WO2007109026A2 - Pten compositions and methods for detecting breast cancer - Google Patents

Pten compositions and methods for detecting breast cancer Download PDF

Info

Publication number
WO2007109026A2
WO2007109026A2 PCT/US2007/006338 US2007006338W WO2007109026A2 WO 2007109026 A2 WO2007109026 A2 WO 2007109026A2 US 2007006338 W US2007006338 W US 2007006338W WO 2007109026 A2 WO2007109026 A2 WO 2007109026A2
Authority
WO
WIPO (PCT)
Prior art keywords
tumor
pten
gene expression
profile
domain containing
Prior art date
Application number
PCT/US2007/006338
Other languages
French (fr)
Other versions
WO2007109026A3 (en
Inventor
Ramon Parsons
Lao Saal
Ake Borg
Sofia Gruvberger-Saal
Original Assignee
The Trustees Of Columbia University In The City Of New York
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by The Trustees Of Columbia University In The City Of New York filed Critical The Trustees Of Columbia University In The City Of New York
Publication of WO2007109026A2 publication Critical patent/WO2007109026A2/en
Publication of WO2007109026A3 publication Critical patent/WO2007109026A3/en

Links

Classifications

    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q1/00—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions
    • C12Q1/68—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions involving nucleic acids
    • C12Q1/6876—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes
    • C12Q1/6883—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes for diseases caused by alterations of genetic material
    • C12Q1/6886—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes for diseases caused by alterations of genetic material for cancer
    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q2600/00—Oligonucleotides characterized by their use
    • C12Q2600/106—Pharmacogenomics, i.e. genetic variability in individual responses to drugs and drug metabolism
    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q2600/00—Oligonucleotides characterized by their use
    • C12Q2600/112—Disease subtyping, staging or classification
    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q2600/00—Oligonucleotides characterized by their use
    • C12Q2600/118—Prognosis of disease development
    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q2600/00—Oligonucleotides characterized by their use
    • C12Q2600/136—Screening for pharmacological compounds
    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q2600/00—Oligonucleotides characterized by their use
    • C12Q2600/158—Expression markers

Definitions

  • Another aspect of the invention provides a method for identifying whether a tumor has a high potential for malignancy, the method comprising: (a) determining gene expression level in a sample of tumor cells of (i) stathmin/oncoprotein 18, (ii) PIK3CA, and (iii) one or more genes selected from the group consisting of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing
  • the invention provides results that show that therapy to the PTEN pathway components may be effective. Furthermore, for BRCAl germline carriers, the risk of cancer is much higher than in the general population, and many people opt to have prophylactic bilateral mastectomy and/or oophorectomy to reduce their risk.
  • a low-dose or infrequent regimen of anti-PTEN pathway therapy e.g., rapamycin
  • the invention provides a method for delaying or preventing the onset of cancer in a subject, the method comprising administering to the subject an effective amount of a compound that increases expression of one or more genes selected from the group consisting of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1, and retinoic acid induced 2.
  • stathmin be used in connection with the PTEN gene signature to predict poor outcome tumors.
  • stathmin itself can be used to reliably identify tumors with poor outcome.
  • the signature genes can also serve as potential therapeutic targets. Inhibition of stathmin, which regulates MT dynamics, could have inhibitory effects on mitosis and/or cell migration. Additionally, studies have shown overexpression of stathmin to affect sensitivity to MT-stabilizing drugs and MT-destabilizing drugs (Orr et al., 2003). Therefore, studies can be designed to determine whether MT-targeting therapies, in combination with PI3 -K- targeted therapies, would be synergistic against rapidly proliferating P13-K pathway- activated tumors. Some of the signature genes encode cell surface proteins, which may be useful as molecular beacons of pathway activation that could be imaged non-invasively using labeled antibodies to monitor disease progression and response to targeted therapies. The future of cancer management is quickly moving towards pathway-based profiling and directed therapy. Moreover, the gains in pathway-specific treatment of cancer are likely to translate to other diseases, as for example the PI3-K pathway is involved in a vast array of human ailments.
  • RNA Preparation and Microarray Hybridization Total RNA was extracted from ⁇ 100 mg grossly dissected frozen tumor tissue by pulverization in a microdismembrator chilled on dry ice, immediately followed by homogenization in TRlzol reagent according to manufacturer's instructions (Invitrogen, Carlsbad, CA). RNA was purified a second round using the RNeasy kit (Qiagen, Hilden, Germany), and the final yield and purity was assessed using a Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA).
  • Data within each array was normalized using a BASE plug-in implementation of the LOWESS algorithm, whereby the 48 subarrays were grouped into 6 groups spatially lengthwise along the slide (8 pins per group, i.e. 2 rows of 4) and the data within each pin- group smoothed independently. The data matrix was then exported and further filtered to remove all rows (features) with missing data in >20% of assays.
  • EXAMPLE 2 A PTEN GENE EXPRESSION SIGNATURE IN HUMAN BREAST
  • the pi 10a catalytic subunit of PI3K encoded by PIKSCA is a transforming oncogene (Chang et al, 1997), the 3q26 region where PlKiCA is located is amplified in tumors (Shayesteh et al, 1999; Ma et al, 2000), and recently, the PIK3CA gene was shown to have activating mutations in five types of cancer (Samuels et al, 2004).
  • PIK3CA mutations were significantly associated with lower tumor grade, lower S-phase fraction, and negative Ki67 staining. This may indicate that, in vivo, PIK3CA mutation is a less potent driver of cell proliferation than, for example, PTEN alteration. It was previously reported that PIK3CA mutations were positively associated to ER, lymph node status, and HER2 status (Saal et al, 2005). The different results for association to ER and node status may be due to the fact that the present study is underpowered compared to the prior report; moreover the strongest association had been seen within a subset of over 150 stage II Swedish BCs. Thus, stage and population effects may also have influenced the present results.

Abstract

The invention provides gene signatures and methods for using the gene signatures for identifying a tumor with high potential for malignancy in a subject based on activation of the PTEN signaling pathway. The invention also provides methods for using the gene signatures to determine the efficacy of anti-tumor therapies. The gene signatures and methods provided by the invention can be used to assess the malignancy potential of breast cancer tumors, including non-hereditary basal-like breast cancer and hereditary BRCAl-related basal-like breast cancer. The invention also provides methods for identifying compounds that regulate PTEN signaling pathways by assessing expression levels of one or more genes in the gene signature.

Description

PTEN COMPOSITIONS AND METHODS FOR DETECTING BREAST CANCER
[0001] This application claims priority to U.S. Provisional Application No.
60/782,503, filed March 15, 2006, which is hereby incorporated by reference in its entirety.
[0002] The invention disclosed herein was made with U.S. Government support under
NIH Grant No. 5T32 GM07367-29. Accordingly, the U.S. Government may have certain rights in this invention.
[0003] All patents, patent applications and publications cited herein are hereby incorporated by reference in their entirety. The disclosures of these publications in their entireties are hereby incorporated by reference into this application in order to more fully describe the state of the art as known to those skilled therein as of the date of the invention described and claimed herein.
[0004] A portion of the disclosure of this patent document contains material that is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever.
BACKGROUND OF THE INVENTION
[0005] The tumor suppressor PTEN is involved in the phosphatidylinositol 3-kinase
(PI3K) pathway and many downstream pathways that regulate cell growth, migration, apoptosis, and the cell cycle. The oncogenic PI3K pathway is activated in a significant proportion of multiple types of human neoplasms.
[0006] The future of cancer management is moving towards pathway-based profiling and directed pharmacological/small molecule therapy. As the PI3K pathway is highly involved in a vast array of human diseases in addition to cancer, it is important to identify markers for PI3K pathway activation that are suitable for clinical use.
SUMMARY OF THE INVENTION
[0007] In one aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy, the method comprising: (a) determining gene expression level in a sample of tumor cells of one or more genes selected from the group consisting of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), phosphatase and tensin homolog (mutated in multiple advanced cancers 1), stathmin 1 /oncoprotein 18, AAA domain containing 1 , DEP domain containing 1 , cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat- containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2, to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and (b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of: (i) upregulation in the profile of phosphoinositide-3 -kinase catalytic alpha polypeptide (PIK3CA), stathmin 1 /oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N- cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat- containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof; (ii) downregulation in the profile of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1 , retinoic acid induced 2, or any combination thereof, then the tumor has a high potential for malignancy.
[0008] In another aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy, the method comprising: (a) determining gene expression level in a sample of tumor cells of (i) PTEN, and (ii) one or more genes selected from the group consisting of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK.3CA), stathmin 1 /oncoprotein 18, AAA domain containing I3 DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2, to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and (b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of: (i) upregulation in the profile of phosphoinositide-3 -kinase catalytic alpha polypeptide (PIK3CA), stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat- containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof; (ii) downregulation in the profile of PTEN, AAA domain containing 1, retinoic acid induced 2, or any combination thereof, then the tumor has a high potential for malignancy.
[0009] In another aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy, the method comprising: (a) determining gene expression level in a sample of tumor cells of (i) PTEN, (ii) ATAD2, and (iii) one or more genes selected from the group consisting of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1/oncoprotein 18, AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4 A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), and chromosome 6 open reading frame 173, to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and (b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of: (i) upregulation in the profile of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, ATAD2, or any combination thereof; (ii) downregulation in the profile of PTEN, AAA domain containing 1, retinoic acid induced 2, or any combination thereof, then the tumor has a high potential for malignancy.
[0010] In another aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy, the method comprising: (a) determining gene expression level in a sample of tumor cells of (i) PTEN, (ii) ATADl, and (iii) one or more genes selected from the group consisting of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1 , importin alpha 1 ), Rac GTPase activating protein 1 , kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2, to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and (b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of: (i) upregulation in the profile of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N- cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4 A, chromosome 20 open reading frame 129, baculoviral IAP repeat- containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof; (ii) downregulation in the profile of PTEN, ATADl, retinoic acid induced 2, or any combination thereof, then the tumor has a high potential for malignancy.
[0011] In another aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy, the method comprising: (a) determining gene expression level in a sample of tumor cells of (i) PTEN, (ii) PIK3CA, and (iii) one or more genes selected from the group consisting of stathmin 1 /oncoprotein 18, AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat- containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2, to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and (b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of: (i) upregulation in the profile of PIK3CA, stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof; (ii) downregulation in the profile of PTEN, AAA domain containing 1, retinoic acid induced 2, or any combination thereof, then the rumor has a high potential for malignancy.
[0012J In another aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy, the method comprising: (a) determining gene expression level in a sample of tumor cells from a subject of (i) stathmin/oncoprotein 18, and (ii) one or more genes selected from the group consisting of phosphoinositide-3 -kinase catalytic alpha polypeptide (PIK3CA), phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1 , DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1 , importin alpha 1 ), Rac GTPase activating protein 1 , kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2, to thereby provide a gene expression profile for the rumor, wherein the profile comprises the expression level of the one or more * genes; and (b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of: (i) upregulation in the profile of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1/oncoprotein 18, DEP domain containing 1 , cadherin 12 type 2 (N- cadherin 2), karyopherin alpha 2 (RAG cohort 1 , importin alpha 1), Rac GTPase activating protein 1 , kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat- containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof; (ii) downregulation in the profile of phosphatase and tensin homolog (mutated in multiple advanced cancers 1 ), AAA domain containing 1, retinoic acid induced 2, or any combination thereof, then the tumor has a high potential for malignancy. [0013] In another aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy, the method comprising: (a) determining gene expression level in a sample of tumor cells of (i) stathmin/oncoprotein 18, (ii) ATAD2, and (iii) one or more genes selected from the group consisting of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), and chromosome 6 open reading frame 173, to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and (b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of: (i) upregulation in the profile of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, ATAD2, or any combination thereof; (ii) downregulation in the profile of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1, retinoic acid induced 2, or any combination thereof, then the tumor has a high potential for malignancy.
[0014] In another aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy, the method comprising: (a) determining gene expression level in a sample of tumor cells of (i) stathmin/oncoprotein 18, (ii) ATADl, and (iii) one or more genes selected from the group consisting of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), phosphatase and tensin homolog (mutated in multiple advanced cancers 1), DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4 A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2, to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and (b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of: (i) upregulation in the profile of phosphoinositide-3 -kinase catalytic alpha polypeptide (PIK3CA), stathmin 1 /oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N- cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat- containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof; (ii) downregulation in the profile of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), ATADl, retinoic acid induced 2, or any combination thereof, then the tumor has a high potential for malignancy.
[0015] Another aspect of the invention provides a method for identifying whether a tumor has a high potential for malignancy, the method comprising: (a) determining gene expression level in a sample of tumor cells of (i) stathmin/oncoprotein 18, (ii) PIK3CA, and (iii) one or more genes selected from the group consisting of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2, to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression, level of the one or more genes; and (b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of: (i) upregulation in the profile of PIK3CA, stathmin 1/oncoprotein 18, DEP domain containing 1 , cadherin 12 type 2 (N- cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat- containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof; (ii) downregulation in the profile of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1, retinoic acid induced 2, or any combination thereof, then the tumor has a high potential for malignancy.
[0016] In one aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of PTEN in a sample of tumor cells, wherein detection of PTEN expression below normal indicates that the tumor has high potential for malignancy.
[0017] In another aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of (i) PTEN, and (ii) ATAD2 in a sample of rumor cells, wherein detection of PTEN expression below normal and ATAD2 expression above normal indicates that the tumor has high potential for malignancy.
[0018] In another aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of (i) PTEN, and (ii) ATADl in a sample of tumor cells, wherein detection of PTEN expression below normal and ATADl expression below normal indicates that the tumor has high potential for malignancy.
[0019] In another aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of (i) PTEN, and (ii) PIK3CA in a sample of tumor cells, wherein detection of PTEN expression below normal and PIK3CA expression above normal indicates that the tumor has high potential for malignancy.
[0020] In another aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of stathmin in a sample of tumor cells, wherein detection of stathmin expression above normal indicates that the tumor has high potential for malignancy.
[0021] In another aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of (i) stathmin, and (ii) ATAD2 in a sample of tumor cells, wherein detection of stathmin expression above normal and ATAD2 expression above normal indicates that the tumor has high potential for malignancy. [0022] In another aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of (i) stathmin, and (ii) ATADl in a sample of tumor cells, wherein detection of stathmin expression above normal and ATADl expression below normal indicates that the tumor has high potential for malignancy.
[0023] In another aspect, the invention provides a method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of (i) stathmin, and (ii) PIK3CA in a sample of tumor cells, wherein detection of stathmin expression above normal and PIK3CA expression above normal indicates that the tumor has high potential for malignancy.
[0024] In one embodiment, a method of the invention further comprises determining gene expression level in the sample of tumor cells of one or more genes selected from the group consisting of Cyclin B2, NIMA (never in mitosis gene a)-related kinase 2, BUBl budding uninhibited by benzimidazoles 1 homolog (yeast), Cyclin A2, Cell division cycle 25 A, Cell division cycle 2, Gl to S and G2 to M, Polo-like kinase 4 (Drosophila), and CHKl checkpoint homolog (S. pombe).
[0025] In one embodiment, at least one gene is in the profile. In another embodiment, the profile comprises 52, 51, 50, 49, 48, 47, 46, 45, 44, 43, 42, 41 , 40, 39, 38, 37, 36, 35, 34, 33, 32, 31, 30, 29, 28, 27, 26, 25, 24, 23, 22, 21, 20, 19, 18, 17, 16, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, or 2 genes. In another embodiment, upregulation of gene expression comprises at least about a 20%, 25%, 30%, 35%, or 40% increase in gene expression, hi another embodiment, downregulation of gene expression comprises at least about a 20%, 25%, 30%, 35%, or 40% decrease in gene expression. In another embodiment, a change in expression is within a p- value of about 0.05, 0.4, 0.3, 0.2, 0.1 or less. In one embodiment, upregulation of gene expression comprises at least about a 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, or 2 fold increase in gene expression. In another embodiment, downregulation of gene expression comprises at least about a 0.8, 0.7, 0.6, 0.4, or 0.3 fold decrease in gene expression.
[0026] In one embodiment, the reference profile is obtained from a sample of non- tumor cells. In another embodiment, the reference profile is obtained from a sample of cells from a PTEN+ tumor. [0027] In another embodiment, the determining comprises measuring mRNA level expressed by one or more of the genes. In another embodiment, the determining comprises measuring protein level expressed by one or more of the genes. In another embodiment, the determining comprises immunohistochernistry, immunoblotting, quantum dots, or nucleic acid hybridization.
[0028] In one embodiment, the tumor is a breast tumor, a prostate tumor, a bladder tumor, a lung tumor, or a diffuse large B-cell lymphoma tumor. In another embodiment, the tumor comprises breast cancer. In another embodiment, the breast cancer comprises non- hereditary basal-like breast cancer. In another embodiment, the breast cancer comprises hereditary BRCAl -related basal-like breast cancer. In another embodiment, the tumor has reduced expression of PTEN.
[0029] In one aspect, the invention provides a method for predicting whether a subject with a tumor will respond to treatment comprising a PI3K inhibitor, the method comprising: (a) determining gene expression level in a sample of tumor cells from a subject of one or more genes selected from the group consisting of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), phosphatase and tensin homolog (mutated in multiple advanced cancers 1 ), stathmin 1 /oncoprotein 18, AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2, to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and (b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of: (i) upregulation in the profile of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1 , importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof; (ii) downregulation in the profile of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1, retinoic acid induced 2, or any combination thereof, then the subject is predicted to respond to treatment comprising a PI3K inhibitor.
[0030] In one embodiment, the PI3K inhibitor comprises rapamycin, or a derivative thereof. In antoher embodiment, the subject has reduced expression or no detectable expression of a BRCAl gene. In antoher embodiment, the subject has reduced expression or no detectable expression of an estrogen receptor (ER) gene, a HER2 gene, and a progesterone receptor (PR) gene.
[0031] In another aspect, the invention provides a method for determining whether a test compound inhibits a PI3K pathway, the method comprising: (a) contacting a cell that has a PI3K pathway with a test compound; (b) determining gene expression in the cell of one or more genes selected from the group consisting of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1 /oncoprotein 18, AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTP ase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4 A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2; and (c) comparing whether the expression of the one or more genes of step (a) is inhibited compared to the expression of the one or more genes in a cell in the absence of the test compound, so as to identify whether the test compound inhibits a PI3K pathway.
[0032] In one embodiment, the determining comprises measuring mRNA level expressed by one or more of the genes. In another embodiment, the determining comprises measuring protein level expressed by one or more of the genes. In another embodiment, the test compound comprises a PI3K inhibitor. In another embodiment, the PI3K inhibitor comprises rapamycin or a derivative thereof.
[0033] Another aspect of the invention provides a method for identifying whether a tumor has a high potential for malignancy, the method comprising detecting a level of stathmin in one or more cells of the tumor, wherein detection of stathmin above normal indicates that the tumor has high potential for malignancy. In one embodiment, the detecting comprises detecting stathmin mRNA, protein or both. In another embodiment, a method further comprises determining gene expression level in the sample of tumor cells of one or more genes selected from the group consisting of Cyclin B2, NIMA (never in mitosis gene a)-related kinase 2, BUBl budding uninhibited by benzimidazoles 1 homolog (yeast), Cyclin A2, Cell division cycle 25 A, Cell division cycle 2, Gl to S and G2 to M, Polo-like kinase 4 (Drosophila), and CHKl checkpoint homolog (S. pombe).
(0034] In another aspect, the invention provides a method for determining whether a test compound inhibits activation of a PI3K pathway, the method comprising: (a) contacting a cell that has a PI3K pathway with a test compound; (b) measuring expression of stathmin in the cell of step (a); and (c) determining whether the expression of stathmin in the cell of step (a) is inhibited compared to the expression of stathmin in a cell in the absence of the test compound, so as to identify whether the test compound inhibits activation of a PI3K pathway. In one embodiment, the determining comprises measuring mRNA level expressed by one or more of the genes. In another embodiment, the determining comprises measuring protein level expressed by one or more of the genes.
[0035] In one aspect, the invention provides methods for identifying the global gene expression profile associated with the presence or absence of the PTEN tumor suppressor. In one aspect, the global gene expression profile is of a breast tumor cell or tissue. Such methods can be used in clinical diagnosis/screening of cancer, grades of cancer, or predisposition to cancer.
[0036] In another aspect, the invention provides methods for identifying expression profiles that serve as useful diagnostic markers as well as markers that can be used to monitor activation of the PTEN/PI3K signaling pathway. This signature in turn can be used to monitor disease states, disease progression, drug toxicity, drug efficacy, and drug metabolism.
[0037] In another aspect, the invention provides methods that identify expression profiles that serve as useful prognostic tools for identifying patients with a predicted favorable or poor survival status. In one aspect, the expression profiles relates to PTEN and genes affected by PTEN expression or lack of expression.
[0038] In another aspect, the invention provides subsets of and individual PTEN signature genes that may be important for tumorigenesis, and may be useful as potential therapeutic targets. Further, they may be utilized as laboratory/clinical diagnostic markers indicating activation of the PTEN/PI3K pathway and/or as a clinical test with diagnostic, pharmacokinetic, and prognostic value. For example, the PTEN signature genes provided by the invention may be used in screening assays to identify compounds that regulate activation of the PTEN/PI3K pathway.
[0039] In one aspect, the invention provides the finding that the tumor suppressor
PTEN is lost in a significant fraction of non-hereditary basal-like breast cancer as well as hereditary BRCAl -related basal-like breast cancer.
[0040] In one aspect, the invention provides a method for identifying a group of non- hereditary breast cancers, wherein these cancers have a poor prognosis and may be particularly sensitive to therapies against the PTEN pathway.
[0041] In one aspect, the invention methods for treating BRCAl -related breast cancer by using agents/therapies against the PTEN pathway. Prophylactic therapy against the PTEN pathway to delay or prevent cancer may be an effective alternative to more extreme and invasive therapies such as prophylactic bilateral mastectomy/oophorectomy for persons who are carriers of the BRCAl germline mutation.
[0042] In one aspect, the invention provides methods of delaying or preventing cancer in subjects with BRCAl germline mutations, the methods comprising a prophylactic therapy to the PTEN pathway.
[0043] In one aspect, the invention provides a method for measuring disease progression, drug toxicity, drug efficacy, or drug metabolism, the method comprising monitoring activation of the PTEN/PI3-K signaling pathway.
[0044] In one aspect, the invention provides stathmin as a prognostic marker for breast cancer.
[0045] In one aspect, the invention provides stathmin as a therapeutic target.
|0046] In one aspect, the invention provides methods of clinical diagnosis, the methods comprising identifying stathmin expression (or correlating stathmin mRNA levels to breast cancers with and without recurrence within 5 years of diagnosis) as being indicative of the PTEN/PD-K pathway.
BRIEF DESCRIPTION OF THE FIGURES
[0047] Figures IA - IE: Biologic and clinical relevance of the PTEN/PI3-K signature in breast carcinoma. Figure IA. Hierarchical clustering of rumor samples
(columns) was performed using the top 246 signature genes (rows) with an average /'-value < 0.02. The two major tumor dendrogram clusters, 'Signature Absent' and 'Signature Present' are indicated by blue and red, respectively. The heatmap represents relative expression ratios in Iog2 space, with relative overexpression in red, underexpression in blue, and missing values in grey. Selected gene symbols are shown to the right of the heatmap with their corresponding signature consensus-rank and average F-value (APV). PTEN immunohistochemistry (IHC), PIK3CA kinase domain (KD) and C2 or helical domain (CDfHD) mutations, ERBB2 overexpression, and estrogen receptor (ER) status are indicated by the boxes below the sample dendrogram (key to right). Figure IB. Scatterplot of the average mRNA expression ratio of the 3 ATADl reporters on the microarrays plotted against the PTEN message levels in Iog2 space. The datapoints were linearly regressed and the Pearson correlation r and P- value are shown. Figure 1C. PTEN anά ATADl gene expression levels appear to be a good marker of the Signature Present group and may help to explain PTEN IHC positive and negative cases that cluster discordantly. Arrows identify cases clustering in the Signature Present group that are PTEN positive by IHC but appear to have low PTENI ATADl message levels, and arrowheads identify Signature Absent PTEN IHC- negative cases that have high PTEh rι 'ATADl levels. Figure ID. Gene set enrichment analysis (GSEA) (Subramanian et al., 2005) identifies gene sets with significant enrichment in the two PTEN IHC status groups. The black bar graph represents the GSEA normalized enrichment score, with positive scores indicating enrichment towards PTEN'hc~ tumors, and negative scores towards PTENlhc+ tumors. Below each black bar, the green bar graphs indicate on a — logio scale the respective nominal P-value computed by GSEA for the enrichment. The gene sets, ordered by ascending P-value (those above the green dashed line have a /><0.05), are indicated to the right of the bar graphs and the number of genes matching the microarrays from the set is indicated within parentheses; see text and Methods for description of the gene sets. Figure IE. Patients with tumors belonging to the two classes, Signature Present (red) and Signature Absent (blue), corresponding to Figure IA, have a significant difference in the rate of distant metastasis formation. Censored events are indicated by tick marks and the P- value is calculated by the log-rank test.
[0048] Figures 2A - 2D: Prediction of breast cancer outcome by the PTEN/PI3-K signature in 2 independent datasets. The top 246 signature genes were mapped to 2 independent breast cancer datasets (van de Vijver et al., 2002; Sotiriou et al., 2003)and nearest centroid classification was used to assign each predicted tumor to either the Signature Present (red line) or Signature Absent (blue line) class. Log-rank P-value at complete follow- up is indicated in the top right corner of each graph, and the P- value at 5-year follow-up is indicated below the double-arrow time bars. Results of Cox regression survival analyses are also presented in Table 1 of Example 1. Figures 2 A and 2B. Kaplan-Meier analysis on the classification of 295 Dutch breast cancers (van de Vijver et al., 2002) with respect to distant disease-free survival (DDFS) and overall survival (OS), respectively. Figures 2C and 2D. Kaplan-Meier analysis on the classification of 99 breast cancers (Sotiriou et al., 2003) with respect to recurrence-free survival (RFS) and breast cancer-specific survival (BCS). For Kaplan-Meier estimates, analogous 3-group analyses are presented in Example 1.
[0049] Figures 3 A - 3D: Prediction of outcome by the PTEN/PI3-K signature in independent datasets of prostate, bladder, and lung carcinoma, and diffuse large B-cell lymphoma. Analysis and annotation of graphs are as in Figures 2 A — 2D and Cox regression survival analyses are also presented in Table 1 of Example 1. Figure 3A. Kaplan-Meier analysis on the classification of 79 prostate cancers (Glinsky et al., 2004) with respect to distant disease-free survival (DDFS). Figure 3B. Kaplan-Meier analysis on the classification of 80 bladder cancers (Blaveri et al., 2005) with respect to overall survival (OS). Here, a 3- group classification was used, as detailed in the text, and the log-rank P-value is given for the comparison between the Signature Present and Absent curves only. Figure 3C. Kaplan-Meier analysis on the classification of 86 lung cancers (Beer et al., 2002) with respect to OS. Figure 3D. Kaplan-Meier analysis on the classification of 240 diffuse large B-cell lymphomas (DLBCL) (Rosenwald et al., 2002) with respect to OS. As in Figure3B, here, a 3- group classification was used. For Kaplan-Meier estimates, analogous 3-group analyses are presented in Example 1.
[0050] Figures 4A - 4H: Stathmin as a marker of the PTEN/PI3-K signature. Figure
4A. Western blots for signature genes across a panel of 8 breast cancer cell lines. The 4 PTEN mutant lines are indicated. Three lines harbor PIK3CA mutations (Saal et al., 2005): MDA-MB-361 (E545K), and MDA-MB-453 and UACC-893 (H 1047R). Adenovirus- mediated overexpression of PTEN (+PTEN) but not lacZ (+lacZ) regulates signature genes. Reduction of phospho-AKT was used as an indicator of functional PTEN induction. Figure 4B. Representative stathmin IHC results for 6 primary breast carcinomas. Composite stathmin IHC score, PTEN IHC status, ERBB2 status, PIK3CA mutational status, and distant tumor recurrence with the time to recurrence in parentheses is shown below each panel. Figure 4C. Stathmin immunostaining scores are significantly higher in PTENlhc~ tumors vs.
PTENlhc+ tumors among all 181 breast cancers analyzed for both proteins. P-value is calculated by the rank-based Mann- Whitney C-test. Figure 4D. Stathmin is significantly overexpressed in PTENlhc~ tumors vs. PTENlhc+ tumors among the 62 ER negative tumors. Figure 4E. Stathmin IHC levels are correlated to the presence of the PTEN/PI3-K signature. ' The hierarchical clustering dendrogram, tumor annotations, and STMNl {stathmin) and PTEN message levels are taken' from Figure IA. Stathmin protein levels are centered at white around the median score, 6, with higher and lower scores pseudocolored in red and blue, respectively (key provided to the right). Figures 4F and 4G. Stathmin IHC levels closely track stathmin message levels, and are significantly inversely related to PTENmRNA expression. The average (ratio) for the two stathmin reporters was used. The data was linearly regressed and the Pearson correlation r and P- values are indicated. Figure 4H. High stathmin protein levels predict poor distant disease-free survival (DDFS), independent of other prognostic factors such as ER status and lymph node involvement (see text and Table 1). Log-rank F-values for complete follow-up (top right corner) and the 2- and 5-year intervals are shown.
[0051] Figure 5: Three-fold cross-validation overview. A 3-fold cross-validation design was used to train a committee of SVMs to predict the PTEN status of the rumors and generate a consensus list of ranked PTEN signature genes. The procedure is carried out as follows: (1) partition the tumors randomly into 3 equal sized groups with roughly an equal distribution of the sample labels of interest in each group; (2) select 2 of the groups for training, and the remaining group is left as the prediction test set; within the training set, rank the genes according to the correlation of their expression measurements to PTEN status using the Mann-Whitney U-test and use this ranked gene expression matrix as input to train a SVM; (3) data from the test group are fed to the SVM and each test tumor is classified as PTEN positive or negative; (4) steps 2-3 are repeated twice, such that each time a different group is put aside as the test group; (6) steps 1 -4 are repeated 9 more times, in which each time a new random partition into thirds is made. In such a way, 30 SVMs in all were trained and each tumor received a prediction output from 10 SVMs.
[0052] Figures 6 A - 6H: Additional Kaplan-Meier analysis for independent datasets. Three group (Figures 6A - 6F) and two group (Figures 6G and 6H) classification survival analyses corresponding to Figures 2 and 3 are shown. Log-rank P-values are given for the comparison between the 'robust' Signature Present group and the 'robust' Signature Absent group for graphs with three groups, with the total number of cases compared in parentheses. Figure 6A. Distant disease-free survival (DDFS) curves for van de Vivjer 295 breast tumor set, P«0.0001 (215). Figure 6B. Overall survival (OS) curves for van de Vivjer 295 breast tumor set; P«0.0001 (215). Figure 6C. Recurrence-free survival curves for Sotiriou 99 breast tumor set; P=0.0718 (46). Figure 6D. Breast cancer-specific survival curves for Sotiriou 99 breast tumor set; P=0.0534 (46). Figure 6E. DDFS curves for Glinsky 79 prostate tumor set; P=0.0814 (19). Figure 6F. OS curves for Beer 86 lung tumor set; P=0327 (54). Figure 6G. OS curves for Blaveri 80 bladder tumor set; /»=0.289 (80). Figure 6H. OS curves for Rosenwald 240 diffuse large B-cell lymphoma set; P=O.822 (240).
[0053] Figure 7: Hierarchical clustering using top 246 PTEN signature genes (see
Example 2).
[0054] Figure 8: Result of gene set enrichment analysis (see Table 11).
[0055] Figures 9A — 9E: Figure 9A. Hierarchical clustering of 105 tumors using the top 246 PTEN signature genes was used to identify two tumor classes, PTEN+-like and PTEN~-like. The identified tumor classes have a significant difference in distant disease-free (or metastasis-free) survival. Figure 9B. Overall survival (OS) curves for van de Vivjer 295 breast tumor set. Figure 9C. Distant disease- free survival (DDFS) curves for van de Vivjer 295 breast tumor set. Figure 9D. Breast cancer-specific survival curves for Sotiriou 99 breast tumor set. Figure 9E. Recurrence-free survival curves for Sotiriou 99 breast tumor set.
[0056] Figures 1OA - 1OG: Figure 1OA. PTEN protein is lost in the majority of
CK5/14+ basal breast cancer. Left, discovery dataset; right, validation dataset. Black bars indicate the rate of PTEN loss in basal positive tumors, the grey bars indicate the rate of PTEN loss in the non-basal tumors. * PO.05; ** P>0.01; *** PO.0001. Figure 1OB. BRCAl mRNA expression is significantly lower in 10 PTEN-IH C-negative sporadic basal breast tumors and 7 PTEN-IHC-positive sporadic basal breast tumors. Green = underexpression; red = overexpression. Figure 1OC. Array CGH identifies a BRCAl- methylated sporadic basal tumors with PTEN homozygous deletion. Figure 10D. Immunohistochemistry example for one BRCAl -hereditary breast tumor. Note the positive staining of normal ductal epithelium (arrow) as compared to the completely PTEN-null staining of the tumor cells (arrowhead). Pie-chart indicates percentage of BRCAl tumors with significant reduction of PTEN protein levels. Figure 1OE. Schema for split-probe FISH analysis to the PTENlocus at 10q23.31. Figure 1OF. Split-probe results for BRCAl-mutant cell lines MDA-MB-436 and SUM- 149 indicating breakage at the PTEN locus. Figure 1OG. High density Agilent oligonucleotide array CGH analysis of one PTEN-IHC-null BRCAl breast tumor indicating a complex duplication of the PTEN gene.
[0057] Figure 11: Hierarchical clustering of breast carcinomas using PTEN/PI3K pathway markers and clinicopathologic markers. Inherent relationships between the pathway markers (PIK3CA mutations, HER2 amplification, PTEN alteration, and EGFR overexpression) and clinical markers (ER, PgR, TP53 mutation status, and CK5/14 basal status) across 116 breast tumors were visualized using complete linkage clustering with the Pearson correlation distance metric. In the heatmap, black represents negative/wild-type status for the given marker, red represents positive or mutant in the case of PIK3CA, and green represents loss of tumor suppressor (PTEN and TP53). Grey signifies missing data. PTEN mutants are further indicated by an asterix. Tumors (columns) with similar profile of marker status cluster together, and markers with similar profiles across tumors cluster together (rows). Four main clusters are evident, marked A, B, C, and D. Cluster A closely corresponds to the basal subtype of breast cancer; cluster B is a mixture of HER2 and luminal B-type tumors, cluster C appears to relate to luminal A, and cluster D to luminal B/C. These clusters and subclusters are discussed in the text. Below the heatmap are mapped additional clinicopathological annotations for each case, including Ki67 status, S-phase fraction (SPF), tumor grade (grade), and lymph node status. For these annotations, red indicates positive, high for SPF, or grade 3; white indicates negative, low for SPF, or grade 1; and violet represents intermediate SPF or grade 2.
[0058] Figure 12: Immunoblotting illustrates differential p-Akt serine 473 activation in breast cancer cell lines harboring PIK3CA helical domain versus kinase domain mutations, consistent with the PTEN signature clustering results. MCF-7 and MDA-MB-361 harbor the helical domain E545K PIK3CA mutation, SUM-159 the H1047L kinase domain mutation, and UACC-893 the H1047R mutation.
[0059] Figures 13A — 13D: Kaplan-Meier analysis for the independent Dutch breast cancer dataset of 295 patients partitioned by STMNl mRNA expression (Figures 13A - 13B) or PTEN mRNA expression (Figures 13C - 13D) using the median expression value as the threshold (high=top 50th percentile; low=bottom 50th percentile). (Figures 13A and 13C) Results for distant disease-free survival and (Figures 13B and 13D) overall survival. Log- rank P- value at complete follow-up is in the top right corner of each graph, and the P- value at the 5-year time interval is below the double-arrow time bars. [0060] Figures 14A - 14D: The PTEN/PI3K pathway regulates stathmin protein levels. Figure 14A. Immunoblotting analysis of stathmin downregulation following adenovirus-mediated PTEN expression (ad-PTEN) in MDA-MB-468 PTEN-null cells compared to control (ad-lacZ). Figure 14B. Stathmin is downregulated in breast cancer lines by treatment with the PI3K-inhibitor LY294002 (LY; 20 μM) for 4 days as detected by immunoblotting. Figure 14C. Immnoblotting analyses show, in vivo, stathmin is downregulated upon PTEN- induction with doxycyclin (Dox) in MDA-468TR-PTEN xenograft tumors and (Figure 14D) following treatment of mice with MDA-MB-468 tumors with the PI3K-inhibitor PWT-458 (100 mg/kg). For C and D, biological replicates were performed for each treatment (independent xenografts from separate animals were analyzed in each sample lane).
DETAILED DESCRIPTION OF THE INVENTION
[0061] It is a discovery of the invention that expression patterns (i.e., upregulation or downregulation) of a particular set of genes can be used to predict whether a tumor will have a high potential for malignancy. Changes in expression of the genes in the gene set are indicative activation of the oncogenic phophatidylinositol 3-kinase (PI3K) pathway and loss of PTEN, a negative regulator or PI3K. Activation of the PI3K pathway indicates an aggressive tumor phenotype. However, there are currently no reliable markers of pathway activation that can be used to predict clinical outcome. Thus, it is a further discovery of the invention that expression patterns of the set of genes can be used, for example, to predict a patient's clinical outcome, to monitor a patient's response to treatment, or to stratify or classify types of cancer in order to target or individualize treatment. A tumor with a high potential for malignancy indicates an aggressive tumor phenotype and a poor clinical outcome for the patient.
[0062] The invention provides gene expression signatures of PTEN loss that can be used to identify tumors with activation of the PI3K pathway. The invention also provides that the degree of correlation that a rumor expresses a PTEN signature provided by the invention is directly related to the malignant potential of the tumor. A gene expression profile associated with the presence or absence of the PTEN tumor suppressor was identified by examining gene expression in breast carcinoma characterized for PTEN status using microarrays. [0063] The invention further identifies the oncoprotein stathmin as a useful marker of
PI3K pathway activation and an important therapeutic target in the treatment of tumors with PBK activation. Stathmin was found to be overexpressed in PTEN negative tumors. The invention provides stathmin as a marker for prognosis prediction in breast cancer.
[0064] The compositions and methods provided by the invention can be used to determine whether a patient will be responsive to a particular therapy. For example, a gene expression pattern in a tumor sample may implicate a cellular pathway, thus allowing a clinician to treat the patient with one or more therapeutic agents that target the implicated pathway. The therapeutic agents that target the pathway may be administered to the patient in combination with other therapies. The advantages to such targeted treatment of tumors has several advantages, for example increased efficiency and decreased cost of treatment.
Terms
[0065] As used herein, a "profile" means a set of one or more genes for which expression levels have been determined in a cell or tissue sample.
[0066] As used herein, "reference profile" means a profile to which another profile is compared. A reference profile can be obtained from, for example, non-tumor cells or cells from a PTEN+ tumor.
[0067] As used herein, "post-treatment profile" means a profile obtained from a tumor in a patient who has undergone treatment for the tumor. Non-limiting examples of treatment include chemotherapy and radiation. Profiles can be obtained, for example, by measuring expression levels in a cell or tissue sample, such as by microarray analysis.
[0068] A tumor or tumor cell has a "high potential for malignancy" if at least one gene in the gene signature exhibits an increase in expression of at least about 30%, or a decrease in expression of at least about 30%, or both, in the tumor or rumor cell.
[0069] As used herein, "expression level" means a measurement of DNA, RNA, protein, or any combination thereof, in, for example, a subject, a cell, or a tissue sample.
PTEN Signature Genes
[0070] Genes in the PTEN signature can be used as targets of tumorigenesis and as clinical markers for activation of the PI3K pathway, clinical outcome markers, and markers for drug toxicity, drug efficacy, and monitoring disease states. Currently, no commercial assays exist to detect activation of the PI3K pathway, and laboratory assays are inadequate due to poor quality and poor reproducibility. For example, outcome prediction for breast cancer is a major clinical problem. Embodiments of PTEN signatures provided herein (Tables 2 and 17), and the expression levels of subsets of genes or individual genes in the PTEN signature, may be useful as a clinical diagnostic test to identify patients with tumors with a high potential for malignancy who would benefit from directed therapies against the PBK pathway.
Figure imgf000023_0001
Table 17. Embodiment of a PTEN gene signature
Level in PTEN-negalives or
Fold Change P-value PI3K pathway active Gene Name Gene Symbol RefSeq
1 992946525 0000005 UP Cadheπn 12, type 2 (N-cadherin 2) CDH 12 NM_004061
I 907051759 0 000161 UP Kinesin family member 2C K.IF2C NM_00684S
I 852468486 0 000007 UP NIMA (never in mitosis gene a)-related kinase 2 NEK2 NMJW2497
I 851979334 0 000055 UP Baculoviral IAP repeat-containing 3 BIRC3 NM 182962
Figure imgf000024_0001
Phosphatase and tensin homolog (mutated in multiple
0.64640981 1 0 EX)WN advanced cancers 1) PTEN NM_000314
0.631772918 0.000931 DOWN Endothelin converting enzyme 1 ECEl NMJ)Ol 397
0.628609944 0.00018 DOWN Dickkopf homolog 3 (Xcnopus lacvis) DK.K3 NMJH3253
0.626918583 0.000063 DOWN EPH receptor A4 EPHA4 NM_004438
0.620625552 0.00071 1 DOWN Dual specificity phosphatase 6 DUSP6 NMJX)1946
0.555536266 0.000136 DOWN Protein kinase C binding protein I PRKCBPI NMJ 83048
0.516191754 0.000277 DOWN Ectonucleotide pyrophosphatase/phosphodiesterase 3 E ENNPPPP33 AA678335
0.489009704 0.000756 DOWN Myosin, heavy polypeptide 11 , smooth muscle MYHU NMJ522844
0.465927051 0.000028 DOWN Fibromodulin FMOD NM 002023
[0071J In one embodiment of the invention, the profile is phosphoinositide-3-kinase catalytic alpha polypeptide, phosphatase and tensin homolog (mutated in multiple advanced cancers 1), stathmin 1/oncoprotein 18, AAA domain containing 1 , DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, and phosphoinositide-3- kinase catalytic alpha polypeptide.
[0072] In other embodiments of the invention, the profile is any combination of 17,
16, 15, 14, 13, 12, 1 1, 10, 9, 8, 7, 6, 5, 4, 3 or 2 genes, or any 1 gene, selected from the group consisting of phosphoinositide-3-kinase catalytic alpha polypeptide, phosphatase and tensin homolog (mutated in multiple advanced cancers 1), stathmin 1/oncoprotein 18, AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat- containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, and phosphoinositide-3-kinase catalytic alpha polypeptide.
[0073] In one embodiment, the profile comprises one or any combination of more than one genes listed in Tables 2, 3, or 17. In another embodiment, the profile comprises PTEN. In another embodiment, the profile is PTEN. In another embodiment, the profile is stathmin.
Cell cycle genes [0074] The cell cycle genes listed in Table 3 were also found to be overexpressed when a poor prognosis PTEN gene signature was observed in a sample. Since poor prognosis gene expression profiles are likely to be identified in a highly proliverative tumor, cell cycle genes may be upregulated as a result of cell proliferation, thus upregulation of cell cycle genes alone will not be indicative of a particular underlying mechanism which can be targeted by specific treatments. However, in embodiments of the invention, one or more cell cycle genes listed in Table 3 can be measured together with one or more PTEN signature genes (Table 2).
Figure imgf000026_0001
Determination of gene expression and gene expression profiles
[0075] Microarray technology is a well-known, widely-used method in the art to measure the expression of a large number of genes simultaneously. A DNA microarray (also referred to as, e.g., biochip, DNA chip, DNA microarray, gene array, expression chip, GeneChip®, and genome chip) provides a technique for matching known and unknown DNA samples or quantifying gene expression based on DNA and RNA base-pairing rules (i.e., A-T and G-C for DNA; A-U and G-C for RNA) or hybridization. This technique allows one to monitor the expression level of thousands of genes simultaneously or to identify unknown genes. A DNA microarray is a collection or arrangement of DNA or segments of DNA (for example, cDNA, oligonucleotides, or PCR products) attached to a solid surface, such as glass, plastic, nylon or silicon. The microarray can be used to detect the expression of genes in a sample by measuring the amount of mRNA bound to each site on the array. The amount of mRNA bound to the spots on the microarray is measured, typically by an automated process, generating a profile of gene expression in the sample (see Example 1). Gene expression profiles can be used, for example, to compare the expression of genes between a normal sample and a pathologic sample to determine changes in gene expression that may be indicative of a pathologic state.
[0076] Preparation of the DNA microarray, purification of mRNA from a sample and quantification of expression are methods known to one skilled in the art. {See Wu and Dewey (2006) From microarray to biological networks: analysis of gene expression profiles. Methods MoI Biol 316:35-48; Hoheisel (2006) Microarray technology: beyond transcript profiling and genotype analysis. NatRev Genet 7(3):200-210; Gabig and Wegrzyn (2001) An introduction to DNA chips: principles, technology, applications and analysis. Acta Biochim Pol 48(3):615-622; Krajewski and Bocianowski (2002) Statistical methods for microarray analysis. J Appl Genet 43(3):269-278; Kaminski and Friedman (2002) Practical approaches to analyzing results of microarray experiments. Am J Respir Cell MoI Biol 27(2): 125-132; and U.S. Patent Nos. 6,953,551, 6,706,867, 6,996,476, 6,731,781.)
Methods for treating or preventing cancer
[0077J The invention provides therapies to PTEN pathway components as a method for treating tumors. The PTEN signature genes in Table 2 present genes which may be used as therapeutic targets to regulate the activation of the PI3K pathway in tumors. The clinical outcome of a tumor may be improved by decreasing or inhibiting the pathway activation via targeting one or more of the genes in the PTEN gene signature.
[0078] The invention also provides the finding that there is a high rate of loss of
PTEN among BRCAl hereditary tumors. Thus, in certain embodiments, the invention relates to prophylactic therapy to the PTEN pathway as a treatment for BRCAl germline carriers to delay or prevent onset of cancer. Due to the fact that basal-like breast cancers are estrogen receptor negative and HER2 negative, few good therapeutic options exist for these women.
[0079] The invention provides results that show that therapy to the PTEN pathway components may be effective. Furthermore, for BRCAl germline carriers, the risk of cancer is much higher than in the general population, and many people opt to have prophylactic bilateral mastectomy and/or oophorectomy to reduce their risk. A low-dose or infrequent regimen of anti-PTEN pathway therapy (e.g., rapamycin) can be an alternative minimally invasive method to reduce risk for cancer.
[0080] In one aspect the invention provides a method for delaying or preventing the onset of cancer in a subject, the method comprising administering to the subject an effective amount of a compound that increases expression of one or more genes selected from the group consisting of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1, and retinoic acid induced 2.
[0081] In another aspect, the invention provides a method for delaying or preventing the onset of cancer in a subject, the method comprising administering to the subject an effective amount of a compound that decreases expression of one or more genes selected from the group consisting of stathmin 1 /oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTP ase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat- containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2.
[0082] In another aspect, the invention provides a method for treating cancer in a subject, the method comprising administering to the subject an effective amount of a compound that increases expression of one or more genes selected from the group consisting of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1, and retinoic acid induced 2.
[0083] In another aspect, the invention provides a method treating cancer in a subject, the method comprising administering to the subject an effective amount of a compound that decreases expression of one or more genes selected from the group consisting of stathmin 1 /oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2.
[0084] In one aspect, the invention provides a method for treating a tumor in a subject, the method comprising administering to a subject in need thereof, an effective amount of an inhibitor of stathmin.
[0085] In one embodiment, the inhibitor is an antibody, a protein, a polypeptide, a peptide, a peptidomimetics, a small molecule, a DNA, a RNA, an antisense RNA, a small interfering RNA (siRNA), a double stranded RNA (dsRNA), a short hairpin RNA, a cDNA, or any combination thereof. [0086] In one embodiment, the cancer comprises breast cancer, prostate cancer, bladder cancer, lung cancer, diffuse large B-cell lymphoma, or any combination thereof. In another embodiment, the subject has reduced expression or no detectable expression of a BRCAl gene. In another embodiment, the subject has reduced expression or no detectable expression of an estrogen receptor (ER) gene, a HER2 gene, and a progesterone receptor (PR) gene.
Compound screening methods
[0087] The gene expression profiles of the invention can be used to screen for compounds that may be useful in treating poor outcome tumors with PTEN loss, PI3K activation or both. For example, if a compound is found to decrease expression of one or more genes that are overexpressed in poor outcome tumors, or increase expression of one or more genes that are underexpressed in poor outcome tumors, or both, this result may indicate that the compound may be useful for decreasing the activation of the PI3K pathway, thus improving the prognosis of a poor outcome tumor.
Stathmin expression
[0088] The invention provides that stathmin be used in connection with the PTEN gene signature to predict poor outcome tumors. The invention also provides that stathmin itself can be used to reliably identify tumors with poor outcome.
[0089] In other embodiments, the invention provides methods for analyzing stathmin protein levels in breast cancer. One method for analysis comprises immunohistomchemistry. Results provided herein show that stathmin immunohistochemical staining is associated to distant-disease free survival. In one embodiment, the invention provides methods for diagnosing a subject for cancer based on the association of stathmin mRNA levels to breast cancer with and without recurrence within 5 years of diagnosis.
[0090] In one embodiment, the invention provides methods that use stathmin to modulate the invasive potential of cancer cells and their propensity for metastasis.
[0091) In one embodiment, the invention provides a clinical diagnostic test comprising analyzing stathmin expression to identify poor-prognosis patients who could benefit from directed therapies against the PI3K pathway.
EXAMPLES EXAMPLE 1 : A PTEN/PI3-K PATHWAY SIGNATURE DETERMINES TUMOR
METASTATIC POTENTIAL
[0092] Pathway-specific targeted therapy is the future of cancer management. The oncogenic phosphatidylinositol 3-kinase (PI3-K) pathway is one of the most frequently activated pathways in solid tumors; however, currently no clinically reliable marker for PI3- K pathway activation exists. Taking advantage of the observation that loss of PTEN, the negative regulator of PI3-K, results in robust activation of this pathway, the invention provides a biologically validated microarray gene expression signature for PTEN loss from human tumor biopsies, hi clinical samples, the signature captures pathologic PI3-K pathway activation by lesions in other PI3-K pathway members, and segregates independent datasets of breast, prostate, lung, and bladder neoplasia into classes with significant differences in patient survival. One signature gene provided by the invention, stathmin, is a durable and reliable novel marker of PI3-K activity by immunohistochemistry and identifies breast rumors that metastasize early, whether or not the patient has lymph node metastases. This Example shows that oncogenic PI3-K pathway signaling is one of the most potent drivers of metastasis formation in cancer, and the signature provided by the invention can be used to elucidate clinically relevant properties of tumors and for targeted therapy patient stratification.
[0093] Cancer mortality is primarily due to the spread of tumor cells to distant sites in the body. Why some tumors progress and metastasize is a key question in the molecular biology of neoplasia, and the identification of molecular markers to stratify patients according to risk for metastasis or death is vitally important for clinical decision-making. Utilizing microarray technologies, numerous gene expression profiling studies have found tumor expression signatures discriminating cancer patients with good vs. poor outcomes (van 't Veer et al., 2002; Pomeroy et al., 2002; Singh et al., 2002; Shipp et al., 2002; Rosenwald et al., 2002; van de Vijver et al., 2002; Beer et al., 2002; Ramaswamy et al., 2003; Huang et al., 2003; Nutt et al., 2003; Eschrich et al., 2005). Despite these results, deciphering the biological basis of the signatures and how the identified risk groups may relate to sensitivity to specific therapies remains a significant challenge. Thus, if personalized medicine is to become a reality, an understanding of the contribution of individual oncogenic pathways to the development and progression of cancer is a fundamentally important concern. [0094] The oncogenic phosphatidylinositol 3-kinase (PI3-K) pathway is activated in a large proportion of multiple types of human cancer (Sansal and Sellers, 2004). Signaling through this pathway frequently occurs in solid tumors by somatic genetic alterations that functionally activate upstream tyrosine kinase receptors such as ERBB2 and EGFR, the PD- K holoenzyme itself (pi 1 Oa and p85α), or downstream members such as AKT, thus highlighting the PI3-K pathway as ideally suited for molecularly-targeted therapy. PTEN plays a pivotal role in suppressing tumors by catalyzing the opposite reaction to PI3-K, thereby reducing the active pool of the PI3-K product, the lipid second messenger phosphatitylinositol-3,4,5-trisphosphate (PIP3). Inactivation of PTEN in cancer occurs frequently by means of somatic mutation and protein silencing, and germline PTEN mutations are also the causative lesion in several cancer-predisposing syndromes (Sansal and Sellers, 2004). Moreover, animal models have demonstrated that Pten-deficient mice develop cancers of multiple organ systems (Di Cristofano et al., 1998; Suzuki et al., 1998; Podsypanina et al., 1999).
[0095] The PI3-K pathway has long been implicated in cellular processes that favor tumorigenesis. In the late- 1980s, it was shown that the cellular transforming ability of several viral oncogenes, v-src, gag-abl, and middle T, was dependent on their ability to activate PI3- K (Cantley et al., 1991). Later, it was elucidated that the PI3-K pathway contributes to malignancy by transducing a mitogenic and cell survival signal via PIP3, largely to activate the proto-oncogene AKT, a kinase with numerous targets that results in potent stimulation of cell proliferation, growth, and protection against apoptosis (Sansal and Sellers, 2004). Relevant to cancer metastasis, it has been demonstrated that tumors arising in a middle-T- driven mammary tumorigenesis mouse model are extremely metastastic and that the tumor phenotype requires activated PI3-K signaling (Webster et al., 1998). Moreover, promotion of carcinoma invasion by the α6β4 integrin requires PI3-K (Shaw et al., 1997), and mechanistically, the modulation of cell motility by the pathway is related to the asymmetrical membrane distribution of PI3-K/PIP3 and PTEN (Funamoto et al., 2002; Iijima and Devreotes, 2002; Liu et al., 2004; Vazquez et al., 2006) and can also involve a lipid phosphatase-independent activity of PTEN (Raftopoulou et al., 2004). Also important for tumor progression and metastasis seeding, evidence exists to suggest both the production of and response to angiogenic factors such as VEGF are mediated, at least in part, by PI3-K signaling (Stephens et al., 2005). [0096] The relative degree of PI3-K pathway activation in human cancers in vivo conferred by individual lesions to the pathway and even by specific types of mutations within one pathway member (e.g. PI3-K pi 10a helical domain vs. kinase domain mutations) is not clear, however the confluence of evidence indicates that loss of PTEN results in highly robust activation and deregulation of the pathway. Despite the known roles of PI3-K signaling in tumorigenesis and in biologic processes that should confer a more aggressive phenotype, the association of lesions to this pathway to markers of pathway activation and of these variables to cancer patient outcome has yielded mixed results (see e.g. Depowski et al., 2001 ; Panigrahi et al., 2004; Cicenas et al., 2005; Bose et al., 2006; Li et al., 2006; Tang et al., 2006; Shah et al., 2005; Latta and Chapman, 2002). The lack of consensus among such studies is most likely due to the heterogeneity/specificity of certain PI3-K lesions within given cancer types, and also due to the lack of a reliable integrative assay of pathway activation that is suitable for use in routine pathology samples. Stratification of cancer based on activation of oncogenic pathways will be important for directing targeted therapies (BiId et al., 2006) as well as for monitoring response to specific therapies. Indeed, the degree of PI3-K pathway activation and the mechanisms by which it is activated may be clinically relevant as levels of PTEN are predictive of response to the ERBB2 inhibitor trastuzumab and the EGFR inhibitors erlotinib and gefitinib in breast and brain cancer, respectively (Nagata et al., 2004; Mellinghoff et al., 2005).
[0097] This Example demonstrates that an in vivo gene expression signature of PTEN loss reflects robust PI3-K pathway activation and is able to identify tumors that have achieved pathologic pathway activation by varied oncogenic lesions to the pathway. The identified PTEN/PI3-K signature was then applied to discern biologically and clinically relevant subgroups of multiple forms of human cancer. One signature gene was validated at the protein l*el as a robust and durable marker of the PTEN/PI3-K signature suitable for clinical assays by immunohistochemistry (IHC), and may be a therapeutic target itself, illustrating that the signature and signature components can lead to the discovery of additional novel biological insights into PTEN/PI3-K-related signaling with relevance to clinical medicine.
PTEN Status of Clinical Samples
[0098] Around 350 stage II primary breast cancers have been analyzed by IHC for
PTEN protein levels, and separated into PTENlhc+ and PTENihc~ (significantly reduced PTEN staining in tumor cells as compared to internal control cells) groups (Saal et al., 2005). In this dataset, loss of PTEN was significantly associated to lack of estrogen receptor (ER) and progesterone receptor expression (Saal et al., 2005; Depowski et al., 2001)). Due to the known influence of ER status on gene expression in breast cancer (Gruvberger et al., 2001 ; West et al., 2001 ; van 't Veer et al., 2002), a subset of 105 cases was selected, comprising 35 PTENlhc~ and 70 PTENlhc+ tumors, for microarray analysis with consideration for hormone
Figure imgf000033_0001
Overabundance of PTEN-Associated Genes
|0099] The non-parametric Mann-Whitney Latest was applied to the entire 105 tumor microarray dataset to ascertain the presence of a PTEN-associated gene expression signal. A >10-fold overabundance of PTEN status-associated genes was found with P<0.001 (184 significant genes when only 16 are expected by chance) and >6-fold overabundance at a cutoff of /M)-Ol. The gene with the lowest P-value was PTEN itself (/^=IxIO"6).
Supervised Analysis with Support Vector Machines
[00100] To generate a more robust ranked PTEN signature gene list and to test whether a tumor's gene expression profile could be used to predict its PTEN status, a machine learning algorithm, support vector machines (SVM) (Vapnik, 1995), was applied to the gene expression data. A 3-fold cross-validation scheme with 10 overall randomizations was employed, whereby in total 30 SVMs were trained and each tumor, having been assigned 10 times to the withheld test set, received a consensus prediction based on 10 trained SVMs
(Figure 5). For each of the 30 sample subsets, the genes were ranked by the Mann- Whitney test. The area under the receiver operating characteristic (ROC) curve was used to measure prediction performance. PTEN status was predicted with high accuracy with a ROC area of 0.758 (P=O-OOOl). To test whether the successful prediction was heavily influenced by the signal from the PTEN reporter alone, PTEN was removed and the entire procedure was repeated, yielding an identical ROC area of 0.758 (/"=0.0005). Furthermore, to confirm that the successful PTEN classification was not driven by an underlying ER-related signal, the PTEN classification was repeated after removing the top 1000 ER status discriminators generated from within the dataset (see below). Verifying the independence of accurate PTEN classification on ER status, an equivalent ROC area, 0.762 (P<0.0001), was obtained. Regardless of the set of genes used, a perfect classification could not be achieved, with 40% of the tumors classified as PTENlhc~ by the SVMs being PTENihc+, suggesting the likelihood that other oncogenic "hits" could also induce the transcriptional program of PTEN loss.
Consensus PTEN/PI3-K Signature Gene List
[00101] A cross-validated gene list is inherently more robust to noise (Ein-Dor et al., 2005); therefore a consensus-ranked PTEN signature gene list was generated by sorting on the average P- value (APV) for each gene across the 30 ranked lists used for training SVMs (see Table 5). An average of P- values is not statistically equivalent to a P- value, as it is not expected that 1% of genes have an APV<0.01 by pure chance. There were 98 reporters with an APV<0.01, which is a >14-fold overabundance compared to chance (7 genes on average had APV<0.01 after permutation simulations; false discovery rate, FDR 6.9%). In the consensus-ranked PTEN signature gene list the PTEN reporter was the top discriminator (APV=5.6xlO~5). This result provides internal validation that the PTEN IHC scoring was accurate and biologically relevant. To illustrate the pattern of expression of the top ranked genes across the tumors, the top 246 reporters (APV<0.02; >15-fold overabundance) were used to partition the PTEN'hc+ and PTENlh*~ tumors by hierarchical clustering (Figure 1 A). The pattern of expression for these genes was biphasic, with the most of PTENlhc+ tumors clustered together in one main branch (denoted 'Signature Absent'), and the majority of PTENlhc~ tumors clustering in another main branch ('Signature Present'). Consistent with the SVMs, some exceptions were observed (Figure 1 A).
Influence of PIK3 CA Mutations and Protein Levels, ERBB2 Overexpression, and ER Status
[00102] To test whether other hits in the PB-K pathway recapitulate the identified gene expression program of PTEN loss, these tumors were analyzed for PIK3CA mutations and for expression of pi 10α and ERBB2 proteins. 67% (8 of 12) PIK3CA kinase domain (KD) mutants, of which 7/8 were PTENlhc+, clustered with the Signature Present tumors, whereas only 19% of helical domain (HD; 2 of 11) and C2 domain (CD; 1 of 5) mutants were in this main branch (/><0.02; note, of these 3 cases, 2 were PTENlhc~). Recent studies demonstrated that PJK3CA HD E542K and E545K mutants fail to exhibit florid hemangiosarcoma-like characteristics and grew at <25% the rate as compared to H1047R KD mutants in an in vivo chicken embryo model (Bader et al., 2006). Although PIK3CA KD and HD mutations appear to have equipotent lipid kinase activity and confer similar oncogenic properties in vitro (Ikenoue et al., 2005; Samuels et al., 2005; Isakoff et al., 2005), the present data combined with the work by Bader et al. suggest there may be differences in potency of PI3-K pathway activation between the different domain mutations in vivo. It has been shown that PIK3CA mutation and loss of PTEN are nearly mutually exclusive in breast cancer (Saal et al., 2005); the relatively infrequent cases (<3%) that harbor lesions to both tend to have non-KD PIK3CA mutations (7/8), indicating that these mutations are not as potent activators of the pathway and thus select for subsequent loss of PTEN (Saal et al., 2005). Together, these results show that mutations in the PIK3CA KD, as compared to mutations in its other domains, result in global gene expression changes that closer recapitulate the transcriptome of PTEN1 c~ tumors. Neither PTEN status nor the presence of the signature was observed to relate to the expression levels of pi 1 Oa protein. However, of the 11 PIK3CA KD mutant tumors with evaluated pi 10a protein status, 5 (45%) had high levels of pi 10a, whereas in contrast only 1 of 15 (7%) PIK3CA HD/CD mutants expressed high levels of pi 10a (P=0.054).
[00103] ERBB2 overexpression (ERBB2ihc+) did not appear to correlate to the presence of the signature, with 54% (13/24) of positive cases clustering among the PTEN'hc~ branch. Moreover, among the ERBB2lhc+ tumors with and without the signature, there was no difference in the distribution of other markers such as ER status (the majority are ER negative; ER~). The distribution of ERBB2lhc+ tumors is consistent with a recent report demonstrating no association between ERBB2 and phospho-AKT protein levels in breast cancer (Cicenas et al., 2005). In light of the reported functional links between PTEN and ERBB2 (Nagata et al., 2004), these data show that in some ERBB2 positive tumors, in the setting of retained PTEN, as yet unidentified hits to the PI3-K pathway can cooperate with ERBB2 to mimic the gene expression profile of PTEN loss. Consistent with this idea, PDKl, an important kinase that contributes to AKT activation in response to PIP3, can cooperate with PI3-K pathway lesions to amplify pathway and phenotype output, and increased DNA copy number of PDKl occurs in -25% of ERBB2-aπιplified breast tumors. ERBB2ihc+ tumors with full PI3-K pathway activation (Signature Present) may be more "addicted" (Weinstein, 2002) to the pathway and thus would be more sensitive to targeted inhibition than ERBB2lhc+ tumors with inadequate PI3-K pathway activation (Signature Absent).
[00104] Although the 105 tumor cohort was selected with consideration of
PTEN and ER status, hierarchical clustering resulted in 78% of the ER+ tumors clustering in the Signature Absent group while 66% of the ER" tumors clustered in the Signature Present group (Figure 1 A). It has been observed in an independent population-based breast cancer cohort that most ER" tumors harbor one or more aberrations that activate the PI3-K pathway (e.g. to PTEN, PIK3CA, ERBB2, PDKl, or EGFR), in contrast to ER+ rumors. More frequent activation of the PI3-K pathway in ER" tumors indicates that ER+ tumors rely on ER-related signaling to promote cellular growth and proliferation; therefore, as a corollary, ER" tumors would be more likely to select for activation of other mitogenic pathways, i.e. the PI3-K. pathway. As shown by the result of PTEN status prediction by SVMs without ER status genes and reinforced by several analyses that are described below, the signature provided by the invention does not merely reflect signaling due to the presence or absence of ER. Rather, it would appear that the identified PTEN status signature is able to integrate various lesions that singly or cooperatively activate PI3-K signaling to a similar extent to that seen with loss of PTEN.
PTEN Protein Levels Are Dictated By PTEN Message Levels
[00105] Studies were designed to determine whether the PTEN signature in breast cancer could show how PTEN is regulated in breast cancer. That the PTEN reporter is the top discriminator (APV=5.6xl0~5) shows that the primary determinant of PTEN protein level, and thus of its function, in sporadic breast cancer is its message level. Therefore, post- translational regulation resulting in accelerated PTEN degradation does not appear to be a significant factor contributing to low PTEN levels in breast cancer.
[00106] Several mechanisms may explain regulation at the transcript level.
Hypermethylation of the PTEN promoter has been reported in several tumor types including breast cancer (Garcia et al., 2004; Khan et al., 2004). ATADl, which encodes a member of the AAA domain-containing ATPase family and is the nearest neighbor (~45kb centromeric) to PTEN at 10q23.31 in a head to head orientation, was significantly downregulated in
PTENihc" tumors (APVO.004). Although lo°s of heterozygosity (LOH) of a ~1 OcM region encompassing the PTEN locus occurs in a significant fraction of invasive breast cancers, it is rarely associated with intragenic mutations of the other allele and is generally not predictive of PTEN protein expression (Bose et al., 1998; Feilotter et al., 1999; Bose et al., 2002). The absolute expression levels of ATADl were closely linked to PTEN message levels (Pearson r=0.6761, PO.0001; Figure IB), whereas the next two genes on either side of ATADl IPTEN (and within the region of frequent LOH) did not correlate to P TEN message levels (0.0195 ^ ..0.1535, P-values>0.10). Thus, ATADl message levels can serve as an accurate surrogate of P TEN message levels in breast cancer. If, as this Example shows, PTENmKNA level is the best determinant of protein level, then the relative message levels of PTEN and ATADl may be a more accurate measure of the functional status of PTEN in the tumor. As can be seen in Figure 1C, PTEN mRNA levels may help explain some of the discrepancies between PTENihc~ and PTENihc+ status and presence of the signature, and shows that induction of the signature in PTENlhc+ cases may be due in part from PTEN haploinsufficiency revealed at the message level.
Biologic Validation: the Cell Cycle, mTOR/Metabolic Processes, and /n Vitro PTEN- Downregulated Targets are Upregulated in PTENlhc" Tumors
[00107] To investigate the landscape of cellular processes and gene expression themes highlighted by the signature, the top 785 signature genes with an APV<0.05 (>9-fold overabundance) were input for gene ontology (GO) analysis using GOMiner (Zeeberg et al., 2003). One-hundred twenty-one 'biological process' GO categories were identified with an overabundance of genes upregulated in PTEN'hc" tumors (Table 6). Of these, 43% (52/122) of the categories were related to cellular metabolism, such as 'DNA metabolism' (27 overexpressed genes, i*=0.0003), Organelle organization and biogenesis' (33 genes, /»=0.0002), and 'RNA processing' (15 genes, P=O-OOo). Forty-six of the 122 (38%; including 15 of the top 20) GO categories were related to the cell cycle, including categories such as 'cytokinesis' (17 genes overexpressed, PO.0001), 'traversing start control point of mitotic cell cycle' (3 genes, P=0.0014), 'cell proliferation' (21 genes, P=0.002), and 'regulation of cell proliferation' (10 genes, P=0.07>02). The multitude of cell cycle-related genes as being upregulated in the PTENlhc~ tumors supports prior evidence that these tumors are more highly proliferative than their PTENlhc+ counterparts (Parsons, 2004). Furthermore, upregulation of genes involved in metabolism, cell growth, and RNA processing is also in accordance to what would be expected in tumor cells with an unchecked PI3-K pathway, e.g., with activation of mTOR, p70 S6 kinase, and inactivation of 4EBP 1 (Hay, 2005).
Figure imgf000038_0001
Figure imgf000039_0001
Figure imgf000040_0001
[00108J GOMiner identified 46 biological process categories for genes overexpressed in the PTENlhc+ group, including 4 categoπes related to MAPK signaling such as 'MAPKKK cascade' (7 genes overexpressed, P=0.003) and 'JNK cascade' (3 genes, P=O 0224), differentiation-related categories 'morphogenesis' (20 genes, P=O 0285) and 'development' (26 genes, P=0.0413), and 4 groups relating to muscle contraction (Table 7). Overexpression of morphogenesis and developmental genes suggests that PTENlhc~ tumors are more poorly differentiated than PTENlhc+ tumors, and is in agreement with reports indicating a correlation between PTEN loss and higher tumor grade and stage (Bose et al., 2002; Lee et al., 2004; Bose et al., 2006). Upregulation of PKC/ERK/MAPK pathway members suggests that, in contrast PTENlhc~ tumors with activated PI3-K signaling, the MAPK pathway plays a more significant role in PTENlhc+ rumors. Closer inspection of members of the muscle related categories revealed genes implicated in both positive and negative regulation of myosins, reflecting a dynamic state of organelle transport and cell motility in these tumors.
Figure imgf000041_0001
Note: Gene Ontology (GO) analysis was performed using the top 785 signature genes. Only 'biological process' GO categories overrepresented as upregulated in PTEN ihc+ tumors and with P<0.05 are shown. Total' = total # of genes annotated to the given category; 'Under" = # of these genes underexpressed in PTEN ihc+ group; Over' = # of genes overexpressed in PTEN ihc+ group; 'Changed' = total # of genes that are under- or overexpressed. Muscle and MAPK pathway-related categories were manually curated.
[00109] Additionally, Gene Set Enrichment Analysis (GSEA) (Subramanian et al., 2005) was applied to further validate the biologic consistency between the present signal, the GO analysis, and the literature (see Methods). As shown in Figure ID, two independent sets of genes identified as downregulated upon PTEN induction in PTEN-defective glioblastoma (Stolarov et al., 2001) and endometrial cancer (Matsushima-Nishiu et al., 2001) cell line models were significantly enriched and upregulated in PTENlhc~ tumors (/><0.028, and P<0.037, respectively). Moreover, a breast cancer proliferation-associated gene set (Perou et al., 2000; Whitfield et al., 2002) was significantly enriched in PTENihc~ tumors (PO.014). Further demonstrating faithful capture of PTEN/PI3-K-related biology, the mTOR gene cassette (genes upregulated by Aktl transgene expression and downregulated by RADOOl, an inhibitor of mTOR), in a mouse model of prostate intraepithelial neoplasia was also significantly enriched in PTENlhc~ tumors (PO.030). Similarly, 3 of 4 gene sets regulated by nuclear FKHR (Ramaswamy et al., 2002), a forkhead transcription factor whose nuclear/cytoplasmic localization is regulated by PI3-K signaling, also trended in the expected direction (Figure 2D). None of the gene sets upregulated upon PTEN overexpression in cell lines was enriched, indicating that the induction of genes by PTEN in cell line models is inherently more noisy, artifactual, or may be cell type-specific. To further rule out that the PTEN status signal has a strong influence from an underlying ER status signal, an ER status consensus gene list was generated in the same 3-fold cross-validated manner as for PTEN. As previously shown (Gruvberger et al., 2001 ; West et al., 2001 ; van 't Veer et al., 2002), an extremely robust ER-associated signal was identified. The top ER status signature genes overexpressed in ER+ tumors with an APV<0.0001 (n=l 15) were selected for GSEA analysis. Reaffirming that the PTEN-associated signal is independent of ER, the gene cassette of top ranked ER signature reporters was not significantly enriched in PTENlhc+ tumors (P=0.203; Figure ID).
[00110] ATADJ, which is the nearest neighbor to PTEN at 10q23.31 in a head- to-head orientation, was significantly downregulated in PTENIHC- tumors (APV<0.004) and the absolute expression levels of ATADl closely tracked PTEN levels (Pearson r=0.676, PO.0001). Moreover, irrespective of PTEN IHC status, PTETVmRNA levels were below the median for 86% of Signature Present compi :d to 20% of Signature Absent tumors (/><0.0001). Thus, presence of the signature in 19/25 PTENiHC+ cases may be in part due to reduced PTEN message levels below the sensitivity of our IHC analysis (Fig. IQ.
[00111] The results presented in the Example show that the signature is highly consistent and relevant to PTEN/PI3-K pathway-regulated biological processes.
Influence of other PI3K pathway lesions
[00112] To test whether other PI3K pathway abnormalities are associated to the activated signature, studies were designed to analyze the tumors for pi 10a (PIK3CA) mutations and for amplification of HER2. Notably, 67% (8 of 12) of the PIK3CA kinase domain (KD) mutants clustered with the Signature Present tumors, whereas only 19% of non- KD mutants [3 of 16: 2/11 helical domain (HD), and 1/5 C2 domain (CD)] were in this main branch (P=O-Ol 9). Consistent with this, downstream pathway signaling differences between PlKiCA domain mutants were also seen in vitro (Figure 12). Moreover, suggestive of a dose- responsive inhibition of the PI3K pathway by PTEN (independent of PTEN IHC status), in the Signature Present cluster 82% (9/1 1) of PIK3CA mutants had PTENmKNA levels below the median compared to only 18% (3/17) of PIK3CA mutants in the Signature Absent cluster (/M)-OOOl). HER2 amplification (HER2+) did not appear to correlate with the presence of the signature, with 58% of positive cases (14/24) clustering among the PTENIHC- branch. However, among these 14 cases exhibiting the signature, 11 had PTEN message levels below the median, in contrast to 3/10 HER2+ Signature Absent cases (/»=0.035). In total, 75% (15/20) of lesions to PIK3CA or HER2 in the Signature Present group had low PTEN message levels, in contrast to 17% (4/24) in the Signature Absent group (/>=0.0002).
[00113] Together, these results show that the identified signature integrates various PI3K pathway lesions, some of which we have discerned but others not yet uncovered, that singly or, particularly in the setting of moderately reduced PTEN message levels, collaboratively activate PI3K signaling to a similar extent to that seen with IHC- detectable loss of PTEN. Conversely, cases with HER2 amplification or PIK3 CA CD/HD mutations rarely exhibit the activated signature in the context of high PTEN expression. The differential clustering of HER2+ cases by the signature could relate to trastuzumab sensitivity. These data demonstrate the important relationship between PTEN dosage and tumor phenotype in man and imply potential clinical utility of therapies that re-activate PTEN.
[00114] Despite controlling for ER status in tumor sample selection and verifying independence of our PTEN-signal from an ER-signal, 78% of ER+ tumors clustered in the Signature Absent group while 66% of ER- tumors clustered in the Signature Present group (Fig. 1 C). This suggests that frequent activation of the PI3K pathway is part of the natural history of ER-BC, and is consistent with our observations in an independent large population-based BC cohort in which most ER- tumors harbor ≥ PI3K pathway activating aberrations, in stark contrast to ER+ tumors.
Clinical Implications of the PTEN/PI3-K Pathway Signature
[00115] As activation of PI3-K signaling has been implicated in numerous processes that favor metastasis formation, studies were designed to investigate whether the signature could predict patient outcome. At a 2-year follow-up period, which corresponded to the length of adjuvant tamoxifen therapy for all 105 patients in the cohort, the patients identified as belonging to the Signature Present group had a significantly higher proportion of distant metastasis formation (Figure 1 E, log-rank /^=0.0248). Since the dataset was not designed for survival analysis, studies, were designed to determine whether the PTEN/PI3-K signature could be a useful prognosticator in other independent breast cancer datasets. The study by van de Vijver and colleagues (van de Vijver et al., 2002) contained primary tumors from 295 Dutch breast cancer patients analyzed on Rosetta oligo microarrays. The 246 top reporters mapped to 187 unique UniGene (build 188) clusters, of which 173 could be mapped to Rosetta probes. To classify samples, a nearest centroid classifier (NCC; see Methods) (van 't Veer et al., 2002) was used whereby each test sample was classified by their Pearson correlation to the Signature Present and Absent centroids trained on the present data. As illustrated by KM survival graphs in Figures 2A and 2B, the signature separated the 295 tumors into the two classes, Signature Present (141 cases, 48%) and Signature Absent (154 cases, 52%). Validating the prognostic utility of the PTEN/PI3-K pathway signature, tumors expressing the signature had a significantly worse DDFS (log-rank /><lxl0"15; hazard ratio, HR, 2.48, /><1.2xlO"5) and overall survival (OS; log-rank /><lxl0-'°; HR 3.69, P<2.3xl0 7; see also Table 1). The proportion of cases displaying the activated signature is the same as the rate seen in the dataset; moreover it is compatible with the expected combined frequency of PTEN loss, PIK3CA mutation, ERBB2 amplification, and EGFR overexpression in an independent population-based breast cancer cohort (-60%). 94% (108/115) of the tumors identified as good prognosis by the Dutch "70-gene prognosis classifier" were identified as belonging to the good prognosis Signature Absent category, and 74% (134/180) of the Dutch- classified poor prognosis tumors were identified as having the Signature Present poor outcome profile. In other words, activation of oncogenic PI3-K signaling may be the biological basis of poor outcome for as many as 74% of the poor prognosis patients identified by the Dutch signature, a signature that was built without any regard to the underlying biological processes which may be driving the metastasis formation. Conversely, the data presented here suggests that the less malignant tumors identified by the Dutch classifier have a good prognosis precisely because they do not exhibit significant activation of the PB-K pathway. Only 4 genes are in common between the Dutch 70-gene classifier and the signature presented in this Example: MELK, MCM6, RFC4, and LGP2.
Figure imgf000045_0001
[00116] As summarized in Table 1, multivariate Cox regression analysis adjusted for two commonly used conventional prognostic markers, ER status and lymph node metastasis status, revealed the PTEN/PI3-K gene expression classifier to be independent of and the most significant prognostic indicator among these markers for both DDFS and OS. Together, these results demonstrate the PTEN/PI3-K pathway signature to be a highly robust prognostic marker and indeed independent of an ER-related signal.
[00117] Since the NCC analysis generates a correlation score for each classified tumor to the Signature Present and Signature Absent centroids of the training dataset, studies were designed to test whether a 'signature score' (see Methods) was predictive of outcome on a cutoff-independent continuous scale in the large Dutch dataset (see Methods). The signature score ranged from -1.44 to +1.45 (positive values indicating greater presence of the signature) and was highly predictive of DDFS (HR 1.80, PM.lOxlO"6; that is, for every 1 point increase the hazard increases by 80%) and OS (HR 2.33, P=\.9$x\0~ 8; Table 1). Since GO and GSEA analysis indicated a significant cell proliferation signal associated to PTENlhc~ tumors, all genes annotated to any of the 304 cell cycle-related categories in the GO hierarchy were removed from the 187 unique UniGene clusters in the signature (29 unique clusters removed) and repeated the NCC procedure. Only 6% (17/295) of tumors changed classification, and all 17 had low original NCC correlation scores (r<0.2 to either Signature Present or Absent centroids) when using all 173 mapped classifier genes. Therefore, these results show that the successful prognostic classification of tumors is not only due to information from proliferation-associated gene expression but also is due to a complex program of transcriptional changes downstream of the PTEN/PI3-K pathway.
[00118] Similar results were obtained when classifying the dataset of 99 British breast tumors published by Sotiriou et al. (Sotiriou et al., 2003). Ninety-two of the 187 unique signature UniGene clusters could be mapped to the Sotiriou microarrays for NCC analysis. Tumors expressing the activated signature (48 tumors, 48%) had a significantly worse relapse-free survival (RFS; log-rank P=0.0386; HR 1.89, P=0.042) and nearly significantly worse breast cancer-specific survival (BCS; log-rank P=O.0648; HR 1.87, P=0.069) over the complete follow-up time, and at the clinically important 5-year follow-up analysis interval, a significant difference in BCS was evident (log-rank P=0.0194; HR 2.70, P=0.025) (Figures 2C and 2D, and Table 1). Moreover, the signature score (-0.96 to +0.95) continuous variable contained significant prognostic information for both RFS (HR 2.13, P=0.0\ 8) and BCS (HR 2.33, /M).O2O; Table 1 ). These results in two independent breast cancer datasets show that the degree of PTEN/PI3-K pathway activation, as measured by the correlation of each tumor to the Signature Present and Absent centroids, is directly related to the metastatic potential of the tumor. Predicting Outcome in Other Cancer Types
[00119] Given the success of the PTEN/PI3-K pathway signature to predict outcome in breast cancer, this signature may be more generally applicable to other cancer types with known high rates of PI3-K pathway activation. To test this, public microarray datasets were obtained on large series of human prostate (Glinsky et al., 2004), bladder (Blaveri et al., 2005), and lung carcinoma (Beer et al., 2002), as well as diffuse large B-cell lymphoma (DLBCL) (Rosenwald et al., 2002), a cancer type in which PI3-K-associated signaling has not been well characterized in the literature. As shown in Figure 3A-3D, the NCC analysis using the matching PTEN/PI3-K pathway signature genes could significantly separate prostate (152 matching signature genes) and bladder (115 genes), but not lung carcinoma (79 genes; trended towards significance with 5-year followup log-rank /*=0.0602, and at complete followup /M).1051) samples or DLBCL (55 genes), into groups with appreciable differences in survival (Cox regression results are presented in Table 1). Moreover, the NCC correlation score was also predictive on a continuous scale for the same tumor types (Table 1). The tumor types in which the PTEN/PI3-K signature was not a prognostic factor were also those with the lowest number of matchable genes.
[00120] When using a binary classification by NCC analysis, bladder cancers and DLBCLs had a considerable number of cases (60% and 54%, respectively) with intermediate correlation scores (r<0.2 to both Signature Present and Absent centroids). Therefore a simple threshold was applied to generate 3 classes: those that are have a more robust Signature Present profile (r>0.2), those that express a robust Signature Absent profile (r>0.2), and an intermediate group (correlation to both centroids <0.2). This yielded significant difference in survival for bladder cancer between the two robustly defined groups (log-rank P=0.0098; HR 3.89, P=0.016; Figure 3B and Table 1) but not DLBCL (Figure 3D and Table 1). The intermediate group of bladder tumors also displayed rather poor outcomes, indicating the presence of perhaps a non-PTEN/PI3-K pathway-associated class of bladder carcinoma with activation of one or more alternative potent oncogenic pathways (Figure 3B). A proliferation signature has been shown to be predictive of outcome in DLBCL (Rosenwald et al., 2002); thus the inability to predict a significant difference in survival for these tumors, despite having a proliferation component in the present signature, suggests that this pathway is not a major oncogenic factor in DLBCL. Analogous 3-group classifications for all independent cancer datasets are presented in Figure 6.
Selection of a Novel Marker for PTEN/PI3-K Pathway Activation [00121] Current markers for PD-K pathway activation are inadequate. The most often used marker is phosphorylated AKT (threonine 308 or serine 473), however the available reagents only work satisfactorily in samples handled under controlled conditions; the epitopes are highly labile in routine specimens from the clinic. Anti-pAKT reagents do not yield reliable data for routine clinical samples. Only specimens that are immediately fixed or frozen upon excision and stored with care, conditions difficult to reproduce in the everyday hospital setting, can be analyzed for pAKT levels. Notwithstanding similar problems with reliable reagents and assays, other PI3 -K pathway markers such as phospho- S6 kinase, phospho-GSK-3β, and phospho-FKHR are several steps downstream in the pathway, and as considerable cross-talk exists to and from other pathways impinging on these effectors, they are imperfect surrogates. Therefore, studies were designed to identify at the protein level PTEN/PI3-K pathway signature genes that would reflect pathway activation. To this end, 5 signature genes, pi 10a (encoded by PIK3CA), discussed above, stathmin (STMNl), involved in microtubule (MT) regulation, the cell cycle, and cell motility (Belmont and Mitchison, 1996; Rubin and Atweh, 2004; Baldassarre et al., 2005; Borghese et al., 2006), mini-chromosome maintenance 6 (MCM6), part of the MCM complex controlling DNA replication (Lei, 2005), NIMA-related kinase 2 (NEK2), a centrosomal kinase found overexpressed in various cancers (Hayward and Fry, 2005), and myeloid leukemia factor 1 (MLFl), involved in the t(3;5)(q25.1 ;q34) NPM-MLFl acute myeloid leukemia translocation (Yoneda-Kato et al., 1996), were selected for immunoblotting experiments.
[00122] These proteins were analyzed across a panel of 4 breast cancer lines with wild-type (wt) PTEN and 4 breast cancer lines with documented PTEN mutations (Figure 4A). Stathmin was very highly overexpressed in the PTEN mutant lines, whereas pi 10α had a modestly higher expression in the PTEN mutant lines (Figure 4A). Of the 3 PTEN wt lines with detectable stathmin protein, two harbored PIK3CA mutations: MDA- MB-453 (H 1047R) and MDA-MB-361 (E545K). UACC-893 is mutated at residue H1047R, however it does not express stathmin protein. MCM6 and MLFl did not show an appreciable association to PTEN mutational status in these lines (Figure 4A). Although pi lOoc displayed a modest increase in protein level in PTEN mutants, IHC analysis for pi 10α in tumor biopsies did not appear to be correlated to PTEN status (P=Q A3, Mann- Whitney test; see also Figure IA). This may be due to decreased sensitivity of the antibody for IHC applications, differences between the context of PTEN mutation in cell lines vs. loss of PTEN protein in tumors by other mechanisms, or may indicate that in the context of in vivo tumor growth, other regulatory processes keep pi 1 Oa levels in check despite relatively higher PIK3CA message levels in PTENlhc~ tumors:
[00123] Other PTEN signature genes are also provided by the invention. That loss of PTEN results in cytoplasmic relocalization of the checkpoint protein kinase CHEKl and associated genomic instability (Puc et al., 2005). In the present study, microarray analysis identified CHEKl to be significantly overexpressed in PTENihc" rumors (APVO.029). This result shows the presence of a nuclear CHEKl sensor, perhaps by feedback inhibition of itself, whereby CHEKl message is upregulated upon reduction of nuclear CHEKl . RAD51C, a recombinational repair gene, was also overexpressed in PTENlhc~ tumors (APV<0.017), suggesting a physiologic upregulation of this gene in response to the increased genetic instability of PTEN low tumors. Given that the regulation of PTEN protein in breast cancer was found to be at the transcriptional level, the fact that several histone-related genes {HIST1H2BA, HDAC2, HlFO, HlSTl H2B J, H3F3A, H2AFZ, HISTl H2BK, H2AFV, and HISTlHlC) are among the top 1000 signature genes and are predominantly overexpressed in the PTENlhc~ tumors, could be indicative of a chromatin-mediated silencing process.
Stathmin is a Marker of PI3-K Pathway Activity
[00124] Given the strong inverse correlation between stathmin and PTEN in breast cancer cell lines (Figure 4A), and the fact that STMNl message was also found to be strongly downregulated upon PTEN induction or LY294002 treatment in glioblastoma cells (Stolarov et al., 2001), stathmin was considered to be a good candidate marker for PI3-K pathway activation. To evaluate stathmin as a marker, in a blinded fashion stathmin protein expression was analyzed by IHC in a set of 191 breast tumors, of which 181 were aiso evaluated for PTEN protein levels (Table 8). Stathmin IHC provided a robustly detectable signal that was amenable to semi-quantitative scoring (Figure 4B). Validating of the microarray result, stathmin staining scores were found to be significantly higher in PTENlhc~ tumors than in PTENihc+ tumors (P=0.0052, Mann- Whitney test; Figure 4C). Since PTEN status and stathmin expression have been independently shown to be associated to ER negativity (Brattsand, 2000; Curmi et al., 2000), the association within the ER" tumors was assessed and it remained significant (F=0.0416; Figure 4D). This shows that stathmin expression is inversely correlated to PTEN expression. Most of the 105 tumors profiled on microarrays were among those evaluated for stathmin protein, so next the stathmin staining scores were mapped to the tumor gene expression data (Figure 4E). Stathmin protein levels closely track stathmin message levels (Pearson r=0.6477, PO.0001 ; Figure 4F), and reflects well the presence of the PTEN/PI3-K pathway signature (better than merely reflecting the loss of PTEN message, r=-0.3540, P=0.0006; Figure 4G). Moreover, very few tumors clustering in the Signature Absent group had marked upregulation of stathmin (Figure 4E). Thus, these results indicate stathmin to be a robust, durable, and specific marker of pernicious PI3-K pathway activation.
Table 8. Clinical Demographics for Stathmin IHC Breast Tumor Set
- All cases With PTEN IHC data P-value
n 191 181
Estrogen receptor positive 126 (66%) 119 (66%) 0.96
Progesterone receptor positive 96 (50%) 92 (51%) 0.91
Lymph node metastasis positive 118 (62%) 113 (62%) 0.90
Pre-menopausal 36 (19%) 35 (19%) 0.90
Median age (range) 64 (26-77) 64 (26-77) 0.88
Median tumor size, mm (range) 25 (2-55) 25 (2-55) 0.90
Median followup, years (range) 6.7 (0.3-13.2) 6.8 (0.3-13.2) 0.98
Note: P-values are calculated using Pearson's Chi-squared test for categorical variables and the student's t-test for continuous variables.
Clinical Implications of Stathmin O verexpression
[00125] Although some reports have indicated that PTEN protein status can carry prognostic information in breast cancer (Depowski et al., 2001; Lee et al., 2004; Shoman et al., 2005), others have not (Panigrahi et al., 2004). In this Example, PTEN status was not a significant marker for DDFS within the set of 181 patients nor in the expanded set of ~350 patients, owing to the relative homogeneity of this cohort (only stage II tumors) and other PI3-K pathway-activating lesions such as the recently uncovered high frequency of PIK3CA mutations in breast cancer (Samuels and Ericson, 2006). Stathmin has been implicated in processes of cellular proliferation (Rubin and Atweh, 2004), MT regulation (Belmont and Mitchison, 1996), and cell motility and migration (Baldassarre et al., 2005; Borghese et al., 2006), has been correlated to high proliferation in breast tumors (Brattsand, 2000) and poor outcome in medulloblastoma (Neben et al., 2004), and the message levels have been shown to be downregulated by inhibition of PI3-K signaling (Stolarov et al., 2001). These are all findings biologically consistent with stathmin being downstream of PI3- K signaling. If, as shown here, stathmin is a surrogate for presence of the PTEN/PI3-K signature, i.e. a marker of oncogenic PI3-K nathway activation, then one would expect it to be an excellent marker for poor prognosis in breast cancer. Therefore, studies were designed to investigate whether stathmin protein levels (dichotomized into low and high groups) were related to outcome in the 191 patient cohort. As shown in Figure 4H, at the 2-year follow-up cutoff corresponding to the end of adjuvant tamoxifen therapy, there was a significant difference in survival for the stathmin overexpressing group (log-rank P=0.0006), as well as over a 5-year followup (P=COl 32). At complete followup, there were very few patients in the stathmin high group due to loss to follow-up; nevertheless the difference in DDFS was nearly significant (P=0.0765). Cox regression analysis indicated that high stathmin had a significantly higher risk for distant metastasis, with a HR >4 for the 2-year interval (P=0.001), HR 2.4 for the 5-year interval CP=0.016), and HR 1.8 at complete follow-up (P=0.081; Table 1). As the choice of a cut-off to yield a binary stathmin IHC variable could result in bias, similar to the PTEN/PI3-K NCC correlation score, the stathmin staining score as a continuous variable (from 0 to 12) was also highly predictive of DDFS at 2 years (HR 1.16, P=0.004; that is, for every 1 point increase in stathmin score, the hazard increases by 16%) and 5 years of follow-up (HR 1.08, P=0.036; Table 1).
[00126] Lymph node involvement at diagnosis is the most important conventional prognostic factor in breast cancer (Schnitt, 2001), however a significant fraction of patients with negative lymph node involvement develop distant metastases, highlighting the critical need for new markers to identify such patients at diagnosis for their proper clinical management. Multivariate Cox regression analyses were performed to investigate the prognostic power of stathmin after adjustment for ER and node status. Stathmin expression was independent of ER status and lymph node status and was the most significant factor at 2- years follow-up (HR 3.54, P=O Ol 9), was independent of ER and node at 5-years (HR 2.45, /•=0.035), and was nearly significantly independent at complete follow-up (HR 2.1 1 , P=0.071 ; Table 1). Among patients with lymph node negative disease, high stathmin was extremely predictive for distant metastasis within 2 years of initial diagnosis, with a HR >9 (95% CI 1.76-46.84, P=0.008). Among patients with lymph node positive disease, high stathmin was also predictive for distant metastasis within 2 years (HR 3.99, 95% CI 1.13- 14.14, P=0.032).
[00127] These results were validated in the independent Dutch BC dataset of
295 tumors. STMNl mRNA levels were higher in the Signature Present-classified poor- prognosis class (PO.0001) and directly related to the signature score (r=0.690, PO.0001; Table 16). STMNl mRNA itself was prognostic, with STMNl -higfr (top 50th percentile) cases having a significantly worse DDFS and OS at 5.ryears and complete follow-up (P=0.0002 and /*=0.0003, respectively, for both time intervals; Figures 13A - 13D). Moreover, supporting our hypothesis that the signature and the signature marker stathmin are pathway-integrative, we note that classification by PTEN message levels by itself was not as prognostic (DDFS: 5- year, /^0.(B 8 and complete follow-up, P=Q.111 ; OS: 5-year, P=0.005 and complete follow- up, .P=0.012; Figures 13A — 13D), whereas the best prognostic separation was achieved with the full signature (Figures 2A-2B and Table 1).
Figure imgf000052_0001
Stathmin is PTEN/PI3K pathway-regulated in vitro and m vivo. [00128] A marker of P.TEN/PI3K pathway activity could have clinical utility, particularly one that demonstrates pharmacodynamic properties. Therefore, experiments were designed to determine whether stathmin protein expression was regulated by the pathway and whether it was an output of pharmacological inhibition of the pathway. To test this, PTEN was expressed in PTENnuXl MDA-MB-468 cells and monitored stathmin levels over time. Compatible with its reported half-life of ~28 hours, stathmin levels were 23-45% downregulated from 24-52 hours post-infection (Figure 14A). Moreover, despite variable downregulation of p-Akt (S473) by PI3K pathway-inhibition using LY294002, stathmin levels were reduced robustly in the 3 .P7E7V-mutant BC cell lines (Figure 145), MDAlO MB- 468 (56% diminished), HCC- 1937 (29%), and BT-549 (32%), and moderately in MDA-MB- 436 cells (15%), which are PTEN wild-type but have undetectable PTEN and low levels of stathmin at steadystate. These results were further validated in vivo utilizing two mouse xenograft models: stathmin levels were appreciably downregulated upon doxycyclin-induced PTEN expression in MDA-468TR-PTEN U~ iors (Figure 14C), and were similarly downregulated in MDA-MB-468 tumors in mice administered the wortmannin-analog PI3K- inhibitor PWT-458 (Figure \4D). Thus, evaluation of stathmin may be an effective way to quantitatively measure POK pathway activity, stratify patients, and monitor treatment response. Moreover, stathmin's role in regulating microtubule dynamics, promoting cell motility and proliferation, and conferring resistance to antimicrotubule drugs are consistent with it being downstream of the PI3K pathway, and may indicate that it is a potential therapeutic target in addition to being a prognostic and pharmacodynamic marker.
Discussion
[00129] The invention provides a robust PTEN/PI3-K pathway gene expression signature in sporadic human breast cancer that faithfully captures the known biologic consequences of activated PI3-K pathway signal transduction. This Example shows that PTEN protein levels in breast cancer are primarily determined by the message level, and the mechanism at play appears to have a simultaneous coordinated effect on PTEN1S nearest neighbor A TADl. Several genes in the signature also encode transcription factors, which may be important downstream effects that contribute to the phenotype of P13-K pathway activation, or, conversely, may be partially responsible for activating the pathway itself {e.g., by modulation of PTEN expression). Thus, the PTEN/PI3-K pathway signature will be a valuable resource for unraveling PI3-K regulated biological processes and can be utilized to identify potential molecules regulated by or cooperating with the PTEN/PI3-K pathway in breast tumorigenesis.
[00130] This Example demonstrates that the PTEN/PI3-K signature successfully segregates two independent breast cancer datasets into pathway-activated and non-activated groups with significant differences in outcome independently of other common prognostic factors. Portions of the 'proliferation cluster' identified by a different study (Perou et al., 2000) appears to be a component of the signature of the invention (see GSEA results) and has been previously shown to be primarily expressed in the 'basal-like', 'luminal subtype C, and to some extent in the 'ERBB2' molecular subgroup of breast cancer, the same three subgroups shown to have the worst outcomes (Sorlie et al., 2001). The results show that these three subtypes have high rates of PI3-K pathway activation. The results show that about half of ERBB2-overexpressing tumors express the signature, and the majority of basal cytokeratin-positive breast cancers have one or more activating hits to the PI3-K pathway.
[00131] The Example also shows the signature to be generally applicable for outcome prediction in other cancer types with known PI3-K pathway involvement. The degree of correlation that a tumor expresses the PTEN/PI3-K signature appears to be directly related to the malignant potential of the tumor. This shows a continuum of PB -K pathway contribution in several cancer types, where tumors with the highest correlation to the activated pathway profile would be most the most "addicted" (Weinstein, 2002) to the pathway and thus the most sensitive to therapies that specifically attack this pathway. The data is also consistent with a recent study showing that the PI3-K pathway is the critical pathway for tumor maintenance in a Ras-mediated model of tumorigenesis (Lim and Counter, 2005).
[00132] As currently available reagents for PI3-K pathway activation are inadequate, a novel marker was identified for the pathway that is more amenable to analysis in routine clinical specimens. The invention provides stathmin to be such a marker and this Example demonstrates stathmin to carry significant prognostic information for breast cancer outcome, particularly among patients with lymph node negative cancer at diagnosis. Evaluation of stathmin, and potentially other signature genes, in clinical specimens in standardized assays may be an effective way to measure P13-K. pathway activity in tumors and requires further study. For example, applying the signature or signature components such as stathmin retrospectively or prospectively to patient cohorts with identified molecular lesions and treated with targeted therapies may provide a meaningful stratification according to therapeutic response, such as in patients with ERBB 2 amplification treated with trastuzumab.
[00133] The signature genes can also serve as potential therapeutic targets. Inhibition of stathmin, which regulates MT dynamics, could have inhibitory effects on mitosis and/or cell migration. Additionally, studies have shown overexpression of stathmin to affect sensitivity to MT-stabilizing drugs and MT-destabilizing drugs (Orr et al., 2003). Therefore, studies can be designed to determine whether MT-targeting therapies, in combination with PI3 -K- targeted therapies, would be synergistic against rapidly proliferating P13-K pathway- activated tumors. Some of the signature genes encode cell surface proteins, which may be useful as molecular beacons of pathway activation that could be imaged non-invasively using labeled antibodies to monitor disease progression and response to targeted therapies. The future of cancer management is quickly moving towards pathway-based profiling and directed therapy. Moreover, the gains in pathway-specific treatment of cancer are likely to translate to other diseases, as for example the PI3-K pathway is involved in a vast array of human ailments.
Materials and Methods [00134] Tissue Samples and Cell Lines. This study was approved by the Lund
University ethics committee. Formalin-fixed paraffin-embedded tumor tissues were retrieved for 343 stage II primary breast cancers assembled by the South Sweden Breast Cancer Group and collected at the Department of Oncology, Lund University, and these tumors were analyzed for PTEN protein by IHC as previously described (Saal et al., 2005). From these, a subset of 105 tumors (35 PTENihc~ and 70 PTENihc+) for analysis with cDNA microarrays were selected roughly matched with respect to ER and lymph node metastasis status where possible (Table 4). Additionally, 191 rumors were also analyzed by IHC for stathmin protein levels, of which PTEN IHC data could be generated for 181 (Table 8). The cell lines BT-549, HCC-1395, HCC-1937, MDA-MB-361, MDA-MB-453, MDA-MB-468, UACC-812, and UACC-893 were obtained from and cultured according to recommendations of the ATCC.
[00135] Microarrays. cDNA microrrays with 27,648 spots were fabricated by the
SWEGENE Microarray Facility, Department of Oncology, Lund University. The printed cDNAs include 24,301 sequence-verified IMAGE clones (Research Genetics, Huntsville, AL), and 1,296 internally-generated clones, together mapping to >15,000 UniGene clusters (build 188). The clones were prepared essentially as described (Khan et al., 1999) with some modifications. Briefly, in 96-well format the clones were PCR amplified using vector- specific primers, visualized by agarose gel electrophoresis, purified by size-exclusion filtration using Montage PCR96 plates (Millipore, Billerica, MA), recovered in water, adjusted to 50% DMSO, and together with a selection of control probes, printed on amino- silane-coated UltraGAPS II slides (Corning Inc., Coming, NY, USA) using a BioRobotics MicroGrid II robot with 48 MicroSpot2500 pins (Genomic Solutions, Cambridgeshire, UK) in a 4 by 12 pin configuration (48 subarrays). The printed slides were then baked and UV cross-linked. All microarray production data were stored in BioArray Software Environment (BASE) (Saal et al., 2002).
[00136] RNA Preparation and Microarray Hybridization. Total RNA was extracted from ~100 mg grossly dissected frozen tumor tissue by pulverization in a microdismembrator chilled on dry ice, immediately followed by homogenization in TRlzol reagent according to manufacturer's instructions (Invitrogen, Carlsbad, CA). RNA was purified a second round using the RNeasy kit (Qiagen, Hilden, Germany), and the final yield and purity was assessed using a Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). For each hybridization, 25 μg of tumor RNA was used to synthesize Cy3-labeled targets, and as the common reference across all experiments, 15 μg of Universal Human Reference RNA (Stratagene, La Jolla, CA) was used to synthesize Cy5-labeled targets, using anchored oligo-dT primers and the CyScribe indirect amino-allyl cDNA synthesis and labeling protocol and GFX purification columns (Amersham Biosciences, Buckinghamshire, UK). The tumor and reference targets were pooled and blocking agents, 12 μg poly-dA, 6 μg yeast tRNA, and 20 μg Cot-1 DNA, were added, and the target mixture cohybridized to the microarrays. Hybridization was carried out for 18 hours under a glass coverslip using humidified Corning hybridization chambers at 42°C and utilizing the Pronto Universal Hybridization System (Corning Inc.). Slides were scanned at 10 μm resolution in an Agilent DNA Microarray Scanner (Agilent Technologies) and the images analyzed using GenePix Pro software (Axon Instruments, Union City, CA). Details of each sample, sample processing and hybridization, and raw images and quantified data matrices were loaded into BASE (Saal et al., 2002).
[00137] Microarray Data Analysis. Data pre-processing and normalization were performed within BASE (Saal et al., 2002). For the tumor and reference channels (Cy3 and Cy5, respectively) the mean foreground spot intensity minus the median local background intensity was used to calculate the background-corrected channel intensities, channel 1 and channel 2. The 105 tumor gene expression measurements of 27,648 spots were filtered to remove features with hybridization artifacts flagged during image analysis and low intensity spots with a mean log intensity A<1.5 (A defined as the log^ channel 1 *channel 2]). Tumor over reference ratio of the remaining values were 1Og2 transformed (Iog2[channel 1 /channel 2]). Data within each array was normalized using a BASE plug-in implementation of the LOWESS algorithm, whereby the 48 subarrays were grouped into 6 groups spatially lengthwise along the slide (8 pins per group, i.e. 2 rows of 4) and the data within each pin- group smoothed independently. The data matrix was then exported and further filtered to remove all rows (features) with missing data in >20% of assays.
[00138] The Mann-Whitney U-test was used to assess the correlation of each gene's expression pattern to a binary sample label {e.g. PTEN positive or negative; ER positive or negative), with a P- value computed for each gene and a sign, +1 or -1, assigned if the gene is correlated or anti-correlated. Similarly to the method described previously (Pavey et al., 2004), a 3-fold cross-validation design was used to train a committee of SVMs to predict the PTEN status of the tumors and generate a consensus list of ranked PTEN signature genes. The procedure was carried out as follows: (1) partition the tumors randomly into 3 equal sized groups with roughly an equal distribution of the sample labels of interest in each group; (2) select 2 of the groups for training, and the remaining group is left as the prediction test set; within the training set, rank the genes according to the correlation of their expression measurements to PTEN status using the Mann-Whitney U-test and use this ranked gene expression matrix as input to train a SVM; (3) data from the test group are fed to the SVM and each test tumor is classified as PTEN positive or negative; (4) steps 2 and 3 are repeated twice, such that each time a different group is put aside as the test group; (6) steps 1 to 4 is repeated 9 more times, in which each time a new random partition into thirds is made. In such a way, 30 SVMs in all were trained and each tumor received a prediction output from 10 SVMs (Figure Sl). The area under the receiver operating characteristic (ROC) curve was used as the measure of prediction performance. A consensus-ranked gene list was generated by sorting on the average P- value (APV) of each gene from the 30 ranked lists. The sum of signs is a measure of the consistency of the correlation and is defined by taking the sum of the correlation/anti -correlation signs (e.g., +30 is correlated in all lists and -30 is anti- correlated in all lists). Permutation tests were used to estimate both the significance of the SVM prediction performance and the consensus-ranked gene list. In these tests, sample labels were randomly permuted. SVMs were built for 10,000 random classification problems and a P-value corresponding to the probability to obtain better performance for random sample labeling was assigned to the original ROC area. Consensus-ranked gene lists were built for 100 random sample labelings and a false discovery rate for an APV was estimated as the average number of genes with smaller or equal APV from the random sample labelings.
[00139] GOMiner was used for identifying overrepresented GO 'biologic process' categories, following the authors' recommended standard procedures (Zeeberg et al., 2003). Comparison to published microarray datasets (Matsushima-Nishiu et al., 2001 ; Stolarov et al., 2001 ; Whitfield et al., 2002; Ramaswamy et al., 2002; Majumder et al., 2004) related to known outputs of PI3-K signaling were performed by first updating all obtained gene lists to UniGene build 188 using the relevant gene identifiers and the ACID database (Ringner et al., 2004) and then matching on unique gene symbols: genes that mapped to zero or to multiple UniGene clusters were discarded. The full dataset from Stolarov et al. (Stolarov et al., 2001) was reanalyzed with less stringent fold-change cut-offs than originally used in their article to yield more genes for comparison to the present data (genes that were regulated by PTEN and LY294002 >1.8-fold and simultaneously <1.5-fold by the PTEN G129R mutant). The mTOR gene cassette was generated by taking all probesets with a correlation >0.7 to the profile of genes upregulated by Aktl and downregulated by RADOOl (Majumder et al., 2004). Gene set enrichment analysis (GSEA) (Subramanian et al., 2005) was performed by first collapsing the gene expression data on unique gene symbols utilizing the default settings in the GSEA program. The analysis was run using the assembled gene sets with default parameters and 1000 random phenotype permutations. A P≤).O5 was considered significant for both GO and GSEA analyses. Hierarchical clustering was performed using Cluster 3.0 software (de Hoon et al., 2004), wherein Iog2(ratios) were polished by median centering genes and arrays and clustered using 1 -Pearson correlation and centroid linkage, and the result visualized with Java Treeview (Saldanha, 2004).
[00140] Prediction of Signature Activation in Independent Datasets. Before testing independent datasets and to avoid 'information leak', one can ensure that a sizable number of top signature genes is selected that predicts well in the dataset and would also be likely to yield a good overlap when mapped to other microarray platforms. Studies were designed to determine whether signature gene sets of different sizes would give tumor clusters with better survival separation in the dataset. PTEN/PI3-K signature gene sets consisting of the top 98 (APVO.01), 500, and 795 (APVO.05) genes were evaluated; all resulting hierarchical dendrograms split the tumor set into two groups with very similar sample distribution to the result when using 246 genes. As the 246 gene-based clustering had the lowest P-value for distant disease-free survival, studies were continued with this set as the signature gene set for PTEN/PI3-K pathway status and outcome prediction for all independent datasets. Samples were classified using a nearest centroid classifier (NCC) (van 't Veer et al., 2002). The NCC is trained on the data set as follows. First, each gene is centralized to mean Iog2(ratio) of zero across the samples. Second, a centroid is calculated for each class of samples {e.g. Signature Present and Signature Absent classes defined by hierarchical clustering using the top 246 signature genes) by, for each gene separately, taking the arithmetic mean of the expression levels across the samples in the class. A centroid serves as a prototypical expression pattern across the genes in the profile to which test samples can be compared. Each independent test dataset was centralized in a similar fashion as the dataset. For each test sample the Pearson correlations between its expression levels and the two class centroids are calculated. A test sample is classified based on to which centroid it is most highly correlated. To evaluate the classification as a continuous correlation score, a 'signature score' was calculated by subtracting the Signature Absent correlation from the Signature Present correlation.
[00141] Immunohistochemistry and Mutational Analysis. IHC staining and scoring for PTEN and ERBB2, and PlKiCA mutational screening, has been described previously (Saal et al., 2005). All IHC evaluation was performed blinded to clinical information. Anti- stathmin IHC on 191 breast cancers was performed similarly as for PTEN, with the exception that the primary antibody (Cell Signaling, Danvers, MA) was used at 1 :250 dilution. Stathmin staining in invasive tumor cells and normal epithelial cells was scored as a composite of percent positive cells (6 bins from 0 to 5: zero cells, 1-10%, 10-25%, 25-50%, 50-75%, and 75-100% positive cells) and staining intensity (6 bins from 0 to 5, with 0 being no and 5 being intense staining). The percent score was weighted by doubling the value (yielding scores from 0 to 10), and then the weighted percent score and intensity score were summed, giving a possible score from 0 to 15 for the tumor cell and normal cell compartments. Fifty- five of 191 cases had evaluable normal compartments. To generate a relative tumor score, the median stathmin score in the 55 normal compartments, 3, was subtracted from each tumor score for all 191 cases, to yield a relative tumor score between 0 and 12 (Figure 5B). Immunostaining with the anti-pl 1 Oa antibody, which detects a single band by Western blot and is specific for the alpha subunit (Cell Signaling), was performed similarly as for PTEN, with the following exceptions: microwave antigen retrieval in 1 mM EDTA; primary antibody dilution 1 :50 for 2 hours; signal detection using the EnVision+ system (DAKO). Anti-pl 1 Oa was membranous/cytoplasmic for almost all cases, with some cases with significant membranous enrichment, and was scored in the normal cell and invasive tumor cell compartments for intensity of stain on a 0 to 8 scale; each case was reduced to a binary variable with a difference in tumor to normal >2 considered pi 10α-elevated and all others pi 10α-normal. For projection onto the hierarchical clustering dendrograms, the stathmin IHC score for the 90 matched cases was centered about the median score, 6 (pseudocolored black), with scores >6 in red and those <6 in blue (Figure 5C), and the pi 10a binary categorization was used (Figures 1 A and 5E).
[00142] Western Blotting. Cells were lysed in 2x Laemmli sample buffer containing
10% 2-mercaptoethanol and 8 M urea. The proteins were separated by No vex 4-20% Tris- glycine or 4-12% Bis-Tris gel electrophoresis and transferred to PVDF membranes following standard manufacturer protocols (Invitrogen, Carlsbad, CA). Primary antibodies directed against the following proteins were used for immunoblotting: PTEN (6H2.1, Cascade Bioscience, Winchester, MA), β-tubulin (clone Tu27, Covance, Berkeley, CA), stathmin, pi 10a (both Cell Signaling), MLFl (GenWay, San Diego, CA), NEK2 (clone 20, BD Transduction Laboratories, San Diego, CA), and MCM6 (clone 753). [001431 Statistical Analysis. The Mann- Whitney U-test was used to assess associations between continuous variables in two sample groups, and the Pearson χ2 test was used for 2x2 tables unless any cells had values ^, in which case Fisher's exact test was used. For survival analysis, NCC output by the signature was used as a dichotomous variable and continuous variable. The prognostic value of stathmin IHC was evaluated as a continuous variable as well as dichotomized into stathmin low (scores 0-10) and stathmin high (scores 11 and 12) groups. The duration of follow-up was computed from the date of primary operation to the date of an event (distant recurrence or death) or for event-free patients to the date of the most recent follow-up. The Kaplan-Meier method was used to estimate relevant event variables and the log-rank test was used to compare survival between two strata. Cox proportional hazards model analysis was used for univariate and multivariate evaluation of prognostic information content. Proportional hazards assumptions were checked and violations were handled using stratification of the time axis. Statistical analyses were carried out using Stata 9.0 (Stata Corporation, College Station, TX), with all tests two-sided and P- values <0.05 considered significant.
EXAMPLE 2: A PTEN GENE EXPRESSION SIGNATURE IN HUMAN BREAST
CARCINOMA
[00144] PTEN is a tumor suppressor and has been shown to be involved in numerous pathways regulating cell growth, migration, apoptosis, and the cell cycle, of which the phosphatidylinositol 3-kinase (PI3-K) pathway has been most studied. The PI3-kinase phosphorylates PIP2 to generate PIP3, a potent second messenger, in response to growth signals, and PTEN functions as a PIP3 phosphatase to negatively regulate the pathway. In breast cancer, PTEN is infrequently mutated, however protein levels are significantly reduced in -30% of cases. Little is known of the in vivo downstream gene expression changes due to alterations of PTEN expression. To better understand the signaling outputs of the PTEN/PI3- K pathway, 27K cDNA microarrays and support vector machines were used to identify a gene expression signature discriminating tumors with PTEN loss (PTEN -) from those with retained PTEN expression (PTEN+) in a set of 105 primary breast tumors characterized for PTEN status by immunohistochemistry.
Materials and Methods
[00145] Patients: 343 stage II primary breast cancers were analyzed for PTEN protein by immunohistochemistry (Saal, et al., 2005). From these, a subset of 105 tumors (35 PTEN" and 70 PTEN+, matched with respect to estrogen receptor, and lymph node status, where possible) was selected for analysis with cDNA microarrays.
[00146] Microarrays: cDNA microrrays with 27,648 spots were fabricated by the
SWEGENE Microarray Facility, Department of Oncology, Lund University. Universal Human Reference RNA (Stratagene) was used as a common reference for all hybridizations. Hybridization, image acquisition and image analysis were performed essentially as previously described (Andersson, et al., 2005).
[00147] Microarray Data Analysis: Data pre-processing and normalization were carried out in BASE (Saal, et al., 2002). A three-fold cross-validation design (Figure 5) employing the Mann- Whitney U-test was used to generate 30 gene lists ranked by the correlation of each gene's expression profile to PTEN status. The 30 ranked lists were used to train support vector machines (SVMs), whereby each tumor was predicted by 10 SVMs. A consensus-ranked PTEN signature gene list was created by ranking each gene by its average P-value across all 30 Mann-Whitney lists. GOMiner (Zeeberg, et al., 2003) and Gene Set Enrichment Analysis (Subramanian et al., 2005) software were used for identifying overrepresented 'biologic process' GO annotations and specific gene sets, respectively.
Results
[00148] Three-fold cross- validated analysis identifies PTEN as a number one discriminator. The three-fold cross-validated analysis is depicted in Figure 5. Average P- values (APV) of each reporter across the 30 MW lists was used to generate a consensus ranked list.
Table 9. Three-fold I cross-validation analysis
AVP cutoff # eenes FDR
0.01 98 7%
0.02 246 7%
0.05 785 11%
PTEN is the number one gene, AVP = = 5.6 x lO"6
Table 10. PTEN Prediction Performance
All genes:
ROC = 0.785 (PO.0001) Without PTEN:
Figure imgf000062_0001
[00149] Hierarchical clustering using top 246 PTEN signature genes. The PTEN signature identifies apparent activation of the pathway by other hits than to PTEN, e.g. ERBB2 amplification and PIK3CA mutation. PlKSCA kinase domain mutations are significantly more common in the PTEN~-like cluster (Figure 7, right main branch) than PIK3CA C2 or accessory domain mutations, which are more common in the PTEN+-like cluster (i><0.019).
[00150] The PTEN signature is highly consistent with in vitro/vivo PTEN literature. For genes overexpressed in PTEN" tumors, GOMiner analysis identified 121 'biologic process' categories with P<0.05. Forty-three per cent of the significant terms related to cell metabolism, e.g. organelle biogenesis (P=O.0002), DNA metabolism (P=0.0003), and RNA processing (P=0.006). Thirty-eight per cent of the significant terms related to cell proliferation, e.g., cytokinesis (P<0.00001) and cell proliferation (P=0.002). For genes overexpressed in PTEN+ tumors, GOMiner analysis identified 46 'biologic process' categories with P<0.05, e.g., MAPKKK cascade (P=0.003), inactivation of MAPK (P=0.0039), and morphogenesis (P=0.0285).
The results of the gene set enrichment analysis are presented in the following table:
Figure imgf000062_0002
[00151] The PTEN signature predicts clinical outcome in independent breast cancer datasets. Hierarchical clustering of the 105 tumors using the top 246 PTEN signature genes was used to identify two tumor classes, PTElsT-like and PTEN" -like, which have significant difference in metastasis-free survival (Figure 9A). To classify other independent breast cancer datasets, a nearest centroid classifier was used. For each test sample, the Pearson correlations between its expression levels and the class centroids are calculated. A test sample is classified based on to which centroid it is most highly correlated. This yielded tumor groups with significant differences in outcome in the van de Vijver (Figures 9B and 9C) and Sotitiou (Figures 9D and 9E) datasets, respectively. Similar results were obtained when removing all Gene Oncology 'cell cycle'-related genes.
Conclusions
[00152] The results presented in Example 2 show that PTEN status is associated with a robust gene expression signature. The results also show that PTEN protein levels are tightly linked to the message level in breast cancer. The PTEN signature recapitulates what would be expected to be regulated by the pathway. This Example further shows that the PTEN status of a tumor can be predicted by the gene expression profile of the tumor. PIK3CA kinase domain mutations have been shown to appear to better recapitulate the signature of PTEN-loss than do C2/acc domain mutations. Finally, this Example shows that the PTEN signature has significant prognostic value in breast cancer.
EXAMPLE 3: PTEN AND BRACAl IN BASAL BREAST CANCER [00153] PTEN and BRCAl are two important proteins involved in preventing neoplasia such as breast cancer. Approximately 15% of non-hereditary breast rumors and 95% of hereditary BRCAl -associated breast tumors belong to the aggressive "basal" subtype, however the mechanisms of sporadic and BRCAl -mediate basal tumorigenesis are unclear. This Example shows that PTEN is targeted for inactivation in the majority of basal tumors, including by the novel mechanism of PTEW-intragenic chromosome breakage in BRCAl- tumors, and that nearly all mammary tumors of Pten+/" mice have basal characteristics. These results show that the PTEN pathway is a major driver of basal tumorigenesis and that PTEN pathway-targeted therapy may be an effective way to treat and possibly prevent this disease.
[00154] PTEN, the key negative regulator of oncogenic PI3K pathway signaling, and
BRCAl (breast cancer, early onset 1), a cancer susceptibility gene involved in DNA repair, are two important proteins that suppress tui~ ~rigenesis (Y. Miki et al., Science 266, 66 (1994); J. Li et al., Science 275, 1943 (1997)). In breast cancer (BC), PTEN is rarely (<5%) inactivated by mutations, however -25% have significantly diminished PTEN protein levels resulting in unrestrained PDK pathway signaling (I. Sansal, W. R. Sellers, J Clin Oncol 22, 2954 (2004)). "BasaP'-subtype breast cancer (BBC), comprising ~15% of all breast tumors and —95% of hereditary BRCAl -associated breast cancer (HBBC), was recently identified by molecular profiling and is so called because they express stratified cytokeratins (CKs) typical of basal epithelial cells of the normal mammary gland, e.g. CK5, 6, and 14 (C. M. Perou et al., Nature 406, 747 (2000); T. Sorlie et al., Proc Natl Acad Sci U S A 100, 8418 (2003)). BBCs are a particularly deadly form of BC because they are characteristically high grade and do not express the two most commonly therapeutically-targeted proteins involved in sustaining >50% of BC, ER (estrogen receptor) and HER2 (human epidermal growth factor receptor 2) (4, 5). Despite recent advances delineating BBC, the precise mechanism of BRCAl -mediated tumor suppression and the molecular lesions and oncogenic pathways driving BBC are incompletely known. The studies described in this Example were designed to investigate the role of PTEN in non-hereditary/sporadic BBC (SBBC) as well as HBBC using pathohistological and genetic approaches.
[00155] PTEN and CK5/14 protein expression was characterized in 298 sporadic stage
II BC. Among all cases and within the ER-negative group, loss of PTEN protein (PTENIHC~) was significantly associated with BBC status (CK5/14-positive) (Figure 10A). This association was validated in an independent dataset of 304 population-based breast cancers evaluated (N. A. Makretsov et al., Clin Cancer Res 10, 6143 (2004)) for PTEN and BBC status (CK5/6-positive) (Figure 10A). Despite the use of different antibodies/scoring methodologies, these data indicate that PTEN loss defines a significant subgroup of BBC.
[00156] Gene expression profiling of 19 SBBC showed BRCAl mRNA levels to be significantly reduced in PTENIHC~cases (Figure 10B) and this trend was also seen at the protein level. Promoter methylation did not fully account for this association: none had PTEN promoter methylation and 5 were BRCAl -methylated, of which 4/5 were PTENIHC". Array comparative genomic hybridization (aCGH) identified 18/20 cases with loss of DNA copy number at the PTEN locus, including one case with homozygous deletion (HD) of PTEN (Figure 10C); this case was BRCA /-methylated. Overall, PTEN1HC" SBBC had significantly lower PTEN DNA copy number than PTENIHC+ SBBC (P=0.043). SBBC were screened for PTEN and PJK3CA mutations: 2 tumors were identified with PTEN mutations (both
PTENIHC~) and another 2 tumors with H 1047RZViOC/. mutations (both PTEN1HC+). Thus, of 36 SBBC, at least 28 (78%) were PTENIHC~, mutant for PTEN or P1K3CA, or had loss of PTEN copy number.
[00157] Given the relationship between PTEN and BRCAl in SBBC, PTEN was investigated in HBBC. Remarkably, 24 of 30 HBBC (80%) were PTENIHC~ (P=I .0x10-7) (6). Moreover, suggestive of a genetic mechanism of PTEN inactivation, 62% of cases had a striking staining pattern with completely undetectable PTEN in the tumor cells in contrast to nearby non-neoplastic cells (Figure 10D). PTEN was sequenced, however no mutations were identified. Therefore, studies were designed to determine whether gene duplications and/or chromosomal breakage intragenic to PTEN could be a mechanism of gene inactivation. Utilizing the BRCAl-mutant MDA-MB-436 cell line as a model (PTEN wild-type but expresses no full-length cDNA), a split-probe fluorescence in situ hybridization (SP-FISH) assay was developed (Figure 1 OE): a large inversion disrupting PTEN, which was localized to be within intron 2 by RT-PCR, was identified in this cell line (Figure 10F). The BRCAJ- mutant cell line SUM- 149, which also is PTEN wild-type but expresses no protein, was also found to have an intragenic chromosome break interrupting PTEN (at or near intron 2 by RT- PCR analysis). 28 BRCAl tumors were analyzed by SP-FISH on tissue microarrays (TMAs); however probe splitting could not be reliably evaluated. High density copy number analysis, using Agilent aCGH arrays with over 1000 probes covering the PTEN locus, of one PTEN-IHC-null BRCAl tumor and one PTENIHC+ tumor revealed the pτENIHC~nu" case to have a focal duplication of most of the PTEN gene indicative of a complex rearrangement (Figure 10G), whereas the PTEN1HC+ had normal copy number. The BxI 1 xenograft, which was determined to harbor a 4 bp deletion in exon 11 yielding a premature stop codon, has previously been shown to have a large PTEN homozygous deletion (J. Li et al., Science 275, 1943 (1997)) which was also confirmed by aCGH analysis. The HCC- 1937 BRCAl breast cancer cell line with known PTEN deletion also was confirmed to have a homozygous deletion of PTEN. Two additional BRCAl tumors were found to harbor large PTEN homozygous deletions by aCGH. Thus, in total, 4 BRCAl tumors and 3 BRCAl cell lines were identified with genetic disruption of PTEN.
[00158] It has been hypothesized that BRCAl can act as a breast stem cell regulator whose loss directs the genesis of BBC. The high rate of PTEN loss in SBBC/HBBC shows that the PTEN/PI3K pathway may be directly involved. To that end, 8 of 10 mammary tumors derived from Pten+/" heterozygous mice were found to be CK5/14-positive, and the remaining 2 were heterogeneous with focal tumor CK5/14 positivity. [00159] Thes studies demonstrate that PTEN is frequently inactivated in BBC, thus defining a novel BBC subgroup and suggesting that activation of the PTEN/PI3K/AKT axis may be necessary for BBC development. BRCAl dysfunction appears to unmask an unknown mechanism which exerts selective pressure to genetically inactivate PTEN, opening up the possibility that other genes may be similarly rearranged in tumors with specific DNA repair deficiencies. These results have important clinical implications: targeted therapy to the PBK pathway may be an effective way to treat and possibly prevent basal breast tumorigenesis.
EXAMPLE 4: THE MAJORITY OF HUMAN BREAST CARCINOMA HAVE
PTEN/PI3K PATHWAY LESIONS
JOO 160] The tumor suppressor PTEN is a lipid phosphatase that negatively regulates the oncogenic phosphatidylinositol 3-kinase (PI3K) pathway. PI3K adds D3-phosphates to phosphatidylinositol-4,5-bisphosphate (PIP2) to generate PIP3, an important lipid second messenger that transduces potent signals, via AKT and other effectors, which act in concert to promote cellular growth, proliferation, motility, angiogenesis, genetic instability, as well as inhibit apoptosis. PTEN negatively regulates all these processes by catalyzing the precise opposite reaction to PI3K, reducing the active pool of PIP 3 (Downward, 2004; Simpson et al, 2001). In addition to these pathway members, numerous upstream receptor tyrosine kinases, such as HER2 and EGFR, have been shown to activate PTEN/PI3K pathway signaling (Bjorge et al, 1990; Peles et al, 1992).
[00161] The importance of this pathway in cancer is highlighted by the following observations: the PTEN tumor suppressor is inactivated by mutations in a variety of sporadic human rumors and cancer-predisposing hereditary syndromes, and loss of PTEN stimulates tumor development in mice (reviewed in Simpson et al, 2001). Moreover, the pi 10a catalytic subunit of PI3K encoded by PIKSCA is a transforming oncogene (Chang et al, 1997), the 3q26 region where PlKiCA is located is amplified in tumors (Shayesteh et al, 1999; Ma et al, 2000), and recently, the PIK3CA gene was shown to have activating mutations in five types of cancer (Samuels et al, 2004). Similarly, HER2 and EGFR are transforming oncogenes (Di Fiore et al, 1987a; Di Fiore, 1987b), the 17ql2 HER2 region and the 7pl 1 EGFR region are focally amplified in BC (Slamon et al, 1987; Bhargava et al, 2005; Ro et al, 1988), and transgenic mice overexpressing HER2 or EGFR develop tumors
(Bouchard et al, 1989; Muller et al, 1988; Brandt et al, 2000). Moreover, the anti-HER2 trastuzumab antibody and EGFR inhibitors such as erlotinib have been demonstrated to be efficacious against HER2 -positive breast cancer (BC) (Piccart-Gebhart et al, 2005; Romond et al, 2005) and EGFR-variant glioblastoma (as well as other tumor types), respectively, and the clinical sensitivity to these agents appear to be PTEN-dependent (Nagata et al, 2004; Mellinghoffet al, 2005).
[00162] Whereas PTEN is infrequently mutated in BC, protein expression is significantly diminished in 20-30% (Perren et al, 1999; Depowski et a, 2001; Bose et al, 2002; Saal et al, 2005). Moreover, recently we and others demonstrated that PIK3CA, has activating mutations in approximately 25% of BC (Saal et al, 2005; Bachman et al, 2004; Broderick et al, 2004) and are mutually exclusive with PTEN loss (Saal et al, 2005). As the sample cohort we studied previously was highly selected with respect to stage and estrogen receptor (ER)-status, our results suggested but did not demonstrate that abrogation at the PTEN/PI3K axis may be present in about half of BCs (Saal et al, 2005).
[00163] Classification of BC using gene expression profiling molecular portraits have identified at least 4 readily discernible subtypes: two ER-positive subtypes, luminal A, and luminal B; and two ER-negative subtypes, HER2 (which typically have amplification of HER2), and basal (which express basal cytokeratins [CK] such as CK5, 6, and 14) (Perou et al, 2000). Some studies also identify a luminal C subtype, as well as a subtype most similar to normal breast. These molecularly-defϊned subtypes appear to have inherent biological characteristics and differ significantly in clinicopathological variables such as TP53 mutational status, treatment response, and patient survival (Sorlie et al, 2001; Sorlie et al, 2003). For example, the basal subtype cannot be targeted by hormonal therapies or trastuzumab and typically are grade 3, have high rates of TP53 mutations, and poor patient outcome, whereas the luminal A subtype responds well to hormonal therapy, are low grade, have low rates of TP53 mutations, and relatively favorable outcomes (Brenton et al, 2005).
[00164] To understand the natural history of PTEN/PI3K pathway alterations and relate them to BC subtypes and other molecular and genetic alterations, a comprehensive analysis of PTEN/PI3K pathway lesions has been performed in an unselected cohort of breast tumors. The incidence of pathway lesions and their relationships to each other were analyzed, and novel associations to standard clinicopathological markers and to BC subtypes were discovered.
Materials and Methods [00165] Samples. For 116 primary breast cancer- patients treated at Columbia
University Herbert Irving Comprehensive Cancer Center/New York-Presbyterian Hospital, formalin-fixed paraffin-embedded tumor blocks and DNA isolated from corresponding frozen tumor specimens were retrieved from the Herbert Irving Comprehensive Cancer Center Tumor Bank, New York (Table 12). All samples were blinded and anonymized and obtained in accordance with Columbia University's Institutional Review Board. This unselected non-consecutive cohort is comprised of patients diagnosed between 1986 and 2003 with all stages of breast cancer and the patients received varied therapies.
Figure imgf000068_0001
Figure imgf000069_0001
[00166] Tissue Microarray Construction. Tissue microarrays (TMAs) were constructed by the Experimental Molecular Pathology Core Facility of the Herbert Irving Comprehensive Cancer Center of Columbia University utilizing a Manual Tissue Arrayer-1 device (Beecher Instruments, Sun Prairie, WI). Tumor and normal tissue areas were identified using hematoxylin and eosin-stained sections, with 3 representative tumor and 3 representative normal tissue cores of 1-mm diameter taken from each formalin-fixed, paraffin-embedded case and inserted into the recipient blocks.
[00167] Immunohistochemical Analysis. Tumors were evaluated for PTEN protein status by immunohistochemistry (IHC) on 4-μm formalin-fixed, paraffin-embedded whole- mount tissue sections using the monoclonal 138G6 PTEN antibody (Cell Signaling, Danvers, MA) at 1 :200 dilution for 2 hours at room temperature. Microwave antigen retrieval was accomplished using the Dako pH 9 solution for 20 minutes, followed by automated staining using the DakoCytomation TechMate 500 staining system with manufacturer's recommended reagents and Dako EnVision+ signal detection (Dako, Carpinteria, CA). PTEN staining intensity scores for invasive tumor and non-neoplastic cells were evaluable for 107 tumors and the tumors classified as PTEN-negative (PTEN") and PTEN-positive (PTEN+) as described in (Saal et al, 2005). Normal epithelial and endothelial cell staining were used as internal positive controls. EGFR IHC was performed using antibody 31G7 (Zymed/lnvitrogen, South San Francisco, CA) at 375 ng/ml (1 :40) on 4-μm TMA sections. Slides were treated for 5 minutes with Dako proteinase K and washed prior to incubating in primary antibody for 45 min at room temperature. Anti-mouse secondary was applied for 30 min, and the signal detected using diaminobenzidine (DAB) chromogen for 3 min followed by DAB Enhancer for 4 min (all Dako). The slides were counterstained with Gils Hematoxylin. EGFR staining was evaluable for 109 cases using a threshold of ≤IO% positive tumor cells as EGFR-negative (EGFR") and >10% positive tumor cells as EGFR-positive (EGFR+). Cytokeratin (CK) 5/14 IHC was performed on TMAs using an antibody cocktail and the cases scored CK5/14-positive (CK5/14+) or CK5/14-negative (CK5/14") as described previously (Laakso et al, 2005). Ki67 mouse monoclonal antibody clone Ki-S5 (Dako) was used at 1 :50. IHC was performed on a Dako autostainer, using a Vector biotinylated secondary anti-mouse antibody (1 :200 for 30 minutes) and Vectastain Elite detection with DAB (Vector Laboratories, Burlingame, CA). Sections were counterstained with methyl green (Sigma, St. Louis, MO). Appropriate positive and negative (staining lacking primary antibodies) controls were used in each batch of staining. Evaluation of Ki67 expression was performed by determining the percentage of positive tumor nuclei as evaluated by the CASS 200 Image Analyzer (Becton Dickinson, San Jose, CA). Cases were considered positive for Ki67 when at least 20% of the tumor cells showed evidence of nuclear expression.
[00168] HER2 amplification. HER2 amplification was assessed for 101 cases using chromogenic in situ hybridization (CISH) on TMAs, with six or more signals per cell in >50% of cancer cells scored as H£7?2-amplified (ΗER2+) (Tanner et al, 2000). Tumors in which the majority of cells contained 5 or fewer signals per nucleus were scored HER2-non- amplified (HER2~). HER2 scores using IHC were used for 6 additional cases for which CISH hybridization failed; HER2 IHC methods are described in Saal et al, 2005. Original clinical pathology workup for HER2 was utilized for another 6 cases with missing HER2 CISH and IHC data.
[00169] PCR and Sequence Analysis. Sequencing of PIK3CA exons 1, 2, 4, 5, 7, 9,
12, 13, 18, and 20 for this cohort of cases has been described previously (Saal et al, 2005). Primers details for PTEN sequencing of exons 1 to 9 are given in Table 13; the PCR reaction mixes and sequencing were performed as for PIK3CA (Saal et al, 2005). Mutational screening of TP 53 exons 2 through 11 was performed using 8 pre- validated primer assays and direct bi-directional sequencing (Agencourt Bioscience, Beverly, MA). Sequence traces were analyzed using Polyphred 5.0 (Stephens et al, 2006) or Mutation Surveyor (Softgenetics, State College, PA)
Table 13. PTEN Primers
Figure imgf000070_0001
Figure imgf000071_0001
Figure imgf000071_0002
[00170] Data Analysis and Statistical Analysis. Several additional variables were created: tumors that were either PTEN" or PTEN-mutant were categorized as 'PTEN altered' or 'abrogated'; cases that were both PTEN+ and PTEN wild-type were 'PTEN normal'. Similarly, tumors that were either PTEN altered, PIK3CA mutated, HER2+, or EGFR+ were categorized as having a PTEN/PI3K 'pathway hit'; cases that were PTEN normal, PIK3CA wild-type, HER2", and EGFR" were 'pathway hit'-negative. Hierarchical clustering of marker data and tumors were performed in Cluster 3.0 (de Hoon et al, 2004) utilizing the Pearson correlation distance metric (centered) and complete linkage algorithm and the data visualized using Java Treeview (Saldanha, 2004). The Pearson χ2 test was used for correlation analyses between two binary variables and the χ test for trend for analyses between variables with more than 2 groups using MedCalc version 9.2.0.2 (MedCalc Software, Belgium). A P-value
< 0.05 was used as the cut-off for decisions of statistical significance. Results
[00171] Distribution of Marker Lesions. The unselected cohort of breast cancer is relatively representative of the population of breast cancers seen at Columbia University (Table 12). The median age at diagnosis was 53 (range 30 to 89), 71% of cases were positive for ER, 61% positive for progesterone receptor (PgR), 31% were smaller than 2-cm and 87% were smaller than 5-cm. The majority of cases were of higher grade, about half had an S- phase fraction above 10%, and 53% had lymph node positive disease at diagnosis. As shown in Table 12, abrogation of pathway members PTEN, PIK3CA, HER2 and EGFR was common and the rates were in accordance with the literature (24%, 25%, 20%, and 24%, respectively) (Slamon et al, 1987; Perren et al, 1999; Depowski et al, 2001; Bose et al, 2002; Saal et al, 2005; Bachman et al, 2004; Campbell et al, 2004; Abd El-Rehim et al, 2004). When these 4 markers were combined into a 'pathway hit' variable, 55% of breast tumors had one or more alterations that could result in PTEN/PI3K pathway activation. Thirty-eight % of cases had deleterious mutations of TP53, in line with the approximate TP53 mutation rate in breast cancer (Borresen-Dale, 2003). Nineteen cases (19%) were positive for the basal cytokeratins CK5/14 (which herein we use to define the basal subtype of BC) in line with rates reported in U.S. populations (Carey et al, 2006).
[00172] Unsupervised Analysis Reveals PTEN/PI3K Pathway-Related Tumor
Clusters. To visualize these marker data we utilized unsupervised hierarchical clustering. As shown in Figure 1 1, the 116 breast tumors were clustered into four cohesive clusters by virtue of the inherent relationships between the markers ER, PgR, PTEN, PIK3CA, HER2, EGFR, TP53, and CK5/14. Cluster A, containing 25 tumors, is largely defined by lack of hormone receptor expression (0% expressed both), high EGFR (22/25; 88%), basal status (18/25; 72%), TP53 mutations (17/24; 71%), PTEN loss (13/24; 54%), and few HER2- amplified cases (4/24; 17%). When matched to additional clinicopathological data not utilized in the clustering algorithm (Figure 1 1), that this cluster also contained primarily grade 3 tumors (24/25; 96%) and had the highest proportion of Ki67-positive tumors (13/19; 68%). Twenty-three tumors formed cluster B, which was characterized primarily by lack of CK5/14 staining (0% positive), few EGFR+ (4/21; 19%), and a mixture of both positive and negative status for PTEN, HER2, and hormone receptors. On the basis of these latter three markers, three group B sub-clusters could be discerned: one primarily containing HER2+/ER77P53-mutant cases (cluster Bl), one containing HER2+/ER" cases (B2), and one containing PTEN" tumors that lacked expression of either ER or PgR (B3). Cluster B tumors were also predominantly high grade (74% grade 3), and 42% were Ki67-positive (8/19). A large group of 40 tumors form cluster C, which are characterized by being ER+, EGFRT, and wild-type for both PIK3CA and TP53 (all 100%). This cluster contains just 1 CK5/14+ basal tumor, 2 cases with PTEN-loss, and 3 //E/?2-positives. Cluster C contains a mixture of grades, with 48% being grade 1 or 2 (19/40); consistent with this, this cluster had mostly Ki67-negative cases (21/31 ; 68%). Finally, cluster D contains 28 rumors characterized largely by mutation of PIK3CA, positive hormone receptors, lack of CK5/14 and EGFR expression, and retention of normal PTEN. TP53 is mutated in 43% of tumors in cluster D, and these mutation-positive cases form a sub-cluster (D2) which also contains the two cases with abrogated PTEN (PTEN" or PTEN-mutated). The adjacent subcluster Dl is characterized by PIK3CA mutations without TP53. mutation. It is notable that cluster D contains very few Ki67-positive cases and few cases with an S-phase fraction above 10%.
[00173] These apparent differences in the distribution of Ki67 status, S-phase fraction, and tumor grade across these 4 main tumor clusters were significant (P=0.0068, P=0.0014, and P^O.0029, respectively; data not shown). However, the distribution of positive lymph node status at diagnosis was essentially random across the 4 tumor clusters (P=0.6435) (Figure 11).
[00174] Associations Between Pathway Lesions. Studies were designed to test the relationship of pathway lesions to each other. It has been previously reported in a series of Swedish stage II sporadic breast carcinomas that mutations of PIK3CA and loss of PTEN are mutually exclusive (Saal et al, 2005). Thus, one goal of this study was to validate and extend this discovery. In the present unselected patient cohort, only 1 tumor was both PIK3CA mutated and had significantly reduced PTEN staining (P=0.0245; Table 14). When including PTEN mutational data only 2 cases had abrogated PTEN and PIK3CA (P=0.0718), indicating that there is little selective pressure to activate the pathway by hitting both enzymes at the PIP3 axis. Interestingly, the 1 case mutated for both PTEN and P1K3CA was PTEN IHC- positive and harbored a PTEN I28T mutation together with the PIK3CA E545K mutation. PTEN mutations at residue 28, to our knowledge, have not been reported in the literature. Given that this PTEN mutant was the only one of 4 mutants that was PTEN IHC-positive, the functional relevance of this mutation is not clear.
Figure imgf000074_0001
[00175] Significantly, more than half of the cases with loss of PTEN protein expression or with altered PTEN also overexpressed EGFR (P=0.0002 and P=0.0004, respectively). No other significant associations were noted between PTEN/PI3K pathway alterations (Table 14).
[00176] Relationship of Pathway Lesions to Clinicopathologic Variables. Given the inherent biological groups identified by unsupervised hierarchical clustering analysis (Figure 11), we queried the correlation of pathway lesions to common clinicopathological markers in breast cancer. Consistent with the literature, PTEN protein loss by IHC and the 'PTEN altered' state was significantly more common in ER-negative tumors (P=0.0016 and P=0.003, respectively) and PgR-negative tumors (P=0.0007 and P=0.0016) (see Table 15). PTEN loss or abrogation was also significantly correlated to higher grade (P=0.0017 and P=0.0012, respectively) and larger tumor size (P=0.0018 and P=0.0005). Similarly, EGFR overexpression was associated to hormone receptor negativity (P<0.0001) and increasing tumor grade (P=COOl 1), but not tumor size (P=0.2202). Seventy-six % of PIK3CA mutants were ER-positive, however this was not a significant enrichment (P=0.6376). Interestingly, PIK3CA mutations correlated with lower tumor grade (P=0.0265) and lower S-phase fraction (P=0.0416) (Table 15). This result was corroborated by Ki67 staining: only 17% of PIK3CA mutants with Ki67 data had positive Ki67 staining compared to 46% of PIK3CA wild-type tumors (P=0.0235, N=90). Using the combined 'pathway hit' variable, 91% (29/32) of ER- negative tumors had one or more pathway alterations compared to 41% of ER-positive tumors (P<0.0001 ; Table 15). Similarly, 74% of PgR-negative tumors had one or more pathway alterations compared to 44% of PgR-positives (P=0.0044; Table 15).
Page 74 missing
Figure imgf000077_0001
[00177] Loss of PTEN to be very common among basal breast tumors (61% vs. 16% in non-basals; P=0.0001; Table 15). Moreover, as expected EGFR overexpression was also significantly associated to the basal subtype (79% vs. 12%; PO.0001). When utilizing the combined pathway hit variable, all but one basal tumor had ^ pathway- activating lesions (94% vs 46%; P=0.0004).
[00178] TP53 mutations were significantly positively correlated to PTEN/PI3K pathway alterations when analyzed as the combined pathway hit variable (P=0.0043), or to PTEN alterations (P=0.0062), HEΛ2-amplification (P=0.0325), and EGFR overexpression (P=0.0006) independently, but not to PIKiCA mutations (P=0.8582). Of note, 3 of 4 PTEN mutants were also TP53 mutated. As noted earlier, EGFR overexpressing tumors frequently had loss of PTEN. After excluding EGFR+ tumors, the remaining PTEN altered cases still tended to have more frequent TP53 mutations than PTEN normal cases (55% vs. 25%, P=0.0936). Consistent with the literature (see Borresen-Dale, 2003 for a review), TP53 mutations were significantly more common in ER-negative tumors compared to ER-positive tumors (62% vs. 28%, P=O-OO 16), were prevalent in CK5/14+ basal tumors (61%; P=0.0638), and increased with grade (0% mutated in grade 1, 28% in grade 2, 47% mutated in grade 3; P=0.0031).
[00179] No significant associations to patient outcome were found for the tumor group clusters, nor when testing individual or aggregated pathway lesions (data not shown). This is likely due to the fact that this cohort of patients were diagnosed over a 17 year period and received varied therapies, and the available clinical follow-up to us was relatively short (median follow-up 2.5 years, range 0 to 12 years).
Discussion
[00180] This Example shows that PTEN/PI3K pathway alterations occur in more than half of an unselected population of 116 human breast carcinomas. Nearly all ER-negative breast tumors have one or more PTEN/PI3K pathway alterations in stark contrast to the rate, in ER-positives. This result may have important clinical implications, as, with the exception of herceptin for //E/?2-amplified breast cancer, clinicians presently have no targeted therapeutic options for patients with ER-negative disease. These results show that therapies that specifically attack the PI3K pathway could be effective in this subtype of breast cancer. [00181] Furthermore, basal tumors and EGFR-overexpressing tumors frequently show loss of PTEN. This result in an unselected cohort corroborates other data has been generated in a much larger series of stage II breast cancers. The data presented herein show that cooperativity between EGFR overexpression and PIK3CA mutation is rare in vivo (5/109 cases). Moreover, it suggests that basal breast cancers may exhibit such aggressive characteristics and have poor prognosis specifically because they have robust activation of two highly oncogenic pathways, the PTEN/PI3K pathway (mediated by loss of the key PI3K. regulator PTEN with input from EGFR) as well as the MAPK pathway (mediated by EGFR). Therefore, basal breast cancers might be highly sensitive to combination therapies against both the MAPK and the PTEN/PI3K pathways. Together with recent in vitro experiments demonstrating synergistic induction of apoptosis if both EGFR and the PTEN/PI3K pathways are inhibited in PTEN-deficient cells (Bianco et al, 2003; She et al, 2003; She et al, 2005), these results provide the in vivo theoretical basis for testing combination therapies for the basal subtype of BC. For example, the FDA-approved EGFR inhibitor gefϊtinib, in combination with RADOOl or CCI-779 (inhibitors of the PI3K pathway downstream effector mTOR and which are currently being tested in clinical trials), should be explored in preclinical models of basal breast cancer or so-called "triple- negative" breast cancer (ERVPgR" /HER2"). As a corollary, these results would argue that anti-EGFR monotherapy against EGFR-overexpressing breast cancers is likely to be ineffective, as most of these tumors will be PTEN-deficient (Mellinghoff et al, 2005); moreover, it is unclear if non-amplified EGFR- positive breast tumors are in fact addicted to EGFR. These results parallel the situation seen in glioblastoma multiforme, where EGFR overexpression/amplification and overexpression of its variant, EGFRvIII, is common and PTEN is frequently deleted or mutated. This suggests that the coexistence of high levels of EGFR with PTEN loss may be a common paradigm in cancer which could potentially be exploited. This hypothesis should be extended beyond breast and brain cancers. For example, EGFR and PTEN appear to be involved in the pathogenesis of lung adenocarcinoma, however the prevalence of both lesions is not well studied. If loss of PTEN is frequent among EGFR-positive cancer, then this may be a contributing factor to the significant resistance to EGFR inhibitors seen in the clinic (Melinghoff et al, 2005) and further stresses the potential importance of combination therapies against both EGFR/MAPK pathway and the PTEN/PI3K pathway.
[00182] In addition to the previously reported high rate of TP53 mutations among
HEΛ2-amplified (Borrensen-Dale, 2003) and EGFR-overexpressing (Fox et al, 1997) BC, TP53 mutations were found to be significantly associated to PTEN abrogation. This last point is highly relevant in the context of recent work which demonstrated in human cells that oncogenic PI3K pathway activation via targeted PTEN disruption or PIK3CA mutation resulted in stabilization of p53 levels and induction of p53-mediated cellular senescence (Kim et al, 2007). It was concluded from this in vitro data that loss of PTEN oτ mutation of PIKSCA could elicit selective pressure on tumors to inactivate TP53 (Kim et al, 2007). The results presented here confirm this hypothesis in human breast tumors, at least for PTEN, and further corroborates the cooperative nature between PTEN lesions and p53 inactivation which has also been observed in mouse models (Chen et al, 2005). Moreover, these results are consistent with the models proposed relating the p53 stress response pathway to the PTEN/PI3K growth and nutrient-sensing pathways (Levine et al, 2006). These data and our results differ, however, to a report that found a mutually exclusive relationship between PTEN mutations and TP53 mutations in microdissected breast tumor cells and stroma (Kurose et al, 2002). However, this prior report in question had an unusually high rate of PTEN mutations (30%, whereas meta-analyses places the rate below 5% in BC), has never been replicated, and the reliability of the methods used has been questioned (Kern et al, 2006). Of 25 breast cancers with abrogation of PTEN and with TP53 mutational data, we found 16 cases (64%) with coincident mutation of TP53, including 3 cases with deleterious mutations to both genes. Therefore, the results presented here demonstrate that the PTEN and p53 tumor suppressors are frequently inactivated in the same individual breast tumors.
[00183] PIK3CA mutations were significantly associated with lower tumor grade, lower S-phase fraction, and negative Ki67 staining. This may indicate that, in vivo, PIK3CA mutation is a less potent driver of cell proliferation than, for example, PTEN alteration. It was previously reported that PIK3CA mutations were positively associated to ER, lymph node status, and HER2 status (Saal et al, 2005). The different results for association to ER and node status may be due to the fact that the present study is underpowered compared to the prior report; moreover the strongest association had been seen within a subset of over 150 stage II Swedish BCs. Thus, stage and population effects may also have influenced the present results. For HER2, in the present study, HER2 status was re-evaluated for the majority of cases using the highly sensitive and specific chromogenic in situ hybridization (CISH) method, and with this new data the prior correlation between HER2 and PIK3CA was negated. A significant proportion of the original clinical diagnostic HER2 evaluations changed after CISH evaluation, primarily with clinical positives now CISH-scored as negative. In light of the fact that we did not see a significant correlation between PIK3CA mutations and HER2 IHC (experimentally evaluated) in the previously-reported Swedish stage II cohort (Saal et al, 2005), it can be concluded that PIK3CA mutations and HER2 amplification are not associated. This highlights the significant improvements that have been made in HER2 status determination.
[00184] All cases except one in the identified cluster A had ≥ϊ PTEN/PI3K pathway activating lesions, and this cluster closely corresponds to the CK5/14-positive basal subtype of BC. It is predicted that cluster A-type tumors would be sensitive to combination therapies to the EGFR and PTEN/PI3K pathways, and this data suggests that it could be clinically feasible to identify such patients using IHC panels such as our own. EGFR+ is not included in this definition for the basal subtype, as has been proposed in the literature (Nielsen et al, 2004). The CK5/6+ and/or EGFR+ definition of basal status has not been tested against cohorts containing non-basal tumors classified by gene expression profiles, thus its true specificity is unknown (Nielsen et al, 2004).
[00185] PTEN/PI3K pathway lesions and our identified tumor clusters may relate to the other luminal and ER-negative BC subtypes. For example, cluster C most closely corresponds to the luminal A subtype, with its low TP53 mutation rate and relatively lower proliferative state (Sorlie et al, 2001). One would hypothesize that the cluster C profile tumors would respond well to selective ER modulators such as tamoxifen or to aromatase inhibitors. Cluster B3 appears similar to the luminal B subtype, which has been defined by some as being ER+HER2+ (Carey et al, 2006). Cluster B2 is essentially the HER2 molecular subtype, which is commonly defined as ER" and HER2+, although a few ER~HER2+ cases are scattered in other clusters. As most tumors in the B2 and B3 clusters have intact PTEN, these patients would be ideal for traztuzumab-containing regimens (Nagata et al, 2004). Where does cluster D fit in the molecular portraits? The Sorlie/Perou luminal C molecular subtype, which is not always discernible in all microarray studies, in at least one report was characterized by a high rate (80%) of TP53 mutations (Sorlie et al, 2001). Cluster D, and specifically subcluster D2, has a high rate of TP53 mutations within main the ER+ luminal branch. Extrapolating, this suggests that the poorer prognosis luminal tumors may have worse outcomes driven by mutation of PIK3CA and/or TP53. Given that PI3K pathway activation has been implicated in tamoxifen resistance (Campbell et al, 2001 ; Clark et al, 2002; Kirkegaard et al, 2005) and TP53 mutations in resistance to polychemo therapy with cyclophosphamide, methotrexate, and 5-fluorouracil (CMF) (Andersson et al, 2005), tumors with a pathway profile like cluster D may require more intensive regimens than adjuvant hormonal therapy alone or plus CMF. For example, addition of an mTOR inhibitor could be highly effective. It has been recently shown that an in v/vø-derived PTEN-loss gene expression signature is highly prognostic in independent breast cancer datasets as well as in other carcinoma types. This gene expression signature captured poor prognosis tumors with bona fide PTEN protein loss, as well as poor prognosis tumors that exhibited a PTEN-loss- like signature profile due to PIK3CA kinase domain mutations and/or HER2 amplification cooperating with moderate reductions in .PZETVmRNA (below the sensitivity of our PTEN IHC assay). Together, these results show that luminal B/C tumors, which have poorer outcome compared to luminal A's, are likely to be enriched for HER2 amplification or PIK3CA kinase domain mutations and have moderate, but not complete, downregulation of PTEN.
[00186] This Example provides the first comprehensive analysis of PTEN/PI3K pathway alterations in human breast carcinoma in vivo. Results show novel relationships between PTEN/PI3K pathway lesions and of these lesions to BC subtypes. Furthermore, the results show that aberrant PTEN/PI3K signaling is closely correlated to mutation of TP53. These results have important clinical implications for stratifying patients and designing clinical trials that target the PTEN/PI3K pathway as well as the MAPK pathway and EGFR/HER2 receptors. Finally, the fact that the majority of breast cancers have oncogenic PTEN/PI3K pathway lesions highlights the enormous need and potential benefit for new drug agents that target this pathway, the pathway associated with the worst prognosis subtypes of breast carcinoma.
Figure imgf000083_0001
Figure imgf000084_0001
Figure imgf000085_0001
Figure imgf000086_0001
Figure imgf000087_0001
Figure imgf000088_0001
Figure imgf000089_0001
Figure imgf000090_0001
REFERENCES
Abd El-Rehim DM, Pinder SE, Paish CE, et al. Expression and co-expression of the members of the epidermal growth factor receptor (EGFR) family in invasive breast carcinoma. Br J Cancer 2004;91 : 1532-42.
Andersson et al, Leukemia 2005: Gene expression profiling of leukemic cell lines reveals conserved molecular signatures among subtypes with specific genetic aberrations.
Andersson J, Larsson L, Klaar S, et al. Worse survival for TP53 (p53)-mutated breast cancer patients receiving adjuvant CMF. Ann Oncol 2005;l 6:743-8.
Bader, A. G., Kang, S., and Vogt, P. K. (2006). Cancer-specific mutations in PIK3CA are oncogenic in vivo. Proc Natl Acad Sci U S A 103, 1475-1479.
Bachman KE, Argani P, Samuels Y, et al. The P1K3CA Gene is Mutated with High Frequency in Human Breast Cancers. Cancer Biol Ther 2004;3.
Baldassarre, G., Belletti, B., Nicoloso, M. S., Schiappacassi, M., Vecchione, A., Spessotto, P., Morrione, A., Canzonieri, V., and Colombatti, A. (2005). p27(Kipl)-stathmin interaction influences sarcoma cell migration and invasion. Cancer Cell 7, 51-63.
Beer, D. G., Kardia, S. L., Huang, C. C, Giordano, T. J., Levin, A. M., Misek, D. E., Lin, L., Chen, G., Gharib, T. G., Thomas, D. G., et al (2002). Gene-expression profiles predict survival of patients with lung adenocarcinoma. Nat Med 8, 816-824.
Belmont, L. D., and Mitchison, T. J. (1996). Identification of a protein that interacts with tubulin dimers and increases the catastrophe rate of microtubules. Cell 84, 623-631.
Bhargava R, Gerald WL, Li AR, et al. EGFR gene amplification in breast cancer: correlation with epidermal growth factor receptor mRNA and protein expression and HER-2 status and absence of EGFR-activating mutations. Mod Pathol 2005;l 8: 1027-33.
Bianco R, Shin I, Ritter CA, et al. Loss of PTEN/MMACl/TEP in EGF receptor-expressing tumor cells counteracts the antitumor action of EGFR tyrosine kinase inhibitors. Oncogene 2003,22:2812-22.
BiId, A. H., Yao, G., Chang, J. T., Wang, Q., Potti, A., Chasse, D., Joshi, M. B., Harpole, D., Lancaster, J. M., Berchuck, A., et al. (2006). Oncogenic pathway signatures in human cancers as a guide to targeted therapies. Nature 439, 353-357.
Bjorge JD, Chan TO, Antczak M, Kung HJ, Fujita DJ. Activated type I phosphatidylinositol kinase is associated with the epidermal growth factor (EGF) receptor following EGF stimulation. Proc Natl Acad Sci U S A 1990;87:3816-20.
Blaveri, E., Simko, J. P., Korkola, J. E., Brewer, J. L., Baehner, F., Mehta, K., Devries, S., Koppie, T., Pejavar, S., Carroll, P., and Waldman, F. M. (2005). Bladder cancer outcome and subtype classification by gene expression. CHn Cancer Res 11, 4044-4055. Borghese, L., Fletcher, G., Mathieu, J., Atzberger, A., Eades, W. C, Cagan, R. L., and Rorth, P. (2006). Systematic analysis of the transcriptional switch inducing migration of border cells. Dev Cell 10, 497-508.
Borresen-Dale AL. TP53 and breast cancer. Hum Mutat 2003;21:292-300.
Bose, S., Chandran, S., Mirocha, J. M., and Bose, N. (2006). The Akt pathway in human breast cancer: a tissue-array-based analysis. Mod Pathol 19, 238-245.
Bose, S., Crane, A., Hibshoosh, H., Mansukhani, M., Sandweis, L., and Parsons, R. (2002). Reduced expression of PTEN correlates with breast cancer progression. Hum Pathol 33, 405- 409.
Bose, S., Wang, S. I., Terry, M. B., Hibshoosh, H., and Parsons, R. (1998). Allelic loss of chromosome 10q23 is associated with tumor progression in breast carcinomas. Oncogene 17, 123-127.
Bouchard L, Lamarre L, Tremblay PJ, Jolicoeur P. Stochastic appearance of mammary tumors in transgenic mice carrying the MMTV/c-neu oncogene. Cell 1989;57:931 -6.
Brandt R, Eisenbrandt R, Leenders F, et al. Mammary gland specific hEGF receptor transgene expression induces neoplasia and inhibits differentiation. Oncogene 2000; 19:2129- 37.
Brattsand, G. (2000). Correlation of oncoprotein 18/stathmin expression in human breast cancer with established prognostic factors. Br J Cancer 83, 31 1-318.
Brenton JD, Carey LA, Ahmed AA, Caldas C. Molecular classification and molecular forecasting of breast cancer: ready for clinical application? J Clin Oncol 2005;23:7350-60.
Broderick DK, Di C, Parrett TJ, et al. Mutations of PIK3CA in anaplastic oligodendrogliomas, high-grade astrocytomas, and medulloblastomas. Cancer Res 2004;64:5048-50.
Campbell IG, Russell SE, Choong DY, et al. Mutation of the PIK3CA gene in ovarian and breast cancer. Cancer Res 2004;64:7678-81.
Campbell RA, Bhat-Nakshatri P, Patel NM, Constantinidou D, AH S, Nakshatri H. Phosphatidylinositol 3-kinase/AKT-mediated activation of estrogen receptor alpha: a new model for anti-estrogen resistance. J Biol Chem 2001 ;276:9817-24.
Cantley, L. C, Auger, K. R., Carpenter, C, Duckworth, B., Graziani, A., Kapeller, R., and Soltoff, S. (1991). Oncogenes and signal transduction. Cell 64, 281-302.
Carey LA, Perou CM, Livasy CA, et al. Race, breast cancer subtypes, and survival in the Carolina Breast Cancer Study. Jama 2006;295:2492-502.
Chang HW, Aoki M, Fruman D, et al. Transformation of chicken cells by the gene encoding the catalytic subunit of PI 3-kinase. Science 1997;276: 1848-50. Chen Z, Trotman LC, Shaffer D, et al. Crucial role of ρ53-dependent cellular senescence in suppression of Pten-deficient tumorigenesis. Nature 2005;436:725-30.
Cicenas, J., Urban, P., Vuaroqueaux, V., Labuhn, M., Kung, W., Wight, E., Mayhew, M., Eppenberger, U., and Eppenberger-Castori, S. (2005). Increased level of phosphorylated akt measured by chemiluminescence-linked immunosorbent assay is a predictor of poor prognosis in primary breast cancer overexpressing ErbB-2. Breast Cancer Res 7, R394-401.
Clark AS, West K, Streicher S, Dennis PA. Constitutive and inducible Akt activity promotes resistance to chemotherapy, trastuzumab, or tamoxifen in breast cancer cells. MoI Cancer Ther 2002;l :707-17.
Curmi, P. A., Nogues, C, Lachkar, S., Carelle, N., Gonthier, M. P., Sobel, A., Lidereau, R., and Bieche, I. (2000). Overexpression of stathmin in breast carcinomas points out to highly proliferative tumours. Br J Cancer 82, 142-150. de Hoon, M. J., Imoto, S., Nolan, J., and Miyano, S. (2004). Open source clustering software. Bioinformatics 20, 1453-1454.
Depowski, P. L., Rosenthal, S. I., and Ross, J. S. (2001). Loss of expression of the PTEN gene protein product is associated with poor outcome in breast cancer. Mod Pathol 14, 672- 676.
Di Cristofano, A., Pesce, B., Cordon-Cardo, C, and Pandolfi, P. P. (1998). Pten is essential for embryonic development and tumour suppression. Nat Genet 19, 348-355.
Di Fiore PP, Pierce JH, Kraus MH, Segatto O, King CR, Aaronson SA. erbB-2 is a potent oncogene when overexpressed in NIH/3T3 cells. Science 1987;237: 178-82.
Di Fiore PP, Pierce JH, Fleming TP, et al. Overexpression of the human EGF receptor confers an EGF-dependent transformed phenotype to NIH 3T3 cells. Cell 1987;51 :1063-70.
Downward J. PI 3-kinase, Akt and cell survival. Semin Cell Dev Biol 2004;l 5: 177-82.
Ein-Dor, L., KeIa, I., Getz, G., Givol, D., and Domany, E. (2005). Outcome signature genes in breast cancer: is there a unique set? Bioinformatics 21, 171-178.
Eschrich, S., Yang, I., Bloom, G., Kwong, K. Y., Boulware, D., Cantor, A., Coppola, D., Kruhoffer, M., Aaltonen, L., Orntoft, T. F., et al. (2005). Molecular staging for survival prediction of colorectal cancer patients. J Clin Oncol 23, 3526-3535.
Feilotter, H. E., Coulon, V., McVeigh, J. L., Boag, A. H., Dorion-Bonnet, F., Duboue, B., Latham, W. C, Eng, C, Mulligan, L. M., and Longy, M. (1999). Analysis of the 10q23 chromosomal region and the PTEN gene in human sporadic breast carcinoma. Br J Cancer 79, 718-723.
Fox SB, Harris AL. The epidermal growth factor receptor in breast cancer. J Mammary Gland Biol Neoplasia 1997;2:131-41. Funamoto, S., Meili, R., Lee, S., Parry, L., and Firtel, R. A. (2002). Spatial and temporal regulation of 3-phosphoinositides by PI 3-kinase and PTEN mediates chemotaxis. Cell 109, 611-623.
Garcia, J. M., Silva, J., Pena, C, Garcia, V., Rodriguez, R., Cruz, M. A., Cantos, B., Provencio, M., Espana, P., and Bonilla, F. (2004). Promoter methylation of the PTEN gene is a common molecular change in breast cancer. Genes Chromosomes Cancer 41, 117-124.
Glinsky, G. V., Glinskii, A. B., Stephenson, A. J., Hoffman, R. M., and Gerald, W. L. (2004). Gene expression profiling predicts clinical outcome of prostate cancer. J CHn Invest 113, 913-923.
Gruvberger, S., Ringner, M., Chen, Y., Panavally, S., Saal, L. H., Borg, A., Ferno, M., Peterson, C, and Meltzer, P. S. (2001). Estrogen receptor status in breast cancer is associated with remarkably distinct gene expression patterns. Cancer Res 61, 5979-5984.
Hay, N. (2005). The Akt-mTOR tango and its relevance to cancer. Cancer Cell 8, 179-183.
Hayward, D. G., and Fry, A. M. (2005). Nek2 kinase in chromosome instability and cancer. Cancer Lett.
Huang, E., Cheng, S. H., Dressman, H., Pittman, J., Tsou, M. H., Horng, C. F., BiId, A., Iversen, E. S., Liao, M., Chen, C. M., et al. (2003). Gene expression predictors of breast cancer outcomes. Lancet 361, 1590-1596.
Iijima, M., and Devreotes, P. (2002). Tumor suppressor PTEN mediates sensing of chemoattractant gradients. Cell 109, 599-610.
Ikenoue, T., Kanai, F., Hikiba, Y., Obata, T., Tanaka, Y., Imamura, J., Ohta, M., Jazag, A., Guleng, B., Tateishi, K., et al. (2005). Functional analysis of PIK3CA gene mutations in human colorectal cancer. Cancer Res 65, 4562-4567.
Isakoff, S. J., Engelman, J. A., Irie, H. Y., Luo, J., Brachmann, S. M., Pearline, R. V., Cantley, L. C, and Brugge, J. S. (2005). Breast cancer-associated PIK3CA mutations are oncogenic in mammary epithelial cells. Cancer Res 65, 10992-11000.
Kern SE, Winter JM. Elegance, silence and nonsense in the mutations literature for solid tumors. Cancer Biol Ther 2006;5:349-59
Khan, J., Saal, L. H., Bittner, M. L., Chen, Y., Trent, J. M., and Meltzer, P. S. (1999). Expression profiling in cancer using cDNA microarrays. Electrophoresis 20, 223-229.
Khan, S., Kumagai, T., Vora, J., Bose, N., Sehgal, L, Koeffler, P. H., and Bose, S. (2004). PTEN promoter is methylated in a proportion of invasive breast cancers. Int J Cancer 112, 407-410.
Kim JS, Lee C, Bonifant CL, Ressom H, Waldman T. Activation of p53-dependent growth suppression in human cells by mutations in PTEN or PIK3CA. MoI Cell Biol 2007;27:662- 77. Kirkegaard T, Witton CJ, McGlynn LM, et al. AKT activation predicts outcome in breast cancer patients treated with tamoxifen. J Pathol 2005;207: 139-46.
Kurose K, Gilley K, Matsumoto S, Watson PH, Zhou XP, Eng C. Frequent somatic mutations in PTEN and TP53 are mutually exclusive in the stroma of breast carcinomas. Nat Genet 2002;32:355-7.
Laakso M, Loman N, Borg A, Isola J. Cytokeratin 5/14-positive breast cancer: true basal phenotype confined to BRCAl tumors. Mod Pathol 2005; 18: 1321-8.
Latta, E., and Chapman, W. B. (2002). PTEN mutations and evolving concepts in endometrial neoplasia. Curr Opin Obstet Gynecol 14, 59-65.
Lee, J. S., Kim, H. S., Kim, Y. B., Lee, M. C, Park, C. S., and Min, K. W. (2004). Reduced PTEN expression is associated with poor outcome and angiogenesis in invasive ductal carcinoma of the breast. Appl Immunohistochem MoI Morphol 12, 205-210.
Lei, M. (2005). The MCM complex: its role in DNA replication and implications for cancer therapy. Curr Cancer Drug Targets 5, 365-380.
Levine AJ, Feng Z, Mak TW, You H, Jin S. Coordination and communication between the p53 and IGF-I-AKT-TOR signal transduction pathways. Genes Dev 2006;20:267-75.
Li, S. Y., Rong, M., Grieu, F., and Iacopetta, B. (2006). PIK3CA mutations in breast cancer are associated with poor outcome. Breast Cancer Res Treat 96, 91-95.
Lim, K. H., and Counter, C. M. (2005). Reduction in the requirement of oncogenic Ras signaling to activation of PI3K/AKT pathway during tumor maintenance. Cancer Cell 8, 381- 392.
Liu, H., Radisky, D. C, Wang, F., and Bissell, M. J. (2004). Polarity and proliferation are controlled by distinct signaling pathways downstream of PI3-kinase in breast epithelial tumor cells. J Cell Biol 164, 603-612.
Ma YY, Wei SJ, Lin YC, et al. PIK3CA as an oncogene in cervical cancer. Oncogene 2000; 19:2739-44.
Majumder, P. K., Febbo, P. G., Bikoff, R., Berger, R., Xue, Q., McMahon, L. M., Manola, J., Brugarolas, J., McDonnell, T. J., Golub, T. R., et al. (2004). mTOR inhibition reverses Akt- dependent prostate intraepithelial neoplasia through regulation of apoptotic and HIF-I- dependent pathways. Nat Med 10, 594-601.
Matsushima-Nishiu, M., Unoki, M., Ono, K., Tsunoda, T., Minaguchi, T., Kuramoto, H., Nishida, M., Satoh, T., Tanaka, T., and Nakamura, Y. (2001). Growth and gene expression profile analyses of endometrial cancer cells expressing exogenous PTEN. Cancer Res 61, 3741-3749.
Mellinghoff, I. K., Wang, M. Y., Vivanco, I., Haas-Kogan, D. A., Zhu, S., Dia, E. Q., Lu, K. V., Yoshimoto, K., Huang, J. H., Chute, D. J., et al. (2005). Molecular determinants of the response of glioblastomas to EGFR kinase inhibitors. N Engl J Med 353, 2012-2024. Muller WJ, Sinn E, Pattengale PK, Wallace R, Leder P. Single-step induction of mammary adenocarcinoma in transgenic mice bearing the activated c-neu oncogene. Cell 1988;54:105- 15.
Nagata, Y., Lan, K. H., Zhou, X., Tan, M., Esteva, F. J., Sahin, A. A., Klos, K. S., Li, P., Monia, B. P., Nguyen, N. T., et al. (2004). PTEN activation contributes to tumor inhibition by trastuzumab, and loss of PTEN predicts trastuzumab resistance in patients. Cancer Cell 6, 117-127.
Neben, K., Korshunov, A., Benner, A., Wrobel, G., Hahn, M., Kokocinski, F., Golanov, A., Joos, S., and Lichter, P. (2004). Microarray-based screening for molecular markers in medulloblastoma revealed STKl 5 as independent predictor for survival. Cancer Res 64, 3103-3111.
Nielsen TO, Hsu FD, Jensen K, et al. Immunohistochemical and clinical characterization of the basal-like subtype of invasive breast carcinoma. Clin Cancer Res 2004; 10:5367-74.
Nutt, C. L., Mani, D. R., Betensky, R. A., Tamayo, P., Cairncross, J. G., Ladd, C.'Pohl, U., Hartmann, C, McLaughlin, M. E., Batchelor, T. T., et al. (2003). Gene expression-based classification of malignant gliomas correlates better with survival than histological classification. Cancer Res 63, 1602-1607.
Orr, G. A., Verdier-Pinard, P., McDaid, H., and Horwitz, S. B. (2003). Mechanisms of Taxol resistance related to microtubules. Oncogene 22, 7280-7295.
Panigrahi, A. R., Pinder, S. E., Chan, S. Y., Paish, E. C, Robertson, J. F., and Ellis, I. O. (2004). The role of PTEN and its signalling pathways, including AKT, in breast cancer; an assessment of relationships with other prognostic factors and with outcome. J Pathol 204, 93- 100.
Parsons, R. (2004). Human cancer, PTEN and the PI-3 kinase pathway. Semin Cell Dev Biol 15, 171-176.
Pavey, S., Johansson, P., Packer, L., Taylor, J., Stark, M., Pollock, P. M., Walker, G. J., Boyle, G. M., Harper, U., Cozzi, S. J., et al. (2004). Microarray expression profiling in melanoma reveals a BRAF mutation signature. Oncogene 23, 4060-4067.
Peles E, Lamprecht R, Ben-Levy R, Tzahar E, Yarden Y. Regulated coupling of the Neu receptor to phosphatidylinositol 3'-kinase and its release by oncogenic activation. J Biol Chem 1992;267: 12266-74.
Perou, C. M., Sorlie, T., Eisen, M. B., van de Rijn, M., Jeffrey, S. S., Rees, C. A., Pollack, J. R., Ross, D. T., Johnsen, H., Akslen, L. A., et al. (2000). Molecular portraits of human breast tumours. Nature 406, 747-752.
Perren A, Weng LP, Boag AH, et al. Immunohistochemical evidence of loss of PTEN expression in primary ductal adenocarcinomas of the breast. Am J Pathol 1999; 155: 1253-60. Piccart-Gebhart MJ, Procter M, Leyland- Jones B, et al. Trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer. N Engl J Med 2005;353:1659-72.
Podsypanina, K., Ellenson, L. H., Nemes, A., Gu, J., Tamura, M., Yamada, K. M., Cordon- Cardo, C, Catoretti, G., Fisher, P. E., and Parsons, R. (1999). Mutation of Pten/Mmacl in mice causes neoplasia in multiple organ systems. Proc Natl Acad Sci U S A 96, 1563-1568.
Pomeroy, S. L., Tamayo, P., Gaasenbeek, M., Sturla, L. M., Angelo, M., McLaughlin, M. E., Kim, J. Y., Goumnerova, L. C, Black, P. M., Lau, C, et al. (2002). Prediction of central nervous system embryonal tumour outcome based on gene expression. Nature 415, 436-442.
Puc, J., Keniry, M., Li, H. S., Pandita, T. K., Choudhury, A. D., Memeo, L., Mansukhani, M., Murty, V. V., Gaciong, Z., Meek, S. E., et al. (2005). Lack of PTEN sequesters CHKl and initiates genetic instability. Cancer Cell 7, 193-204.
Raftopoulou, M., Etienne-Manneville, S., Self, A., Nicholls, S., and Hall, A. (2004). Regulation of cell migration by the C2 domain of the tumor suppressor PTEN. Science 303, 1 179-1181.
Ramaswamy, S., Nakamura, N., Sansal, I., Bergeron, L., and Sellers, W. R. (2002). A novel mechanism of gene regulation and tumor suppression by the transcription factor FKHR. Cancer Cell 2, 81-91.
Ramaswamy, S., Ross, K. N., Lander, E. S., and Golub, T. R. (2003). A molecular signature of metastasis in primary solid tumors. Nat Genet 33, 49-54.
Ringner, M., Veerla, S., Andersson, S., Staaf, J., and Hakkinen, J. (2004). ACID: a database for microarray clone information. Bioinformatics 20, 2305-2306.
Ro J, North SM, Gallick GE, Hortobagyi GN, Gutterman JU, Blick M. Amplified and overexpressed epidermal growth factor receptor gene in uncultured primary human breast carcinoma. Cancer Res 1988;48:161-4.
Romond EH, Perez EA, Bryant J, et al. Trastuzumab plus adjuvant chemotherapy for operable HER2 -positive breast cancer. N Engl J Med 2005;353: 1673-84.
Rosenwald, A., Wright, G., Chan, W. C, Connors, J. M., Campo, E., Fisher, R. L, Gascoyne, R. D., Muller-Hermelink, H. K., Smeland, E. B., Giltnane, J. M., et al. (2002). The use of molecular profiling to predict survival after chemotherapy for diffuse large-B-cell lymphoma. N Engl J Med 346, 1937-1947.
Rubin, C. I., and Atweh, G. F. (2004). The role of stathmin in the regulation of the cell cycle. J Cell Biochem 93, 242-250.
Saal, L. H., Holm, K., Maurer, M., Memeo, L., Su, T., Wang, X., Yu, J. S., Malmstrom, P. O., Mansukhani, M., Enoksson, J., et al. (2005). PIK3CA mutations correlate with hormone receptors, node metastasis, and ERBB2, and are mutually exclusive with PTEN loss in human breast carcinoma. Cancer Res 65, 2554-2559. Saal, L. H., Troein, C, Vallon-Christersson, J., Gruvberger, S., Borg, A., and Peterson, C. (2002). BioArray Software Environment (BASE): a platform for comprehensive management and analysis of microarray data. Genome Biol 3, SOFTW ARE0003.
Saal LH, Holm K, Maurer M, et al. PIK3CA mutations correlate with hormone receptors, node metastasis, and ERBB2, and are mutually exclusive with PTEN loss in human breast carcinoma. Cancer Res 2005 ;65:2554-9.
Saldanha, A. J. (2004). Java Treeview—extensible visualization of microarray data. Bioinformatics 20, 3246-3248.
Samuels, Y., Diaz, L. A., Jr., Schmidt-Kittler, O., Cummins, J. M., Delong, L., Cheong, I., Rago, C, Huso, D. L., Lengauer, C, Kinzler, K. W., et al. (2005). Mutant PIK3CA promotes cell growth and invasion of human cancer cells. Cancer Cell 7, 561-573.
Samuels, Y., and Ericson, K. (2006). Oncogenic PI3K and its role in cancer. Curr Opin Oncol 18, 77-82.
Samuels Y, Wang Z, Bardelli A, et al. High frequency of mutations of the PIK3CA gene in human cancers. Science 2004;304:554.
Sansal, L, and Sellers, W. R. (2004). The biology and clinical relevance of the PTEN tumor suppressor pathway. J Clin Oncol 22, 2954-2963.
Schnitt, S. J. (2001). Traditional and newer pathologic factors. J Natl Cancer Inst Monogr, 22-26.
Shah, A., Swain, W. A., Richardson, D., Edwards, J., Stewart, D. J., Richardson, C. M., Swinson, D. E., Patel, D., Jones, J. L., and O'Byrne, K. J. (2005). Phospho-akt expression is associated with a favorable outcome in non-small cell lung cancer. Clin Cancer Res 11, 2930- 2936.
Shaw, L. M., Rabinovitz, I., Wang, H. H., Toker, A., and Mercurio, A. M. (1997). Activation of phosphoinositide 3-OH kinase by the alpha6beta4 integrin promotes carcinoma invasion. Cell 91, 949-960.
Shayesteh L, Lu Y, Kuo WL, et al. PIK.3CA is implicated as an oncogene in ovarian cancer. Nat Genet 1999;21:99- 102.
She QB, Solit D, Basso A, Moasser MM. Resistance to gefitinib in PTEN-null HER- overexpressing tumor cells can be overcome through restoration of PTEN function or pharmacologic modulation of constitutive phosphatidylinositol 3'-kinase/Akt pathway signaling. CHn Cancer Res 2003;9:4340-6.
She QB, Solit DB, Ye Q, O'Reilly KE, Lobo J, Rosen N. The BAD protein integrates survival signaling by EGFR/MAPK and PI3K/Akt kinase pathways in PTEN-deficient tumor cells. Cancer Cell 2005;8:287-97.
Shipp, M. A., Ross, K. N., Tamayo, P., Weng, A. P., Kutok, J. L., Aguiar, R. C, Gaasenbeek, M., Angelo, M., Reich, M., Pinkus, G. S., et al. (2002). Diffuse large B-cell lymphoma outcome prediction by gene-expression profiling and supervised machine learning. Nat Med 8, 68-74.
Shoman, N., Klassen, S., McFadden, A., Bickis, M. G., Torlakovic, E., and Chibbar, R. , (2005). Reduced PTEN expression predicts relapse in patients with breast carcinoma treated by tamoxifen. Mod Pathol 18, 250-259.
Simpson L, Parsons R. PTEN: life as a tumor suppressor. Exp Cell Res 2001 ;264:29-41.
Singh, D., Febbo, P. G., Ross, K., Jackson, D. G., Manola, J., Ladd, C, Tamayo, P., Renshaw, A. A., D'Amico, A. V., Richie, J. P., et al. (2002). Gene expression correlates of clinical prostate cancer behavior. Cancer Cell 1, 203-209.
Slamon DJ, Clark GM, Wong SG, Levin WJ, Ullrich A, McGuire WL. Human breast cancer: correlation of relapse and survival with amplification of the HER-2/neu oncogene. Science 1987;235:177-82.
Sorlie, T., Perou, C. M., Tibshirani, R., Aas, T., Geisler, S., Johnsen, H., Hastie, T., Eisen, M. B., van de Rijn, M., Jeffrey, S. S., et al. (2001). Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci U S A 98, 10869-10874.
Sorlie T, Tibshirani R, Parker J, et al. Repeated observation of breast tumor subtypes in independent gene expression data sets. Proc Natl Acad Sci U S A 2003;100:8418-23.
Sotiriou, C, Neo, S. Y., McShane, L. M., Korn, E. L., Long, P. M., Jazaeri, A., Martiat, P., Fox, S. B., Harris, A. L., and Liu, E. T. (2003). Breast cancer classification and prognosis based on gene expression profiles from a population-based study. Proc Natl Acad Sci U S A 100, 10393-10398.
Stephens, L., Williams, R., and Hawkins, P. (2005). Phosphoinositide 3-kinases as drug targets in cancer. Curr Opin Pharmacol 5, 357-365.
Stephens M, Sloan JS, Robertson PD, Scheet P, Nickerson DA. Automating sequence-based detection and genotyping of SNPs from diploid samples. Nat Genet 2006;38:375-81.
Stolarov, J., Chang, K., Reiner, A., Rodgers, L., Harmon, G. J., Wigler, M. H., and Mittal, V. (2001). Design of a retroviral-mediated ecdysone-inducible system and its application to the expression profiling of the PTEN tumor suppressor. Proc Natl Acad Sci U S A 98, 13043- 13048.
Subramanian, A., Tamayo, P., Mootha, V. K., Mukherjee, S., Ebert, B. L., Gillette, M. A., Paulovich, A., Pomeroy, S. L., Golub, T. R., Lander, E. S., and Mesirov, J. P. (2005). Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 102, 15545-15550.
Suzuki, A., de Ia Pompa, J. L., Stambolic, V., EHa, A. J., Sasaki, T., del Barco Barrantes, I., Ho, A., Wakeham, A., Itie, A., Khoo, W.. et al. (1998). High cancer susceptibility and embryonic lethality associated with mutation of the PTEN tumor suppressor gene in mice. Curr Biol 8, 1169-1178. Tang, J. M., He, Q. Y., Guo, R. X., and Chang, X. J. (2006). Phosphorylated Akt overexpression and loss of PTEN expression in non-small cell lung cancer confers poor prognosis. Lung Cancer 57, 181 -191.
Tanner M, Gancberg D, Di Leo A, et al. Chromogenic in situ hybridization: a practical alternative for fluorescence in situ hybridization to detect HER-2/neu oncogene amplification in archival breast cancer samples. Am J Pathol 2000; 157: 1467-72. van 't Veer, L. J., Dai, H., van de Vijver, M. J., He, Y. D., Hart, A. A., Mao, M., Peterse, H. L., van der Kooy, K., Marton, M. J., Witteveen, A. T., et al. (2002). Gene expression profiling predicts clinical outcome of breast cancer. Nature 415, 530-536. van de Vijver, M. J., He, Y. D., van't Veer, L. J., Dai, H., Hart, A. A., Voskuil, D. W., Schreiber, G. J., Peterse, J. L., Roberts, C, Marton, M. J., et al. (2002). A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med 347, 1999-2009.
Vapnik, V. N. (1995). The Nature of Statistical Learning Theory (New York: Springer- Verlag).
Vazquez, F., Matsuoka, S., Sellers, W. R., Yanagida, T., Ueda, M., and Devreotes, P. N. (2006). Tumor suppressor PTEN acts through dynamic interaction with the plasma membrane. Proc Natl Acad Sci U S A 103, 3633-3638.
Webster, M. A., Hutchinson, J. N., Rauh, M. J., Muthuswamy, S. K., Anton, M., Tortorice, C. G., Cardiff, R. D., Graham, F. L., Hassell, J. A., and Muller, W. J. (1998). Requirement for both She and phosphatidylinositol 3' kinase signaling pathways in polyomavirus middle T-mediated mammary tumorigenesis. MoI Cell Biol 18, 2344-2359.
Weinstein, I. B. (2002). Cancer. Addiction to oncogenes— the Achilles heal of cancer. Science 297, 63-64.
West, M., Blanchette, C, Dressman, H., Huang, E., Ishida, S., Spang, R., Zuzan, H., Olson, J. A., Jr., Marks, J. R., and Nevins, J. R. (2001). Predicting the clinical status of human breast cancer by using gene expression profiles. Proc Natl Acad Sci U S A 98, 1 1462-11467.
Whitfield, M. L., Sherlock, G., Saldanha, A. J., Murray, J. I., Ball, C. A., Alexander, K. E., Matese, J. C, Perou, C. M., Hurt, M. M., Brown, P. O., and Botstein, D. (2002). Identification of genes periodically expressed in the human cell cycle and their expression in tumors. MoI Biol Cell 13, 1977-2000.
Yoneda-Kato, N., Look, A. T., Kirstein, M. N., Valentine, M. B., Raimondi, S. C, Cohen, K. J., Carroll, A. J., and Morris, S. W. (1996). The t(3;5)(q25.1 ;q34) of myelodysplastic syndrome and acute myeloid leukemia produces a novel fusion gene, NPM-MLFl. Oncogene 12, 265-275.
Zeeberg, B. R., Feng, W., Wang, G., Wang, M. D., Fojo, A. T., Sunshine, M., Narasimhan, S., Kane, D. W., Reinhold, W. C, Lababidi, S., et al. (2003). GoMiner: a resource for biological interpretation of genomic and proteomic data. Genome Biol 4, R28.

Claims

WHAT IS CLAIMED IS:
1. A method for identifying whether a tumor has a high potential for malignancy, the method comprising:
(a) determining gene expression level in a sample of tumor cells of one or more genes selected from the group consisting of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK.3CA), phosphatase and tensin homolog (mutated in multiple advanced cancers 1), stathmin 1 /oncoprotein 18, AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N- cadherin 2), karyopherin alpha 2 (RAG cohort 1 , importin alpha 1 ), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2,
to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and
(b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of:
(i) upregulation in the profile of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof; (ii) downregulation in the profile of phosphatase and tensin homolog
(mutated in multiple advanced cancers 1), AAA domain containing 1, retinoic acid induced 2, or any combination thereof,
then the tumor has a high potential for malignancy.
2. A method for identifying whether a tumor has a high potential for malignancy, the method comprising:
(a) determining gene expression level in a sample of tumor cells of (i) PTEN, and (ii) one or more genes selected from the group consisting of phosphoinositide- 3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1 /oncoprotein 18, AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N- cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2,
to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and
(b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of:
(i) upregulation in the profile of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof; (ii) downregulation in the profile of PTEN, AAA domain containing 1, retinoic acid induced 2, or any combination thereof,
then the tumor has a high potential for malignancy.
3. A method for identifying whether a tumor has a high potential for malignancy, the method comprising:
(a) determining gene expression level in a sample of tumor cells of (i) PTEN, (ii) ATAD2, and (iii) one or more genes selected from the group consisting of phosphoinositide-3 -kinase catalytic alpha polypeptide (PIK3CA), stathmin
1 /oncoprotein 18, AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTP ase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4 A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), and chromosome 6 open reading frame 173,
to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and
(b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of:
(i) upregulation in the profile of phosphoinositide-3 -kinase catalytic alpha polypeptide (PIK3CA), stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, ATAD2, or any combination thereof; (ii) downregulation in the profile of PTEN, AAA domain containing 1 , retinoic acid induced 2, or any combination thereof,
then the tumor has a high potential for malignancy.
4. A method for identifying whether a tumor has a high potential for malignancy, the method comprising:
(a) determining gene expression level in a sample of tumor cells of (i) PTEN, (ii) ATADl , and (iii) one or more genes selected from the group consisting of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin
1 /oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1 , importin alpha 1 ), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2,
to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and
(b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of:
(i) upregulation in the profile of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1 /oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1 , importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof; (ii) downregulation in the profile of PTEN, ATADl, retinoic acid induced 2, or any combination thereof,
then the tumor has a high potential for malignancy.
5. A method for identifying whether a tumor has a high potential for malignancy, the method comprising:
(a) determining gene expression level in a sample of tumor cells of (i) PTEN, (ii) PIK3CA, and (iii) one or more genes selected from the group consisting of stathmin 1 /oncoprotein 18, AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2,
to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and
(b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of:
(i) upregulation in the profile of PIK.3CA, stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat- containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof;
(ii) downregulation in the profile of PTEN, AAA domain containing 1, retinoic acid induced 2, or any combination thereof, then the tumor has a high potential for malignancy.
6. A method for identifying whether a tumor has a high potential for malignancy, the method comprising:
(a) determining gene expression level in a sample of tumor cells from a subject of (i) stathmin/oncoprotein 18, and (ii) one or more genes selected from the group consisting of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1 , importin alpha 1 ), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2,
to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and
(b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of:
(i) upregulation in the profile of phosphoinositide-3-kinase catalytic alpha polypeptide (P1K3CA), stathmin 1/oncoprotein 18, DEP domain containing 1 , cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1 , importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof;
(ii) downregulation in the profile of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1, retinoic acid induced 2, or any combination thereof, then the tumor has a high potential for malignancy
7. A method for identifying whether a tumor has a high potential for malignancy, the method comprising:
(a) determining gene expression level in a sample of tumor cells of (i) stathmin/oncoprotein 18, (ii) ATAD2, and (iii) one or more genes selected from the group consisting of phosphoinositide-3 -kinase catalytic alpha polypeptide (PIK3CA), phosphatase and tensin homolog (mutated in multiple advanced cancers 1 ), AAA domain containing 1 , DEP domain containing 1 , cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), and chromosome 6 open reading frame 173,
to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and
(b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of:
(i) upregulation in the profile of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1 , kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4 A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, ATAD2, or any combination thereof; (ii) downregulation in the profile of phosphatase and tensin homo log (mutated in multiple advanced cancers 1), AAA domain containing 1, retinoic acid induced 2, or any combination thereof,
then the tumor has a high potential for malignancy.
8. A method for identifying whether a tumor has a high potential for malignancy, the method comprising:
(a) determining gene expression level in a sample of tumor cells of (i) stathmin/oncoprotein 18, (ii) ATADl, and (iii) one or more genes selected from the group consisting of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), phosphatase and tensin homolog (mutated in multiple advanced cancers 1), DEP domain containing 1, cadherin 12 type 2 (N- cadherin 2), karyopherin alpha 2 (RAG cohort 1 , importin alpha 1 ), Rac GTP ase activating protein 1 , kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2,
to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and
(b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of:
(i) upregulation in the profile of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1 , importin alpha 1), Rac GTPase activating protein 1 , kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4 A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof;
(ii) downregulation in the profile of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), ATADl, retinoic acid induced 2, or any combination thereof,
then the tumor has a high potential for malignancy.
9. A method for identifying whether a tumor has a high potential for malignancy, the method comprising:
(a) determining gene expression level in a sample of tumor cells of (i) stathmin/oncoprotein 18, (ii) PIK3CA, and (iii) one or more genes selected from the group consisting of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1 , kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2,
to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and
(b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of:
(i) upregulation in the profile of PIK3CA, stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat- containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof;
(ii) downregulation in the profile of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1, retinoic acid induced 2, or any combination thereof,
then the tumor has a high potential for malignancy.
10. A method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of PTEN in a sample of tumor cells, wherein detection of PTEN expression below normal indicates that the tumor has high potential for malignancy.
11. A method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of (i) PTEN, and (ii) ATAD2 in a sample of tumor cells, wherein detection of PTEN expression below normal and ATAD2 expression above normal indicates that the tumor has high potential for malignancy.
12. A method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of (i) PTEN, and (ii) ATADl in a sample of tumor cells, wherein detection of PTEN expression below normal and ATADl expression below normal indicates that the tumor has high potential for malignancy.
13. A method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of (i) PTEN, and (ii) PIK3CA in a sample of tumor cells, wherein detection of PTEN expression below normal and PIK3CA expression above normal indicates that the tumor has high potential for malignancy.
14. A method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of stathmin in a sample of tumor cells, wherein detection of stathmin expression above normal indicates that the tumor has high potential for malignancy.
15. A method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of (i) stathmin, and (ii) ATAD2 in a sample of tumor cells, wherein detection of stathmin expression above normal and ATAD2 expression above normal indicates that the tumor has high potential for malignancy.
16. A method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of (i) stathmin, and (ii) ATADl in a sample of tumor cells, wherein detection of stathmin expression above normal and ATADl expression below normal indicates that the tumor has high potential for malignancy.
17. A method for identifying whether a tumor has a high potential for malignancy in a subject, the method comprising determining a gene expression level of (i) stathmin, and (ii) PIK3CA in a sample of tumor cells, wherein detection of stathmin expression above normal and PIK.3CA expression above normal indicates that the tumor has high potential for malignancy.
18. The method of claim 1, further comprising determining gene expression level in the sample of tumor cells of one or more genes selected from the group consisting of Cyclin B2, NIMA (never in mitosis gene a)-related kinase 2, BUBl budding uninhibited by benzimidazoles 1 homolog (yeast), Cyclin A2, Cell division cycle 25A, Cell division cycle 2, Gl to S and G2 to M, Polo-like kinase 4 (Drosophila), and CHKl checkpoint homolog (S. pombe).
19. The method of claim 1 , wherein at least one gene is in the profile.
20. The method of claim 1 , wherein upregulation comprises at least a 30% increase in gene expression.
21. The method of claim 1, wherein downregulation comprises at least a 30% decrease in gene expression.
22. The method of claim 1, wherein a change in expression is within a p-value of about 0.05 or less.
23. The method of claim 1, wherein the reference profile is obtained from a sample of non- tumor cells.
24. The method of claim 1, wherein the reference profile is obtained from a sample of cells from a PTEN+ tumor.
25. The method of claim 1, wherein the determining comprises measuring mRNA level expressed by one or more of the genes.
26. The method of claim 1, wherein the determining comprises measuring protein level expressed by one or more of the genes.
27. The method of claim 1, wherein the determining comprises immunohistochemistry, immunoblotting, quantum dots, or nucleic acid hybridization.
28. The method of claim 1, wherein the tumor is a breast tumor, a prostate tumor, a bladder tumor, a lung tumor, or a diffuse large B-cell lymphoma tumor.
29. The method of claim 1, wherein the tumor comprises breast cancer.
30. The method of claim 29, wherein the breast cancer comprises non-hereditary basal-like breast cancer.
31. The method of claim 29, wherein the breast cancer comprises hereditary BRCA 1 -related basal-like breast cancer.
32. The method of claim 1, wherein the tumor has reduced expression of PTEN.
33. A method for predicting whether a subject with a tumor will respond to treatment comprising a PI3K inhibitor, the method comprising:
(a) determining gene expression level in a sample of tumor cells from a subject of one or more genes selected from the group consisting of phosphoinositide-3- kinase catalytic alpha polypeptide (PIK3CA), phosphatase and tensin homolog (mutated in multiple advanced cancers 1), stathmin 1 /oncoprotein 18, AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N- cadherin 2), karyopherin alpha 2 (RAG cohort 1, importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2,
to thereby provide a gene expression profile for the tumor, wherein the profile comprises the expression level of the one or more genes; and
(b) comparing the profile to a reference gene expression profile to identify one or more changes in gene expression, wherein if the change in gene expression is one or both of:
(i) upregulation in the profile of phosphoinositide-3-kinase catalytic alpha polypeptide (PIK3CA), stathmin 1/oncoprotein 18, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (R AG cohort 1 , importin alpha 1), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, AAA domain containing 2, or any combination thereof;
(ii) downregulation in the profile of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), AAA domain containing 1, retinoic acid induced 2, or any combination thereof,
then the subject is predicted to respond to treatment comprising a PI3K inhibitor.
34. The method of claim 33, wherein the PI3K inhibitor comprises rapamycin, or a derivative thereof.
35. The method of claim 33, wherein the subject has reduced expression or no detectable expression of a BRCAl gene.
36. The method of claim 33, wherein the subject has reduced expression or no detectable expression of an estrogen receptor (ER) gene, a HER2 gene, and a progesterone receptor (PR) gene.
Ill
37. A method for determining whether a test compound inhibits a PBK pathway, the method comprising:
(a) contacting a cell that has a PI3K pathway with a test compound;
(b) determining gene expression in the cell of one or more genes selected from the group consisting of phosphatase and tensin homolog (mutated in multiple advanced cancers 1), phosphoinositide-3 -kinase catalytic alpha polypeptide (PIK3CA), stathmin 1 /oncoprotein 18, AAA domain containing 1, DEP domain containing 1, cadherin 12 type 2 (N-cadherin 2), karyopherin alpha 2 (RAG cohort 1 , importin alpha 1 ), Rac GTPase activating protein 1, kinesin family member 14, carboxypeptidase Z, centromere protein E, retinoic acid induced 2, kinesin family member 4A, chromosome 20 open reading frame 129, baculoviral IAP repeat-containing 3, Discs large homolog 7 (Drosophila), chromosome 6 open reading frame 173, and AAA domain containing 2; and
(c) comparing whether the expression of the one or more genes of step (a) is inhibited compared to the expression of the one or more genes in a cell in the absence of the test compound, so as to identify whether the test compound inhibits a PI3K pathway.
38. The method of claim 37, wherein the determining comprises measuring mRNA level expressed by one or more of the genes.
39. The method of claim 37, wherein the determining comprises measuring protein level expressed by one or more of the genes.
40. The method of claim 37, wherein the test compound comprises a PI3K inhibitor.
41. The method of claim 40, wherein the PI3K inhibitor comprises rapamycin or a derivative thereof.
42. A method for identifying whether a tumor has a high potential for malignancy, the method comprising detecting a level of stathmin in one or more cells of the tumor, wherein detection of stathmin above normal indicates that the tumor has high potential for malignancy.
43. The method of claim 42, wherein detecting comprises detecting stathmin mRNA, protein or both.
44. The method of claim 42, further comprising determining gene expression level in the sample of tumor cells of one or more genes selected from the group consisting of Cyclin B2, NIMA (never in mitosis gene a)-related kinase 2, BUBl budding uninhibited by benzimidazoles 1 homolog (yeast), Cyclin A2, Cell division cycle 25A, Cell division cycle 2, Gl to S and G2 to M, Polo-like kinase 4 (Drosophila), and CHKl checkpoint homolog (S. pombe).
45. A method for determining whether a test compound inhibits activation of a PI3K pathway, the method comprising:
(a) contacting a cell that has a PI3K pathway with a test compound;
(b) measuring expression of stathmin in the cell of step (a); and
(c) determining whether the expression of stathmin in the cell of step (a) is inhibited compared to the expression of stathmin in a cell in the absence of the test compound, so as to identify whether the test compound inhibits activation of a PI3K pathway.
46. The method of claim 45, wherein the determining comprises measuring mRNA level expressed by one or more of the genes.
47. The method of claim 45, wherein the determining comprises measuring protein level expressed by one or more of the genes.
PCT/US2007/006338 2006-03-15 2007-03-13 Pten compositions and methods for detecting breast cancer WO2007109026A2 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US78250306P 2006-03-15 2006-03-15
US60/782,503 2006-03-15

Publications (2)

Publication Number Publication Date
WO2007109026A2 true WO2007109026A2 (en) 2007-09-27
WO2007109026A3 WO2007109026A3 (en) 2008-11-20

Family

ID=38522921

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2007/006338 WO2007109026A2 (en) 2006-03-15 2007-03-13 Pten compositions and methods for detecting breast cancer

Country Status (1)

Country Link
WO (1) WO2007109026A2 (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2009103790A2 (en) * 2008-02-21 2009-08-27 Universite Libre De Bruxelles Method and kit for the detection of genes associated with pik3ca mutation and involved in pi3k/akt pathway activation in the er-positive and her2-positive subtypes with clinical implications
WO2011022316A1 (en) * 2009-08-20 2011-02-24 The Regents Of The University Of Colorado, A Body Corporate Mirnas dysregulated in triple-negative breast cancer
EP2380991A1 (en) * 2010-04-20 2011-10-26 Universitätsklinikum Hamburg-Eppendorf Method of determining the metastatic potential of a tumor
EP2588110A1 (en) * 2010-07-02 2013-05-08 University Health Network Methods of targeting pten mutant diseases and compositions therefor
EP2823065A4 (en) * 2012-03-09 2016-03-02 Insight Genetics Inc Methods and compositions relating to diagnosing and treating receptor tyrosine kinase related cancers
US9796703B2 (en) 2010-04-06 2017-10-24 University Health Network Synthesis of chiral 2-(1H-indazol-6-yl)-spiro[cyclopropane-1,3′-indolin]-2′-ones
USRE47731E1 (en) 2009-04-06 2019-11-19 University Health Network Kinase inhibitors and method of treating cancer with same

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
ISAKOFF S. ET AL.: 'Breast Cancer Associated P1K3CA Mutation Are Oncogenic in Mammary Epithelial Cells' CANCER RESEARCH vol. 65, no. 23, 2005, pages 10992 - 11000 *
WOENCKHAUS J. ET AL.: 'Genomic gain of P1K3CA and increased expression of p11alpha are associated with progression of dysplasia into invasive squamous cell carcinoma' JOURNAL OF PATHOLOGY vol. 198, no. 3, 2002, pages 335 - 342, XP009040955 *

Cited By (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2009103790A3 (en) * 2008-02-21 2009-11-12 Universite Libre De Bruxelles Method and kit for the detection of genes associated with pik3ca mutation and involved in pi3k/akt pathway activation in the er-positive and her2-positive subtypes with clinical implications
WO2009103790A2 (en) * 2008-02-21 2009-08-27 Universite Libre De Bruxelles Method and kit for the detection of genes associated with pik3ca mutation and involved in pi3k/akt pathway activation in the er-positive and her2-positive subtypes with clinical implications
US8580496B2 (en) 2008-02-21 2013-11-12 Universite Libre De Bruxelles Method and kit for the detection of genes associated with PIK3CA mutation and involved in PI3K/AKT pathway activation in the ER-postitive and HER2-positive subtypes with clinical implications
USRE47731E1 (en) 2009-04-06 2019-11-19 University Health Network Kinase inhibitors and method of treating cancer with same
WO2011022316A1 (en) * 2009-08-20 2011-02-24 The Regents Of The University Of Colorado, A Body Corporate Mirnas dysregulated in triple-negative breast cancer
US8507195B2 (en) 2009-08-20 2013-08-13 The Regents Of The University Of Colorado MiRNAs dysregulated in triple-negative breast cancer
US9796703B2 (en) 2010-04-06 2017-10-24 University Health Network Synthesis of chiral 2-(1H-indazol-6-yl)-spiro[cyclopropane-1,3′-indolin]-2′-ones
US10358436B2 (en) 2010-04-06 2019-07-23 University Health Network Kinase inhibitors and method of treating cancer
US10077255B2 (en) 2010-04-06 2018-09-18 University Health Network Synthesis of chiral 2-(1H-indazol-6-yl)-spiro[cyclopropane-1,3′-indolin]-2′-ones
US9907800B2 (en) 2010-04-06 2018-03-06 University Health Network Kinase inhibitors and method of treating cancer
WO2011131360A3 (en) * 2010-04-20 2012-10-18 Universitatsklinikum Hamburg-Eppendorf Method of determining the metastatic potential of a tumor
CN103003445A (en) * 2010-04-20 2013-03-27 汉堡-艾本德大学医学中心 Method of determining the metastatic potential of a tumor
EP2380991A1 (en) * 2010-04-20 2011-10-26 Universitätsklinikum Hamburg-Eppendorf Method of determining the metastatic potential of a tumor
US9402828B2 (en) 2010-07-02 2016-08-02 University Health Network Methods of targeting PTEN mutant diseases and compositions therefor
US8933070B2 (en) 2010-07-02 2015-01-13 University Health Network Methods of targeting PTEN mutant diseases and compositions therefor
EP2588110A4 (en) * 2010-07-02 2014-04-16 Univ Health Network Methods of targeting pten mutant diseases and compositions therefor
EP2588110A1 (en) * 2010-07-02 2013-05-08 University Health Network Methods of targeting pten mutant diseases and compositions therefor
EP2823065A4 (en) * 2012-03-09 2016-03-02 Insight Genetics Inc Methods and compositions relating to diagnosing and treating receptor tyrosine kinase related cancers

Also Published As

Publication number Publication date
WO2007109026A3 (en) 2008-11-20

Similar Documents

Publication Publication Date Title
US10597729B2 (en) Use of gene expression profiling to predict survival in cancer patient
JP4938672B2 (en) Methods, systems, and arrays for classifying cancer, predicting prognosis, and diagnosing based on association between p53 status and gene expression profile
Zirn et al. Expression profiling of Wilms tumors reveals new candidate genes for different clinical parameters
EP1888787B1 (en) Small interference rna for multiple myeloma
CN106574297B (en) Methods of selecting individualized triple therapy for cancer treatment
WO2007143752A2 (en) Targets in breast cancer for prognosis or therapy
US20100178651A1 (en) Bifunctional Predictors of Cancer Treatment Sensitivity and Resistance
WO2008144345A2 (en) Biomarkers and methods for determining sensitivity to insulin growth factor-1 receptor modulators
EP1917528A2 (en) Biomarkers and methods for determining sensitivity to epidermal growth factor receptor modulators
Andrade et al. Gene expression profiling of lobular carcinoma in situ reveals candidate precursor genes for invasion
CA2556890A1 (en) Breast cancer prognostics
WO2007109026A2 (en) Pten compositions and methods for detecting breast cancer
Chen et al. Development and validation of a novel clinical fluorescence in situ hybridization assay to detect JAK2 and PD-L1 amplification: a fluorescence in situ hybridization assay for JAK2 and PD-L1 amplification
Charkiewicz et al. Gene expression signature differentiates histology but not progression status of early-stage NSCLC
AU2016368696B2 (en) Gene signature of residual risk following endocrine treatment in early breast cancer
WO2017021501A1 (en) Method for the prediction of progression of bladder cancer
Saleh et al. Comparative analysis of triple-negative breast cancer transcriptomics of Kenyan, African American and Caucasian Women
Yin et al. Peripheral blood circulating microRNA‐4636/− 143 for the prognosis of cervical cancer
Gao et al. Alteration and prognostic values of collagen gene expression in patients with gastric cancer under different treatments
Rizzolo et al. Somatic alterations of targetable oncogenes are frequently observed in BRCA1/2 mutation negative male breast cancers
Murray et al. Screening for EGFR mutations in patients with head and neck cancer treated with gefitinib on a compassionate-use program: A Hellenic Cooperative Oncology Group Study
WO2017061953A1 (en) Invasive ductal carcinoma aggressiveness classification
Qin et al. mRNA Expression of FGFR1 as Potential Marker for Predicting Prognosis of Surgical Resection of Small Cell Lung Cancer may be better than Protein Expression and Gene Amplification
Hou Gene Expression Profiling of Early Stage Non-Small Cell Lung Cancer
Pröstling Improvement of the p53 diagnosis in human breast cancer

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 07752998

Country of ref document: EP

Kind code of ref document: A2

NENP Non-entry into the national phase in:

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 07752998

Country of ref document: EP

Kind code of ref document: A2