impact of the reduction in tobacco smoking on lung ...

4 downloads 0 Views 982KB Size Report
IMPACT OF THE REDUCTION IN TOBACCO SMOKING. ON LUNG CANCER MORTALITY IN THE U.S. OVER THE. PERIOD 1975-2000. Suresh H. Moolgavkar,.
Editorial Manager(tm) for PLoS Medicine Manuscript Draft Manuscript Number: PMEDICINE-D-11-00021R1 Title: Impact of the Reduction in Tobacco Smoking on Lung Cancer Mortality in the U.S. over the period 1975-2000 Short Title: Impact of tobacco control on LC mortality Article Type: Research Article Keywords: Tobacco control efforts; Lung cancer risk; Cancer risk prediction models; Cancer Intervention and Surveillance Modeling Network (CISNET) Corresponding Author: Suresh Moolgavkar Corresponding Author's Institution: Fred Hutchinson Cancer Research Center First Author: Suresh Moolgavkar Order of Authors: Suresh Moolgavkar;Theodore R. Holford;David T. Levy;Chung Yin Kong;Millenia Foy;Lauren Clarke;Jihyoun Jeon;William Hazelton;Rafael Meza;Frank Schultz;William McCarthy;Robert Boer;Olga Gorlova;G. Scott Gazelle;Marek Kimmel;Pamela M. McMahon;Harry J. de Koning;Eric J. Feuer Abstract: Background: Considerable effort has been expended on tobacco control strategies in the United States since the mid-fifties. We quantify the impact of the associated changes in smoking habits on lung cancer mortality in the U.S. over the period 1975-2000. Methods and Findings: A consortium of six groups using common inputs and independent models estimated the number of U.S. lung cancer deaths averted over the period 1975-2000 as a result of changes in smoking behavior beginning in the mid-fifties. We also consider the number of deaths that could have been averted had tobacco control been completely effective in eliminating smoking in 1965, following issuance of the first Surgeon General's report (SGR) on Smoking and Health in 1964. Approximately 795,000 deaths were averted over the period 1975-2000 (550,000 among males and 245,000 among females). In the year 2000 alone approximately 70,000 lung cancer deaths were averted (44,000 among males and 26,000 among females). However, only approximately 30% of lung cancer deaths that could have potentially been averted over the period 1975-2000 were actually saved. In the ten-year period 1991-2000, this fraction increased to about 37%. Conclusions: Despite tremendous strides in tobacco control, and an associated major impact on lung cancer deaths, lung cancer remains a major public health problem and continued efforts at tobacco control are key to reducing further the burden of this disease. This disappointing reduction of only a small fraction of deaths is because a) tobacco control efforts did not have an immediate effect on smoking rates, b) the elevated risk of lung cancer persists many years after smoking cessation, and c) over 20% of the U.S. adult population continues to smoke. Enhanced tobacco control efforts are necessary to reduce further the burden of lung cancer in the U.S. population. Suggested Reviewers: Donald A. Berry

M.D. Anderson [email protected] Dik Habbema Erasmus MC [email protected] Philip C. Prorok National Cancer Institute [email protected] Jonathan M. Samet University of Southern California [email protected] Daniel Krewski U. Ottawa [email protected] Opposed Reviewers:

Cover Letter

The research briefly described in the abstract was conducted by a consortium of 6 universities and research institutions, 5 in the U.S. and 1 in Europe, and was funded by the CISNET program of the National Cancer Institute. As the corresponding author, I am submitting this presubmission inquiry on behalf of 17 other co-authors. This paper represents the first systematic attempt at quantifying the impact of changes in tobacco smoking behavior on lung cancer risk in the U.S. This study shows that, while changes in tobacco smoking habits have had a major impact on lung cancer mortality, these gains represent only a small fraction of the lung cancer deaths that could have been avoided had tobacco control been completely successful in ending smoking. As one of the inputs for the models used in the paper, we have developed a smoking history generator that allows the detailed description of smoking behaviors by gender and year of birth starting with the birth cohort of 1890 and ending with the birth cohort of 1970. The smoking history generator will be made freely available to all interested investigators. The development of the smoking history generator and the models used for analyses represents a major advance in methodology that should be applicable to other problems. We would appreciate your assessment of whether a research article based on these results would be appropriate for the journal. The paper is not under consideration at another journal. Key References 1. S. H. Moolgavkar, A. G. Knudson Jr, Mutation and cancer: a model for human carcinogenesis. J. Natl. Cancer Inst. 66, 1037-1052 (1981). 2. R. Meza, W. D. Hazelton, G. A. Colditz, S. H. Moolgavkar, Analysis of lung cancer incidence in the Nurses' Health and the Health Professionals' Follow-Up Studies using a multistage carcinogenesis model. Cancer Causes Control 19, 317328 (2008). 3. B. Rachet, J. Siemiatycki, M. Abrahamowicz, K. Leffondre, A flexible modeling approach to estimating the component effects of smoking behavior on lung cancer. J. Clin. Epidemiol. 57, 1076-1085 (2004). 4. M. J. Thun, A. Jemal, How much of the decrease in cancer death rates in the United States is attributable to reductions in tobacco smoking? Tobacco Control 15, 345-347 (2006). 5. P. M. McMahon et al., Adopting helical CT screening for lung cancer: potential health consequences during a 15-year period. Cancer 113, 3440-3449 (2008). 6. R. Meza, J. Jeon, S. H. Moolgavkar, E. G. Luebeck, Age-specific incidence of cancer: Phases, transitions, and biological implications. Proc. Natl. Acad. Sci. U. S. A. 105, 16284-16289 (2008). 7. W. D. Flanders, C. A. Lally, B. P. Zhu, S. J. Henley, M. J. Thun, Lung cancer mortality in relation to age, duration of smoking, and daily cigarette consumption: results from Cancer Prevention Study II. Cancer Res. 63, 6556-6562 (2003). 8. D. T. Levy, L. Nikolayev, E. Mumford, Recent trends in smoking and the role of public policies: results from the SimSmoke tobacco control policy simulation model. Addiction 100, 1526-1536 (2005).

Manuscript Click here to download Manuscript: Moolgavkarms_PLoS.doc

IMPACT OF THE REDUCTION IN TOBACCO SMOKING ON LUNG CANCER MORTALITY IN THE U.S. OVER THE PERIOD 1975-2000 Suresh H. Moolgavkar,1* Theodore R. Holford,2 David T. Levy,3,4 Chung Yin Kong,5,6 Millenia Foy,7,8 Lauren Clarke,9 Jihyoun Jeon,1 William Hazelton,1 Rafael Meza,1 Frank Schultz,10 William McCarthy,11,12 Robert Boer,10 Olga Gorlova,7 G. Scott Gazelle5,6, Marek Kimmel,8,13 Pamela M. McMahon,5,6 Harry J. de Koning,10 Eric J. Feuer14

1

Program in Biostatistics and Biomathematics, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue North, Seattle, WA 98109, USA

2

Department of Epidemiology and Public Health, Yale University School of Medicine, 60 College Street, New Haven, CT 06511, USA

3

Department of Economics, University of Baltimore, N Charles St Baltimore, MD 21201, USA

4

Pacific Institute for Research and Evaluation, 11720 Beltsville Drive, Suite 900 Calverton, MD 20850, USA

5

Massachusetts General Hospital, Boston, MA 02114, USA

6

Harvard Medical School, 25 Shattuck Street, Boston, MA 02115,USA

7

The Brown Foundation Institute of Molecular Medicine, University of Texas Health Science Center at Houston, 7000 Fannin, Houston, TX 77030, USA

8

Department of Epidemiology, University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd. Houston, TX 77030, USA

1

9

Cornerstone Systems Northwest Inc., 8665 Berthusen Road, Lynden, WA 98264, USA

10

Erasmus MC, Department of Public Health, 3000 CA Rotterdam, The Netherlands

11

Department of Health Services, School of Public Health, University of California, 650 Young Drive, Los Angeles, CA 90095, USA

12

Department of Psychology, University of California, 650 Young Drive, Los Angeles, CA 90095, USA

13

Systems Engineering Group, Silesian University of Technology, Gliwice, Poland

14

Division of Cancer Control and Population Sciences, National Cancer Institute, 6116 Executive Boulevard, Bethesda, MD 20892, USA

Abbreviations: SGR, Surgeon General's report; CISNET, Cancer Intervention and Surveillance Modeling Network; SHG, smoking history generator; ATC, actual tobacco control; NTC, no tobacco control; CTC, complete tobacco control; TSCE, two-stage clonal expansion; SEER, Surveillance, Epidemiology and End Results

*To whom correspondence should be addressed. E-mail: [email protected]

2

Abstract

Background: Considerable effort has been expended on tobacco control strategies in the United States since the mid-fifties. We quantify the impact of the associated changes in smoking habits on lung cancer mortality in the U.S. over the period 1975-2000.

Methods and Findings: A consortium of six groups using common inputs and independent models estimated the number of U.S. lung cancer deaths averted over the period 1975-2000 as a result of changes in smoking behavior beginning in the mid-fifties. We also consider the number of deaths that could have been averted had tobacco control been completely effective in eliminating smoking in 1965, following issuance of the first Surgeon General's report (SGR) on Smoking and Health in 1964.

Approximately 795,000 deaths were averted over the period 1975-2000 (550,000 among males and 245,000 among females). In the year 2000 alone approximately 70,000 lung cancer deaths were averted (44,000 among males and 26,000 among females). However, only approximately 30% of lung cancer deaths that could have potentially been averted over the period 1975-2000 were actually saved. In the ten-year period 1991-2000, this fraction increased to about 37%.

Conclusions: Despite tremendous strides in tobacco control, and an associated major impact on lung cancer deaths, lung cancer remains a major public health problem and continued efforts at tobacco control are key to reducing further the burden of this disease. This disappointing reduction of only a small fraction of deaths is because a) tobacco control efforts did not have an immediate effect on smoking rates, b) the elevated risk of lung cancer persists many years after smoking cessation, and c) over 20% of the U.S. adult population continues to smoke. Enhanced tobacco control efforts are necessary to reduce further the burden of lung cancer in the U.S. population.

3

Introduction

Steadily increasing anti-smoking norms in the form of cumulative restrictions on smoking in public places, large increases in cigarette excise taxes, reduced access to cigarettes as well as increased public awareness of the hazards of smoking – collectively known as “tobacco control,” have been cited as the principal contributors to the observed decline in U.S. adult tobacco use from 1975 to 2003 [1, 2] and to subsequent declines in smoking-related mortality [3]. In this paper, we undertake a detailed analysis of the direct influence of changes in smoking behaviors that began in the mid 1950s on lung cancer mortality rates among males and females in the U.S. over the period 1975-2000. We estimate also the total number of lung cancer deaths averted in males and females over the same period as a direct result of changes in smoking behavior. Finally, we estimate the numbers of avoidable deaths, i.e., the number of lung cancer deaths that could have been averted had smoking been completely eliminated as of 1965.

Our estimates in this paper are based on six different models developed by a consortium of universities and research centers1 selected by the NIH peer review process. The research reported in this paper was funded by the National Cancer Institute’s Cancer Intervention and Surveillance Modeling Network (CISNET - http://cisnet.cancer.gov/). Although the models shared common inputs, each group developed its own model, based on mathematical descriptions of lung carcinogenesis as it relates to smoking behaviors. The models explicitly consider factors associated with the risk of smoking, including the number of cigarettes smoked per day, age of initiation and the number of years quit.

1

There were six centers involved in this project: Erasmus Medical Center (Erasmus MC), Fred Hutchinson Cancer Research Center (FHCRC), Pacific Institute for Research and Evaluation (PIRE), Rice University- M.D. Anderson Cancer Center (Rice-MDA), Massachusetts General Hospital-Harvard Medical School (MGH-HMS), and Yale University.

4

Methods

Figure 1 shows the overall structure shared by all models. The central component of each model is a dose-response module that provides a quantitative description of the agespecific mortality of lung cancer among never smokers, and among continuing smokers and former smokers by detailed history of smoking. This module was used to predict agespecific lung cancer mortality rates under specific smoking scenarios. With the exception of the MGH-HMS group, which used a set of logistic regression models and tumor progression functions [4, 5], the other groups used multistage models [6-9] for the underlying dose-response module. Multistage models, based on mathematical formalisms of the biological paradigm of initiation, promotion, and progression, recognize that carcinogenesis is a process of mutation accumulation and clonal expansion of partially altered cells on the pathway to malignancy [6, 10-15]. These models may be used to explore biological hypotheses regarding the mechanism of tobacco-induced lung cancer and have generally shown clonal expansion (promotion) of partially altered (initiated) cells by cigarette smoke to be the dominant mechanism, and have confirmed the disproportionate importance of smoking duration on lung cancer risk [8,9,16-19]. Both the multistage and the MGH-HMS probabilistic models are capable of accommodating detailed individual-level smoking histories, including temporal factors, such as age at start, age at cessation, and temporal changes in the level of smoking. The parameters of these models were estimated as described in the supporting information (Text S1) by fitting the model to specific epidemiologic cohort (Erasmus MC, FHCRC, PIRE, Yale), case-control (Rice-MDA), or registry (MGH-HMS) data.

Three specific smoking scenarios, the common inputs for the models were simulated using a smoking history generator (SHG) as briefly described below. Each involved a detailed description of smoking behaviors by gender and birth cohort starting with the birth cohort of 1890 and ending with the birth cohort of 1970. The actual tobacco control (ATC) scenario is a quantitative description of the actual smoking behaviors of males and

5

females in the U.S. The no tobacco control (NTC) scenario is a quantitative description of the predicted smoking behaviors of males and females in the U.S. under the assumption that tobacco control efforts starting mid-century had never been implemented. The complete tobacco control (CTC) scenario is a quantitative description of the predicted smoking behaviors of males and females in the U.S. under the assumption that all smoking ceased abruptly in 1965, i.e., all smokers quit permanently at that time and there was no initiation of smoking after 1965.

The models used in CISNET did not incorporate other known or suspected risk factors, such as environmental tobacco smoke and radon exposure, diet, and air pollution [20-22] that could have influenced trends in lung cancer mortality in the U.S. Moreover, the data sets from which the parameters of the individual dose-response modules were estimated may not be representative of the U.S. population. For these reasons the outputs of the models under the ATC scenario cannot be expected to reproduce the observed lung cancer rates in the U.S. population. Rather, without further calibration, these models estimate lung cancer rates in hypothetical populations with the smoking behaviors and the age structure of the U.S. population over the period 1975-2000. To compensate for these limitations, some groups (FHCRC, MGH-HMS, Yale, PIRE) chose to calibrate their models further to describe actual deaths in the U.S. population under the ATC scenario over the period 1975-2000. This calibration was achieved by embedding the doseresponse module in an age-period-cohort model, except for the PIRE model, which used only a period calibration. Other groups (Erasmus MC, Rice-MDA) chose not to perform this additional calibration. The overall model structure and the specific models used by each group are described in greater detail in the supporting information (Text S1).

For each smoking scenario (ATC, NTC, CTC), the SHG provides as its output detailed individual level smoking histories for individuals born between 1890 and 1985 over the period 1975-2000. Figure 2 shows one output of the SHG, the proportion of current

6

smokers, under the three smoking scenarios. Deaths from causes other than lung cancer are also simulated by level of smoking; each individual history is terminated at age 84 or at the age of death from a cause other than lung cancer if the death occurred prior to age 84. Details of the construction of the SHG can be found on the CISNET website (link provided in the supporting information (Text S1)). Using the outputs of the SHG each group estimated the number of lung cancer deaths over the period 1975-2000 while adjusting for other cause mortality for each smoking scenario.

Results

The models yielded a range of results for the numbers of lung cancer deaths under the three smoking scenarios, but the estimates of the fraction of lung cancer deaths averted were reasonably consistent across models. For purposes of illustration, we chose one of the models calibrated against the U.S. data (the Yale model) as an exemplar model to present our results. Based on this model, Figure 3 shows the age-adjusted rates (upper panel) and the actual number of lung cancer deaths (lower panel) among males and females in the U.S. over the period 1975-2000, and also the numbers that would have been expected under the NTC and CTC scenarios. Over the period 1975-2000, there were 2,067,775 lung cancer deaths among males and 1,051,978 lung cancer deaths among females in the U.S. Under NTC, the Yale model estimates 2,670,897 lung cancer deaths among males and 1,273,151 lung cancer deaths among females, while under CTC, the deaths number 958,862 and 438,858, respectively, among males and females. The difference between the NTC and observed numbers provides an estimate of the numbers of lung cancer deaths averted (A), which for the Yale model are 603,122 and 221,173 for males and females, respectively. The corresponding numbers reported in the abstract are averages over all models (Table 1). The difference between NTC and CTC is an estimate of the total number of lung cancer deaths that could have been averted if tobacco control efforts had been immediately and completely successful (B), which are approximately 1,712,035 and 834,293 for males and females respectively for the Yale model.

7

The other models calibrated to U.S. mortality yielded similar estimates of the number of lung cancer deaths under the three scenarios (not shown). Counts of the differences in the number of deaths between scenarios are shown for all models in Table 1 and Figure S1 in the supporting information. The ratio of deaths averted to total deaths that could potentially have been avoided (i.e. A/B) is also presented in Table 1. The models estimate that of all avoidable deaths from smoking-related lung cancer, between 24 and 32 percent among females and between 30 and 37 percent among males were actually averted as a result of the changes in smoking behaviors that actually began in the mid-fifties some years before the first SGR. For both genders combined, approximately 30% (28% to 35% across models) of all avoidable deaths were averted. Table 1 also shows the impact of tobacco control efforts on lung cancer mortality for the decade 1991-2000 and for the year 2000. In the decade 1991-2000, the fraction of lung cancer deaths averted in males and females combined increased to about 37% (34% to 43% across models). In the year 2000, this fraction increased to roughly 45% (39% to 50% across models). The increasing trend in the fraction of lung cancer deaths averted reflects both changes in smoking behaviors and a continuing decrease in risk among former smokers.

Discussion

Using a straightforward demographic projection, Thun and Jemal [23] estimated that reductions in tobacco smoking averted approximately 146,000 lung cancer deaths among U.S. males over the period 1991-2003. Our analyses in this paper are based on detailed recreation of cigarette smoking histories (omitting cigars and pipes) under distinct tobacco control scenarios as inputs for mathematical models relating these smoking histories to lung cancer mortality. We have used a comparative modeling approach to address this complex problem. Comparative modeling produces a range of results across models but, when these are reasonably consistent, enhances their credibility. Table 1 shows that the reductions in lung cancer mortality secondary to reduced tobacco use have

8

been much larger than those estimated by Thun and Jemal. Over the period 1991-2000, approximately 345,000 lung cancer deaths among U.S. males and 175,000 deaths among U.S. females were averted due to changes in smoking behaviors starting in the midfifties2. In the year 2000 alone, approximately 44,000 deaths among U.S. males and 26,000 deaths among U.S. females were averted. Over the period 1975-2000, approximately an additional 1,500,000 lung cancer deaths among males and females combined could have been averted had tobacco control efforts been completely effective in eliminating smoking as of 1965. These numbers greatly under-estimate the overall health impact of tobacco control efforts because they do not consider the substantial impact of tobacco smoking behaviors on diseases other than lung cancer. Smokingassociated diseases other than lung cancer were outside the scope of this work.

Even though other factors, including genetic polymorphisms [24], contribute to lung cancer risk, the vast majority of lung cancer cases could be eliminated by eliminating smoking. The results of this paper show the dramatic impact of the reduction in smoking associated with tobacco control efforts in the second half of the 20th century on lung cancer mortality over the period 1975-2000.

It is not surprising that the various models used in this paper yield a range of estimates of the fraction of lung cancer deaths averted by the tobacco control efforts in the U.S. This range of results represents the uncertainty associated with model choice. First, some of these models were calibrated against U.S. mortality data and, as a consequence, these models describe the lung cancer mortality trends in the U.S. quite well under the ATC scenario. Second, even though five of the six groups used the two-stage clonal expansion (TSCE) version of multistage models [6, 8, 9] as the dose-response module, the estimated parameters were different because they were estimated by fits to different cohorts. The

2

These and other estimates in the discussion are based on averages of the numbers across

models presented in Table 1.

9

Yale group used the models developed by Knoke et al. [7] and Flanders et al. [25] as well with similar results for the relative effect of tobacco control. It is well known that the risks of tobacco smoking have changed over time and could be modified, moreover, by other factors, such as diet, not accounted for in any of the models. Despite these limitations, the estimated numbers of deaths averted and deaths that could have been averted under the assumption of CTC are reasonably consistent across models (Table 1). The main message of these analyses is clear. Tobacco control strategies implemented mid-century have averted hundreds of thousands of lung cancer deaths in the U.S. over the period 1975-2000, but these are only approximately 30% of the lung cancer deaths that could have been averted had all cigarette smoking come to an end in 1965.

The FHCRC, MGH-HMS, and Yale groups calibrated their models to U.S. mortality over the period 1975-2000 using birth cohort and period effects. These calibrations are necessary to describe lung cancer mortality rates and trends in the U.S. and indicate that the lung cancer mortality experience of the entire population cannot be adequately described by extrapolating from the SEER registry in one decade3, or from various cohort and case-control studies of smoking and lung cancer (please see the supporting information for the data sets used by each of the groups for parameter estimation). Particularly among males, U.S. lung cancer mortality is considerably higher than would be expected from the cohort studies against which the dose-response modules were calibrated. In addition, models from cohort studies and available population smoking histories cannot adequately describe temporal components of trend, i.e., the effects of age, period and cohort.

There could be several reasons for the poor prediction of population lung cancer rates without additional model calibrations. First, the data sets used for estimating the

3

The MGH-HMS group estimated parameters of their dose-response module from fits to

the SEER registry data as described in the supporting information (Text S1).

10

parameters of the dose-response modules are almost certainly not representative of the U.S. population. Second, the SHG is based on smoking histories for birth cohorts in the general population inferred by reconstructing through simulation using cross-sectional histories that rely on recall of events that often will have occurred years earlier. Third, potentially important covariates, such as diet, air pollution, radon exposure and occupational exposures including asbestos and ionizing radiation are not available for the overall population, and different exposure distributions may contribute to rate discrepancies. Fourth, the models discussed assume a consistent effect of exposure on lung cancer mortality, but temporal changes in the manufacture of cigarettes and smoking behaviors could explain some of the discrepancies in trend. However, data on changes in cigarette manufacturing and composition are not readily available. Changes in tobacco or cigarette composition, which are not explicitly addressed in these analyses, could be important contributors to population trends in lung cancer mortality. However, one would expect changes in tobacco or cigarette composition to manifest themselves as period effects, whereas models that used age-period-cohort calibrations find that trends are dominated by birth cohort effects. Finally, uncertainty remains with respect to the models themselves, as suggested by model variation among studies.

In particular, the lung cancer rates in the U.S. population under the CTC scenario appear to be higher than would have been expected on the basis of recent work on lung cancer rates among never-smokers [24]. But, for the reasons given above, the never-smoker rates reported by Thun et al. [24] may not reflect the never-smoker rates in the general population. Some confidence in the lung cancer rates under the CTC scenario estimated from the models in this paper can be derived from the fact that the dose-response modules describe lung cancer rates among former smokers well [7, 8].

One limitation of the calibrations is that the same period and cohort parameters are applied to current smokers, former smokers, and never smokers. Factors, such as diet, that could affect trends in lung cancer rates might be expected to have different effects

11

among current smokers, former smokers and never smokers. However, different cohort and period effects cannot be estimated in these sub-groups because of identifiability issues. The FHCRC group did fit period and cohort effects to never smokers alone and to current smokers and former smokers separately, but the original model in which these effects are applied equally to all groups described the data better as judged by the Akaike Information Criterion [26].

Our study shows that changes in smoking behaviors led to a substantial reduction in the lung cancer mortality that would have been expected had the smoking trends in the 1950s continued into the future. Our analysis was conducted through to the year 2000, the latest year for which we were able to obtain sufficiently detailed data when this project was initiated. The trends indicate that the benefits will continue to grow over time, although the models will need to be validated when data for the year 2000 and forward become available in order to make predictions with confidence. Consistent with trends for continued gains due to past tobacco control policies, smoking prevalence continued to fall from 23.2% in 2000 to 20.6% in 2008. Much of this decrease can be attributed to tobacco control policies, especially the cigarette price increases in 1998-1999 [27].

Our results indicate that only approximately 30% of the total lung cancer deaths that could have been averted had tobacco control been complete were actually saved. This is because smoking rates took time to decline after the first SGR in 1965, the elevated risk of lung cancer remains for many years after smoking cessation, and a sizable fraction of the population continues to smoke. Clearly, further reductions in smoking rates are required to reduce lung cancer incidence and mortality rates substantially. The recently reported 20% reduction in lung cancer mortality [28] as a result of early detection using low-dose spiral CT suggests that screening of high risk individuals may have a role in reducing mortality from this disease. However, continued implementation of evidence based tobacco control policies, programs, and services remains the most promising approach to reducing the burden of lung cancer.

12

Supporting Information Figure S1. Comparison of model results for actual and potential deaths avoided over the period 1975-2000. Table S1. Key model attributes.

Text S1.

Technical appendix. The supporting figure and table are available as

individual files (Figure S1 and Table S1), but are also included here for ease of access.

Acknowledgments

We thank Drs David Burns, Jay H. Lubin, Michelle Bloch, Cathy Backinger, Wilson M. Compton, Kevin P. Conway, Brenda K. Edwards for helpful comments.

References

1. Cokkinides V, Bandi P, McMahon C, Jemal A, Glynn T, et al. (2009) Tobacco control in the United States--recent progress and opportunities. CA Cancer J Clin 59: 352-365. 2. (2000) Reducing Tobacco Use: A Report of the Surgeon General. Available: http://www.cdc.gov/tobacco/data_statistics/sgr/2000 Accessed 18 January 2011. 3. Doll R, Peto R, Boreham J, Sutherland I (2004) Mortality in relation to smoking: 50 years' observations on male British doctors. BMJ 328(7455): 1519.

13

4. McMahon PM, Kong CY, Weinstein MC, Tramontano AC, Cipriano LE, et al. (2008) Adopting helical CT screening for lung cancer: potential health consequences during a 15-year period. Cancer 113: 3440-3449. 5. McMahon PM, Kong CY, Johnson BE, Weinstein MC, Weeks JC, et al. (2008) Estimating long-term effectiveness of lung cancer screening in the Mayo CT screening study. Radiology 248: 278-287. 6. Moolgavkar SH, Knudson AG (1981) Mutation and cancer: a model for human carcinogenesis. J Natl Cancer Inst 66: 1037-1052. 7. Knoke JD, Shanks TG, Vaughn JW, Thun MJ, Burns DM (2004) Lung cancer mortality is related to age in addition to duration and intensity of cigarette smoking: an analysis of CPS-I data. Cancer Epidemiol Biomarkers Prev 13: 949957. 8. Hazelton WD, Clements MS, Moolgavkar SH (2005) Multistage carcinogenesis and lung cancer mortality in three cohorts. Cancer Epidemiol Biomarkers Prev 14: 1171-1181. 9. Meza R, Hazelton WD, Colditz GA, Moolgavkar SH (2008) Analysis of lung cancer incidence in the Nurses' Health and the Health Professionals' Follow-Up studies using a multistage carcinogenesis model. Cancer Causes Control 19: 317328. 10. Little MP (1995) Are two mutations sufficient to cause cancer? Some generalizations of the two-mutation model of carcinogenesis of Moolgavkar, Venzon, and Knudson, and of the multistage model of Armitage and Doll. Biometrics 51: 1278-1291. 11. Kopp-Schneider A (1997) Carcinogenesis models for risk assessment. Stat Methods Med Res 6: 317-340. 12. Heidenreich WF, Wellmann J, Jacob P, Wichmann HE (2002) Mechanistic modelling in large case-control studies of lung cancer risk from smoking. Stat Med 21: 3055-3070.

14

13. Luebeck EG, Moolgavkar SH (2002) Multistage carcinogenesis and the incidence of colorectal cancer. Proc Natl Acad Sci U S A 99: 15095-15100. 14. Meza R, Jeon J, Moolgavkar SH, Luebeck EG (2008) Age-specific incidence of cancer: Phases, transitions, and biological implications. Proc Natl Acad Sci U S A 105: 16284-16289. 15. Meza R, Jeon J, Renehan AG, Luebeck EG (2010) Colorectal cancer incidence trends in the united states and united kingdom: Evidence of right- to left-sided biological gradients with implications for screening. Cancer Res 70: 5419-5429. 16. Doll R, Peto R (1978) Cigarette smoking and bronchial carcinoma: Dose and time relationships among regular smokers and lifelong non-smokers. J Epidemiol Community Health 32: 303-313. 17. Rachet B, Siemiatycki J, Abrahamowicz M, Leffondre K (2004) A flexible modeling approach to estimating the component effects of smoking behavior on lung cancer. J Clin Epidemiol 57: 1076-1085. 18. Schollnberger H, Manuguerra M, Bijwaard H, Boshuizen H, Altenburg HP, et al. (2006) Analysis of epidemiological cohort data on smoking effects and lung cancer with a multi-stage cancer model. Carcinogenesis 27: 1432-1444. 19. Deng L, Kimmel M, Foy M, Spitz M, Wei Q, et al. (2009) Estimation of the effects of smoking and DNA repair capacity on coefficients of a carcinogenesis model for lung cancer. Int J Cancer 124: 2152-2158. 20. Krewski D, Burnett RT, Goldberg MS, Hoover K, Siemiatycki J, et al. (2000) Reanalysis of the Harvard Six-Cities Study and the American Cancer Society study of Particulate Air Pollution and Mortality, Part II: Sensitivity Analysis. Rep. No. 129-130. (A Special Report of the Institute's Particle Epidemiology Reanalysis Project. Health Effects Institute, Cambridge, Massachusetts). 21. Darby S, Hill D, Auvinen A, Barros-Dios JM, Baysson H, et al. (2005) Radon in homes and risk of lung cancer: Collaborative analysis of individual data from 13 European case-control studies. BMJ 330: 223.

15

22. (2006) The Health Consequences of Involuntary Exposure to Tobacco Smoke: A Report of the Surgeon General. Available: http://www.cdc.gov/tobacco/data_statistics/sgr/2006. Accessed 18 January 2011. 23. Thun MJ, Jemal A (2006) How much of the decrease in cancer death rates in the United States is attributable to reductions in tobacco smoking? Tobacco Control 15: 345-347. 24. Thun MJ, Hannan LM, Adams-Campbell LL, Boffetta P, Buring JE, et al. (2008) Lung cancer occurrence in never-smokers: An analysis of 13 cohorts and 22 cancer registry studies. PLoS Med 5: e185. 25. Flanders WD, Lally CA, Zhu BP, Henley SJ, Thun MJ (2003) Lung cancer mortality in relation to age, duration of smoking, and daily cigarette consumption: Results from Cancer Prevention Study II. Cancer Res 63: 6556-6562. 26. Sakamoto Y, Ishiguro M, Kitagawa G (1986) Akaike information criterion statistics. Tokyo: KTK Scientific Publishers 290 p. 27. Levy DT, Nikolayev L, Mumford E (2005) Recent trends in smoking and the role of public policies: Results from the SimSmoke tobacco control policy simulation model. Addiction 100: 1526-1536. 28. Available: http://www.cancer.gov/newscenter/pressreleases/NLSTresultsRelease. Accessed 18 January 2011.

16

Figure 1. Shared process flow used by all models. Population and smoking inputs were used to develop the smoking history generator, which, in turn, simulates detailed individual-level smoking and other-cause mortality histories. These individual histories are used by each of the modeling groups to generate lung cancer mortality rates in the population.

Figure 2. US Population percentage of current smokers by gender and birth cohort for three different tobacco control scenarios. This is one of the outputs that can be generated from the smoking history generator. The output from the actual tobacco control scenario describes the observed data well (not shown).

Figure 3. Lung Cancer death rates and counts as observed and for modeled tobacco control scenarios.

17

Table 1. Realized and potential reductions in lung cancer mortality from changes in smoking behavior. Realized Proportion of Potential Benefit from Tobacco Control: 1975-2000 Female

Male

Realized

Potential

(NTC-ATC)

(NTC-CTC)

ERASMUS MC

201,788

FHCRC

202,817

MGH-HMS

Overall

Realized

Potential

(NTC-ATC)

(NTC-CTC)

806,320

Proportion Realized 0.25

Realized

Potential

(NTC-ATC)

(NTC-CTC)

1,757,857

Proportion Realized 0.37

860,317

2,564,177

Proportion Realized 0.34

658,529

862,610

0.24

508,777

1,680,867

0.30

711,594

214,830

854,112

0.25

2,543,477

0.28

487,263

1,597,733

0.30

702,092

2,451,845

PIRE

333,976

1,064,443

0.29

0.31

454,517

1,329,972

0.34

788,493

2,394,415

0.33

RICE-MDA

285,079

YALE

221,173

878,359

0.32

603,236

1,645,651

0.37

888,316

2,524,010

0.35

834,293

0.27

603,122

1,712,035

0.35

824,294

2,546,328

0.32

Realized Proportion of Potential Benefit from Tobacco Control: 1991-2000 Female Realized Potential (NTC-ATC)

(NTC-CTC)

ERASMUS MC

143,273

462,528

Proportion Realized 0.31

FHCRC

152,574

521,040

0.29

MGH-HMS

153,549

511,509

PIRE

253,711

687,156

RICE-MDA

185,782

YALE

157,388

Male Realized

Potential

(NTC-ATC)

(NTC-CTC)

384,882

834,310

Proportion Realized 0.46

318,279

842,602

0.38

0.30

310,210

846,300

0.37

342,558

865,306

461,559

0.40

346,266

507,085

0.31

366,815

Overall Realized Potential (NTC-ATC)

(NTC-CTC)

528,155

1,296,837

Proportion Realized 0.41

470,853

1,363,642

0.35

0.37

463,759

1,357,809

0.34

0.40

596,269

1,552,462

0.38

785,168

0.44

532,048

1,246,727

0.43

871,273

0.42

524,203

1,378,358

0.38

Realized Proportion of Potential Benefit from Tobacco Control for Year 2000 Female Realized Potential (NTC-ATC)

(NTC-CTC)

ERASMUS MC

20,277

55,337

Proportion Realized 0.37

FHCRC

22,271

63,373

0.35

MGH-HMS

21,532

60,774

PIRE

40,496

RICE-MDA YALE

Male Realized

Potential

(NTC-ATC)

(NTC-CTC)

48,897

94,979

Proportion Realized 0.51

39,076

92,434

0.42

0.35

38,375

92,187

90,001

0.45

50,943

28,365

55,988

0.51

23,559

62,628

0.38

Overall Realized Potential (NTC-ATC)

(NTC-CTC)

69,173

150,316

Proportion Realized 0.46

61,347

155,807

0.39

0.42

59,907

152,961

0.39

110,800

0.46

91,439

200,802

0.46

42,351

86,863

0.49

70,716

142,851

0.50

45,165

96,794

0.47

68,723

159,422

0.43

18

Funding: This work was supported by NCI U01 grants: 5U01CA097415-04 (to S.H.M.), 2U01CA097432-04 (to T.R.H), 5U01CA097450-04 (to D.T.L), 5U01CA097416-04 (to R.B. and H.J.K), 2U01CA097431-04 (to M.K.). G.S.G acknowledges support from NCI 5R01CA097337-02, P.M.M. from NCI R00CA126147 and ACS RSG 2008A060554, and C.Y.K. from NIH/NCI CA133141. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing Interests: No authors have any competing interests.

Ethics Statement: An ethics statement was not required for this work.

19

Figure Click here to download high resolution image

Figure Click here to download high resolution image

Figure Click here to download high resolution image

Supporting Information Click here to download Supporting Information: Moolgavkarms_PLoS_TextS1.doc

Supporting Information Click here to download Supporting Information: Moolgavkarms_PLoS_TableS1.doc

Supporting Information Click here to download Supporting Information: Moolgavkarms_PLoS_FigureS1.tif