Gastric cancer (GC) is only second to lung cancer in world-wide cancer-related deaths, however there are great geographical differences in GC distribution. Data from 2010 demonstrate that the GC incidence in Norway is very low (males 6.9, females 3.0 per 100.000) [1] compared to less developed areas, particularly in Eastern Asia, where the incidence is approximately 6-fold (males 42.4, females 18.3 per 100.000) [2].

Gastric adenocarcinoma is remarkably heterogeneous genetically, cytologically and architecturally compared to other gastrointestinal carcinomas. The search for reliable tumor markers and consistent prognostic indicators has proven difficult. Several authors have attempted to predict GC disease and prognosis based on single or multiple genes [3–8], but there are discrepancies between the studies, and currently no gene signature or biomarkers are in routine clinical use. Understanding the mechanisms underlying gastric cancer is one of the major challenges in cancer genomics. The Lauren classification divides adenocarcinomas into three different histological subtypes: intestinal and diffuse types and a mixed variant [9], which are thought to take different pathways of carcinogenesis. The intestinal type is attributable to a multistep progression from chronic gastritis through gastric atrophy, metaplasia, dysplasia and ultimately malignant disease [10]. Diffuse types may arise from chronic inflammation without a clear manifestation of intermediate premalignant steps [11–13]. The mixed type shows non-homogenous mixtures of both intestinal and diffuse type architecture, and might represent a separate cancer category with exclusive gene mutations and a more aggressive course [14, 15]. In spite of extensive research into the genetic changes of GC, the mechanisms underlying the disease are still far from understood, and the disease cannot easily be explained by an adenoma-carcinoma model like in colorectal cancer. There are three molecular mechanisms that drive gastric carcinogenesis: Chromosome instability, microsatellite instability and epigenetic alterations [16]. The net result is activation of oncogenes, inactivation of tumor suppressor genes and deregulation of signaling pathways [11, 12]. Aberrant cell cycle regulation and changes in the expression of growth factors and cytokines regulate differentiation and survival of tumor cells. Mutations of cell-adhesion and angiogenic genes play important roles in the invasive and metastatic behavior of GC cells.

The aim of the current study was to identify the most differentially regulated genes in surgically resected gastric adenocarcinoma compared to matched normal mucosa, using whole genome cDNA microarray profiling. We also attempt to identify genes which influence GC prognosis and survival. The results are compared to the gastric epithelial cell gene response to H. pylori infection, which was analyzed in a previously published paper [17]. This study adds support to the significance of IL-8 and CLDN1 in gastric carcinogenesis, as well as demonstrates important genetic changes in GC and their possible relevance to H. pylori infection.

In this study we identified CLDN1 as one of the most consistently up-regulated genes in GC and a strong correlation between up-regulation of CLDN1 and reduced survival in 20 patients with gastric adenocarcinomas. This correlation is even stronger when adjusted for other parameters such as lymph node stage, tumor size and histological type. Our clinical sample size is small, but the results are consistent.

Claudins are proteins involved in cellular tight junctions and are important for the maintenance of normal epithelium, in particular barrier formation, cell polarity and signal transduction. Dysregulation of these genes have been identified in many different cancers. Based on tumor biology, down-regulation of CLDN1 would result in destruction of tight junctions and loss of cell-to-cell adhesion causing tumor progression [33], however the clinical significance in gastric carcinogenesis is more complex. There is evidence that several of the claudins, CLDN1 included, show increasing levels as gastric epithelium progresses to intestinal metaplasia and early gastric carcinoma [34]. CLDN1 might influence intracellular signalling, demonstrated by Liu et al. who showed that elevated expression of CLDN1 in breast cancer cells contributed to an anti-apoptotic effect through two mechanisms: inhibition of caspase-8 cleavage, and activation of the Wnt/β-catenin signal pathway [35]. CLDN1 has been identified within the nucleus of gastric cancer AGS cells in vitro, suggesting a regulatory role of CLDN1 on cell proliferation, migration and invasiveness at a nuclear level [36]. Some studies on ovarian and colon cancer report a role of CLDN1 on metastatic processes through activation of metalloproteinases, reducing apoptosis and increasing migration [36]. Although there are several papers to support an oncogenic role of CLDN1 in gastric cancer, two studies nevertheless showed reduced CLDN1 staining in metastatic compared to non-metastatic gastric cancer [33], and increased tumorgenicity in CLDN1 negative gastric epithelial cells [37], contrasting our findings.

The expression pattern of CLDN1 differs not only between different stages of carcinogenesis, but also between histological subtypes and between regions of the gastrointestinal tract. Resnick et al. demonstrated increased staining of CLDN1 protein in intestinal compared to diffuse type gastric cancer [38]. In contrast, Jung et al. demonstrated significantly lower CLDN1 expression in intestinal compared to diffuse type [33]. Neither studies found any correlation between CLDN1 and prognosis. Wu et al. demonstrated positive correlation between CLDN1 expression and invasiveness and metastasis in gastric tumors using immunohistochemistry [39]. In two studies on colorectal cancer, low expression of CLDN1 was a predictor of poor prognosis [40, 41], however an association between high CLDN1 expression and depth of tumor invasion was also noted [41]. In summary, the role of CLDN1 in cancer progression and prognosis is far from clear.

Our data demonstrate a marked increase in CLDN1 expression in 19 of 20 tumors compared to normal tissue, with a significant and independent relationship between high CLDN1 expressing tumors and reduced postoperative survival. We found no statistically significant difference between CLDN1 expression and histological subtypes, for that our sample number is insufficient. Nevertheless, tumors with high CLDN1 expression (FC > 2.14) showed an extremely poor prognosis as there were no patients alive at 450 days following curatively intended surgery in this group. In the low-expressing CLDN1 group (FC < 2.14), patients showed significantly longer post-operative survival, and 50% of the patients were still alive at the end of the study. We have reported total mortality and not cancer specific mortality. Our sample number is small, and a much larger study would be required to reveal statistically significant correlation between CLDN1 expression and sample subgroups, such as the histological subtypes, different tumor stages and H. pylori status. The role of CLDN1 in gastrointestinal cancer is controversial, but it seems convincing that the high CLDN1 gene expression conferred a very unfavorable prognosis in our study population. Moreover, CLDN1 was also one of the most significantly up-regulated genes in the previously studied H. pylori-exposed gastric epithelial cells [17], suggesting a possible causal relationship between chronic H. pylori exposure and CLDN1 up-regulation in gastric mucosa.

Other claudins that were up-regulated in the tumor samples were CLDN2 and CLDN7, and these genes might also be implicated in gastric carcinogenesis [36, 42–48]. CLDN7 was also among the most significantly up-regulated gene in the H. pylori exposed gastric epithelial cells [17], suggesting a role of this bactierium in the regulation of this gene.

The most consistently increased gene in the study was IL-8, up-regulated in the tumor in 18 of 20 tissue pairs. IL-8 is one of the major mediators of inflammation and a powerful chemokine that targets neutrophils and lymphocytes through the receptors CXCR1 and CXCR2. IL-8 is paramount in the acute inflammatory response to H. pylori infection and is also increased in chronic gastritis [49]. The increased IL-8 expression in the tumor samples may represent intratumoral inflammation as a normal reaction to an abnormal environment. However, the absence of other acute or chronic inflammatory genes suggests that the up-regulated IL-8 in the tumors can not be entirely explained by an inflammatory process alone. Hence, the role of IL-8 in the gastric cancer is not clear. First, persistent and chronic inflammation in the stomach is associated with an enhanced production of several pro-inflammatory cytokines including IL-8[50] which increases apoptosis, hyperproliferation and production of reactive oxygen and nitrogen species causing DNA damage and mutations. Second, increased vascularization is one of the hallmarks of malignant transformation, and IL-8 may serve an important role in this process. Several authors have demonstrated promotion of angiogenesis in tissue exposed to IL-8 protein [51–53]. A plausible causal role of IL-8 in the growth and vascularization of gastric cancer has also been shown in the work of Kitadai et al., where IL-8 transfected cells that were injected into the gastric wall of mice, rapidly produced growth of highly vascularized tumors [54]. Interestingly, we also found significantly increased and coordinated up-regulation of COL1A1 and COL1A2 in the tumor tissue, both of which are important in blood vessel development. Two authors recently found an association between IL-8 and adhesion, migration, and invasion in gastric cancer cells [55, 56]. Targeting of the IL-8 receptor CXCR2 has been suggested as a novel cancer treatment in several studies [53, 55–57].

In our previous study we identified IL-8 as the single most up-regulated gene in the acute response of gastric epithelial cells exposed to H. pylori in vitro[17]. IL-8 is also up-regulated in the pre-malignant stages of gastric cancer, such as chronic gastritis [58] and intestinal metaplasia [59]. With IL-8 being currently demonstrated as the single most up-regulated gene in surgically resected GC tumors, the up-regulation of this gene throughout gastric cancer progression may constitute an early and important event in the disease, initiated and maintained by H. pylori infection.

The causal relationship between H. pylori colonization of the stomach and GC has been widely accepted [60]. In this study only 2 of the 20 gastric tumors showed active H. pylori infection at the time of surgery, 1 intestinal and 1 diffuse type cancer. While more than 90% of the GC reported in Asian countries are considered attributable to this bacterium [61], our findings might indicate a lesser role of H. pylori in gastric cancer in a country like Norway. However, GC tissue is frequently H. pylori negative, due to the mucosal atrophy caused by the bacteria itself. While H. pylori colonization drives forward the progression of mucosal atrophy and intestinal metaplasia, this process paradoxically also slowly eradicates the same bacteria from the gastric mucosa which causes a decrease in active inflammation [62].

In the current study, we demonstrated up-regulation of several matrix metalloproteinases (MMPs); and KRT17 in the tumor of almost all the tissue pairs. MMPs participate in the degradation of extracellular matrix and the regulation of tumor growth and angiogenesis, and are important in the detachment of malignant cells from adjacent tissue to attain metastatic ability. Keratin 17 has been shown to be over-expressed in several adenocarcinomas including GC, and an association with aggressive tumor behaviour, local invasion, metastasis as well as treatment responsiveness has also been suggested [63–65]. We found no assiciation between MMPS and KRT17 and clinicopathological parameters. However, other studies have demonstrated an association with MMPs and advanced tumor stage, high grade tumors and metastasis, as well as a role for MMP11 as a serum marker in GC disease [66–69]. MMP7 and KRT17 were also among the most up-regulated genes our previous study of gastric epithelial cells exposed to H. pylori[17], which raises the possibility of a role of H. pylori also in the regulation of these genes.

In hiarchical clustering of the samples, the tumor and the control tissues clustered separately. Normal mucosa tissue bore a closer biological resemblance to normal mucosa from the other individuals, than to the tumor counterpart from its own stomach. In similar fashion, most tumor samples clustered together, demonstrating common genetic features between the tumor specimens. In addition, there was a large horizontal distance between the tumor specimen and the normal specimen within each tissue pair, illustrating that a significant shift in gene expression has occurred during the progression from normal mucosa to cancer within the same stomach. Furthermore, the mixed type and the intestinal type of cancers formed two almost exclusive clusters compared to the diffuse type cancers, indicating that each of the histological types have distinct gene expression profiles. Surprisingly, the mixed type cluster showed the greatest difference in gene expression compared to the normal tissue, indicating that the biology of this type of GC has removed itself the most from the original naive tissue. The results are interesting, but the sample number is too small to draw any conclusions.

We performed GO analyses to cluster the most differentially regulated genes according to biological function. Several terms relating to cell adhesion were enriched by genes up-regulated in the tumor samples. KEGG pathway analysis also showed that a number of pathways regulating cell attachment were significantly affected, in agreement with the GO analysis. Disruption of primary cell attachments, and secondly, cell adhesion to distant sites, are two fundamental steps in the ability of tumor cells to develop metastatic disease. Blood vessel development is essential for the tumor’s ability to survive and metastasize to distant sites. This gain of abilities seems to come at the cost of loss of features of differentiated intestinal tissue, such as digestive and excretive processes, which were associated with down-regulated genes. While several of the highly up-regulated genes were mapped to important pathways, none of the 100 down-regulated genes were involved in any significant pathways, indicating that it is the net gain of oncological function that translates into tumor growth and malignant behavior, rather than the loss of tumor suppression.

In our study, CLDN1 was not only highly up-regulated in the gastric cancer samples, but CLDN1 expression was also independently associated with a very poor post-operative prognosis. IL-8 was the single most up-regulated gene in the study, and we have shed new light on the role of both genes in gastric carcinogenesis. IL-8 and CLDN1 may represent important links between GC and the gene response seen in acute H. pylori infection of gastric epithelial cells. Intestinal, diffuse and mixed type of gastric cancer formed separate gene expression clusters, but also showed many genetic similarities. Functional analysis demonstrated that several cellular pathways regulating cell attachment were affected in the tumor tissue.