Long Exposure of Environmental Tobacco Smoke Associated with Activating EGFR Mutations in Never-Smokers with Non–Small Cell Lung Cancer

This article is featured in Highlights of This Issue, p. 1

Although lung cancer is the leading cause of cancer-related death and is predominantly caused by tobacco smoke, approximately 25% of all lung cancers worldwide are not attributable to tobacco smoke (1). In fact, there were about 30% never-smokers in Japan in a large cohort study including more than 20,000 patients with non–small cell lung cancer (NSCLC; ref. 2).

Lung cancer in never-smokers is unique in clinical characteristics and is suggested to be a distinctive disease. When considered as a separate category, lung cancer in never-smokers would constitute the seventh most common cause of cancer death worldwide (3, 4).

Recently, molecular biological studies showed that epidermal growth factor receptor (EGFR) mutations are detected in almost 60% of NSCLC in never-smokers (5), and the mutations happened exclusively in bronchial tree (6, 7). It has been suggested that lung cancer arises through different molecular mechanisms according to smoking status, and EGFR mutations are associated with never-smokers and K-ras mutations in ever-smokers; both mutations are almost exclusively found in adenocarcinomas (3). A high frequency of EGFR mutations in NSCLC in never-smokers leads to the conclusion that these mutations are one of the important biomarkers to clarify the disease.

Although several possible explanations have been proposed, the cause of lung cancer in never-smokers still remains unclear. Explanations include environmental tobacco smoke (ETS) exposure (8), radon exposure (9), occupational exposure (10), oncogenic virus (11), genetic change (12), and estrogen hormone (13, 14). A Japan Public Health Center-based prospective study showed that second-hand smoke exposure is clearly related to the development of lung adenocarcinoma in never-smokers in Japan (8). The study identified a statistically significant dose–response relationship between the quantity and intensity of husbands' smoking and their wives' incidence of lung cancer.

Given this background, we hypothesized that development of EGFR mutations are associated with ETS exposure, and we conducted prospective study using a detailed questionnaire to evaluate the association.

We consecutively enrolled prospective patients with newly diagnosed, histologically or cytologically proven NSCLC between March 2008 and January 2010 in National Kinki-chuo Chest Medical Center. Patients who were never-smokers (<100 cigarettes in the lifetime) and whose tumor tissue samples were available for analysis were eligible for the study. ETS was defined as regular exposure to tobacco smoke produced by an active smoker within a confined space for at least 1 year.

Detailed ETS information was obtained through a standard questionnaire that was carefully supported by interview by trained personnel. The questionnaire included items for years of ETS exposure from parents and other relatives as a child, years of ETS exposure from spouse/partner and/or children at home, and years of ETS exposure from co-workers at workplace. We also included family history of cancer in first-degree relatives (parents, siblings, and offspring) and menopausal status, if the patients are female. This study was approved by the institutional review board of the National Hospital Organization Kinki-chuo Chest Medical Center. All patients gave their written informed consent before enrollment.

A genetic analysis was done to detect EGFR mutations from exons 18 to 21 and codon 12, and 13 mutations in exon 1 of K-ras gene. The nucleotide sequence of the kinase domain of the EGFR gene was determined using PCR-INVADER assay of the individual exons (15), and that of K-ras was done using real-time PCR amplification and genotyping (16).

The cumulative dose of ETS (CETS) exposure was assessed in total smoker-years. This assessment was constructed to add exposure years of ETS from the 3 different parts in the questionnaire: 1) from parents and/or other relatives in his/her childhood; 2) from spouse/partner and/or children at home; and 3) from co-workers at a workplace. In each part, the same year was counted only once to avoid overlap of exposure periods from the different sources (e.g., in part 2, exposure from a spouse at age 30–50 years and his/her child at age 25–35 years would present a total ETS as 25, and if total ETS was 20 from part 1 and 3 respectively, the CETS of the patient would be 20 + 25 + 20 = 65).

Assuming the proportion of patients with activating EGFR mutations to be 60% among never-smokers with NSCLC, we planned to recruit at least 100 patients to detect a 10-year difference in CETS between EGFR mutation positive and negative populations at a significance level of 0.05 with a 90% statistical power.

All patients were divided into quintile groups by CETS. Significant differences in the variables were tested using the Pearson's, chi-square, Fisher's exact, and Mantel extension tests, wherever appropriate.

The odds ratios for the risk of activating EGFR mutations were calculated in a multivariate logistic regression model, including gender, ETS exposure, age, and family history of lung cancer in all the patients and those with adenocarnoma, respectively. To evaluate a dose–response relationship closely, CETS was treated as a continuous variable or quintile in the model. In the quintile model, the odds ratios for the second, third, fourth, and longest quintiles relative to the shortest were calculated. To test for a linear trend across the quintiles, we coded each quartile as 0, 1, 2, 3, or 4, and then included it in the model as a single variable. All the statistical analyses were done with SAS version 9.2 (SAS Institute).

A total of 126 patients were enrolled, and characteristics are summarized in Table 1. Among those, 124 patients (98.4%) had the ETS exposure, and the median CETS was 50 years (range: 0–118). Most patients were female (88.1%) with an adenocarcinoma histologic type (96.8%). The median age was 65 years (range: 29–88 years) in all the patients; 67 years (29–88 years) in females and 54 years (36–75 years) in males. About 23% of the patients were more than 75 years in this population, and age was significantly associated with cumulative ETS, as expected.

Genetic analysis was successfully done for EGFR status on all 126 patients from paraffin-embedded samples. Activating mutations were detected in 79 of 126 patients (62.7%), consisting of 36 in-frame deletions in exon 19, 43 L858R in exon 21, and the other 4 mutations were detected in 2 G719C in exon 18, 1 in-frame deletion in exon 18, and 1 T790M in exon 20. Forty-two samples were positive for EGFR mutations in the 66 surgical specimens, 33 in the 51 biopsy samples from bronchoscopy, and 8 in the 9 cytology samples from pleural effusion. There was no significant difference in EGFR detection rates among sample sources (P = 0.707). The remaining samples after EGFR analysis were 114, enough for K-ras analysis. Mutations were detected in 2 of 114 patients, and both were point mutations in codon 12, one of which was a G to C transversion and the other a G to A transition, respectively. No mutation was detected in codon 13. When the patients were categorized into 2 groups on the basis of EGFR status, incidence of the mutations was significantly higher in females and according to age (P = 0.002, P = 0.005).

When the female patients were divided into 2 groups on the basis of menopausal age, there was no association between them. There was no significant association between family history of cancer and EGFR mutations. When separated by exon 19 and exon 21 mutations in the 79 cases, the elderly tended to have exon 21 EGFR mutations.

We examined the association between cumulative ETS and activating EGFR mutations in the quintile groups separated by cumulative ETS. As presented in Figure 1, the incidence of EGFR mutations increased (shortest group = 44.0%; second group = 64.0%; third group = 48.0%; fourth group = 72.0%; longest group = 84.6%; P = 0.0033).

Next, we examined the association between the type of ETS and EGFR mutations (Table 2). The odds ratios for the EGFR mutations were analyzed for each 1-year increment in the total ETS. In Japan, age of 20 years is traditionally defined as reaching adulthood; also, home and workplace are major, separate places to live. We divided the patients into 2 groups on the basis of CETS in period by age of 20 years (childhood or adulthood) and place (household or workplace). There was a significant association between adulthood exposure (P = 0.0136) and household exposure (P = 0.0049), and EGFR mutations.

In the multivariate logistic regression model, including gender, cumulative ETS, age, and family history of lung cancer, both gender and cumulative ETS were significantly associated with the incidence of EGFR mutations in all cases (Tables 3 and 5) and in adenocarcinoma (Tables 4 and 6). When cumulative ETS was evaluated as continued variable in all cases (Table 3), the odds ratios for the EGFR mutations were 5.13 (95% CI = 1.47–18.0; P = 0.0105) for female gender, 1.02 (95% CI = 1.00–1.04; P = 0.0193) for each 1-year increment in CETS, 1.05 (95% CI = 0.48–2.33; P = 0.9014) for age, and 1.32 (95% CI = 0.46–3.83; P = 0.6040) for family history of cancer. In the quintile models (Tables 5 and 6), gender and CETS were also significantly associated with EGFR mutations. Moreover, the exposure duration of ETS of more than 68 years was the only significant effect on the mutations (Odds ratio = 5.44, 95% CI = 1.40–21.1; P = 0.0144).

We demonstrated the occurrence of activating EGFR mutations were significantly associated with female gender and long exposure of ETS in NSCLC in never-smokers in this prospective study. A large randomized clinical study, IRESSA Pan-Asian Study (IPASS; ref. 17), demonstrated that females and elderly had more EGFR mutations. The study included about 1,200 Asian never/light, former smokers with NSCLC and more than 90% of the patients were never-smokers. The rate was 63.0% for females versus 49.0% for males, and 68.5% for more than 65 years old versus 56.7% for less than 65 years old. Elderly patients have a chance for longer exposure to ETS and females have a higher chance of being exposed to ETS from a spouse, and this study accords well to our results.

EGFR mutations were detected in 65.9% in all the patients, and the 2 activating mutations of exon 19 deletions and L858R point mutations accounted for about 95.2% of all mutations included. Also, in this population, there were 2 patients who had K-ras mutations, and they were all EGFR wild type. These results are consistent with the previous published data (17, 18).

It has been reported that the development of activating EGFR mutations is inversely proportional to pack-years of smoking (19, 20). That carcinogen from cigarettes is similar in toxic effect for active and passive smoking and our results that longer exposure of ETS are significantly associated with EGFR mutations may be contradictory. However, we also demonstrated that the duration of exposure beyond a certain amount only affected the mutations. According to the study of diagnostic accuracy for the mutations, the optimal cutoff for predicting a positive was less than about 15 pack-years in cumulative dose (19, 20). We speculate a mechanism that low dose of tobacco carcinogens and long exposure induce activating EGFR mutations regardless of smoking status (Fig. 2). Another important aspect is that mainstream smoke and sidestream smoke from tobacco combustion are different, and sidestream smoke is produced at a lower burning temperature, and the quantities of its chemical constituents in both vapor and particulate phases differ from those of mainstream smoke (21). Mutational spectrum may possibly be different depending on the intensity and duration of tobacco exposure. In addition, formation of carcinogens indoors by surface-mediated reactions of nicotine with nitrous acid was recently reported to lead to potential third hand smoke hazards (22). Development of EGFR mutations may be associated with this third chemical that may affect healthy subjects in a time-lag manner.

According to our study and other epidemiologic studies (8), ETS play some part in the development of NSCLC in never-smokers. Based on reduced consumption of tobacco in the United States and Japan, the incidence of the disease will decrease as well. However, it is still uncertain that whether the actual number of never-smokers with NSCLC will increase or decrease (23). Defining molecular biomarkers for tumors in never-smokers remains a work in progress. Etinoderm microtubule-associated protein-like 4-anaplastic lymphoma kinase (EML4-AKT) gene dislocation has been discovered apart from EGFR mutations (24) that is also specific for never-smokers, and some of the tumors were reported to relate to K-ras transition and HER2 mutations (25). The influence of ETS was modest, and more comprehensive approaches are required.

Recently, a study on relationship between ETS and activating EGFR mutations was reported from Korea (26). The approach was similar to ours, but the results were inconsistent. Gender was not associated with the mutations, which was more frequently observed numerically in males in the study. Also, long ETS exposure was inversely associated with EGFR mutations, which is opposite to our study. One possible explanation for the difference is duration of the exposure. Frequency of the patients exposed to never-ETS was 24.6% in Korea and 1.6% in ours, and the median total exposure was 30 years in Korea whereas it was 50 years in ours. As mentioned above, the duration of the exposure beyond a certain amount of time was important for mutation development. In fact, EGFR mutation rate in never-smokers was 44.1% in that study that is lower compared with about 60% from Japanese studies (5, 20, 27), and the IPASS (17). Another possible explanation is intensity or concentration of ETS that may be different between the 2 studies. Although intensity of ETS may have a different impact on the development of the mutations, it is still difficult to grade and evaluate it. In fact, our questionnaire included ETS intensity, whereas most of the patients declined to fill them for inaccuracy of their memory, particularly as a child or at workplace. Because of few reply of ETS intensity in the questionnaire, we did not include them for analysis in this study, and further study may be required.

The main limitation of our study is lack of validation for the questionnaire because of the potential for recall bias by the patients. However, because there has been no biomarker to evaluate ETS to date, a detailed questionnaire is essential for study of ETS. Also, support by interview by trained personnel is important to obtain reproducible and accurate information. Moreover, standardized questionnaire will be required for comparison among ethnicity in future global study. Another weak point in this study may be the relatively small number of patients, although sample size was considered statistically in advance of the study. However, multivariate analysis did demonstrate that our data were significant, and the large clinical trial IPASS supports and enhances it.

In conclusion, EGFR mutations are significantly associated with female gender and long exposure of ETS in never-smokers with NSCLC. EGFR mutations are most prevalent and specific mutations in never-smokers. However, defining molecular biomarkers for tumors in never-smokers is still in progress, and a large molecular epidemiologic study will be required to clarify the disease comprehensively.

No potential conflicts of interest were disclosed.

This study was supported in part by a grant-in-aid for Cancer Research from the Ministry of Health and Welfare, Japan.

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.