Amplification of EGFR Wild-Type Alleles in Non–Small Cell Lung Cancer Cells Confers Acquired Resistance to Mutation-Selective EGFR Tyrosine Kinase Inhibitors

EGFR-mutated lung cancers comprise a significant subgroup of non–small cell lung cancer (NSCLC; refs. 1–3). In general, EGFR tyrosine kinase domain somatic mutations activate EGFR by promoting the active conformation of EGFR (4–7). Activated EGFR transduces signals to downstream pathways, such as the phosphoinositide 3-kinase (PI3K)–AKT and MEK–ERK mitogen-activated protein kinase (MAPK) pathways. After the identification of EGFR-activating mutations and response to first-generation EGFR tyrosine kinase inhibitors (EGFR-TKI; refs. 8–10), the treatment strategy for NSCLC patients harboring these mutations changed dramatically. The significant improvement of prognosis of lung cancer patients with EGFR mutations by first- (gefitinib and erlotinib) and second- (afatinib) generation EGFR-TKIs has been repeatedly demonstrated by multiple clinical trials (11–13). However, lung cancer cells inevitably acquire resistance to these inhibitors after approximately one year (14–16).

Multiple mechanisms of acquired resistance to first- and second-generation EGFR-TKIs have been identified to date—including the EGFR T790M gatekeeper mutation (17, 18), transformation to small cell lung cancer (SCLC; refs. 19, 20), and adaptive bypass pathway activation through the MET (21), AXL (22), and FGFR1 (23) axes. Of these, the EGFR T790M mutation is responsible for acquired resistance in approximately 50% of cases. The T790 residue is located at the entrance to a hydrophobic pocket on the posterior side of the ATP-binding cleft. As such, the T790M mutation induces a conformational change within the EGFR ATP-binding pocket, resulting in the steric hindrance of first or second-generation EGFR-TKIs (17). Coincidentally, EGFR T790M also enhances the affinity between ATP and its binding pocket (24), thereby synergistically conferring resistance to EGFR-TKIs. To address this issue, third-generation EGFR-TKIs, such as osimertinib (AZD9291; ref. 25), rociletinib (CO-1686; ref. 26), and nazartinib (EGF816; ref. 27), have been developed that irreversibly bind the EGFR ATP-binding pocket by forming a covalent bond with the C797 residue at pocket periphery, simultaneously blocking ATP binding due to the increased affinity. Interestingly, these third-generation EGFR-TKIs are also effective for the majority of EGFR-activating mutations, such as in-frame deletions in exon 19 and the L858R point mutation in exon 21 (28). Furthermore, mutation selectivity of third-generation EGFR-TKIs has been reported in preclinical and clinical models (25–27). The wide therapeutic window of third-generation EGFR-TKIs mirrors the significant efficacy and safety of these agents performed in multiple clinical trials (29, 30). Third-generation EGFR-TKIs are also subject to acquired resistance. It is important to fully clarify the mechanisms of acquired resistance to third-generation EGFR-TKIs to improve the prognosis of lung cancer patients harboring EGFR mutations. The mechanisms clarified to date include C797S or L798I mutation (31, 32), constitutive MAPK pathway activation by mutated KRAS or MEK (33), or bypass pathway activation via MET or ERBB2 (32, 34–36). However, the mechanisms underlying acquired resistance to third-generation EGFR-TKIs are not fully clarified.

In this study, we elucidated novel mechanisms underlying acquired resistance to third-generation EGFR-TKIs by whole-exome sequencing analysis. Notably, we determined that Src–AKT pathway activation and EGFR wild-type allele amplification can both contribute to EGFR-TKI acquired resistance, the latter of which might possibly be due to the decreased inhibitory pressure for EGFR wild-type by mutation selective third-generation EGFR-TKIs. Our findings highlight the importance of EGFR wild-type allele-mediated signaling in acquired resistance to mutant-selective EGFR-TKIs.

Erlotinib, afatinib, rociletinib, and wortmannin were purchased from LC Laboratories. Osimertinib was purchased from Selleck Chemicals (Houston). Nazartinib was purchased from ApexBio. Bosutinib was purchased from Cayman Chemical). Cetuximab was purchased from Keio University Hospital (Tokyo, Japan). Total EGFR antibody (#2232), EGFR E746-A750del–specific antibody (#2085), total AKT antibody (#9272), phospho-AKT (S473/D9E) antibody (#4060), total p44/42 MAPK antibody (#9102S), phospho-p44/42 MAPK (T202/204) antibody (#9101S), E-cadherin antibody (#3195S), vimentin antibody (#5741S), total Src antibody (#2108), Phospho-(Tyr416)-Src antibody (#2101S), PTEN antibody (#9559), and PI3 kinase p110 alpha antibody (#4255S) were purchased from Cell Signaling Technology). The HA-tag (ab18181) and integrin beta 1 (ab52971) antibodies were purchased from Abcam). Phospho-EGFR (Y1068) antibody (44788G) was purchased from Invitrogen/Life Technologies. Actin antibody was purchased from Sigma-Aldrich.

The PC9 (EGFR E746-A750del) and H1975 (EGFR L858R + T790M) human NSCLC cell lines were obtained from Dr. Susumu Kobayashi (Beth Israel Deaconess Medical Center, Boston, MA) and the ATCC, respectively. Cells were cultured in RPMI1640 growth medium supplemented with 10% FBS at 37°C in a humidified 5% CO₂ incubator. Cell authentication for PC9 and H1975 was performed by the authors in June 2015 using genetic profiling of polymorphic short tandem repeat (STR) loci (Takara).

MTS cell proliferation assays were performed as previously described (37). Briefly, 2 × 10³ cells/well were seeded in 96-well plates and treated with EGFR-TKI or dimethyl sulfoxide (DMSO) vehicle 24 hours later. Absorbance was measured 72 hours after treatment. All experiments were performed at least three times.

The human phospho-receptor tyrosine kinase (phospho-RTK) array kit was purchased from R&D Systems) and screened according to the manufacturer's protocol with 150 μg of protein for each experiment. Signal intensity was calculated using LumiVision Analyzer software.

H1975 cells with stable EGFR wild-type or C797S overexpression were generated by retroviral infection using MigR1-EGFR constructs that harbor a GFP as previously described (5). Briefly, Phoenix-Ampho cells were transfected with 15 μg of wild-type or C797S EGFR using Lipofectamine 2000. Viral supernatant was collected 2 days after transfection and applied to H1975 cells on retronectin-coated plates (Takara) for 24 hours. EGFR-overexpressing H1975 cells were sorted by a FACS MoFlo XDP (Beckman Coulter) based on GFP expression. Purified cells were then grown in RPMI with 10% FBS.

Cells were treated with increasing concentrations of EGFR-TKIs (0.1–1 μmol/L). Cetuximab was used at concentrations of 10 μg/mL. Cells were lysed in Cell Lysis Buffer (Cell Signaling Technology), and equivalent amounts of protein were separated by SDS-PAGE and transferred to polyvinylidene fluoride membranes. The membranes were incubated overnight with primary antibodies at 4°C and then with secondary antibodies for 1 hour. Immunoreactive proteins were visualized with LumiGLO reagent and peroxide (Cell Signaling Technology), then exposed to X-ray film.

Cells were seeded in 6-well plates (50,000/well) and were treated with rociletinib (1 μmol/L) and cetuximab (10 μg/mL) individually or in combination for 72 hours. Control cells were treated with DMSO. Apoptosis was monitored using the TACS Annexin V–FITC Apoptosis Detection Kit (R&D Systems) according to the manufacturer's protocol. The proportion of apoptotic cells was evaluated by flow cytometric analysis using a Gallios Flow Cytometer system (Beckman Coulter).

Total RNA was isolated from cultured cells using an RNeasy Mini Kit (Qiagen) and genomic DNA was isolated using a DNeasy Blood & Tissue Kit (Qiagen). RNA was subjected to reverse transcription using the High-Capacity RNA-to-cDNA Kit (Life Technologies) according to the manufacturer's protocol. Quantitative RT-PCR was performed using fluorescent SYBR Green and an ABI Prism 7000 Sequence Detection System (Life Technologies). Human GAPDH was used to normalize input cDNA. The LINE1 repetitive element was used as a reference gene for EGFR copy-number analysis. The primers used in this study are shown in Supplementary Table S1.

EGFR and chromosome 7 FISH analysis was performed by Genetic lab Co., Ltd. using Vysis LSI EGFR SpectrumOrange and CEP 7 SpectrumGreen probes, respectively.

Isolated cDNA from H1975 cells was used as template in PCR reactions for PIK3CA. Amplified PIK3CA was sequenced and compared with the National Center for Biotechnology Information reference sequence NM_006218.3.

Cells were transfected with PIK3CA-specific siRNA (#S10520; Life Technologies) or negative control siRNA (Ambion Silencer Select Negative Control mix, Life Technologies) using siLentFect transfection reagent (Bio-Rad) according to the manufacturers’ protocols. PIK3CA qRT-PCR was used to confirm gene knockdown.

All animal experiments were approved by the Laboratory Animal Center, Keio University School of Medicine (Tokyo, Japan). Female BALB/c-nu mice were purchased from Charles River. Mice were anesthetized with ketamine and PC9-COR#9 cells were subcutaneously injected in a Matrigel suspension. Calipers were used to measure tumor volume. Once average tumor volume reached 150 mm³, mice were randomized to receive vehicle alone, cetuximab (1 mg/mouse twice per week, intraperitoneally), rociletinib (30 mg/kg daily, orally), or a combination of both. Animals were humanely sacrificed and tumor tissues were harvested.

Whole-exome sequencing libraries were prepared with 3 μg of DNA. The exomes were captured using the SureSelect Human All Exon V5+UTRs Kit (Agilent Technologies) according to the manufacturer's instructions and then sequenced using a HiSeq 1500 system (Illumina) to generate 100-bp paired-end data. The whole-exome sequencing data were deposited in the DDBJ database. Accession numbers: DRA004904, PRJDB5021, and SAMD00056514-SAMD00056521

Sequence reads were aligned to the human reference genome UCSC hg19 using the Burrows-Wheeler Aligner program (BWA, http://bio-bwa.sourceforge.net/). Single-nucleotide variants (SNV) and insertions and deletions (INDEL) were called and annotated using the Genome Analysis Toolkit software package (GATK, http://www.broadinstitute.org/gatk/). Sequencing artifacts were filtered out using custom filters (GATK confidence score, ≥50; number of variant reads in each direction, ≥1; variant allele frequency, ≥10%). Known germline variants were filtered out using data from dbSNP build 131, the 1,000 Genomes Project (Phase 1 exome data, released May 21, 2011), 1 Japanese genome, and 299 in-house Japanese exomes.

CNVs in resistant cells were analyzed from exome sequence data. The log ratio of depth coverage between parental and resistant cells was calculated using the GATK Depth of Coverage tool. CNV segments were then called from the log ratio of depth of coverage using the ExomeCNV package (38).

Statistical analysis was performed using GraphPad Prism software, version 4.0 (GraphPad Software). Two-sided Student t tests were used for comparisons, with P < 0.05 regarded as statistically significant.

To clarify the mechanisms underlying acquired resistance to third-generation EGFR-TKIs, such as osimertinib (AZD9291) and rociletinib (CO-1686), we established third-generation EGFR-TKI–resistant cells using a dose-escalation method. PC9 and H1975 cells were cultured with rociletinib or osimertinib. The initial concentration was 0.03 μmol/L, and this was incrementally increased to 1 μmol/L. After several months of exposure, resistant cell lines, rociletinib-resistant PC9 (PC9-COR), osimertinib-resistant PC9 (PC9-AZDR), rociletinib-resistant H1975 (H1975-COR), and osimertinib-resistant H1975 (H1975 AZDR) were established. Their resistance to a first-(erlotinib), third-generation (rociletinib, osimertinib, or nazartinib) EGFR-TKIs was confirmed by the MTS cell proliferation assay (Fig. 1). The calculated IC₅₀ values are summarized in Supplementary Table S2. Notably, the resistant cells displayed cross-resistance to all third-generation EGFR-TKIs examined in this study, suggesting that the mechanisms of resistance to third-generation drugs may not be inhibitor specific. Further, resistant cells also showed a diminished propensity to undergo treatment-induced apoptosis (Supplementary Fig. S1), indicating that the cells acquired resistance to third-generation EGFR-TKIs.

Bypass pathway activation is consistently reported as one of the mechanisms by which cells acquire resistance to EGFR-TKIs. As such, we first examined the phosphorylation levels of multiple receptor tyrosine kinases in our resistant PC9 and H1975 cells with a human phospho-RTK array kit, but did not observe any activation of reported bypass pathways (HER2, HGFR, IGF-1R, and AXL signaling; Supplementary Fig. S2).

To clarify the heterogeneity of potential resistance mechanisms, the PC9-COR, PC9-AZDR, H1975-COR, and H1975-AZDR lines were subcloned to isolate resistant clones (Supplementary Fig. S3). DNA isolated from the parental and resistant cell clones was then subjected to whole-exome sequencing. Quality of the whole-exome sequencing results and copy number alterations are summarized in Supplementary Table S3 and Supplementary Fig. S4. Several genetic alterations potentially relevant to EGFR-TKIs sensitivity were detected (Table 1), including some already reported. For instance, osimertinib resistance can occur via oncogenic KRAS signaling (33). In our model, KRAS G13D was identified in PC9-AZDR clones (Supplementary Fig. S5), supporting the validity of our findings. EGFR C797S or L798I was not found in the resistant cells.

Parental H1975 cells harbor a heterozygous PIK3CA G118D mutation. Interestingly, the PIK3CA wild-type allele was lost in three of the resistant clones—H1975-AZDR#1 and #3, and H1975-COR#3, leading to a loss of heterozygosity (Supplementary Table S4; Supplementary Fig. S6A and S6B). The activating PIK3CA G118D mutation is found in various cancer types, including endometrium (39) and colorectal (40, 41) cancers. In order to investigate whether acquired homozygosity of PIK3CA G118D could functionally induce resistance to EGFR-TKIs, we examined the impact of PIK3CA inhibition on EGFR-TKI resistant cells using the PIK3CA inhibitor wortmannin. Although effective when used in combination with EGFR-TKI in parental H1975 cells, wortmannin was unable to restore EGFR-TKI sensitivity in the H1975-resistant clones (Supplementary Fig. S6C and S6D). Further, the siRNA-mediated knockdown of PIK3CA gene expression in the H1975-resistant clones also failed to restore EGFR-TKI sensitivity (Supplementary Fig. S6E and S6F). These data indicate that acquired homozygosity of the PIK3CA G118D mutation did not contribute to the acquired resistance to EGFR-TKI in H1975-resistant cells.

While performing these experiments, we noticed a change in cell morphology where the H1975-resistant cells exhibited a spindle cell-like morphology (Fig. 2A). Epithelial-to-mesenchymal transition (EMT) is reported to occur in EGFR-resistant NSCLC (42). To confirm whether H1975 cell clones acquired a more mesenchymal phenotype, we analyzed EMT markers by Western blotting. Consistently, expression of the mesenchymal marker vimentin and the epithelial E-cadherin was upregulated and downregulated in the resistant clones at the protein level, respectively (Fig. 2B). Moreover, both total and phosphorylated Src were elevated in H1975-resistant cells. Next, we examined the effect of EGFR-TKI treatment on EGFR downstream pathway activation, including AKT and MAPK, in H1975-resistant cells. EGFR and ERK1/2 phosphorylation were efficiently inhibited by rociletinib and osimertinib treatment; however, limited effects were observed in AKT phosphorylation in H1975-resistant cells when compared with H1975 parental counterparts (Fig. 2C). Integrin β1/Src/AKT-mediated bypass signaling has previously been reported in erlotinib-resistant NSCLC (43). Consistent with this report, we found that integrin β1 and phospho-Src were increased at the RNA and protein levels, respectively, in H1975-resistant clones (Fig. 2B and D). A loss of PTEN and subsequent AKT pathway activation facilitate EGFR-TKI resistance in mutant lung cancer (44); however, PTEN expression was intact in H1975-resistant cells (Fig. 2B and E). To examine whether Src–AKT pathway activation contributed to the acquired resistance to third-generation EGFR-TKI resistance, we performed MTS proliferation assays on H1975 clones treated with third-generation EGFR-TKIs in the presence or absence of the Src inhibitors dasatinib or bosutinib. Interestingly, both dasatinib and bosutinib partially restored sensitivity to third-generation EGFR-TKIs (Fig. 2F and G), suggesting that Src–AKT pathway activation may contribute to the acquired resistance to third-generation EGFR-TKIs.

Notably, our exome sequencing data revealed an amplification of the EGFR gene locus on chromosome 7 (Fig. 3A). To confirm this finding, chromosome 7 and the EGFR gene were analyzed by FISH and scored using the ratio EGFR signal to that of chromosome 7 (Fig. 3B). PC9 parental cells harbor EGFR gene amplification; however, the ratio of signals became higher in PC9-COR#9 clones when compared with parental counterparts (Supplementary Table S5). In addition, genomic DNA qPCR for the EGFR loci in the PC9-COR clones confirmed the amplification (Fig. 3C). Furthermore, the EGFR wild-type copy number and total EGFR expression were markedly increased in PC9-COR#9 cells (Supplementary Table S6; Fig. 3D). The relative number of tags for the wild-type EGFR allele (31 tags) was about one fourth (0.25 times) that of the mutant allele (127 tags) in PC9 parental cells; however, the wild-type allele was approximately 8.5-fold more prevalent in PC9-COR#9 cells (512 vs. 60 tags for the wild-type and mutant alleles, respectively). These data indicate that the amplified EGFR wild-type allele likely conveyed resistance to third-generation EGFR-TKIs.

We next examined EGFR ligand expression to evaluate the possibility of ligand-induced activation of EGFR (45). Until now, seven EGFR ligands have been identified: epithelial growth factor (EGF), amphiregulin, heparin-binding EGF-like growth factor (HB-EGF), epiregulin, TGFα, epigen, and betacellulin. Quantitative RT-PCR analysis revealed an increased expression of all seven ligands in PC9-COR cells and PC9-COR#9 cells compared with PC9 parental cells (Fig. 3E). As expected, the inhibition of phosphorylation of EGFR, AKT, and ERK by rociletinib was attenuated in the PC9-COR#9 clone when compared with the PC9 parental cells (Fig. 3F). These results indicated that EGFR wild-type allele amplification likely induces the acquired resistance to third-generation EGFR-TKIs through EGFR ligand-induced activation.

To ascertain the plausibility of the aforementioned situation, we expressed HA-tagged wild-type EGFR in H1975 parental cells and evaluated their sensitivity to rociletinib and osimertinib in the presence of EGF. When compared with mock-transfected H1975 cells, wild-type EGFR-overexpressing H1975 cells developed resistance to rociletinib and osimertinib upon EGF administration (Fig. 4A and B), and rescued EGF-induced EGFR, AKT, and ERK1/2 phosphorylation (Fig. 4C). Thus, these data indicate that EGFR wild-type allele amplification and concomitant EGFR ligand expression contribute to acquired resistance to third-generation EGFR-TKIs.

Concomitant EGFR wild-type amplification and ligand expression promoted us to determine whether the inhibition of EGFR ligand binding would be sufficient to restore sensitivity to rociletinib in PC9-COR#9 cells. For this, we performed MTS assays in the presence of cetuximab, an anti-EGFR antibody that blocks ligand binding. As a single agent treatment, cetuximab had no effect on cell proliferation in PC9 parental cells and PC9-COR#9 cells (Fig. 5A), nor did rociletinib/cetuximab combination therapy show a synergistic effect in PC9 parental cells and PC9-AZDR#5 cells that harbor a KRAS G13D mutation (Fig. 5B). However, combination therapy effectively restored rociletinib sensitivity in PC9-COR#9 cells (Fig. 5C). We used Western blotting to investigate whether rociletinib/cetuximab cotreatment suppressed pathway activation downstream of EGFR in PC9-COR#9 cells and found that it efficiently inhibited both AKT and ERK1/2 phosphorylation (Fig. 5D). Analysis of apoptosis by FACS revealed an increase in Annexin V–positive cells with the subsequent addition of cetuximab (12.5 % with 1 μmol/L rociletinib alone, 32.7% with rociletinib/cetuximab combination; Fig. 5E), indicating that ligand-mediated EGFR activation contributed to the acquired resistance to third-generation EGFR-TKIs. We evaluated the effect of afatinib on PC9-COR#9 cells by MTS assay because we previously reported the decreased mutation specificity of afatinib (28). The proliferation of PC9-COR#9 cells was inhibited by lower concentrations of afatinib than rociletinib or nazartinib (Fig. 5F). These findings also indicate that EGFR wild-type mediated signaling contributes to PC9-COR#9 cell resistance.

In addition, we evaluated the efficacy of rociletinib and cetuximab or afatinib cotreatment with H1975 cells expressing wild-type EGFR (H1975 EGFR) or C797S EGFR (H1975 C797S; Supplementary Fig. S7). As observed previously, wild-type EGFR overexpression induced rociletinib resistance following EGF stimulation. The addition of afatinib or cetuximab restored rociletinib sensitivity to both transduced lines. However, the restoration of afatinib was more significant in H1975 EGFR cells compared with H1975 C797S cells.

Finally, mouse xenograft models were used to examine whether the roclietinib/cetuximab combination treatment could suppress EGFR wild-type-mediated resistance in vivo. During treatment, the body weight of the mice and PC9-COR#9 cells derived tumor volumes were monitored (Fig. 6A–C). Notably, tumor volumes increased in the control and rociletinib groups; however, roclietinib/cetuximab cotreatment significantly inhibited tumor growth. Collectively, these data support the possibility that combination therapy may be an effective approach to overcome EGFR wild-type-dependent acquired resistance.

In NSCLC, EGFR mutations activate EGFR, which induce the transformation of lung epithelial cells. EGFR is ubiquitously expressed in organ epithelial cells, including lung, gastrointestinal tract, and skin. As such, it is essential to selectively inhibit mutant EGFR-derived signals to minimize adverse effects of EGFR-TKIs, such as diarrhea and skin rash. Third-generation EGFR-TKIs selectively inhibit EGFR mutants and minimally affect the wild-type EGFR (25, 26). This mirrors the significant efficacy and safety of third-generation EGFR-TKIs observed in recent clinical trials. Specifically, osimertinib and rociletinib showed significant efficacy and safety for previously treated NSCLC patients harboring EGFR-activating mutations, although rociletinib is no longer in development. The response rate and progression-free survival of osimertinib for EGFR T790M-positive patients were 61% and 9.6 months, respectively (29). The response rate of rociletinib for EGFR T790M-positive patients was 45% (46). Neither osimertinib nor rociletinib exhibited a dose-limiting toxicity; therefore, a maximum tolerated dose was not determined (29, 30). Thus, mutation-selective inhibitors offer a clinically relevant advantage of EGFR-TKI treatment with minimal adverse effect.

In this study, we identified several novel mechanisms of acquired resistance to third-generation EGFR-TKIs, including Src–AKT pathway activation and EGFR wild-type allele amplification (Fig. 6D and Supplementary Fig. S8). Most cancer-specific somatic driver mutations, such as those in KRAS and EGFR, exist in heterozygous. Until today, a limited number of studies have focused on the function of the wild-type allele in cancer cells. Here, we show that cancer cells exploit this decreased inhibitory pressure for the wild-type allele.

The mechanisms underlying acquired resistance to third-generation EGFR-TKIs have already been partially clarified. These include the C797S (26) and L798I (32) mutations that presumably prevent the covalent binding of EGFR-TKIs to EGFR, bypass pathway activation (33–35, 47), and mutant EGFR gene amplification (48, 49). Specifically, Piotrowska and colleagues reported EGFR T790M allele amplification in rociletinib-resistant clones (49). To our knowledge, this is the first report describing a role for wild-type EGFR in acquired resistance to mutant-selective third-generation EGFR-TKIs.

Recently, circulating tumor DNA in blood is becoming a study sample for clarification of mechanisms underlying acquired resistance to EGFR-TKIs (26, 32). By using recent next-generation sequencing technology, the identification of cancer cells derived genetic alterations including mutations and gene copy-number alterations become possible. Considering the convenient access to blood samples and sensitivity and specificity of circulating tumor DNA, it will become commonly used as reliable samples in the clinics. However, the contamination of nonmalignant cells derived DNA is inevitable and may limit the accuracy of gene copy-number analysis in cancer cells. Thus, to identify the EGFR wild-type allele amplification, we believe biopsies from either primary tumors or metastatic lesions are necessary.

In summary, the present study characterized novel mechanisms of acquired resistance to mutant-selective third-generation EGFR-TKIs. We propose a novel concept of acquired resistance to mutation selective inhibitors, wild-type allele mediated resistance. In addition, we propose a preclinical rationale for the use of more promiscuous EGFR-TKI, afatinib, or cetuximab combination therapy in EGFR inhibitor–resistant cancers. While these findings underscore the importance of EGFR wild-type–mediated signals in acquired resistance, additional preclinical and clinical trials are necessary to evaluate the efficacy and safety of these treatments to avoid adverse side effects related to EGFR pathway inhibition in nonmalignant cells.

K. Soejima has received speakers bureau honoraria from Chugai, Ono, Taiho, Eli Lilly, AstraZeneca, Pfizer, and Shionogi. No potential conflicts of interest were disclosed by the others.

Conception and design: S. Nukaga, H. Yasuda, K. Tsuchihara, J. Hamamoto, I. Kawada, S. Ikemura, K. Goto, K. Soejima

Development of methodology: S. Nukaga, H. Yasuda, K. Tsuchihara, J. Hamamoto, I. Kawada, S. Matsumoto, S. Ikemura, K. Soejima

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Nukaga, H. Yasuda, K. Tsuchihara, J. Hamamoto, K. Masuzawa, K. Naoki, S. Mimaki, S. Ikemura

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Nukaga, H. Yasuda, K. Tsuchihara, J. Hamamoto, K. Masuzawa, K. Naoki, S. Ikemura

Writing, review, and/or revision of the manuscript: S. Nukaga, H. Yasuda, K. Tsuchihara, I. Kawada, K. Naoki, S. Matsumoto, S. Ikemura, K. Goto, T. Betsuyaku, K. Soejima

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Yasuda, J. Hamamoto, S. Ikemura, T. Betsuyaku

Study supervision: H. Yasuda, K. Tsuchihara, I. Kawada, S. Ikemura, K. Goto, T. Betsuyaku, K. Soejima

We thank Ms. Mikiko Shibuya for her excellent technical assistance. We also thank the Collaborative Research Resources, Keio University, School of Medicine, for cell sorting.

This work was supported in part by Grants-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Culture, Sports, Science, and Technology of Japan to S. Nukaga (Grant #16K19465), T. Betsuyaku (Grant #15H04833), K. Soejima (Grant #22590870), H. Yasuda (Grant #25860656, 15H05666, and 15K14398) and the Practical Research for Innovative Cancer Control from Japan Agency for Medical Research and Development to S. Matsumoto (16ck0106012h0003).

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.