The present prospective study was conducted at a single Institution between August 2017 and June 2020 (Clinicaltrials.gov NCT04703543). The study was approved by the Institutional Review Board (RS 946/17) and specific written informed consent was obtained by each patient before enrollment. Selection criteria as well as other methodological aspects of the study have been reported previously in details (9). In order to be eligible, patients had to have an history of localized prostate cancer treated with RP, followed by an undetectable (<0.1 ng/ml) PSA and a subsequent biochemical recurrence (2 consecutive PSA rises up to 0.2 ng/ml or higher). At restaging with DCE-MRI, all patients had to harbor a presumed local failure defined as an early/fast enhancing discrete lesion on DCE-MRI possibly with a T2w hyperintense lesion (8, 9). The location of each detected lesion was classified as per Connolly et al in perianastomotic, bladder neck and retrovesical (10).
All scans were obtained at our Institution and details can be found elsewhere (9).
All patients then underwent sRT, consisting in an intensity modulated approach to the whole prostatic bed to 66-69 Gy in 30 fractions. Presumed local failures at DCE-MRI were prescribed 73.5 Gy with a simultaneous integrated boost technique. The presumed site of local failure at DCE-MRI was outlined as clinical target volume (CTV) after co-registration to the planning computed tomography (11). The CTV was enlarged by an 8 mm isotropic margin to the planning target volume (PTV). When treated (9), the pelvic nodes (bilateral internal, external and common iliac lymph nodes as well as pre-sacral lymph nodes) were prescribed 54 Gy in 30 fractions. Details on treatment planning and delivery have been reported previously (8).
All patients were offered repeated DCE-MRI (rDCE-MRI) at our Institution as described above and interpreted by the same single observer (L.B.). Response was assessed exclusively on DCE sequences and defined with a qualitative method as it follows: complete response (CR) as the complete disappearance of the target lesion at DCE-MRI, partial response (PR) as any reduction of the size of the area/volume of early/fast enhancement, no response (NR) as no change in size/volume of the target lesion at DCE-MRI.
The first rDCE-MRI was planned at 3 months after sRT completion. In patients without a complete response, imaging was repeated at 3-month intervals until complete dis-appearance or a maximum of 4 repeated scans. In case of complete response, no further scans were offered.
The endpoint of the study is the achievement of a CR at rDCE-MRI during the follow up. Therefore, responses were further dichotomized in CR and noCR, that includes both PR and NR. For patients with multiple local lesions, CR implied the complete disappearance of all lesions seen before sRT. The analysis was initially set for a complete database (no missing observations during the follow up). Unfortunately, some patients did not undergo all planned examinations. Therefore, the analysis has been carried in terms of both actual cumulative incidence (the number of new CRs divided by the total number of individuals at risk) and actuarial cumulative incidence using the Kaplan Meier estimate censoring for patients who did not undergo further testing at the date of last examination. Groups were compared with the chi-squared test, the Mann–Whitney rank test or the log-rank test when appropriate. For proportions, confidence intervals (CI) were computed with the Wilson score method without continuity correction.
Univariable/multivariable binary logistic regression analyses on CR were performed considering the Gleason Grade Grouping at RP (1-2 vs 3 vs 4-5), the PSA value at failure (continuum), the PSA doubling time at failure (continuum), the nodule volume at pre-sRT MRI (continuum), the number of failures at pre-sRT MRI (1 vs multiple), the time interval between RP and sRT (continuum), the time interval between the end of sRT and rDCE-MRI (continuum), the percent decrease of PSA at the 5th week of sRT (continuum) and the location of the failure, anastomotic (yes vs no), in the bladder neck (yes vs no) and retrovesical (yes vs no). For patients with multiple lesions, the total volume was computed by adding up each nodule volume.
PSA recurrence after sRT was defined as a 0.2 ng/ml PSA rise above the nadir (12). Statistical significance was claimed for p values <0.05. All statistical tests were performed using GraphPad (version 8.0.1, GraphPad Software Inc., San Diego, CA) and SPSS (version 25, IBM, Armonk, USA).