Session Item

Integration of imaging modalities in treatment planning
Imaging has an important role in the staging and detection of radio-recurrent prostate cancer. For focal strategies, delineation of the recurrent tumor inside the prostate gland is necessary. Although PSMA-PET is widely used for staging of recurrent disease, and can be helpful to detect local disease, multi-parametric MRI has the benefit of superior soft-tissue contrast and high spatial resolution.
Indeed, multi-parametric MRI is widely used in staging and detection of prostate cancer and is also used for delineating the tumor inside the prostate gland. However, caution is required when interpreting multi-parametric MR images from patients with post-radiotherapy recurrences. Radiotherapy induces changes in the prostatic tissue that influence its appearance on MRI. To distinguish the radio-recurrent cancer from radiotherapy-induced changes in non-cancerous tissue, we studied three cohorts of patients: 33 patients with biochemical failure after external-beam radiotherapy (cases), 35 patients without post-radiotherapy recurrent disease (controls), and 13 patients with primary prostate cancer (untreated). After radiotherapy, values for T2 were significantly decreased in non-cancerous tissue. Some reduction of ADC values was also seen. This reduced the contrast between non-cancerous tissue and the recurrent tumor. However, the reduction in perfusion parameters such as Ktrans in non-cancerous tissue after irradiation improved the visibility of recurrent tumor. Interestingly, a narrow region with high Ktrans is seen around the urethra.
To facilitate contouring of the recurrent tumor, a tumor probability (TP) model was trained using multi-parametric MRI of 21 patients with radio-recurrent prostate cancer. The structure of the model was similar to an earlier published TP model for untreated prostate cancer. The performance of the model was tested in a cohort of 17 patients who received multi-parametric MRI prior to a salvage prostatectomy, thus providing us with a histological ground truth for the tumor delineations. The voxel-wise TP maps were clustered using k-means clustering with 3 clusters, to define a clinical target volume (CTV), a high-risk CTV and the GTV, with increasing tumor risk. This was compared with manual tumor delineations performed by 2 radiologists and with the histopathology-validated contours.
The TP model had a good performance in predicting voxel-wise presence of recurrent tumor. Model-derived tumor risk levels achieved sensitivity and specificity similar to manual delineations in localizing recurrent tumor. Voxel-wise tumor probability derived from multi-parametric MRI can in this way be incorporated for target definition in focal salvage of radio-recurrent prostate cancer.