UC Irvine

Purpose

Colon cancer (CC) is a serious disease burden. The prognosis of patients with CC is different, so looking for effective biomarkers to predict prognosis is vitally important. Ferroptosis is a promising therapeutic and diagnosis strategy in CC. However, the role of ferroptosis in prognosis of CC has not been studied. The aim of the study is to build a prognosis model related ferroptosis, and provide clues for further therapy of CC.

Methods

The RNA-seq data were from TCGA (training group) and GEO (testing group). The R language and Perl language were used to process and analyze data. LASSO regression analysis was used to build the prognosis model. ssGSEA was used to compare the immune status between two groups. Immunohistochemistry was used to detect expression of AKR1C1 and CARS1 in colon cancer tissues and adjacent tissues.

Results

The prognosis model consisted of five ferroptosis related genes (AKR1C1, ALOX12, FDFT1, ATP5MC3, and CARS1). The area under curve (AUC) at 1-, 2-, and 3-year were 0.668, 0.678, and 0.686, respectively. The high- and low-risk patients had significant survival probability and could be clearly distinguished by the PCA and t-SNE analysis. The multivariate cox regression analysis also showed the riskscore is an independent prognosis factor. Importantly, we found that the immune status between high- and low-risk patients were different obviously, such as CD8⁺T cells. And STING, a new promising immune target, was also correlated to our signature genes statistically significantly.

Conclusion

Our ferroptosis prognosis signature could predict survival of CC patients to a certain degree. And the crosstalk between ferroptosis and immune, especially STING need further studies.