Vienna, Austria

ESTRO 2023

Session Item

May 14
10:30 - 11:30
Stolz 1
Corinne Faivre-Finn, United Kingdom;
Daniel Zips, Germany
Post-radiotherapy G3+ esophageal toxicity predictors: Pooled data from 2,131 lung cancer patients
Azadeh Abravan, United Kingdom


Post-radiotherapy G3+ esophageal toxicity predictors: Pooled data from 2,131 lung cancer patients

Azadeh Abravan1, Hitesh Mistry2, Alan Mcwilliam1, Corinne Faivre-Finn1, Ana Vega3, Jose López-Guerra4, Dirk de Ruysscher5, Catharine West6, Ahmed Salem7

1The University of Manchester and The Christie Hospital, Division of Cancer Sciences, School of Medical Sciences, Faculty of Biology, Medicine and Health and Department of Radiotherapy Related Research, MANCHESTER, United Kingdom; 2The University of Manchester, Division of Cancer Sciences, School of Medical Sciences, Faculty of Biology, Medicine and Health, MANCHESTER, United Kingdom; 3Instituto de Investigación Sanitaria de Santiago de Compostela, Fundación Pública Galega de Medicina Xenómica, Santiago de Compostela, Spain; 4University Hospital Virgen del Rocio, Department of Radiation Oncology, Sevilla, Spain; 5Maastricht University Medical Center, Department of Radiation Oncology (MAASTRO), GROW-School for Oncology and Developmental Biology, Maastricht , The Netherlands; 6The University of Manchester, Division of Cancer Sciences, Translational Radiobiology Group, MANCHESTER, United Kingdom; 7Hashemite University, Faculty of Medicine, Zarqa, Jordan

Show Affiliations
Purpose or Objective

Esophageal toxicity is relatively common following lung cancer radiotherapy. Severe esophageal toxicity (G3+) often necessitates hospital admission and significantly detriments quality-of-life. There are no robust predictors for G3+ esophageal toxicity. We investigated predictors of post-radiotherapy acute G3+ esophageal toxicity from a large lung cancer dataset.

Material and Methods

2,687 patients from 5 studies (2 randomized trials: RTOG-0617 and CONVERT and 3 prospective datasets: MAASTRO, REQUITE and Spanish) were reviewed. Inclusion criteria were lung cancer patients treated with curative-intent radiotherapy (prescribed dose≥40Gy) who were followed-up prospectively. Acute toxicity (within 90 days of radiotherapy completion) was assessed according to Common Terminology Criteria for Adverse Events [versions 3.0-4.0]. The primary endpoint was acute G3+ esophageal toxicity. Univariable and multivariable logistic regression analyses of complete cases were performed using a priori selected variables including sex, age, study, lung cancer type, dose per fraction, planning target volume, smoking status, concurrent chemotherapy and esophageal V35Gy (volume receiving≥35Gy), mean and maximum dose. Both physical dose and EQD2,10 (equivalent dose in 2Gy fractions with α/β=10Gy) were tested. Due to low incidence of the outcome, data from all studies were pooled, and sensitivity analyses were performed by removing one study at a time. Model performance was assessed by calibration plots and receiver operating characteristic area under the curve.


Of 2,131 patients analyzed, 202 (9.5%, range 2.0-16.5%) developed acute G3+ esophageal toxicity (Table 1).  There was a strong correlation between esophageal mean dose and V35Gy (90%, range 83-96%). As the CONVERT trial did not collect mean esophageal dose, V35Gy was selected for analyses of all 5 studies. On multivariable analysis, acute G3+ esophageal toxicity was associated with concurrent chemotherapy (OR=2.9, p=0.03) and esophageal V35Gy (OR=1.02, p<0.0001). Moreover, Compared to RTOG-0617, patients in the REQUITE (OR=0.19, p<0.0001) and Spanish cohorts (OR=0.48, p=0.01) had a lower risk of developing acute G3+ esophageal toxicity (C-index: 77%; Figure 1A & B). Esophageal V35Gy remained a significant predictor of acute G3+ esophageal toxicity following all 5 sensitivity analyses. Prediction and discrimination of the model were best when the CONVERT trial dataset was removed (C-index: 80%; Figure 1B & 1C). Rerunning the analyses using mean dose (OR=1.09, p<0.0001) instead of V35Gy for all cohorts removing CONVERT yielded similar results (C-index: 79%). Results using physical dose and EQD2,10 were comparable.


In this large multi-cohort study, acute grade 3+ esophageal toxicity was increased in patients receiving concurrent chemotherapy and was consistently predicted by esophageal V35Gy and mean dose. Our results support using esophageal V35Gy or mean dose as an esophageal toxicity predictor in future trials and clinical practice.