ESTRO 2024 - Abstract Book

S3445

Physics - Dose calculation algorithms

ESTRO 2024

[4] Wasserthal, Jakob, et al. "Totalsegmentator: Robust segmentation of 104 anatomic structures in ct images." Radiology: Artificial Intelligence 5.5 (2023) (DOI: 10.1148/ryai.230024)

2790

Mini-Oral

Lymphopenia after brain radiotherapy: what is the role of lymphocyte irradiation? An in-silico model

François de Kermenguy 1,2 , Nathan Benzazon 1,2 , Pauline Maury 1,2 , Rémi Vauclin 3 , Meissane M’hamdi 1,2 , Vjona Cifliku 1,2 , Elaine Limkin 1,2 , Ibrahima Diallo 1,2 , Candice Milewski 1,2 , Céline Clémenson 1,2 , Michele Mondini 1,2 , Eric Deutsch 1,2 , Charlotte Robert 1,2 1 Gustave Roussy, Radiation Oncology, Villejuif, France. 2 INSERM, U1030, Villejuif, France. 3 TheraPanacea, Research department, Paris, France

Purpose/Objective:

Severe radiation-induced lymphopenia (sRIL) occurs in 40% of patients treated for primary brain tumors with radiochemotherapy and is an independent risk factor for inferior overall survival outcomes [1]. One hypothesis put forward to explain the occurrence of sRIL after brain irradiation is the cytotoxic doses received by the blood lymphocyte pool. Therefore, several research groups have been interested in developing models to evaluate the doses received by the blood lymphocytes [2], [3]. However, none has yet considered the out-of-field (OOF) dose to the head and neck (H&N) region rich in lymphoid structures, or the slow recirculation and homing of lymphocytes between lymphoid organs such as the spleen or lymph nodes and blood (which only contains about 5% of the total lymphocyte pool at any given time [4]). To tackle these shortcomings, this work proposes a new model to assess more accurately the doses received by recirculating lymphocytes during brain irradiation.

Material/Methods:

We propose a new model with highly efficient numerical implementation based on stochastic continuous time Markov chain process, consisting of 2 interconnected compartmental models: M 1 describes the slow recirculation of lymphocytes between lymphoid organs [5], while M 2 describes the bloodstream [6] (Fig. 1). Volumetric Modulated Arc Therapy (VMAT) treatment of 33 patients with primary brain tumors have been considered for dose computation. For each patient, an in-house deep learning neural network (NN), trained on a large dataset of more than 2200 whole body dose maps of patients from the French childhood cancer survivor study (FCCSS) cohort [7], was used to calculate the OOF dose to the lymph nodes of the H&N region that were automatically delineated by the ART-Plan software (Therapanacea). Three variants of the model (Fig. 1), corresponding to different physical and biological assumptions, were tested: H1) no recirculation of lymphocytes outside of blood; recirculation between several lymphoid organs H2) without and H3) with OOF dose to lymphoid structures of the H&N region. Simulations were performed with 10⁵ lymphocytes.

Results: