Introduction

Ten years have passed since three research groups in the USA reported the presence of epidermal growth factor receptor (EGFR) gene mutations in lung cancers (Lynch et al. 2004; Paez et al. 2004; Pao et al. 2004). The EGFR mutation is of both clinical and research interests because its presence strongly predicts the efficacy of the EGFR tyrosine kinase inhibitors (TKIs) that was applied in clinic a few years earlier. The role of EGFR mutations as a strong predictive biomarker of the response to EGFR-TKI treatment was examined in retrospective analyses (as summarized in Mitsudomi and Yatabe 2007) and finally confirmed by the biomarker analyses of the Iressa Pan-Asian Study (IPASS) trial (Fukuoka et al. 2011; Mok et al. 2009). This dramatic treatment effect by EGFR-TKIs reflects the fact that the proliferation and survival of lung cancers with the EGFR mutation solely depend on the aberrant signaling originating from this mutation. This reliance on a single gene is referred to as “oncogene addiction,” and it forms the basis for the clinically impressive results of EGFR-TKI therapy for lung cancers (Weinstein 2002). Moreover, researchers and clinicians have identified many other aberrations in proto-oncogenes leading to oncogene addiction in lung cancers. These include translocations in anaplastic lymphoma kinase (ALK), ROS1, and RET; mutations in ERBB2, BRAF, and mitogen-activated protein kinase kinase 1 (MEK 1); the amplification of MET in lung adenocarcinomas; mutations in discoidin domain receptor tyrosine kinase 2 (DDR2), phosphatidylinositol-4,5-bisphosphate 3-kinase, catalytic subunit alpha (PIK3CA), and EGFR vIII; and aberrations in fibroblast growth factor receptor (FGFR) family members in lung squamous cell carcinomas (as summarized in Suda and Mitsudomi 2014). However, lung cancers with EGFR mutations are still the largest subset of molecularly defined lung cancers, accounting for ~40 % of lung adenocarcinomas in East Asians and ~15 % of those in Caucasians and African Americans (Suda and Mitsudomi 2015).

During the past 10 years, clinicians and researchers have identified many of the characteristics and features of lung cancers with EGFR mutations. These are summarized in Table 1. However, two fundamental questions remain regarding these tumors, and they are discussed controversially among clinicians and researchers: (1) What is the prognostic impact of the EGFR mutation? (2) What is the predictive role of the EGFR mutation regarding tumor response to cytotoxic agents? In this review, we examine the answers that have emerged thus far and summarize the information from the accumulated studies on lung cancers with EGFR mutations.

Table 1 Generally accepted characteristics of lung cancers with EGFR mutations
Full size table

EGFR and its activating mutations: background

EGFR is a transmembrane receptor tyrosine kinase. Upon binding to its ligands (EGF, TGF-α, amphiregulin, etc.), EGFR forms either homodimers or heterodimers with the other ERBB family members (ERBB2, ERBB3, or ERBB4) (Hynes and Lane 2005), which stimulates intrinsic receptor tyrosine kinase activity and triggers the autophosphorylation of specific tyrosine residues within the cytoplasmic regulatory domains of these enzymes. The phosphorylated tyrosine residues activate several downstream signaling pathways, including mitogen-activated protein kinase (MAPK) pathway, phosphatidylinositol 3-kinase (PI3K)⁄AKT pathway, and the signal transducer and activator of transcription (STAT) pathway (Fig. 1). These pathways promote cell proliferation, migration and metastasis, evasion from apoptosis, and angiogenesis, all of which are associated with cancer phenotypes.

Fig. 1
figure 1

Structure and mechanism of activation of the EGFR. The EGFR protein consists of extracellular, transmembrane, tyrosine kinase, and regulatory domains. Conformational changes in the EGFR occur when a specific ligand binds to the extracellular domain, resulting in the formation of either homodimers or heterodimers with other ERBB family members (ERBB2, ERBB3, or ERBB4). During this process, the respective kinase domains dimerize asymmetrically, in a tail-to-head orientation, which stimulates the intrinsic tyrosine kinase activity of the receptors and triggers autophosphorylation of specific tyrosine residues within the cytoplasmic regulatory domains. These phosphorylated tyrosine residues serve as specific binding sites for several adaptor proteins, inducing proliferative or antiapoptotic signaling pathways, such as those of MAPK, phosphatidylinositol 3-kinase (PI3K)⁄AKT, and STAT

Full size image

EGFR mutations in lung cancers usually occur in the first four exons of the tyrosine kinase domain (exons 18–21) and induce the ligand-independent activation of EGFR. The most common EGFR mutations are exon 19 deletion mutations and exon 21 L858R point mutations, accounting for >90 % of all EGFR mutations. Lung cancers with EGFR mutations are highly responsive to EGFR-TKIs, and the type of mutation correlates with the sensitivity of the lung tumors to the various types of these inhibitors (Table 1). Accordingly, patients whose lung cancers have EGFR mutations have a significantly prolonged overall survival (OS) after treatment with EGFR-TKIs compared with patients whose lung tumors are negative for EGFR mutations (Takano et al. 2008).

Prognostic and predictive biomarkers

Prior to the discussion of the prognostic impact of EGFR mutations and the predictive role of these mutations for responsiveness to cytotoxic agents, the terms “prognostic” and “predictive” should be appropriately defined. Stated simply, a predictive biomarker identifies patients who will or will not respond effectively to a certain drug, while a prognostic biomarker identifies patients who have a favorable or poor prognosis irrespective of treatment. It is often difficult to determine whether a biomarker is predictive or prognostic. For example, if a subgroup of patients with biomarker A lives longer than a control group after treatment with drug B, it is not clear whether biomarker A is a predictive biomarker for drug B or if biomarker A merely defines those patients with a favorable prognosis.

The interrelationship between EGFR mutations and clinicopathological prognostic/predictive factors

An additional difficulty in precisely discussing the prognostic or predictive roles of EGFR mutations is the confounding relationship between these mutations and other prognostic/predictive factors (or candidates thereof). An example is smoking status. Smoking status is a well-known poor prognostic factor and probably a poor predictive factor in the treatment for lung cancers (Cuyun Carter et al. 2014). Because the incidence of EGFR mutations in lung cancers is inversely correlated with smoking status (Table 1), the influence of smoking must be excluded in analyses of the prognostic or predictive roles of EGFR mutations. Similarly, sex, which is closely related to smoking status, especially in Asian countries, is another example of a clinical prognostic and predictive factor (Cuyun Carter et al. 2014) that correlates with the presence of EGFR mutations.

Histological subtypes, as defined by the International Association for the Study of Lung Cancer (IASLC)/American Thoracic Society (ATS)/European Respiratory Society (ERS) classification, also influence analyses of the prognostic role of EGFR mutations. Several retrospective studies found that the prognosis of patients with surgically resected lung adenocarcinomas with solid or micropapillary subtypes is poorer than that of patients with lepidic, papillary, or acinar tumor subtypes (as discussed in Suda et al. 2014). The relationship between the presence of EGFR mutations and histological subtypes has been intensively investigated by several groups (Chen et al. 2014; Hu et al. 2014; Jie et al. 2014; Nakamura et al. 2014; Shim et al. 2011; Song et al. 2013; Sun et al. 2012, 2014; Yanagawa et al. 2014; Yoshizawa et al. 2013). Their findings are summarized in Fig. 2, which shows that the percentages of histological subtypes of lung cancers with EGFR mutations are different from those of lung cancers without these mutations.

Fig. 2
figure 2

Correlation between EGFR mutation status and histological subtypes according to the IASLC/ATS/ERS lung adenocarcinoma classification (East Asian population). Histological subtypes of lung adenocarcinomas with EGFR mutations mostly include those with a better prognosis ones, such as lepidic-, papillary-, and acinar-predominant adenocarcinomas, and fewer predominantly solid adenocarcinomas and invasive mucinous adenocarcinomas. Data are from (Chen et al. 2014; Hu et al. 2014; Jie et al. 2014; Nakamura et al. 2014; Shim et al. 2011; Song et al. 2013; Sun et al. 2012, 2014; Yanagawa et al. 2014; Yoshizawa et al. 2013)

Full size image

Thus, in interpreting results regarding the prognostic or predictive role of EGFR mutation status, the influence of these and possibly other closely related factors must be kept in mind.

Prognostic role of EGFR mutations

Many studies have assessed the prognostic role of EGFR mutations in lung cancers, with conflicting results (D’Angelo et al. 2012; Izar et al. 2013; Janjigian et al. 2011; Kim et al. 2013; Kobayashi et al. 2008; Kosaka et al. 2009; Lim et al. 2007; Lin et al. 2014; Liu et al. 2010, 2014; Marks et al. 2008; Sonobe et al. 2007). Most have focused on OS, and some included patients treated with EGFR-TKIs. However, in studies that evaluated OS, the true prognostic significance of the EGFR mutation could not be determined because, as described above, it is virtually impossible to distinguish whether EGFR mutation status is an inherent prognostic factor or a predictive factor of treatment.

The true prognostic implication of a biomarker can be evaluated through a comparison of patient groups, classified by the biomarker, without any treatment (i.e., following the so-called natural history of the disease). Since this type of analysis is ethically unacceptable in cases of malignant disease, the most reasonable method to evaluate the prognostic implication of a certain biomarker is to compare the recurrence-free survival (RFS) rates of patients who have undergone complete tumor resection, preferably without postsurgical adjuvant chemotherapy (Suda et al. 2012b).

One such study was recently carried out in Taiwan. Lin et al. analyzed the RFS rates of 163 patients with pathological stage I, surgically resected lung adenocarcinoma. The tumor size was <2 cm in its maximal dimension, and none of the patients had received adjuvant chemotherapy. The results showed that the presence of an EGFR mutation was not associated with RFS (p = 0.286), while elevated preoperative serum carcinoembryonic antigen levels, the presence of visceral pleural surface invasion, histological differentiation, and a TP53 mutation were significant risk factors for relapse (Lin et al. 2014). Similar results have been reported from Japan (Kobayashi et al. 2008), Korea (Kim et al. 2013), and China (Liu et al. 2014), although some of the included patients had received adjuvant chemotherapy. These studies suggested that the EGFR mutation is frequently associated with other positive prognostic factors, non- or mild-smoking status, female sex, earlier disease stage, and a lepidic tumor pattern (indicative of less invasiveness) and is not an independent prognostic factor.

Conflicting results were obtained by Izar et al. (2013) in their retrospective analysis of 307 patients with completely resected stage I non-small-cell lung cancers (NSCLCs) not treated with adjuvant therapy. In that cohort, Caucasian patients accounted for >90 % of the cases. The authors found that the median RFS in the group negative for EGFR mutations was 7.0 years compared with 8.83 years in the EGFR-mutant-positive group (p = 0.0085). In a multivariate analysis, the presence of the EGFR mutation [hazard ratio (HR) 0.326, p = 0.026] and tumor size (HR 1.37, p = 0.04) was a significant prognostic factor. The discordance between this study and those described above can be explained by the fact that Izar et al. included patients with non-adenocarcinoma NSCLCs in their analyses. However, it may also possible that the prognostic implication of the EGFR mutation differs between different ethnic and racial groups.

Predictive molecular biomarkers for cytotoxic agents in lung cancers

Before discussing the predictive role of the EGFR mutation for tumor responsiveness to cytotoxic agents, we summarize what is known about well-known predictive molecular biomarkers or their candidates (Table 2). Although none of these biomarkers has been generally accepted, some are being evaluated in clinical trials. The rationale for studying molecular biomarkers is that they may be involved in resistance or, conversely, confer sensitivity to certain cytotoxic agents, as described below.

Table 2 Examples of molecular predictive biomarkers for chemotherapeutic agents (Ceppi et al. 2006b; Christoph et al. 2013; Filipits & Pirker 2011; Giovannetti et al. 2005; Mizuuchi et al. 2015; Olaussen et al. 2006; Postel-Vinay et al. 2012; Scagliotti et al. 2008; Vilmar & Sorensen 2011)
Full size table

Because many cytotoxic agents kill cancer cells via DNA damage, DNA repair genes and specifically their expression levels are candidate predictive biomarkers for the efficacy of cytotoxic agents. The high-level expression of DNA repair genes such as the excision repair cross-complementation group 1 (ERCC1), breast cancer 1 (BRCA1), and the MutS homologue 2 (MSH2) has been suggested to protect cancer cells from cytotoxic agents that induce DNA damage, such as those that are platinum-based, resulting in resistance to these drugs. [Note that the currently available anti-ERCC1 antibodies did not specifically detect the unique functional ERCC1 isoform (Friboulet et al. 2013)]. The expression levels of the target genes of cytotoxic agents are a second group of candidate predictive biomarkers. Higher expression of a target gene implies the need for higher concentrations of cytotoxic agents, which may lead to drug resistance. The third group of candidate biomarkers are drug-degrading enzymes (again, specifically their expression levels), such as dihydropyrimidine dehydrogenase (DPD), which reduces the effect of 5-fluorouracil (5-FU). Membrane transporters or drug efflux pumps can increase the sensitivity or resistance, respectively, to cytotoxic agents that are substrates of these molecules. For example, folate receptor alpha (FRA) transports folates and antifolates into cells; therefore, its higher expression may induce the greater efficacy of pemetrexed (antifolates). In the case of drug efflux pumps such as ABC transporters, their high-level expression leads to lower intracellular concentrations of their substrate drugs and, accordingly, resistance to them. The predictive biomarkers commonly used to assess the efficacy of cytotoxic agents in lung cancer treatment are summarized in Table 2 (Ceppi et al. 2006a; Christoph et al. 2013; Filipits and Pirker 2011; Giovannetti et al. 2005; Mizuuchi et al. 2015; Mochinaga et al. 2014; Olaussen et al. 2006; Postel-Vinay et al. 2012; Scagliotti et al. 2008; Vilmar and Sorensen 2011).

Predictive roles of EGFR mutations in response to cytotoxic agents

EGFR mutation is a strong predictive biomarker for the response to treatment with EGFR-TKIs; however, can it also predict sensitivity to cytotoxic chemotherapy? The EGFR mutation does not fall into any of the categories of the molecular predictive biomarkers described above, such as DNA repair genes, target genes of cytotoxic agents, and drug efflux pumps. However, correlations between the presence of the EGFR mutation and molecular predictive biomarkers for the response to cytotoxic agents have been determined through analyses of clinical specimens. As summarized in Table 3, although in some cases the results are controversial, a negative correlation between ERCC1 expression and the presence of the EGFR mutation has been repeatedly reported. [Again, note that the currently available anti-ERCC1 antibodies did not specifically detect the unique functional ERCC1 isoform.] While the mechanism that links the one to the other is unclear, lung cancers with EGFR mutations may be more responsive to platinum-doublet chemotherapy which is currently the gold standard chemotherapeutic regimen for the treatment for NSCLC.

Table 3 Correlation between EGFR mutations and molecular predictive biomarkers for cytotoxic agents
Full size table

The predictive role of the EGFR mutation has also been evaluated retrospectively, using the data of lung cancer patients who were treated with frontline cytotoxic chemotherapies (Dong et al. 2013; Fang et al. 2014; Hotta et al. 2007; Kalikaki et al. 2010). Most of these studies reported that among lung cancer patients with EGFR mutations who were treated with platinum-doublet chemotherapies, progression-free survival (PFS) was better for patients with EGFR-mutation-positive tumors than for those with tumors negative for the mutation. In multivariate analyses adjusting for other predictive factors, such as disease stage, smoking status, sex, and tumor histology, the EGFR mutation was an independent predictive biomarker for responsiveness to platinum-doublet chemotherapy. Some of these studies were relatively large and included >200 patients (Dong et al. 2013; Fang et al. 2014).

Nonetheless, there are limitations in evaluating the predictive role of EGFR mutations retrospectively. For example, the platinum-doublet chemotherapy regimens differed, and some patients were treated with a reduced dose. In addition, there were differences in the follow-up schedule; more frequent follow-ups would be better able to detect disease progression, which would be reflected in shorter PFS. There may also have been publication bias.

The most reliable data to analyze the predictive implication of EGFR mutations for responsiveness to cytotoxic agents are the subgroup analyses of two randomized phase III trials: the above-mentioned IPASS (Mok et al. 2009) and First-SIGNAL (first-line single-agent iressa versus gemcitabine and cisplatin trial in never smokers with adenocarcinoma of the lung) (Han et al. 2012). Both enrolled patients in East Asia with previously untreated lung adenocarcinoma who had never smoked or were former light smokers (IPASS) or never smokers (First-SIGNAL); these criteria increased the number of lung cancer patients with EGFR mutations. Patients were randomly assigned to gefitinib or platinum-doublet chemotherapy (carboplatin–paclitaxel in IPASS and cisplatin–gemcitabine in First-SIGNAL). EGFR mutations were analyzed in tissue samples, and subgroup analyses were performed based on the EGFR mutation status. In the IPASS trial, 214 patients were assigned to the chemotherapy arm and had data on EGFR mutation status compared to 43 in the First-SIGNAL trial. In the two trials, the clinical backgrounds of the patient with/without EGFR mutation were similar, due to the inclusion criteria. The objective response rates (ORRs) in the IPASS trial of patients with and without EGFR mutations who were treated with carboplatin–paclitaxel were 47 and 24 %, respectively. The disease control rates (DCRs) were 88 and 84 %, and PFS was 6.3 and 5.5 months, respectively (Table 4). The difference in PFS was not statistically significant [HR 0.78; 95 % confidence interval (CI) 0.57–1.06; p = 0.11]. In the First-SIGNAL trial, similar ORRs were reported in patients with and without EGFR mutations who were treated with cisplatin–gemcitabine (38 vs. 52 %, respectively; p = 0.36). The PFS of the two subgroups was the same (6.3 vs. 6.4 months, respectively; HR 0.679; 95 % CI 0.34–1.35; p = 0.27).

Table 4 Summary of PFS in prospective studies comparing EGFR-TKIs with platinum-doublet chemotherapies (Han et al. 2012; Maemondo et al. 2010; Mitsudomi et al. 2010; Mok et al. 2009; Rosell et al. 2012; Sequist et al. 2013; Wu et al. 2013; Wu et al. 2014; Zhou et al. 2011)
Full size table

Based on these reports, it can be concluded that lung cancers with EGFR mutation respond better to frontline platinum-doublet chemotherapy. However, the predictive impact of the EGFR mutation is quite small if clinical and pathological factors are strictly aligned.

What is the appropriate cytotoxic agent for lung cancers with EGFR mutations?

Is there a specific cytotoxic agent with higher or lower efficacy in the treatment for lung cancers with EGFR mutations? For the treatment for NSCLCs, platinum (cisplatin or carboplatin) combined with so-called third-generation cytotoxic agents (such as paclitaxel, docetaxel, gemcitabine, and vinorelbine) was the standard treatment before the era of pemetrexed, because the treatment effects of these regimes were similar (Kelly et al. 2001; Ohe et al. 2007; Scagliotti et al. 2002; Schiller et al. 2002). However, in a phase III study that evaluated the efficacy of pemetrexed, a multitargeted antifolate, a preplanned subset analysis showed that cisplatin plus pemetrexed was better than cisplatin plus gemcitabine in the OS of patients with tumors of non-squamous histology (Scagliotti et al. 2008). (For patients with squamous cell carcinoma, cisplatin plus gemcitabine resulted in a longer OS than cisplatin plus pemetrexed.) The molecular basis of this result has been attributed to the lower expression levels of thymidylate synthase (TS), the main target of pemetrexed, in tumors with a non-squamous histology than in squamous cell carcinoma (Ceppi et al. 2006b).

Because almost all lung cancers with EGFR mutations are adenocarcinomas, pemetrexed-based chemotherapy seems to be the most effective regimen. Further support for the use of pemetrexed in this setting comes from the finding of lower TS mRNA expression in lung adenocarcinomas positive for EGFR mutations or with an ALK translocation than in tumors that are negative for either mutation (Table 3). In addition, in a retrospective analysis, patients with lung cancers expressing an EGFR mutation had a better pemetrexed response rate (p = 0.016) and longer PFS (p = 0.030) than those whose tumors expressed the wild-type EGFR (Wu et al. 2011). Prospective evidence was provided by a phase II study (n = 51) that analyzed the efficacy of frontline carboplatin plus pemetrexed in patients with non-squamous NSCLCs. Median PFS was longer in patients with EGFR-mutation-positive disease (7.9 months) than in those whose tumors lacked EGFR mutations (6.3 months), although this difference was not statistically significant (p = 0.09) (Kim et al. 2012).

Table 4 summarizes the results of prospective phase III studies that compared EGFR-TKIs with platinum-doublet chemotherapies. The chemotherapy regimens included carboplatin–paclitaxel, cisplatin–gemcitabine, cisplatin–docetaxel, carboplatin–gemcitabine, and cisplatin–pemetrexed. PFS was 4.6–6.9 months. Although PFS rates determined in different chemotherapy trials cannot be compared directly, the combination of cisplatin and pemetrexed resulted in the numerically longest PFSs. In addition, in the preplanned integrated analysis of the Lux-Lung 3 and Lux-Lung 6 studies, cisplatin plus pemetrexed yielded longer PFS than cisplatin plus gemcitabine in patients with EGFR-mutation-positive lung cancers, whereas the PFS achieved with afatinib was roughly the same between these two trials (Yang et al. 2015). These results support the choice of regimens containing pemetrexed in the treatment for lung cancers with EGFR mutations.

A retrospective analysis reported that among patients with lung adenocarcinoma who underwent curative pulmonary resection followed by uracil–tegafur adjuvant chemotherapy, those whose tumors were negative for EGFR mutations had a significantly prolonged survival, whereas this was not the case in patients with EGFR-mutation-positive tumors (Suehisa et al. 2007). Although the findings of basic research also support this observation (Mochinaga et al. 2014; Suehisa et al. 2007), there is not enough evidence supporting the avoidance of adjuvant 5-FU-based agents in the treatment for lung cancers with EGFR mutations in clinical practices.

Treatment strategy for lung cancers with EGFR mutations

In clinical practice, however, fewer patients with EGFR-mutation-positive lung cancers receive platinum-doublet chemotherapy as a frontline therapy, because in all seven phase III trials, the PFS of patients who received EGFR-TKI treatment (gefitinib, erlotinib, or afatinib) was superior to that of platinum-doublet chemotherapy (Table 4) (Maemondo et al. 2010; Mitsudomi et al. 2010; Rosell et al. 2012; Wu et al. 2013; Wu et al. 2014; Yang et al. 2012; Zhou et al. 2011). However, the emergence of acquired resistance to frontline EGFR-TKI is almost inevitable, such that the median PFS is only in the range of 8.4–13.7 months (Table 4). The disease progression patterns include involvement of the central nervous system (CNS)-only progression, oligo-progression, and systemic progression (Fig. 3). Treatment strategies are usually based on the disease progression pattern, but cytotoxic agents play important roles in all three after the development of resistance to frontline EGFR-TKI. For example, in the NEJ002 trial that compared gefitinib with carboplatin/paclitaxel in the first-line setting, 68 patients (88 %) of the 77 patients who had stopped receiving frontline gefitinib received cytotoxic chemotherapy (Inoue et al. 2013). [However, in a population-based study, only 46 % of patients received cytotoxic chemotherapy after EGFR-TKI treatment failure (Mariano et al. 2014)]. In the next section, we consider whether the chemosensitivity of lung cancers with EGFR mutation is altered after they acquire resistance to EGFR-TKIs.

Fig. 3
figure 3

Patterns of disease progression after a good response to EGFR-TKIs. Disease progression can be classified into three patterns, CNS-only progression, oligo-progression, and systemic progression. Representative treatment strategies, including a role for cytotoxic agents, are also described. (Modified from slides presented by Gandara et al. at the American Society of Clinical Oncology meeting of 2013)

Full size image

Mechanisms of acquired resistance to EGFR-TKIs

Before discussing chemosensitivity after the acquisition of resistance to EGFR-TKIs, it is necessary to briefly review the resistance mechanisms to EGFR-TKIs (Fig. 4) (Suda et al. 2012a). Acquisition of the T790M gatekeeper mutation of the EGFR (as a secondary mutation), which substitutes a methionine for a threonine at amino acid position 790 (Kobayashi et al. 2005; Pao et al. 2005), is the most common mechanism of acquired resistance to the reversible EGFR-TKIs (gefitinib and erlotinib), accounting for 68–83 % according to the high-sensitivity method of detection (Fig. 4b) (Arcila et al. 2011; Su et al. 2012). Initially, the larger methionine residue was thought to sterically block the binding of gefitinib or erlotinib; however, a later study demonstrated the increased ATP affinity of EGFR with a T790M mutation as the mechanism of resistance (reviewed in Suda et al. 2009). Afatinib is an irreversible EGFR-TKI that was expected to overcome acquired resistance by T790M secondary mutations, based on their ability to bind the EGFR irreversibly (Engelman et al. 2007a; Li et al. 2008). However, this drug is also active against the wild-type EGFR (Engelman et al. 2007a; Li et al. 2008), leading to dose limitation and thus a lower clinically achievable concentration. Therefore, a T790M secondary mutation is also considered as an acquired resistance mechanism to afatinib.

Fig. 4
figure 4

Summary of the mechanisms of acquired resistance to EGFR-TKIs in lung cancers with EGFR mutations. Lung cancers with EGFR mutation are highly responsive to EGFR-TKIs (a). However, acquired resistance occurs almost inevitably by: a secondary mutation in the receptor that interferes with the binding of its inhibitor (b), activation of bypass signaling (c), activation of EGFR downstream signaling (d), or other mechanisms (e). Resistance mechanisms thought to induce an altered chemosensitivity are highlighted in red

Full size image

The activation of bypass signaling is another mechanism by which resistance to EGFR-TKIs is acquired. Despite the effective inhibition of EGFR by EGFR-TKI, cancer cells survive due to bypass signaling, that is, by activating other receptor tyrosine kinases. The molecular basis is either MET gene amplification (Bean et al. 2007; Engelman et al. 2007b) or the high-level expression of the ligand (HGF) (Yano et al. 2008), ERBB2 amplification, AXL amplification, and insulin-like growth factor 1 receptor activation (Fig. 4c). The third method of acquired resistance is the activation of downstream molecules of EGFR (Fig. 4d). The candidate molecules are PTEN (down-regulation), CRKL (amplification), ERK (reactivation of signaling by amplification or down-regulation of negative regulators of ERK signaling), and BRAF (mutation). Other acquired resistance mechanisms (Fig. 4e) thus far include epithelial-to-mesenchymal transition (EMT), stem-cell-like changes, small-cell lung cancer (SCLC) transformation, and loss of the mutant EGFR allele (Ercan et al. 2012; Ohashi et al. 2012; Sequist et al. 2011; Shien et al. 2013; Suda et al. 2012a; Tabara et al. 2012; Takezawa et al. 2012; Zhang et al. 2012). Analyses of clinical specimens suggest that these molecular mechanisms of resistance are usually mutually exclusive (Sequist et al. 2011; Suda et al. 2010; Yu et al. 2013).

The altered chemosensitivity of lung cancers with EGFR mutations after acquired resistance to EGFR-TKIs

Convincing data regarding the chemosensitivity of lung cancers with EGFR mutations before and after the acquisition of resistance to EGFR-TKIs have come from in vitro models of acquired resistance. These have been established by chronically exposing sensitive cultured cells to EGFR-TKIs, which gradually induces inhibitor resistance (Engelman et al. 2007b; Suda et al. 2011; Turke et al. 2010). These in vitro isogenic models have been used to analyze the molecular mechanism underlying the chemosensitivity of cells with acquired resistance. A summary of the results of these analyses, which include our own, is that cells with acquired resistance show the same chemosensitivity as the EGFR-TKI-sensitive parental cells if resistance was acquired by specific mechanisms that involve a single proto-oncogene, such as an EGFR T790M mutation, MET amplification, or IGF-1R activation. The exception appears to be PTEN down-regulation, which causes acquired resistance to both EGFR-TKIs and cisplatin (Suda and Mitsudomi 2013) and therefore would result in insensitivity to platinum-doublet chemotherapy. Other in vitro isogenic models of acquired resistance to EGFR-TKI, with EMT or stem-cell-like features, show altered chemosensitivity toward antimicrotubule agents (taxanes and vinorelbine). The overexpression of ABCB1, which encodes a drug efflux pump, is the apparent molecular basis of this cross-resistance (Mizuuchi et al. 2015; Shien et al. 2013). Thus, because “specific resistance mechanisms” are the main cause of acquired resistance to EGFR-TKIs (Yu et al. 2013), the chemosensitivity to cytotoxic agents does not usually change after the acquisition of resistance to these inhibitors in lung cancers with EGFR mutations. Clinical evidence supports this hypothesis. In the above-mentioned NEJ002 study, the protocol recommended a crossover regimen as second-line treatment, although any treatment was permitted. Among the 77 patients who stopped receiving gefitinib, 52 (67.5 %) received carboplatin–paclitaxel as the second-line treatment. The ORR of these patients was 28.8 %, comparable to the 30.7 % determined for patients in the frontline carboplatin–paclitaxel arm (Maemondo et al. 2010). Tseng et al. retrospectively analyzed the data of patients with lung adenocarcinomas with EGFR mutations who had been treated with pemetrexed plus platinum as the frontline therapy and as second-line therapy after EGFR-TKI treatment failure. The ORRs of patients without versus with prior EGFR-TKI were similar (38.6 vs. 24.6 %), as was the DCR (65.9 vs. 62.3 %) and PFS (6.1 vs. 6.1 months) (Tseng et al. 2014).

SCLC transformation is a rare mechanism of acquired resistance to EGFR-TKIs. Clinical cases in which the lung cancer responded to cytotoxic chemotherapy for classical SCLCs after the acquisition of resistance to EGFR-TKI have been reported (Sequist et al. 2011).

Conclusions

This review provided a summary of the clinical and therapeutic features of lung cancers with EGFR mutations, focusing on the prognostic roles and predictive implications for cytotoxic chemotherapies. Lung cancers with EGFR mutations define a subset of those with a better prognosis, probably due to the accompanying clinical characteristics. EGFR-TKIs are the treatment of choice for lung cancers with EGFR mutations, with cytotoxic chemotherapy serving as the second-line treatment. In vitro and clinical evidence suggest that in many cases, the chemosensitivity of lung cancers with EGFR mutations does not change after the acquisition of resistance to EGFR-TKIs.