Although cigarette smoking has long been established as the predominant cause of lung cancer, other risk factors have been identified, some acting to modify the lung cancer risks of smokers. Among these other factors are the presence of chronic lung diseases, including chronic obstructive pulmonary disease and fibrotic disorders—asbestosis, silicosis, and idiopathic pulmonary fibrosis, and a past history of tuberculosis have also been linked to increased lung cancer risk in some studies (1). The hypothesis that diffuse fibrotic disorders of the lung are associated with increased lung cancer risk arose from clinical observations of the simultaneous finding at autopsy of lung cancer and asbestos (2) and of lung cancer and pulmonary fibrosis (3, 4). Epidemiologic studies involving follow-up of asbestos workers and persons with IPF have generally confirmed this hypothesis. In silica-exposed workers, those most at risk for lung cancer in association with this exposure have established silicosis (5) and the presence of asbestosis is an indicator of increased risk as well. Increased lung cancer risk has also been described in persons with collagen vascular diseases causing pulmonary fibrosis, particularly progressive systemic scleredema. The concept of “scar carcinomas” has also been advanced, which attributes the origin of peripheral carcinomas to adjacent areas of scarring (6, 7), although some have questioned whether the cancer induces the apparent fibrosis (8).

The association between fibrosis and lung cancer risk immediately raises intriguing questions: are the locations and histologic types of lung cancer different in persons with and without fibrosis?; does fibrosis have a synergistic effect with smoking?; and what are the underlying mechanisms? There are also potential clinical implications related to the identification of a high risk population that would be in need of effective screening. In persons with exposure to occupational agents causing fibrosis, the presence of fibrosis might also be used to apportion the probability of causation of a particular individual's lung cancer between smoking and the occupational agent of concern. In this regard, there has already been a vigorous debate as to whether the presence of asbestosis is a necessary condition for attributing the causation of a lung cancer to asbestos (9, 10).

In this issue of the Journal, Hubbard and colleagues (11) provide the findings from a new study of IPF and lung cancer risk. Using a general practice patient registry in the United Kingdom, these investigators identified patients with a diagnosis of IPF (termed “cryptogenic fibrosing alveolitis” in their report) and a control series of age- and sex-matched patients drawn from the same practices. Follow-up for lung cancer was accomplished by tracking the cohort through the database retrospectively. Some information on smoking was available in the records for the majority of the IPF patients and the controls. The findings were striking; the incidence of lung cancer was increased seven-fold in the IPF group, compared with the controls, and the excess risk changed little after adjustment for cigarette smoking. Comparing nonsmokers and current smokers, the relative risk of lung cancer associated with IPF was twice as high in the nonsmokers.

The design used by Hubbard and colleagues took advantage of the general-practice database, and consequently the investigators were able to establish a large cohort, 890 IPF patients and 5,884 controls. Limitations are inherent, as acknowledged by the investigators. The diagnosis of IPF was generally made without histological confirmation and is undoubtedly subject to some degree of misclassification. The smoking data, retrieved from the practice dataset, are also likely to have been affected by misclassification. In fact, the authors acknowledge that a substantial proportion of those classified as nonsmokers may have been former smokers. The net consequence of the unavoidable error in the diagnosis of IPF is likely to have been in the direction of reducing the magnitude of its association with lung cancer risk. Only substantial misclassification of persons with asbestosis as having IPF would have tipped the net effect of misclassification towards a positive bias. Smoking patterns of the IPF patients and the controls were similar, and any misclassification of smoking probably did not lead to significant bias in this study in assessing the association of IPF with lung cancer risk.

This new study adds to the small number of epidemiologic reports on IPF and lung cancer risk (12-14). Reports from two studies in the United Kingdom provided similar indication of increased risk. In 1972, Stack and colleagues (12) reported on the natural history of 96 IPF patients from the Edinburgh chest clinics. Over an average follow-up of about four years, five cases of lung cancer were recorded. In a similar study of patients at the Brompton Hospital in London, Turner-Warwick and colleagues (13) found lung cancer in 20 of 205 patients and estimated a 14-fold increased risk of lung cancer in comparison with the general population. Only two of the lung cancers occurred in nonsmoking patients with IPF, and the ratio of observed to expected deaths was lower in nonsmokers (2:9) compared with smokers (9:7).

Two recent studies did not confirm an association between IPF and lung cancer risk, although the methods were distinct from those used in the follow-up studies in the United Kingdom, including the new report by Hubbard and coworkers. Wells and Mannino (14) used the multiple cause of death data for the United States, 1979–1991, to describe the joint mention of lung cancer and pulmonary fibrosis on death certificates. They found that lung cancer was mentioned on only 4.8% of death certificates also mentioning pulmonary fibrosis, a figure lower than for all deaths (6.5%). This approach would appear to have some validity, as supported by the finding of a mention of lung cancer on 26.6% of death certificates listing asbestosis. In a study of similar design involving death certificates for England and Wales for 1985–1986, lung cancer was mentioned with approximately equal frequency, 6 to 8%, on death certificates listing pulmonary fibrosis, coal workers' pneumoconiosis, and silicosis. By contrast, lung cancer was included on 43% of death certificates listing asbestosis. The sensitivity of this study design is diminished, however, by the likely negative bias from incomplete listing of pulmonary fibrosis on death certificates.

The relatively limited observational evidence on the association of IPF with lung cancer risk and the conflicting findings of the follow-up and death certificate–based studies offer a mandate for further research. The strongest evidence would be obtained from follow-up of cohorts of clinically well characterized patients, in order to avoid the potential for diagnostic misclassification that is of concern in the death certificate studies and the practice-registry study of Hubbard and colleagues. Relatively modest sample sizes would be needed to detect an increased risk in the range found by Hubbard and coworkers. Sufficiently large series have been assembled from disease registries and referral institutions (15, 16).

Assuming that IPF is associated with increased lung cancer risk, what mechanisms could underlie this linkage? Both direct and indirect causal pathways could be relevant. Occupational and environmental exposures might increase risk for both apparent IPF and lung cancer. Findings of several case– control studies indicate, for example, that metal dusts may increase risk for IPF. Iwai and colleagues (17) in Japan found evidence in an autopsy series and in a case–control study that metal exposures might be associated with risk for IPF. In a case–control study in the United Kingdom, Hubbard and colleagues (18) found that a questionnaire report of metal dust exposure was significantly associated with risk for IPF (odds ratio = 1.68, 95% confidence interval 1.07–2.65). In a multi center case–control study in the United States, we found that several occupational and environmental exposures were associated with IPF, including self-reported metal dust exposure (odds ratio = 2.0, 95% confidence interval 1.0–4.0) (19). Analyses of tissue specimens for metals might be informative, particularly for IPF cases and appropriate controls with occupational histories captured.

Cigarette smoking might also contribute to the association of IPF with lung cancer risk. Smoking itself has been associated with increased risk for IPF in recent case–control studies (18, 20) and also with increased interstitial fibrosis without clinical consequences (21). The occurrence of IPF in smokers might be an indicator of an exceptionally high cumulative dose of tobacco smoke components or of enhanced susceptibility to the pulmonary consequences of smoking. At least one component of past cigarette filters, asbestos, is associated with both pulmonary fibrosis, and lung cancer and fibers are released from current cellulose filters (22). To further explore the role of cigarette smoking, data are needed that provide estimates of the increased lung cancer risk associated with IPF in smokers and nonsmokers separately.

A third potential explanation for the association of IPF with lung cancer lies in the possibility that the diffuse inflammatory process of IPF increases lung cancer risk. Given the generally short natural history of IPF and the multistage process that leads to lung cancer, the presence of IPF would be predicted to affect the later, promotional stages of carcinogenesis. This possibility would merit exploration, if the IPF–lung cancer association can be confirmed.

It is not satisfying to end with the inevitable call for more research, but the findings in the new report by Hubbard and colleagues (11) need replication and, if confirmed, mechanism-oriented research. Further research on the association of lung cancer and IPF might tell us more about the etiology of both of these disorders.

Jonathan M. Samet, M.S., M.D.

Department of Epidemiology

School of Public Health

Johns Hopkins University

Baltimore, Maryland

References

Section:

ChooseTop of pageReferences <<CITING ARTICLES