ESTRO 2021 Abstract Book

S104

ESTRO 2021

Neurology, Groningen, The Netherlands

Purpose or Objective In The Netherlands, patients with low grade glioma (LGG) are selected for proton therapy (PT) based on tumour and patient characteristics. Additionally, significant dose reduction with PT in relevant brain volumes is required. To this end, a plan comparison (protons vs. photons) is performed for each patient. The purpose of this study was to investigate which LGG patients have the most benefit from PT in terms of dose reduction in brain volumes. Materials and Methods All patients with LGG treated with PT at our centre were included in the study. Prospectively registered characteristics of the cohort were extracted, as well as the mean dose (Dmean) to brain volumes: cerebrum, hippocampus, and the ipsi- and contra-lateral side of these volumes regarding tumour laterality. The Dmean of the PT and photon plan were compared using Wilcoxon Signed Rank Test. The Dmean difference of the PT and photon plan (Δdose) were calculated. The influence of tumour volume (CTV) and tumour location/extension on the magnitude of Δdose was analysed using Spearman correlation and Mann Whitney U Test. Subsequently, multivariable linear regression analysis was performed to predict Δdose. Results From 2018 until Jan 2021, 82 patients were treated with PT. At the time of treatment, the mean age was 40 years (range 19 - 72) and all patients were in a good general condition: WHO-PS grade 0-1, and iADL independent. Forty patients (49%) had oligodendroglioma WHO grade II/III and 38 (47%) IDH-mutated astrocytoma grade II/III, and 4 (4%) other histology. The mean CTV was 110 cc (range 23.54 – 404.79). Sixty-three (77%) tumours were located or extended in the frontal lobe, 28 (34%) in the temporal lobe, and 24 (29%) extended across the midline. The Dmean in all brain volumes was significantly lower for the PT plans (Table 1). Larger CTV correlated with larger Δdose of all brain volumes except for the ipsilateral hippocampus. The strongest association with Δdose was found for the contralateral cerebrum (R 0.643, p ˂ 0.001). The magnitude of Δdose was significantly related with tumour location/extension in several brain regions, with either positive or negative impact (Table 2). Prediction of Δdose (in cGy) was most accurate for the contralateral hippocampus (R = 0.671, R 2 = 0.451, p <0.001). The model included the variables tumour extension in: frontal-basal region (B = 486); hippocampus (B = 357) and throughout multiple lobes (B = 190) and a constant of 276.