Predicting Acute Radio-Induced Toxicity for Head and Neck Cancer Patients: Combining Dosimetry with Biomarker Data, Disclosing a Synergistic Effect

Abstract

We aimed to establish the added value of combining dosimetry with a binary blood assay for radiosensitivity based on quantification of pATM protein (RADIODTECT©) to predict acute radiotherapy (RT) induced mucositis (MUC) and dysphagia (DYS) in head & neck cancer patients (pts). We enrolled 101 pts prospectively scored for acute toxicity with CTCAE. We considered four endpoints: grade≥2 (G2+) and grade≥3 (G3+) MUC, G2+, and G3+ DYS. We dichotomized the pATM concentration to define radiosensitive (RS) vs. radioresistant (RR) pts using previously described cutoffs for G2+ (57.8 ng/mL) and G3+ toxicity (46 ng/mL). We did RADIODTECT© and toxicity scoring blindly. We considered two already published NTCP models, including (i) the Equivalent Uniform Dose to the oral cavity (n = 0.05, EUD_OC, Odds Ratio [OR] = 1.02) and the mean dose to the parotid glands (Dmean_PG, OR = 1.06) for MUC and (ii) EUD_OC (OR = 1.04), the glottic larynx EUD (n = 0.35, EUD_GL, OR = 1.02) and the volume of pharyngeal constrictor muscles receiving>50Gy (V50Gy_CM, OR = 1.02) for DYS. To account for the association of toxicity with the dose distribution in multiple organs at risk (OARs), we derived a "Weighted Dose Score" (WDS) as a linear combination of the dose factors, using their β-coefficients (= lnOR) as weights. WDS for MUC (WDS_OM) = 0.06*Dmean_PG + 0.02*EUD_OC WDS for DYS (WDS_DYS) = 0.02*V50Gy_CM + 0.02*EUD_GL + 0.04*EUD_OC We used WDS as a comprehensive dose feature to fit a dose response and allowed WDS50 (i.e., the WDS associated with 50% toxicity probability) to be different for RR (WDS50_RR) and RS pts (WDS50_RS). The dose-modifying factor (DMF) is the ratio of WDS50_RS/WDS50_RR. It measures the horizontal shift of the dose-response curve when comparing RS vs. RR pts. We scored G2+ and G3+ MUC in 80 and 41 pts; G2+ and G3+ DYS in 73 and 35 pts. The average concentration of pATM was 57.4ng/mL (sd 22.3ng/mL): 53/101 pts were classified as RS for G2+ toxicity and 35/101 as RS for G3+. On the whole cohort, the RADIODTECT© did not significantly associate with the risk of toxicity. However, we found two different dose-response curves at low WDS. There, the intrinsic biological sensitivity significantly affects the toxicity probability: ORs for RADIODTECT© are 2.6/2.4 for G2+/G3+ DYS, 6.4/2.9 for G2+/G3+ MUC. Furthermore, the difference in the incidence of side effects in RR vs. RS pts decreases as the WDS increases, reaching a region where the doses of OARs play a significant role. When combined with WDS, the RADIODTECT© effectively predicted RS pts, with DMF ranging from 0.77 for G3+ DYS to 0.40 for G2+ MUC. These findings support the hypothesis that dose and biomarkers act synergistically; biologically based radiosensitivity plays a significant role when OARs are exposed at lower doses, while high doses of OARs determine toxicity irrespective of the underlying single pt biological characterization.