Correlation of autophagy-related genes for predicting clinical prognosis in colorectal cancer

Abstract

Aim: Autophagy plays a controversial role in cancer. The role of autophagy-related genes (ARGs) in colorectal cancer (CRC) was evaluated based on publicly available data from The Cancer Genome Atlas and the Human Autophagy Database. Materials & methods: After collecting CRC-related transcript and clinical data and a list of ARGs from public databases, the Wilcoxon test was used to identify the differentially expressed ARGs between CRC and paired normal tissues. Gene Ontology and Kyoto Encyclopedia of Genes and Genomes enrichment analyses were used to identify the major biological properties and pathways associated with these genes. Univariate Cox regression was used to identify the prognosis-associated ARGs, and a forest plot was used to visualize the results. Kaplan–Meier analysis of the 5-year survival rate was performed. Univariate and multivariate Cox analyses were used to verify the impact of the prognosis-associated ARGs. Results: A total of 36 differentially expressed genes (16 upregulated and 20 downregulated in CRC) were obtained from among 206 ARGs. There were 53 enriched pathways, including the p53 signaling pathway, platinum drug resistance, apoptosis, EGFR tyrosine kinase inhibitor resistance and ErbB signaling pathway (p- and q-values <0.05). Kaplan–Meier analysis showed that the 5-year survival rate was 46.0% (95% CI: 0.335–0.631) and 76.0% (95% CI: 0.651–0.886) in the high- and low-risk groups, respectively. The high-risk patients had worse survival probability (p = 6.256 × 10-5). Independent-samples t-tests revealed that MAP1LC3C expression was higher in patients aged ≤65 than >65 (p = 0.022); RAB7A expression was higher in patients aged ≤65 than >65 (p = 7.31 × 10-4), higher in M1 than M0 (p = 0.042), higher in N1–3 than N0 (p = 0.002) and higher in stage III and IV than I and II (p = 0.042); risk score was higher in N1–3 than N0 (p = 0.001) and in stage III and IV than I and II (p = 0.002); and WIPI2 expression was higher in M1 than M0 (p = 0.002), higher in N1–3 than N0 (p = 2.059 × 10-7) and higher in stage III and IV than I and II (p = 2.299 × 10-7). There were no differences in risk score between males and females (p = 0.593), T1–2 and T3–4 (p = 0.082) or M0 and M1 (p = 0.072). Univariate and multivariate Cox analyses showed that RAB7A was a lower-risk gene, while MAP1LC3C, WIPI2, DAPK1, ULK3 and PELP1 were high-risk genes. Conclusion: Certain ARGs are potential prognostic molecular markers of poor prognosis in CRC. Additionally, the p53 signaling pathway, platinum drug resistance, apoptosis, EGFR tyrosine kinase inhibitor resistance and ErbB signaling pathway may be critical pathways regulated by ARGs in CRC.