ESTRO 2024 - Abstract Book

S1663

Clinical - Lung

ESTRO 2024

(13/15, 87%). The radiomic score was able to successfully provide a prognostic stratification for all the three considered outcomes, as shown in Fig. 1.

The best predictive performance was achieved for OS (clinical model: AUC 0.87, 95% CI: 0.86-0.88; radiomic model: AUC 0.95, 95% CI 0.93-0.97), and for LPFS, with performances being: AUC 0.71 (95% CI: 0.70-0.73) and AUC 0.77 (95% CI: 0.76-0.78) for the clinical and radiomic models, respectively. For both these outcomes, the combination of clinical and radiomic information led to an overall improvement of model predictive ability compared to individual models (AUC 0.92, 95% CI: 0.91-0.93 and AUC 0.84, 95% CI: 0.80-0.87). Conversely, performances of the clinical and radiomic models were comparable for PFS. However, an improvement in AUC could be demonstrated for the hybrid model, as well (clinical model: AUC 0.68, 95% CI: 0.66-0.71; radiomic model: AUC 0.67, 95% CI 0.66-0.68 and clinico-radiomic model: AUC 0.73, 95% CI: 0.70-0.76).

Conclusion:

Baseline RFs from filtered CT images can improve prognostic stratification in this clinical setting, with hybrid models yielding the best performances. Notably, RFs alone outperform clinical descriptors in predicting both of OS and LPFS, although clinical variables, such as FEV1% seem to play a role.

Keywords: Early-stage NSCLC, radiomics, outcome modeling

1415

Digital Poster

Clinical significance of PD-L1 expression and EGFR mutation in brain metastasis of lung cancer.

fen wang 1 , YiQing Butler- Xu 2

1 university of kansas, radiation oncology, kansas city, USA. 2 university of kasnas, radiation oncology, Kansas City, USA

Purpose/Objective: