Program Information
Targeting Hypoxia in Prostate Cancer Using Multiparametric MRI: Radiomics Meets Bio-Focused Radiotherapy
Y Sun1,2*, S Williams1,2, D Byrne2, C Mitchell2, H Reynolds1,2, D Murphy2, A Haworth2,3, (1) The University of Melbourne, Melbourne, VIC, Australia, (2) Peter MacCallum Cancer Centre, Melbourne, VIC, Australia, (3) The University of Sydney, Sydney, NSW, Australia
Presentations
TU-C1-GePD-J(A)-2 (Tuesday, August 1, 2017) 9:30 AM - 10:00 AM Room: Joint Imaging-Therapy ePoster Lounge - A
Purpose: To investigate the voxel-wise correlation between multiparametric MRI (mpMRI) data and prostate hypoxia, characterised by immunohistochemistry (IHC) staining using hypoxia biomarkers (HIF1a, CAIX and GLUT1).
Methods: In vivo mpMRI data from six patients, scheduled for radical prostatectomy, were obtained which included T2w imaging, diffusion-weighted imaging (DWI), dynamic contrast enhanced MRI (DCE-MRI) and blood oxygen level dependent (BOLD) imaging. Texture features on T2w images (T2w-TFs) based on gray-level co-occurrence matrix (GLCM, n=21), gray-level run length matrix (GLRLM, n=11) and gray-level size zone matrix (GLSZM, n=11) were computed. After prostatectomy, histology obtained from tissue blocks was stained using three hypoxia biomarkers (HIF1a, CAIX and GLUT1). Histology was co-registered with in vivo mpMRI data using a 3D deformable registration framework, and with IHC using affine registration methods. Colour deconvolution on IHC images separated the Diaminobenzidine (DAB) layer, which served as the expression levels of corresponding biomarkers. Voxel-wise correlations between expression levels of all three biomarkers and mpMRI data were analysed.
Results: HIF1a expression level was significantly correlated with DCE-MRI parametric maps (kep: -0.11, p=0.01; ve: 0.04, p=0.04) and 16 of the T2w-TFs. HIF1a positively correlated with R2* map from BOLD (but not significant, 0.15, p=0.053). CAIX was negatively correlated with DCE-MRI parametric maps (Ktrans: -0.18, p<0.01; kep: -0.15, p<0.01; IAUC: -0.15, p<0.01) and six of the T2w-TFs. GLUT1 expression levels were not correlated with any raw mpMRI data, but significantly correlated with 22 T2w-TFs.
Conclusion: We have presented a voxel-based framework to correlate mpMRI data with hypoxia biomarkers expression levels. These data indicate the value of DCE-MRI and T2w radiomics texture features in defining underlying biology. To validate our findings, we are incorporating next generation sequencing that provides additional biological information. We intend to apply these methods, along with the prediction of tumour location, cell density and aggressiveness, for personalised bio-focussed radiotherapy.
Contact Email: