Dose to Heart, Spine, Aorta, and Body Predict for Severe Lymphopenia and Poor Survival in Patients Undergoing Chemoradiation for Esophageal Cancer

Abstract

Esophageal cancer (EC) patients (pts) undergoing chemoradiation (CRT) may be at increased risk of lymphopenia due to unintentional dose to immune-related organs at risk. We investigated the relationship between grade 4 lymphopenia (G4L) and recurrence-free (RFS) and overall survival (OS) and identified dosimetric constraints that are associated with G4L. All pts who received CRT for stage I-III EC and had accessible total lymphocyte counts (TLC) at a single institution from 2003-2018 were identified. G4L was defined using CTCAE 4.0 criteria (< 200/mm3) during CRT. RFS and OS were evaluated with Kaplan-Meier methods, log rank tests, and Cox proportional hazards regression. We defined optimal DVH cutpoints for G4L using ROC curves, and these were compared across G4L groups with univariate (UVA) and multivariable (MVA) logistic regression. 189 pts were included with a median follow-up of 2.3 years. Demographics included: median age 65 (range 35-84), 84% male, 92% white, 20% never smoker, 31% no alcohol use. Most pts had clinical stage II (34%) or III (60%) disease, distal esophageal/GE junction location (85%), adenocarcinoma (78%), and poorly differentiated tumors (40%). Median RT dose was 50.4 Gy (range 41.4-70.2) delivered with carboplatin/paclitaxel (55%) or 5FU based regimens (40%). 68% of pts underwent surgical resection following CRT. G4L was identified in 85 pts (45%). G4L was associated with lower median RFS (1.8 vs 4.0 years, p=0.02) and lower median OS (2.4 vs 4.0 years, p=0.03). On MVA for OS, stage (III vs I/II, HR 1.79 [1.17, 2.74]), surgery (HR 0.40 [0.24, 0.66]), and G4L (HR 1.55 [1.00, 2.41]) were significant. Dosimetric data were available for 119 pts. On UVA, Heart V15 > 73%, T-spine V5 > 72%, Body V10 > 18%, Total Lung V5 > 50%, and Aorta V5 > 93% were strong predictors of G4L (Table 1, all p<0.05). In multivariable models including baseline TLC and other clinical factors, exceeding DVH cutpoints for heart, T-spine, aorta, and body predicted for G4L (all p<0.05). When included separately in MVA for OS, exceeding cutpoints for heart (HR 2.22 [1.17, 4.24]), T-spine (HR 2.38 [1.15, 4.92]), and aorta (HR 2.20 [1.04, 4.66]) independently predicted for worse OS. Dose to the heart, spine, aorta, and body predict for G4L, which is associated with worse RFS and OS. Achieving planning constraints for these parameters may decrease G4L and improve outcomes.Abstract 2475; Table 1Odds Ratio (95% CI) for G4 Lymphopenia, *p<0.05, **p<0.01.VariableUVAMVA1Baseline TLC (100 lymphocyte increase)1.06* (1.01, 1.11)0.90-0.93* (0.83-0.86, 0.97-0.99)Heart V15 > 73%2.87** (1.33, 6.18)2.39* (1.05, 5.46)T-spine V5 > 72%2.80** (1.32, 5.93)2.76* (1.22, 6.28)Body V10 > 18%7.21* (1.57, 33.16)10.59* (1.28, 87.64)Total Lung V5 > 50%2.50* (1.18, 5.30)2.09 (0.94, 4.65)Aorta V5 > 93%2.71* (1.04, 7.09)3.49* (1.14, 10.66)Spleen V20 > 45%2.06 (0.984, 4.295)1.75 (0.785, 3.92)1 Separate models including baseline TLC, smoking (NS), alcohol hx (NS), and each individual DVH cutpoint. Open table in a new tab