Document re0qgNBRB3R8bOazJedyVLKXa

<7/rA^S ^ yoL 284 27 March 1980 exten distorting the epidemiology of cancer: attitajc the need for a more balanced overview 'ritubl id bH little^ were* Idbcl ar tot ausc 3 the* aardVl The politics of cancer are becoming increasingly polarised with environmentalists taking one extreme view and industrialists the other. Richard Peto argues that a more dispassionate and disinterested approach is required for a realistic assessment both of the risks from environmental carcinogens and of the preventative regulations needed There exist, both in the general environment and in certain occupations, chemical or physical agents which increase the likelihood of human cancer. Our political response to this would, in an ideal world where sufficient knowledge was available, depend on the direct and indirect costs of the various possible measures of control, and on how many cancers each such measure would prevent. In the real world, estimates of the direct financial costs of the control of particular agents can easily differ by one or two orders of magnitude, the indirect costs may be grossly exaggerated, or some major indirect costs may be completely overlooked. Wont of all, we have in general no remotely reliable estimates of the numbers of cancers which particular legislative controls would prevent. Consequently, there is wide.scope for pressure groups to have considerable influence on public policy. Historically, the most powerful pressure groups in US society have been the large financial interests, which have almost always put financial advantage before human health. Over the past few yean the major tobacco companies have launched massive sales driver in the Third World, which if successful will loll millions of people. (About one in four Tegular cigarette smoken is killed prematurely by smoking.) In the US, where cigarettes cause well over 100,000 deaths every year from lung cancer or chronic obstructive lung disease, the tobacco industry refuses to accept in public that cigarettes cause cither disease, let alone to collaborate in serious public information, and no epithet can suitably describe the collective efforts of their advertisers. No other industry kills people on anything like the scale that the tobacco industry does, but where other, industries have been found to cause cancer (or dust* induced lung disease) in their workers or in Richard Peto is a reader in cancerstudies at the University of Oxford. This article developed from a review of Samuel Epstein's hook. The Politics of Cancer the consumers of their products, their immediate response has usually been to delay acceptance of the findings, to minimise their relevance to current practice, and in general to delay or obstruct any hygienic measures which will cost money. Even when human danger has been unequivocally demonstrated, industrial consortia may actively lobby for controls so weak that (as with the new UK government regulations limiting inhaled asbestos to 1 fibre/ml from 1981) they leave no reasonable safety margin. Large amounts of money are available to mount press oi TV publicity campaigns about the homely apple-pie virtues of asbestos, journals financed by the tobacco industry run populist articles which misrepresent research results to lay readers, and some US television journalists are explicitly told always to censor all reference to the dangers of smoking. Even the scientific literature is not immune from distortion by financial interests. The decades-long argument that "threshold" dose levels of carcinogens must exist below which the general population is absolutely safe has not been entirely motivated by the scientific plausibility of the hypothesis. With increasing understanding of the derivation of tumours from single cells acted on by mutagenic carcinogens, industry is slowly abandoning "threshold" arguments in favour of arguments1'1 (where the biological fallacies1'4 are somewhat better concealed by the mathematics) that thousandfold reductions in dose can conveniently be "statistically guaranteed" to produce a million-fold or some other enormous reduction in risk. Even if the scientists who propound such models are disinterested, industrial endorsement of them is not. Much excellent toxicology may be done by industry, and many industrial scientists and managers may be directly and honestly concerned with the prevention of hazards. But so many examples of financiallymotivated bias exist that the motives and work of industrial scientists and consultants are inevitably distrusted. Over the past quarter century, the "environmentalist" camp in North American society has pressed for (and achieved) stricter control of particular carcinogens in the work-place, the general environment, and the diet. Usually, their efforts have been directly resisted by the relevant financial interests. Inevitably, in view of the scantiness of reliable quantitative knowledge about the cancer risks to man of nearly aO carcinogens, the argument has polarised. In the politics of cancer, each side takes a very extreme position. Industry usually argues for the irrelevance to man of animal or in vitro cancer tests, or minimisks the quantitative hazards and exaggerates the costs of control. The environmentalists usually exaggerate the likely hazards and are largely indifferent to the costs of control. `The most powerful j pressure groups in the US, have been the large financial interests which have almost always put financial advantage before human health* The vacuum of reliable scientific know ledge is such that each side can find scientists who will maintain in courts, in publichearings or in thescientific literature whatever is politically convenient, and it is important to recognise that scientists bn both tides of this debate now have career interests at stake in it. A particularly good example of the biased writings of politically active environmentalists is Samuel Epstein's The Politics of Cancer (Sierra Club Books, 1978, revised in Anchor Press, 1979). Epstein outlines, for university-educated but not necessarily science-educated readers, his view of our present state of knowledge aboutpreventionofdeath from cancer. He concludes that the cure rates for the major cancers have not been improving much over recent yean, but that we do know enough now about the causes of cancer for the testing and regulation of environmental contaminants to prevent the majority of US cancer deaths, and that the main obstacles to our doing so are political rather than scientific. He therefore reviews the scientific and' political circumstances surrounding a; 29$ Nalurt Vot. 2&4 ij Mann . dozen or so disputed consumer products or occupational hazards, providing much fascinating political detail, and then describes at length the internal politics of the various agencies which research,' regulate and litigate in Washington. Epstein's book has been written to inflame political passions against environmental carcinogens, and parts of it are well worth reading. However, the political punch is often achieved at the expense of scientific accuracy and balance. Despite this, the book has already gained wide and apparently uncritical acceptance even among scientists. For that reason, it is worth considering some of the misrepresentations of scientific evidence . which it contains, and some of the more general defects in the environmentalist perspective on cancer. First, a few details. Epstein's book is have refuted it (see table). This appears to be a deliberate attempt to deceive the reader. It is not a casual slip in a 600-page book, as Epstein devotes twenty pages and two full-page tables to saccharin. His entire table 6.4, entitled "Tumours other than the bladder in rodents fed saccharin", appears to be so subject to artefacts of selection that it provides no evidence that saccharin does cause any rodent tumours other than in the rat bladder. Saccharin is not an isolated example of bias; indeed, I found that in many places where he discussed data with which I was familiar inaccuracies were present, almost always in the direction of accentuating the need for battle with the devils of industry. Turning to more important matters, Epstein asserts that any benefits from the "less dangerous" cigarettes which the tobacco industry has developed will be 'Scientists on both side$*of the environmentalist debate now have career interests at stake* merely be acting as a vehicle for fat-solnkt pesticides and industrial chemicals, whu predictably emerge as his villains yet spj? However, bis suggestion is implausibfc countries with high levels of contamination do not stand out from the general cancer/fat relationship, colorectal sad breast cancer were common longbefore widespread use of pesticides, and-:a* experimental animals high-fat or low^ diets can greatly enhance or reduce the ru^ of cancer11. Something more interesting!-! waiting to be discovered in our diet thaal simple contamination by carcinogeai.^s[ very useful as a source of reference to outweighed by people increasing their The fat-associated and smoking-derived? original papers, but it is not in itself a consumption to get more nicotine, and he cancers collectively account for more thsas reliable secondary reference because the is therefore in many places particularly half of all cancer deaths, and peopled _ material presented is so often distorted. scathing about (or even downright opposed concerned with cancer politics should trytat ~ Sometimes this distortion is due to the to) research into changed cigarette understand them considerably more* effects of Epstein's vigorous campaigning composition. This is a distorted accurately than Epstein appears to. One' style on bis scientific judgement and is, . prerspective on one of the more promising can list many other instances where] perhaps, forgivable; at other times it immediate means of preventing fatal Epstein's content or style are misleading or? appears to be deliberate, which cannot so cancers, a quarter of which in America are unbalanced. For example: - easily be forgiven. currently due to smoking. The discussion of alcoholism and canco^ For example, consider the (by now Smokers of "less dangerous" cigarettes is entirely erroneous because of failure to generally agreed) finding that incor have already been found in various standardise properly for age. poration of about 5ft by weight of epidemiological studies to have disease One of Epstein's chapters endsv saccharin into rat diets gives a few rats rates which are materially lower than "Eisenhower warned against the growing bladder cancer, but has no generally smokers of other cigarettes**7; at autopsy national threat of the military-industrial accepted effects on any other type of they have far fewer "pre-malignant" complex. The medical-industrial complo cancer. By human standards, 5ft repre histological changes in their bronchi* and, now appears to be as serious a threat". t i sents a vast saccharin intake (and so, of perhaps due to the changes in cigarette There is idealisation of the value oflong course, the industrial "Calorie Control composition 10 or 20 years ago male lung term animal tests with concomitant Council'' have tried to shrug off these cancer death rates in early middle age are denigration of the value of the Ames tea findings). To refute the common reaction now decreasing in North America, in and other short-term tests. This is one o. that "anything given in large enough doses Britain and in Finland. Epstein's most inexplicable errors o anil cause cancer in animals", Epstein's After lung, the next commonest fatal scientific judgement, unless he warn ' chief argument is to report that 0.01ft cancers in the US are those of the breast cancer tests to be difficult and expend? saccharin has also been shown to be and the large intestine, for which the most for industry. carcinogenic, la support of this extra striking epidemiological finding to date is a He is irritatingly puritanical, sneering ordinary claim, be presents in tabular form 90ft correlation between fat consumption "tiie plastic age which symbolises bowC the control and 0.01ft data for selected in different countries and their breast or value of our lifestyle has been degraded* cancers from certain multi-group feeding colon cancer rates (see figure; similar He seeks by stylistic tricks to attribute experiments, leaving out the observations correlations exist for' colon cancer10). oral contraceptives some of the establish from those sameexperimeats which would Epstein suggests that dietary fat may hazards of hormone replacement the and of diethylstilboestroL . yt Table Data from OTA report* on the tumour sites selected for presentation in Epstein's book to He claims that cancer costs the substantiate the claim that 0.01 ft saccharin is carcinogenic. Croups selectively omitted in Epstein's economy over $25 billion per year. (In f book are marked with an asterisk.These omissions substantially alter the implications ofthe original / cancer must be prevented for humane, data. for economic reasons; without can Saceberta do** (ft la rat diet; tbase dosca bava * do malarial affact oa longavtty) Male or female lympbotaicomaa (FDA, 1948) Female breast (FDA, 1973) Mite breast (FDA, 1973) there would be three or four million n retired North Americans to supp costing over $25 billion per year). And so on. He seems so certain a- everything--how can anybody bejust 0 (control) 0.01 0.1 0.5 1 5 0/20*(0ft) 1/14 (57ft) 5/16 (31 ft)* 2/15 (13ft)* 1/18 (6ft)* 10/17 (S9ft) 6/26 (23ft) 14/30 (47ft) 13/34 (38ft)* -- 12/30 (40ft) 12/27 (44ft)* 6/29 Olft) 14/25 (56ft) 9/27 (33ft)* -- 8/27 (30ft)* 7/25 (28ft)* *Omiued in Epstein's book. 4 In addition to these 20, since some other sweeteners were being tested concurrently, 34 other control animals were studied in the same experiment, end the 1977 OTA review of these data, considering all the control animals to be equivalent, died 9/54 control lymphosarcomas. (0/10 saccharin controls of each sex were studied, not 0/20 as Epstein inadvertently indicated.) in being so certain about so many de? Sometimes I know he's wrong, but t often (especially in the many places v inconclusive scientific results are prefc as established fact) 1 know that no knows for sure. Lewis Thomas once v in another context: "The scepti. medicine have a hard time of it. It is more difficult to be convincing ignorance concerning disease mech. . ciwoh--ims*- VoL 284 ^7 March 1980 ___________________________ 299 ^.n it U to make claims for full comprebensioo, especially when the ^prehension leads, logically or not, to jooie sort of action. When it comes to serious illness, the public tends, understandably, to be more sceptical about the sceptics, more willing to believe the true believers." Epstein (like many others) is or food additives. Indeed, one of the most intriguing animal findings is the effect in many studies of gross aspects of diet on cancer; for example, mice randomised between Sg of food once daily and ad lib consumption averaging 6g/day of thesame food, had respectively 8W and 644b spontaneous mammary tumours14, and the role of dietary fat11 has already been `Present day cancer rates and trends are probably not dominated by occupational carcinogens or environmental pollutants' unjustifiably definite about three major discussed. issues. He seems certain that even after the Thirdly, Epstein seems certain that at both of whom still seem more ready to effects of cigarettes have been allowed for, least a quarter of all cancer deaths are defend than to repudiate it. Americans live in an era of rapidly attributable to occupational exposure to Entitled "Estimates of the fraction of increasing cancer rates. But trends in carcinogens. There is as yet no good cancer in the US related to occupational recorded cancer death rates (and, perhaps evidence for hazards of anything like this factors", it shows how a group of more so, recorded incidence rates) over the magnitude, and there is dear evidence that reasonable men can collectively generate past quarter century are biased upwards many of his particular claims are an unreasonable report. The group (since the cure rates for the major cancers exaggerated? For example, he says that multiplied the 12 million workers exposed have not materially improved) by asbestos has led to approximately 50,000 in any way in 1972-4to any ofasbestos. As, improvements in medical care and cancer deaths per year in the US from cancer and Ni, Cr, benzene or petrochemicals by the counting. These improvements have related disease (referencing a paper by very large risk factors (e.g. fivefold lung Sdikoff11 which actually neither makes nor cancer excesses) typical of much more `The NCI-NIEHS report shows how a group of reasonable men can implies any such claim). This is sheer nonsense; of 50,000 asbestos deaths, several thousand should be certified as being due to pleural mesothelioma, yet extensivelyexposed industrial populations, finally multiplying again by "4-5` ' to allow for the additional cancers that will be caused by these six agents among workers collectively generate an unreasonable report' only several hundred per year are thus certified. Likewise, Epstein dtes with approval exposed to them at timesotherthan 1972-4. (In casual disregard of dosimetry, this whole process is very like predicting (indeed, be describes it as being of 100.000 occupational lung cancer deaths affected all sectors of US society, but most "epochal significance") Health, per year among non-smokers due to their 2 tO particularly old people andpreviouslypoor Education and Welfare Secretary exposure to other people's cigarette smoke people, especially blacks. Epstein, along Califano's absurd 1978 estimate, in a at work, "since 209* of heavy smokers get with most other US commentators, does speech to labour union leaden, that as hmg cancer"). not allow for these biases even to the limited extent of restricting his attention to trends in age-specific mortality rates among middle-aged whites, where they might be expected to be least prominent. Among middle-aged whites during the past quarter century, cancercurerateshave not materially improved, but some US cancer mortality trends seem to be genuinely down (stomach, cervix) while some seem to be genuinely up (pancreas, melanoma, and certain lymphomas). I can discern no net pattern other than that due to the massive effects of smolring on lung cancer, although no extension to 1978 yet many as half of the 8-11 million workers who have been exposed to asbestos could develop serious diseases such as cancer or asbestosis. In his second edition, Epstein draws extensively on Califano's source, a curiousbutextremely influential document which has been circulating privatdy for the past year or more. This unpublished typescript, with no listed authors, was prepared by a working group of nine wellknown scientists inducting Arthur Upton and David Rail, the National Cancer Institute and National Institute of Environmental Health Sciences directors. These assumptions "predict" over 200.000 extra respiratory tract cancers per year in the near future due to the above six agents alone (and are the sole source of the currently widespread prediction of "20-384b" of cancers being due to occupation). But, in reality, fewer than 90.000 males/year get respiratory tract cancer in the US (and most of these are due to cigarettes and not to occupation at all), and moreover the US male rates in the first 25 years of working life are currently decreasing! A prediction of 100-200,000 extras due to the above six agents is clearly exists of the excellent 1950-67 trend Figure: data? plotting for 39 countries age-venus total dietary fat (animal or standardised breast analyses in NCI monograph 33. In Britain, cancer death rate vegetable). the same is true11 as long as we examine mortality in middle age. (Interestingly, although Epstein emphasises that bladder cancer death rates among whites in the industrial north-eastern US are high, this may not be chiefly due to any current industrial hazard, for they have been decreasing both relatively and absolutely for a quarter of a century11.) Epstein also seems certain that the majority of human cancer is caused by chemical and physical agents in the environment and could be prevented by &eir testing and regulation. There is no sound scientific basis for this certainty. Since the cancers (lung, breast, large intestine) which are commonest in the US are rare in certain other countries, and vice *rar, they probablyreallyarepreventable, hut (see figure) not necessarily by 'emulation ofany environmental pollutants eSKMU/an/i, !- MM very unrealistic. The unreality of the whole method is further highlighted by the fact that the assumptions "predict" 7,300 x 4.5 nickeloxide induced occupational lung cancers/ year. Since nickel refining causes a lot of nasal as well as lung cancer, we would presumably then also soon "expect" something like 13,000 nasal cancer death certificates per year among nickel workers. Nickel has already been widely used for decades, so many thousands of these should presumably already be evident. In reality, each year in the US population there are only 300-odd male and 200-odd female nasal cancer death certificates. The majority of these are not ex-nickel workers and no marked trends are evident. This demonstrates the unsoundness of the NCINIEHS methodology. Unfortunately, such exaggerated estimates are so much what many people want to believe of modern society that they have now achieved a life of their own, and -although they are utterly without foundation they are quoted widely, repeatedly and reputably both in the lay and scientific press. In Geneva the International Labour Organisation reportedly endorsed them, as did the Toxic Substances Strategy Committee in their recent draft report to the US president14, and they have been used in OSHA to emphasise the need for stricter laws. Most recently, the UK trade union the Association of Scientific; Technical and Managerial Staffs (ASTMS) released a policy document which appears to have been strongly influenced by Epstein's book, and which is notable for its prominent but erroneous assertion, derived solely from the foregoing unpublished US typescript, that 20-404h of current UK cancer deaths are caused by occupational carcinogens. Epstein now claims that these exaggerated . estimates seriously underestimate the impact of industrial carcinogens. There is obviously a valid political need to exaggerate the importance of the preventive measures we can already `For most toxic chemicals we now have both qualitative and quantitative uncertainty about the health benefits of restriction' implement, but in the long run we do also still need to discover the main preventable causes of cancer other than smoking. Bearing both these needs in mind, there are limits beyond which it is not even politically wise to distort the science of cancer, and some of today's exaggerations may well transgress those limits. Reliable reviews of the known and suspected causes17-1*'1* of human cancer and of trends in British oozi-oiwio/mocM)3soi.oo death rates12 should be more widelyknown. My criticisms of Epstein's science, however, must be viewed in the light of the continued resistance of many industries to reasonable controls. One has only to read some of his descriptions of industrial behaviour to see where his passion comes from, and a suitably sceptical reader could derive much important information from the dozen or so detailed case-histories of particular carcinogens that make up the bulk of this book. But, although I cannot prove it, I suspect that environmentalists as a whole (including Epstein and those US and British trade unions which have already modelled their public statements closely on his book) would be better `The politics of cancer is dominated on both sides by exaggeration' respected and could therefore press more effectively for controls in particular instances if their overview of cancer was more balanced in the following ways: if they accepted that present-day cancer rates and trends are probably not dominated by occupational carcinogens or environmental pollutants (especially since they could still warn that future rates might perhaps be so dominated unless presentday exposures are regulated21); if their chief concern was with the identification and control ofthe few major determinants of human cancer (or, in an industrial context, the few major industrial carcinogens) rather than the large multitude ofsins which Epstein denounces; if they discussed the costs and benefits of research into the control of toxic substances in the context of other possible ways of improving longevity (bad diet and tobacco in rich countries, malnutrition and infective and parasitic diseases elsewhere); if they took the costs of imposing restrictions on society more seriously. Epstein's general assertion is that the total direct and indirect costs of failing to regulate usually exceed the real costs of the testing and regulation, but I suspect that this is a slogan to avoid political embarrassment rather than a carefully researched conclusion; if they accepted that for most toxic chemicals we have both qualitative and vast quantitative uncertainty about the health benefits of restriction. Particularly, even if the sort of grossly biased interpretation which Epstein applied to the animal saccharin data is avoided, the results of animal experiments simply are not an uncomplicated key to human hazard identification. First, moderate alteration of gross aspects of the diet of animals greatly modifies their spontaneous tumour onset rates11 (so, for example, increasing the sugar content of rat diets may increase their cancer rates more than the equivalent amoum dietary saccharin would have done!* Also, any chemical which causes protif^ I ation or necrosis in any organ that- subject to spontaneous cancers is likely Lii modify the onset rate of tumours in organ, and since the aim of most animJ experiments is to study a dose which 3 nearly, but not quite, sufficient to caui* significant weight loss or mortality withinthree months, it is not surprising that iq many chemicals at such doses can cancer in animals. However, it may be that where ad politics operate one needs views at extremes in order to get a balanced outS come. And there is a possibility that over, emphasis on the avoidance of sdenti error would emasculatethe environmem ist passions and merely lead to my own] rather inactive conservatism. (After all, scruples about scientific certainty of safety . usually precede the widespread introduce ) tion of new chemicals, so should the ' imposition of prudent restrictions require absolute proof?) > Environmentalists with biased judge-.! ments and a quasireligious certainty of . right will fight more battles than any*-, reasonable sceptic would do, and even if; their victories confer 10 or 100 times leuj' benefit on humanity than they imagine, ? they will id the long ran probably do moregood than harm -- unless they materially < reduce food production, distort research-; priorities, or directattention away from the ' smoking problem in the process. Epstein- justly quotes, from the 1st century AD, the Plutarch chronicles: "He who in time of faction takes neither side shall be dise* Si*. paft< cons lOU' *0. t to b com ha*c cons idv; like) this that not Ir it is umJ ior free arc to< * hi. h. ir. T U fi rr rr 111 shv tha ma pre pre inti his are we ih see frachiscd". So, I fed, should bothenviron- mentalists and industrialists who suppress ', hr deliberately misinterpret data, but they > probably won't be. For the moment, the;; `politics of cancer* is dominated on both] sides byexaggeration. 1. GoraltaM. 1. Srtnwr 198,MS (1977). 2. Mawal.N.nHiliibt--M-A. 1379(I97SV S. CmphLlalCNvnanlM,Sninia:^ 4. Ci--.W-A-WWCuwe-->rS7.)47S(l9W). J. Otto of Tutinalpp In--I or At US Cm-- (irntc--rtoiv r--pmwsn*--. -it . ---------* -'* -------------- M 3*-- Hwbor NY (I9H). 7. WyMto.EJ-ffafiaUSt--mDHBW NIH 74-1221 (IVW. - 5. Aoaabock. O.WWMI-- . CaaoB.KJLGoooaritmoM*JS,3374(19?n. ~>4i 10. An--.BJC.atoDea.i.toJ.Ca--'lS.MIO*'*-? 11. ioao D.G. MtoWOar K*. St (I979). ftow* to aUito 1979b--oftto)owtoanof9W.{ u. ions. ee.s ptor., IfTS Ctmar a-- tom tooto a-- v.'Wj' 12. Doe. a. fnt.HtrSac.Umd.rn> MS, 47 (1971). IS. Blot. WJ.toFi--to.SJ.J.NmLCmwMat M. . (ten). * M. Boo. F.I.C. ud Tackar. MJ. --w Mtoo (tor Coot-- Sarto) 111. 171 (1974). IS. SattotT. I.J. io'-OriotoWlliaa Cn-.C*4Srto V Harbor NY (1977). 14. Normaa, A. 7towr2SO.cz> (1*79). 17. Doll. X. Nmmrt MS. S*(l*r7). - It. D0U.8.A--otoCo-- HU. 13 (1979). 7 19. 1ARC rorbiof groo* Caartoi* 4*. I (1900). 20. Dwv, D.L. and Maoto BJt. Snaocr M4,1)34(1979). I* V* \ m sin int on. de: ire co ac * C 90: ClOOOMirantaa JoWbU4;^| t