Document xz3EKkKZ5LkXwXmaMYebLr7q0

Critiques of EPA External Review Draft 600/6-90/006A, M ay 1990 Health Effects of Passive Smoking: Assessment of Lung Cancer in Adults and Respiratory Disorders in Children prepared by Biomedical & Environmental Consultants, Inc. Richland, WA 9 9 3 5 2 February 1991 X00 Contents iii. Preface iv. List of Authors vi. Executive Summary 1. Individual Critiques (in alp h ab etical order of authors) 1. Critique by E.L. Alpen, Ph.D 2. Critique by Michael G. Bissell, M .D., Ph.D., M.P.H. 3. Critique by Martin J. Cline, M.D. 4. Critique by Alvan R. Feinstein, M .D. 5. Critique by Robert W. Gardner, Ph.D. 6. Critique by Alan J. Gross, Ph.D. 7. Critique by Gnter Oberdrster, D.V.M., Ph.D. 8. Critique by Theodor D. Sterling, Ph.D., et al. 9. Critique by Bruce O. Stuart, Ph.D. 10. Critique by Alfred P. Wehner, D.M .D., Sc.D., Cand.Med., DATS. 2. A p p e n d ix -- 3800 Lung C a n c e r Deaths Annually in th e U.S. in Nonsmokers d u e to their Exposure to Environ m ental C ig arette Sm oke, Fa ct or Fiction?-- A lan J. Gross, Ph.D. an d Derek B. Janszen, M.S. s< > 51246 Preface O n November 11,1990, the R.J. Reynolds Tobacco Company (RJR) com missioned Biomedical and Environmental Consultants, Inc. (BEC) to independendy review the EPA External Review Draft 600/6-90/006A, May 1990, titled "Health Effects o f Passive Smoking: Assessment o f Lung Cancer in Adults and Respiratory Disorders in Children." Seven BEC members and three non-BEC scientists/clinicians undertook this task. A workshop was held on January 12/13,1991, in Seattle at which the individual critiques were presented and discussed. These critiques, in alphabetical order o f their au thors, and an executive summary are presented in this report. T he individual critiques reflect the different professional backgrounds and experiences o f the authors but show consensus that the EPA Draft needs to be thoroughly revised. The paramount issue in the EPA Draft is the question whether or not there is an association between environmental tobacco smoke (ETS) and lung cancer in adults. The second part o f the EPA Draft deals with ETS effects on respira tory disorders. This topic is included in individual critiques o f this report but not in the executive summary which focuses exclusively on the primary issue o f ETS and lung cancer. T he Appendix to this report was prepared by Dr. Alan J. Gross, and Mr. Derek Janszen, following the workshop in Seattle. A.P. Wehner, President Biomedical and Environmental Consultants, Inc. February 1991 iii 51246 0003 List of Authors Edward L Alpen, Ph.D. Professor of Biophysics University o f California, Berkeley Professor of Radiology University o f California, San Francisco Michael G. Bissell, M.D., Ph.D., M.P.H. Director o f Clinical Pathology, City o f Hope National Medical Center Duarte, CA 91010-0269 Martin J. Cline, M.D., Bowyer Professor of Medical Oncology UCLA Department of Medicine Division of Hematology-Oncology Los Angeles, CA 90024-1678* *presently at: Ludwig Institute for Cancer Research, London W lP 8BT, Great Britain Alvan R. Feinstein, M.D. Professor of Medicine and Epidemiology Director, Clinical Epidemiology Unit Yale University School of Medicine New Haven, CT 06510 Robert W. Gardner, Ph.D. Professor emeritus Ezra Taft Benson Agriculture and Food Institute Brigham Young University Provo, UT 84602 IV 51246 0004 Alan J. Gross, Ph.D. Professor of Biostatistics Medical University of South Carolina Charlston, SC 29425-2503 Gtlnter Oberdorster, D.V.M., Ph.D Professor of Toxicology University o f Rochester School of Medicine and Dentistry Environmental Health Sciences Center Rochester, NY 14642 Theodor D. Sterling, Ph.D. Professor, Faculty of Applied Science School of Computing Science Simon Fraser University Burnaby, BC V5A 1S6 Bruce O. Stuart, Ph.D. Program Manager, Inhalation Toxicology NSI Technology Services Corporation Brookhaven National Laboratory Upton, NY 11973 Alfred P. Wehner, D.M.D., Sc.D., Cand.Med., D.A.T.S. President, Biomedical 6c Environmental Consultants, Inc. Richland, WA 99352 51246 0005 EXECUTIVE SUMMARY The U.S. Environmental Protection Agency (EPA) document EPA/600/6-90/006A on "Health Effects o f Passive Smoking: Assessment o f Lung Cancer in Adults and Respiratory Disorders in Children" is a preliminary external review draft, hereafter referred to as Draft, issued in May 1990. Based on the review o f data presented in the U.S. Surgeon General's report o f 1986 on environ mental tobacco smoke (ETS), the National Research Council (NRC) report o f 1986 on the same subject, and on subsequently published studies, the Draft concludes that ETS is a Group A carcinogen. EPA's Risk Assessment Guidelines o f 1986 define a Group A carcinogen as a human carcinogen, a designation used according to EPA's 1986 Guidelines "only when there is sufficient evidencefrom epidemiologic studies to support a causal association between exposure to the agents and cancer. * According to the guidelines, "Three criteria must be met before a causal association can be inferred between exposure and cancer in humans: 1. There is no identified bias that could explain the association. 2. T he possibility o f confounding has been considered and ruled out as explaining the association. 3. T he association is unlikely to be due to chance." The Draft based the Group A classification of ETS mainly on the results o f statistical analysis of (1) 19 case-control studies and (2) two cohort studies. The critiques in this report examine the validity o f the data and o f the arguments upon which EPA's conclusion is based. It should be realized that the Draft is not a scientific document containing new evidence, but a selective summary and a selective analysis o f previously published data. It needs to be evaluated whether the original data were adequate or inadequate, accurate or inaccurate, and whether the authors o f the Draft were justified or not justified, accurate or inaccurate, in their analysis and interpretation o f data from others. s< > 51246 Meta Analysis T he Draft combines results o f several previous studies in a statistical meta analysis. M eta analysis describes quantitative methods for combining results across a number of studies. The underlying tenet o f this relatively new and not yet completely evaluated technique is that by combining results from a representative sample o f a multitude of comparable studies a more accurate picture o f reality may be obtai ned. M eta analysis can be applied to increase statistical power by analyzing total evidence from all studies simultaneously, and as a means o f deriving an overall average risk estimate from the separate risk estimates of the individual studies. This statistical tool is useful, for example, when any effect that might exist is likely to be only moderate and when there is an expectation that, while effects o f different studies may have different magnitudes, all o f them may be expected to point in the same direction. Strict criteria have been developed for the appropriate application of meta analysis (see, for example, Yusuf et al, 1985; Chalmers et al, 1989). these include (1) preparation o f a detailed protocol, defining procedures and criteria to be used, before statistical analysis, (2) extensive data search to ensure that the studies pooled aire representative o f all existing studies, including those . showing no effect, (3) comparability of what is measured and compared in the pooled studies, and (4) relative invariance o f results when available data are reorganized or other relevant data added. As shown in the individual critiques o f this report, none of these criteria are met in the Draft. The 19 Case-Control Studies M uch o f EPA's case against ETS rests on the results o f EPA's meta analysis o f 19 case-control studies with nonsmoking females exposed to ETS from smoking spouses. Closer examination shows that these studies vary widely in quality as well as in their definitions o f exposure and endpoint, thereby violating one of the fundamental rules for the appropriate application of meta analysis. The Draft recognizes numerous confounding differences in the case-control studies but nevertheless subjects the data to meta analysis. These differences include (1) definition of ETS exposures, (2) inclusion of former smokers as nonsmokers, if they did not smoke for some minimum period, in some studies but not in others, (3) questions, leading to classification as ETS-exposed subjects, varied across studies, (4) exposure percentages vary from 15 to 84% across studies, (5) reference populations are defined by different parameters, (6) study design, vii 51246 0007 -- i `4l A.V.V (7) experimental protocol, (8) data analysis and interpretation, (9) potential confounding factors, and last but not least (10) confirmation o f primary lung cancers. Sterling et al ( see critique by Sterling et al) counted at least eight different definitions of cases and twelve different definitions o f exposure! Various biases (e.g., publication bias; selection bias; biases due to smoker misclassification, misdiagnosis o f lung cancer, and proxy respondents) can and do lead to distorted and invalid results, as the following examples will show. The combined 19 case-control studies in the Draft yield a pooled risk ratio (RR) o f 1.42 (95% C l 1.24,1.62) following EPA's meta analysis. However, if the Varela study-- a well conducted investigation showing no spousal ETS effects, that was omitted for no valid reasons-- h ad been included in EPA's analysis, the RR would have been markedly reduced, namely to 1.27 (95% C l 1.13,1.42). In addition, EPA elected to restrict its analysis to nonsmoking wives o f smoking husbands, excluding nonsmoking husbands of smoking wives. T he reason given in the Draft for the exclusion of ETS-exposed males is lack of data for the latter situation. However, such data do exist, albeit on a smaller scale. W hen subjected to similar analysis, the data for males show no increased risk of lung cancer for ETSexposed males married to smoking females (RR=1.07, 95% C l 0.86, 1.33; see critique by Ster ling et al in this report). Furthermore, the Draft lumps together studies based on U.S. and nonU.S. populations. However, for many cultural, social, economic, and genetic reasons, it is open to question whether the U.S. and non-U.S. populations can be combined. Estimation o f risk for the U.S. population ought to be based on studies using the U.S. population as subjects. How ever, when this is done, the risk for females fails to be significantly elevated (RR=1.11, 95% Cl 0.96,1.28) which is also true for risk for males. (Risk for males is RR=1.05 95% C l 0.83,1.31; see critique by Sterling et al in this report). The selection of those studies to be pooled that result in significandy elevated risk and the omission o f the pool of studies that fail to support elevated risk is a serious flaw in the D raft's analysis. At the very least the Draft ought to have listed results from applying meta analysis to males and to U.S. and non-U.S studies separately. As is, the pattern o f selection and omissions appears to betray a strong bias by the Draft's authors. The individual critiques in this report will highlight additional biases. In addition to inappropriate use of statistical procedures and biases, the Draft contains a number of other flaws. For example, in attempting an ETS risk assessment for the national nonsmoking population, the Draft extrapolates from the (already questionable) spousal data to the general viii in N<>) s<>s> <oso> ________________________________________________ ___ ___________ population. Given the dose physical proximity in a spousal relationship, this procedure appears inappropriate, introducing a positive bias because non-spousal ETS exposures generally tend to be to markedly lower concentrations than spousal ETS exposures. The definition o f lung cancer cases varies considerably across the case-control studies (e.g., see critique by Sterling et al, p 8-13/14). Indoor air contamination with radon decay products, present at widely differing concentrations in indoor environment, is a confounding variable that is strongly implicated as a causal factor in the development o f respiratory tract cancer. The Draft does not address this problem. Detection bias, one o f the major biases to be expected in observational cohort and case-control studies, is also ignored as a confounding variable: many lung cancers are not identified during life and spouses o f smokers may be more likely than spouses of nonsmokers to seek and receive the tests and examinations that lead to a diagnosis during life. The Cohort Studies The Draft cites die results o f cohort studies as additional evidence for the carcinogenic effect of ETS. At issue are the Hirayama study in Japan and the Garfinkel study in America. Hirayama used 265,118 subjects, Garfinkel used more than 1million. Both studies were subject to criti cism, the former more so than the latter which is generally regarded the technically better study. The Hirayama study has results that vary with the type of analysis designed to correct a number o f basic flaws, most o f them concerned with adequate age adjustment. For some o f these correc tions, the Hirayama study still shows a statistically significantly elevated lung cancer risk for nonsmoking wives exposed to ETS where there are smoking husbands, but for odiers it does not. T he Garfinkel study never showed such a risk. Yet, the Draft authors chose to emphasize the Hirayama findings and to downplay the Garfinkel results. They proceeded with a quantitative ETS risk assessment for the nonsmoking US population, based on the combined results (Hirayama + Garfinkel), claiming a significant ETS risk. This procedure violates basic principles of epidemiological statistics: a valid quantitative ETS risk assessment for the US population must be based on US conditions and US data. The obvious reasons for this postulate are to exclude a host of confounding variables each o f which can affect the results o f such a study. These variables ix 51246 0005 include cultural, dietary, geographic, genetic and many other potential differences (e.g., average room size, interior arrangements, air changes per hour, composition o f Japanese cigarettes) between American and Japanese populations. Such differences can affect the respiratory tract dosimetry of ETS significandy and preclude the uncritical transfer of the Japanese data to the U.S. population for quantitative risk assessment as opposed to qualitative risk assessment (see critique by Oberdorster). An ETS risk assessment for the nonsmoking US population, based on the Garfinkel data, shows no significant lung cancer risk and no dose-response relationship. Yet, the presentation and interpretation of the results of these two studies are rather biased in favor of the Japanese study (for further details see individual critiques). For example, the D raft states on page 3-43, "The Japanese cohort study alone provides compelling (editor's emphasis) evidence o f a lung cancer risk associated with ETS exposure". Two sentences later, the Draft comments on the bigger, seemingly better and much more relevant Garfinkel study as follows: "Results o f the American cohort study are less conclusive" (editors emphasis), instead o f stating the simple fact that there was no evidence of such a risk in the American study. T he Draft authors appear quite unconcerned about the underlying possible causes for the puz zling discrepancy between the results of the two studies. Similar indifference to other discrepan cies is evident elsewhere in the Draft. Foir example, there is a marked difference between relative lung cancer risk derived from epidemiology data and comparative dosimetric estimates from both particulate materials and vapor phase components, these dosimetric estimates being orders o f magnitude lower (see, for example, critique by Stuart). It is scientifically unacceptable to choose among contradictory data those one likes and to disregard the others. Additional Comments The dosimetry o f ETS is discussed in A ppend C of the Draft. Reference is made to an early International Commission on Radiological Protection (ICRP) particle deposition and retention model which, however, is described incorrectly in the text and in an accompanying figure. Appendix C shows a number o f conceptual problems and, in many cases, evidence o f inadequate grasp of the current state of the art of quantitative prediction o f fractional deposition of inhaled particulates. Basic differences in relative deposition in tracheobronchial vs. peripheral lung regions between ETS and mainstream smoke (MS) exist, e.g., relative deposition per unit surface area is higher in the upper conducting airways for MS whereas for ETS it is higher in the transi tional zone o f the lower respiratory tract. 51246 Furthermore, the authors of Appendix C appear to be unaware of the basic precepts of the rapidly developing science o f physiologically based pharmacokinetic models, which describe dynamic and generally first-order interactive processes for distribution o f agents from absorption site into blood and subsequently into target organs. Such models are particularly important for adsorption of substances like nicotine which is absorbed through mucosal sites of the mouth, and o f the nasopharynx when inhaled in the gas phase (as in ETS), an important process which the draft document incorrectly states not to occur, (see critiques by Stuart, p. 9-7, and Bissell, pp. 2-3/4). Assessment o f Lung Cancer Risk T he D raft's assessment o f lung cancer risk appears flawed for several reasons. T o begin with, it is based on inappropriate statistical analysis (see Meta Analysis, above) that yields erroneous figures at the start. In addition, while the Draft acknowledges a number of confounding vari ables, other important confounding variables are ignored. Among them rank radon progeny (see critique by Stuart) and secondary exposure to occupational carcinogens in the home (see Appendix to critique by Sterling et al), to name just two. There appears to be no diligent effort in the Draft to search for potential alternative explanations for lung cancer among spouses of smokers. O f the variables acknowledged in the Draft, those whose adjustment tends to result in lower risk ratios, are played down as essentially inconsequential, supported by a selectively chosen literature reference (example: publication bias, p. 3-33). Referring to the N R C report (1986), the Draft differentiates between "unexposed" and "truly unexposed" subjects. This differentiation is based on cotinine measurements in nonsmokers not knowingly exposed to ETS. These measurements were interpreted to indicate exposure to back ground ETS, based on the assumption that ETS is the only source of nicotine. Ignored is the fact that nicotine also occurs in certain vegetables o f the solanacae family (e.g., egg plant, potatoes, green pepper, tomatoes) that might well account for at least part of the cotinine found in non smokers. Failure to recognize and adjust for this possibility casts doubt on the validity o f the cotinine data as a measure o f ETS exposure. Yet these questionable cotinine values are used in the Draft to convert "unexposed subjects" to "truly unexposed" subjects. The effect which adjustments have on RR is demonstrated in the following paragraph on page 3-32 o f the Draft: xi 51246 0011 "the overall estimate of RR in the N R C report places the excess risk of lung cancer associated with spousal smoking at about 34%. An adjustment for possible misclassification of the neversmoker status reduced the value to 25%. A second adjustment to make this risk relative to a truly unexposed subject, i.e., to take into account a background level o f exposure, raises the increased risk to 42%." O n page 4-28 o f the Draft, it is further assumed that (1) lung cancer risk from passive smoking is linearly related to (these questionable ) cotinine concentrations, (2) background ETS consti tutes about one-third o f the total ETS exposure of a nonsmoker married to a smoker, using (these questionable) cotinine concentrations as an index o f total ETS exposures, and (3^ cotinine is a constant multiple o f the carcinogenic potency o f ETS at low doses. The questionable validity n ftli* rn iin n A v t* ac nrlratrtrc n f F T S mrrwcnre rMirlrts all ttifwA aRsiimntinns niiextihnable. EPA's lung cancer risk assessment is based on linear extrapolation to zero o f findings in smokers, who are exposed to mainstream smoke plus much higher concentrations o f sidestream smoke than nonsmokers. This assumes that there is no threshold for ETS effects or, in other words, that even very low concentrations, such as typically encountered by nonsmokers in ETS, are harmful. It further assumes that mainstream smoke equals sidestream smoke. The latter assumption is demonstrably wrong, and the no-threshold assumption ignores effective defense mechanisms o f the human body (e.g., macrophages, mucociliary clearance, enzyme induction/activation), capable o f dealing with low concentrations o f numerous insults. The no-threshold concept further ignores the phenomenon o f hormesis (see critique by Wehner). W hile the EPA is policybound to adhere to the no-threshold concept, points that atgue against this position should be discussed in the Draft. <s> 51246 Conclusions The consensus o f the authors o f the individual critiques and participants in the Seatde workshop (January 12/13,1991) can be summarized as follows: (1) Attempting to determine biological effects o f ETS is an extremely complex and difficult | task vexed by numerous, often ill-defined confounding variables, insufficient measure ments, lack o f standardized experimental protocols, and controversial data interpretation. . vt ij (2) Because currently available data are incomplete and ambiguous, they cannot bp used for a valid quantitative risk assessment for the general U.S. population. (3) For reasons detailed in this report, EPA's classification o f ETS as a Group A carcinogen is scientifically not justifiable: it does not meet the definition o f a Group A carcinogen in EPS's 1986 Guidelines which requires "sufficient evidence from epidemiologic studies to support a causal (editor's emphasis) association between exposure to the agents and cancer.". A.P. W ehner, Editor A Critique of EPA External Review Draft 600/6-90/006A, M ay 1990 Health Effects of Passive Smoking: Assessment of Lung C a n ce r In Adults . an d Respiratory Disorders in Children prepared by Edward L. A lpen, P h.D . University of California, Berkeley University o f California, San Francisco 51246 0014 Meta Analysis T he method o f analysis o f multiple studies searching for the same end-point is well established in epidemiological usage (first suggested but not applied by Mantel-Haenzsel, 1959), and a M EDLINE search done by me lists over 200 recent applications o f the method in the medical literature. It is important to understand that the method is not limited to, nor was it even originally designed for, studies in the social sciences, as suggested by the authors o f the Industry critique. The authors o f the EPA draft (hereinafter referred to as the "Draft") have undertaken reasonably appropriate analyses o f the data they chose to analyze, even though this is a limited set and many criticisms can be leveled at the individual studies. `K There has not been enough attention paid by the authors o f the Draft to the inevitable short comings of the meta-analytic method. Because o f the widespread use o f meta analysis ih modern I epidemiological analysis o f clinical trials, there has been extensive study o f the strengths and I limitations o f the methodology. I draw, in particular, two studies to our attention. Chalmers ! et al (1989) very explicitly draw up a set o f rules for selection o f the studies that should be j admitted to a meta-analytic retrospective study. Some o f these rules certainly have not been met ; by the authors o f the Draft under consideration. Laird and Mosteller (1990) review the statisti- I cal methodology in detail. T he latter study is part o f an issue o f this referenced journal devoted I to a review of the uses o f meta analysis (in Int. J. o f Technology Assessment in Health Care, vol i 6,1990). Before losing track o f an observation in a more detailed critique o f meta analysis, it is important j to note that there is a reasonably useful test o f the strength o f any particular meta analytic proce dure. This test is the robustness or stability o f the outcome o f the analysis as more data sets are added. Yusuf et al (1985) is the important source the Draft uses for much of its analysis about the design of meta analytic approaches, but they then proceed to ignore most of the important constraints suggested by Yusuf et al. I would like to consider one of the points made strongly by Yusuf et al. That is the question o f "publication bias". This is clearly an important issue, but Yusuf tends to produce selfparalysis by the emphasis he places on this part o f the process. Chalmers (1989), I believe, presents the more balanced view o f the importance o f a study selec- i I ; I 51246 0015 1-1  ' iiliu W U J v .1 L tion and indusion. H e suggests an aggressive search o f the relevant literature indices and sources, j with attention to the possibility o f unpublished studies. This author suggests, quite rightly, in i my opinion, that the most important step in meta analysis is the development o f a written ! protocol defining the selection criteria, specific exdusion criteria, the statistical tests to be used, the hypothesis to be tested and the methods that will be used for establishing the literature i search for the data base. Both Chalmers and Laird suggest the possibility o f a weighting scheme k that could be defined in advance to weight various studies for their shortcomings or advantages. ; T he most important and well known difficulty o f the meta analysis approach is mentioned almost in passing by the Draft authors, but I believe it needs much more attention th a | attrib uted to it. The shortcoming is that if there is a consistent bias present in all o f the studies : subjected to analysis that the meta analysis outcome will be equally biased, in fact Mantel con- ! dudes that in certain circumstances meta analysis can amplify bias. Gross (1989) has discussed a ; few o f these biases, which indude misdassification o f disease, consistent misreporting o f smoking status, socioeconomic dass, weighting and others. O ther biases that are present in the D raft's ; meta analysis indude choices of cases and controls, second person interviews to determine |< smoking status and many others. T he great difficulty is the existence o f an unknown generally existing bias o f design or experiment which causes similar outcome distortions in all o f the i studies meta analyzed. I cannot agree with either Gross (1989) or Layard (1989) that meta analysis is only appropriate if the study populations are homogeneous with respect to exposure indices, demographic, social and other dassification parameters. As long as there has been no selection o f studies by criteria upon which the outcome is dependent the method should be appropriate. I j It is crucially important that the reader examining the outcome o f the meta analysis knows the criteria for selection and inclusion in the study, since one should apply the analytic outcome only to the population groups preselected. This application is both inclusive and exclusive. T hat is to say, a meta analysis devised w ithout regard to ethnic origin can only be used to make inferences about a similarly distributed set o f studies o f differing ethnic populations. It would also exclude the possibility o f making a specialized inference about, for example, American Caucasians. The relative risk o f passive smoking given in the NRG report, namely 1.3, has been criticized for preselection that did not meet these criteria (Letzcl, H.E. et al, 1988). These authors concluded 5124 that three o f the studies included in the NRC (1986) report did not meet the preselection criteria established, and that these particular studies contributed greatly to the meta analysis test o f significance. T he same criticism can be made about the Draft, but in this case the selection criteria have not been shared with us, so we don't know the extent to which they have been violated. Also, the extent o f data manipulation in the original studies used in the Draft's meta analysis is unacceptable, particularly when the nature o f such manipulation has not beep fully disclosed. Koo(1988) reported another important confounding variable which may be generally present throughout most o f the studies included in the Draft's meta analysis: th$ wives o f nonsjnokipg . , r "'^V*^1V-rCV'K'- husbands had generally healthier life styles and less exposure to other possible dietary and other .^. . ,T; sources of lung cancer. T he authors are aware of the bias introduced by the self-reporting of smoking status o f the "never-smoker". Surely few people today would report themselves as a smoker when they were indeed a never-smoker. All o f the bias is in the other direction, In fact, in some cultures, the secret smoker female spouse may be very unwilling to share her smoking status with the interviewer. T he Draft authors follow the lead of others in attempting to correct for this bias, but the results are simply not credible. Furthermore, their analysis is so sketchy that an independent review o f the process is not easy. Letzel et al(1988) undertook the correction ! of smoker misdassification for the data o f the NRC (1988) report and found that it created the necessity for "...accepting the null hypothesis or creating new empirical evidence and performing a really good study." I find Letzel et al's corrections o f the data no more convincing than that o f the EPA study. I am sure it will be noted by all o f our committee members that the use o f "proxy responses" or "surrogate responses" to determine the smoker, nonsmoker dassification is not consistent among the studies induded in the Draft's meta analysis. T he Draft authors suggest that the proxy approach is indeed reliable, but extensive review o f the literature on the subject would not support such a position. W hile considering the smoker misdassification issue I would draw your attention to the report by Friedman et al (1983) which raises an even more important issue about smoker dassification for Americans in passive smoking studies. This report states that for 47% of nonsmoking wives married to smokers, there is no exposure to passive smoke in the home. This represents a strong 5124 W.V1IttCu statement about social norms in the U.S.A. It would suggest strong spousal pressure for the smoker not to smoke while at home. I will not take the time to review the data having to do with socioeconomic status o f spousal non-smokers, since one o f our Committee members has been involved in such studies and will undoubtedly report upon them in detail. Sufficient to say that it is clear that the likelihood o f spousal exposure will be certainly greater in the blue-collar community. W ith such a socioeco nomic bias, it is dear that other aspects o f life style such as work environment exposure and nutrition will play major biasing roles in studies which foil to control for this variable, j Inferences from the Meta Analysis . For the S statistic analysis the authors o f the D raft have chosen to use a "one-tail" test o f signifi cance for accepting or rejecting the null hypothesis that there is no higher risk ratio for lung cancer in those exposed to ETS than those who are not. One might argue with this approach, but it is a criterion often used in epidemiological studies. For example, the one-tail approach is frequently used in low risk ratio studies o f cancer produced as the result o f exposure to environ mental radiations o f one sort or another. Malcolm Pike has, for example, used a one-tail test for the significance o f an increased odds ratio for leukemia in inhabitants o f Utah exposed to the Nevada test site fallout (report in press). j | | Moreover, I believe that the extrapolations made in the Draft from spousal smoking to the national population are well beyond the range o f credibility. ;j Nearly all o f the studies included in the meta analysis are based upon exposure o f a female spouse ! to sidestream smoke from a smoking husband. Even under the best o f circumstances, which are j not to be seen in this case, extrapolation to a national population, both male and female, will be an extremely uncertain process. In those cases where smoking in the home occurs, there is a i special case o f spousal propinquity which would maximize the nonsmoker's exposure to j sidestream smoke. Smoking at the dinner table, in the car, in bed, at the television, will all be ! associated with nearness o f the nonsmoking spouse. It has been suggested by several authors, and it seems a reasonable conclusion, that at the work place and outside the home the non smoker will tend to avoid the smoker, and even place restraints upon the smoker. The propin- 1-4 51246 0018 quity issue and the avoidance issue make, as far as I am concerned, extrapolation o f spousal exposure data nearly useless for quantitative estimates o f associated lung cancer deaths. T he spousal misclassifcation issue (as to smoker-nonsmoker) is treated in such a cavalier fashion in the Draft that it is hard to believe that the authors really examined the extensive review o f the subject in the N RC (1986) report. I commend Table 12-5 from the NRC report for your consideration. This table does a model calculation assuming the real RR to be 1.0 and a misclassifcation among 100 000 women, o f whom 35% are smokers, and 8% o f these smokers . _^ . ' ' ' . J' - misreport themselves as nonsmokers. The observed RR is 1.3 while th real RR is still 1.0. The likelihood becomes even greater that mis reporting is a real bias when proxy reporting ojFsmoking status is allowed. See also Lehnert et al (1984). ::y./-b y The Draft authors have proposed a model for correction o f mostly female exposure data for extrapolation to men, but again their approach is not at all convincing. The Cohort Studies Two cohort studies from the literature, diat of Hirayama (1984) and the study o f Garfnkel (1985) are considered in the Draft. A third cohort study, that o f Gillis et al (1984) is not con sidered and has generally received little attention from the research community even though it reports an RR o f 3.0 for males and 1.00 for females. O f course the reason for ignoring the Gillis study is the small sample size, six females or four males in the exposed groups. T he Hirayama study has received such wide-spread criticism that it is not necessary to dwell at length on the shortcomings o f the study. Layard (1989) lists some o f these as bias in the selection o f the study sample, misclassifcation o f lung cancer diagnosis, misclassifcation o f nonsmoking status and inappropriate statistical analysis o f the data. In spite o f the extensive criticism o f the Hirayama study, the Draft authors dwell extensively on its results. They principally make a point that there appears to be a dose-effect relationship, RR 1.42 for passive exposure to <14 cigarettes per day, RR =1.58 for 15-19 cigarettes per day, and RR 1.91 for passive exposure to >20 cigarettes per day. They do, however, use Garfinkel's data rather extensively. Since Garfnkel refused to assign statistical significance to the determined risk ratio o f 1.17 (c.i. 0.93-1.47) as being different from 1.0, then further consideration of the data is quite inappropriate. The Draft authors rather queerly report that although a risk ratio o f 1.0 is within Garfinkel's confidence limit that risk 51246 0019 1-5 Li' Ii.i i - 'jjiir' J C.'niiU^W"W ' \*ir ' ratios as high as 1.25 to 1.5 arc also in the range. They neglect to point out that risk ratios a good deal less than 1.0 are also in the confidence limit. It is also disturbing to see them report on page 3-43 that "....the American cohort study weakly indicates an increased cancer risk....." when the author himself is much more circumspect. It is surprising to me to see the following in the Draft. j "T he American cohort study appears to contain more statistical uncertainty than the Japanese study. Some o f the general factors contributing to uncertainty in study data are related to sample size, variability in the population sampled, sample design arid protocol, treatment o f massing or incomplete data, accuracy and reliability o f collecting and reporting data, arid methods o f staris- tical analysis." . All o f these purported shortcomings o f the Garfinkel cohort study have been variously ascribed to the Hariyama study. I | Dose-Effect Relationships . j T o establish a dose-effect relationship is the "Holy Grail" o f studies of environmental carcino gens,, and in the Draft there has been an attempt to examine these relationships in the casecontrol studies included in their analysis. Figure 3.4 shows a graphic representation o f the dose-response data which the Draft lists in Table 3-7. W ith only one possible, but questionable exception, (LAMT) there is no statistical significance attributable to the dose-effect relationship. Interestingly, the Draft authors describe those studies (such as K O O , HUM B, LAMT and LEE) in which there is no relationship or an inverse relationship as "variable". In the Draft a great deal is made of the dose-response relationship in the Hirayama (1984a) cohort study. In a further report on the Hirayama data (1984b), there are data that can be analyzed to show that the dose-response relationship issue is somewhat clouded if the age o f the nonsmoking spouse was stratified with respect to age at the time o f entry into the study. The dose-response relationship got weaker and weaker as the age at time o f entry increased. 51246 0020 1-6 = Dose-Effect M odeling The Draft authors consider "cigarette equivalent" dose-response relationship as a possible ap proach to estimating the risk from environmental tobacco smoke (ETS) in Chapter 4 and in Appendix D. They agree that there are significant limitations on the use o f the "cigarette equiva lent" approach. The most important o f these is that there is no useful way to make a comparison between exposure o f the mainstream smoker and the individual passively exposed to sicfestream smoke. It is interesting that after this insightful introduction to the problem, they state: " Legiti mate reservations not withstanding, virtually all analytic approaches bar some assumptions and weaknesses, and most contribute something to our understanding." . ] Repace and Lowrey (1985) start with the affirmed fact that there is an established risk ratio for non-smokers. T he model uses a probability weighted average for exposure to respirable suspended particulates (RSP) and a complicated model o f space air change and number of burning cigarettes per unit volume o f air space. The model predicts that exposure o f U.S. nonsmokers ranges from zero to 14+ mg/day with an "86% exposure probability". Th latter qualification indicates that 14% of adults are completely nonexposed. Their assumption that an average smoker consumes two cigarettes per hour is probably high, but there are more serious concerns than that. Repace and Lowrey base their estimate o f lung cancer deaths on the Seventh Day Adventist (SDA) data o f Phillips et al (1980a, 1980b). Based on the SDA data, Repace and Lowrey calculate that ETS caused 4700 lung cancer deaths among the 62.4 million nonsmokers over 35 years of age in 1980. Based on the assumption o f a linear dose-response relationship, Repace and Lowrey calculated that the ETS dose-response relationship was 5 lung cancer deaths per 100,000 per 1 mg tar per day. There are serious reservations both about the SDA study itself, as used by Repace and Lowrey (Gross, 1989), but there are other fundamental flaws in the model. The first and most impor tant is the assumption that there is a demonstrable RR for ETS exposure. Their independent estimate, based on a single-hit model and extrapolating from the RR for mainstream smoke, gives a dose-response estimate o f 0.6 lung cancer deaths per 100,000 per 1 mg tar per day. These authors suggest that the rate determined from the SDA model is the more reliable, for reasons that are not apparent. 1-7 4t fell&&&*- There are several other dose-effect (cigarette equivalent models) models which can be mentioned, but none are less subject to criticism than the Repace and Lowrey calculations. I mention Wells (1988), Robins et al (1986 N R C report appendix) and Vutuc (1984). The Draft authors mention the Moolgavkar et al (1989) model as a possible approach but it is not extensively developed by them. This model is a two-stage model which includes growth kinetics o f first-stage transformed cells. The model provides a reasonably good fit to the British | Doctor data. The low-dose portion o f the Moolgavkar curve from "doses" o f 5 cigarettes a day j down to 1 cigarette per day is described by a relationship where the relative risk is equal to 1 + i 0.3d, where d is the num ber o f cigarettes per day and the exposure is taken tb be from age 22.5 j years to present age 60 years. I f one accepts the downward extrapolation o f the British p o c to r data to regions well below that at which data exist, the formula just expressed predicts a relative risk of 1.3 for exposure to one cigarette per day. The important caveat for this extrapolation is I that one is extrapolating active smoking data and making application to a passive smoker. W hat : is the passive equivalent o f actively smoking one cigarette per day? McAughey et al (1989) j found, with a radioactive tracer method, that for volunteers exposed continuously to sidestream j smoke from two cigarettes, the tracheobronchial deposition o f particulates was equivalent to 0.006 cigarettes per day. The Moolgavkar model would then predict a risk ratio o f 1.002. From the N R C publication, dosimetry measurements were reported to lead to an estimate o f 0.0001 to 0.005 cigarettes per day. The upper limit from these numbers would be essentially the same as i from McAughey's study. In considering the overall criticisms of the "cigarette equivalent" models one can do little better i than quote Kilpatrick's rebuttal o f the Repace and Lowrey paper (Kilpatrick, 1986). H e points j out that the one-hit model o f these authors has a zero intercept term, implying that ETS is the sole cause o f lung cancer in nonsmokers. He also points out that Repace and Lowrey assume perfect mixing o f smokers and nonsmokers. T hat is to say, nonsmokers will not avoid smokers ! and vice versa. Furthermore, a general criticism o f all of these models is that the ambient mea- ! surements may have litde or no bearing on the amounts that actually reach the target tissue j (Doll, 1985). | Wells, in his modeling, is subject to many o f the criticisms levelled at Repacc and Lowery, but in addition he assumes marriage to a smoker is synonymous with exposure to ETS, which we know j 51246 <As> not to be the case for American spouses in particular. Furthermore, Wells combined morbidity and mortality and combines adjusted and unadjusted risk estimates without any discussion o f the methods o f adjustment when used. For females all cancers are included. For males only lung cancer is counted. All o f these faults leave little to consider of value in the Wells report. Biological Indicators o f Exposure It is relatively easy to relegate all other markers o f exposure to ETS other than nicotine and cotinine to the trash bin. Nicotine and cotinine are, however, nearly tobacco specific. O ne o f the better recent discussions o f these markers is given by Jarvis (1989). O ther in d icato iisu ch 'v as thiocyanate, carbon monoxide and polycyclic aromatic hydrocarbons are simply not spe< ~ V/' enough or sensitive enough to be useful is indicators o f exposure to ETS. T he one exception is that there are dietary sources o f nicotine that can confound measurements. Members o f the family solanacae, such as potatoes, bell peppers, egg plant and others do contain nicotine, and their intake can seriously distort measurements o f nicotine and cotinine in body fluids (Castro and Monji, 1986). In most cases, with reasonable care, the influence o f dietary intake ofthese sources can be controlled. W ithout careful attention to dietary intake serious misinterpretation o f data based on cotinine levels has been shown to be possible (Idle, 1990) 'l'iti''. v rw Nicotine has a half-life o f only 1-2 hours in blood, so that measurements o f this alkaloid are only useful for a brief period after exposure. Cotinine has a half-life o f 10-20 hours in blood, so it is more useful for long term exposure. For a number o f veiy significant reasons cotinine levels in body fluids must be considered to be no more than qualitative, or, at best, semiquantitative measures o f exposure to ETS. It has been well established that the distribution o f nicotine in mainstream smoke is quite different from that in sidestream smoke. In the former, nicotine is associated with the particulate fraction, while in the latter nicotine is in the volatile phase (Tang et al, 1988). O ne could not then, conceivably, try to relate the biological level o f cotinine in the nonsmoker with the number o f cigarettes smoked in his presence, unless extensive correlative experiments were done to establish a useful association o f measurements. The most important general limitation with the use o f cotinine measurements is that it cannot be used as a surrogate for constituents o f ETS other than nicotine w ithout careful study and correlation. The measurements cannot be validly used to estimate Min irfoe CM 0023 1-9 "cigarette equivalents" for risk estimation. Such an estimate, though, has indeed been made by Russel etal (1986). Sepkopvic et al(1986) reported on the rate o f cotinine and nicotine metabolism in both smokers and nonsmokers. This paper states that the rate o f nicotine and cotinine metabolism in smokers and in nonsmokers is greatly different. I f this were not the case it would be very surprising because the detoxification o f most chemicals that do not normally exist in the body occurs via j inducible enzyme systems, the activity level o f which is controlled by the level o f exposure. This fact alone rules out the determination o f cotinine in body fluids as a reliable marker for exposure .to E T S . . ' ; ". , ; ' ' ! There is no methodological barrier to the use o f cotinine concentrations as an indicator o f ETS exposure. The concentrations are well within the present technology, but because they are so low, careful methodology and intercomparisons among laboratories are essential. Interfering substances exist and must be controlled for j Respiratory Disease and Respiratory Function in Children and ETS Section 5 of the Draft deals with the relationship between ETS and respiratory disorders in children. This chapter is, in my opinion, the most superficially written and poorly interpreted part of what is, overall, already a disgracefully nonscientific document. The key document for these effects is the 1986 N R C report on ETS, which draws conclusions j that can be roughly paraphrased as follows: 1. Parental smoking, particularly by the mother, increases the risk of respiratory symptoms and illness in young children. 2. Parental smoking may be associated with small decreases in pulmonary function and may impair pulmonary development in small children. Outside of the Draft there is another good review o f the literature by Witorsch (1989) and Witorsch and W itorsch (1989) I particularly commend to you the extensive bibliography. | 51246 0024 1-10 Also in a relatively recent review (also cited in the Draft), Rubin and Darmus (1988) tentatively I concluded that there might be an association between parental smoking and respiratory disease in their children. They, however, rated each o f 30 studies on the basis o f seven individual criteria of study quality, including among others, data collection, estimates of smoke exposure, patho logical criteria, etc. T he authors found serious flaws of design and conduct in nearly all o f the studies. '| The Witorschs are swayed that there is some substance in the possibility that there is indeed an effect on pulmonary disease in children under five years o f age, but that for older children there | is no such relationship. They point out, however, that all o f the confounding variables that affect j the outcome o f epidemiological studies o f lung cancer and ETS are just as present and just as j important for lung disease in children studies. O f these, the ones to which they assign the highest I significance are socioeconomic status and occupations o f the parents. Harlap and Davis (1974) in a quite old study already reported that, while they found an association o f parental smoking with increased respiratory illness in infants, there was also an association with increased incidence of hospitalization of infants due to injury and poisonings. O f course the latter cannot be attrib- i uted causally to ETS, and is more probably an indicator o f socioeconomic status. W itorsch and W itorsch are convinced that no case can be made for the effect o f parental smok ing on pulmonary function, even for children under five. It has been noted by several authors and commented on in the discussion attached to Witorsch (1989) by Robert Brown, that one cannot deconfound the effects seen in the infant as the result o f ETS on the child from the effects that were established in utero as the result o f maternal Expressing a personal opinion, it is strange to me that, even though the evidence for infant pulmonary disease and ETS is somewhat more convincing than the adult lung cancer associa tion, that the EPA has done such a trivial job analyzing the data and coming to conclusions. Maybe adult lung cancer sells better. 51246 0025 1-11 Conclusions and Recommendations I find the study seriously flawed as a basis for federal regulatory action. The Draft has selectively reviewed the literature in such a fashion that significant bias towards accepting the evidence for a ! risk ratio greater than 1.0 for the association o f adult lung cancer and ETS is introduced. The analysis o f the effects of parental smoking on childhood lung disease and lung function again draws conclusions contrary to the statistical evidence and warps the data to fit the pattern that they seem to have chosen. >. . In the book Environmental Tobacco Smoke, N athan Mantel makes a number o f incisive com- i ments that should be read in the original by the members o f this committee. Firstly, he appropri ately points out that none o f the cohort studies are truly cohort studies in the same sense as the British Doctors Cohort. All o f the presently discussed ETS cohort studies are "record linkage" studies, subject to the usual difficulties with this kind o f study. H e suggests that the wrong study cohort is being assembled in each of the present reports. H e proposes a study cohort of non smoking wives who are then differentiated as being married to smokers or nonsmokers. Finally, Mantel again (1987) suggests that epidemiology is "too blunt a tool" to distinguish a relative risk o f less than 2. The evidence from the radiation literature is all in support o f this view. To appear soon is a study of fallout radiation and leukemia in which the study population o f case-control design was the whole Mormon population of Utah, and it was not possible to statistically confirm a relative risk of about 1.2. References Castro, A. and Monji, N. (1986) Dietary nicotine and its significance in studies on tobacco smoking. Biochem. Arch. 2,91-97. Chalmers. T.C ., Hewett, P., Reitman, D. and Sacks, H.S. (1989) Selection and evaluation of empirical research in technology assessment. Int. J. o f Technology Assessment in Health Care. 5, 521-536. Doll, R (1985) Occupational Cancer, A hazard for epidemiologists. Int. J. Epidemiol. 14, 22-31. ; 51246 026 1-12 i Friedman, G .D., Petitti, D,B. and Bawol, R.D. (1983) Prevalence and correlates o f passive smoking. American J. of Public Health, 73,401-405. Garfinkel, L. (1981) Time trends in lung cancer mortality among non-smokers and a note on passive smoking. J. N at. Cancer Institute, 6,1061-1066. Gross, A J. (1989) Risk assessments relating to environmental tobacco smoke, in Environmental Tobacco Smoke, pp 293-302. D.J. Echibon and J.M.Wu.eds.Lexington Books, Lexington, Mass. ! Harlap, S. and Davis, A.M. (1974) Infant admissions to hospital and maternal smoking. Lancet, ! 1(7857)529-532. 'V .- V;/"'- ' j Hirayama, T. (1984a) Cancer mortality xn non-smoking women with smoking husbands based I on laige scale cohort study in Japan. Prevent. Medicine 13,680-690. Hirayama, T. (1984b) Lung cancer in Japan: Effects o f nutrition and passive smoking, in Lung Cancer, Causes and Prevention, Mizwell and Correa, eds. Vcrlag-Chemie International, New York. Idle, J.R. (1990) Titrating exposure to tobacco smoke using cotinine--A minefield of misunder standing. J. Clin Epidemiol. 43, 313-317. Jarvis, M.J. (1989) Application of biochemical intake markers to passive smoking measurement and risk estimation. M utation Res. 222, 101-110. Kilpatrick, S.J. (1986) Letter to the Editor. Environment Intern. 12, 29-31. Koo, L.C. (1989) Environmental tobacco smoke and lung cancer: Is it smoke or the diet? in Present and Future o fIndoor A ir Quality. Bieva et al, eds. Excerpta Medica, Amsterdam. Laird, N.M. and Mosteller, F. (1990) Some statistical methods for combining experimental results. Int. J. of Technology Assessment in Health Care. 6, 5-30. Layard, M.W. (1989) Environmental tobacco smoke and cancer: The epidemiological evidence.in Environmental Tobacco Smoke, pp 100-115. D.J. Echibon and J.M.Wu.eds.Lexington Books, Lexington, Mass. 51246 027 Lee, P.N. (1988) An alternative explanation for the increased risk of lung cancer in non-smokers married to smokers, in, Indoor and Am bient A ir Quality, pp 149-158 eds, R. Perry and P.W. j Kirk, Selper Ltd, London. Lehnert, G., Garfinkel, L., Hirayama, T., Schmht, D. berla, K., Wynder, E.L., and Lee, P.(1984) Roundtable discussion. Prev. Med. 13,730-746. . Letzel, H ., Blumner, E. and Uberla (1988) M eta analysis on passive smoking and lung cancer: effects o f study selection and misclassification o f exposure. Environmental Technology Letters, 9,491-400. Mantel, N. (1987) Lung cancer and passive smoking. Brit. Med. J. 294 p 440. Mantel, N ., and Haenszel.W . (1959) Statistical analysis o f data from retrospective studies ofdisease.',}.Nat. CancerInstitute^22r719*748.v^ ! . P - . ^ . ; / ^! Repace, J.R. and Lowrey, A.H. (1985). A quantitative estimate o f nonsmokers* lung cancer risk from passive smoking. Envir. Internat., 2, 3-22 Tang, H ., Richards, G., Gunther, K. et al (1988) Determination of gas phase nicotine and 3- ! ethenylpyridine, and particulate phase nicotine in environmental tobacco smoke with the collec- j tion bed capillary gas chromatography system. J. High Resol. Chrom. Commun. 11,775-782. Vsutuc, C. (1984) Quantitative aspects of passive smoking and lung cancer. Prev. Med. 13, 698-704. Wells, A.J. (1988) An estimate o f adult mortality in the United States from Passive Smoking. Environment Internat. 14, 249-265. Witorsch, R.J. (1989) Parental smoking and respiratory health and pulmonary function in children: A review o f the literature and suggestions for future research, in Environmental Tobacco Smoke, pp 205-226. D.J. Echibon and J.M.Wu.eds.Lcxington Books, Lexington, Mass. Witorsch, R.J. and Witorsch, P. (1989) A critical analysis o f the relationship between parental smoking and pulmonary performance in children. Das ffentliche Gesundheitswesen. 51,78-83. Yusuf, S., Peto, S.R., Lewis, J., Collins, R., and Sleight, P. (1985) Beta blockade during and after myocardial infarction: an overview o f the randomized trials. J. o f Progressive Cardiovascular Diseases. 27, 335-371. ui 246 0028 1-14 .v . 4 ijVWaitdSii *: A Critique of EPA External Review Draft 600/6-90/006A, M ay 1990 Health Effects of Passive Smoking: Assessment o f Lung C a n ce r in Adults, and Respiratory Disorders in Children prepared by M ichael G. B issell, M .D ., P h .D ., M .P.H. City of Hope National Medical Center Duarte, CA 91010-0269 51246 0029 s * - i : .'l ---------r ,, J.'; I. The Logic o f the EPA Draft w ith Regard to Use o f Cotinine Measurements. T he issue of the validity and applicability o f body fluid assays o f nicotine metabolites is fundamental to assessing the validity of the EPA Draft's conclusions. Its approach to the estimation of lung caner risk from passive smoking involves the following steps: 1. Estimating the percentage of the population exposed to environmental tobacco smoke (ETS) based on these measurements. 2. Combining the results o f case-control and cohort studies to obtain a summary estimate o f relative risk (RR) o f developing lung cancer for nonsmoking spouses married to smokers. 3. Correcting the summary RR estimate for smoker misclassification based in part on evidence derived from cotinine measurements. (RRM) 4. Further correcting RRM for the effect o f general background exposure to ETS based on cotinine measurements. (RRB) 5. Combining RRB with the estimate o f the population exposed to ETS in (1.) to obtain an estimate o f the population attributable risk (PAR) for lung cancer in never-smoking females exposed to ETS. T he final section o f the EPA Draft deals with respiratory (and related) disorders in children attributable to ETS. In this section, all the estimates o f exposure are based on cotinine measure ments, as are two o f the studies of relative risk for specific conditions. T he EPA Draft draws the following disti nctions in its use o f cotinine measurements: pl-5: "Some mathematical modeling is required to adjust for expected bias from self-reported misclassification status and to account for ETS exposure from sources other than spousal smoking. T he approach, however, does not rely on a mathematical model o f doseresponse or low dose extrapolation o f observations obtained at extraordinarily high exposure levels". 2-1 51246 0030 p4-28: "There is an important distinction between the use o f cotinine as a surrogate dose for ETS to estimate lung cancer risk from background exposure and its use in the cigaretteequivalents approach. In the latter, the contention centers around the assumption that cotinine (or anything else such as RSP) is an equivalent dose surrogate for both pas$ive and active smoking, i.e., that equivalent uptake in passive and active smoking implies equivalent carcinogenic risk". While the approach adopted in the EPA Draft may or may not require assumptions about equivalent effects o f active and passive smoking or risk, it does depend on two other basic . implicit assumptions, namely: ' ~ V^ ' >` i I 1. T hat the only cause o f body fluid, elevations o f nicotine or cotinine in human subjects is tobacco smoke (environmental or mainstream). In other words, that the biological markers, (clinical assays o f nicotine/cotinine) are 100% i sensitive and specific as indicators o f ETS exposure and show no false positives or false negatives due to analytical interferences o f any kind. i i 2. The background exposure to ETS causes lung cancer and is the only cause o f this disease in the nonsmoking population studied; therefore correction o f RR estimates for this background exposure, based on cotinine measurements, is required. It is not immediately obvious, nor is it anywhere explained in the EPA Draft, exactly how this assumption differs in practical terms from "the assumption that cotinine (or anything else such as RSP) is an equivalent dose surrogate for both passive and active smoking". j i II. Examination o f Assumptions A. Assays o f Nicotine and Nicotin e Metabolites. ! O ur understanding o f nicotine metabolism in man is not complete. The elucidation o f the full extent o f these pathways depends upon the isolation and identification o f all o f the metabolites involved. No fewer than twelve such intermediates have been identified so far in the rat by 51246 0031 2-2 radiometric high performance liquid chromatography (HPLC), and at least eighteen have been postulated. (Kyerematen, et al, 1987). Cotinine is one o f these compounds, and has a longer half-life in body fluids than the parent compound. It was for many years thought to be the principal metabolite, but more recent work .j has shown this not to be the case. Nicotine is also converted to nornicotine and to nicotine oxides, and cotinine itself is not an end product, but is further converted to any one o f at least four other compounds (Kyerematen, et al, 1987). O ne o f these, known as "metabolite 5" (Parvainen and Barlow, 1988) and tentatively identified as trans-3-hydroxycotinine (Nurath, j et al, 198) is actually the major nicotine metabolite in smokers (Neurath and Pein, 1987). This i is important in the present context not only as an indication o f one o f the numerous sdurces o f i biological variability in the cotinine assay, but also because this compound shows a 30% cross reaction with the polyclonal anticotinine antibody which is the basis o f the ELISA assay for cotinine (Schepers and Walk, 1988). This enzyme system is not only subject to genetic variabil ity but, more importantly, is known to be very markedly induceable by drugs and other ingested substances (e.g., phnobarbital, Rudell, et al, 1987). O ne of these drugs is nicotine itself, which causes increased microsomal clearance o f drugs like theophylline (Matsunge, et al, 1989) as well as nicotine and cotinine (Sepkovic, et al, 1986). Two other sources o f false positives have major implications for the interpretation o f cotinine assays as surrogates for ETS exposure. The first o f these is caffeine, which is an exogenous source o f analytical interference since it co elutes with cotinine on several different reversed-phase HPLC assays, causing spurious elevations (Thuan, et al, 1989). This problem occurs in assays performed by the methods o f Machecek and Jiang, (1986), Hariharan, et al (1988), Watson (1977), Maskarinec, et al (1978), Kyerematen, I et al, (1982) and Hortsmann (1985). Even more fundamentally problematic, however, is the fact, evidenced by the efficacy o f snuff and Nicorctte chewing gum, that nicotine is quite efficiently assimilated orally (Russell, et al, 1980). This is because of the recent discovery that tobacco and tobacco smoke arc not the only common environmental sources of elevations in body fluid nicotine/cotinine levels. Common dietary sources are now known to exist. A variety of foods, (potatoes, eggplants, green peppers, 2-3 51246 0032 green tea, and certain brands o f instant tea) contain sufficiendy high concentrations o f dietary nicotine to cause significant false positive cotinine elevations and resulting misclassification of ETS exposure (Idle, 1990). It appears that the nicotine is endogenous to the plants, but an additional potential source is absorbed insecticide nicotine on vegetables from certain countries. This may potentially also be a source o f dietary nornicotine and cotinine itself (Idle, 1990). False positive elevations o f cotinine assays by dietaty nicotine provide an alternative explanation ' for certain problematic finding in the ETS exposure literature. (Matsukura, et al, 1984; Pettenger, 1985; Adlkofer, et al, 1985). It will likely necessitate a downward revision in the estimation of the percentage o f the population exposed to ETS based on hicptihe/cotmme assays, and in the estimation o f the rate o f misclassification o f smoking status based on' these assays. B . Problems with Nicotine and Nicotine Metabolites as Surrogates o f Potential ETS Carcinogen Exposure. As stated earlier, it is hard to understand the rationale for the correction of relative risk for "background" ETS exposure without implicitly assuming the relationship that is supposed to be demonstrated, namely that ETS causes lung cancer. This is circular reasoning. The error is only compounded by performing this "background" correction based only on cotinine mea surements and then attempting to deny that. (EPA Draft, p. 4-28). Cotinine measurements are, in fact, being used as dose surrogates in so doing. To quote directly from the EPA Draft; p 4-15: j i i "A difficulty in assessing this approach lies in evaluating the assumption that apparent differences | between passive and active smoking are negligible or have cross-effects that cancel. For example, MS and SS differ in the relative composition o f carcinogens identified in tobacco smoke and in their physicochemical properties in general. The lung and systematic distribution o f chemical agents common to MS and SS are affected by their relative distribution between the vapor and particle phases, which differs between MS and SS as it ages. Passive and active smoking also differ in characteristics o f intake-intermittent (possibly deep) puffing in contrast to normal I (shallow) inhalation. T o help illuminate relationships and identify parameters where additional i information would be helpful on this topic, a mathematical model for comparison of dosimetry of passive and active smoking was constructed as a basis for further study (Appendix C). 51246 0033 2-4 I Several authors have taken issue with the validity o f the cigarette-equivalents approach. For example, Hoffmann et al (1989), in discussing the longer clearance times o f cotinine from passive smokers than from active smokers, concludes `the differences in the elimination time of cotinine from urine preclude a direct extrapolation o f cigarette-equivalents to smoke uptake by involuntary smokers'. A recent consensus report o f an IARC panel o f experts (Saracci, 1989, p.3) states that `lacking knowledge o f which substances are responsible for the well established carci nogenic effect o f MS, it is impossible to accurately gauge the degree o f its similarity to ETS in respect to carcinogenic potential'. T he U.S. SG report devotes a three page section to the con cept o f cigarette-equivalents, quantitatively demonstrating how they can vary as a measure o f exposure (U.S. SG, 1986). It concludes with `these limitations make c ^ p o la tio n fro ^ atmo spheric measures to cigarette-equivalents units o f disease risk complex arid potentially meaning less process'." III. Conclusions/Recommendations The papers upon which the EPA Draft's estimates o f population exposed to ETS, correction for smoker misclassification, and correction for background exposure to ETS are based, should be carefully re-examined. All of these papers which depend upon assays o f nicotine metabolites in body fluids for their conclusions can be meaningfully used for these purposes only if careful ' dietary histories have been obtained. T he analytical results must either have been or be capable of being corrected for the dietary sources o false positive analytical results to avoid a systematic bias toward overestimating ETS exposure in all cases. j References 1. Adlkofer, F. Scherer, G, Hees, U. Passive smoking. (1985). N . Eng. J. Med., 312,719-720. 2. Hariharan, M. Vannoord, T , Greden, JF, (1988). An HPLC method for routine simulta neous determination of nicotine and cotinine in plasma. Clin. Chem., 34,724-729. 3. Hortsmann, M. (1985). Simple H PLC method for rapid determination o f nicotine and cotinine in urine. J. Chromatog., 344, 391-396. ; to *CT*i <<ss>> 2-5 i 4. Idle, JR, (1990). Titrating exposure to tobacco smoke using cotinine-a minefield o f misun derstandings, J. Clin. Epidemiol., 43,313-317. 5. Kycrematen, G A, Damiano, M D , Dworchik, B H , Vcsell, E S, (1982). Smoking-induced changes in nicotine deposition: application o f new HPLC assay for nicotine and its metabolites. Clin. Pharmacol. Ther. 32, 379-382. j 6 . Kyerematen, G A, Taylor, L H , de Bethizy, J D , Vesdl, E S, (1987). Radiometric HPLC assay for nicotine and twelve o f its metabolites. J. Chromatog. 419,191-203. 7. Machecek, D A, Jiang, N S, (1986). Quantification o f cotinine in plasma and saliva by liquid chromatography. Clin. Chem. 32,-379-382. 1 8 . Maskarinec, M P, Harvey, R W , Caton, J E, (1978). A novel method for the isolation and quantification analysis o f nicotine. J. Anal. Toxicol., 2,124-126. ! 9. Matsakura, S, Tom hiku, T , Norikazu, E, et al, (1984). Effects of environmental tobacco smoke on urinary cotinine excretion in nonsmokers N. Eng. J. Med., 311. 828-832. 10. Matsunga, S K, Plezia, P M , Karol, M D , Katz, M .D. Camilli, A E, Benowitz, N L, (1989). Effects of passive smoking on theophylline clearance. Clin. Pharmacol. Ther., 46, 399-407. 1 1 . Nakayama, H , Fujihara, S, Nakashirna, T , Kurogochi, Y, (1987). Formation o f two major nicotine metabolites in livers o f guinea pigs. Biochem. Pharmacol., 36,4313-4317, 1 2 . Neurath, G B, Dunger, M. Krenz, O , O rth, D, Pein, F G, (1988). Trans-3'hydroxycotinine-a main metabolite in smokers. Klin. Wochenschr., 6 6 , (Suppl. 11), 2-4. ; 13. Neurath, G B, Pein F G, (1987). Gas chromatographic determination o f trans-3'hydroxycotinine, major metabolite o f nicotine in smokers. J. Chromatog., 415,400-406. 14. Parvainen, M T . Barlow, R D , (1988). Assessment o f exposure to environmental tobacco smoke using an HPLC method for the simultaneous determination o f nicotine and two o f its metabolites in urine. J. Chromatog., 431, 216-221. 15. Pittenger, D J, (1985). Letter to the editor. N . Eng. J. Med., 312,720. 51246 0035 2-6 16. Rudell, U, Foth, H , Kahl, G F, (1987). Eight-fold induction of nicotine elimination in perfussed rat liver by pretreatment with phnobarbital. Biochem. Biophys. Res. Comm., 1 4 8 ,1 9 2 - 1 9 8 . 17. Russell, M A H , Raw, M , Jarvis, M , (1980). Clinical use o f nicotine chewing-gum. sBrit. Med. *J., 280,1599-1602. 18. Schepers, G, W alk R-A, (1988). Cotinine determination o f immunoassays may be influenced by other nicotine metabolites. Arch. Toxicol., 62,395-397. j | 19. Sepkovic, D W , Haley, N J, Hoffmann, D , (1986). Elimination from the body o f fobacco products by smokers and passive smokers. J. Am. Med. Assoc., 256, 863. ' 20. Thuan, N TL, Migueres, M L, Roche, D , Roussel, G, Mahuzier, G, Chretien, J, Ekindjian, O G, (1989). Elimination o f caffeine interference in HPLC determination o f urinary nicotine and cotinine. Clin. Chem., 35,1456-1459. \ I j 21. Watson, I D , (1977). Rapid analysis o f nicotine and cotinine in urine o f smokers by isocratic high-performance liquid chromatography. J. Chromatog., 143,203-206. ! 51246 0036 2-7 A Critique of EPA Externai Review Draft 600/6-90/006A, May 1990 Health Effects of Passive Smoking: Assessment of Lung C a n c e r in Adults an d Respiratory Disorders in Children prepared by M artin J. C line, M .D ., UCLA Department of Medicine Division of Hematology-Oncology Los Angeles, CA 90024-1678* *presendy at: Ludwig Institute for Cancer Research, London W1P 8BT, Great Britain 51246 0037 Introduction In the past decade, a number o f epidemiologic studies have assessed the relationship between passive smoking and the incidence o f lung cancer (1-17). Fewer studies have addressed the question of a possible relationship between environmental tobacco smoke (ETS) and respiratory infection in infants and children (see 18-20 as examples). The findings in the epidemiologic studies o f ETS and lung cancer have rang;ed from no detectable increase in lung cancer risk (8 ,15) to a modest (less than two-fold) increase (2 -5); however, many studies have reported increases which are o f small magnitude and are statistically insignificant (1,6,7). In 1986, W ald et al. (16) performed a meta analysis o f the data then available and concluded that thet< was a slightly increased risk o f developing lung cancer associated with exposure to ETS. Recently, Janerich et al. (17) reported a case-control study o f 191 lung cancer patients and concluded that only that group o f nonsmokers who were exposed to ETS in childhood and adolescence had an increased risk o f lung cancer (odds ratio 2.07, 1.16-3.68). The authors o f the EPA Draft reviewed the published data with regard to ETS and lung cancer (excluding the Janerich report) and ETS and pulmonary infection. They concluded that ETS is responsible for a large number of lung cancer deaths among nonsmokers, as well as an increased incidence of pulmonary infection among infants and children. Analysis o f EPA Draft The EPA Draft is not a scientific document; rather it is a selective summary and a selective analysis o f data available in the literature. In essence, it is an interpretation or re-interpretation o f the data o f others. Consequently, it may be flawed either because the original data were inad equate or inaccurate, or because the authors o f the Draft were inaccurate in their interpretation o f the data o f others. The Draft relies heavily on a meta analysis o f a series of epidemiologic studies o f ETS and lung cancer risk. It dismisses as invalid the dosimetric approach to analysis which is called the "ciga rette-equivalent" approach. Two reasons are given for this dismissal: 1) The metabolism o f nicotine is different in smokers and nonsmokers. 2) The carcinogens are different in mainstream and sidestream smoke. The Draft 3-1 51246 0038 Z E iia does not consider animal models. This is reasonable in view o f the fact that there are no valid animal models of human lung cancer. Epidemiologic Studies and Meta Analysis After reviewing the D raft and listening to my colleagues, I have concluded that the interpreta tions and conclusions o f the authors o f th e Draft are seriously flawed. T he major defects to be found in the analysis and interpretation o f the data in the literature are the following: 1. The effect o f non-reporting o f negative studies are mentioned but not really considered in the Draft (see pages 3-33); i.c. publication and reporting bias are acknowl ; edged but not taken into account. i 2 . Only females are considered, and the statistically non-significant data for males are dismissed. 3. Confounding and possibly confounding factors which may invalidate the perceived association between ETS and lung cancer are not considered. These may include radon exposure, industrial (including diesel exhaust), exposure and perhaps other factors related to the socioeconomic status of the populations studied. 4. Incorrect diagnosis o f lung cancer where this diagnosis was derived only from death certificates and not confirmed by independent and blinded histologic analysis. 5. Variability in the criteria for defining ETS exposure which should render invalid the pooling o f data from several o f the independent studies. 6 . The failure to consider or explain the lack o f a clear dose-response relationship between ETS and lung cancer incidence (see studies by Koo and Garfinkel as examples). 7. The misidentification o f smoking status by questioning surrogates. 8 . The use o f cotinine as a measure o f defining "truly nonsmokers," whereas it is dismissed as invalid for a cigarette equivalent analysis. 9. The effects o f ETS exposure outside the home are not considered. This is part o f the larger problem o f defining the baseline for nonsmokers and of the extent o f ETS exposure. 10. Studies acknowledged to be o f poor quality (Hirayama et al.) are given emphasis, whereas good-quality studies (Garfinkel) are underemphasized. 11. Differences between Asian and American studies in the quantitative and qualitative aspects o f ETS exposure are not considered. 3--2 51246 1 2 . The significance o f a relative risk ratio o f about 1.2 is not discussed in a critical and unbiased fashion. Cigarettc-Equivaleiit Approach The details of this dosimetric approach are so well known that I shall not discuss them here but basically it has the following elements: 1) W hat is the dose-response curve for lung cancer incidence vs. numbers o f cigarettes smoked per unit time for smokers? 2) Can the numbers of cigarettes per unit time be calculated for nonsmokers exposed to ETS? 3) Can one calculate a risk for nonsmokers based on this cigarette equivalent calculation? This approach has certain requirements: 1) Dosimetry for smokers must be well defined even at low levels o f exposure. 2) O ne must have a valid measurement which would allow one to calculate the exposure o f nonsmokers. 3) The putative carcinogenic materials to which smokers and nonsmokers are exposed must be similar in biologic activity. Let us examine each of these requirements and see whether a cigarette equivalent approach can be used to assess potential risks of ETS. 1. Dose-response curve for smokers. This subject has been studied extensively for more than two decades and there is an abundant literature on the relationship between cigarette consumption in smokers and lung cancer inci dence (see Moolgavkar et al. ref. 21 as a relatively recent summary). 2. The measurement o f tobacco smoke exposure in nonsmokers. Cotinine is the best biologic marker o f exposure to the nicotine in tobacco smoke despite certain limitations: 1) Nicotine can be ingested with certain vegetables; and 2) The metabolism of nicotine/cotinine is different in smokers and nonsmokers. Both these limitations, if recognized and taken into account, will allow a reasonably accurate measurement o f tobacco exposure in nonsmokers if the analytic study is properly constructed. Indeed, both factors will tend to result in an overestimation of the extent of ETS exposure in nonsmokers. N ote that the EPA Draft sometimes uses this approach to define "truly nonsmokers." 3-3 iHn* to er> S> <4S*> S> I- 3. Potential carcinogens in ETS. The concentrations o f a number o f known carcinogenic compounds differ in mainstream smoke, sidestream smoke and in ETS. Furthermore, they may differ in ETS as a function o f time and environmental factors. Nevertheless, the concentration o f a number o f carcinogenic compounds in mainstream smoke and ETS is known with precision for a variety o f environments and is generally if not always o f lower concentration in ETS than in mainstream smoke. W hile I do not have such data at my fingertips, I believe that such data are available. In view o f these considerations I believe that it is possible to determine a m axim alvalue o f cigarette exposure for nonsmokers, and that a cigarette-equivalent analysis is reasonable even if > imperfect. Indeed, such an analysis would be considerably more precise than the crude tools j available to the epidemiologist. M y rough calculation is that a nonsmoker in a household where j 2 0 cigarettes per day were smoked, would be exposed to the equivalent o f less than one cigarette per day, i.e., a level which few people would argue is hazardous. Summary and Conclusions The EPA Draft is a weak document. It is seriously flawed in its analysis and interpretation o f the epidemiologic studies o f ETS and lung cancer. Although I have not discussed the details, the same defects can be found in the analysis o f epidemiologic studies of ETS and respiratory infec tions in infants and children. T he report appears to be biased in its conclusion although one cannot say whether this is an intentional bias or due to a lack o f expertise on the part o f the authors. I do not believe that there is sufficient evidence for classifying ETS as a known human carcinogen or even a "probable" carcinogen. R eferen ces 1. Garfinkel, L.: Tim e trends in lung cancer mortality among nonsmokers and a note on passive smoking. J. Nat. Can. Inst. 6 6 : 1061-6, 1981. 2. Hirayama, T.: Cancer mortality in nonsmoking women with smoking husbands based on a large-scale cohort study in Japan. Prev. Med 13: 680-84,1984. 3. Trichopoulos, D. et al: Lung cancer and passive smoking: conclusion o f a Greek study. Lancet 2: 677-80, 1983. 51246 0041 5-4 4. Correa, P. et al: Passive smoking and lung cancer. Lancet 2: 595-7,1983. 5. Garfinkel, L. et al: Involuntaiy smoking and lung cancer: a case-control study. J. Nat. Can. Inst. 75:463-9,1985. 6 . Akiba, S. et al: Passive smoking and lung cancer among Japanese women. Cancer Res. 46:4804-7,1986. 7 . Dalager, N A et al: T he relation o f passive smoking to lung cancer, ibid 46:4808-11,1986. 8 . Kabat, G .C. and W ynder, E.L.: Lung cancer in nonsmokers. Cancer 53:1214-21,1984. 9 . Sandler, D.P. et al: Cumulative effects o f lifetime passive smoking on cancer risk. Lancet 1:312-5,1985. 10. Sandler, D.P. et al: Cancer risk in adulthood from early life exposure to parents' smoking. Am. J. Publ. Health, 75:487-92,1985 1 1 . Sandler, D.P. et al: Passive smoking; in adulthood and cancer risk. Am. J. Epidem. 121: 37-48,1985. . 12. Pershagen, G. et al: Passive smoking and lung cancer in Swedish women, ibid: 125: 17-24,1987. 13. Koo, L.C. et al: Measurements o f passive smoking and estimates o f lung cancer risk among nonsmoking Chinese females. Int. J. Cancer 39:162-9,1987. 14. Humble, C.G. et al: Marriage to a smoker and lung cancer risk. Am. J. Publ. Health, 77: 598-602, 1987. 15. Chan, W .C.: Lung cancer in nonsmokers in H ong Kong. In: Grundmann et al (eds.), Geographical Pathology in Cancer Epidemiology, Fischer Verlag pp. 199-202,1982. 16. W ald, N.J. et al: Does breathing other people's tobacco smoke cause lung cancer? Brit. Med. J. 293: 1217-23,1986. 17. Janerich. D .T. et al: Lung cancer and exposure to tobacco smoke in the household. NEJM 323: 632-6,1990. 51246 0042 J 3-5 18. Evans, D. et al: The impact o f passive smoking on emergency room visits o f urban children with asthma. Am. Rev. Res. Dis 135: 567-72,1987. 19. Fergusson, D.M . et al: Parental smoking and lower respiratory illness in the first three years o f life. Epidemiol, and Commun. Health 35: 180-4,1981. 20. Ferris, B.G. et al: Effects o f passive smoking on health o f children. Environ. Health Perspect. 62: 289-95,1985. 2 1 . Moolgavkar, S.H. et al: Cigarette smoking and lung cancer: reanalysis o f the British doctors data. J. Natl. Cancer Inst. 81:415-20,1989. ... ' 51246 0043 A Critique of EPA External Review Draft 600/6-90/006A, M ay 1990 Health Effects of Passive Smoking: Assessment of Lung C a n ce r in Adults and Respiratory Disorders in Children prepared by Alvan R. Feinstein, M .D . Ezra Taft Benson Agriculture and Food Institute Yale University School of Medicine New Haven, CT 06510 * 51246 004 5 t s M in . U n .. M y comments refer to four main features o f the EPA document. They are: (1) its status as an editorial review o f existing work rather then a presentation o f new scientific evidence; (2 ) the acceptance, for statistical "meta-analysis", o f data whose distortions and low scientific quality are ignored; (3 ) the absence o f appropriate attention to problems o f recall bias and detection bias; (4) the overt or subtle prejudice with which the authors have selected "evi dence" for citation. 1. Editorial Review ; T he EPA Draft is not a "scientific document" containing new evidence. Instead, the Draft is an editorial review o f existing evidence and o f existing reviews. Accordingly, the EPA Draft cannot be appraised with any o f the standards customarily applied to a "scientific report". As a review and set o f editorial judgments, the EPA Draft must be evaluated for its attention to the scientific quality of the "news" from which the editorial review is derived, and for the quality and fairness with which the editorial is constructed. 2. Statistical Meta-Analyses In contrast to previous editorial reviews (by the Sutgeon General and the NRC), the EPA Draft also presents a statistical meta-analysis o f the existing data. For this purpose, the "news" con tained in existing epidemiologic studies has been accepted at its face value, and then subjected to various forms o f statistical combination and calculation. The process o f statistically combining and re-calculating published data was carried out without any attention to three sets o f generally accepted criteria that have been established as conditions sine qua non for the scientific integrity o f a meta-analysis. The three sets o f criteria are as follows: a. Formal protocol: T o avoid the biases that can arise when investigators selectively choose the publications to be included, a careful, formal protocol must be established before the meta analysis is done. The protocol should indicate the exact procedures to be used in searching the literature, in selecting the studies to be included in the analysis, and in deciding which ones will be excluded. 4-1 51246 084 U1 i b. Precautions to avoid "publication bias**: Because "publication bias" and the "file-drawer phenomenon" may lead to a substantial overemphasis on "positive results" in the published literature, the conductors o f a meta-analysis should also institute a thorough search to find unpublished or otherwise unpublicized evidence o f "negative" results. W ithout such a search, a meta-analysis based on published literature will inevitably be biased toward "positive" results when the data are pooled. c. Random ized-trial evidence: To avoid the diverse problems o f bias that can arise when cause- effect relationships are studied in the absence o f randomized assignment of the compared agents, the evidence pooled in a meta-analysis should preferably come mom randomized trials, vfKr-. Although not everyone agrees about the scientific conclusions that can be drawn from a statisti cal meta-analysis, there is universal agreement that the foregoing three criteria should be fulfilled for the analysis to have scientific integrity and credibility. The EPA meta-analysis did not fulfill any of these three criteria. There was no formal protocol, no attention to publication bias, and, despite the fact that none o f the included studies came from a randomized trial, no appropriate attention to the inevitable biases that must be considered and appropriately adjusted (if possible) when the basic information comes from non-randomized epidemiologic research. 3. Recall and Detection Bias The EPA Draft gives inadequate attention to two o f the major biases to be expected in observa tional cohort and case-control studies. O ne o f them, recall bias about exposure to ETS, is gener ally dismissed by the EPA Draft authors as making odds ratios tend to the null value (i.e., 1). The authors ignore the high likelihood that for the cited clinical conditions, the patterns o f recall bias will almost always go in the same direction, not in random cancellations. A separate prob lem, detection bias, is wholly ignored despite the fact that many lung cancers are not identified during life and that the spouses o f smokers may be much more likely than spouses o f non smokers to seek and receive the tests and examinations that lead to a diagnosis during life. 51246 0046 4-2 /. .<.V*.;> .:-i/; 77 7 7-------------- -------------------------- -- -- :---- - --I-------- ' -------- " . *--' --------- 4. Unfair Selection and Emphasis In several locations and assertions, the authors o f the EPA Draft display their own "editorial' review bias" in choosing the evidence they will emphasize. Perhaps the most obvious example o f this "editorial-review bias" is in the emphasis given to the Hirayama cohort study, ignoring the Garfinkel cohort study. These are the only two pertinent cohort studies available to the Draft ; authors, but the Garfinkel study, which is generally regarded as having much better quality, found a negative result, (i.e., no statistically significant link between ETS and lung cancer), whereas the Hirayama study, which has been heavily criticized, was positive. In their conclusion, the authors focus only on the Hirayama study^i^oring the fitcit that when Hirayama made the ;J 7 necessary corrections to his datai (in a subsequent publication), die result was much ls positive. ; ; Thus, "editorial-review bias" has had a double impact. The EPA Draft authors emphasize the first Hirayama report, ignoring his subsequent correction; and they also ignore the contradictory Garfinkel report. Another obvious example o f "editorial-review bias" is the assertion on page 4-22 that ratios of 1 .0 2 and 1.07 indicate excess risk. Because these ratios are so dose to 1 ,1 cannot think o f a knowledgeable epidemiologist who would take them seriously. Several other statements could be cited to document the bias o f the EPA Draft writers. The only other one I shall note is their use o f terms such as "showed" and "demonstrated" when the results o f a particular study go in the desired direction (i.e. increased risk for passive smoking) and "not helpful" when the results go in the opposite direction. 51246 0047 4-3 . t. ... . A A Critique of EPA External Review Draft 600/6-90/006A, M ay 1990 Health Effects of Passive Smoking: Assessment of Lung C a n ce r in"Adults an d Respiratory Disorders In Children prepared by Robert W. Gardner, Ph.D . Ezra Taft Benson Agriculture and Food Institute Brigham Young University Provo, U T 84602 MU! cOfro>* ) ! Introduction This document was developed to supplement reports by theNational Research Council and the U.S. Surgeon General in assessing the health effects o f exposure to environmental tobacco smoke (ETS). Both o f those reports were published in 1986 and several more recent studies were deemed valuable in aiding the Environmental Protection Agency in making correct judgments as to possible carcinogenic effects o f ETS, as well as respiratory disorders in children. A meta analysis o f results from diverse epidemiological studies was used to establish large num bers which would behdpful statistically in ascertaining the biological significance o f ETS on human health. Comments The use o f meta analysis to statistically analyze studies with such diverse testing methods and parameters for measurements tends to invalidate the statistical analysis. However, there are important findings associated with studies reported. O ne is the trend in relative risk associated with increased exposure to ETS. Obviously, identical environmental conditions to test effects on all smokers vs nonsmokers are needed to satisfy all critics, this because of other possible environ mental contaminants. Precise quantities o f chemicals which subjects actually inhaled from environmental tobacco smoke are not well defined. Housing space, ventilation systems, humidity, and other variables may have led to apparent discrepancies between studies due to different exposure concentrations. There are some trends which are suggestive o f passive smoking being associated with the etiology of cancer. However, more research and more coherent assessment of past research is required before ETS can be classified as a Group A carcinogen. The discussion which follows is presented in an effort to try to identify possible reasons for seeming discrepancies observed in these ETS studies and to identify physiological reasons for respiratory disorders in children exposed to ETS. Phenolic compounds, nicotine, and other aromatic compounds liberated from burning tobaco are potentially toxic and/or pharmacological agents which affect many susceptible individuals. Such volatile chemicals in tobacco are impor- 5-1 51246 0049 .{.--S'* . fX ffi ! # |||;; tant in influencing the taste and aroma o f tobacco smoke (1). O ne o f the effects o f these com pounds is to evoke the release o f catecholamines. This subsequendy can initiate the synthesis and release o f eicosanoids (prostaglandins, thromboxanes, and leukotrienes). These inflammatory agents may in turn cause asthma, irritation o f nasal passages, headaches, cardiovascular disorders, inflamed intestines, inflamed lungs, etc. (2). Some individuals are much more susceptible to these chemical responses than others, and some seem to respond more to the compounds by inhalation than ingestion (and vice versa). Allergic and pharmacological effects are both involved. For example, Ronchetti et al. (3) compared 179 children in fourth grade from three Italian ; towns whose parents smoked with those who were nonsmokers. Those children whose parents smoked had higher IgE levels plus significantly higher total counts and percentages o f eosi- ,, 1` , 1 nophils. Eosinophilic inflammation o f thie airways is correlated with the severity o f asthina (4). m y, Another perspective o f the impact of allergy related to tobacco smoke comes from the study of Murray and Morrison (5) in Canada. They examined 620 children, 1 to 17 years o f age, who were nonsmokers but had a history of atopic dermatitis. Children with a history of atopic dermatitis were much more likely to have asthma if the mother was a smoker than if she was a nonsmoker (79% versus 5 2 %; p=0.001). Similarly, if atopic dermatitis was found on examina tion, the percentages with asthma were 74% and 44%, respectively. By contrast, the children with no history o f atopic dermatitis had asthma as frequently if the mother was a nonsmoker (42%) as when she was a smoker (40%). UH*1 ticotr> <S> <iSn> 5-2 w* - i References 1. Weeks, W W ., Chaplin, J.F., and Campbell, C.R. Capillary chromatography: Evaluation o f volatiles from flue-cured tobacco varieties. J. Agric. Food Chem. 37:1038,1989. j 2 . Ninnemann, J.L. Prostaglandins, leukotrienes, and the immune response. Cambridge Uni versity Press, New York, 1988. 3. Ronchetti, R. et al. Increased serum IgE and increased prevalence o f eosinophilia in 9-yearold children o f smoking parents. J: Allergy Clin. Immunol. 8& .400,1990. : j | j . 4. Bousquet, J. et al. Eosinophilic inflammation in asthma, N Engl. J. Med. 323:1033,1990. 1 5. Murray, A.B., and Morrison, B.J. It is children with atopic dermatitis who develop asthma more frequently if the mother smokes. J. Allergy Clin. Immunol. 86:732,1990. I 51246 0051 5-3 A Critique of EPA External Review Draft 600/6-90/006A, M ay 1990 Health Effects of Passive Smoking: Assessment of Lung C a n ce r in Adults and Respiratory Disorders in Children prepared by A lan J. Gross P h .D . Medical University o f South Carolina Charlston, SC 29425-2503 51246 0052 Abstract The issue of whether there is an association between exposure o f nonsmokers to environmental tobacco smoke (ETS) and their prevalent* of lung cancer is a pervasive one. In 1986, the N a tional Research Council and the Surgeon General both published reports purporting to demon strate such an association. These documents, unfortunately, were flawed on many scientific grounds and when adjustments were made where they could be, there proved to be a total lack o f such an association. Now, four years later, the Environmental Protection Agency (EPA) issued a draft document dealing with this same issue. It becomes clear upon careful scrutiny o f this Draft that many o f the previous flaws in the study design remain uncorrected in the Draft. This j is especially true in the case o f an attempt to use, again, meta analysis to combine the results of 21 case-control studies, each o f which addresses the question o f whether individuals exposed to j ETS at a higher level are more likely to develop lung cancer than those at a lower level. Unfortu nately, this meta analysis does not account for the many biases that have existed and still exist in these studies and so reduces itself to becoming merely an "interesting statistical exercise" with no import. Introduction T he preparation o f this critique is in response to the external review draft Health Effects o fPassive Smoking: Assessment o fLung Cancer in Adults and Respiratory Disorders in Children issued by the United States Environmental Protection Agency, Office of Health and Environmental Assess ment, Office of Atmospheric and Indoor Air Programs, Washington, D C 20460. The principal focus o f this critique is on Chapters 3 and 4 o f the EPA document. In particular, this critique is an assessment o f the epidemiologic evidence o f lung cancer from environmental tobacco smoke (ETS) and the estimates o f relative risk from the epidemiologic data. The question which is raised is whether any direct evidence exists for the relationship between ETS exposure in the general population of the United States and its implications for the public health o f the nation. To address this question, a review and analysis are provided o f the existing epidemiologic studies in which individuals, both cases and controls (noncases), who have higher 6-1 51246 0053 ETS exposures are compared to those with lower exposures, both cases and controls. Typically, the study subjects are married women who presumably never smoked but are married cither to a smoker (higher exposure) or a nonsmoker (lower exposure). T he methodology in the document is a meta analysis o f 21 case-control studies involving mar ried women both exposed and unexposed to ETS. These studies are derived from the literature and have taken place world-wide during the last 15 years. In an article by Fleiss and Gross (1991), the authors indicate that meta analysis, a set o f statistical tools for combining and integrating the results o f independent studies o f a given scientific issue (ETS exposure in this case), can be useful when stringent conditions under which such integra- j tion is valid' are met. T he cond'itions that need to be addressed include: T he aJp. &propriate adsjust ! m ent or control for the biases that frequendy occur in epidemiologic studies such as socio demographic or clin ical differences among study populations, misclassification o f subjects with regard to case-control status and levels o f exposure, factors other than the level o f exposure that may affect whether a subject is a case or a control, i.e., confounding variables and the publication bias/file drawer phenomenon wherein studies that fail to show a positive association are not published and thus are not candidates for inclusion in the meta-analysis. It is the contention o f this critique that the document in question, Health Effects o fPassive Smoking: Assessment o fLung Cancer in Adults and Respiratory Disorders in Children (referred to as the D ocument in the remainder of this report) fails to meet most, if not all, these require ments for the appropriate meta analysis o f the epidemiologic studies in question. Furthermore, it will be demonstrated that, depending on how one proceeds with the meta analysis, the results are inconclusive at best and may lead to unwarranted and potentially frightening conclusions to the lay public without having reached these conclusions in a scientifically objective manner. For example, in Table 3-5 o f the Document, a meta analysis of the 19 raw studies shows a combined relative risk (RR) for lung cancer among nonsmoking females exposed to ETS (being married to a smoking male) o f 1.42 compared to nonsmoking females who are unexposed (being married to a nonsmoking male). T he 95-percent confidence interval for this RR is from 1.24 to 1.63. O n the other hand, the adjusted meta analysis based on Table 3-6 using the 11 studies showing 95percent confidence intervals, i.e., omitting the studies o f Lam (1985) and Shimizu et al. (1988) (as the Document does) produces an RR o f 1.17 with a 95-percent confidence interval from 0.99 6-2 51246 005 to 1.63. Such a conclusion is consistent with the null hypothesis that there is no association between exposure to ETS and the prevalence of lung cancer. Using the DerSimonian-Laird test for homogeneity o f studies, X2 = 13.46, d f = 10, p > 0.10, indicating relative homogeneity, numerically*, among studies. (See DerSimonian and Laird (1986) or Fleiss and Gross (1991) for details.) Thus, the two major tables for meta analysis purposes report somewhat conflicting results with regard to the purported association between nonsmokers' exposure to ETS and the prevalence o f lung cancer. Finally, it is interesti ng to note that in the Document no meta analysis o f the corresponding male exposure data was attempted. Perhaps it was omitted because, as the Document states on p. 3-14, "Data on males is sparse by com parison..." However, there are six studies on nonsmok ing males with lung cancer who were exposed or unexposed to ETS as in the female studies; i.e, the exposed group consists o f nonsmoking men married to smoking women and the unexposed group consists o f nonsmoking men married to nonsmoking women. Using the notation of the Document, these studies are AKIB, BROW , BUFF, CORR, KABA, and LEE. A meta analysis of these six studies provides an RR for males o f 1.14 with a 95 percent confidence interval from 0.61 to 2.16. The DerSimonian-Laird test for numerical homogeneity o f studies shows X2 = 3.136, d f = 5, p > 0.10, indicating that the male studies are relatively homogeneous. O ne sees that there is no evidence whatsoever o f an association between ETS exposure and the prevalence of lung cancer in males. II. Study Biases As pointed out in the introduction, appropriate adjustment or control for the biases that fre quently occur in epidemiologic studies is necessary before analyses of these studies can proceed and the results o f these analyses be interpreted properly. As Fleiss and Gross (1991) (referred to as F-G) indicate, the major potential biases in epidemiologic studies include: (i) sociodemographic or clinical differences among study populations; (ii) misdassification of subjects with regard to case-control status and levels o f exposure; (iii) other confounding factors such as the age and sex o f the study subjects; and (iv) the publication bias in which studies that * This does not imply necessarily homogeneity of study methods, only their numerical homogeneity of study results. 6-3 51246 0055 show no association or a negative association are not published in the available literature and hence never have the opportunity o f being included in an analysis of the epidemiologic studies readily accessible. It is important to comment on how these biases are addressed in the Document. In the first place, due to the inclusion o f worldwide epidemiologic studies on the relationship o f ETS exposure and the prevalence of lung cancer, one would assume that the Document would have included a detailed description o f the existing cultural as well as socioeconomic differences among the populations studied. N o such description was found in the Document. It is well- known, for example, that living conditions differ in different part of the world. Living space is more limited in Europe and Asia that it is in N orth America. Moreover, Asian women tend to fYlAvt*vv* Ar\(awa/uI twkmontt u v PmuwrsAvA^ vMsuni onrl imsu jAumu vAsPiivriouni rwAuimmvuim| / (ourwtet Takmnucf PtviimmAvoe irv iti cooking oils as well as coal stoves (which are still fairly common in both Europe and Asia) contribute to environmental smoke in these non-North American settings. Another major difference among the three cultures is in the brands o f cigarettes smoked. These, o f course, differ from culture to culture. Also, with regard to living quarters in the three separate cultures, it is very common in Asia for elderly parents or in-laws (or both) to dwell in the same residence as the wife who may not only be exposed to her spouse's ETS but, also, to ETS from a smoking father or father-in-law. An attempt is made on p. 3-13 of the Document to quantify the percentage o f controls exposed to ETS by study which, by the way, shows a threefold increase from minimum to maximum exposure. However, cultural exposure differences are not discussed. Finally, there are clear genetic differences among the three populations. It is quite possible that particular ethnic groups are more susceptible to lung cancer, independent o f environmental factors, than other ethnic groups. This issue also is not addressed in any detail. W hat one does read on p. 3 12 is the rather remarkable statement, however, "Study differences do not invalidate statistically testing the hypothesis that exposure to ETS is unrelated to lung cancer occurrence." Considerably more attention is paid to the issue o f misclassification o f smoking status. In par ticular, the status of the female subjects, both controls and cases, is discussed. A formula is developed in Appendix B to correct for misreporting current smokers (CS) and former smokers (FM) as never smokers (NS). The append itself contains much detail and does not, at this time, permit an extensive review. However, the results o f this so-called "reduction formula" do not seem to have been applied in the analyses o f the epidemiologic studies. Instead, a sensitivity 6-4 Mtn N) Cl <s> oo>> / I analysis is performed to demonstrate, under a "worst-case" scenario, that the observed RR from epidemiologic data still indicates a statistically significant association between ETS exposure and the prevalence o f lung cancer in nonsmoking females. The fact remains, however, that a misdassification bias in smoking status ousts in all studies, and the fact also remains that it is more likely that a current smoker or a former smoker is misclassified as a never-smoker than for the never-smoker to be misclassified as a former smoker or current smoker. It should be noted that the Document acknowledges this fact on p. 3-12 as it states: "A few studies include former smokers as nonsmokers if they have abstained from tobacco usage for some minimum period while others do not. Classification o f a subject as ETS-exposed depends on the questions asked which differ across studies." Such an admission o f study-to study heterogeneity is rather interest- j ing in view o f the fact that meta analysis is subsequendy used to combine results o f studies. Another issue concerning misdassification is the definition o f "exposed" versus "unexposed" individuals. O n p. 3 -1 2 one reads: "The relative risk comparison o f exposed to unexposed individuals, however, is implicidy a comparison o f `exposed to both background and spousal smoke' to `exposed to background only'." Although this issue is dealt with in the Document and has been dealt with in the N R C (1986) report, the way in which this is treated, i.e., assuming that an individual with both background and spousal exposure to ETS has three times the amount o f ETS exposure as only background exposure, (NRC (1986) report, p. 291) in no way pertains to the individual studies. It is merely a macro-adjustment applied to all studies simulta neously, not a fine tuning o f individual studies. O ne can easily envision situations in which an "exposed" individual is much less "exposed" than the so-called "unexposed" individual. For example, an "exposed" female may be a housewife whose smoking husband travels a great deal and hence her primary exposure may be only on weekends when they are together. O n the other hand, a "nonexposed" female may work in an area where there is some moderate exposure to ETS and on weekends her primary nonexposure may take place when she is not at work and with her nonsmoking spouse. As F-G point out, other confounding variables should be accounted for and adjusted in prepara tion for performing a meta analysis on epidemiologic studies. As one scans Table 3-1 in the Document (pp. 3-2 - 3-4, inclusive) it becomes clear that variables on which the final sample is matched differ considerably over the di fferent studies. Primarily, however, age and sex are matched. However, some important potentially confounding variables such as ethnicity, place 6-5 51246 0057 of residence, socioeconomic status, and type o f control subject are not matched in the individual studies. Moreover, the matching on sex is done by restricting the study to female subjects. However, if the male studies are considered by themselves, it was shown in the introduction that no statistically significant association exists between exposure to ETS and lung cancer prevalence. . III. Comparison o f Meta-Analyses Although not all biases can be accounted for or eliminated, F-G nevertheless believe that it is important to restrict a meta analysis o f ETS exposure and risk o f lung cancer in nonsmoking subjects to the studies done in the USA. As they state: "There are many reasons for restricting attention to American studies o f whether there is an elevated risk to nonsmokers exposed to ETS relative to nonsmokers not so exposed. O ne is that this is the population on whom policy deci sions will be based, and to whom those decisions will apply. Another is that the summary odds ratios in the individual studies are derived from distributions of smoking amounts and durations, and of brands o f cigarettes and other tobacco products, that pertain to populations within the United States, and may thus be expected to be relatively homogeneous. Odds ratios from studies in other countries, on the other hand, are derived from distributions that may differ markedly from those in the U.S. and thus odds ratios may not be relevant to the American experience. Genetic and lifestyle differences between the U. S. population and the populations studied elsewhere (mainly in east Asia) also argue for a meta-analysis only o f U. S. studies." T he nine U. S. studies in the F-G meta analysis are eight case-control studies identified in Tables 3-5 and 3-6 o f the Document, which arc BROW, BUFF, CORR, GARF, HUM B, KABA, VARE, and W U. The ninth study is the cohort study o f Garfmkel (1981). As these studies are all well described in both the Document and F-G, only the results o f their meta analyses are presented here. Using the method o f analysis in F-G (which is also used in the other meta analyses in this report), the overall RR for the nine American studies is 1.12 with a 95 percent confidence interval from 0.95 to 1.30. The DerSimonian-Laird test for homogeneity yields X2 5.46 with d f = 8 , p > 0.10, indicating relative numerical homogeneity from study to study within the U.S. epidemiologic studies. Thus, based on all the available American epidemiologic evi dence, there is no scientific basis for concluding, as was done in the Document, that (i) there is 51246 0058 6-6 an association between exposure to ETS and the risk of lung cancer in the nonsmoking popula tion in the Untied States, and, hence, that (ii) ETS is responsible for 3,800 lung cancer deaths among nonsmokers in the United States annually. IV. Other Issues In this section o f the critique, specific points made in the Document, or specific statistical procedures that are open to question, are discussed. Unfortunately, time does not permit a complete review of all o f the methodology presented un the Document. First o f all, on p. 1-4 it is stated: "O f the two major cohort studies, the Japanese study I (Hirayama) demonstrates a strong association between passive smoking and lung cancer includ ing an upward trend in dose-response." If the dose-response issue is to be addressed, then why wasn't the other major cohort study by Garfinkel (1981) also referred to at this point? Perhaps the reason is that no clear dose-response relationship was shown in the Garfinkel study and, in fact, the RR is 1.27 for women whose husbands smoked between one and 19 cigarettes a day and 1.10 for women whose husbands smoked a pack a day or more. Neither R R is statistically significant. Furthermore, the Hirayama study has been widely criticized in the literature for its many design flaws. For example, see Kilpatrick (1989). Next, on p. 3 -2 1 one reads: "Table 3-1 identifies the studies with results adjusted for other variables. Some authors have not included complete details, so the choice o f studies for inclusion in this section may be subjective." Many of the so-called "statistically significant" studies that are included, such as AKIB, BUFF, CHAN, CORR, GENG, KABA, KOO, LAMT, AN D TR IC , did not adjust for many of the possibly confounding covariables. As was earlier discussed, com bining unadjusted studies or studies in which confounding variables are not or cannot be con trolled cannot be justified scientifically. O n p. 3 -2 2 the following passage appears: "The values o f S are displayed in Table 3-6 and plotted in Figure 3-3. Five o f the studies had significant values o f S (p < 0.05). The probability o f observing five or more Type I errors in 11 independent studies is less than 0.001. Thus, it is highly unlikely so many significant test results would be observed if there were, in fact, no association between ETS exposure and lung cancer incidence." The 95-percent confidence 6-7 51246 0059 intervals indicate an entirely different situation. Only one study, that o flnoue and Hirayama (1988) , demonstrates a 95-percent confidence interval that excludes unity. This event can happen by chance alone, assuming the null hypothesis of no association between ETS exposure and lung cancer incidence in nonsmokers , about 43 percent o f the time, i.e., p 0.43, hardly a finding inconsistent with the hypothesis o f no association between ETS exposure and the incidence o f lung cancer in nonsmokers. Also, with regard to the application o f the S statistics, one reads on p. 3-22: "The Wilcoxon signed-rank test was also applied to the S statistics of Table 3-6 as conducted previously with the raw data, to provide another statistical test o f the null hypothesis. T he outcome is significant (p = 0.014)." The combining o f the S statistics to obtain a composite statistic is done incorrectly because all S statistics are given the same weight < . regardless of sample size. Thus, the S statistic in a small study receives the same weight as the S statistic in a large study. For example, the largest study (Varela, 1987) receives the same weight as the study by Inoue and Hirayama (1988) which is a small study. The only correct method o f combining study results (if they are combinable) is to weight them properly by their size. The next comment o f interest appears on p. 3-22 as well: "No multiple comparison adjustment is necessary because the choice o f a single exposure level is made without regard to statistical significance. Test results reported at exposure levels other than the one used are not relevant and no adjustment for multiple comparison is needed." This statement requires considerable clarifi cation. T o this point:, not much is written concerning exposure levels. At the very least, Layard (1989) points out that in the Garfinkel (1981) study: "W omen whose husbands smoked 20 or more cigarettes per day had a lower relative risk ( 1. 10) than those whose husbands smoked 1-19 cigarettes per day (1.27)." Concerning levels o f exposure and the attempt in the Document to show an increasing relative risk with an increasing level of ETS, one reads on p. 3-25: "For example, the estimated RRs increase in seven studies: AKIB, CORR, GAO, GENG, PERS, A N D TRIC; decrease slightly in one case, LEE; and are variable in the remaining five plots, GARF, HUM B, K O O , LAMT, and VARE." Further on, it states: "The observed RR at the highest exposure level is less than one in only two o f the 13 studies above. The probability o f two or fewer such occurrences by chance alone is approximately 0 .0 1 2 ." It should be noted, rather strongly, that the lower confidence limit at the highest level o f ETS exposure is less than unity in the following studies: AKIB, HUM B, KOO, LEE, and VARE. N o mention is made o f those studies showing a reversal o f trend, or how large the sample size is in those studies where the RR , exceeds one in the groups with the highest ETS exposure. 6-8 51246 006 The next issue that is o f concern is the accurate diagnosis o f disease. For example, if lung cancer is the disease of study (as it is here), then misclassification due to diagnosis is an important issue. T o this end one reads on p. 3-33: "Some studies addressed this issue by including only patho logically confirmed lung cancers or by considering histological cell type in their analyses (e.g., CORR, GARF, PERS and others)." T he statement is imprecise. W e are given a few studies where one or the other o f these important issues is discussed. Cell type is o f critical importance because lung cancer occurring in nonsmoking individuals is usually adenocarcinoma, which is not the primary cell type for smoking individuals. It is clear that the case-control studies set out for analysis in the Document have not all been looked at from the standpoint o f including only pathologically confirmed lung cancers and giving a breakdown o f the lung cancer types. Another potential source of bias that the Document attempts to address is that o f surrogate responders.* O ne reads on p. 3-33: "T he recent study by Cummings et al. (1989a) o f the passive smoking histories o f 380 NS further supports that conclusion. They report substantial agreement between subjects and surrogates on most exposure measures." In the first place, this study took place at Rosewell Park, which has been a major cancer treatment center for many years. Can its findings be generalized to all the epidemiologic studies presented in the Document? Even in this report, differences between subjects and surrogates exist. For example, the childhood exposure index showed a relative difference o f 21 percent between subjects and surrogates. The adult index shows similar problems. Finally, in their conclusion Cummings et al. (1989a) state: "How ever, it should be kept in mind that while our findings suggest self-reports of exposure to passive smoke can be reproduced by surrogates, we cannot claim that our exposure measures accurately reflect `true* exposure to tobacco smoke." Finally, with regard to publication bias, the Document states on p. 3-33: "Wells (1989a) re viewed the subject and found it unlikely that publication bias has any substantial effect on the RRs that have been calculated from published reports for passive smoking for either men or women." Wells' review of the issue of publication bias is hardly definitive and certainly is not the "last word" on the subject. Wells agrees with Vandenbroucke (1988) that data on males arc sparse and that there is danger in attempting to extrapolate a relative risk for males based on few data. * No breakdown of surrogates is given. That is, we are not informed as to the relationship of the surrogate to the patient. 6-9 51246 0061 D ata for males, as is shown in this critique, are no longer that sparse and some inference can be and is drawn from them. Although Wells attempts to refute Vandenbroucke's arguments in favor o f a publication bias, it is somewhat curious when Wells writes: "If any investigators have data on passive smoking, however, particularly for men, that have not been published or that they have not been able to get published, I would be interested in receiving them for a possible subsequent report." V. Conclusions The Document appears to be an attempt by the EPA to justify and te n d the findings in the N R C (1986) report in which a purported association exists between the exposure to ETS and the risk o f lung cancer in nonsmoking females. Assuming this to be one o f the major goals o f the Document, it again fails to demonstrate any such association, as does the N R C report. Similar criticisms still exist concerning the findings with respect to this so-called "association." The principal tool that is used to attempt to justify this "association" is meta analysis. To reiter ate Fleiss and Gross (1990): "M eta analysis, a set o f statistical tools for combining and integrat ing the results o f independent studies o f a given scientific issue can be useful when the stringent conditions under which such integration is valid are met." The Document failed to meet these stringent conditions. References Akiba, S.; Kato, H .; Blot, W . J. (1986) Passive smoking and lung cancer among Japanese women. Cancer Research, 46:4804-4807. Brownson, R. C., J. S.; Keefe, T . J.; Ferguson, S. W .; Pritzl, J. A. (1987) Risk factors for adenocarcinoma o f the lung. Am. J. Epidemiol. 125:25-34. Buffler, P. A.; Pickle, L. W.; Mason, T . J.; and Contant, C. (1984) The causes o f lung cancer in Texas. Mizell, M. and Correa, P., eds. Lung Cancer: Causes and Prevention. New York: Verlag Chemie International, pp. 83-99. 6-10 51246 0062 Chan, W . C.; Fung, S. C. (1982) Lung ameer in non-smokers in Hong Kong. Cancer Cam paign, Vol. 6 , Cancer Epidemiology, (Grundmann, E., ed.) Gustav Fischer Verlag, Stuttgart pp. 199-202. Correa, P.; Fontham, E.; Pickle, L.; Lin, Y.; Haenszel, W . (1983) Passive smoking and lung cancer. Lancet 2:595-597. Cummings, K. M.; Markello, S. J.; Mahoney, M. C.; Marshall, J. R. (1989a) Measurement o f lifetime exposure t:o passive smoke. Am. J. Epidemiol. 30:122. DerSimonian, R.; Larid, N . (1986) Meta-analysis in clinical trials. Controlled Clin. Trials 7:177* .188. ; ; . - , , v ` '`'-`r.': " . - f , !: y . - S ' X Environmental Protection Agency (1990) Health Effects o f Passive Smoking: Assessment of Lung Cancer in Adults and Respiratory Disorders in Children. EPA External Review Draft 600/6-90/006 A. Fleiss, J. L.; Gross, A . J. (1991) M eta analysis in epidemiology, with special reference to studies o f the association between exposure to environmental tobacco smoke and lung cancer. A cri tique. Journal o f Clinical Epidemiology, 44:127-139. Garfinkel, L. (1981) Tim e trends in lung cancer mortality among nonsmokers and a note on passive smoking. J. N at'l. Cancer Inst. 6:1061-1066. Garfinkel, L.; Auerbach, O.; Joubert, L. (1985) Involuntary smoking and lung cancer: a casecontrol study. J. N at'l. Cancer Inst.75:463-469. Geng, G.; Liang, Z. H .; Zhang, G. L. (1988) O n the relationship between smoking and female lung cancer. In: Smoking and Health, Elsevier Science Publishers, pp. 483-86. Hirayama, T. (1984) Cancer mortality in nonsmoking women with smoking husbands based on a large-scale cohort study in Japan. Prev. Med. 13:680-690. Humble, C. G.; Samet, J. M.; Pathak, D. R. (1987) Marriage to a smoker and lung cancer risk. Am. J. Public Health 77:598-602. Inoue, R.; Hirayama, T . (1988) Passive smoking and lung cancer in women. In: Smoking and Health. Elsevier Science Publishers, pp. 283-285. 51246 0063 6-11 Kabat, G. C.; Wynder, E. L. (1984) Lung cancer in nonsmokers. Cancer, 53:1214-1221. Kilpatrick, S. J. (1989) Model Specifications in ETS/Nutritional Research. Indoor Air Quality (Proceedings o f the International Conference, Tokyo) pp. 256-271. Springer-Verlag, Berlin. Koo, L. C . ; H o, J. H.; Saw, D.; Ho, C. Y. (1987) Measurements o f passive smoking and esti mates o f lung cancer risk among non-smoking Chinese females. Int. J. Cancer, 39:162-169. Lam, T. H .; Kung, I. T . M.j W ong, C. M.; Lam, W ., K.; Kleevens, J. W . L.; Saw, D.; Hsu, C.; Seneviratne, S.; Lam, S. Y.; Lo, K. K.; Chan, W . C. (1987) Smoking, passive smoking and histological types in lung cancer in H ong Kong Chinese women. Br. J. Cancer, 6:673-678. Lam, W . K. (1985) A clinical and epidemiological study o f carcinoma o f lung in H ong Kong. Thesis submitted to the University o f H ong Kong for the degree o f Doctor o f Medicine. Layard, M. (1989) Environmental tobacco smoke and cancer: The epidemiologic evidence. Chapter 6 in Environmental Tobacco Smoke. Proceedings o f the International Symposium at McGill University, Montreal. Lee, P. N.; Chamberlain, J.; Alderson, M . R. (1986) Relationship of passive smoking to risk o f lung cancer and other smoking-associated diseases. Br. J. Cancer, 65:97-105. Lee, P. N. (1988) Misclassification o f smoking habits and passive smoking. Springer Verlag, Berlin. National Research Council (NRC) (1986) Environmental tobacco smoke: Measuring exposures and assessing health effects: National Academy Press, Washington, D. C. Pershagen, G.; Hrubec, Z.; Svensson, C. (1987) Passive smoking and lung cancer in Swedish women. Am. J. Epidemiol. 125(1): 17-24. Shimizu, H.; Morishita, M.; Mizuno, K j et al. (1988). Case-control study o f lung cancer in nonsmoking women. (1988) TobokuJ. Exp. Med. 154:389-397. Trichopoulos, D. (1988) Passive smoking and lung cancer. Scand. J. Soc. Med. 16:75-79. Trichopoulos, D.; Kalandidi, A.; Sparros, L. (1983) Lung cancer and passive smoking: conclu sion o f Greek study. (Letter) Lancet, 667-668. 6-12 51246 006 *> Vandenbroucke, J. P. (1988) Passive smoking and lung cancer: a publication bias? Br. J. Med. 296:391-392. Varela, L. R. (1987) Assessment o f the association between passive smoking and lung cancer. Dissertation to Yale University in candidacy for the degree o f Doctor o f Philosophy. Wells, A. J. (1988a) Re: Passive smoking and lung cancer: a publication bias? (Letter) B r.M ed.J. 296:1128. W u, A. H.; Henderson, B. E.; Pike, M. D.j Yu, M. C. (1985) Smoking and other risk factors for lung cancer in women. J. N at'l. Cancer Inst. 74(4):747-751. j 51246 0065 / 6-13 A Critique of EPA External Review Draft 600/6-90/006A, May 1990 Health Effects of Passive Smoking: Assessment of Lung C a n ce r in Adults an d Respiratory Disorders in Children prepared by G unter Oberdorster, D .V .M ., P h.D . University o f Rochester School o f Medicine and Dentistry Environmental Health Sciences Center Rochester, NY 14642 <crS>> 51246 The EPA draft focuses solely on the carcinogenic potential of environmental tobacco smoke (ETS) and its adverse effects on the respiratory system o f children. Relevant epidemiologic control and cohort studies are reviewed, a meta-analytical approach is described to evaluate the studies together; applied mathematical formulas, dosimetric considerations o f ETS and alterna tive approaches for estimation o f lung cancer deaths due to ETS are appended. Two prior publications, the U.S. Surgeon General's Report on the Health Consequences o f Involuntaty Smoking (1986) and the N RC report on Environmental Tobacco Smoke (1986) arc frequently cited and used as background material to support the major conclusions o f this document. These conclusions are: 1) active smoking is causally associated with lung cancer in adults; 2) exposure o f young children to ETS from parental, smoking, in particular mother's smoking, is associated with a number o f adverse effects in the respiratory tract o f the children, including acute1lower respiratory tract infections, respiratory tract irritations, and reduced lung function. In contrast to the pulmonary carcinogenic effects o f ETS, a causal relationship between ETS exposure and children's adverse respiratory effects was not concluded. Based on these analyses and following U.S. EPA guidelines for carcinogenic risk assessment, EPA concludes that ETS is a Group A (known human) carcinogen. Critical Review Introduction: The EPA draft, hereafter referred to as the Draft, may be viewed as an update j | o f the reports o f N RC and the U.S. Surgeon General mentioned above, including new data which have appeared since 1986. W hile many o f the relevant issues are more or less extensively addressed in this document, several other important areas are only superficially mentioned or not at all, and it would be useful to include these for a better understanding o f problems related to effects o f ETS on the respiratory tract o f nonsmokers. Furthermore, the reader o f the Draft gets the impression that its authors started out with a preconceived opinion about a positive and causative correlation between ETS exposure and lung cancer induction in persons so exposed and that consequently this biased opinion limits an objective discussion in many parts o f the document. The following general and specific remarks are intended to point out the shortcom ings and to suggest additions or a revised version o f the Draft to improve its understanding for the uninformed reader and its acceptability for the scientific community. 51246 0067 7-1 General M y own expectation bias-- prior to reading specific relevant studies--was that ETS should be regarded as a weak human lung carcinogen. This was not simply based on a similarity between ETS and mainstream smoke (MS). Actually, the two forms o f cigarette smoke are dissimilar in many respects which should be made clear in the Draft-- but more specifically on the fact that ETS contains known carcinogens-- such as PAH and nitrosamines-- and respiratory tract irri tants-- such as aldehydes-- as well as particles which could act as carriers to distribute and deposit these potentially toxic and carcinogenic chemicals deep into die respiratory tract. In addition, the fact that MS smoking has dearly been linked with lung cancer in epidemiological studies and a causal relationship between active smoking and lung cancer is universally^accepted in the scientific community, contributes to the plausibility of the link between ETS arid human lung cancer if exposure to ETS is high enough (see, however, comments below). Because expo sure to ETS is definitely much lower than to MS, and because deposition in the respiratory tract is quite different, only a weak positive correlation between ETS and lung cancer would be expected which may be hard to detect in epidemiological studies. Indeed, the case-control studies cited in the EPA document do not uniformly show a significant correlation, and even reversed dose response correlations were found. O f the two large cohort studies, only the Hirayama study (exposed Japanese women) showed a significant correlation and dose-response relationship, whereas the Garfinkel study (American women) did not find a significandy elevated lung cancer risk and no dose-response relationship. Such results would be expected o f a weak pulmonary carcinogen, weak in the sense that exposure and the resulting doses in the respiratory tract are generally rather low. This result is reminiscent of epidemiological studies in diesel-exposed populations where results were either equivocal or did reveal only a weak positive, albeit significant, correlation. Collectively these diesel studies were reviewed by IARC (1989) as showing a positive correlation. However, there were uncertain ties about the exposure assessment. Therefore, the overall evidence that diesel exhaust is a human lung carcinogen was termed limited. W ith respect to ETS, EPA comes to a different conclusion, based mainly on a meta-analytical evaluation of the case-control studies described in the Draft and on the plausibility o f causal I i association. Perhaps not surprisingly, when these case-control studies--which individually did not uniformly show a significant correlation between lung cancer and ETS exposure to spousal smoking--were subjected in the Draft to meta analysis they showed a significant correlation with this combined approach. : Because I do not regard myself as an expert o f meta-analytical methodology, I cannot comment thoroughly on the specific mathematical methodology applied by EPA to the case-control studies. It appears to be reasonable. However, because the individual epidemiological studies were not conducted using the same or at least a similar study protocol and because they differed in their quality, each being afflicted with certain limitations, I am surprised that these studies were used without setting certain criteria regarding a study's acceptance for such analyss prior to initiating it. For example, is the collection o f data comparable (assessment o f exposure), have the studies undergone peer review in the literature (inclusion of dissertations?), were attempts made to correct for misclassification, was the definition o f lung cancer used in the same way in the different investigations (inclusion or exclusion o f adenocarcinoma, adenocarcinoma only, etc.), was workplace co-exposure considered? I think, studies subjected to meta analysis should be o f : equal quality and should have used comparable methodologies when they are weighted equally in a combined analysis. Differences in research quality among the different studies must be consid ered (see: Bangert-Drowns, 1986). Should any study suffering from flaws that may obscure ETS effects be rejected? Possibly yes. "Criteria for study inclusion must be so explicitly stated that . others may replicate the results or evaluate limitations of the study" and "the meta-analyst must test to determine whether differences in quality are related to differences in outcome" (BangertDrowns, 1986) might be some principles o f meta analysis to be discussed at a workshop. H old ing an expert workshop on meta analysis would be useful and should result in valuable rcommendations about meta-analytical methodology of epidemiological studies dealing with TS. j i Thus, it would be extremely desirable and necessary to repeat the meta analysis o f the casecontrol studies according to scientifically acceptable and dfendable methodology. W ithout meta analysis, the results of the different epidemiological case-control studies used in the Draft do not provide sufficient evidence for a causal association between lung cancer and passive smoking as is required for classification o f a Group A carcinogen (The Risk Assessment Guidelines o f 1986, EPA/600/8 87/085, 1987). Rather, this evidence should be termed limited, with ETS being categorized as a probable human lung carcinogen (Group B -l). This assessment is based on 7 -3 51246 0069 I association. Perhaps not surprisingly, when these case-control studies--which individually did not uniformly show a significant correlation between lung cancer and ETS exposure to spousal smoking--were subjected in the D raft to meta analysis they showed a significant correlation with this combined approach. Because I do not regard myself as an expert o f meta-analytical methodology, I cannot comment thoroughly on the specific mathematical methodology applied by EPA to the case-control j studies. It appears to be reasonable. However, because the individual epidemiological studies j were not conducted using the same or at least a similar study protocol nd because they differed in their quality, each being afflicted with certain limitations, I am surprised that these studies were used without setting certain criteria regarding a study's acceptance for such analyses prior to initiating it. For example, is the collection o f data comparable (assessment o f exposure), have the j studies undergone peer review in the literature (inclusion o f dissertations?), were attempts made to correct for misclassification, was the definition o f lung cancer used in the same way in the j different investigations (inclusion or exclusion o f adenocarcinoma, adenocarcinoma only, etc.), ; was workplace co-exposure considered? I think, studies subjected to meta analysis should be o f equal quality and should have used comparable methodologies when they are weighted equally in a combined analysis. Differences in research quality among the different studies must be consid ered (see: Bangert-Drowns, 1986). Should any study suffering from flaws that may obscure ETS effects be rejected? Possibly yes. "Criteria for study inclusion must be so explicitly stated that others may replicate the results or evaluate limitations o f the study" and "the meta-analyst must test to determine whether differences in quality are related to differences in outcome" (Bangert- Drowns, 1986) might be some principles o f meta analysis to be discussed at a workshop. H old ing an expert workshop on meta analysis would be useful and should result in valuable recom mendations about meta-analytical methodology o f epidemiological studies dealing with ETS. Thus, it would be extremely desirable and necessary to repeat the meta analysis of the casecontrol studies according to scientifically acceptable and defendable methodology. W ithout meta analysis, the results o f the different epidemiological case-control studies used in the Draft do not provide sufficient evidence for a causal association between lung cancer and passive smoking as is required for classification o f a Group A carcinogen (The Risk Assessment Guidelines o f 1086, EPA/600/8 87/085,1987). Rather, this evidence should be termed limited, with ETS being categorized as a probable human lung carcinogen (Group B -l). This assessment is based on j 7-3 51246 0069 In this context, the negative outcome o f the Garfinkel cohort study should not be played down as is done repeatedly in the Draft. Q uite the opposite, it should really be emphasized because it may indeed reflect a lower potency o f ETS from exposure to spousal smoking in the US popula tion due to US specific exposure conditions. This is in line with the negative results o f the Janerich et a l study in most o f their study cohorts except the group exposed from childhood on. In summary, the possibility cannot be ruled out that ETS is a pulmonary carcinogen; however, collectively the results o f the epidemiological studies do not support the conclusion o f EPA's Draft o f sufficient evidence for a carcinogenic potential. Finally, experts o f the field o f nieta analysis should re-analyze the available studies with appropriate methodology to strengthen the basis for classification o f ETS. i T he plausibility o f a correlation between ETS exposure and lung cancer which is based on the well-accepted data from mainstream smoke (MS) exposure (active smokers) becomes less obvious upon further analysis o f several underlying issues: 1) Differences in respiratory tract dosimetry between ETS and MS exposure. O n the one hand, we are dealing with high concentrations o f particle clouds (MS), and on the other hand, with i much lower concentrations of diluted smoke (ETS); breathing patterns during exposure to both : forms o f smoke are quite different; in addition, gas phase components also change (example, i nicotine). As a result, bronchial and deep lung deposition o f the respective particles change quite significantly which will be discussed in more detail under the heading Specific Comments. Such ! differences also make the cigarette equivalent approach questionable. (As part o f the dosimetry j issue, low-dose extrapolation also needs further discussion.) ; 2) Effects of freshly generated smoke (MS) vs. aged smoke (ETS). For example, do short-lived radicals play a role? Is there a phenomenon similar to polymer fume fever, where freshly gener ated fumes are highly toxic only during the first one or two minutes after their generation? i 3) Location and histology of lung tumors: There is a prevalence o f more central, bronchogenic lung tumors in MS smokers vs. more peripheral tumors (adenocarcinoma) in nonsmokers (Kuller etal., 1986) and it appears that there is no consensus among the different investigators which tumor type is prevalent in ETS exposed people. Authors o f the different epidemiological 51246 0071 > 7-5 J. studies have different opinions. Based on dosimetry (see above) and on the fact that chronic bronchitis in active smokers may be a significant pathogenetic event in MS-induced lung cancer, j one would expect that the two exposure modes (MS vs. ETS) lead to different target sites o f ! tum or induction in the respiratory tract. Such distinction would be important and would make it less obvious to use results from MS exposure to extrapolate down to ETS effects. O n the other i. hand, there is the issue o f a possible "metabolic overload", i.e., the high concentrations and doses o f carcinogens in inhaled MS as opposed to much lower concentrations in inhaled ETS. It is conceivable, as shown for BaP by W olff et al. (1989), that high doses o f BaP arriving at a target j cell over a short period o f time may overwhelm the cell's ability to metabolize BaP and to form | carcinogenic metabolites. Instead, much o f the BaP is quickly cleared via the blood drdulation 1 (W olff et al., 1989). Apparently, the dose delivery rate o f a carcinogen to the oell is an essential ; parameter as can be deduced from results on the formation o f BaP metabolites after inhalation ! o f the pure compound vs. inhalation o f BaP plus carbon blade particles (Sun etal., 1984; Bond e ta l, 1986). In the latter case, binding o f BaP with macromolecules was increased 10-to 20-fold. 4) Experience and results from epidemiological and experimental studies on the carcinogenic effects o f diesel exhaust may also question the plausibility o f a correlation between inhaled organic carcinogens and lung tumors: the tumor-inducing potential o f inhaled diesel exhaust can probably be explained by a pure partide effect (see attached manuscript by Oberdfirster and Yu). However, further studies on the inducibilty o f DNA-adducts in the lung may still reveal a possible contributory effect o f organic components. It would be useful to indude a discussion o f the issues addressed in the previous paragraph in a revised version o f the EPA document. This would show that difficult issues related, to the possible carcinogenic effects o f ETS have been addressed objectivdy from different viewpoints. T he present Draft is obviously a first attempt to deal with the difficult questions related to the carcinogenic potential of ETS exposure. An indusion o f the recommended changes and topics oudined in these comments would strengthen the document. Attention should also be drawn to the afore mentioned recent paper byjanerich etal. mentioned before (N. Engl. J. Med. 323:632-636, 1990: Lung cancer and exposure to tobacco smoke in the household. Published based on dissertation by Varela, 1987). The authors o f this study report that only exposure during early life (childhood and adolescence, less than 21 years o f age) 51246 0072 7-6 i. smoker years significantly increased the risk o f lung cancer (doubled it to 2.07), but exposure during adulthood (more than 21 years of age) had no significant effect on lung cancer. W hile recall bias may have influenced the result, as discussed by the authors, the study seemed other wise well conducted and should be included in a meta-analysis approach. Finally, it would be useful to include a table at the end o f Section 4 which shows the possible carcinogenic potency o f ETS in relation to other carcinogenic compounds, as is done in other Health Criteria D ocu ments (unit risk estimate). T he derivation o f such unit risk for ETS may be difficult due to the limited data base. However, this would give the reader a perspective about the ranking o f ETS among other inhaled carcinogens. . T he section on respiratory illness in children (respiratory symptoms; acute respiratory illness; impaired pulmonary function) addresses adequately possible confounding variables and biases of the different epidemiological studies. It would be useful and strengthen the document, however, to include (perhaps as an Appendix) a more detailed description o f the individual studies refer enced in this section. The conclusions in the document about a correlation between measured effects and parental, especially maternal, smoking appears to be justified. Obviously, all o f the epidemiological studies have some shortcomings and one can think of several possible confound ing variables which need to be investigated, such as ventilation in homes (sick-building syn drome), pets in household, crowdedness, activity o f child, etc. However, the document stops short o f suggesting a causal relationship between children's respiratory illness and parental smoking although in some studies family history o f coughing or respiratory illness was found to have a large influence, yet could not explain all o f the observed effects. Overall, this section is reasonably well written although several methodological questions still need to be discussed. Such discussion would be facilitated by the inclusion o f a brief description o f the studies as mentioned above. I believe the overall evidence from the different studies supports the condu- sion o f a correlation between ETS exposure and respiratory tract illness in children, which has important implications from a medical preventive point of view. | Specific Comments j Page 2-1, first para: ETS consists o f side stream smoke (SS) and exhaled mainstream smoke (MS) which should be kept in mind throughout the document. Most o f the discussion treats SS as the sole component o f ETS which is not correct. : <S> 7-7 Second para: The plausibility o f a carcinogenic effect o f ETS, based on the known carcinogenic effect o f MS exposure, is attractive and may be justifiable. However, there is also evidence that low concentrations o f known organic carcinogens inhaled together with high concentrations of particles may not contribute to a tumorigenic effect and one has to be cautious not to be biased by a preconceived view when conducting or reviewing a study on the effects o f ETS. For ex ample, diesel exhaust has many known carcinogens, yet the positive (in terms o f tum or induc tion) results in long-term rat inhalation studies with diesel exhaust very likely are not caused by the carcinogens but by the particulate phase, even without the organics. Inhalation add intratracheal instillation studies with carbon black alone without carcinogens gave similar tum or responses in rat studies. In addition, the well-known inflammatory effects in the conducting airways caused by MS may be highly contributory to tumors induced by this mode o f exposure, whereas this inflammatory response is in all likelihood absent, or only present to a much lower degree, in ETS-exposed people. Page 2-3, first para, line 3: T he plausible assumptions for misreported smoking habit used in the NRC report should be described here briefly. Page 2-4, last para1Limitations o f the cigarette-equivalent approach due to incomplete knowl edge o f the biological basis o f ETS- and MS-induced lung cancer are pointed out. This is correct, but the same cautionary note should also be applied when the plausibility o f a carcinogenic potential of ETS exposure based on the MS effect is inferred, as mentioned above. Page 3-12, first para: The last two sentences refer to background exposure with regard to the relative risk comparison o f exposed to unexposed individuals. It may be conceivable, however, that background exposure may be quite different between ETS-exposed people (spouses of smokers) and unexposed. The former are more used to being exposed to an air pollutant, but the latter may be more conscious to avoid also exposures at workplace and other places. Thus, the question arises as to whether background exposure can be assumed to be the same between the two groups. Second para: Large differences in controls classified as exposed between the different case-control studies are noted ranging from 15 to 84%. Although it is stated that such study differences do not invalidate statistically testing the correlation between ETS exposure and lung cancer, the 7-8 51246 0074 not invalidate statistically testing the correlation between ETS exposure and lung cancer, the question arises whether such differences may prohibit using these different studies in am eta- analytical approach. Thus, before a meta. analysis (page 3-14) is performed criteria should be formulated to decide which studies fulfill these criteria and should be used in such an Approach (see general comments). Letzel and Uberla (1990, see below) recendy reported on their meta analysis on passive smoking and lung cancer. They insist that every meta analysis has to state its goals, criteria, and methods before it starts. Consequently, they included five criteria to conduct meta analysis o f the available case-control studies under different aspects (Letzel and Uberla, 1990.) Tire result o f their meta analysis, when using different case-control studies and when applying different criteria, is summarised as follows:. Overall risk o f dying o f lung cancer from nonsmoking women married to smoking meh is: 1.074, based on 6 case-control studies o f reasonable quality 1.013, based on 2 prospective studies, using the Hirayama study with the age selection bias removed, as shown by the authors 1.035, based on 2 prospect studies and 6 case-control studies of reasonable quality 1.076, based on 11 studies with the Trichopoulos study excluded 1.118, based on all 12 studies including the Trichopoulos study. They found that these risk estimates are not statistically different from unity. Their results differ from that of Wald etal. (1986) who did not take into account the quality of the individual studies and did not perform an analysis o f sensitivity. Certainly, more discussions on the use and performance o f meta analysis in EPA's Draft is needed. As stated by Letzel and Uberla, "the whole question o f meta analysis comes down to the question o f the quality o f the individual study." Page 3-34, last para: Reference is made here to P.N. Lee's statement that there is a statistically significant association in lung cancer risk in the study o f Hirayama. However, other criticisms, in particular that calling for independent public examination o f Hirayama's raw data, have not been answered (Kilpatrick, 1990.) Page 3-39, first para: Defining direct and indirect passive smoking by a given distance from the source does not necessarily give an accurate picture. Certainly the size o f the room and also the ventilation rate are o f great importance for exposure to passive sm oke. 0075 tn KM) ch wSUWiri Page 3-43, first para: The statement that the statistical results "solidly" support the conclusion o f the association between lung cancer and ETS exposure is not justified based on the individual case-control studies alone. Second para: As mentioned before, Hirayama still needs to answer criticisms made about his study. The statement that "results o f the American cohort study are less conclusive" is a bit misleading, it should rather read that they are conclusive and do not show a correlation. Third para: Likewise, the statement that the American cohort study "weakly" indicates an increased lung cancer risk is not supported by the results and should be changed. Such state ments give the appearance o f a document written with a rather biased opinion. Page 4-13, last para: Comparing SDA cohorts to nonsmoking non-SDA cohorts with respect to lung cancer incidences implies that a possible difference is solely due to differences in ETS exposure. However, it is as likely that other differences in the lifestyles, other workplace expo sures, etc., may contribute or even folly explain such possible differences. i Page 4-15, second para: T he assumption o f the cigarette-equivalent approach is that the lung cancer risks in passive and active smokers are equivalendy indexed by a common measure of exposure to tobacco smoke, for example, use o f nicotine or its metabolite cotinine. This implies: 1) lung tumors should be the same in MS smokers and ETS smokers; 2) nicotine metabolism and dosimetry should be the same in both exposure modes; 3) additional effects o f MS smoke (e.g., bronchitis) do not contribute to lung cancer incidences. Probably, none o f these implica tions is correct as discussed already before. More on respiratory tract dosimetry will be discussed later (see comments on Appendix C). ] j j A more thorough discussion of the differences between MS and ETS dosimetry would be useful and it would be worthwhile, as stated on page 4-16, last paragraph, to further develop the knowledge base surrounding this issue. A further development o f the cigarette-equivalent ap proach would be worthwhile only if it takes into account compound-specific dosimetry and metabolism in lungs o f MS and ETS smokers. Simply taking the particulate concentration or the inhaled particulate mass or cotinine levels as measures of equivalent exposure is probably not indicated. A better justified approach has to be used. 7-10 9 00 Page 4-22, second para: W hen recommending to use average cotinine concentration in smokers and nonsmokers to extrapolate risk from active smoking to passive smoking, a statement or reference to another section o f the document (4-15) should be included which discusses differ ences in dosimetry and metabolism o f nicotine between MS- and ETS-exposed persons, The two final sentences o f this paragraph on the non-negligibility o f a positive excess risk o f 1.02 and 1.07 are not understandable and need to be explained better. I Page 4-34, first para: The aforementioned different opinions regarding lung cancer types that correlated with ETS exposure become obvious here again. This topic definitely needs sme more discussion in the Draft (see Kuller, 1986). ; . . -vv.., Second para: T he dissertation by Varela (1987) has now been published, as mentioned before (Janerich et a l, 1990). I j Page 4-39, last para: Consistency o f response. The simple statement that the two completed | cohort studies observed a higher risk o f lung cancer among female never-smokers, classified as | exposed to ETS, is misleading because it fails to mention in this summarizing statement that this ! higher risk was not significant in the Garfinkel study and, moreover, there was no increase in response with increasing exposure in that study. j Page 5-18, end o f second para: Citing the increased incidence o f respiratory symptoms in adult smokers as a plausible reason for such symptoms to occur in ETS-exposed children raises again the question of comparative dosimetry. Such symptoms relate to the doses to target sites which, as pointed out before, conceivably are quite different between MS and ETS exposure. However, I agree with the conclusion o f the Draft, that passive smoking in early childhood is associated with respiratory illness in children so exposed. If persistent over a longer period o f time, ETS exposure may indeed reduce their rate of pulmonary growth and development, although no causative association between ETS and these effects is assumed in the Draft. Appendices Page A -l: The question comes up here again about the histological type o f lung tumors associ ated with ETS vs MS exposure. As indicated before, a discussion o f this topic should be included ^1246 0077 11 in the main text o f the document. For example, is the higher relative risk found in the study by Gao et al. (1987) for squamous and oat-cell carcinoma indicative o f active smoking, i.e., misclassification? (page A-3, first para). Quite obviously, when evaluating the individual casecontrol studies described here and on the following pages, there is no uniformity with regard to tum or classification by histology. Page A-6, first para; Is "marginal significance" equivalent to being not significant? This question o f tum or type associated with ETS exposure is an important one because there is great onfusion, as can be seen on these pages A-2 through A -14. Differences exist among the authors o f the different studies in what is regarded as typical lung cancer for ETS-exposed people. It appears that peripherally located tumors (adenocarcinoma) are more likely to occur in nonsmol^ers as opposed to centrally-located tumors which are typical for MS smokers. Because an active smoker is definitely also a passive smoker, one would expect to see peripheral tumors in active stnokers, as indeed is the case. If peripheral tumors are most likely to occur in passive smokers, is the occurrence of centrally-located tumors in this group an indication of previous active smjoking? The existing confusion on this issue becomes apparent in Appendix A which describes some o f the individual case-control studies. As stared before, a discussion o f the biology o f lung cancer would be useful before going into the details o f the epidemiological studies. Page C -l: The reasons for inclusion o f Appendix C are not obvious. In my view, they are two fold: 1) to calculate equivalent doses for ETS and MS; 2) to point out differences in dosimetty between ETS and MS. Page C-4, second para: T he example given here is for nicotine attached to particles, as is also the case in an example given toward the end o f this Appendix C. However, because nicotine in ETS is to more than 95% in the gas phase and in MS to more than 95% in the particulate phase, the dosimetries of deposited and absorbed nicotine are likely to be quite different between the two forms o f smoke. Depending on water solubility and reactivity o f nicotine in the gas phase, ETS ! nicotine may be absorbed quite efficiendy in the upper respiratory tract, including the nasal- I pharyngeal passages, while this is quite different for MS nicotine which may more easily travel down into the deep lung, attached to particles. 7-12 51246 0078 H* - 'Vi * '-* .. Page C-5, last para, first sentence: The meaning o f this sentence is unclear. Any inhaled material will deposit onto the walls o f the airways (not lungs). Page C-6, first para: It should be considered also to express the uptake U per unit surface area of the airways which would allow a better comparison between MS and ETS, as will be shown below. Second para: It should also be mentioned in this paragraph that newer models predicting particle deposition in the lungs, including a generation-by-generation deposition, are available. For example, models by Yeh and Schum (1980) and Schum and Yeh (1980) allow to model deposi tion o f different particle sizes subjected to different breathing patterns in the human luhg. In general, it would be useful to point out the different doses involved in a dose-response'relation- ship in the respiratory tract, i.e., inhaled, deposited, retained doses. The deposited dose is not yet equivalent to an uptake, because many o f the particles deposited will actually be eliminated from the lungs again. Uptake into lung cells involves a separate step. i Page C-8, third para: More defined models of tracheal bronchial clearance are available, for example, Yu etal. (1986) and Lee etal., (1989). . Page C-12: This page, dealing with translocation o f inhaled material to systemic organs, needs some rewriting. For example, the statement in paragraph 1 that material deposited in the pulmo nary region of the lung tends to be translocated into the blood stream is only true for dissolved material and not for particles. In the second paragraph, it is suggested that material deposited in the nasal-pharyngeal region is removed primarily to the GI tract with little absorption o f ETS chemicals direedy into the blood stream. However, that depends on whether a chemical is in the gas or particle phase and how well it is adsorbed as a gas onto the walls o f the airways. Absorp tion from the GI tract is different in that absorbed materials pass first through the liver. The statement that this is due to the "proximity" o f the liver should be changed. Page C-13: All arrows connecting compartments should be labeled. The arrow going from lymph nodes to excreta should be changed to going from lymph nodes to blood. Page C-14, last para: The assumption that deposition in the N P region contributes only insig- nificandy to a systemic dose should not be made summarily. Gases like nicotine in ETS could very well be absorbed in this region and contribute to a systemic dose. : 7-13 51246 0 0 7 9 Pages C-18 and C-19: It appears that SS is used as a surrogate for ETS which, as mentioned before, is not correa. It should be pointed out that ETS consists o f diluted SS and exhaled MS. This is an important distinaion in particular with regard to an aging effea o f smoke charaaeristics. Page C-20, last para: References are made to values o f intakes for passive and active smokers, computed and shown in Tables C-3 on page C-22. It is, however, unclear which deposition fraaions for particles and gases have been used for constructing Table C-3. For example, nico tine in ETS in the vapor phase is probably readily absorbed in the nasal-pharyngeal region while, according to page 0 1 2 , there is little absorption in this region. W hich respiration volume, rates, respiratory pauses, etc., were used for computing the data in Table 3 for active and passive smoking? M outh and/or nose breathing? Although data for those parameters are given in the Appendix, it is not clear whether this table has been constructed using those data. Page C-21, second para: This paragraph compares the relative concentrations o f four chemicals in SS and ETS and concludes that these are rather similar for three o f the chemicals between the two different forms o f smoke. However, the question also is whether these chemicals ate in the same phase of smoke, i.e., vapor or particle phase. This would be important for dosimetric cal culations. However, lacking such data, it may be appropriate to use as a first approach data o f SS as the surrogate for ETS after it has been made clear that there is a distinaion between the two. Page C-28, first para: T he percentage o f inhaled MS particles deposited in the lungs (80%) sounds rather high. It has been found to be on the average 47% by Hinds et al. (1983) which would also be more in line with model prediaions on particle deposition. In addition, It is not quite clear what f N P implies for active smokers, there is no nasal component during aaive smoking except possibly for exhaled MS smoke. | Second para: The assumption that there is no hygroscopic growth of ETS particles, based on the study by Hiller et al. (1982), is questionable because Hiller etal. were working with SS, not ETS. I In addition, they did not ensure that the humidity conditions in their sampling bag o f exhaled ! air was the same as that in the respiratory tra a . j 51246 Third para: I performed some model calculations on deposition o f ETS and MS particles, using the EPA data set as well as slightly changed parameters, based on a modified deposition model <s> of Yeh and Schum. T he data are shown in Tables 1-4. The predicted daily deposition is different from what is given in the EPA document, in particular for the pulmonary regions o f the active and passive smoker, i.e., about half o f the EPA-value for the active smoker and about twice for the passive smoker. Again, it is not clear which parameters EPA used in their model, m outh breathing vs. nose breathing, breath holding included, etc. A comparison o f attached Tables 1 and 2 shows that relative depositions throughout the respira tory tract are quite different for ETS and MS particles. For the ETS exposure, nose breathing is assumed and for MS exposure m outh breathing is assumed with a 3-second respiratory'pause. Higher surface area concentrations after inhalation o f MS occur in generations 3 and 4^of the .conducting airways. This is consistent with prevalent sites for lung tumors (Schlesinger'and Lippmann, 1979), i.e., more centrally located tumors. For ETS exposure, relative highest surface area concentrations are achieved in generations 15-18, i.e., the transitional zone o f the lung . - where one would expea to see peripheral lung tumors to occur (bronchiolo-alveolar tumors). However, as pointed out previously, additional effects o f MS exposure on the conduairig airways have to be considered, i.e., bronchitis and impaired mucociliary clearance (see, for example, Vastag e ta l, 1986). Whereas Tables 1 and 2 assume respiratory values as proposed by EPA in their document, Tables 3 and 4 apply more realistic values for tidal volumes (500 cc for ETS exposure and 1,000 cc for MS exposure) and decreased respiratory frequency (12 for M$ expo sure). The result is essentially similar to the previous prediaions in Table 1 and 2, i.e., relatively higher surface area deposition in the upper conduaing airways for MS exposure and relatively higher deposition in the transitional region (deeper lung) for ETS exposure. O ne important question for such dosimetric calculations is whether the active smoker aaually inhales the MS in a bolus with a volume o f 35-50 cc, which is superimposed on the larger tidal volume, or whether the MS is inhaled over the total tidal volume. T he former case would need some additional modeling which is presently not available in the model of Yeh and Schum, which was the basis for Tables 1-4. The paper by Muller e ta l (1990) compares deposition prediaions o f different breathing rates and smoke particle sizes based on a Weibel lung model. However, their hygroscopic growth, assuming that o f NaCl particles, may be too high for MS particles. Additional influences o f a "cloud" effea may have to be considered too (Martonen, 1989) 51246 0081 7-15 J. In any event, the results in Tables 1-4 show that the highest surface area doses are expected to be j differently located between the two exposure modes for MS and ETS and that consequently one | may expect to see differences in tum or induction between active and passive smokers as indeed seems to be implied by several o f the authors o f the epidemiological studies. Again, parameters used in the EPA D raft to model deposition need to be clearly stated so that the reviewer can follow the calculations performed in this exercise. Page C-30, third para: An alveolar retention half-time o f 17 hours for particles in activ$ smokers is assumed, based on results from Black and Pritchard (1984) and the same value is also assumed for passive smokers. T he study by Black-arid Pritihard,* however, is only $ ^prelimliMur|^nft!rence j * report, no detailed data are given, in particular about the persistence of the label that was used i in their study. The fast dissolution of this label in lung tissue has not been demonstrated to be equivalent to MS particle dissolution. Retention half-times for particles in the alveolar region are much longer than 17 hours which represents only the short-lived fraction o f dissolved particle constituents removal from the deep lung. Normal alveolar retention half-times for particles are on the order o f 300-400 days (Bailey e ta l, 1985). In active smokers the retention half-iime is increased (Bohning e ta l, 1982; Cohen etal., 1979). Therefore, the retention time in the active smoker may even be longer than 300-4000 days. A t any rate, the assumed 17 hours is, at present, not acceptable as an alveolar retention half-time for deposited smoke particles in this region. Consequently, values given in Table C-5 on page C-32 on particulate-phase chemicals and daily interval organ burdens could be considerably different because they are based on the very short retention half-time o f 17 hours, measured for one label only. Page C-33, second para: A discussion o f nicotine absorption for active and passive smokers is included here. If nicotine in the gas phase (passive smoker) is mostly taken up in the upper respiratory tract, then there is no need to model a time for penetrating alveolar cells in the passive smoker, as done in this paragraph. Furthermore, the calculated nicotine dose to the pulmonary region is based on the assumed retention half-time o f 17 hours mentioned before. In addition, both tracheobronchial and pulmonary uptake of nicotine are based on particulate deposition and retention which, as mentioned above, does not apply to gas phase nicotine in ETS. Thus, the comparison made here is not valid. The numbers given in this paragraph need additional clarification. 7-16 51246 0082 Page D -4, first line: T he study by Grimmer et al. (1988) was dealing with SS not ETS. This is rectified in the listed disadvantages of the Grimmer study on page D-8, first line> but should also be made clear here. Page D-8: It has to be kept in mind that the relative potency approach (data shown in Table D - 4) implies that ETS and MS act in the same way with regard to tum or induction, an assumption which can be challenged due to the additional effects o f MS caused in the respiratory tract at sites where tumors are induced, i.e,, in the upper conducting airways. Thus, the relative! potency estimates have to be used with caution. r Summarizing Remarks EPA's Risk Assessment Guidelines o f 1986 state that an agent should be classified as Group A H um an Carcinogen when there is sufficient evidence from epidemiologic studies to support a causal association between exposure to the agent and cancer. The Draft categorizes ETS as a Group A carcinogen based on the weight o f evidence from results o f the epidemiological studies which is supported by the result o f the meta analysis performed by EPA, However, the meta analysis performed in the Draft does not conform to scientifically accepted standards. Because of the highly variable outcome o f the different epidemiological studies they do not support a statement of sufficient evidence for a causal association between lung cancer and passive smoking as is required for classification o f a G roup A carcinogen. Rather, this evidence should be termed limited. This would be in line with the evaluation o f IARC which stated in 1987 that the avail able epidemiological evidence is compatible with either th presence or absence o f lung tancer risk. The strongest evidence that ETS may indeed have a carcinogenic potential is given by the large Japanese cohort study (Hirayama) and by a recendy published study o fJanerich etal. In view o f the negative results o f other epidemiological studies, in particular the large American cohort study (Garfinkel), it appears that a carcinogenic potential o f ETS may only be detectable under exposure conditions resulting in a suificiendy high dose. This means, on the other hand, that the carcinogenic potency o ( ETS depends very much on specific exposure conditions, a conclusion which has to be considered for quantitative risk assessment. Thus, although the possibility cannot be ruled out that ETS is a pulmonary carcinogen, collec tively, the results o f the epidemiological studies do not support the conclusion o f EPA's Draft 51246 0083 7-17 s.*ltiiiXii i'ii 1 U.i k. .r...* S\r`A: o f sufficient evidence for a carcinogenic potential. In view o f the difficulties to express the dose received by a passive smoker from the smoking o f one cigarette (exposure o f a spouse to ETS from one cigarette in a Japanese house is not equivalent--in terms o f delivered dose--to expo sure o f a spouse to ETS from one cigarette in an American home) it is difficult, perhaps even impossible, at present to perform a meaningful quantitative risk assessment for ETS exposure. W ith respect to the biological plausibility of a carcinogenic effect o f ETS, which is bas^d on the well accepted causal relationship between active smoking and lung cancer, it needs to be empha sized that there are differences between MS and ETS, and furthermore that ETS is notjequivalent to SS. Aging and different gas- and particle phase composition are only one aspect o f the differences, and differences in respiratory tract dosimetry are additional ones because nasopha ryngeal, bronchial and deep iung deposition o f ETS and MS particles are quite different. Such differences also make the cigarette equivalent approach questionable. Because such dosimetric differences could explain a shift in tum or location from more central (MS) to more peripheral (ETS) in the respiratory tract, additional attention needs to be given to these issues in the final EPA document . This includes also the pathogenetic importance o f chronic bronchial inflammatory condition in the active smoker which is in all likelihood absent in the passive smoker and which makes an extrapolation o f effects o f MS exposure to those o f ETS exposure questionable if this extrapolation is based on the plausibility of expected similar effects. Furthermore, there is also evidence that organic carcinogens inhaled together with high inhaled particle concentrations may contribute very little or not at all to a tumorigenic effect. Thus, while the plausibility o f a carcinogenic effect o f ETS based on the known carcinogenic effect o f MS exposure is an attractive assumption it needs further discussion o f relevant underlying principles. 51246 0084 7-18 HUMAN MODEL t i t IOO Z N asal B reath in g *** P a r tic le s iz e (eicro n s) * .15 t i l Sigaa 6 * 1.5 m T id al Voluae (c a A3 ) 750 i n R e sp ira to ry Frequency * 15 m B reath-holding pause (s e c .) * 0 i n A erosol C on cen tration * . 2 a g /a A3 h i Date: 01-17-1991; T iie: 16:58:44 Table 1 ETS (EPA data) Nose Breathing \ G en era tio n Nuaber F ra ctio n a l D eposition Surface Area Surface Area at FRC Z of inhaled Z o f inhaled Hicrograas Aerosol depos Aerosol depos Deposited it e d per c a A2 it e d per c a ^ in 8 hr At FRC > Nicrograas o f Aerosol : D eposited per FRC caA2 in 8 hr. . 1 2 .0 1 2 E -0 4 5 3 .9 7 50.75 3.728E-04 3 .9 6 4 E -0 4 2 .1 7 3 E -0 1 4 .2 8 1 6 -0 3 i2 1 .9 5 0 E -0 4 32.88 31.12 5 .9 2 9 E -0 4 6.2656-04 2 . 106E-01 6 .7 6 6 E -0 3 3 1 .7 2 0 E -0 4 19.51 18.46 8 .8 1 6 E -0 4 9 .3 1 7 E -0 4 1.857E-01 1 .0 0 6 E -0 2 4 1 .767E-04 11.51 11.14 1.5356-03 1 .5 8 7 E -0 3 1.908E-01 1.7 HE-02 5 2 .7 B 0 E -0 4 19.27 18.59 1 .4 4 2 E -0 3 1 .4 9 5 E -0 3 3 .0 0 2 E -0 1 1 .6 1 5 E -0 2 6 4.3316-04 41.69 39.65 1 .0 3 9 E -0 3 1 .0 9 2 E -0 3 4 .6 7 7 E -0 1 1 .1 8 0 E -0 2 7 6.204E-04 48.41 46.52 I .2816-03 1 .3 3 4 E -0 3 6.700E-01 1.440E-02 8 1 .0 2 5 E -0 3 102.20 97.27 1 .0 0 3 E -0 3 1 .0 5 4 E -0 3 1.107E+00 1 .1 3 8 E -0 2 9 1 .4 9 2 E -0 3 176.56 166.87 8 .4 5 0 E -0 4 8 .9 4 1 E -0 4 1.611E+00 9 .6 5 6 E -0 3 10 1.9426-03 193.96 184.68 1.0016-03 I . 0516-03 2.0976+00 1.1356-02 11 2 .6 7 3 E -0 3 267.10 253.12 1 .001E-03 1 .0 5 6 E -0 3 2.887E+00 1 .1 4 0 E -0 2 12 3 .S 8 4 E -0 3 341.10 323.04 1 .0 5 1 E -0 3 1.1096-03 3.870E+00 1.1986-02 13 4 .8 1 2 E -0 3 414.83 392.90 1 .1 6 0 E -0 3 1 .2 2 5 E -0 3 5.197E+00 1 .3 2 3 E -0 2 14 6 .2 7 2 E -0 3 498.56 472.01 1 .2 5 8 E -0 3 1 .3 2 9 E -0 3 6.7736+00 1 .4 3 5 E -0 2 15 8 .1 8 1 E -0 3 587.99 558.02 1 .3 9 1 E -0 3 1 .4 6 6 E -0 3 8.836E+00 1 .5 8 3 E -0 2 16 1 .0 8 5 E -0 2 792.81 756.21 1.3696-03 1 .435E-03 1,1726+01 1 .5 5 0 E -0 2 17 1 .2 4 9 E -0 2 819.44 665.65 1 .5 2 4 E -0 3 1 .8 7 6 E -0 3 1.349E+01 2 .0 2 7 E -0 2 18 1 .778E-02 1361.01 1113.50 1.3066-03 1.5966-03 1.920E+01 1.7246-02 19 2 .2 4 0 E -0 2 1870.95 1530.70 1.1986-03 i . 464E-03 2.420E+01 1 .5 8 1 E -0 2 20 2 .9 7 9 E -0 2 3020.39 2471.11 9 .8 6 2 E -0 4 1 .2 0 5 E -0 3 3.217E+01 1 .3 0 2 E -0 2 21 3 .7 5 5 E -0 2 4958.69 4056.91 7.572E-04 9.255E-04 4.055E+01 9 .9 9 6 E -0 3 22 4 .2 6 1 E -0 2 8781.65 7184.65 4.853E-04 5 .9 3 1 E -0 4 4.6026+01 6 .4 0 6 6 -0 3 23 3 .4 4 5 E -0 2 14723.03 12045.55 2 .3 4 0 E -0 4 2.860E-04 3.720E+01 3 .0 8 8 E -0 3 24 6 .2 1 2 E -0 3 33498.67 27406.72 1 .8 5 4 E -0 5 2 .2 6 7 E -0 5 6.7096+00 2 .4 4 8 E -0 4 25 0.000E+00 281286.70 490541.70 0.0006t00 0.000E+00 0.0006+00 O.OOOE+OO P e rc e n t d e p o s ite d in tra c h e o b ro n c h ia l P e rc e n t d e p o s ite d in pulm onary re g io n E x tra th o ra c ic d e p o s itio n (p e rc en t) re g io n 4.290802 20.32789 * 0 Mass (m icrogram s) d e p o s ite d E x tra th o ra c ic - T rach eo b ro n chial Pulm onary - in a irw a ys 0 .0 0 4 6.3 4 2 19 .54 d u rin g 8 hr: 51246 0085 7-19 VA/ ,A..C' U -.,.'. - *** *** *** *** *** *** *** *** *** '-- - HUMAN MODEL O V. N a s a l B r e a t h i n g \ \ P a r t i c l e s i z e ( m i c r o n s ) *= 1 . 4 S ig m a = 1 .5 T id a l V o lu m e (c m A3 ) = 7 5 0 \ R e s p i r a t o r y F r e q u e n c y = 1 5 B r e a th -h o ld in g p a u se ( s e c . ) = 3 A e r o s o l C o n c e n t r t i o n 1 5 mg/m^3 D a te s 0 1 - 1 7 - 1 9 9 1 ; T im es 1 7 : 0 0 : 4 1 Tabic 2 MS (EPA data) M outh Breathing G en eration Nuaber F r a c tio n a l D eposition Surface Area 1 1 .7 6 2 E -0 4 53.97 2 2 .8 7 6 6 -0 3 32.88 0J 3 .2 3 5 E -0 3 19.51 4 2 .7 4 1 E -0 3 11.51 5 2 .8 3 S E -0 3 19.27 6 2 .5 0 2 E -0 3 41.69 7 2 .3 7 7 E -0 3 48.41 8 3 .0 0 7 E -0 3 102.20 9 3 .4 7 7 E -0 3 . 176.56 10 3 .1 5 9 E -0 3 193.96 11 4 .2 2 2 E -0 3 267.10 12 4 .9 5 1 E -0 3 341.10 13 5 .7 1 7 E -0 3 414.83 14 7.198E-03 498.56 IS 7 .9 3 4 E -0 3 587.99 16 9 .7 8 5 E -0 3 792.81 17 9 .2 2 3 E -0 3 819.44 18 I.2 3 7 E -0 2 1361.01 19 1 .5 7 0 E -0 2 1870.95 20 2 .3 9 0 E -0 2 3020.39 21 3 .8 2 1 E -0 2 4958.69 22 6 .3 8 4 E -0 2 8781.65 23 1 .0 0 0 E -0 1 14723.03 24 S .1 4 8 E -0 2 33498.67 25 O.OOOE+OO 2 8 1 286.70 Surface Area at FRC 5 0 .7 5 31.12 18.46 11.14 18.59 39.65 46.52 97.27 166.87 184.68 253.12 323.04 392.90 472.01 558.02 756.21 665.65 1113.50 1530.70 2471.11 4056.91 7184.65 12045.55 27406.72 490541.70 Z of inhaled Z of inhaled Nicrograas Aerosol depos Aerosol depos Deposited it e d per c e A2 it e d per c a A2 in 8 hr At FRO . . Nicrograas; of D ep o sited p er FRO ca A2 | in 8 hr. 3.265E-04 8.7466-03 1 .6 5 8 E -0 2 2.3816-02 1 .4 7 1 E -0 2 6 .0 0 1 E -0 3 4.909E-03 2.943E-03 1 .9 6 9 E -0 3 I.6 2 9 E -0 3 1 .5 8 1 E -0 3 1 .4 5 2 E -0 3 1 .3 7 8 E -0 3 1 .4 4 4 E -0 3 1 .3 4 9 E -0 3 1 .2 3 4 E -0 3 1.1266-03 9 .0 9 0 E -0 4 B.389E-04 7 .9 1 1 E -0 4 7 .7 0 5 E -0 4 7 .2 7 0 E -0 4 6 .7 9 4 E -0 4 1.537E-04 0.0006+00 3 .4 7 3 E -0 4 9 .2 4 1 E -0 3 1.753E-02 2.4616-02 1 .5 2 5 E -0 2 6.3096-03 5 .1 0 9 E -0 3 3 .0 9 2 E -0 3 2.0B4E-03 1.7116-03 1.668E-03 1 .5 3 3 E -0 3 1 .4 5 5 E -0 3 1 .5 2 5 E -0 3 1 .4 2 2 E -0 3 1 .2 9 4 E -0 3 1 .3 8 6 E -0 3 1.1116-03 1 .0 2 5 E -0 3 9 .6 7 0 E -0 4 9.417E-04 8.8866-04 8 .3 0 5 E -0 4 1 .8 7 9 E -0 4 O.OOOE+OO 1.427E+01 2.330E+02 2.620E+02 2.220E+02 2.297E+02 2.027E+02 1.925E+02 2.436E+02 2.817E+02 2.559E+02 3.419E+02 4.0116+02 4.631E+02 S.830E+02 6.427E+02 7.926E+02 7.471E+02 1.002E+03 1.2716+03 1.936E+03 3.095E+03 5.1716+03 8.103E+03 4.170E+03 0.000E+00 1 2.813E-01 7,4851+00 1.420E+01 1.994E+01 1.235E+01 5.1106+00 4.13BE+00 2.5046+00 1.68BE+00 1.386E+00 1.3516+00 1.242E+00 I.179E+00 1.235E+00 1.1526+00 1.048E+00 1.122E+00 , 8 .9 9 9 E -0 1 8 .3 0 6 E -0 1 7.8336-01 7 .6 2 8 E -0 1 7 .1 9 8 E -0 1 6 .7 2 7 E -0 1 1.5226-01 0.000E+00 P e r c e n t d e p o s it e d in t r a r h e o b r o n c h ia l P e r c e n t d e p o s i t e d in p u lm o n a ry r e g io n E x tra th o ra cic d ep o sitio n (p e rcen t) r e g i o n * 6 . 6 1 9 3 6 4 a 3 1 .4 7 5 0 8 8 .3 6 M ass ( m ic r o g r a m s ) d e p o s i t e d in a i r w a y s d u r in g 8 hr : E x tra th o ra cic - 6 7 7 1 .6 0 T racheobronch ial - 5 3 6 1 .6 8 Pul mona r y - 2 5 4 9 4 .8 2 | !i 1 9 8 0 0 9PZis M t i JUii l ju jltu vViW &ijiii. ;. , _ *** *** *** *** *** *** *** *** *** HUMAN MODEL 100 V. N a s a l B r e a t h i n g \ P a r t ic le s iz e (m icro ns) * .1 5 Sigm a Q * 1 .5 T i d a l V o lu m e Ccm^S) 5 0 0 R e s p ir a to r y F requ en cy 15 B re a th -h o ld in g pause (s e c .) = 0 A e ro s o l C o n c e n t r a tio n .2 mg/m^3 D a te : 0 1 -1 7 -1 9 9 1 ? Tim e: 1 7 :0 2 :2 4 Table 3 ETS (EPA data lowerTV) Nose Breathing ! j j | l i ` l'';t*' G en eration Nuaber F r a c tio n a l D eposition Surface Area Surface Area a t FRO Z of inhaled Z of inhaled Hicrograas Aerosol depos Aerosol depos D eposited it e d per ea A2 it e d per c a A2 in 8 hr At FRC > Hicrograas c f D ep o sited pc r : FRC ca A2 in 8 hr. 1 2 .6 0 4 E -0 4 5 2 .9 2 5 0 .7 5 4 .9 2 1 E -0 4 5 . 131E-04 1.875E-01 i 3.694E-03 2 2 .4 0 4 E -0 4 32.31 31.12 7 .4 4 1 E -0 4 7 .7 2 5 E -0 4 1.731E-01 : S.562E-03 3 2 .0 8 0 E -0 4 19.17 18.46 1.085E-03 1 .1 2 7 E -0 3 1 .4 9 7 E -0 1 B.112E-03 4 2 .1 4 9 E -0 4 11.39 11.14 1 .8 8 6 E -0 3 1 .9 3 0 E -0 3 1.547E-01 1 .3 9 0 E -0 2 S 3 .4 1 3 E -0 4 19.06 18.59 1 .7 9 1 E -0 3 1.836E-03 2 .4 5 8 E -0 1 1.322E-02 6 5 .3 6 5 E -0 4 41.03 39.65 1 .3 0 7 E -0 3 1 .3 5 3 E -0 3 3 .8 6 3 E -0 1 9 .7 4 1 E -0 3 7 7 .6 8 4 E -0 4 47.81 46.52 1 .6 0 7 E -0 3 1.652E-03 5 .5 3 2 E -0 1 1 .1 8 9 E -0 2 8 1 .2 6 3 E -0 3 100.60 97.27 1 .2 5 6 E -0 3 1 .2 9 9 E -0 3 9 .0 9 6 E -0 1 9.3528*03 9 1 .B22E-03 173.41 166.87 1 .0 5 1 E -0 3 1 .0 9 2 E -0 3 1.312E+00 7 .8 6 3 E -0 3 10 2 .3 4 4 E -0 3 190.96 184.68 1 .2 2 8 E -0 3 1 .2 6 9 E -0 3 1.688E*00 9 .1 3 8 E -0 3 11 3 .1 7 9 E -0 3 262.56 253.12 1.21IE-03 1 .2 5 6 E -0 3 2.289E+00 9 .0 4 2 E -0 3 12 4 .1 9 6 E -0 3 335.23 323.04 1 .2 5 2 E -0 3 1 .2 9 9 E -0 3 3.021Ef00 9 .3 5 3 E -0 3 13 5 .5 4 4 E -0 3 407.70 392.90 1 .3 6 0 E -0 3 1 .4 1 1 E -0 3 3.991E+00 1 .0 1 6 E -0 2 14 7 .I0 9 E -0 3 489.93 472.01 1 .4 5 1 E -0 3 1 .5 0 6 E -0 3 S . 118E-00 1 .0 8 4 E -0 2 15 9.1I8E -03 578.27 558.02 1 .5 7 7 E -0 3 1 .6 3 4 E -0 3 6.565E+00 1 .1 7 6 E -0 2 16 1 .1 8 7 E -0 2 780.77 756.21 1 .5 2 0 E -0 3 1.S69E-03 8.545E*00 1 .1 3 0 E -0 2 17 1 .3 1 5 E -0 2 769.93 665.65 1 .7 0 8 E -0 3 1 .9 7 5 E -0 3 9.467E+00 1 .4 2 2 E -0 2 18 1 .8 2 8 E -0 2 1281.20 1 1 1 3 .SO 1 .4 2 7 E -0 3 1 .6 4 2 E -0 3 1.3166+01 1 .1 8 2 E -0 2 19 2 .2 2 7 E -0 2 1761.23 1530.70 1.265E-03 1 .4 5 5 E -0 3 1.604E+01 1 .0 4 8 E -0 2 20 2 .8 0 2 E -0 2 2843.27 2471.11 9 .8 5 6 E -0 4 1 .1 3 4 E -0 3 2.018E+01 8 .1 6 5 E -0 3 21 3 .1 7 5 E -0 2 4667.90 4056.91 6 .8 0 3 E -0 4 7.B27E-04 2.286E+01 5 .635E -03 !, 22 2 .6 8 9 E -0 2 8266.68 7184.65 3 .2 5 3 E -0 4 3 .7 4 3 E -0 4 1.936E+01 2.695E-03 23 6 .0 5 6 E -0 3 13859.65 12045.55 4 .3 6 9 E -0 5 5 .0 2 7 E -0 5 4.360E+00 3 .6 2 0 E -0 4 24 O.OOOEtOO 31534.26 27406.72 0.000E+00 O.OOOEtOO O.OOOE+OO O.OOOEfOO 25 0 . 000E+00 264791.60 490541.70 0.000E+00 0 . 000E+00 0.000E+00 0 . 000E+00 P e r c e n t d e p o s i t e d in t r a r h e o b r o n c h ia l P e r c e n t d e p o s it e d in p u lm o n a ry r e g io n E x t r a t h :>r a c i c d e p o s i t i o n ( p e r c e n t ) r e g i o n sa 4 .9 0 1 3 9 1 1 4 .6 4 2 1 =0 1 ! M ass (m icrogram s) d e p o s ite d E x tra th o ra cic T rach eob ron ch ial _ Pulm onary in a ir w a y s d u r in g 8 h r: 0 . OO OOBJP 2 9 1 0 5 .4 2 1 i 1 I 51246 0087 *** *** *** *#* *** *** *** *** HUMAN MODEL 0 V. N a s a l B r e a t h i n g P a r t ic le s iz e (m icrons) = 1 .4 Sigma Q = 1.5 T id a l Volum e <cm^3) = 1000 R e s p i r a t o r y F re q u e n c y " 12 \ B re a th -h o ld in g pause (s e c .) = 3 A e r o s o l C o n c e n t r a t i o n =| 1 5 m g / m ^ S D a te : 0 1 -1 7 -1 9 9 1 ; Tim e: 1 7 :0 4 :0 0 Table 4 MS (EPA data, largerTV, lower resp.) M outh Breathing L> G en eration Nueber T ra ctio n a l D eposition Surface Area Surface Area a t FRO Z of inhaled Z o f inhaled Nicrograas Aerosol depos Aerosol depos D eposited it e d per c a A2 it e d per c a A2 in 8 hr At FRO Nicrograas of Aerosol D eposited f er FRO ca A2 in 8 hr. 1 2 .1 7 4 E -0 4 55.00 50.75 3 .9 5 3 E -0 4 4 .2 8 4 E -0 4 1.878E+01 ! 3.701E -01 2 2 .0 4 3 E -0 3 33.44 31.12 6 .1 0 9 E -0 3 6 .5 6 4 E -0 3 1 .765E +02 5.6716*00 3 2 .1 1 3 E -0 3 19.84 18.46 1 .0 6 5 E -0 2 1 .1 4 5 E -0 2 1.826E+02 9.893E+00 4 1 .788E-03 11.62 11.14 1 .538E-02 1 .6 0 6 E -0 2 l.S45Ee-02 1.3886+01 5 1 .9 7 0 E -0 3 19.48 18.59 1 .0 1 1 E -0 2 1 .0 6 0 E -0 2 1.702E+02 9.154E+00 6 1 .793E-03 42.34 39.65 4 .2 3 6 E -0 3 4 .5 2 2 E -0 3 1.549E+02 3.907E+00 7 1.750E-03 48.99 46.52 3 .5 7 2 E -0 3 3 .7 6 1 E -0 3 1.512E+02 3.250E+00 8 2 .2 7 6 E -0 3 103.75 97.27 2 . 193E-03 2 .3 4 0 E -0 3 1.9666*02 2.0226+00 9 2 .7 4 4 E -0 3 179.65 166.87 1 .5 2 7 E -0 3 1 .6 4 4 E -0 3 2.371E+02 1.421E+00 10 2 .6 5 7 E -0 3 196.87 184.68 1 .3 4 9 E -0 3 1 .4 3 8 E -0 3 2.295E+02 1.2436+00 11 3 .5 1 2 E -0 3 271.53 253.12 1 .2 9 4 E -0 3 1 .3 8 8 E -0 3 3.035E+02 1.199E+00 12 4 . 199E-03 346.84 323.04 1 .2 1 1 E -0 3 1 .3 0 0 E -0 3 3.628E+02 1.1236+00 13 4 .9 0 5 E -0 3 421.79 392.90 1 .163E-03 1 .2 4 8 E -0 3 4.238E+02 1.079E+00 14 6 .3 0 5 E -0 3 506.99 472.01 1 .2 4 4 E -0 3 1 .3 3 6 E -0 3 5.447E+02 1.1546+00 IS 7 .0 9 3 E -0 3 597.46 558.02 1.1B7E-03 1 .2 7 1 E -0 3 6 . 128E+02 1.098E+00 16 8 .8 3 4 E -0 3 804.67 756.21 1 .0 9 8 E -0 3 1 .1 6 8 E -0 3 7.6326*02 1.0096+00 17 8 .8 4 5 E -0 3 867.46 665.65 1 .0 2 0 E -0 3 1 .329E-03 7.642E+02 1.148E+00 18 1 .239E-02 1438.55 1113.50 8.612E-04 1 .1 1 3 E -0 3 1.070E*03 9 .6 1 3 E -0 1 19 1.612E-02 1977.53 1530.70 8 . 149E-04 1 .0 5 3 E -0 3 1.392E+03 9 .0 9 6 E -0 1 20 2 .4 7 1 E -0 2 3192.45 2471.11 7 .7 4 0 E -0 4 1 .0 0 0 E -0 3 2.1356*03 8.6406-01 21 3 .9 1 8 E -0 2 5241.18 4056.91 7.475E-04 9 .6 5 8 E -0 4 3.3856+03 8 .3 4 4 E -0 1 22 6 .3 9 7 E -0 2 9281.93 7184.65 6.892E-04 8 .9 0 4 E -0 4 5.527E+03 7 .6 9 3 E -0 1 23 9 .6 3 0 E -0 2 15561.79 12045.55 6 . 188E-04 7 .9 9 5 E -0 4 8.3206+03 6.907E-01 i 24 I .1666-01 35407.05 27406.72 3 .2 9 2 E -0 4 4.253E-04 1.007E+04 3.675E-01 ! 25 0 . 000E+00 297311.40 490541.70 0.000E+00 0 . 000E+00 O.OOOE+OO 0.000E+00 | 51246 0088 P e rc e n t d e p o s ite d in t r a rh e o b ro n c h ia l P e rc e n t d e p o s ite d in pulm onary re g io n E x t r a th or ac i c d ep o s i t i on ( p e r c e n t ) r e g i o n = 5 . 4 1 9 9 1 4 = 37.80811 6 .9 53 33 4 Mass (m icro gram s) d e p o s ite d in a irw a y s E x t r a t h o r a c ic - >007 . 6 8 T ra c h e o b r o n c h ia l - 4682. 81 P ulm onary - 32666 21 d u rin g 9 hr: | i References 1. Bailey, M.; Fry, F.; James, A. (1985): Long-term retention o f particles in the human respiratory tract. J. Aerosol Sci. 16s. 295-305. 2. Bangert-Downs, R.L. (1986): Review of developments in meta analytical methods. Psychology Bulletin 99* 388-399. 3. Black, A. and Pritchard, J.N . (1984): A comparison o f the regional deposition and short term clearance o f tar particulate material from cigarette smoke, with that o f 2.5 pm polystyrene microspheres. J. Aerosol Science 15s 224-227. ^ '.'v . 4. Bohning, D .; Atkins, H .; Cohn, S. (1982): Long-term particle clearance in man: normal and impaired. Ann. Occup. Hyg. 26s, 259-271. 5. Bond, JA .; Sun, J.D .; Mitchell, C.E.; Dutcher, J.S.; Wolff, R.K. and McClellan, R.O. (1986): Biological fate o f inhaled organic compounds associated with particulate matter. In: Aerosols, pp. 570-592. eds. Lee, Schneider, Grant, Verker. Lewis Publishers, Chelsea, MI. 6. Cohen, D.; Arai, S.F.; Brain, J.D. (1979): Smoking impairs long-term dust clearance from the lung. Science 204:514-517. 7. EPA: T he Risk Assessment Guidelines o f 1986. EPA/600/8-87/045 August 1987. 8. Gao, Y.; Blot, W.J.; Zheng, W.; Ershow, A.G.J Hsu, C.W.; Levin, L.I.; Zhang, R.j Fraumeni, J.F. (1987): Lung cancer among Chinese women. Int. J. Cancer 40* 604-609. 9. Garfmkel, L. (1984): Passive smoking and cancer: American experience. Prev. Med. 13:691-697. 10. Grimmer, G.; Brune, H . ; Dettbarn, G.; Naujack, K.W.; Mohr, U. and Wetzel-Hartung, R. (1988): Contribution o f polycyclic aromatic compounds to the carcinogenicity o f sidestream smoke o f cigarettes evaluated by implantation into the lungs o f rats. Cancer Letters 43: 173-177. 11. Hiller, F.; McCusker, K.; Mazumder, M.; Wilson, J.; Bone, R. (1982): Deposition o f side stream cigarette smoke in the human respiratory tract. Am. Rev. Respir. Dis. 125:406-408. 12. Hirayama, T. (1981): Nonsmoking wives o f heavy smokers have a higher risk o f lung cancer: A study from Japan. British Med. J. 282:183-185. 13. Hinds, W.; First, M .W .; Huber, G.L.; Shea, J.W. (1983): A method for measuring respira tory deposition o f cigarette smoke during smoking. Am. Ind. Hyg. Assoc. J. 44(2): 113-118. 14. IARC International Agency for Research on Cancer (1989): Monographs on the Evaluation o f the Carcinogenic Risk o f Chemicals to Humans. Vol. 46. Diesel and Gasoline Engine Exhaust and some Nitroarenes; IARC,Lyon. V'"-1 15. Janerich, D .T.; Thom pson,W .D .;V arela,L.R .;G reenw ald, P.; Chorost, S.; Tucci, C.; Zam en.M .B.; Melamed, M.R.; Kiely, M. and McKneally, M.F., (1990): Lung cancer and exposure to tobacco smoke in the household. N . Engl. J. Med. 323:623-636. 16. Kilpatrick, S.J. (1980) Model specification effects in ETS/nutrition research, in: Kasuga, H . (ed): Indoor A ir Quality. Springer-Verlag, Berlin, Heidelberg, pp. 256-271. 17. Kuller, L.H.; Garfinkel, L.; Correa, P.j Haley, N .; Hoffmann, D.; Preston-Martin, S.; Sandler, D . (1986): Contribution o f passive smoking to respiratory cancer.; Environ. Hlth. Persp., 70:57-69. 18. Lee, P.S.; Gerrity, T.R.; Hass, F.J.; Lourenco, R.V. (1979): A model for tracheobronchial clearance o f inhaled particles in man and a comparison with data. IEEE Trans, on biomed. Engin. Bme 26(11): 624-630. 19. Letzel, H. and Oberla, K. (1980): M eta analysis on passive smoking and lung cancer. In: Kasuga, H . (ed) Indoor A ir Quality. Springer Verlag, Heidelberg, pp. 316-322. 20. Martonen, T .; Hofmann, W . and Bal&hizy, I. (1989): Physical factors affecting lung deposition o f cigarette smoke (with syncarcinogenic radon progeny effects) and mineral fibers. In: W ehner, A.P. and Felten, D.L. (eds): BiologicalInteraction o fInhaled M ineral Fibers and Cigarette Smoke. Battelle Press, Columbus, O H , pp. 159-181. 21. Muller, W.J.; Hess G.D.; Scherer, P.W. (1990): A model o f cigarette smoke particle deposition. Am. Ind. Hyg. Assoc. J. 51(5): 245-256. 7-24 22. NRC. National Research Council (1986): Environmental Tobacco Smoke: Measuring Exposures and Assessing Health Effects. National Academy Press, W ashington, D C. 23. Schlesinger, R.B.; Lippmann, M. (1970): Selective particle deposition and bronchogenic carcinoma. Environ. Res. 15:424-431. 24. Sun, J.D.; Wolff, R.K.; Kanapilly, G.M . and McClellan, R.O. (1984): Lung retention and metabolic fate of inhaled benzo-a-pyrene associated with diesel exaust. Toxicol. Appl. Pharmacol. 73:48-59. '' . - j 25. U.S. Surgeon General (1986): The Health Consequences o f Involuntary Smoking. ; U.S. Dept, o f Health and Hum an Services, Rockville, M D. ^ :j 26. Wald, N.J.; Nanchahal, K.; Thompson, S.G. and Cuckle, H.S. (1986): Does breathing other peoples' tobacco smoke cause lung cancer? British Med. J. 293:1217-1222. 27. Wolff, R.K.; Bond J A .; Sun, J.D.; Henderson, R.F.; Harkema, J.R.; Griffith, W .C.; Mauderly, J.L.; McClellan, R.O. (1989): Effects o f adsorbtion o f benzo-a-byrene onto carbon black particles on levels o f DNA adducts in lungs o f rats exposed by inhalation. Tox. Appl. Pharma., 97:289-299. 28. Varela, L.R. (1987): Assessment of the association between passive smoking and lung cancer. Dissertation to Yale University for Ph.D. degree. 29. Vastag, E.; Matthys, H.; Zsamboki, G.; Kohler, D.j Daikeler, G. (1986): Mucociliary Clearance in Smokers. Eur. J. Respir. Dis. 69: 107-113. 30. Yeh, M.; Schum H -C (1980): Theoretical evalutation o f aerosol deposition in anatomical models of mammalian lung airways. Bull. Math. Biol., 42:1-5. 31. Yu, C.P.; H u, J.P.; Yen, B.M.; Spektor, D.M.; Lippmann, M . (1986): Models for mucociliary particle clearance in lung airways. In: Aerosols, Ch. 39, pp. 569-578. Eds.: S.D. Lee, T . Schneider, L.D. G rant and P.J. Verkerk. Lewis Publishers, Chelsea, MI. 51246 0091 7-25