Clinical Implications of LET-to-Water Versus LET-to-Medium Calculations In Heterogeneous Head-and-Neck Proton Therapy Scenarios
Abstract
Purpose
Dose-averaged linear energy transfer (LETd) is increasingly used to characterize biological risk in proton therapy. However, LET definitions differ among treatment planning systems. RayStation 2023B reports LET-to-water, whereas Monte Carlo engines such as MCsquare compute LET-to-medium. The clinical impact of these differences has not been systematically evaluated. This study investigates LET discrepancies between these two approaches in clinically relevant heterogeneous head-and-neck scenarios.
Methods
A Sun Nuclear anthropomorphic head phantom was CT-scanned and imported into RayStation and MCsquare. Three single-field proton plans were generated to isolate LET behavior in (1) mandibular bone, (2) soft tissue adjacent to an air cavity, and (3) tissue adjacent to a titanium dental prosthesis. Identical beam geometry, energy modulation, and range settings were applied across both platforms. LETd distributions were extracted along beam paths and within region-of-interest (ROI) volumes at both proximal and distal regions. Comparisons focused on absolute LETd values, relative differences between LET-to-water and LET-to-medium, and spatial patterns near material interfaces.
Results
While LETd differences remained within 10% in homogeneous soft tissue, substantial discrepancies emerged at heterogeneity interfaces. In the mandibular ROI, LET-to-medium values were consistently higher than LET-to-water, with mean differences of approximately 20–30%. Near titanium prostheses, LETd discrepancies increased further, with a mean difference of 28.1% and peak differences reaching 32.5% at the proximal edge. In contrast, LETd deviations near air cavities were lower, in contrast, with a mean difference of approximately 10-20% and lower discrepancies at distal re-buildup regions.
Conclusion
LET-to-water and LET-to-medium calculations exhibit non-negligible differences in heterogeneous head-and-neck conditions, particularly in high-density materials such as bone and titanium. These differences may have implications for LET-based biological modeling and toxicity assessment. Caution is warranted when interpreting LET-derived metrics across different calculation frameworks, and standardization or correction strategies may be necessary for consistent clinical application.