Treatment strategies moving forward will likely involve development of new targeted TKIs with greater potency and specificity against resistance mutations and different kinase selectivity, sequencing of targeted therapies based on the resistance mutations that develop from prior therapy, and development of combination regimens to target bypass signaling tracks.
New therapies that inhibit EGFR and ALK gene rearrangements in non—small cell lung cancer (NSCLC) have changed the paradigm of care in these settings. However, their use has involved a frustrating game of one-upmanship, in which these tumors develop new resistance mechanisms that counter successive generations of inhibitors. Resistance mechanisms that thwarted first-generation ALK inhibitor crizotinib (Xalkori) and the first-generation EGFR tyrosine kinase inhibitors (TKIs) erlotinib (Tarceva) and gefitinib (Iressa) created the need for secondand third-generation ALK and EGFR TKIs that could overcome such barriers (Figure 1, 2).1,2 Furthermore, first-generation EGFR and ALK TKIs had key shortcomings, such as a lack of central nervous system (CNS) penetration.
Superior progression-free survival (PFS) and better CNS penetrance of newer agents have led to approval or consideration of these for frontline therapy. However, the mechanisms of resistance that hinder these drugs are distinctly different from those affecting first-generation agents. Now, selecting the optimal therapy regimen for patients has come to involve careful analysis posttherapy, including repeat biopsies and testing, to identify these mechanisms of resistance.Somatic activating EGFR mutations were the among the first to be described in NSCLC and are found in approximately 10% of patients with NSCLC, but more frequently in nonsmokers, adenocarcinomas, women, and patients of Asian descent.3 Mutations in exons 18 to 21 are the most common mutations sensitive to EGFR TKI therapy, with deletions in exon 19 and the L858R point mutation in exon 21 accounting for 85% to 90%.4 Because there are primary resistance mutations, often occurring in exon 20, that do not respond to EGFR TKIs as expected and de novo or acquired resistance mutations, such as T790M, that occur after treatment with a first- or second-generation EGFR inhibitor, differentiating among EGFR resistance mutations has been important for successful targeted treatment.
The introduction of second- and third-generation EGFR inhibitors, initially designed to target common resistance mutations that developed from erlotinib and gefitinib, also yielded better outcomes in the frontline setting. Afatinib (Gilotrif) showed superior PFS (11.0 vs 10.9 months; HR, 0.73) and time-to-treatment failure (13.7 vs 11.5 months; HR, 0.73) compared with gefitinib in the LUX-Lung 7 trial,5 and dacomitinib (Vizimpro) showed an improvement in PFS (14.7 vs 9.2 months; HR, 0.59) over gefitinib in the ARCHER 1050 trial.6
The third-generation EGFR inhibitor osimertinib (Tagrisso), initially approved for patients with the EGFR T790M mutation on exon 20—the most common mechanism of resistance to erlotinib, gefitinib, or afatinib—also demonstrated numerically longer median PFS compared with gefitinib or erlotinib (18.9 vs 10.2 months; HR, 0.46; 95% CI, 0.37-0.57; P <.001) and was approved for frontline treatment for patients whose tumors have EGFR exon 19 or exon 21 L858R mutations in April 2018.7Despite the advances in EGFR-targeted therapy over the past decade, resistance to EGFR TKI therapy is virtually inevitable. Resistance mechanisms are broadly categorized as secondary alterations within EGFR, activation of an alternative signaling pathway or downstream effector gene, and phenotypic transformation. Although secondary alterations in EGFR are common mechanisms of resistance to first- and second-generation EGFR inhibitors, mechanisms of resistance to osimertinib, while not fully elucidated, appear to be more diverse and include EGFR-independent mechanisms.
Secondary Alterations Within EGFR
The T790M mutation on exon 20 occurs in approximately half of patients treated with erlotinib or gefitinib.8 Osimertinib was initially designed to target this mutation and demonstrated superior PFS compared with platinum/pemetrexed (Alimta) chemotherapy (10.1 vs 4.4 months; HR, 0.30) in patients with disease progression and the EGFR T790M mutation after first-line EGFR TKI therapy in the AURA3 trial.9 Other secondary mutations in EGFR associated with acquired resistance to first-generation EGFR TKIs include EGFR D761Y and T854A.10,11
Use of second- and third-generation agents in the frontline setting has also introduced patterns of resistance. Although the patterns of resistance to dacomitinib have not been investigated extensively, study results revealed EGFR T790M and C797S mutations in cells following chronic dacomitinib exposure.12 EGFR C797S was also identified as a mechanism of resistance to osimertinib, although it was only demonstrated in the cell-free plasma DNA of 6 of 15 patients with advanced lung cancer that progressed on osimertinib.13 This study also identified 5 patients who maintained the T790M mutation but did not acquire the C797S mutation and 4 who lost the T790M mutation despite detection of the underlying EGFR-activating mutation. Findings from another study showed mutations in EGFR G796/C797, L792, and L718/G719 in 24.7%, 10.8%, and 9.7%, respectively, in cell-free DNA samples from 93 patients with osimertinib-resistant disease.14
Activation of an Alternative Signaling Pathways or Downstream Effectors
However, approximately half of patients with acquired resistance mutations have unknown genetic drivers, and many patients likely have co-occurrence of multiple genetic drivers. Alterations in parallel or downstream oncogenes, such as MET, KRAS, and PIK3CA, have been demonstrated in cell-free DNA samples of patients with osimertinib-resistant disease, which may contribute to disease progression in patients without secondary mutations in EGFR.14 Another study showed that 68% of patients with acquired resistance to osimertinib had loss of T790M, which was associated with shorter time to discontinuation of treatment and a smaller decrease in the levels of the EGFR driver mutation, suggesting emergence of preexisting resistant clones.15 This study’s results also showed a wide range of competing resistance mechanisms in patients with loss of T790M, including acquired KRAS mutations and targetable gene fusions.
Another study identified amplification of HER2 and MET along with complete loss of the T790M mutation as potential mechanisms of osimertinib resistance.16 The researchers also identified a T790M mutation at the primary biopsy site and T790M wild-type at the metastatic site, suggesting a scenario in which osimertinib selectively targets the T790M-positive mutation while the wild-type cells mediate resistance through bypass pathways involving HER2 or MET.
The change in histology from adenocarcinoma to small cell lung cancer (SCLC) has been observed in 3% to 10% of cases of EGFR TKI resistance, and sequencing of EGFR from repeat biopsies demonstrates that the activating EGFR mutation is maintained in SCLC, indicating a phenotypic evolution of tumor cells.17 Although the mechanism by which histologic transformation leads to TKI resistance is still unclear, loss of retinoblastoma was demonstrated in all 10 EGFR-mutant NSCLC tumor samples that had transformed to SCLC at the time of drug resistance compared with only 1 of the 9 cases that retained NSCLC histology.18
Phenotypic change via epithelial-to-mesenchymal transition has also been demonstrated in approximately 5% of patients with EGFR TKI resistance and may involve activation of AXL or its ligand GAS6 or IGF1R and SRC/FAK signaling.17ALK rearrangements are encountered in 3% to 7% of patients with stage IIIB/IV NSCLC, which is dependent on the series and also, potentially, on patients selected for molecular testing. In most cases an adenocarcinoma is involved. Frequency is higher for nonsmokers—17% to 20%.19 Although the first-generation ALK TKI crizotinib was considered a breakthrough therapy for the treatment of ALK-positive NSCLC, resistance often occurs within 1 to 2 years from the start of therapy. Additionally, the low CNS penetrance is a major shortcoming of crizotinib because CNS metastases are relatively common with ALK-rearranged NSCLC. Therefore, structurally different ALK TKIs were designed to target mechanisms of crizotinib resistance and treat disease in the CNS. The second-generation ALK TKIs ceritinib (Zykadia), alectinib (Alecensa), and brigatinib (Alunbrig) target ALK L1196M and G1269A mutations, 2 of the most common mutations associated with crizotinib resistance, among others.20
Study findings showing that these second- and third-generation ALK TKIs have better disease control, particularly in the CNS, have prompted the shift toward using them in the frontline setting. Results of the phase III ALEX trial showed that alectinib was associated with improved PFS and overall survival and fewer events of CNS progression compared with crizotinib in the frontline setting, and the phase III ALTA-1L trial’s results showed that brigatinib led to superior PFS and intracranial response rate compared with crizotinib in patients with no exposure to an ALK TKI.21,22 Ceritinib was also approved for first-line therapy based on data from the ASCEND-4 trial, which showed PFS was longer than with platinum-based chemotherapy in patients with treatment-naïve stage IIIB/IV ALK-positive NSCLC.23
Furthermore, findings from a phase II study showed that the third-generation ALK inhibitor lorlatinib (Lorbrena) led to a 90% overall response rate and 63% objective intracranial response rate in the expansion cohort of treatment- naïve patients.24 Enrollment is ongoing for the phase III CROWN trial (NCT03052608), which will compare lorlatinib to crizotinib for first-line treatment of patients with metastatic ALK-positive NSCLC.Although there have been improvements in PFS and the intracranial response rate with secondand third-generation ALK TKIs, resistance eventually develops with these agents and is distinctly different from resistance mechanisms associated with crizotinib. Similar to resistance mutations that follow EGFR-targeted therapy, those that develop following ALK TKIs can be categorized into secondary mutations of the ALK gene; ALK-independent mechanisms, or phenotypic transformation. However, the spectrum of on-target ALK resistance mutations following TKI therapy is much broader than with EGFR inhibitors.25 Additionally, secondary ALK resistance mutations tend to be more common with second-generation ALK inhibitors than with crizotinib, although the results of one study showed that 44% of post—second generation ALK TKI biopsies tested negative for ALK mutations, indicating that alternative pathways likely play a key role in resistance.26
Secondary ALK mutations
Secondary ALK resistance mutations were demonstrated in 54% to 71% of patients progressing on second-generation ALK TKIs (depending on the agent), whereas they only occurred in 20% of patients progressing on crizotinib.26 This study also showed that the frequency of the ALK G1202R resistance mutation was substantially higher with the second-generation ALK TKIs (21%-43%) than with crizotinib-resistant tumors (roughly 2%). However, it is important to note that most patients with resistance to second-generation agents had received prior crizotinib, which may affect resistance patterns differently than if the second-generation agents were used in the frontline setting. Resistance mutations among each of the second-generation inhibitors also vary considerably. For example, ALK I1171 mutations are often present in alectinib-resistant tumors but respond to ceritinib, whereas ALK F1174 mutations tend to be resistant to ceritinib and sensitive to alectinib.26-28
Compound ALK resistance mutations may also occur in patients treated with sequential ALK TKI therapy. An investigation of the clonal evolution of compound ALK mutations at a single site over time showed a patient who demonstrated an ALK E1210K mutation in the post crizotinib specimen, followed by ALK E1210K and S1206C mutations in 1 post-brigatinib specimen and E1210K and D1203N mutation,but no S1206C mutation, in the second post-brigatinib specimen.26 The authors concluded that sequential ALK TKI therapy may lead to compound ALK resistance mutations that confer high levels of resistance to ALK-targeted therapy.
ALK-Independent Mechanisms of Resistance
Activation of bypass signaling tracks through genetic alterations, autocrine signaling, or dysregulation of feedback signaling represent a key category of ALK-independent resistance mechanisms. Increased EGFR activation accompanied by persistent downstream ERK and AKT signaling was identified in crizotinib-resistant cell lines and 4 of 9 patients with pre- and post-treatment biopsies, although EGFR mutations and/or amplification were not observed.25 Unlike crizotinib, second-generation ALK TKIs do not have anti-MET activity, and a case report demonstrated activation of MET as a bypass mechanism of resistance following treatment with alectinib.25 Direct reactivation of the downstream effector protein MEK was found in a patient with ceritinib resistance, and application of the MEK inhibitor AZD6244 to the patientderived cell line resensitized the cells to ceritinib.29 Additionally, a separate study found that dual blockade with ALK and MEK may overcome and delay ALK TKI treatment resistance.30 A series using next-generation sequencing also detected mutations in TP53 in 33% of the 27 specimens tested, as well as mutations in DDR2, BRAF, FGFR2, MET, NRAS, and PIK3CA, each in 1 specimen, after second-generation ALK TKI therapy.26
Changes in Cell Lineage
Cell lineage changes, namely epithelial-tomesenchymal transition (EMT) and SCLC transformation, have also been implicated in ALK TKI resistance. Immunohistochemical staining of ceritinib-resistant biopsies showed evidence of EMT in 5 of 12 cases (although 3 of these had concurrent ALK resistance mutations), and SCLC transformation has been observed in case reports following ALK TKI therapy.26,31Treatment strategies moving forward will likely involve development of new targeted TKIs with greater potency and specificity against resistance mutations and different kinase selectivity, sequencing of targeted therapies based on the resistance mutations that develop from prior therapy, and development of combination regimens to target bypass signaling tracks. Repeat biopsies following each disease relapse will also be instrumental for identifying the development of resistance mutations following a given targeted therapy and selecting the optimal therapeutic regimen.
In one study, 3 ceritinib-resistant, patient-derived cell lines that had ALK resistance mutations demonstrated sensitivity to lorlatinib, whereas the 3 cell lines that lacked ALK resistance mutations were insensitive to lorlatinib, suggesting loss of ALK dependency and resistance to ALK-targeted agents.26 These data highlight the importance of obtaining biopsies following each disease relapse to assess for ALK dependency and predict response to ALK inhibition.
The role of bypass pathway signaling in resistance suggests that combination approaches may be effective for patients who relapse on TKI therapy. The NEJ026 phase III study showed that adding the anti-VEGF therapy bevacizumab (Avastin) to erlotinib improved PFS over erlotinib alone (16.9 vs 13.3 months; HR, 0.605, although the recent approval of osimertinib in the frontline setting may make this regimen obsolete.32 In addition, an ongoing phase I/II trial (NCT02521051) is investigating the combination of alectinib and bevacizumab in patients with ALK-positive NSCLC and at least 1 CNS lesion, with the rationale that bevacizumab may alter systemic and intracranial drug activity by modulating tumor vasculature. Furthermore, an in vitro study showed that lorlatinib-resistant NSCLC cells acquired hyperactivation of EGFR and that administration of erlotinib resensitized the cells to lorlatinib, suggesting the potential for combined targeted therapy if these findings are maintained in the clinical setting.33
Finally, targeting persister cells that survive initial exposure to TKIs may be instrumental for preventing or delaying disease relapse. Persister cells have been implicated in the de novo development of heterogeneous EGFR TKI resistance, and the persister EGFR T790M-mutant cells were biologically distinct from preexisting T790Mmutant cells in that they exhibited a reduced apoptotic response to WZ4002, a third-generation EGFR inhibitor targeting T790M.34 However, the addition of the BCL-2 inhibitor navitoclax was shown to overcome this resistance, and an ongoing phase 1B trial (NCT02520778) is investigating osimertinib plus navitoclax in patients with previously treated EGFR-positive NSCLC.