1143 - Quantifying Pan-Cancer Immune Temperature to Predict Patient-Specific Radiation Response and Outcome

Purpose/Objective(s): Clinical success of radiotherapy (RT) is at least in part due to the synergy with a patient’s immune system. The high heterogeneity in tumor-immune contexture within and across different cancer types highlights the need to quantify an immune temperature – beyond hot or cold – to fully decipher the optimal RT dose, dose fractionation, sequence, and timing of RT +/- other therapies. We hypothesize that we can use pretreatment tumor-immune composition to generate an immune temperature map to predict individual response and outcome to RT. Our objectives are to generate a pan-cancer immune temperature map using a data-driven approach and to validate the map on a separate, well-characterized retrospective clinical cohort.

Materials/Methods: We analyzed untreated biopsy tissues of 10,469 patients across 31 tumor types. We used CIBERSORT to characterize the cell composition of tumor tissues and estimated the presence of anti- and pro-tumor immune phenotypes within tumor mixtures. For each patient, we map the frequency of tumor cells (T), as well as anti-tumor (E) and pro-tumor (S) immune cells on a trilinear coordinate simplex, which represents every possible proportion of T, E, and S. Next, we generated an optimal immune temperature, 0-100°, and validated the pan-cancer immune temperature in a separate cohort of 50 non-small cell lung cancer (NSCLC) patients who underwent conventional fractionated RT (n = 30 with locoregional control (LRC); n = 20 with locoregional failure (LRF)).

Results: The diverse patient-specific tumor-immune compositions across 31 tumor types were sufficient to generate an immune temperature map that is consistent with literature reports and the clinical effectiveness of RT. First, we showed that immune temperature correlates with LRC after RT in NSCLC. The immune temperature distribution for patients with LRC was significantly higher than in the LRF cohort (mean 52.08° vs. 36.03°, p=0.007). Next, we showed that patients with an above-median immune temperature (42.3°) had significantly better overall survival than patients with lower immune temperatures (p=0.014). Finally, we show a negative correlation of the molecular radiosensitivity index (RSI) with immune temperature (r=-0.39, p=0.005).

Conclusion: We present a clinically validated immune temperature to predict patient-specific radiation response and outcome, which will help guide clinical decision-making poised to improve RT personalization.