Abstract Book

S303

ESTRO 37

SP-0578 Patient-reported outcomes and associated dose response models for prostate cancer patients treated with photon vs. proton therapy L. Muren Aarhus University Hospital, Aarhus, Denmark

year OS was performed (‘R’ mle2 package). Three radiobiological models of increasing complexity were fitted: initially, a TCP model with proliferation and disease stage correction and free parameters α/β, λ and T k, and subsequently models including factors accounting for concurrent chemoradiation (CCRT) and late toxicity. Results The model including both CCRT effects and late toxicity was the only one with non-rejectable (p > 0.05) goodness-of-fit, and had fit values: α/β = 3.8 (95% CI 2.1- 5.1) Gy, λ = 0.39 (0.28-0.51) Gy.day -1 and T k ≤ 12 (0-16.5) days. CCRT was found to sharply increase the tumorcidal effect of radiation dose, by a factor 1.27 (1.18-1.40). Positive correlation of OS with study date and negative correlation with normal-tissue EQD2 indicated the importance of historical technique development and late- toxicity. Late toxicity appears to progressively limit 2- year OS beyond EQD2(3.8) = 60 Gy, while TCP continues to increase, resulting in an ‘optimal’ EQD2 of ~80 Gy, for RT-alone/sequential CRT and 63 Gy for CCRT, beyond which OS is predicted to fall.

Abstract not received

SP-0579 Validation of photon-derived normal tissue complication probability models in a head and neck proton therapy cohort P. Blanchard 1 1 Institut Gustave Roussy, Radiation Oncology, Villejuif, France Abstract text Proton therapy for head and neck cancer is a very promising treatment option because of the physical properties of protons that allow a reduction of the integral dose delivered to the patient. Published series report excellent tumor outcomes and lower toxicity compared to conventional photon therapy. A prospective phase III randomized trial is ongoing to demonstrate the benefits of proton therapy in oropharyngeal carcinoma patients. Due to the cost and low availability of proton therapy, it is essential to develop statistical models to predict toxicity following each treatment option. This would enable physicians to assign each patient to the most suited treatment based on the expected toxicity profile, and hence rationalize treatment allocation. We will present the statistical basis of normal tissue complication probability (NTCP) model development and use head and neck proton therapy as an example to show that although NTCP models are currently imperfect and need refinements, they retain enough validity for both photon and proton based treatments and could be used as a way to select patients for the most relevant therapy. OC-0580 Radiation dose-response of locally-advanced NSCLC: Is further dose-escalation useful? M.G. Nix 1 , J.D. Fenwick 2 1 St James Institute of Oncology, Radiotherapy Physics, Leeds, United Kingdom 2 Institute of Translational Medicine- University of Liverpool, Dept. of Molecular and Clinical Cancer Medicine, Liverpool, United Kingdom Purpose or Objective Trials of dose-escalation beyond 60 Gy in LA-NSCLC have produced inconsistent outcomes, the recent RTOG-0617 trial[1] showing an OS detriment following 22% escalation of radiation doses given concurrent with chemotherapy. Meta-analyses[2] report OS gains/detriment for escalation without/with concurrent chemotherapy. Previous radiobiological modelling has often relied on α/β, λ and T k values derived from HNSCC. This work directly fits radiobiological parameters to 2 year OS data from historical NSCLC dose-escalation trials, to establish a dose-response and analyse an apparent lack of efficacy of escalation for doses > 60 Gy. Material and Methods Trials data published since 1995, with n>20 per arm were retrieved from searches for ‘NSCLC radiotherapy dose escalation’ using PubMed, ScienceDirect and Google Scholar. SABR-like trials (dose-per-fraction > 4.0 Gy treating primarily stage I/II disease) were excluded. 68 trial arms (n = 6597) were analysed: 21 RT alone, 27 sequential and 20 concurrent chemoradiation schedules. Nonlinear multivariable maximum-likelihood fitting of 2 Proffered Papers: RB 6: Role and impact of Normal tissue RT response biology

Conclusion Radiobiological parameters have been derived from LA- NSCLC data with narrow confidence limits. Fractionation sensitivity is greater than previously thought (α/β =3.8 Gy vs. 10 Gy) and proliferation seems to start by day 12, or earlier. Therefore dose escalation via moderate hypofractionation would seem preferable to schedule elongation or accelerated hyperfractionatio n. CCRT strongly enhances apparent EQD2, improving T CP, but potentially increasing late-toxicity. Consequently escalation of concurrent schedules requires careful schedule design with tight OAR constraints: escalation of CCRT to physical dose >70Gy is predicted to result in TCP gains of > 15%, which would improve OS if OAR doses could be limited to <60 Gy. [1] J.D. Bradley, et.al. Lancet Oncol. 16 (2015) 187–199 [2] J. Ramroth, et.al., Int. J. Radiat. Oncol. Biol. Phys. 96 (2016) 736–747