Radiomics-based prediction of HCC response to atezolizumab/bevacizumab
Zeitschrift für Gastroenterologie · 01.09.2024 · 62 (09) · e735-e736
DOI: 10.1055/s-0044-1789914 ↗Kurz gefasst
Fachartikel in der Zeitschrift Zeitschrift für Gastroenterologie, erschienen am 01.09.2024. Verfasst von I. Rodriguez und 25 weiteren Autor:innen.
Abstract (Originalsprache)
Background
Atezolizumab/bevacizumab has emerged as a breakthrough therapy for advanced hepatocellular carcinoma (HCC). However, as not all patients respond equally to immunotherapy, predicting treatment response remains an important issue in order to stratify and optimize treatment.
Methods
We investigated 103 HCC patients treated with atezolizumab/bevacizumab, and either disease control (DC) or progressive disease (PD) based on their response to therapy. Radiomics features were extracted from baseline MRI scans. Predictive models using radiomics variables and clinical variables were developed separately and ultimately also integrated to create a composite predictive model. Hyper-parameter tuning and stratified K Fold cross-validation were employed to optimize predictive accuracy and ensure robustness.
Results
While clinical parameters alone showed limited predictive accuracy (40% accuracy, AUC 0.62), radiomics features demonstrated modest performance (55% accuracy, AUC 0.79). However, the hybrid model combining both clinical and radiomics data exhibited significantly improved predictive efficacy, with an overall accuracy of 88% and an AUC of 0.98 for predicting disease progression.
Conclusion
Machine learning significantly enhances the development of hybrid models capable of predicting response to atezolizumab/bevacizumab by leveraging both clinical and radiomics features. Our hybrid model clearly outperformed models solely based on clinical/serological parameters or radiomics features. Publication History Article published online: 26 September 2024 © 2024. Thieme. All rights reserved. Georg Thieme Verlag KG Rüdigerstraße 14, 70469 Stuttgart, Germany
Zitieren
APA
Rodriguez, I., Vellala, A., Itzel, T., Daza, J., Vácha, M., Chang, D., Debic, M., Dill, M. T., Seidensticker, M., Mayerle, J., Munker, S., Schoenberg, S. O., Müller, L., Galle, P. R., Weinmann, A., Tamandl, D., Pinter, M., Scheiner, B., Pech, M., …, & Froelich, M. F. (2024). Radiomics-based prediction of HCC response to atezolizumab/bevacizumab. Zeitschrift für Gastroenterologie, 62(09), e735-e736. https://doi.org/10.1055/s-0044-1789914
Vancouver
Rodriguez I, Vellala A, Itzel T, Daza J, Vácha M, Chang D, et al. Radiomics-based prediction of HCC response to atezolizumab/bevacizumab. Zeitschrift für Gastroenterologie. 2024;62(09):e735-e736. doi:10.1055/s-0044-1789914
BibTeX
@article{Rodriguez2024,
author = {Rodriguez, I. and Vellala, A. and Itzel, T. and Daza, J. and Vácha, M. and Chang, D.-H. and Debic, M. and Dill, M. T. and Seidensticker, M. and Mayerle, J. and Munker, S. and Schoenberg, S. O. and Müller, L. and Galle, P. R. and Weinmann, A. and Tamandl, D. and Pinter, M. and Scheiner, B. and Pech, M. and Sinner, F. and Keitel, V. and Venerito, M. and Santamaria, L. and Ebert, M. P. and Teufel, A. and Froelich, M. F.},
title = {Radiomics-based prediction of HCC response to atezolizumab/bevacizumab},
journal = {Zeitschrift für Gastroenterologie},
year = {2024},
volume = {62},
number = {09},
pages = {e735--e736},
doi = {10.1055/s-0044-1789914}
}
EUVIMED