Gulnur Ungan1, Anne-Flore Lavandier2, Jacques Rouanet3, Constance Hordonneau2, Benoit Chauveau2, Bruno Pereira4, Louis Boyer2, Jean-Marc Garcier2,5, Sandrine Mansard3, Adrien Bartoli1, Benoit Magnin6,7,8. 1. EnCoV, Institut Pascal, UMR 6602 CNRS, Université Clermont Auvergne, 28 place Henri Dunant, 63000, Clermont-Ferrand, France. 2. Department of Medical Imaging, CHU Clermont Ferrand, 1 place Lucie Aubrac, 63100, Clermont-Ferrand, France. 3. Dermatology Department, CHU Clermont Ferrand, 1 place Lucie Aubrac, 63100, Clermont-Ferrand, France. 4. Biostatistics Unit, DRCI, CHU Clermont Ferrand, 58 rue Montalembert, 63000, Clermont-Ferrand, France. 5. Anatomy Department, Université Clermont Auvergne, 28 place Henri Dunant, 63000, Clermont-Ferrand, France. 6. EnCoV, Institut Pascal, UMR 6602 CNRS, Université Clermont Auvergne, 28 place Henri Dunant, 63000, Clermont-Ferrand, France. bmagnin@chu-clermontferrand.fr. 7. Department of Medical Imaging, CHU Clermont Ferrand, 1 place Lucie Aubrac, 63100, Clermont-Ferrand, France. bmagnin@chu-clermontferrand.fr. 8. Anatomy Department, Université Clermont Auvergne, 28 place Henri Dunant, 63000, Clermont-Ferrand, France. bmagnin@chu-clermontferrand.fr.
Abstract
PURPOSE: Immunotherapy has dramatically improved the prognosis of patients with metastatic melanoma (MM). Yet, there is a lack of biomarkers to predict whether a patient will benefit from immunotherapy. Our aim was to create radiomics models on pretreatment computed tomography (CT) to predict overall survival (OS) and treatment response in patients with MM treated with anti-PD-1 immunotherapy. METHODS: We performed a monocentric retrospective analysis of 503 metastatic lesions in 71 patients with 46 radiomics features extracted following lesion segmentation. Predictive accuracies for OS < 1 year versus > 1 year and treatment response versus no response was compared for five feature selection methods (sequential forward selection, recursive, Boruta, relief, random forest) and four classifiers (support vector machine (SVM), random forest, K-nearest neighbor, logistic regression (LR)) used with or without SMOTE data augmentation. A fivefold cross-validation was performed at the patient level, with a tumour-based classification. RESULTS: The highest accuracy level for OS predictions was obtained with 3D lesions (0.91) without clinical data integration when combining Boruta feature selection and the LR classifier, The highest accuracy for treatment response prediction was obtained with 3D lesions (0.88) without clinical data integration when combining Boruta feature selection, the LR classifier and SMOTE data augmentation. The accuracy was significantly higher concerning OS prediction with 3D segmentation (0.91 vs 0.86) while clinical data integration led to improved accuracy notably in 2D lesions (0.76 vs 0.87) regarding treatment response prediction. Skewness was the only feature found to be an independent predictor of OS (HR (CI 95%) 1.34, p-value 0.001). CONCLUSION: This is the first study to investigate CT texture parameter selection and classification methods for predicting MM prognosis with treatment by immunotherapy. Combining pretreatment CT radiomics features from a single tumor with data selection and classifiers may accurately predict OS and treatment response in MM treated with anti-PD-1.
PURPOSE: Immunotherapy has dramatically improved the prognosis of patients with metastatic melanoma (MM). Yet, there is a lack of biomarkers to predict whether a patient will benefit from immunotherapy. Our aim was to create radiomics models on pretreatment computed tomography (CT) to predict overall survival (OS) and treatment response in patients with MM treated with anti-PD-1 immunotherapy. METHODS: We performed a monocentric retrospective analysis of 503 metastatic lesions in 71 patients with 46 radiomics features extracted following lesion segmentation. Predictive accuracies for OS < 1 year versus > 1 year and treatment response versus no response was compared for five feature selection methods (sequential forward selection, recursive, Boruta, relief, random forest) and four classifiers (support vector machine (SVM), random forest, K-nearest neighbor, logistic regression (LR)) used with or without SMOTE data augmentation. A fivefold cross-validation was performed at the patient level, with a tumour-based classification. RESULTS: The highest accuracy level for OS predictions was obtained with 3D lesions (0.91) without clinical data integration when combining Boruta feature selection and the LR classifier, The highest accuracy for treatment response prediction was obtained with 3D lesions (0.88) without clinical data integration when combining Boruta feature selection, the LR classifier and SMOTE data augmentation. The accuracy was significantly higher concerning OS prediction with 3D segmentation (0.91 vs 0.86) while clinical data integration led to improved accuracy notably in 2D lesions (0.76 vs 0.87) regarding treatment response prediction. Skewness was the only feature found to be an independent predictor of OS (HR (CI 95%) 1.34, p-value 0.001). CONCLUSION: This is the first study to investigate CT texture parameter selection and classification methods for predicting MM prognosis with treatment by immunotherapy. Combining pretreatment CT radiomics features from a single tumor with data selection and classifiers may accurately predict OS and treatment response in MM treated with anti-PD-1.
Authors: Caroline Robert; Georgina V Long; Benjamin Brady; Caroline Dutriaux; Michele Maio; Laurent Mortier; Jessica C Hassel; Piotr Rutkowski; Catriona McNeil; Ewa Kalinka-Warzocha; Kerry J Savage; Micaela M Hernberg; Celeste Lebbé; Julie Charles; Catalin Mihalcioiu; Vanna Chiarion-Sileni; Cornelia Mauch; Francesco Cognetti; Ana Arance; Henrik Schmidt; Dirk Schadendorf; Helen Gogas; Lotta Lundgren-Eriksson; Christine Horak; Brian Sharkey; Ian M Waxman; Victoria Atkinson; Paolo A Ascierto Journal: N Engl J Med Date: 2014-11-16 Impact factor: 91.245
Authors: James Larkin; Paolo A Ascierto; Brigitte Dréno; Victoria Atkinson; Gabriella Liszkay; Michele Maio; Mario Mandalà; Lev Demidov; Daniil Stroyakovskiy; Luc Thomas; Luis de la Cruz-Merino; Caroline Dutriaux; Claus Garbe; Mika A Sovak; Ilsung Chang; Nicholas Choong; Stephen P Hack; Grant A McArthur; Antoni Ribas Journal: N Engl J Med Date: 2014-09-29 Impact factor: 91.245
Authors: Vivek Verma; Charles B Simone; Sunil Krishnan; Steven H Lin; Jinzhong Yang; Stephen M Hahn Journal: J Natl Cancer Inst Date: 2017-07-01 Impact factor: 13.506
Authors: Caroline Robert; Jacob Schachter; Georgina V Long; Ana Arance; Jean Jacques Grob; Laurent Mortier; Adil Daud; Matteo S Carlino; Catriona McNeil; Michal Lotem; James Larkin; Paul Lorigan; Bart Neyns; Christian U Blank; Omid Hamid; Christine Mateus; Ronnie Shapira-Frommer; Michele Kosh; Honghong Zhou; Nageatte Ibrahim; Scot Ebbinghaus; Antoni Ribas Journal: N Engl J Med Date: 2015-04-19 Impact factor: 91.245
Authors: Xue Song; Zhongyun Zhao; Beth Barber; Amanda M Farr; Boris Ivanov; Marilyn Novich Journal: Curr Med Res Opin Date: 2015-03-18 Impact factor: 2.580
Authors: Benjamin Weide; Alexander Martens; Jessica C Hassel; Carola Berking; Michael A Postow; Kees Bisschop; Ester Simeone; Johanna Mangana; Bastian Schilling; Anna Maria Di Giacomo; Nicole Brenner; Katharina Kähler; Lucie Heinzerling; Ralf Gutzmer; Armin Bender; Christoffer Gebhardt; Emanuela Romano; Friedegund Meier; Peter Martus; Michele Maio; Christian Blank; Dirk Schadendorf; Reinhard Dummer; Paolo A Ascierto; Geke Hospers; Claus Garbe; Jedd D Wolchok Journal: Clin Cancer Res Date: 2016-05-16 Impact factor: 12.531
Authors: Adi Nosrati; Katy K Tsai; Simone M Goldinger; Paul Tumeh; Barbara Grimes; Kimberly Loo; Alain P Algazi; Thi Dan Linh Nguyen-Kim; Mitchell Levesque; Reinhard Dummer; Omid Hamid; Adil Daud Journal: Br J Cancer Date: 2017-03-21 Impact factor: 7.640
Authors: B Ganeshan; K A Miles; S Babikir; R Shortman; A Afaq; K M Ardeshna; A M Groves; I Kayani Journal: Eur Radiol Date: 2016-07-05 Impact factor: 5.315