Prognostic factors and long-term survivors in EGFR-mutated lung cancer: A multi-institutional retrospective analysis.

e20594 Background: Evolution of tyrosine kinase inhibitors (TKIs) shapes the treatment paradigm of EGFR-mutated lung cancer, but resistance eventually develops. It is important to dissect the prognostic factors and to know the differences between long-term responders (LTR) and non-responders (NR). Methods: Electronic medical records were retrieved from Chang-Gung Medical Foundation from 2011 to 2020. Kaplan-Meier method was used to estimate survival, such as progression-free survival (PFS), and overall survival (OS). Multivariate Cox proportional hazard models was adopted to estimate adjusted hazard ratios and 95% confidence intervals. P 36 months, and 1,036 (29.2%) are NR with OS 18 months and 882 (24.8%) NR with PFS < 6 months. LTR tend to be younger, female, non-smoker, clinical stage 3b, no baseline brain metastasis, exon 19 deletions subtype, subsequent T790M development, bevacizumab usage, and osimertinib exposure (p all < 0.05). The median PFS of all enrolled patients is 12.3 months and median OS is 22.3 months. By multivariate analysis, age older than 65 years (Hazard ratio, HR:1.25 1.15-1.36), male (HR: 1.24 1.13-1.36), clinical stage 4 (2.19 1.78-2.69), baseline brain metastasis (1.25 1.12-1.37) are poor prognostic factors (p < 0.05). Also, T790M is not only predictive to osimertinib but also prognostic factor to OS (p < 0.05), and patients who developed subsequent T790M tend to have better PFS (18.1 v.s. 11.8 months). In patients without detected T790M along their disease course, exposure to osimertinib is related to better OS (33.1 v.s. 18.3 month, log-rank p < 0.05). Conclusions: LTR and NR are of distinctive characteristics, and T790M is prognostic and predictive to osimertinib. Further study is warranted for choosing the optimal treatment by clinical factors and T790M status.