Investigating the Dosimetric Impact of Acuros XB in Lung Cases Before Clinical Implementation

Background This study aims to present a simple and reliable way to study the dosimetric impact of the Acuros XB (AXB) algorithm compared to the Anisotropic Analytical Algorithm (AAA). Volumetric modulated arc therapy (VMAT) plans for lung cancer treatments were studied when changing the planning target volume (PTV) margins, positions, and interface distances. Methodology Three PTV positions were studied: near an air/tissue interface (PTVint), at the upper lobe surrounded by air (PTVair), and at the mediastinum (PTVtis). For the PTV near the interface, four VMAT plans were created, expanding the PTV margin and reducing the air/tissue interface distance at the same time. The margins were set to 0.5 cm, 0.7 cm, 1 cm, and 1.5 cm from the gross target volume (GTV). Differences in dose distributions between AXB and AAA occurred in all PTVs, with the maximum differences seen in PTVs including more air volume. Results For PTVtis, the two algorithms presented similar behavior. Specifically, for PTV mean dose (D_mean), AXB showed lower values of -2.3%, while for PTVair and PTVint, the mean corresponding differences were -18.2% and -10.1%. Concerning the mean lung doses (MLDs), AXB showed lower MLD than AAA for most PTVs, with differences ranging from -5.5% to -13.2%. For the expanding PTVs, MLD increased similarly for the two algorithms, i.e., 2.5 Gy to 4.9 Gy for AAA and 2.2 Gy to 4.4 Gy for AXB (for margins 0.5 cm to 1.5 cm). AAA overestimated the dose to the spine for most PTVs, with mean differences ranging from 4.7% to 15%. Conformity (CI) and Homogeneity indices (HI) also presented differences, with AAA plans showing higher PTV conformity, i.e., CI near unity, while CI values of AXB plans ranged from 0.27 to 0.94. Additionally, AAA plans were more homogeneous, with HI values ranging from 7.0 to 9.6, while HI values of AXB plans ranged from 19.1 to 24.9. Conclusions For lung cancer treatments, the PTV position, PTV margin, interface distance, and algorithm implemented can affect the dose calculation accuracy. Therefore, attention is required to avoid PTV underdosing or normal tissue toxicity. The simple-to-apply method presented here could be clinically applied and proven useful for that analysis. The two main limitations of the present study are the limited sample size and the fact that all plans were recalculated with AXB, keeping the same optimization objectives as AAA. Thus, the results should be interpreted with caution in clinical application.