Immune cell-dependent resistance to LAG3 blockade in SMARCB1-deficient renal medullary carcinoma

Background: Renal medullary carcinoma (RMC) is an exceptionally aggressive kidney cancer of young individuals with sickle cell trait, characterized by SMARCB1 loss, rapid progression, and a median survival of ~13 months. Transcriptomic profiling shows a highly inflamed microenvironment with broad immune-checkpoint upregulation. Among these, PD-1 (log₂ fold-change 1.7; FDR = 0.008) is increased, and LAG-3 (log₂ fold-change 4.6; FDR < 0.0001) shows one of the strongest elevations relative to adjacent kidney tissue. These findings motivated an RMC-specific clinical trial testing anti-PD-1 plus high-dose anti-LAG-3 (NCT05347212), yet mechanisms underlying resistance to LAG-3 blockade remain unknown. Study Design and Methods: To investigate RMC response to PD-1 and LAG-3 inhibition, we used a somatic-mosaic immunocompetent genetically engineered mouse model (GEMM) of RMC on a sickle-trait background. We compared anti-PD-1 and anti-LAG-3 monotherapy with isotype IgG in immunocompetent mice to assess baseline sensitivity to checkpoint inhibition. To dissect mechanisms of LAG-3 resistance, we evaluated anti-LAG-3 treatment in RAG1-knockout mice lacking mature B and T cells and examined anti-LAG-3 under CD4⁺ T-cell depletion to identify the immune subsets driving resistance. Tumor weights were assessed at endpoint, and statistical comparisons were performed using the Mann-Whitney test. Results: Anti-PD-1 monotherapy failed to control tumor growth, confirming intrinsic resistance to PD-1 blockade (mean tumor weight: IgG 0.58 g vs anti-PD-1 0.75 g; p = 0.6682). In contrast, anti-LAG-3 monotherapy exacerbated tumor progression (IgG 0.74 g vs anti-LAG-3 1.27 g; p = 0.0297). Loss of adaptive immunity partially reversed this phenotype: in RAG1-knockout mice, anti-LAG-3 showed a trend toward reduced tumor burden (IgG 1.415 g vs anti-LAG-3 0.61 g; p = 0.127). CD4⁺ T-cell depletion abrogated resistance, yielding significantly lower tumor weights under anti-LAG-3 (anti-LAG-3 1.39 g vs anti-LAG-3 + CD4 depletion 0.75 g; p = 0.0291). Conclusions: Our findings identify an immune cell-dependent mechanism limiting the efficacy of LAG-3 blockade in SMARCB1-deficient RMC. Disrupting this axis, alone or alongside PD-1/LAG-3 inhibition, emerges as a testable strategy to blunt immune checkpoint therapy-induced adaptive reprogramming and restore anti-tumor activity.