Autophagy is a highly conserved evolutionary process in eukaryotic cells, which is involved in a series of cell homeostasis processes. There are three types of autophagy: Macroautophagy, microautophagy and chaperone mediated autophagy. Macroautophagy is the only autophagy that can degrade organelles, which we usually call autophagy. Autophagy related genes LC3, Beclin 1 and ATG5 are all biomarkers of autophagy, which are involved in autophagy regulation [16-18].
Many studies have shown that autophagy protein is closely related to the prognosis of GC patients. The high expression of ATG5 is closely related to the poor prognosis and drug resistance of gastric cancer [19]. It has been found that the disease-free survival rate and the overall survival rate of patients in the Beclin 1 high expression group are significantly increased [20]. However, some studies have come to the opposite conclusion [21]. Considering the importance of autophagy in gastric cancer, we can reasonably speculate that autophagy related genes have broad prospects in the prognosis evaluation of gastric cancer, and the multi gene signature generated by various algorithms will be better than a single molecule in the prediction of GC OS.
In this study, we analyzed the mRNA expression of 204 ATGs in the TCGA gastric cancer dataset. Single factor Cox regression analysis showed that 10 genes were related to the survival of STAD. We used LASSO regression to develop eight prognostic markers for the TCGA-STAD cohort. Finally, the signature of four genes was established by multivariate Cox regression. The risk score of each patient can be obtained by calculating the mRNA expression level and risk coefficient of the selected gene. In the TCGA-STAD cohort, risk scores significantly stratified patient outcomes. More importantly, in two independent geo gastric cancer datasets within the STAD, the prognostic power of the 4-gene signature was verified. Gene signature is often applied to forcast the prognosis of a variety of tumors in the past few years[22], which is even better than TNM staging and histopathological diagnosis in some extent [23]. Gene signatures based on ATGs have been reported in a variety of cancers, such as serous ovarian cancer, breast cancer, colon cancer and glioma [24-27]. For example, Liu and colleagues recently reported a 14 autophagy related signature based on relapse free survival in patients with non-small cell lung cancer [28].
In previous experiments, it has been shown that the genes contained in the signature are all related to cancer. The GluD2 protein encoded by GRID2 is a member of the ionic glutamate receptor family that mediates excitatory synaptic transmission [29]. Research by Ngollo et al. Showed that GRID2 interacts with H3K27me3 in prostate cancer and is significantly overexpressed. [30] The protein encoded by ATG4D belongs to the ATG4 mammalian family (four cysteine proteases, ATG4A4D class), which is closely related to autophagosome maturation and apoptosis pathways [31]. Gil et al's research shows that ATG4D can play a role as a tumor suppressor in the development of colorectal cancer [32]. Similarly, ATG4D's defective expression will eliminate autophagy and promote the growth of human uterine fibroids [33]. The members of the GABARAP (gamma-aminobutyric acid hand-related protein) family (GABARAP, GABARAPL1 / GEC1 and GABARAPL2 / GATE-16) are one of the subfamilies of the ATG8 family of proteins, and are closely related to the intracellular transport of receptors and the autophagy pathway [34]. GABARAP is described as being down-regulated in cancer, and high expression is associated with a good prognosis [35]. The study of Y et al. Showed that GABARAP is overexpressed in colorectal cancer, which is related to the shortened survival time of patients, which shows the prognostic significance of GABARAP [36]. Moreover, the expression of ATG4 isoforms such as ATG4D can regulate post-translationally activated LC3 / GABARAP family proteins [32], which further verifies the internal connection of genes contained in gene signatures. CXC type 4 chemokine receptor (CXCR4), also known as fusion protein (Fusin) or CD184, plays a role in cell proliferation and migration of cells [37]. Yu et al. Found that miR-125b induced by CXCL12 / CXCR4 axis promoted invasion and conferred 5-fluorouracil resistance in colorectal cancer by enhancing autophagy [38]. The CXCR4 / mTOR signaling pathway is also thought to play a role in promoting migration and inducing autophagic cell death in the peritoneal diffusion of gastric cancer cells [39].
Bioinformatics enrichment analysis showed that 28 differentially expressed autophagy related genes(DE-ATGs) were mainly related to cell growth, positive regulation of cell protein localization, neuron death, regulation of cell growth, platinum drug resistance, apoptosis and p53 signaling pathway in STAD. Interestingly, Huang's study found that autophagy plays a vital role in the platinum drug resistance of tumor cells [40]. In tumor treatment, apoptosis tolerance is an important mechanism for tumor drug resistance. Autophagy can prevent apoptosis induced by antitumor drugs and promote tumor drug resistance. However, autophagy cell death may be a death mode of apoptosis tolerant tumor cells, Autophagy has double effects on drug resistance of tumor cells [41]. There is also a lot of evidence implying the interaction between autophagy and apoptosis [42]. Autophagy may promote or hinder apoptosis.
Autophagy inhibited apoptosis when the environmental conditions were less affected. However, when autophagy causes excessive consumption of intracellular proteins and organelles, resulting in the inability of cells to survive, the cells will turn into apoptosis. In some cases, autophagy can also cause cell death. It is worth mentioning that autophagy and apoptosis involve many apoptosis related proteins, such as p53 and BH3 only proteins [32]. In the early stage of cancer cell formation, autophagy can inhibit the formation of cancer cells; after cancer cell formation, cancer cells use autophagy to promote the survival of cancer cells and inhibit cell apoptosis, which may lead to the resistance of cancer cells to chemotherapy drugs. Therefore, if we inhibit autophagy during chemotherapy, it will be beneficial to enhance the therapeutic effect.
Finally, we developed a nomogram to predict individual clinical outcomes including these four autophagy-related gene signature, age, gender, grade, stage, TNM staging and risk score to construct a nomogram to predict the 3-year and 5-year survival of gastric cancer patients. Consistently, the calibration chart shows that the signature can more accurately assess the survival of gastric cancer patients. However, due to the lack of sufficient cases, we were unable to evaluate the predictive power of autophagy gene signatures in other independent gastric cancer data sets. In addition, other potential prognostic variables related to OS in GC, such as neutrophil-to-lymphocyte ratio (NLR), should also be studied. In addition, the expression and prognostic role of these four genes in gastric tissue need further study.
It should be admitted that our research inevitably has some limitations. First, our work is retrospective, not prospective; in addition, some other key clinical pathological features, such as the number of lymph nodes, are not included in the nomogram. Finally, the mechanism and interrelation of autophagy-related genes contained in gene signatures need further study.