Abstract P3-12-04: The genomically adjusted radiation dose (GAD) and its association with distant metastases in breast cancer: A feasible approach to precision medicine in radiation oncology

Abstract

Abstract Background: Clinical validation studies in over 2,200 patients across 8 different disease sites, including breast cancer, have shown the radiosensitivity index (RSI), a gene expression signature, predicts outcomes in patients treated with radiation. We hypothesize that an approach to personalize radiation dose could be developed by integrating RSI into the linear quadratic model of dose and fractionation. Methods: Utilizing the linear quadratic model and RSI, we derived an expression for the genomically adjusted dose (GAD) to model radiation dose effect for individual patients. A higher GAD implies a higher predicted radiation therapy effect. GAD was evaluated as a predictor of clinical outcome in two independent datasets of breast cancer patients treated with surgery and radiation. The association between GAD and distant-metastasis free survival (DMFS) and relapse-free survival (RFS) using univariate (UVA) and multivariate (MVA) Cox proportional hazard models was assessed. Clinical and array-based gene expression were obtained from two independent, previously described cohorts from the Karolinska Institutet and Erasmus University Medical Center. Results: Full radiation treatment details were available for 263 patients in the Erasmus dataset, median follow-up 60 months. GAD-low patients (<75% GAD distribution) were found to have decreased DMFS when compared to GAD-high patients (≥25% GAD distribution) (Hazard Ratio (HR) = 2.31 (95% CI 1.25, 4.25), p=0.006). On MVA, GAD was an independent predictor of DMFS for the whole cohort (HR= 2.11 (1.13, 3.94), p=0.02). When the analysis was restricted to the ER positive cohort, GAD was an independent predictor of outcome both as a continuous (HR=0.977, (0.955, 1.0), p = 0.049) and as a dichotomous variable (HR = 3.42, (1.53, 7.67), p=0.003). These results were independently confirmed in the second Karolinska dataset. The 5 year RFS was 95% for GAD-high patients and 76% in GAD-low patients (p=0.027) and GAD was a significant predictor on MVA for RFS (HR =7.42, (1.41, 137.6), p=0.014). In the Karolinska cohort, we estimate a significant proportion of GAD-low patients (59%) would achieve GAD-high with dose escalation up to 70 Gy. Conclusions: In this study, we develop and validate GAD, a novel and patient-specific measure of radiation dose effect. Importantly, GAD is a clinically actionable metric by adjusting radiation dose. We propose that GAD based radiation dosing is a feasible approach to precision medicine in breast radiation oncology. Citation Format: Ahmed KA, Scott JG, Diaz RJ, Fulp WJ, Torres-Roca JF. The genomically adjusted radiation dose (GAD) and its association with distant metastases in breast cancer: A feasible approach to precision medicine in radiation oncology. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P3-12-04.