The Resistance Mechanisms and Treatment Strategies for ALK-Rearranged Non-Small Cell Lung Cancer

Anaplastic lymphoma kinase (ALK) is a validated molecular target for non-small-cell lung cancer (NSCLC). The use of tyrosine kinase inhibitors (TKIs) has led to significantly improved survival benefits. However, the clinical benefits of targeting ALK using TKIs are limited due to the emergence of drug resistance. The landscape of resistance mechanisms and treatment decisions has become increasingly complex. Therefore, continued research into new drugs and combinatorial therapies is required to improve outcomes in NSCLC. In this review, we explore the resistance mechanisms of ALK TKIs in advanced NSCLC in order to provide a theoretical basis and research ideas for solving the problem of ALK drug resistance.


BACKGROUND
Rearrangement of the anaplastic lymphoma kinase (ALK) gene creates potent oncogenic drivers in patients with non-small cell lung cancer (NSCLC) occurring in approximately 3-7% of all cases. The most common fusion partner is EML4 (echinoderm microtubule associated protein like 4) (1). In addition, at least 20 other types of fusion genes have been discovered and reported, such as TGF-ALK, KIF5B-ALK, and STRN-ALK. ALK+ NSCLC has been associated with the absence of smoking, younger age, and adenocarcinoma histology (2). Tyrosine kinase inhibitors (TKIs) targeting ALK have made significant breakthroughs in recent years such as extending patients' survival periods with ALK-advanced NSCLC. To date, ALK TKIs have received approval from the US Food and Drug Administration (FDA) and European Medicines Agency (EMA) to treat advanced "ALK-positive" NSCLC. These ALK TKIs include crizotinib (first-generation), ceritinib, alectinib, brigatinib (second-generation), lorlatinib (third-generation). Clinical trials demonstrated remarkable responses within this patient population (Table 1). However, the clinical benefits of ALK inhibitors (ALKi) are almost universally limited by the emergence of multi-drug resistance. In this review, we analyze and summarize the mechanisms of resistance, as well as treatment strategies after resistance, in order to provide better therapeutic strategies for clinicians.

MECHANISMS OF RESISTANCE TO ALK TKIS
Resistance is divided into primary and acquired resistance. Primary resistance is defined as the de novo lack of treatment response and can be seen after treatment with a TKI (23). While the mechanism of resistance to ALKi is less well-understood, it can be divided into two categories, on target or ALK dependent alterations and off target or ALK independent alterations.

Secondary Mutations in the ALK Tyrosine Kinase Domain
Resistance mutations in the ALKi account for 30-40% of all known resistance mechanisms (24). These resistance mutations lead to structural changes in the kinase domain that interfere with the binding of the drug. A much broader spectrum of on-target mutations has been identified in ALK-positive NSCLC treated with ALK TKIs. Resistant mutations to crizotinib include L1196M、G1269A、C1156Y、G1202R、I1171T/N/S、 S1206C/Y、E1210K、L1152P/R、V11180L、I1151T、 G1128A、and F1174V (25)(26)(27)(28) (Figure 1). The most common ALK mutations are mutations L1196M and G1269A, where the deep binding pocket of ATP. G1202R was found in 2% of the samples following crizotinib resistance and was the primary mechanism of second-generation ALKi resistance (Figures 2A, B).
After the first-generation ALK inhibitors exhibit resistance within NSCLC, many studies have shown that the sequential second-generation drugs alectinib, ceritinib, brigatinib, and ensatinib can achieve better curative effects and are superior to chemotherapy (11,15,29,30). Of note, the second-generation TKIs alectinib and brigatinib are currently the preferred first-line therapies in Europe (31), while the third-generation compound lorlatinib is also approved as initial therapy by the FDA and an additional preferred first-line drug according to the current NCCN guidelines (32) ( Table 1). The progression free survival (PFS) of alectinib was significantly better than that of crizotinib, response rate (RR) and PFS of 92% and 34.1 months, respectively. The G1202R mutation is the most common secondary resistant ALK mutant in patients post-treatment with second-generation ALK inhibitors, occurring in 21%, 29% and 43% of patients treated with ceritinib, alectinib, and brigatinib (33). It is speculated that although the secondgeneration ALKi have increased activity, one of the costs was the larger molecular volume of their compounds, which is heavily dependent on the direct binding to the solvent front region such as G1202 in order to increase its activity; thus, "inducing" resistance mutations within this region. Resistant mutations to alectinib include G1202R and I1171N. Tumor mutation burden and heterogeneous tumor evolution might be responsible for the rapid acquisition of alectinib resistance (34). Resistant mutations to ceritinib include G1202R、F1174V、 T1151K、and T1151R (27,35,36), and to brigatinib include D1203N, and E1210K (37,38).
Lorlatinib is a reversible third-generation ALK and ROS1 inhibitor that can overcome multiple ALK resistance mutations and penetrate the blood-brain barrier. Lorlatinib has strong activity for common mutations such as L1196M and G1269A. The G1202R mutation is particularly important as it is the primary resistance mechanism to ceritinib, alectinib, and brigatinib, whereas only lorlatinib can inhibit the ALK G1202R mutation ( Figure 2C) (39). The whole exome sequencing of compound ALK mutations occurring in several lorlatinib-resistant patients confirms the stepwise accumulation of ALK mutations during sequential treatment. Several of these ALK kinase compound mutations that have been described include the L1196M/D1203N, F1174L/ G1202R, and C1156Y/G1269A mutations (40). Absolute IC50 values of crizotinib, ceritinib, alectinib, brigatinib, and lorlatinib on cellular ALK phosphorylation in Ba/F3 cells are depicted (33). In Ba/ F3 cells, ALK F1174C and ALK I1171T appear sensitive to ceritinib and alectinib, respectively; however, these mutations may not be susceptible to these agents in vivo based upon prior clinical reports. Therefore, we further combined clinical data at the cellular level in Table 2, which can help with medication selection after resistance. The objective response rate (ORR) was 69% in patients who had received crizotinib or crizotinib plus chemotherapy (21),which means that regardless of the previous use of several first or secondgeneration ALKi or chemotherapy, the efficacy of lorlatinib as a follow-up treatment is superior. Furthermore, the ORR of lorlatinib and crizotinib as a first-line therapeutic for advanced ALK+ NSCLC is 76% and 58%, revealing that lorlatinib has an advantage in regard to efficacy (22).
When patients receive sequential ALKi treatment, the cancer cells accumulate new mutations in addition to the previously acquired mutations, making treatment more complex (33,37,40,60). Takahashi, Ken reported a patient who underwent sequential treatment with crizotinib, alectinib and lorlatinib; thus, developing the double mutations I1171S and G1269A.
Ceritinib and brigatinib have the potential to become the therapeutic agents to treat this double mutation (61). Geeta G. Sharma's team reported a case of ALK-positive NSCLC with the dual mutation ALK L1196M/G1202R after brigatinib treatment. Lorlatinib was effective against the G1202R mutation. Interestingly, this patient's L1196M/G1202R dual mutation also increased primary resistance to lorlatinib, further limiting treatment options (62). ALK D1203N was significantly more common at relapse with lorlatinib than second-generation ALKi's (63). In one case of ALK-positive NSCLC, after the failure of continuous treatment with crizotinib and alectinib, the mutation of the ALK fusion gene L1196M was detected, and no other acquired drug resistance mechanism was found. The patient developed resistance to alectinib, but remained sensitive to ceritinib (64).
However, not all complex mutations increase the difficulty of treatment (Table 3). Interestingly, some compound mutations that lead to lorlatinib resistance led to re-sensitization of the first or second generation ALKi (65). A patient receiving sequential treatment for ALK-positive NSCLC was resistant to crizotinib due to the mutation C1156Y in the ALK kinase region ( Figure 3A). Sequencing revealed the mutation ALK L1198F in addition to C1156Y ( Figure 3B). The L1198F mutation developed resistance to lorlatinib through spatial interference with drug binding. However, the L1198F mutation enhanced its binding to crizotinib ( Figure 3C), making it sensitive to the C1156Y mutation. The patient was treated again with crizotinib, resulting in the successful treatment of cancer-related symptoms and liver failure (45). Other researchers have also demonstrated that the L1198F mutation leads to conformational changes in the inhibitor site as well as changes in the binding affinity of ALK to crizotinib and lorlatinib (66).
For patients with drug resistance after ALK-TKI treatment, a re-biopsy is recommended to provide optimal treatment. Haratake N et al. retrospectively analyzed ALK-TKI treatment patterns and clinical outcomes. Of the 71 patients treated with ALK-TKI for NSCLC, 20 were re-biopsied, and 8 had secondary drug-resistant mutations. The ORR of patients with ALK point mutations receiving ALK-TKI was 88.9%, while patients without the ALK point mutations receiving ALK-TKI or chemotherapy were only 20.0%. However, PFS in patients with secondary drugresistant mutations are relatively short, and their mechanism needs to be further studied (67).

Amplification of ALK
ALK amplification occurs at a low frequency, but it is responsible for acquired resistance to crizotinib. Katayama R reports a high level of ALK amplification in 15 NSCLC patients with crizotinib resistance (24).
The activation of the EGFR pathway is one of the mechanisms of ALK-TKI resistance, such as crizotinib and alectinib. Ceritinib and afatinib combinatorial treatment partially restored the sensitivity to ceritinib (68). Afatinib may be a promising treatment for overcoming ceritinib resistance in ALK or ROS1positive NSCLC cells by inhibiting the neuroregulatory protein (NRG1) signaling pathway (75).
Increased expression of hepatocyte growth factor (HGF) and its physiological receptor tyrosine kinase MET is associated with acquired resistance to various TKIs. MET amplification was detected in 12% and 22% of biopsies of patients using secondgeneration inhibitors or lorlatinib, respectively. Patients treated with second-generation ALKi during first-line therapy were more likely to have MET amplification than those treated with second-generation ALK inhibitors after crizotinib treatment (76). Gab1 is a key effector in the HGF/MET signaling pathway that mediates alectinib resistance. The antidiabetic drug metformin combined with alectinib overcomes HGF/Met-induced alectinib resistance by blocking the complex formation between MET and Gab1, thus inhibiting Gab1 phosphorylation and activating the downstream signaling pathway. These results suggest that metformin combined with alectinib may help overcome alectinib resistance caused by the HGF/MET signaling pathway activation, improving the efficacy of alectinib (77). Cerivastatin, a rate-limiting enzyme inhibitor of the mevalonate pathway, showed anticancer activity against ALK-TKI in vitro and in vivo, accompanied by inactivation of the transcription-assisted regulator YAP. Cerivastatin can significantly induce YAP-targeted oncogenes (EGFR, AXL, CYR61 and TGFbetaR2) in drug-resistant cells, providing a theoretical basis using YAP as a potential therapeutic option in patients with acquired drug-resistant ALK-TKI (74).

Drug Efflux Pump
P-glycoproteins (P-gp) are highly conserved ATP-dependent effluents encoded by the multidrug resistance 1(MDR1) gene, also known as the ATP binding box subfamily B member 1 (ABCB1) (78). The central nervous system (CNS) is the primary site of failure in most patients with crizotinib resistance. P-gp efflux and limited diffusion of crizotinib results in limited bloodbrain barrier penetration (79). In contrast, alectinib is not a P-gp substrate and can achieve higher CNS levels (80, 81).

Lineage Changes
Morphological changes are also one of the mechanisms of ALK-TKI resistance in NSCLC. Many cases have reported drug resistance due to the conversion to small cell lung cancer (SCLC) or squamous cell carcinoma (SCC) after targeted  therapy against ALK-positive adenocarcinoma (82)(83)(84)(85)(86). Deletion of p53 and retinoblastoma (RB) genes is important for SCLC transformation, although the transformation mechanism is not fully understood (87). Mutations in the TP53 and PTEN genes were also found in a patient with SCLC transformation (88). In addition, a patient who underwent alactinib treatment developed transformed SCLC. The levels of gastrin-releasing peptide precursor and neuron-specific enolase in the patients were increased, indicating SCLC transformation during the drug resistance of ALK-tyrosine kinase inhibitors (83). In addition to the conversion of adenocarcinoma to SCC or SCLC, Koyama K. et al. reported a rare case of ALK-positive adenocarcinoma that converted to NSCLC with neuroendocrine differentiation. Histopathological examination of the tumor following alectinib resistance revealed a poorly differentiated carcinoma with insulinoma associated protein 1 (INSM1) expression. The expressions of CD133, Bcl-2, and SOX2 were positive when compared with the initial tumor. SOX2 expression was significantly increased compared to that before treatment. Immunohistochemical results of these markers associated with tumor stem-like cells and neuroendocrine differentiation suggest that tumor stem cells play a role in the histological transformation and acquired resistance mechanisms of ALKreposition-positive tumors (89). HER2 plays an important role in regulating cancer stem cell phenotypes of ALK translocation lung cancer, which is primarily mediated by HER2/HER3 heterodimers (90).
Epithelial-to-mesenchymal transition (EMT) is a morphological change in which epithelial cells lose their polarity and intercellular connections becoming more mobile and invasive. Through EMT, tumor cells acquire mesenchymal morphology and the ability to migrate and invade. There are four pathways associated with EMT: proteoglycan in cancer, HIF-1 signaling, FoxO signaling, and extracellular matrix receptor interactions, related to the drug resistance mechanisms of crizotinib (91) (Figure 4). ALK mutants L1196M and EMT were simultaneously detected in a patient with crizotinib resistance. ALK L1196M primarily existed within the epithelial tumor cells, suggesting that EMT and ALK mutations co-exist as independent mechanisms of drug resistance. EMT was associated with decreased expression of miR-200c and increased expression of ZEB1, leading to cross-resistance of the      (94).
The expression of the ATP-binding domain C-member 11 (ABCC11) in alectinib-resistant cell lines was significantly higher than that in alectinib-sensitive cell lines. This indicated that ABCC11 expression may be involved in the acquired drug resistance of alectinib (95). In addition, the neuroregulatory peptide U (NMU) may make NSCLC resistant to alectinib (96).

Primary ALK TKI Resistance
Progression of ALK-TKI within 3 months is considered primary resistance. In theory, any of these mechanisms of acquired resistance that existed before the use of TKI could also lead to primary resistance (23).
BIM is a Bcl-2 (B lymphocytoma-2) -like protein 11 that activates programmed cell death in cells. The study found that patients with BIM with missing polymorphisms had shortened PFS and reduced objective response rate, which was an independent predictor of patients treated with crizotinib and was related to primary drug resistance (97). In addition, the low minimum allele frequency (MAF) of the EML4-ALK rearrangement may also be a mechanism of primary resistance to crizotinib (98). Rihawi K reported a patient with primary resistance to crizotinib. MYC amplification was a potentially new mechanism of primary ALK-TKI, resistance and proposed as a potential MYCoriented inhibition strategy to overcome primary resistance of advanced ALK-rearrangement NSCLC (99). Similarly, the results of Pilling AB et al. showed a dual oncogene mechanism, in which ALK positively regulates the MYC signaling axis, providing an additional oncogene target (100).

DISCUSSION
The treatment of ALK-rearranged NSCLC with ALK TKIs has significantly changed these patients' outcome and quality of life. However, all patients will inevitably progress in time. Clinicians use imaging to determine whether a patient is resistant so that, if possible, they can switch to the next generation of ALK-TKI quickly. However, caution should be exercised in judging disease progression, as radiological progression may either be nontumor cell proliferation and/or accumulation (101). In patients who have received radiation therapy, sequential ALK-TKI should be recognized as radionecrosis of the central nervous system, since treatment with the next generation of ALK-TKI may increase its severity (102). The brain is the primary site of failure with ALK inhibitors in ALK-positive patients and is considered a sanctuary site owing to the blood-brain barrier (BBB) (103,104). ALK-rearranged NSCLC patients exhibiting a history of prior ALKi treatment are reported to harbor a high incidence of CNS metastases, i.e., from approximately 45 to 70%, suggesting that brain metastasis is the most common form of failure with ALKi therapy. A limitation of crizotinib is that relapse in the brain after treatment was commonly reported (104). Next-generation ALK inhibitors were designed to pass the BBB. The time to CNS progression was significantly longer with alectinib than with crizotinib (cause-specific hazard ratio, 0.16, 95% CI, 0.10 to 0.28; rate of events of CNS progression, 12% with alectinib and 45% with crizotinib), which is attributed to the expression of P-gp's on the luminal side of the BBB endothelium (9,80,105,106).
Acquired resistance has become an important issue. Previous investigations additionally presented the in vitro IC50 values for all available ALK TKIs regarding the different mutations. The findings illustrate that lorlatinib has the broadest activity against the G1202R mutation (33,107). However, whether ALKi is sensitive or resistant is complex within the real world. For example, G1202R has been detected in biopsy specimens from patients with ALK-rearranged NSCLC who relapsed on brigatinib, suggesting that its potency may be compromised with this mutation; however, some cases were effective with brigatinib treatment. This may result from the steric hindrance between the side chain of G1202R and the extended solubilization group of brigatinib ( Figure 2).
The fusion variant background should also be taken into consideration when interpreting ALK resistance mutations. Among>15 EML4-ALK variants have been identified to date, the five most common variants are variant 1 (v1; E13, A20), variant 2 (v2; E20, A20), variant 3 (v3; E6, A20), variant 4 (v4; E15, A20), and variant 5 (v5; E2, A20). The two EML4-ALK variants that together account for up to 70-80% of all EML4-ALK variants are v1 and EML4-ALK v3a/b (108). The ALTA-1L analysis by variants was the first validation of the significance of EML4-ALK variants in the context of a prospective randomized phase 3 study (109). Table 4 shows the differences in PFS between variants 1 and 3. That suggests that the ALK fusion variant may affect clinical outcomes. The reason for this difference may be relatively stable in short EML4-ALK variants, which leads to accumulation and stronger carcinogenic signaling, and their better interactions with cell skeletons, which increases the migration capabilities of V3-positive cancer cells (117,118). However, the molecular basis for this association is unknown. Besides, TP53 mutations and V3 are independently associated with enhanced metastatic spread, shorter TKI responses and inferior overall survival in ALK positive lung adenocarcinoma (115). Furthermore, ALK resistance mutations were significantly more common in variant 3 than in variant 1 (57% v 30%; P = .023). In particular, the ALK G1202R mutation was more common in variant 3 than in variant 1 (32% v 0%; P <.001). Among the patients treated with the third-generation ALK TKI lorlatinib, variant 3 was associated with a significantly longer progression-free survival than variant 1 (hazard ratio, 0.31; 95% CI, 0.12 to 0.79; P = .011) (112,119). These results suggest that among the EML4-ALK v3 patients, we should consider introducing more aggressive therapies earlier on in the course of the disease (Table 4).
Interestingly, more patients are using second-generation TKIs. However, in this final J-ALEX OS analysis, prolongation of OS in the alectinib arm was not observed compared to the crizotinib arm. This indicates that longer PFS does not translate into longer OS, which gives clinicians something to think about when using ALKi's. Therefore, there is much debate about whether PFS can be converted to OS, due to the following reasons. First, after the PFS benefit, disease progression may be faster than in the control group (120). Second, in a study that included 14 studies (N = 12567) in patients with advanced NSCLC submitted to the FDA between 2003 and 2013, a logarithmic scale scatter plot of the therapeutic effects showed no association was observed between PFS and OS in all studies (14, including the targeted studies) (R2 = 0.08; 95% CI 0-0.31) (121).
To expand PFS and OS, it is crucial to think about evidencebased treatment sequencing. Every ALK TKI has its own advantages and disadvantages (122). Therefore, a second biopsy is recommended for gene sequencing when the patient is resistant (123). However, repeated tumor biopsies to identify secondary resistance mutations are invasive and in certain cases not feasible. New tools are needed to evaluate tumor heterogeneity better and monitor tumor mutational profiles over time and throughout disease evolution (124,125). Circulating tumor DNA (ctDNA) can be used as a strategy to identify therapeutic response and drug resistance (107). In contrast to ctDNA, circulating tumor cells (CTC) are either apoptotic or alive, but viable CTCs contain tumorigenic cell clones with high relevance for metastatic progression (126). Besides, copy number variation (CNV) profiling and targeted panel sequencing from cell-free DNA (cfDNA) were also performed to monitor ALK+ NSCLC (127).
Approximately30% of ALK-positive NSCLC patients resistant to crizotinib are related to secondary ALK mutations or amplification. Therefore, the next generation of ALK-TKIs becomes sensitive to some mutations. However, nearly 40% of patients with second-generation TKI resistance are no longer dependent on ALK, so treatment opportunities for these patients are limited. Third-generation ALK-TKI or pemetrexed-based chemotherapy may be beneficial making loratinib more effective in patients with ALK kinase domain point mutation than those without ALK re-mutation (128). An effective longterm strategy may be to pre-treat with third-generation ALK-TKI in order to prevent the emergence of resistance (129). The use of immunotherapies for ALK-TKI is still lacking (130,131). Although patients with advanced NSCLC showed a good response to immune checkpoint inhibitors, this was associated with high PD-L1 expression levels, a high mutant load, and a history of smoking (132). However, ALK-positive patients tend not to smoke, have a low tumor mutation load (133), and have a poor response to PD-1 inhibition (134). Positive PD-L1 expression was associated with unfavorable clinical outcomes in patients with ALK-positive lung adenocarcinoma receiving crizotinib (135). ALK-positive tumors progressing with ceritinib therapy are not immunogenic enough to respond to immune checkpoint inhibitors (136). However, a successful pembrolizumab treatment case of lung adenocarcinoma after becoming resistant to ALK-TKI treatment due to G1202R mutation was reported (137). Therefore, the potential benefits of adding immunotherapy to ALK TKI therapy remains unclear. Activation of bypass signals has emerged as another potential strategy for combating ALK-TKI resistance. Leptomeningeal Carcinomatosis (LMC) often occurs in ALK-positive NSCLC. EGFR bypass activation is known to be the drug resistance mechanism against ALK-TKI therapy. EGFR-TKI in vitro resensitizes cells to alectinib and successfully controls the progression of LMC, indicating the therapeutic potential of new therapies targeting both ALK and EGFR for ALK-TKI resistant LMC (138). In addition, apatinib can restore sensitivity to alectinib by inhibiting the downstream ALK and anti-angiogenic signaling pathway. Furthermore, reversing ALK-TKI and inhibiting angiogenesis in combination with alectinib and apatinib, thus inhibits ALK and VEGF R2 controlling the progression of the EML4-ALK fusion gene lung cancers (139). Furthermore, PFS was more severe in patients with TP53 co-mutations than in patients with wild-type TP53, meaning the combination of proteasome inhibitors with alectinib is a promising therapy for NSCLC with ALK rearrangement/TP53 mutations (49,140).

CONCLUSION
Despite the significant efficacy of ALKi in ALK-positive NSCLC patients, drug resistance is inevitable in some patients. Although the mechanism of drug resistance can be divided into ALKdependent and non-dependent, the specific mechanisms have not been clarified, so there is urgency in developing strategies to overcome or prevent drug resistance. With a growing understanding of the mechanisms of drug resistance, a new generation of ALKi is expected to be more effective in overcoming and suppressing drug resistance. After drug resistance, it is recommended to biopsy again to identify the mutation site. Moreover, variants should also be of concern. In addition, combination therapy is also an option. However, there may be potential problems of increased toxicity or emergence of new toxicities, so these combinatorial treatment regimens still need to be explored. Furthermore, there is much debate about whether PFS can be converted to OS. In targeted therapy, it depends on the PFS1, 2, and 3. In patients with ALK fusion, the first generation may be followed by second generation therapy, or the second generation is followed by another second generation therapeutic. In short, these new approaches are promising at more effectively overcoming and suppressing drug resistance, translating into more profound and more prolonged responses in patients with ALK-driven cancers.

AUTHOR CONTRIBUTIONS
YP participated in the analysis, data interpretation, and wrote the manuscript. CD and ZQ polish the language and search the literature. CC analyzed the data and drew diagrams. FW designed the article ideas analyzed the data. All authors contributed to the article and approved the submitted version.