Abstract
A positive family history is a strong and consistently reported risk factor for gastric cancer (GC). So far, it has been demonstrated that serum pepsinogens (PGs), and gastrin 17 (G17) are useful for screening individuals at elevated risk to develop atrophic gastritis but they are suboptimal biomarkers to screen individuals for GC. The main purpose of this study was to investigate serum metabolomic profiles to find additional biomarkers that could be integrated with serum PGs and G17 to improve the diagnosis of GC and the selection of first-degree relatives (FDR) at higher risk of GC development. Serum metabolomic profiles included 188 serum metabolites, covering amino acids, biogenic amines, acylcarnitines, phosphatidylcholines, sphingomyelins and hexoses. Serum metabolomic profiles were performed with tandem mass spectrometry using the Biocrates AbsoluteIDQ p180 kit. The initial cohort (training set) consisted of n = 49 GC patients and n = 37 FDR. Differential metabolomic signatures among the two groups were investigated by univariate and multivariate partial least square differential analysis. The most significant metabolites were further selected and validated in an independent group of n = 22 GC patients and n = 17 FDR (validation set). Receiver operating characteristic (ROC) curves were used to evaluate the diagnostic power and the optimal cut-off for each of the discriminant markers. Multivariate analysis was applied to associate the selected serum metabolites, PGs, G17 and risk factors such as age, gender and Helicobacter pylori (H. pylori) infection with the GC and FDR has been performed and an integrative risk prediction algorithm was developed. In the training set, 40 metabolites mainly belonging to phospholipids and acylcarnitines classes were differentially expressed between GC and FDR. Out of these 40 metabolites, 9 were further confirmed in the validation set. Compared with FDR, GC patients were characterized by lower levels of hydroxylated sphingomyelins (SM(OH)22:1, SM(OH)22:2, SM(OH)24:1) and phosphatidylcholines (PC ae 40:1, PC ae 42:2, PC ae 42:3) and by higher levels of acylcarnitines derivatives (C2, C16, C18:1). The specificity and sensitivity of the integrative risk prediction analysis of metabolites for GC was 73.47% and 83.78% respectively with an area under the curve of the ROC curve of 0.811 that improves to 0.90 when metabolites were integrated with the serum PGs. The predictive risk algorithm composed of the C16, SM(OH)22:1 and PG-II serum levels according to the age of individuals, could be used to stratify FDR at high risk of GC development, and then this can be addressed with diagnostic gastroscopy.