# Cumulative oesophageal dose and risk of high-grade toxicity in thoracic re-irradiation: a dose/toxicity analysis

## TL;DR

This study models the relationship between cumulative oesophageal dose and severe toxicity in thoracic re-irradiation to help guide safe treatment planning.

## Contribution

A new logistic regression model predicts high-grade oesophageal toxicity based on cumulative dose and chemotherapy use.

## Key findings

- A 5% risk of grade ≥3 toxicity occurs at 94.2Gy3 EQD2 without chemotherapy.
- Concurrent chemotherapy significantly increases the risk of high-grade toxicity.
- The model aligns with existing guidelines for safe re-irradiation dose limits.

## Abstract

•Thoracic re-irradiation can cause high-grade oesophageal toxicity and cumulative dose constraints have limited evidence.•Using multivariable logistic regression, we developed a model to predict grade ≥3 oesophageal toxicity from re-irradiation.•The significant variables were cumulative oesophageal Dmax and the use of concurrent chemotherapy.•The model predicted 5% risk of grade ≥3 toxicity at a cumulative oesophageal Dmax of 94.2Gy3 EQD2 (with no chemotherapy).•This is consistent with current guidelines and may inform re-irradiation planning, patient counselling or future studies.

Thoracic re-irradiation can cause high-grade oesophageal toxicity and cumulative dose constraints have limited evidence.

Using multivariable logistic regression, we developed a model to predict grade ≥3 oesophageal toxicity from re-irradiation.

The significant variables were cumulative oesophageal Dmax and the use of concurrent chemotherapy.

The model predicted 5% risk of grade ≥3 toxicity at a cumulative oesophageal Dmax of 94.2Gy3 EQD2 (with no chemotherapy).

This is consistent with current guidelines and may inform re-irradiation planning, patient counselling or future studies.

Re-irradiation (re-RT) in recurrent thoracic cancer is being performed more often and can cause severe toxicity. Information is lacking on safe re-irradiation dose constraints. We modelled the relationship between cumulative oesophageal dose and grade 3 or worse (≥G3) oesophageal toxicity to develop dose constraints.

We performed a literature search for reports of thoracic re-RT and selected studies that quoted cumulative oesophageal maximum dose (cDmax) in equivalent dose in 2 Gy fractions (EQD2) and ≥G3 toxicity. Additional collected data were inter-treatment interval and concurrent chemotherapy use (conCT) at re-irradiation. Logistic regression analyses were performed. The models were used to predict the cDmax associated with a 5% ≥G3 toxicity rate. Model performance was assessed using the Pearson correlation coefficient (PCC).

We identified 21 studies (505 patients), with 49 ≥G3 toxic events. The median oesophageal cDmax and interval were 84.8 Gy3 (3.7 – 220.6 Gy3 EQD2) and 15.5 months (1–162) respectively. Use of conCT and oesophageal cDmax were significantly associated with toxicity on univariable and multivariable modelling (both p < 0.001). The maximum likelihood doses associated with 5% risk of ≥G3 toxicity with/without chemotherapy were cDmax 43.0 Gy3 (95% CI: −18.5, 108.8) and 94.2 Gy3 (95% CI: 79.6, 142.8) respectively. The model had a PCC of 0.75 (p = 0.013) suggesting good correlation to the dataset.

The models predict a 5% toxicity at cDmax 43.0 Gy3 and 94.2 Gy3 EQD2 with/without chemotherapy. This supports a cDmax constraint of <95 Gy3 EQD2 to limit ≥G3 toxicity to under 5%, consistent with the American Radium Society constraints (<100 Gy3 EQD2). ConCT with re-irradiation has a large radiosensitising effect. Limitations of this study include the use of old, retrospective data resulting in wide CIs around the predictions and insufficient data to predict a volumetric constraint. Further modelling with more detailed dose data is required to refine and validate these predictions.