ALK Rearrangements Are Mutually Exclusive with Mutations in EGFR or KRAS: An Analysis of 1,683 Patients with Non–Small Cell Lung Cancer

Discussion

In NSCLC, genetic alterations in EGFR, KRAS, and ALK are considered to be largely mutually exclusive (7, 10, 33). However, recent series have suggested that EGFR mutations and ALK rearrangements coexist within patients at clinically meaningful frequencies (3, 15–19). The question of whether overlap exists between these driver oncogenes is clinically relevant and would impact therapeutic choices. In vitro studies suggest that concomitant alterations in EGFR and ALK lead to mutual resistance to EGFR or ALK TKI monotherapy (16). In limited clinical series of patients reported to have both alterations, isolated responses to erlotinib, gefitinib, and crizotinib have been described (11, 12, 16, 18, 19). However, to the best of our knowledge, there have been no reports in which individual patients with co-existing EGFR mutations and ALK rearrangements responded to both EGFR and ALK TKIs.

In this study, we examined a large Western population of patients with NSCLC who underwent clinical genotyping. In contrast to the reports mentioned above, we found no cases of coexisting EGFR mutations and ALK rearrangements across 1,687 screened specimens, supporting the mutual exclusivity of these 2 genetic alterations. We did identify a few cases of concomitant KRAS mutations and abnormal ALK FISH. While abnormal, however, these FISH patterns did not meet criteria for an ALK rearrangement. Specifically, most cases (3/4) involved single isolated green 5′ probes, rather than the typical split probe or isolated 3′ probe pattern. To confirm the absence of a functional ALK rearrangement in these specimens, ALK immunohistochemistry was conducted and found to be negative in 3 of 3 cases tested. In a prior series using this antibody, ALK immunohistochemistry had a negative predictive value of 100% (30). Thus, these findings are unlikely to be due to false-negative staining, as the probability of obtaining 3 false-negative immunohistochemical results among these cases would be less than 0.000001. We therefore conclude that these abnormal ALK FISH patterns are unlikely to represent functional ALK gene rearrangements. Altogether, these findings support the lack of overlap between ALK rearrangements and either EGFR or KRAS mutations.

We recognize that these findings are in contrast to some recent reports (14–19). One explanation for such discordant findings may involve ethnic differences between screening populations. Specifically, the higher prevalence of EGFR mutations within Asian NSCLC populations may increase the chance of detecting dual EGFR and ALK alterations (34). Indeed, the highest frequencies of concomitant EGFR mutations and ALK rearrangements have been reported in Asian populations (17–19). For instance, in one study describing concomitant EGFR mutations in 11.8% of patients with ALK rearrangements, the overall prevalence of EGFR mutations in the study population (n = 444) was 51.4% (18).

In addition to ethnic variations, our findings may differ from other reports due to the stage distribution of our patient population. A significant proportion of patients in this study had early-stage disease at the time of mutation testing. It is therefore possible that we did not capture additional genetic alterations that may have accumulated later during the disease course. Another potential explanation for reports of dual genetic alterations may be the presence of multiple separate primary tumors in patients, each harboring different driver mutations. When overlapping genetic alterations in ALK and EGFR are suspected, ALK and EGFR mutant-specific immunohistochemistry may help distinguish whether both abnormalities are colocalized in the same cells. Our findings may also differ from those of prior studies due to possible reporting bias and choice of screening techniques. ALK testing can be conducted using ALK immunohistochemistry, reverse transcription PCR (RT-PCR), and ALK FISH. While ALK FISH is the only clinically validated assay and current gold standard, this technique is technically challenging (35). In prior reports of coexisting EGFR mutations and ALK rearrangements, various methods of ALK testing were used, including ALK FISH, ALK immunohistochemistry, RT-PCR, and rapid amplification of cDNA ends (RACE)-coupled PCR (11–19). In contrast, all patients in our study were screened using ALK FISH. Of note, ALK FISH positivity in this study was defined as more than 15% of cells showing positive signals. This is consistent with the definition of ALK positivity used in clinical trials of crizotinib, and this threshold is above the technical background noise of the assay (21, 36).

As the molecular profile of malignancies may evolve over the course of treatment, we also examined the molecular characteristics of ALK-positive cancers after the development of resistance to crizotinib. Crizotinib-resistant specimens were examined for ALK gene amplification, secondary ALK TK domain mutations, and emergence of EGFR and KRAS mutations. Of particular note, we found that the presence of the ALK fusion gene as assessed by FISH was preserved across all tested specimens despite the presence of acquired crizotinib resistance. This finding is consistent with the oncogene addiction paradigm as well as results from repeat biopsy series in patients with EGFR mutations and acquired resistance to EGFR inhibitors (32, 37). Although we and others have previously identified ligand-dependent activation of EGFR as a potential resistance mechanism to crizotinib (16, 26, 38), we found no cases of EGFR or KRAS mutations among 25 examined specimens, including those cases with upregulation of activated EGFR by immunohistochemistry. Thus, our findings are different from a previous study reporting that ALK-positive patients develop ALK-negative tumors harboring new mutations in EGFR following treatment with crizotinib (27). It remains possible that the patients identified in the other study had 2 separate primary cancers, an EGFR-mutant lung cancer and an ALK-positive lung cancer, and the EGFR-mutant lung cancer emerged upon treatment with crizotinib. In such cases, more comprehensive genetic analyses may inform whether pretreatment and resistant cancers are derived from a common ancestor clone. Nevertheless, our findings on a larger cohort of patients, suggest that such occurrences will be rare in cancers with acquired crizotinib resistance.

Our study has several implications. In rare cases of suspected overlapping genetic alterations in driver oncogenes, our findings underscore the importance of using confirmatory molecular testing, preferably using different genotyping techniques. Our findings may also affect clinical laboratory workflow and resource allocation since the lack of concomitant genetic alterations in EGFR, KRAS, and ALK may permit sequential genetic testing in centers where ALK FISH is not readily available. However, this must be balanced against the need for timely acquisition of genotyping data to guide rapid initiation of therapy, particularly at the time of diagnosis. For this reason, a simultaneous molecular testing strategy also remains a viable approach. Finally, this study has implications for our understanding of crizotinib resistance. Consistent with prior reports, we observed ALK fusion gene amplification and secondary mutations in the ALK TK domain in a subset of ALK-positive patients with acquired crizotinib resistance. Nonetheless, the number of crizotinib-resistant patients examined in this and other studies have been relatively small, and a significant proportion of these patients still have unknown mechanisms of resistance to date. Resistance is likely to be a molecularly complex process, and studies of repeat biopsies in larger cohorts of crizotinib-resistant patients are needed.

There are several potential limitations to our study. It is possible that we did not detect genetic alterations present at very low allelic frequencies within our study population. To minimize this likelihood, all ALK testing was conducted using the current gold-standard assay, ALK FISH. Moreover, EGFR and KRAS mutation testing was conducted using the SNaPshot assay, which has analytic sensitivity to detect mutations present at a frequency of approximately 5% (29). Another consideration is that mutations in EGFR or KRAS may have been present, but they were outside of the mutation hotspot regions designated in our multiplexed assays. This is less likely as more than 95% of mutations in both EGFR and KRAS are captured by these assays. One last consideration is the presence of intratumoral heterogeneity. In this study, molecular testing was conducted on a single biopsy or resection specimen and thus, may not be reflective of all sites of disease. This is mainly relevant to treatment resistant specimens, as studies have shown clonality of driver gene alterations in pretreatment samples (39, 40). However, if this were a common occurrence in resistance, one would have expected to observe an EGFR or KRAS mutation in at least one of the resistant biopsies examined.

Finally, while this manuscript was under review, 2 smaller screening studies were published, confirming similar findings in 208 and 99 patients, respectively (41, 42). Together, our analyses provide support for the mutual exclusivity of alterations in EGFR, KRAS, and ALK within the Western population. By examining a series of biopsy specimens from ALK-positive patients after development of resistance to crizotinib, we also show that this lack of overlap persists in the resistant cancers.