Lung Cancers with Concomitant EGFR Mutations and ALK Rearrangements: Diverse Responses to EGFR-TKI and Crizotinib in Relation to Diverse Receptors Phosphorylation

This article is featured in Highlights of This Issue, p. 1059

Lung cancer accounts for a large number of deaths caused by cancer worldwide (1). Similar to tumors with epidermal growth factor receptor (EGFR) mutations, non–small cell lung cancer (NSCLC) with anaplastic lymphoma kinase (ALK) rearrangements are a molecular subgroup that could benefit from crizotinib (2). Fusion of ALK with the echinoderm microtubule–associated protein-like 4 (EML4) gene was first identified in 2007 and the incidence of ALK rearrangements ranged from approximately 3% to 13% in unselected or selected patients with NSCLC (3–5). ALK rearrangements and EGFR mutations have largely been reported to be mutually exclusive (3–5), and as mutual causes of resistance to EGFR tyrosine kinase inhibitors (TKI) or ALK-TKIs (6, 7). However, such co-alterations did coexist in some clinical cases (3, 8, 9). Although the EGFR mutation rate is higher in East Asian patients as compared with Caucasians (10, 11), coexistence of ALK rearrangements might be more common in East Asian EGFR mutant patients. Both EGFR-TKIs and ALK-TKI have been approved as standards of care for EGFR- or ALK-altered disease. Patients with NSCLC with such co-alterations deserve more attention than before. The prevalence and clinical relevance of co-alterations in these 2 driver genes require detailed investigation.

First-line EGFR-TKIs in EGFR mutant NSCLC have been shown to be superior to chemotherapy in terms of response rate, progression-free survival (PFS), and quality of life (12–19). Patients with ALK rearrangements could greatly benefit from crizotinib in terms of response (2). However, for patients with concomitant EGFR mutations and ALK fusions, few data are available about the clinical activity of EGFR-TKIs and ALK-TKIs, except for limited studies showing conflicting results in terms of the response to EGFR-TKI (20–22). This study was performed to determine the prevalence of EGFR/ALK co-alterations in NSCLC. In addition, we sought to evaluate the clinical activity of EGFR-TKIs and crizotinib and the possible mechanisms in patients with co-alterations.

We prospectively screened consecutive patients from January 2010 to November 2011, for EGFR and KRAS mutations and ALK rearrangements at Guangdong Lung Cancer Institute (GLCI), Guangdong General Hospital (GGH). Histologically proven patients with NSCLC with sufficient tissue were eligible to be enrolled in this study. The prevalence of EGFR/ALK co-alterations and protein expression levels of mutant EGFR, rearranged ALK, phospho-EGFR, phospho-ALK, and downstream molecules were investigated. Objective responses to EGFR-TKI and crizotinib and PFS were also assessed. This study was approved by the Institutional Review Board at GLCI of GGH, and all patients provided specimens with written informed consents.

All advanced patients harboring EGFR/ALK co-alterations received first-line EGFR-TKIs, except for one case enrolled in the crizotinib trial after first-line platinum-based chemotherapy. EGFR-TKIs included gefitinib (250 mg, per os, every day), erlotinib (150 mg, per os, every day), and afatinib (50 mg, per os, every day). Objective responses were assessed every 6 to 8 weeks according to Response Evaluation Criteria In Solid Tumors (RECIST; refs. 23 and 24). PFS was measured from the initiation of EGFR-TKI or crizotinib treatment until radiologic or clinical progression. Four patients were recruited into the A8081005 (NCT0093245) or A8081007 (NCT0093289) trial evaluating crizotinib.

Genomic DNA from each sample was used for sequence analysis of EGFR exons 18 to 21 and KRAS exons 2 and 3. These exons were amplified by PCR as previously described (25), and the resulting PCR products were purified and labeled for sequencing using the BigDye 3.1 Kit (Applied Biosystems) according to the manufacturer's protocol.

Total RNA was extracted from lung tissue samples using the RNeasy Kit (Qiagen). Reverse-transcriptase PCR (RT-PCR) and 5′ rapid amplification c-DNA ends (RACE)-coupled PCR plus sequencing was conducted as reported previously (8). PCR products were then sequenced using a 3730XL Genetic Analyzer (Applied Biosystems). Target sequences of interest were aligned with the ALK reference sequence (NM_004304.3) to determine if a fusion with another gene was present.

Tumor histology was classified using the World Health Organization criteria. Interphase molecular cytogenetic studies using a commercially available ALK probe (Vysis LSI ALK Dual Color, Break Apart Rearrangement Probe; Abbott Molecular) were performed on 4-μm-thick paraffin-embedded sections. Samples were deemed to be FISH-positive if more than 15% of scored tumor cells had split ALK 5′ and 3′ probe signals or isolated 3′ signals (26).

Immunohistochemistry (IHC) was conducted to detect the protein expression in serial sections from formalin-fixed paraffin-embedded (FFPE) tumor samples, according to the protocols recommended by the manufacturer of the anti-mutant-EGFR and anti-ALK antibodies (Cell Signaling Technology). Rabbit monoclonal anti-human ALK antibody (#3633 WP1-01; clone D5F3) was applied at a dilution of 1:100. Staining intensity was scored from 0 to 3+. Tumors with 1+, 2+, or 3+ expression were deemed to be positive for ALK protein expression; tumors with no expression (0) were deemed to be negative (27, 28).

Fresh tumor tissues were homogenized and resuspended in lysis buffer (20 mmol/L Tris, 150 mmol/L NaCl, 1% Nonidet P-40, 10% glycerol, 1 mmol/L EDTA, 1 mmol/L EGTA), incubated on ice for 10 minutes, and centrifuged for 5 minutes (15,000 rpm). Protein concentration determination and immunoblotting were performed according to the manufacturer's protocol using antibodies against total EGFR, phospho-EGFR (p-EGFR Y1068), total ALK, phospho-ALK (p-ALK Y1604), total AKT, phospho-AKT (p-AKT S473/T308), total ERK, and phospho-ERK1/2 (p-ERK T202/Y204; Cell Signaling Technology).

The χ² test was used to compare frequencies of molecular alterations. P < 0.05 was deemed statistical significance. Kaplan–Meier curves were used to estimate PFS. General data analysis was conducted using SPSS version 13.0 (SPSS Institute).

A total of 977 NSCLCs were screened and 336 (32.7%), 70 (6.8%), and 40 (3.9%) patients had EGFR mutations, ALK fusions or rearrangements, and KRAS mutations respectively. Thirteen patients harbored concomitant EGFR mutations and ALK fusions. All of these 13 cases were adenocarcinomas, never or light smokers, of advanced stage, and as old as patients positive for ALK rearrangements alone (P = 0.218; Table 1). Five cases had acinar growth patterns and 2 had solid growth patterns of adenocarcinoma, with 42.8% (3/7) having signet cells. RT-PCR or RACE-PCR followed by sequencing identified EML4-ALK variants in 10 cases with sufficient tissues, with 5 of E13;A20 (V1), 2 of E6a/E6b;A20 (V3a/V3b), 1 of E14;ins124A20 (V4b), 1 of E2;A20 (V5), and 1 of E13;ins90A20 (V6b; Supplementary Fig. S1).

The overall frequency of EGFR/ALK co-alterations was 1.3% (13/977). Of note, the prevalence of co-alterations was 3.9% (13/336) in EGFR mutant patients and 18.6% (13/70) in ALK-positive patients, respectively. A single representative patient with co-alterations in EGFR and ALK is shown in Fig. 1. These data indicate that driver alterations of EGFR and ALK could coexist in a small group of NSCLC, and more frequent in ALK-positive tumors.

Eleven of the 13 cases with EGFR/ALK co-alterations had evaluable clinical data. Of the 10 patients receiving first-line EGFR-TKIs, 8 achieved partial response (4, 3, and 1 treated with erlotinib, gefitinib, and afatinib, respectively); 1 attained stable disease after afatinib treatment; and 1 patient treated with erlotinib had progressive disease. The objective response rate was 80% (8/10). The last follow-up date was January 5, 2012, and the median follow-up duration was 29 months (range, 17.5–40.2 months). Eight patients had progressive disease and then stopped EGFR-TKI treatment. Median PFS for first-line EGFR-TKIs was 11.2 months [95% confidence interval (CI), 6.6–15.8; Fig. 2].

Four patients entered trials to receive crizotinib therapy. Three cases having experienced first-line EGFR-TKIs were treated with crizotinib later during disease course. Among them 1 was de novo resistant to EGFR-TKI, but responsive to crizotinib, whereas 2 were responsive to EGFR-TKI but not responsive to crizotinib. One case achieved partial response and 15.1 months of PFS after the initiation of crizotinib, but did not respond to subsequent EGFR-TKI (Table 2).

To determine the potential expression pattern of EGFR mutant protein in relation to ALK fusion proteins, we tested the 2 oncoproteins by IHC analysis of serial sections in 10 cases with sufficient FFPE slides. Specific antibodies detected mutant EGFR protein in 7 cases, but not in the 3 cases with mutation types other than exon 19 del of 746E-750A or L858R of EGFR. All 7 cases showed EGFR mutant protein coexpressed and colocalized with ALK fusion proteins in the same cell population, although with diverse signal intensities, indicating that these 2 driver oncoproteins might co-operate in the same cancer cells.

To investigate the activation status of the 2 driver oncogenes in cancers with EGFR/ALK co-alterations, EGFR (Y1068) and ALK (Y1604) phosphorylation levels were also assessed by IHC. Three patterns are shown in Fig. 3: high p-EGFR and high p-ALK, high p-EGFR and low p-ALK, and low p-EGFR and high p-ALK.

To identify the molecular characteristics underscoring the efficacy of EGFR-TKI and ALK-TKI in these patients, we carefully checked the relative activation status of EGFR and ALK proteins by IHC analysis of phosphorylated proteins (and Western blotting if there was sufficient tissue; Fig. 3). Of the 8 cases treated with first-line EGFR-TKI, 6 with high levels of p-EGFR had partial responses to EGFR-TKI and 2 with very low levels of p-EGFR (+/−) had progressive disease or stable disease. Of the 4 cases treated with crizotinib, 2 (P7 and P13) had relatively inactivated p-EGFR (−, +) and highly activated p-ALK (+++,+++); one of them showed no benefit from EGFR-TKI, but a partial response to third-line crizotinib, and the other was very responsive to crizotinib, but resistant to subsequent EGFR-TKI. In contrast, 2 cases (P8 and P9) had relative high p-EGFR levels (++,+++) and low p-ALK levels (−, +), corresponding to partial response to first-line EGFR-TKI, but no benefit or short-term stable disease from crizotinib.

Western blotting yielded similar results of IHC in 3 cases (P4, P6, and P7; Figs. 1 and 4). Expression of both p-EGFR and p-ALK was consistent with the IHC data in cases P4 and P6. Notably, in treatment-naïve tissue from case P7, there were high levels of p-ALK and relatively low p-EGFR levels. After progressive disease to EGFR-TKI and partial response to crizotinib, levels of p-EGFR, p-ALK, and p-AKT in the autopsied pulmonary lesions were increased, although p-ERK levels were significantly reduced (Fig. 4). Overall, relative baseline EGFR and ALK activation correlated with the efficacy of EGFR-TKIs or crizotinib in these patients.

Although ALK rearrangements and EGFR mutations were previously reported to be mutually exclusive (4, 26, 29–35), several studies have shown that ALK fusions can occur concurrently with EGFR mutations (1/305, 0.3% or 1/103, 1.0% or 4/444, 0.9%; refs. 8, 9, and 30). Our data demonstrated that the frequency of EGFR/ALK co-alterations in NSCLC was 1.3% (13/977), which is consistent with our previous study (8). However, the frequency of such co-alterations was not described in 3 case reports (20–22). Janne and colleagues reported that 6% (3/50) of ALK-positive and crizotinib-naive NSCLCs had concurrent EGFR mutations (6). In contrast, our study showed a frequency of 18.6% (12/70) for such concomitant alterations in ALK-rearranged NSCLCs. Here, we also showed a co-alteration rate of 3.9% (12/336) in patients with EGFR mutations, which is lower than that of 15.8% (15/95) from Rosell's report at 2012 ESMO conference (36). Thus, the frequency of such co-alterations was considerably high in patients with ALK-positive or EGFR-mutant and possibly higher in Chinese patients with ALK-positive as compared with Caucasians. This observation may be of clinical relevance in terms of treatment strategies because this subgroup has a specific genotype with dual therapeutical targets.

Two or more mutations of driver genes could exist concurrently in NSCLC. In Lung Cancer Mutation Consortium (LCMC) project, 5% of driver alterations in lung adenocarcinoma were concurrently double or multiple mutations (37). Lipson and colleagues also identified 50 alterations in 21 genes, with at least one alteration being present in 83% (20 of 24) of the lung cancers (with a range of 1–7 alterations; ref. 38). Of note, with the development of more sensitive technologies and parallel testing of multiple molecules, more concomitant alterations will be identified in a single test of a given clinical specimen. Coexistence of multiple driver mutations has been taken into consideration by oncologists to obtain an in-depth understanding of cancer mechanisms and for therapeutic developments, such as combinational targeting the molecular driver “hubs” of a cancer (39, 40). How to treat this subgroup may critically depend on the biologic roles of these onco-drivers.

Previous studies revealed that patients with EGFR/ALK co-alterations demonstrated no ALK expression by IHC (6, 21). However, in our study, IHC of serial sections showed coexpression and colocalization of mutant EGFR and ALK fusion proteins in the same cell population in all 7 evaluable patients, although staining intensities varied greatly. Our finding of colocalization was consistent with other studies of cell lines, indicating that the 2 driver alterations could develop in the same clone of tumor cells and might cooperate during cancer development (30). Clarification of the dominant driver receptor(s) is critical to understanding the disease mechanism and clinical decision makings. In our study, 4 patients with co-alterations responded only to either of an EGFR TKI or ALK TKI at different time points, suggesting that 1 of these oncogenes might act as a “dominant” driver. To address this point, phosphorylation of both EGFR and ALK was evaluated by IHC and 3 patterns could be observed: “high p-EGFR and high p-ALK,” “high p-EGFR and low p-ALK,” and “low p-EGFR and high p-ALK”. IHC data showing altered oncoproteins expression and phosphorylation were confirmed by Western blotting in 2 cases. Differential phosphorylation of EGFR or ALK might contribute to differences in sensitivity to EGFR-TKIs or crizotinib in this subgroup. Therefore, we further correlated the efficacy of TKIs with the relative activation status of these receptor kinases.

The objective response rate (80%, 8/10) and median PFS (11.2 months; 95% CI, 5.6–16.8) for first-line EGFR-TKI in EGFR/ALK co-altered tumors were similar to those in previous studies (12–17). Preclinical studies showed coexpression of altered EGFR and ALK in vitro lead to mutual resistance to single-agent ALK or EGFR TKI (6). In contrast, response to either EGFR or ALK-TKI for our patients with co-alterations was achieved. Interestingly, we found that efficacy of first-line EGFR-TKI was associated with EGFR phosphorylation level. Among the 4 cases treated with both an EGFR-TKI and an ALK-TKI, P7 and P13, with a baseline “low p-EGFR and high p-ALK” expression pattern, had de novo or subsequent resistance to EGFR-TKI treatment, but were responsive to ALK-TKI. Alternatively, P8 and P9, with a baseline “high p-EGFR and low p-ALK” expression pattern, achieved partial responses to first-line EGFR-TKI, but had progressive disease or stable disease following crizotinib treatment. Thus, the baseline relative activation of ALK and EGFR was associated with the efficacy of EGFR-TKI and ALK-TKI treatment. To our knowledge, this is the first cohort study showing diverse responses to first-line EGFR-TKIs in patients harboring EGFR/ALK co-alterations. In previous studies, 5 cases with such co-alterations were treated with EGFR-TKIs (1 with first-line gefitinib, ref. 20; 2 with second-line erlotinib, ref. 21 and 22; and 2 with unspecified-line erlotinib, ref. 6), 80% (4/5) achieved partial responses, similar to our results. A Caucasian patient with lung adenosquamous carcinoma harboring such co-alterations was reported resistant to second-line erlotinib treatment (22). No expression of ALK protein tested by IHC in 3 cases of these studies might be because of false-positive results of FISH testing (6, 21). In contrast, ALK protein was detected in our study. We suggest that a relative increase in the p-EGFR level would contribute to a favorable response to first-line EGFR-TKI in this subgroup, although whether the level of benefit of EGFR-TKI is similar to that in patients with pure EGFR mutations requires further evidence (41).

The shortcomings of our study were the small sample size concerning crizotinib treatment and p-EGFR/p-ALK testing, nonprospective design, and the fact that it was not multi-institutional. No NSCLC cell lines with EGFR/ALK co-alterations were used to model the relative activation statuses of driver receptors in relation to the efficacy of targeted therapies. Moreover, re-biopsy at serial time-points would be helpful to clarify the resistance mechanisms and dynamic changes of mutations of these driver molecules in cancer (37, 42).

In summary, the frequency of concomitant EGFR mutations and ALK rearrangements was significantly higher in ALK-rearranged NSCLCs. EGFR/ALK co-alterations could define a specific subgroup that had diverse, although mostly favorable, responses to first-line EGFR-TKIs. Testing of the relative phosphorylation levels of EGFR and ALK might help to guide the selection of TKIs in clinical practice. Molecular mechanisms underlying responsiveness and resistance to EGFR-TKIs and ALK-TKIs, and potential combination or sequential treatment modes, require further investigation in this specific subgroup with co-alterations.

T.S.K. Mok has honoraria from the speakers bureau of Roche, BI, Eli Lilly, Pfizer, and GSK. T.S.K. Mok is a consultant/advisory board member of Astrazeneca, Roche, BI, Eli Lilly, and Pfizer. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J.-J. Yang, X.-C. Zhang, T.S. Mok, Y.-L. Wu

Development of methodology: J.-J. Yang, X.-C. Zhang, H.-X. Tian, Z. Xie, T.S. Mok, Y.-L. Wu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.-J. Yang, X.-C. Zhang, C.-R. Xu, Q. Zhou, H.-X. Tian, Z. Xie, Y.-S. Huang, B.-Y. Jiang, Z. Wang, B.-C. Wang, X.-N. Yang, W.-Z. Zhong, Q. Nie, Y.-L. Wu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.-J. Yang, X.-C. Zhang, J. Su, Z. Xie, R.-Q. Liao, T.S. Mok, Y.-L. Wu

Writing, review, and/or revision of the manuscript: J.-J. Yang, X.-C. Zhang, J. Su, C.-R. Xu, H.-J. Chen, W.-Z. Zhong, R.-Q. Liao, T.S. Mok, Y.-L. Wu

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.-J. Yang, X.-C. Zhang, H.-J. Chen, Y.-L. Wu

Study supervision: T.S. Mok, Y.-L. Wu

The authors thank W.-B. Guo, S.-L. Chen, Y. Huang, and S.-J. Fei for their work on biomarker analysis. The authors also thank B. Gan, Y. Yang, X.-Y. Zheng, S.-F. Luo, X. Li, and Y.-M. Chen for their assistance in collecting clinical follow-up data. The authors also thank two professional editors for editing of this article.

This work was supported by the National Natural Science Foundation of China (grants No. 81071699, to X.-C. Zhang; No. 30772531, to Y.-L. Wu; and No. 81172090, to Q. Zhou), the Guangzhou Science and Technology Project (No. 11BppZXaa6040020, to Y.-L. Wu), and the Wu Jieping Medical Foundation Special for Tumor Targeted Therapy Research (No. 3206720.10001, to Q. Zhou).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.