ESTRO 2024 - Abstract Book

S5052

Physics - Radiomics, functional and biological imaging and outcome prediction

ESTRO 2024

We identified 21 studies including 505 patients, with 49 ≥G3 toxic events. The median interval between initial radiotherapy and re-irradiation was 15.5 months (1-162) and the median cDmax was 84.8Gy (3.7 – 220.6Gy). Use of ConChT and cDmax were significantly associated with toxicity on univariable analysis (both p<0.001). Both factors remained significant when analysed in a multivariable model: ConChT (p<0.001) and cDmax (p<0.001). The multivariable model was selected for dose prediction to determine the effect of ConChT (see Figure 1). The model expression is:

where x 1 = cDmax and x 2 = concurrent chemotherapy.

The maximum likelihood doses associated with 5% risk of G3 toxicity with and without chemotherapy were cDmax 43.0Gy (95% CI: -18.5 to 108.8) and 94.2Gy (95% CI: 79.6 to 142.8) respectively. The doses associated with 20% risk of G3 toxicity with and without chemotherapy were cDmax 79.2Gy (95% CI: 51.6 to 136.0) and 130.4 (95% CI 104.5 to 212.0, see Figure 1). The multivariable model had a Pearson correlation coefficient of 0.75 (p=0.013) suggesting a good correlation between model predictions and observed rates of toxicity by decile. Due to the bootstrapping method used to obtain them, the CIs reflect uncertainties on both the fitted radiation dose-response and the effect of chemotherapy.