A radiosensitivity gene signature and PD-L1 status to predict clinical outcome of patients with invasive breast carcinoma in the cancer genome atlas (TCGA) dataset.

54 Background: A radiosensitivity gene signature that included 31 genes was identified using microarray data from NCI-60 cancer cells; however, this has not been validated in independent datasets for breast cancer patients. We investigated the link between the radiosensitivity gene signature and programmed cell death ligand 1 (PD-L1) status and clinical outcome to identify a group of patients that would possibly receive clinical benefit of radiotherapy (RT) combined with anti-PD1/PD-L1 therapy. Methods: We validated the identified gene signature related to radiosensitivity and analyzed the PD-L1 status of invasive breast cancer in The Cancer Genome Atlas (TCGA) dataset using bioinformatic tools. To validate the gene signature, 1,065 patients (or samples) were selected and divided into two clusters using a consensus clustering algorithm based on their radiosensitive (RS) or radioresistant (RR) designation according to their prognosis. Patients were also stratified as PD-L1-high or PD-L1-low based on the median value of CD274 mRNA expression level as surrogates of PD-L1. The relationship between the RS/RR groups and PD-L1 status was also assessed. The prognostic value was evaluated by Kaplan-Meier analysis and Cox proportional hazard models. Results: Patents assigned to the RS group had better 5-year recurrence-free survival (RFS) rate than patients in the RR group by univariate analysis (89% vs. 75%, p = 0.017) only when treated with RT. The RS group was independently associated with the PD-L1-high group, and CD274 mRNA expression was significantly higher in the RS group (p<0.001) than the RR group. In the PD-L1-high group, the RS group had better 5-year RFS rate compared to the RR group (89% vs. 72%, p = 0.015), and this difference was also significant by Cox-hazard proportional analysis. Conclusions: The radiosensitivity gene signature and PD-L1 status were important factors for prediction of the clinical outcome of RT in patients with invasive breast cancer and may be used for selecting patients who will benefit from RT combined with anti-PD1/PDL1 therapy.