Frontiers of ctDNA, Targeted Therapies, and Immunotherapy in Non-Small-Cell Lung Cancer

Chennianci Zhu; Weihao Zhuang; Limin Chen; Wenyu Yang; Wen-Bin Ou


Transl Lung Cancer Res. 2020;9(1):111-138. 

In This Article

Mechanisms of TKI Resistance


Three generations of TKIs have been designed to target EGFR mutations. As ctDNA has aided the detection and verification of resistance to various drugs, the mechanisms behind resistance to these drugs have been studied for more than a decade. Three generations of EGFR-TKIs are summarized in Table 1.[77,82–85,90] In the following section, we focus on several widely accepted resistance mechanisms along with less discussed novel resistance. Figure 1 shows most of the mechanisms of resistance to different inhibitors through signaling pathways mentioned in this review. Hopefully, the study of these mechanisms will shed light on the clinical treatment of NSCLC patients.

Figure 1.

Resistant mechanisms of TKIs in target therapies of NSCLC. EGFR/ALK mutations, bypass mechanisms such as MET/ERBB2 amplification, NRG1 fusion, IGF1R activation, and EGFR translocation-induced Hippo pathway inhibition result in the activation of their downstream pathways, such as PI3K/AKT/mTOR and RAS/RAF/ERK/MAPK, which directly confer TKI resistance. TKI, tyrosine kinase inhibitor; NSCLC, non-small-cell lung cancer; EGFR, epidermal growth factor receptor; ALK, anaplastic lymphoma kinase; NRG1, neuregulin 1; IGF1R, insulin-like growth factor 1 receptor; KRAS, Kirsten rat sarcoma; YAP, Yes-associated protein; TAZ, transcriptional coactivator with the PDZ-binding motif.

EGFR T790M and C797S. EGFR T790M, the most common mutation against first-generation TKIs, accounts for 50–60% of acquired resistance to TKIs. When the third-generation TKI osimertinib was developed, the response rate to osimertinib was 62.5%, and 52% of patients with the EGFR T790M mutation exhibited a PFS of 12 months.[6] Though osimertinib is a revolutionary drug, there are still cases of osimertinib resistance. Novel osimertinib resistance mechanisms can be divided into different types: (I) the acquisition of tertiary mutations that restore EGFR signaling pathways and (II) bypass mechanisms, such as MET or ERBB2 amplification, Hippo pathway inhibition, or IGF1R signaling activation.

EGFR C797S is sensitive to first-generation TKIs but resistant to second- and third-generation TKIs. The second-generation EGFR-TKI dacomitinib directly induced secondary EGFR T790M or C797S mutations in Ba/F3 cells transfected with EGFR 19 del, L858R, or G719A mutants, and there was no significant difference in the timing of the emergence of the T790M and C797S mutations.[8,78] The C797S mutation occurred in 11% of L858R mutant clones (4 of 35) and 32% of G719A mutant clones (12 of 38) established with low-dose dacomitinib.[8] Afatinib, another second-generation EGFR-TKI, also induced the C797S mutation as a resistance mechanism.[78] First-generation EGFR-TKIs exert their effects independent of the cysteine at position 797.[94] Gefitinib or erlotinib can be used to overcome resistance to the C797S mutation. Erlotinib is particularly effective in treating EGFR C797S-positive tumors.[78–80] Tumors with T790M and the trans C797S mutation responded to a combination of erlotinib and osimertinib. In patients with acquired osimertinib resistance, gefitinib monotherapy was shown to successfully shrink T790M-negative/C797S-positive tumors.[77,95] By studying the T790M and C797S mutations in lung adenocarcinoma, NSCLC containing T790M/cis-C797S mutations was shown to be more aggressive than that containing T790M/trans-C797S mutations.[95] Tumoral heterogeneity plays an important role in the mechanism of dual cis/trans resistance.

Tumor Heterogeneity Reveals Resistance Potential. Overcoming tumor heterogeneity is a major challenge for the personalized treatment of cancer. The most effective treatment protocol can be designed only after tumor DNA profiling, which reveals the heterogeneity of each individual tumor.

Cancer cells simultaneously carry several types of activating EGFR mutations even before EGFR-TKI treatment begins.[68] If these cells possess the ability to develop multiple resistance mechanisms and are more adaptable to TKIs, resistance can emerge earlier.[96] There may be an intrinsic minor population of T790M-positive cancer cells in EGFR-mutated tumor clones carrying multiple mutations.[69] In addition, T790M-negative drug-tolerant cells may persist in patients with drug-resistant EGFR T790M-positive tumors and undergo further evolution to acquire resistance to subsequent therapies.[97] Emergence of the T790M mutation accounts for 50% of acquired resistance. Furthermore, the T790M mutation was more frequent among patients with the activating EGFR 19 del mutation than those with the L858R mutation.[6] By understanding the emergence of EGFR T790M mutation heterogeneity, better clinical decisions can be made during the period of disease resistance.

EGFR T790M Heterogeneity. The presence of resistance mutations, such as the T790M mutation, does not necessarily lead to an overall resistant phenotype in tumor cells. In some cases, despite the selective pressure by TKIs in the treatment process, cancer cells with the T790M mutation still fail to dominate.[68] The heterogeneity within tumors plays a much more important role than previously estimated. For instance, EGFR T790M-positive clones emerge from not only pre-existing clones but also initially EGFR T790M-negative drug-tolerant cells through what is called the de novo acquisition of T790M.[97] This mixed tumor evolution eventually determines the overall response to TKIs although a wide variety of tumor cells originating from different parental cells are contained within the tumor.

Heterogeneous EGFR-mutated cancer cells undergo convergence and divergence in response to EGFR-TKIs. Minor clones are eliminated first, and mutations such as T790M then arise and induce acquired resistance.[68] For instance, the ratio of T790M to exon 19 del mutations changed significantly and fluctuated during treatment. While most EGFR alleles with exon 19 del seem to carry the T790M mutation at the first and second T790M DNA peaks, this does not appear to continue during the development of acquired resistance because exon 19 del alleles may have amplified more rapidly than T790M alleles over time.[68]

Several studies have shown that the T790M mutation can be spatiotemporally heterogeneous in a patient because of selective pressure from EGFR-TKIs.[66,69,97–99] This heterogeneity also reflects the dynamics within resistant tumor cell clones, which are likely made up of TKI-sensitive and TKI-resistant cells. Upon TKI withdrawal, the ratio of TKI-sensitive to TKI-resistant cells will increase as a result of the repopulation of TKI-sensitive cells due to the absence of selective pressure, thus leading to the regain of tumoral TKI sensitivity.[100] In conclusion, long TKI-free intervals may reduce TKI-resistant clones and induce restoration of EGFR-TKI sensitivity. Elimination of the T790M mutation is the result of a significant reduction in TKI resistance, such as that observed in T790M-positive clones. T790M heterogeneity should be taken into consideration when making clinical decisions to apply TKI therapy in patients after the development of acquired resistance. Furthermore, by using T790M-positive clones as a predictive marker during TKI treatment, a better TKI rechallenge scheme with designed "on and off" TKI exposure can maximize elimination of the T790M-positive clones in the tumor.

The heterogeneity of T790M-positive patients is the result of a mixture of T790M-positive and T790M-negative cells that may have been present in heterogeneous tumors before TKI treatment. TKI pressure plays a selective role and controls the ratio between these two types of cells, thus determining the total tumor status.[69] In contrast, the selective pressure exerted by TKIs can never induce T790M-positive cells in a T790M-negative cell clone.

EAI045 is the first allosteric TKI engineered thus far to overcome L858R/T790M and C797S mutations.[93] EAI045 is 1000-fold more selective for mutant EGFR than for wild-type EGFR. Cetuximab, a monoclonal antibody that can block EGFR dimerization by preventing EGF ligand binding, synergizes with EAI045 by converting the inhibitor-resistant receiver population into a monomeric form that is remarkably sensitive to EAI045.[92,93] EGFR mutations at position C797 do not affect the efficacy of EAI045, as C797 is far from the allosteric binding pocket.[91] In addition, EAI045 in combination with cetuximab potently inhibited L858R/T790M/C797S in Ba/F3 cells.[93] More studies are needed to determine the clinical efficacy of EAI045 and shed light on its application.

T790M Loss and C797S. The two main mechanisms of resistance to osimertinib are loss of the T790M mutation and emergence of the C797S mutation. These two mutations cause more than 60% of resistance.[88,101] T790M loss is a common mechanism of resistance in patients treated with osimertinib. Half of the patients in a progressive state exhibited T790M loss during osimertinib treatment.[86,88,102] Twenty-six percent of the patients were observed to harbor EGFR C797 and L792 mutations, and these mutations were exclusive to T790M-preserved cases.[86] The C797S mutation was detected in approximately 40% of EGFR-mutated NSCLC patients with T790M who developed acquired resistance to osimertinib.[88] These results reveal that there are different resistance patterns in cases in which T790M is preserved or lost. In most T790M-preserved cases, resistance was associated with continued EGFR activation through known tertiary mutations that cause resistance, such as C797S, or activation of bypass signaling pathways, whereas resistance in T790M-loss cases occurred through diverse and predominantly EGFR-independent alternative competing mechanisms, such as MET amplification and small cell transformation.[86]

Patients who develop early resistance to osimertinib are likely to have competing resistance mechanisms in other tumor subclones, while patients who develop late resistance to osimertinib are more likely to have maintained T790M and acquired the C797S mutation.[87] Although both T790M-loss and T790M-preserved patients had decreased T790M levels after 1 to 3 weeks of osimertinib treatment, repeated testing for T790M is still required to distinguish between these two biologically distinct types of osimertinib resistance.

Other novel mutations occur at a relatively low frequency. Among these mutations, L792F was detected in 1.76% (6 of 340) of patients with lung adenocarcinoma treated with osimertinib.[89] The L792F mutation results in a level of resistance between that to both first- and third-generation EGFR-TKIs due to the T790M and C797S mutations.[89] Mutations in cis with T790M cannot be inhibited by cetuximab or EAI045.[89] However, L792F-positive tumors were also found to be much less resistant to second-generation TKIs, especially dacomitinib.[78] More precise treatment strategies and additional combinational approaches are required for patients with the EGFR L792F mutation.


Rearrangements of ALK, which are mutually exclusive with mutations in EGFR or KRAS, account for 3–7% of mutations.[13–15] The most common ALK rearrangement occurs in the echinoderm microtubule-associated protein-like 4 (EML4) gene, producing an EML4-ALK fusion.[103] To treat ALK fusion, TKIs of ALK, such as crizotinib, ceritinib, alectinib and brigatinib, have been developed. We examined several clinical trials of ALK-TKIs and EGFR-TKIs and illustrate the clinical efficacies of ALK-TKIs and EGFR-TKIs in Table 2.[111–116] Crizotinib is a competitive ATP inhibitor of ALK and MET tyrosine kinases that received Food and Drug Administration (FDA) approval in 2011.[117,118] A phase I clinical trial of crizotinib demonstrated a high overall response rate (ORR) of 60.8% and a median PFS of 9.7 months. Subsequent phase III trials demonstrated the prolonged PFS and increased ORR of crizotinib compared to those of standard first- or second-line chemotherapy.[110,119]

Resistance to First- and Second-generation ALK-TKIs. Acquired resistance to the first-generation ALK-TKI crizotinib has been identified in approximately 30% to 40% of patients. The occurrence of resistance may be attributed to three causes: (I) the acquisition of secondary resistant mutations that were reported to occur in 22–36% of patients;[120] (II) ALK copy number alterations; and (III) the upregulation of bypass signaling pathways leading to ALK-independent growth, such as the activation of the EGFR, MET, KIT, IGF1R, and/or other pathways.[120–122]

Secondary ALK mutations in NSCLC are distributed throughout the kinase domain. The mutations L1196M, G1269A, F1174, 1151Tins, L1152R, S1206Y, I1171T and G1202R have been determined to confer major resistance to crizotinib.[123–126] L1196M is a mutation of the gatekeeper residue, and G1269A is a mutation of a residue in the ATP-binding pocket that, upon its mutation, can cause changes that prevent crizotinib binding.[127]

To counter resistance to crizotinib, the second-generation ALK inhibitors ceritinib, alectinib and brigatinib were developed. Approximately 40–50% of crizotinib-resistant patients were shown to respond to these TKIs, which exhibit a median PFS of 7–12 months.[128–130] Secondary mutations at I1171 confer resistance to alectinib, and the I1171N and I1171T mutations destabilize ALK inhibitor binding while stabilizing the tyrosine kinase in its activated conformation.[131,132] Ceritinib, however, was found to be capable of overcoming the I1171T mutation as well as several crizotinib-resistant ALK mutations, such as L1196M, G1269A and S1206Y, in preclinical models.[133]

Resistance to Third-generation ALK-TKIs. The third-generation ALK-TKI lorlatinib is an ATP-competitive inhibitor of recombinant ALK and ROS1 kinases. Lorlatinib has the ability to penetrate the brain and central nervous system, where frequent lung cancer metastasis occurs. In preclinical settings, lorlatinib was shown to have low nanomolar potency against wild-type ALK and be highly effective against all clinically acquired ALK mutations, including the highly resistant G1202R mutant, which is resistant to both first- and second-generation ALK inhibitors, by impairing drug binding through steric hindrance.[126,134] Lorlatinib can also potently inhibit wild-type ROS1 and the G2032R ROS1 mutant in vitro and in vivo.[135]

The antitumor efficacy of lorlatinib has been shown in two recent clinical trials.[116,136] In a phase I trial of lorlatinib in NSCLC patients with ALK or ROS1 rearrangement, lorlatinib showed a high response rate and a median duration of response of 11.7 months in 42% of patients (11 of 26) previously treated with first- and second-generation ALK-TKIs. Then, patients with different medical histories were involved in a phase II trial. Among these patients, the treatment-naïve group of ALK-positive NSCLC patients showed the best response rate; 87% (26 of 30) of the patients in this group showed a PR, and only 1 of 30 patients remained at disease progression. Consistent with those in the phase I trial, patients previously treated with crizotinib showed a good response rate that was second to the response of the treatment-naïve group. In all patients previously treated with at least one ALK-TKI, responses were rapid with a median of 1.4 months and durable. Lorlatinib is a new option for patients whose disease has progressed after treatment with crizotinib or second-generation ALK inhibitors.

Bypass Mechanisms

Bypass Mechanisms for EGFR-TKI. Amplification of MET and ERBB2: MET gene amplification and the hyperactivation of MET are mechanisms of resistance to both first- and third-generation EGFR-TKIs, such as erlotinib, rociletinib and osimertinib, in patients that exhibit multiple mechanisms of resistance, such as those in whom tumor cells have undergone epithelial-mesenchymal transition (EMT) or those with small cell lung cancer (SCLC) transformation.[10,38] An inverse correlation between EGFR T790M and MET amplification was observed.[137] EGFR and MET have been shown to act simultaneously to activate downstream effectors, such as PI3K/AKT/mTOR and RAS/RAF/ERK, and ultimately regulate tumor cell proliferation. MET signaling activation likely serves as a compensatory pathway for the loss of the EGFR-driven signaling cascade.[138]

ERBB2 overexpression accounts for approximately 3–26% of acquired resistance to first-generation EGFR-TKIs and 5% of acquired resistance to third-generation TKIs in NSCLC patients.[38,74–76] By activating the EGFR-independent phosphorylation of ERBB3 and downstream activation of the PI3K/AKT pathway, a bypass mechanism is created even in the presence of first-generation EGFR inhibitors.[9,10] The sensitivity of the third-generation EGFR-TKIs rociletnib and osimertinib decreased when they were used at nanomolar concentrations in a PC9/GR NSCLC cell line overexpressing ERBB2.[139] Shi et al.[10] demonstrated that ERBB3 phosphorylation in both HCC827/ER and HCC827/AR cells was minimally inhibited by osimertinib alone and could be fully suppressed only when osimertinib was combined with a MET inhibitor. Hence, the sensitivity to third-generation TKIs is restored by MET inhibition resulting from the suppression of ERBB3 phosphorylation.

Tumor resistance caused by the activation of accessory pathways can be theoretically overcome by a combination of EGFR inhibitors and other involved molecules, which serves as a potential strategy to counter acquired resistance often observed during the treatment of EGFR-mutated NSCLC. Dual inhibition of MET and ERBB has also been performed to determine the intratumor heterogeneity and plasticity in acquired resistance.[140] The combination of capmatinib (MET inhibitor) and afatinib was shown to be more effective than afatinib as a single agent. This drug combination completely suppressed tumor growth in a patient-derived xenograft (PDX) mouse model that showed the necessity of MET amplification to lung cancer cell survival.[140] KRAS G12C mutant clones emerged upon the blocking of two upstream activating components of the MAPK pathway, EGFR and MET, which suggests that the development of resistance in NSCLC cells is a flexible process.

IGF1R Activation: IGF1R stimulates cell proliferation primarily through the PI3K/AKT and RAS/MAPK signaling pathways. Activation of IGF1R enhances PI3K/AKT signaling, giving rise to resistance to EGFR-TKIs in PC9 cells resistant to either PF299804 or WZ4002.[11] IGF1R upregulation as an immediate response to erlotinib was also observed in erlotinib-resistant HCC827 cells with additional acquired features of EMT, whereas MET overexpression and secondary EGFR mutations were absent.[71,81] Exogenous IGF1 also activated IGF1R in the PC9 and H460 cell lines.[72] Targeting IGF1R with AG-1024 and inhibiting EGFR with gefitinib exerted antiproliferative effects in the H1975 cell line via a reduction in AKT phosphorylation and the subsequent upregulation of BCL-2-interacting mediator of cell death (BIM).[141,142]

The loss of IGF binding protein 3 (IGFBP3) has been reported to induce IGF1R activation and EGFR-TKI resistance. Overexpression of IGFBP3 or inhibition of IGF1R increased the sensitivity of NSCLC cell lines to 3rd-generation EGFR-TKIs.[11,73,143] A recent study found that both increased or decreased IGFBP expression induced the activation of IGF1R in response to TKI and served as a bypass mechanism in cells with MET amplification. In addition, the activation of IGF/IGF1R signaling was found in cell lines resistant to both the MET-TKI PHA665752 and the EGFR-TKI gefitinib.[144] IGF1R knockout enhanced MET amplification, resulting in resistance to erlotinib. In addition, IGF1R knockdown attenuated EMT, which involved a decrease in E-cadherin expression and an increase in vimentin, snail, and nuclear β-catenin expression in PC9/GR and H460/ER cells.[72] However, tumoral clones with MET amplification do not always exhibit great advantages. IGF1R hyperactivation and heterogeneous EMT features, but not MET amplification, led to resistance to high-dose erlotinib in the HCC827 cell line. MET amplification tends to emerge instead of EMT as the resistance mechanism in HCC827 cells after their exposure to low concentrations of EGFR inhibitors.[145]

TGFβ1 Pathways: EGFR inhibition can result in an autocrine transforming growth factor β1 (TGFβ1) pathway loop and stimulate the downstream SMAD pathway.[145] Hematopoietic pre-B-cell leukemia transcription factor (PBX)-interacting protein (HPIP/PBXIP1) silencing can suppress TGFβ1 secretion by inhibiting SMAD2 activation.[146] Only continuous TGFβ1 secretion can promote and maintain mesenchymal transition and EGFR-TKI resistance; thus, this induced mesenchymal transition was reversible upon the removal of TGFβ1.[145] The reversal of TGFβ1-induced EMT by E-cadherin overexpression in resistant cells can also restore TKI sensitivity.[72] The expression of EMT-related markers and TGFβ1/SMAD2 was higher in cells transfected with miRNA-132 inhibitor.[147] Furthermore, miRNA-138 knockdown cells exhibited mesenchymal phenotypes.[148] These results imply that miRNA-132 inhibits EMT by regulating TGFβ1/SMAD2 in NSCLC cells, while TGFβ1 downregulates miRNA-138, contributing to an EMT phenotype. This reversibility is clinically significant because the relief of EGFR inhibition could deplete TGFβ1 to reverse EMT, which in turn might resensitize tumors to EGFR-TKIs, thus prolonging the duration of EGFR-TKI therapy.

Hippo Pathway Inhibition: The Hippo pathway consists of a large network of proteins that include neurofibromin-2 (NF2), core kinase cassette containing mammalian STE20-like protein kinase 1/2 (MST1/2), large tumor suppressor 1/2 (LATS1/2), adaptor proteins Salvador homologue 1 (SAV1) and MOB kinase activator 1 (MOB1). These proteins limit tissue growth by promoting LATS1/2-dependent phosphorylation of the oncoproteins Yes-associated protein (YAP) and transcriptional coactivator with the PDZ-binding motif (TAZ).[149–151] YAP and TAZ promote cell proliferation by regulating the activity of different transcription factors such as TEADs and SMADs.[152] The association between Hippo pathway inhibition and the development of EGFR-TKI resistance was discussed in several recent studies.[12,153–157]

Translocation of EGFR from the plasma membrane to the cytoplasm and nuclear membrane inhibits the Hippo pathway. EGF could stimulate the translocation of membranous EGFR (mEGFR) into the cytoplasm (cEGFR) and nucleus (nEGFR) by binding to importin-β through its nuclear localization sequence or binding with YAP.[12,158,159] As a result, the expression of mEGFR decreased, and the expression of cEGFR increased. cEGFR interacted with SIK2 and enhanced its ability to bind to SAV1, which inhibited the interaction between LATS1 and MST1. Furthermore, downstream YAP phosphorylation was inhibited, thus increasing the nuclear translocation of YAP and ultimately inhibiting the Hippo pathway by binding with the transcription factor TEAD.[12] Therefore, resistance against the first-generation EGFR-TKIs gefitinib and erlotinib is associated with inhibition of the Hippo pathway and enhanced YAP activity.[155,156] Furthermore, the combination of the YAP inhibitor verteporfin with erlotinib sensitized the erlotinib-resistant H1975 cell line.[155]

Bypass Mechanisms for ALK-TKI. EGFR Activation: Multiple bypass mechanisms could induce EGFR-TKI resistance, and the emergence of bypass signaling during ALK-TKI treatment also contribute to resistance in ALK-positive NSCLC. EGFR activation as a bypass mechanism for ALK-TKI crizotinib, alectinib, and ceritinib was found in several cell models and one mouse model in recent studies.[160–162] The activation of EGFR pathway induced by TGFα contributed to the resistance to alectinib, and upon TGFα knockdown, the sensitivity of alectinib in H3122-alecinib resistant NSCLC cells was restored.[162] Epidermal growth factor (EGF) was also found to induce resistance to alectinib by activating EGFR signaling.[161] Furthermore, through dual targeting of ALK and EGFR with alectinib and afatinib in mouse xenograft model, EGFR downstream signals to PI3K/AKT and MAPK were inhibited, and tumor volume decreased significantly.[162] Similar study also discussed acquired resistance to ceritinib through EGFR bypass signaling activation in H3122 cells.[160] Besides EGFR signaling, increased expression levels of other members of ERBB family such as ERBB2/3 induced by EGF were reported to be mechanisms contributing to resistance to ALK-TKI in EML4-ALK positive H3122 cells.[163] Dual inhibition of ALK/ERBB family by shRNA and dacomitinib showed further antiproliferative effects in DFCI076 NSCLC cells that are resistant to both ALK inhibitor crizotinib and TAE684.[164] Similarly, crizotinib plus afatinib that inhibited ALK, EGFR, and ERBB2 signaling was able to inhibit the growth of H3122-crizotinib resistant cells,[165] confirming EGFR activation as a resistance mechanism for ALK-TKI. However, if the cell line is resistant to both of the inhibitor, then even a combined inhibition might not work. In a comprehensive study by Katayama et al.,[120] H3122 CR3 cells which were resistant to both crizotinib and gefitinib remained less sensitive to the combination of crizotinib plus gefitinib compared to crizotinib alone.

MET Activation: Although mentioned in a smaller number of studies, MET activation also induces resistance to ALK-TKI. Hepatocyte growth factor (HGF) induced MET activation and triggered resistance to crizotinib and TAE684. HGF stimulated the phosphorylation of MET and its adaptor protein, GAB1, and activated downstream AKT and ERK1/2 pathways thus finally cause resistance to TAE684.[166] Another study also reported that HGF induced resistance to alectinib in H3122 and H2228 cell lines.[167] MET activation but not gene amplification was observed in tissues from patients with ALK rearrangement.[168] However, in another study, circulating tumor cells and ctDNA were analyzed by targeted NGS, and MET amplification up to sevenfold was detected after initiating crizotinib treatment.[169] Further studies are still required to demonstrate whether MET amplification contribute to resistance to ALK-TKI.

Activation of KIT or IGF1R: Besides the activation of EGFR or MET signaling, a few studies discussed KIT or IGF1R signaling, and how they contributed to resistance to ALK-TKI. ALK positive H3122 cells with KIT overexpression showed sensitive to crizotinib. However, in the presence of stroma-derived stem cell factor (SCF), KIT-overexpressing H3122 cells exhibited high resistance to crizotinib through activation of downstream intermediates ERK and AKT.[120] In another study that involved patients receiving crizotinib treatment, a significant decrease in PFS was correlated with high phosphorylation level of KIT in ALK-positive patients.[170] Activating KIT mutation D816G was also identified in crizotinib-resistant cells, however, until now, only in ROS1-positive cell lines.[171]

Through a special case of a patient with ALK-fusion who significantly responded to IGF1R-specific antibody, the combination of ALK plus IGF1R inhibitors was investigated in H3122 cell model and confirmed an enhanced antiproliferative response.[122] This synergistic effect was verified by subsequent study in NSCLC.[172] Furthermore, cellular dependence on ALK decreased because of increased IGF1R signaling induced by stimulation. IGF1R/insulin receptor substrate 1 (IRS1) signaling in the presence of ALK inhibitor therefore became a mechanism by which cells evade ALK blockade.[122] In vitro experiments have confirmed that both ALK and IGF1R activation are inhibited after treatment with ceritinib or TAE684,[173] however, the clinical utility of these two ALK-TKIs in inhibiting ALK/IGF1R still remains to be defined.

Other Novel Resistance Mechanisms

Neuregulin 1 (NRG1) Fusion. CD98hc (SLC3A2, solute carrier family 3 member 2) is the heavy chain of CD98 and forms large neutral amino acid transporter LAT1 (SLC7A5) in cells. Overexpression of SLC3A2 occurs widely in cancer cells and is associated with poor clinical prognosis.[174] SLC3A2 is upregulated in human osteosarcoma and promotes tumor growth through the PI3K/AKT signaling pathway.[175]

The NRG1 gene encodes the growth factor NRG1; members of the NRG1 family are structurally related to EGF and stimulate ERBB3 RTKs.[176] The EGF-like domain of NRG1 in the SLC3A2-NRG1 chimeric protein was shown to be critical for NSCLC proliferation and tumorigenesis.[177] A SLC3A2-NRG1 fusion protein activated the formation and phosphorylation of the ERBB2-ERBB3 heterocomplex and downstream PI3K/AKT/mTOR signaling pathway. Inhibition of both ERBB2 and ERBB3 blocked the downstream signaling intermediates AKT and ERK.[178] Therefore, the dual inhibition of ERBB2/3 might be a suitable strategy to block the signals activated by SLC3A2-NRG1.

CD74 is the most common NRG1 fusion partner, and CD74-NRG1 fusion occurs in approximately 1.7% of patients with lung adenocarcinomas.[179] CD74-NRG1 increases the expression and phosphorylation of the EGF-like domain of NRG1 III-β3 and leads to the heterodimerization of ERBB3 and ERBB2, subsequently activating the downstream PI3K/AKT pathway.[179] An increase in the NRG1 ligand level was directly related to resistance to crizotinib treatment.[180] Furthermore, the treatment of NSCLC cells with second-generation ALK inhibitors activated EGFR family pathways through activation of the NRG1-ERBB3-EGFR axis.[181,182] Therefore, activation of the NRG1/ERBB3 pathway is a potential mechanism of TKI resistance.

Invasive mucinous adenocarcinoma (IMA) is a highly malignant type of lung adenocarcinoma that is mainly caused by KRAS mutations. However, an aberrant, novel tumor driver, the CD74-NRG1 fusion gene, was also found to contribute to KRAS-negative IMA tumorigenesis.[179]CD74-NRG1 and KRAS mutations are mutually exclusive. Expression of the CD74-NRG1 protein not only induced sphere formation in vitro but also enhanced tumor initiation in vivo. The CD74-NRG1 protein activates the PI3K/AKT/NFKB signaling pathway, leading the IGF2 autocrine/paracrine circuit to initiate and maintain cells with cancer stem cell properties. IGF1R, the CD74-NRG1 receptor, was enhanced in an NFKB-dependent manner in cells expressing CD74-NRG1.[183]

Other NRG1 fusion proteins such as SDC4-NRG1 and ALK-NGR1 have been reported in some studies.[184,185] Although SDC4-NRG1 fusion displays a rapid and durable PR to afatinib and NRG1 has been shown to respond to EGFR and ERBB2/3 inhibitors in the preclinical setting,[184,186,187] targeting downstream of NRG1 through the direct inhibition of ERBB3 and other molecules in this pathway is thought to be a better strategy for clinical application than the use of a broad EGFR/ERBB family inhibitor to target mutant tyrosine kinases.[183,184]

Overexpression of MET and BCL-2. AC0010 is a pyrrolopyrimidine-based, irreversible, third-generation EGFR inhibitor that selectively inhibits EGFR-activating and T790M mutations with an up to 298-fold increase in potency compared with its inhibition of wild-type EGFR.[188] A phase I study of AC0010 suggested that AC0010 has a well-tolerated safety profile and shows promising antitumor activity in NSCLC patients with acquired resistance to a first-generation EGFR-TKIs.[189] However, AC0010 cannot overcome resistance caused by the overexpression of MET and BCL-2. The BCL-2 inhibitor navitoclax inhibited cell growth in the AC0010-resistant cell line H1975-AVR1.[97,138] Navitoclax together with gefitinib showed an enhanced ability to eradicate NSCLC cells.[190] Combination treatment with AC0010 and crizotinib inhibited the growth of H1975-P1-R1 cells, and synergistic effects with a 73.5% inhibitory rate at a nontoxic dose were observed.[138] Thus, the overexpression of BCL-2 and MET is responsible for acquired resistance to AC0010 in NSCLC.