LET optimization reduces modeled biological dose to optic structures in sinonasal proton planning

In 16 sinonasal cancer planning cases, LET optimization reduced chiasm LET from 3.1 to 2.3 keV/μm while maintaining target coverage.

KEY POINTS

  • This retrospective in-silico study from Haukeland University Hospital, Norway, evaluated LET-guided intensity-modulated proton therapy (IMPT) planning for sinonasal cancer. Planning CTs and structure sets from 18 patients previously treated with photon radiotherapy were reviewed, with 16 patients included after excluding two whose critical structures did not receive sufficient dose to trigger LET optimization. Prescriptions ranged from 64 to 70 Gy(RBE 1.1) in 32–35 fractions.
  • For each patient, a conventionally robust-optimized IMPT reference plan was compared with a LET-optimized plan. Both strategies used the same beam geometry, physical dose objectives, 2-mm setup uncertainty, and 3.5% range uncertainty. LET optimization introduced objectives to reduce mean dose-averaged LET to <3 keV/μm in selected brainstem and optic structures receiving ≥30 Gy(RBE 1.1). Target coverage and normal-tissue doses were also assessed using five published variable-RBE models with α/β = 3 Gy.
  • LET optimization preserved target coverage and robustness in all 16 patients. Median CTV dose was 67.8 Gy(RBE 1.1) in reference plans versus 67.6 Gy(RBE 1.1) after optimization, without a statistically significant difference. Across 35 optimized critical structures, mean LET decreased by 0.1–1.5 keV/μm in 28, whereas seven structures experienced small increases of 0.1–0.3 keV/μm. Ten patients had successful optimization across all attempted configurations, while six had at least one unsuccessful attempt.
  • The largest consistent LET reduction occurred in the optic chiasm. Median mean LET decreased from 3.1 to 2.3 keV/μm (p=0.009). The right optic nerve showed a reduction from 3.1 to 2.5 keV/μm (p=0.016), while the left optic nerve had a statistically significant paired change (p=0.023) despite both group medians rounding to 2.7 keV/μm. In contrast, brainstem median LET remained 4.6 keV/μm (p=0.483), demonstrating that the optimization benefit was not uniform across critical structures.
  • Variable-RBE modeling demonstrated larger biological-dose differences than conventional fixed-RBE evaluation. Depending on the model, median near-maximum D0.03cm³ decreased by 0.86–4.04 Gy(RBEvar) in the chiasm, 1.85–4.62 Gy(RBEvar) in the left optic nerve, and 0.25–1.23 Gy(RBEvar) in the right optic nerve. Across optimized structures, 29/35 showed lower D0.03cm³ with all five variable-RBE models simultaneously. The models generally predicted 2.1–17.2% higher doses than fixed RBE 1.1, emphasizing substantial model dependence.
  • Several patient-specific plans crossed potentially relevant model-based dose thresholds after LET optimization. Examples included reductions from >56 to <54 Gy(RBEvar) for the brainstem, chiasm, or optic nerve under selected variable-RBE models. However, with fixed RBE 1.1, chiasm and optic nerve near-maximum doses did not show corresponding reductions and in some cases increased slightly. Ten critical structures in six patients remained above 54 Gy(RBE 1.1) in both strategies because target coverage was prioritized over organ sparing.
  • The dosimetric gains do not establish clinical normal-tissue protection. This was a planning comparison rather than a study of patients treated with LET-optimized protons, and no visual, neurological, or radiation-necrosis outcomes were measured. The 3 keV/μm LET objective and 30 Gy threshold were planning assumptions informed by earlier studies, not validated toxicity thresholds. Results also depended on five non-equivalent biological models, while the original cohort had been selected for photon treatment rather than proton therapy.

CLINICAL TAKEAWAY

LET-guided IMPT optimization can reduce dose-averaged LET and model-predicted biological dose in the optic apparatus for selected sinonasal cancer geometries without compromising robust target coverage. However, the benefit varies between patients, critical structures, and variable-RBE models, and conventional fixed-RBE dose metrics may not reflect these changes. Prospective clinical validation is required before such modeled reductions can be interpreted as lower rates of visual toxicity or other radiation-associated complications.

SOURCE

Physics and Imaging in Radiation Oncology

Browse more research Suggest a correction