Prognostic and Therapeutic Implications of Aromatase Expression in Lung Adenocarcinomas with EGFR Mutations

Worldwide, lung cancer is the leading cause of cancer death in males, and the second leading cause of cancer death in females (1). Although tobacco smoking is the predominant risk factor for lung cancer, approximately 25% of lung cancer cases are not attributable to tobacco use (2). The proportion of never-smokers among patients with non–small cell lung cancer (NSCLC) has significantly increased for decades. NSCLC in never-smokers is more frequent in females and the adenocarcinoma cell type, and has a better prognosis compared with NSCLC in ever smokers (3, 4). Furthermore, frequencies of oncogenic drivers, such as mutations in KRAS or epidermal growth factor receptor (EGFR), or echinoderm microtubule–associated protein-like 4 (EML4)–anaplastic lymphoma kinase (ALK) fusion are different between lung cancers in never-smokers and those in smokers (5–7). These striking differences in epidemiologic, clinical, and molecular characteristics suggest that lung cancers associated with smoking and those unassociated with smoking are separate entities (2, 3).

The higher proportion of females among patients with lung cancer who have never smoked suggests a possible role for gender-dependent hormones in lung cancer development. Estrogen reportedly affects differentiation and maturation of the normal lungs (8) and stimulates lung tumor growth in both laboratory-based (9–12) and clinical studies (13–16). Epidemiologic studies also have suggested that endogenous and exogenous estrogen affect development of lung cancer (17). A post hoc analysis of a randomized controlled trial in postmenopausal women showed that hormone replacement therapy (HRT) may increase the risk of death from lung cancer (18). A prospective cohort study confirmed dose-dependent increase in lung cancer risk among women who received HRT (19). A decreased incidence of lung cancer was observed in patients with breast cancer treated with an aromatase inhibitor, exemestane, after tamoxifen therapy compared with patients who continued tamoxifen therapy (20). These data strongly support an important effect of female hormones in lung cancer development.

Aromatase (CYP19A1) is a cytochrome P-450 enzyme that converts androstenedione and testosterone to estrone and estradiol, respectively, and supports breast and endometrial cancer growth via autocrine and paracrine stimulation (21, 22). In not only gonadal tissue, but also in lung tissue, estrogen is synthesized mainly by aromatase (12, 23). Treatment with aromatase inhibitor has been found to suppress growth in lung cancer cell lines and mice tumor xenografts (12). In early-stage lung cancer, high aromatase expression reportedly correlates with poor prognosis in women ages ≥65 years (13). These studies suggest that aromatase affects lung cancer development, although the precise pathway is unclear.

After somatic EGFR mutations were discovered in NSCLC, many studies found higher mutation frequencies in East Asians, women, never-smokers, and adenocarcinomas (5, 6). The prevalence of EGFR mutation among female patients implies interactions between female hormones and EGFR mutations. Interactions between estrogen receptor (ER) and EGFR pathways have been extensively investigated in vitro (10, 11, 24, 25) and in tumor specimens (14, 15, 26, 27), but the influence of aromatase on EGFR-dependent growth is not clear.

We studied expression and prognostic significance of aromatase, with regard to EGFR mutation status, in patients with primary lung adenocarcinoma. We also examined growth inhibition by aromatase inhibitor combined with EGFR tyrosine kinase inhibitor (TKI) on lung cancer cell lines with EGFR mutations.

Primary tumor and corresponding nonneoplastic lung specimens were collected from 110 consecutive patients who underwent complete resections (R0) for primary lung adenocarcinoma from April 2007 to March 2011 at the Department of Surgery and Science, Kyushu University Hospital (Fukuoka, Japan), for whom surgical specimens were available and EGFR mutation status were determined (Table 1). This study included 44 men and 66 women, with a mean age of 67.7 years (range: 37–85 years) at surgical resection. Almost all of the women were postmenopausal. Histologic tumor diagnoses were based on hematoxylin and eosin-stained preparations, using the WHO 2004 classification (28). Pathologic staging was performed according to the 7th edition of the TNM Classification of Malignant Tumors (29). EGFR mutation tests used the peptide nucleic acid–locked nucleic acid (PNA-LNA; Mitsubishi Chemical Medience, Tokyo, Japan) polymerase chain reaction (PCR) clamp method (30) with formalin-fixed paraffin-embedded sections of surgical specimens. No patient was treated with chemotherapy or radiotherapy before surgery. Thirty-nine (35.5%) patients received postoperative chemotherapy: 21 received oral tegafur and uracil, 17 were enrolled into a clinical trial for the postoperative adjuvant chemotherapy (S-1 or cisplatin-S-1), and 1 received paclitaxel. A routine check-up with a physical examination, blood cell counts, serum chemistry, serum tumor markers including carcinoembryonic antigen and cytokeratin fragment 19, and chest X-rays were performed on an outpatient basis 4 times a year for the first 3 years, and thereafter twice annually. Computed tomography was performed twice a year for the first 3 years, and thereafter at least annually. Brain magnetic resonance imaging, and bone scintigram or fluorodeoxyglucose positron-emission tomography were performed annually. This study was approved by the Kyushu University Institutional Review Board for Clinical Research (no. 24–173).

Tumor samples and corresponding nonneoplastic lung tissues (most distant from tumor) were obtained immediately after resection, frozen in liquid nitrogen, and stored at −80°C.

We obtained 21 lung adenocarcinoma cell lines and the breast cancer line MCF-7. A549, LK87, PC-9, and 11-18 cell lines were the kind gift of Dr. M. Takeshita. HCC4006 cell line was the kind gift of Dr. A.F. Gazdar, and was confirmed by identification of the rare EGFR deletion mutation (del L747_E749, A750P) in this cell line (31). The ACC-LC-319 cell line was a kind gift from Dr. T. Hida. Total RNAs from other cell lines were extracted in previous analyses (32, 33) or were the kind gifts of Dr. K. Tomizawa and Dr. T. Mitsudomi.

Driver mutations of the cell lines were KRAS mutations: A549, ACC-LC-94, H23, H358, H2009, LK87, and SK-LU1; EGFR mutations: H3255, HCC827, HCC4006, PC-9, and 11–18; MET mutation: H596; MET amplifications: ACC-LC-319 and H1993; Ros fusion: HCC78; HER2 mutation: H1781; EML4/ALK fusion, H2228; unknown: HCC193, SK-LC-3, and VMRC-LCD.

Cells were maintained in RPMI 1640 medium (Life Technologies) containing 10% fetal bovine serum (Life Technologies), 100 IU/mL penicillin, and 100 μg/mL streptomycin. The cells were maintained in a humidified atmosphere of 5% CO₂ in air at 37 °C.

EGFR-TKI erlotinib and aromatase inhibitor exemestane were purchased from Selleck Chemicals and LKT Laboratories, respectively.

The aromatase mRNA expression levels were evaluated by quantitative RT-PCR. Total RNA was extracted from resected lung tissues and cell lines using ISOGEN (Nippon Gene) according to the manufacturer's protocol. cDNA was synthesized using a SuperScript III First-Strand Synthesis SuperMix (Invitrogen) according to the manufacturer's protocol. Quantitative PCR amplification was performed using Applied Biosystems StepOnePlus real-time PCR system (Life Technologies). TaqMan gene expression assay (Applied Biosystems) for CYP19A1 (Hs00903413_m1) was used and β-actin (Hs99999903_m1) was used as an internal control. The BD qPCR total RNA human reference (Clontech Laboratories, Inc.), corresponding to a standardized mixture of total RNAs from a collection of adult human tissues, was used as a standard for quantitation. Relative aromatase mRNA expression levels of each sample (tissue and cell line) were standardized to those of β-actin and calculated relative to that of the total RNA human reference. Each sample was tested with triplicate measurements, and the mean value of the triplicate measurements was defined as a final value. We divided patients with adenocarcinoma into 2 groups based on the expression level of aromatase compared with human reference; high aromatase expression was defined as being above the human reference expression, whereas low expression was defined as being below it.

HCC4006, 11-18, H358, H2228, and ACC-LC-319 cells (5 × 10³) were plated into each well of 96-well flat-bottomed plates and grown in phenol red-free RPMI 1640 (Life Technologies) containing 10% dextran-coated charcoal-stripped fetal bovine serum (Biological Industries). Twenty-four hours later, dimethyl sulfoxide (DMSO), erlotinib, exemestane, or a combination of these drugs was added to achieve the indicated drug concentration, and cells were incubated for an additional 72 hours. The viability of drug-treated cells was determined by a WST-8 method using Cell Count Reagent SF (Nacalai Tesque) according to the manufacturer's instructions. Percent growth was determined relative to DMSO-treated controls.

Statistical analysis was performed using JMP statistical software version 9.0.2 (SAS Institute Inc.). All variables are expressed as the mean ± standard deviation (SD). Qualitative variables were compared using χ² tests, and quantitative variables were compared using Wilcoxon tests. Multivariate models were constructed using logistic regression, including sex, smoking history, and EGFR mutation status, with aromatase expression (high/low) as the outcome of interest. Survival curves were drawn using the Kaplan–Meier method. Significant differences among subgroups were compared using the log-rank test. The Cox proportional hazard regression model was used to explore the effects of the clinicopathologic variables and aromatase expression on survival. Factors showing prognostic significance in the univariate analyses were adopted as variables in multivariate analysis. P < 0.05 was considered statistically significant.

We first examined the mRNA expression level of aromatase in lung adenocarcinoma and corresponding nonneoplastic lung tissues, using quantitative RT-PCR. Relative aromatase mRNA expression in carcinoma tissues (0.83 ± 1.06) was significantly higher than in corresponding nonneoplastic lung tissues (0.55 ± 0.46; P = 0.025; Fig. 1). Aromatase mRNA expression in nonneoplastic lung tissues did not significantly differ among subgroups divided by age (<70 vs. ≥70), sex, smoking history, or EGFR mutation status (P = 0.07, 0.58, 0.46, and 0.61, respectively).

In univariate analysis, no significant correlation was identified between tumor aromatase expression level and clinicopathologic factors that are associated with smoking-unrelated lung cancer, such as sex, smoking history, and EGFR mutation status (Table 1). The same results were observed for other clinicopathologic characteristics; age, maximum standardized uptake value (SUV_max), tumor size, histologic grade, pleural invasion, lymphatic invasion, vascular invasion, and pathologic stage. In addition, in multivariate analysis, no significant association between high aromatase expression and sex, smoking history, or EGFR mutation status was identified (Supplementary Table S1). We also examined tumor aromatase expression as a continuous variable. However, any statistically significant correlation was still not found between aromatase expression level and clinicopathologic factors, although tendencies were seen in vascular invasion (P = 0.06) and pathologic stage (I vs. ≥II, P = 0.051).

A survival analysis was performed in 110 patients who underwent curative resections. Median follow-up time was 35 months (range: 9–66 months). High expression of aromatase was associated with poor prognosis in terms of both recurrence-free survival (RFS; P = 0.004; Fig. 2A) and overall survival (OS; P = 0.003; Fig. 2B).

Cox regression analyses of clinical variables for RFS and OS are shown in Table 2. Among the variables, sex, aromatase expression, and pathologic stage were identified as potential predictors of RFS. A multivariate analysis that included the above variables also showed aromatase expression to be a significant prognostic factor, with a relative risk of 2.37 [95% confidence interval (CI), 1.05–5.31; P = 0.039] for RFS. Multivariate analysis for OS was not performed because of the small number of events (deaths).

Next, we compared survival between subgroups divided by clinicopathologic factors that are related to smoking-unrelated lung cancer such as sex, smoking history, and EGFR mutation status. High aromatase expression was associated with a poor prognosis in females (P = 0.008 for RFS and P < 0.001 for OS; Fig. 2C and D), in never-smokers (P = 0.009 for RFS and P < 0.001 for OS; Fig. 2E and F), and in patients with EGFR mutations (P = 0.005 for RFS and P = 0.003 for OS; Fig. 2G and H), but not in males (P = 0.14 for RFS and P = 0.65 for OS; Supplementary Fig. S1A and S1B), not in current or former smokers (P = 0.16 for RFS and P = 0.58 for OS; Supplementary Fig. S1C and S1D) and not in patients without EGFR mutations (P = 0.19 for RFS and P = 0.07 for OS; Supplementary Fig. S1E and S1F). Eleven patients with EGFR mutations had recurrent disease, and among them 8 patients received EGFR-TKI gefitinib after recurrence. There was no patient who received EGFR-TKI before recurrence.

Cox regression analyses for potential predictors of survival in patients with EGFR mutations are shown in Table 3. Among the parameters, aromatase expression and pathologic stage were identified as potential predictors of RFS. Multivariate analysis was not performed because of the small number of recurrences.

We examined aromatase mRNA expression in 21 human lung adenocarcinoma cell lines (Fig. 3A). No correlation was demonstrated between aromatase expression level and driver mutation type. Because patients with high aromatase expression had worse prognoses than those with low aromatase expression among those with EGFR mutations (Fig. 2G and H), we next investigated whether aromatase had therapeutic potential in lung adenocarcinomas with EGFR mutations. We chose 11-18 as high-aromatase mRNA-expressing cell line and HCC4006 as low-aromatase mRNA-expressing cell line, both of which have EGFR mutations. To test the growth inhibitory effects of the aromatase inhibitor exemestane, we conducted an MTT assay. HCC4006 was not sensitive to exemestane, either alone or combined with erlotinib (Fig. 3B). By contrast, 11-18 was sensitive to exemestane alone (Fig. 3C), and its cell growth was significantly inhibited by the combination of exemestane with erlotinib. We further tested the growth inhibitory effects of exemestane in high-aromatase mRNA-expressing cell lines without EGFR mutations: H358 (Supplementary Fig. S2A), H2228 (Supplementary Fig. S2B), and ACC-LC-319 (Supplementary Fig. S2C). The antitumor effect of exemestane in H358, H2228, and ACC-LC-319 was much weaker than in 11-18.

Although increasing evidence indicates that female hormones affect development of lung cancer (34–36), to our knowledge, this study is the first report to elucidate the prognostic significance of aromatase expression in patients with lung adenocarcinomas with EGFR mutations. We found that aromatase mRNA expression level was not correlated with clinicopathologic factors, including EGFR mutation status. However, high aromatase expression was associated with poor prognosis in terms of both RFS and OS. Moreover, the prognostic significance of aromatase expression was limited to females, never-smokers, and patients with EGFR mutations, whereas such significance was not observed in their counterparts.

Aromatase is an enzyme that catalyzes the conversion from androgens to estrogens. In NSCLC cells, estrogen is reported to be mainly produced by intrinsic aromatase (23), and stimulates the ER signaling pathway, resulting in tumor development and progression (9–11, 13, 24). Here, we found that aromatase mRNA expression levels in carcinoma tissues were significantly higher than in corresponding nonneoplastic lung tissues. Niikawa and colleagues reported that the estradiol concentration in NSCLC was significantly higher than that in the nonneoplastic lung tissues, and intratumoral estradiol concentration in NSCLC was positively associated with aromatase mRNA expression (23). In another study, levels of aromatase activity tested by radioassay were significantly greater in tumors compared with those in nearby normal tissue (12). These studies indicate that intrinsic aromatase expression levels are closely associated with the estrogen levels in the lung cancer cells. Thus, increased aromatase level may have profound influence in carcinoma tissues through estrogen function.

Most estrogenic actions are mediated by ER, which exists in 2 forms, ERα and ERβ (37). Although immunohistochemical expressions of ERα and/or ERβ has been associated with clinical outcome in some studies (14–16, 26, 38), the findings for expression frequency and subcellular localization (nuclear or cytoplasm) of ERs are inconsistent (14–16, 27, 36, 38, 39). These differences could be because of variation in such factors as (a) antibodies and dilutions, (b) scoring systems for staining, and (c) patient cohort characteristics (16). This discrepancy may obscure the significance of hormone receptor expression in patients' clinicopathologic characteristics or prognoses. For this reason, we found it difficult to clarify the effect of estrogen on the development of lung cancer using immunohistochemical analysis.

In this study, no significant correlation was identified between tumor aromatase expression and clinicopathologic factors, including EGFR mutations, when analyzed as either dichotomized or continuous variables. These results are consistent with previously reports (27, 39, 40). However, a correlation between EGFR mutation and ER expression, both ERα (14) and nuclear ERβ (15), in lung adenocarcinoma, was reported in previous studies, suggesting that some interaction between ER and the EGFR signaling pathway may exist.

High aromatase expression was associated with a poor prognosis in patients who underwent curative resections for lung adenocarcinoma. We also demonstrated that the prognostic significance of aromatase expression was limited to females, never-smokers, and patients with EGFR mutations, whereas such significance was not observed in their counterparts. We conducted Cox proportional hazards analysis in patients with EGFR mutations by RFS, not by OS, because 8 of 11 patients (73%) with EGFR mutations had received EGFR-TKI treatment for recurrence. EGFR-TKI prolonged the survival of such patients; therefore, the follow-up period may not have been enough to evaluate OS in this study. Aromatase expression level and EGFR mutation status did not directly correlate, but aromatase expression only held prognostic significance for lung adenocarcinomas with EGFR mutations, which suggests that estrogenic signaling augments growth that depends on the EGFR pathway. Mah and colleagues reported that lower aromatase levels predicted greater chances of survival in women 65 years and older with NSCLC, particularly among women who had no smoking history (13). Although they did not investigate EGFR mutation status, their results are consistent with ours. Nose and colleagues reported that strong nuclear expression of ERβ correlated with better disease-free survival in patients with EGFR mutations, but found no such prognostic significance in patients without EGFR mutations (15). They also suggested that strong ERβ expression was a surrogate marker for good response to EGFR-TKI (26). These results, along with our own, indicate that female hormone-related factors, such as aromatase and ERβ, affect outcomes only in lung adenocarcinomas with EGFR mutation, which suggests that hormonal and EGFR pathways may contribute in concert to progression of lung adenocarcinoma.

To investigate the influence of differences in aromatase expression between the tumor and normal tissues on patient survival, we classified patients into the following 2 groups: T > N, in whom aromatase expression in adenocarcinoma tissue was higher than in nonneoplastic lung tissue (n = 51); and T < N, in whom it was lower (n = 43). We performed survival analyses between the 2 groups, but saw no significant survival difference in RFS (P = 0.22) or OS (P = 0.27). We further compared RFS and OS between the 2 groups in subsets divided by sex, smoking history, and EGFR mutation status, but saw no significant difference in these analyses (data not shown).

Recently, interactions between the ER and EGFR pathways have been investigated in vitro. A nonnuclear ER pool has been proposed that works via rapid signaling through various kinase cascades, including EGFR pathway and its downstream effectors in the lungs, such as MAPK (10, 11). However, the ER and EGFR pathways seem to act as alternate signaling pathways, with one upregulating when the other is inactivated (10, 41). This bi-directional crosstalk between ER and EGFR signaling suggests that simultaneous or combined therapy that targets both pathways could exert higher antitumor effect in patients with NSCLC.

Both in vitro and in vivo reports have demonstrated that estrogen downmodulator, alone or combined with EGFR-TKI, resulted in enhanced antitumor activity in NSCLCs (9–12, 23, 25, 41–44). Exemestane, an irreversible steroidal inactivator, either alone (41) or in combination with cisplatin (43) showed significant antitumor effects in 2 separate studies. Both letrozole and anastrozole, reversible steroidal inactivators, demonstrated similar antitumor activity in NSCLCs (12, 23, 44). However, lung cancer cell lines used in almost all of these studies were without EGFR mutations. We showed that the 11-18 cell line, which has an EGFR mutation accompanied with a high aromatase mRNA expression, was sensitive to exemestane alone and cell growth was significantly inhibited by the combination of exemestane and erlotinib. We also demonstrated that exemestane's antitumor effects in H358, H2228, and ACC-LC-319, which have high aromatase expressions without EGFR mutations, were much weaker than in 11-18. These results suggest that sensitivity to the aromatase inhibitor may depend on crosstalk between ER and the EGFR pathway; reducing estrogenic signaling by inhibiting aromatase might inhibit cell growth that depends on the EGFR pathway. Therefore, our result suggests that selecting patients with high aromatase expression accompanied by EGFR mutation might improve clinical responses to the combination of EGFR-TKI and aromatase inhibitor. However, we tested the growth inhibitory effects of only one cell line with high aromatase expression and EGFR mutation, thus this study remains limited. Further studies using in vivo and clinical models are needed to elucidate the therapeutic potential of aromatase inhibitor in lung adenocarcinomas with EGFR mutations. Traynor and colleagues reported a pilot study of gefitinib and fulvestrant in the treatment of 22 postmenopausal women diagnosed as NSCLC (45). No significant results were obtained in this small study, but combination therapy was well tolerated. Three of 12 patients tested for EGFR mutation status detected EGFR mutation. A trial of EGFR-TKI and estrogen downmodulator in patients with NSCLC with EGFR mutations may be therefore needed in the future.

In conclusion, high aromatase expression is correlated with poor outcome in patients with lung adenocarcinoma, including those with EGFR mutations. Aromatase may be a useful therapeutic target in lung adenocarcinomas with high aromatase expression and EGFR mutations. Although our results provide potential insights into the influence of aromatase expression in lung cancer, further studies are required to better understand the mechanisms of aromatase expression and interaction with EGFR signaling, and to determine the clinical applicability of aromatase inhibitors.

Y. Maehara reports receiving other research grants from Chugai and Pfizer. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M. Kohno, T. Okamoto, K. Suda

Development of methodology: M. Kohno, K. Suda, T. Yano

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Kohno, T. Yano

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Kohno, K. Suda, M. Shimokawa

Writing, review, and/or revision of the manuscript: M. Kohno, T. Okamoto, M. Shimokawa, M. Takenoyama, T. Yano

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Suda, H. Kitahara, S. Shimamatsu, H. Konishi, T. Yoshida

Study supervision: M. Takenoyama, T. Yano, Y. Maehara

The authors thank Dr. A.F. Gazdar, Dr. T. Hida, Dr. K. Tomizawa, Dr. T. Mitsudomi, and Dr. M. Takeshita for kindly providing cell lines.

This study was supported in part by a Grant-in-Aid for Scientific Research (C) from the Japan Society for the Promotion of Science (21591816).

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.