Impact of Volume and Image Strategy on Dose Prescription Accuracy in Lung Stereotactic Body Radiation Therapy

Despite decades of use, multiple prescription strategies continue to coexist in lung stereotactic body radiation therapy (SBRT). An accurate prescription should reflect the dose delivered throughout the breathing cycle with minimal bias and high precision. This study aimed to evaluate the accuracy of widely used lung SBRT dose prescription methods by assessing which method most accurately represents the true delivered dose to the gross tumor volume (GTV), considering absolute terms (trueness) and consistency across patients (precision).

Fifty lesions were analyzed using combinations of computed tomography (CT) types (single phase, average intensity projection, maximum intensity projection [MIP]), target volumes (GTV, internal target volume, planning target volume [PTV]), dose prescription metrics (D_2%, D_50%, mean, D_95%, D_98%) and dose calculation algorithms (collapsed cone, Monte Carlo, and type A pencil beam). Delivered doses were compared against a 4-dimensional CT-based reference to assess trueness and precision across different prescription methods, including respiratory movement.

Combinations involving internal target volume or GTV with single-phase CT or average CT, and based on D_50%, mean dose, or D_2%, showed biases of less than 2% and precision better than 1.5% when compared to the delivered 4-dimensional GTV dose. This was also valid for D_2% PTV prescriptions. PTV MIP-based prescriptions showed surprising accuracy with primarily a constant 5.3% bias correction and 1.5% precision. However, MIP-based D_2% and D_98% were no longer coherent. Historical type A dose calculations resulted in low accuracy for any prescription method due to algorithm limitations. However, even with current type B/C algorithms, prescriptions based on 98%/95% PTV coverage on any CT still exhibited low precision, approaching that observed with type A calculations. This results in substantial interpatient variability in GTV dose, due to a linear bias correlated with PTV density, reflecting multiple underlying physical factors.

The results of this study can be used not only to improve consistency in lung SBRT prescriptions but also to provide correction factors and weightings to improve accuracy in meta-analyses and dose-response modeling.