Voxel-level linear energy transfer-driven dose conversion factors between LEM- and MKM-based treatment plans in...

Carbon-ion radiotherapy employs distinct relative biological effectiveness (RBE) models-the local effect model (LEM) and the microdosimetric kinetic model (MKM)-which differ in their dependencies on dose-averaged linear energy transfer (LETd) and absorbed dose. Conversion factors (CFs) between these models are essential for translating clinical outcomes across institutions; however, previous studies have neglected LETd-driven spatial heterogeneity.

This study aimed to quantify the impact of LETd on CFs by establishing a voxel-level framework for iso-effective dose conversion and validating its clinical applicability.

RayStation-optimized plans included cubic and spherical targets, as well as clinical cases of prostate and lung cancer. CFs were derived by comparing LEM- and MKM-based RBE-weighted doses. A conversion model correlating LETd with MKM/LEM parameters ( $Z_{1D}^{\ast }$ , $RB{E}_{max}$ ) was developed using cubic plans (1-12 cm modulation depths) and validated against RayStation calculations. Spatial CF variations were analyzed across dose levels (≤10 Gy[RBE]) and LETd values (≤210.4 keV/µm).

CFs showed primary dependence on dose and LETd. Smaller targets (2-cm cubic) demonstrated higher LETd (83.0 keV/µm) compared with larger targets (12-cm cubic, 38.4 keV/µm) and lower CFs (1.03 vs 1.13 at 4 Gy [RBE]). In cube plan targets (e.g., a 6-cm cube at 7 cm depth), the model predicted voxel-level doses with <2% error. LETd variations (36-100 keV/µm) introduced approximately 5% CF uncertainty. Clinical validations showed ≤0.5% dose discrepancies in targets but revealed LETd-driven heterogeneity in Bragg peaks.

LETd-induced spatial CF variability necessitates voxel-level analysis for precise dose conversion. The proposed framework supports LETd-aware treatment planning and cross-model interoperability, although clinical adoption will require tumor-specific CFs and validation in complex anatomies.