ESTRO 37 Abstract book

S167

ESTRO 37

2 Memorial Sloan Kettering Cancer Center, Medical Physics, New York, USA

using an automated, decentralized approach, without exchange of individual patient data. Results Out of 258 patients with esophageal cancer in SCOPE1 trial data, 38 patients developed radiation-induced dyspnea (≥ Grade 2) within 6 months of the end of radiotherapy. Table 1 shows the variables included to develop the lung toxicity models, and their Area Under Curve (AUC) scores in the previous study and on the SCOPE1 data. In addition, the receiver operator curves (ROCs) of using both models on the external validation set of the SCOPE1 are shown in Figure 1 . The curves and AUCs derived by distributed learning were identical to the results from validation on a local host.

Purpose or Objective To 1.) Systematically review tolerance doses for late toxicity following external beam radiotherapy (EBRT) and treatments involving brachytherapy (BT) for prostate cancer in the post-Emami/QUANTEC era with a special emphasis on distinct late gastrointestinal, genitourinary, and sexual dysfunction symptoms (GI/GU/SD), and 2.) Perform a quantitative synthesis of identified dose- volume histogram thresholds (DVH t ). Material and Methods The inclusion criteria for full-text articles published since the Emami paper and the QUANTEC reports were: randomized controlled/case-control/cohort studies with non-aggregated symptom assessments, follow-up time ≥3 months, >20 patients, and primary photon-based treatments for localized prostate cancer with DVH data. The DVH t were converted into equivalent doses in 2Gy fractions (EQD2; GI+SD/GU:α/β=3Gy/6Gy). Within the quantitative synthesis for each symptom, a polynomial function was first fitted to all DVH t (agreement between DVH t : R 2 ), and for symptoms with ≥3 DVH t from ≥2 studies, 2-3 across-studies cohering DVH t (COMT DVHs ) were then identified (cf. equal-weight computational center- of-mass). The review was registered at PROSPERO. Results From 33 studies that fulfilled the inclusion criteria, the majority of DVH t were derived after EBRT and primarily for GI toxicity (N thresholds GI/GU/SD: 97/8/1). Among these five symptoms – Defecation urgency (DU), Diarrhea (DI), Fecal incontinence (FI), Proctitis (P) and Rectal bleeding (RB) – presented with DVH t with reasonable agreement (Median R 2 : 0.94 (range: 0.80-1.00); Figure left). The Identified COMT DVHs (EQD2 (%Vol)) for these symptoms emphasized dose-response relationships both within the low-intermediate, and within the high-dose region (DU: 30(55), 62(25); DI: (28(72), 54(42)); FI: 21(76), 56(32), P: 32(67), 63(24); RB 44(46), 62(32), 70(10)). For BT, DVH thresholds were primarily identified for GI toxicity (N thresholds GI/GU/SD: 14/4/2), but quantitative synthesis was only possible for RB (R 2 : 0.93; COMT DVH : 103(5);

Figure

right).

Conclusion We have externalyl validated previously published dyspnea models using an esophageal cancer dataset. FEV 1 that is not routinely measured in esophageal cancer was imputed using WHO performance status. The prediction performance of the models in esophageal cancer patients treated with high-dose external beam radiotherapy was moderate, AUC of 0.66 (95% CI 0.54 – 0.75.) and 0.68 (95% CI 0.58 -0.77), respectively. Prediction performance was not statistically different from previous training and validation sets, therefore a common thoracic RILD model should be feasible despite a different primary tumor. Risk estimates were strongly determined by WHO score in Model 1 and baseline dyspnea in Model 2. The distributed learning approach gave the same answer as local validation, but is feasible without accessing a validation site’s individual patients-level data. PV-0319 A systematic review and quantitative synthesis of tolerance doses for distinct late toxicities C. Olsson 1 , M. Thor 2 1 University of Gothenburg, Regionalt Cancercentrum RCC väst, Gothenburg, Sweden

Conclusion Overall Rectal bleeding remains the most studied symptom after prostate cancer EBRT and BT, but novel tolerance doses since the Emami paper/QUANTEC reports were found for 17 additional distinct GI, GU, and SD symptoms. Quantitative synthesis was possible for Defecation urgency, Diarrhea, Fecal incontinence, Proctitis, and Rectal bleeding, and suggested coherence between identified DVH thresholds in both the low- intermediate (EQD2: ≤41Gy) and the high-dose region (EQD2: ≥54Gy). Continuous collection of dose-response data for all investigated symptoms is still needed to fully acknowledge the causality of both treatment- and patient-related characteristics on a certain symptom. To accelerate this process, researchers are encouraged to harmonize reporting of tolerance doses and to participate in data sharing initiatives.