Third-generation EGFR and ALK inhibitors: mechanisms of resistance and management

Abstract

The discoveries of EGFR mutations and ALK rearrangements as actionable oncogenic drivers in non-small-cell lung cancer (NSCLC) has propelled a biomarker-directed treatment paradigm for patients with advanced-stage disease. Numerous EGFR and ALK tyrosine kinase inhibitors (TKIs) with demonstrated efficacy in patients with EGFR-mutant and ALK-rearranged NSCLCs have been developed, culminating in the availability of the highly effective third-generation TKIs osimertinib and lorlatinib, respectively. Despite their marked efficacy, resistance to these agents remains an unsolved fundamental challenge. Both ‘on-target’ mechanisms (largely mediated by acquired resistance mutations in the kinase domains of EGFR or ALK) and ‘off-target’ mechanisms of resistance (mediated by non-target kinase alterations such as bypass signalling activation or phenotypic transformation) have been identified in patients with disease progression on osimertinib or lorlatinib. A growing understanding of the biology and spectrum of these mechanisms of resistance has already begun to inform the development of more effective therapeutic strategies. In this Review, we discuss the development of third-generation EGFR and ALK inhibitors, predominant mechanisms of resistance, and approaches to tackle resistance in the clinic, ranging from novel fourth-generation TKIs to combination regimens and other investigational therapies.

Introduction

The evaluation and treatment of patients with non-small-cell lung cancer (NSCLC) has evolved dramatically over the past two decades with the discovery of multiple oncogenic driver alterations and the development of matched molecularly targeted therapies^1-3. The experience with somatic activating mutations in EGFR, detected in approximately 15% of lung adenocarcinomas^4-7, provided the first demonstration that a subset of NSCLCs harbour ‘actionable’ driver oncogenes that confer exquisite sensitivity to targeted therapy using tyrosine kinase inhibitors (TKIs). Chromosomal rearrangements involving ALK were subsequently identified as the oncogenic driver in a further 3–7% of patients with NSCLC^8,9. In the intervening years since the discoveries of EGFR and ALK-driven NSCLCs, increasingly effective targeted small-molecule inhibitors have been developed, culminating in the FDA approvals of highly potent and central nervous system (CNS)-penetrant, third-generation EGFR and ALK TKIs. In parallel, lessons learned from EGFR-mutant and ALK-rearranged NSCLC have informed the paradigm of targeted therapy more generally, providing a framework for understanding the burgeoning number of molecularly defined subsets of NSCLC. In this Review, we provide an overview of the development of EGFR and ALK targeted therapies with a focus on third-generation agents, including the predominant mechanisms of resistance, strategies for overcoming resistance and future research directions.

EGFR mutations in NSCLC

EGFR is a member of the ErbB family of receptor tyrosine kinases (RTKs). Binding of ligands to the extracellular domain induces a conformational change that leads to receptor dimerization, tyrosine phosphorylation and activation of downstream signalling pathways including RAS–MAPK, PI3K–AKT and JAK–STAT, resulting in cell growth and survival¹⁰. Although EGFR is expressed in a range of nonmalignant tissues, including those of the skin and gastrointestinal tract¹¹, it is overexpressed across multiple epithelial cancers 5-fold or greater ¹², including NSCLC^13,14, and thus has long been considered a therapeutic target. Yet, the initial development of effective EGFR-targeted therapies was not straightforward.

Early generation EGFR TKIs

The first-generation reversible EGFR inhibitors, gefitinib and erlotinib, were initially tested in unselected patients with advanced-stage NSCLC in the early 2000s and demonstrated only modest activity (objective response rates (ORRs) of around 10–20% and median progression-free survival (PFS) durations of around 3 months)^15-18, although increased response rates were observed among never-smokers and patients of Asian ethnicity. Subsequent translational studies of exceptional responders led to the discovery of somatic oncogenic alterations in the EGFR kinase domain that were enriched in both of these populations^5-7, thus explaining the previously noted clinical correlates of response. The most common types of activating EGFR mutations were found to be a family of deletions in a small region of exon 19 or the L858R point mutation, both of which confer a 100-fold higher sensitivity to EGFR TKIs compared to the wild-type receptor^5,19. Prospective trials evaluating gefitinib or erlotinib specifically in patients with EGFR-mutant advanced-stage NSCLC soon followed, and the efficacy observed in this population was, at the time, unprecedented (typical ORRs of ~75% and median PFS durations of ~7-12 months). The rapid, deep tumour responses seen in these early studies offered one of the first glimpses into the potential of targeted therapy for NSCLC^4,20-27. Thereafter, randomized clinical trials from the early 2010s demonstrated clearly superior ORRs (typically around 75% versus <40%) and median PFS durations (typically 8–10 months versus 4–8 months) with first-generation EGFR TKIs compared to chemotherapy^28-34, establishing EGFR TKIs as the preferred first-line therapy for patients with EGFR-mutant NSCLC (FIG. 1).

Figure 1. Timeline of genomic discoveries and drug development in EGFR-mutant and ALK-rearranged NSCLCs.

Key events, in a. ∣ EGFR-mutant non-small-cell lung cancers (NSCLC) and b. ∣ ALK-rearranged NSCLC, including the discovery of the oncogenic drivers and gene alterations (orange), FDA approvals of tyrosine-kinase inhibitors for unselected patients (grey), approvals of therapies for biomarker-selected populations in the advanced-stage disease setting (blue), approvals for biomarker-selected populations in the adjuvant setting (green), and important ongoing trials (purple).

Despite excellent initial responses, acquired resistance to first-generation EGFR inhibitors developed, usually within a year. The most abundant mechanism of resistance, seen in ~60% of patients, was the T790M mutation located at the gatekeeper residue of EGFR^35,36, which enhances the affinity of the kinase for ATP and thus limits opportunities for erlotinib and gefitinib to bind³⁷. Subsequent efforts to overcome EGFR^T790M-driven resistance led to the development of second-generation and third-generation EGFR TKIs. The second-generation, irreversible pan-ErbB inhibitors afatinib and dacomitinib were predicted to have activity against EGFR^T790M based on data from preclinical models. However, this early promise was not borne out in clinical trials^38,39, potentially owing to inhibition of wild-type EGFR, resulting in dose-limiting dermatological and gastrointestinal toxicities that might have abrogated the ability to achieve the drug concentrations required to overcome EGFR^T790M. Both TKIs were ultimately approved by the FDA as first-line therapies (owing to improvements in median overall survival (OS) of about 8 months compared with first-generation EGFR TKIs)^40,41, although the role of these agents is now unclear given the successful development of osimertinib.

Osimertinib

Osimertinib is a third-generation, irreversible EGFR TKI with greater potency and selectivity against both the canonical EGFR activating mutations and the T790M resistance mutation⁴². Despite being a substrate of several drug efflux transporters, such as permeability glycoprotein (Pgp) and breast cancer resistance protein (BCRP), the reduced affinity of osimertinib for these transporters enables greater retention within the cerebrospinal fluid after crossing the blood-brain barrier, resulting in greater CNS activity relative to earlier-generation EGFR TKIs^43,44. When compared with standard chemotherapy in patients with T790M-driven acquired resistance to first-generation EGFR TKIs, osimertinib demonstrated a superior ORR (71% versus 31%) and PFS (10.1 months versus 4.4 months), including in patients with CNS metastases (median PFS 8.5 months versus 4.2 months)⁴⁵. Osimertinib was granted accelerated FDA approval in 2015 for patients with EGFR^T790M -positive NSCLC following disease progression on a prior EGFR TKI and subsequently received regular approval in 2017.

Subsequently, osimertinib was compared with gefitinib or erlotinib as first-line therapy for patients with advanced-stage EGFR-mutant NSCLC in the pivotal FLAURA trial, demonstrating improved median PFS (18.9 months versus 10.2 months)⁴⁶, a lower incidence of CNS progression (in 6% versus 15% of patients)⁴⁶ and prolonged OS (median 38.6 months versus 31.8 months)⁴⁷. Osimertinib is now established as the standard-of-care first-line therapy for patients with advanced-stage EGFR-mutant NSCLC (FIG. 1). Other third-generation EGFR TKIs have been developed, with nazartinib (NCT02335944), lazertinib (NCT04248829, NCT04077463, NCT04487080) and aumolertinib (NCT04923906) in ongoing clinical trials, while the development of rociletinib, mavelertinib, avitinib and naquotinib has been halted⁴⁸.

ALK rearrangements in NSCLC

ALK rearrangements were first identified in patients with NSCLC in 2007⁸, more than a decade after the initial discovery of the NPM1–ALK rearrangement in patients with anaplastic large cell lymphoma⁴⁹. Since the initial report of EML4–ALK rearrangements in patients with NSCLC⁸, >90 ALK fusion partners have been identified⁵⁰ across 3–7% of all NSCLCs^9,51,52. Mechanistically, ALK rearrangements result in constitutive activation of the ALK kinase and the associated downstream cellular signalling pathways, such as RAS–MAPK, PI3K–AKT and JAK–STAT, leading to dysregulated cellular proliferation and survival^53,54. Indeed, cancers harbouring ALK rearrangements are dependent on ALK signalling for their survival⁵⁵.

First-generation and second-generation ALK TKIs

In the clinic, ALK rearrangements are associated with adenocarcinomas, a younger age than average at diagnosis and a minimal smoking history⁹. Crizotinib, the first targeted therapy evaluated in patients with ALK-rearranged NSCLC, is a broad-spectrum TKI initially developed as a MET inhibitor but discovered to also be a potent ALK and ROS1 inhibitor in kinase screens⁵⁶. This agent had substantial efficacy in patients with advanced-stage ALK-rearranged NSCLC in phase I/II trials, with ORRs of ~60%^57,58. This led to FDA approval for this indication in 2011, followed by full approval in 2013, illustrating that biomarker-directed treatment approaches can be extended beyond EGFR-mutant NSCLC and supporting the major paradigm shift toward personalized management of advanced-stage NSCLC⁵⁷. Subsequent randomized phase III trials demonstrated crizotinib to be superior to chemotherapy for patients with ALK-rearranged NSCLC^59,60.

As with early generation EGFR TKIs, acquired resistance to crizotinib was common, typically emerging within a year of starting treatment (median PFS 7.7–10.9 months)^59,60. More-potent second-generation ALK TKIs were developed, including ceritinib, alectinib and brigatinib, with the initial goal of overcoming resistance to crizotinib. These second-generation ALK TKIs were active in patients with crizotinib-pretreated disease and had improved CNS efficacy owing to improved CNS penetration ^61-66. Mirroring the experience with osimertinib in EGFR-mutant NSCLC, the second-generation ALK TKIs then supplanted crizotinib as the standard first-line therapy for patients with advanced-stage ALK-rearranged NSCLC, with superior efficacy demonstrated in phase III trials comparing alectinib, brigatinib or ensartinib with crizotinib^67-69 (12-month PFS typically ~65% versus ~45%, median PFS by blinded independent review ranging 24-26 months) (FIG 1).

Lorlatinib

Lorlatinib is a third-generation, macrocyclic, highly potent and selective ALK/ROS1 TKI developed to address acquired resistance to earlier-generation ALK TKIs, with potency against a broad spectrum of ALK kinase domain resistance mutations^70-73. Furthermore, given the known high incidence of CNS metastases in patients with ALK-rearranged NSCLC (lifetime prevalence of ~60% - 73%^74,75, lorlatinib was intentionally designed to accumulate within the CNS by minimizing efflux in Pgp-overexpressing cell lines^70-72 and was confirmed to achieve marked CNS penetration in patients with a mean ratio of cerebrospinal fluid (CSF):plasma (unbound) of 0.75 among four patients in the phase I study⁷². In a global phase II study, lorlatinib demonstrated efficacy in patients who had received at least one prior ALK TKI, with an ORR of 47% and an intracranial response rate of 63%⁷³. Specifically among patients who had received at least one prior second-generation ALK TKI, the ORR and median PFS were 40% and 6.9 months, respectively⁷⁶. Of note, lorlatinib has shown higher efficacy among patients with tumours harbouring mutations associated with resistance to second-generation ALK TKIs at baseline versus those without⁷⁶, suggesting that the former remain dependent on ALK signalling. On the basis of these data, lorlatinib received accelerated FDA approval in 2018 for patients with advanced-stage ALK-rearranged NSCLC with disease progression on first-line alectinib or ceritinib, or on crizotinib followed by at least one other ALK TKI.

More recently, lorlatinib was compared with crizotinib as a first-line therapy in the phase III CROWN trial. Lorlatinib was clearly superior to crizotinib with a 72% reduction in the risk of disease progression or death and a significantly longer time to CNS progression (96% versus 60% of patients were alive without CNS progression at 12 months; HR 0.07, 95% CI, 0.03 – 0.17)⁷⁷, resulting in the regular approval of lorlatinib in March 2021 for patients with advanced-stage ALK-rearranged NSCLC, including those with previously untreated disease (FIG. 1).

Acquired resistance

Despite the efficacy of the highly potent, third-generation EGFR and ALK TKIs, acquired resistance to these agents is inevitable and remains a critical unsolved challenge. Further investigations with the aim of elucidating the underlying mechanisms of resistance to these agents are therefore essential. Broadly speaking, acquired resistance can be categorized as either ‘on-target’ (or target-dependent; involving alterations to the target kinase, such that persistent RTK activation and signalling occurs even when the TKI is present) or ‘off-target’ (or target-independent; involving the activation of one or more bypass signalling pathways following upregulation of signalling proteins downstream of the target, or phenotypic transformation). In the following sections, we focus on mechanisms of acquired resistance to osimertinib and lorlatinib (FIG. 2).

Figure 2. Mechanisms of acquired resistance to osimertinib and lorlatinib in EGFR-mutant and ALK-rearranged NSCLCs, respectively.

Mechanisms of resistance to these agents can be divided into several main categories, including: a ∣ alterations that prevent inhibition of the target receptor tyrosine kinase by the tyrosine kinase inhibitor (TKI); b ∣ Activation of bypass and/or downstream signalling pathways that promote cell survival and proliferation despite adequate TKI binding; and c ∣ Changes in tumour cell lineage, such as transformation from an adenocarcinoma to a squamous cell carcinoma or small-cell lung cancer phenotype and epithelial–mesenchymal transition (EMT). Specific selected examples of these mechanisms are also provided.

On-target resistance

An array of kinase-domain mutations can occur in EGFR and ALK, causing steric hindrance with the binding of a TKI or inducing a conformational change, resulting in drug resistance. Whether these alterations reflect pre-existing resistance mutations present in a few subclones at diagnosis, or the occurrence of one or more de novo alterations during treatment, is beyond the scope of this Review.

On-target resistance to osimertinib.

Osimertinib binds EGFR by forming a covalent bond with the cysteine 797 residue located in the ATP-binding cleft⁴². The C797 residue is therefore a vulnerable site for the development of resistance to osimertinib, and indeed, EGFR^C797S mutations have been identified in 10–26% of patients with disease progression on second-line osimertinib^78-83. Other rarer EGFR mutations causing resistance to osimertinib have also been identified at lower incidences than that of C797S, including but not limited to: G796X, which interferes with osimertinib binding^84-87; the solvent-front mutation L792X ^84,85,88,89, which affects the ‘hinge’ region of the kinase domain, and L718X^84,89,90 located in the p-loop, both of which causes steric hindrance; and G724S in the p-loop, which induces a conformational change in the glycine-rich loop that is incompatible with drug binding^91-93. Nonetheless, the collective incidence of EGFR mutations that confer resistance to osimertinib⁸⁴ is substantially lower than that of the T790M mutation identified in 50–60% of patients with resistance to earlier-generation EGFR TKIs^35,36. EGFR amplifications have also been observed after treatment with third-generation EGFR TKIs^94,95; however, because EGFR amplifications can also co-occur with an activating EGFR mutation in treatment-naïve tumours, detailed comparisons of EGFR copy number between pretreatment and post-treatment samples will be required to define the role of this alteration in resistance to osimertinib⁹⁶.

In the first-line setting, on-target resistance to osimertinib seems to be even less common than in later-line settings^97,98, with EGFR^C797X identified in only 7% of patients in the preliminary analysis of resistance mechanisms from the FLAURA trial⁹⁸. As expected, EGFR^T790M was not detected in biopsy samples from patients with osimertinib-resistant tumours. These findings highlight the role of prior treatment exposures in determining the types of acquired TKI resistance that emerge.

On-target resistance to lorlatinib.

ALK kinase-domain mutations confer resistance to second-generation ALK TKIs in approximately 50–60% of patients, and each second-generation TKI has been associated with a unique distribution of ALK resistance mutations⁹⁹. For example, the most common ALK resistance mutations following alectinib treatment are G1202R (in 25–30% of patients) and I1171X (in 10–15%)⁹⁹. At disease progression on later-line lorlatinib, the overall frequency of on-target resistance mutations is lower compared to that with second-generation ALK TKIs, at only 25–30%^100,101, which likely reflects the high potency of lorlatinib against wild-type ALK as well as a broad spectrum of single ALK resistance mutations including against G1202R and I1171X⁹⁹. Indeed, the majority of on-target ALK mutations conferring resistance to lorlatinib are compound ALK mutations (such as C1156Y/L1198F, G1202R/L1196M, I1171N/D1203N, and others)^100-107. These compound ALK mutations are diverse in range and involve two or more mutations in the kinase domain, which cumulatively reduce the potency of lorlatinib (requiring concentrations of lorlatinib that are not clinically achievable) or steric interference with drug binding^101,102. Whole-exome sequencing and analysis of the clonality of serial biopsy samples obtained from patients with acquired resistance indicate that ALK resistance mutations can accumulate in a stepwise manner following sequential treatment with ALK inhibitors, culminating in highly refractory compound mutations that are challenging to target therapeutically and, in many patients, recalcitrant to the majority of currently available ALK TKIs^100,101.

Lorlatinib was only approved by the FDA for first-line use in 2021 and only 28% of patients in the CROWN study had disease progression at the interim analysis⁷⁷; therefore, insights into mechanisms of resistance to lorlatinib in this setting are currently lacking. However, data from preclinical studies suggest that the incidence of on-target resistance to first-line lorlatinib will be substantially lower than the 25–30% observed with later-line use. In a study using N-ethyl-N-nitrosourea (ENU) mutagenesis to generate Ba/F3 cells expressing non-mutant EML4–ALK (modelling treatment-naive ALK-rearranged tumour cells) no single ALK mutation emerged that conferred high level resistance to lorlatinib following prolonged exposure. However, in Ba/F3 cells expressing EML4–ALK harbouring a single ALK kinase domain mutation (modelling ALK TKI-pretreated tumours) exposure to lorlatinib resulted in the emergence of a range of compound ALK mutations conferring resistance¹⁰⁰. This finding supports the suggestion that the range of single ALK mutations that can emerge de novo and confer resistance to first-line lorlatinib will be narrow. In another study, a novel single ALK mutation (L1256F) was found to confer resistance to lorlatinib in a cell line model, with predictions suggesting that this mutation causes steric clash with the fluorobenzene group of lorlatinib¹⁰⁸; however, this mutation has not yet been reported in patients. The findings by Yoda et al.¹⁰⁰ also suggest that first-line lorlatinib might suppress or delay the development of on-target resistance, because compound ALK mutations are less likely to develop in tumours harbouring wild-type ALK compared with variant forms harbouring single resistance mutations.

Of note, the underlying ALK fusion variants might have implications for the development of on-target resistance to ALK inhibitors¹⁰⁹. In one study, ALK resistance mutations and, in particular, ALK^G1202R (the most frequent ALK mutation arising after treatment with second-generation ALK TKIs), were more common in tumours harbouring EML4–ALK variant 3 (exon 6a/b of EML4 fused with exon 20 of ALK) compared to variant 1 (exon 13 of EML4 fused with exon 20 of ALK)¹¹⁰. Whether ALK fusion variants will influence the distribution of mechanisms of acquired resistance to lorlatinib remains undetermined.

Off-target resistance

The prevalence of off-target resistance is anticipated to increase as third-generation inhibitors become more widely used in the first-line^96,100. Historically, mechanisms of off-target resistance have been more challenging to study as these can involve nongenomic or non-cell-autonomous mechanisms that might not be discernable using tumour sequencing. Indeed, although substantial progress has been made in understanding off-target resistance to EGFR/ALK TKIs, the mechanisms of resistance remain unknown in a substantial proportion of patients (up to 40–50% after first-line osimertinib)^97,98, underscoring the need for further investigations in this area.

Parallel bypass pathways

A key feature of resistance owing to activation of bypass signalling pathways is the requirement for dual inhibition of both the driver oncogene and the bypass pathway in order to successfully overcome resistance. MET amplification was the first mechanism of resistance involving bypass signalling to be identified in EGFR-mutant NSCLC¹¹¹, observed in 7–15% of patients following first-line osimertinib^96,98 and 9.8–30% following later-line osimertinib^83,112,113. Importantly, MET amplification is clinically actionable using a combination approach that incorporates a MET-targeted inhibitor^114-120. For example, the TATTON trial evaluated the combination of osimertinib plus the MET TKI savolitinib, which was tolerable and had encouraging efficacy with an ORR of 30% among patients with EGFR-mutant NSCLC with disease progression and concomitant MET amplification on a third-generation EGFR TKI. MET amplifications have also been identified as an actionable mechanism of resistance across other molecular subsets of NSCLC, including ALK-rearranged NSCLC. In an analysis of 207 tumour biopsy samples, MET amplification was detected in 15% of samples from patients relapsing on next-generation ALK inhibitors (including in 22% with disease relapse on lorlatinib)¹²¹. Experiences with combined ALK and MET inhibition have thus far been limited, although this approach is reported to be feasible¹²¹.

Additional mechanisms of bypass-mediated resistance have been described, including acquisition of gene rearrangements that are established drivers of NSCLC (TABLES 1 and 2). A representative example is provided by the identification of RET rearrangements at disease progression on osimertinib^{83,96,112,113,122-124} or brigatinib¹²⁵. Resistance mediated by acquired RET rearrangements can be overcome using a combination of EGFR and RET inhibitors (such as selpercatinib or pralsetinib)^83,126,127. Several other mechanisms of resistance involving activation of bypass signalling pathways have been reported in patients with EGFR-mutant or ALK-rearranged NSCLCs (TABLES 1 and 2)^{35,36,83,96,98,111-113,121-125,128-165}.

Table 1.

Bypass resistance pathways in EGFR-mutant NSCLC

Bypass mechanism	Prior EGFR TKIa	Prevalence (%)	Refs
MET amplifications	Osimertinib	7–15% in first line	98,96
		9.8–30% in second or later lines	83,112,113,124,128
	First-generation TKIs	5–22%, line of therapy mixed/unspecified	36, 35, 36, 111
RET rearrangements	Osimertinib	3.7% in first line	96
		1–2.4% in second or later lines	83,112,113,122-124
		Not specified, case reports	164 b
	Erlotinib	Unknown, data limited to case reports	129 c
NTRK rearrangements	Osimertinib	1–2.4% in second line	83, 112
	Earlier-generation TKIs	Unknown, data limited to case reports	133,164 b
ALK rearrangements	Osimertinib	3% in first line	96
	Earlier-generation TKIs	Unknown, data limited to case reports	129 c, 164 b
BRAF rearrangements	Osimertinib	3.7% in first line	96
	Erlotinib or osimertinib	2.4% in later lines/mixed cohorts	113,134, 129 c, 165
BRAF mutations	Osimertinib	4–5% in second or later lines	112,113,135
	Nazartinib	Unknown, data limited to case reports	136
ROS1 rearrangements	Osimertinib	Unknown, data limited to case reports	164 b,137,138
FGFR3 rearrangements	Osimertinib	1–2.4%, primarily in later lines	112,113,124,140,164 b
	Earlier-generation TKIs	Unknown, data limited to case reports	129 c, 139, 159
HER2 amplifications	Osimertinib	2% in first line	98
		5% in second line, data from later lines limited to case reports	112, 131
	First-generation TKIs	12% in first-line	130
HER2 mutations	Osimertinib	Unknown, data limited to case reports	132
KRAS mutations	Osimertinib	2–7% in first line	98,96
		1–8.3%, primarily in second line or later	112,113,124,161-163
KRAS gains	Osimertinib	2.4%, primarily in later lines	124
NRAS mutations	Osimertinib	Preclinical data only	158
NRAS gains	Third-generation TKIs	Preclinical data only	158,160
PIK3CA mutations	Osimertinib	4% in first line	98
		2.4–16.7% in second line or later	112,113,124,163
YES1 amplifications	Earlier-generaton TKIs	2.9%, line of therapy mixed/unspecified	141
AXL overexpression	Osimertinib	Preclinical only	142,143
IGF1R activation	Third-generation TKI	Preclinical only	144,145

This table includes selected studies, focusing primarily on those demonstrated as conferring resistance to third-generation EGFR TKIs, and is not meant to represent the entirety of clinical and preclinical work on bypass mechanisms in EGFR-mutant NSCLC.

^aThe EGFR TKI received immediately before biopsy sampling is reported here.

^bThis study included samples from 93 patients with concomitant EGFR alterations and receptor tyrosine kinase (RTK) fusions, 16 of which demonstrated acquired RTK fusions at resistance, including: 6 with RET rearrangements (1 with co-occurring KRAS amplification, 1 with MET amplification and 1 with EGFR amplification), 5 with ALK rearrangements (2 with co-occurring EGFR amplifications and 1 with KRAS G12D), 4 with NTRK1 rearrangements (2 with co-occurring MET amplifications, 1 with HER2 amplification and ROS1 rearrangement, and 1 with EGFR amplification), 1 with ROS1 rearrangement (with co-occurring HER2 amplification and NTRK1 rearrangement), and 1 with FGFR3 rearrangement.

^cThis study included 12 paired pre/post-EGFR TKI samples with acquired RTK fusions identified following treatment: 2 with RET rearrangements, 4 with ALK rearrangements, 4 with FGFR3 rearrangements and 1 with BRAF rearrangement. NSCLC, non-small-cell lung cancer; TKI, tyrosine kinase inhibitor.

Table 2.

Bypass resistance pathways in ALK-rearranged NSCLC

Bypass mechanism	Prior ALK TKIa	Prevalence	Refs
MET amplifications	Second-generation TKIs	12% in first or later lines	121
	Lorlatinib	22% in later lines	121
MET rearrangements	Alectinib or lorlatinib	3% in later lines	121
MET exon 14 mutations	Alectinib	Unknown, data limited to case reports	146
RET rearrangements	Brigatinib	Unknown, data limited to case reports	125
EGFR activation	Crizotinib	44% in first line	151
EGFR mutations	Crizotinib	9–14% in first line	152, 153
HER2 amplifications	Crizotinib, alectinib	Unknown, data limited to case reports	148, 149
KIT amplifications/activation	Crizotinib	15% in first line	151
IGF1R activation	Crizotinib	80% in first line	154
SHP2 signalling	Ceritinib	Preclinical data only	157
NF2 mutations	Lorlatinib	20% in later lines	107
YES1 amplifications	Crizotinib, ceritinib	11.8% in later lines	141
KRAS mutations	Crizotinib	18% in first line	153
BRAFV600E mutations	Alectinib	Unknown, data limited to case reports	147
MAP2K1 mutations	Ceritinib	Unknown, data limited to case reports	150
DUSP6 loss	Crizotinib	83%	166
PIK3CA mutations	Lorlatinib or ceritinib	Unknown, data limited to case reports	100,150
AXL overexpression	Earlier-generation TKIs	Preclinical data only	155,156

This table includes selected studies and is not meant to represent the entirety of clinical and preclinical work on bypass mechanisms in ALK-rearranged NSCLC.

^aThe ALK TKI received immediately before biopsy sampling is reported here. NSCLC, non-small-cell lung cancer; TKI, tyrosine kinase inhibitor

Downstream signalling pathways

Downstream effector pathways of EGFR and/or ALK can become reactivated and mediate resistance, although the path towards targeting these mechanisms of resistance clinically has been less clear. As a key example, the RAS–MAPK pathway is a pivotal downstream effector of both EGFR and ALK that can be reactivated by multiple mechanisms affecting each node of the pathway (such as acquired BRAF fusions^{96,104,112,113,129,134-136,147}, KRAS mutations,^{96,104,112,113,125}, NRAS mutations¹⁵⁸, MAP2K1 mutations¹⁵⁰, DUSP6 loss¹⁶⁶, or gain of wild-type NRAS or KRAS^158,166). Data from preclinical models demonstrate the ability of upregulated ERK signalling to confer resistance to third-generation EGFR inhibitors and that combined MEK/ERK inhibition can restore sensitivity to EGFR TKIs¹⁶⁷.

Histological transformation

Repeat biopsy samples obtained at the time of disease progression on EGFR or ALK TKIs are crucial not only for enabling genomic analyses but also for identifying histological transformation. Transformation to small-cell lung cancer (SCLC) accounts for 3–14% of patients with resistance to early generation EGFR inhibitors^35,36,168, and is likely even more common after first-line osimertinib^169-171. The exact mechanisms leading to this dramatic shift in cellular appearance and biology remain unclear. The founder EGFR mutation is known to persist in the transformed tumour specimens, and yet, the dependence on EGFR no longer exists, rendering EGFR TKIs ineffective¹⁷². Nonetheless, the presence of baseline RB1 and TP53 inactivation confers a strikingly elevated risk (43-fold) of transformation to SCLC¹⁷³. A retrospective review of data from 67 patients with EGFR-mutant SCLC revealed generally poorer outcomes than those with EGFR-mutant NSCLC, including transient responses to chemotherapy and unfavourable OS (mOS following transformation of 10.9 months)¹⁷⁴. Transformation from adenocarcinoma to squamous-cell carcinoma can also occur and has been observed in 5 of 62 patients progressing on first-line osimertinib in one study; this phenotype is also associated with short post-transformation survival durations⁹⁶. Both small-cell and squamous-cell transformation have also been identified following ALK TKIs^101,175-180. In one study evaluating samples from 168 patients with ALK-rearranged NSCLCs with resistance to next-generation ALK TKIs, small-cell transformation was identified in 1.2% of patients, which is lower than for EGFR-mutant NSCLC¹⁸¹. Whether the underlying tumour cell plasticity and the prevalence of histological transformation differs across the various molecular subsets of NSCLC currently remains undetermined. Nevertheless, repeat tissue biopsy sampling at the time of progression on next-generation TKIs should be strongly considered (if feasible), given the possibility of histological transformation that might warrant the use of histology-appropriate therapies. This recommendation also reflects the inability to detect such transformations using current liquid biopsy platforms¹⁸².

Epithelial-to-mesenchymal transition (EMT) has also been detected both in patients with EGFR-mutant³⁶ and ALK-rearranged cancers⁹⁹ and is thought to facilitate invasiveness¹⁸³. EMT is a conserved developmental process in which epithelial cells assume a migratory and invasive mesenchymal phenotype with the loss of cell–cell junctions and polarity¹⁸³. This process has been implicated in both primary and acquired¹⁸⁴ resistance to anticancer therapies, and EMT transcriptional regulators have been identified as possible therapeutic targets in patients with lung adenocarcinomas¹⁸⁴. Over the past few years, EMT has been implicated in resistance to osimertinib¹⁸⁵ and lorlatinib¹⁰⁷. In the case of EMT-mediated resistance to osimertinib, the ATR–CHK1–aurora B signalling cascade has been found to be activated in preclinical studies, with the combined inhibition of EGFR and aurora B kinases overcoming this resistance¹⁸⁶.

Primary Resistance

Across clinical trials of EGFR and ALK TKIs involving patients with EGFR-mutant and ALK-rearranged lung cancers, respectively, primary disease progression (the lack of an initial response to therapy or disease control) on therapy has been observed in as many as 4–10% of newly diagnosed patients, indicative of primary, rather than acquired, TKI resistance^{46,77,187,188}. The biology of primary resistance remains an area of active investigation. Mechanisms implicated in primary resistance to TKIs include the presence of a pre-existing concomitant non-TKI-sensitizing alterations (such as EGFR exon 20 insertions, which do not confer sensitivity to earlier-generation EGFR TKIs)^189,190, co-occurring mutations in other pathways (such as NF-κB)¹⁹¹, or germline deletions of the gene encoding BCL-2-like protein 11 (BIM)^192-194. Conceptually, any pre-existing on-target or off-target alterations that are refractory to a specific TKI could contribute to primary resistance to that inhibitor. Next-generation sequencing offers the advantage of delineating co-occurring alterations in the target genes at baseline, which can help elucidate tumour biology.

Strategies to manage resistance

The clinical pace and pattern of disease progression varies widely following treatment with targeted therapies, including osimertinib and lorlatinib. The behaviour of the progressive cancer, together with the understanding of the mechanisms of resistance outlined above, can inform decision-making regarding the optimal management strategy. TKI-resistant tumours might be biologically complex and harbour polyclonal resistance (with resistance mediated by multiple simultaneous mechanisms) with further diversification of underlying intratumour heterogeneity (as reviewed elsewhere)^195,196, posing additional challenges in overcoming resistance.

Addressing oligoprogression

In the scenario of indolent and/or asymptomatic disease progression, a patient might nonetheless continue to receive therapy if they are deemed to be deriving clinical benefit¹⁸². For example, in patients with oligoprogression, defined as small metastases located at one or a limited number of sites in a patient with disseminated disease that is otherwise well controlled¹⁹⁷, the incorporation of locally ablative therapy can be a valid treatment strategy^198,199. In particular, data from several small-cohort studies demonstrate that stereotactic body radiotherapy can provide meaningful local control of progressive disease located at several sites, with favourable PFS outcomes, in patients with targetable oncogene-driven NSCLCs^198,200.

Addressing intracranial progression

Oligoprogressive CNS disease that occurs despite receiving CNS-active, next-generation EGFR or ALK TKIs might be addressed using stereotactic radiosurgery or surgical resection. By contrast, multifocal CNS progression typically requires an alternative management strategy, such as the addition of pemetrexed (which has some CNS efficacy in patients with metastatic adenocarcinomas)²⁰¹, whole-brain radiotherapy, a switch to a more CNS-penetrant TKI if available, or — for select TKIs — dose escalation in an attempt to increase the CNS drug concentration. The potential for escalation of osimertinib from 80 mg to 160 mg daily dosing in rescuing CNS progression in patients with EGFR-mutant NSCLC has been evaluated in a small subgroup analysis of a phase II trial. In this study, 54% of patients with isolated intracranial disease progression on standard-dose osimertinib had intracranial responses to the escalated dose of osimertinib, although many of these responses were not durable (median intracranial PFS 3.8–7.0 months)²⁰², consistent with findings from a separate multicentre retrospective study demonstrating only modest benefit after escalation of the osimertinib dose following CNS progression²⁰³. In another phase II study in which patients with EGFR^T790M-positive NSCLC (following disease progression on earlier-generation EGFR inhibitors) received 160 mg osimertinib, the intracranial ORR among patients with brain metastases and prior exposure to T790M-targeting agents was 50%²⁰⁴. Conversely, escalation of lorlatinib beyond 100 mg daily is unlikely to rescue CNS progression in patients with ALK-rearranged NSCLC given the already marked CNS penetration of this agent⁷². By contrast, successful re-induction of CNS disease control has been reported with dose-escalation of second-generation ALK TKIs, such as alectinib and brigatinib^205,206. Ultimately, the robust CNS efficacy of later-generation TKIs has meant that any subsequent CNS progression on these therapies is exceptionally challenging to manage with the limited treatment options available, thus highlighting the need for novel strategies.

Addressing systemic progression

For patients with diffuse, systemic disease progression, changes in the approach to systemic therapy are warranted. Here, options can be broadly categorized as an alternative TKI (aimed at overcoming on-target resistance) versus strategies beyond TKI monotherapy (aimed at overcoming off-target resistance).

Novel later-generation TKIs.

A number of fourth-generation EGFR and ALK TKIs are currently in development with the goal of overcoming on-target resistance to third-generation TKIs. EAI045 is an allosteric EGFR inhibitor that binds at a site displaced by the C-helix in the inactive kinase conformation. EAI045 alone has been shown to inhibit L858R–T790M-mutant forms of EGFR in vitro but not in vivo; however, this agent was successful in inhibiting L858R–T790M and L858R–T790M–C797S-driven lung cancers in mouse models when combined with cetuximab^207,208. This apparent lack of activity as a monotherapy and the toxicities associated with inhibition of wild-type EGFR by cetuximab might ultimately limit clinical applicability of this combination. JBJ-04-125-02 is another allosteric EGFR inhibitor currently in development that has been found to inhibit the L858R–T790M–C797S forms of EGFR both in vitro and in vivo. JBJ-04-125-02 might induce more-potent responses when combined with osimertinib²⁰⁹. Most recently, BLU-945 and BLU-701 have emerged as potent, CNS-penetrant and wild-type EGFR-sparing fourth-generation EGFR TKIs. In preclinical studies, BLU-945 has been shown to inhibit the triple-mutant EGFR subtype (harbouring a sensitizing EGFR L858R mutation or exon 19 deletion, T790M and C797S), while BLU-701 is active against the double-mutant (with sensitizing mutations and C797S)^210,211. BLU-701 might be a particularly important agent given the lack of emergence of T790M observed with first-line use of osimertinib. BLU-945 (NCT04862780) and BLU-701 (NCT05153408) are both currently being tested in early phase clinical trials.

Novel ALK TKIs are being developed in an attempt to overcome some of the known compound ALK mutations associated with resistance to lorlatinib. TPX-0131 is a next-generation ALK TKI with a compact macrocyclic structure designed to fit completely within the ATP-binding pocket and inhibit ALK even in the presence of ALK resistance mutations. TPX-0131 has demonstrated enhanced potency against wild-type ALK and various single as well as compound ALK mutations compared with the currently approved ALK TKIs in xenograft models²¹². This compound is currently in phase I testing (NCT04849273). NVL-655 is a selective, CNS-penetrant next-generation ALK TKI with activity against lorlatinib-resistant G1202R–L1196M, G1202R–L1198F and G1202R–G1269A compound ALK mutations that is anticipated to enter phase I testing in 2022²¹³. Of note, a single novel ALK TKI might not be able to overcome the entire spectrum of compound ALK mutations conferring resistance to lorlatinib. Indeed, an analysis involving ALK-rearranged cell lines and patient-derived xenograft models demonstrates that certain lorlatinib analogues are able to more potently inhibit G1202R-based than I1171N/S/T-based compound ALK mutations, and vice versa¹⁰¹. Similarly, TPX-0131 is a potent inhibitor of ALK G1202R as well as G1202R-based compound ALK mutations but not those containing I1171²¹². These findings highlight the challenges associated with effective target inhibition once tumours have acquired compound mutations following disease progression on sequential TKIs.

Cycling or repurposing of TKIs.

Outside of the development of novel TKIs, a potential role for ‘recycling’ of older-generation TKIs or ‘repurposing’ of broad-spectrum kinase inhibitors in overcoming on-target resistance to third-generation EGFR or ALK TKIs has been investigated in certain scenarios. For example, the combination of brigatinib, an ALK TKI that also has potency against ROS1, FLT3 and EGFR in preclinical studies²¹⁴, with cetuximab has been found to inhibit triple-mutant EGFR and overcome resistance to osimertinib in both preclinical studies and case reports^215-218. Certain lorlatinib-resistance mutations can re-sensitize ALK-rearranged NSCLCs to earlier-generation ALK TKIs. For example, a patient who had disease relapse on crizotinib owing to the emergence of an ALK C1156Y mutation subsequently had disease progression on lorlatinib owing to an ALK C1156Y–L1198F compound mutation¹⁰². This compound mutation unexpectedly restored sensitivity to crizotinib in cell-line models, enabling the patient to be successfully re-treated with crizotinib. Other reports describe additional compound ALK mutations that might confer sensitivity to earlier-generation ALK TKIs^106,151, and the aforementioned lorlatinib-resistant ALK L1256F mutation has been shown to be sensitive to alectinib in preclinical studies¹⁰⁸.

Combination therapies

Combination therapies are an alternative treatment strategy that could potentially address, or delay, the onset of resistance to third-generation inhibitors. We highlight that all combination strategies discussed here remain investigational. One approach involves the administration of both an early generation and third-generation TKI targeting the same RTK in order to overcome or delay the emergence of on-target resistance. In EGFR-mutant NSCLC, the combination of osimertinib with a first-generation EGFR TKI could overcome the resistance associated with the C797S mutation in certain allelic contexts. Niederst and colleagues demonstrated that when EGFR T790M and C797S occur in trans (that is, in different EGFR alleles), a combination of a first-generation EGFR TKI (targeting C797S) and a third-generation TKI (targeting T790M) could restore EGFR inhibition²¹⁹. Subsequent case reports indicate that such dual EGFR TKI therapy offers at least a transient clinical benefit^220,221, with some potential for durable responses²²². The phase II multi-arm ORCHARD trial (NCT03944772) is now evaluating the efficacy of osimertinib plus gefitinib in patients with known EGFR C797X mutations after first-line osimertinib. In parallel, other clinical trials are assessing combinations comprising a third-generation plus an earlier-generation EGFR TKI in patients with treatment-naive EGFR-mutant NSCLC, with the goal of delaying the emergence of on-target resistance (TABLE 3). In ALK-rearranged NSCLC, a proof-of-concept clinical trial at two centres in Australia (ACTRN12619000844145) is evaluating an alternating schedule of crizotinib and lorlatinib with the goal of delaying the development of resistance.

A second combination approach involves the use of an EGFR or ALK TKI together with another targeted agent to address off-target resistance; evidence supporting the rationale for such strategies is currently accumulating. The previously mentioned ORCHARD trial (NCT03944772) includes arms assessing the safety and efficacy of osimertinib combined with various non-EGFR-targeted therapies in biomarker-selected patient subpopulations. Several other trials testing osimertinib-based combinations in patients with EGFR-mutant NSCLC are currently either planned or ongoing (Supplementary Table 1). Similarly in patients with ALK-rearranged NSCLC, combinations of lorlatinib with other targeted therapies such as MET inhibitors (targeting MET-driven resistance), MEK inhibitors (targeting RAS–MAPK pathway reactivation), or SHP2 inhibitors (targeting resistance driven by RTK signalling) are all being explored (TABLE 3).

A third combination approach involves the addition of chemotherapy to an EGFR or ALK TKI. After disease relapse on prior third-generation EGFR or ALK TKIs, the goal of chemotherapy is to target TKI-resistant tumour cells, whereas the goal of continuing to administer a CNS-penetrant TKI is often to confer CNS-specific protection or efficacy. Experiences in patients with EGFR-mutant NSCLC suggest that adding chemotherapy to osimertinib following disease progression on osimertinib monotherapy does not lead to intolerable toxicities and might extend PFS^223,224. The randomized phase III COMPEL study is designed to determine whether osimertinib should be continued alongside platinum–pemetrexed chemotherapy following extracranial disease progression on first-line osimertinib (NCT04765059). In ALK-rearranged NSCLC, prospective data on this question are lacking, although data from one small retrospective study suggest that continuing an ALK TKI together with chemotherapy following progression on a second-generation ALK TKI extends PFS²²⁵. A separate question is whether the combination of an EGFR or ALK TKI with chemotherapy as initial therapy — prior to clinical relapse on a TKI — offers a PFS and/or OS advantage compared to sequential use of TKIs followed by chemotherapy. Theoretically, maximal use of cytoreductive therapy upfront could either minimize or eradicate any TKI-tolerant persister cell subclones, thus delaying the emergence of resistant clones and clinical relapse²²⁶. In patients with EGFR-mutant NSCLC, first-line therapy with gefitinib plus chemotherapy has been demonstrated to significantly prolong PFS (by 8–10 months) and OS (by around 12 months) compared with gefitinib alone in two independent randomized trials^227,228. The ongoing phase III FLAURA2 trial (NCT04035486) is evaluating the efficacy of upfront osimertinib plus chemotherapy versus osimertinib alone in patients with previously untreated EGFR-mutant non-squamous NSCLC.

Lastly, combinations of EGFR or ALK TKIs with anti-angiogenic agents are being explored, most notably in the first-line setting, with mixed results to date. Anti-angiogenic agents such as bevacizumab (an anti-VEGFA monoclonal antibody) can modify the tumour vasculature, which can affect both drug delivery and efficacy and also suppress tumour growth. Data from preclinical studies demonstrate synergistic effects of anti-angiogenic agents with both EGFR and ALK TKIs²²⁹. Earlier phase II and III studies in which patients with treatment-naïve EGFR-mutant NSCLCs received erlotinib plus ramucirumab or bevacizumab demonstrated improved PFS, but not OS, from the addition of the anti-angiogenic agent^230-232. However, a randomized study comparing the efficacy of osimertinib plus bevacizumab versus osimertinib alone as later-line therapy failed to demonstrate any significant improvement in PFS or OS^233,234. These results are supported by those from a small cohort in Japan that also did not demonstrate any significant improvement in PFS with osimertinib plus bevacizumab in this setting²³⁵. A phase III trial evaluating osimertinib plus bevacizumab versus osimertinib alone as first-line therapy for patients with advanced-stage EGFR-mutant NSCLC is currently under way (NCT04181060). Similarly, the potential benefit from adding an anti-angiogenic agent to an ALK TKI in patients with ALK-rearranged NSCLC remains to be determined. In phase I/II trials, the combination of alectinib and bevacizumab has been demonstrated to be feasible and tolerable, although the size of both study cohorts was small (11 and 12 patients, respectively)^236,237.

Consideration of immunotherapy

The incorporation of immunotherapy into the treatment of patients with NSCLC harbouring EGFR mutations or ALK rearrangements has remained controversial, despite the widespread adoption of immune-checkpoint inhibitors (ICIs) for patients with NSCLC who lack a targetable driver alteration. Data from previous studies indicate that the efficacy of anti-PD-1/PD-L1 antibodies in patients with EGFR-mutant or ALK-rearranged NSCLC is limited^238-242, and the pivotal phase III trials of first-line ICI monotherapy or ICI–chemotherapy combinations in patients with advanced-stage NSCLC excluded patients with tumours of these genotypes. Potential contributors to the modest efficacy of ICIs in these NSCLC subsets include the disproportional representation of never or minimal smokers with lower tumour mutational burdens and lower levels of tumour-infiltrating lymphocytes compared to those with EGFR/ALK-wild-type NSCLC²⁴³. Moreover, PD-L1 expression can be upregulated in tumours harbouring EGFR mutations or ALK rearrangements through constitutive oncogenic signalling (also known as innate immune resistance), resulting in the predictive correlation between level of PD-L1 expression and benefit from ICIs being lost, thus negating the reliability of PD-L1 level as a biomarker in this population^238,243. Overall, the available data do not support the general use of ICI monotherapy among patients with EGFR-mutant or ALK-rearranged NSCLCs after progression on targeted therapies. The role of ICIs in combination with chemotherapy and/or anti-angiogenic therapy requires further clarification. In the initial subgroup analysis of data from IMpower150, atezolizumab in combination with carboplatin, paclitaxel and bevacizumab (ABCP) demonstrated a higher ORR (71% versus 42%), duration of response (11.1 versus 4.7 months) and median OS (not estimable versus 17.5 months) than those observed without atezolizumab (BCP) in patients with EGFR-mutant NSCLC with disease progression on at least one previous TKI^244,245. However, the updated final exploratory OS analyses with longer follow-up revealed no significant difference in OS with ABCP versus BCP in all patients with EGFR-mutant NSCLC (HR 0.60, 95% CI 0.31–1.14) and in those who had received a previous TKI (HR 0.74, 95% CI 0.38–1.46)²⁴⁶. The global ORIENT-31 trial evaluated a quadruplet regimen comprising the anti-PD-1 antibody sintilimab plus bevacizumab biosimilar and chemotherapy (pemetrexed plus cisplatin) in patients with EGFR-mutant NSCLC and found that median PFS was significantly prolonged for patients who received all four therapies compared with chemotherapy alone²⁴⁷. Whether an ICI plus chemotherapy without an anti-angiogenic agent has meaningful benefit in this patient population also remains unknown²⁴⁸. The phase III KEYNOTE-789 trial (NCT03515837) is assessing the efficacy of chemotherapy with or without pembrolizumab in patients with EGFR-mutant NSCLC with disease progression on osimertinib and the phase III ATLAS trial is evaluating the efficacy of atezolizumab and bevacizumab in combination with chemotherapy in patients with EGFR-mutant or ALK-rearranged NSCLC (NCT03991403). Together, these trials should help inform the evolving conversation on the most effective therapies following disease progression on third-generation TKIs.

A growing number of studies have demonstrated an increased risk of severe toxicities with concomitant or sequential use of ICIs and specific EGFR or ALK TKIs^249-258. For example, ICI–EGFR TKI combinations have been associated with substantial risks of grade ≥3 pneumonitis and hepatitis^259,260. In the TATTON trial, specifically, combination of osimertinib and durvalumab was deemed not feasible due to an increased rate of interstitial lung disease (38% with combination versus 2.9% with osimertinib or 2.0% with durvalumab)²⁵⁹. Sequential rather than concomitant treatment with an ICI followed by TKI also can confer greater risk of adverse events²⁶⁰. For example, in 41 patients treated with immunotherapy followed by osimertinib, 15% developed severe immune-related adverse events, including grade 3 pneumonitis, grade 3 colitis, and grade 4 hepatitis²⁶⁰.

Hence, concomitant use of ICIs and EGFR or ALK TKIs is currently not advised outside of well-designed clinical trials owing to the considerable risks this might pose to the patient. Furthermore, premature exposure of patients to ICIs following an initial diagnosis should be avoided while genotyping data are pending, in order to minimize the risk of incurring toxicities upon subsequent switch to a TKI²⁴³.

Novel therapies and future directions

Antibody–drug conjugates (ADCs) and bispecific antibodies are novel classes of therapies that are currently under development and might be effective in patients with TKI-resistant cancers. These agents are of particular interest owing to the potential to overcome a diverse range of mechanisms of resistance. Amivantamab is a bispecific antibody targeting EGFR and MET that received accelerated approval for patients with NSCLCs harbouring EGFR exon 20 insertion mutations in May 2021 based on the results of the CHRYSALIS trial ²⁶¹. The combination of amivantamab and lazertinib, a third-generation EGFR TKI, has demonstrated preliminary efficacy among 45 chemotherapy-naive patients with NSCLCs harbouring EGFR exon 19 deletions or L858R mutations that had progressed on osimertinib, with an ORR of 36% and a median PFS of 4.9 months²⁶². Of note, although the ORR was higher (47%) among patients with tumours positive for EGFR-based biomarkers (i.e., EGFR mutations and amplification) or MET-based biomarkers (i.e., MET amplification or MET exon 14 skipping mutation), responses were observed even in their absence (ORR 29%), supporting a broader applicability of this approach²⁶³. As another example, patritumab deruxtecan is a HER3-directed ADC with a topoisomerase I inhibitor payload that has activity in patients with disease progression on at least one prior EGFR TKI (86% of whom received osimertinib)²⁶⁴, with efficacy observed across an array of mechanisms of resistance. An analysis of 48 paired pre-treatment and post-treatment tumour samples indicates that HER3 expression is often upregulated in EGFR-mutant tumours with acquired resistance to EGFR TKIs²⁶⁵ and that EGFR inhibition can potentiate the anticancer activity of patritumab deruxtecan in patient-derived xenograft models²⁶⁶, thus supporting the rationale for exploring the combination of patritumab deruxtecan and an EGFR TKI such as osimertinib. The activity of patritumab deruxtecan in patients with previously treated EGFR-mutant NSCLCs is currently being investigated further, both as a monotherapy (NCT04619004) and in combination with osimertinib (NCT04676477). The potential role of alternative ADCs targeting TROP2²⁶⁷ (NCT04152499) and MET²⁶⁸ (NCT03859752) is also being evaluated in patients with lung cancers, potentially including NSCLCs with EGFR mutations or ALK rearrangements. Of note, ALK fusions (which lack the transmembrane domain of native ALK) are not membrane bound; therefore, ADCs directly targeting ALK are not feasible for patients with ALK-rearranged NSCLCs.

Other novel therapeutic strategies, such as cancer vaccines, might hold some promise for patients with EGFR-driven or ALK-driven NSCLCs. For example, data from several studies demonstrate that ALK is antigenic in patients with ALK-positive anaplastic large-cell lymphomas^269-272. Furthermore a DNA-based ALK vaccine has been shown to induce potent immune responses in mouse models of ALK-driven lymphoma and NSCLC^273,274, providing the basis for ongoing efforts to develop anti-ALK vaccines as a novel treatment approach. Personalized neoantigen vaccines, either as monotherapy (NCT04397926) or together with EGFR TKIs (NCT04487093), are currently being tested in early phase studies, and numerous additional trials of personalized cancer vaccines, targeting patient-specific predicted tumour neoantigens are enrolling unselected patients with advanced-stage NSCLC, regardless of tumour genotype. Whether such vaccines are a viable therapeutic strategy for patients with EGFR-mutant or ALK-rearranged NSCLCs remains to be established.

Conclusions

The discovery of activating EGFR mutations and ALK rearrangements as actionable driver oncogenes launched a paradigm shift in the management of NSCLC. Subsequent advances with the development of increasingly potent and CNS-active EGFR and ALK TKIs, and — in tandem — the growing understanding of the mechanisms of resistance to these agents, have come to serve as a roadmap for the discovery of targeted therapies for other genomic subsets of NSCLC. Despite these advances, advanced-stage EGFR-mutant and ALK-rearranged NSCLCs remain incurable. Resistance remains an unsolved challenge. New therapeutic strategies are clearly still urgently needed to extend the lives of patients living with these cancers. Ongoing collaborative laboratory, transitional and clinical research efforts will enable the rational development of novel therapies based on a deeper and broader mechanistic understanding of the biology of TKI-resistant cancers.

Key points.

• Non-small-cell lung cancers (NSCLCs) harbouring oncogenic EGFR mutations or ALK rearrangements can be effectively treated with EGFR and ALK tyrosine kinase inhibitors (TKIs), respectively.

• The third-generation EGFR TKI osimertinib and the ALK TKI lorlatinib are currently the most advanced and effective clinically approved agents in their respective NSCLC subsets; both agents are highly effective including against CNS metastases.

• Resistance to third-generation EGFR or ALK TKIs can be mediated by acquired EGFR or ALK kinase-domain mutations, respectively; fourth-generation TKIs designed to overcome this on-target resistance are currently in development.

• Off-target resistance to third-generation EGFR and ALK TKIs is more prevalent than on-target resistance and is mediated by various mechanisms, including the activation of bypass signalling or phenotypic transformation.

• Novel approaches designed to overcome resistance beyond fourth-generation TKIs, including combination therapies, antibody–drug conjugates, bispecific antibodies and immune-directed approaches, are at various stages of clinical investigation.