: NM 000637.3 (SEQ ID NO.:45), NM 001195102.1 (SEQ ID NO.:46), NM 001195103.1 (SEQ ID NO.: 47) and NM 001195104.1 (SEQ ID NO.: 48)).

MYL9 gene (Gene ID: 10398) encodes a myosin light chain that may regulate muscle contraction by modulating the ATPase activity of myosin heads. Two transcript variants encoding different iso forms have been found for this gene (Genbank accessions: NM 006097.3 (SEQ ID NO.:49) and NM l 81526.1 (SEQ ID NO.:50)).

The inventors have further demonstrated that this dual signature may also comprise additional genes involved in the NRF2-mediated oxidative stress response or involved in hepatic fibrosis.

Thus, in an embodiment, the method of the invention further comprises determining in the cancer sample from the patient, the expression level of one or several additional genes involved in the NRF2-mediated oxidative stress response and selected from the group consisting of DNAJC10, PIK3R3, PRDXl, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKRIAI, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHX1 and STIP1, preferably selected from the group consisting of AKRIAI, PRKCZ and NQOl, more preferably selected from the group consisting of AKRIAI and NQOl, and/or the expression level of one or several additional genes involved in hepatic fibrosis and selected from the group consisting of IGF1, TIMP2, FNl, TNFRSFllB, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB, preferably selected from the group consisting of TIMP2 and FNl. In a particular embodiment, the method of the invention further comprises determining in the cancer sample from the patient, the expression level of one or several additional genes selected from the group consisting of AKRIAI, PRKCZ and NQOl, preferably selected from the group consisting of AKRIAI and NQOl, and/or the expression level of one or several additional genes selected from the group consisting of TIMP2 and FNl. In a further particular embodiment, the method of the invention further comprises determining in the cancer sample from the patient, the expression level of AKRIAI, NQOl, PRKCZ, TIMP2 and FNl genes, preferably the expression level of AKRIAI, NQOl, TIMP2 and FNl genes. In a further particular embodiment, the method of the invention further comprises determining in the cancer sample from the patient, the expression level of AKRIAI and TIMP2 genes. In these embodiments, a high expression of genes involved in the NRF2-mediated oxidative stress response, as defined above, associated with a low expression of genes involved in hepatic fibrosis, as defined above, define the population with the "stress" pattern and is thus indicative of a good prognosis. On the other hand, a low expression of genes involved in the NRF2-mediated oxidative stress response, as defined above, associated with a high expression of genes involved in hepatic fibrosis, as defined above, define the population with the "fibrosis" pattern and is thus indicative of a poor prognosis.

The inventors have further identified additional genes that are expressed at high level in the population with the "stress" pattern and expressed at low level in the population with the "fibrosis" pattern, and thus for which high expression level is indicative of a good prognosis. Inversely, they have also identified additional genes that are expressed at high level in the population with the "fibrosis" pattern and expressed at low level in the population with the "stress" pattern, and thus for which high expression level is indicative of a poor prognosis.

Accordingly, in another embodiment, the method of the invention further comprises determining in the cancer sample from the patient, the expression level of one or several additional genes selected from the group consisting of (i) genes that are expressed at high level in the population with the "stress" pattern and selected from the group consisting of ACAA1, ACAD8, ACADM, ACADSB, AC02, AIFM1, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP5I, ATP5J2, ATP50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBR1, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, DLD, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTKl, HADH, HIBADH, HMGCL, HMGCSl, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT-C02, MT-ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA 7, NDUFA8, NDUFAF1, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHA1, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAK1, UQCR11, UQCRC1, UQCRQ, VCP and XDH, and (ii) genes that are expressed at low level in the population with the "stress" pattern and selected from the group consisting of MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYH11, IGF1R, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, IL1RAP, IGFBP3 and VCAM1.

Alternatively, the method may comprise (i) determining in a sample from said patient the expression level of GSR and MYL9 genes; (ii) performing cluster analysis on standardized values of the expression levels of said genes in said sample and in a population of randomly selected samples, wherein the classification of the cancer sample from the patient in the cluster in which the expression level of GSR gene is above the mean value, and in which the expression level ofMYL9 gene is below the mean value, is indicative of a good prognosis, i.e. a longer progression free survival and/or an increased patient survival.

On the other hand, the classification of the cancer sample from the patient in the cluster in which the expression level of GSR gene is below the mean value, and in which the expression level ofMYL9 gene is above the mean value, is indicative of a poor prognosis, i.e. early disease progression and/or decreased patient survival.

In this embodiment, the method may further comprise determining in the cancer sample from the patient, and optionally in the population of randomly selected cancer samples, the expression level of one or several additional genes selected from the group consisting of DNAJC10, PIK3R3, PRDX1, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKR1A1, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHXl, STIPl, IGFl, TIMP2, FNl, TNFRSFllB, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB, preferably selected from the group consisting of AKR1A1, TIMP2, PRKCZ, NQOl and FNl, more preferably selected from the group consisting of AKR1A1, TIMP2, NQOl and FNl, even more preferably selected from the group consisting oiAKRIAl and TIMP2, and including standardized values of these expression levels to perform the cluster analysis.

The classification of the cancer sample from the patient in the cluster in which expression levels of the group of genes comprising GSR, DNAJC10, PIK3R3, PRDX1, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKR1A1, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHX1 and STIP1, are above the mean value, and in which expression levels of the group of genes comprising MYL9, IGF1, TIMP2, FN1, TNFRSF11B, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB, are below the mean value, is indicative of a good prognosis.

On the other hand, the classification of the cancer sample from the patient in the cluster in which expression levels of the group of genes comprising GSR, DNAJC10, PIK3R3, PRDXl, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKR1A1, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHX1 and STIP1, are below the mean value, and in which expression levels of the group of genes comprising MYL9, IGF1, TIMP2, FN1, TNFRSF11B, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB, are above the mean value, is indicative of a poor prognosis

In this embodiment, the method may also comprise determining in the cancer sample from the patient, and optionally in the population of randomly selected cancer samples, the expression level of one or several additional genes selected from the group consisting of ACAA1, ACAD8, ACADM, ACADSB, AC02, AIFM1, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP5I, ATP5J2, ATP50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBR1, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, DLD, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTKl, HADH, HIBADH, HMGCL, HMGCSl, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT-C02, MT-ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA 7, NDUFA8, NDUFAF1, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHA1, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAK1, UQCR11, UQCRC1, UQCRQ, VCP, XDH, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYHll, IGFIR, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, IL1RAP, IGFBP3 and VCAM1, and including standardized values of these expression levels to perform the cluster analysis.

The classification of the cancer sample from the patient in the cluster in which expression levels of the group of genes comprising GSR, ACAA1, ACAD8, ACADM, ACADSB, AC02, AIFM1, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP5I, ATP5J2, ATP 50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBR1, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, DLD, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTKl, HADH, HIBADH, HMGCL, HMGCSl, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT-C02, MT-ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA7, NDUFA8, NDUFAF1, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHA1, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAK1, UQCR11, UQCRC1, UQCRQ, VCP and XDH„ are above the mean value, and in which expression levels of the group of genes comprising MYL9, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYH11, IGF1R, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, IL1RAP, IGFBP3 and VCAM1, are below the mean value, is indicative of a good prognosis.

On the other hand, the classification of the cancer sample from the patient in the cluster in which expression levels of the group of genes comprising GSR, ACAA1, AC ADS, ACADM, ACADSB, AC02, AIFM1, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP 51, ATP5J2, ATP 50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBR1, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, DLD, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTKl, HADH, HIBADH, HMGCL, HMGCSl, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT-C02, MT-ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA 7, NDUFA8, NDUFAF1, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHA1, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAK1, UQCR11, UQCRC1, UQCRQ, VCP and XDH„ are below the mean value, and in which expression levels of the group of genes comprising MYL9, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYH11, IGF1R, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, IL1RAP, IGFBP3 and VCAM1, are above the mean value, is indicative of a poor prognosis. The method of the invention may further comprise providing a cancer sample from the patient. Optionally, the nucleic acids of the sample may be purified or isolated from the sample. Alternatively, the proteins of the sample may also be purified or isolated from the sample. The expression level of each gene can be determined from a cancer sample by a variety of techniques. In particular, the expression level of each gene may be determined by measuring the quantity or the activity of encoded protein and/or by measuring the quantity of mRNA.

In an embodiment, the expression level of a gene is determined by measuring the quantity or the activity of the encoded protein. The quantity or the activity of the encoded protein may be measured by any methods known by the skilled person and the choice of the method depends on the encoded protein. Usually, these methods comprise contacting the sample with a binding partner capable of selectively interacting with the protein present in the sample. The binding partner is generally a polyclonal or monoclonal antibody, preferably monoclonal. In particular, antibodies specifically recognizing human GSR or MYL9 proteins are commercially available. Preferably, the quantity of protein is measured by semiquantitative Western blots, immunochemistry (enzyme-labeled and mediated immunoassays, such as ELISAs, biotin/avidin type assays, radioimmunoassay, Immunoelectrophoresis or immunoprecipitation) or by protein or antibody arrays. The protein expression level may be assessed by immunohistochemistry on a tissue section of the cancer sample (e.g. frozen or formalin- fixed paraffin embedded material). The reactions generally include revealing labels such as fluorescent, chemiluminescent, radioactive, enzymatic labels or dye molecules, or other methods for detecting the formation of a complex between the antigen and the antibody or antibodies reacted therewith. Specific activity assays may also be used, in particular when the encoded protein is an enzyme.

Glutathione reductase (GSR) increases the glutathione (GSH) pool by reducing oxidized glutathione (GSSG) to GSH and thus increases the GSH/GSSG ratio, which can be easily measured. The activity can be monitored by the NADPH consumption, with absorbance at 340 nm, or the formed GSH can be visualized by Ellman's reagent (Smith et al. 1988). Alternatively the activity can be measured using redox-sensitive Green Fluorescent Protein namely roGFP (Marty et al, 2009).

In another embodiment, the expression level of a gene is determined by measuring the quantity of mRNA. Methods for determining the quantity of mRNA are well known in the art. For example the nucleic acid contained in the sample (e.g., cells or tissue prepared from the patient) is first extracted according to standard methods, for example using lytic enzymes or chemical solutions or extracted by nucleic-acid-binding resins following the manufacturer's instructions. The extracted mRNA may be then detected by hybridization (e.g., Northern blot analysis) and/or amplification (e.g., RT-PCR). Quantitative or semi-quantitative RT-PCR is preferred. Real-time quantitative or semi-quantitative RT-PCR is particularly advantageous. Preferably, primer pairs were designed in order to overlap an intron, so as to distinguish cDNA amplification from putative genomic contamination. Such primers may be easily designed by the skilled person. An example of primer pair which may be used to detect GSR mRNA is constituted by the forward primer 5 '-CACTTGCGTGAATGTTGGATG-3 ' (SEQ ID No.: 1) and the reverse primer 5 '-TGGGATCACTCGTGAAGGCT-3 ' (SEQ ID No.: 2) (PrimerBank, http://pga.mgh.harvard.edu/primerbank/). Another example of primer pair which may be used to detect GSR mRNA is constituted by the forward primer 5 '- GCACTTGCGTGAATGTTGGAT-3 ' (SEQ ID NO.: 27) and the reverse primer 5'- GGCTTGGGATC ACTCGTGAA-3 ' (SEQ ID NO.: 28). An example of primer pair which may be used to detect MYL9 mRNA is constituted by the forward primer 5'- CATCCATGAGGACCACCTCCG-3 ' (SEQ ID No.: 3) and the reverse primer 5'- CTGGGGTGGCCTAGTCGTC-3 ' (SEQ ID No.: 4) (Medjkane et al, 2009). Another example of primer pair which may be used to detect MYL9 mRNA is constituted by the forward primer 5'-AGTTCCACGCACCCAGCGA-3' (SEQ ID NO.: 29) and the reverse primer 5 '-TTGCTGGACATCTTGGCTTCTGGT-3 ' (SEQ ID NO.: 30). Other methods of Amplification include, but are not limited to, ligase chain reaction (LCR), transcription- mediated amplification (TMA), strand displacement amplification (SDA) and nucleic acid sequence based amplification (NASBA). Alternatively, the quantity of mRNA may also be measured using the Nanostring's NCOUNTER™ Digital Gene Expression System (Geiss et al, 2008) which captures and counts individual mRNA transcripts by a molecular bar-coding technology and is commercialized by Nanostring Technologies, or the QuantiGene ^® Plex 2.0 Assay (Affymetrix). The quantity of mRNA may further be determined using approaches based on high-throughput sequencing technology such as RNA-Seq (Wang et al, 2009).

The expression level of each gene may also be determined by measuring the quantity of mRNA by transcriptome approaches, in particular by using DNA microarrays. To determine the expression level of each gene, the samples, optionally first subjected to a reverse transcription, are labelled and contacted with the microarray in hybridization conditions, leading to the formation of complexes between target nucleic acids that are complementary to probe sequences attached to the microarray surface. The labelled hybridized complexes are then detected and can be quantified or semi-quantified. Labelling may be achieved by various methods, e.g. by using radioactive or fluorescent labelling. Many variants of the microarray hybridization technology are available to the man skilled in the art. Examples of DNA biochips suitable to measure the expression level of the genes of interest include, but are not limited to, Human Genome U133 Plus 2.0 array (Affymetrix), Human Gene 1.0 ST Array (Affymetrix) and Human Exon 1.0 ST (Affymetrix). Next Generation Sequencing methods (NGS) may also be used.

Preferably, the quantity of mRNA is measured by quantitative or semi-quantitative

RT-PCR, by real-time quantitative or semi-quantitative RT-PCR, by Nanostring technology or sequencing based approaches, or by transcriptome approaches. More preferably, the quantity of mRNA is measured by quantitative or semi-quantitative RT-PCR, by real-time quantitative or semi-quantitative RT-PCR or Nanostring technology.

The method may further comprise comparing the expression level of GSR and MYL9 genes, and optionally, of any other additional gene as defined above, to a reference expression level.

Optionally, before to be compared with the reference expression level, the expression levels of genes, in particular of GSR and MYL9 genes, are normalized using the expression level of an endogenous control gene having a stable expression in different cancer samples, such as RPLPO, HPRT1, GAPDH and B2M genes.

In an embodiment, the reference expression level is the expression level of a gene having a stable expression in different cancer samples, such as RPLPO, HPRT1, GAPDH and B2M genes.

In another embodiment, the reference expression level is specific of each gene and is the mean expression level of said gene, e.g. GSR or MYL9 gene, among a population of randomly selected cancer samples or among a validated reference population, i.e. a population which belongs to the "stress" pattern group or to the "fibrosis" pattern group. For example, the reference expression level used to evaluate the expression level of GSR gene is the mean expression level of GSR gene among a population of randomly selected cancer samples or among a validated reference population.

The method may further comprise determining whether the expression level of GSR and MYL9 genes, and optionally, of any other additional gene as defined above, is high or low compared to the reference expression level.

If the reference expression level is the expression level of a gene having a stable expression in different cancer samples, such as RPLPO, HPRT1, GAPDH and B2M genes, the expression level of the gene of interest is considered as high if the level or quantity of mRNA is above a cut-off value easily adjusted by the skilled person depending on the gene of interest, the reference gene and the type of cancer. The cut-off value may be defined according to the average and the variance of the expression rates of each gene in each population ("stress" pattern versus "fibrosis" pattern).

If the reference expression level is the mean expression level of the gene of interest among a population of randomly selected cancer samples, the expression level of the gene of interest is considered as high if the expression level (or quantity of mRNA) is higher than the mean expression level of said gene among a population of randomly selected cancer samples. On the other hand, the expression level of the gene of interest is considered as low if the expression level (or quantity of mRNA) is lower than the mean expression level of said gene among a population of randomly selected cancer samples.

As used herein, the term "population of randomly selected cancer samples" refers to a population comprising at least 50 samples, more preferably at least 100, 200 or 250 samples. In a preferred embodiment, the population of randomly selected cancer samples contains only one type of tumors (i.e. tumors derived from the cells of a same organ), preferably of the same type than the tumor of the patient. In a particular embodiment, the patient is affected with an ovarian cancer and the population of randomly selected cancer samples comprises only ovarian cancer samples.

Expression levels of the random population may be obtained from transcriptome databases such as Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo). Examples of datasets comprising transcriptome data of a population of randomly selected ovarian cancer samples include, but are not limited, to microarray datasets with GEO accession numbers GSE 9891, GSE26193 and GSE3149. GSE 9891 corresponds to the expression profile of 285 randomly selected ovarian tumour samples from the AOCS (Australian Ovarian Cancer Study). GSE 26193 corresponds to the expression profile of 107 randomly selected ovarian tumour samples from the Biological Resource Center of the Institut Curie. GSE3149 corresponds to the expression profile of 153 randomly selected ovarian tumour samples from the Duke Cancer Institute (Bild et al., 2006).

In embodiments wherein cluster analysis is performed, expression levels of the genes are standardized, for example by centring values on the mean and rescaling them to have a standard deviation of 1. The cluster analysis is then performed on these standardized values. The clustering may be performed using any method known by the skilled person.

In an embodiment, the cluster analysis is a hierarchical clustering. Preferably, this clustering is performed using standard Pearson's correlation as similarity measure and the average linkage criteria or the Ward method as clustering method. The opposite main tree branches of the hierarchical clustering are used to segregate two distinct clusters of tumors.

In another embodiment, the cluster analysis is a portioning clustering. Preferably, this clustering is performed using Pearson's dissimilarity as distance function and data partition into two clusters. The analysis may be performed, for example, using Partek Genomic Suite software.

The method of the invention may further comprise assessing at least one other prognosis markers such as histological sub-type, tumor grade, tumor stage, p53 status, BRCAl/2 status, mitotic index, tumor size or extent of residual disease after surgery, or expression of proliferation markers such as Ki67, MCM2, CAF-1 p60 and CAF-1 pi 50.

The present invention also relates to a method for providing useful information for predicting clinical outcome of a patient affected with a cancer, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a high expression level of GSR gene and a low expression level of MYL9 gene being indicative of a good prognosis. All the embodiments of the method for predicting clinical outcome of a patient as described above are also contemplated in this aspect.

Based on the dual gene expression signature identified by the inventors, tumors can be segregated in two clusters: (a) tumors with the "stress" pattern, i.e. tumors in which the expression level of GSR gene is high and in which the expression level of MYL9 gene is low, and (b) tumors with the "fibrosis" pattern, i.e. tumors in which the expression level of GSR gene is low and in which the expression level of MYL9 gene is high.

This clustering can also be done using cluster analysis as disclosed above. In this case, tumors are segregated into two clusters: (a) tumors with the "stress" pattern, i.e. tumors in which the expression level of GSR gene is above the mean value and in which the expression levels of MYL9 gene is below the mean value; and (b) tumors with the "fibrosis" pattern, i.e. tumors in which the expression level of GSR gene is below the mean value and in which the expression level of MYL9 gene is above the mean value. Additional genes as defined above can also be used to segregate tumors in these two clusters.

The inventors have demonstrated that patients with "stress" pattern exhibited longer progression free and overall survival than those who displayed the "fibrosis" pattern. The classification of the cancer sample from the patient in the cluster of tumors exhibiting "stress" pattern or "fibrosis" pattern, is thus used to predict the clinical outcome of the patient. The classification of the cancer sample from the patient in the cluster of tumors exhibiting "stress" pattern is indicative of a longer progression free survival and/or increased patient survival. On the contrary, the classification of the cancer sample from the patient in the cluster of tumors exhibiting "fibrosis" pattern is indicative of an early disease progression and/or decreased patient survival. In particular, the inventors showed that the clinical outcome does not depend on the chemotherapy used to treat the patients, as shown in Figure 8. Indeed, various chemotherapies have been applied. The inventors have herein demonstrated that the dual gene expression signature, as disclosed above, exhibits predictive value regarding the response of the tumor to a treatment including surgical debulking and/or chemotherapy inducing acute oxidative stress. They have shown that tumors with the "stress" pattern, i.e. tumors with a high expression level of GSR gene and low expression level of MYL9 gene, were significantly more sensitive to paclitaxel, a ROS-producing chemotherapeutic treatment, than tumors with the "fibrosis" pattern. The results have been further confirmed in vivo. They have further observed that patients with the "stress" pattern exhibit optimal debulking.

Some classes of antineoplastic agents, including anthracyclines, taxanes, alkylating agents, platinum-based drugs, epipodophyllotoxins such as etoposide and teniposide, and camptothecins such as irinotecan and topotecan, are known to generate a high level of oxidative stress in biological systems. Oxidative stress represents an imbalance between the production and manifestation of reactive oxygen species and the ability of the organism to readily detoxify the reactive intermediates or to repair the resulting damage. Radiotherapy also induces acute oxidative stress as a result of the formation of intracellular free radicals.

Thanks to the dual gene expression signature as disclosed herein, it is now possible to select patient affected with a tumor which is particularly sensitive to treatment, especially chemotherapy inducing acute oxidative stress. Furthermore, for patients having the "stress" pattern, dose-intensive and/or dose-dense chemotherapy is particularly indicated because, in this case, the balance between efficacy of the treatment and toxicity is in the patients' favour.

Accordingly, in a second aspect, the present invention discloses a method for selecting a patient affected with a cancer for surgical debulking and/or a dose-intensive and/or dose- dense chemotherapy, a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, or for determining whether a patient affected with a cancer is susceptible to benefit from surgical debulking and/or a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a high expression level of GSR gene and a low expression level of MYL9 gene being indicative that said patient is susceptible to benefit from surgical debulking and/or a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells. The present invention also discloses a method for selecting a patient affected with a cancer for surgical debulking followed by a dose- intensive and/or dose-dense chemotherapy, a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, preferably followed by a chemotherapy inducing acute oxidative stress in cancer cells, or for determining whether a patient affected with a cancer is susceptible to benefit from surgical debulking followed by a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, preferably followed by a chemotherapy inducing acute oxidative stress in cancer cells, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a high expression level of GSR gene and a low expression level of MYL9 gene being indicative that said patient is susceptible to benefit from surgical debulking followed by a dose-intensive and/or dose-dense chemotherapy, a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, preferably followed by a chemotherapy inducing acute oxidative stress in cancer cells.

In an embodiment, the method further comprises determining in the cancer sample from the patient, the expression level of one or several additional genes selected from the group consisting of DNAJCIO, PIK3R3, PRDXl, DNAJA4, MAP2K6, UBE2K, DNAJCll, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKR1A1, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHX1, STIP1, IGF1, TIMP2, FN1, TNFRSF11B, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB, preferably selected from the group consisting of AKR1A1, TIMP2, PRKCZ, NQOl and FN1, more preferably selected from the group consisting of AKR1A1, TIMP2, NQOl and FN1, even more preferably selected from the group consisting of AKR1A1 and TIMP2. In this embodiment, high expression levels of genes from the group consisting of DNAJC10, PIK3R3, PRDX1, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKR1A1, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHX1 and STIP1, associated with low expression levels of genes from the group consisting of IGF1, TIMP2, FN1, TNFRSF11B, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB, is indicative that the patient is susceptible to benefit from surgical debulking and/or dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells.

In another embodiment, the method further comprises determining in the cancer sample from the patient, the expression level of one or several additional genes selected from the group consisting of ACAA1, ACAD8, ACADM, ACADSB, AC02, AIFMl, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP 51, ATP5J2, ATP 50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBRl, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, DLD, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTK1, HADH, HIBADH, HMGCL, HMGCS1, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT- C02, MT-ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA7, NDUFA8, NDUFAFl, NDUFB3, NDUFB4, NDUFSl, NDUFS7, NQ02, OGDH, PARK7, PDHAl, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAK1, UQCR11, UQCRC1, UQCRQ, VCP, XDH, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYH11, IGF1R, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, ILIRAP, IGFBP3 and VCAM1. In this embodiment, high expression levels of genes from the group consisting o CAAl, ACAD8, ACADM, ACADSB, AC02, AIFMl, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP5I, ATP5J2, ATP50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBRl, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, DLD, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTK1, HADH, HIBADH, HMGCL, HMGCS1, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT-C02, MT- ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA7, NDUFA8, NDUFAFl, NDUFB3, NDUFB4, NDUFSl, NDUFS7, NQ02, OGDH, PARK7, PDHAl, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAKl, UQCR11, UQCRC1, UQCRQ, VCP and XDH, associated with low expression levels of genes from the group consisting of MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYH11, IGF1R, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, IL1RAP, IGFBP3 and VCAM1, is indicative that the patient is susceptible to benefit from surgical debulking and/or a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells.

In an alternative embodiment, the method comprises (i) determining in a cancer sample from said patient the expression level of GSR and MYL9 genes, (ii) performing hierarchical cluster analysis on standardized values of the expression levels of said genes in said sample and in a population of randomly selected cancer samples, wherein the classification of the cancer sample from the patient in the cluster in which the expression level of GSR gene is above the mean value, and in which the expression level of MYL9 gene is below the mean value, is indicative that said patient is susceptible to benefit from surgical debulking and/or a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells.

In this embodiment, the method may further comprise determining in the cancer sample from the patient, and optionally in the population of randomly selected cancer samples, the expression level of one or several additional genes selected from the group consisting of DNAJC10, PIK3R3, PRDX1, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGSTl, HSP90AA1, AKRIAI, PRKCH, AKTl, TXN, SODl, GSK3B, NRAS, EPHXl, STIPl, IGF1, TIMP2, FN1, TNFRSF11B, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB, preferably selected from the group consisting of AKRIAI, TIMP2, PRKCZ, NQOl and FN1, more preferably selected from the group consisting of AKRIAI, TIMP2, NQOl and FN1, even more preferably selected from the group consisting of AKRIAI and TIMP2, and including standardized values of these expression levels to perform the cluster analysis.

The classification of the cancer sample from the patient in the cluster in which expression levels of the group of genes comprising GSR, DNAJC10, PIK3R3, PRDXl, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGSTl, HSP90AA1, AKRIAI, PRKCH, AKTl, TXN, SODl, GSK3B, NRAS, EPHXl and STIPl, are above the mean value, and in which expression levels of the group of genes comprising MYL9, IGF1, TIMP2, FN1, TNFRSF11B, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB, are below the mean value, is indicative that said patient is susceptible to benefit from surgical debulking and/or a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells.

In this embodiment, the method may also comprise determining in the cancer sample from the patient, and optionally in the population of randomly selected cancer samples, the expression level of one or several additional genes selected from the group consisting of ACAA1, ACAD8, ACADM, ACADSB, AC02, AIFMl, AKR7A2, AKR7A3, ALDH6A1, APHIA, ATP5D, ATP5G3, ATP5I, ATP5J2, ATP50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBR1, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, DID, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTKl, HADH, HIBADH, HMGCL, HMGCSl, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT-C02, MT-ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA 7, NDUFA8, NDUFAF1, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHA1, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAK1, UQCR11, UQCRC1, UQCRQ, VCP, XDH, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYHll, IGFIR, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, ILIRAP, IGFBP3 and VCAM1, and including standardized values of these expression levels to perform the cluster analysis.

The classification of the cancer sample from the patient in the cluster in which expression levels of the group of genes comprising GSR, ACAA1, AC ADS, ACADM, ACADSB, AC02, AIFMl, AKR7A2, AKR7A3, ALDH6A1, APHIA, ATP5D, ATP5G3, ATP5I, ATP5J2, ATP 50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBR1, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, DLD, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTKl, HADH, HIBADH, HMGCL, HMGCSl, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT-C02, MT-ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA7, NDUFA8, NDUFAF1, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHA1, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAK1, UQCR11, UQCRC1, UQCRQ, VCP and XDH,, are above the mean value, and in which expression levels of the group of genes comprising MYL9, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYHll, IGFIR, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, ILIRAP, IGFBP3 and VCAM1, are below the mean value, is indicative that said patient is susceptible to benefit from surgical debulking and/or a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells. All the embodiments of the method for predicting clinical outcome of a patient as described above are also contemplated in this aspect.

In an embodiment, the method further comprises the step of providing a cancer sample from the patient. Optionally, the nucleic acids of the sample may be purified or isolated from the sample. Alternatively, the protein of the sample may also be purified or isolated from the sample.

Cluster analysis and classification of the cancer sample from the patient based on expression levels of genes may be performed as disclosed above.

As used herein, the term "chemotherapeutic treatment" or "chemotherapy" refers to a cancer therapeutic treatment using chemical or biochemical substances, in particular using one or several antineoplastic agents.

As used herein, the term "dose-intensive chemotherapy" refers to a chemotherapy treatment plan in which patients receive higher dose of drugs that in a standard chemotherapy treatment plan.

The term "dose-dense chemotherapy", as used herein, refers to a chemotherapy treatment plan in which drugs are given with less time between treatments than in a standard chemotherapy treatment plan. A dose-dense chemotherapy regimen may be also dose-intense, i.e. patients may receive higher total dose of drugs than those who receive standard treatment.

Preferably, the antineoplastic agent is known to induce acute oxidative stress. However, most of the antineoplastic agents generate at least some free radicals as they induce apoptosis in cancer cells. Thus, the administration of such compounds generally results in oxidative stress but to a lesser extent than contemplated in the present invention. In contrast, a dose-dense and/or dose-intensive chemotherapy regimen using such compounds induces an acute oxidative stress which is more beneficial for a patient affected with a tumor exhibiting the "stress" signature, i.e. being particularly sensitive to oxidative stress, than standard chemotherapeutic treatment.

The antineoplastic agent may be selected from the group consisting of taxane drugs, anthracyclines, pyrimidine analogues, alkylating agents, vinca alkaloids, aromatase inhibitors, topoisomerase inhibitors, tamoxifen, methotrexate, acronycin, cytotoxic antibiotics such as actinomycin, mitomycin C, plicamycin, acivicin or bleomycin, therapeutic monoclonal antibodies such as trastuzumab or rituximab, interferons, interleukin-2, piperlongumine and piperlongumine analogs, arsenic trioxide, and a combination thereof.

Examples of taxane drugs include, but are not limited to, paclitaxel, docetaxel, larotaxel, XRP6258, BMS-184476, BMS-188797, BMS-275183, ortataxel, RPR 109881A, RPR 116258, NBT-287, PG-paclitaxel, Nab-Paclitaxel, Tesetaxel, IDN 5390, Taxoprexin, DHA-paclitaxel and MAC-321. Preferably, the taxane drug is selected from the group consisting of paclitaxel, docetaxel and combination thereof. More preferably, the taxane drug is paclitaxel.

Examples of anthracyclines include, but are not limited to, doxorubicin, epirubicin, adriamycin, daunorubicin, aclarubicin, idarubicin, amrubicin, pirarubicin, valrubicin, zorubicin, carminomycin, detorubicin, morpholinodoxorubicin, morpholinodaunorubicin, methoxymorpholinyldoxorubicin and mitoxantrone.

Examples of nucleoside analogues include, but are not limited to, 5-fluorouracil (5- FU), cytarabine, gemcitabine and floxuridine.

Examples of alkylating agents include, but are not limited to, platinum-based chemotherapy drugs, cyclophosphamide, chlorambucil, uramustine, estramustine, ifosfamide, melphalan, bendamustine, carmustine, lomustine, semustine, streptozotocin, busulfan, dacarbazine, procarbazine, altretamine, adozelesin, thiotepa, mitozolomide and temozolomide. Platinum-based chemotherapy drugs may include, but are not limited to, cisplatin, carboplatin, iproplatin, spiroplatin, nedaplatin, oxaliplatin, triplatin tetranitrate and satraplatin. Preferably, the alkylating agent is selected from the group consisting of cisplatin, carboplatin, oxaliplatin, cyclophosphamide and ifosfamide, and any combination thereof.

Examples of vinca alkaloids include, but are not limited to, vinblastine, vinorelbine, vincristine, and vindesine.

Examples of aromatase inhibitors include, but are not limited to, aminoglutethimide, testolactone, anastrozole, letrozole, exemestane, formestane and fadrozole.

Examples of topoisomerase inhibitors include, but are not limited to, irinotecan, topotecan. etoposide phosphate and teniposide.

Examples of cytotoxic antibiotics include, but are not limited to, actinomycin, mitomycin C, plicamycin, acivicin, acodazole or bleomycin.

Piperlongumine is an amide alkaloid that can be extracted from the roots of the Piper plant or produced by organic synthesis (Chatterjee and Dutta, 1967). The use of piperlongumine and piperlongumine analogs to treat cancer has been described for example in the international application WO 2009/114126. Recent studies have further demonstrated that piperlongumine selectively kills cancer cells by increasing the level of ROS and apoptotic cell death (Raj et al, 2011).

In a particular embodiment, the chemotherapy comprises a drug selected from the group consisting of anthracyclines, taxane drugs, alkylating agents, piperlongumine and piperlongumine analogs, and any combination thereof. In a more particular embodiment, the chemotherapy comprises a drug selected from the group consisting of taxane drugs and alkylating agents, and any combination thereof. In a preferred embodiment, the chemotherapy comprises a taxane drug, preferably paclitaxel. In another preferred embodiment, the chemotherapy comprises a drug selected from the group consisting of piperlongumine and piperlongumine analogs.

In another embodiment, the chemotherapy comprises a drug selected from the group consisting of paclitaxel, docetaxel, cisplatin, carboplatin, oxaliplatin, doxorubicin, topotecan, fluorouracil, gemcitabine, cyclophosphamide and ifosfamide, and any combination thereof.

The term "radiotherapeutic treatment" or "radiotherapy" is a term commonly used in the art to refer to multiple types of radiation therapy including internal and external radiation therapies or radio immunotherapy, and the use of various types of radiations including X-rays, gamma rays, alpha particles, beta particles, photons, electrons, neutrons, radioisotopes, and other forms of ionizing radiations. In a particular embodiment, the radiotherapeutic treatment is nanoparticle enhanced radiotherapy. Nanoparticle enhanced radiotherapy involves the use of activatable particles, which can generate free radicals and ROS when excited by radiations such as X rays. As an example, it was recently shown that thio-glucose bound gold nanoparticles enhance ovarian cancer radiotherapy by inducing elevated levels of ROS accumulation (Geng et al, 2011).

As used herein, the term "chemotherapy inducing acute oxidative stress" refers to a drug or a combination of drugs that greatly modulates redox homeostasis of cancer cells through direct or indirect increase of ROS generation. Acute oxidative stress may induce apoptosis or necrosis of cancer cells. Preferably, the chemotherapy inducing acute oxidative stress induces apoptosis of cancer cells.

ROS accumulation may be assessed by any method known by the skilled person, for example using a redox-sensitive fluorescent probe such as CM-H2DCFDA (5-6- chloromethyl-2',7'-dichlorodihydro fluorescein diacetate (Gerald et al, 2004; Laurent et al, 2008). Cell apoptosis may also be assessed by any method known by the skilled person, for example using an annexin A5 affinity assay.

The chemotherapy inducing acute oxidative stress may comprise an antineoplastic agent known to induce the accumulation of ROS. Such antineoplastic agents may include, but are not limited to, piperlongumine and its derivatives (Raj et al, 201 1), lanperisone and its derivatives (Shaw et al., 2011), anthracyclines, taxanes such as paclitaxel (Alexandre et al., 2007), alkylating agents such as cisplatin (Berndtsson et al. 2007), epipodophyllotoxins such as etoposide and teniposide (Oh et al, 2007), camptothecins such as irinotecan and topotecan, and cytotoxic antibiotics such as bleomycin. Preferably, the antineoplastic agent known to induce the accumulation of ROS is selected from the group consisting of piperlongumine and its derivatives, lanperisone and its derivatives and epipodophyllotoxins, in particular etoposide and teniposide.

In a particular embodiment, the antineoplastic agent induces apoptosis of cancer cells directly, or essentially, by increasing ROS levels in cancer cells. Examples of such agent include, but are not limited to, piperlongumine and lanperisone, and any derivatives thereof.

The chemotherapy may also comprise a compound having the ability to alter cellular redox status, in particular having the ability to induce or increase oxidative stress. This compound may have intrinsic antineoplastic activity or may be used in combination with an antineoplastic agent, preferably an antineoplastic agent known to induce acute oxidative stress, as described above.

Examples of such compound include, but are not limited to, tolperisone and related compounds such as artemisinin and semi- synthetic or fully synthetic artemisinin derivatives such as artemether, artesunate, artemisone and the synthetic endoperoxide OZ277 (Singh et al. 2004; Effert, 2005; Effert, 2006); darinaparsin (Qunitas-Cardama et al. 2008); motexafm gadolinium (Magda et al., 2006); menadione (Verrax et al., 2006); shikonin (Mao et al., 2008); paracetamol (or acetaminophen) (Vad et al., 2009; Wolchok et al, 2003); acetylsalicylic acid (aspirin) (Vad et al, 2008); geldanamycin and its derivatives (Fukuyo et al, 2008); 3,7-diaminophenothiazinium redox dyes such as methylene blue (Wondrak, 2007); disulfiram (Fruehauf and Trapp, 2008); polysulfide-based anticancer drugs such as calicheamycin (Nicolaou et al, 1993; Stasi, 2008), varacin (Chatterji et al, 2005; Jacob, 2006), leinamycin, S-deoxyleinamycin (Sivaramakrishnan and Gates, 2008), allylmethyltrisulfide, 2-hydroxyethyltrisulfide, diallytetrasulfide, leptosins A, B, E, and F, sirodesmins B and C, lissoclinotoxin A, and NN,dimethyl-5-(methylthio)varacin (Greer, 2001); diallyldisulfide and diallyltrisulfide (Powolny and Singh, 2008); isothiocyanate organosulfur agents such as sulforaphane, β-phenyethyl-isothiocyanate, benzyl-isothiocyanate and 6-methylsulfinylhexyl-isothiocyanate (Kim et al., 2006; Xiao et al., 2008); electrophilic Michael acceptors such as parthenolide, curcumin, dimethylfumarate, cinnamaldehyde, inuviscolide, cinnamaldehyde and neratinib; L-ascorbate (Chen et al, 2008); superoxide dismutase (SOD) inhibitors such as triethylenetetramine, ATN-224 (choline tetrathiomolybdate), 2-methoxyestradiol; SOD mimetics such as M40403, mangafodipir, cis- FeMPy2P2P, Cu(II) diisopropylsalicylate, Mn(III) tetrakis-(5,10,15,20)-benzoic acid porphyrin, Mn(III) tetrakis(N-ethylpyridinium-2-yl)porphyrin, nitroxide TEMPO (2,2,6,6- tetramethyl-piperidine-l-oxyl) and 4-ferrocenecarboxyl-TEMPO (FC-TEMPO); compounds disturbing the glutathione redox-system such as NOV-002, imexon, L-buthionine-S,R- sulfoximine and PABA/NO; compounds disturbing the thioredoxin system such as PX-12, PMX464, PX-916, chaetocin, gliotoxin; compounds inhibiting NQOl (DT-diaphorase) function such as dicoumarol and ES936; compounds inhibiting APE-Refl (a redox enzyme) function such as E3330, PNRI-299, resveratrol, lucanthone and CRT0044876; elesclomol (Kirshner et al, 2008); 2-deoxyglucose (Aykin-Burns et al, 2009); 3-bromopyruvate (Kim et al, 2008); dichloroacetate (Bonnet et al., 2007); redox inactive vitamine E analogues such as a-tocopherylsuccinate ester (Neuzil et al, 2007); 3,3'-diindolylmethane (Roy et al, 2008); Bz-423 (Johnson et al., 2005); erastin and RSL5 (Simamura et al., 2008); and their derivatives and analogues, and any combination thereof.

Chemotherapy and radiotherapy can be used alone or in combination.

Dose-dense and/or dose-intense chemotherapy, radiotherapy and/or chemotherapy inducing acute oxidative stress, can be used in combination with immunotherapy, surgery, in particular surgical debulking, and/or hormone therapy. Radiotherapy, chemotherapy, immunotherapy or hormone therapy may be given as adjuvant treatment after surgical resection of the tumor or as neoadjuvant treatment before surgery. The present invention also relates to a method for providing useful information for selecting a patient affected with a cancer for surgical debulking and/or a dose-intensive and/or dose-dense chemotherapy, a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, or for determining whether a patient affected with a cancer is susceptible to benefit from surgical debulking and/or a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a high expression level of GSR gene and a low expression level ofMYL9 gene being indicative that said patient is susceptible to benefit from surgical debulking and/or a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells.

The present invention also relates to a method for providing useful information for selecting a patient affected with a cancer for surgical debulking followed by a dose-intensive and/or dose-dense chemotherapy, a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, preferably followed by a chemotherapy inducing acute oxidative stress in cancer cells, or for determining whether a patient affected with a cancer is susceptible to benefit from surgical debulking followed by a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, preferably followed by a chemotherapy inducing acute oxidative stress in cancer cells, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a high expression level of GSR gene and a low expression level of MYL9 gene being indicative that said patient is susceptible to benefit from surgical debulking followed by a dose-intensive and/or dose-dense chemotherapy, a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, preferably followed by a chemotherapy inducing acute oxidative stress in cancer cells.

All the embodiments of the method for predicting clinical outcome of a patient and the methods for selecting a patient as described above, are also contemplated in these aspects.

The inventors have herein showed that patients with the "fibrosis" pattern are associated with partial debulking and incomplete treatment response. These patients are more susceptible to benefit from neo-adjuvant chemotherapy before surgical debulking in order to shrink the tumor and facilitate its removal.

Accordingly, the present invention also discloses a method for selecting a patient affected with a cancer for neo-adjuvant chemotherapy before surgical debulking, or for determining whether a patient affected with a cancer is susceptible to benefit from neoadjuvant chemotherapy before surgical debulking, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a low expression level of GSR gene and a high expression level of MYL9 gene being indicative that said patient is susceptible to benefit from neo-adjuvant chemotherapy before surgical debulking.

In an embodiment, the method further comprises determining in the cancer sample from the patient, the expression level of one or several additional genes selected from the group consisting of DNAJC10, PIK3R3, PRDX1, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKRIAI, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHXl, STIPl, IGFl, TIMP2, FNl, TNFRSFllB, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB, preferably selected from the group consisting of AKRIAI, TIMP2, PRKCZ, NQOl and FNl, more preferably selected from the group consisting of AKRIAI, TIMP2, NQOl and FNl, even more preferably selected from the group consisting of AKRIAI and TIMP2. In this embodiment, low expression levels of genes from the group consisting of DNAJC10, PIK3R3, PRDXl, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKR1A1, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHX1 and STIP1, associated with high expression levels of genes from the group consisting of IGF I, TIMP2, FN1, TNFRSF11B, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB, is indicative that the patient is susceptible to benefit from neo-adjuvant chemotherapy before surgical debulking.

In another embodiment, the method further comprises determining in the cancer sample from the patient, the expression level of one or several additional genes selected from the group consisting of ACAAI, ACAD8, ACADM, ACADSB, AC02, AIFMl, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP 51, ATP5J2, ATP 50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBR1, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPTIB, CYCS, DLD, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTK1, HADH, HIBADH, HMGCL, HMGCS1, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT- C02, MT-ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA7, NDUFA8, NDUFAF1, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHA1, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAKl, UQCRll, UQCRCl, UQCRQ, VCP, XDH, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYHll, IGFIR, ICAMl, TGFBl, FGFRl, TNFRSFIA, TGFBRl, TGFB2, ILIRAP, IGFBP3 and VCAM1. In this embodiment, low expression levels of genes from the group consisting o CAAl, ACAD8, ACADM, ACADSB, AC02, AIFMl, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP5I, ATP5J2, ATP50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBR1, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPTIB, CYCS, DLD, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTK1, HADH, HIBADH, HMGCL, HMGCS1, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT-C02, MT- ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA7, NDUFA8, NDUFAF1, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHA1, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAKl, UQCRll, UQCRCl, UQCRQ, VCP and XDH, associated with high expression levels of genes from the group consisting of MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYHll, IGFIR, ICAMl, TGFBl, FGFRl, TNFRSFIA, TGFBRl, TGFB2, ILIRAP, IGFBP3 and VCAM1, is indicative that the patient is susceptible to benefit from neo-adjuvant chemotherapy before surgical debulking.

In an alternative embodiment, the method comprises (i) determining in a cancer sample from said patient the expression level of GSR and MYL9 genes, (ii) performing hierarchical cluster analysis on standardized values of the expression levels of said genes in said sample and in a population of randomly selected cancer samples, wherein the classification of the cancer sample from the patient in the cluster in which the expression level of GSR gene is below the mean value, and in which the expression level of MYL9 gene is above the mean value, is indicative that said patient is susceptible to benefit from neo- adjuvant chemotherapy before surgical debulking.

In this embodiment, the method may further comprise determining in the cancer sample from the patient, and optionally in the population of randomly selected cancer samples, the expression level of one or several additional genes selected from the group consisting of DNAJC10, PIK3R3, PRDX1, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGSTl, HSP90AA1, AKRIAI, PRKCH, AKTl, TXN, SODl, GSK3B, NRAS, EPHXl, STIPl, IGF1, TIMP2, FN1, TNFRSF11B, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB, preferably selected from the group consisting of AKRIAI, TIMP2, PRKCZ, NQOl and FN1, more preferably selected from the group consisting of AKRIAI, TIMP2, NQOl and FN1, even more preferably selected from the group consisting of AKRIAI and TIMP2, and including standardized values of these expression levels to perform the cluster analysis.

The classification of the cancer sample from the patient in the cluster in which expression levels of the group of genes comprising GSR, DNAJC10, PIK3R3, PRDXl, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGSTl, HSP90AA1, AKRIAI, PRKCH, AKTl, TXN, SODl, GSK3B, NRAS, EPHXl and STIPl, are below the mean value, and in which expression levels of the group of genes comprising MYL9, IGF1, TIMP2, FN1, TNFRSF11B, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB, are above the mean value, is indicative that said patient is susceptible to benefit from neo-adjuvant chemotherapy before surgical debulking.

In this embodiment, the method may also comprise determining in the cancer sample from the patient, and optionally in the population of randomly selected cancer samples, the expression level of one or several additional genes selected from the group consisting of ACAA1, ACAD8, ACADM, ACADSB, AC02, AIFM1, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP5I, ATP5J2, ATP50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBRl, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, OLD, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTK1, HADH, HIBADH, HMGCL, HMGCS1, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT-C02, MT-ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA7, NDUFA8, NDUFAF1, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHAl, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAKl, UQCR11, UQCRC1, UQCRQ, VCP, XDH, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYH11, IGF1R, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, IL1RAP, IGFBP3 and VCAM1, and including standardized values of these expression levels to perform the cluster analysis.

The classification of the cancer sample from the patient in the cluster in which expression levels of the group of genes comprising GSR, ACAA1, AC ADS, AC ADM, ACADSB, AC02, AIFM1, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP5I, ATP5J2, ATP 50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBRl, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, DLD, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTK1, HADH, HIBADH, HMGCL, HMGCS1, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT-C02, MT-ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA 7, NDUFA8, NDUFAF1, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHAl, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAKl, UQCR11, UQCRC1, UQCRQ, VCP and XDH,, are below the mean value, and in which expression levels of the group of genes comprising MYL9, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYH11, IGF1R, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, IL1RAP, IGFBP3 and VCAM1, are above the mean value, is indicative that said patient is susceptible to benefit from neoadjuvant chemotherapy before surgical debulking.

All the embodiments of the method for predicting clinical outcome of a patient as described above are also contemplated in this aspect.

In an embodiment, the method further comprises the step of providing a cancer sample from the patient. Optionally, the nucleic acids of the sample may be purified or isolated from the sample. Alternatively, the protein of the sample may also be purified or isolated from the sample. Cluster analysis and classification of the cancer sample from the patient based on expression levels of genes may be performed as disclosed above. The chemotherapy may be as defined above. In particular, the chemotherapy may comprise one or several antineoplastic agents. These antineoplastic agents may be selected from the group consisting of taxane drugs, anthracyclines, pyrimidine analogues, alkylating agents, vinca alkaloids, aromatase inhibitors, topoisomerase inhibitors, tamoxifen, methotrexate, acronycin, cytotoxic antibiotics such as actinomycin, mitomycin C, plicamycin, acivicin or bleomycin, therapeutic monoclonal antibodies such as trastuzumab or rituximab, interferons, interleukin-2, piperlongumine and piperlongumine analogs, arsenic trioxide, and a combination thereof. In a preferred embodiment, the neo-adjuvant chemotherapy comprises a taxane drug, preferably paclitaxel, and/or an alkylating agent, preferably a platinum-based chemotherapy drug.

The present invention also relates to a method for providing useful information for selecting a patient affected with a cancer for neo-adjuvant chemotherapy before surgical debulking, or for determining whether a patient affected with a cancer is susceptible to benefit from neo-adjuvant chemotherapy before surgical debulking, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a low expression level of GSR gene and a high expression level of MYL9 gene being indicative that said patient is susceptible to benefit from neo-adjuvant chemotherapy before surgical debulking. All the embodiments of the method for predicting clinical outcome of a patient and the methods for selecting a patient as described above, are also contemplated in this aspect.

In a third aspect, the present invention discloses a method for selecting a patient affected with a cancer who will be a good responder to a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a high expression level of GSR gene and a low expression level of MYL9 gene being indicative that said patient is a good responder to a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells.

In an embodiment, the method further comprises determining in the cancer sample from the patient, the expression level of one or several additional genes selected from the group consisting of DNAJC10, PIK3R3, PRDXl, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKR1A1, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHX1, STIP1, IGF1, TIMP2, FN1, TNFRSF11B, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB, preferably selected from the group consisting of AKR1A1, TIMP2, PRKCZ, NQOl and FN1, more preferably selected from the group consisting of AKR1A1, TIMP2, NQOl and FN1, even more preferably selected from the group consisting of AKR1A1 and TIMP2. In this embodiment, high expression levels of genes from the group consisting of DNAJC10, PIK3R3, PRDX1, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKR1A1, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHX1 and STIP1, associated with low expression levels of genes from the group consisting of IGF1, TIMP2, FN1, TNFRSF11B, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB, is indicative that the patient is a good responder to a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells.

In another embodiment, the method further comprises determining in the cancer sample from the patient, the expression level of one or several additional genes selected from the group consisting of ACAA1, ACAD8, ACADM, ACADSB, AC02, AIFM1, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP 51, ATP5J2, ATP 50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBR1, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, DLD, DNAJC3, DNAJC4, DNAJC8, ECHSl, FH, FTHl, GCLC, GCLM, GPD2, GSTKl, HADH, HIBADH, HMGCL, HMGCS1, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT- C02, MT-ND6, NDUFAl, NDUFAll, NDUFA2, NDUFA5, NDUFA6, NDUFA7, NDUFA8, NDUFAF1, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHA1, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAK1, UQCR11, UQCRC1, UQCRQ, VCP, XDH, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYH11, IGF1R, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, IL1RAP, IGFBP3 and VCAMl. In this embodiment, high expression levels of genes from the group consisting o CAAl, ACAD8, ACADM, ACADSB, AC02, AIFM1, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP5I, ATP5J2, ATP50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBR1, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, DLD, DNAJC3, DNAJC4, DNAJC8, ECHSl, FH, FTHl, GCLC, GCLM, GPD2, GSTKl, HADH, HIBADH, HMGCL, HMGCS1, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT-C02, MT- ND6, NDUFAl, NDUFAll, NDUFA2, NDUFA5, NDUFA6, NDUFA7, NDUFA8, NDUFAF1, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHA1, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAK1, UQCR11, UQCRC1, UQCRQ, VCP and XDH, associated with low expression levels of genes from the group consisting of MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYH11, IGF1R, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, IL1RAP, IGFBP3 and VCAM1, is indicative that the patient is a good responder to a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells.

In an alternative embodiment, the method comprises (i) determining in a cancer sample from said patient the expression level of GSR and MYL9 genes, (ii) performing hierarchical cluster analysis on standardized values of the expression levels of said genes in said sample and in a population of randomly selected cancer samples, wherein the classification of the cancer sample from the patient in the cluster in which the expression level of GSR gene is above the mean value, and in which the expression level of MYL9 gene is below the mean value, is indicative that the patient is a good responder to a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells.

In this embodiment, the method may further comprise determining in the cancer sample from the patient, and optionally in the population of randomly selected cancer samples, the expression level of one or several additional genes selected from the group consisting of DNAJC10, PIK3R3, PRDXl, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKR1A1, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHX1, STIP1, IGF I, TIMP2, FNl, TNFRSFllB, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COLlAl, LY96, FGFR2, MYH9 and PDGFRB, preferably selected from the group consisting of AKR1A1, TIMP2, PRKCZ, NQOl and FNl, more preferably selected from the group consisting of AKR1A1, TIMP2, NQOl and FNl, even more preferably selected from the group consisting ofAKRIAl and TIMP2, and including standardized values of these expression levels to perform the cluster analysis.

The classification of the cancer sample from the patient in the cluster in which expression levels of the group of genes comprising GSR, DNAJC10, PIK3R3, PRDXl, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKR1A1, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHX1 and STIP1, are above the mean value, and in which expression levels of the group of genes comprising MYL9, IGF1, TIMP2, FNl, TNFRSFllB, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COLlAl, LY96, FGFR2, MYH9 and PDGFRB, are below the mean value, is indicative that the patient is a good responder to a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells. In this embodiment, the method may also comprise determining in the cancer sample from the patient, and optionally in the population of randomly selected cancer samples, the expression level of one or several additional genes selected from the group consisting of ACAA1, ACAD8, ACADM, ACADSB, AC02, AIFM1, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP5I, ATP5J2, ATP50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBRl, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, OLD, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTK1, HADH, HIBADH, HMGCL, HMGCS1, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT-C02, MT-ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA 7, NDUFA8, NDUFAFl, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHAl, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAKl, UQCR11, UQCRC1, UQCRQ, VCP, XDH, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYH11, IGF1R, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, IL1RAP, IGFBP3 and VCAM1, and including standardized values of these expression levels to perform the cluster analysis.

The classification of the cancer sample from the patient in the cluster in which expression levels of the group of genes comprising GSR, ACAA1, AC ADS, ACADM, ACADSB, AC02, AIFM1, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP5I, ATP5J2, ATP 50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBRl, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, DLD, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTK1, HADH, HIBADH, HMGCL, HMGCS1, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT-C02, MT-ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA7, NDUFA8, NDUFAFl, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHAl, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAKl, UQCR11, UQCRC1, UQCRQ, VCP and XDH,, are above the mean value, and in which expression levels of the group of genes comprising MYL9, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYH11, IGF1R, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, ILIRAP, IGFBP3 and VCAM1, are below the mean value, is indicative that the patient is a good responder to a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells..

All the embodiments of the method for predicting clinical outcome of a patient as described above are also contemplated in this aspect. Alternatively, in order to choose the more appropriate treatment for a patient, the present invention also discloses a method for selecting a patient affected with a cancer who will be a poor responder to a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a low expression level of GSR gene and a high expression level of MYL9 gene being indicative that said patient is a poor responder to a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells.

In another embodiment, the method comprises (i) determining in a cancer sample from said patient the expression level of GSR and MYL9 genes, (ii) performing hierarchical cluster analysis on standardized values of the expression levels of said genes in said sample and in a population of randomly selected cancer samples, wherein the classification of the cancer sample from the patient in the cluster in which the expression level of GSR gene is below the mean value, and in which the expression level of MYL9 gene is above the mean value, is indicative that the patient is a poor responder to a radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells.

All the embodiments as described above for the methods for predicting clinical outcome or selecting a patient, are also contemplated in this aspect. In particular, additional genes can be used in this method as disclosed above.

Cluster analysis and classification of the cancer sample from the patient based on expression levels of genes may be performed as disclosed above.

Optionally, the method further comprises a step of administering a therapeutic amount of an alternative treatment to said patient.

In a further aspect, the present invention also relates to a compound inducing the accumulation of ROS, for use in the treatment of a patient affected with a cancer and having high expression level of GSR gene and low expression level of MYL9 gene, said expression levels being determined in a cancer sample from said patient.

In a particular embodiment, the compound is an antineoplastic agent inducing the accumulation of ROS, preferably selected from the group consisting of piperlongumine and its analogs, lanperisone and its analogs, anthracyclines, taxanes, alkylating agents, epipodophyllotoxins, camptothecins and cytotoxic antibiotics. In a particular embodiment, the compound is selected from the group consisting of taxanes and alkylating agents, preferably a taxane drug, more preferably paclitaxel. In a preferred embodiment, the compound is selected from the group consisting of piperlongumine and its analogs, lanperisone and its analogs and epipodophyllotoxins, preferably from the group consisting of piperlongumine and its analogs and lanperisone and its analogs, more preferably is piperlongumine.

All the embodiments as described above for the methods for predicting clinical outcome or selecting a patient, are also contemplated in this aspect.

In a preferred embodiment, the patient is a patient in whom expression levels of GSR and MYL9 genes have been determined in a cancer sample from said patient.

Alternatively, the present invention also relates to a compound inducing the accumulation of ROS, for use in the treatment of a patient affected with a cancer and having a "stress" pattern.

In a particular embodiment, the compound is an antineoplastic agent inducing the accumulation of ROS, preferably selected from the group consisting of piperlongumine and its analogs, lanperisone and its analogs, anthracyclines, taxanes, alkylating agents, epipodophyllotoxins, camptothecins and cytotoxic antibiotics. In a particular embodiment, the compound is selected from the group consisting of taxanes and alkylating agents, preferably a taxane drug, more preferably paclitaxel In a preferred embodiment, the compound is selected from the group consisting of piperlongumine and its analogs, lanperisone and its analogs and epipodophyllotoxins, preferably from the group consisting of piperlongumine and its analogs and lanperisone and its analogs, more preferably is piperlongumine.

In a preferred embodiment, the patient is a patient in whom the stratification of the patient in "stress" pattern has been performed from a cancer sample from said patient.

The stratification of the patient in "stress" pattern may be performed using the method of the invention as disclosed above.

All the embodiments as described above for the methods for predicting clinical outcome or selecting a patient, are also contemplated in this aspect.

The present invention also relates to

- a pharmaceutical composition comprising a compound, preferably an antineoplastic agent, inducing the accumulation of ROS, and optionally a pharmaceutically acceptable carrier, for use in the treatment in the treatment of a patient affected with a cancer and having high expression level of GSR gene and low expression level of MYL9 gene, said expression levels being determined in a cancer sample from said patient;

- the use of a compound, preferably an antineoplastic agent, inducing the accumulation of ROS for the manufacture of a medicament for the treatment of a patient affected with a cancer and having high expression level of GSR gene and low expression level of MYL9 gene, said expression levels being determined in a cancer sample from said patient.

In an embodiment, the patient is a patient in whom expression levels of GSR and MYL9 genes have been previously determined in a cancer sample from said patient.

All the embodiments as described above for the methods for predicting clinical outcome or selecting a patient, are also contemplated in these aspects.

The present invention further relates to

- a method for treating a patient affected with a cancer and having high expression level of GSR gene and low expression level of MYL9 gene, said expression levels being determined in a cancer sample from said patient, comprising administering a therapeutically efficient amount of a pharmaceutical composition comprising a compound, preferably an antineoplastic agent, inducing the accumulation of ROS, and optionally a pharmaceutically acceptable carrier;

- a method for treating a patient affected with a cancer and having high expression level of GSR gene and low expression level of MYL9 gene, said expression levels being determined in a cancer sample from said patient, comprising surgical debulking and/or a dose- dense and/or dose-intensive chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells;

- a method for treating a patient affected with a cancer and having high expression level of GSR gene and low expression level of MYL9 gene, said expression levels being determined in a cancer sample from said patient, comprising surgical debulking followed by a dose-dense and/or dose-intensive chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells;

- a method for treating a patient affected with a cancer and having low expression level of GSR gene and high expression level of MYL9 gene, said expression levels being determined in a cancer sample from said patient, comprising administering neo-adjuvant chemotherapy before surgical debulking.

Optionally, these methods may further comprise the step of determining the expression levels of GSR and MYL9 genes in a cancer sample from said patient. The methods may further comprise the step of selecting a patient according the expression levels of GSR and MYL9 genes in a cancer sample from said patient. By a "therapeutically efficient amount" is intended an amount of therapeutic agent administered to a patient that is sufficient to constitute a treatment of a cancer, preferably an ovarian cancer.

All the embodiments as described above for the methods for predicting clinical outcome or selecting a patient, are also contemplated in these aspects.

In another aspect, the present invention also relates to a DNA microarray comprising a solid support which carries probes that are specific to GSR and MYL9 genes. Preferably, the DNA microarray comprises less than 150 probes with different specificities, preferably less than 100, 50, 40, 30, 20 or 10 probes with different specificities.

In an embodiment, the DNA microarray may further comprise probes that are specific of one or several genes selected from the group consisting of DNAJC10, PIK3R3, PRDX1, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKR1A1, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHX1, STIP1, IGF1, TIMP2, FN1, TNFRSFllB, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COLlAl, LY96, FGFR2, MYH9, PDGFRB, ACAA1, ACAD8, ACADM, ACADSB, AC02, AIFM1, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP5I, ATP5J2, ATP50, ATP6V0B, ATP6V0E1, ATP7A, AUH BCKDHA, CASP8, CASP9, CAT, CBR1, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, DLD, DNAJC3, DNAJC4, DNAJC8, ECHSl, FH, FTHl, GCLC, GCLM, GPD2, GSTKl, HADH, HIBADH, HMGCL, HMGCS1, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT-C02, MT- ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA7, NDUFA8, NDUFAF1, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHA1, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAK1, UQCR11, UQCRC1, UQCRQ, VCP, XDH, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYH11, IGF1R, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, IL1RAP, IGFBP3 and VCAM1, preferably selected from the group consisting of AKR1A1, TIMP2, PRKCZ, NQOl and FN1, more preferably selected from the group consisting of AKR1A1, TIMP2, NQOl and FN1, even more preferably selected from the group consisting of AKR1A1 and TIMP2.

The DNA microarray may further comprise nucleic acids for control gene, for instance a positive and negative control or a nucleic acid for an ubiquitous gene, such as HPRT1, RPLP0, GAPDH and/or B2M, in order to normalize the results. The DNA microarray consists of different nucleic acid probes that are chemically attached to a substrate, which can be a microchip, a glass slide or a microsphere-sized bead. A microchip may be constituted of polymers, plastics, resins, polysaccharides, silica or silica- based materials, carbon, metals, inorganic glasses, or nitrocellulose. Probes comprise nucleic acids such as cDNAs or oligonucleotides that may be about 10 to about 60 base pairs.

The calculations for the methods described herein can involve computer-based calculations and tools. Accordingly, the invention also relates to a computer program and to a computer readable medium comprising computer-executable instructions for performing a method according to the invention, in particular to perform step(s) of standardizing expression level values and/or performing cluster analysis and/or classifying the cancer sample from the patient.

In another aspect, the present invention further concerns a kit (a) for predicting clinical outcome of a patient affected with a cancer and/or (b) for selecting a patient affected with a cancer as a good or poor responder and/or (c) for selecting a patient affected with a cancer for surgical debulking and/or a dose-dense and/or dose-intensive chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells and/or (d) for selecting a patient affected with a cancer for surgical debulking followed by a dose-dense and/or dose- intensive chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells and/or (e) for selecting a patient affected with a cancer for neo-adjuvant chemotherapy before surgical debulking, wherein the kit comprises a pair of primers and/or a probe and/or an antibody, specific to GSR gene and MYL9 gene, and optionally a leaflet providing guidelines to use such a kit.

The kit may also comprise a pair of primers and/or a probe and/or an antibody, specific to one or several genes selected from the group consisting of DNAJC10, PIK3R3, PRDXl, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKR1A1, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHX1, STIP1, IGF1, TIMP2, FN1, TNFRSF11B, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9, PDGFRB, ACAAl, ACAD8, ACADM, ACADSB, AC02, AIFMl, AKR7A2, AKR7A3, ALDH6A1, APH1A, ATP5D, ATP5G3, ATP 51, ATP5J2, ATP 50, ATP6V0B, ATP6V0E1, ATP7A, AUH, BCKDHA, CASP8, CASP9, CAT, CBR1, CLPP, CLYBL, COX17, COX4I1, COX5A, COX5B, COX7A2, COX7B, COX8A, CPT1B, CYCS, DID, DNAJC3, DNAJC4, DNAJC8, ECHS1, FH, FTH1, GCLC, GCLM, GPD2, GSTK1, HADH, HIBADH, HMGCL, HMGCS1, HSD17B4, IDH2, IDH3A, IDH3G, IVD, KRAS, MCCC2, MGST2, MT- C02, MT-ND6, NDUFA1, NDUFA11, NDUFA2, NDUFA5, NDUFA6, NDUFA7, NDUFA8, NDUFAF1, NDUFB3, NDUFB4, NDUFS1, NDUFS7, NQ02, OGDH, PARK7, PDHA1, PIK3CB, PPA2, PPIB, SDHB, SUCLG2, TRAK1, UQCR11, UQCRC1, UQCRQ, VCP, XDH, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYH11, IGF1R, ICAM1, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, ILIRAP, IGFBP3 and VCAM1, preferably selected from the group consisting of AKR1A1, TIMP2, PRKCZ, NQOl and FN1, more preferably selected from the group consisting of AKR1A1, TIMP2, NQOl and FN1, even more preferably selected from the group consisting of AKR1A1 and TIMP2.

The kit may also comprise a pair of primers and/or a probe and/or an antibody, specific to one or several endogeneous control genes such as HPRT1, RPLPO, GAPDH and/or B2M.

The kit may further comprise (i) means for detecting the formation of complexes between proteins encoded by the genes and the antibodies specific of these genes; and/or (ii) means for detecting the hybridization of the probes with mRNA or cDNA molecules of the genes; and/or (iii) means for amplifying and/or detecting the mRNA or cDNA molecules of the genes by using the pairs of primers.

The kit can further comprise control reagents and other necessary reagents.

The kit may also comprise a computer readable medium comprising computer- executable instructions for performing the method of the invention (i) for predicting clinical outcome and/or (ii) for selecting a patient affected with a cancer for surgical debulking and/or a dose-dense and/or dose-intensive chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells and/or (iii) for selecting a patient affected with a cancer for surgical debulking followed by a dose-dense and/or dose-intensive chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells and/or (iv) for selecting a patient affected with a cancer for neo-adjuvant chemotherapy before surgical debulking.

The invention also relates to the use of a DNA microarray or a kit according to the invention in any one of the methods of the invention.

In particular, the invention relates to the use of a kit comprising detection means selected from the group consisting of a pair of primers, a probe and an antibody specific to GSR gene and MYL9 gene, and optionally, a leaflet providing guidelines to use such a kit, for predicting clinical outcome, and/or for selecting a patient affected with a cancer for surgical debulking and/or a dose-dense and/or dose-intensive chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, and/or for selecting a patient affected with a cancer for surgical debulking followed by a dose-dense and/or dose-intensive chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, and/or for selecting a patient affected with a cancer for neo-adjuvant chemotherapy before surgical debulking.

EXAMPLARY ASPECTS OF THE INVENTION

1. An in vitro method for predicting clinical outcome of a patient affected with a cancer, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a high expression level of GSR gene and a low expression level of MYL9 gene being indicative of a longer progression free survival and/or an increased patient survival.

2. An in vitro method for selecting a patient affected with a cancer for surgical debulking and/or a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, or for determining whether a patient affected with a cancer is susceptible to benefit from surgical debulking and/or a dose- intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a high expression level of GSR gene and a low expression level of MYL9 gene being indicative that said patient is susceptible to benefit from surgical debulking and/or a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells.

3. An in vitro method for selecting a patient affected with a cancer for surgical debulking followed by a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, or for determining whether a patient affected with a cancer is susceptible to benefit from surgical debulking followed by a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a high expression level of GSR gene and a low expression level of MYL9 gene being indicative that said patient is susceptible to benefit from surgical debulking followed by a dose-intensive and/or dose-dense chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells

4. An in vitro method for selecting a patient affected with a cancer for neo-adjuvant chemotherapy before surgical debulking or for determining whether a patient affected with a cancer is susceptible to benefit from neo-adjuvant chemotherapy before surgical debulking, wherein the method comprises determining the expression level of GSR and MYL9 genes in a cancer sample from said patient, a low expression level of GSR gene and a high expression level of MYL9 gene being indicative that said patient is susceptible to benefit from neoadjuvant chemotherapy before surgical debulking.

5. The method of any of aspects 1 to 4, wherein the expression level of each gene is determined by measuring the quantity of mR A.

6. The method of aspect 5, wherein the quantity of mRNA is determined by quantitative or semi-quantitative RT-PCR, by real time quantitative or semi-quantitative RT- PCR.

7. The method of aspect 5, wherein the quantity of mRNA is determined using

Nanostring technology.

8. The method of aspect 5, wherein the quantity of mRNA is determined using an approach based on high-throughput sequencing technology

9. The method of aspect 5, wherein the quantity of mRNA is determined by transcriptome approaches.

10. The method of any of aspects 1 to 4, wherein the expression level of each gene is determined by measuring the quantity or the activity of encoded protein preferably by immunochemistry, semi-quantitative Western-Blot or protein or antibody arrays.

11. The method of any of aspects 1 to 10, wherein the method further comprises comparing the expression levels of the genes to a reference expression level.

12. The method according to aspect 11, wherein the expression levels of GSR and MYL9 genes are compared with the mean expression level of GSR and MYL9 genes among a population of randomly selected cancer samples.

13. The method of any of aspects 1 to 12, further comprising determining the expression level of one or several additional genes selected from the group consisting of

DNAJC10, PIK3R3, PRDX1, DNAJA4, MAP2K6, UBE2K, DNAJC11, NQOl, PRKCZ, DNAJC16, MGST1, HSP90AA1, AKR1A1, PRKCH, AKT1, TXN, SOD1, GSK3B, NRAS, EPHX1, STIP1, IGF1, TIMP2, FN1, TNFRSF11B, MYH10, CD40, COL3A1, CCL5, ACTA2, CCR5, COL1A1, LY96, FGFR2, MYH9 and PDGFRB. 14. The method of any of aspects 1 to 12, further comprising determining the expression level of one or several additional genes selected from the group consisting of AKR1A1, TIMP2, NQOl and M.

15. The method of any of aspects 1 to 12, further comprising determining the expression level of one or several additional genes selected from the group consisting of

AKR1A1 and TIMP2.

16. The method of any of aspects 1 to 15, further comprising determining the expression level of one or several additional genes selected from the group consisting of CPT1B, PARK7, SUCLG2, MT-C02, CAT, COX5A, IDH3A, CLYBL, ATP6V0E1, NDUFB4, SDHB, NDUFA7, UQCRll, NDUFA6, AC02, FH, TRAKl, UQCRQ, NDUFAl, PPA2, ATP5J2, NDUFAl 1, CASP9, UQCRC1, NDUFS1, NDUFS7, CASP8, NDUFA8, COX5B, COX7A2, GPD2, ATP5D, OGDH, ATP50, COX8A, IDH3G, NDUFA2, APH1A, ATP5I, XDH COX7B, ATP5G3, NDUFA5, CYCS, AIFM1, ATP7A, NDUFAF1, MT-ND6, DID, IDH2, PDHA1, ATP6V0B, COX4I1, NDUFB3, COX17, MMP2, COL1A2, SMAD7, EDNRA, IGFBP4, PDGFRA, PDGFC, MYL6B, CTGF, VEGFC, IGFBP5, MYHll, IGFIR, ICAMl, TGFB1, FGFR1, TNFRSF1A, TGFBR1, TGFB2, IL1RAP, IGFBP3, VCAM1, ACAA1, ACAD8, ACADM, ACADSB, AKR7A2, AKR7A3, ALDH6A1, AUH, BCKDHA, CBR1, CLPP, DNAJC3, DNAJC4, DNAJC8, ECHS1, FTH1, GCLC, GCLM, GSTK1, HADH, HIBADH, HMGCL, HMGCS1, HSD17B4, IVD, KRAS, MCCC2, MGST2, NQ02, PIK3CB, PPIB and VCP.

17. The method of any of aspects 1 to 16, further comprising assessing at least one other prognosis markers such as histological sub-type, tumor grade, tumor stage, p53 status, BRCAl/2 status, mitotic index, tumor size or extent of residual disease after surgery, or expression of proliferation markers.

18. The method of any of aspects 2, 3 and 5 to 17, wherein the chemotherapy inducing acute oxidative stress in cancer cells comprises an antineoplastic agent inducing the accumulation of ROS, preferably selected from the group consisting of piperlongumine and its analogs, lanperisone and its analogs, anthracyclines, taxanes, alkylating agents, epipodophyllotoxins, camptothecins and cytotoxic antibiotics.

19. The method of aspect 18, wherein the chemotherapy inducing acute oxidative stress in cancer cells comprises an antineoplastic agent inducing the accumulation of ROS, preferably selected from the group consisting of lanperisone and its analogs, anthracyclines, taxanes, alkylating agents, epipodophyllotoxins, camptothecins and cytotoxic antibiotics. 20. The method of aspect 18, wherein the chemotherapy inducing acute oxidative stress in cancer cells comprises an antineoplastic agent inducing the accumulation of ROS, preferably selected from the group consisting of taxanes and alkylating agents.

21. The method of any of aspects 2, 3 and 5 to 20, wherein the chemotherapy inducing acute oxidative stress in cancer cells comprises a compound having the ability to induce or increase oxidative stress, preferably a compound selected from the group consisting of tolperisone; artemisinin; darinaparsin; motexafin gadolinium; menadione; shikonin; paracetamol; acetylsalicylic acid; geldanamycin; 3,7-diaminophenothiazinium redox dyes; disulfiram; polysulfide-based anticancer drugs; diallyldisulfide and diallyltrisulfide; isothiocyanate organosulfur agents; electrophilic Michael acceptors; superoxide dismutase inhibitors; superoxide dismutase mimetics; compounds disturbing the glutathione redox- system; compounds disturbing the thioredoxin system; compounds inhibiting NQOl function; compounds inhibiting APE-Refl function; 2-deoxyglucose; 3-bromopyruvate; dichloroacetate; redox inactive vitamine E analogues; 3,3'-diindolylmethane; Bz-423; erastin and RSL5; and their derivatives and analogues, and any combination thereof.

22. The method of any of aspects 2, 3 and 5 to 21, wherein radiotherapy is nanoparticle enhanced radiotherapy.

23. Use of a kit comprising detection means selected from the group consisting of a pair of primers, a probe and an antibody specific to GSR gene and MYL9 gene, and optionally, a leaflet providing guidelines to use such a kit, (i) for predicting clinical outcome and/or (b) for selecting a patient affected with a cancer for surgical debulking and/or a dose-intensive chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells and/or (c) for selecting a patient affected with a cancer for surgical debulking followed by a dose-intensive chemotherapy, radiotherapy and/or a chemotherapy inducing acute oxidative stress in cancer cells and/or (d) for selecting a patient affected with a cancer for neoadjuvant chemotherapy before surgical debulking.

24. An antineoplastic agent inducing the accumulation of ROS selected from the group consisting of piperlongumine and its analogs, lanperisone and its analogs, anthracyclines, taxanes, alkylating agents, epipodophyllotoxins, camptothecins and cytotoxic antibiotics, for use in the treatment of a patient affected with a cancer and having high expression level of GSR gene and low expression level ofMYL9 gene, said expression levels being determined in a cancer sample from said patient.

25. An antineoplastic agent inducing the accumulation of ROS selected from the group consisting of piperlongumine and its analogs, lanperisone and its analogs, anthracyclines, taxanes, alkylating agents, epipodophyllotoxins, camptothecins and cytotoxic antibiotics, for use in the treatment of a patient affected with a cancer and in whom expression levels of GSR and MYL9 genes have been determined in a cancer sample from said patient and having high expression level of GSR gene and low expression level of MYL9 gene.

26. The antineoplastic agent of aspect 24 or 25, wherein the antineoplastic agent is selected from the group consisting of piperlongumine and its analogs, lanperisone and its analogs and epipodophyllotoxins.

27. The antineoplastic agent of aspect 26, wherein the antineoplastic agent is selected from the group consisting of piperlongumine and its analogs and lanperisone and its analogs.

28. The method of any of aspects 1 to 22, the use according to aspect 23 and the antineoplastic agent according of any of aspects 24 to 27, wherein the cancer is selected from the group consisting of ovarian cancer, cervical cancer, vulvar cancer, vaginal cancer, prostate cancer, lung cancer, pancreas cancer, colorectal cancer and leukemia.

29. The method, the use or the antineoplastic agent according to aspect 28, wherein the cancer is ovarian cancer.

Further aspects and advantages of the present invention will be described in the following examples, which should be regarded as illustrative and not limiting.

EXAMPLES Example 1

Materials and methods

Ovarian cancer sample collection

Tumor samples were obtained from a cohort of consecutive ovarian carcinoma patients, treated at the Institut Curie between 1989 and 2005. For each patient, before chemotherapy, a surgical tumor specimen was taken for pathological analysis and tumor tissue cryopreservation. The median patient's age was 57,8 years (with a range of 31-86 years). Ovarian carcinomas were classified according to the World Health Organization histological classification of gynecological tumors. Pathological analysis showed that tumors corresponded to 82 serous, 8 mucinous, 8 endometrioid, 6 clear cell carcinomas, 2 carcinosarcomas and 1 malignant Brenner tumor. 100 cases were classified as high histological grade (grade 2 and 3) and 7 as low grade (grade 1). Clinical staging showed that 31 cases (29%) were considered as early stages (FIGO I-IIc) and 76 cases (71%) as advanced stage (III/IV). Patients were treated with a combination of surgery and chemotherapy, the latter including alkylating agents +/- taxane as first line treatment in most cases. Two chemotherapeutic regiments were used: alkylating agents alone (referred to as Ifosfamide, 5- FU, platinum) or alkylating agents in association with Taxol. Follow-up was based on 5 clinical, biological (CA12.5 by immunoradiometric assay using OC-125 antibody -Ref: ab693 from Abeam) and imaging data recorded every 6 months. The median follow-up was 98.3 months (with a range of 49-242 months). Progression-free survival (PFS) was defined as the time interval between the date of diagnosis and the first confirmed sign of disease recurrence. Overall survival (OS) was defined as the time interval between the date of diagnosis and the 10 date of death. Prior to inclusion in the study and according to French regulations, all included patients were informed of the research performed using the biological specimens obtained during their treatment and did not express opposition. Experiments were approved by the scientific committee of the Institut Curie.

Cell lines, chemical treatments and transient transfection

15 All cell lines were grown in Dulbecco's modified Eagle's medium (DMEM,

Invitrogen) containing 10% fetal bovine serum (FBS) with the exception of mouse fibroblasts and SKOV3 which were grown in DMEM with 7% FBS and RPMI with 10% FBS, respectively. Penicillin and streptomycin were added to all culture medium unless otherwise stated. Cells were treated with 400μΜ of H202 (Sigma- Aldrich), 50 nM of phorbol myristate

20 acetate (PMA), 25 μΜ of tert-Butylhydroquinone (tBHQ), 7% FBS (after overnight incubation in 0.1% FBS) for the indicated times and 20nM of paclitaxel (Hospira 6mg/ml) ± NAC (2-5mM) (N-Acetyl-L-Cysteine, #A8199, Sigma) for 48 hours. Cells were transiently transfected with ΙΟηΜ of siRNA or miRIDIAN miRNA Mimics (Dharmacon; universal negative control: CN-002000-01-05; miR-141 : C-300608-03-0005; miR-200a: C-300651-05-

25 0005; miR-200b: C-300582-07-0005; miR-200c: C-300646-05-0005) using 9 μΐ of hiperfect (Qiagen) per well in 3 ml final volume (6-wells plates). The expression rates of all transfected miRNA Mimics have been measured by qPCR and reach the same level: 13.4 ± 0.7 Ct (cycle threshold), in average. Experiments are performed 3 days post-transfection. Using similar protocol, cells were transfected with 25nM of miRIDIAN miRNA Inhibitor (Dharmacon)

30 using 4 μΐ of Dharmafect (Dharmacon). Down-regulation of miR-141/200a has been verified in these conditions and dropped to 50%> of the endogenous miR-141/200a levels, in average. RNA extraction from ovarian tumors and cell lines

RNAs from 107 frozen ovarian tumors were extracted by the Biological Resource Center of the Institut Curie with TRIzol reagent (Life Technology, Inc) and purified using a column cleanup kit (Macherey-Nagel). RNA quality was checked on an Agilent 2100 Bioanalyzer. For quality control, each tumor tissue specimen was checked on frozen histological section for evaluation of cellularity and macrodissection before nucleic acid extraction. Only tumors with a high content in epithelial tissue (at least 65%) have been included in our analysis. For cell lines, RNA was extracted using miRNeasy extraction kit (Qiagen) and Phase lock gel Heavy (5PRIME) as recommended by the manufacturer. RNA Integrity was verified using Agilent Bioanalyzer. mRNA and MicroRNAs qRT-PCR analysis in ovarian tumors and cell lines

For mRNA analysis, 0.5 to ^g of total RNA was reverse transcribed using iScript Reverse Transcription kit (Biorad) and quantitative real-time PCR (qPCR) was performed using Power SYBR green PCR Master mix (Applied Biosystems) on a Chromo4 System (Biorad). Data were analyzed using Opticon Monitor and normalized to GAPDH mRNA (cell lines) or to the average of HPRT, HPLP0, GAPDH, and B2M mRNA levels (ovarian tumors). Mature microRNA levels were quantified using TaqMan microRNA Assays (Applied Biosystems), and normalized to the average of U6 snoRNA and miR-16 (cell lines) or to the average of U6B/U6/RNU24/RNU49/RNU48 small RNA levels (ovarian tumors); Fold changes were calculated using 2 ^"AACt . Relative miR-200a and miR-141 levels were quantified from RNA of a subset of 82 ovarian tumors.

Human ovarian microarray datasets and miR-200-dependent gene expression profiling

Ovarian human tumor samples were analyzed on Human Genome U133 Plus 2.0 array (Affymetrix), according to manufacturer's procedures, by Institut Curie's translational research laboratory (Meyniel et al, 2010). Briefly, transcriptome data were normalized using GCRMA algorithm. Only probes with log-intensity value higher than 3.5 (log2) in at least 80% of samples were kept for further analysis. Microarray normalization and analysis were performed using Partek Genomic Suite software (Partek Incorporated, Saint Louis, USA). Institut Curie's ovarian cancer microarray dataset are freely accessible under the following GEO accession number: GSE26193. Using the Curie dataset from which miR-200a- expression values were determined, correlation between mature miR-200a levels and detected probesets was calculated using Pearson correlation coefficient. The probesets, identified as significantly correlated or anti-correlated, were submitted to Ingenuity Pathway Analysis software (IP A, Redwood City, USA). Enriched canonical pathways were selected using Fisher Exact test and adjusted P values (p < 0.05) using Benjamini-Hochberg multiple testing correction. For probesets correlated with miR-200a, the most significant Ingenuity canonical pathway is "Nrf2-mediated oxidative stress response"; for probesets anti-correlated with miR- 200a, the most significant Ingenuity canonical pathway is "Hepatic fibrosis - hepatic stellate Cell activation". Genes, correlated with miR-200a and found in "Nrf2-mediated oxidative stress response" Ingenuity pathway, compose the "stress" signature. Genes, anti-correlated with miR-200a and found in the "hepatic fibrosis" Ingenuity pathway, compose the "fibrosis" signature. The genes are listed in Figure 11.

Hierarchical clustering and survival analysis of human ovarian microarray datasets

Probesets of genes from the "stress" and "fibrosis" signatures were used to build a hierarchical cluster using ovarian cancer patients from Institut Curie and further confirmed using Australian Ovarian Cancer (AOCS) dataset from D.D. Bowtell's laboratory and the Australian consortium (GEO accession number GSE9891) (Tothill et al, 2008). Hierarchical clustering on standardized values (centered on the mean and rescaled to have a standard deviation of 1) was performed using standard Pearson's correlation as similarity measure, and the average method as linkage criteria. Opposite main tree branches were used to segregate two distinct tumors groups of equivalent size. The group associated with high expression of stress-related genes was qualified as "Stress UP - Fibrosis DOWN" and the other as "Stress DOWN - Fibrosis UP". Survival analyses (Kaplan-Meyer and Cox multivariate model) were conducted using Partek, R package or SPSS statistic program (IBM) and significance was calculated using the Log-rank test.

Immunohistochemistry from human ovarian carcinomas

Tissue microarray (TMA) from 56 ovarian serous adenocarcinomas was composed using two cores of tumor tissue per case (1mm of diameter each) and hybridized simultaneously. Sections of paraffin embedded tissue (3μιη) were stained using streptavidin- peroxidase protocol (immunostainer Benchmark, Ventana, Illkirch, France) with specific antibodies recognizing p38a (1/50; #9218, Cell Signaling). For quantification, two sections from distinct areas of each tumor were evaluated independently by two different investigators. An IHC score (0 to 4) was defined as follows: (staining intensity x percentage of positively labeled cells) / 100. miRNA microarray data from mouse fibroblasts

MicroRNA screening was performed on two independent experiments using the miRNA expression profiling assay (Illumina) and Illumina Mouse vl . l microRNA expression beadchip. RNAs were isolated from untreated fibroblasts or cells treated with 400 μΜ H ₂ O ₂ for 0.5, 1, 2, 4, 8 and 24 hours. Briefly, 500ng of RNA was used, processed and hybridized following manufacturer instruction by Integragen compagny (Evry, France). Probe intensities were background-subtracted and normalized by the quantile method using the Beadstudio analysis program (Illumina). For each experiment, signals from early (0.5h, lh), mid (2h, 4h) or late (8h, 24h) time points were averaged. Illumina custom error model was applied to identify differentially expressed microRNAs in each class compared to untreated (P < 0.05). Only microRNA differentially expressed in both experiments were retained for further analysis. The clustering representation corresponds to the mean of values from each experiment (expressed in log2 and normalized to untreated conditions). Hierarchical cluster was generated using R package. H ₂ 0 ₂ -induced miRNA in mouse fibroblast microarray data are freely accessible under the following GEO accession number: GSE26194.

Microarray dataset from miR-141/200a-overexpressing fibroblasts

Analyses of transfected mouse fibroblasts transcriptomes were performed by Integragen (Evry, France) on Illumina Mouse WG6 vl .l microarray following manufacturer instructions. Microarray signals were analysed with Beadstudio analysis program (Illumina). Briefly, signal was normalized using the rank invariant method and significance was established by illumina custom error model with the false discovery rate (FDR) correction. Raw and processed data has been submitted to GEO database under the following accession number GSE26113.

NCI60 microRNA and mRNA correlation analysis

Normalized Agilent microRNA and mRNA dataset 41 from NCI60 cell lines were obtained from Cellminer (http://discover.nci.nih.gov/cellminer/). Pearson correlation and associated P value were calculated between (1) miR-200a and mRNA probesets or (2) MAPK14 probeset (A 24 P397566) and all analyzed microRNAs. For rank analysis, nonsignificant correlations were discarded (P > 0.05) and then lists were ranked by correlation coefficient. Western Blotting

Cell extracts were prepared by harvesting cells in Laemmli Buffer containing 0.1M DTT, phosphatase and protease inhibitors (Roche). Western blotting was performed as described in Gerald et al, 2004. Probing was carried out with specific antibodies recognizing ZEBl (1/1000, #3396, Cell Signaling), E-Cadherin (1/1000, #610181, BD Biosciences), phospho-p38a (1/1000, #4511, Cell Signaling), p38a (1/1000, #9218, Cell Signaling), p-JNK (1/1000, #9255, Cell Signaling), INK (1/1000, #9258, Cell Signaling), GAPDH (1/40000, MAB374, Millipore), MKK4 (#9152, Cell Signaling), P-MKK4 (#9155, Cell Signaling), MKK3 (#9232, Cell Signaling), MKK6 (#1821-1, Epitomics), P-MKK3/6 (#AF4930 R&D system), MKK7 (#1949-1, Epitomics), P-MKK7 (#4171, Cell Signaling).

Xenograft Experiments

ras+Control or ras+miR-141 transformed fibroblasts were obtained by stable transfection of pEGPempty or pEGP-miR-141 expression vector (Cell biolabs), respectively. Briefly, Ras-transformed fibroblasts 2 were transfected with JetPei (Polyplus transfection) following manufacturer instructions. After 1 week, high GFP expressing cells were FACS sorted using Facs Vantage DIVA (BD Biosciences). This enrichment procedure was repeated every two weeks for another 3 times. Overexpression of miR-141 was validated by qRT-PCR. 3 independent clones of each genotype (ras + control or ras + miR-141) have been generated and studied further. For ovarian tumor model, SKOV3 ovarian cancer cells were transfected using pLV-miR-control, pLV-hsa-miR-141 and pLV-hsa-miR-200a (#mir-p000, #mir-pl l2 and #mir-pl59, Biosettia) and further selected using puromycine selection (1 ng/ul) for 3 weeks. Graft experiments were performed by subcutaneous injection of 2x10 ⁵ fibroblasts or 3x10 ⁶ exponentially growing SKOV3 cells to 5/6-weeks-old nude mice. The mice were checked daily for tumor growth and tumors were measured using calipers. Tumor volume was determined using the following equation: 0.5x[lengthx(width)2]. When tumors reached 75 mm ³ , paclitaxel treatment was triggered using a single intraperitoneal injection of paclitaxel (Hospira) at 30mg/kg. Percentage of tumor growth inhibition was calculated using the following method: 100-100x(tumor volume from untreated mice/tumor volume of paclitaxel treated mice). Immunostainings were performed as described in Laurent et al, 2008, using p38a (1/50, #9218, Cell Signaling), phospho-(SerlO) histone H3 (1/200, #9701 , Cell Signaling) and PECAM/CD31 (1/10, 7388-50, Abeam) specific antibodies. The Institut Curie ethical committee approved all experiments. UTR luciferase reporter vectors and luciferase Reporter Assays

Full-length human MAPK14 and ZEB2 3'UTR luciferase reporter plasmids were bought from Switchgear Genomics (Menlo Park, USA). Mouse construct was obtained by amplifying mmu-Mapkl4 UTR by PCR, subcloning it into pCR2.1 by TOPO cloning (invitrogen), and further inserted back into pSGG_3UTR using Xbal-Xhol restriction sites. Truncations of MAPK14 3'-UTR, deletion of either entire microRNA binding sites or their 3 '-half was conducted by site directed mutagenesis using QuickChange lightning multi site mutagenesis kit (Stratagene). 24 hours prior to transfection, 20,000 cells (293T or MDA-MD- 468) were plated in 96-wells without antibiotics. Transient transfection was performed by mixing 0.15μ1 of DharmaFECT Duo Reagent with lOOng of 3 'UTR reporter plasmid and a respective final concentration of either ΙΟηΜ of miRIDIAN microRNA Mimics or 25nM of miRIDIAN miRNA Inhibitor (Dharmacon). As an internal control, 10 ng phRL-TK vector (Promega) were co-transfected and Renilla luciferase activity was used for normalization. 24 hours after transfection, Dual Luciferase Reporter Assay (Promega) was performed and luminescence was recorded with a Fluostar Optima microplate reader (BMG Labtech).

Apoptosis assays

Apoptosis was monitored by Annexin V and DAPI staining. Annexin V staining was performed using AnnexinV-APC antibody (1/20, #561012, BD biosciences) according to manufacturer instructions. DAPI was added at a final concentration of ^g/mL. When indicated, cells were treated with 20nM of Paclitaxel for 48h and daily with 4mM NAC (N- acetylcysteine) from transfection to the end of the experiment. FACS analyses have been conducted after excluding the cellular debris (on SSC-A/FSCA plot) and the doublets (on SSC-A/SSC-W plot). Finally, we quantified the apoptotic population defined as DAPI negative and Annexin positive population on DAPI-A/APC-A plot. Analyses were performed with Flow Jo 9.1 software.

Statistical Analysis

All experiments were performed in triplicates and data shown are means ± sem (unless specified) from at least 3 independent experiments. Differences were considered to be statistically significant at values of P < 0.05 by Student's t test or Mann Whitney test. Single, double and triple asterisks indicate statistically significant differences: *P < 0.05; **P < 0.01; ***p < 0.001. All survival analyses were carried out using Kaplan-Meier method and log- rank test in R (The R development Core Team, R: A language and Environment for Statistical Computing, version 2.12.0, 2011). Univariate or multivariate Cox proportional hazards regression was conducted with SPSS 19.0 software using the enter method.

Results

Expression of miRNA-200 family members is stimulated by oxidative stress

By microarray analysis, the inventors identified a set of 36 miR As whose expression was changed upon exposure of fibroblasts to acute oxidative stress (Fig. la). 20 miRNAs were up-regulated and 16 down-regulated by H ₂ 0 ₂ treatment over time. In particular, the expression of miR-200 family members was stimulated by stress (Fig. la,b). miRNAs from each locus (Fig. 2a,b) followed the same kinetics after induction by stress (Fig. lb). Overall, expression of miR-200s was induced within one hour of treatment, reached its maximum between 2 to 3 hours and was maintained at later time points. miR-200s were up-regulated in fibroblasts and epithelial cells from mouse and human species following similar kinetics (Fig. lb). In cell lines already expressing high basal levels of miR-200s, up-regulation was not detectable (Fig. 2c,d). Up-regulation of miR-200s was specific to H ₂ 0 ₂ exposure and was not observed under other stimuli or stressors (Fig. 2e), further demonstrating the specific role of these miRNAs in response to oxidative stress. miR-141 and miR-200a inhibit p38a

To decipher miR-200 function in oxidative stress, the inventors identified transcriptomic signatures, which were deregulated by miR-141 or miR-200a overexpression upon stress (Fig. 3a). Micro-array analyses indicated that miR-141/200a modulated the p38a/JNK pathways. Indeed, expression ofMapkN (encoding p38a) was significantly down- regulated by miR-141 or miR-200a over-expression, as evaluated by microarrays and qRT- PCR (Fig. 3b). Mapkl4 expression rate remained unchanged by over-expression of miR-200c or miR-200b (Fig. 3b), suggesting that only miR-141/200a could be involved in the MAPK pathway regulation.

miR-141 and miR-200a exhibit high sequence homology (Fig. 2a), suggesting they could target the same proteins. miR-141 or miR-200a over-expression severely reduced the total p38a protein level, either under basal or stressed conditions (Fig. 3c and Fig. 4b). This prevented normal accumulation of p38a phosphorylated form and led to the subsequent decreased phosphorylation of MAPKAPK2, one of its major downstream effectors (Fig. 3c).

Furthermore, miR-141/200a over-expression was associated with the constitutive activation of the JNK pathway (Fig. 3c), as seen previously following Mapkl4 inactivation (Hui et al, 2007). In miR-141/200a expressing cells, as in Mapkl4 ^" cells, the inventors observed earlier and higher phosphorylation rate of JNK1/2 and TNK-targets, such as JUN (Fig. 3c). They confirmed that over-expression of miR-141 or miR-200a led to p38a down- regulation in various mouse or human cell lines, while miR-200c or miR-200b had no effect (Fig. 3d). These results indicate specific control of p38a by miR-141/200a under basal state or stressed conditions in human and mouse cells. Accordingly, cell lines with high endogenous miR- 141/200a levels displayed lower p38a protein levels than those characterized by low miR-141/200a expression (Fig. 3e). MAPK14 mRNA level was significantly inversely- correlated with miR-200a expression (Fig. 3f), when evaluated using NCI-60 database, a panel of 60 diverse human cancer cell lines (Liuet al, 2010). Moreover, miR-200a was the 6 ^th ranked miRNA among 422 miRNAs, with respect to an inverse correlation with MAPK14. Reciprocally, MAPK14 was classified in 3616 ^th rank among the 41,078 probesets for miR- 200a.

Finally, the inventors showed that the 3'UTR of the MAPK14 gene was directly targeted by miR-141 or miR-200a and identified their genuine binding site in human and mouse species (Fig. 5). These data demonstrate that miR-141/200a are important direct regulators of p38a. miR-141/200a promote tumorigenesis in mouse models.

The inventors tested whether the p38a downregulation by miRs could affect transformation using first miR-141-overexpressing K-ras-transformed fibroblasts.

When plated in soft agar, miR-141-overexpressing cells exhibited a markedly enhanced plating efficiency and growth rate compared to controls (Fig. 6a), indicating miR- 141 facilitates cell growth without substrate attachment. Moreover, overexpression of miR- 141 significantly increased tumor size in xenografted nude mice (Fig. 6b). When compared across similar size, miR-141-overexpressing tumors exhibited lower p38a protein levels than controls. Furthermore, these tumors had higher mitotic index and higher number of large blood vessels compared to controls. The effects of miR-141/200a on xenografted ovarian tumors were then assessed. Ovarian cells stably overexpressing miR-141 or miR-200a gave rise to bigger tumors when compared to control ones (Fig. 6c). miR-141/200a expressing tumors appeared earlier and grew faster than control ones (Fig. 6d). Moreover, histological analyses confirmed that miR-141/200a-overexpressing tumors exhibited low p38a protein levels and were associated with high mitotic index. Thus, these data suggest that miR- dependent down-regulation of p38a plays a positive role in tumorigenesis. Human ovarian adenocarcinomas accumulating mir-200a exhibit low p38a protein level and a stress-related signature

The inventors analyzed a large set of human ovarian adenocarcinomas, including mostly serous subtype and high-grade tumors. Clinical details and patient information are provided in Figures 7 and 8. They first observed that there was no correlation between MAPK14 mR A and p38a protein levels (Figures 9a and 8). Interestingly, miR-200a expression rate was significantly inversely correlated with the amount of p38a protein (Spearman correlation coefficient: R = -0.37; P = 6.10-3) (Figures 9b and 10). The same tendency was observed with miR-141, while not significant (R = -0.25; P = 0.07). Tumors with high miR-200a expression exhibited faint p38a staining ; reciprocally, tumors with low miR-200a displayed high p38a epithelial staining (Fig. 9c). Thus, in high-grade ovarian adenocarcinomas, regulation of p38a occurs, at least in part, at post-transcriptional levels in a miR-200a-dependent manner.

The inventors next identified transcriptomic signatures associated with miR-200a in these tumors. In agreement with miR-200 function in MET, genes involved in MET -related pathways were significantly negatively correlated with miR-200a (Figures 9d and 11).

The only relevant pathway significantly positively correlated with miR-200a was involved in oxidative stress response (Figures 9d and 11), indicating that the link between oxidative stress and miR-200a that the inventors uncovered in vitro is relevant in a human tumor context. miR-200a-dependent signatures (stress- and fibrosis-related) predict survival of patients with ovarian cancer

"Stress" and "fibrosis" signatures were defined by the genes positively- and negatively-correlated with miR-200a, respectively. The use of this dual signature enabled the inventors to build a hierarchical cluster of ovarian tumors and to stratify patients into two groups (Fig. 9e). Patients with the "stress" pattern exhibited longer progression-free (PFS) and overall survival (OS) than those who displayed the "fibrosis" pattern (Fig. 9f). In contrast, individual signature did not show differential patient survival (not shown), suggesting that both "stress" and "fibrosis" processes are required for fully recapitulate miR-200-dependent survival. The OS was significantly different according to the level of p38a- targeting miR- 200s (miR-141/200a), the PFS tending to the same observation (Fig. 9f). Importantly, the "stress" and "fibrosis" signatures were validated in another set of patients with ovarian cancer (AOCS) (Tothill et al, 2008) (Fig. 9g), with better prognosis for patients with the "stress" pattern (Fig. 9h).

While miR141/200a-overexpression increased tumor growth in mice (Fig. 6c,d), patients with the "stress" pattern (high miR-200a) exhibit better prognosis than the "fibrosis" 5 ones (low miR-200a) (Fig. 9e-h). This apparent paradox could be explained by different aggressiveness or response to treatment. No correlation was observed between signatures and tumor grade (Figure 12), most probably because there is a bias towards high-grade tumors in Curie and AOCS cohorts (Figure 7). "Fibrosis" patients were associated with partial debulking and high stage (Figure 12). Moreover, when considering tumor residual volume

10 after the first cure of treatment (as evaluated by plasma CA12.5), it was observed that the CA12.5 levels were significantly lower in "stress" patients than in "fibrosis" ones (Figure 12). Similarly, the clinical response, defined by tumor mass variation upon treatment, was considered as complete in a high proportion of "stress" patients (Figure 12). Finally, multivariate analysis including the miR-200a-dependent signatures ("stress"-"fibrosis"), age,

15 grade, histology and chemotherapy demonstrated that the signatures maintained an independent association with PFS (Figure 13). These observations suggest that "stress" and "fibrosis" signatures defined by miR-200a may act on response to treatment.

The inventors next investigated the effect of paclitaxel, a ROS-producing chemotherapeutic agent, on miR overexpressing ovarian cancer cells. Overexpression of miR-

20 141/200a, as p38a inactivation, enhanced cell apoptosis under paclitaxel, while it protected cells against death under untreated conditions (Fig. 14a). These effects were prevented by the use of antioxidant or miR-141/200a-specific inhibitors, further confirming the role of ROS and miR-141/200a in the sensitivity of cancer cells to paclitaxel (Fig. 14a). Importantly, similar effects were observed in vivo. Indeed, while miR-141/200aoverexpression increased

25 significantly tumor growth under untreated conditions (Fig. 14b and Fig. 6c,d), animals bearing miR-141/200a-overexpressing ovarian tumors were significantly more sensitive to paclitaxel than controls (Fig. 14b-d). Thus, miR-141/200a-expression promotes tumor growth but also increases sensitivity to chemotherapy.

Example 2

30 Differential expression of genes in patients with the "stress" pattern or the

"fibrosis " pattern

Samples from patients with the "stress" pattern or the "fibrosis" pattern as defined above were used to identify probesets that are significantly differentially expressed between the "stress" and the "fibrosis" subset (p-value < 0.05). This was calculated using ANOVA test.

Genes upregulated in patients with the "stress" pattern are listed in Table 1 below. Genes upregulated in patients with the "fibrosis" pattern are listed in Table 2 below. Hierarchical clusterings were built using the genes listed in Tables 1 and 2 with cohorts from Institut Curie and AOCS as defined above and the cohort from Duke Cancer Institute (GEO accession number GSE3149 (Bild et al, 2006). Hierarchical clusterings were performed on standardized values (centered on the mean and rescaled to have a standard deviation of 1) and using standard Pearson's correlation as similarity measure and Ward method.

Figures 15 and 16 show Kaplan-Meier of progression- free survival (PFS) and overall survival (OS) curves, according to the signature comprising genes listed in Tables 1 and 2 in the cohorts from Institut Curie (Fig. 15) and AOCS (Fig. 16). Figure 22 shows Kaplan-Meier of overall survival (OS) curves, according to the signature comprising genes listed in Tables 1 and 2 in the cohort from Duke Cancer Institute. Log-rank test was used for Kaplan-Meier.

Example 3

Signature GSR/MYL9

The inventors have further demonstrated that a hierarchical clustering built using the expression levels of only two genes, namely GSR and MYL9 genes, allowed to classify the tumors of the cohorts from Institut Curie, AOCS and Duke Cancer Institute.

The "stress" pattern was defined with both high expression of GSR and low expression of MYL9. The "fibrosis" pattern was defined with both low expression of GSR and high expression of MYL9.

Hierarchical clustering was performed using the two GSR and MYL9 genes on the

Curie cohort, the AOCS cohort and the Duke Cancer Institute cohort, on standardized values (centered on the mean and rescaled to have a standard deviation of 1) and using standard Pearson's correlation as similarity measure and Ward method as linkage criteria (data not shown).

Figures 17 shows Kaplan-Meier of progression- free survival (PFS) and overall survival (OS) curves according to the GSR and MYL9 signature in the cohorts from Institut Curie and AOCS and Kaplan-Meier of overall survival (OS) curves according to the GSR and MYL9 signature in the cohort from Duke Cancer Institute. Log-rank test was used for Kaplan- Meier.

Hierarchical clustering was performed using the genes listed in Figure 11 except GSR and MYL9 genes on the Curie cohort, the AOCS cohort and the Duke Cancer Institute cohort, on standardized values (centered on the mean and rescaled to have a standard deviation of 1) and using standard Pearson's correlation as similarity measure and Ward method as linkage criteria (data not shown).

Figures 18 shows Kaplan-Meier of progression- free survival (PFS) and overall survival (OS) curves, according to the signature defined by the genes listed in Figure 11 except GSR and MYL9 genes in the cohorts from Institut Curie and AOCS and Kaplan-Meier of overall survival (OS) curves, according to the signature defined by the genes listed in Figure 11 except GSR and MYL9 genes in the cohort from Duke Cancer Institute. Log-rank test was used for Kaplan-Meier.

Conclusion

The inventors have herein clearly demonstrated that miR-141 and/or miR-200a expression levels cannot be used alone to predict overall survival (OS) and progression-free survival (PFS). When patients were stratified according to miR-141 levels, p-values computed using logrank-test were 0.396 and 0.868 for OS and PFS, respectively (Figure 19). Using miR-200a levels, the p-values were 0.444 for OS and 0.649 for PFS (Figure 20). Thus, using miR-141 or miR-200a expression levels, results were not statistically significant. In order to avoid potential variations due to inaccuracies in miRNA measurements, the inventors have also considered the average of miR-141 and miR-200a expression values to define prognosis. The p-values computed using logrank-test were 0.053 and 0.293 for OS and PFS, respectively (Figure 21). These results were not statistically significant (i.e. above 0.05). These results thus confirm the controversial results previously published and confirm that expression rates of miR-141 or miR-200a could not be used as satisfying predictive markers.

On the contrary, the use of expression levels of GSR and MYL9 genes according to the invention allows to stratify patients with highly statistically significant p-values. For the Institut Curie cohort, the p-values were 0.0048 for OS and 7e-04 for PFS. For the AOCS cohort, the p-values were 4e-04 for OS and 4e-07 for PFS. For the Duke Cancer Institute cohort, the p-value was 0.016 for OS. (Figure 17). This dual signature was confirmed by using additional sets of genes, as described above (Figures 9f, 15 and 16). Using these additional sets of genes without GSR and MYL9 results in non statistically significant p- values. In this case, the p-values were 0.079 for OS and 0.1441 for PFS, for the Institut Curie cohort, and were 0.4479 for OS 0.0887 for PFS, for the AOCS cohort. For the Duke Cancer Institute cohort, the p-value was 0.2 for OS. (Figure 18).

These results thus clearly show that expression levels of GSR and MYL9 genes can be used as highly significant predictive markers.

Example 4

Materials and methods Quantitative RT-PCR technology

1 ug of total RNA was reverse transcribed using an iScript Reverse Transcription Kit (Bio-Rad), and qRT-PCR was performed using Power SYBR Green PCR Master Mix (Applied Biosystems) on a Chromo4 System (Bio-Rad). Data were analyzed using an Opticon Monitor (Bio-Rad) and normalized to Cyclophilin B mRNA. Fold changes were calculated using the 2-DCt method. Primers: GSR-Forward: 5 '-GCACTTGCGTGAATGTTGGAT-3 ' (SEQ ID NO.: 27); GSR-Reverse: 5 '-GGCTTGGGATCACTCGTGAA-3 ' (SEQ ID NO.: 28); MYL9-Forward: 5'-AGTTCCACGCACCCAGCGA-3' (SEQ ID NO.: 29); MYL9-Reverse: 5 '-TTGCTGGACATCTTGGCTTCTGGT-3 ' (SEQ ID NO.: 30); MYHIO-Forward: 5'- ACTGAGGCGCTGGATCTGTGGT-3 ' (SEQ ID NO.: 31); MYHIO-Reverse: 5'- TCCTCGAGTCC AGTTCTCTGCG-3 ' (SEQ ID NO.: 32); AKR1A1 -Forward: 5'- ATTC ACGCTCTGTGCTTGTG-3 ' (SEQ ID NO.: 33); AKR1A1 -Reverse: 5'- TCCAGGTACCCAGACCAATC-3 ' (SEQ ID NO.: 34); FNl-Forward: 5'- AAACTTGCATCTGGAGGCAAACCC-3 ' (SEQ ID NO.: 35); FNl-Reverse: 5'- AGCTCTGATCAGCATGGACCACTT-3 ' (SEQ ID NO.: 36); TIMP2-Forward: 5'- GTGGGGTCTCGCTGGACGTTG-3 ' (SEQ ID NO.: 37); TIMP2-Reverse: 5'- TGGGTGGTGCTCAGGGTGTCC-3 ' (SEQ ID NO.: 38); NQOl-Forward: 5'- TGAAGGACCCTGCGAACTTTC-3 ' (SEQ ID NO.: 39); NQOl -Reverse: 5'- GAACACTCGCTCAAACCAGC-3 ' (SEQ ID NO.: 40); PRKCZ-Forward: 5'- GGACAGCCGGCCTTCCGTTA-3 ' (SEQ ID NO.: 41); PRKCZ-Reverse: 5'- CCCCGGCGGTAGATAGATTCGGC-3 ' (SEQ ID NO.: 42); CycloB hsa Forward: 5'- AGGCCGGGTGATCTTTGGTCT-3 ' (SEQ ID NO.: 43); CycloB hsa Reverse: 5'- CCCTGGTGAAGTCTCCGCCCT-3 ' (SEQ ID NO.: 44). ROC curve

A binary classifier for the prediction of 'stress / Fibrosis' status was defined using a combination of 4 genes (GSR, MYL9, AKRIAI and TIMP2 genes) and 6 genes (GSR, MYL9, AKRIAI, TIMP2, NQOl and FNl genes).

For each patient included in the study (94 patients from the Institut Curie cohort as described above), a score was calculated as the difference of values for genes 'stress' and values for genes 'Fibrosis' (score = 'Stress' genes - 'Fibrosis' genes).

The performance of the two binary classifiers obtained was then evaluated by Receiver Operating Characteristic (ROC) analysis. The ROC curves were performed by plotting the true positive rate (sensitivity) versus the false positive rate (1 - specificity) at various threshold settings of the functions and the accuracy of the prediction model was assessed using the Area Under the ROC Curve (AUC). AUC is comprised between 0 and 1. An AUC of 1 indicates a perfect prediction.

Results

The inventors measured the expression level of (i) GSR, MYL9, AKRIAI and TIMP2 genes or (ii) GSR, MYL9, AKRIAI, TIMP2, NQOl and FNl genes by qRT-PCR in tumours of the Curie cohort (to date 94 patients out of 107).

They showed that the expression levels of these 4 or 6 genes allowed the identification of patients in the "stress" or "fibrosis" categories. They demonstrated that this stratification method based on expression of these 4 or 6 genes by qRT-PCR exhibits a low percentage of error (less than 12 over 94 patients are misclassified). The area under the ROC curve, graph built by spotting sensitivity versus (1 -specificity) values, reaches 0.84 for the combination of GSR, MYL9, AKRIAI and TIMP2 genes and 0.89 for the combination of GSR, MYL9, AKRIAI, TIMP2, NQOl and FNl genes.

Table 1 : List of genes upregulated in patients with the "stress" pattern.

For each gene are indicated the gene symbol, gene name, probe set number, fold- change ("Fibrosis" subset vs. "Stress" subset) and the corresponding p-value.

Gene Symbol Gene Name p-value Fold Change

ACAA1 acetyl-CoA acyltransferase 1 l,71E-02 -1,271

ACAD8 acyl-CoA dehydrogenase family, member 8 l,13E-02 -1,245

ACADM acyl-CoA dehydrogenase, C-4 to C-12 straight chain 2,01E-02 -1,215

ACADSB acyl-CoA dehydrogenase, short/branched chain 2,33E-02 -1,263 AC02 aconitase 2, mitochondrial 3,15E-03 -1,34

AIFM1 apoptosis-inducing factor, mitochondrion-associated, 1 3,51E-02 -1,309

AKR1A1 aldo-keto reductase family 1, member Al (aldehyde 3,50E-04 -1,351 reductase)

AKR7A2 aldo-keto reductase family 7, member A2 (aflatoxin 2,41E-02 -1,228 aldehyde reductase)

AKR7A3 aldo-keto reductase family 7, member A3 (aflatoxin l,99E-03 -1,566 aldehyde reductase)

AKT1 v-akt murine thymoma viral oncogene homolog 1 2,l lE-03 -1,367

ALDH6A1 aldehyde dehydrogenase 6 family, member Al 2,33E-02 -1,316

APH1A anterior pharynx defective 1 homolog A (C. elegans) 2J3E-02 -1,175

ATP5D ATP synthase, H+ transporting, mitochondrial Fl 2,07E-02 -1,251 complex, delta subunit

ATP5G3 ATP synthase, H+ transporting, mitochondrial Fo 3,24E-02 -1,247 complex, subunit C3 (subunit 9)

ATP5I ATP synthase, H+ transporting, mitochondrial Fo 2J9E-02 -1,198 complex, subunit E

ATP5J2 ATP synthase, H+ transporting, mitochondrial Fo 7,46E-03 -1,241 complex, subunit F2

ATP50 ATP synthase, H+ transporting, mitochondrial Fl 2,39E-02 -1,112 complex, O subunit

ATP6V0B ATPase, H+ transporting, lysosomal 21kDa, V0 subunit b 4,03E-02 -1,218

ATP6V0E1 ATPase, H+ transporting, lysosomal 9kDa, V0 subunit el l,26E-03 -1,326

ATP7A ATPase, Cu++ transporting, alpha polypeptide 3,57E-02 -1,283

AUH AU RNA binding protein/enoyl-CoA hydratase 2,03E-02 -1,242

BCKDHA branched chain keto acid dehydrogenase El, alpha 4,65E-02 -1,207 polypeptide

CASP8 caspase 8, apoptosis-related cysteine peptidase l,42E-02 -1,33

CASP9 caspase 9, apoptosis-related cysteine peptidase 9,07E-03 -1,283

CAT catalase 4,22E-04 -1,464

CBR1 carbonyl reductase 1 l,96E-02 -1,361

CLPP ClpP caseinolytic peptidase, ATP-dependent, proteolytic l,55E-02 -1,24 subunit homolog (E. coli)

CLYBL citrate lyase beta like 6,40E-04 -1,485

COX17 COX 17 cytochrome c oxidase assembly homolog (S. 4,59E-02 -1,199 cerevisiae)

COX4I1 cytochrome c oxidase subunit IV isoform 1 4,24E-02 -1,137

COX5A cytochrome c oxidase subunit Va 4,41E-04 -1,318

COX5B cytochrome c oxidase subunit Vb l,67E-02 -1,214

COX7A2 cytochrome c oxidase subunit Vila polypeptide 2 (liver) l,68E-02 -1,172 COX7B cytochrome c oxidase subunit Vllb 3,l lE-02 -1,188

COX8A cytochrome c oxidase subunit VIIIA (ubiquitous) 2,44E-02 -1,16

CPT1B carnitine palmitoyltransferase IB (muscle) 2,66E-06 -2,067

CYCS cytochrome c, somatic 3,32E-02 -1,219

DLD dihydrolipoamide dehydrogenase 3,79E-02 -1,178

DNAJA4 DnaJ (Hsp40) homolog, subfamily A, member 4 6,67E-06 -2,171

DNAJC11 DnaJ (Hsp40) homolog, subfamily C, member 11 l,17E-02 -1,285

DNAJC16 DnaJ (Hsp40) homolog, subfamily C, member 16 l,16E-05 -1,474

DNAJC3 DnaJ (Hsp40) homolog, subfamily C, member 3 2,35E-02 -1,232

DNAJC4 DnaJ (Hsp40) homolog, subfamily C, member 4 8,39E-03 -1,264

DNAJC8 DnaJ (Hsp40) homolog, subfamily C, member 8 l,36E-02 -1,146

ECHS1 enoyl CoA hydratase, short chain, 1 , mitochondrial 2,65E-02 -1,201

FH fumarate hydratase 3,21E-03 -1,112

FTH1 ferritin, heavy polypeptide 1 l,25E-03 -1,571

GCLC glutamate-cysteine ligase, catalytic subunit 4,70E-02 -1,216

GCLM glutamate-cysteine ligase, modifier subunit l,33E-02 -1,354

GPD2 glycerol-3 -phosphate dehydrogenase 2 (mitochondrial) l,98E-02 -1,355

GSR glutathione reductase 2,20E-04 -1,642

GSTK1 glutathione S-transferase kappa 1 3,64E-02 -1,254

HADH hydroxyacyl-CoA dehydrogenase 2,25E-02 -1,276

HIBADH 3-hydroxyisobutyrate dehydrogenase 4,70E-03 -1,452

HMGCL 3 -hydro xymethyl-3 -methylglutaryl-CoA lyase 9,68E-05 -1,356

HMGCS1 3-hydroxy-3-methylglutaryl-CoA synthase 1 (soluble) 2,83E-02 -1,24

HSD17B4 hydro xysteroid (17-beta) dehydrogenase 4 2,42E-02 -1,232

IDH2 isocitrate dehydrogenase 2 (NADP+), mitochondrial 3,86E-02 -1,299

IDH3A isocitrate dehydrogenase 3 (NAD+) alpha 4,78E-04 -1,557

IDH3G isocitrate dehydrogenase 3 (NAD+) gamma 2,56E-02 -1,186

IVD isovaleryl-CoA dehydrogenase l,98E-02 -1,245

KRAS v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog 2,52E-02 -1,158

MCCC2 methylcrotonoyl-CoA carboxylase 2 (beta) l,22E-03 -1,471

MGST2 microsomal glutathione S-transferase 2 5,89E-03 -1,401

MT-C02 cytochrome c oxidase subunit II 3,17E-04 -1,102

MT-ND6 NADH dehydrogenase, subunit 6 (complex I) 3,75E-02 -1,223

NDUFA1 NADH dehydrogenase (ubiquinone) 1 alpha subcomplex, 6,79E-03 -1,218 l, 7.5kDa

NDUFA11 NADH dehydrogenase (ubiquinone) 1 alpha subcomplex, 8,29E-03 -1,256

11, 14.7kDa

NDUFA2 NADH dehydrogenase (ubiquinone) 1 alpha subcomplex, 2,58E-02 -1,197

2, 8kDa NDUFA5 NADH dehydrogenase (ubiquinone) 1 alpha subcomplex, 3,25E-02 -1,195 5, 13kDa

NDUFA6 NADH dehydrogenase (ubiquinone) 1 alpha subcomplex, 3,01E-03 -1,326

6, 14kDa

NDUFA7 NADH dehydrogenase (ubiquinone) 1 alpha subcomplex, 2,38E-03 -1,345

7, 14.5kDa

NDUFA8 NADH dehydrogenase (ubiquinone) 1 alpha subcomplex, l,45E-02 -1,244

8, 19kDa

NDUFAF1 NADH dehydrogenase (ubiquinone) 1 alpha subcomplex, 3J0E-02 -1,197 assembly factor 1

NDUFB3 NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 3, 4,38E-02 -1,121

12kDa

NDUFB4 NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 4, l,32E-03 -1,236

15kDa

NDUFS1 NADH dehydrogenase (ubiquinone) Fe-S protein 1, Ι,ΙΟΕ-02 -1,311

75kDa (NADH-coenzyme Q reductase)

NDUFS7 NADH dehydrogenase (ubiquinone) Fe-S protein 7, l,32E-02 -1,35

20kDa (NADH-coenzyme Q reductase)

NQOl NAD(P)H dehydrogenase, quinone 1 l,33E-05 -3,063

NQ02 NAD(P)H dehydrogenase, quinone 2 7,60E-04 -1,597

OGDH oxoglutarate (alpha-ketoglutarate) dehydrogenase 2,09E-02 -1,219

(lipoamide)

PARK7 Parkinson disease (autosomal recessive, early onset) 7 2,64E-05 -1,271

PDHA1 pyruvate dehydrogenase (lipoamide) alpha 1 3,91E-02 -1,222

PIK3CB phosphoinositide-3 -kinase, catalytic, beta polypeptide 2,82E-02 -1,251

PIK3R3 phosphoinositide-3 -kinase, regulatory subunit 3 (gamma) l,35E-03 -1,853

PPA2 pyrophosphatase (inorganic) 2 7,42E-03 -1,239

PPIB peptidylprolyl isomerase B (cyclophilin B) 3,57E-02 -1,194

PRDX1 peroxiredoxin 1 l,59E-05 -1,448

PRKCZ protein kinase C, zeta 1J1E-07 -1,768

SDHB succinate dehydrogenase complex, subunit B, iron sulfur 2,00E-03 -1,293

(Ip)

SOD1 superoxide dismutase 1, soluble l,69E-03 -1,266

SUCLG2 succinate-CoA ligase, GDP-forming, beta subunit l,37E-04 -1,471

TRAK1 trafficking protein, kinesin binding 1 4J4E-03 -1,411

TXN (includes thioredoxin l,19E-04 -1,405 EG: 116484)

UBE2K ubiquitin-conjugating enzyme E2K 3,39E-02 -1,159

UQCR11 ubiquinol-cytochrome c reductase, complex III subunit XI 2J4E-03 -1,282

UQCRC1 ubiquinol-cytochrome c reductase core protein I l,07E-02 -1,227 UQCRQ ubiquinol-cytochrome c reductase, complex III subunit 5,52E-03 -1,239

VII, 9.5kDa

VCP valosin containing protein 2,25E-02 -1,205

XDH xanthine dehydrogenase 2,97E-02 -1,579

Table 2 : List of genes upregulated in patients with the "fibrosis" pattern.

For each gene are indicated the gene symbol, gene name, probe set number, fold- change ("Fibrosis" subset vs. "Stress" subset) and the corresponding p-values.

Gene symbol Gene name p-value Fold Change

MYL9 myosin, light chain 9, regulatory 2,09E-14 3,901

TIMP2 TIMP metallopeptidase inhibitor 2 6,41E-11 2,566

MMP2 matrix metallopeptidase 2 (gelatinase A, 72kDa gelatinase, 8J0E-09 3,364

72kDa type IV collagenase)

COL3A1 collagen, type III, alpha 1 l,38E-08 2,763

COL1A1 collagen, type I, alpha 1 2,96E-08 2,903

FN1 fibronectin 1 5,80E-08 2,954

IGF1 insulin-like growth factor 1 (somatomedin C) l,16E-07 3,709

ACTA2 actin, alpha 2, smooth muscle, aorta l,22E-07 2,774

MYH10 myosin, heavy chain 10, non -muscle 2,13E-07 2,365

COL1A2 collagen, type I, alpha 2 5,13E-07 2,935

SMAD7 SMAD family member 7 l,09E-06 1,938

PDGFRB platelet-derived growth factor receptor, beta polypeptide l,99E-06 2,064

EDNRA endothelin receptor type A 3,90E-06 2,558

IGFBP4 insulin-like growth factor binding protein 4 2,56E-05 2,382

PDGFRA platelet-derived growth factor receptor, alpha polypeptide 3J3E-05 3,06

PDGFC platelet derived growth factor C 2,09E-03 1,642

MYL6B myosin, light chain 6B, alkali, smooth muscle and non- 2,10E-03 1,375

muscle

CTGF connective tissue growth factor 2,15E-03 1,794

LY96 lymphocyte antigen 96 2,49E-03 1,673

VEGFC vascular endothelial growth factor C 3,65E-03 1,577

IGFBP5 insulin-like growth factor binding protein 5 4,05E-03 2,177

MYH11 myosin, heavy chain 11 , smooth muscle 4,10E-03 1,952

FGFR2 fibroblast growth factor receptor 2 6,65E-03 1,703

MYH9 myosin, heavy chain 9, non-muscle 7,01E-03 1,248

IGF1R insulin-like growth factor 1 receptor 7,51E-03 1,712

ICAM1 intercellular adhesion molecule 1 7J8E-03 1,596 TGFB1 transforming growth factor, beta 1 8,75E-03 1,663

FGFR1 fibroblast growth factor receptor 1 1J3E-02 1,639

TNFRSF1A tumor necrosis factor receptor superfamily, member 1 A 2,21E-02 1,235

TGFBR1 transforming growth factor, beta receptor 1 2,59E-02 1,222

TGFB2 transforming growth factor, beta 2 2J1E-02 1,543

IL1RAP interleukin 1 receptor accessory protein 3,00E-02 1,387

CD40 CD40 molecule, TNF receptor superfamily member 5 3,l lE-02 1,278

IGFBP3 insulin-like growth factor binding protein 3 3,65E-02 1,653

VCAM1 vascular cell adhesion molecule 1 4,27E-02 1,66

REFERENCES

Alexandre et al. Cancer Res 67: 3512-3517 (2007)

Aykin-Burns et al. Biochem J 418 : 29-37 (2009)

Bendoraite, A., et al. Gynecol Oncol 116, 117-125 (2010).

Berndtsson et al. Int J Cancer 120:175-180 (2007)

Bild et al. Nature. 2006 Jan 19;439(7074):353-7

Bonnet et al. Cancer Cell 11 : 37-51 (2007)

Chatterjee and Dutta, Tetrahedron 23, 1769-1781 (1967)

Chatterji et al. Bioorg Med Chem Lett 15: 3921-3924 (2005)

Chen et al. Proc Natl Acad Sci USA 105: 11105-11109 (2008)

Efferth Drug Resist Updat 8: 85-97 (2005)

Efferth. Curr Drug Targets 7: 407-421 (2006)

Fruehauf and Trapp. Expert Rev Anticancer Ther 8: 1751-1757, 2008.

Fukuyo et al. Cancer Res 68: 6324- 6330, 2008

Geiss et al. 2008 Nat. Biotechnol. 26:317-325

Geng et al. Nanotechnology. 2011 Jul 15;22(28):285101.

Gerald, D., et al. Cell 118, 781-794 (2004).

Greer. J Am Chem Soc 123: 10379-10386, 2001.

Hu, X., et al. Gynecol Oncol 114, 457-464 (2009).

Hui, L., et al. Nat Genet 39, 741-749 (2007).

Iorio, M.V., et al. Cancer Res 67, 8699-8707 (2007).

Jacob. Nat Prod Rep 23: 851-863, 2006.

Johnson et al. Chem Biol 12:485-496, 2005.

Kim et al. Cancer Res 66: 1740-1750, 2006. Kim et al. J Bioenerg Biomembr 40: 607-613, 2008.

Kirshner et al. Mol Cancer Ther 7: 2319-2327, 2008.

Laurent, G., et al. Cell Metab 7, 113-124 (2008).

Liu, H., et al. Mol Cancer Ther 9, 1080-1091 (2010).

Magda and Miller. Semin Cancer Biol 16: 466-476, 2006.

Mao et al. Cell Res 18: 879-888, 2008.

Marty L, et al. Proc Natl Acad Sci U S A. 2009 Jun 2;106(22):9109-14.

Medjkane et al, Nature Cell Biology 11, 257 - 268 (2009)

Meyniel, J.P., et al. BMC Cancer 10, 222 (2010).

Nam, E.J., et al. Clin Cancer Res 14, 2690-2695 (2008).

Neuzil et al. Mol Pharmacol 71 : 1185-1199, 2007.

Nicolaou et al. Proc Natl Acad Sci USA 90: 5881-5888, 1993.

Oh et al. Mol Cancer Ther 6: 2178-2187, 2007.

Powolny and Singh. Cancer Lett 269: 305-314, 2008.

Quintas-Cardama et al. Anticancer Agents Med Chem 8: 904-909, 2008.

Raj et al. Nature 475, 231-234 (2011)

Roy et al. Mol Pharmacol 74: 1292-1307, 2008.

Simamura et al. J Bioenerg Biomembr 40: 213-217, 2008.

Singh et al. Anticancer Res 24: 2277-2280, 2004.

Sivaramakrishnan et al. Bioorg Med Chem Lett 18: 3076-3080,2008.

Shaw et al. Proc Natl Acad Sci U S A. 108 (21):8773-8 (2011)

Smith IK, et al. Anal Biochem. 1988 Dec;175(2):408-13

Stasi, Expert Opin Biol Ther 8: 527-540, 2008.

Tothill, R.W., et al. Clin Cancer Res 14, 5198-5208 (2008).

Vad et al. J Pharm Sci 98: 1409-1429, 2009.

Vad et al. Melanoma Res 18: 386-399, 2008.

Verraxet al. Biochem Pharmacol 72: 671-680, 2006

Wang et al. Nat Rev Genet. 2009 January; 10(1): 57-63.

Wolchok et al. Melanoma Res 13: 189-196, 2003.

Wondrak. Free Radic Biol Med 43: 178-190, 2007.

Xiao et al. J Biol Chem 283: 30151-30163, 2008.