KEY POINTS
- This retrospective, single-centre pilot included 20 patients with World Health Organization grade 2 low-grade glioma treated with actively scanned proton therapy: 10 developed contrast-enhancing brain lesions and 10 remained lesion-free.
- Prescribed doses ranged from 54–60 Gy relative biological effectiveness in 27–30 fractions. Longitudinal apparent diffusion coefficient metrics were extracted from high- and low-risk brain regions defined by the Probability of Lesion Origin model.
- In patients who developed lesions, the apparent diffusion coefficient exceeded by 20% of voxels was higher in high-risk than low-risk tissue: 1,699 versus 1,373 µm²/s (p=0.02). No corresponding difference was observed in lesion-free patients (p=0.91).
- The high-risk minus low-risk diffusion difference separated lesion-positive from lesion-free patients across follow-up (p=0.02), with the strongest discrimination at six months (area under the curve 0.79), approximately 12 months before lesion detection.
- The endpoint was radiological lesion development rather than clinical toxicity. Reported discrimination was exploratory, with no correction for multiple testing, internal validation, or external validation.
CLINICAL TAKEAWAY
Risk-model-guided diffusion imaging may help identify subclinical radiation-induced brain changes before contrast enhancement appears after proton therapy for low-grade glioma. The signal is potentially useful for surveillance, but the very small selected cohort and unvalidated analysis make it hypothesis-generating rather than clinically actionable.