JP2003061678A - Method for gene screening and method for sensitivity judgment - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide new methods for gene screening and disease diagnosis. SOLUTION: In this method for gene screening concerned with medicine sensitivity, the method comprises (1) a process for dividing a plurality of patients having diseases into a first patient group exhibiting sensitivity to a medicine and a second patient group not to exhibiting sensitivity to the medicine, (2) a process for analyzing gene expression profiles of the first patient group and the second patient group and (3) a process for selecting one or more genes significantly different between the first and the second patient groups of expression degree by statistical assay.

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to a gene screening method and a disease diagnosis method. More specifically, the present invention relates to a method for screening a gene and a method for diagnosing a disease, which comprises analyzing the expression profile of the gene in a patient.

[0002]

Despite the recent progress in molecular genetic and biological elucidation of cancer, the number of patients suffering from side effects of anticancer drugs without receiving a beneficial therapeutic effect is still large. While millions of patients with advanced cancer die each year, a significant proportion of them are treated with radiation and / or chemotherapy. Since these radiotherapy and / or chemotherapy are often ineffective, various means have been tried to predict the sensitivity or resistance to adjuvant treatment. Many molecules including multidrug resistance genes have been proposed as predictive markers, but so far none have been demonstrated to be clinically practicable.

Since the nature of cancer cells can vary widely among patients, it is difficult to identify individual tumors with single or several types of molecular markers. Previously, studying the expression profiles of large numbers of genes was cumbersome and impractical, but the development of sophisticated microarray technology allows the simultaneous analysis of thousands of molecular markers in a single experiment. Became. Since the function of all gene products expressed in a patient's cancer cells may be reflected in the properties of the cancer cells, the set of genes whose expression is altered in the patient's tumor is important for response to anti-cancer drugs. May be a determinant.

Based on the above hypothesis, the present inventors have identified tens or hundreds of genes which are differentially expressed in the tumors of patients, thereby providing a novel method for personal treatment of cancer patients. We hypothesized that diagnostic guidelines could be established. Unlike other cancers of the digestive tract, which are primarily adenocarcinomas, the majority of esophageal cancers are squamous cell carcinomas (SCCs), which are generally sensitive to chemotherapy and radiation. In clinical trials, adjuvant and neoadjuvant chemotherapy appeared to improve survival in patients with SCC (Ancona, E. et al. Final report).
on 163 consecutive patients with 5-year follow-u
p. Ann. Surg. 226, 714-723 (1997); and Hainsworth,
JD et al. Cancer85, 1269-1276 (1999)). However, the majority of patients in these studies suffered severe side effects, and in some cases the drug had little or no effect on cancer cells.

Studies on the above problems using various molecular approaches suggest that genetic and epigenetic changes in tumor tissues may be correlated with the prognosis of esophageal cancer patients (al. -Kasspooles, M., Moore, JH, Or
ringer, MB & Beer, DGInt. J. Cancer54 21 3-21
9 (1993); Krishnadath, KK et al., J. Pathol.182,
331-338 (1997); and Hishikawa, Y. et al., Oncology.
54, 342-347 (1997)). However, these markers have not been demonstrated to be effective in predicting susceptibility to chemotherapy or radiation therapy. To avoid the side effects of ineffective drugs on patients, establish diagnostic methods to determine the exact nature of each cancer, including the metastatic potential of the cancer and its sensitivity to chemotherapy and / or radiation. That is a major issue. Microarray analysis is very useful for solving these problems (DeRisi, J. et al. Nat. Genet. 14, 457-460 (199).
6) ； Eisen, MB, Spellman, PT., Brown, PO & Bots
tein, D. Proc. Natl. Acad. Sci. USA 95 14863-14868
(1998); and Golub, TR et al. Science 286, 531-53.
7 (1999)).

[0006]

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems of the prior art.
That is, the present invention should provide a method for screening a gene and a method for diagnosing a disease, for example, a novel method for diagnosing a cancer property such as susceptibility of cancer to an anticancer drug or radiation. It was an issue.

[0007]

[Means for Solving the Problems] In order to solve the above-mentioned problems, the inventors of the present invention have conducted diligent research. First, they assembled a cDNA microarray containing 9216 genes and received adjuvant chemotherapy after surgery. The expression profiles of pre-chemotherapy esophageal cancer and normal esophageal tissue samples from a number of patients were investigated. As a result, these patients were classified into group 1 (30 months or more), group 2 (12 to 24 months) and group 3 (12 months or less) according to the survival time after surgery. A multivariate statistical analysis of the expression profiles of these materials revealed 52 genes that are believed to correlate with patient prognosis (ie, reflect chemotherapy drug sensitivity or efficacy) Was extracted. These results establish a scoring system for predicting patient response to anti-cancer agents prior to treatment. The present invention has been completed based on these findings.

That is, according to the present invention, in the method for screening a gene involved in sensitivity to a drug or radiation, (1) a plurality of patients with diseases are treated with a first group of patients showing sensitivity to the drug or radiation. A step of dividing into a second patient group showing no susceptibility; (2) a step of analyzing gene expression profiles of the first patient group and the second patient group; (3) between the first and second patient groups And a step of selecting at least one gene whose expression level is significantly different by statistical test.

In the step (2) of analyzing the gene expression profiles of the first patient group and the second patient group, the patient-derived cDNA is preferably hybridized to the microarray in which the target gene is arranged. By analyzing the difference in expression of each target gene between the first and second patient groups.

The step (3) of selecting one or more genes whose expression levels are significantly different between the first and second patient groups by a statistical test is preferably the step of selecting the expression of each gene. The degree is classified into one of (1) increased in the patient sample, (2) decreased in the patient sample, (3) equally expressed in the patient sample and the control sample, and (4) non-expressed, and the first and second Compute the Mann-Whitney test amount corresponding to the number of patients with different classifications among
By selecting significant genes (preferably from the corresponding p-values) or by ordering the expression levels of each gene across the patient groups and corresponding to the number of intersections in the order relationship between the two patient groups. Calculate the Whitney calibration amount,
This is done by selecting significant genes (preferably from the corresponding p-values).

According to another aspect of the present invention, there is provided a method for determining the susceptibility of a patient with a disease to a drug or radiation, wherein (1) each of the genes selected by the screening method of the present invention described above, The score corresponding to the expression level is determined, (2) the expression level of the gene is measured in the test patient, and the score obtained in (1) is assigned based on the expression levels measured in (3) and (2), ( 4) A method for determining sensitivity to a drug or radiation is provided by summing scores assigned to all selected genes to obtain a total value, and (5) comparing a predetermined threshold value with the total value. To be done.

The disease in the present invention is preferably cancer, particularly preferably esophageal cancer, and the drug in the present invention is preferably an anticancer agent.

According to another aspect of the present invention, there is provided a gene analysis device for screening a gene involved in sensitivity to a drug or radiation, comprising: (1) a first patient group exhibiting sensitivity to a drug or radiation. Means for analyzing the gene expression profile of the second patient group which does not show susceptibility, and (2) One or more genes whose expression levels are significantly different between the first and second patient groups by statistical test. A gene analyzing apparatus is provided, which comprises:

The means for analyzing the expression profile of the gene is preferably such that each target gene between the first and second patient groups is hybridized by hybridizing the patient-derived cDNA to a microarray in which the target gene is arranged. It is a means to analyze the difference in the expression of.

The means for selecting one or more genes whose expression levels are significantly different between the first and second patient groups by a statistical test is preferably the expression level of each gene (1).
Up in patient sample, (2) down in patient sample,
(3) A means for classifying into any one of (1) a patient sample and a control sample, and (4) a non-expressing expression.
And between the second group of patients by calculating Mann Whitney's test amount corresponding to the number of patients whose classifications overlap and selecting significant genes (preferably from the corresponding p-values), or A means for ordering the expression levels of genes across patient groups and calculating a Mann-Whitney test amount corresponding to the number of intersections of the order relationships in the two patient groups, preferably from significant p-values. It is composed of a means for selecting.

[0016] According to another aspect of the present invention, there is provided a gene analysis apparatus for determining the susceptibility of a patient with a disease to a drug or radiation, which is (1) selected by the method of the present invention described above. For each of the genes, means for determining a score corresponding to the expression level, (2) means for measuring the expression level of the gene for a test patient, (3) (2)
Means for assigning the score obtained in (1) based on the expression level measured in step (4), means for summing the scores assigned for all selected genes, and (5) a predetermined threshold And a means for comparing the total value, and a gene analysis apparatus is provided.

According to yet another aspect of the present invention, a computer program product for screening genes involved in susceptibility to drugs or radiation, comprising:
(1) Code for inputting gene expression profiles of a first patient group showing sensitivity to a drug or radiation and a second patient group showing no sensitivity, (2) expression between the first and second patient groups A code that selects one or more genes having a significantly different degree by a statistical test; and (3) a computer-readable storage medium for accumulating the code,
Computer program products are provided, including:

Preferably, genes whose expression levels are significantly different between the first and second patient groups are 1 by statistical test.
The code to select more than one is either (1) the expression level of each gene is increased in the patient sample, (2) decreased in the patient sample, (3) equally expressed in the patient sample and the control sample, and (4) not expressed. Mann-Whitney test quantities corresponding to the number of patients with overlapping classification between the first and second patient groups are calculated, and significant genes (preferably from the corresponding p-values) are calculated. By selecting, or by ordering the expression levels of each gene across patient groups, and calculating Mann Whitney's test amount corresponding to the number of intersections of the order relationships in the two patient groups (preferably corresponding It consists of a code that selects significant genes (from p-values).

According to another aspect of the present invention, there is provided a computer program product for determining the susceptibility of a patient with a disease to a drug or radiation, which is (1) selected by the method of the present invention described above. Code for determining the score corresponding to the expression level of each gene,
(2) A code for measuring the expression level of the gene in a test patient, (3) A code for assigning the score obtained in (1) based on the expression level measured in (2), (4) Assignment for all selected genes A computer program product is provided which includes a code for summing the points obtained to obtain a total value, and (5) a code for comparing the total value with a predetermined threshold value.

[0020]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described in more detail. The method for screening a gene involved in sensitivity to a drug or radiation according to the present invention includes (1) a plurality of patients with a disease, a first patient group showing sensitivity to the drug or radiation and a second patient not showing sensitivity. Grouping; (2) analyzing gene expression profiles of the first patient group and the second patient group; (3) the degree of expression is significantly different between the first and second patient groups Selecting one or more genes by a statistical test.

In the screening method of the present invention, first, a plurality of patients with diseases are divided into a first patient group showing sensitivity to drugs or radiation and a second patient group not showing sensitivity. It should be noted that the patients may be classified into three groups including a third patient group having a different sensitivity, which shows intermediate sensitivity between the first patient group and the second patient group.

In the present invention, next, the gene expression profiles of the first patient group and the second patient group are analyzed. A preferable method for analyzing the gene expression profile of each patient is to hybridize the cDNA derived from the patient to a microarray in which the target gene is arranged, so that each of the groups between the first and second patient groups can be hybridized. The method of analyzing the difference in the expression of the target gene of E. coli. The microarray fabrication method includes (1) very high density D
A method of arranging NA on the base and analyzing it,
(2) There is a method of directly synthesizing DNA on a substrate. Commercially available spotters can be used to arrange the DNA on the substrate at a very high density. Briefly, one kind of cDNA synthesized is put in each of a large number of holes (for example, 384 holes), the solution is sucked up with a pin (needle), and arranged in order on a glass plate. The principle is to go. The synthesis of DNA on a substrate is based on the principle that a hydroxyl group is formed by using light and an elongation reaction of DNA is performed on the substrate.

In order to analyze expression information using a microarray, first, a large number (for example, tens of thousands of kinds) of genes (cDNA sequences etc.) are arranged on a slide glass by the above method. By hybridizing a labeled sample to the cDNA immobilized on the slide glass, it is possible to examine the increase or decrease in the gene expression level. More specifically, mRNA is extracted from a patient sample, and this is labeled with a red dye to synthesize cDNA. MRNA is also extracted from the control sample, and while it is labeled with a green dye, cDNA is synthesized. By mixing these labeled cDNAs and hybridizing them with the gene on the slide glass, it is possible to detect the ratio of the gene expression level between the respective patient groups. That is, for each spot on the slide glass, the red spot indicates that it is specifically expressed in the patient sample, the green spot indicates that it is specifically expressed in the control sample, the yellow spot ( (Red and green rays become yellow when mixed) indicates that high levels are expressed in both the first and second patient groups, and non-illuminated spots are in the first and second patient groups. It is shown that the expression level is low or not expressed in the patient group.

The degree of gene expression is determined by RT-PCR.
It can be analyzed by the method. The reliability of data obtained by microarray analysis can also be evaluated and confirmed by the RT-PCR method.

Further, in order to analyze the expression profile of tens of thousands of genes by microarray analysis, a large amount (usually 1 to several μg) of mRNA is required, but a large amount of sample can be collected from a patient. It can be difficult. In this case, mRNA can be amplified and used by the RNA amplification method using T7 RNA polymerase. mRNA
When synthesizing the cDNA from T7R
By incorporating a sequence corresponding to the recognition sequence of NA polymerase, the double-stranded cDNA was used as a template for T7RN.
By using A polymerase, one molecule of cDNA
A (corresponding to 1 molecule of mRNA) to about 100 molecules of R
NA is synthesized. By repeating the same experimental procedure based on this RNA, one molecule of mRNA is finally obtained.
It is possible to obtain about 1 million RNA molecules.

Further, when performing microarray analysis, it may be necessary to collect a specific sample from a tissue section. For example, when observing a cancer tissue under a microscope, normal cells and cancer cells are mixed, but it is possible to extract only a population of cancer cells from such a tissue under a microscope and obtain its expression information. is there. Laser capture microdissection (LC) is a means of extracting only a population of cancer cells.
M) method. In the LCM method, (1) attach a transfer film (a thin film for fixing and lifting the target tissue part) onto a tissue piece placed on a glass slide, (2) specify the part to be taken out under a microscope with a computer (3) A method in which the substrate of the film is melted by irradiating only that portion with laser light, the target tissue portion is fixed to the film, and lifted.

In the method of the present invention, one or more genes whose expression levels are significantly different between the first and second patient groups are selected by a statistical test. In this selection step, the expression level of each gene is preferably (1) increased in the patient sample, (2) decreased in the patient sample, (3) equally expressed in the patient sample and the control sample, and (4) not expressed. Between any one of the first and second patient groups,
Either by calculating the Mann-Whitney test amount corresponding to the number of patients with overlapping classification and selecting a significant gene from the corresponding p-values, or by ordering the expression ratio of each gene across the patient group, This is done by calculating Mann-Whitney's test amount corresponding to the number of intersections in the order relation in the patient group and selecting significant genes from the corresponding p-values.

The Mann-Whitney test aims to detect the difference in the distribution of cases between two groups. First, all cases are ordered according to the evaluation value of each case, and the cases were originally included. This is an evaluation method in statistical mathematics that adds up the order overlap between groups. If the evaluation values of the cases between the two groups do not overlap at all, the Mann Whitney u value becomes 0, and the significant difference between the two groups becomes maximum. This method is recommended as an example in this study. In this case, focusing on each gene individually, consider that the group is a patient group and the evaluation value is the expression level or expression ratio of that gene in each patient. To do.

The present invention further provides a method of determining the susceptibility of a patient with a disease to a drug or radiation.
In this determination method, (1) a score corresponding to the expression level of each of the genes selected by the method of the present invention described above is determined, and (2) the expression level of the gene is measured in a test patient, 3) The score obtained in (1) was assigned based on the expression level measured in (2), (4) the assigned scores for all selected genes were summed to obtain a total value, and (5) predetermined It is characterized in that the sensitivity to a drug or radiation is determined by comparing a threshold value and the total value.

In the method of the present study, for each of the genes selected by the method of the present invention described above, a score is determined so that the total score value produces a difference between two patient groups. Preferably, a multivariate analysis is used to determine the score that maximizes the difference in distribution between the two patient groups, or to multiply the sign by a constant multiple depending on the magnitude relationship between the two patient groups in the distribution of gene expression ratios. Use the score or the method of.

According to the present invention, there is provided a gene analysis device for screening a gene involved in sensitivity to a drug or radiation, which is (1) not sensitive to a first patient group showing sensitivity to a drug or radiation. Means for analyzing a gene expression profile of the second patient group;
(2) Means for selecting one or more genes whose expression levels are significantly different between the first and second patient groups by a statistical test, and a gene analysis device are provided. .

First showing sensitivity to drugs or radiation
The means for analyzing the gene expression profile of the second patient group which is not sensitive to the first patient group is a device for performing hybridization using a microarray, and a first device which shows sensitivity to a drug or radiation provided by the device.
And a computer readable storage medium for accumulating a code that provides an expression profile of a gene in a second patient group and a non-sensitive patient group. In addition, the means for selecting one or more genes whose expression levels are significantly different between the first and second patient groups by a statistical test has a significant expression level between the first and second patient groups. It can be constituted by a computer-readable storage medium for accumulating codes for selecting one or more different genes according to a statistical test. The readable storage medium can be connected to a computer to construct a gene analysis device.

Further, according to the present invention, there is provided a gene analysis device for determining the susceptibility of a patient with a disease to a drug or radiation, comprising: (1) each of the genes selected by the above-mentioned method of the present invention. The means for determining the score corresponding to the expression level, (2) the means for measuring the expression level of the gene in the test patient, (3) the score obtained in (1) based on the expression level measured in (2) Means for assigning
(4) means for summing the points assigned to all the selected genes to obtain a total value, and (5) means for comparing a predetermined threshold value with the total value. A gene analysis device is provided. The above means can be constituted by a computer-readable storage medium in which codes for executing the respective steps are accumulated.

The computer system constituting the gene analysis device of the present invention can be constituted by a computer-readable storage medium in which codes for executing the steps of the present invention described above are accumulated. Computer systems typically include monitors, cabinets, keyboards, mice, and the like. The cabinet houses a CD-ROM drive and / or hard drive that can be used to store and retrieve software programs, including computer code, which incorporates the steps of the present invention. The computer-readable medium includes a CD-ROM and the like, but other computer-readable medium (FD, DRAM, hard drive, flash memory, tape, etc.) can also be used. The cabinet also contains the usual computer components (eg, processor, memory, etc.).

A typical computer system that can be used with the present invention includes a monitor and keyboard, and
Central processor, system memory, input / output (I /
O) A controller, a display adapter, a removable disk, a fixed disk (for example, an internal hard drive, etc.), a network interface, etc. may be provided.

[0036]

The present invention will be described in more detail with reference to the following examples, but the present invention is not limited to the examples. (A) Method (1) Preparation of sample All samples were immediately frozen in liquid nitrogen immediately after excision of the sample using Trizol reagent (Gibco BRL), and Rnase-free D
Digested with nase I (Nippon Gene) according to the instruction manual. T7-based RNA amplification was performed with 5 μg of each total RNA
Was used as the starting material and was carried out according to the previously reported method.
Two rounds of amplification were sufficient for the synthesis of the cDNA probe.
The amount of amplified antisense RNA (aRNA) was measured by a spectrophotometer, and the quality of each aRNA was confirmed by agarose gel electrophoresis.

(2) Construction of cDNA microarray The target DNA is the UniGene human database (National Cen
ter for Biotechnology Information)
It was prepared by the RT-PCR amplification method using specific primers. The amplified fragment was c with the repeated sequence removed.
It is composed of the 3-prime region of the DNA-specific sequence.
The length of the PCR product was in the range of 0.2 to 1.1 kb.
CDNA for PCR reaction can be obtained from human brain, heart, liver,
Skeletal muscle, spleen, placenta, small intestine, testis, fetal liver, fetal brain,
Normal poly (A) RNA derived from fetal kidney and fetal lung
(Clontech) pool from oligo dT primer 12-
18 (Gibco BRL) and Superscript ll reverse transcriptase (Gibco BRL)
(BRL). GeneAmp PCR System 9700 (Pe
rKain-Elmer Applied Biosystems)
PCR reaction was performed using Taq DNA polymerase (Takara) according to the standard operating method. After amplification, PCR product is Arraylt
96-well PCR purification kit (TeleChem Internatio
nal, Inc.) according to the instruction manual.
The quantity and quality of each PCR product was confirmed by agarose gel electrophoresis. For spotting onto glass slides, Microarray Crosslinking Reagent D (Amersham Ph
armacia Biotech) at a final concentration of 100 fmol / μl
Added to product. The cDNA array is a dual format Generati on a type 7 glass slide (Amersham Pharmacia Biotech).
on lll Microarray Spotter (Amersham Pharmacia Biot
ech) was used for spotting.

(3) Hybridization and image formation Preparation of a cDNA probe from aRNA was performed by Luo, L. et al.
It was performed according to the method described in al. Nat. Med. 5, 117-122 (1999). 2.5 μg of each aRNA from esophageal cancer tissue
Labeled with Cy5-dCTP (Amersham Pharmacia Biotech).
As a control probe, aRNA (Invitrogen) from the total RNA pool of normal esophagus was treated with Cy3-dCTP (Amersham Pharmacia).
Biotech). The labeled probe was added to the microarray hybridization solution version ll (Amersham
Pharmacia Biotech) and mixed with formamide (Si
The mixture was mixed so that the final concentration of gma) was 50%, and hybridization was carried out at 42 ° C. for 14 hours. Wash glass slides with 2 x SSC, 0.1% SDS for 10 minutes at 55 ° C,
Washed in 0.2 × SSC, 0.1% SDS at 55 ° C. for 10 minutes and 0.1 × SSC at room temperature for 1 minute. Wash the washed slides with the Generation lll Array Scanner (Amersham Pharmac
ia Biotech). The intensity of each hybridization signal is determined by commercially available software (ArrayVision;
Photometric evaluation by Amersham Pharmacia Biotech).

(4) Data analysis Noise was suppressed by statistical analysis and majority voting, and accurate expression values were obtained. A median filter was also applied to suppress the error. Normalization was performed using 60 housekeeping genes to compare slides (Ermo
laeva, O. et al. Nat. Genet. 20, 19-23 (1998)).
By adjusting the parameters so that the latter expression levels were equal, the amount of mRNA from each patient could be normalized. Next, double the spots on each slide,
Averaged to final Cy5 / Cy3 ratio. Reliable analysis is not possible if the signal strength is insufficient, so the cut-off value for each expression level was calculated automatically using the correlation coefficient.

(5) Selection of Genes for Discrimination The expression ratio of each gene was ordered in the whole patient group, and Mann Whitney's test amount corresponding to the number of intersections of the order relations in the two patient groups was calculated, The corresponding p-value was calculated. This calculation was performed for each of all genes, and was performed by selecting genes having a significant difference of p value of less than 1%. These genes have the property of exhibiting different expression ratios between two patient groups, and by combining one or more of them, they can be markers that can distinguish drug sensitivity and radiosensitivity.

(6) Calculation of "prediction score" using genes for identification In order to predict the drug sensitivity of tumor cells of a patient to be diagnosed, the expression ratio of genes selected by the above-mentioned method of the present invention is calculated. Measured with the observation system using the microarray described above, by accumulating a constant multiple of the predetermined sign according to each gene, and by adding up for all selected genes, the drug sensitivity of the tumor cells of the patient to be diagnosed The predictive score was determined. The sign was determined to be positive or negative depending on whether the expression ratio of each gene had a positive correlation or a negative correlation with drug sensitivity or radiation sensitivity between two known patient groups. The constant multiple is 10 in this example
And

The predictive score was calculated by the following formula. During _{X i = 10Σ k S k log} 2 (r ik) formula, r _ik represents the ratio of expression in the patient i gene k, S _k is the sign.

The sign was determined by the following method:
The average expression ratio of the relevant gene was calculated for each of the two patient groups: average for patient group A = Σ _A log ₂ (r _ik ) / number of patients in patient group A, average for patient group B = Σ _B log ₂ ( r _ik ) / number of patients in patient group B. Next, if the average of patient group A> the average of patient group B, then S _k
= + 1, average of patient group B < _Sk = + 1 if average of patient group A
The sign for each of the above genes was determined.

(7) Semi-quantitative RT-PCR Using total RNA (2 μg) from each tissue sample and aRNA (2 μg) from each total RNA, oligo-d (T) 1 was prepared.
Reverse transcription was performed using 2-18 primers and random hexamers according to the manufacturer's instructions. Synthetic cDNA
The mixture of A was diluted with distilled water and 2 μl of each mixture was used as template. For the PCR reaction, the same reverse primer prepared for the construction of the target DNA was used and the forward primer was newly constructed (product size 100-200 bp). The expression of glyceraldehyde-3-phosphate dehydrogenase (G3PDH) served as a control. The PCR reaction was optimized for cycle number so that the product intensity was within the linear phase of amplification.

(B) Results (1) Classification of Patients Table 1 summarizes the clinicopathologic characteristics of 20 patients who underwent surgery for esophageal cancer at Kurume University School of Medicine between 1989 and 1998.

[0046]

[Table 1]

All cases were diagnosed with squamous cell carcinoma (SCC) and all patients were treated following surgery following the same adjuvant chemotherapy protocol: cisplatin (CDDP; 70 mg /
m ² ) and 5-fluorouracil (5-FU; 700 mg / m ² ). Microscopic (R1) or macroscopic (R2) residual tumor was confirmed in 15 patients. Twenty patients were divided into three groups based on post-surgical survival, a parameter likely to reflect responsiveness to anti-cancer drugs. The three groups showed no significant differences in clinicopathologic features.

(2) Expression Profile The 20 primary SSCs originated from the same organ, but the gene expression patterns analyzed by the present inventors' microarrays were different as shown in FIG. . The expression profiles of groups 1 and 3 were compared. The first group of patients was considered to respond well to anti-cancer treatments, and group 3 was considered to respond poorly. By searching the array,
Genes that were either universally down-regulated or up-regulated in Group 1 or Group 3 were identified. Genes most likely to be associated with patient survival include
Selection was made according to our selection algorithm (see Methods).

(3) Evaluation of microarray data In order to confirm the difference in expression shown by microarray analysis, semi-quantitative RT-PCR was performed on 17 out of 52 genes for identification, and microarray analysis was performed on the microarray. It was confirmed that the signals corresponded to the RT-PCR results (Fig. 2b). However, since only a limited amount of total RNA was available in this analysis, sufficient RNA was obtained by performing T7-based RNA amplification and used as a probe for microarray analysis (Van Gelde
r, RN et al. Proc. Nat1. Acad. Sci. USA 87, 1663
-1667 (1990)). Therefore, it was necessary to further investigate whether the amplified RNA actually reflected the proportion of the original RNA extracted from each tumor sample. Therefore, pre-amplified total RNA was used as a template for subsequent RT-PCR experiments. As shown in FIG. 2c, in 13 of the 14 cases forming groups 1 and 3, A4
The results for the 284 gene were in agreement with the results of RT-PCR using the amplified RNA and also with the results of microarray analysis. By comparing the ratio of expression levels of 17 selected genes in normal and cancerous material from these 14 patients (total 238
Example), the ratio was found to be maintained in almost 80% of the cases tested.

(4) Calculation of "predictive score" based on the expression level of selected genes A method for evaluating the expression of all 52 genes selected from a microarray in order to distinguish drug-sensitive groups of tumors from insensitive groups Was adopted. Each gene was labeled into four categories (up, down, down, according to the comparison of expression in cancer tissue and control normal tissue).
(Equal or no expression). The classification pattern of all 52 genes according to the characteristics in each tumor is summarized in FIG. To investigate the possibility that changes in the expression of these genes could be clinically applicable to predict drug sensitivity, we proposed a numerical scoring system called the Drug Responsive Score (DRS). The predictive score is defined as the sum of the points calculated by the function depending on the expression of each gene. The score set is optimized using multivariate analysis or assigned a constant multiple of the sign determined according to the magnitude relationship between the two patient groups of gene expression,
The distribution of DRS in drug-sensitive patients was made different from that in drug-insensitive patients (method). The DRS for each patient was then calculated (Table 2). Group 1 cancer tissues (longer survivors) showed higher scores (14-269). These scores significantly correlate with the results for each patient (Figure 4).

[0051]

[Table 2]

(C) Discussion Due to technical limitations, previously, genetic characterization of cancer cells was mainly based on genetic and / or epigenetic mutations of one or at most several genes. However,
The recent development of cDNA microarray or cDNA chip technology, which is a method for monitoring gene expression at high speed, has made it possible to analyze the expression of thousands of genes at once (DeRisi, J. et al. Nat. . Genet. 14,
457-460 (1996); Schena, M. et al. Science 270, 467-
470 (1995); and Khan, J. et al. Cancer Res. 58, 500.
9-501 3 (1998)). As a result, it has become possible to provide a novel taxonomy of cancer types based on changes in the expression of multiple genes in tumor tissue (Golub, TR et al. S).
cience 286, 531-537 (1999); and Alizadeh, AAet a
l. Nature 403, 503-511 (2000)). In this example, microarrays as an effective tool for solving one of the major problems clinicians face in treating cancer patients (ie, the inability to predict individual patient responsiveness to anti-cancer agents). It has been shown that can be used.

A total of 20 patients with esophageal cancer used in this example were in the clinically advanced stage (TNM stage III-I).
V). Such patients usually have a poor prognosis and very few achieve survival more than 5 years post surgery. For the 8 patients who survived the maximum period, CDDP or
It is considered that the anti-cancer drug treatment after surgery with 5-FU was effective in destroying the remaining cancer cells. Although the biological properties of primary cancer cells in tumors resected by surgery or biopsy do not necessarily reflect the exact composition of the remaining cancer cells, the efficacy of anticancer agents in the resected cancer material prior to treatment is not Being able to evaluate makes it possible to avoid unnecessary treatment and side effects. The main advantage of the described algorithm for generating a predictive score is that after selecting genes that may affect drug response, the score of one or more genes assigned by the algorithm is added to the patient. The point is that it is possible to comprehensively judge the condition, drug sensitivity, and radiation sensitivity. This is because the in-vivo state has a complicated expression pattern, and therefore it is necessary to combine one or more genes for diagnosis. Second, our algorithm determines expression categories by considering the reliability of each spot. Accurate analysis requires the detection of a reliable lower limit of the signal, since microarrays can generate noise due to non-specific hybridization or specs that fluctuate the signal. If each signal did not fluctuate due to noise, gene expression in normal and tumor cells would be well correlated. Our algorithm automatically calculates a lower bound on the reliability with which the correlation disappears.

The present data obtained from the 26 patients analyzed in this example show the genes consistent with the previous reports of gene expression in esophageal cancer, and the prognosis when adjuvant chemotherapy during surgery is applied. It was also possible to identify the reported gene group to be determined. For example, thymidylate synt
hase, dihydropyrimidine dehydrogenase, thymidineph
Although osphorylase and an enzyme involved in the metabolic system of the anticancer drug pyrimidine fluoride were identified, all of them are related to the anticancer drug sensitivity (Salonga, D. et al. Clin. Cancer Res., 6: 1322-
1327 (2000)). Altered expression of MDR1 / P-glycoprotein is associated with anticancer drug sensitivity in advanced ovarian cancer (Baekelandt, M. et.
al. Anticancer Res., 20: 1061-1067, 2000). MRP is a protein capable of transporting many anticancer drugs to the outside of cells (Borst, P. et al. J. Natl. Cancer Inst., 92:
1295-1302, 2000). Taken together, these findings suggest that the results of microarray hybridization are consistent with other approaches. this time,
It was possible to identify a large number of known and unknown genes and pathways that may play important roles in the mechanism of drug resistance and in the medicine and biology of esophageal cancer. These products are likely to be effective targets for the development of new cancer therapies.

[0055]

INDUSTRIAL APPLICABILITY According to the present invention, patients can be freed from ineffective drugs or side effects of radiation treatment, and finally, it becomes possible to select an individualized treatment most suitable for each patient. .

[Brief description of drawings]

FIG. 1 shows an image of a microarray. CDNA derived from cancer tissue was labeled with Cy5-CTP (green), and as a control probe, cD from a pool of normal esophagus
NA was labeled with Cy3-dCTP (red). Differences in gene expression between patients were determined by comparing the Cy5 signal from each patient with its control (Cy3) signal. Images of the same region (including 384 genes) from two patients are shown. The upper image is a patient who survived 98 months after surgery, and the lower image is a patient who died 4 months after surgery. Expanded spots are examples of genes that showed different expression changes between patients.

FIG. 2 shows the expression schema of the selected 52 genes. The color tone was calculated according to the Cy5 / Cy3 expression ratio and the signal intensity. The importance of each gene that distinguishes between patient group 1 and patient group 3 is shown by the Mann-Whitney u value. The sign of each gene was determined by whether the gene had a positive or negative correlation with drug sensitivity.

FIG. 3 shows verification of the reliability of microarray data. (A) shows the experiment of the gene of A4284. The expression ratio of each tumor on the array is shown in color, the amplified RNA is verified by a semi-quantitative RT-PCR experiment, and the total RNA is verified by a semi-quantitative RT-PCR experiment. . (B) shows an experiment of the GAPDH gene. The expression ratio of each tumor on the array is shown in color, the amplified RNA is verified by a semi-quantitative RT-PCR experiment, and the total RNA is verified by a semi-quantitative RT-PCR experiment. .

FIG. 4 shows identification of groups 1-3 based on prediction scores.

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Tatsuhiko Tsunoda             3-43-14 Takada Higashi, Kohoku Ward, Yokohama City, Kanagawa Prefecture F-term (reference) 4B024 AA12 CA02 CA04 HA14 HA19                 4B029 AA07 AA23 BB20 CC03 CC08                       FA12                 4B063 QA12 QA13 QA19 QQ43 QQ58                       QR32 QR55 QS34

Claims (15)

[Claims] 1. A method for screening a gene involved in sensitivity to a drug or radiation, comprising: (1) a plurality of patients with a disease, a first group of patients showing sensitivity to the drug or radiation, and a second group showing no sensitivity. (2) analyzing the gene expression profiles of the first patient group and the second patient group; (3) the degree of expression between the first and second patient groups is significant And a step of selecting at least one different gene by a statistical test. 2. In the step (2) of analyzing the gene expression profiles of the first patient group and the second patient group, the patient-derived cDNA is hybridized to the microarray in which the target gene is arranged. The method according to claim 1, wherein the difference in the expression of each target gene between the first and second patient groups is analyzed by. 3. The step of selecting one or more genes, whose expression levels are significantly different between the first and second patient groups, in the step (3) by a statistical test to determine the expression level of each gene ( The first and second patient groups are classified into one of 1) increased in patient sample, (2) decreased in patient sample, (3) equally expressed in patient sample and control sample, and (4) unexpressed. By calculating the Mann-Whitney test amount corresponding to the number of patients whose classifications overlap and selecting significant genes, or by ordering the expression levels of each gene across the patient group The method according to claim 1 or 2, which is carried out by calculating a Mann-Whitney test amount corresponding to the number of intersections of order relations in a group and selecting a significant gene. 4. The disease is cancer and the drug is an anticancer drug,
The screening method according to claim 1. 5. The screening method according to claim 1, wherein the disease is esophageal cancer. 6. A method for determining the susceptibility of a patient with a disease to a drug or radiation, wherein (1) the expression level of each of the genes selected by the method according to any one of claims 1 to 5. Determine the score corresponding to
(2) measuring the expression level of the gene in a test patient,
(3) The score obtained in (1) is assigned based on the expression level measured in (2), (4) the assigned scores for all selected genes are summed to obtain a total value, and (5) predetermined A method for determining the sensitivity to a drug or radiation by comparing the threshold value and the total value. 7. The disease is cancer and the drug is an anticancer drug,
A method of determining sensitivity according to claim 6. 8. The method for determining susceptibility according to claim 6, wherein the disease is esophageal cancer. 9. A gene analysis device for screening a gene involved in drug or radiation sensitivity, comprising: (1) a first patient group showing a drug or radiation sensitivity and a second patient not showing a sensitivity. A means for analyzing the expression profile of the genes of the group, and (2) means for selecting one or more genes by a statistical test that have a significantly different expression level between the first and second patient groups A gene analysis device characterized by: 10. The means for analyzing the expression profile of a gene comprises hybridizing a cDNA derived from a patient to a microarray in which the target gene is arranged to hybridize each target gene between the first and second patient groups. 10. The gene analysis device according to claim 9, which is a means for analyzing a difference in expression of the gene. 11. A means for selecting one or more genes whose expression level is significantly different between the first and second patient groups by a statistical test is such that the expression level of each gene is (1) in a patient sample. Up, (2) down in patient sample, (3) equally expressed in patient and control samples, (4)
Means for classifying into any one of the unexpressed, by calculating Mann Whitney's test amount corresponding to the number of patients whose classifications overlap between the first and second patient groups, and selecting a significant gene Or, by ordering the expression level of each gene in the entire patient group, by calculating Mann Whitney's test amount corresponding to the number of intersections of the order relations in the two patient groups, by selecting a significant gene The gene analysis device according to claim 10, which is configured. 12. A gene analysis device for determining the susceptibility of a patient with a disease to a drug or radiation, which comprises (1) a gene selected by the method according to any one of claims 1 to 5. For each, a means for determining the score corresponding to the expression level, (2) a means for measuring the expression level of the gene in a test patient, (3) obtained in (1) based on the expression level measured in (2) Means to assign points,
(4) means for summing the points assigned to all the selected genes to obtain a total value, and (5) means for comparing a predetermined threshold value with the total value. Gene analysis device. 13. A computer program product for screening a gene involved in drug or radiation sensitivity, comprising: (1) a first patient group showing a drug or radiation sensitivity and a second patient not showing a sensitivity. A code for entering the expression profile of a group of genes,
(2) A code that selects one or more genes whose expression levels are significantly different between the first and second patient groups by a statistical test; and (3) a computer-readable storage medium for accumulating the code. Computer program products, including ,. 14. (2) A code for selecting one or more genes whose expression levels are significantly different between the first and second patient groups by a statistical test determines the expression level of each gene (1).
Up in patient sample, (2) down in patient sample,
(3) Mann Whitney corresponding to the number of patients having the same classification in the patient sample and the control sample, (4) classified into any one of the non-expression, and having the classification overlapping between the first and second patient groups By calculating the test amount of and selecting the significant genes, or by ordering the expression levels of each gene across the patient groups and corresponding to the number of intersections of the order relations in the two patient groups of Mann Whitney 2. A code which calculates a test amount and selects a significant gene.
A computer program product according to item 3. 15. A computer program product for determining the susceptibility of a patient with a disease to a drug or radiation, which comprises (1) a gene selected by the method according to any one of claims 1 to 5. For each, a code for determining the score corresponding to the expression level, (2) Code for measuring the expression level of the gene in a test patient, (3)
A code that assigns the score obtained in (1) based on the expression level measured in (2), (4) a code that sums up the assigned scores for all selected genes, and (5) predetermines A computer program product comprising a code that compares a threshold value that has been set to the total value.