Modelling Dose and Dose-Averaged Linear Energy Transfer to Predict High-Grade Temporal Lobe Necrosis Following Skull-Base Proton Therapy

Giulia Fontana et al.

This retrospective study investigated whether combining physical dose and dose-averaged linear energy transfer (LETd) could improve prediction of high-grade temporal lobe necrosis (G2-TLN) after proton therapy (PT) for skull-base tumors. The cohort included 79 patients with chordomas (n=61) and chondrosarcomas (n=18) treated at the CNAO National Center for Oncological Hadrontherapy between 2011 and 2020, with prescribed doses of 74 and 70 Gy(RBE), respectively. A constant RBE of 1.1 was used, while temporal lobe dose was constrained to D2cc <71 Gy(RBE). Dose-LETd-volume histograms (DLVH) were generated, and voxel-wise as well as structure-wise analyses were performed to identify parameters associated with CTCAE grade ≥2 TLN. Higher LETd was significantly associated with necrosis in voxels receiving 35–40 Gy(RBE) and 50–55 Gy(RBE). A logistic regression model identified two independent DLVH predictors: the temporal lobe volume receiving 68 Gy(RBE) and the volume receiving ≥19 Gy(RBE) with LETd ≥4.6 keV/µm. The model demonstrated good discrimination, with a cross-validated AUROC of 0.89.

In conclusions, the study demonstrates that LETd provides clinically relevant information beyond physical dose for predicting high-grade TLN after skull-base PT. Incorporating dose and LETd into treatment-plan evaluation may improve normal-brain risk assessment and proton-plan optimization. However, larger prospective cohorts and external validation are required before clinical implementation.

Published by Medical Physics 2026.

https://doi.org/10.1002/mp.70562