Author: Jean-Pierre Bissonnette, Catherine Coolens, Laura Dawson, Ryan A Kuhn, Michael Maddalena, Teo Stanescu 👨🔬
Affiliation: Princess Margaret Hospital, The Princess Margaret Cancer Centre - UHN, University of Toronto, Princess Margaret Cancer Centre 🌍
Purpose: To investigate the generation and reproducibility of 3D hypoxia maps in liver hepatocellular carcinoma (HCC) patients using data derived from an Intravoxel Incoherent Motion (IVIM) MR sequence.
Methods: Thirteen liver patients treated with 1-5 fraction SBRT were imaged using IVIM for 3-6 sessions totaling 72 scans on an MR-Linac system (Unity, Elekta, Stockholm). IVIM calculations were performed for each voxel in the tumor volume generating diffusion and perfusion values that were analysed using a hypoxia threshold derived from literature. Three patients were imaged twice during each online adaptive treatment fraction for a total of 13 pairs of intrafraction test-retest scans. The hypoxia maps were compared with intrafraction test-retest scans to evaluate reproducibility of the hypoxia signal using the reproducibility metric (%RC). This metric represents the minimum significant difference between repeated measurements taken under identical conditions at a confidence level of 95%. %RC was also used to compare interfraction scans taken during simulation and the first online adaptive fraction prior to treatment delivery.
Results: The reproducibility metrics for the intrafraction test-retest scans were 10% for diffusion coefficient and 28% for the fraction of hypoxic tumor volume. Bland-Altman plots showed no significant bias or error in the data. For the interfraction scans, the preliminary analysis showed that using the %RC for the hypoxic fraction, 36.4% of tumors showed a significant difference in hypoxic fraction. Ongoing analysis is assessing the impact of setup changes, motion artifacts and anatomical deformations between the two imaging sessions. 63.6% of tumors showed a significant difference across the entire treatment.
Conclusion: We report high reproducibility of intrafraction IVIM data obtained for HCC liver patients. Further analysis is required to understand the changes observed in the interfraction IVIM data. Preliminary results suggest that significant change in measured tumor hypoxia can take place during the treatment process.