Abstract
Introduction: In tumor mouse models, electron paramagnetic resonance (EPR) imaging has been used to quantify pO2 in-vivo for oxygen-guided radiation therapy. Irradiating the hypoxic tumor sub-regions with a boost of radiation significantly increased survival probability when compared to the boosted radiation of normoxic tumor sub-regions. These results indicate 1) the importance of accurate oxygen imaging in-vivo and 2) an advantage to dose-painting hypoxic tumor sub-regions to improve radiation therapy outcomes of hypoxic tumor treatment. The purpose of this study is to evaluate the accuracy of hypoxia imaging using 18F-misonidazole (FMISO) with PET imaging while using EPR imaging as the reference standard for true hypoxia. We include T2-weighted MRI to define tumor anatomy and dynamic contrast enhanced (DCE)-MRI imaging to model vasculature properties to make FMISO PET more accurate in its measurement of hypoxia.
Methods: We used six SCC7 tumor mouse models in C3H mice, grown in the leg in the range of 200-500 mm3. Under minimal anesthesia, each mouse leg was set in a soft vinylpolysiloxane cast with embedded fiducials for registered images. A custom-built PET insert was inserted into a 720 MHz EPR imager for near-simultaneous imaging with PET and EPR, acquired two hours post-injection of FMISO to allow the radiotracer to bind to hypoxic tumor cells. Then, T2-weighted and gadodiamide DCE-MRI images were acquired and Ktrans and ve parameters were fitted to model tumor vasculature properties using the Tofts model. Data from all modalities were registered using fiducials, and resampled to isotropic (0.5 mm)3 voxels. The thresholds for hypoxia were defined as tumor to muscle ratio (TMR) ≥ 2 and pO2 ≤ 10 torr for PET and EPR, respectively. To correct the PET image using DCE-MRI parameters, we first modeled FMISO retention as a logistic function of pO2 to map the EPR image to PET TMR image, so that its sigmoidal point of inflection was at the threshold of retention. Then we estimated optimal weighting coefficients of DCE-MRI parameters to add or subtract voxel by voxel to the PET data so that its definition of hypoxia is more similar to the EPR image's hypoxia definition. The quality of overlap between hypoxic tumor regions as defined by PET and EPR was assessed using the Dice Similarity Coefficient (DSC) and the Hausdorff Distance (HD), before and after applying a correction to the PET data.
Results: The DSC between hypoxic regions as defined by PET and EPR, before and after applying a correction to PET data, was 0.53 ± 0.2 and 0.80 ± 0.2, respectively. The HD was respectively 3.8 ± 0.5 and 2.1 ± 0.5 mm.
Conclusion: These results indicate that there is poor agreement between FMISO PET hypoxia measurements when compared to hypoxia as defined by EPR pO2 images. Applying our correction method significantly improved the overlap between PET and EPR hypoxic tumor regions.
Citation Format: Inna Gertsenshteyn, Boris Epel, Lara Leoni, Xiaobing Fan, Richard Friefelder, Eugene Barth, Heejong Kim, Marta Zamora, Erica Markiewicz, Darwin Bodero, Mohammed Bhuiyan, Anna Kucharski, Hsiu-Ming Tsai, Mellissa Grana, Subramanian V. Sundramoorthy, Gregory S. Karczmar, Chien-Min Kao, Chin-Tu Chen, Howard Halpern. Multimodal imaging of tumor hypoxia with 18F-misonidazole PET, EPR, and MRI [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1648.