Differential Sensitivity to Crizotinib: Does EML4-ALK Fusion Variant Matter?

Abstract

Recent years have borne witness to the development of numerous anaplastic lymphoma kinase (ALK) tyrosine kinase inhibitors (TKIs) for patients with ALK-rearranged non–small-cell lung cancer (NSCLC). Crizotinib, amultitargeted ALK/ROS1/MET inhibitor, was the first ALK-targeted agent tested in the clinic. In randomized phase III trials, crizotinib showed superior efficacy compared with standard firstand second-line chemotherapy in advanced ALKpositive NSCLC, leading to the widespread adoption of crizotinib as the standard of care for this molecular subset of lung cancer. Since then, two second-generation ALK inhibitors, ceritinib and alectinib, have received accelerated approval by the US Food and Drug Administration for patients with crizotinib-refractory, ALKpositive NSCLC. Several other next-generation ALK TKIs are also under development. Although most patients with ALK-rearranged NSCLC respond to ALK TKIs, a wide range of responses has been observed. For example, although the average duration of crizotinib treatment is 8 to 11 months, some patients experience relapse within only a few months, and others have not yet experienced disease relapse after more than 5 years. The biologic mechanisms underlying this heterogeneity in TKI response are unknown. One of the earliest proposedmechanisms focused on the different variants of the most common ALK rearrangement, EML4-ALK. To date, more than 10 distinct EML4-ALK variants have been identified, each with a discrete breakpoint in EML4. Other fusion partner genes including KIF5B, TFG, and KLC1 have also been reported. In all cases, the fusion oncogene contains the 59 portion of the partner gene, including its promoter, and the entire ALK kinase domain (exons 20 to 29). The fusion partners generally contain oligomerization domains leading to constitutive, ligandindependent ALK activation. Early in vitro studies suggested differential sensitivity of four EML4-ALK variants (v1, v2, v3a, and v3b) to crizotinib and the tool compound TAE684. Interestingly, the differential sensitivity observed correlated with the protein stability of the EML4-ALK variants. This work suggested a potential molecular basis for the differential responses seen with crizotinib. In the article accompanying this editorial, Yoshida et al report the results of a single-institution retrospective analysis addressing the question of clinical impact of EML4-ALK fusion variants on crizotinib sensitivity in patients. The efficacy of crizotinib was examined in 35 patients with ALK-positive NSCLC, whose tumor specimens were assessed for the presence of ALKvariants by reverse transcription polymerase chain reaction with concurrent testing by either fluorescent in situ hybridization (n 5 25) or immunohistochemistry (n5 28). Nineteen patients (54%) had EML4-ALKv1, five (14%) had v2, four (12%) had v3a/3b, and seven (20%) had other variants. Non-EML4 partner proteins were not assessed. Notably, crizotinib was the first-line treatment of 10 patients (29%) and second-line or later treatment for the remaining 25 patients (71%). For the entire cohort, both the response rate (69%) and median progress-free survival (PFS) time (9.7 months) were consistent with published data from crizotinib trials. However, although the response rates were comparable in v1 and non-v1 groups (74% and 63%, respectively; P 5 .7160), there was a statistically significant difference in the disease control rate (95% v 63%, respectively; P5 .0318) and median PFS (11 v 4.2 months, respectively; P, .05). These observations support the notion that the type of ALK fusion could be a determinant of sensitivity to crizotinib in ALKrearranged lung cancers. However, there are several important limitations of this study. First, there is no biologic rationale for categorizing patients with ALK-rearranged lung cancer on the basis of the presence of EML4-ALK v1 versus non-v1. Indeed, the findings of this study are not consistent with prior in vitro studies that showed differences among the EML4-ALK variants v1, v2, v3a, and v3b. On the basis of the in vitro results, EML4-ALK v2 (not v1) was most sensitive to ALK inhibitors, v1 and v3b had intermediate sensitivity, and v3a was least sensitive. At present, there are limited data to suggest how biologically similar, or different, the non-v1 EML4-ALK variants may be compared with one another; therefore, grouping these variants together is somewhat artificial. Second, the study by Yoshida et al was limited by the small sample size, with only four to seven patients in each individual subgroup of non-v1 variants. Within the v1 and non-v1 groups, a remarkably wide range of PFS times was observed. For example, one patient with EML4-ALK v5b had PFS censored at greater than 36 months, which is significantly longer than the median PFS of the v1 group. Furthermore, for three of the 16 patients in the non-v1 group, the PFS was censored at a relatively short follow-up (, 6 months). It is worth noting that other studies examining the role of ALK variants have also been limited by small numbers and have not reported differences in response on the basis of specific variant. The original phase I study of crizotinib included analysis of EML4-ALK variants by reverse transcription polymerase chain