PharmGKB Summary: Very Important Pharmacogene Information for Epidermal Growth Factor Receptor (EGFR)

Introduction

Epidermal growth factor receptor (EGFR) encodes a transmembrane glycoprotein. This protein is a member of the protein kinase superfamily, which consists of EGFR (ErbB1/HER1), HER2 (ErbB2), HER3 (ErbB3) and HER4 (ErbB4). All family members contain an extracellular ligand-binding domain, a single membrane-spanning region, a juxtamembrane nuclear localization signal, and a cytoplasmic tyrosine kinase domain. They are collectively called HER receptors and are ubiquitously expressed in various cell types, primarily in those of epithelial, mesenchymal and neuronal origin. Under homeostatic conditions, receptor activation is tightly regulated by the availability of ligands, which together form the epidermal growth factor (EGF) family [1]. From those ligands, EGF, transforming growth factor alpha and amphiregulin bind specifically to EGFR [2]. Binding of the EGFR or other family members to a ligand induces receptor dimerization and tyrosine autophosphorylation and leads to cell proliferation.

The EGFR involvement in carcinogenesis has been well established and mutations in EGFR can be utilized as predictive markers in the treatment of cancer. In this review, we will focus on the effects of genetic variants (somatic and germline) in the treatment of non-small cell lung cancer (NSCLC) with tyrosine kinase inhibitors (TKI) and will provide a pharmacogenomics overview of EGFR in humans. This Very Important Pharmacogene (VIP) summary is available with interactive links to gene variants and drugs on the PharmGKB website http://pharmgkb.org/gene/PA7360 [3].

EGFR Gene, Molecular Structure and Function

EGFR maps on chromosome 7p11.2; it covers 188.3 kb, from 55,086,725 to 55,275,031, on the positive strand. EGFR is composed of 28 exons and encodes a protein of 1210 amino acids (ENST00000275493, Ensembl v69) [4]. Multiple alternatively spliced transcript variants that encode different protein isoforms have been found [5].

EGRF activation by binding of growth factor leads to the autophosphorylation of the intracellular tyrosine kinase domain and results in the formation of receptor homodimers or heterodimers with other HER family members. The phosphorylated tyrosine residues act as a docking site for various adapter molecules, and this results in the activation of downstream signaling pathways including Ras/Raf/MEK/ERK and PI3K/Akt [6, 7], driving different biological processes including cell cycle progression and differentiation, increased cell invasiveness, apoptosis and angiogenesis [8, 9]. Thus, overexpression of EGFR is believed to have a critical role in tumor progression [8–10].

The principal cause of cancer-related mortality is lung cancer, and non-small cell lung cancer (NSCLC) constitutes almost 80% of all lung cases. NSCLC arises from lung epithelial cells, and comprises diverse histological subtypes including adenocarcinoma, bronchioloalveolar, squamous, anaplastic and large-cell carcinomas. About half of the NSCLC patients manifest advanced disease at the time of diagnosis, thus making treatment difficult [11]. Various oncogenic mechanisms, including EGFR gene mutations, increased EGFR copy number and EGFR protein overexpression may impair the regulation of tyrosine kinase activity of EGFR in tumor cells [12, 13] and may result in increased malignant cell survival, proliferation, invasion and metastasis [14]. The present procedure is that patients with specific types and stages of cancer are treated according to standardized, predetermined protocols [15]. However, understanding the molecular genesis of NSCLC, along with advances in the field of pharmacogenomics, can lead to targeted therapies.

EGFR as cancer drug target

EGFR has been linked to the growth of many human epithelial malignancies, including NSCLC, metastatic colorectal cancer (CRC), head and neck squamous-cell carcinoma (HNSCC), and pancreatic cancer [10, 16, 17]. Intensive laboratory and clinical research have facilitated development of EGFR inhibitors. There are two main types of EGFR inhibitors: tyrosine kinase inhibitors and monoclonal antibodies against EGFR (http://pharmgkb.org/pathway/PA162356267).

Tyrosine Kinase Inhibitors (TKIs)

TKIs are synthetic molecules that block ligand-induced receptor autophosphorylation by binding to the ATP-binding pocket of the intracellular tyrosine kinase domain and disrupting tyrosine kinase activity, thus eliminating intracellular downstream signaling [6, 7]. Gefitinib and erlotinib are specific for EGFR (HER1), whereas afatinib, lapatinib and neratinib inhibit both EGFR and HER2; pelitinib and dacomitinib inhibit EGFR, HER2 and HER4; and vandetanib inhibits EGFR, vascular endothelial growth factor receptor (VEGFR) and the RET-tyrosine kinases [16].

The FDA approved gefitinib through an accelerated process in May 2003 as monotherapy for the treatment of advanced NSCLC patients after failure of both platinum-based and docetaxel chemotherapies. As a condition of accelerated approval, the FDA required demonstration of a survival benefit in a subsequent clinical trial. Three large, prospective studies showed no improvement in overall survival [18–20]; therefore, the original FDA approval for gefitinib was modified. Currently gefitinib is indicated as monotherapy for the continued treatment of advanced NSCLC patients who are benefiting from or who have benefited from gefitinib after failure of both platinum-based and docetaxel chemotherapies [15, 16, 21].

In Europe, gefitinib is not approved for the treatment of patients with locally advanced or metastatic NSCLC unless they also harbor EGFR mutations. In November 2004, erlotinib monotherapy was approved by the FDA for the treatment of advanced NSCLC patients after failure of prior chemotherapy regimen. The FDA also approved erlotinib in combination with gemcitabine for advanced pancreatic cancer patients who have not received previous chemotherapy [15, 16, 21, 22].

Previously, treatment outcomes of erlotinib or gefitinib were studied in unselected patients, which led to conflicting results depending on the type of patient population enrolled in each study. However, the discovery that response to erlotinib or gefitinib is associated with the presence of activating somatic EGFR mutations in NSCLC has led to the design of clinical trials in which patients were selected on the basis of EGFR mutational status [15, 23]. This pharmacogenetic approach and its results are discussed below.

Other TKIs (lapatinib, neratinib, pelitinib and vandetanib) either have been approved or are in clinical trial phases for cancers other than NSCLC [16]. Several clinical trials are continuing for afatinib, and preliminary result of one of these trials [24] will be discussed in the context of treatment of advanced NSCLC harboring activating somatic EGFR mutations.

Monoclonal antibodies

Cetuximab and panitumumab are monoclonal antibodies that specifically target the extracellular domain of EGFR. Cetuximab functions by blocking endogenous ligand binding to the extracellular domain of EGFR and enhancing receptor internalization and degradation [25, 26]. Cetuximab and panitumumab were approved for the treatment of patients, other than NSCLC, with EGFR-expressing metastatic CRC refractory to chemotherapy [27–29]. Cetuximab was also approved for the treatment of advanced HNSCC in combination with radiation therapy [30, 31]. Since cetuximab and panitumumab block the extracellular domain of EGFR, not the TK domain, activating mutations might not affect treatment outcome.

Genetic variation of EGFR: Somatic mutations & germline SNPs

Somatic Mutations

The COSMIC database provides information about somatic mutations and related details of human cancers (release v62, date of access: 12/03/2012, http://www.sanger.ac.uk/cosmic) [32].

Out of 68,986 unique samples deposited in the COSMIC database for EGFR (partial or full sequence and genotype data) including all cancers examined, 13,201 (19.1%) samples have somatic mutations and about 1.3% of all samples have more than one mutation. There are 842 unique location entries for somatic EGFR mutations. Among the mutation bearing patients, six of the mutations have a frequency ≥ 1% and five have 0.1%–1% frequency. The remaining somatic mutations are spread out along EGFR, and are mostly missense substitutions, but there are also insertions and deletions. The six common somatic mutations (≥1%) constitute ~93% of all mutations and are in the tyrosine kinase (TK) domain (between amino acids 712 and 968, exon 18–24) of EGFR (Table 1). The most common set of mutations is in Exon 19 (codon 729–761); it is not a simple mutation, but rather, a collection of different deletions and a few missense substitutions concentrated in codons 744–753 of exon 19. The most frequent mutation in this group is the E746_A750del mutation. Exon 19 mutations comprise 48.3% of all mutations.

Table 1.

Incidence of the Common (≥1%) and Rare (0.1%–1%) Somatic Mutations of EGFR.

Mutation	Incidence
Exon 19 mutations
Collection of different deletions and a few missense substitutions; the most frequent one is E746_A750del	48.3%
L858R: rs121434568
Other missense mutations were also observed in this codon at extremely low frequency	36.2%
T790M: rs121434569	3.8%
Exon 20 mutations
A group of different insertions concentrated in codons 763–774	2.3%
Codon 719 mutation
Mutation at the first base of codon G719S or G719C (rs28929495), 0.82%
Mutation at the second base of codon G719A or G719D, 0.77%	1.6%
L861Q: rs121913444
L861R and L861V mutations were also observed on this codon in extremely low frequency	~1.0%
A289V, G598V, E709K, S768I and L833V: combined incidence	~1.0%

Incidences are derived from COSMIC database. 68,986 unique samples have been deposited for EGFR, of which 69% are from lung cancer tissues.

The second most common mutation is L858R (rs121434568, T-to-G change at the middle base of the codon) which comprises 36.2% of all mutations. Other missense mutations also have been observed in this codon as monoallelic or biallelic mutation combinations (L858K, L858M, L858Q, L858R and L858L) in small numbers of subjects. The third most common mutation, T790M (rs121434569) comprises 3.8% of mutations. Exon 20 mutations, a group of different insertions concentrated in codons 763–774, comprise 2.3% of mutations. Mutations at codon 719, including rs28929495 (G719S/G719C), mutations G719A or G719D, and others comprise 1.6% of mutations. L861Q (rs121913444) comprises ~1% of mutations. L861R and L861V mutations are also observed. Five relatively rare mutations (0.1%–0.3% individual frequency, totaling together ~1% of mutations) are A289V, G598V, E709K, S768I and L833V (Table 1).

Mutations in the TK domain of EGFR (exon 18–21) also appear in non-lung cancers, including 38 different cancer tissues. EGFR mutations in the TK domain are observed in 7.4% of lung cancer samples and 1–2% of salivary gland, eye, peritoneum, upper aerodigestive tract, adrenal gland, and thyroid cancer tissues. Of 39 cancer tissues in COSMIC, 22 of them have EGFR mutations ranging from rare (0.1% in pancreas, hematopoietic and stomach tissues) to moderately frequent (7.4% in lung).

In a recent study, whole exome and genome sequences of 183 lung adenocarcinoma tumor/normal DNA pairs revealed EGFR mutations in 17.5% of patients, with a few patients having more than one mutation. The L858R (rs121434568) and exon 19 deletions constituted half of the EGFR mutations [33]. In contrast, whole exome sequencing of 31 NSCLC revealed a L858R mutation in only one patient (3.2%) [34]. Several somatic mutations were also observed in genes other than EGFR [33, 34].

L858R (rs121434568) and exon 19 deletions:

EGFR mutations that lead to increased response to epidermal growth factors are called activating mutations; these mutations produce a significant and persistent activation of intracellular signaling pathways, resulting in increased cell proliferation. On the other hand, lower concentrations of TKIs are required to inhibit TK phosphorylation, because the mutant receptor has reduced ATP affinity that accounts for increased sensitivity to drugs as compared with wild type EGFR [8, 35, 36]. EGFR kinase domain mutations that are clustered around the ATP-binding pocket of the enzyme (exon 19 mutations, L858R, G719X (G719C, G719S and G719A) and L861Q) increase the kinase activity of EGFR and are activating mutations [35, 36]. There are many rare mutations in this region whose functions have not been determined. The L858R and exon 19 mutations constitute between 85% and 91% of all mutations; therefore many studies use these two mutations in their analysis [32, 37, 38].

Clinical responses to both erlotinib and gefitinib differ among NSCLC patients; approximately 10% of patients had clinical responses when treated with TKIs [18–20, 39, 40]. Sequencing of the EGFR in tumor samples from these responders showed somatic gain-of-function (e.g. activating) mutations, and these findings led to new clinical trials or retrospective analysis in which patients were chosen depending on the activating EGFR mutational status.

In retrospective or prospective studies; patients with an activating somatic EGFR mutation had significantly increased response rate (RR)[41–45] and longer progression-free survival (PFS) [42–49] time when treated with erlotinib or gefitinib compared to patients who have no somatic mutation. Although none of the prospective studies reported a statistical overall survival (OS) advantage, a few relatively small studies suggested that OS was increased in mutation-harboring East Asian NSCLC patients when treated with gefitinib [45, 48, 49]. EGFR mutations were present in most cases of NSCLC patients who responded well to TKIs, yet approximately 10–20% of patients who show a partial response to gefitinib do not have identifiable EGFR mutations, indicating that EGFR mutations are not the sole determinants of TKI response [21]. Most studies present their results according to activating somatic EGFR mutation status, regardless of the mutation type, but all or a majority of these mutations are either rs121434568 (L858R) or exon 19 deletion(s).

In prospective phase III randomized trials comparing TKIs and chemotherapy as first-line therapy in patients with advanced NSCLC who harboured activating EGFR mutations, erlotinib [50, 51] gefitinib [52, 53] and afatinib [24] treatment arms had significantly increased RR and longer PFS time, whereas OS did not show any clinical benefits when compared to standard chemotherapy [50, 52, 53]. Similarly, in a phase III trial where previously untreated East Asian NSCLC patients who were nonsmokers/former light smokers were treated with gefinitib or carboplatin/paclitaxel (IPASS trial), activating mutation-harboring patients treated with gefitinib had significantly longer PFS time compared to the carboplatin/paclitaxel group; on the contrary, the EGFR mutation negative group had significantly shorter PFS time when treated with gefinitib [54, 55]. OS did not differ between two treatments arm (gefitinb vs. carboplatin/paclitaxel)[55].

Many studies analyzed L858R and exon 19 mutations together to increase power. A few studies compared the clinical outcomes for common L858R and exon 19 mutations and failed to show any differential benefits [55–58] between the two types of mutations, except for two studies which suggested that the exon 19 deletions group had longer PFS [59, 60] and OS [60] compared to L858R in TKI-treated NSCLC patients.

Demographic differences of the incidence of EGFR mutations in NSCLC patients were observed. Activating mutations in EGFR are more frequent in women (38% vs. 10% in men), nonsmokers (47% vs. 7% in smokers), adenocarcinomas (30% vs. 2% in non-adenocarcinoma) and Asian populations (26–36% vs. 7–12% in Whites) [32, 36, 38, 61]. EGFR mutations in all NSCLC patients (whether smokers or not) will be important, as inhibition of this receptor has considerable clinical benefits [61]. This observation was particularly clear in Asian patients with EGFR mutations treated with gefitinib in the IPASS trial [54].

T790M (rs121434569): Acquired resistance to TKIs:

The majority of patients with an activating EGFR mutation received clinical benefits when treated with erlotinib/gefitinib, but the magnitude and the duration of the clinical response vary significantly and patients experience disease progression within 9–12 months of treatment [15, 23]. Mutation type (L858R vs. exon 19 del) seems to have little effect on the clinical outcome [55–58].

The most frequent mechanism of acquired resistance to TKIs is the T790M (rs121434569) mutation [62–65]. This mutation may reduce the binding capability of TKIs to the TK domain of EGFR by an allosteric mechanism [63] and increase the affinity to ATP, so that a much higher concentration of TKIs is required to inhibit EGFR [66]. The T790M mutation was originally thought to be acquired by tumors cells during treatment with TKIs; however, when more sensitive methods were used for mutation detection, the presence of the T790M mutation was shown in a small fraction of tumors cells before treatment with TKIs and usually co-exists in these cells with other activating mutations [67, 68]. The tumor cell clones carrying both the activating and the T790M mutations will eventually develop resistance to the TKIs and will be responsible for the progression or recurrence of the disease; this hypothesis was confirmed by findings that patients harboring T790M mutation before the start of the treatment had a significantly shorter PFS [69–71] and decreased response rate (RR) [68] compared with those not having T790M mutation.

Rosell et al. [70] assessed the T790M mutation in pretreatment diagnostic specimens from 129 EGFR TKI treated advanced NSCLC patients with EGFR mutations, and found that the EGFR T790M mutation was present in 45 of 129 patients (35%). PFS was 12 months in patients with and 18 months in patients without the T790M mutation (P = 0.05). Additionally, it was found that low BRCA-1 levels neutralized the negative effect of the T790M mutation and were associated with longer PFS with erlotinib treatment, whereas high levels of BRCA-1 may lead to de novo resistance through increased DNA damage repair capacity, suggesting that pretreatment assessment of both T790M mutation status and BRCA1 expression could be useful to predict outcome [70]. In the EURTAC trial the T790M mutation was detected in 38% of the pre-treatment specimens analyzed [51]. Fujita et al. [72] evaluated the incidence of T790M in pretreatment tumor specimens using the highly sensitive colony hybridization technique, and this mutation was detected in 30/38 (79% of) resected tumor tissues of patients with EGFR mutations. The median time to treatment failure (TTF) was 9 months for the patients with pretreatment T790M and 7 months for the patients without the T790M mutation (p = 0.44), which suggested that patients with the T790M allele may have a relatively favorable prognosis [72].

The T790M (rs121434569) mutation, along with other secondary mutations in EGFR, was observed as a germline mutation in four siblings of European descent family in which multiple members developed NSCLC [73]. The T790M mutation was not observed in a cohort of ~400 subjects [73] and dbSNP (build 137) does not show its existence in the general population.

Other resistance mechanisms to EGFR-targeted therapy

The T790M mutation is detectable in about 50% of patients with NSCLC patients who develop resistance to TKI treatment [64, 65, 67], and may not explain all resistance cases. One of the mechanisms of resistance to TKIs involves the mesenchymal epithelial transition factor gene (MET) amplification that occurs in 5–20% of patients [74]. MET gene amplification leads to EGFR-independent activation of the PI3K/AKT pathway through MET/ErbB-3 heterodimers and may be responsible for the resistance to TKIs [74]. MET amplification in NSCLC was identified in a very small proportion of tumor cells even before exposure to TKIs, and this population of cells expands following TKI treatment [75]. The other mechanism involves the KRAS oncogene, which is mutated in approximately 15–30% of NSCLC [76]. Mutations in KRAS and those in EGFR seem to be mutually exclusive, and KRAS mutation-harboring patients do not respond to TKI therapy [77]. KRAS is a downstream mediator of EGFR-induced cell signaling, and mutations confer constitutive activation of the signaling pathway(s), independent of EGFR activation [78].

BIM (known as BCL2-like 11) is also involved in resistance to TKI treatment and is a proapoptotic protein that is overexpressed in different malignancies [79, 80]. Various chemotherapeutic agents use BIM as a mediating executioner of cell death. Hence, BIM suppression supports metastasis and chemoresistance. BIM upregulation is required for apoptosis induction by EGFR tyrosine kinase inhibitors (EGFR-TKIs) in EGFR-mutant NSCLC. Low BIM mRNA levels can lead to gefitinib resistance in NSCLC with EGFR mutations and may be a marker of primary resistance. The extracellular regulated kinase (ERK) pathway also negatively regulates BIM expression in NSCLC with EGFR mutations [81–83]. Components that cause induction of BIM may have a role to overcome resistance to EGFR TKI in NSCLC with EGFR mutations. Recent studies have showed that HDAC inhibition can epigenetically restore BIM function in vitro and death sensitivity of EGFR-TKI, in cases of EGFR mutant NSCLC where resistance to EGFR-TKI is associated with a common BIM polymorphism [84]. Thus, NSCLC has a significant level of plasticity, being able to activate several different mechanisms leading to resistance to EGFR-TKIs.

EGFR gene copy number and protein expression in NSCLC

Mutations, gene copy number, and protein expression are three EGFR-related biomarkers that have been extensively studied in clinical trials in order to obtain better predictive and prognostic values for treatment modalities. Although EGFR gene amplification frequently correlates with EGFR protein overexpression and tumor progression [85], EGFR gene amplification and protein overexpression studies yielded controversial results in terms of prognostic significance and clinical benefits [15, 23, 47]. In this regard, biomarkers associated with EGFR-TKI activity was evaluated in the IPASS trial (>1200 NSCLC patients) [55]. EGFR mutations were the strongest predictive biomarker for PFS and objective RR to first-line gefitinib versus carboplatin/paclitaxel treatment, and post hoc analysis suggested that the predictive value of EGFR gene copy number was driven by coexisting EGFR mutations [55].

Germline SNPs

EGFR contains over 800 SNPs found in ≥ 1% of samples (dbSNP build 137) and a few of them may have some biological importance.

Intron 1 (CA)n repeat (rs11568315):

This is a simple sequence repeat polymorphism; dinucleotides range from 9 to 23, with the majority clustered around 15 to 21 CA repeats (dbSNP build 137). Association of the intron 1 CA repeat polymorphism to better clinical response in NSCLC patients treated with gefitinib was analyzed in four different studies, all having less than 100 study subjects [86–89]. Alleles with 16 or fewer CA repeats were grouped together and considered to be short alleles, and those with 17 or more CA repeats were considered to be long. NSCLC patients carrying one or two short alleles are more likely to have better clinical response (increased RR, increased PFS and increased OS) when treated with gefitinib as compared to patients who have two long alleles [86–89]. Well-powered studies are needed to replicate the beneficial clinical effect of rs11568315 in NSCLC patients.

The -216G>T (rs712829):

GT and TT genotypes are associated with increased PFS time when treated with gefitinib in NSCLC patients [86] and with decreased severity of diarrhea when treated with erlotinib in neoplasm patients [90] as compared to patients with GG genotypes. Both studies involved few patients, and well-powered studies are needed to replicate suggested associations [86, 90].

GWAS studies on tumor risk and EGFR:

Two well-powered genome wide association studies (GWAS) showed that SNPs in EGFR were significantly associated with risk of glioma, the most common primary brain tumor [91, 92]; implications of these finding on treatment of glioma or other cancers are yet to be seen.

Conclusion

Extensive molecular, cancer genome sequencing and recent GWAS studies demonstrate that EGFR is an important gene for many biological processes and particularly tumorigenesis. NSCLC patients harboring activating somatic EGFR mutations have better clinical outcomes when treated with TKIs compared to the patients who do not have mutations. Clinical and treatment associations with germline EGFR SNPs are not strong; more studies are necessary to clarify the role of germline SNPs in treatment of NSCLC.