Optimizing outcomes in EGFR mutation-positive NSCLC: which tyrosine kinase inhibitor & when? Review progression compared with chemotherapy, and for most patients with baseline brain metastases, the brain was not the site of first disease progression [75] . Furthermore, an additional competing risk analysis in afatinib-treated patients with baseline brain metastases showed that the cumulative incidence of CNS progression was lower than that of non-CNS progression (31 vs 52%; Figure 4 ). Risk of de novo CNS progression with afatinib was also very low (observed in only 6% of patients receiving afatinib in LUX-Lung 3 and 6). Non-CNS progression for patients without baseline brain metastases was 78% [B oehringer I ngelheim , D ata on F ile ]. Overall, both afatinib and osimertinib have demonstrated activity in patients with asymptomatic brain metastases at standard therapeutic doses, while there are also limited data showing activity of erlotinib and gefitinib in patients with brain metastases, especially as high-dose regimens or in combination with WBRT. Further data on the impact of these agents on brain metastases are required, as well as long-term outcomes (including impact of subsequent therapy), to help determine which agent is the most appropriate first-line option for these patients. Factors influencing first-line treatment choice: tolerability profile EGFR TKIs are associated with a predictable, class-related AE profile, based on their mode of action. Inhibition of wild-type EGFR results in characteristic gastrointestinal (e.g., diarrhea) and skin (e.g., rash) AEs, although there are some differences in the AE profile between different EGFR TKIs. For example, in LUX-Lung 7, the most common drug-related grade 3 / 4 AEs with afatinib were diarrhea (13%), rash or acne (9%) and fatigue (6%), while elevated liver enzymes (9%) and rash or acne (3%) were the most common drug-related grade 3 / 4 AEs with gefitinib [28] . Treatment discontinuations due to treatment-related AEs were equally low in both treatment groups (6.3%). This is partly due to the well-defined dose adjustment protocol for management of afatinib-associated AEs. Post hoc analysis from LUX-Lung 3, 6 and 7 showed that dose adjustment resulted in a reduction in the frequency and severity of AEs, as would be expected. Importantly, however, this did not impact on efficacy: median PFS was similar in patients who received a dose reduction within the first 6 months of treatment and those who remained on the 40-mg dose (or 50 mg) [76 , 77] . Osimertinib and other third-generation EGFR TKIs were designed to be EGFR wild-type sparing in an effort to improve tolerability. The overall incidence of treatment-related AEs in FLAURA was similar between osimertinib and gefitinib (91 and 92%, respectively), but with a lower incidence of grade ≥ 3 treatment-related AEs for osimertinib (18 vs 28%). The most common grade 3 / 4 treatment-related AEs with osimertinib were decreased appetite (3%) and diarrhea (2%) [10 , 78] . Information on discontinuations due to treatment-related AEs in FLAURA are not available yet; however, the discontinuation rate due to any AE was 13% in the osimertinib group and 18% in the erlotinib / gefitinib group [10] . Overall, osimertinib has a favorable safety profile. However, it should be considered that the most likely second- line treatment after osimertinib will be chemotherapy. The tolerability profile of chemotherapy can be extremely challenging for patients, especially in later-line treatment, with a detrimental impact on patient quality of life. In contrast, the majority of patients receiving a first- or second-generation EGFR TKI will likely be able to receive osimertinib in second-line, thus prolonging their chemotherapy-free period. Given that EGFR TKIs are generally better tolerated than chemotherapy, sequential use of EGFR TKIs may offer improved tolerability and quality of life in the longer term. Again, data from AURA3 and FLAURA are awaited. One final consideration is the potential for drug–drug interactions. Gefitinib and erlotinib undergo extensive hepatic metabolism via CYP450 enzymes; additionally, erlotinib is a potent inhibitor of CYP1A1 and a moderate inhibitor of CYP3A4 and CYP2C8 [79] . Osimertinib is metabolized by CYP3A4 and CYP3A5, may inhibit breast cancer resistance protein and may induce CYP3A4, CYP1A2 and CYP2C [80 , 81] . In contrast, afatinib undergoes minimal hepatic metabolism and is not a substrate for CYP-dependent enzymes [79] . This would suggest that afatinib has a low potential for drug–drug interactions, which is an important factor in patients with advanced NSCLC, many of whom may have additional co-morbidities and are likely to be receiving multiple medications. Conclusion Gefitinib, erlotinib and afatinib are approved for the first-line treatment of EGFR mutation-positive NSCLC. Emerging evidence suggests that second-generation inhibitors may be more effective than first-generation TKIs. Following disease progression on these agents, osimertinib, a third-generation EGFR TKI, has shown considerable efficacy in patients with T790M-positive disease, and is approved in this setting. For patients with T790M-negative disease, options are more limited, and chemotherapy is the current principal treatment. Recent data from FLAURA showed that osimertinib had superior PFS to first-generation TKIs in the first-line setting, positioning osimertinib future science group 10.2217/fon-2017-0636