58 Tumor immune microenvironment determinants of response to Lenvatinib and aPD-1 blockade in clear cell renal cell carcinoma

Abstract Background Combinations of immune checkpoint blockade (ICB) and anti-angiogenesis tyrosine kinase inhibitors (TKI) are the mainstay of metastatic ccRCC front-line treatment, but heterogenous responses and lack of biomarkers strongly emphasizes the need to understand RCC-specific mechanisms of effective treatment-related anti-tumor immunity. Although prior work consistently highlights tumor associated macrophages (TAMs) in RCC prognosis and treatment response, TAM targeting has yet to experimentally assessed as a viable treatment sensitizer. Methods We therefore assessed the impact of TAM depletion using a clinical-grade colony stimulating factor receptor-1 inhibitor (CSF1Ri) on ICB or TKI sensitivity in a spontaneous mouse model of ccRCC with inducible kidney-specific Vhl, p53 and Rb1 deletions (KVpR). Mice were randomized to receive either no treatment, single agent Lenvatinib (TKI), aPD-1, or combinations with CSF1Ri. We longitudinally monitored a total of 307 tumors from 96 mice using magnetic resonance imaging (MRI) and collected responding and non-responding tumors for single cell RNA sequencing (scRNAseq) to elucidate the underlying mechanisms of response. Results Surprisingly we found that TAM depletion abolished aPD-1 response, with no effect on Lenvatinib response. ScRNAseq analyses suggest that CSF1R selectively depletes antigen presenting TAMs, leading to a striking loss of T cell infiltration. This suggests that TAMs, rather than dendritic cells as shown in other cancer types, may play a dominant role in eliciting T cells during aPD-1 therapy in RCC. On the other hand, Lenvatinib, which elicited a stronger T cell response than aPD-1, was associated with significant increases in cDC1 and plasma cells in responders, together suggesting that these two treatment modalities may operate through distinct antigen-presenting cells. Also unexpectedly, we found that both Lenvatinib and CSF1Ri induced severe intratumor hypoxia, which is typically associated with poor outcomes in most cancers. We identify a hypoxic niche-specific TAM subset that when abundant at baseline is predictive for poor response to ICB/TKI combos across multiple RCC trials, but post-treatment seems to capture hypoxia-mediated tumor cell death in responders in mouse and humans. We therefore wondered whether high levels of existing tumor hypoxia would exacerbate TKI-induced hypoxia leading to worst outcomes and found that in two novel RCC syngeneic models with high-baseline hypoxia, TKI indeed augments metastatic disease. Conclusions Our results offer mechanistic insight into the failure of CSF1R-based TAM depletion approaches in clinical trials and presents cautionary evidence against the use of hypoxia-inducing agents in tumors with high-baseline hypoxia. We show that TKI and ICB-induce distinct TME and hypoxia-related changes which require further study. More generally our study strongly emphasizes the importance of developing mechanism-guided clinical biomarkers. DOD CDMRP Funding: yes