2031 - Role of Whole-Heart Dose on Overall Survival (OS) in Locally-Advanced Non-Small Cell Lung Cancer (LA-NSCLC) Patients (Pts) Treated with Concurrent Definitive Chemoradiotherapy (CDCRT)

T. J. Dilling¹, A. Allen², O. Edrich³, D. Dudas⁴, D. Money⁵, G. Eichelberger⁵, A. Kumar⁵, J. M. Bryant¹, R. J. Cruz-Chamorro⁶, I. El Naqa⁷, E. Keit⁸, V. Feygelman⁶, and D. H. Lee⁹; ¹Moffitt Cancer Center, Department of Radiation Oncology, Tampa, FL, ²University of South Florida School of Medicine, Tampa, FL, ³Moffitt Cancer Center, Department of Machine Learning, Tampa, FL, ⁴Czech Technical University in Prague, Faculty of Nuclear Sciences and Physical Engineering, Prague, Czech Republic, ⁵Morsani College of Medicine, Dept of Internal Medicine, Tampa, FL, ⁶H. Lee Moffitt Cancer Center and Research Institute, Department of Radiation Oncology, Tampa, FL, ⁷H. Lee Moffitt Cancer Center and Research Institute, Department of Machine Learning, Tampa, FL, ⁸Moffitt Cancer Center, Tampa, FL, ⁹Morsani College of Medicine, Division of Cardiovascular Medicine, Tampa, FL

Purpose/Objective(s): In RTOG 0617, researchers found heart V5 and V30 predictive of OS. However, there were no mandated dose constraints on the heart, lungs, or esophagus. We report predictors of OS in LA-NSCLC pts treated with CDCRT at a multi-site tertiary cancer center, with special attention to whole-heart dose variables. Materials/

Methods: We conducted an analysis of LA-NSCLC pts treated with CDCRT (60+ Gy in 30+ fx) between 2009 and 2019. Excluded pts received sequential therapy or had prior lung resection or oligometastatic disease. Pts with prior lung SBRT were allowed if prior max heart dose was < 1000 cGy. We collected whole-heart dosimetry (every 5 Gy) as well as minimum/mean doses, D0.03cc, and gEUD. We performed univariable Cox proportional hazards testing (UVA) on relevant clinical/dosimetric variables, then developed a multi-variable model (MVA) from relevant features, including interaction terms among dosimetric parameters. The Benjamini-Hochberg procedure was used to correct for multiple comparisons (alpha = 0.05). Cutpoint analysis was performed on relevant dosimetric features.
Results: 418 patients met inclusion criteria. Median age was 67, 56% were male, 74% received IMRT, and 12% received durvalumab. 68% received 70 Gy. 2 pts were stage IB, 33 stage IIA/B, and 383 stages IIIA-C (AJCC 8^th ed). Median OS was 22.7 months. In UVA, age and PTV volume were most significant (p < 3e-6) and receipt of durvalumab (p = 0.009). Patients with smaller tumor diameter (p = 0.00168) and 3D-conformal treatment (p = 0.005) had improved OS. Individually, heart mean dose, gEUD, V5, ... V50 were all statistically significant. In MVA, only age, durvalumab, PTV volume, and whole-heart V30/gEUD remain significant (Table 1).
Conclusion: In LA-NSCLC CDCRT pts, only heart V30 and gEUD remain after correcting for age, PTV volume, and receipt of durvalumab. Optimal cutpoints for V30 and gEUD were 15.1% and 2713 cGy, respectively. More sophisticated analysis is necessary to determine which cardiac region(s) is/are relevant. Abstract 2031 – Table 1: Multi-Variable Analysis

Characteristic	HR1	p-value	q-value2
Age	1.03 (1.02 – 1.05)	< 0.001	< 0.001
Receipt of Durvalumab	0.55 (0.37 – 0.83)	0.004	0.014
PTV Volume	1.04 (1.01 – 1.04)	< 0.001	< 0.001
Heart V30	1.02 (1.01 – 1.03)	0.025	0.030
Heart gEUD	1.02 (1.01 – 1.03)	0.025	0.030
1 HR = Hazard Ratio 2 False discovery rate correction for multiple testing