The Impact of Sex and Smoking Status on the Mutational Spectrum of Epidermal Growth Factor Receptor Gene in Non–small Cell Lung Cancer

The epidermal growth factor receptor (EGFR) is a receptor tyrosine kinase (TK) that is frequently overexpressed in malignant tumors, and its signal transduction cascade leads to a multitude of effects, including cell proliferation, cell differentiation, angiogenesis, metastasis, and antiapoptosis (1, 2). Recent findings have shown that EGFR mutations are exclusively present in lung cancer and showed significant association with sensitivity of EGFR TK inhibitor (EGFR-TKI), including gefitinib and erlotinib (3–6). Several clinicopathologic factors were identified to be related to the frequency of EGFR mutations, including adenocarcinoma histology, female sex, and never-smoking status and East Asian ethnicity (6). Most mutations involved nucleotides of exons 18 to 21 in the TK domain, especially deletions in exon 19 and the L858R point mutation in exon 21. We previously reported that exon 19 deletions and exon 21 point mutations were frequent in never smoker compared with ever smoker and in female sex compared with male sex, but there was no bias for sex and smoking status in exon 20 insertion mutation (7). More recently, we found the lack of sex effect for the EGFR-TKI resistance related EGFR T790M mutation located in exon 20 in a minor clone of the tumor (8). These findings suggest the sex, smoking status, or their interactions influence the EGFR mutational spectrum in an exon-specific manner in non–small cell lung cancer (NSCLC) cases. In addition, the clinical impact of EGFR-TKIs on patients with various mutations was different according to the type of EGFR mutant, indicating that biological significance of mutations is not equal according to different mutant types (6, 9–11). These observations indicate that a better understanding of factors influencing the EGFR mutational spectrum may be important to understand the tumor pathogenesis and response for treatment of NSCLC.

We have independently analyzed and reported the relationship between EGFR mutations and clinicopathologic factors, but a large-scale study is necessary to glean novel and important features of EGFR mutations in NSCLC. Because no single study can analyze more than a modest number of mutations, we combined our data to build up database for analysis. In this study, we examined EGFR exons 18 to 21 mutation status in 1,467 NSCLC patients to investigate the impact of sex factor, smoking factor, or interacted factor of sex and smoking status on the EGFR mutational spectrum in NSCLC.

NSCLC samples. Surgically resected specimens of 1,467 NSCLC patients were collected from 1,062 patients in Japan (415 from Aichi Cancer Center Hospital, 413 from Okayama University Hospital, and 234 from Chiba University Hospital), 160 patients in the United States (M.D. Anderson Cancer Center, Houston, TX), 144 patients from Taiwan (Veterans General Hospital, Taichung, Taiwan), and 101 patients from Australia (Prince Charles Hospital, Brisbane, Australia). Regarding ethnicity, all cases of Japan and Taiwan were East Asian, and the majority of cases from the United States and Australia were Caucasian (6). The details of patient characteristics are shown in Table 1. A total of 1,467 patients consisted of 916 males and 551 females, 927 ever smokers and 540 never smokers, and 1,011 adenocarcinoma, 377 squamous cell carcinoma, 41 large cell carcinoma, 23 adenosquamous carcinoma, and 15 other types of NSCLC. Never smokers were defined as those with lifetime exposure of 100 cigarettes or less. Ever smokers were defined as current or former smoker with lifetime exposure of more than 100 cigarettes. Specimens of 1,437 patients were from frozen samples, and those of 30 patients were from paraffin-embedded specimens (6, 12, 13). All cases from the United States and Australia, 93 cases from Taiwan, and 666 cases from Japan were previously reported (6, 12, 13). In this study, 137 cases from the Aichi Cancer Center Hospital, 259 cases from the Okayama University Hospital, 51 patients from the Veterans General Hospital, Taiwan, were added to the previous study. Corresponding nonmalignant lung tissues were examined in 921 patients. Institutional Review Board permission and informed consent were obtained at each collection site.

DNA extraction and sequencing analysis. Genomic DNAs were isolated by digestion with proteinase K followed by phenol-chloroform (1:1) extraction and ethanol precipitation from frozen specimen (14) and by DEXPAT (TaKaRa) from paraffin-embedded tissues following the manufacturer's instructions. For specimens analyzed at the Aichi Cancer Center, the total RNAs were isolated from frozen tissues of the tumor specimens using the RNAeasy kit (Qiagen). DNA-based direct sequencing for exons 18 to 21 of EGFR was done using same PCR primers and conditions at the University of Texas Southwestern Medical Center and Okayama University, except paraffin samples (6, 13). The RNA-based analysis using one-step reverse transcription-PCR for exons 18 to 21 region using Qiagen OneStep reverse transcription-PCR kit (Qiagen) was done at the Aichi Cancer Center (12).

Statistical analyses. All the statistical analyses were conducted by STATA 9.2 (Stata Corp.). A χ² test was applied to test the frequencies if appropriate. A multivariable logistic regression model was applied to estimate odds ratios (OR) and their confidence intervals (CI). Age, sex, histology (adenocarcinoma or non-adenocarcinoma) region (East Asia or United States and Australia), and institutions were adjusted in the model. A model evaluated sex (male versus female) and smoking status (ever versus never) as binary variable, and a model including the interaction between them was also evaluated. We regarded P values <0.05 as statistically significant.

Detection of EGFR mutations in NSCLC patients. We examined the EGFR mutation status in 1,469 lesions from 1,467 NSCLC patients and found 418 mutations in 407 patients. Seven patients had double mutations, and one patient had a triple mutation in a tumor. One patient had three tumors, and each of them harbored a mutation [one exon 19 deletion and two exon 21 mutations (L858R mutations)]. These multiple mutant cases were regarded as each single mutation; thus, our cohort was assumed to have 1,478 cases. There were 197 mutations in exon 19, 176 in exon 21, 21 in exon 18, and 24 in exon 20. Mutations of exon 19 consisted of 196 in-frame deletions involving three to eight codons around codons 747 to 749 (Leu-Arg-Glu sequence) and one point mutation (D761Y; ref. 15). Mutation of exon 20 consisted of 22 insertion and two point mutations (T790M; ref. 16). Other mutations in each exon were point mutations. EGFR mutations were present in 328 (30.9%) patients (336 mutations) of 1,062 patients from Japan, 15 (9.4%) patients (15 mutations) of 160 patients from the United States, 52 (36.1%) patients (52 mutations) of 144 patients from Taiwan, and 12 (11.9%) patients (15 mutations) of 101 patients from Australia. Because EGFR mutation was predominantly present in adenocarcinoma and experimental data also showed a close relationship between EGFR mutations and adenocarcinoma histology, the frequency of EGFR mutations in adenocarcinoma was also examined (6, 17). EGFR mutations were present in 317 (41.0%) patients (324 mutations) of 773 patients in Japan, 15 (15.5%) patients (15 mutations) of 97 patients in the United States, 50 (53.2%) patients (50 mutations) of 94 patients in Taiwan, and 9 (19.1%) patients (11 mutations) of 47 in Australia. These results showed that there was a similar tendency in the prevalence of EGFR mutations in Japan and Taiwan (East Asia) and in the United States and Australia. The EGFR mutation was not found in 921 nonmalignant samples examined (6, 12, 13).

The EGFR mutational spectrum according to sex and smoking status. We divided our cases into four groups based on sex and smoking status: 420 female never-smoker cases, 140 female ever-smoker cases, 130 male never-smoker cases, and 788 male ever-smoker cases. In females, never smokers had 119 mutations in exon 19, 100 in exon 21, 8 in exon 18, and 12 in exon 20. Female ever smokers had 14 mutations in exon 19, 6 in exon 21, 2 in exon 18, and 2 in exon 20. In males, never smokers had 29 mutations in exon 19, 23 in exon 21, 4 in exon 18, and 1 in exon 20. Male ever smokers had 35 mutations in exon 19, 47 in exon 21, 7 in exon 18, and 9 in exon 20. The EGFR mutation spectrum of the total group, East Asia, and the United States and Australia cases stratified by histologic subtypes are shown in Table 2 and Supplementary Material 1.

The impact of smoking status on the mutation spectrum according to sex status. The impact of smoking status on the EGFR mutation spectrum was examined for each sex. Never-smoker female or male status was set as a reference. In females, exon 19 and exon 21 mutations were significantly less frequent in cases from ever smokers compared with never smokers (exon 19: OR, 0.31; 95% CI, 0.16-0.60; P = 0.001; exon 21: OR, 0.16; 95% CI, 0.06-0.38; P < 0.001), and there were no significant differences between exons 18 and 20. In males, exons 19, 21, and 18 mutations were significantly less frequent in cases from ever smokers compared with never smokers (exon 19: OR, 0.14; 95% CI, 0.08-0.25; P < 0.001; exon 21: OR, 0.21; 95% CI, 0.11-0.40; P < 0.001; exon 18: OR, 0.13; 95% CI, 0.03-0.51; P = 0.003; Table 3A).

The same analyses were done among histologic subtypes. Restricted to adenocarcinomas, we had similar results to those of total cases (analyses of histologic subtypes were adjusted in total cases as described). In females, exons 19 and 21 mutations were significantly less frequent in cases from ever smokers compared with never smokers (exon 19: OR, 0.33; 95% CI, 0.17-0.65; P = 0.001; exon 21: OR, 0.14; 95% CI, 0.05-0.37; P < 0.001), and there were no significant differences between exons 18 and 20. In males, exons 19, 21, and 18 mutations were significantly less frequent in cases from ever smokers compared with never smokers (exon 19: OR, 0.13; 95% CI, 0.07-0.24; P < 0.001; exon 21: OR, 0.21; 95% CI, 0.11-0.39; P < 0.001; exon 18: OR, 0.16; 95% CI, 0.03-0.96; P = 0.045; Supplementary Material 2). In non-adenocarcinoma, whereas the number of cases were limited, exon 18 mutations were significantly less frequent in cases from ever smokers compared with never smokers in males (OR, 0.10; 95% CI, 0.01-0.82; P = 0.032; Supplementary Material 2).

The impact of sex difference on the mutational spectrum according to smoking status. The impact of sex status on the EGFR mutational spectrum was examined in ever smoker and never smoker cases separately. Female never-smoker or ever-smoker status was set as a reference. In never smokers, there was no difference between female and male cases on each exon. In ever smokers, exon 19 mutations were significantly less frequent in male compared with female cases (OR, 0.34; 95% CI, 0.16-0.70; P = 0.003; Table 3B). Restricted to adenocarcinomas, similar results were obtained; in ever smokers, exon 19 mutations were significantly less frequent in males compared with females (OR, 0.32; 95% CI, 0.15-0.67; P = 0.003; Supplementary Material 2). In non-adenocarcinoma, the number of cases was too limited to analyze.

The impact of interaction between sex and smoking status on the mutational spectrum. We also examined the impact of interaction between sex and smoking status on each exon when female never smoker was set as a reference. Regarding exon 19 mutations, the effect of male sex and smoking status was significantly decreased compared with an expectation from the independent effect of male sex and smoking status (Interaction OR, 0.41; 95% CI, 0.17-0.99; P = 0.047; Table 2A). On the other hand, there was no significant interactive effect of smoking and sex on exon 21, 18, and 20 mutations. Stratified by histologic subtypes, adenocarcinoma cases showed similar results to total cases that the effect of male sex and smoking status was significantly decreased in exon 19, as compared with an expectation from independent effect of male sex and smoking status (Interaction OR, 0.36; 95% CI, 0.15-0.89; P = 0.027; Supplementary Material 1). There was no significant interactive effect on exons 21, 18, and 20 mutations. In non-adenocarcinoma cases, the number of cases was too small to analyze the impact of interaction between sex and smoking status on each exon.

The impact of smoking and sex differences on the mutational types. We analyzed the relationship between sex and smoking status and types of mutation consisting of deletions, insertions, and point mutation. As shown above, cases with deletion were only present in exon 19; cases with insertion were only present in exon 20; and pointed mutation cases consisted of all cases of exon 18, one of exon 19, one of exon 20, and all of exon 21. Thus, the impact of smoking and sex differences on the mutational types were similar to that on the mutational spectrum (Supplementary Material 3). Indeed, the effect of male sex and smoking status in total cases tended to be decreased in deletion type compared with an expectation from independent effect of male sex and smoking status (Interaction OR, 0.44; 95% CI, 0.18-1.08; P = 0.074). Stratified by histologic subtypes, adenocarcinoma cases showed similar results to total cases that the effect of male sex and smoking status was significantly decreased in exon 19, as compared with an expectation from the independent effect of male sex and smoking status (Interaction OR, 0.39; 95% CI, 0.16-0.97; P = 0.043). There was no significant interactive effect on point mutation and insertion.

Previous studies examined the effects of sex or smoking status among EGFR mutant cases and found that the EGFR L858R mutation in exon 21 was more frequent in females and never smokers as compared with exon 19 deletions (13, 18). By contrast, in this study, NSCLC cases with EGFR wild type were considered as a reference, and the impact of sex and smoking status on the EGFR mutation spectrum was evaluated to identify the different effects of sex and smoking status on mutation of each exon. Furthermore, the discovery of an interaction of sex and smoking on exon 19 mutation (or deletion mutation) implies that the pathogenesis of exon 19 mutations (which mainly consist of deletions) may be different from mutation in other exons (mainly point mutations). Nonetheless, the mechanisms of how deletion or point mutation occurs in genomic DNA are unknown.

The influence of sex and smoking status on lung cancer has been an issue of interest. Zang and Wynder have reported that the ORs for major lung cancer types at every level of exposure to tobacco smoke were higher in women than in men likely due to the higher susceptibility of women to tobacco smoking (19), and a recent review supported this observation (20). Previous molecular data indicate that cancers arising in women smokers harbored significantly more tobacco-related p53 mutations, G:C to T:A transversions, than those in male smokers (20). Furthermore, DNA adduct levels of benzo(a)pyrene diol epoxide (a tobacco carcinogen) in females with lung cancer were markedly greater than those arising in males, suggesting sex differences in susceptibility to DNA damage derived from environmental carcinogen exposure (21). Of note, there was no specific association between smoking status and the type of point mutation (transversion and transition) in EGFR mainly because that majority of point mutation is G to T transversion at the second letter of codon 858.

The biological differences in EGFR mutations have also been identified in experimental models. The in vitro transforming activity of mutants L747-749delA750P (exon 19 deletion) and D770-771insNPG (exon 20 insertion) were higher than that of L858R (exon 21 mutation) and G719S (exon 18 mutation) mutations (9). In a transgenic mouse model, L858R mutants were associated with a more aggressive adenocarcinoma compared with exon 19 deletion mutants (delL747-S752; ref. 17).

Clinically, individual mutant types also show different outcomes following EGFR-TKI treatment. The exon 20 insertion mutant cells revealed resistance to gefitinib or erlotinib, and those of exon 18 mutant revealed intermediate sensitivity (lower than exon 19 or exon 21 mutant cells) in vitro (9). In patients treated with EGFR-TKIs, survival time of exon 19 mutant cases was longer than that of exon 21 mutation (10, 11, 22). These findings strengthen the role and importance of EGFR mutational types in the pathogenesis and treatment response of NSCLC.

In this study, we analyzed cases from four countries, Japan, the United States, Taiwan, and Australia. Our previous study showed the presence of the ethnic differences in the frequency of EGFR mutations (6). Besides ethnicity, there is a possibility that other factors, including differences in lifestyle, may affect the EGFR mutational spectrum. Thus, care is required to account for possible differences found when analyzing such multinational cohorts. Notwithstanding, much useful information may be obtained from a large-scale international database, and our study will encourage the establishment of the worldwide database for the EGFR mutation for common scientific use. Indeed, a recent study conducted by co-authors (K.T., T.M., Y.Y., and T.K.) using a case-control design in a single institution indicated that the impacts of smoking and sex on the risk of NSCLC might differ between EGFR mutant and EGFR wild-type tumor status (23). The results in the study basically support our finding that sex, smoking, and their interaction differentially affected the site of somatically acquired EGFR mutations. However, the reasons for our findings, especially the impact of sex difference on mutation spectrum, are unclear. Further study is warranted from various viewpoints including the effect of sex hormones.

In summary, we found that sex, smoking status, and the interactive effect of sex and smoking status influence the EGFR mutational spectrum, suggesting that individual exons of EGFR may have differing susceptibilities for mutagenesis. In addition, establishment of a large-scale database for EGFR mutations may be a useful resource to study the role of EGFR mutations for understanding the pathogenesis and novel therapeutic strategies for treating and preventing lung cancer.

We thank Makiko Tabata, Cancer and Thoracic Surgery, and Yukinari Isomoto, Central Research Laboratory, Graduate School of Medicine and Dentistry, Okayama University, for excellent technical support. We thank Dr. Makoto Suzuki, Department of Thoracic Surgery, Chiba University, Chiba, Japan, for providing some specimens.