Objective: Autophagy is the catabolic process where the components of eukaryotes experience damage, and the affected or superfluous components undergo self-degradation. However autophagy can promote cancer cell apoptosis or facilitate cell growth. This work aimed to investigat the significance of autophagy-related genes (ARGs) in predicting the prognosis of breast cancer (BC) intervened with Cremastra.
Methods: Active ingredients and action targets were obtained using the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) and SwissTargetPrediction. Then, the BC transcriptome and clinical data were downloaded in The Cancer Genome Atlas (TCGA), whereas ARGs were collected in the Human Autophagy Database (HADb). Meanwhile, Perl and R software were used for data processing and analysis. Firstly, the transcriptome data of BC were mapped to ARGs to screen the BC-ARGs. Secondly, the above genes were mapped to the action targets of Cremastra, ARGs of Cremastra-intervened BC were then screened out. Moreover, an enrichment analysis of biological function was carried out. Univariate Cox regression was carried out on ARGs of BC for preliminarily selecting the independent prognostic genes and constructing the autophagy prognosis model. These genes were mapped to ARGs involved in Cremastra-intervened BC. Finally, those mapped genes were optimized by multi-factor Cox regression, and the key ARGs and potential compounds were obtained. Finally, all cases were classified as low- or high-risk group based on the median risk score. Receiver operating characteristic (ROC) curve, Kaplan-Meier (K-M) survival, independent prognosis and clinical correlation analyses were conducted for model evaluation and identification of factors to independently predict prognosis.
Results: Altogether, 66 active components and 38 targets of the Cremastra-intervened autophagy of BC were screened and the autophagy prognosis model demonstrate good predictive performance. As suggested by the survival curve, low-risk patients had a markedly increased survival rate compared with high-risk patients (P < .01). Besides, the gene expression levels of the high-risk group increased with the increases in patients' risk scores. Upon univariate regression, 34 differentially expressed ARGs related to BC treatment were screened. Multivariate regression identified 4 key ARGs, which were mainly derived from glycosides, lignans, flavonoids, and dibenzyl compounds. Thereafter, key genes were subjected to correlation analysis between clinicopathological features and prognosis, among which BCL2 and TP63, showed independent prognostic value.
Conclusions: In this study, an autophagy prognosis model was established, and BCL2 and TP63 were predicted for the Cremastra intervention of BC by Bioinformatics, which will be applied to further work.