Chemoprevention of Lung Cancer and Management of Early Lung Cancer

Gene Silencing

Another mechanism of lung carcinogenesis is loss of function of tumor suppressor genes. A growing body of evidence suggests that gene silencing through epigenetic means is crucial in lung carcinogenesis. For example, the tumor suppressor gene in NSCLC known as RASSF1A, which is also located in the region of chromosome 3p rich in tumor suppressor genes, encodes a protein that heterodimerizes with Nore-1, an important RAS effector with a proapoptotic effect. Evidence suggests that in NSCLC, RASSF1A can be inactivated by hypermethylation.

Another important gene that can be inactivated by epigenetic means is the retinoic acid receptor-β (RAR-β), which also maps to chromosome 3p. RAR-β is a nuclear retinoid receptor with vitamin A–dependent transcriptional activity. The RAR-β gene is gradually lost during lung carcinogenesis, and this may be caused by loss or hypermethylation. The maintenance of RAR-β in mature tumors is a risk factor for poor prognosis, and the RAR-β gene is differentially regulated in current as opposed to former smokers. For example, a retrospective study on methylation status and the occurrence of second primary tumors (SPTs) in patients with completely resected NSCLC showed an association between RAR-β2 hypermethylation in the development of SPTs only in former smokers. In current smokers, hypermethylation was associated with a protective effect, pointing to the value of context with regard to smoking status in understanding the biologic effects of retinoid receptors in lung cancer.

The role of miRNAs may also prove to be crucial. These small noncoding RNA molecules have been recognized as important in epigenetic control by altering the translation of proteins from messenger RNAs. miRNAs are able to target a multitude of messenger RNAs and have emerged as key posttranscriptional regulators of gene expression involved in cell proliferation, apoptosis, and stress resistance. miRNAs are also located in cancer-associated genomic regions or in fragile sites, suggesting that differences in miRNA expression may be induced by genomic alterations and play either a tumor suppressor or oncogenic role.

Cigarette smoke is a potent source of such molecular injury, and recent studies have shown that the components in cigarette smoke affect miRNA levels in the respiratory tract. An emerging notion is that smoking may lead to downregulation of miRNAs in the airway. The mechanistic understanding of the link between these miRNA changes and early lung carcinogenesis is just beginning to emerge. The fact that miRNA dysregulation is an early event in lung carcinogenesis may offer opportunities for chemopreventive approaches (i.e., reversing aberrant miRNA expression induced by smoking). For example, miRNA changes were found in biopsy specimens obtained in a phase II trial comparing iloprost (a prostacyclin analog) with placebo in high-risk patients, and these changes were related to histologic findings at baseline. Proteomic analyses of preinvasive lesions in the bronchial tree have uncovered patterns clearly different from those in normal epithelium, supporting the rationale for chemopreventive strategies.

Susceptibility to Lung Cancer and Balancing of Antagonistic Pathways

The risk of the development of lung cancer, even among the heaviest smokers, varies widely, perhaps because of genetic and even dietary factors. In fact, 85% of heavy smokers will not develop lung cancer, which suggests important differences among individuals in susceptibility to lung cancer. Several prediction models for lung cancer have been proposed. The Bach model uses variables related to age, smoking, and asbestos exposure in black populations, whereas the Spitz model and Liverpool Lung Project include family history and other occupational exposures in addition to age and smoking history in white populations. Although these models are easy to administer and calculate, their discriminatory power is only moderate, which is not surprising because the models fail to account for many comorbidities and other epigenetic factors. In 2012, Hoggart et al. proposed a prediction model that includes more variables, including biomarker exposure; socioeconomic status; comorbidities including pneumonia, chronic obstructive pulmonary disease (COPD), and emphysema; and body mass index for current smokers, former smokers, and nonsmokers. Despite researchers’ best attempts, a prediction model cannot assimilate all possible epigenetic and genetic factors to make consistently accurate predictions.

Accumulating evidence suggests that genetic and epigenetic factors are crucial in modulating individual susceptibility to lung cancer. An example is the identification of a susceptibility locus for lung cancer at 15q25 in a large genome-wide association study. Locus 15q25 contains nicotinic acetylcholine receptor subunit genes and seems to be involved in carcinogen metabolism. The two major carcinogens from cigarette smoke, benzo[a]pyrene and 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone, require metabolic activation before they can exert full carcinogenic effects.

Various activation pathways compete with detoxification pathways, and the balance between the two is crucial in modulating cancer risk. Cytochrome P450s serve as carcinogen-metabolizing enzymes, whereas glutathione transferases serve as detoxification enzymes. Both sets of these important genes are known to have significant polymorphisms that correlate with variations in lung cancer risk. Gene polymorphisms in microsomal epoxide hydrolase and their patterns of expression in the respiratory epithelium also influence the activity of antioxidant enzymes, skewing the balance between the two redox pathways.

Moreover, many researchers believe that DNA repair capacity may play an important role in susceptibility to lung cancer. CYP1A1, an enzyme of the cytochrome P450 family, plays a role in the development of EGFR mutations, hence altering the risk profile of lung cancer among smokers. Other modulators of risk in lung cancer include diet and gender. Dietary factors function as epigenetic modulators of lung cancer susceptibility. In several case–control studies, defective detoxification and defective repair of genetic damage were associated with increased individual susceptibility to lung cancer. Food constituents, such as those found in fruits and specific vegetables, appear to afford marked protection to individuals with limitations in their detoxification capacity. This relationship between diet and lung cancer has been explored extensively and is being used as a basis for the development of preventive approaches to lung cancer.

Vitamins and Micronutrients

Specific micronutrients have been associated with a lower risk of lung cancer, including vitamin E, selenium, isothiocyanates, polyphenols, betaine, and choline. A number of large randomized trials have sought to delineate which compounds in the human diet may be responsible for the protective effects. Among heavy smokers, a diet high in red meat consumption and with low adherence to the so-called Mediterranean diet has been found to be associated with increased risk of lung cancer. Conversely, a diet high in vegetable fats and fiber was shown to reduce lung cancer incidence among heavy smokers. The role of selenium has been investigated for preventing lung cancer among patients with selenium deficiency. A trial that sought to prevent SPTs after primary skin cancer in a selenium-poor population in Arizona showed a 34% reduction in lung cancer incidence, and because of this success, the role of selenium as a chemoprotective agent was explored in other malignancies. The Selenium and Vitamin E Cancer Prevention Trial was conducted to study the effect of selenium on prostate cancer incidence. In this trial, 32,400 men were randomly assigned to receive oral selenium, vitamin E, or selenium plus vitamin E as compared with placebo in a double-blind fashion and were monitored for a median of 5.46 years. The trial showed no benefit of selenium or vitamin E, or the combination, to confer any protection against prostate cancer. A phase III prevention study was published in 2013 by the Eastern Cooperative Oncology Group in which patients with resected stage I NSCLC were randomly assigned in a 2:1 ratio to receive selenium methionine at 200 μg/day or placebo. In this trial, selenium failed to demonstrate any objective benefit compared with placebo in protecting against the development of lung SPTs.

However, the Eastern Cooperative Oncology Group study was initiated without preliminary studies, and the recommendation must be to continue to build biomarker-driven, targeted, phase II approaches that appropriately pursue modulation of a crucial biomarker, followed by confirmation, before proceeding with large-scale studies.

Initially, researchers found considerable epidemiologic evidence for the potency of carotenoids and retinoids in reducing cancer risk in general, and lung cancer risk in particular. In fact, the original definition of chemoprevention as “attempts to reverse, suppress, and prevent carcinogenic progression to overt cancer” was based on experimental work showing that vitamin A analogs were capable of reversing or preventing epithelial carcinogenesis in animal experiments.

Early chemoprevention trials included broad populations of individuals who were at risk because of their exposure to tobacco and/or asbestos. Others trials focused on specific high-risk populations, such as uranium miners. Thus, with several decades of epidemiologic and dietary investigation augmented by various experimental systems, a number of compounds, including retinoids and carotenoids, were found to be capable of reversing cell damage. Considerable effort was invested in testing this approach in human populations. The trials focused on broad patient populations, including individuals at very high risk. In other words, studies included individuals with known premalignancy of the airway and patients who already had a primary tobacco-related cancer, because these cohorts were known to have extraordinary risk for developing an SPT and therefore offered a more efficient approach for evaluating candidate chemoprevention agents.

Status of Lung Cancer Screening and its Relationship to Chemoprevention

After decades of disappointing results, a robust approach to detect and cure early lung cancer has finally emerged. The National Lung Screening Trial (NLST)—the most expensive screening trial ever conducted by the National Cancer Institute—found that the use of low-dose CT (LDCT) in at-risk populations was associated with a significant reduction in lung cancer mortality. Based on this finding and related data, LDCT screening is now recommended by the USPSTF. Humphrey et al. outlined the underpinnings of the USPSTF draft statement on lung cancer screening in individuals at high risk of lung cancer and summarized the benefits and potential harms of this new screening approach. From a review of this synthesis, a number of evidence-based conclusions emerge that clarify important issues about this new public health service as we begin national dissemination of LDCT screening.

To begin, no strict definition of LDCT exists; it is typically considered to use 10% to 30% of the dose of radiation applied in a standard, noncontrast CT. LDCT has been demonstrated to be as accurate as standard-dose CT for detecting solid pulmonary nodules in most adults. In the NLST, acceptable chest CT screening was accomplished at an overall average effective dose of less than 2 millisievert (mSv), which is considerably less than the average effective dose of 7 mSv used for a typical standard-dose chest CT.

The use of low-dose radiation for screening reflects the context of this imaging application. In a physician’s office, an individual presenting with the appropriate history combined with signs or symptoms of lung cancer is much more likely to have lung cancer than is an asymptomatic individual participating in a lung cancer screening effort. With screening, the balance of risks and benefits is much narrower than in the standard setting of symptom-detected cancer, and so defining the optimal approach to the LDCT process with the greatest efficiency and quality while minimizing costs and harms is the fundamental challenge as we move toward national implementation of lung cancer screening.

The National Comprehensive Cancer Network (NCCN) has provided a useful source of information about how to approach this LDCT screening process. An example of the management and follow-up for one set of findings after LDCT screening is shown in Fig. 9.1 .

NCCN proposes that LDCT screening requires sophisticated multidetector CT scanners and analytic software, professional physicists and staff who can certify equipment and perform studies to a consistent standard at acceptable radiation exposures, qualified radiologists who use standardized terminology and standardized interpretation, appropriate guidelines, reliable communication with primary care physicians, medical environments that can absorb patients who require ongoing management, and the responsibility of tracking screened individuals and documenting outcomes.

NCCN has stratified risk groups that may be eligible for LDCT screening ( Table 9.1 ). The highest-risk group mirrors the eligibility criteria of the NLST, but the lower-risk strata still include people who may develop lung cancer but in whom the cost-to-benefit ratio may be narrower. NCCN also has a patient summary for lung cancer screening, which is an excellent source of objective information for the clinician to guide discussions with potential screening candidates on topics such as who is an appropriate candidate for LDCT screening and what the likely risk is relative to benefit.

An important topic that frequently arises when considering the benefits of screening is the concept of overdiagnosis. Overdiagnosis occurs when a cancer detected during screening does not behave in a lethal fashion, such that an individual may die with, not from, a screen-detected cancer. The USPSTF recently commented on the extent of overdiagnosis with chest x-ray screening in the Prostate, Lung, Colorectal, and Ovarian (PLCO) screening trial. Chest x-ray screening was the initial basis of concern for overdiagnosis in lung cancer screening. In the PLCO trial, among participants at risk for lung cancer due to heavy tobacco exposure, the cumulative incidence of lung cancer after 6 years of follow-up was the same in both the chest x-ray and usual-care groups (606 per 100,000 person-years vs. 608 per 100,000 person-years, respectively; relative risk, 1.00; 95% confidence interval [CI], 0.88–1.13). These results indicate a very small influence of overdiagnosis in this setting, and the PLCO findings suggest that overdiagnosis is not a major confounder in evaluating the benefit of lung cancer screening.

However, many other questions remain unanswered as we consider the implementation of LDCT screening because many misconceptions exist about this new service. One of the most important concepts is understanding the difference between providing screening care for a patient at risk for lung cancer and the more typical situation of treating a symptomatic patient suspected to have lung cancer. The risk–benefit considerations in these two settings are very different, and these differences mandate distinct management approaches.

Regarding other misconceptions, the core design assumption of the NLST, which evaluated more than 53,000 current or former smokers at a cost in excess of $250 million, was widely vetted. The consensus that emerged was that a target reduction in mortality of 20% with LDCT compared with chest x-ray outcomes would constitute compelling evidence of an objective screening benefit. An analysis of the full benefit of LDCT would have been much more expensive and require considerably more time to complete. Therefore it is not surprising that a more recent analysis of NLST outcomes with a rederived eligibility risk model constructed from PLCO case outcomes resulted in a more efficient rate of lung cancer detection. In that report, the mortality reduction was 30%, compared with the 20% benefit reported in the NLST. This example demonstrates how the LDCT screening process can be improved, and the reanalysis of the NLST data set underscores how valuable this data resource is in allowing process improvement for LDCT screening. As screening becomes implemented, it will be essential to continue to build screening registries from the data of as many screening individuals as possible so that continuous process improvement can be sustained.

LDCT screening does entail a range of possible harms. A 2012 review article from a joint society initiative communicated reservations about the balance of risks and benefits of lung cancer screening. The investigators reported that “uncertainty exists about the potential harms of screening and the generalizability of results.” By contrast, a comprehensive synthesis of the issues with screening management was published by the USPSTF in 2013 and conveyed a more optimistic perspective regarding how refined management approaches have improved the process of lung cancer screening. The USPSTF provides an excellent resource for guiding discussions to inform potential screening individuals about risks and benefits of this service.