Diffusion weighted MRI-based hypoxic fraction as a prognostic factor in the FLAME trial prostate cancer cohort

Introduction: Although hypoxia is a prognostic factor in prostate cancer (PCa), it is not routinely assessed because it is difficult to measure. A diffusion weighted imaging (DWI) based model was proposed by Hompland et al. estimating oxygen levels in prostate tumor tissue (hypoxic fraction, HF). We investigated if DWI-based HF is a prognostic factor in a radiotherapy PCa cohort. Materials and methods: We calculated DWI-based HF for FLAME trial patients with available imaging and follow-up data. In a subcohort of patients receiving similar near-minimum dose to the tumor (D98%), we derived an optimal HF threshold to separate patients with low and high HF using maximally selected rank statistic method. Using this threshold, we performed Kaplan-Meier analysis to determine the prognostic value of HF. Furthermore, we estimated tumor control probability (TCP) curves to evaluate the dose–response behavior in these HF subgroups. Results: There were 410 patients with 100 clinical failures included in the analysis. The optimal HF threshold of 0.28 separated the full cohort into 334 low and 76 high HF patients. The 10-year disease-free survival was 77% (95%CI: 72–82%) in the low and 60% (95%CI: 49–73%) in the high HF subgroup (p = 0.0016). The TCP curves showed tumor control increase with increasing D98% in both subgroups with consistently lower TCP in the high HF subgroup. Conclusion: The DWI-based HF is an independent prognostic factor in localized intermediate- and high-risk PCa radiotherapy cohort. Moreover, while high D98% cannot compensate fully for high HF, we still observed a TCP improvement when increasing D98%.