Measuring Radiation Toxicity Using Circulating Cell-Free DNA in Prostate Cancer Patients

Abstract

Currently, radiation therapy (RT) doses are based on population statistics and correlated to dose distributions. Lack of a biomarker-based tool to predict an individual’s RT toxicity is an important limitation in personalized care, as there are known variations in patients’ normal tissue sensitivities to RT. Following RT, cells are killed and spill DNA into the circulation. Circulating plasma cell-free DNA (cfDNA) released in response to RT damage to tissue can be measured using a proprietary bDNA-based method. Our objectives for this phase I/II clinical trial were to determine if plasma cfDNA measured early in an RT course is lower with proton RT compared to photon RT and higher in patients who experience grade 2 or higher RT toxicity. Patients were eligible for enrollment if planned to receive proton or photon RT for non-metastatic prostate cancer in the setting of an intact prostate or after prostatectomy. Patients underwent collection of plasma before RT and on days 1, 2, 3, 4, and 5 of RT for determination of cfDNA levels. Data was analyzed to examine any correlations among cfDNA scores with RT modality (photon vs. proton) and patient-reported toxicities. Fifty-four patients were evaluable for this study. Patients who received photon RT (n=12) had significantly increased cfDNA levels compared to patients who received proton RT when using the peak post-RT cfDNA score (p=0.039), the average post-RT cfDNA score (p=0.040), or the Day 2 cfDNA score (all minus the pre-RT values for each patient). Five patients (9%) and 3 patients (6%) experienced acute and late grade 2 GI toxicity, respectively. Sixteen patients (29%) and 18 patients (35%) experienced acute and late grade 2 GU toxicity, respectively. No patients developed grade 3+ toxicity. The development of any grade 2+ late toxicity, but not acute toxicity, was associated with cfDNA levels obtained on day 3 of RT (p=0.049). Acute maximum GI toxicity score, but not acute GU toxicity, was significantly correlated with cfDNA levels obtained on all days 1, 2, 3, 4, and 5 of RT (p<0.005). Grade 2 late GI toxicity, but not GU or general toxicity, was significantly correlated with cfDNA levels obtained on day 5 of RT (p=0.017) and with the ratio of the maximum patient cfDNA level to the baseline pre-RT cfDNA level (p=0.034). Plasma cfDNA levels correctly predict which patients receive proton versus photon RT and may potentially be utilized as a minimally invasive predictive biomarker for acute and late patient-reported GI toxicity. Given the similarity in toxicity profiles among the various treatments, the data support the expectation that cfDNA will provide a biological estimate to compliment the dose-volume histogram. A test of this hypothesis is under evaluation in an NCI-funded multi-institutional study.