Spirometry, Rapid FEV₁ Decline, and Lung Cancer Among Asbestos Exposed Heavy Smokers

Abstract

We assessed whether spirometric measurements are associated with the development of accelerated FEV₁ decline and lung cancer among active and previous smokers with a wide range of lung function. Bivariate and multivariate analyses that adjusted for age, intervention arm, smoking status at enrollment and smoking history, years exposed to asbestos, and evidence of asbestosis were used to assess whether baseline FEV₁ and FEV₁/FVC ratio were associated with accelerated FEV₁ decline and incident lung cancer. The 3,041 participants enrolled from 1985 to 1994 were followed through April 30, 2005. Baseline FEV₁/FVC ratio < 0.7 was significantly associated with an increased risk for rapid lung function decline (OR = 1.73; 95% CI 1.31-2.28; p < 0.001). Baseline FEV₁/FVC ratio < 0.7 was also significantly associated with an increased risk of developing lung cancer, even when baseline FEV₁ was > 80%. Lung cancer risk among participants with baseline airflow obstruction and FEV₁ < 60% was 4-fold higher than participants without baseline airflow obstruction and FEV₁ > 80% (p < 0.001), even among former smokers. These data indicate an FEV₁/FVC < 0.7 among smokers is significantly associated with faster airflow loss, and an increased risk for developing lung cancer, even among those individuals with a normal FEV₁.

Key words: :

• COPD
• Lung cancer
• Risk factors

Introduction

Cigarette smoking is the major environmental risk factor for developing chronic obstructive pulmonary disease (COPD) and lung cancer. The hallmark feature of COPD is the presence of non-fully reversible airflow obstruction that is attributed to an accelerated rate of lung function decline over time ([1]). These features are reflected by spirometric measurements, the subject of many studies examining whether the one-second forced expiratory volume (FEV₁) and the one-second forced expiratory volume to forced vital capacity ratio (FEV₁/FVC) can be used as predictors of a COPD patient's risk for experiencing rapid loss of lung function ([2]). The degree of decline in these measurements has also been associated with the risk of lung cancer; smokers with COPD, when compared to smokers without COPD, have a higher risk for developing lung cancer ([3], [4], [5], [6]).

Taken together, these studies suggest that among individuals who already have COPD, spirometric measurements may be useful for identifying those at higher risk for developing these diseases. Unfortunately, the majority of these studies have limited public health application. Active and former smokers with relatively normal or mildly abnormal lung function, who represent the majority of the at risk population, are often excluded from these studies. To address this issue, we studied the association of abnormal spirometric measurements with the development of accelerated FEV₁ decline and lung cancer in a cohort of current and previous smokers who were exposed to asbestos and had a wide range of airflow limitation, enrolled in the lung cancer chemoprevention trial, CARET. These analyses represent an extension of findings relating the risk of poor lung function and lung cancer among asbestos exposed individuals that have been previously reported.

METHODS

Study population

The Beta-Carotene and Retinol Efficacy Trial (CARET) was a multi-centered double-blind, placebo-controlled chemoprevention trial that enrolled 18,314 individuals to determine the effect of moderate dose beta-carotene and vitamin A on incident lung cancer among current and former heavy smokers and asbestos-exposed workers ([7]). Participants were enrolled in the study from 1985–1994. The study enrolled two cohorts at high risk for lung cancer ([8]). One cohort consisted of 14,254 men and women who were current or former (quit ≤ 6 years) smokers with a ≥ 20 pack-year history and were aged 50–69 at the time of entry. This cohort was not analyzed because spirometric measurements were not collected. The second cohort, the focus of this study, consisted of 4,060 asbestos-exposed individuals ([9]). Eligibility criteria included age of 45 to 69, active or previous (quit ≤ 15 years) smokers, and an occupational exposure to asbestos at a high risk trade (see below) for a minimum of 5 years beginning at least 10 years prior to enrollment, or a chest radiograph with changes consistent with asbestos-related disease and an occupational history of asbestos exposure. During the pilot phase of the trial, there was no smoking requirement for the asbestos cohort; 133 participants had no history of smoking.

The present analysis is restricted to the 3,157 active and previous smokers exposed to asbestos for whom baseline and at least one follow-up spirometry was obtained. Primary analyses are based on 3,041 participants with a history of smoking. For comparison, the 116 never smokers with two or more spirometries were also assessed. All participants in this cohort had spirometric measurements (see below) prior to enrollment, then annually thereafter.

Chest radiographs

A standard posteroanterior chest radiograph was obtained for each participant at entry into the study. All chest radiographs obtained at a study center (5 total) were interpreted by a local reader trained in the interpretation of radiographs of asbestos-exposed participants according to a modified International Labour Office (ILO) Classification of Radiographs of Pneumonoconiosis (1980) system ([9]). Standard ILO scoring for the classification of small opacities was used, but the system for rating pleural thickening was simplified.

The modified pleural readings included the presence of pleural abnormalities, the extent and width of pleural thickening, whether it was diffuse and/or circumscribed, the presence or absence of pleural calcifications, presence of diaphragmatic plaques and calcifications, and presence of costophrenic angle blunting. In addition, pleural thickening in the standard ILO A category was subcategorized as either < 2 mm (A₁) or > 2 mm (A₂). Assessment of inter-reader variability, which has been described in detail elsewhere ([9]), was performed using a nonrandom sample of 48 chest films selected to assure a mix of radiographic findings and by assessing agreement on profusion rating.

Spirometry

Spirometry, which was attempted twice at baseline (once at the participants' enrollment visit and 3 months later at their randomization visit), was performed by trained technicians according to American Thoracic Society (ATS) standards (1987). All values were expressed as a percentage of the normal predicted values according to published equations ([10]). Quality assurance procedures at all of the study sites have been described in detail previously ([9]). Baseline spirometry was not performed or data were not complete for 7% of the participants. Of the spirometric tests performed on the first visit, 87% met ATS performance criteria; for 7%, variability was > 5%; in 2%, the curve was not smooth; in 2%, both variability and curve did not meet ATS criteria, and in 3%, there were other performance problems. Spirometric tests meeting ATS performance criteria were obtained for an additional 4% of participants at the second visit (after a 3-month run-in period of placebo vitamin administration), for a total of 91%. Tests that did not meet the smooth curve criteria were excluded from analyses. No spirometric tests were excluded on the basis of poor reproducibility alone. Participants who did not have at least one follow-up spirometry after enrollment were excluded from these analyses.

Exposure history and follow-up

At the time of recruitment, all participants provided occupational history information, including trade, type of industry, and years of employment for each job worked. High-risk trades were defined as asbestos worker, boilermaker, plasterboard worker, electrician, plumber/pipe fitter, ship scaler, ship fitter, sheet metal worker, or other. Annual in person visits were conducted from 1985–1996, during which spirometry was repeated. Thereafter participants were followed annually by phone or mail for cancer and death ascertainment. All reports of incident cancer or cancer mortality were confirmed by the CARET Endpoints Committee through review of clinical records and pathology reports. The present analysis is based on follow-up through April 30, 2005.

Statistical methods

All analyses were conducted using SAS, version 8.2 (SAS Institute, Inc., Cary, North Carolina). Two-sided p-values < 0.05 were considered statistically significant. Categorical univariate analyses were conducted using Pearson's χ² tests. Then, t-tests were used for comparing means. All spirometric values were assessed as a percent of predicted normal values. The main predictor variables were baseline FEV₁ and FEV₁/FVC ratio. FEV₁ was categorized according to deciles until < 60% of predicted normal. Participants with a baseline FEV₁/FVC ratio < 0.7 were defined as having airflow obstruction according to the Global Initiative for Chronic Obstructive Lung Disease criteria ([1]). The main outcome variables were rate of FEV₁ decline, rapid FEV₁ decline (see later), and development of lung cancer. A least-squares regression line was used to estimate the annualized rate of FEV₁ decline observed during the study. Participants who experienced an annualized rate of FEV₁ decline during the study that was in the fastest quartile of the cohort were considered to be “rapid decliners.” Main predictor variables assessed were the baseline FEV₁ and FEV₁/FVC ratio. Characteristics assessed for potential association with rapid lung function decline include age at enrollment, race, gender, smoking exposure (smoking status, mean pack years, and years since quit), CARET randomization arm, asbestos related chest radiograph findings, and years in a high-risk trade. Multivariate linear and logistic regression models were used to obtain estimates of adjusted mean change in percent of predicted FEV₁ per year and odds ratios for the development of rapid lung function decline. Cox proportional hazard models were used to estimate relative risk and 95% confidence intervals for the development of lung cancer. The time axis in the models was the time to diagnosis of lung cancer or the date on which the participant was last known to be alive. All models included both age and spirometry values as a percent of the predicted normal to minimize the potential confounding effect of age on the FEV1/FVC ratio. Analysis stratified by baseline smoking status was conducted to explore the effects of tobacco exposure.

RESULTS

Baseline spirometric measurements and rate of lung function loss

Baseline characteristics of the 3,041 participants included in the primary analyses are summarized in Table 1. Despite a mean cigarette smoking history of over 40 pack-years, a large proportion of the tobacco exposed participants had FEV₁ and FVC measurements > 80% of the predicted normal (49% and 67%, respectively). The mean FEV₁, FVC, and FEV₁/FVC ratios among the current and former smokers were similar to those for the group of 116 never smokers. However, 35% had evidence of airflow obstruction defined by an FEV₁/FVC ratio < 0.7. The mean rate of FEV₁ decline per year for the cohort was 0.052 ± 0.092 liters per year, representing a loss of 0.924 ± 2.56% per year. There were 759 participants who experienced a rate of FEV₁ decline at ≥ 1.84% per year (range 1.84% to 28.5%), representing the quartile that experienced the fastest decline (“rapid decliners”). In comparison, the mean rate of FEV₁ decline per year was 0.05 ± 0.046 liters per year for the never smokers, representing a loss of 0.89 ± 1.4% per year, which was not significantly different from the current and former smokers.

Table 1 Baseline characteristics of the asbestos exposed cohort

	N (%) or
Characteristics	mean ± SD
Number of former or current smokers	3,041
Mean years of follow-up per participant	4.4 ± 2.1
Mean age at baseline (± SD)	57.3 ± 7.1
Race
Caucasian	2,727 (90)
African American	197 (6)
Other	117 (4)
Gender, % male	100
Baseline smoking status
Former	1,909 (63)
Current	1,132 (37)
Mean pack years of cigarette smoking	42 ± 24
Mean number of quit years for former smokers	9.8 ± 8.2
CARET intervention assignment
25,000 IU vitamin A + 30 mg beta-carotene, daily	1,537 (51)
Placebo	1,504 (49)
Asbestos exposure, mean years in high risk trades^1	19 ± 13
Asbestos related chest radiograph findings
Negative	1047 (34)
Positive parenchymal only^2	555 (18)
Positive pleural only^3	811 (27)
Positive both parenchymal & pleural	627 (21)
Mean number of spirometry per participant	3.4 ± 1.5
Mean baseline FEV1 (Liters)	2.87 ± 0.74
Mean baseline % predicted FEV1	78 ± 18
≥ 80%	1,503 (49)
70–80%	666 (22)
60–70%	429 (14)
< 60%	443 (15)
Mean baseline FVC (Liters)	3.98 ± 0.83
Mean baseline % predicted FVC	86 ± 15
≥ 80%	2,024 (67)
70–80%	576 (19)
60–70%	287 (9)
< 60%	154 (5)
Baseline FEV1/FVC
≥ 0.70	1,983 (65)
< 0.70	1,058 (35)
Number of never smokers	116
Mean baseline FEV1 (Liters) among never smokers	3.07 ± 0.77
Mean baseline % predicted FEV1 among never smokers	86 ± 17
≥ 80%	76 (66)
70–80%	20 (17)
60–70%	15 (13)
< 60%	5 (4)
Mean baseline FVC (Liters) among never smokers	3.98 ± 0.95
Mean baseline % predicted FVC among never smokers	88 ± 16
≥ 80%	73 (63)
70–80% 60–70%	24 (21) 15 (13)
< 60%	4 (3)
Mean baseline FEV1/FVC among never smokers	0.77 ± 0.08
Baseline FEV1/FVC among never
smokers ≥ 0.70 < 0.70	101 (87) 15 (13)

V₁ = 1-second forced expiratory volume; FVC = forced vital capacity.

¹High-risk trade defined as asbestos worker, boilermaker, plaster board worker, electrician, plumber/pipe fitter, ship scaler, ship fitter, sheet metal worker, or other.

²International Labour Office profusion of small, irregular opacities ≥ 1/0.

³Presence of pleural thickening, calcification, diaphragmatic plaques, and/or costophrenic angle blunting.

Overall, participants who had evidence of airflow obstruction at baseline experienced a faster rate of FEV₁ decline during the course of the study. The rate of FEV₁ decline among participants with baseline airflow obstruction was-1.23 ± 3.0% per year (-0.059 ± 0.109 liters per year), versus a rate of-0.76 ± 2.3% per year (-0.048 ± 0.082 liters per year) among participants without baseline airflow obstruction, p < 0.001. To assess the potential influence of smoking status and the presence of airflow obstruction at baseline on the subsequent rate of FEV₁ decline, we compared the rate of FEV₁ decline according to smoking status and degree of obstruction (defined by FEV₁ deciles), stratified by the presence or absence of airflow obstruction (Table 2). Among all participants, regardless of whether airflow obstruction was present at baseline, the overall rate of FEV₁ decline was significantly faster among current smokers than former smokers. Among all current smokers, the overall rate of FEV₁ decline was significantly faster among those with airflow obstruction at baseline (p = 0.002). This was also true for the former smokers (p = 0.01). These trends in rates of FEV₁ decline were generally observed when adjusted for baseline FEV₁, but did not reach statistical significance.

Table 2 Influence of smoking status and the presence of airflow obstruction at baseline on the subsequent rate of FEV₁ decline

Smoking status	N	Change in FEV1 (%/year)	p-value
Current
FEV1/FVC < 0.7	496	−1.49 (3.44)	0.002
FEV1/FVC ≥ 0.7	636	−0.93 (2.44)
Former
FEV1/FVC < 0.7	562	−1.0 (2.55)	0.01
FEV1/FVC ≥ 0.7	1347	−0.68 (2.17)
Combined
FEV1/FVC < 0.7	1058	−1.23 (3.0)	< 0.001
FEV1/FVC ≥ 0.7	1983	−0.68 (2.17)

FEV₁ = 1-second forced expiratory volume; FVC = forced vital capacity

* mean (standard deviation).

Bivariate assessment that considered the rate of FEV₁ decline as both a continuous variable and categorical variable (“rapid decliners” versus “nondecliners”) demonstrated that older age, current smokers, and higher pack years were significantly associated with faster FEV₁ decline (Table 3). There was also a trend toward a lower risk for rapid FEV₁ decline associated with the active CARET intervention arm active CARET intervention (odds ratio [OR] 0.85; 95% confidence interval [CI] 0.72-1.01; p = 0.06). However, this was not evident when the rate of FEV₁ decline was considered as a continuous variable. Asbestos exposure measured by years in high-risk trade did not appear to influence the rate of FEV₁ decline. However, chest radiographs positive for both parenchymal and pleural manifestations was significantly associated with rapid FEV₁ decline (OR 1.35, 95% CI 1.08–1.7).

Table 3 Univariate analysis of participant characteristics and the rate of FEV₁ decline

	Rate of FEV1 decline		Rapid decline
Characteristics	Coef estimate (95% CI)	p-value	Odds ratio (95% CI)	p-value
Age at baseline (per 10-yr increase)	−0.36 (−0.49, −0.23)	< 0.001	1.48 (1.32, 1.67)	< 0.001
Race (versus white)		0.45		0.33
Black	−0.10 (−0.47, 0.27)		0.98 (0.70, 1.37)
Other	−0.28 (−0.76, 0.19)		1.35 (0.90, 2.02)
Former smoker (vs current)	0.40 (0.21, 0.59)	< 0.001	0.61 (0.52, 0.72)	< 0.001
Pack years (per increase of 10 pack years)	−0.08 (−0.12, −0.04)	< 0.001	1.09 (1.06, 1.13)	< 0.001
Active CARET intervention (vs placebo)	0.08 (−0.10, 0.26)	0.38	0.85 (0.72, 1.01)	0.06
Years in high-risk trade (per 10-yr increase)	0.02 (−0.05, 0.09)	0.51	0.98 (0.93, 1.05)	0.63
Chest radiograph(vs negative)		0.31		0.06
Positive parenchymal only^1	0.11 (−0.16, 0.37)		1.24 (0.97, 1.57)
Positive pleural only^2	0.05 (−0.19, 0.28)		1.19 (0.96, 1.48)
Positive both parenchymal and pleural	−0.16 (−0.41, 0.10)		1.35 (1.08, 1.70)

FEV₁ = 1-second forced expiratory volume; coef = coefficient; CI = confidence interval.

Table 4 Relationship between airflow obstruction and FEV₁ decline, adjusted for smoking status

		Change in FEV1 (%/year)		Rapid decline
	N	Adjusted mean (SE)	p-value	OR (95% CI)	p-value
Former smokers at baseline
FEV1/FVC ≥ 70%	1,341	−0.73 (0.06)	0.23	Referent	−
FEV1/FVC < 70%	561	−0.88 (0.10)		1.16 (0.91, 1.49)	0.24
Current smokers at baseline
FEV1/FVC ≥ 70%	636	−1.01 (0.12)	0.05	Referent	−
FEV1/FVC < 70%	496	−1.39 (0.14)		1.73 (1.31, 2.28)	< 0.001

SE = standard error; OR = odds ratio; CI = c onfidence interval.

* Adjusted for baseline age, pack-years, intervention arm, years of exposure in high-risk trade, and radiographic findings (pleural abnormalities, profusion rating ≥ 1/0, both, or neither).

Given the potential confounding effect of smoking status on the presence of airflow obstruction at baseline and the rate of FEV₁ decline, we estimated the risk of experiencing rapid FEV₁ decline associated with a baseline airflow obstruction in multivariate models, stratified according to smoking status (Table 4). Age at baseline and smoking history in pack-years were included as covariates. We also included years of exposure in high-risk trade and chest radiograph findings in the multivariable model to insure that we accounted for their potential confounding effects. Because our initial results in Table 2 did not demonstrate significant interactions between FEV₁ and FEV₁/FVC, we did not stratify participants based upon baseline FEV₁. Among current smokers, a baseline FEV₁/FVC ratio < 0.7 was associated with a significant increase in risk of developing rapid FEV₁ decline in comparison to participants a baseline FEV₁/FVC ratio ≥ 0.7 (OR 1.73; 95% CI 1.31–2.28; p < 0.001). This was not observed among former smokers.

Baseline spirometric measurements and risk of developing lung cancer

Overall, there were 206 cases of non-mesothelioma lung cancers: 56 (27%) small cell carcinomas, 49 (24%) adenocarcinomas, 29 (14%) squamous cell carcinomas, 37 (18%) other non-small cell carcinomas, and 35 (17%) unknown type (self-reported or medical records/tissue not obtained). The risk of developing lung cancer associated with asbestos exposure in this cohort has been reported previously ([6]). Although that study found lower baseline FEV₁, FVC, and FEV₁/FVC ratio were associated with a higher risk for developing lung cancer, it did not assess the interaction between the FEV₁ and FEV₁/FVC ratio. In our multivariate analysis adjusted for baseline age, smoking status, pack-years, intervention arm assignment, years in high risk trade, and chest radiograph findings, we found significantly higher incidence of lung cancer among patients with airflow obstruction (n = 125, 12%) than among patients without airflow obstruction (n = 80, 4%).

Using Cox-proportional hazards modeling, airflow obstruction was associated with a 2.23-fold (95% CI 1.66 to 3.01, p < 0.001) increased risk of developing lung cancer. In a separate multivariate model, the risk for lung cancer was inversely proportional to the baseline FEV₁. In comparison to a baseline FEV₁ ≥ 80%, the hazard for developing lung cancer increased significantly with each decile decrease of the baseline FEV₁ (FEV₁ 70 to 80% hazard ratio [HR] 1.52, 95% CI 1.01 to 2.26, p < 0.001; FEV₁ 60 to 70% HR 2.19, 95% CI 1.46 to 3.29, p < 0.001; FEV₁ < 60% HR 2.94, 95% CI 2.0 to 4.32, p < 0.001) Incorporation of both the FEV₁ and FEV₁/FVC ratio in a multivariate model revealed that for each lower decile category of baseline FEV₁, the presence of airflow obstruction at baseline was associated with a greater risk of lung cancer when compared to participants without airflow obstruction (Table 5).

Table 5 Lung cancer risk according to FEV₁ and FEV₁/FVC, stratified by smoking status at baseline

	All participants				Current smokers at baseline**				Former smokers at baseline**
	N	cases (%)	HR (95% CI)	p-value	N	cases (%)	HR (95% CI)	p-value	N	cases (%)	HR (95% CI)	p-value
FEV1/FVC ≥ 0.70
Baseline FEV1
≥ 80%	1,281	37 (3)	Referent	–	384	13 (3)	Referent	–	897	24 (3)	Referent	–
70– < 80%	422	21 (5)	1.54 (0.80, 2.63)	0.12	143	9 (6)	1.76 (0.75, 4.12)	0.20	279	12 (4)	1.41 (0.70, 2.83)	0.33
60– < 70%	190	14 (7)	2.25 (1.20, 4.19)	0.01	73	5 (7)	1.81 (0.64, 5.12)	0.26	117	9 (8)	2.61 (1.20, 5.65)	0.02
< 60%	83	8 (10)	3.08 (1.42, 6.69)	0.004	36	4 (11)	3.61 (1.02, 9.75)	0.05	47	4 (9)	3.05 (1.05, 8.85)	0.04
FEV1/FVC < 0.70
Baseline FEV1
≥ 80%	218	19 (9)	2.37 (1.35, 4.16)	0.003	107	12 (11)	2.57 (1.16, 5.69)	0.02	111	7 (6)	2.16 (0.93, 5.03)	0.07
70–80%	242	22 (9)	2.48 (1.45, 4.25)	0.001	108	8 (7)	1.79 (0.73, 4.35)	0.20	134	14 (10)	3.24 (1.66, 6.30)	< 0.001
60–70%	237	30 (13)	3.16 (1.92, 5.19)	< 0.001	114	18 (16)	3.70 (1.80, 7.64)	< 0.001	123	12 (10)	2.62 (1.29, 5.29)	0.007
< 60%	360	54 (15)	3.93 (2.51, 6.15)	< 0.001	167	30 (18)	4.36 (2.23, 8.56)	< 0.001	193	24 (12)	3.53 (1.96, 6.34)	< 0.001

HR = hazard ratio; CI = confidence interval.

* Both FEV₁/FVC and baseline FEV₁ are in the same model adjusted for baseline age, smoking status, pack-years, intervention arm assignment, years in high-risk trade, and chest radiograph findings.

** Both FEV₁/FVC and baseline FEV₁ are in the same model adjusted for baseline age, intervention arm assignment, pack-years, years in high-risk trade, and chest radiograph findings.

Participants who had an FEV₁/FVC ratio < 0.7 and an FEV₁ < 60% at baseline had the highest risk for developing lung cancer. Even the presence of airflow obstruction with a baseline FEV₁ > 80% was associated with a significantly increased risk for developing lung cancer. To assess for the potential effects of tobacco exposure on baseline lung function and lung cancer risk, we stratified this analysis according to smoking status at baseline. The association of increased lung cancer risk associated with baseline FEV₁ and FEV₁/FVC persisted (Table 5).

DISCUSSION

In this analysis, we found that a baseline FEV₁/FVC ratio < 0.7 was significantly associated with an increased risk for rapid airflow decline, and abnormalities in baseline FEV₁ and FEV₁/FVC ratio were significantly associated with an increased risk for lung cancer. Most importantly, we found the presence of an FEV₁/FVC ratio < 0.7 was associated with an increased risk of observing these outcomes, regardless of the degree of airflow obstruction reflected by the FEV₁. Unlike most previous studies examining the predictive value of spirometric measurements for COPD, participant enrollment in CARET was not based upon presence of COPD or respiratory symptoms, and included smokers with normal or mildly abnormal lung function.

Fletcher and Peto first described the variable susceptibility of smokers to the airways effects of tobacco in 1976 ([11]). They reported “many smokers lose FEV₁ almost as slowly as non-smokers and never develop clinically severe airflow obstruction,” suggesting that a select population of smokers are largely resistant to the effects of tobacco smoke on airflow. The baseline lung function of our cohort is consistent with this observation. The majority of our cohort did not have evidence of airflow obstruction (65% had an FEV₁/FVC ratio ≥ 0.7) and 49% of the cohort had a normal FEV₁ (≥ 80% of the predicted normal) despite the impressive average tobacco exposure of 42 ± 24 pack years.

There has been a long-standing interest in using spirometric measurements to assess the risk of developing COPD or experiencing rapid airflow decline. Several longitudinal studies have already found that airflow obstruction in smokers is a strong independent predictor of morbidity and mortality from cardiovascular disease and COPD ([12]). In 1987, in a cohort of nonsmokers and smokers with FEV₁ > 60% of predicted, Burrows confirmed that the “horse-racing effect,” the concept that detecting low lung function will predict future decline in lung function, is observed in spirometric measurements and serves as a reasonable basis for using spirometric measurements as a screening tool. Of interest, he found the FEV₁/FVC ratio was more informative than the FEV₁, contradicting the previous findings of Fletcher and Peto. Burrow's observations were later confirmed by Enright et al., who found that FEV₁/FEV₆ correlated closely with FEV₁/FVC, and was also useful for identifying smokers at risk for experiencing rapid airflow decline (13). Our findings are consistent with these results. We found the presence of airflow obstruction at baseline, defined by an FEV₁/FVC ratio < 0.7, was associated with faster FEV₁ decline only among current smokers in both multivariate linear and logistic models. The lack of significant association among former smokers was not surprising, given the rate of FEV₁ decline improves after smoking cessation ([14]).

Several studies have found that reduced lung function is associated with an increased risk for lung cancer. The largest study, reported by Eberly et al. found the degree of FEV₁ reduction was a strong predictor of lung cancer mortality in smokers, independent of smoking history and quit status; a 5% increase in lung cancer risk was observed for every 100 ml decrease of the FEV₁ ([5]). More recently, Wasswa-Kintu et al. found the risk for lung cancer to be two-to four-fold higher among participants with an FEV₁ < 70% of predicted when compared to participants with an FEV₁ > 100% of predicted ([15]). Even small decreases in FEV₁ to 90% of predicted increased the risk for lung cancer by 1.3-fold in men (95% CI 1.05 to 1.62) and 2.64 fold in women (95% CI 1.30 to 5.31). As mentioned earlier, a previous study also examined this relationship using the CARET cohort ([6]). This analysis found reductions in FEV₁, FVC, and FEV₁/FVC ratio were independently associated with an increased risk of developing lung cancer, after adjusting for tobacco and asbestos exposure.

The current study extends these findings, demonstrating the presence of airflow obstruction, defined by an FEV₁/FVC ratio < 0.7, increases the lung cancer risk associated with a reduced FEV₁, even when the FEV₁ is normal (FEV₁ ≥ 80%) or mildly reduced (FEV₁ 60–80%). We also found that even in the absence of obstruction, defined by an FEV₁/FVC ratio ≥ 0.7, the risk for lung cancer was inversely proportional to the FEV₁. Given such spirometric changes are consistent with a restrictive pattern, we suspect this increase in lung cancer risk may be partially attributable to asbestos exposure, although the number of years at a high-risk trade was included in the model. Unfortunately, lung volume was not assessed in CARET to confirm the presence of restrictive lung disease.

Our results are unique in several aspects. First, this is the only study reporting spirometric risk factors for airflow decline and lung cancer in the same cohort. Although COPD and lung cancer may appear to be different diseases, they have some similarities. An inflammatory component has been suggested as part of both COPD and lung cancer pathogenesis. Lung inflammation is present in active smokers, ([16], [17], [18]) persists among long-term quitters ([16], [19]), and is amplified among the minority of smokers who experience rapid lung function decline ([17]). Chronic inflammation is associated with many cancers ([20]) and may precede the development of lung cancer ([21], [22], [23]). Second, none of the previous studies examined the potential interaction between FEV₁ and the FEV₁/FVC ratio. The influence of tobacco smoke exposure on FEV₁ is well established ([11], [14]). However, this is not true for the FEV₁/FVC ratio, which can be reduced even among never-smokers ([24]).

Although the presence of a reduced FEV₁/FVC ratio among never smokers may be caused by factors such as air pollution and second hand smoke, it may be also explained by genetic predisposition to a reduced FEV₁/FVC ratio ([25]), which may predispose to retention of toxic products in tobacco smoke. The latter theory would suggest that the FEV₁/FVC ratio phenotype might serve as a “biomarker” for identifying heavy smokers at high risk for rapid airflow decline and lung cancer. In the current study, a baseline FEV₁/FVC < 0.7 increased the risk for experiencing rapid FEV₁ decline by 1.73-fold and was associated with significantly higher risks for lung cancer, regardless of the FEV₁.

There are some aspects of this study that should be considered when interpreting these data. First, one of the major inclusion criteria for this cohort was exposure to asbestos, which may influence lung function. The characteristic change in pulmonary function observed in asbestosis is a restrictive impairment ([26]). Large airway function, as reflected by the FEV₁/FVC ratio, is generally well preserved. However, airway obstruction can also be observed and can be seen alone in nonsmokers who have asbestosis, which may be due to asbestos-induced small airway disease ([27]). Therefore, we included asbestos exposure in all of our multivariate models and stratified our analyses according to smoking status to maximally account for the confounding effects of these exposures. Despite this, we believe these findings need to be validated in a nonasbestos exposed cohort before its applicability to all smokers is accepted.

Second, this study was conducted on only males. Given there are likely sex specific differences in susceptibility to COPD ([28], [29]), these results may not be generalizable to women. However, since previous studies that included women found similar results, we suspect our findings will be applicable across genders. Third, our results are subject to the “healthy survivor effect,” where individuals are able and willing to perform spirometric maneuvers tend to be healthier. However, we believe this effect would reduce the likelihood that we would find an association, biasing our results toward the null. Finally, our analysis did not account for changes in smoking behavior and tobacco exposure during the follow-up period. We suspect this is unlikely to influence our results because smokers with more severe lung disease are generally more likely to quit smoking ([30]), which would again reduce the likelihood that we would find an association.

In summary, we found the presence of airflow obstruction among smokers is significantly associated with faster airflow loss, and the presence of airflow obstruction is significantly associated with an increased risk for developing lung cancer, even among those individuals with a normal FEV₁. We hope this report brings attention to the current lack of standard clinical methods for estimating a smoker's risk of experiencing rapid airflow decline and lung cancer. Future prospective studies should focus on whether spirometry is useful in a multistage screening approach in identifying active and previous smokers at risk for COPD and lung cancer.

Financial support came from NIH grants K23HL69860 and CA63673. Dr. Au is supported by a Veterans Administration Health Services Research and Development Career Development Award. We would like to thank members of the CARET Asbestos Research Group, Drs. John R. Balmes, Carl A. Brodkin, Brenda Cartmel, Gilbert S. Omenn, Mark R. Cullen, Carrie A. Redlich, Mark D. Thornquist for making this project possible. Authors Chien, Au, Barnett, and Goodman have no conflicts to disclose.