Lung Cancer Risk Reduction After Smoking Cessation: Observations From a Prospective Cohort of Women By J.O. Ebbert, P. Yang, C.M. Vachon, R.A. Vierkant, J.R. Cerhan, A.R. Folsom, and T.A. Sellers Purpose: We conducted this study because the duration of excess lung cancer risk among former smokers has been inconsistently reported, doubt has been raised regarding the population impact of smoking cessation, and differential risk reduction by histologic cell type after smoking cessation needs to be confirmed. Methods: The Iowa Women’s Health Study is a prospective cohort study of 41,836 Iowa women aged 55 to 69 years. In 1986, mailed questionnaires were used to collect detailed smoking history. Age-adjusted lung cancer incidence through 1999 was analyzed according to years of smoking abstinence. Relative risks were estimated using Cox regression analysis. Results: There were 37,078 women in the analytic cohort. Compared with the never smokers, former smokers had an elevated lung cancer risk (relative risk, 6.6; 95%

confidence interval, 5.0 to 8.7) up to 30 years after smoking cessation for all former smokers. However, a beneficial effect of smoking cessation was observed among recent and distant former smokers. The risk of adenocarcinoma remained elevated up to 30 years for both former heavier and former lighter smokers. Conclusion: The risk for lung cancer is increased for both current and former smokers compared with never smokers and declines for former smokers with increasing duration of abstinence. The decline in excess lung cancer risk among former smokers is prolonged compared with other studies, especially for adenocarcinoma and for heavy smokers, suggesting that more emphasis should be placed on smoking prevention and lung cancer chemoprevention. J Clin Oncol 21:921-926. © 2003 by American Society of Clinical Oncology.

UNG CANCER risk among female former smokers has been shown to decrease with increasing duration of smoking abstinence.1 Case-control and cohort studies have observed a 50% or greater lung cancer risk reduction in the first decade of smoking abstinence for female former smokers compared with current smokers.1-7 However, the time that needs to pass before the risk of lung cancer among former smokers reaches that of never smokers is less clear. Case-control study data have indicated that excess lung cancer risk persists beyond 10 to 15 years of smoking abstinence and that this risk may decline more quickly for lighter than for heavier smokers.2,3,8 Data from cohort studies have indicated both that the lung cancer risk decreases to the level of never smokers by 15 years7 and that excess risk persists beyond 16 years for heavier smokers.1,9 The inconsistent results in the duration of lung cancer risk after smoking cessation may be partially explained by differing rates of risk reduction by histologic cell type. Previous studies have shown a steady decline in risk for small cell and squamous cell carcinoma but not for adenocarcinoma with increasing duration of abstinence.10,11 A meta-analysis found that smoking cessation is associated with a greater risk reduction for small cell and squamous cell lung cancer than for adenocarcinoma.12 More prospective cohort studies are needed to define the duration of excess lung cancer risk among female former smokers and to determine how lung cancer risk decreases with duration of abstinence by histologic cell type. Evidence exists indicating that the population effect of smoking cessation is less than previously expected. Although decreased lung cancer incidence and mortality has been repeatedly observed in studies of former smokers using conventional case-control1 and cohort6,7,9,13,14 study designs, much of the observational data are based on individuals who were already former smokers at the time these studies were initiated. Although randomized, controlled trials theoretically provide the best data regarding the population effect of smoking cessation, negligible

and even increased lung cancer mortality associated with smoking cessation have been observed in these studies.15-19 Results from natural experiments20,21 have indicated that rates of lung cancer death do not significantly decline in a population experiencing substantial declines in smoking prevalence. These results have obfuscated the benefit of smoking cessation. Cohort studies assessing the effect of recent and distant smoking cessation on lung cancer risk may add to the body of evidence supporting the effort to aggressively treat all current smokers at the individual and population levels. We examined lung cancer risk among former smokers in a large prospective cohort of women to estimate the duration of excess lung cancer risk, to determine risk reduction by histologic cell type, and to assess the effect of smoking cessation. Our cohort study provides additional evidence on smoking cessation and lung cancer among women.

L

METHODS

Study Population and Data Collection The Iowa Women’s Health Study (IWHS) is a prospective cohort study of risk factors for cancer and chronic diseases in women aged 55 to 69

From the Nicotine Research Center, Division of Community Internal Medicine, Mayo Clinic Cancer Center, Rochester; and University of Minnesota, Division of Epidemiology, Minneapolis, MN. Submitted May 20, 2002; accepted November 13, 2002. Supported in part by grants CA92049, CA80127, CA84354, CA39741 from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD. Address reprint requests to: Ping Yang, MD, PhD, Department of Health Sciences and Cancer Center at Mayo Clinic/Foundation, 200 First St. SW, Rochester, MN 55905; email: [email protected]. © 2003 by American Society of Clinical Oncology. 0732-183X/03/2105-921/$20.00

Journal of Clinical Oncology, Vol 21, No 5 (March 1), 2003: pp 921-926 DOI: 10.1200/JCO.2003.05.085

Downloaded from ascopubs.org by 37.44.207.9 on January 27, 2017 from 037.044.207.009 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

921

922

EBBERT ET AL

years.22,23 In January 1986, a questionnaire was mailed to 99,826 women randomly selected from a list of women with a valid Iowa driver’s license in 1985. The 41,836 respondents (42.7% response rate) form the cohort under observation. The cohort was 98% Caucasian, and 65% lived in towns of fewer than 10,000 inhabitants. Compared with nonrespondents, respondents were on average 2 months younger and more likely to live in rural areas. The nonrespondents have been demonstrated to have higher mortality rates from smoking-related diseases than the respondents.24 Self-reported items on the baseline questionnaire included reproductive history, height, self-measured body circumferences, weight, education, and physical activity. The survey also included a food frequency questionnaire.25 Data collected for cigarette smoking history included age at initiation, average packs per day, and age at cessation.

Baseline Exclusions and Cohort Follow-Up At study baseline, we excluded women with previous cancers other than nonmelanoma skin cancer (n ⫽ 3,830). We also excluded cases with missing smoking histories (n ⫽ 928). This left a cohort at risk for incident lung cancer of 37,078 women. Follow-up questionnaires were mailed in 1987, 1989, 1992, and 1997 to update smoking status, address changes, and vital status. However, among women who remained eligible at the time of follow-up questionnaire distribution, the response rates were significantly different across the baseline-defined subgroups of smokers. The response rates were 78%, 84%, and 85% for the 1992 questionnaire (P ⬍ .001) and 74%, 80%, and 81% (P ⬍ .001) for the 1997 questionnaire for current, former, and never smokers at baseline, respectively. To avoid potential bias introduced by differential reporting across exposure groups, smoking-related variables were analyzed based on the baseline questionnaire. Incident lung cancer cases occurring from 1986 through 1999 were identified through the Iowa Cancer Registry, which is part of the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) program.26 The Iowa Cancer Registry uses Iowa death certificate data and the National Death Index as part of its data collection and verification procedure. A computer match was performed annually between the list of cohort members and the records of Iowans with incident cancer in the registry. Data regarding the diagnosis were abstracted by registry personnel from medical records and pathology reports according to SEER protocol27 and were coded according to International Classification of Diseases for Oncology, second edition (ICD-O-2).28 Lung cancer (ICD-O-2: 34.0 to 34.9) was categorized as adenocarcinoma (8140 to 8380, 8481, 8480), squamous cell carcinoma (8050 to 8076), small cell carcinoma (8041 to 8045), or other/ unspecified (8000, 8010, 8012, 8020, 8021, 8430, 8560, 9680, and 9691) cell types using ICD-O-2 morphology codes.

Statistical Methods The length of follow-up for each individual in the study was calculated as the time from completion of the baseline questionnaire until date of lung cancer diagnosis, date of death, or date of move from Iowa. If none of these occurred, the woman was assumed to be alive and living in Iowa through December 31, 1999. As of December 31, 1999, 625 (1.7%) were diagnosed with lung cancer; 4,755 (12.8%) died in Iowa; 85 (0.2%) died outside of Iowa; 1,741 (4.7%) moved out of Iowa; and 29,872 (80.6%) were alive and living in Iowa. Each individual was classified according to smoking status: current (smoked at least 100 cigarettes in lifetime and currently smoking), former (smoked at least 100 cigarettes in lifetime and currently not smoking), and never (smoked fewer than 100 cigarettes in lifetime and currently not smoking). Former smokers were further classified according to years of smoking abstinence and pack-years of smoking (pack-years ⫽ total years of smoking multiplied by cigarette packs smoked per day). Comparisons of the smoking-related characteristics of participants who developed various histologic cell types were performed with analysis of variance (ANOVA) and ␹2 tests. For most analyses, squamous cell and small cell carcinomas were combined into one group because of sparse data. Multivariate Cox proportional hazards regression was used to calculate rate ratios representing relative risks (RRs) and 95% confidence intervals (CIs). Two sets of Cox regression analyses were performed: one comparing former smokers with never smokers (referent group) and one comparing former

smokers with current smokers (referent group). For all Cox models, lung cancer incidence was modeled as a function of age.29 Each model included the following potential confounding variables as covariates: physical activity level, education, body-mass index (BMI), waist circumference, alcohol, and fruit consumption.30-32 Tests for trend of lung cancer risk by increasing years of abstinence were performed by ordering the categorical variable and including it in the model as a linear effect. To assess the effect of smoking cessation during the follow-up period, we evaluated the changing patterns of lung cancer risk with respect to duration of abstinence before study baseline and incidence trends during follow-up. Three groups of smokers were examined in relation to never smokers: 1) distant former smokers, who stopped smoking more than 5 years before study baseline; 2) recent former smokers, who stopped smoking within 5 years of baseline; and 3) all baseline current smokers. All statistical tests were two-sided, and all analyses were carried out using the SAS (SAS Institute, Inc, Cary, NC) and Splus (Mathsoft, Inc, Seattle, WA) software systems.

RESULTS

Through 14 years and 470,111 person-years of follow-up, there were 625 lung cancer cases (250 adenocarcinoma, 121 small cell carcinoma, 119 squamous cell carcinoma, and 135 other/unspecified cell types) identified in the cohort at risk. Between 1986 and 1999, lung cancer incidence was 133 per 100,000 in the IWHS cohort and 170 per 100,000 in all women of Iowa in comparable age groups based on SEER data. Comparing lung cancer cases with noncases among all subjects, the baseline mean ages were similar (Table 1). Among former smokers, lung cancer cases initiated smoking at a younger age, achieved smoking abstinence later in life, smoked a higher mean number of cigarettes per day, and had more years of cigarette use on average (compared with noncases; P ⬍ .001). Adenocarcinoma was the most common histologic cell type and accounted for 40% of all diagnosed lung cancers: 70% of the lung cancers among never smokers, 41% among the former smokers, and 32% among current smokers. Adenocarcinoma was the most common lung cancer cell type for all former smokers in the first, second, and third decade of time since smoking cessation. The relative frequency of adenocarcinoma of all lung cancer increased from 34% (35 of 103) in the first decade to 72% (13 of 18) in the third decade of smoking abstinence. Compared with never smokers, former smokers had an elevated risk of lung cancer up to 30 years of follow-up (RR, 6.6; 95% CI, 5.0 to 8.7 over three decades). Patterns of lung cancer risk among former smokers compared with never and current smokers are shown in Table 2. Lung cancer risk generally decreased with increasing time since smoking cessation (test for trend within each pack-year strata and for both strata combined, P ⬍ .001). Former lighter smokers (1 to 19 pack-years) still had a greater than two-fold increased risk up to 30 years after smoking abstinence. Compared with current lighter smokers, former lighter smokers significantly decreased their risk for lung cancer in the first 10 years of smoking abstinence. This decrease was less pronounced among former heavier (ⱖ 20 pack-years) smokers compared with current heavier smokers in the first 5 years but greatly decreased starting at 6-10 years of smoking abstinence. We assessed the effect of antecedent smoking history on estimates for lung cancer risk. Years smoked and cigarettes per day were highly correlated with pack-years (Pearson’s correlation coefficients r ⫽ 0.7 and 0.9, respectively). When the results

Downloaded from ascopubs.org by 37.44.207.9 on January 27, 2017 from 037.044.207.009 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

923

LUNG CANCER RISK AFTER SMOKING CESSATION Table 1.

Smoking-Related Characteristics of Lung Cancer Cases and Noncases in the Iowa Women’s Health Study, 1986-1999

Noncases (N ⫽ 36,453)

Descriptive Variable

All subjects Mean age (SD) at baseline Smoking status at baseline, % Never Former Current Former smokers Mean age (SD) at lung cancer diagnosis Mean age (SD) at smoking initiation Mean age (SD) at smoking cessation Mean number (SD) of cigarettes per day Mean years (SD) of cigarette use Pack-years, % 1-19 20-39 ⱖ40

All Lung Cancer Cases (N ⫽ 625)

Lung Cancer Histologic Cell Type Adenocarcinoma (n ⫽ 250)

Small Cell Carcinoma (n ⫽ 121)

Squamous Cell Carcinoma (n ⫽ 119)

Other/Unspecified (n ⫽ 135)

P

62.1 (4.2)

62.4 (4.2)

61.8 (4.1)

61.8 (4.1)

62.5 (4.3)

64.0 (4.2)

⬍ .001*

67 19 14

15 23 62

27 24 50

4 14 82

8 24 68

11 29 60

⬍ .001†

— 21.7 (6.6) 46.1 (12.5) 15.5 (10.4) 24.4 (12.8)

— 19.9 (4.7) 54.5 (9.9) 21.5 (10.1) 34.6 (11.0)

69.9 (5.3) 20.8 (5.4) 51.5 (12.1) 19.8 (10.8) 30.8 (13.0)

66.9 (5.5) 19.7 (3.3) 56.1 (7.2) 24.4 (9.2) 36.4 (8.1)

68.8 (5.2) 18.2 (2.1) 57.2 (5.6) 24.0 (8.7) 39.0 (6.0)

71.4 (5.4) 19.9 (5.1) 56.3 (8.9) 20.8 (10.3) 36.4 (9.6)

⬍ .02* .12* .03* .18* .003*

58 27 15

21 35 44

34 32 34

6 47 47

3 38 59

21 33 46

.02†

*Analysis of variance, comparing mean values across lung cancer histology groups and/or cancer status. †␹2 test, comparing proportions across lung cancer histology groups and/or cancer status.

from Table 2 were analyzed using cigarettes per day instead of pack-years, both the magnitudes and the trends of the RRs did not change. Age of initiation and age of cessation were less correlated with pack-years (r ⫽ ⫺0.3 and 0.6, respectively). We examined the patterns of RR of former smokers compared with never smokers stratified by age of initiation (tertiles: ⱕ 18, 19 to 21, and ⬎ 21 years of age) and age of cessation (tertiles: ⱕ 40, 41 to 54, and ⬎ 54 years of age). When pack-years was added as a covariate, the RRs for both age variables were reduced to near unity. We conclude that the age of initiation and the age of quitting are unlikely to be independent risk modifiers beyond pack-years. To evaluate the effect of smoking cessation on lung cancer during the 14-year follow-up, we compared RRs between two intervals; years 1 to 7 and years 8 to 14 after baseline (Table 3). A suggestive decline in relative risk (from 18.4 to 15.4; 16.3%) was observed among baseline current smokers, but a more profound effect of smoking cessation was observed among former smokers, particularly for recent former smokers. Differences in lung cancer risk reduction of recent former smokers and distant former smokers compared with current smokers are shown in Fig 1.

Table 2.

The risk for lung cancer among former smokers compared with never smokers by histologic cell type is shown in Table 4. The risk for adenocarcinoma among all former smokers was increased up to 30 years for lighter and heavier smokers. For squamous cell and small-cell cancer, the RRs were elevated for the heavier smokers in first two decades, but there were only two incident cases among lighter smokers in the entire follow-up period. DISCUSSION

In this population-based prospective cohort, increased risk for lung cancer among former smokers remained up to 30 years after smoking cessation, attributable to the sustained higher risk for adenocarcinoma. Our findings indicate that the elevated risk for lung cancer among female former smokers endured for a longer period of time than previously reported. In the Nurses’ Health Study (NHS),7 the former smokers had an age-adjusted 40% risk reduction for lung cancer compared with current smokers after 2 to 5 years of smoking abstinence, which is consistent with our findings. However, the RR of lung cancer among former smokers was observed to approach that of never smokers after 10 to 15 years. Explanations for this discrepancy include differences in

Lung Cancer Risk After Smoking Cessation by Pack-Years of Smoking History in the Iowa Women’s Health Study, 1986-1999 Pack-Years Smoked ⱖ 20

1-19

All

Years of Smoking Abstinence

No. of Cases

RR Versus Never Smokers

RR Versus Current Smokers

No. of Cases

RR Versus Never Smokers

RR Versus Current Smokers

No. of Cases

RR Versus Never Smokers

RR Versus Current Smokers

0-5 6-10 11-20 21-30 ⬎ 30 All

1 6 7 11 5 30

0.7 (0.1 to 4.9) 4.2 (1.7 to 10.5) 2.1 (1.0 to 4.6) 2.8 (1.5 to 5.5) 1.1 (0.4 to 2.9) 2.0 (1.3 to 3.2)

0.04 (0.01 to 0.28) 0.24 (0.10 to 0.58) 0.13 (0.06 to 0.27) 0.16 (0.09 to 0.31) 0.06 (0.02 to 0.17) 0.12 (0.08 to 0.18)

74 22 11 7 0 114

15.3 (11.1 to 21.3) 9.4 (5.7 to 15.3) 4.3 (2.3 to 8.0) 7.3 (3.2 to 16.7) — 10.4 (7.8 to 14.0)

0.9 (0.7 to 1.1) 0.5 (0.3 to 0.8) 0.2 (0.1 to 0.4) 0.4 (0.2 to 0.9) — 0.6 (0.5 to 0.7)

75 28 18 18 5 144

11.6 (8.4 to 16.0) 7.4 (4.7 to 11.7) 3.2 (2.0 to 5.3) 3.7 (2.1 to 6.3) 1.0 (0.4 to 2.8) 5.6 (4.3 to 7.4)

0.7 (0.5 to 0.8) 0.4 (0.3 to 0.6) 0.2 (0.1 to 0.3) 0.2 (0.1 to 0.3) 0.1 (0.02 to 0.2) 0.3 (0.3 to 0.4)

Note: Cox proportional hazards regression adjusting for age, physical activity, education, body mass index, waist circumference, alcohol use, and fruit consumption. Dashes indicate that parameter estimates could not be calculated because of sparse data. Data for RR given as median (range). Abbreviations: RR, relative risk.

Downloaded from ascopubs.org by 37.44.207.9 on January 27, 2017 from 037.044.207.009 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

924 Table 3.

EBBERT ET AL Relative Risk Stratified by the Time of Quitting Smoking for the First 7 and the Second 7 Years of Follow-Up, Iowa Women’s Health Study, 1986-1999 1-7 Years

8-14 Years

All Years

Subgroups of Smokers Compared With Never Smokers

RR

95% CI

RR

95% CI

RR

95% CI

Current smokers at baseline Former smokers at baseline Recent former (quit ⱕ 5 years) at baseline Distant former (quit ⬎ 5 years) at baseline

18.4 6.3 14.0 3.6

12.5 to 26.9 4.2 to 9.5 8.8 to 22.4 2.2 to 6.0

15.4 4.7 8.8 3.3

11.1 to 21.2 3.3 to 6.7 5.6 to 13.6 2.2 to 5.1

16.6 5.6 10.9 3.4

13.0 to 21.2 4.3 to 7.4 8.0 to 15.0 2.5 to 4.8

NOTE: Analyses accounting for age, physical activity, education, body-mass index, waist circumference, alcohol use, and fruit consumption. Abbreviations: RR, relative risk; 95% CI, 95% confidence interval.

cohort age and antecedent smoking history. The NHS was established with nurses aged 30 to 55 years in 1976, and the IWHS was established with women aged 55 to 69 years of age in 1986. The smokers in our cohort may have had a longer duration of previous smoking. A prolonged return to the risk of never smokers among former smokers in the IWHS cohort is consistent with evidence indicating that the risk of lung cancer declines less steeply with increasing age of smoking abstinence33 and that quitting smoking at a younger rather than older age is beneficial.14 In addition, if the former smokers in the IWHS cohort consumed a greater number of cigarettes per day and inhaled more deeply than the subjects in the NHS, then a prolonged duration of return to baseline smoking risk would be observed in our cohort.14 Another possible explanation for the increased duration of risk could be that the former smokers relapsed to smoking. Only four lung cancer cases who were former smokers relapsed between baseline and lung cancer diagnosis. Excluding these subjects did not significantly change the results, confirming that smoking relapse is not influencing the observed prolonged duration of lung cancer risk. Furthermore, among the lung cancer cases, there were 70 current smokers at baseline who became abstinent during follow-up. We considered these former smokers as

current smokers in the analysis and may, therefore, underestimate the lung cancer risk among former smokers. Controversies exist with regard to the population effect of smoking cessation on lung cancer risk,13,20,21,34 which appears to be dependent on study design and population. Conventional case-control1 and cohort6,7,9,13,14 studies have unequivocally observed decreased lung cancer burden with smoking cessation. However, negligible and even increased lung cancer mortality from smoking cessation has been observed from randomized controlled trials15-17 and population- or community-based trials18 as well as from natural experiments.20,21 In a natural experiment,20,21 a population of smokers is considered an intervention group, a portion of which becomes smoking abstinent during the follow-up period, and never smokers are considered a control group. If a significant number of current smokers at baseline undergo smoking abstinence during the follow-up period, then one would observe the RR of current smokers to never smokers to converge toward unity. We attempted to assess the population effect in our cohort by analyzing the follow-up data to assess whether baseline current smokers were achieving smoking abstinence during follow-up. Twenty-nine percent of the baseline current smokers who had sufficient data to answer the follow-up questionnaire in 1992 quit

Fig 1. Relative risk of lung cancer among former smokers compared with current smokers, Iowa Women’s Health Study, 1986-1999. Age-adjusted risk estimates; relative risks (RRs) are 3-year moving averages.

Downloaded from ascopubs.org by 37.44.207.9 on January 27, 2017 from 037.044.207.009 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

925

LUNG CANCER RISK AFTER SMOKING CESSATION Table 4.

Relative Risk Among Former Smokers Compared With Never Smokers by Histologic Cell Type in the Iowa Women’s Health Study, 1986-1999 Pack-Years Smoked, No. of Cases (RR; 95% CI) ⱖ 20

1-19

All

Years of Smoking Abstinence

Adenocarcinoma

Squamous/Small Cell

Adenocarcinoma

Squamous/Small Cell

Adenocarcinoma

Squamous/Small Cell

0-10 11-20 21-30 ⬎ 30 All

4 (2.2; 0.8 to 6.1) 5 (2.2; 0.9 to 5.6) 9 (3.2; 1.5 to 6.8) 2 (0.8; 0.2 to 3.1) 20 (2.1; 1.2 to 3.5)

1 (—*) 0 (—†) 0 (—†) 1 (1.8; 0.2 to 13.9) 2 (0.5; 0.1 to 3.7)

31 (6.8; 4.3 to 10.7) 4 (2.4; 0.9 to 6.7) 4 (5.7; 1.8 to 18.2) 0 (—†) 39 (5.5; 3.6 to 8.5)

39 (30.6; 15.8 to 59.2) 5 (11.1; 3.9 to 31.6) 0 (—†) 0 (—†) 44 (23.0; 12.0 to 44.1)

35 (5.4; 3.5 to 8.3) 9 (2.3; 1.1 to 4.6) 13 (3.6; 1.9 to 7.0) 2 (0.7; 0.2 to 3.0) 59 (3.4; 2.3 to 5.0)

40 (21.8; 11.3 to 42.2) 5 (5.1; 1.8 to 14.5) 0 (—†) 1 (1.8; 0.2 to 13.4) 46 (10.3; 5.4 to 19.7)

Note: Cox proportional hazards regression adjusting for age, physical activity, education, body mass index, waist circumference, alcohol use, and fruit consumption. RR, relative risk; CI, confidence interval. *Relative risk could not be calculated since lung cancer case had missing data on physical activity and could not be included in the multivariate analysis. †Parameter estimates could not be calculated due to sparse data.

smoking. To correlate the rate of smoking cessation with the reduction of lung cancer risk in baseline current smokers relative to never smokers, we compared the RRs between 1986 and 1992 and 1993 and 1999 (Table 3). Although not statistically significant, we observed a 16% RR reduction in lung cancer. The discrepancy between the decline in smoking prevalence among baseline current smokers and the decline in lung cancer risk in this cohort could be explained by the following: 1) an overestimation of smoking cessation resulting from the increased likelihood of nonrespondents being continuing smokers and of quitters relapsing to smoking on later follow-up; 2) an underestimation of risk reduction resulting from the aging of the cohort; 3) baseline current smokers who quit during follow-up because of medical illness35; 4) a follow-up time that was too short to observe the lung cancer risk reduction; and 5) limited lung cancer risk reduction because cohort subjects quit smoking during follow-up, creating a pool of recent quitters. Previous research has shown that lung cancer risk reduction with smoking cessation is more consistent in squamous cell and small cell than with adenocarcinoma,10,11 which we observed in this cohort. There was a nonsignificantly lower risk for adenocarcinoma but a significantly lower risk of squamous cell and small cell among former heavier smokers compared with current heavier smokers in the first 10 years of smoking abstinence. This observation is consistent with the theory that the smokinginduced pathologic changes leading to an increased risk for lung cancer are not reversed after smoking cessation to the same degree for adenocarcinoma as they are for squamous cell and small cell lung cancer. We did not observe a consistent inverse relationship between the risk for adenocarcinoma in former smokers compared with never smokers with time since smoking cessation. This finding may be an artifact caused by the few number of cases of adenocarcinoma diagnosed in the third decade of smoking abstinence or may be influenced by other factors for adenocarcinoma among former female smokers, such as the consumption of filtered or “low tar” cigarettes. Researchers have postulated that one of the chief factors accounting for the persistently increased risk of adenocarcinoma of the lung after smoking cessation is the use of filters, which may result in deeper inhalation and prolonged breath holding to compensate for lower nicotine yields.36-38 Deep inhalation may deposit smoke particles in the periphery of the lung, where most adenocarcinomas arise.

In a study using nine population-based cancer registries, the rates of squamous cell carcinoma and adenocarcinoma along the airways corresponded closely with the deposition pattern of large and small smoke particles, with the smaller smoke particles distributing predominantly to the periphery.39 Because adenocarcinoma tends to arise in the peripheral lung tissue, the clinical diagnosis may be delayed because they are less likely to cause airway obstruction and clinical symptoms such as hemoptysis.40 Furthermore, the mean volume doubling time of adenocarcinoma is twice that of squamous cell carcinoma,41 and these tumors are more likely to arise in a smoking-induced diseased lung.40 All of these factors may contribute to the prolonged duration of lung adenocarcinoma risk and the inconsistent risk decline after smoking cessation. One of the potential limitations of this study relates to the lack of data on environmental tobacco smoke (ETS) exposure. This may result in an underestimation of the RR of lung cancer caused by tobacco smoke inasmuch as IWHS participants without a smoking history but with significant ETS have been classified as never smokers. This effect is likely to be small relative to the effect of mainstream smoke. Another potential limitation relates to the differences between the responders and nonresponders in the cohort with regard to their smoking histories,24 also known as the “healthy responders” effect.42 This phenomenon is reflected by an elevated disease risk in the early years of a prospective cohort study, which decreases with increasing duration of follow-up. During the first 5 years of follow-up in the IWHS cohort, Bisgard et al24 reported a 40% higher lung cancer rate among nonresponders than among responders. During our 14 years of follow-up, lung cancer incidence was 28% lower (133 v 170 per 100,000) in the cohort than among Iowa women in the same age range. This effect should not change the internal validity of the study but could potentially influence the generalizability of our findings to other populations. In addition, our RR estimates of lung cancer risk using the entire cohort of women (N ⫽ 41,836) were similar to those of the analytic cohort after baseline exclusions, indicating that these exclusions did not bias our results by creating an artificially healthy cohort of women at baseline. We confirm the declining risk of lung cancer among former smokers compared with never smokers in a cohort of older Midwestern females, but we observed that this excess lung cancer risk persisted up to 30 years. Health care providers need

Downloaded from ascopubs.org by 37.44.207.9 on January 27, 2017 from 037.044.207.009 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

926

EBBERT ET AL

to continue to be aggressive about diagnosing and treating tobacco use disorders at all ages because the weight of all evidence indicates that smoking cessation decreases lung cancer risk. However, more resources may need to be focused on prevention of smoking initiation and chemoprevention of lung

cancer because former smokers have significantly prolonged elevated risk of lung cancer after smoking cessation. ACKNOWLEDGMENT We thank Susan Ernst for her technical assistance with the manuscript.

REFERENCES 1. The Health Benefits of Smoking Cessation: A Report of the Surgeon General. Washington, DC, United States Department of Health and Human Services, Public Health Service, 1990 2. Wigle DT, Mao Y, Grace M: Relative importance of smoking as a risk factor for selected cancers. Can J Public Health 71:269-275, 1980 3. Alderson MR, Lee PM, Wang R: Risks of lung cancer, chronic bronchitis, ischaemic heart disease, and stroke in relation to type of cigarette smoked. J Epidemiol Community Health 39:286-293, 1985 4. Lubin JH, Blot WJ, Berrino F, et al: Modifying risk of developing lung cancer by changing habits of cigarette smoking. Br Med Bull 288:1953-1956, 1984 5. Jockel KH, Ahrens W, Wichmann HE, et al: Occupational and environmental hazards associated with lung cancer. Int J Epidemiol 21:202-213, 1992 6. Burns DM: Primary prevention, smoking, and smoking cessation: Implications for future trends in lung cancer prevention. Cancer 89:25062509, 2000 7. Speizer FE, Colditz GA, Hunter DJ, et al: Prospective study of smoking, antioxidant intake, and lung cancer in middle-aged women (USA). Cancer Causes Control 10:475-482, 1999 8. Gao YT, Blot WJ, Zheng W, et al: Lung cancer and smoking in Shanghai. Int J Epidemiol 17:277-280, 1988 9. Strauss GM, Dominioni L: Perception, paradox, paradigm: Alice in the wonderland of lung cancer prevention and early detection. Cancer 89:24222431, 2000 (suppl) 10. Higgins IT, Wynder EL: Reduction in risk of lung cancer among ex-smokers with particular reference to histologic type. Cancer 62:23972401, 1988 11. Lubin JH, Blot WJ: Assessment of lung cancer risk factors by histologic category. J Natl Cancer Inst 73:383-389, 1984 12. Khuder SA, Mutgi AB: Effect of smoking cessation on major histologic types of lung cancer. Chest 120:1577-1583, 2001 13. Enstrom JE: Trends in mortality among California physicians after giving up smoking: 1950-79. BMJ 286:1101-1105, 1983 14. Halpern MT, Gillespie BW, Warner KE: Patterns of absolute risk of lung cancer mortality in former smokers. J Natl Cancer Inst 85:457-464, 1993 15. Anthonisen NR, Connett JE, Kiley JP, et al: Effects of smoking intervention and the use of an inhaled anticholinergic bronchodilator on the rate of decline of FEV1. The Lung Health Study. JAMA 272:1497-1505, 1994 16. Shaten BJ, Kuller LH, Kjelsberg MO, et al: Lung cancer mortality after 16 years in MRFIT participants in intervention and usual-care groups. Multiple Risk Factor Intervention Trial. Ann Epidemiol 7:125-136, 1997 17. The Multiple Risk Factor Intervention Trial Research Group: Mortality after 16 years for participants randomized to the Multiple Risk Factor Intervention Trial. Circulation 94:946-951, 1996 18. Rose G, Hamilton PJ, Colwell L, et al: A randomised controlled trial of antismoking advice: 10-year results. J Epidemiol Community Health 36:102-108, 1982 19. Ebrahim S, Smith GD: Systematic review of randomised controlled trials of multiple risk factor interventions for preventing coronary heart disease. BMJ 314:1666-1674, 1997 20. Enstrom JE: Smoking cessation and mortality trends among two United States populations. J Clin Epidemiol 52:813-825, 1999

21. Enstrom JE, Heath CW Jr: Smoking cessation and mortality trends among 118,000 Californians, 1960-1997. Epidemiology 10:500-512, 1999 22. Folsom AR, Kaye SA, Prineas RJ, et al: Increased incidence of carcinoma of the breast associated with abdominal adiposity in postmenopausal women. Am J Epidemiol 131:794-803, 1990 23. Kushi LH, Fee RM, Folsom AR, et al: Physical activity and mortality in postmenopausal women. JAMA 277:1287-1292, 1997 24. Bisgard KM, Folsom AR, Hong CP, et al: Mortality and cancer rates in nonrespondents to a prospective study of older women: 5-year follow-up. Am J Epidemiol 139:990-1000, 1994 25. Kushi LH, Kaye SA, Folsom AR, et al: Accuracy and reliability of self-measurement of body girths. Am J Epidemiol 128:740-748, 1988 26. Ries LAG, Eisen MP: SEER Cancer Statistics Review, 1973-1997. Bethesda, MD, National Cancer Institute, 2000 27. Fritz AG, Ries LAG: The SEER Program Code Manual (ed 3). Bethesda, MD, National Cancer Institute, 1998 28. Percy C, Van Holten V, Muir C: International Classification of Diseases for Oncology (ed 2). Geneva, World Health Organization, 1990 29. Korn EL, Graubard BI: Time-to-event analysis of longitudinal follow-up of a survey: Choice of the time-scale. Am J Epidemiol 145:72-80, 1997 30. Drinkard CR, Sellers TA, Potter JD, et al: Association of body mass index and body fat distribution with risk of lung cancer in older women. Am J Epidemiol 142:600-607, 1995 31. Potter JD, Sellers TA, Folsom AR, et al: Alcohol, beer, and lung cancer in postmenopausal women. The Iowa Women’s Health Study. Ann Epidemiol 2:587-595, 1992 32. Steinmetz KA, Potter JD, Folsom AR: Vegetables, fruit, and lung cancer in the Iowa Women’s Health Study. Cancer Res 53:536-543, 1993 33. Pathak DR, Samet JM, Humble CG, et al: Determinants of lung cancer risk in cigarette smokers in New Mexico. J Natl Cancer Inst 76:597-604, 1986 34. Peto R, Darby S, Deo H, et al: Smoking, smoking cessation, and lung cancer in the UK since 1950: combination of national statistics with two case-control studies. BMJ 321:323-329, 2000 35. Kuller LH: Variance and dissent. Dissent rebuttal to the paper by Enstrom. J Clin Epidemiol 52:827-829, 1999 36. Herning RI, Jones RT, Bachman J, et al: Puff volume increases when low-nicotine cigarettes are smoked. BMJ 83:187-189, 1981 37. Nemeth-Coslett R, Griffiths R: Determinants of puff duration in cigarette smokers. Pharmacol Biochem Behav 21:903-912, 1984 38. Stellman SD, Muscat JE, Thompson S, et al: Risk of squamous cell carcinoma and adenocarcinoma of the lung in relation to lifetime filter cigarette smoking. Cancer 80:382-388, 1997 39. Yang CP, Gallagher RP: Differences in incidence rates of cancers of the respiratory tract by anatomic subsite and histologic type: an etiologic implication. J Natl Cancer Inst 81:1828-1831, 1989 40. Iannuzzi MC, Toews GB: Neoplasms of the Lung, in Stein JH (ed): Internal Medicine (ed 5). St. Louis, MO, Mosby, 1998 41. Steel GG: Growth Kinetics of Tumors: Cell Population Kinetics in Relation to the Growth and Treatment of Cancer. Clarendon, Oxford Press, 1977 42. Prescott E, Andersen PK, Osler M, et al: A sensitivity analysis of secular trends in risk factors and mortality based on cohort studies. Epidemiology 10:176-180, 1999

Downloaded from ascopubs.org by 37.44.207.9 on January 27, 2017 from 037.044.207.009 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.