US 2011 0038862A1 (19) United States (12) Patent Application Publication (10) Pub. No.: US 2011/0038862 A1 SOTRIOU et al. (43) Pub. Date: Feb. 17, 2011

(54) METHOD AND KIT FOR THE DETECTION Related U.S. Application Data OF GENES ASSOCATED WITH PIK3CA MUTATION AND INVOLVED IN PI3K/AKT (63) Continuation-in-part of application No. PCT/EP2009/ PATHWAY ACTIVATION IN THE 052043, filed on Feb. 20, 2009. ER-POSTITIVE AND HER2-POSITIVE (60) Provisional application No. 61/030,450, filed on Feb. SUBTYPES WITH CLINICAL IMPLICATIONS 21, 2008. Publication Classification (75) Inventors: CHRISTOS SOTIRIOU, UCCLE (51) Int. Cl. (BE); SHERENE LOI, A 6LX 39/395 (2006.01) BRUSSELS (BE); GRANT CI2O I/68 (2006.01) MCARTHUR, CAMBERWELL A63L/4535 (2006.01) (AU), BENJAMIN A6IP35/00 (2006.01) HAIBE-KAINS, BRUXELLES A63L/436 (2006.01) (BE) A63L/38 (2006.01) A6II 3/565 (2006.01) Correspondence Address: (52) U.S. Cl...... 424/133.1: 435/6: 514/324; 424/174.1; MERCHANT & GOULD PC 514/291; 514/651: 514/182 P.O. BOX 2903 (57) ABSTRACT MINNEAPOLIS, MN 55402-0903 (US) A method to determine the clinical outcome of breast tumour affecting a patient if treated with an antitumoural agent (73) Assignee: UNIVERSITE LIBRE DE against breast tumour. The method includes the step of assay BRUXELLES, BRUXELLES (BE) ing a sample of a breast tumour from the patient for an expression level of selected genes, by contacting mRNA (21) Appl. No.: 12/860,638 sequences from the cells of this breast tumour with a set of more than 3 nucleotide sequences related to human mutated (22) Filed: Aug. 20, 2010 PIK3CA. Patent Application Publication Feb. 17, 2011 Sheet 1 of 3 US 2011/0038862 A1 Figure 1 There is no COrrelation Higher level between PIK3CA Of PIK3CA andmutation luminal Subtypesstatus signature in HER2+ (lumA/B) and luminal AtumorS

PIK3CA Wild mutation 8B. A 62. 15 O B 58 13

Figure 2 PIK3CA Mutation status (ER+, tamoxifen-treated population)

o

o O CO - c. 's C E S c. 8 n CN O

P=3.8E-01 0 1 2 3 4 5, 6 7 8 9 10 NO. At Risk Time (years) Wild 120 119 110 103 92 79 68 64 52 48 39 mut 28 27 27 25 23 20 18 15 12 11 8 Patent Application Publication Feb. 17, 2011 Sheet 2 of 3 US 2011/0038862 A1 Figure 3 PIK3CA signature (Tamoxifen-treated population)

C Mutated v- signature od O p n1 co so 2 Ese St. Wild-type signature is P=3.8E-04

NO. At Risk 0 1 2 3 4,ime (years) 8 910 LOW 101 99 91 82 73 64 58 52 40 31 22 High 204201 192182 171155 141 127107 82.56

Figure 4 PIK3CA signature Luminal subtype (Tamoxifen-treated population)

C v- } Luminal A od (low proliferation) O E3P CO } Luminal B wild-type (High proliferation) E St. Signature Co signature —LOW/LOW th. S- || LOW/High High/LOW P-40E-07 - High/High 0 1 2 3 4 5, 6 7 8 9 10 NO. At Risk Time (years) LOW/LOW 37 37 36 34 32 31 28 26 19 14 8 LOW/High 64 63 56 49 42 34 31 27 22 1814 High/LOW 117 117 113 108 104 95 89 83 69 5235 High/High 87 85 80 75 68 61 53 45 39 3121 Patent Application Publication Feb. 17, 2011 Sheet 3 of 3 US 2011/0038862 A1 Figure 5 PIK3CA signature (ER+, HER2- untreated population)

o ' '. s Mutated g ti-s, signature s CO "Hisps is states sigs asse g s syss-s: widtype signature - & E NS S of Wild-type signature

QO

csO P-40E-02--.

NO. At RiskS388.S. &: . R& s: Time (yeafrS' sy X Low 14 140 12611102.90 85 78 7 6 48 High 284 278 263251 235220 199 176146 120 104

Figure 6 PIK3CA signature, Luminal subtypes (ER+, HER2- untreated population) se esOe- T } (oppreslieListinia Etition) A Ée itas s (HighgiQfeasierton)Luria B

O 3. - LOW/LLOWILOW is signature “. . ." LOW/High ||---High/Low s P25-06 - - High/High 0 1 2 3 4 56 7 8 9 10 Nasr. . . . The years) LOW/LOW 48 48 48 46 44 41 40 35 34 29 20 Low/High 93 92 79 66 59 50 46 44 38 3328 High/Low 164162 159 154. 146 144 130 115 93 75 68 High/High 120 117 105 98 90 76 70 62 53 47 36 US 2011/0038862 A1 Feb. 17, 2011

METHOD AND KIT FOR THE DETECTION activation of AKT1 and downstream signaling. The clinical OF GENES ASSOCATED WITH PIK3CA relevance of this mutation is unknown. MUTATION AND INVOLVED IN PI3K/AKT 0008 Given the complexity of PI3K signaling, it is impor PATHWAY ACTIVATION IN THE tant to have molecular markers that can predict for prognosis ER-POSTITIVE AND HER2-POSITIVE and therapeutic response for incorporation into future breast SUBTYPES WITH CLINICAL IMPLICATIONS cancer clinical trials with compounds that act on this pathway. 0009. As the Kaplan-Meier analysis of the PIK3CA muta FIELD OF THE INVENTION tion versus the wild-type patients did not reveal any statisti 0001. The present invention is related to a new detection cally significant differences in prognosis, mutation status method and a new detection kit of genes associated with alone may not be a sensitive marker of significant activation PIK3CA mutation(s) and involved in PIK3/AKT pathway of the PI3K/AKT pathway that would affect tumor progres activation in the (luminal-B) ER-positive or HER2 positive Sion. Other downstream interactions of an extra oncogenic Subtypes with clinical implications. The present detection “hit” may be required. method and kit have a predictive clinical outcome (survival outcome) and could be therefore used for identifying if a AIMS OF THE INVENTION patient from which this tumour sample is obtained could be 0010. A first aim of the present invention is to propose a Submitted to a specific antitumoural treatment, especially to a new detection method and new detection means (kit) of tamoxifen or Herceptin treatment (or not). The present inven improving clinical outcome (especially survival outcome) of tion is also directed to the therapeutic application of a class of a human patient following application of this detection active compounds to be applied efficiently to the subtype of method upon a tumour sample obtained from this patient and cancer detected by this method. the defining (selecting among known treatments) the most effective treatment that could be applied to this patient. BACKGROUND OF THE INVENTION 0011. In particular, the present invention aims to provide 0002 Breast cancer (BC) may be subdivided into sub such detection method and kit which allows a better discrimi groups depending on expression profile of several genes and/ nation of clinical outcome of patients with ER-positive and/or or protein. HER2-positive sub-types tumour samples and to identify 0003. Her2 overexpression in tumours results into a worse which type of patients should receive (be prescribed) an anti prognosis. BC patients having Her2 positive status are pref oestrogen, more preferably a tamoxifen or aromatase inhibi erably not treated with anti oestrogens, but with anti Her2 tor therapy or Herceptin related therapy or hormone/chemo, drugs, such as Trastuzumab (Herceptin). radio- or immunotherapy. 0004 For the BC patients whose tumours express ER receptor (ER group) but do not over express Her2, the over SUMMARY OF THE INVENTION expression of several genes related to proliferation results into the classification into the luminal B subgroup, with a 0012. The present invention is related to a method and a kit worse prognosis. These luminal B patients are preferably not for a detection of mutated PIK3CA gene and/or mutated treated with anti-oestrogens, but with more aggressive treat AKT-1 gene and/or gene(s) involved PIK3/AKT pathway ments (chemotherapy). activation(s) in (especially in high proliferative luminal-B) 0005 Deregulated phosphatidylinositol 3-kinase (PI3K)- ER-positive BC or HER2-positive subtype tumour sample AKT signaling has been implicated in many hallmarks of as described in the enclosed set of claims. carcinogenesis as the pathway influences multiple aspects of 0013 The inventors have investigated frequency, pheno cell physiology. Many genomic alterations act on this path type and clinical relevance associated with PIK3CA or way, activating its signaling activity, which contributes to AKT-1 mutations in a large homogenous data set of ER+ tumor progression, metastases and resistance to treatment. (BC), tamoxifen-only treated breast tumours. The inventors 0006 PI3Ks are heterodimeric lipid kinases for which the have examined the associated gene expression profiles to p110C. catalytic and regulatory p85 subunits are encoded by further understand the biology associated with PIK3CA or separate genes. In breast cancer, mutations of the PIK3CA AKT-1 mutations when they activate the PI3K/AKT pathway. gene, which codes for the p110C. catalytic subunit, has been 0014. The inventors have selected a preferred gene set found in 18-40% of human cancers, which makes it one of the related to mutated PIK3CA and/or AKT, comprising the 81 most common genetic changes in breast cancer beside p53 genes of Table 2a or the 65 genes of Table 2b, more preferably mutations and HER2 amplification. Expression of p110C. the 38 genes of Table 4. mutants in human mammary epithelial cells induces multiple 0015 The inventors have selected an alternative preferred phenotypic alterations characteristic of breast tumor cells and gene set related to mutated PIK3CA and/or AKT, comprising in vivo studies with cells expressing PIK3CA mutants result the 278 genes of Table 5 or, more preferably, the 14 genes of in a more active PI3K pathway and induction of tumors. The Table 6. lack of a homogeneous population makes it difficult to inves 0016. The inventors further found other (isolated) genes tigate the prognostic or predictive effect of PIK3CA muta associated with mutated PIK3CA and/or Akt mutations, tions in breast cancer. Extensive cross-talk at multiple levels advantageously, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 or all the 12 with other pathways both upstream and downstream of PI3K genes selected from the group consisting of the following also makes the exact role of PIK3CA mutations in breast genes: PML (Entrez gene ID 5371), PP2A (Entrez gene ID cancer difficult to elucidate. 5523, 5525, 5526), IRS2 (Entrez gene ID 8660), PIK3R1 0007 Mutations in the AKT1 pleckstrin homology (Entrez gene ID 5295), ESR1 (Entrez gene ID 2099), domain (PHD) reported in breast cancer at a frequency of 8% FOXO3A (Entrez gene ID 2309), P21 (PAK2) (Entrez gene may result in PI3K-independent membrane recruitment and ID 5062), RPS6K (Entrez gene ID6198), EIF4E (Entrez gene US 2011/0038862 A1 Feb. 17, 2011

ID 1977), RHEB (Entrez gene ID 6009), P27 (Entrez gene ID (0026. By HER2-positive, it is meant HER2 over expres 1785), PI3K (Entrez gene ID 18708) and possibly their iso sion. This status is measured at gene level {i.e. amplifica forms or variants. tion, at mRNA level or at the protein level and wherein this 0017. The use of the gene set(s) or of the method according overexpression in BC tumours results into a worse prognosis. to the invention allows an efficient establishment of PIK3CA 0027 BC patients having HER2 positive status are gener mutated signature and the use of these gene sets or of the ally not treated with anti oestrogens as they are relatively method according to the invention allows accurate and sen resistant to their effects. sitive determination of the state of activation of the PI3K/ 0028 By luminal B, it is meant a highly proliferative AKT pathway, similar to that induced by a PIK3CA or a tumour (ER+ and Her2-), preferably demonstrated using AKT1 mutation. published prognostic gene expression profiles Such as GGI. 0018. The present invention is further related to a method Oncotype DX, Intrinsic gene set, Amsterdam 70-gene signa to determine the clinical outcome (preferably the survival ture, Rotterdam 76-gene signature, Wound-response signa outcome) of a (breast) tumour affecting a patient, if this ture or having KI67 high expression (>5%). patient is treated with an anti oestrogen agent against this 0029. Luminal B represents a worse prognosis. These (breast) tumour, this method comprising the step of assaying luminal B patients are generally not treated with anti-Oestro a sample of this breast tumour obtained from this patient for gens alone. an expression level of one or more gene(s) or synthesis of 0030. In the method of the invention, the step of assaying corresponding protein(s) encoded by these gene(s), prefer for the expression level of one or more gene(s), protein(s) or ably more than 3,4,5,6,7,8,9, 10, 11, 12, 13, 14, 15, 16, 17, portions thereof comprises a detection of target nucleic acids 18, 19 or 20 genes (including isoforms and variants) or pro prepared by a mRNA amplification from the sample, by a teins selected from the Table 2a, from mutated PIK3CA genes detection of (the amplified) target nucleic acids from the and proteins and/or from mutated AKT-1 genes or proteins sample, by a quantitative PCR, (preferably a qRT PCR), by a involved in the PI3K/AKT pathway. detection of corresponding target proteins or their fragments 0019. By one or more gene(s), it is meant 4,5,6,7,8,9, 10, through specific binding with corresponding capture antibod 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75 and every ies or similar capture molecules (nanobodies, specific Hyper number till 81 genes mentioned and identified in these tables variable portions of antibodies, etc), preferably such detec 2a, 2b or 4, with every possible combination of selected genes tion of target proteins is obtained in patient blood or in breast (or corresponding proteins and fragments thereof). cancer epithelial cells enriched from patient blood. 0020. The present invention is also related to a method to 0031 Advantageously, in the method according to the determine a prognosis and/or prediction of a response of a invention, the detected target genes or proteins are selected patient having a (breast) tumour to a treatment, if treated with from the group consisting of more than 3 genes (9, or 15) or an anti oestrogen agent against this (breast) tumour, and com corresponding proteins of the genes presented in Table 2a, prising the step of assaying for an expression level of one or Table 2b or Table 4 (or Table 5 or 6) or genes involved in more (preferably more than 3, 9, 10, 19 or all of the gene(s) or PI3K/AKT pathway activation. protein(s) of (Table 2a or Table 2b or Table 4 or Table 5 or of 0032. Advantageously, at least one gene selected from the table 6), of mutated PIK3CA/AKT-1 genes or proteins or of group comprising PFN2, ORC5L, MYC, E2F5, ARPP19 and genes or proteins involved in the PI3K/AKT pathway activa MNAT1 (the first group of these genes being under expressed tion from a breast cancer cells obtained from a breast cancer in PIK3CA mutated subjects) and at least one gene selected tumour sample from the patient. from the group comprising SCGB2A2, NOTCH2, TNIK, 0021. The method may also comprise a step of determin GOLPH2, ARHGDIB, GALNT2, SPTLC2 and SCGDAD2 ing a clinical outcome, preferably a Survival outcome corre (these genes of the second group being over expressed in lated to an assaying step of the patient and/or a step of select PIK3CA mutated subjects), wherein at least three gene are ing an antitumoural compound, (especially an anti oestrogen selected. agent), which could be administrated to this patient. 0033. In the method according to the invention, the sample 0022. In the method according to the invention, the expres could be obtained by various techniques, preferably by sion levels are indicative of probabilities of recurrence (or biopsy, more preferably by a minimally invasive technique or relapsing) of cancer, possibly through metastasis. selected from core biopsy, excisional biopsy, ductal lavage 0023. By anti oestrogen, it is meant the administration of a simple, a fine needle aspiration sample or from cells micro selective oestrogen receptor modulator (SERM), a selective dissected from the sample. oestrogen receptor down regulator (SERD), tamoxifen, ral 0034. Another aspect of the present invention is related to oxifene, faslodex, or a mixture thereof. a set of capture nucleotide sequences comprising one or more 0024. In the method according to the invention, the anti strand(s) of sufficient length of about 15 to about 250 or more oestrogen agent against breast cancer is also any compound nucleotides for obtaining an efficient and specific hybridiza that could be used in hormonal therapy of cancer, preferably tion of corresponding target nucleotide sequences (RNA by administration of an efficient compound selected from the sequences) of human mutated PIK3CA/AKT-1 sequences or group consisting of a selective oestrogen receptor modulator sequences involved in PI3K/AKT pathway activation, espe (SERM), a selective oestrogen receptor down regulator cially one or more sequence(s) of Table 2a, Table 2b or Table (SERD), preferably tamoxifen, raloxifene, faslodex or a mix 4 or capture molecules (antibodies, etc) that may bind spe ture thereof, GnRH analog or a aromatase inhibitor (AI) such cifically these corresponding target proteins, wherein at least as Letrozole, Anastrozle or Exemestane. a portion of this set is hybridized to nucleotides quantitatively 0025. In the method according to the invention, the breast amplified from RNA sequences of breast cells. tumour is a HER2+ OR ER+(BC) tumour, especially a lumi 0035 Advantageously, the set of capture nucleotide nal-B ER+ tumour evidenced by high expression of prolif sequences or capture molecules that may bind the proteins is erative genes. immobilised on a solid Support Surface as a microarray. US 2011/0038862 A1 Feb. 17, 2011

0036 Preferably, (specific) hybridisation between capture ETAL. (2006).J. Natl. Cancer Inst., 98, pages 262-272, CHANG and target sequences is obtained under Stringent conditions ET AL. (2004) PLOS BIOL 2: E7, SORLIE ET AL. (2003) (under conditions well-known to the person skilled in the art, PNAS100, pages 8418-8423, IVSHINA ET AL. (2006) Cancer for instance, the one described by SAMBROOK ET AL) which Res. 66, pages 10292-10301, PAWITANET AL., (2005) Breast provides Sufficient binding efficiency of the target sequences Cancer Res; 7: R953-964, FARMER ET AL. (2005) Oncogene on their specific capture probes to the detected with no or very 24, pages 4660-4671, WHITFIELD ET AL. (2006) Nat. Rev. low, preferably lower than 5% and even lower than 1%, cross Cancer 6, pages 99-106, and the expression profiling protein hybridization on non-related target capture probes. used in breast cancer as described in document WO2005/ 0037 Advantageously, the hybridized target nucleotide 07 1419, the expression profiling protein/gene described in sequences are previously amplified from RNA sequences of the document WO2005/021788. The set of the invention may breast cells. also comprise or consist of capture nucleotide sequences or 0038 Advantageously, the breast tumour cell is ER+(BC) capture molecules that can bind specifically the proteins or HER2+ tumour cells. encoded by these genes of these nucleotide or proteins sets. 0039. A further aspect of the present invention is related to 0046. Another aspect of the present invention is related to this set, wherein the mutated genes comprise mutation(s) in a kit or device, preferably a computerised system comprising the PIK3CA mutated sequences are PIK3CA mutations a bioassay module configured for detecting gene expression selected from a group consisting of A3140G, A3150T in (or protein synthesis) from a tumour sample which is based EXON20 of the genomic DNA and/or G1633A, G1624A or upon the genes set including molecules that bind specifically G1634A in EXONS of the genomic DNA and corresponding the proteins encoded by these genes set according to the nucleotide of messenger RNA sequence. invention and a processor module configured to calculate 0040. More preferably, the set according to the invention expression (over or under expression) of these genes and/or comprises more than 3, 5, 10, 15 or all the sequences selected synthesis of corresponding encoded proteins and to generate from the group consisting of the sequences above-described a clinical outcome, preferably a Survival outcome and a risk and present in a Table 2a, Table 2b or Table 4 or present in assessment for a tumour sample (risk assessment to develop a Tables 5 or 6. malignant tumour) or Susceptibility that a patient from which 0041 Advantageously, the set according to the invention this tumour sample has been obtained could be treated by may also further comprise one or more genes selected from efficient therapeutic treatment, especially a treatment based the group consisting of IL17BR (55540), CHDH (55349), upon the administration of an anti oestrogen, preferably QPRT (23475), HOXB13 (10481) genes wherein over tamoxifen. Advantageously, the generated set of genes (and expression of IL17BR and/or CHDH sequences and/or under proteins) according to the invention may also provide a detec expression of QPRT and/or HOXN13 sequences are negative tion of tumour correlated with a Herceptin (Trastuzumab prognosis of selective oestrogen receptor modulator SERM Genentech, California USA) resistance in HER2 tumour treatment of oestrogen receptor positive patient. samples. Therefore, the method of the invention could be used 0042 Another aspect of the invention is identifying a sub in patients presenting a HER2 subtype combined with this group of Her2 positive breast cancer (patients) having the resistance for selecting an appropriate treatment that is not signature of mutated PIK3CA (preferably identified by the based upon administration of Herceptin to the patient method of the invention), that would benefit from anti oestro 0047. The inventors have measured that the signature of gen treatment the invention (mutated signature of PIK3CA and/or Akt) pre 0043. Another aspect of the invention is therefore anti dicts for a better outcome of anti-oestrogen treatments of BC oestrogen for use in the treatment of Her2 positive breast patients. cancer (patients) having the signature of mutated PIK3CA 0048. The inventors have measured that the signature of and being preferably identified by the method of the inven the invention (mutated signature of PIK3CA and/or Akt) pre tion. dicts for a better outcome of anti-oestrogen treatments in 0044 Another aspect of the invention is anti oestrogen for luminal BBC patients. use in the treatment of ER positive, luminal B breast cancer 0049. The inventors have measured that luminal BBC (patients) having the signature of mutated PIK3CA. patients having the mutated PIK3CA and/or Akt signature 0045. The method and kit according to the invention could according to the invention benefit from anti-oestrogen treat be also combined with one or more detection method, kit and mentS. tools already described in the state of the art especially in the document WO 2006/119593, especially prognostic means 0050. Surprisingly, the inventors have observed that the (signature) or gene list (gene set) which could be used for an mutated signature of PIK3CA and/or Akt in ER+ and HER2+ efficient prognosis (prognostic) of cancer in ER+ patient, but BC patients predicts for a better outcome. also possibly in ER- patient such as the one described by 0051 Surprisingly, the inventors have observed that the WANG ET AL. (2005), LANCET 365, page 671-679, VAN T mutated signature of PIK3CA and/or Akt in ER+ (including VEERET AL. (2002), Nature 415, pages 530-536, PAIKET AL. for luminal B patients) and HER2 positive BC patients pre (2004) New ENGL. J. MED. 351, pages 2817-2826, dicts for a better response to anti oestrogens, possibly in TESCHENDORFET AL. (2006). Genome Biol. 7, R101 206, addition for Her2 positive BC patients of anti HER2 drugs, VAN DE VIJVERET AL. (2002) New ENGL. J. MED. 347, Such as Herceptin (trastuzumab). pages 1999-2009, PEROU ET AL. (2000) Nature, 406, pages 0052. The inventors have therefore treated with anti 747-752, SOTIRIOU ET AL (2003) PNAS100, pages 10393 oestrogen BC patients with the PIK3CA mutated signature 10398, SORLIEET AL. (2001) PNAS 98, pages 10869-10874, identified according to the method of the invention. MILLER ET AL. (2005) PNAS102, pages 13550-13555, 0053. The BC patients (with the PIK3CA mutated signa NADERIET AL. (2007) Oncogene 26, pages 1507-1516, PAIK ture) treated (according to the invention) with anti oestrogen ET AL. (2006) J. Clin. Oncol. 24, pages 3726-3734, SOTIRIOU are preferably ER positive. US 2011/0038862 A1 Feb. 17, 2011

0054) The BC patients (with the PIK3CA mutated signa 0065 FIG. 2 presents a Kaplan-Meier analysis revealed ture) treated (according to the invention) with anti oestrogen that PIK3CA mutations were surprisingly not significantly are luminal B ER positive. associated with a better or worse prognosis compared with 0055. The BC (with the PIK3CA mutated signature) those tumours without a mutation. Analyses of exon 9 and 20 treated (according to the invention) with anti oestrogen are mutations separately did not change this result. Univariate Her2 positive. Survival analysis confirmed no significant correlation 0056 Conversely, the inventors have observed that che between PIK3CA mutations and prognosis. motherapy is less effective in BC patients having the gene 0.066 FIG. 3 shows surprisingly that higher expression signature of mutated PIK3CA and/or Akt, according to the levels of the PIK3CA signature were associated with statis present invention. tically better clinical outcome when tamoxifen only treated 0057 The inventors have further observed that radio patient were considered. therapy is less effective in BC patients having the gene sig 0067 FIG. 4 shows the association between the expres nature of mutated PIK3CA and/or Akt, according to the sion levels of the PIK3CA signature and clinical outcome was present invention. better evidenced within the highly proliferative high risk 0058. The inventors have further observed that PI3kinase/ luminal B tumours. Tumours with higher expression levels of AKT/mTOR pathway inhibitors are less effective in BC the signature benefit better from tamoxifen than those with patients having the genesignature of mutated PIK3CA and/or lower expression levels. Akt, according to the present invention. 0068 FIG. 5 presents results that are similar in the breast 0059. The method, gene(protein) set and kit (or tools) cancer patients who had received no systemic treatment and according to the invention could be also used for selecting an that the group of patients with higher expression levels of the adequate therapeutic treatment to get apply to the patient PIK3CA mutation signature had better clinical outcome than from which the tumour sample has been obtained, especially those with lower expression levels as shown by this KM curve selecting an appropriate dose and/or schedule of chemo analysis. therapeutic and/or bio-pharmaceutical and/or targeted agent. 0069 FIG. 6 shows the association between the expres 0060. This treatment could be based upon administration sion levels of the PIK3CA signature and clinical outcome was of anti oestrogens, taxanes, anthracyclines, CHOP or other only seen within the highly proliferative high risk luminal B drugs like velcade, fluorouracil, uracil, vinblastine, gemcit tumours. Surprisingly, tumours with higher expression levels abine, methotrexate, goserelin, irinotecan, thiotepa, topote of the signature showed better clinical outcome than with can, toremifene, anti-EGFR, anti-HER2/neu, anti-VEGF, lower expression levels. RTK-inhibitor, anti-VEGFR, GRH, anti-EGFR/VEGF, HER2/neu, EGF-R or anti-HER2. MATERIALS AND METHODS 0061 The method, set and kit according to the invention could be also used in combination with a method for control Tumour Samples Screened for Mutation Status ling the efficiency of the treated method or an active com 0070 Primary breast cancer tumour samples from a pre pound in cancer therapy. Indeed, the method, set and kit (or viously described “tamoxifen-only treated data set (Loi, tools) according to the invention that apply for an efficient 2007; J. Clin Oncol., 25, 1239-46:) were collected for DNA prognostic of cancer in various breast cancer types could also extraction and mutation analysis. The inventors obtained be used for an efficient monitoring of the treatment applying DNA from 173 samples for PIK3CA mutation sequencing to the patient Suffering from this cancer. and from 131 samples for AKT1 mutation analysis. The 0062. The method according to the invention may require median follow-up of these samples was 9.0yrs (range: 8.2-9.8 a first prognostic step which is applied to the patient before yrs), with 45 (28%) distant metastatic events. Submitting the patient to a treatment or to a second diagnosis step following this treatment. Screening for Mutations 0063. This method could be applied several times (two PIK3CA time, three times, four times, five times, etc) to the mammal Subject (human patient) during the treatment or during the (0071. The vast majority (>85%) of PIK3CA mutations monitoring of the treatment several weeks (one week, two reported in human breast cancers are missense mutations weeks, three weeks, four weeks, etc) or months (one month, clustering in exons 9 (E545K) and 20 (H1047R). (SAAL, two months, three months, etc) after the end of the treatment 2005) These exons we screened for mutations using single to reveal if a modification of gene expression or protein Syn strand conformation polymorphism (SSCP). The PIK3CA thesis in a sample Subject is obtained following the treatment. primer sets are as follows:

DETAILED DESCRIPTION OF THE INVENTION Exon 9 : Figure Legends Forward: {6FAMTGAAAATGTATTTGCTTTTTCTGT; SEO ID N 3 0.064 FIG. 1 shows that there was no correlation between Reverse: {VICTGTAAATTCTGCTTTATTTATTCC; SEO ID N 4 PIK3CA mutations and the luminal subtypes defined using Exon 2 O: the GG values, evidencing that PIK3CA mutations are not Forward: (NEDTCCAAACTGACCAAACTGTTCTT; SEO ID N 5 associated with either molecular ER-positive subtypes. In contrast, the inventors did surprisingly find an association Reverse: {PETCCAGAGTGAGCTTTCATTTTCTC. SEO ID N 6 between the expression levels of the associated PI3K muta Primers labeled with 5' fluorescence (Applied Biosystems). tion signature and the molecular Subtypes. Indeed, higher PCR was carried out with 10ng of genomic DNA in a reaction levels of PI3Kassociated mutation signature were associated volume of 10 uL, with the inclusion of 0.25 units Hot Star with the HER2 and luminal A low proliferate subtypes. Taq DNA polymerase (QIAGEN, Valencia, Calif.). After an US 2011/0038862 A1 Feb. 17, 2011

initial denaturation step of 95°C. for 10 minutes, a “touch samples also treated with tamoxifen monotherapy with cor down' program was used consisting of 2 cycles of amplifi responding Affymetrix gene expression data was also used cation at annealing temperatures of 63°C. to 59°C.; followed for the survival analysis. by 30 amplification cycles at an annealing temperature of 58° 0075 For the survival analysis using breast cancer C. and a final extension cycle of 72°C. for 5 minutes. Samples samples which had received no systemic treatment (hereby were prepared for single-strand conformational polymor referred to as the “untreated dataset), gene expression data phism (SSCP) analysis using the ABI-3130 automated capil was used from datasets described in DESMEDT ET AL., 2007, lary sequencer. The sample, size standard and Hi-DiTM For Clin Cancer Res., 13, 3207-14; WANG ET AL., 2005 and VAN mamide was mixed in each well of sample plate. The PCR DE VIJVERET AL., 2002. product was denatured for 3 minutes at 95°C. and then cooled 0076. The inventors used the normalized data (log2 inten on ice for minutes to avoid re-annealing of the complemen sity in single-channel platforms or log 2 ratio in dual-channel tary strands before being run using the Genemapper fragment platforms) as published by the original studies. Hybridization analysis module on ABI 3130 genetic analyzer. Labelled probes were mapped to Entrez, GeneID. When multiple fragments are visualized on an Applied Biosystems DNA probes were mapped to the same GeneID, the one with the analyzer. The genescan LIZR size standard was used in all highest variance in a particular dataset was selected to repre samples as an internal ladder to align data from different sent the GeneD. capillaries and eliminate capillary-to-capillary or run-to-run 0077. Data analyses between performed using BRB variability. Cases showing aberrant peak shifts by SSCP were ArrayTools version 3.5 developed by Dr. Richard Simon and reamplified and sequenced directly with the BigDye termi Amy Peng Lam (http://linus.nci.nih.gov/BRB-ArrayTools. nator method (Applied Biosystems; Warrington, United html). Differential gene expression between PIK3CA muta Kingdom or Forster City, Calif.) on an auto sequencer (ABI tion carriers versus non-mutation carriers was performed PRISM 3100). using the “class comparison” tool. A two sample t-test was used at a significance value of 0.001 and statistical signifi AKT1 cance of the gene expression profiles between the classes was 0072 The mutation screening for AKT1 exon 4 was car tested by 1000 permutations of the class labels. For this analy ried by High-Resolution Melting (HRM) analysis. These sis, of the 173 sequenced for PIK3CA mutations, 161 had exons were screened for mutations using capillary electro corresponding microarray data. Only those samples with phoresis single strand conformation polymorphism exon 20 mutations were used in the class comparison analy (CESSCP) S1S The AKT1 primer sets are as follows: 0078. The inventors have developed an index called the PIK3CA index that could measure the similarity between the expression profile of any given tumour sample and the PI3K/ Exon 4 AKT pathway activation by breast cancers with a PIK3CA Forward: AGGGTCTGACCCCTAGAGATG SEO ID N 1 mutation. The signature score is the Sum of the expression of the genes up-regulated in the mutated tumours minus the Sum Rewerse: AGAGGGCTCCAGCCAACC SEO ID N 2 of the expression of the genes up-regulated in the wild type 0073 PCR was carried out with 15 ng of genomic DNA in tumourS. a reaction volume of 10 uL, including 5 uL of the High PIK3CA index: Resolution Melting Master (Roche) for amplification and detection of heteroduplex regions in PCR amplicons. The High-resolution melting master contains a dye, Resolight XX, -Xx, that enables detection of double-stranded DNA by fluores iP jew cence, monitoring formation of amplicons during PCR cycling, and melt curve analysis. Samples were carried out in duplicate, in a 96-well plate. After an initial denaturation step where P is the set of genes up-regulated in the mutated of 95°C. for 15 minutes, a touch-down program was used tumours and N is the set of genes up-regulated in the wildtype consisting of 2 cycles of amplification at annealing tempera tumourS. tures of 63° C. to 59° C.; followed by 55 amplification cycles 007.9 The weight of the genes was either +1 or -1 depend at an annealing temperature of 58° C. and a final Melt from ing on their association with PIK3CA mutation status. As a 70° C. to 95° C. PCR cycling and HRM analysis was per result, the index was not optimized to specifically identify formed on the Light Cycler 480 (Roche Diagnostics: F. Hoff mutation positive samples. Advantageously, no clinical out mann-La Roche Ltd.). LightCycler480 Software (v1.3.0. come data was used to identify the genes used in the PIK3CA 0705) was used to analyse results. Samples with variations in index hence the inventors were able to use the tamoxifen DNA sequence are distinguished by discrepancies in melting treated dataset for the survival analyses. curve shape. Samples showing deviations in melt curve were treated with ExoSapIT (GE Healthcare, Buckinghamshire, Interaction Networks and Functional Analysis England) according to the manufacturer's instructions and 0080 Gene oncology and gene interaction analyses were sequenced directly with the BigDye terminator method (Ap carried out using Ingenuity Pathways Analysis (IPA) version plied Biosystems; Warrington, United Kingdom or Forster 3.0 (http://www.ingenuity.com). The gene lists containing the City, Calif.) on an auto sequencer (ABI PRISM 3100). Affymetrix probe, as well as the fold change was inputted into Microarray Analysis IPA and mapped to the corresponding gene object in the database. These focus genes were then used to generate the 0074 Part of the tamoxifen-treated dataset has previously networks based on the curated list of molecular interactions in been described (Loi, 2007). Another 77 primary breast cancer the IPA database. Significance of enrichment is determined US 2011/0038862 A1 Feb. 17, 2011 by a right-tailed Fisher's exact test, using a list of all the genes molecular subtype. There was no significant association with on the array as a reference set. PIK3CA mutation and ERBB2 or PTEN over expression (p=0.4 and 0.1 respectively). Statistical Analysis Microarray Analysis 0081 Statistical analysis was performed using the SPSS 0085. The lack of correlation of PIK3CA mutations with statistical software package (SPSS Inc. Chicago, Ill.) version prognosis with other studies may be due to the unique features 13.0. The chi-square test was used to evaluate for possible of this patient dataset (all ER+ tumours), or that PIK3CA associations between mutation status and the various clinico mutations may predict favourably for tamoxifen treatment. pathological factors. In the univariate and multivariate Cox Another possibility is that PIK3CA mutations alone are not regression, the histologic grade (grade 1 and. 2 VS. 3), tumour prognostic in breast cancer but may need to interact with other size (s2 cm vs. D2 cm), nodal status (positive vs. negative) genetic changes in cancer cells to affect prognosis or other and age (s.50 vs. 50 yrs) were treated as binary variables. properties of the cancer cells. The corresponding gene The PIK3CA gene signature was treated as a continuous expression data was therefore examined with the aim to gain variable. Survival outcomes were also estimated with the further insight into the biology of activation of the PI3K/AKT Kaplan-Meier method and compared using the log-rank sta pathway through PIK3CA mutations in ER+BC. tistic. The PIK3CA gene signature was dichotomized to form two groups for the illustration by Kaplan-Meier survival PIK3CA Mutation-Positive Associated Differential Gene curves using a cut-off at 66:33% as survival of PIK3CA-GS Expression Signature and Interaction Networks highest two tertiles of dataset were similar. The group with the I0086 Firstly, those breast cancers harboring PIK3CA higher and lower expression of the PIK3CA-GS is referred to exon 20 mutations with available transcriptional profiles as “mt-like' and “wt-like' respectively. (n=28) were compared to wild type samples (n=120). Using a 0082 Breast cancer molecular subgroups were defined Supervised analysis, 81 probe sets were found to be signifi using a previously reported method of WIRAPATI ETAL, 2008, cantly differentially expressed at the nominated t-test level Breast cancer cell, 10, R65. The gene expression grade index (see Table 2a or the refined tables 2b and Table 4; the genes of (GGI) was used as a quantification of the expression of pro Tables 2b and of table 4 are the most suitable genes selected liferation genes (SOTRIOUETAL, 2006). For ER+BC subtypes, from the Table 2a). The statistical significance of the class proliferation expression was used to classify tumours repre label permutation was significant at a p value of 0.03, con senting the luminal-A and -B molecular subgroups described firming that the gene expression profiles were significantly by PEROU ET AL., 2000, into luminal low-risk and luminal different between classes. Results were similarif all mutation high-risk subgroups respectively (Loi, 2007). samples (exon 9 and 20) were used. I0087. The inventors then performed another statistical analysis by combining the extent of up- or down-regulation Results (>1.3) and the statistical significance (p<0.05) of a selected Frequency and Location of Mutations gene in mutated cells (Table 5) and further deduce a most preferred signature (Table 6 representing the genes present in 0083 Mutational analysis of the PIK3CA gene was per both Tables 4 and 5). formed in 173 primary ER+BCs. A total of 46 mutations were I0088. The molecular interactions of these differentially found (26%). The majority (71%) of these mutations were expressed genes were examined using Ingenuity Pathways located on exon 20. One sample had mutations in both exon 9 Analysis (IPA). According to Ingenuity Pathways Analysis and 20. Twenty-nine (91%) of mutations on exon 20 were (IPA), the top canonical pathway was insulin receptor signal H1047R substitutions, its high frequency consistent with pre ling (p=0.002) and the top function was protein synthesis vious reports (Table 1). Five AKT1 mutations were found in (p=0.0005) the 131 samples that were able to be tested (3.8%). All 5 0089. Overall, these data were consistent with the notion mutations were E17K substitutions, were found in PIK3CA that the PI3K pathway is activated by PIK3CA mutations and wild type samples. PIK3CA mutations in breast cancer are associated with a Mutations and Correlation with Clinico-Pathological Fea distinct molecular profile. tures Activation of the PI3K/Akt Pathway Due to PIK3Ca Muta 0084. There were no significant correlations between tions Predicts Outcome of ER+Bc Treated with Adjuvant PIK3CA mutations and other important clinico-pathologic Tamoxifen. features, except a borderline association with tumour size 0090. As the molecular profile of the breast cancer (p=0.057) (Table 1). Similarly, AKT1 mutations were not samples with a PIK3CA mutation seemed to represent acti associated with any clinical factors, though the Small num vation of the PI3K/AKT pathway, the inventors went on to bers make this result difficult to interpret. Kaplan-Meier create an index using the differentially expressed genes that analysis revealed that mutations of PIK3CA (FIG. 2), AKT1 would be able to quantify the extent of activation of the or both were not significantly correlated with prognosis com pathway in a given tumour sample. In this way, the inventors pared with those tumours without a mutation. PIK3CA exon were able to encapsulate clinically relevant activation of this 9 and 20 mutations were examined combined and separately pathway through other mechanisms as well as PIK3CA muta and results were similar. Univariate Survival analysis con tions. firmed no significant correlation between both mutations and 0091. The inventors further found the genesignature of the prognosis. There was no correlation between either mutation differentially expressed genes corresponding to mutated and gene expression grade (GGI) values, suggesting that PIK3CA and/or AKT in (PIK3CA and/or AKT) wit patients these mutations are not associated with a particular ER+ and conclude that the mutated signature they evidenced rep US 2011/0038862 A1 Feb. 17, 2011

resents a more physiological read out than the qualitative PHD mutations have been reported to activate the PI3K/AKT identification of a mutation in PIK3CA and/or AKT gene(s). pathway. The incidence here was lower than previously 0092 Firstly, the inventors looked at correlation between reported (3.8% vs. 8%), which makes the results from further the PIK3CA mutated signature and subtypes of cancers. analyses difficult to interpret. Interestingly, all AKT1 muta 0093. The inventors found no correlation with luminal tions occurred in PIK3CA wild type samples. However, nei status of ER+ Breast cancers (FIG. 1). ther PIK3CA and/or AKT1 PHD mutations perse were asso 0094. The inventors found a positive correlation between ciated with prognosis in our ER+BC dataset. These data does Her2 positive status and PIK3CA mutated signature (FIG. 1). not support a recent study which reported that in breast can 0095. The inventors looked at prognostic ability of cer, PIK3CA mutations located on exon 9 conveyed a worse PIK3CA index in the dataset of ER+, (HER2-negative) prognosis that those located on exon 20, though the incidence tamoxifen-treated patients. Surprisingly, an increasing of exon 9 mutations here was less. Given the conflicting data expression level of index was associated with a significantly in the literature and the low incidence of AKT1 mutations, it better outcome in these patients (log rank p value: 0.004— seems it may be impossible to use PIK3CA and AKT1 muta FIG.3). These results were similar in the 405 available patient samples which had not received any systemic treatment (p tion status alone to predict prognosis and treatment response. value: 0.04 FIG. 5). The inventors then went on to look at AKT1 PHD mutations may even predict for a favourable the relevance of PI3K/AKT activation in the 2 molecular prognosis given that cell line and animal models Suggest that subtypes of ER+BC. Interestingly, in the tamoxifen-treated unlike AKT2, AKT1 does not influence invasion and group, the PIK3CA index was able to separate the luminal-B, metaStaSeS. but not the luminal-A group of breast cancers into two prog 0102) However, the PI3K/AKT pathway is complex, nostically distinct groups (p=0.02; FIG. 6). This phenomenon impacting on multiple areas of cell physiology, hence activa was also observed in the untreated dataset (FIG. 4). However, tion of the pathway by different mechanisms is likely to it noteworthy that in thetamoxifen dataset, the outcome of the trigger different cellular functions. Using the corresponding luminal-B group treated with tamoxifen with high expression gene expression data, the inventors were able to identify a of the PIK3CA index (FIG. 6) seemed to nearly approximate molecular profile from PIK3CA mutation positive breast can the survival curves of the luminal-A tumours for the first 5 cers. The inventors then used the 81 genes to form an index years, suggesting that tamoxifen may, in fact, have a benefi that could quantify the level of PI3K/AKT pathway activation cial effect on outcome in this subgroup. Overall, these data of a given tumour similar to that triggered by a PIK3CA suggest that the high expression of the PIK3CA mutation mutation. Interestingly, the inventors found that in ER+BC, index and hence increased activation of the PI3K/AKT path high expression of the index, or activation of the PI3K/AKT way through PIK3CA mutations may predict favourably for pathway seemed to predict for a better outcome and also a tamoxifen treatment in the luminal-B, highly proliferative beneficial effect from tamoxifen treatment. This finding was ER+BCS. most impressive in the luminal-B Subgroup, which normally 0096 Univariate and multivariate analyses confirmed that has a poor prognosis compared with the luminal-A Subgroup. the PIK3CA index was able to provide independent prognos The index was relevant to tumour samples that were negative tic information the tamoxifen dataset (Table 3) for PIK3CA mutations, implying that those tumours with 0097. The inventors furtherlooked at the response of Her2 high expression of the PIK3CA signature had clinically rel positive BC patients treated with tamoxifen and observed that evant PI3K/AKT pathway deregulation through some other the Her2 positive BC patients having the mutated PIK3CA mechanism. The PIK3CA signature could therefore be a bet signature responded better to tamoxifen than the other Her2 ter indicator of pathway dysfunction than mutation status per positive patients. S 0098. The inventors conclude that anti oestrogen treat 0103) Given the multiple levels of molecular interactions ments may be useful for Her2 positive BC patients having the in the PI3K/AKT pathway, it is not inconceivable that differ mutated PIK3CA signature. ent activators of the pathway will be associated with different 0099. The overall incidence of PIK3CA mutations found transcriptional profiles and clinical outcomes. in this current study is within the range of the four other 0104. In cell lines, PIK3CA mutant lines including MCF7 reported large studies on PIK3CA mutations in breast cancer, and T47D, were found to be more sensitive to tamoxifen than even though the analysis of the PIK3CA gene in this study PIK3CA normal lines. These results may be extremely sig was restricted to exons 9 and 20 only. In contrast to other nificant for the luminal-B subtype as it could identify which studies, the majority of mutations found were located on exon tumours may benefit from endocrine therapy and which 20. This is most likely, because almost all of the breast cancers tumours will require other treatments to alter its poor prog in this data set were invasive ductal carcinomas, consistent nosis. It will also be important in the future to determine with previous observations that exon 9 mutations are more whether the PIK3CA gene set can predict response to PI3K common in invasive lobular carcinoma. The incidence of inhibitors. PIK3CA mutations observed to date makes it one of the 0105. The inventors report that PIK3CA mutations and for commonest genetic alterations in breast cancer. the first time, AKT1 mutations do not correlate with progno 0100. This study, similar to others, did not find a particular sis in a large cohort of ER+BCs treated with adjuvant tamox association of PIK3CA mutations with breast cancer clinico ifen monotherapy. The inventors disclose a gene signature pathological characteristics (Table 1). The only consistent that identifies those ER+BCs with clinically relevant activa finding thus far has been the association between PIK3CA tion of the PI3K/AKT pathway, and identify those breast mutations and a positive ER status. cancers that respond favourably to tamoxifen. These findings 0101 The inventors report here for the first time the inci are particularly significant for the luminal-B ER+ Subgroup dence and clinical outcome of AKT1 PHD mutations in and may provide useful stratification in future clinical trials ER+BC treated with adjuvant tamoxifen monotherapy. AKT1 evaluating endocrine therapy in ER+BC. US 2011/0038862 A1 Feb. 17, 2011

0106 Furthermore, the inventors compared the outcome and compared the effect of Everolimus in function of the of breast cancer patients having the wild-type signature with wild-type or mutated signature (according to the present the outcome of patients having the mutated signature (accord ing to the present invention), when treated with a PI3kinase invention). (pathway) inhibitor, being 10 mg per day of Everolimus 0107. Other inhibitors of the PI3K/Akt/mTOR pathway, (Afinitor R or RAD001 from Novartis) (an mTOR inhibitor) such as WYE-354, CCI-779 (from wyeth), Temsirolimus, taken orally. The inventors further selected patients having a GSK105.9615, Deforolimus, KU-0063794, PI-103 and NVP ER+ breast cancer and patients having a Her2+ breast cancer BEZ235 can be used as well.

TABLE 1. (A) Number of cases Exon Nucleotide change Amino acid change (%) 2O A314OG H1047R 29 (91%) 2O A31 SOT H1047L 3 (9%) 9 G1633A ES4SK 7 (50%) 9 G1624A ES42K 6 (43%) 9 G1634A ES4SG 1 (7%) Total 46 (100%) (B) Total cases Variable (n = 173) Mutated (n = 45) Normal (n = 127) P value Age O.3 s50yrs 14 2 (36%) 12 (66%) >50 yrs 157 43 (27%) 114 (73%) Tumour size 0.057 T1s2 cm 76 15 (20%) 61 (80%) T2 > 2 cm 95 30 (32%) 65 (78%) Histologic Grade O.S Grade 1 28 7 (25%) 21 (75%) Grade 2 8O 23 (26%) 57 (74%) Grade 3 32 7 (22%) 25 (78%) Nodal status O.S Node positive 87 23 (26%) 64 (74%) Node negative 83 21 (25%) 62 (75%) ER+ molecular subtype * Luminal A (GGI low) 87 25 (29%) 62 (71%) O.3 Luminal B (GGIhi) 79 17 (22%) 62 (78%)

Note 1) all 173 BC samples were ER+: 2) for some samples, values were missing: 3) for those samples whose gene expression data (n = 161) were available, 41 samples had PIK3CA mutations. * see reference LOIETAL, 2007.

TABLE 2a Parametric Ratio > 1 = up- Affymetrix Rank p-value FDR regulated in WT Probe set Gene symbol 8.OOE-07 O ..111 203012 X at RPL23A 2.54E-OS ..111 208.825 X at RPL23A 3.13E-OS ..111 222327 X at OR7E156P 3.97E-OS 375 219138 at RPL14 4.18E-OS O.S24 206994 at CST4 S.2SE-OS O.2 206378 at SCGB2A2 6.34E-OS 375 208229 at FGFR2 7.94E-OS 0.75 212377 s at NOTCH2 8.86E-OS 8 204992 s at PFN2 1 8.88E-OS 375 211406 at IER3IP1 949E-05 ..111 206447 at ELA2A 9.SOE-OS ..111 202002 at ACAA2 9.76E-OS 375 211212 s at ORCSL 9.78E-OS 222 218238 at GTPBP4. O.OOO1 222 209535 s at O.OOO1 .833 202431 s at MYC O.OOO1 O.667 212415 at SEPT6 O.OOO1 375 202300 at HBXIP

US 2011/0038862 A1 Feb. 17, 2011 10

TABLE 2b

Probe Set ID Gene Symbol Gene Title Entrez Gene ID 202002 a ACAA2 acetyl-Coenzyme A acyltransferase 2 10449 215741 x at AKAP8L A kinase (PRKA) anchor protein 8-like 26993 201288 a ARHGDIB Rho GDP dissociation inhibitor (GDI) beta 397 214553 s at ARPP-19 cyclic AMP phosphoprotein, 19 kD 10776 200058 s at ASCC3L1 activating signal cointegrator 1 complex subunit 3-like 1 23O2O 201171 a ATP6VOE1 ATPase, H+ transporting, lysosomal 9 kDa, VO subunit e1 8992 219242 a. CEP63 centrosomal protein 63 kDa 8O2S4 200998 s at CKAP4 cytoskeleton-associated protein 4 10970 203551s at COX11 COX11 homolog, cytochrome c oxidase assembly protein (yeast) 1353 2O7630 s at CREM cAMP responsive element modulator 1390 207082 a CSF1 colony stimulating factor 1 (macrophage) 1435 206994 a CST4 cystatin S 1472 207407 x at CYP4A11 cytochrome P450, family 4, subfamily A, polypeptide 11 1579 2024.81 a DHRS3 dehydrogenase/reductase (SDR family) member 3 9249 219590 x at DPHS DPH5 homolog (S. cerevisiae) S1611 202810 a DRG1 developmentally regulated GTP binding protein 1 4733 221586 s at E2F5 E2F transcription factor 5, p130-binding 1875 206447 a ELA2A elastase 2A 63.036 218023 s at FAM53C family with sequence similarity 53, member C 51307 208229 a FGFR2 fibroblast growth factor receptor 2 2263 209.046 s at GABARAPL2 GABA(A) receptor-associated protein-like 2 11345 217787 s at GALNT2 UDP-N-acetyl-alpha-D-galactosamine:polypeptide N-acetylgalactosaminyl() 2590 213133 s at GCSH///LOC733) glycine cleavage system protein H (aminomethyl carrier)///similar to Glyc(2) 2653, 73O107 220762 s at GNB1L guanine nucleotide binding protein (G protein), beta polypeptide 1-like S4584 218238 at GTPBP4. GTP binding protein 4 23S60 211406 at IER3IP1 immediate early response 3 interacting protein 1 S1124 20918.4 S at IRS2 insulin receptor Substrate 2 8660 21841.1 s at MBIP MAP3K12 binding inhibitory protein 1 51562 214051 at MGC39900/G2) thymosin beta15b///thymosin-like 8 11013, 286,527 203565 s at MNAT1 menage a trois homolog 1, cyclin Hassembly factor (Xenopus laevis) 4331 202431 s at MYC v-myc myelocytomatosis viral oncogene homolog (avian) 4609 202215 s at NFYC nuclear transcription factorY. gamma 4802 215339 at NKTR natural killer-tumor recognition sequence 482O 202443 x at NOTCH2 Notch homolog 2 (Drosophila) 4853 222115 x at N-PAC cytokine-like nuclear factorn-pac 84656 222327 x at OR7E156P olfactory receptor, family 7, subfamily E, member 156 pseudogene 283491 211212 s at ORC5L. origin recognition complex, Subunit 5-like (yeast) SOO1 208876 s at PAK2 p21 protein (Cdc42/Rac)-activated kinase 2 SO62 204992 s at PFN2 profilin 2 5217 202743 a PIK3R3 phosphoinositide-3-kinase, regulatory subunit 3 (gamma) 8SO3 208.502 s at PITX1 paired-like homeodomain 1 5307 205811 a POLG2 polymerase (DNA directed), gamma 2, accessory subunit 11232 215894 a PTGDR prostaglandin D2 receptor (DP) 5729 221915 s at RANBP1 RAN binding protein 1 5902 203250 a RBM16 RNA binding motif protein 16 22828 2078O1s at RNF10 ring finger protein 10 992.1 219138 a RPL14 ribosomal protein L14 9045 212537 x at RPL17 ribosomal protein L17 6139 203012 X at RPL23A ribosomal protein L23a 6147 200660 a S100A11 S100 calcium binding protein A11 6282 206799 a SCGB1D2 secretoglobin, family 1D, member 2 10647 206378 a SCGB2A2 Secretoglobin, family 2A, member 2 42SO 210779 x at SIP1 Survival of motor neuron protein interacting protein 1 8487 221041 s at SLC17A5 Solute carrier family 17 (anion sugar transporter), member 5 26SO3 203127 s at SPTLC2 serine palmitoyltransferase, long chain base subunit 2 9517 210369 a SWAP70 SWAP-70 protein 23075 204807 a TMEMS transmembrane protein 5 10329 218815 s at TMEM51 transmembrane protein 51 55092 211828 s at TNIK TRAF2 and NCK interacting kinase 23O43 216609 a TXN Thioredoxin 7295 212519 a UBE2E1 ubiquitin-conjugating enzyme E2E 1 (UBC4/5 homolog, yeast) 7324 212756 s at UBR2 ubiquitin protein ligase E3 component n-recognin 2 23304

(2) indicates text missing or illegible when filed US 2011/0038862 A1 Feb. 17, 2011

TABLE 3 TABLE 3-continued (B) (A) Univariate P Multivariate Univariate P Multivariate P Factor HR value HR Factor HR value HR: value Age 0.9 (0.6-1.6) O.8 Tumor size 1.7 (1.1-2.8) O.O3 Age 1.8 (0.6-5.7) 0.3 Nodal status 1.0 (0.4-2.7) O.1 Histologic grade 4.2 (2.2-7.8) O.OOOOOOO7 Tumor size 2.9 (1.6-5.0) 0.0002 3.2 (1.3-8.7) 0.01 PIK3CA index 0.7 (0.5-0.9) O.O3 Nodal status 2.1 (1.2-3.5) 0.0004 0.9 (0.4-2.3) 0.9 Luminal B vs A 3.0 (1.9–4.6) O.OOOOOO6 Histologic grade 1.8 (1.1-2.6) 0.005 Subtypeii PIK3CA mutation 0.6 (0.3-1.4) 0.26 Total patient samples: 425 PIK3CA index 0.4 (0.3-0.7) 0.01 0.5 (0.3-0.8) 0.01 *Only those factors significant in the Univariate analysis were Luminal B vs A 2.2 (1.6-3.0) 0.0000004 2.2 (1.3-3.7) 0.004 used in the multivariate model. histologic grade was not used in the multivariate model has Subtypeft highly correlated to the gene expression grade index (GGI) #as defined by the GGI (see LOIET AL; 2007) Total patient samples: 305 HR: hazard ratio

TABLE 4 Preferred genes associated with PIK3CA mutation Probe Set ID Gene Symbol Gene Title Entrez Gene ID 212500 at ADO 2-aminoethanethiol (cysteamine) dioxygenase 84890 201288 at ARHGDIB Rho GDP dissociation inhibitor (GDI) beta 397 214553 s at ARPP-19 cyclic AMP phosphoprotein, 19 kD 10776 200058 s at ASCC3L1 activating signal cointegrator 1 complex subunit 3-like 1 23O2O 222151 s at CEP63 centrosomal protein 63 kDa 8O2S4 200999 s at CKAP4 cytoskeleton-associated protein 4 10970 203551s at COX11 COX11 homolog, cytochrome c oxidase assembly protein (2) 1353 214508 X at CRE cAMP responsive element modulator 1390 2024.81 at DHRS3 dehydrogenase/reductase (SDR family) member 3 9249 219590 x at DPHS DPH5 homolog (S. cerevisiae) S1611 221586 s at E2F5 E2F transcription factor 5, p130-binding 1875 218023 s at FAM53C family with sequence similarity 53, member C 51307 217787 s at GALNT2 UDP-N-acetyl-alpha-D-galactosamine:polypeptide N-acety(?) 2590 213133 s at GCSH///LOC73) glycine cleavage system protein H (aminomethyl carrier)/(2) 2653///7301.07 217771 at GOLM1 golgi membrane protein 1 S128O 218238 at GTPBP4. GTP binding protein 4 23S60 202300 at HBXIP hepatitis B virus X interacting protein 10542 21841.1 s at MBIP MAP3K12 binding inhibitory protein 1 51562 203565 s at MNAT1 menage a trois homolog 1, cyclin Hassembly factor (Xeno 4331 202431 s at MYC v-myc myelocytomatosis viral oncogene homolog (avian) 4609 202215 s at NFYC nuclear transcription factorY. gamma 4802 215339 at NKTR natural killer-tumor recognition sequence 482O 212377 s at NOTCH2 Notch homolog 2 (Drosophila) 4853 222115 x at N-PAC cytokine-like nuclear factorn-pac 84656 211212 s at ORC5L. origin recognition complex, Subunit 5-like (yeast) SOO1 204992 s at PFN2 profilin 2 5217 202743 at PIK3R3 phosphoinositide-3-kinase, regulatory Subunit 3 (gamma) 8SO3 203250 at RBM16 RNA binding motif protein 16 22828 200660 at S100A11 S100 calcium binding protein A11 6282 206378 at SCGB2A2 Secretoglobin, family 2A, member 2 42SO 210779 x at SIP1 Survival of motor neuron protein interacting protein 1 8487 203128 at SPTLC2 serine palmitoyltransferase, long chain base subunit 2 9517 218815 s at TMEM51 transmembrane protein 51 55092 211828S at TNIK TRAF2 and NCK interacting kinase 23O43 212519 at UBE2E1 ubiquitin-conjugating enzyme E2E 1 (UBC4/5 homolog, ye) 7324 212756 s at UBR2 ubiquitin protein ligase E3 component n-recognin 2 23304 221704 s at VPS37B vacuolar protein sorting 37 homolog B (S. cerevisiae) 79.720 219247 s at ZDFHHC14 Zinc finger, DHHC-type containing 14 79683

ADO is also mentioned as C10orf22; GOLM1 is also mentioned as GOLPH2. (2) indicates text missing or illegible when filed US 2011/0038862 A1 Feb. 17, 2011 12

TABLE 5 Alternative PIK3CA gene signature PIK3CA-GS predictor FC - 1.3 p is 0.05 Entrez rank p Fold Gene Gene Coef accorcC2) value ChangC2) Probe set Description symbol ID ficG) Cytoband 370 -2.16 204637 a glycoprotein hormones, alpha polypeptide CGA 1081 325 91 209242 a. paternally expressed 3 PEG3 5178 248 .79 219109 a sperm associated antigen 16 SPAG16 79582 16 .78 202431 s a v-myc myelocytomatosis viral oncogene MYC 4609 homolog) .78 203987 a rizzled homolog 6 (Drosophila) 8323 .75 211548 s a hydroxyprostaglandin dehydrogenase 15-(NAD2) HPGD 3248 .74 203914 x 8. hydroxyprostaglandin dehydrogenase 15-(NAD2) HPGD 3248 73 204992 s a profilin 2 5217 .71 204688 a sarcoglycan, epsilon SGCE 8910 70 221582 a histone 3, H2a HIST3H2A 928.15 70 214051 a hypothetical protein MGC39900 MGC39900 286527 68 218730s a osteoglycin (osteoinductive factor, mimecan) OGN 4969 66 206110 a histone 1, H3h HIST1H3H 3109 66 202620 s a procollagen-lysine, 2-oxogluterate 5-dioxygenas() PLOD2 5352 63 209185 s a insulin receptor Substrate 2 RS2 8660 62 207156 a histone 1, H2ag. HIST1H2AG 8969 60 206070 s a EPH receptor A3 EPHA3 2042 S8 205280 a glycine receptor, beta GLRB 2743 S8 202619 s a procollagen-lysine, 2-oxogluterate 5-dioxygenas() PLOD2 5352 57 217963 s a nerve growth factor receptor (TNFRSF16) assc(2) NGFRAP1 27.018 57 203895 a phospholipase C, beta 4 PLCB4 5332 53 203913 s a hydroxyprostaglandin dehydrogenase 15-(NAD2) HPGD 3248 S1 204566 a protein phosphatase 1D magnesium-dependen(2) PPM1D 8493 S1 204939 s a phospholamban 5350 49 218541 s a chromosome 8 open reading frame 4 56892 49 20918.4 S a insulin receptor Substrate 2 8660 49 214469 a histone 1, H2ae 3012 49 205279 s a glycine receptor, beta 2743 48 201116 s a carboxypeptidase E 1363 47 204042 a WAS protein family, member 3 10810 46 201030 x a actate dehydrogenase B 3945 46 202708 s a histone 2, H2be 8349 45 218280 x a histone 2, H2aa 8337 .44 203608 a aldehyde dehydrogenase 5 family, member A1 (2) 7915 .44 213564 x a actate dehydrogenase B 3945 .44 214290 s a histone 2, H2aa 8337 43 212154 a Syndecan 2 (heparan Sulfate proteoglycan 1, 6,383 ce(2) O.OO11 43 218732 a Bcl-2 inhibitor of transcription 51651 O.0043 43 211578 s a ribosomal protein S6 kinase, 70 kDa, 61.98 polypeptide(?) 050 42 212859 x a metallothionein 1E (functional) 4493 103 42 202630 a. amyloid beta precursor protein (cytoplasmic APPEBP2 10513 ailC2) 2084 42 204916 a receptor (calcitonin) activity modifying protein 1 RAMP1 10267 250 41 209526 is a hepatoma-derived growth factor, related proteir2) HDGFRP3 SO810 994 41 219312 s a Zinc finger and BTB domain containing 10 ZBTB10 65986 1472 41 206825 a. oxytocin receptor OXTR SO21 1128 41 212589 a Sterol carrier protein 2 SCP2 22800 1172 41 213793 s a homer homolog 1 (Drosophila) HOMER1 9456 38 41 221586 s a E2F transcription factor 5, p130-binding E2F5 1875 533 41 213562 s a squalene epoxidase SQLE 6713 196 40 208456 s a related RAS viral (r-ras) oncogene homolog 2 RRAS2 22800 591 40 209617 s a catenin (cadherin-associated protein), delta 2 CTNND2 1SO1 (ne(2) S16 40 212590 at related RAS viral (r-ras) oncogene homolog 2 RRAS2 22800 1160 40 203510 at met proto-oncogene (hepatocyte growth factor réC) MET 4233 1370 39 200607 s a RAD21 homolog (S. pombe) RAD21 5885 843 39 212816 s a cyStathionine-beta-synthase CBS 875 41 39 216609 at Thioredoxin TXN 7295 933 38 203414 at monocyte to macrophage differentiation MMD 23531 associate?) 923 38 221194 s a PTD016 protein LOCS1136 S1136 17q23.1 23 37 202028 s a 6169 17q23-q25 309 37 210976 s a phosphofructokinase, muscle PFKM 5213 12q13.3 1238 37 203685 at B-cell CLL/lymphoma 2 BCL2 596 18q21.33. 18q21.3 US 2011/0038862 A1 Feb. 17, 2011 13

TABLE 5-continued Alternative PIK3CA gene signature PIK3CA-GS predictor FC - 1.3 p is 0.05 Entrez rank p Fold Gene Gene Coef accorcC2) value ChangC2) Probe set Description symbol ID ficO2) Cytoband 1419 37 214519 s at relaxin 2 RLN2 6019 9p24.1 251 37 210389 x at tibulin, delta 1 TUBD1 S1174 723.1 1137 36 204237 at GULP, engulfment adaptor PTB domain containir 2) GULP1 S1454 2a32.3-q33 51 36 201171 at ATPase, H+ transporting, lysosomal 9 kDa, VO ATP6VOE 8992 5q.35.1 (2) 639 36 205741 s at dystrobrevin, alpha DTNA 1837 1757 36 209292 at Inhibitor of DNA binding 4., dominant negative he(2) ID4 3400 1254 36 208078 s at SNF1-like kinase SNF1-like kinase SNF1LK 1SOO94 753 36 208920 at sorcin SRI 6717 7 35 208229 at fibroblast growth factor receptor 2 (bacteria-exp(2) FGFR2 2263 715 35 36711 at V-maf musculoaponeurotic fibrosarcoma oncoge(2) MAFF 23764 1340 35 205308 at chromosome 8 open reading frame 70 C8orf70 S1101 414 35 202353 s a proteasome (prosome, macropain) 26S subunit() PSMD12 5718 1637 35 204348 s a adenylate kinase 3-like 1 AK3L1 205 130 35 218597 s a chromosome 10 open reading frame 70 C10orf70 55847 871 35 05O13 s a adenosine A2a receptor ADORA2A 135 662 35 7975 at WW domain binding protein 5 WBPS S1186 968 35 O2342 s a tripartite motif-containing 2 TRIM2 23321 477 34 O1946 s a chaperonin containing TCP1, Subunit 2 (beta) CCT2 10576 4 34 91.38 at ribosomal protein L14 RPL14 9045 695 34 20147 s a family with sequence similarity 60, member A FAM60A 58516 O09 34 O761 s a growth factor receptor-bound protein 7 GRB7 2886 574 34 O5047 s a asparagine synthetase ASNS 440 613 34 20145 at ASAP FL21159 79884 O77 34 21523 s a Ras-related GTP binding D RRAGD 58528 10 34 1406 at immediate early response 3 interacting protein 1 ER3IP1 S1124 902 34 04235 s a GULP, engulfment adaptor PTB domain containir(2) GULP1 S1454 420 .33 O5321 at eukaryotic translation initiation factor 2, Subunit EIF2S3 1968 C2) O17 .33 212690 at DDHD domain containing 2 DDHD2 23259 922 .33 21997.4 x a enoyl Coenzyme A hydratase domain containing) ECHDC1 55862 437 .33 201161 s a cold shock domain protein A CSDA 853 745 .33 209849 s a RAD51 homolog C (S. cerevisiae) RADS1C S889 230 .33 205961 s a PC4 and SFRS1 interacting protein 1 PSIP1 11168 13 .33 211212 s a origin recognition complex, subunit 5-like (yeast) ORCSL 500 884 .33 221521 s a DNA replication complex GINS protein PSF2 Pfs2 51659 228 .33 221326 s a tubulin, delta 1 TUBD1 S1174 767 32 213353 at ATP-binding cassette, sub-family A (ABC1), mer(2) ABCAS 2346 184 32 205361 s a prefoldin 4 PFDN4 336 32 205543 at heat shock 70 kDa protein 4-like HSPA4L 12O7 32 205573 s a sorting nexin 7 NXT 60 32 203565 s a menage a trois 1 (CAKassembly factor) NAT1 828 32 218277 s a DEAH (Asp-Glu-Ala-His) box polypeptide 40 X40 1411 32 201117 s a carboxypeptidase E C E 1471 31 213548 s a hypothetical protein H41 4 590 31 218514 at hypothetical protein FLJ10587 273 31 205078 at phosphatidylinositol glycan, class F 299 31 204940 at phospholamban 87 31 221943 x a Ribosomal protein L38 C 3 8 56 31 214553 s a cyclic AMP phosphoprotein, 19 kD 930 31 216693 x a hepatoma-derived growth factor, related proteir2) 820 31 208729 x a major histocompatibility complex, class I, B 1259 31 218788 s a SET and MYND domain containing 3 449 31 203888 at hrombomodulin 949 32 216250 s a eupaxin 2127 32 202149 at neural precursor cell expressed, developmental.(2) 1288 32 38241 at butyrophilin, subfamily 3, member A3 1454 32 213478 at kaZrin 891 32 215536 at major histocompatibility complex, class II, DQ be?) A D Q B 2 709 32 218313 s at UDP-N-acetyl-alpha-D-galactosamine:polypept(2) ALNT7 90 32 207291 at proline rich Gla (G-carboxyglutamic acid) 4 (trar() 734 32 205757 at ectonucleoside triphosphate diphosphohydrolasé2) 119 32 203236 s at ectin, galactoside-binding, soluble, 9 (galectin 9) s T p. 1894 32 210538 s at baculoviral IAP repeat-containing 3 556 32 212256 at UDP-N-acetyl-alpha-D-galactosamine:polypept(2) 428 32 211944 at BAT2 domain containing 1 AT2 D1 393 .33 2066.62 at glutaredoxin (thioltransferase) 1484 .33 217838 s at Enah/Vasp-like 2130 .33 2094.88 s at RNA binding protein with multiple splicing . M S US 2011/0038862 A1 Feb. 17, 2011 14

TABLE 5-continued Alternative PIK3CA gene signature PIK3CA-GS predictor FC - 1.3 p is 0.05 Entrez rank p Fold Gene Gene Coef accorcC2) value ChangC2) Probe set Description symbol ID ficG) Cytoband 481 .33 21083.5 s at C-terminal binding protein 2 CTBP2 1488 482 .34 212841 s at PTPRF interacting protein, binding protein 2 (lip(2) PPFIBP2 849S 221 34 209788 s at type 1 tumor necrosis factor receptor shedding (2) ARTS-1 51752 2004 .34 212240 s at phosphoinositide-3-kinase, regulatory subunit 1 PIK3R1 5295 701 34 203509 at sortilin-related receptor, L(DLR class) A repeats () SORL1 6653 487 34 213931 at inhibitor of DNA binding 2, dominant negative he?) D2 3398 156 .34 211621 at AR 367 436 34 222075 s a ornithine decarboxylase antizyme 3 OAZ3 S1686 403 34 201367 s a Zinc finger protein 36, C3H type-like 2 ZFP36L2 678 1382 34 217478 s a major histocompatibility complex, class II, DMalrC2) HLA-DMA 3.108 72 .35 221041 s a solute carrier family 17 (anion/sugar transporter(2) SLC17AS 26SO3 1788 35 203474 at IQ motif containing GTPase activating protein 2 QGAP2 10788 1957 35 209522 s a carnitine acetyltransferase CRAT 1384 35 35 203128 at serine palmitoyltransferase, long chain base sub(2) SPTLC2 9517 534 35 2101.39 s a peripheral myelin protein 22 PMP22 5376 758 35 204137 at transmembrane 7 superfamily member 1 (upregg) TM7SF1 7107 148 35 269916 at dehydrogenase E1 and transketolase domain cc(2) DHTKD1 555.26 208 35 202421 at immunoglobulin Superfamily, member 3 GSF3 3321 1346 35 209276 is a glutaredoxin (thioltransferase) GLRX 2745 1173 35 201340s a ectodermal-neural cortex (with BTB-like domain(2) ENC1 8507 976 35 211366 x a caspase 1, apoptosis-related cysteine peptidase(2) CASP1 834 824 35 205379 at carbonyl reductase 3 CBR3 874 1451 36 201976 s a myosin X MYO10 4651 798 36 202336 s a peptidylglycine alpha-amidating monooxygenase) PAM SO66 2028 36 202709 at fibromodulin FMOD 2331 423 36 204875 s a GDP-mannose 4,6-dehydratase GMDS 2762 421 .36 218322 s a acyl-CoA synthetase long-chain family member 5(2) ACSLS 51703 348 .36 221042 s a calmin (calponin-like, transmembrane) CLMN 79789 247 36 202638 s a intercellular adhesion molecule 1 (CD54), humar(2) CAM1 3383 8 36 212377 s a Notch homolog 2 (Drosophila) NOTCH2 4853 1258 36 204017 at KDEL (Lys-Asp-Glu-Leu) endoplasmic reticulum () KDELR3 1101S 102 36 205248 at chromosome 21 open reading frame 5 C21orfs 998O 1233 36 203887 s a hrombomodulin THEBD 7056 529 37 210732 s a ectin, galactoside-binding, soluble, 8 (galectin 8) LGALS8 3964 327 37 213462 at neuronal PAS domain protein 2 NPAS2 4862 1908 37 210319 x a mish homeo box homolog 2 (Drosophila) MSX2 4488 999 37 201369 s a Zinc finger protein 36, C3H type-like 2 ZFP36L2 678 495 37 208949 s a ectin, galactoside-binding, soluble, 3 (galectin 3) LS3 , , 3958 769 37 209970 x a caspase 1, apoptosis-related cysteine peptidase 834 986 37 204654 s a transcription factor AP-2 alpha (activating enhar(2) 7020 32 .38 217787 s a UDP-N-acetyl-alpha-D-galactosamine:polypept(2) 2 2590 396 .38 212875 s a chromosome 21 open reading frame 25 rf2 5 25966 1666 .38 217744 s a PERP, TP53 apoptosis effector 6406S 729 .38 41644 at SAM and SH3 domain containing 1 fGRs 23328 1081 .38 204160 s a ectonucleotide pyrophosphatase/phosphodieste?) i. 22875 112S .38 201242 s a ATPase, Na+/K+ transporting, beta 1 polypeptic(2) AT P 481 467 39 212249 a phosphoinositide-3-kinase, regulatory subunit 1 K3R1 5295 297 39 213418 a heat shock 70 kDa protein 6 (HSP7OB') SPA6 3310 827 39 202962 a kinesin family member 13B IF13B 23303 290 39 222258 s a SH3-domain binding protein 4 H3BP4 23677 551 .39 214130 s a phosphodiesterase 4D interacting protein (myom(2) DE4DIP 9659 1655 39 204365 s a chromosome 2 open reading frame 23 2Orf23 65055 1533 39 205668 a ymphocyte antigen 75 Y75 406S 143 39 218451 a CUB domain containing protein 1 DCP1 64.866 948 40 205225 a. estrogen receptor 1 SR1 2099 610 40 211965 a. Zinc finger protein 36, C3H type-like 1 FP36L1 677 511 40 213308 a SH3 and multiple ankyrin repeat domains 2 HANK2 22941 1018 40 208006 a forkhead box I1 FOX1 2299 1374 40 206191 a ectonucleoside triphosphate diphosphohydrolasé(2) ENTPD3 956 950 40 220108 a guanine nucleotide binding protein (G protein), (2) GNA14 9630 964 40 211368 s at caspase 1, apoptosis-related cysteine peptidase CASP1 834 100 40 208683 a calpain 2, (mII) large subunit CAPN2 824 1927 41 214295 a KIAA0485 protein KIAAO485 57235 S61 41 39549 at neuronal PAS domain protein 2 NPAS2 4862 236 41 214129 a Phosphodiesterase 4D interacting protein (myom(2) PDE4DIP 9659 963 41 204446 s at arachidonate 5-lipoxygenase ALOX5 240 317 41 206011 a caspase 1, apoptosis-related cysteine peptidase CASP1 834 US 2011/0038862 A1 Feb. 17, 2011 15

TABLE 5-continued Alternative PIK3CA gene signature PIK3CA-GS predictor FC - 1.3 p is 0.05 Entrez rank p Fold Gene Gene Coef accorcC2) value ChangC2) Probe set Description symbol ID ficO2) Cytoband 803 41 205879 x at ret proto-oncogene (multiple endocrine neoplas() RET 5979 49 41 220066 a caspase recruitment domain family, member 15 CARD15 64127 1151 42 221558 s at lymphoid enhancer-binding factor 1 LEF1 51176 480 42 211110 s. at androgen receptor (dihydrotestosterone receptcG) AR 367 1594 42 O1641 a bone marrow stromal cell antigen 2 BST2 684 416 42 O8997 s at uncoupling protein 2 (mitochondrial, proton carr(2) UCP2 7351 306 43 8273 s at protein phosphatase 2C, magnesium-dependen(2) PPM2C S4704 22 43 3109 a TRAF2 and NCK interacting kinase TNIK 23O43 998 43 O3221 a transducin-like enhancer of split 1 (E(sp1) homo() TLE1 7088 1840 43 2551 a CAP, adenylate cyclase-associated protein, 2 (3) CAP2 10486 2023 .44 OO606 a desmoplakin DSP 1832 110 .44 O1482 a quiescin Q6 QSCN6 5768 341 .44 8918 a mannosidase, alpha, class 1C, member 1 MAN1C1 57134 892 45 4329 x at Tumor necrosis factor (ligand) superfamily, memb3) TNFSF10 8743 719 45 3236 a SAM and SH3 domain containing 1 SASH1 23328 168 45 8084 X at FXYD domain containing ion transport regulato(2) FXYD5 53827 386 46 02017 a epoxide hydrolase 1, microsomal (xenobiotic) EPHX1 2052 17 46 2415 a. septin 6 4OO62 23157 1672 46 2543 a absent in melanoma 1 AIM1 2O2 479 47 02286 s at tumor-associated calcium signal transducer 2 TACSTD2 4070 232 47 O9619 a CD74 antigen (invariant polypeptide of major his(2) CD74 972 249 47 O4352 a TNF receptor-associated factor 5 TRAFS 7188 219 47 4791 a hypothetical protein BC004921 LOC93349 933.49 2O 47 O2743 a phosphoinositide-3-kinase, regulatory subunit 3(2) PIK3R3 8SO3 1140 47 05472 s a dachshund homolog 1 (Drosophila) DACH1 16O2 356 48 O4983 s a glypican 4 GPC4 2239 613 49 213107 a TRAF2 and NCK interacting kinase TNIK 23O43 30 49 201288 a Rho GDP dissociation inhibitor (GDI) beta ARHGDIB 397 1116 49 221841 s a Kruppel-like factor 4 (gut) KLF4 93.14 1032 49 202986 a aryl-hydrocarbon receptor nuclear translocato(2) ARNT2 9915 1630 49 1405 i at chemokine (C-C motif) ligand 5 CCL5 63S2 862 SO 205645 a. RALBP1 associated Eps domain containing 2 REPS2 918S 1231 SO 217966 s a chromosome 1 open reading frame 24 C1orf24 116496 1549 S1 210372 s a tumor protein D52-like 1 TPD52L1 71.64 1456 53 202376 a serpin peptidase inhibitor, clade A (alpha-1 antip(2) SERPINA3 12 821 53 205278 a glutamate decarboxylase 1 (brain, 67 kDa) GAD1 2571 59 54 208.502 s a paired-like homeodomain transcription factor 1 PITX1 5307 321 54 200824 a glutathione S-transferase pi GSTP1 2950 342 55 212325 a. KIAA1102 protein KIAA1102 22998 126 55 208998 a uncoupling protein 2 (mitochondrial, proton carrie? UCP2 7351 1358 S6 217967 s a chromosome 1 open reading frame 24 C1orf24 116496 817 S6 203786 s a tumor protein D52-like 1 TPD52L1 71.64 6q22-q23 776 57 205990s a wingless-type MMTV integration site family, mem) WNTSA 7474 3p21-p14 1462 57 205471 s a dachshund homolog 1 (Drosophila) DACH1 16O2 3d22 1748 57 203354 s a pleckstrin and Sect domain containing 3 PSD3 23362 8pter-p23.3 967 S8 202687 s a tumor necrosis factor (ligand) Superfamily, memb) TNFSF10 8743 3d26 1328 59 218613 at pleckstrin and Sect domain containing 3 PSD3 23362 8pter-p23.3 372 60 O4972 at 2'-5'-oligoadenylate synthetase 2, 69.71 kDa 4939 2d24.2 144 62 O4984 at glypican 4 2239 Xq26.1 291 62 O2688 at tumor necrosis factor (ligand) Superfamily, memb) TNFSF10 8743 3d26 1475 62 4774 x a Erinucleotide repeat containing 9 TNRC9 27324 6d12.1 85 63 21666 s a PYD and CARD domain containing PYCARD 291.08 6p12-p11.2 530 63 03649 s a phospholipase A2, group IIA (platelets, synovia(3) PLA2G2A 532O 3S 46 .67 1828 s a TRAF2 and NCK interacting kinase TNIK 23O43 q26.2-q26.31 151 70 2327 at KIAA1102 protein KIAA1102 22998 13 86 70 04070 at retinoic acid receptor responder (tazarotene induC2) RARRES3 5920 1656 70 20177 s a transmembrane protease, serine 3 TMPRSS3 64699 1q22.3 1538 .71 4440 at N-acetyltransferase 1 (arylamine N-acetyltransf3) NAT1 9 p23.1-p21.3 167 .71 2328 at KIAA1102 protein KIAA 102 22998 13 1514 72 O3355 s a pleckstrin and Sect domain containing 3 PSD3 23362 pter-p23.3 736 73 O1860s a plasminogen activator, tissue PLAT 5327 12 26S .74 9630 at PDZK1 interacting protein 1 PDZK1P1 10158 33 1469 .75 O9016 s a keratin 7 KRT7 3855 2q12-q13 391 .76 04364 S a chromosome 2 open reading frame 23 C2Orf23 65055 p11.2 106 8O 985.0 s a ets homologous factor EHF 26298 28 82 7771 at golgi phosphoprotein 2 GOLPH2 S1280 q21.33 179 84 4428 X a complement component 4A // complement compcG) C4A 720 21.3 C4B US 2011/0038862 A1 Feb. 17, 2011 16

TABLE 5-continued Alternative PIK3CA gene signature PIK3CA-GS predictor FC - 1.3 p is 0.05 Entrez rank p Fold Gene Gene Coef accorcC2) value ChangC2) Probe set Description symbol ID ficQ) Cytoband 172 O.OO21 1.84 213693 s at mucin 1, transmembrane MUC1 4582 260 O.0034 1.90 207847 s at mucin 1, transmembrane MUC1 4582 193 O.OO2S 1.92 208451s at complement component 4A // complement compc3) C4A 721 3 C4B 5 O.OOOO 1.96 206994 at cystatin S CST4 1472 1.21 244 O.OO32 1.97 219759 at leukocyte-derived arginine aminopeptidase LRAP 641.67 392 O.O1 1.98 202357 s at B-factor, properdin BF 629 3 117 O.OO14 2.01 209706 at NK3 transcription factor related, locus 1 (Drosoph2) NKX3-1 4824 328 O.OO45 2.12 220414 at calmodulin-like 5 CALMLS S1806 S.1 1343 O.O3 2.15 206204 at growth factor receptor-bound protein 14 GRB14 2888 -q24 614 O.O1 2.16 201884 at carcinoembryonic antigen-related cell adhesion (2) CEACAMS 104.8 3.1-q13.2 476 O.O1 2.23 204607 at 3-hydroxy-3-methylglutaryl-Coenzyme A synthé(2) HMGCS2 3158 -p12 541 O.O1 2.34 211657 at carcinoembryonic antigen-related cell adhesion (2) CEACAM6 468O 3.2 911 O.O2 2.34 203757 s at carcinoembryonic antigen-related cell adhesion (2) CEACAM6 468O 3.2 258 O.0034 3.23 202018 s at lactotransferrin LTF 4057 31 37 O.OOO3 4.47 206799 at secretoglobin, family 1D, member 2 SCGB1D2 10647 3 O.OOO1 5.19 206378 at secretoglobin, family 2A, member 2 SCGB2A2 42SO

(2) indicates text missing or illegible when filed

TABLE 6 Most preferred genes associated with PIK3CA mutation Probe Set ID Gene Symbol Gene Title Entrez Gene ID Cytoband 201288 at ARHGDIB Rho GDP dissociation inhibitor (GDI) beta 397 12p12.3 214553 s at ARPP-19 cyclic AMP phosphoprotein, 19 kD 10776 15q21.2 221586 s at E2F5 E2F transcription factor 5, p130-binding 1875 217787 s at GALNT2 UDP-N-acetyl-alpha-D-galactosamine:polypeptide N-ace(2) 2590 217771 at GOLM1 golgi membrane protein 1 S1280 203565 s at MNAT1 menage atrois homolog 1, cyclin Hassembly factor (Xe(2) 4331 202431 s at MYC v-myc myelocytomatosis viral oncogene homolog (avian) 4609 8q24.21 212377 s at NOTCH2 Notch homolog 2 (Drosophila) 4853 211212 s at ORCSL origin recognition complex, Subunit 5-like (yeast) SOO1 204992 s at PFN2 profilin 2 5217 202743 at PIK3R3 phosphoinositide-3-kinase, regulatory Subunit 3 (gamma) 8SO3 1p34.1 206378 at SCGB2A2 secretoglobin, family 2A, member 2 42SO 11 q13 203128 at SPTLC2 serine palmitoyltransferase, long chain base subunit 2 9517 14q24.3-q31 211828 s at TNIK TRAF2 and NCK interacting kinase 23O43

(2) indicates text missing or illegible when filed

SEQUENCE LISTING

<16O is NUMBER OF SEO ID NOS: 6

<210s, SEQ ID NO 1 &211s LENGTH: 21 &212s. TYPE: DNA <213> ORGANISM: Artificial sequence 22 Os. FEATURE: <223> OTHER INFORMATION: Akt1 primer, Exon 4 Forward <4 OOs, SEQUENCE: 1 agggtctgac C cctagagat g 21

<210s, SEQ ID NO 2 &211s LENGTH: 18 &212s. TYPE: DNA <213> ORGANISM: Artificial sequence 22 Os. FEATURE: US 2011/0038862 A1 Feb. 17, 2011 17

- Continued <223> OTHER INFORMATION: Akt1 Exon 4 Reverse primer

<4 OOs, SEQUENCE: 2 agagggct Co agccalacc 18

<210s, SEQ ID NO 3 &211s LENGTH: 24 &212s. TYPE: DNA <213> ORGANISM: Artificial Sequence 22 Os. FEATURE: <223> OTHER INFORMATION: PIK3CA Exon 9 Forward primer 22 Os. FEATURE: <221 > NAMEAKEY: misc feature <222s. LOCATION: (1) . . (1) 223 OTHER INFORMATION: FAM 22 Os. FEATURE: <221 > NAMEAKEY: misc feature <222s. LOCATION: (1) . . (1) 223 OTHER INFORMATION: 6 - FAM

<4 OOs, SEQUENCE: 3 tgaaaatgta tittgctttitt ctdt 24

<210s, SEQ ID NO 4 &211s LENGTH: 24 &212s. TYPE: DNA <213> ORGANISM: Artificial Sequence 22 Os. FEATURE: <223> OTHER INFORMATION: PIK3CA Exon 9 Reverse primer 22 Os. FEATURE: <221 > NAMEAKEY: misc feature <222s. LOCATION: (1) . . (1) 223 OTHER INFORMATION: WIC

<4 OOs, SEQUENCE: 4 tgtaaattct gctittattta titcc 24

<210s, SEQ ID NO 5 &211s LENGTH: 23 &212s. TYPE: DNA <213> ORGANISM: Artificial Sequence 22 Os. FEATURE: <223> OTHER INFORMATION: PIK3CA Exon 2 O Forward primer 22 Os. FEATURE: <221 > NAMEAKEY: misc feature <222s. LOCATION: (1) . . (1) 223 OTHER INFORMATION: NED

<4 OOs, SEQUENCE: 5 tccaaactga ccaaactgtt citt 23

<210s, SEQ ID NO 6 &211s LENGTH: 24 &212s. TYPE: DNA <213> ORGANISM: Artificial Sequence 22 Os. FEATURE: <223> OTHER INFORMATION: PIK3CA Exomin 20 Reverse Primer 22 Os. FEATURE: <221 > NAMEAKEY: misc feature <222s. LOCATION: (1) . . (1) 223 OTHER INFORMATION: PET US 2011/0038862 A1 Feb. 17, 2011

- Continued

22 Os. FEATURE: <221 > NAMEAKEY: misc feature <222s. LOCATION: (1) . . (1) 223 OTHER INFORMATION: PE

<4 OOs, SEQUENCE: 6 tccagagtga gctitt cattt totc 24

1. A method to determine a signature of mutated PIK3CA 9. The method of claim 1 further comprising a step of comprising the steps of determining a clinical outcome of breast tumour affecting a measuring an expression level of genes from a biopsy of a patient if treated with an antitumoural agent against breast Breast cancer tumour, by contacting mRNA sequences tumour. from the cells of the said biopsy with a gene set of more 10. The method according to claim 9 wherein the breast than 3 capture nucleotide sequences related to human tumour is ER+. mutated PIK3CA, mutated AKT-1 sequences and/or 11. The method according to the claim 10 wherein the sequences involved in PI3KJAKT pathway activation, breast tumour is obtained from a high proliferative tumour wherein the at least 3 nucleotide sequences related to sample. human mutated PIK3CA, mutated AKT-1 sequences 12. The method according to the claim 10 wherein the and/or sequences involved in PI3K/AKT are selected breast tumour is a luminal B ER+ tumour. within the sequences of Table 6 and wherein at least a 13. The method of claim 12, wherein the breast tumor portion of the capture nucleotide sequences of said cap harbours a mutated PIK3CA signature and which further ture nucleotide sequences can be hybridized to corre comprises a step of administering to the patient, an anti sponding target RNA sequence of a breast cell; and oestrogen agent selected from the group consisting of tamox ifen, raloxifene, faslodex or a mixture thereof. deducing whether said tumour harbours a PIK3CA 14. The method according to claim 10, further comprising mutated Status. a step of administering to the patient, an anti oestrogen agent 2. The method of claim 1, wherein at least 5 nucleotide selected from a group consisting of a selective oestrogen sequences related to human mutated PIK3CA, mutated receptor modulator, a selective oestrogen receptor down regu AKT-1 sequences and/or sequences involved in PI3K/AKT lator, a GnRH analog or an aromatase inhibitor. are selected within the sequences of Table 6. 15. The method of claim 10, wherein the breast tumor 3. The method of claim 1 wherein nucleotide sequences harbours a mutated PIK3CA signature and which further related to human mutated PIK3CA, mutated AKT-1 comprises a step of administering to the patient an anti sequences and/or sequences involved in PI3K/AKT are all the oestrogen agent selected from the group consisting of tamox sequences of Table 6. ifen, raloxifene, faslodex or a mixture thereof. 4. The method of claim 1, wherein the nucleotide 16. The method according to claim 10, wherein the breast sequences related to human mutated PIK3CA, mutated tumor harbours a mutated PIK3CA signature and wherein the AKT-1 sequences and/or sequences involved in PI3K/AKT breast tumor is Her2+, and which further comprises a step of further comprise 1, 2, 3, 4, 5, 6,7,8,9, 10 or all the sequences administering to the patient an antitumoral agent selected of Table 5 that are not present in Table 6. from a group consisting of a selective oestrogen receptor 5. The method of claim 1, wherein the nucleotide modulator, a selective oestrogen receptor down regulator, a sequences related to human mutated PIK3CA, mutated GnRH analog or an aromatase inhibitor. AKT-1 sequences and/or sequences involved in PI3K/AKT 17. The method of claim 16 wherein the antitumoral agent further comprise 1, 2, 3, 4, 5, 6,7,8,9, 10 or all the sequences further comprise an anti Her2 compound, preferably an anti of Table 4 that are not present in Table 6. Her2 antibody. 6. The method of claim 1, wherein the nucleotide 18. The method of claim 17, wherein the anti Her2 com sequences related to human mutated PIK3CA, mutated pound is an anti Her2 antibody. AKT-1 sequences and/or sequences involved in PI3K/AKT 19. The method of claim 18, wherein the anti Her2 com further comprise 1, 2, 3, 4, 5, 6,7,8,9, 10 or all the sequences pound is Trastuzumab. of Table 2a or 2b that are not present in Table 6. 20. The method according to claim 9, wherein the breast 7. The method of claim 1, wherein the nucleotide tumor harbours a wild-type PIK3CA signature and which sequences related to human mutated PIK3CA, mutated further comprises a step of administering to the patient, an AKT-1 sequences and/or sequences involved in PI3K/AKT antitumoural agent selected from the group consisting of a further comprises capture probes for a detection of an expres PI3-kinase inhibitor(s) or a PI3-kinase pathway inhibitor(s). sion level of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 or the 12 corre 21. The method of claim 20, wherein the inhibitor of the sponding target mRNA sequences selected from the group PI3-kinase pathway is a mTOR inhibitor. consisting of PML, PP2A, IRS2 PIK3R1, ESR1, FOXO3A, 22. The method of claim 21, wherein the mTOR inhibitoris P21 (PAK2), RPS6K, EIF4E, RHEB, P27 and PI3K Everolimus. Sequences. 23. The method according to claim 9, wherein the breast 8. The method of claim 1 further comprising a step of tumor harbours a mutated PIK3CA signature and wherein the sequencing of the PIK3CA gene. breast tumor is Her2+, and which further comprises a step of US 2011/0038862 A1 Feb. 17, 2011 administering to the patient an antitumoral agent being an anti further comprises a step of administering to the patient a Her2 compound, preferably an anti Her2 antibody. chemotherapy. 24. The method of claim 23, wherein the anti Her2 com 27. The method according to claim 10, wherein the breast pound is an anti Her2 antibody. tumor harbours a wild-type PIK3CA signature and which further comprises a step of administering to the patient a 25. The method of claim 24, wherein the anti Her2 com radiotherapy. pound is Trastuzumab. 26. The method according to claim 10, wherein the breast tumor harbours a wild-type PIK3CA signature and which