The immunosuppressive landscape of the NOTCH gene family in hepatocellular carcinoma (HCC).

Abstract

e16139 Background: Despite the recent addition of atezolizumab to advanced HCC first-line treatment, response biomarkers are not well studied. The involvement of the NOTCH gene family in HCC tumorigenesis has been previously discussed, yet insights on its effect on the immune microenvironment are still lacking. Methods: We systematically analyzed differentially expressed genes (DEGs) in the KEGG NOTCH pathway across HCC transcriptomics datasets from the HCCDB platform. Immune and prognostic analyses were done using The Cancer Genomic Atlas (TCGA) HCC cohort (n = 372). The TIMER 2.0 tool was used to assess immune infiltration. cBioportal was used for tumor mutational burden (TMB) and microsatellite instability (MSI) correlations. The TIDE tool assessed T cell dysfunction and DEGs efficacy predictive value using the only trial administering atezolizumab (IMvigor210, n = 348). Results: Four DEGs were identified: NCSTN, APH1A, NOTCH3, HDAC2. All four genes were significantly directly associated (spearman’s ρ > 0.1, P < 0.05) with immunosuppressive cells: Tregs, macrophage M2 cells, cancer-associated fibroblasts (CAF), and myeloid-derived suppressor cells (MDSC). HDAC2 was associated with higher T cell dysfunction (z score = 2.07, P = 0.04). High NOTCH3 and low HDAC2 expression served as favorable overall survival (OS) prognostic markers (HR: 0.66, 95% CI: 0.48 - 0.90, P = 0.008 & HR: 1.52, 95% CI: 1.12–2.07, P = 0.006, respectively). Stratification based on the expression of these two genes and strongly associated immunosuppressive cells’ infiltration levels (ρ > 0.5), revealed two unique patients’ subsets with worse OS marked by “low NOTCH3 expression with high CAF levels” and “high HDAC2 expression with high MDSC levels”. Regarding immune checkpoint genes expression, NOTCH3 and HDAC2 expression levels were directly correlated with CTLA-4, PD-1, and TIM-3, whereas NCSTN and APH1A were inversely related with PD-L1, LAG-3, and TIM-3 after correction for multiple hypothesis testing (ρ > 0.2, Q value < 0.05). TMB count was inversely associated with NOTCH3 expression (ρ = - 0.2, P < 0.001), as for the MSI sensor score, NOTCH3 (ρ = -0.29) and APH1A (ρ = 0.29) expression had inverse and direct significant associations, respectively (P < 0.0001). Exploratory subgroup analysis from the IMvigor210 trial on metastatic urothelial carcinoma patients on atezolizumab showed a significant difference in OS favoring low NOTCH3 expression (P = 0.003) and high APH1A expression (P = 0.049). Conclusions: DEGs from the NOTCH family seem to impose an immunosuppressive effect on HCC’s microenvironment. High NOTCH3 and low HDAC2 expression were identified as potential favorable prognostic markers. Interestingly, high NOTCH3 expression translated into worse OS in patients on anti-PD-L1 which could be owed to the inverse relation with TMB and MSI described in our study, thus deserving further translational investigations.