ESTRO 2024 - Abstract Book

S5006

Physics - Radiomics, functional and biological imaging and outcome prediction

ESTRO 2024

5. Winkel D, Gijsbert HB, Kroon PS, Asselen BV, Hackett SS, Werensteijn-Honingh AM et al. Adaptive radiotherapy: The Elekta Unity MR-linac concept. Clinical and Translational Radiation Oncology. 2019; 18: 54-59.

950

Digital Poster

Prediction of radiation induced lymphopenia during chemoradiation therapy for lung cancer patients

Serena Monti 1 , Giuseppe Palma 2 , Radhe Mohan 3 , Ting Xu 4 , Zhongxing Liao 4 , Laura Cella 1

1 National Research Council, Institute of Biostructures and Bioimaging, Napoli, Italy. 2 National Research Council, Institute of Nanotechnology, Lecce, Italy. 3 The University of Texas MD Anderson Cancer Center, Department of Radiation Physics, Houston, USA. 4 The University of Texas MD Anderson Cancer Center, Department of Radiation Oncology, Houston, USA

Purpose/Objective:

Radiation-induced lymphopenia (RIL) is a frequent and significant side effect associated with radiation therapy (RT), with prognostic implications for many tumor types. RIL is influenced by individual patient factors, specific RT plan related factors, and dose to multiple organs involved in the immune system. Our study aims to develop a normal tissue complication probability (NTCP) model for severe RIL based on dose to anatomical regions that have shown correlation with lymphocyte depletion in past voxel-based analysis (VBA) studies [1-3]. This study was conducted in patients with locally advanced Non-Small-Cell Lung Cancer (NSCLC) who underwent concurrent chemotherapy and radiation therapy (RT) [4].

Material/Methods:

We retrospectively analyzed NSCLC patients treated with Intensity Modulated Radiation Therapy (IMRT) or Passive Scattering Proton Therapy (PSPT) to a prescribed dose of 66/74 Gy in conventional 2 Gy daily fractionation with concurrent chemotherapy. For patients included in the study, pre-RT and weekly during-RT complete-blood-counts were obtained. According to CTCAE v. 5.0 definition, severe (≥ grade 4) RIL was defined as absolute lymphocyte count (ALC) at nadir (ALCnadir) <0.2*109 cells/l during RT. Based on VBA findings, dose-volume histograms (DVHs) of the following organs-at-risk (OARs) were extracted: heart, healthy lungs, thymic lodge and bones (i.e., bone marrow, sternum plus vertebrae). The thymic lodge structure was identified on a common anatomical reference (XCAT phantom) and then its contour was propagated from the phantom to each patient native space (i.e., planning CT) by applying an appropriate deformation field, computed to match the two anatomies. The bone and bone marrow structures were automatically segmented on each patient planning CT using a combination of HU thresholding and morphological operation. A NTCP logistic model for G4 RIL was developed based on OAR-DVH metrics and non dosimetric variables (age, sex, RT modality, smoking history, ALC at baseline, adjuvant chemotherapy and induction chemotherapy). The multivariable stepwise logistic regression method and the Leave-One-Out (LOO) cross validation method were applied. The area under the receiver operating characteristic (ROC) curve (AUC) and calibration plots were used to evaluate the model performance.