KEY POINTS
- This retrospective Heidelberg Ion-Beam Therapy Center study evaluated children treated with proton therapy for low-grade glioma or ependymoma between 2010 and 2024. Of 132 eligible patients, 76 had sufficient radiological follow-up for imaging and dosimetric analysis, including 33 with low-grade glioma and 43 with ependymoma; investigators reviewed 724 follow-up MRI examinations.
- Radiation-induced contrast enhancement (RICE) was identified in 6 of 76 patients (7.9%), exclusively among patients with ependymoma (6/43; 14.0%). No RICE was observed among the 33 children with low-grade glioma, although the small sample and wide confidence intervals prevent definitive conclusions about histology-specific risk.
- Four patients developed small transient enhancement visible on a single follow-up MRI, while two had more persistent periventricular enhancement. All six patients demonstrated spontaneous lesion regression, illustrating that RICE can represent transient treatment-related imaging change rather than tumor progression or irreversible necrosis.
- Higher prescribed proton dose was associated with RICE development (mean 58.5 Gy in affected patients versus 55.3 Gy without RICE; p=0.031). However, the ependymoma subgroup also received a higher median prescription dose than the low-grade glioma subgroup (59.4 vs 54 Gy; p<0.001), complicating separation of tumor-type and dose effects.
- Across eight delineated RICE regions in six patients, mean dose-averaged LET was higher in RICE regions than in corresponding PTVs (4.44 vs 2.92 keV/µm). In contrast, mean conventional RBE-weighted dose calculated with fixed RBE 1.1 was 53.21 Gy(RBE) in RICE regions versus 56.42 Gy(RBE) in PTVs, indicating that conventional dose alone did not explain the observed spatial pattern.
- Combined Dose(RBE) × LET metrics were elevated within RICE regions, particularly at low- and intermediate-dose levels. Mean values were 203.64 in RICE regions versus 164.66 in the corresponding PTVs, while D95-related values were 172.76 versus 112.54, supporting a potential contribution from LET-associated biological dose heterogeneity.
- All three investigated variable RBE models, Wedenberg, McNamara, and Carabe, identified biologically weighted dose differences not fully represented by fixed RBE 1.1 calculations. In the broader 132-patient survival cohort, five-year OS was 91.7% for low-grade glioma and 86.6% for ependymoma; however, only six RICE events and the absence of an unaffected-tissue risk comparison limit validation of LET-based toxicity predictors.
CLINICAL TAKEAWAY
These findings add pediatric clinical evidence that proton-associated brain imaging changes may occur preferentially in regions of elevated LET and increased modeled biological effectiveness, even when conventional RBE-weighted dose is not unusually high. However, the small number of RICE events and uncertainty surrounding variable RBE models mean that the results support further biologically informed planning research rather than immediate changes to clinical dose constraints.