PD-0188: Validation of the Prostate Tumorlet Model Based on Multi-Institution Clinical Data

This is a multi-institutional, multi-stage study on prostate focal therapy. The main aim is to validate the tumorlet model, through a foci-based retrospective analysis of actual LDR brachytherapy treatments designed for the whole prostate, the idea being that tumorlets (i.e. intraprostatic lesions or cancer foci) not the whole prostate are the likely targets of our treatments. Secondary aim is to identify possible relationship between dosimetric parameters at the level of the prostate (D90, V100) and tumorlets doses, and to describe the dosimetry and radiobiology of tumorlets in traditional whole prostate cancer treatments. We believe that this is a meaningful way, based on clinical data, to create prescription guidelines for focal therapy. A software platform was built to create in-silico prostate tumorlets, of specified morphology, volume and location. Actual knowledge of tumorlet location was replaced with statistical inference. In this first stage, we analyzed I-125 pre-implant plans for 35 patients, with prostate sizes ranging from 11.9 to 49.4 cm3. Biological growing of tumorlets in 9 volume categories from 0.1 to 10cm3 with 500 random tumorlets per category was simulated. No extra capsular tumor spread was considered. Each tumorlet was superimposed over the 3D dose distribution and positional, morphological, dosimetric and radiobiological quantities were computed. Two formalisms were used to integrate voxel level BED quantities: EUBED and gBEUD. A treatment reference (BEDo) was established using the BED EBRT equivalent of 78Gy regimen in 2Gy/fx. ‘Probabilities of cure' were calculated for each tumorlet, based on EUBED (and gBEUD) > BEDo. The pre-plans studied had a mean D90 of 160.1 Gy (SD: 26.0 Gy) and mean V100 of 94.9% (SD: 11.6%). Treatment planning was performed using volume optimization followed by manual adjustment. The whole prostate EUBED was on average 88.7 Gy3 and range [49.6-137.5] Gy3, significantly lower than BED calculated based on D90, with an average of 161.8 Gy3 and range [96.2-211.8] Gy3 By comparison, BEDo for the EBRT regimen was ∼130 Gy3. In all pre-implant plans studied the percentage of tumorlets receiving gBEUD and EUBED > BEDo shown a monotonic decrease with decreasing D90. When separated in ‘small’ (0.1-1cc) and ‘large’ (2-10cc) volumes, the small tumorlets showed less dramatic drops in ‘cure’ rates than large tumorlets (Figure 1). The slopes for EUBED and gBEUD-based ‘probabilities of cure’ were similar, at 0.55 vs. 0.52 for all tumorlet volumes, 0.40 vs. 0.35 for small volume tumorlets and 0.68 vs. 0.65 for large volume tumorlets. When each of the nine volume categories were analyzed, while the trend was similar, small volume tumorlets seem to show a much more linear ‘cure’ drop with decreasing D90, as evidenced by R2 of the linear fit. Smaller tumorlets, in the setting of prostate brachytherapy seed implants designed for whole prostate seem less sensitive to D90 decreases than large tumorlets. For all tumorlet volumes, there is a linear decreasing of EUBED with prostate D90. It has been reported in the literature, based on clinical outcome data that a BED greater than 200Gy seem to correlate with better outcome. These studies calculated BEDs based on D90, thus such BED > 200Gy would be the equivalent of D90 > 185Gy. Our model based on EUBED (or gBEUD) shows larger than 90% ‘cure’ rates for D90 greater than 185Gy, thus validating not only the model, but also the fact that constraints imposed on tumorlets seem to be equivalent with constraints imposed on the whole prostate.