Title:	Simulation of pediatric CSI proton treatments
DNr:	NAISS 2025/22-1411
Project Type:	NAISS Small Compute
Principal Investigator:	Maria Teresa Romero-Exposito <maria.teresa.romeroexposito@ki.se>
Affiliation:	Karolinska Institutet
Duration:	2026-01-01 – 2027-01-01
Classification:	10301
Keywords:

Abstract

Improvements in oncological treatments have extended the lifetime expectancy of patients after treatment, making the long-term side effects gain greater relevance. Among these, a notable concern is the development of second cancers, particularly those induced by radiation therapy. While radiation-induced second cancers (RISC) are rare in adults, children warrant a detailed study due to their longer lifetime expectancy and higher radiosensitivity. Following leukemia, paediatric cancers commonly manifest in the brain and central nervous system. Medulloblastoma is one of the most prevalent and usually requires a craniospinal irradiation (CSI), which covers a high volume of the patient. Proton therapy (PT) is often recommended for such cases as it minimizes healthy tissue exposure compared to photon irradiation. However, studies on dose distribution, particularly concerning neutron contamination, are scarce, especially in facilities offering scanned beam (SB) technology, which is increasingly becoming the standard. Thus, this project aims to assess the total dose, including neutron exposure, received by paediatric patients under CSI in a SB facility. For that purpose, a Monte Carlo model of the clinical proton beam and the neutron contamination will be created. Subsequently, this model will be used to simulate the PT plans used in a population of selected patients. From these simulations, total organ dose distributions will be evaluated, along with the estimation of the RISC risk using existing dose-risk models. This study will establish the dose and risk ranges associated with the treatments requiring CSI. Moreover, the dosimetric data obtained will enhance future epidemiological studies, improving dose-response models.