Escaping ALK Inhibition: Mechanisms of and Strategies to Overcome Resistance

See allHide authors and affiliations

Science Translational Medicine  08 Feb 2012:
Vol. 4, Issue 120, pp. 120ps2
DOI: 10.1126/scitranslmed.3003728


Mutated anaplastic lymphoma kinase (ALK) drives the development of multiple tumor types, and ALK tyrosine kinase inhibitors such as crizotinib have been validated as targeted therapeutics. Unfortunately, as with other oncogene-driven tumors, therapeutic resistance invariably develops. In Science Translational Medicine, two recent studies provide new insight into mechanisms of resistance to ALK tyrosine kinase inhibitors and possible strategies to overcome this resistance.

One of the greatest success stories in the “War on Cancer” is the identification of small-molecule inhibitors that shut down cell signaling pathways perpetually activated by cancer-specific mutated kinases. However, therapeutic responses are invariably limited by the development of progressive disease while the patient is still receiving therapy, a phenomenon referred to clinically as acquired resistance. The molecular basis of acquired resistance has been characterized in many oncogene-dependent cancers, such as chronic myelogenous leukemia (CML), mutated EGFR (epidermal growth factor receptor)–driven lung cancer, mutated KIT–driven gastrointestinal stromal tumor (GIST), and mutated BRAF–driven melanoma. Mechanisms of resistance include alterations within the gene that encode the drug target (for example, second-site mutations, alternative splicing, gene amplification) and reactivation of the primary signaling pathways through alternative signaling molecules (Fig. 1.) [for review, see (1)]. In this issue of Science Translational Medicine, Katayama and colleagues describe clinical and molecular manifestations of acquired resistance to inhibitors of anaplastic lymphoma kinase (ALK), a tyrosine kinase that is mutated in and drives the development of a subset of non–small cell lung cancers (NSCLCs) (2). We also discuss the recently reported related findings of Bresler and colleagues on the differential inhibitor sensitivities of ALK mutations found in neuroblastoma (3).

Fig. 1. Resisting arrest.

Mechanisms of acquired resistance to crizotinib in NSCLCs described by Katayama and colleagues (2) include mutations within the ALK kinase domain (top left), amplification of the ALK gene (top right), and “bypass” signaling with increased EGFR phosphorylation (bottom right) or KIT amplification (bottom left).


The ALK gene is located on chromosome 2p23 and encodes a receptor tyrosine kinase that is expressed predominantly in neural tissues (4). The gene was originally discovered in anaplastic large-cell lymphoma (ALCL) as part of the t(2,5) chromosomal translocation that fuses the C-terminal kinase domain of ALK to the N terminus of nucleophosmin (NPM) on chromosome 5q35 (5). Subsequently, a variety of ALK fusion proteins were found in other malignancies, including NSCLC (6), inflammatory myofibroblastic tumor (IMT) (7), and more rarely in breast, colon, and renal cell cancers (8, 9). Although multiple different 5′-fusion partners have been found in these malignancies, the ALK kinase domain is conserved in each fusion described. All ALK fusions tested biologically to date have demonstrated gain-of-function properties (5, 6, 10). Activating point mutations within the full-length (that is, nonfusion) ALK gene have also been detected in both familial and sporadic neuroblastoma (1114) and more recently in anaplastic thyroid cancer (15). Most point mutations described in neuroblastoma occur within the tyrosine kinase domain and are transforming in vitro and in vivo (1114).

Together, these findings illustrate that mutated ALK functions as an oncogenic driver across histologically diverse tumors. As such, ALK represents an attractive therapeutic target in human cancer. Recently, in a phase I clinical trial, crizotinib—the first-in-class orally available ALK tyrosine kinase inhibitor (TKI)—was shown to induce a radiographic response in a remarkable 57% of lung cancer patients with advanced ALK + NSCLC, all of whom had previously received standard cytotoxic chemotherapy (16). As a point of reference, the usual response rate for cytotoxic chemotherapy in the setting of relapsed NSCLC is ~10% (17). This study led to the approval of crizotinib by the U.S. Food and Drug Administration (FDA) in August 2011 for the treatment of locally advanced and metastatic NSCLCs that express an abnormal ALK gene. Therapeutic responses to crizotinib have also been reported in patients with ALCL and IMT that harbor ALK rearrangements (18, 19).

Despite these exciting results, most patients whose disease initially responded to crizotinib have developed progressive disease. The mechanisms of acquired resistance to ALK inhibitors are just beginning to be delineated. In their new work, Katayama and colleagues (2) report findings from a series of lung cancer patients with acquired resistance to crizotinib. In the 18 rebiopsy samples analyzed, the authors detected an array of different mechanisms whereby the initially drug-sensitive tumor evades ALK inhibition. The general types of mechanisms are analogous to those found in other kinase-driven cancers (Fig. 1.).

The development of mutations within the drug target that alter drug sensitivity is an important mechanism of acquired resistance to crizotinib. In the series of 18 tumors evaluated at the time of progression, 4 (22%) were found to have distinct mutations that gave rise to changes within the ALK kinase domain, including an L1196M substitution (that is, a leucine-to-methionine substitution at amino acid 1196), a G1202R substitution, an S1206Y substitution, and a threonine insertion at amino acid 1151 (1151Tins). The L1196M substitution (Fig. 2., Table 1.) occurs at the conserved gatekeeper site within the kinase domain and is analogous to mutations that are commonly found in patients with acquired resistance to EGFR inhibitors in mutated EGFR–driven lung cancer (that is, the T790M substitution) (20) or to c-Abl tyrosine kinase inhibitors in BCR-ABL–driven chronic myelogenous leukemia (that is, the T315I substitution) (21). The gatekeeper site is critical for binding of competitive inhibitors to the adenosine triphosphate (ATP)–binding pocket of various kinases (22); therapy-associated mutations at this site interfere with inhibitor binding and, therefore, inhibition of kinase activity by targeted therapeutics. The L1196M mutation was reported previously in three different lung cancer patients with acquired resistance to crizotinib (23, 24) and has also been detected in cell culture models of ALK-TKI resistance (25). Two of the other ALK kinase–domain mutations reported in this study, G1202R and S1206Y, occur at the solvent front (Fig. 2., Table 1.) and are believed to mediate resistance by altering crizotinib binding. The final mutation, 1151Tins, occurs N-terminal to the alphaC-helix and is thought to affect ATP affinity in the mutated kinase. The authors detected 1151Tins both in a clinical sample and in a cell culture model of crizotinib resistance. Importantly, each of these mutations was shown to confer resistance to crizotinib in vitro in standard (Ba/F3) cell models.

Fig. 2. Mapping crizotinib resistance.

Illustrated are alterations (substitutions, insertions) in the ALK kinase domain that result in crizotinib resistance. The bar graph shows the frequency of each mutation that gave rise to an amino acid substitution or insertion in ALK; frequency is indicated as the number of times a selected mutation was detected in the clinical samples described to date. The numbers used indicate the amino acid positions in full-length (that is, nonfusion) ALK.

Table 1. Therapeutic interference.

Compilation of ALK kinase domain amino acid substitutions associated with acquired resistance to crizotinib.

View this table:

Several other acquired ALK mutations have also been reported (Fig. 2., Table 1.). Two nonoverlapping mutations, C1156Y and L1196M, were detected in one patient’s biopsy samples taken at the time of disease progression (23). Another patient’s tumor contained an L1152R mutation (26). In a separate series of 11 tumor rebiopsy samples, 4 were found to have ALK mutations, including the L1196M mutation (n = 2) and a novel G1269A mutation (n = 2). The G1269A substitution occurs in the ATP-binding pocket and confers resistance to crizotinib in vitro (24). Finally, a patient with a crizotinib-sensitive IMT that harbored a RANBP2-ALK fusion was found to have an F1174L substitution at the time of progression on crizotinib therapy (27). The F1174L substitution is the same one found in patients with mutated ALK–driven neuroblastoma (1114). In summary, in studies published to date, kinase-domain mutations associated with acquired resistance to crizotinib have been found in tumor cells from 10 of 31 (32%) patients with ALK–fusion positive lung cancer at the time of disease progression. Thus far, there does not appear to be an obvious way to predict, on the basis of clinical characteristics or genotype (for example, duration of crizotinib therapy or the ALK fusion variant detected, respectively), which acquired-resistance mutation, if any, will develop.

Different ALK mutations appear to confer variable degrees of resistance to crizotinib. For example, Katayama and colleagues (2) demonstrated that cells that express the 1151Tins or L1196M versions of ALK were significantly more resistant to crizotinib in vitro than were cells that expressed the G1202R or S1206Y ALK substitutions. The authors suggest that this finding is reflective of how these different alterations change the ALK kinase domain. Importantly, the reported ALK substitutions also gave rise to fusion proteins that displayed varying affinities to second-generation ALK-TKIs (2). ALK fusions with substitutions that affect the solvent-front (G1202R and S1206Y) were effectively inhibited (albeit at higher concentrations than the wild-type kinase) by two different second-generation ALK inhibitors (CH5425802 and ASP3026); in contrast, ALK fusions with the L1196M and, in particular, the 1151Tins substitutions remained highly resistant. This varying degree of efficacy toward ALK inhibition has also been shown for other mutated ALK proteins and ALK inhibitors (28, 29).

In support of the finding of altered ALK-TKI sensitivity based on the presence of specific ALK kinase–domain alterations, Bresler and colleagues (3) reported in a recent issue of Science Translational Medicine on the differential inhibitor sensitivities of neuroblastoma-associated, mutated ALK receptors. The authors use elegant biochemical and cell biology assays, as well as molecular modeling, to evaluate the effectiveness of ALK inhibitors against ALK F1174L and R1275Q, the most frequent ALK amino acid substitutions found in neuroblastoma. Although both substitutions confer similar levels of ALK activation as assessed by ALK autophosphorylation, the F1174L substitution is significantly less sensitive to the inhibitory effects of crizotinib. The F1174L substitution does not appear to alter the relative binding of inhibitors to the ALK tyrosine kinase domain. Rather, the difference in crizotinib sensitivity appears to be based on the higher affinity for ATP displayed by ALK F1174L compared to ALK R1275Q. As a result, the binding of ATP competitive inhibitors, such as crizotinib, to ALK F1174L is decreased at physiological ATP concentrations. The study by Bresler and colleagues explains how the ALK F1174L functions as both an activating substitution, as well as a TKI-resistant substitution.

ALK amplification represents another mechanism of acquired resistance. One patient in the Katayama et al. (2) series was found to have ALK amplification without concurrent mutation at the time of disease progression. This finding is consistent with the author’s previous in vitro modeling of crizotinib resistance in ALK fusion–positive lung cancer cell lines (25). Doebele and colleagues also recently reported a gain in ALK copy number, both with and without simultaneous ALK kinase–domain mutation, in lung cancer patients with acquired resistance to crizotinib (24). Interestingly, the loss of an ALK genomic rearrangement was also described as a potential mechanism for crizotinib resistance in the latter study; rebiopsy samples from two lung cancer patients with acquired resistance to crizotinib displayed decreased ALK gene rearrangement signals, as assessed by ALK fluorescent in situ hybridization (ALK FISH). One of these patients (with an ALK FISH signal of 80% at diagnosis and 2% at relapse) was found to have a new erlotinib-sensitive EGFR L858R substitution, while the other patient (with an ALK FISH signal of 26% at diagnosis and 8% at relapse) did not have any other identifiable cause for crizotinib resistance (24); these results reveal that we do not yet know the mechanisms of acquired resistance to crizotinib in all of the patient samples studied to date. In support of these clinical findings, loss of ALK gene rearrangement signals, as assessed by ALK FISH, has been observed in an ALK fusion–positive lung cancer cell line (H3122, which harbors an EML4-ALK E13;A20 fusion) experimentally made to be resistant to the ALK-TKI TAE-684 (26).

Bypass signaling is the final mechanism of acquired resistance to crizotinib described both in vitro and in vivo (Fig. 1.). Currently, the signaling pathways downstream of ALK fusion proteins are incompletely understood; however, these aberrant proteins are thought to signal through the mitogen-activated protein kinase (MAPK) and the phosphoinositide-3 kinase (PI3K)-AKT (PI3K/AKT) pathways (25, 26, 29). Upstream regulators of ALK are not clearly defined, and alternative tyrosine kinases that may compensate for ALK are unknown. Initial preclinical studies have demonstrated that resistance to ALK inhibition may be mediated by coactivation of EGFR signaling without concomitant EGFR mutation or gene amplification (26). In the Katayama et al. cohort (2), increased EGFR phosphorylation was found in 4 of 9 rebiopsy specimens at the time of disease progression. There has also been one case reported of a lung cancer patient with acquired resistance to crizotinib whose tumor at the time of progression became ALK gene–rearrangement negative (by ALK FISH) but demonstrated the presence of a new EGFR-activating mutation (24). Interestingly, Sasaki and colleagues have previously reported on a subset (3 of 50; 6%) of tumors from treatment-naïve NSCLC patients who harbor both ALK rearrangements and concurrent EGFR-activating mutations (26). The interplay between EGFR and ALK in the crizotinib-naïve setting, as well as the degree to which coactivation of EGFR confers resistance to ALK inhibitors, remains to be determined.

Other potential bypass mechanisms described to date include KIT gene amplification and KRAS mutation. KIT gene amplification without concurrent KIT mutation was found in one patient in the Katayama series (2). This tumor also harbored an ALK G1202R substitution at the time of relapse, suggesting that multiple mechanisms of resistance may be possible within the same tumor. In addition, there have been two cases of ALK rearrangements co-occurring with a KRAS codon 12 mutation (24). In one case, the KRAS mutation was found only in the postcrizotinib biopsy sample, while in the other case, the KRAS mutation was found in both the pre- and posttherapy samples. The latter patient apparently did have a partial response to crizotinib, and in vitro modeling reveals that the introduction of a KRAS G12C substitution into the H3122 lung cancer cell line (which carries the EML4-ALK E13;A20 fusion gene) does not alter the sensitivity to crizotinib.

Taken together, this mosaic of data from patients and in vitro experiments suggest that escape of ALK fusion–driven tumors from ALK inhibition occurs via diverse mechanisms. Adding further to the complexity, multiple mechanisms of acquired resistance may manifest within a single patient’s tumor sample at the time of disease progression. Therapeutically, this complexity of possible changes at both the gene and protein levels presents a daunting challenge and strongly argues in favor of the need for tumor rebiopsy at the time of disease progression in order to appropriately prioritize therapeutic options. Furthermore, new treatment strategies, including rationally selected combinations of targeted therapies, are needed to delay or prevent resistance in treatment-naïve cancer patients. Potential combinations based on available data include the use of ALK + EGFR inhibitors (2, 24, 26) or ALK + mTOR inhbitors (29).

Lingering questions that remain to be answered include the frequency of ALK kinase–domain alterations in patients with acquired resistance to crizotinib and whether disease in these patients will respond to second-generation ALK-TKIs and other targeted therapies, such as heat shock protein–90 (HSP-90) inhibitors. ALK is known to be an HSP-90 client, and inhibitors of this protein have demonstrated preliminary efficacy in crizotinib-naïve lung cancer patients (30). It also remains to be determined if the specific ALK fusion variant detected within a tumor modifies crizotinib responses or resistance mechanisms and if tumor morphological changes, such as epithelial-to-mesenchymal transition or transformation from NSCLC to small cell lung cancer histology, play a role in the development of resistance to ALK-TKIs, as has been shown for EGFR TKIs in oncogenic EGFR–driven lung cancer (31). Importantly, the collected data have potential implications for an increasing number of histologically diverse tumor types in which ALK is a known oncogenic driver; thus, it is incumbent on clinical and translational researchers to continue to evaluate primary tumor specimens in treatment-naïve and treatment-resistant settings, with the overall goal of improved therapeutic responses in patients with ALK-driven malignancies.

References and Notes

  1. Competing interests: W.P. has received research funding from Xcovery. Funding sources: C.M.L. was supported by a Conquer Cancer Foundation/American Society of Clinical Oncology Young Investigator Award, a Uniting Against Lung Cancer award, and the Vanderbilt University School of Medicine Division of Hematology-Oncology T32 training grant. C.M.L. and W.P. were both supported by the Vanderbilt Specialized Program of Research Excellence in Lung Cancer grant (CA90949) and the Vanderbilt-Ingram Cancer Center Core grant (P30-CA68485).
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article