KEY POINTS
- This retrospective single-center study from Fox Chase Cancer Center, Philadelphia, analyzed 250 treatment fractions in 50 patients with prostate cancer treated with CBCT-guided online adaptive SBRT using the Varian Ethos platform. All patients received 37 Gy in five fractions, with a 3-mm isotropic PTV expansion and nine- or twelve-field IMRT. For each fraction, investigators compared the original scheduled plan recalculated on daily anatomy against a fully reoptimized adaptive plan. The study therefore evaluated competing dosimetric strategies within the same treatment workflow rather than comparing separate patient cohorts.
- Adaptive planning significantly improved PTV target coverage, but the average gain was modest. Mean per-fraction PTV D95 was 740.82 cGy with adaptation versus 719.39 cGy with scheduled planning, corresponding to a paired difference of 21.43 cGy (95% CI 11.62–31.24 cGy; p<0.001). The patient-clustered effect size was Cohen's dz 0.62. Importantly, 83% of scheduled fractions already achieved D95 within approximately 3% of the prescribed fractional dose, suggesting limited incremental benefit from adaptation under routine anatomical conditions.
- A minority of scheduled plans exhibited substantial target undercoverage. Coverage reductions exceeding 5% relative to adapted plans occurred in approximately 11% of fractions, involving eight of the 50 patients. More pronounced undercoverage, defined as PTV D95 ≤80% of the prescription, occurred in 9/250 scheduled fractions (3.6%), while no adapted fraction demonstrated a deviation of this magnitude. Three patients experienced at least one scheduled fraction with D95 below 50%, emphasizing the potential importance of detecting extreme anatomical or contour-related deviations before delivery.
- Large single-fraction deviations generally had limited cumulative effects. For the eight patients with significant scheduled-plan undercoverage, investigators reconstructed the dose that would have been accumulated if all five fractions had been delivered without adaptation. Most cumulative prostate D95 values remained near 37 Gy, with the greatest confirmed reduction occurring in one patient who received a reconstructed 35.2 Gy versus 37 Gy planned, approximately 5% lower. Another patient initially appeared to have accumulated only 81% of the prescription, but offline investigation revealed a contour-dependent isocenter-shift artifact; correcting the shift restored the calculated dose to approximately prescription level.
- Rectal and bladder dose improvements were statistically significant for selected endpoints but small in absolute magnitude. Median adaptive-minus-scheduled differences were −5 cGy for rectal D2cc (p=0.0028), −4 cGy for rectal D20% (p=0.0005), and −4 cGy for bladder D25% (p=0.0002). Conversely, bladder Dmax increased by a median 5 cGy (p<0.001) after adaptation. Most scheduled and adaptive plans already met the principal OAR constraints, indicating that reoptimization primarily reduced occasional threshold violations rather than consistently producing large normal-tissue dose reductions.
- The apparent benefit depended strongly on the definition of plan acceptability. Under strict application of all target and OAR planning objectives, 234/250 adaptive fractions (93.6%) met every criterion, compared with 62/250 scheduled fractions (24.8%). Exploratory 1.6% offline dose renormalization increased scheduled-plan compliance to 135/250 (54.0%). When a 3% allowance was added to planning thresholds, compliance increased to 249/250 (99.6%) with adaptation, 185/250 (74.0%) with scheduled plans, and 197/250 (78.8%) after renormalization. The renormalization procedure was investigational and was not available as a clinical vendor-supported workflow.
- Adaptive treatment required additional workflow resources without demonstrated clinical outcome benefit. In a representative sample of 55 fractions, the complete adaptive session took approximately 23 minutes, including 7–8 minutes of contouring, around five minutes of independent QA, and repeat CBCT verification before delivery. Scheduled plans were evaluated under the institution's relatively relaxed daily bladder and rectal preparation requirements, which may differ from conventional nonadaptive SBRT practice. The study did not compare biochemical control, accumulated OAR toxicity, or patient-reported outcomes between adaptive and nonadaptive delivery strategies.
CLINICAL TAKEAWAY
CBCT-guided online adaptation improved dosimetric robustness in prostate SBRT, but most scheduled plans already achieved acceptable target coverage and organ-at-risk doses. The principal potential benefit appears to be the prevention of occasional large target-coverage deviations rather than systematic improvement of every fraction. Prospective research should establish whether selective adaptation can maintain safety while reducing treatment time and whether the dosimetric advantages translate into clinically meaningful outcomes.