ESTRO 2020 Abstract Book

S409 ESTRO 2020

within a cohort of centrally located NSCLC patients treated with SBRT. Material and Methods A total of 220 patients from 2 centers who underwent SBRT with curative intent (≤12 fractions) for centrally located NSCLC between 2006 – 2016 were eligible. A central tumor was defined as a tumor <2 cm from the esophagus and/or main bronchus or first branches of the bronchial tree. Commonly used fractionation schedules were 8 fractions of 7.5 Gy (31%) and 5 fractions of 9-12 Gy (37%). The majority was diagnosed with stage II disease (52%). External validation was performed in a separate cohort of 92 patients who underwent SBRT for central NSCLC in a 3 rd institute in which 72% of patients underwent 8 fractions of 7.5 Gy. The following parameters were analyzed: age, gender, Charlson Comorbidity Index, WHO performance scale, previous (lung)malignancies, availability of pathology, FEV 1 , tumor localization, operability, endobronchial tumor location, tumor size, PTV volume, disease stage and PTV D max / D min / D mean D 2% / D 50% / D 98% . A Cox proportional hazards model was used to build a nomogram to predict 6-months, 1-year, 2-year and 3-year OS. The bootstrap method was applied to internally validate the nomogram. Discriminatory ability was measured by the concordance index (C-index) while predictive accuracy was assessed with calibration plots. Results The final nomogram ( Figure 1 ) was based on five parameters: age, PTV volume, D mean (<100 Gy BED 10 vs. ≥100 Gy BED 10 ), WHO performance scale (0 vs. 1-4) and tumor localization (upper-/ middle- lobe or mediastinum vs. lower lobe). Median OS of the initial group was 28 months (95% CI 23-35) with a 1-year OS of 76% and a 2-year OS of 55%. The median OS of the external group was 31 months (95% CI 20-42), with a 1-year OS of 80% and a 2- year OS of 63%. The C-index of the nomogram (corrected for optimism) was moderate at 0.61. In the external validation cohort, the C-index was 0.60. Figure 2 shows the calibration plots for the 1-year OS in which the solid line displays the initial group and the dotted line the external validation group. This plot shows an agreement between the model’s predicted and observed values. The closer the points are to the diagonal line, the better the prediction.

Conclusion We presented the first nomogram modeling focusing on only centrally located NSCLC treated with SBRT. This resulted in a nomogram predicting model which is representative in our external cohort and therefore justifies the use of this nomogram within daily practice.