Background: The immune landscape of hepatocellular carcinoma (HCC) is heterogeneous. This study aims to develop the immune type which could improve predictive value of HCC survival. Methods: A total of 208 HCC patients in the testing cohort, 112 patients in the validation cohort and 365 HCC patients in the TCGA database were included in this study. Immune features were assessed by immunohistochemical staining or CIBERSORT method. We constructed prognostic classifiers by LASSO COX analyses in the TCGA cohort, which identified five features out of the 22 types of immune cells. Results: The formulas based on the immunohistochemical staining are as follows: ISOS = 0.648* Macrophagestromal + 0.444*Neutrophilsstromal + 0.218*Tregsstromal - 0.703*Memory T cellsstromal; ISDFS = 0.285*B cellsstromal + 0.494*Neutrophilsstromal + 0.431*Tregsstromal - 0.736*Memory T cellsstromal. We classified HCC patients into immune type A subgroup (IS-A) and type B subgroup (IS-B) based on immune scores. The immune type was an independent prognostic indicator for disease-free survival (DFS) and overall survival (OS) in both testing and validation cohorts. Two nomograms (for OS and DFS) that integrated the immune type and clinicopathologic risk factors also showed good predictive accuracy and discriminatory power. IS-A group was correlated with higher immune checkpoint molecule expression. In addition, patients with IS-A and IS-B had distinct mutation signature. Conclusion: The immune types could predict survival and recurrence of HCC effectively. In addition, the immunosuppressive pathways and mutation signature are distinct between two immune types.
Keywords: LASSO COX; TCGA; hepatocellular carcinoma; prognosis; stromal immunotype.