Risk factors for esophagitis after hypofractionated palliative (chemo) radiotherapy for non-small cell lung cancer

Esophagitis influences quality of life and might cause treatment interruption and hospitalization. Previous studies of risk factors focused on curative treatment for non-small cell lung cancer (NSCLC), which often involves concomitant chemoradiation (CRT). Given the uncertainty around extrapolation of dose constraints, we analyzed risk factors in patients treated with hypofractionated palliative regimens.

A retrospective review of 106 patients treated with palliative radiotherapy or CRT between 2009 and 2017 was performed.

prescribed total dose 30-54 Gy, dose per fraction 2.5-4 Gy, esophageal dose > 1 Gy. Uni- and multivariate analyses were performed in 97 eligible patients to identify predictive factors for acute esophagitis grade ≥ 1 (CTCAE 5.0).

Forty percent of patients were treated with 15 fractions of 2.8 Gy (42 Gy) and 28% also received chemotherapy according to the CONRAD study regimen (induction and concomitant Carboplatin/Vinorelbine) published by the Norwegian Lung Cancer Group. Thirty-four percent were treated with 10 fractions of 3 Gy. Stage IV NSCLC was present in 47%. Esophagus Dmax was 39 Gy (population median) and Dmean 15 Gy. Overall 31% of patients developed esophagitis (26% grade 2-3, no grade 4-5). Several dosimetric parameters correlated with the risk of esophagitis (Dmax, Dmean, D5cc, V20, V30, V35, V40). Dmax outperformed other dosimetric variables in multivariate analysis. Furthermore, concomitant chemotherapy significantly increased the risk of esophagitis, while oral steroid medication reduced it. In patients with Dmax ≥40 Gy a reduced Dmean (≤20 Gy) was beneficial.

In order to reduce esophagitis after hypofractionated palliative treatment lower doses than those recommended in curative NSCLC settings are preferable. Besides esophageal dose, CRT is the main risk factor for esophagitis. Additional work is needed to confirm that steroids are able to modify the risk (or to rule out confounding effects of baseline variables not included in our database).