ABSTRACT: Immunotherapy has opened hitherto unknown possibilities to treat cancer. Whereas some cancer types (e.g. melanoma) can be efficiently treated, others lack measurable positive effects (e.g. PDAC). Moreover, stratification of responders/non-responders is only possible in some cancer types (e.g. melanoma). Hepatocellular carcinoma (HCC) has a dismal prognosis, limited treatment options and survival benefit, and represents a potential cancer entity for successful immunotherapy. Here, we investigated NASH-triggered HCC in the context PD-1-targeted immunotherapy. Using flow cytometry, single cell RNA sequencing, immunohistochemistry and mass spectrometric analyses, we found a progressive increase of CD8+PD-1+ effector T-cells with a unique profile of exhaustion and activation markers rising with murine and human NASH severity. Notably, late-stage HCC treatment with PD-1-targeted immunotherapy enhanced hepatic carcinogenesis in mice. Dissecting potential mechanisms of action during tumor-initiation and -progression we analyzed the effects of PD-1-targeted immunotherapy at HCC initiation. PD-1-targeted immunotherapy induced a pro-tumorigenic environment, enhanced necro-inflammation and increased NAFLD-activation score (NAS), leading to increased liver cancer incidence, tumor number and nodule size. In contrast, anti-CD8 or anti-CD8/anti-NK1.1 treatment reduced NAS and abrogated the development of liver cancer, thus identifying CD8+PD-1+ T-cells as drivers of liver cancer in NASH-triggered HCC. Increased apoptotic signaling, STAT3 phosphorylation and hepatic proliferation were detected in intra-tumoral liver tissue upon PD-1-targeted immunotherapy. In line, PD-1-/- mice challenged with a NASH diet displayed early onset of hepatocarcinogenesis, corroborating the pro-tumorigenic role of absent or reduced PD-1. Mechanistically PD-1-targeted immunotherapy mainly affected hepatic abundance of CD8+PD-1+ T-cells, instead of altering the quality of Tox+CXCR6+ expressing CD8+PD-1+TNF+CD39+Gzmb+ T-cells found in NASH livers, leading to an aggressive, pro-tumorigenic liver environment. Single-cell mapping of human NASH-, borderline NASH- or unaffected livers corroborated our preclinical NASH results. Moreover, in human NASH livers a correlation of hepatic CD8+, PD-1+, TNF+ T-cells with fibrosis and NASH severity was observed. Accordingly, HCC patients with NASH etiology display a sharp increase in intra- and peri-tumoral CD8+ PD-1+ T-cells. In a cohort of 65 patients recruited across 6 centers in Germany and Austria, patients with NAFLD/NASH-driven HCC responded worse to PD-1-targeted immunotherapy by Nivolumab or Pembrolizumab compared to non-NAFLD patients. This resulted in significant reduced overall survival, in trends of faster disease progression and reduced progression free survival. Histological analysis of livers derived from HCC patients treated with PD-1-targeted immunotherapy displayed high levels of intra and peri-tumoral CD8+ PD-1+ T-cells and Ki67+ hepatocytes. Taken together, these data indicate that PD-1-targeted immunotherapy induces immune-related adverse effects in NAFLD/NASH-driven HCC through CD8+PD-1+ T-cells. Our data call for stratification of HCC patients subjected to PD-1-targeted immunotherapy, with NAFLD being a negative predictor.
Mario Oroshi 
