Osimertinib is an oral, irreversible TKI that forms a covalent bond with the cysteine residue in position 797 of EGFR within the ATP-binding site. Osimertinib and its active metabolites also interact with a number of other kinases harboring the cysteine residue. Osimertinib is a potent inhibitor of T790M with little wild-type activity and shows tumor regression in murine models (35).

AURA (a phase I dose escalation study) (36) was performed in patients with EGFR mutation-positive advanced NSCLC with acquired resistance to TKI. No dose-limiting toxicity (DLT) was observed; the most common adverse events were diarrhea, rash, anorexia, and nausea (see Table 1). The overall RR was 51% [95% confidence interval (CI), 45–58]; higher in the T790M mutation-positive group than the T790M mutation-negative group (61 versus 21%). The median PFS was 9.6 months (95% CI, 8.3 to not reached) in EGFR T790M-positive patients.

Updated results of the 80 mg (RP2D) cohorts from three AURA studies were presented (37), and confirm a high RR (66% in phase I study and 71% in phase II studies). Encouraging duration of response (12.5 months in pooled phase II studies) and PFS (11.0 months in same pooled analysis) was seen (37).

AURA2 is a single-arm, open-label phase II study of osimertinib 80 mg daily in T790M mutation positive NSCLC after first-line TKI. A total of 140 (70%; 95% CI 64–77) of 199 patients (with measurable disease) achieved an objective response. There was a disease control rate (dCR) of 92%. Toxicity was manageable with low rates of grade 3 or higher toxicity (see Table 1) (38).

AURA3 was a phase III randomized trial that assessed the efficacy and safety of osimertinib (80 mg daily) versus platinum-doublet chemotherapy after initial TKI failure in 419 patients with T790M mutation-positive advanced NSCLC (39). The trial demonstrated superior PFS 10.1 versus 4.4 months (HR 0.30; 95% CI 0.23–0.41; p < 0.001) and higher RR with osimertinib, 71%, versus chemotherapy, 31% (39). Grade 3 or 4 adverse events occurred in 23% of patients on osimertinib, compared to 47% with chemotherapy (Table 1). Quality of life results are pending.

In November 2015, osimertinib received accelerated approval by the FDA, representing rapid progress through drug development—the first AURA patient was enrolled in March 2013 and FDA accelerated approval was granted in November 2015.

Rociletinib is an irreversible orally delivered third-generation TKI that targets L858R, del19, and T790M mutations of EGFR with little WT activity. Rociletinib also modifies the C797 site through covalent binding. Tumor xenograft and transgenic models documented tumor regression in preclinical studies (40). In the phase I/dose expansion study, TIGERX, 130 patients with progression following EGFR TKI were enrolled but the maximum tolerated dose was not reached (41). The RR was 59% in T790M-positive patients; however, pooled data from this TIGER-X study and the phase II TIGER-2 was initially reported to be 30.2% (42) and later updated and published as 45% (43).

The most frequent grade 3/4 AEs, which occurred in more than 10%, included hyperglycemia and QTc prolongation (Table 1). The hyperglycemia is thought to be mediated by a metabolite of rociletinib that inhibits insulin-like growth factor-1 receptor and to a lesser extent, insulin receptor kinases. Clovis has suspended development of rociletinib and terminated enrollment in clinical trials in 2016, soon after the FDA rejected the request for accelerated approval (44).

Olmutinib is an oral selective inhibitor for EGFR including T790M mutant kinases and acts by binding to a cysteine residue close to the kinase domain. Potent inhibition of representative cell lines and in vivo activity have been reported (45). A phase I trial of 173 patients with EGFR-mutant lung cancer that had failed previous TKI therapy demonstrated a favorable safety profile and promising antitumor activity (46, 47). The MTD was established as 800 mg once daily. Treatment-related adverse events occurred in 87.3% of 165 patients, mainly diarrhea, rash, skin exfoliation, nausea, pruritus, decreased appetite, and dry skin. Grade 3 or greater toxicity was 2% in the initial study report (2/93), although not in the updated results (Table 1) (46–48).

In 34 patients with centrally confirmed T790M tumor mutations who received olmutinib at a dose greater than 650 mg daily, the RR was 59% (10 confirmed, 10 unconfirmed partial responses) and 13 achieved disease stabilization (dCR 97%). The phase II study has been suspended early with three cases of severe skin toxicity, including two reports of toxic epidermal necrolysis (one fatal), and one case of non-fatal Stevens–Johnson syndrome. The future of the drug’s development is uncertain.

EGF816 is an oral, irreversible EGFR TKI that also forms a covalent bond with C797. Low IC₅₀ values and in vivo activity against L858R–T790M and del19–T790M have been reported (49). Data from the phase I/II study of EGF816 in advanced T790M-positive lung cancer are now available (50, 51). Dose escalation began at 75 mg daily up to 350 mg daily for capsules and 100–225 mg daily for tablets. Diarrhea, stomatitis, rash, and pruritus were the most common toxicities (see Table 1), and the confirmed RR was 44% (56/127) and dCR was 91% (NCT02108964). Combination studies with immunotherapy are now recruiting (NCT02323126; NCT02335944; NCT02900664).

ASP8273 is an EGFR TKI that selectively and irreversibly inhibits mutant EGFR kinases including T790M by the formation of a covalent bond with C797. Both in vitro and in vivo studies confirm activity in T790M mutant lung cancer with relative WT sparing (52). In a phase I study, doses were escalated to 500 mg but the RP2D has been deemed 300 mg, although the details of DLT and maximum tolerated dose levels are not published (53). Of 60 patients treated with ASP8273 at the 300 mg dose, there was no DLT. All patients were EGFR positive with 90% having a T790M mutation. PR was demonstrated in 16 of 45 evaluable patients; dCR was 62% (n = 28/45). For the 40 T790M mutation-positive subjects with evaluable data, 38% (15/40) had PR and dCR was 65% (26/40) (NCT02113813). The Phase III SOLAR study is underway comparing initial ASP8273 with a first-generation TKI in patients with EGFR-mutant lung cancer (NCT02588261) (53).

PF0677775 is another oral inhibitor of EGFR T790M with 26-fold increased selectivity of mutant versus wild-type EGFR. It is currently under evaluation in a phase I/II study in patients with advanced EGFR mutation-positive lung cancer (del19 or L858R, T790M positive and negative) (NCT02349633) and early results have demonstrated activity and tolerability (54).

Avitinib is a new, irreversible, EGFR mutation selective TKI being evaluated in a phase I/II clinical trial (NCT02274337). In the reported dose escalation study (55), 25 patients were treated. The most common AEs were diarrhea, rash, and pruritus. Although diarrhea and rash increased in frequency in a dose-dependent manner, the majority of them were grade 1 (Table 1). There was no drug discontinuation in all treated patients. Outcomes for two patients with T790M-positive lung cancer showed partial responses. The clinical characteristics and efficacy outcomes of the remaining patients are not reported (55).

The “uncommon” EGFR mutations represent a heterogeneous group and can account for up to 10–18% of EGFR mutations (56, 57). The most frequent include exon 20 insertions (exon20ins), and point mutations G719X, L861Q, and S768I. The latter three mutations may have a superior response to afatinib (58). The majority of (exon20ins) are thought resistant to EGFR TKIs with the exception of A763_Y764insFQEA. In a preclinical study, osimertinib demonstrated potency with a wide therapeutic window in the exon20ins studied (Y764_V765insHH, A767V769dupASV, and D770_N771insNPG) (59). More recently, it has been revealed that EGFR amplification may occur in a subset of exon20ins. The dual EGFR blockade with osimertinib and cetuximab has demonstrated significant growth inhibition in in vivo models (60). EGF816 has shown both in vitro and in vivo efficacy in a number of exon20ins and in a patient-derived xenograft model, 100 mg/kg dosing resulted in tumor regression of 81% (49). AP32788 has also been shown to inhibit exon20ins in BA/F3 cell lines (61). The activity of third-generation TKIs in preclinical models has led to clinical trials for exon20 insertions including the phase I/II study of AP32788 (NCT02716116).

Approximately 30–50% of patients with NSCLC develop central nervous system (CNS) disease (62, 63). The association between EGFR mutation-positive NSCLC and the incidence of brain metastases is controversial with some studies suggesting an increased risk of CNS disease at diagnosis (64, 65). CNS disease can reduce survival and both brain metastases and loco regional therapies can impact neurological function and quality of life. First-generation EGFR TKIs have shown intracranial activity with erlotinib exhibiting higher CSF concentrations than gefitinib (66). Afatinib in patients pretreated with chemotherapy and a first-generation TKI has demonstrated a CNS dCR of 66% (67). A recent preclinical study has shown superior blood–brain barrier penetration of osimertinib compared to gefitinib, afatinib, and rociletinib (68). Sustained tumor regression in a murine brain metastases model has also been reported with osimertinib, doses of 80 mg in humans were predicted to target human brain metastases using an adaptive pharmacokinetic/pharmacodynamics model (69). AZD3759 is an innovative EGFR TKI developed to penetrate the blood–brain barrier but does not have T790M activity. BLOOM (NCT02228369) is a study testing the safety and efficacy of osimertinib 160 mg/day and AZD3759 in NSCLC patients with leptomeningeal disease; early data have reported disease control in three-quarters of patients and responses in 7 of 20 patients (70).

Osimertinib CNS activity was also confirmed in AURA and AURA2 (71). In the recently reported phase III study of osimertinib versus platinum-pemetrexed doublet, a significant improvement in PFS in patients with brain metastases was evident in the osimertinib group (8.5 versus 4.2 months; hazard ratio 0.32; 95% CI, 0.21–0.49) (39).

The point mutation C797S in exon 20 represents the most common resistance mechanism identified in third-generation EGFR TKIs. Most third-generation TKIs use the site of the cysteine amino acid located at position 797 for covalent binding, and the serine amino acid substitution reduces the capacity for TKI binding. Analysis of both plasma and tissue has confirmed this as a mechanism of resistance to osimertinib and olmutinib (72, 73). C797S as a cause of rociletinib resistance has been found to be much lower, but emergence of other uncommon EGFR mutations including L798I, L692V, and E709K have been implicated (23). Interestingly in preclinical models, if the activating mutation (del19 or L858R) is retained in the presence of C797S but without T790M, the tumor remains sensitive to gefitinib or afatinib. If T790M is present, in vitro analysis has demonstrated partial cetuximab sensitivity (74). Cetuximab with EAI045, a novel EGFR resistance mutation selective allosteric inhibitor was also effective in a mouse model of the triple-mutant EGFR L858R/T790M/C797S (75). Necitumumab is also being trialed in combination with osimertinib (Table 2) (NCT02496663, NCT02789345). Notably, brigatinib with or without the combination of an anti-EGFR antibody has demonstrated activity in preclinical models for the “triple-positive” tumors (76). First- and third-generation TKIs may also be combined effectively, but this is only likely if the C797S and T790M mutations occur in trans (77). Other acquired EGFR mutations have been reported by Chabon et al. including L798I, L762V, and E709K (23). EGFR amplification and copy number alterations are also important resistance mechanisms.

Cell line studies have identified the RAS pathway as important in emerging osimertinib resistance, including mutations in NRAS as well as copy number gains. The BRAF V600E mutation is also a known acquired resistance mutation (78). The addition of selumetinib (MEK inhibitor) delayed and prevented resistance in preclinical models and tumor regression has been documented in an osimertinib-resistant transgenic mouse model (79). The TATTON phase 1b study is a three-arm trial of TKI naive and pretreated patients that includes combinations of osimertinib with selumetinib (AZD6094), savolitinib, a MET inhibitor, and durvalumab (NCT02143466).

MET amplification as a cause for EGFR TKI acquired resistance has been described in case reports and crizotinib led to a response in one of these (80–82). In a further study of rociletinib resistance, MET amplification accounted for 26% of patients. Patient-derived xenograft models in this study were again successfully targeted using crizotinib (83). Notable in the xenografts and in the case report by Planchard et al., selective pressure permitted the emergence of MET amplifications without detectable T790M suggesting that MET may induce resistance to third-generation TKIs. Chabon et al. also described the preexistence of co-occurring MET amplification with T790M, which correlated with inferior responses to rociletinib (23).

Although immune checkpoint inhibitors have made huge advances in shifting the treatment paradigm in NSCLC, their role in EGFR-mutant disease is unclear. The third arm of TATTON investigated the combination of osimertinib with durvalumab and has reported early data (84). Patients were treated with osimertinib 80 mg daily with varying dosing and scheduling of durvalumab. In EGFR TKI naïve patients, the RR was 70% and in pretreated T790M-positive patients and T790M-negative patients RR was 67 and 21%, respectively. The combined rate of interstitial lung disease was 38% and as such this arm is currently on hold. Similarly, the CAURAL (NCT02454933) study investigating the durvalumab combination versus single-agent osimertinib has been halted due to toxicity concerns.

Other potentially targetable resistance mechanisms include HER2 amplification, FGFR1 amplification, and the PIK3CA E545K mutation. Epithelial to mesenchymal transition (EMT) and small cell transformation are also well recognized. Preclinical models have successfully targeted EMT with Akt inhibitors (40). Navitoclax, a BCL-2 inhibitor when combined with WZ4002, was shown to induce greater apoptosis than with the EGFR TKI alone (85). A phase 1b study is accruing (NCT02520778). Vascular endothelial growth factor (VEGF) and EGFR pathways are intimately related. The upregulation of VEGF receptors may be responsible for EGFR resistance and combination studies are ongoing (86) (Table 2). Early trials have already confirmed the benefit of dual inhibition with bevacizumab and erlotinib (87, 88). One small retrospective study has suggested the possibility of rechallenging with EGFR TKIs in addition to bevacizumab to gain further disease control (89).