WO2014047723A1

WO2014047723A1 - Gene expression profiling for prognosis of non-small cell lung cancer

Info

Publication number: WO2014047723A1
Application number: PCT/CA2013/000822
Authority: WO
Inventors: Yohan BOSSE; Michel Laviolette; Philippe Joubert; Jean-Sebastien CARRIER
Original assignee: Universite Laval
Current assignee: Universite Laval
Priority date: 2012-09-27
Filing date: 2013-09-26
Publication date: 2014-04-03
Anticipated expiration: 2015-03-27

Description

GENE EXPRESSION PROFILING FOR PROGNOSIS OF

NON-SMALL CELL LUNG CANCER

Field of the invention

[0001] The present invention relates to markers for the prognosis of lung cancer progression or recurrence and adjustment of therapy. Particularly, the present invention relates to the measurement of KANK4 mRNA expression level in non-tumor lung tissues to predict survival of patients with non-small cell lung cancer (NSCLC), particularly stage I lung adenocarcinoma.

Background of the invention [0002] Lung cancer is the leading cause of cancer deaths worldwide, with more than a million new cases annually¹. The histopathology TNM (Tumor- Node-Metastasis) staging system is currently used to guide treatment decisions for patients with non-small cell lung cancers (NSCLC)². For a technically resectable lesion, surgery remains the first line of treatment. A significant minority of patients (25-30%) with early-stages NSCLC receive surgical intervention alone. However 35-50% of these patients will relapse within 5 years³, which suggests that the TNM staging system is insufficient to guide adjuvant chemotherapy and additional prognostic factors are urgently needed beyond existing diagnostic standard of care. [0003] Gene expression profiling offers great promise to refine the TNM staging system and improve therapeutic decisions following lung resection⁴. Gene expression profiling studies so far have focused on prognostic gene expression signatures in surgically excised tumor samples with a range of histologies and stages^4"24. Tumors are heterogeneous tissues containing multiple subclones and varying levels of cancer and normal cells²⁵,

characteristics that make the development and validation of a robust gene expression signature challenging. Other studies have shown that lung cancer signatures can be detected in the entire respiratory track, including airway epithelial cells²⁶. Accordingly, we hypothesized that non-tumor lung

parenchyma derived from patients undergoing lung cancer surgery bears gene expression signatures that can be used to predict relapse-free and overall survival. In this document we focus particularly on stage I adenocarcinoma. This subtype and stage of lung cancer represent the largest proportion of patients undergoing lung resection and will be increasingly detected with improved molecular and imaging strategies as well as more aggressive computed tomography screening for high-risk patients²⁷.

Summary of the invention

[0004] In a first aspect of the invention, there is provided a method for prognosing a cancer recurrence in a human subject having undergone surgical resection of a non-small cell lung cancer; the method comprising: a) determining KANK4 gene (entrez # 163782) expression level present in a non-tumor lung sample derived from the subject; and b) comparing said level to a control level of KANK4 gene present in a control sample; wherein a decreased level of said KANK4 expression relative to the level in the control sample indicates that the non-small cell lung cancer is at high risk of recurring in said subject.

[0005] In an alternative aspect of the invention, there is provided a method for prognosing a cancer recurrence in a human subject having undergone surgical resection of a non-small cell lung cancer; the method comprising: a) determining a gene expression profile present in a non-tumor lung sample derived from the subject, the gene expression profile comprising the level of expression of at least one gene selected from the group consisting of: ABHD9 (also 3BHD9 or EPHX3): epoxide hydrolase 3; CLEC3B: C-type lectin domain family 3, member B; HIST3H2A: histone cluster 3, H2a; LRP11 : low density lipoprotein receptor-related protein 1 1 ; METTL7B: methyltransferase like 7B; ARHGEF6: Rac/Cdc42 guanine nucleotide exchange factor (GEF) 6; ATP2A2: ATPase, Ca++ transporting, cardiac muscle, slow twitch 2; GPM6B: glycoprotein M6B; PIK3R1 : phosphoinositide-3-kinase, regulatory subunit 1 (alpha); RXFP1 : relaxin/insulin-like family peptide receptor 1 ; CYP3A5: cytochrome P450, family 3, subfamily A, polypeptide 5; KIT: v-kit Hardy- Zuckerman 4 feline sarcoma viral oncogene homolog; PKIA: protein kinase (cAMP-dependent, catalytic) inhibitor alpha; KANK4: KN motif and ankyrin repeat domains 4 (entrez # 163782); P2RY14: purinergic receptor P2Y, G-protein coupled, 14; IGSF9: immunoglobulin superfamily, member 9; ATP6V1C2: ATPase, H+ transporting, lysosomal 42kDa, V1 subunit C2; and MELK: maternal embryonic leucine zipper kinase; and b) comparing said level to a control level of same said gene present in a control sample; wherein a decreased or increased level of said gene expression relative to the level in the control sample indicates the risk of recurrence of said non-small cell lung cancer in said subject.

[0006] Particularly, the gene expression profile comprises at least one or a combination of genes such as for example, at least two, at least 3, at least 5, at least 10 genes. More particularly, the gene expression profile comprises:

KANK4 and/or GPM6B and/or CYP3A5 and/or MELK and/or METTL7B. Most particularly, the gene expression profiles comprises: KANK4; or GPM6B; or CYP3A5; or MELK; or a combination of METTL7B and KANK4; or a combination of all 18 genes herein defined.

[0007] Particularly, the gene expression level consist of GPM6B; CYP3A5; MELK; a combination of METTL7B and KANK4; or a combination of all 18 genes listed herein above.

[0008] In a further aspect, the invention provides the method as defined above further comprising classifying said subject, whereby a difference in the expression profile of the gene in said sample is used for classifying the subject with NSCLC into a poor survival group (high-risk), an average risk or a good survival group (low risk).

[0009] The present invention also provides a composition comprising a plurality of isolated nucleic acid sequences, wherein each isolated nucleic acid sequence hybridizes to: a) at least one RNA product of one of the genes as defined herein; and/or

b) a nucleic acid complementary to a),

wherein the composition is used to measure the level of RNA expression of at least one of said genes.

[0010] In a further aspect, the invention provides a kit for prognosing or classifying a subject with non-small cell lung carcinoma (NSCLC) comprising a set of detection agents capable of measuring the expression level of at least one gene in a test sample, wherein the gene is selected from the group consisting of: ABHD9 (also 3BHD9 or EPHX3): epoxide hydrolase 3; CLEC3B: C-type lectin domain family 3, member B; HIST3H2A: histone cluster 3, H2a; LRP11 : low density lipoprotein receptor-related protein 1 1 ; METTL7B:

methyltransferase like 7B; ARHGEF6: Rac/Cdc42 guanine nucleotide exchange factor (GEF) 6; ATP2A2: ATPase, Ca++ transporting, cardiac muscle, slow twitch 2; GPM6B: glycoprotein M6B; PIK3R1 : phosphoinositide-3- kinase, regulatory subunit 1 (alpha); RXFP1 : relaxin/insulin-like family peptide receptor 1 ; CYP3A5: cytochrome P450, family 3, subfamily A, polypeptide 5; KIT: v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog; PKIA: protein kinase (cAMP-dependent, catalytic) inhibitor alpha; KANK4: KN motif and ankyrin repeat domains 4 (entrez # 163782); P2RY14: purinergic receptor P2Y, G-protein coupled, 14; IGSF9: immunoglobulin superfamily, member 9; ATP6V1C2: ATPase, H+ transporting, lysosomal 42kDa, V1 subunit C2; and MELK: maternal embryonic leucine zipper kinase.

[0011] Particularly, in accordance with the kit of the invention, the detection agent is capable of measuring the expression level of: at least one, at least two, at least 3, at least 5, at least 10 genes from Table 8. Particularly, the gene expression profile comprises: KANK4 and/or GPM6B and/or CYP3A5 and/or MELK and/or METTL7B. More particularly, the gene pexpression profiles comprises: KANK4; or GPM6B; or CYP3A5; or MELK; or a

combination of METTL7B and KANK4; or a combination of all 18 genes herein defined. [0012] In a further aspect, the invention provides a method for treating a human subject being prognosed for cancer recurrence of a non-small cell lung cancer, the method comprising: a) determining a gene expression profile present in a non-tumor lung sample derived from the subject, the gene expression profile comprising the level of expression of at least one gene selected from the group consisting of: ABHD9 (also 3BHD9 or EPHX3): epoxide hydrolase 3; CLEC3B: C-type lectin domain family 3, member B; HIST3H2A: histone cluster 3, H2a; LRP11 : low density lipoprotein receptor-related protein 1 1 ; METTL7B: methyltransferase like 7B; ARHGEF6: Rac/Cdc42 guanine nucleotide exchange factor (GEF) 6; ATP2A2: ATPase, Ca++ transporting, cardiac muscle, slow twitch 2; GPM6B: glycoprotein M6B; PIK3R1 : phosphoinositide-3-kinase, regulatory subunit 1 (alpha);

RXFP1 : relaxin/insulin-like family peptide receptor 1 ; CYP3A5:

cytochrome P450, family 3, subfamily A, polypeptide 5; KIT: v-kit Hardy-

Zuckerman 4 feline sarcoma viral oncogene homolog; PKIA: protein kinase (cAMP-dependent, catalytic) inhibitor alpha; KANK4: KN motif and ankyrin repeat domains 4 (entrez # 163782); P2RY14: purinergic receptor P2Y, G-protein coupled, 14; IGSF9: immunoglobulin superfamily, member 9; ATP6V1 C2: ATPase, H+ transporting, lysosomal 42kDa, V1 subunit C2; and MELK: maternal embryonic leucine zipper kinase; and b) comparing said level to a control level of same said gene present in a control sample; wherein a decreased or increased level of said gene expression relative to the level in the control sample indicates that the non-small cell lung cancer is at high-risk of recurring in said subject, whereby the prognosed cancer recurrence in the subject will be treated. [0013] Particularly, in accordance with the method of treatment, the gene expression profile comprises at least one or a combination of genes such as for example, at least two, at least 3, at least 5, at least 10 genes. Particularly, the gene expression profile comprises: KANK4 and/or GPM6B and/or CYP3A5 and/or MELK and/or METTL7B. More particularly, the gene pexpression profiles comprises: KANK4; or GPM6B; or CYP3A5; or MELK; or a combination of METTL7B and KANK4; or a combination of all 18 genes herein defined.

[0014] According to an alternative aspect of the present invention, there is provided a biomarker for predicting the likelihood of non-small cell lung cancer (NSCLC) progression or recurrence in a human subject, comprising the identification of a variation in expression level of a gene in normal lung tissue from said subject, the gene being selected from the group consisting of:

P2RY14; GPM6B; KANK4; IGSF9; MELK; HIST3H2A; PKIA; KIT; METTL7B; ATP2A2; ARHGEF6; RXFP1 ; PIK3R1 ; ATP6V1 C2; CLEC3B; ABHD9;

CYP3A5; and LRP1 1. [0015] According to a further aspect of the present invention, there is provided a plurality of biomarkers for predicting survival of a subject with non- small cell lung cancer (NSCLC), the plurality of biomarkers comprising at least one of: P2RY14; GPM6B; KANK4; IGSF9; MELK; HIST3H2A; PKIA; KIT; METTL7B; ATP2A2; ARHGEF6; RXFP1 ; PIK3R1 ; ATP6V1 C2; CLEC3B; ABHD9; CYP3A5; and LRP1 1.

[0016] According to a further aspect of the present invention, there is provided a method for predicting survival of a patient with non-small cell lung cancer (NSCLC), the method comprising: a) determining a gene expressing profile from a sample of the patient's non-tumor lung, the gene expression profile comprising the level of expression of at least one gene from Table 8, and b) classifying the gene expression profile as being predictive of good survival (low-risk), average risk or poor survival (high-risk).

[0017] According to a further aspect of the present invention, there is provided a method for preparing a gene expression profile indicative of the need for adjuvant chemotherapy for non-small cell lung cancer (NSCLC) patient, comprising: a) determining the level of expression of at least one gene from Table 8 from a non-tumor lung tissue sample from said patient with NSCLC.

[0018] According to a further aspect of the present invention, there is provided a method of prognosing or classifying a subject with non-small cell lung cancer (NSCLC) comprising the steps: a) determining the expression of a biomarker in a test sample from the subject, wherein the biomarker correspond to at least one gene in Table 8, and b) comparing the expression of the biomarkers in the test sample with expression of reference genes in said sample, wherein a difference in the expression of the biomarkers in samples is used for prognosis or classify the subject with NSCLC into a poor survival group (high-risk) or a good survival group (low risk).

[0019] According to a further aspect of the present invention, there is provided a method of prognosing or classifying a subject with non-small cell lung cancer comprising: a) determining the expression of at least one biomarker in a test sample from the subject, wherein the biomarkers correspond to genes in Table 8, b) combining the expression of each of the biomarkers by a regression coefficient for the corresponding biomarker, c) calculating a risk score for the test sample by summing the values obtained in step (b), and d) deriving the low and high risk groups by comparing the risk score to the risk category cut-off value, wherein a risk score cut-off value is used for prognosis or classify the subject with NSCLC into a poor or good survival groups.

[0020] According to a further aspect of the present invention, there is provided a method of prognosing a subject with NSCLC comprising: a) determining expression levels of at least one biomarkers from Table 8, b) calculating a risk score for the subject from the expression levels of said biomarkers, and c) comparing the risk score to a cut-off value, wherein a risk score cut-off value is used to classify a subject into a high risk (poor survival) or low risk (good survival) groups.

[0021] According to a further aspect of the present invention, there is provided a composition comprising a plurality of isolated nucleic acid

sequences, wherein each isolated nucleic acid sequence hybridizes to: a) an RNA product of one of the 18 genes listed in Table 8; and/or b) a nucleic acid complementary to a), wherein the composition is used to measure the level of RNA expression of each of the 18 the gene listed in Table 8, or a subset thereof.

[0022] According to a further aspect of the present invention, there is provided an array comprising at least one polynucleotide probe hybridizable to an expression product of each of the 18 the genes listed in Table 8, or a subset thereof.

[0023] According to a further aspect of the present invention, there is provided a computer program product for use in conjunction with a computer having a processor and a memory connected to the processor, the computer program product comprising a computer readable storage medium having a computer mechanism encoded thereon, wherein the computer program mechanism may be loaded into the memory of the computer and cause the computer to carry out the method as defined herein. [0024] According to a further aspect of the present invention, there is provided a computer product for implementing a prognostic algorithm predicting a prognosis or classifying a subject with NSCLC comprising: a) a means for receiving values corresponding to a subject expression profile in a subject sample; and b) a database comprising a complete set of algorithm coefficients and cut-off values associated with a prognosis, wherein each value

representing the expression level of a biomarker, wherein each biomarker corresponds to one gene in Table 8; wherein the computer product selects the risk group category to the subject based on biomarker expression profile, thereby predicting a prognosis or classifying the subject. [0025] According to a further aspect of the present invention, there is provided a computer readable medium having stored thereon a data structure for storing the computer implemented product as defined above.

[0026] According to a further aspect of the present invention, there is provided a computer system comprising: a) a complete set of algorithm coefficients and cut-off values to derive the risk categories based on expression values of 18 genes in Table 8; b) a user interface capable of receiving a selection of gene expression levels of the 18 genes in Table 8 to calculate the risk score and discriminate patients into a low and high risk groups, and c) an output that displays a prediction of prognosis or therapy according to the biomarker expression profile of the 18 genes. [0027] According to a further aspect of the present invention, there is provided a kit for prognosis or classify a subject with non-small cell lung carcinoma (NSCLC) comprising a set of detection agents capable of detecting the expression products of at least one biomarker in a test sample, wherein the biomarkers are selected from the group consisting of: P2RY14; GPM6B;

KANK4; IGSF9; MELK; HIST3H2A; PKIA; KIT; METTL7B; ATP2A2; ARHGEF6; RXFP1 , PIK3R1 ; ATP6V1C2, CLEC3B; ABHD9; CYP3A5; and LRP1 1.

[0028] According to a further aspect of the present invention, there is provided a method for adapting a course of treatment of lung cancer in a human subject, comprising the steps of: a) predicting the likelihood of lung cancer recurrence in a human subject in accordance with the method as defined herein, and b) adapting a course of treatment according to whether said subject is at low-risk or high-risk of recurrence, and whereby said subject at high risk will be further treated.

Detailed description of the invention Description of the figures

[0029] Figure 1. Kaplan-Meier analysis for the six most discriminatory risk categories in the discovery set 2. Titles represent genes used to calculate the genetic risk score.

[0030] Figure 2. Kaplan-Meier analysis of overall survival in the validation set according to the six most discriminatory risk categories identified in the validation set. A total of 206 patients with stage I adenocarcinoma are considered in this analysis. Abbreviations and Definitions

Abbreviations

[0031] P2RY14: purinergic receptor P2Y, G-protein coupled, 14; GPM6B: glycoprotein M6B; KANK4: KN motif and ankyrin repeat domains 4 (entrez # 163782); IGSF9: immunoglobulin superfamily, member 9; MELK: maternal embryonic leucine zipper kinase; UBE2C: ubiquitin-conjugated enzyme E2C; HIST3H2A: histone cluster 3, H2a; PKIA: protein kinase (cAMP-dependent, catalytic) inhibitor alpha; KIT: v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog; DNAJC22: DnaJ (Hsp40) homolog, subfamily C, member 22; METTL7B: methyltransferase like 7B; ATP2A2: ATPase, Ca++

transporting, cardiac muscle, slow twitch 2; ARHGEF6: Rac/Cdc42 guanine nucleotide exchange factor (GEF) 6; RXFP1 : relaxin/insulin-like family peptide receptor 1 ; PIK3R1 : phosphoinositide-3-kinase, regulatory subunit 1 (alpha); MFI2: antigen p97 (melanoma associated) identified by monoclonal antibodies 133.2 and 96.5; ATP6V1 C2: ATPase, H+ transporting, lysosomal 42kDa, V1 subunit C2; CLEC3B: C-type lectin domain family 3, member B; ABHD9 (also 3BHD9 or EPHX3): epoxide hydrolase 3; CYP3A5: cytochrome P450, family 3, subfamily A, polypeptide 5; REX01 L1 (also REX1 ): RNA exonuclease 1 homolog (S. cerev/^'s/^'ae)-like 1 ; LRP11 : low density lipoprotein receptor-related protein 1 1.

Detailed description of particular embodiments

Method

[0032] According to a first aspect of the present invention, there is provided a biomarker and a method for predicting the likelihood of lung cancer relapse and overall survival following lung cancer resection surgery in a human subject, comprising the identification in normal lung tissue from said human subject of a variation in a gene selected from the group consisting of: P2RY14; GPM6B; KANK4; IGSF9; MELK; HIST3H2A; PKIA; KIT; METTL7B; ATP2A2; ARHGEF6; RXFP1 ; PIK3R1 ; ATP6V1 C2; CLEC3B; ABHD9; CYP3A5; and LRP1 1.

[0033] Particularly, there is provided a method for predicting survival of a patient with non-small cell lung cancer (NSCLC), the method comprising: a) determining a gene expressing profile from a sample of the patient's non-tumor lung, the gene expression profile comprising the level of expression of at least one gene from the 18 genes defined in Table 8, and b) classifying the gene expression profile as being predictive of good survival or poor survival. [0034] More particularly, the method comprises determining the level of expression of at least one gene from Table 8, still more particularly, at least 2 genes or at least 3 genes or at least 5 genes or at least 10 genes or at least 15 genes. Still, more particularly, the expression levels for all 18 genes from Table 8 are determined. [0035] Particularly, the gene expression profile comprises or consists of: GPM6B.

[0036] Particularly, the gene expression profile comprises or consists of: CYP3A5.

[0037] Particularly, the gene expression profile comprises or consists of: MELK.

[0038] Particularly, the gene expression profile comprises or consists of: a combination of METTL7B and KANK4.

[0039] Particularly, the gene expression profile comprises or consists of: a combination of all 18 genes listed herein above.

[0040] Most particularly, the gene expression profile comprises or consists of: the KANK4 gene (entrez # 163782) .

Control level

[0041] In connection with the different aspects of the present invention, the comparing step b) is carried out once to establish a threshold, or every time an assay is carried out.

[0042] Particularly, the control level is obtained from a control sample, wherein the control level of the selected gene is derived from a cell or a sample (such as lung) from a subject not afflicted with non-small cell lung cancer; or is obtained from a cell or a sample (such as lung) from a subject afflicted with non-small cell lung cancer but for which a low-risk or high survival has been established.

Reference genes

[0043] Particularly, there is provided the method as defined herein, wherein the gene expression profile is normalized with the use of average expression level of one or more reference genes in the sample. More particularly, the reference gene is selected from the group consisting of: TBP (TATA box binding protein), GAPDH (glyceraldehyde-3-phosphate

dehydrogenase), HRPT1 (hyperparathyroidism 1 ), GUSB (beta-glucoronidase) and ACTB (beta-actin).

Subject

[0044] Particularly, in accordance with the method of present invention, the subject is a human. More particularly, the subject has a stage I lung

adenocarcinoma. Biological sample

[0045] Particularly, there is provided the method as defined herein, wherein the test sample from the subject is a non-tumor sample. More particularly, the sample is a tissue or biological fluid, alternatively sputum, saliva or alveolar lavage. More particularly, the sample is a non-tumor lung tissue or a non-neoplastic pulmonary parenchyma sample.

Kit

[0046] There is also provided a kit for performing the method as defined herein. In this kit, the method is qPCR carried out on nucleic acid, such as DNA or RNA with prior amplification. Particularly, the kit comprises reagents. More particularly, the kit is useful for prognosis or classify a subject with non-small cell lung carcinoma (NSCLC) and comprises a set of detection agents capable of detecting the expression products of a set number of biomarkers in a test sample, wherein the biomarkers are selected from the group consisting of: P2RY14; GPM6B; KANK4; IGSF9; MELK; HIST3H2A; PKIA; KIT; METTL7B; ATP2A2; ARHGEF6; RXFP1 ; PIK3R1 ; ATP6V1 C2; CLEC3B; ABHD9;

CYP3A5; and LRP1 1. [0047] More particularly, the detection agents are nucleic acid probes as defined herein, more particularly as defined in Table 7. More particularly, the reagents are PCR primer-probe sets, wherein the primer is selected from the group consisting of: SEQ ID Nos. 1 1 to 58. Computer

[0048] The invention also provides for a computer program product for use in conjunction with a computer having a processor and a memory connected to the processor to carry out the method as defined herein. Particularly, there is provided a computer implemented product for predicting a prognosis or classifying a subject with NSCLC, wherein the computer implemented product selects the biomarker reference expression profile most similar to the subject biomarker expression profile, to thereby predict a prognosis or classify the subject. Additionally, there is provided a computer readable medium having stored thereon a data structure for storing the computer implemented product as defined above, where the data structure is capable of configuring a computer to respond to queries based on records belonging to the data structure, each of the records comprising: a) a value that identifies a biomarker reference expression profile of the 18 genes in Table 8; b) a value that identifies the probability of a prognosis associated with the biomarker reference expression profile.

Adapting course of treatment

[0049] There is also provided a method for adapting a course of treatment of lung cancer in a human subject, comprising the steps of: a) predicting the likelihood of lung cancer progression or recurrence in a human subject in accordance with the method as defined herein; and b) adapting a course of treatment according to whether said subject is a low risk, average risk or high- risk of recurrence. Particularly, when the subject is determined to be at high- risk, the subject is prescribed at least one of: chemotherapy, radiotherapy, active surveillance (watchful waiting). Most particularly, steps a) and b) are carried out after the subject has undergone lung resection.

[0050] The following examples are put forth so as to provide those of ordinary skill in the art with a complete disclosure and description of how to make and use the present invention, and are not intended to limit the scope of what the inventors regard as their invention nor are they intended to represent that the experiments below are all or the only experiments performed. Efforts have been made to ensure accuracy with respect to numbers used (e.g.

amounts, temperature, etc.) but some experimental errors and deviations should be accounted for. Unless indicated otherwise, parts are parts by weight, molecular weight is weight average molecular weight, temperature is in degrees Centigrade, and pressure is at or near atmospheric.

Examples Example 1

Materials and methods

Study participants

[0051] Frozen lung specimens for the discovery and validation sets were obtained from the Respiratory Health Network Tissue Bank of the FRQS (tissuebank.ca), site Institut Universitaire de Cardiologie et de Pneumologie de Quebec (lUCPQ). Lung tissues for the discovery set were explanted from patients who underwent surgical resection at the lUCPQ between February 2004 and December 2008. Only patients with stage I adenocarcinoma were considered. Staging was defined based on the seventh edition of the TNM cancer staging manual²⁸. Similarly, the validation set consisted of patients with stage I adenocarcinoma resected at the lUCPQ between December 1999 and September 201 1. Corresponding clinical variables including demographic data, pathology report and smoking status were taken from the biobank database. Patient's medical charts were abstracted for follow-up starting at the time of surgery, vital status, date and cause of death. Missing data from the file were obtained by contacting patients or their family. Patients were observed until death or last follow-up performed in July 2013. Exclusion criteria for the discovery and validation sets include never smokers, positive neoplastic margins on resected lung tissue and previous cancer of any origins with systemic treatment within 5 years of lung cancer surgery. A representative histologic slide of non-tumor tissue was reviewed by a pathologist (P.J.) and specimens with tumor cells were excluded from the analysis. Following these quality control filters, 1 16 and 206 patients were included in the discovery and validation sets, respectively. Table 1 shows the clinical characteristics of both cohorts used in survival analyses. All patients provided written informed consent, and the study was approved by the ethics committee of the lUCPQ. Lung specimen

[0052] Non-tumor lung specimens explanted during surgery were immediately examined by an experienced pathologist for clinical diagnosis and staging. For each specimen, a non-neoplastic pulmonary parenchyma sample was harvested from a site as distant as possible from the tumor, snap-frozen in liquid nitrogen, and stored at -80°C until further processing.

Genome-wide gene expression in the lung

[0053] Total RNA from specimens in the discovery set 1 was extracted using the SV96 Total RNA Isolation System (Promega). Expression profiling was performed using an Affymetrix custom array (GEO platform GPL10379) and testing 51 ,562 probe sets. Expression values were extracted using the Robust Multichip Average (RMA) method²⁹ as implemented in the Affymetrix Power Tools (APT) software. The quality of the arrays was judged using standard quality control parameters.

Quantitative real-time PCR (qPCR)

[0054] RNA from specimens in the discovery set 2 and validation set was extracted from 30 mg of frozen lung parenchyma specimens using RNeasy Mini kit (Qiagen). RNA concentration and purity was verified by UV 260/280 nm ratio on the NanoVue spectrophotometer (GE Healthcare). Six micrograms of RNA were converted to cDNA using Quantitect Reverse Transcription kit (Qiagen). qPCR was performed on the ABI Prism 7900HT system in 384-well plate format. Distribution of samples into the plates was performed using an 8- channel mulpipette, and distribution of the reaction mix was performed using the epMotion 5075 robot (Eppendorf). The total reaction volume was 12 μΙ containing 6 μΙ of SYBR Green JumpStart Taq ReadyMix (Sigma-Aldrich), 0.48 μΙ of ROX Reference Dye for Quantitative PCR 100X solution (Sigma-Aldrich), 1.26 μΙ of 25 mM MgCI₂, 0.72 μΙ of a 5 μΜ mix of each primer, 0.54 μΙ of water and 3 μΙ of 50X diluted cDNA or as specified in Table 7. Cycling conditions were as follows: 2 min denaturation at 94°C, and 40 amplification cycles consisting of 15 s at 94°C and 1 min at 60°C. Primers were designed using Primer3 (//frodo.wi. mit.edu/primer3/) and ordered from Integrated DNA

Technologies. Five reference genes were considered including TBP, HPRT1 , GAPDH, GUSB1 and ACTB. Primers for the targeted genes and five reference genes are shown in Table 7. Expression values were calibrated according to the standard curve. The standard curves were constructed with a minimum of seven points using serial dilutions (undiluted to 500x) from a pool of the 95 experimental samples. A total of 318 qPCR reactions were performed for each gene including 3 replicates of the following: the corresponding standard curve, a negative control containing no cDNA, a positive control and each of the experimental samples. The targeted acceptance criteria for the calibration curve were as follows: r² > 0.99 and efficacy between 90 and 1 10%. Samples were re-tested if the CV between replicates were >17% and/or found outside the upper and lower limits of the calibration curve. The average expression of the five reference genes was used to normalize expression of the targeted genes. qPCR methods were repeated for the biological validation for 206 experimental samples.

Statistical analyses

Discovery set 1

[0055] Our primary endpoint was overall survival from time of resection. Univariate Cox proportional hazards regression analyses were used to evaluate the association between survival and the level of expression of each gene evaluated on the microarray platform. The Benjamini-Hochberg procedure was used to calculate the false discovery rate (FDR)³¹. Well-annotated genes with FDR < 1 % were considered significant. Significant genes associated with survival were then transferred into a qPCR assay using the same lung specimen (i.e. discovery set 1).

Discovery set 2 (technical validation)

[0056] Univariate and multivariate Cox proportional hazards regression were performed on pPCR-based expression values for technical validation. Multivariate models include genes either all present or selected based on the following algorithms: backward selection, forward selection or penalized Cox regression. Continuous risk score for each patient were calculated from the gene expression values weighted by the regression coefficients derived from the univariate models and from linear combinations of weighted gene expression for the multivariate models. Risk score were scaled on a 1 to 100 interval. Patients were then classified as having a high-risk gene signature or a low-risk gene signature based on the median value of the risk score or as having a high-risk, an average-risk and a low-risk based on the 33^rd and 67^th tertile values of the risk score. Kaplan-Meier analysis was performed to evaluate the discriminatory performance of the risk categories on survival. Risk categories were then ranked according to log-rank test p-values. The six best performing categories were then evaluated in the validation cohort

Validation set.

[0057] Risk scores and risk categories were calculated for each participant based on the regression coefficients and the median or tertiles of the discovery set 2. Risk scores smaller than 1 or greater than 100 were replaced by 1 or 100, respectively. Discriminatory performance was evaluated by Kaplan-Meier analysis and log-rank test. The most significant risk categories (p-value < 0.01 ) were compared to pathologic staging: a Wilcoxon rank sum test was used to compare integrated area under the time-dependent receiver operator curve (iAUROC). Univariate and multivariate Cox proportional hazards regression models were used to assess the effect of available variables on overall survival. All analyses were performed with the R 2.15.3 statistical software³² using the following packages: survival³³, survcomp³⁴ and glmnet³⁵. Example 2

Results

Discovery sets 1 and 2

The mean follow-up time of the discovery set 1 (n=1 16) for survival analysis was nearly 3 years (Table 1). The individual predicting value of all probe sets on the Affymetrix array (n = 51 ,562 probe sets) were evaluated using univariate Cox proportional hazards modelling. Twenty-two (22) genes had a FDR < 1 % and were deemed significant. The individual effect of these 22 genes on survival is showed in Table 2.

Table 1. Clinical characteristics of the discovery and validation cohorts

Characteristic Discovery set 1 Discovery set Validation set

2

No. of patients ΪΪ6 95 206

Age (years; mean ± SD) 63.5 ±8.6 63.6 ± 9.0 63.0 ±9.1

Sex - no. of patients (%)

Male 51 (44) 38(40) 102(50)

Female 65 (56) 57 (60) 104 (50)

Tumor stage - no. of

patients (%)

IA 65(56) 53(56) 101 (49)

IB 51 (44) 42(44) 105(51)

≥ll 0(0) 0(0) 0(0)

Tumor type - no. of

patients (%)

Adenocarcinoma 116(100) 95(100) 206(100)

Other 0(0) 0(0) 0(0)

Smoking status - no. of

patients (%)

Never-smokers 0(0) 0(0) 0(0)

Former-smokers 88 (76) 72 (76) 138 (67)

Current-smokers 28(24) 23(24) 68(33)

Follow-up (days; median, 1136 (771-1526) 1091 (769- 970 (597- 1788)

IQR) 1506)

Deaths - no. of patients 15(13) 11 (12) 29(14)

(%

Table 2. Genes predicting survival at a false discovery rate (FDR) below 1 % in the training set

^•Likelihood ratio tests Example 3 qPCR assay

[0058] In order to develop a more practical and convenient prognostic gene expression signature, the 22 genes predicting survival on microarray- based expression values were validated by qPCR on lung specimens from the discovery set 1. 95 samples (Table 1 ) and 18 genes were available for the technical validation (discovery set 2), i.e. 19 samples were used up and 2 failed RNA quality control, four genes has unacceptable or unobtaineable standard curves, nonspecific fragments amplified or instability between replicates. Table 3 shows the results of the univariate Cox regressions, where 12 out of 18 genes were still associated with survival (5% threshold). Table 4 shows the results of the multivariate Cox regressions using all genes or genes selected based on backward, forward and penalized Cox regression. Continuous risk scores were calculated from a linear combination of the qPCR-based gene expression values weighted by the regression coefficients and scaled according to minimum and maximum values presented in Table 5. Risk categories were obtained from median and fertile values presented in Table 5. Kaplan-Meier analyses of overall survival according to the risk categories were ranked according to log-rank test p-values (Table 8), and the six risk categories with the best discriminatory performance are presented in Figure 1. All six categories were strong predictors of overall survival with p-values bellow 0.005. Low risk groups had only one death or less and high risk groups had up to ten deaths in less than four years.

Table 3. Univariate regression coefficients from qPCR-based gene expression in the discovery set 2

Regression

Gene coefficient Hazard ratio [95% CI] P-value*

KANK4 -3.3 0.0368 [0.00423, 0.32] 0.0004519

KIT -3.15 0.043 [0.00402, 0.459] 0.001316

RXFP1 -2.67 0.0695 [0.00989, 0.488] 0.002821

GPM6B -2.82 0.0594 [0.00667, 0.529] 0.002975

METTL7B 1.88 6.59 [2.15, 20.2] 0.00344

ARHGEF6 -2.59 0.0751 [0.0103, 0.547] 0.006183

PKIA -3.27 0.0381 [0.00315, 0.46] 0.007339

ATP6V1 C2 0.479 1.61 [1.17, 2.23] 0.0138

CYP3A5 -1.87 0.155 [0.0251 , 0.953] 0.0141

CLEC3B -2.05 0.128 [0.0183, 0.899] 0.01564

PIK3R1 -2.51 0.0812 [0.00745, 0.884] 0.01787

P2RY14 -2.13 0.1 19 [0.0145, 0.979] 0.03061

MELK 0.982 2.67 [1.04, 6.88] 0.05414

ATP2A2 1.86 6.42 [0.688, 59.9] 0.1305

ABHD9 1.34 3.84 [0.748, 19.7] 0.1483

HIST3H2A 0.624 1 .87 [0.822, 4.24] 0.191 1

IGSF9 0.278 1 .32 [0.93, 1.87] 0.1915

LRP1 1 0.799 2.22 [0.322, 15.4] 0.4364

*Likelihood ratio tests

Table 4. Multivariate regression coefficients from qPCR-based gene expression in the discovery set 2

Regression

Model Gene coefficient Hazard ratio [95% CI] P-value

Backward CLEC3B -1.73 0.178 [0.023, 1.37] 0.09771

METTL7B 1.45 4.26 [1.37, 13.2] 0.01222

0.0725 [0.00622,

KIT -2.62 0.846] 0.03629

Forward METTL7B 1.43 4.17 [1.26, 13.8] 0.01968

0.0557 [0.00579,

KA K4 -2.89 0.536] 0.0124

Penalized ABHD9 0.748

GPM6B -1.784

All genes ABHD9 3.64 38.2 [0.542, 2690] 0.0934

ATP6V1 C2 0.233 1.26 [0.682, 2.34] 0.4585

CLEC3B -1.65 0.193 [0.00874, 4.25] 0.2968

HIST3H2A -1.92 0.146 [0.0152, 1.41] 0.09648

LRP11 -0.392 0.676 [0.00264, 173] 0.8899

METTL7B 0.766 2.15 [0.433, 10.7] 0.3493

ARHGEF6 -1.58 0.205 [0.00127, 33.1] 0.5415

ATP2A2 -1.75 0.174 [0.000136, 221] 0.6313

GPM6B 1.15 3.16 [0.0289, 346] 0.6307

PIK3R1 1.79 5.96 [0.0941 , 377] 0.399

RXFP1 -1.78 0.168 [0.00266, 10.6] 0.3995

CYP3A5 -1.29 0.275 [0.0306, 2.46] 0.248

KIT -2.88 0.0563 [0.00167, 1.9] 0.1091

MELK 0.106 1.11 [0.141 , 8.75] 0.9195

PKIA -2.36 0.0941 [0.000591 , 15] 0.3608

KANK4 -1.5 0.222 [0.00536, 9.22] 0.4288

P2RY14 2.43 11.4 [0.241 , 534] 0.2164

IGSF9 0.0118 1.01 [0.374, 2.74] 0.9815 Table 5. Risk scores and cutoff values to define risk categories into median and fertile groups

Raw risk score Scaled risk score

Model Minimum Maximum Q33 Q50 Q67

ABHD9 0.294 3.057 21.353 25.601 30.894

CLEC3B -6.907 -0.358 72.292 78.119 83.765

HIST3H2A 0.140 2.688 12.156 14.800 20.438

LRP11 0.497 1.672 21.474 27.112 35.518

METTL7B 0.553 5.162 15.721 19.988 23.157

ARHGEF6 -7.258 -0.536 61.509 67.985 73.931

ATP2A2 1.340 3.693 19.476 26.499 31.257

GPM6B -8.003 -0.856 66.677 73.085 81.058

PIK3R1 -6.193 -0.989 63.754 70.707 78.993

RXFP1 -6.658 -0.346 53.770 62.190 67.350

CYP3A5 -4.118 -0.275 54.952 72.402 80.143

KIT -7.228 -0.823 59.296 68.178 79.471

PKIA -4.927 -0.936 42.279 49.795 60.544

KANK4 -7.759 -0.317 59.034 65.560 74.355

P2RY14 -4.068 -0.571 49.222 59.622 69.819

IGSF9 0.065 1.815 7.438 10.649 13.189

ATP6V1 C2 0.115 2.956 9.559 12.751 18.176

MELK 0.125 2.782 21.125 28.174 37.394

All genes -13.563 -1.283 41.253 52.443 59.129

Backward -6.878 2.177 34.969 45.301 52.368

Forward -5.906 3.055 43.701 50.091 58.776

Penalized -4.203 1.164 51.025 57.885 62.938 Table 6. Cox proportional hazards modeling to predict overall survival in the validation set

Univariate Mutlivariate

Hazard Ratio P-value* Hazard ratio P-value*

Risk Category

based on

KANK4

Average risk 0.5516 0.3143 0.5531

0.0041

High risk 2.5012 0.0415 2.3514

Age (>65) 2.2479 0.0329 2.1687 0.0480

Sex (male) 1.4880 0.2923 1.2556 0.5710

Smoking

(smoker) 1 .2123 0.6134 1.0549 0.7901

Staging (IB) 1.5912 0.2221 1.4855 0.2981

*Wald test for univariate models and maximum likelihood test for multivariate models

Table 7. Primers used for the quantitative real-time PCR

Table 8. Log-rank test p-values and ranking for all risk categories in the discovery set 2

Risk Category Ranking

Model Median Tertile Median based Tertile based

ABHD9 0.486944 0.052072 17 15

CLEC3B 0.178308 0.101 169 15 16

HIST3H2A 0.298723 0.744647 16 21

LRP1 1 0.714154 0.316728 19 20

METTL7B 0.137747 0.015094 14 10

ARHGEF6 0.036545 0.02461 1 10 1 1

ATP2A2 0.800665 0.229129 21 19

GPM6B 0.003968 0.038667 3 13

PIK3R1 0.047408 0.035071 1 1 12

RXFP1 0.013167 0.005214 7 4

CYP3A5 0.002917 0.013174 1 8

KIT 0.068266 0.014172 12 9

PKIA 0.1 19672 0.039476 13 14

KANK4 0.025092 0.0041 16 9 3

P2RY14 0.790573 0.103643 20 17

IGSF9 0.519312 0.799465 18 22

ATP6V1 C2 0.920056 0.174493 22 18

MELK 0.003316 0.008656 2 6

All genes 0.008861 9.4411e-06 5 1

Backward 0.010535 0.01 1296 6 7

Forward 0.022636 0.002877 8 2

Penalized 0.004233 0.006174 4 5 Validation set

[0059] 206 samples and 18 genes were available for the biological validation. Mean follow-up time for the cohort was almost 4 years (1407 days). Patient characteristics were similar to the discovery cohort (Table 1). Continuous risk scores were calculated with qPCR-based gene expression values of the validation set using coefficients and cutoffs of the six best risk models obtained in the discovery set. Kaplan-Meier analyses of overall survival according to the six risk categories are presented in Figure 2.

[0060] Three risk models were validated (p-value < 0.05, Figure 2). The single gene model GPM6B and the two gene model METTL7B+KANK4 were associated with survival. However, the most strongly associated with overall survival was the single gene model based on the expression level of KANK4 (p-value < 0.005). In this model, survival curves were not different between the average risk and low-risk individuals (Figure 2),

suggesting that the qPCR test can delineate high risk individuals from the others, but not low-risk and average-risk individuals. iAUROC of KANK4-base risk categories (0.63) and pathologic staging (0.54) were compared to test if the new risk score was better than pathologic staging to predict overall survival. Wilcoxon rank sum test p-value was smaller than 1 E-5, indicating that KANK4 gene expression values are better at predicting survival than pathologic staging.

[0061] Multivariate and univariate analyses of covariates previously associated with overall survival of lung cancer is presented in Table 6. When covariates were taken separately or together, risk categories derived from KANK4 expression values and age were associated with survival (p-value < 0.05). By univariate analysis, the low-risk and average-risk categories were not significantly different (p-value of 0.31 ).

Example 4 Discussion

[0062] In this study, we identified a prognostic gene expression signature in non- tumor lung specimens that can differentiate patients with early-stage NSCLC beyond clinicopathology staging in order to guide chemotherapy. More specifically, the mRNA expression levels of 22 genes measured in non-tumor lung tissues were found as potential biomarkers to predict overall survival following lung resection for stage I adenocarcinoma. From this discovery, we aim at developing a qPCR assay with optimal discriminatory performance. We identified many models strongly associated with survival that included the expression levels of all genes, selected combinations of genes, and single gene models. The top six models found in the discovery set were locked down and evaluated in an independent set of samples. Three models were validated and the most strongly replicated model was based of the mRNA expression level of a single gene, namely KN motif and ankyrin domain 4 (KANK4). The KANK4-qPCR assay is urgently needed in clinic to improve risk-stratification in patients undergoing surgery for stage I adenocarcinoma.

[0063] The biological function of KANK4 is largely unknown. The gene was first identified with other members of the family based on domain and phylogenetic analyses with KANK1³⁶. KANK1 was originally identified as a tumor suppressor gene in renal cell carcinoma³⁷. The Kank family of proteins consists of four members, Kank1 -Kank4³⁸. The biological function of KANK4 remains to be elucidated, but the gene was found strongly expressed in normal human lung tissues³⁶. The low level of expression associated with poor prognostic following lung tumor resection observed in the current study suggests that KANK4 may act as a tumor suppressor gene akin to KANK1 in renal cell carcinoma.

[0064] Previous studies have identified gene expression signature in lung tumor that predict survival following lung cancer surgery^4"24. Our expression signature was developed on a mature genomic technology, i.e. testing all known genes. Selection of patients were clearly defined in terms of subtype and stage (i.e. stage I adenocarcinoma). More importantly, we studied non-tumor lung tissues, instead of tumors. Tumors are

heterogeneous biological materials consisting of multiple clones that are genetically distinct²⁵. Obtaining a robust gene expression signature from tumors is likely to be very challenging. Non-tumor lung tissues harvested from a site distant from the tumor during the pathological examination are more representative of tissues that remain in the patients after surgery. We thus believe that non-tumor lung specimens are more likely to have a molecular signature that predicts relapse.

[0065] In Canada, more than 25,000 new cases of lung cancer are diagnosed yearly. For patients, surgical resection is the only hope of cure but recurrence rate following resection is very high (35-50% in 5 years)³. This recurrence rate is high compared to resection of other common solid tumors such as colon and breast cancer. Little progress has been made during the past 30 years to reduce disease recurrence and subsequent mortality after lung tumor resection. Adjuvant platinum-based chemotherapy is now a proven therapy to improve survival in advance-stages lung tumors. However, no data support the use of this chemotherapy in patients with early-stages tumors. Our new qPCR assay based on the expression of a single gene allows delineation of high-risk group. Only in Canada, this assay has the potential to expedite the clinical decision-making process for approximately 5,000 patients. Adjuvant chemotherapy in the high risk group of patients undergoing lung cancer resection for stage I adenocarcinoma has the potential to prevent relapse and save social and economic cost associated with second diagnosis and treatment. [0066] In conclusion, we are developing a qPCR prognostic gene expression signature to guide adjuvant chemotherapy following surgical resection of lung tumor for patients with stage I adenocarcinoma. This prognostic test responds to an urgent medical need in order to reduce the socio-economic burden of lung cancer. Large-scale multi-site interventional clinical trials are needed to fully demonstrate the clinical benefit of the qPCR assay, initiate negotiation with regulatory bodies and payers who properly present credentials and accept reimbursements, and finally ensure entry into the health care system.

[0067] While the invention has been described in connection with specific

embodiments thereof, it will be understood that it is capable of further modifications and this application is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the invention and including such departures from the present disclosure as come within known or customary practice within the art to which the invention pertains and as may be applied to the essential features hereinbefore set forth, and as follows in the scope of the appended claims. [0068] All patents, patent applications and publications mentioned in this

specification are herein incorporated by reference to the same extent as if each independent patent, patent application, or publication was specifically and individually indicated to be incorporated by reference.

References 1. Herbst, R.S., Heymach, J.V. & Lippman, S.M. Lung cancer. N Engl J Med 359, 1367-80 (2008). 2. Edge, S.B. ef al. AJCC Cancer Staging Manual, 646 (Springer, 2009).

3. Jemal, A., Siegel, R., Xu, J. & Ward, E. Cancer statistics, 2010. CA Cancer J Clin 60, 277- 300 (2010).

4. Subramanian, J. & Simon, R. Gene expression-based prognostic signatures in lung cancer: ready for clinical use? J Natl Cancer Inst 102, 464-74 (2010).

5. Kratz, J.R. et al. A practical molecular assay to predict survival in resected non-squamous, non-small-cell lung cancer: development and international validation studies. Lancet 379, 823-32 (2012).

6. Boutros, P.C. et al. Prognostic gene signatures for non-small-cell lung cancer. Proc Natl Acad Sci U S A 106, 2824-8 (2009).

7. Chen, H.Y. et al. A five-gene signature and clinical outcome in non-small-cell lung cancer.

N Engl J Med 356, 11-20 (2007).

8. Lau, S.K. et al. Three-gene prognostic classifier for early-stage non small-cell lung cancer.

J Clin Oncol 25, 5562-9 (2007).

9. Mostertz, W. ef al. Age- and sex-specific genomic profiles in non-small cell lung cancer.

JAMA 303, 535-43 (2010).

0. Navab, R. et al. Prognostic gene-expression signature of carcinoma-associated fibroblasts in non-small cell lung cancer. Proc Natl Acad Sci U S A 108, 7160-5 (201 1 ).

1 . Shedden, K. ef al. Gene expression-based survival prediction in lung adenocarcinoma: a multi-site, blinded validation study. Nat Med 14, 822-7 (2008).

12. Zhu, C.Q. ef al. Prognostic and predictive gene signature for adjuvant chemotherapy in resected non-small-cell lung cancer. J Clin Oncol 28, 4417-24 (2010).

13. Roepman, P. et al. An immune response enriched 72-gene prognostic profile for early- stage non-small-cell lung cancer. Clin Cancer Res 15, 284-90 (2009).

14. Sun, Z., Wigle, D.A. & Yang, P. Non-overlapping and non-cell-type-specific gene expression signatures predict lung cancer survival. J Clin Oncol 26, 877-83 (2008).

15. Skrzypski, M. ef al. Three-gene expression signature predicts survival in early-stage squamous cell carcinoma of the lung. Clin Cancer Res 14, 4794-9 (2008).

16. Raz, D.J. et al. A multigene assay is prognostic of survival in patients with early-stage lung adenocarcinoma. Clin Cancer Res 14, 5565-70 (2008).

17. Larsen, J. E. ef al. Expression profiling defines a recurrence signature in lung squamous cell carcinoma. Carcinogenesis 28, 760-6 (2007).

18. Larsen, J.E. ef al. Gene expression signature predicts recurrence in lung adenocarcinoma.

Clin Cancer Res 13, 2946-54 (2007).

19. Guo, L. ef al. Constructing molecular classifiers for the accurate prognosis of lung adenocarcinoma. Clin Cancer Res 12, 3344-54 (2006).

0. Potti, A. ef al. A genomic strategy to refine prognosis in early-stage non-small-cell lung cancer. N Engl J Med 355, 570-80 (2006). 21. Lu, Y. et al. A gene expression signature predicts survival of patients with stage I non-small cell lung cancer. PLoS Med 3, e467 (2006).

22. Raponi, M. ei al. Gene expression signatures for predicting prognosis of squamous cell and adenocarcinomas of the lung. Cancer Res 66, 7466-72 (2006).

23. Tomida, S. ef al. Gene expression-based, individualized outcome prediction for surgically treated lung cancer patients. Oncogene 23, 5360-70 (2004).

24. Beer, D.G. et al. Gene-expression profiles predict survival of patients with lung adenocarcinoma. Nat Med 8, 816-24 (2002).

25. Gerlinger, M. ef al. Intratumor heterogeneity and branched evolution revealed by multiregion sequencing. N Engl J Med 366, 883-92 (2012).

26. Spira, A. ei al. Airway epithelial gene expression in the diagnostic evaluation of smokers with suspect lung cancer. Nat Med 13, 361 -6 (2007).

27. Aberle, D.R. et al. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med 365, 395-409 (2011 ).

28. Edge, S.B. et al. AJCC Cancer Staging Manual, 646 (Springer, New York, NY, 20 0).

29. Irizarry, R.A. ef al. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4, 249-64 (2003).

30. Heber, S. & Sick, B. Quality assessment of Affymetrix GeneChip data. OMICS 10, 358-68 (2006).

31. Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Royal Stat Soc Ser B 57, 289-300 (1995).

32. R Core Team. R: A language and encironment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria (2013).

33. Therneau, T. A Package for Survival Analysis in S+. R package version 2.37-4 (2013). 34. Haibe-Kains, B., Desmedt, C, Sotiriou, C. & Bontempi, G. A comparative study of survival models for breast cancer prognostication based on microarray data: does a single gene beat them all? Bioinformatics 24, 2200-8 (2008).

35. Friedman, J., Hastie, T. & Tibshirani, R. Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Softw 33, 1 -22 (2010).

36. Zhu, Y., Kakinuma, N., Wang, Y. & Kiyama, R. Kank proteins: a new family of ankyrin- repeat domain-containing proteins. Biochim Biophys Acta 1780, 128-33 (2008).

37. Sarkar, S. ef al. A novel ankyrin repeat-containing gene (Kank) located at 9p24 is a growth suppressor of renal cell carcinoma. J Biol Chem 277, 36585-91 (2002).

38. Kakinuma, N., Zhu, Y. , Wang, Y., Roy, B.C. & Kiyama, R. Kank proteins: structure, functions and diseases. Cell Mol Life Sci 66, 2651-9 (2009).

Claims

1. A method for prognosing a cancer recurrence in a human subject having undergone surgical resection of a non-small cell lung cancer (NSCLC); the method comprising: a) determining a gene expression profile present in a non-tumor lung sample derived from the subject, the gene expression profile comprising measuring the level of expression of at least one gene selected from the group consisting of: ABHD9 (also 3BHD9 or EPHX3): epoxide hydrolase 3; CLEC3B C-type lectin domain family 3, member B; HIST3H2A: histone cluster 3, H2a; LRP11 : low density lipoprotein receptor-related protein 1 1 ; METTL7B: methyltransferase like 7B; ARHGEF6: Rac/Cdc42 guanine nucleotide exchange factor (GEF) 6; ATP2A2: ATPase, Ca++ transporting, cardiac muscle, slow twitch 2; GPM6B: glycoprotein 6B; PIK3R1 : phosphoinositide-3-kinase, regulatory subunit 1 (alpha); RXFP1 : relaxin/insulin-like family peptide receptor 1 ; CYP3A5: cytochrome P450, family 3, subfamily A, polypeptide 5; KIT: v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog; PKIA: protein kinase (cAMP-dependent, catalytic) inhibitor alpha; KANK4: KN motif and ankyrin repeat domains 4 (entrez # 163782); P2RY14: purinergic receptor P2Y, G-protein coupled, 14; IGSF9: immunoglobulin superfamily, member 9; ATP6V1C2: ATPase, H+ transporting, lysosomal 42kDa, V1 subunit C2; and MELK: maternal embryonic leucine zipper kinase; and b) comparing said level to a control level of same said gene present in a control sample; wherein a decreased or increased level of said gene expression relative to the level in the control sample indicates the risk of recurrence in the non-small cell lung cancer in said subject.

2. The method according to claim 1 , wherein said gene expression profile is that of KANK4.

3. The method according to claim 1 , wherein said gene expression profile is that of GPM6B

4. The method according to claim 1 , wherein said gene expression profile is that of CYP3A5.

5. The method according to claim 1 , wherein said gene expression profile is that of MELK.

6. The method according to claim 1 , wherein said gene expression profile is that of a combination of METTL7B and KANK4.

7. The method according to claim 1 , wherein said gene expression profile is that of a combination of all 18 genes listed in claim 1.

8. The method according to claim 1 , wherein said gene expression profile is determined using a mRNA sample.

9. The method of claim 1 , wherein the gene expression profile is normalized with the use of average expression level of one or more reference genes in the sample, said reference gene being selected from the group consisting of: TBP, GAPDH, HRPT1 , GUSB and ACTB.

10. The method of claim 1 , wherein the sample is a non-tumor lung tissue.

11. The method of claim 1 further comprising classifying said subject, whereby a difference in the expression of the gene expression profile in said sample is used for classifying the subject with NSCLC into a poor survival group (high-risk), an average risk group or a good survival group (low risk).

12. The method of claim 1 1 , further comprising classifying said subject, whereby a difference in the expression of the gene expression profile in said sample is used for classifying the subject with NSCLC into a poor survival group (high-risk) or a good survival group (low risk).

13. The method of claim 1 , wherein the gene expression profile is determined by a technique selected from the group consisting of: quantitative real-time PCR or microarray analysis.

14. The method of claim 1 , wherein the NSCLC is a stage I adenocarcinoma.

15. The method of claim 1 , wherein the subject is a human.

16. The method of claim 13, wherein the gene expression profile is determined by quantitative real-time PCR using at least one nucleic acid primer selected from the group consisting of : SEQ ID NO. 1 1 to SEQ ID NO. 58.

17. A composition comprising a plurality of isolated nucleic acid sequences, wherein each isolated nucleic acid sequence hybridizes to: a) at least one RNA product of one of the genes as defined in any one of claims 1 to 7; and/or b) a nucleic acid complementary to a), wherein the composition is used to measure the level of RNA expression of at least one of said genes.

18. A kit for prognosing or classifying a subject with non-small cell lung carcinoma (NSCLC) comprising a set of detection agents capable of determining the expression profile of at least 1 gene in a test sample, wherein the gene is selected from the group consisting of: ABHD9 (also 3BHD9 or EPHX3): epoxide hydrolase 3; CLEC3B: C-type lectin domain family 3, member B; HIST3H2A: histone cluster 3, H2a; LRP11 : low density lipoprotein receptor-related protein 1 1 ; METTL7B: methyltransferase like 7B; ARHGEF6: Rac/Cdc42 guanine nucleotide exchange factor (GEF) 6; ATP2A2: ATPase, Ca++ transporting, cardiac muscle, slow twitch 2; GPM6B: glycoprotein M6B; PIK3R1 : phosphoinositide-3-kinase, regulatory subunit 1 (alpha); RXFP1 : relaxin/insulin-like family peptide receptor 1 ; CYP3A5: cytochrome P450, family 3, subfamily A, polypeptide 5; KIT: v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog; PKIA: protein kinase (cAMP-dependent, catalytic) inhibitor alpha; KANK4: KN motif and ankyrin repeat domains 4 (entrez # 163782); P2RY14: purinergic receptor P2Y, G-protein coupled, 14; IGSF9: immunoglobulin superfamily, member 9; ATP6V1C2: ATPase, H+ transporting, lysosomal 42kDa, V1 subunit C2; and MELK: maternal embryonic leucine zipper kinase.

19. The kit according to claim 18, wherein said detection agent is capable of detecting the expression profile of KANK4.

20. The kit according to claim 18, wherein said detection agent is capable of detecting the expression profile of GPM6B.

21. The method according to claim 18, wherein said detection agent is capable of detecting the expression profile of CYP3A5.

22. The method according to claim 18, said detection agent is capable of detecting the expression profile of MELK.

23. The method according to claim 18, wherein said detection agent is capable of detecting the expression profile of a combination of METTL7B and KANK4.

24. The method according to claim 18, said detection agent is capable of detecting the expression profile of a combination of all 18 genes listed in claim 1 .

25. The kit of claim 23, wherein the detection agent is a nucleic acid primer-probe set wherein the primer is selected from the group consisting of : SEQ ID NO. 1 1 to SEQ ID NO. 58.

26. A method for treating a human subject being prognosed for cancer recurrence of a non-small cell lung cancer, the method comprising: a) determining a gene expression profile present in a non-tumor lung sample derived from the subject, the gene expression profile comprising the level of expression of at least one gene selected from the group consisting of: ABHD9 (also 3BHD9 or EPHX3): epoxide hydrolase 3; CLEC3B: C-type lectin domain family 3, member B; HIST3H2A: histone cluster 3, H2a; LRP11 : low density lipoprotein receptor-related protein 1 1 ; METTL7B: methyltransferase like 7B; ARHGEF6: Rac/Cdc42 guanine nucleotide exchange factor (GEF) 6; ATP2A2: ATPase, Ca++ transporting, cardiac muscle, slow twitch 2; GPM6B: glycoprotein M6B; PIK3R1 : phosphoinositide-3-kinase, regulatory subunit 1 (alpha); RXFP1 : relaxin/insulin-like family peptide receptor 1 ; CYP3A5: cytochrome P450, family 3, subfamily A, polypeptide 5; KIT: v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog; PKIA: protein kinase (cA P-dependent, catalytic) inhibitor alpha; KANK4: KN motif and ankyrin repeat domains 4 (entrez # 163782);

P2RY14: purinergic receptor P2Y, G-protein coupled, 14; IGSF9: immunoglobulin super-family, member 9; ATP6V1C2: ATPase, H+ transporting, lysosomal 42kDa, V1 subunit C2; and MELK: maternal embryonic leucine zipper kinase; and b) comparing said amount to a control amount of same said gene present in a control sample; wherein a decreased or increased amount of said gene expression relative to the amount in the control sample indicates that the non-small cell lung cancer will recur in said subject, whereby the prognosed cancer recurrence in the subject will be treated.

27. The method according to claim 26, wherein said gene expression profile is from KA K4.

28. The method according to claim 26, wherein said gene expression profile is from GPM6B.

29. The method according to claim 26, wherein said gene expression profile is from CYP3A5.

30. The method according to claim 26, wherein said gene expression profile is from MELK.

31. The method according to claim 26, wherein said gene expression profile is a combination of METTL7B and KANK4.

32. The method according to claim 26, wherein said gene expression profile is from a combination of all 18 genes listed in claim 26.

33. The method according to claim 26, wherein said gene expression profile is determined using a mRNA sample.

34. The method of claim 26, wherein the gene expression profile is normalized with the use of average expression level of one or more reference genes in the sample, said reference gene being selected from the group consisting of: TBP, GAPDH, HRPT1 , GUSB and ACTB.

35. The method of claim 26, wherein the sample is a non-tumor lung tissue.

36. The method of claim 1 further comprising classifying said subject, whereby a difference in the expression of the gene profile in said sample is used for classifying the subject with NSCLC into a poor survival group (high-risk), an average survival group (average risk) or a good survival group (low risk) prior to adapting the course of treatment.

37. The method of claim 36 further comprising classifying said subject, whereby a difference in the expression of the gene profile in said sample is used for classifying the subject with NSCLC into a poor survival group (high-risk) or a good survival group (low risk) prior to adapting the course of treatment.

38. The method of claim 36, wherein the gene expression profile is determined by a technique selected from the group consisting of: quantitative real-time PCR or microarray analysis.

39. The method of claim 36, wherein the NSCLC is a stage I adenocarcinoma.

40. The method of claim 36, wherein the subject is a human.

41. The method of claim 36, wherein the gene expression profile is determined by quantitative real-time PCR using at least one nucleic acid primer selected from the group consisting of : SEQ ID NO. 1 1 to SEQ ID NO. 58.