EGFR L858R Mutation and Polymorphisms of Genes Related to Estrogen Biosynthesis and Metabolism in Never-Smoking Female Lung Adenocarcinoma Patients

Somatic mutations in the tyrosine kinase domain of epidermal growth factor receptor (EGFR) gene, located from exon 18 to 23, were found in lung adenocarcinoma (1–3). The most common mutations are the in-frame deletion in exon 19 and a substitution of lysine for arginine mutation at amino acid position 858 (L858R) in exon 21. The efficacy of targeted therapies such as gefitinib or erlotinib to non–small cell lung cancer (NSCLC) patients depends on the presence of EGFR somatic hotspot mutations including L858R in exon 21 or in-frame deletion in exon 19. EGFR mutations have been found to be more frequent in adenocarcinoma than other NSCLC, in females than males, and in never-smokers than smokers (4). Furthermore, there was a sex difference in EGFR mutation subtypes (5, 6). The exon 19 in-frame deletion was more frequently associated with male gender, whereas exon 21 mutations were more frequent in females. The sex difference in EGFR mutation subtypes may result from the influence of sex hormone.

Estrogen and its metabolites may play some roles in cell proliferation and genotoxicity to induce breast cancer (7), and estrogen could also potentially promote lung cancer (8). Estrogen receptors (ER) were found to express in normal human lung tissue and NSCLC tissues (9–11). EGFR promoter containing imperfect estrogen response elements (ERE) could be bound by ER (12), and the administration of 17β-estradiol was found to increase the EGFR level and tyrosine kinase activity in vivo (12, 13). These findings suggest that sex hormones play a role in lung carcinogenesis.

One cohort study has found a higher proportion of EGFR hotspot mutation in female patients having experienced a longer fertile life, measured by an early age of menarche, the delayed menopause or the late first parity (14). This finding implies the period of exposure to endogenous hormone may be correlated with the occurrence of EGFR hotspot mutations.

Many studies have used candidate genes approach to assess whether the biosynthesis and metabolism of estrogen is associated with the susceptibility to hormone-related cancer, such as breast cancer. Among these genes, the genetic polymorphisms of CYP17 rs743572, CYP19A1 tetranucleotide (TTTA)_n, COMT rs4680, and ERα rs2234693 are widely studied.

This study aimed to evaluate the associations between EGFR hotspot mutations in lung adenocarcinoma and genetic polymorphisms of CYP17, CYP19A1, ERα, and COMT.

This study included patients affected with lung adenocarcinoma, who were diagnosed between September 2002 and December 2007. This study was approved by the Joint Institutional Review Board (JIRB) at 6 teaching hospitals and tertiary referral centers in Taiwan that participated in the Genetic Epidemiological Study of Lung Adenocarcinoma (GELAC) study and IRB of Taichung Veterans General Hospital, Taiwan. Lung cancer histology was classified according to World Health Organization criteria (15). Demographic characteristics and lifestyle variables, including age, gender, and cigarette smoking status (never-smokers were defined as patients who had never smoked cigarette, whereas ever-smokers defined as those who were current or former smokers), were collected by questionnaire interview. Clinical data were abstracted using a special abstract form according to a standard protocol.

On the basis of multiple candidate genes approach in the estrogen biosynthesis and metabolism pathways, and the selection of candidate genes in this study is based on their physiological importance and absence of isoforms. To this end, 4 genes, including CYP17, CYP19A1, ERα, and COMT, were chosen. CYP17 and CYP19A1 are known to be key genes involved in the biosynthesis of estrogen (16, 17). COMT is the major gene to deactivate catechol estrogen (7). No isoforms of these 3 genes have been identified, and thus, the contribution of these 3 genes in the biosynthetic and metabolic pathways could be considered unique and independent. The ERα serves as a receptor for estrone/estradiol and facilitates the activation of downstream genes (18).

CYP17 rs743572, CYP19A1 (TTTA)_n, ERα rs2234693, and COMT rs4680 were chosen as representative polymorphisms of these genes. Details of SNPs selection were described in the Supplementary methods.

Genomic DNA from tumor-frozen specimen and paraffin-embedded tissues were extracted by QIAamp DNA Tissue Kit (Qiagen) following the manufacturer's protocols. DNA sequencing of exons 18 to 21 of EGFR gene was amplified by polymerase chain reaction, subsequently DNA sequencing reaction and subjected to electrophoresis in the ABI PRISM 3130XL as described previously (4).

The TTTA repeats of the CYP19A1 were determined by polymerase chain reaction using primers as described previously (19), and TaqMan assays were used for genotyping of the CYP17, ERα, and COMT. Details of genotyping procedures were described in the Supplementary methods and Supplementary Table S1.

The associations between EGFR mutation status and demographic and clinical characteristics, including gender, age, cigarette smoking status (never vs. ever smoker), and disease stage and study genotypes, were examined by the chi-square test. The association between EGFR mutation status and sex after adjustment for smoking status was examined by Cochran–Mantel–Haenszel Statistics. The OR with the corresponding 95% CI for each variable was estimated by the unconditional logistic regression model. Multivariate-adjusted OR (aOR) of EGFR mutation status associated with individual genotypes was assessed after adjustment for age.

Patients were classified into 4 groups by their EGFR mutation status: those without mutation in exons 18 to 21 were defined as “wild type,” those with in-frame deletion amino acid positions 746 to 753 in exon 19 as “in-frame deletion,” those with a substitution of lysine for arginine at amino acid position 858 as “L858R mutation,” and those with various mutations other than “in-frame deletion” and “L858R mutation” as “other mutations.”

The effect of genotype was initially evaluated under a codominant model in which each genotype was considered separately. However, the estimates of ORs were relatively imprecise due to the small numbers of cases. The 4 genetic polymorphisms were further classified into 2 groups by pooling the heterozygous group with either homozygous variant or wild-type groups based on the estimated OR of heterozygous genotype: The TTTA repeat in intron 4 of CYP19A1 was dichotomized as “S/L (1 short allele ≤ 7 repeats and 1 long allele > 7 repeats) and L/L (2 long alleles > 7 repeats)” vs. “S/S (2 short alleles ≤ 7 repeats).” The genotypes of ERα, rs2234693, as “C/C and T/C” vs. “T/T”; of COMT, rs4680, as “A/A and G/A” vs. “G/G,” and of CYP17, rs743572, were grouped as “C/C” vs. “T/T and T/C.” This grouping scheme has been commonly used in previous studies (20–22).

The above-mentioned 4 genotypes were further combined to examine whether patients harboring more risk alleles had higher occurrence of EGFR hotspot mutations in a dose–response relationship. The OR for each combinatory group was estimated using women who had 1 or less high-risk genotype as the referent group. The statistical significance of the biological gradient was assessed by the test for trend. Two-sided P < 0.05 was considered statistically significant. All analyses were performed using the SAS statistical software for Windows version 9.2 (SAS institute)

The clinical and demographic characteristics of 617 patients are shown in Table 1. The mean age was 62 years, and the female-to-male ratio was about 1. There were only 32% (195/617) cigarette smokers, and female patients were less likely to be smokers than males (2.9% vs. 63.3%). With regard to EGFR mutation status, there were 146 (23.7%) in-frame deletions, 139 (22.5%) L858R mutations, and 71 (11.5%) other mutations.

The associations with EGFR mutations for sex, cigarette smoking status, and disease are shown in Table 2. There were significant associations with EGFR mutation status for sex and cigarette smoking in univariate analysis. The frequency of L858R and in-frame deletion was higher in females than males (28.1% vs. 16.3% and 24.0% vs. 22.7%, respectively), and higher in never-smokers than ever-smokers (26.6% vs.13.3% and 26.8% vs. 16.9%, respectively). Sex was still significantly associated with EGFR mutation status after adjustment for smoking status (P = 0.04). There was no significant difference in EGFR mutation status among patients at different tumor stages.

To exclude the possible effect of active smoking status, further analyses were limited to never-smokers. The EGFR mutation status among never smokers stratified by sex is shown in Table 3. The percentage of wild type was similar in males (35.2%) and females (33.1%). The percentage of in-frame deletion was higher in males than females (36.1% vs. 23.5%), and the percentage of L858R mutation was higher in females than males (28.8% vs. 20.4%). Sex was significantly associated with EGFR mutation status in never-smokers (P = 0.03). The sex difference suggests that the hotspot mutations of EGFR may be related to genes involved in estrogen biosynthesis and metabolism.

Table 4 shows the associations with EGFR mutations for genetic polymorphisms of CYP17, CYP19A1, ERα, and COMT in never-smokers stratified by sex.

Genotype of CYP17 rs743572 was not significantly associated with any type of EGFR mutation in female and male never-smokers. The aOR (95% CI) of L858R mutation in female never-smokers was 1.5 (0.8–2.7) for C/C genotype vs. T/T and T/C genotypes.

The (TTTA) repeat polymorphism in the intron 4 of CYP19A1 was significantly associated with the L858R mutation in female never-smokers based on either codominant model (aOR, 2.9 with 95% CI, 1.3–6.5 for S/L genotypes vs. S/S genotype) or dominant model (aOR, 2.6 with 95% CI, 1.2–5.7 for S/L and L/L genotypes vs. S/S genotype), but not in male never-smokers. The (TTTA) repeat polymorphism was not significantly associated with in-frame deletion or other mutations in both female and male never-smokers.

With regard to the ERα rs2234693 polymorphism, the C/C genotype was significantly associated with an increased proportion of L858R mutation (aOR, 3.0 with 95% CI, 1.1–8.1), in-frame deletion (aOR, 2.9 with 95% CI, 1.1–7.6), and other mutations (aOR, 4.3 with 95% CI, 1.3–14.0) compared with the T/T genotype in female never-smokers. Based on the dominant model, the ERα rs2234693 polymorphism was still significantly associated with L858R mutation (aOR, 2.1 with 95% CI, 1.1–4.0 for T/C and C/C genotypes vs. T/T genotype) and other mutations (aOR, 2.6 with 95% CI, 1.1–6.0 for T/C and C/C genotypes vs. T/T genotype) in female never-smokers. This ERα rs2234693 polymorphism was not significantly associated with in-frame deletion in female never-smokers or any type of EGFR mutation in male never-smokers.

The COMT rs4680 genetic polymorphism was associated with the L858R mutation (aOR, 6.2 with 95% CI, 1.6–23.5 for A/A genotypes vs. G/G genotype) and other mutations (aOR, 4.7 with 95% CI, 1.1–19.6 for A/A genotypes vs. G/G genotype) in female never-smokers under the codominant model. Based on the dominant model, the COMT rs4680 genetic polymorphism was associated with the L858R mutation in both female (aOR, 1.8 with 95% CI 1.0–3.2 for G/A and A/A genotypes vs. G/G genotype) and male (aOR, 3.6 with 95% CI 1.1–11.3 for G/A and A/A genotypes vs. G/G genotype) never-smokers. This SNP was not significantly associated with in-frame deletion and other mutations in both female and male never-smokers.

Based on known interactions of individual proteins involved in the estrogen biosynthesis and metabolism pathway shown in Figure 1, we further assessed the associations with L858R mutation and in-frame deletion of EGFR gene for the combination of genotypes that were significantly associated with EGFR mutation singly. As shown in Table 5, there was a significant dose–response relationship of the number of risk alleles with L858R mutation (P = 0.0002), but not with in-frame deletion in female never-smokers or any type of EGFR mutation in male never-smokers. The aOR (95% CI) of L858R mutation increased from 2.0 (0.8–4.9), 3.2 (1.3–7.9) to 10.1 (2.6–38.6) for patients carrying 2, 3, and 4 risk alleles compared with those with 1 or less risk allele in female never-smokers.

In the subgroup analysis for never-smokers only, a significant difference was found between EGFR mutation subtypes and sex in never-smokers (P = 0.03). The percentage of L858R mutation was higher in females than males (28.8% vs. 20.4%) and the percentage of in-frame deletion was higher in males than females (36.1% vs. 23.5%), which were in agreement with previous findings (5). This sex difference implies different mechanisms for the mutagenesis EGFR in males and females.

On the basis of multiple candidate genes approach, the L858R mutation of EGFR was found to be significantly associated with genetic polymorphisms of CYP19A1, ERα, and COMT in female never-smokers based on the dominant model. In contrast, the in-frame deletion of EGFR was significantly associated with ERα polymorphism only.

Based on the gene–gene interaction shown in Table 5, the number of risk alleles of CYP17, CYP19A1, ERα, and COMT was significantly associated with an increasing OR of EGFR L858R mutation but not in-frame deletion. The aOR (95% CI) of L858R mutation in male never-smokers also increased from 2.8 (0.6–12.8) to 3.6 (0.8–15.6) for patients carrying 2 and 3 risk alleles compared with those carrying 1 or less risk allele. But the increasing trend was not statistically significant due to the small number of cases. Thus, the possibility that estrogen biosynthesis and metabolism pathway may also contribute to the occurrence of L858R in male never-smokers could not be excluded. In contrast, the in-frame deletion in exon19 was not through such a mechanism.

The difference between L858R mutation and in-frame deletion in carcinogenicity was also found. In-frame deletion has a higher oncogenic ability than that of L858R mutation (23). The mice bearing in-frame deletion has longer tumor latency of tumor formation than those with L858R (24). Tanaka and colleagues proposed that chromosomal recombination that involves DNA repair mechanisms may involve in the occurrence of the in-frame deletion in exon19 (6). The instability of the EGFR gene might link to the deletion of various regions in exon 19. The cytosine and adenine repeats in the EGFR intron 1 were reported to influence the expression of EGFR (25). It is thus hypothesized that CA repeats in intron 1 of EGFR may be associated with the in-frame deletion (unpublished data).

The occurrence of a specific mutation may involve many genes. In this study, 4 genes including CYP17, CYP19A1, ERα, and COMT were chosen. The CYP17 gene codes for cytochrome steroid 17 hydroxylase and C17,20 lyase. They catalyze the conversion of pregnenolone and progesterone to dehydroepiandrosterone (DHEAS) and androstenedione (26). The rs743572 polymorphism is located −34 in the promoter region of CYP17 with a T to C substitution, which was hypothesized to increase the CYP17 m-RNA transcriptional activity by creating a SP1 binding site (27). Patients with C/C homozygous genotype were found to have an increased level of testosterone or estrogen (28, 29). In this study, the C/C genotype was associated with a statistically nonsignificant increase in L858R mutation, and with a statistically nonsignificant decrease in in-frame deletion in both male and female never-smokers.

The CYP19A1 gene codes for an aromatase enzyme, which involves in the conversion of androstenedione and testosterone to estrone (E1) and estradiol (E2), respectively. Aromatase expression is elevated in human lung tumor cell lines as well as lung adenocarcinoma tissues (30, 31). Furthermore, an aromatase inhibitor was reported to suppress cell growth and aromatase activity in lung A549 cells in vitro (30). A (TTTA)_n repeat polymorphism in intron 4 of CYP19A1 was found to be associated with the risk of breast cancer (32) and the prognosis of patients affected by premenopausal breast cancer (20). Women with one 7-repeat allele (S allele) had lower plasma levels of estrone-to-androstenedione ratio (32). Our study found a significant association between L858R mutation and S/L and L/L genotype of CYP19A1. Mah and colleagues found that higher levels of aromatase was associated with a worse disease prognosis in female never-smokers but not male (31), which is consistent with our finding for the sex disparity. Interestingly, in-frame deletion was associated with a statistically nonsignificant association with L/L genotype of CYP19A1 in male never-smokers, which agrees with previous observations that exon 19 in-frame deletions were more frequently found in males (5, 6).

The COMT gene codes for catechol-O-methyl-transferase which involves in the methylation of catechol estrogens into inactive metabolites such as 2-methoxyestradiol (7), which was found to inhibit the proliferation of cancer cells (33). The G to A substitution in exon 4 of COMT, resulting in the substitution of valine to methionine, may decrease the enzyme activity in vitro (34, 35). Thus the subsequent accumulation of hydroxyl-catechol estrogens may generate reactive quinone and semiquinone, which may lead to site-specific oxidative DNA damage such as the binding to the N-7 of guanine to become a depurinating agent [4-OH-E1(E2)-1-(αβ)N7 Gua]. The cleavage of the glycosidic bond may result in apurinic sites (36, 37) and genetic mutation. In this study, the G/A + A/A genotypes of COMT rs4680 were associated with EGFR L858R mutation in both male and female never-smokers.

The ERα gene on chromosome 6q25.1 may act as a nuclear receptor protein with an estrogen-binding domain and a DNA-binding domain. Estrone or estradiol bound to ERα plays a role as a transcriptional factor, which binds to ERE upstream of target genes and affects the gene activity. The rs2234693 T/C (often called PvuII or IVS1–401) polymorphism is located in the intron 1 of the ERα gene. There was an interactive effect of this SNP and ER status on the survival of breast cancer patients (38). It was also found to increase the risk of myocardial infarction (39). C allele was reported to produce a myeloblastosis (myb) binding site and to increase the expression of downstream reporter gene by cotransfection of B-myb (40). Raso and colleagues found a significantly elevated expression of ERα in females, never-smokers and patients with all types of EGFR mutations (11), which are consistent with our findings that the C/C genotype of ERα rs2234693 was significantly associated with all types of EGFR mutations in female never-smokers. In contrary, there were no significant associations for males. However, the number of male never-smokers with EGFR mutations was too small (n = 70) to conclude no association in males. One study with a limited number of cases reported that ERα expression occurred more frequently in female than male lung tissues (41), another study reported that estradiol increased the phospho-serine-118-ERα in H1793 but not A549 lung adenocarcinoma cells derived from a female and male patients, respectively (42). Nevertheless, 2 studies by Stabile and colleagues (8) and Hershberger and colleagues (43) reported that lung adenocarcinoma cell lines from males could be responsive to estrogen and antiestrogen. In addition, another study (30) reported aromatase inhibitors could suppress cell growth of male lung A549 cells. Moreover, Hershberger and colleagues suggested that ERβ is sufficient to mediate both genomic and nongenomic signaling responses to estrogen, and ERβ can cooperate with EGF to promote cell proliferation (43). In addition, a significant correlation between ERβ expression and EGFR mutations was also observed (11). Thus, the mechanisms by which the ERα and ERβ genes might be related to EGFR mutations deserve further exploration.

Cytochrome p450 genes are involved in estrogen hydroxylation, such as CYP1A1. In an association study between estrogen-related genetic polymorphisms and EGFR mutation in lung cancer, the frequency distribution of CYP1A1 genotypes was different among those with different EGFR mutation status (44). But the difference was not statistically significant due to small sample size. Moreover, the CYP1A1, CYP1B1, CYP2C9, CYP3A4 are involved in estradiol hydroxylation (45), which could dilute the associations between CYP1A1 genotypes and EGFR mutation.

In this study, polymorphisms of CYP17, CYP19A1, ERα, and COMT were found to be associated with an increased occurrence of EGFR mutations in female never-smokers. There are several possible explanations for this finding. First, the robust production of estradiol and estrone due to the high expression level of CYP17 and CYP19A1 in combination with the dominant ER genotype may increase the binding of estradiol and estrone to the ER, stimulate the proliferation of cell, and increase the risk of DNA replication errors that may result in the burden of DNA repair. Meanwhile, the excessive amount of estrogen hormones might induce the production of hydroxyl-E1 or E2 through hydroxylation catalyzed by the phase I metabolism enzymes, such as the cytochrome p450. If the unstable hydroxyl-E1 or E2 were not methylated by COMT, they may lead to the formation of semi-quinone or quinone and related DNA adducts resulting in mutation, such as the formation of the L858R mutation. This hypothesis may be supported by our finding that the number of risk alleles of CYP17, CYP19A1, ERα, and COMT was significantly associated with an increasing OR of EGFR L858R mutation.

Raso and colleagues found that ERα expression was associated with EGFR mutations and a worse survival (11). Coombes and colleagues (46) reported a lower incidence of second primary lung cancer among breast cancer patients treated with exemestane after tamoxifen than those treated with tamoxifen alone. In addition to examine the associations between genetic polymorphisms and EGFR mutations in this study, it is also important to explore their associations with the survival of lung adenocarcinoma patients. Nevertheless, in our study, patients underwent surgical procedures and subsequently treated with chemotherapy rather than gefitinib. We were thus unable to assess the association between genetic polymorphisms and the responsiveness to treatment with EGFR inhibitors. Furthermore, previous studies have reported that EGFR mutations were more frequent in East Asians than in non-Asians. In addition to genetic background, environmental factors common among East Asians may contribute to the development of EGFR mutations. Both lifestyle and dietary habit, including eating animal viscera with hormone residues, using plastic bags for hot soup to increase the exposure to low level of bisphenol A, and cooking with high exposure to oil fumes, may increase the mutagenic substrates. As the information of these environmental exposures was not available in this study, the gene–environment interactions could not be assessed. In addition to the genes investigated in this study, ERβ, EGFR, and DNA repair genes mentioned above deserve further comprehensive investigation. This study was limited by the wide confidence interval of estimation, resulting from a relatively small sample size. Further multi-disciplinary prospective studies based on a large number of never-smoking patients are needed to elucidate the gene–environment interactions involved in the formation of the hotspots mutation of EGFR.

In summary, our results indicate that L858R mutation of EGFR is associated with polymorphisms of genes related to estrogen biosynthesis and metabolism in never-smoking female lung adenocarcinoma patients. The number of risk alleles of CYP17, CYP19A1, ERα, and COMT is associated with an increasing OR of EGFR L858R mutation in never-smokers, especially in females. The genetic polymorphism in ERα may be important for various EGFR mutations in female never-smokers. The findings provide a clue for the genesis of EGFR mutations.

No potential conflicts of interest were disclosed

The authors thank the Comprehensive Cancer Center of Taichung Veterans General Hospital, Taichung, Taiwan for the support of clinical data.

Funding Supported by grants (DOH96-TD-G-111-012) from the Department of Health, Executive Yuan, and Academia Sinica, Taipei, Taiwan

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.