Dynamic contrast-enhanced (DCE) MRI DCE enables non-invasive characterisation of tissue microvasculature and has been used in multiple cancers sites to assess treatment response [1]. Although prostate cancer is the fourth most common cancer worldwide, with 1.1 million cases diagnosed in 2012 [2], there is little published data assessing DCE-MRI parameter changes during radiotherapy (RT) in this group. In this pilot work, our aim was to measure plasma flow (F_p) and permeability-surface area product (PS) through the course of RT in prostate tumour and normal tissue to determine whether significant changes could be detected early in treatment, to test feasibility for potential future biologically adaptive radiotherapy.Fifteen patients with prostate cancer stage T2b or greater were recruited after 3 months of androgen deprivation therapy, and subsequently received 60 Gy of RT in 20 fractions. Each patient underwent three MRI examinations: (1) before RT (2) during the 3rd week of RT (3) 8 weeks after the start of RT. Patients were imaged at 1.5 T (Achieva, Philips Medical Systems, Best, The Netherlands) using the cardiac coil and a flat perspex table top (made in-house) to match the radiotherapy treatment position. An imaging volume 400 x 400 x 100 mm was chosen, with the prostate toward the inferior slices. The MR examination began with high resolution T2w imaging (TSE, TR/TE=4800/120 ms, matrix 560 x 560 x 20), then all subsequent images were acquired with matrix 176x176x20 (overcontiguous slices) and SENSE factor 2.5 in the PE (LR) direction. Inversion-recovery turbo field echo (IRTFE) was used to measure T₁ (TR/TE/α=2.38/0.77 ms/12°, shot interval 4 s, ETL=51, TI = 64, 250, 1000, 2500, 3900 ms), and was followed by DCE-MRI images (turbo field echo TR/TE/α =2.47/0.86 ms/30°, temporal resolution 1.2 s for 260 time points) acquired during injection of 0.2 ml/kg gadoterate meglumine at 2 ml/s followed by a saline chaser. The AIF was extracted from the external iliac artery at each visit, T1 was estimated from fitting to the IRTFE data, signal-intensity vs time curves were converted to contrast agent concentration vs time curves and finally the AATH model [3] was fitted on a voxelwise basis using in house-software (Python 3.4). Haematocrit was assumed to be 0.4. The dominant tumour lesion (DIPL), peripheral zone (PZ) and central zone (CZ) were outlined by a radiologist and oncologist with special interest in prostate multiparametric MRI on high-resolution T₂w images and regions were transferred to the parameter maps to evaluate microvascular parameters in each ROI. The Wilcoxon signed ranks test was used to assess significance of the parameter changes across time points, and the Mann-Whitney U test was used to assess differences between regions at each visit.Thirteen patients completed all three scans. One patient had the first scan only, and one patient did not have the third scan. One patient had no identifiable area of normal CZ. Box plots for F_p (figure 1) and PS (figure 2) are shown for the DIPL, CZ and PZ across the three visits. P-values for paired tests between visits and Mann-Whitney U tests between regions at each visit are shown in table 1, with p<0.05 considered significant.Our measurements show considerable variation in parameters between patients after androgen deprivation therapy, consistent with other studies [4,5]. Early increases in measures of perfusion and permeability during RT are consistent with previous work in other tumours. An increase in K^trans (a DCE-MRI parameter that combines information about plasma flow and vessel permeability) has been shown in cervix tumours early in radiotherapy treatment [6,7] and also in animal models, where perfusion was measured using laser Doppler imaging [8]. Following this initial increase values return towards baseline 8 weeks after RT. The increased range in values seen early in treatment may indicate a variable response to therapy. Early identification of radioresistant disease would allow consideration of dose escalation or alternative treatment strategies. In future work, the other microvascular parameters estimated using this model (e.g. plasma volume) will be investigated, along with parameter heterogeneity.We have detected significant changes in F_p and PS during radiotherapy in this small patient group. If non-responders could be identified early in RT treatment this may allow selected dose escalation and minimised treatment associated toxicity. We intend to correlate changes with surrogate outcome measures such as nadir PSA and biochemical failure when these data become available.