Document aDzk3RVGQ7Gxv746jNgaaDXY

nhSiho Sci rice ~ /2xS_. \. - r Brain tumor risk in petrochemical workers Epidemiologists struggle to represent the result of practices long since outmoded. Many naive and sort out tenuous evidence dangerously sloppy procedures of that points to elevated several decades ago have been elimi nated. incidence of brain tumors But sloppy procedures still exist, and other procedures that are gen among petrochemical workers uinely thought to be safe well might be otherwise. Thus, the petrochemi cal industry--and any other industry, for that matter--must rely on im Jeffrey L. Fox perfect early warning systems, such as CAEN, Washington those mustered by epidemiologists, to ferret out potential health hazards in The current effort to show a link be the workplace. tween exposures to petrochemicals Epidemiologists often refer to their and the incidence of brain tumors preliminary surveys as "hypothesis among workers in the petrochemical generating" devices. The current industry is stretching epidemiology to hypothesis, being voiced by epide its limits. miologists working for the Occupa Some critics even argue that epi tional Safety & Health Administra demiologists risk a tumble by leaning tion, for the National Institute for somewhat beyond the reasonable Occupational Safety & Health, and limits of their science. This possibility for the National Cancer Institute, is was apparent at the recent workshop that the incidence of brain tumors sponsored by the New York Academy among petrochemical employees is of Sciences at which was assembled a anywhere from 1.5 to 2.5 times higher blue-ribbon group of epidemiologists than the incidence in the general and other authorities familiar with population. Several other epidemi brain tumor incidence in humans and ologists, some of them employed by in experimental animals (C&EN, the petrochemical industry but others Nov. 3, page 6). Some observers at the without such ties, do not find evi workshop could not help concluding dence for elevated incidence. These that establishing a firm link between disagreements were by no means brain tumor incidence and employ reconciled during the New York ment in the petrochemical industry Academy of Sciences workshop. may prove very difficult. Moreover, "Polarizations came to mind dur even if established, that link may ing preparation of this meeting," says workshop organizer and epidemiolo gist Irving P. Selikoff. "We're con scious of this. But we were taught it | was good to find causes of disease. | Now, when causes in industry are 3 found, there's no automatic pat on the back." There are also no auto matic pats on the back when someone finds the obverse, no evidence for causes, he could have added. Epidemiologic evidence regarding brain tumors is anything but straightforward. Many factors could serve to show the intense analytic ef forts under way at the Texas petro chemical plants, including facilities run by Union Carbide in Texas City and Dow Chemical in Freeport but also at other installations owned by Texaco, Mobil, and Gulf Oil. For instance, the incidence of brain tumors in the general population in Robbing: bring tha best minda logalhar the U.S. is on the rise, according to Maltonl: doaaga and routa ara Important Lawrence Garfinkel of the American Cancer Society. The rise is not a sharp one; it applies more to men than women; and it follows no immediately obvious pattern. Also, the apparent rise in incidence might be due, in part, to improved diagnostic procedures and to better reporting of cancers. Quibbling about statistics and how to analyze them is just the sort of ac tivity that preoccupies many of the epidemiologists now scrutinizing the petrochemical industry. There are plenty of good reasons for the quibbling--scarce data, reliance on old records of questionable validity, and sharp differences on how to ap proach an issue are all factors, "It is clear there is a rise in the risk of brain cancer," asserted NIOSH's Anthony Robbins at the beginning of the workshop. "From the point of view of the workers, we've been too slow in recognizing that risk. We are attracting attention to this issue [by having a public workshop]--not to blame anyone. But until you bring the best minds together, you don't solve the problem." What Robbins said was echoed often during the three-day meeting, particularly by his colleagues from NIOSH and OSHA. Their concerns, quite properly, revolve around the principle of ensuring the health of workers. But those concerns can act like a centripetal force, continually drawing one's attentions back to UCC 096440 Nov 10. 1980 C&EN 33 Animal studi s provid troublesome hints Ideally, the results of carefully controlled animal studies of carcinogens will complement epidemiology, confirming suspicions about humans by referring to systematic data about mice and rats. But a good case may be made that animal studies are sometimes mis leading and can be prejudicial to the epidemiologist. For example, humans might be more, less, or altogether in sensitive to a particular chemical that induces cancer in rats and mice. How often such discrepancies crop up Is not so important here. The point is that the existence of bad or even good animal studies can send epidemiologists off on an oriented search, but a wild goose chase nevertheless. To anyone who is strict about the conduct of scientific Investigations, the question of initial biases--of whatever nature--is bound to be troublesome. Thus, conscientious epidemiologists are likely to find themselves trapped m a paradox: wanting and needing helpful clues about enigmatic human health problems, but also wishing to avoid one Of the worst pitfalls in science--namely, relying on false premises. There is plentiful animal data for the epidemiologists searching for causes of human brain tumors to review. Scien tists have identified several dozen chemical compounds that can cause brain tumors in various animals. In cluding mice, rats, hamsters, rabbits, and nonhuman primates. When tested, however, many of these chemicals do not cause cancer consistently from one type of animal to another. Moreover, the anatomic site of tumors caused by a particular agent can var< from one species of animal to another This lack of consistency is fairly commonplace in animal testing programs, it's a compli cating factor that helps to muddy oth erwise limpid clues now available. Not all brain tumors are alike. The major kind of tumor being followed In the epidemiologic studies is called a glioma. The spontaneous incidence of gliomas in mice is "exceedingly low," according to Italian toxicologist, Cesare Maltonl of the University of Bologna. Thus, the in cidence of gliomas arising after expo sure to various chemicals is a relatively reliable test system, he says. Nonethe less. that incidence also varies ac cording to how the chemicals are ad ministered to the animals--by feeding, injection, implantation, or inhalation. "It is important to know the dosage and route by which a carcinogen gains en trance," Maitoni emphasizes. Other observations of animals un doubtedly will complicate the assess ment of the human health question. For example, there is the matter of varying susceptibility to particular cancercausing agents during an animal's life span. "Embryos and newborn animals are far more sensitive to many agents than are adult animals," Maitoni notes, referring to the incidence of central nervous system tumors. This age-dependent variation in sus ceptibility is pronounced in Erythrocabus patas, a species of reddish, African monkey being studied by Jerry M. Rice and colleagues at the National Cancer Institute in Bethesda. Md. "We cannot get central nervous system tumors by Injection of carcinogens into adults, " he says about such animals. "They need to have very early exposure during gesta tion to develop tumors later, two to three years after birth." To Rice, these results suggest that there may be periods in life when hu mans are particularly or peculiarly sus ceptible to developing certain kinds of tumors. Thus, Induction of brain tumors in adults "might have occurred very much longer In the past," he notes. What ail this means regarding work place exposures to putative carcinogens and the incidence of brain tumors is difficult to say. And when these obser vations are tied into a widely held view of how cancer develops, the story looks even more complex. Many scientists believe that cancer develops in two phases, called initiation and promotion. An early event, initiation, caused by one type of chemical, may not causa a tumor unless followed by exposure to (usually) other types of chemicals, called promoters. Thus, the search for a single type of workplace exposure to explain brain cancer inci dence might be misguided. Imagine the following situation. One group of adults, when children, either had a particular disease, took a certain medicine, or was exposed to some other "initiating agent," whereas another group of adults avoided that agent. Then during adulthood, members from both groups were met with an onslaught of many "promoting agents." Some such promoting agents might be found in the workplace, particularly when that workplace is a petrochemical facility. But other kinds of workplaces and alto gether different kinds of environments also might provide some of the adults with deleterious promoting agents. Sorting through all these possibilities to identify the culprit agents could prove a formidable challenge. Studies of animals are Introducing further complications. For example, some scientists contend that virus par ticles are associated with chemically induced brain tumors in mice. Other scientists dispute the importance and even the existence of such particles. Their importance to human brain tumors is unknown. Other observations are puzzling. Why, for instance, does a particular chemical produce primarily brain tumors without affecting other organs? Access of the chemical to such organs partly accounts for their differing sensitivity. But other metabolic influences, including either different activation or detoxification in various organs, also could be important according to Paul Kleihus of the Uni versity of Frieburg in West Germany, Thus, for instance, enzymes in the brain cells of rats are slow to rid DNA there of certain kinds of damage. Long-lived damage might lead to brain tumors dur ing the later life of such animals. By far, the most puzzling phenome non is one noted by Maitoni and his colleagues. Varying the dose and the means by which it's administered for some carcinogens can cause system atic fluctuations in brain tumor incidence in animals. Thus, an elevated dose of vinyl chloride, for example, induces few (or no) brain tumors, whereas a lower dose may induce many such tumors. This "masking" phenomenon departs significantly from the typical observation that tumor incidence rises directly in proportion to the dose of carcinogen an animal receives. If such a phenomenon pertains to human brain cancer, the search for pertinent workplace hazards may be tortuous Indeed. workers and their environment. Thus, well-intentioned beliefs can predis pose one to look for clues more zeal ously in one place than in another. This is not meant to say that OSHA or NIOSH officials are tailoring their investigations to distort the truth. But these epidemiologists clearly are instilled with the belief that the workplace, particularly a modern petrochemical facility, can and must be safeguarded. That committment, which was on clear display among many of the federal officials at the workshop, is naturally even stronger among offi cials of unions representing the pet rochemical workforce. "We must identify the job places where cases of brain cancer have appeared to see if there's a common denominator," says Raphael Moure of the Oil, Chemical & Atomic Workers International Union. "What are the locations in the plant?" Once that's known, he says, "conditions can be improved and ex posures minimized." Like Moure, Sheldon Samuels of AFL-CIO now assumes that the risk for brain tu mors is elevated for workers in pet- 34 C&EN Nov 10. 1980 UCC 096441 Thomaa: anatyaaa auggaat alavatad risk rochemical plants. "The failure to identify specific agents is a disap pointment, but not an insurmount able difficulty," he says. Although a fear about elevated brain cancer incidence certainly is justifiable, the statements of the two union officials indicate they have al ready gone from fear, to belief, to plan of action. Here again, their profes sional commitments more or less make that course inevitable. But it Many chemicals Induce brain tumors In animals 2-AcetylamlnoMuorene Acrylonitrile 1-Aryl-3,3-dlalkyttr1azenes Azoettiane* Azoxyethane* Azoxymethene Butylnttraeouree Cycasln 1,2-Olethylhydraztne Dlethylnttroaantine* Diethyl auMate* Dfanethyttenzanthracene Dimethyl sulfate* Elakwnycbt Ethyl methanesulphonate EthylnttroeoMuret* Ethymitrosowea* 1-Mettiyl-2-benzyl-hydrazlne* Methyl methanesulphonate Methyimtrosouree* Methylnttroaowethane (Intraplacental) Procarbazine* Propane suitone Propylene (mine Pyrroilzidine alkaloids Vinyl chloride a These chemicals can cross the placenta to induce brain tumors in offspring. Source: Victor Aiewder et 1I..NIOSH comes when many of the epidemiol ogists are arguing over whether there is any elevated risk. If the risk is the same as that faced by the general population, it seems a rather striking challenge to the petrochemical in dustry to improve the workplace. "I'm puzzled by the widespread nature of the problem [of brain tumor incidence]," says H. Michael Utidjian of Union Carbide. "It seems wide spread over many populations and work classes, sometimes extending to white collar workers. 1 don't know of any other case of occupationally re lated cancer where it's appeared over such a wide range of workplaces." It is "oversimplistic," he adds, to say conclusively that brain tumors are due to exposures in the chemical in dustry. Several epidemiologists have con cluded that Drain tumor incidence, at least among some petrochemical workers, is not elevated For example, one such study was conducted by Chi-Pang Wen of Gulf Oil's medical and health resources division of the company's oil refinery in Texas--the same installation included in an NCI survey. Wen says his study "repre sents the largest retrospective cohort study of refinery workers to data in North America in terms of length of observation (44 years of employment experience) and person-years of ob servation." And ha concludes that "the results indicate no increased risk of brain tumors among workers at this Texas refinery." He and NCI's Terry L. Thomas (and her collaborators) differ sharply in their conclusions. "Our analyses suggest that refinery workers have an elevated risk of brain cancer," Thomas says. And it's possible that current studies "underestimate" that risk, she adds. Eventually, their differingconclusions have to be reconciled: They are, after all, looking at overlapping (al beit not identical) populations. Hence, it seems that discrepancies in analytic methods and in deciding what information to include stand in the way of reconciliation. Yet other epidemiologists, looking at different workplaces, come up with all kinds of puzzling dues. Michael R. Aldereon of the Institute of Cancer Research in the U.K. has surveyed eight refineries there and concludes there is no increased risk for brain cancer, although preliminary evi dence for increases in malignant melanoma (skin cancer) is being fol lowed up, he says. "But even if you accept some of the U.S. data, the problem of brain cancer is in no way like other hazards." Hints from other surveys provide a kaleidoscope of incongruent obser Watv raauka Indfcala no tncraaaad riak vations. Women more than men em ployed in the Swedish chemical in dustry might be at greater than nor mal risk for brain cancer. Workers in the rubber tire building industry in the U.S. are also at greater than nor mal risk, says Thomas Mancuso of the University of Pittsburgh School of Public Health. But so are workers in the electrical machinery and also paper (and allied products) indus tries. And so also are workers in cer tain segments of the transportation industry, according to epidemiolo gists at the University of Southern California in Los Angeles. Epidemiology is very much an outgrowth of clinical medicine. Ac cording to one wag, epidemiologic surveys are often really "hunches, shored up with statistics." However, the fact is that many clinicians have hit upon important medical treatments by following up their hunches. The same phenome non may be at work among the epi demiologists scrutinizing brain cancer incidence. Hence, the possibility that they are onto something means that their surveys, no matter how prelim inary, cannot be disregarded. All of this close scrutiny also may serve a broader purpose. For example, Selikoff and his colleague William J. Nicholson at Mount Sinai School of Medicine in New York City find evi dence of higher incidence for other kinds of cancer besides brain tumors among several groups of industrial workers: "Brain tumors may be `the canary' for locating other cancer risks," Nicholson says. Thus elevated risks of diseases, such as lung cancer and malignant melanoma, cited by several of the epidemiologists at the New York Academy of Sciences workshop well could prove to be a more significant public health prob lem than brain cancer. 3 CAEN Nov. 10. 1980 UCC 096442 ANOTHER llCHNOMK.ICAI \I)V\V 1 ( ROM SHARI* MURPHY'S LAW 'S DEFIED. "It anything can go wrong, it \x ill" and nsualK does, especiallx when you're entering an equation into a calculator. Until now, \ou ma\ ha\e been using the error-prone procedure of encoding and decod ing the steps of an Vjfe-mhlv code, 'I Tie Sharp 5lt)0 Scientific (tabulator eliminates this needless jprocedure. It's the iXt programmable calculator that allows xou to enter directlx, using algegtraie, logarithmic, exponential and or her functions in their familiar form. That's Murphx \ I .aw defied. The 5100 is also the first hand-held to give \ou a complete display of the equation: xou see 2 1 characters m a time in a dot matrix display that rolls right or left to accommodate one or more formulas of up to SO steps. Si' that while other programmable calculators displu\ onl\ one pwprain dip at a time, the 5100 displays the actual equation in its entirety, and in the wa\ \ou're accustomed to seeing it. If, anywhere along the line, xou'xc entered an incorrect character (Murphx V I .aw at work), the displax makes it eas\ for you to spot. You can then move the 5100's computer-hke cursor toanx part of the displax to correct it. In addition to the fast solution of frequent lx used formulas, the 5100 is extremely conxeiw ient for tierix ing difference tables, performing data enalx sis, numerical integration, simulations x of linear s\stems, statistical and probabilitx calculations, and manx other procedures. It has so manx' uses, it xxould take a book to fullx describe them, and someone has exert xeritten one.1' Sharp otters xott a complete line of iughlx adxanced scientitics, and it xou xxant the leading edge in scientific calculators, bux the 5100 (under SOO.OOp. i : i w - ; : Step Eknuiu Corp^, K) Keymm Place, FVvrxm, N.J. Q7W2 r-r-4 UCC 096443