Structure-based Classification and Nomogram Prediction for Differential Inhibitor Sensitivity of EGFR Exon 19 Deletions.

8580 Background: EGFR exon 19 deletions (19Del) in non-small cell lung cancer are associated with benefit from tyrosine kinase inhibitors (TKIs), but the relative inhibitor sensitivity of individual EGFR 19Del is unknown. This study predicted the structural features and TKI sensitivity of different EGFR 19Del, compared the clinical efficacy of different TKIs in non-small cell lung cancer (NSCLC) patients harboring EGFR 19Del, and incorporated structural and clinicopathological variables into a nomogram to predict the progression-free survival (PFS) of NSCLC patients with EGFR 19Del. Methods: 1964 patients with advanced NSCLC harboring EGFR 19Del treated with EGFR TKIs from March 2016 to March 2023 were retrospectively screened. Computational model generation and molecular dynamics simulation were used to predict the structural features and TKI sensitivity of different EGFR 19Del variants. Stepwise Cox Proportional Hazards Regression was conducted to select variables for inclusion in the subsequent survival nomogram. Results: A total of 401 patients and 17 subtypes of EGFR 19Del were included in this study. For patients who received first-line 3rd generation TKIs, E746_A750>X and L746_A750>P had significantly shorter median PFS (mPFS) than other patients (10.6 months vs. 22.6 months, p = 0.0069; 12.8 months vs. 22.6 months, p = 0.027). For patients who received first-line 1st or 2nd generation TKIs and second-line TKIs, the PFS was not different among EGFR 19Del. Based on molecular dynamics simulation, E746_A750>X, E746_S752>V, E746_P753>VS and L747_P753>S were grouped into CLASS 1/2G, which had less binding affinity toward 1st or 2nd generation TKIs than other EGFR 19Del variants, while E746_A750>X, E746_S752>V, L747_A750>P and L747_T751>P were grouped into CLASS 3G, which had less binding affinity toward 3rd generation TKIs than other EGFR 19Del variants. For patients who received first-line and second-line 3rd generation TKIs, CLASS 3G showed shorter mPFS than other patients (12.8 months vs. 22.6 months, p = 0.059; 7.73 months vs. 15.1 months, p = 0.041). For patients who received first-line and second-line 1st or 2nd generation TKIs, the PFS was not different between CLASS 1/2G and other patients. For patients who received first-line 3rd generation TKIs, a survival nomogram was generated incorporating histology, degree of differentiation and CLASS 3G. For patients who received first-line 1st or 2nd generation TKIs, a survival nomogram was generated incorporating age, degree of differentiation, liver metastases and number of metastasized organs. Conclusions: We identified a group of EGFR 19Del responded poorly to 3rd generation TKIs based on structural features. Nomogram that incorporated structural and clinicopathological variables was a robust prognostic model. This study highlights the molecular structural analysis as a useful predictive tool of TKIs efficacy.