RSI: A genomic signature of radiosensitivity

Most cancer patients receive radiation therapy (RT) during their illness. Virtually all RT courses are based on techniques and fractionation schemes determined by trial and error, often decades ago. Thus there is a pressing and urgent need for a molecular diagnostic to inform personalized RT delivery. Our team developed a molecular fingerprint of tumor radiosensitivity (RSI), and has subjected it to extensive clinical and analytical validation (1-5). RSI score distribution across disease sites is consistent with their known clinical radio-responsiveness as defined by the surviving fraction after 2 Gy (SF2), and has been validated in 2,200 patients in 12 independent datasets across several disease sites. We have shown that RSI correlates with outcome only in patients treated with RT; it is not prognostic but predictive. The National Cancer Institute designed the Clinical Assay Development Program (CADP) to assist with the development of assays that may predict therapy response or prognostic behavior of a diagnosed cancer; RSI has undergone further development through CADP. Using RSI and the linear quadratic model, our team next modeled the genomically adjusted dose (GAD) to predict RT dose effect at the individual patient level. RSI/GAD has predicted cancer cohorts that will specifically benefit from RT-dose escalation, such as radioresistant luminal lesions in breast cancer (6) and some glioblastomas (7). Current data have also revealed that metastatic lesions have markedly different RSI than the primary lesion, which is further modified by the site of metastasis (8). We will discuss current knowledge of RSI/GAD and describe ongoing current research plans to incorporate RSI, as a predictive biomarker of radiosensitivity, into personalized therapy options for RT patients.