WO2011140662A1 - Cux1 signature for determination of cancer clinical outcome - Google Patents

Abstract

The present application provides a method for determining a clinical outcome in a subject having received an initial diagnosis of cancer based on the wherein the CUX-1 signature gene set comprising at least one gene associated with genetic instability and at least one gene associated with cell proliferation. The method comprises measuring an individualized gene expression level in a plurality of genes in a CUX-1 signature gene set in a sample of the subject, normalizing the individualized gene expression level against at least one control gene expression level, comparing at least one parameter of the normalized individualized gene expression level with data derived from a set of CUX-1 expression profiles having known clinical outcomes divided at least two prognostic classes in accordance with the known clinical outcomes and associating one of the known clinical outcomes with the normalized individualized gene expression level in accordance with the comparison.

Description

CUX1 SIGNATURE FOR DETERMINATION OF CANCER CLINICAL OUTCOME

CROSS-REFERENCE TO RELATED APPLICATIONS

This is the first application filed for the present invention. This application claims priority on U.S. application 61/334,283 filed May 13, 2010, the entire content of which is hereby incorporated by reference.

This application contains a sequence listing submitted herewith electronically. The content of this electronic submission is incorporated by reference in this application.

TECHNICAL FIELD

The disclosure relates to diagnosing cancer and particularly to methods, compositions and kits for classifying subjects with breast cancer according to clinical outcome.

BACKGROUND

Aneuploidy is regarded as one of aberrations in cancer. One route to aneuploidy is believed to involve the generation of tetraploid cells that readily become aneuploid due to frequent chromosome mis-segregation and rearrangements. The presence of multiple centrosomes can lead to multipolar mitosis from which non viable daughter cells are generated, but centrosomes clustering to two poles enable cancer cells to undergo bipolar mitosis. Supernumerary centrosomes, however, increase the possibility of chromosome mis-segregation which results mostly from merotelic kinetochore- microtubule attachments whereby a single kinetochore attaches to microtubules emanating from different poles. This is sometimes manifested at anaphase by the presence of lagging chromosomes and chromosome bridges in a fraction of cells in the population.

Elevated Cut homeobox 1 (CUX1) expression was reported in various human tumors and was associated with poor prognosis (De Vos et al., 2002; Moon et al., 2002; Ripka et al. , 2010; Goulet et al., 2002; Michl et al., 2005; Ripka et al., 2007). The p75 and p1 10 isoforms of CUX1 were shown to cause mammary tumors in the FVB stain of mouse (Cadieux 2009), whereas in mixed genetic backgrounds, many virgin p75 CUX1 mice succumbed to myeloproliferative-disease (MPD)-like myeloid leukemias (Cadieux 2006) . CUX1 plays a role in at least three distinct processes with relevance to cancer: it stimulates cell cycle progression and cell motility (Kedinger and Nepveu, 2010; Michl et al., 2005; Sansregret et al., 2006), and it promotes genetic instability (Sansregret 201 1 ). Up-regulation of stable DNA binding at the G1/S transition involves dephosphorylation by the Cdc25A phosphatase (Coqueret, 1998) and proteolytic cleavage by cathepsin L to generate p1 10 CUX1 (Moon, 2001 ; Goulet 2004), whereas phosphorylation by cyclin A/Cdk1 and cyclin B/Cdk1 inhibits CUX1 activity in the G2 and mitosis phases, respectively (Santaguida et al., 2001 ; Sansregret et al., 2010). Identification of p1 10 CUX1 targets by genome-wide location analysis (ChlP-chip) revealed a striking enrichment for genes that play a role during two phases of the cell cycle: S phase and mitosis (Harada 2008; Sansregret 201 1 ). Indeed, constitutive expression of p1 10 CUX1 was shown to accelerate entry into S phase (Sansregret et al., 2006; Truscott et al., 2007).

SUMMARY

The present applications provides the use of a CUX1 signature gene set for the determination of a prognosis of a patient having already been diagnosed with cancer. Such methods, related systems and products are particularly can be applied to any types of cancer where CUX1 's activity has been implicated in the pathophysiology of the disease.

According to a first aspect, there is provided a method for determining a clinical outcome of a subject having received an initial diagnosis of cancer. Broadly, the method comprises measuring an individualized gene expression level in a plurality of genes of a CUX-1 signature gene set in a sample of the subject, wherein the CUX-1 signature gene set comprises at least one gene associated with genetic instability and at least one gene associated with cell proliferation; normalizing the individualized gene expression level against at least one control gene expression level to provide a normalized individualized gene expression level; comparing at least one parameter of the normalized individualized gene expression level with data derived from a set of CUX-1 expression profiles having known clinical outcomes divided in at least two prognostic classes in accordance with the known clinical outcomes; and associating one of the known clinical outcomes with the normalized individualized gene expression level in accordance with the comparison. In an embodiment, the at least one gene associated with genetic instability is selected from the group consisting of AURKB, BUB1 , BUBR1 , BUB3, CENPA, CENPE, KIF2C, KIFC1 , KNTCI (ROD), MAD1 L1 , MAD2L1 , NEK2, TTK, ZW10 and ZWILCH. In another embodiment, the at least one gene associated with cell proliferation is selected from the group consisting of CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CDKN1 A, MCM3, MCM7, ORCL1 , POLA, and POLA2. In yet another embodiment, the individualized gene expression level is measured in at least 20 different genes of the CUX1 signature gene set. In still another embodiment, the at least two prognostic classes comprises a first prognostic class and a second prognostic class. In yet a further embodiment, the CUX1 expression profiles associated with the first prognostic class comprises an increase, with respect to CUX1 expression profiles associated with the second prognostic class, in the gene expression level of at least one of the following gene AURKB, BUB1 , BUB3, BUBR1 , CENPE, MAD1 L1 , MAD2L1 , TTK, NEK2, KNTC1 , ORCL1 , ZW10, ZWILCH, CENPA, KIF2C, KIFC1 , CAMK2D, COG7, MFAP1 , NUDCD1 , RIN2, SF3B3, CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CTSL1 , MCM3, MCM7, POLA and POLA2. In still another embodiment, the CUX1 expression profiles associated with the first prognostic class comprises a decrease, with respect to the CUX1 expression profiles associated with the second prognostic class, in the gene expression level of the CDKN1A gene. In still a further embodiment, the first prognostic class is associated with a poor prognosis. In another embodiment, the CUX1 expression profiles associated with the second prognostic class comprises a decrease, with respect to the CUX1 expression profiles associated with the first prognostic class, in the gene expression level of at least one of the following gene AURKB, BUB1 , BUB3, BUBR1 , CENPE, MAD1 L1 , MAD2L1 , TTK, NEK2, KNTC1 , ORCL1 , ZW10, ZWILCH, CENPA, KIF2C, KIFC1 , CAMK2D, COG7, MFAP1 , NUDCD1 , RIN2, SF3B3, CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CTSL1 , MCM3, MCM7, POLA and POLA2. In yet another embodiment, the CUX1 expression profiles associated with the second prognostic class comprises an increase, with respect to the CUX1 expression profiles associated with the first prognostic class, in the gene expression level of the CDKN1A gene. In still a further embodiment, the second prognostic class is associated with a good prognosis. In another embodiment, the CUX1 expression profiles comprise gene expression levels for the plurality of genes of the CUX1 signature gene set from an heterogeneous population of subjects having been diagnosed with cancer. In still another embodiment, the individualized gene expression level is measured by determining the mRNA level of the plurality of genes. In a further embodiment, the method further comprises applying a probability model (such as, for example a naive Bayes' modifier) to the normalized individualized gene expression level and generating an output, wherein the data derived from the set of CUX-1 expression profiles comprises a scoring scheme derived therefrom, and wherein comparing at least one parameter of the normalized individualized gene expression level comprises applying the output to the scoring scheme. In still another embodiment, the known clinical outcomes comprises cancer recurrence, cancer metastasis, survival and/or usefulness of a chemotherapy treatment. In still another embodiment, the cancer is breast cancer.

According to a second aspect, the present application provides a system for determining a prognosis in a subject having received an initial diagnosis of cancer. Broadly, the system comprises a reaction vessel for combining a sample from the subject and an analyte-specific reagent (ASR) for measuring the gene expression level of a plurality of genes of a CUX1 signature gene set in a sample of the subject, wherein the CUX1 signature gene set comprises at least one gene associated with genetic instability and at least one gene associated with cell proliferation; a processor in a computer system; a memory accessible by the processor; and at least one application coupled to the processor. The at least one application is configured for receiving a measure of an individualized gene expression level in the plurality of genes of the CUX- 1 signature gene set in a sample of the subject; normalizing the individualized gene expression level against at least one control gene expression level to provide a normalized individualized gene expression level; comparing at least one parameter of the normalized individualized gene expression level with data derived from a set of CUX-1 expression profiles having known clinical outcomes divided into at least two prognostic classes in accordance with the known clinical outcomes; and associating one of the known clinical outcomes with the normalized individualized gene expression level in accordance with the comparison. Embodiments of the at least one gene associated with genetic instability, the at least one gene associated with cell proliferation, the number of genes assayed in the CUX1 signature gene set, the different prognostic classes and the level of gene expression associated thereto, the known clinical outcomes, the set of CUX1 expression profiles used, the measure of the gene expression level, the application of a probability model to the comparison step and the type of cancer have been described above and do apply herein. In a further embodiment, the ASR comprises at least one oligonucleotide listed in Table 4. In still another embodiment, the system further comprises a display device for outputting of the associated known clinical outcome.

According to a third aspect, there is provided a software product embodied on a computer readable medium and comprising instructions for determining a prognosis in a subject having received an initial diagnosis of cancer. It broadly comprises a receiving module for receiving a measurement of an individualized gene expression level in a plurality of genes in a CUX1 signature gene set in a sample of the subject, wherein the CUX1 signature gene set comprises at least one gene associated with genetic instability and at least one gene associated with cell proliferation; a normalizing module for normalizing the individualized gene expression level against at least one control gene expression level to provide a normalized individualized gene expression level; a comparison module for comparing at least one parameter of the normalized individualized gene expression level with data derived from a set of CUX-1 expression profiles having known clinical outcomes divided at least two prognostic classes in accordance with the known clinical outcomes; and a characterization module for associating one of the known clinical outcomes with the normalized individualized gene expression level in accordance with the comparison. Embodiments of the at least one gene associated with genetic instability, the at least one gene associated with cell proliferation, the number of genes assayed in the CUX1 signature gene set, the different prognostic classes and the level of gene expression associated thereto, the known clinical outcomes, the set of CUX1 expression profiles used, the measure of the gene expression level, the application of a probability model to the comparison step and the type of cancer have been described above and do apply herein.

BRIEF DESCRIPTION OF THE DRAWINGS AND TABLES

Having thus generally described the nature of the invention, reference will now be made to the accompanying drawings, showing by way of illustration, a preferred embodiment thereof, and in which:

Figure 1. Tetraploidy and chromosomal instability in p1 10 CUX1 -expressing NmuMG cells. (A) NMuMG cells stably expressing p1 10 CUX1 (a. a. 747-1505) were generated by retroviral infection. More than 500 colonies were pooled after selection which was considered passage 1 . Cell cycle distribution was determined by FACS. Results are shown for passage 5 (left panels) and passage 30 (right panels) for empty vector (upper panels) of CUX1 vector (lower panels). The number on the various panels indicates percentage of >4C cells. (B) FACS analysis (at passage 16) was performed on independent populations of control NMuMG cells (left panel), NMuMG cells expressing either p1 10 CUX1 (middle panel) or p1 0^{S 237}- ^270ACUX (right panel). The number on the various panels indicates percentage of >4C cells. (C) Late passage NMuMG/CUX1 cells were FACS-sorted based on DNA content following Hoechst 33342 staining and put back in culture. 2C represent G1 diploid cells and 8C are G2/M tetraploids. (D-E) The number of centrosomes during interphase (D, n=304) or spindle poles during early mitosis (E, n=250 for each cell line) was determined by indirect immunofluorescence based on γ-tubulin staining, while microtubules and DNA were revealed by a-tubulin and DAPI staining. (F) Chromosome segregation during anaphase was studied by confocal microscopy for late-passage NMuMG/vector (n=31 ), early passage NMuMG/CUX1 (n=50) and late-passage 2C-sorted (n=57) 8C-sorted (n=23) NMuMG/CUX1 cells. Cells were stained for y-tubulin, a-tubulin and DNA. Percentage of normal cells are presented in the grey bars, whereas percentage of abnormal cells are presented in the dotted bars. (G) Chromosome counts were determined on metaphase spreads prepared from low passage NMuMG/p1 10 CUX1 or late-passage 2C or 8C sorted NMuMG/p1 10 CUX1 cells (n > 100 each). Percentage of cells for the early passage are shown in the grey bars, for the late passage sorted 2C cells in the dotted bars, for the late passage sorted 8C cells in the white bars. (H) NMuMG/vector and sorted 8C NMuMG/CUX1 cells were left untreated (-) or exposed (+) to 20mJ/cm2 UV. Cell lysates were prepared 24h later and analyzed by SDS-PAGE and western blotting using the indicated antibodies.

Figure 2. p1 10 CUX1 Expression Enables Bipolar Mitoses in Newly Formed Tetraploid Cells. (A) Proliferation curves of NMuMG/vector (regular line) and NMuMG/CUX1 (dashed line) cells in the absence of blebbistatin, as well as NMuMG/vector (regular bold line) and NMuMG/CUX1 (dotted bold line) cells in the presence of blebbistatin a 12h blebbistatin pulse treatment were determined. t₂ was also determined for NMuMG/vector (12.9 h) and NMuMG/CUX1 (12.8 h) cells in the absence of blebbistatin as well as for NMuMG/vector (17.0 h) and NMuMG/CUX1 (13.7 h) cells in the presence of blebbistatin. (B) The percentage of apoptotic cells was measured 48h after blebbistatin wash-off, based on Annexin V staining and FACS analysis. Cells infected with empty vector are shown on the left (regular line), whereas cells infected with CUX1 vector (bold line) are shown on the right. The number on the panel indicates percentage of >4C cells for each conditions. (C) The outcome of cell divisions after a blebbistatin wash-off was determined by time-lapse microscopy using frames taken every 5 minutes. Multipolar divisions displayed a multipolar anaphase giving rise to three or four daughter cells (data not shown) or a tripolar anaphase followed by cytokinesis failure resulting in two cells, one of which with two nuclei (data not shown). Neighboring mononucleated cells were used as controls. (D) U20S and MCF10A cells expressing p1 10 CUX1 were treated with blebbistatin and the fate of bi-nucleated cells was determined by time-lapse microscopy. (E) NMuMG/CUX1 were pulse-treated with blebbistatin, and imaged by time-lapse microscopy with increasing concentrations of MPS1 -IN-1 . The outcome of cell division and the duration of mitosis were scored for mono- and bi-nucleated cells. (F) U20S/vector cells were pulse-treated 8h with blebbistatin. 16 h later, cells were imaged by time-lapse microscopy during which 10 μΜ MG132 was added for 90 minutes and washed away. The fate of binucleated cells entering mitosis during (+) or just before (-) MG132 treatment or in the presence of a control (DMSO) was examined. Neighboring binucleated cells dividing without being exposed to MG132 were used as. (C-F) Percentage of bipolar division shown in grey in the bars. Percentage of multipolar division shown in dotted in the bars *: p < 0.005; **: p < 0.0002; ***: p < 1 .2x10-5.

Figure 3. CUX1 -Induced Tumorigenicity Involves Chromosomal Instability. (A) Late- passage NMuMG/vector cells (vector), 2C sorted NMuMG/CUX1 (CUX1 2C) or aneuploid 8C sorted NMuMG/CUX1 (CUX1 Sorted 8C) cells were injected subcutaneously in nude mice. Tumor volumes (mm³) were measured 27 days after injection. (B) Early-passage NMuMG cells expressing p1 10 CUX1 (CUX1 ) or not (vector) were pulse-treated with blebbistatin and injected subcutaneously into nude mice. (C) Chromosome number of cells from six mammary gland tumors arising in p1 10 or p75 CUX1 transgenic mice was determined from metaphase spreads. Result shown as percentage of cells in function of the number of chromosomes per cells **: p < 0.0003 for both tumor sizes and tumor numbers.

Figure 4. CUX1 Transcriptional Targets are Highly Expressed in Breast Tumor Subtypes with Poor Prognosis, and Identify a Subset of Luminal Tumors with Poor Outcome. (A) Kaplan-Meier curve displaying significant segregation of good outcome vs. bad outcome patients classified into "low" and "high" groups based on the CUX1 gene signature. HR = 1 .92, p < 1 x 10^"12, ni_ow = 1421 , n_high = 1060. (B1 -B4) Kaplan- Meier curves comparing the outcome of low- and high-expressing groups for patients reported to be ER+/LN- (B1 , HR = 2.70, p = 1 x 10^"8, ni_ow = 388, n_high = 196) and ER+ /LN+ (B2, HR = 2.00, p = 2.3 x 10^"5, ni_ow = 301 , n_high = 186). Below, ER+/LN- patients were further classified into luminal A (B3, HR = 4.00, p = 0.0061 , ni_ow = 168, n_high = 22) and luminal B (B4, HR = 2.13, p = 0.014, ni_ow = 1 10, n_high = 69) subtypes. Low group is presented with a regular bold line, whereas the high group presented with a regular line.

Figure 5. Characterization of polyploid HEK293 and NMuMG Stable Cell Lines. (A) Cell cycle profiles of HEK293/CUX1 (lower panels) or vector controls (upper panel) after one (left panels) or two (right panel) months in culture, where complete tetraploidization was observed (>2 months). (B) Cell cycle profiles of independent populations of HEK293 cells with empty vector (upper panels) or HEK293 cells expressing wild-type p1 10 CUX1 (lower panel) at passage 10 (left panels) or passage 35 (right panels). The number on the various panels indicates percentage of >4C cells. (C) Cell cycle profiles of independent populations of HEK293 cells with empty vector (left panel), HEK293 cells expressing wild-type p1 10 CUX1 (middle panel) or HEK293 cells expressing _{p 1 1 0}cuxisi237^,i270A _{m utant} (_{rjght pane}|) _{gt passage 2}5 (|_eft panels). The number on the various panels indicates percentage of >4C cells. (D) Phosphorylation at serine 1237, which results in inhibition of CUX1 DNA binding, becomes detectable as cells progress in G2. NmuMG cells expressing p1 10 CUX1 were synchronized in GO by serum starvation for 3 days and collected after serum re-stimulation (1 = Oh, 2 = 18h, 3 = 19h, 4 = 20h). Cell cycle profiles are shown on the upper panel. Immunoblotting, shown on the two lower panels, was performed on whole cell extracts using purified CUX1 - pS1237 or CUX1 antibodies using 80 and 5 of cell lysate, respectively. (E) Late- passage NMuMG/vector and NMuMG/p1 10 CUX1 cells were pulse-labeled with 100 μΜ BrdU for one hour, fixed in paraformaldehyde for double staining using propidium iodide and an Alexa 649-conjugated anti-BrdU antibody. Results are shown for BrdU staining in function of propidium iodide staining. (F) Late passage tetraploid NmuMG/CUX1 are larger as determined by FACS analysis based on forward scatter. Results shown for empty vector (regular line) and CUX1 cells (regular bold line). Late passage NmuMG/CUX1 (unsorted population) contain cells with various nuclear sizes (data not shown).

Figure 6. p1 10 CUX1 expression allows cells to sustain a longer mitotic arrest and delays cyclin B degradation upon prolonged exposure to nocodazole. (A) U20S/CUX1 (dotted) and control (grey) cells were treated with nocodazole (10 h, 80 ng/ml) after which mitotic floating cells were harvested (t = Oh) and incubated further in the presence of the microtubule poison nocodazole for the indicated time-points. Cells were fixed and the fraction of mitotic cells was determined by FACS based on phospho-histone H3 staining, and normalized to t = Oh for each cell line. Percentage increase vs. vector is indicated. (B) Western blot analysis for cyclin B and γ-tubulin was done using cell lysates for control cells (U20S/Vect) and U20S/CUX1 cells. (C) Densitometric analysis was done using ImageJ™ and the ratio of cyclin Β/γ-tubulin for each time-point was expressed as a percentage relative to t = Oh. Results are shown for control cells (■) and U20S/CUX1 cells (·) in function of time (hours).

Figure 7. Expression of 21 CUX1 targets are co-overexpressed in epithelial cells isolated from mammary tumors of CUX1 transgenic mice. Expression profiling on microdissected epithelial cells was performed using mammary gland tumors or the corresponding tumor-free adjacent mammary gland of CUX1 transgenic mice and normal mammary glands of non-transgenic littermates. (A) Distribution of the scaled average gene expression of the CUX1 targets in the three different groups (non- transgenic, CUX-1 transgenic (tumor free and tumor). These graphs show that the overall expression of the CUX1 targets is higher in tumor-free epithelial cell from CUX1 transgenic mice compared to normal tissue, and an even higher in CUX1 -driven tumor cells. On average, there was a 1 .75 fold increase in the expression of the 21 CUX1 targets in adjacent tissues of transgenic mice compared to normal non-transgenic tissues (*, p = 0.014), and a 2.31 fold increase in tumors versus transgenic adjacent tissues (**, p = 0.0016). (B-C) Gene Set Enrichment Analysis (GSEA). Each vertical line represents one of the CUX1 targets in the signature, the position of the line indicates its enrichment relative to the enrichment of every other gene on the array with the leftmost position indicating the most enriched gene. The high density of genes at the left of the distribution indicates that the signature as a whole is enriched in this sample. The green line above represents the enrichment score calculated for the signature and the p-value is indicated. GSEA indicated that the 21 CUX1 targets were significantly overexpressed in tissue from CUX1 transgenic mice ((B) tumor-free CUX1 epithelium vs. normal epithelium, p = 0.0065) and was selected during tumorigenesis in CUX1 transgenic mice ((C) CUX1 tumor vs. tumor-free CUX1 epithelium, p = 0.0029). Tests were performed using 3500 gene set permutations on the whole array. Lastly, a hypergeometric test confirmed that the 21 -gene set was significantly over-expressed compared to the entire array (adjacent vs. normal, p = 0.02; tumor vs. adjacent, p = 0.01 ). Figure 8. A CUX1 -gene signature predicts outcome in 8/12 datasets. Kaplan-Meier survival analysis and the log-rank test were used to compare the patients in the "low" versus the "high" classes. (A-H) represents datasets having a significant p-value (A - CHIN (HR = 2.13, p = 0.028, HIGH n = 37, LOW n = 81 ), B - DESMET (HR = 2.22, p = 0.00568, HIGH n = 105, LOW n = 93), C- IVSHINA (HR = 2.13, p = 0.00052, HIGH n = 131 , LOW n = 1 18), D - LOI (HR = 2.17, p = 0.00037, HIGH n = 109, LOW n = 305), E - NKI (HR = 3.13, p = 2 x10^"6, HIGH n = 203, LOW n = 92), F- PARKER (HR = 2.70, p = 0.0012, HIGH n = 80, LOW n = 145), G - SCHMIDT (HR = 1 .85, p = 0.040, HIGH n = 87, LOW n = 1 13), H - PAWITAN (HR = 3.33, p = 0.028, HIGH n = 55, LOW n = 104)) , whereas (l-L) represent datasets having p-value > 0.05 (I - ANDERS (HR = 0.71 , p = 0.55, HIGH n = 35, LOW n = 43), J - BILD (HR = 1 .47, p = 0.24, HIGH n = 1 15, LOW n = 43), K- SOTIRIOU (HR = 1 .82, p = 0.16, HIGH n = 31 , LOW n = 70), L - WANG (HR = 1 .39, p = 0.14, HIGH n = 72, LOW n = 214)). Low class shown in regular bold line, whereas high class shown in regular line. The analysis was completed using the survival package in R/Bioconductor (Gentleman et al., 2004).

Figure 9. Predictive value of a CUX1 -gene signature within intrinsic (molecular) subtypes and subgroups with defined clinico-pathological features. Patients from all 12 datasets were combined and Kaplan Meier curves show the differential outcome of patients with low or high expression of the CUX1 signature. (A) Outcome analysis within intrinsic subtypes defined using the PAM50 classifier (A1 - LUMINAL A (HR = 1 .96, p = 0.013, HIGH n = 72, LOW n = 557), A2 - LUMINAL B (HR = 1 .56, p = 0.0028, HIGH n = 233, LOW n = 364), A3 - HER2+ (HR = 1 .23, p = 0.21 , HIGH n = 251 , LOW n = 139), A4 - BASAL-LIKE (HR = 1 .37, p = 0.16, HIGH n = 471 , LOW n = 83), A5 - NORMALLIKE (HR = 1 .61 , p = 0.19, HIGH n = 33, LOW n = 278)). (B) Outcome analysis based on grade (B1 - GRADE 1 (HR = 2.56, p = 0.0034, HIGH n = 62, LOW n = 302), B2 - GRADE 2 (HR = 1 .89, p = 1 .9 x 10^"6, HIGH n = 284, LOW n = 550), B3 - GRADE 3 (HR = 1 .64, p = 0.0014, HIGH n = 482, LOW n = 206)). (C) Outcome based on lymph node status (C1 - LN- (HR = 2.33, p = 4 x 10^"8, HIGH n = 318, LOW n = 468), C2 - LN+ (HR = 1 .92, p = 2 x 10^"6, HIGH n = 284, LOW n = 345), C3 - LN+/HER2+ (HR = 2.63, p = 0.0014, HIGH n = 71 , LOW n = 49)). (D) Outcome analysis based on ER status (D1 - ER+ (HR = 2.38, p = 5 x 10^"13, HIGH n = 388, LOW n = 705), D2 - ER+/LUMINAL A (HR = 2.38, p = 0.015, HIGH n = 40, LOW n = 308), D3 - ER+/LUMI NAL B (HR = 1 .92, p = 0.00086, HIGH n = 142, LOW n = 208), D4 - ER- (HR = 1 .51 , p = 0.047, HIGH n = 212, LOW n = 1 1 1 )). (E) Outcome based on ER status, lymph node status and HER2+ status (E1 - ER+/LN+ (HR = 2.00, p = 2.3 x 10^"5, HIGH n = 186, LOW n = 301 ), E2 - ER+/LN+/HER2+ (HR = 2.27, p = 0.037, HIGH n = 46, LOW n = 31 )). Survival of patient subsets were analyzed based on molecular subtype and results in which the 29-gene signature showed significant differences in outcome are shown. Low class shown in regular bold line, whereas high class shown in regular line.

Table 1 lists the genes of the CUX1 signature gene set together with corresponding Entrez Gene ID numbers and exemplary Accession Numbers.

Table 2 depicts the validation of p1 10 CUX1 transcriptional targets relevant to the establishment of bipolar mitoses. Putative CUX1 transcriptional targets were identified by chromatin immunoprecipitation (ChIP) in Hs578T cells followed by hybridization on the Human HG18 promoter ChlP-on-chip oligo microarray set of NimbleGen. Column 1 contains the gene symbols and molecular functions; The asterisk (*) indicates genes identified in a previous siRNA screen (Kwon et al., 2008). Column 2 shows the validation of promoter occupancy as determined by real-time PCR, from an independent ChIP experiment in HeLa cells expressing p1 10 CUX1 . Column 3 shows fold difference in mRNA expression measured by real-time PCR between NMuMG/CUX1 and control cells. Column 4 shows the fold difference in mRNA levels measured by real-time PCR between Hs578t cells bearing a doxycyclin-inducible CUX1 specific shRNA, treated or not for 5 days with doxycyclin. Column 5 reveals the fold mRNA difference in target expression after endogenous CUX1 expression was permitted following removal of doxycyclin for 3 days.

Table 3 shows the univariate and multivariate Cox Proportional Hazards analysis performed on patients for which all clinical variables were reported (n = 1474, 8 datasets, see Fig. 4). The "*" (asterisk) denotes the 95% confidence interval.

Table 4 shows the primers used for RT-PCR validation.

Table 5 shows results of cell division when followed by time-lapse microscopy.

Table 6 shows results of blebbistatin treatment and the generation of binucleated cells.

Table 7 shows percentage of cells containing a specific number of chromosomes in tumors arising in p1 10 or p75 CUX1 transgenic mice determined from metaphase spread. Table 8 shows the breakdown of the percentage of patients classified into each molecular subtype for "low" and "high" expression groups.

Table 9 shows the specificity of CUTL1 probes in microarray datasets.

DETAILED DESCRIPTION

I. Definitions

As used herein, the "CUX1 gene set" or the "CUX1 signature gene set" refers to a combination of at least two classes of genes (a genetic instability gene and a cell proliferation gene) whose expression level is associated with the activity of the transcription factor CUX1 . The genes of the CUX1 gene set are associated to CUX1 because their expression is either modulated by CUXI 's expression, activity and/or stability (these genes are also referred to as CUX1 targets) or they modulate CUXI 's expression, activity and/or expression. In an embodiment, the CUX1 signature gene set comprises a combination of at least twenty of the following genes: AURKB, BUB1 , BUB3, BUBR1 (BUB1 B), CENPE, MAD1 L1 , MAD2L1 , TTK (MPS1 ), NEK2, KNTCI (ROD), ORCL1 , ZW10, ZWILCH, CENPA, KIF2C, KIFC1 , CAMK2D, COG7, MFAP1 , NUDCD1 , RIN2, SF3B3, CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CTSL1 , MCM3, MCM7, CDKN1A, POLA, and/or POLA2. In some embodiments, the CUX1 signature get set does not contain information with respect to the CDKN1 A gene, the KNTCI (ROD) gene, the ORCL1 gene and/or the CCNE2 gene. In other embodiments, the CUX1 signature get set does not contain information with respect to the KNTCI (ROD) gene and/or the ORCL1 gene. In yet a further embodiment, the embodiments, the CUX1 signature get set does not contain information with respect to the CCNE2 gene. In still another embodiment, the CUX1 signature gene set does not contain information with respect to the CDKN1A. In some embodiment, the CUX1 signature gene get refers to a 29-gene signature. The 29-gene signature includes the following genes: AURKB, BUB1 , BUB3, BUBR1 (BUB1 B), CENPE, MAD1 L1 , MAD2L1 , TTK (MPS1 ), NEK2, ZW10, ZWILCH, CENPA, KIF2C, KIFC1 , CAMK2D, COG7, MFAP1 , NUDCD1 , RIN2, SF3B3, CCNA2, CDC25A, CDC45L, CDC7, CTSL, MCM3, MCM7, CDKN1A, POLA, and/or POLA2. The CUX1 signature get set also includes nucleic acids (including gene, pre-mRNA, and mRNA), polypeptides, as well as polymorphic variants, alleles and mutants of those genes. Truncated and alternatively spliced forms as well as complementary sequences are also included in the definition. Entrez Gene ID numbers and exemplary accession numbers for the CUX1 signature gene set genes are provided in Table 1 , and are herein specifically incorporated by reference.

The first class of genes in the CUX1 signature gene set are involved in genetic instability. As used herein, genetic instability refers to any change in DNA content or DNA sequence. This includes, for example, the acquisition of extra chromosome copies or chromosome portions and loss of chromosome or chromosome portion. Genetic instability itself involves many cellular processes such as chromosomal segregation defects and more specifically defects in spindle assembly or cytokinesis leading to aneuploidy. The genes involved in the acquisition and maintenance of genetic instability include, but are not limited to, AURKB, BUB1 , BUBR1 (BUB1 B), BUB3, CENPA, CENPE, KIF2C, KIFC1 , KNTC1 (ROD), MAD1 L1 , MAD2L1 , NEK2, TTK (MPS1 ), ZW10 and ZWILCH.

The second class of genes in the CUX1 signature get set are involved with cell proliferation. As used herein, cell proliferation refers to the ability of a cell to divide. Cell proliferation itself involves many cellular processes throughout all phases in the cell cycle, but processes associated with DNA replication during S phase are often used to monitor cell proliferation. The genes involved in cell proliferation include, but are not limited to, CDKN1A, CCNA2, CCNE2, CDC25A, CDC45L, CDC7, MCM3, MCM7, POLA, POLA2, and ORCL1 .

As it will be shown herewith, the level of expression the genes of the CUX1 signature gene set differs between subjects afflicted with cancer having different clinical outcome or prognostic class, e.g. good an poor prognosis. Therefore, the type of modulation (up- or down-regulation) of the level of expression of the genes of CUX1 signature gene set are similar within individuals of the same prognostic class and are different between individuals of different prognostic class. It is recognized herein that the level of all the genes of the CUX1 signature gene set but one (CDKN1 A) is higher in subjects having a poor prognosis when compared to the level of expression of the same genes in subjects having a good prognosis. In the subjects having a poor prognosis, the level of expression of the CDKN1A gene is lower than in subjects having a good prognosis. Conversely, the level of expression of all the genes of the CUX1 signature gene set but one (CDKN1A) is lower than in subjects having a good prognosis when compared to the level of expression of the same genes in subjects having a poor prognosis. In the subjects having a good prognosis, the level of expression of the CDKN1A gene is higher than in subjects having a poor prognosis.

As it will also be recognized, not all the genes of the CUX1 signature gene set are differentially express each in each individuals of a single prognostic class.

As used herein an "expression profile" refers to, for a plurality of genes, gene expression levels and/or pattern of gene expression levels that is for example, useful for determining a clinical outcome such as good prognosis or poor prognosis. For example, an expression profile can comprise the expression levels of genes of the CUX1 signature gene set, and the gene expression levels can be compared to one or more reference profiles, and based on similarity to a reference profile known to be associated with particular classes, be diagnostically or prognostically predicted to belong to a certain class.

A "CUX1 expression profile" or "CUX1 profile" as used herein refers to the expression signature (e.g. gene expression levels and/or pattern) of a plurality of genes or a gene set associated of the CUX1 signature gene set. The CUX1 expression profile is derived from a plurality of samples comprising cancer (e.g. breast cancer) tissue/cells wherein the type and level of gene expression of the CUX1 signature gene set is similar between related samples defining a specific prognostic class and is different to unrelated samples defining a different prognostic class. As such, it is preferable that the CUX1 reference profile be determined from a heterogeneous population of subjects having been diagnosed with cancer. As used herein, the term "heterogeneous" indicates that the population of subjects comprises subjects suffering cancer but having different clinical outcomes, such as good and poor prognosis, and whose gene expression levels of the CUX1 signature set can be useful in the methods set forth below. This heterogeneous population can comprise individuals with different cancer types or a single cancer type. The CUX1 expression profile is accordingly a reference profile or reference signature of the expression of genes of the CUX1 signature gene set, to which the expression levels of the corresponding genes in a patient sample are compared in methods for determining or predicting clinical outcome, e.g. good prognosis or poor prognosis. The CUX1 expression profiles can be determined from publicly available gene expression datasets.

A "CUXI -low expression profile" is also a reference expression profile, but it is associated with a specific outcome, e.g. good prognosis. It can be derived from biological samples or cultured cells in which the expression level/signal is similar between related samples defining an outcome class (e.g. good prognosis) and is different to unrelated samples defining a different outcome class (e.g. poor prognosis). As such, it is preferable that the CUX1 -low expression profiles be determined from a relatively homogeneous population of subject with respect to a specific clinical outcome, e.g. good prognosis. As used herein, the term "relatively homogeneous" indicates that the population of subjects comprises subjects suffering cancer have the same clinical outcome, good prognosis, and whose gene expression levels of the CUX1 signature set can be useful in the methods set forth below. This relatively homogeneous population can comprise individuals with different cancer types or a single cancer type. It can also be to referred to as a "good prognosis profile" or a "good profile". It is derived from one or more samples comprising cancer (e.g. breast cancer) tissue/cells wherein the gene expression profile is similar between samples and is associated with the specific outcome (e.g. good prognosis). In an embodiment, the "CUX1 -low expression profile" presents a decrease, with respect to the "CUX1 -high expression profile", in the gene expression level in at least one, at least five, at least ten or at least fifteen of the following gene AURKB, BUB1 , BUB3, BUBR1 , CENPE, MAD1 L1 , MAD2L1 , TTK, NEK2, KNTC1 , ORCL1 , ZW10, ZWILCH, CENPA, KIF2C, KIFC1 , CAMK2D, COG7, MFAP1 , NUDCD1 , RIN2, SF3B3, CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CTSL, MCM3, MCM7, POLA, and/or POLA2. In still a further embodiment, in order to be considered decreased, the level of gene expression is 50%, 60%, 70 %, 80%, 90%, 93%, 95%, 97% and even 99% lower than the level of expression of the same gene in the "CUX1 -high expression profile". In a further embodiment, the expression of all the genes are not lowered, only 70%, 80%, 85%, 90%, 95% of the genes a exhibit a decreased expression level or signal compared to the "CUX1 -high expression profile". In yet another embodiment, the "CUX1 -low expression profile" presents an increase, with respect to the "CUX1 -high expression profile", in the gene expression level of the CDKN1A gene. In still a further embodiment, in order to be considered increased, the level of gene expression of the CDKN1 A is at least 10%, 20%, 30 %, 40%, 50%, 60%, 70%, 80% and even 100% higher than the level of expression of the same gene in the "CUX1 -high expression profile".

The term "CUX1 signature-low" refers to an expression level, or when in reference to a subject, a subject whose expression level, falls within a group who have decreased signal from the CUX1 signature gene set as determined for example, following hierarchical clustering for example using Euclidean distance and Wards algorithm.

A "CUX1 -high expression profile" is also a reference expression profile, but it is associated with a specific outcome, e.g. poor prognosis. It can be derived from biological samples or cultured cells in which the expression level/signal is similar between related samples defining an outcome class (e.g. poor prognosis) and is different to unrelated samples defining a different outcome class (e.g. good prognosis) such that the reference expression profile is associated with a particular clinical outcome. As such, it is preferable that the CUX1 -high expression profile be determined from a relatively homogeneous population of subject with respect to a specific clinical outcome, e.g. poor prognosis. As used herein, the term "relatively homogeneous" indicates that the population of subjects comprises subjects suffering cancer have the same clinical outcome, poor prognosis, and whose gene expression levels of the CUX1 signature set can be useful in the methods set forth below. This relatively homogeneous population can comprise individuals with different cancer types or a single cancer type. It can also be referred to as a "poor prognosis reference profile" or a "poor profile". It is derived from one or more samples comprising cancer (e.g. breast cancer) tissue/cells wherein the gene expression profile is similar between samples and is associated with the specific outcome (e.g. poor prognosis). In an embodiment, the "CUX1 -high expression profile" presents an increase, with respect to the "CUX1 -low expression profile", in the gene expression level in at least one, at least five, at least ten or at least fifteen of the following gene AURKB, BUB1 , BUB3, BUBR1 , CENPE, MAD1 L1 , MAD2L1 , TTK, NEK2, KNTC1 , ORCL1 , ZW10, ZWILCH, CENPA, KIF2C, KIFC1 , CAMK2D, COG7, MFAP1 , NUDCD1 , RIN2, SF3B3, CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CTSL, MCM3, MCM7, POLA, and/or POLA2. In still a further embodiment, in order to be considered increased, the level of gene expression of the genes is at least 10%, 20%, 30 %, 40%, 50%, 60%, 70%, 80% and even 100% higher than the level of expression of the same gene in the "CUX1 -low expression profile". In a further embodiment, the expression of all the genes are not increased, only 70%, 80%, 85%, 90%, 95% of the genes a exhibit an increased expression level or signal compared to the "CUX1 -low expression profile". In another embodiment, the "CUX1 - high expression profile" presents a decrease, with respect to the "CUX1 -low expression profile", in the gene expression level of the CDKN1 A gene. In still a further embodiment, in order to be considered decreased, the level of gene expression of the CDKN1 A gene is 50%, 60%, 70 %, 80%, 90%, 93%, 95%, 97% and even 99% lower than the level of expression of the same gene in the "CUX1 -low expression profile".

The term "CUX1 signature-high" refers to an expression level, or when in reference to a subject, a subject whose expression level falls within a group who have increased signal of the CUX1 signature gene set as determined, for example, following hierarchical clustering for example using Euclidean distance and Wards algorithm.

As used herein "outcome" or "clinical outcome" refers to the resulting course of disease and/or disease progression and can be characterized for example by recurrence, period of time until recurrence, metastasis, period of time until metastasis, number of metastasis, number of sites of metastasis and/or death due to disease. For example a good clinical outcome includes cure, prevention of recurrence, prevention of metastasis and/or survival within a fixed period of time (without recurrence), and a poor clinical outcome includes disease progression, metastasis and/or death within a fixed period of time.

As used herein "prognosis" refers to an indication of the likelihood of a particular clinical outcome, for example, an indication of likelihood of recurrence, metastasis, and/or death due to disease, overall survival or the likelihood of recovery and includes a "good prognosis" and a "poor prognosis". A "prognostic class" refers to a subset of CUX1 expression profiles associate with the same known clinical outcome.

As used herein, "good prognosis" indicates that the subject is expected (e.g. predicted) to survive and/or have no, or is at low risk of having, recurrence or distant metastases within a set time period. The term "low" is a relative term and, in the context of this application, refers to the risk of the "low" expression group with respect to a clinical outcome (recurrence, distant metastases, etc.). A "low" risk can be considered as a risk lower than the average risk for an heterogeneous cancer patient population. In the study of Paik et al. (2004), an overall "low" risk of recurrence was considered to be lower than 15%. The risk will also vary in function of the time period. The time period can be, for example, five years, ten years, fifteen years or even twenty years after initial diagnosis of cancer or after the prognosis was made.

As used herein, "poor prognosis" indicates that the subject is expected e.g. predicted to not survive and/or to have, or is at high risk of having, recurrence or distant metastases within a set time period. The term "high" is a relative term and, in the context of this application, refers to the risk of the "high" expression group with respect to a clinical outcome (recurrence, distant metastases, etc.). A "high" risk can be considered as a risk Ihigher than the average risk for an heterogeneous cancer patient population. In the study of Paik et al. (2004), an overall "high" risk of recurrence was considered to be higher than 15%. The risk will also vary in function of the time period. The time period can be, for example, five years, ten years, fifteen years or even twenty years of initial diagnosis of cancer or after the prognosis was made.

As used herein, the term "recurrence" refers to the reappearance of cancer, such as breast cancer within a set period of time from initial diagnosis.

As used herein, the term "disease-free survival" refers to the lack of reappearance of cancer, such as breast cancer, within a set period of time from initial diagnosis.

As used herein, the "cancer" or "malignant neoplasm" is a class of diseases in which a group of cells display uncontrolled growth, invasion that intrudes upon and destroys adjacent tissues, and may metastasize to other locations in the body. Cancers can be classified by the type of cell that the tumor resembles and is therefore presumed to be the origin of the tumor. These types include carcinoma: (epithelial cells), carcoma (connective tissue, or mesenchymal cells), lymphoma and leukemia (hematopoietic cells), germ cell tumor (pluripotent cells), blastoma (immature "precursor" or embryonic tissue cells).

As used herein, a "breast cancer" is a cancer originating from a breast tissue. Breast cancer can be classified by their immunohistochemical signature, the presence certain hormonal receptors, transcriptional signatures, etc.

As used herein, the "luminal sub-type" refers to breast cancers that are typically positive immunohistochemically for the estrogen receptor (ER+) and/or comprise an ER+ transcriptional signature as described in Perou et al., 2000; Sorlie et al., 2003 and Sorlie et al., 2001 .

As used herein, the "luminal A subtype" refers to breast cancers that typically display the highest expression of luminal-specific genes including the ER+ signature (Sorlie et al., 2001 , Sorlie et al. , 2003). Immunohistochemically they are classified as ER+ and/or progesterone receptor (PR+) and HER2-. As used herein, the "luminal B subtype" refers to breast cancers that typically display lower expression of the ER+ signature and is further defined in Sorlie et al., 2001 and Sorlie et al., 2003. Immunohistochemically, they are ER+ and/or PR+ and can be HER2+ (in 30-50% of cases).

As used herein, the "basal subtype" refers to a subtype of breast cancer that is typically negative immunohistochemically for the estrogen receptor, the progesterone receptor and the Her2 receptor (ER-/PR-/HER2-) e.g. triple negative, and/or is molecularly characterized as basal based on a transcriptional expression signature as described in Perou et al., 2000; Sorlie et al., 2003, Sorlie et al., 2001 . For example, Figure 1 shows data based on both molecular and immunohistochemical subtyping. Although a large percentage of basal breast cancers are triple negative this is not absolute. The basal subtype can also be determined using the following five markers for the indicated staining pattern: ER-/PR-/HER2-/EGFR+/CK5+ or 6+. The triple negative and basal subtypes are often used interchangeably.

As used herein, the "HER2-positive subtype" (HER2+) refers to breast cancers that are positive for the overexpression of the HER2 protein/receptor.

The term "expression level" of a gene as used herein refers to the measurable quantity of gene product produced by the gene in a sample of the subject, wherein the gene product can be a transcriptional product or a translational product. Accordingly, the expression level can pertain to a nucleic acid gene product such as mRNA or cDNA or a polypeptide gene product. The expression level is derived from a subject's sample and/or a reference sample or samples, and can for example be detected de novo or correspond to a previous determination. The expression level can be determined or measured, for example, using microarray methods, PCR methods (such as qPCR), and/or antibody based methods, as is known to a person of skill in the art.

As used herein, the term "control gene expression level" refers to the gene expression level of a gene, or a combination of genes, whose expression is not modulated in tumor cells with respect to non-tumor cells. Such genes are useful in the normalization of the individualized gene expression level to allow a comparison with the data of the set of CUX1 expression profiles.

The term "altered level" as used herein refers to a difference in a level, or quantity, of a gene product (e.g. mRNA, cDNA or protein) in a sample that is measurable, compared to a control and/or reference sample. The term can also refer to an increase (e.g. overexpression) or decrease (e.g. underexpression) in the measurable expression, level of a given gene marker in a sample as compared with the measurable expression, level of a gene marker in a population of samples. For example, an expression level is altered if the ratio or fold change of the level in a sample as compared with a control or reference is greater than or less than 1 .0 and/or if the fold change or ratio of the level in the sample compared to a reference sample or samples is greater than or less than 1 .0. For example, a ratio of greater than 1 , 1 .2, 1 .5, 1 .7, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 15, 20 or more, or a ratio less than 1 , 0.8, 0.6, 0.4, 0.2, 0.1 , 0.05, 0.001 or less or for example, 20%, 50%, 70%, 100%, 200%, 400%, 900% or more, compared to a reference sample or samples or 20%, 40%, 60%, 80%, 90%, 100%, 200%, 300%, 400%, 900% or less compared to a reference sample or samples. For example, a change of 2 fold refers to a 100% increase or twice as much and a 0.5 fold change refers to a 100% decrease or half as much.

The term "overexpression" or "increased expression" as used herein means a polypeptide or nucleic acid gene expression product that is transcribed or translated at a detectably increased level, in comparison to a reference sample or reference profile derived from, for example in a sample comprising tumor cells compared to a reference sample or profile or samples or profiles associated with a particular outcome. The term includes overexpression due to transcription, post-transcriptional processing, translation, post-translational processing, and/or protein and/or RNA stability. Overexpression can be for example at least a 1 .2, 1 .5, 1 .7, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 15, 20 or more fold change compared to the expression of the corresponding gene of the reference profile or at least a 20%, 50%, 70%, 100%, 200%, 400%, 900%, or more increase, compared to a reference sample. In terms of a profile "overexpression" means a polypeptide or nucleic acid gene expression product that is transcribed or translated at a detectably increased level for each gene or a subset of genes e.g at least 80%, at least 85%, at least 90% of a gene set, for example 80% of genes of a CUX1 signature gene set. For example, as the expression and detection of gene expression can include noise, it would not be expected that each patient would have 100% of the signature. Accordingly, increases in for example at least 80% of the genes in the gene set would be expected to be predictive. The term "underexpressed" and/or "decreased expression" as used herein means a polypeptide or nucleic acid gene expression product that is transcribed or translated at a detectably decreased level, in comparison to a reference sample or sample, for example in a sample comprising tumor cells compared to a reference sample or samples associated with a particular prognosis. The term includes underexpression due to transcription, post-transcriptional processing, translation, post-translational processing, and/or protein and/or RNA stability. Underexpression can be 20%, 50%, 70%, 100%, 200%, 400%, 900% or more decreased, compared to a reference sample.

An individualized gene expression level is considered to be "similar" to a CUX1 -high or CUXI -low expression profile, when the individualized gene expression level has common characteristics with the reference expression profile. Such characteristics can include, for example, an altered level in the expression of a gene or a combination of genes from the CUX1 signature gene set, a signal from the expression level, the identity of the genes whose expression are being altered, etc. An individualized gene expression level has an increased likeliness to only one of the two CUX-high/low reference expression profile. In other words, the level of "similarity" between the individualized gene expression profile is higher for one of the CUX1 -high/low expression profiles.

By opposition, an individualized gene expression level to be "dissimilar" to a CUX1 -high or CUXI -low reference expression profile, when the individualized gene expression level lack common characteristics with one of the reference expression profile. An individualized gene expression level has e decreased likeliness to only one of the two CUX1 -high/CUX1 -low reference expression profile. In other words, the level of "dissimilarity" between the individualized gene expression profile is higher for one CUX1 -high/low reference expression profile than for the other.

The term "long rank test" refers to a hypothesis test to compare the survival distributions of two samples.

The term "Cox proportional hazards regression" refers to a statistical method of analyzing the effect of several risk factors on survival. The probability of the endpoint (death, or any other event of interest, e.g. recurrence of disease) is called the hazard. When the Hazard Ratio score is negative, higher expression correlates with longer survival, whereas a positive score indicates that higher expression correlates with shorter survival. The term "hierarchical clustering" refers to a method of cluster analysis which seeks to build a hierarchy of clusters.

The term "Euclidean distance" is calculated for example using the formula cl(p, q) = v⁷ (pi - q_x)² + (j¾ - q₂)² H h {p_n - q»)^'2 = ∑(Pi - ft)²- =l The "Ward's algorithm" refers to a commonly used procedure for forming hierarchical groups of mutually exclusive subsets. For example see, Joe H. Ward (1963). Hierarchical Grouping to optimize an objective function. Journal of American Statistical Association, 58(301 ), 236-244.

As used herein "sample" refers to any patient sample, including but not limited to a fluid, cell or tissue sample that comprises tumor associated cells, which can be assayed for gene expression levels, particularly genes differentially expressed in patients having a good prognosis and a poor prognosis. The sample includes for example bulk tumor, isolated stromal cells, a biopsy, a resected tumor sample, an aspirate of a tumor cell or a cell sample. The sample can be used fresh, can be fixed and/or paraffin-embedded and even frozen before the gene expression level is measured.

The term "subject" also referred to as "patient" as used herein refers to any member of the animal kingdom, preferably a human being.

The term "hybridize" as used herein refers to the sequence specific non-covalent binding interaction with a complementary nucleic acid. Appropriate stringency conditions which promote hybridization are known to those skilled in the art, or can be found in Current Protocols in Molecular Biology, John Wiley & Sons, N.Y. (1989), 6.3.1 6.3.6. For example, 6.0 x sodium chloride/sodium citrate (SSC) at about 45°C, followed by a wash of 2.0 x SSC at 50°C may be employed. With respect to a chip array, appropriate stringency conditions are known in the art.

The term "stringent hybridization conditions" refers to conditions under which a probe will hybridize to its target subsequence, typically in a complex mixture of nucleic acids, but to no other sequences or only to sequences with greater than 95%, 96%, 97%, 98%, or 99% sequence identity. Stringent conditions are for example sequence- dependent and will be different in different circumstances. Longer sequences can require higher temperatures. An extensive guide to the hybridization of nucleic acids is found in Tijssen, Techniques in Biochemistry and Molecular Biology-Hybridization with Nucleic Probes, "Overview of principles of hybridization and the strategy of nucleic acid assays" (1993). Generally, stringent conditions are selected to be about 5-10°C lower than the thermal melting point (Tm) for the specific sequence at a defined ionic strength pH. The Tm is the temperature (under defined ionic strength, pH, and nucleic concentration) at which 50% of the probes complementary to the target hybridize to the target sequence at equilibrium (as the target sequences are present in excess, at Tm, 50% of the probes are occupied at equilibrium). Stringent conditions may also be achieved with the addition of destabilizing agents such as formamide. For selective or specific hybridization, a positive signal is at least two times background, preferably 10 times background hybridization. Exemplary stringent hybridization conditions can be as following: 50% formamide, 5X SSC, and 1 % SDS, incubating at 42°C, or, 5X SSC, 1 % SDS, incubating at 65°C, with wash in 0.2X SSC, and 0.1 % SDS at 65°C.

Nucleic acids that do not hybridize to each other under stringent conditions can be considered substantially identical if the polypeptides which they encode are substantially identical, e.g. 95%, 95%, 97%, 98% or 99% identical. This occurs, for example, when a copy of a nucleic acid is created using the maximum codon degeneracy permitted by the genetic code. In such cases, the nucleic acids typically hybridize under moderately stringent hybridization conditions.

The term "microarray" as used herein refers to an ordered set of probes fixed to a solid surface that permits analysis such as gene analysis of a plurality of genes. A DNA microarray refers to an ordered set of DNA fragments fixed to the solid surface. For example, the microarray can be a gene chip. Methods of detecting gene expression and determining gene expression levels using arrays are well known in the art. Such methods are optionally automated.

The term "isolated nucleic acid sequence" as used herein refers to a nucleic acid substantially free of cellular material or culture medium when produced by recombinant DNA techniques, or chemical precursors, or other chemicals when chemically synthesized. The term "nucleic acid" is intended to include DNA and RNA and can be either double stranded or single stranded. The term nucleic acid is used interchangeably with gene, cDNA, mRNA, oligonucleotide and polynucleotide according to context. The term "isolated polypeptide" or "isolated protein" used interchangeably as used herein refers to a polymer of amino acid residues.

The term "sequence identity" as used herein refers to the percentage of sequence identity between two or more polypeptide sequences or two or more nucleic acid sequences that have identity or a percent identity for example about 70% identity, 80% identity, 90% identity, 95% identity, 98% identity, 99% identity or higher identity or a specified region. To determine the percent identity of two or more amino acid sequences or of two or more nucleic acid sequences, the sequences are aligned for optimal comparison purposes (e.g., gaps can be introduced in the sequence of a first amino acid or nucleic acid sequence for optimal alignment with a second amino acid or nucleic acid sequence). The amino acid residues or nucleotides at corresponding amino acid positions or nucleotide positions are then compared. When a position in the first sequence is occupied by the same amino acid residue or nucleotide as the corresponding position in the second sequence, then the molecules are identical at that position. The percent identity between the two sequences is a function of the number of identical positions shared by the sequences (i.e., % identity = number of identical overlapping positions/total number of positions.times.100%). In one embodiment, the two sequences are the same length. The determination of percent identity between two sequences can also be accomplished using a mathematical algorithm. A preferred, non- limiting example of a mathematical algorithm utilized for the comparison of two sequences is the algorithm of Karlin and Altschul, 1990, Proc. Natl. Acad. Sci. U.S.A. 87:2264-2268, modified as in Karlin and Altschul, 1993, Proc. Natl. Acad. Sci. U.S.A. 90:5873-5877. Such an algorithm is incorporated into the N BLAST and X BLAST programs of Altschul et al., 1990, J. Mol. Biol. 215:403. BLAST nucleotide searches can be performed with the NBLAST nucleotide program parameters set, e.g., for score=100, wordlength=12 to obtain nucleotide sequences homologous to a nucleic acid molecules of the present application. BLAST protein searches can be performed with the XBLAST program parameters set, e.g., to score-50, wordlength=3 to obtain amino acid sequences homologous to a protein molecule of the present invention. To obtain gapped alignments for comparison purposes, Gapped BLAST can be utilized as described in Altschul et al., 1997, Nucleic Acids Res. 25:3389-3402. Alternatively, PSI- BLAST can be used to perform an iterated search which detects distant relationships between molecules (Id.). When utilizing BLAST, Gapped BLAST, and PSI-Blast programs, the default parameters of the respective programs (e.g., of XBLAST and NBLAST) can be used (see, e.g., the NCBI website). The percent identity between two sequences can be determined using techniques similar to those described above, with or without allowing gaps. In calculating percent identity, typically only exact matches are counted.

The term "analyte specific reagent" or "ASR" refers to any molecule including any chemical, nucleic acid sequence, polypeptide (e.g. receptor protein) or composite molecule and/or any composition that permits quantitative assessment of the analyte level. For example, the ASR can be for example a nucleic acid probe primer set, comprising a detectable label or aptamer that binds to, reacts with and/or responds to a gene listed in Table 1 . A gene specific ASR is herein referred to by reference to the gene, for example a "AURKB ASR" refers to an ASR such as a probe that specifically binds to a AURKB gene product in a manner to permit quantification of the AURKB gene product (e.g. mRNA corresponding of cDNA).

The term "specifically binds" as used herein refers to a binding reaction that is derminative of the presence of the analyte (e.g. polypeptide or nucleic acid) often in a heterogeneous population of macromolecules. For example, when the ASR is a probe, specifically binds refers to the specified probe under hybridization conditions binds to a particular gene sequence at least 1 .5, at least 2 or at least 3 times background.

The term "probe" as used herein refers to a nucleic acid sequence that comprises a sequence of nucleotides that will hybridize specifically to a target nucleic acid sequence e.g. a gene listed in Table 1 . Some embodiments of some probes are presented in Table 3. For example the probe comprises at least 10 or more bases or nucleotides that are complementary and hybridize contiguous bases and/or nucleotides in the target nucleic acid sequence. The length of probe depends on the hybridization conditions and the sequences of the probe and nucleic acid target sequence and can for example be 10-20, 21 -70, 71 -100, 101 -500 or more bases or nucleotides in length. The probes can optionally be fixed to a solid support such as an array chip or a microarray chip.

The term "primer" or "oligonucleotide primer" as used herein refers to a nucleic acid sequence, whether occurring naturally as in a purified restriction digest or produced synthetically, which is capable of acting as a point of synthesis of when placed under conditions in which synthesis of a primer extension product, which is complementary to a nucleic acid strand is induced (e.g. in the presence of nucleotides and an inducing agent such as DNA polymerase and at a suitable temperature and pH). The primer must be sufficiently long to prime the synthesis of the desired extension product in the presence of the inducing agent. The exact length of the primer will depend upon factors, including temperature, sequences of the primer and the methods used. A primer typically contains 15-25 or more nucleotides, although it can contain less. The factors involved in determining the appropriate length of primer are readily known to one of ordinary skill in the art. Some of the primers specific for the genes of Table 1 are presented in Table 3.

The term "antibody" as used herein is intended to include monoclonal antibodies, polyclonal antibodies, and chimeric antibodies. The antibody may be from recombinant sources and/or produced in transgenic animals. The term "antibody fragment" as used herein is intended to include Fab, Fab', F(ab')₂, scFv, dsFv, ds-scFv, dimers, minibodies, diabodies, and multimers thereof and bispecific antibody fragments. Antibodies can be fragmented using conventional techniques. For example, F(ab')₂ fragments can be generated by treating the antibody with pepsin. The resulting F(ab')₂ fragment can be treated to reduce disulfide bridges to produce Fab' fragments. Papain digestion can lead to the formation of Fab fragments. Fab, Fab' and F(ab')₂, scFv, dsFv, ds-scFv, dimers, minibodies, diabodies, bispecific antibody fragments and other fragments can also be synthesized by recombinant techniques.

To produce human monoclonal antibodies, antibody producing cells (lymphocytes) can be harvested from a human having cancer and fused with myeloma cells by standard somatic cell fusion procedures thus immortalizing these cells and yielding hybridoma cells. Such techniques are well known in the art, (e.g. the hybridoma technique originally developed by Kohler and Milstein (Nature 256:495-497 (1975)) as well as other techniques such as the human B-cell hybridoma technique (Kozbor et al., Immunol. Today 4:72 (1983)), the EBV-hybridoma technique to produce human monoclonal antibodies (Cole et al., Methods Enzymol, 121 : 140-67 (1986)), and screening of combinatorial antibody libraries (Huse et al., Science 246: 1275 (1989)). Hybridoma cells can be screened immunochemically for production of antibodies specifically reactive with cancer cells and the monoclonal antibodies can be isolated.

Specific antibodies, or antibody fragments, reactive against particular target polypeptide gene product antigens (e.g. Table 1 polypeptide), can also be generated by screening expression libraries encoding immunoglobulin genes, or portions thereof, expressed in bacteria with cell surface components. For example, complete Fab fragments, VH regions and FV regions can be expressed in bacteria using phage expression libraries (See for example Ward et al., Nature 341:544-546 (1989); Huse et al. , Science 246: 1275-1281 (1989); and McCafferty et al., Nature 348:552-554 (1990)).A "detectable label" as used herein means an agent or composition detectable by spectroscopic, photochemical, biochemical, immunochemical, chemical, or other physical means. For example, useful labels include ³²P, fluorescent dyes, electron-dense reagents, enzymes (e.g., as commonly used in an ELISA), biotin, digoxigenin, or haptens and proteins which can be made detectable, e.g., by incorporating a radiolabel into the peptide or used to detect antibodies specifically reactive with the peptide.

The phrase "therapy or treatment" as used herein, refers to an approach aimed at obtaining beneficial or desired results, including clinical results and includes medical procedures and applications including for example chemotherapy, pharmaceutical interventions, surgery, radiotherapy and naturopathic interventions as well as test treatments. The phrase "breast cancer therapy or treatment" refers to any approach including for example surgery, chemotherapy, hormone therapy, immunotherapy, preventive interventions, prophylactic interventions and test treatments aimed at alleviating or ameliorating one or more symptoms, diminishing the extent of, stabilizing, preventing the spread of, delaying or slowing the progression of, ameliorating or palliating and/or inducing remission of breast cancer and/or any associated complications thereof.

The term a "therapeutically effective amount", "effective amount" or a "sufficient amount" of a compound of the present disclosure is a quantity sufficient to, when administered to a cell or a subject, including a mammal, for example a human, effect beneficial or desired results, including clinical results, and, as such, an "effective amount" or synonym thereto depends upon the context in which it is being applied. For example, in the context of breast cancer, therapeutically effective amounts are used to treat, modulate, attenuate, reverse, or affect breast cancer progression in a subject. For example, an "effective amount" is intended to mean that amount of a compound that is sufficient to treat, prevent or inhibit breast cancer or a disease associated with breast cancer. The amount of a given compound that will correspond to such an amount will vary depending upon various factors, such as the given drug or compound, the pharmaceutical formulation, the route of administration, the type of disease or disorder, the identity of the subject or host being treated, and the like, but can nevertheless be routinely determined by one skilled in the art. Also, as used herein, a "therapeutically effective amount" of a compound is an amount which prevents, inhibits, suppresses or reduces the progression of or symptoms associated with cancer (e.g., as determined by clinical symptoms in a subject as compared to a reference or comparison population. As defined herein, a therapeutically effective amount of a compound may be readily determined by one of ordinary skill by routine methods known in the art.

Moreover, a "treatment" or "prevention" regime of a subject with a therapeutically effective amount of the compound of the present disclosure may consist of a single administration, or alternatively comprise a series of applications. For example, the compound of the present disclosure may be administered at least once a week. However, in another embodiment, the compound may be administered to the patient from about one time per week to one or more, for example one to four, times daily for a given treatment. The length of the treatment period depends on a variety of factors, such as the severity of the disease, the age of the patient, the concentration and the activity of the compounds of the present disclosure, or a combination thereof. It will also be appreciated that the effective dosage of the compound used for the treatment or prophylaxis may increase or decrease over the course of a particular treatment or prophylaxis regime. Changes in dosage may result and become apparent by standard diagnostic assays known in the art. In some instances, chronic administration may be required.

The phrase "usefulness of a chemotherapy" refers to the heightened advantages of using chemotherapy in a individual that has been attributed a prognosis by the methods described herein. Individuals classified in a specific prognostic class will benefit from the chemotherapy because it will limit cancer progression and may even treat cancer. In these individuals, the chemotherapy is considered useful. Not all individuals in this specific prognostic class will benefit from chemotherapy, but those who will are all classified in this specific prognostic class. In opposition, all the individuals classifies in the other prognostic class will not benefit from the chemotherapy, e.g. it will not modify cancer progression or cancer recurrence. In these individuals, the chemotherapy lacks utility and is not considered useful.

As used herein "a user interface device" or "user interface" refers to a hardware component or system of components that allows an individual to interact with a computer e.g. input data, or other electronic information system, and includes without limitation command line interfaces and graphical user interfaces.

In understanding the scope of the present disclosure, the term "comprising" and its derivatives, as used herein, are intended to be open ended terms that specify the presence of the stated features, elements, components, groups, integers, and/or steps, but do not exclude the presence of other unstated features, elements, components, groups, integers and/or steps. The foregoing also applies to words having similar meanings such as the terms, "including", "having" and their derivatives. Finally, terms of degree such as "substantially", "about" and "approximately" as used herein mean a reasonable amount of deviation of the modified term such that the end result is not significantly changed. These terms of degree should be construed as including a deviation of at least ±5% of the modified term if this deviation would not negate the meaning of the word it modifies.

The definitions and embodiments described in particular sections are intended to be applicable to other embodiments herein described for which they are suitable as would be understood by a person skilled in the art.

II. Diagnosis and method for determining prognosis

The present application herewith shows that a specific CUX1 signature get set can be used to predict whether an individual is associated with a certain clinical outcome, such as good cancer prognosis or a poor cancer prognosis. The CUX1 signature set is used to determine if an individual is in the "low" expression group (associated with a good prognosis) or in the "high" expression group (associated with a poor prognosis). First, the gene expression level of a plurality of genes defined in the CUX1 signature gene set (e.g. listed in Table 1 ) is determined in an individual's biological sample containing tumor cells (or suspected of containing tumor cells). For enabling the comparison, this gene expression level is then normalized at least one control gene. Then, a parameter of the normalized expression level is compared to data from a set of CUX1 expression profiles having known clinical outcomes. The set of CUX1 expression profiles defines at least two prognostic classes (good prognosis or poor prognosis). Finally, a particular clinical outcome is associated to the subject submitted to this method based on the comparison. As shown herewith, individuals classified in the "low" expression group have been positively correlated with a good clinical outcome (e.g. good prognosis) and/or negatively correlated with a poor clinical outcome (e.g. poor prognosis). In addition, individuals classified in the "high" expression group have been positively correlated with a poor clinical outcome (e.g. poor prognosis) and/or negatively correlated with a good clinical outcome (e.g. good prognosis).

Since CUX1 plays an important role in various types of cancer, the CUX1 signature can be used for predicting the clinical outcome of various cancer types. It will be appreciated that the CUX1 signature gene set may not be involved in all types of cancers. However, the CUX1 signature gene set can be used in all cancer types where CUXI 's activity, expression and/or stability has been shown to be involved. In an embodiment, the CUX1 signature get set can be specifically used to predict a clinical outcome for breast cancer patients.

The methods described herewith are designed to be used after an initial cancer diagnostic has been made. In this embodiment, the method is particularly useful to predict the individual's clinical outcome, also referred to as a prognosis. Much like what has been described above, a biological sample (tissue or bodily fluid for example) is provided from the individual having been diagnosed with cancer and a gene expression measurement is made in a plurality of genes defined in the CUX1 signature gene set. The measurement is then compared to data from a set of CUX1 expression profiles to determine if the individual is in the "low" expression group or in the "high" expression group. Individuals classified in the "low" expression group are assumed to be positively correlated with a good clinical outcome (e.g. good prognosis) and/or negatively correlated with a poor clinical outcome (e.g. poor prognosis). In addition, individuals classified in the "high" expression group are assumed to be positively correlated with a poor clinical outcome (e.g. poor prognosis) and/or negatively correlated with a good clinical outcome (e.g. good prognosis). This method can be combined with other methods known in the art to determine cancer clinical outcome.

The methods described herein are not only useful in predicting "in general" the prognosis of an individual having been diagnosed with cancer, but can also be used to predict more specifically some clinical outcomes within a specific time period from initial diagnosis or the prognosis (2 years, 3 years, 4 years, 5 years, 10 years, 15 years, 20 years and even greater), such as cancer recurrence, cancer metastasis and survival. Individuals classified in the "low" expression group are more likely not to have cancer recurrence within the specified period, not to have metastasis within the specific period, to have a lower number of metastasis, have a lower number of metastatic sites, and/or to survive longer than individuals classified in the "high" expression group. Similarly, the individuals classified in the "high" expression group are more likely to have cancer recurrence within the specified period, to have metastasis within the specified period, to have a higher number of metastasis, to have a higher number of metastatic sites, and/or to die sooner than the individuals classified in the "low" expression group.

Once a clinical outcome, or a prognosis, is determined, the methods can be used to determine the most efficient course of treatment based on the prognosis. Since the present method indirectly determine the aggressiveness of the cancer in the individual, the information it provides can be useful in determining if a particular treatment would indeed be beneficial in an individual. For example, for individuals classified in the "low" expression group and associated with a good prognosis, more classical and maybe less aggressive treatments would be beneficial in stopping or limiting the progression of cancer, and ultimately, treating cancer. In such individuals, chemotherapy may not even be necessary (or even beneficial) and could even be excluded from the treatment regimen to avoid associated toxic effects. However, for individuals classified in the "high" expression group and associated with a poor prognosis, more aggressive therapeutic methods (such as chemotherapy) would be considered appropriate in stopping or limiting the progression of cancer, and ultimately, treating cancer. In such individuals, chemotherapy may be considered beneficial and should be included in the treatment regimen. It is to be understood that no all patients classified in the "high" expression group will necessarily benefit from the chemotherapy because some of them will be resistant to chemotherapy. However, it is assumed that those who will benefit from chemotherapy will nevertheless be in the "high" expression group.

The methods described herewith can also be used to determine if a therapeutic measure is beneficial for improving the prognosis of an individual that is being treated. In this embodiment, a first prognosis determination is made prior to treatment and indicated that the clinical outcome is poor. Then, the individual is being treated, preferably with a regimen that is suited with the first prognosis. During or after the treatment, a second prognosis is determined. If the second prognosis indicates that the individual is now in the "low" expression class, then it is assumed that the therapeutic measure was beneficial for improving the prognosis.

A further aspect of the application includes a method of identifying agents for use in the treatment of breast cancer. Clinical trials seek to test the efficacy of new therapeutics. The efficacy is often only determinable after many months of treatment. The methods disclosed herein are useful for monitoring the expression of genes associated with prognosis. Accordingly, changes in gene expression levels which are associated with a better prognosis are indicative the agent is a candidate as a chemotherapeutic.

In addition, results from these previous studies have demonstrated a link between a gene expression signature, the rate of distant recurrence at 10 years, and the benefit of chemothepapy for estrogen receptor-positive breast cancer patients (Paik et al., 2004;, Paik et al. 2006; Albain et al., 2010). It is expected that the methods presented herewith can be used to determine if a cancer will be resistant to chemotherapy and therefore provide valuable information for optimizing the treatment regimen.

The use of the CUX1 signature gene set also facilitate the stratification of patients according to clinical outcome with accuracy. Accordingly the methods described herein can be used for example to select or exclude subjects from a cancer clinical trial, such a chemotherapy-based clinical trials.

The methods described herewith first include a step of measuring (or assaying) a gene expression level (also referred to as an individualized gene expression level) of a plurality genes defined in the CUX1 signature gene set from a sample of the individual. The genes of the CUX1 signature gene set are presented in Table 1 and are associated with CUXI 's activity. The plurality of genes that are being assayed preferably comprises one gene associated with genetic instability and another one associated with cell proliferation. In an embodiment, at least twenty different genes from the CUX1 signature gene set are assayed. Gene expression level is preferably a measurement of the mRNA level associated with each gene that is being assayed, but other methods for determining gene expression level (e.g. protein level or activity) can also be used.

Once the individualized gene expression level has been measured it is compared with data from a set of CUX1 expression profiles. The set of CUX1 expression profiles is a compilation of gene expression level for the plurality of genes of the CUX1 signature gene set from a population of subjects having been diagnosed with case and/or from cancer cell lines (or a pre-determined value determined from the population of subjects or cell lines). A parameter for each different gene of the individualized gene expression level can be compared (individually or in combination) to data of the set of CUX1 expression profiles. The parameter of the individualized gene expression that is compared can be, for example, the normalized value of the expression level of each genes (individually or in combination), a signal from such normalized gene expression and/or a transformation by a mathematical algorithm (such as, for example, a probability model).

Various comparisons can be made between the individualized gene expression level the data of the set of CUX1 expression profiles. These comparisons are not mutually exclusive and can even be used in combination.

The data of the set of CUX1 expression profiles can be generated with bioinformatics/biostatistical methodology for integrating the expression levels for the CUX1 gene set. This integration preferably defines at least two prognostic classes each associated with a known clinical outcome. For example, the comparison can comprise applying a probability model to the normalized individualized gene expression level and generating an output. In such embodiment, the data derived from the set of CUX-1 expression profiles defines a scoring scheme for the at least two prognostic classes. As such, the comparison includes applying the output of the scoring scheme for the parameter of the normalized individualized gene expression level. For example, an outcome score between the value of 0 (poor prognosis) and 1 (good prognosis) is generated from the set of CUX1 expression profiles. The outcome score can be derived, for example, from methods from machine learning/statistical inference such as a naive bayes classifier. The naive bayes' classifier facilitates the classification of a subject sample, based on his individualized gene expression level, in such a manner as to predict if the subject belongs to the good or poor outcome group.

In another embodiment, it can be determined if the individualized gene expression level is higher than or lower than a threshold value derived from the set of CUX1 expression profiles. In order to do so, each of the CUX1 expression profiles in the set is attributed a value. The values within a single prognostic class are similar to one another and differ from the values determined in another prognostic class. In addition, each prognostic class is associated with a different clinical outcome. Further, CUX1 expression profiles classified within the same prognostic class have the same clinical outcome. The threshold value is then determined and corresponds to a value which can discriminate between the two prognostic classes. Then, the individualized gene expression level is also attributed a value and is compared to the threshold value to determine the prognostic class and, ultimately, the clinical outcome. For example, if the individualized gene expression level is scored as being higher than the threshold value, then the individual is considered to be classified in the "high" expression group and is associated with a poor prognosis and/or not associated with a good prognosis. If the individualized gene expression level is scored as being lower than this threshold value, then the individual is considered to be classified in the "low" expression group and is associated with a good prognosis and/or not associated with a poor prognosis.

Many biostatistical methods are known to those skilled in the art to determine this threshold value. For example, it can be determined by calculating the Euclidean distance between the CUX1 expression profiles of the set and the average expression profiles (or centroids) representing the at least two prognostic classes. Then, the individualized gene expression level is assigned to the class that minimizes the Euclidean distance.

It can also be determined if the individualized gene expression level has a certain degree of similarity and/or dissimilarity to data from a subset the CUX1 expression profiles associated with a specific prognostic class. In this embodiment, characteristics shared by the CUX1 expression profiles within a single prognostic class but that are different from the characteristics of another subset of CUX1 expression profiles of another prognostic class are identified. Then, the characteristics of the individualized gene expression profile are compared to the characteristics of the CUX1 expression profiles in the different prognostic classes. The individualized gene expression profile is associated to the prognostic class having the most similarity or the least dissimilarity. For example, a subset of profiles referred to as a "CUX1 -high expression profiles" is derived from a relatively homogeneous and representative population of cancer subjects having a poor prognosis and the characteristics of such "high" expression profiles are identified. Another subset of profiles referred to as a "CUX1 -low expression profiles" is derived from an relatively homogeneous and representative population of cancer subjects having a good prognosis and the characteristics of such "low" expression profiles are identified. Then, the level of similarity or dissimilarity of the individualized gene expression level with respect to the "high" and "low" expression profiles is determined. The subject is considered to have a poor prognosis if its gene expression profile is more similar to the "high" expression profiles than to the "low" expression profiles and/or if it is less similar to the "low" expression profiles than to the "high" expression profiles. Alternatively, the subject is considered to have a good prognosis if its gene expression profile is more similar to the "low" expression profiles than to the "high" expression profiles and/or if it is less similar to the "high" expression profiles than to the "low" expression profiles.

A number of algorithms can be used to assess similarity. For example, similarity can be assessed by determining the Euclidean distance of an individualized gene expression level to a class centroid. Wards algorithm can be used for forming hierarchical groups of mutually exclusive subsets of samples.

It can also be correlated if the individualized gene expression level by normalizing this individualized gene expression level with respect to control gene expression level and correlating it with a clinical outcome. For example, an increased normalized individualized gene expression level is positively correlated with a poor clinical outcome and/or negatively correlated with a good clinical outcome. Similarly, a decreased normalized individualized gene expression level is positively correlated with a good clinical outcome and/or negatively correlated with a poor clinical outcome.

The prognostic class to which the individualized gene expression level belongs can also be determined by attributing a value to each CUX1 expression profiles of the set and classifying them. It is understood that CUX1 expression profiles associated with a first prognostic class will segregate to one spectrum of the classification, whereas the CUX1 expression profiles associated with a second prognostic class will segregate at the other end of the spectrum. Then a value is similarly attributed to the individualized gene expression level and is compared to the classified values for each CUX1 expression profiles that has been classified. It is determined to which CUX1 expression profile (or combination thereof) within the set the individualized gene expression level has the most similarity and/or the least dissimilarity based on the value calculated. Once this match has been made, then the clinical outcome associated with selected CUX1 expression profile is associated the individualized gene expression level and, ultimately, to the subject that is being tested.

The CUX1 signature gene set provides a tool that can predict a clinical outcome in a cancer patient. In its entirety, the CUX1 signature gene set comprises the following genes: AURKB, BUB1 , BUBR1 (BUB1 B), BUB3, CAMK2D, CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CENPA, CENPE, CDKNA1 , COG7, CTSL1 , KIF2C, KI FC1 , MAD1 L1 , MAD2L1 , MCM3, MCM7, MFAP1 , NEK2, NUDCD1 , ORCL1 , POLA, POLA2, RIN2, KNTCI (ROD), SF3B3, TTK (MPS1 ), ZW10 and ZWILCH. However, the CUX1 signature gene set does not have to be used in its entirety to be useful in the methods described herewith. In an embodiment, the CUX1 signature gene set or plurality of genes that is being assayed comprises at least 5, 6, 7, 8, 9, 10, 1 1 , 12, 13, 14, 15, 16, 17 , 18 or 19 genes. Preferably, the CUX1 signature gene set comprises at least 20, 21 , 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 or 32 genes. In some embodiment, a 29-gene CUX1 signature gene set has been provided useful to predict a clinical outcome. This 29-gene CUX1 signature gene set comprises the following genes: AURKB, BUB1 , BUBR1 (BUB1 B), BUB3, CAMK2D, CCNA2, CDC25A, CDC45L, CDC7, CENPA, CENPE, COG7, CTSL1 , KIF2C, KIFC1 , MAD1 L1 , MAD2L1 , MCM3, MCM7, MFAP1 , NEK2, NUDCD1 , POLA, POLA2, RIN2, SF3B3, TTK (MPS1 ), ZW10 and ZWILCH.

The fold change between an individualized gene expression level and a CUX1 reference profile can vary within individuals belonging to the same prognostic class ("low" or "high"). For example, the gene expression level can be 1 .2, 1 .5, 1 .7, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 15, 20 or more fold change compared to the CUX1 reference profile or at least a 20%, 50%, 70%, 90%, 95%, 100%, 200%, 400%, 900%, or more increased or decreased, compared to the CUX1 reference profile. On average the expression level of each gene in a subject classified in the "low" group is decreased for example, by about 93% compared to the expression of the gene in a subject falling within "high" group.

Also not all genes in the "high" or "low" expression group are necessarily increased or decreased in all subjects. As such a subjects can be classified in the "high" expression group has an expression profile similar to a CUX1 -high expression profile, such a subject would be classified as having a poor prognosis. Accordingly in an embodiment, 70%, 80%, 85%, 90%, 95% of the genes in the CUX1 signature gene set exhibit increased expression level compared to a CUX1 reference profile. For example, the expression levels of a CUX1 signature gene set in a subject is compared to a CUX1 - high expression reference profile. If the gene level profile is similar to a CUX1 -high expression reference profile, for example, if 16 out of 20 genes assessed (e.g. 80%) have a similar expression level to a CUX1 -high profile, the subject is classified in the "high" group and is associated with a poor prognosis. Conversely, in an embodiment, 70%, 80%, 85%, 90%, 95% of the genes in the CUX1 signature gene set exhibit decreased expression level compared to a CUX1 -low expression reference profile. If the gene level profile is similar to a CUX1 -low expression reference profile, for example, if 16 out of 20 genes assessed (e.g. 80%) have a similar expression level to a CUX1 -low expression reference profile, the subject is classified in the "low" group and are associated with a good prognosis.

It is demonstrated herein that the CUX1 signature get set described herein is useful in different breast cancer subtypes. In an embodiment, the subject has a ER-positive, ER- negative, node positive or node negative, high grade or low grade, luminal A, luminal B, basal, HER2 positive, or any combination of thereof, breast cancer. In an embodiment, the breast cancer subtype is determined prior to CUX1 signature get set classification. For example, the tumor status of ER-positive is determined for example immunohistochemically. In another embodiment, the tumor status of HER2, PR and/or ER is determined. In another embodiment, it is the association of both the CUX1 signature get set determination and the subtype that are used to determine the prognosis and, possibly, the best treatment regimen.

The sample that is being processed to measure gene expression should comprise or consist of cancer cells or of cells suspected to be cancer cells. The sample, in an embodiment, comprises or is a tumor biopsy or a surgical resection. In an embodiment, the sample comprises bulk tumor, including tumor associated stromal cells. In an embodiment, the sample comprises fresh tissue, frozen tissue sample, a cell sample, or a paraffin embedded sample. In an embodiment, the sample is submerged in a RNA preservation solution, for example to allow for storage. In an embodiment, the sample is submerged in Trizol™. Frozen tissue is for example, maintained in liquid nitrogen until RNA can be processed. For RNA preparation, typically a tissue is homogenized immediately in 5M guanidine isothiocyanate (GIT) and purified using commercially- available RNA purification columns (e.g. Qiagen, Invitrogen) according to manufacturer's instructions. RNA is stored for example, at -80°C until use.

The sample, in an embodiment, is processed, for example, to obtain an isolated RNA fraction and/or an isolated polypeptide fraction. For example, the sample can be treated with a lysis solution e.g. to lyse the cells, to allow a detection agent access to the RNA species. The sample can also or alternatively be processed using a RNA isolation kit such as RNeasy™ to isolate RNA or a fraction thereof (e.g. mRNA). The sample is in an embodiment, treated with a RNAse inhibitor to prevent RNA degradation.

Wherein the gene expression level being determined is a nucleic acid, the gene expression levels can be determined using a number of methods for example hybridizing to a probe or a microarray chip (e.g. an oligonucleotide array) or using primers and PCR amplification based methods, optionally, quantitative PCR, multiplex PCR, RT-PCR and a combination thereof. These methods are known in the art and a person skilled in the art would be familiar with the necessary normalizations necessary for each technique. For example, the expression measurements generated using multiplex PCR should be normalized by comparing the expression of the genes being measure to so called "housekeeping" genes, the expression of which should be constant over all samples, thus providing a baseline expression to compare against or other control genes whose expression are known to be modulated with cancer.

Accordingly, in an embodiment, determining the expression profile comprises contacting a sample comprising RNA or cDNA corresponding to the RNA (e.g. a processed sample from the subject) with an analyte specific reagent (ASR), for example an ASR that specifically binds and/or amplifies a nucleic acid product of a gene listed in Table 1 such as BUB1 , for each gene of the plurality of genes (e.g. for each gene of the CUX1 signature gene set) and determining the expression level for each gene. For example, where the ASR specifically binds a nucleic acid expression product, a complex is formed between the ASR and target expression product. The expression level of each gene is thus determined by measuring complexes formed to determine the expression level of the gene. Also for example, where the ASR specifically and quantitatively amplifies a nucleic acid expression product, measuring the amount of the amplification product determines the level of gene expression. Thus contacting for example with a BUB1 ASR, and measuring the complexes formed or the amplification product amounts is used to determine the expression level of the marker (i.e. BUB1 ) in the sample. Similarly contacting with a AURKB ASR is used to determine the expression level of the AURKB marker. In an embodiment, the step of correlating the gene expression levels and/or classifying the subject comprises determining whether or not the expression profile, for example whether the RNA representing 20 or more of the genes listed in Table 1 , is altered in the sample when compared to corresponding RNA expression levels representing each marker nucleic acid of a comparison population of subjects, for example a CUX1 signature-low class or a CUX1 signature-high class.

In an embodiment, the ASR is a nucleic acid molecule (e.g. an oligonucleotide or probe). In an embodiment, the nucleic acid molecule comprises a quantifier. In another embodiment, the ASR comprises a primer set that amplifies a Table 1 nucleic acid gene product (e.g. RNA and/or corresponding cDNA). In another embodiment, the nucleic acid molecule is comprised in an array.

The expression level can also be the polypeptide expression level. A person skilled in the art will appreciate that a number of methods can be used to determine the amount of a polypeptide product of a gene described herein, including mass spectrometry, immunoassays such as Western blots, ELISA, and immunoprecipitation followed by SDS-PAGE, as well as immunocytochemistry or immunohistochemistry.

Accordingly, in an embodiment, determining the expression profile comprises contacting a sample comprising polypeptide (e.g. a processed sample from the subject) with an analyte specific reagent (ASR), for example an ASR that specifically binds a polypeptide product of a gene listed in Table 1 such as BUB1 , for each gene of the plurality of genes (e.g. for each gene of the CUX1 signature gene set) and determining the expression level for each gene. For example, where the ASR specifically binds a polypeptide expression product, a complex is formed between the ASR and target product. The expression level of each gene is thus determined by measuring complexes formed to determine the expression level of the gene. Thus contacting for example with a BUB1 ASR, and measuring the complexes formed is used to determine the expression level of the marker (i.e. BUB1 ) in the sample. In an embodiment, the step of correlating the gene expression levels and/or classifying the subject comprises determining whether or not the expression profile, for example whether the polypeptide level representing 5 or more (preferably at least 20) of the genes listed in Table 1 , is altered in the sample when compared to corresponding polypeptide levels representing each marker polypeptide of a comparison population of subjects, for example a CUX1 signature-low class or a CUX1 signature-high class. In an embodiment, the ASR is an antibody. In an embodiment, the antibody is a monoclonal antibody. In a further embodiment, the antibody is comprised in an array.

An aspect provides a composition comprising a set of probes or primers for determining expression of a plurality of genes of the CUX1 signature gene set. In an embodiment, the plurality of genes comprises and/or consists of at least 20 genes. In yet another embodiment, the set of probed or primers are a combination of the oligonucleotides presented in Table 4.

Another aspect of the disclosure includes an array comprising for each gene in a plurality of genes, the plurality of genes being at least 20 of the genes listed in Table 1 , one or more polynucleotide probes complementary and hybridizable to a coding sequence in the gene. The array can be a microarray, a DNA array and/or a tissue array. In an embodiment, the array is a multi-plex, qRT-PCR-based array.

Another aspect includes a kit for determining prognosis in a subject having breast cancer. The kit comprises an analyte specific reagent for measuring the gene expression level of a plurality of genes of the CUX1 signature gene set and a sample collector. In an embodiment, such kit can be for performing the methods described herein and can comprises instructions to this effect. The specimen collector can comprise a sterile vial or tube suitable for receiving a biopsy or other sample. In an embodiment, the specimen collector comprises RNA preservation solution. In another embodiment, RNA preservation solution is added subsequent to the reception of sample. In an embodiment the RNA preservation solution comprises one or more inhibitors of RNAse. In another embodiment, the RNA preservation solution comprises Trizol™. The analyte specific reagent can comprise at least one primer listed in Table 4 or a combination of primers listed in Table 4. The ASR can also be a micro-array. In an embodiment, the antibody or probe is labeled. The label is preferably capable of producing, either directly or indirectly, a detectable signal. For example, the label may be radio-opaque or a radioisotope, such as ³H, ¹⁴C, ³²P, ³⁵S, ¹²³l, ¹²⁵l, ¹³¹ l; a fluorescent (fluorophore) or chemiluminescent (chromophore) compound, such as fluorescein isothiocyanate, rhodamine or luciferin; an enzyme, such as alkaline phosphatase, beta- galactosidase or horseradish peroxidase; an imaging agent; or a metal ion.

In another embodiment, the detectable signal is detectable indirectly. A person skilled in the art will appreciate that a number of methods can be used to determine the amount of a polypeptide product of a gene described herein, including immunoassays such as Western blots, ELISA, and immunoprecipitation followed by SDS-PAGE, as well as immunocytochemistry or immunohistochemistry. The kit can accordingly in certain embodiments comprise reagents for one or more of these methods, for example molecular weight markers, standards or analyte controls.

II. Computer implementation and associated products

The methods described herein can be adapted for computer implementation. Broadly, the method comprises classifying, on a computer, the subject as having a good prognosis or a poor prognosis based on a gene expression profile comprising measurements of expression levels of a plurality of genes (preferably at least 20) of the CUX1 signature gene set. The method can also comprises displaying or outputting to a user interface device, a computer readable storage medium, or a local or remote computer system; the classification produced by the classifying step. In another embodiment, the method comprises displaying or outputting a result of one of the steps to a user interface device, a computer readable storage medium, a monitor, or a computer that is part of a network.

Another aspect of the disclosure includes a computer product for implementing the methods described herein e.g. for predicting prognosis, selecting patients for a clinical trial, or selecting therapy. Accordingly in an embodiment, the computer product is a non- transitory computer readable storage medium with an executable program stored thereon, wherein the program is for predicting outcome in a subject having breast cancer, and wherein the program instructs a microprocessor to perform the steps of any of the methods described herein.

The present application also provide a system for performing the methods described herein. The system can comprise a reaction vessel for combining a sample with at least one ASR to measure the individualized gene expression level in a plurality of genes in a CUX1 signature gene set. The system also comprises a processor in a computer system, a memory accessible by the computer as well as at least one application coupled to the computer. The at least one application is configured for receiving a measure of the individualized gene expression level, compare this measure and determine the prognosis.

The present application further provides a software product embodied on a computer readable medium and comprising instructions for determining a prognosis. The software product comprises a receiving module (for receiving the measurement of the individualized gene expression level), a normalizing module (for normalizing the individualized gene expression level to at least one control gene expression level), a comparison module (for comparing a parameter the individualized gene expression profile to data from a set of CUX1 expression profiles defining at least two prognostic classes each associated with a known and different clinical outcome) and a characterization module (for associating a known clinical outcome with the individualized gene expression level).

The comparison module and/or the characterization module may comprise a processor and a memory card to perform an application. The processor may access the memory to retrieve data. The processor may be any device that can perform operations on data. Examples are a central processing unit (CPU), a front-end processor, a microprocessor, a graphics processing unit (PPU/VPU), a physics processing unit (PPU), a digital signal processor and a network processor. The application is coupled to the processor and configured to determine the similarity/dissimilarity of the individualized gene expression level to a CUX1 reference profile. An output of this comparison may be transmitted to a display device. The memory, accessible by the processor, receives and stores data, such as measured parameters of the individualized gene expression level or any other information generated or used. The memory may be a main memory (such as a high speed Random Access Memory or RAM) or an auxiliary storage unit (such as a hard disk, a floppy disk or a magnetic tape drive). The memory may be any other type of memory (such as a Read-Only Memory or ROM) or optical storage media (such as a videodisc or a compact disc).

In an embodiment, an application found in the computer system is used in a characterization module. A measuring module extracts/receives information from the reaction vessel with respect to the individualized gene expression level. The receiving module is coupled to a comparison module which receives the value(s) of the individualized gene expression level and determines if this measurement is in the "low" or "high" class and, ultimately provides a prognosis. The comparison module can be coupled to a characterization module.

In a further embodiment, the receiving module, comparison module and characterization module are organized into a single discrete system. In another embodiment, each module is organized into different discrete system. In still a further embodiment, at least two modules are organized into a single discrete system.

A further aspect of the present application includes a computer system comprising a database including records comprising reference expression profiles associated with clinical outcomes, each reference profile comprising the expression levels of a plurality of genes listed in the CUX1 signature gene set; a user interface capable of receiving and/or inputting a selection of gene expression levels of a plurality of genes, the plurality comprising at least 20 genes of the CUX1 signature gene set, for use in comparing to the gene reference expression profiles in the database; and an output that displays a prediction of clinical outcome according to the expression levels of the plurality of genes. III. Therapeutic methods

In another aspect, the present application includes a method of selecting or optimizing a breast cancer treatment. First, it is determined if the individual is in the "high" or "low" prognostic class based on the methods described herewith. Then a therapeutic regimen is indicated (or even administered) based on the respective class. In an embodiment, the individualized expression profile and/or treatment selected is transmitted to a caregiver of the subject. In another embodiment, the expression profile and/or treatment is transmitted over a network.

In yet another aspect, the disclosure provides a method of treating a subject with breast cancer, the method comprising first determining if a subject gene is in the "high" or "low" prognostic class based on the methods described herewith. The, the subject is treated with a treatment indicated by their respective class or prognosis. In an embodiment, the method comprises administering to a subject for treatment of cancer an effective therapeutic amount of a cancer treatment indicated by a CUX1 Signature-high expression profile or a CUX1 Signature-low expression profile.

Where for example, the individualized gene expression level indicates the subject has a poor prognosis, the subject is, in an embodiment, treated with a breast cancer treatment indicated for poor prognosis (such as for example, chemotherapy). In an embodiment, treating the subject comprises administering an effective therapeutic amount of a chemotherapy, hormonal therapy, radiotherapy and/or combinations thereof.

The present invention will be more readily understood by referring to the following examples which are given to illustrate the invention rather than to limit its scope.

EXAMPLES

Example I - Material and Methods

Cell culture. Cell lines were generated by retroviral infection. Following blebbistatin treatment (100 μΜ for 12 h, Sigma) cells were allowed to recover for 4 hours and the media was changed 4 times before live-cell imaging or sub-cutaneous injections. Where indicated, MPS1 -IN-1 was added to the media during cell imaging, while 10 μΜ MG132 was added for 90 minutes and washed-off during imaging. Population doubling was calculated using the formula: N_(t) = N₍₀₎ x e^rt. The proliferation rate (r) was calculated between days 0-4 for untreated cells, and between days 2-6 for blebbistatin-treated cells. Apoptosis assays were done using the Annexin™ V-EGFP Apoptosis Detection Kit (Genscript). Cell sorting was performed on a MoFlo™ cytometer (Dako), after staining nuclei with Hoechst 33342 (2 μg/ml, 1 h).

Live-cell imaging. Time-lapse microscopy was performed on a Zeiss inverted microscope enclosed in a humidified chamber at 37°C, in Leibovitz's media plus 10% FBS using 20X and 32X objectives. Frames were taken every 5 minutes. The duration of mitosis was measured from nuclear envelope breakdown until nuclei were visible in daughter cells.

Chromosome spreads. Cells were treated with 100 ng/ml colcemid for 2 h, trypsinized, washed with PBS and swollen in 0.56% KCI for 8-12 minutes. Cells were centrifuged at 800g for 5 minutes, followed by two rounds of fixation in ice-cold Carnoy's fixative for 10 min. at room temperature. Cells were dropped on glass slides, dried and mounted with DAP I.

Sub-cutaneous injections. 2X10⁶ cells (in 100 μΙ PBS) were injected contra-laterally in 5- week old nude mice (CD1 nu/nu, Charles River Breeding Laboratories). Tumor volumes were measured around 30 days post-injection.

Promoter Occupancy and Gene Expression Analysis. Protocols for chromatin- immunoprecipitation, ChlP-on-chip, cDNA preparation and real-time PCR were described previously (Sansregret et al. , 2006). Validation was done with primers encompassing a <300bp region identified by ChlP-on-chip, with the QuantiTect SYBR Green™ PCR Kit (Qiagen), on a Rotor Gene 3000™ real-time PCR machine (Corbett). For gene expression analysis, RNA was isolated using the RNeasy™ Mini kit (Qiagen), and cDNA was prepared using Superscript II RNase H-reverse transcriptase kit (Invitrogen). Primer sequences are provided in Table 4. Gene Set Enrichment Analysis (GSEA) was performed as described in Gentleman et al., 2004. The significance of enrichment was determined by performing 3500 random gene set membership permutations.

Human Data Analysis. Relative expression of 29 CUX1 targets (Fig. 4A) was examined in 12 publicly available gene expression datasets comprising a total of 2481 patients with breast cancer. In each dataset, hierarchical clustering was performed using Euclidean distance and Ward's algorithm (data not shown). Kaplan-Meier survival analysis and the log-rank test were used to compare the patients in the "low" versus the "high" classes in all five molecular subtypes and in all patients combined. The analysis was completed using R/Bioconductor (Gentleman et al., 2004). Due to the differences in survival characteristics of the datasets (overall survival time vs. time to relapse), the associated times were classified as time to "outcome". The analysis of each individual dataset is shown in Figure 8.

Retroviral infection and cell culture establishment. Cells were maintained in DMEM (U20S, NMuMG, NIH3T3, Rati , HEK293) or McCoy's media (HCT1 16 and HCT1 16 p53-/-) supplemented with penicillin/streptomycin, glutamine and 10% fetal bovine serum (Gibco). Insulin (10 pg/ml, Sigma) was added for NMuMG cells. Mcfl Oa were cultured in DMEM/F12 5% FBS, 10 pg/ml insulin, 20 ng/ml EGF, 100 ng/ml cholera toxin, 0.5 pg/ml hydrocortisone. Stable cell populations were generated with a retrovirus (pREV/TRE vector, Clontech) expressing p1 10 Cux1 (Myc-[aa.747-1505]-HA). After hygromycin selection for five days, over 500 resistant clones were pooled together and the population was considered to be at passage 1 .

Microscopy. Cells were fixed in 3.7% paraformaldehyde, and staining were done in blocking solution (PBS, 5% FBS, 0.5% Triton X-100). Antibodies against γ-tubulin (Sigma T6557), a-tubulin (Abeam ab4074), phospho-histone H3 (Ser28, Cell Signaling #9713), Centrin 3 (Abeam) and secondary detection was done using Alexa-conjugated, species-specific secondary antibodies (Molecular Probes). DNA was stained with DAPI (Sigma). Confocal images were taken using a Zeiss 510 Meta laser scanning confocal microscope (Carl Zeiss, Canada Ltd, Toronto, ON) with a 100X objective. Volocity software (PerkinElmer) was used for image analysis.

FACS Analysis and Sorting. For DNA content analysis, cells were fixed in 75% EtOH and stored at -20°C until analysis. Cells were stained in PBS plus propidium iodide and RNase, then analyzed using a FACScan (Becton Dickinson), using single cell gating. Cell cycle profiles were analyzed using FlowJo™ (Tree Star softwares). Sorting was performed on a MoFlo™ (Dako), after staining with Hoechst 33342 (2 pg/ml, 1 h).

Antibodies, western blot analysis, Electrophoretic mobility shift assay (EMSA). pS1237 CUX1 rabbit polyclonal antibodies were generated using the phospho-peptide Cys- YSQGApSPQPQHQ (SEQ ID NO: 1 ), and purified by affinity chromatography. Rabbit anti-p21 and mouse anti-p53 antibodies were also used. EMSA was performed using end-labeled double stranded oligonucleotides (5'

TCGAGACGATATCGATAAGCTTCTTTTC 3', SEQ ID NO: 2). Western blot analysis. Cells lysates were prepared in RIPA-M (10mM Tris-HCI pH 8, 1 mM EDTA, 0.5 mM EGTA, 150 mM NaCI, 1 % Triton X-100, 0.5% DOC, 0.1 % SDS, 1 mM PMSF, protease inhibitor cocktail tablet and PhosStop™ tablet (Roche)). SDS- PAGE was performed and proteins were transferred to PVDF membranes, blocked in Tris-buffered saline-0.1 % Tween-20 (TBS 0.1 %T) containing 5% milk and 3% BSA. Membranes were probed with antibodies against CUX1 (a-861 ), pS¹²³⁷ CUX1 , Cyclin B (Lab Vision) or γ-Tubulin (Sigma). Primary antibodies were incubated in TBS 0.1 %T and detection was done using a horseradish peroxydase (HRP) conjugated a-rabbit or a- mouse secondary antibody in TBS 0.1 %T. Immuno-reactive proteins were visualized by chemiluminescence with ECL Western Blotting Detection Kit (Amersham Pharmacia Biotech).

Example II - Identification of CUX1 Transcriptional Targets and a corresponding CUX1 signature that predicts clinical outcome in breast outcome patients

All experimental procedures referred in this Example are presented in Example I.

Long-Term Expression of p110 CUX1 Causes Tetraploidization in HEK293 and NMuMG Cells. It was initially noticed that HEK293 cells stably expressing p1 10 CUX1 eventually formed a population composed exclusively of polyploid cells (Fig. 5A). Progressive polyploidization was repeatedly observed following p1 10 CUX1 expression both in NMuMG mouse mammary epithelial cells and HEK293 cells (Fig. lA and Fig. 5B). Moreover, polyploidization was accelerated with a mutant, p1 i o^s1237'^1270A CUX1 , that cannot be phosphorylated by cyclin A/Cdk1 and remains active in G2 (Fig. 1 B and Fig. 5C-D). Tetraploidization was not observed after 35 passages in 5 other cell lines, suggesting that p1 10 CUX1 does not itself induce tetraploidization but could facilitate the survival of polyploid cells in cell lines where cytokinesis failure occurs at a higher frequency. Consistent with this hypothesis, the spontaneous rate of binucleation was significantly higher in NMuMG cells (47/3029) than in Rati (15/3208; p < 0.0001 ) and U20S (22/3541 ; p < 0.0004) cells.

Chromosomal Instability and Aneuploidy in Late-passage NMuMG/CUX1p110 Cells. Subsequent experiments focused on the non-tumorigenic mammary epithelial NMuMG cells because these cells initially displayed a near-diploid karyotype. Diploid (2C) and tetraploid (8C) sub-populations were FACS-sorted from late-passage NMuMG/CUX1 cells (Fig. 1 C). 8C NMuMG/CUX1 cells maintained a 4C-8C DNA content and displayed 2 or 4 centrosomes per cell during interphase based on γ-tubulin staining (Fig. 1 C-D). During early mitosis, 38% of 8C NMuMG/CUX1 cells harbored multiple spindle poles, versus 8% for late-passage NMuMG/vector cells (Fig. 1 E, p < 0.001 ). In time-lapse microscopy, no multipolar divisions in 8C NMuMG/CUX1 cells were observed, indicating that extra-centrosomes were efficiently nucleated into two poles before anaphase (n = 698; Table S1 ). However, the duration of mitosis was extended by 10 minutes in these cells (48 min. vs 38.5 min. , p < 0.0001 , Table 5), in agreement with the notion that a longer mitosis may be an intrinsic characteristic of viable tetraploid cells. Although 8C NMuMG/CUX1 cells underwent bipolar mitoses like the 2C cells, a much higher proportion of 8C cells exhibited chromosome segregation defects during anaphase (Fig. 1 F, 51 % 8C vs 4% 2C, p < 0.001 ), such that almost all 8C cells displayed a sub- tetraploid chromosome count, ranging from 70 to 80 chromosomes per cell (Fig. 1 G). In contrast, chromosomal instability was a rare event in 2C NMuMG/CUX1 cells (Fig. 1 F and 1 G). Of note, sorted 8C NMuMG/CUX1 cells maintained a functional p53 pathway as judged from the stabilization of p53 and the upregulation of p21 following UV irradiation (Fig. 1 H). These results indicate that 8C NMuMG/CUX1 cells are prone to chromosomal instability and evolve to become a heterogeneous population of aneuploid cells.

p110 CUX1 Expression Favors Bi-Polar Mitoses in Newly Formed Tetraploid Cells. To verify the effect of p1 10 CUX1 on newly arising tetraploid cells, the reversible myosin II inhibitor blebbistatin was used to induce cytokinesis failure (Table 6). Twenty-four hours after tetraploidization, NMuMG/vector cells displayed nuclear abnormalities, while NMuMG/CUX1 cells were uniformly mononucleated. NMuMG/CUX1 cells proliferated significantly faster than control cells after treatment (Fig. 2A). This was due, at least in part, to a reduced rate of apoptosis as judged from Annexin V staining (Fig. 2B). Time- lapse microscopy revealed that most (63%) bi-nucleated NMuMG/vector cells underwent a multipolar anaphase (Fig. 2C). In contrast, most (73.5%) bi-nucleated NMuMG/CUX1 cells underwent a bipolar division (p < 0.0001 ). In both cell populations, bipolar division in tetraploid cells was associated with a longer duration of mitosis (Fig. 2C, p < 0.0001 ), whereas mitosis was unaffected in neighboring mononucleated cells (compare Fig. 2C with Table 5). Similar experiments in U20S cells and in the non- transformed human mammary epithelial MCF10A cells confirmed that p1 10 CUX1 and another isoform, p75 CUX1 , can promote bipolar mitoses in tetraploid cells (Fig. 2D, p < 0.0002).

Identification of CUX1 Transcriptional Targets Involved in Bipolar Mitosis. To identify transcriptional targets of p1 10 CUX1 that stimulate bipolar mitosis in tetraploid cells, a list of putative CUX1 transcriptional targets was first established by performing chromatin immunoprecipitation (ChIP) in Hs578T breast tumor cells followed by hybridization on the Human HG18 promoter Chl P-on-chip oligo microarray set of NimbleGen. Independent ChIP experiments were performed in HeLa cells to validate relevant targets. Twelve 12 genes were identified and validated . These 12 genes were previously found in a genome-wide RNAi screen as being required for bipolar mitosis in cells with supernumerary centrioles (Table 2, labeled with "*"). Activation of the spindle assembly checkpoint (SAC) was proposed to favor bipolar division indirectly by extending mitosis and providing sufficient time for centrosome clustering. Therefore, nine other SAC genes from our ChlP-on-chip dataset were also validated. The role of p1 10 CUX1 as a transcriptional activator of these 21 targets was demonstrated from three complementary approaches: overexpression of p1 10 CUX1 , inducible shRNA- mediated inhibition of endogenous CUX1 expression, and re-expression of CUX1 following removal of the shRNA inducer doxycyclin (Table 2, columns 3-5). Note in the latter experiment the striking increase in the expression of all targets upon re- expression of CUX1 (Table 2, column 5). In summary, these experiments demonstrate that many genes involved in the mitotic checkpoint are directly activated by CUX1 .

p110 CUX1 Facilitates Engagement of the Spindle Assembly Checkpoint. The role of CUX1 as a transcriptional activator of many genes involved in the control of bipolar mitosis made us consider the possibility that CUX1 overexpression might enable cells to engage more efficiently the SAC. In agreement with this notion, CUX1 expressing cells sustained an extended mitotic arrest and maintained higher cyclin B levels when the SAC was triggered by the treatment of cells with the microtubule poison nocodazole (Fig. 6A-B).

To verify whether efficient SAC engagement was necessary to allow bipolar division in newly formed tetraploid NMuMG/CUX1 cells, time-lapse microscopy after blebbistatin treatment was performed in the presence of MPS1 -IN-1 , a novel cell permeable inhibitor of the SAC kinase MPS1 . With increasing MPS1 -IN-1 concentrations, the duration of mitosis was progressively shortened and the frequency of multipolar divisions was correspondingly increased up to 3-fold at 5μΜ MPS1 -IN-1 (p < 0.0001 , Fig. 2E). Note that at any drug concentration the average duration of mitosis was significantly longer for bipolar than multipolar divisions (p < 0.0002). Importantly, these concentrations of MPS1 -IN-1 did not affect the outcome nor the duration of mitosis in neighboring mononucleated cells, indicating that tetraploid cells are intrinsically more sensitive to SAC inhibition than diploid cells. These results indicate that mitotic duration and bipolar division in tetraploid NMuMG/CUX1 cells are very sensitive to SAC inhibition. Moreover, these findings suggest that CUX1 promotes bipolar divisions by allowing tetraploid cells to delay mitosis, which in turn would increase the prospect of centrosome clustering. In support of this mechanism of action, the rate of bipolar division (live-cell) and bipolar spindle configuration (fixed cells) in U20S/vector cells was increased to about 80% by simply delaying anaphase onset using the proteasome inhibitor MG132 (Fig. 2F).

The Tumorigenic Potential of p110 CUX1 Is Associated with Chromosomal Instability. Tetraploidy and aneuploidy have previously been associated with increased tumorigenicity. The tumorigenic potential of NMuMG p1 10 CUX1 cells that have become aneuploid or remained diploid were compared. A sub-cutaneous injections in nude mice with late passage populations of cells carrying an empty vector, or the FACS sorted 2C and 8C NMuMG/CUX1 cells were performed (from Fig. 1 C). Significantly more and larger outgrowths were produced by the 8C NMuMG/CUX1 cells (Fig. 3A, p<0.0001 ). The fact that late passage 2C NMuMG/CUX1 cells failed to produce outgrowths strongly suggests that the acquisition of tumorigenic potential in p1 10 CUX1 expressing cells is associated with chromosomal instability. It was therefore directly tested whether p1 10 CUX1 expression enabled tumor outgrowth after cytokinesis failure. Early-passage NMuMG cells expressing p1 10 CUX1 or not were treated with blebbistatin prior to being subcutaneously injected into nude mice. The frequency and size of tumors were significantly higher in cells expressing p1 10 CUX1 than in cells carrying the empty vector (Fig. 3B, p=0.0002). These results together with the assays performed in tissue culture indicate that p1 10 CUX1 promotes the survival and proliferation of tetraploid cells (Fig. 2A-B).

If p1 10 CUX1 contributes to tumorigenicity by promoting chromosomal instability, aneuploidy should be a common feature in tumors of CUX1 transgenic mouse models. We examined the ploidy on metaphase spreads of 6 independent mammary tumors that arose in MMTV-CUX1 transgenic mice. Strikingly, each tumor showed a high level of aneuploidy with the majority of cells containing a sub-tetraploid chromosome content: 82% of the 229 cells scored contained between 60 and 80 chromosomes (Fig. 3C and Table 7). It was next verified whether CUX1 transcriptional targets identified in Table 2 were associated with tumor development in MMTV-CUX1 transgenic mice. An expression profiling on microdissected epithelial cells was performed and a gradual increase in the expression of the 21 genes was observed: a 1 .75 fold increase between normal mammary epithelial cells from non-transgenic and transgenic mice (Fig. 7A, p = 0.014) and in transgenic mice, a 2.31 fold increase between epithelial cells from normal adjacent mammary glands and mammary tumors (p = 0.0016). The significance of these differences was confirmed independently using two other statistical approaches: the hypergeometric test and the Gene Set Enrichment Analysis (GSEA) software (Fig. 7B-C). Together, these results show that transcriptional targets of p1 10 CUX1 identified in cell lines are also up-regulated in the mammary glands and mammary tumors of CUX1 transgenic mice, and most cells of the mammary tumors carry a sub-tetraploid chromosome content.

Targets of CUX1 Predict Clinical Outcome in Human Breast Cancers. A meta-analysis of 12 published gene expression datasets of breast cancer patients was performed to analyze the expression of CUX1 transcriptional targets. Most (1 1/12) datasets did not contain CUX1 specific probes. Moreover, CUX1 transcriptional activity cannot be predicted from mRNA expression since it depends on dephosphorylation and proteolytic processing. Therefore, as a surrogate for CUX1 activity a set of eight well-defined targets of CUX1 at the G1/S transition (CCNA2, CDC7, MCM3, MCM7, CDC45L, CDC25A, POLA and POLA2 and CTSL1 ) was used. For each dataset, patients were hierarchically clustered using this gene list with the Euclidean distance metric via Ward's algorithm. The gene signature showed significant co-variation across the datasets, with the majority of the 29 genes being co-expressed (Figure 8). This enabled us to use low and high expression of the signature to stratify patients and perform survival analysis (Fig. 9). Interestingly, high expression was associated with significantly poorer outcome when patients from all 12 datasets were combined (Fig. 4A, Cox- regression p-value < 10-12, Cox-regression hazard ratio (HR) high vs. low = 1 .92) and in 8 out of 12 individual datasets (Fig. 8). The gene signature was also a strong predictor of outcome in univariate analysis (Table 3, p = 2.2 x 10-16, HR = 2.27). The molecular subtype distribution within the "low" and "high" expression groups was then determined as classified by the gene signature, based on the correlation to centroids created from the Prediction Analysis of Microarray (PAM) method using a 50 gene signature previously described. The breast cancer molecular subtypes were not distributed equally between the two groups. Those subtypes with a poorer outcome, namely HER2+ and basal-like, were over-represented in the high expression group, (7.6-fold, 2.4-fold respectively), while the proportion of the good outcome-enriched luminal A subtype was reduced 5.7-fold (Table 8). Furthermore, the signature identified subgroups of luminal A (72/557 = 1 1 .4%, p = 0.013, HR = 1 .96) and luminal B (233/384 = 37.8%, p = 0.0028, HR = 1 .56) patients with poorer clinical outcome (Fig. 9A). When the analysis was restricted to patients with similar clinico-pathological features, high expression of the signature was associated with poor outcome regardless of grade (grade 1 : p = 0.0034, HR = 2.56; grade 2: p < 10-5, HR = 1 .89; grade 3: p = 0.0014, HR = 1 .64), axillary lymph-node (LN) status (LN+: p < 10-5, HR=1 .92; LN-: p < 10-8, HR 2.33) or estrogen receptor (ER) status (ER+: p < 10-12, HR=2.38; ER- p = 0.047, HR=1 .51 ) (Fig. 9B-D). Importantly, multivariate Cox regression analysis confirmed that the gene signature was a significant predictor of outcome that is independent of grade, LN, HER2 and ER status (Table 3, p = 7.1 x 10-8, HR = 1 .85). The signature was predictive of outcome both among patients considered at risk, like the LN+/HER2+ (Fig. 9C, 49/120, p = 0.0014, HR = 2.63), and among patients at a lower risk, like the ER+/LN- patients (Fig. 4B, 196/584 = 34%, p = 10-8, HR = 2.70). This appears to be not solely due to the enrichment for poor-outcome subtypes within this cluster, since high expression was associated with poor outcome within luminal A and luminal B patients (Fig. 4D bottom panels). Thus, this expression signature is able to identify patients at risk of relapse among ER+/LN- patients.

In the present study, it was shown that p1 10 CUX1 contributes to the establishment of a transcriptional program that enables cells to efficiently engage SAC signaling, thus allowing the survival and proliferation of polyploid cells that evolve to become aneuploid and tumorigenic. Aneuploidy from a tetraploid precursor was thought to arise from multipolar mitoses, but recently such events were found to generate non-viable cells and to occur much less frequently than originally suspected. Yet, although centrosome clustering may succeed in producing bipolar mitosis, transient multipolar spindles were shown to increase the occurrence of merotely which, if not corrected, cause chromosome mis-segregation. Similar observations were made in cells expressing p1 10 CUX1 . While indirect immunofluorescence on fixed cells showed a high proportion of cells with a multipolar spindle configuration (Fig. 1 E), live-cell imaging revealed that centrosome clustering to two poles was eventually achieved in virtually all of these cells (Table 5), but was accompanied by frequent chromosome segregation defects (Fig. 1 F- G).

Two lines of observations suggest that p1 10 CUX1 does not itself induce tetraploidization but affects its outcome once it has occurred. Firstly, constitutive expression of p1 10 CUX1 did not cause a defect in mitosis or cytokinesis, and was associated with tetraploidization in only 2 out of 7 cell lines expressing p1 10 CUX1 (Fig. 1 , 5). Secondly, following induction of tetraploidy with blebbistatin, p1 10 CUX1 enabled a greater proportion of cells to undergo a normal, bipolar, cell division (Fig. 2C), but the protective effect of p1 10 CUX1 was lost in the presence of a SAC kinase inhibitor (Fig. 2E). The extended mitotic arrest in the presence of nocodazole (Fig. 6A), the delay in mitosis following blebbistatin treatment (Fig. 2C, 2E), and the loss of protective effect in the presence of an inhibitor of the SAC kinase MPS1 (Fig. 2E) all concur to suggest that p1 10 CUX1 mediates its effects by establishing a transcriptional program that enables efficient SAC engagement, thereby allowing more time for multiple centrosomes to cluster. In addition, we must consider the possibility that p1 10 CUX1 directly facilitates bipolar mitosis by stimulating the expression of genes that are involved in centrosome clustering. In support of this notion, we reported that KIF2C and KIFC1 are direct targets that are activated by p1 10 CUX1 (Table 2). Unfortunately, it is not easy to verify a direct effect on centrosome clustering because of the close link that exists between the time spent in mitosis and successful clustering. To eliminate the effect of time, we transiently inhibited anaphase onset using MG132 and we analyzed cell division in U20S cells made tetraploid with a blebbistatin treatment (Fig. 2F). In this context, expression of p1 10 CUX1 did not significantly increase the rate of bipolar division over the already high rate observed in control cells carrying the empty vector (Fig. 2F, 80% bipolar for vector vs 83% for CUX1 ). More experiments will be needed to ascertain the effect of p1 10 CUX1 on centrosome clustering.

Because of the unusual structure of the CUTL1 gene, currently most expression microarrays contain oligonucleotides for the Cut-alternatively spliced product, CASP, but not for CUX1 . Therefore, a surrogate for CUX1 expression was used and a set of well-characterized targets that play a role in S phase entry were selected. Across all large-scale breast cancer gene expression datasets, CUX1 transcriptional targets that play a role in mitosis were found to cluster with established targets of CUX1 at the G1/S transition. Linking these two classes of genes produced a gene expression signature that is strongly associated with poor clinical outcome. A group of CUX1 targets was identified that has predictive potential in cancer and that might be important to mediate its oncogenic activity. Not only was high expression of these genes found more frequently in breast tumor subtypes that exhibit a poor prognosis, like the basal-like and HER2+, but it also identified patients with poor outcome within the luminal A and luminal B subtypes (Fig. 4). The finding that these two sets of genes together have prognostic value could have important practical application in the clinic since the luminal subtypes represent close to 50% of breast tumors and there is an urgent need to identify which node-negative patients in this subtype would benefit from adjuvant therapy and which ones should be spared from the toxicities associated with these treatments. Moreover, since our gene signature includes many genes involved in mitotic processes and microtubule-based activity, it has potential to be a useful predictor of treatment response for chemotherapeutic regimens that include anti-microtubule agents. Indeed, the results using the MPS1 inhibitor clearly show that cells with centrosomal aberrations are intrinsically more sensitive than normal cells to spindle checkpoint inhibition (Fig. 2E), thus suggesting a therapeutic window for drug efficacy.

Transcriptional targets of p1 10 CUX1 preventing multipolar divisions in tetraploid cells were identified. A list of putative CUX1 transcriptional targets was first established by performing chromatin immunoprecipitation (ChIP) in Hs578T cells followed by hybridization on the Human HG18 promoter Chl P-on-chip oligo microarray set of NimbleGen. Independent ChIP experiments were performed in HeLa cells to validate relevant targets. Twelve genes previously identified in a genome-wide RNAi screen as being required for bipolar mitosis in Drosophila S2 cells (Table 2, column 1 labeled with "*") were validated. Nine additional genes involved in the Spindle Assembly Checkpoint ("SAC") were identified. The role of p1 10 CUX1 as a transcriptional activator of these targets was demonstrated from three complementary approaches: overexpression of p1 10 CUX1 , inducible shRNA-mediated inhibition of CUX1 expression, and re- expression of CUX1 following removal of the shRNA inducer doxycyclin (Table 2, columns 3-5). This approach revealed that a surprisingly high number of CUX1 targets were directly involved in the spindle checkpoint.

This subset of 21 CUX1 targets exhibited co-expression in human tumors. A metaanalysis of 12 published gene expression datasets of breast cancer patients was also performed (2481 patients) In order to investigate the relationship between CUX1 expression and its targets a surrogate method to determine CUX1 activity needed to be determined as there are two problems associated with CUX1 expression in microarray analysis. Firstly, microarray platforms used in for most datasets (1 1 /12) did not contain CUX1 specific probes. Secondly, mRNA expression does not correlate with CUX1 activity since transcriptional activation by CUX1 requires proteolytic processing by Cathepsin L (CTSL1) and dephosphorylation by Cdc25A. Therefore, as a surrogate for CUX1 activity, a set of nine well-defined targets of CUX1 at the G1/S transition was used (CCNA2, CDC7, MCM3, MCM7, CDC45L, CDC25A, ORCL1 , POLA, POLA2 and CTSL1). Since ORCL1 and KNTC1 (ROD) are not represented on any microarray, the CUX1 Signature includes 29 genes (Fig. 1 A; Table 2). ORCL1 and KNTC1 (ROD) could potentially be added to the CUX1 signature gene set. For each dataset, patients were hierarchically clustered using this gene list with the Euclidean distance metric via Ward's algorithm. The gene signature showed significant co-variation across the datasets, with the majority of the 29 genes being co-expressed. Low and high expression of the signature was used to stratify patients (Fig. 8) and perform survival analysis. Interestingly, high expression was associated with significantly poorer outcome when patients from all 12 datasets were combined (Fig. 4A, Cox-regression p- value < 10^"12, Cox-regression hazard ratio (HR) high vs. low = 1 .92) and in 8 out of 12 individual datasets (Fig. 8). The CUX1 -targets signature was also a strong predictor of outcome in univariate analysis (Table 3, p = 2.2 x 10^"16, HR = 2.27).

Gene-expression profiling studies have led to the molecular classification of breast cancers into subtypes with distinct prognosis: luminal A, luminal B, HER2+, basal-like and normal-like. The molecular subtype distribution was determined within the "low" and "high" expression groups as classified by the CUX1 -target signature, based on the correlation to centroids created from the Prediction Analysis of Microarray (PAM) method using a 50 gene signature previously described (Parker et al., 2009). The breast cancer molecular subtypes were not distributed equally between the two groups. Those subtypes with a poorer outcome, namely HER2+ and basal-like, were over-represented in the high expression group, (7.6-fold, 2.4-fold respectively), while the proportion of the good outcome-enriched luminal A subtype was 5.7-fold reduced (Table 8). Furthermore, the signature identified subgroups of luminal A (72/557 = 1 1 .4%, p = 0.013, HR = 1 .96) and luminal B (233/384 = 37.8%, p = 0.0028, HR = 1 .56) patients with poorer clinical outcome. The analysis was then restricted to patients with similar clinico-pathological features. Surprisingly, high expression of the signature was associated with poor outcome regardless of grade (grade 1 : p = 0.0034, HR = 2.56; grade 2: p < 10^"5, HR = 1 .89; grade 3: p = 0.0014, HR = 1 .64), auxiliary lymph-node (LN) status (LN+: p < 10^"5, HR=1 .92; LN-: p < 10^"8, HR 2.33) or estrogen receptor (ER) status (ER+: p < 10^"12, HR=2.38; ER- p = 0.047, HR=1 .51 ) (Fig. 9). Importantly, multivariate Cox regression analysis confirmed that the CUX1 -target signature was a significant predictor of outcome that is independent of grade, LN, HER2 and ER status (Table 3, p = 7.1 x 10^"8, HR = 1 .85). The signature was predictive of outcome both among patients considered at risk, like the LN+/HER2+ (Table 8, 49/120, p = 0.0014, HR = 2.63), and among patients at a lower risk, like the ER+/LN- patients (Table 8, 196/584 = 34%, p = 10^"8, HR = 2.70). This appears to be not solely due to the enrichment for poor-outcome subtypes (HER2+, basal-like) within this cluster, since high expression was associated with poor outcome within luminal A and luminal B patients (Table 8, Fig. 9). Thus, the CUX1 expression signature is able to identify patients at risk of relapse among ER+/LN- patients.

Example 3 - Clinical use of the CUX1 signature

An illustration of a use of this technology in the clinic is as follows: A patient is diagnosed as having breast cancer by a clinician. At biopsy or at surgery, a tissue sample is removed, processed as described above and the relative expression levels of genes contained within the predictive set as herein presented are measured. For a new patient, the prognosis generated by the CUX1 signature is calculated as follows. For each new patient, the CUX1 signature assigns the corresponding probability of recurrence by determining the Euclidean distance between the measured expression of each gene in the CUX1 signature and a centroid for the "high" and "low" classes.

If the probability of recurrence of the patient is similar to the CUX1 signature-high class then the patient is then considered to have a poor outcome or be recurrent.

If the probability of recurrence of the patient is similar to the CUX1 signature-low class then the patient is then considered to have a good outcome or be non-recurrent. While the present disclosure has been described with reference to what are presently considered to be the preferred examples, it is to be understood that the disclosure is not limited to the disclosed examples. To the contrary, the disclosure is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

All publications, patents and patent applications are herein incorporated by reference in their entirety to the same extent as if each individual publication, patent or patent application was specifically and individually indicated to be incorporated by reference in its entirety. All sequences (e.g. nucleotide, including RNA and cDNA, and polypeptide sequences) of genes listed in Table 1 , for example referred to by accession number are herein incorporated specifically by reference.

Table 1. CUX1 Signature Gene Set

Gene Symbol Gene Name Function Entrez Accession

Gene ID Number

AURKB Aurora Kinase B Genetic 9212 AB01 1446 instability

BUB1 budding uninhibited by Genetic 699 AF043294 benzimidazoles 1 instability

homolog (yeast)

BUBR1 budding uninhibited by Genetic 701 NM_00121 1 (BUB1B) benzimidazoles 1 instability

homolog beta (yeast)

BUB3 budding uninhibited by Genetic 9184 AU 160695 benzimidazoles 3 instability

homolog (yeast)

CAMK2D calcium/calmodulin- 817 AA017093 dependent protein kinase

II delta

CCNA2 cyclin A2 Cell 890 NM_001237 proliferation

CCNE2 cyclin E2 Cell 9134 AF091433 proliferation

CDC25A cell division cycle 25 Cell 993 AI343459 homolog A (S. pombe) proliferation

CDC45L cell division cycle 45 Cell 8318 NM_003504 homolog (S. cerevisiae) proliferation

CDC7 cell division cycle 7 Cell 8317 NM_003503 homolog (S. cerevisiae) proliferation

CDKN1A cyclin-dependent kinase Cell 1026 NM 000389.4 inhibitor 1A proliferation NM_078467.2

CENPA centromere protein A Genetic 1058 NM_001809 instability

CENPE centromere protein E, Genetic 1062 NM_001813

312kDa instability

COG7 component of oligomeric 91949 R61519

golgi complex 7

CTSL1 cathepsin L1 1514 NM_001912

KIF2C kinesin family member Genetic 1 1004 U63743

2C instability

KIFC1 kinesin family member Genetic 3833 BC000712 Gene Symbol Gene Name Function Entrez Accession

Gene ID Number

C1 instability

MAD1L1 MAD1 mitotic arrest Genetic 8379 NM_003550 deficient-like 1 (yeast) instability

MAD2L1 MAD2 mitotic arrest Genetic 4085 NM_002358 deficient-like 1 (yeast) instability

MCM3 minichromosome Cell 4172 NM_002388 maintenance complex proliferation

component 3

MCM7 minichromosome Cell 4176 D55716

maintenance complex proliferation

component 7

MFAP1 microfibrillar-associated 4236 NM_005926

protein 1

NEK2 NIMA (never in mitosis Genetic 4751 NM_002497 gene a)-related kinase 2 instability

NUDCD1 NudC domain containing 84955 BC000967

1

ORCL1 origin recognition Cell 4998 NM 001 190818.1 complex, subunit 1 proliferation NM 001 190819.1

NM_004153.3

POLA DNA polymerase I Cell 8873624 NM_016937 proliferation

POLA2 polymerase (DNA Cell 23649 NM_002689 directed), alpha 2 (70kD proliferation

subunit)

RIN2 Ras and Rab interactor 2 54453 AL136924

KNTC1 (ROD) kinetochore associated 1 Genetic 9735 NM_014708.4 instability

SF3B3 splicing factor 3b, 23450 NM_012426

subunit 3, 130kDa

TTK (MPS1) TTK protein kinase Genetic 7272 NM_003318 instability

ZW10 ZW10, kinetochore Genetic 9183 NM_004724

associated, homolog instability

(Drosophila)

ZWILCH Zwilch, kinetochore Genetic 55055 NM_017975 associated, homolog instability

(Drosophila) Table 2. Validated CUX1 Transcriptional Targets

Promoter mRNA Expression (Fold Difference) Occupancy (Fold

Difference)

ChIP Hela/CUX1 NMuMG Hs578t shCUX1^tetUN

(IgG vs. Input)

Gene CUX1 vs. + Dox (CUX1 - Dox (CUX1

Vector inhibition) re-induction)

Spindle Assembly

Checkpoint

AURKB + 3.8 + 2.6 - 1.3 + 7.0

BUB1* + 3.8 + 3.6 -2.5 + 5.6

BUB3 + 3.5 + 2.4 -2.4 + 7.5

BUBR1 + 1.7 + 3.6 -2.2 + 10.8

CENPE* + 10.8 + 2.6 - 1.6 + 6.7

MAD1L1 + 4.7 + 1.8 - 1.5 + 3.9

MAD2L1* + 2.7 + 2.0 -2.6 + 5.7

TTK(MPS1) + 2.0 + 3.2 - 1.2 + 5.6

NEK2 + 2.5 + 3.8 - 1.4 + 6.5

KNTC1 (ROD) + 3.6 + 3.9 - 1.5 + 5.9

ZW10 + 3.1 + 2.0 - 1.8 + 6.3

ZWILCH + 2.9 + 5.0 -2.5 + 4.3

Centromeric

histone

CENPA* + 3.1 + 1.9 - 1.5 + 5.8

Kinesin

KIF2C* + 2.3 + 3.2 - 1.4 + 6.1

KIFC1* + 1.9 + 2.3 - 1.8 + 9.4

Other functions

CAMK2D* + 2.9 + 1.4 - 1.9 + 2.0

COG7* + 8.1 + 1.6 - 1.4 + 1.3

MFAP1* + 4.3 + 1.4 - 1.7 + 3.9

NUDCD1* + 3.7 + 2.6 - 1.2 + 3.8

RIN2* + 4.2 + 2.3 - 1.6 + 2.4

SF3B3* + 3.0 + 2.0 - 1.9 + 4.6 Table 3. Univariate and Multivariate Analyses

Univariate Multivariate

Variable Hazard Ratio p-value Hazard Ratio p-value

(95% CI*) (95% CI)

Grade 2 (vs.1) 2.26 (1.63-3.15) 1.1 x 10^"B 1.87 (1.34-2.61) 2.6 x10^"4

Grade 3 (vs.1) 2.92 (2.10-4.05) 1.6 x 10^"Ίυ 1.72 (1.19-2.49) 4.0 x 10^"a

Lymph Node 1.63 (1.34-1.98) 6.3 x 10^"' 1.57 (1.29- 1.90) 5.7χ10^"ΰ

HER2 1.60 (1.25-2.05) 1.6 x 10^"4 1.20 (0.93- 1.54) 1.6χ10^"Ί

ER 1.57 (1.28-1.94) 2.1 x 10^"b 1.18 (0.93- 1.49) 1.6χ10^"Ί

CUX1 2.27 (1.87-2.76) 2.2 x 10^"Ίΰ 1.85 (1.48-2.32) 7.1 x 10^"B Signature

Table 4. Primer Sequences

GENE 5'- Human Genomic (SEQ ID Human cDNA (SEQ ID NO) Mouse cDNA (SEQ ID 3' NO) NO)

F

AURKB R

F

BUB1 R

F

BUB3 R

BUB1B F

R

CENPA F

R

F

CENPE

R

F

MAD1L1 R

F

MAD2L1 R

F

TTK R

(MPS1)

F

NEK2 R

F

ROD R

F

ZW10 R

F

ZWILCH R GENE 5'- Human Genomic (SEQ ID Human cDNA (SEQ ID NO) Mouse cDNA (SEQ ID 3' NO) NO)

F

KIFC1 R

F

KIF2C R

F

CAMK2D

R

F

COG7 R

F

MFAP R

F

NUDCD1 R

F

RIN2 R

F

SF3B3 R

Table 5. Cell Division Followed by Time-Lapse Microscopy

The frequency of bipolar and multipolar division was assessed for late-passage NMuMG/vector (n=679), early passage NMuMG/CUX1 (n=596) and late-passage 8C- sorted NMuMG/CUX1 cells (n=698). Images were taken every five minute, and the average time for mitosis was measured (n= 267 for vector; n=267 for early passage CUX1 cells; n=274 for 8C-sorted cells). ** = p<0.0001 . Mitosis was extended by -24% in tetraploid NMuMG/CUX1 cells (48 min) as compared to NMuMG/vector cells (38.6 min) or low passage NMuMG/CUX1 cells (38.5 min, ** = p < 0.0001 ).

Table 6. Blebbistatin Treatment Generates Binucleated Cells

The frequency of mononucleated, binucleated or multinucleated cells after 12h incubation with 100 μΜ blebbistatin was assessed by microscopy using DAPI staining. The low percentage of multinucleated cells, and the low percentage of mononucleated cells suggest that blebbistatin treatment did not encompass more than one doubling time. Table 7. Percentage of cells containing a specific number of chromosomes in tumors arising in p1 10 or p75 CUX1 transgenic mice determined from metaphase spread

Table 8. Breakdown of the percentage of patients classified into each molecular subtype for "low" and "high" expression groups

Low High

Luminal A 39.2 6.8

Luminal B 25.6 22.0

Normal-like 19.6 3.1

HER2+ 9.8 23.7

Basal-like 5.8 44,4 Table 9. Specificity of CUTL1 probes in microarray datasets

Most CUTL1 probes present on microarrays are specific for the CDP/Cut Alternatively Spliced Product (CASP), a Golgi resident protein that possesses an entirely different C- terminus. REFERENCES

Albain, K.S. et al. The lancet oncology 1 1 , 55-65 (2010).

Cadieux, C. et al. Cancer Res 66, 9492-9501 (2006).

Cadieux, C. et al. Cancer Res 69, 7188-7197 (2009).

Cimini, D. et al. J Cell Biol 153, 517-27 (2001 ).

Coqueret, O. et al. EMBO J 17, 4680-4694 (1998).

De Vos, J. et al. Oncogene 21 , 6848-6857 (2002).

Gentleman, R.C. et al. Genome Biol 5, R80 (2004).

Goulet, B. et al. Mol Cell 14, 207-219 (2004).

Goulet, B. et al. Cancer Res. 62, 6625-6633 (2002).

Harada, R. et al. Nucleic Acids Res 36, 189-202 (2008).

Kedinger, V. and Nepveu. Cell Ad h Migr A, 348-352 (2010).

Michl, P. et al. Cancer Cell 7, 521 -532 (2005).

Moon, N.S. et al. Protein. Mol. Cell. Biol. 21 , 6332-6345 (2001 ).

Moon, N.S et al. Int. J. Cancer 100, 429-432 (2002).

Paik, S. et al. N Engl J Med 351 , 2817-2826 (2004).

Paik, S. et al. J Clin One 24, 3726-3734 (2006).

Parker, J.S. et al. J Clin Oncol 27, 1 160-7 (2009).

Perou, CM. et al. Nature 406, 747-52 (2000).

Quintyne, N.J. et al. S.M. & Saunders, W.S. Science 307, 127-9 (2005).

Ripka, S. et al. Carcinogenesis 28, 1 178-1 187 (2007).

Ripka, S. et al. Gut 59, 1 101 -1 1 10 (2010). Sansregret, L. et al. Mol Cell Biol 26, 2441-55 (2006).

Sansregret, L. etal. Proc. Natl. Acad. Sci. USA 108, 1949-1954 (2011). Sorlie, T. et al. Proc Natl Acad Sci U S A 98, 10869-74 (2001).

Sorlie, T. et al. Proc Natl Acad Sci USA.100, 8418-23 (2003).

Truscott, M. et al. J Biol Chem 282, 30216-30226 (2007).

Claims

CLAIMS:

1. A method for determining a clinical outcome of a subject having received an initial diagnosis of cancer, the method comprising:

(a) measuring an individualized gene expression level in a plurality of genes of a CUX-1 signature gene set in a sample of the subject, wherein the CUX-1 signature gene set comprises at least one gene associated with genetic instability and at least one gene associated with cell proliferation;

(b) normalizing the individualized gene expression level against at least one control gene expression level to provide a normalized individualized gene expression level;

(c) comparing at least one parameter of the normalized individualized gene expression level with data derived from a set of CUX-1 expression profiles having known clinical outcomes divided in at least two prognostic classes in accordance with the known clinical outcomes; and

(d) associating one of the known clinical outcomes with the normalized individualized gene expression level in accordance with the comparison.

2. The method of claim 1 , wherein the at least one gene associated with genetic instability is selected from the group consisting of AURKB, BUB1 , BUBR1 , BUB3, CENPA, CENPE, KIF2C, KIFC1 , KNTC1 , MAD1 L1 , MAD2L1 , NEK2, TTK, ZW10 and ZWILCH.

3. The method of claim 1 or 2, wherein the at least one gene associated with cell proliferation is selected from the group consisting of CCNA2, CCNE2, CDC25A,

CDC45L, CDC7, CDKN1A, MCM3, MCM7, ORCL1 , POLA, POLA2 and KNTC1 .

4. The method of any one of claims 1 to 3, wherein, in step (a), the individualized gene expression level is measured in at least 20 different genes of the CUX1 signature gene set.

5. The method of any one of claims 1 to 4, wherein the at least two prognostic classes comprises a first prognostic class and a second prognostic class.

6. The method of claim 5, wherein the CUX1 expression profiles associated with the first prognostic class comprises an increase, with respect to CUX1 expression profiles associated with the second prognostic class, in the gene expression level of at least one of the following gene AURKB, BUB1 , BUB3, BUBR1 , CENPE, MAD1 L1 , MAD2L1 , TTK, NEK2, KNTC1 , ORCL1 , ZW10, ZWILCH, CENPA, KIF2C, KIFC1 , CAMK2D, COG7, MFAP1 , NUDCD1 , RIN2, SF3B3, CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CTSL1 , MCM3, MCM7, POLA and POLA2.

7. The method of claim 5 or 6, wherein the CUX1 expression profiles associated with the first prognostic class comprises a decrease, with respect to the CUX1 expression profiles associated with the second prognostic class, in the gene expression level of the CDKN1A gene.

8. The method of claim 6 or 7, wherein the first prognostic class is associated with a poor prognosis.

9. The method of any one of claims 5 to 8, wherein the CUX1 expression profiles associated with the second prognostic class comprises a decrease, with respect to the CUX1 expression profiles associated with the first prognostic class, in the gene expression level of at least one of the following gene AURKB, BUB1 , BUB3, BUBR1 , CENPE, MAD1 L1 , MAD2L1 , TTK, NEK2, KNTC1 , ORCL1 , ZW10, ZWILCH, CENPA, KIF2C, KIFC1 , CAMK2D, COG7, MFAP1 , NUDCD1 , RIN2, SF3B3, CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CTSL1 , MCM3, MCM7, POLA and POLA2.

10. The method of any one of claims 5 to 9, wherein the CUX1 expression profiles associated with the second prognostic class comprises an increase, with respect to the CUX1 expression profiles associated with the first prognostic class, in the gene expression level of the CDKN1A gene.

11. The method of claim 9 or 10, wherein the second prognostic class is associated with a good prognosis.

12. The method of any one of claims 1 to 1 1 , wherein the CUX1 expression profiles comprise gene expression levels for the plurality of genes of the CUX1 signature gene set from an heterogeneous population of subjects having been diagnosed with cancer.

13. The method of any one of claims 1 to 12, wherein the individualized gene expression level is measured by determining the mRNA level of the plurality of genes.

14. The method of any one of claim 1 to 13, further comprising applying a probability model to the normalized individualized gene expression level and generating an output, wherein the data derived from the set of CUX-1 expression profiles comprises a scoring scheme derived therefrom, and wherein comparing at least one parameter of the normalized individualized gene expression level comprises applying the output to the scoring scheme.

15. The method of claim 14, wherein the probability model is a naive Bayes' modifier.

16. The method of any one of claims 1 to 15, wherein the known clinical outcomes comprises cancer recurrence, cancer metastasis, survival and/or usefulness of a chemotherapy treatment.

17. The method of any one of claims 1 to 16, wherein the cancer is breast cancer.

18. A system for determining a prognosis in a subject having received an initial diagnosis of cancer, said system comprising:

(a) a reaction vessel for combining a sample from the subject and an analyte- specific reagent (ASR) for measuring the gene expression level of a plurality of genes of a CUX1 signature gene set in a sample of the subject, wherein the CUX1 signature gene set comprises at least one gene associated with genetic instability and at least one gene associated with cell proliferation;

(b) a processor in a computer system;

(c) a memory accessible by the processor; and

(d) at least one application coupled to the processor and configured for: i. receiving a measure of an individualized gene expression level in the plurality of genes of the CUX-1 signature gene set in a sample of the subject; ii. normalizing the individualized gene expression level against at least one control gene expression level to provide a normalized individualized gene expression level; iii. comparing at least one parameter of the normalized individualized gene expression level with data derived from a set of CUX-1 expression profiles having known clinical outcomes divided into at least two prognostic classes in accordance with the known clinical outcomes; and iv. associating one of the known clinical outcomes with the normalized individualized gene expression level in accordance with the comparison.

19. The system of claim 18, wherein the at least one gene associated with genetic instability is selected from the group consisting of AURKB, BUB1 , BUBR1 , BUB3, CENPA, CENPE, KIF2C, KIFC1 , KNTC1 , MAD1 L1 , MAD2L1 , NEK2, TTK, ZW10 and ZWILCH.

20. The system of claim 18 or 19, wherein the at least one gene associated with cell proliferation is selected from the group consisting of CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CDKN1A, MCM3, MCM7, ORCL1 , POLA and POLA2.

21. The system of any one of claims 18 to 20, wherein the individualized gene expression level comprises measurements in at least 20 different genes of the CUX1 signature gene set.

22. The system of any one of claims 18 to 21 , wherein the at least two prognostic classes comprises a first prognostic class and a second prognostic class.

23. The method of claim 22, wherein the CUX1 expression profiles associated with the first prognostic class comprises an increase, with respect to CUX1 expression profiles associated with the second prognostic class, in the gene expression level of at least one of the following gene AURKB, BUB1 , BUB3, BUBR1 , CENPE, MAD1 L1 , MAD2L1 , TTK, NEK2, KNTC1 , ORCL1 , ZW10, ZWILCH, CENPA, KIF2C, KIFC1 , CAMK2D, COG7, MFAP1 , NUDCD1 , RIN2, SF3B3, CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CTSL1 , MCM3, MCM7, POLA and POLA2.

24. The method of claim 22 or 23, wherein the CUX1 expression profiles associated with the first prognostic class comprises a decrease, with respect to the CUX1 expression profiles associated with the second prognostic class, in the gene expression level of the CDKN1A gene.

25. The method of claim 23 or 24, wherein the first prognostic class is associated with a poor prognosis.

26. The method of any one of claims 22 to 25, wherein the CUX1 expression profiles associated with the second prognostic class comprises a decrease, with respect to the CUX1 expression profiles associated with the first prognostic class, in the gene expression level of at least one of the following gene AURKB, BUB1 , BUB3, BUBR1 , CENPE, MAD1 L1 , MAD2L1 , TTK, NEK2, KNTC1 , ORCL1 , ZW10, ZWILCH, CENPA, KIF2C, KIFC1 , CAMK2D, COG7, MFAP1 , NUDCD1 , RIN2, SF3B3, CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CTSL1 , MCM3, MCM7, POLA and POLA2.

27. The system of any one of claims 22 to 26, wherein the CUX1 expression profiles associated with the second prognostic class comprises an increase, with respect to the

CUX1 expression profiles associated with the first prognostic class, in the gene expression level of the CDKN1A gene.

28. The system of claim 26 or 27, wherein the second prognostic class is associated with a good prognosis.

29. The system of any one of claims 18 to 28, wherein the CUX1 expression profiles comprise gene expression levels for the plurality of genes of the CUX1 signature gene set from an heterogeneous population of subjects having been diagnosed with cancer.

30. The system of any one of claims 18 to 29, wherein the individualized gene expression level comprises a measure of the mRNA level of the plurality of genes.

31. The system of any one of claim 18 to 30, wherein the at least one application is configured for applying a probability model to the normalized individualized gene expression level and generating an output, wherein the data derived from the set of CUX-1 expression profiles comprises a scoring scheme derived therefrom, and wherein comparing at least one parameter of the normalized individualized gene expression level comprises applying the output to the scoring scheme.

32. The system of claim 31 , wherein the probability model is a naive Bayes' modifier.

33. The system of any one of claims 18 to 32, wherein the known clinical outcomes comprises cancer recurrence, cancer metastasis, survival and/or usefulness of a chemotherapy treatment.

34. The system of any one of claims 18 to 33, wherein the cancer is breast cancer.

35. The system of any one of claims 18 to 34, wherein the ASR comprises at least one oligonucleotide listed in Table 4.

36. The system of any one of claims 18 to 35, further comprising a display device for outputting of the associated known clinical outcome.

37. A software product embodied on a computer readable medium and comprising instructions for determining a prognosis in a subject having received an initial diagnosis of cancer, said product comprising:

(a) a receiving module for receiving a measurement of an individualized gene expression level in a plurality of genes in a CUX1 signature gene set in a sample of the subject, wherein the CUX1 signature gene set comprises at least one gene associated with genetic instability and at least one gene associated with cell proliferation;

(b) a normalizing module for normalizing the individualized gene expression level against at least one control gene expression level to provide a normalized individualized gene expression level; (c) a comparison module for comparing at least one parameter of the normalized individualized gene expression level with data derived from a set of CUX-1 expression profiles having known clinical outcomes divided at least two prognostic classes in accordance with the known clinical outcomes; and (d) a characterization module for associating one of the known clinical outcomes with the normalized individualized gene expression level in accordance with the comparison.

38. The software product of claim 37, wherein the at least one gene associated with genetic instability is selected from the group consisting of AURKB, BUB1 , BUBR1 , BUB3, CENPA, CENPE, KIF2C, KIFC1 , KNTC1 , MAD1 L1 , MAD2L1 , NEK2, TTK, ZW10 and ZWILCH.

39. The software product of claim 37 or 38, wherein the at least one gene associated with cell proliferation is selected from the group consisting of CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CDKN1 A, MCM3, MCM7, ORCL1 , POLA and POLA2.

40. The software product of any one of claims 37 to 39, wherein the individualized gene expression level comprises measurements in at least 20 different genes of the CUX1 signature gene set.

41. The software product of any one of claims 37 to 40, wherein the at least two prognostic classes comprises a first prognostic class and a second prognostic class.

42. The software product of claim 41 , wherein the CUX1 expression profiles associated with the first prognostic class comprises an increase, with respect to the CUX1 expression profiles associated with the second prognostic class, in the gene expression level of at least one of the following gene AURKB, BUB1 , BUB3, BUBR1 , CENPE, MAD1 L1 , MAD2L1 , TTK, NEK2, KNTC1 , ORCL1 , ZW10, ZWILCH, CENPA, KIF2C, KIFC1 , CAMK2D, COG7, MFAP1 , NUDCD1 , RIN2, SF3B3, CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CTSL1 , MCM3, MCM7, POLA and POLA2.

43. The software product of claim 41 or 42, wherein the CUX1 expression profiles associated with the first prognostic class comprises a decrease, with respect to the CUX1 expression profiles associated with the second prognostic class, in the gene expression level of the CDKN1A gene.

44. The software product of claim 42 or 43, wherein the first prognostic class is associated with a poor prognosis.

45. The software product of any one of claims 41 to 44, wherein the CUX1 expression profiles associated with the second prognostic class comprises a decrease, with respect to the CUX1 expression profiles associated with the first prognostic class, in the gene expression level of at least one of the following gene AURKB, BUB1 , BUB3, BUBR1 , CENPE, MAD1 L1 , MAD2L1 , TTK, NEK2, KNTC1 , ORCL1 , ZW10, ZWILCH, CENPA, KIF2C, KIFC1 , CAMK2D, COG7, MFAP1 , NUDCD1 , RIN2, SF3B3, CCNA2, CCNE2, CDC25A, CDC45L, CDC7, CTSL1 , MCM3, MCM7, POLA and POLA2.

46. The software product of any one of claims 41 to 45, wherein the CUX1 expression profiles associated with the second prognostic class comprises an increase, with respect to the CUX1 expression profiles associated with the first prognostic class, in the gene expression level of the CDKN1A gene.

47. The software product of claim 45 or 46, wherein the second prognostic class is associated with a good prognosis.

48. The software product of any one of claims 37 to 47, wherein the CUX1 expression profiles comprise gene expression levels for the plurality of genes of the CUX1 signature gene set from an heterogeneous population of subjects having been diagnosed with cancer.

49. The software product of any one of claims 37 to 48, wherein the individualized gene expression level comprises a measure of the mRNA level of the plurality of genes.

50. The software product of any one of claim 37 to 49, wherein the comparison module is for applying a probability model to the normalized individualized gene expression level and generating an output, wherein the data derived from the set of CUX-1 expression profiles comprises a scoring scheme derived therefrom, and wherein comparing at least one parameter of the normalized individualized gene expression level comprises applying the output to the scoring scheme.

51. The software product of claim 50, wherein the probability model is a naive Bayes' modifier.

52. The software product of any one of claims 37 to 51 , wherein the known clinical outcomes comprises cancer recurrence, cancer metastasis, survival and/or usefulness of a chemotherapy treatment.

53. The software product of any one of claims 37 to 52, wherein the cancer is breast cancer.