1571P - Exploration of immune and metabolism gene signature for prognosis of esophageal carcinoma and establishment of a combined prediction model

Background

Esophageal carcinoma (EC) a high rank common and death cancer. The intricate tumor microenvironment of EC necessitates a better performance signature to distinguish different tumor prognoses, which may facilitate enable appropriate treatment selection for patients. This is the first study aimed to identify the immune and metabolism gene signature for prognosis prediction of EC and constructing a prediction model, from EC transcriptional data.

Methods

Using differential expression analysis to compare the transcriptomic data associated with immune and metabolism from TCGA dataset between 10 normal population and 162 EC patient samples. Screen the differentially expressed genes (DEGs) related to prognosis by univariate Cox regression, the subtype clusters were identified, and the characteristics were analyzed. Through multivariate Cox and LASSO regression, a prediction model of the EC prognostic risk score (RS) was established and validated.

Results

Compared with the normal population, there were 568 immune-related genes of differential expression in EC, with 369 up-regulated and 199 down-regulated; and 810 metabolism-related genes, including 428 up-regulated and 382 down-regulated. Based on these immune and metabolic DEGs, 12 prognosis-related genes identified that could be clustered patients into three subtypes. The prognosis of the three subtypes was significantly different (P= 0.0016). Furthermore, the prediction model of prognostic RS was constructed based on six genes included STC2, APLN, GPER1, FMO1, SNRPB, and FABP3 from above DEGs. The patients with low RS had significantly higher OS (P < 0.001) than those in the high RS group, and it had a certain predictive accuracy (AUC of 5 years OS = 0.818). Those were verified in the GSE53625 cohort.

Conclusions

Immune and metabolism gene were significantly correlated with the prognosis of EC patients, and the prognostic model by six DEGs had good prediction efficiency. These were expected to help clinical diagnosis and therapy, and provides a new perspective to explore the molecular mechanisms of EC.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.