This year, lung cancer will kill approximately 157,200 of the 171,900 people diagnosed with this disease in the United States alone.[1] According to recent Centers for Disease Control and Prevention data, lung cancer deaths have for the first time outstripped coronary artery disease deaths as the leading type of smoking-related death.[2] Despite new laws in many states imposing public smoking bans and higher taxes on cigarettes, young people-especially women-continue to start smoking at an alarming rate, and smoking remains the leading cause of premature death in our society. Lung Cancer, Smoking, and Public Health Lung cancer will continue to grow as a public health problem despite success with tobacco control because in contrast to the risk of cardiac disease, which decreases rapidly after smoking cessation, the risk of lung cancer in former smokers remains persistently elevated.[3] Close to 50 million former smokers in the United States remain at elevated risk for developing lung cancer.[4] At Harvard and other major cancer centers, half of the new cases of lung cancer are diagnosed in former smokers.[5] Moreover, 10% to 15% of all lung cancer cases occur in never-smokers.[ 6-8] Therefore, the number of lung cancer deaths in former and nevnever- smokers represents a solid majority of newly diagnosed lung cancers, resulting in over 90,000 deaths. Thus, sole dependence on tobacco control is an insufficient public health response for lung cancer, because this policy will in no way benefit most people with lung cancer. As lung cancer deaths in former and never-smokers exceeds the number of deaths due to colon and breast cancer combined, renewed efforts to improve lung cancer outcomes are clearly essential.[1] Another major reason for concern about tobacco-related diseases is that the care of these diseases accounts for the majority of health-care costs in our society.[2] With our aging demographics, the expected number of lung cancer cases in our society will continue to increase, exacerbating the current unsatisfactory state of affairs. While it is essential to maintain tobacco control efforts, additional measures are required to address the health-care needs of the many people who have already stopped smoking. Most smokers become addicted in childhood. The implications of this fact have been highlighted in a landmark Surgeon General's report.[9] Today, former smokers are at high risk of developing lung cancer, and when that occurs, their 5-year survival rate is 15%. Lethality of Lung Cancer Given that there is no validated test for early lung cancer detection, the current standard approach to lung cancer detection is to wait for signs or symptoms to develop. In that setting, newly detected lung cancer is generally rapidly fatal. The lethality of lung cancer is attributable to the high frequency of metastatic spread at the time of initial diagnosis. As recently discussed by Carney, there has been only modest objective improvement in outcomes related to treating advanced lung cancer.[10] This situation is particularly unfortunate at the current time, when women in the United States have the highest death rate in the world from lung cancer, with 27.2 deaths per 100,000 women.[11] A recent report has suggested that women may be inherently more susceptible to developing lung cancer than men.[12] Female cohorts in this study exhibited a prevalence- odds ratio of 2.7 vs male cohorts matched for age and smoking history. While it may be premature to definitively conclude that women are at inherently higher risk for developing lung cancer, a recent surgeon general report also pointed out a disproportionate rise in the incidence of lung cancer in women.[13] Clearly, this topic warrants further investigation. The only form of lung cancer that is likely to be curable is localized lung cancer. In other epithelial cancers with much more favorable 5-year survival, including cervical, colon, and breast cancer, techniques exist to routinely detect these diseases prior to metastatic spread. The 5-year survival rate for patients with completely resected stage I lung cancer is approximately 60%.[1,14,15] Finding a more effective way to diagnose premetastatic lung cancer would be a crucial step toward an improved lung cancer-related mortality rate. Based on studies of breast cancer screening, we know that achieving optimal benefit from early cancer detection also involves defining the most effective, efficient, and safest approach to the clinical management of screen-identified lung cancer.[16] In this review, we will consider how to build on the successes of other cancer screening efforts to detect and manage early lung cancer.[17] This involves outlining the specific elements for lung cancer that could make a screening program safe, affordable, and effective. We will also explore the current standards of lung cancer management and target areas where potential pitfalls and opportunities for improvement exist. Why Screen for Lung Cancer? Lung cancer treatment is currently managed on a case-by-case basis in our society, as are other sporadic cancers. When a patient presents to a physician's office with symptoms, a diagnostic work-up is performed at the discretion of the primary physician according to local standards of care. Given that presymptomatic tobacco- related illness in a screening setting will involve a vast number of at-risk individuals, defining how to achieve the best outcomes at the lowest cost is a major challenge. However, in a clinical environment, this more structured approach to managing early lung cancer would constitute a significant paradigm shift. Also, significant clinical research will clearly be required to define the appropriate clinical management that would achieve an optimal balance of risks and benefits in this new area of lung cancer care. Recent efforts based on clinical trials in advanced lung cancer have suggested that better outcomes can be achieved by following practice guidelines- for example, those of the National Comprehensive Cancer Network[18]-but gaining broad adherence to such recommendations is slow. The best precedent in this regard emerges from experience in the clinical management of breast cancer. Many experts attribute the reduction in breast cancer mortality beginning in the early 1990s, at least in part, to wide-based mammographic screening for breast cancer. Indeed, mammography provides a noninvasive, safe, fast way to screen women for breast cancer in advance of clinical symptoms. Mammographic instruments are also portable and can be brought into the community and the workplace to encourage screening compliance. With the maturation of breast cancer screening, it was recognized that the best outcomes were seen in centers that were dedicated to breast cancer screening and its attendant downstream clinical requirements.[16] Furthermore, it was found that disparities in imaging technique and instrument calibration could result in suboptimal case detection rates. To guard against these problems, measures were legislated to govern the quality control of relevant centers. Screening Criteria
Managing early lung cancer in an asymptomatic, high-risk population is likely to be an even more demanding process than that with breast cancer. It is therefore prudent to consider whether these types of measures are fundamental to success in managing preinvasive lung cancer. The precedent in a public health setting is that certain formal criteria must be met before disease screening is justified. In reviewing such criteria as summarized in Table 1,[19] it is evident that these are commonsense provisions, and it is useful to consider their relevance in lung cancer screening. As already discussed, with lung cancer being the most lethal of can- cers, its public health significance is indisputable. The question that emerges is whether we can now detect lung cancer at a point in time that improves cancer-related mortality. With chest x-ray screening, regional or distant metastatic disease was still found in most newly diagnosed lung cancer cases. As a result of rapid refinements in high-resolution spiral computed tomography (CT) imaging, this situation is changing and localized cancers are being found more frequently. That said, is this approach to lung cancer care affordable? A key issue in this regard is the cost of diagnostic evaluations in the screening process. Initial reports outlined significant challenges with the frequency and cost of CT screening.[20,21] A more recent report of the screening experience in Milan outlined a more disciplined approach to work-up of suspicious nodules. In that approach, nodules that were smaller than 6 mm were noted but only worked up the following year if these nodules were growing.[22] Some of those nodules did grow and were subsequently found to be cancers, but in all cases, surgery with curative intent could still be performed and all of those resected cases were still found to be stage I cancers. This is an example in which improving the downstream management of screen-detected early lung cancer can have a favorable impact on the overall process. Moreover, it illustrates that in population-based clinical management, to do less (such as to do fewer diagnostic work-ups with < 6-mm nodules) is sometimes to achieve more efficient overall outcomes. Why Is Lung Cancer Detected So Late? A factor contributing to the difficulty in imaging early lung cancer relates to its location deep in the thoracic cavity. In addition, the presence of complex bony structures and the cardiovascular system further exacerbate the challenge of thoracic imaging. In contrast, other organs such as the breast and the cervix are more easily accessible for diagnostic evaluation. Previous attempts at chest x-ray screening for lung cancer were clearly limited by the insensitivity of this diagnostic tool. Even with repeat chest x-rays performed every 4 months, as in the Mayo Lung Project,[23] only 30% of patients were found to have stage I disease. As the majority of patients were still diagnosed with advanced lung cancer, there was no significant cancer-related mortality reduction observed in that trial.[24] Subsequent reanalysis of aggregate trial data by an international body of screening experts found that the previously completed randomized lung cancer screening trials using chest x-ray and sputum cytology did not offer convincing evidence for or against lung cancer screening.[25] The continued lethality of lung cancer, the absence of informative screening evidence, and the emergence of more potent spiral CT imaging capabilities has mandated the urgent reevaluation of the benefit of lung cancer screening.[25,26] What Is Changing? Specific technical refinements in CT imaging were recently reviewed in detail.[ 26,27] Briefly, improvement in instrument design coupled with vastly more capable microprocessor capabilities allows a much more detailed thoracic imaging study to be acquired much more rapidly. Since the imaging interval is so much shorter (seconds rather than minutes), the confounding influence of respiratory motion is eliminated. Spatial relationships in the thorax can be defined with much greater precision than with early-generation CT scanners. The greatest advantage of spiral CT technology is the reliable detection of very small nodules. With state-of-the-art, high-resolution spiral CT scanning, slice thickness as thin as 0.6 mm can be acquired through the entire chest volume in less than 20 seconds, with radiation exposure comparable to a normal chest x-ray. Spiral CT scanners already exist in most hospitals and free-standing imaging centers in the developed world. In contrast to some screening tests currently done for other cancers (eg, mammography and Pap smear), the CT scan is neither painful nor invasive, making this test more acceptable to the general population. These factors along with the rapidly decreasing cost of a CT scan are making this tool even more attractive for potential application in a public health setting. Small-Volume Primary Tumors
Conventional wisdom tells us that early lung cancer, although seldom found, can be associated with favorable long-term survival. A recent study has suggested that the small size of lung cancer at the time of detection by screening may not result in favorable long-term outcomes.[28] Patz et al evaluated mortality outcomes among patients with stage IA lung cancer at a single institution over an 18-year interval. They concluded that primary tumor size had no significant impact on mortality; however, this cohort had a favorable overall 5-year survival of 80%.[28] It is unclear whether tumors were initially detected by CT scan or by chest x-ray, but as only 26 (of 510) subjects had a subcentimeter nodule at the time of detection, the suggestion that detection of subcentimeter primary tumors (in a screening setting) will not improve survival seems premature. Many investigators are excited about the possible routine detection of small-volume primary lung cancer, including a much higher percentage of subcentimeter primaries. To attempt to generalize about potential outcomes with spiral CT-detected screening cases based on historical experience from a single institution retrospectively evaluating referral cases is, at best, speculative. The Metastatic Process
Several reports from radiologists have suggested that lung cancer is inherently metastatic from its inception.[ 20,29] From an epidemiologic perspective, it is known that the peak incidence of lung cancer lags 2 decades after high levels of cigarette consumption.[30] In this regard, the natural history of lung cancer is similar to that of typical colon or breast epithelial cancers. Epithelial cancers begin as a localized phenomenon. Cancer progression involves many discreet molecular steps to acquire the biologic competence to permit a tumor to grow in three dimensions. Additional acquired steps in metastatic competence may include factors such as the ability of a cancer to degrade matrix structures in the basement membrane. This property allows metastatic cells to escape in the vascular or lymphatic system, where they can "seed" the entire body. Some authors contend that a significant number of lung cancers can remain latent for the entire life span of the patient.[20,31] Other authors have suggested that lung cancer may in fact be more virulent than other cancers, conferring the ability to metastasize much earlier than other tumors of similar size.[28,32] These contentions are at odds with a large body of clinical experience.[15,33,34] Clinical Data
How frequently will subcentimeter lung cancers metastasize prior to detection on screening?[32] In the Memorial Sloan-Kettering experience, only about 10% of subcentimeter lung cancers were found to involve distant metastatic disease.[33] Using the 2000 Surveillance, Epidemiology and End Results (SEER) registry with regard to the impact of tumor size on survival, Wisnivesky et al reviewed data from patients with stage I non-smallcell lung cancer (NSCLC) diagnosed since 1988 who had undergone curative tumor resection. Among stage I malignancies, 12-year survival was inversely proportional to primary tumor size (Figure 1).[34a] High curability with smaller primary tumor size is also evident from the decreased number of deaths from breast cancer and cervical cancer with the advent of mammography and cervival cytomorphologic analysis, respectively. In the recent published experience from Milan,[22] the size of screendetected primaries on prevalence screening averaged about 20 mm, with the average primary size of incidence cases being about 15 mm. In Table 2, we summarize the salient characteristics from recent spiral CT reports.[ 22,35-38] It is apparent that the average lesion size on initial evaluation as well as on annual follow-up represents much smaller-volume cancers than clinicians are accustomed to managing. In this situation, we not only have challenges in defining the optimal clinical care, but we also may not know the natural history of such lesions. Along with the Cornell group, the early pioneers in this field were from several institutions in Japan. Recently, Ryutaro Kakinuma, an investigator from the National Cancer Center, Tokyo, presented the 27-year screening experience of the Anti-Lung Cancer Association.[39] In the first 18 years of this effort, chest x-ray and sputum cytology were used as screening tools. During that time, the Association performed over 26,000 screening evaluations, detecting stage IA cancer in about 42% and resulting in a 49% 5-year overall survival rate. Over the past 9 years, the Association has moved to the use of spiral CT. In over 15,000 screening evaluations during this interval, the stage IA detection rate was 78%, and those patients had a 5-year overall survival rate of 78%. Additional details of this experience are summarized in Table 3.[39] Other centers in Japan have shown comparable results, especially in regard to the frequency of stage IA detection in their screening efforts. While this visionary Japanese experience is clearly promising, it is still necessary to wait for clinical trials evidence of improved lung cancer-related mortality benefit before drawing any conclusions about the real benefit of lung cancer screening. To that end, these data certainly support the wisdom of expeditiously conducting a large randomized trial of lung cancer screening. Indeed, based on the promising data from the pilot screening trials at Cornell and other sites, the National Cancer Institute launched a major randomized study called the National Lung Screening Trial (NLST) to compare spiral CT to chest x-ray in a highrisk tobacco-exposed population.[40] The trial will accrue 50,000 current and former smokers at coordinating centers throughout the country looking for differences in lung cancer-related mortality between the two arms. Study accrual has been proceeding well ahead of schedule. Depending on the outcome, follow-up for as long as 10 years may be needed to reliably evaluate the benefit of spiral CT screening. Costs of Lung Cancer Care
Lung cancer surgery is associated with considerable morbidity and a predictable rate of mortality, so the process of defining less aggressive approaches to surgical evaluation is attractive. In the same vein, when proposing to employ a populationbased screening strategy, the cost of clinical management becomes a critical determinant. As there are approximately 100 million current and former smokers in the United States, this issue can become paralyzing. The cost of starting and sustaining such a program is predicted to be enormous and, in and of itself, has been suggested as a reason not to screen for lung cancer.[21,41] On the other hand, in the United States we are already spending roughly $50 billion on health care for tobaccorelated disease.[2] Despite this expenditure, current treatments for these tobacco-related diseases are suboptimal and tobacco use remains the overwhelmingly leading cause of premature mortality in our society.[30] Because more than $50 billion in economic productivity is also lost annually as a consequence of tobacco-related diseases,[2] the true cost of our tobacco use is not generally appreciated. A premise for screening research in lung cancer is that moving the focus of care from late metastatic disease to earlier disease may effect improvements in health and economic outcomes, as have been recently described for cardiovascular disease screening.[42] With research progress, could the cost of an effective, welldesigned and implemented lung cancer screening system be sustainable within the envelope of resources currently expended in lung cancer care?