ESTRO 2025 - Abstract Book

S3974

Radiobiology - Tumour radiobiology

ESTRO 2025

analyzing both the full cohort and a subset with multiple lung metastases. Feature reduction removed features correlated with volume and each other (| ⍴ | > 0.1 and 0.7, respectively). The MIRV was quantified by maximum Euclidean distance. Treatment response was assessed by change in lesion volume and ctDNA detectability assessed at baseline and post-treatment [3]. Spearman's rank correlation coefficient evaluated the relationship between MIRV and treatment response variability, with multiple testing corrections applied using the Benjamini-Hochberg method to control the false discovery rate (FDR).

Results:

Significant correlations were identified between MIRV and treatment response metrics. In the lung metastasis subset, MIRV was significantly associated with the range of volumetric changes (ρ = 0.46, FDR < 0.05). In the subset of patients with ctDNA data, MIRV was moderately correlated with post-treatment ctDNA detectability (ρ = 0.44, FDR < 0.05); a weak correlation was observed between MIRV and change in ctDNA detectability, but this association did not reach significance (ρ = 0.29, FDR = 0.11). These findings support the hypothesis of MIRV as a marker for differential lesion-specific response; this suggests that MIRV may predict both volumetric response and residual disease burden as indicated by ctDNA. MIRV was not found to be correlated with number of lesions, tumor volume, or other traditional confounders of radiomics analyses. Conclusion: MIRV shows promise as a predictive tool for differential treatment response. Integrating MIRV into predictive models may enhance personalized treatment planning across various cancer types, ultimately improving patient outcomes. Our approach is likely disease-agnostic, highlighting the importance of lesion-specific analysis in personalized treatment strategies, with further extensions to other malignancies warranted for future research. These findings establish MIRV as a key marker for predicting lesion-specific treatment responses, with volumetric response and ctDNA detectability as complementary indicators of therapeutic efficacy.

Keywords: Imaging, Tumor Heterogeneity

References:

[1] W. D. Tap et al. , “Doxorubicin plus evofosfamide versus doxorubicin alone in locally advanced, unresectable or metastatic soft-tissue sarcoma (TH CR-406/SARC021): an international, multicentre, open-label, randomised phase 3 trial,” Lancet Oncol. , vol. 18, no. 8, pp. 1089–1103, Aug. 2017. [2] J. J. M. van Griethuysen et al. , “Computational Radiomics System to Decode the Radiographic Phenotype,” Cancer Res. , vol. 77, no. 21, pp. e104–e107, Oct. 2017.