2942 - Comparison of BED-Based Scheduled Plan Scaling Versus Fully Adaptive Re-Optimization in Abdominal SBRT

Purpose/Objective(s):

Online adaptive radiotherapy typically relies on daily re-optimization to account for anatomical variations and to maintain target coverage while respecting organ-at-risk (OAR) constraints. However, full re-optimization is computationally and operationally demanding. This study investigates an alternative adaptive strategy in which the reference treatment plan is recalculated on daily anatomy and renormalized using biologically effective dose (BED)–based objectives, without performing full daily re-optimization, then the Target and OAR BED’s are compared to BED’s calculated from adaptive treatment.

Materials/Methods:

A cohort of eight patients undergoing abdominal adaptive stereotactic body radiotherapy (SBRT), each treated with five fractions (totaling 40 fractions), was retrospectively analyzed. For each fraction, two plans were available: (1) an adaptive plan generated on the daily anatomy and (2) a scheduled plan recalculated on the same daily anatomy without re-optimization. The biologically effective dose (BED) for the planning target volume (PTV) and organs at risk (OARs) was computed using the linear–quadratic (LQ) model. Then the scheduled dose distribution was uniformly scaled by a fraction-specific factor S_k.. S_k was determined by maximizing the following objective function:

F(S_k)= BED_PTV,k(S_k)-?_i?OAR ?_iBED_i,k(S_k)

where ?_i?[0,1] denotes structure-specific weighting coefficients.

The scaling factor S_k was optimized independently for each fraction to maximize this objective, subject to following constraints:

1. Hard OAR constraints defined in form of OAR specific physical dose Limits
2. Bounds were imposed on the scaling factor to restrict allowable lower limit for target BED such that scaled target BED is no lower than Adaptive Target BED unless it violates the first constraint.
3. Finaly, constrain on scaling factor was also imposed such that if constrains 1 and 2 are met, cumulative Oar value for scaled Scheduled plan is lower than Adaptive Cumulative Oar Value.

In addition, no cumulative inter-fraction optimization was enforced. Clinically delivered fully optimized adaptive plans were used as the reference standard. Optimization was performed using constrained nonlinear programming.

Results:

After scaling of the scheduled plan using the optimized scaling factor, eight cases were evaluated. In one case, the scaled scheduled plan achieved a 29% reduction in OAR BED and a 30% increase in target BED compared to the adaptive optimized session. Although, in this case, the scaled scheduled plan demonstrated a 15% higher OAR BED compared to the original scheduled plan, it resulted in a 32% higher target BED, representing a favorable therapeutic trade-off.

Conclusion:

In number of cases, normalization-based optimization of the scheduled plan produced superior cumulative target BED while maintaining comparable or lower cumulative OAR BED compared with the adaptive plan.