ESTRO 38 Abstract book

S206 ESTRO 38

OC-0404 Dose to vascular calcifications is predictive for overall survival in lung cancer patients E.M. Vasquez Osorio 1 , F. Brewster 2 , A. McWilliam 1 , A. Scaife 3 , K. Banfill 1 , A. Abravan 1 , D. Cobben 1 , C. Faivre- Finn 1 , M. Van Herk 1 1 The University of Manchester, Division of Cancer Studies- School of Medical Sciences- Faculty of Biology- Medicine and Health, Manchester, United Kingdom ; 2 The Christie NHS Foundation Trust, Christie Medical Physics & Engineerring, Manchester, United Kingdom; 3 The University of Manchester, Jodrell Bank Centre for Astrophysics - School of Physics and Astronomy - Faculty of Science and Engineering, Manchester, United Kingdom Purpose or Objective Recent studies suggest that incidental dose to the heart can have an early detrimental effect on overall survival for lung cancer patients receiving radiotherapy. We investigated whether irradiation of vascular calcifications, identified on the radiotherapy planning CT, also affects overall survival. Material and Methods Data from 1002 unselected non-small cell lung cancer patients, all treated with 55Gy in 20 fractions, were used. Calcifications within the thoracic cavity were automatically segmented in the planning CT scans using standard image processing algorithms, including morphological operations, connected pixel analysis and shape analysis. Calcifications for 10 CT scans were manually segmented to evaluate the quality of the automatic segmentation. To explore the interaction of the identified calcifications and the radiation dose on overall survival prediction, the volume of calcifications receiving ≥0, 10, 20, 30, 40 and 50 Gy was determined. We subsequently refer to them as Cx. These variables were included in a Cox-regression analysis alongside log tumour volume, age, mean dose to the lungs, and gender. Heart structures were created for all patients using atlas-based segmentation in ADMIRE v2.0 (Elekta AB, Stockholm, Sweden). The analysis was repeated for calcifications within the heart only. Results he average calcification volume found was 2.6 cm 3 (SD 2.8 cm 3 ), of which 1.3 cm 3 (SD 2.0 cm 3 ) was within the heart. The success rate of the algorithm for identifying calcifications was 81.8%, with an error rate of 8.8%. Figure 1 shows the hazard ratios and significance levels for univariate and multivariate Cox-regressions. Univariate analysis of C10, C20, C30, C40 and C50 were significant for all calcifications, but only C10, C20 and C30 for calcifications within the heart. Tumour volume, age, mean dose to the lungs, and gender were significant in all models. In the multivariate analysis, C20 remained significant when using all calcifications (HR=1.27, CI=1.01, 1.59, p=0.04) but only borderline significant for calcifications within the heart.

Conclusion The data collection and merging in the CAT-system work smoothly and show good potential. The models perform very well on this external patient cohort and could provide doctors and patients with valuable information on patient- specific survival. The model does not include smoking status of the patients, which has been shown to have a major impact on tumor control and survival and therefore could potentially improve the model predictions further.