ESTRO 2020 Abstract Book

S860 ESTRO 2020

Conclusion These preliminary results indicate that it may be possible to predict late xerostomia in H&N cancer patients using IMBs calculated on daily MVCT images. It is interesting to note that the highest AUCs, which may indicate the strongest predictors of late xerostomia, were present after the first 5 fractions. These findings should be confirmed on a larger cohort but open up the prospect of setting adaptation protocols in H&N cancer after just one week. PO-1585 Subregional analysis of the parotid glands: predicting late xerostomia in head and neck cancer T. Berger 1 , D.J. Noble 2 , L.E. Shelley 1 , T. McMullan 1 , M. Romanchikova 3 , L.J. Carruthers 1 , L. Cebamanos 4 , G. Beckett 4 , A. Duffton 5 , C. Paterson 5 , R. Jena 2 , D.B. McLaren 6 , N.G. Burnet 7 , W.H. Nailon 8 1 Edinburgh Cancer Centre, Department of Oncology Physics, Edinburgh, United Kingdom ; 2 The University of Cambridge, Department of Oncology, Cambridge, United Kingdom ; 3 National Physical Laboratory, Data Science, Teddington, United Kingdom ; 4 Edinburgh Parallel Computing Center, High Performance Computing, Edinburgh, United Kingdom ; 5 Beatson West of Scotland Cancer Centre, Oncology, Glasgow, United Kingdom ; 6 Edinburgh Cancer Centre, Department of Clinical Oncology, Edinburgh, United Kingdom ; 7 Division of Cancer Sciences- University of Manchester, Manchester Academic Health Science Centre- and The Christie NHS Foundation Trust- Manchester, Manchester, United Kingdom ; 8 Edinburgh Cancer Centre, Department of Oncology Physics / The University of Edinburgh- School of Engineering, Edinburgh, United Kingdom Purpose or Objective Several studies have investigated the relationship between the clinical symptoms of xerostomia and the radiotherapy dose received by subregions of the parotid glands (e.g. superficial lobe/stem cell region). The purpose of this study was to compare the predictive power of image biomarkers (IMB) calculated on the superficial lobe (SL), deep lobe (DL) and whole parotid (WP) gland for identifying late xerostomia in head and neck (H&N) cancer patients. Material and Methods All patients (N=60) received 30 fractions (fx) on a TomoTherapy HiArt System (Accuray, Sunnyvale, CA, USA) with daily mega-voltage CT (MVCT) images with CTCAE toxicity recorded at 12 months. As illustrated in Figure 1, the MVCT images (0.76 x 0.76 x 6 mm) were used to calculate by textural analysis, seventy-three IMBs consisting of first and higher order features on the SL, DL and WP gland. For each feature, their values for the period 1-20fx were used to perform linear regressions and their slopes considered as potential predictors. LASSO analysis was used (100 times) with 4-fold cross validation to select the best predictors (<9) for each region. To evaluate the robustness of each of the 100 combinations, the patients were split into four folds. The training set made up of 3 folds and the testing set, the remaining fold, were used to test the predictive power of each of the LASSO combinations using a logistic regression model. This was evaluated by the Area Under the Curve (AUC), resulting in 4 AUC test and 4 AUC train values which were averaged (mAUC test and mAUC train ). For robustness, this process was repeated 100 times for each combination of predictors and the median mAUC test and mAUC train calculated. The 50 combinations of predictors with the highest median mAUC were selected for the SL, DL and WP and their median mAUC test distributions compared using a t-test.

changes in daily MVCT IMBs calculated on the parotid glands could predict late xerostomia (grade>1) and 2) establish the predictive power of the IMBs at different intervals of the radiotherapy course. Material and Methods As illustrated in Figure 1, IMBs (N=73) consisting of first and higher order features were calculated by textural analysis on the parotid glands of daily MVCT images (0.76x0.76x6mm) of 60 H&N cancer patients. All patients were treated in 30 fractions (fx) with CTCAE toxicity recorded at 12 months. Linear regression was calculated on the IMBs over 6 different time intervals of treatment (first 5, 10, 15, 20, 25 and 30 fx) and the slope of the regression extracted. To select the best (<9) IMBs for prediction on each interval, LASSO analysis was used with 4-fold cross validation (100 times). To evaluate the robustness of each of the 100 combinations, the patients were split into four folds. The training set made up of 3 folds and the testing set, the remaining fold, were used to test the predictive power of each of the LASSO combinations using a logistic regression model. This was evaluated by the Area Under the Curve (AUC), resulting in 4 AUC test and 4 AUC train values which were averaged: mAUC test and mAUC train . For robustness, this process was repeated 100 times for each combination of predictors and the median mAUC test and mAUC train , calculated. The 30 combinations with the highest median mAUC were selected for each interval and their median mAUC distributions compared using a t-test.

Results Of the 60 patients, 27% reported xerostomia (grade>1) at 12 months. The highest median mAUC train / mAUC test was 0.93/0.80, 0.90/0.75, 0.89/0.73, 0.93/0.82, 0.90/0.75 and 0.94/0.84 for the slopes of the first 5, 10, 15, 20, 25, 30 fractions, respectively. As shown in Figure 2, the best performing 30 combinations of predictors for the first 5, 20 and 30 fractions performed better in median mAUC train and mAUC test compared to the others. However, compared to the first 5 fractions the AUC distributions for 30 and 20 fx were not significantly higher (p>0.05).