ESTRO 2024 - Abstract Book

S902

Clinical - CNS

ESTRO 2024

87 patients with WHO grade 2 or 3 gliomas treated with proton therapy from January 2019 to December 2021 were included (median age 41.3 and range 19 to 67 years). 42 of the patients had oligodendrogliomas and received adjuvant PCV (procarbazine, lomustine and vincristine). The remaining patients had astrocytomas and received adjuvant temozolomide. The prescribed dose was 50.4 to 60 GyRBE. All patients were treated with 3-4 beams using PBS and multi-field optimisation. The PVZ was defined as a 4 mm wall around the lateral ventricles for all patients on the treatment planning MRI scans. Follow-up MR scans were acquired in the referring region and evaluated by an experienced neuro-radiologist. The lesion was biopsied, whenever possible and only patients with necrosis or RT-induced changes were included. MRI scans with the first sign of contrast-enhanced lesions outside the GTV were registered to the treatment plan CT scans and used for the analysis. Monte Carlo calculations 1 of dose and LET were performed for all treatment plans. Dose vs. LET plots were performed according to Niemerko et al. 2 Inter-class correlation coefficient (ICC) was calculated to test for inter-patient variation 3 . If ICC>0.05 a model including patient specific variables was deemed necessary and a two-level mixed-effect model should be used. Patient-specific variables considered in this analysis were histology (oligodendroglioma or astrocytoma), sex and age. 13 patients (15%) had ICs on the follow-up scans. One patient with IC was lost to imaging follow-up. Three of the 13 patients had symptoms that required medication. The incidence of IC in high- dose patients (≥ 54 GyRBE prescribed) was similar to that of low-dose patients (<54 GyRBE), but symptoms were only observed in patients receiving high dose. The inter-patient variation was substantial (ICC=0.16). A two-level mixed-effect model was fitted to the voxel data, from the 12 patients, to account for the inter-patient variation. Only the patient variable oligodendroglioma/astrocytoma improved the model. The model with the lowest Akaike score included intercept and LET as random effects and PVZ, dose, LET and oligodendroglioma as fixed effects. The model coefficients are given in Table 1. The model confirmed the association between IC high dose, high LET and proximity to the ventricles PVZ (OR=7.96). Furthermore, the incidence was higher for oligodendroglioma patients compared to astrocytoma patients (OR=2.78). Results: