Room: Stars at Night Ballroom 4
Purpose: To empower new adopters of knowledge-based planning (KBP) to select the auto-planning routine that best matches their clinical priorities, we quantified the performance of publicly-available prostate KBP routines via automated multi-patient batch planning.
Methods: Four publicly-available prostate RapidPlan routines (UCSD, MIAMI, CCMB, WUSTL) were automatically applied across a 25-patient cohort using Eclipse scripting and a PTV prescription of V81Gy=95%. The institutionsâ€™ model training sets differed in contouring guidelines for planning target volume (PTV) and organs-at-risk (OARs), beam arrangements (VMAT versus IMRT), and sample sizes (41-105 patients). Routines used different optimization parameters. Model-estimated dose-volume histograms (DVHs) and deliverable post-optimization DVHs were extracted from plans to calculate average DVHs across the cohort for each routine. Each routineâ€™s average calculated DVH was subtracted from the average DVH for all plans and from the modelâ€™s average predicted DVH for comparison. Each routineâ€™s DVH metrics for PTV (DMAX,D1%,D99%,DMIN), Rectum (DMAX,V70,V60,V40), Bladder (V75,V40), Femur (DMAX), and PenileBulb (DMEAN) were compared to the average using a 2-sided paired t-test (Bonferroni-corrected p<0.05). To control for contouring effects, the full analysis was conducted for two PTV margin schemas: 5mm uniform and 3mm/7mm posterior/else.
Results: Calculated plans generally aligned with their modelâ€™s DVH estimations, save CCMB OAR Dmaxes. Most dosimetric parameter differences were not significant, with the exception of PTV DMAX [MIAMI=111.1%(p=2e-4);WUSTL=108.6%(n.s.);UCSD=108.5%(n.s.);CCMB=107.9%(n.s.)], PTV D99% [MIAMI=97.4%(p=5e-2);WUSTL=98.1%(n.s.);UCSD=97.4%(p=3e-2);CCMB=98.5%(p=1e-4)], Rectum V40 [MIAMI=19.1%(p=8e-6);WUSTL=39.5%(n.s.);UCSD=22.7%(p=3e-3);CCMB=53.5%(p=4e-10)], and Femur DMAX [MIAMI=42.8%(n.s);WUSTL=48.6%(p=1e-3.);UCSD=44.9%(n.s.);CCMB=37.9%(p=1e-4)]. Overall, UCSD and MIAMI had lower rectum doses while CCMB and WUSTL had higher PTV homogeneity. Conclusions were unchanged with different PTV margin schemas.
Conclusion: Using publicly-available KBP routines spares clinicians the substantial effort to develop a de novo model. Our study demonstrated that publicly-available prostate RapidPlan routines mainly differ in their prioritization of rectal sparing against PTV dose homogeneity. Our results allow clinicians to select the routine that best suits their clinical priorities within the standard-of-care.
Funding Support, Disclosures, and Conflict of Interest: KLM acknowledges funding support from AHRQ (R01 HS025440-01). KLM acknowledges research funding, travel support, and honoraria from Varian Medical Systems.