Epidemiology of Lung Cancer* ACCP Evidence-Based Clinical Practice Guidelines. 2007pdf

8/12/2019 Epidemiology of Lung Cancer* ACCP Evidence-Based Clinical Practice Guidelines. 2007pdf

1/27

Epidemiology of Lung Cancer*

ACCP Evidence-Based Clinical Practice Guidelines(2nd Edition)

Anthony J. Alberg, PhD, MPH; Jean G. Ford, MD, MPH; andJonathan M. Samet, MD

Background: The objective of this study was to summarize the published literature concerning theepidemiology of lung cancer.Methods:A narrative review of published evidence was conducted, identifying and summarizingkey reports that describe the occurrence of lung cancer in populations and factors that affect lungcancer risk.Results:In the United States, lung cancer remains the leading cause of cancer death in both menand women, even though an extensive list of modifiable risk factors has long been identified. Thepredominant cause of lung cancer is exposure to tobacco smoke, with active smoking causingmost cases but passive smoking also contributing to the lung cancer burden.Conclusions: The reductions in smoking prevalence in men that occurred in the late 1960sthrough the 1980s will continue to drive lung cancer mortality rates downward in men during thefirst portion of this century, but rates in women have not yet begun to decrease. Fortunately,exposures to major occupational respiratory carcinogens have largely been controlled, but thepopulation is still exposed to environmental causes of lung cancer, including radon, the secondleading cause of lung cancer death. (CHEST 2007; 132:29S55S)

Key words: air pollution; asbestos; cigarette smoking; epidemiology; lung cancer; nutrition; occupation; passivesmoking; radiation; tobacco

Abbreviations:BMIbody mass index; CIconfidence interval; CLconfidence limit; CPSCancer PreventionStudy; ETS environmental tobacco smoke; FTCFederal Trade Commission; IARC International Agency forResearch on Cancer; ILDinterstitial lung disease; IPFidiopathic pulmonary fibrosis; LET linear energytransfer; RRrelative risk; SScsystemic sclerosis

The vast majority of lung cancer deaths are attrib-utable to cigarette smoking. Any action that

prevents cigarette smoking initiation or promotescessation among dependent smokers is a step to

preventing lung cancer. This includes tobacco con-trol activities to affect policy, such as cigarette taxesand smoke-free workplace legislation, as well asindividual-level interventions to prevent the onset or

continuation of smoking.Epidemiologic evidence is the foundation for pri-mary and secondary disease prevention. Epidemio-logic approaches are used to track the occurrence ofdisease, to characterize natural history, and to iden-tify determinants of disease. The benefits of inter-vention programs, whether based in risk factor inter-

*From the Hollings Cancer Center (Dr. Alberg), Medical Uni-versity of South Carolina, Charleston, SC; and Department ofEpidemiology (Drs. Alberg, Ford, and Samet), BloombergSchool of Public Health, Johns Hopkins University, Baltimore,MD.The authors have reported to the ACCP that no significantconflicts of interest exist with any companies/organizations whoseproducts or services may be discussed in this article.Manuscript received May 30, 2007; revision accepted June 5,2007.Reproduction of this article is prohibited without written permissionfrom the American College of Chest Physicians (www.chestjournal.org/misc/reprints.shtml).

Correspondence to: Anthony J. Alberg, PhD, MPH, HollingsCancer Center, Medical University of South Carolina, 86

Jonathan Lucas St, PO Box 250955, Charleston, SC 29425;e-mail: [email protected]: 10.1378/chest.07-1347

SupplementDIAGNOSISANDMANAGEMENTOF LUNG CANCER: ACCP GUIDELINES

www.chestjournal.org CHEST / 132 / 3 / SEPTEMBER, 2007 SUPPLEMENT 29S

wnloaded From: http://journal.publications.chestnet.org/ on 02/05/2014


2/27

ventions or screening, are also assessed usingepidemiologic approaches. For lung cancer, routinemortality statistics confirmed the clinical impressionthat the disease became more frequent across thefirst half of the 20th century. Case-control andcohort studies, the epidemiologic study designs tha-tare used to evaluate exposure/disease associations,causally linked smoking to lung cancer in investiga-

tions reported from the 1950s onward.13 As we havecontinued to follow lung cancer incidence and mor-tality rates, we have readily shown that their rise anddecline parallel past trends of cigarette smoking.4

The epidemiologic evidence and the complementarybiological understanding of respiratory carcinogene-sis have unassailably supported the conclusion thatsmoking causes lung cancer. Epidemiologic findingsare also relevant to patient care, because skilledclinicians weigh alternative diagnoses depending onrisk factor profiles of patients.

At the end of the 20th century, lung cancer had

become one of the leading causes of preventabledeath.5 It was a rare disease at the start of thatcentury, but exposures to new etiologic agents andan increasing life span combined to make lungcancer a scourge of the 20th century. Althoughtobacco had been widely used throughout the worldfor centuries, the present pandemic of lung cancerfollowed the introduction of manufactured cigaretteswith addictive properties, which resulted in a newpattern of sustained exposure of the lung to inhaledcarcinogens.6 German scientists in Nazi Germanyconducted some of the earliest research on the links

between smoking and lung cancer.7 By the early1950s, epidemiologic studies in Britain and theUnited States using the case-control method hadshown that cigarettes were strongly associated withthe risk for lung cancer810; this association wascorroborated by the pioneering cohort studies ofBritish physicians, US veterans, and volunteers re-cruited by the American Cancer Society.11,12 By1964, the evidence was sufficient to support a con-clusion by the US Surgeon General that cigarettesmoking caused lung cancer.11 The Royal College ofPhysicians had reached the same conclusion 2 years

before.12 Passive smoking, the involuntary inhalationof tobacco smoke by nonsmokers, has also beenfound to cause lung cancer.13,14

Although its predominant cause is now widelyknown (tobacco smoking), there are other causes aswell, some acting in concert with smoking to syner-gistically increase risk. The suspicion that radon wasa cause of lung cancer in underground miners, raisedearly in the 20th century, led to what was probablythe first occupational respiratory carcinogen to beidentified15; radon in indoor environments is nowconsidered as the second-leading cause of lung

cancer in the United States.16 The list of humanoccupational causes of lung cancer also includesarsenic, asbestos, chromates, chloromethyl ethers,nickel, polycyclic aromatic hydrocarbons, radonprogeny, and other agents.17 Outdoor air pollution,which includes combustion-generated carcinogens,is also considered to contribute to the lung cancerburden in urban dwellers. Indoor air contains severalrespiratory carcinogens, including radon, asbestos,and cigarette smoke. In some developing countries,exposure to fumes from cooking stoves and fires isassociated with lung cancer risk. Beginning in the1970s, associations of diet with lung cancer risk havebeen vigorously investigated with the anticipationthat dietary micronutrients that modify the high lungcancer risk in smokers might be found. The biolog-ical basis for prevention of cancer through supple-mentation of micronutrients is addressed in anotherarticle in this supplement.

Even though the epidemiology of lung cancerhas been extensively investigated for 50 years,there are still active areas of research, some quiterelevant to prevention. Investigation of lung can-cer and diet continues, using both observationaland experimental approaches, and concern re-mains over the risk of indoor and outdoor pollut-ants, including, for example, radon and dieselemissions. There has also been a need for researchto track the risks of smoking over time, becausethe cigarette has evolved in its design characteris-tics, and yields of tar and nicotine, as assessed by

standard protocol using a machine, have declinedsince the 1950s. The histologic characteristics oflung cancer in a number of developed countries,including the United States, have also changed inthe past few decades such that the frequency ofadenocarcinoma has risen and that of squamouscell carcinoma has declined.4 There is also emerg-ing evidence on genetic determinants of lungcancer risk. A current research approach, termedmolecular epidemiology, melds the population andlaboratory tools that are used to address suscepti-bility to environmental carcinogens. Whereas the

evidence from the traditional epidemiologic ap-proaches conclusively established the carcinoge-nicity of tobacco smoke, molecular epidemiologyshould characterize the sequence of molecular andcellular changes as a nonmalignant cell becomesmalignant and genetic factors that possibly deter-mine susceptibility to tobacco smoke. Biomarkersof exposure, dosage, susceptibility, and geneticdamage may allow epidemiologic investigations touncover specific pathways of human lung carcino-genesis and provide useful intermediate markersfor prevention studies.

30S Diagnosis and Management of Lung Cancer: ACCP Guidelines



3/27

Materials and Methods

A narrative review of published evidence on the epidemiologyof lung cancer was conducted. Key reports that described theoccurrence of lung cancer in populations and factors that affectlung cancer risk were identified. This was accomplished using acombination of approaches that included cataloguing reportsfrom the authors files and augmented with MEDLINE searches.The MEDLINE searches included a term for lung cancer along

with additional terms for various exposures that have beenstudied in relation to lung cancer (eg, cigarette, smoking,asbestos, radiation). In the updating of recent literature,emphasis was placed on systematic reviews when these wereavailable.

Our objective was to provide a summary of the epidemiologicevidence on lung cancer, with an emphasis on issues that arerelevant to prevention. This literature is now extraordinarilylarge; therefore, we did not attempt to conduct a comprehensivereview and systematic synthesis. Such syntheses have beenperiodically carried out by expert review groups, including thecommittees assembled to prepare the US Surgeon Generalsreports on smoking and health and other federal documents andexpert committees of other governments and organizations,including the UK Royal College of Physicians and Scientific

Committee on Tobacco and the World Health OrganizationsInternational Agency for Research on Cancer (IARC). Severalrelevant reports have been published, including the 2004 IARCmonographs on active and involuntary smoking18 and the 2004report of the Surgeon General.19

The topics covered were agreed on by consensus of the writingcommittee with initial input from the ACCP Guidelines Panel. Asprior versions of this article underwent several rounds of externalreview, additional topics were added as recommended by theexternal reviewers, the ACCP Lung Cancer Guidelines Panel, theThoracic Oncology Network, the Health and Science PolicyCommittee, and the Board of Regents of the American College ofChest Physicians. On the basis of the agreement of all parties, wedid not attempt to grade the evidence or generate formal

guidelines.

Results

Patterns of Occurrence

Survival:The 5-year relative survival rate for lungcancer for the period of 1995 to 2001 was 15.7%,reflecting a steady but slow improvement from12.5% from 1974 to 1976.20 The 5-year relativesurvival rate varies markedly depending on the stageat diagnosis, from 49 to 16 to 2% for local, regional,and distant stage disease, respectively.20 Stage atdiagnosis accounts for the most marked variation inprognosis, but patient characteristics associated withpoorer survival also include being older, male, andAfrican American.20

Temporal Trends:Because of the high case-fatalityrate of lung cancer, incidence and mortality rates arenearly equivalent; consequently, routinely collectedvital statistics provide a long record of the occur-rence of lung cancer. We are amid an epidemic oflung cancer that dates to the first half of the lastcentury.

Sex:Lung cancer was rare until the disease begana sharp rise around 1930 that culminated by mid-century with lung cancer becoming the leading causeof cancer death among men.21 The epidemic amongwomen followed that among men, with a sharp risein rates from the 1960s to the present, propellinglung cancer to become the most frequent cause offemale cancer mortality.21 As the leading cause of

cancer death among women, lung cancer is a majorwomens health issue. As a result of historical ciga-rette smoking patterns, the epidemic of lung cancerstarted later in women than men, but in contrast tothe situation in men, lung cancer incidence rates inwomen have not yet begun to decrease consistent-ly.20 Far more men than women still die from lungcancer each year, but the gender gap in lung cancermortality is steadily narrowing and will eventuallyclose.22,23 This trend is due to historical smokingpatterns, with smoking prevalence having peakedapproximately 2 decades earlier among men than

women.22,23Examination of time trends of age-specific lung

cancer mortality rates in the United States furtherhighlights the differing epidemic patterns in mencompared with women. The sex- and race-specificmortality rates are now almost all decreasing.22 Therates of lung cancer in the younger age groups havebeen declining during the past several decades inmen and during the past decade in women.22 As theyounger birth cohorts age, their reduced risk for lungcancer foreshadows substantial reductions in theoverall occurrence of lung cancer, but the reductions

will be greater for men than for women. Thesepatterns all are consistent with population patterns ofsmoking prevalence over time.22

Tobacco smoking accounts for such a large pro-portion of lung cancer that there have been few dataon the occurrence of lung cancer among nonsmok-ers. Evidence from the American Cancer SocietyCancer Prevention Study (CPS) I and II cohortsindicates that there has not been a strong temporaltrend in lung cancer death rates among male non-smokers, but there has been an upward trendamong female nonsmokers, mostly confined to el-

derly women.23 The data from these cohorts alsoindicate that among nonsmokers, lung cancer deathrates are greater in men than in women and greaterin African-American than white women.

Race and Ethnicity:The patterns of occurrence oflung cancer by race and ethnicity make lung cancera relevant disease for those concerned with thehealth of minorities. Of particular note is thatwhereas lung cancer incidence rates are similaramong African-American and white women, lungcancer occurrence is approximately 45% higher




4/27

among African-American men than among whitemen.20 This racial disparity may be partially due togreater susceptibility of African-American smokersto smoking-induced lung carcinogenesis.24 Thehigher mortality rates of lung cancer in African-American compared with white individuals reflectnot only their higher incidence rate but also thepoorer survival from lung cancer among African-

American compared with white individuals. The5-year relative survival rate was 13% lower inAfrican-American compared with white individualsduring the period 1995 to 2001.20 This racial gappersisted within each stage at diagnosis category andfor men and women.20

Lung cancer mortality rates among Hispanic, Na-tive American, and Asians/Pacific Islander individu-als are significantly lower than rates among African-American and non-Hispanic white individuals.25

Nevertheless, lung cancer poses a considerable pub-lic health burden among these groups.

Socioeconomic Status:Lung cancer is more likely tooccur in the poor and less educated, a pattern that isobserved in many countries worldwide. For example, inCanada, the risk for lung cancer in both sexes wasinversely associated with income, education, and social

class, even after adjustment for cigarette smoking.26 InChina, those who were classified as low income had asixfold increased risk of lung cancer compared withthose in the high-income category.27 In the Nether-lands, the risk for lung cancer was inversely associatedwith attained education, an association that was notattributable to occupational exposures.28 Lower socio-economic status has also been observed to be associ-

ated with later stage at diagnosis.29Socioeconomic status is associated with a constel-

lation of interacting determinants of lung cancer risk,such as smoking, diet, and exposures to inhaledcarcinogens in the workplace and general environ-ment. Lower socioeconomic status is associated withan unfavorable profile for all of these factors. Ad-vancing our understanding of the complex linkagesbetween components of socioeconomic status andlung cancer risk is essential to effectively addressingthis social class disparity and reducing lung cancerrates in the poorer segments of society.

Geographic Patterns: Lung cancer is the mostcommonly diagnosed cancer worldwide,30 but itsgeographic distribution shows marked regional vari-ation: age-standardized incidence rates range 60-fold among men and 30-fold among women (Fig 1,

Figure 1. Age-adjusted lung cancer incidence rates in women worldwide in 2002. Source: IARC,

GLOBOCAN 2002 (www-dep.iarc.fr).




5/27

2).31 Because of differences in cancer registrationbetween countries, caution is needed in interpretingthese data. However, this marked variation in ratescannot be explained on the basis of diagnostic prac-tices and data quality alone. Lung cancer tends to bemost common in developed countries, particularly inNorth America and Europe, and less common indeveloping countries, particularly in Africa andSouth America.31 The low rates of lung cancer inAfrica are comparable to US rates in 1930, whenrates of lung cancer were 5 per 100,000 for bothsexes.32 In contrast, African-American individuals inthe United States, an epicenter, now experience lungcancer incidence rates that are among the highest inthe world. As the lung cancer epidemic begins tosubside in the developed countries, it is on the rise inthe developing world.30

Within countries, lung cancer incidence amongmen invariably exceeds that in women, by well morethan 100% in most nations. The international rank-ings of lung cancer incidence of men and womenfrom the same countries tend to differ only slightly,so the highest rates of lung cancer occur in the sameregions of the world for both sexes.

Substantial geographic variation in lung cancer

mortality rates has also been observed within coun-

tries. For example, during the period 1997 to 2001,the age-adjusted lung cancer mortality rates variedmore than threefold between the state with thehighest rate (Kentucky, 78 per 100,000) and the statewith lowest rate (Utah, 25 per 100,000).20 Trends inits regional distribution can provide clues aboutdeterminants of lung cancer. In the past, ratestended to be highest in urban areas, which led toconjecture that air pollution might be a cause of thelung cancer epidemic.33 Later on, several hypothe-ses34,35 were prompted by patterns observed in asystematic review of US lung cancer mortality rates for

the period 1950 to 1969,36 particularly the rates amongmen. For example, high rates in coastal areas werepostulated to reflect employment in shipyards withattendant asbestos exposure. This hypothesis was thentested in a series of population-based case-controlstudies that showed that employment in the shipbuild-ing industry was indeed associated with an excess riskfor lung cancer.37 Another shift then took place in thedistribution of lung cancer within the United States,with lung cancer mortality rates among white menbecoming highest in the South and lower in theNortheast.38 This temporal fluidity in the geographic

Figure 2. Age-adjusted lung cancer incidence rates in men worldwide in 2002. Source: IARC,GLOBOCAN 2002 (www-dep.iarc.fr).




6/27

variation underscores the need for regularly monitoringlung cancer mortality patterns.

Etiology of Lung Cancer

Although the causes of lung cancer are almostexclusively environmental, there is likely substantialindividual variation in susceptibility to respiratorycarcinogens. The risk for the disease can be concep-tualized as reflecting the joint consequences of theinterrelationship between the following: (1) exposureto etiologic (or protective) agents, and (2) individualsusceptibility to these agents. The environment inits broadest sense may influence the risk for diseasethrough direct exposures or indirectly by affectingthe likelihood of exposure to exogenous agents.Given the multifactorial etiology of lung cancer,synergistic interactions among risk factors may havesubstantial consequences for lung cancer risk. These

interactions have typically been considered on anagent-by-agent basis, such as the synergistic effect ofcigarette smoking on the lung cancer risk fromasbestos exposure.39 Our emerging understanding ofcancer genetics indicates the additional relevance ofgene/environment interactions.

Given the many risk factors that have been iden-tified for lung cancer, a practical question is therelative contribution of these factors to the overallburden of lung cancer. The population attributablerisk approach takes into account the magnitude ofthe relative risk (RR) associated with an exposure

along with the likelihood of exposure in the generalpopulation. These attributable risk estimates includejoint contributions of risk factors that sometimeshave synergistic relationships. For example, the at-tributable risk estimate for cigarette smoking in-cludes the lung cancer risk attributed to the inde-pendent effects of cigarette smoking and furtherincludes the risk for lung cancer from smoking as aresult of its synergistic interactions with factors suchas asbestos and radon. For this reason, the totalpercentage can be 100%. Lung cancer has awell-characterized set of important risk factors and

established synergistic interactions between risk fac-tors, and these reasons contribute to the attributablerisks summing to considerably more than 100%. Asreviewed next, population attributable risk estimatesfor lung cancer indicate that in the United States,active smoking is responsible for 90% of lung cancer;occupational exposures to carcinogens for approxi-mately 9 to 15%; radon for 10% of lung cancer,16 andoutdoor air pollution for perhaps 1 to 2%.40 Thecontribution of nutritional factors cannot yet beprecisely determined; consequently, estimates of therole of dietary factors range widely.41

Environmental and Occupational Agents

Smoking:A single etiologic agent (cigarette smok-ing) is by far the leading cause of lung cancer,accounting for approximately 90% of lung cancercases in the United States and other countries wherecigarette smoking is common.42 Compared with never-smokers, smokers who have smoked without quitting

successfully have an approximate 20-fold increase inlung cancer risk. Few exposures to environmentalagents convey such risks for any disease. In general,trends of lung cancer occurrence closely reflectpatterns of smoking, but rates of occurrence lagsmoking rates by approximately 20 years. Analysesusing statistical modeling techniques show a tightassociation between national mortality rates andsmoking.43 The unequivocal role of cigarette smok-ing in causing lung cancer is one of the mostthoroughly documented causal relationships in bio-medical research.6,44

The burden of lung cancer that is attributable tosmoking has been extensively documented. Using anattributable risk approach, the annual number ofdeaths caused in the United States by smoking-related lung cancer during the period from 1995 to1999 was 122,800.19 Peto et al42 used a differentattributable risk method to quantify the burden ofsmoking-related deaths from lung cancer in themajor developed countries. For 1990, the US totalwas 127,000, the highest in the world, with country-specific estimates ranging down to 150 for Tajikistan.The total for the developed countries was 457,371.42

A staggering future burden of lung cancer has beenforecast for China, where the numbers are predictedto reach several millions by mid-century.45,46

Cigar smoking is also an established cause of lungcancer.47 The lung cancer risks associated with cigarsmoking are substantial but less than the risks ob-served for cigarette smoking as a result of differencesin smoking frequency and depth of inhalation. Thesame pattern holds true for pipe smoking.48 Withrespect to smoking of nontobacco products, thepotential role of smoking marijuana on lung cancerrisk has been of interest. Despite the plausibility of

marijuana as a risk factor for lung cancer, theevidence to date has not documented an associationafter adjusting for tobacco smoking.49

The risk for lung cancer among cigarette smokersincreases with the duration of smoking and thenumber of cigarettes smoked per day.50 This obser-vation has been made repeatedly in cohort andcase-control studies. Risk models have been derivedto estimate quantitatively how lung cancer risk varieswith number of cigarettes smoked, duration of smok-ing, and age. Such models are useful for estimatingthe future burden of lung cancer under various




7/27

scenarios of tobacco control. In one widely citedanalysis, Doll and Peto50 proposed a quantitativemodel for lung cancer risk on the basis of data fromthe cohort study of British physicians. This modelpredicted a stronger effect of duration of smokingthan of amount smoked per day. Thus, a tripling ofthe number of cigarettes smoked per day was esti-mated to triple the risk, whereas a tripling of dura-

tion of smoking was estimated to increase the risk100-fold.51 These quantitative dimensions of thedosage-response relationship between smoking andlung cancer have implications concerning the nowwidespread smoking among youths. Those who startat younger ages have a greater likelihood of becom-ing a heavier smoker and remaining a smoker.52 Theexponential effect of duration of smoking on lungcancer risk markedly increases the lifetime risk forthose who become regular smokers in childhood andplaces them at increased risk at younger ages. Pre-vention approaches that delay the age of onset of

smoking in a population could have substantial im-pact on the incidence of lung cancer by shorteningthe duration of smoking. In considering the likeli-hood of lung cancer in a particular patient, cliniciansshould give more weight to the duration of smokingand less to actual age.

Cigarette smokers can benefit at any age byquitting smoking. The likelihood of lung cancerdeveloping decreases among those who quit smokingas compared with those who continue to smoke.52 Asthe period of abstinence from smoking cigarettesincreases, the risk for lung cancer decreases.53 How-

ever, even for periods of abstinence of40 years,the risk for lung cancer among former smokersremains elevated compared with never-smokers.53,54

The benefits derived from smoking cessation alsodepend on the duration of smoking; for a givenperiod of abstinence, the decrease in risk increases asthe duration of smoking decreases.53 In general,studies55 have shown comparable reductions in riskafter cessation regardless of sex, type of tobaccosmoked, and histologic type of lung cancer.

The benefits of physician (and other clinician)intervention for smoking cessation are well estab-

lished.56 The results of research in this area havebeen translated into an evidence-based clinical prac-tice guideline for treating tobacco dependence onthe basis of the 5 As: ask whether a patientsmokes, assess willingness to quit, advise to quit,assist with quitting, andarrange follow-up.56

The composition of cigarettes has evolved consid-erably since the 1950s. The marketplace has shiftedfrom mainly unfiltered cigarettes to predominantlyfiltered cigarettes. The filters in use in the UnitedStates are predominantly cellulose acetate, whereascharcoal filters are used extensively in Japan and

some other countries.57 In the mid-1960s, ventilationholes were added to the filter, which dilute thesmoke with air drawn through them. However,smokers can readily block the holes with their fin-gers, which are left unblocked by the machines thatare used to test cigarettes. There have also beensubstantial changes in the design of the cigarette andin the tobacco used. Reconstituted tobacco has been

used increasingly since the 1960s, there have beenchanges to the cigarette paper and additives used,and most cigarettes are more ammoniated in theUnited States.57

A concomitant shift toward lowered levels of tarand nicotine, as measured by a smoking machine, hasoccurred.58 Cigarette tar refers to the condensableresidue of cigarette smoke (ie, the total particulatematter of cigarette smoke deposited on the filter ofthe machine, less the moisture and nicotine). Tar isa complex mixture that includes many chemicals thatare cancer initiators and/or promoters.58 Tar and

nicotine yields are measured with a smoking ma-chine according to a standardized protocol estab-lished by the Federal Trade Commission (FTC) thatspecifies such details and puff volume, the frequencyof puffing, and the length to which the cigarette is tobe smoked.59

Studies59 using biomarkers of exposure to anddosage of tobacco smoke components show littlerelationship of levels of these markers with tar ornicotine yield as measured by the FTC protocol.These studies have been conducted in both thepopulation context and during smoking in the labo-

ratory setting. For example, Coultas et al60 collectedsaliva for analysis for cotinine level and end-tidalbreath samples for measurement of carbon monox-ide level in a population sample of New MexicoHispanic individuals who were included in a respi-ratory health survey. After taking account of num-bers of cigarettes smoked, biomarker levels were notassociated with the yields of tar and nicotine of thecurrent brand smoked. Djordjevic et al61 evaluatedsmoking pattern and biomarkers in the laboratorysetting, contrasting smokers of medium-yield andlow-yield cigarettes. The smokers had greater puff

volumes and frequencies than are specified in theFTC protocol and had substantially greater intakes oftar and nicotine than implied by the brand listings.The lack of association of tar and nicotine yields withbiomarker levels partially reflects compensatorychanges in smoking patterns for those who switchfrom higher to lower yield products. The compensa-tion includes blocking of the ventilation holes, morefrequent and deeper puffs, and an increase in thenumber of cigarettes smoked.62

The gradual reduction in machine-measured taryield would be expected to have reduced smokers




8/27

exposures to carcinogens if the FTC test protocolwere predictive of carcinogen dosages delivered tothe lung.58 However, questions remain as to whetherthe FTC test method is informative with regard tolung cancer risk or risks for smoking-caused diseasesmore generally.62,63 Epidemiologic studies havebeen conducted to assess whether the seeminglysubstantial changes in tar and nicotine yield, as

measured by the FTC protocol, have resulted inparallel changes in the risk of smoking. Epidemio-logic studies have been the key source of informationbecause they can provide direct evidence on the risksof smoking cigarettes, as they are actually smokedduring use, including any compensatory behavior.

For lung cancer and for other diseases, three linesof epidemiologic data have been available onchanges in products. The first comes from case-control studies that compared the smoking historyprofiles of people with lung cancer with those ofcontrol subjects. The second comes from cohort

studies that tracked the risk for lung cancer overtime, as the products smoked changed. The thirdcomes from assessment of the temporal changes inage-specific patterns of lung cancer mortality rates incomparison with changes in cigarette characteristics.

The initial evidence came primarily from case-control studies that compared risks in people whohad used filter-tipped cigarettes with people whohad smoked nonfiltered cigarettes exclusively.64,65

This evidence suggests that filtered cigarettes andcigarettes with lower tar yields slightly reduce therisk for lung cancer associated with cigarette smok-

ing compared with nonfiltered cigarettes or withhigher tar yields.6668 This comparison could bemade among smokers in the 1960s because there wasstill a substantial proportion who had not usedfiltered cigarettes at all. For example, in one of thefirst studies, Bross and Gibson64 compared lungcancer risk of smokers of filtered and nonfilteredcigarettes among patients who were seen at RoswellPark Memorial Cancer Institute in Buffalo; individ-uals were classified as filter cigarette smokers whenthey had used these products for at least 10 years.

The relevant cohort studies are the American

Cancer Society CPS I and CPS II studies and theBritish Physicians Cohort. In a 1976 publication,Hammond et al69 compared mortality risks fromlung cancer and other diseases by tar yield ofproducts smoked by CPS I participants. The fol-low-up interval spanned from 1960 to 1972. Smokerswere placed into three categories of productssmoked: low yield ( 17.6 mg per cigarette), highyield (25.8 to 35.7 mg per cigarette), and mediumyield (intermediate). The standardized mortality ratefor lung cancer in low- and medium-yield smokerswas approximately 80% of the rate in high-yield

smokers. A further analysis of tar yield using thesame data set confirmed that risk for lung cancerdeath increased with tar yield.70

Further insights have been gained by comparingthe risks in the two CPS studies of the AmericanCancer Society; this comparison addresses whetherrisks have changed, comparing smokers with diseasedeveloping from 1960 to 1972 with a similar group of

smokers with disease developing during the initialfollow-up of CPS II, from 1980 to 1986.71,72 If therisk for lung cancer associated with smoking isdecreasing over time, then the expectation would bethat risks for smokers would be less in CPS II than inCPS I. In fact, the opposite was observed, withincreasing lung cancer mortality in male and femalesmokers in CPS II compared with CPS I.73

In an analysis with a similar pattern of findings,Doll et al74 compared the risks for death from lungcancer and other causes during the first and second20 years of the 40-year follow-up of the British

physician cohort. Lung cancer mortality increasedamong smokers in the second 20 years (from 1971 to1991), even though products smoked during thisperiod would have had a substantially lower tar andnicotine yield than those smoked during the first 20years (from 1951 to 1971). For the first 20 years, theannual lung cancer mortality rate among currentsmokers was 264 per 100,000, and for the second 20years, it was 314 per 100,000. In 2004, Doll et al75

reported the findings at 50 years of follow-up; com-pared with lifelong nonsmokers, the risk for lungcancer was increased fourfold among former smok-

ers and 14-fold among current smokers. Amongcurrent smokers, the RRs increased from 7.7 to 13.7to 24.5 among smokers of 1 to 14, 15 to 24, and 25cigarettes per day, respectively.

The third line of observational evidence comesfrom descriptive analyses of age-specific trends oflung cancer mortality.18,62,76 Successive birth cohortshave had differing patterns of exposure to cigarettesof different characteristics and yields. For example,the cohort of individuals who were born between1930 and 1940 and started to smoke in the 1950s wasone of the first to have the opportunity to smoke

primarily filter-tipped cigarettes. Subsequent birthcohorts would have had access to the increasinglylower yield products, whereas earlier cohorts hadaccess initially only to nonfiltered cigarettes. Patternsof temporal change in age-specific rates of lungcancer mortality in younger men have been exam-ined to assess whether there has been a declinegreater than expected from changing prevalence,duration, and amount of smoking, thereby indicatinga possible effect of cigarette yield.

Data on lung cancer mortality in younger men inthe United Kingdom have been interpreted as indi-




9/27

cating a possible reduction in lung cancer risk asso-ciated with changes in cigarettes.62,76 A sharp declinein lung cancer mortality has occurred across the pastfew decades in UK men 50 years of age. Thedecline seems greater than anticipated from trendsin prevalence and other aspects of smoking: agestarting and number of cigarettes smoked. A simi-larly steep decline has not taken place in the United

States. Given the ecologic nature of the data underconsideration, uncertainty remains with regard totheir interpretation, and alternative explanationshave been proposed, including less intense smokingat younger ages in more recent birth cohorts.62

This discussion highlights the complexity of isolat-ing the precise effect on lung cancer risk of thecontinually changing cigarette. The data available toevaluate these effects have limitations, particularly incapturing the experience of successive birth cohortsin either case-control or cohort studies that wereappropriately designed. The UK mortality data sug-

gest a greater effect of changes in cigarettes than isfound in the case-control and cohort studies. Asrecommended by the Institute of Medicine,77 sur-veillance is needed to track the health consequencesof the changing cigarette.

Several expert panels have reviewed the findings.The Institute of Medicine77 conducted a compre-hensive review on various harm reduction strategiesfor reducing the disease burden caused by smoking,including lower yield cigarettes. There are also newproducts in various phases of development that areintended to deliver nicotine without direct combus-

tion of tobacco. The Institute of Medicine reportconcluded that smoking lower-yield products hadnot been shown to benefit the health of smokers.This topic was addressed in the 2004 report of theUS Surgeon General,19 with the conclusion thatalthough characteristics of cigarettes have changedduring the last 50 years and yields of tar and nicotinehave declined substantially, as assessed by the Fed-eral Trade Commissions test protocol, the risk oflung cancer in smokers has not declined.

Results of some case-control and screening studieshave suggested a potentially higher risk for smoking-

associated lung cancer in women compared withmen,7880 but methodologic issues cloud the inter-pretation of these studies, particularly a lack of focuson the most informative comparisons.81 Further-more, the evidence from prospective cohort studiesfails to support the notion of a sex differential insusceptibility to lung cancer from smoking.82 Theequal rates of lung cancer mortality in younger USmen and women corresponding to a time of equalsmoking prevalence also provides evidence against animportant sex difference in susceptibility to smoking-induced lung cancer.22 The evidence against this

hypothesis outweighs the evidence in favor of thehypothesis on the basis that the results of studies thathave compared the RR estimates for men andwomen for a specific degree of smoking historydemonstrate very similar associations.82

The development of menthol cigarettes was targetedspecifically at African-Americans and women.83,84 Af-rican-Americans are more likely than white individ-

uals (69 vs 29%) to smoke menthol cigarettes,85 andthe menthol smoke delivery levels of common ciga-rette brands have increased significantly since the1980s.86,87 This has led to the hypothesis that men-thol cigarettes explain the greater susceptibility tolung cancer from cigarette smoking in black vs whiteindividuals24 and thus the disparity in lung cancerrisk between US black and white individuals, espe-cially among men.

Menthol cigarettes may cause a greater increase inlung cancer risk than nonmenthol cigarettes, eitherby increasing systemic exposure to toxicants from

tobacco smoke or by affecting the metabolism ofnicotine and/or tobacco smoke carcinogens. Initially,this hypothesis gained currency because of the po-tential for increased nicotine uptake through theeffects of menthol in the respiratory tract. Theseinclude an increase in the smoothness of tobaccosmoke, which promotes deeper inhalation; stimula-tion of cold receptors, which results in airway coolingeffects that mask the irritation caused by cigarettesmoke, promoting deeper inhalation and alteredinhalation frequency; further masking of irritationthrough anesthetic effects86,88; and increased perme-

ability and diffusibility of smoke constituents.87There is limited information on the molecular

mechanisms by which mentholation might increasethe health risk of smoking. Seventy to 80% ofnicotine is metabolized to cotinine, and cytochromeP450 2A6 is responsible for 90% of this conversion.89

The P450 2A6 gene has multiple functional polymor-phisms that vary by race. The observation that mentholcompetitively inhibits cotinine metabolism by the mon-key analog of a human UDP-glucuronyltransferase90

suggested that inhibition of either CYP2A6 or UDP-glucuronyltransferase by menthol might alter nico-

tine and cotinine metabolism. African-American andwhite menthol smokers have similar baseline cotin-ine levels.91 Human studies89,9193 have suggestedthat smoking mentholated cigarettes inhibits nico-tine metabolism, so smokers experience higher dos-ages of nicotine for a given level of smoking. Mentholinhibits the microsomal oxidation of nicotine tocotinine,92 suggesting that smoking mentholated cig-arettes may lead to inhibition of nicotine metabo-lism. In a randomized, crossover study of sevenAfrican-American and seven white individuals, Be-nowitz et al93 found that the systemic intake of




10/27

nicotine was not affected by mentholation, but smok-ing mentholated cigarettes inhibited the metabolismof nicotine. By slowing the metabolism of nicotineand thereby reducing the need for nicotine fromsmoking, menthol may reduce the number of ciga-rettes smoked per day. A menthol effect mightexplain why African-American individuals smokefewer cigarettes per day than white individuals. It

may also explain, in part, the variation by race andgender in the correlation between cotinine level andcigarettes smoked per day among smokers of men-thol cigarettes,94 possibly reflecting the effect ofmenthol on nicotine inactivation by P450 2A6.89

However, the epidemiologic data suggest that,overall, smokers of mentholated cigarettes do nothave an increased risk for lung cancer compared withsmokers of nonmentholated cigarettes. This evi-dence is based primarily on hospital-based case-control studies,9598 but also includes a population-based case-control study99 and a cohort study within

a health maintenance organization.100 Furthermore,menthol cigarettes have not been associated with anyspecific histologic subtypes of lung cancer.101

Evidence that menthol cigarettes might carrygreater risks were observed in one case-controlstudy97 in which black, male, heavy smokers ofmentholated cigarettes (37.5 pack-years, or 21cigarettes per day) had a higher risk than white menwith similar smoking histories. In the cohort study,100

the RR for lung cancer among men but not womenwas slightly elevated in menthol smokers comparedwith nonmenthol smokers, with a graded increase in

lung cancer risk with increasing duration of mentholcigarette use.

The evidence does not indicate that mentholcigarettes are an important contributor to the highrates of lung cancer in African-American individuals.A more definitive answer to this question willemerge if future studies address several method-ologic challenges, including misclassification of men-thol cigarette exposure as a result of brand ambigu-ity; potential for selection bias in hospital-basedcase-control studies, as a result of lower prevalenceof menthol cigarette use among African-American

patients at university hospitals used for such studiesthan in the general population; and lack of informa-tion about compensatory mechanisms.102

Passive smokers inhale a complex mixture ofsmoke now widely referred to as secondhand smokeor as environmental tobacco smoke (ETS). Passivesmoking was first considered as a possible risk factorfor lung cancer in 1981, when two studies thatdescribed increased lung cancer risk among never-smoking women who were married to smokers werepublished. Hirayama103 reported the findings from acohort study in Japan that showed that among

nonsmoking women, those with a husband whosmoked cigarettes were at higher risk for lung cancerthan those whose husband was a nonsmoker. Acase-control study in Athens reported by Tricho-polous et al104 shortly thereafter replicated thisfinding. Additional evidence rapidly accrued, suchthat by 1986 two important summary reports werepublished. The National Research Council reviewed

the epidemiologic evidence and concluded that non-smoking spouses who were married to cigarettesmokers were approximately 30% more likely to havelung cancer develop than nonsmoking spouses mar-ried to nonsmokers and that this relationship wasbiologically plausible.105 Almost one fourth of lungcancer cases among never-smokers were estimatedto be attributed to exposure to passive smoking.105

The 1986 Surgeon General report also judged pas-sive smoking to be a cause of lung cancer,13 aninference corroborated by the 1992 review of theevidence and risk assessment by the US Environ-

mental Protection Agency, which classified ETS as aknown human (class A) carcinogen.14 Estimates in-dicate that passive smoking accounts for approxi-mately 3,000 lung cancer deaths per year in theUnited States.14 Since these conclusions werereached, several major studies106,107 have been con-ducted to characterize further the association ofpassive smoking with lung cancer, while taking intoaccount some of the limitations of earlier studies,particularly small sample sizes, exposure misclassifi-cation, and omission of some potential confoundingfactors.

Passive smoking is more weakly associated withlung cancer than is active smoking, as expected giventhe generally lower dosages of carcinogens that arepassively received by the lung of the nonsmokercompared with the dosages received by the activesmoker. Because of broad societal implications, theconclusion that this association is causal has gener-ated controversy, some driven by the effort of thetobacco industry to maintain continued questioningof the evidence.108,109 Questions have been raisedabout the method of the epidemiologic studies,including confounding and misclassification of expo-

sure to environmental tobacco smoking. Reviewgroups13,14,106,110 have nonetheless concluded thatthe association between ETS and lung cancer cannotbe attributed to methodologic limitations of epide-miologic data.

Studies have been directed at the specific venueswhere nonsmokers are exposed to tobacco smoke,including the home, workplaces, and public places.Much of the literature has focused on the increasedrisk associated with being married to a smoker, anexposure variable that can be readily ascertained.Metaanalyses have been conducted periodically to




11/27

summarize the evidence from the epidemiologicstudies. A 2002 metaanalysis by Boffetta111 found a25% increased risk associated with marriage to asmoker; this excess risk seemed to be due to expo-sure to passive smoking because it could not beexplained by confounding or misclassification. Thisfinding was consistent with the 29% estimated in-creased risk among women whose husband smoked

in the metaanalysis of Taylor et al,112 who observedthat the association was consistent across study de-signs and in Western and non-Western nations.Workplace exposure to secondhand smoke was asso-ciated with a 17% increase in lung cancer risk in themetaanalysis of Boffetta.111

The studies of passive smoking provide furtherevidence documenting the dosage/response relation-ship between cigarette smoke and lung cancer. Thedosages extend to far lower levels than those ofactive smoking and increased risk is observed, sug-gesting that there is no threshold for tobacco

carcinogenesis.13Lung cancer occurs in multiple histologic types as

classified by conventional light microscopy. The fourmajor types include squamous cell carcinoma, ade-nocarcinoma, large cell carcinoma, and small cellundifferentiated carcinoma; together, these fourtypes of lung cancer account for 90% of lungcancer cases in the United States.113 Notable shiftshave taken place in the incidence rates of lungcancer by histologic type.114 After steadily increasingoccurrence during the period from 1973 to 1987,adenocarcinoma supplanted squamous cell carci-

noma as the most frequent form of lung cancer.114Adenocarcinoma increased markedly in all race andsex subgroups.114

Despite extensive research, the mechanisms thatlead to these different types of lung cancer remainuncertain. Hypotheses have focused on the cells oforigin of lung cancers and on pathways of differen-tiation of malignant cells.113 An area of active interestis characterizing the likelihood that dysplastic lesionsthat are detected by fluorescence bronchoscopy willprogress to invasive cancer115 and relating the distri-bution of these lesions vis a vis the distribution of

invasive lung cancer tumors on the basis of epidemi-ologic findings. CT scans are generally being used toidentify peripheral lesions (usually adenocarcinoma),whereas fluorescence bronchoscopy is being used forthe detection of central airway lesions, predomi-nantly preinvasive squamous cell carcinoma. Smok-ing has been shown to cause each of the majorhistologic types, although the dose/response relation-ship with number of cigarettes smoked varies acrossthe types, being steepest for small cell undifferenti-ated carcinoma.115,116 There are a few suggestivelinks of histologic type with occupational agents:

small cell lung cancer has been reported to be inexcess in workers who are exposed to chloromethylethers and in underground miners who are exposedto radon progeny.113

In the initial decades of the smoking-caused epi-demic of lung cancer, squamous cell carcinoma wasthe most frequent type of lung cancer observed inthe population, and small cell carcinoma was the next

most frequent. In the late 1970s, the first evidence ofa shift toward a predominance of adenocarcinomawas noted,113,117,118 and now adenocarcinoma of thelung is the most common histologic type.112 Thedecline in lung cancer rates has been more rapid forsquamous cell and small cell carcinomas than foradenocarcinoma, which is just beginning to show alower incidence rate.114 In women, the Surveillance,Epidemiology, and End Results4 data from 1973 to1996 indicated that the incidence rates of squamouscell, small cell, and large cell carcinomas at leastreached a plateau, whereas the rate for adenocarci-

noma were still rising.Although changing patterns of diagnosis and clas-

sification of lung cancers could have led to thesechanges over time, most observers have set aside anartifactual change.113,117,118 Beginning in the 1970s,new techniques for the diagnosis of lung cancerbecame available, including the fiberoptic broncho-scope and thin-needle aspiration119; improved stainsfor mucin, the hallmark of adenocarcinoma, werealso introduced. Using data from the ConnecticutTumor Registry, Thun et al119 showed that the rise inadenocarcinoma antedated these diagnostic innova-

tions.Hypotheses concerning the shift in histopathology

have focused on the potential role of changes in thecharacteristics of cigarettes and consequent changesin the dosages of carcinogens inhaled.120 Puff volumehas likely increased in the past few decades with thepossibility that patterns of deposition in the lunghave changed, tending toward enhanced depositionof tobacco smoke in the peripheral airways andalveoli.120 Nitrate levels in tobacco smoke have alsoincreased, which enhances the combustion of to-bacco smoke. Although more complete combustion

decreases the concentrations of polycyclic aromatichydrocarbons, the increased production of nitrogenoxides contributes to increased formation of tobacco-specific nitrosamines. An increase in dosage of thepotent tobacco-specific nitrosamine 4-(methylnitro-samino)-1-(3-pyridyl)-1-butanone has been postulatedas one factor leading to the increase in adenocarcino-ma.120,121 Nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone induces lung carcinomas, predom-inantly adenomas and adenocarcinomas, in mice, re-gardless of route of administration.121,122

Few studies can provide data to test these hypoth-




12/27

eses because of the need for longitudinal observationof lung cancer risk in relation to the characteristics ofthe cigarettes smoked over time. Thun et al119

compared risks for lung cancers of the varioushistologic types among participants in the AmericanCancer Society CPS I and CPS II. They foundmarkedly rising risks associated with smoking foradenocarcinoma of the lung in both men and womenduring the approximate 20 years separating the twostudies. Thun et al119 concluded, The increase inlung adenocarcinoma since the 1950s is more con-sistent with changes in smoking behavior and ciga-rette design than with diagnostic advances. In astudy123 that compared tumor location in lung cancerpatients, lower-tar cigarettes were associated with ahigher likelihood of peripheral than central tumors.

Diet: Research on diet and lung cancer has nowbeen conducted for nearly 3 decades. The possible

role of diet in modifying the risk for lung cancer hasbeen the focus of intensive investigation, driveninitially by the rationale that specific micronutrientsmight have anticarcinogenic activity. The most thor-oughly investigated dietary factors are also those thatseem to have the greatest implications for preven-tion: fruits, vegetables, and specific antioxidant mi-cronutrients that are commonly found in fruits andvegetables. Much of the research on diet and lungcancer has been motivated by the hypothesis thatdiets that are high in antioxidant nutrients mayreduce oxidative DNA damage and thereby protectagainst cancer.124

The results of case-control and prospective cohortstudies have tended to show that individuals withhigh dietary intake of fruits or vegetables have alower risk for lung cancer than those with low fruit orvegetable intake.125 Evidence from cohort stud-ies126130 published since 2000 has tended to rein-force this notion. In the European Prospective In-vestigation Into Cancer and Nutrition Study,131 astrong protective association was observed with fruitbut not vegetable consumption. A stronger protec-tive association was observed for fruit than vegetable

consumption in a pooled analysis of seven cohortstudies.132

To better understand the basis of this protectiveassociation, fruits and vegetables have been groupedinto classes and also examined individually in relationto lung cancer risk. For example, tomatoes133135 andcruciferous vegetables129,135 have been associatedwith a reduced risk for lung cancer in a number ofstudies, at least for the highest vs lowest categories ofconsumption. These food-based analyses can help toclarify whether protection against lung cancer isconferred by the complex mixture contained in fruits

and vegetables or by the presence of specific bio-chemical constituents in particular fruits and vege-tables.

Fruits and vegetables are the major dietary sourceof antioxidant micronutrients. Two different strate-gies are used to evaluate the relationship of micro-nutrients to lung cancer risk in observational epide-miologic studies: (1) using data summarized from

food-frequency questionnaires to estimate micronu-trient intake, and (2) drawing blood samples fromstudy participants and assaying the concentrations ofmicronutrients in circulation. The former approachprovides a better average measure of micronutrientexposure, whereas the latter approach has the advan-tage of measuring micronutrient concentrationscloser to the cellular level, where the postulatedbiological effect occurs. The differences in measure-ment approaches may lead to different results incertain situations. A metaanalysis136 of selenium andlung cancer found that selenium intake as measured

by questionnaire showed no association (RR, 1.0;95% confidence limit [CL], 0.8, 1.3), whereas asso-ciations in the protective direction were observed forselenium concentrations measured in toenails (RR,0.5; 95% CL, 0.2, 0.9) or serum (RR, 0.8; 95% CL,0.6, 1.1).

Studies of both dietary intake137140 and prediag-nostic blood concentrations141,142 suggested a protec-tive association between carotenoids and lung can-cer. The evidence for vitamin C is scant butsuggestive of a protective association, whereas thedata on vitamin A has yielded null findings.143 Re-

ports from cohort studies have tended to reinforcethe previous findings of protective associations withintake of a variety of carotenoids128,135,144 or anantioxidant index.140 However, a pooled analysis139

of seven cohort studies did not find strong protec-tive associations with any carotenoids other than-cryptoxanthin.

More recently, studies have examined phyto-chemicals such as flavonoids and isothiocyanates inrelation to lung cancer risk. Phytochemicals arelow-molecular-weight molecules produced by plants.Of the many classes of phytochemicals, those studied

in relation to lung cancer include phytoestrogens,flavonoids, and glucosinoids. The tumor-promotingeffects of steroid hormones can be blocked byphytoestrogens. Soya beans are a primary source of aspecific class of phytoestrogens known as isofla-vonoids. Flavonoids exhibit potent antioxidant activ-ity. Flavonoid intake has been at least weakly asso-ciated with lung cancer in some of the preliminarystudies145,146 of this topic. Isothiocyanates are me-tabolites of the class of phytochemicals known asglucosinolates. Isothiocyanates could exert antican-cer effects by blocking carcinogens via induction of




13/27

phase 2 detoxification enzymes, such as glutathioneS-transferase. Cruciferous vegetables contain highconcentrations of glucosinolates; therefore, con-sumption leads to higher endogenous isothiocyanateconcentrations. As with cruciferous vegetables,147

lung cancer risk is consistently lower with higherintakes or urinary levels of isothiocyanates.148150

When isothiocyanates have been studied in combi-nation with a common polymorphism in the GSTM1gene, the decreased risk for lung cancer associatedwith isothiocyanates has been especially pronouncedin people with the GSTM1 null genotype.148150 Thisprovides an example of a potential gene/diet inter-action that may be relevant to lung carcinogenesis.

Studies of fruits, vegetables, and micronutrientshave been the centerpiece of studies of diet and lungcancer, but a wide range of dietary and anthropo-metric factors have been investigated. For example,the results of a metaanalysis151 showed that alcohol

drinking in the highest consumption categories wasassociated with increased risk for lung cancer. An-thropometric measures have also been studied, indi-cating a tendency for people with lower body massindex (BMI) to have increased lung cancer riskrelative to heavier people.152,153 However, effects ofboth alcohol drinking and low BMI may be difficultto separate from the concomitant effects of smoking.When considering the possible relationships be-tween lung cancer and factors such as alcohol drink-ing and lower BMI, cigarette smoking cannot bedismissed as a possible explanation.

The overwhelming contribution of cigarette smok-ing as a cause of lung cancer poses a challenge todetecting the role that other lifestyle factors, such asdiet, may play in the cause of lung cancer. Cigarettesmoking is now so closely associated with less health-ful lifestyles in the United States and some othercountries, such as less healthful diets,154 that it isoften difficult to disentangle the dietary factor(s) ofinterest from the effects of smoking. Cigarettesmoke can directly affect circulating concentrationsof dietary factors; for example, smokers tend to havelower circulating concentrations of antioxidant mi-

cronutrients even after accounting for differences indietary intake.154 In addition, associations betweendietary factors and lung cancer risk are likely to befar weaker than the association with active smoking,and diet is measured with much greater error ingeneral than is smoking. Even for a dietary factor,such as vegetable consumption, which is fairly con-sistently associated with a lower risk for lung cancer,the highest exposure category is typically associatedwith at most a halving in the risk for lung cancer.Therefore, in interpreting the evidence, residualconfounding cannot be readily set aside as an expla-

nation for the observed associations between dietaryfactors and lung cancer.155

Chemoprevention Trials: The experimental ratio-nale for trials of beta carotene and retinoids isoffered in another article in thisSupplement(LungCancer Chemoprevention by Gray et al). Experi-

mental data indicated a potential for prevention withthese agents; observational data were supportive ofthe hypothesis that beta-carotene and retinoidsmight have chemopreventive activity.124 However, aprotective association between beta-carotene andlung cancer was not found in three randomized,double-blind, placebo-controlled chemopreventiontrials156158 of beta-carotene reported during the1990s. In fact, beta-carotene supplementation wasassociated with an increased risk for lung canceramong the high-risk populations of heavy smokers inthe -Tocopherol -Carotene Cancer Prevention

Study,156

and smokers and asbestos-exposed workersin the Carotene and Retinol Efficacy Trial.158

In summary, observational evidence suggests thatsmokers who eat more vegetables are at lower riskfor lung cancer than those who consume fewervegetables. The evidence is not as consistent for fruitconsumption. The specific constituents of vegetablesthat confer protection are not known. The results ofthe chemoprevention trials clearly suggest a morecomplex role for micronutrients than previously pro-posed.

Physical Activity: Several studies159161

have re-ported that more physically active individuals have alower risk for lung cancer than those who are moresedentary, even after adjustment for cigarette smok-ing. As with the assessment of any lifestyle factorother than smoking with lung cancer risk, potentialresidual confounding by cigarette smoking needs tobe considered as an alternative explanation.

Occupational Exposures:Investigations of occupa-tional groups, often heavily exposed over a long timeto workplace agents, have provided substantial un-

derstanding of the carcinogenicity of a number ofchemicals and physical agents. Among cancers thatare associated with occupational exposures, cancer ofthe lung is the most common.162 Estimates derivedfrom case-control studies163169 of the proportion oflung cancer that is contributed to by occupationalexposures, via independent or shared causal path-ways, have ranged widely, but most point estimatesor ranges have included values from 9 to 15%.Although disagreement persists concerning specificestimates,170 the message is clear: in industrializednations, the contribution of occupational exposures




14/27

to the lung cancer burden is small compared withthat of cigarette smoking, but large compared withcontributions of most other exposure classes. Ciga-rette smoking potentiates the effect of some knownoccupational lung carcinogens.40

Lung cancer has been observed to be associated withmany workplace exposures. Workers who are exposedto tar and soot (which contains benzoapyrene), such

as coke oven workers,171,172 in concentrations thatexceed those present in urban air173 are at increasedrisk for lung cancer. Occupational exposures to anumber of metals, including arsenic, chromium, andnickel, are also causes of lung cancer.174 For many ofthe worker groups exposed to these agents, there weresubstantial increments in risk. However, in developedcountries, these hazards have largely been controlled.

For some other workplace agents, the evidencehas been less clear. The results of numerous case-control and cohort studies are compatible with aweak association between exposure to diesel exhaust

and the development of lung cancer.175 Althoughinadequate control of cigarette smoking limits theinferences that can be drawn from many of thesestudies, exposure to diesel exhaust remains a likelyexplanation for these findings.175 This associationremains a public health concern because the publicis exposed to diesel exhaust in urban areas, and insome European countries diesel vehicles are increas-ingly used.41

The question of whether silica dust is a risk factorfor lung cancer has been controversial.176178 Atwofold increase in lung cancer risk was estimated

from a metaanalysis179 of the relationship betweensilicosis and lung cancer mortality. Effects of smok-ing were not well controlled in most of the studies.179

The evidence on silica exposure, absent consider-ation of the presence of silicosis, is less clear.180,181 In1997, the IARC did classify crystalline silica as ahuman carcinogen182; however, some still continueto question its carcinogenicity181 and the role of silicaexposure vs that of fibrosis in people with silicosis.180

Asbestos: Asbestos, a well-established occupa-tional carcinogen, refers to several forms of fibrous,

naturally occurring silicate minerals.183 The epide-miologic evidence dates to the 1950s, although clin-ical case series had previously led to the hypothesisthat asbestos causes lung cancer.184,185 In a retro-spective cohort study published in 1955, Doll186

observed that asbestos textile workers at a factory inthe United Kingdom had a 10-fold elevation in lungcancer risk and that the risk was most heavilyconcentrated during the time frame before regula-tions were implemented to limit asbestos dust infactories. A sevenfold excess of lung cancer wassubsequently observed among insulation workers in

the United States.187,188 The risk for lung cancer hasbeen noted to increase with increased exposure toasbestos189 and to be associated with the principalcommercial forms of asbestos.190 Whether asbestosacts directly as a carcinogen or through indirectmechanisms, such as causing chronic inflammationthat eventually leads to cancer development, remainsuncertain.191,192

Asbestos and cigarette smoking both are indepen-dent causes of lung cancer, but in combination theyact synergistically to increase the risk for lung cancerin a manner that is compatible with a multiplicativeeffect.193 Cigarette smoking may increase the lungcancer risk associated with asbestos exposure byenhancing retention of asbestos fibers.194

Radiation: Epidemiologic studies of populationsthat were exposed to high doses of radiation showedthat lung cancer is one of the cancers associated withexposure to ionizing radiation.195 However, the risks

for low-dose radiation, more relevant to contempo-rary workers and the general population, haveproved difficult to characterize.195 Assessing thecancer risk that is associated with low-dose radiationamong humans is methodologically difficult becausethe signal-to-noise ratio is highly unfavorable.196

Nevertheless, large cohort studies,16,197,198 particu-larly the study of Japanese atomic bomb survivors,have provided understanding of the risks of low-doseionizing radiation.

The following two types of radiation, classified byrate of energy transfer to the tissue, are relevant to

lung cancer: low linear energy transfer (LET) radi-ation (eg, x-rays, gamma rays) and high-LET radia-tion (eg, neutrons, radon). High-LET radiation pro-duces ionization of relatively higher density in tissuesthan low-LET radiation, so in equivalent doses, morebiological damage is produced by high-LET thanlow-LET radiation.199 For both types of radiation,the majority of the epidemiologic evidence comesfrom cohorts that were exposed at levels substantiallygreater than those experienced by the general pop-ulation. Risk assessment methods are then used toestimate risks to the population.

Radon is an inert gas that is produced naturallyfrom radium in the decay series of uranium. Two ofthe decay products of radon emitparticles that, byvirtue of their high energy and mass, can causedamage to the DNA of cells of the respiratoryepithelium. Epidemiologic studies200,201 of under-ground miners of uranium and other ores haveestablished exposure to radon daughters as a cause oflung cancer. In the miners who were exposed toradon in past centuries, very high lung cancer riskswere observed; these fell for more recent workers,but the epidemiologic studies16 still show clear evi-




15/27

dence of existing cancer risk. Cigarette smoking andradon decay products synergistically influence lungcancer risk in a manner that is supraadditive butsubmultiplicative.16,201

Radon is of broader societal interest because it is aubiquitous indoor air pollutant that enters buildingsin soil gas. On average, indoor exposures to radon forthe general population are much less than those

received by occupational groups such as uraniumminers. For example, even the lowest historicalradon concentration in a uranium mine is roughly 50to 100 times higher than in the average home.201

Exposure to radon in indoor air is also assumed tocause lung cancer, but the magnitude of the risk isuncertain because of the assumptions underlying theextrapolation of findings from uranium miners to thegenerally lower exposures indoors. These assump-tions relate to dose, dose rate, and dosimetry and alsoreflect the lack of information on risks of exposuresof women and children. Strengthening biological

evidence supports the assumption that a single hit toa cell by an particle causes permanent cellularchange, an assumption that leads to a nonthresholddose/response relationship.

The assumptions made by the EnvironmentalProtection Agency and the Biological Effects ofIonizing Radiation IV and VI Committees of theNational Research Council led to estimates thatapproximately 15,000 to 20,000 lung cancer deathsper year in the United States are caused by radon.202

Case-control studies203,204 concerning indoor expo-sure to radon as a risk factor for lung cancer,

undertaken to assess risks directly, have producedfindings that are generally consistent with downwardextrapolation of risk models based on the under-ground miners. This coherence lends support tousing extrapolation of the miner data to estimate therisk of indoor radon.

Epidemiologic data relating low-LET radiation tolung cancer stem from three principal populations:the atomic bomb survivors in Japan,205 patients withdiseases such as ankylosing spondylitis206 or tuber-culosis207,208 who received multiple radiation treat-ments, and occupational groups in professions that

expose workers to radiation.209 The single, high-doseexposure of the atomic bomb survivors was associ-ated with significant lung cancer risk.205 Regardlessof their age when the atomic bombs were dropped,the excess of lung cancer did not occur until thesurvivors reached older ages, when cancer usuallyoccurs,205 and a consideration of radiation and smok-ing together suggests an additional relationship.198

The risks associated with exposure to lower dosesof low-LET radiation have been estimated in twoways. Statistical models have been used to extrapo-late from the atomic bomb survivors data to lower

doses. Patients who had tuberculosis and receivedradiation therapy have also been studied; they wereintermittently exposed to radiation. Such intermit-tent, low-dose exposures may be most pertinent forthe general population because this exposure patternis the most common in technologically advancedsocieties. Studies of patients with tuberculosis sug-gest that if any risk for lung cancer is associated with

this exposure pattern, then it is small,207,208 suggest-ing that the assumptions on which the higher riskestimates that were obtained from the data of atomicbomb survivors may in actual fact not hold.208

Low-LET radiation therefore seems to be associ-ated with higher lung cancer risk when exposureoccurs at a higher dose rate.208 These results contrastwith those for high-LET radiation, suggesting thatthe two types of radiation have different dose-raterelationships.208

Air Pollution: During a typical day, the average

adult inhales approximately 10,000 L of air.210 Con-sequently, even the carcinogens that are present inthe air at low concentrations are of concern as a riskfactor for lung cancer. Extrapolation of the risksassociated with occupational exposures to the lowerconcentration of carcinogens in polluted ambient airleads to the conclusion that a small proportion oflung cancer cases could be due to air pollution.162,211

Carcinogens that are generated by combustion offossil fuels include polycyclic aromatic hydrocarbonsand metals such as arsenic, nickel, and chromium.174

In considering respiratory carcinogenesis, the con-

stituents of air pollution will vary by locale and overtime depending on the pollution sources.212 Conse-quently, epidemiologic investigations of air pollutionand lung cancer have been limited by the difficulty ofestimating exposure. Nevertheless, descriptive evi-dence is consistent with a role for air pollution incausing lung cancer. Urbanization and lung cancermortality are linked.213215 This association couldarise from differences in the distributions of otherlung cancer risk factors, such as smoking and occu-pational exposures, by degree of urbanization. Ad-justment for these factors may considerably attenu-

ate the effect of urban location,216,217 but an urbaneffect persists in a number of studies.40

Air pollution has been assessed as a risk factor forlung cancer in both case-control and cohort studies.Whereas early evidence from case-control and co-hort studies was found wanting, more recently theevidence supports a causal role for air pollution.218

Two prospective cohort studies219,220 that partiallyaddressed weaknesses of earlier studies add evidencethat suggests air pollution is weakly associated withthe risk for lung cancer. By prospectively studying airpollution levels in relation to risk for lung cancer and




16/27

by controlling for possible confounders such as age,smoking, and socioeconomic status at the individuallevel, these studies surmount some shortcomingsnoted of much previous research.221 In a study of sixUS cities,219 the adjusted risk for lung cancer mor-tality in the city with the highest concentration offine particles was 1.4 times (95% confidence interval[CI], 0.8 to 2.4) higher than in the least polluted city.

Using data from the American Cancer Society CPSII, Pope et al220 observed that compared with theleast polluted areas, residence in areas with highsulfate concentrations was associated with an in-creased risk for lung cancer (adjusted RR, 1.4; 95%CI, 1.1 to 1.7) after adjustment for occupationalexposures and the factors mentioned previously.However, unlike in the Six-Cities Study,222 fine-particulate concentration was not associated withlung cancer risk. In a subsequent update, follow-upwas extended to 1998. In that report, the risk for lungcancer was observed to increase 14% for each 10-

g/m3 increase in concentration of fine particles.By contrast, in the American Cancer Society CPS

I cohort, air pollution was not associated with lungcancer risk; in that study, men were stratified accord-ing to exposures in the workplace, but exposureassessment for air pollution was based on proxy, lessspecific measures of air pollution.221 Some case-control studies223225 have reported indexes of airpollution to be modestly associated with elevatedrisks for lung cancer, but others226 have reported noassociation.

Another research approach to evaluate the risk of

air pollution has been to investigate populations thatreside around point sources of pollution, such asfactories and smelters. Proximity of residence to thepollution source can be used as a proxy for exposure.Many industries have been studied using this ap-proach. Areas surrounding nonferrous smelters,which emit arsenic, have been of particular interest.An ecologic study reported by Blot and Fraumeni35

in 1975 suggested that excess lung cancer occurredin US counties with copper, lead, or zinc smeltingand refining industries. The results of several subse-quent case-control studies227229 lend support to this

hypothesis by showing that the risk for lung cancerincreased the nearer that people lived to nonferroussmelters, after accounting for personal cigarettesmoking and employment at the smelter. Othercase-control studies230,231 did not replicate this find-ing but were also limited by their failure to accountfor smoking and employment at the smelter.

Doll and Peto,162 in their 1981 review of thecauses of cancer, estimated that perhaps 1 to 2% oflung cancer was related to air pollution. Even in lightof more recent findings, this seems to remain areasonable estimate.232 The body of evidence linking

air pollution to lung cancer is solidifying,218 but thepublic health impact of this exposure is small relativeto cigarette smoking, at least in developed countrysettings where research has been conducted. This isto be expected, given that respiratory doses ofcarcinogens from active smoking are significantlygreater than those received from the inhalation ofatmospheric contaminants.

An individuals total exposure to air pollutiondepends on indoor as well as outdoor exposures.Indoor air quality has large potential health implica-tions because people may spend substantial amountsof time indoors. Indoor air pollution may stem fromincoming outdoor air or originate indoors from to-bacco smoking, building materials, soil gases, house-hold products, and combustion from heating andcooking.233 A trade-off exists between energy effi-ciency and indoor air quality because ventilationallows heated/cooled air to escape but improvesindoor air quality.234

As discussed, in more developed countries, two ofthe most important indoor pollutants that moststrongly increase lung cancer risk in never-smokersare passive smoking13 and radon.202 Asbestos expo-sure may pose a risk to building occupants, butconcentrations are generally very low.183 Of majorconcern in the developing world is the indoor aircontamination resulting from the use of unprocessedsolid fuels, notably coal, for cooking and spaceheating.235 Mumford et al236 inferred that smoky coalwas likely to be a major determinant of the geo-graphic distribution of lung cancer in Xuan Wei,

China, a finding corroborated by an animal model.237Evidence supporting a causal association wasstrengthened by the results of a retrospective cohortstudy that showed that switching from use of un-vented fire pits to stoves with chimneys almosthalved the risk for lung cancer.238

Host Factors:Genetic susceptibility to lung cancerhas long been postulated. Environmental agents,even cigarette smoking, cause lung cancer in only aminority of exposed people, leading to the hypothesisthat susceptibility is inherently determined. Epide-miologic studies244 showing that a family history oflung cancer predicts increased risk further support agenetic basis for lung cancer susceptibility. Thislong-postulated hypothesis is now being activelyaddressed using the approach of molecular epidemi-ology. Full coverage of this topic is beyond the scopeof this report; aspects of genetic susceptibility forlung cancer have been reviewed.239243

Familial aggregation of lung cancer has beenprimarily demonstrated in both case-control andcohort studies.244 In these studies, a family history oflung cancer tended to be associated with increased




17/27

risk for lung cancer; most of the studies controlledfor smoking, which is known to aggregate in families.In a large study in Louisiana, segregation analysissuggested that lung cancer inheritance was consis-tent with a Mendelian codominant autosomal genedetermining early onset of disease.245 Conversely,the largest study of lung cancer in twins reported todate did not provide evidence indicating a genetic

basis for susceptibility.246 Follow-up of 15,924 maletwin pairs in the United States did not show greaterconcordance in monozygotic compared with dizy-gotic twins, and death rates from lung cancer weresimilar by zygosity group in surviving twins whosesibling died of lung cancer. The results of a linkageanalysis based on 52 extended pedigrees indicatedthat a locus on chromosome 6q2325 was associatedwith a major susceptibility to lung cancer.247

In a genetic epidemiology study of lung cancer innonsmokers in Detroit, Schwartz et al248 exploredfamilial risk for lung cancer and found an association

between risk and a history of lung cancer in a first-degree relative (odds ratio, 1.4; 95% CI, 0.8 to 2.5). Theassociation was much stronger in those aged 40 to 59years at diagnosis compared with older people. Thispattern of risk with age suggests that genetic factorsmay be more important at younger ages. This generalfinding was confirmed by a subsequent, complex seg-regation analysis of the same data.249

Research Findings on the Genetic Basis of LungCancer: With application of the new and powerfultools of modern molecular and cell biology, researchfindings are now characterizing the changes in cells thatare caused by exposure to tobacco smoke and providinga framework for understanding the genetic and epige-netic basis of lung cancer risk. Figure 3, proposed byHecht,121 offers a general schema for the process ofcarcinogenesis by tobacco smoking. Viewed in theframework set by this type of model, research findingsmirror the predictions of the multistage model in manyrespects and are enhancing understanding of the mech-anisms by which smoking causes cancers of the lungand other organs. A rapidly expanding literature ad-dresses dosimetry and metabolism of tobacco carcino-

gens at the cellular and molecular levels, genetic deter-minants of susceptibility, and patterns of geneticchanges in the tissues of smokers and in the cancersthat the tissues develop.121,241 Much of the researchconducted to date has been based in case-controlstudies that compared the genotypes of lung cancercases with those of control subjects. Studies have alsobeen conducted using cohort designs with affected andnonaffected people sampled from the cohort and bio-logical samples analyzed for the markers of interest.

The understanding of the epigenetic changes thatmay be involved in the causal pathway to lung canceris advancing rapidly. For example, there is increasingevidence that methylation of cytosine in the DNA,leading to hypermethylation of promoter regions, isfrequent in most types of cancers, including lungcancer.250 Promoter regions of many human geneshave loci rich in CpG dinucleotides, regions referredasCpG islands.250,251 Hypermethylation of the CpGislands can be detected by polymerase chain reaction

methods. Cells with abnormal methylation of geneshave been detected in sputum before the diagnosisof lung cancer, suggesting that hypermethylationcould be a useful marker for early detection.252,253

In a general formulation of determinants of cancerrisk, the risk depends on carcinogen exposure andthe factors that determine host susceptibility, includ-ing genetic predisposition.254 For tobacco smokingand lung and other cancers, the elements of thisparadigm all are topics of inquiry, using the combi-nation of laboratory- and population-based studiesindicated in the diagram. Biomarkers are central tothe molecular epidemiology approach; the term re-fers to making measurements of indicators of expo-sure and dose, susceptibility, and response in biolog-ical materials, including tissue samples, blood, urine,and saliva.255 As research evolves within this para-digm, a more complete biological understanding ofthe specific events underlying the multistage model,originally proposed on a conceptual basis, can beanticipated.

This framework indicates multiple points wheregenetically determined host characteristics might beimportant: carcinogen metabolism and

Epidemiology of Lung Cancer* ACCP Evidence-Based Clinical Practice Guidelines. 2007pdf

Documents

Transcript of Epidemiology of Lung Cancer* ACCP Evidence-Based Clinical Practice Guidelines. 2007pdf