Tackling Osimertinib Resistance in EGFR-Mutant Non–Small Cell Lung Cancer

The identification and understanding of molecular mechanisms that drive resistance to targeted therapies in epidermal growth factor receptor (EGFR)-mutant non–small cell lung cancer (NSCLC) have led to novel treatment strategies for patients. Osimertinib is a third-generation tyrosine kinase inhibitor (TKI) and, currently, a standard first-line therapy in treatment-naïve patients, and a second-line therapy in patients whose tumors acquire the gatekeeper resistance T790M mutation driving disease progression to first- or second-generation EGFR TKIs (1, 2). Unfortunately, as with other TKIs, eventually all patients will experience disease progression due to the onset of diverse molecular mechanisms of resistance in EGFR-mutant cancer cells. In this review, we underpin the biological mechanisms of resistance and the rational for innovative drug development strategies to overcome or prevent resistance.

Resistance to EGFR-targeted therapies can be explained by the late acquisition of de novo molecular alterations that drive resistance following prior response or stable disease, which is coined as secondary or acquired resistance; or less frequently the onset of primary progression due to the early selection and rapid proliferation of preexisting clones that harbor resistant alterations, which is defined as primary or innate resistance. At the molecular level, EGFR TKI resistance is mediated by “on-target” mechanisms, “off-target” mechanisms or by phenotypic plasticity. On-target resistance is driven mainly by the acquisition of secondary or tertiary EGFR kinase domain mutations that induce structural modifications in the receptor leading to impaired drug binding, with sustained EGFR kinase domain phosphorylation and downstream signaling (3). Off-target resistance results from the oncogenic activation of other effectors such as tyrosine kinase receptors and intracellular kinases that can be driven in many cases by underlying genomic alterations such as gene amplification, mutations, and rearrangements (4, 5).

Resistance mechanisms also differ whether osimertinib is given as a first-line therapy or in patients whose tumor acquired the EGFR T790M resistance mutation on treatment with a first- or second-generation EGFR TKI, in which tertiary mutations can occur in the same or different allele that T790M. Notably, most patients’ tumors acquire distinct resistance mechanisms either concurrently or in temporal sequence, denoting a high degree of heterogeneity within patient samples (Fig. 1; refs. 6, 7).

Primary or innate resistance is mainly related to the lack of target dependency (e.g., presence of preexisting nonsensitizing alterations), resulting in a rapid treatment relapse within the first three months after TKI initiation (8). In patients treated with osimertinib, primary treatment failure is strongly associated with MET amplification (9–11), leading to the bypass of EGFR downstream signaling pathways to promote cancer cell survival and proliferation.

Co-occurring mutations in other pathways might constitute modifiers of response to osimertinib, resulting in varying degrees of response to treatment in patients. Canonical EGFR driver mutations co-occur with oncogenic alterations in PIK3CA, BRAF, MET, MYC, CDK6, and CTNNB1. TP53 mutations (exons 5–8) are observed in about 60% of EGFR-mutant NSCLC patients (12). Particularly, mutations in exon 8 are associated with decreased disease control rate, progression-free survival (PFS), and overall survival (OS) in patients with EGFR exon 19 deletions (12). In preclinical models, genetic alterations in CTNNB1, PIK3CA, and cell-cycle genes (CDK4 and CDK6) limit EGFR-inhibitor treatment response; however, mutations in CTNNB1 or PIK3CA had no significant impact on osimertinib treatment (10, 12, 13). Furthermore, KEAP1 inactivation reduces osimertinib sensitivity in EGFR-driven tumors in vivo. Concordantly, mutations in the KEAP1/NFE2L2/CUL3 axis correlate with decreased therapeutic response to EGFR-targeted therapy in patients (13).

Following EGFR TKI treatment, EGFR-mutant tumor cells can enter a reversible drug-tolerant state characterized by decreased cell proliferation and reduced drug sensitivity, in the absence of resistance mutations. Drug-tolerant persister cells (DTP) might constitute the essence of residual disease and serve as a reservoir from which resistant cells may emerge (14).

Survival of DTP during chronic treatment is directly related to escape from cell death. In preclinical models, AURKA and AURKB inhibitors increase the magnitude and duration of EGFR-inhibitor response and overcome osimertinib resistance through the upregulation or the stabilization of BIM (15, 16). Similarly, the YAP/TEAD/SLUG complex cooperates to repress the proapoptotic factor BMF and thus suppress apoptosis in response to combined EGFR/MEK inhibition (17). Concordantly, treatment with the combination of WZ4002 (EGFR inhibitor) and navitoclax (BCL-2 inhibitor) induces enhanced tumor regression in vivo, as compared with WZ4002 alone (18). Activation of AXL and differential serine/threonine phosphorylation of IRS1 also promote DTP survival upon osimertinib treatment and constitute a molecular vulnerability of DTP in vitro and in vivo (19, 20).

EGFR-dependent resistance occurs in 10% to 15% of patients treated with first-line osimertinib (21). Mutations in the C797 residue, the site of covalent binding of osimertinib, are the most frequent EGFR-dependent mechanism of resistance to osimertinib (about 7% of patients; refs. 21–23). In patients whose tumors acquire EGFR T790M mutations in the setting of treatment with first- or second-generation EGFR TKIs and are treated with second-line osimertinib, the emergence of on-target resistance occurs in about 15% of cases, of which the EGFR C797S is also the most frequent resistance mutation. In this setting, the EGFR T790M and C797S can occur in the same allele (cis) or in a different allele (trans). EGFR C797S-mutant tumors are sensitive to a combination of first- and third-generation TKIs such as erlotinib and osimertinib in the absence of T790M (first-line osimertinib) or when C797S and T790M are in trans (second-line osimertinib; refs. 24, 25).

Less common mutations are found in ∼2% of the samples from patients who had disease progression to osimertinib (22). These mutations occur in the EGFR solvent front (e.g., G796X) or the hinge pocket (e.g., L792X), leading to steric hindrance to osimertinib, have been also identified as osimertinib resistance mechanisms (23, 25). Additional alterations in codons L718 and G719 (ATP-binding site) and G724 (P-loop domain) are involved in osimertinib resistance but are sensitive to first- and second-generation EGFR inhibitors (10, 21, 23). Recently, structure-based predictive modeling revealed that EGFR G724S emerges in the context of EGFR exon 19 deletions (Ex19Del) but not EGFR L858R mutations (26). In preclinical models, Ex19Del/G724S retains sensitivity to afatinib, providing a potential therapeutic rationale for patients at the time of relapse (Fig. 1; refs. 26, 27).

MET amplification is observed in 7% to 15% of patients following osimertinib treatment (21), while acquired ERBB2 amplification accounts for 2% to 5% of patients (21). A recent phylogenetic analysis of pretreatment and osimertinib-resistant tumors revealed that acquired focal copy-number alterations (FCNA) occur mainly subclonally and earlier in tumor evolution. FCNA are associated with a short-term osimertinib response (<12 months; ref. 6). MET amplification is clinically actionable, and combined MET/EGFR inhibition has shown encouraging efficacy in EGFR-mutant NSCLC patients presenting MET amplification at disease progression on first- to third-generation EGFR TKI (see below; refs. 28–30). MET bypass track resistance may also rarely occur by the acquisition of MET exon 14 skipping alterations (31).

Genetic alterations in downstream effectors of the EGFR signaling pathway are also associated with osimertinib resistance. Mutation or amplification of PIK3CA constitute identified mechanisms of resistance in 3% to 7% patients who progress on osimertinib (21). The RAS–MAPK pathway can be reactivated by mutations in KRAS or in BRAF, which have been identified in 3% to 5% of patients progressing on osimertinib (7, 21). Combined EGFR and MEK inhibition successfully reverted the KRAS^G12S-mediated resistance, in vitro (9). In cases with BRAF-driven osimertinib resistance, recent clinical evidence has shown promising results with the combination of osimertinib with dabrafenib and trametinib, or with trametinib alone (32, 33).

Additionally, gene fusion involving driver tyrosine kinase domains of key oncogenes (e.g., RET, ALK, ROS1, BRAF, and FGFR) has been identified in about 10% of patients at disease progression on osimertinib (34). Some of these fusions are known to represent actionable targets to successfully bypass osimertinib resistance (e.g., CCDC6-RET, FGFR3-TACC3, and STRN-ALK; refs. 7, 35). Interestingly, most of the fusions detected by DNA-based sequencing at disease progression to EGFR-targeted therapy do not result in functional fusion oncogenes that promote drug resistance, which may pose a challenge for genomic profiling report interpretation and assessing drug actionability in the clinical setting (34).

Alterations in cell-cycle regulators have been detected at osimertinib resistance in 10% to 12% of cases (21). In particular, CDK4/6 or CCND/E1 amplifications or CDKN2A loss are associated with worse patient outcomes (36). Other EGFR-independent mechanisms include increased β-catenin signaling and activation of Notch, Stat3, NF-κB, or c-Myc, among others (Fig. 1; refs. 37–41).

Phenotypic transformation into different histologic subtypes leads to target independence. In EGFR-mutant lung adenocarcinomas (LUAD) cases, histologic transformation into small cell lung cancer (SCLC) represents up to 15% of patients presenting disease progression on osimertinib (42). Resistant tumors retain the EGFR-activating mutation and emerge from LUAD with preexisting concurrent RB1 and TP53 mutations, which are associated with a 43-fold increased risk of SCLC transformation (42). In these patients, treatment with systemic chemotherapy has shown limited efficacy (43).

Transformation to other histologic subtypes (e.g., squamous-cell carcinoma, large-cell neuroendocrine carcinoma) was detected in patients relapsing on osimertinib (44–46). Like cases of SCLC transformation, these patients present poorer OS compared with EGFR-mutant LUAD (45). At the molecular level, concurrent activation of PI3K/AKT and MYC induces squamous features in preclinical models of EGFR-mutant LUAD. In this setting, combined pharmacologic inhibition of EGFR and either EZH1/2 or AKT in vivo resensitizes resistant squamous-like tumors to osimertinib (47).

Epithelial-to-mesenchymal transition (EMT) is a conserved embryonic process during which polarized epithelial cells lose cell–cell junctions and polarity and switch to a motile mesenchymal phenotype. Aberrant reactivation of EMT promotes cancer cell plasticity and fuels resistance to therapy (48). In EMT-mediated osimertinib resistance, activation of the ATR–CHK1–Aurora B signaling cascade leads to BIM-mediated mitotic catastrophe. In this setting, combined pharmacologic EGFR/AURKB inhibition overcomes osimertinib resistance (Fig. 1; ref. 15).

In the clinical setting, studying resistance mechanisms to EGFR TKIs is most relevant for deciding on therapeutic options for patients experiencing progression to osimertinib, in a comprehensive and personalized strategy. There are several factors to consider at the time of resistance, regarding the type of sample (plasma and/or tissue) and molecular biology technique to use according to the patient's characteristics and disease patterns.

Next-generation sequencing (NGS) of circulating free DNA (cfDNA) is a minimally invasive tool for genomic profiling of solid tumors and particularly lung cancer (49). From a clinical point of view, cfDNA sequencing in patients with lung cancer is used mainly as genomic profiling for the detection of driver alterations, detection of co-occurring alterations that can drive primary or acquired resistance, and other incidental findings (i.e., pathogenic germline variants) with potential clinical impact currently under investigation (50). Circulating tumor DNA (ctDNA) is a portion of all cfDNA, shedding into the bloodstream from multiple metastatic sites, representing the heterogeneity of resistance mechanisms arising at the time of osimertinib progression, and is the preferred method to initiate the study of resistance mechanisms according to clinical guidelines and recommendations (51, 52). In terms of resistance mechanisms, cfDNA NGS is highly sensitive to detect on- and off-target mutations when plasma cfDNA content is adequate, this technique may be less sensitive for fusion detection and gene amplifications, depending on the assay and allelic frequency, but should be considered if tissue biopsy is not feasible (53–56). Recently, preliminary data of the ELIOS Phase 2 (NCT03239340) study was presented, assessing underlying resistance mechanisms of upfront osimertinib, only 39% of patients had paired plasma and tissue biopsies, highlighting the need for more comprehensive noninvasive testing methods (57).

Some of the factors to consider at the time of requesting cfDNA NGS that can increase the likelihood of false-negative results, due to low tumor DNA shedding below the assay's technical limit of detection, are low tumor burden, only central nervous system (CNS) disease or progression, blood sample hemolysis, and inadequate preanalytical sample conditions (58, 59).

If cfDNA NGS is not informative of resistance mechanisms, then tissue biopsy of a progressive representative metastatic site followed by NGS is recommended as it is a highly sensitive method in a representative tumor sample and is also the only method to diagnose histologic transformation. Comutations of TP53 and RB1 genes in EGFR-mutant lung cancer detected at baseline or at progression with osimertinib should prompt tissue biopsy as the patient is at high risk of SCLC transformation. On the downside, tissue NGS does not capture information on the heterogeneity of resistance genomic mechanisms.

Future clinical applications for cfDNA NGS is plasma monitoring of EGFR mutation detection dynamics during treatment which can be used to predict response or progression to EGFR TKIs, emerging as an important and independent factor associated with clinical outcome (50, 60–62). However, this strategy is not currently being applied worldwide in daily clinical practice (54, 60).

Overcoming on-target osimertinib resistance due to the acquisition of secondary (in first line) or tertiary mutations (for EGFR T790M resistance) is challenging, but there are significant advances in this field. In resistance by tertiary mutations, assessing the allelic distribution by NGS of the T790M and C797S mutations can be relevant. In patients with EGFR-mutant cancer that harbor both mutations in trans (different alleles), sensitivity to EGFR inhibitors could be restored by combining first- and third-generation TKIs (24, 63–65).

More importantly, fourth-generation EGFR inhibitors that bind to the kinase domain in the setting of EGFR C797S mutations are currently in development, including allosteric inhibitors such as JBJ-04-125-02 and EAI001 (binding outside of the ATP-binding pocket) and ATP-competitive inhibitors, that can spare wild-type EGFR, such as BLU-945, BLU-525, BBT-176, and OBX02-011 (66–69). Both allosteric inhibitors were designed to bind and inhibit EGFR L858R/T790M/C797S-mutant cells; however, EAI001 needs to be combined with cetuximab to convey cytotoxic activity in vivo, where JBJ-04-125-02 achieved significant potency in vitro and in vivo as monotherapy, and clinical development of this compound is currently being pursued (67, 68).

ATP-competitive inhibitors have a broader spectrum of targetable alterations. Several reversible, selective, and wild-type EGFR-sparing inhibitors with high CNS penetration are in development aiming to overcome on-target resistance in two scenarios: BLU-945 binds to EGFR in the setting of exon 19 deletions (ex19del) or L858R mutation in cis with T790M/C797S; and BLU-701 binds to EGFR exon 19 deletions or L858R-mutant receptors that also harbor only C797S as seen at progression with first-line osimertinib, where there is no emergence of EGFR T790M mutations (70). BLU-525 is also a new-generation EGFR TKI with a similar target profile as BLU-701 with in vitro activity and high CNS penetration in murine models (71). Given the different selectivity profiles of these TKIs, potential combinations with osimertinib or between them could enhance clinical efficacy and prevent on-target resistance in the first-line setting. BLU-945 is currently being studied in phase I/II SYMPHONY (NCT04862780; Fig. 2 and Table 1). BBT-176 is an ATP-competitive EGFR TKI designed to target EGFR ex19del/T790M/C797S mutations but not L858R/T790M/C797S-mutant cells in vitro and in vivo. Results from the phase I first-in-human trial showed early signs of adequate tolerance and response in patients with the EGFR ex19del/T790M/C797S (72).

In addition to effectively inhibiting the EGFR kinase domain, targeting the extracellular domain of EGFR with monoclonal antibodies is an emerging treatment strategy to overcome osimertinib resistance. Amivantamab is a bispecific antibody that binds to the extracellular domain of both EGFR and MET. In the phase I trial, amivantamab showed early signs of activity as monotherapy in patients previously treated with third-generation EGFR TKIs with an objective response rate (ORR) of 21% (n = 47; ref. 73). Currently, in the setting of osimertinib resistance, it is being clinically developed in combination with the third-generation TKI lazertinib that can maintain EGFR-mediated responses in cancer clones sensitive to osimertinib, and improve brain metastasis control, as a monoclonal antibody like amivantamab has low CNS penetration.

The current clinical evidence on the efficacy of this combination is supported by the CHRYSALIS study (NCT02609776), including patients previously treated with third-generation EGFR TKI who are chemotherapy treatment naïve, and in patients who had also prior chemotherapy (CHRYSALIS-2, NCT04077463; refs. 74, 75). In the CHRYSALIS trial (n = 45), in a biomarker unselected population, the ORR of amivantamab plus lazertinib was 36%, and median PFS was 4.9 months. Importantly, patients with tumors that harbored on-target EGFR resistance mutations or MET amplification (n = 17) had improved benefit with amivantamab, with an ORR of 47%, median duration of response (DOR) of 10.4 months, and PFS of 6.7 months. Per contrary, among 10 patients with detectable bypass alterations other than MET, including in the RAS/RAF and MTOR pathways, cell-cycle gene mutations, and ALK fusions; none experienced an objective response with this combination therapy (75). In the CHRYSALIS-2 trial, in chemotherapy and osimertinib-pretreated biomarker unselected population (n = 162), the ORR was 36%, PFS of 5.1 months, and median OS of 14.8 months (74). (Fig. 2 and Table 2.)

Using other combinations, such as the EGFR-directed monoclonal antibody necitumumab with osimertinib, has shown low activity in two trials, including the ORCHARD in which recruitment was stopped for futility in biomarker selected population (EGFR amplification, L718 or G724 mutation, exon 20 insertion) with 15% ORR (n = 16; refs. 65, 76, 77).

Targeting off-target osimertinib resistance is challenging as it requires to identify the molecular bypass mechanism to deliver a targeted therapy, while sustaining EGFR inhibition. This is the rationale behind the biomarker-directed phase II ORCHARD platform trial (NCT03944772), in which molecular testing is done in biopsies from patients with EGFR-mutant NSCLC who experienced progression to first-line osimertinib. Participants are assigned to receive osimertinib with an additional biomarker-guided targeted therapy for MET, ALK, BRAF, and RET bypass alterations (75–77).

In MET-mediated resistance, several combinations with EGFR-MET TKIs are being developed. In the TATTON trial, osimertinib was studied in combination with the selective type Ib MET inhibitor savolitinib in patients (n = 69) with tumors that acquired MET amplification (defined as MET gene copy number ≥5 or MET–CEP7 ratio ≥2) conveying a 33% ORR with a PFS of 5.5 months (28). Additionally, the phase II SAVANNAH trial included 109 patients whose tumors had high MET overexpression, defined as MET IHC staining in ≥90% of cells by immunohistochemistry (IHC90+) or MET amplification with copy number (GCN) ≥10 (FISH10+). The combination of osimertinib and savolitinib resulted in superior ORR in comparison with patients without MET high criteria (49% vs. 9%) and PFS (7.1 months vs. 2.8 months; Table 2; ref. 78). The multiarm ORCHARD trial previously mentioned also supports the results from the TATTON trial, reporting an ORR of 41% (CI 95%: 25–59, n = 17) with the osimertinib and savolitinib combination (79).

In a similar design, the selective type Ib MET inhibitor tepotinib was studied in combination with osimertinib after progression to osimertinib treatment in patients with tumors that acquired MET amplification (defined as GCN ≥5 or MET/CEP7 ratio ≥2 by FISH or NGS) in the INSIGHT-2 trial. This combination resulted in an ORR of 54.5% in patients (n = 70) with MET-amplified tumors (29). The combination of osimertinib and MET TKIs has also shown clinical activity in overcoming osimertinib resistance by the acquisition of MET exon 14 skipping in case reports (31, 80). In addition to TKIs, amivantamab has also proven activity in combination with lazertinib in patients with tumors that acquire MET amplification as mentioned previously (see section “On-target resistance”; ref. 75).

In the setting of acquired resistance to osimertinib by HER2 amplification, there is currently no effective therapy to treat patients. Attempts to combine the antibody–drug conjugate (ADC) trastuzumab-emtansine (TDM1) with osimertinib (81), the phase II TRAEMOS trial reported a modest 13% ORR and PFS of 2.8 months in 27 patients (82). Trastuzumab deruxtecan, an ADC approved as a standard therapy for patients with NSCLC that harbor HER2 exon 20 insertions, is active in a subset of patients with HER2-expressing tumors (HER2 2+ or 3+) and needs further study in the setting of EGFR resistance (Table 2; ref. 83).

The emergence of targetable oncogenic fusions in EGFR-mutant cancer cells, under selective pressure with osimertinib, that involve the tyrosine kinase domain of ALK, ROS1, RET, NTRK, and FGFR3, and serine-threonine kinases such as BRAF can be challenging to diagnose and treat in the clinical setting. Like with other bypass mechanisms of resistance, targeting both the acquired oncogenic effector together with EGFR is required. This can be achieved by combining osimertinib with selective kinase inhibitors as published in individual case reports (84–89). Currently, this strategy is being studied in the ORCHARD trial including the combination of osimertinib and alectinib in patients with tumors that acquire ALK fusions, osimertinib with selpercatinib in patients with acquired RET fusions, and combined with selumetinib in acquired BRAF fusions (Table 1).

In the setting of osimertinib resistance, the delivery of cytotoxic agents through antibody–drug conjugates (ADC) does not rely on targeting the underlying genomic alterations that drive clonal selection and resistance, but on the expression of target proteins in the cell membrane (90). Several membrane receptors are highly expressed in lung cancer cells including HER3, MET, and TROP2, for which selected ADCs are being developed.

HER3 overexpression occurs in about 85% of EFGR-mutant NSCLC. Patritumab deruxtecan (HER3-DXd) is an ADC directed against HER3 with deruxtecan, an exatecan derivative (topoisomerase I inhibitor) payload, with proven clinical efficacy in patients with EGFR-mutant NSCLC with disease progression to EGFR TKIs, including osimertinib. In the phase I trial including 44 patients, the ORR with this ADC was 39% and PFS was 8.2 months in patients previously treated with osimertinib (91). Interestingly, the activity of patritumab deruxtecan was observed across different on- and off-target mechanisms, and the development of this compound will continue as monotherapy in the HERTHENA-Lung01 phase II trial (NCT04619004) and also in combination with osimertinib (NCT04676477), aiming to maintain EGFR inhibition.

Similarly, datopotamab deruxtecan (Dato-DXd), an ADC directed against Trophoblast cell-surface antigen 2 (TROP2), is a highly expressed transmembrane glycoprotein in lung cancer (92). Treatment with Dato-DXd was studied in a subgroup of patients with tumors that harbored actionable genomic alterations (AGA) in the Tropion-PanTumor01 trial, most patients with EGFR-mutant lung cancer, and a few with ALK- and ROS1-rearranged cancers. In the preliminary analysis (n = 34), the ORR was 35% and the median DOR was 9.5 months (93).

Telisotuzumab vedotin (Teliso-V) is an antibody–drug conjugate directed against the MET extracellular domain delivering the antimicrotubular agent monomethyl auristatin E. MET overexpression is common in lung cancer cells, present in about 37% of EGFR-mutant tumors, and overexpression does not overlap necessarily with oncogenic MET activation mechanisms such as MET exon 14 skipping (5% of cases) or MET amplification (1% de novo or 5%–15% at osimertinib progression), though MET amplification conveys high levels of MET expression in lung cancer (94). Teliso-V given as monotherapy has scant activity in patients with EGFR-mutant NSCLC experiencing progression on osimertinib; however, a recent report of a phase I trial combining osimertinib and Teliso-V shows promising efficacy (NCT02099058; refs. 95, 96). Patients with EGFR-mutant tumors that had intermediate (MET IHC 3+ in 25%–49% of cells) or high-level of MET expression (MET IHC 3+ ≥ 50% of cells) in archival or fresh tissue biopsies and had progressed on osimertinib treatment (n = 25) were included (95). In the preliminary analysis, the combination had a similar toxicity profile than reported for Teliso-V in monotherapy, with an elevated ORR of 58%, which is higher than that reported with Teliso-V alone in this setting. These encouraging preliminary results can be potentially explained by the combination of Teliso-V with osimertinib, in which EGFR inhibition is restored or sustained in EGFR-driven tumors, in a biomarker-selected population. Further data are required to understand this ADC activity according to the underlying resistance mechanism, including a special interest in the impact on MET-driven resistance.

Preventing the emergence of osimertinib resistance is a key objective in trial development. Preclinical and clinical data suggest that combining EGFR TKIs with chemotherapy may act synergistically to restrict the development of acquired resistance (97, 98). In two randomized phase III studies, PFS and OS were statistically and clinically significantly prolonged with gefitinib/chemotherapy combination versus gefitinib in monotherapy at the expense of increased toxicity (99–101). The FLAURA-2 (NCT04035486) is a randomized phase III trial currently addressing the efficacy of platinum pemetrexed combination with osimertinib compared with osimertinib monotherapy as first-line treatment. Run-in data for this trial with a mean duration of osimertinib exposure of 3.81 months and duration of pemetrexed of 4.14 months, shows a similar rate of reported serious adverse events (102).

The TOP study (NCT04695925) is designed to study efficacy and safety of osimertinib and chemotherapy combination in patients with concurrent EGFR and TP53 mutations, which is a molecularly identified high-risk group of poor outcomes with monotherapy.

Combining targeted therapies is also being studied in the first-line setting, like the combination of amivantamab and lazertinib in the MARIPOSA study (NCT04487080), compared with single-agent osimertinib. Data from the treatment-naïve cohort of the CHRYSALIS demonstrated preliminary benefit and safety with combination, including a significant proportion of patients with concurrent EGFR/TP53 mutations (103). In this small cohort, the ORR was 100%, and grade ≥3 adverse events occurring in 5 patients (25%), one patient with interstitial lung disease (103).

Preventing resistance mediated by MAPK signaling pathway activation is also being studied. There is preclinical data supporting the combination of osimertinib with the MEK inhibitor selumetinib as a synergistic strategy to promote apoptosis and delay the emergence of osimertinib resistance both in vitro and in vivo (104). Recently, the TATTON Part B trial demonstrated that using osimertinib plus selumetinib in previously treated patients achieved an ORR of 34% and a PFS of 4.2 months (n = 47). This combination demonstrated antitumor activity supportive of further investigation in patients with MET-negative, EGFR-mutant advanced NSCLC who had progressed on a previous EGFR TKI or even in the first line (105).

The discovery of natural products offers new scaffolds for drug development. Honokiol (HNK) is a natural product purified from Magnolia with potential antitumor activity against different types of cancer. HNK (67) and its derivative Caz-p in combination with osimertinib, effectively reduced the survival and induced the apoptosis of EGFR ex19del/C797S (trans) double-mutant PC-9/2 M cells and EGFR ex19del/T790M /C797S PC-9/3 M cells (106). It is highly encouraging that HNK (65) and its derivatives may overcome clinical resistance to third-generation TKIs (67). With a similar perspective, bufalin a bufadienolide analogues with potent MCL-1 inhibitor has been combined with osimertinib to decrease the emergence of resistance (107, 108).

The introduction of immune-checkpoint inhibition (ICI) has changed the treatment landscape in patients with advanced NSCLC, resulting in unprecedented survival benefits. Unfortunately, NSCLCs harboring EGFR mutations failed to benefit from PD-(L)1 blockade alone in initial phase III clinical trials of second-line immunotherapy versus standard-of-care docetaxel (109–111), and subsequent retrospective series confirmed a limited efficacy of ICI among patients with advanced EGFR-mutant NSCLC, with responses reported in less than 10% of cases (112, 113). In addition, a recent phase II study of pembrolizumab in TKI-naïve patients with EGFR mutation-positive, advanced NSCLC and PD-L1–positive (≥1%, 22C3 antibody) tumors ceased enrollment due to lack of efficacy after 11 of 25 planned patients were treated and no objective responses were noted, including among the subset with PD-L1 expression levels ≥50% (114).

There is scant evidence to support immunotherapy combinations in EGFR-mutant NSCLC. The IMpower 150 trial is a phase III study that randomized patients with NSCLC to receive atezolizumab in combination with platinum-based chemotherapy ± bevacizumab or chemotherapy alone. In a subgroup analysis of this trial, patients with EGFR/ALK driven NSCLC in the chemoimmunotherapy–bevacizumab arm had significant improvement in PFS (9.7 vs. 6.1 months, unstratified HR: 0.59) compared with those in the chemotherapy arm with no significant difference in OS between the treatment groups (114).

The ORIENT-31 is a phase III trial specifically designed to address the efficacy of sintilimab, an anti–PD-1 monoclonal antibody, with the bevacizumab biosimilar IBI305 (anti-VEGF antibody) in patients with EGFR-mutant lung cancer experiencing disease progression to EGFR TKIs. About 28% of patients had previously received a third-generation TKI after EGFR T790M mediated resistance to first- or second-generation inhibitors and 8.1% received first-line treatment with a third-generation TKI. Participants were randomized to receive sintilimab plus IBI305 with pemetrexed and cisplatin, sintilimab plus pemetrexed and cisplatin, or pemetrexed and cisplatin. The PFS was improved with sintilimab, IBI305, and chemotherapy compared with the chemotherapy-alone group (median PFS 6.9 months vs. 4.3 months; P < 0·0001). More data on OS are expected, but this preliminary interim analysis of the ORIENT-31 trial may further support the combination of immune-checkpoint inhibitors with antiangiogenic agents and chemotherapy in the setting of EGFR TKI refractory disease (115).

The CheckMate 722 trial compared the combination of chemotherapy with nivolumab to chemotherapy alone in patients experiencing disease progression on prior EGFR TKIs. No differences in PFS were observed between arms, highlighting the limited role of anti–PD-1 inhibitor and chemotherapy combination without antiangiogenic agents in this disease (116).

There are several potential reasons that explain why immunotherapies alone are not effective in EGFR-mutant NSCLC, including lower tumor mutational burden (TMB; ref. 116) and lower PD-L1 tumor proportion score (TPS) (117, 118), which likely reflects the lack of tobacco exposure and the unique genomic profile of these tumors. In addition, EGFR mutations induce an immunosuppressive tumor microenvironment, by decreasing tumor-infiltrating CD8+ T cells, and increasing Tregs and myeloid-derived suppressor cells (MDSC; ref. 119). Together, these data highlight the need for novel and more effective treatment strategies, for patients with EGFR-mutant NSCLC who have progressed on currently available standard-of-care treatments.

Adoptive cell therapy represents a novel treatment strategy that leverages engineered T cell to recognize and kill tumor cells expressing a tumor-specific antigen. There are currently different types of cell therapies that are either approved or under investigation in patients with cancer, including chimeric antigen receptors (CAR-T), engineered T-cell receptor (TCR), and natural killer cell therapy (120). In NSCLC, potential targets for adoptive cell therapy include EGFR, HER2, mesothelin (MSLN), mucin 1 (MUC1), PD-L1, and CEA, among others (121). Anti-EGFR CAR-T therapies are being explored against the EGFR type III (EGFRvIII) variant resulting from the in-frame deletion of exons 2 to 7 and the generation of a novel glycine residue at the junction of exons 1 and 8 in xenogeneic mouse models of glioblastoma and lung cancer (122–124), EGFRvIII is an EGFR variant with a large deletion in its extracellular domain, which is not shared with wild-type EGFR. By contrast, sensitizing EGFR mutation in lung cancer (e.g., exon 19 deletions, L858R, T790M, C797S) occur in the intracellular kinase domain, and the extracellular structure of these EGFR-mutant receptors is identical compared with EGFR wild-type. This makes CAR-T a challenging approach for this specific subset of EGFR-mutant NSCLC.

TCR-T–based therapies use a natural or minimally engineered TCR to develop T-cell–based adoptive T-cell therapy. TCR-T therapies recognize tumor-specific epitopes bound to MHC, and therefore have broader applicability compared with CAR-T because there are significantly more tumor-unique sequences that are processed and presented via MHC, than potential antigens on the cell surface (125). TCR-T therapy directed against other intracellular oncogenes, such KRAS G12D, have produced dramatic responses in patients with metastatic KRAS-mutant colorectal and pancreatic cancers, suggesting potential for long-term efficacy of this treatment approach in patients with HLA-restricted neoantigens from oncogene drivers (126, 127). Interestingly, in a phase I study of a personalized neoantigen vaccination that included 24 patients with NSCLC, of whom 16 had EGFR-mutant disease, all seven responders harbored EGFR-mutant tumors (128). Immune monitoring demonstrated that 5 of 7 responding patients had strong vaccine-induced T-cell responses against EGFR neoantigens. In addition, two highly shared EGFR mutations (L858R and T790M) were found to be immunogenic in four of the patients who responded, and in all cases, significant increases in peripheral blood neoantigen-specific CD8⁺ T-cell frequencies were observed during the trial. This suggests that EGFR-mutant epitopes can potentially elicit antitumor T-cell responses and that TCR-T–based approach may be explored in this specific patient population.

Resistance mechanisms to osimertinib are heterogeneous and compromise both on-target and off-target mechanisms. Detecting these mechanisms in patients at the time of progression with osimertinib can help design novel treatment strategies, many of which show proven safety and efficacy. Further research and drug development is required to prevent the onset of resistance to third-generation EGFR TKIs and to evoke immune responses against EGFR-mutant cancer cells in this setting.

J.B. Blaquier reports other support from Amgen and Pfizer outside the submitted work. B. Ricciuti reports personal fees from Regeneron outside the submitted work. L. Mezquita reports grants, personal fees, nonfinancial support, and other support from lectures and educational activities from Bristol-Myers Squibb, AstraZeneca, Roche, Takeda, Janssen, and Pfizer; a consulting/advisory role for Roche, Takeda, and Janssen; research grants from Bristol-Myers Squibb, Boehringer Ingelheim, Amgen, Stilla, Inivata, and AstraZeneca; and travel, accommodations, and expenses from Bristol-Myers Squibb, Roche, Takeda, AstraZeneca, and Janssen outside the submitted work. A.F. Cardona reports grants, personal fees, and nonfinancial support from AstraZeneca, Roche, Janssen, Boehringer Ingelheim, Foundation Medicine, Thermo Fisher Scientific Inc, and BMS; grants and personal fees from Takeda; personal fees and nonfinancial support from Pfizer, AbbVie, Celldex, and Illumina; personal fees from Eli Lilly; grants, personal fees, nonfinancial support, and other support from Foundation for Clinical and Applied Cancer Research and Merck Sharp & Dohme; and other support from Rochem Biocare during the conduct of the study as well as grants from Guardant; personal fees from Eisai and Teva; personal fees, nonfinancial support, and other support from Novartis and Merck Serono; grants and nonfinancial support from Idylla; grants, personal fees, and nonfinancial support from Bayer; and grants, personal fees, nonfinancial support, and other support from Amgen outside the submitted work. G. Recondo reports grants from Amgen and Janssen and personal fees from Amgen, AstraZeneca, Bayer, Biocartis, Bristol Meyers Squibb, Janssen, Merck Serono, Merck Sharp & Dohme, Pfizer, Roche, and Takeda outside the submitted work. No disclosures were reported by the other author.

L. Mezquita received support from the Contrato Juan Rodes 2020 (JR20/00019; ISCIII, Ministry of Health); Ayuda de la Acción Estratégica en Salud-ISCIII FIS 2021 (PI21/01653); Ayuda SEOM-Juan Rodés 2020.