Background: Mutations in the epidermal growth factor receptor (EGFR) gene are the most common targetable gene alterations in non-small cell lung cancer (NSCLC). In Japan, approximately 40% of patients who undergo surgical resection for non-squamous NSCLC have EGFR mutations. However, no long-term studies have been conducted including a large number of EGFR-positive NSCLC patients with postoperative recurrence (PR).

Methods: We conducted a retrospective observational study of the data of EGFR-positive NSCLC patients with PR who had undergone surgery at the Shizuoka Cancer Center between October 2002 and November 2017. We evaluated post-recurrence overall survival (PRS) and postoperative overall survival (POS) using the Kaplan-Meier method and identify any associations between the clinical variables at recurrence and PRS using univariate and multivariate analysis.

Results: We enrolled 162 patients. The median observation time for PRS was 4.95 years (range, 0.82-13.25) and POS was 5.81 years (range, 2.84-16.71). The median PRS was 5.17 years (95% confidence interval [CI], 3.90-5.61) and POS was 7.07 years (95% CI, 5.88-8.01). Univariate analysis identified male sex (median PRS: 3.32 vs. 5.39 years; p < 0.05), bone metastasis (median PRS: 2.43 vs. 5.33 years; p < 0.05), and central nervous system (CNS) metastasis (median PRS: 3.05 vs. 5.39 years; p < 0.05) and multivariate analysis identified bone metastasis (hazard ratio [HR], 2.01; 95% CI, 1.23-3.28; p < 0.05) and CNS metastasis (HR, 1.84; 95% CI, 1.14-2.98; p < 0.05) as poor prognostic factors. The pattern of recurrence (oligo vs. non-oligo recurrence) was not a prognostic factor. Logistic regression analysis revealed the association between sex and the presence bone/CNS metastasis at recurrence.

Conclusion: Our data may help visualize future prospects and determine the timing of osimertinib initiation. New treatment strategies need to be developed for patients with bone/CNS metastasis at the first recurrence.