ESTRO 2020 Abstract book

S655 ESTRO 2020

When using the mDSS, 736 (49%) patients were expected to obtain more QALYs with RT, and 764 (51%) with RP. In the base case, the mDSS is estimated to result in lower costs (-€708, CI: -€892- -€524) and more QALYs (0.075 years, CI: -1.353-1.504 years) than randomized treatment selection (Figure 2). The probability that the mDSS is cost- effective is estimated to be 95.2%, when using a willingness to pay threshold of €80,000 per QALY, in accordance with Dutch guidelines. The mDSS was dominant (increase in QALYs and decrease in costs) in 93.8% of the simulations.

Department of Precision medicine the D-lab, Maastricht, The Netherlands ; 4 Maastricht university medical centre+, Oncologiecentrum, Maastricht, The Netherlands ; 5 Maastricht university medical centre+, Department of Radiation Oncology MAASTRO Clinic, Maastricht, The Netherlands Purpose or Objective As primary treatment for prostate cancer (PCa), radical prostatectomy (RP) and external beam radiotherapy (EBRT) are two of the main treatment choices. Studies have found that survival after either treatment is similar, but the risks of different toxicity outcomes are not the same (Hamdy 2016). We hypothesized that there is a clinically relevant increase in cost-effectiveness (CE) when using a multifactorial decision support system (mDSS) (Lambin 2013) to choose treatment compared to randomized treatment selection (i.e. choice between RP and EBRT based on coincidence). This would lead to improved patient quality of life and be a step toward individualized care. The aim of this study was to develop a model-based mDSS that compares patient specific outcomes in terms of quality adjusted live years (QALYs) between RP and EBRT and to perform an initial CE analysis comparing the mDSS to randomized treatment selection. Material and Methods We developed a state transition model to calculate the CE of RP compared to EBRT (Figure 1). The transition probabilities from healthy to recurrence and from healthy to toxicity (impotence, urinary incontinence and rectal bleeding) were calculated from prediction models. These models use clinical parameters (such as age, BMI, pre- treatment PSA, Gleason score) to predict the probability of an event after a set period of time. The toxicity models all predict long term side effects, and we assumed them to be life-long. The time horizon was taken to be 10 years, with a cycle time of one month. We performed analyses on a synthetic dataset of 1500 patients, with realistic clinical parameters. For the CE analyses, we compared patient treatment allocation based on the mDSS versus random allocation. For the mDSS arm, we selected the treatment with the highest number of QALYs. We conducted a multivariate probabilistic sensitivity analysis to assess model uncertainty, using a Monte-Carlo simulation. From this simulation, we calculated the CE acceptability curve.

Figure 2: Cost-effectiveness plane

Conclusion This modelling study shows mDSS based treatment decisions will result in a clinically relevant increase in the patients’ quality of life. Future work should include training predictive models on a single patient dataset for all outcomes, rather than using models from literature, and increasing parameter accuracy to reduce statistical uncertainties. PO-1155 The Impact of Undetectable PSA After Salvage Radiotherapy for Prostate Cancer I. Rodrigues 1 , C. Ferreira 1 , J. Gonçalves 1 , L. Carvalho 1 , J. Oliveira 2 , C. Castro 1 , Â. Oliveira 1 1 Instituto Português de Oncologia do Porto Francisco Gentil, External Radiotherapy Department, Porto, Portugal ; 2 Instituto Português de Oncologia do Porto Francisco Gentil, Urology Department, Porto, Portugal Purpose or Objective To assess the influence of undetectable PSA (<0.01ng/mL) after salvage radiotherapy (sRT) on clinical outcomes, namely biochemical and clinical recurrence-free survival (BC-RFS, C-RFS), hormone therapy-free survival for recurrence after sRT (HT-FS) and overall survival (OS). Material and Methods We performed a retrospective analysis of patients treated with sRT for biochemical/clinical recurrence after radical prostatectomy in our Institution between 2012 and 2017. Statistical analysis was carried using IBM SPSS Statistics 25. Biochemical response was evaluated with PSA, and clinical response was assessed with physical examination, CT, PET- PSMA, PET-choline and/or bone scintigraphy after biochemical recurrence. Results We identified 277 patients who underwent sRT. Median age was 68 years (from 48 to 81). Median time from biochemical recurrence to sRT was 9.6 months (IQR-13.8). Most received 66-70Gy to the prostate bed (78.7%), of which 33.2% underwent a boost to the area of local recurrence and 9.7% also treated the pelvic lymph nodes (most to a total dose of 45Gy). Outcome analysis was

Figure 1: The structure of the state transition model

Results