US20080026950A1

US20080026950A1 - Gene Relating to Estimation of Postoperative Prognosis for Breast Cancer

Info

Publication number: US20080026950A1
Application number: US10/590,219
Authority: US
Inventors: Mitsuru Emi; Masamitsu Onda; Hisaki Nagai
Original assignee: Mitsubishi Rayon Co Ltd
Current assignee: Mitsubishi Rayon Co Ltd; Nippon Medical School Foundation
Priority date: 2004-02-24
Filing date: 2004-08-24
Publication date: 2008-01-31
Also published as: WO2005080570A1; US20110009284A1; US9353416B2; US20130203621A1

Abstract

It is intended to provide a system of predicting the postoperative prognosis in a patient with breast cancer from the viewpoint of gene expression based on the data obtained by genome-wide and comprehensive analysis on gene expression in breast cancer. Expression of human genes is comprehensively analyzed by using a DNA microarray and gene expression functions in various breast cancer conditions are compared, thereby establishing a system of predicting the postoperative prognosis of breast cancer.

Description

TECHNICAL FIELD

The present invention relates to a gene correlated with prediction of the postoperative prognosis of breast cancer. Further, the present invention relates to a method of inspecting the postoperative prognosis of breast cancer using this gene, a method of screening cancer therapeutic medicines for controlling the postoperative prognosis of breast cancer, and a diagnosis kit for the postoperative prognosis of breast cancer.

BACKGROUND ART

Breast cancer is a disease situated as a superior cause for female lethality due to cancer, however, there are found still no dominant reasons for determining the grade of malignancy and survival prognosis from the biological standpoint.
The condition of an estrogen receptor (ER) is one determining element for clinical and biological symptoms of human breast cancer. Adjuvant hormone therapeutics is usually effective in ER-positive breast cancer patients irrespective of age, condition in the menopause, correlation with axillary nodes, and tumor diameter. However, ER-negative breast cancer is resistance to this therapeutic method (J Clin Oncology (2001) 19, 3817-1827, Breast Cancer (2001) 8, 298-304). Patients having an ER-negative tumor do not necessarily show the same response to chemical therapy. Since existent indices cannot classify breast cancer of this type depending on clinical symptom, the postoperative prognosis is recognized to be various (J Natl Inst (1991) 83, 154-155, J Natl Cancer Inst (2000) 93, 979-989).
Prognosis of breast cancer patients with no lymph node metastasis (node-negative breast cancer; n0) is better than that of metastatic breast cancer patients. However, in Japan, the present inventors have found that 16% of node-negative breast cancer patients replapse within 5 years after the initial operation (Clin Cancer Res (2000) 6, 3193-3198).
Prediction of the postoperative prognosis of breast cancer patients shows increasing in importance from the standpoint of adjuvant therapeutics currently utilizable. A gene marker which is useful in identifying patients showing a possibility of relapsing after an operation gives a merit which suitable preoperative adjuvant therapeutics can be applied to a high risk patient, and enables prevention of occurrence of unnecessary, complicated and uncomfortable side effects.
Conventionally, postoperative procedures for individual patients are determined depending on tumor diameter and the stage, metastasis to a lymph node, diagnosis by clinicopathological factors, search of a hormone receptor, and the like, however, they are not critical methods (Cancer (1982) 50, 2131-2138, Histopathology (1991) 19, 403-410, Int J Cancer (1996) 69, 135-141, Am J Clin Oncol (1997) 20, 546-551, Eur J Cancer (2002) 38, 1329-1334, Jpn J Cancer Res (2000) 91, 293-300).
Recently, there is a prognosis marker for postoperative breast cancer patients, intending determination of an importance of mutations of genes. These gene mutations include a mutation of p53 (Breast Cancer Res Treat (2001) 69, 65-68), loss of heterozygosity in several alleles (Int J Clin Oncol (2001) 6, 6-12), and abnormal expressions of a BRCA2 gene (Int J Cancer (2002) 198, 879-882), WTl gene (Clin Cancer Res (2002) 8, 1167-1171), HER2/neu gene (Arch Surg (2000) 135, 1469-1474) and Ki-67 gene (J Pathol (1999) 187, 207-216). However, these would not be recognized as effective prognosis predicting means when taking into consideration a fact which a cancer is a disease owing to accumulation of abnormalities of multiple genes.
Further, in these years, genome projects in various organisms are being progressed, and a lot of genes and their base sequences typically including a human gene are being clarified quickly. The function of a gene having a clarified sequence can be checked by various methods. As one of the effective methods, known is a gene expression analysis method utilizing clarified base sequence information. For example, there are developed methods utilizing various nucleic acid-nucleic acid hybridization reactions and various PCR reactions as typified by Northern Hybridization, and relations between various genes and expressions of their organism functions can be checked by these methods. Though the number of applicable genes is limited in these methods, there have been developed a methodology and a novel analysis method called DNA microarray method (DNA chip method) enabling lump expression analysis of multiple genes, for carrying out comprehensive and systemic analysis of extremely many genes such as one individual level, as being clarified recently through genome projects.
As the DNA microarray, a lot of shapes are known such as that in which DNA synthesis is conducted on many discrete cells applying a lithography technology (U.S. Pat. No. 5,445,934), that in which cells composed of grooves or holes are formed on a board and a probe is fixed to the inner wall of the cell (Tokkyo KOKAI (unexamined Japanese patent application) Nos. 11-108928, 2000-78998), a microarray in which a probe is fixed to a gel such as acrylamide and the like for increasing the amount of a probe to be fixed on a chip (U.S. Pat. No. 5,770,721, Tokkyo KOKAI No. 2000-60554), and the like.
Also known is a microarray obtained by fabricating a nucleic acid fixed gel retaining fiber array which retains a nucleic acid fixed gel, and cutting this array along a direction crossing the fiber axis of the array (Tokkyo KOKAI Nos. 2000-270878, 2000-270879).
Recent studies have found that a cDNA microarray technology is effective for identification of a novel gene marker for cancer diagnosis. To date, some researchers have carried out microarray analysis of breast cancer, however, there is no description about data of breast cancer gene expression property capable of predicting the postoperative prognosis of breast cancer (Proc Natl Acad Sci USA (1999) 96, 9212-9217, Nature (2000) 406, 747-752, Proc Natl Acad Sci USA (2001) 98, 11462-11467, Cancer Res (2001) 61, 5979-5984, Cancer Res (2000) 60, 2232-2238, Cancer Res (2001) 61, 5168-5178, Proc Natl Acad Sci USA (2001) 98, 10869-10874). As one exception, it is shown that a specific profile of a lymph node metastasis negative tumor gives a prediction of a short interval before progressing to distant metastasis (N Engl J Med (2002) 347, 1999-2009).

DISCLOSURE OF THE INVENTION

The present invention has an object of providing innovative means for predicting the postoperative prognosis of breast cancer patients from the standpoint of gene expression, based on results obtained by genome-wide and comprehensive analysis on gene expression in breast cancer.
The present inventors have comprehensively analyzed gene expression of a human gene by a DNA microarray and compared gene expression functions of breast cancers in various conditions, thereby, establishing a system for predicting the postoperative prognosis of breast cancer.
That is, the present invention provides the following genes (groups) (1) to (8).
(1) A gene consisting of at least one of the following definitions correlated with prediction of the postoperative prognosis of breast cancer;
1) a marker gene group capable of establishing classification of genes from breast cancer patients died within 5 years after a surgical operation (5y-D group) and genes from patients survived free of disease for several years or more after the operation (5y-S group), depending on their expression functions, in estrogen receptor-negative breast cancer,
2) a marker gene group capable of establishing classification of genes from n0 breast cancer patients recurred within 5 years after an operation (5Y-R group) and genes from patients survived free of disease for 5 years or more after the operation (5Y-F group), depending on their expression functions, in (node-negative) (n0) breast cancer with no metastasis to a lymph node in the operation,
3) a marker gene group capable of establishing classification of genes from breast cancer patients died within 5 years after a surgical operation (5D group) and genes from patients survived free of disease for several years or more after the operation (5S group), depending on their expression functions, in primary breast cancer.
(2) A gene selected from the following sequences correlated with prediction of the postoperative prognosis of primary breast cancer;

pro-alpha-1 type 3 collagen (PIIIP),
complement component Clr,
dihydropyrimidinase-like 3 (DPYSL3),
protein tyrosine kinase 9-like (PTK9L),
carboxypeptidase E (CPE),
alpha-tubulin,
beta-tubulin,
heat shock protein HSP 90-alpha gene,
malate dehydrogenase,
NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 3 (NDUFB3).

(3) A gene selected from the following sequences highly expressed in a group of good prognosis correlated with prediction of the postoperative prognosis of primary breast cancer;

pro-alpha-1 type 3 collagen (PIIIP),
complement component Clr,
dihydropyrimidinase-like 3 (DPYSL3),
protein tyrosine kinase 9-like (PTK9L),
carboxypeptidase E (CPE),
alpha-tubulin,
beta-tubulin.

(4) A gene selected from the following sequences highly expressed in a group of bad prognosis correlated with prediction of the postoperative prognosis of primary breast cancer; heat shock protein HSP 90-alpha gene,

malate dehydrogenase,
NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 3 (NDUFB3).

(5) A gene selected from the following sequences correlated with prediction of the postoperative prognosis, in (node-negative)(n0) breast cancer with no metastasis to a lymph node in operation;

AF058701/DNA polymerase zeta catalytic subunit (REV3),
AI066764/lectin, galactoside-binding, souble, 1 (galectin 1),
x15940/ribosomal protein L31,
Hs.94653/neurochondrin (KIAA0607),
M13436/ovarian beta-A-inhibin,
Hs.5002/copper chaperone for superoxide dismutase; CCS,
D67025/proteasome (prosome, macropain) 26S subunit, non-ATPase, 3,
M80469/MHC class I HLA-J gene,
Hs.4864/ESTs,
Hs.106326/ESTs.

(6) A gene selected from the following sequences highly expressed in a group of bad prognosis correlated with prediction of the postoperative prognosis, in (node-negative)(n0) breast cancer with no metastasis to a lymph node in operation;

AF058701/DNA polymerase zeta catalytic subunit (REV3),
AI066764/lectin, galactoside-binding, soluble, 1 (galectin 1),
x15940/ribosomal protein L31.

(7) A gene selected from the following sequences highly expressed in a group of good prognosis correlated with prediction of the postoperative prognosis, in (node-negative)(n0) breast cancer with no metastasis to a lymph node in operation;

Hs.94653/neurochondrin (KIAA0607),
M13436/ovarian beta-A-inhibin,
Hs.5002/copper chaperone for superoxide dismutase; CCS,
D67025/proteasome (prosome, macropain) 26S subunit, non-ATPase, 3,
M80469/MHC class I HLA-J gene,
Hs.4864/ESTs,
Hs.106326/ESTs.

(8) A gene selected from the following sequences correlated with prediction of the postoperative prognosis, in estrogen receptor-negative breast cancer;

Hs.108504/FLJ20113/ubiquitin-specific protease otubain I
Hs.146550/MYH9/myosin, heavy polypeptide 9, non-muscle
Hs.194691/RAI3/retinoic acid induced 3
Hs.1975/TDRD3/tudor domain containing 3
Hs.203952/TRRAP/transformation/transcription domain-associated protein
Hs.278607/GSA7/ubiquitin activating enzyme E1-like protein
Hs.429/ATP5G3/ATP synthase, H+ transporting, mitochondrialF0complex, subunitc (subunit9)
isoform3
Hs.75305/AIP/aryl hydrocarbon receptor interacting protein
Hs.81170/PIMI/ pim-1 oncogene
Hs.99987/ERCC2/

excision repair cross-complementing rodent repair deficiency, complementation group 2

Y12781/Transducin (beta) like 1 protein
Hs.104417/KIAA 1205 protein
cl.21783/Hypothetical protein
Hs.112628/Hypothetical protein: MGC43581
Hs. 170345/Hypothetical protein FLJ13710
Hs.53996/weakly similar to zinc finger protein 135
Hs.55422/Hypothetical protein
Hs.112718/EST
Hs.115880/EST
Hs.126495/EST

The present invention also provides a gene selected from the above-mentioned (8), as a gene highly expressed in a group of bad prognosis.
Further, the present invention provides a DNA microarray carrying thereon the gene according to any one of the above-mentioned (1) to (8) and/or a probe specific to the gene, and preferably, the DNA microarray is a fiber type microarray.
The above-mentioned gene and/or probe specific to the gene can be used as a marker in a method of inspecting the postoperative prognosis of breast cancer. Further, it can be also used as a marker for cancer therapeutic medicines for controlling the postoperative prognosis of breast cancer. The above-mentioned microarray can be used in a method of inspecting the postoperative prognosis of breast cancer.
Further, the present invention provides a method of screening cancer therapeutic medicines for controlling the postoperative prognosis of breast cancer using as a marker the above-mentioned gene and/or probe specific to the gene. The above-mentioned microarray can be used in the above-mentioned screening method.
The marker can be included as a reagent, and can be used as a diagnosis kit for the postoperative prognosis of breast cancer. The reagent kit includes a DNA microarray carrying thereon a marker, preferably, a fiber type microarray.
According to the means of the present invention, completely novel breast cancer correlated genes have been found and simultaneously, it has been found that these genes are correlated deeply with malignant degeneration of breast cancer and finally, exert an influence on the prognosis of breast cancer patients. Further, by establishing a mathematical formula for evaluating expression condition of the found gene, a completely novel and effective breast cancer postoperative prognosis predicting system has been developed. The system of the present invention from the standpoint of gene expression is believed to be an innovative prognosis predicting system arresting the biological essentiality of a cancer, utterly different from conventional prognosis evaluation methods, when taking into consideration a fact which a cancer is a disease owing to abnormality of a gene.

BRIEF DESCRIPTION OF DRAWINGS

FIG. 1 is a photograph showing a gene group (A) manifesting increase and a gene group (B) manifesting decrease in expression in 5y-D group as compared with 5y-S group.

FIG. 2 is a photograph showing analysis results of semi-quantitative RT-PCR of RNAs derived from 5y-S group and 5y-D group.

FIG. 3 shows prognosis scores in individual patients.

FIG. 4 is a photograph showing analysis results of semi-quantitative RT-PCR of RNAs derived from 5Y-F group and 5Y-R group.

FIG. 5 is a photograph showing analysis results of semi-quantitative RT-PCR of RNAs derived from 5Y-F group and 5Y-R group.

FIG. 6 shows prognosis scores in individual patients.

FIG. 7 is a photograph showing analysis results of semi-quantitative PCR of 7 genes highly expressed in 5S tumor.

M: marker ladder

S1-S10: newly inspected tissues of patients survived free of disease for 5 years or more after operation.

D1-D10: newly inspected cases of patients died of breast cancer within 5 years after operation.

Difference in expression strength was evaluated by Student's t-test; when p value is 0.05 or less, statistical significance is believed to be present.

FIG. 8 is a photograph showing analysis results of semi-quantitative PCR of 3 genes highly expressed in 5D group. For explanation of marks, please see explanation in FIG. 7.

FIG. 9 shows results illustrating prognosis indices (PI) of newly inspected 20 cases. The indices of all 10 patients survived free of disease for 5 years or more were higher than 7. On the other hand, the indices of patients died of breast cancer within 5 years after operation were lower than 7. Distribution of two groups is statistically significant (p=0.0002).

BEST MODE FOR CARRYING OUT THE INVENTION

The marker gene group correlated with prediction of the postoperative prognosis of breast cancer as one aspect of the present invention is obtained by analysis by cDNA microarray of the expression functions of genes from patients manifesting death or recurring within 5 years after a surgical operation and patients survived for 5 years or more after the operation, in estrogen receptor-negative breast cancer, node-negative breast cancer and primary breast cancer.
Specifically, one aspect of the present invention is a gene consisting of at least one of the following definitions selected from known sequences correlated with prediction of the postoperative prognosis of breast cancer;
1) a marker gene group capable of establishing classification of genes from breast cancer patients died within 5 years after a surgical operation (5y-D group) and genes from patients survived free of disease for several years or more after the operation (5y-S group), depending on their expression functions, in estrogen receptor-negative breast cancer,
2) a marker gene group capable of establishing classification of genes from n0 breast cancer patients recurred within 5 years after an operation (5Y-R group) and genes from patients survived free of disease for 5 years or more after the operation (5Y-F group), depending on their expression functions, in (node-negative)(n0) breast cancer with no metastasis to a lymph node in the operation,
3) a marker gene group capable of establishing classification of genes from breast cancer patients died within 5 years after a surgical operation (5D group) and genes from patients survived free of disease for several years or more after the operation (5S group), depending on their expression functions, in primary breast cancer.
The gene correlated with prediction of the postoperative prognosis of breast cancer of the present invention is obtained by evaluating the data of a cDNA microarray using a Random-permutation test and a Mann-Whitney test. The present invention presents an approach more useful at clinical level, by evaluating gene expression functions by a combination of a cDNA microarray and a semi-quantitative PCR experiment.
In the present invention, a gene correlated with prediction of the postoperative prognosis of primary breast cancer has been identified by evaluating gene expression functions in breast cancer patients.
Specifically, one aspect of the present invention is a gene selected from the following sequences selected from known sequences correlated with prediction of the postoperative

prognosis of primary breast cancer;
pro-alpha-1 type 3 collagen (PIIIP),
complement component Clr,
dihydropyrimidinase-like 3 (DPYSL3),
protein tyrosine kinase 9-like (PTK9L),
carboxypeptidase E (CPE),
alpha-tubulin,
beta-tubulin,
heat shock protein HSP 90-alpha gene,
malate dehydrogenase,
NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 3 (NDUFB3).

In the present invention, “high expression” means that the expression level of a subject gene is, when compared with the average value of the expression levelss of the same gene in a parent population, higher than the average value, for example, 2-fold or more of the average value.
In the present invention, “low expression” means that the expression level of a subject gene is, when compared with the average value of the expression levels of the same gene in a parent population, lower than the average value, for example, 2-fold or less of the average value.
Some of the above-mentioned genes are believed to be correlated with proliferation or distant metastasis of tumor cells, and for example, a heat shock protein HSP 90-alpha is a chaperone for a lot of kinases, and has a possibility of promoting growth of cancer cells (Neckers, L (2002) Trends Mol Med 8, S55-61). Malate dehydrogenase is an important enzyme correlated with energy accompanying aerobic or anaerobic metabolism, and the activity of malate dehydrogenase is correlated with a tumor marker for squamous cell carcinoma (Ross, C. D., et al. (2000) Otolaryngol Head Neck Surg 122, 195-200). NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 3 (NDUFB3) belongs to an mitochondorial electron transport chain , and chromosome abnormality in a region containing NDUFB3 is remarkable in a breast cancer cell line MDA-MB-231 (xie, D., et al. (2002) Int J Oncol 21, 499-507).
The above-mentioned 10 genes correlated with prediction of the postoperative prognosis of primary breast cancer show different expressions in a group of good prognosis (5S group) and a group of bad prognosis (5Y group), and 7 genes among the 10 genes are genes highly expressed in a group of good prognosis (5S group).
Namely, one aspect of the present invention is a gene selected from the following sequences highly expressed in a group of good prognosis selected from known sequences correlated with prediction of the postoperative prognosis of primary breast cancer;

3 genes among the 10 genes correlated with prediction of the postoperative prognosis of primary breast cancer are genes highly expressed in a group of bad prognosis (5Y group). Namely, one aspect of the present invention is a gene selected from the following sequences highly expressed in a group of bad prognosis selected from known sequences correlated with prediction of the postoperative prognosis of primary breast cancer;

heat shock protein HSP 90-alpha gene,
malate dehydrogenase,
NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 3 (NDUFB3).

Here, the prediction index (PI) for primary breast cancer is defined as described below and can be used for prediction of the postoperative prognosis of breast cancer.
Prediction index (PI)=(total of normalized expression ratios of the above-mentioned 7 genes highly expressed in a group of good prognosis in breast cancer tissue)−(total of normalized expression ratios of the above-mentioned 3 genes highly expressed in a group of bad prognosis in breast cancer tissue)
In the present invention, gene expression functions in breast cancer patients have been evaluated and 10 genes correlated with prediction of the postoperative prognosis of node-negative breast cancer have been identified.
Specifically, one aspect of the present invention is a gene selected from the following sequences selected from known sequences correlated with prediction of the postoperative prognosis, in (node-negative) (n0) breast cancer with no metastasis to a lymph node in operation;

AF058701/DNA polymerase zeta catalytic subunit (REV3),
AI066764/lectin, galactoside-binding, soluble, 1 (galectin 1),
x15940/ribosomal protein L31,
Hs.94653/neurochondrin (KIAA0607),
M13436/ovarian beta-A-inhibin,
Hs.5002/copper chaperone for superoxide dismutase; CCS,
D67025/proteasome (prosome, macropain) 26S subunit, non-ATPase, 3,
M80469/MHC class I HLA-J gene,
Hs.4864/ESTs,
Hs.106326/ESTs.

The above-mentioned genes correlated with prediction of the postoperative prognosis of node-negative breast cancer include genes correlated with proliferation and distant metastasis of tumor cells. For example, galectin 1 is an autocrine type cancer repressor for regulating cell differentiation (AxelH, et al. (2003) Int. J. Cancer, 103: 370-379). Further, a gene activating cancer metastasis is included.
The above-mentioned 10 genes correlated with prediction of the postoperative prognosis of node-negative breast cancer show different expressions in a group of good prognosis (5Y-F group) and a group of bad prognosis (5Y-R group), and 3 genes among the 10 genes are genes highly expressed in a group of bad prognosis (5Y-R group). Namely, one aspect of the present invention is a gene selected from the following sequences highly expressed in a group of bad prognosis selected from known sequences correlated with prediction of the postoperative prognosis, in node-negative breast cancer in operation;

7 genes among the 10 genes correlated with prediction of the postoperative prognosis of node-negative breast cancer are genes highly expressed in a group of good prognosis (5Y-F group). Namely, one aspect of the present invention is a gene selected from the following sequences highly expressed in a group of good prognosis selected from known sequences correlated with prediction of the postoperative prognosis, in node-negative breast cancer;

Here, the prognosis score (PS) for node-negative breast cancer is defined as described below and can be used for prediction of the postoperative prognosis of breast cancer.
Prognosis score (PS)=(total of normalized expression ratios of the above-mentioned 3 genes highly expressed in a group of bad prognosis in breast cancer tissue)−(total of normalized expression ratios of the above-mentioned 7 genes highly expressed in a group of good prognosis in breast cancer tissue).
In the present invention, 20 genes correlated with prediction of the postoperative prognosis of estrogen receptor-negative breast cancer have been identified, by evaluating gene expression functions in breast cancer patients.
Specifically, one aspect of the present invention is a gene selected from the following sequences selected from known sequences correlated with prediction of the postoperative prognosis, in estrogen receptor-negative breast cancer;

Hs.108504/FLJ20113/ubiquitin-specific protease otubain 1
Hs.146550/MYH9/myosin, heavy polypeptide 9, non-muscle
Hs.194691/RAI3/retinoic acid induced 3
Hs.1975/TDRD3/tudor domain containing 3
Hs.203952/TRRAP/transformation/transcription domain-associated protein
Hs.278607/GSA7/ubiquitin activating enzyme E1-like protein
Hs.429/ATP5G3/ATP synthase, H+ transporting,
mitochondrialF0complex, subunitc (subunit9) isoform3
Hs.75305/AIP/aryl hydrocarbon receptor interacting protein
Hs.81170/PIM1/pim-1 oncogene
Hs.99987/ERCC2/

Y12781/Transducin (beta) like 1 protein
Hs.104417/KIAA1205 protein
cl.21783/Hypothetical protein
Hs.112628/Hypothetical protein: MGC43581
Hs.170345/Hypothetical protein FLJ13710
Hs.53996/weakly similar to zinc finger protein 135
Hs.55422/Hypothetical protein
Hs.112718/EST
Hs.115880/EST
Hs. 126495/EST

The above-mentioned genes correlated with prediction of the postoperative prognosis of estrogen receptor-negative breast cancer include genes correlated with proliferation and distant metastasis of tumor cells. For example, PIM1 is serine/threonine kinase, and there is a correlation between clinical results of prostate cancer and the expression (Oesterreich, S., et al. (1996) Clin Cancer Res, 2, 1199-1206). TRRAP protein is a subunit of a mammal HTA complex, and antisense RNA against TRRAP inhibits estrogen-dependent growth of breast cancer cells.
The above-mentioned 20 genes correlated with prediction of the postoperative prognosis of estrogen receptor-negative breast cancer show high expression in a group of bad prognosis (5y-D group). Namely, one aspect of the present invention is a gene selected from known sequences correlated with prediction of the postoperative prognosis, in the above-mentioned estrogen receptor-negative breast cancer highly expressed in a group of bad prognosis.
Here, postoperative prognosis of breast cancer can be predicted as described below, based on the expression of the above-mentioned gene correlated with prediction of the postoperative prognosis of estrogen receptor-negative breast cancer;
(1) when the expression levels in breast cancer tissue of the above-mentioned 20 genes correlated with prediction of the postoperative prognosis of estrogen receptor-negative breast cancer are compared with the average value in a parent population, and if the expression level of each gene is 2-fold or more of the average value in a parent population, one point is imparted,
(2) when the procedure of (1) is carried out on 20 genes, and if the total point is 8 points or more, prognosis is decided to be bad.
The above-mentioned gene correlated with prediction of the postoperative prognosis of breast cancer can be used as a marker for inspection of breast cancer postoperative prognosis. Namely, one aspect of the present invention is a method of inspecting the postoperative prognosis of breast cancer using the above-mentioned gene as a marker.
The above-mentioned gene correlated with prediction of the postoperative prognosis of breast cancer can be used as a marker for screening of cancer therapeutic medicines for controlling the postoperative prognosis of breast cancer. Namely, one aspect of the present invention is a method of screening cancer therapeutic medicines for controlling the postoperative prognosis of breast cancer using the above-mentioned gene as a marker.
The above-mentioned gene correlated with prediction of the postoperative prognosis of breast cancer can be used as a marker for diagnosis of the postoperative prognosis of breast cancer. It is also possible to design probes specific to the above-mentioned gene and to use these probes as a marker. These probes can be designed, for example, by Probe Quest (registered trademark) manufactured by Dyna Com. Namely, one aspect of the present invention is a diagnosis kit for the postoperative prognosis of breast cancer containing a reagent using the above-mentioned gene as a marker.
The above-mentioned diagnosis kit can include a microarray. Namely, one aspect of the present invention is the diagnosis kit, wherein the diagnosis kit includes a microarray.
The microarray of the above-mentioned diagnosis kit including a microarray includes a fiber type microarray. Here, for a method of preparing a fiber type microarray, the above-mentioned patent documents 6 to 7 are cited. Namely, one aspect of the present invention is the above-mentioned diagnosis kit wherein the microarray is a fiber type microarray.
Next, aspects of the present invention will be specifically illustrated by examples, but the present invention is not limited to these examples.

EXAMPLE 1

Evaluation of Gene Expression Function for Prediction of the Postoperative Prognosis in Estrogen Receptor-Negative Breast Cancer

(Tissue Sample)

An informed consent was obtained according to a guide line accepted by an ethics committee of Cancer Society and by Nippon Medical School, then, primary breast cancer and tissue from adjacent normal mammary gland were collected from breast cancer patients who undergone an operation in 1995 to 1997 in Cancer Society attached hospital (Tokyo). The tissue was quickly frozen and preserved at −80° C. For 954 patients, all members were clinically traced during a period of 5 years or more or until death, and samples were selected from 10 estrogen receptor-negative breast cancer patients died within 5 years after the operation (5y-D) and 10 patients survived free of disease for 5 years or more after the operation (5y-S). The backgrounds of both the patient groups were allowed to coincide in age, lymph node metastasis, tumor diameter and tissue type (Table 1).

(Clinical Feature of 20 Cases of Breast Cancer)

	TABLE 1

	TNM classification^b

group	Case No.	ER condition	Age	Sex	Process^a	Tumor	Lymph node	TTD	^c

5y-D	3281	Negative	34	Female	a2	T2	N1b		9
	3459	Negative	64	Female	a2	T4	N3		6
	3550	Negative	73	Female	a2	T4	N1b		12
	3892	Negative	62	Female	a2	T2	N1a	21
	3948	Negative	60	Female	a2	T2	N1a	51
	4020	Negative	50	Female	a2	T2	N3	28
	3654	Negative	46	Female	a2	T4	N1b		19
	4118	Negative	53	Female	a2	T1	N1a	21
	4462	Negative	34	Female	a1	T2	N1a		24
	4126	Negative	51	Female	b5	T4	N3		6
5y-S	3656	Negative	31	Female	a2	T2	N1a	>60
	3197	Negative	42	Female	a1	T1	N1a	>60
	3662	Negative	58	Female	a2	T2	N0	>60
	3241	Negative	47	Female	a2	T2	N1a	>60
	3267	Negative	51	Female	a2	T2	N1a	>60
	3329	Negative	60	Female	a2	T2	N1a	>60
	3345	Negative	43	Female	a1	T2	N2	>60
	3556	Negative	59	Female	a2	T3	N0	>60
	3558	Negative	57	Female	a2	T3	N1b	>60
	3658	Negative	42	Female	a1	T2	N1a	>60

^aa1: invasive papillotubular carcinoma. a2: invasive solid-tubular carcinoma. b5: squamous cell carcinoma.
^bTNM classification: clinical classification by Japan Breast Cancer Society
^cTTD: time to death after surgery (months)

All patients underwent postoperative adjuvant therapy according to “Postoperative clinical protocol for breast cancer (nyugan no tameno shujutsugo no rinsho no purotokoru)” of Cancer Society attached hospital. In each case, selection of adjuvant therapy was determined strictly based on surgical operation type, lymph node involvement condition, and presence of local or distant metastasis. In the study of the present invention, all patients did not have distant metastasis before the adjuvant chemical therapy and did not undergo radiation therapy or chemical therapy before the surgical operation.

(Clinicopathological Parameter)

The following parameters were checked: tissue type, tumor diameter and invasion (t factor), lymph node involvement, and conditions of estrogen receptor (ER) and progesterone receptor (PgR). Tumors were classified into the following types according to TNM classification and to tissue classification of Japan Breast Cancer Society (1989); noninvasivetubular (1a), invasivepapillotubular (a1), invasive solid-tubular (a2), invasivescirrhouscarcinoma (a3), and other special types (b). The classification is basically the same as breast cancer tissue classification of WHO. t factors were classified into the following types according to histologial TNM classification; tumor with a maximum size of 2 cm or less (t1), tumor with no invasion into skin or pectoral muscle and with a maximum size of 2 cm or more (t2), and tumor with invasion into skin or pectoral muscle (t3).

(Design and Construction of cDNA Microarray)

From 25344 cDNAs selected from UniGene database, “genome wide cDNA microarray” was constructed. The cDNAs were made by RT-PCR using poly(A)+RNAs separated from various human organs. The PCR products were spotted on slide glasses of type 7 (Amersham Biosciences UK Limited, Buckinghamshire, UK) using Array Spotter Generation III (Amersham Biosciences). Each slide contains 384 house-keeping genes.

(Preparation and Proliferation of RNA)

A tumor raw material was quickly frozen at −80° C. immediately after collection. RNA was extracted using TRIzol (Invitrogen Inc., Carlsbad, Calif., USA), further, purified using RNeasykits (Quiagen Inc., Valencia, Calif.). The purity of each RNA was evaluated by a spectrophotometry and electrophoresis on 1.2% modified formamide gel. The high purity RNA was defined as a sample having an absorbance ratio (260 nm/280 nm) of 1.8 to 2.0 and in which 28S/18S liposomal bands show a ratio of 1.8 or more on formamide gel electrophoresis. After treating with 1 unit of DNaseI (Epicentre Technoloies, Madison, Wis.) (1 unit/μl), RNA amplification by T7RNA polymerase was carried out using 2 μg of RNA from each sample as a starting raw material. Amplification was carried out twice, and the amplified RNA (aRNA) was purified by RNeasykits (Quiagen Inc., Valencia, Calif.). The amount of each aRNA was measured by a spectrophotometer, and the quality was checked by formamide gel electrophoresis.

(Labeling of aRNA, Hybridization and Scanning)

cDNA for microarray analysis was prepared from aRNA. aRNAs (5 to 10 μg) from breast cancer and normal mammary gland tissue were labeled with Cy5 (cancer sample) and Cy3 (normal sample) using aminoallyl-cDNA labelingkits (Ambion, Austin, Tex.). The Cy3-labeled cDNA probe and the Cy5-labeled cDNA probe were mixed and heated at 95° C. for 5 minutes, then, quenched with ice for 30 seconds, and hybridized on a microarray. The mixed probes were added to formamide (Sigma-Aldrich Corp., St. Louis, Mo., USA) having a 50% final concentration of microarrayhybridization solution version 2 (Amersham Biosciences UK Limited, Buckinghamshire, UK). After hybridization at 40° C. for 15 hours, the microarray slides were washed first with 1×SSC and 0.2% SDS at 55° C. for 10 minutes, then, washed twice with 0.1×SSC/0.2% SDS each for 1 minute at room temperature. All treatments were carried out by Automated Slide Processor System (Amersham). The signal strength of each hybridization was scanned by Gene Pix 4000A (Axon Instruments, Inc., Foster City, Calif., USA), and evaluated by Gene Pix. 3.0 (Axon Instruments) by a spectrophotometry. The scanned signals were normalized by a method described in the following literature (the total gene normalization method) (Yang Y H, Dudoit S, Luu P, et al. (2002) Nucleic Acids Res 30, e15; Manos E J, Jones D A. (2001) Cancer Res 61: 433-348).

(Signal Analysis and Selection of Genes Showing Different Expressions)

The signal strength of each hybridization was evaluated by a photometry by Gene Pix 3.0 (Axon Instruments, Inc., Foster City, Calif., USA). For normalizing mRNA expression levels between cancer and control, the Cy5:Cy3 ratio in each gene expression was adjusted. As a result, the averaged Log (Cy5:Cy3 ratio) of the house keeping genes was zero. 27 house keeping genes were adopted from a house-keeping panel in Web site http://www.nhgri.nih.gov/DIR/LCG/ARRAY/expn.html. For each microarray slide, the cut off value of (S/N) ratio was set at 3.0. Genes with signal strengths of Cy3 and Cy5 lower than the cut off value were excluded out of the investigation.

(Mann-Whitney Test)

For investigating genes showing apparently different expressions between 5y-D tumor and 5y-S tumor, Mann-Whitney test was applied to a series of samples X. X represents Cy5/Cy3 signal strength ratio of each gene and each sample (OnoK, Tanaka T, Tsunoda T, et al. (2000) Cancer Res 2000; 60: 5007-5011). The U value was calculated for genes imparting significant signals in at least 5 samples in both groups. Genes showing U values of lower than 23 or larger than 77 were selected. Since the U value is obtained by calculation for 5y-S group based on 5y-D group in each gene based on each X value, U values lower than 23 were evaluated to manifest higher expression in 5y-S group than in 5y-D group. However, genes with U values higher than 77 were evaluated to manifest higher expression in 5y-D group than in 5y-S group. Base on this criterion, 183 genes were highly expressed in 5y-S group and 31 genes were highly expressed in 5y-S group. Thus, only genes in which intermediate expression values show a difference of 2-fold or more between two groups (μXD/μXS≦0.5 or ≧2.0, μXD and μXS represent average X values in 5y-D and 5y-S group, respectively) were defined as genes correlated with prognosis. As a result, 110 genes in total were selected. Of them, 90 genes were expressed at higher level in 5y-D tumor group and 20 genes were expressed at higher level in 5y-S tumor group.

(Random-Permutation Test)

Further, for evaluating values of 110 genes selected by the Mann-Whitney test, a permutation test was carried out. A possibility, Ps of a gene for correlating with a group difference was also assumed. When each gene is represented by an expression vector v(g)=(X1, X2, - - - , X20) (Xi) shows a gene expression level of i-th sample in the first ample group), an ideal expression pattern is expressed by c=(c1, c2, - - - , c20) (ci=+1 or 0, depending on whether i-th sample belongs to S group or D group).
Correlation between a gene and a group difference Pgc was defined as described below. That is, Pgc=(μ_S−μ_D)/(δ_S+δ_D), μ_S(μ_D) and δ_S(δ_D) show standard deviation of log₂X of the gene “g” of each sample in a newly defined S (or D) group.
The permutation test was carried out while substituting the coordinate of c. The correlation values, Pgc were calculated between all permutations. These procedures were repeated for 10000 times. Accidentally, the p value showing a possibility of a gene for classifying two groups was evaluated for each of 110 genes selected. Finally, 71 gene highly expressed in 5y-D case and 15 gene expressed low in 5y-S case were selected.

(Semi-quantitative RT-PCR)

RNA (2 μg) was treated with DNase I (Epicentre Technologies, Madison, Wis., USA), and single-stranded cDNAs were subjected to reverse transcription using Reverscript IIreversetranscriptase (manufactured by Wako Pure Chemical Industries, Ltd., Osaka, Japan) and oligo (dT) 12-18 primer. Single-stranded cDNAs were adjusted in the concentration for the subsequent PCR amplification by monitoring expression of GAPD (glyceraldehyde-3-phosphatedehydrogenase) as a quantitative control. Each PCR was carried out under the following reaction conditions using Gene Amp PCR system 9700 (Applied Biosystems, Foster City, Calif., USA) at an amount of 1×PCR buffer of 30 μl.
94° C. 5 minutes,
(94° C. 30 seconds, 60° C. 30 seconds, and 72° C. 30 seconds) for 25 to 35 cycles.
Primer sequences used in RT-PCR are as described below:

	SEQ ID No. 1 GAPD (control) forward,
	5′-GGA AGGTGA AGG TCG GAG T-3′

	SEQ ID No. 2 reverse,
	5′-TGG GTG GAA TCA TAT TGGAA-3′;

	SEQ ID No. 3 Hs.108504F,
	5′-ACA CTT CAT CTG CTCCCT CAT AG-3′;

	SEQ ID No. 4 Hs.108504R,
	5′CTG CCT AGA CCT GAGGAC TGT AG-3′;

	SEQ ID No. 5 Hs.146550F,
	5′ACT GAG GCC TTT TGGTAG TCG-3′;

	SEQ ID No. 6 Hs.146550R,
	5′TCT CTT TAT TGT GATGCT CAG TGG-3′;

	SEQ ID No. 7 Hs.76607F,
	5′AAA TCC TTC TCG TGT GTTGAC TG-3′;

	SEQ ID No. 8 Hs.76607R,
	5′CAG TCA TGA GGG CTA AAAACT GA-3′;

	SEQ ID No. 9 Hs.1975F,
	5′GAA GAC AAC AAG TTT TAC CGG G-3′;

	SEQ ID No. 10 Hs.1975R,
	5′ATG GTT TTA TTG ACG GCAGAA G-3′;

	SEQ ID No. 11 Hs.203952F,
	5′AGG ACA CGT CCT CTCCTC TCT C-3′;

	SEQ ID No. 12 Hs.203952R,
	5′TAA AGC TAG CGA AGGAAC GTA CA-3′;

	SEQ ID No. 13 Hs.278607F,
	5′TCC CTT CTG TTT CCT CAG TGT T-3′;

	SEQ ID No. 14 Hs.278607R,
	5′CCT GCC CCG ATA AAA ATA TCT AC-3′;

	SEQ ID No. 15 Hs.429F,
	5′TTG ACC TTA AGC CTC TTTTCC TC-3′;

	SEQ ID No. 16 Hs.429R,
	5′ATA ACG TAC ATT CCC ATGACA CC-3′;

	SEQ ID No. 17 Hs.75305F,
	5′ACT TTC AAG ATG GGACCA AGG-3′;

	SEQ ID No. 18 Hs.75305R,
	5′ATA TAC ACA GAA GCATGA CGC AG-3′;

	SEQ ID No. 19 Hs.81170F,
	5′TTG CTG GAC TCT GAAATA TCC C-3′;

	SEQ ID No. 20 Hs.81170R,
	5′TTC CCC TGT ACA GTATTT CAC TCA-3′;

	SEQ ID No. 21 Hs.99987F,
	5′CTG AGC AAT CTG CTCTAT CCT CT-3′;

	SEQ ID No. 22 Hs.99987R,
	5′GTT CCA GAT TCG TGAGAA TGA CT-3′;

	SEQ ID No. 23 Y12781F,
	5′ACC AGT AAC AAC TGT GGGATG G-3′;

	SEQ ID No. 24 Y12781R,
	5′CAA ATG AGC TAC AAC ACACAA GG-3′;

	SEQ ID No. 25 Hs.104417F,
	5′CCC CCT CCA CCT TGTACA TAA T-3′;

	SEQ ID No. 26 Hs.104417R,
	5′GTT TTC GTT TGG CTGGTT GTG-3′;

	SEQ ID No. 27 cl.21783F,
	5′GTC TGA GAT TTT ACTGCA CCG-3′;

	SEQ ID No. 28 cl.21783R,
	5′GGA TGG AGC TGG AGGATA TTA-3′;

	SEQ ID No. 29 Hs.112628F,
	5′ATT GCT AAG GAT AAGTGC TGC TC-3′;

	SEQ ID No. 30 Hs.112628R,
	5′TGT CAG TAT AGA AGCCTG TGG GT-3′;

	SEQ ID No. 31 Hs.170345F,
	5′TTC TTA GGC CAT CCCTTT TCT AC-3′;

	SEQ ID No. 32 Hs.170345R,
	5′GCA TCT GAA TGT CTTTCT CCC TA-3′;

	SEQ ID No. 33 Hs.53996F,
	5′CCA TAG GAT CTT GACTCC AAC AG-3′;

	SEQ ID No. 34 Hs.53996R,
	5′ACT GGG AGT GGA GGAAAT TAG AG-3′;

	SEQ ID No. 35 Hs.55422F,
	5′CTA ATG TAA GCT CCATTG GGA TG-3′;

	SEQ ID No. 36 Hs.55422R,
	5′CAA ACT GCA AAC TAGCTC CCT AA-3′;

	SEQ ID No. 37 Hs.112718F,
	5′AAG ACT AAG AGG GAA AAT GTG GG-3′;

	SEQ ID No. 38 Hs.112718R,
	5′AGG TAA CCC AAA GTG ACA AAC CT-3′;

	SEQ ID No. 39 Hs.115880F,
	5′TTA AGT GAG TCT CCT TGG CTG AG-3′;

	SEQ ID No. 40 Hs.115880R,
	5′AGG GCC CCT ATA TCC AAT ACC TA-3′;

	SEQ ID No. 41 Hs.126495F,
	5′GAT CTT TCA AGA TGAGCC AAG GT-3′;

	SEQ ID No. 42 Hs.126495R,
	5′AGT CAT TCA GAA GCCATT GAG AC-3′

(Measurement of Signal Strength of RT-PCR Product and Calculation of Prognosis Score)

A PCR product was detected by 2% agarose gel electrophoresis and ethidium bromide staining. A gel was scanned by a digital image processing system (Alphalmager 3300; Alpha Innotech, San Leandro, Calif., USA) according to the Spot Density method. A two-dimensional region of each band was constructed, and pixel strength (gene expression) was obtained in which the density was defined as IDV (Integrated Density Value). Importance in a difference in IDV in each group was evaluated by the Student's t-test. As a result, 20 genes showing p values of 0.05 or lower in the t-test were selected as a candidate (Table 2). That is, expression levels of the 20 genes were significantly higher in the 5y-D group than in the 5y-S group. Base on this information, the present inventors have tried to establish a scoring system for predicting the postoperative prognosis. In this procedure, each gene was determined depending on whether the expression level of each sample was higher than the average expression level of 20 samples or not. When the expression level of a sample was 2-fold or more than the average, +1 point was imparted additionally. Next, points of all of the 20 genes were summed up for obtaining the total vote (prognosis score) for each sample. As a result, a case of a sample of 8 points or more was evaluated as an indication of bad prognosis. On the other hand, a case of a sample of 8 points or less was evaluated as an indication of preferable prognosis.

(20 Candidate Genes of Prognosis Scoring System)

TABLE 2

Hs./Accesion No.	kind

Hs.108504	FLJ20113: ublquitin-specific protease otubain 1
Hs.146550	MYH9: myosin, heavy polypeptide 9, non-muscle
Hs.194691	RAI3: retinoic acid induced 3
Hs.1975	TDRD3: tudor domain containing 3
Hs.203952	TRRAP: transformation/transcription
	domain-associated protein
Hs.278607	GSA7: ubiquitin activating enzyme E1-like protein
Hs.429	ATP5G3: ATP synthase, H+ transporting,
	mitochondrial F0 complex, subunit c (subunit 9)
	isoform 3
Hs.75305	AIP: aryl hydrocarbon receptor interacting protein
Hs.81170	PIM1: pim-1 oncogene
Hs.99987	ERCC2: excision repair cross-complementing rodent
	repair deficiency, complementation group 2
Y12781	Transducin(beta)like 1 protein
Hs.104417	KIAA1205 protein
cl.21783	Hypothetical protein
Hs.112628	Hypothetical protein: MGC43581
Hs.170345	Hypothetical protein FLJ13710
Hs.53996	weakly similar to zinc finger protein 135
Hs.55422	Hypothetical protein
Hs.112718	EST
Hs.115880	EST
Hs.126495	EST

(Result)

257 genes highly expressed significantly in estrogen receptor-negative breast cancer tissue were clarified, and 378 genes expressed low were clarified likewise. For identifying genes showing different expressions between the 5y-D group and the 5y-S group, the data of a microarray was analyzed by the Mann-Whitney test and the Random-permutation test. As a result, 71 genes in total (including 10 EST and 9 genes encoding virtual protein) in 5y-D tumor were classified in common into a group of higher expression. In contrast, 15 genes (including 3 EST) were classified in common into a group of lower expression (FIG. 1).
Genes highly expressed in the 5y-D group include the following genes correlated with proliferation and metastasis of cancer cells; matrix metalloproteinase 2 (MMP2), heat shock protein 27 HSPB1), Pim-1 oncogene (PIM1) and transformation/transcription domain-associated protein (TRRAP).
Genes expressed low in the 5y-D group include genes of HLA-C (major histocompatibility complex, class I, C) and specific kinase. A lot of genes having correlations with DNA repair, transcription, signal transduction, cytoskeleton and adhesiveness showed different expressions between two groups.
For confirming reliability of the data of a microarray, 20 genes highly expressed in the 5y-D group were selected (Hs.108504, Hs.146550, Hs.194691, Hs.1975, Hs.203952, Hs.278607, Hs.429, Hs.75305, Hs.81170, Hs.99987, Y12781, Hs.104417, cl.21783, Hs.112628, Hs.170345, Hs.53996, Hs.55422, Hs.112718, Hs.115880, and Hs.126495), and the expression levels of the genes were checked by semi-quantitative RT-PCR. The result coincided with the data of a microarray, and had a statistical significance for classifying the 5y-D group and the 5y-S group (typical data is shown in FIG. 2).
For constructing a scoring system for predicting the postoperative prognosis using the expression profile of a marker gene, prognosis score was calculated by the above-mentioned method. Briefly, a marker gene was selected according to the following standard.
(1) Higher signal strength than cut off level is shown in at least 60% of cases checked;
(2) |μ_D−μ_S| is 1.0 or less. Here, μ_D(μ_S) shows an average value derived from logarithm converted relative expression ratio in the case of 5y-D(5y-S).
Next, for identifying a marker gene capable of classifying the 5y-D group and the 5y-S group depending on the expression function, the Mann-Whitney test and the Random-permutation test were carried out. The result of a microarray correlated was confirmed by a semi-quantitative RT-PCR experiment. By the Student's t-test, 20 genes were selected as a prognosis marker (Table 2).
Depending on the prognosis score (PS) of the present invention, 20 patients were divided into 10 members predicted to show poor prognosis (PS is 11 or more) and 10 members predicted to show excellent prognosis (PS is less than 11). As a result, it was shown by comparison with the postoperative progress which the scoring system of the present invention has reliability with an accuracy of 80% in the 5y-D case and with an accuracy of 100% in the 5y-S case (FIG. 3A).
Using the prognosis scoring system of the present invention, additional 5 cases were checked (FIG. 3B). The system predicted poor prognosis in 2 cases (PS>11; patient TD-1 and patient TD-2), and excellent prognosis in 3 cases (PS<11; patients TD-3, TS-1 and TS-2). As a result, this scoring system showed an accuracy of 80% regarding actual clinical results of these 5 cases.

EXAMPLE 2

Evaluation of Gene Expression Function for Prediction of the Postoperative Progrnosis in Node-Negative Breast Cancer

(Tissue Sample)

A tissue sample was collected in the same manner as described in Example 1. Gene expression was investigated for tumors from 12 patients of node-negative (n0) cancer showed recurrence within 5 years after an operation (5Y-R) and 12 patients survived free of disease for 5 years or more after the operation (5Y-F). The clinical backgrounds of both the patient groups were allowed to coincide in age, lymph node metastasis, tumor diameter, condition of hormone receptor, and pathological tissue (Table 3). The follow up intermediate period was 7.8 years, and the average period between the initial operation and recurring was 2.7 years in the 5Y-R group. All patients underwent the adjuvant therapy described in Example 1.

(Clinical Pathological Data)

TABLE 3

			TNM
			classi-
Climacteric	Histological	Diameter	fication^b	ER	PgR

Case

Age

condition

classification^a

Position

(mm)

T

N

M

Stage

(+/−)

D.F.I.^c

R-1

55

Post.

a2

Rt.

25

2

1a

0

II

+

−

12 m

R-2

50

Pre.

a3

Lt.

25

2

1a

0

II

+

16 m

R-3

42

Pre.

a2

Rt.

25

2

0

II

+

49 m

R-4

39

Pre.

a3

Rt.

35

2

0

II

+

−

20 m

R-5

38

Pre.

a2

Lt.

30

2

0

II

+

52 m

R-6

61

Post.

a3

Lt.

34

2

0

II

−

14 m

R-7

54

Post.

b3

Lt.

30

2

0

II

−

24 m

R-8

37

Pre.

a2

Rt.

23

2

0

II

−

25 m

R-9

54

Post.

a3

Lt.

25

2

1a

0

II

+

47 m

R-10

83

Post.

a2

Rt.

28

2

1a

0

II

+

38 m

R-11

62

Post.

a2

Lt.

23

2

0

II

−

+

40 m

R-12

50

Post.

a3

Lt.

35

2

0

II

−

25 m

F-1

48

Pre.

a2

Lt.

18

2

0

II

+

8 Y

F-2

62

Post.

a2

Rt.

25

2

0

II

+

−

8 Y

F-3

57

Post.

a1

Rt.

20

1

0

I

+

7 Y 10 m

F-4

61

Post.

a2

Lt.

30

2

1a

0

II

−

7 Y 2 m

F-5

42

Pre.

a1

Lt.

12

1

1a

0

I

−

+

7 Y 11 m

F-6

51

Pre.

a2

Rt.

28

2

1a

0

II

−

7 Y 10 m

F-7

59

Post.

a2

Rt.

40

3

0

II

−

7 Y 5 m

F-8

57

Post.*

a2

Rt.

45

3

1b

0

II

−

7 Y 5 m

F-9

42

Pre.

a1

Lt.

48

2

1a

0

II

−

+

7 Y 3 m

F-10

58

Post.

a2

Lt.

13

2

0

II

−

7 Y 3 m

F-11

50

Post.

a2

Lt.

25

2

0

II

+

7 Y 8 m

F-12

55

Post.

a1

Rt.

35

2

0

II

+

7 Y 5 m

^aa1: invasive papillotubular carcinoma, a2: invasivesolid-tubularcarcinoma, a3: invasive schirrhous carcinoma

^bTNM classification: clinically classified according to TNM classification by Japan Breast Cancer Society

^cD.F.I.: period of no pathogeny (disease free interval)

(Clinicopathological Parameter)

The clinicopathological parameter was checked by the method described in Example 1. The histological grade was evaluated by a method of Elastonand Ellis (Abrams J S. Breast Cancer 2001; 8: 298-304). Lymphoduct invasion was evaluated to be deficient or positive (for example, evaluated to be positive when one or more cancer cells are present in lymphoducts around cancer). Fatinvasion was evaluated to be deficient or positive (for example, evaluated to be positive in the case of invasion into interstitial tissue).

(Preparation of cDNA Microarray)

“Genome wide cDNA microarray kit (Amersham Biosciences UK Limited, Buckinghamshire, UK)” with 25344 cDNAs was used. The PCR product was stopped on type 7 glass slides (Amersham Biosciences) using Array Spotter Generation III (Amersham Biosciences).

(Preparation and Proliferation of RNA)

Preparation and proliferation of RNA were carried out in the same method as described in Example 1.

(Labeling of aRNA, Hybridization and Scanning)

Labeling of aRNA, hybridization and scanning were carried out in the same method as described in Example 1.

(Mann-Whitney Test)

For identifying genes showing different expressions between a group of no disease and group of recurrence, normalized signals were analyzed by the Mann-Whitney test applied to a series of Xs. Here, X represents Cy5/Cy3 signal strength ratio of each gene and each sample. Genes showing a difference of 2-fold or more in expression strength between two groups were selected. Genes with signal-noise ratios of 3.0 or less were excluded from analysis.
The U value was calculated for genes imparting significant signals in at least 5 samples in both groups. Genes with U values of lower than 37 or larger than 107 were selected. Since the U value was obtained by calculation for 5Y-F group based on 5Y-R 25 group in each gene based on each X value, genes with U values lower than 37 were evaluated to manifest higher expression in 5Y-F group than in 5Y-R group (first category).
On the other hand, genes with U values higher than 107 were evaluated to manifest higher expression in 5Y-R group than in 5Y-F group (second category).
Based on this method, 78 genes were identified in the first category and 55 genes were identified in the second category. Thus, only genes showing a difference of 2-fold or more of the intermediate expression value between two groups (μX_R/μX_F≦0.5 or ≧2.0, μX_Rand μX_Frepresent average X values in 5Y-R and 5Y-F group, respectively) were defined as genes correlated with prognosis. In total, 98 genes were selected, and of them, 64 genes showed higher expression level in 5Y-F tumor and 34 genes showed higher expression level in 5Y-R tumor.

(Random-Permutation Test)

For evaluating values of genes selected by the Mann-Whitney test, a permutation test was carried out, and correlation to group difference (Ps) of genes selected was evaluated. When each gene is represented by an expression vector v(g)=(X1, X2, - - -, X24) (Xi shows a gene expression level of i-th sample in the first sample set), an idealized expression pattern is expressed by c=(c1, c2, - - - , c24) (ci=+1 or 0, depending on whether i-th sample belongs to F group or R group).
Correlation between a gene and a group difference Pgc was defined as described below. That is, Pgc=(μF+μR)/(sF+sR); μF(μR) and sF(sR) show standard deviation of log₂X of the gene “g” of each sample in a newly defined “F” group or “R” group.
The permutation test was carried out while substituting the coordinate of c. The correlation values, Pgcs were calculated between all permutations. These procedures were repeated for 10000 times. Accidentally, the p value showing a possibility of a gene for classifying two groups was evaluated for each of 58 genes selected.

(Semi-Quantitative RT-PCR)

RNA (5 μg) was treated with DNase I (Epicentre Technologies, Madison, Wis., USA), then, single-siranded cDNAs were subjected to reverse transcription using Reverscript II reversetranscriptase (manufactured by Wako Pure Chemical Industries, Ltd., Osaka, Japan) and 0.5 μg/μl oligo (dT) 12-18 primer. The preparations of single-stranded cDNAs were diluted for the subsequent PCR amplification by monitoring GAPDH as a quantitative control. All PCRs were carried out under the following reaction conditions using Gene Amp PCR system 9700 (Applied Biosystems, Foster City, Calif., USA) at an amount of 1×PCR buffer of 30 μl.
94° C. 2 minutes,
(94° C. 30 seconds, 58-62° C. 30 seconds, and 72° C. 30 seconds) for 27 to 35 cycles
72° C. 5 minutes.
Primer sequences for RT-PCR of GAPDH are as described below:

	SEQ ID No.43 (forward)
	5′-GAA AGG TGA AGG TCG GAG T-3′

	SEQ ID No. 44 (reverse)
	5′-TGG GTG GAA TCA TAT TGG AA-3′

(Primer of Semi-Quantitative PCR (Gene Highly Expressed in Group of no Disease))

TABLE 4A

	SEQ		SEQ
	ID		ID
Ac./HS	No.	Forward	No.	Reverse

M90439	45	CCAGACATCCATGGTACCTATAA	46	TATGCATTGAAACCTTACAGGGG

AF047472	47	CTGTTAAACAAAGCGAGGTTAAGG	48	GGGTTCTGCATCTCGTTTATTAG

Hs.118251	49	GACACATAGCTCATAGGCACACA	50	TTCTGGTACATGGTAAGTGCTCA

D26125	51	TCCGCCATATTGATTCTGCTTA	52	GTTTGCTTTCTGGACCATGGATA

Hs.8619	53	GATAACAACTGGACCACATCCC	54	AACAGGCAGACGAGGTAGACAC

X16135	55	GAGAAGGATGGGTCCACCAGT	56	GTACATGGGCAGCACAAATGTAT

Hs.9006	57	ATTTCATTGGTAGTATGGCCCAC	58	ATACCATGGGACAGGATTGTAAG

M18963	59	GCTCAGACCAGCTCATACTTCAT	60	CCAAAGACTGGGGTAGGTAAAAC

X07979	61	CTGGTGCTTTCTATCACCTCTTC	62	GACTAGTGTGAAACAAGATGGGC

AF018080	63	CTTGAACCCAGGAGTTTGAGAC	64	GTGCCTCAGCTTTCTGAGTAGC

Hs.58464	65	CTGGTGCTGACTATCCAGTTGA	66	CTGGTAAACTGTCCAAAACAAGG

S79867	67	CTCTTACCTGGACAAGGTGCGT	68	GGATGAGCTCTGCTCCTTGAG

J02854	69	CAATGTTTGACCAGTCCCAGA	70	CATGTTGTCTCAGTCCTCTATTGG

Z35309	71	GGACAGCAGCTGGAGTACACA	72	AATCAGATTTGTCGGTGCCTT

Hs.83097	73	GGCTCTGCACTAAGAACACAGAG	74	ACAACTAGCTCTCAGTTCAGGCA

Hs.79137	75	TGGAGCAGTATGACAAAGCTACAA	76	AAGCAGCACTGCATAAACTGTTC

Hs.4864	77	TAAGTACTTTCCTGTGGGTCGCT	78	CCACAAACAGGAAGCTATGTTCT

Y00052	79	GTACTATTAGCCATGGTCAACCC	80	CTACAGAAGGAATGATCTGGTGG

Hs.5002	81	ATCAGTACGGGGACCTTACAAAC	82	CCTGTACTGAGCTCTCCAAAGAC

U43519	83	TCCCTAGCTTCCTCTCCACA	84	AGAATCATGCCTCCCCTTCT

Hs.94653	85	ACCCCTCAAGTGTAAGGAACTG	86	GGATCAAGAGTGTGTGTGTGTGT

X51441	87	CAATGCCAGAGAGAATATCCAGA	88	GATACCCATTGTGTACCCTCTCC

Hs.108623	89	CCACTCCACATAAGGGGTTTAG	90	GAGGTTCTAGCTAAGTGCAGGGT

Hs.5318	91	CCATTGACATTGGAGTTAAGTATGC	92	GGCAAAGACCACATTTAGCAAT

Hs.69469	93	GAAAGCCTATGTGAAAAGCTGGT	94	TTGTTTCCAGGCATTAAGTGTG

AA777648	95	GCATCTTAGTCCACACAGTTGGT	96	GCCCTTACAGGTGGAGTATCTTC

Hs.106131	97	CTCATAGCCAGCATGACTTCTTT	98	GGTTCACTTGTGACTGGTCATCT

X54079	99	ACTTTTCTGAGCAGACGTCCAG	100	TATCAAAAGAACACACAGGTGGC

AI041182	101	ACGTTATTCCCAGTTCCTAAACC	102	AGTCTCGGGTGACTCAATATGAA

AA148265	103	AGTTGAACCCAGGTACCTTTCTC	104	CTAGGCCCTTTTAGAAAACATGG

Hs.4943	105	TACTGGGAACGACTAAGGACTCA	106	TGCTGTGTTGAGTAGGTTTCTGA

Hs.106326	107	TGAGAGTCCTCAGAGGGTATCAG	108	CTTGAAGTCAAGAGTCCTGGTGT

M13436	109	TTTCTGTTGGCAAGTTGCTG	110	CCCTTTAAGCCCACTTCCTC

X99920	111	GATGAGAAGATGAAGAGCTTGGA	112	GAGGAAGCTTTATTTGGGAAGAG

U22970	113	ACTTCCCTCTCTGCCTTTCTG	114	CAGATTGTTTTGGGCTTCTCACT

(Primer of Semi-Quantitative PCR (Gene Highly Expressed in Group of Recurrence))

TABLE 4B

	SEQ		SEQ
	ID		ID
Ac./HS	No.	Forward	No.	Reverse

X75252	115	GTCTGGTCAGCTTTGCTTCC	116	GGCAAGTTCTGCACAGATGA

AA989127	117	CAGCTCAGTGCACCATGAAT	118	GTGGGACTGAGATGCAGGAT

Hs.128520	119	CACGGACTCATGAATGTAGTGAA	120	GTGTAGTGGCACGATCATAGCTT

HSMLN50	121	GGGACCAAACAGACCAAAGA	122	CACCCCACAGAGCCTGTATT

AF058701	123	CGGAAAGGCACTATTTCACAAT	124	ACAGGCCCACAGGTTTGTAAC

AF043473	125	AAGCTCTTCAGCTGCGTCTC	126	CCTCCTCCTTTTCAGCTGTG

Hs.26052	127	TCTGGAACCCTAAAAGTGTCGT	128	TCTTTCAACATCTCTCCACCCTA

Hs.77961	129	AGATACCTGGAGAACGGGAAG	130	GGAAGTAAGAAGTTGCAGCTCAG

Hs.28484	131	ATTAGGTTTCACCCAAAG	132	AGACGAGACTTGTTTTCTC

U44798	133	CAGGGACTTGGTCACAGGTT	134	TTCTTCTCCCTCCCCTTGAT

Hs.77961	135	GATTACATCGCCCTGAACGAG	136	TCCATCAACCTCTCATAGCAAA

X64707	137	GTAAGATCCGCAGACGTAAGG	138	CTGAAGTCAGCCTCTGTAACCTC

Hs.6780	139	ACTGACCCCACTTCTTGTGG	140	ACCCTTCCCTGTTGCTGTC

Hs.153428	141	TCAAAGTATTTAGCTGACTCGCC	142	TAGTCACTCCAGGTTTATGGAGG

AI066764	143	GGGAACTTGAATTCGTATCCATC	144	CTGAATCTCAAACCTGGAGAGTG

cl.5994	145	GATCATCTTTCCTGTTCCAGAG	146	CTGGAAGGTTCTCAGGTCTTTA

D67025	147	GTACGACCAGGCTGAGAAGC	148	ATCTTCGGGGCTATCCAACT

x16064	149	TCAGCCACGATGAGATGTTC	150	TGTGGATGACAAGCAGAAGC

M80469	151	ACCTTAGGAGGGCAGTTGGT	152	AGGGGTCACACCTTGAACAG

E02628	153	GCATCCTACCACCAACTCGT	154	GCAGCATCACCAGACTTCAA

HUMTHYB4	155	ACAAACCCGATATGGCTGAG	156	GCCAATGCTTGTGGAATGTA

Hs.116922	157	TCGGACCATAATCCAAGTTACC

x15940	158	TAACCCGAGAATACACCATCAAC	159	ATGGTTTTATTGACGGCAGAAG

The signal strength of the RT-PCR product was measured and evaluated in the same method as described in Example 1, and 10 genes with p values of 0.05 or lower in the t-test were selected as a candidate; of them, expression levels of 3 genes were higher in the 5y-R group than in the 5y-F group. The expression levels of 7 genes were higher in the 5y-F group than in the 5y-R group. Base on this information, the present inventors have tried to establish a scoring system for predicting the postoperative prognosis of node-negative breast cancer.
For obtaining expression level to be a subject of each gene, the expression ratio (ER) to the GAPDH expression was calculated according to the following formula:
ER of gene A=16 bit imaging score of semi-quantitative PCR (strength of band stained with ethidium bromide) of gene A of cancer sample X/16 bit imaging score of GAPDH of gene A of cancer sample X

(Definition of Scoring System for Predicting Postoperative Prognosis of Node-Negative Breast Cancer)

For obtaining the postoperative gene prognosis index of node-negative breast cancer, prognosis score (PS) was defined; (sum of normalized expression ratios of genes highly expressed in 5Y-R group as compared in 5Y-F group)−(sum of normalized expression ratios of genes highly expressed in 5Y-F group as compared in 5Y-R group)
A significance of the expression ratio between two groups was evaluated by the Student's t-test. All statistical methods were carried out by Statview version 5.0 (SAS Institute, Cary, N.C.).

(Result)

Clinicopathological findings of 24 breast cancer patients whose genome-wide gene expressions have been investigated are summarized in Table 3. The present inventors have investigated the gene expression by a cDNA microarray composed of 25344 human genes, for tumors from node-negative breast cancer patients of 12 cases showing survival free of disease for 5 years or more after an operation (5Y-F) and node-negative breast cancer patients of 12 cases showing recurrence of breast cancer within 5 years after a surgical operation (5Y-R). The clinical backgrounds were allowed to coincide in age, tumor diameter, estrogen receptor and progesterone receptor, and pathology between two groups.
The data of a cDNA microarray was analyzed by the Mann-Whitney test and the Random-permutation test, and genes showing different expressions between 5Y-R tumor and 5Y-F tumor were identified. Through this filter, 58 genes in total were selected, and of them, 21 genes showed significant strong expression in 5Y-R tumor. 37 genes showed higher expression in 5Y-F tumor.
The 37 genes showed higher expression in 5Y-F tumor as compared in 5Y-R tumor had six ESTs and one virtual protein (Table SA, a difference in expression between groups is expresses as “foldchange”).

(Gene with Significant High Expression in 5Y-F Tumor as Compared in 5Y-R Tumor)

TABLE 5A

Ac./HS	kind	fold change	p value

M90439	molecular makar (EPC-1) gene	2.324	0.0014
AF047472	spleen mitotic checkpoint BUB3 (BUB3)	2.889	0.0021
Hs.118251	ESTs	2.121	0.0031
D26125	3 alpha-hydroxysteroid/dihydrodiol dehydrogenase DD4, partial cds	2.084	0.0038
Hs.8619	SRY (sex determining region Y)-box 18	3.375	0.0041
X16135	novel heterogeneous nuclear RNP protein, L protein	4.839	0.0042
Hs.9006	VAMP (vesicle-associated membrane protein)-associated protein A, 33 kDa	3.807	0.0058
M18963	islet of Langerhans regenerating protein (reg)	2.022	0.0060
X07979	integrin beta	1 subunit	2.997	0.0068
AF018080	PYRIN (MEFV)	4.016	0.0071
Hs.58464	ESTs	5.415	0.0079
S79867	type I keratin 16 [human. epidermal keratinocytes, mRNA Partial, 1422 nt]	2.254	0.0090
J02854	myosin light chain (MLC-2)	2.668	0.0090
Z35309	adenylate cyclaseB(brain)	2.264	0.0094
Hs.83097	hypothetical protein FLJ22955	4.979	0.0096
Hs.79137	protein-L isosparate(D-aspartate)o-metyltransferase	2.401	0.0105
Hs.4864	ESTs	2.043	0.0107
Y00052	Peptidylprolyl isomerase A(cyclophilin A)	2.966	0.0107
Hs.5002	copper chaperone for superoxide dismutase; CCS	2.032	0.0114
U43519	dystrophin-related protein 2 (DRP2)	2.022	0.0114
Hs.106326	ESTs	4.733	0.0123
Hs.94653	neurochondrin(KIAA0607)	2.08	0.0129
M13436	ovarian beta-A-inhibin	2.946	0.0135
X51441	serum amyloid A (SAA) protein partial, clone pAS3-alpha	2.383	0.0155
Hs.108623	thrombospondin 2	2.019	0.0174
Hs.5318	ESTs	4.38	0.0174
Hs.69469	GA17 protein	2.279	0.0197
AA777648	peripheral myelin protein 22	2.386	0.0209
Hs.106131	ESTs	2.022	0.0213
X54079	heat shock protein HSP27	5.637	0.0217
D67025	proteasome (prosome, macropain) 26S Subunit, non-ATPase, 3	3.179	0.0359
M80469	MHC class I HLA-J gene	3.572	0.0380
AI041182	ov77e07.x1 Soares_testis_NHT Homo sapiens cDNA clone IMAGE: 1643364	2.321	0.0380
AA148265	RIBOSOMAL PROTEIN L21.	2.019	0.0440
Hs.4943	Inter-Alpha-Trypsin Inhibtor Heavy Chain LIKE gene	2.426	0.0442
X99920	S100 calcium-binding protein A13	3.326	0.0456
U22970	interferon-inducible peptide (6-16) gene	2.741	0.0465

In Table 5B, 21 genes highly expressed in the 5Y-R group are listed. Of them, five genes are ESTs and one gene encodes a virtual protein. From the panel including 58 genes, marker for postoperative prognosis were selected according to the following standard; (1) Having higher signal strength than cut off level situated in at least 60% of cases; (2) |μR−μF|>1.0. Here, μR(μF) shows an average value derived from logarithm converted expression ratio in the case of 5Y-R or 5Y-F.

(Gene with Significant High Expression in 5Y-R tumor as Compared in 5Y-F Tumor)

TABLE 5B

Ac./HS	kind	fold change	p value

X75252	Prostatic Binding protein	4.506	0.0011
AA989127	major histocompatibility complex, class I, C	5.731	0.0060
Hs.128520	ESTs	1.419	0.0067
HSMLN50	ESTs	3.482	0.0071
AF058701	DNA polymerase zeta catalytic subunit (REV3)	2.185	0.0085
AF043473	delayed-rectifier K+ channel alpha subunit (KCNS1), Potassium	4.786	0.0144
	voltage-gated channel, delayed-rectifier, subfamily S, member 1
Hs.26052	hypothetical protein MGC43306	4.829	0.0150
Hs.77961	major histocompatibility complex, class I, B	5.775	0.0152
Hs.26484	HIRA interacting protein 3	5.07	0.0157
U44798	U1-snRNP binding protein homolog (70 kD)	2.615	0.0194
Hs.77961	MHC class I HLA-Bw62	5.775	0.0209
X64707	BBC1 mRNA(ribosomal protein L13)	2.758	0.0210
Hs.6780	PTK9L protein tyrosine kinase 9-like (A6-related protein)	2.749	0.0220
Hs.153428	Ests	3.164	0.0234
AI066764	lectin, galactoside-binding, soluble, 1 (galectin 1)	2.606	0.0275
cl.5994	ESTs	2.844	0.0286
x16064	Tumor protein, translationally-controlled 1	3.567	0.0366
E02628	polypeptide chain elongation factor-1 alpha	4.055	0.0427
HUMTHYB4	thymosin beta-4	4.05	0.0436
Hs.116922	ESTs	2.538	0.0494
x15940	ribosomal protein L31	2.125	0.0499

7 genes highly expressed in 5Y-F tumor as compared in 5Y-R tumor (Hs.94653, M13436, Hs.5002, D67025, M80469, Hs.4864 and Hs.106326; p=0.0018, 0.0011, 0.001, 0.008, 0.0081, 0.0018 and 0.001; each according to Student's t-test) and 3 genes relatively highly expressed in 5Y-R tumor (AF058701, AI066764, and xl5940; p=0.0351, 0.00161 and 0.0001; each according to Student's t-test) coincided with standards, and were selected as a prognosis marker (Table 6).

(Genes Selected as Prognosis Marker for Node-Negative Breast Cancer)

TABLE 6

AF058701	DNA polymerase zeta catalytic subunit (REV3)
AI066764	lectin, galactoside-binding, soluble, 1 (galectin 1)
x15940	ribosomal protein L31.
Hs.94653	neurochondrin(KIAA0607)
M13436	ovarian beta-A-inhibin
Hs.5002	copper chaperone for superoxide dismutase; CCS
D67025	proteasome (prosome, macropain) 26S subunit, non-ATPase,
	3
M80469	MHC class I HLA-J gene
Hs.4864	ESTs
Hs.106326	ESTs

Expressions of these markers were confirmed by a normalized semi-quantitative RT-PCR experiment for GAPDH expression. FIG. 3 shows results of RT-PCR of three marker genes highly expressed in samples from 12 patients showing recurrence of breast cancer (5Y-R group). FIG. 4 shows results of 7 marker genes highly expressed in the 5Y-F group (5 years survival). The expression ratios of these 10 genes were used for definition of prognosis index.
Prognosis score (PS) was defined as described below;
PS=(sum of normalized expression ratios of 3 genes highly expressed in 5Y-R tumor)−(sum of normalized expression ratios of 7 genes highly expressed in 5Y-F tumor)
The prognosis scores of 24 cases investigated are summarized in Table 7 together with the expression ratio of each marker gene. The PS system predicted poor prognosis of cases R1 to R12 having prognosis scores of more than 3. On the other hand, excellent prognosis was predicted for cases F1 to F12 having scores of lower than −16. The predictions coincided with actual clinical results of them with an accuracy of 100% (FIG. 5). The average PS of the 5Y-R group was 9.44 and the average PS of the 5Y-F group was −28.92.

(Prognosis Score for Recurrence of Node-Negative Breast Cancer)

TABLE 7

No.	x15940	AF058701	AI066764	Hs.5002	Hs.94653	M13436	M80469	D67025	Hs.4864	Hs.106326	PS

1n	8.90	2.70	8.35	1.50	0.82	1.47	2.43	2.72	2.60	2.55	5.86
2n	7.02	2.19	7.48	1.14	0.50	1.51	2.32	1.27	1.89	0.62	7.44
3n	7.57	2.36	10.96	1.40	0.55	2.29	3.51	2.38	1.79	0.44	8.53
4n	8.57	2.79	9.78	1.75	1.42	2.02	3.30	3.03	3.44	3.02	3.16
5n	14.96	2.56	18.01	3.88	0.53	0.67	3.96	2.76	3.78	1.83	18.12
6n	16.94	3.97	12.76	0.11	0.73	1.50	3.19	2.01	3.60	4.41	18.12
7n	14.51	3.02	11.62	0.37	2.24	2.05	2.14	1.45	1.64	2.96	16.30
8n	9.50	2.81	10.43	2.86	1.64	1.95	5.40	3.18	1.89	1.79	4.03
9n	8.29	2.96	8.32	0.78	0.55	1.91	1.50	1.31	1.40	2.80	9.32
10n	6.78	2.06	10.59	0.39	1.93	0.70	2.49	3.56	1.27	0.84	8.25
11n	7.30	1.38	10.89	3.03	2.82	0.46	2.18	3.09	2.00	2.16	3.83
12n	8.60	3.81	15.86	3.31	3.46	0.70	3.19	1.82	2.54	2.95	10.30
1nR	4.67	0.81	4.69	4.13	2.98	3.80	7.78	5.34	7.59	8.47	−29.92
2nR	4.32	0.63	3.88	2.82	2.68	2.89	4.51	3.74	4.86	9.28	−21.95
3nR	10.54	0.56	7.28	2.40	2.06	2.10	8.18	6.02	6.02	8.55	−16.95
4nR	5.59	0.56	4.85	3.22	3.69	2.89	11.18	3.31	6.39	11.36	−31.04
5nR	5.56	0.18	4.97	5.57	4.57	1.15	3.18	4.85	5.56	12.68	−26.85
6nR	4.50	0.51	4.01	6.81	2.54	5.45	6.61	7.49	7.16	6.18	−33.22
7nR	5.09	0.97	4.72	3.14	3.74	5.57	7.95	3.94	7.90	9.71	−31.17
8nR	4.93	0.54	4.46	7.53	4.95	5.93	11.03	1.96	6.21	7.75	−35.43
9nR	5.25	1.17	5.15	3.09	3.39	3.30	10.05	2.66	4.76	10.82	−26.50
10nR	5.36	0.59	5.96	3.67	2.78	2.47	4.66	3.12	10.63	8.27	−23.69
11nR	4.99	1.02	5.71	7.48	4.51	6.22	4.61	4.28	10.65	9.20	−35.23
12nR	4.84	0.30	4.98	7.57	6.07	5.04	7.05	3.07	7.42	8.98	−35.08

EXAMPLE 3

Evaluation of Gene Expression Function for Prediction of the Postoperative Prognosis in Primary Breast Cancer

(Tissue Sample)

A tissue sample was collected in the same manner as described in Example 1. Among 954 patients clinically traced during a period of 5 years or more or until death after an operation for breast cancer in a period from 1995 to 1997, 10 cases of death within 5 years after an operation and 10 cases of survival free of disease for 5 years or more after an operation were selected as a sample. The clinical backgrounds between two patient groups were allowed to coincide as strictly as possible regarding age, metastasis to lymph node, tumor diameter and tissue type (Table 8). The clinical backgrounds of additional 20 cases used for testing the final prediction system are summarized in Table 9.

(Clinical Profile of Patients used for Microarray Analysis)

TABLE 8

Case	T	N	M	Stage	Age	NL^a	ly^b	f^c	ER^d

Survived	MS1	2	1	0	II	52	4	1	2	P
	MS2
	2	2	0	II	47	2	0	1	P
	MS3
	2	2	0	II	40	5	0	1	N
	MS4
	2	2	0	II	64	3	0	1	N/A
Dead	MD1
	2	2	0	II	47	5	0	0	P
	MD2
	2	2	0	II	34	3	3	0	N
	MD3
	2	2	0	II	66	4	0	3	N
	MD4
	2	0	0	II	71	2	0	1	P

^aNumber of lymph nodes involved.
^bLymph vessel invasion: 0, no cancer cells in vessels. 3, many cancer cells in vessels.
^cFat invasion: 0, no invasion to fat tissue: 3, severe invasion to fat tissue.
^dEstrogen receptor status: P, positive: N, negative; N/A, not available.

(Clinical Profile of Patients used for RT-PCR Analysis)

TABLE 9

	Case	T	N	M	Stage	ly^a	f^b

Survived	S1	2	0	0	II	0	1
	S2	2	2	0	II	1	0
	S3	2	2	0	11	0	2
	S4	2	1	0	II	1	0
	S5	2	2	0	II	3	2
	S6	2	0	0	II	0	0
	S7	2	1	0	II	0	0
	S8	2	1	0	II	0	2
	S9	2	1	0	II	1	2
	S10	2	1	0	II	0	0
Dead	D1		2	1	0	II	0	1
	D2	2	2	0	II	0	0
	D3	2	2	0	II	3	0
	D4	2	2	0	II	0	3
	D5	2	2	0	II	1	3
	D6	2	1	0	II	0	1
	D7	2	0	0	II	0	1
	D8	2	1	0	II	0	0
	D9	2	4	0	IV	1	0
	D10	2	1	0	II	0	2

	Age^c	lymph node correlation^d

Survived	52.8	7.6
Dead	56.0	5.4

^aph vessel invasion
^bFat infiltration
^cMean of age
^dAverage number of lymph nodes involved

(Clinicopathological Parameter)

The clinicopathological parameters were checked by the method described in Example 1.

(Preparation of cDNA Microarray)

A cDNA microarray was prepared by the method described in Example 2.

(RNA Extraction and RNA Amplification)

RNA was extracted using TRIzol (Invitrogen, Carlsbad, Calif., USA). For removing degenerate RNA, each extracted RNA (1 μg) was subjected to electrophoresis on 3.0% formaldehyde denatured gel. For removing DNA mixing, purification was carried out using RNeasy kit (QIAGEN, Valencia, Calif.). Amplification was carried out based on T7 RNA polymerase base by Messsage Amp aRNA kit (Ambion, Austin, Tex.), and RNA used for microarray analysis was prepared. In the first amplification, RNA (5 μg) was used as a template. Thereafter, the firstly amplified RNA (aRNA) (2 μg) was used as a template for the second amplification. The amplified aRNAs were purified by RNeasy purification kit, and the amount of each aRNA was measured by a spectrophotometer.

(Labeling of aRNA, Hybridization and Data Analysis)

A hybridization probe was produced using aRNA (5 μg) for producing fluorescent probe obtained by second amplification, using Amino Allyl-cDNA labeling kit (Ambion, Austin, Tex.). Probes derived from cancer RNA and normal control RNA were labeled with Cy5 or Cy3 Mono-Reactive Dye (Amersham Bioscience UK Limited, Buckinghamshire, UK), respectively.
For removing an unbound dye, a labeled probe was purified by QIA quick PCR purification kit (QIAGEN, Valencia, Calif.). Each 10 pmol of fluorescent labeled probes from tumor and normal RNA were mixed with 4× microarray hybridization buffer (Amersham (UK)) and de-ionized formamide. The probe mixture was hybridized to a cDNA array at 40° C. for 15 hours. Thereafter, the mixture was washed with 0.1×SSC containing 0.2% SDS once for 5 minutes, then, twice for 10 minutes. All procedures were carried out in Automated Slide Processor System (Amersham). The signal strength of each hybridization was read by Gene Pix 4000 (Amersham), and evaluated by Gene Pix Pro 3.0 (Axon Instruments, Inc., Foster City, Calif., USA). The read signals were normalized by the total gene normalization method (Yang, Y. H., Dudoit, S., Luu, P., Lin, D. M., Peng, V, Ngai, J., and Speed, T.P. (2002). Nucleic Acids Res 30, e15.; Manos, E. J., and Jones, D. A. (2001). Cancer Res 61, 433-438).
For confirming genes showing different expressions between a survival group and a dead group, normalized signals were analyzed by the Mann-Whitney test; the normalized signals were applied to a series of Xs. X represents Cy5/Cy3 signal strength ratio for each gene and each sample (Ono, K., et al. (2000). Cancer Res 60, 5007-5011). Genes showing a U value of 0 in the Mann-Whitney test and genes showing a difference of 2-fold or more in expression strength between two groups were selected. Genes with S/N ratios of less than 3.0 were excluded from investigation.

(Semi-Quantitative RT-PCR Experiment and Gene Expression Ratio)

For verifying the data of a microarray, the present inventors carried out a semi-quantitative RT-PCR experiment by reverse-transcribing RNA (10 μg). For adjusting the concentration of the transcribed cDNA, GAPDH was selected as an internal control, and semi-quantitative RT-PCR was carried out (Ono, K., et al. (2000). Cancer Res 60, 5007-5011). Primers for GAPDH were 5′-ggaaggtgaaggtcggagt-3 (Forward) and 5-tgggtggaatcatattggaa-3 (Reverse). After adjusting the concentration of the primer, semi-quantitative RT-PCR was carried out on selected genes in samples from the survival group and the dead group. Primers for the genes (Table 10) were designed based on sequence information of NCBIGen Bank (http://www.ncbi.nlm.nih.gov/) and primer 3 on website (http ://www.genome.wi. mit. edu/cgi-bin/primer/primer3_www.cgi). Each semi-quantitative RT-PCR experiment was performed using, as a template, cDNA (1 μl) having been adjusted concentration, 5 U TakaraEXTaq (Takara, Otsu, Japan), 1×PCR buffer (10 mM Tris-HCl, 50 mM KCl, 1.5 mM MgCl₂), and 10 nM dNTPs and 10 pmol of forward and reverse primers, in a total amount of 30 μl.

	SEQ ID No. 160	ggaaggtgaaggtcggagt

	SEQ ID No. 161	tgggtggaatcatattggaa

(Primer of Semi-Quantitative PCR)

TABLE 10

	SEQ		SEQ
	ID		ID
gene	No.	Forward	No.	Reverse

PMP	162	CCTCCAACTGCTCCTACTCG	163	TCGAAGCCTCTGTGTCCTTT

C1r	164	GAAGTTGTGGAGGGACGTGT	165	GACTTCCAGCAGCTTCCATC

DPYSL3	166	CATGTACTGAGCAGGCCAGA	167	AAGATCTTGGCAGCGTTTGT

PTK9L	168	TTGTGATTGAGGACGAGCAG	169	AATGGTTTCCCGCTCTAGGT

CPE	170	CTCCTGAGACCAAGGCTGTC	171	TGAAGGTCTCGGACAAATCC

α-tubulin	172	GGAACGCCTGTCAGTTGATT	173	CTCAAAGCAAGCATTGGTGA

β-tubulin	174	TCTGTTCGCTCAGGTCCTTT	175	TGGTGTGGTCAGCTTCAGAG

HSP 90-a	176	AAAAATGGCCTGAGTTAAGTGT	177	TCCTCAATTTCCCTGTGTTTG

MDH	178	TGCACACTAACAGCATGACG	179	GAATTTCTTTCCTCTGCCTGA

NDUFB3	180	GGGATAAACCAGACAAGTAGGC	181	GGACATGAGCATGGACATCA

For evaluating the strengths of gene expressions between the survival group and the dead group, each semi-quantitative PCR product (8 μl) was subjected to electrophoresis on 2.5% agarose gel, and stained with ethidium bromide. The concentration of each stained sample was measured by AlphaImager 3300 (Alpha Inonotech, San Leandro, Calif.) using background correction. For obtaining the expression level of each gene, the expression ratio was normalized with the expression level of GAPDH.
The expression ratio was defined by the following formula: Expression ratio of gene A=16 bit imaging score of semi-quantitative PCR (strength of band stained with ethidium bromide) of gene A in cancer sample X/16 bit imaging score of GAPDH in cancer sample X

(Definition of Prognosis Index (PI) of Primary Breast Cancer)

The present inventors defined the prognosis index (PI) of primary breast cancer by subtracting the sum of normalized expression ratios of genes highly expressed in the 5D group from the sum of normalized expression ratios of genes highly expressed in the 5S group. A significance of expression ratios between two groups was evaluated by the Student's t-test. Comparison of PI between the 5S group and the 5D group was carried out by the Mann-Whitney test. All the statistics were storaged using Statview version 5.0 (SASInstitute Inc., Cary, N.C.).

(Result)

On a cDNA microarray composed of 18432 human genes, genome-wide gene expression functions of tumors from 8 breast cancer patients were examined. Four patients survived free of disease for 5 year or more after an operation (5S), and four patients died of breast cancer within 5 years after the operation (5D). The clinical backgrounds between two patient groups were allowed to coincide as strictly as possible regarding age, tumor diameter, metastasis to lymph node, hormone receptor condition and tissue type (Table 8).
For identifying genes showing different expressions between the 5D group and the 5S group, the present inventors analyzed the data of the cDNA microarray by the Mann-Whitney test. 23 genes in total among which six genes are ESTs/virtual proteins are genes showing a U value of 0 in the Mann-Whitney test and highly expressed in the 5S group (Table 11).

(Gene Group Highly Expressed in Survival Group by Microarray Analysis)

TABLE 11

Gene name and detail	Accession Number	Fold change

IMAGE: 39159 3′ similar to gb: 004173 PHOSPHOGLYCERATE MUTASE, BRAIN FORM	R51864	4.304
IMAGE: 22798 3′, MRNA sequence	R39171	2.918
cDNA clone IMAGE: 1693352 3′, MRNA sequence	AI140851	2.891
CCNDBP1 cyclin D-type binding-protein 1	AF082569	3.202
ESTs	AI446435	3.251
pro-alpha-1 type 3 collagen	X14420.1	3.394
complement component C1r	J04080.1	3.396
DPYSL3 dihydropyrimidinase-like 3	D78014	3.625
ribosomal protein L6	X69391.1	3.807
PTK9L protein tyrosine kinase 9-like (A6-related protein)	Y17169.2	4.143
Homo sapiens full length insert cDNA YN88E09	AF075050.1	4.257
somatostatin receptor isoform 2 (SSTR2) gene	M81830.1	5.475
CPE carboxy peptidase E	NM 001873.1	5.807
YR-29 hypothetical protein YR-29	AJ012409.1	6.333
IMAGE: 4822062, mRNA	BC034811	6.373
KIAA1832 protein, partial cds	AB058735.1	13.352
CREG cellular repressor of E1A-stimulated genes	AF084523.1	2.739
Homo sapiens putative splice factor transformer2-beta mRNA, complete cds	U61267.1	2.55
Human N-ocetyl-beta-glucosaminidase (HEXB) mRNA, 3′ end	M13519.1	2.698
Human cytochrome b5 mRNA, complete cds	M22865.1	2.881
Human pS2 mRNA induced by extrogen from human breast cancer cell line MCF-7	X00474.1	2.702
Human alpha-tubulin mRNA, complete cds	K00558	4.655
Homo sapiens clone 24703 beta-tubulin mRNA, complete cds	AF070561.1	3.917

Table 12 describes 21 genes highly expressed in general in the 5D tumor, including 6 ESTs/virtual proteins, and having a U value of 0 in the Mann-Whitney test. In the table, a difference in gene expression between two groups is shown as “foldchange”.

(Gene Group Highly Expressed in Dead Group by Microarray Analysis)

TABLE 12

Gene name and detail	Accession Number	Fold change

Lyam-1 mRNA for leukocyte adhesion molecule-1	X16150.1	7.459
APM2 adipose specific 2	NM_006829.1	4.853
DNA polymerase gamma mRNA, nuclear gene encoding mitochondrial protein	U60325.1	4.269
FLJ22128 fis, clone HEP19543	AK025781	4.109
actin related protein 2/3 complex, subunit 4, 20 kDa (ARPC4)	NM_005718.2	4.058
Scd mRNA for stearoyl-CoA desaturase	AB032261.1	3.794
novel heterogeneous nuclear RNP protein, L protein	X16135.1	3.771
ENSA endosulfine alpha	AFJ57509.1	3.511
IMAGE: 26483 5′ similar to gb: X15183_cds1 HEAT SHOCK PROTEIN HSP 90-ALPHA	R12732	3.086
malonyl-CoA decarboxylase (MLYCD)	NM_012213	3.067
anion exchanger 3 brain isoform (bAE3)	U05596.1	2.889
IMAGE: 43550 3′, MRNA sequence	H05914	2.345
cDNA FLJ23636 fis, clone CAS07176.	AK074216	2.426
IMAGE: 26366 3′ similar to gb: D16234 PROBABLE PROTEIN DISULFIDE ISOMERASE ER-60	R20554	2.519
PRECURSOR
Similar to hypothetical protein PRO2831, clone MGC: 23813 IMAGE: 4273837, mRNA, complete cds	BC017905.1	2.551
FLJ40629 hypothetical protein FLJ40629	AK097948.1	2.417
ribosomal protein L29 (humrpl29) mRNA, complete cds	U10248.1	2.203
EST, clone IMAGE: 745452, 3′end	AA625869	2.591
KIAA1554 KIAA1554 protein	AB046774.1	2.544
IMAGE: 53316 3′ similar to SP: MDHC_MOUSE P14152 MALATE DEHYDROGENASE,	R15814	2.867
CYTOPLASMIC
NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 3 (12 kD, B12), clone MGC: 9039	BC018183	4.972
IMAGE: 3881592

From 23 genes highly expressed in the 5S group and 21 gene highly expressed in the 5D group, prediction markers for postoperative prognosis were selected according to the following standards; (1) In microarray analysis, a difference in the signal strength between 5S and 5D is larger than 2.0-fold in all cases; (2) The signal strength differs significantly between 5S and 5D in semi-quantitative PCR (p value<0.05 in Student's t-test); (3) The result of semi-quantitative PCR was re-confirmed by independent triple experiments. 7 genes highly expressed in the 5S tumor and 3 genes highly expressed in the 5D tumor satisfied these standards for selecting a prognosis marker.
7 genes highly expressed in the 5S group are constituted of genes encoding pro-alpha-1 type 3 collagen (PIIIP), complement component Clr, dihydropyrimidinase-like 3 (DPYSL3), proteintyrosinekinase 9-like (PTK9L), carboxy peptidase E (CPE), α-tubulin and β-tubulin. The p values in the Student's t-test of these marker genes were 0.00039, 0.0012, 0.0042, 0.036, 0.039, 0.034 and 0.00069, respectively.
3 marker genes highly expressed in the 5D group encoded heat shock protein HSP 90-alpha gene, malatedehydrogenase, and NADH dehydrogenase (ubiquinone) 1 beta subcomplex 3 (NDUFB3). The p values in the Student's t-test of these genes were 0.05, 0.0055 and 0.011, respectively.
The present inventors normalized the experiment results of semi-quantitative RT-PCR by GAPDH as an internal control and evaluated the results, verifying selection of marker genes.
The present inventors carried out semi-quantitative PCR for checking additional 20 cases randomly selected. 10 of these patients died of breast cancer within 5 years after an operation, and remaining 10 patients survived free of disease for 5 years or more after the operation. FIG. 7 shows the results of RT-PCR of 7 marker genes highly expressed in the 5S tumor. FIG. 8 shows the results of RT-PCR of 3 marker genes highly expressed in the 5D tumor.
The present inventors defined the prognosis index (PI) as described below: (sum of normalized expression ratios of genes highly expressed in 5S group)−(sum of normalized expression ratios of genes highly expressed in 5D group). The expression ratios of the selected marker genes are summarized together with prognosis indices for further test examples in Table 13.

(Expression Ratio of Gene and Prognosis Index)

	TABLE 13

		Gene highly
	Gene highly expressed in 5S	expressed in 5D

PIIIP

C1r

DPYSL3

PTK9L

CPE

A-tubulin

B-tubulin

HSP 90

MDH

NDUFB3

Sum of S*

Sum of D**

PI⁺

S1	1.8	4.0	2.1	3.3	2.4	0.8	2.5	1.5	0.2	1.5	16.9	3.2	13.7
S2	5.7	3.5	3.3	3.4	6.0	1.2	5.1	2.2	0.6	2.2	28.1	5.0	23.1
S3	3.1	5.8	2.2	3.4	8.1	1.8	5.3	1.5	0.4	1.5	29.7	3.5	26.2
S4	7.1	10.2	8.6	6.0	16.0	4.1	8.0	3.4	4.8	3.4	60.0	11.5	48.5
S5	6.8	7.4	7.2	6.9	11.2	2.7	7.0	5.5	3.6	5.5	49.1	14.6	34.5
S6	4.0	4.2	1.7	2.2	3.6	0.9	6.0	2.9	0.9	2.9	22.7	6.6	16.1
S7	2.3	4.0	1.1	1.6	0.6	0.7	3.4	0.4	0.3	0.4	13.7	1.1	12.6
S8	3.3	3.6	1.1	0.7	0.8	1.3	5.0	2.3	1.4	2.3	15.9	6.0	9.8
S9	3.1	3.9	2.7	3.7	2.9	1.6	4.1	1.0	1.2	1.0	21.9	3.2	18.8
S10	2.9	3.0	0.9	1.5	1.2	1.0	1.7	1.3	0.4	1.3	12.2	3.0	9.2
D1	0.1	2.9	0.4	1.9	2.9	0.7	0.8	3.4	3.0	3.4	9.6	9.7	−0.1
D2	0.2	0.6	0.1	0.2	0.8	0.2	0.8	1.0	4.9	1.0	2.9	7.0	−4.1
D3	0.2	3.7	0.2	1.0	0.6	0.6	2.8	3.6	6.6	3.6	9.0	13.8	−4.8
D4	0.2	1.4	0.4	0.9	1.0	0.5	1.7	3.5	3.6	3.5	6.1	10.7	−4.6
D5	0.1	1.3	0.1	0.9	0.6	0.5	1.0	3.2	0.3	3.2	4.5	6.7	−2.2
D6	2.2	2.5	1.2	1.9	2.0	0.5	1.7	3.8	3.5	4.2	12.0	11.5	0.5
D7	2.2	2.1	0.9	1.9	2.4	0.3	1.6	1.9	1.4	2.0	11.5	5.3	6.2
D8	1.6	2.7	1.1	2.6	1.8	0.4	1.8	3.4	2.8	3.4	12.0	9.6	2.5
D9	1.2	1.4	0.6	1.6	1.2	0.6	2.4	2.2	0.7	2.2	9.2	5.0	4.1
D10	0.5	0.8	0.4	0.6	0.4	0.4	1.3	3.6	1.6	3.6	4.5	8.9	−4.4

*Sum of ER of PIIP, C1r, DPYSL3, CPE, α and β-tubulin
**Sum of ER of HSP 90, MDH and NDUFP3
+PI: Sum of S − sum of D

PI predicted correctly the actual clinical results of higher prognosis indices (>7) of 10 cases (S1 to S10) in total in the 5S group and prognosis indices (<7) of 10 cases (D1 to D10) in total in the 5D group. PI of the 5S group was 21.2. PI of the 5D group was −0.7. Here, by a PI value of 7 the 5S tumor and the 5D tumor were apparently distinguished (p=0.0002).

INDUSTRIAL APPLICABILITY

The postoperative prognosis prediction system of the present invention is effective for prediction of postoperative risk of a breast cancer patient. Further, the wide-range gene expression list of breast cancer correlated genes of the present invention can provide various information on progress of breast cancer, and a latent target molecule for breast cancer therapy was be predicted by the list.

SEQUENCE LIST FREE TEXT

SEQ ID Nos. 1 to 181: synthesized

Claims

1. A gene consisting of at least one of the following definitions correlated with prediction of the postoperative prognosis of breast cancer;

1) a marker gene group capable of establishing classification of genes from breast cancer patients died within 5 years after a surgical operation (5y-D group) and genes from patients survived free of disease for several years or more after the operation (5y-S group), depending on their expression functions, in estrogen receptor-negative breast cancer,

2) a marker gene group capable of establishing classification of genes from n0 breast cancer patients recurred within 5 years after an operation (5Y-R group) and genes from patients survived free of disease for 5 years or more after the operation (5Y-F group), depending on their expression functions, in (node-negative) (n0) breast cancer with no metastasis to a lymph node in the operation,

3) a marker gene group capable of establishing classification of genes from breast cancer patients died within 5 years after a surgical operation (5D group) and genes from patients survived free of disease for several years or more after the operation (5S group), depending on their expression functions, in primary breast cancer.

2. A gene selected from the following sequences correlated with prediction of the postoperative prognosis of primary breast cancer;

pro-alpha-1 type 3 collagen (PIIIP),

complement component Clr,

dihydropyrimidinase-like 3 (DPYSL3),

protein tyrosine kinase 9-like (PTK9L),

carboxypeptidase E (CPE),

alpha-tubulin,

beta-tubulin,

heat shock protein HSP 90-alpha gene,

malate dehydrogenase,

NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 3 (NDUFB3).

3. A gene selected from the following sequences highly expressed in a group of good prognosis correlated with prediction of the postoperative prognosis of primary breast cancer;

pro-alpha-1 type 3 collagen (PIIIP),

complement component Clr,

dihydropyrimidinase-like 3 (DPYSL3),

protein tyrosine kinase 9-like (PTK9L),

carboxypeptidase E (CPE),

alpha-tubulin,

beta-tubulin.

4. A gene selected from the following sequences highly expressed in a group of bad prognosis correlated with prediction of the postoperative prognosis of primary breast cancer; heat shock protein HSP 90-alpha gene,

malate dehydrogenase,

NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 3 (NDUFB3).

5. A gene selected from the following sequences correlated with prediction of the postoperative prognosis, in (node-negative)(n0) breast cancer with no metastasis to a lymph node in operation;

AF058701/DNA polymerase zeta catalytic subunit (REV3),

AM066764/lectin, galactoside-binding, soluble, 1 (galectin 1),

x15940/ribosomal protein L31.,

Hs.94653/neurochondrin (KIAA0607),

M13436/ovarian beta-A-inhibin,

Hs.5002/copper chaperone for superoxide dismutase; CCS,

D67025/proteasome (prosome, macropain) 26S subunit, non-ATPase, 3,

M80469/MHC class I HLA-J gene,

Hs.4864/ESTs,

Hs.106326/ESTs.

6. A gene selected from the following sequences highly expressed in a group of bad prognosis correlated with prediction of the postoperative prognosis, in (node-negative) (n0) breast cancer with no metastasis to a lymph node in operation;

AF058701/DNA polymerase zeta catalytic subunit (REV3),

AM066764/lectin, galactoside-binding, soluble, 1 (galectin 1),

x15940/ribosomal protein L31.

7. A gene selected from the following sequences highly expressed in a group of good prognosis correlated with prediction of the postoperative prognosis, in (node-negative)(n0) breast cancer with no metastasis to a lymph node in operation;

Hs.94653/neurochondrin (KIAA0607),

M13436/ovarian beta-A-inhibin,

Hs.5002/copper chaperone for superoxide dismutase; CCS,

D67025/proteasome (prosome, macropain) 26S subunit, non-ATPase, 3,

M80469/MHC class I HLA-J gene,

Hs.4864/ESTs,

Hs.106326/ESTs.

8. A gene selected from the following sequences correlated with prediction of the postoperative prognosis, in estrogen receptor-negative breast cancer;

Hs.108504/FLJ20113/ubiquitin-specific protease otubain 1

Hs.146550/MYH9/myosin, heavy polypeptide 9, non-muscle

Hs.194691/RAI3/retinoic acid induced 3

Hs. 1975/TDRD3/tudor domain containing 3

Hs.203952/TRRAP/transformation/transcription domain-associated protein

Hs.278607/GSA7/ubiquitin activating enzyme E1-like protein

Hs.429/ATP5G3/

ATP synthase, H+ transporting, mitochondrialF0complex, subunitc (subunit9) isoform3

Hs.75305/AIP/aryl hydrocarbon receptor interacting protein

Hs.81170/PIM1/pim-1 oncogene

Hs.99987/ERCC2/

Y12781/Transducin (beta) like 1 protein

Hs.104417/KIAA1205 protein

cl.21783/Hypothetical protein

Hs.112628/Hypothetical protein: MGC43581

Hs.170345/Hypothetical protein FLJ13710

Hs.53996/weakly similar to zinc finger protein 135

Hs.55422/Hypothetical protein

Hs.112718/EST

Hs.115880/EST

Hs.126495/EST.

9. A gene selected from claim 8, as a gene highly expressed in a group of bad prognosis.

10. A probe specific to the gene according to any one of claims 1 to 9.

11. A DNA microarray carrying thereon the gene and/or probe according to any one of claims 1 to 10.

12. The microarray according to claim 11, wherein the DNA microarray is a fiber type microarray.

13. A method of inspecting the postoperative prognosis of breast cancer using as a marker the gene and/or probe according to any one of claims 1 to 10.

14. A method of inspecting the postoperative prognosis of breast cancer using the microarray according to claim 11 or 12.

15. A method of screening cancer therapeutic medicines for controlling the postoperative prognosis of breast cancer using as a marker the gene and/or probe according to any one of claims 1 to 10.

16. A method of screening cancer therapeutic medicines for controlling the postoperative prognosis of breast cancer using the microarray according to claim 11 or 12.

17. A diagnosis kit for the postoperative prognosis of breast cancer containing a reagent using as a marker the gene and/or probe according to any one of claims 1 to 10.

18. The diagnosis kit according to claim 17, wherein the kit comprises a microarray.

19. The diagnosis kit according to claim 18, wherein the microarray is a fiber type microarray.