KR102055305B1 - Markers for diagnosis and targeted treatment of adenocarcinoma of gastroesophageal junction - Google Patents

Abstract

Provided are algorithms, markers, and therapeutic targets for diagnosing gastroesophageal border cancer as esophageal or gastric / body cancer. The marker includes ERBB2 or EGFR.

Description

Markers for diagnosis and targeted treatment of adenocarcinoma of gastroesophageal junction

A diagnosis and targeted treatment of gastroesophageal adenocarcinoma.

According to a US SEER database report, esophageal adenocarcinoma has increased rapidly, with a 600% increase in the last 30 years. In particular, most of these growth rates were lower esophageal adenocarcinoma of the lower 1/3 esophagus, and gastric esophageal border cancer also showed a nearly double rate of increase (Pohl H and Welch HG, 2005). According to a report from the National Cancer Center, gastroesophageal adenocarcinoma increased nearly fivefold from 2.3% in 1962-1965 to 10.0% in 2001-2005, especially cardiac cancer (Siewert type II). This share more than doubled from 28.5% to 57.3%.

According to a recent report from the Korean Gastric Cancer Society, the upper gastric cancer, including gastroesophageal border adenocarcinoma, has increased markedly from 11.2% in 1995 to 16.0% in 2014 (Information Committee of Korean Gastric Cancer Association. 2016).

Unlike gastroesophageal gastric adenocarcinoma, which has been commonly observed in Korea, the gastroesophageal border adenocarcinoma still lacks academic consensus on tumor classification, its biological characteristics, and surgical methods, as well as hypothesis on the tumor carcinogenesis mechanism. Until recently, various discussions have been made. However, there is still insufficient research on the biological characteristics that can distinguish gastroesophageal adenocarcinoma from esophageal adenocarcinoma or upper gastric cancer. Actually, these three adjacent carcinomas are expressed with some molecular biological differences and which genes are most effective. There are no clear results as to whether it can be a therapeutic target.

In particular, most of the carcinomas in this position require a very long time of surgery along with gastric resection, which is a poor prognostic factor that increases the recurrence rate by greatly reducing the compliance of the systemic nutrition and chemotherapy after surgery. Works. Therefore, the need for research on the selection of a new diagnostic method or treatment for chemotherapy or targeted therapy before surgery is emphasized.

Recently, large-scale molecular biological analysis of gastric and esophageal cancers was published by the Cancer Genome Atlas group (TCGA), but there are limitations in practical applications such as survival rate. The definition is vague and there is no direct comparative analysis of similarities or differences with esophageal adenocarcinoma in terms of therapeutic targets.

One aspect provides a method comprising measuring the expression level of EGFR or ERBB2 in gastric or esophageal tissue isolated from an individual to provide information necessary for the diagnosis of gastric or esophageal cancer.

Another aspect is a probe, primer set, which specifically binds an antibody, antigen binding fragment, or polypeptide that binds specifically to an EGFR or ERBB2 protein, or a fragment thereof, or a polynucleotide sequence encoding an EGFR or ERBB2 protein, Or a composition for diagnosing gastric cancer or esophageal cancer of a subject comprising a nucleotide and a kit comprising the same.

One aspect provides a method comprising measuring the expression level of EGFR or ERBB2 in gastric or esophageal tissue isolated from an individual to provide information necessary for the diagnosis of gastric or esophageal cancer.

The term “epidermal growth factor receptor (EGFR)” (also referred to as ErbB1 or HER1) is a transmembrane protein or gene encoding it that acts as a receptor for the membrane of the epidermal growth factor family of extracellular protein ligands. Point to. Its sequences are known in the art, for example Genbank Accession No. AAI18666.1 (https://www.ncbi.nlm.nih.gov/genbank/).

The term "ErbB2" (also referred to as human epidermal growth factor receptor 2, or Neu) refers to a receptor tyrosine kinase protein belonging to the ErbB family or a gene encoding the same. Its sequences are known in the art, for example Genbank Accession No. NP_001005862 (https://www.ncbi.nlm.nih.gov/genbank/).

The method relates to the diagnosis and / or targeted treatment of gastric or esophageal cancer. The gastric cancer or esophageal cancer is gastroesophageal adenocarcinoma (adenocarcinoma of gastroesophageal junction), cardia cancer, gastroesophageal border / gate cancer, gastric cancer located at fundus, gastric cancer located at fundus, gastric cancer (gastric cancer located at body), gastric adenocarcinoma located at fundusor body of the stomach (GCFB), esophageal adenocarcinoma (EAC), or a combination thereof. The gastric cancer or esophageal cancer may be classified as adenocarcinoma. The stomach and the esophagus are anatomically connected, and the boundary between the stomach and the esophagus is vague and there is a constant debate about the biological understanding of gastroesophageal border adenocarcinoma and the comparison of molecular biological characteristics with gastric or esophageal adenocarcinoma. However, an embodiment of the present invention can be usefully used for the diagnosis of such cancers accurately using molecular biological markers, and also for the treatment of targeting the markers.

The step of measuring the expression level of EGFR or ERBB2 may be a step of confirming the amount of mRNA of EGFR or ERBB2, a step of confirming the amount of protein, or a step of identifying the number of gene copies.

Measuring the expression level may include contacting a tissue sample with a substance that specifically binds to an EGFR or ERBB2 protein or an mRNA encoding the same, to form a complex, and measuring the level of the formed complex. Thereafter, measuring the level of the complex may be achieved by, for example, attaching a detectable label to the specifically binding substance and detecting a signal therefrom. The measurements are RT-PCR, competitive RT-PCR, Real-time RT-PCR, RNase protection assay (RPA), Northern blotting, Nucleic acid microarrays including DNA, western blotting, enzyme linked immunosorbent assay (ELISA), radioimmunoassay (RIA), radioimmunodiffusion, Ouchterlony immunodiffusion, rocket immunity Electrophoresis, tissue immunostaining, immunoprecipitation assay, complement fixation assay, FACS, mass spectrometry, magnetic bead-antibody immunoprecipitation, protein chip, or combinations thereof Can be.

In addition, measuring the expression level may include reverse phase protein analysis, tissue microarray, immunohistochemistry (IHC), silver in situ hybridization (SISH), Or a combination thereof.

The method may comprise comparing the expression level of EGFR or ERBB2 measured with the expression level of EGFR or ERBB2 measured in the control. The control group may be tissue of an individual who does not have gastric cancer or esophageal cancer, or gastric cancer or esophageal cancer tissue.

The method may also include determining that the subject has gastric cancer or esophageal cancer when the expression level of EGFR or ERBB2 changes compared to the expression level of the control group. The change was 1%, 2%, 3%, 4%, 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100% compared to the control group. , 200%, 300%, 400%, 500%, 600%, 700%, 800%, 900% and 1000% or more, 1%, 2%, 3%, 4%, 5%, 10%, 20 %, 30%, 40%, 50%, 60%, 70%, 80%, 90% and 100% or more reductions. For example, when the number of copies of the ERBB2 gene of tissues collected from an individual has increased (amplified) by 100% or more compared to normal tissues, the individual may be determined to have esophageal cancer.

The method may further include determining to administer an inhibitor of EGFR or ERBB2 when the expression level of the EGFR or ERBB2 is increased. The inhibitor of EGFR or ERBB2 can be used as long as it is commonly used in the art, commercially available, or under clinical trial, for example, lapatinib, trastuzumab, Cetuximab, gefitinib, epitinib, or a combination thereof.

Another aspect is a probe, primer set, which specifically binds an antibody, antigen binding fragment, or polypeptide that specifically binds an EGFR or ERBB2 protein, or a fragment thereof, or a polynucleotide sequence encoding an EGFR or ERBB2 protein, Or compositions and kits for diagnosing gastric or esophageal cancer in a subject comprising nucleotides.

Probes, primer sets, or nucleotides that specifically bind to an antibody, antigen-binding fragment, or polypeptide that specifically binds to the EGFR or ERBB2 protein, or fragments thereof, or to a polynucleotide sequence encoding an EGFR or ERBB2 protein. Can be prepared and screened by the practitioner according to methods known in the art, such as high throughput screening (HTS), or may be commercially available.

The kit may also provide a kit form comprising a packaging unit having one or more reaction reagents. The kit may also include one or more of the following items: substances, buffers, instructions, and positive or negative controls for detecting the subject of detection. The kit may comprise containers of reaction reagents mixed together in an appropriate proportion to carry out the methods described herein. The reaction reagent vessels may preferably include a unit quantity of reaction reagent so that the step of measuring when performing the method may be omitted.

In one embodiment of the present invention, esophageal adenocarcinoma was significantly associated with differentiation and colon cancer, gene mutation of ERBB2 was significantly amplified, and protein expression of ERBB2 and EGFR was significantly increased. Since the expression level of ERBB2 or EGFR is measured, esophageal cancer or gastric / gastric cancer can be distinguished and diagnosed, and a useful target for the treatment of these cancers can be provided.

Another aspect includes measuring an expression profile of a gene of a tissue isolated from the individual to provide information for diagnosing gastric or esophageal cancer of the individual; And obtaining a BCCP score using a computer system executing a Bayesian Compound Covariate Predictor (BCCP) algorithm from the gene expression profile .

The gene may include the following 400 genes: KRT5, KRT13, KRT6A, KRT14, RHCG, KRT4, S100A7, SPRR3, KRT6C, SPRR2A, MUC21, KRT6B, PKP1, SERPINB3, CLCA2, SPRR1B, MUC4, TBX5 , CRNN, SPRR2D, DUOX2, GJB6, SPRR2E, S100A2, KRT16, KRT17, SCEL, DSC3, CALML3, HEPHL1, A2ML1, NKX6-1, DUOXA2, SBSN, TMPRSS11D, CLCA4, ANXA8, FGFBP1, TDU3, DGM , EREG, SERPINB4, KRT78, HOXC10, DUOXA1, ZNF750, SPRR1A, KRT7, TMPRSS11A, CXCL6, KLK13, CRCT1, PAX9, HORMAD1, FAM83A, S100A7A, GPR110, CAPN14, ABCA12, WSSDCB MCE13 , FAT2, IL1RL2, STK31, SPINK5, DDX3Y, PI3, FAM83C, LYPD2, KLK5, IL1F5, KRT15, BBOX1, GPR87, ACTL8, SLC34A2, USP9Y, TGM3, LASS3, TMPRSS13, GUCY1BRT, SPRRDF K23 , SLC5A12, ABCA13, S100A8, GBP6, CEACAM7, IL1F9, MUC15, PPP1R14C, KLK7, DEFB1, MMP3, TM4SF20, SYCP2, TMEM40, CEACAM6, VWA2, CYorf15B, TDRD5, DCDC2, KLKR12, C10, KLKOX12, C1014 , DNAH2, SLC15A1, KLK8, LGALS7B, ALDH1L1, FO XN1, RAET1L, TP63, ATP13A4, UTY, DSG1, ALOX12B, AMY2B, SAA2, SAA1, GPR115, LCN2, KRT24, PRKY, PADI3, SCNN1A, PGLYRP3, PRSS27, ANXA8L2, WISP3, IL20RB, FOX, NCCRP1, SLC44A5, NMU, CDH17, RNASE7, FER1L4, C4BPB, CALML5, CXorf61, SPACA4, ULBP2, KRTDAP, MYO7B,, C7orf54, SLC6A20, LOC221442, GOLGA8B, GPR81, TRLD125 PA2, ALDH3B2 C21 , LOC440905, POU5F1, LOC100131726, KCNH8, SPRR2C, IL13RA2, C3orf67, SAA4, LYPD3, GJB5, ULBP1, PHACTR3, S100A9, TRIM54, KLK11, ITIH5L, BNIPL, LOC284233, LOPL42P5646, FLJ387H , COL7A1, GDA, ARL14, VIP, CLC, PPP2R2B, FABP4, SFRP5, CCL8, CD79A, HIST1H1E, CCL4L2, SYNPO2, SLC47A1, CXCR2P1, C1QTNF7, CDH2, SIGLEC7, GZLR, PNK56 , SOX10, FCGR1A, ST6GALNAC5, CCL4, ACSM5, FOLR2, KCNA1, CCL26, FCGR1C, VSIG4, C1QA, NCF1B, NXPH3, C1QC, TAGLN, PTPRN, TYROBP, ISL1, HCST, RGMA, APNDCL3 , PNMA6A, CD8B, FCGR1B, PCBP3, CD8A, GPR27, DPP6, F13 A1, EPYC, CXCL9, GREM2, SNORA12, DNAJC5B, HLA-DQA2, PLP1, CHGB, APOC2, FAT3, TLR7, CHRDL1, LY86, AGTR1, ISLR, CR1, FCGR3A, DPEP1, PPAPDC1A, ODZ3, MAPF4, CAPJ4, CAPJ4 MSR1, RELN, APOE, C1QB, PHYHIPL, CCDC80, NCF1C, SLITRK5, TREM2, FGF14, PGA5, NKX2-3, PRELP, FCRLA, CCL5, GNAO1, PCDH9, WSCD2, MS4A1, RNF150, KLRD1, KCNK2 IGFBPL1, IGF1, CDO1, COLEC12, FIBIN, CARTPT, VPREB3, PPP1R1A, GDF6, NCAM2, CLEC10A, KIAA0408, BHLHE22, CD52, SULT4A1, SIT1, FNDC1, GRIK3, SUCNR1, TMEMBB, MYP3, MYP3 CADM3, SPOCK1, GREM1, SIGLEC8, ADCY2, CDC26, OGN, FCRL6, PSD, VIPR2, GZMM, ADH1B, CCL19, PLD4, TMEM189-UBE2V1, NKG7, EOMES, STMN2, GZMH, NPTX1, CNTFR1, TMEM130, CTMEM1 ECEL1, ACTG2, COL10A1, ST6GAL2, NRK, PI16, NALCN, GZMK, NCAM1, RMRP, CHRM2, KCNMA1, HSPB7, SLIT2, PDZRN4, CNN1, CHRDL2, OMD, PNNIS, RSPO3, PLA2G2D, SMNN31, SMXN3 HSPB6, C16orf89, PGA3, COMP, C2orf40, C7, GKN2, DES, CILP, COL11A1, LIPF, GATA5, MFAP5, SCRG1, SFRP2, GKN1, PGA4, CCL21, SIX2, NKX3-2, SFRP4, NBLA00301, THBS4, HAND2, or BARX1.

As regards the 400 genes, the compositions, kits, methods of measurement, etc. may be applied to the contents mentioned in the description of the EGFR or ERBB2 unless otherwise specified.

ERBB2 and EGFR are provided as markers and targets for diagnosing gastroesophageal border cancer.

1 shows the anatomical distribution of a study group of TCGA cohorts.
2 shows the anatomical distribution of a study group of SNU cohorts.
3 shows a class prediction model using BCCP.
4 shows a detailed study group according to the analysis summary.
5 shows unsupervised clustering of 5,520 genes between EAC and GCFB of the TCGA cohort.
6 shows the results of performing unsupervised hierarchical clustering of EAC and GCFB in the TCGA cohort using 400 signature gene classification.
7 shows the ROC curve after cross-validation using LOOCV.
8 shows hierarchical clustering of GEJ / segments of a TCGA cohort using BCCP.
9 shows hierarchical clustering of AGEJ or upperthird gastric cancer of the SNU cohort using BCCP.
10 shows the change in replication number between the EAC-prediction and GCFB-prediction groups of the TCGA cohort.
Figure 11 shows the change in replication number between the EAC-prediction and GCFB-prediction groups of the SNU cohort.
12 shows a heat map using reversed phase protein analysis of the TCGA cohort.
13 shows protein expression using IHC staining of tissue microarrays (200 ×).
14 shows the composite H scores of tissue microarrays between the EAC-prediction group and the GCFB-prediction group of the SNU cohort.
15 illustrates external verification of a prediction model using a CCLE database.
16 illustrates hierarchical clustering of CCLE databases between EAC-prediction groups and GCFB-prediction groups.
17 shows the drug response of lapatinib using IC50 data from the CCLE database between the EAC-prediction group and the GCFB-prediction group.

Hereinafter, the present invention will be described in more detail with reference to Examples. However, these examples are for illustrative purposes only and the scope of the present invention is not limited to these examples.

1. Materials and Methods

(One) TCGA Cohort

The Cancer Genome Atlas (TCGA) (https://tcga-data.nci.nih.gov/tcga/) and mRNA expression, somatic mutations, insertions / deletions, replication number changes, pure esophageal adenocarcinoma Adenocarcinoma, gastroesophageal junction or cardiac adenocarcinoma and reverse gas phase of pure gastric adenocarcinoma located at fundusor body of the stomach (GCFB) located in the fundus or body The retrieved information of the protein array (RPPA) was examined. The anatomical location of each of the above mentioned human body parts is as shown in FIG.

(2) Seoul National University Cohort

A clinical information gathered from the Gastric Cancer Biology Laboratory, Cancer Research Institute, and Seoul National University, 1999-2015, to investigate a fresh frozen tissue storage database containing clinical information about the AGEJ and upper third of the stomach. It was. The storage of the fresh tissue was approved by the Organizational Review Board of Seoul National University Hospital (IRB No: H-0806-072-248). Patients with other primary malignancies, recurrent adenocarcinoma or residual gastric cancer at the time of initial diagnosis were excluded. Adenocarcinoma of AGEJ and gastric UT in the cohort of Seoul National University Hospital was classified using distance criteria from the esophageal junction. The anatomical location of each of the above mentioned human body parts is as shown in FIG. 2.

AGEJ II was defined as a tumor located within 1 cm from the gastroesophageal border to the oral cavity and 2 cm opposite the oral cavity. This is the same as the general definition of Siewert type II cancer.

AGEJ III was defined as a tumor located 2-5 cm from the gastroesophageal border, opposite the oral cavity, with or without gastroesophageal border. One third of the upper gastric adenocarcinoma except AGEJ II or AGEJ III was defined as UT. All tumors classifiable with AGEJ II were investigated and next-generation sequencing was performed. AGEJ III and UT examined the most recent equal number of samples. Pathological stage was diagnosed according to 7th AJCC TNM classification. For pathological analysis, papillary, well-differentiated or moderately differentiated types were classified into differentiation groups, and poorly differentiated, poorly mucus, cohesive cell types were classified into non-differentiated groups. This research protocol was approved by the Institutional Review Board of Seoul National University Hospital (IRB No: H-1501-027-639).

(3) nucleic acid treatment

Each frozen tumor and the corresponding normal gastric mucosa from the fresh tissue reservoir of the SNU cohort were prepared to a size of about ² × ² × 1 mm ³ . DNA was extracted using a Qiagen DNA extraction kit (Qiagen, Venlo, Netherlands) using a spin-column protocol. The extracted DNA was quantified using a QUBIT HS dsDNA assay (Life Technologies Gaithersburg, MD, USA) at a minimum of A260 / 280 ≥ 1.7 and amount of dsDNA ≥ 3.0 μg. Isolation of RNA was performed in 5.0 mL of Eppendorf Tubes according to the protocol provided by the manufacturer of TRIzol (user manual TRIzol® Reagent (www.invitrogen.com)). 1 mL TRIzol was added to each fresh tissue for dissolution. 1 mL of starting material was transferred to each tube according to the TRIzol protocol, and 200 μL chloroform was added to each tube. 0.5 mL of isopropanol and 1 mL ethanol (75%) were used in the washing step below to precipitate RNA from the water phase. RNA precipitation and washing steps were performed in 5.0 mL tubes at 12,000 × g. The resulting RNA pellet was resuspended in 50 μL EDPC-treated water. Using NanoDrop ^™ 1000 (Thermo Scientific), OD was measured at 260 nm and 280 nm and the concentration and purity of the sample were determined.

(4) SNU Cohort  all Transcript ( whole transcriptome Sequencing

All tumor samples from the SNU cohort were used to make the entire transcript library using the Illumina Truseq RNA library preparation kit (Ribo-Zero rRNA Removal Kit). All libraries were sequenced using one sample per lane on the Illumina HiSeq2000 platform (a paired-end 2 × 101 bp read length). Tumor RNA and corresponding normal RNA were loaded into the same usually same flow cell. Decryption alignment and processing was performed as a GATK best practice recommendation using STAR aligner and Broad Institute's Picard (http://broadinstitute.github.io/picard).

(5) SNU Cohort  all Exome ( exome Sequencing

Whole exome sequencing of at least 3 μg of dsDNA from tumors and corresponding normal gastric mucosal samples was performed using the Agilent SureSelectHumanAll Exon V5 + UTR region kit. Paired-end pairs of 2 × 101 bp read length were sequenced on the Illumina HiSeq2000 platform. Sequencing target depth was planned to be at least 100 × in both tumor and normal tissue (ideally 200 × for tumor). Decryption alignment and processing was performed as a GATK best practice recommendation using Burrows-Wheeler Aligner (BWA) -mem and Broad Institute's Picard.

(6) Transcript  Predictive Classification Algorithm Using Sequencing

The BRB Array Tool was used to analyze gene expression data. RNA-sequencing data from the TCGA and SNU cohorts were analyzed together to build differentially expressed genes (DEGs) and prediction models. First, DEGs between the EAC and GCFB in the TCGA cohort were identified by Student's t test (P <0.001) and further screened according to whether the fold change was higher or lower. To build the prediction model, we used the leave-one-out cross validation (LOOVC) approach with the previously established Bayesian Compound Covariate Predictor (BCCP) algorithm. The method used at this time is Radmacher MD, McShane LM, Simon R. A paradigm for class prediction using gene expression profiles. J Comput Biol 2002; 9: 505-511, Wright G, Tan B, Rosenwald A et al. A gene expression-based method to diagnose clinically distinct subgroupsof diffuse large B cell lymphoma. Proc Natl Acad Sci U S A 2003; 100: 9991-9996, YOUDEN WJ. Index for rating diagnostic tests. Cancer 1950; 3: 32-35. Robin X, Turck N, Hainard A et al. pROC: an open-source package for R and S + to analyze and compare ROC curves. BMC Bioinformatics 2011; 12:77. doi: 10.1186 / 1471-2105-12-77.: 77-12 et al., which are incorporated herein by reference in their entirety (FIG. 3).

The sensitivity and specificity of the trained model were assessed by the ROC curve. The optimal cut-off value between EAC and GCFB was determined using the Youden Index. External validation of predictive models and their cut-off values is performed using RNA microarrays of gastric and esophageal adenocarcinoma cell lines from the Cancer Cell Line Encyclopedia (CCLE) database (http://www.broadinstitute.org/ccle). It was performed by. Depending on the possibilities of the BCCP model, all tumors of the GEJ / gateway tumor of the TCGA cohort and the SNU cohort were reclassified according to genomic subtypes (EAC-prediction or GCFB-prediction). Differences in genomic subtypes at the clinical and molecular levels were then assessed by analyzing clinical pathological data, mutations, and copy number changes. Pathway analysis was performed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis tool (http://www.kegg.jp). Potential surrogate markers associated with genomic subtypes were identified by reverse phase protein analysis for the TCGA cohort and tissue microarrays for the SNU cohort. Target drug reactivity of these agents was compared using the IC50 of the gastric and esophageal adenocarcinoma cell lines of the CCLE database.

(7) Identification of somatic mutations and insertions / deletions

The TCGA cohort analyzed somatic mutations, including insertions / deletions, using methods previously reported in the art.

The SNU cohort used BAM files for total exome sequencing using Mutect and IndelGenotyper for somatic mutationcalling. 1) exon and splicing modifications based on a reference sequence, or 2) variations in cases with more than 8 sequences and more than 4 alternative alleles.

Within each TGCA and SNU cohort, somatic mutations and insertions / deletions were compared between the EAC-prediction group and the GCFB-prediction group.

(8) Somatic Cloning Number Analysis

The TCGA cohort analyzed copy numberalteration (CNA) using single base polymorphism (SNP) analysis using methods previously reported in the art.

The SNU cohort analyzed CNAs using total exome data based on Read Per Kilobase per Million mapped reads (RPKM) values from CONIFER.

Somatic CNA analysis was performed using the GISTIC 2.0 algorithm for both TCGA and SNU cohorts. Of the genes identified as having local copy number amplification using the GISTIC algorithm, in at least one pair of samples, candidate genes with ≧ 1 greater than the value of the corresponding normal gastric mucosa were selected for at least one pair of samples.

The number of copies of these candidate genes in the EAC-prediction and GCFB-prediction subgroups was compared by Student's t assay.

(9) TCGA Cohort  Reversed-phase protein array

Reverse phase protein array (RPPA) data of 132 of 180 cases (including 44 EAC and 88 GCFB) were obtained from the TCGA database. Clustering analysis was performed after recentered normalization.

10 SNU Cohort  group Microarray ( TMA )

A core tissue biopsy (2 mm in diameter) was obtained from each paraffin-fixed gastric tumor (donation block) and listed in a new receiving paraffin block using a trephine device (Superbiochips Laboratories, Seoul, Korea). Array blocks). Three sets of tissue microarrays were prepared for each sample, taking into account the histological components of the cancer and the diversity of the molecular abnormalities. Target drug response was assessed using IC50 data from the CCLE database for gastric and esophageal adenocarcinoma cell lines. Tissue array blocks included up to 46 cores on three arrays for a total of 138 cases and were subjected to immunohistochemistry (IHC) staining. Tumors with more than 10% core area were considered appropriate. Each paraffin block contained an internal control. IHC was performed using an automated immunostainer (BenchMark XT, Ventana Medical Systems, Tucson, AZ, USA). After taking samples from each core, the staining intensity was 0 (no membrane staining, negative), 1+ (weak / nearly unstained, weak positive), 2+ (weak-medium staining, medium positive), And 3+ (strong staining, strong positive). All IHC staining and silver in situ hybridization (SISH) for each TMA core were evaluated and scored by one expert pathologist who did not know about all clinical information. Staining status for all proteins except ERBB2 was analyzed using complex H-scores by multiplying the staining intensity by the percentage of cells stained with the individual H-scores for each intensity level. Staining status of ERBB2 was considered positive at the strongest staining intensity score when at least 10% of cells were stained in at least one TMA core. Final interpretation using the IHC and SISH results was performed according to previously reported methods. Expression of ERBB2 was determined positive when IHC 3+ or IHC 2+ and SISH Black / Red ratio ≧ 2.0, negative when IHC <2+, or IHC 2+ and SISH Black / Red ratio <2.0. Target drug response was assessed using IC50 data from the CCLE database for gastric and esophageal adenocarcinoma cell lines.

(11) statistical analysis

Comparative analysis was performed using the Student's t test and the chi-squared test. Survival analysis was performed by Kaplan-Meier method and log rank test. Multivariate analysis to identify risk factors for protein expression was performed using binary logistic regression or linear regression analysis. All assays were performed with a two-sided assay and at a significance level of 5% using SPSS version 21.0 (SPSS, Inc., Chicago, IL, USA).

2. Results

4 shows a detailed study group according to the analysis plan. Normal mucosa of AGEJII, AGEJIII, and UT from the SNU cohort, corresponding to pure esophageal adenocarcinoma from the TCGA cohort, pure gastric adenocarcinoma (base / body), GEJ / gateline adenocarcinoma 228 tumors, and 46 pairs (92 samples) of tumors Was analyzed.

For the SNU cohort, nucleic acid extraction from fresh frozen tissues and libraries was repeated before raw sequencing data was obtained.

(1) clinical pathological characteristics

In the TCGA cohort, 78 EAC, 48 GEJ / segment and 102 GCFB samples were identified that could be used for exome and transcript data (Table 1).

Clinical Pathologic Characteristics of the TCGA Cohort EAC
(n = 78) GEJ / department
(n = 48) GCFB
(n = 102) P
Value Last name (male: female) 69: 9 37:11 57:45 <0.001 age 66.8 ± 12.0 66.9 ± 9.2 66.6 ± 9.3 0.985 location Middle esophagus 2 (2.6%) 0 0 <0.001 Esophagus-Horse 1 (1.3%) 0 0 Esophageal extremities 75 (96.2%) 0 0 GEJ / department 0 48 (100%) 0 Upper and Lower Body 0 0 102 (100%) WHO classification Papilloma 0 4 (8.3%) 12 (11.8%) Tubular 0 23 (47.9%) 49 (48.0%) Low cohesion 0 9 (18.8%) 19 (18.6%) Mucus 0 3 (6.3%) 4 (3.9%) Mixed 0 6 (12.5%) 6 (5.9%) Not available 78 (100%) 3 (6.3%) 12 (11.8%) Lauren classification chapter 0 32 (66.7%) 70 (68.6%) Diffusion 0 9 (18.8%) 19 (18.6%) Mixed 0 6 (12.5%) 6 (5.9%) Not available 78 (100%) 1 (2.1%) 7 (6.9%)

The pathologic characteristics of the SNU cohort are shown in Table 2 below.

Clinical Pathologic Characteristics of the SNU Cohort AGEJⅡ
(n = 16) AGEJⅢ
(n = 16) UT
(n = 14) P
Value Last name (male: female) 13: 3 12: 4 11: 3 0.912 age 58.5 ± 10.4 66.5 ± 9.4 63.5 ± 8.1 0.062 WHO classification differentiation 7 (43.8%) 7 (43.8%) 7 (50.0%) 0.919 Non-differentiation 7 (43.8%) 8 (50.0%) 5 (35.7%) Crystal x 2 (12.5%) 1 (6.3%) 2 (14.3%) Lauren classification chapter 5 (31.3%) 6 (37.5%) 7 (50.0%) 0.526 Diffusion 7 (43.8%) 5 (31.3%) 6 (41.9%) Mixed 4 (25.0%) 5 (31.3%) 1 (7.1%)

(2) Predictive  Development of Classification Model

Unsupervised hierarchical clustering of EAC and CGFB in the TCGA cohort revealed 5,520 genes (FIG. 5) (P <0.001).

The top 200 and bottom 200 genes were selected as the 400 signature genes according to the grade of fold change. The list of selected genes is as follows: KRT5, KRT13, KRT6A, KRT14, RHCG, KRT4, S100A7, SPRR3, KRT6C, SPRR2A, MUC21, KRT6B, PKP1, SERPINB3, CLCA2, SPRR1B, MUC4, TBX5, CRNN, SPRR2D DUOX2, GJB6, SPRR2E, S100A2, KRT16, KRT17, SCEL, DSC3, CALML3, HEPHL1, A2ML1, NKX6-1, DUOXA2, SBSN, TMPRSS11D, CLCA4, ANXA8, FGFBP1, TGM1, IVL, DUB1REG, DSG3 KRT78, HOXC10, DUOXA1, ZNF750, SPRR1A, KRT7, TMPRSS11A, CXCL6, KLK13, CRCT1, PAX9, HORMAD1, FAM83A, S100A7A, GPR110, CAPN14, ABCA12, WFDC2, TMPRSS1B, CE2CAMRP, UC12, SERMAL STK31, SPINK5, DDX3Y, PI3, FAM83C, LYPD2, KLK5, IL1F5, KRT15, BBOX1, GPR87, ACTL8, SLC34A2, USP9Y, TGM3, LASS3, TMPRSS13, GUCY1B2, SPRR2F, KDM5D1, KRT5D1, KRT5D12 S100A8, GBP6, CEACAM7, IL1F9, MUC15, PPP1R14C, KLK7, DEFB1, MMP3, TM4SF20, SYCP2, TMEM40, CEACAM6, VWA2, CYorf15B, TDRD5, DCDC2, KLK12, C10orf99, CARD14HC64C13 KLK8, LGALS7B, ALDH1L1, FOXN1, RAET1L, TP63, ATP13A4, UTY , DSG1, ALOX12B, AMY2B, SAA2, SAA1, GPR115, LCN2, KRT24, PRKY, PADI3, SCNN1A, PGLYRP3, PRSS27, ANXA8L2, WISP3, IL20RB, FOXE1, ZFY, ALG1L, LY6K, NCCRP1, NCCH1 , FER1L4, C4BPB, CALML5, CXorf61, SPACA4, ULBP2, KRTDAP, MYO7B,, C7orf54, SLC6A20, LOC221442, GOLGA8B, GPR81, ALDH3B2, TRIM29, C21orf125, BNC1, IL1F6C6, PA6, TR6F6 SPRR2C, IL13RA2, C3orf67, SAA4, LYPD3, GJB5, ULBP1, PHACTR3, S100A9, TRIM54, KLK11, ITIH5L, BNIPL, LOC284233, LOC387646, MRPL42P5, FLJ42393, PCDHAC2, IFNE, BCL454, IFN, BCL454 CLC, PPP2R2B, FABP4, SFRP5, CCL8, CD79A, HIST1H1E, CCL4L2, SYNPO2, SLC47A1, CXCR2P1, C1QTNF7, CDH2, SIGLEC7, GZMA, NKX6-3, PGM5, LILRB4, NGCL1, NMURAOX ACSM5, FOLR2, KCNA1, CCL26, FCGR1C, VSIG4, C1QA, NCF1B, NXPH3, C1QC, TAGLN, PTPRN, TYROBP, ISL1, HCST, RGMA, APOBEC3H, PRND, ADIPOQ, CCL11, GLP2R, CD8B, PNMA6 CD8A, GPR27, DPP6, F13A1, EPYC, CXCL9, GREM2, SNORA12 , DNAJC5B, HLA-DQA2, PLP1, CHGB, APOC2, FAT3, TLR7, CHRDL1, LY86, AGTR1, ISLR, CR1, FCGR3A, DPEP1, PPAPDC1A, ODZ3, MAPK4, C17orf87, KCNJ5, MSB1, RHPL1, RHIPL, RHIPL , CCDC80, NCF1C, SLITRK5, TREM2, FGF14, PGA5, NKX2-3, PRELP, FCRLA, CCL5, GNAO1, PCDH9, WSCD2, MS4A1, RNF150, KLRD1, KCNK2, MATK, HUNK, IGFBPL1, IFIBIN, CDO , CARTPT, VPREB3, PPP1R1A, GDF6, NCAM2, CLEC10A, KIAA0408, BHLHE22, CD52, SULT4A1, SIT1, FNDC1, GRIK3, SUCNR1, TMEM90B, CYP1B1, MKX, SV2B, BMP3, TEMALOCK, SIC2, GL2 , CDC26, OGN, FCRL6, PSD, VIPR2, GZMM, ADH1B, CCL19, PLD4, TMEM189-UBE2V1, NKG7, EOMES, STMN2, GZMH, NPTX1, CNTFR, TMEM130, CTNND2, ITGBL1, ECEL1, ACTG2, COL10A , PI16, NALCN, GZMK, NCAM1, RMRP, CHRM2, KCNMA1, HSPB7, SLIT2, PDZRN4, CNN1, CHRDL2, OMD, PTGIS, RSPO3, PLA2G2D, SMYD1, ZNF683, NRXN3, PCSK1N, HfB89, CSP3 , C7, GKN2, DES, CILP, COL11A1, LIPF, GATA5, MFAP5, SCRG1, SFRP2, GKN1, PGA4, CCL21, SIX2, NKX3-2, SFRP4, NBLA00301, THBS4, HAND2, and BARX1.

 Unsupervised hierarchical clustering of EAC and GCFB in the TCGA cohort was performed using the 400 signature gene classifications, confirming the distinct cluster classification between EAC and GCFB (FIG. 6).

Predictive classification models were developed based on BCCP with 400 gene classifications and trained with LOOCV. The ROC curve using the BCCP score revealed a 0.4535 Youden index as the cut-off value that distinguishes the area under curve of 0.957 (95% confidence interval = 0.93-0.98) and EAC and GCFB (FIG. 7). .

The cut-off value showed a sensitivity of 90.2% and a specificity of 89.7% for predicting EAC. For 400 signature genes, KEGG pathway analysis was performed. The PI3K-AKT signaling pathway associated with GCFB was identified among several cancer-associated pathways containing five or more genes, including CHRM2, COMP, FGF14, IGF1, PPP2R2B, RELN, and THBS4, among 200 genes overexpressed in GCFB. Included. As a result, PI3K and AKT were considered for protein validation using tissue microarrays from the RPPA and SNU cohorts of the TCGA cohort.

(3) using somatic mutation analysis Predictive  Test of classification model

Using the BCCP score with a cut-off value of 0.4535, the clustering of the GEJ / segment of the TCGA cohort was tested (FIG. 8).

Hierarchical clustering of the GEJ / segment of the TCGA cohort showed spectral transitions of the clusters between the EAC-prediction and GCFB-prediction groups, with no fully distinguishable clusters. The GEJ / segment of the TCGA cohort predicted by the EAC was 15/48 (31.2%) and the GCFB predicted 33/48 (68.8%). In terms of somatic mutation, there was no significant difference in TP53, PIK3CA, RHOA, KRAS, and ARID1A between the EAC-prediction group and the GCFB-prediction group of the TCGA cohort. Clustering of AGEJ II, AGEJ III, and UT of the SNU cohort tested spectral transitions of clusters similar to the TCGA cohort between the EAC-prediction group and the GCFB-prediction group (FIG. 9).

In the NUJ II of the SNU cohort, 5/16 (31.2%) was classified as an EAC-prediction group and 11/16 (68.8%) were classified as a GCFB-prediction group. In particular, 15/16 (93.7%) of AGEJ III were classified as GCFB-prediction groups. Considering AGEJ II and III of the SNU cohort, the EAC-prediction group and the GCFB-prediction group were 6/32 (18.8%) and 26/32 (81.2%), respectively. There was no significant difference in TP53, PIK3CA, RHOA, KRAS, and ARID1A between the EAC-prediction group and the GCFB-prediction group of the SNU cohort. In particular, no somatic mutations of RHOA, KRAS and PIK3CA were found in the EAC-prediction groups of both the TCGA and SNU cohorts.

(4) EAC-prediction and GCFB Pathological analysis of predictive groups

Pathological characterization of the SNU cohort confirmed that all AGEJ III including GEJ and 80% of AGEJ III without GEJ were classified as GCFB-predictive groups (Table 3).

Pathological characteristics between the EAC-predictive and GCFB-predictive groups of the SNU cohort EAC-prediction
(n = 10) GCFB-prediction
(n = 36) P
Value location AGEJⅡ 5 (31.3%) 11 (68.8%) 0.231 AGEJⅢ including GEJ 0 11 (100%) AGEJⅢ without GEJ 1 (20.0%) 4 (80.0%) UT 4 (28.6%) 10 (71.4%) WHO differentiation 8 (80.0%) 13 (36.1%) 0.043 Non-differentiation 2 (20.0%) 18 (50.0%) Crystal x 0 5 (13.9%) Lauren chapter 8 (80.0%) 10 (27.8%) 0.009 Diffusion 2 (20.0%) 16 (44.4%) Mixed 0 10 (27.8%)

The distributions of the EAC-prediction group and the GCFB-prediction group did not differ significantly between AGEJ II, AGEJ III and UT. However, the EAC-predicted group had a significantly higher percentage of differentiated and intestinal types, while the GCFB-predicted group had a significantly higher percentage of non-differentiated and diffuse types.

(5) EAC-prediction and GCFC Analysis of the number of replicates between prediction groups

Genome-level copy number analysis was performed in the TCGA cohort. As a result, 435 amplified genes with significant copy number differences between the EAC-predicted and GCFB-predicted groups by the GISTIC algorithm were identified (≥2 fold change and P <0.05).

Filtering these 435 genes with human Cancer Gene Census (http://www.sanger.ac.uk/science/data/cancer-gene-census) revealed six cancer-related genes. (FIG. 10): COX6C (8q22.2, translocation), HNRNPA2B1 (7p15.2, translocation), NDRG1 (8q24.22, translocation), RECQL4 (8q24.3, nonsense, frameshift / splice), TCEA1 ( 8q11.23, translocation), and TFEB (6p21.1, translocation).

In the SNU cohort, using the GISTIC algorithm between the EAC-prediction group and the GCFB-prediction group, we compared 37 genes that were estimated to have focal amplication and identified 37 genes with significant differences in replication numbers. (P <0.05) (Table 4).

Gene showing significant replication number difference between EAC and GCFB among SNU cohorts (P <0.05) Of the EAC-like group
Average Log2 Replications Of GCFB-like groups
Average Log2 Replications BOP1 0.332 0.026 C19orf12 0.279 0.039 DUSP8 0.267 0.002 EGFR 0.460 0.119 ERBB2 1.185 0.227 FOXP4 0.303 0.052 GRB7 0.826 0.219 GSTA1 0.252 0.001 GSTA2 0.320 -0.021 GSTA3 0.273 0.030 GSTA5 0.275 0.012 HIST1H1B 0.228 -0.051 HIST1H2AI 0.245 0.009 HIST1H2AK 0.234 -0.016 HIST1H2AL 0.282 -0.027 HIST1H2AM 0.281 -0.043 HIST1H2BM 0.215 -0.007 HIST1H2BN 0.230 -0.007 HIST1H2BO 0.239 -0.023 HIST1H2BN 0.230 -0.007 HIST1H2BO 0.239 -0.023 HIST1H3H 0.266 0.019 HIST1H3J 0.303 0.005 HIST1H4J 0.254 0.017 LILRA3 0.276 -0.123 LOC100287704 0.399 -0.011 LY86 0.236 -0.045 MDFI 0.381 0.047 MIEN1 1.178 0.205 OR2B2 0.209 -0.019 PI4KAP1 0.284 -0.010 PLEKHF1 0.343 0.048 POP4 0.289 0.053 SSR1 0.216 -0.002 TFEB 0.431 0.075 TMEM191B 0.381 -0.013 TRAM2 0.293 0.041 UGT2B17 0.263 -0.049 VSTM2B 0.270 0.069 ZNF439 -0.242 0.023

Of 37 genes, filtering using humanCancer Gene Census (http://www.sanger.ac.uk/science/data/cancer-gene-census) revealed two cancer-associated genes: ERBB2 (17q12, Amplification), and TFEB (6p21.1, translocation) (FIG. 11).

ERBB1 (EGFR) (7p11.2) in the SNU cohort was amplified simultaneously in the EAC-predictive group and the GCFB-predicted group, but the number of copies of EGFR did not show a significant difference between the two groups. Since the annotated mutation pattern of COX6C, HNRNPA2B1, NDRG1, RECQL4, TCEA1, and TFEB of all cohorts was inconsistent for replication number amplification, ERBB2 and ERRB1 as possible heterodimers of RPPA and SNU cohorts of TCGA cohort Evaluation was performed using a tissue microarray.

(6) reversed-phase protein arrays ( RPPA ) And tissue microarrays ( TMA )

Supervised analysis of the RPPA data consisting of 44 EAC and 88 GCFB of the TCGA cohort identified an apparently isolated expression cluster of 81 proteins between the EAC and GCFB (FIG. 12).

Of the 81 proteins, PIK3CA and AKT1 identified in the pathway analysis of 400 signature genes and ERBB2 and EGFR identified in the copy number analysis showed significant differences between EAC and GCFB in protein expression of RPPA.

Expression of the four proteins was analyzed using three sets of TMAs from the SNU cohort using commercially available antibodies for further validation (Table 5).

Antibody Information Used in Tissue Microarrays Antibodies Clonality
(clonality) Dilution Detection kit source Cat.no EGFR Mouse monoclonal Immediately
use OptiView
Polymer
(Ventana) Roche 790-2988 ERBB2 Rabbit Monoclonal Immediately
use OptiView
Polymer
(Ventana) Ventana medical systems 790-2991 PI3KinaseP110alpha Rabbit Monoclonal 1: 100 OptiView
Polymer
(Ventana) Cell signaling # 4249 AKT1 Rabbit Monoclonal 1:50 OptiView
Polymer
(Ventana) Abcam ab32505

Staining patterns of EGFR, ERBB, PI3Kinasep110alpha, and AKT1 in TMA are shown in FIG. 13. Complex H scores of EGFR, PI3Kinasep110alpha, and AKT1 were calculated using the expression results of three sets of TMA, and the average H score of EGFR was significantly increased in the EAC-predicted group than in the GCFB-predicted group (160.7 ± 108.8 in EAC -like vs. 105.6 ± 81.6 in GCFB-like, P = 0.014, FIG. 14). Immunohistochemistry (ICH) staining for ERBB2 revealed a tendency for ERBB2-positivity to show a higher score in the EAC-prediction group than the GCFB-prediction group (Table 6). However, no significant difference in expression of PI3Kinase and AKT1 was observed.

IHC and SISH of ERBB2 EAC-prediction
(n = 10) GCFB-prediction
(n = 36) P value IHC 0 3 (30.0%) 17 (47.2%) 0.081 1+ 2 (20.0%) 14 (38.9%) 2+ 1 (10.0%) 2 (5.6%) 3+ 4 (40.0%) 3 (8.3%) IHC and SISH IHC <2+, or IHC2 + and
Black / Red ratio <2.0 5 (50.0%) 32 (88.9%) 0.015 IHC 3+, or IHC 2+, and black / red ratio ≥ 2.0 5 (50.0%) 4 (11.1%)

When comparing IHC and SISH together, the EAC-prediction group showed significantly higher positive ERBB2 (IHC 3+, or IHC 2+, and a black / red ratio of SISH ≥ 2.0) compared to the GCFB-prediction group. (50.0% EAC-prediction vs. 11.1% GCFB-prediction, P = 0.015). All significant changes from the univariate analysis of Table 2 were analyzed by multivariate analysis to identify risk factors for the expression of EGFR and ERBB2. The type of prediction (EAC-prediction or GCFB-prediction) for overexpression of EGFR was the only independent risk factor of 0.78 of the adjusted decision coefficient R ² (P = 0.034) (Table 7).

Multivariate Analysis of Overexpression of EGFR change Non-standardized coefficient B ± standard deviation standardization
Coefficient β t P value 95% confidence interval for B WHO classification 1.822 ± 5.722 0.053 0.318 0.752 -9.733-13.378 Lauren classification 26.389 ± 16.886 0.244 1.563 0.125 -7.665-60.443 Peripheral nerve penetration -38.504 ± 27.032 -0.220 -1.424 0.162 -93.057-16.046 Forecast type 62.500 ± 28.509 0.314 2.192 0.034 5.044-19.956

For ERBB2 positive, predictive type and WHO classification were independent risk factors ( P = 0.049 for predictive type and P = 0.029 for differentiation type) (Table 8).

Multivariate analysis for ERBB2 positive change P value Odds ratio
(Odds ratio) 95% confidence interval for odds ratio WHO classification
(vs. decisionx) differentiation 0.029 0.223 0.058-0.856 Non-differentiation 0.002 0.036 0.004-0.309 Lauren classification
(vs. mixed) chapter 0.387 4.156 0.165-105.009 diffusion 0.734 0.581 0.025-13.322 Peripheral Nerve Penetration
(vs. penetration) Non-penetrating 0.576 0.532 0.058-4.870 Type of prediction (vs. GCFB-prediction) EAC-prediction 0.049 6.179 1.1011-37.752

(7) CCLE  External verification using a database

From the CCLE database, esophageal (n = 3) and gastric (n = 38) adenocarcinoma cell lines with IC50 for expression microarray data, SNP array data and lapatinib, EGFR and HER2 tyrosine kinase inhibitor combinations were identified. The data available for each sample is shown in Table 9.

Esophageal and Gastric Adenocarcinoma Cell Line Information from the CLLE Database Cell line Agency BCCP Score prediction ERBB2
Replica ^* EGFR
Replica ^* IC50 † OE33 esophagus 0.546 EAC-prediction Amplification 0 3.538 OE19 esophagus 0.402 GCFB-prediction Amplification 0 N / A JHESOAD1 esophagus 0.484 EAC-prediction N / A N / A N / A FU97 top 0.109 GCFB-prediction fruition 0 8.000 NUGC3 top 0.37 GCFB-prediction 0 0 2.411 IM95 top 0.318 GCFB-prediction 0 0 8.000 AGS top 0.19 GCFB-prediction 0 0 N / A KATOIII top 0.536 EAC-prediction 0 0 N / A SNU16 top 0.351 GCFB-prediction 0 0 6.698 NCIN87 top 0.753 EAC-prediction Amplification 0 0.066 OCUM1 top 0.347 GCFB-prediction 0 0 8.000 SNU5 top 0.291 GCFB-prediction 0 0 N / A GCIY top 0.169 GCFB-prediction 0 0 7.255 SH10TC top 0.152 GCFB-prediction 0 0 8 MKN1 top 0.341 GCFB-prediction 0 0 N / A MKN74 top 0.36 GCFB-prediction 0 Amplification 4.69 KE39 top 0.211 GCFB-prediction Amplification 0 4.056 HGC27 top 0.062 GCFB-prediction 0 0 8 HUG1N top 0.315 GCFB-prediction 0 0 N / A NUGC4 top 0.313 GCFB-prediction Amplification Amplification 0.172 RERFGC1B top 0.365 GCFB-prediction 0 0 8 HS746T top 0.143 GCFB-prediction 0 0 8 NUGC2 top 0.531 EAC-prediction 0 0 N / A SNU1 top 0.176 GCFB-prediction 0 0 8 MKN45 top 0.341 GCFB-prediction 0 Amplification 8 X2313287 top 0.509 EAC-prediction N / A N / A N / A MKN7 top 0.272 GCFB-prediction Amplification 0 8 SNU216 top 0.37 GCFB-prediction Amplification 0 N / A AZ521 top 0.097 GCFB-prediction 0 0 1.66 LMSU top 0.129 GCFB-prediction 0 0 N / A ECC10 top 0.153 GCFB-prediction 0 0 N / A TGBC11TKB top 0.326 GCFB-prediction 0 0 N / A SNU520 top 0.297 GCFB-prediction 0 0 N / A GSS top 0.223 GCFB-prediction 0 Amplification N / A SNU620 top 0.322 GCFB-prediction 0 0 N / A ECC12 top 0.074 GCFB-prediction 0 0 N / A GSU top 0.388 GCFB-prediction 0 0 N / A SNU601 top 0.507 EAC-prediction 0 0 N / A SNU668 top 0.144 GCFB-prediction 0 0 N / A NCCSTCK140 top 0.816 EAC-prediction 0 0 N / A SNU719 top 0.305 GCFB-prediction 0 Amplification N / A

^* 0 indicates that the copy number has not changed and N / A is not available. † IC50 indicates half the maximum inhibitory concentration for lapatinib.

External validation using RNA microarray data from the CCLE database using these cell lines confirmed a significant difference in BCCP scores between esophageal and gastric adenocarcinoma cell lines (using Wilcoxon Rank Sumtest, P = 0.031) (FIG. 15).

Hierarchical clustering of the CCLE database using BCCP scores revealed no significant differences in tissue origin (esophagus or stomach), ERBB2 amplification, or EGFR amplification between the EAC-prediction group and the GCFB-prediction group (FIG. 16).

Target drug responses to lapatinib, EGFR and HER2 tyrosine kinase inhibitor combinations were assessed using IC50 data from the CCLE database for the EAC-prediction (n = 2) and GCFB-prediction groups (n = 17) (FIG. 17). .

As a result, it was confirmed that the IC50 was significantly lower in the EAC-prediction group than the GCFB-prediction group, using Wilcoxon Rank Sumtest (P = 0.044).

Claims (27)

To provide information necessary for the diagnosis of gastric or esophageal cancer,
Measuring the expression level of EGFR and ERBB2 in gastric or esophageal tissue isolated from the individual;
When the expression levels of EGFR and ERBB2 are increased compared to the control group, determining esophageal cancer; And
If the expression level of the EGFR and ERBB2 did not increase compared to the control, the method comprising the step of determining gastric cancer / body cancer The method of claim 1, wherein the gastric cancer or esophageal cancer is gastroesophageal adenocarcinoma (adenocarcinoma of gastroesophageal junction), cardia cancer, gastroesophageal border / gate cancer, upperthird gastric cancer, gastric cancer located at fundus, gastric cancer located at body, gastric adenocarcinoma located at fundusor body of the stomach (GCFB), esophageal adenocarcinoma (EAC), or a combination thereof. The method of claim 1, wherein measuring the expression levels of EGFR and ERBB2 comprises identifying a copy number of EGFR and ERBB2. The method of claim 1, wherein measuring the expression levels of EGFR and ERBB2 comprises measuring the amount of protein of HER1 or HER2. The method of claim 1, wherein measuring the expression levels of EGFR and ERBB2 comprises reverse phase protein analysis, tissue microarray, immunohistochemistry (IHC), and silver in situ matching (silver). in situ hybridization (SISH), or a combination thereof. delete The method according to claim 1, wherein the expression level of EGFR and ERBB2 is increased compared to the control when the number of copies of the ERBB2 gene is increased. delete The method of claim 1, further comprising: measuring a Bayesian Compound Covariate Predictor (BCCP) score of gastric or esophageal tissue isolated from the subject;
To provide information for diagnosing gastric or esophageal cancer in an individual,
Measuring the expression profile of a gene of tissue isolated from the individual; And
A method of analyzing a gene expression profile further comprising obtaining a BCCP score from a gene expression profile using a computer system executing a Bayesian Compound Covariate Predictor (BCCP) algorithm, wherein the gene is composed of the following genes: KRT5 , KRT13, KRT6A, KRT14, RHCG, KRT4, S100A7, SPRR3, KRT6C, SPRR2A, MUC21, KRT6B, PKP1, SERPINB3, CLCA2, SPRR1B, MUC4, TBX5, CRNN, SPRR2D, DUOX2, GJB6 S2, GJB6 S100 , SCEL, DSC3, CALML3, HEPHL1, A2ML1, NKX6-1, DUOXA2, SBSN, TMPRSS11D, CLCA4, ANXA8, FGFBP1, TGM1, IVL, DUOX1, DSG3, EREG, SERPINB4, KRT78, HOXC10, ZDUFKRT7 , TMPRSS11A, CXCL6, KLK13, CRCT1, PAX9, HORMAD1, FAM83A, S100A7A, GPR110, CAPN14, ABCA12, WFDC2, TMPRSS11B, CEACAM5, MUC12, SERPINB13, PADI1, FAT2, IL1RL2, STK31, SPIN3, SPX3 , KLK5, IL1F5, KRT15, BBOX1, GPR87, ACTL8, SLC34A2, USP9Y, TGM3, LASS3, TMPRSS13, GUCY1B2, SPRR2F, KDM5D, KRT23, LY6D, VTCN1, S LC5A12, ABCA13, S100A8, GBP6, CEACAM7, IL1F9, MUC15, PPP1R14C, KLK7, DEFB1, MMP3, TM4SF20, SYCP2, TMEM40, CEACAM6, VWA2, CYorf15B, TDRD5, DCDC2, KLK12, C10R, C14CARD DNAH2, SLC15A1, KLK8, LGALS7B, ALDH1L1, FOXN1, RAET1L, TP63, ATP13A4, UTY, DSG1, ALOX12B, AMY2B, SAA2, SAA1, GPR115, LCN2, KRT24, PRKY, PADI3, SCNN1AWISP IL20RB, FOXE1, ZFY, ALG1L, LY6K, NCCRP1, SLC44A5, NMU, CDH17, RNASE7, FER1L4, C4BPB, CALML5, CXorf61, SPACA4, ULBP2, KRTDAP, MYO7B,, C7orf54, LOCB20, SGA2A20 C21orf125 BNC1 IL1F6 TRPA1 C12orf27 LOC440905 POU5F1 LOC100131726 , FLJ42393, PCDHAC2, IFNE, BCL2L10, FLJ45445, COL7A1, GDA, ARL14, VIP, CLC, PPP2R2B, FABP4, SFRP5, CCL8, CD79A, HIST1H1E, CCL4L2, SYNPO2, SLC47A1, CXCR2P7, C1, CXCR2PN -3, PGM5, LILRB4, NMUR1 , XCL2, SOX10, FCGR1A, ST6GALNAC5, CCL4, ACSM5, FOLR2, KCNA1, CCL26, FCGR1C, VSIG4, C1QA, NCF1B, NXPH3, C1QC, TAGLN, PTPRN, TYROBP, ISL1, HCND, RGPOQ , GLP2R, PNMA6A, CD8B, FCGR1B, PCBP3, CD8A, GPR27, DPP6, F13A1, EPYC, CXCL9, GREM2, SNORA12, DNAJC5B, HLA-DQA2, PLP1, CHGB, APOC2, FAT3, TLR1, CHRDL1TR , CR1, FCGR3A, DPEP1, PPAPDC1A, ODZ3, MAPK4, C17orf87, KCNJ5, MSR1, RELN, APOE, C1QB, PHYHIPL, CCDC80, NCF1C, SLITRK5, TREM2, FGF14, PGA5, NKX2-3, PRELP5, PRELP5 , PCDH9, WSCD2, MS4A1, RNF150, KLRD1, KCNK2, MATK, HUNK, IGFBPL1, IGF1, CDO1, COLEC12, FIBIN, CARTPT, VPREB3, PPP1R1A, GDF6, NCAM2, CLEC10A, KIAA140852 BHL , GRIK3, SUCNR1, TMEM90B, CYP1B1, MKX, SV2B, BMP3, TCEAL2, CADM3, SPOCK1, GREM1, SIGLEC8, ADCY2, CDC26, OGN, FCRL6, PSD, VIPR2, GZMM, ADH1B, CCL19, PLD4, TMEMV1 , EOMES, STMN2, GZMH, NPTX1, CNTFR, TMEM130, CTNND2, ITGBL1, ECEL1, ACTG2, COL10A1, ST6GAL2, NRK, PI16, NALCN, GZMK, NCAM 1, RMRP, CHRM2, KCNMA1, HSPB7, SLIT2, PDZRN4, CNN1, CHRDL2, OMD, PTGIS, RSPO3, PLA2G2D, SMYD1, ZNF683, NRXN3, PCSK1N, HSPB6, C16orf89, Por3, COMP2, C2, C2, C2 CILP, COL11A1, LIPF, GATA5, MFAP5, SCRG1, SFRP2, GKN1, PGA4, CCL21, SIX2, NKX3-2, SFRP4, NBLA00301, THBS4, HAND2, and BARX1. 10. The method of claim 9, further comprising determining esophageal cancer when the BCCP score is at least 0.4535. 10. The method of claim 9, further comprising determining gastric / body cancer when the BCCP score is less than 0.4535. The method of claim 1, further comprising determining to administer an inhibitor of EGFR and ERBB2 when the expression levels of EGFR and ERBB2 are increased. The method of claim 12, wherein the inhibitors of EGFR and ERBB2 include lapatinib, trastuzumab, cetuximab, gefitinib, epitinib, or a combination thereof. How. The method of claim 1, wherein the gastric or esophageal cancer is adenocarcinoma. The method of claim 1, wherein measuring the expression levels of EGFR and ERBB2 is performed by at least one method selected from RT-PCR, RNase protection assay (RPA), northern blotting, and DNA chip. . The method according to claim 1, wherein the step of measuring the expression level of EGFR and ERBB2 is Western blotting, ELISA, radioimmunoassay, radioimmunoassay, oukteroni immunodiffusion, rocket immunoelectrophoresis, immunoprecipitation assay, complement fixation assay Performing at least one method selected from FACS, or protein chip. Probes, primer sets, or nucleotides that specifically bind to EGFR and ERBB2 proteins, or antibodies, antigen binding fragments, or polypeptides that specifically bind to their respective fragments, or polynucleotide sequences encoding EGFR and ERBB2 proteins. Gastric cancer of an individual comprising a substance for measuring the expression profile of a gene of a tissue isolated from the individual, for diagnosing gastric cancer or esophageal cancer of the individual, to provide information for diagnosing gastric cancer or esophageal cancer of the individual. A composition for diagnosing esophageal cancer, wherein the gene comprises the following genes: KRT5, KRT13, KRT6A, KRT14, RHCG, KRT4, S100A7, SPRR3, KRT6C, SPRR2A, MUC21, KRT6B, PKP1, SERPINB3, CLCA2 , SPRR1B, MUC4, TBX5, CRNN, SPRR2D, DUOX2, GJB6, SPRR2E, S100A2, KRT16, KRT17, SCEL, DSC3, CALML3, HEPHL1, A2ML1, NKX6-1, DUOXA2, SBSN, TMPRSS11D, C LCA4, ANXA8, FGFBP1, TGM1, IVL, DUOX1, DSG3, EREG, SERPINB4, KRT78, HOXC10, DUOXA1, ZNF750, SPRR1A, KRT7, TMPRSS11A, CXCL6, KLK13, CRCT1, PAX9, HO83A, SRM14, FAMAN ABCA12, WFDC2, TMPRSS11B, CEACAM5, MUC12, SERPINB13, PADI1, FAT2, IL1RL2, STK31, SPINK5, DDX3Y, PI3, FAM83C, LYPD2, KLK5, IL1F5, KRT15, BBOX1, GPR87, ACTL8, SLC34, ACTL8, SLC34 TMPRSS13, GUCY1B2, SPRR2F, KDM5D, KRT23, LY6D, VTCN1, SLC5A12, ABCA13, S100A8, GBP6, CEACAM7, IL1F9, MUC15, PPP1R14C, KLK7, DEFB1, MMP3, TM4SF20, SDCP2, SYCP2, SYCP2 DCDC2, KLK12, C10orf99, CARD14, HOXC13, LOC642587, GABRR1, DNAH2, SLC15A1, KLK8, LGALS7B, ALDH1L1, FOXN1, RAET1L, TP63, ATP13A4, UTY, DSG1, ALOX12B, AM22 SAA, KN24 PRKY, PADI3, SCNN1A, PGLYRP3, PRSS27, ANXA8L2, WISP3, IL20RB, FOXE1, ZFY, ALG1L, LY6K, NCCRP1, SLC44A5, NMU, CDH17, RNASE7, FER1L4, C4BPB, CALB61, CALB61 , C7orf54, SLC6A20, LOC221442, GOLGA8B, GPR81, ALDH3B2, TRIM 29, C21orf125, BNC1, IL1F6, TRPA1, C12orf27, LOC440905, POU5F1, LOC100131726, KCNH8, SPRR2C, IL13RA2, C3orf67, SAA4, LYPD3, GJB5, ULBP1, PHACTR3, S100A9CCLO, LO3 MRPL42P5, FLJ42393, PCDHAC2, IFNE, BCL2L10, FLJ45445, COL7A1, GDA, ARL14, VIP, CLC, PPP2R2B, FABP4, SFRP5, CCL8, CD79A, HIST1H1E, CCL4L2, SYNPO2, SLC472P1, CCR7A1 NKX6-3, PGM5, LILRB4, NMUR1, XCL2, SOX10, FCGR1A, ST6GALNAC5, CCL4, ACSM5, FOLR2, KCNA1, CCL26, FCGR1C, VSIG4, C1QA, NCF1B, NXPH3, C1QC, ISLG, PSL RGMA, APOBEC3H, PRND, ADIPOQ, CCL11, GLP2R, PNMA6A, CD8B, FCGR1B, PCBP3, CD8A, GPR27, DPP6, F13A1, EPYC, CXCL9, GREM2, SNORA12, DNAJC5B, HLA-DQA2, FAT, APOC2 TLR7, CHRDL1, LY86, AGTR1, ISLR, CR1, FCGR3A, DPEP1, PPAPDC1A, ODZ3, MAPK4, C17orf87, KCNJ5, MSR1, RELN, APOE, C1QB, PHYHIPL, CCDC80, NCF1C, SLITRK5, TREM2, TREM2 3, PRELP, FCRLA, CCL5, GNAO1, PCDH9, WSCD2, MS4A1, RNF150, KLRD1, KCNK2, MATK, HUN K, IGFBPL1, IGF1, CDO1, COLEC12, FIBIN, CARTPT, VPREB3, PPP1R1A, GDF6, NCAM2, CLEC10A, KIAA0408, BHLHE22, CD52, SULT4A1, SIT1, FNDC1, GRIK3, SUCNR1, TMEMBB, CMPBB2, MMPBB2 TCEAL2, CADM3, SPOCK1, GREM1, SIGLEC8, ADCY2, CDC26, OGN, FCRL6, PSD, VIPR2, GZMM, ADH1B, CCL19, PLD4, TMEM189-UBE2V1, NKG7, EOMES, STMN2, GZMH, NPTX1, CNTFRD, CNTFRD ITGBL1, ECEL1, ACTG2, COL10A1, ST6GAL2, NRK, PI16, NALCN, GZMK, NCAM1, RMRP, CHRM2, KCNMA1, HSPB7, SLIT2, PDZRN4, CNN1, CHRDL2, OMD, PTGIS, RSPO3, PLAFDG3D PCSK1N, HSPB6, C16orf89, PGA3, COMP, C2orf40, C7, GKN2, DES, CILP, COL11A1, LIPF, GATA5, MFAP5, SCRG1, SFRP2, GKN1, PGA4, CCL21, SIX2, NKX3-2, SF301, THB HAND2, and BARX1. The gastroesophageal adenocarcinoma of the gastroesophageal junction, cardia cancer, gastroesophageal border / gateway cancer, upperthird gastric cancer, gastric cancer located at fundus, gastric cancer located at body, gastric adenocarcinoma located at fundusor body of the stomach (GCFB), esophageal adenocarcinoma (EAC), or a combination thereof. The composition of claim 18, wherein the composition is for diagnosing esophageal cancer or gastric / body cancer of the subject. A kit for diagnosing gastric or esophageal cancer comprising the composition, reagents and packaging units of claim 17. delete delete delete delete The method of claim 9, wherein measuring the gene expression profile is performed by one or more methods selected from RT-PCR, RNase protection assay (RPA), northern blotting, and DNA chip. delete delete

Images