ESTRO 2021 Abstract Book

S711

ESTRO 2021

1 University Medical Center Freiburg, Department of Radiation Oncology, Freiburg, Germany; 2 University Medical Center Freiburg, Department of Radiation Oncology, Germany, Germany; 3 University Medical Center Freiburg, v, Freiburg, Germany; 4 University Medical Center Freiburg, Department of Radiation Oncology, Division of Medical Physics, Freiburg, Germany; 5 University Medical Center Freiburg, Department of Radiation Oncology, Division of Medical Physics, Freiburg, Germany Purpose or Objective To evaluate the impact of dose and prescription method in early stage non-small cell lung cancer (NSCLC) treated with stereotactic radiotherapy (SBRT). Materials and Methods Patients with early stage NSCLC treated with SBRT in two consecutive prospective trials (02/2011-09/2020) were included in this analysis. All patients were treated in a basis of a 4D PET-CT or a 4D CT with different fractionation and prescription methods such as: in 3, 5 and 8 fractions to the Dmean, or 60% isodose or according to the lungtech trial (V100%≥95% and V90%≥99% with GTV max 110-120%). Follow up (FU) was performed 2 and 7 weeks after SBRT, then 3-monthly for 2 years with assessment of response (incidence of local control (LC) and overall survival (OS)). Impact of predefined patient and treatment related factors such as the prescription method, the prescribed biological effective dose (BED 10 ), the minimum BED (BED 10,Dmin ), the maximum BED (BED 10,Dmax ) the median BED (BED 10,Dmedian. ) on LC and OS were evaluated. Results A total of 107 patients (male: 60, female: 47) with a median age of 74 years were evaluated. 42 patients had central and 65 peripheral tumors. The median planning target volume was 35 ml in both trials. A total of 56% were treated in 3 fractions, 22% in 5 fractions, 21% in 8 fractions and two patients with ultra-central tumors were treated in 12 fractions. The median prescribed biological effective dose (BED 10 ) was 84 Gy. The prescription was performed to the median in 21%, to the 60% isodose in 51% and in 31% according to the lungtech trial. The median BED 10,Dmin to the PTV was 80 Gy, the median BED 10,Dmax 186 Gy and the median BED 10,Dmedian 129 Gy. The median OS was 33 months (97% CI 27-39) and the LC at 2 years was 91.4%. The prescribed BED 10 did not correlate with the LC (HR 0.954 95CI 0.908-1,002, p=0.062) neither the prescribing method, the BED 10,Dmin, BED 10,Dmax or the BED 10,Dmedian. Additionally, none of these parameters correlated with a PD-0875 Novel Classification and Longitudinal Analysis of Radiation-Induced Lung Parenchyma Damage E. Chandy 1,2 , A. Szmul 3 , A. Stavropoulou 3 , J. Jacob 3,4 , D. Landau 1 , C. Veiga 3 , S. Gulliford 5,6 , J. Fenwick 7 , J. Wilson 8 , M. Hawkins 5 , C. Hiley 9,10 , J. McClelland 3 1 University College London, UCL Cancer Institute, London, United Kingdom; 2 University College London Hospital, Radiotherapy, London, United Kingdom; 3 University College London, CMIC, London, United Kingdom; 4 University College Hospital, Radiology, London, United Kingdom; 5 University College London, Dept of Medical Physics & Biomedical Engineering, London, United Kingdom; 6 University College Hospital, Radiotherapy, London, United Kingdom; 7 University of Liverpool, Institute of Translational Medicine, Liverpool, United Kingdom; 8 University College London, Radiotherapy, London, United Kingdom; 9 University College London, UCL Cancer Institute , London, United Kingdom; 10 University College London, Radiotherapy , London, United Kingdom Purpose or Objective Radiation-induced lung damage (RILD) is a common consequence of thoracic radiotherapy (RT). RILD can impact lung parenchyma in a variety of ways, but previous quantitative descriptions have simplified its classification to Hounsfield Unit density changes. We present a novel classification of the parenchymal features of RILD that describe 5 categories of increasing textural density. We developed a deep learning algorithm (DLA) to automate this classification. We quantify the volumetric change in textures after radiotherapy and correlate these with radiotherapy dose and respiratory outcomes. Materials and Methods Diagnostic CTs were available pre-RT, and at 3, 6, 12 and 24-months post-RT, for forty-six subjects enrolled in a clinical trial of dose-escalated chemoradiotherapy for locally advanced non-small cell lung cancer. 5 parenchymal categories were defined and all 230 CT scans were manually labelled (Fig 1). A DLA was trained to automatically annotate the labels. 200 scans (40 patients, at all 5 time points) were used to train and validate the network and the remaining 30 scans (6 patients) were used as a hold-out test set. Blinded qualitative evaluation was performed by a radiologist to assess the clinical acceptability of both the automated and manual labels. Lung registration was performed between scans at each time point and the effect of radiation dose on each tissue class, and correlation with respiratory outcomes, was assessed. Results The DLA automatically labelled the data with Dice Scores of 0.98, 0.43, 0.26, 0.47 and 0.92 for the 5 respective classes. Qualitative evaluation showed that the automated labels were acceptable in over 80% of cases for all tissue classes, and achieved similar ratings to the manual labels. The change in volume of each tissue class over time generated by manual and automated segmentation was calculated. The 5 parenchymal classes showed distinct temporal patterns, for example, classes 2 and 3 show an early peak whereas class 5 became more prevalent up to 24 months. Moderate correlation was seen between change in tissue class volume and clinical and dosimetric parameters. The Pearson’s correlation coefficient was ≤0.49 between the lung V30 Gy and change in Class 2, and 0.39 between change in Class 1 and decline in FVC. The effect of local dose on tissue class revealed a strong dose-dependent relationship (Fig.2). better OS, Conclusion Both OS and LC were high in both prospective trial, irrespectively of the prescription method. None of the parameters tested correlated with a better OS or LC when a sufficient dose to the D2%, D98% and Dmedian is applied.