1 Introduction
Lung cancer is the leading cause of cancer-related death worldwide. Improved understanding of driver mutations of non-small cell lung cancer (NSCLC) has led to more biomarker-directed treatment for patients with advanced stages. The expanding number of drugs targeting these driver mutations offers more opportunity to improve patient’s survival benefit.
To date, NSCLCs, especially those with non-squamous histology, are recommended for testing epidermal growth factor receptor (EGFR) mutations, anaplastic lymphoma kinase (ALK) gene rearrangements, ROS proto-oncogene receptor tyrosine kinase 1 (ROS-1) rearrangements, B-raf proto-oncogene (BRAF) mutations, rearranged during transfection (RET) fusions, Met proto-oncogene (MET) amplification and exon 14 skipping alterations, neurotrophic receptor tyrosine kinase (NTRK) gene fusions, and immunohistochemistry (IHC) testing for the programmed death receptor-ligand 1 (PD-L1) expression.
EGFR-activating mutations are the most common driver mutations in NSCLC. Targeted therapies included first-generation epidermal growth factor inhibitor tyrosine kinase inhibitors (
EGFR-TKIs), erlotinib, gefitinib, and icotinib; second-generation pan-human epidermal growth receptor (
HER) family inhibitor, afatinib and dacotinib; and third-generation
EGFR-TKI, osimertinib, that inhibits both
EGFR-sensitive mutations and resistant mutation
EGFR T790M [
1–
3].
ALK inhibitors included first-generation, crizotinib; second generation, ceritinib, alectinib, and brigatinib; and third-generation, lorlatinib, with increasing capacity for
ALK inhibition generation-by-generation [
4,
5]. The resistance caused by secondary
ALK mutations can be overcome by next-generation
ALK-TKIs [
6]. Moreover, alectinib, brigatinib, and lorlatinib are all recommended as first-line treatment choice for
ALK-rearranged NSCLC for their superior survival compared with crizontinb [
7–
9]. Crizotinib, ceritinib, brigatinib, and lorlatinib can also inhibit
ROS-1 [
10,
11].
BRAF inhibitor dabrafenib combined with
MEK inhibitor trametinib is now the recommended treatment for
BRAF V600E-mutated NSCLC attributing to improved outcomes compared with vemurafenib or dabrafenib monotherapy [
12]. Drugs targeting
MET aberrations with different binding modes included type Ia, crizotinib; type Ib, tepotinib, capmatinib, and savolitinib; and type II, cabozantinib. Entrectinib and larotrectinib are the current standard treatment for
NTRK fusion-positive NSCLC [
13]. Kirsten ratsarcoma viral oncogene homolog (
KRAS) mutations occur in ~20%–30% of patients with NSCLC. Recently, several small molecules including sotorasib (AMG510), adagrasib (MRTX-849) have been developed to specifically target
KRAS G12C [
14].
Although drugs targeting oncogenic driver mutations have significantly improved survival but their long-term responses are still uncommon for most patients, the emergence of acquired drug resistance is inevitable. Therefore, exploration and understanding of the resistance mechanism of target therapy are important to enhance the clinical outcomes and ultimately increase the treatment rate of NSCLC. Meanwhile, improved understanding of biological characteristics in various molecular subtypes of each driver mutation helps explore new classified treatment strategies. Until recently, multiomics analysis has ushered in a new era of precision targeted therapy in lung cancer and led to a deeper understanding of the underlying resistant mechanism. All these scientific and clinical progresses ultimately lead to improved survival in NSCLC and achieve more refined individualized treatment.
We searched PubMed for articles describing clinical trials and reviews of medical therapies for NSCLC with driver mutations form January 1, 2013 to April 1, 2022. The guidelines from major professional societies were also reviewed.
2 Classified management of NSCLC with driver mutation
2.1 Stratification by gene subtypes
EGFR mutations are the most common driver mutations in NSCLC, contributing to approximately 50% of patients with NSCLC in Asians and 15% in Caucasians [
15]. Exon 19 deletion (19del) and L858R mutation in exon 21 (L858R) are the major subtypes of
EGFR-sensitive mutations.
EGFR exon 19 deletion was associated with better response to
EGFR-TKI treatment than L858R mutation [
16]. A meta-analysis including studies comparing
EGFR-TKIs with chemotherapy as the first-line treatment revealed that the benefit of
EGFR-TKI versus chemotherapy was 50% higher for patients with exon 19 deletion (hazard ratio (HR) 0.24, 95% confidence interval (CI) 0.20–0.29) than those with exon 21 L858R alternation (HR 0.48, 95% CI 0.39–0.58,
P interaction < 0.001) [
17]. In the pooled analysis of the LUX-Lung-3 and 6 studies, two phase 3 trials comparing afatinib and chemotherapy, the overall survival (OS) improvement was significant in patients with exon 19 deletion but not in those with exon 21 L858R mutation [
18]. In the FLAURA study, the median OS was significantly prolonged in the first-line osimertinib group than the first-generation
EGFR-TKI comparator. However, the magnitude of OS benefit varied among the two
EGFR mutation subtypes, with the HRs close to 1.00 (95% CI 0.71–1.40) in the L858R subgroup and 0.68 (95% CI 0.51–0.90) in the 19-deletion subgroup [
19]. JO22567 and NEJ026 showed that the bevacizumab plus erlotinib significantly improved progression-free survival (PFS) than erlotinib alone in patients with
EGFR-positive NSCLC. Subgroup analyses by
EGFR mutation subtype suggested that median PFS was longer in patients with L858R mutations in the combination therapy group than those in the erlotinib group (HR 0.57, 95% CI 0.33–0.97). No differences were found between treatment groups in patients with
EGFR exon 19 deletions (HR 0.69, 95% CI 0.41–1.16) [
20]. Similar results were verified in CTONG 1509, a similar design study in the Chinese population [
21]. The INCRESE study showed that double-dose icotinib improved the median PFS (12.9 vs. 9.2 months, HR 0.75, 95% CI 0.53–1.05) and overall response rate (ORR) (73% vs. 48%) in patients with NSCLC harboring 21-L858R mutation compared with the routine dose with acceptable tolerability [
22]. All of these findings suggest fundamental differences in biological characteristics and treatment responsiveness between these two subtypes of activating
EGFR mutations. Therefore, for patients with
EGFR exon 19 deletion, second- or third-generation
EGFR-TKI monotherapy is sufficient; however, for patients with
EGFR L858R mutation, more aggressive treatments including
EGFR-TKIs combined with anti-angiogenesis therapy or dose increase might better extend the PFS and OS.
Most prospective clinical trials of
EGFR-TKIs enrolled NSCLC patients with common (19del or L858R) mutations. However, 10%–20% of
EGFR mutant NSCLC harbor uncommon mutations, which are less sensitive to
EGFR-TKIs treatment. Second- and third-generation
EGFR-TKIs have a broader activity profile across uncommon
EGFR mutations [
23]. A combined analysis of LUX-Lung 2/3/6 indicated that afatinib was active in NSCLC with uncommon
EGFR mutations, especially G719X, L861Q, and S768I but less active in
de novo T790M and exon 20 insertions [
24]. A recent clinical trial indicated that patients harboring uncommon
EGFR mutations including exon 20 insertion were also sensitive to osimertinib. Early clinical activity in
EGFR/HER2 exon 20 has been observed with the dual
EGFR/HER2-TKIs, poziotinib, and TAK-788 [
25]. Amivantamab, a monoclonal bispecific anti-EGFR-MET antibody, showed an ORR of 40% and PFS of 8.3months for patients with
EGFR exon 20 insertion mutant NSCLC with prior platinum-based chemotherapy in the phase I clinical trial, which was certified by the United States Food and Drug Administration as breakthrough therapy for
EGFR exon 20 insertion mutant NSCLC [
26,
27]. A recent study also found that NSCLC with uncommon EGFR mutations associated with a significantly lower incidence of acquired T790M mutation and benefited significantly less from subsequent osimertinib treatment than common
EGFR mutations [
28].
The mutation abundance also affects the therapeutic efficacy of
EGFR-TKI. Studies indicated both tissue and plasma abundance of
EGFR driver mutation affected clinical response and PFS of
EGFR-TKIs [
29,
30]. Meanwhile, the mutation abundance of
EGFR T790M could also predict the efficacy of osimertinib [
31].
Besides
EGFR, other gene alternations also need to stratify patients according to gene subtypes, such as
ALK rearrangement. Currently, eight
EML4-ALK variants (V) have been identified with a number (1, 2, 3a/b, 4', 5a/b, 5′, 7, 8), among them
EML4-ALK variants 1 and 3 being the two predominant variants. Different clinical response to
ALK-TKIs between the “short” (V3 and V5) and “long” (V1, V2, V5′, V7, and V8)
EML4-ALK variants has been reported. According to clinical data concerning crizotinib or lorlatinib, patients with V3a/b seem to have had significantly shorter PFS than did patients with V1. However, these results conflict with those of other studies with crizotinib or alectinib that report insignificant differences in PFS among patients with V1, V2, and V3 [
32]. Meanwhile, V3 was associated with worse overall survival in
ALK-positive NSCLC [
33]. Additionally,
ALK G1202R is more prone to develop among
EML4-ALK V3 after use of
ALK TKIs [
34].
Generally,
BRAF alterations can be categorized into three classes. Class I
BRAF mutations including
BRAF V600 mutations function as RAS-independent monomers, which represent most of
BRAF alterations observed in cancer; non-V600
BRAF mutants can be divided into two categories, class II mutations, including K601E, L597Q, and G469A, signal as RAS-independent activated dimers, and class III mutations, including D594 and G466 mutants, which are kinase impaired but increase signaling through enhanced RAS binding and signal as dimers [
35,
36]. Class I
BRAF mutation well response to
BRAF and
MEK dual inhibition. However, the treatment of NSCLC harboring non-V600 mutations is challenging because of functional heterogeneity and lack of clinical data, though few study indicated that certain non-V600
BRAF alternations (such as L597R mutation, class II mutation) also have durable response to
MEK and
BRAF inhibitors [
37]. These distinct responses of different classes of
BRAF mutations have important implications for future classified management of
BRAF mutant NSCLC.
RET rearrangements are an emerging targetable driver mutation in NSCLC. The most common fusion partner was the kinesin family 5B gene (
KIF5B), followed by coiled-coil domain containing 6 (
CCDC6). Due to the low prevalence of
RET fusions, there is only limited data about the clinical characteristics and outcomes of
RET-rearranged NSCLC. Patients with
CCDC6-RET fusion may have a better prognosis [
38] and longer PFS for chemotherapy [
39] compared with
KIF5B-RET and other subtypes. For target therapy, ORR was similar with
KIF5B and
CCDC6, but lower in other
RET fusions in ARROW study [
40].
Alterations in
MET include MET protein overexpression,
MET gene amplification, and
MET exon 14 skipping mutations or activating mutations. MET overexpression is related to poor clinical outcome in NSCLCs but its role as a therapeutic marker remains controversial.
MET amplification has been regarded as a therapeutic biomarker, with clinical data demonstrating that the high
MET gene copy number indicated a high response rate to targeted therapies.
MET exon 14 skipping mutations also indicate poor outcomes in NSCLC and are presently the most promising driver mutation for the use of
MET-TKIs. As we mentioned before, there are three subtypes of
MET-TKIs and response differently to the
MET mutations: type I
MET inhibitors bind to the activated ATP-pocket of MET, and are subdivided into Ia and Ib; type II inhibitors bind the ATP-pocket in the inactive state; type III inhibitors bind to allosteric sites distinct from the ATP binding site. Mutations in the MET activation loop D1228 and Y1230 confer resistance to type Ia
MET-TKIs while sensitive to type II kinase inhibitors. The solvent front mutation G1163R mediates resistance only to type Ia inhibitor crizotinib but not to type Ib or type II
MET inhibitors. Conversely, mutations in residues L1195 and F1200 confer resistance to type II
MET inhibitors. These findings provide rationality for sequential or combination therapy to avoid resistance to
MET TKIs [
41–
43].
2.2 Stratification by concomitant mutation
EGFR is a member of the ErbB family of receptor tyrosine kinases. Its downstream pathways include PI3K-AKT-mTOR, JAK-STAT, RAF-MEK-ERK, and other pathways. Co-mutations in downstream and other ErbB family members can spur the cell growth through signal transduction, eventually resulting in tumor proliferation and even metastasis [
44]. Clinical data also indicated the concomitant genetic alterations associated with poorer response to
EGFR-TKI treatment among the
EGFR mutant NSCLC [
45]. Some co-mutations such as
TGFBR1,
MTOR, and
RNF43 might indicate primary resistance to
EGFR-TKIs [
46]. In the BENEFIT study [
47], a prospective study using circulating tumor DNA (ctDNA) as direct first-line
EGFR-TKI treatment, patients with tumor suppressor co-mutations and tumors with other driver mutations had shorter PFS (9.3 (7.6–10.5) vs. 4.0 (2.4–9.3) months vs. 13.2 (11.3–15.2)) and OS (21.7 (19.3–27.0)) vs. 15.5 ((10.5–33.7) months vs. 32.0 (29.2–41.5)) than those with pure
EGFR-sensitive mutation [
48]. Preclinical data suggest that
EGFR-TKIs combined with chemotherapy may act synergistically to restrict the development of acquired resistance [
49]. Meanwhile, clinical data also indicated that the adverse events are manageable [
50]. For patients with concomitant mutation, the treatment mode like that of the NEJ009 study,
EGFR-TKIs combined with chemotherapy [
51], might improve the prognosis of patient with concomitant mutations. Some clinical trials aiming to assess the efficacy of combined
EGFR-TKIs and chemotherapy in NSCLC with co-mutations are ongoing (ChiCTR2000032354).
Occasionally, tumors have T790M coexistence with
EGFR-activating mutations at the time of diagnosis. Using more sensitive methods such as DNA mass array, the occurrence of T790M in TKI-naïve patients was higher than that in direct sequencing (25.2% vs. 2.8%,
P < 0.001) [
52]. Naturally, third-generation
EGFR-TKIs are the recommended choice of patients with
de novo T790M.
Besides
EGFR concomitant mutations,
EGFR amplification was also regarded as an
EGFR-dependent resistant mechanism to
EGFR-TKIs [
53].
EGFR amplification is more likely to arise in cases with resistance to third generation of
EGFR-TKIs [
54], and accounts for approximately 20% of the resistant mechanism. The EGFR-MET bispecific antibody amivantamab in combination of third-generation
EGFR-TKIs might improve the outcome of patient with
EGFR amplification and
EGFR sensitive mutations. The clinical trial that is called the MARIPOSA study (NCT04487080) to compare the antitumor activity of amivantamab + lazertinib versus osimertinib alone in patients with
EGFR mutant NSCLC first line treatment is ongoing [
55].
The concomitant sensitive mutations rarely occurred. The frequency of
EGFR/
ALK or
EGFR/
ROS-1 co-alternations was about 1% [
56]. Based on the published data and case report,
EGFR/
ALK co-altered NSCLC still showed response to
EGFR-TKIs [
57,
58], though the
EGFR-sensitizing mutant NSCLC with other driver mutations exhibited shorter PFS (4.0 versus 13.2 months) and OS (15.5 versus 32.0 months) compared with those with
EGFR-sensitive mutation alone [
48,
59]. The co-mutation of
EGFR and
MET (2% prevalence) or
ERBB2 (4% prevalence) amplification was also associated with significantly shorter progression-free survival with first-line 1st or 2nd generation
EGFR-TKIs therapy [
60]. Generally, the first-line
EGFR-TKIs might be a reasonable care for the NSCLC with
EGFR and other driver mutations. The frequency of the co-alterations, the sensitivity of the detection methods, and the phosphorylation levels of the driver gene and its downstream proteins should be considered when making treatment choice [
61].
Like
EGFR mutations, concomitant mutations in the non-
EGFR mutation-driven NSCLC also mostly indicated poor duration of response. For
ALK fusion NSCLC, the presence of
TP53 mutations is a poor predictive and prognostic factor. Though lacks of clinical data, pre-clinical study provided insight that the combination of
ALK-TKIs and proteasome potentially leads to
TP53-independent apoptosis [
34]. The co-occurrent
PIK3CA E542K and
EML4-ALK V3 was also reported to be resistant to
ALK-TKIs [
62]. For
RET-rearranged NSCLC, patient with concomitant
TP53 mutation showed shorter overall survival than those without [
63]. Specially, patient with
STK11 mutation,
KRAS G12C mutation and wild type
KEAP1 benefit more from the sotorasib therapy [
64].
2.3 Stratification by dynamic alternation
The dynamic gene mutation changes are also key predictive factors for target therapy. Liquid biopsy identifying molecular alterations in plasma ctDNA makes the dynamic monitoring possible. In the BENEFIT study, 147 (88%) patients exhibited
EGFR mutation clearance in ctDNA at week 8, and the median PFS was longer for these patients than those with
EGFR mutations that persisted at week 8 (11.0 months (95% CI 9.43–12.85) vs. 2.1 months (1.81–3.65); HR 0.14, 95% CI 0.08–0.23;
P < 0.0001) [
47]. Patients with persistent activating
EGFR mutations in plasma ctDNA after osimertinib initiation also indicated worse prognosis, either in the first-line setting [
65] or T790M-positive patients as the second-line treatment [
66]. Patients without detectable
EGFR mutation in the pretreatment ctDNA possessed the best prognosis from the sequential TKI treatment [
67].
Patients with ctDNA nonclearance at the first visit might require combined therapy because of the limited survival benefit of EGFR-TKI monotherapy. A clinical trial that investigates the efficacy and safety of osimertinib plus carboplatin/pemetrexed versus osimertinib monotherapy in metastatic EGFR mutant NSCLC patients with EGFR mutation persistence in ctDNA 3 weeks after the first-line osimertinib is ongoing (NCT04769388).
Therefore, for EGFR, the target therapy should be stratified based on diverse mutation subtypes, concomitant mutations, and dynamic ctDNA changes. With the in-depth understanding of other driver mutations, this stratified treatment mode can be an effective reference to other target therapy.
2.4 Stratification by special metastasis site
Brain or leptomeningeal metastasis is a generally negative prognostic factor and is refractory to treatment due to the blood–brain barrier. Second- or third-generation
EGFR-TKIs [
52] or
ALK-TKIs are better options for those patients. Second- or third-generation
EGFR and
ALK inhibitors had higher intracranial ORR and disease control rate (DCR) in patients refractory to first-generation TKIs; thus, the next-generation
EGFR/ALK-TKIs are currently replacing first-generation TKIs as the first-line treatment, especially for NSCLC with central nerve system (CNS) metastasis [
68]. Stereotactic radiosurgery and whole-brain radiotherapy are also the treatment options for patients with CNS metastasis.
Oligo-metastatic disease represents a special subtype of metastatic NSCLC that required local treatment [
69]. A series of phase II clinical trials confirmed that local consolidative therapy for patients with oligo-metastatic NSCLC did not progress after initial systemic therapy, which more significantly improved PFS than those without local therapy [
70,
71]. Moreover, for oligo-progressive conditions during target therapy, local ablation of these resistant lesions using either surgery or radiation can prolong the long-term response of the existing target therapy, with an average time to next progression of 6–7 months [
72,
73].
2.5 Stratification by multinomic profiles
With the advent of new genomics, transcriptomics, metabolomics, and proteomics technology, integrated genomic studies have fundamentally contributed to the definition of multinomics landscape of NSCLC. Copy number alterations represent one of the key genetic alterations and included two types: large-scale copy number gain or loss and focal events of gene amplification or loss [
74]. The large-scale copy number gain or loss is consistent in most lung adenocarcinoma [
75], whereas some focal events were associated with negative outcomes, such as
KRAS gene amplification and
MET focal gains [
74]. A previous study also indicated the use of DNA-sequencing, RNA-sequencing, and immunological biomarkers for patients matched to targeted therapies and to identify new potential cancer driver genes [
76,
77]. Mutational signature may also help identify patients. The APOBEC signature was associated with early-onset EGFR wild-type lung cancer in women, which is in line with the finding that the APOBEC signature was associated with better immunotherapeutic response [
78].
3 Resistance mechanisms and coping strategies
Although the target therapy improves survival in NSCLC with driver mutations, only a part of patients respond to these agents and the benefit is temporary. Recognition of the resistance complexity to targeted therapeutics is necessary to understand the tumor biological behavior and to design optional therapeutic strategies.
Based on the duration of disease control, resistance to targeted agents can be classified into intrinsic and acquired resistance [
79]. Primary resistance is defined as resistance in NSCLC with sensitive mutations that progressed within 90 days (3 months) and without showing any evidence of objective responses while receiving target therapy [
80]. The intrinsic resistance might be associated with insensitive driver mutations. For
EGFR-TKIs, intrinsic-resistant mechanisms include
EGFR uncommon mutations, co-mutations that activate downstream or bypass signaling pathways, the presence of germline
Bcl-2-interacting mediator of cell death (
BIM, also known as
BCL2L11), or activation of
NF-κB (nuclear factor-κB) that impairs the apoptosis process in
EGFR-TKI treatment. Acquired resistance refers to tumors that initially responded to the target therapy (usually > 6 months) but ultimately progressed based on Response Evaluation Criteria in Solid Tumors [
81]. Acquired resistance increases both from the selection for pre-existing genetic alterations within a heterogeneous tumor and from the acquisition of new mutations under the therapeutic selective pressure [
79].
Based on the disease failure mode, disease progression in the target therapy can be classified into three subgroups: dramatic, gradual, and local progression. Patients with NSCLC with gradual progression have longer PFS and OS compared with those with local and dramatic progression, and the TKI continuation is superior to switching to chemotherapy in a subsequent setting for patients with gradual progression. For patients with local progressive disease, continued use of the target therapy plus local therapy is recommended. Conversely, studies demonstrated a better survival with chemotherapy instead of continued TKIs for patients with dramatic progression [
82].
Depending on the resistant mechanism, the two major pathways of therapeutic failure are additional genomic mutations in the targeted oncogene (target-dependent resistance) and alternations leading to bypass or downstream activation (target-independent resistance) [
83].
3.1 Target-dependent resistance
The target-dependent resistance mutations include second-site mutations, oncogene amplification or loss, limiting the ability of a drug to inhibit the target activity. The second-site mutations include T790M mutation that leads to acquired resistance to first- and second-generation
EGFR-TKIs and C797S mutation that mediates the resistance to third-generation
EGFR-TKIs. A recent study separated
EGFR mutations into four distinct subgroups based on sensitivity and structural changes, indicating that the on-target mutation might cause EGFR structural changes, thus causing different outcomes following treatment with
EGFR inhibitors [
84]. The on-target mutation is the most common resistant mechanism for third-generation
EGFR-TKIs, such as C797S, G796S, and L792F/H mutations that abolish the covalent binding or confer structural steric hindrance of osimertinib to the EGFR protein. “Fourth-generation”
EGFR-TKIs such as EAI045, JBJ-04-125-02, and BLU-945 are being developed to inhibit these triple mutations (i.e., del19 or L858R/T790M/C797S), yet remain at the early clinical stages of development [
85]. The EGFR/MET bispecific antibody amivantamab has demonstrated activity against a myriad of compound
EGFR mutations [
86].
Meanwhile,
ALK L1196M [
87] and
ROS1 L2026M [
88] that sterically hinder TKI binding also lead to resistance to first-generation
ALK-TKIs. Similarly,
RET G810X mutation is responsible for resistance to
RET-selective inhibitors, such as selpercatinib (LOXO-292) and pralsetinib (BLU-667) [
89]. Both
KRAS R68S and Y96C/D/S drug binding resistance mutations show high-level resistance to both sotorasib and adagrasib [
90–
92].
EGFR amplification and loss of
EGFR T790M mutation have been reported to be resistant to third-generation
EGFR-TKIs [
93]. Similarly,
ALK amplification and copy number gain mediate resistance to crizotinib [
94].
KRAS G12C amplification has been reported at resistance to adagrasib (Tab.1) [
91].
Moreover, for NSCLC with
EML4-ALK fusion, the eight
EML4-ALK variants also demonstrate a differential clinical response to
ALK-TKIs, with variants 3 and 5, indicating poor responses and short PFS [
95].
3.2 Target-independent resistance
Target-independent resistance occurs through the parallel activation or downstream signaling pathways using the driver oncoprotein [
96]. Taking
EGFR (also known as
ERBB1) for example, the alternation in other
ERBB family members such as
HER2,
HER3, and
HER4 all results in the activation of three downstream pathways including PI3K-AKT, JAK-STAT, and MAPK pathways [
97].
In
EGFR-mutated tumors, the MAPK pathway reactivation can occur by acquiring a
BRAF mutation (V600E), loss of
NF1 gene, and activation of
NRAS mutations [
98,
99]. In preclinical studies, this resistance to third-generation
EGFR-TKIs was reversed by combination treatment with
MEK and
EGFR inhibitors [
98]. The
PTEN loss and
PIK3CA mutations are the most common resistant mechanisms in the PIK3CA-AKT pathway. Inhibitors targeting the PI3K-AKT-mTOR pathway are now in clinical trials and have shown potent efficacy. Autocrine interleukin-6 (IL-6) signaling by tumor cells increases JAK-STAT3 activity, which can be activated as an adaptive response to the
EGFR-TKI treatment.
SRC is an oncoprotein implicated in tumor survival and differentiation. SRC activation was reported in
EGFR-TKI-resistant NSCLC cell lines, and the addition of SRC inhibitor dasatinib to erlotinib showed early signs of clinical efficacy in patients with
EGFR mutant NSCLC [
100]. MET and its ligand, hepatocyte growth factor (HGF), can activate MAPK and PI3K-AKT-mTOR signaling.
MET amplification and exon 14 skip mutation account for 10%–20% of
EGFR-TKI-resistant NSCLC. A series of clinical trials concerning combined EGFR and MET inhibitors are ongoing. Meanwhile,
MET variations are also responsible for
ALK-TKI resistance in
ALK-rearranged NSCLC.
The receptor tyrosine kinase AXL also activates MAPK and PI3K-AKT signaling. The overexpression of
AXL in NSCLC resulted in resistance to both
EGFR-TKIs and
ALK-TKIs [
101]. The
AXL-TKI is also at the early phase clinical trials.
HER2 gene amplification/overexpression and
HER2 exon 20 insertion mutations can also lead to bypass activation. Poziotinib and pyrotinib have shown potent inhibitory abilities against
HER2 mutant NSCLC; however, patients with
HER2 overexpression have shown limited responses to
HER2 inhibitors [
96,
102].
The “Target-independent resistance” mechanism can be subdivided into tumor suppressor gene, other driver gene, cell cycle related gene and others. Acquired bypass resistance mechanisms of
KRAS G12C inhibitors included
MET and
HER2 amplification; activating mutations in
NRAS,
BRAF,
MAP2K1, and
RET; oncogenic fusions involving
ALK,
RET,
BRAF,
RAF1, and
FGFR3; and loss-of-function mutations in
NF1 and
PTEN [
91,
103–
105]. The majority of resistance to selective
RET inhibition may be driven by
RET-independent resistance, such as acquired
MET or
KRAS amplification (Tab.1) [
106].
3.3 Other resistant mechanisms
3.3.1 Phenotype transformation
Phenotype transformation is a less common mechanism of
EGFR-TKI resistance and other target therapies, encompassing the histological transformation of adenocarcinoma to small cell lung cancer (SCLC) and to squamous cell carcinoma (SCC) [
83]. The transformed derivatives manifest a more aggressive subtype that can escape from anticancer-targeted therapies. Epithelial-to-mesenchymal transition (EMT) is another phenotypic transformation as a resistance change to both
EGFR and
ALK-TKI therapies [
107], indicating a more invasive phenotype.
RB1 and
TP53 deficiencies indicated a higher incidence of SCLC transformation [
108], and the classical chemotherapy for SCLC is also sensitive to the transformed SCLC [
109].
EMT is a cause of both intrinsic and acquired resistance of
KRAS G12C inhibitors by activating the PI3K pathway [
110]. Moreover, histological transformation from adenocarcinoma to squamous cell carcinoma was observed in patients resistant to adagrasib without determining any genomic resistance mechanisms [
90].
3.3.2 Tumor microenvironment (TME)
TME refers to the tumor-surrounded stromal components, including fibroblasts, immune cells, blood vessels, and cytokines, influencing the response to targeted therapy. Coculture of cancer-associated fibroblasts (CAFs) can induce both EMT and resistance to
EGFR-TKIs in NSCLC cells [
111]. Cytokines, such as HGF, activate the ERK signaling and consequently result in
EGFR-TKI resistance [
112]. The HGF secretion by CAFs can also lead to
ALK-TKI resistance by inducing MET-dependent signaling activation [
113]. Furthermore, the VEGF pathway is a key mediator of tumor angiogenesis.
In vitro studies have found that VEGF and EGFR pathways share common oncogenic downstream signaling and elevated VEGF contributing to the emergence of
EGFR-TKI resistance. In clinical trials, the addition of anti-VEGF therapy to
EGFR-TKIs significantly improved the clinical outcomes [
114]. Angiogenic factors such as the Tie2 receptor also regulated EGFR downstream signaling PI3K/AKT and/or MAPK pathways [
115].
3.3.3 Impaired apoptosis
Germline
BIM (also known as
BCL2L11) deletion polymorphism impairs the apoptotic response to
EGFR-TKI therapy, thus leading to drug resistance [
116]. Activation of NF-κB weakens the apoptosis and rescues
EGFR mutant NSCLC from
EGFR-TKI treatment. The inactivation of NF-κB pathway enhances cell death induced by
EGFR-TKIs [
117]. A recent study has found that YES-associated protein (YAP) activation mediates resistance to combined
EGFR/MEK inhibition. The combination of drugs targeting the YAP pathway, such as transcriptional enhancer associate domain (TEAD) inhibitor, promotes apoptosis of resistant cancer cells [
52].
3.3.4 Barriers and drug–drug interactions (DDIs)
Further challenges beyond the former factors include overcoming barriers, such as blood–brain barriers that limit effective drug delivery to CNS metastases. Increasing drug exposure by dose escalation or exploring new-generation TKIs is a possible solution to overcome body barriers. DDIs should also be under consideration. The combination of drugs interacting with cytochrome P450 and P-glycoprotein should be avoided. Proton pump inhibitors also impair the pH-dependent solution, thus reducing drug absorption [
118].
3.3.5 Resistant mechanism based on multiomics analysis
By analyzing the adaptive-resistant tumor cells using transcriptomic and metabolomics profiling, drug resistance was found to be correlated with cells in proliferative-metabolic quiescence and susceptible to glutamine deprivation and TGFβ2 inhibition [
119]. This study and others [
120] support the co-targeting of bioenergetics and mitochondrial priming to suppress the onset of early drug resistant to
EGFR-TKIs.
4 ICIs in NSCLC with driver mutations
4.1 Immune-checkpoint inhibitors show limited effectiveness in NSCLC with driver mutations
Patients with
EGFR mutation poorly respond to PD-1/PD-L1 inhibitors. In the PACIFIC trial of patients with unresectable, stage III NSCLC, the consolidative durvalumab failed to significantly improve the PFS and OS compared with the placebo group in 6.0% patients with
EGFR mutations [
164–
166]. A meta-analysis, including CheckMate 017/057, KEYNOTE-010, POPLAR, and OAK found that nivolumab, pembrolizumab, and atezolizumab, was associated with prolonged OS, compared with docetaxel in the
EGFR wild-type subgroup but not in the
EGFR mutant subgroup [
167]. ATLANTIC is a phase 2, open-label, single-arm trial concerning durvalumab as third-line or later treatment for advanced NSCLC. Although
EGFR mutant NSCLC with PD-L1 expression of ≥ 25% had a higher response rate than the low expression group (12.2% vs. 3.6%), it still responded worse than the
EGFR/ALK wild-type cohorts [
168]. The WJOG8515L study is a randomized phase II trial comparing nivolumab with carboplatin and pemetrexed treatment in
EGFR mutant NSCLC resistant to
EGFR-TKIs due to mechanisms other than T790M. The PFS of nivolumab was shorter than chemotherapy (1.7–5.6 months), although OS was similar between the two arms [
169].
The reasons for
EGFR mutant NSCLC showed little benefits to ICIs are as following.
EGFR activation induces CD73 expression, which can subsequently reduce the expression of interferon γ mRNA signature [
170]. Meanwhile, high CD73 expression in NSCLC and other tumors is associated with low PD-L1 expression and densities of CD8
+ tumor-infiltrating lymphocytes (TIL) [
171]. Simultaneously, tumor mutational burden (TMB) is positively associated with ICI effectiveness and is lower in the
EGFR mutant tumors compared with
EGFR wild-type tumors (median 3.77 vs. 6.12 mutations/Mb,
P < 0.0001) [
172].
Retrospective studies or subgroup analyses revealed that the limited evidence indicates the characteristics of EGFR mutant tumor that can possibly benefit from ICIs:
(1) Tumors with inflammatory TME. In a study concerting nivolumab monotherapy in tumors with
EGFR mutation, nivolumab responders had a significantly higher CD8
+ TIL density and nonsynonymous mutation burden [
173]. The PFS of anti-PD-1 agents in the
EGFR mutant tumor was longer in patients with high PD-L1 expression and high TMB [
172,
174]. Although patients with
EGFR mutant NSCLC tend to have lower TMB and PD-L1 expression [
172], studies indicated that
EGFR-TKI treatment can change the unfavorable TME. In a cohort of 138 patients with
EGFR mutant NSCLC, the proportion of PD-L1 high expression (≥ 50%) and the TMB level both significantly increased after
EGFR-TKI treatment. Moreover, in this cohort, patients with PD-L1 expression increasing from low to high after that
EGFR-TKI treatment achieved a PFS of > 6 months [
174]. In another cohort of 153 patients with
EGFR mutant lung cancer, TMB also increased at resistance (median 3.42 vs. 6.56 mutations/Mb,
P = 0.008) [
172].
(2) Short PFS to EGFR-TKIs. A study including 58 patients with
EGFR mutant NSCLC that received ICI treatment after
EGFR-TKI treatment found that patients with shorter TKI-PFS (PFS of
EGFR-TKIs was shorter than 10 months) had a significantly prolonged PFS and higher ORR of ICI treatment than those with longer TKI-PFS. Patients with shorter TKI-PFS harbored a higher proportion of CD8
+ T cells and a lower ratio of M1 and M2-like macrophages [
175].
(3) EGFR L858R mutation. A retrospective study reported that patients with
EGFR L858R mutation showed similar clinical outcomes compared with the
EGFR wild-type cohort in anti-PD-1/PD-L1 inhibitor treatment. However, the outcomes were poorer in patients with
EGFR exon 19 deletion [
176].
(4) T790M-negative patients. Clinical data indicated that the PFS of nivolumab after
EGFR-TKIs treatment failure was longer for T790M-negative than T790M-positive patients (2.1 vs. 1.3 months,
P = 0.099) [
173].
For the driver mutation other than
EGFR, the current National Comprehensive Cancer Network (NCCN) guidelines also recommended the use of target therapy as the first-line treatment for advanced NSCLC, such as
ALK,
ROS1,
BRAF,
NTRK1/2/3,
MET ex14 skipping, and
RET. After the disease progression, systematic therapy is recommended. By reviewing the IMMUNOTARGET registry and other published data considering ICIs in NSCLC with oncogenic driver alteration, we found that the ORR and PFS for NSCLC harboring the
KRAS,
BRAF, and
c-MET alterations were comparable to non-mutant NSCLC in ICI registration trials, whereas the ORR in
ALK,
ROS-1 and
RET fusion and NSCLC were relatively low. In the IMMUNOTARGET study, PD-L1 expression was relatively high in those
ALK,
ROS-1 and
RET translocation cases [
177,
178]. Limited data showed variable ORRs of
HER2,
ROS1, and
NTRK-altered NSCLCs, leaving an unclear efficacy of ICIs. Generally, patients with actionable tumor alternations show limited benefit form ICI monotherapy, combination of ICIs and chemotherapy and/or antiangiogenic agents is an option. New relevant biomarkers besides PD-L1 expression should be explored for these oncogenic addicted tumors.
Furthermore, the benefits to ICIs in
KRAS mutant NSCLCs may vary depending on co-mutations. Co-occurring
KRAS mutations and
TP53 inactivation are associated with favorable responses to ICI monotherapy. Conversely,
KRAS mutation co-occurred with liver kinase B1 (LKB1)/STK11 or Kelch-like ECH-associated protein 1 (KEAP1)-nuclear factor erythroid 2-related factor 2 (NRF2) pathway function loss will lead to immunosuppressed phenotypes, thus resulting in resistantance to PD-1/PD-L1 inhibitors [
179–
181].
4.2 Potential mechanisms of low ICI efficacy in EGFR mutant NSCLC
EGFR mutations promote an immunosuppressive TME. First,
EGFR mutant cancer downregulates CXC-chemokine ligand 10 (CXCL10) via phosphatidylinositol-3 kinase (PI3K)–AKT pathways, resulting in low infiltration of CD8
+ T cells. Second, EGFR signaling upregulates the CC-chemokine ligand 22 (CCL22) expression through the JNK–JUN activation, thus recruiting the T-regulatory cells (Tregs). Treg cells can secrete IL-10, IL-35, and TGF-β that reduce antitumor immune responses mediated by natural killer (NK) cells and CD8
+ T cells. Third, the activation of EGFR downstream pathways produces pro-tumoral cytokines and EGFR ligands that promote the macrophages 2 (M2) polarization of tumor-associated macrophages (TAMs), which are associated with suppressed cytotoxic T cell function. Fourth, CD39/CD73 are overexpressed in
EGFR-mutated NSCLCs, which triggers the de-phosphorylation of adenosine triphosphate (ATP) to adenosine diphosphate (ADP), and subsequently results in adenosine monophosphate (AMP) and adenosine (ADO). ADO functions as an immunosuppressive mediator in the TME. The CD73-ADO axis promotes the Treg efficacy and myeloid-derived suppressor cell (MDSC). ADO combined with A2A adenosine receptor (A2AR) also inhibits T cell signal transduction by enhanced immunosuppressive activity of Treg cells and MDSCs [
182–
184].
The immunosuppressive TME induced by EGFR mutation enables tumors to escape from antitumor immune responses. Thus, directly targeting EGFR or upregulating the tumor antigen presentation, inactivating immunosuppressive cells, and downregulating immunosuppressive molecules may lead to antitumor immunity in EGFR mutant tumors.
4.3 Strategies for ICI treatment in NSCLC with driver mutations
NSCLC with driver mutations generally showed limited ICI responses. Strategies for immunotherapy (IO) in NSCLC with driver mutations include the following.
4.3.1 IO + target therapy
A series of studies explored the combination of immunotherapy and target therapy [
185–
187]. However, nearly all clinical trials were discontinued due to severe adverse events and failed to increase the efficacy compared with the target monotherapy. Combination of durvalumab and gefitinib failed to increase PFS and had higher hepatic toxicity than either agent alone [
188]. Pembrolizumab plus gefitinib was also reported to have liver toxicity and failed to increase ORR [
189]. Pembrolizumab or nivolumab plus erlotinib seems tolerable though still does not improve the ORR compared with monotherapy [
189,
190]. TATTON and CAURAL trial explored the combination use of osimertinib plus durvalumab. Both studies showed high incidence of interstitial lung disease (ILD) [
185]. Immunotherapy in combination with first-generation seems prone to cause hepatic toxicity while third-generation
EGFR-TKI with higher incidence of ILD. The mechanisms to explain the synergistic toxicity include that
EGFR-TKIs might enhance T cell-mediated killing of tumor cells by promoting the MHC-I presentation [
192]. In that case,
EGFR-TKIs might also have immunomodulatory effects. EGFR signaling also has an impact on the microenvironment, suppressing immune-mediated anti-tumor responses via cytokines such as IL-6, TGF-β1, and progranulin [
193]. The abnormality of these cytokines may also result in adverse effect.
4.3.2 IO + chemotherapy ± angiogenesis therapy
The IMpower150 study included 124 patients with
EGFR mutant NSCLC. In this
EGFR mutation cohort, the median PFS (10.2 vs. 6.9 months) and response duration (DOR, 11.1 vs. 4.7 months) were significantly improved in the atezolizumab plus bevacizumab plus carboplatin plus paclitaxel (ABCP) group versus the standard-of-care bevacizumab plus carboplatin plus paclitaxel in chemotherapy-naive patients. The difference was not significant when ABCP is compared with atezolizumab plus carboplatin plus paclitaxel group, indicating that the angiogenesis therapy is essential in the combination IO therapy in
EGFR-mutant NSCLC [
194]. ORIENT-31 (NCT03802240) compared sintilimab ± IBI305 plus chemotherapy (pemetrexed + cisplatin) for patients with
EGFR mutant locally advanced or metastasis non-squamous NSCLC after the
EGFR-TKI treatment failed. With a median follow-up time of 9.8 months, ORIENT-31 met its primary end-point of PFS with sintilimab + IBI305 + chemotherapy, achieving a median PFS of 6.9 months versus 4.3 months for chemotherapy alone (HR = 0.464, 95% CI 0.337–0.639) [
195]. Two randomized phase 3 trails, Keynote-789 (NCT03515837) and CT25 (NCT03924050), were designed to assess the beneficial role of the addition of ICIs to chemotherapy in the
EGFR-TKI-resistant setting. CheckMate722 (NCT02864251) evaluated the efficacy of nivolumab + chemotherapy and nivolumab + ipilimumab compared with chemotherapy in patients with
EGFR mutation that progressed on the first-line
EGFR-TKI therapy. These results have not been published.
Several early-stage studies showed promising antitumor activity of IO + chemotherapy in patients with NSCLC who progressed on
EGFR-TKIs. A phase II study evaluated the efficacy of pembrolizumab combined with carboplatin and pemetrexed in recurrent
EGFR/ALK + NSCLC, and results showed that IO + chemotherapy led to a response rate of 42% and a median OS of 22.2 months in
EGFR mutant NSCLC; however, the combination had limited activity in seven patients with recurrent
ALK-positive disease [
196]. Similarly, in a phase II study that evaluated the efficacy of tislelizumab plus chemotherapy in patients with
EGFR-mutated advanced non-squamous NSCLC who failed to
EGFR-TKI therapies, the initial analysis showed that ORR and DCR were 59.4% (95%CI 40.6%–76.3%) and 90.6% (95%CI 75.0%–98.0%), respectively [
197].
4.3.3 IO + angiogenesis therapy
In the phase I study that evaluated the efficacy and safety of TQB2450 (a PD-L1 inhibitor) combined with anlotinib in patients with advanced NSCLC after
EGFR-TKI resistance, the results showed that the combination therapy was well tolerated with potential clinical antitumor activity, with a DCR rate of 77.8% and a median PFS of 9.1 months. Combined apatinib and camrelizumab in a phase II cohort study also exhibited encouraging antitumor activity and acceptable toxicity in
EGFR mutant NSCLC after the TKI resistance, demonstrating an ORR of 18.6% and a median PFS of 2.8 months [
198].
4.3.4 Emerging immunotherapies
NSCLC with driver mutations generally respond poorly to PD-1/PD-L1 inhibitors. The immunosuppressive TME mainly composed by Treg, MDSC, and M2-TAM is a limitation for the T cell-induced antitumor activity. To change the immunosuppressive TME, increasing evidences sustain the role of new additional inhibitory immune checkpoint molecules, such as T cell immunoglobulin and mucin-containing molecule 3 (TIM-3), lymphocyte activation gene-3 (LAG-3), and T cell immunoreceptor with immunoglobulin and ITIM domains (TIGIT) to overcome the resistance to traditional ICIs [
199–
201]. The CD39/CD73/adenosine signaling-mediated dysregulation of energy metabolism in TME is suggested as a potential mechanism involved in the immune escape process. Hence, there is a strong rationale of the combination therapy of ICIs with adenosine signaling inhibitors to overcome immunotherapy resistance in EGFR-mutated NSCLC [
202,
203]. Several clinical trials evaluating the efficacy of several A2AR inhibitors and anti-CD73 monoclonal antibodies for EGFR mutant NSCLC are in progress. Other novel immunotherapy approaches such as the T cell transfer therapy, TIL therapy, and chimeric antigen receptor-modified T (CAR-T) cell therapy, also have potency to control tumor growth in the presence of an appropriate immune context in NSCLC with driver mutations (Tab.2) [
204,
205].
5 New target therapies for NSCLC with driver mutations
Multiple novel oncogenic drivers have been identified in NSCLC. SHP2 (Src homology 2-containing phosphotyrosine phosphatase 2) plays an important role in regulating multiple cascade signaling pathways, including the Ras-Raf-MEK-ERK, PI3K-AKT, JAK-STAT, and PD-1/PD-L1 pathways. SHP2 inhibitors might be a promising drug that could synergistically sensitize the antitumor activity of target drugs in the oncogenic-addicted NSCLC [
206]. To date, several small-molecule SHP2 inhibitors have advanced into clinical trials for mono- or combined therapy of cancers [
207]. Fusions involving the neuregulin-1 (
NRG1) gene result in ErbB pathway activation and therefore can be a candidate for targeted treatment [
208]. Pan-HER tyrosine kinase inhibitor afatinib has reported the activity in NRG1 fusion NSCLC [
208,
209]. Another potential strategy is the use of HER3 inhibitors (GSK2849330) that showed the antitumor activity for
CD74-NRG1 gene fusion tumor in a phase I trial [
210].
Receptor tyrosine-protein kinase ERBB3 (HER3) is wildly expressed in NSCLC.
HER3 overexpression is associated with metastatic progression and disease relapse in NSCLC. Increased
HER3 expression has been observed in EGFR-resistant cell lines, indicating that
HER3 overexpression contributed to drug resistance. HER3 antibody–drug conjugate (ADC) has shown clinical activity in
EGFR TKI-resistant cancers, regardless of resistance mechanisms [
211]. However, many new target agonists or inhibitors and ADCs with new antibodies or payloads are in clinical or preclinical studies that will considerably help the oncogene-driven personalized treatment of NSCLC (Tab.2).
6 Conclusions and perspectives
First, the discovery of oncogenic driver alterations and target therapy have brought significant clinical benefits and established an individualized treatment approach. The management of advanced NSCLC has shifted from a histology based on a biomarker-driven process.
Second, the treatment landscape of oncogenic-addicted NSCLC has become complex. On the one hand, more individualized treatment based on fine stratification has become the focus of research nowadays. The choice of optimal treatment strategy should consider gene subtype, concomitant mutation, dynamic gene alternation, and metastasis site. On the other hand, understanding primary and acquired resistance to targeted therapy provides an insight into the molecular evolution of tumor development. The recognition of resistant mechanisms is the basis to design new drugs or combinatorial therapeutic strategies.
Lastly, combination strategies require integration with immunotherapy, and the immunosuppressive microenvironment should be reversed to improve the sensitivity of ICIs by the drug combination. We should also be aware of the possible risk of combined toxicity and thereby explore the optimal timing and combined regimen of immunotherapy. Meanwhile, the interaction between driver mutations and immune-microenvironment is essential to uncover after drug resistance and to establish robust predictive biomarker for the NSCLC with driver mutations, in order to identify specific oncogenic driving patients with NSCLC who can benefit from immunotherapy.