Tyrosine kinase inhibitors with and without upfront CNS radiation for brain metastases in oncogene-driven non-small cell lung cancer (TURBO-NSCLC).

2019 Background: Newer generation tyrosine kinase inhibitors (TKI) for NSCLC with EGFR mutations and ALK rearrangements have demonstrated encouraging central nervous system (CNS) activity with CNS objective response rates, greatly improved from 1 st generation TKIs. In response to these data, guideline statements have acknowledged a strategy of CNS-penetrant TKI +/- upfront stereotactic radiosurgery (SRS) for the treatment of select patients with BM. However, optimal use of upfront SRS for BM in these patients is controversial since upfront CNS radiation has been the historical standard of care, and there are limited data guiding patient management with upfront TKI alone. Additionally, results from a large multi-institutional series reported inferior overall survival (OS) with the omission of SRS in patients with EGFR-mutated NSCLC treated with first-generation TKI. Methods: Data on TKI-naïve patients with EGFR- and ALK-driven NSCLC with BM treated with CNS-penetrant TKIs +/- upfront SRS were retrospectively collected from 7 centers in the United States. Time to CNS progression (PD), local CNS PD, and OS were analyzed, with multivariable adjustment (MVA) in Fine and Gray and Cox proportional hazards models for baseline factors including age, sex, performance status, mutation, extracranial metastases, prior therapy, neurologic symptoms, and number and size of BM. Results: We identified 317 patients (200 TKI only and 117 TKI+SRS). 250 (79%) and 61 (19%) patients received osimertinib and alectinib, respectively. Patients who received TKI+SRS were more likely to have BM ≥1 cm (p<0.001) and neurologic symptoms (p<0.001) at baseline. The median follow-up from treatment of BM was 23 months and 26 months in the TKI and TKI+SRS groups, respectively. Median OS was similar between the TKI and TKI+SRS groups (median 41 months 95% CI: 35-NR vs 40 months 95% CI: 40-NR, respectively; p=0.5). On MVA, TKI+SRS was associated with a significant improvement in time to CNS PD (HR 0.63; 95% CI: 0.42-0.96; p=0.033). Local CNS control was significantly improved with TKI+SRS (HR 0.30, 95% CI: 0.16-0.55; p<0.001), whereas no significant differences were observed in distant CNS control. Subgroup analyses demonstrated greater CNS control benefits with TKI+SRS in patients with BM ≥1 cm. Conclusions: This is the largest multi-institutional study comparing strategies of CNS-penetrant TKIs +/- upfront SRS in TKI-naïve patients with oncogene-driven NSCLC. The addition of SRS improved time to CNS PD and local CNS control but not OS. Patients with BM ≥1 cm may benefit the most from upfront integration of SRS. Table: see text