2.1 Identification of the differentially Expressed ATGs in Chemotherapy Group and Non-tumor samples
Data from 407 subjects in the STAD cohort in TCGA database were analyzed. 232 ATGs were obtained in our study. A total of 221 ATGs were expressed in TCGA-STAD cohort. The results were 157 patients who received chemotherapy and 32 normal samples. The basic clinical characteristics of these patients in the TCGA database was also compared, as shoen in Table 1. With FDR < 0.05 and |log2 FC| >1 as the screening criteria, 24 ATGs were presented(Figures 1A,B). The upregulated ATGs were IFNG, ATIC, BIRC5, CASP8, VMP1, IL24, CDKN2A, HSP90AB1, VEGFA, CTSB and ERBB2. The downregulated ATGs include: PRKN, CDKN1A, GRID2, HSPB8, NRG3, NRG2, FOS and NKX2-3.
Table 1
Clinical characteristics of GC patients with chemotherapy in TCGA cohort
Gene | Co-ef | HR | HR.95L | HR.95H |
GABARAPL1 | 0.370661 | 1.448692 | 0.912786 | 2.299233 |
GRID2 | 2.358799 | 10.57824 | 0.898029 | 124.6053 |
CXCR4 | 0.302963 | 1.353864 | 1.034964 | 1.771025 |
NCKAP1 | 0.71455 | 2.043268 | 0.967303 | 4.316067 |
ITGA3 | 0.269185 | 1.308897 | 0.971892 | 1.762759 |
GABARAPL2 | 1.334027 | 3.796301 | 1.55472 | 9.26977 |
IRGM | 2.963281 | 19.36138 | 1.362477 | 275.1335 |
BNIP3L | 0.592749 | 1.808954 | 1.091792 | 2.997195 |
ERBB2 | 0.319098 | 1.375887 | 1.105664 | 1.712152 |
2.2 Enrichment Of Atgs
We utilized some techniques to analyze and explore the possible signaling pathways in GC that may be associated with chemotherapy response. Based on GO analysis, the differences in the cellular morphology, neuron death, AGT regulation on cellular membranous surfaces, autophagy and other aspects were studied(Fig. 2A). In the KEGG pathways, ATGs were elucidated with regard to different ailments and pathways(Fig. 2B).
2.3 The construction of Prognostic Markers of ATGs for OS in TCGA GC Chemotherapy Group
221 ATGs were analyzed by some analytical methods. In TCGA-STAD cohort, 13 ATGs had prognostic measures of chemotherapy patients(Fig. 3). 9 ATGs were finally tabulated and pinpointed to Table 2.
Table 2
Multivariate Cox regression analysis of prognostic genes.
Characteristic | Variables | Total | Percentage (%) |
Age | <=65 | 79 | 53.7 |
| > 65 | 67 | 46.3 |
Sex | Male | 92 | 62.6 |
| Female | 55 | 37.4 |
Grade | G1-2 | 49 | 33.3 |
| G3 | 93 | 63.3 |
| GX | 4 | 3.4 |
Stage | I | 10 | 6.8 |
| II | 46 | 31.3 |
| III | 73 | 49.7 |
| IV | 17 | 11.6 |
T stage | T1 | 4 | 2.7 |
| T2 | 29 | 19.7 |
| T3 | 73 | 49.7 |
| T4 | 42 | 27.9 |
N stage | N0 | 28 | 19.0 |
| N1 | 49 | 33.3 |
| N2 | 31 | 21.1 |
| N3 | 38 | 26.6 |
M stage | M0 | 130 | 88.4 |
| M1 | 10 | 6.8 |
| Mx | 7 | 4.8 |
2.4 ATGs and the OS of GC victims in chemotherapeutic group
Risk scores were calculated based on ATGs related mRNA expression levels and risk factors. Patients were classified into related groups. The product limit estimation analysis tool was utilized for data representation. Five-year survival rates were analyzed(Fig. 4A). ROC curves were drawn and plotted to determine the ability of patients in chemotherapy group to predict ATGs (Fig. 4B). The area under the curve was well interpreted. Genetic study, which was well pointed out during the study progression, (Fig. 4C), increases in the number of deaths(Fig. 4D). Heatmaps were created for both groups(Fig. 4E). These results suggested that risk scores accurately reflected patient survival.
To determine whether autophagy-related scoring features were independent prognostic factors in GC patients undergoing chemotherapy, we conducted a study. Similarly, the significant correlation between risk scores and clinical variables was achieved by the utilization of hazard ratio technique sketch diagrams(Fig. 5A). In Cox regression analysis, several Cox regressions factors affecting the prognosis of chemotherapeutic patients with gastric cancer were well plotted(Fig. 5B). Furthermore, the comparison results of the two groups were plotted(Fig. 5C). Cancer pathways were enriched, suggesting that autophagy is involved in the regulation of chemotherapy for high-risk gastric cancer patients.
2.5 Atg’s Progression In Gastric Cancer
The direct effects of ATGs and their correlation with gastric cancer progression, OS,genes and clinicopathological variables was evaluated. Figure 6 showed that BNIP3L, CXCR4, ERBB2, GABRAPL, ITGA3 and NCKAP1 significantly correlated with the pathological classification of GC.On the one hand,BNIP3L, CXCR4, ERBB2, GABRAPL and NCKAP1 were significantly correlated with Lauren typing. ERBB2 and GABRAPL were also significantly correlated with tumor grade. On the other hand, BNIP3L, ERBB2, ITGA3 and NCKAP1 were significantly correlated with TNM staging.
2.6 Prognostic ATGs for DFS of GC Patients in the Chemotherapy Group
Data have been obtained on certain types of biomarkers in GC patients undergoing chemotherapy. GSE26253 dataset was incorporated. According to univariate Cox regression analysis, there was a certain significant correlation among the 9 ATGs(Fig. 7A). 7 ATGs were well obtained and a division was well established in the victims. Kaplan-Meier analysis (product limit estimator) revealed that, P < 0.001(Fig. 7B). Heatmaps were developed for both groups, (Fig. 7C). Results about the chemotherapy of GC patients were summarized.