The results of the expression array studies ... - Wiley Online Library

Cytometry Part B (Clinical Cytometry) 68B:1–17 (2005)

Review Article

The Results of the Expression Array Studies Correlate and Enhance the Known Genetic Basis of Gastric and Colorectal Cancer Orsolya Galamb,* Ferenc Sipos, Krisztina Fischer, Zsolt Tulassay, and Bela Molnar II Department of Medicine, Semmelweis University, Faculty of Medicine Budapest, Hungary

Gastric and colorectal cancers belong to the most frequent cancer types in the world today. This fact emphasizes the importance of identification of useful diagnostic and prognostic markers, in the earliest stage of the disease. The examination of gene expression profile in gastric and colorectal cancer may develop the bases of early diagnosis and of individual therapeutic strategies. In the microarray examinations done so far for these types of cancers, the expression of hundreds and thousands of genes were studied, however, both the sample collection and the results showed wide variations. The diversity of expression array methods and data analysis makes the comparison of microarray results difficult. Beside the exposition of the practical aspects of the chip technology, our aims are the systematization of data that are currently available in the international scientific literature and the description of the results in a comprehensive way. Microarray results show that the gene expression pattern, detected in gastric and colon cancers, highly depends on the histological type and heterogeneity of the sample, array type, and softwares, used for data analysis. Recent experiments point out not just the changes of the alterations of tumor suppression, apoptosis, cell-cycle regulation, and signal transduction, but tumor cell metabolism and cell-microenvironment interactions also. Results show connection to and make more complete the already known molecular background of gastric and colorectal cancers. Based on the accumulation of recent and further data, such kind of multifunctional diagnostic microarrays that can be suited for completing the conventional histological diagnostics and subtypization will certainly become available in the near future. q 2005 International Society for Analytical Cytology Key terms: gastric cancer; colorectal cancer; microarray technology; gene expression

Gastric cancer (GC) and colorectal cancer (CRC) belong to the most frequent cancer types in the world today. Although the prevalence of gastric and colonic cancers is not the same (the former is rare, the latter is frequent, considering the newly detected cases per year), these two cancers are considered as the most death causing gastrointestinal tumors in a number of countries of the world. According to the cancer mortality data from 1990 to 2002 (http://www-depdb.iarc.fr/who/menu.htm) the two most frequent gastrointestinal cancer that cause the most number of deaths in Eastern-Europe, Southern-Europe, in Russian Federation, and in Southern and Eastern Asian countries are the colorectal and gastric cancers. In the United States and in Australia, the incidence rate of the lethal pancreatic cancer cases is higher than lethal gastric cancer cases. Although the increasing number of lethal pancreatic cancers can be observed in Western Europe, the gastric cancer is still the second main death causing cancer among the tumors of the gastrointestinal tract, especially in men.

q 2005 International Society for Analytical Cytology

Both areas of the gastrointestinal system are facing similar exogen carcinogenetic factors like food or bacteria. It is also not clear why gastric cancer is more aggressive and has weaker therapeutic answer than colorectal cancers. We have also chosen these two types of GI cancers because other GI cancers are not in such direct contact with exogenenous carcinogenetic agents like those mentioned earlier. Most GC cases are diagnosed at an advanced stage, resulting in poor prognosis. Currently, surgical resection is the only effective curative modality for primary treatment of GC (1), and
40% of all patients respond to the surgical intervention (2). Five-year survival rate of patients with GC is only 20% (3). *Correspondence to: Orsolya Galamb, II Dept. of Medicine, Semmelweis University, 1088 Budapest, Szentkira´lyi str. 46, Hungary. E-mail: [email protected] Received 22 November 2004; Accepted 10 June 2005 Published online 5 October 2005 Wiley InterScience (www. interscience.wiley.com). DOI: 10.1002/cyto.b.20069

2

GALAMB ET AL.

In advanced cases of CRC, the prognosis is also poor. The 5-year survival rate is between 15 and 65%, on an average of 40% (4). Only a part—mainly in early stage of the disease—of the cases is curable with current therapeutic modalities. CRC is one of the main epithelial cancers where the molecular alterations show a parallel tendency with the progression of the disease (5). Based on this knowledge, basics of new therapies (e.g. gene therapy) were established. However, initial experiments did not show any positive influence for survival. As the molecular basis of gastric and colorectal cancer development has not been completely understood, the identification of further molecular alterations was aimed at in subsequent studies. New biomarkers and new therapeutic targets became a requisite (6). In the last two decades, gastrointestinal carcinogenesis was under intensive research. Both the activation of oncogenes (7) and the inactivation of tumor suppressor genes (8) are involved in gastric and colorectal carcinogenesis. In some cases, the deletion of the mismatch repair system also contributes (9). Additionally, further complicating our understanding, genetic mutations accumulated in tumors are often accompanied by several genetic and epigenetic changes, such as loss of heterozygosity, inactivation of important genes by methylation (10), loss of imprinting (11), and gene amplification, all of which can modify the gene expression profile. Hence, the genome-wide analysis of gene expression is needed. Nowadays, the biochip technology is one of the most rapidly developing fields of medical biology. The recognition of the gene expression differences in the light of different histological alterations gives us a new possibility to identify the disease-specific genes, and, accordingly, to develop relatively quick, exact, and individual diagnostic and (later) therapeutic methods, and makes possible the examination of the gene expression effect of conventional and newly-developed medicaments. The purpose of this review is to foreshow the practical aspects, advantages, and disadvantages of chip technology for gastrointestinal specimen analysis, to offer a survey of the gene expression pattern changes examined by expression microarrays and of their functional classification, and to summarize the genetic knowledge related to gastric and colonic carcinogenesis and progression, and possible routine application fields. TECHNICAL ASPECTS OF APPLICATION OF EXPRESSION MICROARRAY TECHNOLOGY IN GASTROINTESTINAL SAMPLES The microarray technology, which has recently undergone rapid development, allows one to study the expression and activation of large number of genes (or sequences) in a single hybridization (12), in different cells, tissues or distinct stages of the same cell/tissue type (such as physiological or pathological state and with or without drug treatment). Simultaneous monitoring of the expression of thousands of genes by microarray technology offers opportunities for widespread examination of

gene functions and new metabolic pathways, pathogenetic mechanisms, novel indicators for progression of different diseases can be discovered, and, therefore new therapeutic target molecules can be found. Today, two types of arrays are commonly used: cDNA and oligonucleotide arrays. cDNA arrays are generally manufactured by robotic spotting of small quantities of PCR amplified segments of genes, cloned from cDNA libraries, onto glass slides or nylon membranes. Oligonucleotide arrays are prepared by either spotting presynthetized oligonucleotides (typically 50–80 bases) or, alternatively, by synthesis of short oligonucleotides (on the order of 20 bases) directly on glass (13). The in situ synthesis technology allows very high resolution, several different sequences to represent each gene, and, theoretically, a high level of internal controls. Furthermore, the process for manufacturing oligonucleotide arrays is readily standardized, providing relative precision in estimations of expression levels (14). The essence of expression microarray technology is that complementary target oligonucleotides or cDNAs are fixed to solid surface, and single strand cDNA probes that are complementary to the extracted sample RNA and labeled with modified nucleotides (32P, fluorescent dyes) are hybridized to them. Intensity of hybridization refers to the expression level of the tested genes. This rate can be detected according to the labeling of the probes (e.g., Phosphor imager, fluorescence chip readers). Data analysis is carried out with data analyzer program packages (e.g., Array Vision, GenePix) (15). Expression microarrays are semi-quantitative; therefore, it is necessary to verify gene expression data. RTPCR analysis of some candidate genes, which was found over- and underexpressed by chip, can show that the signal intensities on the microarray do (or do not) correlate to the real expression level of these genes (15). Several methods (e.g.: RT-PCR, DD-PCR, Northern blot) have already been evolved for gene expression analysis. Most of them are appropriate to identify specific gene products, but are time- and labor-intensive, and need high amounts of high pure RNA, and by using them one or just a few parameters can be examined abreast (Table 1). Using the conventional gene expression analyzer methods and taking into account that the amounts of clinical samples are limited, the determination of the gene expression pattern of the sample is circuitously difficult or impossible. Contrarily, the chip technology allows us to analyze abreast the expression of hundreds and thousands of genes in cellular or tissue materials from one patient. With the help of expression microarrays, the disease-specific, overexpressed or downregulated genes can be filtered and the development of low-density screening miniarrays becomes accessible. Based on the precise description of gene expression alterations of disease-specific genes, diseases can be divided into subgroups, and the different therapeutic response in these subgroups turns to be understandable. There are several factors that require the verification of the results (mainly with real-time PCR), which may

Table 1 Comparison of Earlier mRNA- and Protein-Based Gene Expression Analytic Technologies

Methods

Sample type

Number of observed parameters per sample at one time

Advantages

mRNS in situ hybridization. Fluorescence in situ hybridisation (FISH)

Tissue

1 (light microscope) 1–6 (fluorescence microscope)

Expression of the gene is located to the histological structure.

Northern blot

Any kind of samples, from which RNA might be extracted

1 (if big difference between mRNA lengths, then 2)

The basic mRNA examiner method that is relatively easy and inexpensive.

Reverse PCR (RT-PCR) Subtractive hybridisation

Any kind of samples, from which RNA might be extracted Tissue or cells

Differential display (DD)

Tissue or cells

Depends on the primer sets used during the reamplification

RAP-PCR (random arbitrarily primed PCR)

Tissue or cells

1 (if big difference between mRNA lengths, then 2)

Direct immunohistochemistry– immunocytochemistry Indirect immunohistochemistry– immunocytochemistry, Enzyme-Linked Immunosorbent Assay (ELISA) Immunoprecipitation

Tissue, cells, or cell culture Tissue, cells, or cell culture

1–6 (fluorescence)

Protein mixture

Immunoblot (Western blot) Electron microscopic immunogold technology

Tissue or cells

1 Allow measuring expression of many genes in parallel

1–2 (light microscope) 1–6 (fluorescence microscope) 1 1 (autoradiography) 1–6 (indirect immunochemistry) 1

Wide-spread method which is exceedingly able to detect RNA viruses and alternative splicing. Isolation and cloning of differencies of gene expression is possible. Isolation and cloning of differences of gene expression is possible. Reproducibility and sensibility is higher than in DD (high stringency). Relatively easy method. Result is located to the morphology. Dynamical experiments can be done (cell movement, cytoskeletal changes, cell adhesion). Using confocal microscope cellular ‘‘CT-images’’ can be done. High specificity and sensitivity.

Synthesis and chemical altering of proteins can be analyzed. Detection and quantification of specific protein in unlabelled protein mixture is possible. The subcellular localization of the protein can be detected.

Disadvantages Transcriptum must be known (preliminary selection of genes is necessary). Parallel examinations are timeand labor intensive. Requires the use of radioactive isotopes. Serial analysis cannot be done. Relatively expensive. Using of high-amounts of 32P. High amounts of RNA are needed. High pure RNA is required. Labor intensive. Relatively large number of false positivity (stringency is low). High pure RNA is required. Labor intensive.

Sometimes specificity and sensitivity are low. Time intensive. Optimalisation of sign conversion is required when light microscopic analysis is required. Radioactive labeling is required. Sometimes radioactive labeling is required. Expensive.

4

GALAMB ET AL.

occur during chip technology (e.g., ischemic degradation of RNA can reduce the signal/noise ratio) (16). Table 2 and 3 summarize the most important parameters of the expression microarray experiments made by different research teams in GCs and CRCs: the array type, number of genes, type of detection, number and origin of the samples, and statistical criteria of up- and down-regulation (tumor/normal ratio). Beside the global arrays, which allow one to study the expression of several thousands of genes at the same time. Offering genome-wide expression profiles, the so-called cancer arrays are useful tools for analysis of some hundreds of tumor-associated genes. The normal and the tumor sample are usually analyzed by the double-fluorescent detecting (Cy3 and Cy5) cDNA microarrays. In most gene expression analysis studies, RNA extracted from the whole tumorous tissue was used for the examinations; no differences were made between the tumor cells and the stromal cells from the surrounding connective tissue (endothelial cells, histiocytes, smooth muscle cells, inflammatory cells etc.). This is understandable, because there are several important signal transduction ways between tumor cells and its stromal and inflammatory cells, and analyzing only the tumor cells may lead to the loss of essential information. In case of cancer studies, there are viewpoints that emphasize the importance of analysis of region-specific gene expression, rather than analyzing tissue from all parts of a tumor. Genes expressed at the invasion front of cancer nests probably contribute to the process of cancer invasion, and genes expressed at either the lymphatic or venous region of cancer invasion are likely to be the candidate metastasis associated genes. Two techniques, which assist the region-specific localization of genes, are currently available. One is laser-microdissection, which allows the sampling of rare cells or extremely small pieces of tissue from the relevant regions. The opinion that the high purity of the observed samples may lead to loss of information due to the loss of intercellular connections is debatable, because during sample processing these connections functionally are out of action. To avoid the loss of information, all cell types in the sample must be analyzed (stromal and inflammatory cells), and this step is frequently missed or defective in the studies done so far. Laser-microdissection has two main disadvantages: small sample quantity and the difficulty in sample handling. The small amount of RNA from laser microdissected samples can be representatively amplified by the T7 RNA amplification method. The second technique is the localization of gene expression itself at the tissue level, using in situ hybridization (ISH). ISH gives region-specific signals and allows histopathological evaluation of the entire tissue, including the region of interest, thereby allowing gene expression to be linked to the results of histopathological examination (41). The main disadvantage of the ISH is the low number of examinations in one sample at the same time.

GENE EXPRESSION ANALYSIS IN GASTRIC CANCER Overexpression of oncogenes associated with the tumor development was detected in several cases. Oncogene v-jun (26), the mutant form of the c-jun, binds the c-fos, and modifies the transcription of the genes contributing to cell proliferation, apoptosis, and oxidative stress response, as a member of AP1 transcription factor complex. High expression of v-raf (26) and growth related oncogen (GRO1) oncogenes was also described (21). V-raf is a transcription factor, while GRO1 is a chemokine that induces uncontrolled cell division and angiogenesis in melanoma and prostate cancers (45). cErbB-2 and -3 receptor tyrosine kinases, two members of epidermal growth factor receptor (EGFR) family are also considered as oncogenes. In addition, in gastric tumors, overexpression of c-ErbB-2 was observed in several other malign tumors including lung, breast, prostate, pancreas, and colorectal cancers. It may play a role not only in the carcinogenesis, but in development of chemotherapyresistance and metastases also (46,47). Parallel to overexpression of oncogenes, deleted in colon carcinoma (DCC) and Ras association domain family 1 (RASSF1) tumor suppressor genes are underexpressed in GC tissues (12,47). GIF growth inhibitor factor—which is underexpressed in gastric tumors—may act as a tumor suppressor gene (19). Changes in ratios of cell-cycle positive and negative regulator molecules participate in the increased division of tumor cells. Tumor cells can pass through the G1/S cell division checkpoint more easily by increasing production of CDK4 and cyclin-D1. Larger amount of CKS1 contributes to this by ubiqutination and degradation of the p27Kip1 molecules, inhibitor of CDK4/cyclin-D1 complex (19,21,48). During the examination of the cell cycle regulatory cyclins it turned out that the gene of cyclin-E is amplified in 15–20% of gastric carcinomas. Overexpression of cyclin-E (18) tends to be correlated with the aggressiveness and invasiveness of the gastric carcinomas, and it is typical in advanced stages (49). Overexpression of DNA synthesis genes (such as PCNA and DNA-topoisomerase2a) can also be detected in GC tissues (19,26). Other genes associated to cell division are downregulated, such as BTG-2 participating in cell cycle control and cell response to DNA damage and showing p53 dependent expression and CHES1 (checkpoint suppressor) transcription factor gene (19). Altered expression of genes contributed to apoptosis inhibition (compared to that of normal tissues) was observed by several teams. In addition, the increased expression of bcl-2 (20,26), downregulation of apoptosis inductor FADD-like apoptosis regulatory molecule and the death-associated protein-6 (DAXX) inhibits the programmed cell death, supporting the survival of the tumor cells (26). Certain components of apoptosis effector kinase-cascade (caspase-8, 9, and 10) are also underexpresssed (12,26). Cellular apoptosis susceptibility (CAS) gene, that has a dual function in apoptosis and

Table 2 Gene Expression Microarray Analyses of Gastric Cancer Samples Research team

Array type

Number of genes

Boussioutas et al. [17]

Global cDNA microarray (own)

Chen et al. [18]

Global cDNA microarray (according to Alizadeh et al.)

30,300

El-Rifai et al. [19]

Cancer cDNA microarray (Atlas Human Cancer 1.2 K) Global cDNA microarray (Ono et al., own protocol)

1,174

Hasegawa et al. [20] Hippo et al. [21] Inoue et al. [22] Kim et al. [23]

Global oligonucleotide microarray (Affymetrix) Cancer cDNA microarray (TaKaRa Shuzo) Global cDNA microarray (own)

Lee et al. [24]

Global cDNA microarray (NEN Company)

Liu et al. [12]

Cancer cDNA microarray (Atlas Human Cancer cDNA Expression Array) Cancer cDNA microarray (own)

Meireles et al. [25] Mori et al. [26]

Cancer microarray (TaKaRa Shuzo)

Norsett et al. [27]

cDNA microarray (Research Genetics)

9,381

23,040 6,800 425 13,376 2,400

Detection type Fluorescent (Cy3, Cy5)

Fluorescent (Cy3,Cy5) Autoradiography (32P) Fluorescent (Cy3, Cy5) Fluorescent (phycoerythrin) Fluorescent (Cy3,Cy5) Fluorescent (Cy3, Cy5) Fluorescent (Cy3,Cy5)

588

Autoradiography (33P)

4,512

Autoradiography (33P) Fluorescent (Cy3,Cy5)

624

2,504

Fluorescent (Cy3, Cy5)

Number of samples Totally 124 gastric cancer (different subtypes) and surrounding mucosa samples 90 primary gastric adenocarcinoma (14 of them with lymph node metastases) 9 xenograft 2 primer adenocarcinoma 20 intestinal-type gastric cancer 20 normal mucosa 22 gastric cancer 8 normal sample 43–43 gastric cancer and normal sample 22–22 gastric cancer and surrounding normal mucosa 22–22 gastric tumor and surrounding normal mucosa 16-16 gastric cancer and normal sample 6–6 gastric tumors and normal gastric mucosa 5–5 gastric adenocarcinoma and normal gastric epithelium 17 gastric cancer

Type of samples

Tumor/normal ratio

Surgical material

Surgical material

At least 2.5-fold differences in at least 3 samples

Surgical material

Up: 2,5–16 down: (2.5)–(16) Up: >2 down: 2.5

Surgical material

Log ratios were analyzed by SAM method

Surgical material

Up: >2 down: 1.5 diff: >10000

Surgical material

t-statistics for each cDNA (tnc values) Up: >2 down: 3x than the phospholipase A2 housekeeping gene expression, down if 2, down: down

Laser-microdissec-ted Excluding 10% of the tissue from surgical expression levels of normal material cells from each side was (
10000cells/sample) determined as normal expression range. Cy3/Cy5 ratio outside this range was differentially expressed. Surgical material

Surgical material

31-31 CRC and Surgical material normal adjacent mucosa 10 normal 20 CRC Surgical material

20 normal 25 CRC

6 normal 21 CRC

22 CRC 23 normal

20 CRC

40 CRC 20 liver metastasis 10 normal 6

Number of samples

Table 3 Gene Expression Microarray Analyses in CRC

6 GALAMB ET AL.

THE RESULTS OF THE EXPRESSION ARRAY STUDIES IN GASTRIC AND COLORECTAL CANCER

cellular proliferation (24), is overexpressed in rapidly dividing cells, such as tumor cells. Beside its essentiality in mitosis, its inhibitory effect on TNF-a- induced apoptosis and the nuclear transport of apoptotic proteins also support its proliferation inductor features. Changes of genes and gene expression of growth factors and their receptors, and the signal transduction pathways associated with them contribute to the development and progression of different tumor types, including GC. Hepatocyte growth factor (HGF) is a pleiotrop growth factor, with motogen, mitogen, and morphogen effect to several cell types and its role in the angiogenesis in vivo is known (50). Amplification of c-met gene encoding HGF receptor frequently occurs in gastric tumors (51). The heparan-sulfate-binding hepatoma derived growth factor (HDGF), the fibroblast growth factor-2, bFGF (FGF-2), the vascular endothelial growth factor (VEGF), and insulin-like growth factor-2 (IGF-2) are all angiogen factors. The IGF-2 is proved to be antiapoptotic in gliomas; moreover, it has a mitogen and invasion enhancer effect. IGF-2 overexpression because of loss of imprinting has also been described (52,53). Results of experiments performed on tumor cell lines suggest that IGF-2 may act as a proto-oncogene. In all these cases, growth factors were found to be upregulated in microarray analyses (19–21,54). Increased expression of transforming growth factor-b3 (TGF-b3) (22,54) was also detected. TGF-b3 firms the fascia breaking strength, increases the collagen deposition, and cellular proliferation, and its expression correlates to the development of metastases (55). Prostate differentiation factor or placental TGF-b (PLAB), other member of TGF-b family is also overexpressed in gastric cancers (20,21). PLAB inhibits tumor cell growth and the inflammation, and participates in cell response to DNA damage (p53-dependent or p53independent apoptotic response). In spite of these facts, PLAB upregulation was detected in numerous tumor types, including prostate cancer, CRC, and GC (56). Among the growth factor receptors, significant overexpression of the fibroblast growth factor receptor-4 (FGFR-4) and insulin-like growth factor receptor-2 (IGFR2) was described (21,26). Activation of FGFR-4 can induce malign transformation, and it shows higher expression level not only in gastric tumors, but in some other cancers, such as breast, pancreas, and renal cancers (57). Increased production of c-ErbB-2 and c-ErbB-3, two receptor tyrosin kinases of EGFR family, has been previously mentioned among the oncogenes. Other receptors were found to be downregulated: lower level of the retinoic acid receptors decreases the growth inhibitory effect of the retinoids (12,58). Overexpression of signal transduction molecules, such as p38 mitogen activated protein kinase (MAPK) and growth factor receptor binding protein (Grb-2), and Grb-7 adaptor molecules was established during microarray analyses performed by several research teams (19,21,54). Grb2 can bind the tyrosin-phosphorylated EGFR and PDGFR, and takes part in Ras activation, together with the SOS

7

GDP/GTP exchange factor. Besides Grb-7 has an EGFRbinding affinity, hereby participates in EGFR signal transduction, it is also important in the regulation of integrinmediated cell migration, by cell adhesion-dependent binding of the focals adhesion kinase (FAK) (59). Metabolism of tumor cells is also changed when compared with that of the normal cells. Alterations of expression level of numerous metabolic enzymes can be detected in different malignancies. Downregulation of several genes that are specific to gastric mucosa and involved in lipid metabolism (apolipoprotein-A4 and -B, microsomal triglyceride transfer protein), carbohydrate metabolism (alcohol-dehydrogenase-3, aldehyde-dehydrogenase-3, aldolase-B, fructose 6-bisphosphatase-1), drug metabolism, drug detoxification and drug resistance (several members of cytochrome P450 family, glutathion-peroxidase, dihydrodiol-dehydrogenase), defense response (Treifol-factor-1 and -2), and transport of small molecules or heavy metals (gastric intrinsic factor, Hþ/Kþ ATPase, Ca2þ ATPase) (18,20,23,24) was observed in gastric tumors. Some of the genes associated with metabolism were found to be upregulated, including transferrin receptor and some members of solute carrier family 2, 16, and 25. Transferrin receptor, which can bind the main iron-transport protein, transferrin, plays a significant role in regulation of cell proliferation by controlling the cell iron metabolism. Besides the activated T-cells, B-cells, and macrophages, other proliferating cells, such as tumor cells express the transferrin receptor. Antitransferrin antibody is suitable for the prediction and prognosis of therapeutic response in cases of breast and cervical tumors and non-Hodgkin lymphomas (60). Among the members of the solute carrier families, one of facilitated glucose transporters, two of monocarboxylic acid transporters, and one of the mitochondrial carriers, an adenine nucleotide translocator associated with energy generation, have been found to be downregulated in GCs (20). Degradation of basal membrane, one of the essential steps of tumor invasion and, hereby, development of metastases, is basically performed and controlled by the matrix metalloproteases (MMPs) and their inhibitors. During the tumor progression, MMPs proteolitically degrade the components of extracellular matrix (ECM), promoting tumor cells to pass through the basal membrane and neovascularization of tumors by endothelial cells (61). Microarray analysis of gastric mucosa samples shows that several types of MMPs are overexpressed in gastric cancerous tissues (12,18,19,21,22,26,54). MMP-1 is an interstitial collagenase, MMP-2 is a gelatinase, MMP3 and -11 are progelatinases, MMP-12 has a metalloelastase activity, while MMP-7 (matrilysin), which is also overexpressed in 80% of CRCs, induces angiogenesis by hydrolysis of human plasminogen, and generation of angiostatin fragments (62), furthermore, matrilysin proteolytically generates active soluble Fas ligand and potentiates epithelial cell apoptosis (63). Regulation of matrilysin production by b-catenin accumulation is a contributing factor to intestinal tumorigenesis (64). The

8

GALAMB ET AL.

expression level of the membrane-type MMP-14 and -16 is also found to be increased (12,19,21,54). Besides bcatenin, which has been previously mentioned, the g-catenin called pacoglobin is also overexpressed in gastric tumors (19). Catenins are cytoplasmatic proteins, which are associated with cytoplasmatic region of E-cadherin. Among the cadherins participating in cell–cell adhesion, E-cadherin (17), OB-, and LI-cadherin (21) show increased expression, while the level of the cadherin-8 (26) transcript decreased. E-cadherin can bind b-catenin and lower free b-catenin concentration within the cell. b-catenin is a member of the Wnt signal transduction pathway that is also altered in over 90% of CRCs (65). Several other genes involved in this pathway show significantly different expression levels between tumor and normal tissues in microarray studies: not only the elevated expression of Axin-2 and EphB2 (ephrin receptor tyrosine kinase B2), but the reduced levels of clathrin and retinoic acid receptor might also contribute to increased Wnt signaling (18,25). Cytokeratin-1 (12), a cytoskeletal structure polypeptide, integrin-a3 (19) adhesion molecule, numerous type of ECM component collagens, such as collagen type I, III, IV, V, VI, VII, and XVIII, and urokinase plasminogen activator (uPA) (21,54), which effectively indicates tumor metastatic potential (66), were established to be upregulated in stomach cancer, compared with that of the normal tissues. Increasing activity of cathepsin-B, which is overexpressed in gastric cancers, correlates with the increasing malignancy of not only gastric tumors, but some other cancer types also (21,67). Cathepsin-B—similarly to MMPs—contributes to tumor cell invasion by the cleavage of the ECM proteins. Besides the increased ECM degradation and remodeling activity, lower expression level of the inhibitor of MMPs tissue inhibitor of metalloprotease-3 also participates in the tumor invasion (19). Table 4 contains the list of commonly over- and underexpressed genes found by different research teams in functional groups. GENE EXPRESSION PATTERN ANALYSIS IN CRC The downregulation of tumor suppressor genes (e.g., p53, CD27-binding protein (SIVA), Fas-induced apoptosis regulator protein (TOSO) (40), and TN (tumor stromal component, it has a tumor suppressor effect) (6,39) was found in both the compact-sized DNA filter (40) and high-density cDNA microarray (6) analytical methods. The downregulation of MXI1 and several apoptoticrelated genes, such as TRAIL receptor 2, TRAIL, caspase 10/Mch4/, was found in CRC. MXI1 is the member of the c-Myc family, and is a putative tumor suppressor, because it negatively regulates the action of c-Myc, an oncogene that activates transcription and stimulates cell proliferation (71). The TRAIL receptor 2 induces apoptosis as a consequence of activation of nuclear factor-kb and c-Jun-NH2-terminal kinase (72). The reduced expression of tumor suppressor and apoptosis inducer genes in CRC may prolong the survival of tumor cells.

Expression changes in cell cycle regulatory genes were observed in colon carcinoma cell lines and in CRC tissue samples as well. Overexpression of cyclin E (40) and the suppression of a DNA synthesis controlling protein topoisomerase IIa (29) were described. The Wee1 gene was downregulated in both cell lines and tissues. It phosphorylates and thereby inhibits the corresponding human kinase CDK1 and the progression from G2 to M phase. Suppression of p21 was detected in cancer tissues (29). The p21 protein binds to cyclin–cyclin-dependent-kinase complexes and arrests cells at the G1–S and G2–M checkpoints; therefore, it has a tumor suppressor and growth inhibitory effect (69). Using laser-microdissection, Kitahara et al. (36) found the 14-3-3s-gene downregulated in CRC. This gene is regulated by p53 and is known to promote premiotic cell-cycle arrest after DNA damage. Because p53 is frequently mutated in CRC, reduced expression of 14-3-3s in the cancer tissues may have resulted from mutations of p53. The genes of equilibrative nucleoside transporter 1 (ENT1), M-phase phosphatase 2 (CDC25B) (34,36,39,43), nitric oxide synthase 2A (NOS2A), and nucleolin (NCL) were also found to be upregulated. ENT1 helps to supply nucleotides required for constitutive DNA synthesis in highly proliferative cells (73). The CDC25B protein is required for progression of the cell cycle (G2–M phase transition) by dephosphorylating cell division cycle 2 (CDC2) (74). Induced NOS2A causes production of nitric oxide, which may be involved in DNA damage as well as several kinds of signal transduction (75). NCL induces chromatin decondensation by binding to histone H1 (76). These genes may have a direct or indirect role in colorectal carcinogenesis including accelerated metabolisms, impaired regulation of cell cycle, production of reactive oxygen species, and disregulated transcription and mitosis. Comparing (29) the gene expression patterns of colon carcinoma cell lines and of surgically removed tumor samples the suppression of transcription factors (TFIIH (subunits 42 and 62 kDa) and CUTL1 (CCAAT-translocation protein gene) was detected in cell lines that was not confirmed in tissues. On the other hand, the expression of transcription factor IIIa (TFIIIA) (6,39) was found to be elevated in large number of CRCs. Alterations of the gene expression of growth factors, cytokines, metabolic enzymes, and transport proteins were found in several studies. Using the serial analysis of gene expression method (SAGE), the overexpression of TGF-b was described. TGF-b has an integrin binding site; therefore, it can play a key role in the adhesion and migration of malignant cells. With the help of the compact-sized DNA array filter, the overexpression of VEGF was described. Using large number of tumor samples, several gene expression pattern similarities were found in two studies (6,39). The GRO1 gene product melanoma-growth stimulatory activity (MGSA), which is structurally related to interleukin 8, was found to be elevated in CRC (33,34,36,39,43,44). Mitochondrial genes that were found to be decreased at all Dukes’ stage are: rhodenase, HMG-CoA (hydroxyl-methylglutaryl coenzyme A synthase), and SCAD (acyl-CoA dehydrogenase) (6). Rho-

Table 4 Results of Gene Expression Analyses in GC and CRC Overexpressed Function

Gastric cancer

Downregulated Colorectal cancer

Oncogenes, tumor suppression, apoptosis

GRO1 [21] v -jun v-raf-1 [26] nibrin [25], humanin [25] bcl-2 [26,20] CAS [24] semaphorin V [12]

GRO-1 [39,36,43,34,44,33] v-myc[43,34,33,28] bcl-2 [43,31], CAS [6,39], survivin[43,34]

Cell cycle control, DNA synthesis, cell division

CDK4, CKS1, CKS2 [21], cyclin C [20], cyclin D1 [19], cyclin E [18] CDC25B [21] Ki-67 [19], MIA [20], PCNA[26,21] DNA topoisomerase -2a [19] NE-DIg [12] Retinoblastoma binding protein – 4 [26] IRF7, HOXB7 [20] NFIL3 [20] SRY-box4 [21] and 9

Cyclin E [40], CDC25B [6,39,36,43,34] ENT1, NCL [36] hystone H3[41]

Transcription, transcription factors, translation Growth factors, cytokines

Signal transduction [membrane receptors, protein kinases and phosphatases, signal transduction intermediers]

IGF-2 [21,18,54] TGF-b3 [54,22] PLAB [21 20] FGF-2 HGF [20] HDGF [19]

nm23 [20,23] C-ERBB2 [21,18], C-ERBB3 [24] FGFR-4 [19] IGFR-2 [26] EphB2 [18] H1Histamin receptor [20] Hemopoietic cell kinase-1 [26] SKY [12] GRB-2 [19], and 7 [21,18] p38 MAPK [19,54] Axin-2 [18]

TFIIIA[6,39]

IGF-2 [68] TGF-b [36,32,37,30] TGF-b1 [6,36,34,32] PLAB [43,44] TGF-b-induced gene b-igh3[68,32,28] GRO-g [39] IFN-induced protein 9–27 [36], Teratocarcinoma derived GF1[34] nm23 [40,33,31] ARA24 [36], CD81(¼TAPA-1) [69] NOS2A [36], HEK2 [36] RACK-1 [41]

Gastric cancer

Colorectal cancer

DCC [12], GIF [19] RASSF1[25] caspase-8, 9 [12,26]-10 [12] Death-associated protein-6 [26] FADD-like apoptosis regulator [26] p27 [18] BTG-2 [19] Checkpoint suppressor-1 (CHES1) [19]

c-fos [68], MXI1[36] FRAT1 p53, SIVA,TOSO [40], TN [6] 14-3-3s [36] caspase 10 [36] TRAIL[36,43], TRAIL-R2 sphingomyelin phosphodiesterase [39]

Host cell factor-2 [20] Kruppel-like factor-7 [20]

CUTL1[29] TFIIH (62 e´s 44kDa) [29] ERF 1, zinc finger protein 139 [36], DRA [6] SLU-7 [6] RPL10 [25] HGF[39]

Mal [Tcell differentiation protein] [20] Retinoic acid receptor [12,25] NOS2A [20]

CSF2Rb [36] GM-CSFR a chain precursor [36] TNF-R [39] vitamin D receptor [36] pancreas polypeptid receptor [36] PTPRH [6] ERK1, CKB, CIB [6] DAG-kinase [39] MAX-3, PCK1 [6,32]

p21[29] WEE1 [29] TP1 [36] DNA topoisomerase-2a [29]

(continued)

Table 4 Results of Gene Expression Analyses in GC and CRC (continued) Overexpressed Function

Gastric cancer

Downregulated Colorectal cancer

Gastric cancer

Colorectal cancer COMT, GUCA1B[6,39,34], NADH-ubiquinon oxydoreductase subunit, CKMT [6] amino acid metabolism: creatinine kinase [6,36,32] Lipid metabolism: Apolipoprtotein-C2 17b-hydroxy-steroid dehydrogenase 2, 11b-dehydrogenase [39] L-FABP1[6,34], HU-K5, phospholipase A2 [6] carboxyl-esterase[6,34] HMG-COA synthase, SCAD Carbohydrate metabolism: Alcohol-dehydrogenase-3 [39,36,34,28] -2[34,28]-1[32,28] glucose-6-phosphatase PGM1[6] Carbon dioxide metabolism: carbonic anhydrase I, II [43,34,70,32] and IV [6,39,34,31 32,28] Drug metabolism: cytochrome p450 IID Transport: MaxiKb potassium hannel subunit MAT8, SBP [6,34] metallothyonein [6,40,43,34,28] a1microglobulin Ceratin 19 [6], cytokeratin 20a [39,32] tetranectin b-catenin [40], dynein heavy chain [36] CEACAM1[6,43], CEACAM7 [6] sialophorin (CD43) [36]

Metabolic enzymes and transport proteins

Transport: Solute carrier family- 2, 16, 25 [20] Transferrin receptor [20]

NADH dehydrogenase 2 subunit [41], LDH B [], ACHY [6,39,34] Transport: lipocalin 2 [6,34]

Glutathion-peroxidase [24], dihydrodiol-dehydrogenase [24] pepsinogen C [18] Defense response: Treifol factor-1, 2 [18,20,23] Lipid metabolism: Apolipoprotein-A4,-B, MTP [20,23] Carbohydrate metabolism: Alcohol-dehydrogenase-3 Aldolase-B, Fructose 6-bisphosphatase 1 [20] Aldehyde-dehidrogenase-3[20,23] Carbon dioxide metabolism: carbonic anhydrase II [18,20] Drug metabolism: cytochrome p450 [20] Transport: Gastric intrinsic factor, Hþ/Kþ ATPase, CaþþATPase [20] clathrin [25]

Cell structure and adhesion, motility, extracellular matrix and ECM remodeling

Cytokeratin-1 [12] catenin-b1 [26,25,19]and g [19] matrix metalloproteases: MMP-1, -2,-3,-7, -10,-11,-12,-14, -16 [12,26,21,19 54,22] uPA [21,54] cathepsin-B e´s –K [21] LI- and OB-cadherin [21], E-cadherin [17] collagenes [I, III, IV, V, VI, VII, XVIII] [21,64] a3-integrin[19] fibroblast collagenase inhibitor [24] S100A4 [19] CD9 [12]

Filamin, fibronectin [68,28] cathepsin H [36], collagen I a2 [41,34,31,28] collagens(III,IV,V,X) [43,34,33,31,32,37] lactadherin [6], HME [39], CD24 (¼nectadrin) [41] TIMP-1 [40,67] a6-integrin MMP-1, -2-3,-7,-11[34,31,32]-13 [31]-10 [32]

MUC5AC, MUC6 [18] cadherin-8 [26] ezrin [24] TIMP-3 [19]

(continued)

Table 4 Results of Gene Expression Analyses in GC and CRC (continued) Overexpressed Function Angiogenesis

Immune system regulation, inflammation

Downregulated

Gastric cancer

Colorectal cancer

Gastric cancer

IL-8 [21], Osteonectin [21] Thrombospondin-2 [21,22] VEGF [10] Osteopontin [24] Interferon induced transmembrane protein-2 [20]

IL-8[34] Osteonectin [39,43,34,31,28] Thrombospondin-2[34,32] VEGF [40] Osteopontin [32,28]

Thrombospondin-2 [26]

Colorectal cancer Thrombospondin-4 precursor [39] IgG Fc-binding protein [6], MHC-III HLA-RP, IgE Fc-fragment I-type receptor [36], adipsin [6,39]

List of abbreviations, not mentioned in the text: ACHY: S-adenosyl-homocysteine hydrolase; ARA: androgen receptor associated protein; BTG-2: B cell translocation gene-2; CEACAM: carcinoembrional antigen family, pp120; CIB: calcium and integrin-binding protein; CKB: creatin kinase brain; CKMT: mitochondrial creatin kinase; CKS1,2: cyclin dependent kinase regulatory subunit 1,2; COMT: cathecol-o-methyl-transferase; CSF2R: colony stimulating factor 2 receptor; DAG: diacyl-glycerol; DRA: colon-mucosa-associated protein; ECM: extracellular matrix; EphB2: ephrin receptor tyrosine kinase; ERF: eukaryotic releasing factor, transcription factor; ERK: protein Ser/Thr kinase; FRAT: frequently rearranged in advanced T-cell lymphomas; GMCSFR: granulocyte-monocyte colony stimulating factor receptor; GRO: growth related oncogen, chemoattractant cytokin; GUCA: guanylate-cyclate activator (guanilyn); HEK: heterogeneous nuclear RNP K-like gene; HLA: human leukocyte antigen; HME: human metalloproteinase; HOXB7: homeobox B7; HU-K5: lysophospholipase homolog; IFN: interferon; IGF: insulin-like growth factor; IRF7: interferon regulatory factor-7; Ki-67: proliferation marker; LDH: lactate-dehydrogenase; L-FABP: fatty acid binding protein liver; MAT: phospholemman-like protein; MAX: Myc-associated factor X; MHC: major histocompatibility complex; MIA: melanoma inhibitory activator, MUC: mucin; MXI: MAX-protein associated protein; NADH: protonated nicotinamid adenin dinucleotide; NE-DIg: neuronal and endocrin dig (Drosophila dig tumor suppressor gene homolog); NFIL3: interleukin 3 regulated nuclear factor; PCNA: proliferative cell nuclear antigen; PGM: phosphoglucomutase; PTPRH: H-type protein tyrosine phosphatase receptor; RACK: activated C kinase receptor; RPL10: ribosomal protein L10; S100A4: placental calciumbinding protein, SBP: selenium binding protein; SLU: II-step splicing factor; SRY: sex determination region Y; TNF-R: tumor necrosis factor receptor; TRAIL: TNF-related apoptosis inducer ligand; TRAIL-R: TRAIL-receptor; TP: telomerase-associated protein Note: Gene systematization based on their main function, but some gene products are involved in multiple functional groups.

12

GALAMB ET AL.

denase (thiosulfate sulfotransferase) is involved in the formation of iron-sulfur complexes and cyanide-detoxification. HMG-CoA synthase catalyses the condensation of acetyl-CoA with acetoacetyl-CoA to produce HMG-CoA and CoA. The mitochondrial form is responsible for ketone body synthesis. SCAD catalyses the b-oxydation of butyryl-CoA to acetyl-CoA. SCAD deficiency results in an increase of butyric acid level. Butyrate is the primary source of energy for colonocytes. At low concentrations, butyrate stimulates growth under glucose- and pyruvate depleted conditions, whereas it causes apoptosis at the same concentrations in the presence of glucose and pyruvate (77). The damage of mitochondrial system, which is showed by underexpression of nuclear encoded mitochondrial genes, was detected in several publications (6,31,35). It is conceivable that one possible cause of CRC carcinogenesis might be located in the mitochondria. Mitochondrial DNA accumulates more damage because of less efficient repair systems in the mitochondria compared with those in the nucleus. Although all of the proteins mentioned earlier are nuclear encoded, the mitochondrial function is altered directly by expression changes of nuclear encoded proteins involved in electron transport and oxydative phosphorylation and altered indirectly because oxydative phosphorylation is linked to many pathways of intermediary metabolism (70). The expression of other metabolic enzymes, such as carbonic anhydrase I, II (32,34,43,68,70), and IV (6,31–34,39), was found to be decreased in CRC. The low expression of carbonic anhydrase I in CRC correlates to the degree of vascular invasion and to the poor prognosis (78). Some elements of common signal transduction pathways were found to be overexpressed. The gene of nm23 (metastasis inhibiting gene in melanoma) (31,33,40) or of TIMP1 (tissue inhibitor of metalloproteinase 1) (31,40) showed high expression levels in cancer tissue. The upregulation of nm23 and TIMPs correlated with reduced patient survival in CRC (79,80). Interestingly, the gene of the cell structure protein cytokeratin 20, which is a marker of complete terminal differentiation of epithelial derived cells, showed downregulation in CRC (32,39,68). The gene of the immune regulatory protein adipsin, which is equivalent to complement factor D and which normally shows high expression level in adipose tissue, was found to be downregulated in CRC (6,39) as well as chromogranin A (CgA, secretion-associated protein, neuroendocrine differentiation marker). Based on international studies, Table 4 shows the main genes that have changes in their expression levels in CRC, classified into functional groups.

COMPARISON OF THE GENE EXPRESSION DATA IN GASTRIC AND COLORECTAL CANCER Despite the different researched target oligo set and different sample collection and preparation, some simi-

larities in the expression pattern of certain genes can be found in gastric and colorectal carcinoma samples (see Table 4) referring to the common cell functions that change during the carcinogenesis, such as apoptosis, cell cycle, growth factor-related proliferation and signal transduction, cell adhesion, and angiogenesis. Table 4 contains the commonly up- or downregulated genes that were identified in earlier available gene expression studies and involved in the carcinogenes and progression of these cancer types. It is important for explanation and understanding of their pathomechanisms. With the help of more newly published data and results generated in the future, microarray analysis will approach the clinical practice. POSSIBILITIES OF CLINICAL APPLICATIONS The chip technology offers opportunities not only for super-early cancer diagnosis, but exact, molecular-based histological diagnosis of the extant malignant alterations. Nowadays, because of DNA-based (polymorphism) and expression microarray analyses, large amount of data have been collected about the development, progression, and metastasis generation of GC and CRC. The preoperative grading and staging of the tumor fail to predict reliably its metastatic potential; therefore, diagnostic tools are needed to link the lymphatic spread and phenotype to gene expression profiles of the biopsy specimen analyzed preoperatively. This allows us to use less aggressive operative tools, such as endoscopy or laparoscopy or smaller resections in the therapeutic process. If a particular gene expression pattern in the primary tumor were proven to predict metachronous metastases, adjuvant treatment options could focus on these genetically high-risk categories and after care programs could be adapted to their individual risk patterns. Figure 1 shows the proposal of modern GC and CRC diagnostic methodology and marker panel, with combination of the conventional diagnostic procedure and the new diagnostic methods are based on the examination of further molecular markers. Examination of certain molecular markers in less advanced stages of the tumorous diseases can be used for super-early diagnosis. CORRELATION OF GENE EXPRESSION MICROARRAY RESULTS WITH THE HISTOLOGICAL SUBTYPE, DIFFERENTIATION GRADE, AND DISEASE STAGE IN GASTRIC CANCER Most of the gastric microarray studies divide the samples into intestinal or diffuse histological types, according to the Lauren classification (17–21,23,24,27). The grade of differentiation, which the prognosis refers to, was shown in a few publications (17,23,24). The stage of the disease is described in different ways according to different methods (Borrman, TNM, for the recommendations of AJCC or Japanese Research Society for Gastric Cancer), and there are articles in which the stage of gastric cancer is not described. Data for every sample, according to all of the three classifying factors men-

THE RESULTS OF THE EXPRESSION ARRAY STUDIES IN GASTRIC AND COLORECTAL CANCER

13

FIG. 1. Proposal of modern GC and CRC diagnostic methodology and marker panel. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

tioned earlier are found in only some articles such as publications of Kim et al. (23) and Lee et al. (24). In most of the studies, only commonly over- or underexpressed genes were shown, regardless of the stage of the gastric cancers (17), or they could not show correlation between these data and the gene expression results (19,23,24), but several of them identified histologicaltype specific (17,23,27) or lymph-node metastasis-related (20–22,24,27) gene expression pattern. Bousssitas et al. (17) used 124 samples for microarray analysis: intestinal type GC was characterized by markers of proliferation (TOP2A, CDC25B), while diffuse GC expression signature reflected active ECM production, remodeling (collagens, biglycan, osteoglycin, cadherin 11, MMPs, fibrillin, Thy-1 serpins), and complex signaling positive and negative regulators of cell growth (SPARC, Wnt5a, DLG5, DKK3, secreted frizzled-related protein 4), and Ecadherin, occludin were downregulated. As Boussitas et al. (17), Chen et al. (18) also worked with higher number of samples, but they came to the conclusion that variation of gene expression pattern is very high in gastric cancer tissues. It was emphasized that gastric cancerous tissue is histologically complex, and the tumor cells and the stromal cells play an important role in tumor biology. The laser microdissection, as we have already mentioned, and the appropriate handling and amplification of small number of cell samples can help explain the differences in the patterns of gene expression on the basis of what cell types they origi-

nated from. Using this method, we avoid, or at least decrease, the expression differences that originate from the differences of the ratios of different cell types only. The work of Inoue et al. (22) is a good example for prognostic score determination, combines the role of several different genes and their significance, and solves the problem of patient follow-up. They found 78 differently expressed genes in gastric cancer samples when compared with those of the normal tissue, and generated prognostic scores summarizing the malignancy-association coefficients of every genes in every cases. The malignancy-association coefficients were established according to the dept of invasion and the lymph node metastasis. MMP-7, osteonectin, TGFb3, IGFBP3, CEArelated cell adhesion molecule 6, fibronectin precursor, and PCNA were found to be principally associated to gastric cancer progression, tumor invasion, and metastatic ability in that study. The authors—in accordance with findings of Lee et al. (24)—raised an issue that overexpression of thrombospondin 2 angiostatic factor can be one of the special indicator of tumor progression and metastasis in gastric cancer. They showed correlation between these prognostic scores and the survival after the surgical intervention. Others found IL8, VEGF, OPN, MMP9, CD44v9 (21), DDOST, GNS, NEDD8, CCT3 and 5, ubiquilin 1 (20), interferon-stimulated protein, 15 kDa, RAB11A, cadherin 2, FAT tumor suppressor gene (27) as the main lymph node metastasis-associated genes.

14

GALAMB ET AL.

GENE EXPRESSION BASED CLASSIFICATION OF COLORECTAL CANCER ACCORDING TO THE DISEASE STAGE, LOCALIZATION, AND METASTASIS STATUS – THE FIRST STEPS Several researchers identified that the genes showed different mRNA expression level between CRC (all stages, all localization, etc.) and the normal tissue samples (29,34,36,39–41,43). They described only the commonly up- and downregulated genes in CRC. Recently, a number of analyses, which aimed to find stage, metastasis status (6,28,30,31,33,35,37,38,44), MSI phenotype, localization specific (31–33,44) and prognostic (42), have been done. Agrawal et al. (28) prepared a microarray study with more number of samples: 10–10 normal, adenoma, CRCB1, CRC-C2, CRC-D (AC stages) and 20 resected liver metastases (totally 70 samples) were involved. One hundred and seven genes were demonstrated as continually up-, or downregulated, according to the disease stages. Among these candidate markers, they emphasized and firstly described the osteopontin, a secreted integrinbinding molecule as a strong marker of CRC progression. The osteopontin stage-specific overexpression was confirmed by Northern-blot on mRNA level and immunohistochemically on protein level. The sample numbers per sample groups were considerably lower in other microarray studies (4-5 Dukes B and C (44), 6-6 Dukes B and C (37), 5-6-6-4 Dukes A, B, C, and D (6), 5-5 Dukes A,B,C and D (35)) for finding lymph-node positivity related (37,38,44) and localization-associated (44) gene expression patterns. Birkenkamp-Demtroder et al. (6) and Frederiksen et al. (35) could only distinguish Dukes’ B and C stages. Birkenkamp-Demtroder et al. (6) determined several genes that are connected to cancer progression. Some of the selected candidate genes were regarded as progression markers and molecular predictors (changes in at least two consecutive Dukes’ stages) or Dukes’ classifiers (major changes in one Dukes’ stage only). Examples of progression markers are phosphoenolpyruvate carboxykinase (PCK1) and monocyte-derived neutrophil chemotactic factor (MDNCF, IL-8). The latter is an angiogenic cytokine, and was shown to be produced by the tumor cells. Many genes that showed a decrease or loss of expression in at least one Dukes’ stage encoded mitochondrial proteins. Some newly appeared articles published in 2005 have focused on correlation between clinicopathologic parameters and gene expression data (32,33,42). Microarray and RT-PCR analysis of 31 CRC/normal tissue pairs were performed by Chiu et al. (33), and upregulated genes were determined, which showed correlation with clinicopathological factors. Increased level of GPX2 (glutathione peroxidase 2) mRNA was related to a higher degree of tumor differentiation. GRO1 oncogene, which was found to be overexpressed in a number of CRC studies (33,34,36,39,43,44), was significantly upregulated in patients younger than 65 years; tumor stage, lymph node metastasis, and serum CEA levels were found to be frequently overexpressed when analyzed in correlation

with GRO1. Overexpression of RNA binding HNRPA1(heterogeneous nuclear ribonucleoprotein A1) was significant in the right side of colon than in the left side. Birkenkamp-Demtroder et al. (32) also desrcibed different gene expression in colon cancer of the caecum versus the sigmoid and rectusigmoid. They found that 30 genes differentially expressed in tumor tissue were common to adenocarcinomas of both sides, including known tumor markers such as MMPs, keratin 8, 19, and 20 as well as carbonic anhidrases (II, IV, VII) showed side-specific expression and were downregulated in the left side tumors, whereas teratocarcinoma growth factor and cyclooxygenase 2 were upregulated in the left side adenocarcinomas. Wang et al. (42), using large number of homogeneous samples (74 Dukes B samples) identified 23-gene signature (including genes involved in cell proliferation and tumor progression such as tyrosine 3-monooxygenase tryptophan 5-monooxygenase activation protein (YWHAH), cell cycle gene RCC1, transcription factor BTEB2; genes involved in local immune response such as immunoglobulin-like transcript 5 protein, gelsolin-like capping protein, LAT) that predicts recurrence in Dukes B patients. By summarization of these data, it can be seen that there are a number of useful markers associated with different stages (osteopontin, GRO1), localization (HNRPA1, MMPs, keratin 8, 19, 20; carbonic anhidrases II, IV, VII, teratocarcinoma growth factor, COX2), lymph node (GRO1) and visceral metastasis status (PCSK7, PRAME, IQ-GAP1, LTPBh, IGFR1), differentiation degrees (GPX2) and predictive outcomes (YWHAH, RCC1, BTEB2, LAT, gelsolin-like capping protein).

CONCLUSIONS In our days, the microarray technology is relatively new and expensive, furthermore, the statistical analysis of the microarray data is not completely settled,. Systematization of microarray results of different research teams is very complicated and different. To make expression microarray data easily comparable, we should standardize each and every steps of the microarray analysis procedure: from the sample collection and storage, through the RNA isolation, cDNA preparation and microarray hybridization to the signal detection and data analysis. Diversity of the target gene sequences, which are fixed to the microarray surface, raises further questions. As the whole-genome microarrays have appeared, interpretation and correlation of huge amount of gene expression information and large number of clinical parameters are waiting to be executed. Selection of stage and histological subtype-specific genes is difficult according to the different gene expression analysis studies. Main reasons are the relatively low sample number and heterogeneous sample groups in most of the analyses, and the lack of the exact and standard histological, differentiation grade, and stage classification. Aims of the presently available publications are also

THE RESULTS OF THE EXPRESSION ARRAY STUDIES IN GASTRIC AND COLORECTAL CANCER

often different, and usually focus on one classification viewpoint only (for example histological type or metastatic or nonmetastatic tumor), the sample groups are often heterogeneous in other respects. Therefore, the expression microarray results are different, and in most of the cases are difficult to find overlapping common markers. For validation of the marker properties of a given gene in a given respect (for example differentiation marker which is important in the disease prognosis), we should use homogeneous high case-number sample groups for microarray analysis, and further experiments (siRNA technology, protein-level analyses), but at least the confirmation of the given gene by RT-PCR, are needed. Nonetheless, more and more stage and subtype-specific data become available in the microarray literature of CRC, less in GC, and expectedly more connection will be found between the results of different research teams. If the systematization of these genetic and epigenetic alterations will be finished, microarrays will also play an important role in daily clinical practice, besides the conventional diagnostic tools. For individual patients, more precise diagnosis and risk assessment based on expression profiles are already achievable, leading to more accurate determination of prognosis and more individually tailored treatment. Genetic profiling could play a role in deciding which patients to be spared from chemotherapy and which regimen offers the highest efficacy and the lowest rate of side effects. The definition of specific gene expression profiles could lead to more effective prevention and early detection strategies in GCs and in aggressive, hereditary forms of CRCs. Microarray technology assists to avoid the intermediate and nondescript cases in diagnostics, which do not belong to any of the conventional diagnostic groups. In addition, the diagnostic routine will be easier, and the prognosis and therapeutic response will also be predictable by creation of disease-specific so-called miniarrays with less number of genes. The future target in microarray technology will be the development of clustered gene chips that specifically characterize each tumor type and focus on gene expression that specifies cell identity (16). Realization of analysis of candidate markers or markergroups determined by microarray technology from peripheral blood or biopsy samples can enhance the diagnostics and prognostics of the gastrointestinal cancers. LITERATURE CITED 1. Wu CW, Lo SS, Shen KH, Hsieh MC, Lui WY, P’eng FK. Surgical mortality, survival, and quality of life after resection for gastric cancer in the elderly. World J Surg 2000;24:465–472. 2. Ajani JA, Mansfeld PF, Ota DM. Potentially resectable gastric carcinoma: current approaches to staging and preoperative therapy. World J Surg 1995;19:216–220. 3. Ries LAG.SEER Cancer Statistics Review, 1973–1997. Bethesda, MD:National Cancer Institute; 2000. 4. Pohl C, Hombach A, Kruis W. Chronic inflammatory bowel disease and cancer. Hepatogastroenterology 2000;47:57–70. 5. Fearon EF, Hamilton SR, Vogelstein B. Clonal analysis of human colorectal tumors. Science 1987;238:193–197. 6. Birkenkamp-Demtroder K, Christensen LL, Olesen SH, Frederiksen CM, Laiho P, Aaltonen LA, Laurberg S, Sorensen FB, Hagemann R,

7. 8.

9. 10. 11. 12.

13. 14. 15. 16. 17.

18.

19. 20.

21.

22. 23.

24.

25.

26.

27.

28.

29.

15

ORntoft TF. Gene expression in colorectal cancer. Cancer Res 2002;62:4352–4363. Forrester K, Almoguera C, Han K, Grizzle WE, Perucho M. Detection of high incidence of K-ras oncogenes during human colon tumorigenesis. Nature 1987;327:298–303. Baker SJ, Fearon ER, Nigro JM, Hamilton SR, Preisinger AC, Jessup JM, vanTuinen P, Ledbetter DH, Barker DF, Nakamura Y. Chromosome 17 deletions and p53 gene mutations in colorectal carcinomas. Science 1989;244:217–221. Thibodeau SN, Bren G, Schaid D. Microsatellite instability in cancer of the proximal colon. Science 1993;260:816–819. Jones PA, Laird PW. Cancer epigenetics comes of age. Nat Genet 1999;21:163–167. Cui H, Horon IL, Ohlsson R, Hamilton SR, Feinberg AP. Loss of imprinting in normal tissue of colorectal cancer patients with microsatellite instability. Nat Med 1998;4:1276–1280. Liu LX, Liu ZH, Jiang HC, Qu X, Zhang WH, Wu LF, Zhu AL, Wang XQ, Wu M. Profiling of differentially expressed genes in human gastric carcinoma by cDNA expression array. World J Gastroenterol 2002;8:580–585. Lipshutz RJ, Fodor SP, Gingeras TR, Lockhart DJ. High density synthetic oligonucleotide arrays. Nat Genet 1998;21(Suppl 1):20–24. Gabrielson E, Berg K, Anbazhagan R. Functional genomics, gene arrays, and the future of pathology. Mod Pathol 2001;14:1294–1299. Galamb O, Molnar B, Tulassay Z. DNA chips for gene expression analysis and their applications in diagnostics. Orv Hetil 2003;144: 21–27. Stremmel C, Wein A, Hohenberger W, Reingruber B. DNA microarrays: a new diagnostic tool and its implications in colorectal cancer. Int J Colorectal Dis 2002;17:131–136. Boussioutas A, Li H, Liu J, Waring P, Lade S, Holloway AJ, Taupin D, Gorringe K, Haviv I, Desmond PV, Bowtell DD. Distinctive patterns of gene expression in premalignant gastric mucosa and gastric cancer. Cancer Res 2003;63:2569–2577. Chen X, Leung SY, Yuen ST, Chu KM, Ji J, Li R, Chan AS, Law S, Troyanskaya OG, Wong J, So S, Botstein D, Brown PO. Variation in gene expression pattern in human gastric cancers. Mol Biol Cell 2003;14:3208–3215. El-Rifai W, Frierson HF, Harper J, Powell SM, Knuutila S. Expression profiling of gastric adenocarcinoma using cDNA array. Int J Cancer 2001;92:832–838. Hasegawa S, Furukawa Y, Li M, Satoh S, Kato T, Watanabe T, Katagiri T, Tsunoda T, Yamaoka Y, Nakamura Y. Genome-wide analysis of gene expression in intestinal-type gastric cancers using a complementary DNA microarray representing 23,040 genes. Cancer Res 2002;62:7012–7017. Hippo Y, Taniguchi H, Tsutsumi S Machida N, Chong JM, Fukayama M, Kodama T, Aburatani H. Global gene expression analysis of gastric cancer by oligonucleotide microarrays. Cancer Res 2002;62: 233–240. Inoue H, Matsuyama A, Mimori K, Ueo H, Mori M. Prognostic score of gastric cancer determined by cDNA microarray. Clin Cancer Res 2002;8:3475–3479. Kim B, Bang S, Lee S, Kim S, Jung Y, Lee C, Choi K, Lee SG, Lee K, Lee Y, Kim SS, Yeom YI, Kim YS, Yoo HS, Song K, Lee I. Expression profiling and subtype-specific expression of stomach cancer. Cancer Res 2003;63:8248–8255. Lee S, Baek M, Yang H, Bang YJ, Kim WH, Ha JH, Kim DK, Jeoung DI. Identification of genes differentially expressed between gastric cancers and normal mucosa with cDNA microarrays. Cancer Lett 2002;184:197–206. Meireles SI, Carvalho AF, Hirata R, Montagnini AL, Martins WK, Runza FB, Stolf BS, Termini L, Neto CE, Silva RL, Soares FA, Neves EJ, Reis LF. Differentially expressed genes in gastric tumors identified by cDNA array. Cancer Letters 2003;190:199–211. Mori M, Mimori K, Yoshikawa Y, Shibuta K, Utsunomiya T, Sadanaga N, Tanaka F, Matsuyama A, Inoue H, Sugimachi K. Analysis of the gene-expression profile regarding the progression of human gastric carcinoma. Surgery 2002;131(Suppl 1):S39–S47. Norsett KG, Laegreid A, Midelfart H, Yadetie F, Erlandsen SE, Falkmer S, Gronbech JE, Waldum HL, Komorowski J, Sandvik AK. Gene expression based classification of gastric carcinoma. Cancer Lett 2004;210:227–237. Agrawal D, Chen T, Irby R, Quackenbush J, Chambers AF, Szabo M, Cantor A, Coppola D, Yeatman TJ. Osteopontin identified as lead marker of colon cancer progression, using pooled sample expression profiling. J Natl Cancer Inst 2002;94:513–521. Backert S, Gelos M, Kobalz U, Hanski ML, Bohm C, Mann B, Lovin N, Gratchev A, Mansmann U, Moyer MP, Riecken EO, Hanski C. Differential gene expression in colon carcinoma cells and tissues detected with a cDNA array. Int J Cancer 1999;82:868–874.

16

GALAMB ET AL.

30. Bandres E, Catalan V, Sola I, Honorato B, Cubedo E, Cordeu L, Andion E, Escalada A, Zarate R, Salgado E, Zabalegui N, Garcia F. Garcia-Foncillas dysregulation of apoptosis is a major mechanism in the lymph node involvement in colorectal carcinoma. J Oncol Rep 2004;12:287–292. 31. Bertucci F, Salas S, Eysteries S, Nasser V, Finetti P, Ginestier C, Charafe-Jauffret E, Loriod B, Bachelart L, Montfort J, Victorero G, Viret F, Ollendorff V, Fert V, Giovaninni M, Delpero JR, Nguyen C, Viens P, Monges G, Birnbaum D, Houlgatte R. Gene expression profiling of colon cancer by DNA microarrays and correlation with histoclinical parameters. Oncogene 2004;23:1377–1391. 32. Birkenkamp-Demtroder K, Olesen SH, Sorensen FB, Laurberg S, Laiho P, Aaltonen LA, Orntoft TF. Differential gene expression in colon cancer of the caecum versus the sigmoid and rectosigmoid. Gut 2005;54:374–384. 33. Chiu ST, Hsieh FJ, Chen SW, Chen CL, Shu HF, Li H. Clinicopathologic correlation of up-regulated genes identified using cDNA microarray and real-time reverse transcription-PCR in human colorectal cancer. Cancer Epidemiol Biomarkers Prev 2005;14:437–443. 34. Croner RS, Foertsch T, Brueckl WM, Guenther K, Siebenhaar R, Stremmel C, Matzel KE, Papadopoulos T, Kirchner T, Behrens J, Klein-Hitpass L, Stuerzl M, Hohenberger W, Reingruber B. Common denominator genes that distinguish colorectal carcinoma from normal mucosa. Int J Colorectal Dis 2004; Dec 22 (Epub ahead of print). 35. Frederiksen CM, Knudsen S, Laurberg S, Orntoft TF. Classification of Dukes’ B and C colorectal cancers using expression arrays. J Cancer Res Clin Oncol 2003;129:263–271. 36. Kitahara O, Furukawa Y, Tanaka T, Kihara C, Ono K, Yanagawa R, Nita ME, Takagi T, Nakamura Y, Tsunoda T. Alterations of gene expression during colorectal carcinogenesis revealed by cDNA microarrays after laser-capture microdissection of tumor tissues and normal epithelia. Cancer Res 2001;61:3544–3549. 37. Kwon HC, Kim SH, Roh MS, Kim JS, Lee HS, Choi HJ, Jeong JS, Kim HJ, Hwang TH. Gene expression profiling in lymph node-positive and lymph node-negative colorectal cancer. Dis Col Rect 2004;47: 141–152. 38. Li M, Lin YM, Hasegawa S, Shimokawa T, Murata K, Kameyama M, Ishikawa O, Katagiri T, Tsunoda T, Nakamura Y, Furukawa Y. Genes associated with liver metastasis of colon cancer, identified by genome-wide cDNA microarray. Int J Oncol 2004;24:305–312. 39. Notterman DA, Alon U, Sierk AJ, Levine AJ. Transcriptional gene expression profiles of colorectal adenoma, adenocarcinoma, and normal tissue examined by oligonucleotide arrays. Cancer Res 2001;61:3124–3130. 40. Okuno K, Yasutomi M, Nishimura N, Arakawa T, Shiomi M, Hida J, Ueda K, Minami K. Gene expression analysis in colorectal cancer using practical DNA array filter. Dis Colon Rectum 2001;44:295–299. 41. Saito A, Fujii G, Sato Y, Gotoh M, Sakamoto M, Toda G, Hirohashi S. Detection of genes expressed in primary colon cancers by in situ hybridization: overexpression of RACK 1. J Clin Pathol Mol Pathol 2002;55:34–39. 42. Wang Y, Jatkoe T, Zhang Y, Mutch MG, Talantov D, Jiang J, McLeod HL, Atkins D. Gene expression profiles and molecular markers to predict recurrence of Dukes’ B colon cancer. J Clin Oncol 2004;22:1564–1571. 43. Williams NS, Gaynor RB, Scoggin S, Verma U, Gokaslan T, Simmang C, Fleming J, Tavana D, Frenkel E, Becerra C. Identification and validation of genes involved in the pathogenesis of colorectal cancer using cDNA microarrays and RNA interference. Clin Cancer Res 2003;9:931–946. 44. Zou TT, Selaru FM, Xu Y, Shustova V, Yin J, Mori Y, Shibata D, Sato F, Wang S, Olaru A, Deacu E, Liu TC, Abraham JM, Meltzer SJ. Application of cDNA microarrays to generate a molecular taxonomy capable of distinguishing between colon cancer and normal colon. Oncogene 2002;21:4855–4862. 45. Wang D, Yang W, Du J, Devalaraja MN, Liang P, Matsumoto K, Tsubakimoto K, Endo T, Richmond A. MGSA/GRO-mediated melanocyte transformation involves induction of Ras expression. Oncogene 2000;19:4647–4659. 46. Yokozaki H, Yasui W, Tahara E. Genetic and epigenetic changes in stomach cancer. Int Rev Cytol 2001;204:49–95. 47. Tsutsui S, Ohno S, Murakami S, Hachitanda Y, Oda S. Prognostic value of c-erbB2 expression in breast cancer. J Surg Oncol 2002;79:216–223. 48. Wilkinson MG, Millar JBA. Control of the eukaryotic cell cycle by MAP kinase signaling pathways. FASEB J 2000;14:2147–2157. 49. Sakaguchi T, Watanabe A, Sawada H, Yamada Y, Yamashita J, Matsuda M, Nakajima M, Miwa T, Hirao T, Nakano H. Prognostic value of cyclin E and p53 expression in gastric carcinoma. Cancer 1998;82:1238–1243.

50. Matsumoto K, Nakamura T. NK4 (HGF-antagonist/angiogenesis inhibitor) in cancer biology and therapeutics. Cancer Sci 2003;94: 321–327. 51. Kuniyasu H, Yasui W, Kitadai Y, Yokozaki H, Ito H, Tahara E. Frequent amplification of the c-met gene in scirrhous type stomach cancer. Biochem Biophys Res Commun 1992;189:227–232. 52. Ellis MJ, Jenkins S, Hanfelt J, Redington ME, Taylor M, Leek R, Siddle K, Harris A. Insulin-like growth factors in human breast cancer. Breast Cancer Res Treat 1998;52:175–184. 53. Freier S, Weiss O, Eran M, Flyvbjerg A, Dahan R, Nephesh I, Safra T, Shiloni E, Raz I. Expression of the insulin-like growth factors and their receptors in adenocarcinoma of the colon. Gut 1999;44:704–708. 54. Yasui W, Oue N, Kuniyasu H, Ito R, Tahara E, Yokozaki H. Molecular diagnosis of gastric cancer: present and future. Gastric Cancer 2001;4:113–121. 55. Li C, Wang J, Wilson PB, Kumar P, Levine E, Hunter RD, Kumar S. Role of transforming growth factor b3 in lymphatic metastasis in breast cancer. Int J Cancer 1998;79:455–459. 56. Tan M, Wang Y, Guan K, Sun Y. PTGF-b, a type b transforming growth factor (TGF-b) superfamily member, is a p53 target gene that inhibits tumor cell growth via TGF-b signaling pathway. Proc Natl Acad Sci USA 2000;97:109–114. 57. Hart KC, Robertson SC, Kanemitsu MY, Meyer AN, Tynan JA, Donoghue DJ. Transformation and Stat activation by derivatives of FGFR1, FGFR3, and FGFR4. Oncogene 2000;19:3309–3320. 58. Widschwendter M, Berger J, Daxenbichler G, Muller-Holzner E, Widschwendter A, Mayr A, Marth C, Zeimet AG. Loss of retinoic acid receptor b expression in breast cancer and morphologically normal adjacent tissue but not in the normal breast tissue distant from the cancer. Cancer Res 1997;57:4158–4161. 59. Han DC, Guan JL. Association of focal adhesion kinase with Grb7 and its role in cell migration. J Biol Chem 1999;274:24425–24430. 60. Sutherland R, Delia D, Schneider C, Newman R, Kemshead J, Greaves M. Ubiquitous cell-surface glycoprotein on tumor cells is proliferation-associated receptor for transferrin. Proc Natl Acad Sci USA 1981;78:4515–4519. 61. Lochter A, Sternlicht MD, Werb Z, Bissell MJ. The significance of matrix metalloproteinases during early stages of tumor progression. Ann NY Acad Sci 1998;857:180–193. 62. Patterson BC, Sang QA. Angiostatin-converting enzyme activities of human matrilysin (MMP-7) and gelatinase B/type IV collagenase (MMP-9). J Biol Chem 1997;272:28823–28825. 63. Powell WC, Fingleton B, Wilson CL, Boothby M, Matrisian LM. The metalloproteinase matrilysin proteolytically generates active soluble Fas ligand and potentiates epithelial cell apoptosis. Curr Biol 1999;9:1441–1447. 64. Crawford HC, Fingleton BM, Rudolph-Owen LA, Goss KJ, Rubinfeld B, Polakis P, Matrisian LM. The metalloproteinase matrilysin is a target of b-catenin transactivation in intestinal tumors. Oncogene 1999;18:2883–2891. 65. Werner M, Becker KF, Keller G, Hofler H. Gastric adenocarcinoma: pathomorphology and molecular pathology. J Cancer Res Clin Oncol 2001;127:207–216. 66. Duffy MJ. Urokinase-type plasminogen activator: a potent marker of metastatic potential in human cancer. Biochem Soc Trans 2002;30: 207–210. 67. Buck MR, Karustis DG, Day NA, Honn KV, Sloane BF. Degradation of extracellular-matrix proteins by human cathepsin B from normal and tumour tissues. Biochem J 1992;282:273–278. 68. Zhang L, Zhou W, Velculescu VE, Kern SE, Hruban RH, Hamilton SR, Vogelstein B, Kinzler KW. Gene expression profiles in normal and cancer cells. Science 1997;276:1268–1272. 69. Eastham JA, Hall SJ, Sehgal I, Wang J, Timme TL, Yang G, ConnellCrowley L, Elledge SJ, Zhang WW, Harper JW. In vivo gene therapy with p53 or p21 adenovirus for prostate cancer. Cancer Res 1995;55:5151–5155. 70. Augenlicht LH, Heerdt BG. Mitochondria: integrators in tumorigenesis? Nat Genet 2001;28:104–105. 71. Eagle LR, Yin X, Brothman AR, Williams BJ, Atkin NB, Prochownik EV. Mutation of the MXI1 gene in prostate cancer. Nat Genet 1995;9:249–255. 72. Hu WH, Johnson H, Shu HB. Tumor-necrosis factor related apoptosisinducing ligand receptors signal NF-kb and LNK activation and apoptosis through distinct pathways. J Biol Chem 1999;274:30603– 30610. 73. Cass CE, Young JD, Baldwin SA. Recent advances in the molecular biology of nucleoside transporters of mammalian cells. Biochem Cell Biol 1998;76:761–770. 74. De Souza CP, Ellem KA, Gabrielli BG. Centrosomal and cytoplasmic Cdc2/cyclin B1 activation precedes nuclear mitotic events. Exp Cell Res 2000;257:11–21.

THE RESULTS OF THE EXPRESSION ARRAY STUDIES IN GASTRIC AND COLORECTAL CANCER

75. Ambs S, Merriam WG, Bennet WP, Felley-Bosco E, Ogunfusika MO, Oser SM, Klein S, Shields PG, Billiar TR, Harris CC. Frequent nitric-oxide synthase-2 expression in human colon adenomas: implication for tumor angiogenesis and colon cancer progression. Cancer Res 1998;58:334–341. 76. Erard MS, Belenguer P, Caizergues-Ferrer M, Pantaloni A, Amalric F. A major nucleolar protein, nucleolin, induces chromatin decondensation by binding to histone H1. Eur J Biochem 1988;175:525–530. 77. Singh B, Halestrap AP, Paraskeva C. Butyrate can act as a stimulator of growth or inducer of apoptosis in human colonic epithelial cell lines depending on the presence of alternative

17

energy sources. Carcinogenesis 1997;18:1265–1270. 78. Mori M, Staniunas RJ, Barnard GF, Jessup JM, Steele GD, Jr, Chen LB. The significance of carbonic anhydrase expression in human colorectal cancer. Gastroenterology 1993;105:820–826. 79. Berney CR, Yang J, Fisher RJ, Russell PJ, Crowe PJ. Overexpression of nm23 protein assessed by color video image analysis in metastatic colorectal cancer: correlation with reduced patient survival. World J Surg 1998;22:484–490. 80. Garbett EA, Reed MW, Brown NJ. Proteolysis in human breast and colorectal cancer. Br J Cancer 1999;81:287–293.