Upper Airway Gene Expression in Smokers: The Mouth as a “Window to the Soul” of Lung Carcinogenesis? Free

Perspective on Boyle et al., p. 266

Tobacco use remains the leading preventable cause of death in the United States, and cigarette smoking is the primary cause of respiratory tract cancers. Exposure to tobacco smoke is widespread, with approximately 45 million current smokers, 46 million former smokers, and millions of environmental (second-hand) tobacco smoke–exposed nonsmokers in the United States (1, 2). Despite the profound causal role of cigarette smoking, it leads to lung cancer in only 10% to 20% of smokers, and there are few clinical or genomic indicators of which smokers are at the highest risk (3–5). Some smokers develop cancer at an early age after relatively low cumulative smoke exposure, and some heavy smokers live into their 90s with no evidence of cancer. Therefore, the biological effect of cigarette smoke exposure on individuals varies, and an accurate biomarker of this variability in host response to tobacco smoke is needed.

Several studies involving cytologic and molecular techniques have established that cigarette smoking creates a field of injury in all exposed airway epithelial cells (6). Various groups have shown that noncancerous bronchial airway epithelial cells of current and former smokers with and without lung cancer display allelic loss (6), p53 mutations (7), and changes in promoter methylation (8) and telomerase activity (9). These observations suggest that the entire respiratory tree is affected by cigarette smoke; therefore, easily obtained proximal airway cells may provide insight into the types and degree of epithelial cell injury that have occurred in an individual smoker.

Our group and others previously have shown that genome-wide gene expression profiling of bronchial airway epithelial cells collected in bronchoscopic brushings reflects the physiologic response to cigarette smoke exposure and the irreversibility of a number of these responses/changes after smoking cessation (10–13). Prior studies have also shown that gene expression profiles in the cytologically normal bronchial airway epithelium can distinguish, with high sensitivity and specificity, smokers with and without lung cancer and can serve as a clinically relevant diagnostic biomarker (14–16). Notwithstanding this success, however, the invasiveness of bronchoscopic brushings prevents its use in large-scale studies of lung cancer early detection or risk assessment or as an intermediate end point in chemoprevention trials.

A potential alternative to bronchial airway cells for measuring the response to tobacco smoke exposure is cells lining the oral or nasal mucosa, which also are directly exposed to tobacco smoke. Our group has shown that genes with smoke-altered expression levels in the bronchial airways can be used to distinguish the oral and nasal mucosae of smokers from those of nonsmokers, suggesting that components of the genomic response to tobacco smoke are shared across the respiratory tract (17). Bhutani et al. further supported this conclusion when they found a strong correlation between bronchial and oral tissue in the degree of p16 and FHIT promoter methylation at baseline and 3 months after intervention in 127 smokers enrolled in a randomized placebo-controlled chemoprevention trial (18, 19); this rather specific study, however, did not profile global epigenetic changes in the bronchi and oral cavity. More recent analyses of genome-wide gene expression profiling of matched nasal and bronchial epithelium showed that the majority of gene expression consequences of smoking are common to both nasal and bronchial epithelium, suggesting that nasal epithelial gene expression may serve as a relatively noninvasive surrogate measure of physiologic responses to cigarette smoke in the lower airway (20).

As reported in this issue of the journal (21), Boyle et al. have shown a striking relationship between the gene expression responses to smoking in buccal and bronchial epithelium, thus adding significant insight into this upper airway “field of molecular injury.” They did whole-genome gene expression profiling of punch biopsies from the buccal mucosa of 40 healthy smokers and 40 healthy nonsmokers, a relatively large study for a global comparison of gene expression. Smoking-related gene expression changes identified in the mouth were very similar to previously published gene expression changes in the bronchial epithelium of active smokers (10). This new study also identified several genes that were more strongly induced or suppressed by smoking in women, an intriguing observation given data that have suggested that women may have an increased lung cancer risk (versus men) in association with a given amount of cumulative tobacco smoke exposure. Furthermore, the authors used the Connectivity Map (a genomic resource that is described later; ref. 22) in identifying the heat shock protein 90 inhibitor geldanamycin as a novel lung cancer chemopreventive agent that could potentially reverse epithelial gene expression changes associated with tobacco smoke exposure.

One of the most important implications of this work is the potential of the buccal mucosa to provide relatively noninvasive surrogate measures of the physiologic response of the lung to tobacco smoke that could be used in large-scale screening and chemoprevention studies in smokers at risk for lung cancer. Fifty-four probe sets (representing 41 genes) on the microarray were differentially expressed in the oral epithelium of smokers (versus nonsmokers), providing novel insights into the genomic networks that are perturbed by tobacco smoke (21). A previously published data set showed that smoking changed many of these genes similarly in the bronchial airway (11). A closer comparison of the two sites (Table 4 of ref. 21) reveals that the relevant genes undergo a much larger fold change in the bronchial airway than in the mouth, suggesting a more robust genomic response to smoking in the bronchial airway and/or a dilution of the epithelial cell–specific changes in buccal biopsies because they contain a mixture of cell types (versus the relatively pure population of bronchial epithelial cells in brushings). Given this difference in cell types between buccal biopsies and bronchial brushings, the authors' detection of a statistically significant relationship between smoking-related changes in the two sites is even more striking and argues strongly for a common respiratory tract–wide genomic response to tobacco smoke.

Another equally important implication of this study derives from its computational strategy, which leveraged the Connectivity Map in identifying potential novel strategies for the chemoprevention of smoking-related cancer. The Connectivity Map is a powerful genomic resource containing gene expression signatures resulting from treatment of multiple cancer cell lines with hundreds of compounds and small molecules (22). This resource can help in identifying potential novel cancer therapy and cancer prevention agents by linking disease-related or carcinogen exposure–related gene expression data with drug-induced perturbations in gene expression, thus identifying which drugs reverse aberrant gene expression signatures associated with cancer or carcinogen exposure (Fig. 1). By this approach, the authors identified a compound (geldanamycin) that might reverse tobacco-related gene expression changes and then found that geldanamycin inhibited the induction by tobacco smoke of CYP1A1 and CYP1B1 in an in vitro model of oral leukoplakia. Given the large number of publicly available microarray data sets profiling numerous human cancers, this strategy could be applied more broadly to identify compounds that may reverse early molecular changes in tumorigenesis and thus may be novel agents for preventing cancer in several sites (Fig. 1).

Although the implications of this work of Boyle et al. are potentially far reaching, there are several caveats that require further studies. Genomic profiling of matched upper and lower airway cells from the same individual (versus from unrelated individuals) is needed to fully characterize the ability of the mouth to serve as a surrogate for the bronchus (18, 20). Further study of buccal sampling methods is also needed. The method used by Boyle et al., punch biopsies after a local anesthetic, could limit the broad application of this tool in large-scale population-based studies. Furthermore, the genomic changes this study identified in buccal tissue reflect, in part, the varied cell types in biopsies (e.g., increased numbers of Langerhans cells in the biopsies of smokers) and not strictly a change in the transcriptome of epithelial cells. Buccal scrapings or brushings would be a less invasive approach and would collect a relatively pure population of epithelial cells. Maintaining RNA integrity in these noninvasively collected cells, however, is a significant challenge given the presence of RNases in saliva (23). Profiling noncoding microRNAs, which are more resistant than is mRNA to degradation, in buccal scrapings may provide a less invasive, more robust biomarker of host response to tobacco smoke exposure, as well as providing insight into the regulation of tobacco-related gene expression changes (24).

Another caveat involves the Connectivity Map findings pointing to a potentially novel chemopreventive agent. As the authors acknowledge (21), it is unclear whether inhibiting the induction of tobacco-related gene expression changes will equate with a reduction in clinically relevant carcinogenic effects. Many of the gene expression changes identified in healthy smokers likely represent appropriate activation of antioxidant and xenobiotic defense systems within the respiratory tract. Limiting the application of the Connectivity Map to airway gene expression changes that are associated with the development of various cancer subtypes in smokers may be a better approach for identifying effective chemopreventive strategies for tobacco-related malignancies.

In conclusion, this work provides another critical piece of evidence for the common airway-wide effect of cigarette smoke and, more important, for the potential of the mouth to serve as a surrogate for the genomic changes occurring within the lower respiratory tract of smokers. Profiling buccal gene expression across larger populations of current and former smokers with variable exposure histories would enable the development of host response biomarkers of the intensity and cumulative dose of, and extent of recovery from, tobacco smoke exposure and would enable the integration of these biomarkers with covariates that influence them (e.g., age, race, and gender) in lung cancer risk models. Longitudinal sampling of smokers could provide a mechanism for exploring the kinetics of these gene expression changes and for deciphering the acute versus chronic effects of smoke exposure. Heterogeneity in gene expression changes in the mouth could help in identifying smokers at a higher lung cancer risk and potentially in developing noninvasive assays of lung cancer risk markers and chemoprevention response biomarkers. If the field of tobacco smoke–related genomic injury in the lower respiratory tract does prove to extend to cells lining the oral cavity, the mouth indeed may provide the molecular “window to the soul” of lung carcinogenesis.

A. Spira: ownership interest and consultant/advisory board, Allegro Diagnostics, Inc.

We want to thank Joshua Campbell for assistance with Figure 1.