ESTRO 2020 Abstract book

S324 ESTRO 2020

trained previously on an European patient cohort : the clinical, radiomic and combined model (Table 1).

models (z-score test: p=0.005). Kaplan-Meier curves are shown for the three models in Figure 1. They showed that the radiomic and combined model were able to stratify the lymph nodes into high- and low-risk groups with significant differences between the nodal control probability prediction between the groups with p-values of 0.003 and < 0.001, respectively.

Lymph nodes were stratified into low-, intermediate- and high-risk groups by using the lower and upper quantile of the prognostic scores from the training cohort. Kaplan- Meier curves were used to show the differences between different risk groups and were compared using log-rank tests. The models’ discriminative power was assessed by Harrell’s concordance index (c-index). Results There were 20 (5.3%) nodal failures from 15 patients after a median follow-up of 36.1 months. Both the radiomic and the combined model showed good agreement between predicted and observed nodal control probabilities (Figure 1). The radiomic (c-index: 0.71; 95% confidence interval (CI): 0.59-0.84) and combined (c-index: 0.71; 95% CI: 0.59- 0.82) models performed better than the clinical model (c- index: 0.57; 95% CI: 0.47-0.68) in this cohort, with a significant difference between the combined and clinical

Conclusion The clinical, radiomic and the combined (including clinical and radiomic features) models were externally validated. The radiomic and combined models performed better in the external cohort than the clinical model. These models could be used to stratify lymph nodes into high- and low- risk groups for nodal failure.