Peng Wang, Guilian Kong

Department of Nuclear Medicine, Henan Provincial People’s Hospital, People’s Hospital of Zhengzhou University, People’s Hospital of Henan University, Zhengzhou, Henan, 450003, People’s Republic of China

Correspondence: Guilian Kong, Henan Provincial People’s Hospital, People’s Hospital of Zhengzhou University, People’s Hospital of Henan University, Zhengzhou, Henan, 450003, People’s Republic of China, Email kongguilian666@163.com

Background: Angiogenesis and ferroptosis are both linked to hepatocellular carcinoma (HCC) development, recurrence, and medication resistance. As a result, a thorough examination of the link between genes associated with angiogenesis and ferroptosis and immunotherapy efficacy is required to improve the dismal prognosis of HCC patients.
Methods: The molecular subtypes were found using a non-negative matrix factorization technique (NMF) based on the genes associated with angiogenesis and ferroptosis. Based on the differentially expressed genes (DEGs) screed between different molecular subtypes, an angiogenesis and ferroptosis-related prognostic stratification model was built using LASSO-COX regression, random forest technique, and extreme gradient boosting (XGBoost), which was further validated in the ICGC and GSE14520 databases. The impact of this model on tumor microenvironment (TME) and immunotherapy sensitivity was also investigated. The expression levels of candidate genes were detected and validated by Real-Time PCR and immunohistochemistry between liver cancer tissues and adjacent non-tumor liver tissues.
Results: Both angiogenesis and ferroptosis-related genes can significantly divide HCC patients into two subgroups with different survival outcomes, mutation profiles, and immune microenvironments. We screened six core genes (SLC10A1, PAEP, DPYSL4, MSC, NQO1, and CD24) for the construction of prognostic models by three machine learning methods after intersecting DEGs between angiogenesis and ferroptosis-related subgroups. In both the TCGA, ICGC, and GSE14520 datasets, the model exhibits high prediction efficiency based on the analysis of KM survival curves and ROC curves. Immunomodulatory genes analysis suggested that the model could be used to predict which patients are most likely to benefit from immunotherapy. Furthermore, the transcriptional expression levels of SLC10A1 in the validation experiment matched the outcomes derived from public datasets.
Conclusions: We identified a new angiogenesis and ferroptosis-related signature that might offer the molecular characteristic information needed for an efficient prognostic assessment and perhaps tailored treatment for HCC patients.

Keywords: HCC, angiogenesis, ferroptosis, prognosis, immunotherapy