ESTRO 2024 - Abstract Book

S2148

Clinical - Upper GI

ESTRO 2024

(p<0.01). Univariable analysis showed relevant associations between several variables and pneumonia: radiotherapy technique (PhRT vs PRT)(p<0.01), baseline COPD (p=0.01), age (p = 0.09), surgical approach (open vs minimal invasive)(p=0.11) and smoking at baseline (p=0.14). A significant interaction was found between baseline smoking and baseline COPD. No confounding was found between the surgical variables and the radiotherapy technique. At multivariable analysis, baseline COPD and radiotherapy technique (PRT vs. PhRT) remained significant (Table 1).

The PCA identified three components that collectively explained 90% of the variance in the DVH data. If, in the multivariable model, radiotherapy technique was replaced by these PCA components, PC1 and PC3 were significantly associated with pneumonia, next to baseline COPD (Table 1). PC1 was correlated to the mean dose in both the lungs and the heart, while PC3 was associated with the heart and lung volume that receives a low dose (≤20Gy) compared to the volume that receives a high dose (>20Gy). PC2 was not significantly associated with pneumonia and, based on the biplot, seemed to represent the distribution of dose between the heart and lungs. In Figure 1, a biplot illustrates the relationship between the DVH parameters and PC1 and PC3. The figure highlights the distinctions between PhRT and PRT patients in terms of the significant DVH parameters.