Chairman Brown: Let us now take up environmental tobacco smoke. And it has been nominated by Review Group 1, recommended for listing in the Ninth Report as a known human carcinogen. And Dr. Bucher.

Dr. Bucher: Environmental tobacco smoke, as you all are pretty well aware, is a complex mixture of gases and particles comprised of sidestream smoke from the burning cigarette, pipe or cigar tip, mainstream smoke that is not inhaled by the smoker, and exhaled smoke. Approximately half of tobacco smoke by weight is sidestream smoke. And it has been estimated that ETS, or environmental tobacco smoke, contains over 4,000 chemicals, a number of which are a known to be carcinogenic and toxic.

DOWNLOAD THE DOCUMENT IN WORD 6.0 FORMAT

The nomination for environmental tobacco smoke is that it be listed as known to be a human carcinogen in the Report on Carcinogens. The source of the nomination, as you have just heard, was the NIEHS Report on Carcinogens Review Group 1. And those of you who were here last year know that active tobacco smoking was considered by this review group and was considered to be recommended as listed in the Report on Carcinogens as a known human carcinogen.

It was recognized by Review Group 1 at that time that there was a considerable body of literature on environmental tobacco smoke and that there were considerable differences and that it would probably be worthwhile to consider this

PAGE 170:

entity separately, and I think that has been borne out.

The rationale for the nomination was previous evaluations of the data on passive smoking or environmental tobacco smoke by IARC in 1986. There has also been a report by the U. S. Surgeon General issued in 1986. There was a report that is not on here issued by the National Research Council also in 1986. There has been a major review of environmental tobacco smoke carried out by the USEPA and published in 1992. And most recently there has been a very extensive evaluation of the health effects of environmental tobacco smoke published by the California EPA in 1997.

Concerning experimental carcinogenesis, there is sufficient evidence that inhalation of tobacco smoke as well as topical application of tobacco smoke condensates causes cancer in experimental animals. As you probably are aware, it has been kind of difficult to generate positive responses in experimental animals to tobacco smoke in inhalation studies. These animals do not care to breathe the smoke. They have difficulty with this. It is fatal in rather low concentrations in animals. There have been a couple of studies that have specifically addressed either sidestream smoke or more recently environmental tobacco smoke itself. A study by--I can't pronounce the name right now, but in 1990 there was a report that equivalent weight amounts of sidestream

PAGE 171:

condensate were more carcinogenic than mainstream smoke to the skin of female MNRI mice at the site of application.

And Witschi more recently in a series of studies has examined the tumor yields in the Strain A/J mouse that were exposed to environmental tobacco smoke. He found that a change in the normal protocol, which calls for a five month exposure and then simply sacrifice and counting of tumor yields, be changed to the protocol to include a four month additional hold after that five month exposure. But the tumor yields were very much enhanced, perhaps giving a model in which environmental tobacco smoke could be studied from a mechanistic standpoint.

Concerning mechanisms of carcinogenesis, we all know tobacco smoke condensates are mutagenic in a wide variety of in vitro mammalian cell system and bacterial systems. I have already indicated tobacco smoke is thought to contain over 4,000 chemicals, many which are carcinogenic, including the ones listed on this slide.

And in a series of very elegant studies by one of our panel members, Dr. Hecht starting in 1993 and has put out papers all through the '90s concerning tobacco specific nitrosamines including detoxification metabolites and metabolites that are capable of forming pyridyloxobutyl adducts with DNA that have been detected in the urine of humans experimentally exposed to machine generated sidestream

PAGE 172:

smoke.

Concerning the human studies that have been reviewed to date, as I indicated, the IARC in 1986 as part of their active smoking document also reviewed several studies of cancers related to spousal smoking. And this involved primarily wives--nonsmoking wives and smoking husbands. Some of these studies showed increased risk in relation to the extent of spousal smoking and others did not.

The IARC conclusion was since the relative risks were low, ETS exposures were not well documented, and other confounders were not controlled, each of the studies was compatible either with an increased risk or the absence of risk. But they also went on to make somewhat of a stronger statement later in the document, which you can read:

"The knowledge of the nature of sidestream and mainstream smoke, of the materials absorbed during passive smoking and of the quantitative relationship between dose and effect that are commonly observed from exposure to carcinogens... leads to the conclusion that passive smoking gives rise to some risk of cancer."

The Surgeon General in 1986, U.S. Surgeon General, indicated in his report:

"The absence of a threshold for respiratory carcinogenesis in active smoking, the presence of the same carcinogens in mainstream and sidestream smoke, PAGE 173:

the demonstrated uptake of tobacco smoke constituents by involuntary smokers, and the demonstration of an increased lung cancer risk in some populations with exposure to ETS lead to the conclusion that involuntary smoking is a cause of lung cancer."

I indicated that the USEPA in the early 1990s carried out a large review of the epidemiology data that had been published to that point. They analyzed data from 27 case control and four cohort studies of lung cancer in nonsmoking women married to smokers or to nonsmokers. This meta-analysis included over 3,000 lung cancer cases in the case control studies and 300,000 women in the cohort studies.

They generated a large number of different relative risks looking at different combinations of studies, but the relative risk that is most often associated with this particular--the USEPA document is the pooled relative risks generated through meta--I'm sorry. The pooled relative risks that were generated in this EPA through a meta-analysis were done so after adjusting for potential bias due to misclassification of smokers as nonsmokers. And this is one of the major criticisms of studies up that time that could have affected the ETS risk for lung cancer.

The pooled relative risks that the EPA determined from this meta-analysis was 1.19 with a 90 percent confidence interval that did not include 1. And this used all of the 11

PAGE 174:

U.S. studies that were considered by the EPA at that time. They concluded that environmental tobacco smoke was a Group 1 carcinogen.

The California EPA in 1997 published a report, and in that report they spent a lot of time dealing with the concerns and issues that had been raised in the literature over a variety of potential systematic biases in studies of spousal smoking--I'm sorry; exposure to a smoking spouse, published prior to 1992. This included the misclassification of smokers as nonsmokers.

Inaccurate estimates of environmental tobacco smoke exposure; you realize that there is no good estimate biomarker of cumulative exposure to environmental tobacco smoke. One needs to estimate based on recall. This of course is better if you are talking to the case than if you are talking to a surrogate respondent. So this could be an introduction of recall bias.

There has also been consideration that there would be a confounding by diet and lifestyle factors that would be common to spouses. There was a consideration that there might be misdiagnosis of a large number of secondary tumors as primary lung cancers. And there has also been the consideration that there was a publication bias and that studies with no effect, showing no effect, would not be published. And of course this would enter into the

PAGE 175:

meta-analysis that was produced by the EPA. This would affect their estimates.

In the time between the USEPA 1992 document and the 1997 California EPA evaluation, there were three large case control studies that were published that addressed potential systematic biases to a great degree. The Fontham et al., study in 1994 was used in part in the USEPA 1992 evaluation. The first three years of that study records were used by USEPA in 1992.

These studies are population based so there is no selection bias. There was a histological confirmation of primary lung cancers for a very large percentage of the cancer cases that were evaluated. The nonsmoking status of the women primarily, the nonsmoking spouses, was confirmed by a variety of different techniques, including unannounced urine evaluations, during collections and evaluations for cotinine or nicotine.

There is a very high quality of information on ETS exposure in a number of these cases, especially the case controls where the information came from the case itself. And there was an extensive evaluation, especially in the Fontham, et al. study, and they concluded that there has been no confounding by diet or other factors that could explain the results.

The Fontham, et al. 1994 study is considered to be the

PAGE 176:

best study in the literature at this time, I believe. There were 653 female lung cancer cases and the odds ratios for lung cancer for tobacco -- for any tobacco use by the spouse is 1.29 with a 95 percent confidence interval that did not include 1.

There was also--the Fontham study as well as the Stockwell and Brownson studies were considered compatible with the causal association of ETS and risk of lung cancer in nonsmokers. And in these studies there were higher risks found with higher exposures. The Cal EPA document also provided evidence that in three studies there was strong evidence, they concluded, linking ETS with nasal sinus cancer in nonsmoking adults.

The potential for publication bias has been examined by a number of authors. One of the better publications is that of Bero et al. In 1994 where they did an exhaustive search for unpublished studies on environmental tobacco smoke. They looked for publications that discussed publication bias for examples of unpublished studies.

They looked at the public comments that were submitted to the USEPA in response to their 1992 report. And they contacted the tobacco industry. And their conclusion was that there is no publication bias against statistically insignificant results on ETS in the scientific literature. One of the other problem areas, I guess if you will, in

PAGE 177:

looking at ETS studies is that there has been somewhat of a disconnect between the results that have been generated by studies looking at a spouse's exposure to ETS from a smoking spouse, nonsmoking women exposed to a smoking spouse, versus nonsmoking women, especially women exposed only in occupational settings.

There have been a number of studies that have published information looking specifically at the issue among--of lung cancer risk of nonsmokers exposed only in occupational settings. A number of the meta-analyses that have been done to this point had not shown significant increases due to occupational exposure to ETS.

In a paper that was sent to you and came out just recently and was not reviewed in the document by Wells in 1998, they developed six criteria to establish minimal study quality. They found that of the 14 available studies that addressed this particular issue five of the 14 were found useful. These five when considered in a meta-analysis gave a pooled odds ratio of 1.39 with a 95 percent confidence interval that did not include 1. Wells also noted why and he noted errors in the previous meta-analyses, and he noted some errors in the underlying data sets that might have contributed to errors in the prior meta-analyses.

Would you cover the slide, Elliot? I just wanted to show you one overhead before I finish. This is a figure from

PAGE 178:

a paper published recently by Hackshaw et al. I don't have a slide on this; I'm sorry. What this is their cumulative meta-analyses done on the studies that have been published looking at women who are lifelong nonsmokers living with a smoker compared with those living with a nonsmoker, looking at the relative risks for lung cancer.

You can see that what they have done is they have simply done a meta-analysis and they have added studies that have come out during these various years. And what you see is a focusing of the overall relative risk down to 1.24, which they believe is the best estimate at that point of the relative risk. And as more studies have come out, you see a smaller and smaller confidence interval around that point estimate.

So the proposal is to list environmental tobacco smoke as a known human carcinogen because studies in humans support a causal relationship between passive exposure to tobacco smoke and human lung cancer. The studies also suggest an association of ETS with cancers of the nasal sinus. The recommendation of the Review Group 1 was to list as known to be a human carcinogen. The vote was 7 to zero with one abstention. RG2 voted known to be a human carcinogen, five for and two opposed.

Subsequent to the reviews of RG1 and RG2 there was a paper published by Boffeta et al. which concerned a

PAGE 179:

multi-center case control study of exposure to ETS and lung cancer in Europe. This is the IARC multicenter study. There were 344 exposed cases in this case control aggregate study to spousal smoke. They reported an odds ratio of 1.16 with a 95 percent confidence level that included 1. These cases included both females and males, males exposed to a smoking wife. He restricted the study to the females exposed to spouses--their spouses' smoke, 321 cases with an odds ratio of 1.11, which also included 1 in the confidence interval.

There was no clear dose-response relationship reported in this study for cumulative exposure, although if you look at the highest exposure category, which is greater than 23 pack years, you did see the highest odds ratio, which was 1.64 with a 95 percent confidence interval that did not include 1. This is for males and females combined.

The diagnoses were confirmed for a number of the cases in this particular study. Current smoking status was well controlled and well evaluated. They controlled for diet, occupational carcinogen exposure, and education. So you see one more study here where the odds ratios are bounding right around 1.15, 1.2, 1.25. And that ends my talk.

Chairman Brown: Thank you, Dr. Bucher. Dr. Zahm.

Dr. Zahm: Just to set the stage, direct exposure to tobacco smoke is a known human carcinogen

PAGE 180

responsible for 80 to 90 percent of all lung cancers as well as a sizable proportion of cancers of the larynx, oral cavity, esophagus, bladder, kidney, pancreas, stomach, and cervical cancer.

Looking at the first criteria, which is the animal data, the environmental tobacco smoke is carcinogenic in animals. It is also reflected in many biomarkers in humans and I believe it is carcinogenic in humans in epidemiologic studies.

The animal data: condensate of sidestream smoke applied to skin, oral mucosa, or lung of experimental animals is carcinogenic. And mice exposed to environmental tobacco smoke of exposed to ETS plus BHT, an agent that enhances lung tumors, had significantly elevated incidence of tumors compared to mice unexposed to environmental tobacco smoke or exposed to BHT but not ETS respectively n some studies.

In terms of humans, there are biomarkers that demonstrate internal exposure to tobacco constituents and metabolites associated with environmental tobacco smoke. These markers include cotinine, thiocyanate, carbon monoxide, and tobacco specific N-nitrosamines. Some work better than others and some show up because of other reasons, but taken together biomarkers do show that exposure occurs.

Obviously most of my comments are going to deal with the epidemiologic data. There have been a number of

PAGE 181:

epidemiologic studies, as you have heard, that have looked at nonsmoking spouses of smokers, have looked at nonsmokers exposed in the workplace, and among persons exposed as children to their parents' smoke.

The spousal research and to a lesser extent the occupational studies have consistently shown excess lung cancer associated with ETS. There does not appear to be excess risk associated with exposure to ETS only during childhood. The spousal exposure to ETS has been consistently associated with increased risks of about 20 to 30 percent with higher risks observed among the heavily exposed. And those risks would be--they go up to around a twofold level. A few are much higher, some of the studies from Asia, the original ones, but those are sort of outliers on the data. The rest would have the twofold excesses in some of the higher exposed categories.

Now, increased risks of 20 percent are extremely difficult to establish as causal associations in epidemiologic research, but I believe they can be so classified for ETS for several reasons. And I am going to go through the reasons, and they do track, you know, the Bradford Hill criteria of judging causality for epidemiologic data.

First of all, I believe there is consistency of the finding across numerous studies, particularly if the recency

PAGE 182:

of exposure to ETS is taken into account. We know that with direct smoking the risk of lung cancer, for example, drops back down to the risk of nonsmokers after about ten years. For bladder cancer it is much quicker. It is about three years.

Well, it appears that for ETS it is even shorter. So one of the critical things in--I mean, it is short. One of the critical things in looking at the epidemiologic data and something to take into consideration with the Boffetta study that was just mentioned is the recency of the exposure to ETS. A weak association was seen in the Boffetta study overall, but it had a greater proportion of subjects whose exposure had ceased several years earlier than that for other studies.

Also, just to talk about the Boffetta study and make another comment, when you look at the exposure, the dose-response data, for that study there are several different metrics that were used. One was just duration only, one is duration times the number of hours the spouse would have been with their spouse per day, and the other is looking at pack years. And kind of a better metric, which is the years plus the hours, is where you see not only a significant excess in the top category, 1.7 to 1.8, but also you see a significant trend overall.

The Nyberg study also demonstrated that timing of

PAGE 183:

exposure is critical with higher risks observed among subjects currently exposed to ETS. So that is one thing to consider when looking across studies and trying to judge consistency.

The next criteria for judging causality would be dose response. Well, there is some evidence of dose-response present in most of the studies, which is a critical feature for this nomination.

The next thing would be how--looking at different biases. Well, one would be the misclassification of exposure. Well, exposure status, either nonsmoker with exposure to ETS, and the amount measures--for example, spouse smoked one pack per day for 20 years--are much more likely to be reported accurately, especially for status, or more accurately than many other exposures in epidemiology when you are dealing with the amount measures.

Proxy reports of smoking history have been shown to be of high validity in methodologic studies. Adjustment for likely levels of misclassification have not eliminated the excess risk. There were several studies, and I have listed them here--Hackshaw, Nyberg, Boffetta--that refer to the work where misclassification has been taken into account and it does not explain away the risk.

Self-reported work exposures, which would be much more difficult to report accurately, the amount of ETS exposure

PAGE 184:

you have there compared to spousal exposure, actually though it is much more difficult, they have been found to correlate well with nicotine are monitors in a study by Willemsen.

Another type of bias that needs to be considered is selection bias. Selection bias is unlikely to explain away the result, particularly for the recent studies that have included all lung cancer cases diagnosed in specified study area. The studies have been population based. They have had high response rates and high rates of histologic confirmation, so selection bias doesn't appear to be an explanation.

Like sources of confounding have been evaluated and have not found to account for the association. For example, the relationship has been demonstrated in nonsmoking women with no exposure to known occupational or other lung carcinogens.

In studies that were able to adjust for diet, a low fruit and vegetable intake has been linked to elevated lung cancer risks, but the ones that have adjusted for diet have still demonstrated the excess lung cancer risks associated with ETS. And the mount of overall 20 percent or so excess in nonsmokers exposed to ETS that might be attributable to dietary differences was indirectly estimated to be only about 2 percent in three different studies that I have listed here.

PAGE 185:

The association is certainly biologically plausible based on animal data and on the well-established relationship with direct tobacco smoking. And there is also no evidence of publication bias against negative studies. So those factors taken together have led me to the conclusion that ETS should be classified as a known human carcinogen.

Chairman Brown: Dr. Yamasaki.

Dr. Yamasaki: Thank you. I thought I would show a transparency, but most of the items I wanted to say was already said by Dr. Zahm, so I will stay here.

First of all, I would like to differentiate from what IARC did and what the NTP is doing here. What IARC did is that tobacco smoke is carcinogenic to humans, and this means IARC just limited ourselves to hazard identification. And we knew that sidestream and the mainstream has similar components. And therefore we restrict ourselves to say tobacco smoke.

And now NTP--what the NTP is doing is to differentiate into two exposure settings, different exposure settings. One is active smoking and the other one is passive smoking. And active smoking we decided last year to be a human carcinogen, and now we are deciding on environmental tobacco smoke. However, because the components of mainstream and sidestream are so similar, it is very biologically--or rather, very plausible that this is also carcinogenic.

But

PAGE 186:

now exposure setting of environmental tobacco smoke of course is very low and therefore the epidemiological power is small enough. And this is small, and therefore IARC in 1986 recommended to have a larger study on passive smoking.

And since then there were many studies. And reading through these documents, I relied heavily on the large scale case control studies, which are Stockwell and Fontham. And I think this was already mentioned in detail by John and Dr. Zahm.

Also Wells has analyzed by meta-analysis five studies leading this exposure to work, not with spouse but work. And these studies it shows not very high but consistent relationship with possible smoking and lung cancers, which is also very strongly supported by the Boffetta study of 1998, which is obviously supported by all the smaller scale studies by Hirayama and others which have been conducted many, many years ago. And so human studies indicate that there is a causal relationship between passive smoking and lung cancer.

Animal carcinogenicity studies I found only one, which was a modified experimental condition with A/J mice, which is a very sensitive mouse for lung carcinogenesis, Witschi in 1997. With or without BHT it increased lung tumors, but this is only in one study. There are two negative studies, by Witschi, et al. and Finch, et al. This has been already known from active smoking that it is very difficult to induce

PAGE 187:

tumors in animals because they are (inaudible) a nose breather.

Other relevant information I think which is very important is the components of sidestream smoke are almost the same as those of mainstream smoke. And therefore we are really dealing with a carcinogen at a low exposure setting. And from this I would agree that this should be classified as knows to be human carcinogen.

Chairman Brown: Thanks you. We now proceed to our public comment, and there are a number of you, I understand, perhaps 13, are there, or 13 plus? Dr. Jenkins. And may I remind you, Dr. Jenkins, but the entire group of the five minute rule.

Dr. Jenkins: Yes, sir. Good afternoon. My name is Roger Jenkins. I am with the Oak Ridge National Laboratory, with is a Department of Energy facility in east Tennessee. I have worked there for about 23 years working on a variety of characterization of airborne materials. I am here today, unlike what it says in the program, representing my own views, although my appearance here today is sponsored by one of our sponsors, the Center for Indoor Air Research.

And the purpose of my presentation today is to contrast the definitive data from a study that we conducted at our laboratory with some of the assumptions made by the EPA in

PAGE 188:

its risk assessment on lung cancer and to try to describe for you the implications of that hard data on those risk assessment.

Now, the so-called 16 city study was a study for which I was a principal investigator, the field work of which was conducted in 1993 and 1994. And we measured personal exposure to environmental tobacco smoke both n the home, away from work, and in the workplace for 100 subjects recruited in each of 16 cities scattered around the United States, and here is a little graphic here to show you what that looked like. And that doesn't show up too well on this view graph, but that is close enough.

We measure d a variety of constituents both in the particle phase of environmental tobacco and the vapor phase as well as we made salivary cotinine measurements before and after the exposures of the individuals. Now, what I want to talk about in the couple of minutes that I have here is this contrast and the differences between the hard data that we have in our study or from our study and the assumptions that the EPA made in its lung cancer risk assessment.

The first example describes what is called the Z factor or the background correction factor. And what you can see here is that basically the EPA estimated that this background correction factor, with is the ratio of exposure to a female if you live with a smoking spouse versus what you get if you

PAGE 189:

live with a nonsmoking spouse. The EPA estimated that risk--or the Z factor to be 1.75. And you can see from the constituents that we measured in our study with the one exception of RSP nothing is close to 1.75. All the Z factors are way above 1.75, again with the exception of RSP, which is not a tobacco specific marker.

Now, what kind of impact does that have on the risk calculations? Well, I am not a risk assessor, but I can plug numbers into formulas provided by EPA. And the formula, if you plug in 1.75 in, here is just a relative risk factor calculation, but you can see is that it is up fairly high, and as it decreases down here to what we would have gotten with FPM or some of the other constituents in our study, you can see a dramatic reduction in the relative risk factor based on a real number for the Z factor as opposed to an assumption.

The other example that I want to describe to you is the misclassification rates. We talked already a little bit about misclassification rates. The EPA estimated the misclassification rate for lifetime never smoking females as 1.09 percent. The estimates from our studies indicated it is much closer to 2.95 percent, and that has a profound effect on the risk assessment.

And her is the 90 percent confidence interval, upper, lower. Here is the actual point estimate. And what you can

PAGE 190:

see is that at a point estimate using a cutoff of 1.06, which is a fairly conservative cutoff for a misclassified lifetime never smoker, you can see that in fact at that level the lower confidence--90 percent confidence interval is almost right at a relative risk of 1.0. If you use a less conservative estimate of 35 ng/mL, what you see is that in fact the lower risk estimate is less than 1.0.

So conclusions, conclusions, that the assumptions that are critical to the EPA's lung cancer risk assessment are simply not supported by the largest personal exposure environmental tobacco smoke study ever conducted in our country. And if you use definitive data rather than the EPA's estimates you in fact would dramatically lower the estimated relative risk. Thank you very much.

Chairman Brown: Thank you very much. Mr. Phillips, please.

Dr. Mirer: May I ask one question?

Chairman Brown: Oh, I'm sorry; a quick question.

Dr. Mirer: Does this apply to the interpretation of the mortality studies or just to the estimated risk from---

Dr. Jenkins: (interposing) Just to the estimate risk.

Dr. Mirer: (interposing) from ETS

PAGE 191:

exposure?

Dr. Jenkins: Right; but I think that--the misclassification rates I think are very important when you are looking at a number of the studies, the epi studies that have been done, and what are the estimated misclassification rates that they use in those studies relative to what the actual ones are.

Chairman Brown: Thank you very much.

Mr. Phillips: Good afternoon, ladies and gentlemen, Mr. Chairman. I have come rather a long way to make this five minute presentation. I am used to speaking for a half an hour per country and I guess to try to do it in five minutes I am not going to do it justice.

However, I think it is important that I attended this meeting really on behalf of not just the company that I represent, Covance Laboratories that is an American company, but really the people who conducted this study. We worked together with the Karolinski Institute from Stockholm while we did these studies. And the studies are very similar to the ones Roger Jenkins has just described except that we this time did housewives working--sort of living with smokers and nonsmokers as well.

We covered ETS particles using solanesol. We covered nicotine, RSP, 3-ethenyl pyridine, cotinine, 32 volatile organic compounds inside and outside at least 10 percent of

PAGE 192:

the homes of the people that we visited. And I would just like to make a comment that ETS may consist of 4,000 compounds, but I have yet to see anyone actually measure them in a real life situation, a very complicated matter indeed.

Again, we did 24 hour personal monitoring. We covered 12 cities, which meant we had to train the people in ten different languages to do these particular studies. We have supplied you with all the data, hopefully, with a list of all the publications, of which there are 17 to date, three of which are in press.

And I guess at the moment I can only try and summarize where we went to in asking the question what does the data we produced mean. Well, as far as I am concerned, it is-- outside the U.S. and probably including the U. S. studies, these are certainly the largest studies ever conducted that we have data for, over 2,600 volunteers.

And I guess the questions I have asked myself, what does this data mean, a very good question indeed. I would like to say that the data does represent some, if you like, the real dose that these people have of environmental tobacco smoke as they go about their daily lives. And what I would like to leave you with is that toxicology is about measuring the dose, we are about measuring the actual amounts in the atmosphere, and risk assessment is something I have to say I am not involved with myself.

PAGE 193:

And what I would like to ask you to do is to really consider this data very, very carefully, having spent four years collecting it throughout the parts of the world that we have been to. And we have done something different than Roger Jenkins, and again, it is controversial because we have converted the concentrations we have found into what we called potentially inhaled quantities using--quite simply the concentrations using the amount of air a person might breathe in over a period for time, et cetera, et cetera. And again, controversially we have changed those or converted those potentially inhaled quantities into cigarette equivalents.

And I think the message I would leave here that although there is a difference between sidestream smoke and mainstream smoke, we have assumed very little difference in order for the general public to get a view of, as I keep getting asked, how much do I really breathe in if I work with someone who smokes, how much do I really breathe in if my husband or my wife smokes.

Seeking the median data from all the people that we have visited over this three to four year period, the levels that we have found vary from a maximum using the man values if you both live and work with a smoker of six cigarette equivalents per year and somewhat closer to .4 or .5 if you live and work with a nonsmoker. Thank you very much.

Chairman Brown: Thank you; a question. Yes.

PAGE 194:

Dr. Russo: These measurement have been done in women or men? What is the sex of the population?

Mr. Phillips: If you go through the papers, you will find that it was a mixture of males and females, approximately 50 percent. But we did also concentrate--different to the 16 cities studies we did concentrate on housewives in their homes, whether they lived both with smokers and nonsmokers.

Dr. Russo: In the female do you consider the menstrual status of the women at the time of measurement?

Mr. Phillips: No.

Dr. Frederick: A quick question: these are fairly massive sorts of analytical chemistry studies and you are a contract laboratory. Who is funding this work and who is the indoor air quality group, which seems to be doing a similar thing? Who is funding that?

Mr. Phillips: The funding for this was made through the Center for Indoor Air Research through to the company that I originally worked for that just changed its name to Covance. It used to be Corning Hazelton.

Dr. Frederick: Where does the money come from Indoor Air? Who puts the money up for them?

Mr. Phillips: I guess that would have to be answered by the Center for Indoor Air Research, I guess.

Dr. Frederick: Give me an answer, somebody.

PAGE 195:

Is this government money? Is it industry money? Is it private money?

Dr. Kelsey: It is tobacco money.

Dr. Frederick: Okay; I just want to be clear on who is funding it.

Chairman Brown: Other questions? (No response.) Thank you very much, sir.

Mr. Phillips: Thank you.

Chairman Brown: Are you Dr. Coggins?

Dr. Coggins: Yes. Mr. Chairman, ladies and gentlemen, my name is Chris Coggins. I am with Lorillard Tobacco Company.

Section 4, Studies of Cancer in Experimental Animals of the NTP background document, relies heavily on animal studies from the University of California at Davis. Our written comments on this subject were not mentioned in the background document even though our submission included numerous other recently published articles relating to the NTP sections on animal carcino-genicity studies, genotoxicity, as well as mechanistic and other relevant studies.

Notably absent from NTP consideration were very recently well designed studies using rates exposed to an ETS surrogate for up to 12 hours per day for over 12 months. The results of these experiments were negative in terms of lung carcinogenesis.

PAGE 196:

Also submitted but excluded was a very recent review of rodent inhalation studies with mainstream smoke. These negative studies were compared in the review with 19 NTP rodent inhalation studies where lung tumors were produced. This collation of positive animal inhalation studies is in our view testament to the general utility and sensitivity of conventional protocols using validated rodent species.

Ignoring the full body of available evidence, the draft background document places complete reliance on very few rodent--on very few recent studies employing a single animal model. This approach is highly unscientific and indicates a major bias, if not a complete predisposition, towards a positive conclusion. Individual studies can never have the scientific strength of a recent examination of all studies.

Our view is that the NTP contractors have stressed the Strain A mouse studies exclusively and that the weaknesses of these few studies have not been considered in relation to the strengths of other studies. Our written text attempts to put the selected studies into perspective examining in detail the choice of test material, test exposure concentrations, experimental design, and animal species.

The responses of the A/J mouse have been compared with the responses in animal strains routinely used by the NTP. Little correlation, if any, has been reported. For example, the authors of one paper concluded that, quote, "There was

PAGE 197:

lack of congruity of results between the strain A pulmonary tumor bioassay and the 2-year rodent carcinogenesis bioassay," quote, along with, quote, "a lack of consistency in strain A bioassay results from two separate laboratories," end quote. Additionally, it was concluded, quote, "Carcinogenicity test date are relevant only to the test model employed," end quote.

The weaknesses itemized in our written comments suggest strongly that as written Section 4 of the background document cannot scientifically support the attempt to classify ETS as a known human carcinogen. We wholeheartedly concur with the authors of the original A/J mouse study, who stated in one of their papers that, quote, "the usefulness of our animal model for the study of human tobacco smoke induced lung cancer remains to be established," end quote.

Clearly then, it would be improper for the NTP to employ experimental findings from a single hypersensitive animal model of uncertain mechanistic relevance to humans as the sole basis for its judgment regarding the animal carcinogenicity of ETS. It is essential that the NTP consider fully and thoughtfully all extant data derived from the full diversity of experimental animal systems employed in the study of ETS to develop a meaningful assessment of potential carcinogenic hazard to humans. Thank you.

Chairman Brown: Thank you. Are there

PAGE 198:

questions? (No response.) If not, let us go to Dr. Levy.

Dr. Levy: My name is Paul Levy. I work at the University of Illinois, Chicago campus. I am a professor of epidemiology and biostatistics at the School of Public Health and I have been a practicing biostatistician and epidemiologist for the past 40 years or so. The work I am presenting is my own. I have been requested to present by R. J. Reynolds Tobacco Company.

My presentation today is about meta-analysis of studies on workplace ETS and lung cancer. Meta-analysis has played an important role in the EPA report and in the Committee's deliberations which we just heard, or the Committee's presentation. Since individual studies have shown low and generally nonsignificant associations, conclusions concerning the carcinogenicity of ETS have been based on the meta-analysis summary odds ratios, at least in human studies.

Prior to 1998 three investigators plus myself have performed meta-analysis on workplace exposures to environmental tobacco smoke. All of us have found odds ratios very close to unity, i.e., 1, with confidence intervals overlapping 1. That is not significant.

The meta-analysis that I performed using 11 U. S. data sets--and that is Exhibit 1, please--is shown there. Notice that I used both fixed effects and random effects models for meta-analyses, and in both the fixed effects and the random

PAGE 199:

effects model the point estimates were low and they were totally nonsignificant.

Recently Dr. Wells, who was mentioned earlier by the reviewer, published a meta-analysis in which he found a summary odds ratio of 1.19 for workplace meta-analysis studies, and his meta-analysis is statistically significant. This was quite different from what I and three others had found later--earlier.

Exhibit 2 shows the odds ratios based on Wells' corrected values that would have been obtained if he had confined his analysis to the U.S. studies that I and the three previous investigators had done. Note that the odds ratio is 1.19 using the fixed effects method and is statistically significant. He did not use the random effects method at all in his paper. The random effects method, however, gives a point estimate of 1.16 with odds--with 95 percent confidence intervals overlapping unity. This is not statistically significant on the U. S. studies.

Exhibit 3 shows a recalculation of my original meta-analysis using Wells' corrected values on the individual studies with the exception of adjustments that I made to Wells' recalculated odds ratios for two of the studies, the Janerich and the Fontham study. The rationale for my adjustments are in my detailed report.

Note that the fixed effects model yields an odds ratio

PAGE 200:

of 1.13. That is just barely statistically significant. And the random effects model yields an odds ratio of 1.12. That is not statistically significant.

Conclusions on the basis of these three exhibits: 1, the size of the estimated meta-analytic odds ratios and confidence intervals were very sensitive to judgments concerning the appropriate data points to use in meta-analysis. This is another example of the fragility of the use of meta-analysis in estimating low associations, which is what we are dealing with ETS and lung cancer.

2, the random effects method gave more conservative nonsignificance confidence intervals than the fixed effects method and is generally the method of choice in combining odds ratios from studies having diverse protocols and controlling for different covariates.

And 3, Wells' meta-analysis which was presented, at least one version of it, used only five of the 16 studies, giving essentially a weight of zero to two thirds of the studies that were available and including the Brownson study, which is highly regarded and one of the better U. S. studies. Thank you.

Chairman Brown: Thank you. Frank.

Dr. Mirer: Two questions: one is does this meta-analysis that you have done weight the studies according to their size? And the second question is do you

PAGE 201:

offhand know what the background risk rate for the nonsmoking men or women in these studies was, the lifetime risk of lung cancer?

Dr. Levy: Okay. It weights--the fixed effects model weights each study by the inverse of the estimated variance of the odds ratio. And essentially that is based on essentially the size of the study.

Dr. Mirer: Okay.

Dr. Levey: The random effects takes into account--this uses a model that each study does not estimate the same thing because the odds ratio--the methods were different, the protocols were different, they controlled for different covariates. So it assumes that they are a sample of possible studies that if you took all of them together would come up with a reason---

Dr. Mirer: (interposing) So it give them equal weight?

Dr. Levy: What?

Dr. Mirer: It gives them equal weight?

Dr. Levy: No.

Dr. Mirer: No?

Dr. Levy: It takes into account the variability among the studies not controlling for the size of them and then also throws in another component due to their individual weights. It is a paper by DerSimonian, Laird. I

PAGE 202:

have done this thing for about 40 years. My thesis was on it. It was originally based on a paper by William Cochran in 1938 and rediscovered 40 years, 50 years later.

Chairman Brown: Thank you very much. Dr. Millner, please. Is Dr. Millner here?

Dr. Frederick: Bud, can I ask a question of Dr. Levy?

Chairman Brown: Okay.

Dr. Frederick: This is totally confined to workplace exposures; is that correct?

Dr. Levy: Yes, and so was Wells, but Wells did all of them including the U. S. plus six others. Mine was on the U. S. But my first slide got the same odds ratio that Wells got.

Dr. Frederick: That is fine.

Chairman Brown: Let us then go to Mr. Repace, who is going to receive a bonus of 30 seconds because he is going to show us a slide that Dr. Bayard would have shown had he come.

Mr. Repace: Thank you. Just so you will know what my position is, there is no question that it is a human carcinogen. I commissioned the EPA study back in 1987 and so I am in a fairly unique position to talk about it. I am just going to review very quickly what Dr. Bayard and I agree to be true. There are a large number of agencies in

PAGE 203:

the United States and aborad that have declared environmental tobacco smoke to be a human carcinogen, including in the U. K. And France and Australia.

EPA's 1992 report was reviewed by an outside science advisatory board of 18 independent experts who unanimously endorsed the report, heard the tobacco industry's comments, and rejected them. All the arguments virtually that were presented today, except the data of Dr. Jenkins and the Covance Laboratories, were presented to the EPA. The NCI endorsed the report. The North Carolina court, which vacated the ETS risk assessment, the judge has no apparent scientific credentials that we are aware of.

The very important thing to emphasize, the relative risks are the highest for the highest exposure groups and when you--the weighted average exposure ratio corrected for smoker status misclassification is 1.8. And when corrected for background ETS in the control group, the adjusted odds ratio is better than 2. In fact, I would place it at 2.5. And one of the papers I presented in my submission contains that estimate.

Here is probably the best estimate of risk of ETS. It is based on smokers. This is in the 1979 Surgeon General's report. Here are smokers who are inhaling. These are--the red bar is smokers who don't inhale. About two thirds of their risk seems to be coming from ETS. And that is the

PAGE 204:

nonsmokers' risk right there. I don't think you need any statistical studies on nonsmokers when you have got good stuff like that.

Here is some work that I and Dr. Ott and Neil Klepheis published in the NCI monograph number 9. This is in a paper on cigar smoke pollution. This is the cigar smoke in a San Francisco parlor here. This is the RSP level. This is the Marlboro cigarette smoke by comparison. Notice how good the model explains the data. Up here these are the polycyclic aromatic hydrocarbon, PAH emissions of the cigar and the cigarette. You can see they are both big emitters of PAHs. And unless I am mistaken, those things are not very good to breathe.

Now, one of the important things that really needs to be emphasized are the epidemiologists do not assess exposure properly. Assessing exposure by spousal smoking status depresses odds ratios and wipes out statistical significance, because many women who are not married to smokers are heavily exposed to ETS.

I am not going to talk about all the things that the models can do because there is not time, but let's just look at the nicotine model. It says that the nicotine concentration--and this could be any other component of ETS-- is directly proportional to the smoker density and inversely proportional to the air exchange rate.

So any venue that has

PAGE 205:

high smoker density and low air exchange rate is going to be very exposed irrespective of whether it is a home or a workplace.

We used this model in a Monte Carlo analysis to compare with data that Kathy Hammond took in 12 Massachusetts workplaces. This dark curve here--and I'm sorry, my laser printer broker down--is the theory--and these diamond shaped points--I'm sorry; these little circles here are the data points. And you can see there is very good agreement between theory and experiment.

We have also developed models, and the papers that I submitted to you show these and I am not going to go into them in any detail, but you can predict the plasma cotinine concentration as a function of nicotine concentration and the time that you spend in that concentration. We can also do it for urinary cotinine. And again, that is in our '93 paper. And you can see that these models in fact predict levels in large clinical epidemiological studies which are very close to what is actually observed within 5 or 10 percent. So these models are actually quite good.

Here is the model that we developed for salivary cotinine. We published it this year, in 1998. And we did a Monte Carlo analysis on this equation as well using our office nicotine data. And you can see again, the round circles are the data points and this black curve is the

PAGE 206:

theory. There is very good agreement between theory and experiment. And so we can not map salivary cotinine into plasmal cotinine and urinary cotinine---

Chairman Brown: (interposing) You have 30 seconds remaining.

Mr. Repace: Excuse me?

Chairman Brown: 30 seconds.

Mr. Repace: 30 seconds; I am going to finish up. We can predict pharmacokinetic data from NHANES III and we can analyze studies like Dr. Jenkins. This is his comparison of NHANES III with what he reported. You can see that it is pretty close. But when you look at the nicotine data that he reports versus the NHANES nicotine data--that is these tall bars--there is very poor agreement.

And so if you look into the details of it, there is something wrong with the pharmacokinetics. His nicotine exposures in the salivary cotinines seem to be dependent on nonlinear pharmacokinetics, and I don't see how possibly that can be true.

The same thing goes for the Covance study or the Corning Hazelton study. If you look at the salivary cotinines that he is reporting, and this the NHANES salivary cotinine over here, they are pretty similar. But if you look at the nicotine levels, they are all much lower than the nicotine level that we would calculate from the NHANES

PAGE 207:

study. Now, this is a log scale, so these are an order of magnitude low. I don't find these studies to be hard data at all. I don't think they should be relied upon.

Chairman Brown: This will have to be your last slide.

Mr. Repace: This is my last slide.

Chairman Brown: Very good.

Mr. Repace: Okay. So in summary you can use--there are dosimetric models that you can use now to assess exposure, and I think that can improve the assessment of epidemiology greatly. Thank you.

Chairman Brown: Thank you. On question?

Mr. Repace: Yes.

Dr. Medinsky: Can you actually just maybe talk a minute about how it might impact epidemiology, because one of the I guess agonies I have with this environmental tobacco smoke data is the really small relative risk associated with the cancer. And so---

Mr. Repace: (interposing) Yea. It isn't small at all. I would place the risk for the average person at about 2.4 and not about 1.25. I can calculate a risk of 1.25 and I have done it in one of the papers that I submitted to you.

If you assess exposure simply saying that we are going to use spousal smoking status, you have got a whole group of

PAGE 208:

women who are married to nonsmokers but who are exposed in the workplace. And that has been true for decades in this country, and it has been true in Europe as well. Where do you find an unexposed control? You can't do it.

And if you look at Mike Cumming's data from Roswell Park where he has compared people who walk into his clinic, women who are married to smokers versus women who are married to nonsmokers, and you look at the urinary cotinine level, it is this much in the women married to smokers and this much in the women married to nonsmokers. So what you are comparing is more exposed to less exposed, not exposed to unexposed.

The epidemiologist do not assess exposure correctly and that is why the odds ratios are so darn low. They are not really that low. It is just an artifact of the way that epidemiologists have chosen to assess exposure.

Dr. Medinsky: Do you mind if I ask a follow up?

Chairman Brown: Go ahead.

Dr. Medinsky: Okay. So you say that if you actually got a true population that was never exposed, that their incidence of cancer---

Mr. Repace: (interposing) Southern California Seventh Day Adventists; that is right. If you compared non-Seventh Day Adventists in Southern Carolina who are lifelong nonsmokers, they have two and a half times the

PAGE 209:

lung cancer rates of a Seventh Day Adventist. And I submit that is a true estimate of the effect on the general population. It is a much larger effect than you would believe.

Unfortunately, the epidemiologists only assess exposure based on spousal smoking status. And if they would read some of the papers on modeling and dosimetry of environmental tobacco smoke, they could improve the assessment of the epidemiology enormously.

Chairman Brown: Thank you very much.

Mr. Repace: You're welcome.

Chairman Brown: Dr. Carchman, please.

DOWNLOAD THE DOCUMENT IN WORD 6.0 FORMAT

Dr. Carchman: Thank you, Mr. Chairman, members of the Committee. I am just going to give some brief comments on one of the studies that has already been talked bout and only one part of that study. The IARC multicenter study actually deals with not simply epidemiology but has a mechanistic component to it, which I won't specifically talk about. So in a certain sense not only is it one of the largest studies, it is probably the most comprehensive study that has been executed to date.

Some of the components of the study is it was in 12 centers in seven countries, 650 cases, over 1500 controls, enrollment from '88 to '94. It generally had a common questionnaire. They had estimates of ETS exposure, some

PAGE 210:

information on occupational exposure, urban versus rural living, but there was some disparities in terms of information on education and diet.

The response rates varied from center to center from being less than 50 percent to being closer to 95 percent. Selection of controls varied from center to center either being community based or hospital based. But the authors of the study did some adjustments in terms of the kinds of models that they would use to adjust for it. And they also did diagnostic criteria and were able to account for 96½ percent of the lung cancers that were evaluated, and they did some adjustments for that as well.

This is pretty much an extraction from that paper looking at childhood, spousal, workplace, workplace or spousal, overall in vehicles, overall public places. And this deals strictly with the odds ratios and the 95 percent confidence intervals.

The first thing that is apparent, there is nothing on here that shows a statistically significant increased risk for lung cancer in any of the categories. The only statistically significant point has to do with a decrease in childhood cancer. There was an intercenter range--that is, between the different centers. And these are all odds ratios.

They could vary by a factor of 5 or more. And even within a country where you had exactly the same

PAGE 211:

questionnaire, three centers in Germany, three centers in Italy, the same language, the same questionnaire, you also had some significant variability between those centers.

As was pointed out by one of the reviewers, the authors did go point by point through a number of sources of bias. And each individual source of bias was concluded by the authors as not being accountable for the effect. Unfortunately, if you leave the analysis bias by bias, that is absolutely true.

But in terms of looking at the entire picture, you need to look at--and they didn't do this for whatever reason--the biases that they did look at. And if you follow this chart for reported spousal, workplace, or the combination, those are the odds ratios. The misclassification biased that they used, they dismissed it as not being--could not account for the effect. And they are right, .03.

Histological confirmation; even with 96½ percent, histological confirmation has a small impact. Unconditional analysis, and this in part has to do with the kinds of controls, has a small and variable effect. All of these are negative. That is, they are going to be lowering the specific odds ratio; confounding, very small or no effect.

If you then subtract, just do a simple addition of those negatives and subtract it, you can see just looking at these single biases reduces the odds ratio from 50 to over 70

PAGE 212:

percent. And these adjust the biases that they looked at.

I have provided more detailed discussion of these points in our submission on this. But our conclusions are it is a very large and well conducted study. There was no statistically significant overall increased risk reported for any sources of reported ETS exposure. IARC collected data on many sources of bias but did not report an odds ratio adjusted for all of these factors. The data presented within the paper indicate that if all the adjustments had been made the reported excess risk would have been reduced by at least 50 percent. Thank you.

Chairman Brown: Thank you. Questions? (No response.) Let us now then proceed to do Dr. Butler.

Dr. Butler: My presentation is going to focus on two epidemiologic studies included in the NTP report, those by Brownson and Fontham. It is appropriate to focus on these studies because, one, they are conducted on human populations and therefore are appropriate in the determination of whether ETS is a human carcinogen.

Second, they are based on U. S. populations. In fact, they are the largest U. S. studies. Third, they are methodologically superior to many of the other studies conducted previously. Fourth, the NTP report relies heavily upon these two studies, as was presented earlier. And fifth and final, the NTP reports these studies as indicating a

PAGE 213:

positive association between ETS and lung cancer when in fact the data indicate the opposite, that there is no association.

I am going to present some details of the analysis here--or excuse me; the final analysis, the details of which are presented in my written submissions. First of all, with respect to Brownson, I have access to raw data from that, getting it from NCI. And I conducted analyses first to replicate what was published by Brownson and then to extend that analysis.

Now, that result that they report starts with a relative risk or an odds ratio of 1.0 indicating no association. They then indicate no dose-response relation-ship, odds ratios going 1.0, 0.7, 0.7, and then the infamous 1.3, which has been alluded to many times here, the highest dose group having the highest relative risk.

That in and of itself does not indicate a dose-response relationship as this does not, and the 1.3 is the only value from Brownson which is listed in the NTP report.

What I did was extend that analysis restricting the cases to self-respondents. By design, the Brownson study--the controls were self-respondents obtained from random digit dialing. The cases had both self-respondents and surrogate respondents. So methodologically it is appropriate to restrict the analysis of the cases to self-respondents to be comparable to the controls. And when you do that, you

PAGE 214:

continue to observe no association between ETS and lung cancer, and the 1.3 is dissolved, it is gone, indicating that the Brownson study shows no association between ETS and lung cancer.

Moving to Fontham, Fontham et al. Presented their results, cross classifying the cases and controls both by childhood ETS exposure, and adult ETS exposure. Those in the lower right-hand corner having both childhood and adult would be expected--since they have the highest combined exposure, would be expected to have the highest risk. Those that are in the upper left-hand corner, having neither childhood nor adult, would be expected to have a lower risk. Those on the off diagonal would be expected to be somewhere intermediate.

However, that pattern is not the way that Fontham analyzed the data. Instead what they did was use two comparison groups defined by childhood exposure. What they found was that among those with no childhood exposure adult ETS was not associated with lung cancer, while among those with childhood exposure it is positively associated, 2.86. When you combine those two, you get the 1.3, the average exposure--excuse me; average odds ratio.

However, the comparison of this 2.86 of implying that those were the joint exposure of a rate 2.86 times greater than those with no exposure is incorrect. Taking the same data and using a common comparison group, you see that the

PAGE 215:

odds ratios for all the adult exposures are 1.0 whether they do or do not have childhood exposure, that the only deviation from the pattern are those with childhood exposure but no adult exposure.

The bottom line is that both Brownson and Fontham are negative studies, they are the largest U. S. studies, and they should be reflected in the NTP report.

Chairman Brown: Thank you. Are there questions? Yes.

Dr. Zahm: Yes. I have one question about your comment on the Brownson study. First, it is true that usually we much prefer self-respondent data to proxy data because it is thought to be less misclassified.

In this case when you are looking at spouse smoking history, are the respondents the spouse, and in which case the data might be even better, which raises the question if it was a 1.3 with everyone and it was 1.0 for the self-respondents, was the proxy data 1.7 or 1.5?

Dr. Butler: The proxy data--well, hold on. It is not appropriate to use all respondents when your comparison group is self--when the controls are self-respondents. If indeed the controls were both self-respondents and surrogates, then you could compare the self-respondent cases with the self-respondent controls and the surrogate cases with the surrogate controls. When you have

PAGE 216:

only self-respondent controls, then you have to use self-respondent cases. Does that make sense?

Dr. Zahm: You are saying there were only self-respondent controls?

Dr. Butler: By design; random digit dialing got the women at home. Interviews were conducted then. But the cases--even though there was a rapid case assessment by the time we got to the hospital, sometimes the case was not in a position to conduct the interview. That is answer number one.

Dr. Zahm: So we don't know what the proxy person---

Dr. Butler: (interposing) Right; what we do know, what we do have valid data on, is self-respondent controls, self-respondent cases. And when we look at those data, not only is there no association overall, it flatlined. Okay. With respect to your second question about what the surrogates look like, the odds ratio goes 1.0, 0.5, 0.7, 1.5.

Dr. Zahm: That is surrogates compared to self-respondents.

Dr. Butler: Right.

Dr. Zahm: We can't use it.

Dr. Butler: Well, your question to me is what--you had two questions to me. One was whether it was appropriate to use--or excuse me; wouldn't the surrogates

PAGE 217:

then be superior to the--and my response to that was no. Your second question, I thought, was wouldn't you expect to see an increase.

Dr. Zahm: No. I understand--the analysis if you could do a proxy proxy would be superior perhaps if these were the spouses themselves. So the question was, you know, are the--who are the proxies? Are they 50 percent spouses, 10 percent daughters, you know, 6 percent--ion any event, we can't tell because you can't do a proxy proxy.

Dr. Butler: We can't do a proxy proxy that is in the data. The majority are spouses and daughters.

Chairman Brown: Thank you very much. Let us know now proceed to Dr. Scherer.

Dr. Scherer: Mr. Chairman, ladies and gentlemen, I am Gerhard Scherer. I am head of the Analytisch-biologisches Forschungslabor in Munich, Germany. Our tobacco-related work is sponsored by the German Association of Cigarette Manufacturers. I shall present some recent results of our biomonitoring studies on the ETS related exposure to three classes of carcinogens, namely polycyclic aromatic hydrocarbons, aromatic amines, and tobacco specific nitrosamines.

We measure three PAH biomarkers, namely urinary 1-hydroxypyrene and adducts of benzo[a]pyrene to albumin and to hemoglobin. Nonsmokers were classified as exposed and

PAGE 218:

unexposed according to their self-report. The cotinine levels, which indicate the exposure to tobacco smoke, were tenfold higher in the exposed group compared to the nonexposed group.

Urinary excretion of 1-hydroxypyrene and the adducts to albumin and hemoglobin levels in nonsmokers were independent of their exposure to ETS. Smokers had about 70-fold higher cotinine levels and significantly elevated levels of pH biomarkers.

We also biomonitored the exposure of pregnant nonsmoking and smoking women to 4-aminobiphenyl, a human bladder carcinogen, by measuring the hemoglobin adduct and in a group of healthy adults the urinary excretion of unchanged 4-aminobiphenyl. Cotinine levels again show a significant difference in exposure to ETS between the two nonsmoker groups.

There was no influence of ETS exposure on the 4-aminobiphenyl biomarker levels in nonsmokers. Smoking significantly increased the adduct level but not the excretion of the unchanged 4-aminobiphenyl. The levels observed in nonsmokers hint at a yet unidentified source of 4-aminobiphenyl or the corresponding nitro-compound in the environment.

NNK is a tobacco specific nitrosamine and induces lung tumors in rodents. NNK exposure can be biomonitored by

PAGE 219:

measuring a hemoglobin adduct or by determining the urinary excretion of the main NNK metabolite, namely NNAL in its free and/or glucuronidated form.

ETS exposure did not influence the HPB hemoglobin adduct levels. Smokers had significantly elevated adduct levels, but the ratio between smokers and nonsmokers is much lower than expected. The reason for this discrepancy is unknown.

ETS exposure significantly increased the urinary excretion of total NNAL. Smokers exhibited more than 100 times higher NNAL excretions than nonsmokers. These results indicate that urinary NNAL excretion in nonsmokers is related to the exposure to ETS.

Conclusions: although ETS contains substances classified as carcinogens by different organizations, the exposure of nonsmokers to ETS under realistic conditions does not exceed the common background exposure to carcinogens unspecific for tobacco smoke. Exposure to tobacco specific nitrosamines such as NNK is measurably increased in ETS exposed nonsmokers.

The biological meaning in terms of cancer risk is uncertain considering the low exposure level compared to everyday burden with nitrosamines and the human metabolism of NNK, which is different from that of rodents, preventing straightforward risk extrapolation to humans. Thank you.

PAGE 220:

Chairman Brown: Thank you very much. Are there questions? Yes.

Dr. Russo: Did you have a chance to measure these in the--for example, in the milk of--did you measure these values in the milk of lactating women or in bronchial lavage or in the--for example, in the cells obtained from the urine, because what is in the (inaudible) is not relevant. What is important is in the tissue. And I think it is very important to determine probably those same things in for example what happened through the milk of a lactating mother or bronchial lavage or even in the cells from the urine.

Dr. Scherer: What I showed here were mainly or only protein adducts, adducts to hemoglobin and to albumin. And these measurements were all done, of course, in blood.

Dr. Russo: No values in those other tissues?

Dr. Scherer: No.

Chairman Brown: Thank you very much. Dr. Gori.

Dr. Gori: Mr. Chairman, the panel, enough is known about mainstream smoke that smokers inhale is a human carcinogen, but mainstream smoke and ETS are not equivalent. Generating temperatures, humidity, oxygen tension and pyrolysis conditions are substantially different.

PAGE 221:

It is true that about 5,000 components have been measured in mainstream smoke, but only two or three dozen have been measured objectively in ETS because of the extreme dilution, so that similarities remain a matter of conjecture.

And there are dramatic exposure differences. Nicotine and cotinine are not useful markers of ETS because low level body burdens can be affected by sources secondary to ETS exposure and even by dietary nicotine.

Based on suspended particles or tar, if you wish, in 1982 the EPA agreed that the prevalent considerations of ETS particulates are on the order of 50 µg/m³. More recent studies suggest that it may be half as much. EPA data alone leads to conclude that the typical exposure to ETS may be equivalent to smoking about one cigarette or so in one year, an exposure that is clearly immaterial in the cumulative context of ordinary exposure and one that could hardly be expected to produce epidemiologic signals that exceed background noise. In fact, epidemiologic studies of ETS and lung cancer, as we have seen, are invaluably equivocal.

The meta-analysis of workplace exposures continue to show practically no elevation or risk. The meta-analysis of childhood studies, arguably the most sensible studies, is actually compatible, continues to be compatible, with a slight reduction of risk.

Only spousal studies register a meta-analysis elevation

PAGE 222:

of some 7 percent for U. S. females. Still, all categories are incompatible with positive and negative studies and more than 90 percent are not statistically significant unless one of course resorts to the gambit of the EPA of using 90 percent confidence intervals.

Moreover, no studies, and I mean no studies if you look one by one at them, have credibly accounted for misclassification of smokers as nonsmokers. No studies have credibly controlled for 30 odd risk factors for lung cancer that are reported in the literature. And all of them are manyfold greater than the risk alleged for ETS. No comprehensive review has accounted for the obvious publication bias that has favored positive results.

It is a combination of problems that prompted, as you have seen before, the International Agency of Research on Cancer to conclude that ETS studies are compatible with a slight increase or a decrease in risk. Clearly epidemiologic studies do not challenge the known hypotheses.

With this in mind, it was surprising that the background document available to this committee, presumably a scientific document, affirms that ETS is a known human carcinogen by quoting the 1992 review of the EPA and one by the California EPA more recently. The issue here is that both reports are not science based.

The EPA review if you read the document at page 5.2,

PAGE 223:

was based, and I quote, on the a priori hypothesis that a positive association exists between exposure to ETS And lung cancer. This is hardly a scientific way of starting a particular review. The California EPA review was based on similar prejudice. In fact, the EPA engaged in egregious contempt of objectivity and science enough to merit the reprimand of a federal court that this past July declared the EPA report invalid, misleading, and void.

Besides the EPA document is a risk assessment exercise that conflicts with the mandate of this committee. In its mission statement, the NTP affirms that listing in the annual report on carcinogens, and I quote, is not intended to constitute a risk assessment but it is a hazard identification only. Excellent intentions, I would say.

So how did we get to consider ETA as a possible carcinogen? Here is my reading of the saga. A public campaign against smoking is definitely justifiable, but the smoker's right to smoke has been a major impediment in achieving a smoke free society. As a remedy, great expectations were placed on the possibility that ETS could be shown to be a significant health risk to nonsmokers.

Today an objective scientific ground shows that the evidence is still not there, and statements to that effect could only be made on the invidious grounds that the end justifies the means. Scientists I hope would not agree.

PAGE 224:

Undoubtedly ETS is an avoidable irritant, but there is no credible evidence that it is a human carcinogen. Thank you.

Chairman Brown: Thank you very much, Dr. Gori. Are there any questions to be directed to Gio?

Dr. Gori: Apparently not.

Chairman Brown: Apparently not; thank you very much. Dr. Marks.

Dr. Marks. Good afternoon. My name is Ronald Marks and I am a professor of statistics and the Director of the Division of Biostatistics at the University of Florida Health Science Center. I have been a biostatistical consultant in health sciences for 25 years. And I have been asked to summarize my independent research findings today at the request of the R. J. Reynolds Company.

I am going to summarize my comments in two areas. First, I have had the opportunity to evaluate the original and complete data sets generated in the Brownson study, the Stockwell study, CPS 1 and CPS 2. This gives us the opportunity to look at a more global basis of the risk factors and their importance in predicting disease.

One of the problems in the scientific literature is that most of the publications tend to focus on one risk factor. And although the authors claim to adjust for potential confounding factors, I think we are all aware of the limitations of being able to appropriately do that in a

PAGE 225:

single paper. So what is needed in an area like this where there is a number of risk factor groupings being studied, it is important to take a more global look and to assess the relative importance of each risk factor on disease outcome.

My findings from doing this analysis, an appropriate step-wise statistical procedure using all the data available in these studies, is that the dietary factors are by far the most common and consistent risk factors for lung cancer. Risk factors are identified at both increased the risk of lung cancer and were determined to be protective against lung cancer. And the findings in these studies were consistent with what has appeared throughout the scientific literature regarding the relationship between dietary factors and lung cancer.

These studies had some information on family history, personal illnesses and workplace exposure. That information was somewhat limited and not consistent between studies, so although a few of these factors came into a few of the models, there was no evidence of any consistency as would be expected due to the limited nature of the data.

The remaining risk factor area is ETS. And all of these studies except for CPS 1 were specifically designed to evaluate ETS and its influence on lung cancer. So there was substantial ETS data and it was collected fairly consistently across these four studies.

PAGE 226:

Yet in the analyses the ETS risk factors seldom came into the developed models. And when they did come in, some of the times they showed an increased risk of disease and some of the times they showed a decreased risk of disease similar to what Dr. Butler showed on one of his transparencies from the Brownson study where there were significantly reduced risks in a number of cases.

So some of the results were positive and some were negative so they cancel out, and the conclusion has to be that there is no evidence that ETS is related to the development of lung cancer based on the analysis of these four largest available data sets in this area.

The second area that I researched was the dose-response relationship between ETS and lung cancer. And in this particular case we have to be careful and we have to realize that this is really more of a trend analysis, which is a lesser form of dose response analysis due to the retrospective nature of the data in this area, and the fact that the doses are so imprecisely measured.

So we are looking at a trend analysis. And there is two requirements that are important to be met before we ever compute the statistical test and generate the p-value for significance of a trend analysis. And those two requirements are that first, all the odds ratios being compared in that trend analysis are greater than 1, and second, that the

PAGE 227:

results are monotonic. That is, as the dose increases the odds ratios increase in a consistent manner, not necessarily linear or any particular form, but that they increase.

In a review of the scientific papers that were used in the EPA report I found that 14--I'm sorry--, 17 percent of the reported trend analyses in those papers included odds ratios less than 1 and 67 percent of the reported trend analyses from the EPA report were nonmonotonic.

It was felt that when the newer studies, the Brownson, the Fontham, the Cardenas, and some of the newer studies cam out, which were of higher quality design and would provide more definitive answers to the question of ETS, in those studies 51 percent of the reported trend analyses had odds ratios less than 1. And 83 percent of their reported trend analyses were nonmonotonic.

The evidence from this is clear that there is no evidence from the published scientific literature that there is an established dose-response relationship between ETS and lung cancer. Thank you.

Chairman Brown: Thank you. Questions? Yes, Frank.

Dr. Mirer: Is any of your work--has any of your work of Dr. Butler's work been submitted for publication?

Dr. Marks: No.

PAGE 228

Dr. Mirer: Thank you.

Chairman Brown: Thank you. Dr. LeVois; and I apologize for mispronouncing you name.

Dr. LeVois: That is perfect; it is LeVois. I am Maurice LeVois and I have been a consultant for about 20 years in epidemiology. I got started in the Agent Orange era at the Centers for Disease Control and attended your earlier meeting with a great deal of interest.

I would like to first draw a couple of distinctions between my views and some things that were said earlier. I have published a paper on recall bias with Paul Switzer. In that paper we noted that among U. S. Case control studies there is actually a reduction in risk, a risk below 1, at the first or the lowest positive level of exposure in the U. S. Studies in the meta-analysis, meaning that there appears to be some shifting of responses to higher levels of exposure than may have actually taken place.

Without belaboring the point, because explaining the paper itself would take my five minutes, this was consistent with the operation of recall bias. It is very hard to explain by any biological means. So I believe that there is some evidence that recall bias is a problem in these studies.

Regarding publication bias, I did a paper on this topic with Max Layard in which we looked at the U. S. Studies of heart disease, not lung cancer, and ETS exposure. And we

PAGE 229:

found a statistically significant association between the size of the study, an inverse association, and the size of the risk reported when this was corrected for study size as is done in the meta-analysis.

Therefore, we concluded that there is evidence based on at least the ETS heart disease literature of a publication bias in the ETS literature. And I would beg to differ that publication bias can only be defined by studies that are not found to be significant that are later detected as not having been submitted for publication. Publication bias unfortunately is more insidious than that. It enters into their whole process of designing and conducting and deciding what to report and how the analyses are done and so forth.

However, I do think that it is important to note that in our study were able to find three very important studies that had not been reported. Two American Cancer Society studies and the National Mortality Follow Back Survey were all three null, very large studies, available to scientists to review.

And these studies were not included in that significant association between the risk and the study size. This was an independent way of judging whether or not there are studies out there that just aren't being reported. So Bero has found no studies. We found some studies. I think it is a lot more significant to find unreported studies than not to.

PAGE 230:

I would also point out that the consistency that has been noted has not universally been admired. There is a very significant lack of consistency geographically. If you do a meta-analysis, there is significant heterogeneity geographically. And unfortunately, the Chinese just published a meta-analysis in which I believe six or seven lung cancer ETS studies showed a risk less than 1. So not only is there significant geographic heterogeneity, but there is not even a consistent increase in risk geographically.

And I think it has been pointed out that the dose response that has been talked about is actually a trend test in all of the ETS literature and the EPA report and so forth.

They don't really do a dose response as toxicologists do a dose response, nor does Brezlow and Day recommend that you do a dose response, and that is exclude the nonexposed group in your trend test. Otherwise you are doing a trend test, and there is a difference, which I won't belabor.

Before my time runs out, I would like to get to one recent paper that I think you should be aware of. It certainly made an impression on me. It is a paper by Shelly Miller, Steven Branoff, and William Nazaroff. The authors are from the University of California at San Francisco, the Department of Civil--well, the University of California-Berkeley, the University of California-San Francisco, and the Department of EPA, Department of Mechanical Engineering

PAGE 231:

rather--I'm sorry; the U. S. Environmental Protection Agency Region 9 is Shelly Miller's affiliation.

They looked at exposures to 17 constituents of tobacco smoke that are also somewhat ubiquitous in the environment. These constituents are by no means unique to tobacco smoke, but they are constituents of interest in tobacco smoke. They did a study using personal air monitoring between 1984 and 1990 and they derived time weighted daily average exposures to 17 constituents. I won't read them all, but it is quite a laundry list: acetaldehyde, croanitrile, 3-butadiene, benzene, acrylate, N-nitrosamethyldiene and so on, toluene.

And one of the most important findings of this study---

Chairman Brown: (interposing) Do you have about one or two sentences that are short?

Dr. LeVois: All right. Let me get to the most important finding, because none of these constituents of tobacco smoke that they measure accounted for more than 8 percent of the daily time weighted average exposure. Most were around 3 percent. I believe that seriously undermines the signal to noise detection ability of spousal smoking studies.

And it also points out that there are--90 percent of the exposure to the substances of interest are not measured in the spousal smoking studies, could not possibly have been controlled by--controlled for in terms of potential

PAGE 232:

confounding due to blue collar lifestyle, lower socioeconomic status and so forth that is known to be associated with the smoking habit.

Chairman Brown: Thank you very much.

Dr. LeVois: You're welcome.

Chairman Brown: Questions? Yes, Frank.

Dr. Mirer: You made reference to a number of published papers. Do we have them?

Dr. LeVois: No, you don't. I will submit them to the record. I got a request to present--it is inaccurate. I am not here representing the Tobacco Institute. I have been a consultant with them over the years and I received a request to present in the last week. So I wasn't able to get my materials to the Committee in time; I apologize.

Chairman Brown: You were speaking of spousal signal to noise. It has been my experience that there is more signal than noise in that relationship. We had one more, Dr. Millner, if he showed up. He was listed. Is he here? (No response). I guess not. And we are about to break for our---

Dr. Butler: (interposing) Excuse me.

Chairman Brown: I'm sorry. You have had your---

Dr. Butler: (interposing) I just want to

PAGE 233:

respond to Dr. Mirer's question. I am Dr. Butler. He asked---

Chairman Brown: (interposing) I know who you are. If you can do this--can you do it briefly?

Dr. Butler: Yes. The question was had I submitted it for publication. And the answer is yes, I had submitted it as letters to the editor for both of the journals. Responses came back saying that because of the time delay they would not accept them as letters. It was not because of a conflict.

Chairman Brown: Thank you very much. Now, this is our last opportunity to break. As matter of fact, the cafeteria is closed. But through the efforts of fairly highly placed people here at NIEHS it is open; right? Sort of open.

I understand that there is some coffee and that there are 100 cookies. And that is it, just 100. Now, there is a cost associated with this. How do we work this out?

Okay. We won't worry about that at the moment. So it is in the private dining room, which is past where the trays are at the very end of the--we will be back here by 4:05.

(A recess was taken from 3:45 p.m. to 4:09 p.m.)

Chairman Brown: I have been exceedingly impressed by not only the content but the discipline that has been inhibited by our public commentators that you were able

PAGE 234:

to stay within your times and the issue at hand very specifically.

Now, we shall proceed to a discussion of the Committee on the issue of environmental tobacco smoke based on the information that you received a month ago and in the weeks since, the days since, and in the discussion that you heard here. Are there comments? (No response.) If there aren't any comments we will proceed at once to a motion.

Dr. Frederick: Well, wait a minute.

Chairman Brown: I somehow thought, you know, we were going to wake you up.

Dr. Frederick: I'm sorry, Bud. But I didn't want to be first this time. There are some other things that are kind of related to the problem at hand. Accompanying the publication of the IARC study in the October issue of JNCI is a very nice editorial by Blot and McLaughlin. They were two of the investigators in the study.

And they indicate in their editorial that they came at this problem from different perspectives, can I say opposing perspectives, and ultimately converged on a common opinion, a consensus opinion or whatever, relative to this particular problem.

And that opinion in the final paragraph of this editorial is that when all the evidence, including the important new data reported in this journal, is assessed the

PAGE 235:

inescapable scientific conclusion is that ETS is a low level lung carcinogen. Thus the reduction in risk of lung cancer following cessation of exposure to ETS in the IARC study is a hopeful sign and suggests that measures aimed at the reduction of smoking may benefit not only workers but also persons with whom then live and work.

But they did clearly converge on that. And they went through a very nice calculation of risk based on cotinine levels and that sort of thing which actually turned out to be very convergent with the relative risk from the epi studies. It came in at a value that--I don't know, it was about 1.2 or something like that, which seems to be consistent with the epi work.

I found this editorial to be very persuasive. I say this because it represents a type of testimony that is not represented physically at this meeting but represents another testimony relative to the data that is in the public literature.

Chairman Brown: Thank you. Other comments, please? Yes, Dr. Hecht.

Dr. Hecht: I will just expand further on that. They discussed in their article the uptake of cotinine and carcinogens, and there is ample literature on that now. And once can estimate that carcinogen uptake from ETS is about 1 percent of what it is from mainstream smoke.

PAGE 236:

And taking that into account, it is not surprising that it is difficult in the epidemiologic studies to really demonstrate the effect because the dose is much lower. So I think that what we know now is very consistent with an effect, but it is a small effect that is difficult to measure.

Chairman Brown: Eula.

Dr. Bingham: I would like to really thank the public presenters. I must say that I am confounded--I will use that word--by the presentation on the biomarkers. That was troubling to me to see that, you know, essentially they were the same number for the unexposed and the ETS exposed and then compare then with smokers. I would have thought we would have seen some differences.

Having said that, I was also brought to reality, because this certainly happens with workers when you are looking at epi studies, with the fact that we don't really have unexposed people as controls that this gentleman over here brought up, and you really have to look at people who do not smoke like the--is it the Seventh Day Adventists?

Mr. Repace: The Seventh Day Adventists.

Dr. Bingham: And when you take that into account, the epi data become more impressive if you look at their rates. So, you know, I found both of these presentations really quite impressive.

PAGE 237:

Chairman Brown: Thank you. Dr. Russo.

Dr. Russo: One thing that we need to really consider is that even though that the values between the unexposed and exposed population look the same, the most important element that has not been measured in this study is the total amount of substance that is produced in the target tissue, because most important is to determine how much, for example in the case of smoking, the mammary glands receive that could be related with cancer of the breast or the epithelial cells or the bronchial epithelium or other kind of epithelium, because the blood has 120 days turnover, especially the erythrocytes, and it is very easy to lose the marker.

But the most important element that is missing in all these studies is how the target tissue produced these adducts and how the target tissue removed these adducts or replaced or repaired the damage in these. And we don't have these values. Therefore, the values that have been presented today could be important but they are meaningless, because they really are not indicating to us the amount of damage that these substances are doing in the target organ.

From my point of view, the values are important from the point of view that the study has been done but really don't tell you anything, don't tell you anything, because of that, because dose values don't reflect the amount of

PAGE 238:

exposure and the amount of damage induced in the target tissue. And I believe we have to state that. I think the dose data are inconclusive.

Chairman Brown: Frank, did you have a comment?

Dr. Mirer: Yeah, a comment for the presenters. Basically we have just spent like an hour doing like an OSHA standard setting hearing. And it is a different--I mean, we are really in a different context. I don't see how we can take into account unpublished data that is presented here at the last minute without--I mean, it is true we have had copies in advance, but it is simply not possible to evaluate that kind of data in the face of a proceeding of this magnitude and rapidity.

Now, in particular we had two presentations that attach the validity of several of the critical--you know, the best of the case control studies where we had data presented criticizing the epidemiologic methods and the like. And I simply think we have to have some rules of the game, as at IARC, that if it is not published it is--we can't take it into account. We can't take into account things that are presented on this kind of--in this kind of setting at this kind of speed. And that is--I mean, there is a process here and it ought to be followed.

If the people who prepared these reviews think that the studies have been particularly damaged by these

PAGE 239:

presentations, they should tell us so we don't rush into something and we could defer action. But if they are not, I think we have to go forward with what has been published.

The second thing is I myself have found the epidemio-logic data persuasive. It is very important this call that we make today, or whoever makes the call--makes is very important because again we are trying to compare high dose to low dose risk rates for one of the know human carcinogens. And it ought to inform, I thin, in about all these carcinogens the same way.

And the third point is that I hope when we get to diesel we will get the same generous interpretation of epidemiology that we are getting here.

Chairman Brown: Don't lobby your cause. Other comments? Yes.

Dr. Hecht: I just want to return to Eula's comment about the biomarkers. That study showed that for the polycyclic aromatic hydrocarbons and the aromatic amines there was no difference between ETS exposed and nonexposed control. But for NNK metabolites, which is a tobacco specific lung carcinogen, the ETS exposed were significantly higher than the nonexposed. And that is a study carried out by industry which found exactly the same as what we found.

Chairman Brown: Any other questions?

Dr. Bingham: I think our troubling PAH one

PAGE 240:

wasn't bigger. It was still troubling PAHs didn't follow, but---

Dr. Hecht: Yeah, but I mean, non--the biggest exposures to PAHs is food; okay? So in the---

Dr. Bingham: (interposing) So?

Dr. Hecht: So in the controls you are going to have exposure through food.

Dr. Bingham: And they both eat.

Dr. Hecht: Yes, right.

Dr. Hooper: That masks the---

Dr. Hecht: (interposing) That masks any difference.

Dr. Medinsky: I just want to make a comment on the epidemiology. I guess the relative risks in this for environmental tobacco smoke are from my perspective quite low. And that was--that was troubling me this entire time. And I think James Repace's comments regarding the Seventh Day Adventists and if we could actually get a control group that was truly unexposed that the relative risk would be up comforted me quite a bit.

I think also I have a level of confidence because a highly exposed group, i.e., maybe a cigarette smoker, although, you know, the exact circumstances under which the exposure takes place is different, but a lot of the carcinogenic--the carcinogenic chemicals in the cigarette

PAGE 241:

smoke and the mainstream smoke are similar.

So I see some sort of a dose response for say a very highly exposed group, and then Steve's comment just a minute ago saying that, well, when you consider that how low these environmental tobacco smoke exposures are that the relative risk falls into line. So I think all that put together makes me feel a higher level of comfort for environmental tobacco smoke than I ordinarily would based on these low relative risks. I just wanted to--so you won't be surprised tomorrow with diesel.

Chairman Brown: Other comments?

Dr. Henry: I guess what was troubling me was the multicenter case control study, the one by Buffetta. And Shelia, you had commented that--because I think you had considered that in your review, and this issue for the length of time from cessation or perception of cessation from exposure to ETS might impact, because this study seems to reflect exactly what Michele is saying in terms of extremely low relative risks and, you know, that the technical material should make a difference to our decision. This would seem to have addressed many of those issues. So if you could discuss that a little bit more and---

Dr. Zahm: Yes. There is actually two points, both stemming from what you have just said. They do make a point in their discussion about the fact that they

PAGE 242:

have more distant smoking exposure than others. They don't quantify that, so it is hard to get beyond that base on the paper.

But one thing we were talking about are truly nonexposed people nonexposed, what is the background. Well, this is a--you know, this is a study in Europe. And I think there is more general exposure than in the studies in the U. S. So I think that could be the reason why, you know, the overall is even more dilute than what we see in some of the U. S. studies.

Dr. Henry: Well, Shelia, are you concerned at all about sort of the category we are talking about. I mean, this is nominated as know human versus reasonably anticipated.

Dr. Zahm: Well, as I said in my initial review, anything down in this range obviously you have to consider very seriously because I think it is at the limits of what we can look at. But there certainly are criteria that I went through that for me pushed me over the edge.

And there are other exposures that we do accept as causal that are down in this range, like--oh, for example, you know, we talk about direct smoking and the risk of heart disease is certainly below, too. It is a lower risk than certainly what we think of for lung cancer and smoking.

Dr. Frederick: Is Dr. Steven Sears in the

PAGE 243:

audience by chance? Okay. Dr. Sears, I have to say that your written comments really impressed something--were impressive to me in this sense. I have felt all along on this issue that we were in a situation where two people could look at a cow and one would call it black and one would call it white.

And as I read your comments and I looked at some of your graphs of the data and this and that, and you were interpreting those in a certain way and you felt those were convincing that this wasn't a major problem, I looked at those and they absolutely convinced me that, yes, it is a significant problem. So I appreciate your comments and they helped form my opinion, so I--but I think--we have weighted these things. We have looked at these things, but I think there is a certain amount of perspective in terms of the interpretation of the data there.

Chairman Brown: You identified an example of cognitive dissonance.

Dr. Frederick: There you go.

Chairman Brown: Other questions or comments? (No response.) Would you like to make a motion, Shelia?

Dr. Zahm: Sure; I move that we--do I have to read this? Let me see; where is the statement? I move that we accept the listing of environmental tobacco smoke as a known human carcinogen as stated in the summary statement.

PAGE 244:

Chairman Brown: Is there a second to that?

Dr. Friedman-Jimenez: Second.

DOWNLOAD THE DOCUMENT IN WORD 6.0 FORMAT

Chairman Brown: Second; Dr. Jimenez?

Dr. Friedman-Jimenez: Uh-huh.

Chairman Brown: Further discussion? (No response.) Hearing none, all those in favor of the motion, please raise your hand.

(Members raise their hands.)

Chairman Brown: 13?

Dr. Hart: 13.

Chairman Brown: That is unanimous. Thank you very much.