The Truth Beyond Prescription and Dose Coverage in High Dose Rate Prostate Brachytherapy. A Tumorlet Perspective

Abstract

PurposeIn prostate brachytherapy, providers use metrics such as prescription dose, D90, V95, and V100 to evaluate implants. In high dose rate (HDR) brachytherapy, different catheter placement strategies, planning target volume (PTV) delineation, and methods of plan optimization can create large variances in final treatment delivery. Dosimetric parameters like Dxx and Vxx, while necessary and widely used, might not be sufficient for a complete description of dose distribution, as HDR produces significant dose gradients over small distances. Acknowledging that multiple foci of disease can be present in the prostate, we analyze the implication of significant dose distributions for these foci of disease (tumorlets).Materials and MethodsTwo representative HDR prostate implants were examined, one small (PTV=46cm3) and one large (PTV=129cm3). Our institution’s standard plan optimization strategy delivered 120-130% of prescription dose to the prostate peripheral zone and allowed for large dose inhomogeneity. Using identical organ at risk constraints, PTV structure, and catheter placement, a second plan was created for each case in which we maximized dose homogeneity (or minimized V150). We evaluated the final treatment plans with standard metrics (D90, V95, and V100). To analyze radiation effects, the equivalent uniform biological effective dose (EUBED) was used as a radiobiological metric. 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. Biological growing of tumorlets in 9 volume categories from 0.1 to 10cm3 with 500 random tumorlets per category was simulated. Each tumorlet was superimposed over the 3D dose distribution and positional, morphological, dosimetric and radiobiological quantities were computed.ResultsThe standard optimization strategy (std) and homogenous optimization strategy (hmg) resulted in similar V95 and V100 for both large (L) and small (S) prostate cases: V95std=99% and 99% vs. V95hmg=99% and 98%, V100std=98% and 97% vs. V100hmg=96% and 95% for the same D90=108.2% (L) and 102.8% (S). The different optimization strategies resulted in significantly different V150 (V150std=42% and 38% vs. V150hmg=22% and 18%. Standard and homogenous plans resulted in similar EUBED for the prostate: 82.8Gy vs. 79.8Gy (L) and 79.4Gy vs 77.3Gy (S). For each tumorlet family, EUBED distributions were compared using the two-sample Kolmogorov-Smirnoff (K-S) test at 5% level of significance. The average EUBED for each distribution significantly decreased with increasing tumorlet volume: 99.3Gy for 0.1cm3 to 87.7Gy for 10cm3 (L) and 92.3Gy to 81.7Gy (S). In both prostate cases, standard and homogenous EUBED distributions were statistically different.ConclusionsDifferent planning optimization strategies have a large impact on final treatment delivery in HDR prostate brachytherapy despite similar or identical prescription doses, D90, V95, V100, and OARs constraints. Radiobiologic metrics at the level of tumorlets may have a role in evaluating implants independent of optimization strategy, allowing for more meaningful comparison between implant styles. The analysis carried out here is a proof of concept which documents that implants with similar classic dose coverage parameters (D90, V95, V100) can vary at the level of foci of disease, depending on the optimization strategy. Different implants (ie. differing catheter number and positioning) can increase this variance. Adding radiobiological tumorlet analysis to the reporting of standard dose and volume coverage parameters improves the capability to compare clinical results within centers and between centers with distinct implant and treatment planning approaches. PurposeIn prostate brachytherapy, providers use metrics such as prescription dose, D90, V95, and V100 to evaluate implants. In high dose rate (HDR) brachytherapy, different catheter placement strategies, planning target volume (PTV) delineation, and methods of plan optimization can create large variances in final treatment delivery. Dosimetric parameters like Dxx and Vxx, while necessary and widely used, might not be sufficient for a complete description of dose distribution, as HDR produces significant dose gradients over small distances. Acknowledging that multiple foci of disease can be present in the prostate, we analyze the implication of significant dose distributions for these foci of disease (tumorlets). In prostate brachytherapy, providers use metrics such as prescription dose, D90, V95, and V100 to evaluate implants. In high dose rate (HDR) brachytherapy, different catheter placement strategies, planning target volume (PTV) delineation, and methods of plan optimization can create large variances in final treatment delivery. Dosimetric parameters like Dxx and Vxx, while necessary and widely used, might not be sufficient for a complete description of dose distribution, as HDR produces significant dose gradients over small distances. Acknowledging that multiple foci of disease can be present in the prostate, we analyze the implication of significant dose distributions for these foci of disease (tumorlets). Materials and MethodsTwo representative HDR prostate implants were examined, one small (PTV=46cm3) and one large (PTV=129cm3). Our institution’s standard plan optimization strategy delivered 120-130% of prescription dose to the prostate peripheral zone and allowed for large dose inhomogeneity. Using identical organ at risk constraints, PTV structure, and catheter placement, a second plan was created for each case in which we maximized dose homogeneity (or minimized V150). We evaluated the final treatment plans with standard metrics (D90, V95, and V100). To analyze radiation effects, the equivalent uniform biological effective dose (EUBED) was used as a radiobiological metric. 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. Biological growing of tumorlets in 9 volume categories from 0.1 to 10cm3 with 500 random tumorlets per category was simulated. Each tumorlet was superimposed over the 3D dose distribution and positional, morphological, dosimetric and radiobiological quantities were computed. Two representative HDR prostate implants were examined, one small (PTV=46cm3) and one large (PTV=129cm3). Our institution’s standard plan optimization strategy delivered 120-130% of prescription dose to the prostate peripheral zone and allowed for large dose inhomogeneity. Using identical organ at risk constraints, PTV structure, and catheter placement, a second plan was created for each case in which we maximized dose homogeneity (or minimized V150). We evaluated the final treatment plans with standard metrics (D90, V95, and V100). To analyze radiation effects, the equivalent uniform biological effective dose (EUBED) was used as a radiobiological metric. 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. Biological growing of tumorlets in 9 volume categories from 0.1 to 10cm3 with 500 random tumorlets per category was simulated. Each tumorlet was superimposed over the 3D dose distribution and positional, morphological, dosimetric and radiobiological quantities were computed. ResultsThe standard optimization strategy (std) and homogenous optimization strategy (hmg) resulted in similar V95 and V100 for both large (L) and small (S) prostate cases: V95std=99% and 99% vs. V95hmg=99% and 98%, V100std=98% and 97% vs. V100hmg=96% and 95% for the same D90=108.2% (L) and 102.8% (S). The different optimization strategies resulted in significantly different V150 (V150std=42% and 38% vs. V150hmg=22% and 18%. Standard and homogenous plans resulted in similar EUBED for the prostate: 82.8Gy vs. 79.8Gy (L) and 79.4Gy vs 77.3Gy (S). For each tumorlet family, EUBED distributions were compared using the two-sample Kolmogorov-Smirnoff (K-S) test at 5% level of significance. The average EUBED for each distribution significantly decreased with increasing tumorlet volume: 99.3Gy for 0.1cm3 to 87.7Gy for 10cm3 (L) and 92.3Gy to 81.7Gy (S). In both prostate cases, standard and homogenous EUBED distributions were statistically different. The standard optimization strategy (std) and homogenous optimization strategy (hmg) resulted in similar V95 and V100 for both large (L) and small (S) prostate cases: V95std=99% and 99% vs. V95hmg=99% and 98%, V100std=98% and 97% vs. V100hmg=96% and 95% for the same D90=108.2% (L) and 102.8% (S). The different optimization strategies resulted in significantly different V150 (V150std=42% and 38% vs. V150hmg=22% and 18%. Standard and homogenous plans resulted in similar EUBED for the prostate: 82.8Gy vs. 79.8Gy (L) and 79.4Gy vs 77.3Gy (S). For each tumorlet family, EUBED distributions were compared using the two-sample Kolmogorov-Smirnoff (K-S) test at 5% level of significance. The average EUBED for each distribution significantly decreased with increasing tumorlet volume: 99.3Gy for 0.1cm3 to 87.7Gy for 10cm3 (L) and 92.3Gy to 81.7Gy (S). In both prostate cases, standard and homogenous EUBED distributions were statistically different. ConclusionsDifferent planning optimization strategies have a large impact on final treatment delivery in HDR prostate brachytherapy despite similar or identical prescription doses, D90, V95, V100, and OARs constraints. Radiobiologic metrics at the level of tumorlets may have a role in evaluating implants independent of optimization strategy, allowing for more meaningful comparison between implant styles. The analysis carried out here is a proof of concept which documents that implants with similar classic dose coverage parameters (D90, V95, V100) can vary at the level of foci of disease, depending on the optimization strategy. Different implants (ie. differing catheter number and positioning) can increase this variance. Adding radiobiological tumorlet analysis to the reporting of standard dose and volume coverage parameters improves the capability to compare clinical results within centers and between centers with distinct implant and treatment planning approaches. Different planning optimization strategies have a large impact on final treatment delivery in HDR prostate brachytherapy despite similar or identical prescription doses, D90, V95, V100, and OARs constraints. Radiobiologic metrics at the level of tumorlets may have a role in evaluating implants independent of optimization strategy, allowing for more meaningful comparison between implant styles. The analysis carried out here is a proof of concept which documents that implants with similar classic dose coverage parameters (D90, V95, V100) can vary at the level of foci of disease, depending on the optimization strategy. Different implants (ie. differing catheter number and positioning) can increase this variance. Adding radiobiological tumorlet analysis to the reporting of standard dose and volume coverage parameters improves the capability to compare clinical results within centers and between centers with distinct implant and treatment planning approaches.