Prospective Study of Pulmonary Function and Lung Cancer1- 3



SUMMARY The role of pulmonary function as an Independent predictor of lung cancer risk wes studied In a community-based cohort of 6,317 Japanese-American men who were aged 45 to 68 at the time of examination. After a follow-up period of about 22 yr, 172Incident cases of lung cancer were Identified. The percentage of the predicted FEY, weslnversely related to lung cancer (p value for trend = 0.01)after adjustment for age and cigarette smoking history. The subjects In the lowest quartile of pulmonary function (% predicted FEY, < 84.5) had a relative risk of 2.1 (95% confidence Interval = 1.3 to 3.5) for lung cancer compared with subjects In the highest quartile (% predicted FEV, = 103.5+). For the 84 cases with a squamous or small cell histologic type of lung cancer, the subjects In the lowest quartile had a relative risk of 2.5 (95% CI = 1.2 to 5.6) compared with subjects In the highest quartile of pulmonary function. For the 84 patients with lung cancer whose tumors were located Within 4 em of the pulmonary hilum, the subjects In the lowest quartile had a relative risk of 4.0 (95% CI = 1.7 to 9.7}. The results suggest that Impaired pUlmonary function in a community-based population Is a predictor of lung cancer.

Although cigarette smoking is linked to both pulmonary airways obstruction and lung cancer (1, 2), it is suspected that impaired pulmonary function is an independent predictor of lung cancer. Several studies have found that airways obstruction, based on a low FEV 1, increases lung cancer risk even after taking cigarette smoking into account (3-6). Tockman and colleagues observed this effect in a follow-up study of 4,395 subjects who either had chronic obstructive pulmonary disease (COPO) or were at high risk for lung cancer (3). In a report from the Mayo Clinic, Skillrud and coworkers Methods noted that chronic obstructive pulmo- The subjects for this study were men of Japnary disease was a risk factor for lung anese ancestry, born between 1900 and 1919, cancer (4). A prospective study of pa- and residing on the Hawaiian island of Oatients with COPO in New York also hu. They were first identified by the Honolushowed that they had four to five times lu Heart Program in 1965with the use of the the incidence of lung cancer than ciga- comprehensive 1942 Selective Service draft rette smokers or patients with bronchitis registration files. Details of the recruitment (5). In another prospective study of 6,075 of the study subjects have been published elsewhere (8). Of 11,136 identified men, 8,006 cigarette smokers who participated in a (71.90/0) were interviewed and examined from cardiovascular disease intervention trial 1965 to 1968, 180 (1.6%) died before they (multiple risk factor intervention trial, could be examined, and 2,950 (26.5%) did MRFIT), subjects with decreased pulmo- not participate in the program. nary function had increased mortality Spirometry was performed on a water-seal caused by lung cancer (6). recording spirometer (Respiratory VitaloIn contrast to these studies, Peto and meter" Model P-600; Warren E. Collins, Inc., colleagues observed only a weak associ- Boston, MA) with subjects standing and withation of initial FEY 1 loss with lung can- out noseclips. Each man performed three cer mortality in 2,718 British men (7). One spirometric tests at examination. Because reason for the difference in the findings standards for calculating FEV 1 from spirometric curves have changed since pulmonary could be the selection of the study popu- function was measured in this cohort, all lation. The British study was done in men. spirograms were reanalyzed using modified living in the community whose eligibili- American Thoracic Society standards that inty to participate in the study was mainly cluded back-extrapolation, correction for based on their occupation. The other body temperature, pressure, and saturation studies included subjects who primarily with water vapor; and the use of a computerhad significantly impaired ventilatory linked digitizer (9). Spirograms were assessed function or werecigarette smokers at high for technical acceptability using specific obrisk for coronary heart disease. We had .jective criteria (smooth and continuous curve, the opportunity to address this issue by apparent maximal effort, good start, and no evidence of the following: false starts, excesconducting a large prospective study of sivehesitation, coughing, glottis closure, early lung cancer among 6,300 community- . termination, leaks, or obstructed mouthbased subjects who were not selected be- piece). cause of their occupation or because of During the interview, a cigarette smoking COPO. history was obtained. Current smokers record-

AM REV RESPIR DIS 1991; 144:307-311

ed the usual number of cigarettes smoked per day and the age they started smoking. Past smokers recorded the age they started smoking, the maximum number of cigarettes smoked per day, the number of years of smoking the maximum number of cigarettes per day, and the age they stopped smoking. The subsequent incidence of lung cancer after examination among the men was determined by continuous surveillance of all general hospitals on Oahu. To reduce the possibility of missing incident cases during the surveillance period, a computer linkage file was established with the Hawaii Thmor Registry, a member of the Surveillance, Epidemiology,and End Results Program of the National Cancer Institute (10). Only those cases that werenewlydiagnosed with histologically confirmed lung cancer were included in the analysis. Based on a 19-yr follow-up survey of the study subjects since their 1965-1968 examination, it was determined that only 1.3% of the men could not be found on Oahu. As a (Received in original form October 29, 1990 and in revised form January 25, 1991) 1 From the Japan-Hawaii Cancer Study and the Honolulu Heart Program, Honolulu, Hawaii, and the Oregon Health Sciences University School of Medicine, Portland. 2 Supported by Grant No. ROI-CA-33644 from the National Cancer Institute and Contract No. N01HC-02901 from the National Heart, Lung, and Blood Institute. 3 Correspondence and requests for reprints should be addressed to A. Nomura, M.D., JapanHawaii Cancer Study, Kuakini Medical Center, 347 North Kuakini Street, Honolulu, HI 96817.



result, the surveillance for incidence cases of cancer should be nearly complete. The following men were excluded from the analysis: two prevalent cases of lung cancer at time of examination and 35 subjects who subsequently received a clinical diagnosis of lung cancer without histologic confirmation of their diagnosis. Among the remaining 7,969 men, those with a technically unsatisfactory pulmonary function test (n = 1,652) werealso excluded. The men with satisfactory and unsatisfactory function tests were similar in their cigarette smoking history (47 and 43070 current smokers, respectively) and experienced similar lung cancer incidence rates after examination. Specifically, subjects with an acceptable pulmonary function test had an ageadjusted lung cancer incidence rate of 27.4 in 1,000 men compared with a rate of 31.4 in 1,000 (p = 0.39) among men with unsatisfactory tests. After these exclusions, the study population consisted of 6,317 men with satisfactory pulmonary function tests, among whom 172 were subsequently diagnosed with lung cancer by June 1989. The lung cancer cases were classified according to histologic type based on the World Health Organization Classification (11). The slides from each case were reviewed by Stemmermann. The cases with insufficient amounts of tissue for typing were listed as unknown. These cases were diagnosed on the basis of cytology, cell blocks, or fine needle aspirates. Of the 172 cases, there were60 squamous cell, 24 small cell, 54 adenocarcinoma, 5 other, and 29 unknown types. A prediction equation for age- and heightadjusted FEV1 in one second wasderived from the values of healthy, asymptomatic men in this cohort who had never smoked cigarettes (9). The prediction equation was FEV 1 (liters) = -1,845 - 0.018 (age, years) + 0.035 (height, centimeters). The percentage of predicted FEV 1 (% predicted FEV 1) was derived for each man from the value for his observed FEV 1 times 100, divided by his predicted FEV I ' This measure of pulmonary function was used in all analyses. The incidence rates of lung cancer werecalculated with Mantel-Haenszel stratified analysis (12). The Cox proportional hazards regression model (13) and the SAS Procedure PHGLM computer program (SAS Institute, Cary, NC) (14) were used to estimate the relative risk (RR) of lung cancer by quartiles of 070 predicted FEV 1 with adjustment for the effect of age at examination and the detailed history of cigarette smoking. The quartiles were determined from the distribution of 0J0 predicted FEV 1 in the study cohort of 6,317 men. Using the highest quartile category (> 103.5%) of % predicted FEV 1 as baseline, the adjusted RR for each of the remaining quartiles was calculated. The 95 % confidence interval (CI) of the RR was computed to determine whether the RR differed from unity. The dose-response trend was also tested by a method of the likelihood ratio test (p ~ 0.05 was taken as statistically significant). Similar analyses were done for subjects stratified



Lung Cancer Incidence


Mean FEV1 (L)*

Mean % Predicted FEV 1


1,557 2,215 1,237 984 324

2.92 2.78 2.61 2.44 2.32

96 94 92 91 90




Rate t

Current Smokers (%)

Past Smokers (%)

Never Smokers (%)

36 50 41 38 7

23.1 22.6 33.1 38.6 21.6

52 49 44 43 39

24 23 27 27 33

24 28 29 30 28






* Adjusted for height. Unadjusted rates per 1,000.



% Predicted FEV 1 103.5+ 94.5-103.4 84.5-94.4 < 84.5

Covariate Adjusted *

Age Adjusted

No. CaseslNoncases


95% CI



20/1,590 22/1,495 66/1,551 64/1,509

1.0 1.2 3.5 3.7

0.7-2.3 2.1-5.8 2.2-6.0

1.0 1.0 2.5 2.1

0.6-1.9 1.5-4.1 1.3-3.5

P Value for linear trend

< 0.001


* Covariates include age at examination, current smoking status, age started smoking (current smokers), number of cigarettes smoked per day (current smokers), past smoking status, maximum number of cigarettes smoked per day (past smokers), and years of smoking maximum number of cigarettes per day (past smokers).

by both smoking status and % predicted FEV1 and by smoking history alone.


The mean FEV. and the mean percentage of predicted FEV. by 5-yr age groups for the 6,317 men in the study are presented in table 1. The unadjusted lung cancer incidence rates and the smoking status of the subjects are also shown in table 1. The overallmean FEV. was 2.70 L, and the overall mean 070 predicted FEV. was 94070. Both these measurements decreased in a linear fashion with age. The lung cancer incidence rate peaked in the men who were aged 60 to 64 at the time of examination and decreased in the oldest age group. There were more current smokers among the younger men and more past and never smokers among the older men. The relative risks for lung cancer by quartiles of percentage predicted FEV. are presented in table 2. The relative risks are adjusted for age at examination and then for detailed history of cigarette smoking. Statistically significant linear trends are shown for both the ageadjusted and covariate-adjusted relative risks, but there is a decrease in the mag-

nitude of the relative risks when cigarette smoking is taken into account. The analysis wasrepeated by histologic types for the 84 cases with squamousor small-cell cancer and the 54 cases with adenocarcinoma. The results are shown in table 3. The p value for a linear trend was significant only for the squamous and small cell cases; but the differences in the relative risks between the squamous- or small-cell, and adenocarcinoma cases were not great. It was possible to record the location of the tumor for 141 (82%) of the 172 lung cancer cases. Of these, 84 werewithin 4 em of the pulmonary hilum and 57 were located more than 4 em from the pulmonary hilum. Of the central tumors, 55 (65070) were squamous or small cell in type, and 32 (56%) of the peripheral tumors were adenocarcinomas. The data on the association of percentage predicted FEV. with lung cancer risk by tumor location are presented in table 4. The linear trend in risk was statistically significant for patients with centrally located tumors. Patients with tumors located more than 4 cm from the hilum also had an increased risk with diminished pulmonary function, but the association was not significant.


309 TABLE 3


Squamous or Small Cell

0Al Predicted FEV,

103.5+ 94.5-103.4 84.5-94.4 < 84.5

No. of Cases

8 9 34 33

P Value for linear trend


1.0 1.0 3.1 2.5

95 0Al CI

No. of Cases

0.4-2.7 1.4-6.8 1.2-5.6

7 8 21 18


1.0 1.1 2.4 1.7

95% CI 0.4-3.0 1.0-5.6 0.7-4.1



• Covariatesincludeage at examination,current smokingstatus, age started smoking(currentsmokers), number of cigarettessmoked per day (currentsmokers), past smoking status, maximumnumber of cigarettessmokedper day (past smokers),and yearsof smoking maximum numberof cigarettesper day (past smokers).


> 4 em from Pulmonary Hilum

" 4 em from Pulmonary Hilum 0Al Predicted FEV,

103S 94.5-103.4 84.5-94.4 < 84.5 P Value for linear trend

No. of Cases


95% CI

No. of Cases

6 7 33 38

1.0 1.1 4.2 4.0

0.4-3.2 1.7-10.0 1.7-9.7

8 10 21 18

< 0.01


1.0 1.2 2.0 1.6

95% CI 0.5-2.9 0.9-4.6 0.7-3.7


• Covariatesinclude age at examination,currentsmokingstatus, age startedsmoking(currentsmokers), number of cigarettes smoked per day (current smokers), past smoking status, maximumnumber of cigarettessmokedper day (past smokers),and yearsof smoking maximumnumberof cigarettesper day (past smokers).

The relation between 070 predicted FEV 1 and lung cancer was further studied by separating the subjects according to their cigarette smoking history (table 5). Only eight cases of lung cancer occurred among never smokers, and there was no increase in risk with decreasing 070 predicted FEV 1 among them. Past smokers had a linear increase in relative risk with decreasing 070 predicted FEV 1 that was statistically significant (p = 0.03). There was also an inverse trend for the two groups of current smokers, but it was significant only for those who smoked lessthan 56 pack-yr of cigarettes. The relativelysmall numbers of lung cancer cases by histologic types and tumor location precluded repeating the analysis in table 5 by subcategories of lung cancer. Next we looked at the association of cigarette smoking with lung cancer, adjusting for the effects of age and 070 predicted FEV r- As shown in table 6, there was a pronounced increase in risk among current smokers, especially those who had smoked for more than 55 pack-

yr (RR = 21.9). However, adjustment for 070 predicted FEV 1 had minimal effects on the magnitude ofthe relative risks associated with cigarette smoking. Discussion

The subjects for this study were identified in the community and volunteered to participate after they were contacted. As a group, they are healthier and have lived longer than the men who did not participate in the study (15). Among the examinees, 79.3070 (6,317 of 7,969) had satisfactory pulmonary function tests, but the remainder did not. The subjects with and without an acceptable pulmonary function test were similar in their smoking habits and their lung cancer incidence rates, indicating that there was no selection bias in the performance of the test. If impaired pulmonary function is defined as having 070 predicted FEV 1 of less than 60070, then only 148of the study subjects or 2.30/0 of the total had impaired pulmonary function. As a result, most

of the study population had adequate lung function, but there wereclear differences in their ability to perform the test. A low percentage of predicted FEV 1 was found to be a statistically significant risk factor for lung cancer in this study. It was observed in table 2 that subjects with a 070 predicted FEV 1 less than 94.5 had an increased chance of developing lung cancer. Adjustment for the effects of cigarette smoking reduced the risk, but it was still at least two times greater in subjects with diminished pulmonary function. The relative risk analysis with covariate adjustment was repeated by deciles instead of quartiles, and the results were very similar. They suggest that there may be a threshold effect of increased risk when the 070 predicted FEV 1 drops below 94.5. The magnitude of the impact of cigarette smoking on lung cancer in the study population can be seen from the results in table 6. Light and heavycurrent smokers had a relative risk of 7.1 and 19.1, respectively, for lung cancer after adjusting for the effects of 070 predicted FEV rThese findings and those of earlier studies (2)leave little doubt that cigarette smoking is a strong causative factor for lung cancer in this community-based population. Because there are usually residual effects caused by smoking that cannot be entirely eliminated with statistical adjustment, we also stratified the subjects by smoking history (table 5). The results showed that past smokers and current smokers had an increase in lung cancer risk with decreasing 070 predicted FEV 1 values, but the trend was not statistically significant for the heavy current smokers. Although the 070 predicted FEV 1 was not related to lung cancer among the neversmokers, there wereonly eight cases in this group, which limited interpretation of the results. Overall, the data still suggest that a diminished 070 predicted FEV 1 in past and current cigarette smokers adds to their lung cancer risk. Our findings can be reconciled with those of past studies. The investigation by Peto and coworkers (7) included 2,718 British men whose recruitment into the study was mainly based on occupation. They found that subjects with a FEV 1 below average had a 30 to 50070 greater risk for lung cancer mortality. Our results showed that subjects with a 070 predicted FEV1 belowthe median value had at least twice the lung cancer risk as subjects above the median. Since both studies included a predominantly healthy, employed population, it is expected that the




% Predicted FEV1 103.5+ 94.5-103.4 84.5-94.4 < 84.5 P Value for linear trend

Current Smokers 55 Pack-Years)


Past Smokers

Never Smokers No. of Cases



No. of Cases

3 3 1 1

1.0 1.3 0.5 0.7

0.3-6.5 0.1-4.5 0.1-6.5

2 4 8 12


95% CI

No. of Cases

1.0 2.4 5.4 8.1

0.4-3.1 1.2-25.5 1.8-36.2

11 8 23 23

1.0 0.7 1.7 1.9 0.04




TABLE 6 ADJUSTED RELATIVE RISK (RR) AND 950Al CONFIDENCE INTERVAL (CI) FOR LUNG CANCER BY CIGARETTE SMOKING HISTORY Smoking History Never smokers Past smoker, 15+yr Past smoker, < 15 yr Current smoker, ~ 55 pack-years Current smoker, 56+ pack-years

Age Adjusted

Covariate Adjusted*

No. of CasesiNoncases


95% CI


95% CI

811,718 5/384 2111,187

1.0 2.8 3.9

0.9-8.6 1.7-8.8

1.0 2.8 3.8

0.9-8.5 1.7-8.5











• Covariates include age at examination and % predicted FEV 1 •

findings would be similar. In the MRFIT study, the men were selected because of their high risk for coronary heart disease. The report, which was limited to cigarette smokers, found that the subjects in the lowest quintile of pulmonary function had a relative risk of 3.6 for lung cancer compared with those in the highest quintile after adjustment for cigarette smoking and other factors (6). The other three studies of pulmonary function and lung cancer mainly involved patients who had significantly impaired ventilatory capacities. The study by Tockman and colleagues (3) followed two groups of subjects: 667 male patients with a clinical diagnosis of COPD and 3,728 male participants who smoked at least one pack of cigarettes per day. A total of 1,031 (23070) of the 4,395 men had a 070 predicted FEV 1 less than 60% and a ratio of FEV 1 to FVC of less than 60%. For this group, the relative risk of lung cancer was 5.3 compared to the subjects without ventilatory impairment after adjustment for age and pack-years of smoking. In another prospective study, 113 patients with COPD, chronic bronchitis, or emphysema had a median 070 predicted FEV 1 of 48070 (4). They were matched by age and cigarette smoking history to

113 subjects who had a median % predicted FEV 1 of 100%. After a followup period of about 10 yr, 9 cases of lung cancer were diagnosed in the first group compared with 2 in the healthier group (p = 0.02). In the third study, 835 subjects with COPD were followed for an average of 4.3 yr (5). They were found to have four to five times the risk for developing lung cancer than other smokers or even smokers with bronchitis from other study series. All three of these studies included patients who had either clinical COPD or moderate to severely impaired 070 predicted _FEV l' The magnitude of the relative risk of lung cancer was uniformly 4 to 5 in these studies. It is not surprising, then, that these studies found a large lung cancer risk in patients with significant pulmonary dysfunction, but the British study and the present study observed risks of a lowermagnitude in study populations with much less pulmonary disability. None of the five cited studies (3-7) analyzed their lung cancer cases separately by histologic type or tumor location. When this was done in the present study, we found a somewhat stronger association of impaired ventilatory function with squamous- or small-cell carcinoma

95% CI 0.3-1.8 0.8-3.5 0.9-4.0

Current Smokers (56+ Pack-Years) No. of Cases 4 7

34 28


95% CI

1.0 1.2 4.2 2.3

0.4-4.2 1.5-11.8 0.8-6.5


than with adenocarcinoma. The location of the cancers was central in 77070 (55 of 71 with recorded location) of the squamous- or small-cell cancers and peripheral in 70070 (32 of 46 with recorded location) of the adenocarcinomas. Additional analysis revealed that the increased risk was more substantial for cancers located within 4 em of the pulmonary hilum than for those located more peripherally. Further work is needed by others to confirm these observations. It is unclear why certain cigarette smokers develop significant pulmonary dysfunction but other smokers do not. Cohen and colleagues studied COPD extensively (16, 17). They found that more of the first-degree relatives of lung cancer patients and COPD patients had impaired ventilatory function than neighborhood control subjects or the relatives of nonpulmonary patients. This difference could not be accounted for by the following adjustment factors: age, sex, ethnicgroup, socioeconomic status, smoking, coffee, tea, alcohol, or any of the tested genetic markers (ai-antitrypsin variants, ABO blood type, ABH secretor status, and phenylthiocarbamide taste ability). Furthermore, the observed difference was present among those who never smoked cigarettes. It was suggested that lung cancer and COPD share a common familial pathogenetic factor associated with pulmonary dysfunction. If this is the case, then it is not unexpected that reduced ventilatory function is a predictor of lung cancer, even beyond the effects caused by cigarette smoking. Acknowledgment The writers thank the following institutions for their helpful cooperation: Castle Medical Center, Kaiser Medical Center, Queen's Medical Center, St. Francis Hospital, Straub Clinic and Hospital, Tripler Medical Center,


Wahiawa General Hospital, and the Hawaii Tumor Registry.

References 1. Higgins MW, Keller JB, Landis JR, et al. Risk of chronic obstructive pulmonary disease. Am Rev Respir Dis 1984; 130:380-5. 2. Loeb LA, Ernster VL, Warner KE, Abbotts J, Laszio J. Smoking and lung cancer: an overview. Cancer Res 1984; 44:5940-58. 3. Tockman MS, Anthonisen NR, Wright EC, Donithan MG, Intermittent Positive Pressure Breathing Trial Group, Johns Hopkins Lung Project for the Early Detection of Lung Cancer. Airwaysobstruction and the risk for lung cancer. Ann Intern Med 1987; 106:512-8. 4. Skillrud OM, Offord KP, Miller RD. Higher risk of lung cancer in chronic obstructive pulmonary disease. Ann Intern Med 1986; 105:503-7. 5. Davis AL. Bronchogenic carcinoma in chronic obstructive pulmonary disease. JAMA 1976; 235:621-2.


6. Kuller LH, Ockene J, Meilahn E, SvendsenKH. Relation of forced expiratory volume in one second (FEV 1) to lung cancer mortality in the multiple risk factor intervention trial (MRFIT). Am J Epidemiol 1990; 132:265-74. 7. Peto R, Speizer FE, Cochrane AL, et al. The relevance in adults of airflow obstruction, but not of mucus hypersecretion, to mortality from chronic lung disease. Am Rev Respir Dis 1983; 128:491-500. 8. Worth RM, Kagan A. Ascertainment of men of Japanese ancestry in Hawaii through World War II Selective Service registration. J Chron Dis 1970; 23:389-97. 9. Marcus EB, MacLean CJ, Curb JD, Johnson LR, Vollmer WM, Buist AS. Reference values for FEV 1 in Japanese-American men from 45 to 68 years of age. Am Rev Respir Dis 1988; 138:1393-7. 10. Young JL, Percy CL, Asire AJ, eds. Cancer incidence and mortality in the United States, 1973-77. Surveillance, epidemiology, and end results: incidence and mortality data, 1973-77. Washington, DC: U.S. Government Printing Of-

fice, 1981. 11. World Health Organization. The World Health Organization histological typing of lung tumours, second edition. Am J Clin Patho11982; 77:123-36. 12. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Nat! Cancer Inst 1959; 22:719-48. 13. Cox DR. Regression models and life tables (with discussion).J R Stat Soc [B]1972;34:187-220. 14. Harrell F. The PHGLM procedure. In: SUGI supplemental library user's guide. Cary, NC: SAS Institute, 1983; 267-94. 15. Heilbrun LK, Nomura A, Stemmermann GN. The effects of nonresponse in a prospective study of cancer. Am J Epidemiol 1982; 116:353-63. 16. Cohen BH. Chronic obstructive pulmonary disease: a challenge in genetic epidemiology. Am J Epidemiol 1980; 112:274-88. 17. Cohen BH, Diamond EL, Graves CG, et al. A common familial component in lung cancer and chronicobstructive pulmonary disease.Lancet 1977; 2:523-6.

Prospective study of pulmonary function and lung cancer.

The role of pulmonary function as an independent predictor of lung cancer risk was studied in a community-based cohort of 6,317 Japanese-American men ...
537KB Sizes 0 Downloads 0 Views