2342 - Post Radiotherapy Lymphocyte Dynamics and Dose Constraints in Locally Advanced Thoracic Malignancies Undergoing Radical Radiotherapy
Presenter(s)
B. M. F. Cheung1, K. K. Yuen1, M. Y. Luk1, D. K. Leung1, and V. H. F. Lee2; 1Department of Clinical Oncology, Queen Mary Hospital, Hong Kong, Hong Kong, 2Department of Clinical Oncology, School of Clinical Medicine, LKS Faculty of Medicine, The University of Hong Kong, Hong Kong, China
Purpose/Objective(s): Radiotherapy-induced lymphopenia (RIL) worsens outcome in thoracic malignancies on radiotherapy(RT). We hypothesised that post-RT lymphocyte nadir and recovery are prognostic for overall survival (OS) in both non-small cell lung cancer (LC) and thoracic esophageal cancer (EC) undergoing radical chemoRT. We also aim to establish and validate robust disease agnostic organ at risk (OAR) constraints for RIL.
Materials/Methods: Two independent cohorts with 191 stage 3 LC patients receiving chemoRT and 144 stage 2-3 EC patients receiving chemoRT from 2013-2022 were retrospectively analysed. Absolute lymphocyte count (ALC) was obtained within 1 week before RT and monitored during and after RT. ALC nadir (AN) was defined as the lowest ALC on RT. ALC recovery index (ARI) was the ALC at 6 months/Pre-RT ALC. OARs including heart, lungs, thoracic vertebra (T spine), sternum, ribs, spleen and great vessels (vena cavae, pulmonary vessels and aorta) were delineated using deep learning assisted algorithm. OAR doses were adjusted to equivalent dose in 2Gy fractions (EQD2). ARI cut off was selected using maximally selected rank statistics by OS. AN cut off was 0.5 x 109 per Common Terminology Criteria for Adverse Events (CTCAE) for grade 3 lymphopenia. Effect of AN and ARI on overall survival (OS) were evaluated by Kaplan-Meier and Cox proportional methods. EQD2 of OARs was correlated with AN and ARI using logistics regression. OAR thresholds were evaluated using receiver operating characteristic curves and selected based on Youden’s index.
Results: Optimal ARI cut off was 65% of baseline ALC. High ARI was associated with improved OS in both cohorts (LC: mOS 71.5m vs 18.4m p= 2.9x10-14; ES: mOS 59.8m vs 10.9m p= 3.4x10-18). On multivariate analysis, high ARI and AN were independently associated with improved OS for both cohorts (ARI: LC HR = 0.81, p= 2.5 x 10-9, EC HR = 0.71, p= 5.1 x 10-13 ; AN: LC HR = 0.8, p = 0.0014, EC HR = 0.7, p=8.3x10-6). In LC, T spine V35Gy and ribs V35Gy were both associated with low ARI (T spine: OR=1.05, p=0.004; Ribs OR = 1.08, p=0.01). Optimal cut off for T spine V35Gy was 25.5%. Ribs V35Gy cut off was 6.7%. T spine V20Gy and Mean lung dose (MLD) were associated with low AN in LC (T spine: OR=1.04, p=0.03; MLD: OR 1.15, p=0.007). MLD cut off was 16.5Gy while T spine V20Gy cut off was 25.6%. For EC, T spine V35Gy was associated with low ARI (OR = 1.025, p=0.005). Optimal cut off was 25.7%, similar to that of LC. Meanwhile, MLD and spleen V5Gy were associated with low AN (MLD OR=1.13, p=8.48x10-7 ; Spleen OR =1.03, p=0.03). MLD cut off was 10.6 Gy. Spleen V5Gy cut off was 7.2%.
Conclusion: Both AN and ARI were independent predictors of OS in two independent cohorts of LC and EC. T spine and MLD were validated to be crucial OARs for ARI and AN with robust disease agnostic dose constraints. Ribs and spleen are disease specific RIL OARs for LC and EC respectively, potentially due to tumor location. These constraints may guide future trials on lymphocyte sparing RT in thoracic malignancies.