KEY POINTS
- The retrospective cohort included 79 patients with skull-base chordoma (n=61) or chondrosarcoma (n=18) treated using pencil-beam scanning proton therapy between 2011 and 2020.
- Chordomas received 74 Gy(RBE) and chondrosarcomas 70 Gy(RBE), calculated with a constant RBE of 1.1. Temporal-lobe D2cm³ was constrained below 71 Gy(RBE) during clinical optimization.
- After a median follow-up of 49.7 months, 13 patients developed grade ≥2 temporal lobe necrosis involving 17 lobes, with a median onset of 21.5 months.
- After matching necrotic and healthy-brain voxels by absorbed dose, seven of 13 affected patients showed a significant LET difference. In five of those seven patients (71.4%), necrotic voxels had higher dose-averaged LET.
- Higher LET was associated with necrosis in the 35–40 Gy(RBE) dose range (OR 7.4, 95% CI 3.5–26.5; p<0.001) and the 50–55 Gy(RBE) range (OR 2.4, 95% CI 1.5–4.0; p<0.001).
- The final model combined the temporal-lobe volume receiving at least 68 Gy(RBE) with the volume receiving at least 19 Gy(RBE) and LET of at least 4.6 keV/µm. Both variables remained independently associated with necrosis.
- Internal ten-fold cross-validation produced an AUROC of 0.89 (95% CI 0.78–0.95), with apparently good calibration. The model was developed from only 17 affected lobes, and no independent cohort or prospective LET-guided planning evaluation was performed.
CLINICAL TAKEAWAY
The findings support evaluating LET alongside physical dose when high-dose skull-base proton fields approach the temporal lobes. The proposed constraints are not ready for clinical adoption; external validation and prospective testing are needed before they can guide beam arrangement or optimization.