JPWO2009084740A1 - Methods and compositions for predicting postoperative recurrence in patients with lung adenocarcinoma - Google Patents

Abstract

The present invention provides methods for predicting recurrence of lung adenocarcinoma after surgical resection of lung cancer in lung adenocarcinoma patients using specific gene sets, and compositions for use in these methods.

Description

  The present invention relates to methods and compositions for predicting postoperative recurrence in lung adenocarcinoma patients.

Lung cancer is the leading cause of cancer-related death in developed countries. Lung cancer is roughly classified into small cell cancer and non-small cell cancer, and non-small cell cancer is mainly classified into adenocarcinoma, squamous cell carcinoma, and large cell carcinoma. Among non-small cell lung cancers, lung adenocarcinoma accounts for about half of non-small cell cancers and is known to occur most frequently in Japan. In addition, the 10-year survival rate of patients with non-small cell carcinoma is very low, about 10%, and about 25-30% of those patients have only undergone surgical resection in stage I patients. About 35-50% have relapsed within 5 years.
In recent years, with the advancement of technologies such as microarrays and mass spectrometry, new technical fields such as bioinformatics and proteomics using these have been developed. In cancer research, comprehensive analysis of gene expression using such techniques is actively performed for the purpose of diagnosis and treatment of cancer.
Regarding lung cancer, research is being conducted to investigate the correlation between classification of lung cancer in patients with lung cancer, prediction of metastasis and prediction of recurrence, and gene expression patterns and clinical parameters (Non-patent Document 1).
To date, the present inventors have also conducted a comprehensive analysis of gene expression patterns mainly in lung cancer, and analyzed expression patterns related to metastasis and prognosis prediction (Non-patent Document 2). In addition, the present inventors recently classified lung adenocarcinoma into TRU-a type, TRU-b type and non-TRU type, and a method for predicting the postoperative prognosis of patients with lung adenocarcinoma based on this classification And a composition are disclosed (Patent Document 1).
With regard to lung adenocarcinoma, yet another group has a gene set for identifying lung adenocarcinoma, a gene set for determining pleural infiltration of lung adenocarcinoma, and a gene set for predicting recurrence of lung adenocarcinoma (Patent Documents 2 to 4). Among these, Patent Literature 4 describes a method for predicting postoperative recurrence of lung adenocarcinoma and a set of 45 genes, and the discrimination rate of lung adenocarcinoma by this gene set is 50% or more. It is described that there is. Non-Patent Document 3 discloses the prediction of recurrence of lung adenocarcinoma by 54 gene sets, which also describes the accuracy of recurrence by a conventional method for determining the prognosis of lung adenocarcinoma patients. The difficulty of prediction is described.
International Publication WO2007 / 043418 JP2007-6791A JP2007-6792A JP2007-135466 D. G. Beer et al., Nature Medicine 2002, 8 (8): 816-824. Shuta Hamada and Takahashi Takahashi, “New development of diagnosis and treatment of lung cancer based on expression profile analysis” Experimental Medicine 2004, 22 (14): 45-50, Yodosha (Japan) J. et al. E. Larsen et al., Clinical Cancer Research 2007, 13: 2946-2954.

Some cancer patients have a very poor prognosis due to metastasis or recurrence after surgical resection of the tumor, and die within 5 years. Conversely, some patients have no metastasis or recurrence, that is, have a good prognosis. The inventors of the present invention have noted that there is no method that enables prediction of recurrence with high accuracy in lung cancer prognosis prediction in lung adenocarcinoma patients, which is relatively common among lung cancer patients in Japan. Predicting recurrence after surgery for lung cancer is very difficult because it involves various factors. If the recurrence of patients with lung adenocarcinoma can be predicted with extremely high accuracy, it is possible to identify a group of patients who need treatment in the group of patients currently undergoing follow-up without post-treatment after surgery Therefore, the prognosis may be improved.
In a conventional method for predicting recurrence of lung adenocarcinoma, a gene set capable of predicting recurrence risk is identified using leave-one-out cross-validation. At this time, if the determination criterion that determination is impossible is used, the apparent discrimination rate and the true discrimination rate may be extremely different depending on the percentage of cases where the determination is not possible. In addition, leave-one-out cross-validation is an effective method for evaluating the accuracy of a prediction model with few repetitive calculations when the data to be analyzed is very small. It is preferable to perform a more accurate evaluation of the model by increasing and performing repeated calculation incorporating the 10-fold cross-validation method and random division.
Furthermore, when comprehensive gene expression profiling of cancer cells using microarrays is associated with clinical outcomes, reproducibility and reliability may be questioned due to sampling methods, platform differences, statistical analysis differences, etc. It has been pointed out (for example, Shigeyuki Matsui, Experimental Medicine 2007, 25 (17): 72-78 (Special Issue) Yodosha (Japan)).
An object of the present invention is to provide a method and composition for predicting recurrence of lung adenocarcinoma with extremely high accuracy when predicting recurrence of lung adenocarcinoma after surgical resection of lung cancer in patients with lung adenocarcinoma. It is.
The present inventors are a method for presenting a determination of either high recurrence risk or low recurrence risk to all cases, which is different from the determination method incapable of determination, and is different from the conventional recurrence prediction gene set, 82 By using a new gene set consisting of different types of genes for the determination, it was possible to predict the presence or absence of recurrence with an accuracy of more than 70% in the evaluation model, thereby completing the present invention.
In summary, the present invention encompasses the following features.
In the first aspect, the present invention is a method for predicting recurrence of lung adenocarcinoma after surgical resection of lung cancer in a lung adenocarcinoma patient, comprising a gene comprising each nucleotide sequence of SEQ ID NOs: 1 to 82 Measuring the expression of a gene set of 2 to 82 genes selected from the group in a biological sample from said patient, wherein SEQ ID NOs: 1-5, 8-9, 12, 14-15 , 17, 19-21, 23-28, 30-33, 35-59, 61-68, 70-82, the expression of the gene is relatively high compared to post-operative relapse-free cases Or the expression of a gene containing any one of the nucleotide sequences of SEQ ID NOs: 6-7, 10-11, 13, 16, 18, 22, 29, 34, 60, 69 is relative to that of a case without recurrence after surgery. Low, lung adenocarcinoma Recurrence and wherein the predicting high, provides the above method.
According to an embodiment of the present invention, the number of genes in the gene set is 5 or more and 82 or less.
According to another embodiment of the present invention, the expression of the gene set is measured by measuring the presence or amount of a nucleic acid or protein corresponding to the gene.
According to another embodiment of the present invention, the measurement of the expression of the gene set is performed by a hybridization method or a quantitative PCR method.
According to an embodiment of the present invention, the hybridization method is a microarray method or a blot method.
According to an embodiment of the present invention, the hybridization method is performed using a nucleic acid that hybridizes with an RNA transcript, cRNA, aRNA or cDNA derived from the gene.
According to another embodiment of the present invention, the measurement of the expression of the gene set is performed by an immunological method.
According to an embodiment of the present invention, the immunological method is performed using an antibody against a protein encoded by the gene or a fragment thereof.
In the second aspect, the present invention provides a composition for predicting recurrence of lung adenocarcinoma after surgical resection of lung cancer in a patient with lung adenocarcinoma, wherein the composition is any one of SEQ ID NOs: 1 to 82. Comprising a plurality of nucleic acids that make it possible to measure the expression of two or more different genes comprising the nucleotide sequence of:
(1) a nucleotide sequence represented by SEQ ID NOs: 1 to 82, a nucleotide sequence complementary thereto, or a nucleic acid having a partial sequence of 15 bases to less than the total number of bases in these nucleotide sequences, and (2) SEQ ID NO: A nucleic acid that hybridizes with a nucleic acid comprising the nucleotide sequence shown in 1-82 or a nucleotide sequence complementary thereto, or a nucleic acid having a partial sequence of less than the total number of bases from 15 consecutive bases of the hybridizing nucleic acid,
There is provided a composition as described above, characterized in that it is selected from the group consisting of:
In an embodiment of the invention, the composition is in the form of a kit, microarray or primer set.
In another embodiment of the invention, the nucleic acid allows measurement of the expression of 5 or more and 82 or less genes.
In another embodiment of the invention, the composition is for use in the method of the invention.
In the third aspect, the present invention provides a composition for predicting recurrence of lung adenocarcinoma after surgical resection of lung cancer in a lung adenocarcinoma patient, wherein the composition is any one of SEQ ID NOs: 1 to 82. A plurality of antibodies that make it possible to measure the expression products of two or more different genes comprising the nucleotide sequence of the protein encoded by any of the nucleotide sequences set forth in SEQ ID NOs: 1-82 or The composition is selected from the group consisting of an antibody that immunologically reacts with a polypeptide comprising at least 8 consecutive amino acid residues in the amino acid sequence of the protein, and a fragment of the antibody. To do.
In an embodiment of the invention, the composition is in the form of a kit.
In another embodiment of the present invention, the antibody probe allows measurement of expression products of 5 or more to 82 or less genes.
In another embodiment of the invention, the composition is for use in the method of the invention.
As used herein, “post-surgical recurrence” means that the patient relapses with lung adenocarcinoma after surgical resection of lung adenocarcinoma. Patients predicted to relapse are determined to be at high risk of relapse and have a low survival rate. In contrast, patients predicted to have no recurrence are determined to be at low risk of recurrence and have a high survival rate.
As used herein, “patient” refers to mammals including humans, and preferred mammals are humans.
In the present specification, the “nucleic acid” is either DNA or RNA, and includes, for example, transcription products (mRNA), cRNA, aRNA, cDNA and the like of each gene.
In the present specification, a “gene” encodes a specific protein and is composed of exons, introns, 5′- and 3′-untranslated regions, etc., and enables the formation of a precursor mRNA by its transcription, and further splicing After an event, it becomes mRNA and is translated into protein.
In the present specification, the “protein” includes a protein encoded by a specific gene, a variant thereof or a fragment thereof. Such variants include variants based on natural events such as splicing, polymorphisms and the like. Moreover, such a fragment is a protein fragment naturally generated by an enzyme action such as protease or peptidase in a living body.
In the present specification, “hybridization” is a reaction for forming a double strand by base pairing between a nucleic acid and a nucleic acid. Hybridization includes DNA-DNA hybridization, DNA-RNA hybridization, and RNA-RNA hybridization. Hybridization is usually performed using a nucleic acid probe immobilized on a solid support. The solid support is made of a material such as plastic, polymer, or glass, and may be in the shape of a test plate, a test tube, a test piece, a chip, or the like. In hybridization, temperature, ionic strength, surfactant concentration, presence or absence of formamide, GC content of nucleic acid, and the like affect the stability and hybridization rate of the hybrid. In general, the higher the temperature or the higher the ionic strength, that is, the more stringent (stringent) conditions, the lower the stability of the hybrid and, conversely, the higher the specificity of hybridization. For hybridization, see Ausbel FM et al., Short Protocols in Molecular Biology (3rd edition) A Complementum Methods from Current Protocols in Molecular Biology, 1995, John Ins. (United States).
In the present specification, “immunological” means forming a complex of an antigen and an antibody. A reaction that forms such a complex is referred to as an immunological reaction, and a method of detecting or quantifying an antigen or antibody using this reaction is referred to as an immunological method. The immunological method can be performed under conditions such as solid phase or liquid phase, homogeneous or heterogeneous, wherein the detection of the antigen-antibody complex is performed by labeling (eg, dye, fluorescent material, radioisotope, Enzymes, etc.), aggregation methods (eg, latex method), turbidity methods, immunoprecipitation methods, secondary antibody methods (eg, sandwich method), etc. Can do.
This specification includes the contents described in the specification and / or drawings of Japanese Patent Application No. 2007-339214, which is the basis of the priority of the present application.

FIG. 1 shows a schematic diagram of a training-effectiveness strategy for identifying recurrence-related signatures using 100 iterations of 10-fold cross-validation that randomly divides a training set into ten.
FIG. 2 shows the results of a training procedure for identifying recurrence-related signatures. FIG. 2A shows the search result of the optimal number of probes for defining the recurrence-related signature. Survival in the training group was assessed using the Kaplan-Meier survival curve. Also shown are patient-related recurrence-related curves at all stages (Figure 2B) and Stage I (Figure 2C).
FIG. 3 shows Kaplan-Meier survival curves in the presence or absence of EGFR mutation (FIG. 3A), K-ras mutation (FIG. 3B) and p53 mutation (FIG. 3C) in the training data set.
FIG. 4 shows the effectiveness of the RRS-82 signature using evaluation set I and evaluation set II. Kaplan-Meier survival curves were shown according to the predicted risk based on RRS-82. 4A shows the relapse-free survival curve for all stages, FIG. 4B shows the overall survival curve for all stages, FIG. 4C shows the relapse-free survival curve for stage I, and the overall survival curve for 4D stage I. In addition, a recurrence-free survival curve using evaluation set II consisting of samples of stage I only is shown in FIG. 4E, and an overall survival curve is shown in FIG. 4F. Furthermore, the recurrence-free survival curve in evaluation set I + II is shown in FIG. 4G, and the overall survival curve is shown in FIG. 4H.
FIG. 5 shows the relationship between the risk group based on RRS-82 and the TNM stage. FIG. 5A shows a relapse-free survival curve for a patient with a low-risk RRS-82 signature, and FIG. 5B shows a relapse-free survival curve for a patient with a high-risk RRS-82 signature.
FIG. 6 shows the validity of the RRS-82 signature by analysis of an independent data set. FIG. 6A shows the results of unsupervised hierarchical clustering analysis performed on the Duke University data set of 39 lung adenocarcinomas using components of the RRS-82 signature, and FIG. 6B shows Kaplan for clusters I and II. -Show Meier survival curves. FIG. 6C shows a conceptual diagram for constructing a prediction model using a Michigan University data set. FIG. 6D shows a Kaplan-Meier survival curve in which 104 samples of gene expression data from the Memorial Sloan-Kettering cancer center were analyzed using the model constructed according to FIG. 6C.
FIG. 7 shows an assessment of Kaplan-Meier survival on a training data set. FIG. 7A shows the overall survival curve with the RRS-82 signature for multiple patients at all stages, and FIG. 7B shows the overall survival curve with the RRS-82 signature for multiple patients at stage I.
FIG. 8 shows an evaluation of Kaplan-Meier survival with RRS-82 signature in stage II-III cases in Evaluation Set I. Of stage II-III patients, recurrence free (FIG. 8A) and overall survival (FIG. 8B) tended to be better for patients with a low risk RRS-82 signature compared to patients with a high risk signature. However, it was not statistically significant.
FIG. 9 evaluated the relationship between RRS-82 signature and stage in evaluation set I using a Kaplan-Meier survival curve. In the low-risk group shown in FIG. 9A, in the comparison between the stage I case and the stage II-III case, the overall survival curve of the stage I case tends to be good, although it is not statistically significant. In the high-risk group shown in FIG. 9B, the overall survival curve is almost the same curve in the comparison between the stage I case and the stage II-III case.

  As described above, the present invention provides a method for predicting recurrence of lung adenocarcinoma with high accuracy after surgical resection of lung cancer in patients with lung adenocarcinoma. In part, it was possible to make a prediction with such a high degree of accuracy. In some cases, patients with lung adenocarcinoma were selected rigorously after and after surgery without recurrence. Based on the discovery of 82 previously unpublished gene sets related to postoperative recurrence / no recurrence in patients with lung adenocarcinoma under conditions such as using reliable statistical methods .
<Determination of gene set>
  In the present invention, a set of 82 different genes for predicting recurrence of lung adenocarcinoma after surgical resection of lung cancer in lung adenocarcinoma patients was found, and the procedure for determining this gene set will be described below. .
  As described in the background art, gene sets for predicting postoperative recurrence of lung adenocarcinoma have been reported by several groups, but gene sets described as predictable from tens of thousands of gene groups Are not overlapped with each other, or the overlap is very low. Regarding the 82 gene set found by the present inventors, for example, there is no gene overlapping with the 45 gene set described in JP-A-2007-135466. E. There is only one gene that overlaps the 54 gene set described in Larsen et al., Clinical Cancer Research 2007, 13: 2946-2954. However, as proved in FIG. 2A, the correct answer rate (that is, [number of correct answers / (number of correct answers + number of incorrect answers)] × 100 (%)) when performing recurrence prediction using a single gene is , Significantly lower. In order to increase the correct answer rate, it is desirable to use a set of a plurality of genes, preferably 5 or more, more preferably 10 or more, further preferably 30 or more, and most preferably 82 genes. Even if more than 82 genes are used, FIG. 2A shows that the correct answer rate tends to deteriorate rather than improve.
  As described in the examples below, the inventors have prepared 87 lung adenocarcinoma samples from lung adenocarcinoma patients who have undergone resection surgery, of which 60 are randomly assigned to the training set and the remaining 27 Was used as an evaluation set. For a sample of the training set, a probe that exceeds the background expression level is selected and a probe signal that satisfies this filtering criterion is used to create a supervised classification technique (weighted voting algorithm; Yanagisawa et al., J. Natl. Cancer Inst. .99: 858-67, 2007) makes 82 genes recurrence-related signatures (ie high risk of recurrence or recurrence) predicting high risk or low risk of recurrence with an average accuracy of 75% or higher, Alternatively, it was identified as having no recurrence or low risk of recurrence). At this time, a 10-fold cross-validation method was used to select a probe signal for identifying a recurrence-free signature. An overview of this method is shown in FIG.
  The 10-fold cross-validation method, for example, divides 100 samples into 10 samples every 10 samples, and selects the characteristics (relapse related signatures) for the remaining 9 divisions by subtracting 1 division from the 10 divisions, and the top 1 selected A classifier consisting of prediction probes up to a maximum of 200 is constructed, and classification is performed by dividing the classifier. Next, signatures are selected in the same manner for the remaining 9 divisions excluding the other 1 division from the 10 divisions obtained by reverting the removed 1 division, and predictions from the selected top 1 to the maximum 200 are performed. A classifier consisting of the probes is constructed, and another one-segment classification is performed. The same operation is performed 10 times in total. Here, for example, in consideration of the fact that there are an infinite number of allocation methods that divide 100 samples into 10, in order to reduce the bias due to the allocation method, 100 samples were divided into 10 by random (random) 100 methods. In this manner, a classifier composed of a maximum of 200 prediction probes was constructed for a total of 1000 independent sets subjected to the 10-fold cross-validation method every 100 random divisions. As a result, the number of predicted probes 82 with the smallest learning error is found (FIG. 2A), and 82 probes that are most frequently shared among 1,000 independent sets are identified as recurrence-related signatures. In this way, a final classifier is constructed.
  The test sample is classified using the classifier constructed as described above. As test samples, 27 evaluation samples of lung adenocarcinoma excised from randomly selected lung adenocarcinoma patients were used. Kaplan-Meier survival curves and Cox proportional hazards model analysis are used to analyze the relationship between recurrence-related signatures obtained from the evaluation samples and overall survival and relapse-free survival. As a result, the 82 gene set of the present invention (referred to as “RRS-82”) has a high-risk patient with relapse / death and a low-risk patient with no relapse / high survival of 75% or more, preferably 80% or more, Preferably, the determination can be made with a high accuracy of 90% or more.
  RRS-82 is a set consisting of 82 genes including each nucleotide sequence of SEQ ID NOs: 1 to 82. This gene set is shown in Table 1.
  No. in the above table. 1 to No. 82 correspond to SEQ ID NO: 1 to SEQ ID NO: 82, respectively. Classifying the 82 genes based on relative expression differences from post-operative relapse-free cases, relapse-related signatures: relapse (death) signature or high recurrence risk signature and no recurrence signature or low recurrence risk signature, It could be divided into the following two groups.
(1) Includes each nucleotide sequence of SEQ ID NOs: 1-5, 8-9, 12, 14-15, 17, 19-21, 23-28, 30-33, 35-59, 61-68, 70-82 A group in which the recurrence of lung adenocarcinoma can be predicted to be high when the gene expression is relatively high compared to post-operative recurrence cases (or a group in which recurrence of lung adenocarcinoma can be predicted to be low if the relative expression is low ),as well as
(2) Expression of the gene containing each nucleotide sequence of SEQ ID NOs: 6-7, 10-11, 13, 16, 18, 22, 29, 34, 60, 69 is relatively low compared to post-operative relapse-free cases Group that can be predicted to have a high recurrence of lung adenocarcinoma (or, if the relative expression is high, the recurrence of lung adenocarcinoma is predicted to be low).
  That is, in the case of the gene of the above group (1), a lung adenocarcinoma patient is predicted as a relapse (death) patient or a high risk patient, or a non-relapse patient or a low relapse risk patient depending on whether the relative expression is high or low. Can be classified. In the case of the gene of the group (2), conversely, a lung adenocarcinoma patient is classified as a non-relapsed patient or a low-risk patient, or a relapsed (death) patient or a high-risk patient depending on whether the relative expression is low or high. Can be predicted and classified.
  If it is determined as recurrence or high risk by the method of the present invention, it becomes possible to devise a treatment plan for suppressing postoperative recurrence of lung adenocarcinoma patients at an early stage, while no recurrence or low risk. Once determined, it can also be used to determine unnecessary chemotherapy and radiation therapy.
  The control in comparing the relative expression levels of the genes is a non-relapsed postoperative case, and if the above prediction is possible, a certain period after the operation (eg 3 years, preferably 5 years, more preferably 10 years) A patient-derived sample (lung adenocarcinoma tissue or cells) determined to have no recurrence.
  No statistically significant association has been found between RRS-82 in the present invention and EGFR, K-ras or p53 gene mutations (FIG. 3).
  When the discriminating ability of RRS-82 in the present invention was verified using an evaluation set (27 lung adenocarcinoma samples), all patients who showed a high risk signature of RRS-82 relapsed with lung adenocarcinoma within 5 years. However, about 70% of patients who showed a low risk signature showed no signs of recurrence after 5 years (Figure 4). In this sense, it can be said that the method of the present invention is extremely effective in predicting a high-risk group of postoperative recurrence in lung adenocarcinoma patients.
  The recurrence / no recurrence prediction by the method of the present invention can be performed regardless of the stage of the tumor, and interestingly, it can be performed even at an early stage.
<Prediction of postoperative recurrence of lung adenocarcinoma>
  In one aspect thereof, the present invention is a method for predicting recurrence of lung adenocarcinoma after surgical resection of lung cancer in a lung adenocarcinoma patient, comprising a gene comprising each nucleotide sequence of SEQ ID NOs: 1 to 82 Measuring the expression of a gene set of 2 to 82 genes selected from the group in a biological sample from said patient, wherein SEQ ID NOs: 1-5, 8-9, 12, 14-15 , 17, 19-21, 23-28, 30-33, 35-59, 61-68, 70-82, the expression of the gene is relatively high compared to post-operative relapse-free cases Or the expression of a gene containing any one of the nucleotide sequences of SEQ ID NOs: 6-7, 10-11, 13, 16, 18, 22, 29, 34, 60, 69 is relative to that of a case without recurrence after surgery. Pulmonary gland Provides the above method characterized by recurrent predict high.
  The 82 genes are human genes including the nucleotide sequences represented by SEQ ID NOs: 1 to 82 as described above. Since it is a gene derived from an organism, it is considered that mutations such as polymorphic mutations, splice mutations and mutations may occur. Thus, the subject gene of the present invention can include not only the nucleotide sequences represented by SEQ ID NOs: 1 to 82 but also deletions, substitutions or additions of one or several nucleotides in these sequences, or It may have at least 95% identity to these sequences, preferably at least 97%, more preferably at least 98%, most preferably at least 99%.
  As used herein, the term “% identity” can be determined using known BLAST or FASTA algorithms with or without gaps introduced (Karlin and Altschul, 1993, Proc. Natl). Acad.Sci.U.S.A., 90: 5873-5877; Karlin and Altschul, 1990, Proc.Natl.Acd.Sci.U.S.A., 87: 2264-2268; J. Mol. Biol., 215: 403; Altschul et al., 1997, Nucleic Acids Res., 25: 3389-3402). In general, identity (%) can be calculated as a percentage of the number of matched bases relative to the total number of bases.
  As used herein, the term “several” refers to an integer from 2 to 10, ie any of 2, 3, 4, 5, 6, 7, 8, 9 or 10, preferably 2 Is an integer from 1 to 5, more preferably 2 or 3.
  In the present invention, the number of genes in the gene set is 2 to the total number (82). If the number of genes is within this range, postoperative recurrence can be predicted. However, in order to further improve the prediction accuracy (or accuracy), the number of genes is preferably 5 to 82, more preferably 10 to 82, and even more preferably 30 to 82, for example 40 to 82, 50 to 82, 60 to 82, 70 to 82, and 82 is the most preferable number of genes.
  Gene expression can be measured for the amount or presence of either nucleic acid or protein. In the measurement of nucleic acid, mRNA that is a transcription product of each gene and its derivative nucleic acid, for example, cDNA, cRNA, and aRNA (antisense amplified RNA) can be measured. On the other hand, in protein measurement, a protein encoded by each gene (also referred to as an expression product or a gene product), a fragment thereof, or a post-translational modification product (for example, modification with a sugar or the like) can be measured.
  Below, the method for measuring the expression of said 82 gene at a nucleic acid and protein level is demonstrated.
  The test sample is a lung adenocarcinoma tissue or cell of a lung adenocarcinoma patient, such as a cancer tissue excised by surgery or a tissue or cell obtained by biopsy.
Measuring gene expression with nucleic acids
  In the method of the present invention, the following nucleic acids can be used as probes or primers in order to detect the expression of 82 genes.
  (1) a nucleotide sequence represented by SEQ ID NOs: 1 to 82, a nucleotide sequence complementary thereto, or a nucleic acid having a partial sequence of 15 bases to less than the total number of bases in these nucleotide sequences, and
  (2) A nucleic acid that hybridizes with a nucleic acid comprising the nucleotide sequence shown in SEQ ID NOs: 1 to 82 or a nucleotide sequence complementary thereto, or a partial sequence that is less than the total number of bases from 15 consecutive bases of the hybridizing nucleic acid Nucleic acid.
  The size of the nucleic acid fragment consisting of the partial sequence is from 15 bases to less than the total number of bases, for example, 17 bases or more, 20 bases or more, 30 bases or more, 50 bases or more, 70 bases or more, 100 bases or more, 150 bases or more. , 200 bases or more, 300 bases or more, 400 bases or more, or 500 bases or more and less than the total number of bases.
  If the nucleic acid is a DNA molecule of about 100 bases or less, it is synthesized using a commercially available DNA automatic synthesizer utilizing the phosphoramidite method or the like, or if it is a DNA molecule of about 100 bases or more, cDNA cloning is performed. Or polymerase chain reaction (PCR). Moreover, cRNA and aRNA can be produced from cDNA using a commercially available kit or the like.
  In the cDNA cloning and PCR methods, total RNA is obtained from lung adenocarcinoma tissue, poly A (+) RNA is obtained by oligo dT cellulose column treatment, and then cDNA live is obtained by reverse transcription-polymerase chain reaction (RT-PCR) method. A library corresponding to the gene of interest is selected / cloned from this library using a probe or primer having a length of 15 bases or more prepared based on the nucleotide sequence shown in SEQ ID NOs: 1 to 82. Or it can be PCR amplified. The probe has a sequence complementary to the cDNA sequence, and the length thereof is preferably 20 bases or more, more preferably 30 bases or more, and further preferably 50 bases or more. Moreover, a primer consists of a sense primer and an antisense primer, The length becomes like this. Preferably it is 15-50 bases, More preferably, it is 17-30 bases, More preferably, it is 20-25 bases.
  The DNA obtained as described above can be introduced into a prokaryotic or eukaryotic host cell after being inserted into a commercially available or documented vector, and amplified, or a vector into which the target DNA has been inserted Can be used as a template for PCR amplification.
  The vector is a vector such as a plasmid, a phage, a plasmid, or a virus, such as a plasmid such as a pBluescript system, a pUC system, or a pBR system. The vector can contain a promoter, a replication origin, a ribosome binding sequence (or Shine-Dalgarno sequence), a terminator, a selection marker sequence (for example, a drug resistance gene), a multicloning site and the like in addition to the target DNA. Since a vector suitable for the host cell is commercially available, it is desirable to use it.
  Prokaryotic host cells are bacterial cells and generally include Escherichia coli, Bacillus subtilis, Pseudomonas bacteria and the like. In addition, eukaryotic host cells include yeast, fungi, basidiomycetes, animal cells (insect cells, mammalian cells, etc.), plant cells and the like.
  Methods for introducing vectors into host cells are also known, and include, for example, electroporation, calcium phosphate method, spheroplast fusion method, protoplast fusion method, microinjection, gene gun method, Agrobacterium method and the like.
  Details of the above technique can be found, for example, in J. Org. Sambrook et al., Molecular Cloning A Laboratory Manual, 1989, Cold Spring Harbor Laboratory Press (USA); M.M. Ausbel et al., Short Protocols in Molecular Biology (3rd edition) A Compendium of Methods from Current Protocols in Molecular Biology, 1995, John Wiley & Son. (US) etc.
  The nucleic acid that can be used in the method of the present invention includes, as described in (2) above, a nucleic acid that hybridizes with a nucleic acid comprising the nucleotide sequence shown in SEQ ID NOs: 1 to 82 or a nucleotide sequence complementary thereto, or the hybridizing nucleic acid A nucleic acid having a partial sequence of 15 bases to less than the total number of bases. Hybridization is preferably performed under stringent conditions. Under these conditions, the nucleotide sequence has 90% or more, preferably 95% or more, more preferably 98 or more, and most preferably 99% or more identity. Hybridization can occur with nucleic acids of the sequence.
  The expression of 82 gene sets is measured using the above nucleic acid. Preferred measurement methods include hybridization methods or quantitative PCR methods (for example, real-time PCR methods).
  Hybridization refers to a technique in which a nucleic acid probe attached or bound on the surface of a solid support is bound to a partially or completely complementary nucleic acid via base pairing to form a double strand. . The solid support includes, for example, a test plate, a test tube, a test piece, an array, and the like, and the material is usually a polymer (or plastic, resin), glass, or the like. The hybridization used in the method of the present invention may be any of DNA-DNA hybridization, DNA-RNA hybridization, or RNA-RNA hybridization, including, but not limited to, a microarray method, Blot methods such as Northern blot, Southern blot, Northern hybridization, Southern hybridization, in situ hybridization and the like are included.
  In the microarray method, a microarray (or DNA chip) in which a nucleic acid probe of 15 bases or more that hybridizes with a gene comprising the nucleotide sequence of SEQ ID NOs: 1 to 82 or a corresponding nucleic acid (mRNA, cDNA, cRNA or aRNA) is bound to a support. ) Is used. A spacer or a crosslinker containing a reactive group for covalently binding a nucleic acid probe can be introduced on the surface of the microarray. The binding between the array surface and the nucleic acid probe can be performed through a chemical reaction by light or heat, for example.
  The hybridization described above enables more specific binding between a nucleic acid probe and a test nucleic acid when hybridization is performed under stringent conditions. Stringent conditions include those defined above, but it is desirable to determine optimal conditions as appropriate according to the type of hybridization method. In general, stringency is highly dependent on temperature, ionic strength of the solution, presence of formamide, relative GC content of nucleic acids, incubation time, and the like. The optimum temperature can be determined from the temperature range before and after the temperature (Tm) when 50% of the oligonucleotide dissociates from its complementary strand. The temperature under stringent conditions is usually within the range of 42 to 68 ° C. High stringency hybridization conditions include, for example, 6 × SSC, 0.01M EDTA, 1 × Denhardt's solution, hybridization in 0.5% SDS at 68 ° C .; 1M NaCl, 0.5% sarkosyl, 30% formamide Medium, hybridization at 60 ° C .; washing in 0.1 × SSC, 0.1% SDS, 55 ° C., etc. Here, 1 × SSC is 150 mM sodium chloride and 15 mM sodium citrate aqueous solution (pH 7.2).
  The solid phase immobilization of the nucleic acid probe is not particularly limited, but a general method such as a method of spotting DNA using a high-density dispenser called a spotter or an arrayer, an ink jet method of ejecting droplets from a nozzle, or the like This method can be used.
  Nucleic acids such as cDNA, cRNA, and aRNA derived from mRNA derived from lung adenocarcinoma tissue test samples surgically excised from lung adenocarcinoma patients are labeled with a fluorescent substance such as Cy dye (eg Cr3, Cy5), and then on the microarray. Hybridize with the probe. The fluorescence intensity is read using a laser scanning reader, and the data is analyzed by a computer.
  In blotting, the above-described nucleic acid probe of the present invention is combined with a radioisotope (eg,³²P and³⁵S) and fluorescent substances (fluoresamine, rhodamine, derivatives thereof, Cy dye, etc.), etc., and then hybridization with nucleic acids such as mRNA, cDNA, cRNA, aRNA in the test sample transferred to the polymer membrane Do. The signal is detected using a radiation detector or a fluorescence detector and its intensity is measured.
  In the quantitative real-time PCR method, primers are annealed with cDNA so as to amplify each target gene region using cDNA prepared from mRNA in a lung adenocarcinoma tissue test sample as a template, and PCR is performed in real time. Double-stranded DNA is detected. The amplification product is amplified exponentially every cycle and eventually reaches a plateau. Real-time PCR is performed using a standard sample that is serially diluted, and an amplification curve for each cycle number is created with the number of cycles on the horizontal axis and the amount of PCR amplification on the vertical axis. A threshold value is set, and a point Ct value (thresholdcycle) where the threshold value and the amplification curve intersect is calculated. The Ct value is the number of cycles when the PCR product reaches a certain amount. Since there is a linear relationship between the Ct value and the initial template DNA amount, a calibration curve is created.
  Detection of the amplification product can be performed using a method such as an intercalator method or a fluorescent labeling probe method. The intercalator method is a method of monitoring the amount of amplification product by utilizing the property that a fluorescent substance binds between double strands of DNA formed by hybridization, for example. In the fluorescent labeling probe method, a probe is labeled with a fluorescent substance, and the amount of amplification product is monitored by a fluorescent signal. Among these methods, a method using a probe having a fluorescent substance at the 5 'end and a quencher bound at the 3' end is also known.
  Since a real-time PCR apparatus is commercially available, a real-time PCR method can be performed using this apparatus.
  As conditions for carrying out PCR, the amplification size should be in the range of 80-150 bp, the primer size should be in the range of 17-25 bases, and the GC content of the entire primer should be in the range of 40-60%, The 3 'end of the primer is G or C, there is no sequence that complements more than 3 bases inside or between the primers, the Tm values of the upstream primer and the downstream primer are within 2 ° C, and a template-specific primer is designed. , Etc. are known.
  PCR conditions are, for example, denaturation: 92 to 94 ° C. for 30 to 60 seconds; annealing: 50 to 55 ° C. for 30 to 60 seconds; extension: 68 to 72 ° C. for 30 to 60 seconds, and 30 to 40 cycles of reaction. including. Reverse transcriptase is a commercially available enzyme such as SuperScript.^TM  III (Invitrogen), AMV Reverse Transscriptase, M-MLV Reverse Transscriptase (Promega), SYBR Premix Ex Taq (Takara Bio) and the like can be used.
  In the present invention, the relative expression level difference between relapse and non-relapse is 1.1 times or more, preferably 1.5 times or more, more preferably 2.0 times or more.
Measurement of protein gene expression
  In the method of the present invention, gene expression can be measured by measuring the amount of a protein encoded by a gene comprising the nucleotide sequence represented by SEQ ID NOs: 1 to 82 or a fragment thereof. An alternative method of measuring this expression level is an immunological method.
  Proteins for producing antibodies can be produced using cDNA cloning and gene recombination techniques.
  Briefly, for example, a cDNA library is prepared from lung adenocarcinoma tissue, a cDNA clone is prepared by PCR using primers synthesized based on each nucleotide sequence of SEQ ID NOs: 1 to 82, and the obtained cDNA clone is It can be obtained from the cell or culture supernatant by culturing a prokaryotic or eukaryotic host cell that is incorporated into an expression vector and transformed or transfected with the vector. In this case, the expression vector can be commercially available or used in the literature, for example, a plasmid, cosmid, phage or the like. The vector can contain DNA encoding the target protein, promoter, polyadenylation signal, ribosome binding sequence, replication origin, terminator, selectable marker, and the like. In order to facilitate separation and purification of the polypeptide, a labeled (poly) peptide (eg, β-galactosidase gene), (His)_6-10Tags, FLAG tags (amino acid sequence: DYKDDDDK (SEQ ID NO: 150)), GFP (green fluorescent protein), and other DNA sequences corresponding to the proteins so that a fusion protein of the target protein is formed with the target protein It can also be contained.
  Host cells include prokaryotic cells such as bacteria (eg, E. coli, Bacillus subtilis, Pseudomonas bacteria, etc.), yeast (eg, Saccharomyces, Schizosaccharomyces, Pichia, etc.), insect cells (eg, Sf cells), mammalian cells ( For example, CHO, COS, BHK, HEK293, NIH3T3, etc.) are included.
  The above-mentioned cDNA cloning and gene recombination techniques are described in Sambrook et al. (Above), Ausbel et al. (Above), and the like.
  The target protein thus obtained may be gel filtration, ion exchange chromatography, affinity chromatography, hydrophobic chromatography, isoelectric focusing, electrophoresis, ultrafiltration, salting out, dialysis, etc. alone or It can be purified in combination.
  The protein encoded by the gene of the present invention comprises the amino acid sequence shown in SEQ ID NOs: 83-149 (the sequences are encoded by the nucleotide sequences of SEQ ID NOs: 15-21, 23-82, respectively). In addition, protein sequences that do not describe amino acid sequences can be obtained by accessing databases such as GenBank (USA) and EMBL (Europe), or cDNA cloning using nucleotide sequences corresponding to them. And the amino acid sequence of the target protein can be determined by sequencing.
  In the immunological method, mammals such as rabbits, mice, rats, and horses are immunized using the protein obtained by the above-described method or a fragment thereof as an antigen, and antibodies against these antigens are produced and purified.
  Antibodies include polyclonal antibodies, monoclonal antibodies, anti-peptide antibodies, antibody fragments thereof and the like.
  Polyclonal antibody immunizes the animal subcutaneously with an appropriate amount (usually in the order of μg) of the antigen, further boosts about 2-3 weeks later, collects blood about 3 weeks to 2 months after the initial immunization, The IgG component containing the target polyclonal antibody can be prepared by separation using ammonium sulfate fractionation and ion exchange chromatography. At this time, in order to increase the specificity of the antibody for the antigen, the obtained IgG is bound to a column prepared by binding the target protein to a carrier such as cellulose or agarose, and then eluted with a high salt concentration buffer. By desalting by a method such as dialysis or ultrafiltration, a specific polyclonal antibody can be obtained. The antibody titer can be measured by a conventional immunoassay, for example, an enzyme immunoassay (ELISA), a radioimmunoassay (RIA), a fluorescent antibody method, or the like.
  The monoclonal antibody can be prepared, for example, by the following general method (for example, Hideaki Nagamune and Hiroshi Terada, monoclonal antibody-preparation and characterization, 1990, Hirokawa Shoten; Yoshikazu Sado (2001) Biochemistry 73: 1163-1167, etc.).
  The target protein or a fragment thereof is administered subcutaneously to mice or rats (for example, Balb / c mice) in the same manner as polyclonal antibody production, and booster immunization is performed about 2 to 4 times at intervals of 1 to 4 weeks. When the antibody titer reaches a plateau, the antigen is injected intravenously or intraperitoneally for final immunization. After 2 to 5 days, antibody-producing cells (eg spleen cells or lymph node cells) are collected. The antibody-producing cells are then fused to a myeloma cell line (preferably a hypoxanthine / guanine / phosphoribosyl transferase (HGPRT) deficient cell line) to produce hybridoma cells, and HAT (hypoxanthine, aminopterin, thymidine) is selected. I do. In cell fusion, antibody-producing cells and myeloma cell lines are mixed at an appropriate ratio in animal cell culture media such as serum-free DMEM and RPMI-1640 media, and cell fusion promoters such as polyethylene glycol are used. Perform in the presence. Confirmation of the target antibody can be performed by the immunoassay described above. Furthermore, in order to proliferate the hybridoma, the hybridoma is administered into the abdominal cavity of the mouse, and after the hybridoma is proliferated, ascites is collected after 1 to 2 weeks. Antibody purification can be performed by appropriately combining methods such as ammonium sulfate fractionation, ion exchange chromatography, affinity chromatography, and gel chromatography.
  Anti-peptide antibodies are antibodies against linear peptides on the surface of proteins and can increase immunological specificity. Such peptides include, for example, the estimation of hydrophilic-hydrophobic regions by Kyte-Doolittle, et al., The probability of being located on the surface of a specific peptide site on a protein molecule by Emini et al., The degree of bending of a polypeptide chain, such as Chou-Fasman Can be estimated using an α helix, β sheet, secondary structure prediction of a protein indicating a turn, etc. alone or in combination. The estimated peptide can then be synthesized using a peptide synthesizer.
  The above antibodies can be used for the detection of target proteins or fragments thereof in lung adenocarcinoma tissue test samples. By creating an antibody microarray in which a large number of antibodies are bound on a microarray support, or by dot spotting a large number of antibodies on a polymer membrane (filter) such as a PVDF membrane, or on a solid phase such as a multiwell plate By adsorbing antibodies on the surface, it becomes possible to detect or quantify a large number of target proteins at once. Also, conventional immunological measurement methods such as enzyme immunoassay (ELISA, EIA), fluorescent antibody method, radioimmunoassay (RIA), luminescence immunoassay, immunoturbidimetric method, latex agglutination, latex turbidimetric The target protein or a fragment thereof in the test sample can be detected or quantified by a method, hemagglutination reaction, particle agglutination reaction, Western blotting, or the like.
  When the reaction is carried out on a solid phase, solid phase supports include polymer films such as polystyrene, polycarbonate, polyethylene, multi-well plates, test tubes, test pieces (test strips), latex, magnetic particles and the like. Immobilization can be performed by physical adsorption or chemically. For chemical coupling, the solid phase can be treated with a reagent such as a maleating reagent or cyanogen bromide to introduce a functional group that reacts with the amino group of the protein into the solid phase.
  Examples of labels include enzymes such as horseradish peroxidase and alkaline phosphatase, fluorescent materials such as fluorescein, rhodamine and derivatives thereof, luminescent materials such as luciferase and luminol,³²P,¹²⁵Radioisotopes such as I are included. Labeling includes, for example, glutaraldehyde method, maleimide method, pyridyl disulfide method, chloramine T method, Bolton Hunter method and the like.
<Composition for predicting recurrence after surgery>
  The present invention also provides a composition for predicting recurrence of lung adenocarcinoma after surgical resection of lung cancer in a patient with lung adenocarcinoma, wherein the composition comprises any nucleotide sequence of SEQ ID NOs: 1 to 82 Comprising a plurality of nucleic acids that allow the measurement of the expression of 2 or more to 82, preferably 5 or more to 82 different genes,
  (1) a nucleotide sequence represented by SEQ ID NOs: 1 to 82, a nucleotide sequence complementary thereto, or a nucleic acid having a partial sequence of 15 bases to less than the total number of bases in these nucleotide sequences, and
  (2) A nucleic acid that hybridizes with a nucleic acid comprising the nucleotide sequence shown in SEQ ID NOs: 1 to 82 or a nucleotide sequence complementary thereto, or a partial sequence that is less than the total number of bases from 15 consecutive bases of the hybridizing nucleic acid Nucleic acid,
The composition is selected from the group consisting of:
  The plurality of nucleic acids constituting the composition of the present invention is 2 to 82, preferably 5 to 82, more preferably 10 or more, wherein the composition comprises any one of the nucleotide sequences of SEQ ID NOs: 1 to 82. Of the nucleic acids described in (1) and (2) above, so long as it is possible to measure the expression of the different genes from 30 to 82, more preferably from 30 to 82, and even more preferably from 40 to 82. Any combination may be used.
  The nucleic acid of (1) above is
  (A) the nucleotide sequence shown in SEQ ID NO: 1-82,
  (B) a nucleotide sequence complementary to the nucleotide sequence of (a) above, and
  (C) The nucleotide sequence of (a) above or the nucleotide sequence of (b) above is less than the total number of bases, preferably 17 bases or more, 20 bases or more, 30 bases or more, 50 bases or more, 70 bases or more. , 100 bases or more, 150 bases or more, 200 bases or more, 300 bases or more, 400 bases or more, or 500 bases or more to a partial sequence of less than the total number of bases,
A nucleic acid having
  The nucleic acid of (2) above is a nucleic acid having a nucleotide sequence that hybridizes with the nucleotide sequence of (a) or (b) of (1) above, or all 15 bases continuous in the nucleotide sequence of the nucleic acid. Less than the number of bases, preferably 17 bases or more, 20 bases or more, 30 bases or more, 50 bases or more, 70 bases or more, 100 bases or more, 150 bases or more, 200 bases or more, 300 bases or more, 400 bases or more or 500 bases or more A nucleic acid having a partial sequence of less than the total number of bases. Preferably, such a nucleic acid has 90% or more, preferably 95% or more, more preferably 98 or more, most preferably 99% or more identity with the nucleotide sequence of (1) (a) or (b) above. Or a fragment of 15 bases to less than the total number of bases.
  A preferred nucleic acid is the nucleic acid of (1) above, particularly the nucleic acid of (1) (c) above.
  The composition of the present invention may be in the form of a kit, a microarray, or a primer set, as well as a mixture in which nucleic acid components are simply mixed. Rather, the latter form is preferred. Here, in the case of a kit, the above nucleic acids can be individually packaged or packaged in a mixed form of a plurality of nucleic acids. The kit further includes a buffer for hybridization, a reagent for detection (such as enzymes), Instructions for use can be included. The microarray and primer set are as described above.
  The composition of the present invention can be used in a method for predicting recurrence of lung adenocarcinoma after surgical resection of lung cancer in lung adenocarcinoma patients as described above. The method of the present invention is as described above.
  The present invention further provides a composition for predicting recurrence of lung adenocarcinoma after surgical resection of lung cancer in a patient with lung adenocarcinoma, wherein the composition comprises any one of the nucleotide sequences of SEQ ID NOs: 1 to 82 A plurality of two or more to 82 species, preferably 5 or more to 82 or less, more preferably 10 or more, even more preferably 30 or more, even more preferably 40 to 82 or less An antibody comprising an antibody, wherein the antibody immunologically reacts with a protein encoded by the nucleotide sequence shown in SEQ ID NOs: 1 to 82 or a polypeptide consisting of at least 8 consecutive amino acid residues of the amino acid sequence of the protein, And a fragment of said antibody, provided with said composition, characterized in that it is selected from the group consisting of
  The production of the antibody is as described above.
  The antibody is a polyclonal antibody, a monoclonal antibody, an anti-peptide antibody or the like as prepared by the above method, but is not limited thereto. The type of antibody may be any type, class, subclass, and includes, for example, IgG, IgM, IgE, IgD, IgA and the like. Antibody fragments are Fab, (Fab ')₂, Fv, scFv and the like.
  The composition of the present invention may be in the form of a kit, an antibody-binding support or an antibody microarray, as well as a mixture in the form of simply mixing antibody components. Examples of the solid support include films of polymers such as polystyrene, polycarbonate, and polyethylene, multi-well plates, test tubes, test pieces (test strips), latex, and magnetic particles. Array materials include polymers, glass, and the like.
  As described above, the antibody-containing composition of the present invention can be used in a method for predicting recurrence of lung adenocarcinoma after surgical resection of lung cancer in a lung adenocarcinoma patient. The method of the present invention is an immunological method as described above.
  The present invention will be described more specifically with reference to the following examples. However, the present invention is not limited to these examples.

Materials and methods
Patients received 87 cases of lung adenocarcinoma from patients who had undergone resection surgery between December 1995 and August 1999, and were approved by the Aichi Cancer Center and Nagoya University Ethics Committee and Obtained consent form and obtained at Aichi Cancer Center (Nagoya City). None of the cases received adjuvant chemotherapy. The median follow-up period for patients who were alive at the time of the last follow-up was 90 months (range: 64-108 months). All tumor specimens were embedded in an OCT compound (Sakura Seiki Co., Ltd., Tokyo, Japan) and stored at −80 ° C. Two-thirds of patients (n = 60) were randomly assigned to the training set, and the remaining one-third (n = 27) was evaluated set I. Similarly, 30 samples diagnosed as stage I among lung adenocarcinoma samples of patients undergoing resection surgery between February 2002 and December 2004 were used as evaluation set II (Table 2).
Isolation of RNA Frozen tissue of tumor specimens was subjected to gross microdissection using Giemsa-stained sections of 10 sheets under the guidance of a pathologist (YY). Total RNA was extracted using the RNeasy kit (Qiagen, Valencia, CA) and treated with DNaseI. RNA levels were measured using a NanoDrop (R) ND-1000 UV-Vis spectrophotometer (NanoDop Technologies, Wilmington, Del.) And RNA properties were observed using an Agilent 2100 Bioanalyzer and expressed as RNA complete values (RIN) (Agilent Technologies, Palo Alto, CA). A large amount of common standard RNA was prepared using 20 lung cell lines showing all major lung cancer histology.
Obtaining EGFR, p53 and K-ras expression profiles and mutational status From 500 ng of total RNA, using poly dT primers containing Moloney murine leukemia virus reverse transcriptase (Agilent Technologies, Palo Alto, Calif.) And T7 promoter, Single-stranded cDNA was synthesized. CRNA was generated using Low RNA Fluorescent Linear Amplification kit (Agilent Technologies) and labeled with Cy3 or Cy5. Cy5-sample cRNA and Cy3-common standard cRNA were hybridized to a Whole Human Genome oligo DNA microarray kit (G4112F, Agilent Technologies) with 41,000 different probes. After hybridization and washing, slides were scanned using an Agilent DNA microarray scanner (G2505B, Agilent Technologies). Expression data was obtained using Agilent Feature Extraction software 9.5.1 (Agilent Technologies). Mutation status of EGFR, p53 and K-ras has already been reported for the same set of patients (Takeuchi et al., JCO, 2006). The Duke data set can be downloaded from Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) under the accession number GSE3593. Regarding clinical information such as survival period (month), situation, cell type, etc. Downloaded from Duke Institute for Genome Sciences and Policy (http://www.genome.duke.edu/). Of the 45 adenocarcinoma samples, 39 adenocarcinoma samples annotated for clinical information were used for this analysis.
Statistical methods First, probes with expression levels above background in at least 90% of the samples were selected. A weighted voting algorithm (each weighted value is calculated as a signal-to-noise ratio, which is an established technique of supervised classification) to identify recurrence-related signatures using signals that have passed this filtering criterion. (Yanagisawa et al., J. Natl. Cancer Inst. 99: 858-67, 2007) In short, it can be used generally without overtraining specific to the training data set. In order to obtain a classifier, we used a 10-fold cross-validation method with 100 random interactions in selecting probe signals to identify relapse-free signatures (AM Molinaro et al. , Bioinformatics 21: 3301-7, 2005). The entire construction process is repeated 1000 times for each classifier with a certain number of predicted genes, using the Kaplan-Meier survival curve and Cox proportional hazards model analysis (Stat version 7.0) for 27 patients. The relationship between the obtained relapse-related signatures in the evaluation set and overall survival and relapse-free survival was analyzed, showing the hazard ratio between death and relapse, respectively.P value presented in multivariable Cox model is similar ratio test All statistical tests were two-sided, using the Cluster program (MB Eisen et al., Proc. Natl. Acad. Sci. USA 95: 14863-8, 1998) between genes and cases. Average associative hierarchical clustering was performed, where the display used the Treeview program (MB isen et al., supra).
result
Identification of Recurrence-Related Signatures We have constructed and verified gene expression signatures for recurrence and death of lung adenocarcinoma patients who received surgical treatment without information leakage between training set / evaluation set I (FIG. 1). First, 87 cases of expression profile data were classified into 60 training sets and 27 evaluation sets I. Next, of the 60 training sets, 21 patients who died within 5 years after surgical resection due to recurrence had a common high-risk signature in the tumor, compared to the entire follow-up period It was assumed that 28 patients who survived for more than 5 years without showing signs of moderate recurrence did not have the aforementioned signature in the tumor. The remaining 11 patients in the training set were excluded from the process of identifying recurrence-related signatures because of the unclearness of the tumor. Of the excluded patients, 5 survived for more than 5 years with some signs of recurrence during the follow-up period, 5 died of cancer after surviving for more than 5 years, and 1 had no evidence of recurrence Died.
Out of 41,000 probes on a whole genome microarray (Agilent; Whole Human Genome (4 × 44K) Oligo Microarray Kit; model number G4112F), 23,828 probes select the first informational probes It was classified based on the SN metric that passed the filtering criteria and was used to identify the best recurrence-related signatures that discriminate between patients who died from recurrence and patients who were cured by surgery. The learning error for each model with an increasing number of predictive probes applied was calculated by 10-fold cross-validation, and the calculation was repeated 100 times using a new randomly divided data set. 1,000 independent sets of up to 200 predictive probes were selected to build a classifier using recurrence-related signatures. As a result, it can be seen that the use of 82 prediction probes causes a minimum learning error (FIG. 2A), which is most frequently shared among 1,000 independent sets of 82 prediction probes each. 82 probes were identified as recurrence-related signatures (hereinafter referred to as “RRS-82”, see Table 1 above). The RRS-82 signature appears to be able to identify patient groups with very poor prognosis when considering all stages or stage I (FIGS. 2B and 2C for relapse-free survival and FIG. 7 for overall survival). See).
Relationship between RRS-82 and various genetic changes and comparison Lung cancer is known to conceal various genetic changes that may affect the clinical nature of patients with genetic changes. Therefore, the present inventors analyzed the relationship between RRS-82 and the presence of EGFR, K-ras or p53 gene mutation. However, no statistically significant association was found between them. Furthermore, the prognostic significance of the genetic change was also analyzed using the same patient sample set, thereby establishing a prognostic classifier using RRS-82. The relationship between the presence of EGFR mutations and postoperative prognosis Not observed (Figure 3A). There was no statistically significant difference in overall survival between patients with and without the K-ras mutation (FIG. 3B) or p53 mutation (FIG. 3C).
Validation of RRS-82 using an evaluation set To assess the robustness of RRS-82, the discriminatory ability of RRS-82 was analyzed using an independent evaluation set I of 27 lung adenocarcinoma samples. The results obtained using Evaluation Set I showed that RRS-82 can distinguish between high-risk and low-risk patients with relapse and death. Relapse-free survival was significantly different between the two groups, with a median relapse-free survival of 8 patients in the high-risk group being 292 days, compared to 108 months in the 18 patients in the low-risk group (P <0.0003, FIG. 4A). The proportion of relapse-free patients in the high-risk group and low-risk group was 38% and 78%, respectively, at 2 years. Of note, all 8 patients with a high risk signature of RRS-82 experienced a relapse during a follow-up period of less than 5 years, compared with 67 patients with a low risk signature of RRS-82. % Showed no signs of recurrence 5 years after surgery. Among high-risk group patients, overall survival after surgical resection was significantly lower than that of low-risk group patients (P = 0.026, FIG. 4B). In a multivariate Cox proportional hazard model, RRS-82 was found to be an independent and significant prognostic factor predicting both relapse-free survival (P = 0.03) and overall survival (P = 0.03). It was. In addition, RRS-82 predicts both relapse-free survival (P <0.001) and overall survival (P = 0.005) in the multivariate Cox proportional hazards model when evaluation sets I and II are added It was found to be an independent and significant prognostic factor (Table 3).
Usefulness of RRS-82 in stage I patients We used 16 samples of stage I in evaluation set I to assess the discriminatory power of RRS-82 in a subgroup of adenocarcinoma cases with early or more advanced disease. And further evaluated. All stage I patients predicted to be high risk based on RRS-82 experienced recurrence within 5 years and died during the follow-up period (FIG. 4C (P = 0.0008) for recurrence-free survival, all Survival is shown in FIG. 4D (P = 0.043), both cases by log rank test). Further, even when 30 samples of evaluation set II consisting of all stages I were used, it was confirmed that they were statistically significant (FIGS. 4E and 4F). Further, in the analysis of 46 samples including 16 samples of stage I of evaluation set I and 30 samples of evaluation set II, it was confirmed that the discriminatory power of RRS-82 was statistically significant (FIGS. 4G and 4H). ).
Of stage II-III patients, relapse-free and overall survival tended to be better in patients with low-risk RRS-82 signatures compared to patients with high-risk signatures, but not statistically significant (FIGS. 8A and 8B).
From another aspect, the present inventors also diagnosed RRS-82 and TNM (T: primary tumor spread; N: presence / absence of regional lymph node metastasis; M: presence / absence of distant metastasis) (for example, lung cancer / esophageal cancer / We investigated the relationship between breast cancer and handling rules (excerpt) published by Kanehara. Among patients in the low-risk group, both the relapse-free survival Kaplan-Meier curve (FIG. 5A) and the overall survival Kaplan-Meier curve (FIG. 9A) tend to show a reasonable relationship with the stage of pathological disease. (P = 0.15 for relapse-free survival, P = 0.18 for overall survival). On the other hand, in the high-risk group of patients, both relapse-free survival (FIG. 5B) and overall survival curve (FIG. 9B) were independent of the pathological stage of disease (P = 0. 71, P = 0.87 for overall survival).
As a result of the above test of the present invention, recurrence was observed in 86% and 100% of cases judged as high-risk RRS-82 signatures in the training set and evaluation set I, respectively. This is because high-risk metadata signatures predicted relapse in 69% and 79% of the ACOSOG and CALGB efficacy assessment data sets reported by Duke et al. (DG Beer, Nat. Med. 8: 816-24, 2002), respectively. Compared to the results, it can be seen that the accuracy (or accuracy) of the determination is significantly improved.
Confirmation of general applicability of RRS-82 using independent data sets The robustness of RRS-82 in predicting lung adenocarcinoma viability was compared to 39 completely different data sets for lung adenocarcinoma And further validation using the Duke University gene expression data set. Since only 46 genes in RRS-82 on the Agilent platform correspond to a relatively new version of the Affymetrix platform Duke University dataset probe (Table 4), applying a classifier using RRS-82 directly It was impossible. Therefore, analysis was performed based on the expression profiles of 46 corresponding genes by unsupervised hierarchical clustering. Twenty-nine samples of adenocarcinoma showed significantly different postoperative viability (P = 0.028, FIG. 6B) and were clearly clustered into two different subsets (FIG. 6A). In addition, 31 genes overlapping with RRS-82 were extracted using a gene expression data set of Michigan University, and a prediction model was constructed (FIG. 6C). Using the model, 104 samples of gene expression data from the Medical Sloan-Kettering Cancer Center were analyzed. The high-risk group and low-risk group identified by this model were statistically significant in terms of recurrence-free survival. Recognized the difference. (P = 0.0003) (FIG. 6D). From these results, it was shown that the gene contained in RRS-82 has versatility (general applicability) related to recurrence risk assessment.

By the method of the present invention, the possibility of recurrence after surgery in lung adenocarcinoma patients can be predicted with extremely high accuracy. This method allows for high accuracy predictions even for stage I patients. This makes it possible to develop a treatment plan for improving the survival rate for patients at a high risk of recurrence at an early stage after the operation.
According to the present invention, it becomes possible to predict postoperative recurrence of adenocarcinoma in lung adenocarcinoma patients, so that a treatment plan suitable for postoperative patients can be made. Thus, the present invention can contribute to the medical industry for the treatment of lung adenocarcinoma.
All publications, patents and patent applications cited herein are incorporated herein by reference in their entirety.
[Sequence Listing]

Claims (16)

  A method for predicting recurrence of lung adenocarcinoma after surgical resection of lung cancer in a patient with lung adenocarcinoma, comprising 2 to 82 selected from the group consisting of genes comprising each nucleotide sequence of SEQ ID NOs: 1 to 82 Measuring the expression of a gene set of genes in a biological sample from said patient, wherein SEQ ID NOs: 1-5, 8-9, 12, 14-15, 17, 19-21, 23-28 , 30-33, 35-59, 61-68, 70-82, the expression of the gene is relatively high compared to postoperative recurrence-free cases, or SEQ ID NOs: 6-7, Lung adenocarcinoma recurrence when the expression of a gene comprising any nucleotide sequence of 10-11, 13, 16, 18, 22, 29, 34, 60, 69 is relatively low compared to post-operative recurrence-free cases Is expected to be high And to the method.   The method according to claim 1, wherein the number of genes in the gene set is 5-82.   The method according to claim 1, wherein the expression of the gene set is measured by measuring the presence or amount of a nucleic acid or protein corresponding to the gene.   The method according to claim 1, wherein the expression of the gene set is measured by a hybridization method or a quantitative PCR method.   The method according to claim 4, wherein the hybridization method is a microarray method or a blot method.   The method according to claim 4 or 5, wherein the hybridization method is performed using a nucleic acid that hybridizes with an RNA transcript, cRNA, aRNA or cDNA derived from the gene.   The method according to claim 1, wherein the expression of the gene set is measured by an immunological method.   The method according to claim 7, wherein the immunological method is performed using an antibody against a protein encoded by the gene or a fragment thereof. A composition for predicting recurrence of lung adenocarcinoma after surgical resection of lung cancer in a patient with lung adenocarcinoma, wherein the composition comprises two or more different genes comprising any nucleotide sequence of SEQ ID NOs: 1 to 82 Comprising a plurality of nucleic acids that allow measuring the expression of
(1) a nucleotide sequence represented by SEQ ID NOs: 1 to 82, a nucleotide sequence complementary thereto, or a nucleic acid having a partial sequence of 15 bases to less than the total number of bases in these nucleotide sequences, and (2) SEQ ID NO: A nucleic acid that hybridizes with a nucleic acid comprising the nucleotide sequence shown in 1-82 or a nucleotide sequence complementary thereto, or a nucleic acid having a partial sequence of less than the total number of bases from 15 consecutive bases of the hybridizing nucleic acid,
The composition is selected from the group consisting of:   The composition according to claim 9, wherein the composition is in the form of a kit, a microarray or a primer set.   The composition according to claim 9, wherein the nucleic acid enables measurement of the expression of 5 to 82 genes.   The composition of claim 9 for use in the method of claim 1.   A composition for predicting recurrence of lung adenocarcinoma after surgical resection of lung cancer in a patient with lung adenocarcinoma, wherein the composition comprises two or more different genes comprising any nucleotide sequence of SEQ ID NOs: 1 to 82 A plurality of antibodies that make it possible to measure the expression product of the protein encoded by any one of the nucleotide sequences shown in SEQ ID NOs: 1 to 82 or the amino acid sequence of the protein at least 8 consecutive The composition is selected from the group consisting of an antibody that immunologically reacts with a polypeptide comprising amino acid residues, and a fragment of the antibody.   The composition of claim 13, wherein the composition is in the form of a kit.   14. The composition of claim 13, wherein the antibody allows measurement of expression products of 5 to 82 genes.   14. A composition according to claim 13 for use in the method of claim 1.