Excited to share our new work on why immunotherapy does not work well in prostate cancer, and how this biology points to our new treatment strategy! This work was led by Abbas Nazir, myself and Ziyu Lu, from Aviv Regev’s group Genentech. Keywords: perturbation, organoids, spatial biology, tumor microenvironment,... AI agent Preprint:show more

1/7 To study this in an immunocompetent setting, we built a new mouse prostate cancer model by orthotopically transplanting Rb1/Trp53/Pten triple-knockout prostate organoids into NINJA mice, enabling controlled neoantigen induction and tumor-specific T cell engagement. The resulting tumors were castration resistant and metastatic.

2/7 We then profiled tumor progression across space and time with scRNA-seq, Xenium spatial transcriptomics, and Zman-seq [1]. In total, we analyzed 21 tissue sections, a 5,100-gene spatial panel helped by SpatialAgent [2], and about 4 million segmented cells, spanning primary tumors and lung metastases. [1] [2]

3/7 One key result was the emergence of an inflammatory niche during tumor progression. Ccl2+Jak2+ inflammatory CAFs, pro-inflammatory macrophages, and malignant basal cells expanded and became spatially organized around dysfunctional CD8 T cells.

4/7 To ask when infiltrating immune cells get reprogrammed by the tumor microenvironment, we used Zman-seq to time-stamp cells entering the tumor. CD8 T cells began shifting toward dysfunction as early as 24 hours after tumor entry.

5/7 Mechanistically, ligand-receptor analysis pointed to strong niche-to-T cell signaling, including PD-L1/PD-1, CD80/CTLA4, CXCL16/CXCR6, Galectin-9/TIM-3, and CD155/TIGIT. The mouse model also showed strong concordance with human prostate tumors, including inflammatory fibroblast states and dysfunctional-like CD8 T cells.

6/7 Finally, guided by the mechanism, we tested therapy. Anti-PD1 alone did not work. But combining anti-PD1 + JAK1/2 inhibition significantly suppressed tumor growth, reduced malignant epithelial cells, reduced pro-inflammatory macrophages, and increased normal Pi16+ fibroblasts in the TME.

7/7 Overall, the study suggests that prostate cancer immune evasion in our model is driven by a spatially organized inflammatory niche, and that breaking this circuitry may help sensitize tumors to checkpoint blockade. Grateful to all collaborators !! Abbas Nazir, Hanchen Wang, Ziyu Lu, Jeff Lau, Frank Peale, Raj Jesudason, Kelli A Connolly, Zaneta Andrusivova, Julia Lau, Sarah Gierke, Linna Peng, Sara Chan, Jian Jiang, Sandra Rost, Eric Lubeck, Marco De Simone, Bench Daniel, Lisa M McGinnis, Danilo Maddalo, Nikhil S Joshi, Levi A Garraway, and Aviv Regev

@ziyu__lu @genentech Quite interesting