ALK tyrosine kinase inhibitors are emerging as effective clinical therapies for cancers containing genetic rearrangements in ALK
including NSCLC, IMT and ALCL (8
). However, the clinical success of this therapeutic approach is uniformly limited by the development of drug resistance. The mechanistic understanding of drug resistance may help to develop effective subsequent clinical treatments and/or rational combination therapeutic strategies.
In the current study, by studying patient derived tumors and cell lines, we uncover novel ALK TKI resistance mechanisms. These include both a secondary mutation (L1152R) in ALK and activation of EGFR signaling. Importantly these can occur together in the same tumor (i.e. the DFCI 076 cell line) highlighting both the complexity of drug resistance mechanisms and the therapeutic challenges in developing strategies to overcome clinical drug resistance.
Secondary mutations in kinases are a common mechanism of drug resistance to kinase inhibitors (4
). A few distinct categories of mutations have so far been identified. These include secondary mutations that alter drug contact residues thus creating a steric hindrance for drug binding (30
). Alternatively secondary mutations can promote conformational changes in the kinase thus disfavoring the binding of a kinase inhibitor (30
). The L1152R mutation is not located in the kinase domain. The currently available crystal structures of ALK do not provide a clear explanation of the mechanistic basis of drug resistance imparted by this mutation. Notably this mutation, unlike the F1174L mutation, is also resistant to TAE684 (14
). Thus structurally distinct ALK inhibitors are needed to overcome this mutation and several are under preclinical development. Additional studies, including solving the crystal structure for the ALK L1152R will be necessary to obtain further insight into how this mutation causes drug resistance.
Prior studies have generated crizotinib resistant H3122 cells and detected both evidence of an ALK
amplification and the presence of the L1196M gatekeeper mutation(31
). We also identify ALK
amplification in a subset of the H3122 TR3 cells (Fig. S3B
), but not the L1196M mutation. Since TAE684, unlike crizotinib, can effectively inhibit the growth of H3122 EML4-ALK
L1196M cells, our findings are consistent with the prior studies(31
). In fact they suggest that a more potent ALK inhibitor may be able to prevent the emergence of this specific drug resistance mechanism. Whether this will ultimately translate into a clinical benefit (such as a prolongation if progression free survival) for NSCLC patients can only be determined from clinical trials.
Our studies identify activation of EGFR signaling as a bypass signaling mechanisms that contributes to ALK inhibitor resistance. Concurrent inhibition of both EGFR and ALK is therapeutically effective in all of the resistant models. Intriguingly different models have differing degrees of apparent EGFR dependence. The DFCI076 cells are mostly ALK dependent () while the H3122 TR3 cells are more EGFR dependent () for their growth. The DFCI032 cells are equally co-dependent with very little effect on growth by only EGFR or ALK inhibition (Fig. S2F
)). These different examples may be representative a dynamic process with variable degrees of adaptation to EGFR signaling in the presence of ALK inhibition. However, we cannot completely exclude the possibility that activation of EGFR signaling in these cell lines did not arise in the process of generating the cell lines. Additional evaluation of tumor specimens for changes in EGFR phosphorylation obtained from patients that have developed crizotinib resistance will be necessary. Further investigation is also needed to study changes in EGFR signaling over time to further understand how this adaptive process evolves. Furthermore, whether the process will revert in the absence of ALK inhibition needs to be determined. Such observation may be clinically significant as it would suggest that drug resistant cancers could regain their sensitivity following a therapeutic holiday from ALK inhibitors. To date we have grown the H3122 TR3 cells 60 passages in the absence of TAE684 and have not observed a reversion to TAE684 sensitivity (data not shown).
It is noteworthy that both in vitro and in NSCLC patients, activated EGFR signaling occurs concurrently with EML4-ALK. Such cancers could still retain sensitivity to single agent EGFR or ALK inhibitors if the tumor was heterogenous and contained two independent populations: one with EML4-ALK and one with an EGFR mutation. Alternatively a tumor could contain both genetic alterations but only expressed one of the mutant proteins. In both instances, such patients may achieve a transient partial response following therapy with either single agent. Our limited studies of crizotinib naïve NSCLC patients, with both genetically confirmed EML4-ALK and EGFR mutations, suggest that ALK is not expressed as both patients treated with erlotinib achieved a clinical response. In contrast, the in vitro studies would predict that co-expression of EML4-ALK and mutant EGFR in the same cells would lead to resistance to both single agent ALK and EGFR inhibitors. Why some cancers harbor an ALK rearrangement which does not lead to ALK expression remains to be determined. It will also be of interest to determine whether the mechanism of erlotinib resistance in our patients, with both an EGFR mutation and ALK rearrangement, will involve reactivation of ALK expression.
Currently there is an ongoing phase I clinical trial of crizotinib and PF299804 (NCT01121575) originally designed to evaluate the therapeutic benefit of inhibiting MET (crizotinib is a potent MET inhibitor) and EGFR T790M in erlotinib resistant EGFR mutant NSCLC patients. However, our studies suggest that combination of crizotinib and PF299804 may represent a rational therapeutic strategy for at least a subset of EML4-ALK NSCLC patients that develop acquired crizotinib resistance.