- Email: [email protected]
Screening for Lung Cancer Revisited and the Role of Sputum Cytology and Fluorescence Bronchoscopy in a High-Risk Group
Screening for Lung Cancer Revisited and the Role of Sputum Cytology and Fluorescence Bronchoscopy in a High-Risk Group* Timothy C. Kennedy, MD, FCCP; York Miller, MD, FCCP; and Sheila Prindiville, MD
Lung cancer is an epidemic disease that is underrepresented in the research funding for early detection and chemoprevention arenas. Screening programs have been discouraged for both financial and political reasons. Yet, increasing evidence suggests that screening and early detection may improve outcome in lung cancer. Sputum cytology examination has been shown in several studies to lead to detection of lung cancer at an earlier stage, resulting in an improved 5-year survival rate. Monoclonal antibody detection, fluorescence bronchoscopy, and low-dose spiral CT increase diagnostic sensitivity and improve the ability to localize early-stage lesions. Utilizing these new techniques and improving the definition of high-risk groups may improve the success and cost-effectiveness of early detection based on sputum cytology. The ultimate goal of improving longterm survival in lung cancer will be achieved only when cancer can be detected in its early stages and lesions can be localized in large numbers. Advances in the last 15 years offer an encouraging vision for the value of early detection and effective treatment for lung cancer. (CHEST 2000; 117:72S–79S) Key words: bronchoscopy; lung cancer; monoclonal antibodies; positron emission tomography; screening; sputum cytology; survival Abbreviations: MoAb ⫽ monoclonal antibody; MSKCC ⫽ Memorial Sloan-Kettering Cancer Center; NCI ⫽ National Cancer Institute; PET ⫽ positron emission tomography; SPORE ⫽ Specialized Program of Research Excellence
the end of the 20th century, few physicians routinely A torder chest radiographs or collect sputum for cytology
examination in the hope of discovering lung cancer in a high-risk patient at an early stage. Effective early detection of disease requires that three essential criteria be met: (1) there must be a preclinical phase of the disease, and (2) technology to detect the disease in the preclinical stage must be available, which (3) enables effective interventions when the disease is discovered within the required window of opportunity. There is ample evidence of a prolonged preclinical phase in lung cancer. Clones of endobronchial cell popu-
*From the University of Colorado Health Science Center, Division of Pulmonary and Critical Care Medicine, Lung Cancer Institute of Colorado, Denver, CO. Supported by National Cancer Institute Grant P50 CA58187. Correspondence to: Timothy C. Kennedy, MD, FCCP, 1721 East 19th Ave, #366, Denver, CO 80218 72S
lations accumulate genetic mutations leading to a progressively more malignant and ultimately invasive malignant state. They may only roughly reflect the Saccommano morphologic criteria (normal cells; hyperplasia; metaplasia; mild, moderate, and severe dysplasia; carcinoma in situ; and invasive carcinoma) as they evolve into malignancy. Neither the critical number of mutations, critical combinations of mutations, or a necessary order of events (if such exists) is known at this time. Elucidating these issues is one of the most exciting areas of current research because of the potential prospect that such understanding may lead to better detection and interventions. Research, however, currently suffers from a lack of scientific surveillance in high-risk populations. It is rather amazing that in this country there is no systematic surveillance of individuals at high risk for lung cancer. If there were, it might aid in the development of new detection technologies, identification of key risk factors, and/or the validation of interventions for precancer or early cancer. In fact, half of the new cases of lung cancer occur in ex-smokers.1 More women die annually of lung cancer than breast cancer, but lung cancer is not considered a woman’s issue. Far more Americans die of lung cancer than of AIDS, but lung cancer victims are not viewed as victims and are underrepresented by advocacy groups. Despite an increase in the number of new cases (171,600/yr), the large numbers of current and ex-smokers at risk in the United States (50 million each), an unchanging 72% mortality rate among lung cancer patients,2 and its position as a predominant cause of death among Americans (6%), there is no strategy and, in fact, little research funding for this epidemic. There are accepted screening strategies for almost all other common solid tumors, including colon, breast, and prostate cancers. However, it is curious that, with some exceptions, few of these screening strategies have been subjected to large, carefully controlled, randomized studies. The data on mammograms remain mixed and controversial in women ⬍50 years old. Fiberoptic colorectal endoscopy screening has not been proven to be of benefit in asymptomatic, occult blood-negative individuals. The benefit of occult blood testing itself remains unsettled, in spite of the Minnesota Colon Cancer Control Study results. The utility of the prostate specific antigen as a screening tool is particularly unresolved, heavily criticized for leading to overdiagnosis. Only the cervical Papanicolaou smear has been clearly shown to reduce case mortality (by 90%). Complicating the issue further, three large, ambitious, randomized, National Cancer Institute (NCI) studies3 from the 1970s and 1980s conducted at Mayo Clinic, Johns Hopkins Oncology Center, and Memorial Sloan-Kettering Cancer Center (MSKCC), as well as a fourth study from Czechoslovakia,4 failed to demonstrate a disease-specific mortality benefit from screening smokers for lung cancer. As a result, the NCI and the American Cancer Society make no recommendation for screening patients for this most important cancer epidemic. Therefore, except for smoking cessation efforts—which offer no benefit to the 50 million Americans who have stopped smoking but Multimodality Approach to Lung Cancer
remain at increased risk for lung cancer—there is no public health strategy for early detection, intervention, or prevention of this disease.
Screening for Lung Cancer There is no question that the earlier lung cancer is discovered, the better are the patient’s chances of survival (Fig 1).5,6 Patients with radiograph-documented stage I lung cancer have a 40 to 80% 5-year survival rate, whether discovered by screening or accident. However, mortality in lung cancer worsens radically with advancing stage at the time of diagnosis. Interestingly, radiologically negative lung cancers discovered by sputum cytology alone in the three NCI studies were associated with an 85 to 90% 5-year survival, even though such patients were few and the histology was usually squamous cell carcinoma. More recently, both Bechtel et al7 and Saito et al8 demonstrated that patients with positive sputum cytology but who were radiologically negative for cancer had a very high survival rate. Squamous cell cancers comprise only 29% of all lung cancer cases.2 Based on this, mortality rates due to squamous cell carcinoma of the lung (n ⫽ 46,081 in 1999) are similar to those of the next most common solid tumors (colon [n ⫽ 47,900 in 1999] or breast cancers [n ⫽ 43,700 in 1999]). These statistics suggest that curing just squamous cell lung cancer would be similar to curing colon or breast cancer. Thus, the significant potential of sputum cytology as an early detection tools warrants reconsidering its role in selected patients. In the NCI randomized studies conducted at MSKCC and at the Johns Hopkins Oncology Center, there was no benefit observed from adding sputum cytology to annual chest radiographs in screening regimens. A low number of cancers were discovered by sputum cytology among subjects randomized to dual screening with annual chest
radiographs and sputum cytology every 4 months vs subjects undergoing a single screen with only an annual chest radiograph. Retrospectively, this may have been due in part to the relatively low risk for lung cancer in the study population. Interestingly, Tockman et al9 used the archived sputa from the group of subjects who eventually developed lung cancer, following moderate dysplasia in sputum to demonstrate the greater sensitivity and potentially earlier diagnosis gained from using monoclonal antibody (MoAb). All three NCI studies demonstrated that periodic screening results in a significant shift toward earlier diagnosis and markedly increased rates of resectability. There is also lengthened survival from the time of diagnosis and improved case fatality. However, there was no disease-specific reduction in mortality: the “gold standard” of efficacy in screening for early fatal disease.
Mayo Clinic Study In the Mayo Clinic study, 10,933 participants underwent an initial prevalence screening with chest radiograph and sputum cytology; 4,618 subjects were randomized to receive chest radiographs and sputum cytology every 4 months, while 4,593 were assigned to a control group advised to have chest radiographs and sputum cytology evaluations annually, since this represented the standard of care at the time. The enrolled subjects (adult males ⬎ 45 years old who had smoked at least 20 cigarettes daily within the year prior to enrollment) underwent 6 years of screening and 3 additional years of follow-up observation. Through the follow-up period, 206 cases of lung cancer were found in the screened group and 160 cases in the control group. There were 122 lung cancer deaths in the screened group and 115 in the control group. There was a shift toward lower stages and better resectability in the screened group. The 5-year survival rate of lung cancer
Figure 1. Survival rates for 2,382 patients after resection of lung cancer, classified by postoperative stage. Differences between groups: stage I vs stage II, p ⬍ 0.01; stage II vs stage IIIA, p ⬍ 0.01; stage IIIA vs stage IIIB, p ⬍ 0.01; stage IIIB vs stage IV, not significant. Reprinted with permission from Naruke et al.5 CHEST / 117 / 4 / APRIL, 2000 SUPPLEMENT
73S
cases in the screened group was much higher than in the control group (33% vs 15%). However, because there were more lung cancer deaths in the screened group than in the control group, it was concluded that there was no advantage for screening. Strictly speaking, this study suggests screening increases the mortality rate. The reason there were more lung cancer deaths in the screened group clearly was that there were more lung cancers in this group, although the reason for this disparity remains a source of dispute. There are a number of valid criticisms of the Mayo Clinic screening study,10 –15 including the following: 1. There were both significant control contamination and screening noncompliance. For instance, 50% of the control group had chest radiographs during the last year of the study, and 73% in the last 2 years of study. On the other hand, 25% of the “screened” group were not compliant with the screening recommendations. 2. It is now accepted that the inclusion criteria provided for a low statistical power. A large number of the participants had a low pack-year smoking history, while other high-risk factors (airflow obstruction, genetic predisposition, occupational exposures) were not considered in the entrance criteria. Thus, the study design provided only a 20% chance of finding a 10% reduction in case mortality. If it had shown a 10% mortality reduction, the low power would have meant daunting screening costs to save 17,800 Americans out of the 100 million smokers and ex-smokers in 1997. One cannot emphasize enough the importance of better defining the risk factors in lung cancer. 3. It appears that the study population was not followed long enough. Major considerations of the counterintuitive results of the study have been lead time bias, length-time bias, and overdiagnosis bias. These issues are common to all cancer screening trials, not just those in lung cancer. The high survival rate among patients with cancer diagnosed by sputum cytology has been discounted based on the concept that these central endobronchial tumors were indolent squamous cell carcinomas with low intrinsic mortality, perhaps so low that they would not have caused significant mortality if undiagnosed and untreated. Overdiagnosis bias, which Eddy16 has suggested is the most likely explanation for the study results, could have been resolved if, with longer follow-up, the incidence cases of the control population caught up with the screened cases. This did not occur within the time frame of the study. Unlike the prostate carcinoma story, however, autopsy evidence by McFarlane et al12 does not suggest a significant incidence of clinically low-level, undiagnosed lung carcinoma. Sobue et al,14,15 as part of a Japanese case-control lung cancer screening trial, also concluded that overdiagnosis bias was an unlikely explanation for the Mayo Clinic results. These authors observed equivalent survival among patients who refused surgical excision for screened vs 74S
symptom-detected cases of stage I lung cancer during a 10-year follow-up period. These cases emphasized radiologic detection in all but a few cases that were sputum detected. These studies lend little support for a significant prevalence of benign lung cancer. The limitations of the technology at that time certainly influenced outcome. After the NCI studies, on reviewing negative chest radiographs from patients discovered to have lung cancer on later radiographs, scientists were able to retrospectively identify the smaller lesion on the earlier radiograph.10 Most lesions were seen on posteroanterior views, and 5% were seen on lateral views only. This suggests insufficient diagnostic lead time in many cases. Some sputum-discovered carcinomas could not be localized on initial bronchoscopy, potentially adding to mortality. In some cases, it appears that the lesion initially was subtle; in others, the lesion was distal to the larger airways accessible by the fiberoptic bronchoscope used at the time of the study (usually 5.4-mm or 6.2-mm outer diameter). Although many tumors are in the large airways, better access to the subsegmental airways would considerably extend the ability to localize some other tumors suggested on sputum cytology. Modern smaller diameter bronchoscopes also may improve lead time.
Czechoslovakian Study Like the Mayo Clinic study, the Czechoslovakian study began with a prevalence screen and randomization to a screened or a control group. The control group was not instructed to obtain a chest radiograph or sputum testing, and thus may have provided a better control than the group in the Mayo Clinic trial. The screening period was only 3 years, and the follow-up period was 3 years, during which time both groups had annual chest radiographs. Interestingly, the incidence of cancer continued to increase in the screened group during the follow-up period, which suggests that overdiagnosis bias is not a compelling explanation for the larger number of lung cancer cases in the screened group (n ⫽ 206) vs the control group (n ⫽ 160) out a total of 6,364 enrollees.4,13
Johns Hopkins and MSKCC Studies It is important that the 5-year survival rate of cancer patients in both the control and the screened groups in the Johns Hopkins and MSKCC studies was nearly 35%, well above the national average of 12 to 13% at the time of the studies. The screened group in the Mayo Clinic study had a similar 5-year mortality rate. Unfortunately, experts and physicians in general in this country still feel that screening high-risk groups is unwarranted, which is not what the data suggest. In fact, because the control group appeared to follow their instructions to get chest radiographs to a substantial degree, and the screened group was somewhat noncompliant, the study, in a sense, compared a very compliant group to a somewhat compliant group, yielding improved diagnostic capacity.
Where Do We Go From Here? Some professionals have argued that recommending lung cancer screening would undercut the impact of Multimodality Approach to Lung Cancer
smoking cessation efforts. However, because current evidence suggests that half of lung cancers occur in exsmokers, ignoring these individuals who have successfully overcome addiction as recommended would seem callous. The incidence of new cases of lung cancer is currently increasing at a time when the prevalence of smoking is decreasing. Because none of these studies addressed a particular high-risk group, the cost per discovery of each lung cancer case was high, discouraging interest in both research and clinical case finding. High costs are particularly counterproductive in the current managed care era of capitated primary care physicians and scarce research dollars. Nonetheless, the NCI has underway a large randomized trial known as the Prostate, Lung, Colon, Ovary Trial, in which 148,000 men and women aged 60 to 74 years will undergo an initial chest radiograph followed by yearly radiographs for 3 years; the control group will not undergo chest radiographs. Unfortunately, no effort is being made to balance the groups for smoking histories or to enroll high-risk groups, thus repeating the problems from the previous randomized studies. Therefore, the design may be ill-suited to the task of addressing the important questions concerning lung cancer, and the study runs the risk of providing yet another expensive trial attempting to resolve the issues but incapable of doing so.
Recent Advances Several recent advances in understanding the biology and genetics of the carcinogenic process have renewed hope that an early detection strategy may be useful.17,18
MoAb in Sputum Tockman et al9 suggested that diagnostic lead time might be extended by adding a specialized procedure as an adjunct to sputum cytology (ie, using two MoAbs directed at a difucosylated Lewis X epitope and a 31-kd protein). Using archived sputum specimens from subjects who developed lung cancer after documented moderate dysplasia during the Johns Hopkins study, the investigators used this procedure to demonstrate positive staining on smears as much as 24 to 48 months prior to the diagnosis of cancer using Saccommano cytology criteria. Studies are underway to prospectively validate this technique.
Genetic Advances An increasing number of important mutations have been identified in lung cancer, including allelic deletions or tumor suppressor gene inactivation in 3p, 5q, 9p, 11q, 17p, 13q, 18q, and 22q. At least three alleles appear important on 3p, with more extensive loss commonly seen in small cell lung cancer and less extensive loss associated with non-small cell lung cancer. p53 alterations appear to be nearly universal in solid tumors, including lung cancers.19 A functional relationship in cell-cycle control between Rb and P16 has been described,20 in which an abnormality in either appears to be associated with any lung malignancy. Array chip technology promises to accelerate the process of identifying important mutations and
critical patterns of mutations. Thus, elucidating molecularbiological events is key to improving the sensitivity of sputum evaluations and may lead to effective chemoprevention strategies, as well as improved treatment of systemic disease.
Fluorescence Bronchoscopy Lam et al21 introduced fluorescence bronchoscopy that is effective without the use of protoporphyrins for the purpose of localizing endobronchial lesions of moderate dysplasia or more malignant tissue. In patients with dysplasia or malignancy in sputum, or in other patients likely to have a high prevalence of endobronchial dysplasia in central airways, such as occupational exposures or previously successful resected lung cancer, this technology may offer increased diagnostic sensitivity using a relatively benign procedure in properly selected patients.
Spiral CT of the Chest Improved chest imaging also may provide increased sensitivity and improved specificity for localizing earlystage lesions. Kaneko et al22 recently screened 1,369 subjects with at least 20 pack-year smoking histories using spiral CT chest examinations. They discovered 15 cancers (14 stage I), including 11 not detectable by plain chest radiographs. Spiral CT imaging takes 15 to 30 s, allowing complete chest imaging in one breath-hold. In addition, it has the radiation exposure of a mammogram and can pinpoint lesions as small as 2 to 3 mm in size. These newer, more sensitive technologies have led some scientists to raise valid concern that overidentification of benign lesions may lead to possible overtreatment morbidity, similar to that experienced with early CT scans. Care must be used to resolve this issue. Shimizu et al23 have suggested using helical CT imaging to improve anatomic distinctions between nodules, vessels, bronchi, and chest wall over conventional CT imaging; this may reduce the incidence of false-negative malignant diagnoses. Bronchoscopy and positron emission tomography (PET) scanning are quite benign and viable approaches to distinguishing between benign and malignant lesions preoperatively, with transthoracic needle aspiration considered less so because of a significant rate of pneumothoraces. A period of follow-up evaluation CT scans have been used successfully to reduce inappropriate surgery. Improved imaging also serves a critical role in addressing the increasing incidence of adenocarcinoma (often arising peripherally), as well as in shrinking the “silent” area of the lung (ie, the subsegmental airways not accessible to the bronchoscope). Thus, although 80% of lung cancer cases in the NCI studies were discovered by chest radiographs, this older technique may be supplanted by newer procedures in those higher-risk groups in whom the enhanced value of sputum cytology has been demonstrated.
Current Studies Sputum in a High-Risk Group Based on data suggesting that airflow obstruction is a risk factor for lung cancer independent of smoking history, CHEST / 117 / 4 / APRIL, 2000 SUPPLEMENT
75S
investigators from the University of Colorado Specialized Program of Research Excellence (SPORE) evaluated sputum cytology from 533 subjects with ⱖ 40 pack-years of smoking and COPD. Enrolled patients were negative for malignancy in the last year based on chest radiograph, but 27.5% had moderate dysplasia or worse (17.1-fold greater than in the Mayo Clinic study; Table 1).24 This included invasive carcinoma or carcinoma in situ in 1.7% of subjects, severe dysplasia in 0.8%, and moderate dysplasia in 25%. These patients required additional clinical investigation. Preliminary analysis suggests this population will have an lung cancer incidence of ⬎ 1%/yr without systematic use of radiographs. Accruals in this study have reached 1,850 subjects with similar distribution of cytology stage. The study will be repositioned for the future as an ongoing incidence study; its goal will be to further the understanding of the molecular biology of preneoplasia and early malignancy, thereby providing a paradigm for intervention and a rationale for earlier detection and treatment. It is hoped that by better defining risk variables for this population, it may be possible to identify a suitable target population for screening or aggressive case finding in the future. Because the prevalence of dysplasia did not increase between the group of patients with a 40 –pack-year habit and those with a 120 –pack-year history, current enrollment includes subjects with ⱖ 30 pack-years. We have observed that this population has an inherently high mortality rate from all causes. Thus, although this group is acceptable for evaluating the biology of premalignancy, the ideal group for cost-effectiveness screening may be different and will perhaps exclude patients with lung impairment and/or comorbidities that would limit the benefit of early diagnosis.
Fluorescence Bronchoscopy A multicenter study including University of Colorado SPORE investigators was completed evaluating the sensi-
Table 1—Cytological Diagnosis of Sputum Samples Collected From Participants Enrolled in the Colorado SPORE Sputum Screening Program*† Diagnosis‡ 0 1 2 3 4 5 6 Total
Ex-Smoker
Current Smoker
Total
47 (8.8) 34 (6.4) 185 (34.7) 95 (17.8) 4 (0.8) 2 (0.4) 2 (0.4) 369 (69.2)
21 (3.9) 27 (5.1) 73 (13.7) 38 (7.2) ⫺ (⫺) 3 (0.5) 2 (0.4) 164 (30.8)
68 (12.8) 61 (11.4) 258 (48.4) 133 (25.0) 4 (0.8) 5 (0.9) 4 (0.8) 533 (100)
*Data are presented as No. (%). A total of 533 participants provided sputum samples deemed adequate for cytopathologic diagnosis (369 ex-smokers and 164 current smokers). These represent the relative frequency of these cytodiagnoses overall and by smoking status. †From Kennedy et al.24 ‡0, no significant abnormalities; 1, regular metaplasia; 2, mild dysplasia; 3, moderate dysplasia; 4, severe (marked) dysplasia; 5, carcinoma in situ; 6, invasive carcinoma. 76S
tivity of the fluorescence bronchoscope in localizing premalignant and malignant endobronchial lesions.25 The bronchoscope provided 271% increased relative sensitivity in detecting lesions when added to white light examination, and even greater sensitivity when obvious, large malignancies were excluded (Fig 2). Interestingly, a University of Colorado SPORE subgroup of patients who underwent bronchoscopy only because of moderate dysplasia in sputum revealed a small but significant prevalence of undetected malignancy (4 of 43 patients), suggesting that such patients may deserve bronchoscopic evaluation on clinical grounds. Our relative sensitivity in localizing significant lesions (moderate dysplasia, severe dysplasia, or carcinoma in situ) in subjects with moderate dysplasia on sputum was 5.75 in favor of fluorescence examination. A subsequent two-bronchoscopist trial, in which each operator is blinded as to the findings of the other and patients are randomly assigned to white light vs fluorescence examination and to a specific operator, is underway at the University of Colorado SPORE. Our preliminary finding of a threefold increase in sensitivity with the fluorescence light system among the first 31 subjects, is similar to our earlier finding.26 It appears that the increased sensitivity is associated with decreased specificity, resulting in many false-positive biopsies. This adds to the expense of analyzing more biopsies and increases the time of the procedure. If there is logic to sputum cytology screening of heavy smokers with airflow obstruction, and investigating those patients with abnormal sputum cytology with fluorescence bronchoscopy, it is to discover a preponderance of squamous cell carcinomas located in the central airways. Because of underlying respiratory impairment, treatment strategies in such cases would likely emphasize local treatment and lung-sparing tactics. A potentially more important use of sputum cytology followed by fluorescence examination is the potential inherent in localizing premalignant lesions in order to treat them prior to becoming invasive.
Implications for Case Finding The very high prevalence of malignancy in our study population leads to intriguing implications for case finding of early malignancies. If 1.7% of the population has signs of carcinoma on sputum culture, as we have demonstrated, and another 2.2% (9% [4 of 43] ⫻ 25%) are discovered after fluorescence bronchoscopy because of moderate dysplasia, and another 1% are discovered by spiral CT in a peripheral location, the potential for early-stage diagnosis arguably exceeds that available with virtually any other non-lung cancer screening program! What other cancer screening program could provide 4.9% discovery rate on a prevalence basis? We emphasize that this argument depends entirely on narrowing the case finding to higher-risk patients.
Future Challenges Future challenges and areas of research in lung cancer screening include the following: Multimodality Approach to Lung Cancer
Figure 2. White-light bronchoscopy image and fluorescence image of the same area, demonstrating identification of (top) normal tissue in the right upper lobe or (middle and bottom) malignant tissue in the right bronchus. Used with permission from Xillix Technologies, Vancouver, British Columbia.
CHEST / 117 / 4 / APRIL, 2000 SUPPLEMENT
77S
1. Advancing technology to improve sputum sensitivity, lead time to diagnosis, and specificity. Monolayering of sputum cytology preparation, improved preservatives and staining, and automated/quantitative evaluation of sputum all have realistic promise to achieve this when applied to the appropriate risk group. 2. Improving definition of risk factors for lung cancer, particularly those useful for narrowing the risk group worth screening. 3. Improving techniques in localizing endobronchial lesions. Next-generation fluorescence bronchoscopy likely will be more sensitive, easier to incorporate into a bronchoscopic examination, and, when used with smaller-diameter bronchoscopes, able to localize more distal lesions. 4. Validating spiral CT chest examinations as an inexpensive, safe, fast, sensitive, and specific screening tool. PET scans, though expensive, may play an important adjunct role for secondary evaluations. The cost (and availability) of PET scans may be more realistic when used at higher volumes. 5. Developing and validating local therapies. Since the University of Colorado Lung Cancer SPORE highrisk population commonly is afflicted with comorbidities in addition to respiratory impairment, establishing acceptable local therapies is needed. Currently available are high-dose brachytherapy, photodynamic therapy with an increasing number of agents including the recently US Food and Drug Administration-approved protoporphyrin, unipolar electrocauterization, cryotherapy, and Nd-YAG laser therapy. Currently, there is no proven answer to the questions, “How should one best treat severe dysplasia?” or “How should one best treat carcinoma in situ?” These questions will only be addressed when lesions can be localized in substantial numbers, and the natural history of such lesions are better understood. The prospect of working on such questions when this arena of pulmonary medicine evolves from its current stagnation is extremely exciting. In the last 15 years, many advances have renewed hope that early detection of lung cancer is possible and may be useful. These include better definition of risk groups, better chest imaging, better localization of endobronchial lesions, and better local treatment modalities. Studies have demonstrated low disease-specific mortality when the disease is discovered by sputum cytology. What is needed now is another major push at the problem using the advances of recent years. The current understanding of lung cancer risk factors allows us to better target the relevant population, but more is needed in this regard. Furthermore, although imaging techniques, localizing techniques, and local treatment strategies clearly are better, these advances will stagnate in the absence of well-designed population-based studies. Trials must be designed to retest the premise that aggressive seeking of early lung cancer will reduce mortality, saving many victims who would otherwise perish, perhaps unnecessarily. 78S
ACKNOWLEDGMENTS: The author acknowledges colleagues Susan Proudfoot, MPH; Wilbur Franklin, MD; and Paul Bunn, MD for reviewing this manuscript and research work done in conjunction with the University of Colorado Lung Cancer SPORE.
References 1 Strauss G, Decamp M, DiBiccaro E, et al. Lung cancer diagnosis is being made with increasing frequency in former cigarette smokers [abstract 1106]. Proc Am Soc Clin Oncol 1995; 14:362 2 American Cancer Society. Cancer Facts and Figures, 1999. Atlanta, GA: American Cancer Society, 1999 3 Melamed MR, Flehinger BJ, Zaman MB. Impact of early detection on the clinical course of lung cancer. Surg Clin North Am 1987; 67:909 –924 4 Kubik A, Polak J. Lack of benefit from semi-annual screening for cancer of the lung: follow-up report of a randomized controlled trial on population of high-risk males in Czechoslovakia. Int J Cancer 1990; 45:26 –33 5 Naruke T, Tsuchiya R, Kondo H, et al. Implications of staging in lung cancer. Chest 1997; 112(suppl)4:242S–248S 6 Mountain CF, Dresler CM. Regional lymph node classification for lung cancer staging. Chest 1997; 111:1718 –1723 7 Bechtel JJ, Kelley WR, Petty TL, et al. Outcome of 51 patients with roentgenographically occult lung cancer detected by sputum cytology testing. Arch Intern Med 1994; 154:975–970 8 Saito Y, Nagamota N, Shin-Ichiron O, et al. Results of surgical treatment from roentgenographically occult bronchogenic squamous cell carcinoma. J Thorac Cardiovasc Surg 1992; 104:401– 407 9 Tockman MS, Gupta PK, Myers JD, et al. Sensitive and specific monoclonal antibody recognition of human lung cancer antigen on preserved sputum cells: a new approach to early lung cancer detection. J Clin Oncol 1988; 6:1685–1693 10 Strauss GM. Measuring effectiveness of lung cancer screening: from consensus to controversy and back. Chest 1997; 112(suppl 4):216S–228S 11 Strauss GM, Gleason RE, Sugarbush DJ, et al. Screening for lung cancer: another look; a different view. Chest 1997; 111:754 –768 12 McFarlane MJ, Feinsein AR, Wells CK. Clinical features of lung cancers discovered as a postmortem “surprise”. Chest 1986; 90:520 –523 13 Prorok PC, Chamberlain J, Day NE, et al. UICC workshop on the evaluation of screening programes for cancer. Int J Cancer 1984; 34:1– 4 14 Sobue T, Suziki T, Matsuda M, et al. Survival for clinical stage 1 lung cancer not surgically treated. Cancer 1992; 69:685– 692 15 Sobue T, Suzuki T, Naruke T, et al. Efficacy of lung cancer screening; comparison of results from a case-control study and a survival analysis. Jpn J Cancer Res 1992; 83:424 – 430 16 Eddy D. Screening for lung cancer. Ann Intern Med 1989; 111:389 –399 17 Bunn P, Carbone D, Grim J, et al. Fifth I.A.S.L.C. lung tumor biology workshop, August 13–17, 1996, Ermatingen, Switzerland. Ann Oncol 1997; 8:27–36 18 Miller YE, Franklin WA. Molecular events in lung carcinogenesis. Hematol Oncol Clin North Am 1997; 11:215–234 19 Weintraub SJ. Inactivation of tumor suppressor proteins in lung cancer. Am J Respir Cell Mol Biol 1996; 15:150 –155 20 Boers JE, Ten Velde GPM, Thunnissen BJM. p53 in squamous metaplasia: a marker for risk of respiratory tract carcinoma. Am J Respir Crit Care Med 1996; 153:411– 416 21 Lam S, MacAulay C, Hung J, et al. Detection and localization of early lung cancer by imaging techniques. J Thorac CardioMultimodality Approach to Lung Cancer
vasc Surg 1991; 105:1035–1040 22 Kaneko M, Eguchi K, Ohmatsu H, et al. Peripheral lung cancer: screening and detection with low dose spiral CT versus radiography. Radiology 1996; 201:798 – 802 23 Shimizu T, Narabayash I, Yesugi Y, et al. Three-dimensional display of pulmonary nodules using helical CT. Diagn Radiol 1995; 13:209 –216 24 Kennedy T, Proudfoot S, Franklin W, et al. Cytopathological analysis of sputum in patients with airflow obstruction and
significant smoking histories. Cancer Res 1996; 56:4673– 4678 25 Lam S, Kennedy T, Unger M, et al. Localization of bronchial intraepithelial neoplastic lesions by fluorescence endoscopy. Chest 1998; 113:696 –702 26 Prindiville S, Miller Y, Cook R, et al. Fluorescence bronchoscopy compared to white light bronchoscopy in the detection of high grade bronchial dysplasia [abstract]. Proc Am Assoc Cancer Res 1998; 39:117
CHEST / 117 / 4 / APRIL, 2000 SUPPLEMENT
79S
Lung cancer is an epidemic disease that is underrepresented in the research funding for early detection and chemoprevention arenas. Screening programs have been discouraged for both financial and political reasons. Yet, increasing evidence suggests that screening and early detection may improve outcome in lung cancer. Sputum cytology examination has been shown in several studies to lead to detection of lung cancer at an earlier stage, resulting in an improved 5-year survival rate. Monoclonal antibody detection, fluorescence bronchoscopy, and low-dose spiral CT increase diagnostic sensitivity and improve the ability to localize early-stage lesions. Utilizing these new techniques and improving the definition of high-risk groups may improve the success and cost-effectiveness of early detection based on sputum cytology. The ultimate goal of improving longterm survival in lung cancer will be achieved only when cancer can be detected in its early stages and lesions can be localized in large numbers. Advances in the last 15 years offer an encouraging vision for the value of early detection and effective treatment for lung cancer. (CHEST 2000; 117:72S–79S) Key words: bronchoscopy; lung cancer; monoclonal antibodies; positron emission tomography; screening; sputum cytology; survival Abbreviations: MoAb ⫽ monoclonal antibody; MSKCC ⫽ Memorial Sloan-Kettering Cancer Center; NCI ⫽ National Cancer Institute; PET ⫽ positron emission tomography; SPORE ⫽ Specialized Program of Research Excellence
the end of the 20th century, few physicians routinely A torder chest radiographs or collect sputum for cytology
examination in the hope of discovering lung cancer in a high-risk patient at an early stage. Effective early detection of disease requires that three essential criteria be met: (1) there must be a preclinical phase of the disease, and (2) technology to detect the disease in the preclinical stage must be available, which (3) enables effective interventions when the disease is discovered within the required window of opportunity. There is ample evidence of a prolonged preclinical phase in lung cancer. Clones of endobronchial cell popu-
*From the University of Colorado Health Science Center, Division of Pulmonary and Critical Care Medicine, Lung Cancer Institute of Colorado, Denver, CO. Supported by National Cancer Institute Grant P50 CA58187. Correspondence to: Timothy C. Kennedy, MD, FCCP, 1721 East 19th Ave, #366, Denver, CO 80218 72S
lations accumulate genetic mutations leading to a progressively more malignant and ultimately invasive malignant state. They may only roughly reflect the Saccommano morphologic criteria (normal cells; hyperplasia; metaplasia; mild, moderate, and severe dysplasia; carcinoma in situ; and invasive carcinoma) as they evolve into malignancy. Neither the critical number of mutations, critical combinations of mutations, or a necessary order of events (if such exists) is known at this time. Elucidating these issues is one of the most exciting areas of current research because of the potential prospect that such understanding may lead to better detection and interventions. Research, however, currently suffers from a lack of scientific surveillance in high-risk populations. It is rather amazing that in this country there is no systematic surveillance of individuals at high risk for lung cancer. If there were, it might aid in the development of new detection technologies, identification of key risk factors, and/or the validation of interventions for precancer or early cancer. In fact, half of the new cases of lung cancer occur in ex-smokers.1 More women die annually of lung cancer than breast cancer, but lung cancer is not considered a woman’s issue. Far more Americans die of lung cancer than of AIDS, but lung cancer victims are not viewed as victims and are underrepresented by advocacy groups. Despite an increase in the number of new cases (171,600/yr), the large numbers of current and ex-smokers at risk in the United States (50 million each), an unchanging 72% mortality rate among lung cancer patients,2 and its position as a predominant cause of death among Americans (6%), there is no strategy and, in fact, little research funding for this epidemic. There are accepted screening strategies for almost all other common solid tumors, including colon, breast, and prostate cancers. However, it is curious that, with some exceptions, few of these screening strategies have been subjected to large, carefully controlled, randomized studies. The data on mammograms remain mixed and controversial in women ⬍50 years old. Fiberoptic colorectal endoscopy screening has not been proven to be of benefit in asymptomatic, occult blood-negative individuals. The benefit of occult blood testing itself remains unsettled, in spite of the Minnesota Colon Cancer Control Study results. The utility of the prostate specific antigen as a screening tool is particularly unresolved, heavily criticized for leading to overdiagnosis. Only the cervical Papanicolaou smear has been clearly shown to reduce case mortality (by 90%). Complicating the issue further, three large, ambitious, randomized, National Cancer Institute (NCI) studies3 from the 1970s and 1980s conducted at Mayo Clinic, Johns Hopkins Oncology Center, and Memorial Sloan-Kettering Cancer Center (MSKCC), as well as a fourth study from Czechoslovakia,4 failed to demonstrate a disease-specific mortality benefit from screening smokers for lung cancer. As a result, the NCI and the American Cancer Society make no recommendation for screening patients for this most important cancer epidemic. Therefore, except for smoking cessation efforts—which offer no benefit to the 50 million Americans who have stopped smoking but Multimodality Approach to Lung Cancer
remain at increased risk for lung cancer—there is no public health strategy for early detection, intervention, or prevention of this disease.
Screening for Lung Cancer There is no question that the earlier lung cancer is discovered, the better are the patient’s chances of survival (Fig 1).5,6 Patients with radiograph-documented stage I lung cancer have a 40 to 80% 5-year survival rate, whether discovered by screening or accident. However, mortality in lung cancer worsens radically with advancing stage at the time of diagnosis. Interestingly, radiologically negative lung cancers discovered by sputum cytology alone in the three NCI studies were associated with an 85 to 90% 5-year survival, even though such patients were few and the histology was usually squamous cell carcinoma. More recently, both Bechtel et al7 and Saito et al8 demonstrated that patients with positive sputum cytology but who were radiologically negative for cancer had a very high survival rate. Squamous cell cancers comprise only 29% of all lung cancer cases.2 Based on this, mortality rates due to squamous cell carcinoma of the lung (n ⫽ 46,081 in 1999) are similar to those of the next most common solid tumors (colon [n ⫽ 47,900 in 1999] or breast cancers [n ⫽ 43,700 in 1999]). These statistics suggest that curing just squamous cell lung cancer would be similar to curing colon or breast cancer. Thus, the significant potential of sputum cytology as an early detection tools warrants reconsidering its role in selected patients. In the NCI randomized studies conducted at MSKCC and at the Johns Hopkins Oncology Center, there was no benefit observed from adding sputum cytology to annual chest radiographs in screening regimens. A low number of cancers were discovered by sputum cytology among subjects randomized to dual screening with annual chest
radiographs and sputum cytology every 4 months vs subjects undergoing a single screen with only an annual chest radiograph. Retrospectively, this may have been due in part to the relatively low risk for lung cancer in the study population. Interestingly, Tockman et al9 used the archived sputa from the group of subjects who eventually developed lung cancer, following moderate dysplasia in sputum to demonstrate the greater sensitivity and potentially earlier diagnosis gained from using monoclonal antibody (MoAb). All three NCI studies demonstrated that periodic screening results in a significant shift toward earlier diagnosis and markedly increased rates of resectability. There is also lengthened survival from the time of diagnosis and improved case fatality. However, there was no disease-specific reduction in mortality: the “gold standard” of efficacy in screening for early fatal disease.
Mayo Clinic Study In the Mayo Clinic study, 10,933 participants underwent an initial prevalence screening with chest radiograph and sputum cytology; 4,618 subjects were randomized to receive chest radiographs and sputum cytology every 4 months, while 4,593 were assigned to a control group advised to have chest radiographs and sputum cytology evaluations annually, since this represented the standard of care at the time. The enrolled subjects (adult males ⬎ 45 years old who had smoked at least 20 cigarettes daily within the year prior to enrollment) underwent 6 years of screening and 3 additional years of follow-up observation. Through the follow-up period, 206 cases of lung cancer were found in the screened group and 160 cases in the control group. There were 122 lung cancer deaths in the screened group and 115 in the control group. There was a shift toward lower stages and better resectability in the screened group. The 5-year survival rate of lung cancer
Figure 1. Survival rates for 2,382 patients after resection of lung cancer, classified by postoperative stage. Differences between groups: stage I vs stage II, p ⬍ 0.01; stage II vs stage IIIA, p ⬍ 0.01; stage IIIA vs stage IIIB, p ⬍ 0.01; stage IIIB vs stage IV, not significant. Reprinted with permission from Naruke et al.5 CHEST / 117 / 4 / APRIL, 2000 SUPPLEMENT
73S
cases in the screened group was much higher than in the control group (33% vs 15%). However, because there were more lung cancer deaths in the screened group than in the control group, it was concluded that there was no advantage for screening. Strictly speaking, this study suggests screening increases the mortality rate. The reason there were more lung cancer deaths in the screened group clearly was that there were more lung cancers in this group, although the reason for this disparity remains a source of dispute. There are a number of valid criticisms of the Mayo Clinic screening study,10 –15 including the following: 1. There were both significant control contamination and screening noncompliance. For instance, 50% of the control group had chest radiographs during the last year of the study, and 73% in the last 2 years of study. On the other hand, 25% of the “screened” group were not compliant with the screening recommendations. 2. It is now accepted that the inclusion criteria provided for a low statistical power. A large number of the participants had a low pack-year smoking history, while other high-risk factors (airflow obstruction, genetic predisposition, occupational exposures) were not considered in the entrance criteria. Thus, the study design provided only a 20% chance of finding a 10% reduction in case mortality. If it had shown a 10% mortality reduction, the low power would have meant daunting screening costs to save 17,800 Americans out of the 100 million smokers and ex-smokers in 1997. One cannot emphasize enough the importance of better defining the risk factors in lung cancer. 3. It appears that the study population was not followed long enough. Major considerations of the counterintuitive results of the study have been lead time bias, length-time bias, and overdiagnosis bias. These issues are common to all cancer screening trials, not just those in lung cancer. The high survival rate among patients with cancer diagnosed by sputum cytology has been discounted based on the concept that these central endobronchial tumors were indolent squamous cell carcinomas with low intrinsic mortality, perhaps so low that they would not have caused significant mortality if undiagnosed and untreated. Overdiagnosis bias, which Eddy16 has suggested is the most likely explanation for the study results, could have been resolved if, with longer follow-up, the incidence cases of the control population caught up with the screened cases. This did not occur within the time frame of the study. Unlike the prostate carcinoma story, however, autopsy evidence by McFarlane et al12 does not suggest a significant incidence of clinically low-level, undiagnosed lung carcinoma. Sobue et al,14,15 as part of a Japanese case-control lung cancer screening trial, also concluded that overdiagnosis bias was an unlikely explanation for the Mayo Clinic results. These authors observed equivalent survival among patients who refused surgical excision for screened vs 74S
symptom-detected cases of stage I lung cancer during a 10-year follow-up period. These cases emphasized radiologic detection in all but a few cases that were sputum detected. These studies lend little support for a significant prevalence of benign lung cancer. The limitations of the technology at that time certainly influenced outcome. After the NCI studies, on reviewing negative chest radiographs from patients discovered to have lung cancer on later radiographs, scientists were able to retrospectively identify the smaller lesion on the earlier radiograph.10 Most lesions were seen on posteroanterior views, and 5% were seen on lateral views only. This suggests insufficient diagnostic lead time in many cases. Some sputum-discovered carcinomas could not be localized on initial bronchoscopy, potentially adding to mortality. In some cases, it appears that the lesion initially was subtle; in others, the lesion was distal to the larger airways accessible by the fiberoptic bronchoscope used at the time of the study (usually 5.4-mm or 6.2-mm outer diameter). Although many tumors are in the large airways, better access to the subsegmental airways would considerably extend the ability to localize some other tumors suggested on sputum cytology. Modern smaller diameter bronchoscopes also may improve lead time.
Czechoslovakian Study Like the Mayo Clinic study, the Czechoslovakian study began with a prevalence screen and randomization to a screened or a control group. The control group was not instructed to obtain a chest radiograph or sputum testing, and thus may have provided a better control than the group in the Mayo Clinic trial. The screening period was only 3 years, and the follow-up period was 3 years, during which time both groups had annual chest radiographs. Interestingly, the incidence of cancer continued to increase in the screened group during the follow-up period, which suggests that overdiagnosis bias is not a compelling explanation for the larger number of lung cancer cases in the screened group (n ⫽ 206) vs the control group (n ⫽ 160) out a total of 6,364 enrollees.4,13
Johns Hopkins and MSKCC Studies It is important that the 5-year survival rate of cancer patients in both the control and the screened groups in the Johns Hopkins and MSKCC studies was nearly 35%, well above the national average of 12 to 13% at the time of the studies. The screened group in the Mayo Clinic study had a similar 5-year mortality rate. Unfortunately, experts and physicians in general in this country still feel that screening high-risk groups is unwarranted, which is not what the data suggest. In fact, because the control group appeared to follow their instructions to get chest radiographs to a substantial degree, and the screened group was somewhat noncompliant, the study, in a sense, compared a very compliant group to a somewhat compliant group, yielding improved diagnostic capacity.
Where Do We Go From Here? Some professionals have argued that recommending lung cancer screening would undercut the impact of Multimodality Approach to Lung Cancer
smoking cessation efforts. However, because current evidence suggests that half of lung cancers occur in exsmokers, ignoring these individuals who have successfully overcome addiction as recommended would seem callous. The incidence of new cases of lung cancer is currently increasing at a time when the prevalence of smoking is decreasing. Because none of these studies addressed a particular high-risk group, the cost per discovery of each lung cancer case was high, discouraging interest in both research and clinical case finding. High costs are particularly counterproductive in the current managed care era of capitated primary care physicians and scarce research dollars. Nonetheless, the NCI has underway a large randomized trial known as the Prostate, Lung, Colon, Ovary Trial, in which 148,000 men and women aged 60 to 74 years will undergo an initial chest radiograph followed by yearly radiographs for 3 years; the control group will not undergo chest radiographs. Unfortunately, no effort is being made to balance the groups for smoking histories or to enroll high-risk groups, thus repeating the problems from the previous randomized studies. Therefore, the design may be ill-suited to the task of addressing the important questions concerning lung cancer, and the study runs the risk of providing yet another expensive trial attempting to resolve the issues but incapable of doing so.
Recent Advances Several recent advances in understanding the biology and genetics of the carcinogenic process have renewed hope that an early detection strategy may be useful.17,18
MoAb in Sputum Tockman et al9 suggested that diagnostic lead time might be extended by adding a specialized procedure as an adjunct to sputum cytology (ie, using two MoAbs directed at a difucosylated Lewis X epitope and a 31-kd protein). Using archived sputum specimens from subjects who developed lung cancer after documented moderate dysplasia during the Johns Hopkins study, the investigators used this procedure to demonstrate positive staining on smears as much as 24 to 48 months prior to the diagnosis of cancer using Saccommano cytology criteria. Studies are underway to prospectively validate this technique.
Genetic Advances An increasing number of important mutations have been identified in lung cancer, including allelic deletions or tumor suppressor gene inactivation in 3p, 5q, 9p, 11q, 17p, 13q, 18q, and 22q. At least three alleles appear important on 3p, with more extensive loss commonly seen in small cell lung cancer and less extensive loss associated with non-small cell lung cancer. p53 alterations appear to be nearly universal in solid tumors, including lung cancers.19 A functional relationship in cell-cycle control between Rb and P16 has been described,20 in which an abnormality in either appears to be associated with any lung malignancy. Array chip technology promises to accelerate the process of identifying important mutations and
critical patterns of mutations. Thus, elucidating molecularbiological events is key to improving the sensitivity of sputum evaluations and may lead to effective chemoprevention strategies, as well as improved treatment of systemic disease.
Fluorescence Bronchoscopy Lam et al21 introduced fluorescence bronchoscopy that is effective without the use of protoporphyrins for the purpose of localizing endobronchial lesions of moderate dysplasia or more malignant tissue. In patients with dysplasia or malignancy in sputum, or in other patients likely to have a high prevalence of endobronchial dysplasia in central airways, such as occupational exposures or previously successful resected lung cancer, this technology may offer increased diagnostic sensitivity using a relatively benign procedure in properly selected patients.
Spiral CT of the Chest Improved chest imaging also may provide increased sensitivity and improved specificity for localizing earlystage lesions. Kaneko et al22 recently screened 1,369 subjects with at least 20 pack-year smoking histories using spiral CT chest examinations. They discovered 15 cancers (14 stage I), including 11 not detectable by plain chest radiographs. Spiral CT imaging takes 15 to 30 s, allowing complete chest imaging in one breath-hold. In addition, it has the radiation exposure of a mammogram and can pinpoint lesions as small as 2 to 3 mm in size. These newer, more sensitive technologies have led some scientists to raise valid concern that overidentification of benign lesions may lead to possible overtreatment morbidity, similar to that experienced with early CT scans. Care must be used to resolve this issue. Shimizu et al23 have suggested using helical CT imaging to improve anatomic distinctions between nodules, vessels, bronchi, and chest wall over conventional CT imaging; this may reduce the incidence of false-negative malignant diagnoses. Bronchoscopy and positron emission tomography (PET) scanning are quite benign and viable approaches to distinguishing between benign and malignant lesions preoperatively, with transthoracic needle aspiration considered less so because of a significant rate of pneumothoraces. A period of follow-up evaluation CT scans have been used successfully to reduce inappropriate surgery. Improved imaging also serves a critical role in addressing the increasing incidence of adenocarcinoma (often arising peripherally), as well as in shrinking the “silent” area of the lung (ie, the subsegmental airways not accessible to the bronchoscope). Thus, although 80% of lung cancer cases in the NCI studies were discovered by chest radiographs, this older technique may be supplanted by newer procedures in those higher-risk groups in whom the enhanced value of sputum cytology has been demonstrated.
Current Studies Sputum in a High-Risk Group Based on data suggesting that airflow obstruction is a risk factor for lung cancer independent of smoking history, CHEST / 117 / 4 / APRIL, 2000 SUPPLEMENT
75S
investigators from the University of Colorado Specialized Program of Research Excellence (SPORE) evaluated sputum cytology from 533 subjects with ⱖ 40 pack-years of smoking and COPD. Enrolled patients were negative for malignancy in the last year based on chest radiograph, but 27.5% had moderate dysplasia or worse (17.1-fold greater than in the Mayo Clinic study; Table 1).24 This included invasive carcinoma or carcinoma in situ in 1.7% of subjects, severe dysplasia in 0.8%, and moderate dysplasia in 25%. These patients required additional clinical investigation. Preliminary analysis suggests this population will have an lung cancer incidence of ⬎ 1%/yr without systematic use of radiographs. Accruals in this study have reached 1,850 subjects with similar distribution of cytology stage. The study will be repositioned for the future as an ongoing incidence study; its goal will be to further the understanding of the molecular biology of preneoplasia and early malignancy, thereby providing a paradigm for intervention and a rationale for earlier detection and treatment. It is hoped that by better defining risk variables for this population, it may be possible to identify a suitable target population for screening or aggressive case finding in the future. Because the prevalence of dysplasia did not increase between the group of patients with a 40 –pack-year habit and those with a 120 –pack-year history, current enrollment includes subjects with ⱖ 30 pack-years. We have observed that this population has an inherently high mortality rate from all causes. Thus, although this group is acceptable for evaluating the biology of premalignancy, the ideal group for cost-effectiveness screening may be different and will perhaps exclude patients with lung impairment and/or comorbidities that would limit the benefit of early diagnosis.
Fluorescence Bronchoscopy A multicenter study including University of Colorado SPORE investigators was completed evaluating the sensi-
Table 1—Cytological Diagnosis of Sputum Samples Collected From Participants Enrolled in the Colorado SPORE Sputum Screening Program*† Diagnosis‡ 0 1 2 3 4 5 6 Total
Ex-Smoker
Current Smoker
Total
47 (8.8) 34 (6.4) 185 (34.7) 95 (17.8) 4 (0.8) 2 (0.4) 2 (0.4) 369 (69.2)
21 (3.9) 27 (5.1) 73 (13.7) 38 (7.2) ⫺ (⫺) 3 (0.5) 2 (0.4) 164 (30.8)
68 (12.8) 61 (11.4) 258 (48.4) 133 (25.0) 4 (0.8) 5 (0.9) 4 (0.8) 533 (100)
*Data are presented as No. (%). A total of 533 participants provided sputum samples deemed adequate for cytopathologic diagnosis (369 ex-smokers and 164 current smokers). These represent the relative frequency of these cytodiagnoses overall and by smoking status. †From Kennedy et al.24 ‡0, no significant abnormalities; 1, regular metaplasia; 2, mild dysplasia; 3, moderate dysplasia; 4, severe (marked) dysplasia; 5, carcinoma in situ; 6, invasive carcinoma. 76S
tivity of the fluorescence bronchoscope in localizing premalignant and malignant endobronchial lesions.25 The bronchoscope provided 271% increased relative sensitivity in detecting lesions when added to white light examination, and even greater sensitivity when obvious, large malignancies were excluded (Fig 2). Interestingly, a University of Colorado SPORE subgroup of patients who underwent bronchoscopy only because of moderate dysplasia in sputum revealed a small but significant prevalence of undetected malignancy (4 of 43 patients), suggesting that such patients may deserve bronchoscopic evaluation on clinical grounds. Our relative sensitivity in localizing significant lesions (moderate dysplasia, severe dysplasia, or carcinoma in situ) in subjects with moderate dysplasia on sputum was 5.75 in favor of fluorescence examination. A subsequent two-bronchoscopist trial, in which each operator is blinded as to the findings of the other and patients are randomly assigned to white light vs fluorescence examination and to a specific operator, is underway at the University of Colorado SPORE. Our preliminary finding of a threefold increase in sensitivity with the fluorescence light system among the first 31 subjects, is similar to our earlier finding.26 It appears that the increased sensitivity is associated with decreased specificity, resulting in many false-positive biopsies. This adds to the expense of analyzing more biopsies and increases the time of the procedure. If there is logic to sputum cytology screening of heavy smokers with airflow obstruction, and investigating those patients with abnormal sputum cytology with fluorescence bronchoscopy, it is to discover a preponderance of squamous cell carcinomas located in the central airways. Because of underlying respiratory impairment, treatment strategies in such cases would likely emphasize local treatment and lung-sparing tactics. A potentially more important use of sputum cytology followed by fluorescence examination is the potential inherent in localizing premalignant lesions in order to treat them prior to becoming invasive.
Implications for Case Finding The very high prevalence of malignancy in our study population leads to intriguing implications for case finding of early malignancies. If 1.7% of the population has signs of carcinoma on sputum culture, as we have demonstrated, and another 2.2% (9% [4 of 43] ⫻ 25%) are discovered after fluorescence bronchoscopy because of moderate dysplasia, and another 1% are discovered by spiral CT in a peripheral location, the potential for early-stage diagnosis arguably exceeds that available with virtually any other non-lung cancer screening program! What other cancer screening program could provide 4.9% discovery rate on a prevalence basis? We emphasize that this argument depends entirely on narrowing the case finding to higher-risk patients.
Future Challenges Future challenges and areas of research in lung cancer screening include the following: Multimodality Approach to Lung Cancer
Figure 2. White-light bronchoscopy image and fluorescence image of the same area, demonstrating identification of (top) normal tissue in the right upper lobe or (middle and bottom) malignant tissue in the right bronchus. Used with permission from Xillix Technologies, Vancouver, British Columbia.
CHEST / 117 / 4 / APRIL, 2000 SUPPLEMENT
77S
1. Advancing technology to improve sputum sensitivity, lead time to diagnosis, and specificity. Monolayering of sputum cytology preparation, improved preservatives and staining, and automated/quantitative evaluation of sputum all have realistic promise to achieve this when applied to the appropriate risk group. 2. Improving definition of risk factors for lung cancer, particularly those useful for narrowing the risk group worth screening. 3. Improving techniques in localizing endobronchial lesions. Next-generation fluorescence bronchoscopy likely will be more sensitive, easier to incorporate into a bronchoscopic examination, and, when used with smaller-diameter bronchoscopes, able to localize more distal lesions. 4. Validating spiral CT chest examinations as an inexpensive, safe, fast, sensitive, and specific screening tool. PET scans, though expensive, may play an important adjunct role for secondary evaluations. The cost (and availability) of PET scans may be more realistic when used at higher volumes. 5. Developing and validating local therapies. Since the University of Colorado Lung Cancer SPORE highrisk population commonly is afflicted with comorbidities in addition to respiratory impairment, establishing acceptable local therapies is needed. Currently available are high-dose brachytherapy, photodynamic therapy with an increasing number of agents including the recently US Food and Drug Administration-approved protoporphyrin, unipolar electrocauterization, cryotherapy, and Nd-YAG laser therapy. Currently, there is no proven answer to the questions, “How should one best treat severe dysplasia?” or “How should one best treat carcinoma in situ?” These questions will only be addressed when lesions can be localized in substantial numbers, and the natural history of such lesions are better understood. The prospect of working on such questions when this arena of pulmonary medicine evolves from its current stagnation is extremely exciting. In the last 15 years, many advances have renewed hope that early detection of lung cancer is possible and may be useful. These include better definition of risk groups, better chest imaging, better localization of endobronchial lesions, and better local treatment modalities. Studies have demonstrated low disease-specific mortality when the disease is discovered by sputum cytology. What is needed now is another major push at the problem using the advances of recent years. The current understanding of lung cancer risk factors allows us to better target the relevant population, but more is needed in this regard. Furthermore, although imaging techniques, localizing techniques, and local treatment strategies clearly are better, these advances will stagnate in the absence of well-designed population-based studies. Trials must be designed to retest the premise that aggressive seeking of early lung cancer will reduce mortality, saving many victims who would otherwise perish, perhaps unnecessarily. 78S
ACKNOWLEDGMENTS: The author acknowledges colleagues Susan Proudfoot, MPH; Wilbur Franklin, MD; and Paul Bunn, MD for reviewing this manuscript and research work done in conjunction with the University of Colorado Lung Cancer SPORE.
References 1 Strauss G, Decamp M, DiBiccaro E, et al. Lung cancer diagnosis is being made with increasing frequency in former cigarette smokers [abstract 1106]. Proc Am Soc Clin Oncol 1995; 14:362 2 American Cancer Society. Cancer Facts and Figures, 1999. Atlanta, GA: American Cancer Society, 1999 3 Melamed MR, Flehinger BJ, Zaman MB. Impact of early detection on the clinical course of lung cancer. Surg Clin North Am 1987; 67:909 –924 4 Kubik A, Polak J. Lack of benefit from semi-annual screening for cancer of the lung: follow-up report of a randomized controlled trial on population of high-risk males in Czechoslovakia. Int J Cancer 1990; 45:26 –33 5 Naruke T, Tsuchiya R, Kondo H, et al. Implications of staging in lung cancer. Chest 1997; 112(suppl)4:242S–248S 6 Mountain CF, Dresler CM. Regional lymph node classification for lung cancer staging. Chest 1997; 111:1718 –1723 7 Bechtel JJ, Kelley WR, Petty TL, et al. Outcome of 51 patients with roentgenographically occult lung cancer detected by sputum cytology testing. Arch Intern Med 1994; 154:975–970 8 Saito Y, Nagamota N, Shin-Ichiron O, et al. Results of surgical treatment from roentgenographically occult bronchogenic squamous cell carcinoma. J Thorac Cardiovasc Surg 1992; 104:401– 407 9 Tockman MS, Gupta PK, Myers JD, et al. Sensitive and specific monoclonal antibody recognition of human lung cancer antigen on preserved sputum cells: a new approach to early lung cancer detection. J Clin Oncol 1988; 6:1685–1693 10 Strauss GM. Measuring effectiveness of lung cancer screening: from consensus to controversy and back. Chest 1997; 112(suppl 4):216S–228S 11 Strauss GM, Gleason RE, Sugarbush DJ, et al. Screening for lung cancer: another look; a different view. Chest 1997; 111:754 –768 12 McFarlane MJ, Feinsein AR, Wells CK. Clinical features of lung cancers discovered as a postmortem “surprise”. Chest 1986; 90:520 –523 13 Prorok PC, Chamberlain J, Day NE, et al. UICC workshop on the evaluation of screening programes for cancer. Int J Cancer 1984; 34:1– 4 14 Sobue T, Suziki T, Matsuda M, et al. Survival for clinical stage 1 lung cancer not surgically treated. Cancer 1992; 69:685– 692 15 Sobue T, Suzuki T, Naruke T, et al. Efficacy of lung cancer screening; comparison of results from a case-control study and a survival analysis. Jpn J Cancer Res 1992; 83:424 – 430 16 Eddy D. Screening for lung cancer. Ann Intern Med 1989; 111:389 –399 17 Bunn P, Carbone D, Grim J, et al. Fifth I.A.S.L.C. lung tumor biology workshop, August 13–17, 1996, Ermatingen, Switzerland. Ann Oncol 1997; 8:27–36 18 Miller YE, Franklin WA. Molecular events in lung carcinogenesis. Hematol Oncol Clin North Am 1997; 11:215–234 19 Weintraub SJ. Inactivation of tumor suppressor proteins in lung cancer. Am J Respir Cell Mol Biol 1996; 15:150 –155 20 Boers JE, Ten Velde GPM, Thunnissen BJM. p53 in squamous metaplasia: a marker for risk of respiratory tract carcinoma. Am J Respir Crit Care Med 1996; 153:411– 416 21 Lam S, MacAulay C, Hung J, et al. Detection and localization of early lung cancer by imaging techniques. J Thorac CardioMultimodality Approach to Lung Cancer
vasc Surg 1991; 105:1035–1040 22 Kaneko M, Eguchi K, Ohmatsu H, et al. Peripheral lung cancer: screening and detection with low dose spiral CT versus radiography. Radiology 1996; 201:798 – 802 23 Shimizu T, Narabayash I, Yesugi Y, et al. Three-dimensional display of pulmonary nodules using helical CT. Diagn Radiol 1995; 13:209 –216 24 Kennedy T, Proudfoot S, Franklin W, et al. Cytopathological analysis of sputum in patients with airflow obstruction and
significant smoking histories. Cancer Res 1996; 56:4673– 4678 25 Lam S, Kennedy T, Unger M, et al. Localization of bronchial intraepithelial neoplastic lesions by fluorescence endoscopy. Chest 1998; 113:696 –702 26 Prindiville S, Miller Y, Cook R, et al. Fluorescence bronchoscopy compared to white light bronchoscopy in the detection of high grade bronchial dysplasia [abstract]. Proc Am Assoc Cancer Res 1998; 39:117
CHEST / 117 / 4 / APRIL, 2000 SUPPLEMENT
79S