Tobacco Documents Online

theory, to facilitate this scientific process of integration . The way out for epidem~ology would be to explicitly recogn,ize the challenge, anddirect research to programmes that focus specifically on the envi, ronment- gene interaction. (2') Specific criticisms of using epidemiological studies to "prove" causation, include the following points: First, epidemiological studies can show a statistical association between factors, but they cannot demonstrate a causal relationship. Even the 196.4 Surgeon General's Report conceded this: "Statistical methods cannot establish proof of a causal relationship in an association." (p. 20) This is self-evident because "cancer is a biologic, not a statistical problem." (3) Second, in epidemiological studies smokers, ex-smokers and nonsmokers are not randomly assigned to their study groups. Instead, they make their own decisions about smokingi,~ and they, therefore, constitute "self-selected" samples. Such self-selection Can introduce considerable bias into an ep~demiological study. Third, many of the large-scale prospective studies examining the relationship between smoking and disease do not adequately account for the potential effects of other factors that can affect the results. These factors include general l~ifestyle, CONFIDENT IAL -- 9 --

alcohol consumption, occupational and envi.ronmental exposures,2 genetics,~ aging and the immune processes. As one statistician has commented: In general it is ironic that while studies of environmental or occupational health effects are constrained to take population smoking habits into account, studies on smoking effects seldom bother to search for a report on occupational exposures. (4) Fourth, if smoking causes lung cancer, it would be reasonable to expect more of the disease in countries where more cigarettes are smoked per capita. That does not always happen. For example, Austria, Belgium, and Finland report higher lung cancer rates but considerably lower per capita tobacco consumption than the United States, Canada, and Australia. (5) Fifth, if cigarette smoking does cause cancer, then the earlier a person starts to smoke and the more he smokes, the sooner he would be expected to get lung cancer. Yet nonsmokers and smokers (whether they smoke a lot or a little) all appear to develop the disease at approximately the same age. (14) According to one expert: "that both the age of starting to smoke, and the rate of smoking, should have no appreciable influence on the average age Indeed, some environmental exposures that now may be considered significant -- such as radon gas in the home -- were not recognized as potentially hazardous at the time the studies were conducted. CONFIDENT IAL

of onset of lung cancer greatly taxes, if it does not destroy, any causal hypothesis." (6) Similarly, one major study indicated that inhalers had a lower lung cancer rate than non-inhalers. Sixth, the same type of epidemiological information used: to suggest that lung cancer has increased in. smokers, suggests that it may also have increased in nonsmokers. In an April 1979 report on this phenomenon, a scientist who believes that smoking causes lung cancer nonetheless conceded that factors besides cigarette smoking must have had a "significant effect" on the mortality rate of lung cancer. (7) Seventh, if cigarettes do cause lung cancer, one might reasonably expect that heavy smoking would lead invariably to the disease. Yet, the studies relied upon by the industry"s critics show that less than 5% of all smokers ever develop lung cancer. From the first to the most recent hearings .regarding smoking and health, Congress has heard scientists and physicians express their skepticism about the relationship between smoking and lung cancer. In 1957, Dr. Joseph Berkson, of the Mayo: Clinic, testified that "there is serious doubt as to the scientific validity" of the conclusion that smoking causes lung cancer. (8) In 1964, Dr. Thomas Burford, Chief of Thoracic and Cardiovascular Surgery at Barnes Hospital in St. Louis, Missouri, said "carefu~l CONF]DENTIAL - ii --

study of the literature will reveal that many thoughtful studies by competent people concur in the opinion that the. cause of lung cancer, or of cancer in any organ, is unknown. Without needlessly belaboring the issue, it is obvious that there is a large volume of good scientific evidence which tends to refute the premise that cigarette smoking is causally related to cancer of the lung." (9) Almost 20 years later, in 1983~ Dr. Walter Booker said "Despite what those in the legislative arena might believe, the cause or causes of cancer of the lung (and other organs) remain unknown. Both smokers and nonsmokers contract cancer and other diseases often associated with cigarette smoking, and we still don't know why." (110)3 The opinion of these scientists that smoking has not been proven to cause lung cancer rests not only on certain of the generic criticisms of the epidemiological "case" described above, but also on factors specific to that particular disease. For example, ostensibly parallel increases in the 20th century between the consumption of cigarettes and the incidence of lung cancer have been cited as one piece of evidence that smoking causes lung cancer. Yet, both pathologists and thoracic surgeons have repeatedly Indeed, the fundamental biological mechanisms of cancer are still largely a mystery. As an a.rticle ~n the Journal of the National Cancer Institute recently acknowledged: "Although a large number of factors have been associated with the development of malignant neoplasms in humans, the mechanisms are still largely unknown." (Ii) CONF )DENT IAL

testified before Congress that the data on lung cancer trends over time are not dispositive because dramatic improvements in physicians' ability to diagnose lung cancer caused more lung cancer to be reported in official health statistics. As Dr. Rosenblatt said in 1965, "statistics showing a tremendous increase in lung cancer during the past 30 years are misleading. The increase is only apparent and is [in part] the result of greater skill in the detection of the disease." (12) Similarly, changes in the International Classification of Diseases in the mid-20th century have inappropriately combined data for cancers that originated in the lung with cancers that spread to the lung from another part of the body. Since majlor internal cancers frequently spread to the lung, these secondary lung cancers have been included in the statistics on. smoking and lung cancer. This combination of primary and secondary lung cancers in some statistics has made it impossible accurately to establish a statistical relationship between primary lung cancer and any agent, including cigarette smoking. In sum., the use of epidemiological data to indict smoking as a "cause" of disease has many flaws, and neither the industry nor independent scientists have acted irresponsibly in noting those problems in the public debate over smoking and health. CONF ]DENT |AL

o Animal Studies Those claiming that cigarette smoking causes lung cancer also rely upon the experiments by Ernst Wynder and other scientists who. claim to have produced skin cancers by painting the skin of certain animals (usually mice)with tobacco '"tar." For several reasons, such skin-painting: studies are not directly applicable to the human smoking experience. First, the mice used in many of the experiments have been specially bred to be extremely sensitive to any possible carcinogenic activity. They therefore may react positively to many substances which are not carcinogenic in humans. Further, in the 1950s, Dr. Jonathan Hartwell of the National Cancer Institute compiled a list of animal carcinogen studies which included such common items as fructose (fruit sugar), glucose (processed sugar), and even lactic acid (made naturally by the human, body). He found that 481 of the 2,.108 common chemical compounds tested (23%) were "carcinogens'" in such. animal studies. But, neither Dr. Hartwell nor any other responsible scientist ever concluded that these many "animal carcinogens" were therefore human carcinogens. Second, the substance that is applied to the backs of mice in such. skin-painting experiments -- condensate created by solidifying tobacco smoke and m~xing it with a chemical solvent -- CONF ]DENT

is not the same substance as the fresh whole smoke a smoker inhales. It is for this reason that Brown & Williamson and the other tobacco companies sponsored animal studies by independent scientists which more closely resemble the human smoking condition- - i.e., experiments in which the animals inhale fresh, whole smoke. Despite numerous animal experiments involving many different species, reproducible human-type lung cancers have never been found in the lungs of animals as the result of inhaling fresh whole tobacco smoke. Third, the organ involved -- mouse skin -- is obviously quite different from a human lung. For example, animal skin lacks the intricate clearance mechanisms of the lungs, such as the mucus blanket which coats the lining of the major airways of the lung. Even the summary report of a study sponsored by the United States government utilizing skin painting techniques conceded that there is an "uncertain relationship between tumors resulting from mouse skin painted with condensate and human lung cancer." (13)i Once again, all scientists recognize the importance of inhalation studies which could tend to support, or as has been the case, cast doubt on the skin-painting results. Finally, the doses applied~ in, mouse skin painting are massively greater than the amounts of particulate matter which a smoker would encounter in a normal smoking situati~on. Such. CONF ]DENT [AL - 15-

experiments involve applying '"the wrong material, in the wrong form,~ in the wrong concentration, to the wrong tissue of the wrong animal." (14) As Dr. Greene. stated at the 1957 hearings, "the [skin-painting] experiments demonstrated that tobacco tar extracted by a special technique induces cancer in the skin of CAF mice, and! nothing more. This point is of little significance to workers in the fiel, d who have found that under certain conditions a multiplicity of substances in everyday use will induce cancer in m~ce of highly susceptible strains." For these reasons, the industry has properly insisted that animal inhalation experiments are more relevant to the question of whether cigarette smoking causes human lung cancer. Over the years, many researchers -- both supported by industry grants and completely independent of any industry connection -- have conducted such experiments where test animals have been forced to take in. large doses of whole smoke. One study recently concluded by Microbiological Associates (with funding from the Council for Tobacco Research) involved thousands of animals, in a lifetime experiment. These studies have not found any significant pathological changes, and none have found credible evidence of cancer CONF ] DEI~T, IAL -- 16 --

production.4 As Arthur Furst, the Director Emeritus of the University of San Francisco's Institute of Chemical Biology, said "For many years, I tried to induce lung cancer in animals with cigarette smoke, with no success, despite the most sophisticated smoking machines available. Not only were my colleagues and I unsuccessful, but every other investigator who. attempted to induce lung: cancer in animals by inhalation of fresh smoke also fail, ed." Thus, he said: "I have concluded that no reliable, reproducible animal studies have shown that the inhalation of cigarette smoke causes lung cancer." In sum, the totality of the evidence fails to establish that smoking is a cause of lung cancer. B. Cardiovascular Disease Independent scientists have similarly expressed skepticism about the alleged causal link between smoking and heart o One of the most thoroughly discussed animal studies was the beagle dog inhalation study conducted by Dr. Auerbach. Dr. Auerbach claimed to find changes which he labeled as cancerous, but his study was severely criticized. Dr. Victor Buhler stated in 1982: "That experiment suffered from. severe design defects, and the photomicrographs published with the article would not permit most pathologists to reach the conclusions stated by the authors." Significantly, Dr. Auerbach and others were unable toreplicate his results in later studies. CONFIDENT I~AL -- 17 --

disease. They have noted the inconsistent statistical correlations between smoking and heart disease in various populations; the strong: genetic influence in heart disease; and the failure of the critics of smoking to establish a mechanism to explain how tobacco smoke might cause heart disease. In 1964, Dr. Henry Russek testified before Congress that "statistical findings linking the tobacco habit to coronary disease afford no proof of a causal relationship.;" and that his research had shown that emotional stress was far more significant than cigarette smoking! in the development of heart attacks. (15) In 1982, Dr. Carl Seltzer, a Fellow of the American Heart Association's Council on Epidemiology, similarly said that the statement "'cigarette smoking is a major cause of heart disease,', is not scientifically valid." (16) Anti-smoking advocates blame nicotine for the development of heart disease. Yet no mechanism by which nicotine, or any other agent, is involved in heart disease has been demonstrated. Serious questions about what role, if any, nicotine plays have been raised as a result of autopsy findings of fatty deposits and other changes in the arteries of individuals who either have not smoked or could have smoked only briefly, such as infants, children, and young men killed in battle. (17) Even the 1983 U.S. Surgeon General's Report, which focused on cardiovascular disease, concedes that "the evid~ence for and against a primary role for nicotine in the development or acceleration of atherosclerosi, s is not conclusive." CONF [DENT [~AL