Is there a causal relationship between genetic changes and radiomics-based image features? An in vivo preclinical experiment with doxycycline inducible GADD34 tumor cells

Background and purpose: The central hypothesis of "radiomics" is that imaging features reflect tumor phenotype and genotype. Until now only correlative studies have been performed. The main objective of our study is to determine whether a causal relationship exists between genetic changes and image features. The secondary objective is to assess whether the combination with radiotherapy (RT) influences these image features. Material and methods: HCT116 doxycycline (dox) inducible GADD34 cells were grown as xenografts in the flanks of NMRI-nu mice. GADD34 overexpression decreases hypoxic fraction. Radiomics analyses were performed on computed tomography images obtained at 40 kVp and again at 80 kVp for validation, before radiotherapy at a volume of 200 mm(3), 4 days post RT (10 Gy) and 500 mm(3). To select reproducible features test-retest experiments were performed at baseline. Results: Gene induction and/or irradiation translated into significant changes in radiomics features. Post irradiation, 17 features for 40 kVp and 9 features for 80 kVp differed significantly between dox+ and dox combined with RT. 8 and 4 of these features remained consistent for 40 and 80 kVp, respectively. Conclusion: Radiomics is able to identify early effects of changed gene expression combined with radiation treatment in tumors with similar volumes which are not visible to human eye.