Identification of potential serum biomarkers for gastric cancer by a novel computational method, multiple normal tissues corrected differential analysis

Background: Genes specifically expressed in one or a few tissues and upregulated in tumors are potentially good serum biomarkers. Methods: By applying a recently developed computational method, called multiple normal tissues corrected differential analysis (MNTDA), we identified genes that are likely to be upregulated in the blood of gastric cancer patients as compared to normal controls. Results: We identified four genes (MMP-1, MMP-3, MMP-12, and CXCL5) as potential serum biomarkers for gastric cancer. Of these four genes, only MMP-1 was significantly upregulated in the sera of 40 gastric cancer patients, as compared to 40 control sera. The same pattern was observed in the second cohort of 80 gastric cancer patients and 80 controls. In a combined analysis, the level of serum MMP-1 in gastric cancer patients was significantly higher than the level in control samples (P<0.0001). The use of MMP-1 was 62.5% sensitive and 62.5% specific in detecting gastric cancer patients. Patients with high serum levels of MMP-1 had a significantly worse outcome than patients with low serum MMP-1 levels. Finally, we determined that preoperative serum MMP-1 levels were prognostic, independent of tumor stage. Conclusions: MMP-1 is a potential prognostic marker for gastric cancer patients after gastrectomy. (C) 2011 Elsevier B.V. All rights reserved.