Screening studies for lung cancer date to the 1950s and 1960s when several studies were undertaken using a variety of screening protocols that combined chest radiography and sputum analysis. The protocols employed different screening time intervals and the study design was either uncontrolled or controlled but nonrandomized. The most widely publicized study was the Philadelphia Pulmonary Neoplasm Research Project, in which only 6 of 94 patients with lung cancer detected at screening survived more than 5 years . No study showed an advantage for lung cancer screening.
The subsequent development of more sophisticated techniques of chest radiography and sputum analysis in the 1960s and the methodologic limitations of the early studies led to the concept that lung cancer screening might prove efficacious if a more rigorous study design was used. In that context, three large randomized controlled studies (National Cancer Institute Cooperative Early Lung Cancer Group) were initiated among male smokers in the 1970s at the Mayo Clinic, Memorial Sloan-Kettering Cancer Center, and the Johns Hopkins Medical Institutions.
In the Mayo Lung Project, 10,933 men who were 45 years of age or older and who smoked more than a pack of cigarettes daily were assessed with chest radiographs and sputum cytology . Lung cancers found in these patients were designated as ''prevalence cases.'' The 9,211 men with negative chest radiographs and sputum cytology were randomized into two groups. The control group of 4,593 patients was given the standard Mayo Clinic recommendation at that time, a yearly chest radiograph and sputum cytologic examination, but no individualized follow-up was pursued. The study group of 4,618 patients was scheduled once every four months for a chest radiograph and a sputum container was sent to collect a 3-day pooled sputum sample, which was returned to the Mayo Clinic. All patients were contacted yearly to assess their status. Approximately 75% of men in the study group complied with the every-4-month protocol.
When the study ended in 1983, lung cancer had been detected in 206
patients in the study group and 160 patients in the control group . Resect-ability was higher in the study group than in the control group (46% vs. 32%) but this advantage was not reflected in mortality rates. The death rates in the two groups were statistically similar: 3.2 per 1000 person-years in the study group compared to 3.0 per 1000 person-years in the control group.
A closer analysis of the data reveals that the every-4-month screening protocol detected a higher proportion of lung cancer at an early stage (42%) than in the control group (25%) and a corresponding 5-year survival benefit was found . However, despite these apparent advantages, no mortality benefit was demonstrated from screening.
Several explanations for the difference between the survival and mortality data have been postulated, including lead-time bias, overdiagnosis, and control-group contamination . Lead-time bias occurs if the lung cancer is detected at an early stage in its natural history but the ultimate time of death is unchanged. In the Mayo Clinic study, the lung cancers in the screened population were detected at an earlier stage than in the control group, resulting in longer survival and apparent 5-year survival benefit. However, assuming the eventual time of death remained unchanged, no mortality benefit would be observed.
Overdiagnosis occurs if cancers that are indolent are disproportionately detected by lung cancer screening. Patients with slow-growing tumors would have a prolonged disease course that would favorably affect 5-year-survival data. Indolent cancers would not be as likely to be detected in the control group because they would remain asymptomatic for an extended period of time and the patient might succumb to other illness. If such indolent cancers were disproportionately found in the screened population, a survival advantage but no mortality benefit would be shown for this population. Length time bias is a related bias but describes detection of indolent cancer over a more limited time frame than overdiagnosis.
Prostate cancer is an example of a disease in which overdiagnosis might occur. Two-thirds of men over 60 years of age that die of other causes have undiagnosed, presumably indolent prostate cancer at autopsy. As for prostate cancer, it was suggested that overdiagnosis of lung cancer might account for the difference between survival and mortality data .
Contamination of the control group may also have been problematic in the Mayo Clinic study . Investigators estimated that approximately 50% of control patients in fact underwent chest radiography during the course of the study and thus took on some characteristics of the screened population (contamination). One-third of the lung cancer in the control group was detected as a result of such ''nonstudy'' chest radiographs.
The Memorial Sloan-Kettering Cancer Center and the Johns Hopkins Medical Institutions studies used similar protocols that were substantively different than that of the Mayo Clinic project [7,8]. These studies were designed to determine any advantage gained by the addition of yearly sputum samples to annual chest radiographs. Because chest radiography was used in both the study and the control populations, the trials were not useful to evaluate the efficacy of annual chest radiographs.
The Memorial Sloan-Kettering Cancer Center study consisted of 10,040 men over 45 years of age that smoked at least one pack of cigarettes . Patients were randomly assigned to two groups. The control group, composed of 5,072 men, underwent annual chest radiography. The screened population of 4,985 received an annual chest radiograph and pooled sputum cytology every 4 months. The 288 lung cancers found in these groups were evenly split between the two groups. There was no significant difference in the opera-bility, 5-year survival, or mortality between the screened and control populations.
The Johns Hopkins study employed a protocol that was nearly identical to that of the Memorial Sloan-Kettering Cancer Center study . Similar numbers of lung cancers were detected in the screened and control populations and the survival and mortality data were not significantly different in the two groups. Thus, neither study demonstrated an advantage for annual screening sputum cytology.
Two other studies have been reported from Europe that assessed the screening potential of chest radiography for lung cancer [9-11]. A randomized controlled study from Czechoslovakia reported in the mid-1980s evaluated 6364 male smokers between the ages of 40 and 64 [9,10]. Both screened and control groups were followed over a 3-year period. The screened group (n = 3172) underwent both a chest radiograph and sputum cytology every 6 months for the duration of the study. The control group (n = 3174) received only a chest radiograph at the end of the 3-year period. Thirty-nine cancers were detected in the screened group compared to 27 in the control group. However, no clear advantage in mortality was demonstrated in the screened group.
A case-controlled study reported from Germany in the late 1980s assessed the rate of lung cancer detection among patients who had undergone biannual surveillance for tuberculosis with chest radiography . The 130 men in this screening program who died of lung cancer were matched with an aged-match control group consisting of men from the same district. No mortality benefit was found for the patients who participated in the screening program.
The failure of these studies to demonstrate a mortality advantage for lung cancer screening with either chest radiography or sputum cytology led most organizations to recommend that routine screening not be undertaken. The American Cancer Society noted that ''the Society did not feel it would be responsible to advocate screening for a large group of people . . . without better evidence that they would derive some benefit'' . Based on the evidence, the view that ''screening for lung cancer ... is not recommended'' has been widely professed . At present, no major organization advocates routine screening for lung cancer.
Over the past several years, several investigators have proposed a reassessment of the data from the four randomized lung cancer trials [13,14]. Strauss et al. noted that although the Johns Hopkins Institute and Memorial Sloan-Kettering Cancer Center studies were not designed specifically to assess the efficacy of chest radiographic screening, both control and screened populations achieved survival rates approximately three times greater than prevailing norms. They suggested that the use of an annual chest radiographic screening protocol might have led to an improved outcome even if a mortality benefit was not shown .
Strauss et al. also contended that the improved resectability, lower stage, and better survival in the Mayo and Czech studies could not be adequately explained on the basis of lead-time bias or overdiagnosis. With respect to overdiagnosis, they cited evidence from an autopsy study that suggests indolent lung cancer occurs only rarely . Based on evaluation of the Mayo Lung Project data, they also noted that patients in that trial with early lung cancer who either were medically unsuitable or refused surgical resection had a much lower rate of survival than those who underwent surgical treatment (10% vs. 70%) . Strauss et al. suggested that the behavior of lung cancer is almost always aggressive even when detected at an early stage and, thus, overdiagnosis is unlikely to be a substantial confounding factor . In the view of the authors, the increases in survival and other parameters of the screened group in the Mayo and Czech studies could not be explained by study design biases. They speculated that the screening protocol itself might have led to a better outcome. They recommended a reappraisal of the role of chest radiographs in early lung cancer detection.
Other methodologic criticisms of the Mayo Lung Project have been raised. The contamination that occurred because 50% of the patients in the control group underwent chest radiographs decreased the distinctiveness of the screened and control group, leading to greater difficulty in detecting a difference between populations. A larger trial might, therefore, be required to show a difference .
Miettinen has suggested that the nine-year period over which cumulative mortality rate was calculated in the Mayo Lung Project is excessive . He has stated that this period of time likely underestimated the maximum effect of screening. Based on his analysis of the data, the time between 3 and 7 years after completion of prevalence screening is optimal because it ''represents a compromise between one that is narrow enough to address the full effect, and one that is wide enough to show a meaningful number of deaths from the disease.'' He believes that if the more appropriate timing of cumulative mortality rates is used, the Mayo Lung data cannot be interpreted as providing direct evidence against screening.
The lingering questions with respect to the major lung cancer screening trials of the 1970s and 1980s in combination with the development of potent new imaging and nonimaging techniques engendered renewed interest in lung cancer screening throughout the 1990s. The remainder of this chapter describes early results using these newer technologies.
Was this article helpful?
Stop Nicotine Addiction Is Not Easy, But You Can Do It. Discover How To Have The Best Chance Of Quitting Nicotine And Dramatically Improve Your Quality Of Your Life Today. Finally You Can Fully Equip Yourself With These Must know Blue Print To Stop Nicotine Addiction And Live An Exciting Life You Deserve!