Gastric cancer (GC) remains a major health problem worldwide, representing the third most common cause of death from all cancers [
]. Mortality rates are higher in Asian and Latin American countries, where cases are usually diagnosed at later stages, leading to very low survival rates. The infection of the gastric mucosa by
represents the major risk factor in over 65% of all distal GC, and recent evidence suggests that it might also play a role in proximal GC [
has the ability to colonize the human stomach and persist for decades, eliciting a chronic long-lasting inflammatory response that varies in magnitude depending on the genetic background of the host, the virulence of the
strain, and environmental factors [
]. The differences in the gastric inflammatory response between hosts may partially explain the different outcomes observed in
infected patients [
]. Understanding the natural history of the infection may help to identify potential biomarkers for the early diagnosis of GC to allow timely treatment to reduce mortality.
The interaction of
with the gastric epithelium induces the production of interleukin (IL)-8, IL-6, and IL-1β. These cytokines are chemotactic for neutrophils and mononuclear cells, and their production leads to a proliferative response with a dense infiltrate of neutrophils and macrophages in the gastric mucosa, resulting in a chronic active gastritis [
also induces gastric mucosal infiltration by dendritic cells and T and B cells, and stimulates secretion of macrophage chemotactic protein (MCP)-1, tumor necrosis factor (TNF)-α, IL-12, IL-10, transforming growth factor (TGF)-β, and interferon (IFN)-γ [
]. Among other deleterious effects that predispose cells to oncogenic transformation, the inflammatory mediators produced by this decades-long gastritis may cause DNA damage, induce proliferation, and inhibit apoptosis [
]. In addition, inflammatory cytokines and chemokines increase the expression of molecular factors such as hypoxia-inducible factor-1, vascular endothelial growth factor, L-selectin, cyclooxygenase-2, and matrix metalloproteinase together with other molecules that contribute to carcinogenesis [
]. Progress of GC can be evaluated using the tumor–node–metastasis (TNM) staging classification of malignant tumors, which is an important prognostic factor for GC [
]. Ideally, biomarkers should detect the early stages (I/II) of disease when opportunities for cure are highest.
The inflammatory mediators produced locally in the gastric mucosa may reach the blood circulation and be detected in plasma samples. In this study, we tested the hypothesis that circulating levels of inflammatory cytokines and chemokines could function as indirect indicators of tissue damage, and that their measurement might be a useful biomarker for the early detection of GC, resulting in a better long-term prognosis.
We analyzed the systemic levels of six cytokines and two chemokines associated with chronic gastric inflammation during
H. pylori infection for potential biomarkers to identify patients with GC. We found that increased levels of IL-1β, IL-6, IFN-γ, and IL-10 and lower levels of MCP-1 significantly differentiated patients with GC from healthy controls. We also classified the GC cases according to their TNM stage and found that higher levels of IL-1β, IFN-γ, IL-10, and IL-8, and lower levels of MCP-1 differentiated late stage IV GC from healthy controls, and importantly, that IFN-γ and IL-10 differentiated patients with early stage (I/II) GC from healthy controls. These results suggest that circulating levels of cytokines may help to identify patients in the early stages of GC, offering the possibility of early detection for timely treatment.
The observed differences of these cytokines between GC patients and healthy controls are consistent with reports in both animal models and human infection. For example, IL-1β is a proinflammatory cytokine required for the efficient control of
]. IL-1R (−/−) mice failed to develop protective immunity but were protected against
-associated gastritis and gastric preneoplasia because of their inability to generate
-specific Th1 and Th17 responses [
]. In addition, the overexpression of IL-1β in the stomach of mice led to spontaneous gastric inflammation and cancer, even in the absence of
], and IL-1β is a potent inhibitor of gastric acid secretion [
], which may also favor the appearance of preneoplastic lesions because of hypochloridia.
The increased levels of IL-6 in GC patients and their strong association with the risk of GC that was identified in this study indicate that IL-6 plays an important role in GC, which is consistent with previous reports in animal models and in patients [
]. IL-6 can upregulate DNA methyltransferases, resulting in modification of the methylation status of genes associated with tumor suppression [
]. In a mouse model of colitis-associated cancer, IL-6 stimulated the survival and proliferation of premalignant intestinal epithelial cells, mainly because of increased signal transducer and activator of transcription (STAT)3 signaling [
]. It has also been observed that IL-6 negatively affects the expression of genes associated with tumor suppression and anchorage-dependence growth in colonic epithelial cells in vitro [
]. Furthermore, it was reported that in patients with GC, expression of IL-6 and STAT3 was increased in GC tumor tissue and that the level was associated with the TNM stage of GC [
The higher level of IFN-γ in sera of GC patients compared with controls in this study could be indicative of its role in establishing a proinflammatory microenvironment in the gastric tissue, although its role in the promotion of GC is still controversial. IFN-γ is upregulated in the gastric mucosa after chronic
], and besides its role in the responses to bacterial infection, it has also an important tumor suppressor activity [
]. Previous studies have shown that specific T cell responses play a critical role in inducing gastric mucosal inflammation [
], and suggested that IFN-γ may exacerbate gastric inflammation and favor progression to GC. However, more recent studies in a mouse model found that overexpression of IFN-γ inhibited gastric carcinogenesis induced by IL-1β and/or
infection by suppressing putative gastric progenitor cell expansion and by reducing epithelial cell apoptosis via induction of an autophagy program [
]. Thus, IFN-γ is a pleiotropic mediator with both pro- and antitumorigenic activities. IL-10 is also a potent pleiotropic cytokine that has the dual ability to suppress or stimulate anticancer activity [
]. The high amount of IL-10 found in GC patients in this study would suggest that it is acting by suppressing anticancer responses. In fact, when IL-10 is elevated in blood during advanced GC, it leads to the inability to eliminate tumor cells [
]. GC cells themselves can also secrete IL-10, which may explain its reduction in GC patients after surgical removal of their tumor [
In contrast to the above cytokines, we found significantly decreased levels of MCP-1 in the sera of patients with GC. This chemokine plays an important role in the progress of
-related gastric diseases [
], and patients infected with
have significantly higher expression of
mRNA in the gastric mucosa than patients without
]. MCP-1 is a CC chemokine that is produced by gastric epithelial cells; it plays a major role in regulating migration of monocytes and lymphocytes into tissues [
] and may also promote angiogenesis and recruitment of tumor-associated macrophages [
]. There are few studies reporting the circulating levels of MCP-1 in patients with GC, although one study reported significantly lower levels in cancer patients than in controls [
], which is consistent with our findings. Furthermore, the authors of that study reported that the concentration of MCP-1 in the serum of GC patients decreased in conjunction with disease progression and suggested that reduced plasma levels reflect the local consumption of MCP-1 by GC tissue [
We found no significant differences in the levels of circulating IL-8, TNF-α, and TGF-β between GC patients and healthy controls. The lack of significant differences in IL-8 contrasts with previous studies where circulating IL-8 levels were found to be increased significantly in patients who developed GC [
]. Although we found no overall difference in IL-8 levels between the GC and healthy control groups, in GC patients there was a significantly higher frequency of IL-8 values in the upper quartile compared with controls. In addition, patients with late stage (IV) GC had significantly higher levels of circulating IL-8. These results are consistent with studies showing that IL-8 is also produced by cancer cells and may promote angiogenesis, tumor growth, tissue invasion, and metastatic spread [
], and that high IL-8 expression directly correlates with a poor prognosis in GC [
]. TNF-α is one of the main mediators of inflammation and has been linked to several steps involved in carcinogenesis, including cellular transformation, survival, proliferation, invasion, angiogenesis, and metastasis [
]. However, similar to previous reports [
], we found no differences in TNF-α levels between GC patients and healthy controls. Although TGF-β can also promote tumor growth, invasion, and metastasis, it has pleiotropic activity and functions as a tumor suppressor during the early stages of GC, although it may promote tumor growth and metastasis during late stage GC [
We then analyzed the differences in cytokines and chemokines between the two types of GC (DGC and IGC) and found that IFN-γ and IL-10 were increased significantly in both types relative to controls. In contrast, IL-1β, IL-6, and TGF-β significantly differentiate IGC but not DGC, whereas MCP-1 was significantly lower in DGC but not in IGC. These results suggest that the pattern of cytokine production associated with GC may differentiate DGC from IGC. The lack of studies analyzing circulating cytokines in the different GC types precludes any comparison, and identifies a need for additional studies to validate our results in larger groups of patients and in different populations.
To evaluate further the utility of these markers for differentiating patients with GC from controls, we next determined the ORs and confirmed that increased levels of IL-1β, IL-6, IFN-γ, and IL-10 were all associated with increased risk of GC, noting that high levels of IL-6 increased the risk of current GC 17.9 times. The analyses also showed no increase in risk associated with MCP-1, IL-8, and TNF-α, whereas there was a trend to an association of diminished TGF-β with a decreased risk of GC. This analysis confirmed the possible value of IL-1β, IL-6, IFN-γ, and IL-10 for differentiating patients with GC in the population studied and suggested that TGF-β may also help in the identification of GC risk. We then tested the utility of these markers as a diagnostic test and found that only IL-6, IFN-γ, and IL-10 had a useful specificity (0.97, 0.53, and 0.82 respectively) and PPV (0.90, 0.64, and 0.67, respectively), but with a low sensitivity (0.39, 0.71, and 0.48, respectively). The other cytokines evaluated had poor value as a diagnostic test. Although a number of studies report the association of inflammation markers with GC, none reports the value of circulating levels of markers as a diagnostic test for GC. Such an analysis is needed for better evaluation of the utility of candidate biomarkers in GC.
In the evaluation of possible biomarkers, there is a need for consensus about the type of analyses that are required to define their utility for the timely diagnosis of patients with GC. Some authors limit their analysis to a description of the differences in candidate markers between GC and controls and report
p-values, whereas others go further and report ORs; however, few studies test the utility of markers as a diagnostic test and report specificity, sensitivity, and predictive values. In this work, we present sequential analyses of the differences, the ORs, and diagnostic utility of circulating levels of cytokines and chemokines in patients with GC to compare the relevance of the different analytical approaches.
A consistent finding for all cytokines and chemokines tested in this study was the high level of variation between individuals in circulating cytokine levels, which, importantly, limits their possible value as candidate GC biomarkers. Another limitation of our study was that the group with GC had an average age almost twice as high as that of the healthy blood-donor control group; and although the multivariate logistic regression analyses adjusted for age confirmed association with GC, results need to be confirmed with control groups adjusted by age and sex. It is also true that despite the investment of hundreds of millions of dollars over the last 10 years in studies of cancer, none of the analyzed biomarkers has yet been approved for clinical use [
]. However, in view of the urgent need for biomarkers in GC and the lack of validated candidates, efforts to identify possible markers are badly needed, particularly in regions like Latin America that have the highest GC mortality rates. Thus, despite their poor sensitivity, our results for IL-6, IFN-γ, and IL-10 might offer a limited but still valuable test with reasonable specificity and PPV to identify patients with GC in high-risk regions.