WO2011065533A1

WO2011065533A1 - Method for determination of sensitivity to pre-operative chemotherapy for breast cancer

Info

Publication number: WO2011065533A1
Application number: PCT/JP2010/071258
Authority: WO
Inventors: 眞三郎野口; 靖人直居; 和希岸; 紘一平尾
Original assignee: 国立大学法人大阪大学; シスメックス株式会社
Priority date: 2009-11-30
Filing date: 2010-11-29
Publication date: 2011-06-03
Also published as: JPWO2011065533A1

Abstract

RNA that is extracted from a sample collected from a subject is amplified to prepare a measurement sample, the expression amounts of genes included in a specific gene group are measured using the measurement sample, and the sensitivity to a pre-operative chemotherapy for breast cancer is determined based on the results obtained by analyzing the expression amounts.

Description

How to determine sensitivity to breast cancer preoperative chemotherapy

The present invention relates to a method for determining sensitivity to breast cancer preoperative chemotherapy.

In breast cancer treatment, breast cancer preoperative chemotherapy with an anticancer agent for the purpose of improving the preservation rate by tumor shrinkage is widely used. In a meta-analysis with Early Breast Cancer Trials' Collaborative Group (EBCCTCG), regarding the suppression of breast cancer recurrence and the improvement of survival rate between patients who received preoperative chemotherapy and those who received postoperative chemotherapy, It is shown that there is no difference.

Moreover, the meta-analysis by EBCCTCG shows that the breast preservation rate is significantly higher in patients who have received preoperative chemotherapy than in patients who have not received preoperative chemotherapy. In addition, the meta-analysis by EBCCTCG also shows that the prognosis afterwards tends to be good in cases where complete pathological response (pCR) is obtained by preoperative chemotherapy. Therefore, pCR is believed to be useful as a prognostic factor for breast cancer in patients who have received preoperative chemotherapy. Against this background, preoperative chemotherapy for breast cancer is expected to become increasingly popular.

However, preoperative chemotherapy may not be effective in all cases of breast cancer, so other cases, including the transition to surgery, are required for cases that do not respond to preoperative chemotherapy. Yes.
In this way, treatment tailoring is required in consideration of the responsiveness of individual tumors to preoperative chemotherapy (sensitivity to preoperative chemotherapy). However, it is difficult at present because a product that accurately determines sensitivity to preoperative chemotherapy has not been realized.

By the way, in the treatment of breast cancer, in order to diagnose whether the tissue is a breast cancer tissue before the start of treatment, for example, biopsy using a puncture aspiration biopsy, core needle biopsy, needle biopsy device with a suction device A small amount of biological sample containing breast cancer tissue can be obtained by, for example, mammotome biopsy.

In recent years, it has become possible to perform comprehensive gene expression information analysis using a microarray using nucleic acids extracted from such a small amount of biological sample. Based on the above comprehensive gene expression information analysis, it is possible to quantify and identify differences in biological characteristics such as prognosis and differences in sensitivity to treatment as individuality of cancer by gene expression level, and multiple cases with similar characteristics Attempts have been made to identify gene groups exhibiting a common expression pattern and to classify cases having gene groups exhibiting similar expression patterns (see, for example, Patent Document 1).

In Patent Document 1, multivariate analysis of the expression of a specific gene in a specimen is performed, and the specimen is classified into groups having similar gene expression patterns using the analysis result as an index. The prediction is described.

International Publication No. 02/072828 Pamphlet

However, there is currently a need for a method for more accurately determining sensitivity to preoperative chemotherapy.

It is an object of the present invention to provide a method for determining sensitivity to breast cancer preoperative chemotherapy that can accurately determine sensitivity to breast cancer preoperative chemotherapy.

That is, the present invention
[1] (A) A step of extracting RNA from a sample collected from a subject,
(B) a step of preparing a measurement sample using the RNA extracted in the step (A),
(C) a step of measuring the expression level of a gene selected from the gene group described in Table 1 and Table 2 using the measurement sample obtained in the step (B), wherein the expression level of the gene A process including at least the expression level of each gene of the gene group described in Table 3,
(D) analyzing the expression level of the gene measured in the step (C), and (E) determining sensitivity to breast cancer preoperative chemotherapy based on the analysis result obtained in the step (D) The process of
A method for determining sensitivity to breast cancer preoperative chemotherapy,
[2] The method according to [1], wherein in the step (D), the expression level is analyzed using a classification method,
[3] The method according to [2], wherein the classification method is Between-group analysis.
[4] In the step (C), the expression level of each gene in the gene group described in Table 1 and Table 2 is measured, and the measured expression level of each gene is expressed by the following formula (1):

{In formula (1), i represents the gene number assigned to each gene in the gene group described in Table 1 and Table 2, and w _i represents the gene number i described in Table 4, Table 5 and Table 6. Indicates the weighting factor corresponding to the gene, and X _i is the following formula (2):

[In the formula (2), j represents the sample number assigned to each sample, y _ij represents the standardized expression level of the gene of gene number i in the sample of sample number j, and min represents the value in parentheses. , Round represents a value rounded to the first decimal place, abs represents an absolute value of the value in parentheses, and y _i represents the following formula (3):

(In formula (3), x _i represents the expression level of the gene of gene number i, u _i represents the average value of the expression level of the gene of gene number i across the specimen, and s _i represents the gene of gene number i It shows the standard deviation of the expression level across the specimen.) Shows the standardized expression level of the gene of gene number i shown in (1). ] Represents the normalized and normalized gene expression level, and Σ _i represents the total over each gene. }
When the solution D of the discriminant is a positive value, it is sensitive to breast cancer preoperative chemotherapy, and when the solution D is zero or a negative value, breast cancer preoperative chemistry The method according to [3], wherein the method is determined to be insensitive to therapy,
[5] In the step (C), the expression level of each gene in the gene group described in Table 3 was measured, and the measured expression level of each gene and the following formula (4):

{In formula (1), i represents a gene number assigned to each gene of the gene group described in Table 3, w _i represents a weighting factor corresponding to the gene of gene number i described in Table 7, X _i is the following formula (5):

[In the formula (2), j represents the sample number assigned to each sample, y _ij represents the standardized expression level of the gene of gene number i in the sample of sample number j, and min represents the value in parentheses. , Round represents a value obtained by rounding off the first decimal place of the value in the parenthesis, abs represents an absolute value of the value in the parenthesis, and y _i represents the following formula (6):

(In formula (6), x _i represents the expression level of the gene of gene number i, u _i represents the average value of the expression level of the gene of gene number i over the specimen, and s _i represents the gene of gene number i This shows the standardized expression level of the gene with the gene number i shown in (2). ] Represents the normalized and normalized gene expression level, and Σ _i represents the total over each gene. }
When the solution D of the discriminant is a positive value, it is sensitive to breast cancer preoperative chemotherapy, and when the solution D is zero or a negative value, breast cancer preoperative chemistry A method for determining sensitivity to breast cancer preoperative chemotherapy according to [3], wherein the sensitivity is determined to be insensitive to therapy,
[6] The method for determining sensitivity to breast cancer preoperative chemotherapy according to [1], wherein the expression level is analyzed by hierarchical cluster analysis in the step (D),
[7] In the step (D), the expression level is analyzed by a scoring method, and the method for determining sensitivity to breast cancer preoperative chemotherapy according to [1],
[8] The method for determining sensitivity to breast cancer preoperative chemotherapy according to [1], wherein the expression level of each gene is measured using a microarray having at least a nucleic acid corresponding to each gene, and [9] The specimen relates to a method for determining sensitivity to breast cancer preoperative chemotherapy according to [1] above, wherein the specimen is a specimen collected from a subject by a pretreatment biopsy.

The method for determining sensitivity to breast cancer preoperative chemotherapy according to the present invention has an excellent effect that the sensitivity of breast cancer to preoperative chemotherapy can be accurately determined.

In Example 1, it is a figure which shows the result which compared the determination result by a discriminant, and the pathological diagnosis result about the sample of 50 breast cancer patients of a training set. In Example 2, it is a figure which shows the result which compared the determination result by a discriminant, and the pathological diagnosis result about the sample of 34 breast cancer patients of a validation set. In Example 3, it is a dendrogram based on the data of the expression level of each sample of a training set. In Example 3, it is a dendrogram based on the expression level data of each specimen of the validation set. In Example 4, it is a scatter diagram of the 1st principal component score and the 2nd principal component score based on the data of the expression level of each specimen of a training set. In Example 4, it is a scatter diagram of the 1st principal component score and the 2nd principal component score based on the data of the expression level of each specimen of a validation set. In Example 5, it is a graph which shows the relationship between the number of probes, and sensitivity x specificity. In Example 6, it is a figure which shows the result which compared the determination result by a discriminant, and the pathological diagnosis result about the sample of 50 breast cancer patients of a training set. In Example 7, it is a figure which shows the result which compared the determination result by a discriminant, and the pathological diagnosis result about the sample of 34 breast cancer patients of a validation set. In Example 8, it is a figure which shows the result of having compared the determination result by a discriminant, and the pathological diagnosis result about the sample of 50 breast cancer patients of a training set. In Example 9, it is a figure which shows the result of having compared the determination result by a discriminant, and the pathological diagnosis result about the sample of 34 breast cancer patients of a validation set.

The method for determining sensitivity to breast cancer preoperative chemotherapy of the present invention comprises (A) a step of extracting RNA from a sample collected from a subject,
(B) a step of preparing a measurement sample using the RNA extracted in the step (A),
(C) a step of measuring the expression level of a gene selected from the gene group described in Table 1 and Table 2 using the measurement sample obtained in the step (B), wherein the expression level of the gene A process including at least the expression level of each gene of the gene group described in Table 3,
(D) analyzing the expression level of the gene measured in the step (C), and (E) determining sensitivity to breast cancer preoperative chemotherapy based on the analysis result obtained in the step (D) The process of
It is a method including.

In Tables 1 and 2, “No” indicates a gene number. The “probe set.ID” is attached to each probe set in which 11 to 20 probes fixed on the base material are grouped in a microarray [trade name: GeneChip (registered trademark) System] manufactured by Affymetrix. Indicates the ID number. “UniGene.ID” indicates the ID number of UniGene, which is a database published by NCBI. The GenBank accession number is the accession number of the public database GenBank used for designing the sequence of each probe fixed on the substrate in the microarray [trade name: GeneChip (registered trademark) System] manufactured by Affymetrix. Indicates. Table 3 shows 13 probe sets included in the probe sets described in Table 1 and Table 2.

The present inventors measured the expression level of each gene of the gene group described in Table 1 and Table 2, and evaluated sensitivity to breast cancer preoperative chemotherapy based on the result of comprehensive analysis of the expression level. In some cases, it has been found that the sensitivity can be accurately determined. Moreover, the present inventors measured at least the expression level of each gene in the gene group described in Table 3 among the expression levels of the genes selected from the gene group described in Table 1 and Table 2, It has been found that the sensitivity can be determined with high accuracy. Furthermore, in particular, it was found that when the expression level was analyzed by a classification method incorporating multivariate analysis, the sensitivity to breast cancer preoperative chemotherapy can be accurately determined. The present invention has been completed based on such findings. In the present specification, the term “expression level of each gene of the gene group described in Table 1 and Table 2” is the probe set described in Table 1 and Table 2. It means the expression level of the gene holding the nucleic acid indicated by the GenBank accession number described in Table 1 and Table 2 corresponding to the ID. Further, in this specification, the term “expression level of a gene selected from the gene group described in Table 1 and Table 2” is the probe set described in Table 1 and Table 2. It means the expression level of a gene holding a nucleic acid selected from among the nucleic acids indicated by the GenBank accession numbers described in Table 1 and Table 2 corresponding to the ID. Furthermore, in this specification, the term “expression level of a gene selected from the gene group described in Table 3” is the probe set described in Table 3. It means the expression level of a gene holding a nucleic acid selected from among the nucleic acids indicated by the GenBank accession numbers described in Table 1 and Table 2 corresponding to the ID. GenBank is a database provided by the National Center for Biotechnology Information, the web page http: //www.GenBank. ncbi. nlm. nih. gov / sites / entrez? It can be used by db = nucleotide or the like. And the arrangement | sequence which attached | subjected the GenBank accession number of the said Table 1 and Table 2 can be obtained from the said database. The GenBank accession number indicates a number in the latest release as of March 11, 2009. In the present specification, the “gene” may be a unit of a base sequence from which RNA as a gene transcription product is extracted, and is a concept including an EST (expressed sequence tag).

In the determination method of the present invention, RNA is first extracted from a sample collected from a subject [step (A)].

The sample is preferably a sample collected from a subject by a pretreatment biopsy. Specific examples of the specimen include tissues collected from a subject by biopsy before treatment. Here, as a biopsy, for example, a puncture aspiration biopsy, a core needle biopsy, a needle biopsy device with a suction device (for example, product name: Mammotome (registered trademark) manufactured by Johnson & Johnson Co., Ltd.) is used. Biopsy (referred to as “mammotome biopsy”) and the like. Among these, since the specimen can be easily obtained with a low load, the mammotome biopsy is preferable.

Extraction of RNA from the specimen can be performed by a known method. For extraction of RNA from a specimen, a commercially available kit for extracting RNA can also be used. Here, as a commercially available kit, a product name: Trizol (registered trademark) manufactured by Invitrogen Corporation, a product name: Qiagen RNeasy kit (registered trademark) manufactured by Qiagen, or the like can be given.

Next, a measurement sample is prepared using the RNA extracted in the step (A) [step (B)].

In this step (B), a measurement sample suitable for measuring the expression level of a gene, that is, the production level of a transcription product (mRNA, cDNA, etc.) corresponding to the gene is prepared. Specifically, for the measurement sample, for example, mRNA is purified from the RNA extracted in the step (A), and the corresponding cDNA is amplified using the RNA extracted in the step (A). It can prepare by. In the present invention, the RNA extracted in the step (A) may be used as it is as a measurement sample as long as the expression level of the gene can be measured.

The purification of the mRNA can be performed using a known purification method. For purification of mRNA, a commercially available purification kit may be used.

Amplification of cDNA can be performed using a known method. A commercially available kit for amplifying cDNA can also be used for amplification of cDNA. Here, examples of the commercially available kit include a product name: WT-Ovation ^™ FFPE System V2 manufactured by NuGEN Technologies.

Next, the expression level of a gene selected from the gene group described in Table 1 and Table 2 is measured using the measurement sample obtained in the step (B). Here, the expression level of the gene includes at least the expression level of each gene in the gene group shown in Table 3 [step (C)].

In the step (C), the gene whose expression level is to be measured only needs to include at least all the genes of the gene group described in Table 3, and the number and types of genes used are particularly limited. is not. That is, you may select all the 70 genes of the gene group of the said Table 1 and Table 2 as a gene used at the said process (C) (1st aspect). Moreover, you may select the gene group of Table 3 as a gene used at the said process (C) (2nd aspect). Furthermore, as a gene used in the step (C), the gene group described in Table 3 and another gene may be selected from the gene groups described in Table 1 and Table 2 (third) Embodiment).

In this step (C), the gene expression level can be measured by, for example, microarray, quantitative RT-PCR, quantitative PCR, Northern blot analysis, or the like. Among these, since the expression level of each gene in the gene group can be measured quickly and easily, it is preferably measured using a microarray. In this case, the expression level may be used as it is in the fluorescence intensity in the microarray in the following steps.

The measurement of gene expression level using a microarray can be performed using a known method. Specifically, for example, a probe set described in Table 1 and Table 2 by using a product name: Human Genome U133 Plus 2.0 Array manufactured by Affymetrix, which is a microarray capable of analyzing human genome expression. . The expression level of each gene in the gene group can be measured at one time by 70 probes indicated by ID.

Next, the expression level of the gene measured in the step (C) is analyzed [step (D)]. Then, based on the analysis result obtained at the said process (D), the sensitivity with respect to a breast cancer preoperative chemotherapy is determined [process (E)].

In the step (D), the expression level can be analyzed using, for example, a classification method, hierarchical cluster analysis, and scoring method. Here, as the expression level, raw data of the measured expression level normalized by, for example, the expression level of a housekeeping gene can be used.

A known method can be used as the classification method. Such classification methods include, for example, Between-group analysis (BGA) [Culhane, AC, et al., Bioinformatics, 2002, Vol. 1600-1608), “see Between-group analysis of microarray data (Between-group analysis of microarray data)”, support vector machine (SVM), diagonal linear discrimination (DLDA) and k nearest neighbor classification (kNN), Examples include decision trees, Random Forest, and neural networks. Among these, BGA is preferable from the viewpoint that it is possible to classify those that are sensitive to breast cancer preoperative chemotherapy and those that are insensitive. When analyzing the expression level using such a classification method, classification is made into a specimen that is sensitive to breast cancer preoperative chemotherapy and a specimen that is insensitive based on the expression level. Therefore, in this case, in the step (E), the sensitivity to breast cancer preoperative chemotherapy can be determined based on the result of such classification. For the analysis, for example, a discriminant constructed using BGA is used according to the number and type of genes used in the step (C) (the first to third aspects). it can.

For example, in the case of the first aspect, first, in the step (C), the expression level of each gene in the gene group described in Table 1 and Table 2 is measured. And it can determine using the measured expression level of each gene, and the discriminant represented by following formula (1).

[In the formula (2), j represents the sample number assigned to each sample, y _ij represents the standardized expression level of the gene of gene number i in the sample of sample number j, and min represents the value in parentheses. , Round represents a value obtained by rounding off the first decimal place of the value in the parenthesis, abs represents an absolute value of the value in the parenthesis, and y _i represents the following formula (3):

(X _i represents the expression level of the gene of gene number i, u _i represents the average value of the expression level of the gene of gene number i across the sample, and s _i represents the expression level of the gene of gene number i. A standard deviation is shown.) A standardized expression level of the gene of gene number i shown in (3) is shown. ] Represents the normalized and normalized gene expression level, and Σ _i represents the total over each gene. }.

In the case of the second aspect, in the step (C), first, the expression level of each gene in the gene group described in Table 3 is measured. And it can determine using the measured expression level of each gene, and the discriminant represented by following formula (4).

{In Formula (4), i represents a gene number assigned to each gene of the gene group described in Table 3, w _i represents a weighting factor corresponding to the gene of gene number i described in Table 7, X _i is the following formula (5):

[In the formula (5), j represents the sample number assigned to each sample, y _ij represents the standardized expression level of the gene of gene number i in the sample of sample number j, and min represents the value in parentheses. , Round represents a value obtained by rounding off the first decimal place of the value in the parenthesis, abs represents an absolute value of the value in the parenthesis, and y _i represents the following formula (6):

(In formula (6), x _i represents the expression level of the gene of gene number i, u _i represents the average value of the expression level of the gene of gene number i over the specimen, and s _i represents the gene of gene number i It shows the standard deviation of the expression level across the specimen.) Shows the standardized expression level of the gene of gene number i shown in (1). ] Represents the normalized and normalized gene expression level, and Σ _i represents the total over each gene. }.

Furthermore, as an example of the third aspect, it is described in Table 8 and Table 9 (50 genes including the gene group described in Table 3 among the genes of the gene group described in Table 1 and Table 2). Examples include determination using each gene in the gene group. In this case, the expression level of each gene in the gene group described in Table 8 and Table 9 is measured. And it can determine using the measured expression level of each gene, and the discriminant represented by following formula (7).

{In Formula (7), i represents the gene number assigned to each gene in the gene group described in Table 8 and Table 9, and w _i corresponds to the gene of gene number i described in Table 8 and Table 9. X _i is the following formula (8):

[In the formula (8), j represents the sample number assigned to each sample, y _ij represents the standardized expression level of the gene of gene number i in the sample of sample number j, and min represents the value in parentheses. , Round represents a value obtained by rounding off the first decimal place of the value in the parenthesis, abs represents an absolute value of the value in the parenthesis, and y _i represents the following formula (9):

(In formula (9), x _i represents the expression level of the gene of gene number i, u _i represents the average value of the expression level of the gene of gene number i across the specimen, and s _i represents the gene of gene number i It shows the standard deviation of the expression level across the specimen.) Shows the standardized expression level of the gene of gene number i shown in (1). ] Represents the normalized and normalized gene expression level, and Σ _i represents the total over each gene. }.

When analyzing the expression level using the discriminant, the value of the expression level of the gene in the sample is sequentially substituted into x _i (eg, i = 1, 2,..., 70) of the discriminant. Then, a solution D is obtained. In this case, in the step (E), when the solution D is a positive value, it is sensitive to breast cancer preoperative chemotherapy, and when the solution D is zero or a negative value, it is sensitive to breast cancer preoperative chemotherapy. Can be determined to be insensitive.

In the hierarchical cluster analysis, for example, the expression level data (or fluorescence intensity data relating to the expression level) of the specimen collected from the subject and the pathological complete response (pCR) were obtained by preoperative chemotherapy. Data on the expression level of the group of known specimens (or fluorescence intensity data relating to the expression level) and data on the expression level of the group of specimens known to be unsuccessful (or fluorescence intensity data relating to the expression level) Using the expression level (or fluorescence intensity related to the expression level) to calculate distances indicating the similarity between samples, forming various clusters based on this distance, integrating the clusters, and creating a dendrogram This can be done by creating. Here, examples of the distance include a Pearson correlation coefficient and a Euclidean distance. Further, cluster integration can be performed by, for example, the Ward method, the farthest neighbor method, the center-of-center distance method, or the like. Among these, by using the Pearson correlation coefficient and the Ward method, those that are sensitive to breast cancer preoperative chemotherapy and those that are insensitive can be well separated. In this case, in the step (E), the sensitivity to breast cancer preoperative chemotherapy can be determined based on the result of the hierarchical cluster analysis.

A known method can be used as the scoring method. Examples of such a scoring method include principal component analysis, multiple regression analysis, logistic regression analysis, Partial Last Square, and the like. Among these, principal component analysis is preferable from the viewpoint that it is possible to satisfactorily distinguish between those that are sensitive to breast cancer preoperative chemotherapy and those that are insensitive. When analyzing the expression level using this scoring method, the score of the specimen sensitive to breast cancer preoperative chemotherapy and the score of the non-sensitive specimen are separated based on the expression level. Is made. Therefore, in this case, in the step (E), the sensitivity to breast cancer preoperative chemotherapy can be determined based on the result of scoring.

As described above, according to the method for determining sensitivity to breast cancer preoperative chemotherapy according to the present invention, the gene expression levels of the gene groups described in Tables 1 and 2 and at least the gene groups described in Table 3 are used. Therefore, the sensitivity to preoperative chemotherapy for breast cancer can be accurately determined.

Hereinafter, the present invention will be described in detail with reference to examples, but the present invention is not limited thereto.

[Example 1]
(1) Collection of specimens from test subjects Aspiration device with a collection needle (size 11G) from each of 90 breast cancer patients who had undergone preoperative chemotherapy at Osaka University Hospital between 2003 and 2008 A specimen was collected using a needle biopsy device (manufactured by Johnson & Johnson, trade name: Mammotome (registered trademark)). Immediately after collection of the specimen, the specimen was placed in liquid nitrogen and stored at −80 ° C. for a long time until use.

(2) Subject classification After the collection of the specimen in (1), the 90 patients were given paclitaxel 80 mg / m ² once a week for 12 weeks as preoperative chemotherapy, followed by 3 weeks. A total of 4 doses of epirubicin 75 mg / m ² , cyclophosphamide 500 mg / m ² and 5-FU 500 mg / m ² were administered each time. After the preoperative chemotherapy, the 90 patients were subjected to breast-conserving surgery or mastectomy, and sentinel lymph node biopsy or axillary lymph node dissection.

Thereafter, histopathological examination is performed to determine the pathological diagnosis and the effect of the anticancer agent, and the 90 patients are classified into a pathological complete response group (pCR group) and a non-response group (npCR group). did. Here, “pCR” refers to a state in which the tumor has completely disappeared, or the tumor remains only in the milk duct and has no infiltration site. “NpCR” refers to a state other than the pCR.

(3) Extraction of RNA from specimen and preparation of cDNA From the specimen (about 20 mg) obtained in (1) above, an RNA extraction reagent [manufactured by Invitrogen, trade name: TRIzol (registered trademark)] The RNA was extracted to obtain an RNA sample. The amount of RNA in the obtained RNA sample was measured with a spectrophotometer [trade name: Nano-Drop ND-1000, manufactured by Biomedical Science Co., Ltd.]. In addition, an RIN (RNA Integrity Number) of RNA in the obtained RNA sample is measured using an analyzer (manufactured by Agilent Technologies, Inc., trade name: Agilent Bioanalyzer), and the RNA is measured based on the RIN. Sample quality was determined. In the following experiments, RNA samples having an RNA amount of 550 ng or more and generally RIN> 6 were used.

Using the RNA sample (corresponding to 50 ng of RNA) and a random primer attached to a transcription product amplification kit (manufactured by NuGEN Technologies, trade name: WT-Ovation ^™ FFPE System V2), first-strand cDNA and second-strand cDNA were synthesized and amplified by Ribo-SPIA ^™ amplification technology. In this way, 90 types of cDNA corresponding to 90 samples were obtained.

(4) Gene expression analysis Using the fragmentation / labeling reagent [manufactured by NuGEN Technologies, trade name: FL-Ovation ^™ cDNA Biotin Module V2], the cDNA obtained in (3) above was obtained. It was labeled with biotin and fragmented.

The obtained fragmented biotin-labeled cDNA was hybridized overnight with a nucleic acid (probe set) on an array for human genome expression analysis [manufactured by Affymetrix, trade name: Human Genome U133 Plus 2.0 Array]. The fragmented biotin-labeled cDNA and the nucleic acid (probe set) on the array were hybridized according to the manufacturer's recommended conditions (Affymetrix).

Next, the array after hybridization is subjected to a microarray washing / staining treatment-dedicated device (manufactured by Affymetrix, trade name: GeneChip (registered trademark) Fluidics Station 450), and nucleic acids (probe set) on the array The cDNA hybridized with was fluorescently stained and washed.

Thereafter, the array is subjected to a laser scanner (trade name: GeneChip (registered trademark) Scanner 3000, manufactured by Affymetrix), and based on a fluorescent labeling substance of cDNA hybridized to the nucleic acid (probe set) on the array. The signal was read and the fluorescence intensity was quantified. The obtained fluorescence intensity data was processed by software [manufactured by Affymetrix, trade name: GeneChip (registered trademark) Operating Software] to obtain a CEL file. The CEL file was used for gene expression analysis and data quality check. In this way, CEL files were obtained for fluorescence intensity data based on nucleic acids corresponding to the probes of the probe set in each of 90 samples.

(5) Construction of discriminant for selection of gene and determination of susceptibility to breast cancer preoperative chemotherapy The following pre-processing (normalization) of the data of 90 CEL files obtained was performed with analysis software [Affymetrix ( Affymetrix (trade name: Affymetrix Expression Console software) manufactured by Affymetrix) was used. All other analyzes were carried out using statistical analysis software R (https://www.r-project.org/) and statistical analysis software Bioconductor (https://www.bioconductor.org/).

Based on the guidelines of the manufacturer [Affymetrix] [(Affymetrix, GeneChip (registered trademark) Expression Analysis Data Analysis Fundamentals, 2004.) Fluorescence intensity data of 90 samples for each sample (intensity parameter quality data) Raw-Q), 3 ′ / 5 ′ ratio such as β-actin gene which is a housekeeping gene, glyceraldehyde 3-phosphate dehydrogenase (GAPDH) gene, scaling factor, Present Call% (P- call%)]. With respect to the fluorescence intensity data of each of the 90 specimens, principal component analysis (PCA) was performed on the data in the case where all 54675 probe sets on the human genome expression analysis array were used, and the quality was evaluated.

Then, among the fluorescence intensity data of each of the 90 specimens, 6 cases of data that are out of the population by the first principal component of PCA are excluded, and the fluorescence intensity data of each of the remaining 84 specimens are as follows. Used for analysis. In the fluorescence intensity data of each of the 84 specimens, among the quality control parameters, the 3 ′ / 5 ′ ratio of the β-actin gene is generally less than 21 and the 3 ′ / 5 ′ ratio of the GAPDH gene is approximately 3 Less than 0.0, P-call% was in the range of more than 60% and less than 70%, and the scaling factor was generally less than 10%. However, in the fluorescence intensity data of each of the six specimens, the 3 '/ 5' ratio of the β-actin gene, the 3 '/ 5' ratio of the GAPDH gene, P-call%, and the scaling factor are one or more items. In the case of the fluorescence intensity data of each of the 84 samples, the value was significantly different from that of the quality control parameter.

Next, the fluorescence intensity data of each of the 84 samples was divided into 50 training sets and 34 validation sets. At this time, the pCR rate in the training set and the pCR rate in the validation set were randomly allocated from the pCR group and the npCR group, respectively. The pCR rate in the training set was 30% (15 cases / 50 cases), and the pCR rate in the validation set was 32.3% (11 cases / 34 cases). The following analysis was applied to the training set data except for those described separately.

Log 2 conversion was performed on the fluorescence intensity data of each of the 50 samples distributed to the training set. In the probe set, except for a probe set (control probe) corresponding to a housekeeping gene or the like, in 30% or more of the specimens of the 50 cases, the Absolute Call which is the reliability of the signal becomes “present”. And the probe set (7983 probe sets / total 54675 probe sets) having a coefficient of variation cv of 0.15 or more were used for gene expression analysis.

Thereafter, the expression level of the gene holding the nucleic acid corresponding to each probe set (fluorescence intensity based on the gene) is expressed by the following formula (10):
[Measured value of expression level (fluorescence intensity) (raw data) −average value of expression level (fluorescence intensity)] / standard deviation (10)
Normalized by

In addition, in order to select a gene corresponding to a probe set in which a difference in fluorescence intensity, that is, an expression level is significantly observed between the pCR group and the npCR group, SAM (Significance Analysis of Microarrays) under the condition of the number of rearrangements 1000. ) To calculate each p-value [For “SAM”, Tusher, V. G. et al., “Significance analysis of microarray applied to the ionizing radiation applied to the ionizing radiation response”. response), Proceedings of the National Academy of Sciences of the United States Breakfast America (Proceedings of the National Academy of Sciences of the United States of America), 2001 years, the first 98 vol., Pp. 5116-5121]. Then, genes having nucleic acids corresponding to 500 probe sets were selected in ascending order of the p value calculated by SAM.

Next, BGA was used to construct the discriminant. The weight of the pCR group was changed from 1 to 10 so as to obtain the optimum determination accuracy. Then, under each weighting, the optimal number of probe sets for determination of sensitivity to breast cancer preoperative chemotherapy was determined by sequential forward filtering.
Specifically, from the 500 probe sets, the probe set is selected while increasing the number of probe sets to be selected in increments of 5 until the p value reaches 500, and the discriminant is constructed. Was done. Then, the number of probe sets that maximized the product of sensitivity and specificity was determined. The sensitivity was obtained by dividing the number of specimens whose pathological diagnosis result was pCR and predicted to be pCR by the number of specimens whose pathological diagnosis result was pCR. The specificity was determined by dividing the number of specimens whose pathological diagnosis result was npCR and predicted to be npCR by the number of specimens whose pathological diagnosis result was npCR.

The process from the selection of the gene to the construction of the discriminant was repeated with 3-Fold Cross-Validation, and the sensitivity and specificity under each weighting condition were estimated as average values. As a result of the estimation, the optimum weighting condition is 4, and the optimum probe set under that condition is shown in Table 10, Table 11 and Table 12, in the descending order of the p value calculated by the SAM in the top 70th. Probe sets (see Table 1 and Table 2).

Based on the above conditions, a final discriminant was constructed using data from all training sets. The constructed discriminant is a discriminant represented by the following formula (1).

(In formula (3), x _i represents the expression level of the gene of gene number i, u _i represents the average value of the expression level of the gene of gene number i across the specimen, and s _i represents the gene of gene number i It shows the standard deviation of the expression level across the specimen.) Shows the standardized expression level of the gene of gene number i shown in (1). ] Represents the normalized and normalized gene expression level, and Σ _i represents the total over each gene. }.

Here, when the solution D of the discriminant represented by the formula (1) is a positive value, it is sensitive to breast cancer preoperative chemotherapy, and when the solution D is zero or a negative value, it is preoperative for breast cancer. Determined to be insensitive to chemotherapy.

(6) Comparison between determination result by discriminant and pathological diagnosis result Expression level data (fluorescence intensity data) measured for all 50 samples distributed to the training set, and the above formula (1) Using the discriminant represented by the formula, it was determined whether all the 50 samples correspond to the samples of breast cancer patients in the pCR group or the npCR group. And the performance of the said discriminant was evaluated by making the pathological diagnosis result into a true value and comparing the said pathological diagnosis result and the determination result by the discriminant represented by said Formula (1). FIG. 1 shows the result of examining the relationship between the determination result by the discriminant and the pathological diagnosis result for the samples of 50 breast cancer patients in the training set in Example 1.

From the results shown in FIG. 1, among the 50 samples distributed to the training set, 23 samples correspond to the samples of breast cancer patients in the npCR group according to the discriminant represented by the formula (1). Then, it is determined that 27 samples are determined to correspond to the samples of breast cancer patients in the pCR group. That is, when using the discriminant represented by the formula (1), the 23 samples are those of breast cancer patients who are insensitive to breast cancer preoperative chemotherapy, while the 27 It can be seen that the example specimen is determined to be a specimen of a breast cancer patient that is sensitive to breast cancer preoperative chemotherapy.

Further, when the performance of the discriminant represented by the formula (1) is evaluated with the pathological diagnosis result as a true value, the sensitivity is 100%, the specificity is 65.7%, the negative predictive value (NPV) is 100%, and It can be seen that the positive predictive value (PPV) is 55.6%.

Therefore, from these results, by using the expression level of each gene of the gene group described in Table 1 and Table 2 and the discriminant represented by the above formula (1), the sensitivity to breast cancer preoperative chemotherapy is improved. It is suggested that the determination can be made with high accuracy.

[Example 2]
(Evaluation of judgment accuracy of discriminant)
Using the expression level data (fluorescence intensity data) measured for all 34 specimens allocated to the validation set and the discriminant represented by the above formula (1), all 34 specimens were pCR. Sensitivity to breast cancer preoperative chemotherapy was determined by determining which group of breast cancer patients was in the group and npCR group. Further, the pathological diagnosis result was evaluated as the true value by comparing the pathological diagnosis result with the determination result based on the discriminant represented by the formula (1). In Example 2, FIG. 2 shows the result of comparing the determination result based on the discriminant and the pathological diagnosis result for the samples of 34 breast cancer patients in the validation set.

From the results shown in FIG. 2, among the 34 samples distributed to the validation set, 15 samples correspond to the samples of breast cancer patients in the npCR group according to the discriminant represented by the above formula (1). Then, it is determined that 19 samples are determined to correspond to the samples of breast cancer patients in the pCR group. That is, when the discriminant represented by the formula (1) is used, the 15 samples are samples of breast cancer patients who are insensitive to breast cancer preoperative chemotherapy, It can be seen that the example specimen is determined to be a specimen of a breast cancer patient that is sensitive to breast cancer preoperative chemotherapy.

Further, when the performance of the discriminant is evaluated with the pathological diagnosis result as a true value, the sensitivity is 90.9%, the specificity is 60.9%, the negative predictive value (NPV) is 93.3%, and the positive predictive value (PPV) ) 52.6%.

Thus, in all 34 specimens assigned to the validation set, the discriminant represented by the above formula (1) is applied to breast cancer preoperative chemotherapy as in the case of the specimen assigned to the training set. It can be seen that the sensitivity can be accurately determined. In addition, by using the expression level of each gene in the gene group described in Table 1 and Table 2 and the discriminant represented by the above formula (1), regardless of the specimen used, it can be used for breast cancer preoperative chemotherapy. It is suggested that sensitivity can be determined with high accuracy.

[Example 3]
Determination by Hierarchical Cluster Analysis In Example 1, the Pearson phase for the fluorescence intensity (expression level) data of each sample distributed to the training set and the fluorescence intensity (expression level) data of each sample allocated to the validation set Hierarchical cluster analysis was performed using the number of relations and the ward method to create a dendrogram. In Example 3, a dendrogram based on the expression level data of each sample of the training set is shown in FIG. Moreover, in Example 3, the dendrogram based on the expression level data of each specimen of the validation set is shown in FIG.

From the results shown in FIG. 3 and FIG. 4, a specimen of a breast cancer patient who is insensitive to breast cancer preoperative chemotherapy and a breast cancer preoperative chemistry, with the thick line attached to divide the dendrogram as a boundary. It can be seen that samples from breast cancer patients who are sensitive to therapy can be separated. Therefore, from these results, it is possible to accurately determine the sensitivity to breast cancer preoperative chemotherapy by performing a hierarchical cluster analysis using the expression level of each gene of the gene group described in Table 1 and Table 2. Recognize.

[Example 4]
Determination by Principal Component Analysis In Example 1, the principal component analysis was performed on the fluorescence intensity (expression level) data of each specimen distributed to the training set using the genes shown in Tables 1 and 2, and A conversion coefficient was calculated, and first and second principal component scores were calculated. In Example 1, the first principal component score and the second principal component score were calculated according to the conversion coefficient of the gene for the fluorescence intensity (expression level) data of each specimen distributed to the validation set. The conversion coefficients calculated in Example 4 are shown in Table 13 and Table 14. Further, in Example 4, a scatter diagram of the first principal component score and the second principal component score based on the expression level data of each specimen of the training set is shown in FIG. Moreover, in Example 4, the scatter diagram of the 1st and 2nd main component score based on the data of the expression level of each specimen of a validation set is shown in FIG. In FIG. 5 and FIG. 6, PCA1 indicates a first principal component score, and PCA2 indicates a second principal component score. In the figure, white circles represent the pCR group, and crosses represent the npCR group.

From the results shown in FIG. 5 and FIG. 6, a sample of a breast cancer patient who is insensitive to breast cancer preoperative chemotherapy and a breast cancer preoperative after the first principal component score on the horizontal axis becomes zero. It can be seen that it can be separated from breast cancer specimens that are sensitive to chemotherapy. Therefore, from these results, it is understood that sensitivity to breast cancer preoperative chemotherapy can be accurately determined by performing principal component analysis using the expression level of each gene of the gene group described in Tables 1 and 2 above. .

[Example 5]
Of the 70 genes obtained in Example 1, sufficient gene combinations to determine susceptibility to breast cancer preoperative chemotherapy were examined by the variable reduction (Backward-elimination) method shown below.

The arbitrary gene A was excluded from the 70 genes obtained in Example 1, and the remaining 69 gene combinations were selected. Next, using the selected 69 gene combinations, the weighting condition of the pCR group is set to 4, and in the same manner as in Example 1, a discriminant is constructed using BGA, and 3-Fold Cross-Validation is performed. Sensitivity and specificity were evaluated. In constructing the discriminant, the same training set as in Example 1 was used for evaluation of sensitivity and specificity. In 3-Fold Cross-Validation, the training set cases were divided into three groups. Two of these groups were used for constructing the discriminant, and the remaining one group was used for evaluation of sensitivity and specificity.

Further, instead of the gene A, except that the gene other than the gene A out of the 70 genes obtained in Example 1 was excluded and a combination of 69 genes was selected, the same as described above, Using a combination of 69 genes out of the 70 genes obtained in Example 1, a discriminant was constructed for each of 69 patterns excluding genes other than gene A. For these 69 discriminants, sensitivity and specificity were evaluated in the same manner as described above. Then, among the 70 combinations of 69 genes including the above-mentioned case excluding gene A, the combination having the maximum product of sensitivity and specificity was selected. In this way, genes that cause a decrease in the product of sensitivity and specificity were excluded from the genes constituting the selected combination.

Thereafter, in the same manner as described above, until the number of genes constituting the combination is two, the gene that causes the reduction of the product of sensitivity and specificity is repeatedly excluded. Then, the product of sensitivity and specificity (sensitivity × specificity) was plotted against the number of genes (number of probes) constituting the combination. FIG. 7 shows the results of examining the relationship between the number of probes and sensitivity × specificity in Example 5.

FIG. 7 shows that the product of sensitivity and specificity (sensitivity × specificity) is greatest when the number of genes (the number of probes) is 13 to 24. That is, it can be seen from this result that the minimum number of genes (the number of probes) having the maximum value of the product of sensitivity and specificity (sensitivity × specificity) is 13. These 13 genes are as shown in Table 3 above. The constructed discriminant is a discriminant represented by the following formula (4).

Here, when the solution D of the discriminant represented by the above formula (4) is a positive value, it is sensitive to breast cancer preoperative chemotherapy, and when the solution D is zero or a negative value, preoperative breast cancer Determined to be insensitive to chemotherapy.

[Example 6]
In Example 1, using the expression level data (fluorescence intensity data) measured for all 50 specimens allocated to the training set and the discriminant represented by the above formula (4), a total of 50 samples were used. It was determined whether the sample of the example corresponds to the sample of the breast cancer patient in the pCR group or the npCR group. Then, with the pathological diagnosis result as a true value, the performance of the discriminant expression was evaluated by comparing the pathological diagnosis result with the determination result based on the discriminant expression represented by the formula (4). FIG. 8 shows the result of examining the relationship between the determination result based on the discriminant and the pathological diagnosis result for the samples of 50 breast cancer patients in the training set in Example 6.

From the results shown in FIG. 8, out of all 50 samples distributed to the training set, 23 samples correspond to the samples of breast cancer patients in the npCR group according to the discriminant represented by the above formula (4). Then, it is determined that 27 samples are determined to correspond to the samples of breast cancer patients in the pCR group. That is, when the discriminant represented by the formula (4) is used, the 23 samples are those of breast cancer patients that are insensitive to breast cancer preoperative chemotherapy, while the 27 It can be seen that the example specimen is determined to be a specimen of a breast cancer patient that is sensitive to breast cancer preoperative chemotherapy.

Further, when the performance of the discriminant represented by the formula (4) is evaluated with the pathological diagnosis result as a true value, the sensitivity is 100%, the specificity is 65.7%, the negative predictive value (NPV) is 100%, and It can be seen that the positive predictive value (PPV) is 55.6%.

Therefore, from these results, by using the expression level of each gene of the gene group described in Table 3 above and the discriminant represented by the formula (4), the sensitivity to breast cancer preoperative chemotherapy is accurately determined. It is suggested that it can be done.

[Example 7]
In Example 1, using the expression level data (fluorescence intensity data) measured for all 34 specimens allocated to the validation set and the discriminant represented by the formula (4), Sensitivity to breast cancer pre-operative chemotherapy was determined by determining whether the sample in the example corresponds to a sample from a breast cancer patient in the pCR group or the npCR group. Further, the pathological diagnosis result was evaluated as a true value, and the performance of the discriminant was evaluated by comparing the pathological diagnosis result with the determination result by the discriminant represented by the formula (4). FIG. 9 shows the result of comparison between the determination result by the discriminant and the pathological diagnosis result for the samples of 34 breast cancer patients in the validation set in Example 7.

From the results shown in FIG. 9, among the 34 samples distributed to the validation set, 14 samples correspond to the samples of breast cancer patients in the npCR group according to the discriminant represented by the above formula (4). Then, it is determined that 20 samples are determined to correspond to the samples of breast cancer patients in the pCR group. That is, when the discriminant represented by the formula (4) is used, the 14 samples are samples of breast cancer patients who are insensitive to breast cancer preoperative chemotherapy, It can be seen that the example specimen is determined to be a specimen of a breast cancer patient that is sensitive to breast cancer preoperative chemotherapy.

Further, when the performance of the discriminant represented by the formula (4) is evaluated with the pathological diagnosis result as a true value, the sensitivity is 81.8%, the specificity is 52.2%, and the negative predictive value (NPV) is 85. 7% and positive predictive value (PPV) of 45.0%.

Thus, in all 34 specimens assigned to the validation set, the discriminant represented by the above equation (4) is used for preoperative chemotherapy for breast cancer as in the case of the specimen assigned to the training set. It can be seen that the sensitivity can be accurately determined. In addition, it is suggested that the sensitivity to preoperative chemotherapy for breast cancer can be determined with sufficient accuracy by using the expression level of each gene in the gene group described in Table 3 and the discriminant represented by Formula (4). Is done.

[Example 8]
In Example 5, when the number of genes (the number of probes) is 50, the combination of genes having a maximum product of sensitivity and specificity (sensitivity × specificity) is the gene group shown in Table 8 and Table 9. It is a combination. The constructed discriminant is a discriminant represented by the following formula (7).

In Example 1, using the expression level data (fluorescence intensity data) measured for all 50 samples distributed to the training set and the discriminant represented by the above formula (7), a total of 50 samples were used. It was determined whether the sample of the example corresponds to the sample of the breast cancer patient in the pCR group or the npCR group. Then, with the pathological diagnosis result as a true value, the performance of the discriminant expression was evaluated by comparing the pathological diagnosis result with the determination result based on the discriminant expression represented by the expression (7). FIG. 10 shows the results of examining the relationship between the determination result based on the discriminant and the pathological diagnosis result for the samples of 50 breast cancer patients in the training set in Example 8.

From the results shown in FIG. 10, among the 50 samples distributed to the training set, 23 samples correspond to the samples of breast cancer patients in the npCR group according to the discriminant represented by the above formula (7). Then, it is determined that 27 samples are determined to correspond to the samples of breast cancer patients in the pCR group. That is, when the discriminant represented by the formula (7) is used, the 23 samples are those of breast cancer patients who are insensitive to breast cancer preoperative chemotherapy, while the 27 It can be seen that the example specimen is determined to be a specimen of a breast cancer patient that is sensitive to breast cancer preoperative chemotherapy.

Further, when the performance of the discriminant represented by the formula (7) is evaluated with the pathological diagnosis result as a true value, the sensitivity is 100%, the specificity is 65.7%, the negative predictive value (NPV) is 100%, and It can be seen that the positive predictive value (PPV) is 55.6%.

Therefore, from these results, by using the expression level of each gene in the gene group described in Table 8 and Table 9 and the discriminant represented by the above formula (7), the sensitivity to breast cancer preoperative chemotherapy is improved. It is suggested that the determination can be made with high accuracy.

[Example 9]
In Example 1, using the expression level data (fluorescence intensity data) measured for all 34 specimens distributed to the validation set and the discriminant represented by the above formula (7), Sensitivity to breast cancer pre-operative chemotherapy was determined by determining whether the sample in the example corresponds to a sample from a breast cancer patient in the pCR group or the npCR group. Further, the pathological diagnosis result was evaluated as the true value by comparing the pathological diagnosis result with the determination result by the discriminant represented by the formula (7). In Example 9, FIG. 11 shows the result of comparing the determination result by the discriminant and the pathological diagnosis result for the samples of 34 breast cancer patients in the validation set.

From the results shown in FIG. 11, of the 34 samples allotted to the validation set, 15 samples correspond to the samples of breast cancer patients in the npCR group according to the discriminant represented by the above formula (7). Then, it is determined that 19 samples are determined to correspond to the samples of breast cancer patients in the pCR group. That is, when the discriminant represented by the formula (7) is used, the 15 samples are samples of breast cancer patients who are insensitive to breast cancer preoperative chemotherapy, It can be seen that the example specimen is determined to be a specimen of a breast cancer patient that is sensitive to breast cancer preoperative chemotherapy.

Further, when the performance of the discriminant represented by the formula (7) is evaluated with the pathological diagnosis result as a true value, the sensitivity is 81.8%, the specificity is 56.5%, and the negative predictive value (NPV) is 86. 7% and positive predictive value (PPV) of 47.4%.

Thus, in all 34 samples distributed to the validation set, the discriminant represented by the above equation (7) is used for preoperative chemotherapy for breast cancer, as in the case of the samples distributed to the training set. It can be seen that the sensitivity can be accurately determined. In addition, by using the expression level of each gene of the gene group described in Table 8 and Table 9 and the discriminant represented by the formula (7), the sensitivity to breast cancer preoperative chemotherapy is determined with sufficient accuracy. It is suggested that it can be done.

From the results described above, by measuring at least the expression level of each gene in the gene group described in Table 3 among the expression levels of genes selected from the gene group described in Table 1 and Table 2, preoperative breast cancer This suggests that the sensitivity to chemotherapy can be accurately determined.

Claims

(A) a step of extracting RNA from a sample collected from a subject;
(B) a step of preparing a measurement sample using the RNA extracted in the step (A),
(C) a step of measuring the expression level of a gene selected from the gene group described in Table 1 and Table 2 using the measurement sample obtained in the step (B), wherein the expression level of the gene A process including at least the expression level of each gene of the gene group described in Table 3,
(D) analyzing the expression level of the gene measured in the step (C), and (E) determining sensitivity to breast cancer preoperative chemotherapy based on the analysis result obtained in the step (D) The process of
A method for determining sensitivity to breast cancer preoperative chemotherapy.
The method according to claim 1, wherein, in the step (D), the expression level is analyzed using a classification method.
The method according to claim 2, wherein the classification method is Between-group analysis.
In the step (C), the expression level of each gene of the gene group described in Table 1 and Table 2 was measured, and the measured expression level of each gene and the following formula (1):

{In formula (1), i represents the gene number assigned to each gene in the gene group described in Table 1 and Table 2, and w i represents the gene number i described in Table 4, Table 5 and Table 6. Indicates the weighting factor corresponding to the gene, and X i is the following formula (2):

[In the formula (2), j represents the sample number assigned to each sample, y ij represents the standardized expression level of the gene of gene number i in the sample of sample number j, and min represents the value in parentheses. , Round represents a value rounded to the first decimal place, abs represents an absolute value of the value in parentheses, and y i represents the following formula (3):

(In formula (3), x i represents the expression level of the gene of gene number i, u i represents the average value of the expression level of the gene of gene number i across the specimen, and s i represents the gene of gene number i It shows the standard deviation of the expression level across the specimen.) Shows the standardized expression level of the gene of gene number i shown in (1). ] Represents the normalized and normalized gene expression level, and Σ i represents the total over each gene. }
When the solution D of the discriminant is a positive value, it is sensitive to breast cancer preoperative chemotherapy, and when the solution D is zero or a negative value, breast cancer preoperative chemistry 4. The method of claim 3, wherein the method is determined to be insensitive to therapy.
In the step (C), the expression level of each gene in the gene group described in Table 3 was measured, and the measured expression level of each gene and the following formula (4):

{In formula (1), i represents a gene number assigned to each gene of the gene group described in Table 3, w i represents a weighting factor corresponding to the gene of gene number i described in Table 7, X i is the following formula (5):

[In the formula (5), j represents the sample number assigned to each sample, y ij represents the standardized expression level of the gene of gene number i in the sample of sample number j, and min represents the value in parentheses. , Round represents a value obtained by rounding off the first decimal place of the value in the parenthesis, abs represents an absolute value of the value in the parenthesis, and y i represents the following formula (6):

(In formula (6), x i represents the expression level of the gene of gene number i, u i represents the average value of the expression level of the gene of gene number i over the specimen, and s i represents the gene of gene number i It shows the standard deviation of the expression level across the specimen.) Shows the standardized expression level of the gene of gene number i shown in (1). ] Represents the normalized and normalized gene expression level, and Σ i represents the total over each gene. }
When the solution D of the discriminant is a positive value, it is sensitive to breast cancer preoperative chemotherapy, and when the solution D is zero or a negative value, breast cancer preoperative chemistry 4. The method of claim 3, wherein the method is determined to be insensitive to therapy.
The method according to claim 1, wherein, in the step (D), the expression level is analyzed by hierarchical cluster analysis.
The method according to claim 1, wherein in the step (D), the expression level is analyzed by a scoring method.
The method according to claim 1, wherein the expression level of each gene is measured using a microarray having at least a nucleic acid corresponding to each gene.
The method according to claim 1, wherein the specimen is a specimen collected from a subject by a pretreatment biopsy.