Quantitative imaging biomarkers: Magnetic resonance imaging for the prediction of radiation therapy treatment response in prostate cancer

Abstract

Recurrence of prostate cancer after primary radiation therapy is not uncommon, and if unmanaged, can result in disease progression. Early detection of localised recurrence enables opportunities for salvage therapies. The current standard of care for monitoring treatment response involves measurement of blood serum prostate specific antigen levels. This method, however, has limited sensitivity and specificity. Quantitative imaging using magnetic resonance imaging has shown potential as a non-invasive method for monitoring treatment response and may provide a means for early detection of local recurrence prior to development of metastases. Measurement uncertainties in the quantitative parameters, however, can impact their reliability in detecting treatment response. This thesis incorporated multi-site data from a longitudinal study of 21 prostate cancer patients who received radiation therapy. Imaging data pre- and post-radiation therapy was co-registered, and the changes were analysed with region-of-interest and voxel-wise measurements to identify candidate features of treatment response. Thresholds to discriminate treatment-related changes from uncertainties due to scanner stability and non-treatment related biological effects were derived from longitudinal phantom studies and in vivo test-retest studies respectively. These thresholds were found to be dependent on tissue type (tumour vs benign) and prostate zone. Candidate features for monitoring treatment response were identified and found to be dependent on timing of imaging, use of hormonal therapy and method of analysis. Future studies will determine the optimal timing of imaging post-treatment to determine likelihood of local recurrence following radiation treatment. The findings presented in this thesis represent the first steps toward developing reliable predictive models for personalisation of treatment response monitoring and early detection of salvageable local recurrence prior to disease progression.