Intrinsic tumor resistance to CAR T-cells is a dynamic transcriptional state that can be exploited with low dose radiation

Purpose: High tumor burden prior to CAR T-cell therapy predicts poor outcome, however it is unclear if this is driven more by cell number or underlying biology. It is also unclear if tumor intrinsic properties that confer CAR T-cell resistance can be clinically manipulated to improve response. Here we capitalize on radiation’fering effects by dose in vivo to examine the relative importance of tumor quantity versus quality prior to CAR T-cell therapy. Experimental Design: Using Nalm6 ALL tumor-bearing mice, we investigate the impact of high dose (20Gy in 5 fractions) radiation to focal bulky disease, versus less cytotoxic low-dose (1.8Gy) to all disease (total tumor irradiation; TTI) on CAR T-cell efficacy. We characterize the dynamics of predictive transcriptional states in mice and patients using tumor RNA-seq under conditions of excellent vs poor response. Results: We find the in vivo sensitivity of leukemia cells to CAR T-cells is not only predicted by a transcriptional score, but this score can be temporarily increased by low dose TTI, which is associated with significantly improved survival. Focal high dose radiation (as is currently done clinically) was more debulking but did not improve survival. The tumor transcriptional state induced by low dose radiation was found in >80% of long-term CAR T-cell survivors. Conclusions: These findings provide a path to 1) identify patients unlikely to respond to CAR T_x0002_cells due to potentially reversible tumor-intrinsic resistance, and 2) potentially improve response using low dose TTI to adjust a tumor’s unfavorable transcriptional state prior to CAR T-cell delivery. 5 replicates each of Nalm6 leukemia cells irradiated with 1.8gy sequenced at 24 hours and 7 days post-treatment and a set of Nalm6 that received sham treatment, 3 sequenced at pre-treatment, 4 at 24 hours, and 1 at 21 days post-treatment.