ESTRO 2024 - Abstract Book

S785

Clinical - CNS

ESTRO 2024

748

Mini-Oral

External validation of a prediction model for radionecrosis after radiosurgery in brain metastases

Elise EMW van Schaik 1 , Jeroen A Crouzen 1 , Mirjam E Mast 1 , Mandy Kiderlen 1 , Noëlle CMG van der Voort van Zyp 1 , Anna L Petoukhova 2 , Jaap D Zindler 1 1 Haaglanden Medical Center, Radiotherapy, Leidschendam, Netherlands. 2 Haaglanden Medical Center, Medical Physics, Leidschendam, Netherlands

Purpose/Objective:

Stereotactic radiotherapy (SRT) is a frequently used treatment option for patients with brain metastases (BM). The most deleterious complication of SRT is radionecrosis (RN), which typically develops after 6-24 months in 5-40% of patients. Despite the clinical importance of RN, diagnosing this condition remains challenging due to a lack of a widely accepted definition. In 2021, a normal tissue complication probability (NTCP) model was established to predict the RN risk in patients with BM after SRT [1]. The V12, which is the volume of normal brain tissue excluding the gross tumour volume receiving a dose of ≥12 Gy, acts as a risk factor in the prediction model on a continuous scale. The objective of this study is to externally validate this NTCP model with two different definitions of RN.

Material/Methods:

Data of 317 patients treated with SRT for solitary BM between 2010 and 2022 was retrospectively reviewed. Patients with leptomeningeal metastases at baseline or a medical history of treatment for BM (previous whole brain irradiation, SRT or resection) were excluded. The prescribed dose was 15/16/18/21Gy in a single fraction or 24/25.5Gy in three fractions. Overall survival (OS) was determined with the Kaplan-Meier method. Two definitions of RN were used to evaluate the NTCP model’s performance, based on radiological information consisting of diagnostic magnetic resonance imaging (MRI) scans and according to the Response Assessment in Neuro-oncology Brain Metastases (RANO-BM) working group [2]. The first one was defined as radiological progression in comparison to the smallest tumour diameter during or after treatment (nadir), both asymptomatic and symptomatic. The second one was described as radiological progression in comparison to the tumour diameter at baseline (baseline), regardless of any symptoms. The predicted and observed RN rates were compared for both the single and multifraction group.

Results:

In total, 231 patients were analysed with a median follow-up of 73 months (95% CI 33-112). Of these, 162 patients were treated with a single fraction and 69 patients with multiple fractions. The definition of RN based on ‘nadir’ suited the NTCP model best. This applied for both the single fraction (Figure 1) as well as the multifraction group (not shown). According to this definition, 55 (24%) patients developed RN with an actuarial risk of 43% and 60% at 1 and 2 years, respectively. For the second definition of RN, diagnosed based on ‘baseline’, the NTCP model showed an overestimation of the predicted RN probability compared to the observed RN probability (Figure 2). According to